Venus Express: Results of the Nominal Mission

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, E00B19, doi:10.1029/2008JE003202, 2008

for

Full Article

Introduction to the special section on Venus Express: Results of the Nominal Mission D. V. Titov,1,2 F. W. Taylor,3 and H. Svedhem4 Received 21 May 2008; accepted 7 August 2008; published 17 December 2008.

Citation: Titov, D. V., F. W. Taylor, and H. Svedhem (2008), Introduction to the special section on Venus Express: Results of the Nominal Mission, J. Geophys. Res., 113, E00B19, doi:10.1029/2008JE003202.

[1] More than 25 spacecraft from the United States and the Soviet Union visited Venus in the 20th century, but in spite of the many successful measurements they made, a great number of fundamental problems in the physics of the planet remained unsolved [Taylor, 2006; Titov et al., 2006]. In particular, a systematic and long-term survey of the atmosphere was missing, and most aspects of atmospheric behavior remained puzzling. After the Magellan radar mapping mission ended in 1994, there followed a hiatus of more than a decade in Venus research, until the European Space Agency took up the challenge and sent its own spacecraft to our planetary neighbor. The goal of this mission, Venus Express, is to carry out a global, long-term remote and in situ investigation of the atmosphere, the plasma environment, and some aspects of the surface of Venus from orbit [Titov et al., 2001; Svedhem et al., 2007]. [2] Venus Express continues and extends the investigations of earlier missions by providing detailed monitoring of processes and phenomena in the atmosphere and near-space environment of Venus. Radio, solar, and stellar occultation, together with thermal emission spectroscopy, sound the atmospheric structure in the altitude range from 150 to 40 km with vertical resolution of few hundred meters, revealing strong temperature variations driven by radiation and dynamical processes. In particular, by stellar occultation, Spectroscopy for Investigation of Characteristics of the Atmosphere of Venus (SPICAV) discovered a warm layer at the mesopause (100 km) apparently caused by adiabatic heating of descending air. Observations by the Visible and Infrared Thermal Imaging Spectrometer (VIRTIS) and radio occultation sounding by Venus Radio Science Experiment (VeRa) indicate remarkable latitudinal temperature variations at the cloud tops, as well as the presence of a convective region within the cloud deck all over the planet. The temperature field in the mesosphere of the southern hemisphere is similar to that observed in the northern hemisphere by earlier missions, thus indicating global north/south symmetry. 1 Max-Planck Institute for Solar System Research, Katlenburg-Lindau, Germany. 2 On leave from the Space Research Institute, Moscow, Russia. 3 Department of Atmospheric, Oceanic, and Planetary Physics, Oxford University, UK. 4 ESA/ESTEC, Noordwijk, Netherlands.

Copyright 2008 by the American Geophysical Union. 0148-0227/08/2008JE003202$09.00

[3] Multispectral imaging is performed to investigate the cloud morphology at different levels and its latitudinal and temporal changes. Venus Monitoring Camera shows evidence of convection in the cloud layer at low latitudes. Streaky cloud features indicate that the convection ceases in the middle and high latitudes. Sequences of images are used to track motions of the cloud features and to derive wind speeds in the 50– 70 km altitude range. The polar orbit of Venus Express provides near-nadir viewing of middle and high latitudes in the southern hemisphere, enabling the first detailed observations of the cloud morphology and dynamics, leading to the discovery and characterization of the complex ‘‘eye’’ of the southern polar vortex by VIRTIS. The imaging instruments also perform thermal mapping of the surface in the near-IR spectral windows on the nightside, searching for surface emissivity anomalies and signs of active volcanism. [4] The chemistry and dynamics of the mesopause region (100 km altitude) is being studied by observing nonlocal thermodynamic equilibrium (non-LTE) emission from O2 and NO molecules in the UV and near-IR. This airglow peaks at the equator, close to midnight and approximately at the mesopause level, indicating a thermospheric solarantisolar component in the atmospheric circulation. Composition measurements over a wide range of altitudes are providing vertical profiles of CO, H2O, HDO, HCl, HF, and SO2 in the mesosphere (70 – 100 km), and global mapping of CO, COS, H2O, and SO2 in the lower atmosphere at heights around 35 km. These results provide powerful tests of dynamical and chemical models of the Venusian atmosphere. [5] The magnetometer (MAG) and Analyzer of Space Plasma and Energetic Atoms (ASPERA) measure the magnetic field and densities of neutral atoms, ions, and electrons in situ. These observations determine the structure and properties of the circumplanetary plasma and characterize escape processes at Venus. They cover the time of solar minimum and thus complement Pioneer Venus investigations at solar maximum. Interestingly, the H/O ratio was found to be approximately two thus suggesting that these ions are escaping in the stoichiometric ratio of water. The magnetometer is detecting whistler signals during 70% of the pericenter passes. This is interpreted as evidence of lightning and the rate is estimated to be at least as frequent as on Earth. [6] The first results from the Venus Express mission were published in nine papers in the special section ‘‘Venus

E00B19

1 of 2

E00B19

TITOV ET AL.: INTRODUCTION

Express’’ in Nature, 450, 629 – 662, 2007. This special section contains original papers describing results and analysis of the observations during the whole nominal mission (4 June 2006 to 2 October 2007), along with supporting modeling efforts. The papers are grouped in topical blocks and are published in two groups. This first group of papers contains a mission overview and papers on atmospheric composition, atmospheric dynamics, non-LTE emissions, and surface investigations. A second group of papers, to appear about 3 months later, will contain papers on atmospheric structure, clouds and hazes, and plasma environment.

References Svedhem, H., et al. (2007), Venus Express: The first European mission to Ve n u s , P l a n e t . S p a c e S c i . , 5 5 , 1 6 3 6 – 1 6 5 2 , d o i : 1 0 . 1 0 1 6 / j.pss.2007.01.013.

E00B19

Taylor, F. W. (2006), Venus before Venus Express, Planet. Space Sci., 54, 1249 – 1262, doi:10.1016/j.pss.2006.04.031. Titov, D. V., E. Lellouch, F. W. Taylor, and L. Marinangeli, H. Opgenoorth, and the Venus Express Team (2001), Venus Express: An orbiter for the study of the atmosphere, the plasma environment, and the surface of Venus, Mission Definition Rep. ESA-SCI (2001)6, Eur. Space Agency, Paris. Titov, D. V., H. Svedhem, and F. W. Taylor (2006), The atmosphere of Venus: Current knowledge and future investigations, in Solar System Update, edited by P. Blondel and J. W. Mason, pp. 87 – 110, Springer, Berlin. H. Svedhem, ESA/ESTEC, Keplerlaan 1, Postbus 299, NL-2200 AG Noordwijk, Netherlands. F. W. Taylor, Department of Atmospheric, Oceanic, and Planetary Physics, Oxford University, Parks Road, Oxford OX1 3PU, UK. D. V. Titov, Max-Planck Institute for Solar System Research, Max Planck Strasse 2, D-37191 Katlenburg-Lindau, Germany. ([email protected])

2 of 2

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114, E00B33, doi:10.1029/2008JE003290, 2009

for

Full Article

Venus Express mission H. Svedhem,1 D. Titov,2 F. Taylor,3 and O. Witasse1 Received 30 October 2008; revised 18 December 2008; accepted 5 January 2009; published 19 March 2009.

[1] Venus Express is well and healthy and has now been providing exciting new data from

Venus, our nearby twin planet, for over 2 years. Many of the new results are presented and discussed in the subsequent papers in this special section. The overall scientific objective of Venus Express is to carry out a detailed study of the atmosphere of Venus, including the interaction of the upper atmosphere with the solar wind and the interaction of the lowest part of the atmosphere with the surface of the planet. In addition, the plasma environment and magnetic fields as well as some aspects of the surface of the planet are addressed. For the first time, investigations make systematic use of the transparent infrared spectral windows in order to probe the atmosphere in four dimensions: three spatial dimensions plus time. The spacecraft design is taken from Mars Express with some modifications necessary owing to the specific environment around Venus. The payload is composed of three spectrometers, a camera, a magnetometer, an instrument for detecting energetic particles, and a radio science package. The orbit is polar and highly elliptic, with a pericenter altitude of about 200 km over the northern polar region and an apocenter altitude of 66,000 km. Presently, the coverage of the southern hemisphere is very good, but important gaps still do exist. The coverage of the northern hemisphere is much less dense. Venus Express is a part of the European Space Agency’s program for the exploration of the inner solar system, which includes missions to study the Sun, Mercury, Venus, the Moon, Mars, and comets and asteroids. Citation: Svedhem, H., D. Titov, F. Taylor, and O. Witasse (2009), Venus Express mission, J. Geophys. Res., 114, E00B33, doi:10.1029/2008JE003290.

1. Introduction [2] Venus was a forgotten planet for more than a decade after the emphasis for investigations of the terrestrial planets shifted from Venus toward Mars during the late 1980s. There were, however, still a large number of fundamental questions to be answered about the past, present and future of Earth’s sister planet, and for an improved understanding of the general evolution of the terrestrial planets better knowledge of Venus is essential [Taylor, 2006; Titov et al., 2006a; Moroz, 2002]. The European Space Agency (ESA) launched Venus Express to open up opportunities for new investigations with a combination of instruments employing completely new techniques and improved versions of conventional instruments. The response to the initial results and findings has been very enthusiastic, generating renewed interest both in the scientific community and in the major space agencies, with new missions being developed or planned in Japan, the United States, and Russia. Venus has come back into the forefront in planetary science.

1

ESA, ESTEC, Noordwijk, Netherlands. Max-Planck Institute for Solar System Research, Katlenburg-Lindau, Germany. 3 Department of Atmospheric, Oceanic, and Planetary Physics, Oxford University, Oxford, UK. 2

Copyright 2009 by the American Geophysical Union. 0148-0227/09/2008JE003290$09.00

[3] Venus Express is an independent follow-up to the successful Mars Express mission, launched a few years earlier, and is reusing much of the basic design for spacecraft, launcher and ground system in order to keep costs down [Titov et al., 2001; Svedhem et al., 2007a, 2009]. It is the third spacecraft in the family that started with the (much larger) Rosetta spacecraft. Both Venus Express and Mars Express benefit greatly from the generic developments carried out for that mission. With only 3 years between mission approval and the launch date, plus about a year of preparatory work preapproval, Venus Express is by far the most rapidly developed scientific project of ESA. The short developments time and the significant design heritage have enabled a powerful mission to materialize at a substantially lower cost compared to a single mission developed in isolation. [4] With the new data from Venus Express, there is a picture emerging of Venus where the existing conditions and the ongoing processes are becoming clearer. Comparisons with the Earth will continue and intensify. Venus and Earth are very different twin planets, and while they will remain different, we are moving closer toward understanding how and why the two planets have their distinct characteristics, particularly with regard to climate [Svedhem et al., 2007b]. [5] This paper describes the mission objectives of the Venus Express mission and gives an overview of the main features of the spacecraft, with emphasis on the differences compared to Mars Express, and the scientific instruments on board. It discusses the mission scenario and the operational

E00B33

1 of 19

E00B33

SVEDHEM ET AL.: VENUS EXPRESS MISSION

orbit and the specific questions related to the choice of orbit and it gives an introduction to the various activities of the scientific operations. It discusses the coverage achieved to date by the different instruments with respect to longitude, latitude and local solar time, for the atmospheric and surface observations as well as the coverage of the different regions in and around the induced magnetosphere. Temporal coverage is discussed for a few special cases but the principles are applicable to most of the measured parameters. It outlines the place of Venus Express in the scientific program of the European Space Agency and finally it discusses topics to be studied in an extended mission, beyond May 2009, that was requested during the latter part of 2008.

2. Mission Objectives and Investigations [6] The overall scientific objective of Venus Express is to carry out a detailed and systematic study of a large number of aspects of the atmosphere of Venus. This includes the interaction of the upper atmosphere with the solar wind and the interaction of the lowest part of the atmosphere with the surface of the planet. In addition, dedicated instruments are focusing on the study of the plasma environment and magnetic fields. Some aspects of the surface are addressed. The objectives have been structured and organized under the following seven themes: (1) atmospheric structure, (2) atmospheric dynamics, (3) atmospheric composition and chemistry, (4) cloud layer and hazes, (5) energy balance and greenhouse effect, (6) plasma environment and escape processes, and (7) surface properties and geology. [7] The evolution of the planet and in particular the atmosphere and the climate are topics of great interest that span over all the themes. Most themes include aspects of comparative planetology and comparison with the other terrestrial planets, the Earth in particular, is an important objective. Direct comparison of data from Mars Express and Venus Express are enabled by the fact that several instruments are identical or similar on the two spacecraft. Also, several of the Venus Express scientists are involved in both missions. This opens up possibilities not found before on interplanetary spacecraft. 2.1. Atmospheric Structure [8] Knowledge about the atmospheric structure is of great importance for the understanding of the state of the atmosphere and the processes active therein. Previous missions provided a basic understanding of the thermal structure, but many questions remained to be answered [Taylor et al., 1980; Seiff et al., 1985]. The atmosphere can be roughly divided into three main layers based on the temperature distribution and the various processes that govern the state of the layers: the troposphere (0 –60 km), the mesosphere (60 – 100 km), and the thermosphere (above 100 km). The isothermal region between the troposphere (where temperature falls with height) and the mesosphere (where temperature increases with height) that would correspond to the stratosphere on the Earth is of very limited vertical extent on Venus and is not usually described as a separate layer. [9] Below 30 km the temperature is believed to be fairly constant all over the planet [Seiff et al., 1985] but the latitudinal coverage of the data is fairly poor. The mesosphere shows a much more variable temperature, especially in

E00B33

latitude, but what the dominating processes are is not clear [Lellouch et al., 1997]. [10] The thermosphere turned out to be surprisingly cool, especially at night (leading to the coining of the term ‘‘cryosphere’’ for the nighttime thermosphere). While it is clear that radiative processes play a major role, additional, and more densely sampled data covering all latitudes and local times are needed to achieve a good understanding [Fox and Bougher, 1991]. [11] Venus Express studies the thermal structure of the atmosphere by means of global, simultaneous, and spatially resolved spectroscopic observations in the wavelength range from UV to thermal IR. Specifically, it (1) investigates the upper atmosphere (140– 90 km) at high vertical resolution through solar and stellar occultation measurements [Bertaux et al., 2007b], (2) sounds the temperature of the middle atmosphere (100– 60 km) using spectroscopic observations of the 4.3 mm CO2 bands, with a coverage in latitude and local solar time corresponding to a spatial resolution of a few tens of kilometers or better [Formisano et al., 2006; Drossart et al., 2007a], (3) sounds the temperature in the altitude range 80– 40 km by radio occultation, providing vertical resolution of a few hundred meters [Ha¨usler et al., 2006; Pa¨tzold et al., 2007], and (4) maps the surface temperature on the nightside as a function of surface elevation [Drossart et al., 2007a]. [12] Together, these different techniques cover the range from 140 km down to 35 km. The optical measurements can be made only during nighttime since it would be very difficult to discriminate between thermal radiation and scattered sunlight. Unfortunately the PFS instrument, which would have used the 15 mm CO2 band for daytime profiling, is not operating. The use of other instruments has been rescheduled in order to as much as possible minimize the impact of this loss. 2.2. Atmospheric Dynamics [13] Data from previous missions and ground-based observations have shown that the general circulation of the atmosphere can be divided into two regimes: A retrograde zonal superrotation in the troposphere and mesosphere [Gierasch et al., 1997] and a solar to antisolar component across the terminator in the thermosphere [Bougher et al., 1997]. The zonal superrotation has a maximum wind velocity of about 100 m/s at the cloud top level (70 km), decreasing to almost zero at the surface. At the same time, there is a ‘‘Hadley-like’’ slower (about 10 m/s) overturning of the atmosphere from the equator to the high latitudes, with giant vortices at each pole recycling the air downward. No attempt to model the superrotation has been completely successful so far, indicating that the basic mechanisms of the phenomenon are unclear. More and better data, with improved spatial and temporal coverage, are needed in order to understand the detailed physical condition in this region and to modify the models to better reflect the reality. [14] Venus Express investigates the atmospheric dynamics by observing clouds at different levels [Drossart et al., 2007a; Markiewicz et al., 2007a], by deriving thermal winds from thermal profiles, and by monitoring airglow of different species. Specifically, it (1) measures the global wind fields in three dimensions and investigates whether the meridional circulation is one large basic ‘‘Hadley’’ cell extending from the surface to the upper atmosphere, or a variation of such a

2 of 19

E00B33

SVEDHEM ET AL.: VENUS EXPRESS MISSION

E00B33

Figure 1. The abundance of the most important minor gases in the atmosphere of Venus as determined by missions before Venus Express. The individual gases are color coded as shown in the bottom left corner. The sizes of the blocks indicate the spread in several measurements rather than measurement errors. The bars with arrows indicate the limits of detection for the Venus Express instruments.

cell, or something completely different; (2) studies the temporal behavior of the polar vortices: how they evolve and how they couple to the two main components of the global circulation [Piccioni et al., 2007]; and (3) studies how the transition between the retrograde superrotation and the solar to antisolar circulation takes place in the interface between the mesosphere and the thermosphere. [Drossart et al., 2007b]. 2.3. Atmospheric Composition and Chemistry [15] The main components of the Venus atmosphere are CO2 (96.5%) and N2 (3.5%). Sulfur-bearing trace gases, carbon and chlorine compounds, and water vapor are present in the atmosphere in amounts from few to few hundred parts per million (ppm) [de Bergh et al., 2006]. Figure 1 shows the mean abundance of the main trace gases and their vertical profiles measured in the Venus atmosphere by earlier missions. [16] Although present in small amounts, the trace gases are involved in important complex chemical cycles. In the region around the cloud tops photochemical reactions between CO2, SO2, H2O, and chlorine compounds lead to the formation of sulfuric acid, which is the main component of the cloud particles. The chemistry of the lower atmosphere is dominated by thermal decomposition of sulfuric acid, and thermochemical cycles that include sulfur and carbon species and water vapor. Surface minerals can also play a significant role in buffering the abundance of certain gases in the lower atmosphere [Fegley et al., 1997]. The main sulfur-bearing gas, SO2, is present in the Venus atmosphere in amounts of a few hundred ppm, which is much more than expected from

thermal equilibrium with the surface minerals. Pioneer Venus measured a strong continuous decline of the SO2 abundance at the cloud tops during its 14 years of operation, indicating possible recent volcanic activity. [17] Venus Express measurements of the chemical composition of the atmosphere address the following: (1) The abundance and spatial and temporal variation of SO2, SO, H2O, HCl, and CO at the cloud tops, to improve the understanding of the physical and chemical processes in this region, including the production of sulfuric acid aerosols [Bertaux et al., 2007a], (2) vertical profiles of SO, SO2, H2O and HDO, HCL, and HF between 80 km and the cloud tops, and vertical profiles of CO from the cloud tops up to about 120 km, by stellar and solar occultation [Bertaux et al., 2007b], and (3) the abundance and spatial variation of H2O, SO2, COS, CO, H2O, HCl, and HF in the lower atmosphere, to improve the understanding of chemistry, dynamics, and radiative balance of the lower atmosphere, and to search for local volcanic activity [Drossart et al., 2007a; Svedhem et al., 2007b]. 2.4. Cloud Layer and Hazes [18] Venus appears completely featureless in visible light owing to a thick cloud layer located between 50 km and 70 km altitude. However, images made in the UV blue spectral range show much structure, both at large scale and at small scale (Figure 2). It has important implications for the energy balance since about half of the solar energy entering into the atmosphere is absorbed in this region. The nature of the UV absorbing matter remains one of the mysteries of Venus. Earlier observations have shown that the upper cloud layer consists of micron-sized droplets of 75% sulfuric acid.

3 of 19

E00B33

SVEDHEM ET AL.: VENUS EXPRESS MISSION

E00B33

radiation penetrates through the atmosphere and heats the surface [Taylor et al., 2007]. The radiation balance, however, is not well constrained and in particular the spectral characteristics and the spatial and temporal variations of the outgoing radiation are not well known. [21] Venus Express measures the outgoing radiative flux (reflected solar light and thermal radiation from the atmosphere and the surface) over the near-IR spectral range and maps the temperature field and the cloud layer in three dimensions. This will constrain the optical models of the atmosphere and give an insight into the radiative and dynamical heat transport, and the role of the different species in the greenhouse mechanism and the planetary heat balance.

Figure 2. An image of Venus made by the Venus Monitoring Camera (VMC) instrument in ultraviolet light at 365 nm from a distance of about 35,000 km. The south pole is at the bottom right in the image and the equator at the top left. The cloud structure is shown to be very different in the polar region compared to the equatorial region. A bright elongated cloud is shown in the middle southern latitudes.

These droplets make up a major portion of the cloud mass but the processes and conditions for their formation are poorly understood, while the physical and chemical processes forming the rest of the cloud population are virtually unknown. [19] Venus Express sounds the structure, composition, dynamics, and variability of the cloud layer. Specifically, it (1) images the spatial distribution and temporal variability of the unknown UVabsorber [Markiewicz et al., 2007b; Drossart et al., 2007a]; (2) measures the vertical structure and microphysical properties of the upper cloud layer and the hazes above it, by stellar and solar occultation, limb observations, and nadir sounding [Bertaux et al., 2007a]; (3) measures the spatial variations of the opacity and particle sizes on the nightside (through the IR ‘‘spectral windows’’); this data together with information on the composition will help to constrain models of the cloud formation and evolution [Titov et al., 2006b]; and (4) searches for correlation in the spatial distribution of the unknown UV absorber with the abundance of SO2 and other trace gases. 2.5. Energy Balance and Greenhouse Effect [20] With a surface temperature of more than 730 K Venus has the strongest greenhouse effect in the solar system. The dense CO2 atmosphere is responsible for this extreme behavior, but water vapor and sulfur dioxide and the cloud layer also play a role [Crisp and Titov, 1997; Titov et al., 2007]. It is even more impressive when considering that Venus, in fact, absorbs less energy from the Sun than the Earth does, owing to the high albedo (76%), which is caused by the thick cloud cover of the planet. Less than 10% of the incoming solar

2.6. Plasma Environment and Escape Processes [22] Today Venus has only very little water, mostly in the form of water vapor. If it all were condensed it would form a global layer of about 3 cm, compared with 3 km on the Earth. There is no good reason why initially the two planets should have been significantly different and it is expected that Venus indeed has had large quantities of water, and possibly other volatile species, perhaps as much as the Earth. If so, how, when and why did this disappear? Pioneer Venus found that deuterium is enhanced relative to hydrogen about a factor 150 compared to the Earth’s value [Donahue et al., 1997]. This is an indication that hydrogen has been lost, while deuterium, having twice the mass, does not escape as easily as hydrogen. Therefore information on the present abundance of deuterium, hydrogen and oxygen and their escape rates are essential to the understanding of the history of water on Venus. [23] The lack of an internal magnetic field causes the solar wind to act on the upper atmosphere in a much more violent way than it does on the Earth. The solar wind interacts with the top of the ionosphere to form a complex system of plasma clouds, tail rays, filaments, and ionospheric holes on the nightside through which a substantial amount of material can leave the planet [Brace and Kliore, 1991]. The escape mechanisms induced by the solar wind are the dominant ones for the loss of heavy atmospheric gases such as oxygen because the gravitational force inhibits both Jeans escape and nonthermal escape. [24] To address the problems of atmospheric escape and investigate the plasma environment, Venus Express (1) determines the positions of the plasma boundaries for the different domains in the planetary environment, and their dependence on the solar activity [Zhang et al., 2008a, 2008b; Martinecz et al., 2008], (2) measures in situ, for the different domains, directional flux and energy of energetic neutral atoms, ions and electrons; of particular importance are the measurements of the escape rate of key species like hydrogen, deuterium and oxygen [Barabash et al., 2007a, 2007b], (3) measures the magnetic field in the different domains in the planetary environment [Zhang et al., 2006, 2007], (4) determines the vertical structure of the ionosphere, by radio occultation measurements [Ha¨usler et al., 2006; Pa¨tzold et al., 2007], and (5) measures the composition and energies of the undisturbed solar wind, as a reference for the Venus measurements and for comparison with similar measurements around the other planets of the solar system. The Venus Express measurements are taken at solar minimum activity, thus complementing the Pioneer Venus plasma studies that were acquired during solar maximum conditions.

4 of 19

E00B33

SVEDHEM ET AL.: VENUS EXPRESS MISSION

2.7. Surface Properties and Geology [25] Venus has a geologically young surface at an age of 700 Ma ± 200 Ma, as estimated on the basis of impact crater statistics. In addition there are indications that the age of the different regions do not differ much from each other. Volcanic and tectonic activities have strongly affected the surface [Solomon et al., 1992] forming highly deformed plateaus, tesserae, and extensive lowlands, planitiae. Many scientists favor a catastrophic global resurfacing of the crust, a mechanism unique among the terrestrial planets [Strom et al., 1994]. Important open questions are related to the surface mineralogy, the existence of present volcanic activity and the role of chemical interaction at the surface-atmosphere interface. Despite the fact that Venus Express is not carrying any instruments specifically devoted to surface science, it contributes to these studies in several ways [Titov et al., 2006b], by [26] 1. Investigating regions of high radar reflectivity, by bistatic measurements, focusing on highland regions like Aphrodite Terra, Beta and Atla Regio, and Maxwell Montes [Ha¨usler et al., 2006]. These areas have shown anomalously high reflectivity in the Magellan radar images. Unfortunately, these measurements have had to be discontinued since mid2007 owing to a still unexplained loss of power in the S band transmitter chain. [27] 2. Investigating the mass distribution in and around Atalanta Planitia by radio science gravity studies, through orbital trajectory analysis. [28] 3. Mapping the surface temperature and estimate the surface emissivity by observations in the 1 mm spectral window. [29] 4. Searching for surface hot spots and local deviations in chemical abundance in the lower atmosphere, in particular SO2, indicating possible volcanic activity. [30] 5. Searching for atmospheric waves generated by seismic activity and coupled to the atmosphere owing to the high atmospheric density at the surface [Drossart et al., 2007a].

3. Spacecraft [31] Venus Express uses the basic design of the Mars Express spacecraft, which was launched in 2003, adapted for the specific conditions at Venus [Sivac and Schirmann, 2009]. The main modifications to the spacecraft are related to the thermal control system and the power system. The most important characteristics are summarized in the following paragraphs. 3.1. Structure and Propulsion [32] The Venus Express spacecraft is based on a box-like structure with the dimensions 1.7 m  1.7 m  1.4 m. The distance from tip to tip of the deployed solar panels is about 8 m. The principal mechanical structures inside the spacecraft are the two fuel tanks, with MonoMethyl Hydrazine (MMH) as the fuel and Nitrogen TetrOxide (NTO) as the oxidizer, and a smaller helium tank for main tank pressurization. On both sides of these tanks two internal shear walls are built. The spacecraft, without solar panels, can be seen during the solar illumination testing in Figure 3 and the fully integrated spacecraft during its mating to the Fregat upper stage can

E00B33

be seen in Figure 4. The mass of the dry spacecraft is 633 kg, including 94 kg for the payload. The tanks were filled at launch with 570 kg of propellant, most of which was consumed during the orbit insertion maneuvers. Together with the launch vehicle adapter this made up the total launch mass of 1241 kg, leaving a comfortable margin to the maximum allowed mass of 1270 kg for the Soyuz Fregat launcher combination. Venus Express as well as Mars Express has been designed to sustain soft aerobraking, at a maximum dynamic pressure of 0.3 N/m2. 3.2. Thermal Control [33] The closer distance to the Sun, and the higher albedo of Venus compared to Mars, made it necessary to completely redesign the thermal control system. A new thermal blanket is used, giving the spacecraft a golden finish, in contrast to the black appearance of Mars Express. Several instruments and subsystems are connected to radiators mounted on different faces on the outside of the spacecraft. One side of the spacecraft, the -X side, carries the largest radiators which act as heat sinks for the coolers for the instruments requiring cryogenic temperatures. This side must never be exposed to the sun at any angle. The -Z side, where the main engine nozzle and the launch vehicle adapter interface reside, also must avoid illumination by the sun since these will absorb the solar heat very quickly. This design, together with specific operational constraints, ensures a low maximum temperature even in Venus orbit. Under some conditions the temperature would be too low, and therefore a set of electrical heaters are fitted to the most critical units. These heaters are either switched by mechanical thermostats or by the on board computer. 3.3. Power System [34] As the solar input in Venus orbit is about 2.6 kW/m2, which is about four times that of Mars or twice that of the Earth, the size of the solar panels could be significantly reduced. The original silicon-based solar cells have been replaced by gallium arsenide, optimized for operation at higher temperature. To keep the temperature in the proper range these cells are mounted in rows interleaved with optical solar reflectors, giving the solar panels a striped appearance. The electrical output in Venus orbit is approximately 1400 W. The remaining part of the power system, including the batteries, required only minor modification from the Mars Express version. 3.4. Telecommunication System [35] The spacecraft uses X band communications for both the telecommand uplink and the telemetry downlink. Fully redundant cross-strapped communication chains, including two 65 W traveling wave tube power amplifiers, feed the signal to the two high-gain antennas of 1.3 m and 0.3 m diameter. S band communications, with dual 5 W solid state power amplifiers, are included as a backup system and for use near the Earth during the first weeks following the launch. The data rate varies between 15 kbps to 228 kbps, corresponding to a downlink capability of between 400 Mbit to 6.5 Gbit per day, depending on the actual distance between Venus and the Earth. The antennas are body fixed and thus the spacecraft will turn toward the Earth for dumping the data

5 of 19

E00B33

SVEDHEM ET AL.: VENUS EXPRESS MISSION

E00B33

Figure 3. The compete spacecraft excluding solar panels during the solar illumination test at Intespace, Toulouse, France. The facility was upgraded for this test in order to be able to produce the required beam with an intensity 2.6 kW/m2 over the full surface of the spacecraft. The front side shows the 1.3 m diameter high-gain antenna covered with its solar shield, a thin foil with germanium deposit to keep the antenna within its operational temperature during all spacecraft attitudes. The back side of the smaller high-gain antenna is the white dish seen on the top platform where all remote sensing instruments are also located. The top and front sides are allowed to be continuously illuminated by the Sun at Venus, while all other sides have strict limitations on the duration of solar illumination. The height of the structure is about 1.4 m.

each telemetry session, usually 8 – 10 h per day. The data are stored in a 12 Gbit solid state data recorder during the observation passes. [36] The radio science investigation makes full use of the telecommunication system and an ultrastable oscillator has been added to one chain of the system to enhance the performance for this purpose. The S band system experienced a drop in signal level during the solar conjunction in July 2006 and has therefore only been used occasionally since then. The reason for this anomaly is still not known. [37] The main ground station is the Cebreros station west of Madrid, Spain, which is used for all uplink and most data downlink activity. The New Norcia station in Australia is used for radio science activities. Occasionally the NASA DSN assists by giving extra coverage, for radio science activities in particular, but also in connection with periods of high data rate requirements, for example, when measurements at high temporal resolution are made, such as highresolution movies of the cloud motions. 3.5. Operational Constraints [38] While pointing to the Earth from Mars during communications the Sun is always within a cone of 40° from the Earth, making it easy to avoid illumination of protected surfaces. In addition the high-gain antenna acts a thermal shield of the spacecraft. From Venus, however, the Sun can appear anywhere in the sky, making it impossible to avoid

illumination of specific surfaces. To deal with this a second, smaller, high-gain antenna has been mounted on top of the +Z face of the spacecraft, pointing in the opposite direction from the main antenna. Then it is possible always to find an attitude where illumination of the forbidden faces can be avoided. Figure 5 shows that inside the quadrature, where the Venus-Sun-Earth angle is less than 45°, the small antenna is used; that is, the spacecraft needs to rotate 180° about the z axis at entry and exit of the quadrature. In addition the spacecraft will need to rotate 180° about the x axis at inferior and superior solar conjunction.

4. Payload [39] The payload is composed of seven instruments in three different categories; spectrometers and (spectral) imagers for remote sensing, plasma and magnetic field instruments for in situ measurements, and the ultrastable oscillator used for radio science. [40] Great care was taken in selecting the instrument complement in order to make optimum use of the limited resources on board. A complication was the short time available for development since the project schedule was very compressed. As it turned out, a very well balanced set of instruments could be assembled from existing designs and even from spare parts and units from other recent missions, mainly Mars Express and Rosetta. In addition, two completely new instruments

6 of 19

E00B33

SVEDHEM ET AL.: VENUS EXPRESS MISSION

Figure 4. The fully integrated spacecraft is being mounted on top of the Fregat upper stage. The black conical structure is the launch adapter, which remains on the Fregat stage after separation. Protective covers are still mounted on the folded solar panels and on the instrument apertures and the two cylindrical baffles of the star trackers. The white recessed surface on the front is the VIRTIS thermal radiator. were designed, developed and built in this short time, namely the VMC and the SOIR. [41] Figures 6 and 7 show how the different instruments complement each other with respect to their fields of view and their spectral range and resolution. Venus Express is the first dedicated atmospheric mission to Venus since the discovery of the near infrared spectral windows [Allen and Crawford, 1984; Baines et al., 2006]. The payload was composed in order to maximize the benefit from this new opportunity to study the atmosphere in three dimensions. Table 1 lists the individual instruments and their main functions and Figure 8 shows the locations of the instruments on the spacecraft. 4.1. Remote Sensing Instruments [42] Spectroscopy for the Investigation of the Characteristics of the Atmosphere of Venus/Solar Occultation in the Infrared (SPICAV/SOIR) [Bertaux et al., 2007a, 2009] is a set of three spectrometers optimized for providing thermal profiles and composition in the upper atmosphere by observations in stellar and solar occultation mode. It is also used in nadir mode, together with other instruments. The solar and stellar occultation technique provides very good vertical

E00B33

resolution and high sensitivity for species of low abundance (like SO2, COS, CO, HCL, HF, and others). Particularly important molecules to measure are H2O and HDO, since the ratio between the two carries information on the history of the total amount of water on the planet. Temperature and density profiles are retrieved by analyzing the absorption in various CO2 bands. The instrument covers three wavelength bands: 110 –310 nm, 0.7 –1.7 mm and 2.2 – 4.4 mm (SOIR). The SOIR unit utilizes a miniaturized Stirling-cycle engine to cool the detector to 90 K, and an acoustooptic tunable filter, together with a grating used at a high order of diffraction, to achieve a spectral resolution (l/dl) of more than 20,000. [43] The basic SPICAV unit has an important heritage from the SPICAM instrument on Mars Express, while the SOIR channel is a completely new development. Important recent findings include the determination of the D/H ratio as a function of altitude, the discovery of a warm layer on the nightside at the base of the thermosphere (90 km) and the discovery of a set of previously unobserved absorption lines of the CO2 isotope 16O12C18O [Bertaux et al., 2008; Wilquet et al., 2008]. [44] Visible and Infrared Thermal Imaging Spectrometer (VIRTIS) is a combination of a visible (0.27– 1.1 mm) and an infrared (1.05– 5.2 mm) medium resolution imaging spectrometer (VIRTIS-M) with a high-resolution infrared (1.8 – 5.0 mm) spectrometer (VIRTIS-H) [Piccioni et al., 2009; Drossart et al., 2007a]. The infrared detectors are actively cooled to 80 K, while the visible sensor is passively cooled to below 200 K. The instrument is almost identical to the VIRTIS instrument presently flying on the Rosetta spacecraft. The imaging spectrometer has a field of view of 64 mrad  64 mrad, with a pixel size of 0.25 mrad  0.25 mrad. The images have a spatial dimension of 256  256 pixels and a spectral dimension of 432 lines. From near apocenter, the field of view covers about 1/3 of the diameter of Venus. Mosaics of 3  3 frames are constructed by repointing of the spacecraft in order to generate global images of the southern hemisphere. VIRTIS is addressing a large number of scientific questions and has generated dramatic images and video sequences of the south polar vortex, nonLTE emission patterns, profiles of abundance of several atmospheric gases, wind field maps and temperature profiles over the southern hemisphere, images of cloud structure at several altitudes and surface temperature and emissivity maps, to mention just a few. [45] Venus Monitoring Camera (VMC) is a small but efficient camera operating simultaneously in four narrow spectral bands at 365, 513, 965, and 1000 nm [Markiewicz et al., 2007a, 2009]. The UV band is mapping the cloud tops in reflected sunlight during daytime, where structure is visible owing to the still unknown UV absorber. These images are used for deriving global wind fields at this altitude and for studying the cloud morphology. The near-IR 1mm band is used for mapping the surface brightness during nighttime. The remaining two bands include O2 airglow emission and possible water vapor absorption. VMC has an unusual design with four optical chains, each one with its own permanent filter, sharing one CCD, with each optical system having its own dedicated quadrant on the CCD. When the CCD is read out in normal mode each image frame actually contains four images. The field of view is 17.5° (0.3 rad), resulting in a pixel footprint size on the surface ranging from 200 m at

7 of 19

E00B33

SVEDHEM ET AL.: VENUS EXPRESS MISSION

E00B33

Figure 5. The orbit of Venus in a Sun-Earth fixed frame, with the Sun at the center and the Earth at the left. The attitude of the spacecraft while Earth pointing will change depending on the relative position in this frame in a way such that the Sun is illuminating only two of the sides of the spacecraft (+X and +Z). To achieve this, the spacecraft is rotated 180° around the z axis at the points marked quadrature and 180° around the x axis at inferior and superior conjunction.

pericenter to 45 km at apocenter. VMC has sampled a large number of images, and important findings include the determination of global wind fields at different times, the characterization of the cloud morphology in different regions and synthesis of surface maps. [46] Planetary Fourier Spectrometer (PFS) is a high spectral resolution double pendulum spectrometer with two channels, a short wave channel (0.9 – 5.5 mm) and a long wave channel (5.5 – 45 mm) [Formisano et al., 2006, 2009]. The history of this instrument goes back to the Russian Mars-96 mission that unfortunately was lost in a launch failure in 1996. It flew again on Mars Express where it performs very well and has made important discoveries. Some of the main objectives for Venus were determination of the temperature field in the altitude range 55– 100 km, both on the nightside and on the dayside, mapping of the surface temperature, and making profiles of the abundance of a large number of gases in the middle and lower atmosphere. It was also to measure the outgoing thermal flux to determine the global radiation budget. Unfortunately, a mechanism that controls the mirror that switches the beam between the different calibration targets and the view toward the planet is stuck in its launch position, pointing toward a blackbody target. In spite of many attempts it has not been possible to move the scanner since the launch. 4.2. In Situ Instruments [47] Analyzer of Space Plasma and Energetic Atoms (ASPERA-4) is the fourth of this series of instruments to

fly in space, but the first of its kind around Venus [Barabash et al., 2007a, 2009]. ASPERA-4 is identical to ASPERA-3 presently in orbit around Mars on Mars Express. The instrument has four different sensors, housed in two separate units. The ion mass analyzer, IMA, measures ions separated by mass up to m/q = 40 for energies from 10 eV/q to 36 keV/q. The ions are mapped at 22.5° resolution over 360° in azimuth and at 4.5° resolution over ±45° in elevation with respect to the instrument orientation. The Neutral Particle Imager (NPI) measures the integral flux of neutral particles between 100 eV and 60 keV with an instantaneous field of view of 9° by 344° at an angular resolution of 4.6° by 11.5°. The Neutral Particle Detector (NPD) consists of two identical units that are basically pinhole cameras for the energy range 0.1 – 10 keV, each with a field of view of 9° by 90° and a resolution of 5° by 30° with separation of hydrogen and oxygen. The Electron Spectrometer (ELS) is a compact electrostatic analyzer for the energy range 1 eV to 15 keV with an energy resolution of 7%. The field of view is 10° in elevation and 360° in azimuth with a resolution of 22.5° in azimuth. ASPERA has made a large number of measurements in all accessible domains around Venus and in the solar wind and has characterized the different regions and their boundaries with respect to flux, energy and composition of the detected particles. A major result is the absolute determination of the escape rates of H+, O+ and He+ which is important for estimating the historic water content on the planet. [48] Magnetometer (MAG) is a dual sensor fluxgate instrument with the main sensor mounted on the tip of a 1 m

8 of 19

E00B33

SVEDHEM ET AL.: VENUS EXPRESS MISSION

E00B33

Figure 6. A comparison of the fields of view (FOV) of the remote sensing instruments onboard and the size of Venus corresponding to a distance of 20,000 km. The background image was taken by VMC at ultraviolet wavelengths from about 35,000 km. The south pole is at the bottom of the image. At apocenter the planetary disk is smaller than the VMC FOV and about three times the VIRTIS-M FOV.

long boom and an auxiliary sensor mounted at the foot of the boom on the outer surface of the spacecraft [Zhang et al., 2006, 2009]. This dual sensor configuration has proven very useful since the spacecraft was developed without a specific magnetic cleanliness program and therefore disturbances from the spacecraft itself are unavoidable. The data from the auxiliary sensor are used to clean the data of the main sensor, and a performance similar to that of a magnetically clean spacecraft is achieved. Venus Express is the first spacecraft where this concept has been used. MAG is the only instrument on board that is making continuous measurements. Samples are taken at 1 Hz frequency when the spacecraft is in the solar wind and at up to 128 Hz around the pericenter. The results include detailed models of the bow shock and the induced magnetopause, and detection of foreshock activities and upstream waves. Particularly interesting is the detection of whistler waves that are interpreted as evidence of lightning in the atmosphere. 4.3. Radio Science [49] Venus Radio science (VeRa) is the radio science investigation [Ha¨usler et al., 2006, 2009]. The main objective is

to determine the temperature and the density of the lower atmosphere in the altitude range 40– 90 km and to determine the electron density of the ionosphere up to the ionopause by means of radio occultation of the telemetry signal. Additional objectives are to carry out bistatic radar measurements over selected areas of the surface, including those highly reflective (at radar wavelength) areas discovered by the Magellan radar, in particular in areas of high elevation, like Maxwell Montes and Thetis Regio. Measurements are made by directing the main S/C antenna toward the area of interest on the surface, and receiving the reflected signal on a large antenna back on Earth. This was attempted successfully several times until the power in the S band channel was lost, as described above. The absorption in the atmosphere at S band is manageable but at X band it is too high for reliable operation. [50] VeRa occultation profiles have been used to map the thermal structure in the lower atmosphere at a high vertical resolution and medium horizontal resolution, separating the daytime and the nighttime regions. The lower parts show little difference between day and night but in the higher regions variations are clearly visible.

9 of 19

E00B33

SVEDHEM ET AL.: VENUS EXPRESS MISSION

Figure 7. A comparison between the wavelength ranges and spectral resolutions of the Venus Express remote sensing instruments.

Figure 8. The Venus Express spacecraft in a semitransparent view, showing the positions of the seven scientific instruments. The top face, the +Z platform, accommodates all the apertures of the remote sensing instruments. This side is turned toward Venus during observations. The two ASPERA units are mounted on the bottom side ( Z platform). 10 of 19

E00B33

SVEDHEM ET AL.: VENUS EXPRESS MISSION

E00B33

E00B33

Table 1. The Seven Instruments of the Venus Express Payloada Name

Function

SPICAV/SOIR

UV-IR spectrometer

VIRTIS VMC PFS ASPERA-4 MAG VeRa

Measured Parameters

Major Topics of Interest

Wavelength 110 – 310 nm, Solar and stellar occultation. Upper atmosphere: 0.7 – 1.65 mm, 2.2 – 4.4 mm Composition, D/H ratio, thermal profiles. Resolving power up to 20,000 UV-Vis-IR Virtis-M, 0.27 – 5.2 mm imaging Atmospheric dynamics, structure and composition. Imaging spectrometer Virtis-H 1.8 –5.0 mm single point at resolving power 2000 Surface temperature, emissivity. 4 band camera Wavelength 365 nm, 513 nm, 965 nm, 1000 nm Wind speed, clouds, surface, global context Fourier spectrometer Wavelength 0.9 – 45mm, resolving power 1200 Thermal profiles, composition. Presently non operational Energetic particles Electrons 1eV – 15 keV, ions 0.01 – 36 keV/q, Atmosphere solar wind interaction, neutral particles 0.1 – 60 keV atmospheric escape. Magnetic fields B field, 8 pT – 262 nT, at 128 Hz Induced magnetosphere. Plasma boundaries. Lightning Occultation of telecom link X- and S-band, Doppler shift, Lower atmosphere density and thermal structure. polarization, amplitude Ionosphere. Surface radar reflectivity.

a The total mass is 92 kg. The instruments together produce between 1 Gbit and 6 Gbit data per day, adapted to the telemetry transfer rate which depends on the actual Venus-Earth distance.

[51] The radio measurements also contribute to surface science by precise tracking of the spacecraft trajectory, from which data the gravity field is determined and gravity anomalies are derived. The gravity field of Venus is in general well known from previous missions, but a few anomalies need to be further investigated, for example in the region of Atalanta Planitia. A secondary objective is to study the solar corona at times near Venus superior conjunction, when the signal from Venus passes close to the disk of the Sun, revealing the structure, density and dynamics of the solar corona.

[54] Many scientific and technical constraints were considered when the operational orbit was selected. To achieve good global coverage, a polar obit was deemed to be essential. The need for a global view and long duration measure-

5. Mission Scenario and Operational Orbit [52] Venus Express was launched by a Soyuz-Fregat combination from Baikonur, Kazakhstan, on 9 November 2005 and arrived at Venus on 11 April 2006 (Figure 9). The injection into the heliocentric transfer orbit by the launcher was very precise and very little fuel had to be spent for trajectory corrections during the cruise to Venus. During the first part of the 153 days cruise all spacecraft and payload elements were checked out for proper operation. [53] The Venus orbit insertion was divided into several steps, starting with a capture maneuver which included a 52 min long burn by the 400 N main engine, to achieve the required delta-v of 1251 m/s [Warhaut and Accomazzo, 2009]. This delta-v is significantly higher than that required for the capture of Mars Express into Mars orbit. The reason is twofold; the differential velocity between the spacecraft and the planet is larger at Venus, and the larger mass of Venus requires more energy to reduce the apocenter height once captured. The insertion burn placed the spacecraft in a 9-day orbit with an apocenter height of 330,000 km. During this initial ‘‘capture’’ orbit, six blocks of scientific observations were included to benefit from the unique opportunity of having a very high apocenter distance that allowed global observations and dynamic studies on a large scale. After finishing the capture orbit, a sequence of smaller engine burns took place to reduce the apocenter height to the operational value. During the capture maneuver and the subsequent orbit reduction burns a total of 482 kg of fuel was consumed. The final operational orbit was reached on 6 May 2006.

Figure 9. Venus Express liftoff on a Soyuz-Fregat launcher from Baikonur Cosmodrome, Kazakhstan, 0333 UT, 9 November 2005. The spacecraft will subsequently be inserted into a perfect heliocentric trajectory toward Venus.

11 of 19

E00B33

SVEDHEM ET AL.: VENUS EXPRESS MISSION

ments for dynamic studies, together with close-up studies and near-planet in situ measurements, drove the requirement toward a highly elliptical orbit. A low circular orbit was not desired but would also not be feasible owing to the high fuel demands. An orbit with a pericenter height of 250 km and an apocenter height of 66,000 km, resulting in a period of 24 h, was selected. Such an orbit also had the advantage of allowing operators work at the same time each day and, since the main ground station is located in Spain, this turns out to be normal daytime working hours, saving a significant amount in the operational costs. [55] The remaining parameter to decide was the latitude of the pericenter. Since Venus rotates very slowly, only one revolution in 243 Earth days, there is very little polar flattening and consequently very little drift in the pericenter latitude, only about 3° per year. This makes the decision more critical for Venus than for Mars or the Earth. The orbital trajectory at arrival left basically only two options free for the choice of pericenter latitude, either 15° or +78°. In order to study one hemisphere at large distance and the other at short distance +78° was selected. This choice also nicely complements Pioneer Venus, which had its pericenter around the equator. The final orbit is shown in Figure 10. [56] A specific feature with a highly elliptical, polar orbit around Venus is the drift in pericenter altitude, due to perturbations by the solar gravity, which causes the pericenter altitude to drift upward or downward at a rate of 1 – 3 km per day. For the case of Venus Express the drift is upward until May 2009, then downward. This drift is compensated for by regular thrusting by four 10 N thrusters, to maintain the pericenter height between 250 km and 400 km and, after August 2008, between 175 km and 275 km. This will ultimately lead to the depletion of the fuel and an end of the mission in late 2013 if no countermeasures are taken. A possible alternative is to use aerobraking in the upper atmosphere of Venus to reduce the apocenter height (and the orbital period), as well as the rate of the drift of the pericenter height. A 12 h (37,000 km apocenter height) or 8 h (27,000 km apocenter height) orbit would dramatically reduce the drift rate and allow operation of the mission for another decade if required.

6. Science Planning and Operations [57] The planning of the scientific operations is a complex process that is carried out at different levels in a sequential manner. At the highest level, governed by the Science Requirements Documents, the Venus Express Science Working Team develops a long-term Science Activity Plan (SAP) typically covering a year or more of observations [Titov et al., 2006b]. This plan identifies specific orbital, and other, characteristics like solar illumination, accessible surface targets, solar eclipse periods, and Earth occultation periods. A scheme for the operations is then designed, on the basis of 10 different building blocks, called Science Cases, which can be put together in different combinations. [58] A medium-term plan (MTP) covering a period of 4 weeks, is agreed between representatives of the instrument teams on the basis of the guidelines in the SAP, approximately 3 months before its execution. [59] The lowest level in the planning cycle is the short-term plan (STP) which is prepared on a weekly basis. Here the

E00B33

Figure 10. The operational 24 h orbit around Venus. There are four major operational phases: (1) Around apocenter, at 10 h to 14 h, full disk imaging is made by VIRTIS and VMC, (2) along the ascending branch, between 14 h and 22 h, longduration observations are made for studies of the dynamics and cloud evolution, and (3) around pericenter, 23 h to 1 h, high spatial resolution observations and occultation measurements are performed. (4) Data transmission to the Earth takes place with the spacecraft Earth pointing from 2 h to 10 h orbital time. detailed command files with all settings for the instruments are included. The whole process is based on an exchange of a large number of files of different complexity between the PI teams, the science operations center (VSOC) and the mission operations center (VMOC), following a well-defined scheme [Koschny et al., 2009]. [60] The orbit is normally split into four parts where different activities take place (Figure 10). The region around apocenter is used for global mapping by VMC and synthesis of 3  3 frame mosaics for full disk coverage by VIRTIS. The ascending branch, from 10 h before pericenter to about 2 h before pericenter, has a long duration view of the same area of the planet and is used for studies of atmospheric dynamics. At times of high data rate, time lapse movies are composed here. The pericenter region, between about 1 h before and 1 h after pericenter, is shared between high spatial resolution

12 of 19

SVEDHEM ET AL.: VENUS EXPRESS MISSION

E00B33

E00B33

Figure 11. The present coverage of the VIRTIS-M images of the southern hemisphere as a function of latitude and local solar time is shown as a color-coded image. The coverage of the latitudes between 50°S and the south pole is obviously very good, while the coverage at the equator and northward of the equator is very limited.

nadir measurements and stellar occultation measurements. If in a solar eclipse period or an Earth radio occultation period, solar and radio occultation measurements also take place here. The descending branch of the orbit, from 2 h after pericenter to apocenter, is normally reserved for telecommunication and data downlink.

7. Coverage of Observations [61] A global and systematic survey of the atmospheric and plasma phenomena is a major goal of the Venus Express mission. To separate spatial variations from temporal variations, good coverage in a multiparameter space is essential for most measurements. Therefore priority is given to areas with poor coverage as far as possible. In general the coverage of the southern hemisphere is good and the coverage of the northern hemisphere is limited. This is inherent in the present orbit that has a larger distance to the planet, allowing more time for measurements over the southern hemisphere. [62] The coverage of VIRTIS spectral images over the southern hemisphere is shown in Figure 11. It can be seen that the polar region is well covered and the nightside of the southern hemisphere is reasonable well covered south of 50° southern latitude. On the day side, the equatorial region and the midlatitudes have very poor coverage. In addition the coverage of the northern hemisphere is very poor. Figure 12 shows the VIRTIS surface coverage from measurements dur-

ing nighttime. These data are essential for making surface maps. It can be seen that the coverage at midlatitudes between longitudes 150° and 30° is fairly good. The rest of the southern hemisphere is less well covered, in particular the longitudes between +45° to +150°. Also the surface coverage in the northern hemisphere is very poor. [63] The VMC dayside images provide uniform coverage of the southern hemisphere. These observations are used for cloud morphology and wind tracking studies. In the north only small-scale imaging is possible owing to the close proximity of the planet and the fast motion of the spacecraft. The VMC surface observations are limited to low latitudes (±40°). These images are made during nighttime, but owing to stray light only images taken during eclipse are useful. The observations complement well VIRTIS thermal mapping of the southern hemisphere. However, full coverage of all longitudes at low latitudes would require about 7 years since the part of the surface seen in eclipse drifts slowly because the rotation period of the planet is close to a Venus year. [64] The coverage of the SPICAV stellar occultation profiles depends on the availability of sufficiently bright UV stars. Presently, about 30 stars are used by SPICAV UV for occultation studies of the upper atmosphere, covering latitude ranging from 50°S to 40°N. A given star will always occult Venus at the same latitude but at varying local time, resulting in a coverage appearing as horizontal lines (Figure 13). The SPICAV and SOIR solar occultations by definition always

13 of 19

SVEDHEM ET AL.: VENUS EXPRESS MISSION

E00B33

E00B33

Figure 12. The present coverage of the VIRTIS-M nighttime images of the southern hemisphere as a function of latitude and longitude is shown as a color-coded image. The coverage of the latitudes between 40°S and 80°S for longitudes between 150° and 30° is fairly good, while the coverage of other locations is limited. These data are required for seeing the lower atmosphere and the surface. occur at the terminator, i.e., 0600 or 1800 local solar time (Figure 14b), except very close to the poles owing to the slight tilt of the axis of rotation of the planet. Full coverage in local solar time is thus not possible, although improved temporal coverage, to study both short-term and long-term variations, can be achieved by more measurements. [65] Figure 14a shows the coverage of the VeRa Earth radio occultations. It can be seen that even after three full occultation seasons the coverage is not very dense, in particular in the northern hemisphere, from the equator up to about 60° northern latitude. [66] ASPERA has a very good coverage inside the bow shock, except for in a small corridor straight south of the planet. It also has a good set of reference data taken in the undisturbed solar wind, well outside the bow shock (Figure 15). [67] MAG is the only instrument that is almost constantly on. In the solar wind it samples at a fixed frequency of 1Hz and close to pericenter it samples at 128Hz for a few minutes. This thorough coverage provides data for a multitude of processes within and outside the induced magnetosphere.

8. Mission Extension Topics [68] Presently, the Venus Express mission is funded for operations until May 2009. Most of the original primary objectives will have been met by then, but a number of old and new questions that have arisen would benefit from more data and longer total mission duration. As described above, the coverage for many parameters both in local solar time and

in latitude and longitude is not yet sufficient for detailed analysis. The present solar minimum has surprised us by being longer than normal, and the solar activity is expected to increase soon. It is very valuable to monitor the response by the atmosphere to this increase and to compare it to the situation with the low solar activity we have had over the past 2 years, and to the Pioneer Venus measurements at high solar activity in the years around 1980. An improved knowledge on the influence of the solar activity on the escape rate of various species may have important implications for the understanding of the evolution of the atmosphere. [69] Long-duration monitoring of key species, for example SO2, and of cloud properties and wind fields is essential to make conclusions on secular variation of composition and dynamics. Likewise, long-duration observations of the surface increase the likelihood of finding local hot spots and/or volcanic activity. An extended operational life will also allow time for a further reduction of the pericenter altitude, perhaps to 170 km in order to improve the basis for the in situ measurements of energetic particles and magnetic fields. The thermal structure and density in the otherwise inaccessible region between 170 and 190 km can be studied by atmospheric drag measurements. [70] An extension beyond 2010 would open up the possibility for joint observations with the Japanese Planet-C spacecraft. Planet-C is focusing on small-scale atmospheric dynamics and weather phenomena [Nakamura et al., 2007], while Venus Express aims at a more general study of the atmosphere. Planet-C will be placed in a highly elliptical

14 of 19

E00B33

SVEDHEM ET AL.: VENUS EXPRESS MISSION

Figure 13. Coverage of the SPICAV stellar occultations in the ultraviolet channel; latitude versus local time. From these measurements, thermal profiles in the altitude range 90 km to 140 km are derived together with and abundances of several atmospheric gases. The coverage is poor at high latitudes, and more data will be needed for a full coverage. These measurements are only effective over the dark hemisphere.

15 of 19

E00B33

E00B33

SVEDHEM ET AL.: VENUS EXPRESS MISSION

E00B33

equatorial orbit, complementing well the polar orbit of Venus Express. Joint observations would allow simultaneous highresolution imaging and global context imaging, and detailed 3-D imaging of the morphology of the upper clouds. This would be of particular interest for the studies of cloud dynamics and cloud evolution. [71] As discussed above an attractive way to extend the lifetime of the mission is to significantly lower the apocenter altitude by aerobraking. This would allow both an extended duration of the present observations and allow new investigations from the vantage of a different orbit. In addition density and temperature of the upper atmosphere can be retrieved from the drag exerted on the spacecraft by the atmosphere, parameters that in this altitude (140– 170 km) range cannot easily be measured remotely. Aerobraking in 2012 is under consideration and will be studied in detail in the near future.

9. Venus Express and the ESA Solar System Science Program

Figure 14. (a) Coverage of VeRa radio occultation data; latitude versus local time. From these measurements, profiles of atmospheric density and temperature are derived in the altitude range 35 km to 90 km. Additional data will be needed for a full coverage. (b) Coverage of the SOIR solar occultation measurements. By principle, only data along the terminator are available. More data will be needed for coverage at high southern latitudes and between 40°N and 80°N.

[72] Since Venus Express is very similar to Mars Express, with several instruments in common and simultaneous operations, interesting comparisons can be made. Together with data from Earth orbiting science satellites like Envisat and the Cluster spacecraft, an important basis for the comparative planetology of the terrestrial planets with atmospheres is laid. Possibilities also open up for simultaneous investigations of the solar wind and atmosphere interactions, and ‘‘space weather’’ studies on the three planets. The Solar System Science program of ESA is addressing all major bodies of the inner solar system from the Sun to Mars, and asteroids and comets. The three planets mentioned above all have ESA spacecraft presently orbiting them and the Sun is being observed, jointly with NASA, by the SOHO spacecraft from L1. [73] The last of the terrestrial planets, Mercury, will be studied in depth, jointly with Japan, by the dual spacecraft mission BepiColombo, which will be launched in 2014. The Rosetta spacecraft is already on its way to a comet and is due to arrive at its target, comet 67P/Churyumov-Gerasimenko, in 2015. During the cruise two asteroid flybys are scheduled. Thus, a fairly complete portion of the inner solar system is being researched, and synergistic effects can also be expected even if not predictable in advance. The information gathered will form a basis for the definition of other missions still to come. [74] ESA’s science program for the future, ‘‘Cosmic Vision,’’ with launches foreseen in the timeframe 2015 – 2025, is addressing a number of basic scientific themes with questions of a fundamental nature, like ‘What are the conditions for life and planetary formation?’ and ‘How does the Solar System work?’. A large number of missions were proposed for Cosmic Vision and several of them are now being studied in a competitive phase for a final selection in 2009. As discussed above, Venus Express already contributes to parts of these objectives, in particular in the field of comparative planetology and planetary evolution. [75] The ESA Planetary Science Archive, PSA, stores the data from all missions and makes them available to the world wide community. The first data from Venus Express, as well

16 of 19

SVEDHEM ET AL.: VENUS EXPRESS MISSION

E00B33

E00B33

Figure 15. The ASPERA coverage shown in Venus solar cylindrical coordinates (positive x axis pointing toward the Sun and the y axis showing the distance toward the Venus-Sun line). The coverage is excellent in most of the volume inside the bow shock, with a slightly weak coverage only straight south of the southern hemisphere. A very good set of solar wind reference data are available from measurements around the apocenter. as the other missions, are available for immediate electronic download at www.rssd.esa.int/psa.

10. Conclusions [76] Originally born as a ‘‘mission of opportunity’’ reusing the Mars Express design and available instruments, Venus Express has proven to be the most powerful mission ever in Venus orbit. A fortunate combination of a versatile spacecraft, a state of the art payload, and an efficient ground segment allows scientists to carry out a complex and systematic survey of the planet from the surface to the thermosphere and above. The observations include nadir and limb geometry, stellar, solar and Earth occultation measurements, and in situ plasma investigations. These studies have unveiled details of the atmospheric structure, composition, cloud morphology, dynamics and escape processes never observed before. [77] The papers following this one in this special section of JGR report on the most important findings to date. Many of the original objectives have been addressed to a significant

depth, while additional objectives are being formulated for the extended mission presently being planned. The analysis of the large amounts of data collected is a formidable task, and the science teams are engaged with both operations and processing the data. There is much more to be done with the data sets already acquired and new types of observations still to be carried out, such as atmospheric drag measurements and coordinated measurements with Planet-C, are promising new exciting results. [78] Everyone involved in the mission is appreciating the worldwide revival of the interest in Venus science that Venus Express appears to have triggered. The spacecraft and the instruments are in a good condition, and should continue to provide new data for scientists worldwide for several years to come. [79] Acknowledgments. The authors are grateful for the professionalism and enthusiastic commitment shown by all colleagues in the experimental teams, at EADS-Astrium, ESTEC, ESOC, and ESAC, which contributed greatly to the positive spirit and the success of the Venus Express mission.

17 of 19

E00B33

SVEDHEM ET AL.: VENUS EXPRESS MISSION

References Allen, D. A., and J. W. Crawford (1984), Cloud structure on the dark side of Venus, Nature, 307, 222 – 224, doi:10.1038/307222a0. Baines, K. H., et al. (2006), To the depths of Venus: Exploring the deep atmosphere and surface of our sister world with Venus Express, Planet. Space Sci., 54, 1263 – 1278, doi:10.1016/j.pss.2006.04.034. Barabash, S., et al. (2007a), The Analyser of Space Plasmas and Energetic Atoms (ASPERA-4) for the Venus Express mission, Planet. Space Sci., 55, 1772 – 1792, doi:10.1016/j.pss.2007.01.014. Barabash, S., et al. (2007b), The loss of ions from Venus through the plasma wake, Nature, 450, 650 – 653, doi:10.1038/nature06434. Barabash, S., et al. (2009), ASPERA-4: Analyser of Space Plasmas and Energetic Atoms, in Venus Express, edited by A. Wilson, Eur. Space Agency Spec. Publ., ESA SP-1295, in press. Bertaux, J.-L., et al. (2007a), SPICAV on Venus Express: Three spectrometers to study the global structure and composition of the Venus atmosphere, Planet. Space Sci., 55, 1673 – 1700, doi:10.1016/j.pss.2007. 01.016. Bertaux, J.-L., et al. (2007b), A warm layer in Venus’ cryosphere and highaltitude measurements of HF, HCl, H2O and HDO, Nature, 450, 646 – 649, doi:10.1038/nature05974. Bertaux, J.-L., et al. (2008), First observation of 628 CO2 isotopologue band at 3.3 mm in the atmosphere of Venus by solar occultation from Venus Express, Icarus, 195, 28 – 33, doi:10.1016/j.icarus.2008.01.001. Bertaux, J.-L., et al. (2009), SPICAV: Spectroscopy for the investigation of the characteristics of the atmosphere of venus, in Venus Express, edited by A. Wilson, Eur. Space Agency Spec. Publ., ESA SP-1295, in press. Bougher, S. W., M. J. Alexander, and H. G. Mayer (1997), Upper atmosphere dynamics: Global circulation and gravity waves, in Venus-II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 259 – 292, Univ. of Ariz. Press, Tucson. Brace, L. H., and A. J. Kliore (1991), The structure of the Venus ionosphere, Space Sci. Rev., 55, 81 – 163, doi:10.1007/BF00177136. Crisp, D., and D. V. Titov (1997), The thermal balance of the Venus atmosphere, in Venus-II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 353 – 384, Univ. of Ariz. Press, Tucson. de Bergh, C., V. I. Moroz, F. W. Taylor, D. Crisp, B. Be´zard, and L. V. Zasova (2006), The composition of the atmosphere of Venus below 100 km altitude: An overview, Planet. Space Sci., 54, 1389 – 1397, doi:10.1016/j.pss.2006.04.020. Donahue, T. M., D. H. Grinspoon, R. E. Hartle, and R. R. Hodges (1997), Ion/neutralescape of hydrogen and deuterium: Evolution of water, in Venus-II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 385 – 414, Univ. of Ariz. Press, Tucson. Drossart, P., et al. (2007a), Scientific goals for the observation of Venus by VIRTIS on ESA/Venus Express mission, Planet. Space Sci., 55, 1653 – 1672, doi:10.1016/j.pss.2007.01.003. Drossart, P., et al. (2007b), A dynamic upper atmosphere of Venus as revealed by VIRTIS on Venus Express, Nature, 450, 641 – 645, doi:10.1038/nature06140. Fegley, B., Jr., G. Klingenhoefer, K. Lodders, and T. Widemann (1997), Geochemistry of surface-atmosphere interactions on Venus, in Venus-II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 591 – 636, Univ. of Ariz. Press, Tucson. Formisano, V., et al. (2006), The Planetary Fourier Spectrometer (PFS) on board the European Venus Express mission, Planet. Space Sci., 54, 1298 – 1314, doi:10.1016/j.pss.2006.04.033. Formisano, V., et al. (2009), PFS: The Planetary Fourier Spectrometer, in Venus Express, edited by A. Wilson, Eur. Space Agency Spec.Publ., ESA SP-1295, in press. Fox, J. L., and S. W. Bougher (1991), Structure, luminosity, and dynamics of the Venus Thermosphere, Space Sci. Rev., 55, 357 – 489, doi:10.1007/ BF00177141. Gierasch, P., et al. (1997), The general circulation of the Venus atmosphere: An assessment, in Venus-II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 459 – 500, Univ. of Ariz. Press, Tucson. Ha¨usler, B., et al. (2006), Radio science investigations by VeRa on board the Venus Express spacecraft, Planet. Space Sci., 54, 1315 – 1335, doi:10.1016/j.pss.2006.04.032. Ha¨usler, B., et al. (2009), Venus atmospheric, ionospheric, surface and interplanetary radio-wave propagation studies with the VeRA radioscience experiment, in Venus Express, edited by A. Wilson, Eur. Space Agency Spec. Publ., ESA SP-1295, in press. Koschny, D., H. Svedhem, R. Hoofs, D. V. Titov, D. Merritt, A. Hulsbosch, and VSOC team (2009), Venus Express science planning and commanding, in Venus Express, edited by A. Wilson, Eur. Space Agency Spec. Publ., ESA SP-1295, in press.

E00B33

Lellouch, E., T. Clancy, D. Crisp, A. J. Kliore, D. Titov, and S. W. Bougher (1997), Monitoring of mesospheric structure and dynamics, in Venus-II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 295 – 324, Univ. of Ariz. Press, Tucson. Markiewicz, W. J., et al. (2007a), Venus Monitoring Camera for Venus Express, Planet. Space Sci., 55, 1701 – 1711, doi:10.1016/j.pss.2007. 01.004. Markiewicz, W. J., et al. (2007b), Morphology and dynamics of the upper cloud layer of Venus, Nature, 450, 633 – 636, doi:10.1038/nature06320. Markiewicz, W. J., et al. (2009), VMC: The Venus Monitoring Camera, in Venus Express, edited by A. Wilson, Eur. Space Agency Spec. Publ., ESA SP-1295, in press. Martinecz, C., et al. (2008), Location of the bow shock and ion composition boundaries at Venus: Initial determinations from Venus Express ASPERA-4, Planet. Space Sci., 56, 780 – 784, doi:10.1016/j.pss.2007.07.007. Moroz, V. I. (2002), Studies of the atmosphere of Venus by means of spacecraft: Solved and unsolved problems, Adv. Space Res., 29, 215 – 225, doi:10.1016/S0273-1177(01)00571-3. Nakamura, M., et al. (2007), Planet-C: Venus climate orbiter mission of Japan, Planet. Space Sci., 55, 1831 – 1842, doi:10.1016/j.pss.2007.01.009. Pa¨tzold, M., et al. (2007), The structure of Venus’ middle atmosphere and ionosphere, Nature, 450, 657 – 660, doi:10.1038/nature06239. Piccioni, G., et al. (2007), South polar features on Venus similar to those near the north pole, Nature, 450, 637 – 640, doi:10.1038/nature06209. Piccioni, G., et al. (2009), VIRTIS: The Visible and Infrared Thermal Imaging Spectrometer, in Venus Express, edited by A. Wilson, Eur. Space Agency Spec. Publ., ESA SP-1295, in press. Seiff, A., J. T. Schofield, A. J. Kliore, F. W. Taylor, S. S. Limaye, H. E. Revercomb, L. A. Sromovsky, V. V. Kerzhanovich, V. I. Moroz, and M. Y. Marov (1985), Models of the structure of the atmosphere of Venus from the surface to 100 km altitude, Adv. Space Res., 5, 3 – 58, doi:10.1016/ 0273-1177(85)90197-8. Sivac, P., and T. Schirmann (2009), The Venus Express spacecraft system design, in Venus Express, edited by A. Wilson, Eur. Space Agency Spec. Publ., ESA SP-1295, in press. Solomon, S., et al. (1992), Venus tectonics: An overview of Magellan observations, J. Geophys. Res., 97, 13,199 – 13,255. Strom, R. G., G. G. Schaber, and D. D. Dawson (1994), The global resurfacing of Venus, J. Geophys. Res., 99, 10,899 – 10,926, doi:10.1029/ 94JE00388. Svedhem, H., et al. (2007a), Venus Express - The first European mission to Venus, Planet. Space Sci., 55, 1636 – 1652, doi:10.1016/j.pss.2007. 01.013. Svedhem, H., D. V. Titov, F. W. Taylor, and O. Witasse (2007b), Venus as a more Earth-like planet, Nature, 450, 629 – 632, doi:10.1038/nature06432. Svedhem, H., et al. (2009), Venus Express: Mission overview, in Venus Express, edited by A. Wilson, Eur. Space Agency Spec. Publ., ESA SP1295, in press. Taylor, F. W. (2006), Venus before Venus Express, Planet. Space Sci., 54, 1249 – 1262, doi:10.1016/j.pss.2006.04.031. Taylor, F. W., et al. (1980), Structure and meteorology of the middle atmosphere of Venus: Infrared remote sounding from the Pioneer Orbiter, J. Geophys. Res., 85, 7963 – 8006, doi:10.1029/JA085iA13p07963. Taylor, F. W., H. Svedhem, and D. Titov (2007), Venus Express and terrestrial planet climatology, in Exploring Venus as Terrestrial Planet, edited by L. W. Esposito, E. R. Stofan, and Th. E. Cravens, Geophys. Monogr. Ser., vol. 176, pp. 157 – 170, AGU, Washington D. C. Titov, D. V., E. Lellouch, F.W. Taylor, L. Marinangeli, H. Opgenoorth, and Venus Express Team (2001), Venus Express: An orbiter for the study of the atmosphere, the plasma environment, and the surface of Venus, Mission Definition Rep. ESA-SCI (2001)6, Eur. Space Agency, Paris. Titov, D. V., H. Svedhem, and F. W. Taylor (2006a), The atmosphere of Venus: Current knowledge and future investigations, in Solar System Update, edited by P. Blondel and J. W. Mason, pp. 87 – 110, Springer, Berlin. Titov, D. V., et al. (2006b), Venus Express science planning, Planet. Space Sci., 54, 1279 – 1297, doi:10.1016/j.pss.2006.04.017. Titov, D. V., M. Bullock, D. Crisp, N. Renno, F. W. Taylor, and L. V. Zasova (2007), Radiation in the atmosphere of Venus, in Exploring Venus as Terrestrial Planet, edited by L. W. Esposito, E. R. Stofan, and Th. E. Cravens, Geophys. Monogr. Ser., vol. 176, pp. 121 – 138, AGU, Washington D. C. Warhaut, M., and A. Accomazzo (2009), The ground segment and mission operations, in Venus Express, edited by A. Wilson, Eur. Space Agency Spec. Publ., ESA SP-1295, in press. Wilquet, V., et al. (2008), Line parameters for the 01111-00001 band of C12O16O18 from SOIR measurements of the Venus atmosphere, J. Quant. Spectrosc. Radiat. Transfer, 109, 895 – 905, doi:10.1016/j.jqsrt. 2007.12.021.

18 of 19

E00B33

SVEDHEM ET AL.: VENUS EXPRESS MISSION

Zhang, T. L., et al. (2006), Magnetic field investigation of the Venus plasma environment: Expected new results from Venus Express, Planet. Space Sci., 54, 1336 – 1343, doi:10.1016/j.pss.2006.04.018. Zhang, T., et al. (2007), Little or no solar wind enters Venus’ atmosphere at solar minimum, Nature, 450, 654 – 656, doi:10.1038/nature06026. Zhang, T. L., et al. (2008a), Initial Venus Express magnetic field observations of the magnetic barrier at solar minimum, Planet. Space Sci., 56, 790 – 795, doi:10.1016/j.pss.2007.10.013. Zhang, T. L., et al. (2008b), Initial Venus Express magnetic field observations of the Venus bow shock location at solar minimum, Planet. Space Sci., 56, 785 – 789, doi:10.1016/j.pss.2007.09.012.

E00B33

Zhang, T. L., et al. (2009), MAG: The fluxgate magnetometer of Venus Express, in Venus Express, edited by A. Wilson, Eur. Space Agency Spec. Publ., ESA SP-1295, in press. H. Svedhem and O. Witasse, ESA, ESTEC, Keplerlaan 1, NL-2200 AG Noordwijk, Netherlands. ([email protected]) F. Taylor, Department of Atmospheric, Oceanic, and Planetary Physics, Oxford University, Oxford OX1 3PU, UK. D. Titov, Max-Planck Institute for Solar System Research, Max-PlanckStrasse 2, D-37191 Katlenburg-Lindau, Germany.

19 of 19

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, E00B07, doi:10.1029/2008JE003074, 2008

for

Full Article

A latitudinal survey of CO, OCS, H2O, and SO2 in the lower atmosphere of Venus: Spectroscopic studies using VIRTIS-H E. Marcq,1,2 B. Be´zard,1 P. Drossart,1 G. Piccioni,3 J. M. Reess,1 and F. Henry1 Received 14 January 2008; revised 8 April 2008; accepted 10 July 2008; published 23 September 2008.

[1] The high-resolution channel (R ’ 2000) of the Visible and Infrared Thermal Imaging Spectrometer (VIRTIS) instrument (VIRTIS-H) aboard Venus Express has provided numerous spectra of the nightside infrared thermal emission in the 2.3-mm window. Mixing ratios of various minor species in the 30–40 km range could therefore be inferred using this spectral window at higher latitudes accessible to the spacecraft but which cannot be observed from Earth. The previously known enhancement in carbon monoxide (CO) toward high latitudes is confirmed and extended up to 60° with a mixing ratio varying from 24 ± 3 to 31 ± 2 ppmv at 36 km. Measurements of carbonyl sulfide (OCS) also agree with the previously suspected latitudinal variations that are anticorrelated with those of CO, ranging between 2.5 ± 1 and 4 ± 1 ppmv at 33 km. New constraints were also derived on the mean abundance of water vapor (H2O, 31 ± 2 ppmv) and sulfur dioxide (SO2, 130 ± 50 ppmv) in the probed altitude range. CO and OCS variations are interpreted as caused by large-scale vertical motions, an explanation under current testing by various chemical and dynamical modeling. In such a case, these variations may help constrain the chemical time scale of those species in the lower troposphere. Citation: Marcq, E., B. Be´zard, P. Drossart, G. Piccioni, J. M. Reess, and F. Henry (2008), A latitudinal survey of CO, OCS, H2O, and SO2 in the lower atmosphere of Venus: Spectroscopic studies using VIRTIS-H, J. Geophys. Res., 113, E00B07, doi:10.1029/2008JE003074.

1. Introduction [2] The discovery of Venus’ nightside near infrared emission in 1983 by Allen and Crawford [1984] provided a new, valuable method to investigate the deep atmosphere of the planet. The physical origin of this emission was rapidly understood in the 1980s, on the basis of its physical properties and successful modeling [Crisp et al., 1989; Kamp et al., 1988]. The hot and deep layers of the lower atmosphere emit strong thermal radiation that can partly escape the thick atmosphere in a few narrow spectral windows thanks to the sub-Lorentzian behavior of CO2 line shape and the weakly absorbing scattering by the H2SO4 particles within the overlying cloud layers. In the 2.3-mm window considered hereafter, the emission originates from the 30– 40 km altitude range (average temperature of about 500 K). [3] Besides the imaging of the lower cloud deck whose opacity modulates the intensity of these emissions [Carlson 1 Laboratoire d’E´tudes Spatiales et d’Instrumentation en Astrophysique, UMR8109, CNRS, Observatoire de Paris, Meudon, France. 2 Laboratoire de Me´te´orologie Dynamique, UMR8539, CNRS, Universite´ Paris 6, Paris, France. 3 Istituto di Astrofisica Spaziale e Fisica Cosmica, Istituto Nazionale di Astrofisica, Rome, Italy.

Copyright 2008 by the American Geophysical Union. 0148-0227/08/2008JE003074$09.00

et al., 1991], the main scientific interest comes from its spectroscopic analysis. The comparison with radiative transfer synthetic models [Be´zard et al., 1990; Kamp and Taylor, 1990; Pollack et al., 1993; de Bergh et al., 1995] provided numerous constraints on the abundance of minor gaseous constituents. In the 2.3-mm window, carbon monoxide (CO), carbonyl sulfide (OCS), water vapor (H2O, HDO), sulfur dioxide (SO2) and hydrofluoric acid (HF) were detected and measured thanks to their spectral signatures. Furthermore, the spectral behaviors of CO and OCS are rich enough so that not only their mean abundance level, but also their respective vertical gradients could be derived, and were shown to be opposite by Pollack et al. [1993], with CO increasing with altitude as it is produced by the photochemistry of CO2 above 60 km. [4] The study of local compositional variations began later, in the 1990s, when Collard et al. [1993] noticed in the NIMS/Galileo spectra recorded during the flyby that the spatial variations of the CO band shape indicated a northern enrichment at higher latitudes of about 35%. More recently, spectra have been acquired at the NASA IRTF using the spectrometer SpeX during the quadratures that occurred since 2003. Their better spectral resolution compared to NIMS allowed us to confirm the CO northern enrichment and report a similar enrichment in the southern hemisphere, as well as a correlated depletion of OCS [Marcq et al., 2005, 2006] in the portions of Venus’ nightside that can be

E00B07

1 of 9

E00B07

MARCQ ET AL.: VIRTIS-H STUDY OF TRACE GASES VARIATIONS

Table 1. Summary of Mentioned Science Cases Science Case

Description

1 2 3 5 7

pericenter nadir observations off-pericenter nadir observations apocenter mosaic observations stellar occultations limb observations

observed from Earth (below 40° in latitude). A latitudinal variability of the vertical gradient of these two species in the probed layers could also be measured thanks to the high signal-to-noise ratio in the best data sample, and it appears that both gradients were steeper at higher latitudes than in the equatorial region [Marcq et al., 2006]. Spatial variations in water vapor have also been sought by several observers [Drossart et al., 1993] since the localized enrichment reported by Bell et al. [1991], which was hypothesized to be due to cloud subsidence and evaporation. [ 5 ] Keeping in mind that CO increases and OCS decreases with increasing altitude, the latitudinal variability of CO and OCS can be qualitatively understood as being primarily caused by the global meridional dynamics. At high latitudes, downward fluxes from the Hadley cell circulation bring air depleted in OCS and enriched in CO, whereas the ascending branch of these cells has the opposite effect close to the equator. Quantitative modeling is currently progressing (Y. Yung et al., Modeling the distribution of OCS in the lower atmosphere of Venus, submitted to Journal of Geophysical Research, 2008; E. Marcq et al., Simulations of the latitudinal variations of CO and OCS below the clouds of Venus using a general circulation model, submitted to Journal of Geophysical Research, 2008). Besides the use of minor species as tracers of the deep circulation, the study of the compositional variability can also be related to the geological activity of the surface; a local enrichment in water vapor and SO2 and a correlated modification of the D/H ratio of water may indicate an ongoing volcanic eruption [Donahue et al., 1997; Grinspoon, 1993; Donahue, 1999]. [6] After more than a year of activity, the VIRTIS instrument aboard ESA’s Venus Express spacecraft has provided numerous spectra from the nightside of Venus very similar to the SpeX/IRTF ones. Using very similar methods to our previous studies, we present in this paper the new constraints we have derived from these data about the spatial variations of CO, OCS and water vapor, as well as a semiquantitative interpretation of the aforementioned variations.

E00B07

probe, there is an asymmetry in the observation times for both hemispheres: the northern one is seen from a nearer vantage point, but for shorter durations. [8] Within each 24-hour orbit, 8 hours are dedicated to communication with the Earth. The remaining 16 hours are used by the six functioning instruments (ASPERA, MAG, SPICAV-SOIR, VeRa, VIRTIS, VMC) for the various Venus-related observations (such as limb or nadir sounding, solar or stellar occultations) according to preplanned schedules named ‘‘science cases.’’ A brief summary of the science cases mentioned in this article is available in Table 1 [from Titov et al., 2006b]. 2.1.2. VIRTIS-H [9] The VIRTIS spectral imager is described in detail by Coradini et al. [1998]. Here we describe the characteristics of the H-channel that are relevant to our analysis. The specific optical device of the H-channel consists of an echelle grating spectrometer, and dispersion orders from 8 to 15 are focused on the detector plane. The 2.3-mm emission can be seen on three among these orders, designated as orders 5, 6 and 7. The spectral resolution of each of these spectra varies with the wavelengths, as shown on Figure 1. The spectra are then sampled on 432 discrete wavelengths for each dispersion order, although the first tens of pixels on the right side (short wavelengths) of the focal plane array usually yield smaller radiances than expected because of the low grating efficiency on the edges of the orders, thus reducing their effective spectral ranges. Another peculiarity of the array reading system that we had to take into account in our processing methods is the difference in fluxes between odd and even pixels along a given spectrum. Nevertheless, this peculiarity is partially corrected by the calibration process, yielding a few percent difference in the absolute intensities between even and odd pixels. 2.2. Quality of the Observations 2.2.1. Orbits [10] We have currently reviewed all VIRTIS-H spectra between orbits 7 (27 April 2006) and 298 (13 February 2007). The proportion of orbits during which nightside spectra could be successfully acquired is close to 40%. This

2. Observations 2.1. VIRTIS and the Venus Express Mission 2.1.1. Brief Overview of the Mission [7] A detailed description of the Venus Express mission is provided by Titov et al. [2006a]. Here we summarize the mission characteristics relevant to this study. The orbit of the spacecraft is quasi-polar, with a periapsis located at 78°N, and highly elliptical (the distance to the surface ranging from 250 km up to 66000 km) with a 24-hour period. As a consequence of the Keplerian motion of the

Figure 1. Spectral resolution versus wavelength for orders 5 (red), 6 (green), and 7 (blue).

2 of 9

MARCQ ET AL.: VIRTIS-H STUDY OF TRACE GASES VARIATIONS

E00B07

E00B07

Table 2. Processed Orbits Orbit

Number of Processed Spectra

Science Case

Latitude on Venus

Single Spectrum S/N Ratio

007 024 027 033 040 042 060 061 073 074 076 077 078 079 082 088 095 096 097 098 100 102 103 104 110 110 111 113 116 125 128 129 134 136 139 143 148 243 248 249 258 277 284 298

128 100 35 32 50 40 160 140 230 230 128 192 300 200 80 80 50 1728 40 1664 30 30 30 50 2880 150 640 50 50 32 1400 320 1200 200 40 96 50 160 1100 150 250 96 51 200

N/Aa 1 1 7 7 7 7 2 1 5 1 2 1 1 1 7 2 2 1 2 2 2 5 2 2 2 2 2 2 5 2 2 2 2 2 2 2 1 3 1 5 1 2 1

29°S – 26°S 20°S – 55°N 20°N – 60°N 20°N – 50°N 0° – 60°N 40°N – 60°N 10°S – 70°N 30°S – 50°N 30°S – 20°N 30°S – 20°N 35°N – 55°N 44°S – 40°S 10°S – 70°N 0° – 70°N 25°N – 60°N 30°N – 70°N 40°N – 70°N 58°S – 44°S 40°N – 60°N 55°S – 28°S 45°N – 55°N 30°N – 55°N 30°N – 50°N 10°N – 55°N 50°S – 15°S 10°N – 60°N 49°S – 24°S 35°N – 60°N 35°N – 60°N 35°N – 60°N 60°S – 45°S 11°S – 50°N 60°S – 45°S 56°S – 54°S 20°N – 70°N 10°S – 5°N 11°N – 25°N 25°S – 40°N 43°S – 40°S 0° – 60°N 30°S – 0° 20°S – 5°S 20°S – 50°N 10°S – 40°N

10 3 – 10 8 – 12 4–8 4–6 6 3–8 4–8 4 – 12 10 – 30 4–8 5 2–4 4 – 25 2–6 4 – 18 4–8 3–4 3 3–4 4 5 5 2.5 4 4 2–8 3 3 3 4 4 4 10 3–4 3 5 12 10 20 6 20 4 – 10 4 – 14

a

During insertion orbit.

proportion and the temporal distribution of these orbits are consistent with the expected requirements (VIRTIS operational and pointing to a location on the nightside). Nevertheless, a subset of these orbits (14% of the total number) show a S/N ratio significantly better than those of the typical spectra taken at similar latitudes. They are summarized in Table 2. The retrievals shown in section 3.2 have therefore been conducted on this subset. 2.2.2. Influence of the Observing Mode (‘‘Science Case’’) [11] Quite surprisingly, the periapsis observations (case 1) were not the ones which consistently yielded the best spectra. It can be understood when we consider that the limiting factor in the processing of the VIRTIS-H spectra is the signal-to-noise ratio. In order to improve it, we have to integrate signal for as long as possible: at least a couple of minutes (around 20 frames in case 1) and up to half an hour, and a few hundred spectra for off-pericenter (case 2) or apoapsis mosaic observations (case 3), when the orbital velocity is slower than for periapsis observations. Longer

integration times are not suitable for a study of spatial variations, since the location on Venus of the various spectra would be blurred owing to the motion of the spacecraft. As a comment, it should be noted that even if it could theoretically be done, it would be useless to shorten the integration time to enhance the spatial resolution below roughly 30 km, since multiple scattering of the thermal emission by the cloud particles causes an intrinsic blurring of the signal over such a horizontal scale. Actually, our integration signal causes a latitudinal extension of the spectra’s location over a few degrees at most (longitudinal extension can be neglected owing to the quasi-polar orbital motion), thus enabling an integrated S/N ratio around 100 (individual spectra’s S/N ratios are given in Table 2). [12] Considering this trade-off between spatial resolution and signal-to-noise ratio, it appears that case 1 observations are well suited to low-latitude regions, where the signal is strong enough to cope with the moderate integration durations available in this science case owing to the high orbital velocity. On the other hand, midlatitudinal to high-latitudinal

3 of 9

E00B07

MARCQ ET AL.: VIRTIS-H STUDY OF TRACE GASES VARIATIONS

Figure 2. Superimposition of averaged nightside spectra of the same location on orders 5 (red), 6 (green), and 7 (blue) and their respective zero level. Note also the difference of spectral resolution between the various orders, in agreement with Figure 1. regions require a slower orbital velocity which can ensure a longer integration in spite of an increased distance. This explains why high-latitude regions have been mostly successfully observed during science cases 2, 5 or even 3. 2.3. Comparison with Similar Earth-Based Spectra [13] Sets of very similar spectra in terms of spectral range (1.92 – 2.52 mm) and resolution (R = 2000) have been acquired since 2003 using SpeX/IRTF by Marcq et al. [2005]. These observation campaigns were intended as a benchmark for the methods used hereafter, so that a comparison between these previous studies [Marcq et al., 2005, 2006] and the present one is very interesting. However there are some differences between both data sets that should be highlighted. 2.3.1. Spatial and Temporal Coverage [14] First of all, the spatial and temporal coverage is of course much better for VIRTIS, a dedicated instrument aboard a space probe orbiting Venus, than for the groundbased observations. From Earth, the nightside of Venus can be successfully observed only for a few weeks between inferior conjunction and maximal elongation. This period of favorable observing geometry occurs only twice in every 19.2 months. In contrast, VIRTIS has been providing spectra of the nightside of Venus on a regular basis (40% of the orbits on average) since April 2004 and will continue until at least the end of the extended mission in 2009. Another interesting feature of the Venus Express observations is that the quasi-polar orbit enables the observation of northern and southern circumpolar regions which are otherwise unreachable; the farthest observations with respect to the Venusian equator using SpeX could only reach 40 degrees of latitude. 2.3.2. Preliminary Processing [15] Another major advantage of space-based observations is the absence of any sunlight reflected from the dayside in virtually all nightside spectra, thanks to the apparent size of Venus as seen from the spacecraft. For SpeX observations, the much smaller angular diameter of

E00B07

Venus and the fact that a significant portion of the visible Venusian disk is illuminated near maximal elongation can lead to a strong scattering of the much brighter light from the solar illuminated portion of the planet onto the nightside, reducing the attainable S/N. The removal of the diffused component is a significant step in the processing and a major source of uncertainty for the retrievals of Marcq et al. [2005] and Marcq et al. [2006]: it prevented the observation of the nightside hemisphere close to the terminator, thus contributing to the narrowness of the latitudinal range of Earth-based observations. [16] Nevertheless, some processing is also needed for VIRTIS-H spectra, since the 2.3 mm window is covered by 3 different orders of dispersion that should agree on the actual intensity observed within their common spectral ranges. A typical superimposition is shown on Figure 2 for orders 5, 6 and 7. The differences on the actual value of the zero-intensity level are clearly visible between order 6 on one hand and orders 5 and 7 on the other hand. In the subsequent processing, the actual zero level for order 6 is computed from the zero levels measured on the orders 5 and 7, whose mutual agreement despite being on opposite sides of the 2.3-mm window make more trustworthy than order 6’s short-wavelength zero level. [17] As a last comment, Earth-based observations must also be corrected for the spectral signature of the terrestrial atmosphere, especially for water vapor which is much more abundant on Earth than on Venus. Such a concern is absent for VIRTIS observations, which makes them more reliable, especially for water vapor retrievals. 2.3.3. Signal-to-Noise Ratio [18] One major asset of Earth-based observations compared to VIRTIS-H ones is their quality. The longer integration time and especially the much greater light collecting area available more than compensate for the much greater distance between the instrument and Venus. A typical individual VIRTIS-H spectrum in a low-latitude to midlatitude region has a S/N ratio of between 5 and 10. Averaging over a few tens of spectra improves this ratio to about 50, at the expense of a having a spatial precision roughly equivalent to that of SpeX observations with a seeing of about 1 arcsec. SpeX spectra which are comparable in terms of integration time, location on Venus and latitudinal extension show an S/N ratio of 100 or higher. Thus, rather than supplementing Earth-based observations, VIRTIS provides a long-needed complementary spatial and temporal coverage of nightside deep atmosphere.

3. Retrievals 3.1. Algorithms 3.1.1. Synthetic Model [19] The forward synthetic model that we use is exactly the same as described by Marcq et al. [2006]. A typical fit is shown on Figure 3. [20] A major flaw of all radiative transfer models used to investigate the 2.3-mm window is the poor fit in the CO2 dominated spectral domain between 2.2 and 2.3 mm. Therefore, we have tried the new CDSD-750 database for CO2, suited for high-temperature calculations as encountered in the probed altitude region. Unfortunately, the fit we obtained was worse compared to the same model using the

4 of 9

E00B07

MARCQ ET AL.: VIRTIS-H STUDY OF TRACE GASES VARIATIONS

E00B07

Figure 3. Comparison in the 2.2– 2.5 mm interval of a good quality order 6 spectrum (from orbit 277, in black) with the synthetic model. The relatively poor agreement at wavelengths shortward of 2.3 mm is due to the poor knowledge of CO2 opacity. older HITEMP database, whether a continuum absorption for CO2 was considered or not. So we discarded the CDSD-750 database, reverted to the HITEMP database and kept the continuum absorption of CO2 to the usual constant value of 3.5  108 cm1.amagat2 throughout the window [Pollack et al., 1993; Marcq et al., 2006], owing to the lack of experimental measurements at the relevant pressures and temperatures. [21] An update of the temperature profile that we consider in all of our simulations has also been tried from the older mean VIRA profile [Seiff, 1983] to the newer VIRA-2 [Moroz and Zasova, 1997]. A comparison between the two profiles in the altitude range where the emission originates from (30 to 40 km) showed very little difference, especially for the vertical gradient dT/dz. This latter could have been a concern since altering dT/dz distorts the simulated spectra in almost the same way as a change in CO abundance; it affects the relative brightness temperature of neighboring atmospheric layers where CO mixing ratio differs. On the contrary, changing the mean temperature of the 30– 40 km region while maintaining the temperature gradient produces almost no shape distortion of the spectra, but merely alters the integrated flux in the whole window, as expected. As far as we are concerned, the differences were too weak to justify a change of our temperature profile. [22] Still, the assumption of a horizontally uniform temperature profile could be invalid in the polar regions (above 60° in latitude), where some variability of T(z) is expected even in the lower atmosphere [Pa¨tzold et al., 2007]. But these regions yield a too weak signal in VIRTIS-H to be studied, so that we maintained this assumption. It is also worth noting that the weaker fluxes of the polar region are caused by a thicker lower cloud cover, whose modal distribution of aerosols is poorly known. A proper modeling of nightside polar spectra would therefore require a good understanding of both temperature profile and vertical

and modal distribution of cloud particles, which are both difficult to assess yet. [23] Finally, the synthetic spectra were sampled and convolved according to the instrumental function to allow straightforward comparison between synthetic and observed spectra. The instrumental functions differ for orders 5 and 6, and their width and shape vary with wavelength in a complex way. 3.1.2. One-Parameter Retrieval [24] The algorithm is based on a simple calculation of the residuals between an observed spectrum and various sets of precomputed synthetic spectra of various prescribed atmospheric parameters (e.g., vertical profiles of minor constituents, cloud opacity) within a few specific spectral intervals. This algorithm is almost exactly similar to the one described in section 4.1.1. by Marcq et al. [2006], except for some minor adaptations due to the differences between VIRTIS-H and SpeX. [25] Actually, besides the different determination of the true zero level (see 2.3.2) and, more generally, the completely different preprocessing, especially regarding diffused sunlight removal for SpeX, the key differences come from the fact that VIRTIS-H nightside spectra are available on two dispersion orders (5 and 6; order 7 was only used for zero-level determination since it does not cover the minor species absorption bands). Furthermore, the totality of the 2.3-mm window is only available on order 6, as can be seen on Figure 2: the 2.2-2.3 mm interval is not accurately rendered in order 5. From these considerations, the following adaptations were required: [26] 1. The former version of the algorithm used the same spectral interval to compute the free-scaling parameter ai (designed to have the best possible agreement between a given synthetic spectrum Si and the observed spectrum E) and to compute the residuals between E and Si. Owing to the lack of coverage for order 5, the algorithm is now able to use spectral intervals that are distinct for these two uses, the

5 of 9

MARCQ ET AL.: VIRTIS-H STUDY OF TRACE GASES VARIATIONS

E00B07

E00B07

Figure 4. Spectral intervals used by our algorithm for the various species. The thick lines correspond to the residual computation intervals, whereas the whole lines (thin and thick) correspond to the scaling parameter determination intervals.

scaling interval typically being a superset of the residual estimation interval. The actual values, which can differ from order to order, are shown on Figure 4. [27] 2. Orders 5 and 6 of VIRTIS-H overlap between 2.3 and 2.5 mm. Although the analysis on both orders is done separately since they yield two distinct spectra of the same location, the estimates are considered as two independent measurements of the same spectrum: the resulting estimation e and its 2 uncertainty s are given by the following s6 e5 þs25 e6 5 s6 ffi , where the subscripts formulæ: e = s2 þs2 and s = psffiffiffiffiffiffiffiffiffi 2 2 6

5

s5 þs6

5 and 6 refer to the considered order. To ensure the consistency of such a combination, the retrievals were dismissed if their maximal normalized probability density, 2 5Þ ], was inferior to a threshold depending given by exp[ðes62e þs25 6 on the species, usually between 0.8 and 0.9. Note that coadding spectra from orders 5 and 6 would not make sense since their spectral resolution differ on their common wavelength range. 3.1.3. Strategy for Retrieving Gaseous Profiles [28] The global retrieval strategy is mostly unchanged since Marcq et al. [2006], where the 2.3-mm window was studied sequentially from the shorter wavelengths to the longer ones. The spectral intervals used for the retrievals of these three species are shown on Figure 4. Note that wavelengths less than 2.35 mm could not be used for order 5-based retrievals, since the instrument underestimates the intensity at these wavelengths for this order. 3.1.3.1. Carbon Monoxide Retrieval [29] The CO absorption band centered at 2.33 mm cannot be properly observed on order 5, so that only order 6 can be used for the retrievals. Spectral distortions due to the variation of the lower cloud opacity between the various spectra are taken into account as by Marcq et al. [2006] by first measuring the cloud opacity using integrated spectral radiance between 2.2 and 2.5 mm and then accordingly interpolating the synthetic spectra to the measured opacity. The considered set of CO vertical profiles are all proportional to the standard VIRA-2 profile. 3.1.3.2. Water Vapor Retrieval [30] Spectra from both orders 5 and 6 can be used for this species. Variations of lower cloud opacity, which affect spectral radiance in the whole window, and carbon monoxide, measured at the previous step and whose absorption band overlaps the water absorption band, are taken into account in the same way as previously stated. The set of synthetic profiles are here also proportional to the standard VIRA-2

profile, with no attempt to explore various D/H ratios at this stage. 3.1.3.3. Carbonyl Sulfide Retrieval [31] Spectra from orders 5 and 6 were also processed for the retrievals. Spectral overlapping with water vapor and carbon monoxide requires that the variations of these two species are taken into account, not to mention the consideration of cloud opacity variations here also. The set of considered synthetic profiles are various vertical translations of the standard VIRA-2 profile. 3.1.3.4. Sulfur Dioxide Retrieval [32] Its spectral signature being very narrow, sulfur dioxide retrievals are performed tentatively with a very low accuracy, using orders 5 and 6. Variations of carbonyl sulfide, water vapor and lower cloud opacity are taken into account, with the SO2 profiles assumed to be vertically uniform. 3.2. Results 3.2.1. CO [33] The mixing ratio of carbon monoxide at 36 km inferred at various locations is plotted on Figure 5. The minimal value of 24 ± 3 ppmv appears to be slightly off the equator, between 0° and 20°S. Its abundance then rises symmetrically with respect to this minimum toward higher latitudes, reaching 31 ± 2 ppmv at 60°S and between 30 and

Figure 5. Variations of the CO abundance at 36 km according to various VIRTIS-H spectra (orbits shown in various colors). Error bars stand for 1-s uncertainty levels.

6 of 9

E00B07

MARCQ ET AL.: VIRTIS-H STUDY OF TRACE GASES VARIATIONS

E00B07

3.2.3. H2O [35] As concerns water vapor, no similar latitudinal trend could be observed on our results shown on Figure 7. Nevertheless, the combination of order 5 and 6 retrievals has enabled us to enhance the accuracy on its mean value significantly: 31 ± 2 ppmv between 30 and 40 km. Some temporal and/or longitudinal variability may also be possible with slightly drier conditions (29 ± 1 ppm) past 20°N, but this departure with the mean value is not significant: we have no evidence for any variations within this data set. 3.2.4. SO2 [36] Sulfur dioxide measurements could not reach a high accuracy considering the narrowness of its spectral band on the long-wavelength edge of the 2.3-mm spectral window, so that no latitudinal variation could be detected (see Figure 8). The assumption that the mixing ratio is uniform yields a mean mixing ratio of 130 ± 50 ppmv. Figure 6. Variations of the OCS abundance at 33 km according to various VIRTIS-H spectra (orbits shown in various colors). Error bars stand for 1-s uncertainty levels. 40 ppm at 60°N, where the coverage is not dense enough to ensure as much accuracy as in its southern counterpart. The dispersion of the measurements suggests some longitudinal and/or temporal variability within less than the 300-day interval during which the processed spectra have been acquired, although never exceeding 20% of the zonal mean value nor the pole-to-equator gradient. 3.2.2. OCS [34] Carbonyl sulfide latitudinal variations are shown on Figure 6. These variations appear to be anticorrelated to those of carbon monoxide, showing a maximal value of 4 ± 1 ppmv between 20°S and the equator, which was also the latitude where CO was minimal; the asymmetry of CO and OCS variations with respect to the equator could be related to the presence of high-elevation landmasses (namely, Aphrodite Terra, which is centered around 10° S) through the perturbations they cause to the general circulation. The minimum value of 2.5 ± 1 ppmv is reached near 60°S. The relative uncertainty on these retrievals is quite high, higher than what the S/N value or the accuracy of the fitting could account for, and such that we cannot dismiss a latitudeindependent mixing ratio between 3 and 3.5 ppm. Our current interpretation of this poor precision is related to the choice of the 1-D vertical synthetic profile set mentioned in section 3.1.3, which implies a univocal relation between mean abundance and vertical gradient. This relation prevents the exploration of the whole 2-D parameter space to which spectra are sensitive (i.e., vertical gradient and, independently, mean abundance in the 30– 36 km range). The explored subset is probably out of the optimal fit region since the dispersion of measurements is consistently inferior to the 1-s error bars, which points out that statistical noise is not the only source of discrepancy between synthetic and observed spectra. The same concern was raised by Marcq et al. [2006], and was solved by enabling the vertical gradient and the abundance to be retrieved separately. Such a study requires very high S/N ratio, and is therefore still in progress for the best VIRTIS-H spectra currently available.

3.3. Comparison with other Spectroscopic Studies 3.3.1. Lower-Resolution Spectra [37] Observations at lower spectral resolution could constrain with a similar accuracy the species with the most prominent band in the whole 2.3-mm window, namely, CO. H2O and OCS retrievals are currently in progress at a lesser precision compared to VIRTIS-H retrievals. Nevertheless, these studies require less signal to work on, so their horizontal coverage is substantially larger. The observations of Collard et al. [1993] using NIMS/Galileo were the first of this kind, and pointed to a northern enrichment in CO of 35 ± 15% from the equator to 47°N, in a good agreement with ours (30 ± 15% in the same latitudinal range). More recently, Tsang et al. [2008] has used VIRTIS-M spectra to map the CO abundance over large portions of the night hemisphere, and their conclusions about the latitudinal trends for CO corroborate ours in both hemispheres: from 23 ± 2 ppm near the equator to 32 ± 2 ppm at 35 km around 60° –70°. 3.3.2. Similar Resolution or Higher-Resolution Spectra [38] Spectra at a resolution similar or higher than ours in the 2.3-mm spectral window have been acquired and pro-

Figure 7. Variations of the water vapor at 33 km according to various VIRTIS-H spectra (orbits shown in various colors). Error bars stand for 1-s uncertainty levels.

7 of 9

E00B07

MARCQ ET AL.: VIRTIS-H STUDY OF TRACE GASES VARIATIONS

Figure 8. Variations of the sulfur dioxide at 35 km according to various VIRTIS-H spectra (orbits shown in various colors). Error bars stand for 1-s uncertainty levels. cessed before, but focused first on obtaining horizontally averaged spectra rather than studying their spatial variations. Comparisons are nevertheless useful, and agree with our results. The mean values of the CO mixing ratio at 36 km given by Pollack et al. [1993], Be´zard et al. [1990] and Taylor et al. [1997] range between 20 and 30 ppm, consistent with our own measurements. Using the same spectra, OCS and SO2 were also constrained, and respectively found to be 4.4 ± 1 ppm at 33 km (OCS [Pollack et al., 1993]) and 130 ± 40 ppm between 30 and 40 km (SO2 [Be´zard et al., 1993]). These determinations are also in full agreement with our own averages. The mean H2O mixing ratio between 30 and 40 km was found using high-resolution (R  20000) CFHT spectra in the 1.74 and 2.3-mm windows by de Bergh et al. [1995] to be 30 ± 10 ppm, which is also fully compatible with ours. In contrast, the lower estimate of 26 ± 4 ppm found by Marcq et al. [2006] is possibly due to contamination by terrestrial water vapor: the mean abundance is probably closer to 30 ppmv than to 26. [39] More recently, latitudinal studies in this spectral window were conducted using SpeX/IRTF spectra [Marcq et al., 2005, 2006] in order to prepare the processing of VIRTIS-H spectra. These results also agree with ours, at least in the latitude interval that can be observed from Earth: CO was found to vary between 26.5 ± 0.5 ppm (36 km) near 40°S and 23.75 ± 0.75 ppm near the equator, which matches our results. OCS variations were similarly anticorrelated with CO, although quantitative comparisons are harder considering the large vertical gradient of OCS in this region since their results are given at a different altitude of reference (36 km instead of 33). 3.4. Dynamical Interpretations 3.4.1. Qualitative Interpretation [40] The current interpretation of the anticorrelated variations of CO and OCS relies upon our knowledge of the global-scale circulation, especially vertical motions. It is known [Pollack et al., 1993; Marcq et al., 2006] that in the 30– 40 km altitude range, CO mixing ratio increases with height whereas OCS decreases. These observed vertical gradients are also predicted by most chemical models

E00B07

[Krasnopolsky, 2007; Krasnopolsky and Pollack, 1994]: CO is produced by photo-dissociation of CO2 by UV radiation above the clouds and consumed in the lower atmosphere as a reducing agent, whereas OCS is produced at ground level by reactions involving pyrite, CO and CO2 whereas it is strongly oxidized by SO3 (coming from the evaporation of H2SO4 at the lower cloud floor). The Hadley-cell vertical circulation of Venus causes upward motion in the equatorial region (due to solar heating) and downward motion at higher latitudes; most global circulation models (GCMs) predict the peak of downward velocity around 60° (S. Lebonnois et al., Superrotation of Venus’ atmosphere analysed with a full General Circulation Model, manuscript in preparation, 2008). Therefore, there is a tendency to deplete the CO mixing ratio around the equator (and increase that of OCS, which is more abundant at lower levels) whereas the trends are opposite at higher latitudes, which is precisely the observed pattern for CO and OCS. We can also note that the other measured species (H2O and SO2), which do not exhibit latitudinal variations, have no detectable vertical gradient either. 3.4.2. Estimate of Typical Relaxation Times [41] If the explanation given above is true, we can estimate the order of magnitude of the chemical relaxation time toward thermochemical equilibrium profiles for CO and OCS, assuming that the latitudinal contrasts we observe are caused only by the large-scale vertical motions. This coarse approach neglects other phenomena such as diffusion, and is very simplistic as for the chemical modeling, so it cannot yield more than order-of-magnitude estimates of the chemical time scale. [42] The advected net flux of a tracer whose mixing ratio ~ (q~ is q is given in a 1-D vertical model by r v) = w dq dz  dw v stands for the wind speed vector, w for the dz q, where ~ vertical wind velocity and z for the altitude. Neglecting the variation of w between 30 and 40 km, this flux is in balance 0 ð zÞ where t with the chemical relaxation @q@tðzÞjchem = qðzÞq t stands for the characteristic time and q0(z) the expected vertical profile in the absence of any dynamical perturbation. In a stationary regime, this yields t¼

q0  q w dq dz

q  q0 is typically around a few ppm, jdq/dzj about 1 ppm/ km for CO and OCS [Pollack et al., 1993; Taylor et al., 1997; Marcq et al., 2006] and GCMs predict typical velocities below 1 mm/s. This gives for both species characteristic times t  106 – 107, s in order to allow the formation of the equator-to-pole gradients in CO and OCS seen here. We would like to stress that this value is not related to a definite chemical process, but should merely be seen as a threshold: a faster chemistry (with a smaller t) would not allow enough time for the vertical circulation to establish the latitudinal contrasts described in section 3.2. Other characteristic times could be expressed regarding other processes (e.g., horizontal circulation, eddy diffusion), but a more detailed description is beyond the scope of this paper. 3.4.3. Quantitative Studies in Progress [43] To progress further in the interpretation of CO and OCS variations, a number of modeling efforts are currently

8 of 9

E00B07

MARCQ ET AL.: VIRTIS-H STUDY OF TRACE GASES VARIATIONS

in progress. The one from Yung et al. (submitted manuscript, 2008) focuses on OCS and adopts an accurate chemical scheme embedded in a 2-D circulation model, whereas Marcq et al. (manuscript in preparation, 2008) models both CO and OCS-type tracers with a simple relaxation scheme within a full 3-D radiatively consistent GCM. The results of these two models can be found in the above-quoted papers.

4. Conclusion [44] First of all, these results stress the complementary role of VIRTIS-M and VIRTIS-H channels of Venus Express, the precision of abundance retrievals and the number of species being the main assets of VIRTIS-H in this field whereas the CO mapping capabilities of VIRTIS-M over large extents on the Venus nightside are not accessible here. We have also significantly improved the accuracy of water vapor retrievals thanks to the availability of two spectral orders over the 2.3-mm spectral window. Even using a single order, CO variations were confirmed and validated up to 60° in latitude for both hemispheres, still fully compatible with previous estimates by Collard et al. [1993] and Marcq et al. [2005] (30 ± 15%). Anticorrelated OCS variations first noticed from Earth were also tentatively confirmed and extended in the same latitudinal range, though the relatively poor accuracy of the retrievals cannot rule out a constant value; the accuracy of the OCS retrievals should greatly benefit from a more detailed study with OCS abundance and vertical gradient being derived separately as it was the case for Marcq et al. [2006]. [45] With the oncoming extended phase of Venus Express, the prospective of future observations are very good. Exploration of high-latitude region (beyond 60°) could be possible thanks to the implementation of new science cases (enabling the tracking of a specific location during longer times), although such a study would probably require new additional information (better knowledge of possible horizontal variations of the T(p) profile, modal and vertical distribution of aerosol particles). Supportive studies from space and/or Earth in the other spectral windows could also help in constraining the vertical profiles of several minor species at different altitudes, especially H2O. [46] The first results of coupled dynamical-chemical simulations are also expected to improve significantly our understanding of the chemistry below the clouds of Venus, especially regarding the relative time scales of various dynamical and chemical phenomena. Observational constraints provided by the present paper are also of paramount interest for developing and improving such complex coupled simulations. [47] Acknowledgments. This work has been supported and funded by the Centre National d’E´tudes Spatiales (CNES).

References Allen, D. A., and J. W. Crawford (1984), Cloud structure on the dark side of Venus, Nature, 307, 222 – 224. Bell, J. F., P. G. Lucey, T. A. Ozoroski, W. M. Sinton, and D. Crisp (1991), Spectroscopic observations of bright and dark emission features on the night side of Venus, Science, 252, 1293 – 1296. Be´zard, B., C. de Bergh, D. Crisp, and J.-P. Maillard (1990), The deep atmosphere of Venus revealed by high-resolution nightside spectra, Nature, 345, 508 – 511, doi:10.1038/345508a0.

E00B07

Be´zard, B., C. de Bergh, B. Fegley, J.-P. Maillard, D. Crisp, T. Owen, J. B. Pollack, and D. Grinspoon (1993), The abundance of sulfur dioxide below the clouds of Venus, Geophys. Res. Lett., 20, 1587 – 1590. Carlson, R. W., et al. (1991), Galileo infrared imaging spectroscopy measurements at Venus, Science, 253, 1541 – 1548. Collard, A. D., et al. (1993), Latitudinal distribution of carbon monoxide in the deep atmosphere of Venus, Planet. Space Sci., 41, 487 – 494, doi:10.1016/0032-0633(93)90033-X. Coradini, A., et al. (1998), VIRTIS: An imaging spectrometer for the ROSETTA mission, Planet. Space Sci., 46, 1291 – 1304. Crisp, D., W. M. Sinton, K.-W. Hodapp, B. Ragent, F. Gerbault, and J. H. Goebel (1989), The nature of the near-infrared features on the Venus night side, Science, 246, 506 – 509. de Bergh, C., B. Be´zard, D. Crisp, J. P. Maillard, T. Owen, J. Pollack, and D. Grinspoon (1995), Water in the deep atmosphere of Venus from highresolution spectra of the night side, Adv. Space Res., 15, 79 – 88. Donahue, T. M. (1999), New analysis of hydrogen and deuterium escape from Venus, Icarus, 141, 226 – 235, doi:10.1006/icar.1999.6186. Donahue, T. M., D. H. Grinspoon, R. E. Hartle, and R. R. Hodges (1997), Ion/neutral escape of hydrogen and deuterium: Evolution of water, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher, D. M. Hunten, and R. J. Philips, pp. 385 – 414, Univ. of Ariz. Press, Tucson. Drossart, P., et al. (1993), Search for spatial variations of the H2O abundance in the lower atmosphere of Venus from NIMS-Galileo, Planet. Space Sci., 41, 495 – 504, doi: 10.1016/0032-0633(93)90032-W. Grinspoon, D. H. (1993), Implications of the high D/H ratio for the sources of water in Venus’ atmosphere, Nature, 363, 428 – 431, doi:10.1038/ 363428a0. Kamp, L. W., and F. W. Taylor (1990), Radiative-transfer models of the night side of Venus, Icarus, 86, 510 – 529, doi: 10.1016/0019-1035(90)90231-W. Kamp, L. W., F. W. Taylor, and S. B. Calcutt (1988), Structure of Venus’s atmosphere from modelling of night-side infrared spectra, Nature, 336, 360 – 362, doi:10.1038/336360a0. Krasnopolsky, V. A. (2007), Chemical kinetic model for the lower atmosphere of Venus, Icarus, 191, 25 – 37, doi:10.1016/j.icarus.2007.04.028. Krasnopolsky, V. A., and J. B. Pollack (1994), H2O-H2SO4 system in Venus’ clouds and OCS, CO, and H2SO4 profiles in Venus’ troposphere, Icarus, 109, 58 – 78, doi:10.1006/icar.1994.1077. Marcq, E., B. Be´zard, T. Encrenaz, and M. Birlan (2005), Latitudinal variations of CO and OCS in the lower atmosphere of Venus from near-infrared nightside spectro-imaging, Icarus, 179, 375 – 386, doi: 10.1016/ j.icarus.2005.06.018. Marcq, E., T. Encrenaz, B. Be´zard, and M. Birlan (2006), Remote sensing of Venus’ lower atmosphere from ground-based IR spectroscopy: Latitudinal and vertical distribution of minor species, Planet. Space Sci., 54, 1360 – 1370, doi:10.1016/j.pss.2006.04.024. Moroz, V. I., and L. V. Zasova (1997), VIRA-2: A review of inputs for updating the Venus International Reference Atmosphere, Adv. Space Res., 19, 1191 – 1201. Pa¨tzold, M., et al. (2007), The structure of Venus’ middle atmosphere and ionosphere, Nature, 450, 657 – 660, doi:10.1038/nature06239. Pollack, J. B., et al. (1993), Near-infrared light from Venus’ nightside-A spectroscopic analysis, Icarus, 103, 1 – 42, doi:10.1006/icar.1993.1055. Seiff, A. (1983), Thermal structure of the atmosphere of Venus, in Venus, edited by S. W. Bougher, D. M. Hunten, and R. S. Phillips, pp. 215 – 279, Univ. of Ariz. Press, Tucson. Taylor, F. W., D. Crisp, and B. Be´zard (1997), Near-infrared sounding of the lower atmosphere of Venus, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher, D. M. Hunten, and R. J. Philips, pp. 325 – 351, Univ. of Ariz. Press, Tucson. Titov, D. V., et al. (2006a), Venus Express: Scientific goals, instrumentation, and scenario of the mission, Cosmic Res., 44, 334 – 348, doi:10.1134/ S0010952506040071. Titov, D. V., et al. (2006b), Venus Express science planning, Planet. Space Sci., 54, 1279 – 1297, doi:10.1016/j.pss.2006.04.017. Tsang, C. C. C., et al. (2008), Tropospheric carbon monoxide concentrations and variability on Venus from Venus Express/VIRTIS-M observations, J. Geophys. Res., doi:10.1029/2008JE003089, in press. 

B. Be´zard, P. Drossart, F. Henry, E. Marcq, and J. M. Reess, Laboratoire d’E´tudes Spatiales et d’Instrumentation en Astrophysique, UMR8109, CNRS, Observatoire de Paris, 5 place Juleps Janssen, F-92195 Meudon, France. ([email protected]) G. Piccioni, Istituto di Astrofisica Spaziale e Fisica Cosmica, Istituto Nazionale di Astrofisica, Via Fosso del Cavaliere 100, I-00133 Roma, Italy.

9 of 9

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, E00B08, doi:10.1029/2008JE003089, 2008

for

Full Article

Tropospheric carbon monoxide concentrations and variability on Venus from Venus Express/VIRTIS-M observations Constantine C. C. Tsang,1 Patrick G. J. Irwin,1 Colin F. Wilson,1 Fredric W. Taylor,1 Chris Lee,2 Remco de Kok,1 Pierre Drossart,3 Giuseppe Piccioni,4 Bruno Bezard,3 and Simon Calcutt1 Received 26 January 2008; revised 10 June 2008; accepted 25 June 2008; published 1 October 2008.

[1] We present nightside observations of tropospheric carbon monoxide in

the southern hemisphere near the 35 km height level, the first from Venus Express/Visible and Infrared Thermal Imaging Spectrometer (VIRTIS)-M-IR. VIRTIS-M data from 2.18 to 2.50 mm, with a spectral resolution of 10 nm, were used in the analysis. Spectra were binned, with widths ranging from 5 to 30 spatial pixels, to increase the signal-to-noise ratio, while at the same time reducing the total number of retrievals required for complete spatial coverage. We calculate the mean abundance for carbon monoxide at the equator to be 23 ± 2 ppm. The CO concentration increases toward the poles, peaking at a latitude of approximately 60°S, with a mean value of 32 ± 2 ppm. This 40% equator-to-pole increase is consistent with the values found by Collard et al. (1993) from Galileo/NIMS observations. Observations suggest an overturning in this CO gradient past 60°S, declining to abundances seen in the midlatitudes. Zonal variability in this peak value has also been measured, varying on the order of 10% (
3 ppm) at different longitudes on a latitude circle. The zonal variability of the CO abundance has possible implications for the lifetime of CO and its dynamics in the troposphere. This work has definitively established a distribution of tropospheric CO, which is consistent with a Hadley cell circulation, and placed limits on the latitudinal extent of the cell. Citation: Tsang, C. C. C., P. G. J. Irwin, C. F. Wilson, F. W. Taylor, C. Lee, R. de Kok, P. Drossart, G. Piccioni, B. Bezard, and S. Calcutt (2008), Tropospheric carbon monoxide concentrations and variability on Venus from Venus Express/VIRTIS-M observations, J. Geophys. Res., 113, E00B08, doi:10.1029/2008JE003089.

1. Introduction 1.1. Past Observations of Tropospheric CO [2] The near-infrared thermal emissions were discovered by Allen and Crawford [1984] and were shown to be emanating from the deep troposphere of Venus by radiative transfer modeling by Kamp et al. [1988] and Bezard et al. [1990]. These emissions, spanning spectral wavelengths from 1.01 to 2.50 mm, are attenuated as they pass through the cloud layer, so the radiance observed from space is sensitive to changes in the total cloud optical depths. It is also sensitive to abundances of minor species, namely CO, H2O, HDO, OCS, SO2, HF, and HCl, at altitudes ranging from the surface to the base of the cloud layer at 45 km. A compre-

1 Atmospheric, Oceanic and Planetary Physics, Clarendon Laboratory, Department of Physics, University of Oxford, Oxford, UK. 2 Geophysics and Planetary Sciences, California Institute of Technology, Pasadena, California, USA. 3 LESIA, Observatoire de Paris, Meudon, France. 4 INAF-IASF, Rome, Italy.

hensive review of the near-infrared windows can be found in the work of Taylor et al. [1997] and Baines et al. [2006]. [3] While Kamp et al. [1988] were the first to propose retrieving abundances of these minor species, it was the subsequent observations by Bezard et al. [1990] and Pollack et al. [1993] using the Canada France Hawaii Telescope, who first measured the CO abundance in the troposphere, estimating mean values at 36 km of 40 ppm and 23 ± 5 ppm, respectively. Pollack et al. [1993] also retrieved a CO vertical profile, with increases of 1.20 ± 0.45 ppm km1 increasing with height. At approximately the same time, the Galileo spacecraft flew past Venus on its trajectory to Jupiter. Onboard, the NIMS imaging spectrometer took the first images from space of the nightside of Venus, most notably at the wavelengths of 1.74 and 2.30 mm. [4] Collard et al. [1993] measured the latitude distribution of CO from the Galileo/NIMS VJBAR data set. Rather than using a spectral fitting technique, Collard et al. [1993] positively detected a latitude CO gradient using a method of ratios of spectra at two different wavelengths. It was shown that a CO meridional gradient exists in the lower atmosphere, with an enhancement mainly poleward of 47°N. This work

Copyright 2008 by the American Geophysical Union. 0148-0227/08/2008JE003089$09.00

E00B08

1 of 13

E00B08

TSANG ET AL.: TROPOSPHERIC CARBON MONOXIDE ON VENUS

remained unverified until only relatively recently, when Marcq et al. [2005, 2006] using the NASA/IRTF/SpeX, observed the same CO gradient from ground-based highresolution nightside spectra. From this analysis, a lower limit for the equator-to-pole CO gradient was found to increase by approximately 15% from the equator to 40°N/S, less than the 35% increase found by Collard et al. [1993]. While this is a lower gradient, it does not cover the high latitudes covered by Galileo/NIMS. In addition, neither Collard et al. [1993] nor Marcq et al. [2006] were able to measure any zonal distribution or variability of CO because of the limited horizontal and temporal sampling of their observations. 1.2. CO as a Dynamical Tracer [5] Gierasch [1975] postulated that a mean meridional circulation would exist in the Venus atmosphere, transporting momentum from the equator to the poles. This was verified by the in situ measurements of meridional winds by the Pioneer large and small probes [Counselman et al., 1980]. Images of Venus’ dayside in the UV reveal cloud features at 65–75 km altitude; tracking these features with potential vorticity data revealed meridional wind speeds of >10 m s1 [Limaye, 2007] at this altitude. Nightside images of Venus at 1.7 and 2.3 mm exhibit contrast because of variability in the lower cloud layer at 50 km. Analysis of Galileo/NIMS [Carlson et al., 1991] found that the motion of the features indicated that meridional flow is still poleward at this altitude. This would indicate the Hadley cell extended to a depth of at least this altitude. It is this Hadley cell circulation which maintains the midlatitude jets [Lee et al., 2007], which in turn support the super-rotation, with high zonal wind speeds. In addition, because the Hadley cell is transporting angular momentum toward the poles, it may also be responsible for maintaining the polar vortex seen from a latitude of 70° poleward. There are strong indications, from both observations and general circulation models (GCM), that this equator-to-pole circulation exists in both hemispheres. However, both the vertical and latitudinal extent of the Hadley cells are yet to be fully understood. Outstanding questions include how far in latitude do the Hadley cells extend, how deep in altitude they go, are the southern and northern hemisphere cells symmetric about the equator, and are there any timedependent horizontal variations of these cells. [6] The tropospheric CO enhancement was interpreted by Taylor [1995] to be caused by the descending arm of the Hadley cell, bringing down CO from above the cloud tops where the photolysis of CO2 by UV, with an energetic threshold for photo-dissociation near 224 nm [Von Zahn et al., 1983], creates CO (noting that the CO2 cross section is dominated by Rayleigh scattering longward of 205 nm [Shemansky, 1972; Karaiskou et al., 2004]). As CO descends, it is believed to be converted to OCS and CO2 [Krasnopolsky, 2007; Fegley et al., 1997; Hong and Fegley, 1997; Y. Yung et al., Modeling the distribution of OCS in the lower atmosphere of Venus, submitted to Journal of Geophysical Research, 2008]. It is therefore possible to use CO as a dynamical tracer, which is entrained by the circulation of the lower atmosphere. This would not only provide mean abundances of CO in the lower atmosphere,

E00B08

but also provide information on the size and variability of the Hadley cell in the deep atmosphere.

2. Venus Express/VIRTIS-M Observations [7] Venus Express entered Venus orbit in April 2006. Onboard are seven science instruments devoted to observing the atmosphere and surface of Venus from UV to infrared and radio wavelengths. Visible and Infrared Thermal Imaging Spectrometer (VIRTIS), which is also flying on the Rosetta mission [Coradini et al., 1998], is one of these instruments. Covering wavelengths from 0.27 to 5.19 mm, VIRTIS is divided into two subsystems; the nonimaging H channel covering the spectral ranges from 1.84 to 4.99 mm, with a spectral resolution of 1nm, and the imaging M channel, which is further divided into two channels. One channel, viewing in the UV and visible wavelengths, is named M-Vis, while the 1.05 to 5.19 mm spectral region is sounded by the M-IR channel. The spectral resolution of VIRTIS-M is approximately 2 nm and 10 nm for the VIS and IR sub-channels, respectively. In this analysis, we will be using data from the VIRTIS-M-IR subsystem only. A comprehensive review of VIRTIS is given by Drossart et al. [2007] and Piccioni et al. [2007]. [8] While the ability of VIRTIS-H to obtain higher resolution spectra has obvious advantages, such as better retrieval of the absolute abundances of minor species as well as their vertical distribution, the multispectral imaging capabilities of VIRTIS-M makes it uniquely able to map global-scale properties such as cloud opacity and wind field measurements. It is this capability not only to measure the global-scale values of minor species, but also to do so as a function of time, which we intend to take advantage of. Indeed, this has always been one of the key measurements to be made with VIRTIS [Baines et al., 2006]. [9] The single linear array of VIRTIS-M has an instantaneous field of view of 0.25  64 mrad, divided into 256 spatial pixels. At an apogee of 66,000 km, this instantaneous field of view spans one third of the diameter of Venus. Using a scan mirror, with 256 step positions, the field of view expands to 64  64 mrad. Therefore, at apogee, a full image of the Venus disc is created when nine images are taken sequentially to create a 3  3 mosaic image. [10] Table 1 shows the list of observations used in this analysis; two orbit insertion observations, CIOB00 taken on 12 April 2006, and CIOB03 on 16 April 2006, and two nominal orbit observations in orbit number 99 and 121. The orbit insertion observations were used because the whole disc of Venus was captured in the field of view, as the initial orbit of Venus Express in April 2006 was extremely large. The distance from Venus at the time of these two observations was approximately 214,000 km and 315,000 km, respectively. Nominal orbit observations were MTP03-99-2, taken on 28 July 2006 from a distance of 61,000 km, and MTP04121-0 taken on 12 April 2006 at a distance of 65,000 km. [11] The raw observations were spectrally and spatially binned to increase the signal-to-noise ratio of the data, as well as to reduce the number of spectra needed to achieve complete spatial coverage over the observation image. As the spatial scale of each observation varied, a bin of various widths was used for different observations. A typical bin width of 20  20 pixels would be used, which is then shifted

2 of 13

TSANG ET AL.: TROPOSPHERIC CARBON MONOXIDE ON VENUS

E00B08

E00B08

Table 1. Summary of the Observations Used in This Analysis and Their Important Parametersa Observation Name

Integration Time (s)

Number of Spectra

Longitude Ranges (°)

Latitude Ranges (°)

Subsolar Longitude (°)

CIOB00 – 0 CIOB03 – 7 MTP03-99-2 MTP04-121-0

8.0 8.0 8.0 3.3

300 375 300 289

0.0: 359.9 0.0: 359.9 196.71: 286.17 0.00: 360.0

0.77: 90 13.16: 89.65 22.6: 76.07 13.13: 82.69

70.74 83.69 39.71 107.69

a

To compare the local time aspect of each observation, subtracting the longitudes by the subsolar longitude will yield the local time in units of longitude.

by 10 pixels across the image for adequate Nyquist sampling. An average of some 300 spectra per observation would be generated in this manner. The associated latitudes, longitudes, and geometries is also averaged in this way. The result can be seen in an example (Figure 1) from observation CIOB00.

3. Radiative Transfer Modeling 3.1. Spectral Data [12] We employ a radiative transfer model to generate synthetic spectra, initially developed for the analysis of Galileo/NIMS data [Irwin et al., 2004], which uses the method of correlated-k [Lacis and Oinas, 1991] to calculate

the absorption coefficients of the gases in the atmosphere. The k tables are precalculated from line-by-line spectral databases. The tables are calculated on a grid of 20 evenly spaced log-pressure levels (e17 to e5 bars) and 20 evenly spaced temperature levels (150 K to 760 K). These values correspond to approximately the 120 km and surface height range. The spectral resolutions are calculated at 10 nm and 1 nm which correspond to the resolutions of the VIRTIS-M and H subsystems. For gaseous opacity, the HITEMP database is used for CO2, while the remaining trace gases use HITRAN2K. A CO2 sub-Lorenztian lineshape from Tonkov et al. [1996] was implemented, with a cutoff of 150 cm1 from the line center. CO2-CO2 self-broadening

Figure 1. A typical binning scheme used to reduce the data to manageable sizes, as well as increasing the signal-to-noise ratio. This example is taken from orbit insertion observation CIOB00. Only nightside data can be used for this analysis, hence we only use data from half of the hemisphere. 3 of 13

E00B08

TSANG ET AL.: TROPOSPHERIC CARBON MONOXIDE ON VENUS

E00B08

Figure 2. The Venus a priori temperature profile used in the retrieval processes taken from Seiff [1983]. half-widths were taken from the HITEMP database, while the CO2-H2O foreign half-widths were taken from Delaye et al. [1989]. The other remaining gases use the air-broadened half-widths given in HITRAN2K. CO2 collision induced

absorption is parameterized according to Bezard et al. [1990], with a value of 3.5  108 cm1 amagat2 used as the opacity factor.

Figure 3. A priori concentrations of minor constituents as a function of height. 4 of 13

E00B08

TSANG ET AL.: TROPOSPHERIC CARBON MONOXIDE ON VENUS

3.2. Atmospheric Profiles [13] The temperature profile used as the a priori was taken from Seiff [1983] and is plotted in Figure 2. The a priori vertical distribution of gases is given in Figure 3. The amount of CO2 is 0.965 fractional abundance, which is assumed to be vertically uniform. The abundance of CO and HF has been taken from Collard [1993], where HF is uniform, and kept uniform during the retrieval process, at 0.005 ppm. The remaining vertical profiles have been taken from Marcq et al. [2005], who have taken their profiles from a mixture of Oyama et al. [1980], Von Zahn et al. [1983], and Taylor et al. [1997]. 3.3. Cloud Profile [14] We use an a priori cloud model which follows the vertical profiles of Pollack et al. [1993]. The clouds in the model extend from 50 km to some 80 km. There are four modes in the model, which are distinguished by their size parameter. The refractive indices used in the model have been taken from Palmer and Williams [1975], for a mixture of 75% sulphuric acid and 25% water vapor. This is in line with other models of the near-infrared windows such as Pollack et al. [1993] and Marcq et al. [2006]. These refractive indices are used to calculate the single scattering albedo and absorption cross sections as look-up tables assuming Mie theory. Limited tests have been conducted for different mixtures of sulphuric acid/water vapor and their impact on the 2.3 mm spectrum. Tests show, for concentrations of 75%, 85%, and 96% H2SO4, the change to the ratio value at 2.30/2.33 mm (radiance outside the CO absorption band to radiance inside the band) is on the order of much less than 1%. This shows that we are really not that sensitive to potential changes in concentrations of H2SO4/ H2O in the Venus atmosphere. [15] To account for the multiple-scattering atmosphere and aerosols, as well as to accurately model high emissionangle observations, we use a matrix operator method to calculate for multiple scattering of the radiation [Plass et al., 1973]. We calculate this using ten streams (five up, five down).

4. Analysis [16] The retrieval model used in this work, named Nonlinear Optimal Estimator for Multivariate Spectral Analysis (NEMESIS), follows the formulation of Rodgers [1976]. The retrieval scheme starts with the prescribed forward model synthetic spectrum, generated with an a priori atmosphere, described above. The convergence limit f (denoted flimit) is also required by the user to start the iterative process. This produces a predicted spectrum yn with an initial cost function fi, given by T 1 f ¼ ðym  yn ÞT S1 m ðym  yn Þ þ ðx  xa Þ Sx ðx  xa Þ

ð1Þ

where ym is the measured radiance, yn is the predicted radiance, Sm is the measurement covariance matrix, x is the state vector, Sx is the state vector covariance matrix, and xa is the a priori state vector. The measurement covariance matrix Sm contains both the measurement errors and the forward modeling errors. The forward modeling error is

E00B08

included to compensate for any lack of physics within the forward model, as well as any inaccuracies incurred when using the correlated-k approximation. The calculation and estimates of the forward modeling and a priori errors can be found in the work of Irwin et al. [2008a] and Tsang et al. [2008], respectively, and are negligible overall. [17] The first term on the right-hand side of the cost function is known as c2 and is a measure of how well the synthetic spectrum is fitting to the observed spectrum. It is f which is minimized to arrive at a possible solution. Ultimately, the optimized solution to the nonlinear problem can be found for the unknown state vector by solving xiþ1 ¼ xa þ Sx KTi Ki Sx KTi þ Sm


1

ðym  yi  Ki ðxa  xi ÞÞ ð2Þ

where Ki is the K matrix associated with the ith iteration state vector xi. A modification is added to this GaussNewton iteration with the addition of a ‘‘braking parameter,’’ l, to keep the iteration stable for cases where the problem is nonlinear and is given by a LevenbergMarquardt constraint of the form x0iþ1 ¼ xi þ

ðxiþ1  xi Þ 1þl

ð3Þ

where x0i+1 is the modified state vector. If x0i+1 makes the fit worse, l is increased and x0i+1 is calculated again. If x0i+1 239 yields a better fit, l is decreased and the iteration continues. [18] The iterative process then proceeds to take a predefined a priori covariance error Sa, the measured spectrum ym, and the accompanied measurement covariance (error) Sm. This produces an estimate of the state vector x, with which a new spectrum yn is calculated using the radiative transfer equations. This yn has an accompanied fn. If this fn is greater than the previous f(n1), the braking parameter (equation (3)) is increased by tenfold and the iteration begins again. If fn is less than the previous f value, braking parameter l is decreased to 0.3l and the next iteration begins. This iteration scheme terminates when fn1  fn < flim it fn

ð4Þ

where flimit is a percentage change. This yields a possible state of the atmosphere xi, given the initial measured spectrum ym. In this work, the retrieval model terminates when the number of runs reaches a maximum of 60 iterations, or if the change in cost function f between iterations drops below 0.1%, whichever is reached first (these values in practice change depending on the level of fit specified and time allotted). In practice, a solution is reached usually after approximately 10– 20 iterations, when the cost function drops below the allotted 0.1% minimum. A full review of the radiative transfer and retrieval model used in this work can be found in the work of Irwin et al. [2008a].

5 of 13

E00B08

TSANG ET AL.: TROPOSPHERIC CARBON MONOXIDE ON VENUS

E00B08

Figure 4. (left) The contribution functions at 2.35 mm for the background temperature, CO at nadir, and CO at an emission angle of 80°. (right) The absorbing line strengths of CO in this spectral region. [19] Figure 4, left, shows the nadir contribution functions for 2.35 mm, produced by the radiative transfer model, which shows the background emissions coming from 30 km, with a sensitivity to CO at 35 km. Between 2.30 and 2.5 mm, the rotational-vibrational lines of the 2-0 band of carbon monoxide are active [Bezard et al., 1990], the spectral lines of which are also shown in Figure 4, right. It is these lines which provide us with the ability to retrieve the concentration of CO in the 35 km height range. [20] We take the entire spectral range, from 2.20 to 2.50 mm, to retrieve the tropospheric carbon monoxide concentrations. A sample spectrum can be seen in Figure 5. From wavelengths of 2.18 to 2.30 mm, the absorption is dominated by the presence of CO2. The radiation is also attenuated by the intervening cloud layer. This attenuation has a predominantly gray effect over this entire spectral window [Bezard et al., 1990; Marcq et al., 2006; Pollack et al., 1993]. Thus, the spectra covering these wavelengths are used to fit to the varying optical depths of the clouds across the planet. Once this is done, the rest of the spectra, from 2.30 to 2.50 mm, are assumed to have the same absorption as that caused by the clouds from 2.18 to 2.30 mm. [21] The remaining absorption in this region is due to the presence of trace gases which are active absorbers in this spectral region. The sensitivity of this spectrum in wavelength and height of trace gases can be found in the work of

Tsang et al. [2008] and Baines et al. [2006]. While we fit the 2-0 band of CO from 2.30 to 2.35 mm, we also simultaneously fit the rest of the spectrum to retrieve abundances for H2O and OCS, the results of which will not be shown in this work as it has no impact on the retrieved CO abundances. An example of our ability to fit our synthetic spectra to the data is shown in Figure 5. The residual, a simple difference between the measured and the synthetic spectrum, is also shown in Figure 5. Once f has been minimized, a best estimate for the CO abundance is found. In the case of CO, we retrieve a scaling factor which is multiplied by the a priori profile to yield a new abundance. Since the peak of the contribution function for CO is at approximately 35 km, we will show the abundances for that altitude.

5. Results [22] Figure 6 shows the observation and the results from the first orbit insertion observation, CIOB00. The 2.3 mm radiances resulting from the variations in the optical depths of the clouds that are attenuating the thermal radiation from above 35 km are shown at the spatial resolution of the binned spectra. After spectral fitting using the method described above, we can also display the c2/n values for different areas of the map, showing our ability to fit spectra

6 of 13

E00B08

TSANG ET AL.: TROPOSPHERIC CARBON MONOXIDE ON VENUS

E00B08

Figure 5. An example Visible and Infrared Thermal Imaging Spectrometer (VIRTIS)-M spectrum between 2.18 and 2.50 mm used in this analysis. (top) Solid line (no symbol) is the measured spectrum, with dotted lines showing the limits of uncertainty in the radiances which arise from forward modeling and the measurement of the radiances. Solid, star symbol line is the model fit when fitting to cloud opacity. Solid, diamond symbol line is the model fit when fitting to CO, H2O, and OCS, assuming a cloud opacity derived from the solid star line. (bottom) The radiance residual between the measured and fitted spectrum. at those precise locations. The c2/n map is quite homogeneous, with values of between 1 and 3. Much higher c2/n values appear near the day-night terminators and at the poles because of bad spectra and poor signal-to-noise. It should be noted here that the c2/n fitting is still in an acceptable range, even though the optical depth has increased because of the presence of the polar cloud and vortex. The primary driver of the retrieved errors is the cloud opacity. If the cloud opacity is extremely high, like at the polar vortex, the signal-to-noise reduces dramatically, and thus the retrieved errors increase as well. [23] The resulting CO abundances and related errors are plotted on Figure 6. The mean retrieved errors in the abundances are of the order of 2 to 3 ppm (approximately 10% of the retrieved variance to mean ratio abundances) in the midlatitudes. The errors then increase toward the poles and the equator. The large errors at the equator are due to the geometry of the observation during this orbit insertion phase, where the subspacecraft point was above the poles. This results in radiances with high emission angles at the equator, with reduced signal-to-noise ratios. Correspondingly, the polar regions beyond 70°S are also subject to low signal-to-noise because of the increase in optical depths of the polar clouds. [24] We see that the mean equatorial values of 23 ± 3 ppm increase toward the poles, reaching a maximum near 60°S, of 32 ± 3 ppm. The distribution of CO poleward of 60°S

also indicates a possible decrease in abundance. This general first-order trend of increasing from equator, reaching maximum near/at 60°S, and inversion toward the poles is best seen in Figure 6. This map also shows a degree of variability of CO in solar longitude. We can best see this in Figure 7, where we plot the retrieved CO abundances as a function of the solar longitude. Solar longitude of 100° corresponds to the dusk terminator, while 260° corresponds to the dawn terminator. The first-order trends are apparent, but variability seen in Figure 7 as a function of solar longitude indicates a number of general features which are of note: (1) The mean abundance of CO on longitudes near the dusk terminator has greater absolute value than post local midnight and near dawn. (2) The maximum of the CO abundance near dusk is at 60°S, while the CO maximum near dawn is closer to 70°S. (3) Midlatitude (approximately 40°S) abundances of CO at approximately 26 ± 2 ppm have less variation than at other latitudes. [ 25 ] The same picture is painted with observation CIOB03, taken some 96 h later, again from orbit insertion (Figure 8). The same evening/morning asymmetry in the CO abundance can be seen, although it is less pronounced in this observation than CIOB00. Less of the equator is seen as CIOB03 was taken in the ascending branch of the Venus Express orbit. We also note that the midlatitude values are very constant across solar longitude, while variability can be seen in the dusk and dawn part of the nightside disc. It

7 of 13

E00B08

TSANG ET AL.: TROPOSPHERIC CARBON MONOXIDE ON VENUS

Figure 6. Orbit Insertion Observation CIOB00: (top left) 2.30 1 m radiance, a proxy for cloud opacity, plotted after binning; (top right) c2/n values when fitting for CO between 2.30 and 2.35 mm; (bottom left) the retrieved CO abundance at 35 km; and (bottom right) the error in the retrieved CO abundance. Areas of the observation where the retrieved error exceeds 8 ppm are colored white. Regions of high c2/n are reflected in the map of the retrieved errors (in white).

8 of 13

E00B08

E00B08

TSANG ET AL.: TROPOSPHERIC CARBON MONOXIDE ON VENUS

E00B08

Figure 7. The retrieved CO gradient versus latitude from CIOB00 observation, plotted as a function of solar longitude, interpolated in spacings of 5°. These retrieved CO abundances are referenced to 35 km. Dusk terminator at 100° and dawn terminator at 260°. should also be noted that the mean abundances at 60°S for CIOB00 and CIOB03 are 32 and 34 ± 2 ppm, respectively. [26] To investigate some of these features further, to enhance the sampling and to see the if the degree of variability of CO concentrations changes over time, we take observations from the nominal orbit during ‘‘case 2’’ observations, MTP03-99-2 and MTP04-121-0. These are shown in Figures 9 and 10, respectively. MTP03-99-2 shows a constant band of maximum CO abundance of between 35 to 36 ± 2 ppm at a latitude of 60°S. Zonal variability at this latitude can just be made out, with local maxima of CO at 90 and 135°W and a minimum at approximately 112°W, with a value of 34 ppm. However, it should be noted that the retrieved error is of the order of these fluctuations and therefore is not entirely without doubt. In addition, the maximum in CO commencing at 60°S continues at a steady value until as high as 70°, where a slight decrease in the abundance past 70°S can be made out. The retrieved errors in this region have not increased significantly. Observation MTP04-121-0 shows a similar trend. Again, the CO maximum is located at 60°S, with mean upper limit of 36 ± 3 ppm, but in this instance seems more localized around a longitude of 45°W, decreasing by 3 ppm toward 68°W. However, caution should be taken again as this decrease is

of the order of the retrieved error. The decrease from 65°S in the CO abundance to 30 ppm toward the poles is matched by increased error caused by the increase in the cloud optical depth of the polar vortex, decreasing the signal-tonoise ratio. Therefore, we cannot fully confirm the validity of the overturning from this image alone.

6. Discussion [27] We first confirm the equator-to-pole increase of CO in the troposphere first seen by Collard et al. [1993] from Galileo/NIMS data, validated by Marcq et al. [2006] from IRTF/SpeX data. It is also consistent with observations made by Marcq et al. [2008], using much higher spectral resolution subsystem of VIRTIS-H, who have shown very similar results for both equatorial and polar values of CO. The authors of the above work found a 30 ± 15% in- crease from 0° to 60°, corresponding to 24 ± 3 to 31 ± 2 ppm at 36 km. This is very consistent with the results of this paper, which show on first order, an increase in the same latitude range of 23 ± 2 to 32 ± 2 ppm at the same altitude. [28] In addition, because of the unique imaging capabilities of the VIRTIS-M subsystem, this analysis has also possibly revealed two new features in the tropospheric CO

9 of 13

E00B08

TSANG ET AL.: TROPOSPHERIC CARBON MONOXIDE ON VENUS

Figure 8. Orbit Insertion Observation CIOB03: (left) Retrieved CO abundance at 35 km with (right) the associated retrieval errors. Areas of the observation where the retrieved error exceeds 8 ppm are colored white.

Figure 9. (left) The retrieved CO abundance at 35 km and (right) the errors for observation MTP03-99-2. Variability in the CO abundance at 60°S across different longitudes can be seen. 10 of 13

E00B08

E00B08

TSANG ET AL.: TROPOSPHERIC CARBON MONOXIDE ON VENUS

E00B08

Figure 10. (left) Retrieved CO from observation MTP04-121-0, with (right) the retrieved errors. Areas of the observation where the retrieved error exceeds 8 ppm are colored white. distribution which have not been previously observed and warrant further discussion: the evening/morning asymmetry in the CO zonal profiles and the reversal in CO concentration poleward of 60°. Both these features can be explained by a combination of the Hadley-type circulation, as depicted in Figure 11, and photochemistry from the cloud tops.

[29] In regard to the overturning of CO past 60°S, the mechanism for this is likely to be purely dynamical in nature. The general idea is that the descending branch of the Hadley cell occurs at 60°. This is consistent with GCM models such as Lee et al. [2007] where the maximum downwelling occurs at 60°N/S. It is also consistent with measurements of the dayside meridional winds from Pioneer

Figure 11. A pictorial representation of the Hadley cell type circulation which exists in the Venus atmosphere. A complex interplay between dynamical and chemical cycles combines to yield the features seen in the retrieved CO maps. Photo-disassociation reaction taken from Huebner et al. [1992]. 11 of 13

E00B08

TSANG ET AL.: TROPOSPHERIC CARBON MONOXIDE ON VENUS

Venus which indicate the horizontal convergence of the meridional flow is greatest between 50 and 60° [Limaye, 2007]. We expect the maximum downwelling of the Hadley cell to be associated with regions of maximum vertical winds, and therefore would bring the most CO rich air from mesosphere (>65 km) to the troposphere. If this downwelling is greatest at 60°, it implies a poleward decrease in CO concentration past 60°S, and therefore would be consistent with our observations. [30] The situation with respect to the east-west asymmetry in the CO abundances is probably a mixture of photochemistry and dynamics processes, which is responsible for the concentration of CO at this altitude since the dominant chemistry occurs way above or way below this 35 km region. The supply of CO is governed by the upper atmosphere source. The two options to create this east-west asymmetry in CO at 35 km is either an east-west gradient in the source of CO in the source region and constant downwelling, or a constant source of CO in the source region and an east-west gradient in the downwelling across longitudes. It is indeed likely to be a combination of both processes. [31] The evening/morning asymmetry seen in Figures 6 and 8 can be, on first order, simply a function of the photochemistry which is occurring at the cloud tops at 65 km. If we consider what is occurring at the cloud tops, during the early morning, we would expect a minimum in the early morning with a ramp up of CO production in the morning sunshine due to the photolysis by solar UV, and would depend on the photolysis rate of CO above the clouds. This would peak in the early evening (before sunset probably). Maps of CO on the nightside of Venus above the clouds [Irwin et al., 2008b] indicate enhanced CO at the evening terminator and are consistent with these observations of the tropospheric CO and photochemical mechanisms above the clouds. As the photolysis stops because of the lack of sunlight, no more CO is produced on the nightside. The meridional circulation, however, will continue to pump mesospheric air into the lower troposphere. At the start of the evening you would therefore expect high abundances of CO in the lower atmosphere at the location of the downwelling of the Hadley cell and for this concentration to decay as the night progresses, assuming the vertical velocity of this downwelling was constant across the nightside (which is most likely not the case). However, CO is constantly being destroyed by sink reactions, most notably by OCS [Pollack et al., 1993] and S2 [Hong and Fegley, 1997], and reconverted back to CO2. Therefore, we also need to know something about the creation and destruction rates for CO in the lower atmosphere, as well as the meridional overturning rates (10 m s1 poleward winds at 100 mbar). From the observations of Figures 6 and 8, a change in the concentration of CO of 2 ppm from the evening to morning terminator can be seen. There may be some problem accounting for this given what is known about the reaction rates and atmospheric dynamics.

7. Conclusions [32] Observations of the nightside of Venus taken from Venus Express/VIRTIS-M instrument at wavelengths between 2.20 to 2.50 mm have been used to measure and map the distribution of CO in the troposphere of Venus. We have

E00B08

taken spectra from four different observations, two from orbit insertion, and two from the nominal orbit, which are then coadded and binned for increased signal-to-noise and homogeneous spatial sampling. We use a spectral fitting algorithm to first account for variations in the cloud optical depth by fitting spectra from 2.20 to 2.30 mm. This is then followed by fitting the remaining spectra from 2.30 to 2.35 mm, which is sensitive to the 2-0 CO band, by scaling a known a priori profile. [33] We first confirm the presence of the CO enhancement from equator to pole, with a mean increase of between 40% and 45% from 23 ± 2 ppm to 32 ± 2 ppm, which is consistent with observations made by the Galileo/NIMS flyby of Venus in 1990. We find the peak value of CO to occur between 55 and 65°S. We also find evidence of two possibly new features in this CO gradient; decreases of CO poleward of 60°S and variability in longitude. [34] A mixture of photochemistry and dynamics will be responsible for the concentration of CO at this altitude because the dominant chemistry does not occur near 35 km. Advection of CO from below is unlikely because of a lack of a continuous source, leaving photolysis of CO2 above 60 km as the most likely source of CO in the lower atmosphere. The CO is then advected into the lower atmosphere, particularly in the downwelling region near 60°, where it is destroyed. Given the need for sunlight to photolyse the CO2, the distribution of CO at the source altitude probably follows the solar zenith angle, peaking in the afternoon hemisphere because of the background flow. Poleward advection in the source region and downwelling at the poles then transports the CO into the lower atmosphere, producing the observed latitudinal gradient at 35 km. These observations would then place the downwelling branch of the Hadley cell near 60°, consistent with observations of peak westward flow at 60° and GCM simulations such as Lee et al. [2007]. It is also consistent with measurements of the dayside meridional winds from Pioneer Venus which indicate the horizontal divergence (converging air mass) of the meridional flow is greatest between 50 and 60° [Limaye, 2007]. Poleward of 60°, the vortex ‘‘barrier’’ and decreasing sunlight would contribute to a negative latitudinal gradient in CO, as observed. In addition, the fact that we observe the CO gradient at 35 km, which is a proxy for the return branch of the Hadley cell, implies the vertical extent of the cell(s) is at least 40 km. [35] Acknowledgments. I would like to thank Emmanuel Marcq for his discussions of CO and his work with VIRTIS-H spectra and its findings. We would also like to thank Frank Mills and Robert Carlson for their discussions on this subject. This work was made possible from funding given by the United Kingdom Science Technology Facilities Council, as well as the continued support of CNES and ASI.

References Allen, D., and J. Crawford (1984), Cloud structure on the dark side of Venus, Nature, 307, 222 – 224, doi:10.1038/307222a0. Baines, K., S. Sushil Atreya, R. Carlson, D. Crisp, P. Drossart, V. Formisano, S. Limaye, W. Markiewicz, and G. Piccioni (2006), To the depths of Venus: Exploring the deep atmosphere and surface of our sister world with Venus Express, Planet. Space Sci., 54, 1263 – 1278, doi:10.1016/ j.pss.2006.04.034. Bezard, B., C. de Bergh, D. Crisp, and J. Maillard (1990), The deep atmosphere of Venus revealed by high-resolution nightside spectra, Nature, 345, 508 – 511, doi:10.1038/345508a0. Carlson, R., K. Baines, T. Encrenaz, F. Taylor, P. Drossart, L. Kamp, J. Pollack, E. Lellouch, and A. Collard (1991), Galileo infrared imaging

12 of 13

E00B08

TSANG ET AL.: TROPOSPHERIC CARBON MONOXIDE ON VENUS

spectroscopy measurements at Venus, Science, 253, 1541 – 1548, doi:10.1126/science.253.5027.1541. Collard, A. (1993), Thermal emission from the nightside of Venus, D.Phil. thesis, Univ. of Oxford, UK. Collard, A., F. Taylor, S. Calcutt, R. Carlson, L. Kamp, and K. Baines (1993), Latitudinal distribution of carbon monoxide in the deep atmosphere of Venus, Planet. Space Sci., 41(7), 487 – 494. Coradini, A., et al. (1998), VIRTIS: An imaging spectrometer for the Rosetta mission, Planet. Space Sci., 46(9 – 10), 1291 – 1304. Counselman, C., S. Gourevitch, R. King, G. Loriot, and E. Ginsberg (1980), Zonal and meridional circulation of the lower atmosphere of Venus determined by radio interferometry, J. Geophys. Res., 85, 8026 – 8030, doi:10.1029/JA085iA13p08026. Delaye, C., J. Hartmann, and J. Taine (1989), Calculated tabulations of H2O line broadening by H2O, N2, O2, and CO2 at high temperature, Appl. Opt., 28(23), 5080 – 5087. Drossart, P., et al. (2007), Scientific goals for the observation of Venus by VIRTIS on ESA/Venus Express mission, Planet. Space Sci., 55, 1653 – 1672, doi:10.1016/j.pss.2007.01.003. Fegley, B. J., M. Y. Zolotov, and K. Lodders (1997), The oxidation state of the lower atmosphere and surface of Venus, Planet. Space Sci., 125, 416 – 439. Gierasch, P. (1975), Meridional circulation and the maintenance of the Venus atmospheric rotation, J. Atmos. Sci., 32, 1038 – 1044, doi:10.1175/ 1520-0469(1975)032<1038:MCATMO>2.0.CO;2. Hong, Y., and B. Fegley (1997), Formation of carbonyl sulphide (OCS) from carbon monoxide and sulphur vapour and applications to Venus, Icarus, 130, 495 – 504, doi:10.1006/icar.1997.5824. Huebner, W. F., J. J. Keady, and S. P. Lyon (1992), Solar photon rates for planetary atmospheres and atmospheric pollutants, Astrophys. Space Sci., 195, 1 – 294, doi:10.1007/BF00644558. Irwin, P., P. Parrish, T. Fouchet, S. Calcutt, F. Taylor, A. Simon-Miller, and A. Nixon (2004), Retrievals of Jovian tropospheric phosphine from Cassini/CIRS, Icarus, 172, 37 – 49, doi:10.1016/j.icarus.2003.09.027. Irwin, P. G. J., N. Teanby, R. de Kok, L. Fletcher, C. Howett, C. C. C. Tsang, C. F. Wilson, S. B. Calcutt, C. Nixon, and P. Parrish (2008a), The NEMESIS planetary atmosphere radiative transfer and retrieval tool, J. Quant. Spectrosc. Radiat. Transf., 109, 1136 – 1150, doi:10.1016/ j.jqsrt.2007.11.006. Irwin, P. G. J., R. de Kok, A. Negra˜o, C. C. C. Tsang, C. F. Wilson, P. Drossart, G. Piccioni, D. Grassi, and F. W. Taylor (2008b), Spatial variability of carbon monoxide in Venus’ mesosphere from Venus Express/VIRTIS Visible and Infrared Thermal Imaging Spectrometer measurements, J. Geophys. Res., 113, E00B01, doi:10.1029/ 2008JE003093. Kamp, L., F. Taylor, and S. Calcutt (1988), Structure of Venus’s atmosphere from modelling of night-side infrared spectra, Nature, 336, 360 – 362, doi:10.1038/336360a0. Karaiskou, A. C., C. Vallance, V. Papadakis, I. M. Vardavas, and T. P. Rakitzis (2004), Absolute absorption cross-section measurements of CO2 in the ultraviolet from 200 to 206 nm at 295 and 373 K, Chem. Phys. Lett., 400, 30 – 34. Krasnopolsky, V. A. (2007), Chemical kinetic model for the lower atmosphere of Venus, Icarus, 191, 25 – 37, doi:10.1016/j.icarus.2007.04.028. Lacis, A., and V. Oinas (1991), A description of the correlated k-distribution method for modeling non-grey gaseous absorption, thermal emission and multiple scattering in vertically inhomogeneous atmospheres, J. Geophys. Res., 96, 9027 – 9063, doi:10.1029/90JD01945. Lee, C., S. Lewis, and P. Read (2007), Superrotation in a Venus general circulation model, J. Geophys. Res., 112, E04S11, doi:10.1029/ 2006JE002874. Limaye, S. (2007), Venus atmospheric circulation: Known and unknown, J. Geophys. Res., 112, E04S09, doi:10.1029/2006JE002814.

E00B08

Marcq, E., B. Bezard, T. Encrenaz, and M. Birlan (2005), Latitudinal variations of CO and OCS in the lower atmosphere of Venus from nearinfrared nightside spectro-imaging, Icarus, 179, 375 – 386, doi:10.1016/ j.icarus.2005.06.018. Marcq, E., T. Encrenaz, B. Bezard, and M. Birlan (2006), Remote sensing of Venus’ lower atmosphere from ground-based spectroscopy: Latitudinal and vertical distribution of minor species, Planet. Space Sci., 54, 1360 – 1370, doi:10.1016/j.pss.2006.04.024. Marcq, E., B. Bezard, P. Drossart, and G. Piccioni (2008), A latitudinal survey of CO, OCS, H2O and SO2 in the lower atmosphere of Venus: Spectroscopic studies using VIRTIS-H, J. Geophys. Res., doi:10.1029/ 2008JE003074, in press. Oyama, V., G. Carle, F. Woeller, J. Pollack, R. Reynolds, and R. Craig (1980), Pioneer Venus gas chromatography of the lower atmosphere of Venus, J. Geophys. Res., 85, 7891 – 7902, doi:10.1029/JA085iA13p07891. Palmer, K., and D. Williams (1975), Optical constants of sulfuric acid; Application to the clouds of Venus?, Appl. Opt., 14(1), 208 – 219. Piccioni, G., et al. (2007), VIRTIS: The Visible and Infrared Thermal Imaging Spectrometer, ESA Spec. Publ. SP-1295, pp. 1 – 27, Eur. Space Agency, Paris. Plass, G. N., G. W. Kattawar, and F. E. Catchings (1973), Matrix operator theory of radiative transfer: 1. Rayleigh scattering, Appl. Opt., 12, 314 – 329. Pollack, J., J. Dalton, D. Grinspoon, R. Wattson, R. Freedman, D. Crisp, D. Allen, B. Bezard, C. de Bergh, and L. Giver (1993), Near-infrared light from Venus’ nightside: A spectroscopic analysis, Icarus, 103, 1 – 42, doi:10.1006/icar.1993.1055. Rodgers, C. (1976), Retrieval of atmospheric temperature and composition from remote measurements of thermal radiation, Rev. Geophys., 14(4), 609 – 624, doi:10.1029/RG014i004p00609. Seiff, A. (1983), Thermal structure of the atmosphere of Venus, in Venus, pp. 215 – 279, Univ. of Ariz., Tucson. ˚, Shemansky, D. E. (1972), CO2 extinction coefficient 1700 – 3000 A J. Chem. Phys., 56, 1582 – 1587, doi:10.1063/1.1677408. Taylor, F. (1995), Carbon monoxide in the deep atmosphere of Venus, Adv. Space Res., 16(6), 81 – 88, doi:10.1016/0273-1177(95)00253-B. Taylor, F., D. Crisp, and B. Bezard (1997), Near-infrared sounding of the lower atmosphere of Venus, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, pp. 325 – 351, Univ. of Ariz. Press, Tucson. Tonkov, M., N. Filippov, V. Bertsev, J. Bouanich, V.-T. Nguyen, C. Brodbeck, J. Hartmann, C. Boulet, and F. Thibault (1996), Measurements and empirical modeling of pure CO2 absorption in the 2.3 mm region at room temperature: Far wings, allowed and collision- induced bands, Appl. Opt., 35(24), 4863 – 4870. Tsang, C. C. C., P. G. J. Irwin, F. W. Taylor, and C. F. Wilson (2008), A correlated-k model of radiative transfer in the near infrared windows of Venus, J. Quant. Spectrosc. Radiat. Transf., 109, 1118 – 1135, doi:10.1016/ j.jqsrt.2007.12.008. Von Zahn, U., S. Kumar, H. Niemann, and R. Prinn (1983), Composition of the Venus atmosphere, in Venus, pp. 299 – 430, Univ. of Ariz., Tucson. 

B. Bezard and P. Drossart, LESIA, Observatoire de Paris, 5 place Jules Janssen, F-92195 Meudon, France. S. Calcutt, R. de Kok, P. G. J. Irwin, F. W. Taylor, C. C. C. Tsang, and C. F. Wilson, Atmospheric, Oceanic and Planetary Physics, Clarendon Laboratory, Department of Physics, University of Oxford, Parks Road, Oxford OX1 3BH, UK. ([email protected]) C. Lee, Geophysics and Planetary Sciences, California Institute of Technology, MC 150-21, Pasadena, CA 91125, USA. G. Piccioni, INAF-IASF, Via del Fosso del Cavaliere 100, I-00133 Rome, Italy.

13 of 13

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, E00B01, doi:10.1029/2008JE003093, 2008

for

Full Article

Spatial variability of carbon monoxide in Venus’ mesosphere from Venus Express/Visible and Infrared Thermal Imaging Spectrometer measurements P. G. J. Irwin,1 R. de Kok,1 A. Negra˜o,2 C. C. C. Tsang,1 C. F. Wilson,1 P. Drossart,3 G. Piccioni,4 D. Grassi,2 and F. W. Taylor1 Received 1 February 2008; revised 16 April 2008; accepted 5 May 2008; published 23 July 2008.

[1] Observations of Venus’ mesosphere by the Visible and Infrared Thermal Imaging

Spectrometer (VIRTIS)-M instrument of Venus Express have been used to investigate the spatial distribution of CO above Venus’ nightside cloud tops by fitting the CO absorption in the (1–0) CO band around 4.7 mm. We find little spatial variation in the abundance of CO at midlatitudes, with a retrieved abundance of approximately 40 ± 10 ppm just above the cloud tops between 65 and 70 km altitude. Unfortunately, we find it very difficult to constrain the abundance of CO in the cold polar collar, centered at about 70°S, as the retrieved temperature structure in the CO line-forming region masks the absorption lines. However, there is a possibility that CO increases toward the poles, as we detect a significant signature of high levels of CO over Venus’ south polar dipole feature in all the observations analyzed so far. To constrain the abundance of CO more closely will require the analysis of higher-resolution VIRTIS-H observations. In addition, limb observations would greatly help to resolve any possible temperature/cloud ambiguities and allow us to assess vertical variations in the abundance of CO. Citation: Irwin, P. G. J., R. de Kok, A. Negra˜o, C. C. C. Tsang, C. F. Wilson, P. Drossart, G. Piccioni, D. Grassi, and F. W. Taylor (2008), Spatial variability of carbon monoxide in Venus’ mesosphere from Venus Express/Visible and Infrared Thermal Imaging Spectrometer measurements, J. Geophys. Res., 113, E00B01, doi:10.1029/2008JE003093.

1. Introduction [2] Carbon monoxide is produced at equatorial and midlatitudes in the dayside upper atmosphere of Venus by UV photolysis (l < 224 nm) of CO2 [von Zahn et al., 1983; Huebner et al., 1992]. It is thought to be transported toward the poles by the planet’s Hadley cell circulation, where it is then subducted and recombined with O near Venus’ hot surface, perhaps via sink reactions with either carbonyl sulfide (OCS) [Pollack et al., 1993], or S2 [Hong and Fegley, 1997]. It is thus expected that the volume mixing ratio of CO decreases as we move deeper into the atmosphere, both owing to such recombination reactions and also owing to the CO-rich air from high altitudes being steadily diluted by mixing with CO-poor air at lower altitudes. Superimposed on this meridional circulation, CO is also rapidly transported longitudinally in Venus’ upper atmosphere by Venus’ strong zonal winds at these pressure levels. Thus, even though it is produced on Venus’ dayside, CO is expected to be quickly transported to Venus’ night-

1 Atmospheric, Oceanic, and Planetary Physics, Clarendon Laboratory, University of Oxford, Oxford, UK. 2 INAF-IFSI, Rome, Italy. 3 LESIA, Observatoire de Paris, Meudon, France. 4 INAF-IASF, Rome, Italy.

Copyright 2008 by the American Geophysical Union. 0148-0227/08/2008JE003093$09.00

side where it can be detected at thermal-IR wavelengths in the (1 –0) CO band around 4.7 mm. [3] Existing observations of CO prior to the arrival of the Venus Express mission at Venus in 2006 are reviewed by de Bergh et al. [2006] and issues concerning the modeling of the distribution of CO are reviewed by Mills and Allen [2007]. The abundance of CO at high altitudes (around 100 km altitude) is estimated from observations at millimeter [e.g., Clancy and Muhleman, 1991; Gurwell et al., 1995] and submillimeter [e.g., Clancy et al., 2003] wavelengths to be highly variable and of the order of 100 – 1000 ppm, while deeper in the atmosphere, the CO abundance has been inferred from measurements of the near-infrared emission from Venus’ night side. Be´zard et al. [1990] and Pollack et al. [1993] used data from the Canada France Hawaii Telescope to infer mean values of 40 ppm and 23 ± 5 ppm at 36 km respectively, with Pollack et al. [1993] also estimating that the abundance of CO increased with height at a rate of 1.20 ± 0.45 ppm/km. These observations roughly coincided with the flyby of Venus of NASA’s Galileo spacecraft, on its way to Jupiter. During the encounter, the NIMS (Near-Infrared Mapping Spectrometer) instrument made the first space-based observations of Venus’ night side near-infrared emission and Collard et al. [1993] used these data to determine the latitudinal distribution of CO in the deep atmosphere, finding that it increased toward the poles with a maximum abundance at approximately 60°N.

E00B01

1 of 11

IRWIN ET AL.: VENUS EXPRESS/VIRTIS MESOSPHERE CO

E00B01

E00B01

Table 1. Summary of Observations Used in This Studya Name

Orbit

VI0029_06 VI0029_07 VI0029_08 VI0038_00 VI0067_00 VI0067_01 VI0067_02 VI0067_03 VI0067_04 VI0067_05

29 29 29 38 67 67 67 67 67 67

Date 19 19 19 28 26 26 26 26 26 26

May May May May June June June June June June

2006 2006 2006 2006 2006 2006 2006 2006 2006 2006

Start Time, UT

End Time, UT

nl

ns

nl

texp (s)

1658 1713 1729 1641 1449 1503 1517 1534 1542 1553

1708 1724 1740 1651 1454 1509 1523 1540 1548 1559

432 432 432 432 432 432 432 432 432 432

256 256 256 256 256 256 256 256 256 256

253 254 254 228 126 126 126 126 126 126

0.36 0.36 0.36 0.36 0.36 0.36 0.36 0.36 0.36 0.36

a

Here nl is the number of wavelengths, ns is the number of spatial samples, and nl is the number of spatial lines; texp is the exposure time.

[4] New observations of Venus’ near-infrared emission have been made by the VIRTIS instrument on Venus Express and these data have been used to investigate further the spatial variation of CO in the deep atmosphere by C. C. C. Tsang et al. (Tropospheric Carbon Monoxide Concentrations and Variability on Venus from Venus Express/VIRTIS-M Observations, submitted to Journal of Geophysical Research, 2008) and E. Marcq et al. (A latitudinal survey of CO, OCS, H2O and SO2 in the lower atmosphere of Venus: Spectroscopic studies using VIRTIS-H, submitted to Journal of Geophysical Research, 2008). Higher in the atmosphere, SPICAV/SOIR UV data have been used to determine the vertical distribution of CO above 60 km, finding values of 10 –20 ppm at midlatitudes at an altitude of 70 km, decreasing to
4 ppm at 90 km, and then increasing rapidly at higher altitudes [Svedhem et al., 2007; Bertaux et al., 2007].

2. Measurements [5] The Venus Express spacecraft was launched on a Soyuz-Fregat rocket from Baikonur, Kazakhstan on 9 November 2005 and went into a near-polar 24 h orbit about Venus on 11 April 2006, where it will remain until at least May 2009. [6] The spacecraft carries a number of remote sensing instruments including VIRTIS (Visible and Infrared Thermal Imaging Spectrometer), which covers the wavelength range 0.27 to 5.19 mm. The VIRTIS instrument is split into two main subsystems, the high-resolution subsystem, VIRTIS-H, and the mapper subsystem, VIRTIS-M. The VIRTIS-H subsystem is an Echelle spectrometer covering the 1.84 to 4.99 mm range at a spectral resolution of 0.001 mm, with an instantaneous field of view (IFOV) of 0.45  2.25 mrad. The VIRTIS-M subsystem, has an IFOV of 0.25  64 mrad sampled by 256 rows of a CCD array, which can be scanned to generate an image covering 64  64 mrad with an angular resolution of 0.25  0.25 mrad. At an apogee of 66,000 km, Venus’ disk extends approximately 180 mrad and thus a complete image of Venus’ disk may be generated by mosaicking 3  3 VIRTIS-M observations. The VIRTIS-M subsystem is itself split into two components: one component covers the 0.27 to 1.0 mm wavelength range with a spectral resolution of 0.002 mm, while an infrared component covers the 1.05 to 5.19 mm range at a lower spectral resolution of 0.01 mm. [7] To investigate the distribution of CO in Venus’ mesosphere we analyzed VIRTIS spectra from 4– 5 mm.

In this spectral range, there is a strong CO2 absorption band at 4.3 mm and a weaker one at 4.8 mm, with the (1 – 0) absorption band of CO observable at around 4.7 mm. No other gases have a significant contribution in this range. At other wavelengths where gas absorption becomes negligible, the absorption of the upper cloud decks becomes important. Day side observations at these wavelengths are considerably complicated by reflected sunlight from the cloud tops and so we chose to concentrate upon VIRTIS-M night side observations where only thermal emission is important, and for which there exist several wide area maps from the south pole to the equator. We chose two sets of observations in particular, VI0029 and VI0067, whose characteristics are summarized in Table 1. These mapping observations were chosen because, at the time of writing, they covered the largest area and had the longest integration time, and consequently best signalto-noise ratio. The signal-to-noise ratio of these spectra typically varies from about 1.0 in the center of the strong CO2 band at 4.3 mm to roughly 40 where the radiance is greatest, near 4.9 mm. To improve the signal-to-noise ratio, the 4.0– 4.95 mm spectra from the data ‘cubes’ sampling Venus’ night side for these orbits (six cubes for VI0067 and three for VI0029) were averaged into 20  20 pixel boxes and stepped by 10 pixels in each direction to achieve Nyquist sampling. The areas covered by both sets of observations are shown in Figure 1. pThe ffiffiffi error on the spectral points was set to either the NESR/ n or the variance of the measured points pffiffiffi divided by n, whichever was larger, where n is the number of points averaged in a single bin and NESR is the noise equivalent spectral radiance. [8] We also searched the measured data set for VIRTIS-M limb-sounding observations of Venus’ night side, but at the time of writing could only identify one suitable observation, which was of the equatorial region and had insufficient vertical sampling to improve upon what could be extracted from the VIRTIS-M nadir observations.

3. Retrieval Model [9] To analyze the recorded spectra, we used the NEMESIS correlated-k radiative transfer and retrieval model [Irwin et al., 2008], which has been successfully applied to numerous planetary remote sensing missions, and has also been used by Tsang et al. (submitted manuscript, 2008) to determine the variability of CO in Venus’ deep atmosphere. The NEMESIS retrieval code is based upon the optimal estimation formalism of Rodgers [2000], which assumes that a good knowledge exists of the expected

2 of 11

E00B01

IRWIN ET AL.: VENUS EXPRESS/VIRTIS MESOSPHERE CO

E00B01

Figure 1. Sampling of Venus’ southern hemisphere by the Venus Express/VIRTIS (left) VI0067 observations and (right) VI0029 observations. variation in the retrieved parameters from other sources. Such a situation does not exist in this case and hence the a priori guessed profiles and covariance matrices are instead tuned to provide a reasonable balance between precision and vertical smoothing, in the manner of the constrained linear inversion technique [Conrath et al., 1998; Hanel et al., 2003], as described by Irwin et al. [2008]. [10] To use the model, k-distribution tables were first generated with VIRTIS-M resolution from line data for the different spectrally active gases in this spectral and pressure region. We used line data from the HITRAN 2004 database [Rothman et al., 2005] with a Voigt line shape for all gases except CO2, for which we used a sub-Lorentzian line shape of Tonkov et al. [1996] with a cutoff at 150 cm1 from the line centers. Although this line shape was originally developed for analyzing spectra at 2.3 mm, we find that it allows us to reproduce the entire 4 – 5 mm spectrum very well, in particular the section from 4 to 4.3 mm, which is not wellproduced by other line shapes. Our final retrieved temperature maps match closely those retrieved by Grassi et al. [2008] from the 4.3 to 5.1 mm range (excluding the CO band at 4.7 mm) who use a different sub-Lorentzian correction. Hence, the use of the Tonkov et al. [1996] line shape does not significantly effect our temperature retrievals, and allows us to better model the weak CO absorption feature. The k-tables were calculated on a grid with 20 temperatures spread equally between 100 and 350 K, and 20 pressure levels spread logarithmically between 4  108 and 2.7 bar, with a square spectral resolution of width 10 nm. When calculating spectra, NEMESIS convolved the output with a further square bin of width 10 nm, to achieve an effective instrument function close to that experimentally determined for VIRTIS-M. [11] For these tests we assumed that the upper cloud optical depth was dominated by a lognormal distribution of H2SO4 droplets of mean radius of 1.0 mm and a variance of 1.29 (i.e., the mode 2 particles of Grinspoon et al. [1993]) and computed the extinction cross-section spectrum from Mie theory. Scattering effects were found to be negligible and so synthetic spectra were calculated assuming thermal emission only. [12] Our first-guess, or a priori, temperature profile was taken from Seiff [1983], while for clouds we assumed the

model of Roos et al. [1993], who used limb-darkening observations from Galileo/NIMS observations to determine that the mean cloud structure at equatorial latitudes was best matched by a cloud with a scale height of approximately 5.2 km. The a priori cloud profile density was thus set to decrease with height at this rate and the profile scaled to give a visible optical depth, at 630 nm, of 1.0 at an altitude of 65 km. For gaseous profiles the mole fraction of CO2 was set to 0.965 [von Zahn et al., 1983] and the remaining gases included were H2O and SO2, whose a priori profiles were set as recommended by von Zahn et al. [1983] and Krasnopolskii and Parshev [1983], although these gases were found to contribute negligibly to this spectral region. For CO we assumed a constant mole fraction with an a priori abundance of 100 ppm, based on determinations of the abundance in the lower atmosphere of 30 ppm (Tsang et al., submitted manuscript, 2008) and assuming the abundance increases with height. [13] NEMESIS can simultaneously retrieve multiple variables and can model atmospheric constituents either as continuous or parameterized profiles. To assess the retrieval model and the validity of the correlated-k approximation [Lacis and Oinas, 1991], NEMESIS was first used to retrieve the vertical temperature profile alone from VIRTIS-M observations and the results compared with the temperature retrievals of Grassi et al. [2008]. Very good agreement was achieved, adding confidence to both retrieval models. In addition, we found that the effects of variable cloud opacity and temperature at altitudes less than 70 km were indistinguishable since the calculated extinction cross section of the assumed cloud particles has very little variation with wavelength across the 4 – 5 mm window. [14] To investigate whether the CO abundance in Venus’ mesosphere could indeed be retrieved from VIRTIS-M observations of Venus’ night side from 4 to 5 mm, a series of retrieval tests was first undertaken whereby a set of synthetic spectra were generated from a set of randomly varied atmospheres. Synthetic Gaussian noise was added to these spectra, whose amplitude was set by the expected NESR of VIRTIS for typical observations. These synthetic ‘measured’ spectra were then fed into NEMESIS to retrieve some or all of the quantities varied and the retrieved properties compared with the values originally used to

3 of 11

E00B01

IRWIN ET AL.: VENUS EXPRESS/VIRTIS MESOSPHERE CO

E00B01

Figure 2. Correlation of retrieved CO with synthetic (or ‘true’) CO for the tests where temperature and mean CO abundance were retrieved with NEMESIS from synthetic spectra generated by randomly varying the temperature, mean CO abundance, and cloud opacity and scale height about their a priori values. The agreement between retrieved and ‘true’ CO can be seen to be very good. Cases where the retrieved CO is significantly greater than the true CO are cases where the cloud scale height exceeded 7 km. generate the synthetic spectra. The error on the ‘measured’ spectra was taken to be the same NESR used to generate the random noise together with a degree of forward modeling error, which accounts for uncertainties in the line data and correlated-k approximation. [15] Analysis of the spectrum generated from our a priori atmosphere showed that we are sensitive to the abundance of CO just about the cloud tops, whose lines usually appear in absorption as temperature falls with height at this altitude for most regions of the planet. Unfortunately, we found that these observations could reveal very little about the vertical distribution of CO and were thus forced to assume a vertical profile and decided to assume that the mole fraction varies little with height in this region, retrieving a single scaling factor. [16] In the first test, the cloud profile was kept constant and a range of spectra generated by varying the temperature profile and CO mean mixing ratio. For the temperature profile, sine waves of temperature (varying with respect to height) were added to the a priori temperature profile with a random variation of phase, wavelength and amplitude. Similarly, the CO mean mixing ratio was varied randomly about the a priori value. NEMESIS then used the resulting synthetic spectra (to which Gaussian noise had been added) to retrieve the temperature profile and mean CO abundance, and the retrieved CO values were compared to the true values. In all cases the retrieved CO closely matched the ‘true’ CO values. However, we were concerned that variation in the cloud profile might be aliased as CO variations and so an additional set of test spectra was generated where

temperature, CO and the cloud profile (represented by two parameters: (1) the integrated visible optical depth from space to an altitude of 65 km and (2) a scale height above and below that altitude) were randomly varied. These test spectra were then used (1) to again retrieve the temperature profile and mean CO mole fraction and (2) to retrieve the temperature profile, mean CO mole fraction and the two cloud parameters. In both cases the correlation between the retrieved and true CO was found to be very good (Figure 2) and can be explained by the fact that variations in the cloud profile produced spectral effects that are indistinguishable from thermal variations and also that we are most sensitive to CO just above the clouds. The cases in Figure 2 where the fitted CO greatly exceeded the true value were where the cloud scale height exceeded 7 km, which is a value far greater than anticipated for Venus, especially near the poles, where values of less than 5 km are expected. Hence, when analyzing the real observations, discussed in the next section, the cloud profile was kept fixed at its a priori profile and only the vertical temperature profile and mean CO mole fraction retrieved.

4. Retrievals [17] Before analyzing the extended maps, the measurements in the VI0029 and VI0067 data sets, which had already been averaged into 20  20 pixel bins as described earlier, were further averaged by latitude into bins of width 10° and spaced every 5°, again to achieve Nyquist sampling. Since the Venus Express spacecraft was virtually over the south pole for both sets of observations, the emission

4 of 11

E00B01

IRWIN ET AL.: VENUS EXPRESS/VIRTIS MESOSPHERE CO

E00B01

Figure 3. Variation of the retrieved CO volume mixing ratio scaling factor (of an assumed profile with 100 ppm CO at all altitudes) with latitude for the VI0067 and VI0029 observations. The horizontal dashed line indicates the a priori scaling factor and the dash-dotted lines indicate the a priori error limits. The retrieved CO abundance is indicated by the solid line, with retrieved error limits indicated by the dotted lines. angle did not vary much with longitude and so the measured spectra could simply be averaged and a mean value of the emission angle assigned to each latitude bin. The a priori temperature profile was set as described earlier with an assumed temperature error of 10 K at all altitudes and a correlation length of 1.5 scale heights. For CO we initially assumed a scaling factor of 0.1 ± 0.2 of the a priori profile where the mole fraction was set to 100 ppm at all altitudes. To prevent the retrieved mole fraction ever going negative

NEMESIS was set to retrieve the logarithm of the CO scaling factor. [18] Figure 3 shows the variation of retrieved CO factor with latitude in the southern hemisphere for the VI0029 and VI0067 observations. It can be seen that in both cases at low latitudes the mean retrieved CO abundance is reasonably well constrained at a level of 40– 50 ppm with an error of about 15 ppm. However, poleward of 50°S, the retrieved abundance drops rapidly to around 10 ppm, although the

Figure 4. (left) Fitted temperature profile contour map, (middle) CO functional derivative (rate of change of mean radiance in the range 4 – 4.95 mm (mW cm2 sr1 mm1) with CO mole fraction) contour map, and (right) sections of the CO functional derivatives at 20°S (dotted line) and 70°S (solid line) for the combination of the VI0067 and VI0029 observations. 5 of 11

E00B01

IRWIN ET AL.: VENUS EXPRESS/VIRTIS MESOSPHERE CO

E00B01

Figure 5. Quality of fit to the VI0067 observations at 20°S and 70°S. Measured spectra are shown as points and the units of radiance are mW cm2 sr1 mm1. Fitted spectra are shown by the solid lines. The spectrum calculated for the same temperature profile, but with 0 ppm of CO is shown as the dotted line, while the spectrum calculated for the same temperature profile, but with 100 ppm of CO is shown as the dash-dotted line. The top row compares the spectra, while the bottom row shows the residual difference between the measured and fitted spectra. The small absorption spike at 4.69 mm, which remains even for the 0 ppm CO case, is due to a weak Q-branch absorption of CO2. retrieved error increases markedly, approaching its a priori value, indicating the solution is poorly constrained by the measurements. Intriguingly, while Figure 3 shows the low value of CO extending to the pole for the VI0067 observations, the retrieved abundance for the VI0029 observations seems to increase toward the pole. [19] To understand why the model has difficulty in retrieving a CO scaling factor in the polar collar around 70°S we need to look at the retrieved temperature profile, shown in Figure 4. From Figure 4 we can see that there is a marked temperature inversion in the 65– 70 km altitude region at these latitudes associated with Venus’ polar collar. Figure 4 also shows the CO functional derivative, that is to say the sensitivity of the calculated radiance (integrated over 4 – 4.95 mm) with respect to the CO abundance at all latitudes and altitudes, and sections are also shown at 20°S (typical midlatitude) and 70°S (in the middle of the polar collar). At 20°S we can see that the sensitivity to CO abundance is greatest at
68 km and that increasing the CO fraction at all altitudes reduces the radiance, indicating the

CO spectral lines to be in absorption. In contrast, at 70°S we can see that while the derivative of radiance with CO abundance is negative at some altitudes, it becomes positive in the temperature inversion region. Given that we can only retrieve a mean CO fraction from these measurements, having a situation where CO increases the mean radiance at some altitudes and reduces it at others leads to limited and ambiguous sensitivity to this species. As a test, we took the retrieved temperature profile and then recalculated the spectra assuming a CO abundance of (1) 0 ppm and (2) 100 ppm, shown in Figure 5, together with the spectrum calculated with the fitted CO value. In the region of the polar collar (70°S), very little difference between the spectra calculated with the retrieved CO abundance and the two trial abundances could be seen, whereas at midlatitudes (20°S) varying the mean CO abundance had a very clear effect on the spectra. Hence, while we can be reasonably confident of the retrieved CO abundance at low to mid latitudes, the abundance in the polar collar is effectively

6 of 11

E00B01

IRWIN ET AL.: VENUS EXPRESS/VIRTIS MESOSPHERE CO

E00B01

Figure 6. Measured radiance (in units of mW cm2 sr1 mm1) averaged over the range 4.8– 5.0 mm for the (left) VI0067 and (right) VI0029 observations (binned into 20  20 pixel bins and stepped every 10 pixels in both directions). unconstrained as is indicated by the retrieved errors differing little from the assumed a priori limits. [20] As a final test, we also attempted to fit the observations with a temperature profile alone, keeping the CO value fixed at 100 ppm. We found that allowing the CO abundance to vary led to significantly improved fits to the observations in regions where we are sensitive to CO, showing that CO indeed has a significant and measurable effect on the observed spectrum. [21] Having investigated the fitting of temperature and CO to the longitudinally averaged spectra, and determined at

which latitudes CO might be retrieved, we then took the mapped observations and attempted to retrieve maps of the mean CO abundance in the southern hemisphere. Figure 6 shows the measured radiance (integrated from 4.8 to 5.0 mm) for the two sets of observations, indicating the position of the southern polar vortex and Figure 7 shows the precision to which the spectra could be fitted with our model expressed n c2 1 X ððmi  si Þ=ei Þ2 , where mi are the measured ¼ as n 1 n radiances of estimated error ei, si are the calculated synthetic radiances, and n is the number of observations.

Figure 7. Fitted c2/n for the (left) VI0067 and (right) VI0029 observations. 7 of 11

E00B01

IRWIN ET AL.: VENUS EXPRESS/VIRTIS MESOSPHERE CO

E00B01

Figure 8. Retrieved CO scaling factor for the (left) VI0067 and (right) VI0029 observations, where areas with c2/n > 3 have been masked (black areas). Note that a scaling factor of 1.0 corresponds to 100 ppm of CO. Areas where c2/n is less than
1 are regions where our model fits the observed spectra well and thus where we can have some confidence in our retrieved temperatures and CO abundances, if CO is detectable. Tests were conducted with a priori CO scaling factors of 1.0 ± 2.0, 0.1 ± 2.0 and 0.1 ± 0.2. In all cases very similar results were achieved since there is no trade-off to be made between vertical resolution and precision in these retrievals as the shape of the CO profile is fixed, resulting in a simple least squares solution for CO abundance that does

not depend on the initial guess. Here we will show the results for the case where the CO a priori scaling factor was set to 0.1 ± 0.2. Since we fit the log of the scaling factor, it is important to remember that the a priori error assumed by NEMESIS was actually the fractional error in the abundance, in this case 2.0. In regions where there is little CO information, the retrieved fractional error remains close to 2.0, while in regions where CO is better constrained the fractional error reduces to less than 0.5. Figure 8 shows the retrieved CO factor regardless of retrieved error, while

Figure 9. Fitted CO scaling factor for the (left) VI0067 and (right) VI0029 observations, where regions with a fractional error exceeding 0.4 and c2/n > 3 have been masked. 8 of 11

E00B01

IRWIN ET AL.: VENUS EXPRESS/VIRTIS MESOSPHERE CO

E00B01

Figure 10. Retrieved values of T(70 km)  T(65 km) for the (left) VI0067 and (right) VI0029 observations, showing the inverted temperature profile at these altitudes over the polar collar. Figure 9 shows the retrieved CO factor for cases where the fractional error has been reduced to less than 0.4 and thus where we can be confident that the CO abundance really is being constrained by the observations. The large black areas in Figure 9, mostly coincide with regions of the polar collar temperature inversion around 70°S, as shown in Figure 10. [22] Comparing the retrieved CO maps of Figures 8 and 9 with the mean latitudinal variation for the two sets of observations shown earlier in Figure 3 we can see many

of the same features: roughly constant CO abundance at low to mid latitudes, decreasing in the polar collar (but with much less constraint) and then increasing again toward the pole in the case of the VI0029 observations. Both maps seem to indicate a higher abundance of CO near the evening terminator at 50°S – 70°S, although since the abundance of CO at similar latitudes on the morning terminator is not well constrained (as can be seen from Figure 9) owing to the vertical temperature structure there, it is not possible to determine if this is a real increase; it is possible that the CO

Figure 11. (left) Measured radiance (in units of mW cm2 sr1 mm1) averaged over the range 4.8 – 5.0 mm for the VI0038_00 observation, together with (right) the fitted CO abundance, again showing high CO over one of the two bright dipoles at 75°N, 0°E, indicative of rapid downwelling. 9 of 11

E00B01

IRWIN ET AL.: VENUS EXPRESS/VIRTIS MESOSPHERE CO

abundance is higher at all latitudes between 50°S and 70°S, but only observable near the evening terminator. [23] The retrieved CO abundance near the pole for the VI0029 observation is particularly interesting as it seems to pick out high CO values over the polar dipole feature at 80°S, 45°E and 85°S, 45°W. This feature would appear to be absent in the VI0067 observation owing to the polar dipole being mostly on the day side in this case as indicated in Figure 6. To check if this was a repeatable feature, an additional observation, VI0038_00, which covers the south polar dipole, was analyzed in the same way to yield the fitted CO map shown in Figure 11. Again, we see high CO over one of the dipole hot spots at 75°N, 0°E, which would be consistent with this being a region of rapid downwelling dragging CO-rich air from higher altitudes to just above the cloud tops where it can be observed.

5. Discussion and Conclusions [24] Observations of the (1 – 0) absorption band of CO at 4.7 mm on Venus’ night side by the VIRTIS-M instrument of Venus Express have been used to investigate the spatial distribution of this gas above Venus’ cloud tops. Retrieval tests showed the effects of cloud top temperature and cloud opacity to be indistinguishable in the 4 – 5 mm range and that the retrieved CO abundance differed little if clouds were included or omitted. Hence, the cloud opacity was fixed and a vertical temperature profile fitted simultaneously with the CO abundance. [25] We find little spatial variation in the abundance of CO at midlatitudes, with a retrieved abundance of approximately 40 ± 10 ppm, a figure slightly higher than the levels observed by SPICAV/SOIR [Svedhem et al., 2007] of 10– 20 ppm at an altitude of 70 km. However, SPICAV/SOIR estimated the CO abundance at this altitude to decrease with height and as we are most sensitive to the abundance just above the cloud tops at
65 km, our results are broadly consistent. [26] Unfortunately, we find it very difficult to constrain the abundance of CO in the cold polar collar at 70°S as the retrieved temperature structure in the CO line-forming region masks the lines. However, the observations would suggest that CO increases toward the poles as we detect a significant signature of CO over Venus’ south pole, coinciding with the areas of the bright dipole feature, which would be consistent with rapid downwelling in these features bringing CO-rich air from high altitudes to just above the clouds where it can be detected. In addition, both observations suggest the possibility of higher abundances of CO between 70°S and 50°S near Venus’ evening terminator, although it is possible that this enhancement exists at all other longitudes, but cannot be detected owing to the CO lines being masked by the vertical temperature profile. [27] A caveat to the conclusion of high CO over the polar dipole is that the shape of the vertical profile of CO had to be assumed in these retrievals, owing to the observations having effectively no vertical resolution assumption in the line forming region (65 – 70 km). It was assumed that the abundance of CO did not vary with altitude, but if the CO abundance actually decreases rapidly with altitude and if cloud tops in the polar dipole are much deeper than elsewhere, as is thought to be the case, then we would see

E00B01

a longer path length of CO in this region due simply to CO’s vertical profile. However, since the polar dipole is thought to be a region of rapid downwelling and since we know the abundance of CO increases with height in the high atmosphere, where CO is formed by UV photolysis, it could be argued that this possibility is unlikely. To answer some of these questions and constrain the abundance of CO more closely will require more work and the analysis of many more observations than has been possible in this paper. One approach will be to analyze also the VIRTIS-H observations, which have much better spectral resolution, but poorer spatial sampling. This work is currently in progress, but the results are not yet mature enough to be presented here. In addition, limb observations would allow us to directly determine the vertical profile of CO above the cloud tops around the planet and help greatly to resolve any possible temperature/cloud ambiguities and we hope that such measurements can be conducted by VIRTIS during the remainder of its mission. [28] Acknowledgments. The Venus Express/VIRTIS team is acknowledged for their work. This research was made possible by funding from the United Kingdom Science and Technology Facilities Council.

References Bertaux, J.-L., et al. (2007), A warm layer in Venus’ cryosphere and high altitude measurements of HF, HCl, H2O and HDO, Nature, 450, 646 – 649, doi:10.1038/nature05974. Be´zard, B., C. de Bergh, D. Crisp, and J. Maillard (1990), The deep atmosphere of Venus revealed by high-resolution night side spectra, Nature, 345, 508 – 511, doi:10.1038/345508a0. Clancy, R. T., and D. O. Muhleman (1991), Long-term (1979 – 1990) changes in the thermal, dynamical and compositional structure of the Venus mesosphere as inferred from microwave spectral line observations of 12CO, 13CO, and C18O, Icarus, 89, 129 – 146, doi:10.1016/ 0019-1035(91)90093-9. Clancy, R. T., B. J. Sandor, and G. H. Moriarty-Schieven (2003), Observational definition of the Venus mesopause: Vertical structure, diurnal variation, and temporal instability, Icarus, 161, 1 – 16, doi:10.1016/ S0019-1035(02)00022-2. Collard, A., F. Taylor, S. Calcutt, R. Carlson, L. Kamp, and K. Baines (1993), Latitudinal distribution of carbon monoxide in the deep atmosphere of Venus, Planet. Space Sci., 41, 487 – 494, doi:10.1016/ 0032-0633(93)90031-V. Conrath, B. J., P. J. Gierasch, and E. A. Ustinov (1998), Thermal structure and para hydrogen fraction in the outer planets from Voyager IRIS measurements, Icarus, 135, 501 – 517, doi:10.1006/icar.1998.6000. de Bergh, C., V. I. Moroz, F. W. Taylor, D. Crisp, B. Bezard, and L. V. Zasova (2006), The composition of the atmosphere of Venus below 100 km altitude: An overview, Planet. Space Sci., 54, 1389 – 1397, doi:10.1016/j.pss.2006.04.020. Grassi, D., P. Drossart, G. Piccioni, N. I. Ignatiev, L. V. Zasova, A. Adriani, M. L. Moriconi, P. G. J. Irwin, A. Negrao, and A. Migliorini (2008), Retrieval of air temperature profiles in the Venusian Mesosphere from VIRTIS-M data: Description and validation of algorithms, J. Geophys. Res., doi:10.1029/2008JE003075, in press. Grinspoon, D. H., J. B. Pollack, B. R. Sitton, R. W. Carlson, L. W. Kamp, K. H. Baines, T. Encrenaz, and F. W. Taylor (1993), Probing Venus’s cloud structure with Galileo NIMS, Planet. Space Sci., 41, 515 – 542, doi:10.1016/0032-0633(93)90034-Y. Gurwell, M. A., D. O. Muhleman, K. P. Shah, G. L. Berge, D. J. Rudy, and A. W. Grossman (1995), Observations of the CO bulge on Venus and implications for mesospheric winds, Icarus, 115, 141 – 158, doi:10.1006/ icar.1995.1085. Hanel, R. A., B. J. Conrath, D. E. Jennings, and R. E. Samuelson (2003), Exploration of the Solar System by Infrared Remote Sensing, 2nd ed., Cambridge Univ. Press, Cambridge, U. K. Hong, Y., and B. Fegley (1997), Formation of carbonyl sulfide (OCS) from carbon monoxide and sulfur vapor and applications to Venus, Icarus, 130, 495 – 504, doi:10.1006/icar.1997.5824. Huebner, W. F., J. J. Keady, and S. P. Lyon (1992), Solar photo rates for planetary atmospheres and atmospheric pollutants, Astrophys. Space Sci., 195, 1 – 289, 291 – 294.

10 of 11

E00B01

IRWIN ET AL.: VENUS EXPRESS/VIRTIS MESOSPHERE CO

Irwin, P. G. J., N. A. Teanby, R. de Kok, L. N. Fletcher, C. J. A. Howett, C. C. C. Tsang, C. F. Wilson, S. B. Calcutt, C. A. Nixon, and P. D. Parrish (2008), The NEMESIS planetary atmosphere radiative transfer and retrieval tool, J. Quant. Spectrosc. Radiat. Transf., 109, 1136 – 1150, doi:10.1016/j.jqsrt.2007.11.006. Krasnopolskii, V. A., and V. A. Parshev (1983), Photochemistry of the Venus atmosphere, in Venus, pp. 431 – 458, Univ. of Ariz. Press, Tucson. Lacis, A. A., and V. Oinas (1991), A description of the correlated-k distribution method for modeling nongray gaseous absorption, thermal emission, and multiple scattering in vertically inhomogeneous atmospheres, J. Geophys. Res., 96, 9027 – 9063, doi:10.1029/90JD01945. Mills, F. P., and M. Allen (2007), A review of selected issues concerning the chemistry in Venus’ middle atmosphere, Planet. Space Sci., 55, 1729 – 1740, doi:10.1016/j.pss.2007.01.012. Pollack, J., J. Dalton, D. Grinspoon, R. Wattson, R. Freedman, D. Crisp, D. Allen, B. Bezard, C. de Bergh, and L. Giver (1993), Near-infrared light from Venus’ night side: A spectroscopic analysis, Icarus, 103, 1 – 42, doi:10.1006/icar.1993.1055. Rodgers, C. D. (2000), Inverse Methods for Atmospheric Sounding: Theory and Practice, World Sci., Hackensack, N. J. Roos, M., P. Drossart, T. Encrenaz, E. Lellouch, B. Bezard, R. W. Carlson, K. H. Baines, L. W. Kamp, F. W. Taylor, and A. D. Collard (1993), The upper clouds of Venus: Determination of the scale height from NIMSGalileo infrared data, Planet. Space Sci., 41, 505 – 514, doi:10.1016/ 0032-0633(93)90033-X. Rothman, L. S., et al. (2005), The HITRAN 2004 molecular spectroscopic database, J. Quant. Spectrosc. Radiat. Transf., 96, 139 – 204, doi:10.1016/j.jqsrt.2004.10.008.

E00B01

Seiff, A. (1983), Thermal structure of the atmosphere of Venus, in Venus, pp. 215 – 279, Univ. of Ariz. Press, Tucson. Svedhem, H., D. V. Titov, F. W. Taylor, and O. Witasse (2007), Venus as a more Earth-like planet, Nature, 450, 629 – 632, doi:10.1038/nature06432. Tonkov, M., N. Filippov, V. Bertsev, J. Bouanich, V.-T. Nguyen, C. Brodbeck, J. Hartmann, C. Boulet, and F. Thibault (1996), Measurements and empirical modeling of pure CO2 absorption in the 2.3 mm region at room temperature: Far wings, allowed and collision-induced bands, Appl. Opt., 35, 4863 – 4870. von Zahn, U., S. Kumar, H. Niemann, and R. Prinn (1983), Composition of the Venus Atmosphere, in Venus, pp. 299 – 430, Univ. of Ariz. Press, Tucson. 

R. de Kok, P. G. J. Irwin, F. W. Taylor, C. C. C. Tsang, and C. F. Wilson, Atmospheric, Oceanic, and Planetary Physics, Clarendon Laboratory, University of Oxford, Parks Road, Oxford OX1 3PU, UK. ([email protected]) P. Drossart, LESIA, Observatoire de Paris, 5 place Jules Janssen, F-92195 Meudon, France. D. Grassi and A. Negra˜o, INAF-IFSI, Via del Fosso del Cavaliere, 100, I-00133 Rome, Italy. G. Piccioni, INAF-IASF, Via del Fosso del Cavaliere, 100, I-00133 Rome, Italy.

11 of 11

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, E00B23, doi:10.1029/2008JE003140, 2008

for

Full Article

Composition of the Venus mesosphere measured by Solar Occultation at Infrared on board Venus Express A. C. Vandaele,1 M. De Mazie`re,1 R. Drummond,1 A. Mahieux,1 E. Neefs,1 V. Wilquet,1 O. Korablev,2 A. Fedorova,2 D. Belyaev,2 F. Montmessin,3,4 and J.-L. Bertaux3,4 Received 14 March 2008; revised 20 June 2008; accepted 4 September 2008; published 27 December 2008.

[1] Solar Occultation at Infrared (SOIR), which is a part of the Spectroscopy for

Investigation of Characteristics of the Atmosphere of Venus (SPICAV) instrument on board Venus Express, combines an echelle-grating spectrometer with an acoustooptical tunable filter. It performs solar occultation measurements in the IR region at a high spectral resolution better than all previously flown planetary spectrometers. The wavelength range probed allows for a detailed chemical inventory of the Venus atmosphere above the cloud layer, with an emphasis on the vertical distribution of the gases. A general description of the retrieval technique is given and is illustrated by some results obtained for CO2 and for a series of minor constituents, such as H2O, HDO, CO, HCl, and HF. Detection limits for previously undetected species will also be discussed. Citation: Vandaele, A. C., et al. (2008), Composition of the Venus mesosphere measured by Solar Occultation at Infrared on board Venus Express, J. Geophys. Res., 113, E00B23, doi:10.1029/2008JE003140.

1. Introduction [2] Venus is a very warm and dry planet with a dense atmosphere composed mainly of carbon dioxide (CO2, 96.5%) and Nitrogen (N2, 3.5%). Chemically active species, such as sulfuric bearing gases (OCS and SO2) and halides (HCl and HF) have already been reported (see de Bergh et al. [2006] for a general review on the composition of the atmosphere of Venus below 100 km altitude). Measurements have been performed essentially in the mesosphere below 100 km and below the clouds. Information about minor atmospheric constituents, their concentration, reactions, sources and sinks is incomplete, as for example only scarce measurements have been performed above 100 km altitude. In particular, photochemical models of the middle atmosphere would benefit from abundance measurements of Cl-bearing gases. [3] The Solar Occultation at Infrared (SOIR) spectrometer is an extension mounted on top of the Spectroscopy for Investigation of Characteristics of the Atmosphere of Venus (SPICAV) instrument [Bertaux et al., 2007a]. SPICAV/ SOIR is one of the seven instruments on board Venus Express, a planetary mission of the European Space Agency (ESA) that was launched in November 2005 and inserted into orbit around Venus in April 2006 [Titov et al., 2006]. 1

Belgian Institute for Space Aeronomy, Brussels, Belgium. Space Research Institute, Moscow, Russia. 3 Service d’Ae´ronomie du CNRS, Verrie`res-le-Buisson, France. 4 Institut Pierre Simon Laplace, Universite´ de Versailles-Saint-Quentin, Saint Quentin en Yvelines, France. 2

Copyright 2008 by the American Geophysical Union. 0148-0227/08/2008JE003140$09.00

[4] SOIR [Nevejans et al., 2006] is designed to measure at high resolution (0.15 cm
1) the atmospheric transmission in the IR (2.2 – 4.3 mm) using solar occultations. This technique allows for the derivation of unique information about the vertical structure and composition of the Venus mesosphere. SOIR is the first high-resolution NIR spectrometer on board a spacecraft investigating the Venusian atmosphere and it enables a sensitive search for new minor species from the top of the clouds up to about 125 km of altitude.

2. Description of the Instrument [5] The instrument has already been extensively described elsewhere [Bertaux et al., 2007a; Mahieux et al., 2008; Nevejans et al., 2006] and will only be briefly described here. SOIR is an echelle-grating spectrometer operating in the IR, combined with an acoustooptic tunable filter (AOTF) for the selection of the diffraction-grating orders. The free spectral range (FSR) of the echelle spectrometer, i.e., the spectral interval in which there is no interference or superposition of light from adjacent orders equals 22.38 cm
1, whereas the bandwidth of the AOTF was originally designed to be 20 cm
1, as measured on ground before launch [Nevejans et al., 2006]. The real measured bandwidth of SOIR is 24 cm
1 [Mahieux et al., 2008], creating some order leakage on the detector. The wave number domain that can be investigated by the SOIR instrument extends from 2256 to 4369 cm
1, and is divided into 94 smaller ranges corresponding to the different orders (from 101 to 194). The detector width for orders 101 to 122 is smaller than the FSR of 22.38 cm
1 and hence the detector will miss part of the spectrum. For orders 123 to

E00B23

1 of 16

E00B23

VANDAELE ET AL.: VENUS MESOSPHERE COMPOSITION BY SOIR

E00B23

Figure 1. Example of spectra obtained during one occultation (sunset 15 April 2007). Each transmittance is obtained by making the ratio of the solar spectrum seen through the Venus atmosphere to the unattenuated solar spectrum measured above the atmosphere. The selection of a spectral interval is achieved through the acoustooptic tunable (AOT) filter, tuned in this case to (a) 15809 kHz for diffraction order 121, (b) 19869 kHz for diffraction order 149, (c) 23031 kHz for diffraction order 171, and (d) 25742 kHz for diffraction order 190. In these particular ranges of wave number, the main absorption lines are from HDO (Figure 1a), CO2 (Figure 1b), H2O (Figure 1c), and CO (Figure 1d). 194 the inverse happens: the detector width is equal to or larger than the FSR and the detector will not be completely covered by the selected order. [6] The SOIR detector has 320 columns along the wave number axis and 256 rows along the spatial axis. The slit is projected on 32 fixed rows only. Since, owing to imposed telemetry limitations, only a data volume equivalent to 8 rows of 320 pixels can be retrieved per second, one is forced to bin the rows in eight groups of four rows, provided only one order (or AOTF frequency setting) is used during a given second. It is however possible to select up to four different orders (not necessarily sequential) per second, allowing us to gather a more versatile set of absorption lines. This reduces the maximum measurement time per order to 250 ms and implies that only 2 larger bins of 16 rows will be used if the complete slit height has to be covered. Later, the binning was changed to 2 bins of 12 rows because the outside rows of the illuminated part of the detector received a lower signal, because that part of the slit was too close to the edge of the Sun [Mahieux et al., 2008]. Background measurements are subtracted onboard from the measurements themselves. [7] Raw spectra, registered by SOIR and transmitted to Earth, need dedicated processing in order to upgrade them to a calibrated data set. This involves detector nonlinearity correction, spectral calibration and division by a reference solar spectrum. Ideally, the reference spectrum that is taken

outside the atmosphere would be measured with an identical relative slit position with respect to the solar disk. Attitude drift of the spacecraft, however, makes the slit float which results in a gradual linear change of the intensity. This effect is also corrected for Mahieux et al. [2008]. [8] A SOIR occultation observation can be taken either at sunset or sunrise. In the case of a sunset, the measurement cycle is started well before the instrument’s line of sight to the Sun intersects with the top layers of the atmosphere, and reference spectra are recorded (at a rate of 1 spectrum s – 1). Once the top of the atmosphere is reached, solar light is absorbed and the intensity of the recorded signal starts to decrease until the Sun gets so flattened by refraction that the spectrometer slit moves out of the refracted solar disk. One of the main advantages of solar occultations is that it is a self-calibrated technique in terms of transmission: dividing a spectrum obtained during the occultation by a reference solar spectrum recorded outside the atmosphere removes the solar signature and leaves a transmittance containing only information about the composition of the Venus atmosphere. The reference spectrum is in fact defined by selecting spectra recorded within the 40 s before the level at 220 km is reached. [9] Figure 1 gives an example of the evolution of the spectra through one occultation (sunset 15April 2007) in the orders 121, 149, 171, and 190 corresponding to the 2725– 2750, 3330 – 3357, 3820 – 3855, and 4245 – 4283 cm
1

2 of 16

E00B23

VANDAELE ET AL.: VENUS MESOSPHERE COMPOSITION BY SOIR

E00B23

Figure 2. Latitudinal distribution of measurements of (a) CO2, (b) CO, (c) HF, and (d) HCl. Dots represent measurements performed when Venus Express is separated from limb at 65 km tangent height by less than 5000 km and crosses when this distance is larger. In the latter case, the vertical resolution is coarser. Only measurements corresponding to the smallest distances have been considered in this study. Most of those correspond to north polar air masses. ranges, respectively. These transmittances show the characteristic behavior observed on all occultation series measured by SOIR. At the beginning of the series, the light path does not cross the atmosphere. No absorption signatures are present and transmittances are equal to unity. As the Sun

sets, the light path goes deeper and deeper into the atmosphere, and two absorption processes take place: the overall signal decreases owing to extinction by aerosols and absorption signatures appear. At the end of the observation, no light is captured anymore when the Sun disappears behind

3 of 16

E00B23

VANDAELE ET AL.: VENUS MESOSPHERE COMPOSITION BY SOIR

E00B23

Figure 3. Typical examples of the vertical resolution for different types of orbits. The geometry of the orbit defines the field of view (instantaneous range of altitude encompassed by the instrument entrance slit taken at the beginning of the measurement, solid line) and the vertical resolution (dashed line). This corresponds to the portion of atmosphere sounded during one measurement, thus 250 ms. It highly depends on the distance from the planet and the velocity of Venus Express (VEX). Figures 3a and 3b show the vertical resolution and the field of view during orbits 223 (30 November 2006; distance to the limb at 65 km tangent height is 2000 km; 81°N) and 332 (19 March 2007; distance to the limb is 11198 km; 4°N) as a function of time. the cloud deck or moves out of SOIR’s field of view owing to diffraction. The structures seen in the spectra of Figure 1 are mainly attributed to HDO (Figure 1a), CO2 (Figure 1b), H2O (Figure 1c), and CO (Figure 1d). From Figure 1, it can also be seen that in general, the SOIR spectra contain information on the Venus atmosphere between 65 and 110 km for molecules such as HDO or HCl. For H2O and CO signatures are still observable up to 130 km altitude and CO2 features are seen up to 125 –130 km [Bertaux et al., 2007b; Wilquet et al., 2007]. [10] Most of the measurements of SOIR occur at high northern latitude because of the shape of the orbit with its pericenter located at about 250 km above the northern pole and its apocenter at about 65,000 km. When solar occultation occurs, a sunset or a sunrise can be observed. When the satellite is close to the planet the vertical resolution is less than 1 km. Measurements correspond to latitudes ranging from 60° to 90°N. When the satellite is located far from Venus, measurements have a poorer vertical resolution and occur at lower latitudes, typically from 70°S to 60°N. Figure 2 illustrates the latitudes and longitudes, corresponding to the tangent altitude of 65 km, of the measurements yielding information on CO2, CO, HCl, and HF. We have distinguished two types of geometries, when the distance between

Venus Express (VEX) to the limb (at 65 km tangent height) is less than 5000 km (dots), and when it is larger (crosses). Figure 2 will be discussed in more detail in section 4, where results for each species will be described individually. [11] We define the vertical resolution at the tangent point as the total altitude range scanned during one measurement. Because the slit is not always parallel to the limb, but rotating slightly, and because the measurement lasts for 250 ms, this vertical resolution may vary between 100 m and several kilometers in the worst cases. Another variable is the field of view, which is defined as the instantaneous height encompassed by the instrument slit at the beginning of the measurement. This parameter always gives a lower limit to the vertical resolution. The vertical resolution is mainly a function of the distance of the satellite to the planet. The distance of the spacecraft to the planet and the velocity of the spacecraft influence the vertical resolution in the following ways: (1) the further the spacecraft is from the limb, the larger the size of the instantaneous height measured in the atmosphere; (2) the velocity of the spacecraft projected at the limb on the atmospheric local vertical depends on the position of the spacecraft on its orbit around Venus at the moment of the occultation.

4 of 16

E00B23

VANDAELE ET AL.: VENUS MESOSPHERE COMPOSITION BY SOIR

E00B23

Figure 4. Geometry of solar occultation measurements and definition of the onion- peeling method. [12] Figure 3 shows the variation of the field of view and vertical resolution as a function of time of orbit 223 (30 November 2006; distance to the limb at 65 km tangent height is 2000 km; 81°N) and 332 (19 March 2007; distance to the limb is 11,198 km; 4°N). During orbit 332, the spacecraft was far away from the planet which implies a large atmospheric height seen by the instrument slit. The field of view and vertical resolution are very similar because the Sun is rising almost vertically in the atmosphere and the velocity of the spacecraft is not too high (the spacecraft is almost at the equator). During orbit 223, spacecraft is close to the planet and the instantaneous height of the sounded atmosphere is much smaller. The velocity is higher, because the spacecraft is almost at pericenter (87°N). For the reasons explained just above, we will limit the retrieval to spectra obtained under favorable conditions, i.e., corresponding to low distance between VEX and the limb at 65 km tangent height (<5000 km) and with vertical resolution of less than 2 km. As can be inferred from Figure 2, this condition limits the selection to regions near the north pole, with latitudes above 60°N.

3. Description of the Retrieval Technique [13] The retrieval technique is based on the onion peeling method, illustrated in Figure 4. The atmosphere is treated as an onion-like composite of spherical layers, in which the temperature, pressure, and mixing ratios of the constituents are held constant.

model based on the fundamental knowledge of the optical properties of the atmosphere. In the forward model, SOIR spectra are simulated using a line-by-line (LBL) code developed initially for Earth exploration [Vandaele et al., 2006] and adapted for the conditions on Venus. The general equation describing radiative transfer through the atmosphere can be written as I ðn Þ ¼ I0 ðn Þe
tðn;0;sobs Þ þ

Z

Sobs

Bðn; T ðsÞÞaðn; sÞe
tðn;0;sÞ ds ð1Þ

0

where I0 represents the light intensity of the source (here the Sun) placed at the starting point of the raypath situated at the distance sobs from the observer, a(n, s) is the absorption coefficient, B(n, T) is the Planck function, and Z

s2

t ðn; s1 ; s2 Þ ¼

aðn; sÞds

ð2Þ

s1

is the optical depth along the path between the points s1 and s2. In solar occultation, the second term is often negligible with respect to the Sun intensity. The optical depths t of the absorbing constituents along the line of sight (see Figure 4) are calculated at high resolution in each layer considering the temperature and pressure determined by the ray-tracing procedure:

3.1. Forward Model [14] Quantitative analysis of the recorded spectra needs first the calculation of synthetic spectra through a forward 5 of 16

t ð8; n Þ ¼

Z að PðsÞ; T ðsÞ; n Þds s

ð3Þ

VANDAELE ET AL.: VENUS MESOSPHERE COMPOSITION BY SOIR

E00B23

where t is a function of wave number and depends on the tangent height. This, in turn, depends on the solar zenith angle 8 and on the atmospheric refraction characteristics, which, on Venus, are far from negligible. Starting from known temperature and pressure vertical profiles, the raytracing calculations are carried out on a finer grid (with a 200 m step) and the final results, i.e., the effective temperature and pressure in each layer as well as the effective densities, are obtained using the Curtis-Godson approximations [Goody and Yung, 1995]. In this work, we have used temperature and pressure vertical profiles from the VIRA model for altitudes up to 100 km [Seiff et al., 1985]. For higher altitudes (from 140 km and upward), data were taken from the model of Hedin et al. [1983] as suggested by Mueller-Wodarg and Tingle [2008]. The transition between the two data sets was performed by spline interpolating the temperature and reconstructing the pressure through the hydrostatic law. In the following we will refer to this composite model as the VENUSREF model. [15] The extinction coefficient a(P, T, n) is a function of the temperature and pressure prevailing at altitude z. It represents all absorption processes, including Rayleigh aR and aerosol aA extinction and absorption by molecular species: að Pð zÞ; T ð zÞ; n Þ ¼ aR ð z; n Þ þ aA ð z; n Þ þ

M X

si ð Pð zÞ; T ð zÞ; n ÞNi ð zÞ

ð4Þ

i¼1

where M is the number of absorbing species, si and Ni are the absorption cross section and number density of species i. The contribution of the aerosols will be further described and discussed in a later paper (V. Wilquet et al., Characterization of the upper Venusian haze from UV to mid-IR by SPICAV/SOIR on Venus Express, submitted to Journal of Geophysical Research, 2008). The contribution of each species is determined using a line-by-line model based on a line catalog specific to Venus, as explained later (see section 3.2). The absorption coefficient kij (cm2 molecule
1) for a particular line j of species i is given as   kij ðn; T ; P; pi Þ ¼ Sij ðT Þ  f n; vij ; T ; P; pi

ð5Þ

where 8(. . .) is a normalized line shape. The intensity Sij(T) exhibits a temperature dependence which can be described as     c2 n 0j T Qi ðT0 Þ c2 Ej T10
T1 1
e
 Sij ðT Þ ¼ Sij ðT0 Þ e
c n 2 0j Qi ðT Þ 1
e T0

ð6Þ

where c2 is the second Planck constant hc/kB (1.4387 cm K), with h the Planck constant, kB the Boltzmann constant, and c the speed of light, n 0,j is the central wave number of the jth transition, Ej is the energy of the lower state (cm
1), and Sij(T0) is the intensity at the reference temperature T0 (cm
1 molecule
1 cm2). Q(T) and Q(T0) are the total partition functions under local thermodynamic equilibrium condi-

E00B23

tions, at temperature T and T0, respectively. These functions describe the temperature dependence of the line intensity of the transition. They are approximated by QðT Þ ¼ a0 þ a1 T þ a2 T 2 þ a3 T 3

ð7Þ

where a0, a1, a2, and a3 are tabulated coefficients [Gamache et al., 1990]. ASIMUT allows the user to select between different line profiles, the Voigt profile being the default. In the case of the H2O and CO2 molecules, sub- and superLorentzian line profiles have been observed [Clough et al., 1989; Pollack et al., 1993], which are characterized by a symmetric or asymmetric c function: fc ðn Þ ¼ cðn Þ  fL ðn; P; T Þ

ð8Þ

This correction factor has been introduced to take into account the fact that far from the line center, the line displays marked deviations from the Lorentzian behavior 8L. Typically, CO2 displays a sub-Lorentzian behavior (the opacity far from the line center is less than that predicted by a Lorentzian profile) whereas H2O shows a superLorentzian behavior. The c function is usually defined on large spectral intervals. [16] The spectral grid, on which the profile must be determined, must be fine enough so that the narrowest line be adequately represented. In the upper atmosphere, the Doppler width is the limiting factor for selecting an adequate sampling value. As Doppler width depends on temperature, it varies with altitude; the line profile must therefore be sampled with different steps as the altitude varies. ASIMUT determines for each layer, of temperature T and pressure P, the optimized sampling step as DnðP; T Þ ¼

1 4

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi a2D ðT Þ þ a2L ð P; T Þ

ð9Þ

with aD the Doppler width obtained for a molecule of mass 20.0, and aL the Lorentzian width obtained for a molecule characterized by no self-broadening, a foreign broadening of 0.04 cm
1 atm – 1, and a temperature coefficient of 0.5. However, for some particular temperature and pressure conditions, this sampling step might be too large compared to the desired final resolution. In that case, the sampling is set to the value of final resolution/15. LBL calculations often require the computation of a large number of line shapes over large to very large spectral intervals. The number of points might then become prohibitive. However, it can be observed that the line profile changes slower at a distance from the line center than it does near the center. The solution implemented in ASIMUT is the use of a nonuniform grid: near the center of the line, the optimized step derived with the help of equation (9) is used; the step is then progressively enlarged as one goes away from the center. Our algorithm is based on the study of Fomin [1995], which splits the spectral grid in a series of subintervals. Let us assume that the line shape must be calculated on the interval D, for which a value of 25 cm
1 is considered sufficiently large for most of the simulations, except when applying a c factor. The cutoff value D is then determined by the interval on which this function is given.

6 of 16

E00B23

VANDAELE ET AL.: VENUS MESOSPHERE COMPOSITION BY SOIR

E00B23

Figure 5. Description of the order overlap occurring in the instrument owing to the combined presence of the AOTF filter and the different diffraction orders of the echelle grating. In the top, a simulated spectrum is shown containing only CO2. It spans several diffraction orders whose limits are also indicated. This spectrum is first filtered by the AOTF, and only the portion under the filter bandwidth enters the spectrometer. Because this bandwidth is larger than the free spectral range (FSR) of the echelle spectrometer, more than one order of diffraction is transmitted. This is illustrated in the middle, where the contributions of the different orders are represented. Finally, in the bottom, the sum of all of these contributions is measured on the detector. The limits of the center zone are defined by [n - C, n + C] with C = 2/3 (aD + aL), aD and aL being the Doppler and Lorentzian widths of the line. The grid is divided into 2L portions located at unequal intervals:



 n
D; n
2L
1 C ; . . . ; n
22 C; n
2C ; ½n
2C; n
C ;

 central zone; ½n þ C; n þ 2C ; n þ 2C; n þ 22 C ; . . . ;

 ð10Þ n þ 2L
1 C; n þ D

The number of intervals is related to D and C through the following expression: 2L  C ¼ D

ð11Þ

In the central zone, the sampling is set to the optimized value determined with equation (9). Then in each subsequent interval, the sampling is doubled. This drastically reduces the number of points on which the line profile is calculated, without losing accuracy at the center of the line. At the end, the absorption line shape is interpolated to correspond to the sampling used for the determination of the optical depth (OD). The latter is chosen by considering the step asked for the radiance or transmission simulation and an oversampling factor either provided by the user or chosen such that the OD wave number step is a factor 10 lower than the final radiance step.

[17] Radiances are finally convolved by the instrumental function of the instrument, which is chosen to be a Gaussian, whose width varies between 0.13 and 0.25 cm
1 depending on the spectral interval simulated [Mahieux et al., 2008]. [18] At this point, one must take into account the effect of the AOTF on the measured spectrum. As already noted, the filter has a bandwidth larger than the free spectral range of the echelle spectrometer, implying some order overlap on the detector. This is illustrated in Figure 5 where the contribution of four adjacent orders on top of the central one are considered. The incoming spectrum is represented in the Figure 5 (top), along with the position of the different orders of diffraction spanned by the AOTF filter bandwidth. The position of the maximum of the AOTF bandwidth is determined by the selected RF applied to the device. It is necessary to determine the best possible (AOTF frequency – wave number) calibration curve to increase the accuracy of the simulation. This is done through the careful analysis of specific measurements of some selected solar lines, whose positions are well known [Mahieux et al., 2008]. In the case illustrated here, the RF applied to the AOTF corresponds to the central order number 121. The entire signal received on the detector comes from the adjacent orders as no absorption is present in the incoming spectrum in the range of the central order of diffraction. The contribution coming from the noncentral orders is certainly not negligible and has to

7 of 16

E00B23

VANDAELE ET AL.: VENUS MESOSPHERE COMPOSITION BY SOIR

be taken into account when simulating SOIR spectra. In general, we consider a total of 7 contributing orders to determine the final transmittance. [19] The resulting transmittance is then interpolated to correspond to the wave number values of the observed spectrum. It is possible to fit a wave number shift between the observed and simulated transmittances to optimize the correspondence. 3.2. Spectroscopic Data Sets [20] Spectroscopic data, i.e., line parameters, have been taken from the latest version of the HITRAN database [Rothman et al., 2005]. However, broadening coefficients have been modified in order to take into account the presence of CO2 as main buffer gas, whereas data reported in HITRAN are given for Earth-like air conditions. [21] Sung and Varanasi [2005] reported CO2-broadened half widths and CO2-induced line shifts for the fundamental (1 – 0) of 12C16O at 201, 244, and 300 K, as well as for the first (2 – 0) overtone and the second (3 –0) overtone bands at 298 K. We have considered their values for the lines in common with the HITRAN database. Those which are not reported by Sung and Varanasi have been corrected by a conversion factor air width (shift) to CO2 width (shift) derived from the comparison of available common values. The temperature coefficient has been held constant for all lines (n = 0.73), as suggested by Sung and Varanasi. [22] Toth and Darnton [1974] have performed measurements of the HCl line widths by CO2 in the 1 – 0 and 2 – 0 bands of HCl. Their values have been taken for the lines in common with the HITRAN database. A correction similar to the one devised for CO was applied to the lines not reported by those authors. [23] Air-broadening measurements performed by the Brussels-Reims group [Fally et al., 2003; Jenouvrier et al., 2007; Me´rienne et al., 2003] up to 25,000 cm
1 have revealed that there is a large vibrational dependence of the width for most of the H2O transitions. This is also the case under a CO2 rich atmosphere. Various attempts [Brown et al., 2007; Gamache et al., 1995] have tried to give simple relations between the quantum vibrational numbers and the width and shift of the lines. However, the spectrum of water is so complex that it is virtually impossible to obtain measured values for each band, reducing the validity of the proposed relations. A simplistic solution is to scale the broadening coefficients obtained with other perturbing gases, such as N2 or air, although it has been pointed out by Brown et al. [2007] that ‘‘simple scaling of existing values of air- or nitrogen-broadened parameters will not achieve sufficiently reliable CO2-broadened H2O coefficients’’, as these authors found considerable scatter of the ratio of CO2 to air broadened width (from 0.95 to 3.07) around the mean value of 1.67. However, no data, neither experimental nor calculated, are available in the spectral region investigated by SOIR and the constant factor of 1.67 was applied to correct the air-broadening coefficient given in HITRAN. This value has to be compared to the value of 1.3 usually used in Venus studies [Pollack et al., 1993]. 3.3. Onion-Peeling Method [24] The onion-peeling method was implemented to coherently treat a series of spectra recorded during one

E00B23

occultation. In this method, one starts the analysis in the uppermost layer, i.e., with the first spectrum containing absorption structures due to the constituents of the atmosphere, deriving concentrations in that layer, and progressively goes deeper into the atmosphere taking into account the results from the layers above. Vertical profiles of several key species of the Venusian atmosphere have been obtained by applying this technique, as will be demonstrated hereafter. [25] For the sake of clarity, we will consider in the following that the second right term in equation (1) is negligible. The observed transmittance Tr1 corresponding to ray 1 passing through the uppermost layer (layer 1 in Figure 4) is then given by

 Tr1 ðn Þ ¼ exp
a1 ðn ÞDs11

ð12Þ

where a1(n) stands for a(T1, P1, n) determined using equation (4) and Ds11 is the length of the raypath in layer 1 obtained by the ray-tracing procedure already explained. In this expression, the only unknowns are the concentration Ni of each species and the aerosol loading in layer 1, which are retrieved from the analysis of this first layer. Transmittance observed for ray 2 will result from the combination of the absorption of light in layer 1 (Ds21) and layer 2 (Ds22). If we moreover consider the atmosphere as spherical and homogeneous, we can further write

 Tr2 ðn Þ ¼ exp
a2 ðn ÞDs22
a1 ðn ÞDs21

ð13Þ

in which the only unknowns are the concentration Ni of each species and the aerosol loading in layer 2. By going down progressively, the vertical profiles of the interacting species can be derived.

4. Results [26] In this section, we will present some results of our analysis concerning the various species unambiguously detected with SOIR. CO2, being the most prominent absorber in the IR region, can be probed in different spectral regions. Combining different diffraction orders in one occultation moreover allows the possibility to use different bands presenting very distinctive absorption levels, therefore permitting the extraction of the CO2 vertical profile from the top of the clouds up to 170 km high. There is also the clear possibility to probe different absorption bands originating from the different isotopologues of the CO2. CO has been shown to act as a potential tracer of the dynamical processes occurring in the Venus atmosphere and its detection in the high-altitude range is of high interest. HCl and HF have also been detected in the SOIR spectra and vertical profiles are described. Some results will be briefly described concerning H2O and HDO but we refer to the paper of Fedorova et al. [2008] for a detailed discussion. Finally the detection limits of a series of species, which have not yet been unambiguously detected, are reported and discussed. 4.1. CO2 [27] CO2 is the main component of the atmosphere of Venus (96.5%). The first measurements of the atmospheric composition of Venus were made by Adams and Dunham [1932] using the 100 inch reflector at Mount Wilson. They

8 of 16

VANDAELE ET AL.: VENUS MESOSPHERE COMPOSITION BY SOIR

E00B23

Table 1. Possible Orders Where CO2 Can Be Detecteda Order

Isotopologue

Intensity

Temperature Dependence

101 102 103 104 105 106 107 108 109 111 112 115 116 117 118 140 141 142 143 147 148 149 150 151 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171

628 628 626 (628) 626 (628) 626 (628) 626 (628) 626 626 626 628 628 628 628 628 628 626 626 626 626 626 626 626 626 626 626 626 626 626 626 626 626 626 626 626 626 626 626 626 626 626 626 626

S S S S S S M W W W W W W W W W W W W W M M M W M M M M M M S S S M S S S S M W W W

M M S S S S W W W W W W W W W W W W W W W W W W W W W M M/W M M M M M M/W M/W M/W M M W W W

Category 1 1 1 1 1 1 2–3 3 3 3 3 3 3 3 3 3 3 3 3 3 2–3 2–3 2–3 3 2–3 2–3 2–3 1 2–3 1 1 1 1 1 2 2 2 1 1 3 3 3

a Indication on the altitude range probed and on the isotopologue measured. W, M, and S are for weak, medium and strong, respectively. Category is defined as the following: 1 means S/M intensity with S/M temperature dependence (cases for probing high altitudes CO2 and temperature); 2 means S/M intensity but W temperature dependence (cases for probing high altitudes CO2, less sensitive to temperature); 3 means M/W intensity with W temperature dependence (cases for probing lower altitudes CO2).

discovered three bands that they tentatively attributed to CO2. Since then, CO2 has been proven to be the main absorber in the infrared region. Its absorption bands are present throughout the spectral domain covered by SOIR, with intensities varying over a wide range of values. Combining different spectral intervals (orders of diffraction) in which the CO2 line strengths differ widely, the CO2 vertical profile can be obtained from lower altitudes around 65 km to higher altitudes of about 170 km. Indeed, the interval where the CO2 absorption is the largest will lead to information on the highest layers of the atmosphere, but will saturate for lower tangential height and, on the contrary, spectral intervals where the CO2 lines are weaker will provide information on the deepest layers. Moreover, the choice of the spectral interval has to be done with great care, because of the high temperature dependence of some

E00B23

absorption bands. This, in turn, could lead to the retrieval of temperature profiles if CO2 bands characterized by high temperature dependency are selected. In general, a mix of intervals is chosen such that one is optimal for the CO2 detection at high altitude, a second for CO2 detection at lower altitudes, and finally two intervals, with higher temperature dependency, are chosen to give potential information on the temperature as a function of altitude. A summary of the possible spectral intervals and their characteristics i.e., the altitudes sounded and their temperature dependence is given in Table 1. [28] The possibility to retrieve different isotopologues of CO2 has already been illustrated by the discovery of the 01111 – 00001 band of 12C16O18O in the SOIR spectra [Bertaux et al., 2007b; Wilquet et al., 2007]. Besides this new band, 12C16O18O abundances can be derived from well known features around 2500 cm
1. A series of occultations dedicated to the measurement of the isotopologues which can be detected by our instrument were performed during MTP 21 (orbits 583 to 598) which occurred in January 2007. The spectral region in which SOIR is active contains a large number of lines owing to the three of the four main isotopologues of CO2. Spectral signatures of 12C16O16O, 12 16 18 C O O, and 12C16O17O have been clearly identified in the SOIR spectra, as shown in Figure 6. Simultaneous measurements of the different isotopologues will lead to the determination of the 17O/16O and 18O/16O isotopic ratios as a function of altitude. 4.2. HCL and HF [29] HCl and HF were observed for the first time in the Venus atmosphere by Connes et al. [1967], who estimated their mixing ratios at the cloud top as being 0.6 ppm (refined to 0.4 ± 0.07 ppm by Young [1972]) and 5 ppb, respectively. More recent nightside observations [Be´zard et al., 1990] provided measurements of HCl and HF below the clouds. These authors reported values of 0.5 ± 0.15 ppm and 5 ± 2 ppb, respectively, which are similar to the values found by Connes et al. [1967]. Preliminary measurements of Bjoraker et al. [1992] corresponding to altitudes above 72 km yielded a HF mixing ratio of 6.5 ± 0.3 ppb, in agreement with the values found previously. Recently, Iwagami et al. [2008] measured hemispheric distributions of HCl mixing ratio above and below the Venus cloud deck. These authors reported mean values of 0.76 ± 0.1 ppm at altitudes between 61 and 67 km and 0.4 ± 0.05 ppm at about 20 km. They argued that the larger HCl mixing ratio found above the clouds than that existing below the cloud requires the presence of a production process of HCl in the cloud region or above. [30] HCl abundances are retrieved from SOIR data using a series of lines belonging to the 1 – 0 band in the 2905– 2995 cm
1 spectral range (orders 130 to 133), whereas HF determination is based on two lines of the 1 –0 transition (R1 and R2 at 4038.96 cm
1 and 4075.29 cm
1, respectively, in the orders 180 and 181). The measurement points reproduced in Figure 2 correspond to occultations where one of these orders was observed. Moreover, the P3 line at 3833.66 cm
1 is also used to derive the HF abundance. This line, however, lies in a spectral region rich with H2O lines which render the determination of the HF abundance less accurate.

9 of 16

E00B23

VANDAELE ET AL.: VENUS MESOSPHERE COMPOSITION BY SOIR

E00B23

Figure 6. The isotopologues 12C16O16O, 12C16O18O, and 12C16O17O have been observed during MTP 21 which was fully dedicated to the measurement of CO2 and the determination of the 17O/16O and 18 16 O/ O isotopic ratios as a function of altitude. Top shows observed spectra recorded by SOIR, and bottom presents the corresponding simulations. The asterisks correspond to 12C16O17O lines absorbing in the same region as 12C16O18O, and the circles correspond to 12C16O18O lines absorbing in the same region as 12C16O17O. [31] Figure 7 shows some typical vertical profiles found for HCl corresponding to the orbits 341 (28 March 2007; distance to the limb at 65 km tangent height is 3346 km; 82°N), 356 (12 April 2007; distance to the limb is 3298 km; 84°N), and 366 (22 April 2007; distance to the limb is 3898 km; 73°N). The actual parameters which are retrieved are the HCl densities in each of the sounded layers and volume mixing ratios are obtained by considering the air densities calculated from pressures and temperatures given by the VENUSREF model. The HCl profiles show very similar evolution with altitude. Almost systematically we observe a depletion feature around 90 km altitude. For comparison, a profile corresponding to a constant volume mixing ratio (vmr) (0.5 ppm) is also represented. The interpretation of the depletion feature observed around 90 km is not clear. An analysis with respect to temperature has been performed, which is illustrated in Figure 8, where we have considered four different vertical profiles for the temperature: the VENUSREF model, the VENUSREF model +20% and -20%, and a more realistic profile obtained by the Vera instrument [Pa¨tzold et al., 2007]. The four corresponding density profiles are compared in Figure 8: the absolute values change, because of the temperature changes, but the depletion feature does not disappear. This would indicate that the feature is not directly temperaturedependent. We have also investigated the sensitivity of the retrieval to inaccuracies in the determination of the position of the maximum of the band pass of the AOTF function. The results, also plotted in Figure 8, show that a displace-

ment of 1.0 cm
1 of this maximum (the accuracy on this parameter has been estimated to be 0.83 cm
1) [see Mahieux et al., 2008], implies a decrease of the HCl retrieved density by 10%, but the depletion feature does not disappear. [32] If we consider the evolution of the vmr with altitude of Figure 7, this corresponds to a more or less constant vmr comprised between 0.1 and 0.2 ppm, except around 90 km where it is slightly lower. The values at the lower boundary are somewhat lower than the values found in the literature, 0.4 ppm obtained by Connes et al. [1967] at an altitude of 64 km or 0.76 ppm at 61– 67 km from Iwagami et al. [2008]. Following the interpretation proposed by Iwagami et al. [2008] suggesting that there should be a chemical source of HCl inside or above the clouds to explain the gradient in HCl mixing ratio that they observed, our results seem to indicate that this potential source is restricted to the cloud region. [33] Profiles obtained in the case of HF are presented in Figure 9 for the occultations 357 (13 April 2007; distance to the limb is 3325 km; 83°N), 462 (27 July 2007; distance to the limb is 1864 km; 88°N), and 484 (18 August 2007; distance to the limb is 2670 km; 70°N). The HF abundance shows a more varying vertical distribution than HCl. Values found with SOIR seem to be in very good agreement with the values reported earlier, as for example, by Bjoraker et al. [1992], who measured HF abundances of 6.5 ± 0.3 ppb above 72 km. We note that both for HCl and HF, there is a

10 of 16

E00B23

VANDAELE ET AL.: VENUS MESOSPHERE COMPOSITION BY SOIR

E00B23

Figure 7. Vertical profiles of the HCl density (left) and volume mixing ratio in ppm (right) obtained during the three occultations 341 (solid line, 28 March 2007; distance to the limb at 65 km tangent height is 3346 km; 82°N), 356 (dotted line, 12 April 2007; distance to the limb is 3298 km; 84°N), and 366 (dashed line, 22 April 2007; distance to the limb is 3898 km; 73°N). For comparison, the solid line with dots corresponds to a profile of constant volume mixing ratio (vmr) (0.5 ppm). The vmr in this plot correspond to the ratios of the HCl densities to the total air densities obtained from the temperatures and pressures of the VENUSREF model. Error bars correspond to the fitting error on the single-fitted parameter, i.e., the density in the corresponding layer, and does not take into account the errors on the densities in the upper layers. region of minimum vmr. However, their altitudes are different: 90 km for HCl, and 80 km for HF. 4.3. CO [34] The primary source of CO in the atmosphere of Venus is the photodissociation of CO2 by solar UV at altitudes higher than 120 km. The mixing ratio of CO was measured by the Pioneer Venus gas chromatograph [Oyama et al., 1980] at different altitudes in the lower atmosphere (between 22 and 62 ppm at 52 km, 30 ± 18 ppm at 42 km and 20 ± 3 ppm at 22 km). The gas chromatograph on Venera 12 [Gel’man et al., 1980] confirmed the low value found below 42 km (28 ± 7 ppm). Connes et al. [1968] reported a value of 45 ± 10 ppm at 64 km from Earth-based observations in the near-infrared. This value was corrected to 51 ± 1 ppm by Young [1972], who reanalyzed the spectra recorded by Connes et al. [1968]. These measurements seem to indicate the presence of a gradient in the mixing ratio of CO at least in the altitude range sounded. This was confirmed by observations of microwave lines of CO, which yielded CO mixing ratio for altitudes between 75 and 105 km. The CO mixing ratio increases from 55 ppmv at 75 km, to 130 ppmv at 85 km and 200 to 1000 ppmv at 105 km. Moreover it was shown that CO exhibits a significant diurnal variation but also strong year to year

variations [Clancy and Muhleman, 1991; Clancy et al., 2003; Gurwell et al., 1995]. Ground-based observations of the night side of Venus [Marcq et al., 2005; Marcq et al., 2006] have reported that the CO abundances in the lower atmosphere (below the clouds) showed a pronounced latitudinal enhancement of more than 10% when going toward the poles. Observations by the VIRTIS-M instrument on Venus Express [Irwin et al., 2008] have shown that there was little spatial distribution of CO just above the cloud (approximately 65– 70 km) at midlatitudes, with abundances of the order of 40 ± 10 ppm, with higher values at the poles, consistent with rapid downwelling bringing CO-rich air from higher altitudes. [35] Retrieval of the vertical CO profile from SOIR spectra relies on several absorption lines of the (2 – 0) CO band located between 4178 cm
1 (order 187) and 4325 cm
1 (order 192). All orbits during which this spectral region is investigated are shown in Figure 2b, distinguishing orbits corresponding to low and high (>5000 km) distance to the limb. Typically the vertical profiles are obtained from 70 to 125 km altitude, as can be seen from the examples reproduced in Figure 10. Those profiles correspond to the orbits 341 (28 March 2007; distance to the limb is3346 km; 82°N), 356 (12 April 2007; distance to the limb is 3298 km; 84°N), and 366 (22 April 2007; distance to the limb is 3898 km;

11 of 16

E00B23

VANDAELE ET AL.: VENUS MESOSPHERE COMPOSITION BY SOIR

E00B23

Figure 8. Investigation of the temperature effect on the detection of HCl. In the left, the corresponding density profiles are shown. For comparison, the line with dots corresponds to a profile of constant vmr of 0.5 ppm. The feature around 85 km does not disappear. In the right, the relative difference in the retrieved HCl densities is shown for the different test cases. Four different temperature profiles have been considered: VENUSREF, VENUSREF with
20% and +20% excursions, and a vertical temperature profile obtained by the Vera instrument on board Venus Express [Pa¨tzold et al., 2007]. The effect of displacing the AOTF function maximum by 1.0 cm
1 is also indicated. 73°N). CO vmr have been obtained from the retrieved CO densities using the air densities calculated from temperature and pressure from the VENUSREF model at the considered level. All three profiles show values for the CO vmr between 10 and 50 ppm below 90 km, and then increasing up to 4 – 8  104 ppm at 125 km, consistent with the existence of a source of CO, the photodissociation of CO2, at high altitudes. These profiles were obtained at the northern polar region and seem to be consistent with the observations performed by VIRTIS-M [Irwin et al., 2008] which indicate high values of the CO abundance at the south pole for altitudes between 65 and 70 km. They found values ranging from 100 to 200 ppm, these values being mostly influenced by the position relative to the brightest features of the polar vortex dipole. [36] Our value at 75 km (30 to 50 ppmv) is slightly lower than that derived from ground-based microwave (55 ppmv) by Clancy et al. [2003]. However, there is a major discrepancy above 75 km: our measurements show a decreasing vmr with increasing altitude, while microwave measurements indicate a steady increase. We find a minimum of CO (at 10 ppmv around 85 km). At higher altitudes, we find a steady increase of the vmr with altitude up to the limit of our measurements (4,000 – 8000 ppmv at 125 km). This steady exponential increase, when extrapolated upward to 140 km, would give a mixing ratio of 1, while in situ measurements with the Pioneer Venus Bus Mass Spectrometer yielded 0.4

[von Zahn et al., 1980]. The increase of vmr from 90 km upward is not due to diffusive separation of CO from CO2, because the homopause as determined by N2 measurements is at an altitude of 136 km. Rather, the CO gradient is dominated by the production rate of CO from CO2 which increases with altitude. More puzzling is the strong minimum region (80 – 90 km) found by SOIR, which had escaped detection up to now. It should be recognized that the technique of solar occultation provides an unprecedented vertical resolution. It could be of dynamic origin or it could be the result of a strong chemical sink. 4.4. H2O and HDO [37] Water is scarcely present today and it is not yet known if Venus was already dry at its formation or evolved slowly to its present state. In order to refine theories and models describing the formation and evolution of Venus and its atmosphere, to characterize the escape of D atoms from the upper atmosphere and to provide a robust explanation to the problem of the origin of water on Venus, high-resolution vertically resolved measurements of H2O and HDO and their temporal variations are needed. As illustrated in Figure 1, SOIR is able to measure both isotopologues of water simultaneously during the same occultation. Three absorption lines owing to HDO are clearly seen in the spectra recorded in the 2720 – 2725 cm
1 region (order 121), whereas most of the features in the 3825 – 3855 cm
1 region

12 of 16

E00B23

VANDAELE ET AL.: VENUS MESOSPHERE COMPOSITION BY SOIR

E00B23

Figure 9. Vertical profiles of the HF density (left) and volume mixing ratio in ppb (right) obtained during the occultations 357 (solid line, 13 April 2007; distance to the limb at 65 km tangent height is 3325 km; 83°N), 462 (dotted line, 27 July 2007; distance to the limb is 1864 km; 88°N), and 484 (dashed line, 18 August 2007; distance to the limb is 2670 km; 70°N). For comparison, the solid line with dots corresponds to a profile of constant vmr (5 ppb). Error bars correspond to the fitting error on the single fitted parameter, i.e., the density (vmr) in the corresponding layer, and does not take into account the errors on the densities (vmr) in the upper layers. The following values from the literature are also indicated: Connes et al. [1967] (circles) and Bjoraker et al. [1992] (inverted triangles). (order 171) originate from the main isotopologue of water H2O and CO2. Such simultaneous measurements have been performed on a regular basis and are described in more detail by Fedorova et al. [2008]. 4.5. Other Minor Constituents [38] Spectra were recorded in different spectral intervals where some key constituents for the chemistry of the Venus atmosphere have prominent absorption features. However, some of those gases are only present as traces in the atmosphere and only upper limits of detection could be derived. As an example, OCS absorbs in the 2900–2950 cm
1 range, corresponding to the order 130, but spectra recorded with SOIR contain no signature of this species, as illustrated in Figure 11. The upper limit of detection for OCS has been estimated to be 1.6 ± 2 ppb between 70 and 90 km and 0.02 ± 0.01 ppm above 90 km. OCS has up to now never been observed above the cloud deck. Measurements performed at lower altitudes [Marcq et al., 2005; Marcq et al., 2006] indicate that the OCS shows latitudinal variations with abundances between 5 and 20 ppm at 30 km decreasing with altitude. At 36 km, OCS amounts to 0.55 ± 0.15 ppm. The same spectral interval could also be used to derive the upper limit of detection of H2CO leading to the values of 3 ± 2 ppb below 90 km and 0.015 ± 0.01 ppm above 90 km.

[39] Similarly, it is also possible to determine the upper limit of other trace gases. In the case of SO2, the procedure is somewhat more complex since the SO2 signature is overlapped by a band of CO2, as explained and discussed by Belayev et al. [2008], who found a positive detection of 0.3 to 3 ppmv in some occultation profiles. 4.6. Error Analysis [40] The most critical parameters influencing the retrieval are related to the definition of the AOTF function: its formulation, the position of its maximum, and its extension in wave numbers (number of orders taken into account in the forward model). To test the sensitivity of our retrieval toward these parameters, several test cases have been investigated: (1) the function has been approximated by a sinc2 function or by a more complex function resulting of the sum of 7 sinc2, with different positions of their maxima and widths; (2) the number of adjacent orders has been varied between 2 and 4; (3) the position of the function maximum has been shifted by 1.0 cm
1, a slightly higher value than the estimated accuracy on the determination of this position (0.83 cm
1) [Mahieux et al., 2008]. The reference test case corresponds to the complex AOTF function using 3 adjacent orders (simulation on 7 orders in total), which corresponds to the normal settings used for retrieval. The results of this sensitivity study are the

13 of 16

E00B23

VANDAELE ET AL.: VENUS MESOSPHERE COMPOSITION BY SOIR

E00B23

Figure 10. Vertical profiles of the CO density (left) and volume mixing ratio in ppm (right) obtained during the three occultations 341 (solid line, 28 March 2007; distance to the limb at 65 km tangent height is 3346 km; 82°N), 356 (dotted line, 12 April 2007; distance to the limb is 3298 km; 84°N), and 366 (dashed line 22 April 2007; distance to the limb is 3898 km; 73°N). For comparison, the solid line with dots corresponds to the CO in the VENUSREF model. Error bars correspond to the fitting error on the single fitted parameter, i.e., the density (vmr) in the corresponding layer, and does not take into account the errors on the densities (vmr) in the upper layers. following: (1) choosing a sinc2 function with 2 or 3 adjacent orders give rise to an underestimation of the CO2 density of 20 to 50% depending on the altitude, the maximum value

being obtained at around 90 km; (2) using the complex AOTF function but on only 2 adjacent orders implies an underestimation of 5 to 40%; (3) including more adjacent

Figure 11. SOIR spectra recorded around 2915 cm
1 (order 130) where OCS and H2CO signatures should be observed. The main absorption structures observed in this region are due to HCl. Absorption features due to OCS and H2CO are also shown. From such spectra, only upper limits of detection can be derived. 14 of 16

E00B23

VANDAELE ET AL.: VENUS MESOSPHERE COMPOSITION BY SOIR

orders leads to higher densities by 1 – 2%, but lengthens the time required for the forward modeling; (4) changing the position of the AOTF maximum implies a difference of 5 to 20%. From this discussion, it is clear that the most critical factor in the simulation or retrieval of SOIR spectra is the accurate definition of the AOTF function, in particular its width and the positions and intensities of its sidelobes. In the near future, significant efforts will be devoted to a better yet determination of these parameters from specifically dedicated in-flight calibration measurements.

5. Conclusions [41] The SOIR spectrometer, which is part of the SPICAV/ SOIR instrument on board Venus Express has proven its high potential for the detection of minor key species for the understanding of the chemical and dynamical processes occurring in the Venus mesosphere. Detection of CO, HCl, HF, H2O, and HDO has been confirmed at altitudes ranging from 65 to 105 km, even 125 km depending on the species. Measurements have shown that the instrument was also sensitive to temperature through its observations of CO2 absorption lines, although this will require future development of the retrieval algorithm. SOIR is also able to differentiate between three isotopologues of CO2, namely 12 16 16 C O O, 12C16O18O, and 12C16O17O. This will in fine provide the vertical distribution of the 17O/16O and 18O/16O isotopic ratios. Vertical profiles of HCl, HF, and CO in the mesosphere above the northern pole show low variability in time. The CO vertical profiles measured by SOIR indicate values of CO abundances from 10 to 50 ppm at 90 km with a pronounced minimum of 10 ppm at altitudes of the order of 85 km, followed by an increase toward the higher altitudes. HCl volume mixing ratio is about 0.1 – 0.2 ppm and that of HF decreases from 1 to 7 ppb showing some variability from one orbit to the other. Because of its high sensitivity and its wide spectral coverage, SOIR is also a good tool to determine upper limit of detection for a series of trace gases: for example, OCS upper limit of detection has been estimated to be 1.6 ± 2 ppb below 90 km and 0.02 ± 0.01 ppm above 90 km. [42] Acknowledgments. The research program was supported by the Belgian Federal Science Policy Office and the European Space Agency (ESA, PRODEX program, contracts C 90268, 90113, and 17645). Procurement of AOTF was funded by CNES. Russian team also acknowledges RFBR grant 06-02-72563.

References Adams, W. S., and T. Dunham (1932), Absorption bands in the infrared spectrum of Venus, Publ. Astron. Soc. Pac., 44, 243 – 247, doi:10.1086/ 124235. Belyaev, D., O. Korablev, A. Fedorova, J.-L. Bertaux, A.-C. Vandaele, F. Montmessin, A. Mahieux, V. Wilquet, and R. Drummond (2008), First observations of SO2 above Venus’ clouds by means of Solar Occultation in the Infrared, J. Geophys. Res., doi:10.1029/2008JE003143, in press. Bertaux, J.-L., D. Nevejans, O. Korablev, E. Villard, E. Que´merais, E. Neefs, F. Montmessin, F. Leblanc, J. P. Dubois, and E. Dimarellis (2007a), SPICAV on Venus Express: Three spectrometers to study the global structure and composition of the Venus atmosphere, Planet. Space Sci., 55, 1673 – 1700, doi:10.1016/j.pss.2007.01.016. Bertaux, J.-L., A. C. Vandaele, V. Wilquet, F. Montmessin, R. Dahoo, E. Villard, O. Korablev, and A. Fedorova (2007b), First observation of 628 CO2 isotopologue band at 3.3 mm in the atmosphere of Venus by solar occultation from Venus Express, Icarus, 195, 28 – 33, doi:10.1016/ j.icarus.2008.01.001.

E00B23

Be´zard, B., C. de Bergh, D. Crisp, and J. P. Maillard (1990), The deep atmosphere of Venus revealed by high-resolution nightside spectra, Nature, 345, 508 – 511, doi:10.1038/345508a0. Bjoraker, G. L., H. P. Larson, M. J. Mumma, R. Timmermann, and J. L. Montani (1992), Airborne observations of the gas composition of Venus above the cloud tops: Measurements of H2O, HDO, HF and the D/H and 18O/16O isotopic ratios, Bull. Am. Astron. Soc., 24, 995. Brown, L. R., C. M. Humphrey, and R. R. Gamache (2007), CO2-broadened water in the pure rotation and n 2 fundamental regions, J. Mol. Spectrosc., 246, 1 – 21, doi:10.1016/j.jms.2007.07.010. Clancy, R. T., and D. O. Muhleman (1991), Long-term (1979 – 1990) changes in the thermal, dynamical and compositional structure of the Venus mesosphere as inferred from microwave spectral line observations of 12CO, 13CO, and C18O, Icarus, 89, 129 – 146, doi:10.1016/00191035(91)90093-9. Clancy, R. T., B. Sandor, and G. H. Moriarty-Schieven (2003), Observational definition of the Venus mesopause: Vertical structure, diurnal variation, and temporal instability, Icarus, 161, 1 – 16, doi:10.1016/S00191035(02)00022-2. Clough, S. A., F. X. Kneizys, and R. Davies (1989), Line shape and the water vapor continuum, Atmos. Res., 23, 229 – 241, doi:10.1016/01698095(89)90020-3. Connes, P., J. Connes, W. S. Benedict, and L. Kaplan (1967), Traces of HCl and HF in the atmosphere of Venus, Astrophys. J., 147, 1230 – 1237, doi:10.1086/149124. Connes, P., J. Connes, L. Kaplan, and W. S. Benedict (1968), Carbon monoxide in the Venus atmosphere, Astrophys. J., 152, 731 – 743, doi:10.1086/149590. de Bergh, C., V. I. Moroz, F. W. Taylor, D. Crisp, B. Be´zard, and L. V. Zasova (2006), The composition of the atmosphere of Venus below 100 km altitude: An overview, Planet. Space Sci., 54, 1389 – 1397, doi:10.1016/j.pss.2006.04.020. Fally, S., P.-F. Coheur, M. Carleer, C. Clerbaux, R. Colin, A. Jenouvrier, M.-F. Me´rienne, C. Hermans, and A. C. Vandaele (2003), Water vapor line broadening and shifting by air in the 26000 – 13000 cm
1 region, J. Quant. Spectrosc. Radiat. Transfer, 82, 119 – 131. Fedorova, A., O. Korablev, D. Belayev, A. C. Vandaele, A. Mahieux, E. Neefs, V. Wilquet, F. Montmessin, and J.-L. Bertaux (2008), HDO and H2O vertical distribution and isotopic ratio in the Venus mesosphere by Solar Occultation at Infrared spectrometer on board Venus Express, J. Geophys. Res., doi:10.1029/2008JE003146, in press. Fomin, B. A. (1995), Effective interpolation technique for line-by-line calculations of radiation absorption in gases, J. Quant. Spectrosc. Radiat. Transfer, 53, 663 – 669. Gamache, R. R., R. L. Hawkins, and L. S. Rothman (1990), Total internal partition sums for atmospheric molecules in the temperature range 70 – 2005 K: Atmospheric linear molecules, J. Mol. Spectrosc., 142, 205 – 219, doi:10.1016/0022-2852(90)90178-S. Gamache, R. B., S. P. Neshyba, J.-J. Plateaux, A. Barbe, L. Re´galia, and J. B. Pollack (1995), CO2-broadening of water vapor lines, J. Mol. Spectrosc., 170, 131 – 151, doi:10.1006/jmsp.1995.1060. Gel’man, B., V. Zolotukhin, N. Lamonov, B. Levchuk, A. Lipatov, L. Mukhin, D. Nenarokov, V. Rotin, and B. Okhotnikov (1980), An analysis of the chemical composition of the atmosphere of Venus on an AMS of the Venera-12 using a gas chromatograph, Cosmic Res., Engl. Transl., 17, 585 – 589. Goody, R. M., and Y. L. Yung (1995), Atmospheric Radiation: Theoretical Basis, Oxford Univ. Press, New York. Gurwell, M., D. O. Muhleman, K. Shah, G. Berge, D. J. Rudy, and A. W. Grossman (1995), Observations of the CO bulge on Venus and implications for mesospheric winds, Icarus, 115, 141 – 158, doi:10.1006/ icar.1995.1085. Hedin, A. E., H. B. Niemann, and W. T. Kasprzak (1983), Global empirical model of the Venus thermosphere, J. Geophys. Res., 88, 73 – 83, doi:10.1029/JA088iA01p00073. Irwin, P. G. J., R. de Kok, A. Negra˜o, C. C. C. Tsang, C. F. Wilson, P. Drossart, G. Piccioni, D. Grassi, and F. W. Taylor (2008), Spatial variability of carbon monoxide in Venus’ mesosphere from Venus Express/Visible and Infrared Thermal Imaging Spectrometer measurements, J. Geophys. Res., 113, E00B01, doi:10.1029/2008JE003093. Iwagami, N., et al. (2008), Hemispheric distributions of HCl above and below the Venus’ clouds by ground-based 1.7 mm spectroscopy, Planet. Space Sci., 56, 1424 – 1434. Jenouvrier, A., L. Daumont, L. Re´galia-Jarlot, V. G. Tyuterev, M. Carleer, A. C. Vandaele, S. Mikhailenko, and S. Fally (2007), Fourier transform measurements of water vapor line parameters in the 4200 – 6600 cm
1 region, J. Mol. Spectrosc., 105, 326 – 355. Mahieux, A., et al. (2008), In-flight performance and calibration of SPICAV SOIR on board Venus Express, Appl. Opt., 47(13), 2252 – 2265, doi:10.1364/AO.47.002252.

15 of 16

E00B23

VANDAELE ET AL.: VENUS MESOSPHERE COMPOSITION BY SOIR

Marcq, E., B. Be´zard, T. Encrenaz, and M. Birlan (2005), Latitudinal variations of CO and OCS in the lower atmosphere of Venus from nearinfrared nightside spectro-imaging, Icarus, 179, 375 – 386, doi:10.1016/ j.icarus.2005.06.018. Marcq, E., T. Encrenaz, B. Be´zard, and M. Birlan (2006), Remote sensing of Venus’ lower atmosphere from ground-based IR spectroscopy: Latitudinal and vertical distribution of minor species, Planet. Space Sci., 54, 1360 – 1370, doi:10.1016/j.pss.2006.04.024. Me´rienne, M.-F., A. Jenouvrier, C. Hermans, A. C. Vandaele, M. Carleer, C. Clerbaux, P.-F. Coheur, R. Colin, S. Fally, and M. Bach (2003), Water vapor line parameters in the 13000 – 9250 cm
1 region, J. Quant. Spectrosc. Radiat. Transfer, 82, 99 – 118. Mueller-Wodarg, I., and S. Tingle (2008), Vertical structure above the clouds: Towards a unified model, paper presented at XLIII Rencontres de Moriond, Venus Express Scientific Workshop, Eur. Space Agency, La Thuile, Italy, 3 – 7 Mar. Nevejans, D., et al. (2006), Compact high-resolution space-borne echelle grating spectrometer with AOTF based on order sorting for the infrared domain from 2.2 to 4.3 micrometer, Appl. Opt., 45(21), 5191 – 5206, doi:10.1364/AO.45.005191. Oyama, V., G. Carle, F. Woeller, J. B. Pollack, R. Reynolds, and R. Craig (1980), Pioneer Venus gas chromatography of the lower atmosphere of Venus, J. Geophys. Res., 85, 7891 – 7902, doi:10.1029/JA085iA13p07891. Pa¨tzold, M., et al. (2007), The structure of Venus’ middle atmosphere and ionosphere, Nature, 450, 657 – 660, doi:10.1038/nature06239. Pollack, J. B., et al. (1993), Near-infrared light from Venus’ nightside: A spectroscopic analysis, Icarus, 103, 1 – 42, doi:10.1006/icar.1993.1055. Rothman, L. S., et al. (2005), The HITRAN 2005 molecular spectroscopic database, J. Quant. Spectrosc. Radiat. Transfer, 96, 139 – 204. Seiff, A., J. T. Schofield, A. Kliore, F. W. Taylor, S. S. Limaye, H. E. Revercomb, L. A. Sromovsky, V. V. Kerzhanovich, V. I. Moroz, and M. Y. Marov (1985), Models of the structure of the atmosphere of Venus

E00B23

from the surface to 100 kilometers altitude, Adv. Space Res., 5(11), 3 – 58, doi:10.1016/0273-1177(85)90197-8. Sung, K., and P. Varanasi (2005), CO2-broadened half-widths and CO2induced line shifts of 12C16O relevant to the atmospheric spectra of Venus and Mars, J. Quant. Spectrosc. Radiat. Transfer, 91, 319 – 322, doi:10.1016/j.jqsrt.2004.05.063. Titov, D. V., et al. (2006), Venus Express science planning, Planet. Space Sci., 54, 1279 – 1297, doi:10.1016/j.pss.2006.04.017. Toth, R. A., and L. A. Darnton (1974), Linewidths of HCl broadened by CO2 and N2 and CO broadened by CO2, J. Mol. Spectrosc., 49, 100 – 105, doi:10.1016/0022-2852(74)90099-X. Vandaele, A. C., M. Kruglanski, and M. De Mazie`re (2006), Simulation and retrieval of atmospheric spectra using ASIMUT, paper presented at Atmospheric Science Conference, Eur. Space Agency, Frascati, Italy. von Zahn, U., K. H. Fricke, D. M. Hunten, D. Krankowsky, K. Mauersberger, and A. O. Nier (1980), The upper atmosphere of Venus during morning conditions, J. Geophys. Res., 85, 7829–7840, doi:10.1029/JA085iA13p07829. Wilquet, V., A. Mahieux, A. C. Vandaele, V. Perevalov, S. Tashkun, A. Fedorova, O. Korablev, F. Montmessin, R. Dahoo, and J.-L. Bertaux (2007), Line parameters for the 01111 – 00001 band of 12C16O18O from SOIR measurements of the Venus atmosphere, J. Quant. Spectrosc. Radiat. Transfer, 109, 895 – 905. Young, L. (1972), High resolution spectra of Venus: A review, Icarus, 17, 632 – 658, doi:10.1016/0019-1035(72)90029-2.























D. Belyaev, A. Fedorova, and O. Korablev, Space Research Institute, 84/32 Profsoyuznaya Street., 117997, Moscow, Russia. J.-L. Bertaux and F. Montmessin, Service d’Ae´ronomie du CNRS, BP3, 91371, Verrie`res-le-Buisson, France. R. Drummond, A. Mahieux, M. De Mazie`re, E. Neefs, A. C. Vandaele, and V. Wilquet, Belgian Institute for Space Aeronomy, 3 avenue Circulaire, B-1180 Brussels, Belgium. ([email protected])

16 of 16

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, E00B22, doi:10.1029/2008JE003146, 2008

for

Full Article

HDO and H2O vertical distributions and isotopic ratio in the Venus mesosphere by Solar Occultation at Infrared spectrometer on board Venus Express A. Fedorova,1 O. Korablev,1 A.-C. Vandaele,2 J.-L. Bertaux,3,4 D. Belyaev,1 A. Mahieux,2 E. Neefs,2 W. V. Wilquet,2 R. Drummond,2 F. Montmessin,3,4 and E. Villard3,4 Received 18 March 2008; revised 13 June 2008; accepted 26 August 2008; published 25 December 2008.

[1] Vertical distributions of the molecular density and mixing ratios of H2O and HDO

in the Venus mesosphere have been obtained using Solar Occultation at Infrared (SOIR), a high-resolution (with l/dl
20,000) echelle spectrometer on Venus Express. The atmosphere is sounded in solar occultation in the range of altitudes from 65 to 130 km. Simultaneous measurements of water vapor lines in the spectral range around 2.61 mm (3830 cm1) at altitudes between 70 and 110 km and HDO lines around 3.58 mm (2715cm1) at altitudes 70–95 km have been performed. During 1 1/2 years, from April 2006 to August 2007, 54 such measurements have been carried out at different locations of Venus from the north pole to middle south latitudes. Most of the observations at morning and evening terminator correspond to high northern latitudes. We report values of mixing ratio and isotopic ratio obtained for 22 of those measurements occurring in the northern polar area. The average value of the volume mixing ratio of H2O is 1.16 ± 0.24 ppm and that of HDO is 0.086 ± 0.020 ppm. A depletion in the mixing ratio for both H2O and HDO is observed at 85 km, which can be related to a depletion of CO2 density above (
95 km) and a possible temperature inversion at these altitudes. The vertical variation of HDO and H2O mixing ratio is within a factor of 2–3 for the analyzed set of observations. The temporal variations have been investigated, and no noticeable variability of H2O is reported at high northern altitudes. The average ratio of HDO/H2O obtained in this work, 240 ± 25 times the terrestrial ratio, is higher (1.5 times) than the value of 157 ± 30 times terrestrial reported for the lower atmosphere. This could be explained by a lower photodissociation of HDO and/or a lower escape rate of D atoms versus H atoms. Citation: Fedorova, A., et al. (2008), HDO and H2O vertical distributions and isotopic ratio in the Venus mesosphere by Solar Occultation at Infrared spectrometer on board Venus Express, J. Geophys. Res., 113, E00B22, doi:10.1029/2008JE003146.

1. Introduction [2] Venus’ atmosphere is significantly drier than the atmosphere of the Earth: it contains from 1 to 100 ppm of water depending on altitude and location. As reviewed by de Bergh et al. [2006], the water vapor content measured below the cloud level is within 20– 30 ppm, with much larger controversy at the cloud tops. [3] Most of the H2O mesospheric observations were obtained from ground-based millimeter heterodyne spectroscopy [Encrenaz et al., 1991, 1995; Sandor and Clancy, 2005]. Using the HDO line at 225.9 GHz and assuming a 1

Space Research Institute, Moscow, Russia. Belgian Institute for Space Aeronomy, Brussels, Belgium. 3 Service d’Ae´ronomie du CNRS, Verrie`res-le-Buisson, France. 4 Also at Institut Pierre Simon Laplace, Universite´ de Versailles Saint Quentin en Yvelines, Guyancourt, France. 2

Copyright 2008 by the American Geophysical Union. 0148-0227/08/2008JE003146$09.00

D/H enrichment of 120 with respect to the terrestrial value, Encrenaz et al. [1991] obtained a mixing ratio of 3.5 ± 2.0 ppm at 60– 95 km. Later, observing the 187.31 GHz transition of H2O, Encrenaz et al. [1995] obtained 1 ppm above the clouds and 7 ppm (+5.0, 4.0) from the HDO line at 225.9 GHz, also assuming D/H = 120 terrestrial value. From a series of microwave observations, Sandor and Clancy [2005] reported a strong global variability of H2O on a 1 – 2 month time scale: at 65– 100 km the mixing ratio ranged from 0 ± 0.06 to 3.5 ± 0.3 ppm. Gurwell et al. [2007] also reported an extreme variability of the Venus mesosphere dramatically demonstrated by submillimeter wave astronomy satellite (SWAS) observations from December 2002. Over the course of 5 days, a deep ground-state water absorption feature consistent with a water abundance of 4.5 ± 1.5 ppm suddenly has transformed into a significantly shallower absorption, implying a decrease in the water abundance by a factor of 50 in less than 48 h.

E00B22

1 of 16

E00B22

FEDOROVA ET AL.: HDO AND H2O IN VENUS’ MESOSPHERE BY SOIR

[4] The isotopic ratio of HDO to H2O in standard mean ocean water (SMOW) equals 3.1153  104 and corresponds to ([D]/[H])smow = 1.5576  104 [Hagemann et al., 1970]. On Venus, first measurements of the D/H ratio were performed by the Pioneer Venus mass spectrometer LNMS in the low atmosphere (100 (±12.5) times terrestrial, corrected in 1997 to 157 (±30) times terrestrial) [Donahue et al., 1982, 1997] and the ion mass spectrometer on the Pioneer Venus Orbiter at 155– 160 km ([D+]/[H+] = (1.7 ± 0.6)  102) or [D]/[H] = (2.2 ± 0.6)  102) at the turbopause level of about 132 km) [Hartle and Taylor, 1983]. These values have been confirmed later by ground-based observations in the 2.3 mm nightside windows [de Bergh C. et al., 1991]. Both in situ and remote-sensing determinations agree with one other and point to a strong deuterium enrichment in the Venus’ atmosphere. [5] In the near IR range above the clouds only one measurement of H2O and HDO has been reported [Bjoraker et al., 1992]. High-resolution spectra were obtained between 2.59 and 2.65 mm (3860 – 3770 cm1). A volume mixing ratio of H2O of 2.09 ± 0.15 ppm at 72 km was reported and a D/H ratio of 157 ± 15 times the value in the terrestrial ocean. Unfortunately, these results have never been published in any refereed literature. [6] The enrichment of
150 of deuterium to hydrogen supports the idea of dramatic escape of water from Venus. Two scenarios are possible: either Venus has had at least 0.3% of the terrestrial ocean and lost its water during a catastrophic process, or Venus was dry from the beginning, and the present isotopic ratio is explained by cometary impacts, degassing, and escape processes. The unknown vertical distribution of H2O in the middle and upper atmosphere precludes accurate theoretical descriptions of the escape of D and H atoms. New measurements of HDO and H2O are necessary to understand the evolution of the Venusian climate. [7] SOIR is a part of the SPICAV/SOIR experiment on the Venus Express spacecraft, operating on the orbit around Venus from April 2006 [Titov et al., 2006]. It is a high-resolution IR spectrometer working in the range of 2.2– 4.3 mm [Bertaux et al., 2007a; Nevejans et al., 2006]. The experiment is dedicated to vertical sounding of the Venus’ mesosphere in the range of 60– 120 km by means of solar occultation. One of the main scientific goals of SOIR is a simultaneous measurement of H2O and HDO vertical profiles to retrieve the isotopic ratio in the mesosphere. The first results on water vapor measurements with SOIR have been reported by Bertaux et al. [2007b]. In this paper we present a new analysis of the observed H2O content, employing a better instrument calibration and using longer series of observations.

2. Measurements 2.1. Instrument Description [8] Echelle spectrometer SOIR with an acoustooptic filtration of light is the first instrument with a spectral resolution above 20,000 orbiting another planet. It was first proposed for Venus Express mission by Korablev and Bertaux [2002], and built at the Belgian Institute for Space Aeronomy in collaboration with a Belgian industry in a very short time.

E00B22

[9] SOIR is designed to operate during solar occultations, when the instrument entrance optics is pointed toward the Sun as the latter goes down or up, allowing measurement of the atmospheric transmission of different layers of the atmosphere. The spectral range of SOIR from 2.32 to 4.25 mm (2353– 4310 cm1) allows measuring a number of atmospheric constituents showing up as absorbers in the transmission spectrum. High dispersion is provided by a 4 grooves mm1, arctan(2) incidence angle, echelle grating operating in diffraction orders from 101 (corresponding to the wavelengths of 4.4 mm) to 194 (2.3 mm). Each diffraction order covers a spectral interval from 20 to 40 cm1. The diffraction orders are separated by an acoustooptic tunable filter (AOTF). In the AOTF the radiation is filtered owing to volume diffraction on the acoustooptic wave excited within a birefringent crystal by a piezoelectric transducer at high frequency (RF = 14– 26MHz). The diffracted light is then analyzed by the echelle spectrometer. The central wavelength of the AOTF band-pass function is determined for the RF applied from the frequency-wavelength calibration as described by Mahieux et al. [2008]. The full width at half maximum FWHM of the AOTF band pass is around 24 cm1, and the profile of the AOTF band-pass function is close to a (sinx/x)2 function with a number of significant side lobes. As a result, spectral features leaking from several adjacent diffraction orders contribute to the spectrum observed on the detector with different weights that will be discussed in section 3. [10] The SOIR detector (from SOFRADIR) has 320 columns oriented along the spectral dispersion (wavelengths) and 256 rows along the spectrometer’s slit (spatial dimension). To avoid saturation, short integration times are used (20 to 30 ms), depending on the wavelength at which the measurement is taken. The background signal (Dark Current + Thermal emission of optics) is measured and subtracted onboard. In order to improve the signal-to-noise ratio (SNR), a number of measurements can be accumulated as long as the total measuring time remains below 250 ms. [11] The slit height is 30 arc min, and it is projected onto 32 rows of the detector. The slit width is equivalent to two detector pixels in the spectral direction. The spectral resolution of the spectrometer is very high and equals
0.13 cm1 at 2500 cm1 (order 111) and 0.27 cm1 at 4300 cm1 (order 192). The corresponding resolving power [l/Dl = n/Dn] is
20,000 [Mahieux et al., 2008]. [12] Owing to telemetry limitations, only eight spectra, each of 320 pixels long can be downloaded per second. During most of observations on Venus orbit, these eight spectra are taken in four different diffraction orders (different tunings of the AOTF), each corresponding to two large bins of 16 or 12 rows on the detector. [13] The detailed description of the instrument can be found by Nevejans et al. [2006] and Bertaux et al. [2007a]. Calibrations and in-flight performances of the instrument, including data handling, on board background subtraction, calibrations of the AOTF and the echelle spectrometer are described in detail by Mahieux et al. [2008]. 2.2. Observations [14] For accurate measurements of the isotopic ratio, simultaneous observations of H2O and HDO absorption lines are required. As discussed above, only four 20– 30 cm1

2 of 16

E00B22

FEDOROVA ET AL.: HDO AND H2O IN VENUS’ MESOSPHERE BY SOIR

E00B22

Figure 1. Synthetic spectra of H2O (green), HDO (blue), and CO2 (red) at the target altitude of 85 km in solar occultation geometry. The top corresponds to the spectral interval 3760– 3940 cm1 (2.6 mm) dedicated to H2O measurements; the bottom shows the spectral interval 2660 –2840 cm1 (3.6 mm) dedicated to HDO measurements. Order boundaries are shown as solid (beginning) and dashed (end) black lines. The numbers above each plot show the order number and are located at the center of the order. Orders 171 and 121 are used mainly for retrieval of H2O and HDO abundance, respectively. portions of the spectrum within the spectral range of SOIR (four diffraction orders) can be acquired during a single occultation. We therefore carefully chose the spectral range for H2O and HDO detection attributing at least one order for HDO and another one for H2O. The priority was given to spectral ranges with strong lines of the gases of interest, and minimal contamination from other gases. There are several strong transitions of H2O within the spectral range of SOIR. The most preferable for the H2O retrieval is the strongest 3600 – 3900 cm1 band, which includes lines of n 1 and n 3 fundamental transitions. Below 3750 cm1 a strong CO2 2.7 mm band prevents the accurate determination of H2O, being completely saturated below 85 km of tangent altitude in solar occultation geometry. [15] For HDO we can consider the two strongest transitions within SOIR spectral range: the fundamental bands n 3 (001– 000) near 3707 cm1 (2.69 mm) and n 1 (100– 000) near 2723 cm1 (3.67 mm). The first one is blended by the 2.56 mm band of the main H2O isotope, and the HDO transitions are weak. The best candidate is the strong fundamental HDO band n 1 (100 – 000), completely isolated from H2O transitions. Moreover, this spectral range is free from detectable CO2 lines. Considering the other molecules, HCl lines are also located in the range of 3.67 mm. But these lines are isolated and do not contaminate HDO lines. [16] Synthetic spectra of Venus atmosphere limb transmission expected at the tangent altitude of 90 km, showing separately H2O and HDO lines, and all isotopologues of CO2 are presented in Figure 1 for the selected ranges

discussed above. The diffraction orders of SOIR are indicated. The orders numbers from 170 to 172 and 174 have been used for H2O measurements. Most of the observations so far have been performed in the order 171. The strongest lines of H2O in this range correspond to the rotational structures of (001) – (000) transition. For the retrieval of the HDO density, the orders 121 and 125 have been chosen, most of the observations having been performed in the order 121. 2.3. Observation Coverage [17] We have considered the operations from April 2006 to August 2007. During this period 54 orbits have been dedicated specifically to simultaneous measurements of H2O and HDO in the Venus’ mesosphere. Among them, on 47 orbits an additional spectral range dedicated to CO2 was scanned simultaneously in order to measure independently the atmospheric density profile and to reduce spacecraft pointing uncertainties. We address an interested reader to the paper by Vandaele et al. [2008, Figure 1] for the illustration of transmission evolution during an occultation in these diffraction orders. Vandaele et al.’s Figure 1 gives an example of the evolution of one occultation (sunset 15 April 2007) in the order 121, 171, 149, and 190. It is a typical observation dedicated to H2O, HDO, and CO measurements. Order 149 has been chosen for CO2 retrieval in the range of 3330 – 3357 cm1 where strong P and Q branches of 21102– 00001 transition allow the measurement of the atmospheric density at the altitudes of 70– 120 km.

3 of 16

E00B22

FEDOROVA ET AL.: HDO AND H2O IN VENUS’ MESOSPHERE BY SOIR

E00B22

Figure 2. Spatial coverage for the simultaneous measurements of H2O and HDO. The maps of longitude versus latitude and local time versus latitude are presented. Most of the observations are located at high northern latitudes, corresponding to low distances from spacecraft to the limb. It reflects the peculiarity of the Venus Express polar orbit with the pericenter near the north pole. [18] The list of the main CO2 bands used for retrieval of atmospheric density is presented by Vandaele et al. [2008, Table 1]. In case of H2O and HDO, for the retrieval of the CO2 density several orders (depending on the orbit) have been chosen: 111, 112 (isotope O16C12O18, spectral range 2470 – 2520 cm1), 123 (spectral range 2748 – 2773 cm1) and 148, and 149 (a transition 21102 – 00001 at 3285 – 3360 cm1). [19] The geographical distribution of observations for which H2O and HDO were measured simultaneously, and the dependence on local time are presented in Figure 2. [20] Venus Express spacecraft is on highly elongated orbit with a pericenter located at about 250 km near the north pole, and an apocenter at 65000 km above the southern hemisphere of Venus [Titov et al., 2006]. From such an orbit, most of the solar occultations are observed at high northern latitudes with smaller distance to limb (near the pericenter), another part corresponding to relatively large distances to the limb and low latitudes. The vertical resolution of the occultation depends directly on the distance to the limb. The attitude of the spacecraft is maintained so that the slit of SOIR would be nearly parallel to the limb at 65 km. Near the pericenter the vertical resolution is generally better than 1.5 km. But because of the rotation of the slit with respect to the limb during one occultation and as the distance to limb is changing with the spacecraft motion, the vertical resolution during one observation usually varies from several hundred meters to several kilometers. This effect is described in detail in the paper of Vandaele et al. [2008, section 3]. In the present study we have considered only the measurements taken near the pericenter with small distances to the limb in order to minimize uncertainties. About 35 H2O measurements have been performed near the pericenter with the distance to the limb below 6000 km. All of them are located close to the north pole. We have further constrained the analyzed set to identical spectral ranges. We concentrated on orbits where the orders numbers 171 and 121 have been used for H2O and HDO respectively. In all, 22 orbits have been analyzed. The list of reported observations including universal time, latitude, longitude, local time, and distance to the limb at the altitude of 80 km is presented in Table 1. Within the considered set of observations which includes both sunrises

and sunsets, latitudes vary from 60° to 86° N, and the local time is 5 –6 h and 17– 19 h.

3. Retrieval Process [21] In solar occultation the transmission spectrum is directly obtained from relative measurements, and the photometric calibration of the instrument is generally not required. The processing of raw SOIR data, including nonlinearity and other corrections up to the construction of transmission spectra, and spectral calibration is described by Mahieux et al. [2008]. The reference spectrum is obtained as a weighted average of solar spectra recorded outside of the atmosphere (at the altitudes from 160 to 200 km) [Vandaele et al., 2008]. A forward model of gas absorption and the retrieval of vertical profiles of atmospheric constituents based on ‘‘onion-peeling’’ method are described in detail by Vandaele et al. [2008]. In the present paper we discuss only the details related specifically to H2O and HDO spectroscopy. 3.1. Spectroscopic Data [22] As a basis, the HITRAN 2004 database [Rothman et al., 2005] with the update of 2006 for H2O has been used. Nevertheless, no significant modification in the spectral range of interest was found in the 2006 update. Among the spectroscopic data like line strengths and positions, temperature coefficients and pressure shifts, HITRAN database contains line widths for air and self broadening. The widths of spectral lines broadened by CO2 which is the main component of the Venus’ atmosphere (96.3%) are not well known. The first measurements of CO2 broadening for H2O lines have been done by Howard et al. [1956a, 1956b, 1956c, 1956d]. This work recommended a constant factor of 1.3 to translate from the air to CO2 broadening. This factor has been then widely used in the 1990s for investigation of the Venus’ atmosphere [Pollack et al., 1993]. Later accurate laboratory measurements of CO2 broadening in the near-infrared range were made by Gamache et al. [1995]. The authors compared their laboratory measurements of CO2-broadened half widths with some recent experimental work and theoretical calcu-

4 of 16

FEDOROVA ET AL.: HDO AND H2O IN VENUS’ MESOSPHERE BY SOIR

E00B22

E00B22

Table 1. Venus Express Orbits During Which Simultaneous Observations of H2O, HDO, and CO2 Have Been Performed With Low Distances to the Limb, Near the Pericenter of an Orbita Orbit

Date

Obs

Longitude

Latitude

LT

Ls

DTL

O1

O2

O3

O4

244 247 251 255 262 345 347 349 358 434 435 438 440 442 443 445 447 456 462 471 486 487

21 Dec 2006 24 Dec 2006 28 Dec 2006 1 Jan 2007 8 Jan 2007 1 Apr 2007 3 Apr 2007 5 Apr 2007 14 Apr 2007 29 Jun 2007 30 Jun 2007 3 Jul 2007 5 Jul 2007 7 Jul 2007 8 Jul 2007 10 Jul 2007 12 Jul 2007 21 Jul 2007 27 Jul 2007 5 Aug 2007 20 Aug 2007 21 Aug 2007

IN IN IN IN IN E E E E IN IN IN IN IN IN IN IN IN IN IN IN IN

248.91 256.82 261.13 268.06 282.80 13.69 36.86 81.05 194.10 264.07 266.74 274.52 279.49 284.24 286.51 290.81 294.66 298.02 254.94 232.25 251.33 253.47

84.80 84.39 82.09 79.19 70.76 85.93 87.34 88.13 82.30 70.83 72.54 76.11 77.84 79.30 79.97 81.18 82.31 86.52 87.68 79.49 66.41 63.01

19.96 19.88 19.34 18.99 18.54 7.56 8.69 11.23 16.90 5.88 5.85 5.76 5.68 5.59 5.53 5.41 5.26 3.64 23.55 18.98 18.42 18.36

63.08 67.82 74.14 80.46 91.52 224.28 227.52 230.76 245.36 7.99 9.59 14.36 17.54 20.72 22.31 25.49 28.67 42.93 52.39 76.02 90.16 91.78

2160.52 1558.68 1627.02 1729.67 2111.65 3315.49 3293.40 3281.31 3434.61 2060.91 1980.57 1848.39 1805.21 1777.70 1768.19 1758.40 1751.90 1780.27 1833.11 2088.11 2766.10 3001.76

121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121 121

171 171 171 171 171 171 171 171 171 171 171 171 171 171 171 171 171 171 171 172 171 171

111 149 112 148 111 149 149 149 149 111 112 149 118 149 111 149 111 112 149 123 149 149

112 192 181 191 133 133 133 190 133 125 190 132 151 184 188 130 136 145 180 180 190 190

H2O-1 (ppm) 1.08 ± 1.38 ± 1.00 ± 0.89 ± 1.02 ± 0.84 ± 0.84 ± 0.80 ± 0.75 ± 1.28 ± 1.14 ± 1.54 ± 0.96 ± 1.47 ± 1.22 ± 1.66 ± 1.22 ± 1.24 ± 1.60 ± 0.94 ± 1.54 ± 1.11 ±

0.10 0.13 0.10 0.05 0.06 0.06 0.04 0.05 0.05 0.11 0.03 0.10 0.09 0.05 0.07 0.10 0.05 0.04 0.09 0.12 0.10 0.05

H2O-2 (ppm) 1.24 1.36 1.28 1.03 1.08 0.68 0.74 0.63 0.64 1.15 1.21 1.50 1.25 1.35 1.20 1.53 1.22 1.26 1.59 1.36 1.31 0.99

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

0.13 0.17 0.05 0.05 0.07 0.06 0.04 0.04 0.05 0.12 0.03 0.15 0.05 0.07 0.07 0.14 0.04 0.02 0.13 0.04 0.18 0.05

HDOb (ppm) 0.097 ± 0.010 0.098 ± 0.012 0.092 ± 0.004 0.081 ±0.003 0.086 ± 0.004 0.056 ± 0.007 0.070 ± 0.014 0.054 ± 0.006 0.057 ± 0.011 0.075 ± 0.011 0.084 ± 0.002 0.106 ± 0.010 0.084 ± 0.003 0.091 ± 0.005 0.087 ± 0.004 0.106 ± 0.010 0.088 ± 0.006 0.086 ± 0.003 0.106 ± 0.011 0.118±0.004 0.081 ± 0.006 0.079 ± 0.005

a Latitude, longitude, and local time are given for the altitude of 80 km. Obs can be sunrise as E or sunset as IN; LT is the local time on Venus in Venusian hours; Ls is the solar longitude; DTL is the distance to the limb in km; O1, O2, O3, and O4 are the number of orders observed during the orbit; H2O-1 is the H2O mixing ratio in ppm averaged for altitudes between 75 and 112 km; H2O-2 is the H2O mixing ratio in ppm averaged for altitudes between 75 and 95 km; and HDO is the HDO mixing ratio in ppm averaged for altitudes between 75 and 95 km.

lations, and investigated the resulting temperature dependence of the half widths. The scaling factor from air broadening varies from 1.3 to 2.0, depending on a particular transition. Recently, Brown et al. [2007] published the CO2-broadened water parameters (half width, line shift, and temperature dependence of the widths) for the transitions between 200 and 900 cm1. They obtained ratios of CO2broadened to N2-broadened widths which varied widely from 0.95 to 3.07 with an average ratio of 1.67. To account for the CO2 broadening of lines in the Venusian atmosphere, we have multiplied the air-broadened half widths from the HITRAN 2004 database by 1.7 [Gamache et al., 1995; Brown et al., 2007]. The HDO line strengths in the HITRAN are scaled to the isotopic ratio of HDO to H2O in standard mean ocean water (SMOW) (3.1115  104). 3.2. Forward Modeling of H2O and HDO Transmittance Spectra [23] Using the forward model [Vandaele et al., 2008], the synthetic spectra of water vapor and HDO are calculated for the geometry of the solar occultation. Figures 3 and 4 illustrate the synthetic models for HDO in the order 121 (2703 – 2727 cm1) and for H2O in the order 171 (3823 – 3853 cm1) at a tangential altitude of 90 km. CO2 absorption lines have also been considered in those simulations. Atmospheric temperature and pressure profiles were taken from the Venus International Reference Atmosphere (VIRA) [Keating et al., 1985] for the dayside. H2O and HDO are assumed uniformly mixed with the volume mixing ratio of 1 ppm and 100 ppb, respectively. Monochromatic spectra have been converted to SOIR spectral resolution. Figures 3 and 4 (top) show modeling of gaseous absorption in ‘‘clear’’ case assuming there is no order mixing on detector. [24] The contribution of adjacent diffraction orders complicates the spectra measured by SOIR with respect to the

synthetic model shown in Figures 3 and 4 (top). The mixing of diffraction orders occurs owing to a wider than specified AOTF band-pass function (FWHM of 25 cm1), and the side lobes of this function (see discussion in the section 3.1 of Vandaele et al. [2008]). Figure 5 clearly demonstrates the contribution of different orders (actually ±3 orders) around the main order 121. Moreover, as described by Mahieux et al. [2008], the shape of the AOTF band-pass function is different for the beams coming from different parts of the spectrometer’s slit. As a result the contribution of adjacent orders is different for bins 1 and 2 recorded simultaneously during an observation (see section 2.1), where bin1 and bin 2 correspond to the top and bottom parts of the slit. [25] We reconstruct SOIR spectra in the 121 and 171 diffraction orders for bins 1 and 2 taking into account the overlapping of three orders from each side in Figures 3 and 4 (middle and bottom). In all, the seven diffraction orders contribute into the shown transmission spectra (from 3 to +3). [26] With the mixing of orders the 121 diffraction order, which is originally free from CO2 lines, becomes contaminated by CO2 lines coming from adjacent orders 122– 123 (compare Figure 1). This emphasizes the need of high accurate AOTF function taken into account more orders. [27] The mixing of diffraction orders becomes even more important in the order 171 (2.6 mm) located close to a strong 2.7 mm CO2 band. In Figure 5 we compare the synthetic model assuming the contribution of three diffraction orders from each side (Figure 5a) with a model assuming contribution of six orders (Figure 5b) with SOIR spectrum from the orbit 462 recorded at the altitude of 107 km. At this altitude the signal is not decreased by aerosol extinction, and the signal-to-noise of SOIR for order 171 between 100 and 290 pixels is better than 1000. It allows detecting very weak absorption features with a relative depth below 0.1–

5 of 16

E00B22

FEDOROVA ET AL.: HDO AND H2O IN VENUS’ MESOSPHERE BY SOIR

E00B22

Figure 3. Synthetic models of HDO and CO2 in the order 121 (2703 – 2727 cm1). Frequency preset on the acoustooptic tunable filter (AOTF) is 15822.62 kHz. Solar occultation geometry is considered with the tangential altitude of 90 km. The VIRA atmospheric model was used. Volume mixing ration of HDO was assumed to be equal to 100 ppb. Monochromatic spectra have been converted to SOIR resolution. The thick solid line corresponds to HDO, and the thin solid line corresponds to CO2. H2O lines are also taken into account (dashed thin line) but are weak in this range. Top shows a ‘‘clear’’ spectrum of gaseous absorption in the range of 2703 – 2727 cm1 without any order mixing. Middle shows a synthetic spectrum taking into account the mixing of 7 orders (from 3 to +3) for bin 1 (spectra obtained in the bottom part of the slit). Bottom shows a synthetic spectrum taking into account the mixing of 7 orders (from 3 to +3) for bin 2 (spectra obtained in the top part of the slit). The different calibration of frequency-wavelength functions for different bins (or rows on the matrix) introduces differences in the spectra which correspond to different rows of the detector. 0.2%. Figure 5 demonstrates that the structures in order 171 are not a noise but CO2 lines coming from the orders 165 to 167. These orders correspond to the third side lobe of the AOTF function which has an amplitude of only 1% of the central lobe. However, the strong absorption in the CO2 2.7 mm band (up to 50%) results in a quasi-chaotic structure with relative depths of 0.1 – 0.3% inside the order 171, contaminating the measured H2O spectrum. [28] Although the AOTF band pass function is not well constrained for such distant side lobes, taking into account six adjacent orders allows to reduce the residual error. We therefore take into consideration the contribution from ±6 orders in all further modeling of H2O and HDO spectra in orders 121 and 171.

3.3. Fitting [29] The local densities of considered molecules (CO2, HDO, and H2O) are retrieved using the traditional onionpeeling method [e.g., Rodgers, 2000]. The main idea of this method in application to SOIR data is described by Vandaele et al. [2008] (see section 3.3 and Figure 4). The local densities of molecules are retrieved from the top of the atmosphere downward. The retrieved quantities are used for simulating the spectra corresponding to the lower layers. To calculate the absorption cross sections of the selected molecules, the atmospheric temperature-pressure profile was taken from VIRA [Keating et al., 1985]. [30] The measured spectra are fitted to the model using minimization of c2 and the simplex algorithm [Press et al., 1992] with two variables: local density of molecules (H2O,

6 of 16

E00B22

FEDOROVA ET AL.: HDO AND H2O IN VENUS’ MESOSPHERE BY SOIR

Figure 4. Synthetic models of H2O and CO2 in the order 171 (3823– 3853 cm1). The frequency preset on the AOTF is 22946.44 kHz. Solar occultation geometry is considered with the tangential altitude of 90 km. VIRA atmospheric model was used. Volume mixing ratios of H2O were assumed to be equal to 1 ppm. Monochromatic spectra have been converted to the SOIR resolution. The thick solid line corresponds to H2O, and the thin solid line corresponds to CO2. HDO lines are also taken into account (dashed thin line) but are weak in this range. The presentation of different top, middle, and bottom is the same as for Figure 3.

Figure 5. Contamination of adjacent orders in the main order 171 used for H2O retrieval. (a) ±3 orders around the main order have been taken into account. (b) ±6 orders around the main order have been taken into account. Those data were obtained during orbit 462 (27.07.2007) at the location latitude 87°N, longitude 245°E, 23.55 h Venusian local time (LT), distance to the limb 1830 km. 7 of 16

E00B22

FEDOROVA ET AL.: HDO AND H2O IN VENUS’ MESOSPHERE BY SOIR

E00B22

E00B22

Figure 6. Examples of data fitting for three orders 121 (HDO), 149 (CO2), and 171 (H2O) at the altitudes 86– 89 km. Those data were obtained during orbit 462 (27.07.2007) at the location latitude 87°N, longitude 245°E, 23.55 h LT, distance to the limb 1800 km. The light boxes relate to the spectral range chosen for HDO and H2O retrieval.

HDO, or CO2) and aerosol extinction that determines the continuum level of the transmittance spectrum. The best density value for every measurement is retrieved automatically, minimizing c2 statistically weighted according to the uncertainties of the measured quantity: c2 ¼

p320 X

ðTdata ðiÞ  Tmodel ðiÞÞ2 =sðiÞ2

p1

where Tdata(i) and Tmodel(i) are the measured and modeled spectrum of transmittance for pixel i. s(i) is the statistical error determined from the signal-to-noise of the measured spectra which vary depending on diffraction order and detector pixel. Examples of the best fit spectra for H2O, HDO, and CO2 are presented in Figure 6. The light gray boxes in the top and bottom indicate the spectral range chosen for the retrieval of H2O and HDO density. The preference has been given to the lines with higher intensity, lower sensitivity to temperature, which are less contaminated with CO2 lines. Some lines appear broader in the model than in the observed spectrum that could be explained by the uncertainty of the assumed temperature profile (as partially described by Vandaele et al. [2008]), and possibly by sampling on the detector’s pixels. For the retrieval we used several lines that minimize the discrepancy in the line widths, as described in more detail in section 3.4. The curvature of measured spectrum in Figure 6 (top) is due to undercorrection of detector’s nonlinearity apparent at the edges of order’s spectral range. Figure 7 demonstrates a typical result of the retrieval of the gaseous densities for one

selected orbit, showing the simultaneous vertical distributions of the local densities for HDO, H2O, and CO2. 3.4. Uncertainty of Results [31] There are several sources of uncertainties in our results: spectroscopic data set (in particular the broadening factor for H2O in the CO2 atmosphere), atmospheric data set and assumed temperature-pressure profiles, and instrumental calibration uncertainties. The calibration uncertainties include AOTF frequency-wavelength calibration that determines the position of the maximum of the AOTF function versus the preset frequency for bin 1 and bin 2. This has been obtained with a precision of 0.83 cm1 [Mahieux et al., 2008]. Uncertainties on the shape of the AOTF function, including the contribution of the side lobes are still poorly estimated. Detailed sensitivity analysis on this parameter will be performed in the near future. [32] The uncertainties due to a shift of the maximum of the AOTF function of 1 cm1 give an error of
8% on the retrieved values at altitude from 70 to 95 km for HDO and 7% for H2O. The uncertainties due to the shape of the AOTF function (the cutoff of the far wing in ±3 orders during the retrieval that corresponds to removal of the contribution of the far side lobes) is the most important for H2O retrieval and put the error to 4% at the altitude from 70 to 100 km, which increases up to 40% at the altitude of 110 km. For HDO uncertainties from order mixing are below than 4%. [33] The uncertainties coming from the assumed spectral instrument function of the echelle spectrometer also give

8 of 16

E00B22

FEDOROVA ET AL.: HDO AND H2O IN VENUS’ MESOSPHERE BY SOIR

E00B22

Figure 7. Example of retrieved densities of HDO, H2O, and CO2 in molecules cm3 for orbit 255 corresponding to the northern latitude 79.2°, longitude 261°E, and 19 h LT and distance to the limb 1730 km at altitude 80 km. rise to some systematic bias in the result. The sensitivity to variation of the instrument spectral function half width was investigated for each molecule. A variation of the half width by 20% results on average 8.5% of error in gaseous density, slightly varying with altitude and diffraction order. For different orders the widths of this function varies from 0.16 cm1 for the order 121 to 0.22 cm1 for the order 171 that may also bias the relation between different species detected in distant orders. [34] The comparison of results obtained for bin 1 and bin 2 also can provide a good estimation of uncertainties. For HDO (order 121) the total uncertainty on the average amounts to 8 – 14% and varies a little with altitude and from orbit to orbit that could be explained by different calibrations for these spectra and rotation of slit with respect to the limb. [35] Uncertainties on spacecraft pointing and in fact the unknown real pressure and temperature at the location of measurements add systematic biases to the retrieval. To account for this, we should use CO2 density retrieved from CO2 lines during the same observation as presented in Figure 7. The absorption cross sections of the molecules are sensitive to temperature and pressure. In particular, the line strengths depend on the temperature and the power of dependence is determined by the energy of transition (see equation 6 by Vandaele et al. [2008]). As already mentioned, in the case of H2O and HDO retrieval, CO2 density is retrieved from several different orders: 111, 112, 123, 148, and 149. The line strengths inside these ranges are not very sensitive to temperature variations. The intensity of the H2O and HDO lines are also not very sensitive to temperature, no more than 50% increase for a variation of temperature of 100 K. The uncertainties on temperature can be estimated by comparing the retrievals obtained with two different temperature profiles (Figure 8). We have chosen an extreme case that is assuming a deep inversion of temperature at the altitudes between 85 and 95 km. Retrievals of CO2 (order

149, top) and H2O (order 171, bottom) obtained for orbit 462 are presented in Figure 8. Left shows the temperature profiles used for the retrievals, middle shows the retrieved densities, and right shows the relative difference in % calculated as N1 ð zÞ  N2 ð zÞ N1 ð zÞ

where N1(z) and N2(z) are the local densities for the first and second temperature profiles. For H2O, variations due to temperature are not very high and do not exceed 10%. For CO2, depending on the order, the error can reach 20% at the altitudes of the inversion (85 – 100 km). [36] The importance of the retrieved density profiles for the determination of the volume mixing ratios of the minor species is shown in Figure 9. Mixing ratios of H2O from orbit 442 obtained by division by the model density from VIRA or by the CO2 density retrieved in this work are compared.

4. Results [37] We report here results for 22 orbits obtained from January 2006 to August 2007. All the orbits have been performed at the latitudes of 63°– 88°N on morning and evening terminators with the distance to limb below 3500 km. [38] To detect variability of HDO and H2O abundance with time, several sets of observations have been done. In the present analysis the following sets are considered: 5 orbits from 244 to 262 at December 2006 to January 2007, 4 orbits from 345 to 358 at April 2007, and 15 orbits from 434 to 487 at June– August 2007. Figure 10 demonstrates the density retrieved for 5 consecutive orbits from 244 to 262 and 3 orbits from 442 to 445 obtained close to the north pole. No strong variations of the H2O and HDO

9 of 16

E00B22

FEDOROVA ET AL.: HDO AND H2O IN VENUS’ MESOSPHERE BY SOIR

Figure 8. Sensitivity of the retrieved parameters to the temperature profile. The results are presented for retrieval of CO2 (order 149, top) and H2O (order 171, bottom). They were obtained during orbit 462 (27 July 2007 at latitude 87.7°N, longitude 255°E, and 23.6 h LT, distance to the limb at altitude 80 km is around 1830 km). Left shows the temperature profiles used for the retrievals; middle shows the retrieved densities; right shows the residual error in % calculated as (N1(z)-N2(z))/N1(z), where N1(z) and N2(z) are the densities obtained for two different temperature profiles.

Figure 9. Comparison of the volume mixing ratios (vmr) of H2O using VIRA densities and CO2 densities derived from measurements recorded on the same orbit. The example shows the results obtained for orbit 442 (7 July 2007, at latitude 79°N, longitude 284°E, and 20.7 h LT, distance to the limb at altitude 80 km is around 1780 km). 10 of 16

E00B22

E00B22

FEDOROVA ET AL.: HDO AND H2O IN VENUS’ MESOSPHERE BY SOIR

E00B22

Figure 10. Vertical distributions of the H2O and HDO local density for eight orbits: five consecutive orbits from MTP09 (from 19 December 2006 to 10 January 2007) and three consecutive orbits from MTP016 (from 1 July 2007 to 27 July 2007), where MTP is the medium term plan. The conditions of observations are listed in Table 1. profiles have been detected for the altitude range of 75– 110 km. The latitude of observations varies for these orbits from 70° to 88°N. The mixing ratios of H2O and HDO for the same orbits are presented in Figure 11. [39] A marked depletion of H2O is observed in the range 80– 90 km, for which we have no explanation yet, other than noting that this altitude range coincides with the mesospheric minimum temperature and the top of the haze layer. The bump of H2O and HDO at 90– 100 km relates to depletion of CO2 density retrieved from CO2 order. Figure 9 demonstrates the difference between mixing ratios obtained assuming CO2 density from observations and from VIRA models. This depletion could be related to a strong temperature inversion with minimum at altitude
85– 90 km and we keep this study for a separate paper dedicated to the retrieval temperature and pressure profiles from CO2 lines. The curves of both isotopes show little variability. [40] The mean mixing ratio of H2O has been calculated for each orbit by averaging at altitudes from 75 to 112 km. The values are varying from 0.8 to 1.5 ppm. The temporal variation from orbit to orbit and the latitude variation of mixing ratios are presented in Figure 12. There is no systematic behavior in the distribution detected and the variations could reflect uncertainties of the SOIR observations. For HDO, mixing ratios were averaged from 75 to 95 km, the values varying from 50 to 130 ppb. Table 1 summarizes the averaged values of H2O and HDO volume mixing ratios (vmr). To make the comparison easier, the averaging of H2O for the same altitude range as HDO is

also presented in Table 1. The error bar corresponds to the error of the average in this case. [41] The vertical distribution of D/H ratio can be easily obtained from H2O and HDO density profiles. The complete set of results is presented in Figure 13. There are some vertical variations of the isotopic ratio that have been discussed by Bertaux et al. [2007b]. The average isotopic ratios scaled to the standard mean ocean water (SMOW) ratio of 3.10693  104 and averaged at the range of altitudes from 75 to 90 km are shown in Figure 14 as a function of orbit number and latitude. The obtained values range between 200 and 300 times terrestrial with the average of 240 ± 25.

5. Discussion [42] There were no continuous sets of H2O observations in the Venusian mesosphere before Venus Express. Sandor and Clancy [2005] and Gurwell et al. [2007] have made systematic observations of the mesosphere from groundbased and orbital telescopes. The summary of previous measurements is presented in Table 2. Most of the groundbased observations correspond to measurements of the Venusian disc and the direct comparison with the local observations of SOIR is not possible. Ground-based observations cannot directly resolve the vertical distribution of the species and inversion techniques are required to derive the profile from a single observation. The advantage of SOIR is that direct measurements of vertical profiles from the orbit are made possible. Moreover, simultaneous obser-

11 of 16

E00B22

FEDOROVA ET AL.: HDO AND H2O IN VENUS’ MESOSPHERE BY SOIR

E00B22

Figure 11. Vertical distributions of the H2O and HDO volume mixing ratios for eight orbits: four consecutive orbits from MTP09 and four consecutive orbits from MTP016. The conditions of observations are listed in Table 1. vations in spectral ranges corresponding to both H2O and HDO are sometimes difficult to realize from the Earth. Several observations of H2O [Encrenaz et al., 1991; Sandor and Clancy, 2005] are, in fact, measurements of HDO lines and rely on the HDO/H2O ratio obtained from other measurements. The volume mixing ratio of H2O (
1 ppm)

measured by SOIR is well consistent or lower than obtained from the Earth. The HDO mixing ratio of 0.08 ppm from SOIR is also lower than values obtained by Encrenaz et al. [1991, 1995] and by Sandor and Clancy [2005], but all values are in agreement within the error bars.

Figure 12. Temporal and latitudinal evolution of the H2O volume mixing ratios averaged as described in the text for orbits listed in Table 1. The H2O mixing ratio has been obtained by division by CO2 densities from the same observations. 12 of 16

E00B22

FEDOROVA ET AL.: HDO AND H2O IN VENUS’ MESOSPHERE BY SOIR

E00B22

Figure 13. Vertical distributions of the HDO/H2O ratio for several orbits. [43] High variability of H2O in the Venus mesosphere was recently detected by Sandor and Clancy [2005] and Gurwell et al. [2007]. This variability is not supported by SOIR observations at least near the north pole region. The

SOIR values show variations, but no more than a factor of 2 around the mean value (1 ppm), staying inside the SOIR uncertainties (except perhaps in the lowermost range 70– 75 km, where errors are greater). But the direct compar-

Figure 14. Temporal and latitudinal evolution of the HDO/H2O ratio scaled to Earth’s value. 13 of 16

FEDOROVA ET AL.: HDO AND H2O IN VENUS’ MESOSPHERE BY SOIR

E00B22

E00B22

Table 2. Comparison of Averaged Values of the H2O and HDO Mixing Ratios With Previous Observations Altitude (km)

Spectral Range

H2O (ppm)

HDO (ppm)

Reference

1990

60 – 95

225.9GHz (HDO)

0.13

Encrenaz et al. [1991]

1990 – 1993

60 – 95

187.31GHz (H2O) 225.9GHz (HDO) 2.59 – 2.65 mm (HDO and H2O) 226 GHz (HDO) and 335 GHz 547.676GHz (H2O) and 556.936GHz 2.61 mm (H2O) 3.58 mm (HDO)

3.5 ± 2 (D/H = 120 terrestrial value) 1 ppm

0.26

Encrenaz et al. [1995]

2.09 ± 0.15

0.102

Bjoraker et al. [1992]

0 ± 0.06 3.5 ± 0.3

0 0.17

Sandor and Clancy [2005]

4.5 ± 1.5

-

Gurwell et al. [2007]

0.8 – 1.5

0.05 – 0.13

This research

Years

1992?

72

1998 – 2004

65 – 100

2002 – 2004

65 – 100

2006 – 2007

75 – 95

ison with the ground-based observations is not possible. SOIR measures only morning or evening profiles at terminator, whereas values of H2O from millimeter observation may reflect disc averaged variability. It is equally difficult to compare our results to the H2O afternoon bulge observed by Pioneer Venus orbiter at the altitudes of 62– 70 km in low and middle latitudes from 10°S to 60°N [Koukouli et al., 2005] Therefore, more SOIR occultations in wider latitude range are required to understand spatial variations of water vapor in the mesosphere. [44] Several measurements of the D/H ratio in the Venus’ mesosphere have already been performed before. The 2.3 mm window has been used to observe lines from both H2O and the deuterated water HDO and to measure the D/H ratio below the clouds from ground-based observations. de Bergh et al. [1991] derived a value of 120 ± 40 times the terrestrial SMOW value from high-resolution CFHT spectra on the nightside. Donahue et al. [1982, 1997] has reported the first determinations of the D/H ratio performed in situ by the neutral mass spectrometer of Pioneer Venus, the analysis of their data yields a value of D/H = 100 ±12.5 corrected later to 157 ± 30 times the ratio in Earth’s oceans. Bjoraker et al. [1992] have obtained the value of 157 ± 15 times terrestrial using near-IR dayside ground-based observations. The isotopic ratio retrieved from SOIR measurements (240 ± 25) times terrestrial is higher (factor
1.5) than the results of all previous measurements in the lower atmosphere 157 ± 30 [Donahue et al., 1997] and the measurements by Bjoraker et al. [1992] at the effective altitude of 72 km but they support the higher abundance of HDO to H2O compared to Earth. [45] One motivation to HDO and H2O measurements in the upper atmosphere of Venus was to determine if HDO was present up to the altitude of photodissociation, providing D atoms for further escape. Clearly this D escape needs to be quantified, if we want to extrapolate back in time what was the original content of water on Venus. In the Earth’s stratosphere, there is a cold trap at the tropopause, and preferential condensation of HDO versus H2O is an important factor of fractionation, which is circumvented by deuterated methane (CH3D) passing through the tropopause without condensing [Moyer et al., 1996]. On Mars, the observed depletion of atomic D in the upper atmosphere [Krasnopolsky et al., 1998] was explained by the lower HDO photolysis rate [Cheng et al., 1999],

preferential condensation of HDO [Fouchet and Lellouch, 2000; Bertaux and Montmessin, 2001], and the smaller observed abundance of H2 [Krasnopolsky and Feldman, 2001] than expected from models. Thermal and nonthermal escapes of D are weaker than those of H [Krasnopolsky, 2002] and tend to increase D/H. [46] Our present results show that on Venus, there is no condensation of H2O nor HDO, and no cold trap preventing HDO to be photodissociated in the region above 80 km. On the contrary, one has to explain why the ratio HDO/H2O is found 1.5 times higher than lower in the atmosphere, with a possible trend of increasing with altitude (Figure 13). Bertaux et al. [2007b] proposed two explanations: the higher photodissociation rate of H2O [Cheng et al., 1999] will preserve more HDO; also, if D atoms are not at all escaping at the top of the atmosphere, they may eventually recombine with OH radicals, generating a downward HDO flow, and further decrease the importance of HDO photodissociation. In any case, detailed modeling of these mechanisms is necessary, but it is beyond the scope of the present paper.

6. Conclusions [47] We report vertical distributions of the molecular density and mixing ratios of H2O and HDO in the Venus mesosphere by SOIR, a high-resolution (with R
20,000) echelle spectrometer on board Venus Express. The spectrometer operates in solar occultation mode and sounds the atmosphere at the range of altitudes from 65 to 130 km. Simultaneous measurements of water vapor lines in the 2.61 mm range (3830 cm1) at altitudes of 70– 110 km and HDO lines in the 3.58 mm range (2715cm1) at altitudes of 70– 95 km have been performed. For a year and a half, from April 2006 to August 2007, 54 such measurements have been carried out at different locations on Venus from the north pole to middle south latitudes mainly in the high northern latitudes at the morning and evening terminator. After in-flight recalibration of the spectrometer during summer 2007, a new analysis of the observed H2O has been undertaken. We analyzed the mixing ratio and isotopic values obtained from 22 orbits corresponding to high northern latitudes. The temporal variations have been investigated. The averaged volume mixing ratios of H2O = 1.16 ± 0.24 ppm and HDO = 0.086 ± 0.020 ppm have been obtained for this set of orbits. The depletion in the mixing ratios of H2O and HDO near 85 km is related to the

14 of 16

E00B22

FEDOROVA ET AL.: HDO AND H2O IN VENUS’ MESOSPHERE BY SOIR

depletion in the CO2 density near 95 km and a possible temperature inversion at these altitudes. No large variability of H2O has been detected for the high northern latitudes observations. The time variation of the mixing ratios of HDO and H2O do not exceed 2 –3 times the mean value. The obtained HDO/H2O ratio equals 240 ± 25 times the ratio in the Earth’ ocean on the average, which is higher than previously reported values by a factor
1.5. [48] Future analysis of SOIR data corresponding to middle and lower latitudes of Venus and to a wider set of observations will allow understanding the possible long-term variations. We also count on more accurate instrumental calibration and improved simultaneously retrieval of temperature from SOIR data. [49] The resulting vertical profiles of water vapor and its isotopologue could give an exciting challenge to the development of new photochemical models of the upper atmosphere of Venus. Such investigations have already started with models such the one developed by M.-C. Liang and Y. L. Yung (Modeling the distribution of H2O and HDO in the upper atmosphere of Venus, submitted to Journal of Geophysical Research, 2008). Moreover, combination of VIRTIS and SPICAV nadir measurements of H2O below and above the clouds could reconstruct a global distribution of water vapor in the Venus atmosphere from the lower atmosphere to the upper mesosphere [Drossart et al., 2007; Bertaux et al., 2007a; Marcq et al., 2008]. [50] These profiles give also hints to estimation of past and present escape from the planet. If there were no escape of D atoms, now and in the past, the present 1 cm (equivalent liquid) and D/H ratio  0.025 (enrichment 150) would imply in the past only 1.5 m, compared to 2.8 km on Earth. With the present observation of plenty of HDO in the photodissociation region, D atoms are certainly present in the thermosphere. It would be important to quantify the escape of D atoms, possibly by measuring mass 2 ions escape as could do ASPERA on board Venus Express [Barabash et al., 2007]. In spite of the possible confusion with H+2 ions, it would provide a useful upper limit to D+ ion escape [McElroy et al., 1982; Hartle and Taylor, 1983]. [51] Acknowledgments. We would like to thank our reviewers for helpful comments that improved the manuscript. We thank our collaborators at the three institutes for the design and fabrication of the SOIR instrument, which was mainly built in Belgium by OIP company, under the direction of IASB-BIRA. Russian team acknowledges RFBR grant 06-0272563. Belgium team was supported by the Belgian Federal Science Policy Office and the European Space Agency (ESA, PRODEX program, contracts C 90268, 90113, and 17645). Procurement of AOTF was funded by CNES and the French authors are sponsored by CNRS and CNES. We thank A. Bensoussan (COB) for procuring in due time the SOFRADIR detector.

References Barabash, S., et al. (2007), The analyser of space plasmas and energetic atoms (ASPERA-4) for the Venus Express mission, Planet. Space Sci., 55, 1772 – 1792. Bertaux, J. L., and F. Montmessin (2001), Isotopic fractionation through water vapor condensation: The Deuteropause, a cold trap for deuterium in the atmosphere of Mars, J. Geophys. Res., 106(E12), 32,879 – 32,884, doi:10.1029/2000JE001358. Bertaux, J.-L., et al. (2007a), SPICAVon Venus Express: Three spectrometers to study the global structure and composition of the Venus atmosphere, Planet. Space Sci., 55, 1673 – 1700, doi:10.1016/j.pss.2007.01.016.

E00B22

Bertaux, J.-L., et al. (2007b), A warm layer in Venus’ cryosphere and high altitude measurements of HF, HCl, H2O and HDO, Nature, 450, 646 – 649, doi:10.1038/nature05974. Bjoraker, G. L., H. P. Larson, M. J. Mumma, R. Timmermann, and J. L. Montani (1992), Airborne observations of the gas composition of Venus above the cloud tops: Measurements of H2O, HDO, HF, and the D/H and 18O/16O isotopic ratios, Bull. Am. Astron. Soc., 24, 995. Brown, L. R., C. M. Humphrey, and R. R. Gamache (2007), CO2-broadened water in the pure rotation and n2 fundamental regions, J. Mol. Spectrosc., 246, 1 – 21, doi:10.1016/j.jms.2007.07.010. Cheng, B.-M., E. P. Chew, C.-P. Liu, M. Bahou, Y.-P. Lee, Y. K. Yung, and M. F. Gerstell (1999), Photo-induced fractionation of water isotopomers in the Martian atmosphere, Geophys. Res. Lett., 26(24), 3657 – 3660, doi:10.1029/1999GL008367. de Bergh, C., B. Bezard, T. Owen, D. Crisp, J.-P. Maillard, and B. L. Lutz (1991), Deuterium on Venus: Observations from Earth, Science, 251, 547 – 549. de Bergh, C., V. I. Moroz, F. W. Taylor, D. Crisp, B. Be´zard, and L. V. Zasova (2006), Composition of the atmosphere of Venus below the clouds, Planet. Space Sci., 54, 1389 – 1397. Donahue, T. M., J. H. Hoffman, R. R. Hodges, and A. J. Watson (1982), Venus was wet: A measurement of the ratio of deuterium to hydrogen, Science, 216, 630 – 633. Donahue, T. M., D. H. Grinspoon, R. E. Hartle, and R. R. Hodges Jr. (1997), Ion/neutral escape of hydrogen and deuterium: Evolution of water, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by Stephen W. Bougherer al., pp. 385 – 415, Univ. of Ariz. Press, Tucson. Drossart, P., et al. (2007), Scientific goals for the observation of Venus by VIRTIS on ESA/Venus Express mission, Planet. Space Sci., 55, 1653 – 1672, doi:10.1016/j.pss.2007.01.003. Encrenaz, T., E. Lellouch, G. Paubert, and S. Gulkis (1991), First detection of HDO in the atmosphere of Venus at radio wavelengths: An estimate of the H2O vertical distribution, Astron. Astrophys., 246(1), L63 – L66. Encrenaz, T. H., E. Lellouch, J. Cernicharo, G. Paubert, S. Gulkis, and T. Spilker (1995), The thermal profile and water abundance in the Venus mesosphere from H2O and HDO millimeter observations, Icarus, 117, 162 – 172, doi:10.1006/icar.1995.1149. Fouchet, T., and E. Lellouch (2000), Vapor pressure isotope fractionation effects in planetary atmospheres: Application to deuterium, Icarus, 144, 114 – 123, doi:10.1006/icar.1999.6264. Gamache, R. R., S. P. Neshyba, J. J. Planeaux, A. Barbe, L. Regalia, and J. B. Pollack (1995), CO2-broadening of water vapor lines, J. Mol. Spectrosc., 170, 131 – 151, doi:10.1006/jmsp.1995.1060. Gurwell, M. A., G. J. Melnick, V. Tolls, E. A. Bergin, and B. M. Patten (2007), SWAS observations of water vapor in the Venus mesosphere, Icarus, 188, 288 – 304. Hagemann, R., G. Nief, and E. Roth (1970), Absolute isotopic scale for deuterium analysis in natural waters: Absolute D/H ratio of SMOW, Tellus, 22, 712 – 715. Hartle, R. E., and H. A. Taylor (1983), Identification of deuterium ions in the ionosphere of Venus, Geophys. Res. Lett., 10(10), 965 – 968, doi:10.1029/GL010i010p00965. Howard, T. N., D. E. Burch, and D. Willams (1956a), Infrared transmission of synthetic atmospheres: 1. Instrumentation, J. Opt. Soc. Am., 46, 186 – 190. Howard, T. N., D. E. Burch, and D. Willams (1956b), Infrared transmission of synthetic atmospheres: 2. Absorption by carbon dioxide, J. Opt. Soc. Am., 46, 237 – 241. Howard, T. N., D. E. Burch, and D. Willams (1956c), Infrared transmission of synthetic atmospheres: 3. Absorption by water vapor, J. Opt. Soc. Am., 46, 242 – 245. Howard, T. N., D. E. Burch, and D. Willams (1956d), Infrared transmission of synthetic atmospheres: 4. Application of theoretical band models, J. Opt. Soc. Am., 46, 334 – 338. Keating, G. M., et al. (1985), VIRA (Venus International Reference Atmosphere) models of Venus neutral upper atmosphere: Structure and composition, Adv. Space Res., 5(11), 117 – 171. Korablev, O., and J.-L. Bertaux (2002), Soir experiment for Venus Express: Scientific objectives for solar occultations, paper presented at XXVII General Assembly, Eur. Geophys. Soc., Nice, France, 21 – 26 April 2002, Abstract 3506. Koukouli, M. E., P. G. J. Irwin, and R. W. Taylor (2005), Water vapor abundance in Venus’ middle atmosphere from Pioneer Venus OIR and Venera 15 FTS measurements, Icarus, 173, 84 – 99, doi:10.1016/ j.icarus.2004.08.023. Krasnopolsky, V. A. (2002), Mars’ upper atmosphere and ionosphere at low, medium, and high solar activities: Implications for evolution of water, J. Geophys. Res., 107(E12), 5128, doi:10.1029/2001JE001809.

15 of 16

E00B22

FEDOROVA ET AL.: HDO AND H2O IN VENUS’ MESOSPHERE BY SOIR

Krasnopolsky, V. A., and P. D. Feldman (2001), Detection of molecular hydrogen in the atmosphere of Mars, Science, 294, 1914 – 1917, doi:10.1126/science.1065569. Krasnopolsky, V. A., M. J. Mumma, and G. R. Gladstone (1998), Detection of atomic deuterium in the upper atmosphere of Mars, Science, 280, 1576 – 1580, doi:10.1126/science.280.5369.1576. Mahieux, A., et al. (2008), In-flight performance and calibration of SPICAV SOIR on board Venus Express, Appl. Opt., 47, 2252 – 2265, doi:10.1364/ AO.47.002252. Marcq, E., et al. (2008), A latitudinal survey of CO, OCS, H2O and SO2 in the lower atmosphere of Venus: Spectroscopic studies using VIRTIS-H, J. Geophys. Res., 113, E00B07, doi:10.1029/2008JE003074. McElroy, M. B., M. J. Prather, and J. M. Rodriguez (1982), Escape of hydrogen from Venus, Science, 215, 1614 – 1615, doi:10.1126/science. 215.4540.1614. Moyer, E. J., F. W. Irion, Y. L. Yung, and M. R. Gunson (1996), ATMOS stratospheric deuterated water and implications for troposphere-stratosphere transport, Geophys. Res. Lett., 23(17), 2385 – 2388, doi:10.1029/ 96GL01489. Nevejans, D., et al. (2006), Compact high-resolution space-borne echelle grating spectrometer with AOTF based order sorting for the infrared domain from 2.2 to 4.3 micrometer, Appl. Opt., 45, 5191 – 5206, doi:10.1364/AO.45.005191. Pollack, J. B., et al. (1993), Near-infrared light from Venus’ nightside: A spectroscopic analysis, Icarus, 103, 1 – 42.

E00B22

Press, W. H., B. P. Flannery, S. A. Teukolsky, and W. T. Vetterling (1992), Numerical Recipes in FORTRAN: The Art of Scientific Computing, 2nd ed., 992 pp., Cambridge Univ. Press, Cambridge, U. K. Rodgers, C. D. (2000), Inverse Methods for Atmospheric Sounding: Theory and Practice. Atmos., Oceanic Planet. Phys.Ser., vol. 2, World Sci., Hackensack, N. J. Rothman, L. S., et al. (2005), The HITRAN 2004 molecular spectroscopic database, J. Quant. Spectrosc. Radiat. Transfer, 96, 139 – 204, doi:10.1016/j.jqsrt.2004.10.008. Sandor, B. J., and R. T. Clancy (2005), Water vapor variations in the Venus mesosphere from microwave spectra, Icarus, 177, 129 – 143. Titov, D. V., et al. (2006), Venus Express science planning, Planet. Space Sci., 54, 1279 – 1297, doi:10.1016/j.pss.2006.04.017. Vandaele, A. C., et al. (2008), Composition of the Venus mesosphere measured by SOIR on board Venus Express, J. Geophys. Res., doi:10.1029/2008JE003140, in press. 

D. Belyaev, A. Fedorova, and O. Korablev, Space Research Institute, 117997, 84/32 Profsoyuznaya Street, Moscow, Russia. ([email protected]. rssi.ru) J.-L. Bertaux, F. Montmessin, and E. Villard, Service d’Ae´ronomie du CNRS, BP 3, F-91371, Verrie`res-le-Buisson, France. R. Drummond, A. Mahieux, E. Neefs, A.-C. Vandaele, and W. V. Wilquet, Belgian Institute for Space Aeronomy, 3 avenue Circulaire, B-1180 Brussels, Belgium.

16 of 16

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, E00B25, doi:10.1029/2008JE003143, 2008

for

Full Article

First observations of SO2 above Venus’ clouds by means of Solar Occultation in the Infrared D. Belyaev,1 O. Korablev,1 A. Fedorova,1 J.-L. Bertaux,2,3 A.-C. Vandaele,4 F. Montmessin,2,3 A. Mahieux,4 V. Wilquet,4 and R. Drummond4 Received 18 March 2008; revised 29 July 2008; accepted 2 September 2008; published 31 December 2008.

[1] Solar Occultation in the Infrared (SOIR) is a part of the Spectroscopy for Investigation

of Characteristics of the Atmosphere of Venus (SPICAV)/SOIR occultation experiment on board Venus Express dedicated to the study of gaseous and aerosol vertical structure of Venus’ mesosphere. SOIR is an echelle spectrometer with acoustooptic selection of diffraction orders operating in the wavelengths range of 2.2–4.3 mm at high spectral resolution (l/Dl
20,000). Detection of minor constituents such as CO, H2O, HDO, HCl, HF, and SO2, at altitudes between 65 and 130 km has been demonstrated. We report results from a series of six occultations with observations of the 4-mm SO2 band at latitudes 69°–88°N and 23°–30°N. It is the first time when the vertical distribution of SO2 is retrieved above the clouds with the help of solar occultation direct method. The sulfur dioxide transmission spectrum is measured on a background of strong CO2 absorption. Each retrieved vertical profile of SO2 is characterized by few points; the mixing ratio of SO2 being
0.1 ppm with scale height 1 ± 0.4 km for polar measurements (evening observations) and
1 ppm with scale height 3 ± 1 km at low latitudes (morning observations) at the altitude of about 70 km. Upper limits of <
0.05 ppm are established around 75 km. Citation: Belyaev, D., O. Korablev, A. Fedorova, J.-L. Bertaux, A.-C. Vandaele, F. Montmessin, A. Mahieux, V. Wilquet, and R. Drummond (2008), First observations of SO2 above Venus’ clouds by means of Solar Occultation in the Infrared, J. Geophys. Res., 113, E00B25, doi:10.1029/2008JE003143.

1. Introduction [2] Sulfur dioxide is an important component of Venus’ atmosphere because this gas is the chemical precursor of H2SO4 droplets clouds which completely enshroud Venus. Any significant change in SO2 above and within the clouds of Venus can affect photochemistry and dynamics of the clouds [Esposito et al., 1997]. Moreover, sulfur dioxide behavior in the atmosphere may be an indicator of possible geological activity on the surface of the planet. [3] Sulfur dioxide in the Venusian atmosphere was first detected from ultraviolet Earth-based observations of the solar light scattered by cloud particles and Rayleigh scattering with a high spectral resolution
0.1 nm [Barker, 1979], and the detected mixing ratio was 0.02 – 0.5 ppm at the cloud top. Space-based identifications of SO2 UV absorption followed soon from Pioneer Venus orbiter (PV) [Stewart et al., 1979] and International Ultraviolet Explorer (IUE) [Conway et al., 1979]. The first one provided a retrieval of SO2 vertical column density of about 1017 cm2 with a 1

Space Research Institute, Moscow, Russia. Service d’Ae´ronomie du CNRS, Verrie`res-le-Buisson, France. 3 Also at Institut Pierre Simon Laplace, Universite´ de Versailles-SaintQuentin, Guyancourt, France. 4 Belgian Institute for Space Aeronomy, Brussels, Belgium. 2

Copyright 2008 by the American Geophysical Union. 0148-0227/08/2008JE003143$09.00

spectral resolution of 1.3 nm [Esposito et al., 1979] while the second one registered a column density of 4  1016 cm2 (0.1 – 0.8 ppm above the clouds) with a higher resolution of 0.3 nm. These abundances were larger by orders of magnitude than previously established upper limits delivered from rocket soundings in 1967 [Jenkins et al., 1969; Anderson et al., 1969] and Orbiting Astronomical Observatory (OAO) in 1972 [Owen and Sagan, 1972]. Continuous observations of the PV UV spectrometer from 1978 to 1986 showed a steady decline of the cloud top SO2 content down to 20– 50 ppb [Esposito et al., 1988]. This decline was confirmed by 1987 – 1988 IUE observations [Na et al., 1990]. Inbetween observations of sulfur dioxide in the UV range were made in 1983 by a balloon-borne spectrophotometer with a resolution of 0.4 nm [Parisot et al., 1986]. These measurements gave a mixing ratio of 70 ppb and, hence, confirmed the decrease of SO2 at that moment. Later rocket measurements in the UV range with a resolution of 0.2 nm resulted in SO2 content of 80 ± 40 ppb in 1988 and 120 ± 60 ppb at the equator in 1991 [McClintock et al., 1994; Na et al., 1994]. That was in agreement with PV and IUE results. One more UV measurement was made in 1995 from Hubble Space Telescope (HST) [Na and Esposito, 1995]. The SO2 abundance was 20 ± 10 ppb that is lower than values obtained in 1978– 1991 by a factor of 2 – 5 again confirming the decrease of SO2. [4] Sulfur dioxide was also measured in the infrared spectral range by the Fourier Spectrometer on board Venera

E00B25

1 of 10

E00B25

BELYAEV ET AL.: FIRST OBSERVATIONS OF SO2 BY SOIR

15 orbiter [Moroz et al., 1985]. The instrument detected SO2 absorption bands at 519, 1150, and 1360 cm1 with a spectral resolution of 5 – 7 cm1. Nadir measurements covered the high northern latitudes almost completely; the sounding at altitude was around 69 km (40 mbar pressure). A strong dependence of SO2 content on latitude was reported: from 20 ppb at equator to 0.5 ppm in polar region. These data were found consistent with Pioneer Venus and IUE results [Zasova et al., 1993], taking into account the different effective sounding altitudes. [5] The goal of this paper is to present the first results concerning the detection of sulfur dioxide above the Venus’ cloud top by means of solar occultation in the infrared range using SOIR on board Venus Express orbiter (VEX) [Titov et al., 2006; Bertaux et al., 2007] The use of solar occultations probing (at the planet terminator) allows us to measure the atmospheric transmission and to retrieve information about the vertical structure and the composition at altitudes between 65 and 130 km (see Vandaele et al. [2008] and Fedorova et al. [2008] for the description of the technique and results for other key constituents of the Venus atmosphere). The slant density of SO2 is rapidly decreasing with altitude because of photolysis by sunlight. In occultation geometry this gas originating from the clouds is, therefore, expected to be observed only near the cloud top. The high photometric accuracy of SOIR instrument allows to measure transmitted spectrum down to slant optical thicknesses of t
6 (from aerosols), making it possible to detect SO2 with a satisfactory signal/noise ratio down to 68 km.

2. Instrument [6] SOIR instrument is a compact high-resolution infrared echelle spectrometer, in which an acoustooptical tunable filter (AOTF) is used for preliminary sorting of echelle diffraction orders [Korablev et al., 2004; Nevejans et al., 2006]. The instrument with spectral range 2300– 4500 cm1 and resolution of
0.15 cm1 is capable to detect important minor gaseous constituents such as CO, SO2, HCl, HF, H2O and HDO by means of solar occultation. The detailed description of the spectrometer, its in-flight performance, and calibrations are given in papers by Nevejans et al. [2006] and Mahieux et al. [2008]. We describe briefly the main parameters of the instrument, and some details related to SO2 detection. [7] The spectrometer’s rectangular field of view (FOV) is 30 arc minutes in spatial dimension, by 2 arc min in spectral dimension. This FOV is projected into 256 by 320 elements of an HgCdTe matrix detector. The elements of the detector, corresponding to the height of the slit are binned to form two groups each corresponding to 15 arc minutes. The angular diameter of the Sun from Venus orbit is
40 arc min. The width of the slit corresponds to 2 arc min. The spectrum is dispersed by the echelle onto the 320 spectral elements of the detector. The resolution of the echelle spectrometer, operating in diffraction orders 104– 204 is 0.15 cm1. The SO2 absorption bands were studied in the 4-mm range (2500 cm1), corresponding to 111 and 112 diffraction orders. The AOTF serves as an electrically tuned filter with a passband of
25 cm1, which is slightly larger than the widths of a single diffraction order [Mahieux et al., 2008]. Moreover, the spectral profile of the AOTF trans-

E00B25

mittance with side lobes includes also a few adjacent orders. Therefore, the measured spectrum contains a mixture of several orders. In the present study we consider the contribution of five nearby diffraction orders from each side. Calibration of the AOTF spectral function is made as described by Mahieux et al. [2008].

3. Calculations 3.1. Modeling of Transmission Spectra [8] The solar occultation technique allows measuring the atmospheric transmission at the terminator of a planet (either sunrise or sunset). For a spacecraft moving along its orbit the Sun is occulted by the planetary limb. During the occultation the FOV of the spectrometer is maintained by the inertial spacecraft attitude in Sun direction. A solar spectrum, measured out of the atmosphere (in our case, above
150 km of tangent altitude), is considered as a reference spectrum F0(n) (n denotes wave number). Then, the instrument registers a set of solar spectra F(n, z), passed through the atmosphere at various altitudes of the target point z. The solar occultation method is self-calibrated with respect to the atmospheric transmission which for a given altitude z is determined as: T ðn; zÞ ¼ F ðn; zÞ=F0 ðn Þ:

ð1Þ

The strongest absorption band of SO2 within the spectral range of SOIR is located around 4 mm (2500 cm1). This range is contaminated by a strong absorption of O16C12O18 isotope of CO2 (20003 – 00001 transition) and the SO2 spectrum is registered on the background of the CO2 absorption (Figure 1). The contribution of other atmospheric gases such as H2O and its isotopes, HCl, HF is negligible in this range; among the molecules listed in HITRAN 2004 spectral database [Rothman et al., 2005] the strongest absorption is expected from HDO, and it does not exceed 102. The transmission may be then theoretically expressed by the following combination: T ðn; zÞ ¼ TCO2 ðn; zÞTSO2 ðn; zÞT t ðn; zÞ;

ð2Þ

where 2 TCO2 ðn; zÞ ¼ exp42

Z1

3 sCO2 ðT ðl Þ; pðlÞÞnCO2 ðl Þdl5;

z

2 TSO2 ðn; zÞ ¼ exp42

Z1

3 sSO2 ðT ðl Þ; pðl ÞÞnSO2 ðl Þdl5;

z

Tt ðn; zÞ ¼ exp½t a ðn; zÞ :

Integration is performed along the optical trajectory l in the atmosphere. Gaseous local density n(l) is defined from respective slant density using the ‘‘onion peeling’’ method [Be´zard et al., 1987; Vandaele et al., 2008] assuming spherical symmetry of the atmosphere, and hydrostatic equilibrium. The absorption cross section si(T, p) can be

2 of 10

E00B25

BELYAEV ET AL.: FIRST OBSERVATIONS OF SO2 BY SOIR

E00B25

Figure 1. Synthetic spectra of SO2 (for several mixing ratios, 0.05 (red), 0.1 (blue), and 1 ppm (black)) are presented at occultation altitude of 70 km on a background of CO2 spectrum (green curve) in range 2470 –2525 cm1 corresponding to the echelle orders 111 and 112.

calculated using line-by-line simulations of gaseous absorption using line parameters from spectroscopic database. The slant optical thickness of aerosol t a(n, z) is determined from the occultation procedure as

contribution of adjacent diffraction orders of the echelle spectrometer. Separate CO2 and SO2 simulations with the orders mixing are shown as well (green and black curves, respectively).

t a ðn; zÞ ¼  lnðTa ðn; zÞÞ;

3.2. Algorithm of Fitting [11] As it is evident from the synthetic model (Figures 1 and 2), the SO2 absorption features are overlapped by a strong CO2 structure. Not a single isolated line of SO2 is available for the analysis. Therefore, a combined synthetic model TCO2TSO2 including extinction and the mixing of diffraction orders was compared to the measured spectrum at each altitude. The fitting was performed in the spectral range of 2483 –2500 cm1 in the diffraction order 111, and of 2505 – 2523 cm1 in the order 112. The procedure to compute the synthetic model including aerosol extinction and the order mixing consists of the following steps: [12] 1. Computing of theoretical TCO2(n, z) and TSO2(n, z) transmissions separately for a specified altitude z. The CO2 density nCO2 and the SO2 mixing ratio fSO2 = nSO2/nCO2 are considered as variable parameters. [13] 2. Combined transmission TCO2SO2(n, z) = TCO2(n, z) TSO2(n, z) is calculated taking into account the mixing of echelle orders with the contribution of five diffraction orders ord (n, z). The full from each side: TCO2SO2(n, z) ! TCO2SO2 spectral range which contributes to the combined spectrum amounts to 2370 – 2640 cm1. mod data (n, z) and Idif (n, z) [14] 3. Two ‘‘differential’’ continua Idif ord are extracted from both theoretical TCO2SO2(n, z) and measured Tdata(n, z) spectra (equation (1)), in a similar manner as in the method of differential optical absorption spectroscopy (DOAS) [Plat, 1994]. Continuum curves are determined at a number of spectral intervals between absorption lines where absorption is minimal. Square polynomial interpolation is used in between.

ð3Þ

where Ta(n, z) is the continuum derived from the measured transmission spectra (equation (1)) following a retrieval procedure described below. In the present analysis we do not take into account the atmospheric refraction, which may introduce an error of a fraction of km into the effective sounded attitude at 70 km [Mahieux et al., 2008]. [9] All necessary spectroscopic gaseous parameters for calculations in equation (2) are taken from HITRAN 2004 database. Within the spectral range of interest (2450 – 2550 cm1) the SO2 absorption is observed on the background of carbon dioxide absorption dominated by isotopic band of 16O12C18O (628), and the mixing ratio of SO2 is derived with respect to this isotope. The isotopic ratio of 628 isotope on Earth is 3.9  103, and it is almost the same on Venus with an accuracy of
5% (as discussed by Vandaele et al. [2008]). Further in the text, for simplicity, CO2 will signify the 628 isotope. [10] A set of SO2 synthetic spectra of Venus atmospheric transmission in occultation geometry for the tangent altitude of 70 km is presented together with the CO2 model (including all isotopes) in Figure 1. We assumed VIRA-2 empirical atmospheric model [Moroz and Zasova, 1997; Zasova et al., 2006] for atmospheric density and uniformly mixed SO2 with several relative abundances: 0.05, 0.1 and 1 ppm (from the VIRA data this value is 0.1 ppm for the same altitude). In Figure 2 the spectrum of the combined TCO2TSO2 transmission (blue curves for 111 and 112 orders) at the altitude 68 km is modeled taking into account

3 of 10

BELYAEV ET AL.: FIRST OBSERVATIONS OF SO2 BY SOIR

E00B25

E00B25

Figure 2. Synthetic spectra of combined transmission (CO2*SO2) (dashed curves) are presented for echelle orders (top) 111 and (bottom) 112 at the altitude of 68 km together with separated models of CO2 (thin curves) and SO2 (for 1 ppm; thick curves). All the spectra are calculated taking into account the contribution from ±5 adjacent echelle orders due to nonzero side lobes of the AOTF passband function. [15] 4. Calculation of additional extinction factor (due to mod (n, z) that attenuation in the atmosphere) A(n, z) = 1/Idif describes difference between gaseous ‘‘differential’’ and aerosol continua. The aerosol continuum is then defined as data (n, z) A(n, z), and, therefore, we can find the Taer(n, z) = Idif atmospheric optical thickness t a(n, z) following equation (3). [16] 5. Finally, a combined synthetic transmission is ord (n, z)Tt(n, z) (equation (2)). found as Tmod(n, z) = TCO2SO2 [17] The resulted synthetic model Tmod(n, z) is fitted to each measured spectrum to retrieve the values of nCO2 and fSO2 corresponding to the minimum of the function S ðnCO2 ; fSO2 Þ ¼

X Tdata ðn i ; zÞ  Tmod ðn i ; zÞ2 si

i

;

ð4Þ

where i is a number of a spectral point (1 – 320), si is the error corresponding to this point. The aerosol optical thickness t a is defined as a by-product of the retrieval at

each altitude. We will use value t a = 1 as reference to define the cloud top level (though it is rather the top of the haze, above the classical cloud deck).

4. Occultations 4.1. Strategy of Observations [18] From December 2006 to August 2007 more then 20 occultations covering the spectral range relevant to SO2 were performed. We have analyzed a subset of six occultations available before July 2007. Two measurements correspond to low latitudes (23°– 30°N) and four to high latitudes of Venus (69° – 88°N) (see Table 1). Observations nearby the pole were made from the closest distance to the limb (
2000 km, that corresponds to altitude resolution about 1 km), while observations at low latitudes were made from larger distances (>5000 km with vertical resolution of >3 km). During an occultation two spectra are recorded simultaneously by two bins of the detector corresponding to

Table 1. Tabulation of Analyzed Occultations With Main Parameters of Observationa Orbit (in/eg) 235 244 251 262 434 435

(in) (in) (in) (in) (eg) (eg)

Date

Local Time

Lat

Lon

Dist, km

12 Dec 2006 21 Dec 2006 28 Dec 2006 8 Jan 2007 29 Jun 2007 30 Jun 2007

0149 1953 1916 1830 0558 0558

88° 85° 82° 69° 30° 23°

309° 246° 260° 282° 265° 267°

2130 2242 1741 2274 5502 6179

Spectral Range, cm1 (Order) 2483 – 2500 2483 – 2500 2505 – 2523 2483 – 2500 2483 – 2500 2505 – 2523

(111) (111) (112) (111) (111) (112)

Altitude Range, km

Altitude of t = 1, km

fSO2, ppm

HSO2, km

68 68 68 – 71 69 – 72 70 – 76 72 – 78

76 75 76 75 83 83

0.5 0.7 0.3 0.7 1 3

— — 1.1 0.8 2.8 3.3

a Orbit (in/eg), number of orbit (ingress/egress); Date, date of observation; Local time, local time on Venus; Lat, latitude; Lon, longitude; Dist, distance to limb; Spectral range (order), spectral range (cm1) with corresponded echelle order; Altitude range, altitude range of SO2 detection (km); Altitude at t = 1, altitude at which optical thickness is one; fSO2, mixing ratio of SO2 (ppm) at level of
69 km; HSO2, scale height of SO2 at 70 km.

4 of 10

BELYAEV ET AL.: FIRST OBSERVATIONS OF SO2 BY SOIR

E00B25

E00B25

Figure 3. An example of transmission spectrum measured during an occultation on the orbit 244 at the altitude of 68 km is presented for the echelle order 111 (blue curve) with level of noise (red curve). ord (n, z) Tt (n, z) (see step 5 of the Simulated spectrum of combined transmission Tmod (n, z) = TCO2SO2 fitting algorithm) corresponds to the best fit of 0.7 ppm SO2 mixing ratio (black curve). Also, a fit in CO2 (n, z)Tt (n, z) is which SO2 absorption is forced to be zero and only the CO2 absorption remains Tmod shown by the green curve for the reference.

both parts of the spectrometer slit (see above). For the six occultations selected in present study the spectra measured in the two bins coincide within 10%. It means that for these occultations the slit is kept nearly parallel to the limb, and we have averaged spectra from the two bins. For the remaining observations the slit cannot be considered parallel to the limb; more sophisticated geometry calculations and appropriate data processing for this case will be addressed to a future paper about SOIR performance. 4.2. Data Analysis [19] Figure 3 displays a transmission spectrum measured at the altitude of 68 km during the occultation at orbit 244 (blue line). The spectrum is compared with a synthetic CO2 (n, z)Tt (n, z)) and of combined spectra of CO2 (Tmod CO2*SO2 (Tmod(n, z)) with retrieved optimal values of nCO2 and fSO2 as described above. The observed transmission is of the order of 0.3% only: the solar flux is highly attenuated, mainly by aerosols, with a horizontal optical thickness t a  5.8. The two synthetic spectra (for CO2 and CO2*SO2), which are built to mimic the observations, contain an aerosol attenuation, and the contribution of the adjacent orders, not completely eliminated by the AOTF function. This contribution amounts to 50% of the transmission in the main order observed. [20] The signature of SO2 in the measured spectrum may be extracted by dividing the observed transmission by the CO2 simulation using: SO2 Tdata ðn; zÞ ¼

Tdata ðn; zÞ CO2 ðn; zÞTt ðn; zÞ Tmod

ð5Þ

and comparing the result with what is obtained using the simulated transmission: SO2 Tmod ðn; zÞ ¼

Tmod ðn; zÞ CO2 ðn; zÞT ðn; zÞ Tmod t

ð6Þ

This is illustrated in Figure 4. The resulting model transmission of SO2 is slightly different from the one displayed on Figure 1, because here the contribution of adjacent orders leaking through the AOTF orders is taken into account. [21] An alternate estimation of SO2 contents was performed by comparing the measured SO2 transmission spectrum (equation 5) with the simulated one (equation 6) using SO2 a ‘‘scatter plot’’ where each point has for x coordinate Tmod SO2 (n, z), and for y coordinate the Tdata (n, z) (Figure 5a and equation (4)). The same exercise was performed with differential transmission dT, where dT = T – average(T(n)) (Figure 5b), where the average value is a sliding average over 1 cm1 along the whole 2484 – 2500 cm1 range. When the slope on the plot is one at some variable parameter fSO2, it means that the SO2 mixing ratio during the measurement is equal (or very near) to this fSO2. The slope is found using least squares method. In the example shown, the slope is k = 0.94 ± 0.20 for T, and k = 0.90 ± 0.16 for d(T). 4.3. Results [22] Vertical profiles of the sulfur dioxide mixing ratio are presented in Figure 6, together with aerosol slant optical thickness retrieved from the same measurement sequences as described above. Sometimes the fitting process does not allow finding a distinct minimum; in such cases only upper

5 of 10

E00B25

BELYAEV ET AL.: FIRST OBSERVATIONS OF SO2 BY SOIR

E00B25

SO2 Figure 4. SO2 spectrum Tdata (n, z) (thin curve) retrieved from the same measured spectrum as in SO2 (n, z) (thick curve) following Figure 3 following equation (5) is compared with SO2 synthetic model Tmod equation (6). Arrows mark the most prominent absorption features of sulfur dioxide. Only the gaseous transmission without contribution of aerosol extinction is presented.

limits of the SO2 abundance can be found (indicated by triangles). All the remaining points signify a positive detection with error bars. For all four high-latitude occultations available (69° – 88°N) about 0.1 ppm of SO2 is detected at the level of 70 km, rapidly decreasing with a scale height of HSO2 = 1 ± 0.4 km above. At low latitudes (23° –30°N) we detect significantly higher contents of about 1– 3 ppm and HSO2 = 3 ± 1 km. We used only the detected SO2 values to estimate the scale height; upper limits were not taken into account. [23] The clouds top level seems to be a bit higher at low latitudes where mixing ratio of SO2 is found larger. The

altitude where t a = 1, a possible definition of clouds top level, is found to be of the order 75– 76 km for latitudes >69° and 83 km for latitudes 23° and 30°. Usually the clouds top is known to be at a lower altitude for high latitudes [Crisp, 1986]. It must be recognized that the clouds top is usually defined with a vertical depth of unity, at a much lower level than where the horizontal optical thickness is t a = 1. In fact, at this level we probe the upper haze, which stands above the main clouds. [24] In transmission we are sensitive to the total slant density of SO2 molecules, and not to the mixing ratio. This is why we would not be able to see SO2 at higher

Figure 5. ‘‘Scattered plots’’ illustrating the retrieval of SO2 transmission are presented for the same SO2 plotted versus the modeled one measurements as in Figures 3 and 4. (a) The measured transmission Tdata SO2 Tmod , with a linear regression of this dependence (solid line). (b) The transmission dT, where dT = T – average(T) is plotted in the same manner. When the slope of the linear approximation k is close to unity, the measured and theoretical SO2 contents are equal to each other (or very near). 6 of 10

BELYAEV ET AL.: FIRST OBSERVATIONS OF SO2 BY SOIR

E00B25

E00B25

Figure 6. Vertical profiles of SO2 mixing ratio retrieved from all analyzed occultations (points with error bars) together with aerosol profiles (solid curves). Triangles denote the point where only the upper limits of the SO2 detection. The conditions of observations are indicted on the plots: d, distance to limb; Lat, latitude; Lon, longitude; LT, local time (see Table 1). altitudes, even if the SO2 mixing ratio would be >1 ppmv above 75 km.

5. Uncertainties [25] There are several aspects that impact on the accuracy of retrieved results. Some of them deal with specific features of the spectrometer and some others with uncertainties of the observation geometry. [26] The signal/noise ratio (SNR) of the instrument is about 103 for solar spectrum outside the atmosphere in the center of diffraction orders 111 – 112. On the edges of the detector SNR decreases by a factor of 4 – 5 due to combined effect of AOTF passband and echelle blaze functions. The extinction by aerosol and CO2 at the altitude of 68– 70 km (where SO2 is detected) decreases the signal by a factor of 5  103. The resulting SNR effective for the altitude range of SO2 detection varies from 2 to 3 in the center of the detector to 10– 15 at the edges. These uncertainties for each pixel are taken into account as si in equation (4), and they are translated into
40% of random error of retrieved SO2 abundance. [27] Systematic uncertainties considered are AOTF calibrations (different for the two parts of the slit), and altitude determination. The different AOTF calibrations [Mahieux et al., 2008] are minimized by averaging spectra from the two bins and can be estimated as
10% of relative error in the final gaseous content. The altitude uncertainties depend on the distance of the spacecraft to the limb (±0.5 km at

2000 km and ±2 km at 6000 km); also we have neglected the atmospheric refraction (see above). However, for the retrieval of the mixing ratio the altitude uncertainty results in extraction of CO2 local density with errors
5%. Summarizing all mentioned uncertainties we obtain errors of sulfur dioxide mixing ratio of about 50 –70%.

6. Discussions [28] The measurements of sulfur dioxide above clouds were performed during almost 40 years (Figure 7 and Table 2) with some gaps. Summarizing the history of SO2 observations on Venus clouds top, an immediate conclusion is that the SOIR measurements indicate higher mixing ratios by a factor 5 to 25 with respect to all previous measurements. The solar occultation measurements are made at the terminator of the planet; that is, around one half of the optical path is in the shade, and another is sunlit. Most of previous nadir and limb observations, with the exception Venera 15 IR measurements were performed in the UV range on the dayside of the planet. The increased contents of SO2 measured on the terminator may be explained by its photochemical behavior. Indeed, the photochemical lifetime of SO2 exposed to sun light at 68– 70 km altitude is
3  103 s (i.e., 1 h) [Mills et al., 2007], and its contents should depend on local time. However, other photochemical cycles in which the SO2 is involved are more complicated: with UV solar light SO2 is rapidly produced again from SO, and the timescale for net loss of SO2 via oxidation to SO3

7 of 10

BELYAEV ET AL.: FIRST OBSERVATIONS OF SO2 BY SOIR

E00B25

E00B25

Figure 7. Measurements of SO2 on Venus above clouds, available from 1969 up to now (see Table 2). The SO2 content in ppb volume is considered at the level of 40 mbar (
69 km of altitude). SOIR results indicate values for 68 and 70 km (only at high latitudes) and SPICAV UV results were derived from 100 to 110 km. is much longer, 106 – 107 s [Winick and Stewart, 1980; Krasnopolsky and Pollack, 1994; Pernice et al., 2004]. These estimations for the lifetime depends on the column abundance of SO2 and the production rate for SO3, both of which are not well constrained. [29] Comparing SOIR results on morning and evening terminators (see Table 1 and Figure 6), for morning occultations (at low latitudes) we measured
1 ppm of SO2 mixing ratio while in the evening (high latitudes)
0.1 ppm is detected at 70 km of altitude. Unfortunately, at present we do not have considerable statistics of observations to compare morning and evening occultations at similar latitudes. An extensive set of daytime and nighttime SO2 measurements was obtained by Fourier spectrometer (FS) on Venera 15. A strong latitude dependence of SO2 was retrieved from 20 ppb at the equator up to 0.5 ppm in polar regions [Zasova et al., 1993], but diurnal cycle dependencies were not analyzed. [30] Regarding the altitude distribution of SO2, on the average SOIR measurements indicate a scale height of 1 ±

0.4 km for polar measurements and 3 ± 1 km at low latitudes at the altitude about 70 km. The latter is comparable with the scale height retrieved by Venera 15 FS at 69 km [Zasova et al., 1993]. For polar regions the scatter of Venera 15 scale heights is large, data being clustered around 1 km and 4 – 6 km. The estimations of scale height from PV ultraviolet measurements [Esposito et al., 1984] ranges from 1.5 to 4 –5 km; the IUE measurements [Na et al., 1990] are consistent with a scale height of 3 ± 1 km. Our data are not in conflict with these previous measurements. The peculiarities of the solar occultation technique are operations at the planet terminator, hence, at local morning or evening, and a significant horizontal optical path, reaching 450 km at 70 km. Therefore, SOIR measurements might reflect evening/morning, and also horizontal inhomogeneities, the latter may be associated with small-scale haze structures. VIRTIS has observed such structures at a scale down to few kilometers [Piccioni, 2008]. [31] Several hypothesis have been put forward in order to explain the amazing evolution of sulfur dioxide content

Table 2. List of Previous Detections of Sulfur Dioxide Above Venus’ Clouds by Several Methods and in Several Spectral Rangesa Reference

Instrument

Spectral Range (nm)

SO2 Content (ppb)

Anderson et al. [1969] Owen and Sagan [1972] Esposito et al. [1988] Na et al. [1990] Parisot et al. [1986] Moroz et al. [1985]; Zasova et al. [1993] McClintock et al. [1994] Na and Esposito [1995] Bertaux et al. [2008] This paper

Rocket OAO UV spectrometer (PV) UV spectrometer (IUE) Balloon Fourier Spectrometer (Venera 15) Rocket HST SPICAV UV (VEX) SOIR (VEX)

200 – 300 200 – 360 200 – 320 200 – 220 200 – 320 7350, 8700, 19270

<50 <10 50 – 400 50 – 380 70 200

190 – 230 207 – 216 200 – 320 4000

80 – 120 20 300 – 1000 100 – 550

a

SO2 content correspond to altitude of
69 km except SPICAV UV results that were derived from 100 to 110 km.

8 of 10

E00B25

BELYAEV ET AL.: FIRST OBSERVATIONS OF SO2 BY SOIR

above Venus clouds on a timescale of tens of years from 1967 to 1995 (Figure 7). After a sharp increase between 1967 and 1979, a stable decline down to
20 ppb at 1995 was observed, with values close to upper limits of 1967. Three possible explanations of such behavior were recently recalled by Mills et al. [2007]: active volcanism [Esposito, 1984], changes in the effective eddy diffusion within the cloud layers [Krasnopolsky, 1986, p. 147], and changes in atmospheric dynamics [Clancy and Muhleman, 1991]. The volcano hypothesis uses the volcanic eruption as a source of buoyancy that allows the abundant SO2 below the Venus clouds to break through the stable upper cloud layer. The entrained SO2 is then observable remotely at the cloud top in the UV. Similarly, because the observed SO2 mixing ratio may differ by as much as 4 orders of magnitude from the base to the top of the cloud layers [Bertaux et al., 1996; Esposito et al., 1997], a small change in the effective eddy diffusion within the cloud layers may significantly alter the cloud top abundance of SO2 [Krasnopolsky, 1986, p. 147]. The average SO2 abundance below the clouds varies probably much more slowly, connected to the evolution of volcanic activity over geologic timescales [Fegley et al., 1997a]. [32] Large values of SO2 (up to 2 ppm) observed by SOIR on Venus Express in a limited set are supported by recent microwave observations by Sandor et al. [2007], indicating also a large variability of SO2, and by SPICAV UV stellar occultations with retrieval from 0.3 to 1 ppmv of the gas at 100– 110 km [Bertaux et al., 2008]. At present it would be premature to conclude that in 2004– 2007 we observe another global increase of SO2 contents, like one suspected in between 1970 and 1977 or a leftover of such an increase in the past. Sulfur dioxide content at Venus’ clouds top might express significant variations in the haze cover. Sulfuric acid droplets are a potential reservoir of SO2, and a surge of SO2 may results from an increased evaporation of these droplets, triggered by a temperature change of dynamical origin. [33] Further monitoring of SO2 by SOIR and SPICAV at high altitudes may clarify these dependences. [34] Acknowledgments. We thank our collaborators of the three institutes for the design and fabrication of the SOIR instrument, which was mainly built in Belgium by OIP Company, under the direction of IASB-BIRA (IASB-BIRA/Belgium, Service d’Ae´ronomie/France, and IKI/ Moscow). Russian team acknowledges RFBR grant 06-02-72563. Belgium team was supported by the Belgian Federal Science Policy Office and the European Space Agency (ESA, PRODEX program, contracts C 90268, 90113, and 17645). Procurement of AOTF was funded by CNES.

References Anderson, R. C., J. G. Pipes, A. L. Broadfoot, and L. Wallace (1969), ˚ , J. Atmos. Sci., 26, Spectra of Venus and Jupiter from 1800 to 3200 A 874 – 888, doi:10.1175/1520-0469(1969)026<0874:SOVAJF>2.0.CO;2. Barker, E. S. (1979), Detection of SO2 in the UV spectrum of Venus, Geophys. Res. Lett., 6(2), 117 – 120, doi:10.1029/GL006i002p00117. Bertaux, J.-L., T. Widemann, A. Hauchecorne, V. I. Moroz, and A. P. Ekonomov (1996), VEGA 1 and VEGA 2 entry probes: An investigation of local UV absorption (220 – 400 nm) in the atmosphere of Venus (SO2, aerosols, cloud structure), J. Geophys. Res., 101, 12,709 – 12,745, doi:10.1029/96JE00466. Bertaux, J.-L., et al. (2007), SPICAV on Venus express: Three spectrometers to study the Global structure and composition of the Venus atmosphere, Planet. Space Sci., 55, 1673 – 1700, doi:10.1016/j.pss.2007.01.016. Bertaux, J.-L., et al. (2008), SPICAV/SOIR on board Venus Express: An overview of two years of observations, paper presented at 37th COSPAR Scientific Assembly, Montreal, Canada, July 13 – 21.

E00B25

Be´zard, B., J.-P. Baluteau, A. Marten, and N. Coron (1987), The 12C/13C and 16O/18O ratios in the atmosphere of Venus from high-resolution 10-mm spectroscopy, Icarus, 72(3), 623 – 634, doi:10.1016/0019-1035(87)90057-1. Clancy, R. T., and D. O. Muhleman (1991), Long-term (1979 – 1990) changes in the thermal dynamical and compositional structure of the Venus mesosphere as inferred from microwave spectral observations of 12 CO, 13CO and C18O, Icarus, 89, 129 – 146, doi:10.1016/0019-1035(91) 90093-9. Conway, R. R., R. P. Mccoy, C. A. Barth, and A. L. Lane (1979), IUE detection of sulfur dioxide in the atmosphere of Venus, Geophys. Res. Lett., 6(7), 629 – 631, doi:10.1029/GL006i007p00629. Crisp, D. (1986), Radiative forcing of the Venus mesosphere. I - Solar fluxes and heating rates, Icarus, 67, 484 – 514, doi:10.1016/00191035(86)90126-0. Esposito, L. W. (1984), Sulfur dioxide: Episodic injection shows evidence for active Venus volcanism, Science, 223, 1072 – 1074, doi:10.1126/ science.223.4640.1072. Esposito, L. W., J. R. Winick, and A. I. Stewart (1979), Sulfur dioxide in the Venus atmosphere: Distribution and implications, Geophys. Res. Lett., 6(7), 601 – 604, doi:10.1029/GL006i007p00601. Esposito, L. W., M. Copley, R. Eckert, L. Gates, A. I. F. Stewart, and H. Worden (1988), Sulfur dioxide at the Venus cloud tops, 1978 – 1986, J. Geophys. Res., 93, 5267 – 5276, doi:10.1029/JD093iD05p05267. Esposito, L. W., J.-L. Bertaux, V. Krasnopolsky, V. I. Moroz, and L. V. Zasova (1997), Chemistry of lower atmosphere and clouds, in Venus II, edited by S. W. Bougher, D. M. Hunten, and R. J. Phillips, pp. 415 – 458, Univ. of Ariz. Press, Tucson. Fedorova, A., et al. (2008), HDO and H2O vertical distributions and isotopic ratio in the Venus mesosphere by Solar Occultation at Infrared spectrometer on board Venus Express, J. Geophys. Res., doi:10.1029/ 2008JE003146, in press. Fegley Jr., B., G. Klingelhofer, K. Lodders, and T. Widemann (1997a), Geochemistry of surface-atmosphere interactions on Venus, in Venus II, edited by S. W. Bougher, D. M. Hunten, and R. J. Phillips, pp. 591 – 636, Univ. of Ariz. Press, Tucson. Jenkins, E. B., D. C. Morton, and A. V. Sweigart (1969), Rocket spectra of ˚ , Astrophys. J., 157, 913 – 924, Venus and Jupiter from 2000 to 3000 A doi:10.1086/150123. Korablev, O. I., J.-L. Bertaux, I. I. Vinogradov, Y. K. Kalinnikov, D. Nevejans, E. Neefs, T. Le Barbu, and G. Durry (2004), Compact high-resolution echelle-AOTF NIR spectrometer for atmospheric measurements, in Proceedings of the 5th ICSO, Eur. Space Agency Spec. Publ., ESA SP-554, 73 – 80. Krasnopolsky, V. A. (1986), Photochemistry of the Atmosphere of Mars and Venus, 147 pp., Springer, Berlin. Krasnopolsky, V. A., and J. B. Pollack (1994), H2O-H2SO4 system in Venus’ clouds and OCS, CO, and H2SO4 profiles in Venus’ troposphere, Icarus, 109, 58 – 78, doi:10.1006/icar.1994.1077. Mahieux, A., et al. (2008), In-flight performance and calibration of SPICAV SOIR on board Venus Express, Appl. Opt., 47(13), 2252 – 2265, doi:10.1364/AO.47.002252. McClintock, W. E., C. A. Barth, and R. A. Kohnert (1994), Sulfur dioxide in the atmosphere of Venus, Icarus, 112, 382 – 388, doi:10.1006/ icar.1994.1192. Mills, F. P., L. W. Esposito, and Y. L. Yung (2007), Atmospheric composition, chemistry, and clouds, in Exploring Venus as a Terrestrial Planet, Geophys. Monogr. Ser., vol. 176, edited by L. W. Esposito, E. R. Stofan, T. E. Cravens, pp. 73 – 100, AGU, Washington, D. C Moroz, V. I., and L. V. Zasova (1997), VIRA-2: A review of inputs for updating the Venus International Reference Atmosphere, Adv. Space Res., 19(8), 1191 – 1201, doi:10.1016/S0273-1177(97)00270-6. Moroz, V. I., et al. (1985), Venera 15 and Venera 16 infrared experiment. 4. Preliminary results of spectral analyses in the region of H2O and SO2 absorption bands, Cosmic Res., Engl. Transl., 23(2), 202 – 211. Na, C. Y., and L. W. Esposito (1995), UV observations of Venus with HST (abstract), Bull. Am. Astron. Soc., 27, 1071. Na, C. Y., L. W. Esposito, and T. E. Skinner (1990), International Ultraviolet Explorer observation of Venus SO2 and SO, J. Geophys. Res., 95(D6), 7485 – 7491, doi:10.1029/JD095iD06p07485. Na, C. Y., L. W. Esposito, W. E. McClintock, and C. A. Barth (1994), Sulfur dioxide in the atmosphere of Venus, II. Modeling results, Icarus, 112, 389 – 395, doi:10.1006/icar.1994.1193. Nevejans, D., et al. (2006), Compact high-resolution spaceborne echelle grating spectrometer with acousto-optical tunable filter based order sorting for the infrared domain from 2.2 to 4.3 mm, Appl. Opt., 45(21), 5191 – 5206, doi:10.1364/AO.45.005191. Owen, T., and C. Sagan (1972), Minor constituents in planetary atmospheres: Ultraviolet spectroscopy from the Orbiting Astronomical Observatory, Icarus, 16, 557 – 568, doi:10.1016/0019-1035(72)90102-9.

9 of 10

E00B25

BELYAEV ET AL.: FIRST OBSERVATIONS OF SO2 BY SOIR

Parisot, J. P., P. Rigaud, and D. Huguenin (1986), Balloon observations of Venus from 200 to 320 nm, Astron. Astrophys., 166, 333 – 336. Pernice, H., et al. (2004), Laboratory evidence for a key intermediate in the Venus atmosphere: Peroxychloroformyl radical, Proc. Natl. Acad. Sci. U. S. A., 101, 14,007 – 14,010, doi:10.1073/pnas.0405501101. Piccioni, G. (2008), Observations of Venus by VIRTIS aboard Venus Express, paper presented at 37th COSPAR Scientific Assembly, Montre´al, Canada, 13 – 20 July. Plat, U. (1994), Differential optical absorption spectroscopy, in Air Monitoring by Spectroscopic Techniques, Chem. Anal. Ser., vol. 127, edited by M. W. Sigrist, pp. 27 – 84, John Wiley, New York. Rothman, L. S., et al. (2005), The HITRAN 2004 molecular spectroscopic database, J. Quant. Spectrosc. Radiat. Transfer, 96(2), 139 – 204, doi:10.1016/j.jqsrt.2004.10.008. Sandor, B. J., T. Clancy, G. H. Moriarty-Schieven (2007), SO and SO2 in the Venus mesosphere: Observations of extreme and rapid variation, paper presented at 3th DPS Meeting, Am. Astron. Soc., Orlando, Fla. Stewart, A. I., D. E. Anderson, L. W. Esposito, and C. A. Barth (1979), Ultraviolet spectroscopy of Venus: Initial results from the Pioneer Venus orbiter, Science, 203, 777 – 779, doi:10.1126/science.203.4382.777. Titov, D. V., et al. (2006), Venus Express: Scientific goals, instrumentation, and scenario of the mission, Cosmic Res., Engl. Transl., 44(4), 334 – 348, doi:10.1134/S0010952506040071.

E00B25

Vandaele, A. C., et al. (2008), Composition of the Venus mesosphere measured by Solar Occultation at Infrared on board Venus Express, J. Geophys. Res., doi:10.1029/2008JE003140, in press. Winick, J. R., and A. I. Stewart (1980), Photochemistry of SO2 in Venus’ upper cloud layers, J. Geophys. Res., 85, 7849 – 7860, doi:10.1029/ JA085iA13p07849. Yung, Y. L., and W. B. Demore (1982), Photochemistry of the stratosphere of Venus—Implications for atmospheric evolution, Icarus, 51(2), 199 – 247, doi:10.1016/0019-1035(82)90080-X. Zasova, L. V., V. I. Moroz, L. W. Esposito, and C. Y. Na (1993), SO2 in the middle atmosphere of Venus: IR measurements from Venera-15 and comparison to UV data, Icarus, 105(1), 92 – 109, doi:10.1006/icar.1993.1113. Zasova, L. V., et al. (2006), Structure of Venusian atmosphere from surface up to 100 km, Cosmic Res., Engl. Transl., 44(4), 364 – 383, doi:10.1134/ S0010952506040095. 

D. Belyaev, A. Fedorova, and O. Korablev, IKI, 117997, 84/32 Profsoyuznaya Street, Moscow, Russia. ([email protected]) J.-L. Bertaux and F. Montmessin, Service d’Ae´ronomie du CNRS, BP 3, F-91371 Verrie`res-le-Buisson, France. R. Drummond, A. Mahieux, A.-C. Vandaele, and V. Wilquet, Belgian Institute for Space Aeronomy, 3 avenue Circulaire, B-1180 Brussels, Belgium.

10 of 10

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114, E00B34, doi:10.1029/2008JE003094, 2009

for

Full Article

Evidence for carbonyl sulfide (OCS) conversion to CO in the lower atmosphere of Venus Yuk L. Yung,1 M. C. Liang,2,3 X. Jiang,4 R. L. Shia,1 C. Lee,1 B. Be´zard,5 and E. Marcq5 Received 31 January 2008; revised 30 December 2008; accepted 9 January 2009; published 26 March 2009.

[1] The chemical regimes in the atmosphere of Venus vary from photochemistry in the

middle atmosphere to thermal equilibrium chemistry in the lower atmosphere and the surface. Many chemical cycles have been proposed, but few details about these cycles are fully verified by comparison between observations and modeling. Recent high-quality data of carbonyl sulfide (OCS) and CO from ground-based and Venus Express observations provide a unique opportunity to test our understanding of chemistry and transport in the lower atmosphere of Venus. The spatial distributions of OCS and CO in the atmosphere reflect a sensitive balance between chemistry and transport. On the basis of our updated photochemical model and winds from Lee et al.’s (2007) general circulation model, we study the chemistry and transport in a simplified two-dimensional chemistry-transport model. OCS is produced by heterogeneous reactions on the surface; the middle atmosphere is a net sink for OCS. The combination of data and modeling provides strong evidence for the loss of OCS by conversion to CO. The detailed chemical mechanism is currently unknown, although a number of speculations have been proposed. The sensitivity of the distributions of OCS and CO to model parameters is reported. Citation: Yung, Y. L., M. C. Liang, X. Jiang, R. L. Shia, C. Lee, B. Be´zard, and E. Marcq (2009), Evidence for carbonyl sulfide (OCS) conversion to CO in the lower atmosphere of Venus, J. Geophys. Res., 114, E00B34, doi:10.1029/2008JE003094.

1. Introduction [2] Sulfur chemistry is critical to the composition of the Venus atmosphere, and four sulfur species have been firmly identified: SO2, SO, OCS, and H2SO4 (vapor and aerosol). Mills et al. [2007] have recently carried out an extensive review of chemistry in the atmosphere of Venus, and the reader is referred to this paper for a summary of previous results. As noted in this review, there are two parts to the chemistry of sulfur species in the atmosphere of Venus. On the surface of Venus (and possibly in the dense, hot lower atmosphere) the chemistry is dominated by thermodynamic equilibrium chemistry [Fegley et al., 1997]. Near and above the cloud tops, the chemistry is driven by solar UV radiation. Thus, the partitioning of sulfur among the different species represents a competition between thermodynamic equilibrium chemistry at the surface and in the lower atmosphere and photochemistry in the middle atmosphere. This underscores the importance of transport and mixing in

1 Division of Geological and Planetary Sciences, California Institute of Technology, Pasadena, California, USA. 2 Research Center for Environmental Changes, Academia Sinica, Taipei, Taiwan. 3 Graduate Institute of Astronomy, National Central University, Jhongli, Taiwan. 4 Department of Earth and Atmospheric Sciences, University of Houston, Houston, Texas, USA. 5 LESIA, Observatoire de Paris, Meudon, France.

determining the distribution of chemical species in the atmosphere of Venus. [3] Early works on sulfur chemistry on Venus by Prinn [1975, 1978, 1979] is based on a prediction by Lewis [1970] of sulfur species with mixing ratios of 60 ppmv for OCS, 6 ppmv for H2S, and 0.3 ppmv for SO2. Prinn [1975] suggested a scheme of photochemical formation of sulfuric acid and polysulfur from carbonyl sulfide OCS. [4] The primary sulfur carrier from the deep atmosphere and surface to the middle atmosphere is OCS. Near and above the cloud tops, OCS readily dissociates, releasing the S atom.
 OCS þ hn ! CO þ S 1D

ð1Þ

where S(1D) is the first electronically excited state of the S atom and is 1.145 eV above the ground S(3P) state. The most likely fate of S(1D) is quenching:

 S 1D þ CO2 ! CO2 þ S 3P

ð2Þ

The other branch of (2) forming CO + SO is endothermic. If there is a source of oxygen atoms (such as photodissociation of CO2), S undergoes oxidation to SO2, which subsequently forms SO3 and H2SO4 and condenses to form the cloud layers as follows. [5] The S atom gets oxidized to SO by reacting with O and O2

Copyright 2009 by the American Geophysical Union. 0148-0227/09/2008JE003094$09.00

E00B34

S þ O þ M ! SO þ M

ð3Þ

S þ O2 ! SO þ O

ð4Þ 1 of 16

YUNG ET AL.: OCS AND CO ON VENUS

E00B34

Table 1. Mixing Ratios of CO2 and Trace Gases at the Lower Boundary of 58 km Species

Mixing Ratio

Reference

CO2 H2 O CO HCl SO2 OCS

0.965 2.6  105 3.0  105 4.0  107 1.3  104 3.0  107

Yung and DeMore [1982] Mills et al. [2007], Bertaux et al. [2007] Yung and DeMore [1982] Yung and DeMore [1982] Mills et al. [2007], Be´zard et al. [1993] Krasnopolsky [2008], this work

Further oxidation to SO2 can proceed via the three-body reaction SO þ O þ M ! SO2 þ M

E00B34

emphasis on carbonyl sulfide (OCS) and carbon monoxide (CO). Section 2 describes the mechanistic aspects of this work, consisting of three key components: (1) an updated 1-D photochemical model above the cloud tops, (2) a 2-D transport model based on winds and diffusivities derived from a general circulation model (GCM), and (3) a simplified chemical model for OCS and CO for the entire atmosphere. The main modeling results for OCS and CO, along with comparisons with recent observations, and model sensitivity are discussed in section 3. Section 4 examines the implications of the model, especially on the nature of the unidentified sink reaction for OCS. More speculative results are discussed in Appendix A. The conclusions are stated in section 5.

ð5Þ

2. Photochemistry and Transport Catalytic oxidation by ClO is also possible SO þ ClO ! SO2 þ Cl

ð6Þ

The rate for reaction (6) has been measured in laboratory studies [Sander et al., 2006], and reaction (6) accounts for 10– 20% of the loss of SO at 66– 80 km altitude in recent photochemical models [Pernice et al., 2004; Mills and Allen, 2007]. Note that the net result is the oxidation of S to SO2, and eventually to H2SO4. [6] The above scenario dominates primarily above the cloud tops (located at 60 km above the surface). Below the cloud tops, oxygen is scarce. In the absence of an oxygen source, S reacts with other S bearing species to form polysulfur, Sx. S atoms can react to produce S2 or S2 may be produced via coupled S-Cl chemistry involving chlorosulfanes [Mills and Allen, 2007]. Production of S3 is possible through successive addition reactions. S3 is the chemical analog of ozone, known as thiozone. As the number of sulfur atoms increases, the polyatomic sulfur compounds tend to have lower-saturation vapor pressures. It is convenient to name all sulfur species beyond S3 ‘‘polysulfur’’ or Sx. The production of Sx is part of what has been termed the ‘‘slow atmospheric sulfur cycle’’ [von Zahn et al., 1983], which is completed by decomposition reactions in the lower atmosphere. In the UV region Sx absorbs strongly, and it may be the principal constituent of the unidentified UV absorber in the upper atmosphere of Venus [Toon et al., 1982]. Krasnopolsky and Pollack [1994] and Krasnopolsky [2007] presented detailed models of OCS and Sx chemistry. However, critical components of their chemical model are not based on known laboratory kinetics. [7] Therefore, the large picture is that OCS is produced at the surface, transported to the middle atmosphere, where it is destroyed by photolysis above the cloud tops. However, between 1 and 10 bars (30 to 50 km), there may be an additional sink, whose identity is currently unknown and which is a major focus of this paper. The compelling observational evidences are (1) vertical gradient of OCS between 30 and 50 km, (2) latitudinal gradient of OCS in this region, and (3) complementary latitudinal gradient in CO, suggesting OCS to CO conversion. [8] In this paper, we report a combined study of photochemistry and transport in a two-dimensional (2-D) chemistrytransport model (CTM) for the atmosphere of Venus, with

2.1. 1-D Photochemical Model [9] The UV radiation is averaged diurnally before photolytic calculations, and the latitude dependence is carefully taken into account (see Liang et al. [2005] and references contained therein for a detailed description of the model.) The 1-D model [Yung and DeMore, 1982; Mills et al., 2007] is from 58 to 112 km. The vertical eddy mixing coefficients equal those obtained previously [Yung and DeMore, 1982]. Additional UV attenuation caused by absorbers estimated by Crisp [1986] is included. The chemical scheme used in this paper is taken from Mills et al. [2007]. The volume mixing ratios of CO2, H2O, CO, HCl, SO2, and OCS at the lower boundary are given in Table 1. For the other species, free escape through the lower boundary is allowed. The upper boundary is closed for all species. [10] Figure 1 shows the mixing ratios of CO2 and the most abundant trace gases in the model. The abundance of CO increases with height because there is a source of CO from photolysis of CO2 and OCS higher up. CO is removed by catalytic chemistry involving Cl radicals derived from HCl photolysis, which results in a decline of HCl with altitude. Both SO2 and H2O are removed by the formation of H2SO4. As H2SO4 aerosols are removed from the model, they represent a net sink of SO2 and H2O, thereby explaining their rapid decrease with altitude. OCS is destroyed by photolysis above the cloud tops. The secondary peak around 85 km is caused by the reaction S þ ClCO ! OCS þ Cl

ð7Þ

with the S atom ultimately derived from SO2 þ hn ! SO þ O

ð8Þ

SO þ hn ! S þ O

ð9Þ

Figure 2 shows the mixing ratios of O, O2, and O3. The mesosphere is a net source of oxygen from CO2 photolysis. Oxygen is rapidly consumed by catalytic recombination with CO and oxidation of SO2 to SO3, followed by formation of H2SO4. The latter is removed from the model. [11] Figure 3 presents the mixing ratios of major hydrogen species. H2O is the major carrier of hydrogen to the

2 of 16

E00B34

YUNG ET AL.: OCS AND CO ON VENUS

E00B34

Figure 1. Vertical profiles of mixing ratios of CO2, CO, HCl, SO2, and OCS predicted by the 1-D photochemical model. The boundary conditions are summarized in Table 1.

middle atmosphere above the cloud tops. Photolysis of H2O produces H2 and radicals H, OH, HO2, and H2O2 (odd hydrogen). The odd hydrogen species play a minor role in the photochemistry of Venus. Reactive chlorine bearing radicals are shown in Figure 4. They are derived from HCl photolysis, which also contributes to the production of odd hydrogen. The chlorine bearing species play a major role in the stability of the CO2 atmosphere [Yung and

DeMore, 1982]. There is a possible coupling to sulfur chemistry but the kinetics is highly uncertain [Mills and Allen, 2007]. [12] The possibility of Sx formation is illustrated in Figure 5. The principal sources of S are photolysis of OCS and SO. In the absence of O2, formation of higher polymers of sulfur is possible. The column production rates are summarized in Table 2. The Sn species are in photo-

Figure 2. Same as Figure 1 for oxygen-bearing molecules, O2, O, and O3. 3 of 16

E00B34

YUNG ET AL.: OCS AND CO ON VENUS

E00B34

Figure 3. Same as Figure 1 for hydrogen-bearing molecules, H2O, H2, H, OH, HO2, and H2O2.

chemical equilibrium; that is, production and loss (primarily by photolysis) are equal, except for S8. The model stops at S8, which is removed from the atmosphere in the model. The possible role of this chemistry in the lower atmosphere is discussed in Appendix A. [13] The 1-D photochemical model with full chemistry provides the basis for constructing a simplified chemical model for OCS and CO for the 2-D chemistry-transport model. This will be taken up in section 2.3.

2.2. Transport Model [14] The general circulation model (GCM) used to calculate the stream function uses the dynamical core of the Hadley Centre general circulation model [Cullen, 1993] with a 5° horizontal grid staggered as an Arakawa B grid [Arakawa and Lamb, 1977] from pole to pole and a 31-level hybrid sigma pressure vertical coordinate from surface to about 100 km. Forcing and dissipation are provided by linear parameterizations of the radiative forcing (cooling to space) and boundary layer (Rayleigh friction). A qualita-

Figure 4. Same as Figure 1 for chlorine-bearing molecules, Cl, Cl2, ClO, ClCO, and ClCO3. 4 of 16

E00B34

YUNG ET AL.: OCS AND CO ON VENUS

E00B34

Figure 5. Same as Figure 1 sulfur-bearing molecules, SO, S, S2, S3, S4, S5, S6, and S7. tively realistic radiative forcing is used to drive an equatorial superrotation and low wave number planetary waves in the middle atmosphere. No topography or diurnal cycle is used, and no active chemical processes are considered within the GCM. We briefly summarize below the essential parts of the model and make a comparison with other model results available in the literature. Full details of the GCM and this experiment are given by Lee et al. [2007]. [15] When a similar radiative forcing is used, the circulation produced by other GCMs [Yamamoto and Takahashi, 2003; Hollingsworth et al., 2007; Herrnstein and Dowling, 2007] is at least as strong as the circulation produced by Lee et al.’s [2007] GCM shown here, suggesting the circulation is not unreasonable for the radiative forcing used. When a radiative forcing with less heating in the lower atmosphere is used [e.g., Hollingsworth et al., 2007] the circulation in the lower atmosphere is much weaker, but the superrotation is also much weaker and does not compare well to the observed Venus atmosphere. However, the middle atmosphere circulation still resembles a Hadley-like overturning. [16] Observations by Venus Express [Limaye, 2007] and Pioneer Venus [Schubert, 1983] suggest that the circulation, at least at cloud level, is stronger than in the GCM shown here. However, there are few observations at depth, making the calculations in the deep atmosphere somewhat speculative. These results suggest that the ‘‘age of air’’ calculation shown here (see later discussion) should be more accurate in the middle atmosphere than in the lower atmosphere, but physical and observational constraints suggests even the lower atmosphere values presented should be within an order of magnitude of the actual values. [17] The meridional circulation is calculated from wind and temperature from GCM [Lee et al., 2007]. The data from the GCM are in 5°  5° latitude-longitude resolution. It has 31 vertical levels from 92 bar to 103 bar. We choose the same method as that of Jiang et al. [2004] to calculate

the transformed meridional circulation for the Venus. First, the three-dimensional (3-D) meridional mass flux, y P (l, 8, p), is determined by y P ðl; 8; pÞ ¼

2p a cos 8 g

Z

p

V ðl; 8; p0 Þdp0

ð10Þ

o

where a is the Venus radius, l, 8, and p are the longitude, latitude and pressure, V is the meridional velocity, and g is the gravitational acceleration rate. Then we interpolate the 3-D meridional mass flux to isentropic surfaces, using a mass-conserving linear interpolation scheme [Juckes et al., 1994]. The 2-D isentropic mass stream function, y q (8, q), is derived by zonal averaging of the 3-D isentropic meridional mass flux, y q (l, 8, q), along isentropes. Finally, we interpolate the 2-D isentropic mass stream function, y q (8, q), to pressure coordinates and scale by the density to produce the pressure surface stream function, y P (8, p), which is used to drive the 2-D CTM. The resulting pressure surface meridional circulation, y P (8, p), shown in Figure 6 (left), is like that for the Hadley cell in the Table 2. Column Production Rates of Sulfur Species Above the Cloud Topsa Species

Principal Reactions

Column Rate

S S2 S3 S4 S5 S6 S7 S8

SO + hn, S2 + O S + S + M, S + OCS S + S2 + M S2 + S2 + M S + S4 + M S + S5 + M, S3 + S3 + M S + S6 + M S4 + S4 + M, S2 + S6 + M

8.2  1011 2.4  1011 1.6  1010 2.7  1010 7.3  109 6.1  109 1.6  109 5.2  109

a Sn, sulfur species. Units are in molecules cm2 s1. The principal source of Sn is COS photolysis, whose column rate is 7.5  1010 molecules cm2 s1. At higher altitudes, SO photolysis provides an additional source of S atoms.

5 of 16

E00B34

YUNG ET AL.: OCS AND CO ON VENUS

E00B34

Figure 6. (left) Mass stream function y and (right) horizontal eddy mixing coefficient Kyy calculated from Lee et al. [2007]. Units are 109 kg s1 for y and 109 cm2 s1 for Kyy. Minimum Kyy is set to be 1010 cm2 s1, a value that is chosen to be large enough that small-scale transport (resulting from the derivation of y) is reduced but small enough that the transport remains dominated by y. terrestrial atmosphere. There is upwelling in the tropics, followed by downwelling in the high latitudes. The total mass flux is 3000  109 kg s1, a value that should be compared to the terrestrial value of 200  109 kg s1 [see Peixoto and Oort, 1992, Figure 7.19]. Since the atmosphere of Venus is about 100 times more massive than the terrestrial atmosphere, this implies that the atmosphere turns over in about a decade, as the Earth’s atmosphere turns over in a year. This result is consistent with the age of air experiment described later. Note that the latitudinal extension of the circulation is much greater than that for the terrestrial atmosphere owing to the near absence of the Coriolis effect. [18] The horizontal mixing coefficients, Kyy, is calculated on the isentropic surfaces Jiang et al. [2004, equation (2)]. Then we interpolate it from the isentropic surfaces to the pressure surfaces for use in the 2-D CTM. Figure 6 (right) shows a latitude-altitude profile of Kyy. The values for Kyy are larger in the midlatitudes than the tropics owing to the presence of enhanced wave activities in the midlatitudes. Although the transport in the atmosphere is dominated by advection, the effect of horizontal diffusion is non trivial in smoothing out gradients in chemical tracers. [19] To visualize the effect of the circulation on transport of chemical species in the atmosphere, we compute the age of air, a quantity that has proved to be extremely useful in understanding the stratosphere-troposphere exchange in the terrestrial atmosphere [Hall and Waugh, 1997]. The age of air is obtained by following the trend of an inert tracer, whose abundance at the surface increases linearly with time. Figure 7 shows the age of air in the lower atmosphere of Venus derived from the circulation shown in Figure 6. It

takes about 10 (Earth) years for an air parcel released at the surface to reach the middle atmosphere at the cloud tops. Since the radiative forcing does not include a diurnal cycle, the GCM does not simulate the large temperature variations seen above 90 km [Bertaux et al., 2007] associated with the variation in radiative forcing. As a result the age of air is only approximate in the upper atmosphere. [20] There is a simple physical interpretation of the mass stream function [see Shia et al., 1989, equation (12)]: y P (8, p) is the integral from latitude = 90 to 8 for the vertical mass flux across the pressure = p level, or the integral from the surface to pressure = p for the horizontal mass flux across the latitude = 8. The convention is positive for upward and northward flows. For later reference, it is convenient to divide the atmosphere into nine boxes, as shown in Figure 8. Using the values of the stream function from Figure 6 (left) and the formulas by Shia et al. [1989], we derive the appropriate mass fluxes in and out of each box in units of 109 kg s1. The numbers have been rounded off, and so the values are approximate. Note that the divergence of fluxes from each box is zero owing to conservation of mass. As we will show later, the central box is crucial for our investigation of the OCS and CO budget. This simple diagram will allow us to make a qualitative estimate of the chemical production and loss rates of OCS and CO in this region of the atmosphere from observed gradients in the concentrations of the tracers. 2.3. 2-D Chemistry-Transport Model With Simplified Chemistry [21] The 2-D Caltech/JPL photochemical model has been employed to simulate the meridional distribution of hydro-

6 of 16

E00B34

YUNG ET AL.: OCS AND CO ON VENUS

E00B34

Figure 7. Age of air (Earth year) derived from the circulation shown in Figure 6. carbons in the atmosphere of Jupiter [Liang et al., 2005] and the isotopic composition of ozone in the terrestrial atmosphere [Liang and Yung, 2007]. In principle, such 2-D models are similar to that used for Titan [Lebonnois et al., 2001] and have been useful in modeling the interaction of

chemistry and transport in planetary atmospheres. As noted earlier, we obtain our transport component of the model from a GCM. [22] We extend the base model [Yung and DeMore, 1982; Mills et al., 2007], which sets the lower boundary at the

Figure 8. Mass fluxes in units of 109 kg s1 across the boundaries of nine boxes in the atmosphere of Venus. The boundaries at the bottom, top, and sides are impermeable. The numbers are taken and rounded off from Figure 6. 7 of 16

YUNG ET AL.: OCS AND CO ON VENUS

E00B34

E00B34

Figure 9. Model profiles of (top) OCS and (bottom) CO, in units of ppmv, using the model that includes reaction (R11). See text.

cloud tops, to the entire atmosphere from 0 to 112 km. Above the planetary boundary (2 km), the vertical eddy mixing coefficients are uniformly set at 100 cm2 s1 and we assume that the transport in this region is dominated by the advection. In the planetary boundary, the vertical eddy coefficients are set at 104 cm2 s1, a value high enough simulating for the mixings/transport of matters between the troposphere and the boundary, where the mixing ratios of OCS and CO are set to be 23 and 16.6 ppmv, respectively. Since we are primarily interested in the chemistry of OCS and CO the lower atmosphere, the chemistry in the atmosphere above the cloud tops only serves as the upper boundary and may be greatly simplified. Thus, the chemical model is reduced to about 10 key reactions involving production and destruction of OCS and CO. Similar simplification has been carried out for the chemistry of ozone in the Earth’s atmosphere [Liang et al., 2006]. In order to successfully simulate the distribution of OCS and CO in the lower atmosphere, we have to postulate a loss reaction for converting OCS to CO via reaction with X: OCS þ X ! CO þ Y

ð11Þ

If we assume that R11 is a bimolecular reaction, then the loss coefficient is given by L11 ¼ k11  ½X

ð12Þ

where k11 is the bimolecular rate coefficient [see Yung and DeMore, 1999, chapter 3]. By trial and error, we found that L11  108 s1 in the region around 30 km. We will discuss the chemical and physical mechanisms for this additional sink for OCS later. The identity of X in (12) remains elusive. Candidate molecules include SO3, (SO)2 and S.

3. Model Results 3.1. Modeling Results and Comparison With Observations [23] Figure 9 shows the latitude-altitude plots of OCS and CO in the region from 30 to 40 km. The vertical distribution of OCS is consistent with its source at the surface, transport to the middle atmosphere, where it is destroyed. The falloff of OCS with altitude is not as large as the observed value. This may be due to inadequacy in the transport model or

8 of 16

E00B34

YUNG ET AL.: OCS AND CO ON VENUS

E00B34

Figure 10. Vertical profiles of OCS computed by the 2-D model: global mean (solid curves) and at 65°N (dashed curves). (left) Loss by photolysis of OCS alone. (right) Additional sink (R11) is included. See text.

loss rate. According to the schematic in Figure 8, the export of OCS poor air from the tropics and via the downward portion of the Hadley cell results in lower OCS mixing ratios at middle and high latitudes. The pattern of CO is similar to that for OCS, except that the gradients are now reversed. In this case, there is a source of CO from the mesosphere and an additional source from reaction (R11). Comparison between model predictions and observations will be discussed as follows. [24] Figure 10 (left) shows the vertical profiles of OCS from our model with loss only by photolysis. Although the model correctly simulates the falloff of OCS above the cloud tops, the scale height of the OCS density at 30 km is close to the atmospheric scale height of 10 km (i.e., small vertical gradient in mixing ratio), in contradiction to the observed scale height of about 2 km (see below). The reason is that the photolysis coefficient of OCS falls rapidly below the cloud tops. In order to produce the correct scale height of OCS at 30 km, we include an additional parameterized sink for OCS in the model. The results are shown in Figure 10 (right) and are in agreement with the observations. [25] The falloff of the OCS profile around 30 km is a well-documented observation [see Marcq et al., 2005, 2006, and references therein]. The slope is dlog½OCS=dlogP ¼ 5 1

ð13Þ

where P is pressure. The corresponding scale height at this altitude is about 2 km. Higher up in the atmosphere, above 60 km, the OCS concentrations are in fair agreement with recent measurements by Krasnopolsky [2008]. The poorer

agreement at 70 km may be due to the unrealistic circulation above the cloud tops. [26] The latitudinal distribution of OCS at 33 km from our model is presented in Figure 11 (top). The Venus Express VIRTIS (asterisks) and ground-based telescope IRTF (diamonds) are shown by special symbols. Dotted curve represents the cosine function of latitude. The model is in satisfactory agreement with the data. The reasons for the increase in the mixing ratio of OCS away from the tropics were discussed earlier. The error bars are 1-s standard deviation intervals, for both VIRTIS and IRTF retrievals. [27] We assume that the variation of OCS at the probed altitude (33 km) was caused by vertical translations of the reference profile given by Pollack et al. [1993]. A more detailed discussion of the considered set of OCS vertical profiles is available in the work by Marcq et al. [2005, 2008]. Orbits 98, 110, 111, 134, 136, 258, and 277 of the Venus Express spectra, have been used, mostly in science case 2 (off-pericenter observations), because of the need of long integration times. The spectra were provided by the high spectral resolution IR channel from the VIRTIS instrument (VIRTIS-H). The resolving power R in the used order of dispersion (order 5) is close to 1500 between 2.42 and 2.46 mm. The interpretation of these spectra is still in progress, so the retrievals in upcoming publications may differ significantly, although the main trends should persist at least qualitatively. On the other hand, the Earth-based retrievals have already been published by Marcq et al. [2005, 2006]. The used spectra were acquired on 13 August 2004 using the order 3 of the SpeX spectrometer (R  2000

9 of 16

E00B34

YUNG ET AL.: OCS AND CO ON VENUS

E00B34

Figure 11. (top) Model OCS distribution at 33 km (solid curve). The Venus Express (asterisks) and ground-based telescope IRTF (diamonds) are shown by symbols. Dotted curve represents the cosine function of latitude. (bottom) Similar results for CO. from 1.92 mm to 2.48 mm, which includes the whole 2.3 mm spectral window) at the NASA IRTF. Only the retrievals from slit positions 1 and 2, as defined by Marcq et al. [2006, Figure 1], are used. [28] Figure 11 (bottom) shows the model latitudinal distribution of CO at 36 km. The latitudinal structure of CO is the complement of that of OCS in Figure 11 (top). This gives rise to the hypothesis that the chemical destruction of OCS leads to the production of CO, as suggested by reaction (R11). We will return to this issue later. 3.2. Sensitivity of Model to Input Parameters [29] The agreement between the model and data shown in Figure 11 is far from perfect. It is important to investigate the model sensitivity of these results to input parameters. The most important input to the model is the stream function that is responsible for advection in the model. We note that the model profile of OCS in Figure 11 is flatter than the data; this is true also for CO (see Figure 11). One way we can enhance the curvature of the chemical species is to fine-tune the stream function and Kyy and the chemical reaction rates. There are many ways to modify the parameters in the model. After many trials, we conclude that the latitudinal gradients in OCS and CO are sharpened by (1) narrowing the width and increasing the amplitude of the Hadley cell, (2) enhancing the unidentified chemistry R11, and (3) reducing horizontal diffusion.

[30] The modifications of the standard model are summarized in Table 3. Model A is same as the standard model, except that the peak of the stream function is modified by the following factors. We have nine latitude boxes evenly spaced between south pole (SP) and equator (EQ). Let the stream function at box i be multiplied by a factor fi, where i = 1 is the SP box at 85° and i = 9 is the EQ box at 5°. In model A, we set f6 = 2, allowing it to fall back linearly to 1 at SP and EQ. The changes in the northern hemisphere, for i = 10 to 18, are a mirror image of those in

Table 3. Sensitivity of Latitudinal Gradients in OCS and CO at 30 km to Model Parametersa Model

D[OCS] (ppmv)

D[CO] (ppmv)

Parameter Change

Standard A B C D E F G

1.65 1.83 1.77 2.15 2.24 2.63 2.80 1.22

1.51 1.84 1.88 1.57 1.78 2.50 3.00 1.08

y mid  2 y mid  3 Same as B, R11  3 Same as C, y mid  3 + y trop  2 Same as D, Kyy/4 Same as D, Kyy/10 Same as Standard, R11 = 0

a

D[OCS] = [OCS](0°) – [OCS](60°); D[CO] = [CO](0°) – [CO](60°). The atmosphere is divided into three vertical regions (see Figure 8). Stream functions in the middle and bottom regions of the atmosphere are y mid and y trop, respectively.

10 of 16

E00B34

YUNG ET AL.: OCS AND CO ON VENUS

E00B34

Figure 12. Same as Figure 11 for model F (see Table 3 and text).

the southern hemisphere. The purpose of this change is to make the Hadley cell narrower, so that we can sharpen the latitudinal gradient of OCS and CO. A measure of the latitudinal gradient is Dc = c(0°) – c(60°), where c is the mixing ratio of OCS or CO. In model A, we have D[OCS] = 1.83 ppmv, as compared to the value of 1.65 ppmv in the standard model. The corresponding values for D[CO] are 1.84 and 1.51 ppmv. Therefore, the gradients of OCS and CO in model A are sharper than those in the standard model. [31] Further accentuation of the gradients is achieved if we set f6 = 3 in model B (see Table 3). In model C, we enhance the chemical reaction (R11) by a factor of 3. The chemistry change has a larger effect on OCS than CO. In model D, we enhance the tropical stream function by a factor of 2; that is f9 = 2. Horizontal diffusivity acts to remove the latitudinal gradient in the model. model E studies the effect of reducing Kyy by a factor of 3. This results in increases of the absolute values of Dc in the model. This factor is increased to 10 in model F, which represents the model with the greatest gradients studied so far. Figure 12 presents a comparison of data and model results for OCS and CO in model F. The agreement is somewhat better than those in Figure 11, but crucial data at higher latitudes are lacking. The sharp edges in the model profiles are produced by advection and they could not be smoothed out owing to the low Kyy in this model.

[32] Finally, in model G we test the relative importance of chemistry (R11) versus boundary conditions in determining Dc in the model. This run is same as the standard model, except that tropospheric chemistry R11 = 0. We note that in this case there remains a residual gradient due to downwelling of air at middle and high latitudes from the stratosphere (see Figure 8), where OCS (CO) is low (high). 3.3. Is the Sum of OCS and CO a Constant? [33] The above discussion strongly suggests the possibility that the latitudinal variation in OCS and CO seen in Figure 11 is driven by the conversion of OCS to CO. In this case the sum of [OCS] and [CO] should be a constant. Figure 13 shows the data and model for [OCS] + [CO], demonstrating that the sum is approximately a constant. Inspection of Table 3 suggests that D[OCS] + D[CO] is roughly zero, further verifying that the changes are the result of OCS to CO conversion.

4. Implications of the Model 4.1. Sulfur Budget [34] Our model which is consistent with the observations suggests that the global rate of destruction of OCS is 23,000 Tg-S a1 (teragram of sulfur atom equivalence per year, i.e., 1 Tg-S a1 = 3  1010 moles a1 = 2  1034 OCS molecules a1), a value that should be compared to the total volcanic source of 10 Tg-S a1 for the Earth [see,

11 of 16

E00B34

YUNG ET AL.: OCS AND CO ON VENUS

E00B34

Figure 13. Comparison of model OCS + CO and data. Solid and dashed curves are for standard model and model F, respectively. Symbols are for two observations. e.g., Seinfeld and Pandis, 1998]. The mean destruction rate is equivalent to 3.0  1012 molecules cm2 s1, which is 43 times the corresponding rate of OCS destruction above the cloud tops (see Table 2). The middle atmosphere is a net sink for OCS. We argue that the large implied source of OCS is unlikely to be supplied by volcanic emission. For example, a large volcano like Pinatubo delivered about 10 Tg-S into the terrestrial stratosphere in 1991 [Bluth et al., 1992]. To produce fluxes consistent with our model would require 2300 Pinatubo-sized eruptions per year. There is no geochemical or geological evidence in support of enhanced volcanic activities of this magnitude on Venus. [35] It is more likely that OCS is produced by heterogeneous reactions on the surface from CO and polysulfur (Sx), or between CO and CO2 and surface minerals (e.g., pyrite), as discussed by Mills et al. [2007, section 3.2]. The reaction S2 + CO ! OCS + S is favored by Hong and Fegley [1997] and Krasnopolsky [2007]. The CO ultimately comes from CO2 photolysis above the cloud tops and destruction of OCS in the middle atmosphere. Its mixing ratio remains fairly high, 105 near the surface [see Mills et al., 2007, Figure 3]. It is a challenge for the laboratory chemists to confirm and quantify these surface reactions. [36] We shall argue that the OCS sink predicted by the model is a direct consequence of the circulation of the atmosphere and the observed vertical and horizontal gradients in OCS, independent of the details of the model. Referring to the middle box in Figure 8, the mass flux entering the box is 8in = 1630  109 kg s1. Thus, the S mass flux entering this box is 8in  [OCS]in  32/44. The mass fluxes out of this box are the upward flux at 50 km (where there is almost no OCS; see Figure 10) and the two horizontal fluxes 8out = 410 + 390 = 800  109 kg s1 at

midlatitudes. Therefore, the S mass flux out of the center box is 8out  [OCS]out  32/44. Putting in approximate values from the data presented in Figure 11, [OCS]in = 3.3 ppmv and [OCS]out = 2.5 ppmv, we have the next sink for OCS equal to 76, 000 Tg-S a1, a value that is within a factor of 3 of our earlier number. The overestimate is not surprising, as we attribute the cause of the latitudinal gradient entirely to loss by R11, whereas comparison between model G and the standard model suggests that part of the gradient is caused by the upper boundary via the downward fluxes at middle to high latitudes. 4.2. Previous Work on the Unidentified Reaction [37] Krasnopolsky and Pollack [1994] were the first to propose chemical destruction of OCS in the lower atmosphere of Venus. Krasnopolsky [2007] quantitatively modeled the following reactions for destroying OCS in the lower atmosphere: H2 SO4 ! H2 O þ SO3

ð14Þ

SO3 þ OCS ! CO þ ðSOÞ2

ð15Þ

ðSOÞ2 þOCS ! CO þ SO2 þ S2

ð16Þ

———————————————————— net H2 SO4 þ 2 OCS ! H2 O þ CO2 þ CO þ SO2 þ S2 ð17Þ

According to Krasnopolsky’s [2007] estimate, [SO3] 5  1013 cm3 and k15  5  1023 cm3 s1 at 30 km, resulting

12 of 16

YUNG ET AL.: OCS AND CO ON VENUS

E00B34

in L15 = k15[SO3] = 2.5  109 s1. This is of the same order of magnitude but somewhat smaller than L11 from (11). Similar estimate shows that the second Krasnopolsky reaction (16) yields an L16 = k16[(SO)2] that is larger than L15. The combination of L15 and L16 is of the same magnitude as L11. [38] A major weakness of this scheme is that reactions (15) and (16) are not based on laboratory studies. Laboratory experiments are urgently needed to verify or disprove these reactions. In view of these uncertainties associated with reactions (15) and (16), it is reasonable to propose alternatives. Three chemical schemes are discussed as follows. 4.3. Other Possibilities for the Unidentified Reaction [39] We will briefly describe three additional possible mechanisms for the destruction of OCS on Venus. The first two are based on heterogeneous reactions on or in particles. The third mechanism is the photosensitized dissociation of OCS via the photochemistry of Sx. [40] Chen et al. [2007] measured the rate of oxidation of OCS to CO2 on the surface of hematite. The uptake coefficient (g) lies in the range 107 to 1011. The loss rate of OCS in the atmosphere can be estimated using the expression L11 ¼ 1=4gvANa

ð18Þ

where v isffi the mean speed of the impacting molecule = pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 8kT =pm, A is the mean surface area of each hematite grain, and Na is number density of hematite grains. Since L11  108 s1 and v  104 cm s1, ANa must be  105 to 101 cm2 cm3. For comparison, ANa is  3  106 in the terrestrial stratosphere for dust after the eruption of the El Chichon volcano [Michelangeli et al., 1991]. Therefore, while not ruled out, this reaction is probably too slow to have an impact for the destruction of OCS in lower atmosphere. [41] Dalleska et al. [2000] measured the rate coefficient for the oxidation of OCS by H2O2 in sulfuric acid solution. While the abundance of sulfuric acid is high on Venus, the limiting chemical may be H2O2, which is produced in small abundance at 80 to 90 km (see Figure 3). Thus, unless we have evidence otherwise, we consider this mechanism unlikely. [42] The third proposal is the photosensitized dissociation of OCS via the photochemistry of Sx. The idea is based on with analogy with the photosensitized dissociation of H2O in the troposphere of the Earth:
 O3 þ hn ! O2 þ O 1D

ð19Þ


 H2 O þ O 1D ! OH þ OH

ð20Þ

In this case, the dissociation of H2O could not occur directly because the short-wavelength UV photons required to dissociate H2O (l < 200 nm) are absorbed by O2 and O3 in the stratosphere. However, the long-wavelength UV photons (l > 310 nm) can penetrate to the troposphere and photolyze O3 to produce O(1D), which then reacts with H2O

E00B34

[see Sander et al., 2006; Yung and DeMore, 1999, chapter 10]. We argue that, by analogy with (19) and (20) on Earth, the photochemistry of Sx could lead to photosensitized dissociation of OCS using long-wavelength photons that can penetrate to the lower atmosphere of Venus. The details of the Sx scheme are referred to in Appendix A. [43] The key reaction in this case is the reaction (R7) in Appendix A, OCS + S ! CO + S2. This reaction was considered by Prinn [1975] and Krasnopolsky [2007], and was measured by Lu et al. [2006] in the temperature range 298– 985 K. This is adequate for our modeling. The crucial requirement is that the number density of S atoms at 30 km must be 105 cm3. Without a source of S atoms from the Sx photochemistry in Appendix A, Krasnopolsky [2007] predicts [S]  10 cm3, which is too low to have impact on the OCS distribution. 4.4. Testable Hypotheses [44] The nature of the unidentified reaction in the lower atmosphere of Venus can only be resolved with better atmospheric measurements and laboratory experiments. Therefore, the proposed chemical schemes constitute testable hypotheses that are useful for guiding future experiments. 4.4.1. Krasnopolsky’s Scheme [45] Reactions (15) and (16) must be demonstrated in the laboratory. The existence of SO3 follows from the thermal decomposition of H2SO4, but the existence of (SO)2 is less obvious. Both must be measured in the atmosphere of Venus. 4.4.2. Heterogeneous Reactions [46] The two heterogeneous reactions discussed in section 4.3 were studied at room temperature. It is conceivable that they could be faster at higher temperatures. Also, there may be other types of heterogeneous reactions not considered here. A thorough laboratory search is needed. 4.4.3. Photosensitized Dissociation [47] A comprehensive study of the UV radiation and Sx photochemistry in the lower atmosphere is needed to make a realistic assessment of the concentration of S atoms at 30 km. Measurements of sulfur speciation as well as UV irradiance in the lower atmosphere would be valuable.

5. Concluding Remarks [48] A simple two-dimensional chemistry and transport (2-D CTM) model is used to study the spatial distributions of OCS and CO in the lower atmosphere of Venus. The residual circulation and horizontal eddy diffusivities are derived from winds from Lee et al.’s [2007] general circulation model. The Hadley circulation provides rapid transport in the lower atmosphere of Venus. Mixing between the surface and the cloud tops occurs in as short as 10 years, which is significantly shorter than previous estimates of this time constant. The results for the latitude-altitude distributions of OCS and CO from our 2-D CTM are compared to recent observations (see Figure 11). A total of eight models were studied to test the sensitivity of model results to input parameters, as summarized in Table 3. High-latitude data for OCS and CO are urgently needed to distinguish between the current models, shown in Figure 11 (standard model) and Figure 12 (model F), or a more realistic

13 of 16

E00B34

YUNG ET AL.: OCS AND CO ON VENUS

E00B34

Figure A1. Schematic diagram illustrating the Sn chemistry in the lower atmosphere.

CTM. Further work is also required on the GCM used in this work in order to improve the simulated circulation in our model so that we may better constrain the effects of the unknown reaction(s). [49] There is compelling evidence for an unidentified chemical reaction that destroys OCS in the lower atmosphere of Venus around 30 km. There is strong evidence that this unidentified reaction converts OCS to CO (but not to CO2), as both data and model suggest that the sum of OCS and CO is roughly a constant (see Figure 13 and section 3.3). A major implication of the model is that the total loss rate of OCS in the lower atmosphere is about 23,000 Tg-S a1. This is a robust result that does not depend on the details of the 2-D CTM. The value can be derived from heuristic arguments based on the strength of the Hadley circulation and observed chemical gradients (see section 4.1). [50] The nature of the unidentified reaction remains controversial. A survey of recent literature and this work reveal three testable hypotheses: (1) Krasnopolsky’s [2007] reactions of OCS with SO3 and (SO)2 (see section 4.2), (2) heterogeneous reactions on hermatite or sulfuric acid particles (see section 4.3), and (3) photosensitized dissociation of OCS driven by absorption of soft UV by Sx (see section 4.3 and Appendix A). New laboratory experiments and observations are urgently needed to prove or disprove the proposed hypotheses (see section 4.4). We note that if the last speculative chemistry were confirmed, there would important implications for the modeling of the evolution of the terrestrial atmosphere [Kasting et al., 1989; Farquhar et al., 2002].

Appendix A: Hypothesis of an Innovative Chemical Cycle [51] In order to account for the additional loss of OCS below the cloud tops (see Figure 10 (right)), we propose an innovative photosensitized dissociation of OCS driven by

photolysis of Sn using photons at long wavelengths that are able to penetrate through the clouds 2½S3 þ hn ! S2 þ S ðR3Þ S4 þ hn ! S3 þ S ðR5Þ S2 þ S þ M ! S3 þ M ðR8Þ S2 þ S2 þ M ! S4 þ M ðR9Þ 2½OCS þ S ! CO þ S2  ðR7Þ   net 2 OCS ! 2 CO þ S2

ðA1Þ

The details of this innovative chemistry are summarized in Table A1 and shown in Figure A1. The reactions (Rn) refer to the reaction set listed in Table A1. Note that the scheme has the following characteristics: (1) photolysis at UV wavelengths releases S from OCS, (2) S reacts with OCS to form S2 and subsequent reactions produces Sn, (3) photolysis of Sn occurs at near UV and longer wavelengths, producing S, and (4) cycle repeats via characteristic 2. [52] The net result is summarized by (9). This scheme is known as photosensitized dissociation (for a discussion of this process see, e.g., Yung et al. [1984, section 2]). The rate coefficient for the key reaction (R7) has recently been measured in the temperature range of 298 – 985 K by Lu et al. [2006] and their expression is given in Table 1. At 500 K, k7 = 5.41  1014 cm3 s1 is much smaller than the value 5.30  1013 cm3 s1 used by Prinn [1975] but is close to the value 6.29  1014 cm3 s1 estimated by Krasnopolsky [2007]. We must emphasize the speculative nature of the proposed chemistry. There remain major gaps in our knowledge of the chemistry of Sn in Table A1 and Figure A1; new laboratory studies are needed to close the gaps. [53] In a crude model we have for Sn around 30 km the following concentrations, [S] = 1  105 cm3, [S2] = 2  108 cm3, [S3] = 7  109 cm3, [S4] = 7  109 cm3, [S8] = 4  1014 cm3. We do not go beyond S8, which has no atmospheric sink (and therefore a large concentration). It is

14 of 16

YUNG ET AL.: OCS AND CO ON VENUS

E00B34

E00B34

Table A1. List of Reactions Related to the Chemistry of Sn as Shown in Figure A1 Reaction

Rate Coefficienta 5

(R1) (R2) (R3) (R4)

OCS þ h ! CO þ S S2 þ h ! S þ S S3 þ h ! S2 þ S S4 þ h ! S2 þ S2

J1 = 3.81  10 J2 = 9.93  103 J3 = 1.15 J4 = 1.07  101

(R5)

S4 þ h ! S3 þ S

J5 = 1.07  101

(R7) (R8)

OCS þ S ! CO þ S2 S2 þ S þ M ! S3 þ M

(R9)

S2 þ S2 þ M ! S4 þ M

(R10)

S4 þ S4 þ M ! S8 þ M

(R11) (R12) (R13)

S3 þ S ! S2 þ S2 S4 þ S ! S3 þ S2 CO þ S þ M ! OCS þ M

k7 = 6.63  1020T2.57e1180/T k8 (0) = 1.00  1030 k8 (1) = 3.00  1011 k9 (0) = 2.20  1029 k9 (1) = 1.00  1011 k10 (0) = 1.00  1030 k10 (1) = 3.00  1011 k11 = 3.00  1011 k11 = 3.00  1011 k13 (0) = 4.00  1036 e1940/T k13 (1) = 1.00  1015

Reference Molina et al. [1981] Moses et al. [2002, and references therein] Moses et al. [2002, and references therein] Moses et al. [2002, and references therein]; Branching ratio estimated Moses et al. [2002, and references therein]; Branching ratio estimated Lu et al. [2006] Mills [1998] Mills [1998] Mills [1998] Mills [1998] Mills [1998] Mills [1998]; Estimated on the basis of analogy with O + CO + M ! CO2 + M

The photolytic coefficients are given for a diurnally averaged model at 30°N at the top of the atmosphere (s1). Two-body and three-body rate coefficients are given in units of cm3 s1 and cm6 s1, respectively; k(0) and k(1) refer to rate coefficient in the low and high pressure limit, respectively. a

transported to the surface, where surface reactions will recombine the sulfur and CO to form OCS, thus completing the cycling of OCS. Note that S3 has been tentatively identified by UV data from Venera 14, and Sx is a leading candidate for the unidentified UV and blue absorber [see Mills et al., 2007, sections 3 and 6]. The amount quoted for S3 by Maiorov et al. [2005] is 0.03 to 0.1 ppb. Bertaux et al. [1986] reported 5 to 25 ppmv of S8 from 25 to 45 km from analysis of UV data obtained by Venera 11 and 12. Note that 10 ppmv of S8 at 30 km corresponds to [S8]  1  1015 cm3, which is of the same order magnitude as what is needed in the model. However, the authors concluded that such values might be 2 orders of magnitudes too high. One resolution of the conflict is that the UV absorption in the lower atmosphere of Venus may not be homogeneous, creating gaps through which UV may penetrate to much deeper levels than an average model would predict. [54] Acknowledgments. We thank K. Baines, P. Drossart, J. Moses, C. Parkinson, V. Natraj, and X. Zhang for helpful discussions, D. Crisp for providing UV absorber profiles, and K. F. Li, D. Yang, and X. Zhang for assistance in preparing the manuscript. Special thanks are due to W. B. DeMore for a critical discussion of sulfur chemistry and S. Lebonnois and three anonymous referees for raising fundamental issues the resolution of which led to a much better paper. This research was supported by NASA grant NNX07AI63G to the California Institute of Technology. M. Liang was supported by NSC 97-2628-M-001-001 grant to Academia Sinica.

References Arakawa, A., and V. R. Lamb (1977), Computational design of the basic dynamical processes of the UCLA general circulation model, Methods Comput. Phys., 17, 173 – 265. Bertaux, J. L., et al. (1986), Active spectrometry of the ultraviolet-absorption within the Venus atmosphere, Sov. Astron. Lett., 12, 33 – 37, Engl. Transl. Bertaux, J. L., et al. (2007), A warm layer in Venus’ cryosphere and highlatitude measurements of HF, HCl, H2O, and HDO, Nature, 450, 646 – 649, doi:10.1038/nature05974. Be´zard, B., et al. (1993), The abundance of sulfur-dioxide below the clouds of Venus, Geophys. Res. Lett., 20, 1587 – 1590, doi:10.1029/93GL01338. Bluth, G. J. S., et al. (1992), Global tracking of the SO2 clouds from the June, 1991 Mount-Pinatubo eruptions, Geophys. Res. Lett., 19, 151 – 154, doi:10.1029/91GL02792. Chen, H., et al. (2007), Heterogeneous uptake of carbonyl sulfide on hematite and hematite-NaCl mixtures, Environ. Sci. Technol., 41, 6484 – 6490, doi:10.1021/es070717n.

Crisp, D. (1986), Radiative forcing of the Venus mesosphere: 1. Solar fluxes and heating rates, Icarus, 67, 484 – 514, doi:10.1016/00191035(86)90126-0. Cullen, M. J. P. (1993), The unified forecast climate model, Meteorol. Mag., 122, 81 – 94. Dalleska, N. F., A. J. Colussi, A. M. Hyldahl, and M. R. Hoffmann (2000), Rates and mechanism of carbonyl sulfide oxidation by peroxides in concentrated sulfuric acid, J. Phys. Chem. A, 104, 10,794 – 10,796, doi:10.1021/jp002333o. Farquhar, J., B. A. Wing, K. D. McKeegan, J. W. Harris, P. Cartigny, and M. H. Thiemens (2002), Mass-independent sulfur of inclusions in diamond and sulfur recycling on early earth, Science, 298, 2369 – 2372, doi:10.1126/science.1078617. Fegley, B., Jr., G. Klingelhofer, and L. Lodders (1997), Geochemistry of surface-atmosphere interactions on Venus, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 591 – 636, Univ. of Ariz. Press, Tucson. Hall, T. M., and D. W. Waugh (1997), Timescales for the stratospheric circulation derived from tracers, J. Geophys. Res., 102, 8991 – 9001, doi:10.1029/96JD03713. Herrnstein, A., and T. E. Dowling (2007), Effects of topography on the spin-up of a Venus atmospheric model, J. Geophys. Res., 112, E04S08, doi:10.1029/2006JE002804. Hollingsworth, J. L., et al. (2007), A simple-physics global circulation model for Venus: Sensitivity assessments of atmospheric superrotation, Geophys. Res. Lett., 34, L05202, doi:10.1029/2006GL028567. Hong, Y., and B. Fegley Jr. (1997), Formation of carbonyl sulfide (OCS) from carbon monoxide and sulfur vapor and applications to Venus, Icarus, 130, 495 – 504, doi:10.1006/icar.1997.5824. Jiang, X., et al. (2004), Quasi-biennial oscillation and quasi-biennial oscillation-annual beat in the tropical total column ozone: A twodimensional model simulation, J. Geophys. Res., 109, D16305, doi:10.1029/2003JD004377. Juckes, M. N., I. N. James, and M. Blackburn (1994), The influence of Antarctica on the momentum budget of the southern extratropics, Q. J. R. Meteorol. Soc., 120, 1017 – 1044, doi:10.1002/qj.49712051811. Kasting, J. F., K. J. Zahnle, J. P. Pinto, and A. T. Young (1989), Sulfur, ultraviolet-radiation, and the early evolution of life, Origins Life Evol. Biosphere, 19, 95 – 108, doi:10.1007/BF01808144. Krasnopolsky, V. A. (2007), Chemical kinetic model for the lower atmosphere of Venus, Icarus, 191, 25 – 37, doi:10.1016/j.icarus.2007.04.028. Krasnopolsky, V. A. (2008), High-resolution spectroscopy of Venus: Detection of OCS, upper limit to H2S, and latitudinal variations of CO and HF in the upper cloud layer, Icarus, 197, 377 – 385, doi:10.1016/ j.icarus.2008.05.020. Krasnopolsky, V. A., and J. B. Pollack (1994), H2O – H2SO4 system in Venus clouds and OCS, CO, and H2SO4 profiles in Venus troposphere, Icarus, 109, 58 – 78, doi:10.1006/icar.1994.1077. Lebonnois, S., D. Toublanc, F. Hourdin, and P. Rannou (2001), Seasonal variations of Titan’s atmospheric composition, Icarus, 152, 384 – 406, doi:10.1006/icar.2001.6632.

15 of 16

E00B34

YUNG ET AL.: OCS AND CO ON VENUS

Lee, C., S. R. Lewis, and P. L. Read (2007), Superrotation in a Venus general circulation model, J. Geophys. Res., 112, E04S11, doi:10.1029/ 2006JE002874. Lewis, J. S. (1970), Venus: Atmospheric and lithospheric composition, Earth Planet. Sci. Lett., 10, 73 – 80, doi:10.1016/0012-821X(70)90066-X. Liang, M.-C., and Y. L. Yung (2007), Sources of the oxygen isotopic anomaly in atmospheric N 2 O, J. Geophys. Res., 112, D13307, doi:10.1029/2006JD007876. Liang, M. C., et al. (2005), Meridional transport in the stratosphere of Jupiter, Astrophys. J., 635, L177 – L180, doi:10.1086/499624. Liang, M. C., et al. (2006), Isotopic composition of oxygen of stratospheric ozone, J. Geophys. Res., 111, D02302, doi:10.1029/2005JD006342. Limaye, S. S. (2007), Venus atmospheric circulation: Known and unknown, J. Geophys. Res., 112, E04S09, doi:10.1029/2006JE002814. Lu, C. W., Y. J. Wu, Y. P. Lee, R. S. Zhu, and M. C. Lin (2006), Experimental and theoretical investigation of rate coefficients of the reaction S (P-3)+OCS in the temperature range of 298 – 985 K, J. Chem. Phys., 125(16), 164329, doi:10.1063/1.2357739. Maiorov, B. S., et al. (2005), A new analysis of the spectra obtained by the VENERA missions in the Venusian atmosphere: I. The analysis of the data received from the VENERA-11 probe at altitudes below 37 km in the 0.44 – 0.66 mm wavelength range, Sol. Syst. Res., 39, 267 – 282, doi:10.1007/s11208-005-0042-1. Marcq, E., B. Be´zard, T. Encrenaz, and M. Birlan (2005), Latitudinal variations of CO and OCS in the lower atmosphere of Venus from nearinfrared nightside spectro-imaging, Icarus, 179, 375 – 386, doi:10.1016/ j.icarus.2005.06.018. Marcq, E., T. Encrenaz, B. Be´zard, and M. Birlan (2006), Remote sensing of Venus’ lower atmosphere from ground-based IR spectroscopy: Latitudinal and vertical distribution of minor species, Planet. Space Sci., 54, 1360 – 1370, doi:10.1016/j.pss.2006.04.024. Marcq, E., B. Be´zard, P. Drossart, G. Piccioni, J. M. Reess, and F. Henry (2008), A latitudinal survey of CO, OCS, H2O, and SO2 in the lower atmosphere of Venus: Spectroscopic studies using VIRTIS-H, J. Geophys. Res., 113, E00B07, doi:10.1029/2008JE003074. Michelangeli, D. V., M. Allen, and Y. L. Yung (1991), Heterogeneous reactions with NaCl in the El Chichon volcanic aerosols, Geophys. Res. Lett., 18, 673 – 676, doi:10.1029/91GL00547. Mills, F. P. (1998), I. Observations and photochemical modeling of the Venus middle atmosphere. II. Thermal infrared spectroscopy of Europa and Callisto, Ph.D. thesis, Calif. Inst. of Technol., Pasadena, Calif. Mills, F. P., and M. Allen (2007), A review of selected issues concerning the chemistry in Venus’ middle atmosphere, Planet. Space Sci., 55, 1729 – 1740, doi:10.1016/j.pss.2007.01.012. Mills, F. P., L. W. Esposito, and Y. L. Yung (2007), Atmospheric composition, chemistry, and clouds, in Exploring Venus as a Terrestrial Planet, Geophys. Monogr. Ser., vol. 176, edited by L. W. Esposito, E. Stofan, and T. Cravens, pp. 73 – 100, AGU, Washington, D. C. Molina, L. T., J. J. Lamb, and M. J. Molina (1981), Temperature-dependent UV absorption cross-sections for carbonyl sulfide, Geophys. Res. Lett., 8, 1008 – 1011, doi:10.1029/GL008i009p01008. Moses, J. I., M. Y. Zolotov, and B. Fegley (2002), Photochemistry of a volcanically driven atmosphere on Io: Sulfur and oxygen species from a pele-type eruption, Icarus, 156, 76 – 106, doi:10.1006/icar.2001.6758. Peixoto, J. P., and A. H. Oort (1992), Physics of Climate, 520 pp., Am. Inst. of Phys., New York.

E00B34

Pernice, H., et al. (2004), Laboratory evidence for a key intermediate in the Venus atmosphere: Peroxychloroformyl radical, Proc. Natl. Acad. Sci. U. S. A., 101, 14,007 – 14,010, doi:10.1073/pnas.0405501101. Pollack, J. B., et al. (1993), Near-infrared light from Venus’ nightside: A spectroscopic analysis, Icarus, 103, 1 – 42, doi:10.1006/icar.1993.1055. Prinn, R. G. (1975), Venus: Chemical and dynamical processes in the stratosphere and mesosphere, J. Atmos. Sci., 32, 1237 – 1247, doi:10.1175/1520-0469(1975)032<1237:VCADPI>2.0.CO;2. Prinn, R. G. (1978), Venus: Chemistry of the lower atmosphere prior to the Pioneer Venus mission, Geophys. Res. Lett., 5, 973 – 976, doi:10.1029/ GL005i011p00973. Prinn, R. G. (1979), On the possible roles of gaseous sulfur and sulfanes in the atmosphere of Venus, Geophys. Res. Lett., 6, 807 – 810, doi:10.1029/ GL006i010p00807. Sander, S. P., et al. (2006), Chemical kinetics and photochemical data for use in stratospheric modeling evaluation number 15, JPL Publ. 06 – 2, 523 pp. Schubert, G. (1983), General circulation and the dynamical state of the Venus atmosphere, in Venus, edited by D. M. Hunten et al., pp. 681 – 765, The Univ. of Ariz. Press, Tuscon. Seinfeld, J. H., and S. N. Pandis (1998), Atmospheric Chemistry and Physics: From Air Pollution to Global Change, John Wiley, New York. Shia, R. L., Y. L. Yung, M. Allen, R. W. Zurek, and D. Crisp (1989), Sensitivity study of advection and diffusion coefficients in a twodimensional stratospheric model using excess C-14 data, J. Geophys. Res., 94, 18,467 – 18,484, doi:10.1029/JD094iD15p18467. Toon, O. B., R. P. Turco, and J. B. Pollack (1982), The ultraviolet absorber on Venus: Amorphous sulfur, Icarus, 51, 358 – 373, doi:10.1016/00191035(82)90089-6. von Zahn, U., S. Kumar, H. Niemann, and R. G. Prinn (1983), Composition of the Venus atmosphere, in Venus, edited by D. M. Hunten et al., pp. 299 – 430, Univ. of Ariz. Press, Tucson. Yamamoto, M., and M. Takahashi (2003), Superrotation and equatorial waves in a T21 Venus-like AGCM, Geophys. Res. Lett., 30(9), 1449, doi:10.1029/2003GL016924. Yung, Y. L., and W. B. DeMore (1982), Photochemistry of the stratosphere of Venus: Implications for atmospheric evolution, Icarus, 51, 199 – 247, doi:10.1016/0019-1035(82)90080-X. Yung, Y. L., and W. B. DeMore (1999), Photochemistry of Planetary Atmospheres, Oxford Univ. Press, New York. Yung, Y. L., M. Allen, and J. P. Pinto (1984), Photochemistry of the atmosphere of Titan: Comparison between model and observations, Astrophys. J. Suppl. Ser., 55(3), 465 – 506, doi:10.1086/190963. 

B. Be´zard and E. Marcq, LESIA, Observatoire de Paris, Baˆtiment 18, pie`ce 111, F-92195 Meudon CEDEX, France. X. Jiang, Department of Earth and Atmospheric Sciences, University of Houston, Houston, TX 77004, USA. C. Lee, R. L. Shia, and Y. L. Yung, Division of Geological and Planetary Sciences, California Institute of Technology, Pasadena, CA 91125, USA. ([email protected]) M. C. Liang, Research Center for Environmental Changes, Academia Sinica, 128 Academia Road, Section 2, Taipei 115, Taiwan.

16 of 16

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114, E00B28, doi:10.1029/2008JE003095, 2009

for

Full Article

Modeling the distribution of H2O and HDO in the upper atmosphere of Venus Mao-Chang Liang1,2 and Yuk L. Yung3 Received 31 January 2008; revised 30 June 2008; accepted 2 December 2008; published 24 February 2009.

[1] The chemical and dynamical processes in the upper atmosphere of Venus are poorly

known. Recently obtained vertical profiles of trace species from the Venus Express mission, such as HCl, H2O, and HDO, provide new information to constrain these processes. Here, we simulate these profiles, using the model we have developed and described in a related paper by Yung et al. (2008), with special emphasis on the modeling of H2O and HDO. A new mechanism, the photo-induced isotopic fractionation effect (PHIFE) of H2O and HCl, is incorporated into our model. The observed enhancement of HDO could be attributed to (1) preferential destruction of H2O relative to HDO via PHIFE and (2) escape of hydrogen that enhances the abundance of D and hence its parent molecule HDO. Over a wide range of the sensitivity of the results to the changes of the two mechanisms, we find that the observed profiles of HDO and H2O profiles cannot be explained satisfactorily by current knowledge of chemical and dynamical processes in this region of the atmosphere. Several conjectures to tackle the problems are discussed. Citation: Liang, M.-C., and Y. L. Yung (2009), Modeling the distribution of H2O and HDO in the upper atmosphere of Venus, J. Geophys. Res., 114, E00B28, doi:10.1029/2008JE003095.

1. Introduction [2] Venus provides a window of opportunity in the solar system for studying the end-member of water evolution. Its highly enhanced D/H ratio, compared with the terrestrial value, suggests that about one Earth ocean might have been lost [e.g., Kasting and Pollack, 1983; Donahue, 1999, and references therein] via nonthermal escape of hydrogen (such as hydrodynamic escape, charge exchange, and collisionally induced escape). Hydrogen is produced by the photolysis of water, a process that is known to preferentially destroy the light isotopologue, resulting in the enrichment of the heavy parent molecule in the atmosphere [Cheng et al., 1999]. The process of enhancing the abundance of parent molecules is similar in nonthermal escape of hydrogen, and hence over the course of Venusian history, the D/H ratio is enriched as compared with the primordial value. A combination of photolytic and nonthermal escape processes fractionates the ratio in a way faster than each of them alone. In this paper, we investigate the D/H ratio affected by the photoinduced isotopic fractionation effect (PHIFE) of H2O/HDO and HCl/DCl in the present atmosphere of Venus. The reader is referred to Miller and Yung [2000] for a detailed explanation of PHIFE. The observed D/H in water in the upper atmosphere from Venus Express, along with other 1 Research Center for Environmental Changes, Academia Sinica, Taipei, Taiwan. 2 Graduate Institute of Astronomy, National Central University, Jhongli, Taiwan. 3 Division of Geological and Planetary Sciences, California Institute of Technology, Pasadena, California, USA.

Copyright 2009 by the American Geophysical Union. 0148-0227/09/2008JE003095$09.00

molecules (HCl, HF), provides additional constraints for understanding the relative importance of two dynamical processes: atmospheric circulation and escape. The former conserves the bulk D/H ratio in an air parcel and the latter enhances the ratio. As a result, the D/H ratio in water has less latitude dependence, compared with the case involving hydrogen escape. To evaluate the processes quantitatively in the upper atmosphere of Venus, one-dimensional (1-D) and two-dimensional (2-D) models are used. The data used for the study are described by Bertaux et al. [2007], Vandaele et al. [2008], and Fedorova et al. [2008].

2. Models [3] One-dimensional photochemical models are used to simulate the vertical profiles of H2O/HDO, HCl/DCl, H/D, H2/HD, OH/OD, HO2/DO2, CO2, CO, O2, O, O(1D), O3, Cl, ClO, Cl2, ClCO, and ClCO3 in the upper atmosphere (58 – 112 km) of Venus. The current model is based on Yung and DeMore [1982] and Mills [1998a, 1998b], and a subset of the chemistry is selected from Yung et al. [2008] to account for the UV attenuation in the upper part of the atmosphere. The selected hydrogen/deuterium chemistry is summarized in Table 1. The PHIFE of H2O/HDO and HCl/ DCl are taken from Cheng et al. [1999] and Bahou et al. [2001], respectively. The rest of the hydrogen and deuterium chemical reactions are assumed to be isotopically neutral. The transport and boundary conditions are primarily taken from Yung and DeMore [1982] and Mills [1998a, 1998b]. The CO2 mixing ratio is relatively uniform at 0.96 throughout the entire atmosphere. The fixed mixing ratios of 1, 0.1, 0.15, and 0.0075 ppmv are used for H2O, HDO, HCl, and DCl, respectively, at the lower boundary. The selected

E00B28

1 of 9

E00B28

LIANG AND YUNG: D/H ON VENUS

E00B28

Figure 1. Modeled vertical profiles of H2O (black) and HDO (gray) in 1-D model with prescribed winds depicted by solid and dashed curves (see text). Data are from Bertaux et al. [2007]. D/H ratio used to better represent the Venus Express profiles (Figure 1) is a factor of 2 higher than the bulk ratio, which is 0.05 for [HDO]/[H2O]. The rest of the species are transported downward at the lower boundary at a velocity determined by dynamics; the velocities are 0.05 and 0.016 cm s
1 for 1-D and 2-D models, respectively. The upper boundary is impermeable to all species. This is chosen to be our reference case. A model with hydrogen escape is also run. In this model, species other than atomic H and D have zero flux at the upper boundary. The escape fluxes of H and D are, respectively, 3.5  106 and 3.1  104 atoms cm
2 s
1 [Gurwell and Yung, 1993] or higher [Donahue, 1999]. Sensitivity of the model results with respect to the changes of transport and upper boundary conditions is studied using the 1-D model. The UV radiation is averaged diurnally before photolytic calculations, and the latitude dependence is carefully taken into account. (See Liang et al. [2005, and references therein] for a detailed description of the model.) The modeled latitude for 1-D models is set at 45°N, where we think it should better represent the condition at the polar region if the time constant of the large-scale meridional circulation is not small compared with the time of vertical transport and the lifetimes of H2O and HDO against photolysis. The major effect on the selection of the modeled latitude is the total photolysis rates (or J values) for molecules. The effect on isotopic ratios can be ignored under the current uncertainties of models and observations. We test the isotopic effect at a higher latitude at 80°N, and the largest difference of 5%, compared with that at 45°N, occurs above 110 km. [4] The 2-D version [Liang et al., 2005] of the Caltech/JPL photochemical mode is employed to simulate the meridonal distribution of H2O/HDO and HCl/DCl in the upper atmosphere of Venus. The current model and the adopted transport are described in a companion paper [Yung et al., 2008]. We solve the model at latitudes from pole to pole and altitudes from 56 to 112 km. The chemical species and reactions are taken from the 1-D model

described in the previous paragraph. The boundary conditions are the same as the 1-D case with hydrogen escape. This is selected to be our reference 2-D model. [5] The vertical profile of H2O provides insight into transport of the upper atmosphere of Venus. The SOIR data, taken near the north pole, show that H2O mixing ratio decreases with increasing altitude until 85 km and then increases above that [Bertaux et al., 2007]. Since there is no known source and sink of this magnitude (50%) above the cloud tops at 50 km and below 100 km (where photochemical processes become important), the profile has to be caused by transport. Several ad hoc advections have been tested, and 1-D models are used for such sensitivity study. One proposal that fits the data is having a downward advection (
0.3 cm s
1) between 75 and 85 km and upwelling (0.5 cm s
1) above until 95 km where the transport becomes downwelling (
0.5 cm s
1). This qualitatively agrees with the fact that there is a temperature inversion layer at 100 km. General circulation models [e.g., Lee et al., 2007] predict, in general, that air ascends at low latitudes and descends at high latitudes (Hadley cell). The heating at 100 km results from the wind profile (see later discussion on the relation between wind and temperature). The resulting H2O and HDO profiles from this prescribed advection (dashed curves) are shown in Figure 1.

3. Results [6] The photolysis of H2O/HDO and HCl/DCl tends to enhance their isotopic composition d, which is defined by the deviation of the ratio of an isotopically substituted species and its normal molecule from that of the prescribed standard d  ½ D=½ H =ð½ D=½ H Þ0
1;

ð1Þ

where [D] and [H] are the concentrations of [HDO] or [DCl] and [H2O] or [HCl], respectively. The subscript ‘‘0’’ refers

2 of 9

LIANG AND YUNG: D/H ON VENUS

E00B28

E00B28

Table 1. List of Hydrogen and Deuterium Chemical Reactionsa Reaction (R22) (R25) (R41) (R42) (R46) (R47) (R49) (R50) (R51) (R52) (R53) (R54) (R55) (R56) (R58) (R61) (R63) (R64) (R65) (R66) (R67) (R68) (R70) (R75) (R78) (R86) (R90) (R91) (R95) (R104) (R105) (R106) (R115) (R116) (R127) (R128) (R130) (R139) (R140) (R141) (R142) (R143) (R144) (R146) (R147) (R149) (R150) (R151) (R152) (R153) (R154) (R155) (R156) (R157) (R158) (R159) (R160) (R163) (R164) (R168) (R169) (R171) (R172) (R173) (R174) (R175) (R176) (R177) (R178) (R179) (R181) (R182) (R184)

H2O ! H + OH HCl ! H + Cl O(1D) + H2O ! 2OH O(1D) + H2 ! H + OH H + O2 + CO2 ! HO2 + CO2 H + O3 ! OH + O2 O + HO2 ! OH + O2 O + OH ! O2 + H OH + CO ! CO2 + H OH + H2 ! H2O + H OH + O3 ! HO2 + O2 HO2 + O3 ! OH + 2O2 H + HO2 ? 2OH H + HO2 ! H2O + O H + HO2 ! H2 + O2 OH + HO2 ! H2O + O2 2H + CO2 ! H2 + CO2 H + HCl ! H2 + Cl OH + HCl ! Cl + H2O O + HCl ! OH + Cl Cl + H2 ! HCl + H Cl + OH ! HCl + O Cl + HO2 ! HCl + O2 ClO + OH ! HO2 + Cl Cl + H + M ! HCl + M ClCO + H ! HCl + CO H + Cl2 ! HCl + Cl Cl + HO2 ! OH + ClO ClCO3 + H ! CO2 + Cl + OH O + H2 ! OH + H 2OH ! H2O + O O(1D) + HCl ! Cl + OH O(1D) + HCl ! O + HCl O(1D) + HCl ! H + ClO ClO + H2 ! HCl + OH O + H + M ! OH + M H + OH + CO2 ! H2O + CO2 HDO ! H + OD HDO ! D + OH DCl ! D + Cl O(1D) + HDO ! OH + OD O(1D) + HD ! H + OD O(1D) + HD ! D + OH D + O2 + CO2 ! DO2 + CO2 D + O3 ! OD + O2 O + DO2 ! OD + O2 O + OD ! O2 + D OD + CO ! CO2 + D OD + H2 ! HDO + H OH + HD ! HDO + H OH + HD ! H2O + D OD + O3 ! DO2 + O2 DO2 + O3 ! OD + 2O2 D + HO2 ! OH + OD H + DO2 ! OH + OD D + HO2 ! HDO + O H + DO2 ! HDO + O D + HO2 ! HD + O2 H + DO2 ! HD + O2 OD + HO2 ! HDO + O2 OH + DO2 ! HDO + O2 D + H + CO2 ! HD + CO2 D + HCl ! HD + Cl H + DCl ! HD + Cl OD + HCl ! Cl + HDO OH + DCl ! Cl + HDO O + DCl ! OD + Cl Cl + HD !DCl + H Cl + HD !HCl + D Cl + OD ! DCl + O Cl + DO2 ! DCl + O2 ClO + OD ! DO2 + Cl Cl + D + M ! DCl + M

Rate Coefficient
6

J22 = 2.5  10 J25 = 1.8  10
6 k41 = 2.2  10
10 k42 = 1.1  10
10 k46 = 2.0  10
31 (T/300)
1.6; k8 = 7.50  10
11 k47 = 1.4  10
10 e
470/T k49 = 2.9  10
11 e200/T k50 = 2.2  10
11 e120/T k51 = 1.5  10
13 k52 = 5.5  10
12 e
2000/T k53 = 1.6  10
12 e
940/T k54 = 1.1  10
14 e
500/T k55 = 7.3  10
11 k56 = 1.6  10
12 k58 = 6.4  10
12 k61 = 4.7  10
11 e250/T k63 = 5.0  10
29 T
1.3 k64 = 1.5  10
11 e
1750/T k65 = 2.6  10
12 e
350/T k66 = 1.0  10
11 e
3300/T k67 = 3.7  10
11 e
2300/T k68 = 1.2 e
510/T k70 = 1.8  10
11 e170/T k75 = 1.1  10
11 e120/T k78 = 1.0  10
32 k86 = 1.0  10
11 k90 = 1.4  10
10 e
90/T k91 = 4.1  10
11 e
450/T k95 = 1.0  10
11 k104 = 9.9  10
32 T6.5 e
1460/T k105 = 4.2  10
12 e
240/T k106 = 1.0  10
10 k115 = 1.4  10
11 k116 = 3.6  10
11 k127 = 1.0  10
12 e
4800/T k128 = 1.3  10
29 T
1 k130 = 7.7  10
26 T
2

(1/2)J22

(1/2)J22

J25 k41 k42 k42 k46 k47 k49 k50 k51 k52 (1/2)k52 (1/2)k52 k53 k54 k55 k55 k56 k56 k58 k58 k61 k61 k63 k64 k64 k65 k65 k66 (1/2)k67 (1/2)k67 k68 k70 k75 k78

3 of 9

Referenceb 4 1 2 2 2, 10 2 2 2 8 2 2 2 2 2 2 8 6, 9, 10 10 2 2 2 3 5 8 10 10 3 5 10 7 2 8 8 8 8 8 8 4 4 1 assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed

LIANG AND YUNG: D/H ON VENUS

E00B28

E00B28

Table 1. (continued) Reaction (R185) (R186) (R187) (R188) (R189) (R190) (R191) (R192) (R193) (R194) (R195) (R196) (R197) (R198) (R200) (R201) (R202) (R205) (R206)

ClCO + D ! DCl + CO D + Cl2 ! DCl + Cl Cl + DO2 ! OD + ClO ClCO3 + D ! CO2 + Cl + OD O + HD ! OD + H O + HD ! OH + D OD + OH ! HDO + O O(1D) + DCl ! Cl + OD OD + HCl ! HDO + Cl OH + DCl ! HDO + Cl DO2 + HCl ! HDO + ClO HO2 + DCl ! HDO + ClO O(1D) + DCl ! O + DCl O(1D) + DCl ! D + ClO ClO + HD ! DCl + OH ClO + HD ! HCl + OD O + D + M ! OD + M D + OH + CO2 ! HDO + CO2 H + OD + CO2 ! HDO + CO2

Rate Coefficient k86 k90 k91 k95 (1/2)k104 (1/2)k104 k105 k106 k110 k110 k111 k111 k115 k116 (1/2)k127 (1/2)k127 k128 k130 k130

Referenceb assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed assumed

a The photolytic coefficients (J) are given for a diurnally averaged model at 45°N at the top of the atmosphere (s
1). J value is defined by the product of the species’ cross section and the solar spectrum. Two-body and three-body rate coefficients are given in units of cm3 s
1 and cm6 s
1, respectively. b 1, Bahou et al. [2001]; 2, Baulch et al. [1980]; 3, Baulch et al. [1981]; 4, Cheng et al. [1999]; 5, Lee and Howard [1982]; 6, Prather et al. [1978]; 7, Robie et al. [1990]; 8, Sander et al. [2006]; 9, Trainor et al. [1973]; 10, Yung and DeMore [1982].

to the reference standard, which is set at the lower boundary. Figure 2 shows the ratios of the photolytic coefficients (J values) of HDO/H2O (black curve) and DCl/HCl (gray curve). The ratios decrease with decreasing altitude, demonstrating that their d values can be enhanced accordingly by photolytic processes. The photolytic isotope fractionation studied in this article is caused primarily by self-shielding by the most abundant molecules. The photodissociation of H2O occurs primarily at altitudes between 80 and 110 km and that of HCl at altitudes less than 70 km; the latter is more significant below 95 km in terms of photolysis rate (Figure 3). In the regions below 95 km, the photolysis of HCl provides a source of H atoms that eventually form H2O. As this source favors the formation of H2O over HDO, this results in a depletion of d(HDO). The resulting column-averaged (above the lower boundary of 58 km) d(HDO) is
2%. As a result, HDO, HD, and D are all isotopically depleted; and d(DCl) is enhanced by 25%. The factor of 10 between d(HDO) and d(DCl) is due to higher abundance of H2O relative to HCl. Above 95 km, H2O photolysis is significant and d(HDO) increases with altitude. Including H2O/HDO photolysis enhances its isotopic composition, and consequently, the total d(HDO) above 58 km becomes
1% (Figure 4). [7] Sensitivity of the results to atmospheric transport is shown by the dashed curves in Figures 1 – 4. Though the transport provides a good fit to the observed H2O and HDO profiles, it is negligible in modifying d(HDO). This is caused by a high eddy diffusion coefficient (a few times 105 cm2 s
1, or a transport time of 105 sec) that greatly dilutes the isotopic fractionation resulting from H2O/HDO and HCl/DCl photolytic processes (see an explanation of the dilution effect by Liang et al. [2007]). Several other modifications in transport have been tested but none gives a satisfactory explanation to both H2O and HDO profiles. For example, the decreasing mixing ratios of H2O and HDO between 70 and 85 km can be caused by advection and diffusive separation. The former can be prescribed by a downwelling transport (adopted in this paper). The latter

requires a reduction of eddy mixing coefficients (so that molecular diffusion becomes more important than advective transport). The increase above 85 km can be attributed either to upwelling transport (adopted in this paper) or to downwelling of air from a higher region in the atmosphere where H2O/HDO sources are needed. To account for the increase, the required H2O source is  5  108 molecules cm
2 s
1 (the total photolysis rate of H2O approximately the photolysis rate of HCl above 85 km; Figure 4), a value that is unreasonably high. Furthermore, the increase of the ratio of [HDO]/[H2O] above 70 km observed by Venus Express implies that either H2O photolysis is not small (i.e., at least 0.1 times HCl photolysis) or a hitherto unknown mechanism that transports HDO to the region. Further a general circulation model with the correct physics (e.g., heating between 90 and 120 km) is urgently needed. The laboratory measurements of the photolytic cross sections of H2O/HDO/ HCl/DCl at temperatures (150–200 K) similar to the Venus’ are also required, because of the temperature-dependent nature of PHIFE [e.g., Kaiser et al., 2002; Liang et al., 2004]. [8] Additionally, atomic hydrogen can escape from the atmosphere of Venus. The escape fluxes of H and D are estimated to be 3.5  106 and 3.1  104 atoms cm
2 s
1, respectively [Gurwell and Yung, 1993]. Higher fluxes determined by Donahue [1999] are also tested, but no noticeable effect is observed. This additional process, that enhances atomic D abundance (i.e., increase D/H ratio) in an air parcel, followed by subsequent chemical reactions of H/D with oxygen compounds favors the production of HDO. However, the escape flux is 2 orders of magnitude lower than the photolysis of H2O/HDO and HCl/DCl. Consequently, the hydrogen escape plays a small role in modifying the isotopic composition of water. This effect is identical to the influx of water described in the previous paragraph. [9] Figure 5 shows the modeled HDO/H2O ratios in the 2-D model. The ratio is relatively uniform in the regions below 95 km and increases above. The increase is caused by the preferential photolysis of H2O over HDO, as de-

4 of 9

LIANG AND YUNG: D/H ON VENUS

E00B28

E00B28

Figure 2. Ratios of the photolytic coefficients (J values) of deuterated and normal molecules. Black and gray curves are for the ratios of HDO/H2O and DCl/HCl, respectively. Line designation is same as that in Figure 1. The solid and dashed curves are indistinguishable in the current plotting scale. scribed above. In addition, we see that the ratio of HDO/ H2O also decreases with increasing latitude, demonstrating that in the current model, transport plays a negligible role in modifying the isotopic composition of water. As in the reference 1-D model, we obtain no kink feature in H2O and HDO mixing ratio profiles observed by Venus Express. The transport was obtained by assuming Newtonian cooling [Lee et al., 2007]. Future work on the transport is required.

4. Discussion [10] In our standard 1-D model, the eddy diffusion coefficient in the upper atmosphere of Venus is 106 cm2 s
1, and this corresponds to a mixing time 105 s [see Yung and DeMore, 1982]. The prescribed advection transport has a time constant of 106 s (velocity of 0.4 cm s
1 and atmospheric scale height of 4 km), an order of magnitude lower than the eddy mixing time. This confirms that the prescribed advection can explain well the vertical mixing ratio profile of H2O but not the ratio profile of HDO/H2O. We will demonstrate in section 4.2 that the tentatively observed enhancement of HDO relative to H2O [Bertaux et al., 2007] could be explained by the PHIFE of water alone, through hitherto unknown processes that transport the isotopic signature to a lower altitude. Moreover, our model also suggests that in order to explain the observed H2O profile, an upward transport of 0.5 cm s
1 between 85 and 95 km and a downward transport of 0.5 cm s
1 above 95 km are needed. The existence of downward transport is supported by the temperature anomaly observed by Venus Express [Bertaux et al., 2007] (see section 4.1), but the upward transport remains a puzzle. We note that though the temperature inversion is obtained on the night side at midlatitudes, this inversion probably extends to the polar region on the basis of the analogy with the terrestrial middle atmosphere [see, e.g., Holton et al., 1995, Figure 3]. A 3-D

general circulation model that includes realistic heating is urgently needed to resolve these issues. 4.1. Temperature Anomaly and Descent Rate [11] SPICAV data show a large temperature anomaly 20– 50 K at around 100 km, taken on the night side at low to midlatitudes [see Bertaux et al., 2007, Figure 1]. The authors interpret this as evidence for heating by air subsidence. The associated vertical velocity may be estimated as follows. The thermodynamic equation can be written as follows [Andrews et al., 1987, p. 115]: DT =Dt þ kwT=H ¼ J =Cp ;

ð2Þ

where D/Dt is material derivative, T is temperature, w is vertical velocity, H is scale height, J is diabatic heating rate, k is (g – 1)/g, g is Cp/Cv, and Cp and Cv are heat capacity at constant pressure and volume, respectively. For a CO2 atmosphere, k = 1/4. For vertical advection only, we have DT =Dt ¼ @T =@t þ w@T =@z:

ð3Þ

Substituting (3) into (2) and simplifying, we have @T =@t þ wð@T =@z þ kT =H Þ ¼ J =Cp :

ð4Þ

Since we are interested in dynamical heating, we can set J = 0. Equation (4) allows us to estimate the change of temperature due to dynamical heating DT ¼
wð@T =@z þ kT =H ÞDt:

ð5Þ

The maximum change in T occurs for Dt = trad, where trad is the Newtonian cooling time constant [see, e.g., Goody and Yung, 1989, p. 252]. If Dt  trad cooling by radiation becomes important, thereby reducing the effect of dynami-

5 of 9

LIANG AND YUNG: D/H ON VENUS

E00B28

E00B28

Figure 3. The photolysis rates of H2O (black) and HCl (gray). Line designation is same as that in Figure 1.

cal heating. For Venus around 100 km (0.03 hPa), we have the following values for the quantities in equation (5): T ¼ 175K;

ð6Þ

@T =@z ¼
4K=km;

ð7Þ

H ¼ 4km;

Dt ¼ trad  1 day;

H2 O þ hv ! OH þ H;

ð11Þ

HDO þ hv ! OD þ H;

ð12aÞ

HDO þ hv ! OH þ D:

ð12bÞ

ð8Þ

ð9Þ

where in (9), we have used estimates based on Crisp and Titov [1997]. Substituting (6) – (9) into (5) we have DT ¼
54w;

summarized in Figure 6. The higher signal-to-noise ratio data are obtained at altitudes between 75 and 85 km (blue symbols). An offset between the solid line and the blue symbols can be accounted for by the mixing ratio of H2O and the ratio of [HDO]/[H2O] set at the lower boundary. Both H2O and HDO are destroyed by photolysis

ð10Þ

where the units for DT and w are K and cm s
1, respectively. Thus, if we have downwelling velocity 1 cm s
1, then DT is 54 K, which is also the correct order of magnitude to account for the H2O profile, as discussed in the previous section. On the other hand, if we use the descent velocity of 0.43 m s
1 suggested by Bertaux et al. [2007], we have DT is 2322 K, which is clearly too large compared with the observations! Thus, the H2O profile offers a potentially valuable clue to the descent rate of air in the mesosphere of Venus. 4.2. Isotopic Enrichment [12] SOIR data show evidence for changes in H2O, HDO and the HDO/H2O ratio. There appears to be a decrease in H2O mixing ratio between 70 km and 85 km, even though the mixing ratio of H2O or HDO could be seriously affected by the uncertainty in CO2 retrieval, as pointed out by Vandaele et al. [2008]. This is explained as follows and

Let x and y be the concentrations of H2O and HDO, respectively. In the absence of production, the loss of x and y is described by dx=dt ¼
J11 x;

ð13Þ

dy=dt ¼
J12 x;

ð14Þ

where J11 and J12 denote the photodissociation coefficients for reactions (11) and (12). Solving these equations, we have xðt Þ ¼ xð0Þ expð
J11 t Þ

ð15Þ

yðt Þ ¼ yð0Þ expð
J12 t Þ;

ð16Þ

and

6 of 9

Rðt Þ ¼ ½ yðt Þ=xðt Þ=½ yð0Þ=xð0Þ ¼ exp½ðJ11
J12 Þt :

ð17Þ

LIANG AND YUNG: D/H ON VENUS

E00B28

E00B28

Figure 4. Model HDO/H2O corresponding to cases in Figure 1. Data are from Bertaux et al. [2007]. The two curves are indistinguishable below 100 km. Line designation is the same as that in Figure 1.

Let r(t) = x(t)/x(0). From (15), we have rðt Þ ¼ expð
J11 t Þ:

ð18Þ

From (17) and (18), we can derive a simple relation Rðt Þ ¼ ½rðt Þ
f ;

ð19Þ

f ¼ ðJ11
J12 Þ=J11 :

ð20Þ

where

Referring to the data from SOIR, let us assume that the mixing ratio of H2O at the cloud tops is 1 ppm; at 95 km,

the value has decreased to 0.7 ppm. Therefore, r(t) = 0.7. From our model, J12/J11 = 0.54; thus f = 0.46 (The total column integrated photolysis rates of HDO and H2O are 2.6  107 and 4.8  108 molecules cm
2 s
1, respectively, giving the averaged ratio of J12/J11 = 0.54. Since the vertical transport time is significantly shorter than the photolytic lifetimes of H2O and HDO, the column averaged J values are relevant to use.). The expected R(t) from (19) is 1.2, which is close to the observed enrichment between 70 and 75 km and 90– 95 km. This simple theory offers a satisfactory explanation of the increase of the HDO/H2O ratio as a result of the preferential destruction of H2O relative to HDO. Figure 6 provides a quantitative prediction for the isotope ratio of water (R) to

Figure 5. Reference case for simulation of the ratio of HDO and H2O by the 2-D model. 7 of 9

LIANG AND YUNG: D/H ON VENUS

E00B28

E00B28

Figure 6. The isotope ratio of water (R) to the fraction of remaining water (r). See text for details.

the fraction of remaining water (r). This relation is consistent with the limited observations currently available. It is desirable to test this relation over a wider range of R and r.

5. Concluding Remarks [13] Simple one-dimensional and two-dimensional chemistry and transport models are used to study the spatial distribution of HCl, H2O, and HDO in the upper atmosphere of Venus. The distributions of these molecules in the atmosphere reflect the influence of chemistry and transport. A large subsidence at around 85 km is needed to explain the observed vertical profile of H2O. The magnitude of the descent rate is consistent with the temperature anomaly observed by SPICAV. Preferential photolysis of H2O over HDO provides a driving force for isotopic enrichment. Laboratory measurements of temperature-dependent cross sections for H2O over HDO are needed. The correct simulation of the distribution of HCl, H2O, and HDO ultimately requires a 3-D model. This paper identifies the most important physical and chemical processes that a model must incorporate. [14] The emphasis of this work is in the region of the atmosphere above 70 km, where the H2O abundance is 1 ppm. As the abundance of H2O in the deep atmosphere is 100 ppm, there is another removal mechanism that is not related to photolysis. It is known that H2SO4 aerosols are formed above the cloud tops. They represent a net sink of SO2 and H2O, thereby explaining their rapid decrease with altitude above the cloud tops. We do not know whether this process can cause fractionation. Laboratory studies are needed to determine the fractionation associated with this chemical dehydration, if any occurs. [15] Acknowledgments. Special thanks are due to J.-L. Bertaux for providing H2O, HDO, HCl, and HF profiles from Venus Express. We thank H. Hartman, N. Heavens. K. F. Li, V. Natraj, C. Parkinson, R. L. Shia, and X. Zhang for critical comments. This research was supported in part by

NSC grant 97-2628-M-001-001 to Academia Sinica and NASA grant NNX07AI63G to the California Institute of Technology.

References Andrews, D. G., J. R. Holton, and C. B. Leovy (1987), Middle Atmosphere Dynamics, 489 pp., Academic Press, Orlando, Fla. Bahou, M., et al. (2001), Absorption cross sections of HCl and DCl at 135 – 232 nanometers: Implications for photodissociation on Venus, Astrophys. J., 559(2), L179 – L182, doi:10.1086/323753. Baulch, D. L., et al. (1980), Evaluated kinetic and photochemical data for atmospheric chemistry, J. Phys. Chem. Ref. Data, 9(2), 295 – 471. Baulch, D. L., J. Duxbury, S. J. Grant, and D. C. Montague (1981), Evaluated kinetic data for high-temperature reactions, Vol. 4—Homogeneous gas-phase reactions of halogen-containing and cyanide-containing species, J. Phys. Chem. Ref. Data, 10(Supplement 1), 1 – 721. Bertaux, J. L., et al. (2007), A warm layer in Venus’ cryosphere and highaltitude measurements of HF, HCl, H2O and HDO, Nature, 450, 646 – 649. Cheng, B. M., et al. (1999), Photo-induced fractionation of water isotopomers in the Martian atmosphere, Geophys. Res. Lett., 26(24), 3657 – 3660, doi:10.1029/1999GL008367. Crisp, D., and D. Titov (1997), The thermal balance of the Venus atmosphere, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher, D. M. Hunten, and R. J. Phillips, pp. 353 – 384, Univ. of Ariz. Press, Tucson, Ariz. Donahue, T. M. (1999), New analysis of hydrogen and deuterium escape from Venus, Icarus, 141(2), 226 – 235, doi:10.1006/icar.1999.6186. Fedorova, A., O. Korablev, A.-C. Vandaele, J.-L. Bertaux, D. Belyaev, A. Mahieux, E. Neefs, W. V. Wilquet, R. Drummond, F. Montmessin, and E. Villard (2008), HDO and H2O vertical distributions and isotopic ratio in the Venus mesosphere by Solar Occultation at Infrared spectrometer on board Venus Express, J. Geophys. Res., 113, E00B22, doi:10.1029/ 2008JE003146. Goody, R. M., and Y. L. Yung (1989), Atmospheric Radiation: Theoretical Basis, Oxford Univ. Press, New York. Gurwell, M. A., and Y. L. Yung (1993), Fractionation of hydrogen and deuterium on Venus due to collisional ejection, Planet. Space Sci., 41(2), 91 – 104, doi:10.1016/0032-0633 (93)90037-3. Holton, J. R., P. H. Haynes, M. E. McIntyre, A. R. Douglass, R. B. Rood, and L. Pfister (1995), Stratosphere-troposphere exchange, Rev. Geophys, 33(4), 403 – 439. Kaiser, J., T. Rockmann, and C. A. M. Brenninkmeijer (2002), Temperature dependence of isotope fractionation in N2O photolysis, Phys. Chem. Chem. Phys., 4(18), 4420 – 4430, doi:10.1039/b204837j. Kasting, J. F., and J. B. Pollack (1983), Loss of water from Venus. Part 1. Hydrodynamic escape of hydrogen, Icarus, 53(3), 479 – 508, doi:10.1016/0019-1035 (83)90212-9. Lee, Y. P., and C. J. Howard (1982), Temperature-dependence of the rateconstant and the branching ratio for the reaction Cl+HO2, J. Chem. Phys., 77(2), 756 – 763, doi:10.1063/1.443892.

8 of 9

E00B28

LIANG AND YUNG: D/H ON VENUS

Lee, C., S. R. Lewis, and P. L. Read (2007), Superrotation in a Venus general circulation model, J. Geophys. Res., 112, E04S11, doi:10.1029/ 2006JE002874. Liang, M. C., G. A. Blake, and Y. L. Yung (2004), A semianalytic model for photo-induced isotopic fractionation in simple molecules, J. Geophys. Res., 109, D10308, doi:10.1029/2004JD004539. Liang, M. C., et al. (2005), Meridional transport in the stratosphere of Jupiter, Astrophys. J., 635(2), L177 – L180, doi:10.1086/499624. Liang, M. C., G. A. Blake, B. R. Lewis, and Y. L. Yung (2007), Oxygen isotopic composition of carbon dioxide in the middle atmosphere, Proc. Natl. Acad. Sci. U.S.A., 104(1), 21 – 25, doi:10.1073/pnas.0610009104. Miller, C. E., and Y. L. Yung (2000), Photo-induced isotopic fractionation, J. Geophys. Res., 105(D23), 29,039 – 29,051. Mills, F. P. (1998a), Observations and photochemical modeling of the Venus middle atmosphere, Ph.D. thesis, Calif. Inst. of Technol, Pasadena, Calif. Mills, F. P. (1998b), Thermal infrared spectroscopy of Europa and Callisto, Ph.D. thesis, Calif. Inst. of Technol, Pasadena, Calif. Prather, M. J., J. A. Logan, and M. B. McElroy (1978), Carbon-monoxide in Jupiter’s upper-atmosphere: An extraplanetary source, Astrophys. J., 223(3), 1072 – 1081, doi:10.1086/156340. Robie, D. C., S. Arepalli, N. Presser, T. Kitsopoulos, and R. J. Gordon (1990), The intramolecular kinetic isotope effect for the reaction O(3P)+HD, J. Chem. Phys., 92(12), 7382 – 7393, doi:10.1063/1.458224.

E00B28

Sander, S. P., et al. (2006), Chemical kinetics and photochemical data for use in atmospheric studies, Jet Propul. Lab. Pub. 06 – 2, Jet Propul. Lab., Pasadena, Calif. (Available at http://jpldataeval.jpl.nasa.gov/.) Trainor, D. W., D. O. Ham, and F. Kaufman (1973), Gas-phase recombination of hydrogen and deuterium atoms, J. Chem. Phys., 58(10), 4599 – 4609, doi:10.1063/1.1679024. Vandaele, A. C., et al. (2008), Composition of the Venus mesosphere by SOIR on board Venus Express, J. Geophys. Res., 113, E00B23, doi:10.1029/2008JE003140. Yung, Y. L., and W. B. DeMore (1982), Photochemistry of the stratosphere of Venus: Implications for atmospheric evolution, Icarus, 51(2), 199 – 247, doi:10.1016/0019-1035(82)90080-X. Yung, Y. L., M. C. Liang, X. Jiang, C. Lee, B. Bezard, and E. Marcq (2008), Modeling the distribution of OCS in the lower atmosphere of Venus, J. Geophys. Res., doi:10.1029/2008JE003094, in press.























M.-C. Liang, Research Center for Environmental Changes, Academia Sinica, 128 Academia Road, Section 2, Taipei 115, Taiwan. (mcl@rcec. sinica.edu.tw) Y. L. Yung, Division of Geological and Planetary Sciences, California Institute of Technology, MC 170-25, 1200 East California Boulevard, Pasadena, CA 91125, USA.

9 of 9

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114, E00B39, doi:10.1029/2008JE003251, 2009

for

Full Article

Water vapor abundance near the surface of Venus from Venus Express/VIRTIS observations Bruno Be´zard,1 Constantine C. C. Tsang,2 Robert W. Carlson,3 Giuseppe Piccioni,4 Emmanuel Marcq,1 and Pierre Drossart1 Received 13 August 2008; revised 16 December 2008; accepted 13 March 2009; published 1 May 2009.

[1] Nightside observations of the 1.18-mm atmospheric window by the Visible and

Infrared Thermal Imaging Spectrometer (VIRTIS) aboard the Venus Express spacecraft were analyzed to measure and map the water vapor abundance in the lower atmosphere. Thermal emission in this window originates partly from the surface and partly from the first scale height (0–15 km) of the atmosphere. Constraints on the CO2 continuum absorption, which is the dominant source of gaseous opacity in the window, were obtained from the variation of the 1.185-mm intensity with surface elevation. An absorption coefficient of 1 ± 0.4
109 cm1 amagat2 best fits the observed variation. We retrieved a water vapor mole fraction of 44 ± 9 ppm from various selections of VIRTIS spectra in the southern hemisphere, in agreement with previous analyses of the nightside emission. This value is somewhat larger than that previously determined at higher altitudes from the 2.3- and 1.74-mm nightside windows, but the error bars still allow a constant with height H2O mole fraction from the surface up to 40 km. Using the intensity ratio in the two wings of the 1.18-mm window as a proxy, we searched for horizontal variations of the H2O abundance in various VIRTIS observational sequences. We derived stringent upper limits for any possible latitudinal variations on the night side: ±1.5% in the range 60°S–25°N and ±3% for the broader range 80°S–25°N. The lack of detectable latitudinal variations is consistent with a constant with height water profile in the lower atmosphere and probably precludes any strong concentration gradient near the surface. Citation: Be´zard, B., C. C. C. Tsang, R. W. Carlson, G. Piccioni, E. Marcq, and P. Drossart (2009), Water vapor abundance near the surface of Venus from Venus Express/VIRTIS observations, J. Geophys. Res., 114, E00B39, doi:10.1029/2008JE003251.

1. Introduction [2] The near-infrared windows centered at 1.01, 1.10 and 1.18 microns provide a means of probing the lower atmosphere and surface of Venus. In these windows, thermal emission from the surface and the lowest scale height (0 – 15 km) of the atmosphere can leak through the thick sulfuric acid clouds and be detected on the night side of the planet [Taylor et al., 1997]. The 1.10- and 1.18-mm windows are limited on one side by the n 1 + n 2 + n 3 H2O band centered at 1.13 mm and on the other side by CO2 bands (respectively the 2n 1 + 3n 3 band at 1.05 mm and the n 1 + 3n 3 band at 1.21 mm). Imaging and spectroscopic observations of these windows can thus provide information on the water vapor abundance near the surface and its variations. [3] Be´zard and de Bergh [2007] recently reviewed the determinations of the water vapor abundance in Venus’ deep 1 LESIA, Observatoire de Paris, UPMC, Universite´ Paris-Diderot, CNRS, Meudon, France. 2 Atmospheric, Oceanic and Planetary Physics, Department of Physics, University of Oxford, Oxford, UK. 3 Jet Propulsion Laboratory, Pasadena, California, USA. 4 INAF, IASF, Rome, Italy.

Copyright 2009 by the American Geophysical Union. 0148-0227/09/2008JE003251$09.00

atmosphere. In the 1990s, various ground-based observations of Venus’ night side established that a H2O mole fraction around 30 ± 10 ppm can reproduce the 1.18-mm window’s spectrum [Crisp et al., 1991; Pollack et al., 1993; de Bergh et al., 1995]. From spectroimaging data covering a broad range in surface altitude, Meadows and Crisp [1996] derived a slightly larger H2O mole fraction, 45 ± 10 ppm. They also argued that the temperature lapse rate in the lowest 6 km is shallower than that measured in situ [e.g., Seiff, 1983] and used in earlier analyses of nightside observations. Reanalyzing Venera 11, 13 and 14 optical spectra, Ignatiev et al. [1997] concluded that the H2O mixing ratio lies between 20 and 40 ppm in the altitude range 5 – 60 km. Below 5 km, the quality of the Venera data is worse but they seem to indicate an increase of the H2O mixing ratio up to 50– 70 ppm at the surface. All these results show that significant uncertainties remain on the water abundance close to the surface. In particular, analysis of near-infrared windows is hampered by our poor knowledge of the CO2 ‘‘continuum’’ opacity at high pressures, high temperatures, and long path lengths, arising from collisioninduced bands and extreme far wings of allowed CO2 bands. [4] Drossart et al. [1993] searched for horizontal variations of the H2O abundance using Galileo/NIMS data

E00B39

1 of 12

E00B39

BE´ZARD ET AL.: WATER VAPOR NEAR THE SURFACE OF VENUS

E00B39

Figure 1. Two averages of 135 Venus’ nightside spectra recorded during session 390– 13 over Alpha Regio with a mean altitude of 1.30 km (thick line) and about 5° to the east at a mean altitude of 0.47 km (thin line). Emission angle is 45° for both selections. Spectra have been shifted by 7 nm to correct for a slight wavelength calibration error. The location of the main CO2 and H2O bands that limit the nearinfrared windows is shown along with the position of the O2 airglow emission. recorded over a strip on the night side extending from 60°S to 40°N. Analyzing the 1.18-mm window, they concluded that the water abundance in the 0 –15 km range did not vary by more than 20% over the limited area covered. No other searches for horizontal or temporal variations of water vapor near the surface have been published. [5] The Venus Express spacecraft, orbiting Venus since April 2006, is currently monitoring the dynamics and composition of Venus’ atmosphere. In particular, the Visible and Infrared Thermal Imaging Spectrometer (VIRTIS) is repeatedly mapping Venus’ nightside emission through its infrared imaging channel (M-IR), giving access to the composition of the deep atmosphere over large horizontal and temporal scales. We present here an investigation of the water vapor abundance near the surface of Venus using various VIRTIS-M observational sets. We focused on the 1.18-mm window, which is the most favorable, being 3 times more intense than the 1.10-mm one. In a first step, we derived constraints on the CO2 so-called continuum opacity by investigating the variation of the 1.18-mm emission as a function of the surface elevation. We then determined the H2O mixing ratio that best reproduces observed spectra. Finally, we analyzed several VIRTIS-M spectral images to search for possible weak horizontal abundance variations, using intensity ratios in the 1.18-mm window as a proxy.

2. VIRTIS Observations [6] VIRTIS is a dual spectrometer comprising two subinstruments with their own telescope: an imaging spectrometer (VIRTIS-M) with visible and infrared channels and a high-resolution spectrometer (VIRTIS-H) limited to the interval 2 – 5 mm [Piccioni et al., 2007]. The infrared channel of VIRTIS-M covers the range 1.05 – 5.2 mm with a spectral sampling of 9.5 nm. The spatial sampling is

0.25 mrad and the instantaneous field of view (FOV) is 0.25
64 mrad, with 256 pixels along the slit. Scanning in the direction across the slit with 256 step positions at 0.25 mrad per step yields a 256
256 image with a 64
64 mrad FOV. [7] The VIRTIS-M spectral cubes used in this study are listed in Table 1. We selected observational sequences offering all together a good latitudinal coverage from south pole to low northern latitudes (10 – 20°N). They usually correspond to off-pericenter observations (Science case 2) except for sequence 093– 01, which corresponds to Science case 3 (global spectroimaging from apocenter). The Venus Express science planning is described by Titov et al. [2006]. The integration time varies from 3 to 8 s, ensuring a good signal-to-noise ratio on the night side. The sequences marked in bold incorporate regions of relatively high elevation (Alpha Regio, Imdr Regio, Themis Regio) along with regions of low elevation, and have been used to constrain the CO2 continuum opacity (see section 4.2). Figure 1 shows an example of VIRTIS spectra averaged over two small areas observed in session 390 – 13. [8] A small amount of scattered sunlight is visible in the spectra at short wavelengths (Figure 1). To remove it from the nightside spectra, we used the intensity at 1.061 mm (Spectel 4) and 1.232 mm (Spectel 22), two wavelengths at which no nightside emission is expected from synthetic models. We assumed a linear spectral variation for this scattered component between these two wavelengths. The so-calculated scattered light also reproduces the residual intensity at 1.356 mm (Spectel 35), another wavelength at which models predict no nightside emission. [9] We found that the wavelength scale of the observations was slightly shifted from that given in the spectral cubes after the latest calibration of the VIRTIS VEX archive (v2.1, 18 March 2008). In the sequences selected, this shift

2 of 12

BE´ZARD ET AL.: WATER VAPOR NEAR THE SURFACE OF VENUS

E00B39

E00B39

Table 1. VIRTIS Nightside Sequencesa Sessionb 093 – 01 157 – 03 390 – 09 390 – 13 392 – 05 392 – 06 571 – 08 571 – 11 577 – 06 577 – 08 577 – 10 579 – 02 579 – 06 579 – 08 579 – 10 579 – 12

Date 22 25 16 16 17 17 12 12 18 18 18 20 20 20 20 20

Jul 2006 Sep 2006 May 2007 May 2007 May 2007 May 2007 Nov 2007 Nov 2007 Nov 2007 Nov 2007 Nov 2007 Nov 2007 Nov 2007 Nov 2007 Nov 2007 Nov 2007

Science Case

Exposure Time (sec)

Maximum and Minimum Latitudes

3 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2

3.3 3.3 8 8 8 8 3 3 8 8 8 8 8 8 8 8

7; 61 8; 44 9; 68 17; 56 34; 85 29; 82 5; 61 8; 70 10; 56 14; 55 15; 50 46; 90 9; 57 14; 55 16; 50 23; 31

Easternmost and Westernmost Longitudes 178; 331; 323; 331; 328; 337; 199; 245; 279; 223; 186; 178; 276; 228; 192; 229;

247 6 29 23 119 114 269 311 327 281 229 340 333 287 243 265

Comment includes Imdr Regio includes Alpha Regio includes Alpha Regio includes Imdr Regio includes Themis Regio

a

Visible and Infrared Thermal Imaging Spectrometer. Sessions in bold are those used to derive the CO2 continuum opacity (section 4.2).

b

varies between approximately 0.75 and 0.95 time the spectel size (9.5 nm). It varies reproducibly along the 256-pixel slit by 0.1 pixel, reaching a maximum around Pixel 100 (see section 5 and Figure 7). We considered this spectral shift as a free parameter in the fitting procedure used to derive the water vapor abundance.

assumed the following sublorentzian lineshape (with a cutoff at 180 cm1 from line center):

3. Radiative Transfer Model

where Ds is the distance from line center expressed in cm1 and c is the factor by which we multiply the Voigt lineshape. This c factor was found to best reproduce the low-frequency side of the 1.18-mm window, dominated by CO2 absorption, as observed in high-resolution groundbased spectra [de Bergh et al., 1995]. Various analyses of the Venus nightside windows have shown that an additional continuum opacity, likely due to collision-induced CO2 bands and/or extreme far wings of strong allowed CO2 bands, is a major contributor to the gas opacity. Its strength may be relatively well determined in the 2.3- and 1.74-mm windows by fitting the amplitude of allowed CO2 lines in high-resolution spectra [e.g., Be´zard et al., 1990; Pollack et al., 1993; de Bergh et al., 1995]. This is not the case for the 1.18- and 1.1-mm windows where no marked CO2 feature is present. Fortunately, it is possible to take advantage of the transparency of these two windows and thus of their sensitivity to the surface elevation to determine this ‘‘continuum’’ opacity, as shown in section 4. We parameterize it with a constant binary absorption coefficient, expressed in cm1 amagat2. [12] For each spectral selection we analyzed, we used the mean outgoing intensity from our model (i.e., the upward flux divided by p) and rescaled it by a factor f to reproduce the observed 1.18-mm intensity. This factor accounts for the variation in emission angle and cloud opacity from one selection to another. This simplified approach is valid because these two parameters are spectrally neutral and uniformly modulate the outgoing intensity. To check this, we calculated the radiance ratios of spectra at various emission angles up to 80° to that at zero emission angle. Figure 2 shows that these ratios are quasi-constant across the 1.18-mm window, with a variation less than 1%. The water vapor signature is thus fully preserved in changes of

[10] Synthetic spectra were calculated using a line-by-line radiative transfer model with scattering originally developed to analyze high-resolution spectra of Venus’ night side [Be´zard et al., 1990; de Bergh et al., 1995]. We improved over these earlier works by employing the Discrete Ordinates Radiative Transfer Program for a MultiLayered PlaneParallel Medium (DISORT) algorithm [Stamnes et al., 1988] with eight streams to solve the equation of transfer in place of a Delta-Eddington adding algorithm. A HenyeyGreenstein phase function was used for particle scattering. The temperature profile, assumed to be horizontally uniform, is based on entry probe data at low latitudes, as compiled by Seiff [1983, Table A1]. We did not update these data, the VIRA-2 profile [Moroz and Zasova, 1997] being the same as the VIRA one in the lower atmosphere. The CO2 mole fraction is 0.965 and the H2O profile is assumed constant with height below the clouds. Rayleigh scattering by CO2 is included. Surface emissivity was arbitrarily set to 0.95. We used the simplified cloud model described by Crisp [1986] that extends from 30 to 80 km. Extinction efficiencies, single scattering albedos and asymmetry parameters were generated layer by layer by calculating Mie scattering by four modes of 75% H2SO4 particles. [11] For CO2 line opacity, we used the high-T database presented by Pollack et al. [1993], while the H2O line parameters come from Geisa 97 [Jacquinet-Husson et al., 1999]. The CO2 self-broadened half width was taken as 0.1 cm1 atm1 and varying as T0.75. For CO2 – broadened H2O line half widths, we used a routine provided by R.H. Tipping and R. Freedman (private communication, 2000). We used a Voigt profile for the H2O lineshape with a cutoff at 120 cm1 from line center. For CO2 lines, we

Ds < 3 : c ¼ 1 3 < Ds < 60 : c ¼ 1:051 expðDs=60Þ 60 < Ds : c ¼ 0:6671 expðDs=110Þ;

3 of 12

E00B39

BE´ZARD ET AL.: WATER VAPOR NEAR THE SURFACE OF VENUS

E00B39

Figure 2. Ratios of synthetic spectra calculated at 17-nm resolution for different emission angles (solid lines), twice and half the optical depth of the lower cloud layer (dashed lines), and for surface elevation of 1 and 2 km (lines with triangles). In all cases, the reference spectrum corresponds to the nominal cloud model described in the text, a zero surface elevation, and a zero emission angle. the emission angle and the mean intensity throughout the window is proportional to the intensity at any emission angle. This occurs because the clouds are optically thick and located well above the line-forming region. In Figure 2, we also show the radiance ratios of models in which the cloud optical depth in the middle and lower cloud layers (below 57 km) is equal to half and twice that in the nominal model. A negligible variation of these ratios across the window is found (less than 1%). Cloud scattering therefore acts a gray attenuator (or a scaling factor) and does not modify the spectral shape of the window because it takes place well above the line-forming region at 1.18 mm with negligible thermal emission by the particles themselves. This also means that our water vapor retrievals are not sensitive to uncertainties in our cloud model.

4. Determination of the Water Vapor Abundance 4.1. Spectral Resolution of VIRTIS-M Observations [13] A precise knowledge of the spectral resolution of the VIRTIS spectra is important for an accurate determination of the H2O abundance, all the more as the wavelength calibration of the data is not known to better than about half a spectel. Water vapor absorption affects the width of the 1.18-mm window so that an error in the spectral resolution can be offset to first order by a change in the H2O abundance and a slight variation in the pixel registration. The spectral profile of the VIRTIS-M infrared channel, around the center of the slit, has been measured before launch. The profile could be fit with a Gaussian having a full width at half maximum (FWHM) of 11.0 ± 1.2 nm. However, synthetic spectra convolved with this profile do not match the VIRTIS Venus nightside spectra, as evidenced at 1.74 mm, the narrowest window. To determine the actual resolution of the observations, we used this window and compared VIRTIS spectra to high-resolution ground-based

spectra recorded at the Canada-France-Hawaii telescope (CFHT) by one of us (BB), Catherine de Bergh, Dave Crisp and Jean-Pierre Maillard [Taylor et al., 1997, Figure 4]. We convolved the latter spectrum with Gaussian functions having different FWHMs and determined the best value through a least squares fit. Figure 3 shows the result for the Alpha Regio spectrum of Figure 1: The CFHT spectrum convolved at 11 nm resolution is too narrow compared to the VIRTIS spectrum and does not produce enough emission in the wings of the window whereas a FWHM of 17 nm yields a very good fit. We found that, among the VIRTIS sessions marked in bold in Table 1, the FWHM of the 1.74-mm window varies between 15 and 19 nm. We used the so-inferred resolution to analyze the 1.18-mm window. The actual resolution of the VIRTIS-M infrared spectra thus appears to be at least 50% larger than measured in the laboratory owing to different temperatures of the spectrometer (135 K for the laboratory measurement versus around 150– 160 K in orbit). 4.2. Determination of the CO2 Continuum Absorption [14] As discussed in section 3, the presence of an unknown continuum opacity in the windows, likely due to CO2 collision-induced bands or extreme far wings of strong allowed bands, adds some uncertainty in the composition retrievals. To constrain this quantity, we analyzed VIRTIS 1.18-mm spectra taken over regions of different elevations during the same session. The contrast between high and low elevation spectra, due to differences in surface temperatures, is sensitive to the gas opacity, e.g., a larger continuum absorption increases the atmospheric opacity, attenuates more strongly the surface emission from lower elevations, and thus diminishes the contrast. In five different sessions, we selected spectra over areas at relatively high elevation (Alpha Regio, Themis Regio or Imdr Regio) and at low elevation. We used the altimetry derived from Magellan and

4 of 12

BE´ZARD ET AL.: WATER VAPOR NEAR THE SURFACE OF VENUS

E00B39

E00B39

Figure 3. The 1.74-mm VIRTIS spectrum of Alpha Regio in Figure 1 (squares) is compared with CanadaFrance-Hawaii telescope (CFHT) spectra convolved with a Gaussian function having full width at half maximum (FWHMs) of 11 nm (dashed line) and 17 nm (solid line). In each case, a multiplying factor f that minimizes the sum of squares of residuals is applied to the CFHT data, a 0.0007 W sr1 m2 mm1 is added to account for residual scattered sunlight in the instrument, and a wavelength shift lshift = 6.8 nm is applied to the VIRTIS spectrum. smoothed it with a boxcar average of width 80– 100 km to account for blurring by the overlying cloud layers [e.g, Moroz, 2002]. The spectral samples, listed in Table 2, comprise 12 to 285 spectra. [15] Besides elevation, the intensity of the emission at 1.18 micron is also dependent on the emission angle and the cloud opacity. We then corrected the spectra to zero emission angle using the limb-darkening model derived by Carlson et al. (manuscript in preparation, 2008) and we used the peak intensity in the windows at 1.27, 1.31 and 1.74 microns, not probing down to the surface, as an indicator of the cloudiness. The 1.31-mm window (Spectel 30) is in principle best suited for this purpose [Meadows and Crisp, 1996] but it is weak and has a relatively low

signal-to-noise ratio. The 1.28-mm intensity at Spectel 27 is stronger but slightly contaminated by the 1.27-mm O2 airglow peaking at Spectel 26. The 1.74-mm window (Spectel 76) has a very good signal-to-noise ratio but, being at much longer wavelengths, it is not a perfect indicator of the cloud opacity at 1.18 mm: spectra with similar 1.74-mm radiances can differ at 1.18 mm owing to different particle size characteristics as was investigated from Galileo/NIMS data in the case of the 2.3- and 1.74-mm windows by Carlson et al. [1993]. To select low elevation regions having similar cloudiness as the high elevation regions, we used a combination of these three windows and empirically imposed intensity variations less than 4% at 1.28 mm, 6% at 1.31 mm and 15% at 1.74 mm. In each session, we

Table 2. VIRTIS Nightside Spectra Used to Determine the CO2 Continuum Opacity and H2O Mole Fraction a

Session

Pixel Range

093 – 01 093 – 01 093 – 01 093 – 01 157 – 03 157 – 03 157 – 03 390 – 13 390 – 13 390 – 13 571 – 08 571 – 08 571 – 11 571 – 11 571 – 11

120 – 122; 70 – 73 85 – 92; 73 – 79 134 – 143; 95 – 103 144 – 158; 61 – 66 64 – 71; 214 – 220 88 – 96; 177 – 183 19 – 27; 88 – 96 112 – 130; 149 – 163 141 – 159; 126 – 140 69 – 80; 147 – 163 218 – 220; 78 – 81 248 – 254; 58 – 65 93 – 97; 137 – 145 88 – 98; 33 – 43 131 – 139; 138 – 159

Mean Latitude and Longitude 46.4; 43.4; 41.4; 48.8; 26.0; 26.5; 36.9; 26.6; 25.9; 31.7; 46.2; 46.8; 40.8; 51.5; 43.5;

214.8 220.7 208.7 208.2 359.2 355.0 351.9 359.9 354.8 4.2 214.6 206.8 277.4 291.1 270.2

Mean Local Time

Mean Altitude (km)

Mean Emission Angle (deg)

23.10 22.68 23.48 23.51 2.76 3.04 3.24 21.38 21.04 21.66 22.61 22.09 21.22 20.30 21.69

2.58 0.03 1.35 0.14 1.68 0.37 1.11 1.30 0.47 0.13 2.68 0.02 2.06 0.07 0.17

36 38 43 35 28 26 15 45 45 41 32 33 37 32 32

a

Along the slit; across the slit. Assuming a CO2 continuum absorption of 1.1
109 cm1 amagat2.

b

5 of 12

H2O Mole Fractionb (ppm) 42 43 41 39 39 41 47 43 42 45 49 51 49 54 48

± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

3 3 3 3 4 4 5 3 3 3 4 4 4 5 5

E00B39

BE´ZARD ET AL.: WATER VAPOR NEAR THE SURFACE OF VENUS

E00B39

Figure 4. Relative variations of the intensity at 1.185 mm as a function of surface elevation for the Visible and Infrared Thermal Imaging Spectrometer (VIRTIS) spectral selections in Table 2 (symbols). Model variations for different continuum opacity (0.5, 1.0, and 1.5
109 cm1 amagat2) are shown for comparison. chose a region around 0 km altitude and, if possible, one at a lower elevation. Figure 1 shows two averaged spectra from session 390– 13, one at 1.30 km and the other one at 0.47 km. One clearly sees that the intensities at 1.1 and 1.18 mm are larger for the low elevation spectrum whereas the intensities in the 1.27, 1.31 and 1.74-mm windows are similar in the two spectra, which indicates similar cloudiness. [16] Figure 4 shows the relative intensity observed at Spectel 17 (1.185 mm) for the highest and lowest elevations of each session with respect to that at 0 km altitude in the same session. We estimate that each datum point bears an uncertainty of ±0.02 owing to the cloud opacity. A least square fit to these data indicates a variation of the intensity of 5.7 ± 0.4% per km. Comparison with model calculations indicates that a constant absorption coefficient of 1 ± 0.4
109 cm1 amagat2 best reproduces the observed variation. The corresponding optical depth at 0 km is 0.92, larger than that due to H2O lines (0.36 for a mole fraction of 40 ppm) and CO2 lines (0.68) at 1.182 mm and 17-nm resolution, which shows that this continuum opacity is an important contributor to the gaseous absorption in this window. The Rayleigh scattering optical depth is 0.65. We checked that changing the surface emissivity (from 0.95 to 0.80) does not significantly affect the derived continuum absorption coefficient. 4.3. Determination of the H2O Mole Fraction [17] To determine the water vapor abundance, we fitted the spectra in Table 2 with synthetic models calculated with the above determined continuum opacity and various H2O mole fractions. For each value of H2O, we determined the values of the spectral shift lshift and scaling factor f (related to cloud opacity and emission angle) that provide the best fit of the 1.18-mm VIRTIS spectrum and calculate the residuals of the fit. We then retained the value of the H2O mole fraction that minimizes the sum of squares of these resid-

uals. In Figure 5, the Alpha Regio spectrum (Figure 1) is compared to synthetic spectra calculated with 30, 40 and 55 ppm of H2O, which shows the strong effect of water vapor absorption on the spectra. Figure 6 shows that the two spectra of Figure 1 that have different surface elevations are well fitted with a continuum absorption coefficient of 1
109 cm1 amagat2 and a water mole fraction of 40 ppm (the same spectral shift and scaling factor are applied to both spectra). [18] The H2O mole fraction that best reproduces the spectral selections, assuming a continuum absorption coefficient of 1.1
109 cm1 amagat2, is indicated in the last column of Table 2 with the 1 SD error bar derived from the residuals of the fit. The weighted mean value is 45 ppm (44 ppm if a continuum absorption coefficient of 1
109 cm1 amagat2 is used) and the standard deviation of the sample is 5 ppm. The ±40% uncertainty on the CO2 continuum absorption adds a ±3 ppm uncertainty on the H2O mole fraction and the estimated ±10% uncertainty on the VIRTIS-M spectral resolution at 1.18 mm adds another ±7 ppm. Combining quadratically all these uncertainties, we conclude that the H2O mole fraction in the lowest scale height of the atmosphere is 44 ± 9 ppm. Note that this uncertainty does not take into account any possible inaccuracies or incompleteness in the H2O spectral database used here (Geisa 97).

5. Search for Horizontal Variations [19] We searched for possible variations of the H2O abundance in the sequences listed in Table 1. Besides the H2O abundance, the intensity of the 1.18-mm window emission is sensitive to the cloud optical depth, emission angle and surface elevation (Figure 2), all of which vary spatially. It is thus necessary to correct the images for the variations in these parameters or to work on quantities that

6 of 12

E00B39

BE´ZARD ET AL.: WATER VAPOR NEAR THE SURFACE OF VENUS

E00B39

Figure 5. The Alpha Regio spectrum of Figure 1 is compared to synthetic spectra with H2O mole fractions of 30, 40, and 55 ppm. The spectral shift lshift = 7.9 nm and scale factor f = 0.78 are those that yield the best fit of the data with the 40-ppm calculation. are least sensitive to them. The effect of the first two parameters is spectrally neutral across the window (Figure 2) and can therefore be eliminated by ratioing the intensity at two wavelengths. A complication arises from the fact that the spectral registration is not constant along the slit as mentioned above and that the spectral resolution may also slightly vary. The first step is thus to determine the spectral shift for each pixel of the images. This was done by determining the peak of the 1.18-mm window from a parabolic fit of the intensity at spectels 16, 17 and 18 and assigning it to 1.182 mm, as determined from synthetic

calculations (Figure 6). The result is shown for a particular sequence (579 – 08) in Figure 7. In all sequences, the spectral shift increases by 1 nm (0.1 spectel) along the slit from Pixel 6 up to Pixels 80 –100 and slightly decreases beyond. The variation of the spectral shift in the scan direction, i.e., during the acquisition, is negligible for this sequence (Figure 7) and all those investigated here. [20] We first considered the ratio R1 of the intensity at 1.169 mm to that at the peak of the window (1.182 mm) to map water vapor variations. The first wavelength lies in the wing of the window that is dominated by H2O absorption.

Figure 6. The two spectra of Figure 1 (session 390 –13) at elevations of 1.30 and 0.47 km are compared with our best fit model that includes a constant absorption coefficient of 1
109 cm1 amagat2 and a H2O mole fraction of 40 ppm. The same spectral shift and scale factor as in Figure 5 are used for both spectra. 7 of 12

E00B39

BE´ZARD ET AL.: WATER VAPOR NEAR THE SURFACE OF VENUS

Figure 7. (top) Map of the 1.182-mm intensity (W sr1 m2 mm1) in sequence 579 – 08 and of the spectral shift, expressed as a fraction of a spectel, with respect to the VIRTIS/Venus Express archive calibration (v2.1). (bottom) Plots of the shift as a function of detector line number along the slit for two different frames (i.e., image lines). The intensity at 1.182 mm is derived from the above mentioned parabolic fit and that at 1.169 mm is interpolated linearly between the two adjacent spectels. Synthetic calcu-

E00B39

lations show that a 10% variation of the H2O mixing ratio yields a 3.5% variation of this ratio. Because this ratio is potentially very sensitive to remaining uncertainties in the spectral registration, we considered a second parameter (W), which is the FWHM of the window. Calculations indicate that a 10% variation of the H2O mixing ratio yields a 2.3% variation of this quantity. The maps of these two quantities are shown in Figure 8 for sequence 579 – 08. Clearly, the effect of cloud opacity and emission angle variations is very well removed as attested by comparing to the 1.182-mm image. However, features remain that are not due to water vapor variations. First, there is a slight increase of the two parameters R1 and W with increasing pixel number along the slit (from left to right in Figure 8) for all image lines. This probably reflects a slight increase of the spectral width of the instrumental function from the beginning (small detector line numbers) to the end (large detector line numbers) of the 256-pixel slit. Second, topographic features are clearly seen in the R1 and W images and actually a very good correlation is observed with the altimetry map also shown in Figure 8. This occurs because altitude variations affect more the peak of the window, where the gas opacity is at minimum, than its wings. A higher altitude region is therefore characterized by a larger value of R1 and a larger FWHM as the 1.182-mm intensity is more strongly reduced than those at half maximum in the wings. Consequently, we defined a third parameter R2 to minimize the influence of the topography and of the spectral resolution, that is the ratio of the intensity in the CO2 wing at 1.1945 mm to that in the H2O-dominated wing at 1.169 mm. These two wavelengths exhibit similar intensities and probe similar atmospheric levels so that their intensity ratio is not too sensitive to the surface altitude. Synthetic calculations indicate that this ratio shows little sensitivity to small variations of the spectral resolution and that a 10% variation in the H2O mixing ratio causes a 5% variation of this intensity ratio R2. [21] The map of the R2 ratio, shown in Figure 8, is bland except for the hot pixels, which confirms that all the effects above mentioned have been removed with a very good accuracy. Figure 9 shows cuts through this image with peakto-peak variations of about 3%, attributable to noise and a

Figure 8. Maps of the 1.182-mm intensity I (W sr1 m2 mm1), smoothed surface elevation Z (km), ratio of the intensity at 1.169 and 1.182 mm (R1), FWHM of the window (spectels), and ratio of the intensity at 1.1945 and 1.169 mm (R2) for sequence 579 –08. 8 of 12

E00B39

BE´ZARD ET AL.: WATER VAPOR NEAR THE SURFACE OF VENUS

E00B39

from the day side and we set an envelope of ±1% for any possible latitudinal variation in the whole range 80°S – 25°N. These upper limits translate into ±1% (±0.5 ppm) for the H2O mixing ratio in the range 60°S – 25°N and ±2% (±1 ppm) for the broader range 80°S–25°N.

6. Discussion

Figure 9. Image of the 1.1945- to 1.169-mm intensity ratio (R2) shown in Figure 8, and cuts of this image at line (frame) 70 and at column (detector line) 20. residual altitude signature at the 1% level. We have calculated the median of R2 in latitude boxes 5° wide (Figure 10) and found no variations beyond ±0.3% from the mean from 60°S to 15°N for this sequence 579– 08. The same analysis was performed for all sequences in Table 1 (examples shown in Figure 10). Data southward of 80°S cannot be used owing to a large amount of scattered light from the dayside. Note that for some sequences, having a lower spectral shift than in most cases, we used other wavelengths to calculate the R2 ratio (namely 1.1905 and 1.1685 mm), which allows us to better correct the above mentioned effects. We do not see any significant variations of the R2 intensity ratio with latitude in any sequence and we place an upper limit of ±0.5% for the range 60°S – 25°N. The data poleward of 60°S are somewhat affected by scattered light

[22] The absorption coefficient we inferred for the additional continuum opacity in the 1.18-mm window (1 ± 0.4
109 cm1 amagat2) agrees with that derived independently by Carlson et al. (manuscript in preparation, 2008) from VIRTIS and Galileo/NIMS data. Both analyses assume a temperature lapse rate from the VIRA model [Seiff, 1983] in all investigated regions. If it is shallower (and thus more stable) than in the VIRA model, as argued by Meadows and Crisp [1996], a smaller continuum opacity would be needed. Also, our analysis relies on the assumption of constant surface emissivity at 1.18 mm over the regions we investigated. If, for example, the surface emissivity instead decreases with elevation in the near-infrared as it generally does in Magellan radar images, a larger continuum opacity would be needed to reproduce the observed intensity variation. To our knowledge, we provide here the first determination of this continuum opacity, which is very difficult to measure in the laboratory as it requires very large optical path lengths and high temperatures. [23] Having constrained this important source of opacity, we could determine the H2O mole fraction with a relatively good accuracy. The result, 44 ± 9 ppm, pertains to the 0– 15 km range. It agrees with most previous determinations as reviewed by Be´zard and de Bergh [2007], in particular with that of Meadows and Crisp [1996] (45 ± 10 ppm). We do not have any vertically resolved information given the small range of surface elevation in the data sets we analyzed (3 km) and therefore we cannot discriminate between the somewhat different conclusions of Meadows and Crisp [1996] and Ignatiev et al. [1997]. From ground-based observations of the 1.0, 1.1 and 1.18-mm windows and using the topography for altitude resolution, Meadows and Crisp concluded that the H2O mole fraction was constant (within ±10%) in the lowest scale height (0 – 15 km) at a value of 45 ± 10 ppm. In contrast, from a reanalysis of the Venera optical spectra, Ignatiev et al. [1997] argued that the H2O mole fraction probably increases from about 20 ppm at 10– 20 km to 50 –70 ppm below 5 km. Both distributions yield column-averaged mixing ratios (45 and 37 ppm respectively) that agree with our result within error bars. [24] It is important to note that our determination was obtained using the Geisa 97 database for the H2O absorption. Given the large optical paths and high temperatures involved, it is possible that weak high-energy lines, not included in Geisa (or Hitran), provide a significant contribution in the H2O wing of the 1.18-mm window. If this was the case, the mole fraction needed to reproduce the 1.18-mm observations would probably be lower than the 44 ppm we derived. [25] The 2.3-mm nightside window gives the opportunity to measure the water abundance at higher altitudes, in the range 30– 45 km. A recent analysis of VIRTIS-H spectra indicates a mixing ratio of 31 ± 2 ppm [Marcq et al., 2008], where the error bars derive from the residuals of the fits and

9 of 12

E00B39

BE´ZARD ET AL.: WATER VAPOR NEAR THE SURFACE OF VENUS

E00B39

Figure 10. Variations of the intensity ratio R2 with latitude in sequences 579– 08, 577– 06, 577 – 08, and 577– 10. Images of these sequences at 1.182 mm are also shown. do not account for systematic errors such as in the CO2 continuum absorption. This result is consistent with previous ground-based observations, which indicated 30 ± 10 ppm (see review by Taylor et al. [1997]). Emission in the 1.74-mm window is also sensitive to water vapor absorption and originates from 15 to 25 km, a region intermediate between those probed at 2.3 and 1.18 mm. Ground-based observations of this window again indicate a mole fraction of 30 ± 10 ppm [Pollack et al., 1993; de Bergh et al., 1995]. The 2.3- and 1.74-mm results preclude the large abundance (67 ppm) derived from the Pioneer Venus mass spectrometer measurements [Donahue and Hodges, 1992]. [26] Taken at face value, our H2O mole fraction is about 50% larger than that derived at 2.3 and 1.74 mm, which suggests an increase of the water abundance in the lowest scale height of the atmosphere. However, given the error bars, we feel that a constant mole fraction around 30 –40 ppm from below the clouds down to the surface is still consistent with all analyses of Venus’ nightside emission. A uniform profile agrees with chemical models, which do not predict a significant variation of the H2O mole fraction between 0 and 38 km [Krasnopolsky, 2007]. Below the clouds, the only significant sink for H2O is formation of H2SO4 vapor above 38 km. At 40 km, the H2O mole fraction is depleted by about 2 ppm and at 45 km by about 7 ppm [Krasnopolsky, 2007]. The 2.3-mm window has some weak sensitivity to the 40– 45 km region and high spectral resolution observations might be able to measure this decrease with height. If real, the strong increase of the water mole fraction below 5 km suggested by Ignatiev et al. [1997] would imply a source at or near the surface and a sink above 5 km that are not predicted by existing chemical models and remain to be identified. [27] We have obtained very stringent upper limits for any variation of the H2O mole fraction at 0– 15 km with latitude in several VIRTIS-M sequences: ±0.5 ppm in the range 60°S – 25°N and ±1 ppm if we extend the range to 80°S– 25°N. Drossart et al. [1993] previously searched for H2O variations using the same 1.18-mm window in the Galileo/ NIMS data. This study was limited to three narrow north to south strips covering small areas between 40°S and 50°N.

No variation exceeding ±20% (±6ppm) was found in this latitude range (±10% for the smaller range 30°S – 30°N). Marcq et al. [2006] searched for horizontal variations of H2O in their NASA/IRTF observations of the 2.3-mm nightside window, probing the 30– 45 km region. They did not see any beyond ±15% (±4 ppm) in the latitude range 40°S – 40°N. In contrast, CO shows a 30 – 40% increase from the equator to 60°S at 36 km and an anticorrelated decrease of OCS at 33 km [Marcq et al., 2006, 2008; Tsang et al., 2008]. These variations are interpreted as the signature of the large-scale circulation, with upwelling at the equator and downwelling at high latitudes, in the presence of vertical gradients of CO and OCS due to chemical sources and sinks. The fact that we do not see any variations of the H2O abundance in the first scale height of the atmosphere is fully consistent with the constancy of the vertical profile with height below 40 km and the absence of significant sources and sinks. A contrario, our result probably precludes the factor of 2 –3 increase of the H2O mole fraction from 10 to 0 km advocated by Ignatiev et al. [1997]: General Circulation Model (GCM) calculations by Lebonnois et al. (manuscript in preparation, 2008) suggest an alternance of upward and downward motions from equator to pole in the lowest 20 km, which would probably induce some measurable latitude variation of the H2O abundance we determined, with larger (smaller) abundances in upwelling (downwelling) regions.

7. Conclusions [28] Observations of the night side of Venus by the Venus Express/VIRTIS-M instrument in the 1.18-mm atmospheric window have been used to measure and map the distribution of H2O near the surface of Venus. We have first determined the continuum opacity present in this window from the variation of the peak intensity with surface elevation. This opacity likely results from CO2 collision-induced bands and far wings of strong CO2 bands outside the window (beyond 180 cm1, which is the cutoff we used here), as is the case at 2.3 mm [Tonkov et al., 1996]. The absorption coefficient we inferred is 1 ± 0.4
109 cm1 amagat2. At the

10 of 12

E00B39

BE´ZARD ET AL.: WATER VAPOR NEAR THE SURFACE OF VENUS

VIRTIS-M resolution (17 nm), the corresponding optical depth is similar to that due to molecular bands of CO2 and H2O. Up to now, CO2 continuum opacity has been measured only in the 2.3-mm window and at room temperature [Tonkov et al., 1996]. Laboratory measurements in the 1.18and 1.1-mm windows at temperatures relevant to Venus’ deep atmosphere would be extremely useful but are very difficult owing to the huge CO2 path lengths and high temperatures (600 – 730 K) required. [29] Analyzing spectral selections in various areas of Venus’ southern hemisphere, we derived from radiative transfer calculations a water vapor mole fraction of 44 ± 9 ppm, which pertains to the altitude range 0 to 15 km. The quoted error bars do not include possible errors or incompleteness of the H2O spectroscopic data used here (Geisa 97). It is highly desirable to assess whether the Geisa or Hitran databases are sufficient to model the H2O absorption in the 1.18-mm window. If weak hot lines, not included in these databases, contribute significantly to the opacity, the H2O mole fraction needed to reproduce the VIRTIS data would probably be lower than inferred here. Besides this caveat, the main source of uncertainty lies in the spectral resolution of the VIRTIS-M observations that varies from orbit to orbit and was determined for each sequence using the narrow 1.74-mm window. On the other hand, the VIRTIS data have the advantage of an excellent spatial resolution, which permits to obtain spectra at welldefined surface elevations. The limited range of surface elevations (0.5 to 2.7 km) covered by the present analysis did not allow us to retrieve a vertical profile for water vapor following Meadows and Crisp’s [1996] approach. An extension of the Venus Express mission should provide observations of higher terrains, such as Aphrodite Terra, which could constrain the H2O profile in the 0 – 10 km range. This is important to discriminate between a constant with height profile and a large increase or decrease of the mole fraction below 5 km as suggested by some in situ measurements [Ignatiev et al., 1997; Donahue and Hodges, 1992]. The water mole fraction we derived is somewhat larger but still consistent, within error bars, with that determined at higher altitudes from the 2.3-mm (30 – 45 km) and 1.74-mm (15 – 25 km) nightside emission, which suggests that the water vapor mole fraction is constant below the clouds between 0 and 40 km, as predicted by chemical models [Krasnopolsky, 2007]. [30] We finally searched for spatial variations of the H2O abundance in various VIRTIS sequences covering altogether a large latitude range (80°S – 25°N). To do so, we used the ratio of the intensity in the two wings of the 1.18-mm window as a proxy. This search was negative and we derived stringent upper limits for any variation of the H2O mole fraction with latitude: ±1% for the range 60°S –25°N and ±2% if we consider the broader range 80°S – 25°N. These constraints are much stronger than those derived by Drossart et al. [1993] (±20%) from Galileo/NIMS spectra over a more limited spatial coverage. The horizontal uniformity of the H2O abundance is consistent with an abundance profile constant with altitude and with the expected lack of significant sources and sinks below 40 km. In fact, our results likely preclude any strong vertical gradient of the H2O concentration in the lowest scale height as vertical motions predicted by GCM calculations would then produce

E00B39

detectable horizontal variations as is the case for CO and OCS near 35 km [Marcq et al., 2006, 2008; Tsang et al., 2008]. [31] Acknowledgments. B.B., E.M., and P.D. acknowledge support from Centre National d’E´tudes Spatiales (CNES). G.P. was supported by Agenzia Spaziale Italiana (ASI). We gratefully thank the VIRTIS technical team for their continuous support.

References Be´zard, B., and C. de Bergh (2007), Composition of the atmosphere of Venus below the clouds, J. Geophys. Res., 112, E04S07, doi:10.1029/ 2006JE002794. Be´zard, B., C. de Bergh, D. Crisp, and J.-P. Maillard (1990), The deep atmosphere of Venus revealed by high-resolution nightside spectra, Nature, 345, 508 – 511, doi:10.1038/345508a0. Carlson, R. W., L. W. Kamp, K. H. Baines, J. B. Pollack, D. H. Grinspoon, T. Encrenaz, P. Drossart, and F. W. Taylor (1993), Variations in Venus cloud particle properties: A new view of Venus’s cloud morphology as observed by the Galileo Near-Infrared Mapping Spectrometer, Planet. Space Sci., 41, 477 – 485. Crisp, D. (1986), Radiative forcing of the Venus mesosphere: I. Solar fluxes and heating rates, Icarus, 67, 484 – 514, doi:10.1016/0019-1035(86)90126-0. Crisp, D., D. A. Allen, D. H. Grinspoon, and J. B. Pollack (1991), The dark side of Venus: Near-infrared images and spectra from the AngloAustralian Observatory, Science, 253, 1263 – 1266, doi:10.1126/science. 11538493. de Bergh, C., B. Be´zard, D. Crisp, J.-P. Maillard, T. Owen, J. Pollack, and D. Grinspoon (1995), Water in the deep atmosphere of Venus from highresolution spectra of the night side, Adv. Space Res., 15(4), 79 – 88, doi:10.1016/0273-1177(94)00067-B. Donahue, T. M., and R. R. Hodges Jr. (1992), Past and present water budget of Venus, J. Geophys. Res., 97, 6083 – 6091. Drossart, P., et al. (1993), Search for spatial variations of the H2O abundance in the lower atmosphere of Venus from NIMS-Galileo, Planet. Space Sci., 41, 495 – 504. Ignatiev, N. I., V. I. Moroz, B. E. Moschkin, A. P. Ekonomov, V. I. Gnedykh, A. V. Grigoriev, and I. V. Khatuntsev (1997), Water vapour in the lower atmosphere of Venus: A new analysis of optical spectra measured by entry probes, Planet. Space Sci., 45, 427 – 438, doi:10.1016/S00320633(96)00143-2. Jacquinet-Husson, N., et al. (1999), The 1997 spectroscopic GEISA databank, J. Quant. Spectrosc. Radiat. Transfer, 62, 205 – 254, doi:10.1016/ S0022-4073(98)00111-3. Krasnopolsky, V. A. (2007), Chemical kinetic model for the lower atmosphere of Venus, Icarus, 191, 25 – 37, doi:10.1016/j.icarus.2007.04.028. Marcq, E., T. Encrenaz, B. Be´zard, and M. Birlan (2006), Remote sensing of Venus’ lower atmosphere from ground-based IR spectroscopy: Latitudinal and vertical distribution of minor species, Planet. Space Sci., 54, 1360 – 1370, doi:10.1016/j.pss.2006.04.024. Marcq, E., B. Be´zard, P. Drossart, and G. Piccioni (2008), A latitudinal survey of CO, OCS, H2O, and SO2 in the lower atmosphere of Venus: Spectroscopic studies using VIRTIS-H, J. Geophys. Res., 113, E00B07, doi:10.1029/2008JE003074. Meadows, V. S., and D. Crisp (1996), Ground-based near-infrared observations of the Venus nightside: The thermal structure and water abundance near the surface, J. Geophys. Res., 101, 4595 – 4622, doi:10.1029/ 95JE03567. Moroz, V. I. (2002), Estimates of visibility of the surface of Venus from descent probes and balloons, Planet. Space Sci., 50, 287 – 297, doi:10.1016/S0032-0633(01)00128-3. Moroz, V. I., and L. V. Zasova (1997), VIRA-2: A review of inputs for updating the Venus International Reference Atmosphere, Adv. Space Res., 19, 1191 – 1201, doi:10.1016/S0273-1177(97)00270-6. Piccioni, G., et al. (2007), VIRTIS: The Visible and Infrared Thermal Imaging Spectrometer, Eur. Space Agency Spec. Publ., ESA SP 1295, 1 – 27. Pollack, J. B., et al. (1993), Near-infrared light from Venus’ nightside: A spectroscopic analysis, Icarus, 103, 1 – 42, doi:10.1006/icar.1993.1055. Seiff, A. (1983), Thermal structure of the atmosphere of Venus, in Venus, edited by D. M. Hunten et al., pp. 215 – 279, Univ. of Ariz. Press, Tucson. Stamnes, K., S. C. Tsay, W. Wiscombe, and K. Jayaweera (1988), Numerically stable algorithm for discrete-ordinate-method radiative transfer in multiple scattering and emitting layered media, Appl. Opt., 27, 2502 – 2509, doi:10.1364/AO.27.002502. Taylor, F. W., D. Crisp, and B. Be´zard (1997), Near-infrared sounding of the lower atmosphere of Venus, in Venus II: Geology, Geophysics, Atmo-

11 of 12

E00B39

BE´ZARD ET AL.: WATER VAPOR NEAR THE SURFACE OF VENUS

sphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 325 – 351, Univ. of Ariz. Press, Tucson. Titov, D. V., et al. (2006), Venus Express science planning, Planet. Space Sci., 54, 1279 – 1297, doi:10.1016/j.pss.2006.04.017. Tonkov, M. V., N. N. Filippov, V. V. Bertsev, J. P. Bouanich, N. Van-Thanh, C. Brodbeck, J. M. Hartmann, C. Boulet, F. Thibault, and R. Le Doucen (1996), Measurements and empirical modeling of pure CO2 absorption in the 2.3-mm region at room temperature: Far wings, allowed and collisioninduced bands, Appl. Opt., 35, 4863 – 4870, doi:10.1364/AO.35.004863.

E00B39

Tsang, C. C. C., et al. (2008), Tropospheric carbon monoxide concentrations and variability on Venus from Venus Express/VIRTIS-M observations, J. Geophys. Res., 113, E00B08, doi:10.1029/2008JE003089. 

B. Be´zard, P. Drossart, and E. Marcq, LESIA, Observatoire de Paris, 5 place Jules Janssen, F-92190, Meudon, France. (Bruno.Bezard@ obspm.fr) R. W. Carlson, Jet Propulsion Laboratory, MS 183-601, 4800 Oak Grove Drive, Pasadena, CA 91109, USA. G. Piccioni, INAF, IASF, via del Fosso del Cavaliere 100, I-00133 Rome, Italy. C. C. C. Tsang, Atmospheric, Oceanic and Planetary Physics, Department of Physics, University of Oxford, Clarendon Laboratory, Parks Road, Oxford, OX1 3PU, UK.

12 of 12

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114, E00B31, doi:10.1029/2008JE003117, 2009

for

Full Article

Venus cloud top winds from tracking UV features in Venus Monitoring Camera images R. Moissl,1 I. Khatuntsev,1,2 S. S. Limaye,3 D. V. Titov,1,2 W. J. Markiewicz,1 N. I. Ignatiev,2 T. Roatsch,4 K.-D. Matz,4 R. Jaumann,4 M. Almeida,5 G. Portyankina,1 T. Behnke,4 and S. F. Hviid1 Received 15 February 2008; revised 9 October 2008; accepted 10 November 2008; published 17 March 2009.

[1] To date dynamical observations of the Venus clouds have delivered mainly either only

short-term or long-term averaged results. With the Venus Monitoring Camera (VMC) it finally became possible to investigate the global dynamics with a relatively high resolution in space and time on a long-term basis. Our findings from manual cloud feature wind tracking in VMC UV image sequences so far show that the details of the mesospheric dynamics of Venus appear to be highly variable. Although the general rotation of the atmosphere remained relatively stable since Mariner 10, more than 30 years ago, by now, there are indications of short-term variations in the general circulation pattern of the Venus atmosphere at cloud top level. In some cases, significant variations in the zonal wind properties occur on a timescale of days. In other cases, we see rather stable conditions over one atmospheric revolution, or longer, at cloud top level. It remains an interesting question whether the irregularly observed midlatitude jets are indeed variable or simply become shielded from view by higher H2SO4 haze layers for varying time intervals. Winds at latitudes higher than 60°S are still difficult to obtain track because of low contrast and scarcity of features but increasing data is being collected. Over all, it was possible to extend latitudinal coverage of the cloud top winds with VMC observations. Thermal tides seem to be present in the data, but final confirmation still depends on synthesis of Visible and Infrared Thermal Imaging Spectrometer and VMC observations on night and dayside. Although poorly resolved, meridional wind speed measurements agree mainly with previous observations and with the presence of a Hadley cell spanning between equatorial region and about 45°S latitude. Citation: Moissl, R., et al. (2009), Venus cloud top winds from tracking UV features in Venus Monitoring Camera images, J. Geophys. Res., 114, E00B31, doi:10.1029/2008JE003117.

1. Introduction [2] The cloud level super rotation of the atmosphere, first determined from ground based images [Boyer and Guerin, 1969] has been measured from Mariner 10 [Limaye and Suomi, 1981], Pioneer Venus and Galileo missions [Rossow et al., 1990; Belton et al., 1991; Toigo et al., 1994; Limaye, 2007; Peralta et al., 2007]. [3] During its flyby at Venus in 1974 Mariner 10 collected image data with resolutions down to 30 km/pixel during the 8 days around its closest approach to the planet. The most 1 Max-Planck-Institut fuer Sonnensystemforschung, Katlenburg-Lindau, Germany. 2 Space Research Institute, Moscow, Russia. 3 Space Science and Engineering Center, University of WisconsinMadison, Madison, Wisconsin, USA. 4 Institut fur Planetenforschung, Deutsches Zentrum fur Luft- und Raumfahrt, Berlin, Germany. 5 European Space Astronomy Centre, European Space Agency, Madrid, Spain.

Copyright 2009 by the American Geophysical Union. 0148-0227/09/2008JE003117$09.00

extensive data set so far was obtained by the Pioneer Venus OCPP instrument between 1979 and 1986. The Galileo probe, which performed a flyby in 1990, acquired data during the 16 h around closest approach to Venus. The UV filter was centered at a wavelength of 400 nm. [4] The super rotation refers to the fact that bulk of the atmosphere of Venus rotates faster than the underlying solid planet from the surface to an altitude of
80 km, reaching speeds above the visible cloud top at rates
50 times faster than the planet. How this super rotation is maintained has been a puzzle since its discovery. The combination of the strong zonal flow and the weaker meridional flow explains the spiraling streaks seen in the ultraviolet images [Suomi, 1975; Smith and Gierasch, 1996] and is likely responsible for, or an artifact of the hemispheric vortex organization of the circulation centered over each pole of Venus [Limaye and Suomi, 1981; Limaye, 2007]. [5] The inferred global circulation is that the atmosphere is organized vertically in at least one Hadley circulation cell, wherein the solar heating in low latitudes causes rising motion and near the cloud level atmosphere flows toward the pole where radiative cooling leads to sinking to com-

E00B31

1 of 13

E00B31

MOISSL ET AL.: WIND TRACKING RESULTS FROM VMC UV IMAGES

plete the return flow (i.e., thermally direct circulation). The indirect evidence of this cell extending to the limiting altitude of the observations (
40 km) was determined from the detection of the solenoidal circulation from the temperature profiles determined from the radio occultation technique from the Pioneer Venus orbiter through the determination of the angle between the pressure and density surfaces [Limaye, 1985]. Winds inferred from tracking entry probes are inconclusive about whether there may be multiple cells between the surface and the 40 km level below which the thermal structure data from radio occultations are not available, but have been postulated [Schubert, 1983]. [6] The previous studies of the cloud motions have revealed aspects of the cloud level circulation such as Kelvin and gravity waves [Del Genio and Rossow, 1990; Smith and Gierasch, 1996] and presence of thermal tides [Limaye et al., 1982; Rossow et al., 1990; Limaye, 2007]. Investigations on these phenomena are part of the continuing effort to understand the processes that maintain the super rotation of the atmosphere [Schubert, 1983; Schubert et al., 2007] These include observations of the circulation below the clouds from entry probes or balloons and, more recently, on the night side from tracking of features in near infrared observations of Venus from Galileo [Carlson et al., 1991; Sa´nchez-Lavega et al., 2008] and Venus Express as well as from telescopes on Earth [Crisp et al., 1991; Limaye et al., 1988]. [7] Schubert [1983] discusses the different suggestions for the maintenance of the super rotation of the atmosphere, which include processes such as equatorward transport of angular momentum by eddies and angular momentum transport through solar thermal tides. Thus much of the effort continues to be directed at establishing the roles of eddies and solar tides in the transport of angular momentum. Limaye [2007] discussed the challenges posed by the lack of sufficient and complete observations due to the fact that until recently the thermal tides in the winds could be detected only from the cloud motions measured on the day side, which likely introduced a bias in the estimates of meridional transport of angular momentum by the mean and eddy circulation if the night side wind distributions were strong and different enough to cause the true zonal average to be substantially different from the day side average. The absence of night side information also impacts the inferences about the Kelvin waves [Del Genio and Rossow, 1990] to some degree and in the understanding of true temporal variations in the cloud level circulation on Venus. [8] One of the main goals of the Venus Express mission is to provide a global and systematic study of the atmospheric circulation [Svedhem et al., 2007; Titov et al., 2006]. We use sequences of UV images taken by Venus Monitoring Camera [Markiewicz et al., 2007a, 2007b] to measure the wind speeds at the cloud tops by tracking apparent motions of the cloud features. In comparison to the earlier investigations these observations provide extended and systematic latitude and local time coverage including high latitudes in nadir viewing geometry, significantly better temporal resolution and long-term coverage, and have higher spatial resolution. The preliminary results of the UV digital cloud tracking were presented by Markiewicz et al. [2007b]. Here we describe in detail the Venus Monitoring Camera (VMC) imaging sequences, the cloud tracking techniques, and the

E00B31

results of cloud tracking in VMC images from two slightly different manual tracking techniques during the primary mission of Venus Express. The digital tracking results will be briefly addressed in this work and presented in more detail at a later date pending further analysis.

2. Observations and Data Set [9] The Venus Monitoring Camera is a wide angle camera with a field of view of 17°, imaging Venus simultaneously in four narrow-band filter channels. All four channels are sharing one CCD detector, each one using one quadrant of the CCD. Two narrow band filters are in the near-IR (965 nm and 1010 nm), one in the visible (513 nm) and one in UV (365 nm). The UV channel is centered at the spectral signature of the unknown UV absorber. We use this filter to observe and track the UV cloud markings on the dayside, which are detected at around 70 km of altitude, where the optical depth in the UVapproaches unity [Tomasko et al., 1985; N. I. Ignatiev et al., Altimetry of the Venus cloud tops from the Venus Express observations, submitted to Journal of Geophysical Research, 2008]. A more detailed description of the VMC instrument, its properties and science goals has been presented in the work of Markiewicz et al. [2007a; 2007b]. [10] Because of the highly elliptical polar orbit of the Venus Express spacecraft the distance from the planet varies from
250 km at pericenter to
66.000 km in apocenter in the course of one 24 h orbit. Therefore the VMC angular resolution of 0.74 mrad per pixel translates into spatial resolution of approximately 0.2 km/pixel at pericenter and
50 km/pixel around apocenter. Consequently, VMC data yields high-resolution close-up images of the Northern Hemisphere as well as a global view of Venus from the South Pole. [11] In October 2007 Venus Express completed its primary mission [Svedhem et al., 2007; Titov et al., 2006]. In the course of 510 revolutions about the planet, the Venus Monitoring Camera acquired a total inventory of roughly 67,000 images with 31,800 in the UV among them. The data and results presented in this paper are based on the images obtained at 60,000 – 26,000 km distance from Venus, corresponding to a pixel size of 20– 50 km. Typical time intervals between image pairs used for cloud tracking are
40 min that corresponds to 5 – 10 pixel displacement of cloud features between the images. This selection is a compromise between the error in velocity measurements and temporal resolution. The planet coverage, regarding tracked UV markings, ranges from about 10°N to 80°S in latitude and 8 – 16 h local time. The contrast in VMC UV images varies between 5% and 30%, with typical values of 15– 30% for the tracked features. [12] Because of some minor damage to the detector due to solar irradiation, all images are subject to an in-flight flat fielding procedure before evaluation. In order to eliminate all artifacts in the VMC data, a series of images is acquired at high latitudes where the observed contrast is at a minimum. The images of these sequences are then averaged to achieve a uniformly gray background on which all persistent artifacts stand out prominently. All images in the corresponding orbit are then divided by the (normalized)

2 of 13

E00B31

MOISSL ET AL.: WIND TRACKING RESULTS FROM VMC UV IMAGES

E00B31

Figure 1. Coverage by VMC cloud-tracking measurements per 5° latitude  0.5 h local time bin. flat field. Because of a temporal evolution of some artifact patterns minor residuals may remain. [13] The Venus Express orbit can be divided roughly into three parts: apocentric (66,000 – 50,000 km distance from the planet), ascending branch (50,000 – 10,000 km) and pericentric (10,000 – 250 km). In these three orbital segments VMC images are acquired in order to study various types of features and observe different aspects of the cloud top dynamics. In apocentric images Venus is seen almost directly from beneath the South Pole, so the polar region is imaged without any foreshortening (unlike previous missions), but the low latitudes are seen in oblique view. This part of the VMC data set is used to study near-polar global dynamics and cloud morphology. In the ascending branch VMC observes middle and low latitudes in nadir geometry. Local time coverage depends on the sub spacecraft longitude that slowly changes from one orbit to another (
1.5 deg or 6 min per day). The ascending branch images are predominantly used to track winds from medium and small-scale features, ranging from about thousand down to several hundred kilometers in diameter. At close approach cloud features on scales down to few tens of kilometers in size can be identified in the images. However, at distances closer than 10,000 km wind tracking becomes increasingly more difficult as the rapid motion of the spacecraft decreases the overlap between images and features can be kept in the field of view only for short times. [14] Figure 1 shows the coverage of the latitude-localtime field by the VMC cloud tracking. The number of vectors peaks near downstream (to the west) of the sub solar region where a large number of discrete features of convective nature is regularly observed and drops toward higher latitudes as well as the morning and evening terminator. The main limiting factor for coverage in local time in the dusk

and dawn regions are the steep brightness gradients which, even in high-pass-filtered images, dampen contrast of the UV markings. At polar latitudes the morphology of the clouds is different altogether, presumably due to a decrease in or even absence of convective activity and presence of submicron haze [Kawabata et al., 1980]. This results in lack of discrete features, very low contrasts and drawn-out diffuse feature boundaries. In total, about 20,000 vectors have been extracted from
450 image pairs. [15] During the primary mission the descending branch of the Venus Express orbit was reserved for data transmission to Earth. In conjunction with the relative change in local solar time of the orbit plane, this lead to separation of the VMC wind tracking data into three periods in which it was possible to observe the dayside of Venus in ascending branch from a sufficient distance. These three periods are separated by gaps of almost 140 orbits in which VMC was pointed either to the night side or too close to the terminator to allow for tracking sequences to be taken in the required orbit segment. The first period spans from orbit 29 to 72 (containing tracking sequences from orbits 29, 30, 31, 34, 38, 46, 51, 56, 60, 61, 72, May– June 2006), the second period from orbit 208 to 295 (containing 208, 210, 230, 246, 250, 257, 258, 260, 263, 265– 267, 279, 281– 284, 295, November 2006 to January 2007), the third period from orbit 439 to 530 (containing 439, 440, 442, 453, 460– 463, 469, 471, 530, July – September 2007). Table 1 is listing the according dates of each orbit in the data set, together with the approximate coverage of the wind tracking in local time (LTR) and latitude (LAR), as well as the local time of the subspacecraft point at ascending node (LTAN). [16] Usually the individual wind tracking sequences are separated from each other by several days. But also sets of up to 4 consecutive orbits are present in the data set. These

3 of 13

MOISSL ET AL.: WIND TRACKING RESULTS FROM VMC UV IMAGES

E00B31

Table 1. List of All Orbits in the Data Seta Orbit

Date

LTR (h)

LAR (°)

LTAN (h)

29 30 31 34 38 46 51 56 60 61 72 208 210 230 246 250 257 258 260 263 265 266 267 279 281 282 283 284 295 439 440 442 453 460 461 462 463 469 471 530

19 May 2006 20 May 2006 21 May 2006 24 May 2006 28 May 2006 5 Jun. 2006 10 Jun. 2006 15 Jun. 2006 19 Jun. 2006 20 Jun. 2006 1 Jul. 2006 15 Nov. 2006 17 Nov. 2006 7 Dec. 2006 23 Dec. 2006 27 Dec. 2006 3 Jan. 2007 4 Jan. 2007 6 Jan. 2007 9 Jan. 2007 11 Jan. 2007 12 Jan. 2007 13 Jan. 2007 25 Jan. 2007 27 Jan. 2007 28 Jan. 2007 29 Jan. 2007 30 Jan. 2007 10 Feb. 2007 4 Jul. 2007 5 Jul. 2007 7 Jul. 2007 18 Jul. 2007 25 Jul. 2007 26 Jul. 2007 27 Jul. 2007 28 Jul. 2007 3 Aug. 2007 4 Aug. 2007 2 Oct. 2007

10.0 – 15.1 08.8 – 16.5 09.3 – 16.7 08.2 – 16.2 08.9 – 15.5 08.1 – 15.8 09.3 – 15.6 10.1 – 16.7 09.4 – 16.4 11.5 – 16.3 09.7 – 17.4 07.8 – 15.4 07.8 – 14.6 08.0 – 15.3 08.0 – 16.3 07.8 – 16.0 08.6 – 16.3 08.8 – 15.8 08.1 – 16.4 08.3 – 16.6 09.9 – 16.3 08.6 – 16.3 10.6 – 16.4 07.9 – 17.2 09.5 – 17.2 08.5 – 17.2 09.0 – 17.0 10.3 – 16.8 09.6 – 17.0 08.2 – 15.3 08.2 – 16.1 07.6 – 15.2 07.1 – 16.7 07.3 – 15.9 08.6 – 15.8 08.0 – 16.7 07.4 – 16.7 07.6 – 16.3 07.8 – 16.7 08.9 – 17.2

78 to 15 78 – 01 76 – 13 79 to 02 65 – 05 81 – 05 79 to 02 69 – 00 73 – 07 67 – 12 74 – 02 73 – 04 75 – 05 73 – 04 71 – 08 78 – 09 67 – 11 75 – 07 76 – 05 77 – 11 75 – 10 72 – 15 73 – 06 82 – 08 73 – 11 84 – 13 77 to 06 76 – 07 80 – 01 78 to 02 80 – 01 79 – 00 82 – 01 74 – 04 72 to 02 73 – 06 74 – 01 74 – 08 75 – 03 74 to 03

13.6 13.7 13.8 14.1 14.6 15.4 15.9 16.5 16.9 17.0 18.2 8.8 9.0 11.2 12.8 13.3 14.0 14.1 14.3 14.6 14.8 14.9 15.0 16.3 16.5 16.6 16.7 16.8 18.0 9.5 9.6 9.8 11.0 11.7 11.8 11.9 12.0 12.7 12.9 19.1

a

The LTR column shows the approximate coverage of the wind tracking in local time (decimal representation). The LAR column gives approximate latitudinal coverage (negative values denote southern latitudes). LTAN refers to the subspacecraft local time at ascending node.

small irregularities in coverage were compensated by the fact that the field of view included broad range of local times in distant imaging.

3. Wind Tracking Procedures 3.1. Description of the Employed Tracking Methods [17] We used both digital (automatic) and visual (manual) tracking of the UV cloud markings to derive the wind speeds. Three independent experimenters used imaging sequences from 39 orbits and applied two different visual tracking techniques and automatic wind tracking software. The first step in all cloud tracking methods was to establish navigation for the used images. We used the postprocessed SPICE data (see: http://naif.jpl.nasa.gov/naif/) to calculate VMC boresight vectors and their intersection with the cloud layer, assigning planetary coordinates to each image pixel. The cloud top height was set to a fixed value of 70 km above the planetary surface for navigation purposes. We expect the actual cloud top height to vary by about ±5 km

E00B31

around this level, which results in negligible uncertainties in coordinate determination. [18] For historical reasons and software availability we use two different procedures for visual wind speed measurements. The ‘‘sequential method’’ traces back to the wind speed measurements from previous missions, especially from Pioneer Venus OCPP observations [Limaye, 1988; Limaye et al., 1982, 1988]. As done for the OCPP wind speed measurements a loop of an entire image sequence from one VEX orbit was displayed in either equiangular (rectilinear, 0.2°/pixel) or polar projection onto a fixed coordinate grid (Figure 2 (right)). In this method motion of a cloud marking is observed in the way of a short movie clip, giving the experimenter an overview of the path of a selected cloud feature as well as the global pattern of motion. After familiarizing with the general flow pattern the experimenter puts small markers on distinctive cloud features in one image from the given orbit. Then the loop is continued, switching to a later image from the same sequence. There, another group of markers is put on the same cloud features. These steps are then repeated until the selected cloud features have changed significantly from their initial shapes and sizes and cannot be identified anymore. To enhance visibility of the cloud features a contrast enhancing high pass filter had been applied to the images after care was taken that the filtering did not alter any other image properties. This technique was used by two observers for about 40% (
8000) cloud vectors. [19] The ‘‘paired method’’ is a straightforward approach in which two different images from the same sequence are displayed side by side on a monitor and the cloud features in both are observed and compared simultaneously (Figure 2 (left and middle)). Also in this method, markers are set in order to track the corresponding points of cloud features in the compared image pairs. One significant difference is that here the images did not undergo any filtering, contrast enhancement or projection. Thus, basically unaltered VMC images were used for tracking with this method. The images were only slightly resized in order to achieve an approximately constant display size of the planet in each image. The paired method was used by one observer for about 60% (
12,000) of cloud vectors. [20] Although the basic principle, visual correlation between two images, is the same in both approaches, there are subtle but important differences. In the first case the observer tracks clouds by blink comparison between two images, trying to follow displacement of the tracked features on the fixed coordinate grid. In the second case the tracked feature is seen in both images simultaneously, thus emphasizing visual pattern recognition. [21] In both methods the experimenters tracked a great variety of feature types. Basically all features that did not span on a globally significant scale were used to track the cloud movements. We did not differentiate between ‘‘bright’’ and ‘‘dark’’ features, tracking both alike. The main criteria for feature selection were as follows: (1) Significant, well discernible contrast between the feature and its immediate surroundings. No fixed criterion was applied but the vast majority of tracked features had contrasts in excess of 15 % with respect to their surrounding. (2) Features size of 100– 300 km that allows for reliable identification of the feature center (only in equatorial regions). (3) Prominent

4 of 13

E00B31

MOISSL ET AL.: WIND TRACKING RESULTS FROM VMC UV IMAGES

E00B31

Figure 2. Typical VMC UV images used for tracking purposes. (left and middle) Pair of original VMC images used in the paired method. (right) Image in rectilinear projection used in sequential method. shapes like protrusions or clearly edged parts of boundaries for larger features. (4) Minimum time intervals between images were chosen to allow for resolving wind speed differences of about 10 – 20 m/s (depending on image resolution). (5) No features from the limb regions were used because they are subject to foreshortening. [22] The wind speeds in both methods were derived in similar ways. First the navigation data was linked to the images (for both, the projected and unprojected images). Then latitude and longitude of the respective marked pixels were extracted for all measurements of the sequence. From these values and the time intervals between images the zonal and meridional cloud speeds were calculated as U ðl; t; 8Þ ¼ DlRC =Dt cosð8Þ; V ð8; tÞ ¼ D8RC =Dt;

where U and V are zonal and meridional wind speed components, l and 8 are longitude and latitude, RC is the distance from the planetary center to 70 km above the surface (RC = 6121.8 km), and Dt is the time between images. [23] For the digital tracking in VMC images we used the same algorithm as applied by Rossow et al. [1990]. The analysis was performed systematically in pairs of images with at least one hour time difference using cross correlation as a metric with a ‘‘cloud’’ target of 10  10 pixels in the first image (2°  2° latitude by longitude). The displacement between the target location in the first image and the location of the maximum cross correlation coefficient in the second image between the search and the target was taken as the best estimate of the movement of the selected target over the interval between both images. The displacement over the two images leads to the determination of the zonal and meridional components of the cloud level flow. The search window in pixels was set to a window equivalent to a maximum flow of ±100 m/s in the meridional and 200 m/s in the zonal direction, accounting for the time interval between the images. Here the digital tracking results on
205,000 vectors obtained from images acquired on nine orbits (30, 31, 34, 38, 46, 51, 56, 60, and 61) are presented in Figure 3 to show the generally good agreement between the results from visual and digital tracking. Preliminary results from orbits 262– 267 were presented by Markiewicz et al. [2007b].

[24] Figure 2 shows examples of images used in the two different visual tracking methods. After the tracking sequences have been compiled from the orbital data set and markers have been set, the marker positions are determined and recorded. From the resulting set of vectors, latitudinal and/or longitudinal wind profiles and maps can be determined, by averaging all wind vectors inside a latitude bin. From the average density of retrieved vectors per orbit, we decided to use latitude bins of 5° and longitude bins of 7.5° (corresponding to 0.5 h of local solar time) in order to maintain good levels of statistics. 3.2. Comparison of Wind Tracking Measurements [25] As is obvious from Figure 2, the appearance of the cloud deck strongly differs between the original and the projected images. While the original images show all parts of the planet true to the perspective of the VMC, the rectilinear projection tends to distort the polar region quite drastically. Respectively, the polar projection distorts the equatorial region to a point where features get too distorted to be used for tracking. The basic properties of both methods will be addressed below. [26] Since Venus cloud features evolve over time, visual tracking relies on a somewhat subjective approach to identify a cloud feature in two or more images acquired at different times. In the case of VMC data, the orbital motion of the Venus Express spacecraft results in a slightly different ‘‘view’’ of a given cloud because of both its displacement due to ambient flow and due to the change in perspective. The paired method thus visually measures the combined displacement from perspective and ambient flow. And the part due to perspective change is accounted for by the image navigation implicitly. From the preserved perspective in the original images, the experimenter has good insight into which parts are seen heavily foreshortened, especially close to the limbs. This allows for a well founded possibility to select only the weakly or nonforeshortened features for tracking. On the other hand the apparent feature size and resolution change along the orbit. Therefore one has to be careful about reidentifying the same part of a feature in different images. This is especially the case in the lowcontrast areas at higher latitudes. There it is possible for faint feature boundaries to interfere with so-called ‘‘flat field

5 of 13

E00B31

MOISSL ET AL.: WIND TRACKING RESULTS FROM VMC UV IMAGES

E00B31

Figure 3. Latitude profiles of zonal winds derived from the paired (red line) tracking, sequential (blue line) visual tracking, and digital (black line) tracking. Error bars represent standard deviation due to uncertainties of individual measurements.

remainders’’ which result from the flat field correction process applied to VMC images. [27] In the sequential tracking method the flat field remainders become distinctively visible after high pass filtering, because they represent high-frequency variations. So they can be easily identified and avoided, since their apparent movement is governed by the change of the VMC perspective on the planet. Thus they mimic a movement completely different from all real atmospheric flows. [28] Figure 3 compares the latitude profiles, averaged in 5° latitude bins, of zonal wind speeds derived from the agglomerate data sets of the sequential (red curve) and paired (blue curve) visual tracking methods and those from digital tracking (black curve). The sequential method shows slightly higher speeds on average, which might be an indication for minimal bias due to either method. The biggest differences occur at high latitudes, between 70 and 80°S. Although the two methods do not actually agree there very well within their respective measurement accuracies, this is more likely due to the difficulties in finding adequate features for tracking in these latitudes than to differences in the tracking methods. Therefore, these deviations seem to originate from observer bias. Regarding the above mentioned risk of interference of faint, streaky cloud features with flat field remainders, some influence of this interference might possibly also play a role in this. Also the digital and the visual tracking results from the VMC data are in very good agreement up to about 40° latitude, and show a difference of
15 or 20 ms1 between 40 and 60° latitude from the visual tracking. There are likely two causes for this (1) the morphology and low contrast at these latitudes makes pattern matching by digital cross correlation difficult and (2) a selection effect in that

visually only the targets that are moving fast are discernible, but in longitude regions where the flow may be slower, no targets are seen as is suggested by the longitudinal distribution of the visual vectors. The digital results south of 60°S are not very reliable yet and are therefore to be viewed as entirely preliminary at this point. [29] Further, the average standard variation in all latitude bins is approximately the same for both visual tracking methods. This is a clear indication that measurement accuracy is nearly the same for both methods. [30] In conclusion, the deviations between the two visual methods are well on the order of the expected measurement uncertainty for single measurements in the according latitude regions. Despite the obvious discrepancy in the 70– 80°S latitude range between the two profiles from the two visual tracking methods, we decided that merging also these results will improve the poor statistics in this region and minimize observer bias. From this we conclude that the results from both methods are in reasonable agreement to allow for regarding all measured wind vectors as one data set, regardless of the employed method. In consequence later on in the paper we do not differentiate the results by tracking method. 3.3. Error Sources and Systematic Uncertainties [31] A number of measurement properties are influencing the accuracy of the wind speed measurements. Main sources of uncertainty are image resolution, measurement accuracy and feature evolution. Additionally some bias from the different methods and/or observers might introduce systematic errors. [32] Velocity uncertainty of an individual measurement due to technical limitations, such as image resolution and

6 of 13

E00B31

MOISSL ET AL.: WIND TRACKING RESULTS FROM VMC UV IMAGES

marker position error in the image is easily assessed via first-order Taylor expansion as follows: dU ¼

E00B31

nounced features and up to three pixels for fuzzy lowcontrast features in high latitudes.

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi pffiffiffi ð 2Rc da cosð8Þ=DtÞ2 þ ðdaRC sinð8Þ=DtÞ2 þ ðadRc cosð8Þ=DtÞ2 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffi dV ¼ ð 2Rc da=DtÞ2 þ ðadRc =DtÞ2 ;

where d values denominate total uncertainties for velocities (U,V), measured angular differences (a), and distance from Venus center to the cloud top (RC) and 8 is the latitude at measuring point. [33] For equatorial latitudes the first term, assuming a 1 pixel error (corresponding to approximately 0.2– 0.4° in latitude and longitude, depending on observing distance) and 40 min between images further away and 20 min when closing in, ranges from about 36 m/s at apocenter to roughly 10 m/s at 30,000 km distance from the planet, being 20 m/s on average. At higher latitudes the accuracy in latitudinal direction decreases further because of the convergence of latitude circles. This accuracy limit poses a great problem for deriving reliable meridional speeds, since they are supposed to be of the order of 10 m/s. Of course this source of uncertainty could be further suppressed through choice of image pairs with further increased time intervals. The early tracking sequences where too short to allow for reasonable statistics with image pairs separated by more than one hour, thus the chosen time intervals represent a trade-off between statistics and precision of the measurements. The second term, assuming 3 – 5 pixels between images and 10 km uncertainty in cloud deck height, remains well below 1 m/s in any case and can therefore be neglected. [34] Sources for uncertainty of an individual measurement could arise from the following: [35] 1. Pointing and mapping inaccuracies are considered to be negligible since both have proven to be extremely accurate and stable. A very conservative estimate would be an inaccuracy amounting to one pixel. [36] 2. Pixelation and noise are assumed to amount to an error of one pixel if physical feature boundaries lie close to the edge of a pixel. Depending on flat field quality, which is varying from orbit to orbit since images of nearly featureless cloud areas are needed, the uncertainty by noise can reach up to two pixels. [37] 3. Morphology and morphological evolution also influence the measurement accuracy. Since cloud features seldom have clear boundaries, there is always some possible (and variable) inaccuracy in determining the center or exact boundary line of a tracked feature. Furthermore, evolution of cloud features is observed, but because of time intervals between images being always smaller than 4 h, no significant inaccuracy should arise from them. This has been discussed already for the Pioneer Venus OCPP data set [Rossow et al., 1990]. These issues are especially valid for higher latitudes, since the low contrasts there make identification of the feature boundaries more difficult. It is difficult to quantify the influence of these morphological uncertainties, since they vary significantly with latitude and feature type. As a rule of thumb, the corresponding uncertainty varies between one pixel for small, sharply pro-

[38] 4. Systematic and random errors from measurement methods are likely to have some effect, as mentioned above. To minimize the effects of erroneous wind speeds from faulty measurements, we decided to reject all vectors outside the interval of 160– 0 m/s. The constraints for meridional wind speeds were 60– +60 m/s. These selection criteria lead to rejection of about 1.4% of all measured vectors. [39] As in previous works about cloud top winds inferred from the cloud feature tracking, we find that the variability in the derived and the averaged wind speed profiles are larger than the expected uncertainties from individual measurements. Therefore we chose to apply the standard deviation of wind speeds in one latitude bin as a measure of uncertainty for all wind speed profiles.

4. Results of the Wind Tracking 4.1. Zonal Wind Profiles [40] Figure 4 shows the averaged latitude profiles of zonal wind speed for the aggregate visual and digital tracking data sets, binned in latitude bins of 5° width. Also shown in Figure 4 is the profile obtained from the Visible and Infrared Thermal Imaging Spectrometer (VIRTIS) 380 nm channel data [Sa´nchez-Lavega et al., 2008] and the corresponding rotation period derived from the digital and visual profiles. UV winds derived from both experiments are in good agreement although VIRTIS tends to show systematically stronger winds by
10 m/s. At low latitudes, zonal wind speed is 85 – 90 m/s and almost constant with latitude. The latitude profile shows a gradual increase to 100 m/s, peaked at
45°S, indicating the presence of a weak midlatitude jet which is also seen in the inferences from the VIRTIS thermal wind analysis [Piccialli et al., 2008]. South of this latitude the wind speeds steadily decrease toward the pole. Although the profile is close to a solid body rotation curve at first glance, a closer look on the rotation period in Figure 4b reveals significant differences, where solid body rotation profiles would show up as horizontal lines. The kink at
10°N in the zonal wind profile is very likely an artifact, due to the rather poor sampling as it is at the very edge of VMC coverage of Venus in the tracking image sequences. [41] The error bars represent standard deviation of the zonal wind speed in each latitude bin. The deviation increases toward higher latitudes, mostly due to the changes in morphology and difficulties in finding well defined UV markings there, as discussed above. The standard deviations exceed the uncertainties of individual measurements orbit (compare to Figure 3), indicating a likely orbit to orbit variability of the latitudinal profiles of the zonal wind component. Also variability with local time is likely to be present because of influences from planetary-scale waves [Del Genio and Rossow, 1990].

7 of 13

E00B31

MOISSL ET AL.: WIND TRACKING RESULTS FROM VMC UV IMAGES

E00B31

Figure 4. (a) Average zonal wind speed profiles from VMC visual (solid line) and digital (dotted line) in comparison with data from VIRTIS (dashed line) [Sa´nchez-Lavega et al., 2008]. VMC and VIRTIS data is binned by 5° and 2°, respectively. The error bars represent standard deviation of the whole data set, including both measurement inaccuracies and orbit-by-orbit variations. (b) Corresponding rotation period profile from VMC wind tracking data (solid line is visual and dashed line is digital). [42] Figure 5 shows a selection of different single orbit profiles and the average profiles for each of the three observation periods. Standard deviations per latitude bin divided by the respective square root of the number of measurements are shown in the bottom Figures 5a and 5b, giving a measure of the standard error for the average value per bin. The mean zonal speeds can change by as much as
30 m/s from orbit to orbit and show significant variability in strength, position and even presence of the mid latitude wind speed maximum. Since these differences are well outside the standard error, they are most likely representing true variations of the zonal flow. These rather strong shortterm variations in the wind speed profiles leads to slightly different characteristics in the average profiles for each of the three observation periods. In periods 1 and 3, pro-

nounced wind speed maxima have been observed in the mid latitude regions, which is not the case for period 2. 4.2. Meridional Wind Component [43] Since the measured tracking sequences do not allow for individual measurements with accuracies better than ±10 m/s, all results on the meridional winds have to be considered preliminary. The standard deviation in each 5° latitude bin amounts to ±13 m/s. These are about as large as the uncertainties expected to arise from the chosen combination of image resolutions and time intervals. This indicates that measurement inaccuracies are likely larger than the real variability in meridional wind speeds. Better accuracy meridional component profiles are possible from extended mission data and will be published in future.

Figure 5. Latitude profiles of zonal wind for (a) individual orbits and (b) means for three tracking periods. The lines in the bottom of the images show average standard deviation divided by the square root of the number of measurements of the compared profiles for each latitude bin. 8 of 13

E00B31

MOISSL ET AL.: WIND TRACKING RESULTS FROM VMC UV IMAGES

E00B31

Figure 6. Average meridional wind profile. Error bars represent RMS deviation of measurements in one bin. [44] As one can see from Figure 6 the average meridional wind component ranges from 0 to 10 m/s, which is roughly consistent with previous results. Meridional speeds are slightly increasing from about 5 m/s at the equator to peak values of
10 m/s at mid latitudes between 40 and 55°S and then decreasing again toward the pole. The observed profile is in agreement with the assumption of a Hadley cell circulation between the equator and the mid latitudes [Schubert, 1983; Gierasch et al., 1997]. [45] Future tracking sequences acquired at shorter distances from Venus will help to improve on accuracy of the

meridional wind measurements by increasing image resolution which until now is not sufficient to satisfactory resolve wind speeds on the order of 0 – 10 m/s. We hope to be able to investigate temporal variability once higherresolution tracking sequences have been evaluated. 4.3. Dependence of the Zonal Component on Local Solar Time [46] Figure 7 shows the average zonal wind field as a function of latitude and local time after binning the wind speed vectors in ‘‘boxes’’ of 0.5 h local solar

Figure 7. Local solar time versus latitude field of the zonal wind in the Southern Hemisphere. 9 of 13

E00B31

MOISSL ET AL.: WIND TRACKING RESULTS FROM VMC UV IMAGES

E00B31

Figure 8. Local time profile of average zonal wind speeds between 0 and 20°S. VMC measurements in comparison with best fit for diurnal and semidiurnal tidal model (solid line) and diurnal model only (dashed line). time  5° ltitude each. The zonal wind field has a pronounced minimum at
11.5 h slightly upstream of the subsolar point and accelerates in the afternoon in low latitudes. In the middle latitudes velocities are higher on average over the whole observed local time range and show less local time variability. In higher latitudes an increase in speed with local time is apparent down to 60°S. Toward the pole zonal speeds are decreasing rapidly. [47] The data displayed in the Figure 7 has been divided into three latitude bands (0 – 20°S, 20 –40°S, and 40– 60°S) in order to evaluate tidal components in the zonal wind field

in the different latitude regions (Figures 8, 9, and 10). In each of the latitude bands we compared the local time profiles to a tidal model that includes both diurnal and semidiurnal components U ðlÞ ¼ U0 þ U1 sinðl þ F1 Þ þ U2 sinð2l þ F2 Þ;

where U0 is the zonal mean value, l is the longitude, U1 and U2 are the diurnal and semidiurnal velocity amplitudes, and F1 and F2 are the diurnal and semidiurnal phase angles.

Figure 9. Local time profile of average zonal wind speeds between 20 and 40°S. VMC measurements in comparison with best fit for diurnal and semidiurnal tidal model (solid line) and diurnal model only (dashed line). 10 of 13

E00B31

MOISSL ET AL.: WIND TRACKING RESULTS FROM VMC UV IMAGES

E00B31

Figure 10. Local time profile of average zonal wind speeds between 40 and 60°S. The dashed line indicates the general trend with local time.

[48] In Figure 8, the local-time-dependent zonal wind component in the latitude strip between 0 and 20°S is being compared to fits from the diurnal-only part of the model and both diurnal and semidiurnal. Both fit curves show considerable agreement with the observed wind speeds, with higher agreement when both tidal wave modes are employed. Figure 9 shows the same comparison for the 20– 40°S latitude strip. Here the data is in about equal agreement with both model curves. For the high-latitude region no conclusive correlation for diurnal and semidiurnal wave models could be obtained. The data shows, however, an overall trend toward increased wind speeds in the afternoon hours. [49] In summary the observed local time variations of the zonal wind component is indicating the presence of solar thermal tides in VMC observations. Since only about 8 h of local time are covered from VMC measurements no reliable coefficients can be obtained from the employed model at this moment. We are continuing our efforts to expand the local time coverage of our tracking results to regions closer to the terminator regions.

5. Discussion and Conclusions [50] Figure 11 shows a comparison between zonal wind speed profiles from VMC with previous observations from Pioneer Venus and Galileo. Although maximum resolutions of Mariner 10 Images lie within the same range as VMC data, lack of sufficient overlap with required time interval prevented measurements from the high-resolution images. Therefore we did not include results from Mariner 10 wind tracking in this comparison. [51] The Pioneer Venus profile represents an average of the results from 1980 and 1982 imaging seasons of the OCPP polarimeter [Limaye, 2007] which is in good agreement with the VMC profile. The general shape of the Pioneer Venus orbit was quite similar to the one of Venus

Express, with a pericenter at about 15°N the planet was observed mainly from a near-equatorial perspective. In contrast to the Pioneer Venus orbit, the 80°N pericenter of Venus Express allows VMC to deliver improved coverage of the Southern Hemisphere in nadir geometry. The average resolution of the OCPP images used for cloud tracking also lies around 30 km/pixel. One significant limitation of the OCPP image data is the 4 h time interval between images [Rossow et al., 1980, 1990]. Whereas the instantaneous image acquisition of VMC allows for arbitrary intervals between images. [52] Also the results from the high-resolution cloud tracking in Galileo images are in good general agreement with our findings. Prior to the VMC observations the Galileo images comprised the best resolved (down to 15 km/pixel) sequences used for cloud tracking and have recently been reused for high-resolution wind tracking [Toigo et al., 1994; Peralta et al., 2007]. Time intervals between Galileo images range from 15 min to 2 h, thus having in principle the same temporal resolution as the VMC sequences. [53] So far, dynamical observations of the Venus clouds have delivered either only highly resolved short-term or broadly averaged long-term results. With VMC it finally became possible to investigate the global dynamics with a relatively high resolution in space and time on a long-term basis. Our findings from manual wind tracking in VMC UV image sequences so far show that the details of the cloud top level dynamics of Venus appear to be highly variable, down to a timescale of days. On the other side the average general rotation of the atmosphere remained effectively stable since Mariner 10, especially with regard to the remarkable agreement between VMC measurements and results from previous missions. [54] In some cases significant variations in the zonal wind properties were observed especially in the mid latitude regions on an orbit to orbit basis, indicating possible

11 of 13

E00B31

MOISSL ET AL.: WIND TRACKING RESULTS FROM VMC UV IMAGES

E00B31

Figure 11. (a) Comparison of zonal wind speed profiles from VMC, Pioneer Venus, and Galileo measurements. (b) Corresponding rotation periods. influences from global-scale wave modes. In other cases we measured more constant conditions over one atmospheric revolution or even longer periods, at cloud top level. These short-term variations between results from individual orbit tracking sequences appear to have an impact on the average zonal wind component profiles over longer-term observation periods. So far the search for Kelvin gravity wave modes in the consecutive measurements from orbits 281 – 284 and 460– 463 did not yield any conclusive results. Most likely longer sequences of coherent tracking data will be required for clear identification of these global wave modes in VMC data. Also it remains an interesting question whether the irregularly detected mid latitude jets are indeed variable with time or simply become, in parts or in total, shielded from view by higher H2SO4 haze layers for varying time intervals. [55] Winds at latitudes higher than 60°S are still difficult to track because of low contrast and scarcity of features, but increasing data is being collected. Over all, it was possible to extend latitudinal coverage of the cloud top winds with VMC observations. [56] A first analysis indicated that diurnal and semidiurnal thermal tide modes appear to be present in the data, but a more profound confirmation still depends on synthesis of VIRTIS and VMC observations on night and dayside in order to bridge the large data gap in local time coverage of the VMC UV data between 17 and 9 h local solar time. [57] Although poorly resolved, results on average meridional wind speed components are in general agreement with previous observations and with the presence of a Hadley cell spanning between equatorial region and about 45°S latitude. Since measurement accuracy is very low so far, so we decided not to use the results on meridional winds for detailed investigations. We expect future data to improve the accuracy up to a degree that allows for more reliable meridional profiles fit for being used to draw more conclusions. [58] As VMC continues to acquire data, not only more UV data will become available for dayside cloud top level tracking purposes. In addition we are investigating whether data from the two IR channels of the instrument could be used for tracking lower-level features on the day side and

extend cloud tracking to the night side. Although contrasts range only from
1 – 5% for the IR features in day side images, it might be feasible to improve contrasts through carefully applied filtering processes. Tracking on the night side would prove even more difficult, since VMC is registering signals from the surface simultaneously with those from the lower cloud levels. Efforts for both tracking possibilities are in progress at the moment and will be presented in future works. [59] Meanwhile operations are increasingly adapted and custom fitted for wind tracking in the UV channel, improving steadily on sequence length and also on timing between individual images. Furthermore wind tracking sequences are extended down to lower altitudes above the planet to allow for higher resolutions and more accurate tracking. One method to achieve this goal is to follow cloud features to compensate for the small area covered by pericenter observations. We hope to achieve tracking sequences with resolutions as high as about 7 km/pixel. [60] Acknowledgments. We would like to thank A. Sanchez-Lavega and R. Hueso for their valuable cooperation and making their latest wind tracking results available to comparison for us. Thanks also go to the International Max Planck Research School on Physical Processes in the Solar System and beyond, the rest of the VMC team for making this work possible, and to NASA for contributions to this paper from grants NNG06GC68G and NNX07AF27G.

References Belton, M. J. S., P. J. Gierasch, M. D. Smith, P. Helfenstein, P. J. Schinder, J. B. Pollack, K. Rages, D. Morrison, K. P. Klaasen, and C. B. Pilcher (1991), Images from Galileo of the Venus cloud deck, Science, 253, 1531 – 1536, doi:10.1126/science.253.5027.1531. Boyer, C., and P. Guerin (1969), E´tude de la rotation re´trograde, en 4 jours, de la couche exte´rieure nuageuse de Ve´nus, Icarus, 11, 338 – 355, doi:10.1016/0019-1035(69)90067-0. Carlson, R. W., et al. (1991), Galileo infrared imaging spectroscopy measurements at Venus, Science, 253, 1541 – 1548, doi:10.1126/science.253. 5027.1541. Crisp, D., S. McMuldroch, S. K. Stephens, W. M. Sinton, B. Ragent, K.-H. Hodapp, R. G. Probst, L. R. Doyle, D. A. Allen, and J. Elias (1991), Ground-based near-infrared imaging observations of Venus during the Galileo encounter, Science, 253, 1538 – 1541, doi:10.1126/science.253. 5027.1538. Del Genio, A. D., and W. B. Rossow (1990), Planetary-scale waves and the cyclic nature of cloud top dynamics on Venus, J. Atmos. Sci., 47, 293 – 318, doi:10.1175/1520-0469(1990)047<0293:PSWATC>2.0.CO;2.

12 of 13

E00B31

MOISSL ET AL.: WIND TRACKING RESULTS FROM VMC UV IMAGES

Gierasch, P., et al. (1997), The general circulation of the Venus atmosphere: An assessment, in Venus II, edited by Stephen W. Bougher, D. M. Hunten, and R. J. Philips, pp. 459 – 500, Univ. of Ariz. Press, Tucson, Ariz. Kawabata, K., D. L. Coffeen, J. E. Hansen, W. A. Lane, M. Sato, and L. D. Travis (1980), Cloud and haze properties from Pioneer Venus polarimetry, J. Geophys. Res., 85, 8129 – 8140, doi:10.1029/JA085iA13p08129. Limaye, S. S. (1985), Venus atmospheric circulation: Observations and implications of the thermal structure, Adv. Space Res., 5(9), 51 – 62, doi:10.1016/0273-1177(85)90270-4. Limaye, S. S. (1988), Venus: Cloud level circulation during 1982 as determined from the Pioneer Cloud Photopolarimeter images. Part II. Solar longitude dependent circulation, Icarus, 73, 212 – 226, doi:10.1016/00191035(88)90094-2. Limaye, S. S. (2007), Venus atmospheric circulation: Known and unknown, J. Geophys. Res., 112, E04S09, doi:10.1029/2006JE002814. Limaye, S. S., and V. E. Suomi (1981), Cloud motions on Venus: Global structure and organization, J. Atmos. Sci., 38, 1220 – 1235, doi:10.1175/ 1520-0469(1981)038<1220:CMOVGS>2.0.CO;2. Limaye, S. S., C. J. Grund, and S. P. Burre (1982), Zonal mean circulation at the cloud level of Venus: Spring and fall 1979 OCPP observations, Icarus, 51, 416 – 439, doi:10.1016/0019-1035(82)90092-6. Limaye, S. S., C. Grassotti, and M. J. Kuetemeyer (1988), Venus: Cloud level circulation during 1982 as determined from the Pioneer Cloud Photopolarimeter images. Part I. Time and zonally averaged circulation, Icarus, 73, 193 – 211. Markiewicz, W. J., et al. (2007a), Venus Monitoring Camera for Venus Express, Planet. Space Sci., 55(12), 1701 – 1711, doi:10.1016/j.pss. 2007.01.004. Markiewicz, W. J., D. V. Titov, S. S. Limaye, H. U. Keller, N. Ignatiev, R. Jaumann, N. Thomas, H. Michalik, R. Moissl, and P. Russo (2007b), Morphology and dynamics of the upper cloud layer of Venus, Nature, 450, 633 – 636, doi:10.1038/nature06320. Peralta, J., R. Hueso, and A. Sanchez-Lavega (2007), A reanalysis of Venus winds at two cloud levels from Galileo SSI images, Icarus, 190, 469 – 477, doi:10.1016/j.icarus.2007.03.028. Piccialli, A., D. V. Titov, D. Grassi, I. A. Khatuntsev, P. Drossart, G. Piccioni, and A. Migliorini (2008), Cyclostrophic winds from the VIRTIS temperature sounding: A preliminary analysis, J. Geophys. Res., 113, E00B11, doi:10.1029/2008JE003127. Rossow, W. B., A. D. Del Genio, S. S. Limaye, L. D. Travis, and P. H. Stone (1980), Cloud morphology and motions from Pioneer-Venus images, J. Geophys. Res., 85, 8107 – 8128, doi:10.1029/ JA085iA13p08107.

E00B31

Rossow, W. B., A. D. Del Genio, and T. Eichler (1990), Cloud-tracked winds from Pioneer Venus OCPP images, J. Atmos. Sci., 47, 2053 – 2084, doi:10.1175/1520-0469(1990)047<2053:CTWFVO>2.0.CO;2. Sa´nchez-Lavega, A., et al. (2008), Variable winds on Venus mapped in three dimensions, Geophys. Res. Lett., 35, L13204, doi:10.1029/ 2008GL033817. Schubert, G. (1983), General circulation and the dynamical state of the Venus atmosphere, in Venus, edited by D. Hunten, L. Colin, T. Donahue, and V. Moroz, pp. 651 – 765, Univ. of Ariz. Press, Tucson, Ariz. Schubert, G., S. W. Bougher, C. C. Covey, A. D. Del Genio, A. S. Grossman, J. L. Hollingsworth, S. S. Limaye, and R. E. Young (2007), Venus atmosphere dynamics: A continuing enigma, in Exploring Venus as terrestrial planet, Geophys. Monogr. Ser., vol. 176, edited by L. W. Esposito, E. R. Stofan, and T. E. Cravens, pp. 121 – 138, AGU, Washington, D. C. Smith, M. D., and P. J. Gierasch (1996), Global-scale winds at the Venus cloud top inferred from cloud streak orientations, Icarus, 123(2), 313 – 323, doi:10.1006/icar.1996.0160. Suomi, V. E. (1975), Cloud motions on Venus, in The Atmosphere of Venus, edited by J. E. Hansen, 42 pp., NASA, Washington, D. C. Svedhem, H., et al. (2007), Venus Express—The first European mission to Venus, Planet. Space Sci., 55(12), 1636 – 1652, doi:10.1016/j.pss. 2007.01.013. Titov, D. V., et al. (2006), Venus Express science planning, Planet. Space Sci., 54(13 – 14), 1279 – 1297, doi:10.1016/j.pss.2006.04.017. Toigo, A., P. J. Gierasch, and M. D. Smith (1994), High resolution cloud feature tracking on Venus by Galileo, Icarus, 109(2), 318 – 336, doi:10.1006/icar.1994.1097. Tomasko, M. G., L. R. Doose, and P. H. Smith (1985), The absorption of solar energy and the heating rate in the atmosphere of Venus, Adv. Space Res., 5(9), 71 – 79, doi:10.1016/0273-1177(85)90272-8. 

M. Almeida, European Space Astronomy Centre, European Space Agency, P.O. Box - Apartado de Correos 78, E-28691 Madrid, Spain. T. Behnke, R. Jaumann, K.-D. Matz, and T. Roatsch, Institut fur Planetenforschung, Deutsches Zentrum fur Luft- und Raumfahrt, Rutherfordstrasse 2, D-12489 Berlin, Germany. S. F. Hviid, W. J. Markiewicz, R. Moissl, G. Portyankina, and D. V. Titov, Max-Planck-Institut fuer Sonnensystemforschung, Max-PlanckStrasse 2, D-37191 Katlenburg-Lindau, Germany. ([email protected]) N. I. Ignatiev and I. Khatuntsev, Space Research Institute, 84/32 Profsoyuznaya Street, Moscow 117997, Russia. S. S. Limaye, Space Science and Engineering Center, University of Wisconsin-Madison, 1017 Atmospheric Oceanic and Space Sciences Building, 1225 West Dayton Street, Madison, WI 53706, USA.

13 of 13

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, E00B11, doi:10.1029/2008JE003127, 2008

for

Full Article

Cyclostrophic winds from the Visible and Infrared Thermal Imaging Spectrometer temperature sounding: A preliminary analysis A. Piccialli,1 D. V. Titov,1,2 D. Grassi,3 I. Khatuntsev,2 P. Drossart,4 G. Piccioni,5 and A. Migliorini5 Received 25 February 2008; accepted 25 July 2008; published 22 October 2008.

[1] Visible and Infrared Thermal Imaging Spectrometer (VIRTIS) onboard the Venus

Express spacecraft has been operating since April 2006 providing new observations of the temperature structure of Venus mesosphere (60–95 km). Zonal winds in the middle atmosphere of Venus have been retrieved from VIRTIS temperature profiles using the cyclostrophic approximation. The wind field is characterized by three main features: (1) a midlatitude jet connected to the thermal feature known as the cold collar; the jet reaches a maximum speed of 80–90 ± 10 m/s near the cloud top (
70 km altitude) at 50°S latitude; (2) a strong decrease of wind from 60°S toward the pole reaching zero velocity at
70°S; and (3) the decrease of the wind above the jet with increasing altitude. Local time dependence of the temperature and wind field has been analyzed. Temperatures show at cloud tops a cooling of
15 K during the night which propagates also on the zonal wind field. Comparison with cloud-tracked winds from Venus Monitoring Camera images indicates a first-order agreement. Citation: Piccialli, A., D. V. Titov, D. Grassi, I. Khatuntsev, P. Drossart, G. Piccioni, and A. Migliorini (2008), Cyclostrophic winds from the Visible and Infrared Thermal Imaging Spectrometer temperature sounding: A preliminary analysis, J. Geophys. Res., 113, E00B11, doi:10.1029/2008JE003127.

1. Introduction [2] Early observations showed that the Venus atmosphere has two different dynamic regimes. In the thermosphere above
100 km diurnal temperature gradients force global solar-antisolar circulation [Bougher et al., 1997; Lellouch et al., 1997]. Troposphere and mesosphere (0 – 100 km) are involved in the retrograde almost purely zonal motion with maximum wind speed of about 100 m/s at the cloud tops, called superrotation. The physical mechanisms that maintain this global regular motion are still not known [Schubert, 1983; Gierasch et al., 1997; Schubert et al., 2007]. Leovy [1973] was first to notice that on a slowly rotating planet, where Coriolis force is negligible, the global zonal circulation is a result of local balance of pressure gradient and centripetal force which is called cyclostrophic balance. The thermal wind equation that describes this balance relates vertical shear of zonal wind to latitudinal temperature gradient on a constant pressure surface [Leovy, 1973; Schubert, 1983]. Although the cyclostrophic balance seems to clearly describe the observed state of Venus superrotation, it does not explain what originally brought the atmo-

1 Max Planck Institute for Solar System Research, Katlenburg-Lindau, Germany. 2 Space Research Institute, Moscow, Russia. 3 Istituto di Fisica dello Spazio Interplanetario, INAF, Rome, Italy. 4 LESIA, Observatoire de Paris, Meudon, France. 5 Istituto di Astrofisica Spaziale e Fisica Cosmica, INAF, Rome, Italy.

Copyright 2008 by the American Geophysical Union. 0148-0227/08/2008JE003127$09.00

sphere to this state or which mechanisms maintain the vertical wind shear. [3] Temperature structure of the Venus mesosphere was sounded remotely by radio occultation and thermal emission radiometry and spectroscopy on the Pioneer Venus and Venera-15 orbiters [Seiff, 1983; Taylor et al., 1983; Seiff et al., 1985; Lellouch et al., 1997; Zasova et al., 2007]. The mesospheric temperature field showed two remarkable peculiarities. Temperatures above
75 km monotonically increase from equator to pole thus creating positive latitudinal temperature gradient at constant pressure level. Inside the upper cloud layer (60 –75 km) the trend reverses and temperature field forms a bulge of cold air at 60° – 80° latitude which was called ‘‘cold collar.’’ [4] The measured temperatures were used to derive zonal wind field using the cyclostrophic approximation [Chub and Yakovlev, 1980; Seiff, 1983; Newman et al., 1984; Limaye, 1985; Roos-Serote et al., 1995; Zasova et al., 2007]. These studies constrained the winds in the poorly explored region of transition to the solar-antisolar thermospheric circulation. The main features of the cyclostrophic wind field are (1) the presence of strong midlatitude jet centered at
50° latitude and
70 km altitude and (2) the decrease of zonal wind with altitude above the cloud top. This behavior is governed by the latitudinal temperature contrasts. In the midlatitudes the atmosphere is accelerated at the cloud level by negative latitudinal gradient of temperature associated with the cold collar. Above the cloud tops on the contrary positive temperature gradient forces the winds to fade out with altitude. The cyclostrophic balance was found to break above 70– 75 km in high (>70°) and low

E00B11

1 of 9

E00B11

PICCIALLI ET AL.: CYCLOSTROPHIC WINDS IN VENUS MESOSPHERE

E00B11

Figure 1. Contours of temperature field averaged over the range 2000 – 2200 LT for the VIRTIS-M orbit VI0072. Contour interval is 5 K. In gray the cloud layer is represented schematically. (<40°) latitudes. Here the latitudinal temperature gradient cannot provide centripetal force for the air parcels involved in the zonal motion. The magnitude of the midlatitude jet changes from
160 m/s in the work by Newman et al. [1984] to
100 m/s in the work by Zasova et al. [2007]. Position of the jet core showed meandering in the latitude range of 45° – 65° and altitudes of 65– 70 km. Zasova et al. [2007] also reported about diurnal and semidiurnal harmonics present in the thermal wind pattern derived from Venera15 data. [5] Since April 2006, Venus Express carries out systematic remote sensing observations of the Venus mesosphere using both radio occultation and thermal emission spectroscopy techniques [Svedhem et al., 2007; Ha¨usler et al., 2006; Drossart et al., 2007]. Polar orbit of the satellite allows the experiments to achieve full latitude coverage and especially focus of the southern hemisphere poorly studied in the earlier missions [Titov et al., 2006]. Here we present the first results of calculations of the cyclostrophic wind field from the temperature sounding by VIRTIS experiment [Grassi et al., 2008].

2. Temperature Field in the Mesosphere [6] Visible and Infrared Thermal Imaging Spectrometer (VIRTIS) is one of the experiments on board of Venus Express. It consists of two channels. VIRTIS-M is an imaging spectrometer with moderate spectral resolution (R
200) and high spatial resolution (0.25 mrad) operating in the spectral ranges 0.25 – 1 mm and 1 –5 mm. VIRTIS-H is a high-resolution spectrometer (R
1200) operating in the spectral range 1.84– 4.99 mm [Drossart et al., 2007; Piccioni et al., 2008]. VIRTIS-M acquires thermal emission spectra in the region of 4.3 mm CO2 band. These measurements were then used to retrieve vertical profiles of air temperature at 67 pressure levels in the Venusian mesosphere. The VIRTIS temperature sounding covers altitude range from 85 to 65 km (1 – 100 mbar). The overall error in retrieved temperature is<4 K in the range 100 –0.1 mbar and

is better than 1 K between 70 and 7 mbar. The main sources of errors and systematic uncertainties in the temperature retrievals are instrumental noise in the spectral ranges sounding high altitudes and uncertainties in the aerosols densities within the cloud deck (70 – 60 km) [Grassi et al., 2008]. The VIRTIS observations analyzed in this work cover the night side in the southern hemisphere. The measurements on the day side have nonnegligible solar component and were excluded from our study. [7] Figures 1 and 2 show example of the latitude-altitude temperature field retrieved from the VIRTIS-M observations in orbit 72 and corresponding latitude dependence of temperature at constant pressure levels. We used Chebyshev polynomials of degree 4 to fit the data points and to evaluate the latitudinal temperature gradient. Above 75 km (
20 mbar) temperatures on isobaric surfaces generally increase toward the pole. Below this level the temperature pattern becomes more complex and at 50°– 70° latitude shows the region of temperature inversions at the cloud top (cold collar). In Figure 2 the cold collar is clearly visible as dip between 50° and 70° latitude at 40 and 90 mbar. Both positive temperature gradient in the higher mesosphere and cold collar right at the cloud tops were observed by Pioneer Venus [Taylor et al., 1983; Newman et al., 1984] and Venera-15 [Zasova et al., 2007] missions in the northern hemisphere. VIRTIS observations in the south strongly suggest global hemispheric symmetry of the temperature field.

3. Zonal Thermal Winds [8] Following Newman et al. [1984] we used cyclostrophic approximation to derive zonal wind pattern in the mesosphere from the VIRTIS temperature field. If we indicate with u the zonal wind speed in m/s, with z =  ln (p/pref) the logarithmic pressure vertical coordinate, with R the gas constant and with (@T/@f)p = const the latitudinal temperature gradient, it is possible to describe the zonal wind u by the thermal wind equation [Schubert, 1983]

2 of 9

E00B11

PICCIALLI ET AL.: CYCLOSTROPHIC WINDS IN VENUS MESOSPHERE

E00B11

Figure 2. Plots of temperature at different pressure levels for 2000 –2200 LT. Chebyshev polynomial approximation is represented in gray. Error bar shows random error of air temperature retrieval. Data refer to the VIRTIS-M orbit VI0072. @u R @T 2u ¼ @z tan f @f p¼const

ð1Þ

The retrieved air temperatures T(f) were approximated by Chebyshev polynomials of degree 4 between 25° and 75° of latitude at 49 pressure levels. The lower boundary condition needed to solve differential equation (1) was

Figure 3. Functions used as lower boundary condition. Curve 1 is the function described by equation (2) used for the nominal case. Curve 2 is the solid body rotation function u0 = 90 cos f; as can be seen, it perfectly fits the curve 1 for latitudes higher than 70°, while at the equator it results 10 m/s stronger. Curve 3 represents the cloud-tracked wind derived from Galileo SSI NIR images [Peralta et al., 2007]. This curve has the same magnitude of curve 1 near the equator, but a strong decrease in the wind speed is seen at latitudes higher than 40°. 3 of 9

E00B11

PICCIALLI ET AL.: CYCLOSTROPHIC WINDS IN VENUS MESOSPHERE

E00B11

Figure 4. Contours of the difference absolute value (m/s) between the zonal thermal wind speed derived from VIRTIS temperature retrievals assuming u0 = (45 sec h((f  56)/9) + 75) cos f as lower boundary condition and the zonal thermal wind derived assuming (a) u0 = 90 cosf and (b) cloud-tracked winds derived from Galileo SSI NIR images as lower boundary condition. Data refer to the VIRTIS-M orbit VI0072. taken at the reference pressure level pref = 275 mb (
58 km) using the equation adopted by Counselman et al. [1980]; coefficients were selected to fit the Venus

Monitoring Camera (VMC) direct measurements of wind profile and were provided by I. Khatuntsev (personal communications, 2008):

Figure 5. Latitude dependence of temperatures at a pressure level of 90 mb for 2000 – 2200 LT. Chebyshev polynomial approximation is represented in black together with the ±1s, ±2s curves represented in gray. Error bar shows random error of air temperature retrieval. Data refer to the VIRTISM orbit VI0072. 4 of 9

E00B11

PICCIALLI ET AL.: CYCLOSTROPHIC WINDS IN VENUS MESOSPHERE

E00B11

Figure 6. Contours of the zonal thermal wind speed (m/s) for 2000– 2200 LT derived from VIRTIS temperature retrievals assuming cyclostrophic balance. Contour interval is 5 m/s. Data refer to the VIRTIS-M orbit VI0072. In gray the cloud layer is represented schematically.

U0 ðfÞ ¼ ½45 sech ððf  56Þ=9Þ þ 75 cos f

ð2Þ

In order to test our retrieval code we used temperature field retrieved from Venera-15 Fourier Spectrometer (FS)

data [Zasova and Khatuntsev, 1997]. Our test calculations were in good agreement with original results. [9] We studied sensitivity of calculated thermal winds to numerical parameters and lower boundary conditions. First, we found that the Chebyshev polynomial of degree 4

Figure 7. Plots of latitude dependence of temperatures at selected pressure levels; different colors correspond to different local time: (blue) 1800– 2000 LT; (green) 2000– 2200 LT; (yellow) 2200 –2400 LT; (red) 0000 –0200 LT; (light blue) 0200 – 0400 LT; and (orange) 0400– 0600 LT. Data refer to the VIRTIS-M orbit VI0072. 5 of 9

E00B11

PICCIALLI ET AL.: CYCLOSTROPHIC WINDS IN VENUS MESOSPHERE

E00B11

Figure 8. Contours of the zonal thermal wind speed (m/s) for (a) 1800 – 2000 LT and (b) 0300 –0500 LT. Winds have been derived from VIRTIS temperature retrievals assuming cyclostrophic balance. Contour interval is 5 m/s. Data refer to 13 VIRTIS orbits acquired between May and December 2006. provided the best fit to the data (Figure 2). Higher power of polynomial did not improve the data fit and resulted in highfrequency oscillations. Second, we considered different functions as lower boundary condition (Figure 3). [10] We calculated the difference between the zonal wind field derived from VIRTIS temperature profiles using curve 1 as lower boundary condition and that obtained using curves 2 and 3 (Figures 4a and 4b). Both cases show small discrepancy with the nominal case (equation (2)) confirming that the retrieved wind weakly depends on the lower boundary condition. In Figure 4a the difference reaches maximum value of 40 m/s because of the absence in curve 2 of the midlatitude bulge present in curve 1 that forces winds to slow down with altitude faster than in the nominal case. In Figure 4b a bigger discrepancy is observed caused by the strong wind decrease at midlatitudes seen in Galileo NIR images. Both test cases show that lower boundary condition does not affect the region of the jet. [11] Errors in the temperature retrievals are the source of uncertainty on the derived wind speed. We followed the approach used by Newman et al. [1984] to assess the propagation of temperature retrieval error on the wind field.

On every pressure level air temperatures were averaged within 5° latitude bins. Mean value and standard deviation of each data point from approximating Chebyshev polynomial were calculated and centered in the interval. Another Chebyshev polynomial was used to fit the standard deviations and was added to the original polynomial approximation curve to produce the +1s curve. The  1s and ±2s curves were calculated in a similar way. Figure 5 shows that almost all the data points are within ±2s range. Since scattering of the retrieved temperatures depends on latitude the gradients of ±1s, ± 2s curves differ from that of the original approximation curve that results in distortions of the wind field. [12] Wind speeds were retrieved from ±1s, ±2s curves, using equation (1) and the boundary conditions from equation (2) to assess the error in wind velocity determination. For all the curves, the position of the jet is not changed within 2° latitude. The wind field calculated for 1s curve has a midlatitude jet speed of 89 m/s at 51° latitude. The speed of the jet for the 2s curve is slightly altered. On the other hand for the  1s and  2s curves the speed of the jet reaches a value of 98 and 102 m/s

6 of 9

E00B11

PICCIALLI ET AL.: CYCLOSTROPHIC WINDS IN VENUS MESOSPHERE

E00B11

Figure 9. (a) Latitude of midlatitude jet versus local time for thirteen VIRTIS orbits acquired between May and December 2006; (b) Altitude of midlatitude jet versus local time; (c) Wind speed of midlatitude jet versus local time.

respectively. As an esteem of retrieval error on the midlatitude jet speed, it seems reasonable to take a value of ± 10 m/s which is comparable to the uncertainties on cloud-tracked winds derived from the Venus Monitoring Camera (R. Moissl et al., Cloud top winds from tracking UV features in VMC images, submitted to Journal of Geophysical Research, 2008). [13] Figure 6 shows an example of thermal wind for 2000 – 2200 local time (LT) derived from the VIRTIS-M observations in orbit 72. The main feature in the plot is the midlatitude jet centered at 50° latitude and 67 km of altitude with a maximum speed of 90 m/s. Comparison of the wind field (Figure 6) with the temperature field (Figure 1) shows that the midlatitude jet is related to the cold collar. [14] The thermal feature known as the warm polar mesosphere, characterized by an increase of temperature toward the pole on isobaric surfaces between 75 and 90 km of altitude, forces the retrieved zonal wind to decrease to zero very fast at higher latitudes. It is important to note, however, that for latitudes lower than 30° and higher than 75° the cyclostrophic balance fails, thus wind retrievals at these latitudes should not be taken in account.

[15] The VIRTIS observations in orbit 72 completely covered the night side of the southern hemisphere that allowed us to estimate the dependence of temperature field on local time. Figure 7 shows latitude profiles of temperatures at selected pressure levels for 2-h local time bins. They suggest that the atmosphere at the cloud tops cools down by about 15 K during the night. Theses changes in temperature structure also affect the thermal wind field. [16] An example of the effect of diurnal variations of temperature structure is shown in Figures 8a and 8b. Wind field has been retrieved for two different local times combining the temperature profiles of thirteen VIRTIS orbits acquired between May and December 2006. [17] Last, we examined in detail the dependence of the midlatitude jet on local time. Temperature profiles have been divided in local time intervals of 3 h and the wind speed has been retrieved for each local time bin. The behavior of midlatitude jet position and speed with local time has been analyzed in Figures 9a – 9c. A change in the position and speed of the midlatitude jet can be clearly observed, however the magnitude of the variations lays within the uncertainties on retrieved wind speeds. There-

7 of 9

E00B11

PICCIALLI ET AL.: CYCLOSTROPHIC WINDS IN VENUS MESOSPHERE

fore, any conclusion on local time dependence should be taken cautiously.

4. Discussion and Conclusions [18] We presented the preliminary results of the cyclostrophic wind calculations from the VIRTIS-M temperature sounding in the southern hemisphere on the night side [Grassi et al., 2008]. The main features of the wind field are (1) the presence of the midlatitude jet with peak velocity of 80 –90 m/s centered at
50°S at the cloud tops (
70 km); (2) fast decrease of zonal winds poleward from 60°S with zero velocity reached at
70°S; and (3) gradual decrease of thermal wind with altitude above the jet. Our results are in general agreement with those based on the earlier observations [Newman et al., 1984; Zasova et al., 2007]. These features correlate with the behavior of the temperature field (Figure 1). According to the thermal wind equation (1) the negative latitude gradient within the upper cloud (<70 km) associated with the cold collar accelerates the wind. This trend changes with reversal of the temperature gradient above the cloud that causes thermal winds to fade out toward the top of the mesosphere. This deceleration is the strongest in high latitudes (>65°S) that results in that cyclostrophic balance breaks down in polar mesosphere (Figure 6). [19] The magnitude of the jet in our results (80 – 90 m/s) is smaller than that reported by Zasova et al. [2007] (90 – 110 m/s) and especially by Newman et al. [1984] (140 – 160 m/s). This discrepancy can be partially attributed to the differences in numerical schemes and boundary conditions used in calculations, but seems to mainly result from peculiarities of temperature sounding techniques. Radio occultation data used by Newman et al. [1984] provide temperature structure with vertical resolution of few hundred meters that allows the measurements to completely resolve deep temperature inversions typical for the cold collar regions. Vertical resolution in thermal emission spectroscopy in the CO2 bands used by Zasova et al. [2007] and in this work does not exceed few kilometers. This smoothes temperature inversions and effectively reduces the latitudinal gradient of temperature that eventually accelerates the wind. [20] Comparison of the thermal wind field to the cloudtracked winds is the test bench for cyclostrophic balance assumption. Results of our calculations in first order agree with the Venus Express measurements of the cloud top winds [Sanchez-Lavega et al., 2008; R. Moissl et al., submitted manuscript, 2008] that gave the wind speed of 80– 100 m/s at
50°S and fast decrease of zonal wind poleward. However, our thermal wind deviates from the observed one in low latitudes (<40°S), indicating that cyclostrophic balance is not valid here. The core of the cyclostrophic jet is located almost exactly at the cloud tops (
70 km) in all thermal wind calculations. We note, however, that the latitudinal profiles of thermal wind at this altitude have much sharper maxima than those observed in cloud-tracked winds. [21] Following the earlier studies, especially the one by Zasova et al. [2007], we searched for local time variability in the properties of the midlatitude jet. The temperature field clearly indicates radiative cooling by
15 K of the night side atmosphere at the cloud tops (Figure 7) which also

E00B11

propagates to the thermal wind field (Figures 9a – 9c). However, we prefer to be cautious in making conclusions on this basis. First, the coverage by VIRTIS temperature sounding that we used so far is limited by the night side. And, second and probably more important, the weak meandering of the thermal wind field of 10– 20 m/s that results from diurnal variations of temperature structure seems to be within the ‘‘ignorance’’ range of the cyclostrophic hypothesis itself. For instance, the equation (1) was derived ignoring the meridional wind component which is of
15 m/s. So there could be some doubt that the wind field variations of 10 – 20 m/s, although derived by correct numerical procedures, are physically meaningful. [22] The future work will include the use of VIRTIS temperature sounding extended to the northern hemisphere and the day side. This would allow us to better constrain the thermal wind at 20°– 40°S and to extend the coverage to the northern hemisphere and to study hemispheric symmetry of the wind field. Days side temperature sounding would be needed to complete local time coverage and would allow comparison with simultaneously derived cloud-tracked winds. And, finally, extensive radio occultation sounding by VeRa/Venus Express would allow the thermal wind calculations to be extended to as deep as 40 km. Comparison of the thermal wind field derived from the combined VIRTIS/VeRa temperature structure with the winds derived from cloud features tracking in UV (
70 km) and near IR on the night side (
50 km) would eventually allow one to conclude about the regions of validity of the cyclostrophic balance. [23] Acknowledgments. We thank S. Limaye for fruitful discussions of the results. A. Piccialli is grateful to the Max Planck Research School for providing the opportunity to carry out this study. VIRTIS-Venus Express is an experiment developed jointly by IASF-INAF (Italy) and LESIA, Observatoire de Paris (France). The project is funded by ASI and CNES.

References Bougher, S. W., M. J. Alexander, and H. G. Mayr (1997), Upper atmosphere dynamics: Global circulation and gravity waves, in Venus II, edited by S. W. Bougher, et al., pp. 259 – 291, Univ. of Ariz. Press, Tucson. Chub, E. V., and O. I. Yakovlev (1980), Temperature and zonal circulation of the atmosphere of Venus based on the data of radio probe experiments, Cosmic Res., Engl. Transl., 18, 331 – 336. Counselman, C. C., S. A. Gourevitch, R. W. King, G. B. Loriot, and E. S. Ginsberg (1980), Zonal and meridional circulation of the lower atmosphere of Venus determined by radio interferometry, J. Geophys. Res., 85, 8026 – 8030, doi:10.1029/JA085iA13p08026. Drossart, P., et al. (2007), Scientific goals for the observation of Venus by VIRTIS on ESA/Venus Express mission, Planet. Space Sci., 55, 1653 – 1672, doi:10.1016/j.pss.2007.01.003. Gierasch, P., et al. (1997), The general circulation of the Venus atmosphere: An assessment, in Venus II, edited by S. W. Bougher, et al., pp. 459 – 500, Univ. of Ariz. Press, Tucson. Grassi, D., P. Drossart, G. Piccioni, N. I. Ignatiev, L. V. Zasova, A. Adriani, M. L. Moriconi, P. G. J. Irwin, A. Negrao, and A. Migliorini (2008), Retrieval of air temperature profiles in the Venusian mesosphere from VIRTIS-M data: Description and validation of algorithms, J. Geophys. Res., 113, E00B09, doi:10.1029/2008JE003075. Ha¨usler, B., et al. (2006), Radio science investigations by VeRa onboard the Venus Express spacecraft, Planet. Space Sci., 54, 1315 – 1335, doi:10.1016/j.pss.2006.04.032. Lellouch, E., T. Clancy, D. Crisp, A. Kliore, D. Titov, and S. Bougher (1997), Monitoring of mesospheric structure and dynamics, in Venus II, edited by S. W. Bougher, et al., pp. 295 – 324, Univ. of Ariz. Press, Tucson. Leovy, C. B. (1973), Rotation of the upper atmosphere of Venus, J. Atmos. Sci., 30, 1218 – 1220, doi:10.1175/1520-0469(1973)030<1218:ROTUAO>2.0.CO;2.

8 of 9

E00B11

PICCIALLI ET AL.: CYCLOSTROPHIC WINDS IN VENUS MESOSPHERE

Limaye, S. S. (1985), Venus atmospheric circulation: Observations and implication of the thermal structure, Adv. Space Res., 5, 51 – 62, doi:10.1016/0273-1177(85)90270-4. Newman, M., G. Schubert, A. J. Kliore, and I. R. Patel (1984), Zonal winds in the middle atmosphere of Venus from Pioneer Venus radio occultation d a t a , J . A t m o s . S c i . , 4 1 , 1 9 0 1 – 1 9 1 3 , d o i : 1 0 . 11 7 5 / 1 5 2 0 0469(1984)041<1901:ZWITMA>2.0.CO;2. Peralta, J., R. Hueso, and A. Sa´nchez-Lavega (2007), A reanalysis of Venus winds at two cloud levels from Galileo SSI images, Icarus, 190(2), 469 – 477, doi:10.1016/j.icarus.2007.03.028. Piccioni, G., et al. (2008), VIRTIS (Visible and Infrared Thermal Imaging Spectrometer) for Venus Express, Eur. Space Agency, Noordwijk, Netherlands, in press. Roos-Serote, M., P. Drossart, T. Encrenaz, E. Lellouch, R. W. Carlson, K. H. Baines, F. W. Taylor, and S. B. Calcutt (1995), The thermal structure and dynamics of the atmosphere of Venus between 70 and 90 km from the Galileo-NIMS spectra, Icarus, 114(2), 300 – 309, doi:10.1006/ icar.1995.1063. Sanchez-Lavega, A., et al. (2008), Variable winds on Venus mapped in three dimensions, Geophys. Res. Lett., 35, L13204, doi:10.1029/ 2008GL033817. Schubert, G. (1983), General circulation and the dynamical state of the Venus atmosphere, in Venus, edited by D. M. Hunten, et al., pp. 651 – 765, Univ. of Ariz. Press, Tucson. Schubert, G., S. W. Bougher, C. C. Covey, A. D. Del Genio, A. S. Grossman, J. L. Hollingsworth, S. S. Limaye, and R. E. Young (2007), Venus atmosphere dynamics: A continuing enigma, in Exploring Venus as a Terrestrial Planet, Geophys. Monogr. Ser., vol. 176, edited by L. W. Esposito, E. R. Stofan, and T. E. Cravens, pp. 121 – 138, AGU, Washington, D. C. Seiff, A. (1983), Thermal structure of the atmosphere of Venus, in Venus, edited by D. M. Hunten, et al., pp. 215 – 279, Univ. of Ariz. Press, Tucson.

E00B11

Seiff, A., J. T. Schofield, A. J. Kliore, F. W. Taylor, S. S. Limaye, H. E. Revercomb, L. A. Sromovsky, V. V. Kerzhanovich, V. I. Moroz, and M. Y. Marov (1985), Models of the structure of the atmosphere of Venus from the surface to 100 kilometers, Adv. Space Res., 5, 3 – 58, doi:10.1016/0273-1177(85)90197-8. Svedhem, H., et al. (2007), Venus Express—The first European mission to Ve n u s , P l a n e t . S p a c e S c i . , 5 5 , 1 6 3 6 – 1 6 5 2 , d o i : 1 0 . 1 0 1 6 / j.pss.2007.01.013. Taylor, F. W., D. M. Hunten, and L. V. Ksanfomaliti (1983), The thermal balance of the middle and upper atmosphere of Venus, in Venus, edited by D. M. Hunten, et al., pp. 650 – 680, Univ. of Ariz. Press, Tucson. Titov, D. V., et al. (2006), Venus Express science planning, Planet. Space Sci., 54, 1279 – 1297, doi:10.1016/j.pss.2006.04.017. Zasova, L. V., and I. V. Khatuntsev (1997), Thermal zonal wind in the Venus middle atmosphere according to Venera 15 IR-spectrometry, Adv. Space Res., 19, 1181 – 1190, doi:10.1016/S0273-1177(97)00269-X. Zasova, L. V., N. Ignatiev, I. Khatuntsev, and V. Linkin (2007), Structure of the Venus atmosphere, Planet. Space Sci., 55, 1712 – 1728, doi:10.1016/ j.pss.2007.01.011. 

P. Drossart, LESIA, Observatoire de Paris, Section de Meudon, 5 place Jules Janssen, F-92195 Meudon, France. D. Grassi, Istituto di Fisica dello Spazio Interplanetario, INAF, Via del Fosso del Cavaliere 100, I-00133 Rome, Italy. I. Khatuntsev, Space Research Institute, 84/32 Profsojuznaja Str, 117997 Moscow, Russia. A. Migliorini and G. Piccioni, Istituto di Astrofisica Spaziale e Fisica Cosmica, INAF, Via del Fosso del Cavaliere 100, I-00133 Rome, Italy. A. Piccialli and D. V. Titov, Max Planck Institute for Solar System Research, Max Planck Strasse 2, D-37191 Katlenburg-Lindau, Germany.

9 of 9

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114, E00B32, doi:10.1029/2008JE003073, 2009

for

Full Article

Gravity waves in the upper atmosphere of Venus revealed by CO2 nonlocal thermodynamic equilibrium emissions R. F. Garcia,1 P. Drossart,2 G. Piccioni,3 M. Lo´pez-Valverde,4 and G. Occhipinti5 Received 10 January 2008; revised 29 December 2008; accepted 14 January 2009; published 17 March 2009.

[1] The imaging capabilities of the Visible and Infrared Thermal Imaging Spectrometer-

Mapper (VIRTIS-M) onboard Venus Express mission are used to analyze perturbations of CO2 nonlocal thermodynamic equilibrium emissions in the thermosphere of Venus. These emissions with a wavelength of 4.3 mm originate from the 110–140 km altitude range and are sensitive to density perturbations. They show wave-like perturbations of about 0.5% root-mean-square amplitude of the background signal with horizontal wavelengths in the 90–400 km range. The horizontal phase velocities are similar in magnitude and direction from one orbit to the next, with averages of 70 m/s westward and 30 m/s northward. The geographical wave distribution and the orientation of wavefronts demonstrate that the polar vortex at work in the cloud layer is the source of these gravity waves. The large westward zonal phase velocity in the 0900–1500 local time range argues in favor of the superrotation dynamics within 30° from the south pole to an altitude of at least 115 km. The gravity wave dispersion relation and the geographical distribution of wavefront amplitudes suggest the presence of a solar to antisolar wind close to the south pole. Because the centroid altitude and the sensitivity of these emissions to gravity wave perturbations are only roughly estimated, it is not possible to make a quantitative estimate of the upward energy transfer. However, this study demonstrates the strong influence of the polar vortex on the circulation in the atmosphere of Venus up to the thermosphere. Citation: Garcia, R. F., P. Drossart, G. Piccioni, M. Lo´pez-Valverde, and G. Occhipinti (2009), Gravity waves in the upper atmosphere of Venus revealed by CO2 nonlocal thermodynamic equilibrium emissions, J. Geophys. Res., 114, E00B32, doi:10.1029/2008JE003073.

1. Introduction [2] The atmosphere of Venus is a very active and turbulent medium which is expected to show competition between subsolar to antisolar and superrotation circulations [Gierasch et al., 1997]. The clues to understanding the mechanisms of this competition are in the lower thermosphere of Venus. However, up to now, the dynamics of Venus’ lower thermosphere have only been investigated by indirect evidence relying mainly on the distribution of neutral species, temperature and O2 airglow [Bougher and Borucki, 1994; Bougher et al., 1997]. The nonlocal thermodynamic equilibrium (non-LTE) emissions of CO2 on the dayside upper atmosphere of Venus provide a way of investigating the structure and dynamics of Venus’ 1 Laboratoire de Dynamique Terrestre et Plane´taire, UMR5562, Observatoire Midi-Pyre´ne´es, Universite´ de Toulouse, CNRS, Toulouse, France. 2 LESIA, Observatoire de Paris, UPMC, Universite´ Paris-Diderot, CNRS, Meudon, France. 3 INAF, IASF, Rome, Italy. 4 Instituto de Astrofı´sica de Andalucı´a, CISC, Granada, Spain. 5 Institut de Physique du Globe de Paris, Universite´ Denis Diderot Paris 7, Paris, France.

Copyright 2009 by the American Geophysical Union. 0148-0227/09/2008JE003073$09.00

lower thermosphere [Rolda´n et al., 2000; Gilli et al., 2009]. Moreover, the imaging capabilities of the Visible and Infrared Thermal Imaging Spectrometer-Mapper (VIRTIS-M) onboard the Venus Express mission can be used to investigate the evolution over time of these emissions on successive images taken at nadir or at the limb [Drossart et al., 2004, 2007]. [3] The models of thermosphere circulation [Zhang et al., 1996; Bougher et al., 1997] are sensitive to the dynamics and breaking of atmospheric gravity waves. These waves have been observed in Venus troposphere [Gierasch, 1987; Young et al., 1987; Hinson and Jenkins, 1995] and thermosphere [Kasprzak et al., 1993] by different exploration missions. The modeling of the upper atmosphere gravity wave propagation suggests a source in the lower atmosphere [Mayr et al., 1988] which is probably related to the turbulence in the cloud layer [Woo et al., 1980, 1982]. Gravity waves are also used to explain the atmospheric superrotation [Alexander, 1992]. [4] The CO2 non-LTE emissions, which are mainly sensitive to density perturbations [Rolda´n et al., 2000], provide a unique opportunity for direct observation of the dynamics of gravity waves in the 110– 140 km altitude range. Despite the limitations induced by the integrated information content along the line of sight, and the restricted knowledge on both the space/time distribution of these emissions [Drossart et al.,

E00B32

1 of 11

E00B32

GARCIA ET AL.: WAVES IN VENUS UPPER ATMOSPHERE

E00B32

Figure 1. (left) Limb emission profiles at 4.277 mm (solid line) and 4.315 mm (dashed line) on orbit 315. (right) Spectral profiles at 132 km and 118 km altitudes on the limb profile (vertical bar indicates the position of the wavelength at 4.277 mm). The spectral profiles are shifted by about 0.007 mm relative to values indicated in the text owing to a recent correction of spectral registration implemented in the text but not shown. 2007] and their sensitivity to density perturbations, the perturbations of these emissions can be used to characterize the wave content and dynamics of the upper atmosphere and its background circulation. [5] This study first describes the structure of CO2 nonLTE emissions at the 4.3 mm wavelength as seen by the VIRTIS-M spectral imaging instrument. The structure and dynamics of wave-like perturbations are then extracted through a specific data processing scheme. Finally, the source of wave excitation and the consequences on Venus circulation models are discussed.

2. CO2 Non-LTE Emissions at 4.3 mm [6] The non-LTE CO2 emissions in the upper atmosphere of Venus are described in detail by Gilli et al. [2009] in this issue. These emissions are due to the excitation of CO2 molecules by dayside solar radiations. Owing to the low atmospheric density above 90 km altitude, these molecules return to their ground state in a noncollisional manner by releasing photons in a series of emission bands around the 4.3 mm wavelength. Non-LTE situations of pure radiative excitation, like the present CO2 emissions, are relatively insensitive to the actual temperature, while their density dependence is more significant [Rolda´n et al., 2000; LopezPuertas et al., 2000]. In this section, we will recall only the main features of these emissions as seen by VIRTIS-M. [7] At the spectral resolution of VIRTIS-M, the limb profiles present two main peaks at 4.277 mm and 4.315 mm in the altitude range 110 to 140 km. The first emission peak is weaker than the second and presents a maximum on limb data at higher altitudes. The maximum emission strength and the altitude of the emission centroid vary with solar zenith angle and local time. The centroids of the emissions seen at the limb are in the 125–135 km range at the 4.277 mm wavelength and the 115–120 km range at 4.315 mm. An example of these emissions measured by VIRTIS-M at the

limb is shown in Figure 1 for the two emission wavelengths at 1300 local time (LT) and at 50.8° solar zenith angle on orbit 315. In Figure 1, the spectra around the 4.3 mm wavelength clearly show the two emission peaks at 4.277 mm and 4.315 mm plus additional non-LTE emissions above 4.4 mm. [8] When observed at nadir, CO2 non-LTE emissions at 4.3 mm show a clear variation with solar zenith angle, as can be seen in Figure 2 from a south polar view acquired during orbits 473 and 474. Some wavelike features, hardly discernible in Figures 2a and 2b, appear over the background emission. The rest of this study will discuss the structure and dynamics of these features. The structure of the background emission is not investigated.

3. Characteristics of Wave Patterns 3.1. Data Processing [9] This study uses the calibrated VIRTIS-M data for which continuum effects have been removed and absolute spectral radiances are expressed in W/(m2 mm sr). These data are extracted from the VIRTIS database and processed in the following steps: (1) correction of residual instrument discrepancies between odd and even samples of VIRTIS-M measurement line by an algorithm computing a smooth correction factor for each sample, (2) selection of image parts far from the limb to avoid limb effects, (3) enhancement of wave-like features through a filtering process based on the removal of low-pass filtered background emission, (4) projection of the filtered image into a geographical coordinate system centered on the subspacecraft point to avoid image distortions, (5) computation of wave displacements between two consecutive images for the same wavelength through an algorithm based on cross correlation, and (6) back-projection of the images and estimated displacement and velocity vectors into the true geographical system. [10] Calibrated VIRTIS-M data are corrected for instrument artifacts such as discrepancies between odd and even

2 of 11

E00B32

GARCIA ET AL.: WAVES IN VENUS UPPER ATMOSPHERE

E00B32

Figure 2. (a and b) Venus south polar views of radiance at the 4.315 mm wavelength averaged over four VIRTIS-M bands, for images taken along orbits 473 (Figure 2a) and 474 (Figure 2b). Color scale from black to white ranges from 0 to 0.06 W/(m2m msr). (c and d) Same images after wave enhancement by the data processing scheme. Color scales range from
1.5 to +1.5% of the background signal on the left (Figure 2c) and from
3 to +3% on the right (Figure 2d). Latitude contours are indicated in degrees, and the terminator line is almost horizontal. samples along the VIRTIS measurement line. However, owing to the low amplitudes of the wave signals investigated here, an additional correction of this instrument effect is applied in the first step of the algorithm. It is based on the idea that each pixel value at one wavelength pij is related to the true pixel radiance Rij by a correction factor cij in the following way: pij = Rijcij. This correction factor is estimated under the following hypotheses. cij is smooth along the sample direction (j) because the correction of a given sample is approximately the same for two consecutive measurement lines; and cij is close to 1 because the relative difference between odd and even samples of VIRTIS-M images is below 10% for more than 95% of the pixels. An algorithm is implemented on the basis of these two assumptions and the p true radiance is computed by Rij = cijij . [11] The second step is carried out manually to exclude regions of the image outside the 50° arc limit around the subspacecraft point. The third step is designed to enhance the wave features on nadir images at the 4.3 mm wavelength. In order to improve the signal-to-noise ratio, the images are first averaged over four VIRTIS-M wavelengths around the first and the second emission peaks (respectively from 4.26 mm to 4.29 mm and from 4.305 mm to 4.335 mm). The background radiation Bij is estimated by low-pass filtering of the radiance values Rij with a median filter 16 pixels wide. The filtered image is then computed through Fij = Rij
Bij, and the normalized filtered image is estimated through Nij = Fij/Bij. The wave features in these images taken far from the planet, at spacecraft altitudes greater than 20,000 km, have

been shown to be almost insensitive to the median filter window for widths larger than 16 pixels. However, even if the use of a median filter is justified to reduce data spike effects, its frequency response is difficult to estimate. The normalized filtered images Nij are used to estimate the amplitudes of the waves relative to the background. Two examples of wave enhancement are presented for CO2 non-LTE emissions at 4.315 mm in Figures 2c and 2d during orbits 473 and 474. On the nightside, the low level of background signal creates spurious features not shown here. On the dayside, wavefront like features are visible which correspond to the small-scale features observed on the original images. [12] The projection and back-projection at steps 4 and 6 are performed to avoid distortions of pixels when plotted in geographical coordinates close to the south pole. The average subspacecraft point during the image acquisition is computed, and the pixels coordinates are simply rotated and back-rotated from the true geographical coordinates to a geographical coordinate system in which this point has zero latitude and longitude. In this manner, the latitudes and longitudes of pixels are kept within a range of ±30°. The images are projected on a regular (latitude, longitude) grid of 0.15° resolution in this new coordinate system. [13] When the VIRTIS data include two images of the same region at different times along the same orbit, the horizontal displacement of the waves can be estimated by cross correlation of the two images. The typical time difference between two consecutive images is about 1800 s. Similarly, if two wavelengths sample different altitudes, the

3 of 11

GARCIA ET AL.: WAVES IN VENUS UPPER ATMOSPHERE

E00B32

E00B32

Figure 3. Focus on two consecutive VIRTIS-M images at 4.315 mm wavelength filtered and projected at nadir along orbits (a and c) 473 and (b and d) 474. Color scales range from
1.5 to +1.5% of the background signal on the left and from
3 to +3% on the right. Black grid lines are plotted with a step of 25 pixels equal to the size of the correlation window. apparent displacement of the wavefront between the two wavelengths on the same image can be used to estimate the vertical wavelength [Vargas et al., 2007]. An important assumption is that the wavefront shape remains constant between the two images. Our method computes the pattern correlation coefficient (PCC) [Schmetz et al., 1993] between two images, defined by PCCij ðn; mÞ ¼

s12 ðn; mÞ s1 s2 ðn; mÞ

ð1Þ

For each point on the first image, located at coordinates (i, j), a square area of K*K pixels centered on this point is defined, and the same square area is defined on the second image centered on the coordinates (i + n, j + m). s1 is then the standard deviation of the signal over the area of the first image centered on the (i, j) coordinates, and s2(n, m) is the standard deviation of the signal over the area of the second image centered on (i + n, j + m) coordinates. s12(n, m) is the covariance of zero mean signals over the two areas. Values of PCCij(n, m) range from
1 to +1 and are calculated over the range [
M2 ; M2 ] of indices n and m. The maximum value = PCCij(n0, m0)) indicates the image of this function (PCCmax ij shifts (n0, m0) at which the two images are best correlated. In order to improve the quality of image shifts retrieved, a minimum value criterion on the correlation is required by > 0.8, and only points with all neighbors fitting PCCmax ij this criterion are retained to avoid image border effects. Additionally, the zero shifts values (n0 = 0 and m0 = 0) are

excluded because instrument effects generally report a signal at zero shifts. Once the best correlated parts in the images are determined and selected, the image shifts are converted into real displacement between the pixels, and then interpreted as wavefront velocities. This image correlation algorithm has been validated by comparison with outputs of COSI-CORR software [Leprince et al., 2007]. Figure 3 shows examples of consecutive images after filtering and projection at Nadir. In Figure 3, the displacements of wavefronts are clearly visible. 3.2. Wave Structure and Distribution [14] The VIRTIS-M data containing wave-like features after processing are summarized in Table 1 with corresponding data parameters. These data have been selected from the large VIRTIS-M database because the same region is mapped twice during the same orbit, enabling the computation of the horizontal displacement of wave features at the same wavelength. The VIRTIS-M data covering the 4.3 mm emission on the dayside at nadir are restricted to south pole or terminator views owing to thermal constraints on the instrument. [15] The two non-LTE emission peaks at the 4.277 mm and 4.315 mm wavelengths show the same patterns after data processing. The image correlation algorithm was not able to detect any coherent horizontal displacement of these patterns between these two wavelengths on the same image. In the rest of the study, only results at the 4.315 mm wavelength are presented because the signal to noise ratio is lower at the 4.277 mm wavelength.

4 of 11

GARCIA ET AL.: WAVES IN VENUS UPPER ATMOSPHERE

E00B32

E00B32

Table 1. Summary of VIRTIS-M Data Providing More Than 100 Horizontal Velocity Estimates and Corresponding Processing Parameters Orbit

Session

Pixel Size (km)

Range or Dominant Horizontal Wavelength (km)

K/2

M/2

Mean Local Time (h)

219 220 220 232 233 251 473 474 476

04 01 04 01 01 08 00 – 01 00 – 01 00 – 01

16 16 16 16 16 16 16 16 16

340 250 230 210 275 230 90 – 400 110 – 320 90 – 400

10 10 10 10 10 10 12 12 12

10 10 10 10 10 10 12 12 12

11.3 11.5 11.3 11.8 11.8 12.0 12.3 13.5 12.2

[16] The horizontal wavelengths of the upper atmosphere wave features are in the range 90 –400 km. The maximum pixel size of VIRTIS-M data in high space resolution mode (about 16.5 km) should make it possible to resolve horizontal wavelengths below 90 km, but such features are rarely seen. Moreover, the various filtering procedures tested do not reveal horizontal wavelengths larger than 400 km. We therefore believe that the horizontal wavelength range of 90– 400 km is not biased by instrument resolution or data processing. [17] Figure 4 shows an example of the typical distribution of wave features observed from south polar view. For comparison between the three images at the 4.315 mm wavelength, the normalized filtered images are plotted with the same scale (±2.5%). On the composite image, coherent circular wavefronts with horizontal wavelengths larger than 200 km are visible in the middle of the image indicating a possible source region close to the south pole. The wave activity is low on the morningside (on the right). On the eveningside (on the left), small horizontal wavelength features (90 – 200 km) are observed close to dusk with wavefronts almost parallel to the terminator line. The geographical distribution of wave features varies between one orbit and the next, and a wide range of horizontal wavelengths can be observed on the same image, as shown in Figure 2. However, the orientation of wavefronts and the low level of wave activity on the morningside are persistent features. [18] The root-mean-square (RMS) amplitudes of wave features on all images processed are presented in Figure 5 for the two emission peaks at the 4.277 mm and 4.315 mm wavelengths. The areas over which RMS amplitudes are

computed correspond to the image points at which a horizontal velocity vector has been retrieved. The amplitudes of wave-like perturbations of CO2 non-LTE emissions are slightly larger at the 4.315 mm than at 4.277 mm wavelength. The ratio of RMS wave amplitudes is about 3/4. If the emission radiances are directly interpreted in terms of wave amplitudes, this observation may suggest a wave dissipation between the two emission centroids. However, the unknown sensitivities of non-LTE emissions to density perturbations convolute the wave signal preventing the direct interpretation of emission perturbation amplitudes in terms of wave amplitudes. 3.3. Wave Dynamics [19] The horizontal displacement of the wave features is computed for the second peak of CO2 non-LTE emissions (around 4.315 mm) by using the algorithm described in step 5 of data processing. The parameters of the cross-correlation analysis must be adapted to the dominant horizontal wavelength of the signal and to the phase shift between the two images. As shown in Figure 3, the image correlation window size (K) is chosen to be larger than the dominant horizontal wavelength. The displacement range of the correlation window (±M/2) is set to be equal to half the size of the of correlation window (K/2) in order to avoid aliasing. The validity of this parameter was visually checked by comparing image pairs (Figure 3). Parameters K/2 and M/2 chosen for each image pair are given in Table 1, and vary in the range 10– 12 pixels. In addition to the quality tests described in the algorithm, the relative difference between horizontal displacement vectors estimated at 4.277 mm and 4.315 mm is computed. Displacement estimates with a relative difference

Figure 4. Venus south polar views of normalized filtered images (Nij) at the 4.315 mm wavelength on orbit 232. Black and white color scale ranges from
2.5 to 2.5% on the three images. Latitude contours are indicated in degrees, the terminator line is almost horizontal at the lowest border of the images, and the eveningside (morningside) is on the left (right) side. The data processing generates spurious vertical stripes at the limb owing to the strong contrast there. 5 of 11

E00B32

GARCIA ET AL.: WAVES IN VENUS UPPER ATMOSPHERE

Figure 5. Root-mean-square amplitude (in percent) of wave-like features at the 4.315 mm wavelength (circles) and at 4.277 mm (grey dots) on each image processed as a function of average local time (in hours) in reverse scale (eveningside on the left).

E00B32

larger than 15% are excluded. From the horizontal displacement estimates and knowing the time lag between the two image points, horizontal phase velocity vectors are calculated. [20] Some examples of velocity vector estimates are shown in Figure 6 for orbits 473 and 474. The wave velocities are obtained in the part of the image that shows wave features with strong contrasts and which maintain the same shape during the time lag between the two images (1800 s). The phase velocity vector is not perpendicular to the wavefront and it presents a strong westward zonal component. [21] The horizontal velocity vectors for the whole data set are plotted in Figure 7. The amplitudes of the phase velocity vectors are in the 50– 110 m/s range. The direction of velocity vectors is zonal westward in the direction of lower atmosphere superrotation, with a small meridional component. The coherency of the velocity vectors is remarkable, because these estimates are obtained from different images along different orbits with different conditions of observation. Meridional and zonal velocities are plotted as a function

Figure 6. Venus south polar views of enhanced wave features at the 4.315 mm wavelength, and corresponding estimates or horizontal wave velocities along orbits (a) 473 and (b) 474. One velocity estimate in 15 is presented to clarify the plots. Local time is converted into longitude (zero longitude is noon) with evening on the left side. 6 of 11

E00B32

GARCIA ET AL.: WAVES IN VENUS UPPER ATMOSPHERE

E00B32

2.9 mm wavelength. So, the reflected light on cloud top is not the main contribution at the 4.3 mm wavelength. [23] Our data processing scheme is limited to integer pixel displacements, but this limitation will be overcome in future subpixel versions by interpolation of the correlation function or by analysis in the frequency domain [Leprince et al., 2007]. The main limitation comes from the low signal to noise of the wave features partly owing to their low amplitude relative to background emission and partly owing to operational constraints on the spacecraft orientation and on VIRTIS-M exposure time. However, the strong consistency of our results, obtained along different orbits for different observation conditions, excludes the possibility of these wave features having been created by noise or artifacts. [24] Another limitation of our analysis arises from the integrated character of the 4.3 mm radiance along the line of sight. In order to fully interpret these data it is necessary to know the large wavelength space/time variations of nonLTE CO2 volumetric emission, and the sensitivity of these emissions to density perturbations. The first parameter will be obtained from an analysis of the whole VIRTIS data set, including limb profiles, and the second will be modeled as an extension of previous modeling [Rolda´n et al., 2000; Lopez-Puertas and Lopez-Valverde, 1995; Lo´pez-Valverde et al., 2007].

4. Interpretation Figure 7. Estimates of horizontal wave velocities. (top) Norm of the horizontal velocity vector (m/s) as a function of local time (h) in reverse scale (eveningside on the left). (bottom) Norm and direction of horizontal velocity vectors as a function of latitude and local time converted into longitude (zero longitude is noon) with evening on the left side. In this plot, one velocity estimate in 15 is presented to clarify the plots. of local time and latitude on Figure 8. The meridional component is positive (in a northern direction) with an average value of about 30 m/s. The zonal component is westward with an average value of about 70 m/s. The only possible trend in these plots is an increase of the meridional component when approaching dusk, but the low number of images processed in this study did not allow us to validate this feature fully. 3.4. Possible Artifacts and Limitations [22] The radiance observed at nadir is the sum of the direct non-LTE emission and possible other components coming from the reflection of these emissions and/or the sunlight on cloud top at these wavelengths. In order to check the possible contribution of light reflected on cloud top, the wave enhancement procedure was applied on the same images at the 2.9 mm wavelength and in the visible for which the reflected sunlight is the main contribution. Some wave features were detected on these images, but the positions and orientations of these features are completely different from the ones observed at the 4.3 mm wavelength. Similarly, the background emission observed at the 4.3 mm wavelength and its large horizontal wavelength variations are different from the ones observed in the visible and at the

4.1. Wave Type [25] In order to identify the type and structure of the observed wave features, the results are analyzed in the framework of gravity and acoustic wave propagation theories. The difficulty when discussing the wave type (gravity or acoustic waves) lies in the following two major unknowns: the vertical wave structure and the background circulation. The vertical wave structure can be constrained by the thickness of the emission layers and the fact that the two main non-LTE emissions at 4.277 mm and 4.315 mm display the same wave features whereas their centroid altitudes seen at the limb are separated by about 15 km. These two observations impose a minimum vertical wavelength of approximately 30 km. Concerning the background thermospheric circulation, previous models [Zhang et al., 1996; Bougher et al., 1997] predict a strong subsolar to antisolar circulation with winds in the range 60– 130 m/s in this altitude range (110 –140 km) and in this angular distance range to the terminator (15°– 35°). However, the particularities of the atmospheric dynamics in the polar region [Piccioni et al., 2007] can modify this simple picture significantly. [26] The wavefronts are almost parallel to lines of constant latitude, so the horizontal component of the apparent phase velocity (Ch0) can be approximated by the meridional component of the phase velocity which is perpendicular to the wavefronts. An average apparent horizontal phase velocity is Ch0 = 30 ± 10 m/s. If we assume that our wave features are acoustic waves, the intrinsic horizontal phase velocity is Chi = Ch0
U, with the background meridional wind U in the range [
130;0] m/s, which is a very conservative hypothesis. We deduce that Chi is in the range [20;170] m/s taking into account error bars. The speed of sound at these altitudes is about 215 m/s. Because this value is a minimum estimate of the intrinsic horizontal phase

7 of 11

E00B32

GARCIA ET AL.: WAVES IN VENUS UPPER ATMOSPHERE

E00B32

Figure 8. (top) Meridional and (bottom) zonal components of the apparent phase velocities as a function of (left) local time in reverse scale and (right) latitude. velocity Chi for acoustic waves, obtained only for horizontal wave propagation, our wave features cannot be acoustic waves. So, the gravity wave hypothesis is the only remaining possibility. [27] In conclusion, the wave features presented here are indicative of gravity wave phenomena. We will explore in the next two sections the source and propagation of these waves and their links to the background atmospheric dynamics. 4.2. Gravity Wave Source [28] As shown in Figures 4 and 6 the main wavefronts are almost circular and parallel to lines of constant latitude. This observation suggests a source close to the south pole. Figure 9 shows a comparison between the enhanced wave features at the 4.315 mm wavelength and the polar vortex observed at the 5.05 mm wavelength in the same images. This comparison has also been performed on polar vortex features observed at the 3.3 mm wavelength (which is less sensitive to saturation problems) for all the images processed in this study. The direction perpendicular to wavefronts is always pointing in the direction of the region of strongest polar vortex activity, whatever the position and shape of the vortex. These observations clearly identify the polar vortex as the source of the gravity waves observed in the 110–140 km altitude range. This conclusion is also supported by theoretical studies [Baker et al., 1999; Nappo, 2002] which demonstrate that turbulence is able to generate gravity waves propagating upward. For such a source close to south pole, the gravity waves propagate with horizontal components of intrinsic phase and group velocities along meridional direction from the pole to the equator.

[29] Moreover, the analysis of VIRTIS-M images has shown that the wave features at the 4.315 mm wavelength are more numerous with larger amplitude when the polar vortex is on the dayside than on the nightside. As an example, Figures 10a and 10b show that only a few wave features of low amplitude are visible when the polar vortex is completely on the nightside. This observation suggests that the gravity wave propagation from the cloud layer up to the 110 –140 km altitude range is more favorable on the dayside than on the nightside. Figure 10c shows the BruntVa¨isa¨la¨ frequency profile computed for day and night Venus atmosphere parameters [Hunten et al., 1983]. In the 100– 140 km range, it increases on the nightside and decreases on the dayside. The shape of the Brunt-Va¨isa¨la¨ frequency profile on the dayside suggests that gravity waves may be trapped between altitudes of 60 km to 140 km for wave frequencies between 3 and 3.5 mHz. In this layer, trapped waves propagate horizontally with horizontal wavelengths depending on the trapped mode number, and with low attenuation along their path [Lighthill, 2001]. The presence of these trapped waves at frequencies higher than 3 mHz could explain the difference between day and nightside observations. Moreover, the presence of strong background winds and vertical wind variations may also play an important role in the day/night difference of wave propagation by creating wind ducts [Nappo, 2002; Snively et al., 2007; Zhou and Morton, 2007]. 4.3. Implications for Background Circulation [30] The gravity waves at these heights propagate within the global thermosphere circulation. The horizontal phase

8 of 11

E00B32

GARCIA ET AL.: WAVES IN VENUS UPPER ATMOSPHERE

E00B32

Figure 9. Venus south polar views along orbits (a and b) 220, (c and d) 232, and (e and f) 251 of (left) enhanced wave features at 4.315 mm wavelength and (right) emission at 5.05 mm wavelength which reveals the structure and position of the south polar vortex. Latitude contours are indicated in degrees, and the terminator line is clearly visible through the spurious features created by the data filtering on the nightside. speeds obtained therefore include the effects of both wave propagation and background wind. Consequently, the observed gravity wave propagation can constrain the background wind structure. [31] If we assume that the source of the gravity waves is the turbulent area of the polar vortex close to the south pole, their propagation without wind should be mainly along

meridians. However, a strong zonal component of the horizontal phase velocity is observed in Figures 7 and 8. This observation implies that a zonal westward background wind of about 70 m/s is present at the altitude of 4.315 mm wavelength non-LTE emission perturbations in the 0900 – 1500 LT range in the polar regions. The gravity wave propagation therefore suggests that the zonal lower

9 of 11

GARCIA ET AL.: WAVES IN VENUS UPPER ATMOSPHERE

E00B32

E00B32

Figure 10. Venus south polar views along orbit 253 of (a) enhanced wave features at the 4.315 mm wavelength and (b) emission at the 5.05 mm wavelength revealing the structure and position of the south polar vortex. Latitude contours are indicated in degrees. (c) Plot of the Brunt-Va¨isa¨la¨ frequency (in mHz) as a function of altitude (in km) on the dayside (solid line) and on the nightside (dotted line), computed for Venus atmosphere models [Hunten et al., 1983]. thermosphere circulation in the 0900 – 1500 LT range is in the direction of the lower atmosphere superrotation up to 30° from the south pole and up to an altitude of at least 115 km. [32] Moreover, because of the absence of phase difference between the wave features observed at the 4.277 mm and 4.315 mm wavelengths, the vertical wavelength of these waves must be larger than twice the centroid altitude difference of these two emissions (15 km at the limb) [Vargas et al., 2007]. In order to use this constraint, the gravity wave dispersion relation with constant background wind should be written [Li et al., 2007] N2 2

ðC0
U Þ

¼ m2 þ k 2 þ

1 4H 2

ð2Þ

where m = 2p lz is the vertical wave number with lz the vertical wavelength, N = 2p f = 0.0215 rad/s (or f = 3.4 mHz) is the Brunt-Va¨isa¨la¨ pulsation which is estimated from a dayside Venus atmosphere model at a latitude of 120 km [Hunten et al., 1983], C0 is the apparent wave velocity, U is the meridional wind, k = 2.513  10
5 rad/m is the horizontal wave number assuming an average 250 km horizontal wavelength, and H = 4 km is the density scale height at 120 km altitude. If we assume an intrinsic wave propagation from the pole to the equator, and a lower bound on the vertical wavelength lz > 2 * 15 km = 30 km and on the apparent phase velocity C0 > Ch0 = 30 m/s, by using equation (2) we deduce an upper bound on the meridional phase velocity U <
58 m/s. This analysis suggests that the meridional wind is opposite to the wave propagation direction with an amplitude greater than about 60 m/s. This conclusion is in favor of a subsolar to antisolar thermospheric meridional circulation consistent in direction and amplitude with previous results [Zhang et al., 1996; Bougher et al., 1997]. [33] In addition to the gravity wave dispersion analysis, the previous conclusion is also supported by the geographical distribution of gravity wavefronts. Sun et al. [2007]

demonstrated recently that gravity waves propagating against the wind direction are amplified relative to other propagation directions. Assuming a wave source close to the south pole and a strong subsolar to antisolar wind, this prediction is consistent with our observation of gravity wave features above the noise level only in the 0900– 1500 LT range. The wave feature distribution therefore also supports the presence of a strong subsolar to antisolar wind at the south pole.

5. Conclusion [34] The nadir images of non-LTE CO2 emissions present wave-like perturbations of about 0.5% RMS amplitude relative to background signal. These features are mainly observed in the 0900 – 1500 LT range, and their horizontal wavelengths are in the 90 – 400 km range. The dynamics of these wave features are similar for all the images processed along different orbits. Average zonal and meridional components of the horizontal phase velocities are respectively 70 m/s westward and 30 m/s northward. The horizontal phase velocities and horizontal wavelengths are compatible with gravity waves propagating in the 110 – 140 km altitude range which is sampled by non-LTE CO2 emissions at the 4.3 mm wavelength. The considerable uncertainty on the altitude of these perturbations restricts the acquisition of information on the vertical structure of the waves. However, the geographical distribution and the orientation of wavefronts clearly identify the polar vortex at work in the cloud layer as the source of gravity waves. Finally, the characteristics of gravity wave propagation can be used to infer the background atmospheric winds. The strong zonal component of gravity wave phase velocities is not due to their propagation from a source close to the south pole, and can consequently be attributed to the background atmospheric circulation in the 110 – 140 km altitude range. The gravity wave dispersion relation and the geographical distribution of wavefronts strongly argue in favor of the presence of a subsolar to antisolar wind of amplitude larger than 60 m/s close to the south pole.

10 of 11

E00B32

GARCIA ET AL.: WAVES IN VENUS UPPER ATMOSPHERE

[35] This first direct mapping of thermospheric gravity waves demonstrates the strong influence of the polar vortex on the atmospheric polar circulation of Venus [Piccioni et al., 2007], and illustrates an upward energy transfer from the cloud layer to the thermosphere. Moreover, this study seems to demonstrate the persistence of superrotation dynamics at least up to 115 km in polar regions (up to 30° from the south pole). Furthermore, our data support the existence of the subsolar to antisolar circulation predicted by thermosphere circulation models [Zhang et al., 1996; Bougher et al., 1997], with meridional wind speeds greater than 60 m/s in this polar region. [36] The increase of the VIRTIS database will improve the statistics of wave features, and its combination with VMC data [Markiewicz et al., 2007] will make it possible to follow gravity wave propagation from the top of the cloud layer up to the thermosphere. In order to improve our description and understanding of these wave phenomena, a large-scale threedimensional model of CO2 volumetric emission rates should be developed by using the limb and nadir VIRTIS observations, the sensitivity of these emissions to gravity wave perturbations should be modeled, and the gravity wave propagation within the background circulation should be computed. Finally, these high-altitude emissions can also be used to investigate transient infrasonic signals generated by quakes and volcanic events [Garcia et al., 2005]. [37] Acknowledgments. Two anonymous reviewers improved this paper considerably by their constructive comments. We thank Ste´phane Erard and Alejandro Cardesin for computation and validation of VIRTIS-M data and corresponding observation geometry, Ricardo Hueso Alonso for his help with IDL data projection routines, F. Ayoub for his help with COSI-CORR software, Se´bastien Lebonnois for helpful discussions on the atmosphere dynamics, and P.E. Mallet for validating the correlation software. This study was funded by CNES through space research scientific projects.

References Alexander, M. (1992), A mechanism for the Venus thermospheric superrotation, Geophys. Res. Lett., 19, 2207 – 2210. Baker, R., G. Schubert, and P. Jones (1999), High rayleigh number compressible convection in Venus’ atmosphere: Penetration, entrainment, and turbulence, J. Geophys. Res., 104, 3815 – 3832. Bougher, S., and W. Borucki (1994), Venus O2 visible and IR nightglow: Implications for lower thermosphere dynamics and chemistry, J. Geophys. Res., 99, 3759 – 3776. Bougher, S., M. Alexander, and H. Mayr (1997), Upper atmosphere dynamics: Global circulation and gravity waves, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 259 – 291, Univ. of Ariz. Press, Tucson. Drossart, P., et al. (2004), VIRTIS imaging spectrometer for the ESA/Venus Express mission, in Earth Observing Systems IX, vol. 5583, pp. 175 – 185, Int. Soc. for Opt. Eng., Bellingham, Wash. Drossart, P., et al. (2007), A dynamic upper atmosphere of Venus as revealed by VIRTIS on Venus Express, Nature, 450, 641 – 645, doi:10.1038/nature06140. Garcia, R., P. Lognonne´, and X. Bonnin (2005), Detecting atmospheric perturbations produced by Venus quakes, Geophys. Res. Lett., 32, L16205, doi:10.1029/2005GL023558. Gierasch, P. (1987), Waves in the atmosphere of Venus, Nature, 328, 510 – 512. Gierasch, P., et al. (1997), The general circulation of the Venus atmosphere: An assessment, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 459 – 500, Univ. of Ariz. Press, Tucson. Gilli, G., M. Lo´pez-Valverde, P. Drossart, G. Piccioni, S. Erard, and A. Cardesı´n-Moinelo (2009), Limb observations of CO2 and CO non-LTE emissions in the Venus atmosphere by VIRTIS/Venus Express, J. Geophys. Res., doi:10.1029/2008JE003112, in press. Hinson, D., and J. Jenkins (1995), Magellan radio occultations measurements of atmospheric waves on Venus, Icarus, 114, 310 – 327. Hunten, D., L. Colin, T. Donahue, and V. Moroz (Eds.) (1983), Venus, 1143 pp., Univ. of Ariz. Press, Tucson.

E00B32

Kasprzak, W. T., H. B. Niemann, A. E. Hedin, and S. W. Bougher (1993), Wave-like perturbations observed at low altitudes by the Pioneer Venus Orbiter Neutral Mass Spectrometer during orbiter entry, Geophys. Res. Lett., 20, 2755 – 2758. Leprince, S., S. Barbot, F. Ayoub, and J. Avouac (2007), Automatic and precise ortho-rectification, coregistration, and subpixel correlation of satellite images: Application to ground deformation measurements, IEEE Trans. Geosci. Remote Sens., 45, 1529 – 1558. Li, F., G. Swenson, A. Liu, M. Taylor, and Y. Zhao (2007), Investigation of a wall wave event, J. Geophys. Res., 112, D04104, doi:10.1029/ 2006JD007213. Lighthill, J. (2001), Waves in Fluids, 520 pp., Cambridge Univ. Press, Cambridge, U.K. Lopez-Puertas, M., and M. A. Lopez-Valverde (1995), Radiative energy balance of CO2 non-LTE infrared emissions in the Martian atmosphere, Icarus, 114, 113 – 129, doi:10.1006/icar.1995.1047. Lopez-Puertas, M., M. Lopez-Valverde, R. Garcia, and R. G. Roble (2000), A review of CO2 and CO abundances in the middle atmosphere, in Atmospheric Science Across the Stratopause, Geophys. Monogr. Ser., vol. 123, edited by D. E. Siskind, S. D. Eckermann, and M. E. Summers, pp. 83 – 100, AGU, Washington D.C. Lo´pez-Valverde, M.A., P. Drossart, R. Carlson, R. Mehlman, and M. RoosSerote (2007), Non-LTE infrared observations at Venus: From NIMS/ Galileo to VIRTIS/Venus Express, Planet. Space Sci., 55, 1757 – 1771, doi:10.1016/j.pss.2007.01.008. Markiewicz, W., et al. (2007), Morphology and dynamics of the upper cloud layer of Venus, Nature, 450, 633 – 636, doi:10.1038/nature06320. Mayr, H. G., I. Harris, W. T. Kasprzak, M. Dube, and F. Varosi (1988), Gravity waves in the upper atmosphere of Venus, J. Geophys. Res., 93, 11,247 – 11,262. Nappo, C. (2002), An Introduction to Atmospheric Gravity Waves, 276 pp., Academic, San Diego. Piccioni, G., et al. (2007), South-polar features on Venus similar to those near the north pole, Nature, 450, 637 – 640, doi:10.1038/nature06209. Rolda´n, C., M. Lo´pez-Valverde, and M. Lo´pez-Puertas (2000), Non-LTE infrared emissions of CO2 in the atmosphere of Venus, Icarus, 147, 11 – 25. Schmetz, J., K. Holmlund, J. Hoffman, B. Strauss, B. Mason, V. Gaertner, A. Koch, and L. Van De Berg (1993), Operational cloud-motion winds from Metosat infrared images, J. Appl. Meteorol., 32, 1206 – 1225. Snively, J. B., V. P. Pasko, M. J. Taylor, and W. K. Hocking (2007), Doppler ducting of short-period gravity waves by midlatitude tidal wind structure, J. Geophys. Res., 112, A03304, doi:10.1029/2006JA011895. Sun, L., W. Wan, F. Ding, and T. Mao (2007), Gravity wave propagation in the realistic atmosphere based on a three-dimensional transfer function model, Ann. Geophys., 25, 1979 – 1986. Vargas, F., G. Swenson, A. Liu, and D. Gobbi (2007), O(1S), OH, and O2(b) airglow layer perturbations due to AGWs and their implied effects on the atmosphere, J. Geophys. Res., 112, D14102, doi:10.1029/ 2006JD007642. Woo, R., J. W. Armstrong, and A. Ishimaru (1980), Radio occultation measurements of turbulence in the Venus atmosphere by Pioneer Venus, J. Geophys. Res., 85, 8031 – 8038. Woo, R., J. W. Armstrong, and A. J. Kliore (1982), Small-scale turbulence in the atmosphere of Venus, Icarus, 52, 335 – 345, doi:10.1016/00191035(82)90116-6. Young, R. E., R. L. Walterscheid, G. Schubert, A. Seiff, V. M. Linkin, and A. N. Lipatov (1987), Characteristics of gravity waves generated by surface topography on Venus: Comparison with the VEGA balloon results, J. Atmos. Sci., 44, 2628 – 2639. Zhang, S., S. Bougher, and M. Alexander (1996), The impact of gravity waves on the Venus thermosphere and O2 IR nightglow, J. Geophys. Res., 101, 23,195 – 23,205. Zhou, Q., and Y. Morton (2007), Gravity wave propagation in a nonisothermal atmosphere with height varying background wind, Geophys. Res. Lett., 34, L23803, doi:10.1029/2007GL031061.























P. Drossart, LESIA, Observatoire de Paris, UPMC, Universite´ ParisDiderot, CNRS, 5 place Jules Janssen, F-92195 Meudon, France. R. F. Garcia, Laboratoire de Dynamique Terrestre et Plane´taire, UMR5562, Observatoire Midi-Pyre´ne´es, Universite´ de Toulouse, 14 avenue Edouard Belin, F-31400 Toulouse, France. ([email protected]) M. Lo´pez-Valverde, Instituto de Astrofı´sica de Andalucı´a, CISC, 50 Camino Bajo de Hue´tor, ES-18080 Granada, Spain. G. Occhipinti, Institut de Physique du Globe de Paris, Universite´ Denis Diderot Paris 7, 4 place Jussieu, F-75252 Paris CEDEX 05, France. G. Piccioni, INAF, IASF, via del fosso del cavaliere 100, I-00133 Rome, Italy.

11 of 11

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, E00B18, doi:10.1029/2008JE003185, 2008

for

Full Article

Characterization of mesoscale gravity waves in the upper and lower clouds of Venus from VEX-VIRTIS images J. Peralta,1 R. Hueso,1 A. Sa´nchez-Lavega,1 G. Piccioni,2 O. Lanciano,2 and P. Drossart3 Received 9 May 2008; revised 11 August 2008; accepted 22 August 2008; published 11 December 2008.

[1] Images obtained from the Visible and InfraRed Thermal Imaging Spectrometer

(VIRTIS)-M instrument onboard Venus Express present visible trains of alternating bands of cloud brightness in two different layers: at the upper cloud tops (
66 km altitude) observed in the dayside hemisphere using reflected ultraviolet light (380 nm) and in the lower cloud (
47 km altitude) observed in the nightside hemisphere using thermal radiation (1.74 mm). The waves are nearly zonal (with the bands perpendicular to latitude circles), have wavelengths of 60–150 km, propagate westward with low phase velocities relative to the zonal flow, and are confined in wave packets of 400 to 1800 km in length. The waves in the lower cloud observed in the infrared are widely distributed around the planet, and their appearance varies widely throughout the VIRTIS data set. The locations of both types of waves seem not correlated with latitude, local times, surface topography, or the structure of the wind. In both cases the characteristics of the waves correspond to gravity waves propagating in confined stable layers of the atmosphere. We examine the properties of these waves in terms of a linear model and perform a simple analysis to discuss the vertical stability of the atmosphere within Venus clouds. Citation: Peralta, J., R. Hueso, A. Sa´nchez-Lavega, G. Piccioni, O. Lanciano, and P. Drossart (2008), Characterization of mesoscale gravity waves in the upper and lower clouds of Venus from VEX-VIRTIS images, J. Geophys. Res., 113, E00B18, doi:10.1029/2008JE003185.

1. Introduction [2] Atmospheric internal gravity waves are wave disturbances whose restoration force is buoyancy under stable stratification of the atmosphere [Holton, 1992]. They are commonly observed in the stratosphere of most planets as oscillations on the temperature field and they have also been observed as wavy structures in cloud fields of the Earth [Houze, 1993], Jupiter [Hunt and Muller, 1979; Flasar and Gierasch, 1986; Reuter et al., 2007] and Venus [Gierasch et al., 1997; Markiewicz et al., 2007]. Since they are supported by such a fundamental force (buoyancy) they are of great interest to atmospheric dynamics. On the one hand, from their properties one can infer the static stability of the atmosphere (since these waves can only propagate in regions where the static stability is positive). On the other, they are responsible of a number of important dynamical phenomena including the transfer of energy and momentum vertically. They are of particular importance to the atmosphere of Venus, where their role as a mechanism able to transport momentum from the surface to the atmospheric 1 Departamento de Fı´sica Aplicada I, Escuela Te´cnica Superior de Ingenierı´os, Universidad del Paı´s Vasco, Bilbao, Spain. 2 Istituto di Astrofisica Spaziale e Fisica Cosmica, Istituto Nazionale di Astrofisica, Rome, Italy. 3 Laboratoire d’Etudes Spatiales et d’Instrumentation en Astrophysique, Observatoire de Paris, UPMC, Universite´ Paris-Diderot, CNRS, Meudon, France.

Copyright 2008 by the American Geophysical Union. 0148-0227/08/2008JE003185$09.00

mean circulation has been examined by several teams, although without a final consensus on their role in powering the atmospheric superrotation [Hou and Farrell, 1987; Gierasch, 1987; Leroy and Ingersoll, 1995]. Space probes determined during their descent at different latitudes and local times the existence of at least two stable layers: between 30 and 48 km, and upward of 55 km [Kliore and Patel, 1980; Seiff et al., 1985; Gierasch, 1987]. These altitude levels are home to dense clouds on Venus [Esposito et al., 1997] and gravity waves arising on them can manifest as regular patterns on the clouds albedo fields [Markiewicz et al., 2007; McGouldrick and Toon, 2008]. The altitudes below 30 km and the intermediate layer between 48 and 55 km have small static stabilities and are not suitable candidates for the development of gravity waves. [3] Gravity waves observed at the upper cloud level were first inferred from analysis of Mariner 10 [Belton et al., 1976] and Pioneer Venus [Rossow et al., 1980] ultraviolet images of the upper cloud field at z
65 km. More recently, the Venus Monitoring Camera (VMC) onboard Venus Express has also detected regular cloud patterns identified as gravity waves from their observed characteristics [Markiewicz et al., 2007]. Mariner 10 images of equatorial elongated cloud patterns of about 5000 km in the longitudinal direction and separated
500 km in latitude (circumequatorial belts) were interpreted as evidence of internal gravity waves. These structures propagated southward near the equator with phase speeds of
20 m/s [Belton et al., 1976]. The same circumequatorial belts were seen during the Pioneer Venus mission in addition

E00B18

1 of 12

E00B18

PERALTA ET AL.: VENUS GRAVITY WAVES OBSERVED BY VEX-VIRTIS

to a new type of train of wavelike features composed of dark streaks of
2000 km, separated by
200 km, but in this case lying at a large angle to the constant latitude circles [Rossow et al., 1980]. No evidence for small-scale internal gravity waves or for circumequatorial belts was found in images taken by the Galileo spacecraft in 1990 [Belton et al., 1991; Toigo et al., 1994; Peralta et al., 2007a]. This was linked to a lower abundance of convective clouds visible during the Galileo flyby. The high-resolution images obtained recently by the VMC instrument onboard Venus Express display vigorous convection in near equatorial latitudes and visible wave activity (wavelengths of just a few tens of kilometers) in tropical latitudes in the south hemisphere and in middle to subpolar latitudes in the north hemisphere [Markiewicz et al., 2007]. [4] Evidence for gravity waves in the Venus atmosphere does not come only from imagery. The four Pioneer Venus probes, as they descended through the atmosphere, obtained altitude profiles of temperature (upward of 110 km) and wind velocity (lower cloud region) that showed wavelike variations with vertical wavelengths of 5 – 10 km [Seiff at al., 1980; Counselman et al., 1980]. Similar vertical variations can be seen in infrared temperature soundings [Taylor et al., 1980], and in radio occultation temperatures from Pioneer Venus [Kliore and Patel, 1980], Venera 9 [Kolosov et al., 1980] and Magellan [Hinson and Jenkins, 1995] sensitive to mesospheric vertical levels between the cloud tops (z
65 km) and the thermosphere base (z
100 km). Above these levels, gravity waves with wavelengths between 100 and 600 km have also been detected on the thermosphere as density perturbations studied by the mass spectrometer on the Pioneer Venus orbiter [Niemann et al., 1980; Kasprzak et al., 1988], and from non LTE CO2 emissions observed by the Visible and InfraRed Thermal Imaging Spectrometer (VIRTIS) instrument on Venus Express (R. Garcı´a et al., Gravity waves in Venus upper atmosphere revealed by CO2 non LTE emissions, submitted to Journal of Geophysical Research, 2008). [5] Although several candidates could act as sources of gravity waves (flow over a mountain range, KelvinHelmholtz instability around a jet stream), convective processes generated in the low-stability layers between 48 and 55 km and below 30 km are considered to play the main role for gravity wave generation in the middle and upper regions of the clouds layer [Schubert, 1983; Gierasch, 1987; Leroy and Ingersoll, 1995, 1996; Baker et al., 2000a, 2000b]. [6] In this paper we present the observed characteristics of wavelike features present in the atmosphere of Venus at two cloud levels detected on images at different wavelengths obtained by the VIRTIS instrument [Drossart et al., 2007] onboard Venus Express [Svedhem et al., 2007]. We identify these features as gravity waves and present a simple theoretical approach to interpret them in terms of a ‘‘textbook’’ classical linear model. This is the first time that the abundant and variable mesoscale gravity waves within the lower cloud level of Venus have been characterized. From the comparison between the observations and model we infer the existence of a widely (from latitudes 40° to 80°S) and temporarily extended vertical stable region between 30 and 48 km, in agreement with

E00B18

previous inferences [Gierasch, 1987; Leroy and Ingersoll, 1995].

2. Observations [7] The instrument VIRTIS on board Venus Express is an imaging spectrometer (VIRTIS-M) combined with a highresolution IR-spectrometer (VIRTIS-H). VIRTIS-M is capable of obtaining images over several visible and infrared wavelengths in two channels: visible (0.3 to 1 mm) and near infrared (1 to 5 mm) [Drossart et al., 2007]. Dayside observations in the ultraviolet show cloud features in the upper cloud (380 nm), corresponding to an altitude of
(66 ± 4) km. At the near infrared 980 nm, solar photons are more penetrating and reach the base of the upper cloud at
(61 ± 3) km [Belton et al., 1991; Peralta et al., 2007b]. Night side images of Venus at 1.74 mm are sensitive to the opacity of the lower cloud layer at
44– 48 km, with the radiance coming from altitude levels beneath the clouds and being attenuated as it passes through the lower cloud layer [Carlson et al., 1991; Crisp et al., 1991]. Sa´nchez-Lavega et al. [2008] present a detailed account of the global cloud dynamics observed in these filters and the relation between wavelength and cloud altitude determination. [8] Because of the highly elliptical orbit of the VEX spacecraft, with nearly 80°N pericenter, only the south hemisphere can be covered by the VIRTIS instrument. The spatial resolution of VIRTIS images vary from 15 km/pix at polar and subpolar latitudes to 45 km/pix at equatorial latitudes. The nightside images used in this study cover the period from 12 April 2006 (Orbit Insertion) to 9 March 2007. The dayside images cover the period from Orbit Insertion to 28 July 2007, extending during a large fraction of the nominal mission. The examined period for the visible images was extended compared to the infrared images because of the lower number of orbits containing high-resolution well-contrasted images of the dayside. From these periods, and attending to image quality and spatial resolution, we selected observations corresponding to 112 orbits in the visible and 116 orbits in the infrared. [9] Figure 1 shows maps of the number of observations for each latitude and local time with a high spatial resolution suitable for gravity waves search (spatial resolution better than 30 km/pix). This roughly corresponds to latitudes poleward of 30°S in both day and nightside. [10] The images were navigated, corrected from defects (mainly striations and hot and dark lines), processed and mapped into cylindrical and polar projections using the ‘‘Planetary Laboratory Image Analysis’’ (PLIA) software (R Hueso et al., The Planetary Laboratory for Image Analysis (PLIA), submitted to Planetary and Space Science, 2008.). Each image was processed individually by a combination of contrast enhancement, unsharp mask and smoothing filters.

3. Results [11] Examples of systems of waves observed in the cloud fields in ultraviolet (380 nm), near-infrared (980 nm) and infrared (1.74 mm) wavelengths (hereafter, UV, NIR and IR,

2 of 12

E00B18

PERALTA ET AL.: VENUS GRAVITY WAVES OBSERVED BY VEX-VIRTIS

E00B18

Figure 1. Global resolution maps and frequency of high-resolution observations made with Visible and InfraRed Thermal Imaging Spectrometer (VIRTIS) on the (a) dayside and (b) nightside of Venus. The number of observations of a specific latitude-local time with a high spatial resolution of 30 km/pix is indicated. This spatial resolution is suitable for observations of waves with wavelengths larger than
100 km. Observations with better resolutions of the order of 15 –20 km/pix were also present in the data set with much less frequency. The position of the waves encountered is shown as symbols (triangles for the waves in the upper cloud, and crosses for the waves in the lower cloud) and correlates with regions with high number of observations at this resolution. respectively) are displayed in Figure 2. The spatially distributed regular patterns of alternative bright and dark stripes represent reflectivity variations in the ultraviolet and near infrared (observed during dayside) and opacity variations in the infrared (observed during nightside). The images show trains of crests and troughs generally ordered in longitude (oriented in the zonal direction) and extending a few degrees in latitude nearly perpendicularly to the wave packet. In many cases the troughs are wider than the crests, or they simply disappear. Some of the wave packets are very regular while others have nonequidistant crests, change their latitudinal dimensions over the packet or show a slight curvature. Beside on Venus, these irregularities have also been observed in mesoscale cloud waves observed in Jupiter and interpreted in terms of ducted gravity waves [Flasar and Gierasch, 1986]. [12] We performed a systematic search of waves (see Figure 1) as those shown on Figure 2, finding 6 distinctive and separate wave packets in ultraviolet images, one wave packet in the near infrared (this wave system was observed simultaneously in the ultraviolet and the infrared with nearly identical characteristics) and 30 wave packets in infrared images. In some cases, a single infrared image presents wave systems at various latitudes. The larger abundance of waves in the infrared images results in a larger diversity of wave properties. The average image contrast variations that characterize these wave systems are on the order of 1% in the ultraviolet, less than 1% in the near infrared and vary widely in the infrared with some faint wavy structures with contrast variations of 1%, subtle to detect, to clearly observable systems with contrast of 15%. This study does not rule out the possible existence of less conspicuous wave variations in the cloud fields with less contrast or shorter wavelengths as found in the upper cloud by the VMC instrument [Markiewicz et al., 2007]. [13] We measured the observable properties of the wave packets such as their latitude and local time location, packet length (i.e., their extension perpendicular to the wavefronts),

packed width (i.e., their extension in the wavefront direction) and orientation respect to the parallels (lines of constant latitude). Results are listed in Tables 1a–1c. [14] Figure 3 shows the location of the clouds in terms of latitudes, local times and surface topography. These maps show abundant waves at the lower cloud with no correlation in their position with surface topography (Figure 3a) or local time (Figure 3b). This is in contrast to the northern hemisphere, where the Vega balloon registered the influence of the surface in atmospheric motions with oscillations at the lower cloud level interpreted as gravity waves generated by the flow over specific mountainous regions [Sagdeev et al., 1986; Young et al., 1987, 1994]. In the southern hemisphere, where the surface is nearly flat compared to the northern hemisphere, especially in the sampled latitudes, the waves seem to be present all along the nightside hemisphere except inside the South Polar Dipole [Piccioni et al., 2007]. This could be due to an observational effect since the dipole interior is particularly dark on the infrared images. In any case, gravity waves have been detected up to the dipole’s outer edge (80°S in orbit 84) and the global dynamics of the polar area will be presented elsewhere. [15] The Venus Monitoring Camera on VEX has found abundant cloud waves of higher frequency (wavelengths of a few tens of km) in the upper clouds from south tropical to north polar latitudes [Markiewicz et al., 2007]. These waves, if also present in the south hemisphere, cannot be resolved by VIRTIS observations. The scarcity of large waves detected on the upper clouds does not allow extracting a clear conclusion about their distribution but they seem clustered in the early morning hours and late afternoon. In both cloud layers, we do not find wave packets equatorward of 40°S. This is probably caused by the comparatively low number of high-resolution observations of those latitudes. However, we would also like to point out that at those latitudes there seems to be a transition to different cloud structures, with dark midlatitude bands in the upper cloud (UV), and a mottled turbulent morphology in the lower

3 of 12

PERALTA ET AL.: VENUS GRAVITY WAVES OBSERVED BY VEX-VIRTIS

E00B18

E00B18

Figure 2. Examples of cloud waves observed on VIRTIS images projected into polar maps for size comparison. Figures 2a and 2b show waves observed at 380 nm in the upper cloud level. The singular detection of a wave at 980 nm images is shown in Figure 2c and correlates with the same structure observed in 380 nm and shown in Figure 2b. Either this is a vertically extended wave, or the observed levels in those wavelengths can be closer than expected at specific latitudes. Large wave packets observed in 1.74 mm in the lower cloud are shown in Figures 2d and 2e. cloud (IR) that on the one hand may difficult the detection of gravity waves but could also represent a region of active cellular convection where gravity waves might not be generated. [16] The extension covered by the wave packets, i.e., the packet length, varies between 640 and 1200 km in the upper cloud (UV) with a mean value of 920 km and from 280 to 1700 km in the lower cloud (IR) with a mean value of 760 km. Their average width are 320 and 230 km in the in

the upper (UV) and lower cloud (IR), respectively. We further measured the wavelength of each packet, the number of crests and, when possible, their zonal motions. Since the winds at cloud level derived from VIRTIS are variable in terms of the local time [Sa´nchez-Lavega et al., 2008] we also measured the local velocity at nearby longitudes and latitudes of each individual wave packet to better estimate their phase speed relative to the local wind. This was done only on those orbits where we could observe the same cloud

Table 1a. Summary of Wave Packets Properties From UV Images by VEX-VIRTISa Orbit

Date (mm/dd/yy)

Latitude (deg)

Local Hour

Packet Length (km)

Packet Width (km)

Orientation (deg)

59 59 170 255 388

06/18/2006 06/18/2006 10/07/2006 12/31/2006 05/13/2007

41 58 62 70 74

16,4 15,7 06,8 15,0 08,1

>860 >865 1275 980 640

170 340 340 420 335

25 30 25 40 2

a

Visible and InfraRed Thermal Imaging Spectrometer (VIRTIS).

4 of 12

PERALTA ET AL.: VENUS GRAVITY WAVES OBSERVED BY VEX-VIRTIS

E00B18

Table 1b. Summary of Wave Packets Properties From NIR Images by VEX-VIRTIS

Orbit

Date (mm/dd/yy)

Latitude (deg)

Local Hour

Packet Length (km)

Packet Width (km)

Orientation (deg)

59

06/18/2006

58

15.4

>860

350

35

field with time differences large enough to track the motions (1 h). Repeated observations of the same wave features indicate they conserve their morphology and albedo variations over these timescales being nondispersive. These properties appear detailed in Tables 2a– 2c. [17] Observed wavelengths are within the range of 90 to 170 km in the upper cloud (UV) and 60 to 180 km in the lower cloud but we cannot rule out waves of higher frequencies (wavelengths shorter than 50 km) not resolved by our observations. At both cloud levels the phase velocities turn out to be generally low respect to the averaged zonal flow. This is in agreement with our interpretation in terms of gravity waves that will be later discussed. [18] The overall properties of these wave systems do not seem to vary with latitude or local time. The wavelength behavior and wave packet properties in terms of latitude appear on Figure 4. Neither the wavelength (Figure 4a), nor the wave packet length (Figure 4b) or the wave packet width (Figure 4b) seem to be correlated with latitude. Only the wave packet orientation (Figure 4d) seems to slightly grow at higher (more poleward) latitudes. The wave system detected on the outer edge of the dipole is aligned with this

E00B18

structure and presents the largest inclination with respect to the parallels. [19] The derivation of the properties detailed above critically depends on the quality of the data and, for the determination of the wavelengths and phase speeds, in the ability to accurately track specific structures in the wavefronts and outside them. We estimate the navigation error to be lower than 1 pixel but since all the measurements were done by a human operator the measurement error can be larger. We therefore estimate the wavelengths to be accurate to a 10%, packet lengths and widths accurate to a few tens of km and phase speeds to have a mean measurement error of 8 m/s. The sometimes subtle identification of wave patterns in ultraviolet images relied on the confirmation of all wave structures by a second operator.

4. Discussion [20] On the basis of the characteristics of the wave systems described above we interpret them as internal gravity waves lying at regions of positive static stability. On the one hand, the high frequency of the waves and the slow rotation of the planet discard the observed waves as manifestations of Rossby waves and on the other, these frequencies are much smaller than the acoustic cut-off frequency. Also their latitudes and short sizes discard Kelvin waves as the underlying mechanism for the regular cloud patterns. Roll convection on Earth sometimes produces structures similar in their shape to the regular alignment of clouds observed on our images but this phenomenon is characteristic of the planetary boundary layer and depends on a critical equilibrium between a small static stability and

Table 1c. Summary of Wave Packets Properties From IR Images by VEX-VIRTIS Orbit

Date (mm/dd/yy)

Latitude (deg)

Local Hour

Packet Length (km)

Packet Width (km)

Orientation (deg)

84 96 97 97 100 112 112 112 112 112 113 114 118 141 142 161 164 166 228 231 261 290 313 315 317 317 317 318 321 323

07/13/2006 07/25/2006 07/26/2006 07/26/2006 07/29/2006 08/10/2006 08/10/2006 08/10/2006 08/10/2006 08/10/2006 08/11/2006 08/12/2006 08/16/2006 09/08/2006 09/09/2006 09/28/2006 10/01/2006 10/03/2006 12/04/2006 12/07/2006 01/06/2007 02/04/2007 02/27/2007 03/01/2007 03/04/2007 03/04/2007 03/04/2007 03/05/2007 03/07/2007 03/10/2007

80 48 39 61 55 60 49 49 53 58 54 54 65 65 64 58 50 53 51 78 59 63 60 61 53 59 63 60 49 41

21,9 02,0 22,1 19,8 03,5 23,8 23,1 22,3 21,4 21,1 01,5 05,1 00,2 20,5 03,8 03,6 03,6 03,2 00,9 03,3 21,8 20,6 02,8 01,2 19,1 19,7 20,5 19,5 19,7 19,4

>600 1750 660 570 >1525 610 410 730 785 575 >1140 >645 635 520 >1415 1050 335 >790 690 280 620 >440 >860 760 >1230 685 450 640 >1300 615

225 520 160 160 270 105 170 320 200 205 180 205 265 370 160 310 380 225 370 130 240 250 315 335 170 200 145 150 210 200

50 15 35 40 2 20 40 25 8 7 17 14 2 11 13 17 20 12 10 30 13 4 11 0 10 15 7 0 12 0

5 of 12

E00B18

PERALTA ET AL.: VENUS GRAVITY WAVES OBSERVED BY VEX-VIRTIS

E00B18

Table 2a. Summary of Wave Properties From UV Images by VEX-VIRTIS Orbit 59 59 170 255 388

Latitude Number of Wave Length Crest Width cx (deg) Crests (km) (km) (m/s) 41 58 62 70 74

8 7 5 11 4

95 115 210 90 170

45 60 125 45 100

113 108 62 29 76

j~ cju (m/s) 12 40 0.5 14 36

a dynamic instability due to the vertical wind shear (low values of the Richardson number). Our wave systems are located several tens of km above the surface and at regions of high Richardson numbers. [21] In order to gain insight into the nature of the waves and further determine their characteristics, we will examine their wave dispersion relation using a simple ‘‘textbook’’ classical linear model of gravity waves. The wave dispersion relation can be deduced by neglecting the effects of rotation under the Boussinesq approximation and assuming two-dimensional motion. Linear perturbations over a basic state with constant zonal flow and density yield the dispersion relation [Holton, 1992]: ðw   uk Þ2 ¼

k2

k2 N 2; þ m2

ð1Þ

 is the where w is the frequency of the wave perturbation, u mean wind projected in the direction of the wave phase propagation, N2 is the square of the buoyancy or BruntVa¨isa¨lla¨ frequency, and k and m are the horizontal and vertical wave numbers related to the horizontal and vertical wavelengths, lx and lz, by k = 2p/lx and m = 2p/lz. From (1) it is straightforward to deduce the horizontal and vertical components of the phase speed vector relative to the mean flow as,

Figure 3. Polar map of cloud waves’ locations in terms of their longitude, local time, morphology of the clouds, and surface topography. Figure 3a shows the wave packets central positions (triangles for the waves in the upper cloud, and crosses for the waves in the lower cloud) over a surface map obtained by the Magellan IDRS instrument [Saunders et al., 1991]. The topography varies from 1.3 km in the darkest low plains to +2.5 km at the top of the highest mountains in this projection. The higher concentration of waves in the sector between longitudes 210 – 300° is due to a bias in the spatial sampling. Figure 3b shows the cloud waves in terms of latitude and local time displaying the typical cloud structure at these levels with a background image from the Venus Orbit Insertion covering the whole southern hemisphere. Given the spatial resolution and sampling of the polar region, the lack of cloud waves at these latitudes might be significant.

N cx ¼  pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 k þ m2

ð2aÞ

N k cz ¼  pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; k 2 þ m2 m

ð2bÞ

where cx is the phase speed relative to the mean wind flow. We now proceed to verify if those waves for which we have information about their phase speeds verify this simple model of internal gravity waves. [22] The clouds observed in the ultraviolet are located at
66 km altitude. These levels are close to the transition between the troposphere and the mesosphere, where the static stability of the atmosphere abruptly increases. A reasonable value of the Brunt-Va¨isa¨la¨ frequency from

Table 2b. Summary of Wave Properties From NIR Images by VEX-VIRTIS Orbit 59

6 of 12

Latitude Number of Wave Length Crest Width cx j~ cju (deg) Crests (km) (km) (m/s) (m/s) 58

7

125

75

-

-

PERALTA ET AL.: VENUS GRAVITY WAVES OBSERVED BY VEX-VIRTIS

E00B18

Table 2c. Summary of Wave Properties From IR Images by VEX-VIRTIS

Orbit

Latitude (deg)

Number of Crests

Wave Length (km)

Crest Width (km)

cx (m/s)

j~ cju (m/s)

84 96 97 97 100 112 112 112 112 112 113 114 118 141 142 161 164 166 228 231 261 290 313 315 317 317 317 318 321 323

80 48 39 61 55 60 49 49 53 58 54 54 65 65 64 58 50 53 51 78 59 63 60 61 53 59 63 60 49 41

6 12 8 8 12 8 5 7 6 4 6 5 7 4 18 14 4 6 5 5 4 4 7 6 9 9 6 5 12 5

63 155 100 80 130 60 103 90 125 105 140 100 90 180 75 75 100 130 130 105 145 120 135 100 100 90 90 100 105 110

40 70 35 30 65 30 50 20 70 45 90 40 50 75 44 40 50 60 65 65 65 66 65 55 55 50 45 50 50 55

26 64 59 58 44 61 64 45 42 60 64 59 60 66 54 50

1.0 9.0 3.1 2.7 0.2 5.1 1.8 29.7 9.0 5.0 3.1 2.4 1.1 6.1 7.5 3.3

different probes for these levels is N2 260  106 s2 [Seiff et al., 1985; Del Genio and Rossow, 1990; Gierasch et al., 1997]. Above these levels the stability of the atmosphere increases to higher values characteristic of the mesosphere. Below the clouds the stability decreases to nearly zero allowing for a region between 48 and 55 km where vertical convection can develop. The lower clouds acting as a source of opacity for the atmospheric emitted radiation are located just below, in altitudes (43 – 48 km) characterized by high values of the static stability. We consider as a representative value for the Brunt-Va¨isa¨la¨ frequency in the lower cloud N2 65  106 s2. This is close to the peak values measured by the Pioneer Venus Night Sounder at the height of 45 km [Seiff et al., 1985; Gierasch et al., 1997] over the stable layer situated in the height range 30– 48 km, and it is within the range of 50– 100  106 s2 obtained at these altitudes by the ensemble of Pioneer Venus sounders. This stable layer may act as a duct where waves can form excited by convection from the lower levels or from downdrafts from the unstable levels above [Baker et al., 1998]. [23] Figure 5 shows the behavior of the gravity wave dispersion relationship for different values of the vertical wavelength lz and the assumed values of the Brunt-Va¨isa¨la¨ frequency. The upper cloud waves (Figure 5a) seem to be distributed in two groups of short (5 km) and large (15 km) vertical wavelengths or in regions with small and large values of the atmospheric static stability. We note that the cloud wave simultaneously detected in the ultraviolet and

E00B18

near infrared images corresponds to one of the two cases of long vertical wavelengths. The lower cloud waves (Figure 5b) are more abundant and are homogeneously distributed in the plot in regions of short vertical wavelengths (5 – 10 km). This is coherent with the vertical extent of the duct and consistent with the fact that more vertically extended waves would be dispersed by the interaction with the vertical wind shear [Leroy and Ingersoll, 1995]. [24] In both cases, phase velocities relative to the mean zonal flow are low, as one would expect, for convectively driven gravity waves typically have small phase speeds relative to the mean zonal wind where they are generated. Also the observed nondispersive behavior of the waves is easily explained by Figure 5 and points to relatively small variations of the vertical wavelength and vertical stability variations across the wave packet extending several hundred kilometers. [25] Although a full nonlinear approach is beyond the scope of this paper we note that in the limit jc – uj!0 the waves encounter a critical level where linearity does not hold and the waves break in the atmosphere. The more slowly traveling waves on Venus are therefore close to the critical level (but not breaking since we are able to observe them for 1 – 2 h in the VIRTIS frames of the same atmospheric area). Notably under the WKB approximation, as the critical level is approached the vertical wavelength diminishes and finally collapses. However, the wave stalls and the limit is only reached after a time that approaches infinite [Salby, 1996]. This process is accompanied by an increase of the wave amplitude. For the IR waves in our survey it seems that the better they can be seen the slower the phase velocities they have, what might be a signature of high-amplitude waves moving close to their critical level. [26] For the lower cloud waves we can study the double dependence of the phase speed on the two unknowns N2 (constrained from measurements from previous spacecrafts) and lz (constrained by using physically reasonable arguments). This is not feasible on the upper cloud with less observations and apparently larger variations on lz or N2. Figure 6 shows a plot of the total c2 deviation of the phase speeds of the waves from the predictions by the wave dispersion relation (equation (2a)). This was calculated for different values of N2 and lz using the equation below which also considers the estimated error in phase speeds determination. 2 N  0  2  X cx  cx ðN 2 ; mÞ c N ;m ¼ ; sc i 2

ð3Þ

where the index i runs for all of the wave system measured, c0x is the measured phase speed, cx is the theoretical value from equation (2a) and sc is the variance of the measured phase speeds. [27] The contours of constant Dc2 around its minimum value can be used to determine confidence regions in the values of the varying parameters N2 and lz. A confidence region is a region within the space of parameters that contains a large percentage of the probability distribution of finding the true parameter values within it [Press at al., 1992]. If all the wave systems measured had the same values of N2 and lz, and measurement errors obeyed a

7 of 12

E00B18

PERALTA ET AL.: VENUS GRAVITY WAVES OBSERVED BY VEX-VIRTIS

E00B18

Figure 4. Wave properties in terms of latitude. Figures 4a, 4b, 4c, and 4d show the wavelength, wave packet longitudinal extent, wave packet latitudinal extent, and the wave packet orientation relative to the parallels, respectively. Gravity waves at the upper and lower cloud level are displayed as triangles and crosses, respectively. normal error distribution, the region surrounded by the c2 = 2 curves would have a 68% probability of containing the right values of N2 and lz. The more extended regions surrounded by the c2 = 5 and c2 = 10 curves would have a 90 and 95% probability of containing the real values. From Figure 6 it seems that, when fixing N2 with the value introduced previously (N2 = 65  106 s2), the confidence region with the best fit between experiment and theory limits the range of vertical wavelengths lz between 2 and 5 km (consistent with Figure 5b). This range of vertical wavelengths matches the values of 2.5 km found from temperature scintillations measured by Magellan above the middle cloud deck [Hinson and Jenkins, 1995]. These vertically short waves have vertical wave numbers m much larger than the horizontal wave numbers k and equation (2a) can be further simplified to cx 

N ; m

ð4Þ

which results in nondispersive horizontally traveling waves in accordance with the observations. [28] Convection seems the most plausible source of gravity waves in the lower atmosphere of Venus [Gierasch et al., 1997]. The possibility of internal gravity waves generated by penetrative convection has been previously considered in the terrestrial case [Stull, 1976], and it has been extensively studied in Venus through several theoretical and numerical works [Leroy and Ingersoll, 1995; Baker et al., 1998, 2000a, 2000b; Yamamoto, 2003; McGouldrick and Toon, 2008; Baker et al., 1998] which showed that compressible convection in the Venus atmosphere can lead to a significant penetration of the stable layer between 30 and 48 km both by upward convective movements from the unstable layer between 20 and 30 km, and by downward convective movements from the unstable layer between 48 and 55 km. They also pointed out the possibility that the latter convective downwellings could also affect the stable region above 55 km via convective entrainment. Baker et al. [2000b] showed that this mechanism works better

8 of 12

E00B18

PERALTA ET AL.: VENUS GRAVITY WAVES OBSERVED BY VEX-VIRTIS

E00B18

producing enhanced overshooting of the convection in regions with high vertical wind shear, which may allow the interaction of upward and downward plumes within the 30– 48 km stable layer and generate gravity waves of higher wavelengths. In that case we could anticipate a

Figure 6. Chi-square tests exploring the space parameter defined by equation (2a) on the lower cloud waves. Confidence regions corresponding to values of chi-square larger than a given value are displayed with contour lines. The horizontal gray area indicates the range of possible buoyancy frequencies at the cloud waves’ location, and the vertical gray area indicates the corresponding range of vertical wavelengths within the confidence region of the best fit (bounded by c2 = 2 lines).

certain degree of correlation between the locations of gravity waves in the upper and lower clouds. Unfortunately the scarce gravity waves observed in the upper cloud layer may not be enough to closely examine this issue. Also, since the lower clouds are observed in the nightside hemisphere and the upper cloud are observed in the dayside hemisphere, our VIRTIS observations do not cover the same locations in high resolution and close in time.

Figure 5. Internal gravity wave dispersion relation applied to the cloud waves in the (a) upper cloud and (b) lower cloud. Phase velocities are relative to the zonal wind. Lines represent solutions to equation (2a) varying the vertical wavelengths in the ranges lz = 2 – 15 km for both clouds. We used a fixed value of the buoyancy frequency representative of the two cloud levels (N2 = 260  106 s2 for the upper cloud and N2 = 65  106 s2 for the lower cloud; details are given in section 4). 9 of 12

E00B18

PERALTA ET AL.: VENUS GRAVITY WAVES OBSERVED BY VEX-VIRTIS

Figure 7. Waves’ phase velocity and mean zonal wind profiles at the two cloud levels. Wave phase velocities and the mean zonal wind profile at the upper cloud level are displayed with dark triangles and continuous line. Wave phase velocities and the mean zonal wind profile at the lower cloud level are displayed with light crosses and discontinuous line. Those waves with no dynamical information are also displayed along a discontinuous line at zero zonal velocity. The error bars in the zonal wind profiles represent measurement errors and variance of the wind at each latitude [Sa´nchez-Lavega et al., 2008]; the error bars of the wave phases represent estimated measurement error in each wave packet.

[29] Our observations, coupled to data from previous spacecraft probes that extend in time and cover ample locations (Pioneer-Venus, Venera and Vega landers, see, e.g., Gierasch et al. [1997]), suggests that the presence of the two stable layers where the clouds lie are a persistent feature of Venus’s vertical cloud structure. Our gravity wave detections in the south hemisphere from latitudes 40°S to 75°S imply that the layer sampled with infrared images favors the vertical stability. The region equatorward of 40°S is not sampled well enough to extend this conclusion and could be dominated by stronger convective motions, whereas poleward of 75° the dipole feature may mask the presence of the waves. Although abundant and recurrent, the waves observed in the lower cloud are not permanent. In view of the stability measurements obtained from the

E00B18

different previous probes we interpret this variability as a perturbation distinct to changes in the atmospheric stability and affecting the mechanisms responsible of exciting the waves. [30] The relation between wave location and wind is examined in Figure 7. Here we show the zonal wind profiles in the upper and lower cloud layers obtained by extensive analysis of the VIRTIS data set [Sa´nchez-Lavega et al., 2008] compared to the zonal phase speeds of the wave systems in both levels. The position of the waves is correlated neither with changes in the zonal profile, nor with values of the vertical wind shear. Additionally, the width of the wave packets does not correlate with the meridional shear of the wind, what may imply that the packet width could be related with meridional variations in the atmospheric static stability. [31] The significantly larger phase speeds of the waves in the upper cloud is probably a consequence of the stronger stability in this cloud level, with much higher values for the buoyancy frequency [Seiff et al., 1985; Gierasch et al., 1997]. Contrary to the lower cloud waves, which are vertically confined in a duct of static stability, these upper cloud waves are not bounded vertically. This may result in vertical propagation to levels where they are critically absorbed by the vertically varying mean flow, or dissipated [Schubert and Walterscheid, 1984]. The absorbed waves transport energy and momentum. Their role in powering the atmospheric superrotation was examined by Leroy and Ingersoll [1995] who concluded that the waves are probably not responsible for the atmospheric global winds. [32] Most of the numerical studies of gravity waves on Venus focus on waves of smaller wavelengths (no larger than 30 km, see Baker et al. [1998, 2000a, 2000b]) than those here reported (typically 100 km). While our observations cannot distinguish these high-frequency waves (our best resolution images have a pixel size of 15– 20 km) we do detect significant wave activity of larger wavelengths (100 km). McGouldrick and Toon [2008] suggest that gravity waves with greater horizontal wavelengths comparable to our observations could be generated by broad convective plumes or a large magnitude of the vertical shear of the zonal wind but we do not observe significant convection on the upper or lower clouds. The vertical wind shear of the wind from VIRTIS data is mainly confined to the 61 to 66 km altitude range with values @ < u >/@z = 8 ± 2 ms1 per km at equatorial to midlatitudes and less than 2 ms1 per km at subpolar latitudes [Sa´nchez-Lavega et al., 2008]. [33] Gravity waves transmit energy and momentum perpendicularly to the direction of phase propagation in the direction of constant phase lines (considering the traveling direction, mainly zonal, and the vertical). Most of the waves are zonally aligned but some others are tilted and may transport some momentum meridionally. In the zonal direction some of the waves move slightly faster while others slightly more slowly than the average wind. In the meridional direction some waves move slightly northward and some other slightly southward. We also do not have information about the vertical component of the wave motion. For these reasons it was difficult to evaluate the

10 of 12

E00B18

PERALTA ET AL.: VENUS GRAVITY WAVES OBSERVED BY VEX-VIRTIS

average role of the waves in the global transport of energy and momentum in the atmosphere.

5. Conclusions [34] We have presented a study of the mesoscale waves present in the atmosphere of Venus at different cloud levels by using images at three wavelengths obtained by the VIRTIS instrument onboard Venus Express. We interpret them as gravity waves, and this is the first time that mesoscale waves within three different cloud levels are characterized in Venus at the same time, these levels corresponding to wavelengths 380 nm, 980 nm, and 1.74 mm. [35] The abundant waves observed in IR images display a large diversity of morphologies and wave properties with many observations revealing wave activity at a wide range of latitudes (40°S75°S). None of the wave properties seem to vary systematically with latitude or the zonal wind profile except from an observational bias of the VIRTIS data set that is unable to observe these waves at tropical and equatorial latitudes. In particular the location of the waves seems uncorrelated with topographic features on the surface or the local time that combined with the negligible effect of the solar heating at the lower cloud could indicate a crucial role of deep convection as the source of wave activity. [36] We have presented a first account on wave properties using a classical ‘‘textbook’’ linear model of gravity waves [Holton, 1992] that confirms the nature of the waves. From our statistical analysis of the wave dependence on the static stability and vertical wavelength (Figures 5b and 6) we concluded that the vertical wavelengths are on the order of 2 to 5 km in the lower cloud and 5 to 15 km in the upper cloud from the less numerous measurements of wavelength and phase speeds (Figure 5a). These results are consistent with previous determinations of the vertical wavelengths of temperature scintillations in the middle cloud measured by the Magellan spacecraft [Hinson and Jenkins, 1995]. [37] This study opens questions about the specific source of the wave activity observed, the latitudinal and temporal distributions of atmospheric stability and convection and the role of these waves in the overall dynamics of Venus atmosphere. [38] Acknowledgments. This work has been funded by Spanish MEC AYA2006-07735 with FEDER support and Grupos Gobierno Vasco IT-464-07. J.P. acknowledges a UPV fellowship, and R.H. acknowledges a ‘‘Ramo´n y Cajal’’ contract from MEC.

References Baker, R. D., G. Schubert, and P. W. Jones (1998), Cloud-level penetrative compressible convection in the Venus atmosphere, J. Atmos. Sci., 55, 3 – 18, doi:10.1175/1520-0469 (1998)055<0003:CLPCCI>2.0.CO;2. Baker, R. D., G. Schubert, and P. W. Jones (2000a), Convectively generated inertial gravity waves in the lower atmosphere of Venus. Part I: No wind shear, J. Atmos. Sci., 57, 184 – 199, doi:10.1175/15200469(2000)057<0184:CGIGWI>2.0.CO;2. Baker, R. D., G. Schubert, and P. W. Jones (2000b), Convectively generated inertial gravity waves in the lower atmosphere of Venus. Part II: Mean wind shear and wave-mean flow interaction, J. Atmos. Sci., 57, 200 – 215, doi:10.1175/1520-0469(2000)057<0200:CGIGWI>2.0.CO;2. Belton, M. J. S., G. R. Smith, G. Schubert, and A. D. Del Genio (1976), Cloud patterns, waves and convection in the Venus atmosphere, J. Atmos. Sci., 33, 1394 – 1417, doi:10.1175/1520-0469(1976)033<1394:CPWACI> 2.0.CO;2.

E00B18

Belton, M. J. S., et al. (1991), Images from Galileo of the Venus cloud deck, Science, 253, 1531 – 1536, doi:10.1126/science.253.5027.1531. Carlson, R. W., K. H. Baines, L. W. Kamp, P. R. Weissman, W. D. Smythe, A. C. Ocampo, T. V. Johnson, D. L. Matson, J. B. Pollack, and D. Grinspoon (1991), Galileo infrared imaging spectroscopy measurements at Venus, Science, 253, 1541 – 1548, doi:10.1126/science.253.5027.1541. Counselman, C. C., S. A. Gourevitch, R. W. King, G. B. Loriot, and E. S. Ginsberg (1980), Zonal and meridional circulation of the lower atmosphere of Venus determined by radio interferometry, J. Geophys. Res., 85, 8026 – 8030, doi:10.1029/JA085iA13p08026. Crisp, D., S. McMuldroch, S. K. Stephens, W. M. Sinton, B. Ragent, K. W. Hodapp, R. G. Probst, L. R. Doyle, D. A. Allen, and J. Elias (1991), Ground-based near-infrared imaging observations of Venus during the Galileo encounter, Science, 253, 1538 – 1541, doi:10.1126/ science.253.5027.1538. Del Genio, A. D., and W. B. Rossow (1990), Planetary-scale waves and the cyclic nature of the cloud top dynamics on Venus, J. Atmos. Sci., 47, 293 – 318, doi:10.1175/1520-0469(1990)047<0293:PSWATC>2.0.CO;2. Drossart, P., et al. (2007), Scientific goals for the observation of Venus by VIRTIS on ESA/Venus express mission, Planet. Space Sci., 55, 1653 – 1672, doi:10.1016/j.pss.2007.01.003. Esposito, L. W., J. L. Bertaux, V. Krasnopolsky, V. I. Moroz, and L. V. Zasova (1997), Chemistry of lower atmosphere and clouds, in VENUS II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, pp. 415 – 458, Univ. of Ariz. Press, Tucson. Flasar, F. M., and P. J. Gierasch (1986), Mesoscale waves as a probe of Jupiter’s deep atmosphere, J. Atmos. Sci., 43, 2683 – 2707, doi:10.1175/ 1520-0469(1986)043<2683:MWAAPO>2.0.CO;2. Gierasch, P. J. (1987), Waves in the atmosphere of Venus, Nature, 328, 510 – 512, doi:10.1038/328510a0. Gierasch, P. J., et al. (1997), The general circulation of the Venus atmosphere: An assessment, in VENUS II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, pp. 459 – 500, Univ. of Ariz. Press, Tucson. Hinson, D. P., and J. M. Jenkins (1995), Magellan radio occultation measurements of atmospheric waves on Venus, Icarus, 114(2), 310 – 327, doi:10.1006/icar.1995.1064. Holton, J. R. (1992), An Introduction to Dynamic Meteorology, 535 pp., Elsevier, New York. Hou, A. Y., and B. F. Farrell (1987), Superrotation induced by critical-level absorption of gravity waves on Venus: An assessment, J. Atmos. Sci., 44, 1049 – 1061, doi:10.1175/1520-0469(1987)044<1049:SIBCLA> 2.0.CO;2. Houze, R. A. (1993), Cloud Dynamics, 573 pp., Elsevier, New York. Hunt, G. E., and J. P. Muller (1979), Voyager observations of small-scale waves in the equatorial region of the jovian atmosphere, Nature, 280, 778 – 780, doi:10.1038/280778a0. Kasprzak, W. T., A. E. Hedin, H. G. Mayr, and H. B. Niemann (1988), Wavelike perturbations observed in the neutral thermosphere of Venus, J. Geophys. Res., 93, 11,237 – 11,245, doi:10.1029/JA093iA10p11237. Kliore, A. J., and I. R. Patel (1980), Vertical structure of the atmosphere of Venus from Pioneer Venus orbiter radio occultations, J. Geophys. Res., 85, 7957 – 7962, doi:10.1029/JA085iA13p07957. Kolosov, M. A., O. I. Yakovlev, A. I. Efimov, S. S. Matyugov, T. S. Timofeeva, E. V. Chub, A. G. Pavelyev, A. I. Kucheryavenkov, I. E. Kalashnikov, and O. E. Milekhin (1980), Investigation of the Venus atmosphere and surface by the method of radiosounding using VENERA-9 and 10 satellites, Acta Astronaut., 7, 219 – 234, doi:10.1016/ 0094-5765(80)90062-4. Leroy, S. S., and A. P. Ingersoll (1995), Convective generation of gravity waves in Venus’s atmosphere: Gravity wave spectrum and momentum transport, J. Atmos. Sci., 52, 3717 – 3737, doi:10.1175/1520-0469(1995) 052<3717:CGOGWI>2.0.CO;2. Leroy, S. S., and A. P. Ingersoll (1996), Radio scintillations in Venus’s atmosphere: Application of a theory of gravity wave generation, J. Atmos. Sci., 53, 1018 – 1028, doi:10.1175/1520-0469(1996)053<1018:RSIVAA> 2.0.CO;2. Markiewicz, W. J., D. V. Titov, S. S. Limaye, H. U. Keller, N. Ignatiev, R. Jaumann, N. Thomas, H. Michalik, R. Moissl, and P. Russo (2007), Morphology and dynamics of the upper cloud layer of Venus, Nature, 450, 633 – 636, doi:10.1038/nature06320. McGouldrick, K., and O. B. Toon (2008), Observable effects of convection and gravity waves on the Venus condensational cloud, Planet. Space Sci., 56, 1112 – 1131, doi:10.1016/j.pss.2008.02.010. Niemann, H. B., W. T. Kasprzak, A. E. Hedin, D. M. Hunten, and N. W. Spencer (1980), Mass spectrometric measurements of the neutral gas composition of the thermosphere and exosphere of Venus, J. Geophys. Res., 85, 7817 – 7827, doi:10.1029/JA085iA13p07817. Peralta, J., R. Hueso, and A. Sa´nchez-Lavega (2007a), Cloud brightness distribution and turbulence in Venus using Galileo violet images, Icarus, 188, 305 – 314, doi:10.1016/j.icarus.2006.12.005.

11 of 12

E00B18

PERALTA ET AL.: VENUS GRAVITY WAVES OBSERVED BY VEX-VIRTIS

Peralta, J., R. Hueso, and A. Sa´nchez-Lavega (2007b), A reanalysis of Venus winds at two cloud levels from Galileo SSI images, Icarus, 190, 469 – 477, doi:10.1016/j.icarus.2007.03.028. Piccioni, G., et al. (2007), South-polar features on Venus similar to those near the north pole, Nature, 450, 637 – 640, doi:10.1038/nature06209. Press, W. H., S. A. Teukolsky, W. T. Vetterling, and B. P. Flannery (1992), Numerical Recipes in Fortran 77: The Art of Scientific Computing, pp. 687 – 693, Cambridge Univ. Press, New York. Reuter, D. C., et al. (2007), Jupiter cloud composition, stratification, convection and wave motion: A view from New Horizons, Science, 318, 223 – 225, doi:10.1126/science.1147618. Rossow, W. B., et al. (1980), Cloud morphology and motions from Pioneer Venus images, J. Geophys. Res., 85, 8107 – 8128, doi:10.1029/ JA085iA13p08107. Sagdeev, R. Z., V. M. Linkin, J. E. Blamont, and R. A. Preston (1986), The VEGA Venus balloon experiment, Nature, 328, 510 – 512. Salby, M. L. (1996), Fundamentals of Atmospheric Physics, 627 pp., Elsevier, New York. Sa´nchez-Lavega, A., et al. (2008), Variable winds on Venus mapped in three dimensions, Geophys. Res. Lett., 35, L13204, doi:10.1029/ 2008GL033817. Saunders, R. S., R. E. Arvidson, J. W. Head III, G. G. Schaber, E. R. Stofan, and S. C. Solomon (1991), An overview of Venus geology, Science, 252, 249 – 252, doi:10.1126/science.252.5003.249. Schubert, G. (1983), General circulation and the dynamical state of the Venus, in VENUS, pp. 681 – 765, Univ. of Ariz. Press, Tucson. Schubert, G., and R. L. Walterscheid (1984), Propagation of small-scale acoustic-gravity waves in the Venus atmosphere, J. Atmos. Sci., 41, 1202 – 1213. Seiff, A., D. B. Kirk, R. E. Young, R. C. Blanchard, J. T. Findlay, G. M. Kelly, and S. C. Sommer (1980), Measurements of thermal structure and thermal contrasts in the atmosphere of Venus and related dynamical observations: Results from the four Pioneer Venus probes, J. Geophys. Res., 85, 7903 – 7933, doi:10.1029/JA085iA13p07903. Seiff, A., J. T. Schofield, A. J. Kliore, F. W. Taylor, S. S. Limaye, H. E. Revercomb, L. A. Sromovsky, V. V. Kerzhanovich, V. I. Moroz, and M. Y. Marov (1985), Models of the structure of the atmosphere of

E00B18

Venus from the surface to 100 kilometers altitude, Adv. Space Res., 5(11), 3 – 58, doi:10.1016/0273-1177(85)90197-8. Stull, R. B. (1976), Internal gravity waves generated by penetrative convection, J. Atmos. Sci., 33, 1279 – 1286, doi:10.1175/15200469(1976)033<1279:IGWGBP>2.0.CO;2. Svedhem, H., et al. (2007), Venus Express: The first European mission to Venus, Planet. Space Sci., 55, 1636 – 1652, doi:10.1016/j.pss.2007. 01.013. Taylor, F. W., et al. (1980), Structure and meteorology of the middle atmosphere of Venus: Infrared remote sensing From the Pioneer Orbiter, J. Geophys. Res., 85, 7963 – 8006, doi:10.1029/JA085iA13p07963. Toigo, A., P. J. Gierasch, and M. D. Smith (1994), High resolution cloud feature tracking on Venus by Galileo, Icarus, 109, 318 – 336, doi:10.1006/icar.1994.1097. Yamamoto, M. (2003), Gravity waves and convection cells resulting from feedback heating of Venus’ lower clouds, J. Meteorol. Soc. Jpn., 81, 885 – 892, doi:10.2151/jmsj.81.885. Young, R. E., R. L. Walterscheid, G. Schubert, A. Seiff, V. M. Linkin, and A. N. Lipatov (1987), Characteristics of gravity waves generated by surface topography on Venus: Comparison with the VEGA balloon results, J. Atmos. Sci., 44, 2628 – 2639, doi:10.1175/15200469(1987)044<2628:COGWGB>2.0.CO;2. Young, R. E., R. L. Walterscheid, G. Schubert, L. Pfister, H. Houben, and D. L. Bindschadler (1994), Characteristics of finite amplitude stationary gravity waves in the atmosphere of Venus, J. Atmos. Sci., 51, 1857 – 1875, doi:10.1175/1520-0469(1994)051<1857:COFASG>2.0.CO;2. 

P. Drossart, Observatoire de Paris, LESIA, 5 place Jules Janssen, F-92195 Meudon, France. R. Hueso, J. Peralta, and A. Sa´nchez-Lavega, Departamento de Fı´sica Aplicada I, Escuela Te´cnica Superior de Ingenierı´os, Universidad del Paı´s Vasco, Alameda Urquijo s/n, E-48013, Bilbao, Spain. (javier.peralta@ ehu.es) G. Piccioni and O. Lanciano, Istituto di Astrofisica Spaziale e Fisica Cosmica, INAF, via del Fosso del Cavaliere 100, I-00133 Rome, Italy.

12 of 12

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, E00B14, doi:10.1029/2008JE003113, 2008

for

Full Article

Venus Express/VIRTIS observations of middle and lower cloud variability and implications for dynamics K. McGouldrick,1 K. H. Baines,2 T. W. Momary,2 and D. H. Grinspoon1 Received 1 February 2008; revised 27 June 2008; accepted 20 August 2008; published 18 November 2008.

[1] We present an analysis of Venus Express Visible and Infrared Thermal Imaging

Spectrometer (VIRTIS) data, carried out to characterize the morphological, geographical, and evolutionary trends of the middle and lower cloud features that are observed in the atmosphere of Venus as variations in brightness temperatures in specific near-infrared wavelengths. In this preliminary study, we analyze only data collected over the span of 11 orbits. The mean radiance as a function of latitude is consistent with previous ground-based observations, indicating that the overall global distribution of mean cloud cover is stable, at least on a 10- to 20-year time scale. In contrast with the consistent level of radiance at high latitudes, a significant amount of variability to the radiance exists at lower latitudes, consistent with significant convective activity in the lower and middle cloud decks. The morphology of the holes tends from highly variable orientations of features with aspect ratios of nearly one at low latitudes, to very large aspect ratios and zonally oriented features at higher latitudes. The peak radiance of the holes appears not to demonstrate a latitudinal tendency. There is evidence of more variability to the morphology and radiance of features at lower latitudes. To investigate the evolution of the holes, we examine a sequence of images taken over a 5 h span of a single orbit. If this limited amount of data is representative, then the typical e-folding time scale for the evolution of a hole is about 1 day. Citation: McGouldrick, K., K. H. Baines, T. W. Momary, and D. H. Grinspoon (2008), Venus Express/VIRTIS observations of middle and lower cloud variability and implications for dynamics, J. Geophys. Res., 113, E00B14, doi:10.1029/2008JE003113.

1. Introduction [2] Holes in the middle and lower cloud decks of Venus were first discovered as spatial inhomogeneities in the nearinfrared brightness temperatures [Allen and Crawford, 1984]. Thermal emission from the lower atmosphere and surface of Venus escapes through narrow spectral windows among the near-infrared absorption bands of carbon dioxide and water vapor. This emission is scattered by the sulfuric acid cloud decks. From tracking of the features as they traversed the planet and detailed radiative transfer analysis of the spectra of those features, it was recognized that they represented variations in the lower and middle cloud opacity [Crisp et al., 1989]. [3] The clouds are a defining characteristic of the atmosphere of Venus. They completely enshroud the planet between the altitudes of about 50 and 70 km. Consequently, except for the variations noted in the previous paragraph, the clouds render the lower atmosphere difficult to observe in infrared, visible, and ultraviolet wavelengths. The clouds of Venus have been implicated as playing a significant role 1 Department of Space Sciences, Denver Museum of Nature and Science, Denver, Colorado, USA. 2 Jet Propulsion Laboratory, California Institute of Technology, Pasadena, California, USA.

Copyright 2008 by the American Geophysical Union. 0148-0227/08/2008JE003113$09.00

in the maintenance of the Venus greenhouse effect [Pollack et al., 1980]. Thus, breaks in the clouds, which allow a greater flux of radiation to escape (and absorb a smaller amount of solar radiation), can have a significant effect on the thermal balance of Venus. This can have significant ramifications, especially in terms of the long-term climate evolution of Venus [Crisp, 1989; Hashimoto and Abe, 2000; Bullock and Grinspoon, 2001]. [4] The atmosphere of Venus exhibits a global superrotation at most altitudes. In order to sustain the global superrotation, there must be vertical transport of horizontal momentum to the upper atmosphere, specifically between about 40 and 80 km. Vertical transport of momentum by eddies and by waves has been suggested [Schubert et al., 1980]. However, both the eddies and the waves are going to interact in some way with the clouds that occupy the altitudes between 50 and 70 km. Most likely, vertically traveling waves and eddies will be dissipated by the regions of instability in the vicinity of the clouds. The radiative dynamical feedback that supports the cloud [Pollack et al., 1980; McGouldrick and Toon, 2007], maintains a region of dynamic instability in the middle and lower cloud region, which can limit the efficiency of vertical momentum transport by wave propagation. The shear profile and the static stability profile, both of which affect the vertical transport of energy that is brought about by convection, will have the largest effect on what happens to these vertically propagating waves and eddies. Furthermore, the morphological appear-

E00B14

1 of 12

E00B14

MCGOULDRICK ET AL.: VIRTIS OBSERVATIONS OF VENUS CLOUDS

ance of cloud features (aspect ratio and ‘‘tilt’’ relative to the zonal direction) can offer information on the nature of these atmospheric characteristics (shear profile, static stability profile, and the amount of convective overturning). Thus, the morphology of the clouds can provide insight to the dynamical environment; hence, the efficiency of vertical momentum transport by these processes. [5] For these reasons, an understanding of what sustains the clouds, and what sorts of changes are responsible for the formation of holes in those clouds, is necessary in order to understand the Venus atmosphere as a whole. In this paper, we compare the results and predictions of several recent theoretical studies of the clouds of Venus, to recent observations made by the Visible and Infrared Thermal Imaging Spectrometer (VIRTIS) on Venus Express. We analyze the possible existence of latitudinal tendencies of cloud morphology and overall cloud opacity, as well as the typical cloud feature lifetimes. 1.1. Previous Observations of the Clouds and Holes [6] Crisp et al. [1989] showed that the near-infrared features discovered by Allen and Crawford [1984] were caused by variations in the opacity of the middle and lower cloud decks. The observations by Crisp et al. [1989] also provided an early general characterization of the morphology and distribution of the cloud features at wavelengths of 1.74 mm and 2.3 mm in the near infrared. Crisp et al. [1989] observed an apparent latitudinal dependence of the overall cloud opacity, in which the opacity was greatest at high latitudes, and a relatively featureless bright band associated with diminished cloud opacity that often existed at mid latitudes (around 40° to 60°). Crisp et al. [1989] also noted occasional significant asymmetry between the brightness (hence, opacity) of the northern and southern hemispheres. [7] The flyby of the Galileo spacecraft in February of 1990 provided an excellent opportunity for the study of these near-infrared cloud features. Belton et al. [1991] observed the clouds of Venus with the Galileo solid-state imaging (SSI) camera. It is important to note, however, that the features observed by SSI likely are due to cloud variability at a higher altitude (i.e., about 60 to 65 km) than the cloud variability observed in the near-infrared spectral windows on the nightside (i.e., around 50 to 55 km). This is partly because the dayside observations detected solar infrared radiation that has been reflected from near the tops of the clouds, whereas the nightside observations detected surface-emitted infrared radiation transmitted through the clouds from below. Gaseous absorption by the atmosphere extinguishes the emitted or reflected radiation at much higher altitudes outside of the narrow spectral windows. Thus, the broadband observation exhibits contributions from a greater range of altitudes (mostly higher) than does the narrow-band observation. Nevertheless, Belton et al. [1991] noted that cloud features tended to appear ‘‘patchy’’ at low latitudes and ‘‘streaky’’ at high latitudes. Viewing the cloudy atmosphere at spatial resolutions as fine as 17 km per pixel, Belton et al. [1991] detected changes in cloud morphology on time scales only in excess of an hour or two; generally on the order of 1 day. They also noted the presence of ‘‘bright-rimmed cellular features,’’ suggestive of cellular convective elements that had lifetimes in excess of a day, and sizes on the order of a few hundred kilometers.

E00B14

Blamont et al. [1986] also noted a tendency of cloud features at higher latitudes to tilt relative to the zonal direction in a way that suggested a spiral pattern. Such a spiral (i.e., the polar vortex) is known to exist over both the north and south poles of Venus [Taylor et al., 1979; Piccioni et al., 2007]. The latitudinal extent of the Piccioni et al. [2007] observations is only poleward of about 50° latitude. Analysis of the tilting of cloud features in images centered over middle and lower latitudes can place constraints on the equatorward influence of these polar vortices. [8] Carlson et al. [1991], observing the nightside of Venus with the Galileo Near-Infrared Imaging Spectrometer (NIMS), noted that contrast between the brightest and darkest regions of the nightside of Venus at 1.74 mm was about 5:1. Carlson et al. [1991] also noted the existence of linear features whose tilt relative to the parallels of latitude implied meridional transport, presumably by Hadley circulation. These features seem to be similar in appearance to those described by Blamont et al. [1986] and Belton et al. [1991]; but Carlson et al. [1991] noted that establishment of a direct correlation of features between the two instruments (NIMS and SSI) was uncertain. [9] While the Galileo Near-Infrared Mapping Spectrometer (NIMS) was able to obtain only, essentially, a snapshot of the planet during the flyby, ground-based observations by Crisp et al. [1991a], spanning the weeks leading up to and following the flyby, were able to provide some context for the NIMS observations. Crisp et al. [1991a] detected a latitudinal distribution of cloud opacity similar to that observed by Crisp et al. [1989]. That is, they noted that high-latitude regions tended to be ‘‘dark and featureless,’’ whereas midlatitude regions tended to be ‘‘occupied by bright quasi-zonal bands.’’ In addition, low latitudes were noted to exhibit the most variability, both in terms of feature opacity and size. Crisp et al. [1991a] also noted the existence of a very large (hemisphere-scale) dark region that had persisted throughout the duration of the observing sessions (about 7 weeks). Chanover et al. [1998] also noted the existence of such a hemispherical asymmetry. Finally, Crisp et al. [1991a] noted that the features that were visible to Galileo during the flyby were of a relatively small scale that tended not to persist for more than about a week (i.e., about one atmospheric rotational period). [10] Markiewicz et al. [2007b] described observations of Venus cloud structure observed with the Venus Monitoring Camera (VMC) [Markiewicz et al., 2007a] on board Venus Express. While the infrared channels of VMC can be used to probe the middle and lower clouds of Venus, the observations reported by Markiewicz et al. [2007b] are mostly in the ultraviolet channel. Thus, their observations are of cloud features located largely above 65 km altitude. They observe the polar vortex as a dark spiral structure in the ultraviolet images. They observe a mottled cloud structure at low latitudes, which they attribute to vigorous convection in the upper cloud deck, with the sizes of typical convective cells being about 30 km. They also report waves with wavelengths on the order of 10 to 30 km in the upper cloud region. 1.2. Previous Simulations of the Clouds and Holes [11] Simulations suggest that the global dynamics of Venus have a significant effect on the formation and distribution of the Venus clouds and their holes. Imamura

2 of 12

E00B14

MCGOULDRICK ET AL.: VIRTIS OBSERVATIONS OF VENUS CLOUDS

and Hashimoto [1998] showed that the Hadley circulation could result in enhancements of cloud opacity in the vicinity of both the ascending branch and the descending branch of the near-global Hadley cell. McGouldrick and Toon [2008b] showed that differences in the structure of the vertical shear of the zonal winds could have an effect on the evolution of large (
2000 km) holes in the clouds of Venus. McGouldrick and Toon [2008a] showed that convective cells with typical size and separation of about 30 km tended to generate holes with lifetimes on the order of an hour or two, and with optical depths that were consistent with observations. McGouldrick and Toon [2008a] also showed that gravity waves, with characteristics similar to those of waves putatively launched by convection that have been both observed [Hinson and Jenkins, 1995] and simulated [Baker et al., 2000], can generate wave-like features in the clouds with
10 to
30 km wavelengths, visible in the emitted 1.74 mm radiation. [12] Imamura and Hashimoto [1998] suggested that the Hadley circulation determines the latitudinal variation in the middle and lower cloud deck opacity. Upwelling in the ascending branch near the equator leads to enhanced opacity because of the influx of sulfuric acid vapor into the cloud region. Meridional transport toward the descending branch of the Hadley circulation, which they reasoned to occur poleward of 55° latitude, also leads to enhanced cloud opacity because of an accumulation of upper cloud mass that has been transported poleward by the upper level Hadley circulation (because the time scale for the sedimentation of the upper cloud particles was considerably longer than the time scale for poleward transport of those particles by the Hadley circulation). Furthermore, the bright region between about 40° and 60° latitude could be explained as an absence of these cloud-thickening mechanisms. They also found that if the Hadley circulation were slower, then the contrast between the brightness at midlatitudes and that at high latitudes would be less severe, as a result of a thinning of the upper cloud at high latitudes. If the Hadley circulation were less efficient, then the time for the upper level flow to travel from equator to pole no longer would be large compared with sedimentation times of the upper cloud particles. Hence, less photochemically produced mass is transported to high latitudes. They also found that the bright band was located more equatorward in the case of a less efficient Hadley circulation. Hence, it may be possible to determine the magnitude of the Hadley circulation by determining the latitudinal location of the high-latitude peak in total cloud opacity [Imamura and Hashimoto, 1998]. [13] The middle and lower cloud decks of Venus are sustained by a radiative dynamical feedback, whereby heating of the cloud base by the very warm deep atmosphere drives an unstable lapse rate within the region occupied by the middle and lower cloud decks [McGouldrick and Toon, 2007]. Furthermore, the time scales for growth and for dynamical transport (by winds, and by convection, when present) are much shorter than the radiative time scale. Hence, in the simulations by McGouldrick and Toon [2007], the structure of the clouds was dominated by brief, intense periods of vertical transport (i.e., convection), separated by longer periods of quiescence during which the thermal instability developed.

E00B14

[14] McGouldrick and Toon [2008b] suggested that the nature of the vertical wind profile affects the longevity and possibly the morphology of large holes (
2000 km) in the clouds. Specifically, the vertical shear of the zonal wind, coupled with the radiative dynamical feedback, can effectively dissipate large holes in the clouds. Furthermore, the nature of the vertical shear profile affected the time required to smooth out a simulated hole in the clouds. In simulations exhibiting negligible vertical shear throughout the middle cloud deck (as have been observed at low latitudes and midlatitudes by Pioneer Venus [Counselman et al., 1980] and measured via multispectral wind tracking with VIRTIS on board Venus Express [Sa´nchez-Lavega et al., 2008]), the lifetimes of holes were longer than in simulations in which the shear was at a constant magnitude throughout the entire cloud domain (i.e., shear similar to what has been observed at higher latitudes, especially by the Pioneer Venus north probe). Although the radiative dynamical feedback was able to limit the zonal ‘‘stretching’’ of holes in the simulations by McGouldrick and Toon [2008b] that exhibited minimal shear within the middle and lower cloud decks, the holes were stretched out across large zonal distances in simulations that exhibited greater amounts of shear. Such differences in the aspect ratios of features in the Venus clouds observed by VIRTIS could indicate the vertical structure of the zonal winds in the region of the middle and lower cloud decks of Venus. That is, holes that become highly zonally elongated may be in regions of high vertical shear of the zonal wind; and holes that maintain a more blocky appearance may be in regions of low vertical shear of the zonal wind. [15] An alternative cause for this zonal stretching of cloud features could be horizontal shear of the zonal wind. However, observations by both the Venus Monitoring Camera and VIRTIS on Venus Express indicate that there is negligible meridional shear of the zonal wind equatorward of about 50° latitude [Markiewicz et al., 2007b; Sa´nchez-Lavega et al., 2008]. Thus, at latitudes equatorward of 50°, the only contribution to stretching by the meridional shear of the zonal wind arises from the change in planetary circumference with latitude (f): @w 86400 u tan f ¼ ; @f RV cos f

ð1Þ

where RV = 6100  103 m is the radius of the planet at the altitude of the clouds, w is the angular velocity of the atmosphere in units of ° longitude per day, and the factor 86400 is needed to convert the units of the RHS to day1. For f  50°, and u = 60 m s1 from Sa´nchez-Lavega et al. [2008], this rate does not exceed 1.6° longitude per day per degree of latitude. Thus, among features at midlatitudes to low latitudes, only those holes that persist for several days, can be stretched significantly by the meridional shear. Furthermore, the stretching of the cloud features cannot be the result of zonal shear of the zonal wind. Such a situation would result in horizontal convergence or divergence, which would force vertical motion of atmospheric parcels. Such vertical motion can create significant changes in cloud opacity [McGouldrick and Toon, 2007, 2008a], but will not cause a zonal elongation of the features. [16] McGouldrick and Toon [2008a] analyzed the effect of gravity waves and convection cells on the appearance of

3 of 12

MCGOULDRICK ET AL.: VIRTIS OBSERVATIONS OF VENUS CLOUDS

E00B14

Table 1. Data Analyzed in Morphological Analyses File Name

Orbit

VI0363_00 VI0364_00 VI0365_00 VI0366_00 VI0367_00 VI0368_00 VI0369_00 VI0370_00 VI0371_00 VI0372_00

363 364 365 366 367 368 369 370 371 372

Start Date 18 19 20 21 22 23 24 25 26 27

Apr Apr Apr Apr Apr Apr Apr Apr Apr Apr

2007 2007 2007 2007 2007 2007 2007 2007 2007 2007

Start Timea

Exposure Timeb (s)

Distance (km)

2017:56 2017:10 2016:25 2015:42 2014:57 2014:12 2013:25 2012:41 2011:55 2011:10

0.36 3.3 0.36 3.3 0.36 3.3 0.36 3.3 0.36 3.3

67021.2 66775.9 66468.1 66226.3 67209.0 67226.6 67218.5 67223.6 67203.3 67220.6

a

Times and dates are UTC. b Exposure time is per slice.

the clouds of Venus. They found, from a simple kinematical model used to drive the microphysical model utilized by McGouldrick and Toon [2007], that convective cells with sizes on the order of 30 km and vertical velocities of ±2 m s1 produce optical depth variations comparable in magnitude to those that have been observed in the clouds of Venus [Grinspoon et al., 1993], and which experience lifetimes on the order of hours. These simulated convective cells are consistent with those that have been observed with the Vega balloons [Linkin et al., 1986], and that have been generated in convective simulations by Baker et al. [2000]. McGouldrick and Toon [2008a] also found that gravity waves, such as those found in the simulations of Baker et al. [2000], and in the occultation observations by Hinson and Jenkins [1995], would lead to the appearance of wave-like structures in the lower clouds of Venus, with wavelengths comparable to the size of the convective cells. 1.3. VIRTIS [17] The VIRTIS-M IR channel is a medium (spatial) resolution (0.25 mrad) infrared mapping spectrometer that covers the range of wavelengths from 1 mm to 5 mm, with a spectral resolution of roughly l/Dl
250. This is sufficient spectral resolution to resolve the spectral windows [Drossart et al., 2007]; and the spatial resolution allows for the detection of features as small as about 30 km when the spacecraft is at apoapse (d
66,000 km). When the spacecraft is within about 10,000 km of the planet during the course of its highly eccentric orbit, it is moving too quickly for VIRTIS to operate very effectively in its mapping modes [Drossart et al., 2007]. Thus, all of the data we analyze in this paper are obtained at spacecraft distances greater than
10,000 km. [18] With clever pointing over the course of multiple orbits, a significant latitudinal and longitudinal coverage of the clouds of the southern hemisphere may be built up, which allows for the determination of possible latitudinal tendencies in the location of and optical thickness of the clouds. In the course of a single orbit, it is possible to observe several hours worth of cloud evolution (up to about 6 h) of a small region of the Venus atmosphere (defined by the field of view common to all of the images in the series: the maximum possible region being the total field of view of the image taken when the spacecraft is closest to the planet). Analysis of longer-term evolution is possible, but requires a repointing of the spacecraft on successive orbits, since the

E00B14

previously observed region of the atmosphere will have rotated out of the relatively narrow field of view by the time the spacecraft next arrives in position for observation by the mapping modes of VIRTIS. Even with successive repointings that follow the rotation of the atmosphere, significant changes in cloud morphology during the time that VIRTIS-M is unable to observe the planet frequently make difficult the recovery of features common to observations on successive orbits.

2. Observational Goals and Methods [19] Specific latitudinal variations in overall cloud opacity have been predicted to be a consequence of the Hadley circulation [Imamura and Hashimoto, 1998], and have been recorded by ground-based observations [Crisp et al., 1991a]. Does a large-scale latitudinal dependence of cloud cover exist in the VIRTIS data, as suggested by previous ground-based observations? Are there significant spatial or temporal variations in the latitudinal distribution of cloud cover? Tendencies in cloud shape (aspect ratio) and orientation (angle of ‘‘long axis,’’ relative to the zonal direction) can indicate differences in the shear structure or in the level of convection that is occurring. Does a latitudinal tendency to cloud shape or orientation exist in the VIRTIS data? Since the clouds represent a significant contributor to the greenhouse effect on Venus, their variability can alter the thermal balance of the planet. How variable are the clouds, and how does cloud variability affect the thermal balance? We analyze a subset of the VIRTIS-M IR observations to attempt to answer these questions concerning the interactions between the atmospheric dynamics and the cloud structure of Venus. [20] In all cases, we perform our analyses using the 1.74 mm image of the VIRTIS-M IR data. We choose this wavelength band over those in the other spectral windows because the 1.74 mm band possesses the best combination of high peak radiance and sufficient contrast between cloudy and clear observations; thus, the signal-to-noise ratio of both the brighter holes and the darker clouds is greater than it is for any other spectral image. Furthermore, while the images in the 2.3 mm window region demonstrate a greater overall brightness contrast (around 20:1 in the Galileo NIMS data, compared with a roughly 5:1 ratio at 1.74 mm [Carlson et al., 1991]), the radiance in the cloudiest regions, is barely distinguishable (and possibly indistinguishable) from the background noise. In order to analyze the latitudinal distribution of cloud cover and cloud morphology on Venus, we use images/cubes taken near the apoapse (
66,000 km) of several successive orbits (orbits 363 through 372), as summarized in Table 1. We use images obtained near apoapse in order to maximize our spatial coverage of the planet, with only a small loss in spatial resolution. The data are calibrated using the VIRTIS PDS/IDL software library, as described by Drossart et al. [2007]. [21] Figure 1a demonstrates the overall coverage of the middle atmosphere by these selected observations. In generating this plot, we assume that the atmosphere rotates at a solid body rate of once per 5.5 days (a speed of about 80 m s1 at the equator, or about 60 m s1 at 40° latitude), and shift the geographic longitudes of each observation to the east, according to the time separation of each image from

4 of 12

E00B14

MCGOULDRICK ET AL.: VIRTIS OBSERVATIONS OF VENUS CLOUDS

E00B14

[22] Figure 1b shows the frequency of observations by location in the atmosphere. We see that among these images there is significant overlap, such that about 35% of the atmosphere between +10° latitude and 90° latitude has been imaged at least twice; and some regions (such as the one centered near 50° longitude and 40° latitude) have been observed three times or more. Such repeated coverage of the same part of the atmosphere gives hope to the possibility of ultimately extending the analysis of the variation in feature morphology and evolution from time scales no greater than
6 h (the extent of observation possible during a single orbit) to time scales on the order of days.

Figure 1. Distribution and frequency of coverage of the atmosphere by the images analyzed in this paper. (a) The regions of the atmosphere that have been imaged at least once over the course of the considered images (crosshatched regions). (b) The frequency of coverage of the atmosphere. The numbered contours indicate the number of images among those considered here that view the same region of the atmosphere. Both plots assume the atmosphere rotates as a solid body with a period of 5.5 days.

the first image (orbit 363). We choose a planet-circling period of 5.5 days for the clouds because that period is consistent with previous observations of lower and middle cloud tracking [Allen and Crawford, 1984; Crisp et al., 1991b]. According to the recent analysis of wind speeds from VIRTIS data by Sa´nchez-Lavega et al. [2008], a zonal wind speed of about 80 m s1 corresponds to an altitude between 58 and 70 km, for latitudes equatorward of 50°. A 5.5 day period at 40° latitude is consistent with the 60 m s1 winds reported by Sa´nchez-Lavega et al. [2008] at altitudes between 46 and 61 km. Though it is an oversimplification to assume solid body rotation, doing so allows us to obtain at least a first-order estimate of the atmospheric coverage of the observations. It also helps to facilitate the identification of features from image to image, despite their sometimes significant evolution (see section 2). Figure 1a shows that the images that we consider for this particular analysis provide us with coverage of a significant fraction of the southern hemisphere atmosphere (about 72% coverage of the area between the latitudes of +10° and 90°, with most of the missing regions lying poleward of 70°). This extensive coverage of the atmosphere allows us to have confidence in the conclusions that we draw regarding the latitudinal distributions of the features. If these images did not cover so much of the planet longitudinally, our analyses might have been blind to hemispherical variations such as those that have been reported by Crisp et al. [1989, 1991b].

2.1. Latitudinal Distribution of Cloud Cover and Hole Morphology [23] Crisp et al. [1989, 1991b] noted the appearance of a significant latitudinal distribution of cloud cover in which midlatitudes tended to be brighter, and higher latitudes tended to be darker. Imamura and Hashimoto [1998] provided a theoretical explanation of such a latitudinal distribution of cloud opacity, which, they determined, is caused by the action of the Hadley circulation. Here again, the tremendous volume of the VIRTIS data can play a significant role in confirming or refuting the existence and consistency of such a distribution of cloud opacity. Additionally, the VIRTIS data can be used to more finely tune such simulations of the Hadley circulation on Venus. To analyze the latitudinal distribution of cloud cover, we remap the VIRTIS data to a cylindrical projection and correct for the emission angle by applying the correction used by Carlson et al. [1993]. We then find the mean and standard deviation of the radiance as a function of latitude for each image. We also calculate a longer-term average latitudinal distribution of radiance, in which we average over all 10 images considered here. [24] Belton et al. [1991] suggested that there was a trend in the shapes of the features as a function of latitude. Specifically, the features at lower latitudes appeared ‘‘blocky,’’ whereas those at higher latitudes appeared streaky. The great volume of VIRTIS data from Venus Express can indicate whether these trends noted by Belton et al. [1991] are real and/or persistent. We investigate these possible morphological tendencies of the clouds by identifying features in each image, and determining a longitudinal versus transverse aspect ratio for each feature. The features are identified by using the latitudinal profile of the mean radiance of each image, as described in section 2.1. We subdivide the image into latitude bands that are no greater than 15° in extent. We find the mean and the standard deviation of the radiance over those latitudes in that image, and draw contours of the mean and the mean plus one standard deviation onto the image (Figure 2). A feature is identified if the mean plus standard deviation contour is closed (or nearly so) and convex. At higher latitudes, where the longitudinal variation is smaller and the latitudinal variation is greater, we use narrower bands to determine the defining contours. We determine the longitudinal versus transverse aspect ratio by measuring the distances along the long and short axes of each feature. The feature is assigned latitude and longitude coordinates that identify its left (L), right (R), top (T), and bottom (B) extent. The aspect ratio is then the ratio of the LR arc to the TB arc. Thus, a large aspect ratio indicates a highly elongated feature,

5 of 12

MCGOULDRICK ET AL.: VIRTIS OBSERVATIONS OF VENUS CLOUDS

E00B14

E00B14

Figure 2. (right) 1.74 mm VIRTIS image VI0383_00. (left) The blowup highlights and zooms in on the boxed region of the VIRTIS image. The parameters measured in the course of the morphological analyses are shown in the zoomed image at left. whereas an aspect ratio closer to one indicates a more blocky feature. [25] We also accumulate information on the tilt of the features relative to the parallels of latitude. The tilt angle is taken to be the angle that the LR arc of the feature makes relative to the parallels of latitude. Depending upon the magnitude of the meridional and zonal winds, a tilt measured from the parallels of latitude can indicate meridional flow or shear of cloud material or energy or a significant vertical shear of the zonal wind. 2.2. Cloud/Hole Lifetimes [26] The duration of holes in the clouds of Venus can indicate the nature of the processes that sustain them. McGouldrick and Toon [2008b] suggested that differences in the vertical profile of the zonal wind could lead to differences in the longevity of the features. Furthermore, McGouldrick and Toon [2008b] noted that a combination of vertical shear of the zonal wind and the vertical motions driven by the radiative dynamical feedback was sufficient to dissipate a large (
2000 km) hole in a time scale of under two weeks. Crisp et al. [1991b] reported that a
2000 km hole persisted for at least two weeks. However, since the initial formation of that particular feature was not observed, this is only a lower limit. Furthermore, the spatial resolution of VIRTIS on Venus Express, even at apoapse, is far

superior to what Crisp et al. [1991b] were able to achieve from Earth. Thus, the analysis of the evolution of smaller features is now possible. Since the absorption of upwelling infrared radiation plays a role in sustaining the middle and lower clouds [Pollack et al., 1980; McGouldrick and Toon, 2007], the longevity of holes affects the magnitude of possible horizontal temperature or pressure variations that would result from such heating. Strong horizontal temperature or pressure gradients will affect the magnitude of vertical motions triggered by the radiative dynamical feedback. [27] To determine the typical lifetime of the cloud features, we analyze the VIRTIS-M data from orbit 383. This orbit contains a sequence of cubes that spans 5 uninterrupted h, and each cube in the sequence exhibits exposure times long enough that the nightside (rather than the dayside) is exposed ideally (Table 2). Since the Venus Express spacecraft is approaching Venus throughout the series of acquired cubes in each orbit, there is only a small area of the planet’s atmosphere, which cannot exceed the field of view of the last cube acquired (i.e., when the spacecraft is at its closest to Venus), that contains features that can be tracked for the entirety of the sequence. [28] We first identify features that persist throughout the series of observations. That is, we limit our analysis to features that persist for at least 5 h, and do not stray beyond the instrumental field of view. The zonal superrotation

Table 2. Data Analyzed in Evolution Analyses File Name

Orbit

VI0383_00 VI0383_01 VI0383_02 VI0383_03 VI0383_04 VI0383_05 VI0383_06 VI0383_07 VI0383_08 VI0383_09 VI0383_10

383 383 383 383 383 383 383 383 383 383 383

Start Date 8 8 8 8 8 8 8 8 9 9 9

May May May May May May May May May May May

2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007

Start Timea

Exposure Timeb (s)

Distance (km)

Resolution (km pixel1)

2028:12 2058:12 2128:12 2158:12 2228:12 2258:12 2328:12 2358:12 0028:12 0058:12 0128:12

3.3 3.3 3.3 3.3 3.3 3.3 3.3 3.3 3.3 3.3 3.3

67106.1 66178.0 65055.9 63735.1 62426.3 60678.6 58710.8 56515.1 54343.1 51631.1 48668.4

16.8 16.5 16.3 15.9 15.6 15.2 14.7 14.1 13.6 12.9 12.2

a

Times and dates are UTC. Exposure time is per slice.

b

6 of 12

MCGOULDRICK ET AL.: VIRTIS OBSERVATIONS OF VENUS CLOUDS

E00B14

E00B14

Figure 3. All 10 images considered in the morphological analyses, combined into a single image. The atmosphere is assumed to rotate as a solid body, with a planet-circling period of 5.5 days, as in Figure 1. makes it difficult to be certain of the identity of each feature from image to image, so we crudely correct for the motion of the atmosphere by shifting each image longitudinally at a corresponding rate of about 2.73° hr1 (which equates to a 5.5 day period). This adjustment renders many of the features nearly stationary in the sequence of images, thus identification of the same hole in subsequent images was facilitated. The boundary of each feature is determined in the same manner as the feature boundaries in the morphological analysis above (section 2.1). We measure the peak radiance and the contrast between the peak radiance and the background radiance for each feature that we were able to track in this sequence of images.

well as the overall mean and standard deviation of all of the data considered here. The overall mean indicates that the 10 images that we have analyzed for this paper demonstrate the same tendencies that have been noted by previous

3. Results 3.1. Latitudinal Distribution of Cloud Cover [29] Figure 3 demonstrates the approximate spatial coverage of cloudy and clear regions in the data considered in this paper. All of the considered 1.74 mm images are plotted onto a cylindrical projection. As in Figure 1, the atmosphere is assumed to rotate as a solid body with a period of 5.5 days, so each image has been advected longitudinally according to the time it was acquired relative to the first image (orbit 363). Overlapping regions are averaged together. There is possibly a hint of longitudinal hemispherical asymmetry, similar to that observed by Crisp et al. [1991a], whereby longitudes east of 180° in these data are brighter than those west of 180°; but this conclusion might be biased by the existence of a single very large and very bright feature that spans the region from 220° to 300° longitude and 20° to 5° latitude. Although the data cover nearly two full atmospheric rotations, the coverage and consistency seems to be insufficient to determine whether a longitudinal hemispherical asymmetry exists in these data. [30] Figure 4 shows the latitudinal profile of the mean radiance for each of the images considered in this paper as

Figure 4. Latitudinal profile of the mean radiance for each of the 1.74 mm images considered in this paper. The profiles of the various images are plotted as points. The mean of all of the data considered is shown as a solid line, and range of ±1 standard deviation of all of the data considered is shown with a pair of dashed lines. Also shown here, for comparison, is an estimate of the emitted radiance that results from an application of the cloud properties as derived from the Pioneer Venus Lower Cloud Particle Size Spectrometer (LCPS) by Knollenberg and Hunten [1980]. The vertical dotted line is the calculated radiance; the horizontal dotted line is the latitude of the probe’s descent (the probe actually descended in the northern hemisphere, but we assume roughly hemispherical symmetry only for the purpose of making this comparison).

7 of 12

E00B14

MCGOULDRICK ET AL.: VIRTIS OBSERVATIONS OF VENUS CLOUDS

Figure 5. Aspect ratio (long axis divided by short axis) as a function of latitude. The data also have been binned into 20° latitude regions, with error bars indicating the standard deviation of aspect ratio in each region. One feature was identified poleward of 60°, which is shown in Figure 5, but obviously could not be sorted into any of the defined latitude bins. observers [Crisp et al., 1991a; Chanover et al., 1998]. Low latitudes tend to be fairly bright, as well as middle to high latitudes (between about 40° to 60°). There is typically less radiance emitted in middle to low latitudes (between about 20° and 40°). The polar collar, which is not easily detected in ground-based observations, can be seen in these images as a region of significantly diminished radiance poleward of about 60°. [31] Also shown in Figure 4, for comparison, is the estimated radiance at the location of the Pioneer Venus Large Probe. Although the Large Probe descended in the northern hemisphere, the location is plotted, in Figure 4, at about 5° latitude. We assume a north-south atmospheric symmetry here because the planetary obliquity is only 3°, indicating minimal seasonal variations, and because previous observations are consistent with such a north-south symmetry, with only a few, probably transient, exceptions. Recent observations by Venus Express indicate that the dynamics of the south polar region are similar to those observed of the north polar region by Pioneer Venus, further supporting the likelihood of north-south hemispheric symmetry [Piccioni et al., 2007]. This calculation of radiance experienced by the Pioneer Venus Large Probe assumes the cloud properties derived by Knollenberg and Hunten [1980], along with a factor of three reduction in the number of mode 3 particles [Pollack et al., 1980]. This represents the radiance that results from a typical assumed set of Venus cloud properties and mass. The optical depth of the middle cloud (i.e., the optical depth below about 57 km) that was used to generate this radiance is approximately t
23. This radiance is calculated using the model of McGouldrick and Toon [2007, 2008a, 2008b]. For a lack of an alternative, previous studies have taken this in situ observation (or reanalyses thereof) of the Venus cloud to be representative of the planetary cloud cover, as a whole [Toon et al., 1982; Grinspoon et al., 1993; James et al., 1997; Imamura and

E00B14

Hashimoto, 1998; McGouldrick and Toon, 2007]. Figure 4 shows that this clearly is not the case, since the predicted radiance of the cloud observed in situ by the Pioneer Venus Large Probe is closer to that of the maximum cloudiness (minimum radiance) encountered in the course of the 10 orbits analyzed here. Furthermore, the significant variability in radiance demonstrates the necessity of taking into account both the spatial and temporal cloud variability when determining the radiative balance of the middle atmosphere of Venus. [32] As mentioned in section 1.2, Imamura and Hashimoto [1998] explained that the latitudinal brightness variations seen by Crisp et al. [1991a] likely were a consequence of the global-scale Hadley cell that exists in the atmosphere of Venus. The overall average of the latitudinal distribution of radiance in these data (solid line in Figure 4) is remarkably similar to the results of those simulations. The bright zone around 50° in the data corresponds with the minimum of opacity located around 45° in the simulations of Imamura and Hashimoto [1998]. The diminished radiance poleward of about 60° may be reaching a minimum around 75°. The location of this radiance minimum also corresponds well with the increase in opacity in the nominal simulations of Imamura and Hashimoto [1998], which peak around 70° to 75°. Imamura and Hashimoto [1998] also performed a simulation in which they reduced the effectiveness of the Hadley circulation by reducing by half the speed of the meridional circulation in their simulations. In contrast with the nominal simulation, this second simulation exhibited a peak of cloud opacity somewhat more equatorward, at about 60° latitude (compared with about 75° in their nominal simulation). Although the data considered here represent only 10 days worth of observation, these data indicate that, compared with their test case with reduced meridional speeds, the nominal case of Imamura and Hashimoto [1998] is a better representation of the Hadley circulation of Venus. The increase in the variability of the observed radiance at low latitudes is consistent with the observations by Crisp et al. [1991a] and also suggests the existence of significant convective activity in the clouds of Venus. 3.2. Latitudinal Distribution of Cloud Morphology [33] In Figure 5, we plot the aspect ratio that we measured for approximately 130 features in the 1.74 mm VIRTIS images from orbits 363 through 372, as a function of the latitude of the center of the feature. In addition to plotting each individually measured aspect ratio as symbols, we also bin the measured data into 20° latitude bins. The mean and standard deviation of these three bins are also plotted in Figure 5. We see in Figure 5 that the aspect ratio of the measured holes in the clouds increases with latitude. The typical hole becomes more elongated poleward of 40° (of the 16 features with aspect ratios greater than 5:1, all but three are poleward of 40°). At low latitudes, in addition to holes having aspect ratios closer to 1.0, there also are several features with aspect ratios less than one (i.e., longer in the transverse direction than in the longitudinal direction), something that does not at all occur poleward of 40° in the images considered here. Thus, we observe a preponderance of highly elongated holes at high latitudes, and a

8 of 12

E00B14

MCGOULDRICK ET AL.: VIRTIS OBSERVATIONS OF VENUS CLOUDS

E00B14

it will have affected the average brightness in this latitude bin, but not the variability.

Figure 6. Tilt angle (relative to parallels of latitude) as a function of latitude. The data also have been binned into 20° latitude regions, with error bars indicating the standard deviation of tilt angle in each region. An angle of zero means zonally oriented features. The one feature poleward of 60° was not sorted into any bin. tendency for more equal aspect ratios at lower latitudes. These results are consistent with the observations reported by Belton et al. [1991], as summarized in section 1.1. [34] In Figure 6, we plot the angle that each feature makes relative to the parallels of latitude, as a function of latitude. A small ‘‘tilt angle’’ indicates that a feature is aligned nearly zonally, whereas a large tilt angle indicates a feature whose long axis is more vertically oriented. As in Figure 5, we plot each feature individually, as well as the mean and standard deviation of the tilt angle in 20° latitude bins. In Figure 6, we see that the average tilt angle is somewhat larger at low latitudes, becoming smaller at higher latitudes. This, in combination with the trends in Figure 5, indicates that the features not only become more elongated at higher latitudes, they also become more zonally oriented. There is significantly greater variation in the tilt angle at latitudes equatorward of 40°. The smaller variability poleward of 40° suggests that the polar vortex exerts its influence mainly on the poleward side of 40°. Furthermore, the variability equatorward of 40° suggests a complex interaction, likely involving convection, between the local atmospheric dynamics and the formation and evolution of the lower and middle clouds at lower latitudes. [35] Figure 7 shows the peak radiance emerging from holes as a function of latitude. As in Figures 5 and 6, we plot both the individual features themselves, as well as the averages binned over 20° latitude bins. The mean of the feature peak radiance is about 0.2 W m2 mm1 sr1, fairly consistent across all three latitude bins considered here. However, there is significantly more variability in the peak brightness of holes at low latitudes, as compared with the variability seen at higher latitudes. This too suggests that lower latitudes are demonstrating the effects of convection. Since the lower latitudes are viewed most obliquely in these data, the lower latitudes will be most affected by any errors that may occur in the correction for emission angle. However, even if our correction for the emission angle is flawed,

3.3. Feature Lifetimes [36] We might also be able to obtain information about the ages (and, ultimately, the lifetimes) of holes in the clouds from the morphology analyses. Recall, from equation (1), that the change in the atmospheric rotation period with latitude is a function of latitude only, if the zonal velocity is approximately constant. Cloud-tracking analysis by Sa´nchez-Lavega et al. [2008] suggests that the zonal wind is indeed a relatively constant 60 m s1 at latitudes equatorward of about 50°. If we assume that a hole is circular when it is formed, and the zonal velocity is constant, then the poleward edge of the hole advances farther than the equatorward edge of the hole, at a rate given by equation (1) times the latitudinal extent of the hole (Df). If we assume that there is no meridional motion, then the ratio of this translation in degrees longitude to the latitudinal extent of the hole is precisely our previous definition of the tilt angle of the hole! For example, a circular hole centered at 50° latitude, having a meridional extent of 2°, exhibits a roughly 3.15° longitude day1 difference between its top and bottom edges. After 1 day, this equates to a tilt angle of q = tan1(2/3.15) = 32°. After two days, the tilt angle would be 18°, and after three days, 12°. Granted, this is an extreme case. Holes that are closer to the equator experience less shearing because of this effect, since the angular velocity changes less quickly with latitude. A corresponding effect on the aspect ratio can also be calculated. Thus, these measurements of the aspect ratio and of the tilt angle, coupled with knowledge of the horizontal winds in the atmosphere, can provide a rough estimate of the ages of the holes in the clouds. [37] Using the procedure outlined in section 2, we have identified 10 holes, in the series of images taken during orbit 383, whose evolution we have been able to analyze for this paper. This is such a small number of holes that it will be difficult to determine any global trends of hole evolution,

Figure 7. Peak radiance of the holes as a function of feature latitude. The data also have been binned into 20° latitude regions, with error bars indicating the standard deviation of feature peak radiance in each region. The one feature poleward of 60° was not sorted into any bin.

9 of 12

E00B14

MCGOULDRICK ET AL.: VIRTIS OBSERVATIONS OF VENUS CLOUDS

Figure 8. Derived evolutionary time scales versus peak radiance in the holes. Both axes are parameters of a log linear fit to the measured data. The error bars indicate the standard deviation error estimates of the fitted parameters. whether as a function of position or size. Furthermore, the difficulty in establishing a method of hole definition that is consistent across successive images (i.e., that identifies the same hole, and is not fooled by changes in morphology or surroundings) may introduce small additional uncertainties. For the present investigation, we quantify these numerous error sources in the uncertainties of the fits to the data, as described in subsequent paragraphs. In future investigations we will improve the level of quantification of the error in these analyses of hole evolution, as well as improve our method of feature identification and tracking. Nevertheless, as the first quantitative measurement of cloud evolution, this analysis provides the first benchmark of the lifetimes of the holes in the clouds of Venus. [38] We fit the sequence of the measured peak radiance for each hole according to an exponential variation: ln Ipeak ðt Þ ¼ ln Ipeak ð0Þ 

1 t t

E00B14

fit to the curve at the time of the first image acquisition. It does not represent an actual measurement of radiance, nor does the variance shown indicate a known measurement error. Figure 8 may exhibit a peak in the evolutionary time constant (i.e., both thicker and thinner holes than this tend to evolve more quickly) at about Ipeak
0.15 W m2 mm1 sr1. However, the small number of holes that have been analyzed, and the large error bars on some of the features, cause us to have only a little confidence in the existence of such a peak, at this time. [40] For Figure 9 we replace Ipeak in equation (2) with the contrast between the peak radiance and the background radiance. We define the contrast as the ratio between the peak radiance of the hole (which we have measured directly from the images), and the local background radiance. We determine the local background radiance to be the mean minus one standard deviation of the radiance in the image at the latitude of the center of the hole. Recall from Figure 4 that the mean minus one standard deviation as a function of latitude tracks remarkably close to the minimum radiance as a function of latitude. Thus, the mean minus one standard deviation stands as a reasonable proxy for the minimum radiance (or thickest cloud cover). In Figure 9, we see a possible trend indicating that holes with greater contrast evolve more quickly than holes with smaller contrast. This is consistent with simulations of hole evolution [McGouldrick and Toon, 2008b]. It also suggests that a uniform cloud cover is the equilibrium situation in the Venus atmosphere, since large contrasts between cloudy and clear regions are quickly reduced, and small contrasts evolve rather slowly. [41] In Figure 10, we plot the evolution time constant obtained from the analysis of the peak radiance as a function of the latitude of the analyzed holes. In Figure 10, the horizontal error bars indicate the range of latitudes at which a given measured hole was located during the course of the image sequence. That is, a large horizontal error bar may be indicative of a significant meridional motion. If meridional

ð2Þ

where the initial peak radiance [ln Ipeak(0)] and the evolutionary time constant (t) are free parameters to be fit to the data for each hole. We find that, for the holes we have analyzed during orbit 383, the average time scale for evolution of a typical hole is 0.95+4.88 0.51 days. Thus, since the typical e-folding time scale of a hole is approximately 1 day, the peak radiance of a typical hole will decrease to approximately 1% of its initial value in a time of about 4.5 days. Note, however, that there is a large upper error bar to this time scale, indicating that a significant fraction of holes can be expected to exhibit time constants for evolution as large as 5 – 6 days. Such slowly evolving holes will remain evident for many weeks. [39] Figure 8 shows the relationship between the initial peak radiance and the evolutionary time scale. Each point in Figure 8 represents the best linear fit to equation (2) for each measured hole. The vertical error bars indicate the one standard deviation variation in the slope (1/t), and the horizontal error bars indicate the one standard deviation variation in the initial peak radiance [Ipeak(0)]. Note that the initial peak radiance in this case refers to the peak radiance

Figure 9. Derived evolutionary time scales versus contrast between the peak hole radiances and the background radiances. Background radiance is defined in the text. Both axes are parameters of a log linear fit to the measured data. The error bars indicate the standard deviation error estimates of the fitted parameters.

10 of 12

E00B14

MCGOULDRICK ET AL.: VIRTIS OBSERVATIONS OF VENUS CLOUDS

Figure 10. Derived evolutionary time scales versus the latitudes of the measured holes. The horizontal error bars indicate the range of latitudes measured for each hole during the sequence of images. The vertical error bars indicate the standard deviation error estimates of the fitted time scale. motion can be ruled out, then a significant variation in measured feature latitude suggests that we may not be measuring the same feature in each image. With sufficient statistics and independent means of determining wind velocities, such an analysis can provide a further check on the validity of the analysis of feature evolution. Even with the small number of features measured in this instance, there appears to be a trend in which holes at lower latitudes evolve more quickly than those at higher latitudes. This behavior is consistent with the earlier assessment of increased cloud variability at equatorial latitudes. [42] In each of these analyses, our conclusions are necessarily tentative, as they derive from measurements of only 10 holes over the span of only 5 h. We hope to improve on these statistics by measuring many more features in many more images in the future. Nevertheless, these analyses represent a first quantitative assessment of the evolution of holes in the middle and lower clouds of Venus.

4. Conclusions [43] We have analyzed a series of ten 1.74 mm VIRTIS-M IR images, taken on successive orbits, covering a span of 10 days, to identify tendencies to the morphology and the brightness of the holes in the clouds of Venus. We have also analyzed a separate series of eleven 1.74 mm VIRTIS-M IR images, taken over the course of a single orbit, spanning a time of 5 h, to investigate the evolution of the holes in the clouds of Venus. [44] We find that the latitudinal distribution of radiance in which midlatitudes are relatively brighter, and low latitudes exhibit somewhat greater overall variability, is consistent with that observed by Crisp et al. [1991a], indicating that this tendency is a fairly stable phenomenon. The latitudinal distribution of average radiance that we observe in the images considered here is largely consistent with the simulations of Imamura and Hashimoto [1998], indicating that the global Hadley circulation plays a significant role in the distribution of the cloud mass of Venus. We observe an

E00B14

increase in the variability of radiance at lower latitudes, suggestive of the importance of convection to the nature of the middle and lower clouds of Venus. While the distribution of potential cloud constituents and their precursors (i.e., H2SO4, SO2, H2O, and amorphous sulfur) are not constrained by the data and analyses presented here, variations in these constituents, driven either by chemistry or by the atmospheric dynamics already noted, might also play a role in the cloud variability. [45] We find that the morphology of the holes in the clouds of Venus trends from features having a wide variety of orientations (relative to the zonal axis) and aspect ratios of approximately 1.0 at lower latitudes, to highly elongated and zonally oriented features at high latitudes. The greater variability in the morphology of features at lower latitudes is possibly indicative of a greater amount of convective activity there. We fail to find a correlation between the peak radiance of a hole and its latitude. A hint of a longitudinal hemispheric asymmetry, similar to that observed by Crisp et al. [1991a] and Chanover et al. [1998] was seen; but the amount of data considered here is insufficient to be certain of its existence here. [46] We find that, if the data analyzed here are typical, holes in the middle and lower clouds of Venus evolve with a time constant of approximately 1 day. We find that holes exhibiting greater contrast between their peak radiance and the background radiance, and holes at lower latitudes evolve more quickly. This is consistent with the increased variability seen at lower latitudes that results from enhanced convection. It is also consistent with the notion that the middle and lower clouds are supported by a radiative dynamical feedback, whereby regions exhibiting large contrasts in clouds cover will also experience large contrasts in radiative heating because of the absorption of upwelling infrared radiation. However, the data considered here represent only 5 h of observation of a fairly small area (about 2% of the atmosphere). This is hardly sufficient data to allow us to draw any broad conclusions. We intend to carry out future efforts to improve both the methods of the analysis as well as the volume of analyzed data. [47] The features in the images analyzed here strongly indicate the activity of convection in the lower and middle cloud decks of Venus. The spatial variability and short time scale for their evolution are somewhat inconsistent with typical terrestrial stratus, to which the clouds of Venus are often compared. However, previous radiative modeling by Pollack et al. [1980] and microphysical modeling of the middle and lower clouds by McGouldrick and Toon [2007] suggests the existence of a radiative dynamical feedback, which can support convection in the lower and middle cloud decks of Venus. The clouds of Venus may be more analogous to terrestrial stratocumulus ‘‘cloud streets,’’ whereby convection exists in the midst of strong horizontal flow [Houze, 1993]. Depending upon the relative strengths of the horizontal flow and the convection, either rolls or cellular patterns may be formed. Similarly, the nature of terrestrial thunderstorm development (whether single-cell, multicell, or supercell) is known to be somewhat dependent upon the local wind shear and static stability [Houze, 1993]. Perhaps processes analogous to cloud street formation or the development of multicell thunderstorms are occurring in the atmosphere of Venus in the vicinity of the lower and middle clouds.

11 of 12

E00B14

MCGOULDRICK ET AL.: VIRTIS OBSERVATIONS OF VENUS CLOUDS

[48] Here we have presented analyses of only a small subset of the VIRTIS data. We have analyzed data from only 11 orbits, from among more than 500: barely 2% of the total available VIRTIS data set. However, given that at this writing, Venus Express has been orbiting Venus for more than three full Venus years, there likely is sufficient data in the archive to seek out potential tendencies of hole formation and/or appearance as a function of geographic or solar longitude. We intend to carry out further analyses of the VIRTIS data to determine whether such tendencies exist. Furthermore, comparisons with observations at other wavelengths, and concurrent Venus Monitoring Camera observations, can be utilized to gain greater understanding of the properties and evolution of the Venus clouds. In the longer term, the anticipated arrival of the Japanese Space Agency’s Venus Climate Orbiter (VCO) in 2010 promises to enhance the study of the Venus clouds [Nakamura et al., 2007]. The understanding of the nature of the middle and lower clouds of Venus could gain much from the potential coordinated observation of Venus by the two spacecraft. Subsequent observations of the same region of the atmosphere could extend the analysis of cloud evolution beyond the limits that either VCO or Venus Express could accomplish by itself, by extending the baseline of time over which a particular feature is viewable. Concurrent observations of features by multiple spacecraft could further elucidate the cloud properties. For example, it may be possible to determine the difference between the properties of a profile of a hole versus that of a cloud by performing occultation measurements with one spacecraft, while the other identifies the types of features that are passing over the limb of the planet. Alternately, long-lived balloon observatories can be placed in the atmosphere and possibly maneuvered to sample the atmosphere in regions that are identified by the concurrently orbiting spacecraft. [49] Acknowledgments. We thank two anonymous reviewers for their careful reading of the manuscript and their suggestions for making this a stronger paper. We also acknowledge conversations with Colin Wilson, Con Tsang, and Robert Carlson that have helped to improve the analysis of the data with respect to the correction for the emission angle. We also thank The European Space Agency for the opportunity to work with Venus Express. We thank Pierre Drossart, Giuseppi Piccioni, and the VIRTIS team for their assistance in enabling us to reduce and analyze the VIRTIS data. A portion of the work described in this paper was carried out at the Jet Propulsion Laboratory, Pasadena, California, under contract with NASA. K.M. and D.H.G. were supported by NASA in support of ESA’s Venus Express mission under grant NNX07AI61G.

References Allen, D. A., and J. W. Crawford (1984), Cloud structure on the dark side of Venus, Nature, 307, 222 – 224. Baker, R. D., G. Schubert, and P. W. Jones (2000), Convectively generated internal gravity waves in the lower atmosphere of Venus, part II: Mean wind shear and wave-mean flow interaction, J. Atmos. Sci., 57, 200 – 215. Belton, M. J. S., et al. (1991), Images from Galileo of the Venus cloud deck, Science, 253, 1531 – 1536. Blamont, J. E., et al. (1986), Implications of the VEGA balloon results for Venus atmospheric dynamics, Science, 231, 1422 – 1425. Bullock, M. A., and D. H. Grinspoon (2001), The recent evolution of climate on Venus, Icarus, 150, 19 – 37, doi:10.1006/icar.2000.6570. Carlson, R. W., et al. (1991), Galileo infrared imaging spectroscopy measurements at Venus, Science, 253, 1541 – 1548. Carlson, R. W., L. W. Kamp, K. H. Baines, J. B. Pollack, D. H. Grinspoon, Th. Encrenaz, P. Drossart, and F. W. Taylor (1993), Variations in Venus cloud particle properties: A new view of Venus’s cloud morphology as observed by Galileo Near-Infrared Mapping Spectrometer, Planet. Space Sci., 41, 477 – 485, doi:10.1016/00320633(93)900306.

E00B14

Chanover, N. J., D. A. Glenar, and J. J. Hillman (1998), Multispectral nearIR imaging of Venus nightside cloud features, J. Geophys. Res., 103, 31,335 – 31,348. Counselman, C. C., S. A. Gourevitch, R. W. King, G. B. Loroit, and E. S. Ginsberg (1980), Zonal and meridional circulation of the lower atmosphere of Venus determined by radio interferometry, J. Geophys. Res., 85, 8026 – 8030. Crisp, D. (1989), Radiative forcing of the Venus mesosphere, Icarus, 77, 391 – 413, doi:10.1016/00191035(89)900961. Crisp, D., et al. (1989), The nature of the near-infrared features on the Venus night side, Science, 246, 506 – 509. Crisp, D., D. A. Allen, D. H. Grinspoon, and J. B. Pollack (1991a), The dark side of Venus: Near-infrared images and spectra taken from the Anglo-Australian Observatory, Science, 253, 1263 – 1266. Crisp, D., et al. (1991b), Ground-based near-infrared imaging observations of Venus during the Galileo encounter, Science, 253, 1538 – 1541. Drossart, P., et al. (2007), Scientific goals for the observation of Venus by VIIS on ESA/Venus Express mission, Planet. Space Sci., 55, 1653 – 1672, doi:10.1016/j.pss.2007.01.003. Grinspoon, D. H., J. B. Pollack, B. R. Sitton, R. W. Carlson, L. W. Kamp, K. H. Baines, Th. Encrenaz, and F. W. Taylor (1993), Probing Venus’s cloud structure with Galileo NIMS, Planet. Space Sci., 50, 515 – 542, doi:10.1016/00320633(93)90034Y. Hashimoto, G. L., and Y. Abe (2000), Stabilization of Venus’ climate by a chemical-albedo feedback, Earth Planets Space, 52, 197 – 202. Hinson, D. P., and J. M. Jenkins (1995), Magellan radio occultation measurements of atmospheric waves on Venus, Icarus, 114, 310 – 327, doi:10.1006/icar.1995.1064. Houze, R. A. (1993), Cloud Dynamics, Academic, San Diego, Calif. Imamura, T., and G. L. Hashimoto (1998), Venus cloud formation in the meridional circulation, J. Geophys. Res., 103, 31,349 – 31,366. James, E. P., O. B. Toon, and G. Schubert (1997), A numerical microphysical model of the condensational Venus cloud, Icarus, 129, 147 – 171, doi:10.1006/icar.1997.5763. Knollenberg, R. G., and D. H. Hunten (1980), The microphysics of the clouds of Venus: Results of the Pioneer Venus Particle Size Spectrometer Experiment, J. Geophys. Res., 85, 8038 – 8058. Linkin, V. M., et al. (1986), Vega balloon dynamics and vertical winds in the Venus middle cloud region, Science, 231, 1417 – 1419. Markiewicz, W., et al. (2007a), Venus Monitoring Camera for Venus Express, Planet. Space Sci., 55, 1701 – 1711, doi:10.1016/j.pss.2007.01.004. Markiewicz, W. J., et al. (2007b), Morphology and dynamics of the upper cloud layer of Venus, Nature, 450, 633 – 636, doi:10.1038/nature06320. McGouldrick, K., and O. B. Toon (2007), Investigation of possible causes of the holes in the condensational Venus cloud using a microphysical cloud model with a radiative-dynamical feedback, Icarus, 191, 1 – 24, doi:10.1016/j.icarus.2007.04.007. McGouldrick, K., and O. B. Toon (2008a), Observable effects of convection and gravity waves on the Venus condensational cloud, Planet. Space Sci., 46, 1112 – 1131, doi:10.1016/j.pss.2008.02.010. McGouldrick, K., and O. B. Toon (2008b), Modeling the effects of shear on the evolution of the holes in the condensational clouds of Venus, Icarus, 196, 35 – 48, doi:10.1016/j.icarus.2008.02.020. Nakamura, M., et al. (2007), Planet-C: Venus Climate Orbiter mission of Japan, Planet. Space Sci., 55, 1831 – 1842, doi:10.1016/j.pss.2007.01.009. Piccioni, G., et al. (2007), South-polar features on Venus similar to those near the north pole, Nature, 450, 637 – 640, doi:10.1038/nature06209. Pollack, J. B., O. B. Toon, and R. Boese (1980), Greenhouse models of Venus’ high surface temperature, as constrained by Pioneer Venus measurements, J. Geophys. Res., 85, 8223 – 8231. Sa´nchez-Lavega, A., et al. (2008), Variable winds on Venus mapped in three dimensions, Geophys. Res. Lett., 35, L13204, doi:10.1029/ 2008GL033817. Schubert, G., et al. (1980), Structure and circulation of the Venus atmosphere, J. Geophys. Res., 85, 8007 – 8025. Taylor, F. W., et al. (1979), Temperature, cloud structure, and dynamics of Venus middle atmosphere by infrared remote-sensing from pioneer orbiter, Science, 205, 65 – 67. Toon, O. B., R. P. Turco, and J. B. Pollack (1982), The ultraviolet absorber on Venus: Amorphous sulfur, Icarus, 51, 358 – 373, doi:10.1016/ 00191035(82)900896. 

K. H. Baines and T. W. Momary, Jet Propulsion Laboratory, California Institute of Technology, M/S 183-601, 4800 Oak Grove Drive, Pasadena, CA 91109, USA. D. H. Grinspoon and K. McGouldrick, Department of Space Sciences, Denver Museum of Nature and Science, 2001 Colorado Boulevard, Denver, CO 80205, USA. ([email protected])

12 of 12

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, E00B05, doi:10.1029/2008JE003137, 2008

for

Full Article

Whistler mode waves from lightning on Venus: Magnetic control of ionospheric access C. T. Russell,1 T. L. Zhang,2 and H. Y. Wei1 Received 29 February 2008; revised 29 May 2008; accepted 19 June 2008; published 18 September 2008.

[1] The fluxgate magnetometer on Venus Express samples the magnetic field near

periapsis at 128 Hz. Bursts of plane-polarized magnetic waves in the vicinity of 100 Hz are observed propagating at small angles to the magnetic field. The magnetic field is generally horizontal in the region around periapsis, located at high northern latitudes. When the magnetic field remains within 15° of horizontal during the 2-min periapsis pass, no such waves are observed; but when there are brief periods during which the local magnetic field dips into the atmosphere by more than 15°, the bursts begin to appear. Such radial excursions of the magnetic field occur 25% of the time in the region around periapsis. The bursts are seen only on passes with these excursions. We interpret this magnetic control in terms of the coupling between the electromagnetic wave from lightning discharges refracted vertically by the increasing electron density and the nearly horizontal ionospheric magnetic field along which the energy is guided to the spacecraft. The inferred rate of electric discharges in the Venus atmosphere is about 20% of that seen in the Earth’s atmosphere. Citation: Russell, C. T., T. L. Zhang, and H. Y. Wei (2008), Whistler mode waves from lightning on Venus: Magnetic control of ionospheric access, J. Geophys. Res., 113, E00B05, doi:10.1029/2008JE003137.

1. Introduction [2] While the Venus atmosphere is quite dry, containing very little water, Venus is also quite cloudy, with hydrated sulfuric acid droplets shrouding the planet some 50 –60 km above the surface. At low altitudes, the winds are slow, but in the clouds, the horizontal velocities are observed to range to over 100 ms 1. These ‘‘4-d’’ winds produce much wind shear and probably are accompanied by significant vertical transport as well. Thus, the clouds may well be electrified as terrestrial clouds are. The observations reported herein are over the polar vortex that is possibly associated with rapid downwelling [Piccioni et al., 2007]. [3] Many researchers have reported optical and electromagnetic signals that could be produced by discharges in or from these clouds. Krasnopolsky [1983] reported flashes seen in the Venera-9 visible spectrometer; Hansell et al. [1995] reported detections with an Earth-based telescope. The landing probes of Venera 11, 12, 13, and 14 detected electromagnetic pulses as they descended through the atmosphere and as they sat on the surface [Ksanfomaliti, 1983]. They have been most extensively studied from orbit on Pioneer Venus (PVO) using an electric antenna [Taylor et al., 1979; Scarf et al., 1980]. Finally, at radio frequencies that are able to propagate through the ionosphere with minimal interaction with the plasma, Galileo recorded radio 1 Institute of Geophysics and Planetary Physics, University of California, Los Angeles, California, USA. 2 Space Research Institute, Austrian Academy of Sciences, Graz, Austria.

waves similar to terrestrial radio frequency emissions from lightning [Gurnett et al., 1991]. However, as Gurnett et al. [2001] later reported, these high-frequency radio waves are weaker in strength and less frequent in occurrence than terrestrial lightning. [4] The waves detected in the magnetized Venus ionosphere by PVO were of two distinct types [Russell, 1991]. One class appeared only when the magnetic field had a significant radial component and behaved like an electromagnetic wave would. This wave diminished only slightly in estimated electromagnetic energy flux with increasing altitude. The second class was apparently electrostatic and attenuated rapidly with altitude. It was seen in all four narrowband frequency channels: 100 Hz, 730 Hz, 5.6 kHz, and 30 kHz. We interpret these electrostatic waves as being associated with cloud-to-ionosphere strikes, probably ‘‘local.’’ The electromagnetic waves were most probably due to intracloud discharges, possibly at some distance. [5] The occurrence of lightning on Venus remained controversial after the Venera and PVO missions ceased because some searches were unsuccessful. A search for scattered light in the PVO star sensor came up empty [Borucki et al., 1991], albeit the total time the star sensor was active and able to see flashes was only a few minutes in total, with very little coverage over the region defined as active by the PVO electric field instrument. The photometer on the Venus balloons also did not observe flashes [Sagdeev et al., 1986], but the balloons were in the clouds and were not looking down on the clouds from above. Thus any light from flashes could be scattered before reaching the detectors on the balloons.

Copyright 2008 by the American Geophysical Union. 0148-0227/08/2008JE003137$09.00

E00B05

1 of 8

E00B05

RUSSELL ET AL.: LIGHTNING ON VENUS

Figure 1. Waveform of 8-ms samples of the magnetic field at periapsis on 1 July 2006, bandpass filtered from 42 to 60 Hz and rotated into principal axis coordinates.

[6] The waves observed by Pioneer Venus were quite strong. In a vacuum, the energy flux of electromagnetic waves is evenly divided between the electric and magnetic components. In the ionosphere where the index of refraction is high, the wave slows down and the energy is mainly carried by the magnetic component. On Pioneer Venus, the plasma density, the magnetic field, and the electric component of the wave were measured and we could calculate the electromagnetic energy of the waves identified as propagating in the whistler mode [Russell et al., 1989]. Furthermore, we could check this estimate with data obtained in the atmosphere beneath the ionosphere on the last few orbits when Pioneer Venus began to enter the atmosphere because of gravitational perturbations and eventually because of drag on the spacecraft [Strangeway et al., 1993]. These calculations indicated that the Pioneer Venus electric waves near 100 Hz could be detected by a state-of-the-art fluxgate magnetometer in the Venus ionosphere [Russell et al., 2006]. The Pioneer Venus magnetometer did not have the bandwidth to detect these waves. When it was decided to include a magnetometer on Venus Express, the decision was made to include a 128-Hz sampling mode that would allow magnetic waves near 100 Hz to be sampled [Zhang et al., 2006]. These data were obtained initially for only 2 min around periapsis. In late December 2006, this interval was extended. As expected, waves with the expected amplitude and temporal structure were observed at periapsis [Russell et al., 2007]. In this paper we report on the properties of the waves seen in these 2-min sampling intervals.

E00B05

has been described in detail by Zhang et al. [2006], and the cleaning process has been described by Zhang et al. [2008]. The Alfvenic character of interplanetary fluctuations has been used to establish the magnetometer’s DC level to better than 1 nT [Leinweber et al., 2008]. [8] In addition to the DC spacecraft fields, there is AC interference, in particular from the reaction wheels. These waves are found in the passband in which we expect to see the waves due to lightning. They also can vary in frequency from pass to pass, and at times, during a pass. Examples of these waves have been published by Russell et al. [2008]. A very simple test of whether a wave is in the ambient plasma or due to spacecraft noise is simply to subtract the measurements of the two magnetometers. Natural waves have the same amplitude at the two locations and disappear in the difference wave. The reaction wheel noise is stronger in the sensor on the spacecraft deck and does not disappear in subtraction [Russell et al., 2008]. As we demonstrate below in section 3, the bursts we discuss in this report pass this test. [9] In this study we use the outbound sensor and pass the time series through a bandpass filter from 42 to 60 Hz. This range of frequencies is chosen to avoid strong signals from the reaction wheels [Russell et al., 2008]. The magnetometer has an antialiasing filter, but this filter does allow strong signals above the Nyquist frequency of 64 Hz to enter the telemetry stream. These signals will appear to be below 64 Hz, but their polarization will be reversed. When reaction wheel noise appears in this passband, we simply eliminate this pass from consideration. We examine herein observations obtained in 2006, after the initial commissioning of the instruments in orbit.

3. Observations [10] In the Pioneer Venus observations, whistler mode waves could not be detected in the day-lit hemisphere because of solar-induced noise in the plasma wave electric antenna. One of the advantages of the Venus Express magnetic measurement is that the waves can be observed in sunlight as well as darkness. Figure 1 shows a filtered

2. Instrument and Data Processing [7] Because Venus Express was a low-cost mission and employed an existing bus, there was no magnetic cleanliness program. Instead, two magnetometer triads were installed: one on the top deck and the second on the end of a 1-m boom. The two magnetometers were aligned and used in a gradiometer mode to identify spacecraft-associated waves so that they could be removed. The magnetometer

Figure 2. Three seconds of the record shown in Figure 1 showing substructure in the burst.

2 of 8

E00B05

RUSSELL ET AL.: LIGHTNING ON VENUS

E00B05

Figure 3. Hodogram of 400 ms of the burst shown in Figure 2 in the principal axis system.

waveform from the outer fluxgate sensor on 1 July 2006, at 308-km altitude and 0642 LT. The wave packets have been rotated into their principal axis system where the minimum variance direction is along the third axis, and the maximum variance direction is along the first. Clearly, the waves are confined to two directions and have similar amplitudes in these two directions. Figure 2 focuses on 3 s of this record revealing substructure within the burst. Overall, the wave packets last about 1/4 to 1/2 s, but clearly substructure can occur on 100-ms scales. A hodogram of these waves is shown in Figure 3, over 400 ms of data. The waves are confined to a plane as is expected of whistler mode signals. The waves’ phase velocity (k vector) along the minimum variance direction is propagating at only 9° to the background magnetic field. As mentioned above, the magnetometer samples continuously in a gradiometer mode with simultaneous samples at the end of the boom and on the spacecraft upper deck. Ambient waves will be the same

Figure 4. By component of the 128-Hz magnetic field in spacecraft coordinates measured on the outboard and inboard sensors and the difference between the two readings. Naturally occurring, nonspacecraft sources would be identical at the two locations and cancel in subtraction. The burst’s absence indicates that the burst in Figure 3 is not due to spacecraft sources.

Figure 5. Bandpass-filtered measurements in principal axis system for 400 ms at periapsis on 23 December 2006. Comments of Figure 1 apply.

amplitude at the two magnetometers so that subtracting the two time series will result in a time series of only noise. In Figure 4, we demonstrate this subtraction for the event shown in the hodogram of Figure 3. The top trace is the output of the y component outboard sensor. The middle trace is the y component inboard sensor. The bottom trace is the difference. In the difference trace, the burst of waves vanishes into the noise level, indicating it is an ambient signal. [11] An event observed on 23 December 2006, at 284-km altitude, 88° latitude, 1042 LT, and 88.5° solar zenith angle is shown in Figure 5. This event is propagating at only 6° to the local magnetic field. Figure 6 shows a hodogram of 250 ms of data at the maximum of the event. Figure 7 repeats our gradiometer difference test for the burst in the hodogram in Figure 6. Again the burst disappears in the difference trace indicating its source is not on the spacecraft. [12] Figure 8 shows projections of the background magnetic field along the orbit in the plane containing the solar direction and the orbit plane normal. The 8-s averaged magnetic field is shown every minute. The event shown in Figure 7 occurs at the second vector before 0750 UT

Figure 6. Hodogram of 250 ms of the burst shown in Figure 5.

3 of 8

E00B05

RUSSELL ET AL.: LIGHTNING ON VENUS

E00B05

where the field is dipping most strongly into the atmosphere. This suggests that we should examine the control of the occurrence of these waves by the magnetic field orientation. We demonstrate this control statistically in section 4.

shows that on some days, there just are not waves present near the dipping field lines. July 2 has an ambient magnetic field that is quite horizontal and says little about the possible presence of lightning near the satellite at this time. We note that more reaction wheel noise is present here than on other passes, but we feel we would have seen a typical amplitude burst if present. [16] Figure 11 shows three consecutive days in December 2006, when the spacecraft was near 251 km, 0141 LT, and 96.8° solar zenith angle. December 22 has a horizontal ambient magnetic field and no wave events. December 23 has a field inclined nearly 40° to the horizontal and has a strong wave event, near 1 nT peak-to-peak. December 24 deviates nearly 20° to the horizontal, and no wave bursts are seen. Again we have a pattern of inclined field being associated with occasional wave events and horizontal fields being associated with quiet intervals. [17] Figure 12 summarizes our survey of the 2006 Venus Express data. Plotted is the peak-to-peak amplitude of the strongest bursts on each pass versus the maximum deviation of the ambient field from the horizontal. We see that when the field deviation was less than 15° from the horizontal direction everywhere on the pass, no whistler mode wave bursts were detected. When the deviation was greater than 17°, the wave events could occur. We attribute this magnetic control to the effect of the ionosphere on the electromagnetic waves propagating from the clouds below and the properties of the propagation of whistler mode waves at the conditions in the Venus ionosphere as discussed in section 5.

4. Magnetic Field Control of Burst Occurrence

5. Discussion

[13] To illustrate the nature of the magnetic field control of the appearance of the whistler mode waves, we display the entire 2-min passes for three consecutive orbits together with the angle between the ambient magnetic field and the local horizontal direction for the same interval plotted with 1-s resolution. These orbits have very similar altitudes and local times but may differ in the steadiness and orientation of the magnetic field. [14] Figure 9 shows the measurements on 8, 9, and 10 June 2006. On 10 June, the ambient magnetic field is quiet and close to horizontal for the entire 2-min pass. No wave bursts are seen in the right-hand panel. In contrast on 9 June, the ambient magnetic field is time varying moving up to 30° away from horizontal. Strong wave packets accompany these variations at the beginning of the interval. These wave packets are analyzed in detail by Russell et al. [2007]. June 8 has similar field variations to those of 9 June, but no wave events are seen. This observation indicates that the magnetic orientation is not the only controlling factor. In fact, the most probable explanation of the lack of waves on this pass is that there are no atmospheric discharges below the satellite on this day. [15] Another 3-day interval, 30 June, 1 and 2 July, is shown in Figure 10 in the same format as Figure 9. In this interval, there is a burst on 1 July, but no activity on 30 June or 2 July. The most inclined fields occur on 30 June, but they are not accompanied by whistler mode bursts. The 1 July ambient fields do deviate from the horizontal by about 20° and are sufficient to allow the entry of waves into the ionosphere as in the previous example, but clearly 30 June

[18] The evidence for the necessity of inclined ionospheric magnetic fields for the access of whistler mode

Figure 7. By component of the 128-Hz field in spacecraft coordinates measured on the outboard and inboard sensors and the difference between the two readings. The absence of the burst on the bottom trace indicates that the burst in Figure 6 is not due to spacecraft sources.

Figure 8. Venus Express orbit together with projections of 8-s samples of the magnetic field once per minute in the x-z plane in Venus Solar Orbital coordinates. Magnetic field is seen to be dipping into atmosphere around time of burst of noise.

4 of 8

E00B05

RUSSELL ET AL.: LIGHTNING ON VENUS

E00B05

Figure 9. Three consecutive periapsis passes on 8, 9, and 10 June. The left-hand panels show the angle of the field to the radial direction with 90° being horizontal as calculated from 1-s samples of the field. The right-hand panels show bandpass-filtered 8-ms magnetic field measurements covering the frequency range from 40 to 62 Hz. The component shown is from the y axis sensor in spacecraft coordinates. This is the sensor with the lowest noise level. waves from the atmosphere into the ionosphere is clear from our statistical survey. However, in order to understand why this access control occurs, we need to examine the expected characteristics of the electromagnetic waves that enter the

ionosphere from below. Figure 13 shows a cartoon of the wavefronts produced by an intracloud discharge in Venus’ atmosphere. The waves at first propagate with spherical wavefronts at the speed of light. When they encounter the

Figure 10. Three consecutive passes on 30 June, 1 July, and 2 July 2006. Comments of Figure 9 apply. 5 of 8

E00B05

RUSSELL ET AL.: LIGHTNING ON VENUS

E00B05

Figure 11. Three consecutive passes on 22, 23, and 24 December 2006. Comments of Figure 9 apply. ionosphere, the part of the wavefront that enters the ionosphere slows down considerably to a very small fraction of the velocity of light depending on the electron density and the strength of the magnetic field. The flat wavefront corresponds to vertical propagation of the wave. The

Figure 12. Peak-to-peak amplitudes of burst maxima on each pass versus maximum inclination of the magnetic field to the horizontal.

Figure 13. Cartoon of electric discharge in the Venus clouds and the resultant wavefronts of electromagnetic waves. When the wavefronts enter the ionosphere where the phase velocity decreases substantially, the wavefront flattens and the wavenormals become vertical.

6 of 8

E00B05

RUSSELL ET AL.: LIGHTNING ON VENUS

E00B05

Figure 14. Phase and group velocities of whistler mode waves at 100 Hz propagating at various angles to the magnetic field in a plasma with a density of 5000 cm 3 and a magnetic field strength of 20 nT. magnetic field in the ionosphere, as we have seen, tends to be horizontal so long as the interplanetary magnetic field is steady. Inclined fields can be produced by time variations of the magnetic fields that in turn can lead to interconnection of different field directions within the ionosphere. [19] Figure 14 shows the phase and group velocities of a whistler mode wave at 100 Hz propagating at various angles to the magnetic field in a plasma with a density of 5000 electrons cm 3 and a magnetic field strength of 20 nT. These velocities are calculated from the full dispersion relation for cold plasma waves [Stix, 1962]. On the left is the phase velocity. This wave does not travel perpendicular to the magnetic field. In fact, there is a cone of nonpropagation of the order of 10°, an angle similar to the angle of inclination needed for the waves to reach Venus Express. On the right is the group velocity in a similar display. Once the energy couples to the magnetic field, it is strongly guided by the magnetic field. Thus, our observations are consistent with what one would expect for the waves in the ionosphere that would be produced as a result of atmospheric electrical discharges. [20] We can use the first year’s statistics of the magnetic field orientation and the number of discrete whistler wave bursts during the 2-min sampling intervals to make a rough comparison with terrestrial rates. If we use the conservative assumption that access to the ionosphere occurs only when the field is inclined 15° or more to the horizontal, then the statistics of cone angle occurrence indicates that access is allowed only 25% of the time. In the first Venus year of operation up to 16 December 2006, we received 12,223 s of 128-Hz data that could be analyzed. During this interval, 61 individual bursts of noise were detected, giving a rate of 0.005 per second. Normalizing this rate for access time, i.e., the duration of the fields inclined more than 15° to the horizontal, we get a rate of 0.02 per second below the clouds. If the spacecraft can see a circle of 200-km radius in the clouds below it, a distance equal to its height above them, it is seeing only 0.027% of the surface. If we use this ratio to normalize our observations, we get a global rate of 18 strokes s 1 or 20% the terrestrial rate. We do not know if our region of study near Venus’ polar vortex is representative of the entire planet or if our area of seeing is correct, but these numbers do tell us that lightning is a significant and important process in the Venus atmosphere. Finally, we note

that our measurements were gathered over a full Venus year covering all local times. While we would not necessarily expect the results of our observations to agree with the rate seen on a brief Cassini swing-by, observing radio waves and not the whistler mode waves studied here, we do agree qualitatively with the result of Gurnett et al. [2001] that the rate at Venus may be lower than on Earth.

6. Summary [21] In summary, Venus Express observes bursts of whistler mode noise in the Venus ionosphere. These waves are almost circularly plane-polarized and propagate at a small angle to the magnetic field. They were seen only when the local ionospheric magnetic field was inclined 17° or more to the horizontal. Such a cone of evanescence would be expected for electromagnetic radiation produced in the atmosphere and refracted vertically as it enters the ionosphere. The observations are consistent with lightning being prevalent in the Venus atmosphere and creating the whistler mode waves seen in the ionosphere, both those observed with the electric antenna on Pioneer Venus at low latitudes and those detected with the Venus Express magnetometer at high latitudes. [22] Acknowledgments. This work was supported by the National Aeronautics and Space Administration under research grant NNG06GC62G.

References Borucki, W. J., et al. (1991), Pioneer Venus Orbiter search for Venusian lightning, J. Geophys. Res., 96(A7), 11,033 – 11,043, doi:10.1029/ 91JA01097. Gurnett, D. A., et al. (1991), Lightning and plasma wave observations from the Galileo fly by of Venus, Science, 253(5027), 1522 – 1525, doi:10.1126/science.253.5027.1522. Gurnett, D. A., et al. (2001), Non-detection at Venus of high-frequency radio signals characteristic of terrestrial lightning, Nature, 409(6818), 313 – 315, doi:10.1038/35053009. Hansell, S. A. W. K. W., et al. (1995), Optical detection of lightning on Venus, Icarus, 117, 345 – 351, doi:10.1006/icar.1995.1160. Krasnopolsky, V. A. (1983), Lightnings and nitric oxide on Venus, Planet. Space Sci., 31, 1363 – 1369, doi:10.1016/0032-0633(83)90072-7. Ksanfomaliti, L. V. (1983), Electrical activity of the atmosphere of Venus: I. Measurements on descending probes, Kosm. Issled., 21, 279 – 296. Leinweber, H., C. T. Russell, K. Torkar, J. L. Zhang, and V. Angelopoulos (2008), An advanced approach to finding magnetometer zero levels in the interplanetary magnetic field, Meas. Sci. Technol., 19, doi:10.1088/09570233/19/5/055104.

7 of 8

E00B05

RUSSELL ET AL.: LIGHTNING ON VENUS

Piccioni, G., et al. (2007), South-pole features on Venus similar to those near the north pole, Nature, 450(7170), 637– 640, doi:10.1038/nature06209. Russell, C. T. (1991), Venus lightning, Space Sci. Rev., 55, 317 – 356. Russell, C. T., et al. (1989), VLF bursts in the night ionosphere of Venus: Estimates of the Poynting flux, Geophys. Res. Lett., 16(6), 579 – 582, doi:10.1029/GL016i006p00579. Russell, C. T., R. J. Strangeway, and T. L. Zhang (2006), Lightning detection on the Venus Express Mission, Planet. Space Sci., 54, 1344 – 1351, doi:10.1016/j.pss.2006.04.026. Russell, C. T., T. L. Zhang, M. Delva, W. Magnes, R. J. Strangeway, and H. Y. Wei (2007), Lightning on Venus inferred from whistler-mode waves in the ionosphere, Nature, 450(7170), 661 – 662, doi:10.1038/ nature05930. Russell, C. T., T. L. Zhang, R. J. Strangeway, H. Y. Wei, M. Delva, and W. Magnes (2008), Electromagnetic waves observed by Venus Express at periapsis: Detection and analysis techniques, Adv. Space Res., 41(1), 113 – 117, doi:10.1016/j.asr.2007.08.032. Sagdeev, R. Z., et al. (1986), Overview of VEGA Venus balloon in-situ meteorological measurements, Science, 231(4744), 1411 – 1414, doi:10.1126/science.231.4744.1411. Scarf, F. L., et al. (1980), Lightning on Venus: Orbiter detection of whistler signals, J. Geophys. Res., 85(A13), 8158 – 8166, doi:10.1029/ JA085iA13p08158.

E00B05

Stix, T. H. (1962), Theory of Plasma Waves, 283 pp., McGraw-Hill, New York. Strangeway, R. J., et al. (1993), Observation of intense wave bursts at very low altitudes within the Venus nightside ionosphere, Geophys. Res. Lett., 20(23), 2771 – 2774, doi:10.1029/93GL02702. Taylor, W. W. L., et al. (1979), Absorption of whistler mode waves in the ionosphere of Venus, Science, 205(4401), 112 – 114, doi:10.1126/ science.205.4401.112. Zhang, T. L., et al. (2006), Magnetic field investigation of the Venus plasma environment: Expected new results from Venus Express, Planet. Space Sci., 54, 1336 – 1343, doi:10.1016/j.pss.2006.04.018. Zhang, T. L., et al. (2008), Initial Venus Express magnetic field observations of the Venus bow shock location at solar minimum, Planet. Space Sci., 56, 785 – 789, doi:10.1016/j.pss.2007.09.012. C. T. Russell and H. Y. Wei, Institute of Geophysics and Planetary Physics, University of California, 603 Charles Young Drive East, 3845 Slichter Hall, Los Angeles, CA 90095-1567, USA. ([email protected]. edu) T. L. Zhang, Space Research Institute, Austrian Academy of Sciences, Schmiedlstrasse 6, A-8042 Graz, Austria.

8 of 8

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, E00B02, doi:10.1029/2008JE003081, 2008

for

Full Article

Morphology and dynamics of Venus oxygen airglow from Venus Express/Visible and Infrared Thermal Imaging Spectrometer observations R. Hueso,1 A. Sa´nchez-Lavega,1 G. Piccioni,2 P. Drossart,3 J. C. Ge´rard,4 I. Khatuntsev,5 L. Zasova,5 and A. Migliorini2 Received 21 January 2008; revised 21 March 2008; accepted 2 May 2008; published 26 July 2008.

[1] Images obtained by the Visible and Infrared Thermal Imaging Spectrometer

(VIRTIS)-M channel instrument onboard Venus Express have been used to retrieve maps and apparent motions of the O2 (1D) infrared nightglow on Venus at 1.27 mm. The nightglow distribution is highly inhomogeneous with the regions of brightest emission generally located at low latitudes near the midnight meridian. Unexpectedly some orbits show also intense airglow activity over the south polar region. The spatially resolved airglow is spectacularly variable not only in its morphology and intensity but also in the apparent motions of the airglow small- and large-scale structures. Visual tracking of the bright features allowed to obtain mean zonal and meridional motions related to the subsolar to antisolar circulation. The zonal velocity is dominated by an intense prograde jet (contrary to the retrograde planetary rotation) from dawn to midnight extending up to 22 hours in local time with lower velocities and reversed sign from dusk. Typical zonal velocities range between +60 (prograde) to 50 (retrograde) m/s, whereas most meridional velocities range from 20 (poleward) to +100 m/s (equatorward) with an average meridional circulation of +20 m/s toward low latitudes. The brightest small-scale (100 km) features appear correlated with locations of apparent convergence which may be a signature of compression and downwelling, whereas this is not evident for the largescale structures suggesting slow subsidence over large areas mixed with horizontal motions. We argue that part of the tracked motions are representative of real motions at the mesosphere over an altitude range of 95–107 km. Citation: Hueso, R., A. Sa´nchez-Lavega, G. Piccioni, P. Drossart, J. C. Ge´rard, I. Khatuntsev, L. Zasova, and A. Migliorini (2008), Morphology and dynamics of Venus oxygen airglow from Venus Express/Visible and Infrared Thermal Imaging Spectrometer observations, J. Geophys. Res., 113, E00B02, doi:10.1029/2008JE003081.

1. Introduction [2] The circulation of the upper atmosphere of Venus is traditionally decomposed into two distinct flow patterns at different altitudes: A subsolar-to-antisolar (SS-AS) thermospheric circulation cell driven by solar heating, stable at levels above the mesosphere (z > 120 km [Bougher et al., 1997], and a superrotating zonal (SZ) flow in the planetary rotation sense at the upper troposphere (z  65 km) [see, e.g., Gierasch et al., 1997; Markiewitz et al., 2007; Limaye, 1 Departamento de Fı´sica Aplicada I, E.T.S. Ingenieros, Universidad del Paı´s Vasco, Bilbao, Spain. 2 Istituto di Astrofisica Spaziale e Fisica Cosmica, INAF-IASF Roma, Rome, Italy. 3 Laboratoire d’ Etudes Spatiales et d’ Instrumentation en Astrophysique, Observatoire de Paris, Universite´ Paris-Diderot, CNRS, UPMC, Meudon, France. 4 Laboratoire de Physique Atmosphe´rique et Plane´taire, Universite´ de Lie`ge, Lie`ge, Belgium. 5 Space Research Institute, Moscow, Russia.

Copyright 2008 by the American Geophysical Union. 0148-0227/08/2008JE003081$09.00

2007; Sa´nchez-Lavega et al., 2008]. In the transition region, the mesosphere (70 < z < 120 km), the interaction of both motions results in a complex circulation that greatly varies over time [Lellouch et al., 1997; Bougher et al., 2006]. The global circulation in the mesosphere is probably influenced by a number of physical processes such as breaking gravity waves [Alexander, 1992] and diffusive processes, but their effects over the mesospheric circulation are not well understood or quantified [Bougher et al., 2006]. Additionally, the return branch of the SS-AS upper circulation must lie somewhere in the mesosphere but its location has not been clearly identified. [3] While the dynamics of the mesosphere of Venus is hardly accessible to in situ measurements it can be studied by means of Doppler shifts in certain absorption lines sensitive to mesospheric levels [e.g., Clancy et al., 2007; Widemann et al., 2007] but this technique lacks spatial resolution, requires a sophisticated analysis of the observations and a precise treatment of systematic errors. A complimentary technique suited for exploration from a spacecraft is the study of nightglow molecular emissions which trace the motions of different chemical species in the

E00B02

1 of 7

E00B02

HUESO ET AL.: DYNAMICS OF VENUS OXYGEN AIRGLOW

upper atmosphere with the drawback that the emissions are coupled with photochemical processes. [4] Among the many nonthermal emissions produced in Venus, the oxygen airglow at 1.27 mm is the most intense. Its was first detected in ground-based observations of Venus by Connes et al. [1979]. This airglow is produced by the recombination of oxygen atoms dissociated by photolysis of CO2 at thermospheric altitudes on the sunlit hemisphere. In the upper mesosphere (95 – 110 km) three-body recombination of O atoms leads to O2 formation in a particular excited state, a1Dg, which is followed by airglow emission as the molecule relaxes to its X 3Sg ground state. The radiative lifetime of O2 singlet state is 70 min. [Lellouch et al., 1997; Miller et al., 2001]. The O2 airglow is observed in the night-side hemisphere at 1.267 microns [Allen et al., 1992; Bougher and Borucki, 1994] and with far less intensity in O2 Herzberg II visible wavelengths [Krasnopolsky et al., 1976; Bougher et al., 1997]. [5] It was soon realized that the intense oxygen IR nightglow is highly variable from day to day [Crisp et al., 1996], its intensity distribution is generally not symmetric in latitude [Allen et al., 1992], often exhibits multiple local maxima, and shows variations on timescales as short as 1 hour [Lellouch et al., 1997]. The related nitric oxide nightglow emission was mapped during the Pioneer Venus mission [Stewart et al., 1980] statistically presenting larger brightness close to the Equator and shifted by about 2 hours from midnight toward dawn. Both nightglows are decoupled in height providing information about different vertical levels: 95– 105 km for O2 airglow (G. Piccioni et al., Venus oxygen airglow vertical profile, submitted to Journal of Geophysical Research, 2008) and 115 km for NO [Ge´rard et al., 2008b]. [6] In this work we have used night side nadir observations from the Visible and Infrared Thermal Imaging Spectrometer (VIRTIS) [Drossart et al., 2007a] on board the Venus Express orbiter [Titov et al., 2006; Svedhem et al., 2007] to construct maps of the Venus atmosphere in the O2 emission band. The airglow maps were used to study the morphology and temporal variability of the airglow. In those orbits where the airglow is particularly intense we also measured the displacement of ‘‘cloud-like’’ airglow features. [7] Preliminary measurements of apparent motions were presented by Drossart et al. [2007b] for selected orbits. Ge´rard et al. [2008a] gave a precise account of the average distribution of the O2 nightglow. Other works in this special section (Piccioni et al., submitted manuscript, 2008) present results obtained in limb view which lead to vertical profiles of the airglow emission.

2. Observations and Methods [8] We analyzed data obtained during 28 consecutive orbits between 1 July 2006 and 28 July 2006, one orbit on 6 August 2006 presenting a good view of the equator, six orbits between December 2006 and January 2007 and five orbits from 7 April 2007 to 11 April 2007 that cover the polar region in higher detail than the rest of the VIRTIS data set. The data were selected according to the availability of high-resolution observations, fast repetition in each orbit (two or more images of the same region separated by 1hour) and the strong airglow emission present during most of the

E00B02

selected orbits. The Sun F10.7 flux (a convenient measurement of the solar activity in radio wavelengths commonly used as a proxy of the Sun’s irradiance in the Extreme UltraViolet) averaged over July 2006, August 2006, January 2007 and April 2007 was 76  10 22, 79  10 22, 84  10 22 and 72  10 22 W/m2 Hz respectively, reflecting homogeneous conditions for the Sun’s activity close to the solar minimum. [9] Oxygen airglow at 1.27 mm superimposes with thermal radiation coming from the lower atmosphere and partially filtered by the lower clouds. In order to subtract the thermal contribution from the total airglow signal we compare VIRTIS fluxes measured in the 1.26 – 1.28 mm wavelength range to the flux on the 1.18 mm observation window which provides a nearby region with flux dominated by the lower thermal emission. In addition, O2 photons emitted downward are subsequently backscattered by the underlying clouds and enhance the measured airglow radiation. A correction factor of 2.7 was used for nadir observations and geometry effects were corrected to first order by multiplying each pixel by the cosine of the emission angle [Crisp et al., 1996]. This crude estimation of brightness emission is adequate for VIRTIS observations of subpolar and midlatitudes (emission angles lower than 40°) but may fail in the few observations containing information about the tropics and the equator. Finally, depending on the latitudes being sampled we projected the images in longitude-latitude maps or polar stereographic projections to compare images obtained from different observational geometries. [10] Images separated by 1 hour allowed to track and retrieve the displacements of the airglow features in 23 of the 40 orbits, typically the orbits with higher intensity airglow structures, being the remaining orbits characterized by low intensity of the airglow or low contrast of the observed structures. The apparent motions of distinct airglow features were measured visually using the PLIA software [Peralta et al., 2005].

3. Morphology of the Oxygen Airglow and Temporal Variability 3.1. Low and Mid Latitudes [11] As a consequence of the quasi-polar VEX orbit, the airglow activity at low latitudes is only partially covered by VIRTIS observations with a resolution of 100 km pix 1 in some cases, while mid to subpolar latitudes are observed on many orbits with spatial resolution down to 15 km pix 1. [12] The statistical distribution of Venus oxygen nightglow is inhomogeneous with the regions of brightest emission located at low latitudes near and dawnward of the midnight meridian [Ge´rard et al., 2008b]. Maps of airglow brightness distribution of different regions are presented in Figure 1 using data from orbits that sample nearly simultaneously most of the Southern Hemisphere. On this data set the equatorial region at midnight always presents high values of brightness. Large structures with typical sizes of 3000 km extending from the equator to midlatitudes are often observed. In most cases the large-scale structures observed on Figure 1a are similar to the global structures seen in the global map of airglow brightness presented by Ge´rard et al. [2008a] and constructed by averaging several

2 of 7

E00B02

HUESO ET AL.: DYNAMICS OF VENUS OXYGEN AIRGLOW

E00B02

Figure 1. Airglow intensity map obtained during orbits (a) 87 (16 July 2006) and (b) 91 (20 July 2006). The wide coverage of the maps is close to half the night-side hemisphere and was obtained using data from three different VIRTIS data cubes for each map. The mean brightness of the sampled regions close to the equatorial midnight can change from 5 MR to 1 in a single day but the large values are more common. hundred maps from different dates. Remarkably, though largely expected, the mean airglow brightness in VIRTIS images (centered at the same latitude and local time and covering areas as big as 1.5  107 km2) changes dramatically from one orbit to other. Changes observed on Figure 1 are sometimes observed on timescales as short as 1 day over the VIRTIS data set. On the contrary, large-scale structures as seen on Figure 1a are stable on timescales of 2 hours or longer indicating sustained formation of excited oxygen molecules. 3.2. Polar Latitudes [13] Ge´rard et al. [2008a] give an average value of O2 airglow brightness at 70°S of 1.2 MegaRayleighs (MR). In contrast, intense airglow activity can also arise over polar latitudes in particular orbits. This was unexpected since the polar regions are far from the antisolar point and should contain lower densities of oxygen atoms. Figure 2 shows two images of a sequence of VIRTIS observations of the most intense airglow brightness over polar latitudes and viewed with an emission angle lower than 20° for all the sequence. A gaussian fit to the brightness temporal evolution of this structure following its motion gives a time decay of 3 hours with a peak brightness of 7 MR over a 100 km

elongated structure, moving toward lower latitudes with a meridional velocity of v = 50 ± 10 m/s. The brightness decay in time of this structure was about 2.5 times the radiative lifetime of the oxygen airglow indicating continuous production of excited molecules over this small region. [14] The period from 7 to 11 April 2007 was used for detailed research of the polar area finding intense airglow activity in all the VIRTIS images and sustained during all orbits. Figure 3 shows a sequence of observations where apparent motions of the airglow lead to a polar to low latitudes apparent motion with v  50 ± 5 m/s. Detailed motions in the whole movie during this orbit (composed of 12 images at time steps of 30 min to 1 hour) shows that the airglow evolves by a combination of apparent motions at small scale and activation and deactivation of airglow emission structures over extended regions.

4. Apparent Motions and Mesospheric Circulation [15] Figure 4 shows apparent motions tracked over airglow structures on different dates, latitudes and local times. The panels show that the morphology of the airglow is related with the apparent motions tracked. Large-scale

3 of 7

E00B02

HUESO ET AL.: DYNAMICS OF VENUS OXYGEN AIRGLOW

E00B02

morning hours (0600 0200 local time) and reversed in direction from 1800 to 2000 local time. The meridional motions are more intense at subpolar latitudes (70°S) with average velocities of v  20 m/s, decreasing toward lower latitudes and reaching a minimum value of v  5 m/s at 15°S and increasing again at equatorial latitudes. [18] In the orbits with high intensities of the airglow emission (notably on the equatorial case and the five polar orbits) the displacements of the airglow brightness patterns seem to take place at different altitude levels since the images show a mixture of structures moving with different speeds that visually seem to superimpose. In support of this interpretation, limb observations of airglow activity some-

Figure 2. Polar projections and time evolution of the brightest airglow structure seen on polar latitudes. Longitudes are given in terms of local times. Observations correspond to orbit 80, 9 July 2006, and the time difference between both observations is 3 hours. structures move slowly on one direction (generally perpendicular to the airglow brightness front) while smaller-scale structures move more rapidly in the direction defined by the front (see Figure 4a). [16] The equatorial region covered on Figure 4d looks turbulent with bright features colliding suggesting that the observed motions may come from a variety of different altitude levels. Some of the brightest structures (notably on Figure 4e, but also on Figures 4c and 4d and particular regions on Figure 4a) correlate with locations where convergence of the apparent motions is present. This behavior may not apply to the larger-scale structures, since they appear surrounded by far less intense airglow features and thus, it is not possible to retrieve motions around them. [17] Figure 5 shows the mean horizontal circulation in Venus’ night-side hemisphere obtained by spatially and temporally averaging the ensemble of apparent motions obtained in 23 orbits with 1000 tracked motions. The upper part of Figure 5 also shows the local time-latitude location of the specific regions displayed in Figure 4. The large variability in intensity and direction of the velocities, which sometimes can be as high as 100 m/s, results in low values of the average speeds. The most remarkable feature in this plot is a repeatedly observed jet at 35°S with an average zonal speed of u = 30 m/s opposite to the planetary rotation and with motions twice as intense in the early

Figure 3. Polar projections and time evolution of the 1.267 mm radiance over orbit 355 obtained on 27 August 2007 at local times: (a) 1539, (b) 1739, (c) 1939, and (d) 2209. The brightest airglow features have an intensity of 1.8 MR on the first image and longitudes are given in terms of local times. The global apparent motions are toward higher latitudes. Some of the structures like the bright arm at 68°S in Figure 3d appear suddenly and are not related with horizontal motions but with the sudden airglow brightening of new regions.

4 of 7

HUESO ET AL.: DYNAMICS OF VENUS OXYGEN AIRGLOW

E00B02

E00B02

Figure 4. Average apparent motions and correlation with the airglow intensity for selected orbits: (a) orbit 96 obtained on 25 July 2006; (b) orbit 99 obtained on 28 July 2006; (c) orbit 76 obtained on 5 July 2006; (d) orbit 108 obtained on 6 August 2006; (e) orbit 84 obtained on 13 July 2006; and (f) orbit 80 obtained on 9 July 2006. The maps cover a significant portion of the local time-latitude plane showing repeatedly observed motions, like the zonal jet at 30°S – 45°S opposite to the planetary rotation (Figures 4a, 4b, 4e, and 4f). Opposite motions between 45°S and 75°S that seem part of a close circulation are observed at different local times on different orbits (Figure 4b); meridional motions are toward the pole (Figure 4a) but generally directed toward low latitudes (Figures 4a, 4b, 4c, and 4f). The few observations of the equator show chaotic motions with a mean prograde (in the planetary rotation sense) component (Figure 4d). Brightness emission given are only approximative for Figure 4d with emission angles 70°. times presents peaks at two different altitude levels, one at 96 ± 1 km altitude and a second peak at 103 –107 km altitude (Piccioni et al., submitted manuscript, 2008).

5. Discussion [19] The average motions obtained are in agreement with a subsolar to antisolar circulation at the IR oxygen airglow levels (96 km) with evident meridional motions from polar to low latitudes and large variability of the zonal and meridional components. [20] Caution must be taken in interpreting the airglow brightness patterns displacements as tracers of flow motions in the lower mesosphere. They could be influenced by the activation and/or deactivation of airglow emission at specific regions by other phenomena (i.e., gravity waves affecting atmospheric density, localized or extended regions of downwelling, atmospheric diffusivity, etc). [21] An additional level of complexity is introduced by the radiative lifetime (70 min) of the O2 singlet state which introduces difficulties in the interpretation of the observed rates of brightness variation. The e-folding time constant of any local change (decrease or increase) cannot be less than 70 min without advection of air masses containing different proportion of glowing O2 molecules. For airglow brightness decreasing more slowly the brightness reduction could be

caused by a combination of horizontal transport and a decrease of the flux of O atoms modifying the local O density and airglow brightness. In those locations where the airglow intensity increases there must be enhanced downward flux of O atoms. [22] The observations show two types of behavior for the Venus oxygen airglow temporal evolution. On the one hand, the large-scale 1000 –3000 km structures survive at least for a few hours (longer than the radiative decay time) but change dramatically from day to day. Since limb scans of O2 emission slightly depend on latitude with higher latitudes characterized by higher altitudes of the airglow (Piccioni et al., submitted manuscript, 2008) these large structures may descend progressively as they travel from polar latitudes to the equator, compressing and increasing the concentration of free O atoms. On the other hand, smallscale bright structures with smaller scales (100 km) seem related with the apparent convergence of motions tracked in the airglow. In fact, the time evolution of the airglow maps seems composed of two types of motions, the large-scale structures moving like ‘‘fronts’’ or waves, and the bright structures immersed on them or isolated, apparently moving like passive tracers. We argue that at least part of the later motions are real motions at the mesosphere as the newly formed oxygen molecules are advected by the mesospheric circulation.

5 of 7

E00B02

HUESO ET AL.: DYNAMICS OF VENUS OXYGEN AIRGLOW

E00B02

Figure 5. Mean apparent circulation of the mesospheric oxygen airglow brightness. (a) Averages of all apparent motions tracked on the airglow and also marks the sampled regions appearing on Figure 4. A reference vector of 100 m/s is included on the upper left corner of the map. (b) A schematic model of the mean circulation of Venus mesosphere from these measurements. The background image corresponds to a map of airglow brightness obtained on orbit 93, 22 July 2007, that can be compared with those on Figure 1. Regions with no arrows indicate regions with not much available data. The apparent rotating motions indicated at 50°S are found at a variety of local times depending on the particular orbit. The jet from dawn to midnight is a repetitive structure while the jet from dusk to midnight is less regular from one orbit to another. [23] The brightest airglow patches may be regions of strong subsidence (and local adiabatic warming) associated with downdrafts that increase the volume concentration of O atoms. If this is correct, spectral maps of the O2 IR nightglow should show a positive correlation between the brightness and the rotational temperature of the emission line. Partial confirmation of this point comes from stellar occultation experiments performed by Venus Express [Bertaux et al., 2007] which result on warm temperatures at the airglow emission levels (95 – 100 km). Ground-based observations have been made to correlate the rotational temperature and brightness with results sometimes indicating such a correlation [Ohtsuki et al., 2005] and sometimes not with some dependence on the morphology of the airglow region [Bailey et al., 2007]. [24] Although the present work does not fully disentangle the different contributions to the airglow variability and therefore is not able to fully retrieve the mesospheric true motions it does show that such an approach is possible.

VIRTIS-M observations obtained with shorter time differences than the radiative lifetime of the oxygen airglow (70 min) will partially help to solve this problem. Although part of the airglow motions tracked on this work were obtained on image pairs separated by 30 min the short time difference between the images translate into larger errors in the derived motions and most of the data corresponds to images separated by 1 hour. Both data sets produce similar results on terms of the mean derived circulation. [25] Further analysis should concentrate on identifying any correlation between local temperature anomalies and O2 intensities. Also, detailed comparison with Venus GCMs should allow to better constrain the role of gravity waves as drivers of the airglow variability. The Venus Express Extended mission should run during 2008 – 2009 with higher levels of solar activity. According to NOAA Space Weather Prediction Center (SWPC) a reasonable estimate is an increase of a 60% of the F10.7 index for the end of 2009 (http://www.sec.noaa.gov/SolarCycle/SC24). This

6 of 7

E00B02

HUESO ET AL.: DYNAMICS OF VENUS OXYGEN AIRGLOW

will allow a higher airglow activity and a better sampling of Venus mesosphere through its spatial distribution and variability. [26] Over the next years the combination of results from Venus Express instruments (VIRTIS O2 IR maps, SPICAV thermal profiles and NO nightglow observations, VeRa thermal data combined with cyclostrophic balance over mid to high latitudes and cloud top wind tracking with VMC and VIRTIS) and ground-based observations (nightglows and Doppler winds retrieval) will allow a significant improvement of our understanding of the mesospheric dynamics and the underlying mechanisms driving its variability. [27] Acknowledgments. Venus Express is a mission of the European Space Agency. We gratefully thank all members of the ESA Venus Express project and of the VIRTIS technical team. This work has been funded by Spanish MEC AYA2003-03216, with FEDER support and Grupos UPV 15946/2004 and is supported by the CNES and ASI national space agencies. R.H. acknowledges a Ramo´n y Cajal contract from MEC. J.C.G. acknowledges funding from the Belgian Fund for Scientific Research and the PRODEX program managed by the European Space Agency with the help of the Belgian Federal Space Science Office.

References Alexander, M. J. (1992), A mechanism for the Venus thermospheric superrotation, Geophys. Res. Lett., 19, 2207 – 2210. Allen, D., D. Crisp, and V. S. Meadows (1992), Variable oxygen airglow on Venus as a probe of atmospheric dynamics, Nature, 359, 516 – 519. Bailey, J. A., V. S. Meadows, S. Chamberlain, A. Simpson, and D. Crisp (2007), Variability of the Venus Oxygen Airglow, Bull. Am. Astron. Soc., 38, 526. Bertaux, J. L., et al. (2007), A warm layer in Venus’ cryosphere and highaltitude measurements of HF, HCl, H2O and HDO, Nature, 450, 646 – 649. Bougher, S. W., and W. J. Borucki (1994), Venus O2 visible and IR nightglow: Implications for lower thermosphere dynamics and chemistry, J. Geophys. Res., 99, 3759 – 3776. Bougher, S. W., M. J. Alexander, and H. G. Mayr (1997), Upper atmosphere dynamics: Global circulation and gravity waves, in Venus II: Geology, Geophysics, Atmospheres, and Solar Wind Environment, edited by S. W. Bougher, D. M. Hunten, and R. J. Philips, Univ. of Ariz. Press, Tucson. Bougher, S. W., S. Rafkin, and P. Drossart (2006), Dynamics of the Venus upper atmosphere: Outstanding problems and new constraints expected from Venus Express, Planet. Space Sci., 54, 1371 – 1380. Clancy, R. T., B. J. Sandor, and G. H. Moriarty-Schieven (2007), Dynamics of the Venus upper atmosphere: Global-temporal distribution of winds, temperature, and CO at the Venus mesopause, Bull. Am. Astron. Soc., 39, abstract 61.07. Connes, P., J. F. Noxon, W. A. Traub, and N. P. Carleton (1979), O2 (1D) emisssion in the day and night airglow of Venus, Astrophys. J., 233, L29 – L32. Crisp, D., V. S. Meadows, B. Be´zard, C. deBergh, J.-P. Maillard, and F. P. Mills (1996), Ground-based near-infrared observations of the Venus nightside: 1.27 mm O2 (a1Dg) airglow from the upper atmosphere, J. Geophys. Res., 101, 4577 – 4594. Drossart, P., et al. (2007a), Scientific goals for the observation of Venus by VIRTIS on ESA/Venus express mission, Planet. Space Sci., 55, 1653 – 1672. Drossart, P., et al. (2007b), Venus upper atmospheric emissions from VIRTIS spectral imaging observations, Nature, 450, 641 – 645.

E00B02

Ge´rard, J.-C., A. Saglam, G. Piccioni, P. Drossart, C. Cox, S. Erard, R. Hueso, and A. Sa´nchez-Lavega (2008a), Distribution of the O2 infrared nightglow observed with VIRTIS on board Venus Express, Geophys. Res. Lett., 35, L02207, doi:10.1029/2007GL032021. Ge´rard, J.-C., C. Cox, A. Saglam, J.-L. Bertaux, E. Villard, and C. Nehme´ (2008b), Limb observations of the ultraviolet nitric oxide nightglow with SPICAV on board Venus Express, J. Geophys. Res., doi:10.1029/ 2008JE003078, in press. Gierasch, P. J., et al. (1997), The general circulation of the Venus atmosphere: An assessment, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher, D. M. Hunten, and R. J. Philips, pp. 459 – 500, Univ. of Ariz. Press, Tucson. Krasnopolsky, V. A., A. A. Krysko, V. N. Rogachev, and V. A. Parshev (1976), Spectroscopy of the Venus night airglow from the Venera 9 and 10 satellites, Cosmic Res., 14, 789 – 795. Lellouch, E., T. Clancy, D. Crisp, A. Kliore, D. Titov, and S. W. Bougher (1997), Monitoring of mesospheric structure and dynamics, in Venus II: Geology, Geophysics, Atmospheres, and Solar Wind Environment, edited by S. W. Bougher, D. M. Hunten, and R. J. Philips, pp. 295 – 324, Univ. of Ariz. Press, Tucson. Limaye, S. S. (2007), Venus atmospheric circulation: Known and unknown, J. Geophys. Res., 112, E04S09, doi:10.1029/2006JE002814. Markiewitz, W. J., D. V. Titov, S. S. Limaye, H. U. Keller, N. Ignatiev, R. Jaumann, N. Thomas, H. Michalik, R. Moissl, and P. Russo (2007), Morphology and dynamics of the upper cloud layer of Venus, Nature, 450, 633 – 635. Miller, H. C., J. E. McCord, J. Choy, and G. D. Hager (2001), Measurement of the radiative lifetime of O2 (a1Dg) using cavity ring down spectroscopy, J. Quant. Spectrosc. Radiat. Transfer, 69, 305 – 325. Ohtsuki, S., N. Iwagami, H. Sagawa, Y. Kasaba, M. Ueno, and T. Imamura (2005), Ground-based observation of the Venus 1.27-mm O2 airglow, Adv. Space Res., 36, 2038 – 2042. Peralta, J., R. Hueso, N. Barrado, and A. Sa´nchez-Lavega (2005), Introducing PLIA: The planetary laboratory for image analysis, Bull Am. Astron. Soc., 37, 653. Sa´nchez-Lavega, A., et al. (2008), Variable winds on Venus mapped in three dimensions, Geophys. Res. Lett., 35, L13204, doi:10.1029/ 2008GL033817. Stewart, A. I. F., J. C. Gerard, D. W. Rusch, and S. W. Bougher (1980), Morphology of the Venus ultraviolet night airglow, J. Geophys. Res., 85, 7861 – 7870. Svedhem, H., et al. (2007), Venus Express—The first European mission to Venus, Planet. Space Sci., 55, 1636 – 1652. Titov, D. V., et al. (2006), Venus Express science planning, Planet. Space Sci., 54, 1279 – 1297. Widemann, T., E. Lellouch, and A. Campargue (2007), New wind measurements in Venus’ lower mesosphere from visible spectroscopy, Planet. Space Sci., 55, 1741 – 1756. P. Drossart, Laboratoire d’ Etudes Spatiales et d’ Instrumentation en Astrophysique, Observatoire de Paris, Universite´ Paris-Diderot, CNRS, UPMC, 5 place Jules Janssen, F-92195 Meudon, France. (pierre.drossart@ obspm.fr) J. C. Ge´rard, Laboratoire de Physique Atmosphe´rique et Plane´taire, Universite´ de Lie`ge, 5 Avenue de Cointe, B-4000 Lie`ge, Belgium. ([email protected]) R. Hueso and A. Sa´nchez-Lavega, Departamento de Fı´sica Aplicada I, E.T.S. Ingenieros, Universidad del Paı´s Vasco, Alameda Urquijo s/n, E-48013 Bilbao, Spain. ([email protected]; [email protected]) I. Khatuntsev and L. Zasova, Space Research Institute, Profsouznaya 84/ 32, Moscow, Russia. ([email protected]; [email protected]) A. Migliorini and G. Piccioni, Istituto di Astrofisica Spaziale e Fisica Cosmica, INAF-IASF Roma, Via del Foso del Cavaliere 100, I-00133 Rome, Italy. ([email protected]; giuseppe.piccioni@ iasf-roma.inaf.it)

7 of 7

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114, E00B29, doi:10.1029/2008JE003112, 2009

for

Full Article

Limb observations of CO2 and CO non-LTE emissions in the Venus atmosphere by VIRTIS/Venus Express G. Gilli,1 M. A. Lo´pez-Valverde,1 P. Drossart,2 G. Piccioni,3 S. Erard,2 and A. Cardesı´n Moinelo3 Received 11 February 2008; revised 29 October 2008; accepted 13 November 2008; published 7 March 2009.

[1] We report and analyze here observations of strong infrared emissions from the limb of

the Venus upper atmosphere during daytime, taken by the Visible and Infrared Thermal Imaging Spectrometer (VIRTIS) aboard Venus Express. We focus on the measurements taken during the first 4 months of nominal operations. The emissions observed at 4.3 mm and at 2.7 mm are attributed to CO2 fluorescence of solar radiation and are detected up to about 160 km and 130 km, respectively, while the CO fluorescence at 4.7 mm is observed up to about 120 km. The emissions are detected in both the channels of VIRTIS, at different spatial and spectral resolutions (resolving powers about 1800 and 400), for the periapsis and the apoapsis of the Venus Express orbit. From these data sets we built up 2-D maps of the emissions as well as vertical profiles, which are then studied in order to characterize their variations with geophysical parameters, like solar illumination and emission altitude. Several analyses are performed in order to understand the VIRTIS behavior, to determine systematic effects in the data, and to propose appropriate corrections. We also present comparisons with a theoretical nonlocal thermodynamic equilibrium (non-LTE) model of the Venus upper atmosphere. The agreement is very encouraging, in general, and the main variability observed in the data, with solar zenith angle and altitude, can be understood with the model. We conclude that the present data set opens brilliant perspectives for deriving densities and rotational temperatures in the upper mesosphere and lower thermosphere of Venus. Citation: Gilli, G., M. A. Lo´pez-Valverde, P. Drossart, G. Piccioni, S. Erard, and A. Cardesı´n Moinelo (2009), Limb observations of CO2 and CO non-LTE emissions in the Venus atmosphere by VIRTIS/Venus Express, J. Geophys. Res., 114, E00B29, doi:10.1029/2008JE003112.

1. Introduction [2] The European mission Venus Express arrived at Venus in April 2006, and started its scientific phase in June 2006, being the first mission in more than 20 years to study the Venus atmosphere systematically from orbit [Svedhem et al., 2007]. In contrast to the Earth’s mesosphere and lower thermosphere, the equivalent density layers on Venus from about 90 to 150 km, are regions scarcely observed using remote sounding in the infrared (IR). The most abundant species CO2, is well known to present strong IR vibrationalrotational bands, and more than 25 years ago, Pioneer Venus made systematic nadir observations using the fundamental band of CO2 at 15 mm in order to retrieve atmospheric temperatures in the mesosphere of Venus [Taylor et al., 1980]. Also, NIMS/Galileo permitted a sounding of the lower mesosphere temperature at 4.3 mm in nadir and nighttime [Roos-Serote et al., 1995]. However, at higher altitudes, these emissions are expected to be out of thermo1 2 3

Instituto de Astrofı´sica de Andalucı´a, CSIC, Granada, Spain. Observatoire de Paris, Meudon, France. IASF, INAF, Rome, Italy.

Copyright 2009 by the American Geophysical Union. 0148-0227/09/2008JE003112$09.00

dynamic equilibrium, with particularly large deviations during daytime, owing to solar pumping of the corresponding vibrational states [Dickinson, 1972; Deming et al., 1983; Rolda´n et al., 2000]. As a consequence, their inversion in terms of geophysical parameters is not straightforward, or in other words, their thermodynamic information content is limited [Lo´pez-Puertas and Taylor, 2001]. The understanding of these non-LTE emissions is important, not only for its own interest, but also they can give rise to a large radiative heating, or cooling, of the atmosphere. In the case of Venus, they are a key ingredient of the energy balance between 80 and 150 km [Bougher et al., 1994]. Rolda´n et al. [2000] showed that the CO2 bands at 4.3 mm and the ones at 2.7 mm dominate the solar heating at those altitudes; much less important are the CO absorptions at 4.7 mm, given the lower atmospheric abundance of CO. Further observational problems challenged an infrared sounding of these layers from the Venus orbit so far; among them, the pointing in the limb (the optimum geometry of observation at these low atmospheric densities) and the sensitivity of the infrared detectors. [3] Progress toward limb sounding of planetary atmospheres has been made during the last 2 decades, both theoretically and observationally. Triggered by ground-based

E00B29

1 of 19

E00B29

GILLI ET AL.: VIRTIS/VEX NON-LTE LIMB OBSERVATIONS

observations of CO2 laser bands at 10 mm, a number of nonLTE models for atmospheric IR emissions were developed for Mars and Venus in the 1980s [Deming et al., 1983; Gordiets and Panchenko, 1983; Stepanova and Shved, 1985]. Later, more comprehensive tools were developed in preparation for sounding of the Martian atmosphere in the IR during the 1990s [Lo´pez-Valverde and Lo´pez-Puertas, 1994], which were adapted later to the Venus atmosphere [Rolda´n et al., 2000]. These models, which included many more radiative bands and CO2 energy states, represent adequate tools for simulation and analysis of limb data. Some of the few non-LTE observations available so far, and amenable for such studies, were taken by the Near Infrared Mapping Spectrometer (NIMS) instrument [Carlson et al., 1992] on board the Galileo sounder, during its flyby of Venus in 1990 [Carlson et al., 1991]. A few limb spectra of the CO2 4.3-mm band were obtained during daytime and were explained recently using an improved version of the non-LTE model of Rolda´n and colleagues [Lo´pez-Valverde et al., 2007] (hereinafter LVEA). Recent analysis of CO2 non-LTE emissions have also been carried out for Mars using data from the Planetary Fourier Spectrometer (PFS) on board Mars Express [see Lo´pez-Valverde et al., 2005; Formisano et al., 2006], which helped to partially validate the non-LTE model. Regarding CO, non-LTE emission lines in the near-IR were first discovered in the Venus atmosphere about 20 years ago by de Bergh et al. [1988]. They were recently modeled and analyzed by Crovisier et al. [2006], who identified the fundamental CO(1-0) and the first hot CO(2-1) vibrational-rotational bands, derived rotational temperatures from them, and predicted that the instrument VIRTIS/Venus Express might detect these emissions. [4] VIRTIS is one of the instruments aboard Venus Express with capabilities to sound the upper atmosphere of Venus [European Space Agency (ESA), 2001; Titov et al., 2006]. It is an imaging spectrometer in the visible and near infrared, inherited partly from the Rosetta mission [Coradini et al., 1998]. VIRTIS is devoted to a large number of investigations at Venus, like the cloud deck morphology and dynamics, the lower atmospheric composition, and the derivation of the temperature structure in the lower mesosphere. First studies of these aspects are presented in companion papers in this issue. Most of these studies are carried out by nadir observations. However, VIRTIS is designed to perform limb observations aboard Venus Express as well [Titov et al., 2006]. This mode of observation represents the first systematic sounding of the upper layers of the Venus atmosphere in the infrared, and is devoted to understand the Venus high-atmosphere emissions [Drossart et al., 2007b]. The detection of the non-LTE CO2 4.3-mm band by VIRTIS at thermospheric altitudes has been confirmed by Drossart et al. [2007a], in consonance with the theoretical expectations. [5] In this paper, we focus on a detailed and systematic analysis of VIRTIS observations of CO2 and CO non-LTE emissions from the upper atmosphere of Venus using such a limb geometry. This is a unique data set, and our goals include, first of all, performing an internal validation of the data of our interest, second, characterizing the behavior of the instrument during limb sounding, and also describing the data set available, its quality and quantity, and its

E00B29

scientific potential. These CO2 non-LTE emissions have also been observed by VIRTIS in nadir, which show signatures of gravity waves propagation into the thermosphere, as studied in a companion work in this issue [Garcı´a et al., 2009]. The study of the O2 infrared system in the Venus mesosphere is also presented by Hueso et al. [2008] and in this issue (G. Piccioni et al., Near IR oxygen nightglow observed by VIRTIS in the Venus upper atmosphere, submitted to Journal of Geophysical Research, 2009). [6] In this paper we also present and discuss comparisons with the non-LTE model results in LVEA, using both their published results and a small number of specific simulations performed here with such model. Although a more detailed comparison between model and data is under preparation, this is the first time that this model is confronted with an extensive data set of such a high spectral and spatial resolution. This work is, therefore, also devoted to test the model’s main theoretical predictions, and to identify improvements and lines of exploration for non-LTE models. [7] In section 2, we first describe VIRTIS observations in the limb, including the presentation of 2-D maps and vertical profiles of radiances. Their validation analysis, including consistency tests, actual noise level determination, as well as internal correlations and comparisons with the non-LTE model, are presented in section 3. Further discussions, future prospects and conclusions, are presented in section 4.

2. VIRTIS Observations [8] VIRTIS on Venus Express is an imaging spectrometer whose precursor is currently en route to comet 67P/ Churyumov-Gerasimenko, as part of the scientific payload of the Rosetta mission. Its detailed description and calibration can be found elsewhere [Coradini et al., 1998; Piccioni et al., 2006; Titov et al., 2006] and see also the companion papers in this issue. We briefly describe here those characteristics and mode of operation on board Venus Express which are needed for our analysis. 2.1. VIRTIS/VEX Characteristics and Data Set [9] The instrument consists of two channels: VIRTIS-M (hereinafter V-M), a mapping spectrometer at a resolving power R  400, working in the visible (0.3 –1 mm) and in the near infrared (1 – 5 mm), and VIRTIS-H (hereinafter V-H), a high-resolution spectrometer (R  1800) working in the spectral range 2 –5 mm. Table 1 of the work by Drossart et al. [2007b], summarizes their main characteristics; let us recall some of them here. The nominal field of view (FOV) is 64  64 mrad2 for V-M (whole frame, with 256 spatial pixels in each direction), and 0.58  1.75 mrad2 for V-H. The V-M spectral range is sampled at 432 spectral wavelengths, while a typical V-H spectrum obtained in the nominal mode contains a sequence of 3456 measurements. The nominal noise equivalent radiance (NESR), for 1 second of integration time, is 5000 mW m 2 sr 1 mm 1 in both channels, according to the ESA [2001]. All V-H and V-M data taken during one orbit, or terrestrial day, are stored in the so-called ‘‘qubes.’’ The V-M ones generally contain a set of data in 3 dimensions, 2 spatial and 1 spectral. During each orbit the number of V-H/V-M data qubes

2 of 19

E00B29

GILLI ET AL.: VIRTIS/VEX NON-LTE LIMB OBSERVATIONS

E00B29

Figure 1. Geographical coverage with VIRTIS-M (V-M) during orbit 25, from apoapsis. (top, left) Map of radiances at 4.32 mm from qubes 0, 1, and 2. (top, right) Altitude-SZA location of all the limb data above the cloud layer (all pixels of seven qubes of this orbit); (bottom) Location of the data in SZA, longitude, latitude, and local time. The numbers identify the different qubes. obtained varies from 1 to 20. A detailed description of data handling and archiving is presented by Drossart et al. [2007b]. [10] VIRTIS observations are separated into two categories, corresponding to distances of Venus Express to the planet lower than 12,000 km (Spectral Mode) or higher (Spectral Imaging Mode). During the Spectral Mode only a partial coverage of the surface is normally obtained, but thanks to the proximity to the planet, a very small field of view (projected) is achieved, ideal to study atmospheric variability and composition, and for tangential limb observations. Several instruments on board Venus Express share this operation mode. The Spectral Imaging Mode permits unique and unprecedented maps of the disk of Venus, and has been used to obtain global observations from apoapsis (‘‘mosaic construction’’). We will show below how these VIRTIS 2-D images contain useful information on the limb of the planet as well. [11] A note regarding the distinction between limb and nadir observations is needed in order to clarify that, in a general sense, a data qube from VIRTIS will contain both kind of observations. Our distinction merely tells whether the line of sight touches the disk of Venus (nadir data) or crosses its atmospheric limb. Therefore, some pixels may represent actual limb sounding while nearby pixels in the same V-M image will supply ‘‘disk’’ observations. This will become clearer in section 2.5. For this reason, analysis of nadir observations, described in companion papers of this issue, share similar difficulties to our study and all of them are beneficial to the overall validation of VIRTIS data.

2.2. Geographical and Temporal Coverage [12] At present, V-M data are available in calibrated form up to the orbit 631 (13 May 2006 to 1 Jan 2008), but in this work we focused our analysis on a selected number of orbits up to orbit 295 (10 February 2007). For V-H, at the time of preparing this paper, the data from orbit 23 to 127 (13 May 2006 to 25 August 2006) were already calibrated, so we focused on them, while only a fraction was calibrated and available for the rest of the mission. As mentioned above, our purpose is to perform an exploration of V-M and V-H limb data in the Venus upper atmosphere, from the cloud top, around 60 km, up to upper thermospheric layers. With this aim, we made a catalog of orbits/qubes/pixels with actual limb sounding and a number of diagrams of the geometrical and spatial coverage of all limb data obtained. Examples are shown and commented next. 2.2.1. VIRTIS-M [13] Starting with V-M, Figures 1 and 2 show two examples of the V-M coverage for two different observational geometries, both sampling the Venus limb, one from orbit 25 (15 May 2006), during the apoapsis, and the other from orbit 43 (2 June 2006) near the periapsis. They can be considered as representative of the two principal VIRTIS modes of observation (see section 2.1). [14] The first orbit, from apoapsis, contains 8 qubes of V-M data (7 with limb data), each one consisting of set of pixels, or a 2-D image at each wavelength, as mentioned above. Let us recall that each pixel is associated to one individual spectrum. In Figure 1a, data at the wavelength of 4.32 mm from three of the qubes, building up a map of radiances; we will discuss them below. Each qube includes a

3 of 19

E00B29

GILLI ET AL.: VIRTIS/VEX NON-LTE LIMB OBSERVATIONS

E00B29

Figure 2. V-M coverage in tangent altitude versus SZA during periapsis of orbits 43 and 287. Data are shown between 60 and 180 km only. large number of pixels (245  256), and the fraction of those pixels with actual pointing to the Venus limb are shown in Figures 1b– 1d. We see that when the 8 qubes are considered, the latitude, longitude, and local time of the data cover an ample range of values. In this work we pay particular attention to the altitude and solar zenith angle (SZA) coverage. These two parameters are expected to produce the maximum variation in the non-LTE emission of both CO2 and CO in the atmosphere, according to the model of LVEA. The most intense signal is expected between 90 and 140 km, and for low SZA values; for nighttime the signal would be very low, and we confirmed all of this with VIRTIS data. Figure 1b shows that the number of limb daytime spectra is large for this orbit. The vertical resolution from each row of pixels (for each qube) is between 15 and 20 km, typically. This is larger than an atmospheric scale height, but the superposition of adjacent rows may improve significantly this resolution, as we will show below. Also the SZA sampling is not completely even, but the large density of points permits suitable averaging if they are needed. [15] Figure 2 illustrates the V-M coverage in the qube 0 of orbit 43 (2 June 2006), during the periapsis of Venus Express and in the qubes 0 and 1 of the orbit 287 (1 February 2007). We use again altitudes at the tangent point versus SZA, which describe our preferred data set for limb sounding. The available daytime limb data during the orbit 43 are about 5410 while this number is larger for the 287 (about 8000). The vertical resolution obtained at the limb from one single row of pixels (300 m), improves by a very large factor with respect to the apoapsis case, and is therefore particularly useful for studying the vertical variation of the non-LTE emissions and testing the theoretical predictions. 2.2.2. VIRTIS-H [16] Observations of the limb of the planet by VIRTIS in the Spectral Mode are not as frequent as in the Spectral

Imaging Mode owing to pointing constraints; the field of view of the V-M frame is much larger than the V-H. The amount of V-H spectra which actually touches the limb of the planet is therefore very limited in most orbits, sometimes only 2 or 3 spectra of our interest are available per orbit. [17] Figure 3 illustrates the V-H geometrical coverage at the limb, again as a function of SZA and tangent altitude, combining all the orbits during the first four months of operation. A total of 651 points are shown, each one representing a V-H spectrum. We have marked 4 specific orbits (43, 76, 80, and 88) to indicate 4 different special observational coverages. The first one corresponds to a special limb case, the so-called ‘‘Tangential Limb Scenario’’ [see Titov et al., 2006], where the satellite is fixed pointing to the limb of Venus during the periapsis. This is a specially useful kind of orbit for limb sounding and for non-LTE studies, given the high rate of tangent point observations. Orbit 76 (6 July 2006) corresponds to a ‘‘stellar occultation’’ case, where again a large number of observations is obtained at the limb, but they correspond to the nighttime hemisphere. They do not show a distinct emission level and have not been considered in the analysis. The other two orbits correspond to the most common limb scenario. [18] For a number of studies, averaging of these data may be required, in order to achieve a good SZA or altitude coverage. From the 651 V-H limb spectra available, a total of 145 are daytime observations (SZA < 90°), and 44 of these correspond to strong solar illumination conditions (SZA < 60°). The last ones are, in principle, optimum for non-LTE studies. Let us recall that, within Venus Express, V-H data are the measurements with the highest spectral resolution. The latitude and longitude coverage of the daytime spectra are shown in Figure 3 as well. Except for orbits 43 (2 June 2006) and 80 (10 July 2006), with highlatitude sampling of nearby points, all others are disperse, with a preference for low latitude.

4 of 19

E00B29

GILLI ET AL.: VIRTIS/VEX NON-LTE LIMB OBSERVATIONS

E00B29

Figure 3. (left) Coverage in tangent altitude and SZA of the VIRTIS (V-H) data available from all the orbits from 23 to 127. Some orbits are indicated with special symbols. (right) Latitude-longitude location of the data, for daytime only. Symbols as in the left. See section 2.2.2 for details. 2.3. VIRTIS-H Spectra and Profiles [19] We describe here typical V-H limb spectra which show strong non-LTE signals, and we build up vertical profiles also, in order to characterize the emissions. We focus on three emission bands, those at 4.3 mm and 2.7 mm both due to CO2, and that at 4.7 mm, due to CO. All of them are suitable to be contrasted with the recent non-LTE model predictions of LVEA. 2.3.1. CO2 4.3 mm Emission [20] Figure 4 (left) shows a sample of seven V-H spectra in the 4.3-mm region, corresponding to different orbits, and pointing at different tangent altitudes. The selection corresponds to orbits 33, 40, and 47 (24 May, 31 May, and 7 June 2006, respectively) and to a SZA interval from 20° to 30°. First, it is noticeable an intensity of the signal well above the noise level at all wavelengths between 4.20 and 4.50 mm. The emission falls to zero shortward and longward of these values, respectively, where the small oscillation observed can be considered as the measurement noise and which seems to be close to 2000 mW m 2 sr 1 mm 1. Only the two lowest altitude spectra show signal in the wings, though surely from solar scattering in the clouds. At the uppermost altitude shown in Figure 4, 167 km, the spectrum is still above noise level, which demonstrate that V-H can sound up to those thermospheric altitudes in Venus before averaging.

[21] It is clear that the spectral shape of the 4.3-mm band, the peak intensity, and the wavelength of the peak of the band, vary with altitude. The maximum emission is obtained in this sample around 110 km tangent altitude. This is close to the altitude predicted by the CO2 non-LTE model by LVEA. Also the spectral shape is similar to the model prediction, with three clear peaks in the 110 km spectrum, around 4.28, 4.32 and 4.35 mm. According to the model, the first peak is due to the second hot band of the main CO2 isotope; the mechanism being solar absorption at 2.7 mm, exciting CO2 states which later relax radiatively, emitting in the 4.3-mm region. The second peak, around 4.32 mm, is produced by radiative relaxation of the CO2 vibrational states excited after solar absorption at 2.7 mm and 2.0 mm. The third peak is mostly due to solar pumping at 2.0 mm. The non-LTE model predicts that at high tangent altitudes the 4.3-mm band would change this shape to a simpler two emission peaks separated by a dent around 4.29 mm, due to the dominance of the 2.7 mm solar pumping. This is indeed observed in Figure 4, above about 130 km. At higher altitudes, the model predicts a further change, due to the direct solar absorption in the fundamental band of the main isotope, whose central dip at 4.26 mm is observed in Figure 4 in the 167 km spectrum. [22] Vertical profiles can be constructed combining spectra at different altitudes. They, however, will not represent

5 of 19

E00B29

GILLI ET AL.: VIRTIS/VEX NON-LTE LIMB OBSERVATIONS

E00B29

Figure 4. (left) Selection of seven V-H spectra from orbits 33, 40, and 47 at different altitudes and for 20° < SZA < 30°. (right) Vertical distribution of the 145 daytime measurements at 4.32 mm from orbits 23 to 127 (May – August 2006). The solid line on the right represents a model simulation. Orbits and symbols as in Figure 3. real 1-D profiles of the Venus atmosphere. Only a couple of orbits (43 and 80) contain a sufficient number of limb spectra to build a vertical profile for one single orbit; in general, data from different dates and geographical locations will be combined. This will have to be taken into account when performing detailed comparisons with non-LTE models, which usually provide the radiance structure at a single geographical location. In Figure 4 (right) we combine diverse data at 4.32 mm. The data and the symbols correspond to those in Figure 3. A tentative grouping by SZA values did not show any clear trend, given the scarcity of data; the SZA variation is better observed in V-M data and discussed below. [23] The cloud of V-H data presents a dispersion not present in the 1-D radiance simulations by LVEA, also shown in Figure 4. The model simulation corresponds to SZA = 80° similar to that of orbit 80. This orbit can be considered as a good 1-D profile extending from 30 to 200 km. Notice that data from orbit 43 can also be used to build vertical profiles, but only above about 100 km. Both data and model present a clear increase with altitude in the lower mesosphere and a peak emission around 120 km. Above this altitude, the emission declines quickly to zero, following the decrease in atmospheric density. The actual atmospheric density profile is critical in determining the precise altitude of the peak emission, and this is the most likely reason for the mismatch with the prediction of LVEA. Regarding the peak radiances, not much comparison can be performed at this stage, as it should be the topic of a more rigorous comparison of the non-LTE model, and should account for suitable atmospheric variability in the density profiles.

2.3.2. CO2 2.7 mm Emission [24] According to LVEA, CO2 emissions at 4.3 mm and at 2.7 mm are both produced by solar pumping. The emission level at 2.7 mm is, however, much lower and we have only found a small number of cases/spectra with a distinct emission above noise levels. Figure 5 shows two of those spectra, which correspond to SZA of about 45° and 60°. One of them presents a bias, which is not corrected here. A non-LTE model simulation has been performed for these solar illumination conditions, using the VIRA reference atmosphere, and is also presented in Figure 5. The spectral shape of the two fundamental bands of the main CO2 isotope can be discerned, by comparing with the non-LTE model simulation. The data are very noisy for further conclusions at this stage. Hopefully, a more extended data set of V-H spectra will be obtained during the rest of the mission, and a statistically sounded comparison with the theoretical prediction will be a good test for non-LTE models. This 2.7 mm non-LTE emission has also been detected in the V-M signal, as we discuss in section 3.2. 2.3.3. CO 4.7 mm Emission [25] Figures 6 and 7 illustrate a similar study for the 4.7-mm spectral region which, according to LVEA is dominated by non-LTE emissions of CO in the upper mesosphere and lower thermosphere. The emission observed is lower than the CO2 emission, and the number of V-H CO spectra with good signal to noise ratio is much smaller. About 20 spectra, out of the 145 available at the limb during daytime, show a clear signal. Many of them show a small signal above the noise level or above the daylight scattered by the aerosols and clouds. This is why it is difficult to detect this non-LTE

6 of 19

E00B29

GILLI ET AL.: VIRTIS/VEX NON-LTE LIMB OBSERVATIONS

E00B29

Figure 5. Two selected spectra at 2.7 mm observed by VIRTIS-H (thin lines) compared to a non-LTE model simulation (thick line). The upper spectrum is from orbit 88, qube 2 at an altitude of about 107 km and SZA about 60°. The lower spectrum is from orbit 96, qube 20 at an altitude of 118 km and SZA about 45°. The theoretical simulation has been performed at 110 km and for similar solar illumination conditions. No shift has been applied to the observed data. See section 2.3.2. emission below about 70 km. The highest altitude where the emission can be clearly identified is about 120 km. [26] Three of the most interesting spectra are shown in Figure 6 (top). The two spectra with the strongest emission correspond to two observations at approximately the same tangent height, around 85 km, but for SZA of 60° and 76°. The third one, shifted downward for clarity, is the spectrum at the highest altitude available in the data set explored so far, about 120 km. In spite of the noise level, very noticeable in the high-altitude spectrum, and which amounts to 5000 mW m 2 sr 1 mm 1, in all of them we can see two branches of what seems a typical vibrational-rotational band. The simulations of LVEA for the CO(1-0), or fundamental band (FB), of the main isotope showed such structure as well. However, their prediction has a minimum emission at the FB center, at 4.67 mm, while the spectra in Figure 6 present their central minimum around 4.73 mm. This points to a different band of CO, the CO(2-1), or first hot (FH), of the main isotope, as responsible for these data. Such emission was not simulated by LVEA, who focused their analysis in the FB. These V-H measurements indicate that, if the fundamental transition is present, its intensity is smaller than the first hot. A similar effect was observed with ground-based measurements at much higher spectra resolution [see Crovisier et al., 2006]. [27] The spectral resolution of V-H allows to identify individual lines of the P and R branches of the FH CO band, and this is shown Figure 6 (bottom). A number of vertical

lines are drawn at the wavelengths of the spectral lines centers of the two bands, according to HITRAN [Rothman et al., 2005]. For clarity, we focus on the spectral interval occupied by the P branch of the FH band. We can identify most of the lines of the P branch, up to P20 at least. It is interesting to notice, however, that in the center of the FH P branch, at 4.727 mm, a weak emission line is observed, which coincides with the line P7 of the CO fundamental band. No FH lines are present there. Moreover, all the 5 lines observed around this wavelength are better fit by the P5 –P9 lines of the FB than by the FH band. In other words, the emission from the FB seems to be also detected in the central portion of the FH band. The absence of FH lines may be due to the smaller strength of the lines with low rotational number. This is a tentative explanation which requires careful quantitative modeling. [28] As previously done by Crovisier et al. [2006], it might be possible to derive temperatures from this emission after a correct model fit. This is not possible with the model of LVEA in this moment, but we have attempted to estimate the rotational temperature from some VIRTIS data. First, we assumed that the rotational levels of CO are in thermal equilibrium at the pressures of relevance. Then, the usual expression for the rotational temperature, given by the ratio of line intensities (see an example for the Venus atmosphere by Deming et al. [1983]), was applied to the spectrum shown in Figure 6 (bottom). A selection of lines from the P branch of the FH band is required. The criteria we

7 of 19

E00B29

GILLI ET AL.: VIRTIS/VEX NON-LTE LIMB OBSERVATIONS

E00B29

Figure 6. (top) Selection of three V-H spectra at 4.7 mm. The two closer ones correspond to about 85 km and to SZA of 60° (solid line) and 76° (dashed line). The lowest spectrum, shifted 20,000 radiance units downward for clarity, corresponds to a tangent altitude of 118 km and SZA of 46°. (bottom) Zoom of the SZA 60° spectrum, with lines marking the positions of the Hitran line centers for the fundamental (dashed marks) and first hot (solid marks) bands of CO. followed were a small overlapping with the FB lines and a peak emission well above the noise levels. This method rules out high and low J lines. Also, adjacent rotational lines should be avoided in order to reduce numerical error. For this spectrum, the best subset of lines that we found are those between P3 and P8. By forming pairs between the more extreme ones, P3, P4 and P7, P8, several values of rotational temperature were found, all around 237 K, but with uncertainties of 50 K or larger. This large error is due to the large sensitivity to the peak intensity in the center of the lines. This intensity value is uncertain because it is not directly available from the data, given the V-H spectral resolution and sampling, but obtained by an extrapolation to the radiance values in the centers of the selected lines. Notice that the P branch of this spectrum apparently has a maximum at P9 and a local minimum at P6; we have excluded these two lines in this analysis since they seem specially affected by the V-H sampling and noise. This error in the lines intensities can be reduced in the future, by using a non-LTE model simulation for the FH band, as was done by Crovisier et al. [2006].

[29] Regarding the vertical variation, profiles can be constructed from the two dozens of spectra with good S/N ratio. They are shown in Figure 7 at 4.678 mm. The data have a clear dispersion, and in contrast to the model prediction for the FB, it does not show any clear peak: the plateau between 95 and 120 km, at about 6000 mW m 2 sr 1 mm 1 is very close to the noise level. As it happened with the CO2 data in Figure 4, the small number of points does not allow to discern a clear SZA variation. 2.4. VIRTIS-M Maps During Periapsis [30] In this section we present measurements of V-M from the periapsis, where the maximum spatial resolution can be achieved, as shown in Figure 2. In the Spectral Mode, near periapsis, the V-M slit is fixed, no scanning is performed, and a row of pixels is taken. Therefore, the mapping is composed of a set of successive ‘‘V-M slit frames,’’ each one taken about 10 s apart, while the spacecraft is moving along the orbit. [31] This is illustrated in Figure 8, where we plotted a map of CO2 radiances at 4.32 mm, in the center of the strong

8 of 19

E00B29

GILLI ET AL.: VIRTIS/VEX NON-LTE LIMB OBSERVATIONS

Figure 7. Vertical distribution or V-H radiances at a central wavelength of the 4.7-mm band of CO. Orbits and symbols as in Figure 3. See section 2.3.3. CO2 bands system. The vertical lines correspond to the actual pointing of each pixel, similarly to Figure 2. We will build similar maps with V-M data but from apoapsis observations, in the next section. [32] Two strong variations are observed in the radiances, one along the vertical axis and the other along the horizontal, and both can be explained by the non-LTE model.

E00B29

Considering first the vertical, a peak emission is observed, in the lower thermosphere, 110– 120 km. As the SZA varies along the track (from 70° to nighttime), the peak intensity, and the whole profile, decrease. This is predicted by the non-LTE model and this data set offers a perfect benchmark for a quantitative validation of the model; a work which is in progress. [33] This is the first time that maps of IR radiation with this geometrical resolution are obtained in the Venus upper atmosphere. In order to illustrate more clearly such a resolution, we constructed a vertical profile with data from a single slit view, that indicated with an arrow in Figure 8. They are shown in Figure 9 (right), at three wavelengths. A log scale is used for the radiances, to confirm that a good signal to noise is achieved even at 160 km from one single spectrum. These data correspond to SZA  75°– 80°, a value typical of daytime observations at high latitudes. The altitude of the peak is maximum in the center of the band, at 4.32 mm, and it is lower in the wings, at 4.25 and 4.40 mm, according to model expectations. Figure 9 (left) shows three V-M spectra at three tangent altitudes. The spectral resolution is lower than V-H, but the change in the overall shape of the band with altitude is clearly observed, as discussed in the previous section. We have also tried to build up maps of radiances in the 2.7-mm and 4.7-mm region. Unfortunately such V-M maps are very noisy at all altitudes, contain a very strong component of scattering up to 90 km, and are not further studied here. 2.5. VIRTIS-M Maps and Profiles During Apoapsis [34] V-M observations from the apocenter of the Venus Express orbit are good examples to show the large potential of an imaging instrument like V-M for mapping large portions of the Venus disk. However, some images also contain information about atmospheric limb emissions, and

Figure 8. Map of V-M radiances from periapsis (orbit 43) at 4.32 mm versus SZA and tangent altitude. The color code ranges from 0.03 to 0.34 mW m 2 sr 1 mm 1. Each line corresponds to a column of pixels along the V-M slit. V-H data for the same orbit are overplotted (triangles). The arrow indicates the column of pixels from which the vertical profiles in Figure 9 have been extracted. 9 of 19

E00B29

GILLI ET AL.: VIRTIS/VEX NON-LTE LIMB OBSERVATIONS

E00B29

Figure 9. V-M spectra and profiles at 4.3 mm from periapsis, orbit 43. (left) Spectra at three different altitudes, as indicated. (right) Profiles at three wavelengths extracted from the pixels indicated in Figure 8. See section 2.4 for details. allowed us to extract vertical profiles with an excellent vertical resolution. 2.5.1. 2-D Images From Apoapsis [35] Figure 10 shows a map of V-M radiances at 4.32 mm, built up with the data from three consecutive qubes or V-M images (2, 1, and 0) of orbit 25, all of them acquired from

apoapsis, at about 66,000 km from the center of Venus [Svedhem et al., 2007]. The instrument takes about 10 min to take every qube, or image, before the satellite pointing switches to the next. The three images coincide with the data qubes shown in Figure 1a. The current map looks very different for two main reasons. First, now we are plotting

Figure 10. Maps of V-M limb radiances at 4.32 mm from apoapsis, from the three qubes of orbit 25 shown in Figure 1, as a function of tangent altitude and longitude. Vertical lines indicate SZA in degrees. Color scale is given in Figure 11. 10 of 19

E00B29

GILLI ET AL.: VIRTIS/VEX NON-LTE LIMB OBSERVATIONS

E00B29

Figure 11. Smiling effect in V-M images from apoapsis and its correction for two cases: (top) qube 1 of orbit 25 and (bottom) qube 0 of orbit 29. (left) Original map of radiances at 4.32 mm. (middle) Location of pixels with radiance between the ranges indicated, and fit of a parabolic line to the cloud of pixels. (right) Corrected maps. See section 2.5.2 for details. just the portion of the image which corresponds to actual limb sounding, i.e., tangent altitudes above about 60 km. And second, instead of pixel number, in Figure 10 we used altitude above the surface for the y axis, and a geographical parameter, longitude, in the x axis. Since the latitude is almost constant for all these limb data, Figure 10 represents a geometrically realistic view of such altitude-longitude cross section of the Venus upper atmosphere. [36] Although the data acquisition and the image construction are different, this map is similar to the map from periapsis shown in Figure 8. Again, we observe a strong emission in the lower thermosphere, with a peak around 110– 120 km. Also a large variation is observed in the horizontal axis. The SZA variation is, however, very small in each of the three qubes/images shown in Figure 10. Therefore, the horizontal variation may be of a different nature than that seen in the periapsis case, and might indicate a real atmospheric variability. [37] However, extreme care is required before such an interpretation. The observations are taken from very far away, in a geometry which the instrument was not optimized for limb sounding; the data might contain some geometrical effects. First of all, these maps contain what appears to be a systematic bending of the radiances in the central part of the image. This bump has also been observed at other wavelengths, qubes, and dates, and it seems to be systematic. It is specially apparent in the uppermost layers.

A second problem is shown by the sharp change in radiance between adjacent qubes. The three qubes shown here were taken with some minutes apart, and there is no obvious physical reason to expect changes of that magnitude between them. Investigation of these instrumental effects is ongoing. A likely explanation for the ‘‘smiling effect’’ is discussed in the next section. A probable cause for the second problem could be some flat field effect not yet identified. 2.5.2. Geometrical Correction of V-M Images [38] The apparent bending observed in Figure 10 is of the order of 20 km, precisely the vertical resolution per pixel from the apoapsis, in other words, this error is within the size of one pixel. Although we cannot rule out small variations of the spacecraft pointing, the effect may be due to a slight bending of the V-M slit. This ‘‘smiling effect,’’ which cannot be avoided, is of special importance for this limb sounding. Fortunately, some correction is possible, precisely by using these limb non-LTE emissions. We basically used two theoretical predictions of these emissions: first their strong decrease with tangent altitude, and second their variation with SZA, as explained below. The idea is to fit the bending observed with a simple function and then use that function to replot the radiances. We illustrate two examples of such a correction in Figure 11. [39] Figures 11a and 11d show two different V-M images, those for orbit 25 (15 May 2006), qube 1 and orbit 29

11 of 19

E00B29

GILLI ET AL.: VIRTIS/VEX NON-LTE LIMB OBSERVATIONS

E00B29

Figure 12. Vertical profiles of V-M radiance at (right) 4.33 mm after combining the (left) rows of pixels. (top) Qube 2 of orbit 25. (bottom) Qube 1. (19 May 2006), qube 2; these are the original maps to transform. The correction has to be applied to them on a pixel basis, as we will assume it to be symmetric around the central pixel of the frame 128. Then we have selected all those pixels with radiances within a small bracket of values; they are marked with crosses in Figures 11b and 11e. According to the non-LTE model, if the atmospheric densities and the SZA illumination do not change very much, a given limb radiance corresponds to a given tangent altitude. Actually, this is the reason why we discovered this smiling effect. We can observe how the crosses marking the 0.15 – 0.20 radiance interval indicate a bending approximately symmetric around pixel 128. We fit a simple parabolic function to this cloud of pixels, and used it to correct for the tangent altitude of all the 256 pixels in each row of the frame. A perfect fit cannot be obtained owing to the different sources of error, which produce some dispersion in the measurements. In addition, the cloud of data does not present a random distribution but a peculiar dispersion produced by the particular geometry of the observations, which is specific for each qube. Some atmospheric variability cannot be ruled out either. For these reasons, we aim here at a first-order correction. Let us notice, also, that it will be difficult to find a single fit function with matches all the radiance intervals of all the orbits and qubes. In our example, the radiance maps resulting after the correction are shown in Figures 11c and 11f. The bending of about 20 km is greatly reduced. This correction function introduces some uncertainty in the absolute pointing altitudes. In the center of the frame, around pixel 128, the correction is very

small, but the pointing may be inaccurate by around 5 km for the pixels at the extremes of the V-M frame. Still, we believe that the corrected images can be used for scientific studies, like atmospheric variability, possible dependence of the emissions on non-LTE parameters, etc, which would be impossible without this geometrical correction. In the VIRTIS team we are considering the implementation of this procedure in future operational processing. 2.5.3. Vertical Profiles From V-M 2-D Images [40] Lets analyze now the vertical profiles of radiance from the limb that can be obtained from these V-M images at the apoapsis. As an example, Figure 12 shows two vertical profiles extracted from two different data qubes, already shown in Figure 1, qubes 2 and 1 of orbit 25. The profiles, at the wavelength of 4.32 mm, are shown on the right, while on the left we plot a small portion of the qube/ image and marked the precise pixels used to build the profiles. The portion of the Venus disk selected corresponds to an almost constant solar illumination, with a small change of SZA, from 10° to 20°. To build up the profiles, we considered two adjacent sets of 10 pixels, taken at the same time, from two adjacent rows of the V-M frame. The selected pixels are pointing at varying altitudes, from the planet’s surface up to the Venus thermosphere, with a vertical resolution around 15 km, and symbols in Figure 12 just point to the center of each pixel. If the orientation of the limb of Venus is optimum, as we can see in Figures 12a and 12b, a very good altitude resolution can be achieved by mixing adjacent pixels; otherwise, the gain in vertical resolution is lower or nonexistent. The improvement of

12 of 19

E00B29

GILLI ET AL.: VIRTIS/VEX NON-LTE LIMB OBSERVATIONS

E00B29

Figure 13. (top) Vertical profiles of V-M radiances at four wavelengths for qube 2, orbit 25. (bottom) Spectra of the 20 pixels composing the vertical profiles shown on the top; they belong to 2 adjacent rows of pixels, marked with stars and circles as in Figure 12. Four triangles mark the wavelengths of the four vertical profiles. the vertical resolution is at the expense of losing horizontal resolution, obviously. [41] Other four vertical profiles are shown in Figure 13 (top), obtained from the same rows of pixels in the top of Figure 12 but at four different wavelengths in the 4.3-mm region. The wavelength values are marked in Figure 13 (bottom), which shows the 20 spectra considered. We can get a lot of useful information about the non-LTE emissions from an image like this, since it combines spectral and vertical variations. For instance, the peak altitude is shown to vary with wavelength, being higher at 4.29 mm or 4.33 mm (see Figure 12) than in the wings of the band. This was predicted by the non-LTE model, and was observed also in periapsis observations. Although the qualitative behavior seems to agree with the model, the vertical profile at 4.45 mm does not coincide with the model, but this suggests a much lower emission compared with the center of the band. We think this may be solved by increasing the number of weak CO2 bands in the non-LTE model, a task for a forthcoming revision of the non-LTE model that we are currently undertaking. [42] A systematic problem in V-M data can be noticed in Figure 13 (bottom). There seems to be an oscillation in the spectral domain, with high and low values of intensity from one spectral point to the next. We checked that this ‘‘oddeven effect’’ is present at all VM near-IR wavelengths, from 2 to 5 mm, and it amounts to about 0.05 radiance units in Figure 13, or 50,000 mW m 2 sr 1 mm 1, which is about 10 times larger than the measurement noise. The technical cause of the odd-even seems to be an asymmetry of the

Read-Out Integrated Circuit (ROIC) of the Focal Plane Array (FPA) of the instrument; see Coradini et al. [1998] for a more detailed description of the instrument. In particular, there is an asymmetry in the clock feedthrough and the unit cells between pixels in odd columns and pixels in even columns, which causes some differences in gain and offset. Most part of this effect is corrected through the radiometric calibration where the responsivity of each pixel is taken into account. However, there seems to be also a problem of the offset/reset being dependent with the signal, which makes the difference of the odd-even effect bigger for extreme conditions, which is the case of very high signals or very low exposure times. During periapsis, for example, the exposure time is about 10 times larger than in apoapsis. This is the reason why the odd-even effect is much smaller (at or below the noise level) in Figure 9. Let us recall that odd-even effects are common in imaging spectrometers, in the spectral or spatial domains [Moutou et al., 2003; Siebenmorgen et al., 2007]. We will return to this point in the next section.

3. Validation Analysis [43] Validation analysis traditionally comprises direct comparisons with independent observations. This is not possible with our VIRTIS limb data, although eventually, some indirect comparisons with related measurements by other Venus Express instrument, like neutral densities from VERA and SPICAV, will be possible. At this stage, however, our analysis focused on other aspects. These include (1) internal consistency tests and comparison with averages,

13 of 19

E00B29

GILLI ET AL.: VIRTIS/VEX NON-LTE LIMB OBSERVATIONS

E00B29

Figure 14. V-H and V-M noise and bias in a selection of spectra around 4.3 mm from periapsis, at a tangent altitude of 118 km. Radiance units are mW m 2 sr 1 mm 1. See section 3.1 for further details. (2) estimation of noise/data quality, (3) cross correlation between the two VIRTIS signals, (4) qualitative comparisons with theoretical non-LTE models, and (5) examination of the repeatability and variability of the measurements. The goal is to characterize the behavior of the VIRTIS instrument, to confirm that the measurements are physically meaningful, and to detect potential bias or systematic errors. The most important of these analysis, and their conclusions, are presented next. 3.1. Noise and Bias in V-H and V-M [44] One of our basic objectives was to evaluate the measurement noise in orbit, in the spectral ranges and emissions of our interest, to compare it with the nominal one obtained by ground calibration, and at the same time, to analyze possible systematic effects. A first step was the examination of individual spectra for the orbits with limb data. A second task was to average homogeneous sets of data, considering appropriate boxes in SZA and tangent altitude, and to calculate their mean radiance and dispersion. [45] For the evaluation of the noise level and the bias in the 4.3-mm region, we looked at the two wings of this CO2 system of bands, shortward and longward of 4.20 and 4.50 mm, and at altitudes where no significant emission is expected there. The mean radiance of each spectrum can be considered as the bias, and its standard deviation (SD) as an estimation of the noise. An example for orbit 43 (periapsis) is shown in Figure 14, where two individual spectra are shown, one from V-H and a nearby one from V-M. The tangent altitude is around 120 km. [46] Regarding V-H, the noise values obtained for this particular example are similar in the left and right wings of the band, and slightly lower than the nominal noise of 5000 mW m 2 sr 1 mm 1. We have evaluated these values

for a diverse set of spectra, for different orbits, solar illumination conditions and tangent altitudes, and have found that they are basically constant up to orbit 47 (7 June 2006) and decrease to about half those values after orbit 79 (8 July 2006). This is probably related to the spacecraft thermal evolution along the mission, with higher noise levels at the beginning of the mission, when the cold box temperature was higher. [47] Regarding the V-M data, the noise in this example is also lower than the nominal value, although there is a significant difference between the left and right wings of the band. Regarding its bias, the availability of a large number of spectra from V-M at a given position, allows us to estimate it for a close subset of pixels. In Figure 14 we added another spectrum, that obtained by averaging 25 V-M pixels/spectra within a small range of tangent altitudes (118 –120 km) and SZA (78° –81°) around the V-M spectrum selected. The mean value coincides well with the bias obtained from one single spectrum. [48] Similar noise and bias analyses for V-M at apoapsis are difficult at present owing to the large odd-even effect mentioned above, much larger than those parameters. In fact, the averaged V-M spectrum in Figure 14 shows more clearly such oscillation in the intensity than one single spectrum. This effect has a magnitude (amplitude between peaks) of about 2000 mW m 2 sr 1 mm 1, of the order of the noise. The V-M calibration group of the VIRTIS team is currently working on this point, in order to reduce its impact on apoapsis data. 3.2. Correlation Between V-H and V-M Spectra [49] There is a small number of observations in the Spectral Mode which correspond to the special Tangential Limb sounding, with simultaneous measurements of V-H

14 of 19

E00B29

GILLI ET AL.: VIRTIS/VEX NON-LTE LIMB OBSERVATIONS

E00B29

Figure 15. Correlation between V-M and V-H spectra taken during periapsis, orbit 43. (right) Location in tangent altitude and SZA, with approximate size of the V-H and V-M field of view. (left) Spectra of the 15 selected V-M pixels, those with SZA of 78.2°, 79.8°, and 81.3°, and of the V-H spectrum (convolved to V-M spectral resolution). Differences between V-H and V-M spectra are also plotted below. See section 3.2 for further details. and V-M. The proximity to the planet during periapsis, allows for a precise geographical collocation of these profiles, and therefore, permits a specially appropriate comparison between them. Orbit 43 offers one example, and Figure 8 shows how close V-M and V-H measurements can be. Figure 15 shows a number of VM spectra close to one particular V-H spectrum, obtained at that orbit, around SZA = 80° and 119 km tangent altitude. Five adjacent slit images were considered, with five pixels on each. The vertical size of the pixel’s FOV, about 300 m for V-M and 2 km for V-H, is shown on the right. On the left we plot 15 of the VM spectra, those from the groups at SZA of 78.2°, 79.9°, and 81.3°, and compare them with the V-H spectrum. This was degraded/convolved to the lower spectral resolution of V-M for the comparison. The shape of the whole CO2 band is very well reproduced by all of them. We think the overall comparison seems fairly good, given the dispersion among the 15 V-M spectra plotted. These present, first of all, a small dispersion within each group, which is similar or slightly larger than typical noise and bias levels. In addition there is a small variation between the three groups, significantly larger than the noise, which is surely due to the small SZA difference between them; the radiance at larger SZA being smaller, as expected. Comparing the V-H spectrum with its closest V-M group of spectra, there is a small shift between them, of about 40,000 mW m 2 sr 1 mm 1 around the peak at 4.33 mm, larger than the noise and bias of both, V-H and V-M. However, this difference is only a factor 2 larger than the SD of this particular group of V-M data, and equivalent to a small fraction of a degree in SZA. This gives us a good level of confidence on both signals.

[50] The emission at 2.7 mm can also be used to test the V-H and V-M agreement. Although the number of spectra with good signal is very limited, we found some spectra in the orbit 43 which are useful for the present discussion. They are shown in Figure 16 and correspond to SZA = 80° approximately. Figure 16 (top) shows vertical profiles at three wavelengths in the 2.7 mm spectral region and show a neat altitude variation, with a clear peak around 115– 120 km. This was predicted by the LVEA model. Figure 16 (bottom) shows the mean and standard deviation of a set of 15 V-M spectra, at altitudes around 120 km. The error bars of individual V-H and V-M are slightly larger than the nominal values, and they have been omitted in Figure 16. The agreement between V-H (convolved to V-M resolution) and V-M is good within error bars; there may be a slight mismatch around 2.8 mm but the data seem noisier there. Moreover, this emission is lower and noisier than that shown in Figure 5, at all wavelengths, since the solar excitation is lower now. We performed a simulation with the non-LTE model, also shown in Figure 16, for the same solar illumination conditions. The agreement is very encouraging, although the model seems to overestimate the radiance; this is due to the specific atmosphere structure assumed in the simulation, and therefore is of no relevance in this analysis. [51] Related to this correlation, we investigated the correlation between different V-H spectra, wherever they can be collocated. The amount of V-H spectra is not large, but we identified a couple of interesting correspondences between spectra with good S/N ratio. On Figure 17 (left), they correspond to four consecutive V-H measurements (acquisition 84 to 87) obtained at the closest approach to Venus

15 of 19

E00B29

GILLI ET AL.: VIRTIS/VEX NON-LTE LIMB OBSERVATIONS

E00B29

Figure 16. (top) Vertical profiles of V-H and V-M radiances at three wavelengths in the 2.7 mm for qube 0, orbit 43 and SZA around 80°. Triangles and crosses mark V-H and V-M data, respectively. (bottom) One V-H spectrum, convolved to V-M resolution (solid line) and comparison with the mean of 15 V-M spectra (dashed lines), from the data composing the vertical profile in the top, at a tangent altitude of 120 km. The dashed-dotted line represents a model simulation. Radiance units are mW m 2 sr 1 mm 1 in both the top and bottom. Three triangles in the bottom indicate the wavelengths of the three vertical profiles. during orbit 43, and point to the lowest tangent altitudes for this orbit, around 99 km (as indicated in the picture) and around SZA = 83° (see Figure 8). The geometrical distance along the track of the satellite between the spectra 84 and 87 is about 250 km, and they correspond to latitudes of 73.0° and 71.8°, longitudes of 302° and 307°, respectively, and a difference in SZA of 1.4°. The four spectra are very similar, with differences of the order of the noise at almost all wavelengths, except in the central region. Around 4.3 mm there is a clear trend, from 84 to 87, the last one being about 30,000 mW m 2 sr 1 mm 1 lower than the former. We think, again, that this is due to SZA changes; close to the terminator small changes in SZA produce noticeable variations in the solar pumping of the CO2 vibrational states. This reflects small changes in the atmospheric structure between those four locations. [52] Figure 17 (right) presents another two V-H spectra, very closely located to each other, both at a latitude of 77.27°S and longitudes of 257.8°E and 260.8°E, respectively, but obtained 2 days apart, on orbits 40 and 42. Their tangent altitudes and SZA are also very close, at 108.4 and 107.8 km, and 82.7° and 82.3°, respectively. Their noise and bias levels are below 5000 mW m 2 sr 1 mm 1 and 8000 mW m 2 sr 1 mm 1, respectively. It is surprising the excellent agreement between them in the overall shape and in the wings of the band; there is only a small difference, in

the central portions of the CO2 emission, with the spectrum of orbit 42 slightly larger than the other and the noise. Again, the small SZA differences is a possible candidate for the observed difference. This points, once more, to a relatively calm and stable atmospheric structure, fully repeatable in times scales of 2 days, at least in this case. It would be interesting to explore further coincidences of this type in future measurements of V-H, since the number of them available to date is small (see Figure 3). 3.3. Comparisons With a Non-LTE Model [53] As discussed above, we have used in this work the non-LTE model of the Venus atmosphere developed at the Instituto de Astrof ´ısica de Andalucı´a/CSIC, in Granada [Rolda´n et al., 2000; Lo´pez-Valverde et al., 2007], to perform a small number of simulations. These simulations and those reported by LVEA have guided most of the analysis presented in previous sections. From these comparisons we found a good overall agreement with the predictions of the spectral shape of V-H and V-M data (Figures 4, 9, and 17), and with the altitude profiles of the radiances in the central part of the 4.3-mm band (Figures 4 and 9). Also the SZA variation of the peak radiance (Figures 8 and 15) qualitatively agrees with expectations. The vertical profiles of the CO2 emission shown in Figures 9 and 13 do agree qualitatively with those of LVEA but a quantitative fit

16 of 19

E00B29

GILLI ET AL.: VIRTIS/VEX NON-LTE LIMB OBSERVATIONS

E00B29

Figure 17. Geographical coincidences between different V-H spectra. (left) Four spectra from orbit 43, around 99 km. (right) Example of two spectra taken at very close points but 2 days apart. Variations attributable to SZA changes. See section 3.2 for details. would be desirable and future work should focus on this aspect and exploit such fit, once the weak bands required are implemented into the model. Other non-LTE emissions predicted by LVEA have also been examined, detected and compared with the model. Figures 5 and 16 show simulations of the CO2 2.7 mm bands for a given reference atmosphere; they indicate a good agreement in the spectral shape of the bands, and for both V-H and V-M spectral resolutions. Although these data are noisy, the main features are captured by the model, and the vertical profiles are also in agreement, with similar altitudes of the peak emission in the data and in the model. The absolute emission, as it happens with the 4.3 mm bands, is not considered in this comparison, since it depends on the actual atmospheric structure assumed in the non-LTE model. The third nonLTE emission detected and studied in this work, that of CO, contains a first hot component not simulated yet by the non-LTE model. [54] Let us highlight here some other results which, either were not predicted or seem to pose interesting challenges to the theoretical modeling. One of them is the intensity in the longer wavelength wing of the 4.3 mm band, where contributions from weak, isotopic and hot transitions are expected. The predictions by LVEA produce a relatively minor contribution of that band’s wings compared to the central band, but the VIRTIS spectra shown in Figures 4 and 9 indicate a more significant contribution. The model surely needs a larger number of CO2 bands in this spectral region. Such a model extension is underway. [55] Certainly, another extension of the non-LTE model is needed to include the first hot band of carbon monoxide, CO(2-1), in order to study the spectra shown in Figure 6.

This is also a minor band compared to the strongest fundamental transition, but it seems to be highly excited by direct solar radiation, like the weaker transitions of CO2 discussed above. As discussed in section 2.3.3, a proper simulation of the CO emission lines will be very valuable for the derivation of rotational temperatures. These could be obtained from this band’s structure, but only after fitting a theoretical model which permits small radiance errors in the line peaks. This is the only molecular band detected by VIRTIS in this infrared portion of the spectrum with well isolated lines, and is therefore a very interesting data set.

4. Conclusions and Future Work [56] VIRTIS on Venus Express is an innovative instrument which, combining imaging and spectroscopy in the near infrared, is obtaining very exciting new data on the upper atmosphere of Venus. We have analyzed the limb measurements taken by VIRTIS in the regions from 4 mm to 5 mm and around 2.7mm from both, the periapsis and the apoapsis of the Venus Express orbit, and in the two signals, V-H at high spectral resolution, and V-M at a lower resolution. We have focused in this work on a subset of VIRTIS data, which includes limb measurements with V-H taken at the periapsis from orbits 23 (13 May 2006) to 127 (25 August 2006), and a small sample of V-M data qubes between orbits 23 (13 May 2006) and 295 (10 February 2007). Extension of the analysis presented here to the rest of the data is ongoing. [57] We clearly identified the strong emission by CO2 at 4.3 mm in the upper mesosphere and lower thermosphere, and of the first hot band of CO at 4.7 mm in the upper

17 of 19

E00B29

GILLI ET AL.: VIRTIS/VEX NON-LTE LIMB OBSERVATIONS

mesosphere. Individual spectra show good signal to noise up to above 160 km in the case of CO2, and up to about 120 km in the case of CO. Single spectra show noise values slightly below the value expected from ground calibration, and a small bias too. We also showed that the V-H spectral resolution is good enough for a number of tests of our nonLTE models. The spectral resolution of V-M is lower, however the vertical resolution is much better from V-M data. During periapsis this is typically 300 m for a single row of pixels, while from apoapsis, the grouping of V-M adjacent rows of data permits a resolution at least two to three times better than the FOV of individual pixels, depending on the geometry of the V-M images and on the relative orientation of the limb of Venus. These observations characterize the Venus atmosphere directly, since they will permit to calculate densities and temperatures from measurements, and also indirectly, because they can give information about the collisional processes, energy transfer and the upper atmosphere energy balance, after a more detailed comparison with a non-LTE model. [58] The variability observed by VIRTIS and mentioned previously by Drossart et al. [2007a] have been confirmed and analyzed here for the first time with this new and extended data set. The maps studied in this paper permit to describe altitude variations at several wavelengths, variations with SZA, and the detailed spectral shape of the CO and CO2 bands. Actually, the good agreement between model and data, on the basic non-LTE features, indicates that our current understanding of the physical processes in the upper atmosphere is essentially correct. All these results will be perfect tests for future non-LTE models, and suggestions for more detailed future comparisons are given in section 3.3. We foresee that such detailed comparisons, specially of the shape of the V-H emissions, can give us information about non-LTE collisional parameters in the upper mesosphere and lower thermosphere, where these processes are specially relevant. Also, comparisons between spectra at similar conditions but obtained with a span of 2 days, and between data more than 200 km apart, show very good matches, which indicates a good behavior of VIRTIS. When applied to a future and extended data set, similar comparisons may supply interesting information about atmospheric variability in the lower thermosphere. The high density of points in the limb in some maps during periapsis offer excellent chances for derivation of density maps in the thermosphere. Such density profiles are foreseen to be obtained by a rigorous ‘‘retrieval’’ process which uses inversion techniques under non-LTE conditions, an exercise similar to other upper atmosphere experiment on Earth, like the MIPAS/Envisat data analysis [Funke et al., 2005]. Densities of CO would be derived in a limited altitude range (90 – 120 km) owing to the low signal of the CO 4.7 mm emission (see section 2.3.3). However, densities of CO2 can be derived in principle in a wider altitude range. Nevertheless, the accuracy of such derivation will depend on a sensitivity study to be performed with the revised non-LTE model. Limb views from V-M during apoapsis should also be very useful for density retrievals, although they are affected at present from a ‘‘smiling effect’’ which deforms the actual emission field by about 20 km, approximately the size of the projected FOV of one pixel. We showed how the non-LTE model has been essential to detect this effect which could be reduced in

E00B29

future processing thanks to a correction technique using these CO2 non-LTE emission. These V-M data from apoapsis are also affected by a large odd-even effect, about 10 times larger than the noise level. During periapsis this effect is below noise due to the much larger exposure time there. This aspect requires further work within the VIRTIS team in the future. [59] The study of the first hot CO(2-1) emission is also very promising, since the rotational structure is resolved and it can therefore be used to derive rotational temperatures in the upper mesosphere. This will be confined to daytime and to the altitude range 90– 120 km, where such signal is strongest. This derivation does not require a full non-LTE retrieval but a fit of the shape of the CO emission band with the model, as explained in section 2.3.3. At present, uncertainties in individual lines of a CO spectrum produce very large temperature errors. Comparisons with temperatures obtained from ground-based measurements of these bands [Crovisier et al., 2006] will also be interesting, as a validation exercise of these data and in order to future check the non-LTE models. [60] VIRTIS/Venus Express will continue acquiring new data at least until 2009. The exploration of the new data and their analysis will continue, hopefully with a version of the non-LTE model extended to weaker bands, and with better understanding of some of the current uncertainties in the data that we highlighted here. The application of retrieval techniques and systematic search for atmospheric variability, makes the VIRTIS observations a very promising data set for studying the non-LTE processes and the structure of the upper atmosphere of Venus. [61] Acknowledgments. We thank the space agencies ASI and CNES for their support. The work of the IAA-CSIC team has been carried out under project ESP2004-01556 and was partially funded by the project AYA2008-03498/ESP of the Spanish Ministery of Science and Innovation and by EC-FEDER funds. GG has been partially supported by a fellowship from Spanish National Research Council (CSIC-JAE Program).

References Bougher, S. W., D. M. Hunten, and R. G. Roble (1994), CO2 cooling in terrestrial planet thermospheres, J. Geophys. Res., 99, 14,609 – 14,622. Carlson, R. W., et al. (1991), Galileo infrared imaging spectroscopy measurements at Venus, Science, 253, 1541 – 1548. Carlson, R. W., P. R. Weissman, W. D. Smythe, and J. C. Mahoney (1992), Near-Infrared Mapping Spectrometer experiment on Galileo, Space Sci. Rev., 60, 457 – 502. Coradini, A., F. Capaccioni, P. Drossart, A. Semery, G. Arnold, and U. Schade (1998), VIRTIS: An imaging spectrometer for the Rosetta mission, Planet. Space Sci., 46, 1291 – 1304. Crovisier, J., E. Lellouch, C. de Bergh, J.-P. Maillard, B.L. Lutz, and B. Be`zard (2006), Carbon monoxide emissions at 4.7 mm from Venus’ atmosphere, Planet. Space Sci., 54, 1398 – 1414, doi:10.1016/ j.pss.2006.04.027. de Bergh, C., J. Crovisier, B. L. Lutz, and J.-P. Maillard (1988), Detection of CO infrared emission lines in spectra of Venus, Bull. Am. Astron. Soc., 20, 831. Deming, D., F. Espenak, D. Jennings, T. Kostiuk, M. Mumma, and D. Zipoy (1983), Observations of the 10-um natural laser emission from the mesosphere of Mars and Venus, Icarus, 55, 347 – 355, doi:10.1016/00191035(83)90107-0. Dickinson, R. E. (1972), Infrared radiative heating and cooling in the Venusian mesosphere: I. Global mean radiative equilibrium, J. Atmos. Sci., 29, 1531 – 1556. Drossart, P., et al. (2007a), A dynamic upper atmosphere of Venus as revealed by VIRTIS on Venus Express, Nature, 450, 641 – 645, doi:10.1038/nature06140. Drossart, P., et al. (2007b), Scientific goals for the observation of Venus by Virtis on ESA/Venus Express mission, Planet. Space Sci., 55, 1653 – 1672, doi:10.1016/j.pss.2007.01.003.

18 of 19

E00B29

GILLI ET AL.: VIRTIS/VEX NON-LTE LIMB OBSERVATIONS

European Space Agency (ESA) (2001), Venus Express: An orbiter for the study of the atmosphere, the plasma environment, and the surface of Venus, Eur. Space Agency Sci. Tech. Rep., ESA STR 6, 46 pp. Formisano, V., A. Maturilli, M. Giuranna, E. D’Aversa, and M. A. LopezValverde (2006), Observations of non-LTE emission at 4 – 5 microns with the Planetary Fourier Spectrometer aboard the Mars Express mission, Icarus, 182, 51 – 67, doi:10.1016/j.icarus.2005.12.022. Funke, B., et al. (2005), Retrieval of stratospheric NOx from 5.3 and 6.2 mm nonlocal thermodynamic equilibrium emissions measured by Michelson Interferometer for Passive Atmospheric Sounding (MIPAS) on Envisat, J. Geophys. Res., 110, D09302, doi:10.1029/2004JD005225. Garcı´a, R., M. A. Lo´pez-Valverde, P. Drossart, and G. Piccioni (2009), Gravity waves in Venus atmosphere revealed by CO2 non-LTE emission, J. Geophys. Res., doi:10.1029/2008JE003073, in press. Gordiets, B. F., and V. I. Panchenko (1983), Non-equilibrium infrared emission and the natural laser effect in the Venus and Mars atmospheres, Kosm. Issled., 21, 929 – 939. Hueso, R., A. Sa´nchez-Lavega, G. Piccioni, P. Drossart, J. C. Ge`rard, I. Khatuntsev, L. Zasova, and A. Migliorini (2008), Morphology and dynamics of Venus oxygen airglow from Venus Express/Visible and Infrared Thermal Imaging Spectrometer observations, J. Geophys. Res., 113, E00B02, doi:10.1029/2008JE003081. Lo´pez-Puertas, M., and F. W. Taylor (2001), Non-LTE Radiative Transfer in the Atmosphere, Ser. Atmos., Oceanic, Planet. Phys., vol. 3, World Sci., Singapore. Lo´pez-Valverde, M. A., and M. Lo´pez-Puertas (1994), A non-local thermodynamic equilibrium radiative transfer model for infrared emission in the atmosphere of Mars: 2. Daytime populations of vibrational levels, J. Geophys. Res., 99, 13,117 – 13,132. Lo´pez-Valverde, M. A., M. Lo´pez-Puertas, J. J. Lo´pez-Moreno, V. Formisano, D.Grassi, A. Maturilli, E. Lellouch, and P. Drossart (2005), Analysis of CO2 non-LTE emissions at 4.3 mm in the Martian atmosphere as observed by PFS/Mars Express and SWS/ISO, Planet. Space Sci., 53, 1079 – 1087, doi:10.1016/j.pss.2005.03.007. Lo´pez-Valverde, M. A., P. Drossart, R. Carlson, R. Mehlman, and M. RoosSerote (2007), Non-LTE infrared observations at Venus: From NIMS/ Galileo to VIRTIS/Venus Express, Planet. Space Sci., 55, 1757 – 1771, doi:10.1016/j.pss.2007.01.008.

E00B29

Moutou, C., A. Coustenis, J. Schneider, D. Queloz, and M. Mayor (2003), Searching for helium in the exosphere of HD 209458b, Astron. Astrophys., 405, 341 – 348. Piccioni, G., et al. (2006), VIRTIS: The Visible and Infrared Imaging Spectrometer, Eur. Space Agency Spec. Publ., ESA-SP 1295, 1 – 27. Rolda´n, C., M. A. Lo´pez-Valverde, M. Lo´pez-Puertas, and D. P. Edwards (2000), Non-LTE infrared emissions of CO2 in the atmosphere of Venus, Icarus, 147, 11 – 25, doi:10.1006/icar.2000.6432. Roos-Serote, M., P. Drossart, Th. Encrenaz, E. Lellouch, R. W. Carlson, K. H. Baines, F. W. Taylor, and S. B. Calcutt (1995), The thermal structure and dynamics of the atmosphere of Venus between 70 and 90 km from the Galileo-NIMS spectra, Icarus, 114, 300 – 309, doi:10.1006/icar.1995.1063. Rothman, L., et al. (2005), The Hitran 2004 molecular spectroscopic database, J. Quant. Spectrosc. Radiat. Transfer, 96, 139 – 204. Siebenmorgen, R., et al. (2007), Very large telescope paranal science operations CRIRES user manual, Rep. VLT-MAN-ESO-14500-3486, 56 pp., Eur. South. Obs., Garching, Germany. Stepanova, G. I., and G. M. Shved (1985), Radiation transfer in the 4.3-mm CO2 band and the 4.7-mm CO band in the atmospheres of Venus and Mars with violation of LTE: Populations of vibrational states, Sov. Astron., Engl. Transl., 29, 248 – 422. Svedhem, H., et al. (2007), Venus Express: The first European mission to Venus, Planet. Space Sci., 55, 1636 – 1652, doi:10.1016/j.pss.2007.01.013. Taylor, F. W., et al. (1980), Structure and meteorology of the middle atmosphere of Venus infrared remote sensing from the Pioneer orbiter, J. Geophys. Res., 85, 7963 – 8006. Titov, D. V., et al. (2006), Venus Express science planning, Planet. Space Sci., 54, 1279 – 1297, doi:10.1016/j.pss.2006.04.017. A. Cardesı´n Moinelo and G. Piccioni, IASF, INAF, via del Fosso del Cavaliere 100, I-00133 Rome, Italy. P. Drossart and S. Erard, Observatoire de Paris, F-92195 Meudon, France. G. Gilli and M. A. Lo´pez-Valverde, Instituto de Astrofı´sica de Andalucı´a, CSIC, Camino Bajo de Huetor 50, E-18008 Granada, Spain. ([email protected])

19 of 19

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114, E00B38, doi:10.1029/2008JE003133, 2009

for

Full Article

Near-IR oxygen nightglow observed by VIRTIS in the Venus upper atmosphere G. Piccioni,1 L. Zasova,2 A. Migliorini,1 P. Drossart,3 A. Shakun,2 A. Garcı´a Mun˜oz,4 F. P. Mills,4,5 and A. Cardesin-Moinelo1 Received 29 February 2008; revised 9 January 2009; accepted 17 February 2009; published 30 April 2009.

[1] We present observations of both the (0–0) and (0–1) bands at 1.27 and 1.58 mm of the O2(a1Dg
X3S
g ) nightglow made with the Visible and Infrared Thermal Imaging Spectrometer (VIRTIS) instrument aboard Venus Express. The observations were conducted in both nadir and limb viewing modes, the latter constituting the first systematic investigation into the vertical distribution of the volume emission rate of the infrared oxygen nightglow in Venus’ upper atmosphere. Limb measurements from 42 orbits covering the latitude range 7°S to 77°N are analyzed. The peak altitude of the volume emission rate occurs typically between 95 and 100 km, with a mean of 97.4 ± 2.5 km. The vertical profile is broader near the equator, with a full width at half maximum of 11 km, a factor 2 larger than at middle latitudes. A double peak is frequently observed, with the lower and upper peaks occurring near 96–98 km and 103–105 km, respectively. On average, the nightglow appears brightest in the vicinity of the antisolar point. This conclusion is consistent with past ground-based observations and nadir measurements by VIRTIS. We mapped the global mean O2 nightglow intensity from VIRTIS data collected during 880 orbits. Patchy features of the nightglow intensity observed in nadir view are correlated with the thermal brightness at 4.23–4.28 mm. The observed positive correlation is consistent with downwelling (upwelling) of oxygen atoms accompanying compressional heating (expansion cooling) or with modulation by gravity waves. Finally, from simultaneous measurements of the 1.27 and 1.58 mm bands, we have estimated the ratio of the transition probabilities A00/A01 to be 63 ± 8. Citation: Piccioni, G., L. Zasova, A. Migliorini, P. Drossart, A. Shakun, A. Garcı´a Mun˜oz, F. P. Mills, and A. Cardesin-Moinelo (2009), Near-IR oxygen nightglow observed by VIRTIS in the Venus upper atmosphere, J. Geophys. Res., 114, E00B38, doi:10.1029/ 2008JE003133.

1. Introduction [2] Intense nightglow emission in the (0 – 0) band of the O2 (a1Dg
X3S
g ) electronic transition near 1.27 mm was first detected on Venus in 1975 [Connes et al., 1979]. Subsequent ground-based observations of the airglow emission on the nightside (nightglow) have shown that the intensity and spatial distribution are highly variable on time scales of hours to days [Crisp et al., 1996; Ohtsuki et al., 2005, 2008; Bailey et al., 2008]. Nadir-looking observations by the Visible and Infrared Thermal Imaging Spectrometer (VIRTIS) on Venus Express (VEX) have provided more

1

INAF, Istituto di Astrofisica Spaziale e Fisica Cosmica, Rome, Italy. Space Research Institute of Russian Academy of Sciences, Moscow, Russia. 3 LESIA, Observatoire de Paris, UPMC, Universite´ Paris-Diderot, CNRS, Meudon, France. 4 Research School of Physics and Engineering, Australian National University, Canberra, ACT, Australia. 5 Fenner School of Environment and Society, Australian National University, Canberra, ACT, Australia. 2

Copyright 2009 by the American Geophysical Union. 0148-0227/09/2008JE003133$09.00

detailed coverage of the morphology of the nightglow, and visual tracking of bright features has enabled the estimation of apparent motions [Hueso et al., 2008]. Limb observations by VIRTIS of the O2 nightglow [Drossart et al., 2007a; Ge´rard et al., 2008] provide a complementary view of the chemistry and dynamics of Venus’ upper atmosphere by determining the vertical distribution of the O2 nightglow. However, no systematic study of a large number of limb observations of the O2 nightglow has been published to date. In this work, we examine limb observations spanning 7°S to 77°N latitude and 2000 to 0400 local times. These observations were collected during the period from July 2006 until August 2008. This much larger set of data has allowed us to determine how the vertical distribution of O2 nightglow varies with latitude. We have also been able to make an observational estimate of the ratio of the transition probabilities for the (0 – 0) and (0– 1) bands near 1.27 and 1.58 mm, respectively. Our results provide new observational insight into the chemical and dynamical processes occurring in Venus’ upper atmosphere and new constraints on chemical transport models. [3] On Venus’ nightside, O2(a) is believed to be produced primarily via reaction (1) [e.g., Connes et al., 1979; Mills

E00B38

1 of 18

E00B38

PICCIONI ET AL.: VENUS O2 NIGHTGLOW FROM VIRTIS

and Allen, 2007], where O*2 denotes an electronically excited state of O2: 2O þ M ! O*2 þ M

ð1aÞ

O*2 þ M ! O2 ðaÞ þ M

ð1bÞ

There are two recent estimates of the net yield of O2(a) from reaction (1). Crisp et al. [1996] inferred net yields of 0.63 ± 0.19 and 0.6– 0.75 for M = N2 and CO2, respectively, on the basis of laboratory studies, while Huestis [2002] recommends a net yield of 0.94 – 0.99 for M = N2 and CO2 on the basis of a combination of laboratory and atmospheric studies. The atomic oxygen consumed in reaction (1) is transported to the nightside via the upper atmospheric circulation from the dayside [Bougher and Borucki, 1994] where atomic oxygen is produced by photodissociation of CO2 at wavelengths <200 nm. Nightglow in the O2(a-X) bands results from O2 radiatively decaying from the electronically excited O2(a) state to the ground electronic state. The vertical profile of the a-X nightglow is determined by the balance between production of O2(a) via reaction (1) and its loss due to radiative decay of the O2(a) state and collisional quenching. There has been controversy over the lifetime of O2(a) against radiative decay, but recent laboratory measurements [Cheah et al., 2000; Lafferty et al., 1998; Miller et al., 2001; Newman et al., 1999; Spalek et al., 1999] appear to have converged to a consistent value. The mean lifetime we have computed from these studies, 74 ± 3 min is consistent with recent assessments [Gamache and Goldman, 2001; Slanger and Copeland, 2003]. Only an upper limit on the rate constant for collisional quenching of O2(a) by CO2 has been determined in laboratory studies [Sander et al., 2006], so coordinated limb observations of the O2(a-X) nightglow, total density, and temperature may provide the best estimate to date of this rate constant. A recent calculation estimated 80% of the O2(a) produced at 96-km altitude decay via emission in the a-X bands [Ge´rard et al., 2008]. More detailed studies of VIRTIS’ limb observations of O2(a-X) nightglow should provide further insight into both the physical and chemical conditions in the nightglow region and the processes involved. [4] The upper atmospheric circulation on Venus is a composite of two patterns: a subsolar (SS) to antisolar (AS) flow and a retrograde zonal flow. Long-term averages of NO and O2 nightglow emissions suggest that the longterm average circulation at 115– 150 km altitude is dominated by strong subsolar to antisolar flow with a weaker retrograde zonal flow [Bougher and Borucki, 1994]. Tracking of cloud features at 65– 70 km shows the cloud top circulation is dominated by the retrograde zonal flow [Gierasch et al., 1997; Markiewicz et al., 2007; Limaye, 2007; Sa´nchez-Lavega et al., 2008]. At intermediate altitudes, the relative strength of these two patterns varies with altitude and with time [Lellouch et al., 1997; Bougher et al., 2006]. Part of the temporal variability is likely due to breaking gravity waves transporting energy and momentum upward from the cloud tops [Alexander, 1992], but significant further work is required to understand and quantify

E00B38

their effects [Bougher et al., 2006]. VIRTIS’ limb observations of O2(a-X) nightglow may provide insight into the vertical and horizontal transport processes and rates, complementing the horizontal transport results obtained using nadir-looking images from VIRTIS [Hueso et al., 2008]. [5] The first reported detection of nightglow emission in the a-X (0 – 1) band of O2 near 1.58 mm on Venus was based on VIRTIS data [Piccioni et al., 2008]. This band has not been observed on Mars but has been observed on Earth [Vallance Jones and Harrison, 1958; Winick et al., 1985]. No prior observation on any planet, however, has detected simultaneously the a-X (0 – 0) and (0– 1) bands. Significant uncertainties exist in both laboratory and theoretical determinations of the relative transition probabilities for these bands, so VIRTIS’ observations of the two bands should advance our understanding of the molecular physics of the a and X states of O2.

2. Instrument Description and Observation Geometry [6] VIRTIS is an imaging spectrometer which covers the spectral range 0.27 to 5.1 mm [Piccioni et al., 2009; Drossart et al., 2007b]. It includes two spectrometers: VIRTIS-M, a mapping spectrometer with medium spectral resolution, and VIRTIS-H, an echelle spectrometer with higher spectral resolution than VIRTIS –M but no imaging capability. VIRTIS-M includes two channels: a visible channel spanning 0.27 to 1 mm and an infrared channel spanning 1 to 5.1 mm. The results reported here were obtained using the infrared channel of VIRTIS-M, with spectral sampling interval of about 10 nm, a spectral full width at half maximum (FWHM) of about 15 nm, and an instantaneous field of view (IFOV) of 250 mrad. As is typical for an imaging spectrometer, a single instantaneous acquisition by VIRTIS-M collects a frame, which is composed of 432 bands (wavelengths) for each of 256 samples (spatial pixels along the instrument slit). If the slit is oriented perpendicular to the limb of Venus, then a single limb observation will obtain a full spectrum at each of 256 tangent altitudes, so the limb profiles of the a-X (0 – 0) and (0– 1) nightglow emissions can be derived from a single frame. A scanning mirror (or the spacecraft’s motion) is used to image multiple sites with different latitudes, longitudes, and/or local times and, thus, obtain a vertical cross section of the line of sight integrated nightglow emission. A similar process is used in nadir imaging mode to obtain 256 pixel  256 pixel  432 band spectral imaging cubes. A full spatial scan across the 64-mrad field of view (FOV) of the mirror requires around 50 min with an 8-s exposure time for each frame. The mirror moves nominally by 250 mrad during the acquisition of an 8-s frame, so that the typical pixel is square, but both the mirror’s scan rate and the integration time can be varied. The orientation of the VIRTIS-M slit is fixed relative to the spacecraft, so it is not always possible to orient the VIRTIS-M slit perpendicular to Venus’ limb when conducting limb observations. The least favorable observation geometry is when the VIRTIS-M slit is tangential to Venus’ limb because then every spatial pixel samples approximately the same tangent altitude. However, even a 10° angle between the VIRTIS-M slit and Venus’ limb is sufficient to obtain a nearly complete limb profile.

2 of 18

E00B38

PICCIONI ET AL.: VENUS O2 NIGHTGLOW FROM VIRTIS

The absolute pointing accuracy for each pixel is much better than the IFOV. This is achieved thanks to the three-axis stabilized capability of the Venus Express spacecraft, supported by accurate star trackers which provide an absolute pointing accuracy better than ±35 mrad (about ±1/7 of a VIRTIS-M pixel) in the timeframe of the observation. Other observation geometries are possible [Titov et al., 2006]. [7] Limb and nadir observations provide complementary information on nightglow emissions. During limb observations, the line of sight integrated volume emission is about 50 times larger than in nadir mode, so faint emissions from minor species are more easily detected. The vertical spatial resolution (2.5 km at 10,000-km slant distance), however, limits the maximum distance from which high-quality limb observations can be made. Venus Express’ polar orbit has its pericenter at about 75°N, so almost all limb observations are made in the northern hemisphere. Nadir observations provide an extended view of the geographic distribution of nightglow emission but no information on its vertical distribution. Spatial resolution decreases when the spacecraft is further from Venus (15 km at apocenter), but areal coverage increases, so nadir observations are typically made over the southern hemisphere and areal coverage is extremely poor north of 20°N. Examples of observations in limb and nadir modes are shown in Figure 1 for orbits 76_18, 271_09, and 44_01 (prefix before the underscore is orbit number, suffix is session number, which is the period of time of continuous acquisition of data resulting in a spectral imaging cube). Figures 1a and 1b show an example of an observation made with the VIRTIS-M slit nearly perpendicular (in the central part of the scan, executed from top to bottom) and tangent to Venus’ limb, respectively. Figure 1c (right) shows an example of nadir view.

3. Data Selection [8] For this study, we have used both limb and nadir observations. The limb observations were acquired during 72 sessions from 42 orbits between July 2006 and August 2008, Table 1. Each orbit lasts 24 h and orbit number 5, the first stabilized nominal orbit, corresponds to 25 April 2006. Usually a few sessions are acquired within a single orbit. Each spectral imaging cube typically extends over 30° latitude, and the 72 sessions selected span 7°S to 77°N. These 72 sessions include all VIRTIS-M limb observations with exposure time of 8 s. Long exposure times (8 s) provide an excellent signal-to-noise ratio (SNR), typically better than 100, at wavelengths shorter than 4 mm, which enabled derivation of well defined vertical profiles, including the high-altitude tail, for fainter nightglow emissions, such as the a-X (0– 1) band at 1.58 mm. The data from the 72 sessions selected have a vertical spatial resolution of about 2 km. The limb profiles of the O2 emission are obtained for each 5° latitude interval as the moving average within 1 km of altitude (in vertical direction). Vertical profiles are retrieved from the observed limb profiles using an onion peeling technique. Most of the limb mode observations in Table 1 had a limb profile with a single peak. These observations were used for retrieval and study of the emission layer’s characteristics. In Table 1 we report only the average of the parameters for each session. The three limb profiles in Table 1, selected as an example among the

E00B38

many others with double peaks but not reported, are discussed below but were not used for our statistical analyses. [9] The nadir observations that were examined in greatest detail in this study are reported in Table 1. For simplicity here we call ‘‘nadir observations’’ all those where the instrument line of sight (LOS) intercepts the solid body of the planet, although the term ‘‘nadir’’ is more appropriate for perpendicular view only. All have exposure times of 3 to 18 s, since the a-X(0– 0) band at 1.27 mm is sufficiently bright to provide enough signal for mapping its geographic distribution with these exposure times. The spatial resolution for the selected nadir observations is typically about 10 km. [10] The global nadir view map for nightglow was constructed using all data collected during 880 orbits between 14 May 2006 and 14 September 2008 with emergence angle less than 80° (0° = nadir). The spanned period corresponds to about 4 Venusian years. The oxygen nightglow map was constructed from observations with exposure times of at least 0.1 s.

4. Data Processing and Separation of Thermal and Nightglow Emissions [11] Nightside infrared spectra of Venus contain nightglow emission from multiple species and thermal emission, which originates in the lower atmosphere [Allen and Crawford, 1984]. In the spectral regions where absorption by CO2 is small, the thermal emission can be observed outside Venus’ atmosphere after having undergone multiple scattering within Venus’ cloud layers [Kamp et al., 1988]. One window in the CO2 absorption spectrum lies near 1.27 mm [Meadows and Crisp, 1996], so emission originating from below 20 km altitude [Tsang et al., 2008] and scattered by higher-altitude clouds overlaps spectrally and, for nadir observations, spatially with the a-X(0 – 0) nightglow emission. Although the peak of the a-X(0 – 0) nightglow emission occurs at a shorter wavelength than the maximum of the thermal emission, it is not possible to spectrally resolve and quantitatively separate the two sources with VIRTIS-M’s spectral resolution. However, nearby spectral windows, at 1.18 and 1.74 mm, provide information on the thermal emission from the lower atmosphere that may be used to assess the contribution of thermal emission to the observed 1.27 mm radiance and to empirically remove the thermal emission contribution when significant. The nightside radiance at 1.18 mm originates from the surface and the lowest 20 km of the atmosphere [Meadows and Crisp, 1996]. The 1.74 mm radiance originates from near 20– 28 km [Meadows and Crisp, 1996]. Given existing uncertainties in the CO2 spectrum near 1.27 mm [Tsang et al., 2008], an empirical approach for removing the thermal emission contribution at 1.27 mm is probably as accurate as more detailed radiative transfer calculations. [12] For limb observations, as shown in Figure 2a, the thermal emission at 1.18 and 1.74 mm is much smaller than the radiance at 1.27 mm between 90 and 103 km. The difference is attributed to the nightglow and therefore, given the expected correlation among these bands for the thermal emission, it is safe to neglect the thermal emission at 1.27 mm in this altitude range. Some typical limb profiles of the oxygen nightglow intensity at 1.27 mm are shown in

3 of 18

E00B38

PICCIONI ET AL.: VENUS O2 NIGHTGLOW FROM VIRTIS

Figure 1. Limb and nadir view modes. (a) Orbit 76_18, case of limb observation with slit (parallel to the horizontal direction) almost perpendicular to the limb in the central part of the image. The image of the oxygen nightglow emission at 1.27 mm and the images in 1.18 and 1.74 mm (thermal emission of the lower atmosphere scattered by the clouds) windows are shown. The isolines are the tangent altitude above the Venus ellipsoid. Scattered thermal emission extends up to altitudes exceeding 80 km. (b) Orbit 271_09, case of limb observation with slit approximately parallel to the limb. (top) Image at 1.27 mm. (bottom) Image of the same cube but at 1.74 mm, same scale. The isolines are the tangent altitude. (c) Orbit 44_01, case of nadir mode. Several spectra in the 1.27 mm range are shown for different points of the image. Different colors of spectra coincide with the same color of area in the red-green-blue (RGB) image. Positions of the R, G, and B bands are shown on the spectra. The area of O2 emission appears white in this composite. The shift to shorter wavelengths of the nightglow maximum emission compared to the maximum of the thermal emission from below the clouds can be seen (compare for example the green and magenta spectra). The small peak in the spectra at about 1.31 mm is a thermal emission in another atmospheric window.

4 of 18

E00B38

PICCIONI ET AL.: VENUS O2 NIGHTGLOW FROM VIRTIS

E00B38

E00B38

Table 1. Summary of the Selected Dataa Volume Emission Rate

Limb Session

Start Time

Latitude Range (deg)

VI0076_18b VI0271_09 VI0275_06 VI0317_06b VI0317_07b VI0317_08b VI0320_03b VI0320_04b VI0321_03b VI0321_04b VI0321_05b VI0322_06b VI0322_07b VI0322_08b VI0323_07b VI0323_08b VI0324_06b VI0324_07b VI0324_08b VI0327_05b VI0327_06b VI0327_07 VI0330_04b VI0330_05b VI0330_06 VI0364_07 VI0364_08 VI0364_09 VI0371_10b VI0377_11b VI0383_12 VI0443_06b VI0449_10b VI0457_08 VI0460_08 VI0461_07 VI0499_11b VI0507_11 VI0507_12b VI0516_05b VI0517_05b VI0519_05b VI0565_18b VI0565_19b VI0565_20b VI0600_00 VI0600_02b VI0602_00 VI0602_02b VI0623_00 VI0623_01b VI0718_03 VI0724_02 VI0733_02 VI0742_02 VI0754_01 VI0754_02 VI0764_01 VI0764_02 VI0766_01 VI0766_02 VI0792_01 VI0792_02 VI0796_02 VI0796_03 VI0802_00 VI0802_01 VI0802_02 VI0806_01 VI0806_02

2006-07-06T01:34:11.885 2007-01-17T07:36:25.808 2007-01-21T07:33:11.848 2007-03-04T06:12:02.447 2007-03-04T06:26:02.459 2007-03-04T06:40:02.483 2007-03-07T06:09:36.409 2007-03-07T06:23:36.415 2007-03-08T06:08:48.416 2007-03-08T06:22:48.431 2007-03-08T06:36:48.443 2007-03-09T06:07:58.525 2007-03-09T06:21:58.413 2007-03-09T06:35:58.536 2007-03-10T06:21:10.344 2007-03-10T06:35:10.475 2007-03-11T06:06:22.483 2007-03-11T06:20:22.504 2007-03-11T06:34:22.594 2007-03-14T06:03:57.504 2007-03-14T06:17:56.492 2007-03-14T06:31:57.442 2007-03-17T06:01:32.528 2007-03-17T06:15:32.429 2007-03-17T06:29:32.559 2007-04-20T05:36:38.960 2007-04-20T05:50:45.877 2007-04-20T06:04:52.891 2007-04-27T05:42:00.927 2007-05-03T05:37:23.199 2007-05-09T05:51:47.236 2007-07-08T07:36:05.904 2007-07-14T07:38:35.961 2007-07-22T07:14:22.048 2007-07-25T06:58:28.022 2007-07-26T06:53:09.967 2007-09-02T03:13:58.502 2007-09-10T02:34:10.606 2007-09-10T03:07:10.565 2007-09-19T02:04:08.786 2007-09-20T01:58:48.667 2007-09-22T01:53:38.770 2007-11-07T00:50:16.235 2007-11-07T01:12:15.305 2007-11-07T01:26:16.217 2007-12-12T02:07:05.806 2007-12-12T02:45:05.782 2007-12-14T02:10:56.781 2007-12-14T02:48:55.817 2008-01-04T02:56:19.960 2008-01-04T03:14:19.984 2008-04-08T04:36:17.693 2008-04-14T04:30:38.081 2008-04-23T04:06:09.731 2008-05-02T04:20:36.192 2008-05-14T02:54:24.287 2008-05-14T03:08:24.287 2008-05-24T03:02:37.862 2008-05-24T03:16:38.266 2008-05-26T03:05:07.274 2008-05-26T03:19:07.289 2008-06-21T03:37:10.343 2008-06-21T03:50:10.343 2008-06-25T03:42:10.367 2008-06-25T03:55:10.371 2008-07-01T03:32:44.307 2008-07-01T03:49:44.389 2008-07-01T04:03:44.411 2008-07-05T03:54:48.428 2008-07-05T04:08:48.562

30.0:65.0 15.0:35.0 15.0:40.0 15.0:30.0 25.0:50.0 50.0:75.0 15.0:35.0 25.0:50.0 15.0:30.0 25.0:50.0 50.0:75.0 15.0:30.0 25.0:50.0 60.0:70.0 30.0:50.0 50.0:60.0 15.0:30.0 25.0:50.0 50.0:60.0 15.0:30.0 35.0:50.0 50.0:75.0 15.0:30.0 25.0:45.0 50.0:75.0 15.0:30.0 30.0:45.0 55.0:65.0 35.0:60.0 35.0:60.0 35.0:50.0 0.0:20.0 0.0:20.0 0.0:25.0 0.0:25.0 0.0:25.0 10.0:40.0 10.0:40.0 0.0:20.0 0.0:30.0 0.0:25.0 0.0:25.0 10.0:15.0 25.0:40.0 45.0:70.0 10.0:25.0 50.0:65.0 10.0:25.0 50.0:75.0 10.0:25.0 20.0:50.0 0.0:20.0 0.0:5.0 5.0:30.0 5.0:15.0 25.0:45.0 45.0:70.0 25.0:45.0 45.0:70.0 25.0:45.0 45.0:70.0 25.0:40.0 45.0:70.0 25.0:40.0 45.0:70.0 10.0:20.0 25.0:40.0 45.0:65.0 25.0:40.0 45.0:70.0

Local Time Range (h)

Peak From Limb (km)

21.8:23.9 23.8:2.4 0.3:2.9 23.8:0.7 23.7:0.4 0.03:7.1 0.18:1.1 0:0.8 0.31:1.2 0.1:0.9 0.46:7.4 0.43:1.3 0.21:0.97 0.57:7.5 0.32:1.08 0.68:7.6 0.67:1.5 0.43:1.2 0.78:7.7 1.05:1.8 0.77:1.5 1.11:7.99 1.4:2.1 1.1:1.8 1.4:8.26 21.4:21.9 21.6:22.3 17.8:21.8 21.9:23.5 22.6:0.16 18.03:0.8 0.9:2.86 1.6:3.5 19.2:21.0 19.5:21.4 19.6:21.5 23.9:3.1 0.8:3.3 23.6:1.7 0.9:3.0 1.0:3.1 1.3:3.4 2.6:3.5 2.0:2.9 2.2:5.1 22.2:23.5 20.2:23.7 22.4:23.7 20.4:23.9 0.8:2.2 1.4:2.4 22.4:0.3 22.4:0.8 0.7:2.2 1.3:2.9 22.0:23.0 22.2:0.7 23.1:0.1 23.2:1.9 23.3:0.3 23.5:2.1 2.3:3.1 2.3:4.7 2.8:3.6 2.7:5.2 20.1:20.8 20.5:21.3 18.9:21.3 20.9:21.8 19.5:21.8

95.6 95.7 94.5 94.3 94.2 95.2 93.0 94.8 92.3 96.0 98.2 92.7 96.4 98.0 96.5 97.0 96.7 96.8 98.0 96.3 93.3 93.2 92.7 93.3 93.8 97.3 97.0 98.0 97.2 97.0 94.0 94.8 92.5 93.0 93.2 93.8 98.8 98.5 94.8 94.7 94.8 95.4 100.0 97.3 93.0 96.3 100.3 97.0 93.2 93.3 95.0 93.5 94.5 97.6 94.5 93.5 93.0 93.0 91.6 93.8 92.8 98.0 93.5 91.3 91.2 92.5 94.7 98.8 96.3 95.6

5 of 18

Peak (km)

FWHM (km)

Peak Intensity (MR/km)

Total VER (MR)

96.7 97.3 97.3 96.0 96.6 97.4 95.3 96.2 95.3 98.3 101.4 96.3 99.4 101.0 99.8 101.0 98.7 99.6 101.0 101.0 95.0 94.6 95.7 95.0 95.4 99.7 99.3 98.5 99.6 99.6 95.3 98.3 95.8 96.6 96.4 97.0 100.2 100.3 98.3 96.8 97.0 98.2 103.0 98.7 95.0 99.3 101.3 101.7 94.4 96.7 97.3 98.0 98.5 99.8 97.5 95.8 95.0 95.8 93.6 96.3 94.2 100.7 95.0 94.7 94.0 95.0 96.3 99.8 98.3 96.8

3.9 5.7 6.3 9.7 9.0 5.8 8.5 6.2 9.3 8.7 6.4 9.3 6.6 5.0 6.3 7.0 9.3 7.8 8.0 11.0 7.3 6.4 9.7 7.0 5.6 7.7 7.0 3.5 5.6 7.4 6.7 9.8 10.3 10.0 10.4 9.8 6.5 7.2 10.3 9.7 9.0 8.0 9.0 6.0 7.4 8.0 3.7 11.0 5.4 9.3 7.3 11.0 12.5 8.4 10.0 7.3 6.5 8.8 8.2 7.0 6.8 8.3 7.5 8.7 7.6 9.0 6.0 4.3 9.3 4.6

0.06 0.14 0.06 0.12 0.05 0.04 0.03 0.04 0.03 0.06 0.13 0.03 0.03 0.09 0.08 0.14 0.10 0.08 0.08 0.12 0.03 0.02 0.02 0.03 0.02 0.09 0.11 0.10 0.11 0.10 0.04 0.11 0.08 0.01 0.01 0.02 0.10 0.11 0.11 0.10 0.15 0.12 0.09 0.04 0.01 0.08 0.14 0.06 0.02 0.03 0.06 0.05 0.10 0.14 0.07 0.06 0.01 0.03 0.02 0.03 0.02 0.07 0.03 0.01 0.01 0.06 0.05 0.16 0.10 0.04

0.27 0.91 0.43 1.10 0.49 0.26 0.31 0.26 0.29 0.45 0.85 0.25 0.24 0.47 0.64 1.01 0.96 0.59 0.58 1.19 0.28 0.10 0.23 0.21 0.11 0.73 0.77 0.48 0.78 0.81 0.29 1.03 0.90 0.12 0.13 0.17 0.71 0.87 1.16 1.03 1.30 0.99 0.86 0.27 0.10 0.74 0.69 0.68 0.15 0.32 0.44 0.58 1.32 1.28 0.73 0.47 0.11 0.26 0.17 0.24 0.11 0.67 0.25 0.10 0.09 0.55 0.37 0.74 0.98 0.26

PICCIONI ET AL.: VENUS O2 NIGHTGLOW FROM VIRTIS

E00B38

E00B38

Table 1. (continued) Volume Emission Rate Local Time Range (h)

Peak From Limb (km)

Limb Session

Start Time

VI0843_03 VI0847_03 VI0320_05 VI0321_04 VI0323_06

2008-08-11T05:19:18.518 2008-08-15T05:28:28.552 2007-03-07T06:37:36.350 2007-03-08T06:22:48.431 2007-03-10T06:07:10.379

25.0:45.0 25.0:45.0 60.0:70.0 35.0:45.0 20.0:30.0

1.3:2.0 1.8:2.4 0.4:7.3 0.1:0.9 0.5:1.4

Nadir Session

Start Time

Latitude Range (deg)

Local Time Range (h)

Exposure Time (s)

VI0044_01 VI0093_00 VI0093_01 VI0093_02 VI0264_04 VI0344_01 VI0574_05 VI0586_01

2006-06-03T15:06:56.666 2006-07-22T15:00:09.328 2006-07-22T15:28:09.283 2006-07-22T15:56:09.355 2007-01-09T20:36:56.857 2007-03-30T23:32:32.734 2007-11-15T18:42:55.387 2007-11-27T16:53:15.434


79.4:
9.9
59.1:
2.5
60.9:7.3
45.2:11.3
78.6:
10.6
66.2:
9.3
40.1:15.7
62.3:8.6

19.4:2.7 1.3:4.7 20.9:1.5 19.3:22.8 19.6:3.1 19.6:23.7 21.5:0.4 21.6:2.6

3.3 3.3 3.3 3.3 3.3 18.0 8.0 8.0

93.5 94.3

Peak (km) 95.3 97.0 Limb profile Limb profile Limb profile

FWHM (km)

Peak Intensity (MR/km)

Latitude Range (deg)

6.3 7.0 with double with double with double

0.03 0.06 peak peak peak

Total VER (MR) 0.23 0.45

a List of the selected data in limb and nadir mode and some measured parameters. The session name is shown. The first two letters of the session name (VI) are fixed and they mean ‘‘Virtis Infrared’’; the four numbers after VI are the orbit number; the two suffix numbers after the underscore are the session number, which is the progressive observation number within the considered orbit. The start time of the acquisition is shown. Each limb observation lasts typically 30 – 40 min. The range of latitude spanned by the considered observation is given. The local time spanned range by the considered observation is given. The observed peak altitude for the limb profile, averaged over the latitude range of the session, is given (only limb mode). The parameters from the retrieved volume emission rate are the peak altitude, the FWHM, the peak intensity, and the integrated volume emission rate along the perpendicular direction. All the parameters reported are an average of all the grid point values within the range of latitude of each session. For the peak altitudes and the FWHM, an accuracy of ±1 km can be assumed. For the peak intensity and VER, an accuracy of ±10% can be assumed. The exposure time for all the limb observations is 8.0 s. b Used to infer the ratio 1.27 mm/1.58 mm. The ratio is calculated from a portion of the session where the intensity is brighter and not from the full session. For nadir data, the exposure time is 3.3, 8.0, or 18.0 s.

Figure 2b. Figure 2 includes limb profiles from the sessions shown in Figures 1a and 1b (sessions 76_18 and 271_09). [13] In nadir mode observations, the thermal emission cannot be neglected so it must be removed from the observed radiance to quantify the oxygen nightglow emission. Our first step in determining the thermal emission was to construct a false color composite red-green-blue (RGB) image using data from the 1.285 (red), 1.275 (green), and 1.265 (blue) mm wavelengths, as shown in Figure 1c. Regions dominated by nightglow emission appear blue or white. Those dominated by thermal emission are pink or red. To quantitatively remove the thermal contribution at 1.27 mm, we have scaled the 1.18-mm radiance by an empirical factor ranging from 0.27 to 0.38. This empirical scaling factor was determined by selecting spectra with a small intensity at 1.265 mm and a high correlation between the radiances at 1.18 and 1.285 mm (reddish areas without O2 emission on the RGB images). Variations in the scaling factor across the disk are due mostly to topography but there is a weak dependence on cloud structure. [14] The scaling factor to remove the thermal contribution at 1.27 mm also has been estimated using radiative transfer model calculations. Synthetic spectra were calculated using a surface temperature of 735 K and the typical cloud structures found at middle to low latitudes [Zasova et al., 2007]. The total vertical cloud opacity at 1.18 mm was varied from 20 to 50 by varying the opacity of each layer without altering the structure of the clouds, which were assumed to be composed of mode 1 and mode 2 particles with scale height of 4 km. The calculations used the ARS code [Ignatiev et al., 2005], which is based on DISORT

[Stamnes et al., 1988], with 48 streams and the HITEMP database [Pollack et al., 1993] for CO2 absorption. For the synthetic spectra, the ratio of the integrated emission in the 1.27- and 1.18-mm windows, r = I(1.27)/I(1.18), is in the range from 0.28 to 0.34, consistent with the empirical ratio from 0.27 to 0.38. The ratio depends only weakly on the total vertical cloud opacity because the single scattering albedo of the particles is very high (0.99999 at 1.18 mm). The single scattering albedo in the 1.74-mm window is smaller, 0.99978, so it does not work as well for the thermal correction. The ratio is most sensitive to the surface temperature. If the surface temperature is changed by 15 K, which is equivalent to a 2 km change in surface altitude, then the ratio changes by 0.03 for a cloud opacity of 30. [15] Vertical profiles of the volume emission rate are retrieved from the observed limb profiles using an onion peeling technique [Sharma et al., 1988]. Figure 3 shows the vertical profile of the volume emission rate of the (0 – 0) oxygen nightglow retrieved for orbit 317_06. The emission rate is expressed in MR/km (or 107 photons cm
3). The technique assumes the atmosphere is spherically symmetric, including the airglow emission layer. Nadir-view images, however, indicate the nightglow emission is spatially inhomogeneous. The effective consequence of our assumption of spherical symmetry is to distribute the nightglow emission observed at higher altitudes over a larger region than may be true in reality. As a result, the retrieved emission at lower altitudes is underestimated. Clear evidence for this is found when the retrieved intensity below the peak altitude becomes negative. For several orbits where the spherical symmetry assumption led to negative values for the retrieved nightglow emission below 90 km, we imposed

6 of 18

E00B38

PICCIONI ET AL.: VENUS O2 NIGHTGLOW FROM VIRTIS

Figure 2. Limb profiles distribution of the O2 (0 –0) nightglow at 1.27 mm. (a) Limb profiles of radiance at 1.18 and 1.74 mm, shown in comparison with the O2 (0 – 0) limb profile at 1.27 mm. The radiance at 1.18 and 1.74 mm is due to the scattered thermal emission from the deep atmosphere scattered to VIRTIS-M by the upper haze particles. Both radiances decay rapidly in the transition region of the upper boundary of the aerosol at about 90 km altitude. The thermal contribution at 1.27 mm should have a similar behavior; therefore, in the typical altitude range where the nightglow emitter resides (90 – 105 km), an almost pure nightglow emission is observed in limb view. (b) The first three numbers of the curve names are the number of the orbit, while the given range after the orbit number is the latitude range where the average was taken. Data from all local times spanned in the session were included in the average. One can see that the peak altitude emission and its brightness significantly vary from one orbit to the next one and even in the same orbit. From Figure 2, the profile seems wider in altitude at lower latitudes, and a double peak is often present with a peak at 96 and 103 km.

E00B38

Figure 3. Vertical profiles of the O2 (0 – 0) retrieved volume emission rate. Black curves are volume emission rate in O2 (0 – 0) band at 1.27 mm, for orbit 317_06 in different latitude ranges; red curve is normalized radiance at 1.18 mm, which describes the upper boundary for the lower atmosphere thermal emission scattered into the VIRTIS-M field of view limb view. The normalized limb profile of thermal emission at 1.18 mm coincides well with the retrieved emission at 1.27 mm below 90 km, which is also thermal emission from the lower atmosphere. set of limb observations from different orbits over a period of about 2 weeks (in contour plot style). [17] These results reveal several characteristics of the O2(a-X) nightglow emission. As was previously known, the intensity of the emission is highly variable. Figure 2b shows the maximum observed peak limb intensity varying from about 15 to 80 MR, and the minimum and maximum observed peak limb intensities in the 72 sessions studied are 4 and 100 MR, respectively. Figure 2b also demonstrates that the observed altitude of peak brightness can vary. This

the additional constraint that the retrieved nightglow emission must be nonnegative. From these test cases, the upper limit on the geographic size of the emitting area was 500 – 1500 km, consistent with nadir images. When the retrieval is satisfactory, the shape of the vertical profile of the 1.27-mm radiance below 90 km coincides with that of the 1.18 mmradiance, both providing the profile of thermal radiance from the deep atmosphere as scattered in the upper haze layer.

5. Vertical Distribution of the O2(a-X) Nightglow Emission at 1.27 mm [16] In Figure 4 we show observations from a single orbit of the (0 – 0) limb profiles averaged over 5° latitude intervals which were also averaged over the full local time span at each latitude, about 10– 20 min; Figure 5 shows a selected

Figure 4. Limb profiles distribution of the O2 (0 – 0) nightglow in limb mode. The intensity in limb view for orbit 317 at 35 –73°N, session 7 and 8, is shown. Different peak altitudes are observed at different latitudes with a regular trend of correlation within this orbit. Also, the full width at half maximum (FWHM) appears wider toward the equator.

7 of 18

E00B38

PICCIONI ET AL.: VENUS O2 NIGHTGLOW FROM VIRTIS

E00B38

Figure 5. Limb profiles distribution of the O2 (0– 0) nightglow at 1.27 mm versus latitude. The intensity is integrated over all the local times at each latitude. The color bar is the intensity along the line of sight. The orbit and session number are given in the titles. The intensity can differ significantly from one limb observation to another, as, for example, in the case of Figures 5a and 5b. The altitude can be almost constant, like in Figure 5f, and moderately constant like in Figure 5c, or more frequently a slope is observed like in Figures 5d and 5e. Some small ‘‘wave-like’’ features, as in Figure 5b at 27°N, are real and may possibly be attributed to gravity waves. is new information that cannot be obtained from nadir observations. In Figure 2b the observed peak altitude of the limb profile ranges from 96 to 103 km (for the higher peak of the double-peak profile), and the minimum and maximum altitudes for the observed peak of the nightglow limb profile in the 72 sessions studied are 90 and 102 km

(for the single-peak profiles), respectively. Further study is required to assess whether the variations in the observed peak altitude may explain the temperature variations derived from ground-based observations of the emission [Bailey et al., 2008]. Finally, significant variations have been detected in the shape of the observed limb profile (Figure 2b). Most

8 of 18

E00B38

PICCIONI ET AL.: VENUS O2 NIGHTGLOW FROM VIRTIS

observed limb profiles have a single peak, as has been reported previously [Drossart et al., 2007a; Ge´rard et al., 2008], but some observed limb profiles have a double peak. This has not been reported previously and is discussed later in this paper. [18] A view of the longer-term average characteristics of the (0 – 0) limb observations can be derived by using all of the limb data available in the VIRTIS data set to investigate the statistical characteristics of the retrieved vertical emission rate. The observed limb profile for each session, averaged over all local times (Table 1) was determined, assigned to a point on a 5° latitude range, and inverted to retrieve the vertical profile. Each retrieved vertical profile was then analyzed to derive the retrieved peak altitude, the peak altitude brightness, the FWHM of the retrieved vertical profile, and the total (column integrated) vertical emission rate. The mean values for each of these parameters for each latitude range are calculated and shown in Figure 6. Figure 6a shows the number of available limb observations for each latitude range. In Figure 6b, we show the meridional distribution of the mean peak altitude brightness. The abscissa is latitude, as for the other plots of Figure 6, and the ordinate is the intensity of the vertical emission rate at the retrieved peak altitude. The mean peak emission rate ranges from about 0.03 to 0.11 MR/km. Figure 6c shows the meridional distribution of the mean values of the altitudes at which the integrated line of sight (as observed, red curve) and vertical emission (as retrieved, blue curve) rates are maximum. For an ideally thin emitting layer, the observed peak altitude would coincide with the retrieved peak altitude. The observed peak altitude is systematically lower than the retrieved peak altitude by 2 – 3 km. Most of this bias is due to the geometrical effect of a not ideally thin emitting layer. In Figure 6d is shown the meridional distribution of the mean FWHM of the retrieved vertical profile of the oxygen nightglow. Figure 6e shows the total (column integrated) vertical emission rate, which, in turn, is equivalent to the intensity that would be seen in nadir view for pure oxygen nightglow emission. Finally, in Figure 6f, isolines of the mean total (column integrated) vertical emission rates are shown as a function of latitude and the retrieved peak altitude. The uncertainty in the peak altitudes and the FWHM is ±1 km, while that for the emission rate is ±10%. In all cases, however, the observed fluctuations are dominated by temporal variability in the nightglow emission. For this reason the error bar in the plots reports the standard deviation of this fluctuation. [19] While the mean values discussed above represent the general morphology, there are occasional but significant deviations. For example in many cases the peak intensity decreases with increasing latitude, but sometimes it is observed an inverse behavior, as the one for example shown in Figure 2b, sessions 321_04 and 321_05. In some orbits of Figure 5, the maximum of the nightglow intensity is at subequatorial latitudes but there are secondary local maxima at higher latitudes. For example, session 324_06 (Figure 5d) has one maximum near 30°N and orbit 330 (Figures 5e and 5f) has several maxima at 25 –45°N. Figure 5 also suggests an episodic variation in the maximum nightglow intensity with time for the region that was observed. The maximum intensity decreased by a factor of 4 over four days from session 317_06 (Figure 5a) to session 321_03 (Figure 5b)

E00B38

with almost the same local time; increased over the next three days, so that the observed maximum intensity in session 324_06 (Figure 5d) is similar to that in session 317_06; and then decreased again in session 330_04 (Figure 5e) to values similar to those observed in session 323_06 (Figure 5c). Some dynamical factors, such as breaking gravity waves at a local scale and tides at larger scales may disrupt the general picture, so long-term averages may reveal different dynamical information than the single orbit snapshots. [20] The spatial distribution of the nightglow intensity derived from limb observations (Figure 6) is consistent with that found from the previous analysis of VIRTIS’ nadir observations [Ge´rard et al., 2008] and our analysis of a more complete set of nadir observations, as discussed later. In general, there is a negative correlation between the intensity of the emission and latitude. As shown in Figure 6b, the most intense emission observed is found at low latitudes. The peak altitude and the FWHM of the nightglow’s vertical profile also show regular trends with latitude. As shown in Figure 6c, the mean retrieved peak altitude is highest at 45– 50°N with a value of about 99 km, and appears to be relatively constant equatorward with a value of 97– 98 km, while it decreases poleward with a relative minimum of 95 km at 65– 70°N. The vertical FWHM of the emission shows a clearer trend, increasing from high to low latitudes with the largest FWHM, 11 km, at 5– 10°N and the smallest, 6 km, at 60– 65°N (see Figure 6d). The decrease of the observed peak altitude and its larger deviation (to lower altitudes) from the retrieved peak altitude toward the equator, see Figure 6c (left), can be explained by the increase of the vertical FWHM in this region. The thicker the emitting layer, the larger the difference between the observed and retrieved peak altitudes. A similar trend is found in Figure 4, where the observed peak altitude of the nested profiles increases with latitude, while FWHM and intensity decrease. The decrease in the peak altitude poleward of 45°N (Figure 6c) may indicate a different dynamical regime exists at the altitude of the emitting layer. This may be related to the influence of the polar vortex. [21] Regions of intense nightglow where the peak altitude is approximately constant are likely to be regions of convergence for the subsolar to antisolar flow with strong downwelling. In terms of pressure, a constant peak altitude is consistent with the conditions to have very low horizontal pressure gradient forces which will suppress horizontal flow and give predominantly vertical flow. This is illustrated by the distribution of the total (column integrated) vertical emission rates derived from our limb observations (Figure 6e) and the distribution of peak altitudes (Figure 6c). The former increases by a factor of 6 from about 0.2 MR at 60–70°N to about 1.2 MR near the equator. The latter is relatively constant equatorward of 45°N. The overall picture derived from our limb observations is consistent with that derived from our own and previous analyses of nadir observations. In addition, we have been able to determine that the enhanced total (column integrated) vertical emission rate near the equator is due to both a more intense vertical emission rate at the retrieved peak altitude (Figure 6b) and a larger vertical FWHM for the emitting layer (Figure 6d). The former contributes a factor of three to the equatorial enhancement and the latter a factor of two. Finally, as shown in Figure 6f, the most intense O2 nightglow emission is

9 of 18

E00B38

PICCIONI ET AL.: VENUS O2 NIGHTGLOW FROM VIRTIS

Figure 6. Mean vertical profile distribution of the O2 (0 – 0) nightglow at 1.27 mm. (a) Number of available sessions in the latitude range is shown to indicate the number of samples over which the average is calculated for each latitude range. (b) Mean emission rate at peak altitude as a function of latitude. The error bar is the standard deviation. The maximum is found in the equatorial region, and the intensity decreases poleward. (c) Mean peak altitude of the volume emission rate (blue curve) and in limb profile (red curve) as a function of latitude. The error bar is the standard deviation. Figure 6 indicates the peak altitude of the volume emission is relatively constant and tends to decrease poleward of 45– 50°N. The weak decrease of the peak altitude of the limb profile (red curve) is likely related to the wider FWHM at low latitudes. (d) Mean FWHM of the vertical profile as a function of latitude. The error bar is the standard deviation. The results show the FWHM is largest at low latitudes and decreases poleward with some relative peak at 45– 50° and 70°N. (e) Mean total integrated vertical emission rate as a function of latitude. The error bar is the standard deviation. The maximum is found in the equatorial region, and the intensity decreases poleward. The slope is less pronounced than in Figure 6b owing to the amplification factor of Figure 6d. (f) Same as Figure 6e but as a function of latitude and peak altitude of the vertical profile. An uncertainty of ±10% in emission can be assumed. The intensity is highest at low latitudes when the peak altitude is higher.

10 of 18

E00B38

E00B38

PICCIONI ET AL.: VENUS O2 NIGHTGLOW FROM VIRTIS

E00B38

Figure 7. Population distribution of the oxygen nightglow parameters. (a) Distribution of the peak altitude emission rate intensity, same parameter as Figure 6b. It is remarkable the large variability of the intensity which is spread over a wide range of emission. The global mean value of the peak emission rate intensity is 0.064 ± 0.050 MR/km. (b) Distribution of the peak altitude of the vertical profile. Variability is less pronounced than in Figure 7a, and the global mean value of the peak altitude of emission is 97.4 ± 2.5 km. (c) Distribution of the FWHM of the vertical profile. The global mean value of FWHM is 7.6 ± 2.2 km. (d) Distribution of the total integrated vertical emission rate. Also, in this case the large variability is noticeable. The global mean value of the integrated emission rate is 0.52 ± 0.40 MR. observed at near equatorial latitudes when the retrieved peak altitude is at least 98 km. These are the most favorable conditions for radiative deexcitation as opposed to collisional quenching. [22] The statistical distributions of the parameters are shown in Figure 7. In average, the peak vertical emission rate intensity (peak altitude brightness) is 0.064 ± 0.050 MR/km (Figure 7a); the retrieved peak altitude is 97.4 ± 2.5 km and the observed peak altitude is 95.0 ± 2.4 km (Figure 7b); the vertical FWHM is 7.6 ± 2.2 km (Figure 7c); the total (column integrated) vertical emission rate is 0.52 ± 0.40 MR (Figure 7d). The observed and retrieved peak altitudes vary by ±2.5% across the range of latitudes observed while the total (column integrated) vertical emission rate varies by more than ±75%. A factor of 50 has been inferred from the limb to the perpendicular direction views of the intensity of the emitting layer in average.

6. Detection of Double-Peaked Nightglow [23] In many cases two emission peaks are found in the observed limb profiles. Some examples are shown in Figures 2 and 8. In Figure 8a the measured limb profiles are shown on the left and the corresponding limb images at 1.27 mm on the right. The separation of the two layers is

clear. On the occasions when a double peak is detected, the lower and higher peaks occur typically at 96– 98 and 103– 105 km, respectively, see Figure 8a and Figure 8b (left) where the retrieved vertical profiles are shown. Double peaks are more likely to occur at high latitudes, 45– 70°N, although there are also occasional examples at lower latitudes. The higher peak is often the more intense of the two. [24] The origin of the double peak is not yet understood. However, some suggestions can be provided. Given the uneven global distribution of the nightglow, one may think that the higher and lower peaks are the results of probing two separate emitting layers, each at a different altitude level. If so, the double peak would be a consequence of the viewing geometry. It is also possible the viewing geometry could produce an apparent double peak if both emitting regions are at the same radial distance from Venus but with a gap between them where there is less nightglow emission. Supposing this to be true for the case shown in Figure 8a, for the session 320_05, let’s assume the higher peak as coming from a single-peaked layer located in a region at the LOS tangent altitude. The lower peak might be explained then with a local maximum of another emitting region placed off by 3° (about 310 km) from the former and at the same radial distance. However, this region of local maximum, estimated from the black curve of Figure 8a to

11 of 18

E00B38

PICCIONI ET AL.: VENUS O2 NIGHTGLOW FROM VIRTIS

E00B38

Figure 8. Profiles of the oxygen nightglow with double peak. (a) Comparison of different limb profiles for the (left) O2 (0 – 0) nightglow at 1.27 mm and (right) the image view of the same observations. The green isolines in the images are the tangent altitude. The yellow isolines of the same images are latitudes, and they mark the latitudinal range where the limb profiles are taken. Color scale intensity is not the same for the images, in order to optimize the contrast. Double peaks are noticed in the limb profile on the left plots, and on the right, spatially separated layers can be seen in the acquired images. (b) Vertical emission rate retrieved for the profiles shown in Figure 8a. The double peak is observed also for the retrieved profiles and it results even more enhanced with relative peaks very well separated vertically.

be about 2.5 times brighter in total VER than the region responsible of the higher peak, should develop in a long strip precisely at this same distance for more than 20° latitude, corresponding to about 2000 km of spatial displacement (see Figure 8a, top right). Therefore, although these combinations can reproduce a double-peaked limb profile in some cases, this scenario appears to be very unlikely to explain the two distinct emitting layers of Figure 8 and several other similar observed cases not reported here. [25] A third possible explanation, which is somewhat simpler and appears to be more likely than the other two, involves vertically propagating gravity waves. Gravity waves can modulate the conditions in the atmosphere, thereby modifying density, temperature, atomic oxygen abundance, and hence the collisional rate of recombining oxygen atoms and collisional quenching of the O2(a) state. As a matter of fact, wave structures at Venus were observed by the Pioneer Venus Orbiter star trackers [Bougher and

Borucki, 1994] in the same range of middle and high latitudes where double peaks are often found. Finally, it must be said that the retrieved atomic oxygen number density vertical profile is often double peaked even when the observed limb profile is single peaked [Ge´rard et al., 2009].

7. O2 Emission Rate From Nadir Observations [26] Figure 9 shows a nadir-looking view of the nightglow. The emission rate has been corrected for thermal emission in the way described above, while emergence angle and cloud backscattering are corrected as was done by Crisp et al. [1996]. Various structures with clearly defined boundaries are visible in Figures 9a– 9d that denote the complex interplay of dynamical and, possibly, chemical factors in the formation of the nightglow. Notwithstanding the significant variability observed, it seems that the distri-

12 of 18

E00B38

PICCIONI ET AL.: VENUS O2 NIGHTGLOW FROM VIRTIS

Figure 9

13 of 18

E00B38

E00B38

PICCIONI ET AL.: VENUS O2 NIGHTGLOW FROM VIRTIS

bution of the nightglow over the planet is not entirely random but shows structured recurrent features. The emission patches may appear elongated, as seen in Figures 9a– 9c, in the form of jets over extensions of 1700 to 2300 km, or as wavy features as in Figure 9d. [27] We have searched for a connection between the nightglow and the local temperature. For this purpose, we have retrieved the average thermal brightness (TB) from the radiance in the 4.23 – 4.28 mm spectral interval which probes, according to radiative transfer models [Grassi et al., 2008], the 90– 93 km altitude range. Our approach is somewhat simpler than that used in the formal retrieval in that work, but the TB provides reliable relative values. In Figures 9e and 9f, intense nightglow seems to be correlated with elevated temperatures. Other authors have reported similar conclusions [Ohtsuki et al., 2005, 2008; Bailey et al., 2008; Bertaux et al., 2007], and it has been suggested that the intense nightglow may be caused by strong downwelling and resulting compressional heating. Conversely, the dim nightglow regions, if affected by upwelling, may be involved in expansion cooling with consequently lower temperatures than the surrounding environment. The same effects may also be a consequence of the atmospheric temperature structure modulation induced by vertically propagating gravity waves. The circular dark region of faint nightglow has on average a local TB of 182 K, lower by about 3 K than the surroundings with a TB of 185 K. The TB may differ by a few K from the formally retrieved actual absolute atmospheric temperatures, but their relative accuracy is estimated to be better than 1 K, so these temperature differences are considered significant. The bright region at the top (Figure 9f) where TB is on average about 189 K, corresponds to the region in Figure 9e where the nightglow is brightest and therefore the downwelling is a maximum in this region. [28] We have also performed a long-term study of the nightglow global distribution. For this purpose, a series of observations made with exposure times of at least 0.1 s and emergence angles smaller than 80° have been selected from 880 orbits. The radiances are subsequently corrected for thermal radiation (a constant factor of 0.3 was used in this case for scaling the 1.18 mm radiance), emission angle and cloud backscattering as previously discussed. The resulting image is shown in Figure 10a. The brightest region has a total (column integrated) vertical emission rate of 1.2 MR, consistent with the maximum derived from limb observations (Figure 6e). The center of the brightest region is about 5°S of the AS point, the proximity to which is an expected outcome of the subsolar to antisolar circulation. Figure 10b

E00B38

shows the map of coverage for the data used in Figure 10a, where the color scale bar represents the log10 of counts.

8. Oxygen Band Ratio [29] The (0 – 1) band of the a-X system that occurs at 1.58 mm is weaker than the 1.27 mm band. It is the former, however, that was first observed in the terrestrial atmosphere in the evening twilight ground experiment by Vallance Jones and Harrison [1958]. The atmospheric transmission at 1.27 mm is significantly lower, making it difficult for photons at this wavelength to reach the Earth’s surface. Some years would have to pass before the in-flight detection of the (0 –0) band by Noxon and Vallance Jones [1962], and some more for Lowe [1969] to resolve it in a spectrum recorded from the ground. The nondetection of the (0– 0) band in the pioneering work of Vallance Jones and Harrison [1958] was interpreted by the authors to mean a ratio of transition probabilities A00/A01 = 10, which is the value still in place in the current version of the HITRAN database [Rothman et al., 2005]. Observations of the (0 – 1) band in the terrestrial atmosphere continue, driven by the interest in investigating the oxygen chemistry of the mesosphere and lower thermosphere. [30] There have been a few subsequent estimates of the ratio, that have come either from laboratory experiments, e.g., A00/A01 = 46.0 ± 0.7 [Findlay, 1969], 49.5 ± 4 [Becker et al., 1971], 80 ± 20 [Haslett and Fehsenfeld, 1969], and 55 ± 10 [Thomas and Thrush, 1977]; or from airglow observations, some times in combination with photochemical modeling, e.g., A00/A01 = 52 ± 25 [Winick et al., 1985], and 80 ± 15 [Pick et al., 1971]. More recently, Kassi et al. [2005] have determined the line strengths of a few lines belonging to the strong Q branch in a high-resolution absorption spectrum experiment. By comparison to the respective lines in HITRAN, and after the correction of a numerical inconsistency in the conversion of absorption strengths to transition probabilities (confirmed by the authors), the work of Kassi et al. [2005] leads to A00/A01 = 55– 60. Recalling that on the basis of the Franck-Condon principle the predicted ratio is A00/A01 = 148 [Kassi et al., 2005] and that the principle is less likely to operate in longlived states, it is important to note the fundamental interest that the discrepancy between theory and experiments bears. [31] The VIRTIS instrument can resolve simultaneously the absolute radiances from both emission bands in the nighttime spectrum of Venus’ upper atmosphere. This was utilized by Piccioni et al. [2008] to infer A00/A01 = 63 ± 6. We now provide the ratio from a more extended data set and also how the ratio was determined.

Figure 9. Observations of the oxygen nightglow in nadir view geometry. O2 (0 –0) nightglow observations at 1.27 mm, after thermal, emergence angle, and cloud backscattering corrections: (a) orbit 264, session 04; (b) orbit 586, session 01; (c) orbit 574, session 05; (d) orbit 344, session 01. The green and yellow isolines are latitude and local time, respectively. (e) Orbit 93, mosaic of session 00, 01, and 02. The isolines from left to right are local time while the others from top to bottom are latitude. (f) Thermal brightness of the same observation as in Figure 9e, averaged from 4.23 to 4.28 mm which probes, according to radiative transfer calculation, the 0.1 mbar pressure level, corresponding to about 90– 93 km altitude. The green and red isolines are local time and latitude, respectively. The extension of the long stripes in Figure 9a is of the order of 1700 – 2300 km. The thermal brightness (TB) of the dark patch indicated by the arrow is about 182 K, which is 3 K lower than the surrounding. This is visibly correlated with the corresponding dark patch in Figure 9e of the oxygen nightglow. The diagonal dimension of the dark circular feature is about 800 km. The brighter region in the top middle of Figure 9e has the highest TB of the image, about 189 K, which is consistent with the maximum downwelling zone. 14 of 18

E00B38

PICCIONI ET AL.: VENUS O2 NIGHTGLOW FROM VIRTIS

E00B38

bution of the thermal emission scattered by the clouds and haze, only observations with tangent altitude in the 90– 100 km range have been used. The continuum has been subtracted by linearly interpolating the observed radiances between 1.24 and 1.3 mm for the (0 – 0) band and 1.55 and 1.60 mm for the (0 –1) band. At the relevant temperatures, no contribution from the oxygen nightglow is expected outside these intervals. Subsequently we integrated the radiances for each transition to obtain the total emission rate. In Figure 11a the observed limb profiles from session 317_06 of the (0 – 0) and (0 – 1) bands at 1.27 and 1.58 mm, respectively, are shown as an example. In Figure 11b the ratios of the integrated radiances are shown. The mean radiance ratio integrated over the selected altitude range of 90– 100 km, and weighted by the inverse square of the uncertainty is 78 ± 10, which results in 63 ± 8 for the ratio of transition probabilities A00/A01. The uncertainty is largely due to the difficulty of evaluating the continuum of the 1.58-mm band so it is largest when the SNR in this spectral region is low.

9. Conclusions

Figure 10. Global map of the oxygen nightglow over 880 orbits in nadir geometry. (a) Mean global O2 (0 –0) nightglow at 1.27 mm, after thermal, emergence angle, and cloud backscattering corrections, averaged over 880 orbits. The map is constructed with all the available nadir observations in nighttime with an exposure time greater or equal than 0.3 s. The region of maximum emission, 1.2 MR, is near the antisolar (AS) point, a clear indication of the subsolar (SS) to AS circulation, but with a slight shift of the patch center toward the morning sector, to about 0015, and south, to 5°S. (b) Counts of the coverage map for the same data set as Figure 10a. Color scale bar is log10 (counts). The geometry of the orbit permits more frequent observations of the southern polar region owing to the longer persistence and farther distance of the apocenter which is located approximately over the south pole. However, it is shown that even in the less covered portion in the equatorial region, hundreds of observations are present. [32] For the analysis we have used portions of the 36 limb observations from Table 1 with relatively high intensity for the (0 – 0) band, of the order of 0.5 MR or greater, in the period March 2007 to January 2008. To avoid the contri-

[33] A study of Venusian molecular oxygen nightglow in the (0 –0) and (0 – 1) bands at 1.27 and 1.58 mm, respectively, from limb and nadir geometry has been conducted with the VIRTIS instrument on board Venus Express. Limb geometry data in the period from July 2006 to August 2008 as well as some nadir data, acquired in the period from 3 June 2006 to 27 November 2007, have been shown, described, analyzed, and compared with the thermal brightness at 4.23 to 4.28 mm, which probes the 0.1 mbar level of pressure, corresponding to about 90– 93 km altitude. All nadir data acquired in the period from 14 May 2006 to 14 September 2008, about 4 Venusian years, have been used to globally map the 1.27 mm emission averaged over this period of time (after correction for thermal emission, emergence angle and cloud backscattering). [34] The shape and the intensity of the 1.27 mm vertical profile are highly variable, however some regularities have been identified. The mean distribution of the total (column integrated) vertical emission rate inferred from limb observations is consistent with that derived from nadir observations over a 2-year period with the most intense nightglow emission found near the antisolar region. The retrieved peak altitude is relatively independent of latitude equatorward of 45– 50°N, with a mean value of 97.4 km while a different dynamical regime is identified poleward of this latitude. The vertical FWHM shows a significant increase with decreasing latitude and it reaches its maximum of 11 km in the equatorial region. The observed emission is most intense at low latitudes when the retrieved peak altitude is above 98 km. It has been shown that the intense total (column integrated) vertical emission rate that is observed near the antisolar point is due to both a more intense volume emission rate at the retrieved peak altitude and a thicker emitting layer. Limb profiles with two peaks have been observed. These could be related to vertically propagating gravity waves modulating the atmospheric structure. [35] Significant variability in the instantaneous local distribution of 1.27 mm emission rate in nadir mode is found, however features frequently appear very well struc-

15 of 18

E00B38

PICCIONI ET AL.: VENUS O2 NIGHTGLOW FROM VIRTIS

E00B38

Figure 11. O2 (0 – 0) and (0 – 1) limb profiles and radiance ratio. (a) Intensity of limb profiles for total integrated emission in the (0 – 0) and (0 – 1) O2 transitions at 1.27 and 1.58 mm are shown for session 317_06. The intensity was averaged over the full range of latitude and local time of the session. The red scale shows the intensity value for the (0 – 1) transition, in MR. Their peak altitudes are in agreement. (b) O2 (0 –0) and (0 – 1) radiance ratio in limb mode at 1.27 and 1.58 mm is shown. The statistic takes in account all the data analyzed in the period March 2007 to January 2008 marked with a superscript "b" in Table 1. The ratio is

calculated in a selected bright region of the session only and not over the full spanned latitude and local time range of the session. This is why there is no direct correspondence from the ratio calculated from Figure 11a and the relevant point of the session in Figure 11b. The weighted mean value of the radiance ratio is 78 ± 10, which translates into 63 ± 8 in terms of the ratio of the transition probabilities A00/A01. tured, with very elongated bright stripes similar to jets and wavy type formations. A correlation of the nightglow intensity and the thermal brightness probed at the altitude of about 90– 93 km is often found in the instantaneous snapshot. Dark circular shape features occur frequently in the nightglow map and usually have a thermal brightness that is lower than the surrounding bright nightglow region. This suggests these dark regions may be places at which upwelling flow has inhibited the recombination of oxygen atoms and produced localized cooling. These dark features could also be produced by vertically propagating gravity waves. Finally, both the (0 – 0) and (0– 1) bands have been simultaneously measured in the VIRTIS spectra and a ratio of the transition probabilities A00/A01 = 63 ± 8 has been inferred.

[36] Acknowledgments. We gratefully acknowledge the work of the entire VIRTIS and Venus Express teams that allowed these data to be obtained and discussed. We wish to thank ASI, CNES, and the other national space agencies that have supported this research. Russian coauthors acknowledge Russian Foundation of Basic Research for grant RFFI 08-02-00850. Partial funding for this research was provided by the Australian Research Council. Helpful comments from two anonymous reviewers, D. Titov and R. Carlson, are gratefully acknowledged.

References Alexander, M. J. (1992), A mechanism for the Venus thermospheric superrotation, Geophys. Res. Lett., 19, 2207 – 2210, doi:10.1029/92GL02110. Allen, D. A., and J. W. Crawford (1984), Cloud structure on the dark side of Venus, Nature, 307, 222 – 224, doi:10.1038/307222a0. Bailey, J., V. S. Meadows, S. Chamberlain, and D. Crisp (2008), The temperature of the Venus mesosphere from O2(a1Dg) airglow observations, Icarus, 197, 247 – 259, doi:10.1016/j.icarus.2008.04.007. Becker, K. H., W. Groth, and U. Schurath (1971), The ratio of the FranckCondon factors q(0, 0)/q(0, 1) of the infrared atmospheric band system of

16 of 18

E00B38

PICCIONI ET AL.: VENUS O2 NIGHTGLOW FROM VIRTIS

oxygen, Planet. Space Sci., 19, 1009 – 1010, doi:10.1016/00320633(71)90150-4. Bertaux, J. L., et al. (2007), A warm layer in Venus’ cryosphere and high altitude measurements of HF, HCl, H2O, and HDO, Nature, 450, 646 – 649, doi:10.1038/nature05974. Bougher, S. W., and W. J. Borucki (1994), Venus O2 visible and IR nightglow: Implications for lower thermosphere dynamics and chemistry, J. Geophys. Res., 99, 3759 – 3776, doi:10.1029/93JE03431. Bougher, S. W., S. Rafkin, and P. Drossart (2006), Dynamics of the Venus upper atmosphere: Outstanding problems and new constraints expected from Venus Express, Planet. Space Sci., 54, 1371 – 1380, doi:10.1016/ j.pss.2006.04.023. Cheah, S.-L., Y.-P. Lee, and J. F. Ogilvie (2000), Wavenumbers, strengths, widths, and shifts with pressure of lines in four bands of gaseous 16O2 in 1 + 3
the systems a1Dg
X3S
g and b Sg
X Sg , J. Quant. Spectrosc. Radiat. Transfer, 64, 467 – 482. Connes, P., J. F. Noxon, W. A. Traub, and N. Carleton (1979), O2(a1Dg) emission in the day and night airglow of Venus, Astrophys. J., 233, L29 – L32, doi:10.1086/183070. Crisp, D., V. S. Meadows, B. Be´zard, C. de Bergh, J.-P. Maillard, and F. P. Mills (1996), Ground-based near-infrared observations of the Venus nightside: 1.27 mm O2(a1 Dg) airglow from the upper atmosphere, J. Geophys. Res., 101, 4577 – 4594, doi:10.1029/95JE03136. Drossart, P., et al. (2007a), A dynamic upper atmosphere of Venus as revealed by VIRTIS on Venus Express, Nature, 450, 641 – 645, doi:10.1038/nature06140. Drossart, P., et al. (2007b), Scientific goals for the observation of Venus by VIRTIS on ESA/Venus express mission, Planet. Space Sci., 55, 1653 – 1672, doi:10.1016/j.pss.2007.01.003. Findlay, F. D. (1969), Relative band intensities in the atmospheric and infrared atmospheric systems of molecular oxygen, Can. J. Phys., 47, 687 – 691. Gamache, R. R., and A. Goldman (2001), Einstein A coefficient, integrated band intensity, and population factors: Application to the a1Dg
X3S
g (0, 0) O2 band, J. Quant. Spectrosc. Radiat. Transfer, 69, 389 – 401, doi:10.1016/S0022-4073(00)00072-8. Ge´rard, J.-C., A. Saglam, G. Piccioni, P. Drossart, C. Cox, S. Erard, R. Hueso, and A. Sa´nchez-Lavega (2008), Distribution of the O2 infrared nightglow observed with VIRTIS on board Venus Express, Geophys. Res. Lett., 35, L02207, doi:10.1029/2007GL032021. Ge´rard, J.-C., A. Saglam, G. Piccioni, P. Drossart, F. Montmessin, and J.-L. Bertaux (2009), Atomic oxygen distribution in the Venus mesosphere from observations of O2 infrared airglow by VIRTIS-Venus Express, Icarus, 199, 264 – 272, doi:10.1016/j.pss.2006.04.023. Gierasch, P. J., et al. (1997), The general circulation of the Venus atmosphere: An assesment, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 459 – 500, Univ. of Ariz. Press, Tucson. Grassi, D., P. Drossart, G. Piccioni, N. I. Ignatiev, L. V. Zasova, A. Adriani, M. L. Moriconi, P. G. J. Irwin, A. Negra˜o, and A. Migliorini (2008), Retrieval of air temperature profiles in the Venusian mesosphere from VIRTIS-M data: Description and validation of algorithms, J. Geophys. Res., 113, E00B09, doi:10.1029/2008JE003075. Haslett, J. C., and F. C. Fehsenfeld (1969), Ratio of the O2(a1Dg
X3S
g) (0, 0), (0, 1) transitions, J. Geophys. Res., 74, 1878 – 1879, doi:10.1029/ JA074i007p01878. Hueso, R., A. Sa´nchez-Lavega, G. Piccioni, P. Drossart, J. C. Ge´rard, I. Khatuntsev, L. Zasova, and A. Migliorini (2008), Morphology and dynamics of Venus oxygen airglow from Venus Express/Visible and Infrared Thermal Imaging Spectrometer observations, J. Geophys. Res., 113, E00B02, doi:10.1029/2008JE003081. Huestis, D. L. (2002), Current laboratory experiments for planetary aeronomy, in Atmospheres in the Solar System: Comparative Aeronomy, Geophys. Monogr. Ser., vol. 130, edited by M. Mendillo, A. Nagy, and J. H. Waite, pp. 245 – 258, AGU, Washington, D. C. Ignatiev, N. I., D. Grassi, and L. V. Zasova (2005), Planetary Fourier Spectrometer data analysis: Fast radiative transfer models, Planet. Space Sci., 53, 1035 – 1042, doi:10.1016/j.pss.2004.12.009. Kamp, L. W., F. W. Taylor, and S. B. Calcutt (1988), Structure of Venus’s atmosphere from modelling of night-side infrared spectra, Nature, 336, 360 – 362, doi:10.1038/336360a0. Kassi, S., D. Romanini, and A. Campargue (2005), Very high sensitivity CW-cavity ring down spectroscopy: Application to the a1Dg(0)
X 3 S
g (1) O2 band near 1.58 mm, Chem. Phys. Lett., 409, 281 – 287, doi:10.1016/j.cplett.2005.05.033. Lafferty, W. J., A. M. Solodov, C. L. Lugez, and G. T. Fraser (1998), Rotational line strengths and self-pressure-broadening coefficients for the 1.27-mm, a1Dg
X3S
g , v = 0 – 0 band of O2, Appl. Opt., 37, 2264 – 2270, doi:10.1364/AO.37.002264.

E00B38

Lellouch, E., T. Clancy, D. Crisp, A. J. Kliore, D. Titov, and S. W. Bougher (1997), Monitoring of mesospheric structure and dynamics, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 295 – 324, Univ. of Ariz. Press, Tucson. Limaye, S. S. (2007), Venus atmospheric circulation: Known and unknown, J. Geophys. Res., 112, E04S09, doi:10.1029/2006JE002814. Lowe, R. P. (1969), Interferometric spectra of the Earth’s airglow (1.2 – 1.6 mm), Philos. Trans. R. Soc. London, Ser. A., 264, 163 – 169. Markiewicz, W. J., D. V. Titov, S. S. Limaye, H. U. Keller, N. Ignatiev, R. Jaumann, N. Thomas, H. Michalik, R. Moissl, and P. Russo (2007), Morphology and dynamics of the upper cloud layer of Venus, Nature, 450, 633 – 636, doi:10.1038/nature06320. Meadows, V. S., and D. Crisp (1996), Ground-based near-infrared observations of the Venus nightside: The thermal structure and water abundance near the surface, J. Geophys. Res., 101, 4595 – 4622, doi:10.1029/ 95JE03567. Miller, H. C., J. E. McCord, J. Choy, and G. D. Hager (2001), Measurement of the radiative lifetime of O2(a1Dg) using cavity ring down spectroscopy, J. Quant. Spectrosc. Radiat. Transfer, 69, 305 – 325. Mills, F. P., and M. Allen (2007), A review of selected issues concerning the chemistry in Venus’ middle atmosphere, Planet. Space Sci., 55, 1729 – 1740, doi:10.1016/j.pss.2007.01.012. Newman, S. M., I. C. Lane, A. J. Orr-Ewing, D. A. Newnham, and J. Ballard (1999), Integrated absorption intensity and Einstein coefficients for the O2 a1Dg
X3S
g (0, 0) transition: A comparison of cavity ringdown and high resolution Fourier transform spectroscopy with a long-path absorption cell, J. Chem. Phys., 110, 10,749 – 10,757, doi:10.1063/1.479018. Noxon, J. F., and A. Vallance Jones (1962), Observation of the (0, 0) band of the (a1Dg
X3S
g ) system of oxygen in the day and twilight airglow, Nature, 196, 157 – 158, doi:10.1038/196157a0. Ohtsuki, S., N. Iwagami, H. Sagawa, H. Kasaba, Y. Ueno, and M. Imamura (2005), Ground-based observation of the Venus 1.27-mm O2 airglow, Adv. Space Res., 36, 2038 – 2042, doi:10.1016/j.asr.2005.05.078. Ohtsuki, S., et al. (2008), Imaging spectroscopy of the Venus 1.27-mm O2 airglow with ground-based telescopes, Adv. Space Res., 41, 1375 – 1380, doi:10.1016/j.asr.2007.10.014. Piccioni, G., et al. (2008), First detection of hydroxyl in the atmosphere of Venus, Astron. Astrophys., 483, L29 – L33, doi:10.1051/0004-6361: 200809761. Piccioni, G., et al. (2009), VIRTIS: The Visible and Infrared Thermal Imaging Spectrometer, Eur. Space Agency Spec. Publ., ESA SP-1295, in press. Pick, D. R., E. J. Llewellyn, and A. Vallance Jones (1971), Twilight airglow measurements of the OH and O2 bands by means of balloon-borne instruments, Can. J. Phys., 49, 897 – 905. Pollack, J. B., et al. (1993), Near-infrared light from Venus’ nightside: A spectroscopic analysis, Icarus, 103, 1 – 42, doi:10.1006/icar.1993.1055. Rothman, L. S., et al. (2005), The HITRAN 2004 molecular spectroscopic database, J. Quant. Spectrosc. Radiat. Transfer, 96, 139 – 204, doi:10.1016/j.jqsrt.2004.10.008. Sa´nchez-Lavega, A., et al. (2008), Variable winds on Venus mapped in three dimensions, Geophys. Res. Lett., 35, L13204, doi:10.1029/ 2008GL033817. Sander, S. P., et al. (2006), Chemical kinetics and photochemical data for use in atmospheric studies evaluation number 15, JPL Publ., 06-2, 523 pp. Sharma, R. D., H. B. Harlow, and J. P. Riehl (1988), Determination of atomic oxygen density and temperature of the thermosphere by remote sensing, Planet. Space Sci., 36, 531 – 538, doi:10.1016/0032-0633(88)90022-0. Slanger, T. G., and R. A. Copeland (2003), Energetic oxygen in the upper atmosphere and the laboratory, Chem. Rev., 103, 4731 – 4765, doi:10.1021/cr0205311. Spalek, O., J. Kodymova´, P. Stopka, and I. Micek (1999), Experimental verification of the Einstein A-coefficient used for evaluation of O2(a1Dg) concentration in the chemical oxygen-iodine laser, J. Phys. B At. Mol. Opt. Phys., 32, 1885 – 1892, doi:10.1088/0953-4075/32/8/309. Stamnes, K., S.-C. Tsay, W. Wiscombe, and K. Jayaweera (1988), Numerically stable algorithm for discrete-ordinate-method radiative transfer in multiple scattering end emitting layered media, Appl. Opt., 27, 2502 – 2509, doi:10.1364/AO.27.002502. Thomas, R. G., and B. A. Thrush (1977), Energy transfer in the quenching of singlet molecular oxygen: I. Kinetics of quenching of singlet oxygen, Proc. R. Soc. London, Ser. A, 356, 287 – 294, doi:10.1098/ rspa.1977.0133. Titov, D. V., et al. (2006), Venus Express science planning, Planet. Space Sci., 54, 1279 – 1297, doi:10.1016/j.pss.2006.04.017. Tsang, C. C. C., P. G. J. Irwin, F. W. Taylor, and C. F. Wilson (2008), A correlated-k model of radiative transfer in the near-infrared windows of Venus, J. Quant. Spectrosc. Radiat. Transfer, 109, 1118 – 1135.

17 of 18

E00B38

PICCIONI ET AL.: VENUS O2 NIGHTGLOW FROM VIRTIS

Vallance Jones, A., and A. W. Harrison (1958), 1Dg
X3S
g O2 infrared emission band in the twilight airglow spectrum, J. Atmos. Terr. Phys., 13, 45 – 60, doi:10.1016/0021-9169(58)90024-2. Winick, J. R., R. H. Picard, R. D. Sharma, and R. M. Nadile (1985), Oxygen singlet Delta 1.58-mm (0 – 1) limb radiance in the upper stratosphere and lower mesosphere, J. Geophys. Res., 90, 9804 – 9814, doi:10.1029/JA090iA10p09804. Zasova, L., et al. (2007), Structure of the Venus atmosphere, Planet. Space Sci., 55, 1712 – 1728, doi:10.1016/j.pss.2007.01.011.

E00B38

P. Drossart, LESIA, Observatoire de Paris, UPMC, Universite´ ParisDiderot, CNRS, 5 place Jules Janssen, F-92195 Meudon, France. F. P. Mills and A. Garcı´a Mun˜oz, Research School of Physics and Engineering, Australian National University, Canberra, ACT 0200, Australia. A. Shakun and L. Zasova, Space Research Institute of Russian Academy of Sciences, Profsojuznaja 84/32, 117997 Moscow, Russia.

























A. Cardesin-Moinelo, A. Migliorini, and G. Piccioni, INAF, Istituto di Astrofisica Spaziale e Fisica Cosmica, via del Fosso del Cavaliere 100, I-00133 Rome, Italy. ([email protected])

18 of 18

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, E00B03, doi:10.1029/2008JE003078, 2008

for

Full Article

Limb observations of the ultraviolet nitric oxide nightglow with SPICAV on board Venus Express J.-C. Ge´rard,1 C. Cox,1 A. Saglam,1 J.-L. Bertaux,2 E. Villard,2 and C. Nehme´2 Received 17 January 2008; revised 15 April 2008; accepted 30 May 2008; published 20 August 2008.

[1] Limb observations of the spectrum of nightglow emission in the d (190–240 nm) and

g (225–270 nm) bands of nitric oxide have been made with the Spectroscopy for Investigation of Characteristics of the Atmosphere of Venus (SPICAV) ultraviolet spectrometer on board Venus Express. These emissions arise from radiative recombination between O(3P) and N(4S) atoms that are produced on the dayside and recombine to form excited NO molecules on the nightside. No other emission feature has been identified. The mean altitude of the emission layer is located at 113 km, but it varies between 95 and 132 km. The mean brightness of the total NO emission at the limb is 32 kR, but it is highly variable with limb intensities as large as 440 kR observed at low latitude and values below 5 kR seen at northern midlatitudes. No systematic dependence of the brightness with latitude is observed, but the mean altitude of the emission maximum statistically drops with increasing latitude between 6° and 72°N. Typical observed limb profiles are compared with simulations based on a one-dimensional chemical-diffusive atmospheric model. From model fits to observed profiles, we find that the downward flux of N atoms at 130 km typically varies between 1
108 to 4
109 atoms cm2 s1. Comparisons of observed airglow topside scale heights with modeled profiles smoothed by the instrumental field of view indicate that the observations are compatible with a downward flow of O and N atoms by molecular and turbulent transport above the peak of emission. The K coefficient deduced from comparisons to limb profiles is less than that determined from the observations made with the Pioneer Venus UV spectrometer at low latitude during periods of high solar activity. Citation: Ge´rard, J.-C., C. Cox, A. Saglam, J.-L. Bertaux, E. Villard, and C. Nehme´ (2008), Limb observations of the ultraviolet nitric oxide nightglow with SPICAV on board Venus Express, J. Geophys. Res., 113, E00B03, doi:10.1029/2008JE003078.

or

1. Introduction [2] The presence of the delta and gamma bands of nitric oxide in the Venus nightglow was detected and identified by Feldman et al. [1979] using the ultraviolet spectrograph on board the International Ultraviolet Explorer (IUE). It was also observed by Stewart and Barth [1979] in spectra obtained with the ultraviolet spectrometer on board the Pioneer Venus Orbiter (PV-OUVS). The emission process is radiative recombination through reverse predissociation of nitrogen N(4S) and oxygen O(3P) atoms, yielding excited NO* molecules which emit the ultraviolet d and g bands:
 N þ O ! NO C2 P

ð1Þ

giving rise to

 NO C2 P ! NO X2 P þ d-bands;

1 Laboratoire de Physique Atmosphe´rique et Plane´taire, Universite´ de Lie`ge, Lie`ge, Belgium. 2 Service d’Ae´ronomie du CNRS, Universite´ Versailles Saint Quentin, Verrie`res-le-Buisson, France.

Copyright 2008 by the American Geophysical Union. 0148-0227/08/2008JE003078$09.00



 NO C2 P ! NO A2 S; v0 ¼ 0 þ 1:22 mm;

followed by

 NO A2 S; v0 ¼ 0 ! NO X2 P þ g-bands

[3] As a consequence, the total emission rate of the NO bands is proportional to the rate of recombination of O and N atoms and thus depends on the nitrogen and oxygen densities. A detailed study of reaction (1) [Dalgarno et al., 1992] suggests a rate coefficient for the N + O recombination equal to 1.92
1017
(300/T)1/2
(1– 0.57/T1/2) cm3 s1. The N and O atoms are produced by dissociation of N2, CO2 and CO on the dayside. [4] Spin-scan images in the wavelength of the d (0,1) band at 198 nm obtained from Pioneer Venus near apoapsis by Stewart et al. [1980] indicated important day-to-day variations. Patches of enhanced intensity appeared to vary in intensity and location without correlation with solar activity. The location of the brightest spots ranged from 2130 to 0300 LT and 39°S to 60°N. This variability was attributed to temporal charges in the Venus thermospheric circulation on

E00B03

1 of 10

E00B03

GE´RARD ET AL.: VENUS NITRIC OXIDE NIGHTGLOW FROM VENUS EXPRESS

timescales of one Earth day or less. It appears to be caused by instabilities in the large-scale circulation, possibly as a result of wind shears near the terminator or time-varying wave drag from gravity waves. However, no quantitative modeling for this variability has been made so far. Stewart et al. [1980] also showed that, statistically, the emission is concentrated in a bright spot located near 0200 local solar time, just south of the equator. Following revision of the calibration of the instrument [Bougher et al., 1990], the emission rate of this region was estimated to 1.9 kR. From these observations, a general picture emerged where production of O and N atoms by solar EUV and possible solar particle impact occurs on the dayside, followed by global transport to the nightside, downward transport and radiative recombination. It was found that parameters such as the downward flux of O and N atoms and the strength of turbulent transport in a one-dimensional model could be adjusted to reproduce the average NO nightglow intensity. Since the attitude of the spacecraft was not known with sufficient accuracy, the determination of the altitude of the emission peak was indirect and based on the absorption of the NO emission by CO2. Ge´rard et al. [1981] observed that it lies close to 115 km. They showed that downward transport by molecular diffusion alone is not sufficient to reproduce the observations. They parameterized turbulent mixing using an eddy diffusion coefficient K 8
1012/n1/2 cm2 s1, where n is the total number density. This value was close to that deduced by von Zahn et al. [1979] from the analysis of the neutral composition measured near the morning terminator, which sets the altitude of the homopause close to 135 km. [5] Globally, the general picture of production of O and N atoms followed by transport to the nightside by the subsolar to antisolar circulation, downward turbulent mixing and radiative recombination appeared quantitatively consistent with the PV-OUVS observations. Further quantitative validation of this scheme was obtained by three-dimensional simulations using the Venus Thermospheric General Circulation Model (VTGCM) [Bougher et al., 1990]. Onedimensional chemical-transport calculations of odd nitrogen species on the dayside indicated that the daytime N average production is 1.3
1010 atoms cm2 s1 for solar maximum activity (F10.7 = 180 – 200) [Ge´rard et al., 1981]. Parameter adjustments in the VTGCM lead to an average production of about 9
109 atoms cm2 s1, approximately 30% of which are needed for transport to the nightside to reproduce the observed average nightglow brightness. The statistical location of the bright spot was also reasonably well reproduced by the 3-D model and implied zonal winds of about 50– 75 m s1 in the 115– 150 km region. However, it was found that model parameters could be adjusted to reproduce either the ultraviolet nightglow or the observed N density on the dayside but not both. Predictions for solar minimum conditions indicated a reduction by a factor of 3 from solar maximum. This is a direct consequence of the lower production of N and O atoms on the dayside and of the predicted reduced strength of the thermospheric circulation. The observed shift toward dawn of the statistical location of the airglow maximum, presumably as a consequence of the residual atmospheric superotation in the upper mesosphere, was reproduced by the VTGCM.

E00B03

[6] The NO d and g bands were also detected on Mars by Bertaux et al. [2005] using the SPICAM (Spectroscopy for Investigation of Characteristics of the Atmosphere of Mars) spectrometer on board the Mars Express spacecraft. The distribution and variability of the NO Mars airglow were recently described by Cox et al. [2008]. They found that the peak of the airglow layer is located between 55 and 92 km with lower values observed at higher latitude and larger values at low latitudes and midlatitudes. The limb brightness was also found to be very variable, ranging from less than 0.2 to 10.5 kR with the large values occurring at southern latitudes higher than 50°. [7] The NO airglow is thus a tracer of the thermospheric circulation and its distribution and intensity provide constraints on transport models in a region of the Venusian atmosphere difficult to probe with other in situ or remote sensing methods [Bougher et al., 2006]. In this study, we describe new observations of the NO nightglow collected in the northern hemisphere with the Spectroscopy for Investigation of Characteristics of the Atmosphere of Venus (SPICAV) spectrometer on board the Venus Express spacecraft [Svedhem et al., 2007]. In particular, we investigate the distribution of the altitude and brightness of the airglow and its spatial variability. We examine possible relationships between the emission intensity and its brightness and a possible latitudinal dependence. We also compare limb profiles with results of a one-dimensional chemical-diffusive model solving the odd nitrogen and oxygen continuity equations and draw conclusions concerning the importance of transport by other processes.

2. SPICAV Observations [ 8 ] The observations described in this study were obtained with SPICAV, the ultraviolet and infrared spectrometer on board Venus Express. The instrument and its performances were described by Bertaux et al. [2007a]. The ultraviolet component covers the range extending from 118 nm to 320 nm and includes the NO d (C 2P - X 2P) and g (A 2S - X 2P) emission bands. For reasons of telemetry limitations but also because of the time needed to read all the lines of the CCD, only 5 adjacent zones of the CCD detector are usually read out. Each zone (or spatial bin) is made of 1, 2, 4, 8, 16 or 32 lines of the CCD following a preselected mode. Therefore each spatial bin covers a different region of the atmosphere separated by an angle ranging from 1.4 to 22.4 arcmin, depending on the spatial binning. In the case of the NO airglow limb scans reported here, each spatial bin includes 2, 4, 8, 16 or 32 adjacent pixel lines. These lines are seen through the large (500 mm) or the small (50 mm) slits, which provides a spectral resolution of 12 or 1.5 nm, respectively. The SPICAV CCD is read every second and therefore five spatial bins corresponding to five adjacent portions of the SPICAV field of view are recorded each second. In each individual spectrum non-uniform dark current and offset values are subtracted. Finally, the intensities are calibrated in kiloRayleighs (kR) using well-known hot stars spectra observed by SPICAV during the mission. [9] The VEX spacecraft moves along a quasi-polar eccentric orbit with a 24-hour period. The pericenter and apocenter are located at 250 km and 66,000 km,

2 of 10

E00B03

GE´RARD ET AL.: VENUS NITRIC OXIDE NIGHTGLOW FROM VENUS EXPRESS

Figure 1. Variation of the altitude of the tangent point of the center of the field of view for SPICAV spatial bins 1 and 5 during tangential limb nightside observations of orbit 322. The altitudes of the other three spatial bins are within the gray shaded area. respectively, and the orbit precesses in local time at the rate of 1.6°/day. The precession of the orbital plane leads to a wide variety of configurations on the nightside as well as on the dayside. Several observation modes (‘‘science cases’’) may be selected including nadir observations, star pointing for stellar occultations by Venus’ atmosphere, fixed point tracking and limb observations [Titov et al., 2006]. A particular mode is well suited for high spatial resolution airglow observations. In this grazing (tangential) limb mode, the line of sight is at some angle to the orbital plane and moves in such a way to maximize the time spent in the atmosphere, so that the time of observation of the limb is larger than in the usual limb mode in the orbital plane [Bertaux et al., 2007a]. The signal recorded by the CCD detector is integrated over 640 ms periods typically during about 10 min. In this way, the line of sight scans a range of altitudes, generally between 70 km and 400 km and each second a full UV spectrum is obtained. For most of the observations reported here, the spacecraft was oriented in such a way that the SPICAV line of sight scanned the dark limb several times during the ascending portion of the VEX orbit. Since the line of sight crossed the same altitudes once during the ingress and once during the egress segments owing to the geometry of the grazing limb observation type, SPICAV supplied several sets of two (one for ingress, one for egress) altitude scans of five (for the 5 spatial bins) altitude profiles at each orbit. Consequently, as many as 6
5 limb profiles of the NO airglow were obtained on a same orbit between 6.1° and 71.8°N. At this date, a total of 17 orbits and 201 limb scans with suitable night airglow observations have been collected. As an example, Figure 1 shows the altitude variation of the altitude of the tangent point of the central point of the field of view for SPICAV spatial bins 1 and 5 during orbit 322. Other spatial bins scan the region indicated by the dotted gray area. Once all profiles are retrieved, the finite aperture of the field of view projected onto the limb is taken into account. For a given airglow layer, the apparent altitude of the emission peak and

E00B03

the maximum brightness depend on the value of the field of view projected on the limb. The angular field of view of one pixel is equal to 0.7 arcmin and the bin parameter varies between 2 and 32. The distance between the spacecraft and the observed tangent point lies within the 3000 to 12000 km range and the angle between the SPICAV slit and the limb is between 15° and 72°. For each observation, this set of parameters is used to determine the field of view projected on the limb. It varies from 3 to 27 km, but generally remains between 12 and 22 km, with a mean value of 14.9 km. We have investigated the effects of the smoothing by the field of view of an emission layer having a vertical Chapman profile integrated along the line of sight. Each Chapman profile is constrained to show the same topside scale height as the observation. It is found that the peak intensity observed by SPICAV decreases by a factor between 1 and 2 and the altitude of the peak is slightly above the unconvolved value. Data points in Figures 6 to 10 have been corrected for this smoothing effect by setting the peak intensity and altitude to the values they would have if the limb profiles had been observed with a negligibly small field of view. [10] For this study, we have integrated each spectrum over the entire d and g bands emission between 190 nm and 260 nm, including both the d and the g bands. As discussed in section 1, the total emission rate of the two molecular systems is equal to the N + O recombination rate. Figure 2 shows a sum of all (771) individual nightglow spectra obtained between 90 and 120 km on orbit 516. Comparison with the laboratory spectrum of the N(4S) + O(3P) recombination [Groth et al., 1971] confirms that this region of the Venus nightglow ultraviolet spectrum contains no other emission than those arising from the C 2P and A 2S states of NO. Integration of each individual spectrum over this spectral domain provides the total NO emission brightness in kR for each scanned altitude. Figure 3 shows limb profiles of the NO nightglow measured on ingress and egress segments of randomly selected orbit 271. In this plot, since the limb profiles in the five spatial bins are very similar as it is generally the case, the data points for spatial elements 1 and 2 have been grouped together. The data points were first binned in altitude (in 2 km steps), and these binned profiles were then co-added, separating egress and ingress data. Depending on altitude, each bin contains between 10 and 20 spectra, so that the error bar are quite small. The latitudes of the peak emission at ingress and egress are 21.7°N and 29.7°N and the local times are 2348 and 2330 LT, respectively. The maximum intensities at the peak are 46 and 35 kR, indicating the presence of a substantial gradient over a distance of only 8.0 degrees of latitude. The limb intensities have been numerically inverted using an Abel algorithm assuming a horizontally homogeneous distribution to calculate the corresponding vertical emission rates of 0.8 and 0.9 kR. These values may be compared with the 1.9 kR statistically observed by PVOUVS brightness in the bright spot. As just mentioned the limb profiles observed in the spatial bins are generally close to each other. However, strong horizontal gradients are occasionally present and noticeable differences between the profiles in the spatial bins may then be observed. An example is given in Figure 4 which shows four ingress smoothed profiles obtained during VEX orbit 322. The line of sight for spatial bin 5 did not intersect the altitude of the

3 of 10

E00B03

GE´RARD ET AL.: VENUS NITRIC OXIDE NIGHTGLOW FROM VENUS EXPRESS

E00B03

Figure 2. Spectrum of the nitric oxide ultraviolet nightglow obtained by summing all 771 individual spectra collected with the SPICAV instrument between 90 and 120 km on VEX orbit 516.

maximum of the airglow layer and is thus not represented in Figure 4. Moving from spatial bin 1 to 4, the limb intensity first increases then drops between adjacent bins angularly separated by only 5.6 arcmin. This separation maps into 1.8 (between bin 1 and 2), 1.4 (between bin 2 and 3) and 3.5 (between bin 3 and 4) degrees of increasing planetary latitude at the emission peak, which corresponds to 197, 158, and 374 km, respectively, of horizontal distance of the tangent points along the line of sight. Similarly, the altitude of the airglow peak varies from a bin to the next one. The differences in shape, peak altitude and brightness observed in the 4 spatial bins indicate that, in this case, the NO airglow distribution is significantly non uniform even over a restricted range of latitudes. It is important to note, however, that latitudinal gradients are usually much smaller, as was shown for example in Figure 3. [11] Each observed limb profile can be characterized by the altitude, brightness, latitude and local time of the maximum emission brightness. We first apply a filter to each limb profile to remove some of the high spatial frequency fluctuations through the data point following noise and background subtraction. From these fits we determine the peak altitude, the topside scale height of the emission as well as the total NO peak brightness. In this data set, covering the period between 17 January and 19 September 2007, a total of 201 good quality limb profiles were obtained. Their local time covers the range from 2000 to 0300 LT. The local time and latitudinal distribution of the observed limb profiles included in the database is represented in Figure 5. It shows that most observations were collected in two separate groups. Group 1 includes profiles between 6.1°N and 71.8°N observed in the post-midnight sector (2400 to 0300 LT), while group 2 which is more widespread corresponds to midlatitude to low-latitude observations between 2000 and 2200 LT. It should be noted that no observations were made in the

region of the NO bright spot observed with PV-OUVS south of the equator in the post-midnight period.

3. Correlations [12] We first examine the statistical distribution of the altitude of the peak of the airglow emission, corrected for the smoothing effect of the finite field of view. Figure 6 shows the distribution, where the data points have been distributed into 2.5 km altitude bins. The distribution is

Figure 3. Limb profile of the NO airglow measured on the ingress and egress of VEX orbit 271 on 17 January 2007. The latitudes of the peak emission are 21.7°N and 29.7°N and local times 2348 LT and 2330 LT for ingress and egress, respectively. The raw data are represented by open triangles and diamonds, while the smooth profile is obtained following Fourier filtering of high spatial frequencies. The error bars include the effect of the photon statistics and the detector thermal noise.

4 of 10

E00B03

GE´RARD ET AL.: VENUS NITRIC OXIDE NIGHTGLOW FROM VENUS EXPRESS

Figure 4. Smoothed limb profiles observed during ingress of the line of sight during VEX orbit 327 in four adjacent spatial bins on the CCD detector. The error bars include the effect of the photon statistics and the detector thermal noise.

E00B03

Figure 6. Distribution of the altitudes of the tangent points corresponding to the peak in the airglow limb profiles analyzed in this study. The altitudes have been corrected to account for the smoothing effect of the SPICAV field of view (see text).

quite symmetric about the mean value, with some values above 125 km and a few profiles with a peak below 100 km. The mean value is 112.6 km, quite close to the median value of 112.9 km, with a 1-s deviation of 5.8 km. It should be noted that we implicitly assume a horizontal homogeneous layer. If the emission is not present (or weaker) at the limb but present (or stronger) in the foreground or in the background behind the limb, then the derived altitude of the peak is necessarily lower than the actual altitude. Therefore, this effect cannot explain the high-altitude peaks which are shown in Figure 6. [13] The distribution of the peak brightness with a 6 kR bin size is shown in Figure 7. It is much less symmetric, showing an extended tail of high-intensity values. The mean limb total NO brightness is 32 kR and the median value is 19 kR, with a standard deviation of 60 kR. The large emission rate of 440 kR observed at the limb on VEX orbit 516 (19 September 2007) has not been included in this nor in the following plots for reasons of graphical representa-

tions. The large scatter suggests that the large-scale winds which carry the O and N atoms from the dayside are intrinsically variable. As the wind pattern and velocity change, the amount of atoms transported to a given location and the position of the subsidence region considerably fluctuate, generating variability of the bright airglow spot location and intensity. [14] We now search for possible correlations between the altitude of the airglow, its brightness and latitude. Figure 8 shows the altitude of the airglow peaks plotted versus its latitude between 11° and 85°N. A slow decrease is observed toward the polar region as indicated by the linear regression obtained by minimizing the chi-squares, which shows a drop of about 5 km from low to high latitude. However, the linear correlation coefficient is equal to only 0.26. Figure 9 shows the lack of correlation (r = 0.19) between the

Figure 5. Local time and latitudinal coverage of the limb profiles used in the statistical study of the NO airglow distribution.

Figure 7. Distribution of the brightness of the peaks in the airglow limb profiles analyzed in this study. The intensities have been corrected to account for the smoothing effect of the SPICAV field of view.

5 of 10

E00B03

GE´RARD ET AL.: VENUS NITRIC OXIDE NIGHTGLOW FROM VENUS EXPRESS

E00B03

lation carries atoms from the day to the night side, where the combination of the global circulation and local turbulent transport carries the atoms in the region where they recombine near 115 km. The one-dimensional model developed for this study brings into play diffusion equations as well as production and loss by chemical reactions. It is similar to that described by Cox et al. [2008] to model the NO nightglow on Mars. The continuity equation for a minor constituent i in the thermosphere may be written @ni @f @ ðni wÞ ¼  i þ Pi  L i  @t @z @z

ð2Þ

with the vertical diffusive flux fi of a minor constituent i given by     @ni ni @T Di K  ni fi ¼ ðDi þ K Þ þ þ @z T @z Hi H

Figure 8. Relationship between the altitude of the tangent point of the emission peak and its latitude. airglow peak intensity and the latitude of the tangent point. However, the largest intensities are observed at latitudes between 20° and 50°N and 65° and 70°N. It should be noted that the large values between 65° and 70° are located far from the statistical bright spot observed with PV-OUVS. Figure 10 illustrates how these brightness data points are distributed with the altitude of the airglow peak. Although a considerable scatter is observed (r = 016), it appears that the large intensity limb profiles are generally concurrent with a peak altitude located in the vicinity of 113 ± 4 km, close to the mean altitude of the airglow layer. By contrast, profiles showing weaker emission rate correspond to situations where the peak altitude is outside this range of central values.

4. One-Dimensional Modeling [15] The use of a one-dimensional model makes it possible to evaluate the downward flux of nitrogen and oxygen atoms that recombine to produce the NO airglow emission. As described in the introduction, the current global view is that the subsolar to antisolar (SSAS) circu-

Figure 9. Relationship between the brightness of the emission peak and its latitude.

ð3Þ

with Di the molecular diffusion coefficient for constituent i, K is the vertical eddy diffusion coefficient, Hi is the local scale height of the ith constituent. H is the atmospheric scale height, T is the neutral gas temperature, ni is the number density of the ith species, z is the altitude, t is time, Pi is the production rate of species i, Li the loss rate and w is the vertical velocity positive upward. The last term in equation (2) corresponds to the vertical advective flux. We adopt the vertical variation of the eddy diffusion coefficient K similar to that used in earlier studies [von Zahn et al., 1979; Ge´rard et al., 1981], that is, A K ð zÞ ¼ pffiffiffiffiffiffiffiffiffi cm2 s1 nð zÞ

ð4Þ

where n is the total number density and A is a free parameter of the model which is independent of altitude. Therefore, K only depends on altitude and is identical for all constituents. The model solves equation (2) for O(3P), N(4S), NO and O2 (1D). The Pi and Li coefficients depend on the choice of the different reactions that come into play and which are listed in Table 1 with their corresponding reaction coefficients. The temperature dependence of the

Figure 10. Relationship between the tangent point altitude of the emission peak and its brightness.

6 of 10

GE´RARD ET AL.: VENUS NITRIC OXIDE NIGHTGLOW FROM VENUS EXPRESS

E00B03

E00B03

Table 1. Chemical Reactions and Rate Coefficients Rate Coefficienta (cm3 s1)

Reaction ðR1Þ ðR2Þ ðR3Þ ðR4Þ

N þ O ! NO þ hv N þ O þ CO2 ! NO þ CO2 N þ NO ! N2 þ O O þ O þ CO2 ! O2 þ CO2

1.9 2.8 2.5 2.8







1017
(300/T)1/2
(1 – 0.57/T1/2) 1032 (300/T)1/2 1010
(T/300)1/2
exp(600/T) 1032

Reference Dalgarno et al. [1992] Campbell and Thrush [1966]b Fox [1994] Campbell and Gray [1973]c

Except for three-body recombination reactions (R2) and (R4) in cm6 s1. The rate coefficient of (R2) is the value measured with M = N2 as a third body, multiplied by 2.5 for M = CO2 with an inverse square root dependence on temperature. c The rate coefficient is the value measured for M = N2 at 196 K multiplied by 2.5, a factor recommended as appropriate for Mars atmosphere models [Nair et al., 1994]. a

b

N + O + CO2 reaction was introduced by Stewart et al. [1980] and used in subsequent models such as Yung and DeMore [1982]. For reasons of continuity, we have adopted the same temperature dependence as earlier studies. Equation (2) is solved numerically between 130 and 80 km using the finite volume method on a constant grid. For O and N, we apply a flux boundary condition through the upper boundary, and assume photochemical equilibrium at the lower boundary and we use density null vectors for initial conditions (except at the boundaries). The model is then parameterized by A, w(z), Fo and FN. In fact, the choice of the densities at 80 km is relatively arbitrary and it does not influence the solution further than one scale height from the lower boundary if it is kept in a reasonable range. At the upper boundary, we leave the O(3P) and the N(4S) fluxes as free parameters to be determined by fitting the modeled limb profile to the observations. The limb profile of the nitric oxide emission is obtained by integrating the k[O][N] product along the line of sight, where k is the rate coefficient of the O + N recombination. We note that in all limb profiles most of the emission is located below 135 km, the altitude of the homopause determined from the composition measurements made by the Pioneer Venus large probe during its descent [von Zahn et al., 1979]. Onedimensional models usually do not include advection terms. Consequently, in this representation, vertical transport is solely the result of the molecular and eddy diffusions, and

advection is implicitly contained in the K coefficient, despite the lack of physical meaning. [16] We first present results of modeled limb profiles using CO2 density and temperature taken from the Venus Reference Atmosphere (VIRA) model in the code formulation given by Hedin et al. [1983]. Figure 11 shows a comparison between three observed limb profiles and those simulated using the one-dimensional model described above, following integration of the calculated volume emission rate along the line of sight. Absorption of the NO bands by CO2 is less than 1% in the altitude region considered here and is thus negligible. The simulated distributions match reasonably well the peaks of the observed limb profiles. As in earlier studies [Stewart et al., 1980; Ge´rard et al., 1981], we have assumed that the flux of N(4S) is 1% of the O flux, in agreement with the average O/N number density ratio measured in the dayside thermosphere with the neutral mass spectrometer on board Pioneer Venus. The observational data have been smoothed as described in section 2 to remove most of the scatter associated with instrument noise. In Figure 11, three comparisons are made with profiles that have been selected to cover a range of altitudes of the emission peak. The profile for orbit 320 obtained on 7 March 2007 at 32.3°N and 0036 LT shows a maximum at 112 km and is typical of a vertical distribution frequently observed. The peak intensity reaches 39 kR, a value larger than the median but close to

Figure 11. Examples of airglow limb profiles with different peak altitudes (circles). The solid lines show the corresponding model simulations that best fit the altitude and brightness of the airglow maximum. 7 of 10

E00B03

GE´RARD ET AL.: VENUS NITRIC OXIDE NIGHTGLOW FROM VENUS EXPRESS

Figure 12. Distribution of the apparent topside scale heights of the airglow limb profiles in the region located above the emission peak. For comparison, the dotted line indicates the airglow scale height for diffusive equilibrium of O and N, while the dashed line is the expected value for a downward flow of atoms in the homosphere (see text). the mean value, as was shown in Figure 7. The calculated limb profile represented by the solid line was obtained using the CO2 density and temperatures profiles from the Venus Reference Atmosphere (VIRA) model in the code formulation given by Hedin et al. [1983], for the appropriate latitude, local time and solar activity level. The calculated limb profile has been convolved by the value of the field of view projection on the planet calculated for the pixel binning and the spacecraft-Venus distance at the time of the observation. The parameters used for this simulation were FO = 2.1
1011 cm2 s1, FN = 2.1
109 cm2 s1, A = 7.4
1011. The altitude distribution at the peak and the topside scale height are well reproduced by the model, whereas the calculated bottomside scale heights are somewhat larger than the observed values. Bertaux et al. [2007b] showed that SPICAV stellar occultations by CO2 in the Venus thermosphere suggest that temperatures are as much as 40 K above the VIRA values in the region of the antisolar point. The effect of this different temperature distribution has been assessed by substituting the CO2 and temperature profiles observed by Bertaux et al. on orbit 104 (near midnight) to the VIRA distribution. The shape and absolute values of the predicted intensity remains almost unaffected when the warmer distributions deduced from the SPICAV occultation are used. [17] To document the question of the scale height, for each limb profile, we have determined the value of the apparent emission scale height H in the region extending between Zp + 5 km and Zp +15 km (except if the scale height varies rapidly in this altitude range), where Zp is the altitude of the emission peak. The resulting distribution is shown in Figure 12. The distribution of H values covers more than an order of magnitude, ranging from about 2 to more than 30 km. If the homopause is located above the airglow peak region, as suggested by the determination of the homopause altitude from the Pioneer large probe descent, the airglow layer is in the homosphere dominated

E00B03

by eddy mixing. In these conditions, the analytical solution to equations (2) and (3) indicates that the topside scale height of the constituents flowing downward toward the chemical sink region varies as 1/K. With the altitude dependence adopted in our model, this implies that both the O and N densities vary as the square root of the CO2 density. Consequently, since the airglow volume emission rate is proportional to the product of the O and N densities, the airglow scale height above the peak altitude is close to the value of the background atmosphere (mostly CO2), which, according to the VIRA model, is on the order of 3.1 km at 0000 LT near the equator. Instead, if the emission is located above the homopause and controlled by molecular diffusion in the absence of wind, the airglow scale height is that of the geometric mean 4.6 km of the O and N individual scale heights. Both values are indicated as vertical bars in Figure 12. The scale height determined from the analysis of the limb scans observed closer to the planet with PV-OUVS by Ge´rard et al. [1981] was 3.2 km, in close agreement with the value expected for turbulent downward transport of the atoms. It is clear that a large fraction of the observations (over 90%) exhibit an apparent scale height in excess of this value. Most of these larger values are observational effects due to the combination of (1) a finite field of view of the instrument whose value depends on the spatial bin size used for a given observation, (2) the distance between the spacecraft and (3) the slit orientation relative to the planetary limb. As mentioned before, the projected aperture of the field of view on the limb ranges between 3 and 27 km, which tends to smooth the observed limb profile and to increase the apparent topside scale height. [18] Figure 13 shows the O(3P) and N(4S) vertical distribution calculated from the best fit to the limb profile of orbit 320 in Figure 11. The maximum atomic oxygen density is predicted at 111 km where it reaches 4
1011 cm3. This value is in good agreement with the O density determined by Drossart et al. [2007] and Ge´rard et al. [2008] from the vertical distribution at the limb of the O2(1D) infrared airglow observed with the VIRTIS spectral imager. The atomic nitrogen density profile presents a peak of 2
108 cm3 at 122 km. This result may be compared with the N density profiles calculated previously by Stewart et al. [1980] or by Bougher et al. [1990] for solar maximum activity at 2400 LT with the Venus TGCM and VIRA models and reported in their Figure 5. The calculated maximum N density was about 7
108 cm3 at 112 km, in reasonably good agreement with the result of Figure 13, considering the different solar activity levels and local times. In Figure 11, two other simulations are compared with observed NO limb profiles. One (orbit 321, lat 53.8°N, LT = 0048) shows a peak at 105 km and is very well reproduced by the 1-D model both in magnitude and shape. The last one (orbit 324, lat 17.9°N, LT = 0118) presents a shallow peak at 119 km, while the best fit model produces maximum limb intensity at 112 km. The parameters of the fit for orbit 321 are fO = 6.6
1010 cm2 s1, A = 4
1012. For orbit 324, the parameters are fO = 8.2
1010 cm2 s1, and A = 0. In this latter case, downward transport solely by molecular diffusion alone is even slightly too strong to match the emission peak at 119 km. This result suggests that eddy diffusion may be locally quite weak or that upward

8 of 10

GE´RARD ET AL.: VENUS NITRIC OXIDE NIGHTGLOW FROM VENUS EXPRESS

E00B03

E00B03

Figure 13. Vertical distribution of the atomic oxygen and nitrogen number density obtained from the model fitting the limb profile of orbit 320 shown in Figure 11. vertical winds may act to raise the altitude of the airglow layer. The observed airglow topside scales heights are 13.3 km (orbit 320), 16 km (321) and 14.5 km (324). All three simulated profiles smoothed by the appropriate projected field of view exhibit an apparent scale height in agreement with the observed intensity drop above the emission peak. The modeled emission scale heights before smoothing are 3.6 km, 3.7 km and 4.6 km for orbits 320, 321 and 324, respectively. These values lie between the diffusive and the turbulent scale heights of 4.6 km and 3.1 km as discussed before.

5. Conclusions [19] The NO nightglow emission is a good tracer of atmosphere dynamics in the Venus thermosphere. The nightside lower thermosphere of the planet remains relatively unexplored, especially the minor constituents densities and their dynamics. SPICAV observations of the NO nightglow provide a set of limb profiles data that have been used to analyze the NO vertical distribution in detail. From the comparison with a one-dimensional chemicaldiffusive model, we have determined values for the eddy diffusion coefficient. The value of the A coefficient in formula (4) is 7.4
1011, 4
1012 and 0 for the cases modeled in this study. The first two values are 0.09 to 0.5 smaller than the earlier determination based on NO airglow observations from Pioneer Venus. We note that the PVOUVS limb observations were made at low latitudes near maximum solar activity. The profiles described here were obtained in the northern hemisphere, generally away from the bright spot reported by Stewart et al. [1980]. The nitrogen downward fluxes at 130 km range between 1
108 and 4
109 cm2 s1 with typical values of 2
109 cm2 s1. The deduced O fluxes are less than the average column production rate of O atoms on the Venus dayside which was estimated at 8
1012 cm2 s1 by Leu and Yung [1987], in agreement with the discussion by Ge´rard et al. [2008]. Similarly, the daytime N average column production was found to be 1.3
1010 atoms cm2 s1 by Ge´rard et al. [1988] at solar maximum and

about 9
109 atoms cm2 s1 in the VTGCM photochemistry [Bougher et al., 1990]. The N vertical fluxes on the nightside derived from these observations thus also appear compatible with the availability of N atoms on the dayside. We also find that the topside scale height of the NO airglow layer is in agreement with that expected for constituents in diffusive equilibrium or diffusing downward through the homosphere. [20] One of the main results in the Venus Express observations is the large variability in the brightness of the NO nightglow emission. This feature was already apparent in the individual NO brightness maps obtained from PV-OUVS presented by Stewart et al. [1980] and in Figure 9 by Bougher et al. [1990]. It is not related to concurrent solar activity which remained very low throughout the period of these observations. It implies timescales as short as an Earth day or less. A similar variability is present in the O2 (1D) emission arising from three-body recombination of O atoms at 1.27 mm that have been observed with VIRTIS-M, also on board Venus Express. Both limb and nadir observations [Drossart et al., 2007; Ge´rard et al., 2008; Hueso et al., 2008; G. Piccioni et al., Molecular oxygen nightglow from the VIRTIS near IR observations in the Venus upper mesosphere, submitted to Journal of Geophysical Research, 2008] indicate that the brightness of the O2 emission and its peak altitude exhibit substantial variations sometimes on timescales on the order of an hour or so. This variability may be caused by time variations in gravity waves breaking which potentially decelerate the SSAS flow. [21] The SPICAV limb observations confirm and complete this picture and they also show the presence of strong horizontal gradients. Such gradients may be interpreted as signatures of localized regions of enhanced or decreased downward fluxes of atoms producing the patchy structure that had been observed on individual maps. In the southern hemisphere, the larger distance of the spacecraft from the planet precludes high-resolution limb observations. Therefore, the statistical bright spot near midnight slightly south of the equator was generally not captured in this tangential limb observing mode. It is thus not possible at this stage to

9 of 10

E00B03

GE´RARD ET AL.: VENUS NITRIC OXIDE NIGHTGLOW FROM VENUS EXPRESS

confirm whether the emission maximum is at the same location during solar minimum periods. If so, the comparison with the morphology of the O2 (1D) emission due to three-body recombination of O atoms which has a peak near midnight near the equator would suggest that the zonal winds which shift the NO airglow at 113 km toward dawn have a different regime near 96 km. In any case, the features described in this study should be additional constraints to future three-dimensional models of the Venus nightside. [22] Acknowledgments. The authors thank the SPICAV and the Venus Express teams for the excellent quality of their work. J.-C. Ge´rard is supported by the Belgian Fund for Scientific Research (FNRS). This research was funded by the PRODEX program of the European Space Agency (ESA) managed with the help of the Belgian Space Policy Office. The construction of the SPICAV instrument was funded by CNRS, CNES, and ESA/PRODEX.

References Bertaux, J.-L., et al. (2005), Nightglow in the upper atmosphere of Mars and implications for atmospheric transport, Science, 307, 566 – 569, doi:10.1126/science.1106957. Bertaux, J.-L., et al. (2007a), SPICAVon Venus Express: Three spectrometers to study the global structure and composition of the Venus atmosphere, Planet. Space Sci., 55, 1673 – 1700, doi:10.1016/j.pss.2007.01.016. Bertaux, J.-L., et al. (2007b), A warm layer in Venus’ cryosphere and high altitude measurements of HF, HCl, H2O and HDO, Nature, 450, 646 – 649, doi:10.1038/nature05974. Bougher, S. W., J. C. Ge´rard, A. I. F. Stewart, and C. G. Fesen (1990), The Venus nitric oxide night airglow: Model calculations based on the Venus Thermospheric General Circulation Model, J. Geophys. Res., 95, 6271 – 6284, doi:10.1029/JA095iA05p06271. Bougher, S. W., S. Rafkin, and P. Drossart (2006), Dynamics of the Venus upper atmosphere: Outstanding problems and new constraints expected from Venus Express, Planet. Space Sci., 54, 1371 – 1380, doi:10.1016/ j.pss.2006.04.023. Campbell, I. M., and C. N. Gray (1973), Rate constants for the O (3P) recombination and association with N (4S), Chem. Phys. Lett., 18, 607 – 609, doi:10.1016/0009-2614(73)80479-8. Campbell, I. M., and B. A. Thrush (1966), Behaviour of carbon dioxide and nitrous oxide in active nitrogen, Trans. Faraday Soc., 62, 3366 – 3374, doi:10.1039/tf9666203366. Cox, C., A. Saglam, J.-C. Ge´rard, J.-L. Bertaux, F. Gonza´lez-Galindo, F. Leblanc, and A. Reberac (2008), Distribution of the ultraviolet nitric oxide Martian night airglow: Observations from Mars Express and comparisons with a one-dimensional model, J. Geophys. Res., doi:10.1029/ 2007JE003037, in press. Dalgarno, A., J. F. Babb, and Y. Sun (1992), Radiative association in planetary atmospheres, Planet. Space Sci., 40, 243 – 246, doi:10.1016/ 0032-0633(92)90062-S. Drossart, P., et al. (2007), Infrared spectral imaging observations of Venus by VIRTIS reveal a dynamical upper atmosphere, Nature, 450, 641 – 645, doi:10.1038/nature06140.

E00B03

Feldman, P. D., H. W. Moos, J. T. Clarke, and A. L. Lane (1979), Identification of the UV nightglow from Venus, Nature, 279, 221 – 222, doi:10.1038/279221a0. Fox, J. L. (1994), Rate coefficient for the reaction N + NO, J. Geophys. Res., 99, 6273 – 6276, doi:10.1029/93JA03299. Ge´rard, J.-C., A. I. F. Stewart, and S. W. Bougher (1981), The altitude distribution of the Venus ultraviolet airglow and implications on vertical transport, Geophys. Res. Lett., 8, 633 – 636, doi:10.1029/GL008i006p00633. Ge´rard, J.-C., E. J. Deneye, and M. Lerho (1988), Sources and distribution of odd nitrogen in the Venus daytime thermosphere, Icarus, 75, 171 – 184, doi:10.1016/0019-1035(88)90135-2. Ge´rard, J.-C., A. Saglam, G. Piccioni, P. Drossart, C. Cox, S. Erard, R. Hueso, and A. Sa´nchez-Lavega (2008), The distribution of the O2 infrared nightglow observed with VIRTIS on board Venus Express, Geophys. Res. Lett., 35, L02207, doi:10.1029/2007GL032021. Groth, W., D. Kley, and U. Schurath (1971), Rate constant for the infrared emission of the NO (C2P ! A2S+) transition, J. Quant. Spectrosc. Radiat. Transfer, 11, 1475 – 1480, doi:10.1016/0022-4073(71)90109-9. Hedin, A. E., H. B. Niemann, W. T. Kasprzak, and A. Seiff (1983), Global empirical model of the Venus thermosphere, J. Geophys. Res., 88, 73 – 83. Hueso, R., A. Sa´nchez-Lavega, G. Piccioni, P. Drossart, J. C. Ge´rard, I. Khatuntsev, L. Zasova, and A. Migliorini (2008), Morphology and dynamics of Venus oxygen airglow, J. Geophys. Res., 113, E00B02, doi:10.1029/2008JE003081. Leu, M.-T., and Y. L. Yung (1987), Determination of O2 (a1Dg) and O2 (b1S+g) yields in the reaction O + ClO ! Cl + O2: Implications for photochemistry in the atmosphere of Venus, Geophys. Res. Lett., 14(9), 949 – 952, doi:10.1029/GL014i009p00949. Nair, H., M. Allen, A. D. Anbar, Y. L. Yung, and R. T. Clancy (1994), A photochemical model of the Martian atmosphere, Icarus, 111, 124 – 150, doi:10.1006/icar.1994.1137. Stewart, A. I., and C. A. Barth (1979), Ultraviolet night airglow of Venus, Science, 205, 59 – 62, doi:10.1126/science.205.4401.59. Stewart, A. I. F., J.-C. Ge´rard, D. W. Rusch, and S. W. Bougher (1980), Morphology of the Venus ultraviolet night airglow, J. Geophys. Res., 85, 7861 – 7870, doi:10.1029/JA085iA13p07861. Svedhem, H., et al. (2007), Venus Express-The first European mission to Venus, Planet. Space Sci., 55, 1636 – 1652, doi:10.1016/j.pss.2007.01.013. Titov, D. V., et al. (2006), Venus Express science planning, Planet. Space Sci., 54, 1279 – 1297, doi:10.1016/j.pss.2006.04.017. von Zahn, U., K. H. Fricke, H. J. Hoffmann, and K. Pelka (1979), Venus: Eddy coefficients in the thermosphere and in the inferred helium content of the lower atmosphere, Geophys. Res. Lett., 6, 337 – 340, doi:10.1029/ GL006i005p00337. Yung, Y. L., and W. B. DeMore (1982), Photochemistry of the stratosphere of Venus: Implications for atmospheric evolution, Icarus, 51, 199 – 247, doi:10.1016/0019-1035(82)90080-X. 

J.-L. Bertaux, C. Nehme´, and E. Villard, Universite´ de Versailles SaintQuentin-UVSQ, Service d’Ae´ronomie du CNRS, Boite Post 3, F-91371 Verrie`res-le-Buisson, France. C. Cox, J.-C. Ge´rard, and A. Saglam, Laboratoire de Physique Atmosphe´rique et Plane´taire, Universite´ de Lie`ge, Baˆtiment B5c, Alle´e de 6 Aouˆt, 17, B-4000 Lie`ge, Belgium. ([email protected])

10 of 10

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114, E00B44, doi:10.1029/2009JE003371, 2009

for

Full Article

Concurrent observations of the ultraviolet nitric oxide and infrared O2 nightglow emissions with Venus Express J.-C. Ge´rard,1 C. Cox,1 L. Soret,1 A. Saglam,1 G. Piccioni,2 J.-L. Bertaux,3,4 and P. Drossart5 Received 19 February 2009; revised 4 May 2009; accepted 18 June 2009; published 22 September 2009.

[1] Two prominent features of the Venus nightside airglow are the nitric oxide d and g

bands produced by radiative association of O and N atoms in the lower thermosphere and the O2 infrared emission generated by three-body recombination of oxygen atoms in the upper mesosphere. The O2 airglow has been observed from the ground, during the Cassini flyby, and with VIRTIS on board Venus Express. It now appears that the global structure of the two emissions shows some similarities, but the statistical location of the region of strongest emission is not coincident. The Spectroscopy for Investigation of Characteristics of the Atmosphere of Venus (SPICAV) ultraviolet spectrograph has collected a large number of spectra of the Venus nitric oxide nightside airglow. Visible and Infrared Thermal Imaging Spectrometer (VIRTIS) images have been obtained at the limb and in the nadir-viewing mode and have provided new information on the horizontal and vertical distribution of the emission. We present the first concurrent observations of the two emissions observed with Venus Express. We show that nadir observations generally indicate a low degree of correlation between the two emissions observed quasi-simultaneously at a common location. A statistical study of limb profiles indicates that the altitude and the brightness of the two airglow layers generally do not covary. We suggest that this lack of correlation is explained by the presence of strong horizontal winds in the mesosphere-thermosphere transition region. They carry the downflowing atoms over large distances in such a way that regions of enhanced NO emission generally do not coincide with zones of bright O2 airglow. Citation: Ge´rard, J.-C., C. Cox, L. Soret, A. Saglam, G. Piccioni, J.-L. Bertaux, and P. Drossart (2009), Concurrent observations of the ultraviolet nitric oxide and infrared O2 nightglow emissions with Venus Express, J. Geophys. Res., 114, E00B44, doi:10.1029/2009JE003371.

1. Introduction [2] Planetary airglow is a powerful way to remotely probe the characteristics of atmospheres from an orbiting or flying-by spacecraft that complements in situ measurements. In particular, the study of airglow morphology, time variations and brightness may provide key observations concerning atmospheric composition, temperature structure, transport processes and their response to the solar photon and particle inputs. In the case of Venus, the Venera and Pioneer Venus missions have shed new light on some of these aspects more than 25 years ago. The presence of the delta and gamma bands of nitric oxide in the Venus nightglow was detected and identified by Feldman et

al. [1979] using the ultraviolet spectrograph on board the International Ultraviolet Explorer (IUE). It was also observed by Stewart and Barth [1979] with the ultraviolet spectrometer on board the Pioneer Venus Orbiter (PVOUVS). The emission is produced by radiative recombination through inverse predissociation of nitrogen N(4S) and oxygen O(3P) atoms and dominates the middle ultraviolet nightglow spectrum. In this process, excited NO molecules radiate the ultraviolet d and g bands between 180 and 310 nm:
N þ O ! NO C 2 P

giving rise to

NO C 2 P ! NO X 2 P þ d-bands

1

Laboratoire de Physique Atmosphe´rique et Plane´taire, Universite´ de Lie`ge, Lie`ge, Belgium. 2 INAF, IASF, Rome, Italy. 3 LATMOS, Verrie`res-le-Buisson, France. 4 Institut Pierre Simon Laplace, Universite´ de Versailles Saint Quentin, Guyancourt, France. 5 LESIA, Observatoire de Paris, Meudon, France. Copyright 2009 by the American Geophysical Union. 0148-0227/09/2009JE003371$09.00



NO A 2 S; v0 ¼ 0 ! NO X 2 P þ g-bands

Emission from the C 2P (v = 0) ! A 2S (v = 0) transition at 1.224 mm was recently observed with the Visible and Infrared Thermal Imaging Spectrometer (VIRTIS) -M in the

E00B44

1 of 10

E00B44

GE´RARD ET AL.: NO AND O2 VENUS NIGHTGLOW OBSERVATIONS

Venus nightglow [Garcı´a Mun˜oz et al., 2009]. The N and O atoms are mainly produced by dissociation of N2 and CO2 on the dayside by extreme ultraviolet (EUV) photons and photoelectrons. [3] Stewart et al. [1980] obtained images of the Venus nightside in the d (0,1) band at 198 nm every 24 h with PVOUVS when Pioneer Venus was near apoapsis. They showed that the emission was highly variable in brightness and morphology over consecutive 24 h periods. The location of the brightest spots ranged from 2130 to 0300 LT and 39°S to 60°N [Bougher et al., 1990]. This variability appeared to be caused by instabilities in the large-scale circulation, possibly as a result of wind shears near the terminator or time-varying wave drag from gravity waves. Stewart et al. [1980] built up a statistical global map of the UV nightglow showing that the emission is concentrated in a bright spot located near 0200 local solar time, south of the equator. Bougher et al. [1990] estimated the emission rate of this enhanced emission to be 1.9 kilorayleighs (kR) whereas the average hemispheric nightside intensity is 0.48 kR. These observations confirmed the general picture where production of O and N atoms by solar EUV on the dayside is followed by global circulation to the nightside, downward transport and radiative recombination. The shift of the statistical bright spot toward dawn was interpreted as a signature of a residual superrotation into the lower thermosphere. A determination of the altitude of the emission peak by Ge´rard et al. [1981] concluded that the emission peak is located close to 115 km. Using a one dimensional model, they derived an eddy diffusion coefficient K  8  1012 n1/2 cm2 s1, where n is the total number density. The general picture of production of O and N atoms followed by transport to the nightside by the subsolar to antisolar circulation, downward turbulent mixing and radiative recombination appeared quantitatively consistent with the PV-OUVS observations. This concept was numerically validated by three-dimensional simulations using the Venus Thermospheric General Circulation Model (VTGCM) [Bougher et al., 1990]. The statistical location of the bright spot was reasonably well predicted by the threedimensional model and implied zonal winds of about 50– 75 m s1 in the 115 –150 km region. The observed shift toward dawn of the statistical location of the airglow maximum was reproduced by the VTGCM. [4] Limb observations of the spectrum of nightglow emission in the d and g bands of NO with the Spectroscopy for Investigation of Characteristics of the Atmosphere of Venus (SPICAV) ultraviolet spectrometer on board Venus Express [Titov et al., 2006; Svedhem et al., 2007] have been recently reported by Ge´rard et al. [2008a]. The mean altitude of the emission layer was found to be located at 113 km, with variations between 95 and 132 km. The mean limb brightness of the total NO emission at the limb was 32 kR, but it is highly variable with limb intensities as large as 440 kR and values below 5 kR at northern midlatitudes. It was found that the mean altitude of the emission peak statistically drops with increasing latitude between 6° and 72°N. From model fits to observed profiles, they determined that the downward flux of N atoms at 130 km typically varies between 1  108 to 4  109 atoms cm2 s1. The eddy diffusion coefficient K deduced from comparisons to the observed limb profiles was significantly less than that

E00B44

determined from the observations made with the Pioneer Venus UV spectrometer at low latitudes during periods of high solar activity. [5] The oxygen airglow (0– 0) emission at 1.27 mm is the most intense nonthermal component in the Venus atmosphere. It belongs to the a1Dg – X 3Sg Atmospheric Infrared system and corresponds to an electric dipole forbidden transition with a radiative lifetime of about 70 min [Miller et al., 2001]. It was first discovered in ground-based observations of Venus by Connes et al. [1979] and subsequently imaged with ground-based telescopes [Alien et al., 1992; Crisp et al., 1996; Lellouch et al., 1997; Ohtsuki et al., 2008; Bailey et al., 2008]. It was measured from space during the Cassini flyby with a local maximum brightness of about 4 MR [Drossart et al., 1993]. The oxygen IR nightglow appeared patchy, highly variable with asymmetries, often exhibiting multiple local maxima, with variations on time scales as short as 1 h. A much weaker O2 airglow was also detected in the Herzberg II visible wavelengths [Krasnopolsky, 1986; Bougher and Borucki, 1994]. The 1.27 mm emission is produced by recombination of oxygen atoms created by photodissociation of CO2 and CO at thermospheric altitudes on the dayside. As previously explained, O atoms are transported to the nightside by the global thermospheric circulation. Three-body recombination of O atoms in the upper nightside mesosphere (95 –110 km) leads to O2 formation in excited states, followed by airglow emissions as the molecules relax to the X 3Sg ground state: O þ O þ M ! O2 * þ M

followed by O2 * ! O2 þ hn

where O2* indicates one of the excited states of the O2 molecule and M is any neutral constituent. A fraction of the O2 molecules, estimated to be about 7%, is formed directly in the a1Dg metastable state. A substantial fraction of the upper excited states cascades into the 1D state, so that the net efficiency of the production of this state in the threebody recombination may be close to 100% [Crisp et al., 1996]. Below the emission peak, O2 1Dg molecules may be deactivated by collisions with CO2, causing nonradiative transitions to the O2 ground state. The altitude of the peak of the 1.27 mm emission is thus controlled by the competition between vertical transport, recombination and quenching of O atoms. The emission rate is related to the downward flux of oxygen atoms. [6] The VIRTIS-M O2 airglow limb observations have been presented by Drossart et al. [2007a], Ge´rard et al. [2008b, 2009], and Piccioni et al. [2009a]. Drossart et al. [2007a] determined that the O2 peak emission is located near 96 km, which is consistent with three-body recombination of oxygen atoms. Ge´rard et al. [2008b] found that limb profiles observed at northern midlatitudes exhibit large intensity variations over short time periods. Ge´rard et al. [2009] described further emission limb profiles extracted from the images. They determined the vertical distribution of O2 (1Dg) atoms using an Abel inversion of the radiance limb profiles. Assuming photochemical equilibrium for O2 (1D), they used these density profiles combined with the CO2

2 of 10

E00B44

GE´RARD ET AL.: NO AND O2 VENUS NIGHTGLOW OBSERVATIONS

vertical distribution to determine the atomic oxygen density. Piccioni et al. [2009a] analyzed limb measurements from 42 orbits. They found that the peak altitude of the O2 (1Dg) volume emission rate is typically located between 95 and 100 km, with a mean value of 97.4 ± 2.5 km. The vertical profile is broader near the equator, with a full width at half maximum of 11 km, a factor 2 larger than at middle latitudes. They reported that a secondary peak is frequently observed between 103 and 105 km. [7] VIRTIS nightside observations from Venus Express have complemented ground-based observations at much higher spatial resolution. In addition, limb observations from an orbit around Venus have given unprecedented access to the vertical distribution of the airglow layer and provided key constraints on the models. Drossart et al. [2007a] confirmed that the O2 nightglow exhibits a large spatial and temporal variability. Observations by VIRTIS in the nadir mode have been used to construct extensive maps of the Venus atmosphere in the O2 emission band [Ge´rard et al., 2008b; Piccioni et al., 2009a]. In nadir viewing geometry, the contamination of the O2 airglow by the thermal emission of the deeper atmosphere has to be subtracted to obtain clean O2 airglow images. The mean value, integrated over the nightside of the southern hemisphere, is typically about 0.8 MR, which is in agreement with the early groundbased observations giving a mean brightness of 1.2 MR for the night side. Hueso et al. [2008] found that the airglow is highly inhomogeneous with the regions of highest intensity generally located at low latitudes near the midnight meridian. They showed that zonal velocity derived from the motion of airglow features is dominated by an intense prograde jet from dawn to midnight extending up to 22 h in local time, with lower velocities and reversed sign from dusk. The brightest small-scale (100 km) features appeared correlated with locations of apparent convergence which may be a signature of compression and downwelling. Piccioni et al. [2009a] described the characteristics of the horizontal distribution of the airglow and showed that regions of high O2 airglow intensity are associated with downwelling causing an increase of the infrared brightness temperature. A similar conclusion was reached by Bailey et al. [2008], who derived rotational temperatures in excess of the VIRA values in regions of enhanced O2 1D emission rate. They associated these regions with conditions of larger downflow velocities where local temperature is increased by compressional heating. Similar and larger temperature enhancements were observed with SPICAV from measurements of UV CO2 absorption measurements during stellar occultations by Bertaux et al. [2007a]. The nonhomogeneous, time-dependent distribution of the O2 1Dg nightglow indicates that the local downward flow of oxygen may differ substantially from the mean value, in response to variations in the efficiency of the global day-to-night transport, the focusing effect of the nightside subsidence, changing zonal wind speeds, eddy transport efficiency, and gravity wave breaking [Bougher et al., 2006]. [8] The NO and O2 nightglows do not occur at the same altitude and thus provide information about different vertical levels: 95 – 105 km for O2 airglow [Piccioni et al., 2009a] and 115 km for NO [Ge´rard et al., 2008a]. In this study, we address the question of the covariation of the two airglow layers. Earlier studies have established that the

E00B44

statistical location of the nightglow bright spots is not coincident. This result was unexpected since the two emissions are produced by recombination of atoms created on the dayside by photodissociation and transported to the nightside by the subsolar to antisolar global circulation. In this study, we take advantage of the unique opportunity offered by two instruments of the Venus Express mission (SPICAV and VIRTIS) to observe almost simultaneously the two emissions and to investigate if their characteristics covary in time and space. The observations reported here were obtained both in nadir-viewing geometry, where horizontal variations of the emission rate can be mapped, and at the limb, where vertical variations are best investigated. We describe both types of observations and draw conclusions on the level of covariance we have observed and transport processes occurring in the Venus mesospherethermosphere transition region.

2. Parallel Nadir Observations [9] The Venus Express spacecraft moves along a quasipolar eccentric orbit with a 24-h period. The apocenter is located at 66,000 km, while the altitude of the pericenter (at 80°N) has varied between 250 km and 185 km. The orbit is fixed in the inertial space and therefore precesses at the rate of 1.6° d – 1. The precession of the orbital plane leads to a wide variety of configurations on the nightside as well as on the dayside. Several observation modes (science cases) may be selected including nadir observations, star pointing for stellar occultations by Venus’ atmosphere, fixed point tracking and limb observations [Titov et al., 2006]. [10] The SPICAV instrument and its performances were described by Bertaux et al. [2007b]. The ultraviolet spectrometer covers a spectral range extending from 118 nm to 320 nm including the NO d (C 2P  X 2P) and g (A 2S  X 2P) emission bands, the only spectral features with Lymana observed in the Venus nightglow [Ge´rard et al., 2008a]. The detector is a 407  288 pixel CCD and the angular field of view of one pixel is equal to 0.7  0.7 arcmin. For reasons of telemetry limitations and because of the time needed to read all the lines of the CCD, only 5 adjacent zones of the CCD detector are usually read out. In these nadir observations, the width of each spatial bin is 32 pixel lines, corresponding to a field of view of 3.7°. These lines are seen through the large (500 mm) slit, providing a spectral resolution of about 12 nm. The planetary area intercepted by the field of view depends on the location on its orbits. The spacecraft altitude ranged between 7350 and 9050 km during the nadir observations reported here. The SPICAV CCD is read out every second, but the actual integration period of each spectrum is 640 ms. The nonuniform dark current and offset values are carefully subtracted in each individual spectrum, using similar observations performed with a null amplification. The absolute calibration obtained by observing well-known hot stars spectra is then applied to obtain nitric oxide emission rates in kR [Bertaux et al., 2007b]. [11] Spectral images have been regularly obtained in nadir geometry with VIRTIS mostly from segments of the orbit near apocenter. The VIRTIS [Drossart et al., 2007b; Piccioni et al., 2009b] pixel size of 0.25 mrad gives a spatial resolution of 15 km on Venus from apocenter. For this

3 of 10

GE´RARD ET AL.: NO AND O2 VENUS NIGHTGLOW OBSERVATIONS

E00B44

E00B44

Table 1. Time, Location, Maximum Brightness and Correlation Coefficient R of the Nitric Oxide and O2 (1D) Airglow Concurrent Observations by SPICAV and VIRTIS-M at Nadir

Orbit

Start Time (UT)

Stop Time (UT)

Local Time

Minimum Latitude (deg)

243 341 342 343 345 346 453 459 567 571 592 901 905 907

20 Dec 2006; 0752:46 28 Mar 2007; 0549:01 29 Mar 2007; 0547:05 30 Mar 2007; 0545:17 1 Apr 2007; 0542:15 2 Apr 2007; 0540:55 18 Jul 2007; 0710:30 24 Jul 2007; 0650:06 9 Nov 2007; 0055:00 13 Nov 2007; 0057:30 4 Dec 2007; 0213:35 8 Oct 2008; 0811:21 12 Oct 2008; 0819:58 14 Oct 2008; 0824:18

20 Dec 2006; 0806:29 28 Mar 2007; 0558:43 29 Mar 2007; 0606:39 30 Mar 2007; 0557:22 1 Apr 2007; 0556:04 2 Apr 2007; 0605:27 18 Jul 2007; 0735:42 24 Jul 2007; 0718:12 9 Nov 2007; 0123:40 13 Nov 2007; 0121:04 4 Dec 2007; 0231:45 8 Oct 2008; 0830:11 12 Oct 2008; 0838:48 14 Oct 2008; 0843:08

0035 2307 2314 2318 2330 2336 2306 2343 2312 0010 0156 2301 2328 2339

1.8 9.3 10.2 10.9 11.7 12.0 16.4 18.8 10.9 16.4 13.4 13.1 12.8 12.7

study, we use the VIRTIS-M mode which provides spectral cubes between 0.25 and 5 mm at a spectral resolution R  200. Each spectral channel is 9.5 nm wide in the region of the O2 Atmospheric Infrared system emission. A spatial scan, covering a 64 mrad  64 mrad field of view is obtained using a scanning mirror. However, even from apocenter, only a fraction of the Venus disk is observed during a mirror scan of the instrument and a spacecraft repointing is needed to collect a more extended coverage. For each VIRTIS image, the thermal contribution from the lower atmosphere is subtracted from the total signal using the VIRTIS fluxes measured in the three adjacent channels centered on 1.27 mm. The count rate is expressed in radiative flux units and MR using the measured instrumental calibration and the O2 (1Dg) relative line intensity for a temperature of 200 K. Airglow radiation emitted downward and subsequently backscattered by the underlying clouds is

Maximum Latitude (deg)

Imax NO (kR)

Imax O2 (MR)

R

48.9 5.1 10.9 0.4 2.3 14.1 38.1 37.5 31.6 19.0 48.3 20.2 20.2 20.4

2.7 1.0 3.1 1.0 2.1 3.7 1.1 2.2 3.5 2.5 6.2 1.7 2.5 3.9

1.1 0,6 1.1 2.2 0.9 3.5 0.6 0.9 1.1 1.1 1.4 0.8 0.7 0.8

0.55 0.84 0.41 0.74 0.23 0.32 0.13 0.08 0.86 0.59 0.25 0.44 0.49 0.26

accounted for using the correction factor derived by Crisp et al. [1996]. [12] The SPICAV and VIRTIS databases have been searched to identify periods when the fields of view of the nadir-viewing observations of both instruments overlapped over a significant time span. Table 1 lists the orbit numbers, times, and locations of these occurrences. As an example, Figure 1 illustrates the spatial coverage in the Venus atmosphere of the SPICAV slit (in gray) and the VIRTIS images (in black) during orbit 243. As can be seen, a common region was observed by both instruments northward of 2°N and southward of 8°S. The shape of the VIRTIS image coverage is defined by the combination of the decreasing spacecraft altitude during the 688 s of the VIRTIS exposure and a reorientation of the spacecraft close to the equator. Similarly, the footprint of the SPICAV slit moved at a nearly constant longitude, with a small deviation

Figure 1. Spatial coverage of SPICAV (in gray) and VIRTIS-M (in black) nadir observations of the Venus nitric oxide and O2 (1D) airglow during Venus Express orbit 243. 4 of 10

E00B44

GE´RARD ET AL.: NO AND O2 VENUS NIGHTGLOW OBSERVATIONS

E00B44

Figure 2. Examples of concurrent observations of the latitudinal distribution of nightglow intensities at nadir by VIRTIS-M and SPICAV as a function of latitude: (a) orbit 243, (b) orbit 592, (c) orbit 342, and (d) orbit 459. Note the different brightness scales used for the NO and the O2 airglow emission rates. The brightness at 1.27 mm has been corrected for backscattered emission and both emission rates have been corrected for the view angle.

at low latitudes. Other cases of parallel observations with the two instruments present a similar pattern of spatial coverage. Once the regions of observation overlap have been identified, the brightness information is extracted from the VIRTIS nadir images along the track of the SPICAV slit. Since a VIRTIS image is constructed by combining adjacent pixel lines corresponding to successive positions of the mirror, the observations are not exactly coincident in time. The time difference is usually on the order of a few minutes. In the particular case of orbit 243 which covers a wide range of latitudes, the maximum time delay between measurements of the intensity with the two instruments at any given location varies between 4 and 16 min. [13] Figure 2a shows the latitudinal distribution of the NO and O2 (1Dg) nadir emission rates for this orbit measured at 0035 LT between 0752 and 0806 UT on 20 December 2006. The NO brightness is sampled once per second and a smoothing function over 10 s has been applied. The O2 signal has been extracted by averaging the intensity in the processed VIRTIS image over an area corresponding to the projection of one of the SPICAV spatial bins, as was shown

in Figure 1. The region of parallel observations extends from close to the equator up to nearly 50°N. In this case, two different regions are identified. The first one, southward of 25°N presents three successive peaks in the NO intensity with brightness ranging from 1 to 2.7 kR. The O2 airglow latitudinal variation shows structural similarities with the NO variations up to about 14°N, with two peaks nearly coincident with the NO maxima. Further north, no clear increase of the O2 brightness corresponds to NO maximum at 17°N. Northward of the location, no correlation is observed between the two signals. In particular, no feature is observed in the NO intensity at 33°N where the O2 airglow increases by over a factor two. In the lowlatitude region of the two common intensity peaks, the O2/ NO airglow intensity ratio is on the order of 500. A similar case is illustrated by Figure 2b for orbit 592 where the two emissions show a different latitudinal trend equatorward of 24°N, followed by a nearly coincident maximum near 30° reaching 5.6 kR in the NO bands and 1.2 MR at 1.27 mm. The two distributions show little correlation poleward of 35°N. In this example, the average O2/NO intensity ratio

5 of 10

E00B44

GE´RARD ET AL.: NO AND O2 VENUS NIGHTGLOW OBSERVATIONS

is again close to 500. An example of uncorrelated structures of the two airglow emissions is illustrated in Figure 2c which shows the latitudinal distribution measured during orbit 342 (29 March 2007) in the premidnight sector (2314 LT). In this case, no correlation is observed between 10°S and 11°S. The O2 airglow exhibits a peak reaching 1 MR on the equator and a decrease on either side of this maximum. A secondary peak is observed at 5°N. The NO airglow presents an equatorial dip with larger intensities up to 3.2 kR at 9°N. In this example, the brightness ratio of the two emissions is on the order of 800. Figure 2d is another example showing no correlation between the NO and the O2 emissions observed during orbit 459. It extends from 20°S to 38°N, close to local midnight (2343 LT). Following correction for the thermal emission component, the O2 (1D) emission rate is very weak southward of 10°N and hardly distinguished from the noise level. It continuously increases toward middle northern latitudes and nearly reaches the 1.2 MR level at the end of the observation sequence. Interestingly, the NO airglow shows a bright maximum of 2.2 kR near 2°N, a region where the O2 emission is very weak. Inversely, the O2 airglow increases poleward of 30°N, which corresponds to a region where the NO airglow drops with increasing latitude. [14] Concurrent sequences of the two airglow emissions have been collected during 14 Venus Express orbits. None of them shows latitudinal distribution of the two emissions which are correlated over the full observation sequence. Instead, the two features are either totally uncorrelated as in Figures 2c and 2d or they exhibit some correlation between the locations of the intensity peaks over a limited latitudinal extent as in Figures 2a and 2b. Table 1 summarizes the dates, times and locations of the parallel airglow observations and the correlation coefficients derived from each orbital sequence. From the examples in Figure 2 and Table 1, we conclude that no large-scale correlation is generally observed in the latitudinal distribution of the vertical emission rate of the NO and O2 airglow. Some of the brightness enhancements are colocated over a restricted region, such as on orbit 343, but the two latitudinal distributions may also show quite a different morphology over regions exceeding 50 degrees of latitude. We now examine concurrent limb observations of the two emissions to verify if the same conclusion holds and increase the sample size of parallel SPICAV-VIRTIS observations.

3. Parallel Limb Observations [15] The observations used for this study were collected in the grazing (tangential) limb mode, where the line of sight is at some angle to the orbital plane and moves in such a way to maximize the time spent in the atmosphere [Titov et al., 2006; Bertaux et al., 2007b]. The SPICAV line of sight scans a range of altitudes, generally between 70 km and 400 km and each second a full UV spectrum is obtained. In this mode, the line of sight crosses the dark limb several times during the ascending portion of the VEX orbit. Therefore, SPICAV supplied several sets of two (one for ingress, one for egress) altitude scans of five altitude profiles at each orbit [Ge´rard et al., 2008a]. During these limb observations, the bin parameter varies between 2 and 32. The apparent altitude of the emission peak and its

E00B44

Figure 3. Sketch illustrating the methodology to generate the data point for the study of the O2 and NO concurrent limb observations shown in Figures 5 and 6. The projection of the SPICAV field of view in the atmosphere is represented by the yellow dotted line. The green dotted line shows the parallel line shifted downward by a distance Dh along which the values of the O2 intensity is extracted to trace emission limb profiles such as illustrated in Figure 4.

brightness depend on the value of the field of view projected on the limb. As was discussed by Ge´rard et al. [2008a], the SPICAV field of view projected on the limb intercepts a vertical region whose size depends on the spacecraft limb distance, the orientation of the slit and the bin parameter. It varies from 3 to 27 km, with a mean value of 14.9 km. This effect is accounted for by smoothing by the field of view of an emission layer having a vertical Chapman profile integrated along the line of sight. Each Chapman profile is constrained to show the same topside scale height as the observation. SPICAV data points have been corrected for this smoothing effect by setting the peak intensity and altitude to the values they would have had if the limb profiles had been observed with a negligibly small field of view. The 0.25 mrad pixel size of the VIRTIS-M detector projected on Venus limb corresponds to a spatial resolution of 1.9 km for a spacecraft distance of 7500 km, a typical value for a VIRTIS observation at 40°N. Analysis of the spectral cubes at the limb has indicated that the contribution of thermal radiation from the lower atmosphere is very small in the vicinity of 1.27 mm for altitudes of the tangent point above 85 km and corrections are negligible above 95 km [Piccioni et al., 2009a]. [16] The possible correlation of the altitude and peak intensity of the NO and O2 airglows at the limb has been investigated on a statistical basis. Periods when both SPICAV and VIRTIS-M were observing the same limb region have been identified and concurrent limb profiles of the two airglows layers have been extracted. The methodology consists in first determining the limb profile of the NO airglow distribution during ingress and egress of the SPICAV line of sight in the lower thermosphere and upper mesosphere. Figure 3 illustrates the example of a profile measured for a line of sight ingress. In this sketch, the emission peaks of the two airglow layers are separated by a vertical distance taken to be equal to 15 km on the basis of the NO and O2 limb statistics. Once the NO emission profile is constructed, the corresponding VIRTIS limb image is scanned and pixels with a tangent point altitude located at an altitude Dh below each successive SPICAV data points are extracted from the VIRTIS cube in the 1.27 mm channel.

6 of 10

E00B44

GE´RARD ET AL.: NO AND O2 VENUS NIGHTGLOW OBSERVATIONS

Figure 4. Example of concurrent limb profiles of the nitric oxide and O2 infrared airglow measured during Venus Express orbit 323.

In this way, an O2 limb profile is extracted from the VIRTIS-M cube so that the effects of horizontal inhomogeneity of the airglow is minimized. The time separating the acquisition of a SPICAV data point and the underlying O2 intensity is, at most, of a few minutes. In this way, quasisimultaneous profiles of the two emissions are obtained for every ingress or egress when the two instruments were simultaneously operating. Figure 4 shows a example obtained for orbit 323 (10 March 2007) between 0035 and 0047 UT at 2258 LT for an airglow layer separation of Dh = 15 km. The NO limb profile shows a peak of 21 kR at 111 km. The O2 profile reaches 69 MR at 99.8 km with a fast intensity drop above the peak. Statistical error bars are indicated on the VIRTIS and SPICAV data points. A total of 249 such parallel limb profiles have been obtained between 17 January 2007 and 9 January 2008. For each profile, the altitude and brightness of the emission peaks are determined and included in the database. The results of this statistical study are summarized in Figures 5 and 6. [17] The altitudes of the simultaneously observed NO and O2 emission profiles are shown in Figure 5. The scatterplot indicates that the intensity of the two airglow layers is not correlated, as confirmed by the very low value of the correlation coefficient R = 0.05. The dashed line indicates equal altitudes for the NO and O2 emission peaks. This plot also confirms that the O2 emission peak is in most circumstances located below the NO layer. However, the distance separating the two emission peaks varies from nearly zero to as much as 28 km, with an average of 15 km. A detailed analysis of the four data points where the O2 peak is above the NO peak has been looked at in detail. These are specific cases when the NO airglow intensity shows a considerable latitudinal gradient as evidenced by the different peak altitudes obtained with the different SPICAV spatial bins. Since the method we use to extract the O2 limb profiles from the VIRTIS images is such that the NO and O2 profiles do not exactly correspond to the same observed volume, it is possible that the actual O2 emission peaks are not really located above the NO emission. The star indicates that the average altitude of the NO airglow peak for this data set is 113 km and 96.4 km for O2(1D). The large scatter in the distance between the two airglows layers probably reflects

E00B44

Figure 5. Altitude distribution of the emission peaks of the NO and O2 (1D) airglow layers measured quasisimultaneously at the limb. The observations corresponding to each point have been collected as illustrated in Figures 3 and 4 (see text). The dashed line indicates equal altitudes for the two airglow emission peaks. The correlation coefficient is 0.05, indicating the lack of covariation of the altitude of the two emissions. The full square indicates the mean value of the peak altitude of the two emissions.

the widely changing dynamical regime prevailing in the transition region between the upper mesosphere and the lower thermosphere. A similar result is obtained when comparing the brightness of the emission peaks in parallel observations shown in Figure 6. For clarity, the NO and O2 brightness has been plotted on a logarithmic scale since they vary by over a wide range of values. As it was found for the peak altitude, this plot indicates that the limb brightness of the two emissions is not significantly correlated (R = 0.29). The O2/NO intensity ratio varies by nearly 3 orders of magnitude from 55 to 40,000. The black square indicates

Figure 6. Brightness distribution of the emission peaks of the NO and O2 (1D) airglow layers measured quasisimultaneously at the limb. The observations corresponding to each point have been determined as illustrated in Figures 3 and 4 (see text). The correlation coefficient is 0.29, indicating the lack of covariation of the brightness of the two emissions. The full square indicates the mean value of the peak intensity of the two emissions.

7 of 10

E00B44

GE´RARD ET AL.: NO AND O2 VENUS NIGHTGLOW OBSERVATIONS

that the average limb brightness of the NO and O2 is 45 kR and 35 MR, respectively, in this data set, in good agreement with the statistical values of 32 kR for NO [Ge´rard et al., 2008a] and 29 MR for O2(1D) (A. Saglam et al., Characteristics of the OH infrared nightglow in the Venus mesosphere and correlation with the O2 (1D) emission at 1.27 mm based on VIRTIS limb observations, submitted to Advances in Space Research, 2009). The sensitivity of the results to the value of the distance Dh separating the two emission peaks has been tested by varying it by ±5 km. No significant difference was found in the results. It is thus concluded that neither the altitude of the emissions peaks nor the peak intensities at a given location in the Venus nightside atmosphere are correlated. This result is in full agreement with the independent conclusion derived from the nadir observations reported in section 2. We now examine possible explanations for these differences in the section 4.

4. Discussion [18] The observations collected with the SPICAV spectrograph and the VIRTIS-M spectral imager provide evidence that the molecular oxygen and the nitric airglow emissions are only weakly correlated. This conclusion is based on three different sets of observations. First, the statistical location of the NO and the O2(1D) regions of enhanced emission are not coincident. This was demonstrated by the difference between the O2 spot centered on the equator at midnight [Ge´rard et al., 2008b; Piccioni et al., 2009a] and the NO maximum which is shifted by about two hours toward dawn and southward of the equator [Stewart et al., 1980; Bougher et al., 1990]. Second, nadir quasi-simultaneous observations of the two emissions reported in section 2 of this study demonstrate that the distribution of the intensity along latitudinal cuts exhibits significant differences even though the two emissions may show similarities over a limited range. Third, limb observations indicate that neither the brightness nor the altitude of the emission peak covary in the northern hemisphere. The first aspect requires additional studies and modeling. The shift of the statistical region of bright NO emission from the antisolar point [Stewart et al., 1980; Bougher et al., 1990; Bougher and Borucki, 1994] is an indication that the dawnward superotation observed at the cloud level persists in the upper hemisphere in such a way that the subsidence region of the global thermospheric circulation is statistically displaced by 2 h. This result needs further SPICAV observations to confirm that the shift is still observed in a period of low solar activity conditions. If the difference between the locations of the NO and O2 bright regions is still observed during the Venus Express era, the picture of the vertical wind structure in the upper mesosphere-lower thermosphere transition region has to be revised accordingly. [19] The decoupling between the characteristics of the two emissions at a given location and time raises a different question. At first glance, the results reported in this study contradict the concept of a global subsolar-to-antisolar circulation carrying the O and N atoms from their dayside source region to the nightside location where they recombine to produce the NO and O2 (1D) airglow emissions. In this view, the region of subsidence of the two species would

E00B44

Figure 7. Sketch illustrating the role of horizontal wind in the mesosphere-thermosphere transition region as a source of spatial decorrelation between a bright spot of NO airglow and a region of enhanced O2 (1D) nightglow. The region initially enriched in oxygen atoms has traveled a horizontal distance DL by the time the blob of O-rich gas reaches the altitude of the O2 nightglow layer located Dz km below the O2 emission peak.

be nearly coincident in the vicinity of the antisolar point. We note however that, at any given time, airglow images indicate that the location of the regions of bright emissions may considerably vary as was shown for NO [Stewart et al., 1980; Bougher et al., 1990] and O2 (1D) [Hueso et al., 2008; Piccioni et al., 2009a]. This important feature shows that the location of the subsidence of the SS-AS circulation is very variable, presumably as a consequence of an intrinsic variability of the circulation (possibly caused by the drag from gravity waves) and horizontal winds. As mentioned before, apparent wind velocities locally as large as about 100 ms1 have been deduced by Hueso et al. [2008] from the displacement of O2 airglow features between successive VIRTIS-M images. [20] We now take a close look at the role of horizontal winds in the decorrelation between the two airglow emissions. The two airglow layers are separated in altitude typically by 16 km. Although this distance is limited, the time required for vertical down transport of atoms from the NO layer to the O2 emission peak is fairly long, leaving the possibility that a region richer in oxygen at 113 km has moved over considerable horizontal distances by the time it reaches the 97 km level. This situation is illustrated by the sketch in Figure 7 where an air parcel of enhanced NO emission (in white-blue) is carried over an horizontal distance DL by the time it has reached the altitude of maximum O2 recombination Dh km below the NO layer. The effect of this horizontal transport in a situation of nonuniform downward flux above the NO airglow layer is potentially able to explain the observed decorrelation. We first need to estimate the downward velocity of an air parcel to traveling the distance Dh separating the two airglow layers. The vertical transport velocity w has not been directly measured but may be estimated using models and circumstantial evidence. We examine three possible ways to derive typical vertical velocity values. First, in the one-dimensional chemical

8 of 10

E00B44

GE´RARD ET AL.: NO AND O2 VENUS NIGHTGLOW OBSERVATIONS

transport model used by Cox et al. [2008] and Ge´rard et al. [2008a], vertical transport below the homopause is parameterized by an eddy diffusion coefficient K. We adopt a standard value of DZ = 16 km for the distance separating the average altitudes of the NO and O2 emission peaks. On the basis of numerical simulations using the expression K = A/n1/2, we find a velocity on the order of 3 cm s1. A second approach to estimate w is based on results from the three-dimensional model by Bougher et al. [1990] where eddy diffusion is relatively small. In a recent version of this model, the vertical advection velocity between 115 and 95 km is on the order of 15–20 cm s1 near the antisolar point (S. W. Bougher, private communication, 2009). Finally, an estimate of the downflow velocity was also given by Bailey et al. [2008] for a nightside region with a 20 K temperature enhancements observed over a 2-day period. They obtained a vertical flow velocity of 20 cm s1 for such a region of airglow brightening associated with enhanced vertical transport. The corresponding transit time for atoms flowing from the NO to the O2 airglow peak ranges between 22 h and 6 Earth days. [21] The horizontal wind velocity v at the level of the O2 airglow layer is very variable as was recently summarized by Lellouch et al. [2008]. Winds velocities, derived from CO millimeter observations, are typically on the order of 30– 50 m s1 at 93 km and 90– 120 m s1 near 102 km. Hueso et al. [2008] derived values of the effective wind velocity at 97 km of a few tens of m/s from the motions of bright spots of O2 airglow. Horizontal winds calculated with the 3-D model values are on the same order in the region separating the two emission layers. A crude estimate of the horizontal distance DL crossed by moving air parcel is then given by DL = v/w DZ. Adopting a value of v = 100 m s1 as an upper limit, we find values on the order of 53,000 km if vertical transport is parameterized by eddy diffusion or 8000 km if the estimate for w from Bailey et al. adopted. Using v = 10 m s1 as a wind velocity probably closer to the average, we estimate a typical horizontal transport range of 800 to 5300 to km. Actually, the situation is further complicated by the occurrence of chemical reactions and collisional quenching which limit the chemical lifetimes of the ground state O atoms and the excited O2(1D) molecules, respectively. In any case, it appears that the downward flow is much slower than the horizontal transport and atoms may travel considerable horizontal distances during their transit from 113 to 97 km. It is therefore a direct explanation of the lack of correlation between the two emissions observed concurrently by SPICAV and VIRTIS-M. The very fact that the two emissions exhibit significant differences in their horizontal distribution strongly argues for the presence of strong horizontal winds in the thermosphere-mesosphere transition region. Other factors such as the different altitude where the N and the O atoms are formed on the dayside may play an additional role in the lack of correlation since the streamlines followed during their transport to the nightside are slightly different. [22] On the basis of these arguments, it may now appear more difficult to understand the covariation occasionally observed in latitudinal cuts such as in Figures 2a and 2c. We speculate that the two emissions may covary at least under three particular conditions. First, when the distance between the two airglow layers is less than the average 16 km. The

E00B44

existence of such conditions is testified by Figure 6, which shows the occurrence of cases when this distance may be reduced to only a few kilometers. Under such circumstances, the vertical downflow time and the horizontal traveled distance may also be reduced accordingly. Second, maps of the horizontal displacements of bright O2 features have indicated that regions with much smaller horizontal velocities, sometimes quasi-null values, have been observed near 97 km. It is possible that these horizontal stagnation regions persist over a sufficiently long period of time, allowing a more vertical downflow of the oxygen atoms. Finally, local enhancements in the brightness of the O2 airglow suggest the presence of strong downflows, possibly associated with an increased vigor of local downward turbulent transport. These conditions may well explain that the two emissions can exhibit some degree of covariation, which is otherwise absent in areas of strong horizontal winds and/or weaker vertical velocity.

5. Conclusion [23] The O2(1D) and the NO d and g bands night airglows of Venus have been observed concurrently for the first time by two instruments on board the Venus Express spacecraft. These observations include both nadir and limb viewing geometries. During nadir observations, the O2 (a1D) emission intensity has been extracted from VIRTIS images observing the same locations as the footprint of the SPICAV slit in the Venus atmosphere on the same orbit. Occasional positive correlations between the latitudinal distributions of the two emissions have been observed over a limited latitudinal range. However, the overall covariability of the two airglows is low, as shown by some of the results illustrated in this study and by the globally low correlation coefficients obtained between the two sets of observational sequences. A similar conclusion is reached from the statistical comparison of the altitude and peak brightness of the two emissions simultaneously observed in the grazing limb geometry. The very low correlation coefficients of both the altitude and the intensity of the two airglow layers indicate that the transport of O and N atoms considerably deviates from a steady state vertical flow, even in the region close to the antisolar point. We suggest that the airglow decorrelation frequently observed is a consequence of the transport of the downward moving air mass by strong horizontal winds in the transition region that dynamically decouples the airglow in the two layers in a given vertical column. Our simple calculation of horizontal transport indicates that the atoms may travel considerable distances during the transit time of their vertical transport by vertical advection and eddy diffusion. Occasional correlations of the latitudinal intensity variations may be associated with situations when the distance between airglow layers is small or when the ratio of the horizontal and vertical transport velocity components is reduced. Simulations with a two- or threedimensional chemical transport model are needed to assess this scenario and quantify this effect. [24] Acknowledgments. We gratefully thank all members of the ESA Venus Express project and of the VIRTIS ( http://servirtis.obspm.fr/Venus_ Express/VIRTIS_Team.html) and SPICAV scientific and technical teams. J.C.G. acknowledges funding from the Belgian Fund for Scientific Research (FNRS). A. Saglam, C. Cox, and L. Soret were supported by

9 of 10

E00B44

GE´RARD ET AL.: NO AND O2 VENUS NIGHTGLOW OBSERVATIONS

the PRODEX program managed by the European Space Agency with the help of the Belgian Federal Space Science Policy Office. This work was funded by Agenzia Spaziale Italiana and the Centre National d’Etudes Spatiales.

References Alien, D., D. Crisp, and V. Meadows (1992), Variable oxygen airglow on Venus as a probe of atmospheric dynamics, Nature, 359, 516 – 519, doi:10.1038/359516a0. Bailey, J., V. S. Meadows, S. Chamberlain, and D. Crisp (2008), The temperature of the Venus mesosphere from O2 (a1Dg) airglow observations, Icarus, 197, 247 – 259, doi:10.1016/j.icarus.2008.04.007. Bertaux, J.-L., et al. (2007a), A warm layer in Venus’ cryosphere and high altitude measurements of HF, HCl, H2O and HDO, Nature, 450, 646 – 649, doi:10.1038/nature05974. Bertaux, J.-L., et al. (2007b), SPICAV on Venus Express: Three spectrometers to study the global structure and composition of the Venus atmosphere, Planet. Space Sci., 55, 1673 – 1700, doi:10.1016/j.pss.2007. 01.016. Bougher, S. W., and W. J. Borucki (1994), Venus O2 visible and IR nightglow: Implications for lower thermosphere dynamics and chemistry, J. Geophys. Res., 99, 3759 – 3776, doi:10.1029/93JE03431. Bougher, S. W., J. C. Ge´rard, A. I. F. Stewart, and C. G. Fesen (1990), The Venus nitric oxide night airglow: Model calculations based on the Venus Thermospheric General Circulation Model, J. Geophys. Res., 95, 6271 – 6284, doi:10.1029/JA095iA05p06271. Bougher, S. W., S. Rafkin, and P. Drossart (2006), Dynamics of the Venus upper atmosphere: Outstanding problems and new constraints expected from Venus Express, Planet. Space Sci., 54, 1371 – 1380, doi:10.1016/ j.pss.2006.04.023. Connes, P., J. F. Noxon, W. A. Traub, and N. P. Carleton (1979), O2 1D emission in the day and night airglow of Venus, Astrophys. J., 233, L29 – L32, doi:10.1086/183070. Cox, C., A. Saglam, J.-C. Ge´rard, J.-L. Bertaux, F. Gonza´lez-Galindo, F. Leblanc, and A. Reberac (2008), The distribution of the ultraviolet nitric oxide Martian night airglow: Observations from Mars Express and comparisons with a one-dimensional model, J. Geophys. Res., 113, E08012, doi:10.1029/2007JE003037. Crisp, D., V. S. Meadows, B. Be´zard, C. de Bergh, J. P. Maillard, and F. P. Mills (1996), Ground-based near-infrared observations of the Venus nightside: 1.27 mm O2(1Dg) airglow from the upper atmosphere, J. Geophys. Res., 101, 4577 – 4594, doi:10.1029/95JE03136. Drossart, P., et al. (1993), Search for spatial variations of the H2O abundance in the lower atmosphere of Venus from NIMS-Galileo, Planet. Space Sci., 41, 495 – 504, doi:10.1016/0032-0633(93)90032-W. Drossart, P., et al. (2007a), Infrared spectral imaging observations of Venus by VIRTIS reveal a dynamical upper atmosphere, Nature, 450, 641 – 645, doi:10.1038/nature06140. Drossart, P., et al. (2007b), Scientific goals for the observation of Venus by VIRTIS on ESA/Venus express mission, Planet. Space Sci., 55, 1653 – 1672, doi:10.1016/j.pss.2007.01.003. Feldman, P. D., H. W. Moos, J. T. Clarke, and A. L. Lane (1979), Identification of the UV nightglow from Venus, Nature, 279, 221 – 222, doi:10.1038/279221a0. Garcı´a Mun˜oz, A., F. P. Mills, G. Piccioni, and P. Drossart (2009), The near-infrared nitric oxide nightglow in the upper atmosphere of Venus, Proc. Natl. Acad. Sci. U. S. A., 106, 985 – 988, doi:10.1073/pnas. 0808091106. Ge´rard, J.-C., A. I. F. Stewart, and S. W. Bougher (1981), The altitude distribution of the Venus ultraviolet airglow and implications on vertical trans port, Geo phys. Re s. Lett. , 8 , 63 3 – 636 , do i:1 0.1 029 / GL008i006p00633. Ge´rard, J.-C., C. Cox, A. Saglam, J.-L. Bertaux, E. Villard, and C. Nehme´ (2008a), Limb observations of the ultraviolet nitric oxide nightglow with

E00B44

SPICAV on board Venus Express, J. Geophys. Res., 113, E00B03, doi:10.1029/2008JE003078. Ge´rard, J.-C., A. Saglam, G. Piccioni, P. Drossart, C. Cox, S. Erard, R. Hueso, and A. Sa´nchez-Lavega (2008b), The distribution of the O2 infrared nightglow observed with VIRTIS on board Venus Express, Geophys. Res. Lett., 35, L02207, doi:10.1029/2007GL032021. Ge´rard, J.-C., A. Saglam, G. Piccioni, P. Drossart, F. Montmessin, and J.-L. Bertaux (2009), Atomic oxygen distribution in the Venus mesosphere from observations of O2 infrared airglow by VIRTIS-Venus Express, Icarus, 199, 264 – 272, doi:10.1016/j.icarus.2008.09.016. Hueso, R., A. Sa´nchez-Lavega, G. Piccioni, P. Drossart, J.-C. Ge´rard, I. Khatuntsev, and L. Zasova (2008), Morphology and dynamics of Venus oxygen airglow, J. Geophys. Res., 113, E00B02, doi:10.1029/ 2008JE003081. Krasnopolsky, V. A. (1986), Oxygen emissions in the night airglow of the Earth, Venus, and Mars, Planet. Space Sci., 34, 511 – 518, doi:10.1016/ 0032-0633(86)90089-9. Lellouch, E., T. Clancy, D. Crisp, A. Kliore, D. Titov, and S. W. Bougher (1997), Monitoring of mesospheric structure and dynamics, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 295 – 324, Univ. of Ariz. Press, Tucson. Lellouch, E., G. Paubert, R. Morenoaand, and A. Moullet (2008), Monitoring Venus’ mesospheric winds in support of Venus Express: IRAM 30-m and APEX observations, Planet. Space Sci., 56, 1355 – 1367, doi:10.1016/j.pss.2008.06.010. Miller, H. C., J. E. McCord, J. Choy, and G. D. Hager (2001), Measurement of the radiative lifetime of O2 (a1Dg) using cavity ring down spectroscopy, J. Quant. Spectrosc. Radiat. Transfer, 69, 305 – 325. Ohtsuki, S., N. Iwagami, H. Sagawa, M. Ueno, Y. Kasaba, T. Imamura, and E. Nishihara (2008), Imaging spectroscopy of the Venus 1.27-mm, O2 airglow with ground-based telescopes, Adv. Space Res., 41, 1375 – 1380, doi:10.1016/j.asr.2007.10.014. Piccioni, G., L. Zasova, A. Migliorini, P. Drossart, A. Shakun, A. Garcı´a Mun˜oz, F. P. Mills, and A. Cardesin-Moinelo (2009a), Near-IR oxygen nightglow observed by VIRTIS in the Venus upper atmosphere, J. Geophys. Res., 114, E00B38, doi:10.1029/2008JE003133. Piccioni, G., et al. (2009b), The Visible and Infrared Thermal Imaging Spectrometer, Eur. Space Agency Spec. Publ., ESA SP-1295, in press. Saglam, A., J.-C. Ge´rard, L. Soret, G. Piccioni, and P. Drossart (2009), Characteristics of the OH infrared nightglow in the Venus mesosphere and correlation with the O2 (1D) emission at 1.27 mm based on VIRTIS limb observations, submitted to Advances in Space Research. Stewart, A. I., and C. A. Barth (1979), Ultraviolet night airglow of Venus, Science, 205, 59 – 62, doi:10.1126/science.205.4401.59. Stewart, A. I. F., J.-C. Ge´rard, D. W. Rusch, and S. W. Bougher (1980), Morphology of the Venus ultraviolet night airglow, J. Geophys. Res., 85, 7861 – 7870, doi:10.1029/JA085iA13p07861. Svedhem, H., et al. (2007), Venus Express: The first European mission to Venus, Planet. Space Sci., 55, 1636 – 1652, doi:10.1016/j.pss.2007. 01.013. Titov, D. V., et al. (2006), Venus Express science planning, Planet. Space Sci., 54, 1279 – 1297, doi:10.1016/j.pss.2006.04.017. 

J.-L. Bertaux, LATMOS, Re´duit de Verrie`res, BP3, Route des Gaˆtines, F-91371 Verrie`res-le-Buisson, France. ([email protected]) C. Cox, J.-C. Ge´rard, A. Saglam, and L. Soret, Laboratoire de Physique Atmosphe´rique et Plane´taire, Universite´ de Lie`ge, B-4000 Lie`ge, Belgium. ([email protected]; [email protected]; [email protected]; lauriane. [email protected]) P. Drossart, LESIA, Observatoire de Paris, 5 Place Jules Janssen, F-92195 Meudon, France. ([email protected]) G. Piccioni, INAF, IASF, Via del Fosso del Cavaliere 100, I-00133 Rome, Italy. ([email protected])

10 of 10

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, E00B17, doi:10.1029/2008JE003118, 2008

for

Full Article

Venus surface thermal emission at 1 mm in VIRTIS imaging observations: Evidence for variation of crust and mantle differentiation conditions N. Mueller,1 J. Helbert,2 G. L. Hashimoto,3 C. C. C. Tsang,4 S. Erard,5 G. Piccioni,6 and P. Drossart5 Received 15 February 2008; revised 10 July 2008; accepted 20 August 2008; published 10 December 2008.

[1] The Venus Express spacecraft images the nightside thermal emissions using

the Visible and Infrared Thermal Imaging Spectrometer (VIRTIS). At 1.02, 1.10, and 1.18 mm, thermal emission from the surface is observed. The signal is attenuated by scattering and absorption in the dense atmosphere. The measured flux at the top of the atmosphere is positively correlated with surface temperature and surface emissivity. The surface temperature of Venus is relatively well constrained as being mainly a function of altitude with a gradient lesser or equal to the adiabatic lapse rate. This study examines the correlation of VIRTIS images showing a signal of the surface at 1.02 mm with viewing geometry, stray sunlight, cloud opacity, and topography and applies semiempirical relations to remove their influence. The remaining contrast can be either ascribed to surface emissivity or unexpected temperature variations. Temperature variations due to active volcanism are unlikely to be persistent over the time of observations; therefore, the mosaic of all processed images is here interpreted in terms of surface emissivity variation. The emissivity variation found is correlated with geomorphological features established from Magellan synthetic aperture radar images. It is generally lower at tessera terrain. Some, but not all, volcanic edifices show increased emissivity. Large lava flows in the Lada terra-Lavinia planitia region also show an increased thermal emission. This might indicate a more felsic surface composition of tessera highlands and large-scale extrusive volcanism of ultramafic composition. Citation: Mueller, N., J. Helbert, G. L. Hashimoto, C. C. C. Tsang, S. Erard, G. Piccioni, and P. Drossart (2008), Venus surface thermal emission at 1 mm in VIRTIS imaging observations: Evidence for variation of crust and mantle differentiation conditions, J. Geophys. Res., 113, E00B17, doi:10.1029/2008JE003118.

1. Motivation [2] The search for exoplanets yields an ever increasing number of known planets. With further improvements in observation techniques it will hopefully soon be possible to detect Earth sized planets orbiting their stars at distances where liquid surface water is stable. The special interest in planets with these attributes is of course motivated by the desire to find habitable planets or even life. Size and orbital distance comparable to Earth is by no means an ideal indicator of habitability as the Earth’s neighboring planets demonstrate. Venus with eight tenths of Earth’s mass and 1 Joint Planetary Interior Physics Research Group of the University Muenster and DLR Berlin, Berlin, Germany. 2 Institute of Planetary Research, DLR, Berlin, Germany. 3 Laboratory for Earth and Planetary Atmospheric Science, Department of Earth and Planetary Sciences, Kobe University, Kobe, Japan. 4 Atmospheric, Oceanic, and Planetary Physics, Department of Physics, Clarendon Laboratory, University of Oxford, Oxford, UK. 5 LESIA, Observatoire de Paris, Paris, France. 6 IASF-INAF, Rome, Italy.

Copyright 2008 by the American Geophysical Union. 0148-0227/08/2008JE003118$09.00

nearly three fourths of Earth’s orbital distance is quite inhospitable. The surface of Venus is extremely hot owing to greenhouse climate imposed by its dense CO2 atmosphere and sulfuric acid clouds. Surface and atmosphere are also dry compared to Earth [Taylor, 2006]. [3] Comprehensive understanding of why Venus so utterly failed to become remotely habitable has not yet been achieved [Crisp et al., 2002]. One missing piece for the reconstruction of the evolution of Venus is global knowledge on the composition of the crust. The surface has been extensively mapped at radar wavelengths by the Magellan mission [Pettengill et al., 1991] which further revealed a unique geology [e.g., Saunders et al., 1991; Head et al., 1992; Solomon et al., 1992; Schaber et al., 1992; Phillips et al., 1992; Stofan et al., 1992, 1997; Hansen et al., 1997; Smrekar et al., 1997; Grimm and Hess, 1997; Basilevsky et al., 1997]. Radar is however not very sensitive to mineral composition. The discovery of spectral windows in the atmosphere (Figure 1) that allow measurable amounts of near-infrared thermal emission from the surface to escape [Carlson et al., 1991] presents the opportunity for limited remote sensing of surface mineralogy [Carlson et al., 1993a; Lecacheux et al., 1993; Baines

E00B17

1 of 21

E00B17

MUELLER ET AL.: VENUS SURFACE THERMAL EMISSION

Figure 1. Modeled spectrum of the near-infrared (NIR) windows at VIRTIS-H spectral resolution [Drossart et al., 2007]. Model is described by Tsang et al. [2008]. The different windows feature different amounts of emission originating from the surface: at 1.02 mm more than 95% comes from the surface, and at 1.27 and 1.31 mm a negligible part of the radiation comes from the surface [Meadows and Crisp, 1996]. et al., 2000; Moroz, 2002; Hashimoto and Sugita, 2003; Hashimoto et al., 2008; Arnold et al., 2008; A. T. Basilevsky et al., Geologic interpretation of the near-infrared images of the surface taken by the Venus Monitoring Camera, Venus Express, submitted to Journal of Geophysical Research, 2008]. [4] The window most suited for this purpose is at 1 mm which coincides with a absorbtion band of FeO, one of the primary compounds of mafic minerals. Abundance of refractory mafic minerals gives evidence of the conditions under which the ultramafic mantle material differentiated and formed the crust. It is possible to distinguish three different types of crustal differentiation [Taylor, 1989]. Primary differentiation is the formation of floating crust in a fully molten mantle or magma ocean. Evidence for this is found in the lunar highland anorthosite rocks returned by the Apollo missions [Taylor, 1974]. Secondary differentiation occurs when the convecting mantle material partially melts, the liquid phase moves upward and forms a basaltic crust either by volcanism or plutonic intrusion, e.g., at midocean ridges on Earth. Tertiary differentiation requires recycling of basaltic crust and water into the mantle. Water lowers the liquidus temperature of secondary crust which creates partial melts with a felsic composition that generated the bulk of the continental crust on Earth [Taylor and Campbell, 1983]. The greater part of the venusian crust is thought to be formed by secondary differentiation [Head et al., 1994; Grimm and Hess, 1997]. Arguments for this are lava viscosity inferred from morphology, in situ measurements and the unimodal hypsometry. Most in situ measurements indicate a mafic composition comparable to common terrestrial basalts [Surkov and Barsukov, 1985; Surkov et al., 1986, 1987]. Two out of seven, Venera 8 and Venera 13, indicate a more complex differentiation process that involves enrichment in abundance of alkalines [Kargel et al., 1993]. Abundance of alkalines is not reflected in the felsic-mafic description.

E00B17

[5] Crustal differentiation is counteracted by thermal buoyancy, which on Earth leads to remixing of crustal material into the mantle by oceanic plate subduction. It is not very well constrained whether crustal recycling takes or took place on Venus [Grimm and Hess, 1997]. Crater frequency gives strong evidence for relative recent resurfacing [Phillips et al., 1992; Schaber et al., 1992; Strom et al., 1994]. The crust mapped by radar appears young because extrusive volcanism and tectonic deformation have erased the early crater history. The processes responsible for the resurfacing are not fully understood, and the fact of a seemingly young crust does not constrain crustal recycling. For instance, the concept of episodic plate tectonics [Turcotte, 1993] involves recycling of most of the crust and simultaneous creation of new crust. In other models considering vertical rather than horizontal accretion [e.g., Parmentier and Hess, 1992; Head et al., 1994], recycling takes place only if crustal growths exceeds a thickness of 60 to 80 km and crustal basalt minerals undergo a phase transition to denser eclogite [Spohn, 1991]. Mapping of geochemically old crust (i.e., granite or anorthosite) might constrain rate of crustal formation and recycling. [6] Another product of mantle differentiation is the silica and FeO depleted material remaining in the mantle after the liquid basaltic phase migrated upward. Parmentier and Hess [1992] assume that this residual material forms a compositionally buoyant layer below the crust. On Earth this reservoir is thought to be small owing to lithosphere subduction. On Venus plate tectonics is not unambiguously identifiable [Kaula and Phillips, 1981; Solomon et al., 1992], and the lithosphere is assumed at present to form a stagnant lid [Solomatov and Moresi, 1996]. The reservoir of depleted mantle material on Venus is therefore likely greater than on Earth and steadily increasing during secondary differentiation as it is not remixed into the mantle by subduction. Cooling of this conducting layer leads to negative net buoyancy and foundering of the layer drives episodic tectonic and volcanic resurfacing. In the context of this resurfacing model, Head et al. [1994] predict composition of lava to be of ultramafic, e.g., picritic or komatiitic composition during the periods of quiescence as partial melting in mantle diapirs occurs in deeper regions under different conditions. [7] Regardless of the stagnant lithosphere the surface is strongly influenced by endogenous processes. Mantle diapirs supposedly lift the volcanic dome shaped highlands through dynamic support and lithosphere thinning and imprint coronae structures [Smrekar et al., 1997; Stofan et al., 1997]. The wide range of topographic expressions of coronae can be explained by mantle diapirism triggering a gravitational instability and delamination of lithosphere [Smrekar and Stofan, 1997]. Lithosphere delamination can also account for the wide range of inferred lava viscosities requiring a diversity of melt compositions [Elkins-Tanton et al., 2007]. Their modeling of the temperature-pressure conditions in and around a diapir of delaminating lithosphere demonstrates the possibility of generation of melt varying in SiO 2 abundance from 60% (intermediary between felsic and mafic) to 40% (ultramafic). [8] The process of differentiation and recycling of crustal material is relevant for the thermal evolution of the interior of the planet. One side is the effective cooling of the interior

2 of 21

E00B17

MUELLER ET AL.: VENUS SURFACE THERMAL EMISSION

by recycling of cool crustal material. The other side is the distribution of radioactive heat sources throughout the planet. Important radioactive heat sources such as 40K, 140 Th, and 239U are incompatible, during partial melting these isotopes are enriched in the liquid, more felsic phase. Thus, on Earth the oceanic basaltic and especially the continental granitic crust contain higher abundances of radioactive elements than the mantle. This concentration of heat sources toward the top decreases the heat flux in the deep interior of the planet and also influences amount of magmatism by depletion of mantle heat sources [Spohn, 1991]. [9] The convective heat flux in the outer core driven by overall cooling of the planet is the energy source for the magnetic field of the Earth and presumably Mercury [Stevenson et al., 1983]. The geodynamo process is currently not active on Venus but it is conceivable that it was in the past when distribution of the radioactive elements or the cooling of the planetary interior by lithosphere recycling was different. An intrinsic magnetic field would greatly influence the process of atmospheric erosion and allow Venus to retain its primordial water longer [Donahue and Russell, 1997]. In such an primordial climate and tectonic regime tertiary differentiation might have occurred and led to granitic cratons still existing in the Venusian highlands [Taylor and Campbell, 1983]. Hashimoto and Sugita [2003] concluded that it is possible to distinguish mafic basalt from felsic granite or rhyolite from orbit when given a sufficient but not impossible accuracy of measurement of thermal emission on the nightside of Venus. [10] The aim of this study is to analyze VIRTIS images to find whether there is a spatial signal that can be plausibly attributed to content of mafic minerals of the topmost crust in the context of surface morphology from Magellan synthetic aperture radar (SAR) images.

2. Background and Theory 2.1. Atmosphere and Surface Temperature [11] The thermal emission is dominated by surface temperature T as it is at any wavelength l the product of surface emissivity el and blackbody spectral radiance given by the Planck function Bl(T). At the wavelengths and temperatures relevant here (1 mm and 735 K) the Planck function varies strongly with temperature. To retrieve surface emissivity from remote sensing of the thermal emission, the physical surface temperature must be well known. On Venus atmospheric temperature is well constrained owing to the insulating greenhouse climate and measurements at different local times and locations by the Venera, Pioneer Venus, and Vega descent probes. The surface temperature is thought to be in thermal equilibrium with the atmosphere owing to high heat capacity of the atmosphere and little diurnal variations [Lecacheux et al., 1993]. [12] Stone [1975] estimated the different time scales of radiative heating and cooling and convective heat transport of the atmosphere of Venus and concluded that temperature was governed by the latter. Change of temperature by solar heating and cooling during the night is effectively distributed through the whole atmosphere by advection. Average diurnal and meridional temperature variations in the lower atmosphere were estimated to be less than 0.1 K.

E00B17

[13] The in situ measurements of Venera 9 to 11 and the four Pioneer Venus descent probes reviewed by Seiff [1983] agree with each other within approximately 12 K at the same pressure level in the atmosphere below the clouds. The temperature increases toward the surface close to the adiabatic gradient. Diurnal variations are less than 1 K and latitudinal variations are less than 5 K in the data from the Pioneer Venus probes. [14] Observations combined with atmospheric models lead to the Venus International Reference Atmosphere (VIRA). The model temperature structure is presented by Seiff et al. [1985]. The atmosphere in contact with the surface close to the mean planetary radius has a temperature of 735.3 K at 92.1 bar, adiabatical lapse rate is
8.06 K km
1, no diurnal or latitudinal variations are included in the model atmosphere. [15] Little is known however about the planetary boundary layer of the atmosphere since Venera probes were designed to travel through the hot atmosphere at great speed to maximize lifetime on the surface and the vertical temperature sampling rate is low. The higher-resolution measurements of the Pioneer Venus probes ceased at 12 km height. The Vega 2 temperature profile is the only measurement in the lower part of the atmosphere comparable in resolution to Pioneer Venus measurements. It shows a stratified atmosphere with dynamically unstable superadiabatic lapse rates in the lowest 6.5 km [Seiff, 1987]. In the lowest 1.5 km the lapse rates vary between
1.5 and
10 K km
1. These measurements are not verified and no explanation for them has been found yet [Crisp and Titov, 1997]. Theoretical consideration by Gierasch et al. [1997] of the planetary boundary, where vertical convective motion is inhibited by the surface of the planet, gives an upper estimate of a diurnal temperature oscillation of 6 K corresponding to a 170 m thin thermal boundary layer. [16] In conclusion, the small variation of temperature in the lowest layer of the atmosphere is well constrained but the temperature structure is not. The single high-resolution in situ measurement points toward possible superadiabatic lapse rates near the surface but is on average
8 K km
1. Temperatures retrieved from observation of the thermal surface emission at 1.18 mm via radiative transfer modeling suggest a subadiabatic lapse rate of less than
7.5 K km
1 [Meadows and Crisp, 1996]. Measurement variations and theoretical diurnal variations near the surface are both less than 10 K at the same pressure level. The atmospheric boundary layer is probably not very relevant for the nightside owing to the long duration of nights on Venus (58 Earth days). No significant near-infrared (NIR) thermal emission variation with local time or latitude has been reported by Lecacheux et al. [1993], Meadows and Crisp [1996], or Hashimoto et al. [2008]. 2.2. Radiative Transfer in the Atmosphere of Venus [17] The atmosphere of Venus consists mostly of CO2 (96.5 wt %) and N2 (3.5 wt %) [von Zahn et al., 1983] with a pressure of 93.2 bar at the zero altitude mean planetary radius (MPR) [Seiff et al., 1985]. From approximately 50 to 70 km above MPR there are several cloud layers composed mostly of droplets of concentrated sulphuric acid. The clouds have spatially varying number densities and size distributions resulting in optical depths between 30 and

3 of 21

E00B17

MUELLER ET AL.: VENUS SURFACE THERMAL EMISSION

Figure 2. The altitude where brightness temperature of the lower atmosphere without clouds corresponds to atmospheric temperature gives an approximation for the source region of radiation. Spectrum calculated from the same model as in Figure 1 but without clouds. Isolines of Rayleigh scattering optical depths t r calculated following Hansen and Travis [1974]. Rayleigh scattering is less relevant for the 1.31 mm window than for the 1.02 mm window. Brightness temperature varies little between 1.01 and 1.03 mm in this model with a constant continuum absorption coefficient. Compare with spectral dependance of brightness temperature in Figure 8. 50 at visible wavelengths, for a review see Ragent et al. [1985]. A three modal size distribution with modal radii of 0.3, 1.0, and 3.7 mm is well established from in situ and remote observations [Hansen and Hovenier, 1974; Marov et al., 1980; Knollenberg and Hunten, 1980; Carlson et al., 1993b; Grinspoon et al., 1993], but it is unclear whether this is a comprehensive description of all cloud particles. In particular the nature of the UV absorber leaving distinct cloud shaped marks on dayside images is not known [Esposito et al., 1997]. Also, for mode 1 and 3 particles, compositions different from H2SO4 solutions and including solid crystals have been proposed [Ragent et al., 1985; Esposito et al., 1997]. In clouds models usually a mode 20 is included that features a slightly larger modal radius and size variation than mode 2. Above and below the clouds are haze layers of mostly submicrometer particles. Hazes below the clouds have a total optical depth of less than 3 [Ragent et al., 1985] that, however, might have been as high as 10 in Venera 8 near terminator measurements [Moroz, 2002]. Ragent et al. [1985], reviewing the different in situ measurements of hazes, gives no information for altitudes below 10 km. Grieger et al. [2004] reexamined the zenith and nadir radiance measurements of Venera 13 and Venera 14 and found in both a distinct peak of extinction between 1 and 2 km that they concluded to be a near surface cloud deck or haze layer. [18] In conclusion, the atmosphere is optically thick at all visible and infrared wavelengths. Light scattering in the clouds and hazes prohibits direct or indirect surface observation in visible light. Scattering is conservative [Ragent et al., 1985] which allows a small fraction of incident sunlight to reach the surface [see, e.g., Gierasch et al., 1997]. Most

E00B17

sunlight is reflected at the cloud layer but Rayleigh scattering in the deep atmosphere is also not negligible [Sagan, 1962; Moroz, 2002]. Sunlight reflected from the surface is not significantly distinguishable against this background down to an altitude of 2 km at 0.65 mm [Moroz, 2002]. Figure 2 shows altitudes of Rayleigh scattering optical thicknesses 0.1 and 1 in the wavelengths studied in here. [19] In the infrared range beyond 2.5 mm gases and H2SO4 cloud particles become strongly absorbing, thermal emission corresponds approximately to the temperatures of the cloud tops. The average brightness temperature of Venus in this range is 220 to 250 K on both dayside and nightside [Pettit and Nicholson, 1955; Moroz et al., 1985] which corresponds to atmospheric temperature at 60 to 75 km height [Seiff et al., 1985]. In the near infrared region there are several windows between CO2 and H2O absorption bands that allow measurable thermal emission from deeper layers to escape. The emissions from these windows were first discovered by Allen and Crawford [1984] at 1.74 mm and 2.30 mm. On the basis of shape, distribution, and revolution period of the bright and dark markings they concluded the observable contrast to be due to the illumination of the variable lower cloud layer by thermal emissions of the hot matter below the clouds. Kamp et al. [1988] found the emitting matter at these wavelengths to be CO2 in the deep atmosphere. Absorption/emission increases with depth owing to higher density and pressure and temperature broadening of absorption bands. On the basis of their calculations they predicted further windows in the 1 mm region that were subsequently observed at 1.10 mm, 1.18 mm, 1.27 mm and 1.31 mm by ground-based observations [Crisp et al., 1991] and during the Galileo flyby [Carlson et al., 1991] at 1.2 mm, 1.01 mm, and 0.8 mm. Two more windows were observed by Cassini VIMS at 0.85 mm and 0.9 mm [Baines et al., 2000]. [20] In addition to the cloud like markings simultaneously seen in all window regions, in those shortward of 1.2 mm the contrast of the thermal emission of cold highlands and hot lowlands is observed [Crisp et al., 1991; Carlson et al., 1993a]. Lecacheux et al. [1993] investigated this contrast in ground-based observations of the 1.02 mm window. Using a radiative transfer model they found the contrast correlated with surface elevation to be consistent with thermal emission of the surface in thermal equilibrium with the atmosphere and less than 10% variation in emissivity. [21] One obstacle in modeling of the windows into the lower atmosphere of Venus is that little or no laboratory data exist on gaseous absorption at comparable pressures and optical paths. In spectral window regions far from absorption bands, absorption coefficients cannot fully be determined from existing absorption line databases and standard line shapes [Taylor et al., 1997]. Parameters describing the continuum absorption are often determined empirically, i.e., chosen such that modeled spectra best fit the observed spectra [see, e.g., Pollack et al., 1993; Meadows and Crisp, 1996; Marcq et al., 2006; Tsang et al., 2008]. [22] The optical properties of the clouds are better understood and consistent with Mie scattering at spheres with the refractive index of 75 wt % solution of H2SO4 [e.g., Hansen and Hovenier, 1974; Grinspoon et al., 1993; Pollack et al., 1993; Carlson et al., 1993b]. On the basis of imaging observations of the windows at 1.74 mm and 2.30 mm with

4 of 21

E00B17

MUELLER ET AL.: VENUS SURFACE THERMAL EMISSION

Galileo NIMS Carlson et al. [1993b] attributed the cloud contrast not only to varying particle number densities but also to a shift in size or modal distribution. Grinspoon et al. [1993] calculated the scattering properties of cloud particles using Mie theory. Subsequent radiative transfer modeling yielded best results to NIMS data by variation of modes 2’ and 3 in the middle and lower clouds, confirming the findings of Carlson et al. [1993b]. [23] Figure 1 shows a synthetic spectrum of the nightside of Venus based on gaseous absorption line databases and Mie scattering calculations for cloud particles as sulfuric acid droplets. The model is described by Tsang et al. [2008], with surface emissivity 0.8, surface temperature 740 K, a continuum absorption coefficient of 3.0  10
9 cm
1amagat
2, cloud model from Pollack et al. [1993], and no subcloud haze. [24] In windows shortward of 1.74 mm, cloud particles scatter almost conservatively. Single scattering albedos differ from unity less than 10
3 in magnitude [Grinspoon et al., 1993, Figure 1d]. Subsequently, cloud optical properties are approximately the same at the wavelengths of these windows. Meadows and Crisp [1996] exploit this to diminish cloud contrast in imaging observations via the ratio of surface windows (1.02, 1.10, and 1.18 mm) to the atmospheric window at 1.31 mm. At 1.31 mm the atmosphere becomes strongly absorbing between clouds and surface. Atmospheric brightness temperature can assumed to be horizontally constant below the clouds wherefore observable contrast of flux at top of atmosphere gives local transmittance of flux through the clouds. To interpret thus ‘‘declouded’’ images, Meadows and Crisp [1996] construct a synthetic image based on their model of the transfer of surface thermal emission. Temperatures were calculated from Pioneer Venus altimetry and a uniform emissivity of basalt (0.85) was chosen. The ratio of synthetic to declouded observed image showed less than 10% variation which is presented as upper limit of surface emissivity variation at 1.18 mm. There is still a slight negative correlation with the cloud contrast to be seen in the declouded image. Additionally, for the effect of topography or surface temperature to vanish in emissivity, a lapse rate with 7.5 K km
1 has to be chosen, which is different from the 8.06 K km
1 given by the VIRA model [Seiff et al., 1985]. [25] As Moroz [2002] and Hashimoto and Sugita [2003] point out, this approach employing simple ratios to account for cloud and topography contrast neglects, that emerging flux at a surface window is neither proportional to surface emissivity nor to the cloud transmittance determined from 1.31 mm because of the different lower boundary conditions. While at 1.31 mm the deep atmosphere below the clouds reflects no downwelling radiation, at 1.02 mm atmospheric absorption is small enough for multiple reflections between atmosphere and surface to be relevant for emerging flux. Because cloud reflectivity and surface albedo are negatively correlated to cloud transmittance and surface emissivity, respectively, the contrast of emerging flux that contains significant contribution from multiple reflections is diminished compared to the direct proportionality implicitly assumed by Meadows and Crisp [1996]. Hashimoto and Sugita [2003] conclude that the slight negative correlation with cloud contrast in the emissivity

E00B17

presented by Meadows and Crisp [1996] is likely due to this effect and also that actual variation of emissivity might be higher than reported by Meadows and Crisp [1996]. Hashimoto and Sugita [2003] also employ a two-stream approximation to quantify the effect of multiple reflections and to study the observability of the surface emissivity from orbit. This approximation is here adapted to parameterize the effect of the atmosphere on the emerging thermal emission at 1.02 mm. 2.3. Atmosphere Parametrization [26] Moroz [2002] and Hashimoto and Sugita [2003] give an approximation for the relation between surface thermal and emerging flux Fl at the top of a emission flux Ftherm l highly reflective atmosphere Fl ¼

t F therm 1
al r l

ð1Þ

where t is transmission of the atmosphere, r is reflectivity of atmosphere and al is lower boundary albedo. Equation (1) is exact for plane parallel conservative atmospheres which is approximately true for atmospheric layers composing the clouds at the window wavelengths and it is t = 1-r. Hashimoto and Sugita [2003] assume that cloud reflectivity r is the same for both surface window and the atmospheric window at 1.31 mm. therm is constant because the atmosphere becomes [27] F1.31mm opaque owing to absorption and emits in local thermal equilibrium with the constant atmospheric temperature. Brightness of the thermal emission from radiative transfer modeling [Tsang et al., 2008] gives a rough approximation for the height of the transition from a conservative to absorbing atmosphere and thus for the source region of radiation (Figure 2), for a better estimate see the contribution functions by Tsang et al. [2008]. The isoline of t r = 0.1 in Figure 2 illustrates that Rayleigh scattering is not very important in the source region of the 1.31 mm window at approximately 20 km height. Little downwelling radiation is scattered upward in the source region; this allows us to assume a1.31mm = 0 [Hashimoto and Sugita, 2003]. Equation (1) then becomes r ¼1


F1:31mm therm F1:31mm

ð2Þ

and allows us to calculate to calculate the atmospheric reflectivity r above the source region of the 1.31 mm window. [28] Cloud reflectivity r contains all required information on the variable cloud transmittance and can be used to remove cloud contrast at 1.02 mm with equation (1). Cloud reflectivity r does not account for scattering below the source region of the 1.31 mm window, however there is significant scattering at 1.02 mm below 35 km due to Rayleigh scattering (Figure 2) and possibly due to aerosols [Moroz, 2002; Grieger et al., 2004]. Additionally, there is some small but poorly constrained amount of gaseous absorption due to the high pressure. Therefore, the paramdo not directly represent the surface in eters al and Ftherm l the frame of equations (1) and (2), but rather a composite of surface and lowest atmosphere that cannot be disentangled

5 of 21

MUELLER ET AL.: VENUS SURFACE THERMAL EMISSION

E00B17

E00B17

Figure 7 shows that a constant al is consistent with the data and a certain fit of Tl(Z). [30] Equation (3) relates the observable flux Fl to three parameters, upper atmosphere reflectivity r, lowest atmosphere and surface reflectivity al, and temperature parameter Tl(Z), none of which is directly indicative of surface composition. Since surface composition variability is the aim of this study, equation (3) has to be interpreted in light of the implicit assumption of no surface emissivity variability. Any local deviation of VIRTIS data from the solution of equation (3) with empirically found values for the three unknown parameters is related either to local variations in the properties of the atmosphere or, since there is yet no evidence for such variations, to surface properties. To quantify this deviation, VIRTIS data are introduced as Fl and thermal flux anomaly Al is defined as deviation from unity of the ratio of both sides of equation (3) Al ¼ 1


Figure 3. Sketch illustrating the atmosphere parametrization. Layer 1 is the lowest part of the atmosphere, from the surface to 35 km altitude. Its average behavior, including the surface, is described empirically. Source of thermal radiation is assumed to be a blackbody emitting flux p Bl[Tl(Z)]; scattering in the lowest atmosphere is represented by al. Layer 2 is the atmosphere from 35 km to the top of the atmosphere, which is approximately conservative and grey. This allows us to account for the cloud contrast at 1.02 mm using the VIRTIS images at 1.31 mm to determine cloud reflectivity r. Deviation of data from top of atmosphere flux derived using this model may indicate surface temperature or emissivity anomalies. easily. Instead, global average behavior of these two parameters is determined empirically from VIRTIS data in section 4.4. [29] For a better understanding of the parameters al and , consider an additional atmospheric layer (Figure 3, Ftherm l layer 1) below the part of the atmosphere that is spectrally grey and conservative (Figure 3, layer 2). There is no evidence for horizontal variation in optical properties other than surface topography Z in this layer. Surface topography affects extinction optical depth but also determines average surface temperature. For sake of simplicity, the surface is represented by a blackbody with hemispherically integrated thermal emission pBl[Tl(Z)], with a temperature term Tl(Z) that encompasses all brightness variations due to topography. Topography-independent scattering is represented by al. When conservative scattering is assumed, this allows us to rewrite equation (1): Fl ¼

ð1
rÞ ð1
al ÞpBl ½Tl ðZ Þ 1
al r

ð3Þ

Tl(Z) is an approximation for surface temperature. This approximation deviates from real surface temperature owing to absorption, multiple reflections within the lowest atmosphere, and the dependance of optical depth on topography Z. In reality, al will depend on surface topography too, but

Fl ð1
al rÞ ð1
rÞð1
al ÞpBl ½Tl ðZ Þ

ð4Þ

Surface emissivity is positively correlated with Al, but surface temperature variation not expressed by Tl(Z) also affects it.

3. Surface Observations by VIRTIS on Venus Express [31] The VIRTIS instrument on Venus Express is a flight spare of the instrument of the same name of the Rosetta mission [Drossart et al., 2007]. The versatility of the instrument originally designed to observe coma and nucleus of a comet at moderate spectral and spatial resolution [Coradini et al., 1998] allows imaging of all NIR windows longward of 1 mm wavelength from Venus orbit. The infrared mapping subsystem VIRTIS-M IR that acquired the data for this study diffracts a light bundle selected by a slit with on a detector array. Data sets acquired at a given time (frames) therefore have one spatial and one spectral dimension, here referred to as samples and bands with index b. The detector array has 256 samples and 431 bands, in total giving a field of view of 64 mrad, corresponding to approximately a third of the diameter of Venus at apocenter, and a spectral range from approximately 1 mm to 5 mm. A scanning mirror with 256 position allows the instrument to construct hyperspectral image cubes by appending frames acquired consecutively at different viewing angles, thus adding a third dimension of lines. Neither mission nor instrument are specifically designed to observe the thermal emission of the surface but the manifold objectives of the mission requires different viewing geometries and integration times [Svedhem et al., 2007]. The signals of the surface temperature and moreover of surface emissivity are small compared to the reflected sunlight on the dayside and require several seconds of exposure on the nightside to gain sufficient signal-to-noise ratio (SNR). Additionally a correction for variable cloud opacity such as attempted here is an additional source of noise that is especially critical since radiance and SNR at 1.31 mm is even lower. The error in cloud reflectivity derived from 1.31 mm is a major contribution to overall uncertainty in retrieval of surface properties [Hashimoto and Sugita, 2003]. There is an atmospheric

6 of 21

E00B17

MUELLER ET AL.: VENUS SURFACE THERMAL EMISSION

Figure 4. VIRTIS data spectrum from cube VI0112_01, line 0, sample 0. Dotted line is stray light spectrum Sb scaled to fit radiance I½36;39 . Stray light is removed by subtraction of individually scaled stray light spectra from all data spectra, similar to the approach by Meadows and Crisp [1996]. window at 1.27 mm with comparable properties and better SNR. Unfortunately it is coincident with oxygen airglow in the upper atmosphere [Crisp et al., 1996] and unsuitable for a straightforward determination of cloud transmittance. The long duration of the mission and repeated observations of the same area of surface gives the unprecedent opportunity to combine observations over a long period of time to enhance the time invariant signal of the surface and to offset the poor SNR. [32] The 24 h eccentrical polar orbit with pericenter at approximately 80°N exclusively allows for imaging of the southern hemisphere. Science planning of the mission is described by Titov et al. [2006]. Observation schemes are designated as science cases. All relevant cases are briefly summarized here: The northern hemisphere is only accessible in case 1 observations that are composed of isolated lines of spectra perpendicular to the spacecrafts track. Case 3 observations are apocenter mosaics of either the entire visible southern hemisphere or alternatively of either dayside or nightside. Owing to the eccentrical polar orbit these observations show the equatorial regions only at very high emission angles and accordingly low spatial resolution. Case 2 observations are done on the ascending or descending branch of the orbit and therefore able to observe regions up to 30°S with fairly low emergence. During most observations, especially case 3, nadir is at low latitudes and on average observed latitude is correlated to emission angle. At low latitudes the terminator is in or close to the field of view and observations are affected by sunlight scattered in the upper atmosphere and instrumental stray light originating from the bright side of Venus. In addition to the viewing geometry affecting the measured radiance, there is a shift from observation to observation in band positions due to varying thermal strain in the instrument.

4. Data Processing

E00B17

1.0255 mm are processed. A uniform
7.5 nm shift is applied to the wavelengths table in the VIRTIS data cubes to better reconcile data with synthetic spectra. Parameters empirically derived during processing either have the subscript b, giving band number as integer starting at zero, or l giving band position in micrometer. The first pertains to all images of that band regardless of individual band positions while l denotes that the parameter in question is seen as a function of individual band position wavelength of individual observations. [34] Several different VIRTIS bands are situated inside the window regions. The band with the best ratio of surface to atmospheric influence is b = 0, on average at l = 1.021 mm. The band closest to the center of the 1.31 mm is at b = 30 on average at l = 1.307 mm. Analysis of VIRTIS data is performed only for band 0 for the surface signal and band 30 for the cloud signal, although the data processing mechanism may as well be extended to other windows and bands farther from the center (e.g., Figure 4). It is possible to use the average of several or all bands showing a signal from the surface with this approach. This would increase SNR for individual observations and help in the detection of transient temperature variations, e.g., thermal signature of active volcanism. The contrast related to surface emissivity, however, is decreased in the averaged image as the transparency of the atmosphere is greatest in the small wavelength range used here. [35] Only observations with 3.3 s exposure duration or greater have a sufficient SNR ratio. In total 1161 case 2 and 3 observations have been selected for processing. Observations span a period from May 2006 to December 2007. Some observations contain frames, where all spectra of one line are offset relative to the neighboring frames. These frames are not processed and appear as horizontal lines of missing data in VIRTIS images (Figure 9). 4.1. Scattered Sunlight [36] VIRTIS image cubes on the nightside are affected to some extend by sunlight. While sunlight scattered by the upper atmosphere extends to at least 95° of incidence, the bright illuminated crescent of Venus close to the field of view of VIRTIS also causes stray light within the instrument. The same is true for direct sunlight, which is however, due to the geometry of the orbit, less relevant for the case 2 and 3 observations than for case 1 [Arnold et al., 2008]. Meadows and Crisp [1996] use a spectrum extracted from the sunlight crescent as template for the removal of stray light from the 1 mm region. The spectrum is scaled to fit the radiance at the 1.4 mm CO2 absorption band for each of their spectra and then subtracted. [37] In a similar approach here the spectral average of the VIRTIS bands 36 to 39 is used as parameter I½36;39 for the spatial distribution of stray light. To determine the spectrum of stray light, a linear regression is performed over all VIRTIS observations for the radiances of VIRTIS bands 0 to 35 against this parameter assuming a direct proportionality between the stray light seen at I½36;39 and the radiance of other bands Ib. The linear regression retrieves the best fitting parameters a to the equation

[ 33 ] To minimize the influence of the wavelength shift due to instrument temperature, only observations with the first band situated between 1.0175 mm and 7 of 21

Ib ¼ Vb þ Sb I½36;39

ð5Þ

E00B17

MUELLER ET AL.: VENUS SURFACE THERMAL EMISSION

E00B17

4.2. Limb Darkening [39] Limb darkening due to scattering in the upper clouds affects all VIRTIS images, most severely in off-nadir case 2 and in case 3 southern hemisphere mosaics. The limb darkening of synthetic spectra agrees well with VIRTIS radiances (Figure 5) and shows that the angular distribution of emitted radiance I0b does not significantly vary between the 1.02 mm and the 1.31 mm window. The angular distribution of radiance with respect to cosine of emission angle x can be described as product of nadir radiance I0b(1) and limb darkening function P(x). Ib0 ðx Þ ¼ Ib0 ð1ÞPðxÞ

ð7Þ

Required for the two stream approximation employed here is hemispherically integrated flux [Goody and Yung, 1989] Fb

¼ 2p

R1

0 0 Ib ðx Þxdx

¼ 2pIb0 ð1Þ

R1 0

ð8Þ

PðxÞxdx

To account for viewing angles, the radiance I0b(x) measured at each pixel is divided by the limb darkening function P(x) determined by interpolation of synthetic spectra at different emission angles scaled to unity at nadir (x = 1), see solid line in Figure 5. The angular integration is performed by approximating P(x) = 0.31 + 0.69x which in conclusion yields Figure 5. Distribution of surface window radiance I0, atmospheric window radiance I30, and corresponding cosines of emission angles x; only data points with negligible stray light I½36;39  0.005 W m
2sr
1 mm
1 are considered. Frequency distribution is scaled to yield one at maximum. Solid line is created from synthetic spectra of the nightside at different emission angles, and dashed lines are function P(x) = 0.31 + 0.69x scaled to fit synthetic spectra at nadir. This separates the flux Ib spectra into two terms, first thermal emission of Venus Vb which is not correlated to flux at bands 36 to 39 but represents average thermal emission, and second stray light spectrum Sb as the coefficients of proportionality of Ib to I[36,39]. The linear relation extends to a value of approximately I[36,39] = 0.01 W m
2mm
1sr
1 and only points in this range are used for the regression and further processing. [38] Comparison of VIRTIS spectra with the stray light spectrum Sb scaled to fit radiance at bands 36 to 39 demonstrates that there is a good fit at absorption bands while at window wavelengths, stray light is slightly increased above the lower envelope of the VIRTIS data spectrum, see Figure 4. Individually scaled stray light spectra are subtracted from each data spectrum to isolate the local flux of thermal emission: Ib0 ¼ Ib
Sb I½36;39

ð6Þ

The values relevant for data processed here are S0 = 3.30 and S30 = 1.32.

Fb ¼ p0:77

Ib0 ðxÞ Pðx Þ

ð9Þ

4.3. Projection and Smoothing of Magellan Altimetry [ 40 ] The Magellan global topography data record (GTDR) from the planetary data system (PDS) gives topography of the surface of Venus [Ford and Pettengill, 1992]. For the VIRTIS observations, the data are projected on the VIRTIS field of view and incorporated in the ancillary geometry data files distributed together with VIRTIS hyperspectral cubes. Since the thermal emission of the lower atmosphere and surface is scattered in the clouds, the radiance observed by VIRTIS originates in the cloud layer illuminated by the surface and atmosphere below. This has two consequences: First, the altitude relevant to a pixel is not given by the intercept of the line of sight (LOS) with the surface, but by the altitude directly below the intercept of the LOS with the cloud layer. Second, not only the point directly below LOS cloud intercept but the neighboring area contributes to illumination of the clouds at this point. This results in a smearing of the image of the surface. Simulations by Hashimoto and Imamura [2001] describe this effect as photons from a point source on the surface scattered into a gaussian distribution with full wide half maximum (FWHM) of 90 km above cloud level. Moroz [2002] approximates a spatial resolution of 100 to 200 km. [41] To simulate this effect, the Magellan altimetry in the VIRTIS geometry is projected according to VIRTIS image viewing geometry to a height of 65 km corresponding to the upper main cloud deck. The spatial resolution of VIRTIS images varies on the southern hemisphere between approx-

8 of 21

E00B17

MUELLER ET AL.: VENUS SURFACE THERMAL EMISSION

E00B17

0.04%. In some regions, especially rifts, the difference is higher but does not exceed 2%. No strong dependance of the difference to surface altitude has been found, the error in the determination of global brightness temperature is assumed to be small. The width of the smoothing weighting function is likely a more critical parameter but no study of its influence has been made yet.

Figure 6. VIRTIS fluxes F30 binned according to band position l with 1s uncertainty. Solid line is synthetic flux spectrum from plot 1 reduced in spectral resolution by convolution with a gaussian with full wide half maximum (FWHM) of 18 nm and scaled with a factor of 1.36. Assuming cloud transmittance of 0.18 [Hashimoto and Sugita, 2003], this spectrum is used for approximation of spectral dependance of T30. imately 10 and 60 km. The spatial resolution of the projected altimetry has to be aligned with the spatial resolution of the thermal emission flux penetrating the atmosphere. To achieve this task, a moving weighted average is applied. The weighting function is a gaussian with respect to distance from pixel center on the surface and has a FWHM of 100 km. There are some complications to the problem of connecting thermal emission flux and altimetry. [42] First, flux is calculated by angular integration of the blackbody radiation. The hemispherical integration through the factor of p is only exact for plane surfaces, here no attempt has been made to account for curvature of the topography. Second and more important, thermal emission radiance is not linear to surface altitude. Because thermal emission varies strongly with temperature, low altimetry areas contribute unproportionately more to the brightness of a region than higher areas in the same region. On the other side atmospheric transmittance at highland regions is higher. Averaging over altimetry instead of thermal flux leads to an error depending on distribution of altimetry values within the smoothing radius; it is most severe in regions with high slopes and altitude differences. [43] To minimize this error without detailed simulations of radiative transfer, the smoothing algorithm is used in two different ways. For the determination of global atmosphere and surface parameters (section 4.4), the moving average is applied directly to altimetry. Thus, smoothed Magellan altimetry is denoted by Z in units of meters relative to mean planetary radius (MPR) of 6051.84 km [Ford and Pettengill, 1992]. For the calculation of local flux anomaly (section 4.5), the thermal emission flux is calculated on the basis of global brightness temperature to altimetry relation 8 and the original Magellan altimetry projected on the VIRTIS field of view. The moving average is then applied to the thermal flux. [44] The difference between flux calculated from the smoothed altimetry, and the smoothed flux is on average

4.4. Atmospheric Parameter Determination [45] Equation (3) relates emerging flux at surface window to three parameters: r, al, and Tl(Z). Typical cloud reflectance of r = 0.82 based on a nominal cloud model is given by Hashimoto and Imamura [2001]. Adopting this value in = 0.77pBl[Tl] leads to equation (2) and assuming of Ftherm l a temperature of the atmosphere below the clouds Tl. Shape of the synthetic spectrum (Figure 1) is used to fit spectral dependance of Tl to average VIRTIS flux at different wavelengths (Figure 6). The resulting range of T30 from 528 K to 531 K approximately corresponds to VIRA atmospheric temperature at 26 km height for all wavelengths covered by band 30. Flux F30 measured with VIRTIS band 30 and estimated T30 allow to calculate local cloud reflectivity r with equation (2). [46] The two remaining parameters are estimated from VIRTIS and Magellan altimetry data. While atmospheric temperature at the mean planetary radius of 735 K and surface emissivity e0 > 0.8 are relatively well constrained by in situ measurements, adopting these values as e0 = 1
a0 and T0(0 km) = 735 K does not work well because lowest atmosphere scattering and absorption is not accounted for. Albedo a0 is expected to be greater than 0.15, which Meadows and Crisp [1996] used to represent the basaltic plains of the surface. Temperature T0 with respect to equation (3) will be less than real surface temperature since the lowest atmosphere is not fully transparent. [47] To estimate a0, the distribution data points in the F0, F30, and Z space is determined using all VIRTIS observations. Average value of F0 in each F30 interval is given with 0.5s error bars for different topography intervals in Figure 7. The position on the F30 axis translates with equation (2) into cloud reflectivity r. Figure 7 demonstrates that it is possible to reproduce the relation of observed fluxes F0 and F30 within 0.5s with a lowest atmosphere and surface albedo a0 = 0.22 independent from topography and temperature T0(Z) as monotonic function of altitude. Variation of a0 with band position l has not been investigated. Preliminary analysis of the windows at 1.10 and 1.18 mm not presented here yields similar values of lowest atmosphere albedos, which leads to the conclusion that this parameter does not vary strongly with wavelength. [48] Using this constant, a0, equation (3) is solved for T0 for each data point. Next, the distribution of T0 values in the topography Z and band position l space is determined from all processed data points. A two dimensional polynomial of degree 2 with coefficients knm and parameters Z and l is fitted to the averages of T0 to obtain an analytical approximation of Tl(Z):

9 of 21

Tl ðZ Þ ¼

2;2 X i;j¼0;0

kij Z j li

ð10Þ

MUELLER ET AL.: VENUS SURFACE THERMAL EMISSION

E00B17

E00B17

the following conditions are met: (1) emission angle is less than 80°, (2) incidence angle is greater than 95°, and (3) sunlight parameter I½36;39 is less than 0.01 W m
2mm
1sr
1. [51] Two different quantities are retrieved, either surface temperature T0 or the thermal flux anomaly A0 which is thermal emission corrected for the global relation of surface temperature Tl(Z) to altimetry and band position. Figure 9 illustrates the process of retrieving surface and lowest atmosphere brightness temperature. Both results are map projected for each observation using Lambert’s azimuthal equal area projection centered on the south pole. The projected individual images are then combined by using the median of all different observations of the same place. The spatial distribution of processed images is shown in Figure 10.

5. Results Figure 7. Average of surface window flux for different altitudes as a function of cloud reflectivity r. Flux uncertainty is given as 0.5s. Size of topography bins is 500 m. Curves are calculated with equation (3) with a0 = 0.22 for all heights and T0(
1 km) = 712 K, T0(0 km) = 708 K, T0(1 km) = 703 K, and T0(2 km) = 698 K. Figure 7 shows that equation (3) describes the relation between surface and atmosphere window flux well and thus can be used for removal of cloud contrast. The i  j matrix of coefficients is 0 B B kij ¼ B B @

63:6  103


12:3  103

60:3  103

13:0


25:6

12:5


1:83  10
3

3:62  10
3


1:79  10
3

1 C C C ð11Þ C A

Units of kij are m
jmm
i K where i is index of column and j is index of row starting with zero. [49] The Z-l isolines determined directly from the data and from the fit of equations (10) and (11) are illustrated in Figure 8 for seven temperatures. The lapse rate of
4 to
5 K km
1 is not consistent with the
8.06 K km
1 in the VIRA model but this is not surprising since scattering, absorption and emission in the atmosphere lowers the lapse rate of observed flux. Furthermore scattering and absorption optical path from surface to top of atmosphere is determined by topography. Usually this is negligible, as topography is only a fraction of total optical path lengths, but extinction is here strongly concentrated toward the lowest part of the atmosphere owing to Rayleigh scattering being proportional to density and the continuum absorption being proportional to square of density [Pollack et al., 1993]. 4.5. Image Processing [50] After the parameters necessary for atmospheric correction are either assumed or determined from the statistical behavior of VIRTIS data points, the next step is the application of equation (3) on each observation to analyze the local thermal flux. Data pixels are processed only when

[52] Surface temperature parameter T0 averaged over time is highly correlated with altimetry in Figure 11. Signal to noise is better in the hemisphere west of 0° longitude as is expected owing to greater coverage (Figure 10). SNR generally decreases toward the limb, where less images as well as less spectra per unit area have been acquired owing to the generally oblique viewing angles. The polar region is usually seen at low emission angles but stray sunlight increases the error although spectra surpassing a certain threshold of stray light are discarded during processing. [53] The map of flux anomaly Figure 12 presents in principle the same data as Figure 11 but corrected for the global temperature to topography relation (Figure 8). There are however some correlations with topography remaining when comparing the altimetry contours in Figure 11 with flux anomaly in Figure 12. There is lower flux in the sector between
120° and
110° longitude starting at about 75° to 60°S latitude and extending at least to the equator. This coincides with an increased uncertainty in the Magellan altimetry investigated by Rappaport et al. [1999]. Several rifts, e.g., Artemis Chasma in V-48, show contrast in the flux anomaly map. This might be related to the difficulties in smoothing of the radar altimetry (section 4.3). The error

Figure 8. Relation of lower atmosphere surface brightness temperature T0 (for seven different values of T0) to topography Z and band position l. Solid lines are isolines of average data temperature. Dashed lines are polynomial fit to data.

10 of 21

E00B17

MUELLER ET AL.: VENUS SURFACE THERMAL EMISSION

E00B17

Figure 9. VIRTIS image VI0373_01 (second image acquired in orbit 373) during several processing steps: (a) band 0 at 1.02405 mm; (b) spectral median of bands 36 to 39 showing distributions of stray light; (c) band 0 corrected for stray light and limb darkening; (d) band 30 corrected for stray light and limb darkening; (e) brightness temperature of lower atmosphere and surface derived from band 0 and band 30; and (f) surface temperature map constructed from Magellan altimetry and Tl(Z). Horizontal dark lines are data points not processed owing to an offset of the whole frame. due to the smoothing is expected to be greatest in regions with steep slopes and large altitude differences. [54] The equatorial regions generally show a greater residual correlation of flux and topography. This might be indicative of a latitudinal component in the flux to topography relation, which might be due to latitudinal temperature or atmospheric composition variation. Most data points used for determination of the global flux to topography relation are in the midlatitude to low-latitude regions between 25° and 75°S where influence of topography seems to be well accounted for (compare Figure 10 and Figure 12). Spectrometer temperature and thus band position l varies

with orbital position. The equatorial regions have a bias to be observed at wavelengths displaced from the optimum. Determination of Tl(Z) is therefore possibly less reliable for the equatorial regions. The lower latitudes are, with some exceptions that are described in sections 5.1 – 5.3, very well represented by the global flux to topography relations as the flux anomaly is very near to zero. 5.1. In Situ Sites [55] There are four in situ measurements of surface rock composition on the southern hemisphere, see Figure 12. The Venera 8 landing site in quadrangle V-43 is in a region with

11 of 21

E00B17

MUELLER ET AL.: VENUS SURFACE THERMAL EMISSION

Figure 10. Spatial distribution of processed images in Lamberts azimuthal equal area projection centered on the south pole and extended to the equator; labels denote Venus mapping quadrangles. average to low thermal flux. Venera 13 in quadrangle V-42 is in the region with low flux adjacent to Doliya tessera, Venera 14 is on the gradual transition from this same diffuse low flux area to the diffuse high flux area over Navka planitia. Vega 2 landed on Rusalka planitia in quadrangle V37 which shows an unreliably determined flux. All landing sites are in the equatorial regions and not very well covered by VIRTIS observations. Possible interpretations of rock types are summarized on the basis of the work of Kargel et al. [1993]. Venera 8 measured abundances of natural radioactive isotopes of K, U, and Th, which where consistent with several terrestrial analogues, including rhyolite (volcanic rock of felsic composition), monzonite (intermediary composition) and leucitite, an on Earth rare alkaline mafic volcanic rock in which felsic minerals are mostly replaced by the feldspathoid mineral leucite. Leucitite is also proposed for the Venera 13 landing site, where elemental abundances were more directly measured by X-ray fluorescence spectroscopy. The Venera 14 and Vega 2 samples also analyzed with XRF spectroscopy are interpreted to have basaltic composition similar to mid-ocean ridge basalts. Though it might be tempting to attribute areas of low flux anomaly to the exceptional samples of Venera 8 and Venera 13 situated in the vicinity, very little difference is seen in NIR flux anomaly between the Venera 8, Venera 13, and Venera 14 sites. 5.2. Relation to Geological Settings [56] The thermal flux anomaly shows some correlation with geomorphological terrain types and also individual features. Over large areas of tessera terrain the flux adjusted for topography is generally lower than in neighboring areas. The tessera terrain outlines in Figure 12 are taken from the

E00B17

electronic global geological map accompanying the work of Tanaka et al. [1997]. Large tessera terrains with relative good SNR are Phoebe regio and Doliya tessera in V-41, Alpha regio in (among others) V-32, Lhamo and Cocomama tessera in V-56. Tessera boundaries are however not well defined in NIR flux, negative flux anomalies are interspersed with average flux and occasional positive anomalies. For instance, at Doliya tessera no relation of its boundaries to flux is obvious, but instead the tessera area is situated directly on the boundary between larger areas of positive and negative flux anomalies. [57] The negative flux anomaly over tessera regions is consistent with the relative low surface emissivity at highlands retrieved by Hashimoto et al. [2008] from the Galileo NIMS flyby data. Tessera are more frequent in highlands [Ivanov and Head, 1996]. Their approach does not correct for topography empirically but uses a flux to topography relation derived from the adiabatic lapse rate and radiative transfer modeling and is therefore not directly comparable to this result. [58] Tessera are not extensively covered by VIRTIS observations; there is very little data yet of the large tessera highlands in Aphrodite terra. Most VIRTIS spectra were acquired on midlatitudes to low latitudes <25°S between 180° and 360°E longitude, dominated by tectonically modified lowland plains. Highlands in this area are Imdr, Ishkus, Thetis, and Dione regios where some, but not all, large volcanic edifices show a flux exceeding average flux at comparable altitude by 5%, listed in Table 1. Two coronae are also listed. The area of these flux anomalies is significantly smaller than the topographic rise and concentrated on the flank. [59] A similar but larger anomaly is found at the southern flank of the Quetzalpetlatl corona rise in Lada terra, 0°E 70°S. The southern boundary of this anomaly is not clearly defined as it borders on the polar area with small SNR ratio. The fan shape nevertheless correlates with the large lava flows Juturna and Cavilaca fluctus, extending from the rim of Boala corona, nested inside Quetzalpetlatl corona. For most recent radar imaging see the work by Kratter et al. [2007]. The contours of digitate plains units from the electronic material accompanying Tanaka et al. [1997] are outlined in Figure 13. Ivanov and Head [2006] characterized the unit containing the positive flux anomaly south of Quetzalpetlatl as tectonically undeformed lobate plains. Lavinia planitia in V-55 shows an overall increased flux compared to other planitia regions at altitudes below mean planetary radius and especially prominent at the eastern basin rim. This basin rim is characterized by several lobate lava flows and plains (Figure 13), emanating from coronae, volcanoes or the Lada rift [Magee and Head, 1995]. These units are designed stratigraphically young, undeformed lobate plains in the map by Ivanov and Head [2001]. The spatial correlation is, however, less significant than at the lava streams south of Quetzalpetlatl and the anomaly does not exceed 10%. For a more detailed analysis of the NIR flux anomaly at Quetzalpetlatl see Helbert et al. [2008]. 5.3. High-Altitude Radiothermal Anomaly [60] Radiothermal emissivity as measure by the Magellan mission [Pettengill et al., 1992] distinctly drops off from a value around 0.85 to a value of 0.4 above surface elevations

12 of 21

E00B17

MUELLER ET AL.: VENUS SURFACE THERMAL EMISSION

E00B17

Figure 11. Surface lower atmosphere brightness temperature T0 mosaics in Lamberts azimuthal equal area projection centered on the south pole. Magellan altimetry is plotted with 1 km contours. Black hachures denote missing altimetry data. Temperature is well correlated with altimetry, although influence of wavelength shift is not equal in all regions. of 4 to 5 km [Wood, 1997]. This height is rarely reached in the area mapped by VIRTIS with good SNR. Areas with radiothermal emissivity less than 0.7 are marked on the flux anomaly map in Figure 12 by dark hachures from up left to down right. A single region with a radiothermal emissivity anomaly, quadrangle V-41, Yunya-mana mons (285°E, 18°S) is mapped with reasonable SNR and shows no significant flux anomaly in NIR. However, the extent of this area is close to spatial resolution of the NIR data and observed flux might be influenced by the neighboring tessera terrain. The Venus Monitoring Camera (VMC) also on Venus Express did not observe resolvable emissivity variations in this region (Basilevsky et al., submitted manuscript, 2008). [61] Repeated coverage of large regions showing a radiothermal emissivity, e.g., Thetis regio in V-36, is required to analyze the radiothermal anomaly in NIR. The approach used here is however not very well suited for this investi-

gation. The influence of any present altitude-dependent variation of emissivity will be to some extent removed during the processing owing to the empirical determination of Tl(Z). A nonempirical approach that correctly accounts for absorption and scattering independent from VIRTIS data in the lowest atmosphere, as in the work of Hashimoto et al. [2008] or Arnold et al. [2008], is required. Data at higher latitudes and altitudes, e.g., Maxwell Montes are also desirable, but will not be available in observations other than case 1 northern hemisphere sparse imaging.

6. Discussion [62] Interpretation of the thermal flux anomaly is highly ambiguous. First, a flux anomaly due to surface emissivity cannot be distinguished from surface temperature effects with this approach. The long period of one and one half year over which the data are averaged and the properties of the

13 of 21

E00B17

MUELLER ET AL.: VENUS SURFACE THERMAL EMISSION

E00B17

Figure 12. Map showing flux anomaly A0 in relation to in situ measurements and tessera terrain. Landing site coordinates of Venera and Vega probes are from Basilevsky and Head [2003]. Tessera terrain outlines are from the electronic material accompanying the article by Tanaka et al. [1997]. Relation to in situ measurements is inconclusive. Tessera show a negative flux anomaly. Flux over tessera terrain is less than in plains and volcanic rises of the same Magellan altimetry values. Only areas with a coverage of three or more VIRTIS images are mapped here. lower atmosphere of Venus (see section 2.1) preclude, to first approximation, local surface temperature variation due to atmospheric or insolation effects. Endogeneous heat, e.g., active volcanic vents or even lava flows, could also be a source of thermal NIR flux and the coincidence of positive flux anomalies and relatively young volcanic features presents this as an attractive explanation. However, the recent rate of extrusive volcanism is comparable to intraplate volcanism on Earth [see Grimm and Hess, 1997, and references therein]. Cooling heat flux of liquid lava on the surface is slightly lower than on Earth [Snyder, 2002], but cooling time scale of the lava is estimated to be in the order of 1 day; that is, an eruption is not detectable for more than 1 day after its end [Hashimoto and Imamura, 2001]. Areas of increased flux are up to the order of hundreds of

Table 1. Volcanic Edifices Correlated With Positive Flux Anomalya Diameter (km) Name Mertseger mons Shiwanokia corona Shulamite corona Idunn mons Hathor mons Ininni mons

Coordinates 270°E, 278°E, 284°E, 215°E, 324°E, 328°E,

37°S 42°S 39°S 46°S 38°S 34°S

Anomaly

Structure

125 200 200 150 250 125

450 500 275 250 333 339

a Names, coordinates, and diameters of volcanic edifices are from the Web page of the USGS Astrogeology nomenclature database (available at http://planetarynames.wr.usgs.gov/). Two coronae are included in this compilation; their flux anomaly is located on the flank of the topographic rise but outside the annulus.

14 of 21

E00B17

MUELLER ET AL.: VENUS SURFACE THERMAL EMISSION

E00B17

Figure 13. Map showing flux anomaly A0 in relation to plains with a flow-like morphology and to the radiothermal emissivity anomaly in high regions. Digitate plains outlines from the material accompanying Tanaka et al. [1997]. Correlation of radiothermal end NIR emissivity is inconclusive owing to the insufficient coverage and the nature of the topographic correction. Most prominent coincidence of positive flux anomaly and digitate plains is on the southern flank of Quetzalpetlatl corona, 0°E and 70°S. This plains unit is composed of the large lava flow fields fan shaped extending from Boala corona on top of the topographic rise [Ivanov and Head, 2006]. thousands of square kilometers, temporal sampling rate is at most one or two images per day, typically much less. The map is a temporal median of up to several tens of images, and therefore filters the effects of short or even singular events with significantly increased temperatures. All individual observations (i.e., the result presented in Figure 9) have been searched for clear evidence of an active volcanic eruption but none has been found yet. Volcanic vent activity can be continuous but is also difficult to reconcile with the extend of the flux anomaly. Owing to these considerations, here flux anomaly is interpreted in terms of surface emissivity. [63] Second, interpretative ambiguity is that variation of surface emissivity known at only one wavelength is difficult to assign to a certain surface material. As only one point in

the surface spectrum is spatially mapped, the emissivity is interpreted as a result of relative ratio of felsic to mafic mineral abundances. Felsic minerals (i.e., feldspars and quartz) generally have a low emissivity at 1 mm while mafic (Mg- and Fe-rich) minerals tend to high emissivities, which would allow to distinguish between felsic granite and mafic basalt [Hashimoto and Sugita, 2003]. The poorly known chemical weathering environment complicates the interpretation of surface emissivity with respect to composition of terrestrial analogues. Minerals affecting emissivity such as pyrite or magnetite might be unstable depending on surface temperature and atmospheric redox state [Fegley et al., 1997; Hashimoto and Abe, 2005]. Wood [1997] relates the high-altitude radiothermal emissivity anomaly to volatile

15 of 21

E00B17

MUELLER ET AL.: VENUS SURFACE THERMAL EMISSION

transport of magnetite from the hot lowlands to the mountaintops. This would affect NIR emissivity as well but no evidence has been found by Lecacheux et al. [1993], Meadows and Crisp [1996] or Basilevsky et al. (submitted manuscript, 2008). Owing to the empirical nature of this approach, it is very difficult to investigate any present altitude-dependent emissivity trend, though a comparison of VIRTIS data of comparable high altitudes with and without radiothermal anomaly would be very interesting but difficult owing to the orbit of Venus Express. [64] Weathering age might affect surface emissivity independently from altitude. If emissivity constantly decreases with time, e.g., by slow and steady volatile transport of magnetite from the lowlands to the highlands above 4 km height [Wood, 1997], the observed thermal flux anomaly will denote differences in surface age. Increased thermal flux is more frequently found at tectonically undeformed and probably younger areas as opposed to the lower flux over the stratigraphically old and deformed tessera. However, the small thickness of material contributing to NIR surface emissivity dictates that the weathering effects in emissivity take place relatively rapid and then reach an equilibrium state. The large areas with increased emissivity would require that weathering time scale is not small against volcanic resurfacing time scale. From a very general and preliminary view on stratigraphic age, a weathering process that lowers emissivity over time scales of the average age of the surface of about 0.5 Ga is consistent with observed flux anomalies. Weathering should not be disregarded as possible explanation for the flux anomalies. Further hypotheses, as to which kind of surface material regardless of any weathering effects might be responsible for observed flux anomalies, are presented in sections 6.1 and 6.2, and their plausibility is discussed. 6.1. Relative Felsic Composition of Tessera Terrains [65] In the plains, a basaltic composition is supported by the low viscosity inferred from radar images and in situ measurements. No direct morphological evidence for composition and no in situ measurement exists for tessera terrain, which is characterized by complex tectonic deformation patterns. The crater frequency of tessera terrain is not distinctly different from that of plains, but this does not necessarily imply same age since tessera terrain might have undergone tectonic deformation that reset the crater history. At most boundaries tessera terrain is embayed by neighboring plains and therefore stratigraphically older as well as elevated in topography when visible. It more frequently occurs at higher surface altitudes [Ivanov and Head, 1996]. Several plateau shaped highland regions are composed mostly of tessera terrain. These highlands generally show a small geoid to topography ratio which implies shallow depth of compensation due to either density or thickness variations of the crust [Smrekar and Phillips, 1991]. [66] On the basis of these properties, tessera may represent the granitic cratons that Taylor and Campbell [1983] hypothesized to exist on Venus. If tessera terrain were indeed the venusian analogue to archean continental nuclei on Earth the implications would be far-reaching. Under current dry surface conditions and stagnant lid tectonic regime creation of silicic (i.e., granitic) magmas is unlikely since subduction of water rich crust is required [Taylor and

E00B17

Campbell, 1983]. D/H ratio measured in the atmosphere of Venus is consistent with evaporation of a primordial ocean of several tens of meters deep and subsequent preferential loss of hydrogen by atmospheric erosion [Donahue and Russell, 1997]. Under cooler and wetter conditions the accordingly different temperature and rheological contrasts in the mantle and crust might have resulted in continuous plate tectonics [Solomatov and Moresi, 1996; Stein et al., 2004]. The geochemical signature of such a hypothetical setting predating the current greenhouse climate era and stagnant lid regime is most likely to be preserved in tessera terrain and would manifest as low emissivity at 1 mm due to the low content of FeO and high SiO2 content [Hashimoto and Sugita, 2003; Hashimoto et al., 2008]. [67] The qualitative result of relative low thermal emission flux above tessera terrain does not dictate the conclusion of granitic surface composition. Absolute surface emissivity is not known and not also very conclusive even if known at only one wavelength. Other materials can have a significantly lower emissivity at 1 mm than basalt. Anorthosite rock is to 90 to 100% composed of the feldspars anorthite and albite, which both have reflectances at 1 mm of about 0.8. Anorthosite maybe even more plausible an explanation for low emissivity regions than granitic composition since its creation has less prerequisites. Anorthositic crust is supposed to differentiate on the top of a fully molten planetary mantle of terrestrial composition [Taylor, 1974]. In this case the situation on Venus would resemble the lunar crustal dichotomy of bright anorthosite highlands and dark basalt mares [Nikolaeva et al., 1992]. [68] The negative flux anomaly observed here is not perfectly aligned with tessera boundaries, see Figure 12. This does not preclude an interpretation of tessera terrain to be mainly composed of less dense felsic material. Basaltic material might have been emplaced by volcanism, accretion of terranes or obduction on the buoyant and thickened crustal block. The basaltic material then would have undergone the same tectonic deformation history characteristic of tessera while the isostatically elevated topography prevented burial by lava floods. On the other hand buried granitic material might undergo remelting leading to felsic volcanism outside of tessera boundaries. [69] However, there are serious concerns to the hypothesis of felsic tessera terrain. Aeolian transport of crater ejecta is known to occur on Venus. Finer grained ejecta and volcanic ash will be transported over larger distances. It is difficult to imaging that over time period of several Ga, that are required for either anorthosite or granites hypotheses, the tesserae had not been covered by a blanket of material closer to average composition. Basilevsky et al. [1997] gives an resurfacing rate of 0.1 to 10 cm per Ma from the fading of radar bright decimeter to decameter ejecta. Though higher winds at higher altitudes might sweep the tessera highlands [Basilevsky et al., 1992] and tectonic deformation may expose the rock below compacted soil, compositional mixing of surface materials will be more relevant for areas significantly older than the relatively young plains and volcanic features. [70] A second and more serious concern is that radar altimetry seems to be less reliable over tessera terrain, with a bias toward lower than actual surface altitudes. Figure 14 shows Magellan SAR mosaic and global topography data

16 of 21

E00B17

MUELLER ET AL.: VENUS SURFACE THERMAL EMISSION

E00B17

formed plain slightly northeast of the center of the image, is partly depressed relative to the plains. Emplacement of the plains without covering the depressed terrain seems improbable. Further support for the hypothesis of systematically too low radar altimetry is given by the negative NIR flux anomaly, that is in shape similar to the suspicious depressions, see Figure 14. The eastern, elevated block of tessera shows nonuniform flux anomaly. Isolated depressions circling the southern and western rim of the elevated tessera block have lower altimetry readings than embaying plains. Assuming that these readings are too low, smearing of Magellan altimetry with a 100 km FWHM gaussian weight function during data processing will lead to a diffuse, crescent shaped negative flux anomaly along the southern border of the tessera. Such a flux anomaly is observed. Similar considerations can be made for the other negative flux anomalies in the area. Magellan altimetry of tessera terrain needs to be examined more closely before any definite conclusions on emissivity of tessera can be made. [71] Independent examination with a different approach analyzing VMC images did not reveal NIR contrast in the shape of tessera terrains (Basilevsky et al., submitted manuscript, 2008). Such contrast would be expected if these terrains were fully composed of anorthositic outcrops with an emissivity of 0.5.

Figure 14. Qualitative comparison of different data sets in the Cocomama tessera region, sinusoidal projection from 66° to 60°S and 13° to 30°E. (a) Magellan left-look SAR (image data from NASA/JPL/Magellan). (b) Magellan GTDR altimetry (image data from NASA/JPL/Magellan). (c) VIRTIS thermal flux anomaly. Outlined is the boundary between mostly pristine plains and areas dominated by tectonic deformation. Deformed terrain is mostly tessera, the central and southeastern region with predominant north – south strike is characterized in electronic maps by Tanaka et al. [1997] as ridge belt. Magellan GTDR possibly has a bias to too low values over deformed areas. tesseraSAR-GTDR. record GTDR of Cocomama tessera. Both seem difficult to reconcile with each other. The easternmost part of the tessera terrain is elevated above the surrounding plains. The deformed terrain encircling the smooth, nearly unde-

6.2. Large Scale Ultramafic Volcanism [72] The high thermal flux over young volcanic flows associated with coronae in Lada terra and less clearly over the nearby Lavinia basin might indicate a composition more mafic than the average plains, which are here used to define normal thermal flux. Most evidence, from morphology and in situ measurements, hints to a basaltic composition of plains [Grimm and Hess, 1997]. Basalts are created by partial melting of ultramafic mantle material, the amount of refractory mafic minerals in the liquid phase is determined by temperature and pressure conditions. The cratering record indicates that the most of the plains were created during an episode of increased resurfacing with little magmatic activity afterward [Schaber et al., 1992; Strom et al., 1994; Basilevsky et al., 1997]. Parmentier and Hess [1992] furthermore propose that the upper mantle below the crust is depleted of felsic minerals and FeO through basalt generation during the resurfacing event. The depleted mantle material is assumed compositionally buoyant and not partaking in convection. Head et al. [1994] argue that pressure release melting in mantle diapirs at the base of this layer might produce large volumes of ultramafic, MgO-rich magma that might account for the large lowviscosity lava flows. [73] Terrestrial analogues are komatiite or picrite that have been proposed by Komatsu et al. [1993] as lavas that could thermally erode basaltic surface and thus produce the incised sinuous channels found, e.g., on the young lava flows in or close to Quetzalpetlatl corona. These sinuous rilles are north of and with flow direction away from the thermal flux anomaly seen on the southern flank of Quetzalpetlatl. The mapping of Komatsu et al. [1993] does, however, not extend below 70°S which is unfortunately also the northern rim of the flux anomaly. The lava flows on the rim of Lavinia planitia feature simple and complex channels

17 of 21

E00B17

MUELLER ET AL.: VENUS SURFACE THERMAL EMISSION

with flow margins that share some characteristics with sinuous rilles but do not necessarily require thermal erosion [Baker et al., 1997]. [74] The Lada-Lavinia regio, which features the most prominent thermal flux anomalies, has been characterized by Magee and Head [1995] as geological setting favorable for the generation of melt. Mantle downwelling below the Lavinia planitia basin is thought to be associated with the extensional Lada rift on the southern and eastern rim of Lavinia [Baer et al., 1994]. The lava flows with anomalous IR flux are in the vicinity of coronae. Coronae are interpreted to be imprints of diapir activity and possibly linked to extensional rifting [Stofan et al., 1997]. Smrekar and Stofan [1997] modeled the range of topographic signatures of coronae as result of an evolutionary process starting from an upwelling mantle plume, its cessation and as latest stage lithospheric delamination. Elkins-Tanton et al. [2007] modeled possible melt compositions in the context of lithosphere delamination and found, among others, ultramafic melts. [75] Constraints on composition further than the assumption of a higher emissivity than that of basalt cannot be made with this simple atmospheric correction model, wherefore it might be idle to speculate on exact surface rock composition at this point. Furthermore the positive flux anomaly over Lavinia planitia is to some extent correlated with altimetry that given the empirical nature of this approach, rouses suspicions of systematical errors due to the assumption of an incorrect temperature. A similar error might pertain to the flux anomaly seen at volcanic edifices with a diameter in the magnitude of the assumed spatial resolution. Assumption of a lower spatial resolution, i.e., smoothing of the radar altimetry with a wider radius will lower the flux anomalies seen at these places. [76] This is not true for the anomalies seen at the flank of topographic rises. The flux anomaly at the southern flank of Quetzalpetlatl is below 70°S, where VIRTIS images generally are affected by stray light. Also in considering the coverage plot (Figure 10) together with the usual viewing geometry of VIRTIS (the slit oriented parallel to the terminator) it becomes obvious that the Lavinia region has a certain bias to be observed at certain samples of the detector array. If the distribution of stray light on the detector array is not homogeneous the removal of stray light might lead to an systematic error that is not removed by averaging because the coverage of the southern hemisphere is not homogeneous. Further collection of data might resolve this issue.

7. Conclusions [77] The atmospheric correction employed here and more importantly, the capability of VIRTIS on Venus Express to observe the same surface area repeatedly, provide consistent indications that some brightness variation in the 1.02 mm window cannot be accounted for by cloud transmittance and surface temperature depending only on altitude. Contrary to previous results in studies by Lecacheux et al. [1993] and Meadows and Crisp [1996], significant contrast, uncorrelated to cloud opacity and surface altitude, remains after atmospheric correction. The above statement owes much, if not everything, to the higher spatial resolution and area

E00B17

coverage achieved through the VIRTIS data set. The method employed here does not allow us to retrieve absolute surface emissivity, as it does not accurately address radiative transfer in the lowest atmosphere [Hashimoto and Sugita, 2003]. Furthermore the empirical determination of thermal flux to topography relation may lead to an error depending on distribution of surface emissivity with surface elevation; see Hashimoto et al. [2008] for evidence for such a distribution. [78] Regardless of these intrinsic uncertainties, a relation exists between NIR flux and morphological units from Magellan radar images at medium altitudes. The lower than average flux over tessera terrain is possibly indicative of felsic composition such as granite or anorthosite, which have been suggested to exist on Venus by Taylor and Campbell [1983], Nikolaeva et al. [1992], and Hashimoto and Sugita [2003]. A similar observation and interpretation based on Galileo NIMS data is made by Hashimoto et al. [2008], who find generally lower emissivity in the highlands. A systematical bias toward lower altitude values in the Magellan topography data set over the highly tectonized tessera terrain is also conceivable and consistent with this result. Comparison with other altimetry data sets, e.g., Pioneer Venus, Venera 15/16, and possibly topography derived from Magellan stereo radar imaging might help to resolve this issue. [79] Higher than average flux is found at some volcanic edifices in Lada terra and Themis, Imdr, and Phoebe regions, the latter three classified as active hot spots by Smrekar et al. [1997]. Lavinia planitia is characterized by a anomalously high flux likely caused by systematic errors. The positive flux anomalies located on the flanks of Shulamite, Shiwanokia, and Quetzalpetlatl coronae, are difficult to attribute to any systematical problem. The Quetzalpetlatl anomaly is well correlated with the young lava flows Cavillaca and Juturna flucti. The high flux anomaly at young lava flows is consistent the hypothesis of large volume komatiite or picrite volcanism predicted by Head et al. [1994] as consequence of chemical differentiation of the upper mantle in the course of secondary crust formation. Another possible interpretation would be that lithospheric delamination in association with the Quetzalpetlatl-Boala corona formation [Smrekar and Stofan, 1997] produced the ultramafic lavas [Elkins-Tanton et al., 2007] that lead to the increased NIR flux. [80] Overall probability of these compositional interpretations is however difficult to estimate, as no absolute surface emissivity is retrieved. To retrieve emissivity, an improved model of radiative transfer has to be implemented similarly to the work of Hashimoto et al. [2008] or Arnold et al. [2008]. Crucial input parameters of radiative transfer models, e.g., continuum absorption coefficient, subcloud haze and even surface temperature are poorly constrained. Analysis of VIRTIS data together with Magellan altimetry data might improve knowledge on these parameters [Carlson et al., 1993a]. When a reliable estimate of Venus surface emissivity is achieved, the next step has to be comparison with near infrared emissivity or laboratory samples measured at high temperatures. Little data exist yet on high temperature emissivity of minerals, but more will be hopefully available soon [Maturilli et al., 2007; Helbert and Maturilli, 2008]. This comparison will constrain surface rock

18 of 21

E00B17

MUELLER ET AL.: VENUS SURFACE THERMAL EMISSION

composition if near infrared emissivity is not dominated by chemical weathering and volatile transport of minerals. In this case some knowledge on the nature of the high-altitude radiothermal emissivity will be gained if data coverage can sufficiently be extended to these regions. [81] Acknowledgments. We thank S. Smrekar for an insightful discussion of the geology of Venus and also for kindly providing many suggestions for the improvement of language. We acknowledge the financial support for the VIRTIS instrument from ASI and CNES. A significant part of this work was made possible by a grant from the Wernher von Braun foundation (Germany). We are grateful to the whole VIRTIS on Venus Express team for providing this excellent data set.

References Allen, D. A., and J. W. Crawford (1984), Cloud structure on the dark side of Venus, Nature, 307, 222 – 224. Arnold, G., R. Haus, D. Kappel, P. Drossart, and G. Piccioni (2008), Venus surface data extraction from VIRTIS/Venus Express measurements: Estimation of a quantitative approach, J. Geophys. Res., 113, E00B10, doi:10.1029/2008JE003087. Baer, G., G. Schubert, D. L. Bindschadler, and E. R. Stofan (1994), Spatial and temporal relations between coronae and extensional belts, northern Lada Terra, Venus, J. Geophys. Res., 99, 8355 – 8369. Baines, K. H., et al. (2000), Detection of sub-micron radiation from the surface of Venus by Cassini/VIMS, Icarus, 148, 307 – 311, doi:10.1006/ icar.2000.6519. Baker, V. R., G. Komatsu, V. C. Gulick, and T. J. Parker (1997), Channels and valleys, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 757 – 793, Univ. of Ariz. Press, Tuscon. Basilevsky, A. T., and J. W. Head (2003), The surface of Venus, Rep. Prog. Phys., 66, 1699 – 1734, doi:10.1088/0034-4885/66/10/R04. Basilevsky, A. T., O. V. Nikolaeva, and R. O. Kuz’min (1992), Resurfacing, in Venus Geology, Geochemistry, and Geophysics: Research Results From the USSR, edited by V. L. Barsukov et al., pp. 153 – 160, Univ. of Ariz. Press, Tuscon. Basilevsky, A. T., J. W. Head, G. G. Schaber, and R. G. Strom (1997), The resurfacing history of Venus, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 1047 – 1084, Univ. of Ariz. Press, Tuscon. Carlson, R. W., et al. (1991), Galileo infrared imaging spectroscopy measurements at Venus, Science, 253, 1541 – 1548. Carlson, R. W., K. H. Baines, M. Girard, L. W. Kamp, P. Drossart, T. Encrenaz, and F. W. Taylor (1993a), Galileo/NIMS near-infrared thermal imagery of the surface of Venus, Proc. Lunar Planet. Sci. Conf., 24th, 253. Carlson, R. W., L. W. Kamp, K. H. Baines, J. B. Pollack, D. H. Grinspoon, T. Encrenaz, P. Drossart, and F. W. Taylor (1993b), Variations in Venus cloud particle properties: A new view of Venus’s cloud morphology as observed by Galileo Near-Infrared Mapping Spectrometer, Planet. Space Sci., 41, 477 – 485, doi:10.1016/0032-0633(93)90030-6. Coradini, A., et al. (1998), VIRTIS: An imaging spectrometer for the Rosetta mission, Planet. Space Sci., 46, 1291 – 1304. Crisp, D., and D. Titov (1997), The thermal balance of the Venus atmosphere, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 353 – 384, Univ. of Ariz. Press, Tuscon. Crisp, D., D. A. Allen, D. H. Grinspoon, and J. B. Pollack (1991), The dark side of Venus: Near-infrared images and spectra from the AngloAustralian Observatory, Science, 253, 1263 – 1266. Crisp, D., V. S. Meadows, B. Be´zard, C. de Bergh, J.-P. Maillard, and F. P. Mills (1996), Ground-based near-infrared observations of the Venus nightside: 1.27-mm O2 (aDg ) airglow from the upper atmosphere, J. Geophys. Res., 101, 4577 – 4594. Crisp, D., et al. (2002), Divergent evolution among Earth-like planets: The case for Venus exploration, in The Future of Solar System Exploration (2003 – 2013): First Decadal Study Contributions, edited by M. V. Sykes, Astron. Soc. Pac. Conf. Ser., 272, 5 – 34. Donahue, T. M., and C. T. Russell (1997), The Venus atmosphere and ionosphere and their interaction with the solar wind: An overview, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 3 – 32, Univ. of Ariz. Press, Tuscon. Drossart, P., et al. (2007), Scientific goals for the observation of Venus by VIRTIS on ESA/Venus express mission, Planet. Space Sci., 55, 1653 – 1672, doi:10.1016/j.pss.2007.01.003.

E00B17

Elkins-Tanton, L. T., S. E. Smrekar, P. C. Hess, and E. M. Parmentier (2007), Volcanism and volatile recycling on a one-plate planet: Applications to Venus, J. Geophys. Res., 112, E04S06, doi:10.1029/2006JE002793. Esposito, L. W., J.-L. Bertaux, V. Krasnopolsky, V. I. Moroz, and L. V. Zasova (1997), Chemistry of lower atmosphere and clouds, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 415 – 458, Univ. of Ariz. Press, Tuscon. Fegley, B., Jr., G. Klingelho¨fer, K. Lodders, and T. Widemann (1997), Geochemistry of surface-atmosphere interactions on Venus, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 591 – 636, Univ. of Ariz. Press, Tuscon. Ford, P. G., and G. H. Pettengill (1992), Venus topography and kilometerscale slopes, J. Geophys. Res., 97, 13,103 – 13,112. Gierasch, P. J., et al. (1997), The general circulation of the Venus atmosphere: An assessment, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 459 – 500, Univ. of Ariz. Press, Tuscon. Goody, R. M., and Y. L. Yung (1989), Atmospheric Radiation: Theoretical Basis, 2nd ed., Oxford Univ. Press, New York. Grieger, B., N. I. Ignatiev, N. M. Hoekzema, and H. U. Keller (2004), Indication of a near surface cloud layer on Venus from reanalysis of Venera 13/14 spectrophotometer data, in Planetary Probe Atmospheric Entry and Descent Trajectory Analysis and Science, edited by A. Wilson, Eur. Space Agency Spec. Publ., 544, 63 – 70. Grimm, R. E., and P. C. Hess (1997), The crust of Venus, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 1205 – 1244, Univ. of Ariz. Press, Tuscon. Grinspoon, D. H., J. B. Pollack, B. R. Sitton, R. W. Carlson, L. W. Kamp, K. H. Baines, T. Encrenaz, and F. W. Taylor (1993), Probing Venus’s cloud structure with Galileo NIMS, Planet. Space Sci., 41, 515 – 542, doi:10.1016/0032-0633(93)90034-Y. Hansen, J. E., and J. W. Hovenier (1974), Nature of the Venus clouds as derived from their polarization, in Exploration of the Planetary System, edited by A. Woszczyk and C. Iwaniszewska, Symp. Int. Astron. Union, 65, 197 – 200. Hansen, J. E., and L. D. Travis (1974), Light scattering in planetary atmospheres, Space Sci. Rev., 16, 527 – 610. Hansen, V. L., J. J. Willis, and W. B. Banerdt (1997), Tectonic overview and synthesis, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 797 – 844, Univ. of Ariz. Press, Tuscon. Hashimoto, G. L., and Y. Abe (2005), Climate control on Venus: Comparison of the carbonate and pyrite models, Planet. Space Sci., 53, 839 – 848, doi:10.1016/j.pss.2005.01.005. Hashimoto, G. L., and T. Imamura (2001), Elucidating the rate of volcanism on Venus: Detection of lava eruptions using near-infrared observations, Icarus, 154, 239 – 243, doi:10.1006/icar.2001.6713. Hashimoto, G. L., and S. Sugita (2003), On observing the compositional variability of the surface of Venus using nightside near-infrared thermal radiation, J. Geophys. Res., 108(E9), 5109, doi:10.1029/2003JE002082. Hashimoto, G. L., M. Roos-Serote, S. Sugita, M. S. Gilmore, L. W. Kamp, B. Carlson, and K. Baines (2008), Galileo Near-Infrared Mapping Spectrometer (NIMS) data suggest felsic highland crust on Venus, J. Geophys. Res., doi:10.1029/2008JE003134, in press. Head, J. W., L. S. Crumpler, J. C. Aubele, J. E. Guest, and R. S. Saunders (1992), Venus volcanism: Classification of volcanic features and structures, associations, and global distribution from Magellan data, J. Geophys. Res., 97, 13,153 – 13,197. Head, J. W., E. M. Parmentier, and P. C. Hess (1994), Venus: Vertical accretion of crust and depleted mantle and implications for geological history and processes, Planet. Space Sci., 42, 803 – 811, doi:10.1016/ 0032-0633(94)90061-2. Helbert, J., and A. Maturilli (2008), The heat is on: In the planetary emissivity laboratory, Proc. Lunar Planet. Sci. Conf., 39th., 2408. Helbert, J., N. Mu¨ller, P. Kostama, L. Marinangeli, G. Piccioni, and P. Drossart (2008), Surface brightness variations seen by VIRTIS on Venus Express and implications for the evolution of the Lada Terra region, Venus, Geophys. Res. Lett., 35, L11201, doi:10.1029/2008GL033609. Ivanov, M. A., and J. W. Head (1996), Tessera terrain on Venus: A survey of the global distribution, characteristics, and relation to surrounding units from Magellan data, J. Geophys. Res., 101, 14,861 – 14,908. Ivanov, M. A., and J. W. Head (2001), Geologic map of the Lavinia Planitia Quadrangle (V-55), Venus, U.S. Geol. Surv. Geol. Invest. Map I-2684. Ivanov, M. A., and J. W. Head (2006), Geologic map of the Mylitta Fluctus Quadrangle (V-61), Venus, U.S. Geol. Surv. Sci. Invest. Map I-2920. Kamp, L. W., F. W. Taylor, and S. B. Calcutt (1988), Structure of Venus’s atmosphere from modelling of night-side infrared spectra, Nature, 336, 360 – 362, doi:10.1038/336360a0.

19 of 21

E00B17

MUELLER ET AL.: VENUS SURFACE THERMAL EMISSION

Kargel, J. S., G. Komatsu, V. R. Baker, and R. G. Strom (1993), The volcanology of Venera and VEGA landing sites and the geochemistry of Venus, Icarus, 103, 253 – 275, doi:10.1006/icar.1993.1069. Kaula, W. M., and R. J. Phillips (1981), Quantitative tests for plate tectonics on Venus, Geophys. Res. Lett., 8, 1187 – 1190. Knollenberg, R. G., and D. M. Hunten (1980), The microphysics of the clouds of Venus: Results of the Pioneer Venus particle size spectrometer experiment, J. Geophys. Res., 85, 8039 – 8058. Komatsu, G., V. R. Baker, V. C. Gulick, and T. J. Parker (1993), Venusian channels and valleys: Distribution and volcanological implications, Icarus, 102, 1 – 25, doi:10.1006/icar.1993.1029. Kratter, K. M., L. M. Carter, and D. B. Campbell (2007), An expanded view of Lada Terra, Venus: New Arecibo radar observations of Quetzalpetlatl Corona and surrounding flows, J. Geophys. Res., 112, E04008, doi:10.1029/2006JE002722. Lecacheux, J., P. Drossart, P. Laques, F. Deladerriere, and F. Colas (1993), Detection of the surface of Venus at 1.0 micrometer from ground-based observations, Planet. Space Sci., 41, 543 – 549, doi:10.1016/00320633(93)90035-Z. Magee, K. P., and J. W. Head (1995), The role of rifting in the generation of melt: Implications for the origin and evolution of the Lada Terra-Lavinia Planitia region of Venus, J. Geophys. Res., 100, 1527 – 1552. Marcq, E., T. Encrenaz, B. Be´ zard, and M. Birlan (2006), Remote sensing of Venus’ lower atmosphere from ground-based IR spectroscopy: Latitudinal and vertical distribution of minor species, Planet. Space Sci., 54, 1360 – 1370, doi:10.1016/j.pss.2006.04.024. Marov, M. I., V. E. Lystsev, V. N. Lebedev, N. L. Lukashevich, and V. P. Shari (1980), The structure and microphysical properties of the Venus clouds: Venera 9, 10, and 11 data, Icarus, 44, 608 – 639, doi:10.1016/ 0019-1035(80)90131-1. Maturilli, A., J. Helbert, and L. Moroz (2007), The Berlin Emissivity Database (BED): A collection of emissivity spectra for planetary analogue minerals, Proc. Lunar Planet. Sci. Conf., 38th, 1281. Meadows, V. S., and D. Crisp (1996), Ground-based near-infrared observations of the Venus nightside: The thermal structure and water abundance near the surface, J. Geophys. Res., 101, 4595 – 4622. Moroz, V. I. (2002), Estimates of visibility of the surface of Venus from descent probes and balloons, Planet. Space Sci., 50, 287 – 297. Moroz, V. I., A. P. Ekonomov, B. E. Moshkin, H. E. Revercomb, L. A. Sromovsky, and J. T. Schofield (1985), Solar and thermal radiation in the Venus atmosphere, Adv. Space Res., 5, 197 – 232, doi:10.1016/02731177(85)90202-9. Nikolaeva, O. V., M. A. Ivanov, and V. K. Borozdin (1992), Evidence on the crustal dichotomy, in Venus Geology, Geochemistry, and Geophysics: Research Results From the USSR, edited by V. L. Barsukov et al., pp. 129 – 139, Univ. of Ariz. Press, Tuscon. Parmentier, E. M., and P. C. Hess (1992), Chemical differentiation of a convecting planetary interior: Consequences for a one plate planet such as Venus, Geophys. Res. Lett., 19, 2015 – 2018. Pettengill, G. H., P. G. Ford, W. T. K. Johnson, R. K. Raney, and L. A. Soderblom (1991), Magellan: Radar performance and data products, Science, 252, 260 – 265. Pettengill, G. H., P. G. Ford, and R. J. Wilt (1992), Venus surface radiothermal emission as observed by Magellan, J. Geophys. Res., 97, 13,091 – 13,102. Pettit, E., and S. B. Nicholson (1955), Temperatures on the bright and dark sides of Venus, Publ. Astron. Soc. Pac., 67, 293 – 303. Phillips, R. J., R. F. Raubertas, R. E. Arvidson, I. C. Sarkar, R. R. Herrick, N. Izenberg, and R. E. Grimm (1992), Impact craters and Venus resurfacing history, J. Geophys. Res., 97, 15,923 – 15,948. Pollack, J. B., et al. (1993), Near-infrared light from Venus’ nightside: A spectroscopic analysis, Icarus, 103, 1 – 42, doi:10.1006/icar.1993.1055. Ragent, B., L. W. Esposito, M. G. Tomasko, M. I. Marov, and V. P. Shari (1985), Particulate matter in the Venus atmosphere, Adv. Space Res., 5, 85 – 115, doi:10.1016/0273-1177(85)90199-1. Rappaport, N. J., A. S. Konopliv, A. B. Kucinskas, and P. G. Ford (1999), An improved 360 degree and order model of Venus topography, Icarus, 139, 19 – 31, doi:10.1006/icar.1999.6081. Sagan, C. (1962), Structure of the lower atmosphere of Venus, Icarus, 1, 151 – 169, doi:10.1016/0019-1035(62)90015-5. Saunders, R. S., R. E. Arvidson, J. W. Head, G. G. Schaber, E. R. Stofan, and S. C. Solomon (1991), An overview of Venus geology, Science, 252, 249 – 252. Schaber, G. G., et al. (1992), Geology and distribution of impact craters on Venus: What are they telling us?, J. Geophys. Res., 97, 13,257 – 13,301. Seiff, A. (1983), Thermal structure of the atmosphere of Venus, in Venus, edited by D. M. Hunten and M. S. Matthews, pp. 215 – 279, Univ. of Ariz. Press, Tuscon.

E00B17

Seiff, A. (1987), Further information on structure of the atmosphere of Venus derived from the VEGA Venus balloon and lander mission, Adv. Space Res., 7, 323 – 328, doi:10.1016/0273-1177(87)90239-0. Seiff, A., J. T. Schofield, A. J. Kliore, F. W. Taylor, and S. S. Limaye (1985), Models of the structure of the atmosphere of Venus from the surface to 100 kilometers altitude, Adv. Space Res., 5, 3 – 58, doi:10.1016/0273-1177(85)90197-8. Smrekar, S. E., and R. J. Phillips (1991), Venusian highlands: Geoid to topography ratios and their implications, Earth Planet. Sci. Lett., 107, 582 – 597, doi:10.1016/0012-821X(91)90103-O. Smrekar, S. E., and E. R. Stofan (1997), Coupled upwelling and delamination: A new mechanism for coronae formation and heat loss on Venus, Science, 277, 1289 – 1294. Smrekar, S. E., W. S. Kiefer, and E. R. Stofan (1997), Large volcanic rises on Venus, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 845 – 878, Univ. of Ariz. Press, Tuscon. Snyder, D. (2002), Cooling of lava flows on Venus: The coupling of radiative and convective heat transfer, J. Geophys. Res., 107(E10), 5080, doi:10.1029/2001JE001501. Solomatov, V. S., and L.-N. Moresi (1996), Stagnant lid convection on Venus, J. Geophys. Res., 101, 4737 – 4754. Solomon, S. C., et al. (1992), Venus tectonics: An overview of Magellan observations, J. Geophys. Res., 97, 13,199 – 13,255. Spohn, T. (1991), Mantle differentiation and thermal evolution of Mars, Mercury, and Venus, Icarus, 90, 222 – 236, doi:10.1016/00191035(91)90103-Z. Stein, C., J. Schmalzl, and U. Hansen (2004), The effect of rheological parameters on plate behaviour in a self-consistent model of mantle convection, Phys. Earth Planet. Inter., 142, 225 – 255, doi:10.1016/ j.pepi.2004.01.006. Stevenson, D. J., T. Spohn, and G. Schubert (1983), Magnetism and thermal evolution of the terrestrial planets, Icarus, 54, 466 – 489, doi:10.1016/ 0019-1035(83)90241-5. Stofan, E. R., V. L. Sharpton, G. Schubert, G. Baer, D. L. Bindschadler, D. M. Janes, and S. W. Squyres (1992), Global distribution and characteristics of coronae and related features on Venus: Implications for origin and relation to mantle processes, J. Geophys. Res., 97, 13,347 – 13,378. Stofan, E. R., V. E. Hamilton, D. M. Janes, and S. E. Smrekar (1997), Coronae on Venus: Morphology and origin, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 931 – 965, Univ. of Ariz. Press, Tuscon. Stone, P. H. (1975), The dynamics of the atmosphere of Venus, J. Atmos. Sci., 32, 1005 – 1016. Strom, R. G., G. G. Schaber, and D. D. Dawsow (1994), The global resurfacing of Venus, J. Geophys. Res., 99, 10,899 – 10,926. Surkov, I. A., L. P. Moskaleva, V. P. Khariukova, A. D. Dudin, and G. G. Smirnov (1986), Venus rock composition at the VEGA 2 landing site, J. Geophys. Res., 91, 215 – 218. Surkov, I. A., F. F. Kirnozov, V. N. Glazov, A. G. Dunchenko, and L. P. Tatsy (1987), Uranium, thorium, and potassium in the Venusian rocks at the landing sites of VEGA 1 and 2, J. Geophys. Res., 92, 537 – 540. Surkov, Y. A., and V. L. Barsukov (1985), Composition, structure and properties of Venus rocks, Adv. Space Res., 5, 17 – 29, doi:10.1016/ 0273-1177(85)90237-6. Svedhem, H., et al. (2007), Venus Express-The first European mission to Ve n u s , P l a n e t . S p a c e S c i . , 5 5 , 1 6 3 6 – 1 6 5 2 , d o i : 1 0 . 1 0 1 6 / j.pss.2007.01.013. Tanaka, K. L., D. A. Senske, M. Price, and R. L. Kirk (1997), Physiography, geomorphic/geologic mapping and stratigraphy of Venus, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 667 – 694, Univ. of Ariz. Press, Tuscon. Taylor, F. W. (2006), Venus before Venus Express, Planet. Space Sci., 54, 1249 – 1262, doi:10.1016/j.pss.2006.04.031. Taylor, F. W., D. Crisp, and B. Be´zard (1997), Near-infrared sounding of the lower atmosphere of Venus, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 325 – 351, Univ. of Ariz. Press, Tuscon. Taylor, S. R. (1974), Geochemical evolution of the Moon, LPI Contrib., 195, 7 – 9. Taylor, S. R. (1989), Growth of planetary crusts, Tectonophysics, 161, 147 – 156. Taylor, S. R., and I. H. Campbell (1983), No water, no granites: No oceans, no continents, Geophys. Res. Lett., 10, 1061 – 1064. Titov, D. V., et al. (2006), Venus Express science planning, Planet. Space Sci., 54, 1279 – 1297, doi:10.1016/j.pss.2006.04.017. Tsang, C. C. C., P. G. J. Irwin, F. W. Taylor, and C. F. Wilson (2008), A correlated-k model of radiative transfer in the near-infrared windows of venus, J. Quant. Spectrosc. Radiat. Transfer, 109, 1118 – 1135, doi:10.1016/j.jqsrt.2007.12.008.

20 of 21

E00B17

MUELLER ET AL.: VENUS SURFACE THERMAL EMISSION

Turcotte, D. L. (1993), An episodic hypothesis for venusian tectonics, J. Geophys. Res., 98, 17,061 – 17,068. von Zahn, U., S. Kumar, H. Niemann, and R. Prinn (1983), Composition of the Venus atmosphere, in Venus, edited by D. M. Hunten and M. S. Matthews, pp. 299 – 340, Univ. of Ariz. Press, Tuscon. Wood, J. A. (1997), Rock weathering on the surface of Venus, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 637 – 661, Univ. of Ariz. Press, Tuscon.























P. Drossart and S. Erard, LESIA, Observatoire de Paris, 61 Avenue Observatoire, F-75014 Paris, France.

E00B17

G. L. Hashimoto, Laboratory for Earth and Planetary Atmospheric Science, Department of Earth and Planetary Sciences, 1-1 Rokkodai-cho, Kobe University, Kobe 657-8501, Japan. J. Helbert and N. Mueller, Institute of Planetary Research, German Aerospace Center, Rutherfordstrasse 2, D-12489 Berlin, Germany. ([email protected]) G. Piccioni, IASF-INAF, via del fosso del cavaliere 100, Rome I-00133, Italy. C. C. C. Tsang, Atmospheric, Oceanic, and Planetary Physics, Department of Physics, University of Oxford, Clarendon Laboratory, Parks Road, Oxford OX1 3PU, UK.

21 of 21

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, E00B10, doi:10.1029/2008JE003087, 2008

for

Full Article

Venus surface data extraction from VIRTIS/Venus Express measurements: Estimation of a quantitative approach Gabriele Arnold,1,2 Rainer Haus,3 David Kappel,2 Pierre Drossart,4 and Giuseppe Piccioni5 Received 24 January 2008; revised 17 April 2008; accepted 15 July 2008; published 7 October 2008.

[1] Nightside emission measurements of the Visible and Infrared Thermal Imaging

Spectrometer (VIRTIS) on the Venus Express spacecraft were used to estimate the potential for surface data extraction. A selection of orbits over the northern hemisphere was performed for footprints that cover different scales of surface elevation variations. A correction method was used to remove stray light from the measured spectra that is due to direct sunlight striking the instrument. A preliminary radiative transfer calculation technique was applied to simulate Venus nightside radiation. The basic features of the measured spectra are well reproduced. Present limitations of the algorithm are discussed. The variability of the emission window radiances with respect to cloud opacity and surface elevation is modeled and discussed in direct comparison with the measurements. It is demonstrated that a multispectral analysis in the surface and deep atmosphere window ranges (1.0–2.3 mm) and the use of radiance ratios are well suited to decloud the data and to extract surface information from the VIRTIS measurements. This method allows a mapping of surface topography and the retrieval of the surface temperature. A preliminary topography retrieval at Beta Regio was performed and compared with Magellan radar data. Differences are possibly due to emissivity variations on the surface. Citation: Arnold, G., R. Haus, D. Kappel, P. Drossart, and G. Piccioni (2008), Venus surface data extraction from VIRTIS/Venus Express measurements: Estimation of a quantitative approach, J. Geophys. Res., 113, E00B10, doi:10.1029/2008JE003087.

1. Introduction [2] Most of Venus’s surface consists of gently rolling plains with little relief. There are several depressions and two large highlands, Ishtar Terra in the northern hemisphere and Aphrodite Terra along the equator. The Magellan radar images indicated a violent volcanic activity in the planet’s past [Saunders et al., 1992; Solomon et al., 1992]. Extended lava plains, dotted with isolated shield volcanoes and volcanic constructs dominate the geology. The deformations on Venus are not a result of Earth like plate tectonic. They are probably related to the dynamic forces within the planet’s mantle [Basilevsky and Head, 1988, 2002]. [3] Little is known about the surface material. Most of it appears to consist of basalt, but the primary material is poorly classified. The volcanic landforms are consistent with low-viscosity eruptions, which are characteristic of mafic materials like basalt [Head et al., 1992]. However, 1 Institut fu¨r Planetologie, Westfa¨lische Wilhelms-Universita¨t Mu¨nster, Mu¨nster, Germany. 2 Institute of Planetary Research, German Aerospace Center, Berlin, Germany. 3 Department of Marine Remote Sensing, Remote Sensing Technology Institute, German Aerospace Center, Berlin, Germany. 4 LESIA, Observatoire de Paris, CNRS, UPMC, Universite´ ParisDiderot, Meudon, France. 5 Instituto di Astrofisica Spaziale e Fisica Cosmica, Rome, Italy.

Copyright 2008 by the American Geophysical Union. 0148-0227/08/2008JE003087$09.00

Magellan data suggest that some high-viscosity lava formed pancake domes and festoons [McKenzie et al., 1992; Taylor, 2006], which could be related to the presence of more felsic materials like rhyolite. [4] Recent estimates have demonstrated that there may be significant spatial variations in the surface emissivity as large as 20%, which correspond to the difference between granitic and basaltic rocks [Hashimoto and Sugita, 2003]. The chemical and mechanical erosion of the primary material is controlled by reactions of the surface with the hot and dense atmosphere. The study of the chemical composition and the physical properties of the lower atmosphere and surface of Venus are crucial for the understanding of the surface evolution and the climate of the planet. [5] Although Venus Express primarily focuses on atmospheric science, it also supplies valuable information about the surface. VIRTIS, the Visible and Infrared Thermal Imaging Spectrometer [Drossart et al., 2007; Piccioni et al., 2007], is one of the important experiments dedicated to atmospheric and surface data extraction. VIRTIS is the first instrument operating in an orbit around Venus with the capability to systematically investigate the nightside emission of the planet in the near-infrared atmospheric windows. It performs the first detailed global exploration of the depths of the thick Venusian atmosphere. It provides for the first time clues to the emissivity of surface materials and it may provide direct evidence of active volcanism if present [Baines et al., 2006]. The feasibility of such studies was demonstrated first by Galileo/NIMS [Carlson et al., 1991]

E00B10

1 of 13

E00B10

ARNOLD ET AL.: VENUS SURFACE DATA FROM VIRTIS/VEX

E00B10

Figure 1. The three orbit groups with different topographical variations along their footprints. Group 1 orbits (left to right) are 82, 86, 90, 92, 94, 97, 100. Group 2 orbits are 110, 113, 116. Group 3 orbits are 146– 154. The colored background is the surface topography as obtained by Magellan. Each spatial pixel footprint center corresponds to a white dot. The pixel density is higher near the equator because the spacecraft is slower there with respect to the surface. The black edging is used to stress the footprints. Magellan data are available at http://pds-geosciences.wustl.edu/missions/magellan. and Cassini/VIMS [Baines et al., 2000] during the Venus flybys. [6] VIRTIS measurements of the thermal emission from the Venus nightside reveal a high variability of near-infrared spectra over the planet. This is mainly a consequence of spatial variations of cloud opacity and surface elevation, but differences are also due to changes in atmospheric temperature and absorber contents as well as variations of the surface emissivity. While the maximum information on surface and deep atmosphere-surface interaction is obtained from the spectral windows located between 1.00 and 1.35 mm, the windows at 1.74 and 2.3 mm provide information about atmospheric temperature and composition below the main cloud deck. [7] We focus our work on the extraction of surface data from VIRTIS measurements. The present approach is based on a data selection from orbits over the northern hemisphere where the footprints cover a broad range of surface elevation variations including deep valleys like Guinevere Planitia (50°N, 290°E) and high mountain regions like Ishtar Terra (70°N, 340°E) and Beta Regio (30°N, 280°E). A quantitative evaluation of the measurements requires detailed radiative transfer simulations that include appropriate spectral line databases, deep atmosphere continuum absorption features, and multiple scattering effects due to the dense cloud deck. The main intention of part I of the paper, however, is a feasibility study and an estimation of a

quantitative approach to extracting surface information instead of elaborated quantitative retrievals.

2. Data Selection [8] The data were selected from VIRTIS-M-IR measurements over the northern hemisphere of Venus. According to Figure 1, each narrow stripe extends roughly from the equator to the North Pole. It is part of the measurements obtained during one Venus Express (VEX) orbit. One stripe consists of a series of exposures during one VEX orbit. Each exposure yields a frame of 64 spatial pixels times 432 spectral pixels. The 64 pixel wide swath corresponds to a width of roughly two degrees of longitude in the surface footprint, while the latitude remains approximately constant for any pixel in the frame. Thus, the coverage of the pixel footprints of the selected data resembles a latitudinal cross section of spectra at a fixed longitude. Because of the highly elliptical shape of the orbits, a pixel footprint on the surface is of the order of approximately 4 km in longitudinal and 1 km in latitudinal direction at 20°N, but only 0.4 km times 0.1 km at 83°N. [9] The required stripes were found by searching for pushbroom observations with small observation angles close to nadir. This ensures a minimal atmospheric influence on the measured signatures. In this way, the viewing geometry is approximately constant within one stripe, except for a framewise progression of the latitude of the

2 of 13

E00B10

ARNOLD ET AL.: VENUS SURFACE DATA FROM VIRTIS/VEX

E00B10

Figure 2. Initial and stray light-corrected nightside spectrum of VIRTIS (orbit 149, 9°S). The main NIR window positions are 1, 1.02 mm (9765 cm
1); 2, 1.10 mm (9105 cm
1); 3, 1.18 mm (8465 cm
1); 4, 1.28 mm (7843 cm
1); 5, 1.31 mm (7657 cm
1); 6, 1.74 mm (5770 cm
1); 7, 2.30 mm (4357 cm
1). pixel footprints. Furthermore, only measurements at solar incidence angles exceeding 98 degrees were considered for nightside emission studies to ensure that they had been recorded at an adequate distance from the scattered sunlight close to the terminator. [10] To simplify matters, the stripes are named for the corresponding orbit number. Considering the above defined search criteria, three groups with different topographical variations along each group were selected on the northern hemisphere, as shown in Figure 1, to investigate the influence of surface elevation on the spectra. Group 1 contains the orbits 82, 86, 90, 92, 94, 97, and 100, where the topographical variation with latitude is fairly small. These orbits cover lowlands from 10°S to 80°N and 237°E to 265°E. Stronger topographical variations with latitude occur in groups 2 and 3, respectively. Group 2 comprises orbits 110, 113, and 116. The footprints range from 5°S to 83°N and 270°E to 288°E. The surface elevation rises up to 5600 m at Beta Regio. It goes down to
1.6 km at Guinevere Planitia and it is fairly uniform in the north. Group 3 contains orbits with similar topographical variations with latitude as in group 2, but these orbits, with the numbers 146 to 154, span high mountains in high latitudes (Ishtar Terra region at an elevation of up to 6000 m) and lowlands down to
1.5 km elsewhere. They range from 10°S to 83°N and 332°E to 345°E. [11] Some of the selected spectra show a strong increase of radiance with decreasing wavelength in the 1 – 2.5 mm range; that is, the lower envelope in that spectral range is lifted as it is illustrated by the dashed line in Figure 2. A scrutiny of the viewing geometry reveals that this effect is due to direct sunlight close to, but outside the VIRTIS field of view, although there is dark night on the planet’s surface. An examination of the most extreme examples (orbits 137 and 139) shows that the spectral characteristic changes significantly over a small footprint distance and time interval. The change is correlated with the exposure point, as similar occurrences along other orbits confirm. When the Sun angle (i.e., the angle between observation direction and the instrument-to-Sun vector) decreases, the stray light

initially increases, but it disappears abruptly at the point where the Sun becomes eclipsed by Venus itself from the VIRTIS point of view. This stray light effect must not be confounded with upper atmosphere scattering close to the terminator [Meadows and Crisp, 1996] and with indirect light from the illuminated crescent of Venus scattered into the VIRTIS field of view [Mu¨ller et al., 2008], respectively. It is not a surface or atmospheric feature, but is due to direct sunlight striking the VIRTIS instrument. [12] The orbits 137 and 139 with long (3.3 s) and short (0.36 s) exposure times, respectively, can be used to remove the stray light effect to a large extent, because it is most pronounced here and vanishes quickly upon entering eclipse. The difference in radiation measured immediately before and after the eclipse is a good indicator of the amount of stray light. The correction algorithm averages the difference over all pixels of the corresponding frames to ensure a good statistical relevance. Approximately 1 min elapses between the two measurements. This corresponds to a footprint distance of just 125 km. Using the difference between the two measurements seems to be justified, because it will rarely be possible to identify real surface or atmospheric parameter variations in the measured signals along that distance, since the predicted IR surface resolution limit is about 100 km due to the strong cloud scattering [Moroz, 2002a]. [13] The resulting correction curves for the two exposure times are scaled for each individual spectrum to fit it as a lower envelope. This correction is applied to all the orbits and spectra investigated in this work, regardless of the degree of direct sunlight illumination. This prevents a possible discontinuity between spectra that have, or have not, been corrected. The solid line in Figure 2 displays the same spectrum as the dashed line but with the effect of scattered sunlight being removed. Some low-level radiance discontinuities were also eliminated. [14] Because of the limited surface resolution mentioned above it is reasonable to enhance the statistical stability by averaging the spectra of adjacent pixels, where atmospheric conditions are assumed to be similar. The basic data of

3 of 13

E00B10

ARNOLD ET AL.: VENUS SURFACE DATA FROM VIRTIS/VEX

E00B10

Figure 3. Variability of VIRTIS spectra (orbit 151) at different northern latitudes. Window signatures at 2.3 and 1.7 mm are sometimes very small and may even disappear. every orbit are confined to less than two degrees in longitude. The orbits are divided into bins of one degree in latitude. The mean over all pixels of each bin is then written as a function of latitude. This procedure is applied to the radiance values as well as to the corresponding Magellan elevations. [15] Figure 3 presents a set of four VIRTIS-M spectra in the wavelength range from 1.0 to 2.6 mm taken along orbit 151 (group 3) at different latitudes over the northern hemisphere. This orbit crossed the equatorial lowlands as well as high mountain regions at Ishtar Terra. Figure 3 shows that VIRTIS measurements of the thermal emission from the Venus nightside reveal a high variability of nearinfrared spectra over the planet. The main reasons for the observed differences are spatial variations of cloud optical depth and surface elevation, but their influence can be very clearly distinguished in dependence on wavelength as will be shown below. Window signatures at 2.3 and 1.7 mm are sometimes very small and may even disappear. This is due to an extreme cloud thickness in these cases. Radiation changes may be also due to changes in atmospheric temperature profile and absorber contents and also due to variations in the surface emissivity, but their influence is usually much smaller.

3. Preliminary Radiative Transfer Simulations [16] A radiative transfer model is used to simulate VIRTIS-M spectra. It allows for absorption, emission, and multiple scattering by atmospheric gaseous and particulate constituents in planetary atmospheres. The algorithm was described in detail by Haus and Titov [2000], who applied it to the atmosphere of Mars. Although Earth, Mars, and Venus differ substantially with respect to their total atmospheric thermal regime and composition, the basic method of radiative transfer simulation is the same. It can be applied to quantitative analyses even in the field of environmental and air pollution research [Haus et al., 1994, 1998]. The atmosphere of Venus above the dense cloud deck is rather similar to the conditions in the Earth’s stratosphere and mesosphere with respect to its thermal regime. Therefore, only minor

modifications are required in the radiative transfer algorithm to simulate VIRTIS dayside measurements. The extreme temperature and pressure regime in the deep atmosphere of Venus, however, makes high demands on new theoretical and methodical work including hot gaseous absorption bands, far spectral line wings and pressure-induced absorption. Much effort has been devoted to explaining the effects of spectral line shape under high temperature and pressure conditions [e.g., Burch et al., 1969; Tonkov et al., 1996; Filippov and Tonkov, 1998; Ma et al., 1999; Tvorogov and Rodimova, 1995; Afanasenko and Rodin, 2005], but up to now, the related works do not offer a self-sufficient model that is suitable for practical calculations over the full spectral range that is of interest for surface and deep atmosphere studies. This is the reason why currently most simulation algorithms make use of empirically determined continuum opacities to fit the observations. [17] Unless satisfactory models or laboratory measurements are available, the measured VIRTIS radiation spectra are used to determine the relative magnitude of continuum absorption as well as cloud absorption and scattering features from the signatures themselves. Therefore, the radiative transfer simulation that has been applied throughout this work is a preliminary one, but it will be further developed in near future. The main intention of this paper is to provide a feasibility study and an estimation of a quantitative approach to extracting surface information from VIRTIS data instead of elaborated quantitative retrievals. Nevertheless, the simulation results obtained so far look very promising as is shown below. [18] Look up tables of quasi-monochromatic absorption cross sections k (n, p, T) of gaseous constituents in the atmosphere of Venus are calculated on the basis of a lineby-line procedure, where n is the wave number [cm
1], p the atmospheric pressure [mbar], and T the temperature [K] covering the altitude range 0 – 140 km. Cross sections of SO2, HF, HCl, and OCS are evaluated based on the HITRAN 2004 catalog [Rothman et al., 2005] where a Voigt profile with a line cut of 125 cm
1 is used. H2O and CO cross-section calculations (line cut 125 cm
1) use the HITEMP 1995 catalog [Rothman et al., 1995]. Since this

4 of 13

E00B10

ARNOLD ET AL.: VENUS SURFACE DATA FROM VIRTIS/VEX

E00B10

Figure 4. Radiative transfer simulation results in a purely gaseous Venus atmosphere. Top three lines are pure CO2 for different spectral line databases; line with diamonds is H2O added to CO2-CDSD, bottom curve (triangle dn) is with continuum included. H2O mixing ratio is 25 ppm (0 – 46 km), Voigt line profiles, line cut 125 cm
1. older database only contains H2O lines of the main isotope 161, the minor isotopic species were added from HITRAN04. The HDO abundance is considered to be 150 times the telluric one [deBergh et al., 1991]. The even higher abundance in the upper atmosphere [Bertaux et al., 2007] is not considered at present. HITRAN04 and HITEMP95 were found to be inadequate for simulating near-infrared emissions from Venus nightside with respect to the main constituent, carbon dioxide. Carbon dioxide opacities were finally calculated by using line parameters from the high-temperature carbon dioxide spectroscopic databank (CDSD) compiled by Tashkun et al. [2003]. Figure 4 compares the results of a pure CO2 radiative transfer simulation in the spectral range 1.0– 1.45 mm when the three different databases are used. The high number of additional week spectral lines that are contained in the CDSD database yields many additional absorption features under the high temperature and pressure conditions in the deep Venusian atmosphere. Significant changes in comparison with HITRAN04 and HITEMP95 also occur at longer wavelengths. [19] The diamond line in Figure 4 marks the strong influence of water vapor on the 1.10 and 1.18 mm windows and the intermediate range. It is evident that the striking 1.10 mm window feature only appears when H2O is considered in the simulation. A constant volume mixing ratio of 25 ppm from 0 to 46 km altitude was used. A comparison of this pure gas simulation with VIRTIS measurements (compare Figures 2 and 3) reveals a disagreement of spectral profiles in the range between the 1.10 and 1.18 mm windows. It is known that water vapor spectral lines exhibit a collision-induced superLorentzian behavior leading to higher wing absorptions compared to a pure Lorentz (or Voigt) profile. This effect was not considered here. Contrary to water vapor, carbon dioxide lines are rather characterized by a sub-Lorentz profile, which provides less absorption in the distant line wings. The calculated absorption cross sections k (n 0, p, T) also strongly depend on the line cut criterion that is

applied to suppress contributions from very distant lines relative to the wave number n 0. Different sub-Lorentz line profiles of the simple form fL(n
n 0)exp[
a(jn
n 0j
s)] for jn
n 0j > s as well as more sophisticated profiles from the literature [Fukabori et al., 1986; Perrin and Hartmann, 1989; Pollack et al., 1993; Meadows and Crisp, 1996; Tonkov et al., 1996], combined with different line cut assumptions (125 –1000 cm
1), are currently under investigation with respect to individual band positions of CO2, but a final definition of a ‘‘best choice model’’ could not be provided up to now. As was mentioned above, pressure-induced absorption is currently not included in the simulations. [20] Empirical continuum absorption coefficients have been used so far in window ranges where it is needed. The continuum cross section kc is currently determined from a Venus ‘‘reference spectrum’’ (compare Figure 5) when the clouds are included in the simulation procedure. The bottom line (inverted triangle) in Figure 4 shows the influence of continuum absorption on a pure gas simulation in the 1.0– 1.45 mm range. With the continuum included, the simulated relative peak values and profiles of the window signatures are much closer to the observed ones. [21] Mie scattering theory is applied to derive the microphysical parameters of the H2SO4 clouds, where a four modal log normal size distribution is used (modes 1, 2, 20, and 3 according to Pollack et al. [1993]). Optical depths of cloud absorption, scattering and extinction are calculated on the basis of a particle number density model that essentially corresponds to that described by Moroz [2002b]. The total cloud opacity of a so-called standard model, which is applied throughout this paper, is 26.9 at 1 mm. Unity total optical depth occurs near the upper limit of the main cloud deck at an altitude of 70 km. There are many open questions with respect to the chemical composition and the particle size distribution of the clouds [Moroz, 2002b; Taylor, 2006]. It is well established that the upper cloud particles (mainly mode 2 and mode 20) consist of a water solution of sulphuric acid (75% H2SO4 by weight). The composition of the

5 of 13

E00B10

ARNOLD ET AL.: VENUS SURFACE DATA FROM VIRTIS/VEX

E00B10

Figure 5. Comparison of VIRTIS measurement for orbit 90, 30°N (solid line) and radiative transfer simulation (dashed line). The surface elevation is
1.0 km. The integers refer to the band number introduced in Figure 2. This spectrum was used as reference spectrum for the simulation model development. smaller mode 1 droplets is unknown. The large mode 3 particles are probably highly variable in composition and spatial and temporal distribution. [22] Rayleigh scattering optical depths are considered according to Hansen and Travis [1974]. The total Rayleigh optical depth at 1 mm is 1.54; that is, it is not negligible within the short-wavelength windows. Radiation multiple scattering in the dense cloudy atmosphere is considered by a successive order procedure where the source functions are calculated in a two-stream approximation of the angulardependent radiation field [Arnold et al., 2000; Haus and Titov, 2000]. [23] The synthetic quasi-monochromatic intensity spectra at the model top level of the atmosphere (140 km) that are generated by the radiative transfer code are convolved with the VIRTIS spectral response function. A constant spectral resolution of 10 nm and a Gaussian instrumental response function have been used so far. In the wave number frame, 10 nm correspond to 100 cm
1 at 10000 cm
1 (1 mm), 25 cm
1 at 5000 cm
1 (2 mm), and 4 cm
1 at 2000 cm
1 (5 mm), respectively. [24] Figure 5 compares a synthetic spectrum with the VIRTIS measurement performed at 30°N on orbit 90. The surface elevation according to Magellan topography is 1.04 km there. This spectrum was selected as the ‘‘reference spectrum’’ to investigate the combined effects of cloud scattering and gas continuum absorption. The spectral range from 1 to 2.6 mm shown in Figure 5 covers the atmospheric windows located at 1.02, 1.10, 1.18, 1.28, 1.31, 1.74, and 2.3 mm. The simulation was performed for the Venus standard atmosphere [Seiff et al., 1985]. An update of these VIRA data by VIRA-2 [Moroz and Zasova, 1997] was not required for the investigations intended here, since the temperature and pressure profiles below 50 km remained unchanged. The calculations indicate that temperature changes within and above the clouds do not influence the surface and deep atmosphere signals. [25] The preliminary quantitative simulation approach yields a more or less good fit to the observed spectrum,

although there are some discrepancies (mainly at 1.7 and 2.3 mm). They are probably due to an inappropriate choice of spectral line profiles, line cut and continuum parameters. Additionally, the lower atmospheric temperature profile may be slightly different from the VIRA standard profile. Small changes can result in significant window flank shifts as corresponding calculations have shown. This will be investigated in more detail in near future. The present gas simulation uses Voigt line shapes and a line cut of 125 cm
1 everywhere for each molecular constituent. Pressure-induced continuum absorption is assumed to depend on the total molecular density squared. It is determined as a correction value that must be applied to fit the measured reference spectrum when the total aerosol and Rayleigh optical depths of the atmosphere are 26.9 and 1.54 at 1 mm, respectively. The following values (in units of [cm
1 amagat
2]) were used in the simulation shown in Figure 5: windows 3 – 5 (1.18, 1.28, 1.31 mm), 1.56  10
09; window 6 (1.74 mm), 6.23  10
09; window 7 (2.30 mm), 4.17  10
08. Pollack et al. [1993] derived the values 7.00  10
09 and 2.50  10
08 for windows 6 and 7, respectively, while Marcq et al. [2006] used a value of 3.5  10
8 at 2.3 mm. The water vapor signatures, which appear in the pure gas simulation between 1.10 and 1.18 mm (compare Figure 4) due to the neglect of super-Lorentz line shapes, have been currently removed by applying a continuum in the order of 1.0  10
8 at 1.13 mm. All the continuum absorption coefficients are assumed to be independent of wavelength throughout a given window range. They are also not varied from spectrum to spectrum; that is, the values are held constant for simulations with respect to different VIRTIS measurements.

4. Results [26] Figure 6 illustrates the simulated radiances at the top level of the Venus atmosphere for cloud optical depths (opacities) that differ from the standard model by a factor of 0.7, 1.0, 1.4, 2.0, and 2.5, respectively. The surface

6 of 13

E00B10

ARNOLD ET AL.: VENUS SURFACE DATA FROM VIRTIS/VEX

E00B10

Figure 6. Simulated nightside radiance spectra in dependence on cloud opacity factor. Radiances decrease with increasing factor. Factor 1 corresponds to a total cloud optical depth of 26.9 at 1 mm. The surface elevation is
1.0 km. elevation is constant at a value of
1.04 km. Each window is sensitive to changes in the cloud opacity, but the contrasts increase toward longer wavelengths. Cloud contrast is most apparent within the 2.3 mm window. The aerosol single scattering albedo below 1.5 mm is very close to unity; that is, conservative scattering is the dominating cloud feature. In other words, the clouds do not contribute to thermal emission, but act as a strong attenuator of radiation that originates from below the cloud deck and emerges into space. Little spectral dependence of upwelling radiation is added by the clouds below 1.5 mm. This is the reason why it should be possible to use multispectral radiation measurements of the Venus surface and lower atmosphere in the 1.0 – 1.35 mm window complex to eliminate cloud interferences. [27] Figure 7 illustrates the change of averaged window radiances in dependence on the cloud opacity factor. Averaging means that mean values over each window range are

used instead of the maximum value. The ranges [mm] were defined as follows. W1, 1.028– 1.035; W2, 1.090 – 1.110; W3, 1.160 – 1.190; W4, 1.255 – 1.289; W5, 1.300 – 1.309; W6, 1.691 – 1.780; W7, 2.200 – 2.290. Only the shortwavelength part of window 7 at 2.3 mm was used here to eliminate possible minor constituent variations (H2O, CO, COS, SO2, HF) in the long-wavelength part at the right. The left-hand part is exclusively determined by CO2. Water vapor concentration changes, however, which may affect the windows at 1.74, 1.18, and 1.10 mm, respectively, were not eliminated at present. [28] As expected, there is a dependence of radiation on cloud opacity in the windows 1– 5 that leads to constant radiance ratios. There is an increasing contribution of cloud absorption with increasing wavelength. Thus, the linear dependence is lost and the radiances at 1.7 and 2.3 mm decrease much faster with growing cloud opacity. As a

Figure 7. Simulated averaged window radiances as a function of cloud opacity factor. The integers in the box refer to the band number introduced in Figure 2. The surface elevation is
1.0 km. 7 of 13

E00B10

ARNOLD ET AL.: VENUS SURFACE DATA FROM VIRTIS/VEX

E00B10

Figure 8. Comparison of measured (scatter points) and simulated (lines) averaged window radiance ratios as a function of 2.3 mm averaged window radiance. The scatterplot of binned measured ratios includes all latitudes from the 19 orbits of groups 1 – 3 (compare Figure 1), where the surface elevation is constrained to the interval [
1500 m,
500 m]. The integers in the box refer to the band numbers introduced in Figure 2. Values included in parentheses denote the corresponding nominal wavelengths. F is the cloud opacity factor. Nearly constant ratios for the surface windows are apparent. consequence, a cloud correction algorithm, which makes use of radiance ratios, tends to fail for longer wavelengths. [29] Figure 8 compares measured and simulated averaged window radiance ratios as a function of the wavelengthaveraged radiance that was measured at 2.3 mm. The radiance at 2.3 mm is most sensitive to changes in cloud optical depth as is clearly visible in Figures 6 and 7, and also in Figure 3. It is insensitive to changes in surface elevation (compare Figures 9 and 10). Cloud opacity is not a free parameter in the measurements as it is in the simulations. It cannot be measured directly by the VIRTIS instrument. But the observed strong variability at the lefthand flank of the 2.3 mm window is a good indicator of cloud thickness on the Venus nightside. Thus, a scatterplot of measured radiance ratios against the measured 2.3 mm

radiance, which decreases with increasing cloud opacity, has the same physical background as a plot of simulated ratios against the cloud opacity. This is convincingly demonstrated in Figure 8. The scatterplot is restricted to measurements from the 19 orbits shown in Figure 1, and the surface elevation according to Magellan topography is constrained to the interval [
1500 m,
500 m]. The restriction to a narrow elevation interval is necessary to omit strong radiation changes due to topographical changes (compare Figure 9). It is assumed that the quantitative differences between measurement plot and simulation result, which are visible in the case of 6/1 and 7/1 ratios, are due to present inaccuracies in the spectral line profiles and continuum coefficients at these wavelengths.

Figure 9. Simulated nightside radiance spectra in dependence on surface elevation for the standard cloud model (opacity factor 1.0). Radiances decrease with increasing elevation in the surface windows. The 1.74 and 2.3 mm windows are unaffected. 8 of 13

E00B10

ARNOLD ET AL.: VENUS SURFACE DATA FROM VIRTIS/VEX

E00B10

Figure 10. Simulated averaged window radiances as a function of surface elevation. The integers in the box refer to the band numbers introduced in Figure 2. The cloud opacity factor is 1.0. [30] Another important conclusion can be derived from Figure 8. All the measurement points are distributed up to a maximum 2.3 mm radiance value of about 0.065 W/(m2 sr mm). This value corresponds to a cloud opacity factor of 0.95 (compare Figure 7) and seems to indicate that the assumed standard cloud optical depth of 26.9 (opacity factor 1.0) is rather the lower limit of what is present in the real Venusian atmosphere. Since continuum parameters were derived based on this standard model, a possible conclusion is that not only higher cloud opacities, but also lower continuum values are closer to reality than those adopted in the present ‘‘standard case model.’’ This will be addressed in future investigation. [31] The possibility to eliminate cloud signals from the VIRTIS measurements below 1.5 mm by the use of radiance ratios is of decisive importance for our future work that will mainly focus on Venus surface data extraction and the study of surface and lower atmosphere interactions. [32] Figures 9 and 10 illustrate the change of simulated radiances with surface elevation and wavelength. The cloud opacity was held constant in the simulations. The windows at 1.02, 1.10, and 1.18 mm exhibit a clear dependence of transmitted radiation on topographical features and, thus, on surface thermal emission. The radiance decreases considerably with increasing elevation, since high-elevation surfaces are substantially cooler and emit less thermal radiation. An elevation change of 12 km results in a surface temperature change of almost 100 K. The window at 1.28 mm still shows a minor influence of surface temperature, while it is mainly determined by features of the lower atmosphere (15 – 30 km). The surface elevation (temperature) does not observably influence the 1.31, 1.74, and 2.3 mm windows radiances, respectively. A simulation was performed, where the surface contribution was excluded assuming a surface emissivity value of zero. Thus, a quantitative estimate of the surface thermal emission contribution to emission within the Venus atmospheric windows was obtained. In the windows 1 (1.02 mm), 2 (1.10 mm), 3 (1.18 mm), and 4 (1.28 mm), the surface contributes 97 (95) %, 59 (60) %, 27 (40) %, and 1.3 (<2) %

to the total emission. The values in parentheses are the corresponding results of Meadows and Crisp [1996]. [33] Figure 11 compares measured and simulated averaged window radiance ratios as a function of surface elevation. Since the radiance at 1.02 mm is most sensitive against elevation changes, the ratios are not constant and, thus, a valuable measure of surface topography and surface temperature, respectively. The scatterplot contains all measurement data bins from the 19 orbits shown in Figure 1. As Figure 11 shows, these orbits pass predominantly over lowlands with an elevation between
1 km and 0 km. Another maximum occurs between 2 and 4 km (Beta Regio, Ishtar Terra). There is again a very good qualitative correlation between measurements and simulation.

Figure 11. Comparison of measured (scatter points) and simulated (lines) averaged window radiance ratios as a function of surface elevation. The scatterplot of measured ratios includes all bins from the 19 orbits of groups 1 – 3 (compare Figure 1). The integers in the box refer to the band numbers introduced in Figure 2. Values included in parentheses denote the corresponding nominal wavelengths. The cloud opacity factor in the simulation is 1.0.

9 of 13

E00B10

ARNOLD ET AL.: VENUS SURFACE DATA FROM VIRTIS/VEX

E00B10

Figure 12. Averaged window radiances at 1.02, 1.18, and 2.30 mm and the radiance ratios 1.18/1.02 in comparison with surface elevation according to Magellan topography as a function of latitude for orbits (a – b) 90, (c– d) 113, and (e – f) 150. The ratios are based on measured radiances, which have been displayed and discussed before by Arnold et al. [2007]. [34] Figures 12a– 12f yield additional evidence of the important finding, that emission window radiance ratios can be used to separate atmospheric and surface properties in the measured Venus spectra. It shows the radiances at 1.10, 1.18, and 2.30 mm and the radiance ratio 1.18/1.02 in comparison with surface elevation according to Magellan topography as a function of latitude for three selected orbits. Orbit 90 (Figures 12a and 12b) at 250°E is representative for group 1 measurements over predominantly low elevation areas in the northern hemisphere of Venus. Orbit 113 (Figures 12c and 12d) touched the center of Beta Regio and the deep depressions of Guinevere Planitia at 283°E, while orbit 150 (Figures 12e and 12f) passed over the highlands of Ishtar Terra at 339°E. Higher elevations should result in

lower radiances of the windows 1 (1.02mm) and 3 (1.18 mm), and vice versa. This is the case at Beta Regio, Guinevere Planitia, and Ishtar Terra in comparison to the surrounding terrains. But it is rather hard or even impossible to verify this behavior for smaller topographical variations. The window 1 and 3 measurements on orbit 90 (Figure 12a), for example, indicate higher radiances at 35°N, but this has nothing to do with an elevation decrease. The high window 7 (2.3 mm) radiance value at this location proves that much lower cloud opacity is the cause. The same holds true on this orbit at 60°N. The ratio 1.18/1.02 (Figure 12b), on the other hand, follows the elevation profile in most details. The elevation subsidiary summit at Ishtar Terra on orbit 150 at 65°N (Figure 12e) is not reflected in decreasing window 1

10 of 13

E00B10

ARNOLD ET AL.: VENUS SURFACE DATA FROM VIRTIS/VEX

E00B10

Figure 13. Retrieval of surface topography of group 1 and 2 orbits near Beta Regio using 1.18 mm to 1.02 mm radiance ratios. (a) ‘‘The target state’’ of topography retrieval according to Magellan data, which were smoothed over the swaths with respect to the 100 km resolution limit to be comparable with the VIRTIS data. (b) ‘‘The actual state’’ of topography retrieval. The ratio for each radiance measurement on a pixel basis is mapped to the elevation according to the linear fit of measurement data shown in Figure 11 (1.18/1.02 versus Magellan elevation) and depicted in the same color coding as the Magellan elevation. The black edging is used to stress the footprints. and 3 radiances, since it is masked by a local cloud decrease (enhanced window 7 radiance). The ratio 1.18/1.02 (Figure 12f) shows this feature very well. [35] The high correlation between the ratios of radiances measured in the surface windows between 1.0 and 1.35 mm and the planetary topography can be used to extract surface data of Venus. Figure 13 demonstrates this capability for some parts of orbits in the vicinity of Beta Regio including the orbits of groups 1 and 2. Figure 13a shows the topographical ‘‘target state’’ as it is defined by the Magellan data, which were smoothed over the swaths with respect to the 100 km resolution limit to be comparable with the VIRTIS data. Figure 13b is the ‘‘actual state’’ of the topography retrieval from VIRTIS data, which is obtained from a linear regression of the r3/r1 (1.18/1.02) scatterplot shown in Figure 11. The ratio r3/r1 as a function of the Magellan elevation h reads r3/r1(h) = (3.08 ± 0.01) + (0.319 ± 0.008)  h/km, with an error of twice the standard deviation. The ratio for each radiance measurement on a pixel basis is mapped to the elevation according to the linear fit of measurement data shown in Figure 11 and depicted in the same color coding as the Magellan elevation. The black edging in Figures 13a and 13b is used to stress the footprints.

[36] In general, the VIRTIS and Magellan topographies are largely in agreement with each other, but differences occur in certain small areas. These areas will be of special interest for future data analyses. The examination of differences at comparable elevations and surface temperatures may give hints either to local variations of the deep atmospheric temperature profiles and composition or to different surface materials. [37] As discussed above, the window 1 at 1.02 mm contributes more than 95% to the total emission of the Venus nightside. Thus, any relative change in the surface emissivity corresponds to approximately the same relative radiance change. It should be possible to detect a difference of a few percent. According to Hashimoto and Sugita [2003], there may be spatial variations in the surface emissivity of Venus as large as 20%, which correspond to the difference between granitic (rhyolitic) and basaltic rocks due to a different iron oxide content in the rocks.

5. Conclusions [38] The VIRTIS-M nightside IR data form a valuable basis for systematic and continuous surface and deep atmosphere studies of Venus. The observed high variability

11 of 13

E00B10

ARNOLD ET AL.: VENUS SURFACE DATA FROM VIRTIS/VEX

of measured signatures is mainly due to spatial variations of cloud optical depth and surface elevation. The conservative character of cloud scattering below 1.5 mm, where the clouds do not contribute to thermal emission, makes it possible to use multispectral radiation measurements of the Venus surface and lower atmosphere emission in the 1.0– 1.35 mm window complex to separate cloud interference and surface influences. [39] The radiance in the left-hand flank of the 2.3 mm window does not depend on surface features and minor constituent distribution changes in the lower atmosphere. It is, therefore, a good indicator of cloud thickness on the Venus nightside and can be used for quantitative estimates. From the comparison of measured and simulated averaged radiance ratios as a function of 2.3 mm averaged radiance and, hence, of cloud opacity, it can be concluded that a cloud optical depth in the order of 27 is a lower limit with respect to the VIRTIS data that have been investigated so far. [40] Measurements as well as radiative transfer simulations have proven a high correlation between the radiance ratios in the emission windows between 1.0 and 1.35 mm and the surface elevation. This allows a mapping of surface topography and the retrieval of surface temperature. On the basis of measured radiances at 1.18 and 1.02 mm and their ratios, the surface topography in selected regions around Beta Regio has been retrieved and compared with the Magellan data. The way the results largely agree proves that the ratio method to decloud the VIRTIS data is well suited to extract surface information. Differences have been observed in certain small areas. To the first order, the radiance ratios are a function of surface temperature. Small deviations from this first-order dependence are possibly due to surface materials that exhibit different emissivity characteristics. Mu¨ller et al. [2008] and G. L. Hashimoto et al. (Galileo near-infrared mapping spectrometer data suggests felsic highland crust on Venus, submitted to Journal of Geophysical Research, 2008) deal with observed emissivity variations. [41] The preliminary radiative transfer simulations have demonstrated the capability of the algorithm to investigate the surface of Venus. Future studies will focus on a separation of atmospheric and surface radiance contributions in the measured VIRTIS signals and a possible separation of surface temperature and emissivity by a comparative investigation of different surface windows and their radiance ratios. Since the orbit repetition increases during the mission operation, it will be possible to distinguish between static and dynamic local variations and to gather more detailed information about the surface of Venus on a global scale. [42] These objectives can only be achieved by further progress in the radiative transfer simulations to eliminate the masking of the Venus nightside windows by far wing and pressure-induced absorptions of the deep atmosphere constituents. A great deal of theoretical and numerical effort will be required, including taking appropriate spectral line databases and line profiles into account, as well as revised models of cloud composition. [43] Acknowledgments. We acknowledge the work of the entire Venus Express team which allowed these data to be obtained.

E00B10

References Afanasenko, T. S., and A. V. Rodin (2005), The effect of collisional line broadening on the spectrum and fluxes of thermal radiation in the lower atmosphere of Venus, Sol. Syst. Res., 39(3), 187 – 198, doi:10.1007/ s11208-005-0034-1. Arnold, G., R. Haus, and H. Hirsch (2000), The Planetary Fourier Spectrometer (PFS): Studies of the Martian Atmosphere and Surface, Res. Rep. 2000-18, 114 pp., German Aerosp. Cent., Inst. fu¨r Weltraumsensorik und Planetenerkundung, Berlin. Arnold, G., W. Do¨hler, R. Haus, D. Kappel, P. Drossart, G. Piccioni, and the VIRTIS Team (2007), Estimation of a quantitative approach for Venus surface data extraction from VIRTIS measurements using topographical variations, Geophys. Res. Abstr., 9, 07972. Baines, K. H., et al. (2000), Detection of sub-micron radiation from the surface of Venus by Cassini/VIMS, Icarus, 148, 307 – 311, doi:10.1006/ icar.2000.6519. Baines, K. H., S. Atreya, R. W. Carlson, D. Crisp, P. Drossart, V. Formisano, S. S. Limaye, W. J. Markiewicz, and G. Piccioni (2006), To the depths of Venus: Exploring the deep atmosphere and surface of our sister world with Venus Express, Planet. Space Sci., 54, 1263 – 1278, doi:10.1016/j.pss.2006.04.034. Basilevsky, A. T., and J. W. Head III (1988), The geology of Venus, Annu. Rev. Earth Planet. Sci., 16, 295 – 317, doi:10.1146/annurev.ea. 16.050188.001455. Basilevsky, A. T., and J. W. Head III (2002), Venus: Timing and rates of geologic activity, Geology, 30, 1015 – 1018, doi:10.1130/00917613(2002)030<1015:VTAROG>2.0.CO;2. Bertaux, J. L., et al. (2007), A warm layer in Venus’ cryosphere and highaltitude measurements of HF, HCl, H2O and HDO, Nature, 450, 469 – 646, doi:10.1038/nature05974. Burch, D. E., D. A. Gryvnak, R. R. Patty, and C. E. Bartky (1969), Absorption of infrared radiant energy by CO2 and H2O. IV. Shapes of collision broadened lines, J. Opt. Soc. Am., 59, 267 – 280. Carlson, R. W., K. H. Baines, L. W. Kamp, P. R. Weissman, W. D. Smythe, A. C. Ocampo, T. V. Johnson, D. L. Matson, J. B. Pollack, and D. Grinspoon (1991), Galileo infrared imaging spectroscopy measurements at Venus, Science, 253, 1541 – 1548, doi:10.1126/science.253.5027.1541. deBergh, C., B. Be´zard, T. Owen, D. Crisp, J.-P. Maillard, and B. L. Lutz (1991), Deuterium on Venus: Observations from the Earth, Science, 251, 547 – 549, doi:10.1126/science.251.4993.547. Drossart, P., et al. (2007), A dynamic upper atmosphere of Venus as revealed by VIRTIS on Venus Express, Nature, 450, 641 – 645, doi:10.1038/nature06140. Filippov, N. N., and M. V. Tonkov (1998), Theory of band shapes in molecular spectra of gases: Application to band wings, J. Chem. Phys., 108, 3608 – 3619, doi:10.1063/1.475755. Fukabori, N., T. Nakazawa, and M. Tanaka (1986), Absorption properties of infrared active gases at high pressures—I. CO2, J. Quant. Spectrosc. Radiat. Transfer, 36, 265 – 270, doi:10.1016/0022-4073(86)90074-9. Hansen, J. E., and L. D. Travis (1974), Light scattering in planetary atmospheres, Space Sci. Rev., 16, 527 – 610, doi:10.1007/BF00168069. Hashimoto, G. L., and S. Sugita (2003), On observing the compositional variability of the surface of Venus using nightside near-infrared thermal radiation, J. Geophys. Res., 108(E9), 5109, doi:10.1029/2003JE002082. Haus, R., and D. V. Titov (2000), PFS on Mars Express: Preparing the analysis of infrared spectra to be measured by the planetary Fourier spectrometer, Planet. Space Sci., 48, 1357 – 1376, doi:10.1016/S00320633(00)00116-1. Haus, R., K. Scha¨fer, W. Bautzer, J. Heland, H. Mosebach, H. Bittner, and T. Eisenmann (1994), Mobile FTIS-monitoring of air pollution, Appl. Opt., 33(24), 5682 – 5689. Haus, R., R. Wilkinson, J. Heland, and K. Scha¨fer (1998), Remote sensing of gas emissions on natural gas flares, Pure Appl. Opt., 7(4), 853 – 862, doi:10.1088/0963-9659/7/4/020. Head, J. W., L. S. Crumpler, J. C. Aubele, J. E. Guest, and R. S. Saunders (1992), Venus volcanism: Classification of volcanic features and structures, associations, and global distribution from Magellan data, J. Geophys. Res., 97, 13,153 – 13,198. Ma, Q., R. H. Tipping, C. Boulet, and J.-P. Bouanich (1999), Theoretical far-wing line shape and absorption for high-temperature CO2, Appl. Opt., 38, 599 – 604, doi:10.1364/AO.38.000599. Marcq, E., T. Encrenaz, B. Be´zard, and M. Birlan (2006), Remote sensing of Venus’ lower atmosphere from ground-based IR spectroscopy: Latitudinal and vertical distribution of minor species, Planet. Space Sci., 54, 1360 – 1370, doi:10.1016/j.pss.2006.04.024. McKenzie, D., P. G. Ford, F. Liu, and G. H. Pettengill (1992), Pancakelike domes on Venus, J. Geophys. Res., 97, 15,967 – 15,976. Meadows, V. S., and D. Crisp (1996), Ground-based near-infrared observations of the Venus nightside: The thermal structure and water abundance

12 of 13

E00B10

ARNOLD ET AL.: VENUS SURFACE DATA FROM VIRTIS/VEX

near the surface, J. Geophys. Res., 101(E2), 4595 – 4622, doi:10.1029/ 95JE03567. Moroz, V. I. (2002a), Estimates of visibility of the surface of Venus from decent probes and balloons, Planet. Space Sci., 50, 287 – 297, doi:10.1016/S0032-0633(01)00128-3. Moroz, V. I. (2002b), Studies of the atmosphere of Venus by means of spacecraft: Solved and unsolved problems, Adv. Space Res., 29, 215 – 225, doi:10.1016/S0273-1177(01)00571-3. Moroz, V. I., and L. V. Zasova (1997), VIRA-2: A review of inputs for updating the Venus International Atmosphere, Adv. Space Res., 19, 1191 – 1201. Mu¨ller, N., J. Helbert, G. L. Hashimoto, C. C. C. Tsang, S. Erard, G. Piccioni, and P. Drossart (2008), Venus surface thermal emission at one micrometer in VIRTIS imaging observations: Evidence for variation of crust and mantle differentiation conditions, J. Geophys. Res., doi:10.1029/2008JE003118, in press. Perrin, M. Y., and J. M. Hartmann (1989), Temperature-dependent measurements and modeling of absorption by CO2-N2 mixtures in the far linewings of the 4.3 mm CO2 band, J. Quant. Spectrosc. Radiat. Transfer, 42, 311 – 317. Piccioni, G., et al. (2007), South polar features on Venus similar to those near the north pole, Nature, 450, 637 – 640, doi:10.1038/nature06209. Pollack, J. B., et al. (1993), Near-infrared light from Venus’ nightside: A spectroscopic analysis, Icarus, 103, 1 – 42, doi:10.1006/icar.1993.1055. Rothman, L. S., R. B. Wattson, R. R. Gamache, J. Schroeder, and A. McCann (1995), HITRAN, HAWKS and HITEMP high-temperature molecular database, Proc. Soc. Photoopt. Instrument. Eng., 2471, 105 – 111. Rothman, L. S., et al. (2005), The HITRAN 2004 molecular spectroscopic database, J. Quant. Spectrosc. Radiat. Transfer, 96, 139 – 204, doi:10.1016/j.jqsrt.2004.10.008. Saunders, R. S., et al. (1992), Magellan mission summary, J. Geophys. Res., 97, 13,067 – 13,090. Seiff, A., J. T. Schofield, A. J. Kliore, F. W. Taylor, S. S. Limaye, H. E. Revercomb, L. A. Sromovsky, V. V. Kerzhanovich, V. I. Moroz, and

E00B10

M. Y. Marov (1985), Models of the structure of the atmosphere of Venus from the surface to 100 kilometers altitude, Adv. Space Res., 5, 3 – 58, doi:10.1016/0273-1177(85)90197-8. Solomon, S. C., et al. (1992), Venus tectonics: An overview of Magellan observations, J. Geophys. Res., 97, 13,199 – 13,256, doi:10.1029/ 91JA02410. Tashkun, S. A., V. I. Perevalov, J.-L. Teffo, A. D. Bykov, and N. N. Lavrentieva (2003), CDSD-1000, the high-temperature carbon dioxide spectroscopic databank, J. Quant. Spectrosc. Radiat. Transfer, 82, 165 – 196, doi:10.1016/S0022-4073(03)00152-3. Taylor, F. W. (2006), Venus before Venus Express, Planet. Space Sci., 54, 1249 – 1262, doi:10.1016/j.pss.2006.04.031. Tonkov, M. V., N. N. Filippov, V. V. Bertsev, J. P. Bouanich, N. Van-Thanh, C. Brodbeck, J. M. Hartmann, C. Boulet, F. Thibault, and R. Le Doucen (1996), Measurements and empirical modeling of pure CO2 absorption in the 2.3 mm range at room temperature: Far wings, allowed and collisioninduced bands, Appl. Opt., 35, 4863 – 4870. Tvorogov, S. D., and O. B. Rodimova (1995), Spectral line shape. I. Kinetic equation for arbitrary frequency detunings, J. Chem. Phys., 102, 8736 – 8745, doi:10.1063/1.468977.























G. Arnold and D. Kappel, Institute of Planetary Research, DLR, Rutherfordstrasse 2, D-12489 Berlin, Germany. ([email protected]; [email protected]) P. Drossart, LESIA, Observatoire de Paris, CNRS, UPMC, Universite´ Paris-Diderot, 5 place Jules Janssen, F-92195 Meudon, France. R. Haus, Department of Marine Remote Sensing, Remote Sensing Technology Institute, DLR, Rutherfordstrasse 2, D-12489 Berlin, Germany. G. Piccioni, INAF-IASF, via del Fosso del Cavaliere 100, I-00133 Rome, Italy.

13 of 13

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114, E00B41, doi:10.1029/2008JE003156, 2009

for

Full Article

Venus Express bistatic radar: High-elevation anomalous reflectivity Richard A. Simpson,1 G. Leonard Tyler,1 Bernd Ha¨usler,2 Riccardo Mattei,2 and Martin Pa¨tzold3 Received 31 March 2008; revised 1 April 2009; accepted 17 April 2009; published 27 June 2009.

[1] Magellan (MGN) bistatic radar observations in 1994 confirmed earlier Pioneer

Venus reports of unusual Venus surface reflectivity and emissivity at elevations above 6054 km radius. They also revealed that the anomalous values of surface dielectric constant e near Cleopatra Patera included a large imaginary component (e
i 100) at 13 cm wavelength, consistent with a semiconducting surface material. The MGN observations were conducted using a linearly polarized wave, canted at 45° with respect to the plane of incidence and radiated by the MGN synthetic aperture radar antenna toward the specularly reflecting region of the mean planetary surface. In 2006 similar experiments were conducted using 13 cm circularly polarized transmissions from Venus Express (VEX). The VEX signal-to-noise ratio (SNR) was lower than that of MGN, but elevated jej has been inferred broadly over Maxwell Montes. A quasi-specular echo was detected near Cleopatra but with insufficient SNR to address the question of conductivity. An early failure of the VEX 13 cm radio system precludes further measurements with VEX. Citation: Simpson, R. A., G. L. Tyler, B. Ha¨usler, R. Mattei, and M. Pa¨tzold (2009), Venus Express bistatic radar: High-elevation anomalous reflectivity, J. Geophys. Res., 114, E00B41, doi:10.1029/2008JE003156.

1. Introduction [2] Early Earth-based radar images of Venus revealed three very bright areas: ‘‘Alpha’’ and ‘‘Beta’’ near the equator [Goldstein, 1965; Rogers and Ingalls, 1970] and ‘‘Maxwell’’ at 70°N latitude [Jurgens, 1970]. On the basis of then-current understanding, investigators hypothesized that these were regions of very rough surface texture, where highly tilted surface elements preferentially backscattered radar signals toward Earth. On the Moon the large, young crater Tycho had been identified in range-Doppler data as a bright scattering center [Pettengill, 1962]. But the bright features on Venus did not have the appearance of craters, nor did the brightness of Alpha and Beta (both near the equator) vary with the planet’s rotation, as might have been expected had there been a significant specular component to the scattering. [3] The Pioneer Venus Orbiter (PVO) radar altimeter provided the first topographic model for Venus [Pettengill et al., 1980] as well as measurement of the planet’s backscattering parameters and emissivity. Ford and Pettengill [1983] reported that the areas of unusually high reflectivity corresponded to those with low emissivity and that both depended on altitude. With these results, opinion within the planetary community shifted to favor the presence of high 1 Space, Telecommunications, and Radioscience Laboratory, Stanford University, Stanford, California, USA. 2 Institut fu¨r Raumfahrttechnik, Universita¨t der Bundeswehr Mu¨nchen, Neubiberg, Germany. 3 Rheinisches Institut fu¨r Umweltforschung, Cologne, Germany.

Copyright 2009 by the American Geophysical Union. 0148-0227/09/2008JE003156$09.00

values of surface dielectric constant, e, rather than roughness, as the important surface property underlying the anomalous brightness. The question then became ‘‘What materials could be present on Venus that had e values well above the normal range for terrestrial analogs?’’ Iron pyrite was an early candidate [Ford and Pettengill, 1983]; ferroelectric and other materials have been proposed subsequently [Shepard et al., 1994]. [4] Magellan (MGN) used its side-looking radar, nadir altimeter, and radiometer to map Venus at much higher resolution than PVO, confirming that anomalous reflectivity on Venus occurred globally at radii greater than about 6054 km [Pettengill et al., 1992]. Magellan also showed that the anomalous behavior was less pronounced at the highest elevations, although the thresholds varied somewhat with latitude [Klose et al., 1992]. A unique Magellan contribution was implementation of several bistatic radar (BSR) experiments near the end of its mission. Forwardscattered, quasi-specular reflections from the area around Cleopatra Patera on Maxwell Montes contained a circularly polarized component which could be explained only if the surface material were moderately conducting. For this, Pettengill et al. [1996] proposed a surface composition including elemental tellurium, a trace constituent of terrestrial volcanic gases with a melting point of 723 K. This temperature corresponds to 6054 km elevation in the atmosphere of Venus, above which tellurium can exist as a solid. Schaefer and Fegley [2004] have since shown that elemental tellurium is not stable on Venus and favor certain lead and bismuth compounds. [5] Venus Express (VEX) entered orbit in April 2006, providing a new opportunity for bistatic radar experiments

E00B41

1 of 13

SIMPSON ET AL.: VENUS EXPRESS BISTATIC RADAR

E00B41 Table 1. Bistatic Radar Parameters

Magellan Venus Express Venus Express S Band S Band X Band

Parameter

Spacecraft Transmitter Frequency, f (MHz) 2298 Wavelength, l (cm) 13.1 Polarization Linear 5 Power, PT (watts) Antenna diameter (m) 3.7 35.9 Antenna gain, GT (dB) 1.1 Antenna beam widtha (°) Earth Receiver Antenna diameter (m) 70 63.3 Antenna gain, GR (dB) System temperature (K) 22 – 26 Polarization RCP and LCP

2296 13.1 RCP 5 1.3 25.7 3.0

8420 3.6 RCP 65 1.3 37.0 0.8

70 63.3 22 – 30 RCP and LCP

70 74.3 24 – 26 RCP and LCP

E00B41

insufficient RCP power to support estimation of jej at incidence angles 8
35°. Bistatic radar experiments were then suspended. This paper summarizes the methods used for analyzing the circularly polarized data and the application to 13 cm data acquired from Maxwell Montes in June 2006.

2. Experiment Overview

a

Antenna beam width is given as the angular distance between the boresight (peak value) and the half power point.

[Ha¨usler et al., 2006]. Magellan’s experiments were conducted from a near-circular, 93 min orbit at both 13 and 3.6 cm wavelengths (Table 1), but only 13 cm echoes were detected from Maxwell Montes. The VEX 13 cm transmitter radiated 5 W, the same power as Magellan but through a smaller antenna, while VEX polarization was the more conventional right circular (RCP). By applying a methodology comparable to determination of the Brewster angle, at which VEX RCP and LCP (left-circularly polarized) echoes should have equal powers, we proposed to carry out additional studies of the high-jej areas found by Magellan. The initial, highly elliptical, 24 h VEX orbit had its periapsis near 70°N latitude, which was favorable for observations of Maxwell. We also anticipated study of other high-elevation targets such as Ovda Regio and Thetis Regio with eight and three times, respectively, the anomalous high-altitude surface area of Maxwell [Pettengill et al., 1992, Table 2]. [6] Nine bistatic experiments were conducted after VEX orbit insertion (Table 2). Five were at incidence angles 8
70° suitable for studying surfaces with large jej. At 8
70° the 13 and 3.6 cm atmospheric absorptions along the propagation path are about 1 and 20 dB, respectively, rendering detection of surface echoes at 3.6 cm impractical with current equipment. A failure aboard the VEX spacecraft in late 2006 reduced the radiated power at 13 cm wavelength by
15 dB. Strong 3.6 cm LCP echoes were seen in the final experiments in August 2007; but there was

[7] The scattering geometry is shown in Figure 1. Transmissions from the VEX spacecraft were directed toward Venus’ surface by the high-gain antenna (HGA). HGA pointing was updated throughout the experiment so that the antenna beam tracked the specular point (8 = 8i = 8r) on the mean spherical surface as the spacecraft moved in its orbit. Because the direct and reflected path lengths changed at different rates, the differential Doppler shift between fd and fr usually allowed easy separation of the two signals. [8] The scattering process is characterized as quasispecular; that is, the total echo comprises many individual reflections from randomly located, specularly oriented, mirror-like facets. There is not a single fr but a spread of frequencies, where each frequency component represents the Doppler shift of one or more individual facets located near the nominal specular point on a smooth planet. The degree of frequency dispersion and the resulting total power received depend on the roughness of the surface weighted by the HGA illumination pattern [Tyler and Ingalls, 1971]. [9] On Earth the echo signals were captured by the 70 m antenna at the NASA Deep Space Network (DSN) site near Canberra, Australia. Signals were amplified, adjusted in frequency to remove the predicted sources of Doppler shifts, converted to baseband, and sampled at the Nyquist frequency (100 kHz) for later processing. Calibration signals using ambient loads and nominal 12.5 K value noise diodes were recorded as part of the data stream to provide references for later scaling to absolute echo powers. [10] During data processing, power spectra hViVi*i were computed for each RCP and LCP sample stream, where h i denotes time averaging, Vi is the baseband complex voltage spectrum from the ith Fourier component, and Vi* is its complex conjugate. Typically, 1024-point spectra were averaged over 10 s (Figure 2). Power in frequencies adjacent to the surface echo was used to establish a noise baseline, which, when compared with the calibration signal output, yielded the absolute received echo signal power.

Table 2. Venus Express Bistatic Radar: Overviewa Date

VEX Orbit Number

Earth Receive Time (UTC)

Earth-Venus Distance (A.U.)

Target

Elevation (km)

Latitude (°N)

Longitude (°E)

Incidence Angle 8 (°)

23 May 2006 15 Jun 2006 17 Jun 2006 19 Jun 2006 18 Mar 2007 20 Mar 2007 12 Jun 2007 2 Aug 2007 3 Aug 2007

32 55 57 59 331 333 417 468 469

0251 0159 0200 0201 0633 0633 0713 0609 0604

1.126 1.281 1.294 1.307 1.305 1.293 0.684 0.325 0.321

‘‘Commissioning’’ Maxwell Montes Maxwell Montes Maxwell Montes Rhea Mons Rhea Mons Ozza Mons Theia Mons Rhea Mons

N/A 6058.7 6058.7 6058.7 6055.5 6055.5 6056.5 6055.8 6055.5

31.3 66.69 67.09 67.44 22.54 24.52 0.15 22.46 21.56

298.3 0.15 3.54 6.97 281.36 285.75 197.99 279.27 280.78

50.83 72.44 72.22 72.00 69.02 70.11 12.95 35.87 35.19

a

Elevations (column 6) are from Table 2 of Pettengill et al. [1992]; targets were chosen for study of anomalous emissivity/reflectivity. The first experiment was conducted to test spacecraft and ground operations; target location was incidental. Earth receive time, latitude, longitude, and incidence angle are nominal values neglecting atmospheric refractivity. Observations typically spanned 1 h, providing profiles several thousands of kilometers in length, including the identified target. Experiments beginning 2007-03-18 were degraded at 13 cm wavelength as the result of a 15 dB loss in spacecraft radiated power (see text).

2 of 13

SIMPSON ET AL.: VENUS EXPRESS BISTATIC RADAR

E00B41

E00B41

Figure 1. Bistatic radar geometry. Transmitter, receiver, and the center of Venus define a plane within which the incidence angle is 8i and reflection angle is 8r. From orbit, Venus Express pointed its high-gain antenna (HGA) toward the instantaneous spot on the mean spherical surface where specular reflection occurred (8 = 8i = 8r). The reflection process converts incident RCP waves to echoes with right circular polarization (RCP) and left circular polarization (LCP) components. A weak signal often reached Earth directly, radiated via an antenna sidelobe, with Doppler shift fd. Doppler shift along the specular point reflection path was fr. On Earth, the RCP and LCP signals were captured by separate receivers and digitally recorded for later processing. For Magellan bistatic radar, the significant difference was that the spacecraft radiated a linearly polarized wave. The baseline noise was subtracted from the total power to obtain the echo power. [11] Additionally, complex cross spectra between the RCP and LCP channels hVRVL*i were computed at both wavelengths (Figure 3). The magnitude of the cross spectrum is a measure of the fractional surface which behaves as specular reflectors, and its phase is sensitive to the argument of e. The cross spectra of the noise in the RCP and LCP channels showed about 1 percent correlation, presumably from imperfect polarization isolation between the ground antenna feeds. We used these measured correlations, calculated over intervals of 7 –15 min, to correct for the coupling between the feeds, thereby improving the effective feed isolation, our ability to measure polarization of the surface echoes, and our ability to infer the argument of e.

3. Analysis [12] Specular scattering from the surface converts an incoming monochromatic VEX RCP wave into an outgoing wave with both RCP and LCP components, the magnitudes and phases of which depend on the Fresnel reflection coefficients [Stratton, 1941]: 
 
 RH ¼ cos 8  sqrt e  sin2 8 = cos 8 þ sqrt e  sin2 8 ð1aÞ 
 
 RV ¼ e cos 8  sqrt e  sin2 8 = e cos 8 þ sqrt e  sin2 8 ð1bÞ

RH and RV are the reflection coefficients for the horizontal and vertical components of the incident electromagnetic wave and 8 is the angle of incidence. The RCP and LCP

echo components, when RCP is incident, are given by [Simpson, 1993] RR ¼ ðRV þ RH Þ=2 and RL ¼ ðRV  RH Þ=2

ð2Þ

For partially conducting interfaces, equations (1) and (2) may be used by replacing the ‘‘real’’ dielectric constant with its complex equivalent, i.e., e — > e0  is/2pf = jej exp[i( f )], where e0 is the real part of the dielectric constant, s is the surface material conductivity, f is the wave frequency, and q( f ) is the argument of e. [13] The first step in finding the surface dielectric constant e from measurements is to solve (1) and (2) using the powers in the received RCP and LCP echo signals [Simpson et al., 2006]. The ratio of RCP to LCP power depends strongly on the magnitude of the dielectric constant but not its argument (Figure 4a). The angle at which RCP and LCP echo powers are equal, the Brewster angle for real e, increases from 60° to 75° to 84.3° for jej = 3, 10, and 100. [14] The phase of the received cross spectrum depends initially on the phase of the product of the circular reflection coefficients, RRRL*. With RCP incident on the surface, the difference in the two reflection coefficients imposes a phase shift between the RCP and LCP echo components. When the spacecraft rolls about the HGA axis during transmission, the phase difference between the RCP and LCP signals changes by 4p for each full rotation (VEX rolls for thermal protection). The two components may undergo further, small differential phase shifts as they pass through magnetized plasma between Venus and Earth. The separate paths through microwave circuits and electronics on the ground are certainly different as well. It is usually adequate to assume that spacecraft roll, plasma effects, and electronic drift add no more than a slowly time-varying bias to the

3 of 13

E00B41

SIMPSON ET AL.: VENUS EXPRESS BISTATIC RADAR

E00B41

Figure 2. Example VEX 13 cm RCP 1024-point 100 kHz power spectrum from 15 June. Spectrum is the average of 976 individual power spectra collected within ±5 s of 0156:16, when the specular point was near (78°N, 25°E) in a plains area northwest of Maxwell Montes. The surface echo is identified by the horizontal bar over 38– 47 kHz. Within the echo window a narrow carrier signal propagating directly from the spacecraft (Figure 1) is visible at
43 kHz. Baseline noise power density (No = 0.299 zeptowatts/Hz, where 1 zW is 1021 W) is calculated from the windows on either side of the echo. The carrier power, when present, is removed by linear interpolation across five frequency bins, after which the noise pedestal is removed and the echo power (Pe = 164.8 zW) is calculated by summing the remainders. The formal uncertainty in Pe (3.99 zW) is based on No and the amount of time and frequency averaging in the noise and echo windows. The uncertainty does not include the fluctuations in Pe, caused by the scattering process, which are typically 3 – 4 times larger (Table 3).

phase difference initiated by the RCP and LCP reflection coefficients. If the elevated regions on Venus are conducting while the surrounding plains are not, we would expect to see clear changes in the echo phase in hVRVL*i, mimicking the changes in the phase of hRRRL*i and occurring suddenly as the specular point moves into and out of the anomalously reflecting regions. [15] Figure 4b shows the predictions of cross-spectrum phase y from (1) and (2) for the nine choices of e (three each of magnitude and argument) used in Figure 4a. The cross-spectrum phase is ±180° for real dielectric constants. For imaginary e, y
135°, varying slightly with incidence angle 8 for the smallest e. The cross-spectrum phase is approximately the average of the real and imaginary values when e itself is midway between real and imaginary. Thus, to first order, y
180  q=2 þ bias terms

ð3Þ

The corresponding cases for Magellan’s 13 cm bistatic radar experiment are shown in Figure 5. Figure 5a shows the RCP/LCP power ratio for the three example choices of jej and three choices for q when the incident wave is linearly polarized at 45° with respect to the incidence plane. For the Magellan configuration the power ratio would be unity when the dielectric constant is real. If the dielectric constant were purely imaginary, RCP echo power can become
10 times larger than LCP near the Brewster angle for the corresponding real e, with the difference being most pronounced for the smallest jej. Figure 5b shows the phase of the echo cross spectrum, which is most sensitive to 8 and jej and least sensitive to q.

4. Data [16] Figure 6 shows example HGA ‘‘footprints’’ across Maxwell Montes for the bistatic radar experiments. The MGN footprints moved from southwest to northeast; since the experiments were conducted on alternating 93 min

4 of 13

E00B41

SIMPSON ET AL.: VENUS EXPRESS BISTATIC RADAR

E00B41

Figure 3. Example VEX 13 cm cross spectrum on 19 June at 0158:55 ± 05, when the specular point was near (76°N, 9°E). (a) The directly propagating carrier, visible at 43 kHz in Figure 2, can be seen in the magnitude plot at
73 kHz. (b) Phase values are clustered at y
100°, where the surface echo is well defined, but are distributed randomly elsewhere.

orbits, the tracks are nearly identical. The Venus Express footprints moved from northwest to southeast on alternating 24 h orbits. Of particular interest is the VEX track on 19 June, which crossed the MGN tracks near Cleopatra, where the specular echoes implying conducting surface material were obtained. Here we focus on 19 June. The results from 15 and 17 June and are similar but without the Cleopatra connection. [17] Figure 7 provides an overview of the 13 cm VEX results on 19 June. Details for Cleopatra Patera, Maxwell Montes, and the immediate surroundings are expanded in Figure 8 and listed in Table 3. The solid lines in Figure 7 show amplitudes and phases averaged over 25 frequency bins spanning 2441 Hz centered on the echo peak, consistent with the echo widths in Figures 2 and 3. Results for the peak bin alone (Df
98 Hz) are shown by the open circles. Averages of noise in 25 bins approximately 10 kHz above and below the surface echo are indicated by ‘‘plus’’ and ‘‘cross’’ symbols, respectively. [18] A predominantly specular echo is expected to have a cross-spectrum magnitude that is close to the geometric mean of the RCP and LCP echo powers and a crossspectrum phase that has a small standard deviation. As the scattering process becomes less specular, we expect the

correlation between RCP and LCP echo components to decrease, the cross-spectrum magnitude to drop below the RCP-LCP geometric mean, and the cross-spectrum phase to approach a uniform distribution. The extreme case of no echo is illustrated in the noise portions of spectra. For example, over 55 – 65 kHz in Figure 3, the RCP and LCP noise power densities are 0.299 and 0.416 zW/Hz, respectively, the cross-spectrum magnitude is small, and the phases are distributed randomly. In Figure 8a, the noise power has been subtracted to leave the RCP and LCP echo powers. The majority of measurements for t < 2.011 and t > 2.039 in Figure 8 satisfy the criteria for specular echoes in that the cross-spectrum magnitude is close to the RCP-LCP geometric mean and the crossspectrum phase is narrowly distributed. Also, measurements in this range have larger RCP than LCP echo power density, as is expected for reflection at incidence angles greater than the Brewster angle. [19] For 2.011 < t < 2.039, at least seven of the ten measurements in Figure 8a show LCP echo power density larger than RCP, implying a much larger e than in the surrounding terrain. Only the last measurement (t = 2.0375) meets the criteria for specular reflection, however. Half of the measurements have cross-spectrum magnitudes smaller

5 of 13

E00B41

SIMPSON ET AL.: VENUS EXPRESS BISTATIC RADAR

E00B41

Figure 4. (a) Ratio of hVRVR*i to hVLVL*i calculated from equations (1) and (2) for RCP incident (the VEX case), three choices of dielectric constant magnitude (jej = 3, 10, 100), and three choices of dielectric constant argument (q = 0°, 45°, and 90°). (b) Phase y of the cross spectrum hVRVL*i is given for the same conditions.

than both RCP and LCP echo powers, and only the last measurement has a cross-spectrum phase standard deviation less than 40° (Table 3, rows 5 – 14). Of particular concern is the cross spectrum that we associate most closely with Cleopatra (t = 2.0208; Table 3, row 8). There, a broad maximum in cross-spectrum magnitude (0.008 zW/Hz) is centered on Cleopatra; but the RCP and LCP echo power densities are also relatively large with values 0.018 and 0.025 zW/Hz, respectively (Figure 8a). Taken together, these measurements imply that less than 40% of the echo power is from specular scattering. The standard deviation of the cross-spectrum phase at t = 2.0208 is 40°, the second smallest within the Maxwell group, and not unreasonable if there is a mix of specular and ‘randomizing’ elements within the illuminated footprint. The cross-spectrum phase itself is 92°, midway between the four-row averages at the beginning (106°) and end (82°) of Table 3. Although 92° could be taken as evidence against a sudden shift in electrical properties, we note that the measurements immediately before and after yielded cross-spectrum phases of 154 ± 51° (Table 3, row 7) and 127 ± 44° (row 9), which diverge from the linear trend apparent in Figure 7d. However, all three values have overlapping one-sigma uncertainties. [20] To calculate the total echo power Pe and its uncertainty in each polarization we used the method illustrated in

Figure 2. The dielectric constant can then be obtained by solving (1) and (2). In practice, we used the numerical RCP/ LCP ratio and the incidence angle 8 to obtain e by interpolation from a precomputed lookup table. A total of nine solutions for e was sought for each (RCP, LCP) pair of Pe values (Table 3, Notes). When any of the nine (RCP, LCP) pairs failed to yield a solution, ‘N/A’ was entered in the table. [21] Before solving for e, we adjusted the incidence angle for the effects of atmospheric refraction, including differences in refraction resulting from surface elevation changes (specular point latitude and longitude are changed very little by refraction). The incidence angle correction is as much as 4° ten minutes before the specular point entered Maxwell Montes. The maximum topography of Maxwell reduces a 2.3° correction to 1.3°. The observational geometry is almost identical among the three VEX orbits except for a 30 s/day delay in time and a 1.8°/day eastward shift in longitude of the specular point, resulting primarily from the rotation of Venus. [22] The 10 s spacing of data points in Table 3 represents a compromise between long integration times needed to improve signal-to-noise ratio and short integration times needed to preserve temporal and spatial resolution. Averages over 30 s consistently resulted in solutions for e, but the enhanced echoes from Cleopatra were not well resolved.

6 of 13

SIMPSON ET AL.: VENUS EXPRESS BISTATIC RADAR

E00B41

E00B41

Figure 5. (a) Ratio of hVRVR*i to hVLVL*i when a linearly polarized wave is incident for three choices of dielectric constant magnitude (jej = 3, 10, 100) and three choices of the dielectric constant argument (q = 0°, 45°, and 90°). The incident angle of polarization is halfway between horizontal and vertical: the Magellan case. Note all ratios approach unity for small angles of incidence. In order to facilitate display, we have displaced the curves for jej = 3 upward and the curves for jej = 100 downward. (b) Phase y of the cross spectrum hVRVL*i under the same conditions, but without displacements.

The 10 s points are separated by about 85 km along the specular track (Figure 6). The circularly polarized antenna beam from VEX was about 84 km across, half power to half power, where it intercepted the surface at periapsis (slant range
820 km on 19 June at t = 2.033). The HGA pattern, projected onto the surface at 8 = 71°, results in an elliptical footprint with dimensions approximately 84 260 km; five minutes earlier and seven minutes later the ellipse dimensions were doubled, chiefly because of increased slant range. The long axis of the ellipse was approximately perpendicular to the specular point track during the experiment, so the measurements listed in Table 3 can be considered independent.

5. Discussion 5.1. Magellan [23] We reprocessed the Magellan bistatic radar data [Simpson, 1997] using the same software that was applied to the VEX BSR data. One source of concern has been a possible ground receiver problem, which might have resulted in an apparent excess in Magellan RCP echo power and led to the conclusion that the surface near Cleopatra is

electrically conducting. We also know that the ground antenna system is imperfect. However, our experience since 2004 with a number of bistatic radar experiments and tests at DSN 70 m antennas is that cross polarization between RCP and LCP is approximately 25 dB below the direct polarization level. [24] Our reprocessing of the Magellan data included calibration of the absolute power levels. Compared with an original ad hoc assumption that the two receiver channels had equal gains and system noise (neither of which was known), the reprocessed results are less stable than the originals because of variations in system gain (especially LCP) of up to 1 dB on time scales of seconds to minutes. Also, because RCP and LCP are not balanced, the differences between channels appear as a weakly polarized (and variable) component of the noise where none should be expected. These instabilities dominate the physical cross polarization in the Magellan data in both an average sense and in terms of minute-to-minute fluctuations. [25] When the specular point is outside Maxwell, typical Magellan echoes have power densities larger than 0.1 zW/Hz (Figure 9a). Because the transmitted polarization was linear, both RCP and LCP have similar echo powers at

7 of 13

E00B41

SIMPSON ET AL.: VENUS EXPRESS BISTATIC RADAR

E00B41

Figure 6. Magellan and Venus Express ground tracks, shown as a sequence of antenna beam footprints superimposed on a topographic map of the Maxwell area [U.S. Geological Survey, 1984]. Footprints are separated by 30 s. Magellan footprints moved from southwest to northeast along three almost identical tracks spaced by 186 min in time (only one orbit is shown). Venus Express footprints moved from northwest to southeast on 15 –19 June (only the first and last orbits are shown).

the receiver; the power ratio is unity for latitudes less than 60° and larger than 70° (Figure 9b). The echo powers from within Maxwell drop to about a tenth those from areas immediately adjacent to Maxwell. Nonetheless, it is possible to identify the excess RCP power during the latter half of the Maxwell passage (latitudes 65– 67° in Figure 9), which Pettengill et al. [1996] associated with Cleopatra. Although this excess may sometimes be associated with the short-term gain fluctuations, the amount of excess is clearly different in spectral regions occupied by echo signals as compared with nearby noise. In addition, the RCP excess recurs on each of the three Magellan orbits over Cleopatra Patera, and no similar feature appears in echoes at other surface locations. On this basis, we support the original conclusion by Pettengill et al. [1996] regarding the excess RCP power in the Maxwell echoes. [26] From Figure 5 we can confirm that the 100° crossspectrum phase change over Maxwell (Figure 9c) is consistent with jej
100 (or larger); a power ratio of 1.4 is consistent with equal real and imaginary parts, while a power ratio of 1.8 favors an entirely imaginary dielectric constant. Neither of these results is strongly dependent on incidence angle, which is approximately 2° larger than the latitude during this time.

5.2. Applicability of Fresnel Reflection [27] The methods presented here are based on an assumption that the surface echo from Venus is dominated by Fresnel reflection, equations (1) and (2), reflection by many surface ‘facets’ having radii of curvature that are large compared to the wavelength [Tyler, 1976]. Over most of Venus, the magnitude of the cross spectrum is near the geometric mean of the RCP and LCP powers and the crossspectrum phase is well defined, validating the assumption. The data from Cleopatra Patera, however, imply that less than 40% of the echo originates from such surface elements. Around Cleopatra the model may be inadequate, and over other parts of Maxwell the breakdown may be even more severe. [28] However, the VEX data do show that there is a reversal of the sense of circular polarization over Maxwell generally and near Cleopatra specifically. We do measure a nonzero cross spectrum with constrained (albeit not small) phase distributions: at least in the vicinity of Cleopatra. These are results we would expect from Fresnel reflection even though we admit that (1) and (2) cannot be applied rigorously. A literature of Fresnel alternatives exists, scattering by discrete objects (e.g., wires and rocks) and

8 of 13

E00B41

SIMPSON ET AL.: VENUS EXPRESS BISTATIC RADAR

E00B41

Figure 7. Summary results from Venus Express 13 cm bistatic radar experiment on 19 June. (a) RCP power density, (b) LCP power density, (c) cross-spectrum magnitude, and (d) and cross-spectrum phase. Horizontal axis is Earth receive time in hours past midnight. Solid lines are average values over 25 frequency bins (2441 Hz) centered on the surface echo peak. Open circles are values in the central echo bin (Df = 98 Hz). Plus and cross symbols are 25 bin averages 10 kHz above and below the echo, respectively, where only noise is present. Note in Figure 7d that surface echo phase drifts toward smaller values during the experiment because of spacecraft attitude and other slow changes. For more detail on Maxwell Montes, including Cleopatra Patera, see Figure 8 and Table 3.

irregular structure (edges and cracks), but the results do not explain the Venus observations. Dedrick et al. [1978] have shown that the cross-polarized radar cross section of an ‘‘atmosphere’’ of randomly oriented wires is one-third that of the direct cross section in both the forward and backward directions. This sets a limit on the amount of polarization change that can be attributed to discrete objects and fails to explain the polarization reversal we see at Maxwell Montes. Refraction has been proposed by Tryka and Muhleman [1992] and Peters [1992] to explain unexpected phenomena on Venus and icy satellites, respectively, but the models require a regolith which is nearly lossless and are more attractive for backscatter. [29] Cost [1965] and Ulaby et al. [1988] have studied the microwave bistatic response of terrestrial surfaces both with

and without vegetation. Their experiments were typically conducted in a controlled (outdoor laboratory) environment with orthogonal linear polarizations. This and other work has been summarized by Weiner [2007]. Almost without exception the maximum bistatic radar cross section of these surfaces was observed to be consistent with specular theory: even for ‘rough’ surfaces. For incidence angles 8 > 40° the response in the specular direction was typically 10 dB or more than in other directions. [30] Lacking a detailed, viable alternative model we propose that many of the key features of Fresnel reflection are being reproduced in Maxwell Montes by its very rough surface. Although (1) and (2) cannot be applied because none of the ‘facets’ is large enough to qualify as a quasi-

9 of 13

SIMPSON ET AL.: VENUS EXPRESS BISTATIC RADAR

E00B41

E00B41

Figure 8. The 13 cm VEX bistatic radar echo measurements collected near and within Maxwell Montes. Maxwell is assumed to extend over 2.011 < t < 2.039, with Cleopatra Patera at t
2.02. (a) RCP (plusses) and LCP (crosses) echo power density averaged over 25 frequency bins centered on the echo peak; noise power has been subtracted from total power listed in Table 3, columns 6– 7. Cross-spectrum magnitude density (circles) is taken from Table 3, column 8. (b) Cross-spectrum average phase (circles) and its standard deviation from calculations over the same 25 frequency bins.

specular reflector, the bulk properties of the surface material may still be sensed. A surface saturated with relatively flat rocks would have a hybrid scattering response, partly from the dipole nature of the rocks and partly from the exposed flat side, which can be responsible for polarization reversal in the bistatic geometry. Working the problem backward, we can interpret changes in the RCP/LCP ratio in terms of the material dielectric constant without being able to say exactly what the values are. We have previously investigated the scattering by populations of discrete rocks on and within a regolith [Baron et al., 1998]. Similar methods may be applied here; but the model must incorporate rock-rock interactions, which makes an already computationally intensive calculation an order of magnitude more difficult.

6. Conclusions [31] The three Venus Express bistatic radar experiments conducted over Maxwell Montes confirm the presence of high dielectric material in elevated regions through measurements of opposite sense, circularly polarized echo powers when a circularly polarized wave was incident.

The echo power ratio yielded values in the range 4 < jej < 25 for Maxwell, but the weak echoes and relatively low cross-spectrum magnitude caution against a strict Fresnel-based interpretation. Away from Maxwell Montes, where criteria for specular scattering are more easily met, values jej
4 are common. [32] Attempts to detect electrical phase changes in the VEX cross spectrum hVRVL*i, expected if the surface material were conducting, have not been successful. Magellan observations of excess RCP power, which require that e have a large imaginary component, have been confirmed near Cleopatra crater. VEX results from the same area are inconclusive because of inadequate signal-to-noise ratio. [33] Continuing the VEX BSR explorations, particularly over areas other than Maxwell and at smaller Earth-Venus distances remains a desirable goal. The VEX opportunity has been lost because of failure in the spacecraft radio system, which after November 2006 radiated only 200 mW at 13 cm wavelength. Future electromagnetic probing of the unusual high-elevation terrains on Venus must await the arrival of a new spacecraft.

10 of 13

11 of 13

2.0014 2.0042 2.0069 2.0097 2.0125 2.0153 2.0181 2.0208 2.0236 2.0264 2.0292 2.0319 2.0347 2.0375 2.0403 2.0431 2.0458 2.0486

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

71.52 71.10 70.95 70.73 70.46 70.17 69.87 69.55 69.15 68.77 68.42 67.93 67.40 66.83 66.32 65.88 65.51 65.07

8 (°) 71.73 71.00 70.27 69.52 68.76 67.98 67.20 66.41 65.60 64.79 63.98 63.15 62.31 61.47 60.63 59.79 58.96 58.13

Lat (°N) 2.29 3.38 4.37 5.28 6.12 6.83 7.50 8.11 8.63 9.11 9.56 9.92 10.25 10.54 10.80 11.06 11.30 11.50

Lon (°E) 0.334 0.334 0.313 0.306 0.305 0.306 0.307 0.319 0.312 0.304 0.303 0.303 0.306 0.312 0.356 0.325 0.308 0.313

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

0.016 0.018 0.009 0.010 0.011 0.010 0.009 0.010 0.007 0.012 0.009 0.013 0.010 0.011 0.030 0.012 0.010 0.014

RCPd 0.441 0.434 0.417 0.418 0.429 0.424 0.431 0.443 0.434 0.425 0.419 0.421 0.430 0.434 0.464 0.438 0.429 0.426

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

0.021 0.017 0.011 0.011 0.010 0.010 0.013 0.012 0.015 0.010 0.016 0.014 0.013 0.015 0.025 0.013 0.014 0.015

LCPd 0.029 0.027 0.009 0.007 0.003 0.004 0.007 0.008 0.008 0.006 0.002 0.001 0.001 0.015 0.052 0.022 0.015 0.015

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

0.012 0.012 0.007 0.006 0.009 0.009 0.009 0.012 0.008 0.007 0.008 0.008 0.006 0.011 0.026 0.011 0.006 0.011

Peak Magnitude (zW/Hz) 110.2 ± 17.0 110.0 ± 18.3 93.4 ± 46.6 111.7 ± 50.2 74.6 ± 69.8 114.9 ± 60.6 154.5 ± 51.3 92.2 ± 40.0 126.5 ± 43.6 101.7 ± 49.2 137.7 ± 79.7 132.9 ± 85.7 25.9 ± 79.1 78.2 ± 21.9 83.3 ± 7.4 79.4 ± 15.0 86.1 ± 29.0 79.8 ± 26.2

Phase (°)

Cross-Spectrum Average

131.23 ± 10.87 96.69 ± 10.95 24.64 ± 10.89 62.69 ± 10.83 4.23 ± 10.86 43.22 ± 10.77 23.07 ± 10.83 60.97 ± 10.79 52.10 ± 10.73 36.38 ± 10.75 6.48 ± 10.75 12.53 ± 10.72 36.31 ± 10.68 54.49 ± 10.66 162.26 ± 10.64 100.07 ± 10.56 54.23 ± 10.54 72.33 ± 10.54

RCPt

85.28 ± 15.28 70.63 ± 15.16 36.92 ± 15.18 15.46 ± 15.11 75.36 ± 15.10 54.95 ± 15.03 37.68 ± 15.12 130.38 ± 14.98 70.14 ± 15.06 69.00 ± 14.97 41.05 ± 14.94 45.06 ± 14.83 40.92 ± 14.86 103.63 ± 14.77 187.01 ± 14.73 94.71 ± 14.74 39.56 ± 14.73 47.43 ± 14.73

LCPt

Total Echo Power per Figure 2 (zW)

e 6.150 ± 0.769 6.511 ± 1.014 N/A N/A N/A 9.849 ± 2.529 13.252 ± 6.386 14.619 ± 2.185 9.172 ± 1.865 12.363 ± 3.274 N/A 47.770 ± 51.2 6.682 ± 2.098 9.809 ± 1.640 5.877 ± 0.383 4.791 ± 0.554 3.803 ± 0.938 3.347 ± 0.646

From RCPt/LCPt

Columns 1 – 2: Values in each row are 10 s averages centered on the time in column 2. Columns 3 – 5: Geometry was corrected for atmospheric refraction (including differential refraction effects of topography) using a ray tracing model, the Magellan model atmosphere [Lyons, 1991], and topography derived from Magellan radar altimetry [Ford, 1992]. Columns 6 – 7: Power density and its standard deviation were computed using 25 frequency bins (2441 Hz) centered on the fitted position of the echo peak; of these totals, 0.299 – 0.301 and 0.416 – 0.420 zW/Hz were RCP and LCP noise power, respectively (see Figure 2). Columns 8 – 9: Cross-spectrum magnitude and phase averaged over 25 bins. Statistics were computed using real and imaginary parts; surface echoes are associated with small phase standard deviations, while noise has phase standard deviations
90°. Columns 10 – 11: Echo powers as calculated from components illustrated in Figure 2. Column 12: Dielectric constant was derived by solving equations (1) and (2) (see text) for each of nine RCP/LCP ratios and the corrected incidence angle (column 3). The nine ratios were constructed by pairing RCP power, RCP power less one standard deviation, and RCP power plus one standard deviation with the corresponding LCP values. Solutions were given a weighting of e0.5 for each occurrence of a standard deviation in the ratio. Values in column 12 are the weighted mean and standard deviation from the nine solutions. ‘N/A’ appears if any of the nine solution attempts failed. Rows 5 – 14 are associated with Maxwell Montes, row 8 with Cleopatra Patera.

a

Time (Hours Past 0 h)

Row

Average Power Density Noise + Echo Peak (zW/Hz)

Table 3. Venus Express Bistatic Radar Power and Cross-Spectrum Results 13 cm Wavelength 19 June 2006a

E00B41 SIMPSON ET AL.: VENUS EXPRESS BISTATIC RADAR E00B41

SIMPSON ET AL.: VENUS EXPRESS BISTATIC RADAR

E00B41

E00B41

Figure 9. Magellan bistatic radar results after reprocessing in the Maxwell Montes region, plotted versus specular point latitude, which is about 2° less than angle of incidence. Echo properties were averaged over 25 frequency bins (
610 Hz at sampling rate 25 kHz). (a) Echo power from the sum of the RCP and LCP power densities. (b) Ratio of RCP to LCP power densities, expected to be near 1.0 for surfaces with real dielectric constants. (c) Cross-spectrum phase. Cleopatra Patera (65 – 67° latitude) is unique in having moderately strong echoes, excess RCP power (ratio 1.5), and a large phase change.

[34] Acknowledgments. This work was supported at Stanford University by the NASA Venus Express Participating Scientist Program (grant NNG06GG04G-000001). Estimates of absorption by the Venus atmosphere were based on a model kindly provided by J.M. Jenkins of the SETI Institute. Comments on an earlier version of the manuscript by Peter Ford, Gordon Pettengill, and an anonymous reviewer are gratefully acknowledged.

References Baron, J. E., R. A. Simpson, G. L. Tyler, H. J. Moore, and J. K. Harmon (1998), Estimation of Mars radar backscatter from measured surface rock populations, J. Geophys. Res., 103, 22,695 – 22,712, doi:10.1029/ 98JE02221. Cost, S. T. (1965), Measurements of the bistatic echo area of terrain at X-band, Rep. 1822 – 2, Ohio State Univ. Antenna Lab., Columbus. Dedrick, K. G., A. R. Hessing, and G. L. Johnson (1978), Bistatic radar scattering by randomly oriented wires, IEEE Trans Antennas Propag., 26, 420 – 426. Ford, P. G. (1992), MGN V RDRS 5 Global Data Record Topographic V1.0 (MGN-V-RDRS-5-GDR-TOPOGRAPHIC-V1.0), NASA Planet. Data Syst., Greenbelt, Md. Ford, P. G., and G. H. Pettengill (1983), Venus: Global surface radio emissivity, Science, 220, 1379 – 1381, doi:10.1126/science.220.4604.1379. Goldstein, R. M. (1965), Preliminary Venus radar results, J. Res. Natl. Bur. Stand., Sect. D, 69, 1623 – 1625. Ha¨usler, B., et al. (2006), Radio science investigations by VeRa onboard the Venus Express spacecraft, Planet. Space Sci., 54, 1315 – 1335, doi:10.1016/j.pss.2006.04.032. Jurgens, R. F. (1970), Some preliminary results of the 70-cm radar studies of Venus, Radio Sci., 5, 435 – 442, doi:10.1029/RS005i002p00435. Klose, K. B., J. A. Wood, and A. Hashimoto (1992), Mineral equilibria and the high radar reflectivity of Venus mountaintops, J. Geophys. Res., 97, 16,353 – 16,369, doi:10.1029/92JE01865.

Lyons, D. T. (1991), Magellan planetary constants and models, JPL D-2300 Rev. D, Jet Propul. Lab., Pasadena, Calif. Peters, K. J. (1992), The coherent backscatter effect: A vector formulation accounting for polarization and absorption effects and small or large scatterers, Phys. Rev. B, 46, 801 – 812, doi:10.1103/PhysRevB.46.801. Pettengill, G. H. (1962), Enhancement of radar reflectivity associated with the lunar crater Tycho, J. Geophys. Res., 67, 4881 – 4885, doi:10.1029/ JZ067i012p04881. Pettengill, G. H., E. Eliason, P. G. Ford, G. B. Loriot, H. Masursky, and G. E. McGill (1980), Pioneer Venus radar results: Altimetry and surface properties, J. Geophys. Res., 85, 8261 – 8270, doi:10.1029/ JA085iA13p08261. Pettengill, G. H., P. G. Ford, and R. J. Wilt (1992), Venus surface radiothermal emission observed by Magellan, J. Geophys. Res., 97, 13,091 – 13,102, doi:10.1029/92JE01356. Pettengill, G. H., P. G. Ford, and R. A. Simpson (1996), Electrical properties of the Venus surface from bistatic radar observations, Science, 272, 1628 – 1631, doi:10.1126/science.272.5268.1628. Rogers, A. E. E., and R. P. Ingalls (1970), Radar mapping of Venus with interferometric resolution of the range-Doppler ambiguity, Radio Sci., 5, 425 – 433, doi:10.1029/RS005i002p00425. Schaefer, L., and B. Fegley Jr. (2004), Heavy metal frost on Venus, Icarus, 168, 215 – 219, doi:10.1016/j.icarus.2003.11.023. Shepard, M. K., R. E. Arvidson, R. A. Brackett, and B. Fegley Jr. (1994), A ferroelectric model for the low emissivity highlands on Venus, Geophys. Res. Lett., 21, 469 – 472, doi:10.1029/94GL00392. Simpson, R. A. (1993), Spacecraft studies of planetary surfaces using bistatic radar, IEEE Trans. Geosci. Remote Sens., 31, 465 – 482. Simpson, R. A. (1997), Magellan Bistatic Radar Raw Data Records V1.0 (MGN-V-RSS-1-BSR-V1.0), NASA Planet. Data Syst., Greenbelt, Md. Simpson, R. A., G. L. Tyler, M. Pa¨tzold, and B. Ha¨usler (2006), Determination of local surface properties using Mars Express bistatic radar, J. Geophys. Res., 111, E06S05, doi:10.1029/2005JE002580. Stratton, J. A. (1941), Electromagnetic Theory, McGraw Hill, New York.

12 of 13

E00B41

SIMPSON ET AL.: VENUS EXPRESS BISTATIC RADAR

Tryka, K. A., and D. O. Muhleman (1992), Reflection and emission properties on Venus: Alpha Regio, J. Geophys. Res., 97, 13,379 – 13,394, doi:10.1029/92JE01163. Tyler, G. L. (1976), Wavelength dependence in radio-wave scattering and specular-point theory, Radio Sci., 11, 83 – 91, doi:10.1029/ RS011i002p00083. Tyler, G. L., and D. H. H. Ingalls (1971), Functional dependencies of bistatic-radar frequency spectra and cross sections on surface scattering laws, J. Geophys. Res., 76, 4775 – 4785, doi:10.1029/JB076i020p04775. Ulaby, F. T., T. E. Van Deventer, J. R. East, T. F. Haddock, and M. E. Coluzzi (1988), Millimeter-wave bistatic scattering from ground and vegetation targets, IEEE Trans. Geosci. Remote Sens., 26, 229 – 243. U.S. Geological Survey (1984), Topographic and shaded relief maps of Venus (map I-1562), Reston, Va.

E00B41

Weiner, M. W. (2007), Clutter, in Advances in Bistatic Radar, edited by N. J. Willis and H. D. Griffiths, chap. 9, pp. 230 – 319, SciTech, Raleigh. 

B. Ha¨usler and R. Mattei, Institut fu¨r Raumfahrttechnik, Universita¨t der Bundeswehr Mu¨nchen, D-85577 Neubiberg, Germany. (bernd.haeusler@ unibw-muenchen.de; [email protected]) G. L. Tyler and R. A. Simpson, Space, Telecommunications, and Radioscience Laboratory, Stanford University, Packard Building, MC 9515, Stanford, CA 94305-4020, USA. ([email protected]; rsimpson@ magellan.stanford.edu) M. Pa¨tzold, Rheinisches Institut fu¨r Umweltforschung, Abt. Planetenforschung, Aachener Strasse 209, D-50931 Ko¨ln, Germany. (mpaetzol@ uni-koeln.de)

13 of 13

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, E00B24, doi:10.1029/2008JE003134, 2008

for

Full Article

Felsic highland crust on Venus suggested by Galileo Near-Infrared Mapping Spectrometer data George L. Hashimoto,1 Maarten Roos-Serote,2 Seiji Sugita,3 Martha S. Gilmore,4 Lucas W. Kamp,5 Robert W. Carlson,5 and Kevin H. Baines5 Received 2 March 2008; revised 29 July 2008; accepted 18 September 2008; published 31 December 2008.

[1] We evaluated the spatial variation of Venusian surface emissivity at 1.18 mm

wavelength and that of near-surface atmospheric temperature using multispectral images obtained by the Near-Infrared Mapping Spectrometer (NIMS) on board the Galileo spacecraft. The Galileo NIMS observed the nightside thermal emission from the surface and the deep atmosphere of Venus, which is attenuated by scattering from the overlying clouds. To analyze the NIMS data, we used a radiative transfer model based on the adding method. Although there is still an uncertainty in the results owing to the not well known parameters of the atmosphere, our analysis revealed that the horizontal temperature variation in the near-surface atmosphere is no more than ±2 K on the Venusian nightside and also suggests that the majority of lowlands likely has higher emissivity compared to the majority of highlands. One interpretation for the latter result is that highland materials are generally composed of felsic rocks. Since formation of a large body of granitic magmas requires water, the presence of granitic terrains would imply that Venus may have had an ocean and a mechanism to recycle water into the mantle in the past. Citation: Hashimoto, G. L., M. Roos-Serote, S. Sugita, M. S. Gilmore, L. W. Kamp, R. W. Carlson, and K. H. Baines (2008), Felsic highland crust on Venus suggested by Galileo Near-Infrared Mapping Spectrometer data, J. Geophys. Res., 113, E00B24, doi:10.1029/2008JE003134.

1. Introduction [2] Venus is the nearest neighbor of Earth, and these twins are close in mass, size, and bulk composition. However, the surface environment of Venus is completely different from that of Earth, where the mean surface temperature is 735 K and water is almost absent. In contrast, the Earth surface is covered with ocean and is habitable. It is not clear whether the surface environments on these twin planets differed from beginning or they evolved from a similar environment into the current different states. Understanding the evolution of these twin planets is one of the most important and elusive problems in planetary climatology. [3] The chemical and mineralogical composition of the Venus surface can provide valuable information on the climatic evolution of Venus. If geologic signatures that are produced only under unique conditions are found in the chemical and mineralogical composition, they could be used 1 Laboratory for Earth and Planetary Atmospheric Science, Organization of Advanced Science and Technology, Kobe University, Kobe, Japan. 2 Lisbon Astronomical Observatory, Lisbon, Portugal. 3 Department of Complexity Science and Engineering, Graduate School of Frontier Science, University of Tokyo, Kashiwa, Japan. 4 Department of Earth and Environmental Sciences, Wesleyan University, Middletown, Connecticut, USA. 5 Jet Propulsion Laboratory, California Institute of Technology, Pasadena, California, USA.

Copyright 2008 by the American Geophysical Union. 0148-0227/08/2008JE003134$09.00

to reconstruct the Venus paleoenvironment. Additionally, a global map of surface composition will be a very important diagnostic of bulk planetary composition, chemical differentiation, and evolution of the planetary interior. [4] It has been also argued that the controlling mechanism of the present Venus surface environment is a connection between the atmosphere and the chemical characteristics of planetary surface [e.g., Hashimoto and Abe, 2005, and references therein]. The high surface temperature of Venus enhances the importance of the gas-solid reactions on the planetary surface in controlling the atmospheric composition, including infrared-active species which control the greenhouse effect. Venus climate models that combine chemical reaction and radiative transfer in the atmosphere have demonstrated that such a coupling between surface chemical reactions and greenhouse effect will play an important role in controlling the surface environment [e.g., Hashimoto and Abe, 2000; Bullock and Grinspoon, 2001]. [5] However, chemical characteristics of the Venus surface are not well constrained. In situ measurements of the surface using gamma-ray and X-ray fluorescence spectroscopy were performed by Venera and Vega landers [Surkov et al., 1984, 1986, 1987], but it is difficult to infer the global characteristics from lander data collected at 7 sites on the planet. Although the radar reflectivity and radiothermal emissivity were measured nearly globally [e.g., Pettengill et al., 1982, 1992], origins of the microwave emissivity differences are highly controversial and appear to be related to both grain size and dielectric constant of constituent

E00B24

1 of 10

E00B24

HASHIMOTO ET AL.: FELSIC HIGHLAND CRUST ON VENUS

minerals [e.g., Campbell et al., 1997; Pettengill et al., 1997]. We need a new observation capable of determining the chemical and mineralogical composition of the Venus surface. [6] It has been discovered that thermal emission from the surface of Venus is detectable at the CO2-free windows in the near-infrared wavelengths (0.85, 0.90, 1.01, 1.10, and 1.18 mm) [e.g., Carlson et al., 1993a, 1993b; Meadows and Crisp, 1996; Baines et al., 2000]. Thermal emissivity is a function of surface temperature, grain size, and mineral composition. Previous investigations of surface emissivity using Earth-based telescopic data have demonstrated variations that appear to be correlated with topography and therefore surface temperature [Lecacheux et al., 1993; Meadows and Crisp, 1996]. Although the variations in the surface emissivity have been investigated, no signature of surface emissivity differences was found at near-infrared wavelengths in contrast with the microwave wavelengths [Lecacheux et al., 1993; Meadows and Crisp, 1996]. However, these analyses neglect multiple reflection between the planetary surface and the clouds which has been demonstrated to significantly obscure variations in surface emissivity [Moroz, 2002; Hashimoto and Sugita, 2003]. Reexamination of previous analyses demonstrated that there may be a significant spatial variation in the surface emissivity as large as
20% [Hashimoto and Sugita, 2003]. [7] Here in this study, we evaluated the Venus surface emissivity at 1.18 mm wavelength using data obtained by the Near-Infrared Mapping Spectrometer (NIMS) aboard the Galileo spacecraft. Because emissivity depends on the mineralogy, a mapping of surface emissivity may constrain the chemical characteristics of the Venus surface. In the following, we first describe the procedure of our analysis which includes the reflection by the planetary surface. Then, we evaluate the surface emissivity, and discuss the errors and assumptions that may affect the estimation. Finally, the existence of paleo-ocean on early Venus is inferred from the spatial variation of surface emissivity.

2. Data and Analysis 2.1. Data [8] We analyzed the multispectral nightside image identified as VPDIN-1 (for Venus Partial Disk Imaging-NightSide) which covers the disk from approximately 20°W to 90°E longitude [Carlson and Taylor, 1993] and is distributed by the Planetary Data System (PDS). Those data were obtained by Galileo NIMS during its flyby of Venus on 10 February 1990. NIMS is an imaging spectrometer, operating in the spectral range from 0.7 to 5.2 mm [Carlson et al., 1992], and images of nightside are obtained with 17 spectral channels [Carlson et al., 1991]. Some channels measured the thermal radiation in the well-known spectral windows at 1.18, 1.74, and 2.3 mm. In these windows, thermal emission originating from both the surface and lower atmosphere are observed [e.g., Allen and Crawford, 1984; Crisp et al., 1989; Carlson et al., 1991; Lecacheux et al., 1993; Pollack et al., 1993; Baines et al., 2000]. [9] NIMS channel 3 observed the radiation of the 1.18 mm window, in which thermal emission from the surface leaks through the Venus clouds (Figure 1a). Initial results from the VPDIN observations are described by Carlson et al.

E00B24

[1991], which mention that the high-altitude (12 km above mean planetary radius) surface feature Maxwell Montes is detectable in the NIMS channel-3 image. The dark features near the top and the right of the image are likely due to the Ishtar and Aphrodite terra, respectively (Figures 1a and 1e). However, there are also some features irrelevant to the surface. They are probably produced by the effects of scattering by the Venus clouds, which is the dominant effect in channels 5 and 7 (Figures 1c and 1d). Other effects detectable in channel 3 are scattered sunlight and limb darkening. These effects need to be corrected separately. 2.2. Overview of Analysis [10] The analysis procedure for evaluating surface emissivity is the following. (1) We first remove the contribution of scattered sunlight, (2) we make a correction for emission angle, and (3) we evaluate the surface emissivity. Since the radiation at 1.18 mm window is affected by both clouds and surface topography, these contributions are evaluated to isolate the contribution of surface composition to thermal emissivity. [11] In this analysis procedure, we use the NIMS channels 4, 5, and 7 in addition to channel 3. Figure 2 provides a schematic representation of the characteristics of radiation observed by the four channels. NIMS channel 4 is used to remove the contribution of scattered sunlight, and the NIMS channels 5 and 7 are used to estimate the cloud properties. 2.3. Radiative Transfer Model [12] We calculated synthetic spectra by means of a lineby-line radiative transfer program including both scattering and absorption. The computational code was developed by G. L. Hashimoto and others at the University of Tokyo. The algorithm known as the adding method is used to calculate the radiation in a vertically inhomogeneous plane-parallel atmosphere [Goody and Yung, 1989]. The number of radiative streams is 12 in each hemisphere. Taking into account a triangular spectral bandpass with the width of 0.025 mm [Carlson et al., 1992], we calculate radiances for each channel from synthesized spectra in which molecular absorption features are resolved. [13] Following the method by Pollack et al. [1993], opacities of gases and particles were calculated. In all our simulations, we used the vertical profiles of gas mixing ratios developed by Pollack et al. [1993]. Line absorption data for CO2, H2O, CO, SO2, HF, and OCS were taken from the HITEMP and HITRAN2000 [Rothman et al., 2003; L. S. Rothman et al., HITEMP, the high-temperature molecular spectroscopic database, manuscript in preparation, 2008]. Total internal partition sums (TIPS) are calculated by the TIPS code developed by Fischer et al. [2003]. Rayleigh scattering cross sections for atmospheric molecules are calculated from the refractive indices [Keady and Kilcrease, 2000; Penndorf, 1957; Vardavas and Carver, 1984]. Optical properties of cloud particles are calculated by the Mie scattering code [Bohren and Huffman, 1983] based on the baseline cloud model [Pollack et al., 1993] and the optical constants of 75% H2SO4 solution [Palmer and Williams, 1975]. [14] We also introduced a continuum absorption to reproduce an acceptable fit to the NIMS data, since an additional source of continuum opacity is present owing to the far

2 of 10

E00B24

HASHIMOTO ET AL.: FELSIC HIGHLAND CRUST ON VENUS

E00B24

Figure 1. Multispectral images of nightside of Venus. These data were obtained during the first NIMS imaging sequence (VPDIN-1) at a resolution of
50 km/pixel. (a) An image at 1.18 mm (channel 3). (b) An image at 1.47 mm (channel 4). (c) An image at 1.74 mm (channel 5). (d) An image at 2.31 mm (channel 7). (e) A Magellan topographic map. This image covers from 20°W to 90°E of longitude. The topographic high near the north pole is the Ishtar Terra. The western edge of Ovda Regio can be seen near the rightmost part of the image. wings of strong CO2 bands and collision-induced CO2 opacity [Pollack et al., 1993]. Binary absorption coefficients for continuum opacity are determined to match the measured flux with the cloud model of Pollack et al. [1993]. The values

used in our simulations are 1.0  109 cm1 amagat2 for 1.18 mm window, 1.0  108 cm1 amagat2 for 1.74 mm window, and 7.0  108 cm1 amagat2 for 2.3 mm window, respectively, though these values are somewhat different

Figure 2. A schematic picture of radiation in Venus’s atmosphere and clouds. 3 of 10

HASHIMOTO ET AL.: FELSIC HIGHLAND CRUST ON VENUS

E00B24

E00B24

from those used in other analyses [e.g., Be´zard et al., 1990; Pollack et al., 1993]. 2.4. Scattered Sunlight [15] Nightside emissions at the near-infrared wavelengths contain scattered sunlight reflected off the dayside of the planetary disk. In particular, a strong component that brightens gradually toward the terminator is found in channel 4 (1.47 mm) (Figure 1b). Atmospheric radiative transfer models demonstrate that Venus does not emit thermal radiation at this wavelength. The Venusian CO2 atmosphere is too opaque to allow radiation to penetrate the atmosphere, though the surface and lower atmosphere of Venus are hot enough to emit thermal radiation at this wavelength. The radiance observed in channel 4 can serve as a good approximation for the scattered sunlight. [16] We corrected the effect of scattered sunlight, by assuming that the contributions of this effect to channel 3, 5, and 7 are proportional to that of channel 4. Taking into account the solar spectrum, we use the following correction: IxS ¼ Ix 

Sx  I4 S4

ð1Þ

where suffix denotes the NIMS channel, ISx is the radiance after the correction of scattered sunlight, Ix is the radiance observed by the NIMS, and Sx is the solar spectrum. 2.5. Limb Darkening [17] We computed limb darkening by using our radiative transfer model. Figure 3 shows our calculated limb darkening curve and NIMS observation that is corrected for the scattered sunlight. The observed limb darkening is well reproduced by our simulation. Although the particle size and number density in the lower clouds would vary in time and space, they affect only scales of radiance and do not affect the shape of limb darkening. Since the upper cloud layer is opaque enough to determine the distribution of emission angle at the cloud top, the shape of limb darkening curve is independent of the property of the lower clouds. [18] The equation for limb darkening correction is given by IxSL ¼

Jx ð0Þ S  Ix Jx ðqÞ

ð2Þ

where ISL x is the radiance after the correction of scattered sunlight and limb darkening, Jx (q) is the computed radiance for emission angle of q, respectively. 2.6. Estimation of Surface Emissivity [19] To evaluate the surface emissivity at 1.18 mm window, we need to take into account the effects of clouds and surface topography. The overlying clouds modulate the radiation mostly by scattering. The lower atmosphere of Venus absorbs the radiation emitted by the surface and also emits thermal radiation, since the opacity of the thick Venus atmosphere is not negligible at 1.18 mm wavelength [e.g., Taylor et al., 1997]. The effects of atmospheric absorption and emission increase with the column density of the atmosphere, which is largely determined by the surface

Figure 3. Theoretical and observed limb darkening for the NIMS channel 3. The radiances are plotted as a function of the cosine of the emission angle. The NIMS observation is corrected for the scattered sunlight. topography owing to hydrostatic balance. For example, the atmospheric column density is smaller in a highland area, leading to smaller atmospheric absorption and emission. In addition the intensity of thermal radiation from the surface is a function of the temperature which in turn decreases with height, so that highlands are cooler and emitting less. [20] We cannot separate the clouds’ correction from the evaluation of surface emissivity, since multiple reflection between the surface and clouds affects the intensity of radiation observed above the clouds [Hashimoto and Sugita, 2003]. Here, multiple reflection refers to phenomenon where a fraction of the upward incident radiation at the bottom of cloud layer has experienced reflections between clouds and the planetary surface more than once. This effect has a significant influence on the observed radiance under the Venus conditions, where the reflectivity of overlying clouds is high. Although multiple reflection also occurs on Earth and Mars, the degree of its effect is very small. Since the atmospheric absorption affects the multiple reflection, the correction for modulation induced by clouds is also connected to the surface topography via atmospheric opacity. [21] To estimate the cloud properties we follow the method of Carlson et al. [1993b] which used NIMS channels 5 and 7 (1.74 and 2.31 mm, respectively). At these wavelengths, visible features are almost entirely due to clouds, which modulate the emission from the lower atmosphere, with no significant surface contribution [e.g., Taylor et al., 1997]. It is not necessary to take into account the effect of multiple reflection at these channels, since photons reflected by clouds are absorbed by lower atmosphere (Figure 2). Using the NIMS channel 5 and 7, we can estimate the cloud properties without the influence from the surface emissivity. [22] Then, we calculate the upward radiance at the top of the atmosphere in channel 3 as a function of surface emissivity. The cloud properties estimated from the NIMS channels 5 and 7 and the surface topography are input to the radiative transfer model. The surface emissivity is evaluated in accordance with the computed relation between

4 of 10

E00B24

HASHIMOTO ET AL.: FELSIC HIGHLAND CRUST ON VENUS

the surface emissivity and the radiance at the top of the atmosphere. 2.6.1. Evaluation of Cloud Properties [23] The clouds of Venus have been observed by remote sensing as well as in-situ measurement techniques [e.g., Esposito et al., 1983]. The observations indicate that the clouds of Venus are vertically stratified with three layers, which are called the upper, middle, and lower clouds, respectively [e.g., Knollenberg and Hunten, 1980]. These clouds are made up of a few different particle size components, which are often referred with mode 1, 2, 20, and 3 [e.g., Knollenberg and Hunten, 1980; Pollack et al., 1993]. The upper and middle clouds are generally uniform and featureless, while considerable spatial variations in cloud properties are observed in the lower cloud [Marov et al., 1980; Ragent and Blamont, 1980]. [24] We start with a cloud model developed by Pollack et al. [1993] and assume that the spatial variations in cloud properties are caused by variations in the number densities of mode 20 and 3 particles in the lower clouds. Carlson et al. [1993b] demonstrated that the NIMS observation is well reproduced by varying the number densities of mode 20 and 3 particles. Their results are verified by our radiative transfer model (Figure 4). Since the wavelength dependence of scattering varies with cloud particle size, channels 5 and 7 enable us to distinguish the variation in the abundance of mode 20 and 3 particles (Figure 5). 2.6.2. Surface Topography and Atmospheric Structure [25] The surface topography of Venus is measured by the Magellan spacecraft with vertical accuracy of about 50 m [Pettengill et al., 1991]. We assume that the Venusian lower atmosphere is horizontally uniform and the vertical temperature profile is given by the Venus International Reference Atmosphere (VIRA) [Seiff et al., 1985]. Below the zero altitude (6052.0 km from the center of Venus), the temperature lapse rate in the lowermost layer is extrapolated. For

E00B24

Figure 5. Venus cloud maps. (a) A map of the optical thickness of lower clouds. (b) A map of the modal radius of lower cloud particles. pressure extrapolation, the hydrostatic pressure relation is integrated using the temperature profile. We also assume that the temperature of the planetary surface equals to the temperature of atmosphere in contact with it, since the temporal variation of the temperature in the Venusian lower atmosphere is expected to be extremely small [e.g., Seiff, 1983]. 2.6.3. Emissivity Estimate and Spatial Smoothing [26] Using our radiative transfer model, we can estimate the surface emissivity based on the band-3 radiance which has been corrected for the effects of scattered day-side light and limb darkening. However, because the band-3 radiance depends only weakly on the surface emissivity due to multiple reflection between clouds and planetary surface [Hashimoto and Sugita, 2003], small errors in both observation and correction lead to a large error in surface emissivity estimate. In fact, not all the data points in the band-3 radiance have values that correspond to surface emissivity values between 0 and 1. Thus, we calculated the surface emissivity  beyond the physically significant range (i.e., 0 through 1) by extrapolating the results of radiative transfer model and discarding data points whose  is smaller than 1 or larger than 2. Such a criterion for data selection increases the available data and helps us to obtain a sufficient sample to infer the Venus surface composition. We reduced the random noise by averaging the data within a circle of radius 250 km. This smoothing procedure should not change general trends of the data since the spatial resolution of the Venus surface observation is limited to about 100 km owing to cloud blurring [Hashimoto and Imamura, 2001].

3. Results and Discussion Figure 4. Scatterplot of the NIMS channel 5 against the channel 7. The NIMS data are corrected for the effect of limb darkening. Six theoretical curves are also shown. For each curve, the ratio of mode 3 to mode 20 content is constant, and the total opacity of the lower cloud is varied. The lowermost curve is for a cloud that contains no mode 3 particles, while the uppermost curve is for a cloud in which the lower cloud is thoroughly composed of mode 3 particles.

3.1. Surface Emissivity [27] The estimated surface emissivity at 1.18 mm window wavelength is shown in Figure 6. Although the effect of scattered sunlight increases toward the terminator, our result does not show such a bias. Also there is no noticeable tendency varying with the emission angle, while the effect of limb darkening depends on the emission angle. These facts indicate that both the corrections for the scattered sunlight and limb darkening work fairly well.

5 of 10

E00B24

HASHIMOTO ET AL.: FELSIC HIGHLAND CRUST ON VENUS

E00B24

erties. However, the Northern Hemisphere does not exhibit such a behavior associated with the modal radius in the lower clouds. [33] The good fit of Carlson’s method (Figure 4) strongly suggests that it works well for evaluating the cloud properties. We did not identify any apparent error in the estimation of surface emissivity, though there may be a problem with the correction for the effect of clouds. It would be an important future work to develop a declouding method that uses the spectra of 1.73 and 2.3 mm window. Some instruments aboard the Venus Express spacecraft are observing the spectra of 1.73 and 2.3 mm window [e.g., Baines et al., 2006], though the Galileo NIMS did not observed them in a mapping mode [Carlson and Taylor, 1993]. Figure 6. Surface emissivity at 1.18 mm window wavelength. (a) A map of surface emissivity. (b) A map of surface emissivity that is averaged for a region of a circle of radius 250 km. [28] However, the mean surface emissivity in the Southern Hemisphere is greater than that in the Northern Hemisphere. This feature may be an artifact due to incorrect evaluation of cloud properties, since the modal radius of lower clouds also exhibit such behavior (Figure 5b). The result of declouding that is a correction for this effect is discussed further in section 3.2. [29] Although the surface emissivity map exhibits a considerable amount of noise, there is a substantial regional variation. For example, Ishtar Terra, Eistla Regio, and Alpha Regio have relatively low emissivity. In contrast, Bell Regio and a band of region from Tahmina Planitia to Fonueha Planitia have higher emissivity values. We will not further discuss a regional difference in emissivity, since surface emissivity estimation might be affected by overlying clouds. 3.2. Declouding [30] We created a declouded image corrected for cloudinduced contrast (Figure 7a). The cloud-induced contrast was corrected on the basis of the cloud properties evaluated by the method by Carlson et al. [1993b]. Although the cloud-induced contrast depends on the surface emissivity, we assumed that surface emissivity is uniform ( = 0.85) during this declouding procedure. [31] It is clearly shown that all the major features in the topography can be recognized in the declouded image, even though the declouding process is apparently a conspicuous noise source. For comparison, the thermal emission at the NIMS channel 3 wavelength was computed on the basis of the Magellan surface topography, assuming no clouds and uniform surface emissivity (Figure 7b). The correspondence between the declouded image (Figure 7a) and the synthesized image (Figure 7b) indicates that our declouding procedure effectively removes the influence of clouds. [32] The declouded image shows a north-south asymmetry that is not noticeable before the declouding. Since there is also a hemispheric asymmetry in the modal radius of lower clouds, their appearance of high surface emissivity in the southern hemisphere, especially the region along the limb, might be attributed to insufficient evaluation of cloud prop-

3.3. Temperature in the Lower Atmosphere [34] The intensity of leaking radiation at the 1.18 mm window depends not only on the surface emissivity but also on the temperature of lower atmosphere [e.g., Taylor et al., 1997]. A deviation from the assumed temperature profile will cause an error in the surface emissivity estimation. To obtain a rough estimate of the sensitivity to the temperature variation, we computed the intensities of the leaking radiation with varying the temperature of the lower atmosphere. The range of temperature deviation from the VIRA profile is roughly estimated to be about ±2 K, if the spatial variation in the declouded NIMS channel 3 image is entirely attributed to the deviation in the atmospheric temperature. [35] In other words, our analysis of the NIMS channel 3 indicates that horizontal temperature variation in the lower atmosphere is no more than ±2 K. This is consistent with results from observational and theoretical studies. While measurement accuracies are no less than ±4 K, temperature data from the Venera probes indicated that surface temperatures scatter over no more than a few K [Seiff et al., 1985]. Extrapolation of the temperature profile measured by Pioneer probes indicates surface temperatures from 731 to 735 K [Seiff et al., 1985]. Theoretical study also indicated that variations of temperature in the deep atmosphere is as

Figure 7. Thermal emission at 1.18 mm window wavelength from the surface and the lower atmosphere. (a) A declouded image that is corrected the cloud-induced contrast. (b) A synthesized image based on the Magellan topographic map.

6 of 10

E00B24

HASHIMOTO ET AL.: FELSIC HIGHLAND CRUST ON VENUS

E00B24

small as 0.1 K, because of the large heat capacity of the deep atmosphere [Stone, 1975]. [36] The NIMS channel 3 image also shows no clear evidence for a variation in temperature in the deep atmosphere with latitude or with local time; temperatures in the deep atmosphere are almost constant in spite of the variation in the insolation as a function of distance from the subsolar point. There might be a small hemispheric asymmetry in the temperature of deep atmosphere causing the north-south asymmetry in the declouded image, though that asymmetry would be attributed to the declouding process. A further study is needed to evaluate the actual horizontal variation in the temperature of deep atmosphere and its influence on the circulation of Venusian atmosphere. 3.4. H2O Abundance [37] In the estimation of surface emissivity, we assumed that H2O mixing ratio is 30 ppmv and is constant below the cloud deck. Since water vapor absorbs some radiation at 1.18 mm window, inhomogeneous distribution of water vapor causes an error in surface emissivity estimation. We computed the intensities of the leaking radiation with varying H2O mixing ratio, and found that variation of about ±10 ppm in H2O mixing ratio is required to reproduce the observed contrast in the declouded image. [38] However, it is unlikely that variation in the H2O mixing ratio is a major contributer to the spatial variation in the NIMS channel 3 image. Spatial variation of the H2O abundance in the lower atmosphere had been evaluated by Drossart et al. [1993]. They searched for its spatial variations from the analysis of the NIMS complete spectra at maximum spectral resolution using the 1.18 mm window. The result of their analysis is that the water vapor abundance shows no horizontal variation exceeding 20% over a wide latitude range on the nightside of Venus. Hence, the variation in the H2O mixing ratio is not enough to explain the spatial variation in the NIMS channel 3 image. 3.5. Emissivity and Altitude [39] Probability densities of surface emissivity (after smoothing) for three different altitude ranges are shown in Figure 8. We can see a general trend of negative correlation between emissivity and altitude; the emissivity of lowlands is generally higher than that of highlands. Although it is difficult to discuss the difference in emissivity in local scales (e.g., a few hundred km) due to noise emerging from the declouding procedure, the majority of lowlands have higher emissivity than the majority of highlands. Whereas there is an uncertainty in the results due to the not well known parameters of the atmosphere, the difference of emisivity between lowlands and highlands is more than 0.3. [40] The temperature lapse rate of the lowermost atmosphere can cause an altitude-dependent bias in the surface emissivity estimation. When a larger temperature lapse rate was used in the surface emissivity estimation, emissivities of highlands and lowlands are estimated to be higher and lower, respectively. If we use the temperature profile in which the lapse rate of the lowermost atmosphere is about 1 K/km larger than that of VIRA, the difference in the estimated surface emissivity between lowlands and highlands will disappear. However, such a temperature profile

Figure 8. Surface emissivity distribution for three different altitude ranges. Solid curve is for lowlands (z < 0 km); dashed curve is for intermediate altitudes (0 km < z < 2 km); and dotted curve is for highlands (z > 2 km). is statically unstable compared with the adiabatic lapse rate [Seiff et al., 1985]. It is unlikely that unstable stratification is observed all over the nightside of Venus, since there is no heat source which maintains the superadiabatic lapse rate. [41] It is also unlikely that horizontal variations in the H2O abundance and/or atmospheric temperature cause an altitude-dependent bias in the surface emissivity estimation. If H2O abundance and/or atmospheric temperature vary in association with surface topography, the estimated emissivity might have a bias related to the altitude. However, no observation indicates that the variation in the H2O abundance or atmospheric temperature correlates with the surface topography. Thus, it is likely that there is a difference in surface emissivity between lowlands and highlands. [42] Although we will not argue a regional difference in emissivity, a regional analysis will be useful to test the validity of emissivity estimate independent from any uncertainties in the retrieval. For example, there is no reason to expect that the volcanoes and coronae would be granitic, particularly as they are geologically young (erupted during today’s dry Venus). It would be useful to plot the emissivity of the tesserae separately from the other highlands to see if they are similar. Also, a study on the correlation of morphology and retrieved emissivity would be interesting, since morphology of many volcanic structures allows us to infer the chemical composition of the lavas [e.g., Head et al., 1992]. Using the gravity data, we can examine whether there is a systematic difference in emissivity in the same altitude region between isostatically compensated and dynamically supported highlands [e.g., Smrekar and Phillips, 1991]. [43] It is worth noting that material properties such as grain size could alter the emissivity, which may correspond to surface roughness or age (older terrains may acquire impact debris over time). It is also possible some differences might be due to weathering or that older terrains are covered by more impact debris which subdues the spectral absorptions. Chemical weathering is a function of altitude, and might be important in connection with anomalous radar

7 of 10

E00B24

HASHIMOTO ET AL.: FELSIC HIGHLAND CRUST ON VENUS

reflectivity [e.g., Klose et al., 1992; Schaefer and Fegley, 2004]. These issues will require further consideration. 3.6. CO2 Continuum Absorption [44] The opacities of pressure-induced continuum absorption are important and poorly known factor for calculating the emission from the lower levels of the Venus atmosphere. We used the cloud model of Pollack et al. [1993] to determine the continuum absorption coefficients, but these values are somewhat different from those used in other analyses [e.g., Be´zard et al., 1990; Pollack et al., 1993]. The emissivity contrast between highlands and lowlands revealed by our analysis will decrease with decreasing the continuum opacity. Since more radiation is absorbed above lowlands than above highlands, the decrease in the continuum opacity allows lowlands to emit more radiation. Therefore, smaller continuum opacity leads to an estimation of smaller emissivity in the lowlands. There is still a large uncertainty in the parameters of the clouds, and ambiguity between average cloud opacity and continuum opacity. If the total cloud opacity is larger than that of the Pollack’s cloud model, the emissivity difference between highlands and lowlands retrieved by our analysis may decrease or even disappear. [45] To analyze the thermal emission from the Venus nightside, Meadows and Crisp [1996] used a different approach for modeling the CO2 far wings, instead of a constant binary continuum absorption coefficient. They determined the coefficients of a sub-Lorentzian profile by fitting laboratory data. Assuming constant surface emissivity, they derived that temperature lapse rate is dynamically stable (subadiabatic) as opposed to the dynamically unstable (superadiabatic) lapse rate found in this study. If temperature of the Venusian lower atmosphere is horizontally uniform, the results of Meadows and Crisp [1996] indicates a trend of emissivity increasing with altitude. The difference between their and our results indicates that modeling of the CO2 far wings is crucial to sense the surface and lower atmosphere of Venus. [46] It is quite obvious that we need detailed modeling and more accurate laboratory measurements for pressureinduced continuum absorption. Although the approach of Meadows and Crisp [1996] is no less empirical than the constant binary coefficient, it may provide insights into the processes responsible for the CO2 continuum absorption. There is also certainly a need for constraining the total cloud opacity. Either by new in situ measurements or possibly by spectroscopy with higher spectral resolution, more reliable estimates of the Venusian surface and lower atmosphere will be obtained.

4. Implication for Venus Geology and Evolution [47] Near-infrared emissivity can be used to discriminate the rock compositions [e.g., Ross et al., 1969; Baird, 1984b]. The emissivity, , is related to the reflectivity, r, by Kirchhoff’s law ( = 1  r). The reflectance spectra at near-infrared wavelength is most sensitive to the presence of iron-bearing mafic minerals [e.g., Burns, 1993]. For mafic rocks with large amounts of mafic minerals, the emissivity around 1 mm is high; while for felsic rocks with small amounts of mafic minerals, the emissivity around 1 mm is low [e.g., Baird, 1984a].

E00B24

[48] Our results indicate that there is a significant difference in emissivity between highlands and lowlands. The relative difference in surface emissivity between lowlands and highlands are consistent with lowlands materials being composed of generally mafic rocks, whereas highlands materials are composed of generally felsic rocks. We cannot yet determine which value of  corresponds to which types of rocks, such as basalt, andesite, and granite. Nevertheless, the observed large disparity in  between the lowland and highland is very difficult to be accounted for by the emissivity variation of basalts alone but requires presence of more felsic rocks, such as granites [e.g., Hashimoto and Sugita, 2003]. [49] This inference is consistent with several observations. The radar images of the Venusian surface by Magellan revealed that most landforms in the lowland plains suggest low-viscosity materials, which are characteristic of mafic magmas [e.g., Head et al., 1992]. The data of Venera and Vega landers are interpreted as that lowland materials are dominated by mafic rocks [e.g., Pieters et al., 1986; Kargel et al., 1993]. The inferred composition contrast between lowlands and highlands is also consistent with the principle of isostasy that is the application of Archimedes’ principle to the crust. Isostatic compensation at the surface can be achieved either by variation of crustal density or by variation of thickness of homogeneous crust. The estimated emissivity indicates that highland materials are less dense than lowland materials, since density of mafic rocks is higher than that of felsic rocks. [50] One interpretation for the felsic highland crust is that highlands are composed of granitic rocks. The presence of granitic rocks would imply that Venus had have an ocean and subduction in the past, since it has been suggested that remelting of oceanic crust combined with water along subduction would have caused the formation of granitic magmas [e.g., Campbell and Taylor, 1983]. If Venus had been an Earth-like planet where the surface was covered with ocean and subduction processes were working, granitic magmas would be generated like the present Earth. Although the present Venus is a dry-planet where granitic magmas would not be generated any more, ancient continents might have survived until today. [51] A wet origin for Venus is also suggested by numerical simulations of late-stage accretion which begins after the end of oligarchic growth of planetary embryos [e.g., Morbidelli et al., 2000; Raymond et al., 2004]. Such models indicated that Venus’s initial endowment of volatiles is likely similar to that of Earth. In fact, there is similarities between Venus and Earth in the near-surface inventories of carbon and nitrogen [e.g., Le´cuyer et al., 2000]. It indicates that Venus’s inventory of water was once similar to that of the Earth at the end of planetesimal accretion. In addition, loss of Venus’s hydrogen, that is necessary for early wet Venus to evolve into the current dry Venus, is suggested by the hydrogen isotopic ratio (D/H) measured in the atmosphere of Venus [e.g., DeBergh et al., 1991; Donahue and Hodges, 1992]. The factor of 100 enrichment in D/H compared to the Earth’s ocean is generally considered to be evidence that Venus started with at least 100 times as much hydrogen as it has now [Donahue et al., 1982]. Since early Sun’s luminosity is fainter than the present, early Venus

8 of 10

E00B24

HASHIMOTO ET AL.: FELSIC HIGHLAND CRUST ON VENUS

would have avoided going into a state of runaway greenhouse [e.g., Kasting, 1988]. If Venus was endowed with sufficient water, early Venus would have had an ocean for more than a billion years [Hashimoto et al., 2007]. [52] It would be an important future work to look for other evidences of ancient ocean on Venus. Unfortunately, we cannot infer the ancient Venus from the observation of Venusian surface morphology, since the age of the Venus surface is no more than 1 billion years [e.g., Basilevsky et al., 1997, and references therein]. However, chemical signatures of water such as hydrous minerals may remain over a longer time period [e.g., Johnson and Fegley, 2000], even as the surface morphology has been erased. It is also reasonable to expect that ancient ocean has left its traces in the planetary interior, since the presence of water should have a great influence on the evolution of planets through controlling plate tectonics and mantle convection.

5. Summary [53] We evaluated the emissivity of the Venus surface at 1.18 mm wavelength using multispectral images obtained by the Galileo NIMS. Although the surface of Venus is shrouded in a thick atmosphere and clouds, our declouding procedure adequately remove the influence of clouds. Our analysis has revealed that the emissivity of lowlands is generally higher than that of highlands. This observation indicates that highland materials are generally composed of felsic rocks, while lowland materials are dominated by mafic rocks. Such a composition contrast between lowlands and highlands is consistent with the principle of isostasy. Since formation of granitic magmas is likely related to water, there is an implication of ancient wet Venus in the presence of granitic terrains. [54] It would be an important future work to confirm the presence of granitic rocks with the intention of revealing the history of the Venus surface environment. There are five spectral windows in the Venus atmosphere between 0.85 and 1.18 mm that are sensitive to the surface property. We will be able to evaluate the surface emissivity at these spectral windows in the same way as described in this paper. Such spectral information will help us estimate the mineralogic composition of the Venus surface. [55] It is also worth mentioning that the declouding procedure used in this study is also useful in retrieving temperature profiles in the Venus lower atmosphere. There are several spectral windows in the near-infrared wavelengths that can sense temperatures at several altitude levels below the cloud top. A multilevel global temperature fields retrieved from multiple-wavelength images will provide new insights into the general circulation of the Venus atmosphere. [56] Acknowledgments. The authors express their thanks to anonymous reviewers for their constructive comments on this manuscript. G.L.H. would like to thank Teruyuki Nakajima for detailed discussion of radiative transfer modeling, Naomoto Iwagami for helpful discussion of gas absorption, Yutaka Abe for fruitful discussion of early evolution of terrestrial planets, Yukari Tsutsumi for her help in developing computational code, and Constantine Tsang for his technical aid in preparing figures. G.L.H. and S.S. thank Grants-in-Aid for Scientific Research and The 21st Century COE Program of Origin and Evolution of Planetary Systems of Ministry of Education, Culture, Sports, Science and Technology (MEXT) for support. M.R.S. is supported by Portuguese FCT project PDCTE/FNU/49822/2003 and FCT scholarship BSAB/584/2006.

E00B24

References Allen, D. A., and J. W. Crawford (1984), Cloud structure on the dark side of Venus, Nature, 307, 222 – 224. Baines, K. H., et al. (2000), Detection of sub-micron radiation from the surface of Venus by Cassini/VIMS, Icarus, 148, 307 – 311. Baines, K. H., S. Atreya, R. W. Carlson, D. Crisp, P. Drossart, V. Formisano, S. S. Limaye, W. J. Markiewicz, and G. Piccioni (2006), To the depths of Venus: Exploring the deep atmosphere and surface of our sister world with Venus Express, Planet. Space Sci., 54, 1263 – 1278. Baird, A. K. (1984a), Iron variation within a granitic pluton as determined by near-infrared reflectance, J. Geol., 92, 344 – 350. Baird, A. K. (1984b), Rapid discrimination of granitic rock compositions by low-resolution near-infrared reflectance, J. Geophys. Res., 89, 2491 – 2496. Basilevsky, A. T., J. W. Head, G. G. Schaber, and R. G. Strom (1997), The resurfacing history of Venus, in Venus II, edited by S. W. Bougher, D. M. Hunten, and R. J. Phillips, pp. 1047 – 1084, Univ. of Ariz. Press, Tucson. Be´zard, B., C. de Bergh, D. Crisp, and J.-P. Maillard (1990), The deep atmosphere of Venus revealed by high-resolution night-side spectra, Nature, 345, 508 – 511. Bohren, C. R., and D. R. Huffman (1983), Absorption and Scattering of Light by Small Particles, 530 pp., John Wiley, New York. Bullock, M. A., and D. H. Grinspoon (2001), The recent evolution of climate on Venus, Icarus, 150, 19 – 37. Burns, R. G. (1993), Origin of electronic spectra of minerals in the visible to near-infrared region, in Remote Geochemical Analysis: Elemental and Mineralogical Composition, edited by C. M. Pieters and P. A. J. Englert, pp. 3 – 29, Cambridge Univ. Press, New York. Campbell, B. A., R. E. Arvidson, M. K. Shepard, and R. A. Brackett (1997), Remote sensing of surface processes, in Venus II, edited by S. W. Bougher, D. M. Hunten, and R. J. Phillips, pp. 503 – 526, Univ. of Ariz. Press, Tucson. Campbell, I. H., and S. R. Taylor (1983), No water, no granites — No oceans, no continents, Geophys. Res. Lett., 10, 1061 – 1064. Carlson, R. W., and F. W. Taylor (1993), The Galileo encounter with Venus: Results from the Near-Infrared Mapping Spectrometer, Planet. Space Sci., 41, 475 – 476. Carlson, R. W., et al. (1991), Galileo infrared imaging spectrometer measurements at Venus, Science, 253, 1541 – 1548. Carlson, R. W., P. R. Weissman, W. D. Smythe, and J. C. Mahoney (1992), Near-Infrared Mapping Spectrometer experiment on Galileo, Space Sci. Rev., 60, 457 – 502. Carlson, R. W., K. H. Baines, M. Girard, L. W. Kamp, P. Drossart, T. Encrenaz, and F. W. Taylor (1993a), Galileo/NIMS near-infrared thermal imagery of the surface of Venus, Proc. Lunar Planet. Sci. Conf., 24th, 253. Carlson, R. W., L. W. Kamp, K. H. Baines, J. B. Pollack, D. H. Grinspoon, T. Encrenaz, P. Drossart, and F. W. Taylor (1993b), Variations in Venus cloud particle properties: A new view of Venus’s cloud morphology as observed by the Galileo Near-Infrared Mapping Spectrometer, Planet. Space Sci., 41, 477 – 485. Crisp, D., W. M. Sinton, K. W. Hodapp, B. Ragent, F. Gerbault, J. Goebel, R. Probst, D. Allen, and K. Pierce (1989), The nature of the near-infrared features on the Venus night side, Science, 246, 506 – 509. DeBergh, C., B. Be´zard, T. Owen, D. Crisp, J.-P. Maillard, and B. L. Lutz (1991), Deuterium on Venus: Observation from Earth, Science, 251, 547 – 549. Donahue, T. M., and R. R. Hodges Jr. (1992), Past and present water budget of Venus, J. Geophys. Res., 97, 6083 – 6091. Donahue, T. M., J. H. Hoffman, R. R. Hodges Jr., and A. J. Watson (1982), Venus was wet: A measurement of the ratio of D to H, Science, 216, 630 – 633. Drossart, P., et al. (1993), Search for spatial variations of the H2O abundance in the lower atmosphere of Venus from NIMS-Galileo, Planet. Space Sci., 41, 495 – 504. Esposito, L. W., R. G. Knollenberg, M. Y. Marov, O. B. Toon, and R. P. Turco (1983), The clouds and hazes of Venus, in Venus, edited by D. M. Hunten et al., pp. 484 – 564, Univ. of Ariz. Press, Tucson. Fischer, J., R. R. Gamache, A. Goldman, L. S. Rothman, and A. Perrin (2003), Total internal partition sums for molecular species in the 2000 edition of the HITRAN database, J. Quant. Spectrosc. Radiat. Transfer, 82, 401 – 412. Goody, R. M., and Y. L. Yung (1989), Atmospheric Radiation, 519 pp., Oxford Univ. Press, Oxford, U. K. Hashimoto, G. L., and Y. Abe (2000), Stabilization of Venus’ climate by a chemical-albedo feedback, Earth Planets Space, 52, 197 – 202. Hashimoto, G. L., and Y. Abe (2005), Climate control on Venus: Comparison of the carbonate and pyrite models, Planet. Space Sci., 53, 839 – 848.

9 of 10

E00B24

HASHIMOTO ET AL.: FELSIC HIGHLAND CRUST ON VENUS

Hashimoto, G. L., and T. Imamura (2001), Elucidating the rate of volcanism on Venus: Detection of lava eruptions using near-infrared observations, Icarus, 154, 239 – 243. Hashimoto, G. L., and S. Sugita (2003), On observing the compositional variability of the surface of Venus using nightside near-infrared thermal radiation, J. Geophys. Res., 108(E9), 5109, doi:10.1029/2003JE002082. Hashimoto, G. L., Y. Abe, and S. Sugita (2007), The chemical composition of the early terrestrial atmosphere: Formation of a reducing atmosphere from CI-like material, J. Geophys. Res., 112, E05010, doi:10.1029/ 2006JE002844. Head, J. W., L. S. Crumpler, J. C. Aubele, J. E. Guest, and R. S. Saunders (1992), Venus volcanism: Classification of volcanic features and structures, associations, and global distribution from Magellan data, J. Geophys. Res., 97, 13,153 – 13,198. Johnson, N. M., and B. Fegley Jr. (2000), Water on Venus: New insights from tremolite decomposition, Icarus, 146, 301 – 306. Kargel, J. S., G. Komatsu, V. R. Baker, and R. G. Strom (1993), The volcanology of Venera and VEGA landing sites and the geochemistry of Venus, Icarus, 103, 253 – 275. Kasting, J. F. (1988), Runaway and moist greenhouse atmospheres and the evolution of Earth and Venus, Icarus, 74, 472 – 494. Keady, J. J., and D. P. Kilcrease (2000), Radiation, in Allen’s Astrophysical Quantities, edited by A. N. Cox, pp. 95 – 120, Springer, New York. Klose, K. B., J. A. Wood, and A. Hashimoto (1992), Mineral equilibria and the high radar reflectivity of Venus mountaintops, J. Geophys. Res., 97, 16,353 – 16,369. Knollenberg, R. G., and D. M. Hunten (1980), The microphysics of the clouds of Venus: Results of the Pioneer Venus particle size spectrometer experiment, J. Geophys. Res., 85, 8039 – 8058. Lecacheux, J., P. Drossart, P. Laques, F. Deladerrie`re, and F. Colas (1993), Detection of the surface of Venus at 1.0 mm from ground-based observations, Planet. Space Sci., 41, 543 – 549. Le´cuyer, C., L. Simon, and F. Guyot (2000), Comparison of carbon, nitrogen and water budgets on Venus and the Earth, Earth Planet. Sci. Lett., 181, 33 – 40. Marov, M. Y., V. E. Lystsev, V. N. Lebedev, N. L. Lukashevich, and V. P. Shari (1980), The structure and microphysical properties of the Venus clouds: Venera 9, 10 and 11 data, Icarus, 44, 608 – 639. Meadows, V. S., and D. Crisp (1996), Ground-based near-infrared observations of the Venus nightside: The thermal structure and water abundance near the surface, J. Geophys. Res., 101, 4595 – 4622. Morbidelli, A., J. Chambers, J. I. Lunine, J. M. Petit, F. Robert, G. B. Valsecchi, and K. E. Cyr (2000), Source regions and timescales for the delivery of water to the Earth, Meteorit. Planet. Sci., 35, 1309 – 1320. Moroz, V. I. (2002), Estimates of visibility of the surface of Venus from descent probes and balloons, Planet. Space Sci., 50, 287 – 297. Palmer, K. F., and D. Williams (1975), Optical constants of sulfuric acid: Application to the clouds of Venus?, Appl. Opt., 14, 208 – 219. Penndorf, R. (1957), Tables of the refractive index for standard air and the Rayleigh scattering coefficient for the spectral region between 0.2 and 20.0 m and their application to atmospheric optics, J. Opt. Soc. Am., 47, 176 – 182. Pettengill, G. H., P. G. Ford, and S. Nozette (1982), Venus: Global surface radar reflectivity, Science, 217, 640 – 642. Pettengill, G. H., P. G. Ford, W. T. K. Johnson, R. K. Raney, and L. A. Soderblom (1991), Magellan: Radar performance and data products, Science, 252, 260 – 265. Pettengill, G. H., P. G. Ford, and R. J. Wilt (1992), Venus surface radiothermal emission as observed by Magellan, J. Geophys. Res., 97, 13,091 – 13,102. Pettengill, G. H., B. A. Campbell, D. B. Campbell, and R. A. Simpson (1997), Surface scattering and dielectric properties, in Venus II, edited by S. W. Bougher, D. M. Hunten, and R. J. Phillips, pp. 527 – 546, Univ. of Ariz. Press, Tucson.

E00B24

Pieters, C. M., et al. (1986), The color of the surface of Venus, Science, 234, 1379 – 1383. Pollack, J. B., et al. (1993), Near-infrared light from Venus’ nightside: A spectroscopic analysis, Icarus, 103, 1 – 42. Ragent, B., and J. Blamont (1980), The structure of the clouds of Venus: Results of the Pioneer Venus nephelometer experiment, J. Geophys. Res., 85, 8015 – 8089. Raymond, S. N., T. Quinn, and J. I. Lunine (2004), Making other earths: Dynamical simulations of terrestrial planet formation and water delivery, Icarus, 168, 1 – 17. Ross, H. P., J. E. M. Adler, and G. R. Hunt (1969), A statistical analysis of the reflectance of igneous rocks from 0.2 to 2.65 microns, Icarus, 11, 46 – 54. Rothman, L. S., et al. (2003), The HITRAN molecular spectroscopic database: Edition of 200 0 including updates through 2001, J. Quant. Spectrosc. Radiat. Transfer, 60, 665 – 710. Schaefer, L., and B. Fegley Jr. (2004), Heavy metal frost on Venus, Icarus, 168, 215 – 219. Seiff, A. (1983), Thermal structure of the atmosphere of Venus, in Venus, edited by D. M. Hunten et al., pp. 215 – 279, Univ. of Ariz. Press, Tucson. Seiff, A., J. T. Schofield, A. J. Kliore, F. W. Taylor, S. S. Limaye, H. E. Revercomb, L. A. Sromovsky, V. V. Kerzhanovich, V. I. Moroz, and M. Y. Marov (1985), Models of the structure of the atmosphere of Venus from the surface to 100 kilometers altitude, Adv. Space Res., 5, 3 – 58. Smrekar, S. E., and R. J. Phillips (1991), Venusian highlands — Geoid to topography ratios and their implications, Earth Planet. Sci. Lett., 107, 582 – 597. Stone, P. H. (1975), The dynamics of the atmosphere of Venus, J. Atmos. Sci., 32, 1005 – 1016. Surkov, Y. A., V. L. Barsukov, L. P. Moskalyeva, V. P. Kharyukova, and A. L. Kemurdzhian (1984), New data on the composition, structure, and properties of Venus rock obtained by Venera 13 and Venera 14, Proc. Lunar Planet. Sci. Conf. 14th, Part 2, J. Geophys. Res., 89, suppl., B393 – B402. Surkov, Y. A., L. P. Moskalyova, V. P. Kharyukova, A. D. Dudin, G. G. Smirnov, and S. Y. Zaitseva (1986), Venus rock composition at the Vega 2 landing site, Proc. Lunar Planet. Sci. Conf. 17th, Part 1, J. Geophys. Res., 91, suppl., E215 – E218. Surkov, Y. A., F. F. Kirnozov, V. N. Glazov, A. G. Dunchenko, L. P. Tatsy, and O. P. Sobornov (1987), Uranium, thorium, and potassium in the Venusian rocks at the landing sites of Vega 1 and 2, Proc. Lunar Planet. Sci. Conf. 17th, Part 2, J. Geophys. Res., 92, suppl., E537 – E540. Taylor, F. W., D. Crisp, and B. Be´zard (1997), Near-infrared sounding of the lower atmosphere of Venus, in Venus II, edited by S. W. Bougher, D. M. Hunten, and R. J. Phillips, pp. 325 – 351, Univ. of Ariz. Press, Tucson. Vardavas, I. M., and J. H. Carver (1984), Solar and terrestrial parameterizations for radiative-convective models, Planet. Space Sci., 32, 1307 – 1325. 

K. H. Baines, R. W. Carlson, and L. W. Kamp, Jet Propulsion Laboratory, California Institute of Technology, M/S 183-601, 4800 Oak Grove Drive, Pasadena, CA 91109, USA. ([email protected]; rcarlson@lively. jpl.nasa.gov; [email protected]) M. S. Gilmore, Department of Earth and Environmental Sciences, Wesleyan University, 265 Church Street, Middletown, CT 06459, USA. ([email protected]) G. L. Hashimoto, Laboratory for Earth and Planetary Atmospheric Science, Organization of Advanced Science and Technology, Kobe University, 1-1 Rokkodai, Nada, Kobe 657-8501, Japan. ([email protected]) M. Roos-Serote, Lisbon Astronomical Observatory, Tapada da Ajuda, P-1349-018, Lisbon, Portugal. ([email protected]) S. Sugita, Department of Complexity Science and Engineering, Graduate School of Frontier Science, University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa, Chiba, 277-8561, Japan. ([email protected])

10 of 10

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114, E00B40, doi:10.1029/2008JE003316, 2009

for

Full Article

Climate evolution of Venus F. Taylor1 and D. Grinspoon2 Received 19 December 2008; revised 5 March 2009; accepted 16 March 2009; published 20 May 2009.

[1] The processes in the atmosphere, interior, surface, and near-space environment that

together maintain the climate on Venus are examined from the specific point of view of the advances that are possible with new data from Venus Express and improved evolutionary climate models. Particular difficulties, opportunities, and prospects for the next generation of missions to Venus are also discussed. Citation: Taylor, F., and D. Grinspoon (2009), Climate evolution of Venus, J. Geophys. Res., 114, E00B40, doi:10.1029/2008JE003316.

1. Introduction [2] A dramatically hot and dry climate is found on Venus at the present time, probably the result of the atmosphere following a divergent evolutionary path from a more Earthlike beginning. A key question is to what extent the two planets and their early inventories of gases and volatiles were physically alike, given their common origin in the same region of the young Solar System. Venus and Earth have nearly the same size and mass, and most current models suggest they have similar chemical compositions and interior structures. However, factors such as the small discrepancy in mean density (after allowing for compressional effects, see Ringwood and Anderson [1977]), and the absence of an internal dynamo [Luhmann and Russell, 1997], as well as discrepancies in the abundances of the noble gases [Pepin, 2006], fuel a lively debate about the extent to which the two planets can be considered to be a close match owing to essentially identical origins. The common assumption of identical origins is also clouded by the possibility of stochastic variations in late accretion history leading to unequal volatile inventories [Morbidelli et al., 2000] or volatile loss and interior processing through catastrophic early impacts [Davies, 2008; Zahnle, 2006; Alemi and Stevenson, 2006]. Even if we understood these issues, deriving the path and time scales of Venus’ divergent evolution to its present state would still present numerous challenges, not least in terms of explaining the high temperature and pressure and low water abundance at the surface. [3] Taylor et al. [2007, p. 160] posed a thought experiment: Suppose Venus and Earth had been swapped at birth – that is, at the time when they had accumulated virtually all of their present mass but before their atmospheres were fully evolved. Venus, with its slow retrograde rotation would then be at one astronomical unit from the Sun, and Earth somewhat closer. Venus would still have any bulk compositional differences it may have acquired as a result of forming at the closer position to the center of the protosolar cloud. What would this part of the Solar System look like today?

1 2

Department of Physics, Oxford University, Oxford, UK. Denver Museum of Nature and Science, Denver, Colorado, USA.

Copyright 2009 by the American Geophysical Union. 0148-0227/09/2008JE003316$09.00

In particular, would the atmospheres of Earth and Venus still be as dramatically different in composition, surface temperature and pressure? Would a repositioned Earth, its climate provoked by twice as much solar irradiance and its atmosphere exposed to a more aggressive solar wind, have developed a climate like that of present-day Venus? Would Venus, left more to its own devices at 1 AU from the Sun, be the planet with plate tectonics and oceans, and the one whose inhabitants were conducting our current explorations? To what extent might the chance formation of Earth’s singular moon have broken any potential symmetry? [4] In seeking to understand questions like this about the formation and initial states of the two planets, and how and by what stages their atmospheres progressed to the presently observed conditions, considerable use is made of the findings of each mission flown to Venus by spacecraft. Many of the measurements still needed (for instance, seismic determinations of the interior structure of Venus) will require substantial investment in new technologies, while other important investigations (such as the history and present activity of volcanism) can be addressed with available techniques. In this paper, we look at the current state of understanding of the atmosphere, interior, and nearspace environment of Venus, and the interactions between them that produce the climate at the surface and control its evolution (Figure 1). We review actual and potential progress on these topics, particularly that beginning to be made in the light of new results from the European Venus Express mission [Svedhem et al., 2009], and the important gaps that could be addressed by future planned projects. [5] The plan of the paper is as follows. Section 2 summarizes briefly what is known about the present-day climate, implicitly defined as the state of the atmosphere, and the main processes thought to be responsible for the energy balance and stability of the system. In section 3, some major topics relevant to possible long-term changes in the stability of the atmosphere are tackled, in particular the state of the interior of Venus, its thermal evolution and consequent resurfacing rates and outgassing into the atmosphere; surface-atmosphere chemical reactions and possible equilibrium states; and the loss processes at the top of the atmosphere, in particular as they concern the history of water on Venus. These are all very uncertain and hard to understand with the current very limited measurement base,

E00B40

1 of 22

E00B40

TAYLOR AND GRINSPOON: CLIMATE EVOLUTION ON VENUS

E00B40

However, the internal processes leading to critical climate elements like volcanism, global tectonics and magnetic field generation seem to be quite different, at least in outcome. New data from missions more advanced and challenging than anything attempted to date will be needed to address these; they are the subject of section 5.

2. Current Atmosphere and Climate

Figure 1. A conceptualization of the problem addressed in this paper and the formidable nature of its long-term goals. The properties and processes in and between the interior, surface, atmosphere, and near-space environment are all known to be important for determining the climate, in particular, the surface temperature. We want to understand these and their interrelationships not only for current conditions, but also throughout time from the formation of the planet to some future state when Venus may become more, or less, Earth-like. Each box in the diagram can be expanded into a range of different disciplines; for instance, the role of the atmosphere depends on its radiative transfer properties, on heterogeneous chemistry involving gas-liquid and gas-solid reactions, and on the circulation and dynamics. A further hidden dimension exists in the form of the same matrix for the Earth, and perhaps Mars also, and their relationship to Venus through common origins and processes. but incremental advances are made with each successful mission, and Venus Express was developed with the goal of making a key contribution. Section 4 reviews the incorporation of key processes into evolutionary climate models, which, in principle at least, are capable of simulating the present-day climate and showing how this may have developed over the history of the Solar System to date. Such models can also be used, within their limitations, to obtain some idea of how the climate may evolve in the future. A further dimension is added by our wish to understand, within the same physical framework, the important similarities and differences between this climate system and the corresponding one for the Earth (and also Mars and Titan, although we will not consider them in depth here). [6] The tools at our disposal for Venus climate modeling include not only the fruits of previous Venus missions and studies but also the methods that have been derived as part of the huge contemporary effort to predict the evolution of climate on the Earth. While the current emphasis in the latter is mainly on the very short term, decades to centuries, the similarities in size, composition, internal, and atmospheric processes mean that much of the same physics applies. Important contributions to the surface temperature result from an atmospheric CO2 – H2O – aerosol driven greenhouse effect, for example, and the input of solar energy to the climate system on each planet is similar.

2.1. Historical Background [7] Before the space age, many astronomers expected that the surface environment on Venus would resemble a more tropical version of the Earth. The Swedish Nobel Laureate Svante Arrhenius wrote nearly a century ago that the surface temperature ‘‘is calculated to be about 47°C’’, compared to an average of around 26°C in the Congo on Earth, and he inferred that the humidity on Venus is about six times higher than on Earth [Arrhenius, 1918]. Patrick Moore, in his book The Planet Venus, published in 1954, wrote that Venus could be a world in a ‘‘Cambrian’’ state, possibly complete with primitive organisms. At about the same time, however, Urey [1952, p. 222] noted that ‘‘the presence of carbon dioxide in the planet’s atmosphere is very hard to understand unless water were originally present, and it would be impossible to understand if water were present now.’’ [8] Beginning in 1956 at the U.S. Naval Research Laboratory, Earth-based microwave observations of Venus showed that the planet had an equivalent blackbody temperature of about 575 K. The scientists planning the microwave radiometer to be carried on the first spacecraft mission to Venus, Mariner 2, took this to be the probable surface temperature of the planet and, weighing up all of the available observational and theoretical evidence, planned their experiment around atmospheric models in which the surface pressure ranged from 2 to 20 bars and the composition was 75% CO2, 24% N2, and 1% H2O [Barrett et al., 1961]. The results from the experiment confirmed a high surface temperature [Barath et al., 1964] and the first direct measurements by the lander Venera 4 in 1967 delivered an improved estimate of around 675 K. Modern values for the mean surface temperature on Venus are around 730 K, which is higher than the melting point of the metals lead and tin, with excursions of more than 100 K, due primarily to topography. [9] Much of our current knowledge of the details of the Venus atmosphere and climate system was accrued by the Pioneer Venus orbiter and entry probe missions of the late 1970s and early 1980s [Russell, 1992]. Four probes sounded the clouds and lower atmosphere, returning chemical, physical, and meteorological data on the Venus atmosphere. The orbiter observed the surface of Venus with a radar altimeter and sounded the atmosphere in the infrared and ultraviolet regions of the spectrum. It also provided in situ data on the upper atmosphere, ionosphere and solar wind interaction. [10] After the Soviet Venera and VEGA missions in the early 1980s (notwithstanding the Magellan surface mapping mission which arrived at Venus in August 1990) there was a gap of 2 decades before another mission with an atmospheric focus was launched. In May 2006 Venus Express became the 28th spacecraft to arrive successfully at Venus since Mariner 2 in 1962, and the first mission to employ the near infrared spectral windows, discovered in the 1980s (see

2 of 22

E00B40

TAYLOR AND GRINSPOON: CLIMATE EVOLUTION ON VENUS

Figure 2. A comparison of measured atmospheretemperature profiles on Earth and Venus, where the vertical scale is pressure in millibars (1000 mb equals the mean surface pressure on Earth). The solid line is derived from remote sounding measurements made by the Pioneer Venus Orbiter Infrared Radiometer [Taylor et al., 1979a], extrapolated assuming a dry adiabatic lapse rate below 500 mb, and the dashed profile is derived from similar measurements by the Improved Stratospheric and Mesospheric Sounder on the Upper Atmosphere Research Satellite [Taylor et al., 1993]. the review by Taylor et al. [1997]) from orbit to carry out systematic remote sensing observations of the Venusian atmosphere below the clouds. Potential climate-related advances and serendipitous discoveries were to be used, inter alia, as a basis for (1) producing improved greenhouse models of the energy balance in the lower atmosphere, (2) validating and improving general circulation models of the atmosphere, with improved treatment of the zonal superrotation, the meridional Hadley circulation, and the polar vortices, (3) generating new climate evolution models using simple physics and chemistry constrained by measurements, and (4) comparative studies in all three areas with the other terrestrial planets including Earth. 2.2. Atmospheric Temperature Structure [11] The Bond albedo of Venus is around 2.5 times that of Earth (about 0.76 compared to about 0.3), so that Venus absorbs rather less radiative energy than Earth, despite its greater proximity to the Sun. The net effect of this, and the differences in atmospheric composition, is that the temperature profiles for the atmospheres of Earth and Venus are actually quite similar in the range where they overlap in pressure (Figure 2). The main difference is due to the effect of radiative heating in the Earth’s ozone layer, which produces a local maximum near the 1 mb pressure level that has no equivalent on Venus. [12] The remarkably un-Earth-like climate at Venus’ surface is a consequence of the fact that the pressure, and hence the temperature, both continue to rise with depth below the 1 bar level. The profile roughly follows the

E00B40

hydrostatic and adiabatic formulae, as would be expected, leading to a temperature increase of about 10 K for each km of depth below the 1 bar level. This amounts to some 450 K altogether at Venus’ surface pressure of 92 bars. If the Earth had such a high surface pressure, it too would be extremely hot, even without the increased proportion of greenhouse gases that is found on Venus. About 96% of this is carbon dioxide, which, along with water vapor and other minor constituents, and the net radiative effect of the ubiquitous cloud cover, drives the radiative energy balance at the surface in the direction of elevated temperatures. [ 13 ] Measured temperature profiles for both Venus and Earth conform reasonably well to the predictions of radiative-convective model calculations like those discussed in section 4. This confirms that the processes at work are basically the same in both cases and that, unlike many aspects of the climate on Venus, there are no mysteries, at least to first order. The factor that was so surprising when it was first discovered, the high surface temperature on Venus, is largely a consequence of the large mass of the atmosphere, rather than any mysterious thermal process. As discussed below, the atmospheric bulk may not be too surprising either, provided we can account for the history of water on Venus. [14] Conditions in the upper atmosphere are crucial for determining loss rates for atmospheric species and hence understanding the composition as a function of time, specifically issues such as the long-term water budget and the evolution of the atmospheric oxidation state and surface pressure. A number of processes are involved: dissociation, ionization, thermal and nonthermal escape, solar wind and cosmic ray erosion, meteoritic and cometary impacts. For determining the nature and extent of current losses, key measurements are temperature and composition as a function of height in the upper atmosphere, and the abundance and distribution of atoms and ions of atmospheric origin in the magnetosphere (Figure 3). [15] Temperature profiles from 80 to 140 km altitude deduced from stellar occultations observed by the SPICAV instrument on Venus Express, combined with previous measurements, show the maximum temperature (90 – 100 km), increasing with the value of solar zenith angle. A sharp maximum is seen in the temperature profile near the antisolar point, corresponding to the adiabatic heating expected in the subsolar to antisolar circulation regime that occurs above the mesopause at around 90 km [Bertaux et al., 2007]. Overall, the thermosphere of Venus is cooler than Earth’s, because of the greater abundance of carbon dioxide, which is very efficient at radiating heat to space. Above about 150 km, the temperature is approximately constant with height on the dayside at about 300 K. The terrestrial thermosphere is the seat of rapid winds, up to 1000 m s
1 or more, and this tends to redistribute energy originally absorbed from the Sun over the dark as well as the sunlit hemisphere. The result is a day-night difference of around 200 K about a mean temperature of 1000 K. On Venus however, the nighttime temperature in the thermosphere is very low, around 100 K, in contrast to 300 K on the dayside. The transition from the day to nightside values of temperature on Venus also show remarkably steep gradients [Keating et al., 1979] and modelers have great difficulty

3 of 22

TAYLOR AND GRINSPOON: CLIMATE EVOLUTION ON VENUS

E00B40

E00B40

Figure 3. Thermal structure, composition, and cloud measurements have defined the main parameters of the circulation and other key features of the climate system, as illustrated schematically here and discussed in the text. in reproducing both the minimum temperature and the short distance across the terminator with which it is attained. 2.3. Atmospheric Composition [16] The primordial atmosphere of Venus that originally formed with the solid body, like those of the other terrestrial planets, will have been lost or radically altered in the distant past as both the young Sun and the hot, postaccretional young planets went through phases of high activity. The present atmosphere would have been produced much later by outgassing associated with volcanism, a process that may still be ongoing on Venus, as it is to some extent on Earth, augmented by the influx of unknown amounts of cometary and meteoritic material. The relative contributions of these distinct sources can, to some extent, be deduced from the data which are gradually being accrued on the composition, and in particular the isotopic ratios, in the contemporary terrestrial planet atmospheres, and in comets and meteorites. [17] Comets are a rich source of volatile compounds such as carbon dioxide, water vapor, methane, and ammonia. If the last of these was the source of the nitrogen now present, allowing for processes such as the production of argon by the decay of radioactive potassium in the crust, the contemporary atmosphere could all be of external origin. On the other hand, the high abundance of sulfur in Venus’

clouds indicates extensive volcanic activity, as discussed below. Volcanoes are also prolific sources of carbon dioxide and the other gases required to explain the present-day climate of Venus, SO2 and H2O in particular. [18] Apart from carbon dioxide and water vapor, Venus’ atmosphere consists primarily of inert gases, particularly nitrogen and argon (Table 1). The amount of water present as gas and bound up with sulfuric acid and other compounds in the clouds is roughly one hundred thousand times less than exists in the oceans and atmosphere of the Earth. Thus, assuming most of the primordial water is not retained in the interior, Venus is overall very dry compared to the Earth while, at the same time, deuterium is more than one hundred times more abundant on Venus than Earth. This suggests that large amounts of water have escaped and that, unless the inventory is dominated by nonprimordial sources, Venus had much more water initially [Grinspoon, 1993; Donahue, 1999]. Loss takes place by dissociation of the water in the upper atmosphere by solar ultraviolet radiation, and the subsequent escape of the hydrogen. Both deuterium and normal hydrogen escape from the atmosphere, but the heavier isotope escapes less efficiently, leading to the observed fractionation. [19] The loss rate of the water depends strongly on its abundance in the relatively cool middle atmosphere, as well

Table 1. Composition of the Atmospheres of Venus and Earth as Fractional Abundances Except Where Parts Per Billion Is Stateda Species

Venus

Earth

Climate Significance

Carbon dioxide Nitrogen Argon Neon Water vapor Sulfur dioxide Carbonyl sulfide Carbon monoxide Atomic oxygen Hydroxyl Atomic hydrogen

0.96 0.035 0.00007 0.000005 0.000030 0.00015 0.000004 0.00004 trace trace trace

0.0003 0.770 0.0093 0.000018 .01 0.2 ppb

Major greenhouse gas Similar total amounts Evolutionary clues Evolutionary clues Volcanic, cloud, greenhouse Volcanic, cloud, greenhouse Volcanic, cloud Deep circulation High circulation, escape processes High circulation, escape processes Escape processes

a

0.00000012 trace trace trace

All except the noble gases argon and neon are observed by Venus Express instruments.

4 of 22

E00B40

TAYLOR AND GRINSPOON: CLIMATE EVOLUTION ON VENUS

as the intensity of the solar ultraviolet flux. Models of the process suggest, with considerable uncertainty however, that Venus could have lost an ocean of present-day terrestrial proportions in only a few hundred million years [Kasting et al., 1984]. Such potentially important processes as cloud-albedo feedback in the water-rich early atmosphere have yet to be included in models of early water loss from Venus. The oxygen produced at the same time is too massive to escape at any significant rate, according to Jeans’ formula, and would remain on the planet, presumably most of it bound chemically within the crust, if thermal escape were the only process available to remove it. However, recent results from the ASPERA instrument on Venus Express show that oxygen is escaping at a rate nearly half that of the hydrogen escape flux, suggesting that large amounts of O could have escaped over time through nonthermal processes. So long as liquid water remained available, the formation of carbonates would remove atmospheric carbon dioxide efficiently, as it does on the Earth. Once the surface water was gone, the mixing ratio of water vapor in the upper atmosphere would have fallen sharply and the loss rates of both forms of hydrogen, and the take up of oxygen into minerals, would have begun declining toward the present relatively low levels. With the loss of water, the removal mechanism for CO2 would be eliminated, and carbonate rocks on the surface would presumably eventually be subducted and lost to thermal decomposition, with the CO2 being irreversibly returned to the atmosphere through outgassing. [20] In the modern atmosphere of Venus, chemical cycles coupled with transport and radiative processes control the abundances of the minor constituents. The most important are the cycles involving water vapor, sulfuric acid, and their products, which maintain the cloud layers and probably also involve reactions between the atmosphere and the surface. Another is the sequence that gives rise to the observed distribution of carbon monoxide. CO is very abundant (mixing ratios on the order of a few parts per thousand by volume) in the upper atmosphere of Venus, as would be expected from the action of solar ultraviolet radiation on carbon dioxide. It is strongly depleted in the cloud layers (<1 ppmv), again not too surprisingly, since it is involved in reactions with SO2 and the other species that make up the sulfur cycle. Below the clouds, and near the surface, however, the carbon monoxide value recovers to an average value of around 30 ppmv, but with a marked decline from pole to equator. The reason for the gradient may be that CO is transported rapidly down from the thermosphere in the polar vortices and the poleward branch of the Hadley circulation, to the troposphere where it is gradually removed by reactions in the hot lower atmosphere and at the surface. [21] The issue of the bulk abundances of water and carbon dioxide, where Venus appears to have lost most of the former but, as a result, retained in its atmosphere much more of the latter, is of primary importance. Without liquid water, many of the weathering processes that affect, and possibly stabilize, the climate on the Earth [cf. Walker et al., 1981] would not operate. The relatively small amounts of water and other hydrogen containing species that exist as vapor above, within, and below the clouds, plus an unknown quantity bound up with sulfuric acid and probably other compounds in the liquid or solid cloud particles

E00B40

themselves, can reveal, through their abundance and distribution, key production and loss processes, act as tracers of the dynamics, and define the cloud chemistry. [22] Other potential, and poorly quantified reservoirs for planetary water include hydrated minerals in the crust and the interior. The indications from spectroscopic and entry probe data are that the H2O abundance is fairly constant across the globe near the surface, but highly variable in the clouds and above [Bezard et al., 2009]. The water vapor measurements prior to Venus Express above, below and within the cloud layers show a baffling disparity that is presumably, by analogy with Earth, linked to cloud formation and dissipation processes and meteorological activity in Venus’ atmosphere [Ignatiev et al., 1999; Koukouli et al., 2005]. Systematic new measurements from the Venus Express extended mission, sounding within and below the clouds for the first time, could radically improve our understanding of these. [23] Several of the other minor constituents in Venus’s atmosphere also exhibit striking amounts of temporal and spatial variability, with hints of terrestrial analogies that can be followed up with new data. During the Galileo flyby in 1991, near infrared measurements revealed an equator-topole gradient in the abundance of tropospheric carbon monoxide [Collard et al., 1993], which Taylor [1995] showed was unlikely to be volcanic in origin but could be the result of a hemispherical Hadley circulation that extended from the lower thermosphere at around 100 km all the way down to the surface. While the Galileo data had large uncertainties and limited latitude coverage, early Venus Express data [Tsang et al., 2008] are confirming the equator-to-pole gradient seen by Galileo and has revealed the symmetry between hemispheres we would expect on a planet without seasons if the dynamical explanation is correct. Measurements of the seasonal CO profile on the Earth show a related similar effect, known to be due to the descent of air rich in CO from CO2 dissociation in the mesosphere. The main differences from Venus are the generally smaller CO abundances, and the fact that enhanced values are found on Earth only over the winter pole, since the terrestrial vortex breaks up in the summer. 2.4. Clouds and Radiative Balance [24] Enough sunlight diffuses through the cloud layers on Venus to provide about 17 W per m2 of average surface insolation, about 12% of the total absorbed by Venus as a whole when the atmosphere is included [Crisp and Titov, 1997; Titov et al., 2007]. Most of the energy deposited at depth cannot escape as long-wave radiation but must instead be raised by convection along an approximately adiabatic temperature-pressure profile to a level near the cloud tops where it can radiate to space. An airless body with the same albedo and heliocentric distance as Venus would reach radiative equilibrium at a mean surface temperature of only about 230 K. This is close to the actual temperature at the Venusian cloud tops, as we should expect if they are the most important source of thermal infrared opacity in the tropopause region. Global measurements by the Pioneer Venus Orbiter of the net infrared emission and the total reflected solar energy [Schofield and Taylor, 1982] confirmed that the planet is in overall energy balance to within

5 of 22

E00B40

TAYLOR AND GRINSPOON: CLIMATE EVOLUTION ON VENUS

E00B40

Figure 4. The different components of the radiative energy budgets of Venus and Earth are shown as planet-wide averages, taking the solar irradiance at Venus as 100% and Earth as half that. (Actually, the insolation at Earth relative to that at Venus varies between 50% and 55% when the orbital eccentricities of 0.0167 and 0.007, respectively, are taken into account.) the accuracy of the measurements, limited by incomplete spectral and spatial coverage. [25] On Earth, half of the radiant energy from the Sun is deposited at the surface (50%), with smaller proportions absorbed in the atmosphere (20%) or reflected back into space (30%). On Venus, however, the proportions are more like 3%, 21%, and 76% respectively, with the bulk of the energy absorbed by the planet deposited well above the surface in the principal cloud layers (Figure 4). In an energy balance calculation for the planet as a whole these values lead to an equilibrium temperature of about 230 K for Venus and 250 K for Earth, depending on the values taken for the hemispherical Bond albedo in each case, a number which is quite uncertain even for Earth. Elaborate schemes have been proposed for measuring the total energy reflected and emitted from Earth to obtain an improved value and to monitor its variability and trends as a key component of climate research. Despite the fact that this is a difficult measurement, requiring multiple spacecraft in different orbits to do it properly, a similar project for Venus would be very valuable. It could also settle the question of whether any significant part of the high surface temperature is attributable to the release of heat from the interior, in the event (thought unlikely, but not known) that this release is enormously greater for Venus than for Earth. Interior heat release cannot, in any case, be the dominant influence in surface temperature, because of the constraints placed by the energy balance measurements mentioned above. [26] Other key differences between Venus and Earth concern the composition, microstructure and optical properties of the different types of cloud [Esposito et al., 1997]. This is a major research topic for the Earth, where climate change projections depend crucially on understanding the role of different cloud types and how they may change along with temperature, circulation, and pollution loading of the atmosphere. On Venus, corresponding studies are of course at an earlier stage, but are likely to be just as crucial for understanding the climate and its evolution. It is already clear that Venus has more than one type of cloud, with the distribution depending on depth, latitude and time [e.g.,

Titov et al., 2008; Wilson et al., 2008]. The absence of major seasonal variations in the incoming radiative flux on Venus is in contrast to the Earth and another factor that needs to be taken into account. 2.5. Circulation and Dynamics [27] As is now well known from studies of terrestrial climate change, including the most recent ice ages, variations in the circulation regimes in the atmosphere and oceans of Earth can lead to significant variations in surface temperature. Whether there is any important analog to this type of behavior on Venus is not known, but variations in cloud structure and winds are clearly seen in early Venus Express observations and their interpretation in terms of general circulation models that include the dense and in some ways ocean-like lower atmosphere, is a topic of considerable importance that is being addressed by groups in several different countries. [28] The first-order differences between the atmospheric general circulation regimes on Venus and Earth can be explained by the differences in the rotation rates of the solid bodies and in the optical depths and masses of their atmospheres [Rossow, 1985; Gierasch et al., 1997; Schubert et al., 2007]. The relative unimportance of Coriolis forces on Venus allows a single Hadley cell to extend much closer to the pole than on Earth, apparently reaching right to the edge of the polar vortex without the intermediate Ferrell cell. Carbon monoxide measurements in the deep atmosphere by the NIMS experiment on the Galileo spacecraft [Collard et al., 1993], and now by Venus Express [Tsang et al., 2008], are consistent with a deep Hadley circulation on Venus that extends from well above the clouds to the surface, and from the equator to the edge of the polar vortex (Figure 5). [29] Vortex behavior occurs in the polar region of any terrestrial planet, owing to general subsidence of cold, dense air and the propagation of zonal angular momentum in the meridional flow. On Venus, the small obliquity and the large superrotation lead to an extreme version of this effect, manifest by a sharp transition in the circulation regimes in both hemispheres at a latitude of about 65°. There, the

6 of 22

E00B40

TAYLOR AND GRINSPOON: CLIMATE EVOLUTION ON VENUS

E00B40

Figure 5. Some of the global-scale meteorological features observed on Venus that may be coupled to the general circulation and affect the climate. Hadley cell stops and we find the circumpolar collar, a belt of very cold air that surrounds the pole at a radial distance of about 2500 km and has a predominantly wave number 1 structure locked to the Sun [Taylor et al., 1980]. The vertical extent of the collar must be much less than its 5000 km diameter, and the indications from Pioneer Venus and early Venus Express data are that it may be only about 10 km deep, with a complex vertical structure [Piccioni et al., 2007]. The temperatures that characterize the collar are about 30°C colder than at the same altitude outside, so the feature generates pressure differences that would cause it to dissipate rapidly were it not continually forced by some unknown mechanism. [30] Inside the collar, the air at the center of the vortex must descend rapidly to conserve mass, and we expect to find a relatively cloud-free region at the pole, analogous to the eye of a terrestrial hurricane but much larger and more permanent. Interestingly, however, the ‘‘eye’’ of the Venus polar vortex is not circular but elongated, and with typically two brightness maxima (possibly corresponding to maxima in the downward flow) at either end of a quasi-linear feature connecting the two. This wave 2 characteristic gives the polar atmosphere a ‘‘dumbbell’’ appearance in infrared images that use the thermal emission from the planet as a source, and has led to the name polar dipole for the feature. A dipole was first seen at the north pole by Pioneer Venus [Taylor et al., 1979b], and now a similar feature has been discovered and extensively studied at the south pole as well by Venus Express [Piccioni et al., 2007]. The northern dipole was observed in successive images obtained in

1979– 1980 to be rotating about the pole with a period whose dominant component, among several, was 2.7 Earth days [Schofield and Diner, 1983], i.e., with about twice the angular velocity of the equatorial cloud markings. If angular momentum were being conserved by a parcel of air as it migrated from equator to pole the dipole might be expected to rotate five or six times faster. In fact, the ultraviolet markings are observed to keep a roughly constant zonal velocity (solid body rotation) from the equator to at least 60° latitude, and must be accelerating poleward of this if the rotation of the dipole represents the actual speed of mass motions around the pole and not simply the phase speed of a wavelike disturbance superimposed on the polar vortex. At the time of writing, many new details of the dipole collar structure are emerging from Venus Express VIRTIS maps, soundings, and movies that must, after detailed analysis, reveal much more of its true nature. In particular, it has become clear that the ‘‘dipole’’ description is too simplistic: more complicated shapes, as well as monopoles and tripoles, also occur, with remarkably rapid (for such a large feature) morphing between them, although wave 2 does seem to dominate as some theories [e.g., Elson, 1982] expect. [31] The thermal tide on Venus around the equatorial regions also has two maxima and two minima. This may not be directly connected with the polar phenomenon, since the two regions are separated by a narrow latitude band apparently free of planetary-scale waves, as well as by the predominantly wave number one collar. The Earth’s atmosphere has a small wave number two component superposed

7 of 22

E00B40

TAYLOR AND GRINSPOON: CLIMATE EVOLUTION ON VENUS

on the familiar early afternoon maximum to postmidnight minimum cycle, but this component dominates on Venus. In fact the dynamical theory of atmospheric tides, as developed for Earth, shows when applied to Venus that the observed state of affairs can be explained as primarily a consequence of the long solar day on Venus [Fels et al., 1984]. [32] The tracking of meteorological features on Venus was, for many decades, limited to the transient and quasipermanent features seen in the ultraviolet images of the cloud top region, where they revealed structures identified with Rossby and gravity wave activity [e.g., Belton et al., 1976]. In the mid-1980s, this changed with the discovery of the near-infrared windows, which permitted imaging of the deep cloud structure (see the review by Taylor et al. [1997], and references therein). The spectral imaging instruments on Venus Express have exploited this to investigate the meteorological activity that is clearly present in the deep atmosphere of Venus, revealing chaotic convective and wave activity near the equator where most of the solar energy is deposited in the clouds, with an abrupt switch to a more laminar flow at midlatitudes, and then finally a further transition to the polar vortex complex near the poles [Markiewicz et al., 2007; Sanchez-Lavega et al., 2008].

3. Processes Affecting Atmospheric Evolution 3.1. Internal Structure and Thermal Evolution [33] As already noted, Venus is generally taken to have essentially the same internal structure as the Earth. Venus certainly has a metallic core, which may however be slightly smaller than Earth’s given the difference of a few percent in mean density, and may be in a different physical state. The latter could account for the apparent absence of dynamo action, as evidenced by the lack of an intrinsic magnetic field, although the details are obscure. For understanding atmospheric evolution, the key questions are (1) if Venus had a planetary magnetic field in the past, for how long, and whether it might again have one in the future, and (2) what mechanisms are primarily responsible for the removal of heat from the interior? The first of these has a key role to play in determining the rate at which atmospheric gases including, crucially, water vapor, have escaped owing to solar wind erosion. The second is related to the history of volcanic activity on Venus and the venting of gases from the interior into the atmosphere. [34] On Venus, where lithospheric temperatures are higher than on Earth, and large excursions in surface temperature are apparently possible, the thermal evolution of the interior may even be influenced by changes in climate. If this leads to changes in the mechanisms or efficiency of heat release or outgassing, then interesting feedbacks between climate and interior evolution could potentially result [Grinspoon, 1997; Solomon et al., 1999]. [35] Phillips et al. [2001] investigated climate-interior coupled evolution models for Venus by merging a partial melting/parameterized mantle convection model with a gray radiative-convective atmospheric model. They found that positive feedback can operate by the release of water to the atmosphere via mantle melting, leading to an increase in atmospheric thermal opacity and radiative temperature gradient. The amplification of the greenhouse effect raises surface and mantle temperature, leading to an increase in the

E00B40

partial melting rate. These very simple models demonstrate the potential for a long-term, complex interplay between the interior evolution and climate. [36] It is likely to be some time before we have direct measurements of the heat flow in the crust of Venus. The Pioneer Venus Orbiter measurements of the albedo and of the heat flux from the top of the atmosphere into space, over a large but incomplete range of latitude, longitude, wavelength, and solid angle, showed that the planet as a whole is in radiative balance with the Sun to within about ±15%. This is not surprising, of course, since geothermal sources on the Earth account for only about 1 part in 5000 of the energy radiated to space, and most origin and interior models show Venus and Earth containing essentially identical heat sources. The difficulty comes when trying to explain how a terrestrial-like heat flux might be transferred from the interior to the atmosphere on Venus in the absence of plate tectonics, which plays this role on the Earth. On the basis of geochemical arguments, specifically the reaction of SO2 with calcite, CaCO3, on the surface, Fegley and Prinn [1989] argued that the answer cannot be primarily through volcanoes since, although Venus does show very extensive signs of volcanic activity, the atmosphere would have to be even richer than it is in volcanic gases like SO2 at present if this was the principal, steady state mechanism. [37] Simpler calculations based on heat flow, and which do not depend on assumptions about the composition and chemical state of Venus’ surface, do seem to add up, however. A long-term study of volcanic output on the Earth produced a number of 4  1010 W for the mean flux of energy from volcanoes, equal to about 0.1% of the total heat flux from the Earth’s interior [Pyle, 1995]. If Venus has the same total flux but it is all accounted for by volcanoes, then we would expect 1000 times as much gas, in particular sulfur dioxide, to be released. In fact, there is approximately 100,000 times as much SO2 in Venus’ atmosphere compared to Earth’s, but this could be explained by the fact that this is also the ratio of the lifetimes of the gas on the two planets when the efficient rain out mechanism that applies only on Earth is taken into account. Loss on Venus is principally by the much slower process of conversion to sulfuric acid. [38] Clearly, the uncertainties in the above argument can only be resolved with much more data on surface composition, atmospheric chemistry, and volcanic activity on Venus. Meanwhile, the other major possibility to be considered for heat loss from the interior is one in which the rigid lithosphere below the surface on Venus is relatively thin, 45 km or less on average [Schubert et al., 1997], so that conduction can transfer heat at the necessary rate, which it could not if the layer were thicker. However, Magellan observations of structures on the surface, like the massive Maxwell Montes, are inconsistent with such a thin solid crust unless there are large spatial or temporal variations in lithospheric thickness, with Maxwell and other uncompensated structures representing the areas of maximum thickness or of high plume activity. [39] Volcanism on Venus may release a different mix of gases from Earth (where each volcano is at least slightly different from every other, in any case) and determining the amount or composition of a volcanic contribution to the atmosphere may provide clues to the internal dynamics. For

8 of 22

E00B40

TAYLOR AND GRINSPOON: CLIMATE EVOLUTION ON VENUS

example, the idea that a resurfacing around 0.5 Ga ago was related to the transition from upper mantle convection to whole mantle convection implies the release at that time of gases that were trapped in the lower mantle 4.5 Ga ago. Models that predict the implications of such an event on Venus’ atmosphere composition may be constrained by the measurements made by VIRTIS and SPICAV on Venus Express [Svedhem et al., 2007]. [40] The observed abundances of atmospheric gases such as SO2 and H2O put some rough constraints on the resurfacing rate, although a quantitative study would require assumptions or petrologic calculations of the composition of Venusian volcanic gases, which is not known. The existence of SO2 and H2SO4 in the amounts already seen requires some level of outgassing in the last 30 Ma [Bullock and Grinspoon, 2001], and with some reasonable assumptions for the sulfur content of volcanic gases and the volume of intrusive versus extrusive volcanism, this can lead to a specific derivation of lava flux. Thus, SO2 and H2O measurements, and the flux required for the maintenance of the thick global cloud deck, may provide a crude lower limit, and infrared surface maps a crude upper limit, on resurfacing rates, which can then be linked to interior models and histories. 3.2. Volcanism and Volcanic Emissions [41] Atmospheric composition and chemistry models based on accurate data about the history and current level of volcanism on Venus are essential for understanding the current climate, and measurements that provide at least reliable estimates are being sought urgently, from Venus Express in the first instance. At the present time, the flux of volcanic gases into the atmosphere remains unknown, and could conceivably be as low as zero, although that seems very unlikely as it would require, among other adjustments, major revision in contemporary understanding of the sulfur cycle and the provenance of cloud aerosols. The evidence for volcanism on Venus is threefold: (1) the abundance of pristine volcanic structures seen on the surface; (2) the high level of volcanic gases, primarily sulfur containing, and especially SO2; (3) constraints on overall thermal and geological evolution provided by analogy with the Earth. [42] The global radar images obtained to date, especially from the Magellan mission, reveal thousands of volcanic features covering most of the surface of Venus. In addition to over 150 large shield volcanoes, with lava flows that often extend for hundreds of kilometers, intermediate-sized volcanic features are seen in large numbers and categorized as anemones, ticks, arachnids, etc. depending on their appearance. In many places, large numbers of small domes or vents are clustered together to form shield fields that collectively cover an area of more than 10,000 km2. There are hundreds of these on Venus, some with extensive lava flows surrounding them, while others are located within tectonic structures. Finally, around a hundred volcanic calderas have been identified on Venus, which are apparently sources of lava flows but not associated with cones or domes [Head et al., 1992]. [43] Unless they can be observed in the act of changing, the volcanoes and lava flows on the surface of Venus are difficult to date with any precision because the usual method of crater counting is unreliable, as discussed in

E00B40

section 3.3. The high sulfur content of the atmosphere, including the H2SO4 clouds, is on the other hand a powerful indicator of current, or geologically recent, activity, since gases like sulfur dioxide have a short lifetime in the atmosphere before they are removed by interaction with the surface. The latest values for the deep atmosphere abundance of SO2 from VIRTIS on Venus Express are about 180 ppm [Belyaev et al., 2008], which is more than 2 orders of magnitude too high to be at equilibrium with the surface [Fegley et al., 1997], a problem which does not exist for CO2. The time constant for the decline of the sulfur abundance in the atmosphere if the source was removed is in the range 1 – 10 Ma [Fegley and Prinn, 1989], much shorter than the period since the hypothetical global resurfacing event 500 Ma ago, indicating that the atmospheric sulfur must be of recent origin. [44] Pioneer Venus ultraviolet spectra from the first 5 years of operation show a decline by more than a factor of ten in sulfur dioxide abundance at the cloud tops, accompanied by a fall in the amount of submicron haze above the clouds [Esposito, 1984]. Venus Express SPICAV has also detected large, short-term variations in SO2 near the 100 km altitude level [Belyaev et al., 2008]. While it is not possible at present to associate these with specific eruptions on the surface, and transport effects due to local meteorology are probably a more likely cause at this great height, it is certainly true that large SO2 variations are noted in the terrestrial upper atmosphere following large eruptions. For instance, the injection of an estimated 20 million tons of SO2 into the stratosphere by the eruption of Mount Pinatubo in the Philippines in 1991 left localized contrasts of more than ten times the mean abundance, even 100 days after the event [Read et al., 1993]. [ 45 ] The mean volcanic flux of carbonic gases, mostly CO2, into the Earth’s atmosphere is estimated to be 2.4  1011 kg a
1 [Sigurdsson and Laj, 1992], while the mass of sulfur compounds is 20 times less, 1.2  1010 kg a
1. Frankel [1996] reports that a single large eruption like that of El Chichon in 1982 emits 6 million tonnes of SO2 per day, adding up to around 200 million tonnes (and possibly a similar amount of chlorine) over the main phase of the event. The fluxes for Venus may be much larger, in light of the evidence of extensive volcanism revealed in the Magellan maps of the surface, the high and variable concentrations of sulfur compounds in the atmospheric gases and clouds, and the apparent absence of plate tectonics to provide an alternative means to release heat from the interior. Until more progress is made to quantify the volcanic input to the atmosphere, its role in the climate, past, present and future, remains hard to estimate. [46] Measurements from spectroscopic instruments on orbiters like Venus Express, and by the Near-Infrared Mapping Spectrometer on the Galileo spacecraft which observed Venus during a flyby in 1990 [Hashimoto et al., 2008], can help to constrain estimates of Venus’ internal activity and investigate current volcanic emissions by searching for evidence of plumes from active vents containing high concentrations of sulfurous or other volcanic gases. Water vapor or carbon monoxide might be equally good tracers for this purpose. Unfortunately, the spectral plume detection objective has been made more difficult by the loss of the high-resolution Planetary Fourier

9 of 22

E00B40

TAYLOR AND GRINSPOON: CLIMATE EVOLUTION ON VENUS

Spectrometer instrument at the outset of the mission. A different approach involves searching for hot spots on the surface that might be fresh lava fields by mapping the thermal emission in the infrared bands that see the surface. Even a negative observation can help constrain the nature of geologic activity on Venus by providing an upper limit on the current rate of volcanism. Simulations by Hashimoto and Imamura [2001] showed that a lava flow covering 100 km2 or more, at a temperature of 915 K or more, could in principle be detectable by surface mapping in the near infrared spectral windows near 1 mm, if it can be distinguished from atmospheric opacity variations (due mainly to cloud) and from surface elevation and emissivity effects. [47] Observations have been made with the VIRTIS instrument on Venus Express through atmospheric windows at 1.02, 1.10 and 1.18 mm. While the source of most of the spatial variations seen clearly arise from topographically mediated surface temperature differences, careful declouding of the images and comparison with Magellan topography maps has begun to reveal some residual anomalies in surface emissivity which are believed to be related to composition. Helbert et al. [2008] found that emissivity variations observed in the Lada Terra region are correlated with surface geology, and that positive emissivity anomalies appear to be associated with relatively young lava flows, whereas negative anomalies seem to be found in more heavily fractured, and probably older, terrains. Hashimoto et al. [2008] also report compositional variability of surface units, with the highlands being more felsic (having a high silica content, as opposed to mafic or iron-rich minerals) than the lowlands, which they suggest implies large reservoirs of water in Venus’s distant past prior to the global resurfacing event. Further observations of this type with the Venus Express extended mission may help to clarify the resurfacing history. [48] Because the radiative balance of Venus depends sensitively on the abundances of several trace volcanic gases, including SO2 and H2O, and on the properties of the global cloud deck, itself a by-product of volcanogenic SO2 and H2O, the resurfacing history is probably linked to geologically forced climate change in Venusian history. 3.3. Resurfacing History [49] Over the past 15 years, developments in unravelling the geological history of Venus, primarily from the cratering record as revealed by the 1989 –1994 Magellan mission, have permitted quantitative assessments of the magnitude and timing of sources of volcanically derived gases to the atmosphere [e.g., Bullock et al., 1993; Bullock and Grinspoon, 1996; Kreslavsky and Head, 1999]. The surface is 80% covered by various types of volcanic plains [Basilevsky and Head, 1998], and the cumulative crater count of approximately 1000 indicates an average surface age of 700 ± 300 My [McKinnon et al., 1997]. Thus, the Magellan data set shows clearly that most of the planet has been volcanically resurfaced within the last billion years, raising the possibility that the transfer of heat, as well as of lava and atmospheric gases, is episodic rather than quasi continuous [Turcotte et al., 1999]. If this is so then the climate on Venus may also have significant periodic variations driven by changes in atmospheric composition, density, and cloud properties.

E00B40

[50] The spatial distribution of impact craters is indistinguishable from a random distribution, suggesting that, to first order, the plains areas of Venus are of uniform age [Schaber et al., 1992; Strom et al., 1994]. However, the small total number of craters and, in particular, the absence of small craters (less than 4 km across) due to atmospheric filtering of the smaller impactors makes age dating of discrete areas impossible and has led some researchers to question the conclusion of uniform age [Campbell, 1999]. Most of the craters are in a pristine state, with only a small percentage showing signs of tectonic (7%) or volcanic (33%) modification or embayment [Schaber et al., 1992]. This supports the conclusion that the ages of the plains are highly uniform and that the global resurfacing happened on a very short time scale relative to the cratering age, with a production population of mostly unaltered craters forming after an early burst of resurfacing faded [Bullock et al., 1993; Basilevsky and Head, 1996, 1998, 2000, 2002, 2003, 2006]. [51] An alternative view has been proposed in which volcanism has been random in space and time [Guest and Stofan, 1999; Addington, 2001]. Intersection relationships among sets of wrinkle ridges have been interpreted as suggesting that the regional stress domains that produced them differ significantly in age [McGill, 1993], in contrast to the view that much of the plains deformation was globally coherent [Bilotti and Suppe, 1999; Solomon et al., 1999]. Herrick and Sharpton [2000] found on the basis of stereo-derived topography of Venusian impact craters that the number of unmodified craters may have been previously underestimated, which could lead to an underestimation of the resurfacing rate over the past several hundred My. An analysis of resurfacing rates and styles based on an assessment of 18 mapped Venusian quadrangles suggests a global resurfacing history more complex than that advocated by Basilevsky and Head [1996], with a wide range of volcanic styles occurring throughout a period of Venusian history with a duration that is likely to exceed 100 My [Stofan et al., 2005]. The latter researchers, however, conclude that the lack of ancient impact basins implies planet-wide resurfacing by lava to depths of at least 1 km. [52] Although individual features cannot be reliably dated, several researchers have attempted to derive relative ages of different terrain and feature types, using the fact that the cumulative area of these spatially discontinuous features is sometimes large enough for more meaningful cratering statistics. This technique relies on the extreme assumption that the formation ages of similar features are identical. However, even if this assumption is not strictly true, the technique may still detect real trends in age relationships that would otherwise be invisible to our current arsenal of observational techniques. Large shield volcanoes appear to have fewer impact craters and therefore have been interpreted to be on average younger than the plains [Namiki and Solomon, 1994], and the highly deformed tesserae are most likely the oldest terrain on the planet [Ivanov and Basilevsky, 1993]. Price and Suppe [1994] and Price et al. [1996] estimated average ages of (1.1 ± 0.1) T for the ridged and shield plains, (0.3 ± 0.2) T for large volcanoes, and (0.5 ± 0.3) T for lobate and smooth plains, where T is the average crater retention age of the entire planet’s surface.

10 of 22

E00B40

TAYLOR AND GRINSPOON: CLIMATE EVOLUTION ON VENUS

[53] Other researchers have used the state of preservation of impact craters and superposed aeolian features such as parabolic and circular halos around craters to infer chronology [Izenberg et al., 1994]. Dark parabolas, often extending several tens of crater diameters, are most likely due to the dispersion of impact generated fine particles that become entrained in the superrotating winds. Izenberg et al. [1994] concluded that owing to either continuing aeolian activity or weathering, the dark parabolas become dark and then light halos. They inferred that dark parabolas have ages of less than 0.1 to 0.15 T, and that dark halos around craters are between 0.5 and 0.1 T in age. Craters with no halos are older that about 0.5 T. Phillips and Izenberg [1995] suggested that dark halos are removed by volcanic, and other endogenic, processes. They observed that (presumably older) areas of high crater spatial density also have high fractions of halo-free free craters. They also found more halo-free craters in some regions of low crater spatial density, especially in the Beta-Atla-Themis region, which has a relatively high fraction of modified craters, suggesting more recent geological activity. Basilevsky and Head [2003] have also applied the prevalence of radar-dark deposits associated with craters in some areas to the dating of relatively recent surfaces and structures. [54] The plains of Venus are almost entirely lacking in severely embayed craters: those which are completely flooded except for the crater rim, in contrast to say, the lunar maria, where such features are common. Those severely embayed craters that do exist are shown to have been flooded by flows from discreet volcanic edifices that most likely postdate the formation of the plains. If one accepts that the majority of impact craters, while perhaps showing some minor degree of volcanic modification, do not predate the plains volcanism which resurfaced most of the planet in the last 300 –1000 My, then it is clear that, regardless of the (still contentious) details, the rate of resurfacing has declined precipitously over this time period. Several of the analyses described above support the view that the resurfacing activity peaked strongly within 10 to 100 million years of the mean surface age. Clearly, further detailed observations, both orbital and in situ, are required before the history of resurfacing and volcanic outgassing on Venus can be confidently known. 3.4. Geological Constraints on Atmospheric Evolution [55] To first order, a plausible explanation for the apparent superabundance of CO2 on Venus relative to Earth is not particularly difficult to find. It has been estimated that the carbonate rocks on the Earth hold the equivalent of roughly 60 bars of CO2 [Kasting, 1988], but since a volcanic source has clearly been active on Venus and since the conversion of atmospheric to crustal carbonate occurs much more efficiently in the presence of liquid water to dissolve the CO2 first, the relatively water-depleted state of Venus may be responsible for so much of the gas remaining in the atmosphere. However, Venus has not always been so dry. The evidence from the D/H ratio, plus the cosmogonical argument that Venus should have accreted with similar amounts of H2O to the Earth, both suggest that Venus, too, was once covered by oceans to a considerable depth. How long this state survived is not known; nor is the abundance of carbonates in the component of Venus’ crust

E00B40

that is, or has been, in contact with the atmosphere and hydrosphere [Donahue et al., 1997]. As noted above, some or all of any carbonate formed in the early stages could have been recycled into atmospheric CO2 by high-temperature thermal processes during subduction of the crust. [56] However, the crucial question of whether the current surface pressure on Venus is stable remains an interesting and important one. It is well known that the CO2 abundance in Earth’s atmosphere can vary, owing to natural and anthropogenic factors, and that it is increasing at the present time, with likely consequences for the global climate. Over time scales greater than 106 years, the terrestrial CO2 abundance is regulated by the carbonate weathering cycle, and has gradually decreased over billions of years as the Sun’s main sequence brightness has increased [Walker et al., 1981]. If the climate on Venus is stable in the long-term then it is likely that some mechanism provides a buffer that stabilizes the atmospheric carbon dioxide content. Since Urey [1952] proposed the exchange between atmospheric CO2 and common minerals in the surface, it has been shown that the reaction (CaCO3 (calcite) + SiO2 (quartz) $ CaSiO3 (wollastonite) + CO2) reaches equilibrium at precisely the temperature and pressure found on the surface of Venus. Either this is a coincidence or the reaction proposed by Urey, augmented or dominated by other surface chemical reactions, is actively buffering the atmospheric pressure. [57] Problems have been raised with this theory however [see, e.g., Hashimoto and Abe, 2005], including the question of how a sufficiently intimate contact between atmosphere and lithosphere is achieved. Any answer to the latter depends on a much better understanding of the actual mineralogical composition and physical state of the exposed material on the surface of Venus, and of weathering and possible subduction and effusion rates, than will be available without future in situ studies at the Venus surface. Bullock and Grinspoon [1996] showed that although the surface temperature and pressure are indeed at an equilibrium point with the calcite-wollastonite mineral reaction, it is actually an unstable equilibrium, suggesting that unknown mechanisms may be providing the stability, requiring a more complex model of surface-atmosphere interactions that are linked to the history of volcanism and the nature of the interior. [58] In addition to any contribution to maintaining the high surface density of carbon dioxide made by coupling between the surface and the atmosphere, there must certainly be an effect on the abundance of more reactive trace species. Small changes in radiatively active atmospheric gases can change the magnitude of the Venusian greenhouse effect and shift the temperature-dependent equilibrium points of key mineral buffers, as well as the kinetics of heterogeneous reactions, resulting in climate feedbacks. Heterogeneous reactions between sulfur dioxide and the surface are seen to proceed rapidly, relative to geologic time scales, in chemical kinetics experiments performed under Venus-like conditions in the laboratory [Fegley and Prinn, 1989; Fegley and Treiman, 1992]. Since the deep atmosphere abundance of SO2 is 1– 2 orders of magnitude higher than can be accounted for by equilibrium with surface minerals [Fegley and Treiman, 1992], this implies active sources and sinks of sulfur. If surface reactions are indeed active in altering atmospheric SO2, it is of interest to assess

11 of 22

E00B40

TAYLOR AND GRINSPOON: CLIMATE EVOLUTION ON VENUS

the impact they may have on the climate of Venus using evolutionary models. 3.5. Escape Fluxes, Fractionation, and the History of Water [59] Venus has one hundred thousand times less water in its atmosphere than exists in the oceans and atmosphere of the Earth [Donahue and Pollack, 1983; Pollack et al., 1993]. The fact that, at the same time, deuterium is approximately 150 times more abundant than on Earth [McElroy et al., 1982; Donahue et al., 1982; de Bergh et al., 1991] suggests that the current complement of water derives from a reservoir that has been severely depleted. This is consistent with Venus having a much higher primordial water abundance, although it could also reflect loss of water supplied more recently by exogenous or endogenous sources [Grinspoon, 1993; Donahue, 1999]. The low water abundance (about 30 ppm) suggests a water lifetime of several hundred million years, the precise value depending on the time-averaged hydrogen escape flux. This is much shorter than the lifetime of Venus’ atmosphere, suggesting that water on Venus is currently in a steady state between source and loss processes [Grinspoon, 1987, 1993; Donahue, 1999]. Yet if water is indeed in a steady state, what is the source? Two obvious candidates are volcanic outgassing and cometary infall. If water is in a steady state then the escape flux also measures the time-averaged sum of these sources. If further information can be brought to bear on discriminating between these sources, for example through placing quantitative limits on the recent exogenous contribution through additional isotopic clues or other constraints on the impact flux in the inner solar system, then data on planetary escape fluxes can be used to quantify outgassing rates. Combined with geologically determined estimates of magma production rates, this can constrain magma volatile content. Grinspoon [1993] used such an approach in deriving a rough upper limit on average magma water content of 50 ppm by mass. [60] The loss processes involve dissociation to form hydrogen and oxygen followed by escape from the planet of hydrogen, a process which depends strongly on the abundance of water in the middle atmosphere. According to Kasting et al. [1984], Venus could have lost an ocean of present-day terrestrial proportions in less than 500 million years. These authors also suggest a reason why the D/H ratio on Venus is only greater by 100 than that on Earth. It would be much larger if all of the deuterium in the primordial Venusian ocean had been retained. However, deuterium as well as hydrogen can escape from the atmosphere in large amounts through nonfractionating hydrodynamic escape when there is free water on the surface, if the heating of the upper atmosphere by solar UV radiation is sufficiently intense. Once the free water is all gone, the mixing ratio of vapor in the upper atmosphere falls and the escape processes become highly fractionating between the two isotopes. In Kasting et al.’s [1984] model, with the simplifying assumption that all of the deuterium is lost until the last of the ocean evaporates and then none thereafter, the predicted enhancement is almost exactly that observed. These authors further point out that an extensive ocean on Venus would facilitate the disposal of the oxygen produced by water vapor dissociation. It was thought at the time that

E00B40

this could not escape efficiently, an assumption that Venus Express results now challenge, and that large amounts would have to be bound chemically in the crust through weathering processes involving liquid water. Grinspoon [1987, 1993] and Donahue [1999] have pointed out that after a phase of massive water loss, evolution of the D/H ratio during the subsequent steady state phase of water evolution would likely have at least partially obscured the primordial signal, complicating efforts to derive a relationship between this observed quantity and the evolution of Venusian water. [61] The observed high D/H ratio may be at least partly the signature of the catastrophic resurfacing and associated outgassing that apparently occurred within the past 0.5– 1 billion years, presumably accompanied by a massive injection of water followed by fractionating escape [Grinspoon, 1993]. If this occurred recently compared to the deuterium lifetime, which is longer than the hydrogen lifetime by a factor determined by the relative escape efficiency of deuterium and hydrogen, then the enhanced D/H from this episode would be largely preserved at present. Alternatively, the interior and surface may simply have been continuously more active before that time. An extremely large comet impact, or a comet shower caused by a gravitational perturbation to the Oort cloud or the breakup of a massive, volatile rich object, could also potentially leave such a signature. [62] The ability to discriminate between these different interpretations of the enhanced D/H ratio, with very different implications for water evolution, was long hampered by the large uncertainties in the current escape flux. Accurate evolutionary modeling also requires some knowledge of how the escape flux and deuterium fractionation efficiency have varied with time over a range of time scales. The ASPERA experiment on Venus Express has found evidence that a surprisingly large flux of oxygen ions is currently escaping from the upper atmosphere of Venus through nonthermal processes, calling into question the earlier assumption that massive hydrogen escape must necessarily have left behind large quantities of oxygen [Barabash et al., 2007]. Provisional estimates by the ASPERA team suggest a planetary average column hydrogen escape flux which, if it also represents the time averaged flux, is an order of magnitude lower than those previously assumed in evolutionary models [Donahue, 1999]. If substantiated, the new values would imply a residence time of atmospheric water of approximately 109 years, roughly equal to the apparent average age of the volcanic plains that dominate the surface. [63] Furthermore, if the water abundance is currently in steady state with outgassing from postplains volcanism, this low escape flux would imply magmas that are, in bulk, 2 orders of magnitude drier than the driest terrestrial magmas. This assumes a resurfacing rate of 0.4 km3 a
1, which would be consistent with mapping of volcanic features in Magellan images [Head et al., 1992; Phillips et al., 1992] when combined with simulations of the observed crater population, and roughly equivalent to the current terrestrial intraplate magmatic flux [Bullock et al., 1993], as described by Grinspoon [1993]. [64] However average escape fluxes are not so straightforward to quantify, as they cannot be specified completely by any one single instrument, which must always measure

12 of 22

E00B40

TAYLOR AND GRINSPOON: CLIMATE EVOLUTION ON VENUS

at a single place or time. The actual total hydrogen escape flux may include the escape of neutral species, ions of lower energy than can be measured by ASPERA, and may show significant variations with solar cycle and unique solar events such as coronal mass ejections. Thus, at present the ASPERA observation must be considered to represent a lower limit on the H+ escape flux, pending further observations. The coverage possible with the extended mission, plus more detailed analysis and modeling, should lead to a more representative value for the average in space and time. [65] With SPICAV/SOIR’s very high spectral resolution it is possible to study the ratio of HDO and H2O, which may shed light on the escape of H and D. In particular, by measuring simultaneous vertical profiles of H2O and HDO above the clouds, SPICAV is examining D/H fractionation. Bertaux et al. [2007] show several measured profiles of H2O and HDO from 70 to 95 km altitude in which the averaged HDO/H2O ratio equals a factor of 240 ± 25 times the ratio in Earth’s ocean, or nearly a factor of 2 times the bulk atmospheric value measured in the lower atmosphere. This surprising result could be due to some combination of (1) preferential destruction of H2 relative to HD, perhaps from photolysis induced isotopic fractionation [Liang and Yung, 2009]; (2) preferential escape of H relative to D, leaving a residue of enhanced HDO at these altitudes; or (3) selective condensation, a process that has recently been found to be important for fractionating D and H on Mars, and also on Earth [Bertaux and Montmessin, 2001]. Solving this problem will depend on obtaining a better understanding of both global dynamics and photochemistry in the upper atmosphere. At present the observations are limited to latitudes from 70 to 86°N. Venus Express extended mission observations that sample a wider range of latitudes will help to distinguish between these two fractionation mechanisms, and allow a clearer understanding of the potential of the D/H results to resolve more definitively between competing models for the history of water on Venus. [66] Observations over a significant fraction of a solar cycle will also be important for deriving a time averaged escape flux for recent epochs and for understanding the relative importance of several escape mechanisms. This will allow improved modeling of the variation of escape rate and fractionation efficiency with changing atmospheric composition, structure and solar inputs, which will be necessary for improved reconstructions of water evolution. Eventually, direct observations of surface materials may be able to find evidence for an early period of Venus history when surface water was stable and abundant. As noted above, the possibility that the highlands on Venus are more felsic than the lowlands, inferred by Hashimoto et al. [2008] using Galileo spectroscopic data, may support the existence of large reservoirs of water in Venus’s distant past, since aqueous processes are involved in the formation of felsic provinces on Earth.

4. Climate Models and Parameterizations 4.1. General Circulation Models [67] The overall goal for Venus climate modeling must be the development of fully three-dimensional, time-dependent general circulation models in which all of the relevant sources and sinks, and all radiative, dynamical, and

E00B40

chemical processes, are included with high precision and resolution. This is a goal which is being approached, although so far not attained, for terrestrial climate models, which of course are much better tested and constrained by data. Nevertheless, modified, and where necessary simplified, terrestrial GCMs are being used to model the dynamical component of the current climate of Venus, and to help understand common processes with the Earth [Yamamoto and Takahashi, 2003; Lebonnois et al., 2005; Lee et al., 2007]. [68] Experiments with these models show that global superrotation tends to develop in optically thick atmospheres on slowly rotating planets as different as Venus and Titan. However, the present state of model development, including the details of energy deposition profiles required in the model specification, is deficient in that the predicted wind speeds are too slow, by a factor of 2 or more. The features seen in ultraviolet images of Venus rotate around the planet in a period of only 4 to 5 days, corresponding to wind velocities of more than 100 m s
1 at the cloud tops, while the solid surface of Venus rotates at only about 2 m s
1, or once every 243 days. More information about cloud variability and wave modes in the atmosphere below the visible cloud tops, from repeated UV and IR mapping, should permit progress in understanding issues such as the role of the surface topography in maintaining or opposing the superrotation and the role of waves or eddies in the transport of angular momentum. 4.2. Evolutionary Models [69] Eventually, helped by the massive effort being applied to model the changing climate of the Earth, Venus GCMs will incorporate the relationships between dynamics, volcanism, exospheric escape, surface-atmosphere reactions, composition, clouds and radiative balance. For the time being, however, our attempts to trace the origins and evolution of Venus’ atmosphere depend on simplified one-dimensional evolutionary climate models that incorporate the global-scale processes and their interrelations in one (altitude) rather than three spatial dimensions, neglecting or simplifying dynamics so they can model the complex set of time-dependent feedbacks that control the planetary climate. [70] The current state of the art with 1-D evolutionary models is represented by that of Bullock and Grinspoon [2001]. In this, a radiative transfer code calculates the radiative-convective equilibrium temperature structure as a function of atmospheric composition, and is coupled to a chemical/microphysical model of Venus’ clouds, models of volcanic outgassing, models of heterogeneous reactions of atmospheric gases with surface minerals, and a model of the escape of hydrogen from the exosphere. Figure 6 shows the various modules and their coupling in the model. An atmospheric radiative transfer code is used to describe the transport and balance of energy within the atmosphere, calculating thermal infrared fluxes, heating rates, and temperature profiles that are tested for consistency with spacecraft and ground-based observations. The code must be flexible and fast enough to predict these quantities with respect to variations in solar flux and atmospheric composition as they change over time, which involves making some simplifying parameterizations. Bullock and Grinspoon used a one dimensional, two-stream model of infrared

13 of 22

E00B40

TAYLOR AND GRINSPOON: CLIMATE EVOLUTION ON VENUS

Figure 6. A block diagram of the Venus climate evolution model of Bullock and Grinspoon [2001]. At each time step the atmospheric composition is adjusted for the effects of volcanic exospheric loss, volcanic outgassing, and by reactions with the surface, then the coupled cloud and radiative-convective models are allowed to reach equilibrium. radiative transfer employing correlated-k gaseous absorption coefficients to describe the spectral properties of nine molecular species found in Venus’ atmosphere: CO2, H2O, SO2, CO, OCS, HDO, H2S, HCl, and HF. Net infrared fluxes were then calculated using the hemispheric mean approximation appropriate to an emitting, highly absorbing and scattering atmosphere, balanced using an iterative variational method against the observed solar net flux profile from the radiometer on the Pioneer Venus entry probes. [71] The resulting radiative equilibrium profile of temperature as a function of altitude calculated by this model matches the Venus International Reference Atmosphere [Kliore et al., 1986], which is based on measured temperature profiles, very well everywhere except above 70 km (Figure 7). To some extent this may be fortuitous, since the model makes a large number of simplifying assumptions, and each of these modules can be developed, and the overall scheme refined, using new data and mission findings. Previous models have obvious limitations, including those in the following list, and should be reexamined, using the new data and improved computational techniques now available: [72] 1. Convection was assumed to reduce the lapse rate in the radiative equilibrium temperature profile to the adiabatic value wherever it tended to be larger. [73] 2. Tomasko et al. [1980], and others, have shown that an extra source of opacity above the cloud tops has to be arbitrarily introduced in models before they will accurately predict the upper atmosphere temperature structure. [74] 3. Energy deposited in the atmosphere by absorption of UV radiation, mostly above 70 km, was not accounted

E00B40

for, since the net solar fluxes from Pioneer Venus have a cutoff at 0.4 mm [Tomasko et al., 1980]. [75] 4. The arbitrary addition of large ‘‘mode 3’’ cloud particles above 65 km, with a scale height of 4 km, was necessary to achieve agreement between the radiative transfer model and the VIRA temperature structure. [ 76 ] 5. The direct reactions of atmospheric CO 2 with surface silicates were neglected, though Bullock and Grinspoon [2001] noted that such reactions are possible, even likely. The kinetics of such reactions are poorly known, making it difficult to include them; future laboratory experiments to determine these rate constants would allow potentially important improvements to the model. [77] 6. The number of spectral and vertical increments in the model had fairly low maximum values of 68 and 20, respectively, and simple spectral and hemispherical integration schemes were used. [78] Further tests of a model with these and other improvements can be made through comprehensive comparisons to the radio occultation temperature profiles already available from Magellan and Venus Express orbiters [Jenkins et al., 1994; Ha¨usler et al., 2006]. 4.3. Cloud Models [79] Cloud properties, including the vertical and horizontal distribution, composition, microphysics, and variability, are notoriously difficult to model in terrestrial climate models, since they depend simultaneously on temperature, composition (including the number and composition of condensation nuclei), and dynamics. However, the difficulties must be faced because, on Venus as on Earth, changes in the thickness of clouds have two important effects on climate. They alter the visual albedo of the planet, changing the input of solar energy, and they alter the thermal infrared opacity of the mid atmosphere, affecting the temperature in the atmosphere and at the surface. [80] For their evolutionary calculations, Bullock and Grinspoon [2001] combined a thermochemical model of Venus’ cloud aerosols by Krasnopolsky and Pollack [1994] with a simple microphysical model, to predict the number

Figure 7. Temperature (solid line) calculated with the radiative transfer model of Bullock and Grinspoon [2001]. For comparison, the Venus International Reference Atmosphere [Kliore et al., 1986] is plotted with a dashed line.

14 of 22

E00B40

TAYLOR AND GRINSPOON: CLIMATE EVOLUTION ON VENUS

E00B40

Figure 8. Mixing ratios assumed in the baseline models of Bullock and Grinspoon [2001] shown as a function of altitude. The points are the SO2 mixing ratio as a function of altitude derived by Bertaux et al. [1996] from VEGA 1 and 2 entry probe data. density profile with altitude of aerosol particles as a function of the atmospheric abundances of H2O and SO2. Then, changes in cloud structure, infrared opacity, and albedo could be incorporated into the radiative transfer model using optical constants for H2SO4/H2O from Palmer and Williams [1975] and Mie scattering calculations. In addition to the large uncertainties in the thermochemistry and microphysics of the Venusian clouds, improved simulations would add the large-scale global variations in cloud structure and optical thickness that are apparent in the VIRTIS and VMC maps from Venus Express [Titov et al., 2008]. These need to be analyzed statistically, and their overall affect on radiative balance quantified, before they are incorporated in a new generation of 1-D climate models. Eventually, the realistic inclusion of these spatially and temporally variable elements will require the development of 3-D models combining the dynamical code of a GCM with full climate physics. 4.4. Interior/Surface/Atmosphere Interactions [81] Bullock and Grinspoon [1996, 2001] included the reaction of sulfur dioxide with surface calcite (CaCO3 + SO2 $ CaSO4 (anhydrite) + CO) using kinetic data measured by Fegley and Prinn [1989], who showed that this reaction proceeds rapidly under Venus surface conditions. However, the actual bulk reaction rate will depend not only on chemical kinetics but also on the ability of the gas to diffuse to new reaction sites on buried grains once the easily available surface has reacted. To solve this problem, Bullock and Grinspoon used a diffusion/reaction formalism which takes into account the temperature-dependent lifetime for SO2 reaction with surface carbonate, t, as well as the time required for the diffusion of SO2 (with temperature and porosity-dependent diffusion coefficient, D) into the planetary surface. In this formalism, the abundance n of SO2 is determined by @n @2n n ¼D 2
@t @z t

The choice of diffusion coefficient requires assumptions about soil porosity and the effectiveness with which forming CaSO4 rinds will reduce pore space. These can be tested by future surface missions and further laboratory experiments. The above equation reveals that the effect of the SO2-anhydrite buffering mechanism is temperaturedependent, through both the reaction rate and the diffusion coefficient. [82] There is another set of possibilities tied to some uncertainties and controversies about the current sulfur abundance in the lower atmosphere and the stability of sulfur-bearing minerals at the surface. Bullock and Grinspoon [2001] assumed the lower atmosphere mixing ratios of reactive gases shown in Figure 8. Here, the lower mixing ratio of SO2 is assumed to be constant below the clouds at a value of 180 ppm, as measured by Pioneer Venus, and SO2 at the surface is more than two orders more abundant than required for equilibrium with calcite, but it is close to equilibrium with pyrite and magnetite. Hashimoto and Abe [2005] have suggested that the SO2 abundance may in fact be controlled by a pyrite buffer, in which case near equilibrium may exist. However, for either of these reactions to be in equilibrium the reactants must exist at the surface. While there are no unequivocal data for the existence of either of these phases, CaCO3 is a possible interpretation of the Venera X-ray fluorescence data. FeS2 has been shown in laboratory experiments to have a lifetime of 100 days at Venus surface conditions [Fegley et al., 1995], so it is unlikely to exist for geologically relevant time scales. [83] By contrast, Bertaux et al. [1996] reporting results from VEGA 1 and 2, found a steep decline in SO2 toward the surface. If the SO2 abundance in the lowest part of the atmosphere in contact with the surface is actually only 30 ppm, as the VEGA team reported, rather than 180 ppm, then it is conceivable that SO2 is not very far out of equilibrium from the calcite-anhydrite buffer. However, the steep lower atmosphere gradient of SO2 inferred by

15 of 22

E00B40

TAYLOR AND GRINSPOON: CLIMATE EVOLUTION ON VENUS

Bertaux et al. [1996] would have to be maintained by some unknown dynamical or chemical process. [84] Preliminary results by Marcq et al. [2008], using VIRTIS on Venus Express, have pointed toward sulfur abundances just below the clouds more consistent with the Pioneer Venus, constant mixing ratio profile. Although the loss of the Planetary Fourier Spectrometer makes such determinations of lower atmosphere abundances more challenging, further results from VIRTIS on Venus Express could help to resolve between these two different pictures of lower atmosphere SO2, with important consequences for the nature of the sulfur cycle and surface/atmosphere interactions. 4.5. Exospheric Escape [85] Bullock and Grinspoon [2001] utilized the diffusion limit approximation [Chamberlain and Hunten, 1987] to calculate the loss of H and D from the top of the atmosphere. The current escape flux of H from Venus is due to two mechanisms: an electric field-driven flow of ions in the nightside hydrogen bulge, and charge exchange. Each of these processes has a different solar cycle average loss rate. Prior to the Venus Express mission, estimates of the average escape flux over time were 1.6  107 cm
2 s
1 [Donahue et al., 1997; Donahue, 1999] for H, and about a tenth of this for D. For diffusion-limited escape, where the loss rate is limited by the ability of H and D to diffuse to the exobase, these amount to H and D lifetimes in the atmosphere of 170 million years and 1.7 billion years, respectively. As discussed in section 3.5, these values will eventually need to be revised in the light of Venus Express observations. 4.6. Model Experiments [86] Their Venus climate evolution model was used by Bullock and Grinspoon [2001] to predict how surface temperatures and cloud structure responded to large-scale volcanic injections of radiatively active gases. They found that for volcanic outgassing associated with the emplacement of the largest plains units on Venus, surface temperature excursions of 100 K were possible. Outgassing was modeled as a sudden pulse of water and sulfur dioxide to the atmosphere, declining exponentially with a time constant of 100 million years, and assuming that the total amount of lava erupted onto the surface is equal to a global layer 1 km in thickness. The water content of the lava was assumed to be 50 ppm by weight and the sulfur dioxide content to be 0.2%, typical for terrestrial Ocean Island basalts and large igneous provinces, which Bullock and Grinspoon [2001] argue are likely to be the best terrestrial compositional analogs to plains magmas on Venus. Atmospheric sulfur dioxide is lost rapidly to temperature-dependent reactions with surface carbonates, while atmospheric water is lost more slowly owing to its dissociation by solar UV and the exospheric escape of H. At 735 K, the residence time for sulfur dioxide is approximately 30 million years; the residence time for atmospheric water was assumed to be 160 million years. Both atmospheric constituents decline in abundance more slowly than this owing to the continued but exponentially declining outgassing rate. The initial conditions for the model included an abundance for atmospheric water at the surface of 30 ppmv (today’s value for Venus), and an atmospheric sulfur dioxide abundance, in thermo-

E00B40

chemical equilibrium with the surface, of 18 ppmv (1/100 of today’s value for Venus). [87] These initial conditions yield a starting surface temperature of 780 K, but subsequent evolution of surface temperatures are fairly independent of the starting conditions. The surface temperature initially declines from 780 K to 750 K owing to the formation of thick clouds and increased albedo. However, after about 150 million years, the thinning clouds lower the planetary albedo, increasing surface temperatures to between 800 K and 850 K for about 400 million years. A further drop in surface temperatures after 600 million years is due to the loss of clouds and their infrared scattering, which helps to maintain warmer surface temperatures. [88] Several groups have investigated whether surface temperature changes of this magnitude could have significant effects on surface geology and geophysics. Diverse puzzling aspects of the surface geology of Venus can potentially be explained by such extreme climate changes. Climate-driven variations in thermal stress are consistent with the formation of wrinkle ridges on the most widespread volcanic plains units due to the propagation of a climateinduced thermal pulse that deformed the surface within 100 My of their formation [Solomon et al., 1999]. Others include (1) the origin of ‘‘ribbon terrain’’ in ancient crustal plateaus, which may indicate large changes in the depth of the brittle-ductile transition [Phillips and Hansen, 1998; Brown and Grimm, 1999], (2) the origin of extensive canali thousands of kilometers in length which could have been carved by carbonatite flows that would be stable in a somewhat warmer climate regime [Kargel et al., 1994], (3) widespread and apparently coherent formation of polygonal and gridded terrains [Anderson and Smrekar, 1999; Smrekar et al., 2002; Moreels and Smrekar, 2003], and (4) steep-sided dome morphology which could be consistent with rhyolitic composition (a volcanic rock resembling granite) only if surface temperatures were high enough to inhibit crust formation during extrusion [Stofan et al., 2005]. [89] Taken collectively these independent suggestions of possible climate influence on geology provide strong motivation for further investigation of the links between outgassing, climate, and surface records of climate change. [90] Such coupling between climate change and thermal stress provides an avenue for testing models of outgassing history against the geological record of deformation. In this way the geologic history of the planet becomes an additional tool for exploring how the physics of planetary-wide feedbacks have driven Venus’ climate evolution, perhaps occasionally driven its tectonic evolution, and led to the present atmospheric state. [91] The range of temperatures found by Bullock and Grinspoon’s [2001] experiment are indicated on Figure 9, which also shows two simple radiative-convective models from Taylor [2006], and a measured temperature profile for the middle atmosphere of Venus from the Magellan radio occultation experiment [Jenkins et al., 1994]. The simple models have a stratosphere in radiative equilibrium with the Sun, overlying a deep atmosphere in which the profile follows a dry adiabat. The solid line is such a model calculated assuming present-day conditions; the dashed line is an imaginary scenario in which the surface pressure on

16 of 22

E00B40

TAYLOR AND GRINSPOON: CLIMATE EVOLUTION ON VENUS

E00B40

Figure 9. Two simple models for the temperature profile in Venus’ atmosphere: one calculated for the present-day (solid line) and the other (dashed line) for a hypothetical scenario in which the surface pressure and the CO2 mixing ratio both relax to Earth-like values. The heavy double-headed arrow at the surface shows the range of temperatures that appeared in model experiments by Bullock and Grinspoon [2001] in which enhanced amounts of volcanic gases were injected into the atmosphere. A middleatmosphere temperature profile from the Magellan radio occultation experiment [Jenkins et al., 1994] is shown for comparison with all of these. Venus falls to 1 bar and the planetary albedo falls to 0.52, that is, to a less cloud-reflective state, perhaps as the result of continued exospheric loss and chemical erosion of the atmosphere following a cessation of the volcanic source at some distant point in the future. This is roughly the scenario imagined by Arrhenius a century ago, and gives rise to a surface temperature of 320 K, precisely his assessment (What Arrhenius actually wrote was, ‘‘assuming the sun constant to two calories per cubic centimeter (0.061 cu. in.) per minute.’’ Verifying that this corresponds to an albedo of 0.52 is left as an exercise for the reader). [92] As a final example of the use of climate models, and the motivation for improving them to obtain accurate results, we show in Table 2 the effect on the surface temperature of removing various key atmospheric constituents from the model of Bullock and Grinspoon [2001]. The leading role of carbon dioxide in maintaining the greenhouse is not surprising, although its predominance perhaps is, while the cloud number is very model-dependent and may be variable, as no doubt are the smaller contributions from trace species. The larger effect of water vapor relative to the others comes from its very rich infrared spectrum (which makes it the principal greenhouse gas on Earth), which compensates for its low mixing ratio on Venus. In contrast, notwithstanding its key role in cloud formation, sulfur dioxide makes a relatively small spectral contribution on both planets, even on Venus where its abundance is relatively high, because it has few infrared bands. Bullock and Grinspoon [1996] found that if all of the CO2 were removed from the atmosphere, leaving only N2, water vapor and other trace constituents, the surface temperature would be roughly 400 K.

5. Future Observations and Evolutionary Analyses [93] The discussions in the previous sections confirm the expectation that understanding the climate system on Venus

is a vast and complex undertaking that will proceed gradually in tandem with the similar undertaking for the Earth, but with an especial need for much more data. The data will come mainly from planetary missions, including the current Venus Express, the Japanese Climate Orbiter now under construction, and those future missions still under discussion that eventually fly. The latter must in time include landed and buoyant probes of long duration, which can make very precise measurements of atmospheric composition, surface and cloud properties. Further off, but essential, are the missions that will sample the geochemistry of the surface and probe the deep interior using seismic and other measurements. It is profoundly to be hoped that progress toward these goals will be faster during the next few decades than it has been in the last two, when Venus Express ended a long period of benign neglect of a nearby, Earth-like world that is uniquely instructive for so many of our crucial environmental issues. 5.1. Venus Express Extended Mission and Venus Climate Orbiter [94] A strong case has been put forward for extending the Venus Express mission beyond its original span of 500 days, about 2 Venus years, to 1000 and then to about 2000 days, Table 2. Evolution of Surface Temperature With Composition in the Bullock and Grinspoon [2001] Modela Species Removed

Change in Surface Temperature (K)

HCl CO SO2 Clouds H2 O OCS CO2

1.5 3.3 2.5 142.8 68.8 12 422.7

a The temperature decreases listed assume that the indicated species is removed entirely from the atmospheric greenhouse calculation, while the albedo of the planet remains unchanged (even when clouds are removed).

17 of 22

E00B40

TAYLOR AND GRINSPOON: CLIMATE EVOLUTION ON VENUS

ending in December 2012 [Svedhem et al., 2009]. It is worth considering how this long extension will help to meet the original goals of the mission in terms of a new picture of the climate on Venus. [95] The goal of Venus Express as originally stated was the acquisition and dissemination of new knowledge about the Venusian climate and its place in our understanding of the climate regimes on all of the terrestrial planets (including Earth, Mars, and, for some purposes, Titan), specifically in the following key areas and objectives: (1) Detection of volcanic activity and better quantification of the volcanic gas inventory in the atmosphere; (2) improved knowledge of vertical cloud structure, microphysics, and variability; (3) updated inventories of minor constituent abundances; (4) atmospheric temperature fields above, in, and below the clouds; (5) new observational constraints from mapping on the general circulation and dynamical phenomena like the polar vortices and deep atmosphere ‘‘weather’’; (6) improved estimates of atmospheric loss rates for O, C, H, and D; (7) interaction with the solar wind and escape processes; and (8) detection of any interannual and interhemispheric asymmetries and trends in all of the above. [96] Another 4 years of operation should lead to further advances in all of these areas, particularly where timedependent phenomena are involved. Meteorological activity takes place on all time scales, and studies of the long-term behavior of the polar vortices and of the global circulation, in particular, require long-term observations. Wind information over more latitudes, time of day, longitudes, and times are needed to ascertain the time and spatial variability and periodicities in the dynamics. Separation of standing versus traveling wave phenomena, the study of the stability of the superrotation, and fluctuations in wind profiles will need long temporal sequences to establish what periodicities are present. Rare dynamical events such as volcanic eruptions and bright cloud surges like that seen in January 2007 [Titov et al., 2008], but observed only once, require dedicated periods of continuous mapping to either detect volcanism or give a reliable upper limit for the current volcanic activity of Venus. [97] Completely new data will result from upper atmosphere in situ measurements and joint operations with the Japanese Venus Climate Orbiter [Nakamura et al., 2007]. The VCO, also known in Japan as Planet-C, is due to launch in the first half of 2010 and to arrive in December of the same year, with a payload of atmospheric sounding instruments. These are designed to obtain an understanding of the atmospheric circulation and meteorology on Venus, in particular the driving force behind the zonal superrotation. VCO has an equatorial orbit while Venus Express (VEX) has a polar orbit, and these will be synchronized for studies of the dynamics of the cloud motions. VCO can obtain much longer uninterrupted observations of a particular area than VEX, while VIRTIS and SPICAV have spectral capabilities that VCO lacks, so global contextual information from VCO, coupled with local spectral information and vertical profiles of temperature and density from stellar and radio occultations by VEX will enable improved studies of the motions and evolution of structure in the cloud features, and consequent advances in understanding of cloud formation and destruction mechanisms, including radiative dynamic feedbacks in the middle and lower clouds.

E00B40

[98] The controversial question of lightning on Venus has implications for the climate through its effect on atmospheric chemistry and composition. Current VEX magnetometer observations include signals attributed to lightning, possibly cloud-to-cloud rather than cloud-to-ground, but the phenomenon has not yet been detected optically. VCO has a high-speed lightning observation camera and simultaneous observations of optical flashes by VCO and whistlers by VEX would present an irrefutable detection of lightning, as well as further clues as to the source regions and mechanisms. [99] The other climate-related investigations by Venus Express that are enabled by an extended mission include observations of the plasma environment and atmospheresolar wind interaction as the Sun moves toward solar maximum conditions. The upper atmosphere dynamical regime has been monitored in a period of very low solar activity; extension of the observations will allow the intensity and morphology of the O2 and OH airglow features, in particular, to be correlated with solar activity [cf. Stewart et al., 1980]. [100] There is also the opportunity, considered too risky for the main mission, to reduce the pericenter altitude to dip into the upper atmosphere. This will extend the measurements of local magnetic fields and plasma parameters to relatively low altitudes and high densities in the region where the atmosphere is impacted by the solar wind. Atmospheric drag measurements from orbit perturbations and the onboard accelerometer will provide unique information on density and temperature in the range 150– 200 km, which is not accessible by other means. [101] Venus Express and Venus Climate Orbiter will not address, let alone resolve, every one of the key questions about Venus that have accumulated as a result of exploration by the Venera, VEGA, Pioneer and Magellan missions. The knowledge gaps that will remain, that can be predicted in advance, are mostly in the area of atmospheric evolution (addressable by accurate measurements of noble gas isotopic ratios, for instance) and composition (a full understanding of surface-atmosphere interactions, cloud composition and chemistry will require in situ trace constituent abundance measurements, especially at the surface and in the clouds). Other areas that will be largely untouched by Venus Express are surface geology, geochemistry, and interior structure, and surface-atmosphere and surfaceinterior interactions. For these investigations, high-pressure balloons, landed missions and sample return may be the optimum way forward. 5.2. Entry Probes and Floating Stations [102] ESA recently turned down a proposal for a Venus Entry Probe mission [Chassefiere et al., 2007] in favor of in-depth studies of a new Outer Planets mission to follow Galileo and Cassini. The best prospect now for obtaining essential data on, for instance, the cloud chemistry and the isotopic composition of noble gases, rests with NASA, where the most recent decadal survey called for a Venus In Situ Explorer mission, which will seek to (1) understand the physics and chemistry of Venus’ atmosphere through measurement of its composition, especially the abundances of its trace gases, sulfur, light stable isotopes, and noble gas isotopes, below the clouds and all the way down to the

18 of 22

E00B40

TAYLOR AND GRINSPOON: CLIMATE EVOLUTION ON VENUS

surface with more detail than is possible using remote sensing; (2) constrain the coupling of thermochemical, photochemical, and dynamical processes in Venus’ atmosphere and between the surface and atmosphere to understand radiative balance, climate, and dynamics, and to characterize the chemical cycles involving clouds, surface and atmospheric gases; (3) understand the physics and chemistry of Venus’ crust through analysis of near-IR descent images from below the clouds to the surface and through measurements of elemental abundances and mineralogy from a surface sample; (4) understand the properties of Venus’ atmosphere down to the surface through meteorological measurements and improve our understanding of Venus’ zonal cloud level winds through temporal measurements over several Earth days; (5) understand the weathering environment of the crust of Venus in the context of the dynamics of the atmosphere of Venus and the composition and texture of its surface materials; and (6) map the mineralogy and chemical composition of Venus’ surface on the planetary scale for evidence of past hydrological cycles, oceans, and life and constraints on the evolution of Venus’ atmosphere. [103] Much of what we now know about the history of Earth’s atmosphere has been inferred from measurements of abundances and isotopic ratios for the noble gases. Not only are these chemically inert, which greatly simplifies the range of potential sources and sinks for any given isotope, but also some are produced at well-defined rates by the radioactive decay of parent molecules with a range of halflives that spans most of the history of the planet. The wide range of atomic masses (from 2He to 130Xe) among the commonest of these gases, and the convenient mass scale (for instance, 20Ne/21Ne/22Ne) across measurable abundances of the same element, make them a convenient yardstick for determining mantle degassing and atmospheric loss rates over time. [104] It follows that measurements of noble gases in the atmosphere of Venus are a powerful tool for tracing Venus’ evolution in the same way. Direct comparisons of the relative abundances of neon, krypton, xenon, argon and helium and their isotopes between the two planets highlight differences in their histories, and tell us something about the nature and timing of the events that produced them. For instance, the Pioneer Venus probes discovered that Venus is rich in neon and nonradiogenic argon compared to Earth and Mars, prompting speculation that they may have been brought in during the collision with Venus of a very large comet from the cold outer reaches of the solar system, where substantial quantities of these species can be trapped in water ice as clathrates. More isotopic ratio measurements, especially if they are more accurate than the 10% or so achieved by Pioneer Venus, will refine this theory and distinguish it from rival explanations. [105] As another example, Watson et al. [2007] argue that argon compatibility with rock forming minerals has interesting implications for interpretation of argon ratios on Venus versus Earth. What has long been interpreted as implying a difference in total cumulative outgassing may actually say more about the history of the crust and weathering. Many other instances can be cited where accurate measurements of trace gas abundances will improve our understanding of Venus’ atmosphere and

E00B40

climate [see, e.g., Baines et al., 2007]. They will also make it much clearer which events are common between Venus and Earth and which may be unique, like the massive cometary impact described above, to one or the other. However, the data required can only be obtained in situ, using Venus entry probes, buoyant stations, and landers, and not by orbiters like the current generation of missions from Europe and Japan. 5.3. Surface Missions and Sample Return [106] The hostile conditions on the surface of Venus, particularly the high temperature, have in the past limited the lifetime of landed missions to about 1 h on the surface. This is the time it takes for a well-insulated payload to rise in temperature to the point where electronics and other systems fail. Advanced technology that can overcome this problem and permit a long-lived lander on Venus is under development, but still some considerable distance away in practical terms. Ten years ago, when the European Space Agency decided to study a mission to land on Venus, drill a core sample, and return it to Earth, the conclusion was that the only realistic option was to carry out the surface phase quickly enough that conventional electronics, packed with thermally insulating phase change material, could be used [Coradini et al., 1998]. Numerous NASA studies over a 40-year period reached similar conclusions, and the Venus In Situ Explorer, currently the most likely mission to fly to Venus after the Japanese Climate Orbiter, follows the same path to obtain surface samples which are carried to a platform floating at an altitude where a more comfortable temperature for analysis can be found. NASA has recently commissioned a Science and Technology Definition Team to study a possible Flagship Mission to Venus to be launched in the 2025 time frame. Such an ambitious mission would likely include a large orbiter equipped with a radar interferometer, and multiple landers and floating stations. [107] Taking the ESA sample return study as an example, two launches using the most powerful version of the Ariane vehicle would be required, one to carry an orbiter and Earth return vehicle, and the other to insert a lander directly into the Venusian atmosphere. The latter, with a landed mass of 4 tons, would acquire three 100-g samples from on and below the surface, and a bottle of near-surface atmosphere, before ascending by balloon to the 1 bar level near the cloud tops. A double-balloon arrangement might be employed, using a metal bellows filled with helium to traverse the lowest 12 km and then a more conventional Teflon-coated Kapton balloon for the rest of the ascent. From the float level around 55 to 60 km above the surface, a rocket would carry the samples into orbit where they would rendezvous with the return vehicle and be transferred for the flight back to Earth and a landing by parachute. The rendezvous, in particular, is a slow affair and the total elapsed time for the mission, from takeoff to retrieval of the samples, was estimated to be 6 years. [108] This brief summary is enough to illustrate the complexities of a sample return mission and to make clear why there are no plans to implement one soon. Still, it is essential that the technological challenges are systematically studied so that eventually samples can be analyzed in terrestrial laboratories and key issues related to the climate and its evolution answered. The variety of analyses that can

19 of 22

E00B40

TAYLOR AND GRINSPOON: CLIMATE EVOLUTION ON VENUS

be conducted in a laboratory, as opposed to on Venus, and their greater accuracy and precision, will make it possible to determine the ages of Venusian surface units, a vital area of information. The analysis of atmospheric samples would provide much improved data on rare gas isotopic ratios, and with several samples spaced vertically in altitude by about 10 km we could address the key questions of water vapor abundance, cloud chemistry, and the role of currently active volcanism. In situ correlative gas chromatography and mass spectroscopy will be needed for the more reactive species that are expected to be present, especially near the surface. [109] Returning a core of the surface to Earth would enable a determination of the amount of weathering that occurs and also would allow analysis to be done on Venusian rock samples from below the surface that were unaffected by the atmosphere. This would provide the composition and structure of the near-surface material and help us to understand the differences in bulk density, atmospheric constituents and absolute abundances and water contents of the terrestrial planets. The importance to the climate history of Venus of understanding the original abundance of water has already been discussed, and is also relevant to the even more profound question of whether life developed on Venus under more benign conditions in the past. [110] Acknowledgments. F. W. T. acknowledges support from the UK Science, Technology, and Facilities Research Council. D. H. G. acknowledges support from the NASA Venus Express Interdisciplinary Scientist Support Program.

References Addington, E. A. (2001), A stratigraphic study of small volcano clusters on Venus, Icarus, 149, 16 – 36, doi:10.1006/icar.2000.6529. Alemi, A., and D. Stevenson (2006), Why Venus has no Moon, Bull. Am. Astron. Soc., 38, 491. Anderson, F. S., and S. E. Smrekar (1999), Tectonic effects of climate change on Venus, J. Geophys. Res., 104, 30,743 – 30,756, doi:10.1029/ 1999JE001082. Arrhenius, S. (1918), The Destinies of the Stars, Putman, New York. Baines, K. H., S. K. Atreya, R. W. Carlson, D. Crisp, D. Grinspoon, C. T. Russell, G. Schubert, and K. Zahnle (2007), Experiencing Venus: Clues to the origin, evolution, and chemistry of terrestrial planets via in situ exploration of our sister world, in Exploring Venus as a Terrestrial Planet, Geophys. Monogr. Ser., vol. 176, edited by L. W. Esposito, E. R. Stofan, and T. E. Cravens, pp. 171 – 189, AGU, Washington, D. C. Barabash, S., et al. (2007), The loss of ions from Venus through the plasma wake, Nature, 450, 650 – 653, doi:10.1038/nature06434. Barath, F. T., A. H. Barrett, J. Copeland, D. E. Jones, and A. E. Lilley (1964), Mariner 2 microwave radiometer experiment and results, Astron. J., 69, 49 – 58, doi:10.1086/109227. Barrett, A. H., J. Copeland, D. E. Jones, and A. E. Lilley (1961), Objectives of the Mariner Venus microwave radiometer experiment, Tech. Rep. 32 – 156, Jet Propul. Lab., Pasadena, Calif. Basilevsky, A. T., and J. W. Head (1996), Evidence for rapid and widespread emplacement of volcanic plains on Venus: Stratigraphic studies in the Baltis Vallis region, Geophys. Res. Lett., 23, 1497 – 1500, doi:10.1029/96GL00975. Basilevsky, A. T., and J. W. Head (1998), The geologic history of Venus: A stratigraphic view, J. Geophys. Res., 103, 8531 – 8544, doi:10.1029/ 98JE00487. Basilevsky, A. T., and J. W. Head (2000), Geologic units on Venus: Evidence for their global correlation, Planet. Space Sci., 48, 75 – 111, doi:10.1016/S0032-0633(99)00083-5. Basilevsky, A. T., and J. W. Head (2002), Venus: Timing and rates of geologic activity, Geology, 30, 1015 – 1018, doi:10.1130/00917613(2002)030<1015:VTAROG>2.0.CO;2. Basilevsky, A. T., and J. W. Head (2003), The surface of Venus, Rep. Prog. Phys., 66, 1699 – 1734, doi:10.1088/0034-4885/66/10/R04. Basilevsky, A. T., and J. W. Head (2006), Impact craters on regional plains on Venus: Age relations with wrinkle ridges and implications for the

E00B40

geological evolution of Venus, J. Geophys. Res., 111, E03006, doi:10.1029/2005JE002473. Belton, M. J. S., G. R. Smith, G. Schubert, and A. D. Del Genio (1976), Cloud patterns, waves and convection in the Venus atmosphere, J. Atmos. Sci., 33, 1394 – 1417, doi:10.1175/1520-0469(1976)033<1394:CPWACI> 2.0.CO;2. Belyaev, D., O. Korablev, A. Fedorova, J.-L. Bertaux, A.-C. Vandaele, F. Montmessin, A. Mahieux, V. Wilquet, and R. Drummond (2008), First observations of SO2 above Venus’ clouds by means of Solar Occultation in the infrared, J. Geophys. Res., 113, E00B25, doi:10.1029/ 2008JE003143. Bertaux, J.-L., and F. Montmessin (2001), Isotopic fractionation through water vapor condensation: The deuteropause, a cold trap for deuterium in the atmosphere of Mars, J. Geophys. Res., 106, 32,879 – 32,884, doi:10.1029/2000JE001358. Bertaux, J.-L., T. Widemann, A. Hauchecorne, V. I. Moroz, and A. P. Ekonomov (1996), Vega-1 and Vega-2 entry probes: An investigation of local UV absorption (220 – 400 nm) in the atmosphere of Venus (SO2 aerosols, cloud structure), J. Geophys. Res., 101, 12,709 – 12,745, doi:10.1029/96JE00466. Bertaux, J.-L., et al. (2007), A warm layer in Venus’ cryosphere and highaltitude measurements of HF, HCl, H2O and HDO, Nature, 450, 646 – 649, doi:10.1038/nature05974. Bezard, B., et al. (2009), Water vapor abundance near the surface of Venus from Venus Express/VIRTIS observations, J. Geophys. Res., 114, E00B39, doi:10.1029/2008JE003251. Bilotti, F., and J. Suppe (1999), The global distribution of wrinkle ridges on Venus, Icarus, 139, 137, doi:10.1006/icar.1999.6092. Brown, C. D., and R. E. Grimm (1999), Recent tectonic and lithospheric thermal evolution of Venus, Icarus, 139, 40 – 48, doi:10.1006/ icar.1999.6083. Bullock, M. A., and D. H. Grinspoon (1996), The stability of climate on Venus, J. Geophys. Res., 101, 7521 – 7530, doi:10.1029/95JE03862. Bullock, M. A., and D. H. Grinspoon (2001), The recent evolution of climate on Venus, Icarus, 150, 19 – 37, doi:10.1006/icar.2000.6570. Bullock, M. A., D. H. Grinspoon, and J. W. Head (1993), Venus resurfacing rates: Constraints provided by 3-D Monte Carlo simulations, Geophys. Res. Lett., 20, 2147 – 2150, doi:10.1029/93GL02505. Campbell, B. A. (1999), Surface formation rates and impact crater densities on Venus, J. Geophys. Res., 104, 21,951 – 21,955, doi:10.1029/ 1998JE000607. Chamberlain, J. W., and D. M. Hunten (1987), Theory of Planetary Atmospheres: An Introduction to Their Physics and Chemistry, Academic, Orlando, Fla. Chassefiere, E., et al. (2007), The European Venus Explorer (EVE) mission proposal, Eur. Planet. Sci. Congr. Abstr., 2, EPSC2007-A-00548. Collard, A. D., F. W. Taylor, S. B. Calcutt, R. W. Carlson, L. Kamp, K. Baines, T. Encrenaz, P. Drossart, E. Lellouch, and B. Be´zard (1993), Latitudinal distribution of carbon monoxide in the deep atmosphere of Venus, Planet. Space Sci., 41, 487 – 494. Coradini, M., G. Scoon, and J.-P. Lebreton (1998), Venus sample return assessment study report, ESA SCI(98)3, Eur. Space Agency, Noordwijk, Netherlands. Crisp, D., and D. V. Titov (1997), The thermal balance of the lower atmosphere of Venus, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 353 – 384, Univ. of Ariz. Press, Tucson. Davies, J. H. (2008), Did a mega-collision dry Venus’ interior?, Earth Planet. Sci. Lett., 268, 376 – 383, doi:10.1016/j.epsl.2008.01.031. de Bergh, C., B. Be´zard, T. Owen, D. Crisp, J. P. Maillard, and B. L. Lutz (1991), Deuterium on Venus: Observations, Earth Sci., 251, 547. Donahue, T. M. (1999), New analysis of hydrogen and deuterium escape from Venus, Icarus, 141, 226 – 235, doi:10.1006/icar.1999.6186. Donahue, T. M., and J. B. Pollack (1983), Origin and evolution of the atmosphere of Venus, in Venus, edited by D. M. Hunten et al., pp. 1003 – 1036, Univ. of Ariz. Press, Tucson. Donahue, T. M., J. H. Hoffman, R. R. Hodges Jr., and A. J. Watson (1982), Venus was wet: A measurement of the ratio of deuterium to hydrogen, Science, 216, 630 – 633, doi:10.1126/science.216.4546.630. Donahue, T. M., D. H. Grinspoon, R. E. Hartle, and R. R. Hodges Jr. (1997), Ion neutral escape: Evolution of water, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 385 – 414, Univ. of Ariz. Press, Tucson. Elson, L. S. (1982), Wave instability in the polar region of Venus, J. Atmos. Sci., 39, 2356 – 2362, doi:10.1175/1520-0469(1982)039<2356:WIITPR>2.0. CO;2. Esposito, L. W. (1984), Sulfur dioxide: Episodic injection shows evidence for active Venus volcanism, Science, 223, 1072 – 1074, doi:10.1126/ science.223.4640.1072.

20 of 22

E00B40

TAYLOR AND GRINSPOON: CLIMATE EVOLUTION ON VENUS

Esposito, L. W., J.-L. Bertaux, V. Krasnopolsky, V. I. Moroz, and L. V. Zasova (1997), Chemistry of lower atmosphere and clouds, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 415 – 458, Univ. of Ariz. Press, Tucson. Fegley, B., and R. G. Prinn (1989), Estimation of the rate of volcanism on Venus from reaction rate measurements, Nature, 337, 55 – 59, doi:10.1038/337055a0. Fegley, B., and A. H. Treiman (1992), Chemistry of atmosphere-surface interactions on Venus and Mars, in Venus and Mars: Atmospheres, Ionospheres, and Solar Wind Interactions, Geophys. Monogr. Ser., vol. 66, edited by J. G. Luhmann, M. Tatrallyay, and R. O. Pepin, pp. 7 – 71, AGU, Washington, D. C. Fegley, B., K. Lodders, A. H. Treiman, and G. Klingelho¨fer (1995), The rate of pyrite decomposition on the surface of Venus, Icarus, 115, 159 – 180. Fegley, B., G. Klingelhofer, K. Lodders, and T. Widemann (1997), Geochemistry of surface-atmosphere interactions on Venus, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 591 – 636, Univ. of Ariz. Press, Tucson. Fels, S. B., J. T. Schofield, and D. Crisp (1984), Observations and theory of the solar semidiurnal tide in the mesosphere of Venus, Nature, 312, 431 – 434, doi:10.1038/312431a0. Frankel, C. (1996), Volcanoes of the Solar System, Cambridge Univ. Press, Cambridge, U. K. Gierasch, P. J., et al. (1997), The general circulation of the Venus atmosphere: An assessment, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 459 – 500, Univ. of Ariz. Press, Tucson. Grinspoon, D. H. (1987), Was Venus wet? Deuterium reconsidered, Science, 238, 1702 – 1704, doi:10.1126/science.238.4834.1702. Grinspoon, D. H. (1993), Implications of the high D/H ratio for the sources of water in Venus’ atmosphere, Nature, 363, 428 – 431, doi:10.1038/ 363428a0. Grinspoon, D. H. (1997), Venus Revealed: A New Look Below the Clouds of Our Mysterious Twin Planet, Addison-Wesley, New York. Guest, J. E., and E. R. Stofan (1999), A new view of the stratigraphic history of Venus, Icarus, 139, 55 – 66, doi:10.1006/icar.1999.6091. Hashimoto, G. L., and Y. Abe (2005), Climate control on Venus: Comparison of the carbonate and pyrite models, Planet. Space Sci., 53, 839 – 848, doi:10.1016/j.pss.2005.01.005. Hashimoto, G. L., and T. Imamura (2001), Elucidating the rate of volcanism on Venus: Detection of lava eruptions using near-infrared observations, Icarus, 154, 239 – 243, doi:10.1006/icar.2001.6713. Hashimoto, G. L., M. Roos-Serote, S. Sugita, M. S. Gilmore, L. W. Kamp, R. W. Carlson, and K. H. Baines (2008), Felsic highland crust on Venus suggested by Galileo Near-Infrared Mapping Spectrometer data, J. Geophys. Res., 113, E00B24, doi:10.1029/2008JE003134. Ha¨usler, B., et al. (2006), Radio science investigations by VeRa onboard the Venus Express spacecraft, Planet. Space Sci., 54, 1315 – 1335, doi:10.1016/j.pss.2006.04.032. Head, J., L. Crumpler, J. Aubele, J. Guest, and R. Saunders (1992), Venus volcanism: Classification of volcanic features and structures, associations, and global distribution from Magellan data, J. Geophys. Res., 97, 13,153 – 13,197. Helbert, J., N. Mu¨ller, P. Kostama, L. Marinangeli, G. Piccioni, and P. Drossart (2008), Surface brightness variations seen by VIRTIS on Venus Express and implications for the evolution of the Lada Terra region, Venus, Geophys. Res. Lett., 35, L11201, doi:10.1029/2008GL033609. Herrick, R. R., and V. L. Sharpton (2000), Implications from stereo-derived topography of Venusian impact craters, J. Geophys. Res., 105, 20,245 – 20,262, doi:10.1029/1999JE001225. Ignatiev, N. I., V. I. Moroz, L. V. Zasova, and I. V. Khatuntsev (1999), Water vapor in the middle atmosphere of Venus: An improved treatment of the Venera 15 IR spectra, Planet. Space Sci., 47, 1061 – 1075, doi:10.1016/S0032-0633(99)00030-6. Ivanov, M. A., and A. T. Basilevsky (1993), Density and morphology of impact craters on tesserae terrain, Venus, Geophys. Res. Lett., 20, 2579 – 2582, doi:10.1029/93GL02692. Izenberg, N. R., R. E. Arvidson, and R. J. Phillips (1994), Impact crater degradation on Venusian plains, Geophys. Res. Lett., 21, 289 – 292, doi:10.1029/94GL00080. Jenkins, J. M., P. G. Steffes, D. P. Hinson, J. D. Twicken, and G. L. Tyler (1994), Radio occultation studies of the Venus atmosphere with the Magellan spacecraft: 2. Results from the October 1991 experiments, Icarus, 110, 79 – 94, doi:10.1006/icar.1994.1108. Kargel, J. S., R. L. Kirk, B. Fegley, and A. H. Treiman (1994), Carbonatesulfate volcanism on Venus?, Icarus, 112, 219 – 252, doi:10.1006/ icar.1994.1179. Kasting, J. F. (1988), Runaway and moist greenhouse atmospheres and the evolution of Earth and Venus, Icarus, 74, 472 – 494, doi:10.1016/00191035(88)90116-9.

E00B40

Kasting, J. F., J. B. Pollack, and T. P. Ackerman (1984), Response of Earth’s atmosphere to increases in solar flux and implications for loss of water from Venus, Icarus, 57, 335 – 355, doi:10.1016/00191035(84)90122-2. Keating, G. M., F. W. Taylor, J. Y. Nicholson, and E. W. Hinson (1979), Short-term cyclic variations and diurnal variations of the Venus upper atmosphere, Science, 205, 62 – 64, doi:10.1126/science.205.4401.62. Kliore, A. J., V. I. Moroz, and G. M. Keating (1986), The Venus International Reference Atmosphere, Pergamon, Oxford, U. K. Koukouli, M., P. G. J. Irwin, and F. W. Taylor (2005), Water vapor abundance in Venus’ middle atmosphere from Pioneer Venus OIR and Venera 15 FTS measurements, Icarus, 173, 84 – 99, doi:10.1016/j.icarus. 2004.08.023. Krasnopolsky, V. A., and J. B. Pollack (1994), H2O – H2SO4 system in Venus’ clouds and OCS, CO, and H2SO4 profiles in Venus’ troposphere, Icarus, 109, 58 – 78, doi:10.1006/icar.1994.1077. Kreslavsky, M. A., and J. W. Head III (1999), Morphometry of small shield volcanoes on Venus: Implications for the thickness of regional plains, J. Geophys. Res., 104, 18,925 – 18,932, doi:10.1029/1999JE001042. Lebonnois, S., F. Hourdin, V. Eymet, R. Fournier, and J.-L. Dufresne (2005), A new Venus general circulation model, in the context of the Venus Express mission, Bull. Am. Astron. Soc., 37, 742. Lee, C., S. R. Lewis, and P. L. Read (2007), Superrotation in a Venus general circulation model, J. Geophys. Res., 112, E04S11, doi:10.1029/ 2006JE002874. Liang, M. C., and Y. L. Yung (2009), Modeling the distribution of H2O and HDO in the upper atmosphere of Venus, J. Geophys. Res., 114, E00B28, doi:10.1029/2008JE003095. Luhmann, J. G., and C. T. Russell (1997), Venus: Magnetic field and magnetosphere, in Encyclopedia of Planetary Sciences, edited by J. H. Shirley and R. W. Fainbridge, pp. 905 – 907, Chapman and Hall, New York. Marcq, E., B. Be´zard, P. Drossart, G. Piccioni, J. M. Reess, and F. Henry (2008), A latitudinal survey of CO, OCS, H2O, and SO2 in the lower atmosphere of Venus: Spectroscopic studies using VIRTIS-H, J. Geophys. Res., 113, E00B07, doi:10.1029/2008JE003074. Markiewicz, W. J., et al. (2007), Morphology and dynamics of the upper cloud layer of Venus, Nature, 450, 633 – 636, doi:10.1038/nature06320. McElroy, M. B., M. J. Prather, and J. M. Rodriguez (1982), Escape of hydrogen from Venus, Science, 215, 1614 – 1615, doi:10.1126/ science.215.4540.1614. McGill, G. E. (1993), Wrinkle ridges, stress domains, and kinematics of Venusian plains, Geophys. Res. Lett., 20, 2407 – 2410, doi:10.1029/ 93GL02635. McKinnon, W. B., K. J. Zahnle, B. A. Ivanov, and H. J. Melosh (1997), Cratering on Venus: Models and observations, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 969 – 1014, Univ. of Ariz. Press, Tucson. Morbidelli, A., J. Chambers, J. I. Lunine, J. M. Petit, F. Robert, G. B. Valsecchi, and K. E. Cyr (2000), Source regions and timescales for the delivery of water to the Earth, Meteorit. Planet. Sci., 351, 1309 – 1320. Moreels, P., and S. E. Smrekar (2003), Watershed identification of polygonal patterns in noisy SAR images, IEEE Trans. Image Process., 12, 740 – 750, doi:10.1109/TIP.2003.814254. Nakamura, M., et al. (2007), Planet-C: Venus climate orbiter mission of Japan, Planet. Space Sci., 55, 1831 – 1842, doi:10.1016/ j.pss.2007.01.009. Namiki, N., and S. C. Solomon (1994), Impact crater densities on volcanoes and coronae on Venus: Implications for volcanic resurfacing, Science, 265, 929 – 933, doi:10.1126/science.265.5174.929. Palmer, K. F., and D. Williams (1975), Optical constants of sulfuric acid: Application to the clouds of Venus?, Appl. Opt., 14, 208 – 219. Pepin, R. O. (2006), Atmospheres on the terrestrial planets: Clues to origin and evolution, Earth Planet. Sci. Lett., 252, 1 – 14, doi:10.1016/ j.epsl.2006.09.014. Phillips, R. J., and V. L. Hansen (1998), Geological evolution of Venus: Rises, plains, plumes, and plateaus, Science, 279, 1492 – 1497, doi:10.1126/science.279.5356.1492. Phillips, R. J., and N. R. Izenberg (1995), Ejecta correlations with spatial crater density and Venus resurfacing history, Geophys. Res. Lett., 22, 1517 – 1520, doi:10.1029/95GL01412. Phillips, R. J., R. Raubertas, R. Arvidson, I. Sarkar, R. Herrick, N. Izenberg, and R. Grimm (1992), Impact craters and Venus resurfacing history, J. Geophys. Res., 97, 15,923 – 15,948. Phillips, R. J., M. A. Bullock, and S. A. Hauck (2001), Climate and interior coupled evolution on Venus, Geophys. Res. Lett., 28, 1779 – 1782, doi:10.1029/2000GL011821. Piccioni, G., et al. (2007), South-polar features on Venus similar to those near the north pole, Nature, 450, 637 – 641, doi:10.1038/nature06209.

21 of 22

E00B40

TAYLOR AND GRINSPOON: CLIMATE EVOLUTION ON VENUS

Pollack, J. B., et al. (1993), Near-infrared light from Venus’ nightside: A spectroscopic analysis, Icarus, 103, 1 – 42, doi:10.1006/icar.1993.1055. Price, M. H., and J. Suppe (1994), Mean age of rifting and volcanism on Venus deduced from impact crater densities, Nature, 372, 756 – 759, doi:10.1038/372756a0. Price, M. H., G. Watson, J. Suppe, and C. Brankman (1996), Dating volcanism and rifting on Venus using impact crater densities, J. Geophys. Res., 101, 4657 – 4671, doi:10.1029/95JE03017. Pyle, D. M. (1995), Mass and energy budgets of explosive volcanic eruptions, J. Geophys. Res., 22, 563 – 566. Read, W., L. Froidevaux, and J. Waters (1993), Microwave limb sounder measurement of stratospheric SO2 from the Mt. Pinatubo volcano, Geophys. Res. Lett., 20, 1299 – 1302, doi:10.1029/93GL00831. Ringwood, A. E., and D. L. Anderson (1977), Earth and Venus: A comparative study, Icarus, 30, 243 – 253, doi:10.1016/0019-1035(77)90156-7. Rossow, W. B. (1985), Atmospheric circulation of Venus, Adv. Geophys., 28, 347 – 379, doi:10.1016/S0065-2687(08)60230-7. Russell, C. T. (1992), The Pioneer Venus mission, in Venus and Mars: Atmospheres, Ionospheres and Solar Wind Interactions, Geophys. Monogr. Ser., vol. 66, edited by J. G. Luhmann, M. Tatrallyay, and R. O. Pepin, pp. 225 – 236, AGU, Washington, D. C. Sanchez-Lavega, A., et al. (2008), Variable winds on Venus mapped in three dimensions, Geophys. Res. Lett., 35, L13204, doi:10.1029/ 2008GL033817. Schaber, G. G., R. G. Strom, H. J. Moore, L. A. Soderblom, R. L. Kirk, D. J. Chadwick, D. D. Dawson, L. R. Gaddis, J. M. Boyce, and J. Russell (1992), Geology and distribution of impact craters on Venus: What are they telling us?, J. Geophys. Res., 97, 13,257 – 13,302. Schofield, J. T., and D. J. Diner (1983), Rotation of Venus’s polar dipole, Nature, 305, 116 – 119, doi:10.1038/305116a0. Schofield, J. T., and F. W. Taylor (1982), Net global thermal emission from the Venus atmosphere, Icarus, 52, 245 – 262, doi:10.1016/00191035(82)90111-7. Schubert, G., V. S. Solomatov, P. J. Tackley, and D. L. Turcotte (1997), Mantle convection and the thermal evolution of Venus, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 1245 – 1287, Univ. of Ariz. Press, Tucson. Schubert, G., S. W. Bougher, C. C. Covey, A. D. Del Genio, A. S. Grossman, J. L. Hollingsworth, S. S. Limaye, and R. E. Young (2007), Venus atmosphere dynamics: A continuing enigma, in Exploring Venus as a Terrestrial Planet, Geophys. Monogr. Ser., vol. 176, edited by L. W. Esposito, E. R. Stofan, and T. E. Cravens, pp. 101 – 120, AGU, Washington, D. C. Sigurdsson, H., and P. Laj (1992), Atmospheric effects of volcanic eruptions, in Encyclopedia of Earth System Science, vol. 1, edited by W. A. Nierenberg, pp. 183 – 199, Academic, San Diego, Calif. Smrekar, S. E., P. Moreels, and B. J. Franklin (2002), Characterization and origin of polygonal fractures on Venus, J. Geophys. Res., 107(E11), 5098, doi:10.1029/2001JE001808. Solomon, S. C., M. A. Bullock, and D. H. Grinspoon (1999), Climate change as a regulator of tectonics on Venus, Science, 286, 87 – 90, doi:10.1126/science.286.5437.87. Stewart, A. I. F., J. C. Ge´rard, D. W. Rusch, and S. W. Bougher (1980), Morphology of the Venus ultraviolet night airglow, J. Geophys. Res., 85, 7861 – 7870, doi:10.1029/JA085iA13p07861. Stofan, E. R., A. W. Brian, and J. E. Guest (2005), Resurfacing styles and rates on Venus: Assessment of 18 Venusian quadrangles, Icarus, 173, 312 – 321, doi:10.1016/j.icarus.2004.08.004. Strom, R. G., G. G. Schaber, and D. D. Dawson (1994), The global resurfacing of Venus, J. Geophys. Res., 99, 10,899 – 10,926, doi:10.1029/94JE00388. Svedhem, H., D. V. Titov, F. W. Taylor, and O. Witasse (2007), Venus as a more Earth-like planet, Nature, 450, 629 – 633, doi:10.1038/nature06432. Svedhem, H., D. V. Titov, F. W. Taylor, and O. Witasse (2009), Venus Express mission, J. Geophys. Res., 114, E00B33, doi:10.1029/ 2008JE003290. Taylor, F. W. (1995), Carbon monoxide in the deep atmosphere of Venus, Adv. Space Res., 16(6), 81 – 88, doi:10.1016/0273-1177(95)00253-B. Taylor, F. W. (2006), Venus before Venus Express, Planet. Space Sci., 54, 1249 – 1262, doi:10.1016/j.pss.2006.04.031.

E00B40

Taylor, F. W., D. J. Diner, L. S. Elson, D. J. McCleese, J. V. Martonchik, P. E. Reichley, J. T. Schofield, S. P. Bradley, J. C. Gille, and M. T. Coffey (1979a), Temperature, cloud structure and dynamics of Venus middle atmosphere by infrared remote sensing from the Pioneer Orbiter, Science, 205, 65 – 67, doi:10.1126/science.205.4401.65. Taylor, F. W., D. J. McCleese, and D. J. Diner (1979b), Polar clearing in the Venus clouds observed from the Pioneer Venus Orbiter, Nature, 279, 613 – 614, doi:10.1038/279613a0. Taylor, F. W., et al. (1980), Structure and meteorology of the middle atmosphere of Venus: Infrared remote sounding from the Pioneer Orbiter, J. Geophys. Res., 85, 7963 – 8006, doi:10.1029/JA085iA13p07963. Taylor, F. W., et al. (1993), Remote sensing of atmospheric structure and composition by pressure modulator radiometry from space: The ISAMS experiment on UARS, J. Geophys. Res., 98, 10,799 – 10,814, doi:10.1029/92JD03029. Taylor, F. W., D. Crisp, and B. Be´zard (1997), Near-infrared sounding of the lower atmosphere of Venus, in Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 325 – 351, Univ. of Ariz. Press, Tucson. Taylor, F. W., H. Svedhem, and D. Titov (2007), Venus Express and terrestrial planet climatology, in Exploring Venus as a Terrestrial Planet, Geophys. Monogr. Ser., vol. 176, edited by L. W. Esposito, E. R. Stofan, and T. E. Cravens, pp. 157 – 170, AGU, Washington, D. C. Titov, D., M. A. Bullock, D. Crisp, N. O. Renno, F. W. Taylor, and L. V. Zasova (2007), Radiation in the atmosphere of Venus, in Exploring Venus as a Terrestrial Planet, Geophys. Monogr. Ser., vol. 176, edited by L. W. Esposito, E. R. Stofan, and T. E. Cravens, pp. 1121 – 1138, AGU, Washington, D. C. Titov, D. V., F. W. Taylor, H. Svedhem, N. I. Ignatiev, W. J. Markiewicz, G. Piccioni, and P. Drossart (2008), Atmospheric structure and dynamics as the cause of ultraviolet markings in the clouds of Venus, Nature, 456, 620 – 623, doi:10.1038/nature07466. Tomasko, M. G., et al. (1980), The thermal balance of Venus in the light of the Pioneer Venus mission, J. Geophys. Res., 85, 8187 – 8199, doi:10.1029/JA085iA13p08187. Tsang, C. C. C., P. G. J. Irwin, F. W. Taylor, C. F. Wilson, C. Lee, R. de Kok, P. Drossart, G. Piccioni, B. Bezard, and S. B. Calcutt (2008), Tropospheric carbon monoxide concentrations and variability on Venus from Venus Express/VIRTIS-M observations, J. Geophys. Res., 113, E00B08, doi:10.1029/2008JE003089. Turcotte, D. L., G. Morein, D. Roberts, and B. D. Malamud (1999), Catastrophic resurfacing and episodic subduction on Venus, Icarus, 139, 49 – 54, doi:10.1006/icar.1999.6084. Urey, H. C. (1952), The Planets, Their Origin and Development, Yale Univ. Press, New Haven, Conn. Walker, J. C. G., P. B. Hays, and J. F. Kasting (1981), A negative feedback mechanism for the long-term stabilization of the Earth’s surface temperature, J. Geophys. Res., 86, 9776 – 9782, doi:10.1029/JC086iC10p09776. Watson, E. B., J. B. Thomas, and D. J. Cherniak (2007), 40Ar retention in the terrestrial planets, Nature, 449, 299 – 304, doi:10.1038/nature06144. Wilson, C. F., S. Guerlet, P. G. J. Irwin, C. C. C. Tsang, F. W. Taylor, R. W. Carlson, P. Drossart, G. Piccioni, and R. C. Holmes (2008), Evidence for anomalous cloud particles at the poles of Venus, J. Geophys. Res., 113, E00B13, doi:10.1029/2008JE003108. Yamamoto, M., and M. Takahashi (2003), The fully developed superrotation simulated by a general circulation model of a Venus-like atmosphere, J. Atmos. Sci., 60, 561 – 574, doi:10.1175/1520-0469(2003)060< 0561:TFDSSB>2.0.CO;2. Zahnle, K. J. (2006), Earth after the Moon forming impact, Geochim. Cosmochim. Acta, 70, A729, doi:10.1016/j.gca.2006.06.1311.























D. Grinspoon, Denver Museum of Nature and Science, 2001 Colorado Boulevard, Denver, CO 80205, USA. F. Taylor, Department of Physics, Oxford University, Oxford OX1 3PU, UK. ([email protected])

22 of 22

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, E00B09, doi:10.1029/2008JE003075, 2008

for

Full Article

Retrieval of air temperature profiles in the Venusian mesosphere from VIRTIS-M data: Description and validation of algorithms Davide Grassi,1 P. Drossart,2 G. Piccioni,3 N. I. Ignatiev,4 L. V. Zasova,4 A. Adriani,1 M. L. Moriconi,5 P. G. J. Irwin,6 A. Negra˜o,1,7 and A. Migliorini3 Received 14 January 2008; revised 9 May 2008; accepted 2 June 2008; published 1 October 2008.

[1] We present here methods developed for the retrieval of air temperature profiles in the

Venusian mesosphere from the absolute radiances measured by the Visual and Infrared Thermal Imaging Spectrometer (VIRTIS) on board the Venus Express satellite. The infrared M channel of the instrument acquires multispectral images between 1000 and 5000 nm. In nighttime measurements, radiance in the range 3800–5000 nm is dominated by the thermal emission and absorption by the clouds and carbon dioxide. Since the latter is the main atmospheric component, it is possible to exploit the strong variability of its opacity in this spectral range, as resolved by the instrument, to reconstruct the vertical air temperature profile as a function of pressure. In this context we decided to adopt the Twomey et al. (1977) relaxation scheme. The resulting code was extensively tested on a set of simulated VIRTIS-M data. Comparison of the known input conditions with the results of analysis code allowed us to evaluate the systematic and random errors affecting the retrievals procedures on a statistical basis. The code returns the vertical air temperature profile with an uncertainty of less than 1 K in the region between 70 and 7 mbar (66 and 77 km above the reference surface) and less than 4 K throughout the entire range 100–0.1 mbar (64–95 km). Finally, we present the first examples of the code applied to actual measured Venusian data, demonstrating its capability to achieve a satisfactory modeling of the observations and provide physically reasonable results. Citation: Grassi, D., P. Drossart, G. Piccioni, N. I. Ignatiev, L. V. Zasova, A. Adriani, M. L. Moriconi, P. G. J. Irwin, A. Negra˜o, and A. Migliorini (2008), Retrieval of air temperature profiles in the Venusian mesosphere from VIRTIS-M data: Description and validation of algorithms, J. Geophys. Res., 113, E00B09, doi:10.1029/2008JE003075.

1. Introduction [2] The Visual and Infrared Thermal Imaging Spectrometer (VIRTIS) is one of the experiments included in the scientific payload of the Venus Express ESA mission [Titov et al., 2006]. Its design is largely based on the heritage of the twin instrument on board the ESA Rosetta satellite. The instrument actually consists of an ensemble of different

1 Istituto di Fisica dello Spazio Interplanetario, Istituto Nazionale di Astrofisica, Rome, Italy. 2 Laboratoire d’Etudes Spatiales et d’Instrumentation en Astrophysique, Observatoire de Paris, Universite´ Pierre et Marie Curie, Universite´ Paris 7, CNRS, Meudon, France. 3 Istituto di Astrofisica Spaziale e Fisica Cosmica, Istituto Nazionale di Astrofisica, Rome, Italy. 4 Space Research Institute, Russian Academy of Sciences, Moscow, Russia. 5 Istituto di Scienze dell’Atmosfera e del Clima, Consiglio Nazionale delle Richerche, Rome, Italy. 6 Clarendon Laboratory, Atmospheric, Oceanic and Planetary Physics, University of Oxford, Oxford, UK. 7 Faculdade de Engenharia da Universidade do Porto, Porto, Portugal.

subsystems, with specific measuring capabilities. VIRTIS-H is a high-resolution grating spectrometer operating in the range 3000 –5000 nm, with a typical resolution and sampling step of 1.5 nm (variable along the range and with spectral order). VIRTIS-M is a spectro-imager, able to acquire a stack of monochromatic images that allow the spectrum to be reconstructed for each pixel. VIRTIS-M operates simultaneously in the visible and infrared part of the spectrum. The ranges covered are 280– 1100 and 1000– 5000 nm, with sampling steps of 19 and 11 nm, respectively. The VIRTIS-M and -H instantaneous fields of view (IFOV) are 0.25
0.25 mrad (for an individual pixel) and 1.74
0.58 mrad. These figures, together with the Venus Express orbital parameters, lead to a horizontal resolution for individual pixels of 16.5
16.5 and 115
38 km, respectively, in the case of measurements acquired at the apocenter. A complete description of the instrument and its radiometric performances is given by Piccioni et al. [2007a]. The noise equivalent radiance pertinent to each measurement depends on several factors such as the instrument temperature and exposure time. For the purpose of this study, a value of 5104 erg/(sec. cm2 ster. nm) at 4300 nm was assumed.

Copyright 2008 by the American Geophysical Union. 0148-0227/08/2008JE003075$09.00

E00B09

1 of 12

GRASSI ET AL.: VENUS VIRTIS-M T(P) RETRIEVALS

E00B09

Figure 1. Example of a VIRTIS-M IR spectrum acquired on the Venus nightside. [3] Once calibrated, the radiation field measured by VIRTIS in orbit around Venus is given by the radiative transfer equation [Hanel et al., 2003, chap. 4]:

E00B09

environment has been demonstrated by the wide literature derived from the data acquired by Pioneer Venus Infrared Radiometer and Venera 15 Fourier Transform Spectrometer (FTS). Some examples include a first study of air temperature fields in the mesosphere as a function of latitude and altitude [Taylor et al., 1980; Moroz et al., 1986], aerosol vertical distributions [Zasova et al., 1999], and water vapor mapping [Ignatiev et al., 1999]. The experience accumulated by the corresponding scientific teams represents therefore the fundamental starting point for this kind of study. [5] The full exploitation of the VIRTIS data is a task well beyond the ambition of an individual article. In this paper we present the first successful retrievals derived from the VIRTISM IR subsystem data, aiming to determine the temperature structure of the Venusian atmosphere. Several concepts presented here have their roots in the previous studies by Grassi et al. [2005], Zasova et al. [1999] and Roos-Serote et al. [1995]. For the sake of brevity, we will hereinafter use the term ‘‘VIRTIS’’ as synonymous of ‘‘VIRTIS-M IR subsystem.’’

2. Information Content of VIRTIS Data 2.1. Example of Venusian VIRTIS Spectra [6] Figure 1 presents an example of an actual VIRTIS Venusian nighttime observation. A visual inspection allows


 In t n;total ; m; f ¼ tZn;total 
 
 0 1 A: 1  vo;n t n ¼ t 0n eðtn;total tn Þ=m Bn T t n ¼ t 0n dt 0n þ m 0

B:

C:

D:

1 4pm Fn;0 4p 1 4pm

tZn;total Z2p

Z1

0 0 tZn;total

1



 0 eðtn;total tn Þ=m~ pn t n ¼ t 0n ; m; f; m0 ; f0 In t n ¼ t 0n ; m0 ; f0 dm0 df0 dt 0n þ ð1Þ


 0 0 ~pn t n ¼ t 0n ; m; f; m0 ; f0 eðtn;total tn Þ=m eðtn;total tn Þ=m0 dt 0n þ

0 tZn;total Z2p

0

0

Z1


 0 eðtn;total tn Þ=mSn t n ¼ t 0n ; m; f; m0 ; f0 dm0 df0 dt 0n

1

ð1Þ

where t n is the optical thickness, pn is the aerosol phase function, v0, n is the aerosol single scattering albedo, Bn is the Planck function and F0,n is the solar flux density. [4] This expression includes the terms describing the thermal emission by the surface and atmosphere (A), as well as a modeling of the reflection of the solar radiation (C), multiple scattering phenomena (B) and other nonthermal emissions (D) such as radiation due to non-LTE (local thermal equilibrium) phenomena. Term (B) makes (1) an integral equation, without analytical solutions. Even with a complete knowledge of the vertical distributions of aerosols and gases, air temperatures (driving thermal emission described by Planck function B) and optical properties of suspended materials, the analytical determination of the expected radiation field is not usually possible; consequently, it has to be estimated numerically. The inversion of (1), aiming to retrieve the above mentioned quantities, becomes therefore an extremely complex problem, often without unique solution [Rodgers, 1976]. Nevertheless, the scientific potential of infrared measurements in the study of Venusian

us to identify the main spectral features and relate them with specific parameters of Venusian environment. [7] 1. For wavelengths less than 3000 nm, the H2SO4 droplets (the main cloud component) become much less absorbing and they behave, effectively, as conservative scatterers. Thermally emitted radiation from lower atmosphere (i.e., below the main cloud deck) and even from the surface may therefore escape to space in the windows such as those at 1180, 1740 and 2300 nm that lie between strong atmospheric absorption bands. [8] 2. The signal level measured in the range 3500 – 5100 nm follows roughly a Planckian shape (in its Wien tail), corresponding to the temperature of clouds where the optical thickness reaches a unity value. [9] 3. Two strong CO2 absorptions bands are centered at 4300 and 4800 nm. [10] 4. The main CO band is evident as a continuum depression centered at 4600 nm. [11] 5. Again in the range 3500 – 5100 nm, deviations from an ideal Planckian shape, in the spectral intervals not

2 of 12

E00B09

GRASSI ET AL.: VENUS VIRTIS-M T(P) RETRIEVALS

Figure 2. Levels of unit optical depth of different VIRTISM IR channels for typical Venusian conditions. The absorption of both clouds and CO2 is included in the computations. severely affected by gases, are ascribed to variations of cloud optical properties. 2.2. Assessment of Retrieval Capabilities [12] The use of Bayesian formalism [Rodgers, 2000] allows a precise assessment of retrieval capabilities offered by individual radiometric measurements, but its usage in the Venusian case is de facto precluded by the lack of reliable a priori information. The data sets collected by previous missions, despite their obvious scientific values, do not allow the compilation of a set of climatology statistics with adequate seasonal, longitudinal and local time coverage. Consequently, a heuristic analysis, to be confirmed by numerical tests, represents the most reasonable method to assess the information content offered by VIRTIS data. [13] At VIRTIS resolution, the CO2 4300 – 4800 nm bands are covered by about 90 sampling channels (hereinafter, the term ‘‘channel’’ will be used as a synonym for ‘‘sampling position along the wavelength grid’’). For typical Venusian temperatures and aerosol densities (Seiff [1983] and models references therein), the VIRTIS channels where S/N exceeds 3 reach a unity opacity in the pressure range between 100 and 0.1 mbar ( 64 – 95 km), when both aerosols and CO2 are considered (Figure 2). These opacity values are derived from averages of monochromatic opacities weighted by the instrumental line shapes of individual channels. Our data may therefore carry information of the thermal structure of the atmosphere in the same pressure range, as far as term (A) in equation (1) dominates the emerging radiation field. [14] This condition is true only for nighttime data. During daytime, the 4300-nm band is affected by the scattering of solar radiation by the Venusian clouds as well as by intense non-LTE emission by CO2 molecules in the higher atmosphere [Lopez-Valverde et al., 2007]. The cumulative effects of these terms may easily reach a value of 40% of the total signal at 4500 nm. Even if both effects could possibly be included in the modeling of observed radiances, the induced computational load is so heavy that we prefer to limit

E00B09

ourselves to the analysis of nighttime data in this first phase of data processing. [15] Analysis of the deep atmosphere temperature structure is also severely limited by current modeling issues. Radiances measured in the atmospheric windows commonly show variations of the order of 30 –40%, which are interpreted as being due to variable thickness of the cloud decks [Taylor et al., 1997]. On the other hand, these lower clouds are mainly composed of mode 3 particles [Esposito et al., 1983], for which substantial uncertainties of optical properties, shape and composition exist. In this study, we considered the Venus aerosols model given by Crisp [1989]: (Mode 1: reff = 0.49 mm, Mode 20: reff = 1.18 mm; Mode 2: reff = 1.14 mm; Mode 3: reff = 3.85 mm). [16] Absorption properties of CO2 at the high temperatures and pressures of Venus lower atmosphere are poorly known at the present date [Taylor et al., 1997] and current data analysis attempts largely rely on empirical corrections. In these conditions, even if a temperature retrieval scheme may possibly be envisaged, its application to actual data processing would be questionable. Its development has therefore been postponed to later phases of our work and will not be described here.

3. Characteristics of Retrieval Code 3.1. Preliminary Considerations [17] According to equation (1), for nighttime observations, the observed radiance level depends on the behavior of atmospheric temperatures T as function of total opacity t, i.e., I = f(T(t)). For a given vertical distribution of optically active species, this is equivalent to the atmospheric temperatures as a function of altitude or pressure. As long as we neglect the CO band, the opacity between 3500 and 5100 nm is due to CO2 and clouds. [18] 1. Carbon dioxide is the main constituent of Venus atmosphere, and has a constant mixing ratio of 0.965 in the mesosphere. Therefore, once a temperature profile is assumed, we can compute its density on a fixed pressure grid (and the related opacities) on the basis of hydrostatic equation and the perfect gas law. [19] 2. In the absence of CO2 absorption, unity optical thickness would be reached in the spectral range of our interest in the upper cloud deck [Esposito et al., 1983]. Even though this structure appears less variable than its lower counterpart, variations of effective altitude and density decay scale heights have been reported by Zasova et al. [1999] on the basis of Venera 15 FTS data. [20] Since both air temperatures and aerosol densities are variable and contribute to total opacity, it appears that the two unknowns have to be retrieved simultaneously in order to achieve a self-consistent modeling of experimental data. Unfortunately, the optical properties of mode 2 particles (the main components of upper cloud deck) are almost constant in the VIRTIS spectral range and therefore it is extremely hard to discriminate between the effects of these two parameters. For example, in the case of a positive temperature lapse rate close to the cloud top, the same radiance level can be achieved either with a higher air temperature and high cloud or with a lower temperature and lower cloud. In numerical (noise-free) radiance simulations these two cases lead to different spectral intensities and shapes.

3 of 12

E00B09

GRASSI ET AL.: VENUS VIRTIS-M T(P) RETRIEVALS

However, in the case of actual data, the magnitude of these differences is comparable to the modeling errors due to uncertainties on aerosol optical properties, radiative transfer approximations or residual calibration offset. Therefore, a ‘‘nonuniqueness-of-solution’’ problem exists at altitudes where the cloud opacity reaches unity values. On the basis of analysis of aerosol densities retrieved from Venera 15 FTS data [Zasova et al., 1999], we can see that this level of unity opacity is located, for most cases, around 100 mb. 3.2. Initialization [21] The initialization of the state vector (i.e., input quantities for the radiative transfer computation) is performed in two main phases: (1) definition of the quantities to be assumed as known and (2) setting of a first guess for the quantities to be retrieved. [22] The following list describes the preferred initialization options for phase 1 adopted during our analysis of measured Venusian spectra: [23] 1. Clouds were assumed to consist of liquid droplets of a mixture composed of 75% of H2SO4 and 25% H2O, whose complex refractive indices were taken from Palmer and Williams [1975]. [24] 2. Since the upper cloud is mostly composed by Mode 2 particles, only this aerosol component has been included in our simulations. Following Crisp [1989], a lognormal distribution with reff = 1.14 mm and variance of 0.23 mm was adopted. [25] 3. Gaseous CO2 lines data were extracted from the HITRAN 2004 database [Rothman et al., 2005]. These data were used to build an expansion of preconvolved transmittance profiles in the Venusian mesosphere, according to the methods described by Zasova et al. [1999]. [26] During phase 2 of initialization, we should keep in mind that owing to the nonlinearity of the radiative transfer equation, an initial guess as close as possible to the true value is highly desirable to avoid convergence toward a purely mathematical, but unphysical solution. The analysis of VIRTIS data currently conducted in our team adopts the following guidelines: [27] 1. A study carried out on a wide population of simulated spectra (see section 4.1) allowed us to assess the correlation coefficient between the air temperatures at different pressure levels of the atmosphere with the observed brightness temperature at the different VIRTIS sampling points. The maximum correlation level as a function of sampling point matches closely the level of unit optical depth, plotted in Figure 2. About fifteen sampling points outside the CO contaminated region were selected to provide initial guesses of air temperatures at corresponding levels. These estimates were then fitted with a third degree polynomial in the entire pressure range considered in our computations. [28] 2. The aerosol density as a function of pressure was initialized to the average values derived from Venera 15 FST data. In the version of the retrieval code presented here, no effort was made to provide a more refined guess on the basis of the data. Future improvements of our code could benefit from a correlation analysis on simulated radiances carried out during our validation tests. This study demonstrated that the ratios of brightness temperatures measured in symmetric positions of the 4800 nm CO2 band with

E00B09

respect to its center are highly sensitive to effective scale height and altitude of the cloud deck. 3.3. Retrieval Algorithms [29] In our code, air temperatures and aerosol densities are retrieved in two separate steps, nested inside an outer iterative loop. This design is justified by the different retrieval algorithms employed for the two unknowns. [30] 1. The computation of vertical air temperature profile has been widely discussed in literature for the case of the Earth’s atmosphere as well as for other planets. A general introduction on these topics is given by Hanel et al. [2003, chap. 7]. In our case we adopted the algorithm proposed by Twomey et al. [1977], in the formulation presented by Zasova et al. [1999]. In this scheme, the temperature update between two consecutive iterative cycles is given by m P

Tliþ1

¼

TBch ðIobs;ch Þ




i ch¼1 TBch Iexp ected;ch Tli m P

 Wch;l ð2Þ

Wch;l

ch¼1

where j, ch, and l are the cycle, sampling channel and pressure level indices, respectively, and TB are the brightness temperatures corresponding to the radiances observed by VIRTIS (Iobs) or modeled according the current temperature profile (Iexpected). W is the weighting function, i.e.,  @Tch  Wch;l ¼  @ log10 pp¼pl

ð3Þ

where T is the total atmospheric transmittance between space and a given pressure level. The Twomey et al. [1977] algorithm falls into the general class of relaxation methods, which are based on empirical corrections of a first guess profile, aiming mainly to achieve a perfect correspondence between the observed and synthetic spectrum, computed on the basis of the current value of the T(p) function. These methods are usually very robust, but suffer from weak theoretical basis. They are not able to take into account correlations between the temperatures at different levels of the atmosphere or the instrumental noise equivalent radiance (NER hereinafter). The latter point implies that they cannot provide an analytical estimation of the error on the retrieved solution, which shall therefore be assessed by means of numerical tests. Consequently, the only criteria to stop their iterations comes from the comparison between the NER, observed and modeled spectra. In our case, the atmosphere is modeled by a stack of 67 pressure levels logarithmically spaced between 1200 and 0.005 mbar ( 49– 106 km). The air temperature retrieval takes into account the VIRTIS radiances measured between 4250 and 5000 nm, excluding the spectral region dominated by CO centered roughly at 4600 nm. The low-wavelength shoulder of the 4300 nm CO2 band proved very difficult to model, as observed by Roos-Serote et al. [1995]. These authors suggested line mixing as a possible cause of misfit. Since our radiative transfer model is not yet able to account for this effect, the 4000 – 4250 nm region was not considered for our analysis purposes.

4 of 12

E00B09

GRASSI ET AL.: VENUS VIRTIS-M T(P) RETRIEVALS

E00B09

Figure 3. Distribution of cases considered for code validation in the latitude-local time space. Points, Venera 15 FTS measurements; triangles, selected cases considered for this study.

Figure 4. Random (dashed line) and systematic (solid line) components of air temperature retrieval errors for the developed algorithm. Values refer to VIRTIS-M data once spatial pixels have been averaged in 10
10 bins.

[31] 2. The aerosol density profile is modeled as an exponential decay:

[34] These data were used as input for a direct radiative transfer code with the same characteristics of that used in the retrieval algorithm, namely: [35] 1. CO2 transmittances have been evaluated according to the preconvoluted transmittances approach. [36] 2. Aerosols have been assumed to be fully described by mode 2 particles. [37] 3. Scattering has been treated according to the TWOSTR algorithm by Kylling et al. [1995]. [38] For all cases, an emission angle equal to zero has been assumed. [39] The correspondence of the radiative transfer scheme in simulation and retrieval phases ensures that we are

nðzl Þ ¼ nðz0 Þe

ðzl z0 Þ ha

ð4Þ

where z0 is the nominal altitude of 1200 mbar pressure level. With this assumption, n(z0) and ha become the two free parameters to be computed on the basis of the data. In this case, we adopted a classical Gauss-Newton retrieval scheme, where the partial derivatives of simulated radiances with respect to the two free parameters are evaluated by explicit differentiation at each iterative step. The iteration is stopped when variations of both free parameters between two successive iterative steps become less than 1%. The aerosol retrieval takes into account the VIRTIS radiances measured between 4450 and 5100 nm, but again excluding the spectral region dominated by CO.

4. Retrieval Code Performances [32] The retrieval code performances were assessed on statistical basis by applying the software on a wide population of simulated spectra. The comparison between the retrieval outcomes and the known input conditions allowed us to estimate the random and systematic components of the retrieval errors for Venus conditions. 4.1. Reference Simulated Spectra Set [33] Temperature and aerosol density profiles derived from analysis of the Venera 15 FTS were selected in order to achieve, as far as possible, a uniform coverage in the latitude and local time space. A total of 187 cases have been considered for the subsequent analysis (Figure 3). The low values of Venus orbit eccentricity and inclination of rotation axis ensure limited seasonal cycles as well as, to a first approximation, a symmetry in the atmospheric structures of the two hemispheres. Since the profiles are derived from actual data, we are confident that they retain the actual characteristics of Venus mesosphere and its variability.

Figure 5. Correlation coefficients between retrieval errors at different altitudes.

5 of 12

E00B09

GRASSI ET AL.: VENUS VIRTIS-M T(P) RETRIEVALS

E00B09

systematic error component in our retrieval scheme, while the standard deviation gives a measure of its random part. Both are presented in Figure 4. Cumulative errors remain below 4 K in the entire range 100– 0.1 mbar ( 64– 95 km), and below 1 K in the indicative span 70– 7 mbar ( 66– 77 km). [42] Further tests, not shown here for shake of brevity, allowed us to identify the main causes of limitation in retrieval performance. [43] 1. At the higher altitudes, the limiting factor is represented by the instrumental noise. The higher part of the atmosphere is probed by the channels with stronger CO2 absorption located at the center of the band, which have a very low signal level. A test carried out using the VIRTIS NER pertinent to individual spectra (i.e., not averaged on 10
10 bins) demonstrated that the upper boundary of region effectively probed by retrieval shrinks downward from 0.1 to 1 mbar (from 95 to 85 km).

Figure 6. (a) Comparison between the retrieved and true value of cloud deck altitude, defined as the pressure level where unity optical thickness at 5000 nm is achieved. Code performances are heavily affected by the nonuniqueness of the solution once we try to determine simultaneously air temperatures in the same pressure range. (b) The same as Figure 6a, but assuming the true air temperature profile during retrieval. actually investigating the performances of the algorithm itself and not other possible sources of error due to the required assumptions. [40] Random noise, with the same statistic of instrumental NER, was finally added to the simulated data. In actual operative conditions, VIRTIS-M monochromatic images are spatially degraded by averaging pixels on 10
10 bins, to achieve acceptable processing times. To account for that, the actual instrumental NER has been divided by 10 in order to simulate the increase in signal-to-noise ratio derived from the averaging process. 4.2. Performance Test Results [41] The case-by-case comparison between the retrieved and input temperature profiles allowed us to build, for each pressure level, a statistical population of retrieval errors. The average value of this population is interpreted as a

Figure 7. Examples of the retrieval code capability to properly model the simulated observations. In the bottom panels, the solid line is the true spectrum, and the dashed line is the spectrum as modeled by retrieval code. A positive offset of 5
103 was added to the latter curve for clarity. The top panels show the difference between true and modeled spectra.

6 of 12

E00B09

GRASSI ET AL.: VENUS VIRTIS-M T(P) RETRIEVALS

E00B09

retrieved temperature profiles were compared on a separated test and demonstrated to be less than 1 K above the reference level of 100 mbar. [48] The code, in its current implementation, performs a tentative retrieval of aerosol density profile, where actual free parameters are n(z0) and ha (see section 3.3 and equation (4)). This modeling is, however, intended only to define a continuum level for the CO2 absorption. Numerical tests on simulated radiances actually demonstrated that the code is unable to retrieve properly the altitude of cloud deck; retrieved and true values of pressure level where unity optical thickness is achieved appear completely uncorrelated (Figure 6a). This fact has to be related with the nonuniqueness of solution discussed in section 3.1; assuming for each test case the corresponding ‘‘true’’ temperature profile allows us to properly retrieve aerosol densities in a variety of situations (Figure 6b). Similar numerical tests demonstrated Figure 8. Statistics of c2 values computed from the fit between observed (simulated) and modeled spectra in the wavelength range [4250, 4450] nm. [44] 2. Errors in the retrieved aerosol densities are the main cause of temperature retrieval errors in the lower mesosphere. 100 mbar ( 64 km) is the indicative level where aerosol clouds reach a unity optical thickness at 4500 nm. Assuming for each test case the corresponding ‘‘true’’ aerosol density (i.e., retrieving only the temperature profile) leads to a decrease from 100 to 300 mbar (from 64 to 59 km) of the lower boundary of the region effectively probed by retrieval. [45] Different options of the temperature profile initialization have a very limited impact on the retrieval performance, affecting mainly the behavior of systematic components outside the reference 100– 0.1 mbar pressure range. Inside these boundaries, differences in retrieval performance with respect to the reference case of Figure 4 are less than 0.3 K. [46] Further insights on code behavior are provided by correlation studies on retrieval errors at different altitudes (Figure 5). This plot (1) shows how errors at different levels may compensate each other (regions of negative correlations) and (2) illustrates the vertical resolution of retrieval scheme. Actually, nonzero correlations between different levels indicate the occurrence of systematic components in retrieval errors along the vertical grid. These components have their ultimate cause in the finite width of weighting functions in equation (2), or, in other words, in the limited vertical resolution of retrieval. [47] The latter point is quite ambiguous owing to the presence of negative lobes, but it provides at least in the lower atmosphere an indicative figure similar to a gaseous scale height (of the order of 10 km for typical Venus conditions) and slightly less above 2 mbar ( 83 km). The error estimate shown in Figure 4 is considered as a minimum, since other potential sources of errors such as (1) radiative transfer approximations, (2) possible systematic calibration uncertainties or artifacts, and (3) uncertainties in the HITRAN parameters were not considered in this validation test. The effects of different radiative transfer schemes (namely, two streams and multi streams) on

Figure 9. Examples of retrieval code capability to properly model the actual Venusian VIRTIS observations. In the bottom panels, the solid line is the true spectrum, and the dashed line is the spectrum as modeled by retrieval code. A multiplicative factor of 1.5
103 was applied to the latter curve for clarity. The top panels show the difference between true and modeled spectra. Both examples are extracted from the VI0038_02 frame.

7 of 12

E00B09

GRASSI ET AL.: VENUS VIRTIS-M T(P) RETRIEVALS

E00B09

Figure 10. Two-dimensional air temperature maps at (a) 90 and (b) 35 mbar levels ( 65 and 70 km), derived from the analysis of the VI0038_02 frame. that in most cases, errors in the aerosol retrieval do not add errors in the temperature retrieval greater than 3 K above the 100 mbar level. [49] The code has been demonstrated to be capable of providing satisfactory modeling of simulated data in a variety of situations (Figures 7a and 7b). Nevertheless, the percentage of cases where the code was not able to reproduce observations within instrumental error is far from being negligible (Figure 8). A direct inspection shows how the most problematic cases are those related to weak thermal inversion and low altitude of cloud deck. In these conditions, the initial guess of aerosol density was demonstrated to be particularly poor and the code was not able to converge toward a realistic solution, oscillating until the maximum number of iteration was achieved. In these cases, however, systematic differences between observed and modeled radiances well above NER level appear around 5000 nm, allowing the results to be effectively filtered.

pling grid or to approximate values for instrumental resolution of individual sampling channels. [51] Figure 10 shows two typical two-dimensional maps of air temperature at two different pressure levels, derived from an individual frame acquired during orbit 38. The maps present quite smooth behavior, despite the fact that the analysis of each pixel was carried out completely independently from the surrounding ones. This fact demonstrates the stability of the retrieval code and its robustness against occasional calibration issues (dead columns, caused mainly by impinging cosmic rays). This fact is further confirmed by the analysis of partially overlapping frames that, when available and acquired within small time intervals, show

5. Examples of Temperature Mapping From Venus Data [50] The purpose of this section is to provide to the reader an idea of the actual behavior of the code once applied to real Venus measurements. The modeling of Venus’ spectra provides usually very satisfactory fits (Figure 9). Systematic effects can be appreciated in the following spectral regions: (1) main CO band around 4600 nm, not included in our simulations, (2) above 5000 nm, in a region with weak CO2 absorption, probably related to the approximate aerosol models adopted during retrieval or radiative transfer algorithms not being completely adequate, and (3) around the radiance local maximum at 4270 nm, likely to be ascribed to residual registration errors in the VIRTIS wavelength sam-

Figure 11. Average temperature field along the meridian 180°, as derived from frames VI0063_01 to VI0063_07. Local time is about 1900.

8 of 12

Figure 12. Comparison of a VIRTIS frame documenting (a) the occurrence of the polar dipole (l = 5 mm) and the air temperature fields derived from the same frame at (b) 90 mbar and (c) 35 mbar. Temperature fields were derived from the VI0038_02 frame and represent higher-resolution versions of Figure 10.

E00B09 GRASSI ET AL.: VENUS VIRTIS-M T(P) RETRIEVALS

9 of 12

E00B09

Figure 13. Sequence of air temperature fields along the 0° meridian. The arrows highlight the patterns of the air temperature isosurface that appears to propagate between consecutive frames (cubes VI0038_01 to VI0038_04). The time interval between frames is exactly 1 hour.

E00B09 GRASSI ET AL.: VENUS VIRTIS-M T(P) RETRIEVALS

10 of 12

E00B09

E00B09

GRASSI ET AL.: VENUS VIRTIS-M T(P) RETRIEVALS

E00B09

Figure 14. (a) Standard deviation of the air temperature fields, as derived from the time sequence of Figure 13, local time 0400. (b) The same as Figure 14a, but sequence VI0063_01 to VI0063_07, local time 1900. differences in the temperature fields in the order of retrieval errors described in section 4.2.

6. Preliminary Analysis of Air Temperature Fields [52] A complete scientific analysis of results is currently being undertaken by the VIRTIS team, but cannot be considered as conclusive, owing to the limited numbers of observations processed so far. Moreover, owing to the orbital characteristics of the Venus Express orbiter, the available temperature fields document mainly the atmosphere conditions close to the southern pole. Preliminary findings are as follows: [53] 1. The main features of the air temperature fields (Figure 11) are in quantitative agreement with the previous results of Pioneer Venus Infrared Radiometer and Venera 15 FTS for the northern hemisphere [Taylor et al., 1980; Zasova et al., 1999]. Namely, the occurrence of a cold collar at 100 mbar ( 64 km) centered at 65°, and the rise of temperature toward the pole for a wide range of altitudes are observed. Further discussion about the general characteristics of the atmospheric temperature fields, their local time dependences and the derived winds are presented in the companion paper by Piccialli et al. [2008]. [54] 2. The pattern of the polar dipole, as observed in VIRTIS frames at 5 mm [Piccioni et al., 2007b], has an immediate correspondence in the thermal fields only in the lowest parts of probed range (around 90 mbar, 65 km), and is already barely visible at 35 mbar, i.e., 70 km (Figure 12). This evidence may put constraints on the vertical scale at which the dipole develops. The relevant pressure range is characterized by moderate to negative lapse rates. The higher part of probed pressure range is usually much less rich in detail. [55] 3. Comparison of temperature fields derived from frames closely spaced in acquisition times shows the occurrence of constant patterns, with a consistent evolution in time (Figure 13), which strongly suggest a wavelike activ-

ity. Related temperature variations are of the order of 5 K. Bearing in mind the retrieval errors of Figure 4, these values are considered as statistically meaningful. [56] 4. The standard deviation of air temperature derived from these sequences of frames allows us to assess the region of the atmosphere characterized by the strongest variability on timescales of an hour (Figure 14). The region around 1 mbar ( 86 km) is apparently characterized by the strongest variability, well above the random retrieval errors given in Figure 4. Notably, the variability of the atmosphere appears higher just after sunset than before dawn.

7. Conclusions [57] This paper describes the retrieval code in the form that is currently adopted by our team for the analysis of Venus VIRTIS-M data. The extensive validation and error characterization described here make it suitable for routine data processing for scientific purposes. Planned or inprogress studies include (1) phenomenological characterization of air temperature fields as a function of season, (2) their correlations with topography, airglow emissions, cloud altitudes and UV markings, (3) mesosphere energy budget and, in a longer-term perspective, (4) validation and assimilation in Venus global circulation models. Moreover, several improvements are already under development or can be envisaged for the near future: (1) testing of an adding/ doubling algorithm for radiative transfer, (2) extension of the algorithm concept to VIRTIS-H, (3) better initialization of the aerosol densities on the basis of thermal brightness ratios and statistics derived from Venera 15 FTS data, and (4) use of more realistic initial guesses for retrievals, as derived from the analysis of the measurements by other Venus Express instruments (SPICAV and VeRA). These improvements should allow a better characterization of the thermal status and aerosol densities in the lower mesosphere and therefore a more complete exploitation of the information content offered by the VIRTIS data set.

11 of 12

E00B09

GRASSI ET AL.: VENUS VIRTIS-M T(P) RETRIEVALS

[58] Acknowledgments. D.G. carried out this research during an 11 month postdoctoral period in LESIA-Observatoire de Paris, funded by a grant kindly provided by the Paris City Hall (Ville de Paris). VIRTISVenus Express is an experiment developed jointly by IASF-INAF (Italy) and LESIA-Observatoire de Paris (France). The project is funded by ESA, ASI, and CNES. Russian coauthors acknowledge Russian Foundation of Basic Research for financial support (grant RFFI 08-02-01383). The work presented here would not have been possible without the efforts of IASF and LESIA VIRTIS technical staffs.

References Crisp, D. (1989), Radiative forcing of the Venus mesosphere. II. Thermal fluxes, cooling rates, and radiative equilibrium temperatures, Icarus, 77(2), 391 – 413, doi:10.1016/0019-1035(89)90096-1. Esposito, L. W., et al. (1983). The Clouds and Hazes on Venus, in Venus, edited by D. M. Hunten et al., pp. 484 – 564, Univ. of Ariz. Press, Tucson. Grassi, D., N. I. Ignatiev, L. V. Zasova, A. Maturilli, V. Formisano, and M. Giuranna (2005), Methods for the analysis of data from the planetary Fourier spectrometer on the Mars Express Mission, Planet. Space Sci., 53(10), 1017 – 1034, doi:10.1016/j.pss.2005.01.006. Hanel, R. A., B. J. Conrath, D. E. Jennings, and R. E. Samuleson (2003), Exploration of the Solar System by Infrared Remote Sensing, 2nd ed., Cambridge Univ. Press, Cambridge, U. K. Ignatiev, N. I., V. I. Moroz, L. V. Zasova, and I. V. Khatuntsev (1999), Water vapour in the middle atmosphere of Venus: An improved treatment of the Venera15 IR spectra, Planet. Space Sci., 47(8 – 9), 1061 – 1075, doi:10.1016/S0032-0633(99)00030-6. Kylling, A., K. Stamnes, and S.-C. Tsay (1995), A reliable and efficient two-stream algorithm for spherical radiative transfer: Documentation of accuracy in realistic layered media, J. Atmos. Chem., 21(2), 115 – 150, doi:10.1007/BF00696577. Lopez-Valverde, M., P. Drossart, R. Carlson, R. Mehlman, and M. RoosSerote (2007), Non-LTE infrared observations at Venus: From NIMS/ Galileo to VIRTIS/Venus Express, Planet. Space Sci., 55(12), 1757 – 1771, doi:10.1016/j.pss.2007.01.008. Moroz, V. I., D. Spankunch, and V. M. Linkin (1986), Venus spacecraft infrared radiance spectra and some aspects of their interpretation, Appl. Opt., 25(10), 1710 – 1719. Palmer, K. F., and D. Williams (1975), Optical constants of sulfuric acid: Application to the clouds of Venus?, Appl. Opt., 14(1), 208 – 219. Piccialli, A., D. V. Titov, D. Grassi, I. A. Khatunsev, P. Drossart, G. Piccioni, and A. Migliorini (2008), Cyclostrophic winds from the VIRTIS temperature sounding: A preliminary analysis, J. Geophys. Res., doi:10.1029/ 2008JE003127, in press. Piccioni, G., et al. (2007a), VIRTIS: The Visible and Infrared Thermal Imaging Spectrometer, Eur. Space Agency Spec. Publ., in press. Piccioni, G., et al. (2007b), South-polar features on Venus similar to those near the north pole, Nature, 450(7170), 637 – 641, doi:10.1038/ nature06209.

E00B09

Rodgers, C. D. (1976), Retrieval of atmospheric temperature and composition from remote measurements of thermal radiation, Rev. Geophys., 14, 609 – 624, doi:10.1029/RG014i004p00609. Rodgers, C. D. (2000), Inverse Methods for Atmospheric Sounding: Theory and Practice, World Sci., Singapore. Roos-Serote, M., P. Drossart, T. Encrenaz, E. Lellouch, R. W. Carlson, K. H. Baines, F. W. Taylor, and S. B. Calcutt (1995), The thermal structure and dynamics of the atmosphere of Venus between 70 and 90 KM from the Galileo-NIMS spectra, Icarus, 114(2), 300 – 309, doi:10.1006/ icar.1995.1063. Rothman, L. S., et al. (2005), The HITRAN 2004 molecular spectroscopy database, J. Quant. Spectrosc. Radiat. Transfer, 96(2), 139 – 204, doi:10.1016/j.jqsrt.2004.10.008. Seiff, A. (1983), Thermal structure of the atmosphere of Venus, in Venus, edited by D. M. Hunten, et al., pp. 215 – 279, Univ. of Ariz. Press, Tucson. Taylor, F. W., et al. (1980), Structure and meteorology of the middle atmosphere of Venus: Infrared remote sensing from the Pioneer orbiter, J. Geophys. Res., 85(A13), 7936 – 8006. Taylor, F. W., D. Crisp, and B. Be´zard (1997), Near infrared sounding of the lower atmosphere of Venus, in Venus II, edited by S. W. Bougher, D. M. Hunten, and R. J. Phillips, pp. 215 – 279, Univ. of Ariz. Press, Tucson. Titov, D. V., et al. (2006), Venus Express science planning, Planet. Space Sci., 54(13 – 14), 1279 – 1297, doi:10.1016/j.pss.2006.04.017. Twomey, S., B. Herman, and R. Rabinoff (1977), An extension of the Chahine method of inverting the radiative transfer equation, J. Atmos. Sci., 34, 1085 – 1090, doi:10.1175/1520-0469(1977)034<1085:AETTCM> 2.0.CO;2. Zasova, L. V., I. A. Khatountsev, V. I. Moroz, and N. I. Ignatiev (1999), Structure of the Venus middle atmosphere: Venera 15 Fourier spectrometry data revisited, Adv. Space Res., 23(9), 1559 – 1568, doi:10.1016/ S0273-1177(99)00169-6. 

A. Adriani, D. Grassi, and A. Negra˜o, Istituto di Fisica dello Spazio Interplanetario, Istituto Nazionale di Astrofisica, Via del Fosso del Cavaliere 100, I-00133 Rome, Italy. ([email protected]) P. Drossart, Laboratoire d’Etudes Spatiales et d’Instrumentation en Astrophysique, Observatoire de Paris, Universite´ Pierre et Marie Curie, Universite´ Paris 7, CNRS, 5 place Jules Janssen, F-92195 Meudon CEDEX, France. N. I. Ignatiev and L. V. Zasova, Space Research Institute, Russian Academy of Sciences, Profsojuznaja 84/32, 117997, Moscow, Russia. P. G. J. Irwin, Clarendon Laboratory, Atmospheric, Oceanic and Planetary Physics, University of Oxford, Parks Road, Oxford OX1 3PU, UK. A. Migliorini and G. Piccioni, Istituto di Astrofisica Spaziale e Fisica Cosmica, Istituto Nazionale di Astrofisica, Via del Fosso del Cavaliere 100, I-00133 Rome, Italy. M. L. Moriconi, Istituto di Scienze dell’Atmosfera e del Clima, Consiglio Nazionale delle Richerche, Via del Fosso del Cavaliere 100, I-00133 Rome, Italy.

12 of 12

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114, E00B36, doi:10.1029/2008JE003204, 2009

for

Full Article

Structure of the Venus neutral atmosphere as observed by the Radio Science experiment VeRa on Venus Express Silvia Tellmann,1 Martin Pa¨tzold,1 Bernd Ha¨usler,2 Michael K. Bird,3 and G. Leonard Tyler4 Received 23 May 2008; revised 2 February 2009; accepted 18 February 2009; published 23 April 2009.

[1] The European Space Agency Venus Express Radio Science experiment (VeRa)

obtained 118 radio occultation measurements of the Venusian atmosphere between July 2006 and June 2007. Southern latitudes are uniformly sampled; measurements in the northern hemisphere are concentrated near the pole. Radial profiles of neutral number density derived from the occultations cover the altitude range 40–90 km, which are converted to profiles of temperature (T) and pressure (p) versus height (h). Profiles of static stability are found to be latitude-dependent and nearly adiabatic in the middle cloud region. Below the clouds the stability decreases at high latitudes. At an altitude of 65 km, the VeRa T[p(h)] profiles generally lie between the Venus International Reference Atmosphere (VIRA) and VIRA-2 models; the retrieved temperatures at any given pressure level typically are within 5 K of those derived from the Pioneer Venus Orbiter Radio Occultation experiments. A large equator-to-pole temperature contrast of
30 K is found at the 1-bar (1000 hPa) level. The VeRa observations reveal a distinct cold collar region in the southern hemisphere, complementing that in the north. At the latitudes of the cold collars, the tropopause altitude increases relative to higher and lower latitudes by 7 km while the temperature drops roughly 60 K. The observations indicate the existence of a wave number 2 structure poleward of ±75° latitude at altitudes of 62 km. Citation: Tellmann, S., M. Pa¨tzold, B. Ha¨usler, M. K. Bird, and G. L. Tyler (2009), Structure of the Venus neutral atmosphere as observed by the Radio Science experiment VeRa on Venus Express, J. Geophys. Res., 114, E00B36, doi:10.1029/2008JE003204.

1. Introduction [2] The exploration of Venus from space began in the early 1960s. Several American and Soviet flyby missions, orbiters, landers, balloons, and probes explored the Venusian atmosphere over the following decades, supported by ground-based observations. The Pioneer Venus Orbiter (PVO) spacecraft orbited Venus between 1978 and 1992. Extensive data returned by PVO and the related atmospheric probes, together with results from the Venera and Mariner missions, provided the base for the Venus International Reference Atmosphere (VIRA) [Seiff et al., 1985]. [3] The first radio occultation investigation of Venus was conducted during the flyby of Mariner 5 in 1967 [Stanford Mariner Group, 1967; Kliore et al., 1967; Fjeldbo et al., 1971]. This was followed by Mariner 10 [Howard et al., 1974; Nicholson and Muhleman, 1978], Venera-9 and -10 [Kolosov et al., 1979], Venera-15 and -16 [Yakovlev et al., 1 Abteilung Planetenforschung, Rheinisches Institut fu¨r Umweltforschung, Universita¨t zu Ko¨ln, Cologne, Germany. 2 Institut fu¨r Raumfahrttechnik, Universita¨t der Bundeswehr Mu¨nchen, Neubiberg, Germany. 3 Argelander Institut fu¨r Astronomie, Universita¨t Bonn, Bonn, Germany. 4 Department of Electrical Engineering, Stanford University, Stanford, California, USA.

Copyright 2009 by the American Geophysical Union. 0148-0227/09/2008JE003204$09.00

1991], and Magellan [Jenkins et al., 1994; Steffes et al., 1994; Hinson and Jenkins, 1995]. To date, the most extensive campaign of radio occultation experiments at Venus was carried out with PVO [Kliore and Patel, 1980, 1982; Kliore, 1985]. A number of authors have addressed important issues arising from investigation of the Venus atmosphere [e.g., Hunten et al., 1983; Bougher et al., 1997; Moroz, 2002; Taylor et al., 1979, 1980; Taylor, 2002, 2006; Titov et al., 2001; Svedhem et al., 2007]. [4] The European Space Agency (ESA) spacecraft Venus Express (VEX) began orbiting Venus in April 2006. The Venus Express Radio Science experiment (VeRa) uses radio signals at 3.6 and 13 cm wavelength (X and S bands, respectively) to investigate the surface, atmosphere, and ionosphere of Venus. A detailed description of the experiment can be found in the work by Ha¨usler et al. [2006, 2007]. Pa¨tzold et al. [2007] give preliminary VeRa results from the first year in orbit. [5] Temperature changes in lower atmosphere, the troposphere, and the transition region above the tropopause, the middle atmosphere or the mesosphere, are mainly dependent on latitude; several regions of temperature variation, presumably resulting from eddy motions, can be found throughout the atmosphere [Seiff, 1983; Hinson and Jenkins, 1995; Young et al., 1987]. Both small- and global-scale variations associated with gravity waves are present in the lower and middle atmosphere of Venus [Hinson and Jenkins, 1995]. Also, latitude-dependent temporal fluctuations with

E00B36

1 of 19

E00B36

TELLMANN ET AL.: VENUS ATMOSPHERE FROM VERA

periods of 2.48, 2.9, and 5.3 days are apparent in the middle atmosphere between the cloud tops and the upper atmosphere above 100 km [Taylor et al., 1980; Seiff, 1983; Piccioni et al., 2007]. [6] The circulation in the lower and middle atmosphere below the cloud tops is driven mainly by retrograde zonal circulation in near cyclostrophic balance with the pressure gradient [Schubert et al., 1980; Counselman et al., 1979]. A strong retrograde jet with wind speeds exceeding 100 m/s can be found between latitudes ±50 and 55° at a height of about 70 km. This is equatorward of the cold collar regions, located between ±65 and 75° latitude at 60-km altitude, where the zonal retrograde rotation gives way to strong polar vortex structures [Newman et al., 1984]. Hadley circulation is found above the clouds, presumably the result of radiative heating in the clouds. The combination of the zonal rotation and the descending branches of the Hadley circulation, which move poleward at the cloud tops [Limaye and Suomi, 1981; Rossow et al., 1980; Svedhem et al., 2007], forms the strong polar vortex [Suomi and Limaye, 1978; Taylor et al., 1980; Taylor, 2006]. [7] In the remainder of this paper, following an introduction to the method, section 2, the data set, section 3, and an overview of the lower atmosphere, section 4, we provide specifics as follows: [8] 1. Section 5 investigates the static stability in the lower and middle atmosphere of Venus which is a key parameter for understanding the dynamical behavior of an atmosphere. [9] 2. Section 6 compares the tropopause values retrieved from VeRa with previous results from Pioneer Venus Orbiter Radio Occultation (PV-ORO) [Kliore and Patel, 1982] and Venera-15 and -16 [Yakovlev et al., 1991]. The static stability of an atmosphere is closely associated with the behavior of the tropopause. [10] 3. Section 7 investigates the dominant variability of the atmosphere as manifest in changes within the temperature structure as a function of latitude. The VeRa data are compared with earlier data from the PV-ORO experiment, which provided the previously most extensive radio occultation data set [Kliore and Patel, 1980, 1982; Kliore, 1985]. [11] 4. Section 8 describes an investigation of the thermal structure between 75 and 85° latitude for evidence of solarinduced structures in the VeRa data. Diurnal variations are significant above an altitude of 100 km, but circulation features within the Venus middle atmosphere, such as the polar collar and the hot dipoles, also show some solar related variability [Taylor et al., 1980; Piccioni et al., 2007]. [12] 5. Section 9 compares the VeRa profiles with the predictions from two previously established atmospheric reference models, VIRA [Seiff et al., 1985] and VIRA-2 [Moroz and Zasova, 1997; Zasova et al., 2006]. Data retrieved after the publication of the Venus International Reference Atmosphere were used by Zasova et al. [2006] to develop an improved version, VIRA-2.

planetary disk only to reappear at the opposite limb of the planet. During the early and late stages of these events, radio signals from VEX, sound the ionosphere and neutral atmosphere, and are received in closed-loop receiver mode at an ESA or NASA ground station. VeRa employs a oneway, dual-frequency radio link stabilized by a dedicated onboard ultrastable oscillator (USO), which provides precise control of the transmitted signal frequency so that perturbations in the received signal frequencies as small as 3 – 4  1013 can be attributed to propagation effects within the atmospheric path. [14] Within the atmosphere and ionosphere the gradient of refractive index of the gas incrementally deflects the direction of propagation, resulting in a curved propagation path. In this way atmospheric refraction alters the path geometry by an amount that is controlled by the radial variation of refractivity. The result is a subtle but easily measured shift in the frequency of the signal received on Earth. [15] Assuming local spherical symmetry, the degree of bending, the ray periapsis and impact parameter, and the projection of the raypath onto the planetary surface can be computed from geometrical optics and the known Earth spacecraft geometry. The refractivity of the atmosphere at the ray periapsis is obtained from the bending angle via an Abel transform [Fjeldbo et al., 1971]. [16] The refractivity profile m(h) as a function of the altitude h is determined by the local state of the neutral atmosphere and the electron density distribution Ne(h) in the ionosphere as

mðhÞ ¼ C3 

Ne ðhÞ þ C1  nðhÞ  k f02

ð1Þ

where k is the Boltzmann constant, n(h) is the vertical distribution of neutral number density, C3 = 40.31 m3/s2 and fo is the frequency of the radio signal. The first term in (1) describes the effect of the ionospheric electron density. In (1), the neutral number density is related to the refractivity profile through the constant factor C1 which depends on the atmospheric composition of the atmosphere. This proportionality factor has the value C1  1312  106 Kms2/kg for an atmospheric composition of 96.5% CO2 and 3.5% N2 [Essen and Froome, 1951]. At Venus the ionospheric contribution is much weaker than the one from the atmosphere given by the second term in (1). In the reduced data, the neutral and ionized regions are well separated in altitude, the atmospheric term dominating below about 100 km, the ionosphere above 100 km. The VeRa investigations of the ionospheric vertical structure are described by Pa¨tzold et al. [2007]. [17] The ideal gas law relates the pressure, temperature, and neutral number density of an atmosphere pðhÞ ¼ k  nðhÞ  T ðhÞ

2. Radio Science Observation Method [13] VeRa radio occultation observations make use of the spacecraft radio subsystem working with Earth receivers to study the atmosphere. During these observations, as seen from the Earth, the spacecraft disappears behind the

E00B36

ð2Þ

Since hydrostatic equilibrium is very accurate throughout a well-mixed planetary atmosphere, height profiles of pressure and temperature can be derived directly from the neutral number density profile [Eshleman, 1973; Jenkins et

2 of 19

TELLMANN ET AL.: VENUS ATMOSPHERE FROM VERA

E00B36

Figure 1. Example T[p(r)] at high latitudes (day of year (DOY) 150 2007). The occultation point at the 1-bar level (50 km) corresponds to 8 = 85.85° N, solar zenith angle is 89.5°, and local time (LT) is 1550 h. Altitude is approximate relative to the mean Venus radius of 6051.8 km. Three profiles, calculated for upper boundary temperatures of 170, 200, and 230 K, merge at an altitude near 90 km (0.3 hPa). Mean lapse rate 10 K/km obtains below the tropopause. The middle atmospheric lapse rate in the middle atmosphere is much lower; the temperature is nearly isothermal up to the 10-hPa level (10 mbar at roughly 75 km). Small-scale fluctuations are often found in the middle atmosphere. Temperature continues to decrease at higher altitudes up to the upper boundary of VeRa sensitivity near 0.02 hPa (100 km). al., 1994; Ahmad and Tyler, 1998]. In practice, the temperature profile is derived from mup m T ðhÞ ¼  Tup þ mðhÞ k  nðhÞ

Zhup

nðh0 Þ  gðh0 Þdh0

ð3Þ

h

and the pressure follows from 1 pðhÞ ¼ mðhÞT ðhÞ C1

ð4Þ

where m represents the mean molecular mass of the mixed neutral atmospheric species, and g(h) the altitude-dependent acceleration of gravity [Lipa and Tyler, 1979]. [18] The integration parameter Tup = T(hup) in (3) represents the temperature at the upper boundary of the detectable atmosphere. Figure 1 shows the influence of this

E00B36

parameter on the retrieval of the temperature profiles. The three temperature curves have been computed from (3) but with different upper boundary conditions separated by 30 K. The curves merge below about 80– 90 km so that the effect from the upper boundary temperature is negligible below this altitude level. For the work below we chose a boundary temperature of 200 K at an altitude level of 100 km for our retrieval. Except as noted, we relate all altitudes to the mean Venus radius of 6051.8 km. [19] The dense Venus atmosphere is critically refractive at an altitude of roughly 32 km, rendering these lower-altitude levels inaccessible to occultation measurements [Fjeldbo et al., 1971; Ha¨usler et al., 2006]. With reference to the convergence in temperature for solutions with different boundary conditions (Figure 1), we see that the VeRa profiles cover the altitude range from about 40 km up to about 90 km, within which the dependence on the upper boundary condition can be neglected. [20] In this paper we have not provided an analysis of errors. Previous work has shown that occultation measurements at Venus are accurate [e.g., Jenkins et al., 1994; Hinson and Jenkins, 1995; Lipa and Tyler, 1979; Steffes et al., 1994]. The uncertainties in retrieved values of temperature and pressure vary with altitude, but characteristically are 1 K and 0.1 percent, respectively. Appendix B provides a brief discussion of error sources and citations to pertinent literature for understanding this topic.

3. Data Set [21] A total of 118 profiles from the ionosphere and the neutral atmosphere were retrieved from the VeRa observations over the three occultation seasons between July 2006 and June 2007 of the nominal VEX mission. Figure 2 shows the latitudinal distribution of the occultation ingress and egress locations. While the southern hemisphere could be observed with good coverage in latitude and longitude during each occultation season, observations in the northern hemisphere are mainly constrained to latitudes near the pole. The concentration of occultation events near the northern pole, however, provides a good spatial coverage owing to the large number of measurements with high vertical resolution in this area. The VEX orbital configuration presents the opportunity to cover each latitude range at different local times and illumination conditions during the same occultation season. In fact, half of the neutral atmospheric observations are for solar zenith angles (SZA) >90°. This large range of variations supports systematic study of latitudinal, local time and long-term variations in the retrieved profiles.

4. Large-Scale Temperature Structure of the Middle and Lower Venus Atmosphere [22] Figure 1 shows a typical temperature profile of the Venus atmosphere between 40 and 100 km at high latitudes. It is clear that the atmosphere in this altitude range can be divided into two distinct regions according to the thermal structure [Pa¨tzold et al., 2007]: (1) the troposphere, extending from the surface to the large temperature inversion in Figure 1, the tropopause at about 60 km, which is also the upper boundary of the middle clouds [Seiff, 1983], and (2) a

3 of 19

E00B36

TELLMANN ET AL.: VENUS ATMOSPHERE FROM VERA

E00B36

Figure 2. Spatial distribution of ingress and egress occultation points. Points for the neutral atmosphere are given for ray periapsis location at 1-bar level (altitude 50 km). (top) Latitudinal distribution of ingress and egress occultation points by orbit number. (bottom) Latitude versus solar zenith angle of ingress and egress occultation points. transition region, the middle atmosphere (sometimes called the mesosphere), extending from the temperature minimum at the tropopause in Figure 1 to about 100 km, where the diurnal temperature variation becomes extremely pronounced. [23] The temperature in the troposphere decreases monotonically from a surface value of about 735 K [Seiff et al., 1985; Moroz and Zasova, 1997] to values in the range 210 –280 K at the tropopause (Figure 1). The lapse rate has a value near 10 K/km below the tropopause and shows several small abrupt changes. These changes coincide roughly with the boundaries of the highly variable cloud layers in the lower and middle atmosphere at 50– 70 km. The cloud layers, consisting mainly of sulphuric acid droplets, have a strong influence on the thermal structure and stability of the Venus atmosphere, even at heights well above and below the clouds. The circulation patterns in the troposphere and mesosphere are believed to be also strongly dependent on the thick cloud layers. [24] Stratification of the middle atmosphere, the transition region between the tropopause and about 100 km (0.02 hPa), is stable over the VEX observation time, but displays a high thermal variability, especially with regard to latitudinal

changes. The lapse rate is always lower than in the troposphere and nearly isothermal in the latitude range close to the poles. Several small inversions and wave-like structures are also seen (Figure 1). The latter have been attributed to dynamical variations on local or global scales, eddy motions, or gravity waves [Hinson and Jenkins, 1995].

5. Static Stability [25] The static stability S of an atmosphere is defined as the difference between the temperature gradient of the atmosphere dT/dz and its dry adiabatic temperature gradient G: S¼

dT G dz

ð5Þ

where z is the geometric height of the atmosphere above the mean radius. The value of G used for this comparison is from Seiff et al. [1980] using a pure CO2 atmosphere; the difference between a mixed atmosphere and the pure CO2 atmosphere is negligible in this context. While positive values of S indicate a stably stratified atmosphere, negative

4 of 19

E00B36

TELLMANN ET AL.: VENUS ATMOSPHERE FROM VERA

E00B36

Figure 3. Height profiles of static stability for the first occultation season OCC-1. (left) Low, (middle) middle, and (right) high latitudes, respectively (see legend). values of S represent an atmosphere that is unstable against convective overturning. [26] For the first occultation season (OCC-1), Figure 3 compares the stability in three latitude ranges in both hemispheres: 0° > 8 > ±30° (equatorial), ±30° > 8 > ±60° (midlatitudes), ±60° > 8 > ±90° (polar latitudes). Figures 4 and 5 show similar plots for OCC-2 and OCC-3. The atmosphere is generally stable at all altitudes above 45 km, although there is considerable variation in the lowest 20 km probed. The altitude levels displaying the most abrupt changes coincide with the boundaries of the cloud layers, as determined by cloud tracking instruments [Ragent and Blamont, 1980]. [27] A stable region in the troposphere, beginning just below the middle cloud deck near 50 km, extends well below the clouds. Results from in situ measurements have shown that the stability is maximum at about 45 km, just below the cloud layer; the entire stable region is about 20 km in vertical extent [Seiff et al., 1980; Young et al., 1987]. VeRa results show that the upper boundary may depend on the latitude, as upper boundary rises at high latitudes during the three occultation seasons, in good agreement with the tropopause altitude (section 6). [28] The lapse rate is close to adiabatic over the altitude range of the middle cloud, approximately from 49 to 59 km, and is latitude-dependent. Convective regions (S < 0) are found only in very shallow layers within the middle cloud

deck. Figure 4 (middle), between 52 and 54 km, shows typical examples where S  1. This result is in good agreement with previous observations that show a stable, stratified atmosphere with a shallow convective region in the middle cloud deck [Schubert et al., 1980; Seiff et al., 1980; Seiff, 1983; Hinson and Jenkins, 1995]. The altitude range of adiabatic behavior appears to expand toward higher latitudes and the upper boundary increases from the equator toward the cold collar region. While the convection region thickness is close to 5 km at equatorial latitudes, it doubles to 10 km at polar latitudes. [29] We note the presence of two regions of high stability. The uppermost occurs above
64-km altitude, coincident with the upper boundary of the cloud layer. The lower is located just below the middle cloud layer. All profiles show a highly stable layer above the middle cloud in the altitude range 60– 65 km. The regions above about 70 km have a mean stability of S  10 K/km with several superimposed wavelike structures that might be associated with local- or planetary-scale dynamics such as gravity waves or eddy motions. The stable layer below the clouds does not appear in all high-latitude profiles, in agreement with Yakovlev et al. [1991]. [30] The transition between stable and adiabatic regions is fairly smooth at equatorial latitudes, but abrupt at middle and polar latitudes. The change in lapse rate from adiabatic to stable defines the tropopause altitude (section 6).

5 of 19

E00B36

TELLMANN ET AL.: VENUS ATMOSPHERE FROM VERA

E00B36

Figure 4. Same as for Figure 3, but for the second occultation season OCC-2.

[31] The approximately adiabatic behavior in the middle cloud can be explained by the influence of clouds on atmospheric heating. Infrared radiation from the deep atmosphere is trapped within the clouds only to be partly reemitted at longer wavelengths to adjacent areas below. This process cools the upper part of the middle cloud, leading to the observed adiabatic behavior in the cloud. This process can explain the thick stable layer just below the clouds and the nearly adiabatic layer within the cloud [Schubert et al., 1980; Seiff, 1983]. [32] The existence of the highly stable layer, S > 0, just below the clouds may have an important connection with the dynamics acting to separate the altitude ranges above and below with regard to atmospheric wave propagation and meridional circulation [Schubert et al., 1980]. Hadley circulation, which may extend into the cloud deck, with equatorward flow at an altitude of 45– 55 km [Seiff et al., 1980] and poleward flow above the upper cloud, is cut off below this level, thereby isolating the deeper atmosphere from the dynamics of the upper regions. At the same time, convective motions may occur at the same time in the nearly adiabatic region in the middle cloud.

6. Tropopause [33] The lapse rate in the troposphere is slightly less than the dry adiabatic lapse rate at most altitude levels, indicating

an atmosphere which is stable against convective overturning. The tropopause level can be defined as the altitude at which the temperature lapse rate shows a significant decrease (Figures 3 – 5). In order to specify a working threshold, the tropopause is identified as the altitude at which the absolute difference between the observed temperature lapse rate and the adiabatic lapse rate exceeds a value of 8 K/km. This threshold is consistent with that used by Kliore [1985]. [34] The overall stability profile above about 65 km, in the middle atmosphere, is invariant within each VeRa occultation season (Figures 3 – 5) with only occasional small-scale differences from profile to profile. Large-scale differences from occultation season to occultation season occur above the cloud tops, however. These differences do not affect the overall stability of the atmosphere above the tropopause. [35] Comparing the upper boundaries of the nearly adiabatic region in Figures 3– 5 illustrates the latitudinal change of the tropopause altitude. The tropopause structure is very weak and occurs at lower altitudes near the equator. The neutrally stable layer below the tropopause expands vertically from equator to pole. The lapse rate in the middle cloud is extremely close to adiabatic, especially at midlatitudes. The high latitudes show an enhanced level of smallscale fluctuations in the lapse rate, possibly connected with the intermittent polar vortex structures.

6 of 19

E00B36

TELLMANN ET AL.: VENUS ATMOSPHERE FROM VERA

E00B36

Figure 5. Same as for Figure 3, but for the third occultation season OCC-3. [36] The tropopause and the temperature inversion region in this altitude range depend strongly on latitude [Kliore and Patel, 1982; Yakovlev et al., 1991]. This behavior is coincident with the strong zonal winds and the superposition of the zonal retrograde rotation and the descending branch of the Hadley circulation in the collar region leading to the strong polar vortex structures in the polar regions. [37] Figure 6 shows the variation of the tropopause temperature as a function of latitude for both hemispheres. The cubic spline fits are calculated for both hemispheres separately in order to bring out possible hemispheric asymmetries. Each profile is assigned to the latitude at which the closest approach of the radio raypath crosses the tropopause altitude as defined by the stability criterion. The spatial coverage in the northern and southern hemispheres is quite different owing to the geometrical distribution of the occultation points over the occultation season. While the VEX orbit provides a broad spatial distribution of measurements in the southern hemisphere, measurements in the northern hemisphere are mainly clustered in the polar region. [38] Our observations show that the higher latitudes of the two hemispheres containing the cold collar and the polar regions are very similar. The tropopause temperature decreases from a maximum in the lower latitudes near ±15° to a minimum near ±70° latitude, the cold collar region, and rises again going closer to the poles. The minimum temperatures of the spline curves in both cold collar regions are 219.0 and 218.0 K at 73°N and 74°S,

respectively. The tropopause temperatures within 5° of the poles are higher, with 232.8 ± 12.4 K and 233.6 ± 4.1 K for the northern and southern hemispheres, respectively. An overview of tropopause conditions can be found in Table 1. The VeRa tropopause temperatures in the polar region are slightly lower than those reported from PV-ORO [Kliore and Patel, 1982], and those from Yakovlev et al. [1991], but the two are in agreement within the limitations of the data. In the low latitudes the VeRa values are lower than those from Kliore and Patel [1982]. These differences reach values of 20 K (3.0 km) where the data are sparse. VeRa, PV-ORO, and Venera-15 and -16 show good agreement in the cold collar region. The data sets from both hemispheres seem to indicate a lower tropopause temperature at the equator than at roughly ±15° latitude. More precisely, the spline fit indicates on average a maximum temperature of 280 K at 8 = 17° in the north and 267 K at 8 = 15° in the south. The number of profiles in the equatorial region, however, is definitely too small to draw any definite conclusion. [39] We cannot adequately compare the two hemispheres between the equator and about 8 = ±60° because of the lack of data in the northern hemisphere. Another uncertainty at lower latitudes comes from the weakness of the temperature inversion at the tropopause because the change in the measured temperature lapse rate is not as sharp as in the higher-latitude region. The tropopause altitude and temperature thus depend critically on the choice of the

7 of 19

E00B36

TELLMANN ET AL.: VENUS ATMOSPHERE FROM VERA

E00B36

Figure 6. Tropopause temperature as a function of latitude. (top left) Northern hemisphere (dots) and the resulting least median of squares (LMS) spline fit (dashed line). (bottom left) Southern hemisphere data (circles) and the resulting spline fit (solid line). (top right) Same as left, but sorted by OCC season and entry/exit. (bottom right) Superposition of northern (dots) and southern (circles) hemisphere data together with LMS spline fits (northern hemisphere, dashed line; southern hemisphere, solid line). stability threshold criterion. The latitudinal behavior of the tropopause temperature is virtually mirrored by the tropopause altitude shown in Figure 7. [40] The tropopause altitude rises from 59 km at low latitudes to about 62 km in the cold collar region, and then drops toward the near polar and polar region to a mean value of 58 km. The cubic spline fits for both hemispheres are slightly different, again mainly induced by the different spatial distribution of the data sets. [41] The northern polar region appears in some profiles warmer than the corresponding region in the south (Figure 6), but the cold collar region on both hemispheres are in good agreement. Large fluctuations in the thermal structure of this latitude range are induced by strong fluctuations of the temperature near the pole resulting from the rotating hot dipole structures. The polar dipoles, which follow a wave-2 dependence with a rotation period of 2.9 days in the north [Taylor et al., 1980] and 2.48 days in the south [Piccioni et al., 2007], are poleward of the cold collar and contrast with its predominantly wave-1 behavior [Taylor et al., 1980]. The VeRa tropopause altitudes in this region are in good

agreement with those reported by Kliore and Patel [1982] and Yakovlev et al. [1991] (Table 1). We note the positive bias of 0.2 km in our values resulting from the revised radius used here. These differences do not affect the

Table 1. Tropopause Conditions Region 8 8 8 8

= 0° – 40° = 40° – 55° = 55° – 77°  77°

8 8 8 8

= 0° – 40° = 40° – 55° = 55° – 77°  77°

8 of 19

a

Altitude (km)

Temperature (K)

Pressure (hPa)

Northern Hemisphere 58.9 ± 1.0 269.5 ± 8.5 291.0 ± 47.2 60.6a 244.7 190.7 61.5 ± 0.9 220.0 ± 8.4 149.7 ± 26.5 59.3 ± 1.5 228.5 ± 11.7 210.8 ± 63.2 Southern Hemisphere 59.3 ± 1.2 263.6 ± 10.6 270.0 ± 52.1 61.1 ± 1.2 241.5 ± 11.3 191.6 ± 40.0 62.3 ± 0.9 220.5 ± 8.1 141.3 ± 19.2 59.7 ± 1.3 228.2 ± 7.6 196.9 ± 45.5

Only one value.

Depth Inversion (K) 3.1 ± 1.3 N/A 20.9 ± 7.7 16.5 ± 6.1 3.0 ± 2.1 7.8 ± 3.7 19.9 ± 5.6 16.3 ± 5.4

E00B36

TELLMANN ET AL.: VENUS ATMOSPHERE FROM VERA

E00B36

Figure 7. Tropopause altitude as a function of latitude. (top left) Northern hemisphere (dots) and resulting spline fit (dashed line). (bottom left) Southern hemisphere data (circles) and LMS spline fit (solid line). (top right) Same as left, but now separated according to OCC season and entry/exit measurements. (bottom right) Superposition of northern (dots) and southern (circles) hemispheric data together with LMS spline fits (northern hemisphere, dashed line; southern hemisphere, solid line). Values are relative to 6051.8 radius. determination of the temperature and pressure values in Table 1. [42] The tropopause pressure as a function of latitude is shown in Figure 8. The same characteristics can be observed in the pressure differences as in the altitude differences. The significant differences in the spline fits are affected in an unknown way by the differences in the hemispheric data distribution. [43] The depth of the temperature inversion, defined as the temperature difference between the tropopause and the temperature maximum in the middle atmosphere above, shown in Figure 9, also suggests a strong latitudinal dependency. The inversions in the northern hemisphere are somewhat deeper than those in the southern cold collar region but, again, this impression may be caused by the limited set of data points in this highly variable region. Given a longitudinal open cold collar and a high variability in the observed inversion depths at fixed high latitudes, we

speculate that the cross section in latitude varies and is at its thickest opposite the opening. [44] The three occultation seasons cover 350 Earth days,
1.5 Venus years. No significant long-term temporal changes are observed in the thermal structure of the atmosphere within this period (Figures 6 –9, top right).

7. Latitudinal Variations in the Venus Atmosphere [45] Latitudinal variations dominate the structure of the lower and middle atmosphere of Venus. These are evident at an altitude of 30 km [Seiff, 1983] and are significant in the cloud layer (45– 70 km) and above, mainly in association with cyclostrophic balance of the high-velocity zonal winds [Newman et al., 1984]. [46] Figure 10 shows a temperature map based on 12 temperature profiles from the southern hemisphere obtained during the third occultation season during nighttime (day of

9 of 19

E00B36

TELLMANN ET AL.: VENUS ATMOSPHERE FROM VERA

E00B36

Figure 8. Tropopause pressure as function of latitude. See Figures 6 and 7 for explanation. year (DOY) 116– 146 2007). Some typical features of the Venusian atmosphere are apparent in this map. [47] The temperature in the troposphere from the equator to 60° latitude remains nearly constant for a constant pressure level below 2  102 hPa, 60-km altitude. In the polar region (j8j > 60°), the temperature drops significantly, for example, at the pressure level 1000 hPa (50 km) DT  40 K between 60° and 90°. The temperature at pressure levels above 30 hPa (70 km) increases with increasing latitude. The ‘‘cold collar’’ [Taylor et al., 1980] is evident at the pressure level 100 hPa (ca. 64 km) between 60° and 75°. [48] The characteristic poleward decrease in temperature on surfaces of constant pressure is apparent in Figure 11, which shows the temperature at the 1-bar level as a function of latitude for the three occultation seasons. The temperature is nearly constant equatorward of 50° latitude and decreases from 50° poleward. We see no significant differences between the three occultation seasons, nor between the two hemispheres at any pressure level in the troposphere (the differences between the mean temperature values of the three occultation seasons are not greater than 2 K (8 ±50°, p = 1 bar)). These results are consistent with the PV-ORO results [Kliore and Patel, 1982], suggesting that

there have been no significant long-term temporal changes in the four decades since 1967 in the lower atmosphere of Venus. [49] The most striking feature in Figure 10 is the presence of a strong inversion layer at a pressure level near 100 hPa (60– 70 km) that becomes apparent in the latitude range between 60 and 80° latitude. This inversion layer corresponds to the cold collar region first observed in the northern hemisphere by Pioneer Venus [Taylor et al., 1980; Kliore and Patel, 1982; Newman et al., 1984]. The cold collar was not seen in some southern hemisphere profiles from Pioneer Venus [Kliore and Patel, 1982] and was not detected in the three northern hemisphere height profiles retrieved from Magellan occultation data [Jenkins et al., 1994]. The Radio Science experiment on Venera-15 and -16, however, did observe the cold collar in the southern hemisphere [Yakovlev et al., 1991]. The Orbiter Infrared (OIR) experiment on Pioneer Venus described this inversion region as a ‘‘crescent-shaped collar region’’ [Taylor et al., 1980], which seems to have a solar-fixed component and is also influenced by spiral streaks from the hot dipole regions around the poles. [50] The dynamics supporting the cold collar inversions are not understood. The strong zonal winds in and below the

10 of 19

E00B36

TELLMANN ET AL.: VENUS ATMOSPHERE FROM VERA

E00B36

Figure 9. Depth of temperature inversion in the middle atmosphere (mesosphere) as a function of latitude. The depth of inversion is defined as the temperature difference between the tropopause and the maximum temperature in the middle atmosphere above this level. See Figures 6 and 7 for explanation. clouds in connection with the downwelling branch of the Hadley circulation likely have an association with the observed thermal behavior and particularly the transition to pronounced vortices around the poles. [51] Cooling of the atmosphere at 60-km altitude is affected by the rotating pair of infrared emission windows found at slightly higher latitudes than the cold collar structures. This, together with radiative absorption at higher levels, could be related to the strong inversion [Taylor et al., 1980]. The emission windows have been attributed to cloud depressions resulting from the downwelling Hadley circulation. [52] The collar region also divides the atmosphere vertically. Below the collar, the atmosphere cools with increasing latitude. Above, the temperature gradient is reversed (Figure 10). Again the origin and dynamics of this structure remain unknown. Diabatic heating or dynamically controlled processes could be responsible for the observed structures [Schubert et al., 1980]. At the altitude of the collar and closer to the pole the atmosphere is almost isothermal and also much warmer than the surrounding areas.

[53] Reconstructed orbit solutions for Venus Express have formal accuracies along the projected path in the plane of the sky of less than 100 m, which are more precise than earlier Venus missions. Uncertainties resulting from the orbit inaccuracy and further retrieval approximations, such as the assumption of spherical symmetry, lead to errors in the altitude of the retrieved occultation profiles that are at the same level of 100 m or less, which contributes to systematic errors in the retrieved profile parameters, especially the temperature. [54] Much weaker limitations on the altitude determination were reported by Kliore and Patel [1980] and Yakovlev et al. [1991], who indicated discrepancies of some kilometers. Yakovlev et al. [1991] report disagreements in the altitude assignment in the order of ±1 km in most cases up to ±5 km in certain orbits. Kliore and Patel [1980] observe inconsistencies of 2 – 3 km, although some orbits show differences by about 10 km. On the basis of a lapse rate of 10 K/km, this translates to temperature uncertainties of several kelvins in the lower atmosphere. [55] Systematic errors due to solar tidal distortions of the Venus atmosphere arising both from heating of the atmo-

11 of 19

E00B36

TELLMANN ET AL.: VENUS ATMOSPHERE FROM VERA

E00B36

Figure 10. Nighttime temperature in the southern hemisphere for the third occultation season. Local times range from 2230 at low latitudes to 0500 at high latitudes. Contour map is based on measurements from 12 occultations in the interval DOY 116-146 2007. sphere and the Sun’s gravity gradient are thought to be small but need further analysis. [56] For reasons of convenience and because the uncertainties associated with earlier measurements are not well documented, we choose to present our results preferably as a function of the pressure, even though in most cases we indicate the approximate altitude level. The agreement between the VeRa temperatures and previous results under

similar conditions and at the same pressure level is typically within a few kelvins. [57] In this connection, it should be mentioned that the reduction of the PV-ORO occultation measurements employed a different atmospheric gas constant in (1), on the basis of the accepted atmospheric composition at that time (see Appendix A). [58] Figure 12 shows the pressure at 60-km altitude versus latitude for VeRa, PV-ORO, and VIRA. The VeRa

Figure 11. Temperature at the 1-bar level (
50 km) as a function of latitude. Different symbols mark the different occultation seasons separated for ingress (entry) and egress (exit). 12 of 19

E00B36

TELLMANN ET AL.: VENUS ATMOSPHERE FROM VERA

E00B36

Figure 12. Pressure at the 60-km altitude. VeRa (solid dots), PVO (open dots), and VIRA model (dashed line). Compare text. VeRa data here are for a Venus radius of 6052 km for consistency with PVO and VIRA. points are derived from the available observations after the first three OCC seasons, which have been averaged over ten latitude bins before comparison with the PV-ORO data. Hemispheric differences are neglected. The VeRa results fall between the VIRA model and the ORO observations. The

difference between VeRa and VIRA at the 60 km level is in the order of 10 hPa at low latitudes and increases to 20 hPa at the higher latitudes. [59] Figure 13 shows a comparison of VeRa and PV-ORO profiles from both the northern and southern hemispheres,

Figure 13. Mean VeRa temperature profiles (solid lines) from three occultation seasons binned for 10 latitude ranges in both hemispheres. PVO mean temperature profiles from ORO data (dashed lines) are from Seiff et al. [1985]. 13 of 19

E00B36

TELLMANN ET AL.: VENUS ATMOSPHERE FROM VERA

E00B36

Figure 14. Temperature and pressure as a function of solar-fixed longitude. 0° is subsolar point; 90° is 0600; 180° is antisolar point; 270° is 1800. Data from latitude range 75° – 85°. averaged over 10° latitude bins. Although Figure 12 shows substantial differences in the pressures at fixed altitude, the overall agreement below the 100 hPa altitude is within ±5 K with maximum differences of 13 K. Overall, the VeRa results indicate a slightly warmer atmosphere at pressure levels > 1000 hPa (<50 km), in the equatorial and midlatitudes. The differences between VeRa and PV-ORO increase in the inversion region and the middle atmosphere at p 100 hPa where the discrepancies can be as large as 13 K (at 8 = 45–55° at p  100 hPa). At altitudes above roughly the 100 hPa level, the VeRa data are systematically colder by a few kelvins in nearly all latitude bins in the middle atmosphere. [60] After the first three occultation seasons the present VeRa data set contains only a limited number of profiles in the middle- and low-latitude range. The PV-ORO data set is based on about 123 profiles mainly resulting from the first and second occultation season [Kliore, 1985]. These measurements provide good spatial coverage in the high northern latitudes and the low southern latitudes. One reason for the observed differences between the data sets, aside from any differences resulting from the sampling bias, may be the increasing natural variability of the atmosphere above the tropopause. The presence of small-scale wavelike structures, for example, could lead to significant differences between the individual measurements.

8. Solar-Forced Structures [61] During the nominal mission of VEX, 32 profiles were obtained within the latitude ranges of ±75°85°. These provide the opportunity to study the influence of the Sun on the thermal structure of the planetary atmosphere. The particular latitude range selected here is based on the distribution of available observations and the need to have a sufficiently large sample to support meaningful

conclusions concerning the solar-driven variability of the atmosphere. The data were binned into 12 solar-fixed longitude ranges of 30° width, centered at 0°, 30°, 60°, etc. The solar-fixed longitude is the equivalent of local time, but expressed in degrees. The resulting temperature map is shown as a function of solar-fixed longitude and pressure (Figure 14). The subsolar point establishes the solar-fixed longitude of 0°, with dawn at 90°, midnight at 180°, and dusk at 270°. The temperature structure at the 100-hPa level (64 km) appears to be dominated by a wave-2 solar tide with an amplitude of 30 K the minimum aligned with the subsolar and the antisolar points. [62] Figure 15 shows a slice of the temperature in Figure 14, at a fixed pressure of 108.3 hPa (62 km), again as a function of solar-fixed longitude. The dots show the temperature values of the individual profiles while the triangles mark the values of the mean temperature profile in each solar-fixed longitude bin. A sinusoidal temperature variation over two periods is clearly evident (dashed curve). Wave-2 structures have been reported previously by the OIR experiment on PVO [Schofield and Taylor, 1983] who found indications of increasing wave-2 amplitudes in the high latitudes in this altitude range, but the infrared retrieval is unreliable near the cloud tops. [63] Other atmospheric dynamics may also play a role in this context. For example, the influence of the rotating polar dipole structures in this latitude region or other atmospheric variability, such as traveling waves or eddy motions, could be superimposed on the diurnal effect. Additional observations are needed to sort out these issues.

9. Comparison With the Venus International Reference Atmosphere [64] VIRA summarizes results obtained prior to 1985 by four missions to Venus [Seiff et al., 1985]. The model

14 of 19

E00B36

TELLMANN ET AL.: VENUS ATMOSPHERE FROM VERA

E00B36

Figure 15. Temperature at p = 0.1083 bar (62 km) versus solar-fixed longitude for latitudes in the range 8 = ±75 – 85°. Individual measurements are shown as dots; average values over solar-fixed longitude ranges of 30° are plotted as triangles. The dashed line shows an LMS sinusoidal fit to the data. describes the atmosphere from the surface to 100 km on the basis of the data from the Pioneer Venus Probes and Orbiter (PVO) and the Venera 10, 12 and 13 landers [Seiff et al., 1985]. [65] The VIRA-2 model [Moroz and Zasova, 1997; Zasova et al., 2006] incorporates the atmospheric data obtained after the publication of VIRA such as the vertical temperature profiles from the VEGA spacecraft, VEGA balloon measurements, temperature profiles of infrared spectrometry obtained by Venera-15 and -16, by radio occultation measurements with Magellan, and by Venera-15 and -16. [66] Given additional observations, the VIRA-2 model subdivides the atmosphere according to Venus local time. Diurnal variations monitored by the OIR experiment on Pioneer Venus [Taylor et al., 1980; Schofield and Taylor, 1983] include solar-related variations dominated by wave-2 structures. These diurnal variations, however, are superimposed on rotating structures such as the 5.9 d cold collar and the 2.7 d hot dipole structures, as well as other perturbations including gravity waves, eddies, and dynamical effects. [67] Figure 16 provides comparisons between the VeRa profiles, the VIRA model [Seiff et al., 1985], and the VIRA-2 model [Zasova et al., 2006] for different latitude ranges. Figure 16 (left) shows the average of at least three VeRa profiles compared with the VIRA and VIRA-2 models for the same range of solar-fixed longitude. Figure 16 (middle) shows the differences between the VeRa observations and the models as a function of pressure. Figure 16 (right) shows three individual VeRa profiles in comparison with the models. Typically, the temperature differences between the models and VeRa are less than 10 K, with the maximum deviations located near the altitude of the thermal inversion near the troposphere. Above an altitude of about 90 km (0.3 hPa) the

differences increase owing to the influence of the upper boundary condition on the profile retrieval and the increasing dynamical variability. The largest deviations between VIRA-2 and VeRa exceed 20 K; these occur at the altitude range of the tropopause in the polar regions. We note that these deviations are at the approximate altitude of the inversion layer. [68] The currently available maximum of at least three profiles in each range of latitude and solar-fixed longitude is insufficient to draw any final conclusions, however. In particular, the temperature profile at high latitudes shows a significant temporal variability induced by the rotating dipole and cold collar structures. These influences may be much stronger than the solar-related effects, and are likely to be superimposed on the diurnal effects in individual profiles.

10. Discussion [69] This study is based on VeRa occultation profiles from the first three VEX occultation seasons in 2006 and 2007. First results reported by Pa¨tzold et al. [2007] were based on 15 occultations from the first occultation season in 2006. The significant temperature differences between day and night for OCC-1 previously reported by Pa¨tzold et al. [2007] could not be confirmed for the following occultation seasons. Subsequent examination of the data reduction procedures led to the discovery that the previously reported large temperature differences [Pa¨tzold et al., 2007, Figure 3] were the result of an error in the computed observational geometry that propagated into the altitude assignments of the temperature profiles. Correction of this error, further small improvements in the retrieval algorithms, and reprocessing of all profiles have reduced the observed altitude differences to less than 1 km between the different occultation seasons; typical altitude differences are of the order of

15 of 19

E00B36

TELLMANN ET AL.: VENUS ATMOSPHERE FROM VERA

Figure 16. Comparison of VeRa measurements with VIRA and VIRA-2 models. (left) averaged VeRa profiles (solid lines with circles), VIRA (solid lines) and VIRA-2 (dashed lines). (middle) Differences between VeRa and VIRA (solid lines) and VeRa and VIRA-2 (dashed lines). (right) Three individual VeRa profiles (solid lines with symbols), VIRA (solid lines), and VIRA-2 (dashed lines). Plots are ordered by three ranges of latitude and solar-fixed longitude: (top) low-latitude region (lat < 30°), solarfixed longitude 20° –90°; (middle) midlatitudes (35° –55°), solar-fixed longitude 200° –270°; (bottom) polar latitudes (lat > 80°), solar-fixed longitude 90° –130°.

16 of 19

E00B36

E00B36

TELLMANN ET AL.: VENUS ATMOSPHERE FROM VERA

E00B36

Figure 17. Temperature distribution as a function of latitude and pressure for the first occultation season. The approximate altitude levels are also shown. (left) Daytime profiles from DOY 196 – 216 2006. (right) Nighttime profiles from DOY 216– 239 2006. Figure is revised version of Figure 3 from Pa¨tzold et al. [2007]. 0.2 km. In particular the reanalysis of OCC-1 has resolved the unexpectedly large day/night temperature variations reported earlier. [70] Figure 17 provides corrected results from the first occultation season. In addition to the correction of the geometry, the following other changes were included in the processing: (1) Every temperature value is assigned to the specific latitude and pressure level of the measurement point given by the projection of the closest approach of the radio link to the planetary surface. In the work by Pa¨tzold et al. [2007] each temperature profile was assigned to a single latitude value. (2) The profiles are now displayed as a function of log pressure. No significant local time differences are observed in the temperature-pressure profiles. The data sets are slightly different from those used by Pa¨tzold et al. [2007] owing to the refined latitude assignments of the profiles. [71] The improvements applied to Figure 17 are now incorporated in our standard data reduction procedures. We now find the temperature-pressure profiles from the three occultation seasons to be highly consistent among each other, as can be seen in the earlier sections of this paper.

11. Summary and Conclusions [72] The VeRa experiment sounded the Venus neutral atmosphere over three occultation seasons during the Venus Express nominal mission. In total, 118 profiles of neutral number density were retrieved over a meaningful range of local times. These were used to derive density, temperature, and pressure profiles with fairly uniform coverage of the southern hemisphere and high northern latitudes. [73] Comparing the results of the VeRa observations with those from PV-ORO, we find the latitudinal variations from

the two missions to be in general agreement. Overall, the VeRa temperatures are slightly lower than those from PV-ORO by differences that do not exceed a few kelvins, with the exceptions noted above. A comparison of the pressure distributions at 60-km altitude, VeRa vis-a-vis VIRA, reveals small deviations, with a difference of about 10 hPa in the lower latitudes, but with larger deviations in the vicinity of the poles (25 hPa). [74] The high vertical resolution of the VeRa radio occultation profiles offers the opportunity to compare the observed temperature gradient with the adiabatic lapse rate in order to investigate the stability of the atmosphere. These studies show an extremely stable atmosphere in the topmost cloud layer and above, a nearly adiabatic behavior in the middle cloud, and a stably stratified atmosphere below the middle clouds. These results are consistent with former results from the Pioneer Venus probes [Seiff et al., 1980] and Magellan [Hinson and Jenkins, 1995], which show only very small altitude regions of instability in the middle cloud. The unstable regions are extremely thin in the VeRa profiles. The static stability shows a latitudinal dependency corresponding to the more pronounced inversion structures in the higher-latitude range. The nearly adiabatic lapse rate in the middle cloud at altitudes of
50 –55 km is most obvious at midlatitudes. The near-polar regions show several small-scale perturbations and the stable layer below the middle cloud is almost undetectable. [75] The static stability is used to determine the variability of the tropopause with latitude. Referring to Figures 6 and 7 (bottom, left), any differences appear to be concentrated in the lower midlatitude and within a few degrees of the poles. Additional observations are needed to sort this out. [76] The VeRa results are consistent, in an overall sense, with the VIRA and VIRA-2 models as the temperature differences at the same pressure seldom exceed 10 K. From

17 of 19

E00B36

TELLMANN ET AL.: VENUS ATMOSPHERE FROM VERA

the limited data available at the current stage it is not possible to determine whether or not overall the VeRa data agree better with the VIRA or the VIRA-2 model. In particular, it is not clear that VIRA-2 is an improvement over VIRA. At present the VeRa observations are too limited to draw final conclusions about a possible improvement from the implementation of diurnal effects. Additional data are required in order to determine the extent to which temporal effects resulting from rotating structures, eddies or gravity waves are affecting the VeRa observations. [77] A first investigation of the latitude range between ±75° and 85° indicates the presence of a solar-induced effect above the tropopause. VeRa observations imply the presence of a wave-2 structure with the coldest temperatures located at the subsolar and antisolar points. Further observations are required to refine and extend these preliminary results.

Appendix A [78] Kliore and Patel [1980, 1982] used an atmospheric composition of 96.0% CO2 and 4.0% N2. Using the refractive indices from Essen and Froome [1951] this leads to an atmospheric constant of C1 = 1.30943  106 Kms2/kg. The implementation of this C1 value in our retrieval procedure results in temperature changes less than 2 K at all altitude levels in comparison with our usual results. The VeRa temperature profiles would be colder by 2 K if we would use the atmospheric composition used by Kliore and Patel [1980, 1982]. Therefore, differences in the atmospheric composition cannot explain the systematic bias between the VeRa results and the ORO results in the middle atmosphere.

Appendix B [79] The occultation results presented here are derived from observations of a series of measurements of the radio signal traveling between the Venus Express spacecraft and the ground station; for VeRa the measurement interval typically is 0.5 s. During this interval, the observed occultation ray always remains in the instantaneous plane defined by the positions of the spacecraft and Earth receiver with respect to Venus and the direction of local gradient of atmospheric refractivity, essentially the direction to the center of Venus. The altitude and lateral position of the ray varies with changes in the locations of the spacecraft and receiver. The atmosphere sampled by the occultation signal corresponds to the geometric area swept out during the measurement interval. Typical changes in the location of the ray during the measurement interval are of the order of kilometers in the horizontal, and at most a few hundred meters in the vertical, decreasing rapidly with lowering altitude. As the average horizontal atmospheric gradient is very small, this generally is not of concern except for measurements made at very large bending angles. Vertical gradients are quite large, so a reported measurement must be considered as an average over the altitude range seen by the ray during the measurement interval. [80] More difficult to visualize is the manner in which the atmosphere is represented in profile retrieved by Abelian inversion. Using ray theory, Fjeldbo et al. [1971, Appendix C] showed that the vertical refractivity profile of a spherically symmetric atmosphere could be retrieved by Abelian inver-

E00B36

sion from measurement of the ray asymptotes obtained by radio occultation. That is, it is possible to obtain the value of refractive index at the lowest level of each ray in a continuous sequence of observed rays. Given the variation in refractive index one can then calculate the corresponding state of the atmosphere from hydrostatic equilibrium and the gas laws. Lipa and Tyler [1979] provided an analysis of statistical error for Mariner 10 radio occultation measurements of Venus atmosphere. More recently, Ahmad and Tyler [1998] found the inversion kernel for atmospheric occultation and used this to calculate the weighting function for errors due to departures from spherical symmetry. A perturbation theory of systematic errors Abelian inversion due to systematic departures from spherical symmetry in ‘‘thick’’ and ‘‘thin’’ atmospheres is given by Ahmad and Tyler [1999], with results using the atmospheres of Earth and Mars as example cases. The Mars example is comparable to the Venus case above about the 1-bar pressure level; a detailed comparison should include a correction for the electrical and mass differences between the predominately nitrogen and carbon dioxide gases. [81] Acknowledgments. This paper presents results of a research project partially funded by the Deutsches Zentrum fu¨r Luft- und Raumfahrt (DLR) under contract 50 OV 0601. Our investigation could not have been successful without the efforts of the ESA Venus Express Science and Mission Operations teams. We particularly appreciate the support of the VEX Project scientist H. Svedhem (ESTEC) and VEX Mission Operations Manager F. Jansen (ESOC). The VeRa experiment has benefited greatly from the continued support of the NASA Deep Space Network. In this respect, it is a pleasure to thank T. W. Thompson and D. P. Holmes (JPL) for their special efforts.

References Ahmad, B., and G. L. Tyler (1998), The two-dimensional resolution kernel associated with retrieval of ionospheric and atmospheric refractivity profiles by Abelian inversion of radio occultation phase data, Radio Sci., 33, 129 – 142, doi:10.1029/97RS02762. Ahmad, B., and G. L. Tyler (1999), Systematic errors in atmospheric profiles obtained from Abelian inversion of radio occultation data: Effects of large-scale horizontal gradients, J. Geophys. Res., 104, 3971 – 3992, doi:10.1029/1998JD200102. Bougher, S. W., D. M. Hunten, and R. J. Philips (Eds.) (1997), Venus II: Geology, Geophysics, Atmosphere, and Solar Wind Environment, Univ. of Ariz. Press, Tucson. Counselman, C. C., III, S. A. Gourevitch, R. W. King, G. B. Loriot, and R. G. Prinn (1979), Venus winds are zonal and retrograde below the clouds, Science, 205, 85 – 87, doi:10.1126/science.205.4401.85. Eshleman, V. R. (1973), The radio occultation method for the study of planetary atmospheres, Planet. Space Sci., 21, 1521 – 1531, doi:10.1016/00320633(73)90059-7. Essen, L., and K. D. Froome (1951), The refractive indices and dielectric constants of air and its principal constituents at 24,000 Mc/s, Proc. Phys. Soc. London, Sect. B, 64, 862 – 875, doi:10.1088/0370-1301/64/10/303. Fjeldbo, G., A. J. Kliore, and V. R. Eshleman (1971), The neutral atmosphere of Venus as studied with the Mariner V radio occultation experiment, Astron. J., 76, 123 – 140, doi:10.1086/111096. Ha¨usler, B., et al. (2006), Radio science investigations by VeRa onboard the Venus Express spacecraft, Planet. Space Sci., 54, 1315 – 1335, doi:10.1016/j.pss.2006.04.032. Ha¨usler, B., et al. (2007), Venus atmospheric, ionospheric, surface and interplanetary radio-wave propagation studies with the VeRa Radio Science experiment, Eur. Space Agency Spec. Publ., ESA SP 1295, 30 pp. Hinson, D. P., and J. M. Jenkins (1995), Magellan radio occultation measurements of atmospheric waves on Venus, Icarus, 114, 310 – 327, doi:10.1006/icar.1995.1064. Howard, H. T., et al. (1974), Venus: Mass, gravity field, atmosphere and ionosphere as measured by the Mariner 10 dual-frequency radio system, Science, 183, 1297 – 1301, doi:10.1126/science.183.4131.1297. Hunten, D. M., L. Colin, T. M. Donahue, and V. I. Moroz (1983), Venus, Univ. of Ariz. Press, Tucson. Jenkins, J. M., P. G. Steffes, D. P. Hinson, J. D. Twicken, and G. L. Tyler (1994), Radio occultation studies of the Venus atmosphere with the

18 of 19

E00B36

TELLMANN ET AL.: VENUS ATMOSPHERE FROM VERA

Magellan spacecraft: 2. Results from the October 1991 experiments, Icarus, 110, 79 – 94, doi:10.1006/icar.1994.1108. Kliore, A. (1985), Recent results on the Venus atmosphere from Pioneer Venus radio occultations, Adv. Space Res., 5(9), 41 – 49, doi:10.1016/ 0273-1177(85)90269-8. Kliore, A., and I. R. Patel (1980), Vertical structure of the atmosphere of Venus from Pioneer Venus Orbiter radio occultations, J. Geophys. Res., 85, 7957 – 7962, doi:10.1029/JA085iA13p07957. Kliore, A., and I. R. Patel (1982), Thermal structure of the atmosphere of Venus from Pioneer Venus radio occultations, Icarus, 52, 320 – 334, doi:10.1016/0019-1035(82)90115-4. Kliore, A., G. S. Levy, D. L. Cain, G. Fjeldbo, and S. I. Rasool (1967), Atmosphere and ionosphere of Venus from the Mariner V S-band radio occultation measurement, Science, 158, 1683 – 1688, doi:10.1126/ science.158.3809.1683. Kolosov, M. A., O. I. Yakovlev, A. I. Efimov, A. G. Pavelyev, and S. S. Matyugov (1979), Radio occultation of the Venusian atmosphere and bistatic radiolocation of the surface of Venus using the Venera-9 and Venera-10 satellites, Radio Sci., 14, 163 – 173, doi:10.1029/ RS014i001p00163. Limaye, S. S., and V. E. Suomi (1981), Cloud motions on Venus: Global structure and organization, J. Atmos. Sci., 38, 1220 – 1235, doi:10.1175/ 1520-0469(1981)038<1220:CMOVGS>2.0.CO;2. Lipa, B., and G. L. Tyler (1979), Statistical and computational uncertainties in atmospheric profiles from radio occultation: Mariner 10 at Venus, Icarus, 39, 192 – 208, doi:10.1016/0019-1035(79)90163-5. Moroz, V. I. (2002), Studies of the atmosphere of Venus by means of spacecraft: Solved and unsolved problems, Adv. Space Res., 29(2), 215 – 225, doi:10.1016/S0273-1177(01)00571-3. Moroz, V. I., and L. V. Zasova (1997), VIRA-2: A review of inputs for updating the Venus International Reference Atmosphere, Adv. Space Res., 19(8), 1191 – 1201, doi:10.1016/S0273-1177(97)00270-6. Newman, M., G. Schubert, A. J. Kliore, and I. R. Patel (1984), Zonal winds in the middle atmosphere of Venus from Pioneer Venus radio occultation data, J. Atmos. Sci., 41, 1901 – 1913, doi:10.1175/1520-0469 (1984)041<1901:ZWITMA>2.0.CO;2. Nicholson, P. D., and D. O. Muhleman (1978), Independent radio-occultation studies of Venus’ atmosphere, Icarus, 33, 89 – 101, doi:10.1016/00191035(78)90026-X. Pa¨tzold, M., et al. (2007), The structure of Venus’ middle atmosphere and ionosphere, Nature, 450, 657 – 660, doi:10.1038/nature06239. Piccioni, G., et al. (2007), South-polar features on Venus similar to those near the north pole, Nature, 450, 637 – 640, doi:10.1038/nature06209. Ragent, B., and J. Blamont (1980), Structure of the clouds of Venus: Results of the Pioneer Venus nephelometer experiment, J. Geophys. Res., 85, 8089 – 8105, doi:10.1029/JA085iA13p08089. Rossow, W. B., A. D. Del Genio, S. S. Limaye, L. D. Travis, and P. H. Stone (1980), Cloud morphology and motions from Pioneer Venus images, J. Geophys. Res., 85, 8107 – 8128, doi:10.1029/JA085iA13p08107. Schofield, J. T., and F. W. Taylor (1983), Measurements of the mean, solarfixed temperature and cloud structure of the middle atmosphere of Venus, Q. J. R. Meteorol. Soc., 109, 57 – 80, doi:10.1002/qj.49710945904. Schubert, G., et al. (1980), Structure and circulation of the Venus atmosphere, J. Geophys. Res., 85, 8007 – 8025, doi:10.1029/JA085iA13p08007. Seiff, A. (1983), Thermal structure of the atmosphere of Venus, in Venus, edited by D. M. Hunten et al., pp. 215 – 279, Univ. of Ariz. Press, Tucson. Seiff, A., D. B. Kirk, R. E. Young, R. C. Blanchard, J. T. Findlay, G. M. Kelly, and S. C. Sommer (1980), Measurements of thermal structure and

E00B36

thermal contrasts in the atmosphere of Venus and related dynamical observations: Results from the four Pioneer Venus probes, J. Geophys. Res., 85, 7903 – 7933, doi:10.1029/JA085iA13p07903. Seiff, A., J. T. Schofield, A. J. Kliore, F. W. Taylor, and S. S. Limaye (1985), Models of the structure of the atmosphere of Venus from the surface to 100 kilometers altitude, Adv. Space Res., 5(11), 3 – 58. Stanford Mariner Group (1967), Venus: Ionosphere and atmosphere as measured by dual-frequency radio occultation of Mariner V, Science, 205, 14 – 19. Steffes, P. G., J. M. Jenkins, R. S. Austin, S. W. Asmar, D. T. Lyons, E. H. Seale, and G. L. Tyler (1994), Radio occultation studies of the Venus atmosphere with the Magellan spacecraft, Icarus, 110, 71 – 78, doi:10.1006/icar.1994.1107. Suomi, V. E., and S. S. Limaye (1978), Venus: Further evidence of vortex circulation, Science, 201, 1009 – 1011, doi:10.1126/science.201.4360.1009. Svedhem, H., D. V. Titov, F. W. Taylor, and O. Witasse (2007), Venus as a more Earth-like planet, Nature, 450, 629 – 632, doi:10.1038/nature06432. Taylor, F. W. (2002), Some fundamental questions concerning the circulation of the atmosphere of Venus, Adv. Space Res., 29(2), 227 – 231, doi:10.1016/S0273-1177(01)00572-5. Taylor, F. W. (2006), Venus before Venus Express, Planet. Space Sci., 54, 1249 – 1262, doi:10.1016/j.pss.2006.04.031. Taylor, F. W., D. J. Diner, L. S. Elson, D. J. McCleese, J. V. Martonchik, J. Delderfield, S. P. Bradley, J. T. Schofield, J. C. Gille, and M. T. Coffey (1979), Temperature, cloud structure and dynamics of Venus middle atmosphere by infrared remote sensing from Pioneer orbiter, Science, 205, 65 – 67, doi:10.1126/science.205.4401.65. Taylor, F. W., et al. (1980), Structure and meteorology of the middle atmosphere of Venus: Infrared remote sensing from the Pioneer Orbiter, J. Geophys. Res., 85, 7963 – 8006, doi:10.1029/JA085iA13p07963. Titov, D., et al. (2001), Venus Express: An orbiter for the study of the atmosphere, the plasma environment, and the surface of Venus, Rep. ESA-SCI (2001) 6, Eur. Space Agency, Paris. Yakovlev, O. I., S. S. Matyugov, and V. N. Gubenko (1991), Venera-15 and -16 middle atmosphere profiles from radio occultations: Polar and nearpolar atmosphere of Venus, Icarus, 94, 493 – 510, doi:10.1016/00191035(91)90243-M. Young, R. E., R. L. Walterscheid, G. Schubert, A. Seiff, V. M. Linkin, and A. N. Lipatov (1987), Characteristics of gravity waves generated by surface topography on Venus: Comparison with the VEGA balloon results, J. Atmos. Sci., 44, 2628 – 2639, doi:10.1175/1520-0469 (1987)044<2628:COGWGB>2.0.CO;2. Zasova, L. V., V. I. Moroz, V. M. Linkin, I. V. Khatuntsev, and B. S. Maiorov (2006), Structure of the Venusian atmosphere from surface up to 100 km, Cosmic Res.. Engl. Transl., 44(4), 364 – 383, doi:10.1134/ S0010952506040095. 

M. K. Bird, Argelander Institut fu¨r Astronomie, Universita¨t Bonn, Bonn D-53121, Germany. B. Ha¨usler, Institut fu¨r Raumfahrttechnik, Universita¨t der Bundeswehr Mu¨nchen, Neubiberg D-85577, Germany. M. Pa¨tzold and S. Tellmann, Abteilung Planetenforschung, Rheinisches Institut fu¨r Umweltforschung, Universita¨t zu Ko¨ln, Ko¨ln D-50931, Germany. G. L. Tyler, Department of Electrical Engineering, Stanford Universtity, Stanford, CA 94305-4020, USA.

19 of 19

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, E00B13, doi:10.1029/2008JE003108, 2008

for

Full Article

Evidence for anomalous cloud particles at the poles of Venus C. F. Wilson,1 S. Guerlet,2 P. G. J. Irwin,1 C. C. C. Tsang,1 F. W. Taylor,1 R. W. Carlson,3 P. Drossart,2 and G. Piccioni4 Received 30 January 2008; revised 25 May 2008; accepted 24 July 2008; published 4 November 2008.

[1] An analysis of near-infrared emissions on the nightside of Venus observed by the

Visible and Infrared Thermal Imaging Spectrometer (VIRTIS) instrument on board Venus Express reveals anomalous cloud particles in the polar regions of Venus. These anomalous particles are found within the centers of polar vortices at both poles and are either larger or different in composition from those elsewhere in the planet. We find no persistent latitudinal variation in cloud properties at low to midlatitudes, nor do we find asymmetry between the southern and northern hemispheres. These findings arise from analysis of the relative brightness of 1.74 and 2.30 mm infrared radiation thermally emitted from the deep atmosphere of Venus. Larger cloud particles cause relatively more attenuation at 2.30 mm than at 1.74 mm, so we use a ‘‘size parameter,’’ m = (I1.74mm)/(I2.30mm)0.53, as a proxy for particle size. This methodology follows that of Carlson et al. (1993), supported by new radiative transfer modeling. Citation: Wilson, C. F., S. Guerlet, P. G. J. Irwin, C. C. C. Tsang, F. W. Taylor, R. W. Carlson, P. Drossart, and G. Piccioni (2008), Evidence for anomalous cloud particles at the poles of Venus, J. Geophys. Res., 113, E00B13, doi:10.1029/2008JE003108.

1. Introduction [2] The clouds of Venus have been extensively studied, by means of spacecraft entry probes, orbiter and flyby missions, and ground-based observation (see comprehensive reviews by Esposito et al. [1983, 1997]). Entry probes revealed that almost all particulate matter is located in a series of cloud layers starting at 45 km altitude, stretching up to 65 – 70 km. Data from the Pioneer Venus lander particle size spectrometer (LCPS) instrument [Knollenberg and Hunten, 1980] were interpreted as indicating four populations of particles. Mode 1 particles have a mean radius of 0.3 mm, and make up the bulk of the upper cloud layers. The larger mode 2 particles (mean radius 1.0 mm) make up the particulate mass in the upper clouds, while the slightly larger mode 20 particles (mean radius 1.4 mm) are found in the middle and lower clouds. The scattering properties of the mode 2 and 20 particles indicate that they are composed of liquid H2SO4/H2O mixture, with 75– 85% H2SO4 by weight [Esposito et al., 1983]. [3] Mystery still surrounds the large mode 3 particles found in the base of the clouds at altitudes 45– 52 km. The original analysis of LCPS data concluded that these must be crystalline particles [Knollenberg and Hunten, 1980]; however, a later study [Toon et al., 1984] concluded that the mode 3 particles could indeed be composed of liquid 1

Department of Physics, Oxford University, Oxford, UK. Laboratoire d’Etudes Spatiales et d’Instrumentation en Astrophysique, Observatoire de Paris, Meudon, France. 3 Jet Propulsion Laboratory, California Institute of Technology, Pasadena, California, USA. 4 Istituto di Astrofisica Spaziale e Fisica Cosmica, Rome, Italy. 2

Copyright 2008 by the American Geophysical Union. 0148-0227/08/2008JE003108$09.00

sulfuric acid droplets, once possible instrumental errors were taken into account. [4] Although it is clear that sulfuric acid comprises the majority constituent of the mode 2 and 20 cloud particles and possibly the mode 3 particles as well, there is also evidence that minority constituents are present. X-ray fluorescence measurements on the Vega probes found significant abundances of chlorine, phosphorus, and iron in lower cloud particles [Andreychikov et al., 1987]; an analysis from Krasnopolsky [1989] concluded that this suggested that the clouds contained compounds including Fe2Cl6, H3O4, and Al2Cl6. For the present work, though, we have ignored these minority constituents and assumed clouds comprising of 85 wt % H2SO4 to 15 wt % H2O, copying the modal size distributions as used by Grinspoon et al. [1993], as shown in Table 1. [5] In this work we take advantage of the near-infrared spectral window regions at wavelengths of 1.0 –2.5 mm, in which thermal emission from the deep atmosphere of Venus escapes through the clouds to space. These window regions exist in spectral regions between broad CO2 and H2O absorption features. The radiation is attenuated by a combination of gaseous absorption and scattering as it passes upward through the clouds to space; therefore the optical thickness and scattering properties of the clouds both have a significant effect on the observed radiation, as will be shown. Initially discovered fortuitously by Allen and Crawford [1984], the near-infrared emission windows have been extensively exploited in ground-based observations [Be´zard et al., 1990; Pollack et al., 1993], as well as by the Galileo Near Infrared Mapping Spectrometer (NIMS) during that spacecraft’s Venus flyby in 1990 [Carlson et al., 1991]. However, Venus Express is the first Venus orbiter equipped with a spectrometer capable of observing at these wavelengths.

E00B13

1 of 12

WILSON ET AL.: ANOMALOUS CLOUD PARTICLES AT VENUS POLES

E00B13

Table 1. Cloud Parameters Used in This Worka Mode Mode Mode Mode

1 2 20 3

Mean Radius r (mm)

Variance s (mm)

Altitude Range (km)

0.3 1.0 1.4 3.65

1.56 1.29 1.23 1.28

48 – 84 57 – 84 48 – 57 48 – 57

a Parameters are as per Grinspoon et al. [1993]. Unless otherwise noted, all cloud particles are assumed to be composed of 85% H2SO4 15% H2O.

[6] Radiative transfer modeling has been quite successful at replicating the observed spectra of these nightside emissions [Pollack et al., 1993; Grinspoon et al., 1993; Be´zard et al., 1990; Tsang et al., 2008a]. However, while the spectrally narrow near-IR window regions include enough spectral features to allow unambiguous mapping of several constituents of the deep atmosphere, they do not provide much information on cloud composition, because variations in cloud composition tend to cause broad changes in the spectra which cannot be observed within the narrow spectral range of a single near-IR transmission window. [7] All the thermal emissions in all of the window regions are attenuated as they pass upward through the clouds. This attenuation arises both because of absorption by cloud particles themselves (nonconservative scattering), and by gaseous absorption along the path taken by the photons. Because the particles are liquid and thus spherical, and because the wavelength of the light is of the same order as the radius of the particles, one can assume Mie scattering for the calculation of the scattering parameters. The single scattering albedo and extinction cross sections for the different modes are shown in Figure 1. The first thing to note is that the single scattering albedo, w0, is extremely close to 1 for all these particles at wavelengths below 2.5 mm, particularly at the shorter wavelength windows near 1.0 mm. The optical depth of the clouds is typically very high (t  30 as defined at l = 0.63 mm, in the model

E00B13

cloud defined by Pollack et al. [1993]), so the thermal emissions undergo a large number of scattering events as they pass upward through the clouds; it is only due to the high albedo of the H2SO4 droplets at wavelengths <2.5 mm that the light escapes through the cloud layer. [8] In Mie scattering, the extinction cross section exhibits a peak at wavelengths similar to the radius of the particle (lpeak  r). As can be seen in Figure 1, mode 20 particles have a significantly higher scattering cross section than the smaller mode 2 particles at l = 2.30 mm, while there is little difference at l = 1.74 mm, while the single scattering albedo for the two modes is almost the same. Therefore a cloud composed entirely of mode 20 particles will attenuate more of the 2.30 mm radiance than a cloud with the same optical depth (t 0.63mm) composed of the smaller mode 2 particles. Carlson et al. [1993] (hereinafter referred to as C93) proposed that the particle size was the dominant factor affecting the 1.74/2.30 mm ratio, and that this ratio could thus be used to infer information about the mean particle size. C93 found that the relationship between 1.74 and 2.30 mm radiances was well described by

m ¼ I1:74mm = I2:30mm P0:53 ;

ð1Þ

where m was defined as a ‘‘branching parameter’’ or ‘‘size parameter.’’ Radiative transfer modeling was used to show that larger values of m were associated with clouds consisting of larger particles. The exponent 0.53 is chosen empirically to fit the data. Its value is diagnostic of the relative importance of different mechanisms of attenuation in the cloud, and so, as we show in section 5, it should really vary with optical depth. For the present purposes (and for consistency with C93), though, we shall continue to assign to it a constant value of 0.53. [9] In sections 3 and 4, we use this approach as defined by C93 to evaluate the size parameter m for a range of VIRTIS data, and examine its spatial variation. In section 5,

Figure 1. Variation with wavelength of (a) the extinction cross section Qext, normalized to its value at 0.63 mm and (b) the single-scattering albedo w0 of cloud particles. Note that because the single scattering albedo w is close to 1, the data are presented as (1  w0) for clarity. The curves are calculated assuming Mie scattering, assuming modal size distributions and particle composition as defined in Table 1. 2 of 12

WILSON ET AL.: ANOMALOUS CLOUD PARTICLES AT VENUS POLES

E00B13

Table 2. Summary of Observations Used Observation VI0073_03 VI0077_03 VI0085_01 VI0094_01 VI0096_01 VI0082_18 VI0086_19 VI0113_15 VI0146_15 VI0148_15 VI0150_15 VI0152_15

Date Obtained 2 Jul 6 Jul 14 Jul 23 Jul 25 Jul 12 Jul 16 Jul 12 Aug 14 Sep 16 Sep 18 Sep 20 Sep

2006 2006 2006 2006 2006 2006 2006 2006 2006 2006 2006 2006

Minimum Latitude

Maximum Latitude

Exposure Time (s)

88.7° 86.7° 83.8° 80.4° 79.7° 12.5° 15.4° 2.1° 12.9° 13.0° 13.0° 13.0°

17.7° 28.3° 10.6° 6.5° 5.5° 89.9° 89.9° 89.6° 90.0° 90.0° 90.0° 90.0°

3.3 3.3 3.3 3.3 3.3 3.3 3.3 3.3 3.3 3.3 3.3 3.3

E00B13

default operation mode for Venus Express is to keep the field of view of VIRTIS pointed at the nadir during the pericenter pass. This results in a 1-D ‘‘slice’’ through the northern hemisphere, obtained at near-nadir viewing geometry.

3. Southern Hemisphere Images

we present new radiative modeling results which serve both to confirm broadly the previous interpretation of the size parameter and to examine its sensitivity to the vertical distribution of the cloud.

2. Measurements [10] The Venus Express spacecraft has been in orbit around Venus since 11 April 2006 [Svedhem et al., 2007]. Its orbit is near polar, with a 24-h period, and highly elliptical, with a periapsis altitude of 200 – 400 km above the northern hemisphere and an apoapsis altitude of some 66,000 km. [11] The spacecraft carries a number of remote sensing instruments including the Visible and Infrared Thermal Imaging Spectrometer (VIRTIS), which covers the wavelength range 0.27 to 5.19 mm. The VIRTIS instrument is split into two main subsystems, the High Resolution Subsystem, VIRTIS-H, and the Mapper subsystem, VIRTIS-M. The VIRTIS-H subsystem is an Echelle spectrometer covering the 1.84 to 4.99 mm range at a spectral resolution of 1 nm, with an instantaneous FOV of 0.45  2.25 mrad. The VIRTIS-M subsystem, has an IFOV of 0.25  64 mrad sampled by 256 rows of a CCD array, which can be scanned to achieve generate an image covering 64  64 mrad, with an angular resolution of 0.25  0.25 mrad. At an apogee of 66,000 km, Venus’ disc extends approximately 180 mrad and thus a complete image of Venus’ disc may be generated by mosaicking 3  3 VIRTIS-M observations. The VIRTIS-M subsystem is itself split into two components: one component covers the 0.27 to 1.0 mm wavelength range with a spectral resolution of 2 nm, while an infrared component covers the 1.05 to 5.19 mm range at a lower spectral resolution of 10 nm. [12] In this paper we have used the M infrared channel due to its wide spatial coverage. A limited number of observations are discussed in this work; they are outlined in Table 2. Because of the orbital geometry, VIRTIS performs imaging almost exclusively of the southern hemisphere. When the spacecraft is over the northern hemisphere, it is at low altitude (<10,000 km) leading to a small field of view for VIRTIS, and its tangential velocity is too great to allow 2-D imaging using the scan mirror (‘‘pushbroom’’ imaging can be used instead, but is not optimal due to the rapidly changing altitude of the spacecraft). When over the northern hemisphere, then, the

[13] Figure 2 shows a typical VIRTIS image of the southern high latitudes (VIORB0096_01) at 1.74 mm and 2.30 mm. Both of these wavelengths are window regions, so the images show thermal emission from the deep atmosphere (15– 30 km altitude for the 1.74 mm window and 26– 45 km for the 2.30 mm window [Taylor et al., 1997]). The contrast in the images is due to attenuation as the radiation passes upward through the clouds, so bright regions of the image correspond to relatively cloud-free areas. It can immediately be seen that the 1.74 mm and 2.30 mm values show similar patterns, but that the 2.30 mm radiance is much more attenuated at the polar region. The purpose of this paper is to explain why this differential attenuation occurs, and what this can tell us about the cloud properties. [14] To process the data, we first apply a limb-darkening correction, using equations copied directly from C93 (these were based on a radiative transfer model of Kamp and Taylor [1990], validated by using data from Galileo NIMS): 0 I1:74 ¼ I1:74 ½0:316 þ 0:685 cos qe 1 0 I2:30 ¼ I2:30 ½0:232 þ 0:768 cos qe 1

ð2Þ

Once corrected for limb darkening, we create a ‘‘correlation plot’’ by plotting 2.30 mm radiances on the x axis against 1.74 mm radiances on the y axis, as shown in Figure 3. We

Figure 2. Observation VIORB0096_01 for (a) R1.74, (b) R2.30, and (c) R3.50 and (d) the size parameter m calculated using equation (1). This is a near-polar view of the southern hemisphere; the south pole itself is at the bottom of the image.

3 of 12

E00B13

WILSON ET AL.: ANOMALOUS CLOUD PARTICLES AT VENUS POLES

E00B13

Figure 3. Correlation plots for the data of Figure 1 (observation VI0096_01) in (a) linear radiance scales and (b) logarithmic radiance scales. The dashed lines represent lines of constant size parameter m as defined by equation (1).

note that main population of points tends to fall along welldefined ‘‘branches.’’ If plotted on logarithmic axes as in Figure 3b, it can be seen that the main branches are well described by the relationship (I1.74mm) = m  (I2.30mm)0.53, where m is a size parameter as described in equation (1). Any given branch encompasses both regions of high 1.74 mm radiance (corresponding to thin cloud) and low radiance (corresponding to regions of thick cloud). [15] From the data of Figure 3, we can see that there is a very distinctly separate branch in the correlation plot above and to the left of the main branch, indicating that it has a larger size parameter m. Turning to the data of Figure 2, a map of the size parameter is shown in Figure 2d. It can be seen that the area of increased size parameter is found very close to the pole, in a region colocated with the core of polar vortex. [16] The region we refer to as the core of the polar vortex can most easily be identified by using images at 3.5 mm or 5 mm (see, e.g., Figure 2c). Nightside images at these wavelengths show thermal emission from the cloud tops; the bright region at the pole therefore indicates a higher cloud top temperature, which may indicate lower or thinner clouds due to downwelling. This polar hot spot in the thermal infrared is often elliptical or hourglass shaped and thus is often known as the polar dipole, although VIRTIS images have revealed that its shape is variable [Piccioni et al., 2007]. [17] Figure 4 shows side-by-side plots of the size parameter m and the 3.5 mm radiance for four more observations of the southern hemisphere. In two of the observations (VI0077_03 and VI0094_01), it can clearly be seen that the size parameter again increases sharply at the edge of the polar vortex. In the other two images (VI0073_03 and VI0085_01), one can again see an increase in the size parameter at the edge of the vortex core. However, the calculated size parameter then falls to a minimum value in the core of the vortex. This unexpected behavior can be explained in part by looking at the correlation plot shown for VI0085_01 in Figure 5. This correlation plot shows that the minimum size parameter occurs where the 1.74 and

2.30 mm radiances are at their least. At the time of writing, detailed radiometric error calculations are not yet available for the VIRTIS data; however, the signal to random noise ratio here is of the order of 10 and 50 for the 1.74 and 2.30 mm radiances, respectively, so the calibration of the data seems to be robust. As we show in section 5, when we present radiative transfer calculations, it is also possible that the simple size parameter formulation is no longer appropriate in regions of thickest cloud. [18] To highlight once again the latitude variability of the size parameter, we have plotted the zonally averaged size parameter as a function of latitude in Figure 6. Figure 6 clearly shows that there is little variability of the size parameter found outside the polar region.

4. Northern Hemisphere Data [19] The variation of size parameter with latitude was examined further using northern hemisphere data. As discussed in section 2, the northern hemisphere observations are obtained during the pericenter pass when the spacecraft is traveling fast and low above the planet. Therefore the data consist not of images but rather as a series of spectra, in nadir viewing geometry, all grouped along one meridian. The large range of latitudes covered, all at nadir viewing geometry, are ideal for examining the latitude dependence of the data. [20] In Figure 7, the size parameter is plotted as a function of latitude, for six different observations of the northern hemisphere (again, details of the observations used can be found in Table 2). As in the southern hemisphere, the size parameter is roughly constant at low latitudes, with a sharp increase at 70 –80° latitude. The increased variability seen in these plots of the northern hemisphere data, compared to the southern hemisphere data of Figure 6, occurs simply because the S hemisphere data are averaged over many tens of degrees of longitude, while the N hemisphere data are zonally averaged only over a much narrower swath. [21] In Figure 8, we examine two of the observations in greater detail, showing radiances at 1.74, 2.30 and 3.5 mm

4 of 12

E00B13

WILSON ET AL.: ANOMALOUS CLOUD PARTICLES AT VENUS POLES

E00B13

Figure 4. Plots of the (left) the size parameter and (right) the 3.5 mm radiance for several orbits.

as well as the size parameter. It can be seen that, as found in the southern hemisphere images, the increase in the size parameter occurs at roughly the same latitude as the edge of the core of the polar vortex. [22] It is instructive also to look at the 1.74 and 2.30 mm radiance plots. In both observations, there are very large variations in the 1.74 and 2.30 mm radiances at low to midlatitudes. Clearly, the 1.74 and 2.30 mm radiances are correlated with each other; after all, a region of thick cloud will cause reduced radiance at both 1.74 and 2.30 mm. However, it can be seen that despite this great variation in the cloud optical thickness, there is no corresponding variation evident in the size parameter. This finding is consistent with the earlier analysis by Carlson et al. [1993] which also found little correlation of the size

parameter to cloud optical depth in Galileo NIMS data. This is perhaps somewhat surprising because standard models of convective cloud suggest that the largest cloud droplets, and the most optically thick clouds, should be found in regions of updraft [see, e.g., McGouldrick and Toon, 2007; Imamura and Hashimoto, 2001]. In the VIRTIS data, then, one might expect to find a correlation between particle size and optical thickness. We see no signs of such a correlation in the data of Figure 6; despite the great variation in the optical thickness of the clouds as evidenced by the 2.30 mm radiance plots, there is no corresponding variation in the size parameter. This nondetection may simply indicate the observed variation in 2.30 mm radiance is not due to convective cells, or that our analysis with its

5 of 12

E00B13

WILSON ET AL.: ANOMALOUS CLOUD PARTICLES AT VENUS POLES

E00B13

Figure 5. Correlation plots as for Figure 3 but for observation VI0085_01. typical pixel size of 50 km was too coarse to resolve individual convective cells. [23] We note that the Carlson et al. [1993] reported an asymmetry in the size parameter between the north and south hemispheres, with northern hemisphere data showing size parameters up to 20% higher than those in the southern hemisphere (this increase of 20% was calculated by comparing data at latitudes in the range 45°S – 45°N). Looking at the data presented in Figures 5 and 6, we do not find a significant difference between the mean sizes in the two hemispheres. However, we do see temporal variation from orbit to orbit of roughly 10– 20% in the size parameter at low latitudes. Therefore we suggest that the north-south asymmetry at low latitudes reported by Carlson et al. [1993] can be ascribed to temporal variability of the clouds. [24] Carlson et al. [1993] also found that the size parameters at high latitudes might be hemispherically asymmetric. However, this finding relied on data from 60°S, which due to the Galileo/NIMS observing geometry meant that the emission angle of these observations was 75°. At these high emission angles the validity of some of the radiative transfer

calculations are called into doubt. The Venus Express observing geometry is much better suited to study of high latitudes because of its polar orbit.

Figure 6. Variation of the zonally averaged size parameter with latitude for southern hemisphere data. Numbers on the graph show the orbit numbers of the observations.

Figure 7. Variation of the size parameter with latitude for northern hemisphere data. The observations used are listed in Table 2.

5. Radiative Transfer Modeling [25] We have used a radiative transfer model to examine the effect of different atmospheric parameters on the radiances observed in the 1 – 2.5 mm spectral windows. [26] The radiative transfer tool used is a full multiple scattering code using correlated k parameterization for computational efficiency, which has been developed at Oxford for many planets [Irwin et al., 2008]. For application of this code to the deep atmosphere of Venus, we use spectral line data from HITEMP for carbon dioxide, and from HITRAN2k for all other gases. A sub-Lorentzian line shape is specified for CO2 as per Tonkov et al. [1996] to account for line mixing in the far wings of the CO2 aborption bands. Collision-induced absorption is included, using continuum absorption coefficients of 6  109 cm1 amagat2 and 4  108 cm1 amagat2 for the 1.7 and

6 of 12

WILSON ET AL.: ANOMALOUS CLOUD PARTICLES AT VENUS POLES

E00B13

E00B13

Table 3. Sensitivity Test for Different Atmospheric Parametersa Temperature raised by 1 K Temperature lowered by 1 K H2O increased by 20% CO doubled zbase lowered by 2 km zbase raised by 2 km FSH = 0.5 FSH = 2.0 25% H2O in cloud droplets 4% H2O in cloud droplets

DR (1.308 mm)

DR (1.740 mm)

DR (2.294 mm)

Dm

+3.3% 3.3%

+3.2% 3.2% 2.1% 0.0% 2.3% + 1.2% 11.1% +10.6% +8.2% +6.4%

+2.4% 2.4% 0.0% 2.4% 1.4% +0.7% 12.7% +15.5% +24.9% +49.9%

+1.8% 1.8% 2.4% +1.3% 1.5% +0.8% 4.5% +2.5% 3.9% 14.2%

4.7% +3.1% 5.2% +5.9% +3.6% 2.9%

a Sensitivity of the peak radiances in each of the 1.3, 1.7, and 2.3 mm spectral window regions to cloud base height, fractional scale height, atmospheric temperature (for the temperature sensitivity, the entire T(z) profile was increased or decreased by 1 K), CO and H2O abundance, and cloud droplet water fraction. DR is the percentage change in peak radiance, while Dm is the change in the calculated size parameter.

2.3 mm window regions, respectively. These parameterizations, and the validation of this spectral transfer model for Venus, are described in detail elsewhere [Tsang et al., 2008a]. [27] We have started by assuming a single-mode cloud structure, using either modes 1, 2, 20 or 3 as defined in Table 1. For each mode of particles, we have assumed a lognormal size distribution for the particles, given by n(r)  (1/r) exp {[ln (r/rg)]2/2 [ln (s)]2}, where rg is the mean radius of the cloud particles and s is the variance. Scattering parameters were calculated assuming Mie scattering, using refractive index data for H2SO4 from Palmer and Williams [1975]. In the base model, we have set the cloud base at 46 km, with the aerosol number density falling off with a constant scale height above the cloud base. The aerosol fractional scale

height (FSH) is defined as the ratio of the aerosol scale height to the atmospheric scale height; for the default model, we have set the FSH to equal 1 (note that in some other papers, this parameter is referred to as the ‘‘particle-togas scale height’’ or PGS). For each mode of cloud particles, the number density of particles at the base of the cloud is varied in order to obtain a range of vertically integrated optical depths. [28] In Figure 9 we have plotted the synthetic spectrum from 1 to 2.5 mm for the base cloud model of mode 20 particles (with a vertically integrated optical depth of 30, as defined at 0.63 mm). In Figure 10 we plot 2.30 mm peak radiance against 1.74 mm peak radiance for cloud models of different optical depths and different modes. Also plotted on Figure 10 are lines of constant size parameter m. These

Figure 8. Radiance at 1.74 mm, 2.30 mm, and 3.5 mm and the calculated size parameter, plotted as a function of latitude, for two pericenter passes of the northern hemisphere. Note that there is no apparent correlation between the radiance and the measured size parameter and that the size parameter increases sharply inside the polar hot spot as defined from the 3.5 mm radiance plots. 7 of 12

E00B13

WILSON ET AL.: ANOMALOUS CLOUD PARTICLES AT VENUS POLES

E00B13

Figure 9. Modeled spectrum for the 1 – 2.5 mm spectral region, using a single-mode cloud model of mode 20 particles, with a scattering optical depth of t 0.63mm = 30, cloud base at 46 km, fractional scale height of 1. Note that the spectrum is plotted at a spectral resolution of 1 nm, while the resolution of VIRTIS-M is 10 nm. lines, which are straight on a log-log plot, should represent a constant particle size according to the analysis of C93. [29] It can be seen that the two studies agree in their fundamental conclusion, which is that larger particles attenuate 2.30 mm radiation more than they attenuate 1.74 mm radiation. In regions of higher radiation, our branches do form straight lines on a log-scale plot indicating that they exhibit a constant size parameter m. However, it can be seen that the ‘‘branches’’ predicted by the present study deviate away from this behavior, particularly in regions of

thick clouds (low radiance); in these regions, the 2.30 mm/ 1.74 mm radiance ratio is higher than predicted by C93. [30] The deviation from the previously simple power law behavior may be due to thermal emission from the base of the clouds themselves. To investigate this, we examined the sensitivity of the modeled radiance to the atmospheric temperature. Figure 11 shows dR/dT, the functional derivative of radiance with respect to atmospheric temperature, for wavelengths of 1.74 and 2.30 mm. The main peak of dR2.30/dT occurs below the clouds at 20– 35 km, which shows that the bulk of the radiation at this wavelength is due

Figure 10. Correlation plot for 1.74/2.30 mm radiances. Experimental data from orbit 85, shown in Figure 5, are reproduced here. The solid lines represent synthetic branches calculated for clouds of different optical thicknesses, composed solely of mode 2, mode 20, and mode 3 particles. All modeled clouds have a cloud base height of 46 km and a fractional scale height of 1. For comparison, dashed lines show lines of constant size parameter as defined in equation (1). 8 of 12

E00B13

WILSON ET AL.: ANOMALOUS CLOUD PARTICLES AT VENUS POLES

Figure 11. Temperature functional derivative, i.e., the sensitivity to temperature changes at different heights in the atmosphere. The horizontal gray line shows the base of the lower cloud.

to thermal emission from these altitudes. However, there is a secondary peak in dR2.30/dT at the cloud base (45 – 50 km) which occurs due to thermal emission from this layer due to the high optical thickness (dt/dz) of this layer. The percentage of the total radiance observed at 2.30 mm is roughly 10% for the ‘‘standard’’ cloud model (i.e., that which has an optical depth of t 0.63mm = 30), but can reach over 30% in regions of thick cloud as are found in the polar vortex. This additional radiance from the cloud base might explain the deviation from the behavior of equation (1) in regions of thick cloud. [31] If we assume that there is no variation in the composition of the particles, we can use the data of Figure 10 to find a mean particle size, on the basis of the mode 2, mode 20 and mode 3 particles, knowing that this comparison should not be trusted in regions of particularly low radiance. We can thus state that the effective mean particle radius is typically (1.3 ± 0.5) mm at latitudes of 0 –70°, with a peak value some 50% larger in the polar regions. However, this statement should be interpreted cautiously, with the understanding that the ‘‘effective mean particle radius’’ is a representation of the combined effects of scattering from several different particle modes. [32] The radiative transfer work described in C93, based on the radiative transfer model of Kamp and Taylor [1990] was optimized for the clouds encountered at the low latitudes seen by Galileo NIMS and from Earth observation. It was not used to simulate clouds with optical thicknesses as high as those encountered in the center of the polar vortices. Although there is clearly scope for further investigation, the agreement between the present radiative transfer modeling and that of C93 is sufficiently close, as can be seen in Figure 8, that the principal conclusions of this work are unaltered. [33] We performed several sensitivity studies to test the sensitivity of the size parameter with respect to temperature, cloud structure, and, to a limited extent, gas and cloud composition. The sensitivity of the emitted radiance with respect to atmospheric temperature can be seen from the dR/

E00B13

dT plot in Figure 11. On the basis of these data, the sensitivity of the calculated size parameter to a change in atmospheric temperature has been calculated to be roughly 2% KP1, as shown in Table 3. This value represents the error in the calculated size parameter that would occur if the temperature at all heights in the atmosphere were increased by 1 K. [34] We performed a limited sensitivity study to atmospheric composition by calculating the effect of first a 20% change in H2O abundance and second to a doubling of CO. The increased CO results in a significantly reduced radiance in the region 2.31– 2.42 mm; however, its effect on the 2.30 mm radiance is outside the main CO absorption band (indeed, that is why the peak radiance of the 2.3 mm region occurs here), so its sensitivity to CO is weak, as shown in Table 3. The 1.74 mm radiance is comparatively much more sensitive to H2O abundance at 15– 30 km altitude. Drossart et al. [1993] found that tropospheric H2O abundance varies by less than 20% over a latitude range of 40°S – 50°N; such a small variation in H2O abundance will only have a small (2.4%) effect on the calculated size parameter as shown in Table 3. In order to explain the 50% increase in size parameter found at high latitudes, one would require that H2O abundance at high latitudes be depleted by a factor of 5 with respect to its low-latitude value, which is unlikely in the context of previous observational constraints. [35] The vertical distribution of the cloud would also be expected to affect the radiances. In a first test, the cloud base was moved from 46 km to either 44 or 48 km, without changing any of the other parameters. The effect on the 2.30 mm spectrum is shown in Figure 12a, while the effect on peak radiances in 1.3, 1.74, and 2.30 mm window regions is tabulated in Table 3. When the cloud base was lowered to 44 km, the observed radiation at 1.74 mm and 2.30 mm dropped by 1.4% and 2.3%, respectively. This is complicated to evaluate, because it is understood that what occurs is that more of the optical path of each photon is deeper in the atmosphere, where the atmospheric density is 20% higher. One might thus expect the absorption to be 20% greater. Balanced against this increased absorption, however, is increased thermal emission from the 44– 46 km region due to the higher dt/dz in this region. [36] In a second test, the fractional scale height (FSH) of the cloud was changed. The fractional scale height determines the vertical extent of the cloud; if the FSH is doubled, the cloud falls off twice as fast, so the vertical extent of the cloud is effectively halved. This means that the number of cloud particles is unchanged and the number of scattering events for each photon are unchanged; but that the mean photon path length is shorter, leading to less absorption of the radiation. The computed spectra are shown in Figure 12b, with peak radiances again tabulated in Table 3. It is interesting to note that the relative magnitude of the center and edges of the spectral emission feature, i.e., its shape, varies with different vertical distributions of the cloud, due to difference in the gaseous absorptance between the center of the spectral window and its edges. Further treatment of this is beyond the scope of the present paper; however, we note that the vertical structure of cloud should be considered when mapping abundances in the deep atmosphere using the 1.7 and 2.3 mm windows. As to the size parameter, we note that its sensitivity to cloud structure variations as

9 of 12

E00B13

WILSON ET AL.: ANOMALOUS CLOUD PARTICLES AT VENUS POLES

E00B13

Figure 12. Sensitivity to (a) the height of the cloud base and (b) the fractional scale height. Parameters are as for Figure 7 except for varying values of the parameter indicated. Spectra have been convolved to the VIRTIS-M resolution of 10 nm. The change in the peak radiance values for this and other window regions is tabulated in Table 3. shown in Table 3 is below 5% for all the changes which were considered; this is a relatively small error compared to the observed spatial variation of the size parameter. Therefore we conclude that the observed spatial variation in the size parameter cannot be due merely to changes in the vertical structure of the cloud. [37] We have also conducted radiative transfer calculations for more complex clouds consisting of different mixtures of modes 1, 2, 20, and 3. In correlation plots of the 1.74/2.30 mm radiances, there is not enough information to distinguish between a multimode cloud of large and small particles (for example a cloud consisting of modes 1 and 3), and that of a single-mode cloud with an intermediate size of cloud particles. Therefore we do not present any of these modeling results here. It is hoped that use of further wavelengths within each window region (e.g., 2.25 mm), as well as use of shorterwavelength window regions (e.g., 1.31 mm), may provide enough information to constrain the vertical distribution of cloud and/or the modal mix of cloud particles. [38] Until now, we have ignored the possibility of variations in the composition of the cloud particles. We have shown that it is possible to interpret the observed diversity of 1.74 and 2.30 mm radiances by varying only the cloud optical thickness and the mean size of the scattering particles. However, we cannot eliminate the possibility that the composition of the cloud particles varies across the planet. In a final sensitivity test, we altered the proportion of H2SO4:H2O ratio assumed in the cloud particles. Instead of using the 85% H2SO4:15%H2O assumed until now, we have also performed radiative transfer calculations with 96% H2SO4:4% H2O and with 75% H2SO4:25% H2O. The refractive index data used were taken from Palmer and Williams [1975], as before. The effect of this change, for the basic cloud model (mode 20 particles with t 0.63mm = 30) is shown in Table 3. It may seem surprising that the 1.74 mm and 2.30 mm radiances increase both if the cloud water fraction is increased to 25% and if it is decreased to 4%; this is because the imaginary part of the refractive index is lower

in both of these cases than in the baseline case of 15% water content, leading to less attenuation. [39] It can be seen that the change in composition has a large effect on the radiances, in particular in the 2.30 mm radiance. This then has a correspondingly significant effect on the calculated size parameter: changing the H2SO4 fraction to 96% caused a 14% change in the inferred size parameter. Unsurprisingly, this shows that the 2.30 mm/1.74 mm ratio is much more sensitive to the composition of the cloud droplets than to their vertical extent. We have not considered the effect of adding other impurities to the cloud particles (e.g., HCl, S, Fe), but these would greatly affect the relative radiances, and thus the calculated size parameter. This sensitivity to cloud particle composition is limited to the regions of the cloud where the cloud optical thickness (and thus the attenuation) is greatest. Therefore, the size parameter as defined here may be sensitive to cloud composition in the lower cloud (below 55 km) but should be relatively insensitive to the composition of the upper cloud where the abundance of scatterers is much smaller. For completeness, though, we intend to test in future the effect of inclusion of sulphur allotropes (Sx) which may be a constituent of mode 1 particles in the upper cloud [Toon et al., 1982]. [40] We do not have enough information to determine the composition of the cloud particles; at this point we can only conclude that while it is possible to interpret the variation in 1.74 mm and 2.30 mm radiances purely by varying the size distribution of the particles, we cannot rule out the possibility of variations in the composition of the particles.

6. Discussion [41] In this discussion we consider how the present analysis fits in with our understanding of the dynamics of the polar vortex. [42] In the troposphere of Venus, as on Earth, larger cloud particles are usually found in regions of updraft. This is

10 of 12

E00B13

WILSON ET AL.: ANOMALOUS CLOUD PARTICLES AT VENUS POLES

partly because the updrafts bring warm moist air up to colder regions where condensation occurs, and partly because large cloud particles quickly settle out of the atmosphere unless held aloft by upward winds. Therefore, the increase in particle size found in the center of the polar vortex would seem to indicate rising air in the center of the vortex. [43] The general understanding of the meridional circulation of Venus is that of a Hadley-type overturning, with heated air rising at the equator and sinking at the high latitudes. Air traveling poleward from the equator is accelerated in a zonal direction by Coriolis acceleration, giving rise to a vortex in midlatitudes. This poleward transport has been seen at cloud top levels by cloud tracking in the UV and visible wavelengths [e.g., Limaye, 2007]. However, very few data were obtained from near the poles, due to the near-equatorial geometry of most past observations. Therefore it is not clear at what latitude the descending branch of the Hadley Cell is located. [44] One possibility is simply that the maximum downwelling is in the center of the polar vortex. This is consistent with the thermal IR imagery because strong downwelling would push down the cloud top altitude to regions of higher temperature, hence resulting in the polar hot spot as seen in the 3.5 mm and 5 mm images. However, it is inconsistent with the finding that cloud particles are largest in the center of the vortex, since this would imply rising air. It thus seems that a more complex cloud model, incorporating more than one cloud-forming constituent, may be required to explain the complexity of the cloud structure in this region. For example, one could postulate a population of nonvolatile large ‘‘mode 3’’ cloud particles in the lower clouds. Strong downwelling in the center of the vortex would lead to evaporation of the smaller mode 2 and mode 20 particles, composed of volatile sulfuric acid particles, leaving only the larger nonvolatile particles in the cloud base. This would result in large particle sizes in the center of the polar vortex despite downwelling. However, this explanation is not satisfactory since it does not explain the regions of very high optical depth sometimes seen at the pole in the 1.74 and 2.30 mm images. [45] Another possibility is that the downwelling branch of the Hadley Cell is strongest at 60– 70° of latitude, which is where the polar collar is found. Here, too, the cloud top altitude is 2 – 3 km lower than it is elsewhere on the planet [Zasova, 1995], which could be associated with downwelling. A stronger argument for downwelling at this latitude comes from maps of CO at 30 km altitude [Tsang et al., 2008b]. Formed at high altitude from photolysis of CO2, CO is then transported downward into the deep atmosphere, where it has a limited lifetime. CO thus acts as a tracer of meridional circulation. The CO maps show a peak at 60° latitude, implying that this is the latitude at which the maximum downwelling occurs. This is consistent with the results at least one global circulation model of Venus, which found that there was strong downwelling at 60°S, with upwelling at the pole itself [Lee et al., 2007]. If one assumes upward velocity transport at the pole, however, it is not clear how to explain the polar hot spot seen in thermal IR images. [46] To resolve the nature of polar circulation and cloud processes using only near-IR remote sounding data will be

E00B13

difficult, because so little information is available. Some further information on both the particle size and vertical structure of cloud will be sought by examining more wavelengths in the near-IR regions. Vertical profiles of temperature can be obtained from radio occultation; however, these will not have a sufficiently high horizontal resolution to capture the fine structure of the vortex region. However, a deeper understanding of this region may require data either from a microwave sounder, which would be able to see through the clouds to map temperature fields, or from in situ measurements.

7. Conclusion [47] Observations of near-infrared emissions from the nightside of Venus using the VIRTIS-M instrument on Venus Express have been analyzed in order to gain information on cloud properties. Using radiative transfer modeling, we have confirmed that increasing cloud particle sizes result in more attenuation of 2.30 mm radiation than of 1.74 mm radiation; and that the relationship between these radiances can be used to constrain spatial variation of particle size. The present modeling work has found that the relation between 1.74 and 2.30 mm radiances takes a more complex form of than the simple power law of (I1.74mm) = m  (I2.30mm)0.53 as proposed by Carlson et al. [1993], especially in regions of optically thick clouds. However, we confirm that the size parameter m proposed in that work is, as a first approximation, indicative of the mean cloud particle size. [48] The size parameter has been evaluated for a range of VIRTIS data from both the southern and northern hemispheres. In both hemispheres, we find no consistent spatial variation of the size parameter at low to midlatitudes; however, we consistently find a sharp increase of the size parameter within the polar hot spot (as defined using thermal infrared images). This is seen in both southern and northern hemisphere data. [49] The data are consistent with the explanation that the cloud particle size increases sharply in the polar vortices; however, we note that the observed variation could also be related to a variation in the composition of the cloud particles. [50] Acknowledgments. We acknowledge the work of the entire Venus Express team, who allowed these data to be obtained. This work was made possible by funding from the UK Science and Technology Facilities Council, as well as national space agencies CNES, ASI, and NASA.

References Allen, D. A., and J. W. Crawford (1984), Cloud structure on the dark side of Venus, Nature, 307, 222 – 224, doi:10.1038/307222a0. Andreychikov, B. M., I. K. Akhmetshin, B. N. Korchuganov, L. M. Mukhin, B. I. Ogorodnikov, I. V. Petryanov, and V. I. Skitovich (1987), X-ray radiometric analysis of the cloud aerosol of Venus by the Vega 1 and 2 probes, Cosmic Res., Engl. Transl., 25, 721 – 736. Be´zard, B., C. de Bergh, D. Crisp, and J. Maillard (1990), The deep atmosphere of Venus revealed by high-resolution nightside spectra, Nature, 345, 508 – 511, doi:10.1038/345508a0. Carlson, R., K. Baines, T. Encrenaz, F. Taylor, P. Drossart, L. Kamp, J. Pollack, E. Lellouch, and A. Collard (1991), Galileo infrared imaging spectroscopy measurements at Venus, Science, 253, 1541 – 1548, doi:10.1126/science.253.5027.1541. Carlson, R. W., L. W. Kamp, K. H. Baines, J. B. Pollack, D. H. Grinspoon, T. Encrenaz, P. Drossart, and F. W. Taylor (1993), Variations in Venus

11 of 12

E00B13

WILSON ET AL.: ANOMALOUS CLOUD PARTICLES AT VENUS POLES

cloud particle properties: A new view of Venus’s cloud morphology as observed by the Galileo Near-infrared Mapping Spectrometer, Planet. Space Sci., 41(7), 477 – 485. Drossart, P., et al. (1993), Search for spatial variations of the H2O abundance in the lower atmosphere of Venus from NIMS-Galileo, Planet. Space Sci., 41(7), 495 – 504. Esposito, L. W., R. G. Knollenberg, M. Y. Marov, O. B. Toon, and R. P. Turco (1983), The clouds and hazes of Venus, in Venus, pp. 484 – 564, Univ. of Ariz. Press, Tucson. Esposito, L. W., J.-L. Bertaux, V. Krasnopolsky, V. I. Moroz, and L. V. Zasova (1997), Chemistry of lower atmosphere and clouds, in Venus II, pp. 415 – 458, Univ. of Ariz. Press, Tucson. Grinspoon, D. H., J. B. Pollack, B. R. Sitton, R. W. Carlson, L. W. Kamp, K. H. Baines, T. Encrenaz, and F. W. Taylor (1993), Probing Venus’s cloud structure with Galileo NIMS, Planet. Space Sci., 41(7), 515 – 542. Imamura, T., and G. L. Hashimoto (2001), Microphysics of Venusian clouds in rising tropical air, J. Atmos. Sci., 58, 3597 – 3612, doi:10.1175/1520-0469(2001)058<3597:MOVCIR>2.0.CO;2. Irwin, P. G. J., N. A. Teanby, R. de Kok, L. N. Fletcher, C. J. A. Howett, C. C. C. Tsang, C. F. Wilson, S. B. Calcutt, C. A. Nixon, and P. D. Parrish (2008), The NEMESIS planetary atmosphere radiative transfer and retrieval tool, J. Quant. Spectrosc. Radiat. Transfer, 109(6), 1136 – 1150, doi:10.1016/j.jqsrt.2007.11.006. Kamp, L. W., and F. W. Taylor (1990), Radiative transfer models of the nightside of Venus, Icarus, 86, 510 – 529, doi:10.1016/00191035(90)90231-W. Knollenberg, R. G., and D. M. Hunten (1980), The microphysics of the clouds of Venus: Results of the Pioneer Venus Particle Size Spectrometer experiment, J. Geophys. Res., 85(A13), 8039 – 8058, doi:10.1029/ JA085iA13p08039. Krasnopolsky, V. A. (1989), Vega mission results and chemical composition of Venusian clouds, Icarus, 80, 202 – 210, doi:10.1016/00191035(89)90168-1. Lee, C., S. R. Lewis, and P. L. Read (2007), Superrotation in a Venus general circulation model, J. Geophys. Res., 112, E04S11, doi:10.1029/ 2006JE002874. Limaye, S. S. (2007), Venus atmospheric circulation: Known and unknown, J. Geophys. Res., 112, E04S09, doi:10.1029/2006JE002814. McGouldrick, K., and O. B. Toon (2007), An investigation of possible causes of the holes in the condensational Venus cloud using a microphysical cloud model with a radiative-dynamical feedback, Icarus, 191, 1 – 24, doi:10.1016/j.icarus.2007.04.007. Palmer, K. F., and D. Williams (1975), Optical Constants of Sulfuric Acid: Application to the clouds of Venus?, Appl. Opt., 14(1), 208 – 219. Piccioni, G., et al. (2007), South polar features on Venus similar to those near the north pole, Nature, 450(7170), 637 – 640, doi:10.1038/ nature06209.

E00B13

Pollack, J., J. Dalton, D. Grinspoon, R. Wattson, R. Freedman, D. Crisp, D. Allen, B. Be´zard, C. de Bergh, and L. Giver (1993), Near-infrared light from Venus’ nightside: A spectroscopic analysis, Icarus, 103, 1 – 42, doi:10.1006/icar.1993.1055. Svedhem, H., et al. (2007), Venus Express – the first European mission to Venus, Planet. Space Sci., 55(12), 1636 – 1652, doi:10.1016/j.pss.2007. 01.013. Taylor, F., D. Crisp, and B. Be´zard (1997), Near-infrared sounding of the lower atmosphere of Venus, in Venus II, pp. 325 – 351, Univ. of Ariz. Press, Tucson. Tonkov, M., N. Filippov, V. Bertsev, J. Bouanich, V.-T. Nguyen, C. Brodbeck, J. Hartmann, C. Boulet, and F. Thibault (1996), Measurements and empirical modeling of pure COB2B absorption in the 2.3 mm region at room temperature: Far wings, allowed and collision-induced bands, Appl. Opt., 35, 4863 – 4870. Toon, O. B., R. P. Turco, and J. B. Pollack (1982), The ultraviolet absorber on Venus: Amorphous sulphur, Icarus, 51, 358 – 373, doi:10.1016/00191035(82)90089-6. Toon, O. B., B. Ragent, D. Colburn, J. Blamont, and C. Cot (1984), Large, solid particles in the clouds of Venus – do they exist?, Icarus, 57(2), 143 – 160, doi:10.1016/0019-1035(84)90063-0. Tsang, C. C. C., P. G. J. Irwin, F. W. Taylor, and C. F. Wilson (2008a), A correlated-k model of the Venus nightside near-infrared window emissions from 1.0 to 2.5 mm, for the retrieval of minor species from Venus Express/VIRTIS, J. Quant. Spectrosc. Radiat. Transfer, 109(6), 1118 – 1135, doi:10.1016/j.jqsrt.2007.12.008. Tsang, C. C. C., P. G. J. Irwin, C. F. Wilson, F. W. Taylor, C. Lee, R. de Kok, P. Drossart, G. Piccioni, B. Be´zard, and S. B. Calcutt (2008b), Tropospheric carbon monoxide concentrations and variability on Venus from Venus Express/VIRTIS-M observations, J. Geophys. Res., 113, E00B08, doi:10.1029/2008JE003089. Zasova, L. V. (1995), The structure of the Venusian atmosphere at high latitudes, Adv. Space Res., 16(6), 89 – 98, doi:10.1016/0273-1177(95) 00254-C. 

R. W. Carlson, Jet Propulsion Laboratory, California Institute of Technology, MS 183-601 4800 Oak Grove Drive, Pasadena, CA 911090000, USA. P. Drossart and S. Guerlet, Laboratoire d’Etudes Spatiales et d’Instrumentation en Astrophysique, Observatoire de Paris, Place Jules Janssen, Meudon F-92195, France. P. G. J. Irwin, F. W. Taylor, C. C. C. Tsang, and C. F. Wilson, Department of Physics, Oxford University, Parks Road, Oxford OX1 3PU, UK. ([email protected]) G. Piccioni, Istituto di Astrofisica Spaziale e Fisica Cosmica, I-00133 Rome, Italy.

12 of 12

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114, E00B27, doi:10.1029/2008JE003116, 2009

for

Full Article

Multivariate analysis of Visible and Infrared Thermal Imaging Spectrometer (VIRTIS) Venus Express nightside and limb observations S. Erard,1 P. Drossart,1 and G. Piccioni2 Received 14 February 2008; revised 31 May 2008; accepted 3 November 2008; published 7 January 2009.

[1] Spectral signatures measured by Visible and Infrared Thermal Imaging Spectrometer

(VIRTIS) on the nightside and at the limb of Venus are analyzed with Independent Component Analysis. A methodology has been set up to minimize instrumental effects and to interpret the results on the basis of studies of the most common situations in the data set. The main spectral components commonly retrieved on the nightside include the bulk signal modulated by atmospheric opacity variations, photometric variations in the long-wavelength atmospheric windows, a branching parameter describing particle size variations, and O2 emission at 1.26 and 1.58 mm. Faint atmospheric windows are detected at 1.51, 1.55, 1.78, and 1.82 mm for the first time. The polar vortex structure is outlined, with two main circular areas made of many concentric rings with alternating particle sizes. Discrete clouds about 100 km across are observed in low opacity conditions. High-altitude, warm clouds are tentatively observed from the polar vortex down to 55°S. At the limb, the two signatures of CO2 nonlocal thermodynamic equilibrium emission are directly mapped, and the thermal structure of the cloud layers and upper atmosphere is apparent. Surface emission is detected with a spatial resolution limited by atmosphere blurring, reaching
35 km in exceptional conditions. Horizontal offsets indicate that the radiation propagates mostly vertically, consistent with the large optical depth and vertical extent of the cloud layer. Intense scattering is suspected to take place at the bottom of the atmosphere, at least in the southern plains. Citation: Erard, S., P. Drossart, and G. Piccioni (2009), Multivariate analysis of Visible and Infrared Thermal Imaging Spectrometer (VIRTIS) Venus Express nightside and limb observations, J. Geophys. Res., 114, E00B27, doi:10.1029/2008JE003116.

1. Introduction [2] Early space observations of Venus have shown that the atmosphere is stratified in three main cloud layers which differ in particular in terms of scattering effects related to different particle density and size. Particle sizes aggregate around three modal values ranging from 0.6 mm (mode 1, forming a haze above the main cloud deck) to 7 mm (mode 3, dominant source of opacity in the lower/middle clouds). Mode 2 and 20 particles (
2 and 3 mm in size) are present mostly in the upper and lower/middle clouds respectively, and make them uniformly opaque and relatively featureless spectrally at infrared wavelengths [e.g., Esposito et al., 1983; Grinspoon et al., 1993]. [3] The radiance of Venus’ nightside is dominated by black body emission of the upper clouds longward of 3.5 mm, in the 70– 75 km altitude range. CO2 absorption and strong scattering by the haze and clouds blocks thermal emission from lower, warmer layers, except in discrete spectral windows where intensity peaks are observed. The shape of these peaks is influenced by specific species at specific altitudes, and by scattering effects. Signatures from various 1 2

LESIA, Observatoire de Paris, CNRS Meudon, France. INAF, IASF, Rome, Italy.

Copyright 2009 by the American Geophysical Union. 0148-0227/09/2008JE003116$09.00

atmospheric layers are therefore present at various wavelengths, with spatial distributions related to the dynamics or absorbers in these layers [e.g., Allen and Crawford, 1984; Carlson et al., 1991]. [4] Most notably, peaks centered at 1.02, 1.10, 1.18, and 1.28 mm are dominated by the emission of the very deep atmosphere and of the surface itself. Longer-wavelength peaks are strongly modulated by the atmospheric layers opacity, and allow tracking cloud motions in these layers [Baines et al., 2006]. Increased radiance at 1.74 mm and 2.2– 2.5 mm is thus indicative of opacity in the lower and middle clouds region (47 – 57 km); a smaller and narrower peak is also present at 1.31 mm. The ratio of peak intensities at 2.3, 1.74, and 1.31 mm is on the first-order a measurement of scattering effects, most notably related to particle size. Spatial variations in these ratios define latitudinal zones with different size distributions (mode 2 and mode 3 particle abundances [Carlson et al., 1993]). [5] H2O in deeper layers affects the range from 1.10 to 1.19 mm, and therefore modifies the profile of the 1.10 and 1.18 mm peaks. H2O at higher altitudes (
35 km) affects the intensity of the 1.74 mm peak. The wide 2.2 – 2.5 mm structure is sensitive successively to H2O, CO, HF, OCS, and SO2 in the 35– 50 km altitude range [e.g., Pollack et al., 1993]. Deexcitation of O2 at 1.27 mm forms a rapidly varying airglow at
95 km altitude, centered around the antisolar point, and overlapping the 1.28 mm window

E00B27

1 of 20

E00B27

ERARD ET AL.: MULTIVARIATE ANALYSIS OF VENUS NIGHTSIDE

[Connes et al., 1979]. Finally, thermal emission from the surface is expected to contribute to the first three peaks (at 0.98, 1.10 and 1.18 mm) and is controlled by surface elevation [e.g., Meadows and Crisp, 1996]. Lower areas are warmer, which translates in larger intensity in the shortwavelength peaks whenever atmospheric opacity is moderate enough to allow the signal to cross the entire path length. [6] Visible and Infrared Thermal Imaging Spectrometer (VIRTIS) limb observations allow measurement of highaltitude emission from various species [Drossart et al., 2007b]. These include CO2 and CO fluorescence [Gilli et al., 2008], deexcitation of O2 recombined on the nightside [Ge´rard et al., 2008], and OH emission in the so-called Meinel bands [Piccioni et al., 2008]. In addition to the 1.27 mm band observed in nadir view another, a much fainter O2 feature is located at 1.58 mm and can be observed at the limb [Piccioni et al., 2008]. Nonlocal thermodynamic equilibrium (non-LTE) emission of CO2 is also observable at the limb depending on the Sun direction, with a significant magnitude in the 4.25 – 4.5 mm range and additionally at 2.7 and 2.0 mm. CO has a similar emission feature in the 4.6– 4.7 mm range. Finally, the very faint OH bands have been observed in some situations in the 1.42 –1.46 mm and 2.7– 3.1 mm ranges [Piccioni et al., 2008]. [7] The clouds are organized in zonal bands with elongated features at midlatitudes up to the polar vortex area, with occasional latitudinal mixing [Piccioni et al., 2007]. Mottled, irregular and warped turbulent structures are present at lower latitudes, and packets of gravity waves are detected at different latitudes in the upper and lower clouds [Sanchez-Lavega et al., 2008]. Poleward of 75°S, the polar vortex is observed with its characteristic dipole pattern displaced from the pole, and surrounded by a dark, cold cloud collar extending down to 60°S [Piccioni et al., 2007]. This is the counterpart of a similar structure observed in the northern hemisphere by Pioneer Venus [Taylor et al., 1979]. [8] The spectral signatures measured by VIRTIS on the nightside are hereafter analyzed with Independent Component Analysis (ICA) to determine their spectral associations and spatial regularities. These phenomena are usually studied through simple spectral parameters such as peak intensities measured at a given wavelength. The use of ICA is expected to improve data analysis for three reasons: [9] 1. Increased signal-to-noise ratio. Multivariate analyses take advantage of the complete peak profiles, not only the central wavelength; besides, they gather together all peaks with a consistent spatial distribution, which further enlarges the signal. [10] 2. Intrinsic spectral parameter corrections. Opposition between measurements at different wavelengths is similar to computing more elaborate spectral parameters, e.g., stray light removal, band depth computation by subtracting a local continuum, removal of a background cloud pattern. [11] 3. Optimized spectral associations. The ICA evidences associations of spectral measurements in a more robust and flexible way than correlation between peak intensities, for instance. Given enough spectra and variability, associated signatures correspond to the same physical phenomenon. [12] A practical problem is raised by the increasing capacities of modern imaging spectroscopy experiments. The data sets have become so large that they are difficult to

E00B27

analyze, especially because the new and interesting information consists in small and local signatures hidden inside highly correlated spectral data, and carry very little variance. For the above reasons, ICA may offer a practical way to analyze such large data sets, at least as a quick-look tool, with an ability both to identify the major spectral types in the data, and to extract minor but markedly different signatures. [13] The objectives of this work are therefore to assess the capacities of the method to study a thick planetary atmosphere, to identify the type of preprocessing potentially required, and to create a methodology to interpret the results. This involves (1) identifying the main associations observed in typical situations on Venus, (2) studying small signatures carrying little variance and observed with low signal-to-noise ratio, and (3) assessing the effects of noise, random events, and remaining instrumental effects in the data. The present work focuses on the overall regularities encountered so far in the VIRTIS data set, and will be used in future works as a general frame to analyze unusual situations.

2. Method 2.1. Principle of the Analysis [14] ICA is a multivariate method which is focused on signal unmixing (blind source separation). Like Principal Components Analysis (PCA) it starts from mixed signals only, and returns a set of components and mixing coefficients [Comon, 1990]. In contrast to PCA though, it does not look for the directions of maximal variance, but it searches for independent components. Statistical independence implies an absence of correlation, but is a more stringent condition: it means that the distribution of one variable does not tell anything about the distribution of the other variables (marginal probability distribution functions are separable). [15] Technically, this is done in two steps: first, the data are decorrelated and scaled (whitened); this is achieved by projecting the data in their eigenspace, or in its first dimensions. Second, ICA searches a rotation of this subspace that maximizes a measure of non-Gaussianity, because any mixture of independent components is expected to be more Gaussian than the individual components; this however assumes that at most one component has a Gaussian probability distribution function. Several solutions have been implemented, using various non-Gaussianity estimates. In this paper, we use the JADE algorithm [Cardoso, 1999], which is based on joint diagonalization of the fourthorder cumulant tensor. JADE has some practical advantages over other ICA algorithms, in particular it is more efficient (less demanding in terms of CPU time) and it uses direct computation (not starting from random nuclei), which makes the results reproducible from one run to the next. As all ICA algorithms however, JADE can only identify directions in the data space but does not provide their relative magnitudes. More explicitly, ICA defines a transform: X ¼ AS

ð1Þ

where X is the data, S is independent components, and A is the matrix of mixing coefficients. ICA identifies A and S

2 of 20

E00B27

ERARD ET AL.: MULTIVARIATE ANALYSIS OF VENUS NIGHTSIDE

from X alone, but only the products A  S can be identified. Like most ICA algorithms, JADE scales the signals S so that their variance equals one. As a consequence, the signs of the returned components are arbitrary; in addition, they are ordered according to a non-Gaussianity parameter, not according to their contributions to the data. Results can be refined using extra knowledge of the physical problem involved. The only free parameter in ICA is the number of components retained in the first step. [16] ICA has been increasingly used in the recent years. Although originally applied to observational cosmology, it has proven particularly well adapted to analysis of planetary remote sensing data [e.g., Forni et al., 2005; Moussaoui et al., 2008]. Experimentally, applications of ICA to spectral cubes are more precise and easier to interpret than PCA: independent components are closer to physical associations of variables than principal components; ICA clearly separates components introducing similar variance in the data set, whereas PCA does a terrible job at inverting signal mixtures in these conditions; for the same reason, independent components with limited spatial exposure (and therefore limited weight) are readily identified, whereas PCA usually fails to identify these small variations; finally, ICA is by construction very robust to random noise in the data, which can be considered Gaussian; in contrast, a simple PCA does not separate low-variance signals from noise and often ascribes a higher weight to noisy components than to physical phenomena with minor expression. The Maximum Noise Fraction method, which is essentially a PCA applied to noise-whitened data, reduces this problem but does not appear as efficient as ICA in this context. 2.2. Implementation for VIRTIS [17] ICA is hereafter applied to spatially extended data from the infrared imaging channel, VIRTIS M-IR. VIRTISM data are organized in spectral cubes, with one spatial dimension and the spectral dimension acquired simultaneously at each time step using bidimensional detectors [Coradini et al., 1998; Drossart et al., 2007a]. The spectral range (1.02– 5.12 mm) is sampled with a constant spectral step of 9 nm, and an actual channel width of
16.5 nm. The image dimension and the spatial resolution depend on slant distance and binning mode. The noise equivalent radiance is usually on the order of 3.104 W m2 sr1 mm1, allowing to resolve very subtle features in emission or absorption. [18] Nightside and limb observations are used in this paper. In terms of spacecraft operations [Titov et al., 2006], nightside observations usually correspond to science case 2 (off-pericenter disk observation, including movies), sometimes to science case 3 (observation from apocenter, including global mosaics); limb data are from either science case 2 or 7 (dedicated limb observations). Two different observing modes are used during operations: JHK mode (long exposure times to optimize signal-to-noise ratio at shorter wavelengths) and LM mode (short exposure times to avoid saturation at longer wavelengths). In JHK mode, only short-wavelength spectral channels are retained, to filter saturation; all such sessions are analyzed in the spectral range from 1.02 to 3.9 mm. For disk observations, pixels in the dark sky are filtered out; although they only contain Gaussian noise, the presence of outliers slightly affects the definition of the average disk component.

E00B27

[19] The spectral cubes are calibrated in radiance (in W m2 sr1 mm1) using the latest version of the standard VIRTIS scheme, and saturated and unreliable pixels are set to zero. Multivariate analyses of this sort assume that spectra are directly comparable, and therefore that at least dark current and flat field corrections are of high quality. Conversely, they can shed light on calibration inaccuracies, and indeed the amount of details provided by the ICA has increased as the calibration scheme got more accurate. Spatial coefficients are here plotted as images (e.g., Figure 1b) or projected on geographic coordinates (e.g., Figure 3). In image view, the X axis corresponds to the spatial dimension of the detector while the Y direction is acquired through time. Small variations of the pixel to pixel response thus propagate through time and produce a vertical striation; defective pixels produce random patterns in vertical bands; localized events such as cosmic rays translate as points or short segments; flat field inaccuracies or temporal variations produce low-frequency variations in the horizontal direction. The same may apply in the spectral direction. The analysis is therefore restricted to the part of the detector which responds nominally, while columns producing systematic patterns are removed (for example, the first six columns are always removed). An additional difficulty is that spectral registration must also be uniform inside a cube (i.e., the channel-to-wavelength correspondence should be constant along the spatial direction of the detector), which is not always found to be the case at the nm level. This issue can only be approximately fixed by precise spectral resampling, which is not performed with the current calibration scheme. [20] The ICA algorithm used here is the IDL implementation of JADE [Moudden et al., 2005], with minor adaptations. A single additional constraint is applied to VIRTIS data in order to recover consistent signs of the components S and coefficients A from session to session. In general, both the spectral components and their spatial coefficients have positive and negative values. The normalizing convention applied here consists in forcing the spatial coefficients of each component to be positive in average. In particular, this convention results in positive features for the main contributions in the atmospheric windows, the O2 emission band, and the upper layer thermal emission. [21] Since most phenomena involved in this data set are multiplicative in nature, a straightforward idea is to analyze the logarithm of the signal. This potentially allows to separate a layer transmission from the emission of the underlying layers, and emissivity from temperature variations in a given layer. In practice however, this scheme does not provide satisfying results. The signal logarithm is dominated by random variations and instrumental patterns in the channels where the measured radiance is small, i.e., where the atmosphere is essentially opaque. Independent components derived this way are hardly usable, and have extremely noisy coefficient maps. More fundamentally, the objectives of the present study are not to derive fractions of the various components, but rather to evidence the modes of spectral variability and their spatial distributions. A more quantitative scope, such as deriving gas abundances or opacities, can only be reached through a radiative transfer approach. Such an approach cannot be applied to all spectra owing to long processing times, but a statistical analysis is

3 of 20

E00B27

ERARD ET AL.: MULTIVARIATE ANALYSIS OF VENUS NIGHTSIDE

helpful in a first step to identify the diversity of situations, and possibly to select example spectra to be inverted. The approach adopted here is therefore to analyze the radiance directly, which still allows us to detect, e.g., variations of temperature with local time in a linear approximation. [22] Owing to the rather demanding algorithm involved, JADE is applied in the spatial direction rather than the spectral one, and is limited to the first 10 components. This means that the images of all spectral channels are decomposed as mixtures of statistically independent images, while the mixing coefficients are spectra providing the actual components in the signal (i.e., equation (1) is transposed). Since only the first components of the whitened data are retained, the computation in both directions may not be exactly equivalent however. Comparisons have been performed on smaller VIRTIS cubes with ICA applied in the spectral direction, or in the image direction by retaining up to 30 components. In all cases, the differences proved to be small, and do not affect the interpretation. [23] Studies based on ICA do not usually focus on component magnitudes, because of the scaling performed during the whitening phase. To interpret remote sensing data, it is however interesting to know the average contributions of the various spectral components to the measured signal. We use here the quadratic mean of the mixing coefficients to estimate these contributions. Since we are working in the spatial direction, this is done after proper scaling of the coefficient matrix so as to retrieve results similar to those of an ICA performed in the spectral direction (i.e., spectral components are first normalized to unit variance). Tests performed on controlled mixtures show that complete analyses in both directions are very similar, and that the contributions retrieved in these conditions are usually within
15% of the actual mixing coefficients. These contributions are, therefore, representative of the percentage of radiance owing to each component in the spectral cube, but components are sorted by the ICA according to their inhomogeneity, which is usually related to the existence of clustering around several mean values. In this sense (which is different from variance analysis) the first components introduce more variability among spectra. [24] In the following discussions, the physical meaning is derived from the spectral components themselves (normalized to unit variance) and the coefficient maps are used only to study the spatial distributions of the retrieved components. Other quantities used to interpret the results are the weighted components (normalized components  contributions, providing the spectral contributions in radiance) and the components’ variability (normalized components  coefficients variance, used to estimate the spectral contrast).

3. Results [25] This section presents extensive ICA results in several situations common in the VIRTIS data set: midlatitude observations with short integration time, adjusted to longwavelength measurements; midlatitude observations with long integration time, stopping at 3.9 mm; observations of the south pole with long integration time. The main regularities encountered in these situations are outlined in this section.

E00B27

[26] The VIRTIS data archive includes geometry files associated with each data file, which provide viewing angles and coordinates of each pixel on two reference surfaces: the Venus ellipsoid and an arbitrary cloud layer located at 60 km altitude. In addition, surface elevation is provided for each pixel as the average of the Magellan Global Topography Data Record (GTDR) [Ford and Pettengill, 1992] over the nominal FOV projected at the surface. This projection is performed along the geometrical line of sight neglecting refraction and scattering effects in the atmosphere. For limb observations, surface elevation is replaced by the tangent altitude, and all other quantities are computed at the tangent point. These quantities are computed with the Spice system, using navigation kernels provided by ESA. The nominal pointing accuracy is on the order of 0.01°, better than the IFOV size (see VIRTIS archive documentation [Erard, 2008] and associated documents). This information is used to draw maps of the data and to study variations related to viewing geometry (emergence angle and local time) and to surface elevation. 3.1. Midlatitude Views: Long Wavelengths [27] A nine-cubes mosaic of the whole disk was acquired on orbit 67. The second of these cubes, analyzed here, is entirely composed of nightside data. It was acquired with short integration time (0.36 s, LM mode) therefore it is noisy at short wavelengths but does not saturate in the longer-wavelength channels. When applied to the complete spectral range, the ICA identifies three main components (Figure 1). All three components have spatial coefficients with average far from zero, and are always positive. Higherorder components are obviously dominated by instrumental effects (noisy lines, hot pixels, and stray light) or cosmic rays. [28] The first component has a 58% contribution. It dominates the thermal signal (black body radiance, including absorptions by CO2 -CO) and the shortest-wavelength emission peaks, and is a major contributor also in the midrange peaks at 1.74 and 2.3 mm. The coefficients have a spatial distribution which is rather uniform, displaying mostly low-frequency variations and only a subdued cloud pattern. The decrease toward the right of the image corresponds to the decrease in temperature to the west, as the atmosphere cools down with increasing local time in the night; this is by far the largest variability observed in the data. The long-wavelength end of the spectrum is fit by a 240 K black body, corresponding to an altitude of 59 km according to the VIRA profile; this appears to be the average altitude of the main thermal contribution to the spectra, i.e., the altitude where opacity equals 1 longward of 3 mm. The shortwavelength contribution to this component is the average radiance in the atmospheric windows. [29] The second component has an overall contribution of 27%. It is mainly concentrated in the peaks, especially at 1.74 and 2.3 mm where this is the main contributor, and to a lesser extent in the black body. Its effect is to reduce the intensity of the black body, to add radiance in the CO2 band wings, and to reduce the radiance in the atmospheric peaks. It therefore represents variations in the temperature of the emitting layer, plus variations in cloud opacity. The spatial coefficients have marked cloud structure at midlatitudes,

4 of 20

E00B27

ERARD ET AL.: MULTIVARIATE ANALYSIS OF VENUS NIGHTSIDE

E00B27

Figure 1. (a) Main three components for session 67_01 in LM mode, normalized to unit variance. The first component is the main contribution and responds to the blackbody and short-wavelength peaks. The other two components decrease the blackbody emission and either reduce or increase the shortwavelength peaks. (b) Spatial coefficients for these components in image format (west is on the right side). The bottom frame is the radiance measured at 1.74 mm plotted for comparison. In this case, all coefficients are positive. related to opacity variations, and a maximum contribution at latitudes higher than 60°S, related to temperature. [30] The third component has a 11% contribution. It is again related to the atmospheric peaks and to the wings of the CO2 absorptions above 4 mm. Although the contributions are located at the same wavelengths as those of the second component, there is an opposite relationship between the long- and short-wavelength parts of the spectra: this component contributes essentially to increase the signal in the short-wavelength peaks, but also to remove intensity in the black body and to add radiance in the wing of the CO2 bands. Spatially, it mostly displays an area located equatorward of 45°S where the 1.74 mm and 2.3 mm peaks have reduced intensity. [31] Longward of 3 mm, components 2 and 3 are fit by black bodies at 230 and 220 K respectively. It is remarkable that the spectral components follow a Planckian function at long wavelengths, suggesting that they actually represent different temperatures although they are added to each other. It is likely that they represent spatial variations in temperature and altitude of the upper, emitting layer: the overall gradient in component 2 represents the cooling of the emitting layer with increasing latitude, but the variations in CO2 band profiles qualitatively suggest an increase of altitude as well. This remains to be assessed by more quantitative radiative transfer. A part of the variability in the atmospheric windows is included in this component because it follows a similar latitudinal pattern, although it is related to opacity variations in a much deeper layer (lower to middle clouds). In contrast, component 3 is mostly responsive to the intensity in the peaks, and the black body contributes here mainly to remove the large-scale latitudinal pattern. [32] As a conclusion, the various physical phenomena are not entirely separated in this situation: components 2 and 3 must be recombined to recover either the variations in the 1.74 mm peak or the latitudinal variations in the black body. Altogether, these two components define three areas: a midlatitude band dominated by warm black body and intense peak emission (or reduced cloud opacity); a high

southern latitude area which has opposite properties; a lower-latitude area with warm black body and reduced peak intensity. These three areas are obviously related to zonal circulation. In addition, component 1 defines a cooler, deeper night region. Overall, the three components identified in this case do not appear related to different layers, but rather to the average signal and its variations with increasing local time (first component), or to latitudinal variations (next two components). [33] ICA of the other night cubes of this mosaic reveal consistent patterns, with additional information at the limb (O2 emission). In any case, the main contributors to the nightside signal appear to be the thermal black body, the peaks in the atmospheric windows (especially the 1.74 mm one and the 2.2– 2.5 mm structure, representative of opacity in the lower and middle cloud layer), and the O2 emission at 1.27 mm. Observations performed at short integration times are dominated by these variations. In the rest of this paper, only disk observations acquired with long integration times will be used, so as to evidence the more subtle signatures located at short wavelengths. 3.2. Midlatitudes: Short Wavelengths [34] Successive global views of a given area (‘‘movie’’ mode) were acquired on the nightside at reduced resolution during orbit 110. The area is centered at low to midlatitudes (15°N to 45°S) between Themis and Beta Regio (240°– 275°longitude). Session 5 of this series is analyzed in this section. A long exposure time is used to optimize the signalto-noise ratio at short wavelengths (JHK mode: summation of 4 frames acquired with an integration time of 3.3 s). Therefore, the long-wavelength channels are saturated, and the effective spectral range analyzed is in this case 1.02– 3.9 mm, which includes just the onset of the thermal emission. The particularities observed in this session are (1) an extremely low opacity; (2) a large range of emergence angles, affecting both the thermal emission and peak intensities at 1.74 and 2.3 mm. These quantities appear to be strongly correlated at large scales, with a relatively sudden drop at latitude 20°S, as emergence reaches 60°.

5 of 20

E00B27

ERARD ET AL.: MULTIVARIATE ANALYSIS OF VENUS NIGHTSIDE

E00B27

Figure 2. (a) Main components for session 110_05, acquired in JHK mode at middle to low latitudes, under low opacity. (b) Spatial coefficients for these components in image format (west is on the right side). The last frame is the radiance measured at 1.74 mm plotted for comparison. [35] In this session the first 7 components have continuous spatial distributions (Figure 2). The second component has only large positive values, therefore it carries most of the signal and varies with the overall brightness but it only ranks second in terms of inhomogeneity introduced among the spectra. The third component (O2 emission) also has

positive values, and is therefore simply added in some locations. All other components are either positive or negative on the disk, with average and median values much closer to zero, and therefore represent modulations around the main component. These components represents more than 96% of the measured signal.

6 of 20

E00B27

ERARD ET AL.: MULTIVARIATE ANALYSIS OF VENUS NIGHTSIDE

E00B27

Figure 3. Session 110_05. Cylindrical maps of (a) component 4 (small-scale variability in the 2.3 mm structure, inverted), (b) component 7 (surface, inverted) compared to (c) Magellan topography integrated on the pixel areas (scale in km). The contours are from Magellan altimetry, contour interval is 500 m. [36] The first component (10% contribution), in spite of its moderate contribution, carries most signal variability in the 2.2 –2.5 mm structure, in the 1.74 mm peak, and in the black body emission. It has no contribution from the 1.28 mm peak, but is slightly opposed to the O2 emission at 1.27 mm (i.e., it includes a negative contribution at this wavelength). This component depicts a strong variation with latitude: intense thermal emission and low cloud opacity observed at middle to high latitudes, while the reverse situation is found equatorward of 20°S. As mentioned above thermal emission and cloud opacity are correlated, although it is unclear if it is only an effect of viewing geometry (implying that the upper layer is not Lambertian), or if the temperature of the upper layer is decreased by the opacity of the underlying cloud layers. Another pattern in the component’s distribution is an area centered at latitude 25°S which has maximum flux at 1.74 and 2.3 mm, and therefore minimum cloud opacity. [37] The second component (60% contribution) is the main nightside signal, with large, positive coefficients on the whole disk, which dominates the radiance at all wavelengths except in the O2 emission band. Although roughly similar to the first component, it includes a much larger (
50%) contribution in particular at short wavelengths, both in the atmospheric windows and between them. Interpeak radiance is mostly daylight signal scattered on the nightside, and is rather small in this session; it is only present in this component. This is also the only component which contributes significantly at 1.28 mm. Finally, this component includes faint but distinct structures at 1.31, 1.51, 1.55, and 1.78 mm. In spite of its dominant contribution to the signal, it carries only
25% of the variability in the short-wavelength peaks, and even less in the 2.2– 2.5 mm structure. [38] The third component (19% contribution) carries the only significant contribution to the 1.27 mm peak and reflects the variability at this wavelength. Its contributions are negative in all peaks except at 1.27 mm. It stands for the O2 emission from high altitude, which is especially marked at the limb and in a yin and yang shaped pattern at the top right of the image. This contribution is further discussed in section 3.3. [39] The fourth component (3.5% contribution) has positive contributions from the first three peaks and includes the most intense thermal emission, opposed to the long-

wavelength peaks at 1.27, 1.74, and 2.3 mm. Detailed examination of the data show that it is related to smallscale variations in the intensity of the 2.3 mm structure (Figure 3a). The black body contributes mainly to remove the large-scale variations between equatorial regions and higher latitudes, which are described by component 1. The resulting features have unusually small sizes (on the order of 100 km), and they are probably observed only because of the very low opacity in this session. They are tentatively interpreted as variations in the lower/middle cloud layer with either higher opacity, or different particle size. [40] The fifth component (2.2% contribution) displays very unusual peak profiles in the atmospheric windows. In particular it detects a structure in the 1.27 – 1.29 mm peak (with different responses to the O2 emission and to the aerosol window), but also in the 1.73 – 1.76 mm peak (opposing the two wings of the peak), inside the 2.2– 2.5 mm structure and to a lesser extent in the 1.10 and 1.18 mm peaks. The spatial coefficients image is dominated by a strong gradient in the left part of the image. Coefficients are distributed around zero, therefore this component is either added or subtracted to the first ones. Altogether, this is the signature of a gradual shift of spectral registration along the spatial axis of the detector, which translates as a derivative effect in the sharp peaks and in the 2.3 mm structure. [41] The sixth component (2% contribution) again exhibits unusual peak profiles, this time opposing a central minimum to maxima in the wings of the peaks. This is the typical signature of a second derivative effect (Laplacian), demonstrating that the spectral misregistration is not uniform along the detector. Spatially, it mostly affects the right part of the image. Both registration effects are small, less than 1 spectral step wide. [42] The seventh component (1.5% contribution) is dominated by strong negative contributions from the 1.74 and 1.28 mm peaks, and positive contributions from the 1.02, 1.10, and 1.18 mm peaks. It is also opposed to the black body emission at long wavelengths. Comparison with the altimetry map demonstrates that this component includes a surface contribution, originating from the positive contributions in the short-wavelength peaks (Figures 3b and 3c). The comparison is particularly good in the southernmost region, where the emergence angle is minimum. Two

7 of 20

E00B27

ERARD ET AL.: MULTIVARIATE ANALYSIS OF VENUS NIGHTSIDE

E00B27

Figure 4. (a) Main components for session 96_01, acquired in JHK mode, acquired at high to low latitudes under large opacity and (b) spatial coefficients for these components in image format (west is on the right side). The last frame is the radiance measured at 1.74 mm plotted for comparison. summits in Themis Regio (Shiwanokia and Shulamite Coronae) are identified without ambiguity, as well as smaller reliefs, although they are relatively modest in size (culminating at
2000 m over the datum). Depressions are also identified clearly, e.g., the one at 45°S, 245°E, which is 1600 m below the datum.

[43] The night sessions include much more details when acquired with longer integration times, owing to the higher signal-to-noise ratio available in the short-wavelength peaks. The components are in this case related to 3 different atmospheric structures, O2 emission from high altitude, and a surface component. The three atmospheric components are: (1) the average atmospheric spectrum, varying with

8 of 20

E00B27

ERARD ET AL.: MULTIVARIATE ANALYSIS OF VENUS NIGHTSIDE

normal brightness and controlled by the overall opacity; (2) the large-scale variations followed by all atmospheric windows and the black body emission, correlated with the emergence angle; (3) small-scale features mainly observed in the 2.3 mm structure. The latter two introduce small variations around the main contribution, but represent most of the signal variability. In addition, the O2 contribution is extremely variable depending on the session, and can reach a 20% contribution. The very low and uniform opacity for this session highlights the variation of the signal with viewing geometry at all wavelengths, and allows to detect departures from this variation. Those are mainly observed at 1.02 and 1.10 mm, and at 1.74 and 2.3 mm. The former are identified as surface relief from their correlation with Magellan topography, the latter are tentatively interpreted as small heterogeneities in the lower cloud layer. The way various phenomena are detected and separated is based on differential detection, and is discussed in more details below (sections 4.3 and 4.4). [44] Component 4 highlights small-scale structures about 100 km wide mostly related to the 2.3 mm peak intensity. This is reminiscent of holes observed in the lower/middle cloud layer, which are interpreted as localized evaporations [see McGouldrick and Toon, 2007]. There are two differences with these structures however: first, both positive and negative anomalies exist at this scale; second, these anomalies have much smaller size than the structures observed by NIMS on Galileo, for instance. Brighter structures in Figure 3a, which are larger in average, could correspond to holes in the cloud layer previously reported; the smaller dark spots in Figure 3a have larger opacity, and may be discrete clouds. [45] The surface contribution is clearly detected in this session, mainly from its contribution to the 1.02 mm peak. As demonstrated here, elevation can be retrieved down to very low values, close to the minimum elevation on Venus, at least in situations of low opacity. This will be discussed further below. A last point worth noticing is that spectral misregistration, although detected with remarkable accuracy (including second-order effects) does not preclude the detection of lesser patterns in the data, because the various phenomena are efficiently decoupled by the ICA. 3.3. South Polar View [46] Session 96_01 encompasses low to high latitudes from –10° to – 80° around the 0h meridian, which is located at longitude 210°. Although it does not extend to the pole the session covers a part of the polar vortex, which is not centered at the pole. The range of emergence angles is even larger than on the previous example and the opacity is more typical of what is usually observed by VIRTIS, with a marked zonal cloud pattern. Five among the first seven components have continuous spatial distributions (Figure 4). [47] The first component (44% contribution) dominates all the atmospheric peaks, including the secondary ones at 1.31 and 1.51 mm. It is interpreted as previously as the main Venus signal, with modulation by the lower and middle clouds. In contrast to the previous session though, it has only a limited contribution in the thermal range (
30% above 3 mm). This difference is partly due to the larger opacity variations but mostly to the presence of the warm polar area, which decouples the peaks from the black body.

E00B27

Although the component is controlled by the 1.74 and 2.3 mm peaks, the very prominent cloud pattern observed in the polar region at these wavelengths is much attenuated. It is preserved only at midlatitudes (equatorward of 50°S) where the intensity of the two peaks varies in proportion. Apart from these variations, the component is globally correlated with the emergence angle, like short-wavelength peaks are, and to a lesser extent with local time. [48] The second component (9% contribution) is related to a pair of hot pixels which behave erratically in this session. This information is entirely separated from the actual signal from Venus. [49] The third component (14% contribution) is dominated by the 1.74 and 2.3 mm peaks only, with no thermal contribution. Since it has positive coefficients with a negative spectrum, it represents a decrease of radiance in these peaks (large values correspond to large opacity). The component is mainly related to varying relative contributions in the two peaks, while the short-wavelength contribution compensates the large-scale dependence with emergence and local time. The spatial distribution mostly opposes a polar collar (with maximum values) to both the vortex and a middle to high latitudes zone (55°S to 70°S), but it also evidences a finescale concentric structure inside the polar vortex. This pattern is essentially present in the distribution of the 1.74 mm radiance, therefore this component is certainly responsive to varying cloud opacity at high latitudes, perhaps amplified by photometric effects or a different size distribution. [50] The fourth component (9% contribution) has a strong signature at short wavelengths, in opposition to the longwavelength peaks, and no thermal contribution. The spatial pattern is rather complicated: it mainly opposes middle and high latitudes (50°S) to a zone of minimum values from 30°S to 40°S, but also two concentric areas in the vortex itself (delimited at 78°S). Besides, it evidences a nonzonal opposition in the external area of the vortex and the polar collar; this latter pattern is congruent with a similar distribution of component 5, with a subtle boundary located at
195°longitude, running from the pole to 50°S. It suggests the presence of a high-altitude haze in the polar areas, although it is only present in the 1.74 mm peak. The concentric pattern observed in the polar vortex is not present in maps of the 1.74 and 2.3 mm radiance, and appears to result from different variations at these two wavelengths. According to Carlson et al. [1993], radiances at 1.74 and 2.3 mm vary with the emergence angle and their ratio increases with larger particle size [see also Wilson et al., 2008]. The photometric effects are expected to be very small in the polar areas, owing to the narrow range of emergence, so that the variability in this component is mainly related to particle size variations: the outer vortex area appears to be dominated by larger particles, while smaller ones are found in average in the inner region; but both areas are made of interleaved rings with changing particle size. A similar opposition occurs at lower latitudes, on much larger scales. [51] The fifth component (11% contribution) is like the first one essentially positive, but has near-zero contribution from the 2.3 mm structure, significant signal at short wavelengths, and the only noticeable thermal contribution (this is the main contributor longward of 2.9 mm). It is therefore controlled by the temperature of the upper layer. Short-

9 of 20

E00B27

ERARD ET AL.: MULTIVARIATE ANALYSIS OF VENUS NIGHTSIDE

E00B27

Figure 5. Maps of components (a) 3 and (b) 4 in session 96_01, evidencing the concentric structure in the polar region. wavelength peaks are associated with it because they reach maximum values on the vortex, suggesting a different size distribution. They also appears to be affected by especially intense limb darkening, which is centered at the pole in this view. Spatially, the component opposes the vortex area to the rest of the scene, with a border located at about – 75°S. This component also includes a feature observed only at long wavelengths: tenuous, narrow features suggesting high-altitude, warm clouds ranging down to 55°S. [52] The sixth component (3.9% contribution) describes a severe cosmic ray event on the matrix, extending in the spectral and spatial dimensions but limited to a single time step (3 spectral channels wide, extending on
20 pixels). [53] The seventh component (4.8% contribution) is dominated by the short-wavelength side of the 1.27 mm peak, in opposition to the other atmospheric windows. It represents the O2 airglow, with a very strong limb signal and a characteristic nonzonal, high-altitude cloud pattern entirely decoupled from the other components. [54] This example differs from the previous one by higher opacity and larger extension toward the south pole. The analysis results in five main components all related to the atmosphere: the bulk signal and its main variations (component 1); two components related to additional opacity variations (components 3 and 4, see Figure 5); a high-altitude component related to polar temperatures and limb darkening (component 5); the O2 emission, as usual completely independent and easily separated (component 7).

Not unexpectedly, the large opacity precludes the observation of surface features (see Figure 6 and section 4.5). [55] Photometric dependence is expected to be different in each atmospheric window depending on opacity [see Grinspoon et al., 1993], so that the relationship between the peaks is a function of emergence angle. In this session with large opacity variations, the variations at large emergence are included in components 1 and 5, while opacity variations under low emergence define another component 3. Besides, a changing size distribution also affects the photometric behavior, and is included here in components 4 and 5 (the latter mostly driven by temperature variations of the upper layer in the polar area). Similar secondary photometric variations in the northern polar region have been interpreted by Carlson et al. [1993] as related to the ratio of mode 20 and mode 3 particles in the clouds. The map of component 4 (Figure 5b) is actually similar to their image of branching parameter in the northern hemisphere. Such a variation can result either from varying particle densities in individual layers, or from changing layers thickness. [56] The general pattern observed in most sessions covering the vortex typically defines five main areas: [57] 1. A middle- to low-latitude zone (50°S to 10°S) which usually displays whirling patterns of middle to lower clouds and includes mottled structures due to gravity waves. The southern limit roughly corresponds to a sharp transition in zonal velocity [Sanchez-Lavega et al., 2008]. [58] 2. A high-latitude zone (70°S to 50°S) with more regular flow.

10 of 20

E00B27

ERARD ET AL.: MULTIVARIATE ANALYSIS OF VENUS NIGHTSIDE

Figure 6. Magellan topography for sessions 96_01 and 112_01. The background map is a lowresolution version outside the area of interest. The scale is in km.

Figure 7. Matching spectral components in sessions 84_01 and 84_03, two parts of a movie covering approximatively the same area. Apart from the fourth one which concentrates a small, changing, dayside contamination, these spectral signatures are very similar between the two sessions. 11 of 20

E00B27

ERARD ET AL.: MULTIVARIATE ANALYSIS OF VENUS NIGHTSIDE

E00B27

E00B27

Figure 8. Maps of component 1 in (a) session 84_01 and (b) session 84_03. The same components are retrieved by ICA, although in a different order. Paired components such as those have nearly identical spatial distributions, demonstrating that the method is extremely robust to noise, random events, and small variations in the signal. [59] 3. The polar collar (74°S to 70°S) here delimited on components 3 and 4. [60] 4. The polar vortex (74°S), globally warmer, which appears composed of an external and an internal areas separated at latitude
78°. The contrast is related to the average particle size, with larger particles in the outer vortex. All the vortex is made of many concentric rings with contrasted opacity. This concentric pattern at high latitudes is illustrated by the polar maps of components 3 and 4 in Figure 5.

4. Discussion 4.1. Robustness and Consistency [61] The comparison of successive cubes of the same movie demonstrates a very strong stability of the signal decomposition relative to noise, random events such as cosmic rays, and small variations of signal contents. This is illustrated by ICA of sessions 84_01 and 84_03 on the south polar area, which have been acquired 80 min apart and cover slightly different areas. Five of the first six components are related to the main atmospheric signatures (Figure 7): the lower/middle cloud signature bearing most of the variability (component 1, inverted on the plots); the warm and strong scattering spectrum typical of the vortex (component 2); the middle/upper clouds component which always dominates the signal (component 3); the O2 emis-

sion component (component 2); a specific day side spectrum which is observed in all sessions approaching the terminator, and apparently also includes the particle size variations. [62] Although paired components represent nearly the same fraction of the signal in both sessions, they do not appear in the same order because their histograms are slightly different, and because random events can produce intermediate components. Components describing the same information in two such sessions can be paired by looking for the best correlations between the two sets of components, and rejecting those which do not have an expression in both sessions. The maps of paired components usually compare very well. Figure 8 shows the maps of the first component in sessions 84_01 and 84_03, plotted with the same scale. The same cloud structures are present and easily identified, with little deformation at this time scale. Other components have equally similar maps, including the O2 emission which varies at a time scale of some hours. [63] In this typical example, cosmic rays and random signal from hot pixels vary noticeably between the two sessions, and produce some of the major components in only one session. However, the main physical components are still identified and their signatures are preserved (see also components 2 and 6 in session 96_01 above, Figure 4b). Of particular interest is the robustness of the analysis to changing information content. In this case, a larger portion

12 of 20

E00B27

ERARD ET AL.: MULTIVARIATE ANALYSIS OF VENUS NIGHTSIDE

E00B27

Figure 9. Four selected components from session 110_05 at short wavelengths. The first three are from the atmosphere. Component 7 carries the surface signal and is multiplied by 10 for clarity. This plot illustrates the separation of the 1.27 and 1.28 mm channels (components 3 and 2) and the differential detection performed by the algorithm. The scale in radiance provides the actual contributions of the components. Solid arrows indicate secondary atmospheric windows which are associated to the main signal in all sessions. of the observed area is contaminated by day side signal in the first session, which modifies the spectrum of this component (it contains larger scattering and absorption signatures). In spite of this, the other spectral components remain unaffected. [64] Another, more critical assessment of robustness consists in comparing several analyses of the same cube, with different type of filtering applied. At this level of calibration, several defective pixels and clusters are present and left uncalibrated in the cubes (either saturated pixels set to 0, or hot pixels with occasional abnormal output level). We compare here analyses of these cubes in their complete form, and after removal of the columns presenting the largest problems. A first situation is encountered when hot pixels affect spectral channels lying outside the atmospheric windows: in this case their are described by a specific component, and removing these columns do not change the definition of the other components. The second situation occurs when dead pixels affect channels inside the atmospheric windows, which takes place at least in two instances: a cluster affecting 3 columns and 2 channels located at 1.10 mm; another cluster affecting 2 columns and 2 channels located at 1.19 mm. Both fall in the short-wavelength peaks, which include atmospheric and surface contributions. When these pixels are preserved and set to 0, components which are defined mainly through a specific spectral shape in the corresponding range (surface contributions in both cases) have coefficients 0 in these columns; other components which include a minor contribution from these channels have slightly different values in these columns. The main result is therefore that components defined by the dead pixels include a part of the information relevant to the shortwavelengths peaks, which is especially sensitive when the surface signal is very apparent (e.g., session 411_02 analyzed below). In this case, the surface signatures are diluted in several components, some of which are driven by instrumental effects. In such situations at least, filtering of the affected columns is clearly preferred to simplify the analysis and its interpretation. [65] When dead pixels are filtered, the overall results improve but are not entirely different. Components domi-

nated by other spectral ranges (e.g., cloud opacity or thermal emission) are almost unchanged spectrally and have the same coefficient maps, as well as the same contribution to the signal. Components dominated by the short-wavelength peaks may be merged, and this information is then more readable. In session 411_02, three of the first ten components appear to include a strong surface signal when dead pixels are present, whereas the last two are combined when dead pixels are filtered, and represent about the same contribution to the signal. These contributions will be analyzed below, but the important point here is that the current method appears extremely robust to the addition of spectrally localized artifact, even when they affect essential spectral signatures: in the worst case, only the affected signatures are split among several components but they are still identifiable and the other ones are unchanged. 4.2. Minor Atmospheric Windows [66] The component which carries the main contribution to the signal always corresponds to the emission of the lower atmosphere modulated by the cloud opacity. It therefore includes peaks in the main atmospheric windows, and sometimes a weak contribution from scattered daylight at short wavelengths. Other, subdued features appear to be systematically present in the spectrum of this component. Three such features are of particular interest: two small features centered at 1.51 and 1.55 mm, and a wide feature ranging from the 1.74 mm peak up to 1.86 mum, with secondary maxima located at 1.78 and 1.82 mm (Figures 9 and 10). The 1.51 mm feature and the 1.78 mm maximum have intensity ranging from 20 to 30% that of the 1.31 mm window, and the 1.55 and 1.82 mm features are another factor of 3 fainter in general. The widest of these signatures is the 1.51 mm one, which is detected on two spectral channels, and is on the order of 20 nm wide (FWHM). [67] These features are not systematic artifacts, and they are not observed in the flat field nor in the calibration data. They are always associated to the main component, even when scattered daylight is large enough to define a specific component (e.g., component 4 of session 84_01, Figure 7). The intensity images in these channels, although very noisy, are correlated to those of the main atmospheric windows,

13 of 20

E00B27

ERARD ET AL.: MULTIVARIATE ANALYSIS OF VENUS NIGHTSIDE

E00B27

Figure 10. Individual spectra in session 110_05, sampled from the dark sky to the disk across the limb. The average noise is about 150 mW m2 sr1 mm1 in each channel. The 1.51 and 1.55 mm features are very faint atmospheric windows which have the same spatial distribution as the 1.74 mm peak. These features are detected on component 1 (solid line). The 1.58 mm O2 emission feature is unambiguously detected at the limb and stands above the noise without summation in this case. The feature is detected on component 3 (dashed line). and do not increase toward the terminator, like scattered daylight does. This behavior is different from a probable artifact which is easily identified at 1.44 mum: this latter feature is present on several components, and is particularly marked in the one representing scattered daylight. The 1.44 mm location also corresponds to the border of a filter in the M-IR optical path, and is therefore expected to be affected by spurious effects in some situations. Finally, similar peaks are present at 1.51 and 1.52 mm in a theoretical spectrum published by Kamp et al. [1988], although these small features are not discussed in this paper. [68] We therefore interpret the features at 1.51, 1.55, and 1.77 – 1.86 mm as minor atmospheric windows, where cloud opacity is not large enough to block the radiation from the underlying layers completely. Their intensity is very small and they hardly stand above the noise in VIRTIS measurements (Figure 10), which explains why they have not been reported from previous observations. Owing to the dissymmetry of the 1.74 mm peak, the secondary maximum at 1.78 mm could be related to an absorption on the shoulder of the peak; such absorption is expected, e.g., from H2O. Although their intensity is very small, these windows may prove to be useful to study atmospheric composition in a specific range of altitude. 4.3. O2 Airglow [69] Recombined O2 is detected through its main emission at 1.27 mm, partly overlapping the 1.28 mm atmospheric window at VIRTIS resolution [Ge´rard et al., 2008]. The specific O2 component has therefore its main positive contribution from this channel. The O2 emission is concentrated in some areas where is dominates the signal in this channel. In other regions the lower cloud pattern is visible at this wavelength, and is extremely similar to the one measured at 1.74 and 2.3 mm, which is incorporated in another component. In order to optimize the independence of the components, the algorithm subtracts a part of the flux measured in the atmospheric windows to remove the cloud pattern from the O2 component. Conversely, a part of the flux at 1.27 mm is subtracted from the main cloud components (Figure 9). The resulting O2 maps are usually decoupled from other components even when the underly-

ing vortex and cloud pattern is very strong (e.g., Figure 4a). The 1.27 mm band of O2 is almost always observed on nightside sessions, except in rare occurrences (e.g., session 112_00). [70] Another, much fainter signature of O2 is located at 1.58 mm, and is detected by VIRTIS in some instances [Piccioni et al., 2008]. In session 110_05 analyzed above, component 2 is responsive to O2 emission. This component is the only one which has a larger relative weight than component 3 in a part of the spectral range. Interestingly, this occurs not only at 1.27 mm, but also at 1.58 mm in the fainter O2 band. Indeed, a very marginal maximum is detected around the limb at this wavelength. Examination of the individual spectra however clearly evidence this feature without summation, at a level which is at most 10 times the noise (Figure 10). The ratio of integrated bands measured on this component is 102?20, to be compared with a value of 78?8 measured in better conditions at the limb in orbit 317 by Piccioni et al. [2008]. The spectral component is therefore representative of an actual emission spectrum. What is remarkable here is the ability of ICA to identify and locate an extremely small signature which is present in less than 1% of the spectra. More usual variance analysis methods would not extract such marginal features from the noise. 4.4. Surface [71] A surprising characteristic of the data set is that the surface signature is clearly observed in many sessions, at least in a part of the observed areas. This signature stands out readily with ICA. The identification is unambiguous because, although based on spectral interpretation, it can also be checked by comparison of the coefficient maps with the GTDR. The strength of this surface contribution is due to VIRTIS improvements in terms of spatial resolution and signal to noise with respect to previous observations. [72] Specific study of surface elevation from VIRTIS data [Mueller et al., 2008] is based on the short-wavelength peaks intensity, mostly at 1.02 mm. It includes subtraction of scattered light measured between the short-wavelength peaks, correction of limb darkening in these peaks, and removal of the cloud pattern. The limb darkening rate is the

14 of 20

E00B27

ERARD ET AL.: MULTIVARIATE ANALYSIS OF VENUS NIGHTSIDE

E00B27

Figure 11. Component 2 in session 112_0, the main atmospheric signal. Idunn Mons is visible as a circular spot with lower intensity at
215°E, 45°S. variation with emergence fit on the data; cloud pattern removal is performed through division by the radiance measured in a channel where the surface contribution is negligible, and accounts for multiple reflections between the surface and the atmosphere (following a two-stream model by Hashimoto and Sugita [2003]). This process involves assumptions of atmospheric temperature, cloud reflectivity, and surface emissivity, which have to be adjusted on the basis of the data. [73] The ICA performs similar corrections, although because ICA is a linear analysis, they are only based on differences of radiance at various wavelengths. As expected, the surface is only detected through the short-wavelength peaks, which have the only positive contributions. The interpeak radiance is subtracted to remove scattered light, a large-scale limb darkening correction is performed through removal of the black body emission, and radiance in the long-wavelength atmospheric windows is subtracted to minimize the remaining cloud pattern. An enlargement at short wavelengths is provided in Figure 9 for session 110_05 (component 7). The derivative effect seen in the short-wavelength peaks corresponds to a larger surface contribution from the long-wavelength side of the peaks (according to Meadows and Crisp [1996]). The ICA detection is expected to be more robust than estimates based on a single peak intensity, because it takes advantage of all the surface-related information in the spectra. The result is however not directly given in terms of black body irradiance; it is a weighted average of all contributing wavelengths, and is therefore not easily translated as a surface temperature. [74] In the data files analyzed so far, three situations may occur: (1) no surface component identified; this is usually related to short integration times (low signal-to-noise ratio in the short-wavelength peaks) and high opacity, (2) limited detection, usually in situations of high opacity (translating in intense short-wavelength peaks), or (3) complete detection, in sessions with low opacity. In the latter case, even the main atmospheric components may actually include a surface contribution identified in the coefficient maps. [75] Examples of high opacity situations include sessions 96_01 (discussed in section 3.3), 112_0, and 84 (discussed in section 4.1).

[76] Session 96_01 encompasses an area with little marked topography, except for Atahensik Corona in Aphrodite Terra (170°E, 20°S) which is observed under extremely large air mass, and an isolated plateau (Imdr Regio) which culminates with Idunn Moons at more than 3000 m (215°E, 45°S). The rest of the area is made of low plains with subdued topography, ranging from – 1600 to 0 m. As mentioned in section 3.3, the atmospheric components include intense scattering at short wavelengths. However, careful examination of components 1 (the lower to middle clouds comprising the bulk of the signal), 3, 4, and 5 (main atmospheric variations) shows a circular spot of low values over Idunn Mons (Figures 5a and 5b, and comparison with Figure 6). On component 1, the large elevation contrast in the Atahensik Corona area is also barely detected under large emergence. Session 112_00 is another polar observation with a large overlap on the previous one, in particular over Idunn Mons. Again, the summit is clearly identified as a circular feature on the main middle to high clouds component (Figure 11). Other possible surface detections may exist in these sessions but are marginal or not entirely convincing. [77] In both cases an indirect detection is performed, the highest relief being detected as a local minimum of radiance in the main atmospheric signature. This is surprising. In both sessions, the feature is present in the 3 peaks at 1.02, 1.10, and 1.18 mm. The main atmospheric component is particularly responsive to the 1.18 mm peak and exhibits more contrast than direct measurements. The detection is thus not related to a particular spectral signature, but to the thermal flux from the surface: the presence of a topographic high results in missing signal from the strongly scattering layers located at much higher altitude. The surface therefore always contributes significantly to the outgoing flux, and colder local boundary conditions at the bottom of the atmosphere over higher areas translate as local minima of the outgoing flux. Consequently, low areas surrounding Idunn Mons are expected to be visible, and even more contrasted because of their higher temperature, but these variations are not observed. [78] Because Idunn Mons is detected with its apparent size unaffected by blurring processes, we can conclude that blurring by cloud layers only affect spatial scales smaller

15 of 20

E00B27

ERARD ET AL.: MULTIVARIATE ANALYSIS OF VENUS NIGHTSIDE

E00B27

Figure 12. Cylindrical projection of component 9 (surface related) from session 411_02: (a) projection on the ellipsoid, (b) comparison with Magellan altimetry at the same resolution, and (c) projection of the data on the upper cloud layer at 60 km altitude. Atahensik Corona appears slightly displaced on Figure 12a, which is ascribed to refraction effects, while the correlation is much better for Figure 12c. The background image and contours are the low-resolution topography from Magellan. than this (
100 km). This blurring scale is not expected to vary greatly with surface elevation, which is small compared to the altitude of the cloud layers. In order to mask the contrast in low-elevation areas, blurring should take place below the summit of Idunn Mons, which implies the presence of a high opacity layer near the surface in some occasions. This is reminiscent of reanalysis of Venera 13/14 descent measurements which indicates a layer of increased extinction at altitude
1.5 km above the surface, in lowplain areas [Grieger et al., 2004]. [79] Similar observations are performed on several occasions in the Idunn Mons area, while larger elevation contrasts are discernible over Aphrodite Terra in the same sessions (e.g., Atahensik Corona in session 96_01, Arthemis in session 84_03). This suggests that the inferred scattering layer is occasionally present only on topographic lows of the southern hemisphere, but not on highlands at lower latitude. Since low reliefs areas are observed in some instances (e.g., session 110_05, Figure 3a) this blurring is unlikely to result from CO2 absorption or Rayleigh scattering at the bottom of the atmosphere, which are related to pressure and are not expected to vary greatly with time. Aerosols present at high altitude are not stable at such pressures, therefore the inferred scattering is most likely related to uplifted dust in the first kilometer above the surface. Another possibility is related to the condensation of volatile metals at the bottom of the atmosphere [Schaefer and Fegley, 2004]: in particular, condensation of Pb or Bi sulfides at 2.6 km altitude have been considered consistent with Venus geochemistry, and could form a scattering layer above the low plains. [80] The surface is more visible when scattered daylight is reduced, i.e., when most components are flat at short wavelengths. This is the case for session 110_05 discussed above, but also for a set of sessions acquired during orbits 410 and 411 far from nadir viewing, when the atmosphere was particularly transparent. These sessions encompass the region around Atahensik Corona and Rusalka Planitia, including the Vega 2 landing site. The components are similar to those retrieved for session 110_05, with a specific surface detection in component 9 mostly responsive to the flux at 1.02 mm. The strong similitude of the coefficient map with Magellan altimetry confirms the surface origin of

this component (Figure 12). In particular, the very specific surface features around Atahensik Corona are clearly identifiable. [81] Comparisons between the GTDR surface features and those observed by VIRTIS allow to assess atmospheric

Figure 13. Surface-related information in session 411_02. (top left) Magellan Global Topography Data Record averaged on pixels surface (then box averaged on 6 pixels); (top right) measured radiance at 1.03 mm (inverted); (bottom right) measured radiance at 1.185 mm (inverted); (bottom left) component 9, inverted (filtered clusters appear as white columns). Although the main surface features are clearly identified in the radiance images, they are mixed with marked, unrelated cloud patterns. The surface component retrieved by the ICA is a simple linear combination of channels which minimizes these patterns and improves the contrast in some areas (e.g., bottom right corner).

16 of 20

E00B27

ERARD ET AL.: MULTIVARIATE ANALYSIS OF VENUS NIGHTSIDE

E00B27

Figure 14. Components 8 (solid line) and 10 (dashed line) in tangential limb session 43_00, which are related to different regimes of CO2 fluorescence in the upper atmosphere. scattering effects. Session 411_02 is particularly adapted to this purpose, because of the low opacity and large emergence angles. The good fit at the limb between signal cutoff and emergence computed on the upper cloud layer demonstrates that the data are registered with an accuracy better than one pixel (Figure 13). However, the observed surface features are slightly displaced limbward, and the shift increases with the emergence angle in the northern direction (Figures 12a and 12b). The first version of the GTDR [Ford and Pettengill, 1992] (distributed by the PDS Geoscience node) is used in the archive; further refinements of the GTDR from improved Magellan orbit reconstruction [e.g., Rappaport et al., 1999] do not fit better: relief locations are identical in successive versions, with differences concentrated at the sharp transitions. However, the shift disappears when the data are projected not on the ellipsoid, but on a 60 km altitude shell representing the top of the scattering atmosphere (Figure 12c). This confirms that light travels mostly in the vertical direction through the atmosphere, owing to intense scattering effects. [82] The strong bending of light rays in the atmosphere can be described using an effective refraction index. The shift of about 4 pixels (2.5°) observed for an emergence of
50° on the cloud top corresponds to a refractive index (at 1 – 1.20 mm) ranging from 1.03 to 1.04. Extrapolation of Earth’s variations at Venus pressure and temperature conditions yields a value of
1.003. The much larger value observed by VIRTIS is related to strong scattering on Venus, and perhaps to the compositional difference. Finally, the data compare in resolution with the GTDR averaged on the pixel surface and convolved with a 6 pixels box (Figure 13). This convolution kernel provides an estimate of blurring effects in the atmosphere in these particularly favorable conditions, with a PSF about 5° wide (FWHM) for an emergence of 50° (or
1.5 airmass). The corresponding resolution at the surface is
35 km, and is probably close to the maximum achievable with this observing technique. 4.5. Lightning [83] Optical detection of lightning has been performed by the Venera 9 visible spectrometer [Krasnopolskii, 1983] then from Earth-based observations [Hansell et al., 1995]. Visible spectra are roughly white, with radiance at least 4 times that of neighboring areas. Electromagnetic wave detections suggest that the activity is concentrated near the dusk limb (see review by Russell et al. [2006]).

[84] Electromagnetic activity is reported routinely from Venus Express, and interpreted as lightning detection [Russell et al., 2007]. VIRTIS could therefore observe flashes during nighttime observations, possibly concentrated near the dusk terminator. The frequent detection of cosmic rays and other punctual features by ICA shows that lightning with a specific signature should be readily detected if present in the cubes, as long as they do not saturate the detector. ‘‘Specific’’ in this context may mean an unusual intensity ratio in the atmospheric windows. If lightning events have regular, simply brighter spectra, they would appear as isolated maxima on one of the main components and may be difficult to identify. [85] So far no firm detection has been found, although not all transient events in the data set are fully understood. Because the instrument is a slit spectrometer with limited field of view, the probability of detection is also rather small. Positive detection, or even reliable statistics, may require the analysis of a large number of sessions, and this activity is still ongoing. 4.6. Limb Observations [86] Dedicated limb sessions allow study of the upper atmosphere and its vertical profile. Although deexcitation of newly recombined O2 is observed on the nightside, most signatures are due to fluorescence and are present only in daytime limb observations. [87] We focus here on session 43_00, a tangential limb observation acquired at short integration time (0.36 s). This mode is adapted to long-wavelength studies, while the short wavelengths are affected by noise, by a nonlinearity pattern in the dark current, and by intense scattering. The present analysis is therefore limited to the 3.5– 5.12 mm range. The data are acquired over the northern hemisphere, in the Guinevere Planitia quadrangle, entirely during day time. [88] Even in this situation where the signal is small and concentrated on restricted locations, the actual signatures are correctly separated from the noise and artifact. The largest contributions are located below 70 km and are related to thermal emission and CO2 absorption in the cloud layers. Two components appear to be controlled by CO2 emission only: they dominate the signal in the 4.2– 4.5 mm spectral range, and are only marginally affected by other wavelengths (Figure 14). [89] Component 8 is the main contribution to the CO2 complex at 4.2– 4.5 mm, peaking at 4.32 mm and 4.35 mm,

17 of 20

ERARD ET AL.: MULTIVARIATE ANALYSIS OF VENUS NIGHTSIDE

E00B27

E00B27

Figure 15. Coefficients for the two components in Figure 14 as a function of tangent altitude and incidence angle. Larger contributions are figured in white tones. with a large additional contribution to the long-wavelength side of the complex, peaking at 4.44 mm (Figure 14). Component 10 is a lesser contribution (
half of the previous one) with a distinct spectral shape peaking at 4.28 and 4.32 mm. These spectral shapes are consistent with predictions of non-LTE models ([Lo`pez-Valverde et al., 2007], see discussion by Gilli et al. [2008]). [90] These two components are detected at latitudes in the 60°– 75°N range. Figure 15 shows their distribution as a function of tangent altitude and incidence angle (i.e., solar zenith angle). The two components appear to be spatially decoupled, with the major contribution peaking at
110 km, and the low-wavelength one at 130 km, again in agreement with theoretical predictions. The attenuation with increasing incidence angle is also predicted. These plots compare very well with the band intensity plots in the work by Gilli et al. [2008, Figure 7]. A third component exhibits significant variations with altitude above the 80 km level, but it includes strong thermal and absorption signatures; it more likely represents the cold collar surrounding the polar vortex at latitudes 60– 75°N. Finally CO fluorescence peaking at 4.68 and 4.78 mm does not stand out clearly, but could be mixed with thermal emission in other components; it appears to be difficult to detect at the spectral resolution of VIRTIS-M. Apart from the strong, main O2 band at 1.27 mm, other limb signatures detected with VIRTIS are very faint and require spatial summation to be identified [see Piccioni et al., 2008]. Those are not identified in limb data by ICA so far.

5. Conclusions [91] The VIRTIS-M infrared observations of the nightside and limb of Venus have been analyzed with ICA [Cardoso, 1999]. A methodology has been set up to minimize instrumental effects and to interpret the results. [92] The analyses are run directly on data calibrated in radiance, with saturated pixels set to 0. Data acquired with long exposure times are analyzed in the 1.02– 3.9 mm range. Isolated dead or hot pixels have minor impact on the

components definition. However, two potential sources of artifact are filtered prior to analysis: spectra of the dark sky, and clusters of dead pixels. In this work, we removed the columns affected from the data cubes and maintained the complete spectral information whenever possible. This precaution results in a noticeable improvement of the analyses: the spectral components are more easily interpreted, and their spatial distributions are more consistent. [93] The method proves to be extremely robust to noise and to large, random variations (cosmic rays). Those are for the most part isolated in specific components and do not interfere with the signal analysis. Typically, 5 to 7 components among the first 10 are related to the signal from Venus, and the others describe instrumental effects and transient events in the data. Uncalibrated instrumental effects are usually clearly identified and are decoupled from the signal in good approximation, at least in low opacity situations. In the current stage of data calibration, the main effects include irregular spectral registration along the spatial dimension of the detector, with first (gradient) and second (Laplacian) orders of deformation directly evidenced. [94] The ICA always identifies the bulk signal from the atmosphere, modulated by the major opacity variations: zonal cloud pattern and limb darkening. Limb darkening may be detected separately whenever the cloud pattern is reduced, i.e., in situations of low opacity. The main component represents between 40 and 60% of the signal in general, but seldom introduces the largest heterogeneity between spectra. All contributions in the atmospheric windows are grouped in this component, which evidences faint, nearly opaque windows not reported before Venus Express. Such signatures are confirmed at 1.51 and 1.55 mm, with intensities
30% and 10% of the 1.31 mm window. Besides, significant signal is measured on the longward foot of the 1.74 mm peak, up to 1.86 mm, with secondary maxima located at 1.78 and 1.82 mm. [95] Other components are defined either through intensity balance in the main atmospheric windows (at 1.74 and 2.3 mm), or through differential detection of small signatures. In polar views, the peaks’ intensity (controlled by

18 of 20

E00B27

ERARD ET AL.: MULTIVARIATE ANALYSIS OF VENUS NIGHTSIDE

limb darkening) and the black body emission (with a strong maximum in the polar region) are detected together on a separate component. At high latitudes, two additional components are required to describe the main variations in the atmospheric signal: one which responds to opacity variations seen under low emergence and related to the difference of photometric functions in the long-wavelength peaks, the other one similar to branching parameter maps, and related to changing particle size distribution. The latter displays the structure of the polar vortex, which appears to be composed of two main concentric areas, with larger average particle size in the outer shell; the two shells of the vortex are composed of many concentric rings with a similar opposition at a smaller scale. [96] Local atmospheric features are evidenced by the analysis in specific conditions. Under clear atmosphere, small-scale structures are detected separately in the cloud layers at 2.3 mm at low latitudes. They are consistent with localized evaporations of the lower/middle cloud layer, but also suggest the presence of discrete clouds about 100 km in dimension. At high latitudes, tenuous and elongated veils are observed at 1.74 mm and in the black body emission, suggesting high-altitude, warm clouds ranging from the polar vortex down to 55°S. [97] High-altitude emission signatures from various species are correctly reported. Emission from recombined O2 is always detected separately at 1.27 mm, in association with the much fainter 1.58 mm feature observed mostly at the limb. Limb observations show CO2 non-LTE emission with its two characteristic signatures well separated, peaking at 110 and 130 km altitude. Other vertical variations are present at the limb, which are probably related to the thermal structure of the cloud layers and the upper atmosphere, and are still under study. [98] For low cloud opacity, the surface is detected directly down to the minimum elevation. Under high opacity however, only the upper summits are detected through minimum radiance in the outgoing flux. This suggests that intense scattering takes place near the surface, below the summits, in some situations. This scattering at the bottom of the atmosphere is most likely related to uplifted dust. From the limited number of sessions analyzed so far, the effect seems to be present in the topographic lows of the southern hemisphere, but not over Aphrodite Terra. Spatial resolution at the surface is limited by atmosphere blurring. The maximum resolution achieved so far, in exceptional conditions, is on the order of 35 km, and is probably close to the limit. In these conditions, strong light bending is observed with an effective refraction index on the order of 1.03. [99] This preliminary study of the VIRTIS nightside and limb observations demonstrates the power of Independent Component Analysis applied to imaging spectroscopy data of a thick atmosphere, and qualifies it to analyze spectral remote sensing data. Without any a priori assumptions, the main physical phenomena are identified and located spatially, and their relative intensities are correctly estimated. Small structures are enhanced, because the various sources of variability which affect the same spectral channels are separated, and because the signal-to-noise ratio is increased by summing all channels carrying related information. In particular, association of very faint spectral signatures are clearly evidenced (down to signal-to-noise ratios
10 or

E00B27

less), thanks to the quality of the instrument and data calibration. Specific spectral signatures with limited spatial expression are also highlighted. Finally, ICA automatically performs adequate corrections to untangle the main physical phenomena, similar to empirical methods based on best-fit parameters. [100] This ‘‘assessment study’’ is successful in the sense that the main regularities in the data set, studied independently by more elaborated means, are retrieved consistently with a blind source separation technique, down to low levels of variance and signal-to-noise ratio. Given minimal precautions, the method can therefore be confidently applied, at least as a quick-look tool, to large data sets produced by modern imaging spectroscopy experiments. Future, more detailed and systematic applications to VIRTIS data are expected to ease and enhance both dynamical studies (e.g., evolution of small clouds) and measurement of faint spectral features. [101] Acknowledgments. The VIRTIS experiment is supported by CNES and ASI. The experiment has been built and operated by the science and technical team listed at the following address: http://servirtis.obspm.fr/ Venus_Express/VIRTIS_Team.html. We gratefully thank the VIRTIS technical team for continuous and high level support as well as Bruno Bzard and Colin Wilson for constructive discussions and editorial help. The VIRTIS data archive is distributed online by ESA on the Planetary Science Archive Web site (http://www.rssd.esa.int/index.php?project=PSA&page= vexIndex).

References Allen, D. A., and J. W. Crawford (1984), Cloud structure on the dark side of Venus, Nature, 307, 222 – 224. Baines, K. H., S. Atreya, R. W. Carlson, D. Crisp, P. Drossart, V. Formisano, S. S. Limaye, W. J. Markiewicz, and G. Piccioni (2006), To the depths of Venus: Exploring the deep atmosphere and surface of our sister world with Venus Express, Planet. Space Sci., 54, 1263 – 1278, doi:10.1016/ j.pss.2006.04.034. Cardoso, J.-F. (1999), High-order contrasts for independent component analysis, Neural Comput., 11, 157 – 192. Carlson, R. W., K. H. Baines, L. W. Kamp, P. R. Weissman, W. D. Smythe, A. C. Ocampo, T. V. Johnson, D. L. Matson, J. B. Pollack, and D. Grinspoon (1991), Galileo infrared imaging spectroscopy measurements at Venus, Science, 253, 1541 – 1548. Carlson, R. W., L. W. Kamp, K. H. Baines, J. B. Pollack, D. H. Grinspoon, T. Encrenaz, P. Drossart, and F. W. Taylor (1993), Variations in Venus cloud particle properties: A new view of Venuss cloud morphology as observed by Galileo Near-Infrared Mapping Spectrometer, Planet. Space Sci., 41, 477 – 485, doi:10.1016/0032-0633(93)90030-6. Comon, P. (1990), Analyse en composantes inde´pendantes et identification aveugle, Trait. Signal, 7, 435 – 450. Connes, P., J. F. Noxon, W. A. Traub, and N. P. Carleton (1979), O2 (1 D) emission in the day and night airglow of Venus, Astrophys. J., 233, L29 – L32, doi:10.1086/183070. Coradini, A., et al. (1998), VIRTIS: An imaging spectrometer for the Rosetta mission, Planet. Space Sci., 46, 1291 – 1304. Drossart, P., et al. (2007a), Scientific goals for the observation of Venus by VIRTIS on ESA/Venus Express mission, Planet. Space Sci., 55, 1653 – 1672, doi:10.1016/j.pss.2007.01.003. Drossart, P., et al. (2007b), A dynamic upper atmosphere of Venus as revealed by VIRTIS on Venus Express, Nature, 450, 641 – 645, doi:10.1038/nature06140. Erard, S. (2008), VEx-VIRTIS to planetary science archive interface control document, Rep. VIRTIS VIR-LES-IC-2269, Lab. dEtud. Spatiales et dInstrum. en Astrophys., Paris. Esposito, L. W., R. G. Knollenberg, M. I. Marov, O. B. Toon, and R. P. Turco (1983), The clouds are hazes of Venus, in Venus, pp. 484 – 564, Univ. of Ariz., Tuscon. Ford, P. G., and G. H. Pettengill (1992), Venus topography and kilometerscale slopes, J. Geophys. Res., 97, 13,103 – 13,114. Forni, O., F. Poulet, J.-P. Bibring, S. Erard, C. Gomez, Y. Langevin, B. Gondet, and the OMEGA science team (2005), Component separation of OMEGA spectra with ICA, Lunar Planet. Sci., XXXVI, Abstract 1623. Ge´rard, J.-C., A. Saglam, G. Piccioni, P. Drossart, C. Cox, S. Erard, R. Hueso, and A. Sa`nchez-Lavega (2008), Distribution of the O2 infrared nightglow

19 of 20

E00B27

ERARD ET AL.: MULTIVARIATE ANALYSIS OF VENUS NIGHTSIDE

observed with VIRTIS on board Venus Express, Geophys. Res. Lett., 35, L02207, doi:10.1029/2007GL032021. Gilli, G., M. A. Lopez-Valverde, P. Drossart, G. Piccioni, S. Erard, A. Cardesin, and the VIRTIS team (2008), Limb observations of non-LTE emissions in the Venus atmosphere by VIRTIS/Venus Express, J. Geophys. Res., doi:10.1029/2008JE003112, in press. Grieger, B., N. I. Ignatiev, N. M. Hoekzema, and H. U. Keller (2004), Indication of a near surface cloud layer on Venus from reanalysis of Venera 13/14 spectrophotometer data, in Planetary Probe Atmospheric Entry and Descent Trajectory Analysis and Science, Eur. Space Agency Spec. Publ., 544, 63 – 70. Grinspoon, D. H., J. B. Pollack, B. R. Sitton, R. W. Carlson, L. W. Kamp, K. H. Baines, T. Encrenaz, and F. W. Taylor (1993), Probing Venus’s cloud structure with Galileo NIMS, Planet. Space Sci., 41, 515 – 542, doi:10.1016/0032-0633(93)90034-Y. Hansell, S. A., W. K. Wells, and D. M. Hunten (1995), Optical detection of lightning on Venus, Icarus, 117, 345 – 351, doi:10.1006/icar.1995.1160. Hashimoto, G. L., and S. Sugita (2003), On observing the compositional variability of the surface of Venus using nightside near-infrared thermal radiation, J. Geophys. Res., 108(E9), 5109, doi:10.1029/2003JE002082. Kamp, L. W., F. W. Taylor, and S. B. Calcutt (1988), Structure of Venus’s atmosphere from modelling of night-side infrared spectra, Nature, 336, 360 – 362. Krasnopolskii, V. A. (1983), Venus spectroscopy in the 3000 – 8000 region by Veneras 9 and 10, in Venus, pp. 459 – 483, Univ. of Ariz., Tuscon. Lo`pez-Valverde, M. A., P. Drossart, R. Carlson, R. Mehlman, and L. W. Roos-Serote (2007), Non-LTE infrared observations at Venus: From NIMS/Galileo to VIRTIS/Venus Express, Planet. Space Sci., 55, 1757 – 1771, doi:10.1016/j.pss.2007.01.008. McGouldrick, K., and O. B. Toon (2007), An investigation of possible causes of the holes in the condensational Venus cloud using a microphysical cloud model with a radiative-dynamical feedback, Icarus, 191, 1 – 24, doi:10.1016/j.icarus.2007.04.007. Meadows, V. S., and D. Crisp (1996), Ground-based near-infrared observations of the Venus nightside: The thermal structure and water abundance near the surface, J. Geophys. Res., 101, 4595 – 4622, doi:10.1029/ 95JE03567. Moudden, Y., J.-L. Starck, P. Abrial, and J.-F. Cardoso (2005), MRS: MultiResolution on the sphere, 102 pp., DAPNIA Comm. l’Energ. At., Gif-suryvette, France. Moussaoui, S., H. Hauksdottir, F. Schmidt, C. Jutten, J. Chanussot, D. Brie, S. Doute´, and J. A. Benediksson (2008), On the decomposition of Mars hyperspectral data by ICA and Bayesian positive source separation, Neurocomputing, 71, 2194 – 2208.

E00B27

Mueller, N., J. Helbert, G. L. Hashimoto, C. C. C. Tsang, S. Erard, G. Piccioni, and P. Drossart (2008), Venus surface thermal emission at 1 mm in VIRTIS imaging observations: Evidence for variation of crust and mantle differentiation conditions, J. Geophys. Res., 113, E00B17, doi:10.1029/2008JE003118. Piccioni, G., et al. (2007), South-polar features on Venus similar to those near the north pole, Nature, 450, 637 – 640, doi:10.1038/nature06209. Piccioni, G., et al. (2008), First detection of hydroxyl in the atmosphere of Venus, Astron. Astrophys., 483, L29 – L33. Pollack, J. B., et al. (1993), Near-infrared light from Venus nightside: A spectroscopic analysis, Icarus, 103, 1 – 42, doi:10.1006/icar.1993.1055. Rappaport, N. J., A. S. Konopliv, A. B. Kucinskas, and P. G. Ford (1999), An improved 360 degree and order model of Venus topography, Icarus, 139, 19 – 31, doi:10.1006/icar.1999.6081. Russell, C. T., R. J. Strangeway, and T. L. Zhang (2006), Lightning detection on the Venus Express mission, Planet. Space Sci., 54, 1344 – 1351, doi:10.1016/j.pss.2006.04.026. Russell, C. T., T. L. Zhang, M. Delva, W. Magnes, R. J. Strangeway, and H. Y. Wei (2007), Lightning on Venus inferred from whistler-mode waves in the ionosphere, Nature, 450, 661 – 662, doi:10.1038/nature05930. Sanchez-Lavega, A., et al. (2008), Variable winds on Venus mapped in three dimensions, Geophys. Res. Lett., 35, L13204, doi:10.1029/ 2008GL033817. Schaefer, L., and B. Fegley (2004), Heavy metal frost on Venus, Icarus, 168, 215 – 219. Taylor, F. W., D. J. Diner, L. S. Elson, D. J. McCleese, J. V. Martonchik, J. Delderfield, S. P. Bradley, J. T. Schofield, J. C. Gille, and M. T. Coffey (1979), Temperature, cloud structure, and dynamics of Venus middle atmosphere by infrared remote sensing from Pioneer Orbiter, Science, 205, 65 – 67. Titov, D. V., et al. (2006), Venus Express science planning, Planet. Space Sci., 54, 1279 – 1297, doi:10.1016/j.pss.2006.04.017. Wilson, C., S. Guerlet, P. G. J. Irwin, C. C. C. Tsang, F. W. Taylor, R. W. Carlson, P. Drossart, and G. Piccioni (2008), Evidence for anomalous cloud particles at the poles of Venus, J. Geophys. Res., 113, E00B13, doi:10.1029/2008JE003108. 

P. Drossart and S. Erard, LESIA, Observatoire de Paris, 5 place Jules Janssen, Meudon, F-92195 France. ([email protected]) G. Piccioni, INAF, IASF, Via del Fosso del Cavaliere 100, I-00133 Rome, Italy.

20 of 20

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114, E00B43, doi:10.1029/2008JE003320, 2009

for

Full Article

Altimetry of the Venus cloud tops from the Venus Express observations N. I. Ignatiev,1,2 D. V. Titov,2 G. Piccioni,3 P. Drossart,4 W. J. Markiewicz,2 V. Cottini,3,5 Th. Roatsch,6 M. Almeida,7 and N. Manoel2,8 Received 16 December 2008; revised 17 March 2009; accepted 29 May 2009; published 13 August 2009.

[1] Simultaneous observations of Venus by Visible and Infrared Thermal Imaging

Spectrometer and Venus Monitoring Camera onboard the Venus Express spacecraft are used to map the cloud top altitude and to relate it to the ultraviolet (UV) markings. The cloud top altitude is retrieved from the depth of CO2 absorption band at 1.6 mm. In low and middle latitudes the cloud top is located at 74 ± 1 km. It decreases poleward of ±50° and reaches 63–69 km in the polar regions. This depression coincides with the eye of the planetary vortex. At the same latitude and hour angle, cloud top can experience fast variations of about 1 km in tens of hours, while larger long-term variations of several kilometers have been observed only at high latitudes. UV markings correlate with the cloud altimetry, however, the difference between adjacent UV dark and bright regions does not exceed several hundred meters. Surprisingly, CO2 absorption bands are often weaker in the dark UV features, indicating that these clouds may be a few hundred meters higher or have a larger scale height than neighboring clouds. Ultraviolet dark spiral arms, which are often seen at about
70°, correspond to higher altitudes or to the regions with strong latitudinal gradient of the cloud top altitude. Cloud altimetry in the polar region reveals the structure that correlates with the thermal emission maps but is invisible in UV images. This implies that the UV optically thick polar hood is transparent in the near IR. Citation: Ignatiev, N. I., D. V. Titov, G. Piccioni, P. Drossart, W. J. Markiewicz, V. Cottini, Th. Roatsch, M. Almeida, and N. Manoel (2009), Altimetry of the Venus cloud tops from the Venus Express observations, J. Geophys. Res., 114, E00B43, doi:10.1029/2008JE003320.

1. Introduction [2] A remote observer sees numerous features in the ultraviolet (UV) spectral range in the top layers of Venus’s thick cloud deck. The nonuniform vertical and horizontal distribution of an unknown UV absorber produces contrast markings that have puzzled scientists since their discovery [Rossow et al., 1980; Esposito, 1980]. Variations of temperature and cloud structure above the tropopause (60 km) appear as specific features in the thermal infrared range [Taylor et al., 1980]. These features indicate temperature conditions and processes at the cloud tops, but poor knowl1 Space Research Institute for Russian Academy of Sciences, Moscow, Russia. 2 Max Planck Institute for Solar System Research, Katlenburg-Lindau, Germany. 3 INAF, IASF, Rome, Italy. 4 LESIA, Observatoire de Paris, Meudon, France. 5 Now at NASA Goddard Space Flight Center, Greenbelt, Maryland, USA. 6 Institute of Planetary Research, German Aerospace Center, Berlin, Germany. 7 ESA, ESAC, Madrid, Spain. 8 Department of Mathematics, University of E´vora, E´vora, Portugal.

Copyright 2009 by the American Geophysical Union. 0148-0227/09/2008JE003320$09.00

edge of the altitude where they form makes their interpretation difficult. This uncertainty affects altitude assignment to the cloud tracked winds and trace gas abundances. The question of cloud top altitude is also directly related to the problem of location of the unknown UV absorber. [3] Earlier observations provided several tools to determine the cloud top altitude (i.e., the altitude of unit cloud optical depth t) and aerosol structure. The measurements of solar fluxes, spectra of scattered radiation, and aerosol microphysical properties by the Pioneer Venus and Venera descent probes enabled the construction of consistent models of particulate and optical properties of the cloud layer at low and middle latitudes [Esposito et al., 1983; Tomasko et al., 1985; Ragent et al., 1985; Moroz et al., 1985]. These models placed the t  1 level in the visible range at about 70 km. Kawabata et al. [1980] and later Sato et al. [1996], Knibbe et al. [1998], and Braak et al. [2002] used the Pioneer Venus Orbiter Cloud Photopolarimeter (OCPP) observations to constrain the microphysical and optical properties and structure of the upper haze. They concluded that in the polar regions (70 – 90°N) a submicron haze with radius r  0.25 mm is responsible for UV opacity of 0.8 above the main cloud which top is located at about 40 mbar (68 km). In equatorial regions the submicron haze opacity is about an order of magnitude lower and the level t  0.8

E00B43

1 of 10

IGNATIEV ET AL.: ALTIMETRY OF VENUS CLOUDS

E00B43

E00B43

Figure 1. (a) VIRTIS-M measured (black) and synthetic (gray) spectra of the Venus dayside. The synthetic spectrum is shifted by +5 W m
2 sr
1 mm
1. Dotted lines show continuum and band depth (continuum
S(lD)). Positions of CO2 bands are shown. (b) Synthetic VIRTIS-M spectra calculated for several cloud top altitudes. is located inside the main cloud layer at 28 mbar (70 km). Thus the OCPP observations implied that the cloud top is by about one atmospheric scale height lower in the polar regions than in low latitudes. Pioneer Venus radio occultation measurements showed that the cloud tops at the pole are 100 mbar deeper than those at the equator [Cimino et al., 1980]. Thermal emission spectrometry onboard Venera-15 orbiter suggested also that the cloud top altitude at 1200 cm
1 (8.3 mm) descends from 70 km in low latitudes to 60 km in the polar region [Zasova et al., 2007]. [4] Imaging of Venus simultaneously in the near-IR by Visible and Infrared Imaging Spectrometer (VIRTIS) [Drossart et al., 2007] and in the UV range (0.365 mm) by Venus Monitoring Camera (VMC) [Markiewicz et al., 2007a] onboard Venus Express [Svedhem et al., 2007; Titov et al., 2006] provides a novel tool to determine the cloud top altitude and its variations over the planet and allows one to correlate them with observed UV features. Here we present the results of the cloud top altimetry investigations using the VIRTIS mapping in the 1.6 mm band of CO2.

2. Observations [5] On 11 April 2006 Venus Express was inserted in a polar orbit around Venus, and few months later it began nominal science operations [Svedhem et al., 2007]. Nadir observations from apocenter and in ascending branch of the orbit provide global imaging of the planet [Titov et al., 2006]. VIRTIS takes the images and mosaics (VIRTIS-M channel) of the Venus disc in a broad spectral range from UV (0.3 mm) to thermal infrared (5 mm) with a spectral resolution in the IR of 15 nm, as well as high-resolution spectra (VIRTIS-H channel) in the interval of 2 – 5 mm with the spectral resolution of 1 – 3 nm [Drossart et al., 2007]. A VIRTIS-M image has 256 elements (pixels) in the direction along the slit of the spectrometer. In some measurement sequences 256 pixels are binned in groups of 4 pixels to give 64 elements in the image. Scanning in the direction normal to the slit gives an image. A typical size of the VIRTIS-M images is about 4000  4000 km at apocenter, decreasing as the spacecraft approaches the planet [Piccioni et al., 2007]. For each spatial element of

20 km and smaller VIRTIS-M near-IR spectra were used to derive the cloud top altitude. VMC performs simultaneous observations in the narrow-band filter at 0.365 mm with spatial resolution of 50 km or better, thus providing full-disc context images of the Venus cloud tops in the spectral band of the unknown UV absorber [Markiewicz et al., 2007b]. [6] We processed 840 dayside VIRTIS-M measurement sequences with the exposure time of 20 and 300 ms taken on 120 orbits from the beginning of the mission up to orbit 853, from April 2006 to August 2008. Dayside of the southern hemisphere is completely covered by the VIRTIS-M observations, while on the dayside of the northern hemisphere VIRTIS-M image footprints look like narrow stripes. The number of such observation sessions is also limited. The lack of the data in the northern hemisphere can be compensated with the VIRTIS-H measurements, which have been analyzed by Cottini et al. [2008]. Here we also used VIRTIS-H data for cross check of the results. [7] The VIRTIS instrument is optimized for observations of the nightside of the planet. Dayside spectral intensities are several orders of magnitude more intense than those for the nightside, and contain a sawtooth odd/even spectel (spectral element) component, which was not completely removed by the calibration procedure. For this reason, here we used an undersampled spectrum composed of odd spectels only (Figure 1). The spectral resolution appeared to be a bit worse than the nominal value obtained in the preflight laboratory calibrations. The instrumental function of the VIRTIS-M used here was represented by a Gaussian curve with the FWHM of 15 nm at 1.6 mm.

3. Method of the Cloud Top Altitude Determination [8] Spectrum of the solar radiation reflected from the dayside of Venus shows several strong CO2 absorption bands in the near IR range (Figure 1). Their relative depth increases with the total number of CO2 molecules on the line of sight and, thus, depends on the effective path of radiation in the atmosphere. This path is a function of the cloud altitude, its vertical structure, aerosol optical proper-

2 of 10

IGNATIEV ET AL.: ALTIMETRY OF VENUS CLOUDS

E00B43

Figure 2. Modeled dependence of the CO2 1.6 mm normalized band depth index on the cloud top altitude for different cloud structure/scale heights. ties, atmospheric temperature and pressure, and geometry of observations. Using the depths of CO2 bands and some assumptions about properties and atmospheric parameters, the vertical structure of the clouds can be evaluated. Here we use an exponential model of aerosol vertical distribution with two parameters: cloud top altitude h, defined as the altitude of unit optical depth at 1.51 mm, and aerosol scale height H. We adopted the temperature and pressure profiles from the Venus International Reference Atmosphere (VIRA) model [Seiff et al., 1985] and assumed cloud particles composed of 75% sulfuric acid with complex refraction index from Palmer and Williams [1975]. Particle size distribution was represented by ‘‘mode 2,’’ i.e., lognormal function with rm = 1.05 mm and s = 1.21 [Pollack et al., 1980]. These assumptions are evident simplifications: both temperature profile and particle size distribution in the region of cloud tops vary with latitude [e.g., Ragent et al., 1985; Zasova et al., 2007]. However, the effect of a variable temperature profile on the dayside reflectance spectrum is much weaker than that owing to cloud top pressure variations. The submicron haze dominating at higher altitudes is almost transparent in the IR. Larger mode 3 particles can sometimes dominate the aerosol population at the cloud top at high latitudes, but substituting them with equivalent number of mode 2 particles would just slightly bias the measured cloud top altitude. However, we also tested the retrievals with complete 4-modal particle size distribution (Figure 2). To characterize the relative depth of CO2 band, we define the depth index D as difference between the continuum cont(lD) and the spectrum inside the band S(lD) normalized to the continuum at wavelength lD: D¼

cont ðlD Þ
S ðlD Þ ; cont ðlD Þ

where lD = 1.61 mm, and the continuum is defined by a straight line drawn between the two points on the sides of the band: 1.51 and 1.70 mm.

E00B43

[9] Synthetic spectra were calculated for a discrete set of parameters (h,H,m0,m,8), where m0 and m are the cosines of the zenith angles of incident and outgoing radiation, and 8 is the relative azimuth angle of the outgoing radiation. This yielded 5-dimentional matrix D = D(h,H,m0,m,8) that was used in the analysis of VIRTIS spectra. Our radiative transfer model is based on the discrete ordinate code DISORT [Stamnes et al., 1988] coupled with a line-by-line model of gaseous absorption. CO2 line parameters were taken from the preliminary version of Carbon Dioxide Spectroscopic Database (CDSD) for Venus (S. A. Tashkun and V. I. Perevalov, personal communication, 2007) which is available at the anonymous ftp site of Institute of Atmospheric Optics of Russian Academy of Sciences (ftp://ftp.iao.ru/pub/Venus), the CO2 high-temperature database [Pollack et al., 1993], and HITRAN (http:// www.hitran.com). Data on other gases (CO, H2O) were taken from HITRAN [Rothman et al., 2005], version of 2006. Voigt line profiles were used. For CO2, the line profile was multiplied by a sub-Lorentzian c factor [Pollack et al., 1993, equations 1 – 4]. [10] Figure 2 shows the dependence of the computed band depth index D on the cloud top altitude for the simplified mode 2 cloud model and three values of the aerosol scale height. It is compared to the complete 4-mode cloud model by Pollack et al. [1980], with modifications by Zasova et al. [1985], which around the altitude of the unit optical depth has H  2.5 km. Determination of both aerosol scale height and cloud top altitude using one wavelength is ambiguous. The same band depth index D may be obtained with lower clouds having sharp upper boundary (H = 0) and with higher clouds that are more diffuse (i.e., large H). Radiation in the center and in the wings of CO2 bands comes from the levels separated by few kilometers enabling us, in principle, to derive both parameters from the spectrum. Earlier observations indicated that the cloud scale height at the cloud tops can vary from less than 1 km in some regions at high latitudes [Zasova et al., 1993, 2007] to 4 – 5 km at the equator [Ragent et al., 1985; Zasova et al., 1993, 2007; Koukouli et al., 2005]. However, this behavior does not follow from our data: both low- and high-latitude spectra were better fitted with the scale height of 4 – 6 km. We concluded that a reliable determination of at least two parameters requires improvements in the fit of the whole band or several bands and therefore improvements in the spectral and radiometric calibration. In the absence of these improvements, we assumed a constant scale height of 4 km, being a reasonable value for low and middle latitudes and allowing us to perform first-order cloud height determinations. An uncertainty that follows from this assumption is considered below. [11] Once the scale height is fixed, monotonic dependence of the band depth on the cloud top altitude makes the solution of the inverse problem straightforward. For each VIRTIS-M spectrum, we interpolate the function D = D(h,H,m0,m,8) given by the precalculated 5-dimensional matrix to the actual values of H,m0,m,8, thus defining function D(h) on a set of discrete h levels. Interpolation of inverse function h(D) gives h that corresponds to the measured value of D (Figure 2). [12] The principal uncertainty of our approach results from the assumption about the constant scale height of

3 of 10

E00B43

IGNATIEV ET AL.: ALTIMETRY OF VENUS CLOUDS

Figure 3. Venus Monitoring Camera ultraviolet (VMC UV) images with overplotted cloud altimetry maps for orbits (a and b) 243, (c) 462, and (d) 48. Note correlation of the cloud altimetry and UV dark and bright features.

4 of 10

E00B43

E00B43

IGNATIEV ET AL.: ALTIMETRY OF VENUS CLOUDS

E00B43

Figure 4. Cloud altimetry and UV images for three consecutive orbits during the polar brightening event. 4 km. For low clouds (65 km) the altitude uncertainty corresponding to the scale height range from 1 to 5 km is equal to 4 km, while for high clouds (75 km) it increases up to 6 km (Figure 2). Smaller scale heights result in lower cloud top altitudes. However, large variations of scale height values are encountered only in high latitudes [Zasova et al., 1993, 2007]. Assuming that possible variations of the scale height in low latitudes are equal to 3 –5 km and in high latitudes to 1 – 5 km, we may evaluate this uncertainty as ±1.5 km and +1/
3 km in low and high latitudes, respectively. Another systematic bias is caused by a single temperature and pressure vertical profile used in our model. At high latitudes the level of 100 mb is located by 1 – 2 km lower than that at low latitudes [Seiff et al., 1985; Zasova et al., 2006], which means that at high latitudes our results should be corrected by
1 – 2 km. [13] Instrument calibration uncertainties resulted in the dependence of the measured band depth on sample number (position of a pixel in the direction along the slit). It can be caused by imperfect flat fielding and spectral calibration. These two effects are difficult to separate. The above defined matrix D was calculated assuming fixed positions

of spectral channels, which actually depend on temperature of the instrument and sample number. Variations of a spectel wavelength around the average value can be ±0.5 spectel size (9.5 nm). Interpolation of the measured spectrum to the predefined set of wavelengths introduces an error evaluated as a few hundred meters. A similar effect is caused by a systematic shift of the wavelength scale present in the calibrated data. Despite the systematic behavior of the shift, we found it difficult to evaluate with a precision better than ±0.2 spectel size because of the odd-even spectel problem, spectral resolution uncertainty, and other factors that might cause residual differences between measured and synthetic spectra. The total uncertainty of this kind expressed in terms of the retrieved cloud top altitude is evaluated as ±1 km and will be illustrated in section 4 (Figures 3d and 7b). On some orbits where the images taken with 20 ms exposure were interleaved with those of 300 ms, we detected a difference up to several hundred meters in the cloud top altitudes derived from the adjacent measurement sessions. [14] Provided the model and data calibration are perfect, any other reasonable choice of spectral bands or channels used to derive the cloud top altitude should lead us to

5 of 10

IGNATIEV ET AL.: ALTIMETRY OF VENUS CLOUDS

E00B43

E00B43

Figure 5. Comparison of the (left) cloud top altimetry with (right) thermal IR (5 mm) VIRTIS image, both overplotted on the simultaneously captured VMC UV image (orbit 674). similar results. We have demonstrated this using a longer wavelength band observed by the VIRTIS-H channel (see section 4). An alternative to the relative band depth index D is the traditional equivalent width of the band. We found that using the latter value moves down the retrieved cloud top altitude by 1 km. [15] Summarizing all the uncertainties we conclude that the absolute precision of the cloud top measurement does not exceed 2 km at low and middle latitudes. At high latitudes a lower-altitude error limit should be extended by another 2 km. For a single pixel, the relative precision limited by the sensitivity of the spectrum to the cloud top variations and the instrumental noise is few hundred meters for 20 ms exposure time, and tens of meters for 300 ms.

4. Results and Discussion [16] The first results on the global cloud top altimetry were reported by Titov et al. [2008]. Here we applied routinely the above described method to the VIRTIS-M dayside observations. Typical examples of the cloud top altimetry maps are shown in Figure 3. The mosaics consist of overlapping VIRTIS-M images taken at ascending branch of the orbit while the field of view moved from the South pole to the middle latitudes (Figures 3b and 3c) or global mosaics that completely covered the Southern hemisphere (Figure 3d). The altimetry maps are overplotted on the simultaneously taken VMC UV image that allowed us to relate the cloud top altitude to the UV markings. Some altimetry maps, whenever necessary, are processed with a

simple median filter with a window size up to 11 pixels in the direction along the slit and up to 5 pixels in the perpendicular direction to improve the image quality at the minimum degradation of spatial resolution. [17] The cloud top is located at remarkably constant altitude of 74 ± 1 km over most of the planet (Figure 3). Poleward from 50 to 60 degrees latitude the cloud top quickly descends reaching its minimum of 65– 68 km in the huge polar depression 2000– 3000 km in size. The elevation change from equator to pole is about 1 – 2 atmospheric scale heights. The general trend is quite typical and is reproduced from orbit to orbit within 1 km. Variations of that magnitude at a given latitude and solar hour angle may occur on time scales of tens of hours and can be sometimes clearly seen when comparing images taken at the same geometry from consecutive orbits and therefore separated by 24 h (e.g., Figure 4). [18] The global pattern did not change much during the polar brightening event that occurred on 13 January 2007 [Markiewicz et al., 2007b] when both brightness and latitude extension of the polar hood significantly increased for a few days (Figure 4). Surprisingly, these drastic changes in UV appearance had no considerable effect on the cloud top altitude derived from the VIRTIS near-IR spectra: their changes were within the limits of usual variations of 1 – 2 km. Also surprising is that the sharp boundary between the UV dark low latitudes and bright midlatitudes is not pronounced in the cloud altimetry maps (Figure 4). This implies that the UV brightening was not associated with significant changes in cloud top altitude.

6 of 10

E00B43

IGNATIEV ET AL.: ALTIMETRY OF VENUS CLOUDS

Figure 6. Correlation of the fine structure of the vortex eye (orange image, 5 mm) with the cloud top pattern (blue isolines). Altimetry data are absent in the nightside (top) and in a bright region of the dayside (bottom), where the signal is saturated at 300 ms exposure. Contours are drawn every 0.2 km starting from 67.8 km in the center. This also means that either the thickness of the bright UV hood does not exceed few hundred meters or that the hood is transparent at thermal IR wavelengths, the latter indicating that it consists of submicron particles. Markiewicz et al. [2007b] tentatively explained the brightening event by quick formation of fine sulfuric acid haze by homogeneous nucleation and its clearing due to coagulation decay of aerosol number density. [19] UV dark spiral and circular features usually present at
70° are clearly seen in the cloud altimetry maps as variations of several hundred meters on the global slope to the pole (Figures 3a and 3b). Contrary to the analysis of the Pioneer Venus polarization measurements [Esposito and Travis, 1982], these dark features are either located higher than the adjacent UV bright clouds (e.g., Figure 3b) or characterized by a higher latitudinal gradient of the cloud top altitude (e.g., Figure 3c). [20] The polar depression in the cloud tops always coincides with the vortex eye observed by VIRTIS at thermal IR wavelength (Figure 5) [Piccioni et al., 2007]. The eye, which usually appears almost featureless in the UV (e.g., Figure 3a), shows complex patterns in the cloud altimetry maps that show high correlation with the thermal IR images at 5 mm (Figure 6). Hot spiral arms, which have almost the same temperature as the core, are located higher than it and characterized by the strong gradient of the cloud top altitude, thereby being the boundary of the polar vortex.

E00B43

[21] We analyzed 545 VIRTIS frames taken with 20 ms exposure in 72 orbits from the beginning of the mission till orbit 858 (April 2006 to August 2008). We averaged all analyzed data except for high emission angles (q > 85°), where our plane parallel radiative transfer model is no longer valid. Figure 7 shows mean map of the cloud top altitude derived from this data set. On average, the cloud tops in middle and low latitudes are located at a constant altitude of 74 km with orbit-to-orbit variations of about 1 km. Note that uncertainties discussed above can result in a systematic shift of this altitude by 1 – 2 km. Figure 7 shows no considerable local time dependence of the cloud top altitude, except for tentative indication of slightly higher altitudes in the afternoon at low latitudes, that coincides with the region of strongest convective activity observed by VMC [Markiewicz et al., 2007b]. [22] Since the Venus Express orbit is fixed in inertial space, optimal conditions for dayside observations at ascending branch of the orbit are naturally grouped in 100 orbits interleaved with the same number of orbits devoted to nightside observations. This makes continuous monitoring of the cloud altitudes impossible. After orbit 610 we observed a systematic raise of the cloud tops at high latitudes by 3 km with respect to orbits before the gap 505– 609 in the dayside observations. Figure 8a compares the mean latitude trend derived from the first 500 orbits to that observed after the orbit 610. After scrutinizing the data and VIRTIS behavior we exclude instrumental effects and conclude that polar clouds can show strong long-term variability, as known from previous observations in the UV [Kawabata et al., 1980; Esposito et al., 1988].

Figure 7. Mean cloud top altitude as a function of latitude and local time. Linear patterns are the traces of VIRTIS-M image frames.

7 of 10

E00B43

IGNATIEV ET AL.: ALTIMETRY OF VENUS CLOUDS

E00B43

Figure 8. Latitudinal behavior of the cloud top altitude. (a) Mean value in the southern hemisphere for two periods: from April 2006 to September 2007 (orbits 0 – 503) (black) and from December 2007 to August 2008 (orbits 614 – 858) (gray). The curves are moving averages with 5° latitude window. Error bars represent scattering of individual measurements due to both temporal variations and instrumental errors. (b) Observations in the northern hemisphere in orbit 724 with by VIRTIS-M (dots) and VIRTIS-H (squares).

[23] Relatively random VIRTIS-M observations that passed over the northern hemisphere showed no principal difference in the cloud altimetry between the north and the south (Figure 8). Simultaneous VIRTIS-H measurements in orbit 724 allowed us to evaluate independently the cloud top altitudes from the weak CO2 band at 2.5 mm. Figure 8b demonstrates good agreement between both data sets. The VIRTIS-H cloud top altitudes shown in Figure 8b, although derived from 2.5 mm band, are referenced to 1.51 mm for comparison with the VIRTIS-M results. In the northern hemisphere, owing to a small distance to the planet, the VIRTIS-M image footprint on the clouds is a 50 km wide stripe stretched along meridian from equator to pole (similarly, VIRTIS-H measurements cover a 3 km wide stripe approximately in the center of the VIRTIS-M footprint). The scatter of the VIRTIS-M data points in Figure 8b is thus caused by the flat field and spectral calibration errors in the direction along the slit rather than the cloud altitude variability, which gives us an estimate of the effects of those errors on the results. The same effect appears as the difference between the adjacent frames in Figure 3d. [24] The above described method provides the altitude of the t = 1 level at 1.5 mm. We studied the relation between the cloud top altimetry derived from the near-IR observations and altitude of the UV markings. Optical properties of the submicron haze with rm  0.15 –0.25 mm are quite different from those of the main cloud in which the particles with 1 – 1.4 mm dominate. The extinction coefficient of the submicron haze in UV is several times greater than that in the near-IR. Thus the upper haze is almost transparent at 1.5 mm, and at these wavelengths VIRTIS sounds the main cloud deck all over the planet. Calculations of the altitude of the unit optical depth level in the broad spectral range using the standard 4-modal aerosol distribution by Pollack et al. [1980] with low-latitude number density profiles by Zasova et al. [1985] show that the cloud top in UV is located almost at the same altitude as in the near-IR (Figure 9). This is representative of low and middle latitudes where the upper haze is generally of low density. Doubling

the number of submicron particles lifts the UV cloud top by 1 km. To simulate polar clouds we removed mode 2 particles, so that the main cloud deck is now composed of lager mode 2’ particles and located below 65 km. It is overlaid by submicron upper haze with opacity of 1. In this case there is a significant difference of several kilometers between the levels of unit t in the near IR and UV. This condition is often observed at high latitudes, where the UV opacity of submicron upper haze can reach 0.8 [Kawabata et al., 1980]. We note that Markiewicz et al. [2007b] estimated UV opacity to be of 1 and haze vertical extension of 2 km from the analysis of the brightening event (Figure 4). [25] Simultaneous observations by the VIRTIS and VMC instruments onboard Venus Express provide an opportunity to assign altitudes to mesoscale UV markings and to

Figure 9. Cloud top altitude as a function of wavelength for the low-latitude 4-mode model of Venus clouds, the same model with the upper haze enhanced by 2 times, and the polar cloud model.

8 of 10

E00B43

IGNATIEV ET AL.: ALTIMETRY OF VENUS CLOUDS

constrain location of the unknown absorber that produces these markings within the cloud deck. Surprisingly the analysis shows that the formal cloud top altitude increases in UV dark features. This relation is clearly seen in the region of dark spiral arms (Figures 3a – 3b) in high latitudes. Such behavior, although obvious in the VIRTIS altimetry, contradicts the earlier observations suggesting that the UV absorber is located within the main cloud layer [Esposito and Travis, 1982; Esposito et al., 1983]. There are several possible explanations of this disagreement. First, Pioneer Venus and Venus Express observations were focused on different latitudes: from low to middle and from middle to high, respectively. Another plausible explanation is that our simple model description of the clouds is insufficient and should include variations of the cloud scale height and optical properties of the medium rather than the cloud top altitude only. In any case cloud top altitude variations related to dark UV markings do not exceed few hundred meters.

5. Conclusions [26] Analysis of the VIRTIS and VMC observations showed that they provide a powerful tool to determine the cloud top altitude. These measurements assign the altitude of the cloud tracked winds and constrain the location of the unknown UV absorber with the cloud. Further investigations should include the feasibility study of the simultaneous scale height evaluation using several CO 2 IR absorption bands, and detailed cloud altimetry measurements at higher spatial resolution. [27] Acknowledgments. Venus Express is a mission of the European Space Agency. We thank the Agenzia Spaziale Italiana (ASI) and the Centre National d’E´tudes Spatiales (CNES) for the support and funding of the VIRTIS experiment. We gratefully thank all members of the Venus Express project and of the VIRTIS and VMC technical teams. N.I. was supported by the Russian Foundation of Basic Research grant 08-02-01383. We thank L.W. Esposito, R.W. Carlson, and an anonymous reviewer for their valuable comments that greatly helped us to improve the manuscript.

References Braak, C. J., J. F. de Haan, J. W. Hovenier, and L. D. Travis (2002), Spatial and temporal variations of Venus haze properties obtained from Pioneer Venus Orbiter polarimetry, J. Geophys. Res., 107(E5), 5029, doi:10.1029/ 2001JE001502. Cimino, J. B., C. Elachi, A. J. Kliore, D. J. McCleese, and I. R. Patel (1980), Polar cloud structure as derived from the Pioneer Venus Orbiter, J. Geophys. Res., 85(A13), 8082 – 8088, doi:10.1029/JA085iA13p08082. Cottini, V., N. Ignatiev, D. Grassi, G. Piccioni, and P. Drossart (2008), Venus mesospheric water vapor from VIRTIS-H VEX dayside measurements, Eos Trans. AGU, 89(53), Fall Meet. Suppl., Abstract #P33A1429. Drossart, P., et al. (2007), Scientific goals for the observation of Venus by VIRTIS on ESA/Venus express mission, Planet. Space Sci., 55, 1653 – 1672, doi:10.1016/j.pss.2007.01.003. Esposito, L. W. (1980), Ultraviolet contrasts and the absorbers near the Venus cloud tops, J. Geophys. Res., 85(A13), 8151 – 8157, doi:10.1029/JA085iA13p08151. Esposito, L. W., and L. D. Travis (1982), Polarization studies of the Venus UV contrasts: Cloud height and haze variability, Icarus, 51, 374 – 390, doi:10.1016/0019-1035(82)90090-2. Esposito, L. W., R. G. Knollenberg, M. Ia. Marov, O. B. Toon, and R. P. Turco (1983), The clouds are hazes of Venus, in Venus, edited by D. M. Hunten et al., pp. 484 – 564, Univ. of Ariz. Press, Tucson. Esposito, L. W., M. Copley, R. Eckert, R. Gates, A. I. F. Stewart, and H. Worden (1988), Sulfur dioxide at the Venus cloud tops, 1978 – 1986, J. Geophys. Res., 93(D5), 5267 – 5276, doi:10.1029/ JD093iD05p05267.

E00B43

Kawabata, K., D. L. Coffeen, J. E. Hansen, W. A. Lane, M. Sato, and L. D. Travis (1980), Cloud and haze properties from Pioneer Venus polarimetry, J. Geophys. Res., 85(A13), 8129 – 8140, doi:10.1029/ JA085iA13p08129. Knibbe, W. J. J., J. F. De Haan, J. W. Hovenier, and L. D. Travis (1998), Analysis of temporal variations of the polarization of Venus observed by Pioneer Venus Orbiter, J. Geophys. Res., 103(E4), 8557 – 8574, doi:10.1029/98JE03558. Koukouli, M. E., P. G. J. Irwin, and F. W. Taylor (2005), Water vapor abundance in Venus’ middle atmosphere from Pioneer Venus OIR and Venera 15 FTS measurements, Icarus, 173, 84 – 99, doi:10.1016/ j.icarus.2004.08.023. Markiewicz, W. J., et al. (2007a), Venus Monitoring Camera for Venus Express, Planet. Space Sci., 55, 1701 – 1711, doi:10.1016/ j.pss.2007.01.004. Markiewicz, W. J., D. V. Titov, S. S. Limaye, H. U. Keller, N. Ignatiev, R. Jaumann, N. Thomas, H. Michalik, R. Moissl, and P. Russo (2007b), Morphology and dynamics of the upper cloud layer of Venus, Nature, 450, 633 – 636, doi:10.1038/nature06320. Moroz, V. I., A. P. Ekonomov, B. E. Moshkin, H. E. Revercomb, L. A. Sromovsky, J. T. Schofield, D. Spaenkuch, F. W. Taylor, and M. G. Tomasko (1985), Solar and thermal radiation in the Venus atmosphere, Adv. Space Res., 5(11), 197 – 232, doi:10.1016/0273-1177(85)90202-9. Palmer, K. F., and D. Williams (1975), Optical constants of sulfuric acid: Application to the clouds of Venus?, Appl. Opt., 14, 208 – 219. Piccioni, G., et al. (2007), South-polar features similar to those near the north pole, Nature, 450, 637 – 640, doi:10.1038/nature06209. Pollack, J. B., O. B. Toon, R. C. Witten, R. Boese, B. Ragent, M. Tomasko, L. Esposito, L. Travis, and D. Wiedman (1980), Distribution and source of the UV absorption in Venus’ atmosphere, J. Geophys. Res., 85(A13), 8141 – 8150, doi:10.1029/JA085iA13p08141. Pollack, J. B., et al. (1993), Near-infrared light from Venus’ nightside: A spectroscopic analysis, Icarus, 103, 1 – 42, doi:10.1006/icar.1993.1055. Ragent, B., L. W. Esposito, M. G. Tomasko, M. Y. Marov, V. P. Shari, and V. N. Lebedev (1985), Particulate matter in the Venus atmosphere, Adv. Space Res., 5(11), 85 – 115, doi:10.1016/0273-1177(85)90199-1. Rossow, W. B., A. D. Del Genio, S. S. Limaye, L. D. Travis, and P. H. Stone (1980), Cloud morphology and motions from Pioneer Venus images, J. Geophys. Res., 85(A13), 8107 – 8128, doi:10.1029/ JA085iA13p08107. Rothman, L. S., et al. (2005), The HITRAN 2004 molecular spectroscopic database, J. Quant. Spectrosc. Radiat. Transfer, 96, 139 – 204, doi:10.1016/j.jqsrt.2004.10.008. Sato, M., L. D. Travis, and K. Kawabata (1996), Photopolarimetry analysis of the Venus atmosphere in polar regions, Icarus, 124, 569 – 585, doi:10.1006/icar.1996.0231. Seiff, A., J. T. Schofield, A. J. Kliore, F. W. Taylor, S. S. Limaye, H. E. Revercomb, L. A. Sromovsky, V. V. Kerzhanovich, V. I. Moroz, and M. Y. Marov (1985), Models of the structure of the atmosphere of Venus from the surface to 100 kilometers altitude, Adv. Space Res., 5(11), 3 – 58, doi:10.1016/0273-1177(85)90197-8. Stamnes, K., S.-C. Tsay, W. Wiscombe, and K. Jayaweera (1988), Numerically stable algorithm for discrete-ordinate-method radiative transfer in multiple scattering end emitting layered media, Appl. Opt., 27, 2502 – 2509, doi:10.1364/AO.27.002502. Svedhem, H., et al. (2007), Venus Express: The first European mission to Ve n u s , P l a n e t . S p a c e S c i . , 5 5 , 1 6 3 6 – 1 6 5 2 , d o i : 1 0 . 1 0 1 6 / j.pss.2007.01.013. Taylor, F. W., et al. (1980), Structure and meteorology of the middle atmosphere of Venus: Infrared remote sensing from the Pioneer Orbiter, J. Geophys. Res., 85(A13), 7963 – 8006, doi:10.1029/JA085iA13p07963. Titov, D., et al. (2006), Venus Express science planning, Planet. Space Sci., 54, 1279 – 1297, doi:10.1016/j.pss.2006.04.017. Titov, D. V., F. W. Taylor, H. Svedhem, N. I. Ignatiev, W. J. Markiewicz, G. Piccioni, and P. Drossart (2008), Atmospheric structure and dynamics as the cause of ultraviolet markings in the clouds of Venus, Nature, 456, 620 – 623, doi:10.1038/nature07466. Tomasko, M. G., L. R. Doose, and P. H. Smith (1985), The absorption of solar energy and the heating rate in the atmosphere of Venus, Adv. Space Res., 5(9), 71 – 79, doi:10.1016/0273-1177(85)90272-8. Zasova, L. V., et al. (1985), VENERA-15 and VENERA-16 infrared experiment: Some spectral analysis results on the cloud structure, Cosmic Res., Engl. Transl., 23, 189 – 201. Zasova, L. V., V. I. Moroz, L. W. Esposito, and C. Y. Na (1993), SO2 in the middle atmosphere of Venus: IR measurements from VENERA-15 and comparison to UV data, Icarus, 105, 92 – 109, doi:10.1006/ icar.1993.1113. Zasova, L. V., V. I. Moroz, V. M. Linkin, I. V. Khatuntsev, and B. S. Maiorov (2006), Structure of the Venusian atmosphere from surface up

9 of 10

E00B43

IGNATIEV ET AL.: ALTIMETRY OF VENUS CLOUDS

to 100 km, Cosmic Res., Engl. Transl., 44, 364 – 383, doi:10.1134/ S0010952506040095. Zasova, L. V., N. Ignatiev, I. Khatuntsev, and V. Linkin (2007), Structure of the Venus atmosphere, Planet. Space Sci., 55, 1712 – 1728, doi:10.1016/ j.pss.2007.01.011.























M. Almeida, ESA, ESAC, PO Box 78, E-28691 Villanueva De La Can˜ada, Madrid, Spain. ([email protected]) V. Cottini, NASA Goddard Space Flight Center, 8800 Greenbelt Road, Building 2, Code 693, Greenbelt, MD 20771, USA. (valeria.cottini@ gmail.com) P. Drossart, LESIA, Observatoire de Paris, Section de Meudon, place Jules Janssen 5, F-92195 Meudon CEDEX, France. (Pierre.Drossart@ obspm.fr)

E00B43

N. Ignatiev, Space Research Institute for Russian Academy of Sciences, Profsoyuznaya 84/32, GSP-7, 117997 Moscow, Russia. (ignatiev@irn. iki.rssi.ru) N. Manoel, Department of Mathematics, University of E´vora, Office 258, R. Roma˜o Ramalho, 59, P-7000 E´vora, Portugal. ([email protected]. uevora.pt) W. J. Markiewicz and D. V. Titov, Max-Planck-Institut fu¨r Sonnensystemforschung, Max-Planck-Strasse 2, D-37191 Katlenburg-Lindau, Germany. ([email protected]; [email protected]) G. Piccioni, INAF, IASF, via del Fosso del Cavaliere 100, I-00133 Rome, Italy. ([email protected]) Th. Roatsch, Institute of Planetary Research, German Aerospace Center, Rutherfordstrasse 2, D-12489 Berlin, Germany. ([email protected])

10 of 10

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114, E00B37, doi:10.1029/2008JE003184, 2009

for

Full Article

Cloud structure in Venus middle-to-lower atmosphere as inferred from VEX/VIRTIS 1.74 mm data T. Satoh,1 T. Imamura,1 G. L. Hashimoto,2 N. Iwagami,3 K. Mitsuyama,3 S. Sorahana,3 P. Drossart,4 and G. Piccioni5 Received 8 May 2008; revised 6 October 2008; accepted 10 December 2008; published 25 April 2009.

[1] We have analyzed 1.74 mm nightside emission of Venus recorded using Visible and

Infrared Thermal Imaging Spectrometer (VIRTIS) onboard European Space Agency’s (ESA) Venus Express (Orbit 344, 30–31 March 2007). Attention was paid to how infrared radiance, intense at the center of the 1.74 mm ‘‘window’’, dims at an off-center wavelength (1.71 mm). Cloud models are required to simultaneously reproduce the emission intensity at 1.74 mm and the ratio of intensities (I1.71mm/I1.74mm). Our best-fit model (5 km vertical resolution) has located the main cloud opacity in 40–45 km altitude, lower than previous studies. This may be due to the use of CO2 line parameters from a relatively new source (Carbon Dioxide Spectroscopy Databank) which may also be responsible for weaker continuum absorption, 5.6
109 cm1 amagat2. The data are reproduced well by models of which total aerosol optical thickness is 30–50 plus subcloud haze at 30–40 km altitude. We have mapped the subcloud haze opacity (approximately 0–4) and found that the opacity basically anticorrelates with the 1.74 mm intensity. There are regions of ‘‘positive’’ correlation which may imply enhanced production of aerosols due to penetration of more sunlight in less cloudier regions. Venus Express, now with a capability of sensing ‘‘from the top to the bottom’’ of Venus cloud system, will greatly enhance our knowledge about the current status of Venus atmosphere. Citation: Satoh, T., T. Imamura, G. L. Hashimoto, N. Iwagami, K. Mitsuyama, S. Sorahana, P. Drossart, and G. Piccioni (2009), Cloud structure in Venus middle-to-lower atmosphere as inferred from VEX/VIRTIS 1.74 mm data, J. Geophys. Res., 114, E00B37, doi:10.1029/2008JE003184.

1. Introduction [2] Venus, often called the Earth’s twin planet, is totally covered by aerosol layers. On the basis of in situ measurements by descent probes [see, e.g., Esposito et al., 1983], the lowermost layer (haze) starts as low as 30 km altitude and an upper haze layer reaches up to 90 km. The main cloud deck extends from 70 km (the level of unit optical depth as observed from the space in the ultraviolet) down to altitudes 45– 50 km [Esposito et al., 1997]. A substantial amount of energy absorbed within this enormous cloud system is thought to control the atmospheric dynamics and climate of Venus [Crisp and Titov, 1997]. [3] The top haze layer of mode 1 particles, of which effective radius (reff) is 0.3 mm, exhibits great variability both 1 Institute of Space and Astronautical Science, Japan Aerospace Exploration Agency, Kanagawa, Japan. 2 Laboratory for Earth and Planetary Atmospheric Science, Organization of Advanced Science and Technology, Kobe University, Kobe, Japan. 3 Department of Earth and Planetary Science, Graduate School of Science, Tokyo University, Tokyo, Japan. 4 LESIA, Observatoire de Paris, Meudon, France. 5 INAF, Istituto di Astrofisica Spaziale e Fisica Cosmica, Rome, Italy.

Copyright 2009 by the American Geophysical Union. 0148-0227/09/2008JE003184$09.00

in space and time [Kawabata et al., 1980; Sato et al., 1996]. The upper cloud layer is dominated by mode 2 particles (reff = 1.0 mm) as discovered by the analysis of polarization of the reflected sunlight [Hansen and Hovenier, 1974]. Polarization data were utilized to constrain the refractive index of aerosols to 1.45, and this has been regarded most important evidence that the aerosols, at least in the upper cloud, are likely to be droplets of sulfuric acid [Sill, 1972; Young and Young, 1973]. Mode 2 particles are, as indicated by descent probe measurements, abundant also in the middle cloud layer. Existence of much larger (mode 3) particles in the lower cloud has been inferred from Pioneer Venus LCPS experiment [Knollenberg and Hunten, 1980] with continuing controversy [Toon et al., 1984]. Below the lower cloud is the lower haze layer [Golovin and Ustinov, 1982], of which composition as well as that of the lower cloud remains uncertain to date. To answer outstanding questions regarding this enormous cloud system, observations of the entire system with comprehensive coverage in time and space are essential. [4] Discovery of near-infrared windows in the CO2 absorption spectrum [Allen and Crawford, 1984] opened a new era for remote sensing. Such windows allow thermal emissions from deeper atmosphere of Venus to be remotely detected if observations are made on the night (unillumi-

E00B37

1 of 13

SATOH ET AL.: AEROSOLS IN VENUS’ LOWER ATMOSPHERE

E00B37

E00B37

Figure 1. VIRTIS band 77 images acquired while the spacecraft was approaching Venus from the south in its Orbit 344 (30 – 31 March 2007). The south pole of Venus is below the bottom edge and the equator is over the top edge in each image. Circles in white dashed lines indicate locations of the same cloud feature. Note the saturated pixels are blacked out. nated) hemisphere: Radiations at 1.74 and 2.3 mm windows originate primarily from 15– 30 km and 26 –45 km altitudes, respectively [Kamp et al., 1988; Kamp and Taylor, 1990]. These altitudes are well below the main cloud deck so that the opacity variations within clouds appear as contrasts between brighter and darker regions. [5] High spatial resolution maps in these windows were obtained by Near-Infrared Mapping Spectrometer (NIMS) onboard the Galileo spacecraft during its flyby of Venus in February 1990 [Carlson et al., 1991; Carlson and Taylor, 1993]. These high-resolution maps exhibited intensity variations of about a factor 20 (at its best resolution of about 25 km) between the brightest and the darkest features in 2.3 mm window. Analyzing the NIMS data in the 1.7, 2.3, and 3.75 mm, Grinspoon et al. [1993] have constrained the amplitude and vertical distribution of the optical depth anomalies. Their findings include (1) the cloud opacity variations are largely confined to altitudes between 48 and 50 km and (2) mode 3 particles (reff = 3.65 mm) are primarily responsible to the cloud opacity variations. [6] ESA’s Venus Express arrived at Venus on 11 April 2006 and has been continuously observing Venus with a variety of instruments from an elongated polar orbit [Svedhem et al., 2007]. Observations in the near-infrared windows (1.7, 2.3 mm, and other wavelengths) are done with the Visible and Infrared Thermal Imaging Spectrometer (VIRTIS) [Piccioni et al., 2007; Drossart et al., 2007]. Quality of the VIRTIS data (spatial resolution and signal-tonoise ratio) well exceeds that of NIMS. In addition, its coverage in space and time makes the VIRTIS data suitable to study spatial and temporal variability of Venus clouds, an important key to understand the evolution of Venus climate. [7] In this paper, we investigate the aerosol distribution (vertical and horizontal) by analyzing VIRTIS-M-IR 1.74 mm window data. Because we deal with only VIRTIS-M-IR

(medium-resolution infrared) data, hereafter we refer to VIRTIS-M-IR as just VIRTIS.

2. Data 2.1. Selecting VIRTIS Data Cubes for Analysis [8] We analyze a small subset from the VIRTIS data archive, which is so chosen as to satisfy the following criteria: (1) Venus Express science case is 2 or 3, (2) VIRTIS pointing mode ‘‘nadir’’, and (3) VIRTIS exposure time is 18 s for each frame. [9] The longest exposure (18 s) of VIRTIS data acquisition is favored in order that highest possible signal-to-noise ratio can be achieved even for the data off the center of the 1.74 mm window. In Figure 1, we display one of such data, acquired during Orbit 344 (30 – 31 March 2007). Three data cubes were acquired in Orbit 344 while the spacecraft was approaching Venus from the south in the ascending portion of the trajectory. The data cover a wide range in planetary latitudes from near the south pole to near the equator. Basic parameters such as the data acquisition time and the latitude coverage for three data cubes are summarized in Table 1. 2.2. Center and Wing of 1.74 mm Window [10] A spectrum around the 1.74 mm window recorded in VI0344_01 data cube is shown in Figure 2. The center (i.e., the strongest signal pixel) of this window is at the band 77 with a close second at an adjacent band, 76. We pay our attention to how this strong emission (up to 0.2 W m2 sr1 mm1 at the band center) dims as the CO2 absorption increases for off-center wavelengths. In order to include an adequate amount of CO2 absorption, while keeping a fair signal-to-noise ratio, we choose the band 74 as the secondary data wavelength (Figure 2). The wavelengths calculated for VIRTIS’s bands 77 and 74 are given in Table 1. Note that VIRTIS’s wavelength registration is temperature-dependent:

Table 1. Parameters for Orbit 344 VIRTIS-M-IR Data Start

Stop

l (mm)

Latitude

IDa

Date

Time

Date

Time

Science Case

Minimum

Maximum

T b (K)

Band 74

Band 77

VI0344_00 VI0344_01 VI0344_02

30 March 2007 30 March 2007 31 March 2007

2132:32.7 2332:32.7 0132:32.8

30 March 2007 31 March 2007 31 March 2007

2306:52.5 0106:52.6 0306:52.6

2 2 2

82.8° 66.2° 49.1°

27.7° 9.3° 3.2°

156.94 158.53 159.15

1.7202 1.7189 1.7184

1.7487 1.7474 1.7469

a

Data ID is in the form of orbit number_cube number. Average of 5 HK values under a keyword of M_SPECT_TEMP.

b

2 of 13

E00B37

SATOH ET AL.: AEROSOLS IN VENUS’ LOWER ATMOSPHERE

Figure 2. A spectrum of Venus nightside emission from VI0344_01 cube. (top) All near-infrared windows; (bottom) an enlarged view of the 1.74 mm window. Bands 74 and 77 used in our analysis are indicated in the bottom with their nominal wavelengths. Also shown in the bottom (dotted line) is the CO2 opacity (optical thickness per 10 km path) for temperature and pressure at 35 km altitude. Note the horizontal offset between the intensity peak around bands 76– 77 and the smallest CO2 opacity around bands 74– 76 (see text).

E00B37

a 10 K higher spectrograph temperature will offset the wavelength shortward by 0.008 mm, or 8 nm, for the 1.74 mm window. This almost compares to VIRTIS’s one spectral resolution element, 9 nm. 2.2.1. Assigning Wavelengths to VIRTIS Bands [11] The nominal wavelengths for VI0344_01 data cube are 1.747 mm (band 77) and 1.719 mm (band 74), respectively (Table 1). These assignments may, however, need some adjustments as suggested by a slight offset between the brightest signal pixels (bands 77 and 76) and wavelengths of the smallest CO2 opacities (Figure 2). [12] We allow a small horizontal offset to the observed spectrum so that the brightest pixels overlap with wavelengths of the smallest CO2 opacities. By examining by eye, it is found that an offset of 1.5 bands shortward brings the observed spectrum to a closer fit with the CO2 opacity profile. This imply that VIRTIS’s wavelength registration for VI0344_01 includes an offset of 1.5 bands, part of which has already been noticed by the VIRTIS team (of the order of 0.01 mm) but not formularized yet. Taking this into account, we assign the wavelengths to bands 74 and 77 as follows: (1) Band 74: 1.719 mm ! 1.705 mm (labeled 1.71 mm). (2) Band 74: 1.747 mm ! 1.735 mm (labeled 1.74 mm). We hereafter label the band 77 intensity as I1.74mm and the band 74 intensity as I1.71mm, respectively. 2.2.2. Characteristics of Wing-to-Center Ratio [13] A set of (I1.71mm/I1.74mm) ratio images are displayed in Figures 3a – 3c. We hereafter refer to this ratio as wing-tocenter (W2C) ratio. Relief-like appearances of cloud edges are due to a slight vertical offset between two images. They nearly disappear when we shift the band 77 image upward by 0.35 pixels before taking a ratio (Figures 3d– 3f). It has also been pointed out that VIRTIS’s spectral registration is nonuniform across a frame (B. Bezard, private communication, 2008). The effect is obvious in Figures 3a – 3c as systematic brightening toward the right edges. We correct

Figure 3. The W2C ratio images from Orbit 344. Figures 3a – 3c all exhibit systematic brightening toward the right edge. This may be due to a slight wavelength change (0.15 nm) from the left edge to the right. Also corrected for is a slight vertical offset which is the cause of relief-like appearances of cloud edges in images Figures 3a – 3c. In the corrected Figures 3d– 3f, such effects are no longer noticed. 3 of 13

E00B37

SATOH ET AL.: AEROSOLS IN VENUS’ LOWER ATMOSPHERE

Figure 4. W2C ratios, from three data cubes acquired during the Orbit 344, are plotted against the observed band 77 intensity. Points from VI0344_01 and VI0344_02 are horizontally offset by 0.10 and 0.30 W m2 sr1 mm1, respectively, so the data from different cubes can easily be compared. The W2C ratio tends to decrease as the spectrograph temperature gets higher, indicating that the temperature-dependent wavelength shift has significant effects on the W2C ratio. for this effect by applying a small tilt to the spectra, 0.08 band (0.7 nm) on one end and +0.08 band (+0.7 nm) on another end of a scan. Although the correction is not perfect, most of systematic brightening seen in the ‘‘uncorrected’’ images can be removed. In the following analysis, the corrected data are used. [14] In Figure 4, we plot the W2C ratio against I1.74mm. There exist 2 distinct branches in the plot of the VI0344_01 data (Figure 4, middle). These can be characterized as follows: (1) The ‘‘main’’ branch has W2C ratio of 0.18, and the ratio increases to 0.20 as the I1.74mm gets smaller (low-intensity region). (2) The ‘‘upper’’ branch has nearly constant W2C of 0.20 for all the range of I1.74mm. [15] There are noticeable discrepancies (vertical offsets) between the data from different cubes. These discrepancies may likely be attributed to the shift of observing wavelength due to temperature changes. To assure the internal consistency of data, we use the data from a single cube, VI0344_01, for the analysis. [16] What we try in the followings is to construct atmospheric models, with fewest free parameters, which can simultaneously reproduce the observed I1.74mm and W2C ratio.

E00B37

tion of 96.5% CO2 and 3.5% N2 as well as the hydrostatic equilibrium are assumed to interpolate values given in the table. The water vapor abundance is assumed to be constant (25 ppm) below 50 km altitude (near the middle of aerosol layers) and its opacity becomes negligible at higher levels [Taylor et al., 1997; Svedhem et al., 2007]. [18] Once the temperature and pressure are specified, the absorption coefficients can be obtained by performing standard line-by-line computations. Our model includes three opacity sources: (1) CO2 line absorption, (2) H2O line absorption, and (3) continuum absorption. [19] Other minor gases are neglected since they only slightly modulate the observed intensity at VIRTIS-M’s spectral resolution, 0.01 mm. CO2 line absorption is computed by referring to the Carbon Dioxide Spectroscopy Databank (CDSD; http://spectra.iao.ru). The line shape (Doppler and pressure broadened) is simulated with a Voigt function. We truncate the profile (force the absorption to zero) in the far wing at a distance of 50 times the half width from the line center. H2O absorption is computed using HITEMP database. Wavelength-independent continuum absorption of the type introduced by Pollack et al. [1993] is used in the computations. The strength of this for the 1.74 mm window is 7.0
109 cm1 amagat2. Rayleigh scattering is neglected as its optical depth for the entire atmospheric column amounts only 0.4 in the 1.74 mm window [Hansen and Hovenier, 1974; Hansen and Travis, 1974]. Opacities computed for two wavelengths, 1.705 and 1.735 mm, corresponding to a 5 km optical path, are shown as a function of the altitude in Figure 5. 3.2. Aerosol Vertical Distribution [20] On the basis of direct measurements by descent probes to Venus atmosphere, several distinct layers of aerosol concentration have been identified [Marov et al., 1980; Ragent and Blamont, 1980; Knollenberg and Hunten, 1980]. (1) Upper haze (UHZ), above 66 km, (2) upper cloud (UCL), 56– 66 km, (3) middle cloud (MCL), 50– 56 km, (4) lower cloud (LCL), 47– 50 km, lower haze (LHZ), below 47 km. [21] To keep our model simple enough, we set 5 km vertical grids in our model. The first layer starts at the ground surface (z = 0 km) and extends upward to z = 5 km, and the ith layer is between zi,bottom = (i  1)
5 km and zi,top = i
5 km. The topmost layer in our model is the 13th (65 – 70 km). Anything above this is neglected since neither significant opacity source nor emission source for the 1.74 mm window exists. In doing this, altitude assignments for nominal aerosol layers are as follows (Figure 6): (1) UHZ, 65– 70 km, (2) UCL, 55– 65 km, (3) MCL, 50– 55 km, (4) LCL, 45– 50 km, (5) LHZ, 30– 45 km. [22] Within a layer, gaseous opacity, including CO2, H2O and continuum, is computed for every 0.1 km (Figure 5) and is subsequently summed over the altitude range (5 km): Z

ztop

t ð zÞdz:

t gas ¼

ð1Þ

zbottom

3. Model Computations 3.1. Atmospheric Structure and Opacity Sources [17] Atmospheric structure (pressure/temperature profile) is taken from Seiff [1983]. Nominal atmospheric composi-

The scale height of aerosols is assumed to be the same as that of the gas within a layer.

4 of 13

SATOH ET AL.: AEROSOLS IN VENUS’ LOWER ATMOSPHERE

E00B37

E00B37

Figure 5. Line absorption opacity (CO2 and H2O) as well as the continuum opacity as a function of the altitude from the ground surface are shown. The H2O abundance is 25 ppm. Note that the continuum absorption (dotted line) is common in two wavelengths, (left) 1.705 mm and (right) 1.735 mm, so the comparison is easy.

3.3. Radiative Transfer Computation [23] We employ the Discrete Ordinate Radiative Transfer (DISORT) program developed by Stamnes and colleagues [Stamnes et al., 1988]. The single-scattering phase functions for aerosols are approximated using the Henyey-Greenstein analytic function: PHG ðqÞ ¼

1  g2 ð1 þ

g2

 2g cos qÞ3=2

;

ð2Þ

where q is the scattering angle. The anisotropy factor, 1 g 1, determines the degree of forward scattering (g is positive) or backward scattering (g is negative). On the basis of a series of Mie computations for Venusian aerosols, with an assumption of sulfuric acid droplets [Palmer and Williams, 1973], values of g are found to fall within the range of 0.60.9 in the visible and near-infrared wavelengths. [24] Because we deal with transmission of infrared radiation which is affected primarily by forward scattering by the aerosols, using a Henyey-Greenstein function with properly adjusted g is just adequate for Venus nightside simulations. For example, the following phase functions are almost identical when used in simulations of the nightside disk emission: (1) A single-peak Henyey-Greenstein function with g = 0.75. (2) A combination of forward-scattering and backward-scattering Henyey-Greenstein functions with g1 = 0.8, g2 = 0.8, f1 = 0.979 (yielding an averaged anisotropy factor of 0.75) [Tomasko et al., 1978]. (3) Numerically derived Mie scattering phase functions for mode 2, 2’, and 3 particles at 1.74 mm wavelengths (all within the range 0.76 g 0.80). [25] An important parameter to control the performance of DISORT computation is the number of streams, NSTR. This parameter determines how radiation field within a layer is decomposed. According to Levoni et al. [2001], the amplitude of residual errors, for models with HenyeyGreenstein phase function, converges to 0 for a choice

of NSTR 32. We therefore use NSTR = 32 for our computations with Henyey-Greenstein functions.

4. Computational Results 4.1. Types A – C: Models With Three Cloud Layers [26] First, we simplify the model by omitting haze layers (UHZ and LHZ) as optically thin hazes may be expected to have minimal effects. We call such a model ‘‘type A’’: In model A1, the optical thickness in UCL is allowed to vary,

Figure 6. A simplified cloud model in which aerosol layers are organized with vertical resolution of 5 km. According to the atmospheric opacity displayed in Figure 5, location of scattering medium above 50 km altitude is not uniquely determined by this analysis.

5 of 13

SATOH ET AL.: AEROSOLS IN VENUS’ LOWER ATMOSPHERE

E00B37

Table 2. Cloud Parameters for Model A Series Z (km)

UCL MCL LCL

65 – 70 60 – 65 55 – 60 50 – 55 45 – 50 40 – 45 35 – 40

g

0.75 0.75 0.75

w

0.997 0.997 0.990

t A1

variable 5 5

E00B37

Table 3. Cloud Parameters for Model B Series t A2

5 variable 5

t A3

Z (km)

5 5 variable

65 – 70 60 – 65 55 – 60 50 – 55 45 – 50 40 – 45 35 – 40

while models A2 and A3 have variable optical thicknesses in MCL and LCL, respectively (Table 2). The initial results indicate the nominal continuum absorption, 7.0
109 cm1 amagat2, too strong (Figure 7). It is found that half the nominal value yields models’ W2C ratios at the level of observed main branch. We therefore adopt 3.5
109 cm1 amagat2 in the following computations. [27] By examining the results shown in Figure 7, it is found that the model is insensitive to the exact location of cloud opacity within 50– 60 km altitude range (models A1 and A2). Model 3, on the other hand, exhibits an increase of W2C ratio for low-intensity region, similar to the behavior of observed main branch. This may be suggesting that the main opacity is within the lower part of the clouds, consistent with Grinspoon et al.’s [1993] finding. This motivates us to lower the cloud layers by one altitude grid (‘‘type B’’ models in Table 3). Other parameters remain the same as type A. By comparing the results shown in Figure 8 with Figure 7, we find that model B3 with the variable opacity in LCL (40 – 45 km) yields the behavior of

Figure 7. W2C ratios computed with type A models are compared with the observation. The curve for model A3 with the nominal continuum absorption motivates us to half the strength, down to 3.5
109 cm1 amagat2. Curves from models A1 and A2 are almost identical, indicating that the location of aerosols within 50– 60 km altitude range cannot be precisely determined in this analysis.

UCL MCL LCL

g

w

t B1

t B2

t B3

0.75 0.75 0.75

0.997 0.997 0.990

variable 5 5

5 variable 5

5 5 variable

W2C curve similar to the observed main branch. Through these experiments we believe that it is essential to place in our model the main cloud opacity at z = 40– 45 km. This is lower than previously proposed altitudes [cf. Grinspoon et al., 1993], however, it should be emphasized that this likely depends on the choice of continuum absorption strength and/or on other adjustables. We therefore may not be sensitive to the ‘‘absolute’’ altitudes of cloud layers. Still, it should be possible to argue, in ‘‘relative’’ sense, how cloud structure differs between the observed main and upper branches. [28] Now, we examine, on the basis of model B3, how cloud single-scattering albedo affects the results. Variations in this ‘‘type C’’ are shown in Table 4: Table 5 [29] 1. wLCL = 0.970. A strong-absorption case (may imply a considerable amount of contaminant). [30] 2. wLCL = 0.990. The nominal albedo for the LCL particles as obtained from a Mie scattering computation with the mode 3 particle parameters. This value makes the model identical to B3. [31] 3. wLCL = 0.997. All the clouds (UCL, MCL, and LCL) have the same ‘‘high’’ single-scattering albedo.

Figure 8. Same as Figure 7 but for type B models (with a slightly different y axis range). The curve from model B3 shows a good agreement with the observed main branch. Curves from B1 and B2 look just similar as those in Figure 7 but slightly lower. This suggests that the main (variable) opacity needs to be located within 40– 45 km altitude range.

6 of 13

SATOH ET AL.: AEROSOLS IN VENUS’ LOWER ATMOSPHERE

E00B37

E00B37

Table 4. Cloud Parameters for Model C Series Z (km)

UCL MCL LCL

65 – 70 60 – 65 55 – 60 50 – 55 45 – 50 40 – 45 35 – 40

g

t

wC1

wC2

wC3

0.75 0.75 0.75

5 5 variable

0.997 0.997 0.970

0.990

0.997

[32] The results (Figure 9) show that loci of models C1 and C3 envelop almost all the points in the main branch. This means that the observed variation in W2C ratio could be explained by variation of cloud single-scattering albedo within a range, 0.970 w 0.997. To reproduce the upper branch, single-scattering albedos lower than 0.970 are required. Validity of this model will be discussed in the later section. 4.2. Types D and E: Models With Hazes [33] In ‘‘type D’’ models, we introduce a layer of small aerosols (LHZ) at an altitude range 30– 40 km, just below the LCL of model C2 (= B3). The anisotropy factor (g = 0.65) and the single-scattering albedo (w = 0.997) are those obtained for mode 1 particles through Mie scattering computation. The optical thickness of LHZ is allowed to vary and 3 different values, 1, 2, and 4, are tested (Table 4). The results shown in Figure 10 are surprising because a small amount of opacity introduced below the cloud base significantly lower the model’s W2C ratio. [34] Then, we include both lower and upper hazes in the model (‘‘type E’’). An upper haze layer (w = 0.997 and g = 0.65) is added to models D1 – D3. The optical thickness for the upper haze is fixed to t UHZ = 5. This value is chosen arbitrarily because the purpose of type E models are just to know how upper haze could affect the model intensities and W2C ratios. Model parameters are summarized in Table 6 and the results shown in Figure 11. [35] These models do not significantly improve type D models. Probably, the horizontal portion of the locus may be slightly closer to the observations than type D models. To summarize the models A through E, there seems to be many ways of reproducing the main branch. On the other hand, none of these models can reproduce the upper branch. Recall that we halved the strength of continuum absorption by just referring to the initial results of type A model computation (Figure 7). We now believe that more continuum absorption is needed, probably between the nominal strength and a half of it. So we are motivated to adjust the continuum absorption again.

Figure 9. Same as Figure 8 but for type C models. The model C2 is identical with B3. Models C1 and C3 envelope the observed main branch. The upper branch, on the other hand, cannot be reproduced unless very low wLCL is assumed.

4.3. Type F: Models With More Continuum Absorption [36] We have readjusted the continuum absorption to 5.6
109 cm1 amagat2, 80% of the value of Pollack et al.

Table 5. Cloud Parameters for Model D Series Z (km)

UCL MCL LCL LHZ

65 – 70 60 – 65 55 – 60 50 – 55 45 – 50 40 – 45 30 – 40

g

0.75 0.75 0.75 0.65

w

0.997 0.997 0.990 0.997

t D1

5 5 variable 1

t D2

2

t D3

4

Figure 10. Same as Figure 8 but for type D models. The model D0 is identical with C2 and B3. It is surprising that only a small amount of aerosol opacity introduced in 30– 40 km altitude range substantially lowers the model’s W2C ratio.

7 of 13

SATOH ET AL.: AEROSOLS IN VENUS’ LOWER ATMOSPHERE

E00B37

E00B37

Table 6. Cloud Parameters for Model E Series

UHZ UCL MCL LCL LHZ

Z (km)

g

w

t E1

t E2

t E3

65 – 70 60 – 65 55 – 60 50 – 55 45 – 50 40 – 45 30 – 40

0.65 0.75 0.75 0.75 0.65

0.997 0.997 0.997 0.990 0.997

5 5 5 variable 1

2

4

[1993]. The new value is so chosen that the model E1, without changing the aerosol parameters, can fit the observed upper branch. This model, ‘‘type F1’’ (t LHZ = 1), is plotted in Figure 12 and matches nicely to the right half of the upper branch. After adjustment of the continuum absorption, model F2 (the same parameters as E2) is found to reproduce the left half of the upper branch. We have found that model F3 (from E3) with t LHZ = 4 reproduces the upper edge of the main branch. Additionally, we have constructed the model F4 with t LHZ = 8 which traces the lower edge of the main branch. [37] On each locus in Figure 12, total aerosol optical thickness (35, 40, 47, and 55) is marked. This is the sum of optical thicknesses in UHZ, UCL, MCL, LCL, and LHZ and no gaseous opacity is included. As first three are fixed to 5 (Table 7), t LCL is given as t LCL = t total  15  t LHZ. [38] These ‘‘type F’’ models are favored as the observation can be explained with a minimal number of variables. To explain the observed intensity variation, the optical thickness of the lower cloud (t LCL) plays the key role. However, by varying only the LCL opacity, we obtain just one locus as seen in Figure 12. To jump from one locus to another, we need to change the lower haze opacity in the

Figure 12. Same as Figure 11 but after adjusting the continuum absorption to 5.6
109 cm1 amagat2. The model E1 (Figure 11) becomes F1 and E2 becomes F2. Both models reproduce the observed upper branch well. The model E3 becomes F3 and traces the upper edge of the main branch. The lower edge is traced with model F4 (t LHZ = 8).

range 1 t LHZ 8. We conclude that this is the simplest model which can reasonably explain the observations.

5. Discussion 5.1. Comparison Between Type C and Type F Models [39] We compare models C1– C3 with models F2– F4. The former explains variation of W2C ratio with variation of aerosol single-scattering albedo, while the latter with variation of subcloud aerosol opacity. 5.1.1. Limb Darkening on the Nightside Disk [40] In Figure 13, the limb-darkening curves for models C3 and F4 are compared with an empirical linear approximation derived by Carlson et al. [1991], I(m)/I(m = 1.0) = 0.316 + 0.685m (open circles). Both cloud models yield the limb-darkening curves (C3 and F4 nearly identical) which are in an excellent agreement with Carlson et al.’s [1991] approximation. Assuming such an approximation is valid for VEX/VIRTIS data acquired 17 years after Galileo/NIMS data, it can be concluded that models C3 and F4 are equally

Table 7. Cloud Parameters for Model F Series

Figure 11. Same as Figure 8 but for type E models. Adding the upper haze (55 – 60 km altitude) only slightly lowers the model’s W2C ratio.

UHZ UCL MCL LCL LHZ

8 of 13

Z (km)

g

w

t F1

t F2

t F3

t F4

65 – 70 60 – 65 55 – 60 50 – 55 45 – 50 40 – 45 30 – 40

0.65 0.75 0.75 0.75 0.65

0.997 0.997 0.997 0.990 0.997

5 5 5 variable 1

2

4

8

E00B37

SATOH ET AL.: AEROSOLS IN VENUS’ LOWER ATMOSPHERE

Figure 13. Limb-darkening curves for the 1.74 mm nightside emission computed with model F4 (solid line) and C3 (dotted line). Computed intensities are plotted against cosine of the emergent angle (m) after normalized to the disk center (m = 1.0) intensity. Also shown (open circles) is Carlson et al.’s [1991] linear approximation derived for the Galileo/NIMS observation (see text). good to reproduce the limb-darkening curves observed on the nightside of Venus in the 1.74 mm window. 5.1.2. Reflectivity on the Dayside Disk [41] When applying the same models to the dayside disk, it should be reminded that a ‘‘single-peak’’ Henyey-Greenstein function used in our models has limitations. As described in section 3.3, phase functions with similar anisotropy g behave almost identical in the nightside simulations. On the other hand, they yield substantial differences in reflectivities (a factor 2 at m = 0.5) when used in simulations of the dayside reflectivities. Therefore, we may only be able to compare ‘‘differences’’ (or contrast) in reflectivities of different models. [42] Total aerosol opacities are adjusted as follows: t total = 30 for model C1, 45 for C2, 70 for C3, and t total = 40 for models F2, F3, and F4, in order that the models give simulated nightside emission I1.74mm  0.1 W m2 sr1 mm1. Then, the models are used to compute the intensity distribution across the dayside disk (Figure 14). Curves for type C models are normalized to the peak intensity of model C3 curve (at m = 1.0). Curves for type F models are represented by the model F4 curve as other two (F2 and F3) are just identical in their shapes and also the absolute values with F4. [43] It is found that type C models predict substantial contrast up to 20% between C1 and C3, while type F models allow no noticeable contrast on the dayside disk. Such high contrast markings on the dayside disk at 1.74 mm have not been reported. Examinations of VIRTIS 1.74 mm images acquired with a shorter exposure time (0.02 s) also seem to support ‘‘featureless’’ dayside disk of Venus. Therefore, it can be concluded that type F models are more favored than type C.

E00B37

5.2. Optical Properties of the Lower Haze 5.2.1. Single-Scattering Albedo [44] One of constraints on the subcloud haze we have obtained is high single-scattering albedo. To be influential on the W2C ratio (Figure 12), the subcloud haze particles should allow radiation they scatter to continue traveling in the atmosphere. Then, en route an extended path due to scattering, the radiation would be more attenuated by the gaseous absorption at 1.71 mm than at 1.74 mm, resulting in a higher contrast between 2 bands (or, in other words, lower W2C ratio). Lowering the subcloud haze albedo results in higher W2C ratio, the same effect as obtained by increasing the wavelength-independent continuum absorption. Our value w = 0.997 well compares to those by Ekonomov et al. [1983]: the single-scattering albedo in the altitudes 15– 48 km is 0.9986 at a wavelength 0.726 mm (Venera 11) and 0.9969 at 0.634 mm (Venera 12). 5.2.2. Anisotropy of Single Scattering [45] The anisotropy of the scattering phase function allows a similar insight. With strongly forward-scattering aerosols (i.e., larger g), scattered radiation may propagate in nearly the same direction as before the scattering. This causes just minimal increase in the traveling light path. In contrast, weakly forward-scattering aerosols (smaller g) make the radiation more ‘‘scattered.’’ This makes the light path much longer than the case of strongly forward-scattering aerosols, resulting in greater attenuation of radiation. [46] To strengthen the case, an additional test with ‘‘type G’’ models is performed (Figure 15). The anisotropy factor, g, is decreased to 0 [Ekonomov et al., 1983], making the

Figure 14. Limb-darkening curves of the reflected sunlight on the dayside disk at 1.74 mm computed with type C and type F models. Curves for models C1 – C3 are normalized to model C3’s disk center brightness. Models F2– F4 are represented by the curve for model F4 as they are all identical with each other (normalized to model F4’s disk center brightness). Note the large contrast predicted by models C1 – C3 for the dayside disk.

9 of 13

E00B37

SATOH ET AL.: AEROSOLS IN VENUS’ LOWER ATMOSPHERE

E00B37

By inverting this, we can map C values over an image frame. Then, the lower haze opacity (t LHZ) can be obtained as follows:  t LHZ ¼

Figure 15. Same as Figure 12 but with ‘‘isotropically scattering’’ lower haze (g = 0 for the scattering phase function). This model requires smaller optical thicknesses, compared to models F1 –F4, for the lower haze (t LHZ = 0.5  3) and for the lower cloud (t total = 30  50). These models seem to be consistent with previous probe measurements and, therefore, are favored as the best-fit models. subcloud aerosols isotropically scattering to mimic extremely fine particles. All other parameters remain unchanged from type F models. The results same as shown in Figure 12 can be obtained by models with smaller optical thicknesses for the lower haze (t LHZ) and for the lower cloud (t LCL). We find the following: [47] 1. Observed W2C ratios can be explained with smaller lower haze opacity, t LHZ 3. This is in a better agreement with small optical thicknesses (t  3) inferred from Venera probe measurements [Marov et al., 1980]. [48] 2. The range of observed intensities, I1.74mm, can be explained with the total aerosol opacity of approximately 30– 50. This does not seem to be too far from previously suggested range, 25– 35 for the visible wavelengths [Ekonomov et al., 1983]. [49] The observations seem to support smaller anistropy (and higher single-scattering albedo) for the subcloud haze which may imply concentration of fine particles although it is not all that simple to infer, from only g values, the actual particle size. We now conclude that the best-fit model is type G.

 C  0:054 2 : 0:030

ð4Þ

A grayscale map of t LHZ for VI0344_01 data is compared with the 1.74 mm image in Figure 16. [51] Variability of t LHZ in the derived map is substantial, from almost 0 to 4 (Figure 16). This can be compared with previous probe measurements. In the Pioneer Venus probe experiments, the nephelometer (Day probe) and LCPS (Large probe) detected the lower haze. Because the nephelometer’s lower limit of detection is 2 orders of magnitude higher than LCPS, the lower haze at the Day probe site had 2 orders of magnitude stronger backscatter (at 900 nm) than at the Large probe site. The great spatial variability we have obtained seems to be consistent with this. Note that a large area of low t LHZ (labeled f), south of approximately 40°S latitude, corresponds to the upper branch in the W2C – I1.74mm plot. [52] The lower haze opacity basically anticorrelates with the 1.74 mm intensity. Locations indicated with labels c and e are dark in I1.74mm and have higher t LHZ (up to 4). This anticorrelation may imply that subcloud aerosols provide condensation nuclei for the lower cloud. There are, however, some exceptions. A dark band (b) seen in I1.74mm does not have a counterpart in the t LHZ map. The same is true for dark features near the image center (d). Such features may, therefore, be produced by local temperature fluctuations caused by some sort of waves. [53] There is one place where high I1.74mm corresponds to large t LHZ (labeled a). We think this very important. This may imply that penetration of sunlight through the clouds triggers a chain of photochemical reactions, producing subcloud aerosols in less cloudier regions. One of such hypothesis has been proposed by Yung and Liang [2008]. We discuss possible composition of the subcloud haze later. 5.3.1. Examining Other VIRTIS Data [54] We have examined additional data to find more cases that exhibit positive correlations between I1.74mm and t LHZ.

5.3. Lower Haze Opacity Maps and Their Implications [50] Now we map the lower haze opacity (t LHZ) using type G models. It is found that the four curves in Figure 15 can reasonably be approximated by     I1:74mm þ 0:005 1 ; W 2C ¼ 0:28 þ C
exp  0:023

ð3Þ

where C is 0.075, 0.084, 0.096, and 0.106 for models G1 (t LHZ = 0.5), G2 (1.0), G3 (2.0), and G4 (3.0), respectively.

Figure 16. A map of (left) lower haze opacity is compared with (right) 1.74 mm image. The lower haze opacity basically anticorrelates with the 1.74 mm intensity with some exceptions, such as those labeled a, b, and d (see text).

10 of 13

E00B37

SATOH ET AL.: AEROSOLS IN VENUS’ LOWER ATMOSPHERE

Figure 17. Same as Figure 16 but for VI0382_02 data cube. Normal regions (h and j) exhibit anticorrelations between t LHZ and I1.74mm. A large area of positive correlation, surrounded by sharp edges, exists (labeled k). Note that saturated pixels are blacked out on the right and appear white on the left.

There are only seven orbits during which VIRTIS data were acquired with 18 s exposures. Among those, we have found a very interesting feature in VI0382_2 cube (Orbit 382, 07– 08 May 2007). A grayscale map of t LHZ for this cube is compared with the 1.74 mm image in Figure 17. The latitude range is from 65°S (bottom edge of the map) to 23°S (top edge), similar to that of VI0344_01 (Table 1). [55] A large area, with sharp edges, of enhanced t LHZ is obvious (labeled k). Inside of this area appears rather flat in spite of the fact that the I1.74mm map indicates cloud opacity variations in it. If the area corresponds to where production of subcloud aerosols is enhanced owing to penetration of incident sunlight, it may be expected to find more aerosols when observed in the day. We speculate that the observation by nephelometer on board Pioneer Venus Day probe was made in one of such regions (the probe descended the atmosphere at 30°S, not too far from the latitude of this feature). [56] Labels h and j indicate ‘‘normal’’ regions where the anticorrelation between t LHZ and I1.74mm is seen. No systematic decrease of lower haze opacity toward the south pole is observed in this map, very different from that is shown in Figure 16. Temporal and spatial variability is therefore substantial. 5.3.2. Atmospheric Circulation and the Lower Haze [57] The fact that t LHZ only weakly correlates with the 1.74 mm intensity may imply that the dynamical coupling between the lower cloud and the subcloud haze is not strong. This is not surprising because, in the probe measurements [Tomasko, 1983], the altitude range 31 – 48 km is considered ‘‘stable’’ (i.e., the measured temperature lapse rate dT/dz is larger than the adiabatic lapse rate G). In such a region, a upwelling motion that an air parcel gets when it is heated will soon be suppressed because the air parcel cools down with a rate larger than the lapse rate of its surroundings. [58] Schubert [1983] proposed a hypothesis of several (vertically stratified) circulation cells in Venus atmosphere. One within the cloud layer (50 – 70 km altitude) which is directly driven by absorption of the solar energy by the clouds. Another direct cell is in the lowermost atmosphere

E00B37

which is driven by absorption of the solar energy by the ground. This cell is expected to extend up to 40 km altitude. Between two direct cells (40 – 50 km altitude) is an indirect cell frictionally driven by direct cells above and below it. The wind directions (alternating equatorward and poleward) measured by the Pioneer Venus probes [Counselman et al., 1980] in the middle-to-lower atmosphere are consistent with this hypothesis. [59] According to Schubert’s proposal, equatorward motion is expected for the altitude 50 km near the lower limb of the direct cell. On the other hand, poleward motion is expected for the altitude 40 km near the lower limb of the indirect cell. It should therefore be of great interest to track the meridional motion of features in the lower cloud and in the lower haze. 5.4. Possible Composition of Subcloud Haze [60] Very little is known to date about the composition of the subcloud aerosols. Golovin and Ustinov [1982] analyzed spectrophotometric data obtained from Venera 9 – 12 and have constrained some properties of subcloud aerosols. They have found that the particles are small (0.05 – 0.1 mm radii) and are mainly in altitudes above 30 km. They also note detection of some larger particles (0.1 – 0.3 mm) detected at Venera-12 landing site. Refractive index has only crudely been constrained to n 1.8 by attenuation effectiveness factor measured by Venera 12 for larger (0.2 – 0.3 mm) particles. They listed a variety of candidate species, such as chlorine-containing salts (KCl, NaCl, MgCl2, PbCl2, etc.), Pb, PbBr2, PbS, PbO, Zn and some others. [61] Other possible candidates for the lower haze are H2SO4 plus some contaminants as proposed by Esposito et al. [1983] and Fe2(SO4)3 proposed by Krasnopolsky [1985]. Neither of these as well as those listed by Golovin and Ustinov [1982] has been confirmed to date. [62] Recently, Yung and Liang [2008] proposed that sulfur atoms and molecules (up to S8) are produced from OCS by photosensitized dissociation processes in the 30 km region. Such processes may significantly reduce OCS in the lower atmosphere, consistent with the observed OCS vertical profile. Polymeric and alpha-sulfur, products of such processes, are nonabsorbing at 1.7 and 2.3 mm, and thus may be good candidates for the subcloud aerosols. Increases of t LHZ found for regions a (Figure 16) and k (Figure 17), where overlying clouds are thinner, may support this hypothesis. Another support to this comes from Bertaux et al. [1986]. They analyzed the absorption spectrum in the ultraviolet obtained from spectrometers on board Vega 1 and 2 and have found that the principal absorbent in the subcloud region is S8 allotrope of sulfur in the gas phase, with mixing ratios of several to a few tens of ppm. [63] Although the composition of subcloud aerosols remain unclear at this moment, continuous mapping/monitoring (as demonstrated here) and detailed analysis of high-resolution spectroscopy (lower atmosphere chemistry), combined with laboratory measurements and theories would improve our knowledge on this.

6. Conclusions [64] We have developed a new method of constraining Venus cloud structure in the middle-to-lower atmosphere

11 of 13

E00B37

SATOH ET AL.: AEROSOLS IN VENUS’ LOWER ATMOSPHERE

based on analysis of nightside emission in the 1.74 mm window. Ratios of intensities at 2 spectral bands, one at the window center (1.74 mm) and another off center (1.71 mm), are found to be affected by the distribution and properties in the lower cloud and haze. This method may make it possible to map variations in the horizontal distribution of the lower haze with the remote sensing data. Owing to limitations of our current knowledge on gaseous absorptions (line shapes, line mixings, collision-induced continuum absorption, etc.), determining absolute values for the lower haze parameters is needless to say difficult. The analysis in this paper finds variation in the W2C ratio within a single VIRTIS data cube, VI0344_01 (30 – 31 March 2007), so we feel confident that the best explanation is spatial variability of aerosol distribution in the lower atmosphere. [65] The data favor models with shiny and nearly isotropically scattering particles (i.e., very small in size) for the lower haze of which composition is to date quite uncertain. The subcloud haze plays a key role, together with the lower cloud, to explain remarkable variations in the observed W2C ratios and the 1.74 mm intensities. The range of optical thickness found for the lower haze (t LHZ  0.5 –3) is consistent with the Venera probe measurements [Marov et al., 1980]. The total aerosol optical thickness is found to be in the range 30– 50, also not too far from the typical range, 25– 35, inferred from in situ solar-radiation flux measurements [Ekonomov et al., 1983]. [66] The question is how stable or unstable these variations of the lower cloud/haze are. It is at this moment unclear how variabilities of aerosols in the middle-to-lower atmosphere are related (or unrelated) to the atmospheric circulations. As this study is the first demonstration of mapping the lower haze distribution using VEX/VIRTIS remote sensing data, further monitoring observations, combined with measurements of trace gas species in the lower atmosphere, would greatly contribute to understandings on the current status and physics/chemistry in this hidden world. [67] Acknowledgments. We are grateful to Robert Carlson and Colin Willson for very useful discussion about the Venus aerosol properties and distributions. Thanks go to Bruno Bezard for providing his analysis on the ‘‘spectral shift’’ of VIRTIS. Dima Titov kindly gave us comments which are of great value to improve our analysis. The VEX/VIRTIS project has been made possible by all-out support from ASI and CNES. T.S. is one of the Interdisciplinary Scientists (IDS) to Venus Express mission and is grateful to ESA for providing such a great opportunity.

References Allen, D., and J. Crawford (1984), Discovery of cloud structure on the dark side of Venus, Nature, 307, 222 – 224. Bertaux, J.-L., et al. (1986), Active spectrometry of the ultraviolet absorption within the Venus atmosphere, Sov. Astron. Lett., Engl. Transl., 12(1), 33 – 36. Carlson, R. W., and F. W. Taylor (1993), The Galileo encounter with Venus: Results from the Near-Infrared Mapping Spectrometer, Planet. Space Sci., 41, 475 – 476. Carlson, R. W., et al. (1991), Galileo infrared imaging spectroscopy measurements at Venus, Science, 235, 1541 – 1548. Counselman, C. C., III, A. A. Gourevitch, R. W. King, and G. B. Loriot (1980), Zonal and meridional circulation of the lower atmosphere of Venus determined by radio interferometry, J. Geophys. Res., 85, 8026 – 8030. Crisp, D., and D. Titov (1997), The thermal balance of the Venus atmosphere, in Venus: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 353 – 384, Univ. of Ariz. Press, Tucson.

E00B37

Drossart, P., et al. (2007), A dynamic upper atmosphere of Venus as revealed by VIRTIS on Venus Express, Nature, 450, 641 – 645. Ekonomov, A. P., Yu. M. Golovin, V. I. Moroz, and B. Ye. Moshkin (1983), Solar scattered radiation measurements by Venus probes, in Venus, edited by D. M. Hunten et al., pp. 632 – 649, Univ. of Ariz. Press, Tucson. Esposito, L. W., R. G. Knollenberg, M. Ya. Marov, O. B. Toon, and R. P. Turco (1983), The clouds and hazes of Venus, in Venus, edited by D. M. Hunten et al., pp. 484 – 564, Univ. of Ariz. Press, Tucson. Esposito, L. W., J.-L. Bertaux, V. Krasnopolsky, V. I. Moroz, and L. V. Zasova (1997), Chemistry of lower atmosphere and clouds, in Venus: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 415 – 458, Univ. of Arizona Press, Tucson. Golovin, Yu. M., and E. A. Ustinov (1982), Subcloud aerosol on Venus, Cosmic Res., 20, 104 – 110. Grinspoon, D. H., et al. (1993), Probing Venus’s cloud structure with Galileo NIMS, Planet. Space Sci., 41, 515 – 542. Hansen, J. E., and J. W. Hovenier (1974), Interpretation of the polarization of Venus, J. Atmos. Sci., 31, 1137 – 1160. Hansen, J. E., and L. D. Travis (1974), Light scattering in planetary atmospheres, Space Sci. Rev., 16, 527 – 610. Kamp, L. W., and F. W. Taylor (1990), Radiative transfer models of the night side of Venus, Icarus, 86, 510 – 529. Kamp, L. W., F. W. Taylor, and S. B. Calcutt (1988), Structure of Venus atmosphere from modelling of night side infrared spectra, Nature, 336, 360 – 362. Kawabata, K., D. L. Coffeen, J. E. Hansen, W. A. Lane, M. Sato, and L. D. Travis (1980), Cloud and haze properties from Pioneer Venus polarimetry, J. Geophys. Res., 85, 8129 – 8140. Knollenberg, R. G., and D. M. Hunten (1980), Microphysics of the clouds of Venus: Results of the Pioneer Venus particle size spectrometer experiment, J. Geophys. Res., 85, 8039 – 8058. Krasnopolsky, V. A. (1985), Chemical composition of Venus clouds, Planet. Space Sci., 33, 109 – 117. Levoni, C., E. Cattani, M. Cervino, R. Guzzi, and W. Di Nicolantonio (2001), Effectiveness of the MS-method for computation of the intensity field reflected by a multi-layer plane-parallel atmosphere, J. Quant. Spectrosc. Radiat. Transfer, 69, 635 – 650. Marov, M. Ya., V. E. Lystsev, V. N. Lebedev, N. L. Lukashevich, and V. P. Shari (1980), The structure and microphysical properties of the Venus clouds: Venera 9, 10, and 11 data, Icarus, 44, 608 – 639. Palmer, K. F., and D. Williams (1973), Optical constants of sulfuric acid: Application to the clouds of Venus?, Appl. Opt., 14, 208 – 219. Piccioni, G., et al. (2007), South-polar features on Venus similar to those near the north pole, Nature, 450, 637 – 640. Pollack, J. B., et al. (1993), Near-infrared light from Venus’ nightside: A spectroscopic analysis, Icarus, 103, 1 – 42. Ragent, B., and J. Blamont (1980), Structure of the clouds of Venus: Results of the Pioneer Venus nephelometer experiment, J. Geophys. Res., 85, 8089 – 8105. Sato, M., L. D. Travis, and K. Kawabata (1996), Photopolarimetry analysis of the Venus atmosphere in polar regions, Icarus, 124, 569 – 585. Schubert, G. (1983), General circulation and the dynamical state of the Venus atmosphere, in Venus, edited by D. M. Hunten et al., pp. 681 – 765, Univ. of Ariz. Press, Tucson. Seiff, A. (1983), Models of Venus’s atmospheric structure, in Venus, edited by D. M. Hunten et al., pp. 1045 – 1048, Univ. of Ariz. Press, Tucson. Sill, G. T. (1972), Sulfuric acid in the Venus clouds, Comm. Lunar Planet. Lab., 9, 191 – 198. Stamnes, K., S.-C. Tsay, W. Wiscombe, and K. Jayaweera (1988), Numerically stable algorithm for discrete-ordinate-method radiative transfer in multiple scattering and emitting layered media, Appl. Opt., 27, 2502 – 2509. Svedhem, H., et al. (2007), Venus Express: The first European mission to Venus, Planet. Space Sci., 55, 1636 – 1652. Taylor, F. W., D. Crisp, and B. Bezard (1997), Near-infrared sounding of the lower atmosphere of Venus, in Venus: Geology, Geophysics, Atmosphere, and Solar Wind Environment, edited by S. W. Bougher et al., pp. 325 – 351, Univ. of Ariz. Press, Tucson. Tomasko, M. G. (1983), The thermal balance of the lower atmosphere of Venus, in Venus, edited by D. M. Hunten et al., pp. 604 – 631, Univ. of Ariz. Press, Tucson. Tomasko, M. G., R. A. West, and N. D. Castillo (1978), Photometry and polarimetry of Jupiter at large phase angles: 1. Analysis of imaging data of a prominent belt and a zone from Pioneer 10, Icarus, 33, 558 – 592. Toon, O. B., B. Ragent, D. Colburn, J. Blamont, and C. Cot (1984), Large, solid particles in the clouds of Venus: Do they exist?, Icarus, 57, 143 – 160. Young, A. T., and L. D. G. Young (1973), Comments on the composition of the Venus cloud tops in light of recent spectroscopic data, Astrophys. J., 179, 39 – 43.

12 of 13

E00B37

SATOH ET AL.: AEROSOLS IN VENUS’ LOWER ATMOSPHERE

Yung, Y., and M. Liang (2008), Chemical sources and sinks of OCS in the lower atmosphere and on the surfaces of Venus, paper presented at General Assembly 2008, Eur. Geosci. Union, Vienna, 13 – 18 Apr. 

P. Drossart, LESIA, Observatoire de Paris, F-92195 Meudon, France. G. L. Hashimoto, Laboratory for Earth and Planetary Atmospheric Science, Organization of Advanced Science and Technology, Kobe University, 1-1 Rokkodai-cho, Nada-ku, Kobe 657-8501, Japan.

E00B37

T. Imamura and T. Satoh, Space Plasma Physics Department, Institute of Space and Astronautical Science, Japan Aerospace Exploration Agency, 3-1-1 Yoshinodai, Sagamihara, Kanagawa 229-8510, Japan. (satoh@stp. isas.jaxa.jp) N. Iwagami, K. Mitsuyama, and S. Sorahana, Department of Earth and Planetary Science, Graduate School of Science, Tokyo University, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan. G. Piccioni, INAF, Istituto di Astrofisica Spaziale e Fisica Cosmica, I-00133 Rome, Italy.

13 of 13

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114, E00B42, doi:10.1029/2008JE003186, 2009

for

Full Article

Preliminary characterization of the upper haze by SPICAV/SOIR solar occultation in UV to mid-IR onboard Venus Express V. Wilquet,1 A. Fedorova,2 F. Montmessin,3,4 R. Drummond,1 A. Mahieux,1 A. C. Vandaele,1 E. Villard,3,4 O. Korablev,2 and J.-L. Bertaux3,4 Received 9 May 2008; revised 19 January 2009; accepted 22 April 2009; published 10 July 2009.

[1] The Spectroscopy for Investigation of Characteristics of the Atmosphere of

Venus/Solar Occultation at Infrared (SPICAV/SOIR) suite of instruments onboard the Venus Express spacecraft comprises three spectrometers covering a wavelength range from ultraviolet to midinfrared and an altitude range from 70 to >100 km. However, it is only recently (more than 1 year after the beginning of the mission) that the three spectrometers can operate simultaneously in the solar occultation mode. These observations have enabled the study of the properties of the Venusian mesosphere over a broad spectral range. In this manuscript, we briefly describe the instrument characteristics and the method used to infer haze microphysical properties from a data set of three selected orbits. Discussion focuses on the wavelength dependence of the continuum, which is primarily shaped by the extinction caused by the aerosol particles of the upper haze. This wavelength dependence is directly related to the effective particle radius (cross section weighted mean radius) of the particles. Through independent analyses for the three channels, we demonstrate the potential to characterize the aerosols in the mesosphere of Venus. The classical assumption that the upper haze is only composed of submicron particles is not sufficient to explain the observations. We find that at high northern latitudes, two types of particles coexist in the upper haze of Venus: mode 1 of mean radius 0.1
rg
0.3 mm and mode 2 of 0.4
rg
1.0 mm. An additional population of micron-sized aerosols seems, therefore, needed to reconcile the data of the three spectrometers. Moreover, we observe substantial temporal variations of aerosol extinction over a time scale of 24 h. Citation: Wilquet, V., A. Fedorova, F. Montmessin, R. Drummond, A. Mahieux, A. C. Vandaele, E. Villard, O. Korablev, and J.-L. Bertaux (2009), Preliminary characterization of the upper haze by SPICAV/SOIR solar occultation in UV to mid-IR onboard Venus Express, J. Geophys. Res., 114, E00B42, doi:10.1029/2008JE003186.

1. Introduction [2] Venus is completely enshrouded in clouds which show an enormous vertical extent of more than 50 km. Venus’ clouds are mainly found in a cloud deck located between 45 and 70 km of altitude, with thin hazes above and below. Much of the clouds are composed of H2SO4 aerosol particles showing a multimodal particle size distribution. The upper haze (70 – 90 km) is composed of submicron aerosol particles with an effective radius below 0.3 mm [Lane and Opstbaum, 1983; Sato et al., 1996]. The composition of these particles was derived from the Pioneer Venus Orbiter Cloud photopolarimeter experiment [Kawabata et al., 1980]. A refractive index of 1.4 was measured, which is in agreement with a haze consisting of 1

Belgian Institute for Space Aeronomy, Brussels, Belgium. Space Research Institute, Moscow, Russia. 3 Service d’Ae´ronomie du CNRS, Verrie`res-le-Buisson, France. 4 Institut Pierre Simon Laplace, Universite´ de Versailles Saint Quentin, Guyancourt, France. 2

Copyright 2009 by the American Geophysical Union. 0148-0227/09/2008JE003186$09.00

concentrated sulfuric acid (75%). Kawabata et al. [1980] showed that the haze is thicker at high latitudes and very diffuse in the tropics. Larger particles are found at lower altitudes. Particles with a mean radius of about 1.0 mm are the principal constituent of the main cloud deck. Even bigger particles with a mean radius larger than 4.0 mm have been observed at levels below the tropopause. [3] The Spectroscopy for Investigation of Characteristics of the Atmosphere of Venus (SPICAV) instrument [Bertaux et al., 2007], which is composed of three independent spectrometers, namely the UV and IR spectrometers and the Solar Occultation at Infrared (SOIR) instrument, was developed in view of investigating the Venus atmosphere from the ground up to the uppermost limit of the atmosphere. Each channel of the instrument can perform, vertical profiling by mean of solar occultation, this is the first time at Venus. Besides the detection of CO2, H2O, and HDO, as well as other minor constituents, this instrument provides unrivalled information on the aerosol loading of the atmosphere. Taken separately all three channels, operating in the solar occultation mode, can contribute to the characterization of the aerosol extinction as well as, in some cases, of

E00B42

1 of 13

E00B42

WILQUET ET AL.: UPPER VENUSIAN HAZE BY SPICAV/SOIR

the aerosol particle size. We will show that if we combine their measurements, it is moreover possible to define a bimodal distribution of the aerosol particles size. This information is of great interest as it is known that aerosol particles, through absorption and scattering of solar radiation, play a major role in the energetic balance of the atmosphere.

2. SPICAV/SOIR Instrument [4] SPICAV is a suite of three spectrometers in the UV and IR range dedicated to the study of the atmosphere of Venus from ground level to the outermost hydrogen corona at more than 40,000 km. It is derived from the SPICAM instrument flying on board Mars Express (MEX), with the addition of a new IR high-resolution spectrometer SOIR [Bertaux et al., 2007]. The UV spectrometer (118– 320 nm, resolution 1.5 nm) is identical to the MEX version. It is dedicated to nadir viewing, limb viewing and vertical profiling by stellar and solar occultation. The SPICAV IR sensor (0.65– 1.7 mm, resolution 0.5– 1.2 nm) which makes use of an Acoustooptical tunable filters (AOTF), investigates the dayside solar radiation reflected from the clouds and the nightside thermal emission coming through the clouds in nadir mode, and in solar occultation mode, this channel studies the vertical profiles of H2O, CO2, and aerosols. The pioneer SOIR spectrometer is a solar occultation IR spectrometer with the highest spectral resolution on board VEX. This new concept of instrument includes a combination of an echelle grating and an AOTF crystal to select one order at a time. One of the main objectives of SOIR is to measure HDO and H2O in solar occultation [Bertaux et al., 2007; Fedorova et al., 2008a; Svedhem et al., 2007], in order to characterize the escape of D atoms from the upper atmosphere and give more insight about the evolution of water on Venus. However, the instrument is also well adapted to investigate the isotopologues of CO2 [Bertaux et al., 2008; Wilquet et al., 2008] and other minor species [Vandaele et al., 2008]. The three channels of the SPICAV/SOIR instrument will be described in the following sections more specifically in view of their capabilities to characterize the aerosols present in the Venus atmosphere. The instrument is on the Venus Express spacecraft operating on the orbit around Venus from April 2006. This orbit lasts about 24 h and is highly elliptical with nearly 80°N pericenter [Titov et al., 2006]. The vertical resolution of the occultation is better when the distance to the limb is small, which is the case, for such an elongated orbit, at high northern latitudes. 2.1. UV Channel [5] The ultraviolet channel of SPICAV is a versatile instrument, operating in a wide range of modes: nadir, limb, and solar/stellar occultations. A detailed instrument description can be found in the paper by Bertaux et al. [2007]. The UV channel covers the 118 to 320 nm spectral range with a spectral sampling of 0.54 nm per pixel while the actual spectral resolution determined by optical aberration is three pixels (i.e., 1.5 nm). The UV channel is a compact imaging spectrometer equipped with an image intensifier and a Thomson CCD array of 288  384 pixels.

E00B42

[6] The focal length of the UV telescope is such that one CCD pixel covers a field of view (FOV) of 0.01  0.01°. The total useful FOV of the instrument is, however, limited by a slit located in the focal plane, whose width equals two CCD pixels as imposed by Nyquist-Shannon sampling theorem. Total FOV varies between 0.02°  0.1° and 0.02°  0.4° depending on CCD readout configuration. [7] Owing to downlink limitations, the equivalent of only 5 out of the 288 available pixel lines can be transmitted to Earth. To comply with this restriction while making use of a larger portion of the photosensitive area, pixel lines are first binned together during readout. The number of lines per bin (also referred to as bands) is determined upon consideration of photometry. For solar occultations, five adjacent bins of two to eight lines are used. In order to maintain the same vertical resolution at the limb of 1 km, a particular combination of integration time (ranging between 80 and 450 ms) and line binning (from two to eight) is set to adjust with the altitude of the spacecraft during solar occultation. See the numerous articles describing the binning procedure in more detail [Bertaux et al., 2007; Montmessin et al., 2006a, 2006b; Que´merais et al., 2006]. [8] The signal-to-noise ratio (SNR) has been determined by observing the sun outside the atmosphere where the signal is stable and varies only according to photon statistical noise. This analysis showed that pixel SNR can exceed 150 (SNRmax) in the range 200 to 300 nm for a single measurement. Below 200 nm, solar emission drops rapidly and yields no detector count shortward of 170 nm. As the line of sight traverses progressively deeper regions of the atmosphere, SNR decreases as SNRmaxx(exp(t))1/2 where t is the slant opacity of the atmosphere. [9] UV data treatment is performed through a succession of steps. First, dark charge (DC) and offset are subtracted from pixel content using masked pixels located at the extremity of each pixel line and dedicated to dark charge estimation. A predetermined non uniformity pattern allows extraction of an instantaneous DC value for any pixel of the CCD from reading of the masked CCD pixels. In addition, when a significant portion of the occultation sequence remains behind Venus with respect to the sun, i.e., SPICAV line of sight aims at Venus nightside, the electronic readout noise pattern (with a periodicity of 14 pixels) can be isolated and subtracted throughout the sequence, thereby improving the scientific analysis of low-statistics signals. [10] Atmospheric transmission spectrum is computed as the ratio of the spectrum collected at a given altitude to the reference spectrum of the sun formed by coadding hundreds of solar spectra recorded outside the atmosphere (Figure 1a.). [11] On some sequences, the solar spectrum recorded outside the atmosphere shows a linear drift of the signal amplitude throughout the sequence. The cause of this drift is still not clearly established but appears related to center-tolimb darkening effect of the sun. Intensity variation is then produced by spacecraft line of sight scanning across the solar disk as a result of either the spacecraft slewing in one direction or, most likely, the orbital course of the spacecraft itself projecting at different locations at the surface of the sun. In any case, this effect is corrected for by performing a linear regression of the solar signal variation outside the atmosphere and by extrapolating the linear variation to the reference spectrum over the whole sequence.

2 of 13

WILQUET ET AL.: UPPER VENUSIAN HAZE BY SPICAV/SOIR

E00B42

126 km

a 1

104 km 95 km

0.8

Transmittance

E00B42

91 km

0.6

88 km

0.4

0.2 82 km

0

60 km

200

220

240

260 wavelength (nm)

280

300

Figure 1. Example of transmittance spectra for the three channels during occultation 486. (a) SPICAVUV: transmittance at several altitudes during the occultation. (b) SPICAV-IR: sequential scanning of 664 spectral points recorded by defining three windows (spectral ranges 1288.5 –1309.3 nm, 1334.4 – 1442.5 nm, and 1461.5 – 1471.1 nm) and 11 individual measurements described as ‘‘dots’’ (wavelengths 650.8, 757.2, 852.7, 982.3, 1101.1, 1159.6, 1197.2, 1273.5, 1323.0, 1553.7, and 1626.0 nm). Last dot is not used for the retrieval owing to its sensitivity to a CO2 absorption band. (c) SOIR: a series of transmittances throughout the occultation. Each image corresponds to one selected diffraction order (spectral window). 2.2. Near-IR Channel [12] The SPICAV-IR instrument is part of the SPICAV/ SOIR experiment on board the Venus Express mission. It is a new generation of the AOTF spectrometer (SPICAM-IR) on MEX [Fedorova et al., 2008b; Korablev et al., 2006].

SPICAV-IR works in the range 0.65 –1.7 mm with a spectral resolution of 5.2 cm1 in the LWL channel (1 – 1.7 mm) and 7.8 cm1 in the SWL channel (0.6 –1.05 mm). Two photodiodes register light in the two orthogonal polarizations at the AOTF output. The spectrometer is designed to work in

3 of 13

E00B42

WILQUET ET AL.: UPPER VENUSIAN HAZE BY SPICAV/SOIR

E00B42

Figure 1. (continued) nadir or solar occultation mode. The entry optics (a small lens with an useful diameter of 3 mm and a circular diaphragm) provides an angular field of view of about 40 or 0.07° for solar occultation observations, which corresponds to a vertical resolution of 1 –15 km on the limb of Venus, depending on the spacecraft altitude. SPICAV- IR works on the principle of sequential scanning of the spectrum. Each spectral point is recorded in 2.8 ms. To shorten the spectrum recording time to 2 s (664 spectral points), a regime of ‘‘windows’’ and ‘‘dots’’ has been used (Figure 1b) [see also Korablev et al., 2006]. First, the instrument scans the spectral windows (from longer wavelength downward) then the reference wavelengths are recorded in reversed order. In solar occultation mode, the possible windows are 1461.5 – 1471.1 nm, 1334.4 – 1442.5 nm and 1288.5 – 1309.6 nm, primarily dedicated to measurements of CO2 and H2O abundances from the 1.43 mm and 1.38 mm bands, respectively. Aerosol properties and vertical distributions are determined using 10 continuum points outside the gas absorption bands (Figure 1b). 2.3. SOIR Channel [13] The instrument [Bertaux et al., 2007; Nevejans et al., 2006] and its in-flight performances [Mahieux et al., 2008], including data handling, onboard background subtraction, calibrations of the AOTF and the echelle spectrometer, have already been extensively described, whereas the retrieval technique and some results are discussed in the paper by Vandaele et al. [2008]. SOIR is an echelle grating spectrometer operating in the IR, combined with an AOTF for the selection of the diffraction grating orders. It is designed

to measure the atmospheric transmission in the IR (2.2 – 4.3 mm) at high resolution (0.15 cm1), using the method of solar occultation (Figure 1c). [14] The SOIR detector has 320 columns along the wave number axis and 256 rows along the spatial axis. To avoid saturation, short integration times are used (20 to 30 ms), depending on the wavelength at which the measurement is taken. The background signal is measured and subtracted onboard. In order to improve the signal-to-noise ratio (SNR), a number of measurements can be accumulated as long as the total measuring time remains below 250 ms. [15] The slit height is 30 arc min and it is projected onto 32 rows of the detector. The slit width is equivalent to 2 detector pixels in the spectral direction. The spectral resolution of the spectrometer is very high and varies between 0.13 cm1 at 4.0 mm (order 111) and 0.27 cm1 at 2.3 mm (order 192). The corresponding resolving power [l/Dl = u/Du] exceeds 20,000 [Mahieux et al., 2008]. [16] Owing to telemetry limitations, only eight spectra, each 320 pixels long, can be downloaded per second. During most observations of Venus, these eight spectra are taken in four different diffraction orders (four different RFs applied to the AOTF), each corresponding to two large bins of 16 or 12 rows on the detector.

3. Data Sets and Observation Strategy [17] Figure 2 shows the spatial distribution of all solar occultations performed by the three channels of SPICAV/ SOIR since March 2007, i.e., since it was rendered possible to perform simultaneous observations for SOIR and SPI-

4 of 13

E00B42

WILQUET ET AL.: UPPER VENUSIAN HAZE BY SPICAV/SOIR

E00B42

Figure 2. Spatial map of measurements in the solar occultation mode, acquired since March 2007, for the three channels of the SPICAV/SOIR instrument. Coordinates of the limb tangent points at an altitude of 120 km for SPICAV-UV (circles), SPICAV-IR (inverted triangles), and SOIR (asterisks). CAV in the solar occultation mode. This paper is aimed at demonstrating the capability of the instrument to sound the vertical structure of aerosols in the upper haze through occultation spectroscopy. We will therefore concentrate on a few observations for which we had simultaneous data for the three channels (see Table 1). The line of sight to the Sun is slightly different for the three channels so that the portion of the atmosphere probed is not exactly the same, yet the geophysical positions of the tangent point (the point along the optical path that is the closest to the surface of the planet) are very close. The analysis of the data sets from each channel of the instrument is made independently in order to eliminate any instrumental issues. All these observations are located at high northern latitudes (between +60° and +75°). [18] During a solar occultation, e.g., for a sunset, the measurement cycle is started before the instrument’s line of sight intersects with the top layers of the atmosphere, and a reference spectrum can be recorded for each channel. Below that level, solar light is absorbed by atmospheric components and the intensity of the recorded signal starts to decrease. As spectra obtained during the occultation are divided by the reference solar spectrum recorded outside the atmosphere, any solar or instrumental signature gets removed and the transmittance contains only information about the Venus atmosphere. This is illustrated in Figure 1c where the transmittances obtained during occultation 486 for 4 different spectral windows centered at 2715, 3343, 3838, and 4264 cm1 (orders 121, 149, 171, and 190) are plotted. The evolution of the light transmission during solar

occultation 485 is depicted in Figure 3 for the three channels of the SPICAV/SOIR instrument: two wavelengths for the UV channel, two wavelengths for the IR channel, and the average signal for two diffraction orders recorded during this occultation for SOIR. The vertical resolution depends on the geometry of the solar occultations and the distance to the limbs of VEX. In the orbits analyzed in this paper the vertical resolution is 3.0 km for SPICAV-UV, 3.4 km for SPICAV-IR, and 2.0 km for SOIR.

4. Method of Analysis 4.1. Retrieval of the Extinction Coefficients [19] All three instruments, taken together, provide useful information on the absorption due to aerosols. In this section, we will first indicate how aerosol extinction coefficients are obtained from the measurements, which are Table 1. Distance to the Limb, Longitude, and Latitude for Each Orbit at a Tangent Height of 65 km and 120 km Showing How These Values Vary During the Occultation Distance to the limb (km) Longitude (deg) Orbit

Date

482 484 485 486 487

16 Aug 2007 18Aug 2007 9 Aug 2007 20 Aug 2007 21 Aug 2007

5 of 13

Latitude (deg)

65 km 120 km 65 km 120 km 65 km 120 km 2474 2657 2790 2973 3249

2131 2277 2380 2515 2702

242 246 249 251 253

243 248 250 252 254

+73 +70 +68 +65 +61

+75 +73 +71 +69 +66

E00B42

WILQUET ET AL.: UPPER VENUSIAN HAZE BY SPICAV/SOIR

E00B42

Figure 3. Solar light transmission through the Venus atmosphere of the solar occultation performed during orbit 485 by the SPICAV/SOIR instrument. Two wavelengths for each channel are plotted here: SPICAV-UV at 220 nm (black circles), SPICAV-UV at 300 nm (gray circles), SPICAV-IR at 757 nm (black inverted triangles), SPICAV-IR at 1553 nm (gray inverted triangles), SOIR at 2560 nm (black asterisks), and SOIR at 3978 nm (gray asterisks). different in kind for the three channels of the instrument. Then we will investigate how particle size and number density can be inferred from these measurements. 4.1.1. UV Channel [20] The inversion procedure is almost identical to the one applied to the SPICAM stellar occultations at Mars, with the exception that solar occultations provide a narrower spectral range to extract the CO2 concentration. The reader is therefore referred to the numerous publications dedicated to the subject [Montmessin et al., 2006a, 2006b; Que´merais et al., 2006]. Inverting occultation data only requires consideration of the simple Beer-Lambert’s law, which states that source attenuation scales exponentially with opacity. Spectral inversion is performed independently for every altitude. For every level, a least squares fitting technique (Levenberg-Marquardt) is used to infer the amount of CO2, which possesses a prominent electronic transition between 120 and 200 nm, and the aerosol opacity; both quantities being integrated over the line of sight. Profiles of CO2 local density and aerosol local extinction (b) are obtained after vertical inversion using Abel’s integrals. [21] For each integration, five spectra are recorded simultaneously by the instrument, each corresponding to a distinct position relative to the limb. Each band of the

detector therefore provides an independent profiling of the atmosphere. However, it was found in a number of cases (usually when the spacecraft was far from the limb) that a significant shift existed between the transmission profiles of each band. This was most likely caused by a relative error of altitude registration between the pixels of the CCD. A 5-point boxcar average has therefore been applied to the aerosol profiles to smooth out the scatter of values. 4.1.2. SPICAV IR and SOIR Channels [22] The first step is to determine the extinction (b) in each atmospheric layer, supposed to be homogenous, where we have a measurement by using an onion-peeling method R [Vandaele et al., 2008] and by assuming that bdz = ln(T) where T is the observed transmittance from which line absorptions have been removed, and dz is the thickness of the layer under consideration. Cumulative and local extinction coefficients were retrieved from each of the series of transmittances recorded during one occultation: the values correspond to the average value of the signal, for SOIR, in each selected diffraction order and, for SPICAV IR, for each selected spectral point. Except for some details due to the instrument specificity, the inversion procedure for SPICAV-IR is identical to the one described for the SPICAM solar occultations on Mars [Fedorova et al., 2008b].

6 of 13

WILQUET ET AL.: UPPER VENUSIAN HAZE BY SPICAV/SOIR

E00B42

E00B42

Figure 4. Spectral dependence of the extinction coefficient b during orbit 485. The extinction was obtained from data acquired at two wavelengths with the SPICAV-UV channel (circles), at 10 wavelengths with the SPICAV IR channel (inverted triangles), and at four wavelengths with the SOIR channel (asterisks). The various altitudes of the solar occultation are mentioned in the top right corner. 4.1.3. Spectral Dependence [23] First, we investigate the spectral dependence of the extinction for the different wavelengths probed by the instrument. The extinction coefficients deduced, as explained earlier, from the measurements performed by the three channels are plotted as a function of wavelength in Figure 4, for various altitudes of solar occultation performed during orbit 485. [24] The Angstro¨m exponent (a) is usually used to describe the dependency of the aerosol extinction coefficient on wavelength: b ¼ b0

 a l l0

aerosol extinction for the SOIR data is slight and not monotonic, and therefore is not described by equation (1). The occultations chosen for this paper are selected so that the 4 spectral windows probed by SOIR are well spread over its entire spectral range, in order to maximize the aerosol extinction variations in function of the wavelength (Figure 4). 4.2. Determination of the Radius and of the Total Number Density of the Particles [27] The extinction coefficient is expressed as Zþ1 bð z; lÞ ¼ N ð zÞ f ðrÞQðr; lÞdr

ð1Þ

ð2Þ

0

[25] In the case of the UV channel only two points are shown to facilitate reading. The spectral dependence follows a power law (see Table 2). Usually, all the other points fall in between the two plotted points, defining a clear spectral dependence which allows the unambiguous determination of the particle size, see thereafter. [26] The SPICAV near-IR data also exhibit an obvious spectral dependency, while the spectral dependency of the

Table 2. Values of the Angstro¨m Exponent in the Near-IR and in the UV for Various Sizes of Particles r (mm)

l = 0.3 mm; l0 = 0.2 mm

l = 0.75 mm; l0 = 0.85 mm

l = 1.00 mm; l0 = 0.85 mm

l = 1.55 mm; l0 = 0.85 mm

0.2 0.7 1.0

2.3 0 0

2.07 0.65 0.3

2.4 0.48 0.56

2.8 0.12 0.50

7 of 13

WILQUET ET AL.: UPPER VENUSIAN HAZE BY SPICAV/SOIR

E00B42

E00B42

Figure 5. Refraction index model (T = 215 K) assuming 75% H2SO4 spherical particles and classical Mie theory [Hummel et al., 1988] in the wavelength range covered by the SPICAV/SOIR instrument. The boxes show the spectral windows of the instrument; dark gray box for the SPICAV-UV channel, gray for the SPICAV-IR channel, and light gray for the SOIR channel. where N(z) is the particle total number density at altitude z and Q(r; l) is the extinction cross section at wavelength l. A lognormal distribution has been used to describe the size distribution of particles of radius r in the Venus upper haze:  2 !   ln r  ln rg 1 f r; rg ; sg ¼ const  r exp  2 ln2 sg

ð3Þ

where rg and sg are the mean radius and variance. [28] The calculation of the extinction cross section Q(r, l) is based on Mie theory, assuming that the upper haze aerosols are composed of 75% H2SO4 aqueous spherical particles, whose refraction index is derived from Hummel et al. [1988] for T = 215 K (see Figure 5). The spectral ranges probed by the three channels of the SPICAV/SOIR instrument are also indicated in Figure 5 and clearly show the complementarity between them. [29] The aerosol extinction coefficients b derived for each of the sampled wavelength ranges (i.e., for one channel) are then normalized to a reference wavelength lr: ben ¼

bð z; lÞ bð z; lr Þ

ð4Þ

This reference is chosen as b (852.7 nm) for SPICAV IR and b (250 nm) for SPICAV UV. In the case of SOIR, the reference value corresponds to the higher wave number probed, since the four spectral intervals in one occultation are not always the same. [30] Finally, an estimation of the particle radius was obtained as described in the paper by Bingen et al. [2003], by fitting the normalized experimental aerosol extinctions ben, obtained from equation (4), to their corresponding theoretical values:   bt rg ; sg ; l ¼

Zþ1

  f r; rg ; sg Qðr; lÞdr

ð5Þ

0

The fit was obtained by minimization of the merit function:  2 4 e   X bn ð z; li Þ  btn rg ; sg ; li M rg ; sg ; z ¼ Dben ð z; li Þ i¼1

ð6Þ

This leads to the determination of the aerosol parameters sg (z) and rg (z), where Db is the 1-sigma error on the retrieved extinction coefficients. Once the radius that gives the best fit at each altitude has been extracted, the

8 of 13

WILQUET ET AL.: UPPER VENUSIAN HAZE BY SPICAV/SOIR

E00B42

E00B42

Figure 6. Vertical profiles of the b extinction coefficient obtained with the three channels of the SPICAV/SOIR instrument for the orbits (a) 485, (b) 486, and (c) 487. The extinction was obtained from data acquired with SPICAV-UV at 220 nm (black circles), SPICAV-UV at 300 nm (gray circles), SPICAVIR at 757 nm (black inverted triangles), SPICAV-IR at 1553 nm (gray inverted triangles), on Figure 6a with SOIR at 2560 nm (gray asterisks) and SOIR at 3978 nm (black asterisks), and on Figure 6b and Figure 6c with SOIR at 2345 nm (gray asterisks) and SOIR at 3682 nm (black asterisks). corresponding model of particle number density N(z) that best fits the data can also be found.

5. Results 5.1. Vertical Profiles of Aerosol Extinctions [31] The extinction coefficients b retrieved as described in section 4.1 are plotted in Figure 6 for the selected orbits, when all three channels of the instrument were measuring simultaneously. For clarity, b for only two wavelengths per channel of the SPICAV/SOIR instrument are shown. [32] The general trend is a decrease of the extinction with increasing altitude except when aerosols layers are observed. This phenomenon was more pronounced at longer wavelengths (SOIR channel) than at shorter wavelengths (SPICAV-UV). It could indicate the presence of particles with a size larger than the wavelengths of the UV channel (>300 nm), which is less sensitive to such particles. In general, less temporal variability of the b coefficient was observed in the UV than in the IR. [33] These aerosol layers have been observed in several occultations, mainly during summer 2007, showing temporal and/or spatial variations in terms of the altitude of the layer (Figure 7). Figure 7 illustrates the apparition and the displacement of such layers observed with the SOIR channel in the upper haze during orbits 482 and 484 to 487. While the layers were almost not visible in the UV, they

were also detected with the SPICAV-IR channel (Figure 6), yet a small difference in the altitude of the layer is observed between the SPICAV-IR and the SOIR channels. This can be explained by a weak accuracy of the calibration of the FOV of the SPICAV-IR channel. For orbit 482, the extinction coefficient, for the four spectral windows investigated during the solar occultation, increased fairly smoothly with decreasing altitude. In addition, the extinction showed little spectral dependency; that is, the extinction is very similar for the four diffraction orders. In contrast, for orbits 484 to 487, one or two layers appeared in the extinction profiles in which extinction decreased with decreasing altitude before it increased further (Figure 7). The altitude of the layers varied from orbit to orbit, they were located at 82 km and 74 km for orbit 484, 78 km and maybe 74 km for orbit 485 and, at 84 km and 78 km for orbit 486 and at 83 km and maybe 72 km for orbit 487. Moreover, the spectral dependency of the extinction was greater for these occultations than for the one during orbit 482 and was strikingly higher within the layers than at other altitudes (Figure 7). For example, in orbit 486, at an altitude of 78.5 km, there is about 1 order of magnitude for b between diffraction order 171 (4  105 km1) and order 121(3  104 km1). We are currently retrieving temperature, density and pressure from these peculiar orbits, as a layer was not observed in all occultations, to see if there is any correlation with the observed variations of the layer.

9 of 13

E00B42

WILQUET ET AL.: UPPER VENUSIAN HAZE BY SPICAV/SOIR

E00B42

Figure 7. Temporal variation of a layer in the extinction profiles observed with the SOIR channel. Extinction coefficients in function of the altitude for orbit 482 (brown), diffraction orders 111 (circles), 121 (triangles), 137 (squares), 171 (inverted triangles); orbit 484 (blue), diffraction orders 125 (circles), 149 (triangles), 172 (squares), 180 (inverted triangles); orbit 485 (green), diffraction orders 112 (circles), 119 (triangles), 149 (squares), 174 (inverted triangles); orbit 486 (gray), diffraction orders 121 (circles), 149 (triangles), 171 (squares), 190 (inverted triangles); and orbit 487 (orange), diffraction orders 121 (circles), 149 (triangles), 171 (squares), 190 (inverted triangles).

[34] Detached layers of aerosols were already observed in the past above the Venus clouds [Lane and Opstbaum, 1983]. It is preliminary here to speculate on the mechanisms of formation of such layers but gravity waves are known to create stratified layers in the atmosphere. Recently, internal gravity waves have been identified in the lower and upper clouds of Venus [Peralta et al., 2008]. 5.2. Retrieval of Particle Size [35] The use of the three channels allowed us to investigate a broad range of wavelengths in the UV and the IR and to make use of the spectral dependence of the extinction (Figure 5) to infer aerosol size(s). It is worth mentioning that the simultaneous use of the three channels in the solar occultation mode was not possible at the beginning of the VEX mission but through a programming adaptation, it has been in operation for about 1 year. This means that for the earlier measurements, aerosol characterization has to be performed from data acquired by two channels (SOIR and SPICAV-IR). We therefore make first an attempt to derive independently particle size(s) and number densities of the Venusian upper haze from the extinction coefficients obtained for each of the three channels of the instrument. This analysis is illustrated using a number of orbits with coordinated solar occultations and using the method described in section 4.3, i.e., by fitting the aerosol extinction

curve with a Mie theory model (equation (5)). A lognormal distribution (equation (3)) was assumed where sg was variable for SPICAV-UV and fixed to 0.1 for SPICAV-IR and SOIR (similar particle sizes were found for a fixed variance of 0.2 or of 0.3, data not shown), and the refractive indices of H2SO4 compiled by Hummel et al. [1988] was used. [36] Figure 8 shows the vertical size distribution of the particles for solar occultations performed during orbits 485, 486, and 487. The channels cannot discriminate particles with a radius higher than the wavelength at which the data were acquired, i.e., for these particular orbits, a maximal radius of 0.3 mm for SPICAV-UV, of 1.6 mm for SPICAV-IR and of 3.7 mm for SOIR. [37] The data obtained with the UV channel indicate the presence of particles with a radius increasing from 0.1 mm at 100 km up to 0.3 mm at 75 km (Figure 8). The vertical distribution for this channel exhibits low variability from one orbit to another. [38] From the SOIR and SPICAV-IR channels, the vertical distribution of particle size is limited to altitudes comprised between 70 km and 90 – 95 km. The radii determined with both channels are in fairly good agreement and are comprised between 0.4 and 1.0 mm. Particles of 1 mm were detected at 74– 77 km of altitude during all orbits. For each orbit analyzed here, the maxima in the

10 of 13

E00B42

WILQUET ET AL.: UPPER VENUSIAN HAZE BY SPICAV/SOIR

E00B42

Figure 8. Vertical distributions of particle size for orbits (a) 485, (b) 486, and (c) 487. Radii derived from normalized extinction values obtained at 2 wavelengths with the SPICAV-UV channel (circles), at 10 wavelengths with the SPICAV-IR channel (inverted triangles), and at 4 wavelengths with the SOIR channel (asterisks). vertical particle size distribution (Figure 8) correlate with the maxima of the extinction vertical profiles (Figure 6). The aerosol layers also exhibited temporal and/or spatial variations in terms of the size of the particles within the layer, from 0.4 up to 1.0 mm radius. [39] The particle sizes found with the UV channel in this study are in good agreement with values obtained previously. Sato et al. [1996] compiled Venus cloud and haze properties deduced from Pioneer Venus, Venera, and Vega mission results. Sato et al. [1996] reported haze particles in the north polar region of rg = 0.25 mm ± 0.05 mm and sg = 0.25 mm ± 0.05, while Esposito [1983] identified particles of rg = 0.2 mm and Kawabata et al. [1980] estimated haze particles of rg = 0.23 mm ± 0.04 mm and sg = 0.18 mm ± 0.1. They all refer to an altitude range comprised between 70 and 90 km, called the upper haze. However, our measurements tend to suggest that the two modes found in the upper clouds (properties summarized by Esposito [1983]), usually defined as mode 1 (rg = 0.2 mm) and mode 2 (rg = 1.0 mm), are still both present at altitudes higher than 70.0 km. Earlier works reported that modes 1 and 2 only coexist in the upper cloud region (55 – 70 km) or below and that only mode 1 is found above. Whether the merging of these two modes account for only one global

mode of particles or two separate distributions cannot be firmly established on the basis of this analysis alone. 5.3. Total Particle Number Density [40] Figure 9 shows the particle number densities derived in the UV spectral range for aerosol particles of about 0.1– 0.3 mm size from the SPICAV-UV channel and in the IR for aerosol particles of about 0.4– 1.0 mm size from the SPICAV-IR and the SOIR channels depending on the altitude. SPICAV-UV measurements lead to mode 1 particle number density between 10 and 30 cm3 below 90 km, decreasing at higher altitudes. These values are 1 order of magnitude lower than those reported by Esposito [1983] but more consistent with the measurements from Pioneer Venus limb scans [Lane and Opstbaum, 1983] at northern midlatitude, close to the orbit latitudes analyzed in this paper (Table 1). [41] Number density for the larger particles (mode 2) decreases smoothly from 10 to 15 cm3 at 70 km down to less than 1 cm3 at 90 km. It is important to mention that the results obtained for the two IR channels are very comparable although they were calculated from independent data sets, yet from simultaneous measurements. The values found at about 70 km, which is the boundary between the upper cloud and the upper haze, is close to, yet smaller than the one reported by Esposito [1983] for

11 of 13

WILQUET ET AL.: UPPER VENUSIAN HAZE BY SPICAV/SOIR

E00B42

E00B42

Figure 9. Vertical profiles of the particle total number density in the Venus upper haze during two successive orbits of Venus Express. The radius of particles found by fitting normalized extinction to a haze model allowed derive N for mode 1 particles with the SPICAV-UV channel (circles), N for mode 2 particles with the SPICAV-IR channel (inverted triangles), and N for mode 2 particles with the SOIR channel (asterisks) during orbit 485 (black) and orbit 486 (gray). mode 2 particles in the upper cloud region, i.e., 50 cm3. Nevertheless, the values found in this restricted analysis (only three orbits) should be taken with caution and are indicative of an order of magnitude for the number density for mode 2 particles. A systematic multichannel retrieval is currently ongoing. The combination of extinction coefficients obtained from UV up to mid-IR and the use of a bimodal particle size distribution will probably allow retrieving unique aerosols parameters and solving the discrepancies between nadir [Esposito, 1983] and limb viewing [Lane and Opstbaum, 1983].

6. Conclusion [42] The number of orbits analyzed was restricted for this paper, aimed at proving the capability of the SPICAV/SOIR instrument on board Venus Express to characterize the upper haze of Venus’ atmosphere. The ability of the instrument since March 2007 to perform simultaneous solar occultations with the three channels will allow us to investigate the upper haze in greater detail in the future and to plan solar occultations specifically for this purpose. This could be achieved by making use of the full spectral range of the instrument, from 170 nm up to 4 mm, by choosing appropriate, regularly spread wavelengths for each channel, and by taking maximum advantage of the spectral dependency of the extinction studied in this paper. [43] One of the main advantages of the solar occultation technique is the high vertical resolution and in comparison

with nadir viewing, a much longer optical path resulting in an improved sensitivity. In addition, data collected at terminators could lead to a better understanding of the known morning/evening asymmetry of the upper haze [Esposito, 1983]. [44] The determination of the particles sizes performed for this paper was made independently for the different channels of the SPICAV/SOIR instrument. The first results look promising for further analysis of the aerosols in the mesosphere of Venus as, in the future, the radii of particles could be retrieved in one single minimization of the merit function (see section 4.3) from the extinction coefficients obtained with the three channels. This study demonstrated the existence of at least two types of particles, one type with a radius comprised between 0.1 and 0.3 mm as inferred by the UV channel and the second type, detected in the IR, with a radius varying between 0.4 and 1 mm depending on the altitude. Therefore, the model describing the upper haze on Venus should include a bimodal population for putting together the three channels in retrieving particle sizes from the extinction profiles. To our knowledge, this is the first time that the existence of mode 2 particles at altitudes above 70 km in the atmosphere of Venus is shown. It was found independently for the two IR channels of the SPICAV/SOIR instrument, giving strong evidence for its existence. [45] Our measurements also showed the high temporal variability of the aerosol loading in the Venus mesosphere. The SPICAV/SOIR is well suited to study the vertical distribution of the different types of aerosols, as well as such temporal and latitudinal variability.

12 of 13

E00B42

WILQUET ET AL.: UPPER VENUSIAN HAZE BY SPICAV/SOIR

[46] Acknowledgments. The research program was supported by the Belgian Federal Science Policy Office and the European Space Agency (ESA-PRODEX program contracts C 90268, 90113, and 17645). The Russian team acknowledges RFBR grant 06-02-72563.

References Bertaux, J. L., D. Nevejans, O. Korablev, E. Villard, E. Que´merais, E. Neefs, F. Montmessin, F. Leblanc, J. P. Dubois, and E. Dimarellis (2007), SPICAV on Venus Express: Three spectrometers to study the global structure and composition of the Venus atmosphere, Planet. Space Sci., 55(12), 1673 – 1700, doi:10.1016/j.pss.2007.01.016. Bertaux, J.-L., A. C. Vandaele, V. Wilquet, F. Montmessin, R. Dahoo, E. Villard, O. Korablev, and A. Fedorova (2008), First observation of 628 CO2 isotopologue band at 3.3 mm in the atmosphere of Venus by solar occultation from Venus Express, Icarus, 195(1), 28 – 33, doi:10.1016/ j.icarus.2008.01.001. Bingen, C., F. Vanhellemont, and D. Fussen (2003), A new regularized inversion method for the retrieval of stratospheric aerosol size distributions apllied to 16 years of SAGE II data (1984 – 2000): Method, results and validation, Ann. Geophys., 21, 797 – 804. Esposito, L. W. (1983), The clouds and hazes on Venus, in Venus, edited by D. M. Hunten et al., pp. 484 – 564, Univ. of Ariz., Tucson. Fedorova, A., et al. (2008a), HDO and H2O vertical distribution and isotopic ratio in the Venus mesosphere by Solar Occultation at Infrared spectrometer on board Venus Express, J. Geophys. Res., 113, E00B22, doi:10.1029/2008JE003146. Fedorova, A. A., O. I. Korablev, J.-L. Bertaux, A. V. Rodin, F. Montmessin, D. A. Belyaev, and A. Reberac (2008b), Solar infrared occultations by SPICAM experiment on Mars-Express: Simultaneous measurement of the vertical distributions of H2O, CO2 and aerosol, Icarus, 200, 96 – 117. Hummel, J. R., E. P. Shettle, and D. R. Longtin (1988), A new background stratospheric aerosol model for use in atmospheric radiation models, AFGL-TR-88 – 0166, Air Force Geophys. Lab., Hanscom Air Force Base, Mass. Kawabata, K., D. L. Coffeen, J. E. Hansen, W. A. Lane, M. Sato, and L. D. Travis (1980), Cloud and haze properties from Pioneer Venus polarimetry, J. Geophys. Res., 85(A13), 8129 – 8140, doi:10.1029/JA085iA13p08129. Korablev, O., et al. (2006), SPICAM IR acousto-optic spectrometer experiment on Mars Express, J. Geophys. Res., 111, E09S03, doi:10.1029/ 2006JE002696. Lane, W. A., and R. Opstbaum (1983), High altitude Venus haze from Pioneer Venus limb scans, Icarus, 54, 48 – 58, doi:10.1016/00191035(83)90071-4. Mahieux, A., et al. (2008), In-flight performance and calibration of SPICAV SOIR on board Venus Express, Appl. Opt., 47(13), 2252 – 2265, doi:10.1364/AO.47.002252.

E00B42

Montmessin, F., et al. (2006a), Sub-visible CO2 ice clouds detected in the mesosphere of Mars, Icarus, 183, 403 – 410, doi:10.1016/j.icarus. 2006.03.015. Montmessin, F., E. Que´merais, J. L. Bertaux, O. Korablev, P. Rannou, and S. Lebonnois (2006b), Stellar occultations at UV wavelengths by the SPICAM instrument: Retrieval and analysis of Martian haze profiles, J. Geophys. Res., 111, E09S09, doi:10.1029/2005JE002662. Nevejans, D., et al. (2006), Compact high-resolution space-borne echelle grating spectrometer with AOTF based on order sorting for the infrared domain from 2.2 to 4.3 micrometer, Appl. Opt., 45(21), 5191 – 5206, doi:10.1364/AO.45.005191. Peralta, J., R. Hueso, A. Sa´nchez-Lavega, S. Pe´rez-Hoyos, G. Piccioni, O. Lanciano, P. Drossart, and t.V.-V.E. team (2008), Characterization of gravity waves in the upper and lower clouds of Venus using VIRTISVEX images, Geophys. Res. Abstr., 3, Abstract EPSC2008-A-00095. Que´merais, E., J.-L. Bertaux, O. Korablev, E. Dimarellis, C. Cot, B. R. Sandel, and D. Fussen (2006), Stellar occultations observed by SPICAM on Mars Express, J. Geophys. Res., 111, E09S04, doi:10.1029/ 2005JE002604. Sato, M., L. D. Travis, and K. Kawabata (1996), Photopolarimetry analysis of the Venus atmosphere in polar regions, Icarus, 124, 569 – 585, doi:10.1006/icar.1996.0231. Svedhem, H., et al. (2007), Venus Express—The first European mission to Venus, Planet. Space Sci., 55(12), 1636 – 1652, doi:10.1016/j.pss.2007. 01.013. Titov, D. V., et al. (2006), Venus Express science planning, Planet. Space Sci., 54(13 – 14), 1279 – 1297, doi:10.1016/j.pss.2006.04.017. Vandaele, A. C., et al. (2008), Composition of the Venus mesosphere measured by SOIR on board Venus Express, J. Geophys. Res., 113, E00B23, doi:10.1029/2008JE003140. Wilquet, V., A. Mahieux, A. C. Vandaele, V. I. Perevalov, S. A. Tashkun, A. Fedorova, O. Korablev, F. Montmessin, R. Dahoo, and J.-L. Bertaux (2008), Line parameters for the 01111 – 00001 band of 12C16O18O from SOIR measurements of the Venus atmosphere, J. Quant. Spectrosc. Radiat. Transfer, 109, 895 – 905, doi:10.1016/j.jqsrt.2007.12.021. 

J.-L. Bertaux, F. Montmessin, and E. Villard, Service d’Ae´ronomie du CNRS, BP3, F-91371, Verrie`res-le-Buisson, France. R. Drummond, A. Mahieux, A. C. Vandaele, and V. Wilquet, Belgian Institute for Space Aeronomy, 3 Avenue Circulaire, B-1180 Brussels, Belgium. ([email protected]) A. Fedorova and O. Korablev, Space Research Institute, 84/32 Profsoyuznaya Street, 117997, Moscow, Russia.

13 of 13

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, E00B04, doi:10.1029/2008JE003092, 2008

for

Full Article

Venus Express observations of atmospheric oxygen escape during the passage of several coronal mass ejections J. G. Luhmann,1 A. Fedorov,2 S. Barabash,3 E. Carlsson,3 Y. Futaana,3 T. L. Zhang,4 C. T. Russell,5 J. G. Lyon,6 S. A. Ledvina,1 and D. A. Brain1 Received 29 January 2008; revised 21 April 2008; accepted 27 May 2008; published 26 August 2008.

[1] The solar wind interaction contributes to the loss of Venus atmospheric constituents,

especially oxygen, by direct antisunward acceleration of planetary ions and possibly by related sputtering of neutrals. Both comet-like ‘‘ion pickup’’ and related sputtering processes may have played a key role in Venus’ atmosphere evolution, but the significance of their effects, as well as other proposed escape mechanisms, is still uncertain. In particular, recent reports of only modest ion escape rates measured on Venus Express (VEX) during the current low-activity phase of the solar cycle make it important to reconsider the evidence seen in both Pioneer Venus Orbiter and VEX observations suggesting significantly enhanced escaping O+ ion fluxes during periods of disturbed interplanetary conditions. At present, the most extreme interplanetary conditions result from the effects of coronal mass ejections (CME), which may have been more prevalent in the first 1–2 billion years of the Sun’s history. The Analyser of Space Plasmas and Energetic Atoms 4 (ASPERA-4) Ion Mass Analyzer and Magnetometer on Venus Express have now made detailed measurements during several periods when CME disturbances encountered Venus. The observations and models described in this report provide further insights into the possible response of oxygen ion escape to solar activity. In particular, they illustrate nuances of in situ sampling of large-gyroradius pickup ions, related to the atmospheric source properties, spacecraft orbit geometry, and the prevailing interplanetary conditions, that make the estimation of the variable global escape fluxes due to that process particularly challenging. In three of the four cases examined in some detail, ionospheric oxygen ions were either unobservable or below the limit of detectability for passes well within interplanetary coronal mass ejection (ICME) intervals. In the fourth example, where ionospheric ions were observed as expected from the model, O+ pickup ions were observed in greater abundance than is typical in undisturbed solar wind. Other escape processes are not considered here, although it is assumed the source population for the modeled pickup ions is a preaccelerated upper atmosphere component. Analysis of many more ICME events, expected to be obtained as the Sun becomes more active in future years, are necessary to resolve the question of the importance of ICME to Venus oxygen escape. Citation: Luhmann, J. G., A. Fedorov, S. Barabash, E. Carlsson, Y. Futaana, T. L. Zhang, C. T. Russell, J. G. Lyon, S. A. Ledvina, and D. A. Brain (2008), Venus Express observations of atmospheric oxygen escape during the passage of several coronal mass ejections, J. Geophys. Res., 113, E00B04, doi:10.1029/2008JE003092.

1. Introduction 1

Space Sciences Laboratory, University of California, Berkeley, California, USA. 2 Centre d’Etude Spatiale des Rayonnements, Toulouse, France. 3 Swedish Institute of Space Physics, Kiruna, Sweden. 4 Space Research Institute, Austrian Academy of Sciences, Graz, Austria. 5 Institute of Geophysics and Planetary Physics, University of California, Los Angeles, California, USA. 6 Department of Physics and Astronomy, Dartmouth College, Hanover, New Hampshire, USA. Copyright 2008 by the American Geophysical Union. 0148-0227/08/2008JE003092$09.00

[2] One of the key planetary science issues for Venus (and Mars) is the evolutionary importance of the solar wind interaction [e.g., Chassefiere, 1997; Lammer et al., 2006]. Our current picture of planetary atmospheric ion scavenging by direct interaction of the solar wind with weakly magnetized planets was developed in large part from observations obtained by the Pioneer Venus Orbiter, PVO [e.g., Moore et al., 1990; Brace et al., 1995], although the Venera mission first detected Venus’ planetary ion wake [e.g., Vaisberg et al., 1995]. The observations confirmed the presence of a comet-like interaction where the convection electric field of the solar wind, E = VXB (where V and B are the bulk

E00B04

1 of 15

E00B04

LUHMANN ET AL.: VEX—OXYGEN ESCAPE AND CME

velocity of the solar wind and the interplanetary magnetic field vector, respectively) accelerates or ‘‘picks up’’ planetary ions near and above the exobase (200 km altitude).The ions that become pickup ions may be either transported into the acceleration region by other forces such as those associated with pressure gradients, or they may be produced in situ from the local neutral gases. If they are accelerated to at least the 11 km/s escape velocity for Venus, and are on outward trajectories that do not reimpact the exobase [e.g., Luhmann and Kozyra, 1991], they can be permanently lost. Other mechanisms for Venus ion escape have also been suggested that generally fall under the headings of ‘ionospheric ion outflow’ [e.g., Barabash et al., 2007a] or ‘bulk ionospheric escape’ [e.g., Terada et al., 2002]. However, many of the PVO observations at both high ( keV) and low (10s of eV) ion energies can be modeled by assuming that the pickup process dominates [Luhmann et al., 2006, 2007] we focus on that process in the present analysis. [3] It was also inferred from PVO measurements of both the upper atmosphere (observed by UV remote sensing and by the Orbiter Neutral Mass Spectrometer), the ionosphere (observed by the retarding potential analyzer, thermal ion mass spectrometer, and Langmuir Probe), and the ion fluxes moving antisunward in the nightside wake (observed by the Plasma Analyzer and Neutral Mass Spectrometer in its ion mode), that singly ionized atomic and molecular oxygen ions are the primary species involved in the solar windinduced escape of the Venus upper atmosphere [e.g., Brace et al., 1987, 1995; Mihalov et al., 1995]. This is a consequence of the ion chemistry of a CO2 atmosphere, where oxygen forms the bulk of the upper atmosphere and ionosphere [e.g., Fox and Kliore, 1997]. [4] Typical escape rates of oxygen ions from Venus today are low from an atmospheric evolution standpoint, little more than the equivalent of a weak cometary ion tail at approximately 1024 1026 O+ s 1 [e.g., Moore et al., 1991; Barabash et al., 2007a]. However, the Venus solar wind interaction represents one of the only ways to accelerate oxygen and other heavy atmospheric elements to escape speeds. The need for an effective loss mechanism for Venus oxygen has lately become a matter of special importance as arguments mount for a possible early ocean [e.g., Grinspoon and Bullock, 2007; Kulikov et al., 2006]. One key question is whether this solar wind-enabled escape mechanism could have been sufficient, or whether instead, important quantities are sequestered and buried in oxidized regolith. This question bears on the expectations for future surface and subsurface mineralogy measurements at Venus in pursuit of a history of water on the planet. The broader consequences relate to general unmagnetized and weakly magnetized planetary bodies and satellites with atmospheres that interact directly with surrounding magnetized external plasma flows of any origin. [5] Luhmann et al. [2006, 2007] revisited the PVO escaping O+ ion observations to investigate the possible impact of disturbed solar wind plasma and interplanetary magnetic field conditions on escaping ion fluxes. The observations used in that study consisted of > 36 eV O+ ion fluxes (10 eV is required for O+ escape from Venus) measured by the Orbiter Neutral Mass Spectrometer in its ion detection mode. The results suggested that enhanced

E00B04

incident solar wind dynamic pressures and interplanetary magnetic fields, such as those that occur during passage of interplanetary coronal mass ejections (ICME) and solar wind stream interaction regions are sometimes followed by detection of enhanced (by 100) escaping O+ fluxes. What could not be determined from PVO measurements owing to the limited instrument capabilities, and orbital and temporal sampling coverage, is both the full range of the escape fluxes and the definitive escaping ion composition. In particular, evaluation of the escape flux requires knowledge of the details of its spatial and temporal variations, which are difficult to obtain with in situ measurements, while the composition measurements require ion mass spectrometry over an energy range from 10 eV to > several 10s of keV. [6] While several statistical analyses carried out on the PVO data revealed general patterns of the inferred O+ escape flux spatial distribution, and average values of the fluxes along the spacecraft orbit [e.g., Mihalov and Barnes, 1982; Moore et al., 1990; Luhmann et al., 2006], it is not possible to reconstruct from these the global response to relatively rare and/or short-lived extremes in the solar and interplanetary conditions. Observation-validated models that allow the interpretation of global loss rates from local measurements provide the best alternative. In addition, the ASPERA-4 Ion Mass Analyzer, IMA, on Venus ExpressVEX [Barabash et al., 2007b] now routinely measures escaping O+ fluxes with definitive mass identification in an energy range (0.01 – 30. keV) exceeding that available on PVO (e.g., see the discussion by Luhmann et al. [2006]). Since its orbit insertion in April 2006, Venus Express has witnessed a number of significant solar events [e.g., Futaana et al., 2007], providing an excellent opportunity to investigate, with the supporting VEX magnetometer observations [Zhang et al., 2006], the response of the escaping ions. [7] The motivation for the study described here is to further constrain present rates of oxygen ion escape during ICME passages, toward evaluating the evolutionary importance of these conditions for Venus. VEX ASPERA-4 IMA and magnetometer observations are used, together with an MHD field-based test particle model of O+ pickup ions, to analyze the escaping ion response during several events. The results illustrate that making the case for increased loss rates at these times depends critically on our ability to use models to interpret the observations. However, the observations reported here are independent of the assumption of any particular process and are of interest in themselves. Truly definitive results on the ion escape response will depend on the analysis of data obtained during a large number of different ICME as well as case studies. This will require measurements over at least one solar maximum period, the next of which will be underway in 2010 – 2012, together with models that take into account whatever processes are found to be significant.

2. ICME Observations on VEX [8] Details of the VEX instruments used in this study can be found in papers by Barabash et al. [2007b] and Zhang et al. [2006]. The magnetometer is a dual triaxial fluxgate magnetometer where one triad is used to correct the

2 of 15

E00B04

LUHMANN ET AL.: VEX—OXYGEN ESCAPE AND CME

E00B04

Figure 1. A set of four single orbit VEX magnetic field time series showing the ICME cases used here: (a) 5 July 2006, (b) 18 July 2006, (c) 11 September 2006, and (d) 14 February 2007. Near the beginning of each time series Venus perturbation is seen, including the bow shock, magnetosheath, dayside magnetic barrier, and magnetotail as sampled by VEX. The ICME field components exhibit an initial interplanetary shock and field pileup in the ICME sheath, followed by the smoothly rotating, often enhanced, solar ejecta fields sometimes lasting a day or more, although the start of the ICME is not always visible in these plots. measurements for the spacecraft-generated magnetic fields [Zhang et al., 2006]. The ASPERA-4 instrument is a close copy of the ASPERA-3 instrument on Mars Express [Barabash et al., 2006], and includes both the ion mass spectrometer, called the IMA for ion mass analyzer, and an electron spectrometer, although here we concentrate on the data from the former. The mass range of the IMA is from 1 to 60 amu and the energy range is from 10 eV to 30 keV. The angular response of the IMA, which includes a deflection system, allows detection of ions coming from all azimuthal directions with respect to the Venus-Sun line, and from elevation angles of ±45°. Although the IMA is also a capable solar wind ion analyzer, ASPERA-4 operates only around periapsis to conserve instrument resources and to stay within its telemetry allotment (although there are occasional special full-orbit operations). Additional details of the IMA characteristics and operations are given by Fedorov et al. [2006] and Barabash et al. [2007b].

[9] VEX orbits Venus in a high-inclination elliptical orbit with 250 – 300 km periapsis close to the north pole [Svedhem et al., 2007]. While the plasma analyzer operates only around periapsis, the magnetometer makes continuous measurements. To identify occurrences of ICME events in the VEX data we visually scanned magnetometer vector time series from 2006 and 2007 for the typical field signatures of ICME. These often include a leading shock jump and sheath (compressed solar wind) region, followed by larger than average, low-variance, smoothly rotating ejecta or driver fields. Typical duration of the entire disturbance is 1 – 2 days. Examples found in the PVO data at 0.73 AU are described in earlier papers by Lindsay et al. [1995], Mulligan et al. [1998] and in a recent paper by Jian et al. [2008]. The same signatures are used to identify ICME near the Earth [e.g., Jian et al., 2006]. [10] Most unambiguous ICME events found in the VEX data occurred in 2006, when the Sun was more active. From

3 of 15

E00B04

LUHMANN ET AL.: VEX—OXYGEN ESCAPE AND CME

E00B04

Figure 2a. Case 1: 5 July 2006 ICME ASPERA-4 IMA observations, together with the preceding and following orbits (days). The top panels show a 2-h orbit segment for these data, near periapsis. The top energy-time spectrograms include ions with masses 12– 60 amu, while the bottom spectrograms include all ions detected. The latter is dominated by solar wind H+. The bottom panels show the magnetic field magnitude from the magnetometer for context. these, three cases were selected for detailed analysis because they exhibit the smooth, slowly rotating ejecta fields useful for analyzing and modeling the effects of the ICME conditions on pickup ions. A single case from 2007 was also included. In addition to the large, smooth ejecta fields, high dynamic and thermal pressures are found in the preceding postshock sheath of an ICME due to the ejecta’s high velocity, solar wind density compression ahead of the fast-moving ejecta, and shock heating of the sheath solar wind plasma. However, these sheath periods last only a

fraction of a day and so it is difficult to capture the Venus interaction at the VEX periapsis during the times of their passage. However, the high bulk speeds, and related highconvection electric fields and enhanced solar wind pressures, can persist throughout the passage of the ejecta, although they have typical solar wind densities. These attributes are not easily distinguished in the perturbed planetary interaction environment that ASPERA-4 samples near the VEX periapses, but their presence can still be inferred from the intervals of IMA data obtained outside the

Figure 2b. Case 2: Same as Figure 2a but for the 18 July 2006 ICME. 4 of 15

E00B04

LUHMANN ET AL.: VEX—OXYGEN ESCAPE AND CME

E00B04

Figure 2c. Case 3: Same as Figure 2b but for the 11 September 2006 ICME. (In this case the pass in the first panel is a day earlier because ASPERA-4 data were not available the day before this ICME). bow shocks. In the cases of the four ICME events analyzed here, the plasma velocities outside of the Venus bow shock were not especially high, an attribute not surprising for events occurring near solar minimum. Thus the main feature of these particular ICME is their steadier and larger than normal (by 2x) interplanetary magnetic fields lasting a day or more, plus the expected (but not observed) shorter period of enhanced dynamic pressure from the preceding sheath higher densities. [11] What are the possible consequences of ICME passage expected for Venus? In addition to a reduced ionopause altitude and ionospheric magnetization in response to the increased solar wind pressures in the ICME sheath [e.g.,

Luhmann and Cravens, 1991], ICME conditions are known to produce the largest convection electric fields and interplanetary field magnitudes detected in the interplanetary medium at 0.73 AU [e.g., Lindsay et al., 1995]. Thus any processes affected by the convection electric field or ionospheric magnetization should be enhanced. In addition, the interplanetary shock-heated solar wind is expected to result in higher than usual Venus sheath electron temperatures and related impact ionization of any neutrals exposed to them, while the increased solar wind plasma densities in the ICME sheath can increase the rate of charge exchange (creating both additional ionization of the neutral O and energetic neutral hydrogen atoms, ENAs). The slowly varying, un-

Figure 2d. Case 4: Same as Figure 2c but for the 14 February 2007 ICME. 5 of 15

E00B04

LUHMANN ET AL.: VEX—OXYGEN ESCAPE AND CME

E00B04

Figure 3a. ASPERA-4 IMA energy-mass diagrams for ICME cases 1 and 2. These show the ion composition for the 15 min around periapsis for the three passes (before, during, and after the ICME passage, from left to right), for (top) case 1 (Figure 2a) and (bottom) case 2 (Figure 2b). The heavy ions can be seen in the right part of the diagrams for the passes on either side of the central passes in Figures 2a and 2b. They consist mainly of O+ and O2+, but heavier ions are also present. usually steady magnetic fields of the ejecta meanwhile provide exceptionally simple conditions for analyzing and interpreting the behavior of observed ions in terms of the pickup process. [12] Time series of VEX magnetometer vector components in the VSO system (x directed toward the Sun, y opposite the direction of planetary motion, and z northward from the Venus orbit plane) are shown in Figures 1a– 1d for the days containing the selected passes. In all of these cases the ICME is already in progress when the day begins. The ICME are easily distinguished here by the previously mentioned low variance, slowly rotating ejecta fields, and by their field magnitudes, especially in the third example. For the first case shown, from 5 July 2006, the conditions surrounding the Venus interaction are dominated by highinclination, northward (+Bz, clock angle 0) ejecta fields, while the 18 July 2006 case is dominated by highinclination, southward ( Bz, clock angle 180) ejecta fields. The 11 September 2006 case has By and Bz (clock angle 225) dominated fields, while the 14 February 2007 case has +By and +Bz (clock angle 45) fields. Thus the selected cases provide a variety of external magnetic field orientations during ICME passages. [13] ASPERA-4 IMA energy-time spectrograms for the four selected passes within the ICME events are shown in Figures 2a– 2d, together with the total magnetic field and orbit near periapsis in the cylindrical view (where the y and

z positions are converted into a cylindrical coordinate projected into a plane perpendicular to the Venus-Sun (x) axis). As controls, we also show the spectrograms for the adjacent periapsis passes the day before and the day after the main event passage. (The VEX orbital period is approximately an Earth day, also the approximate duration of a typical ICME.) The orbit segments at the top of each pass illustrate the spacecraft sampling in local time, solar zenith angle, and radius. The spectrograms are shown for both all detected ions (0– 60 amu) and heavy (>12 amu) ions only. The total ion spectrograms are included to provide the solar wind plasma context of the measured heavy planetary ions. Note that the light ions may also include some picked-up planetary H+, H+2 and He+ ions [e.g., Barabash et al., 2007a], but we exclude these from consideration in our present analysis focused on oxygen. The total ion spectrograms are dominated by the standard solar wind interaction features including proton heating at the bow shock (seen as broadening of the main solar wind proton band in the energy-time spectrogram), the compression and deceleration in front of the ionospheric obstacle, and the reduction or disappearance of solar wind proton densities in the wake. The periodic dropouts in the detected fluxes are a result of the directional sampling cycle of ASPERA-4 and the anisotropy of the detected ions, most of which flow approximately antisunward. Near-periapsis magnetic field magnitude time series are plotted at the bottom to provide

6 of 15

E00B04

LUHMANN ET AL.: VEX—OXYGEN ESCAPE AND CME

E00B04

Figure 3b. Same as Figure 3a but for ICME cases 3 and 4 in Figures 2c and 2d. For case 3 (top row), the pre-ICME pass is two days ahead. For case 4 (bottom row) it is a day ahead. Case 4 is the only one of the four with high (saturated counts) oxygen ion content in the central diagram, during the main portion of the ICME passage. a perspective on the location of the ion features relative to the magnetic pileup signatures in the sheath. [14] The heavy ion spectrograms in the second panels from the top of each example in Figures 2a, 2b, 2c, and 2d are dominated by atomic oxygen, but may also contain contributions from O+2 , CO+, and CO+2 [Carlsson et al., 2006]. As shown by the IMA mass-energy plots of Barabash et al. [2007b], these also include ‘shadows’ of the solar wind due to the high flux of protons and the resulting overlap in mass-energy space. The species separations used for the heavy ion spectrograms are obtained with Gaussian distribution fits to the overlapping mass peaks in the mass-energy plots [e.g., Carlsson et al., 2006]. These spectrograms exhibit several distinctive features including a period dominated by lo—energy ions within the 10– 20 min interval around periapsis (roughly the plot center), and a more energetic heavy ion signature with a highly variable appearance. The low-energy ions are typically in the 10eV to a few 10s of eV energy range, and are thus already at escape energies for oxygen. The latter may appear as a faint high-energy component during the same time period that the low-energy ions are observed, or it may be observed outside the periapsis interval in the adjacent magnetosheath, together with the solar wind protons. The magnetosheath cases closely resemble what was observed in the PVO plasma analyzer data in similar timeenergy spectrograms without mass separation [e.g., Luhmann et al., 2006]. In a few passes the energies of the

heavy ions around periapsis appear to dip below the IMA energy threshold. A possible relationship between the lowenergy and high-energy components will become apparent in the discussion of the modeling results below. [15] The key observational result derived from these events is best seen with the help of mass-energy analyses of the data in Figures 2a– 2d. Figures 3a and 3b show massenergy plots obtained from the 15 min around periapsis for the four event passes and for passes before and after the main event pass. The color scale indicates the detector counting statistics for these samples, which are approximately proportional to the flux. It is seen that three of the four event passes show a near or total absence of counts at the heavy ion masses compared to the passes during relatively typical interplanetary conditions. The one exception out of the four, on 14 February 2007, in contrast shows an exceptionally high (saturated) count rate for the heavy ions in the same near-periapsis sample. The possible reasons for the nondetection of the mainly oxygen ions at these times for the other cases, both at periapsis and in the adjacent sheath, are best investigated with models of the global distribution of escaping ions.

3. Model Analysis of the VEX Ion Observations [16] As noted in the introduction, we here assume that the pickup process describes the bulk of escaping heavy ions observed by the ASPERA IMA around Venus. This as-

7 of 15

E00B04

LUHMANN ET AL.: VEX—OXYGEN ESCAPE AND CME

sumption provides a starting point for analysis that has observational support from PVO results, and can also lead to insights on whether other processes are important. [17] Venus pickup ions include all planetary ions that are accelerated by the action of the underlying convection

E00B04

electric field, whether they are escaping or on trajectories that return them to the atmosphere. It is important to appreciate that the pickup process is not restricted to the magnetosheath and solar wind outer regions. If the solar wind magnetic field penetrates the upper atmosphere between the ionopause and the exobase (or even below it), and a flow is present, an ion can still be accelerated by E = VXB. Different processes may also combine. For example, some ions may first move in response to other forces such as pressure gradient forces, and then attain escape speed when they are transported into a region where the convection electric field leads to their further acceleration. It is also important to remember that, unless they are observed in the mid-to-distant wake, pickup ions detected along a spacecraft orbit do not necessarily contribute to an escape flux. In determining whether an observed O+ ion is escaping or not, one must consider its energy, its location with respect to the planet, and its direction of motion. One also needs to know how representative the orbital sampling of the escaping ions is of the total escaping ion population. A global model of the ions is thus extremely useful for determining both whether a particular observed O+ flux is escaping, and what the implications of the local measurements are for the overall loss rates. [18] As also mentioned above, one of the striking features of the heavy ion spectrograms in Figures 2a–2d is their diverse appearances. Most show a period of enhanced fluxes around periapsis at the lowest energies and altitudes measured. These probably represent the pickup O+ source region in the upper ionosphere that is dominated by ionization and nightward transport of the dayside thermosphere neutral O source, rather than by the much more rarefied exospheric O from dissociative recombination of ionospheric O+2 . The coexistence of cold and hot oxygen atmospheric components in the pickup ions has been discussed previously (e.g., recently by Luhmann et al. [2007]). The ion spectrogram features at the higher altitudes exhibit the inbound-outbound asymmetries expected from the pickup process and the finite heavy ion gyroradius effects found for the more energetic pickup ions. When the interplanetary magnetic field component perpendicular to the solar wind velocity rotates from one angle (or clock angle) to another, the appearance of the pickup ion wake signature can change dramatically [e.g., Kallio et al., 2006]. But we do not even know how the low- and higher-energy ion observations are related to one another, or if their variation on the global scale is coupled. Such key outstanding questions only emphasize the difficulty of inferring global ion escape rates from one or two orbits during special events such as ICME passages.

Figure 4. Orthogonal views of some MHD model field lines (black) with superposed test particle O+ trajectories for MHD fields without (red) and with (blue) mass loading near the Venus obstacle. The model interplanetary field is northward (clock angle = 0). (a) Noon– midnight plane projection; (b) ecliptic plane projection; and (c) view from the Sun, looking tailward. The trajectories are traced to 3 Rv. These illustrate both the asymmetry and the spatial concentration of the oxygen pickup ion ‘‘wake.’’ 8 of 15

E00B04

LUHMANN ET AL.: VEX—OXYGEN ESCAPE AND CME

E00B04

Figure 5. (top) Cross sections of the picked up O+ ion wake at the (right) terminator plane and (left) down-tail color coded by ion energy. This case uses the MHD model with mass loading effects and is for 90° (+y) interplanetary field clock angle. The northward pointing convection electric field produces the north-south asymmetry. (bottom) Same as top but using the MHD fields and flows without mass loading effects. [19] The model of the global pickup process that we use here has been validated by comparisons with magnetic fields measured on PVO [Kallio et al., 1998], and with PVO plasma analyzer measurements of O+ [Luhmann et al., 2006]. The background fields and flows are provided by an MHD model of the typical 0.73 AU solar wind interacting with a spherical conductor whose radius is at the nominal subsolar ionopause height for Venus, with an interplanetary magnetic field perpendicular to the incident flow direction. The trajectories of planetary O+ ions are traced as test particles in these background magnetic fields and flows. We use the same numerical equation of motion solver, including the Lorentz force and Venus gravity, as used by Luhmann et al. [2006]. The test particle pickup ions’ properties (energy and directional distributions) are the collective result of the assumed distribution of O+ starting points, their initial velocities, the geometry and strength of the background MHD model magnetic fields, and the bulk plasma flows that together with these magnetic fields describe the E = VXB convection electric field. The effects of the Venus-Sun (x) component of the interplanetary magnetic field are not included in this model. This

allows us to rotate the model results around the Venus-Sun or x axis to approximate the effects of orbital sampling of the pickup ion population for different interplanetary field clock angles (relative to Venus North). The x component of the field is known to introduce some dawn-dusk asymmetries into the plasma interaction with Venus and so its neglect should be kept in mind, but for many analyses this is a secondary effect. [20] Two versions of the MHD model were used to calculate the pickup ion trajectories: one assumes the solar wind interacts with the conducting spherical obstacle with no effect of the production of heavy ions in the inner magnetosheath. Such a scenario might apply if the ionopause is at higher altitudes due to a low solar wind pressure or an enhanced ionospheric pressure. The other version assumes mass loading of the underlying flow by heavy O+ ions consistent with a simplistic approximation to photoion production from both thermospheric and exospheric atomic oxygen above the dayside of the planetary obstacle. (See Kallio et al. [1998] for further details.) [21] To illustrate the basic nature of the pickup ion spatial distributions and their sampling by the VEX orbit, Figure 4

9 of 15

E00B04

LUHMANN ET AL.: VEX—OXYGEN ESCAPE AND CME

E00B04

Figure 6. Illustration of VEX periapsis pass sampling of model O+ trajectories (obtained with the MHD field and flow model with no mass loading) for four interplanetary field clock angles: (a) 0°; (b) 45°; (c) 90°; and (d) 270°. The view is from the Sun. These suggest how an in situ ion observer’s perspective could be biased by a particular orbit’s geometry of intersection with the ion wake.

compares the trajectories of several hundred O+ ions launched on a spherical shell within a few hundredths of a planetary radius of the MHD model’s inner boundary for MHD model fields with (blue) and without (red) mass loading. The trajectories are traced to 3 Rv (Rv = the Venus obstacle in the model (planet plus ionosphere) 6300 km), a distance that covers the VEX orbit sampling of near-Venus space. Three projections, one as seen from the Sun, one in the ecliptic, and one in the noon –midnight plane are shown. The pickup ion distribution global asymmetry seen here, with the highest concentration of large-gyroradius trajectories initially moving toward the hemisphere where E = VXB points, is a well known signature of the pickup process. While the two cases produce similar trajectories, the mass loaded case trajectories are more spatially concentrated when seen in cross section from the Sun. In both cases it is apparent that one’s assessment of the escaping ion fluxes in such a distribution will depend on where and when they are sampled as the spacecraft orbits and the interplanetary field (and ion wake) rotate. [22] Figure 5 displays complementary plots of the locations where the modeled pickup ions cross thin cross-

sectional slabs at the terminator and downstream for the cases with (Figure 5a) and without (Figure 5b) mass loading. Here the interplanetary field is along the +y axis so that the maximum density of the asymmetric pickup ion wake seen in Figure 4 is in the north or +z half plane. The existence of a wedge-like feature and a low-energy central wake concentration at several Rv downstream is consistent with VEX results reported by Barabash et al. [2007a] from statistical analyses of the oxygen ions in the Venus wake, while the terminator cross section asymmetric bulge in the direction of the upstream convection electric field is consistent with reported ionopause cross section asymmetries seen on PVO [Phillips et al., 1988]. These cross sections, together with the ion trajectories in Figure 4, emphasize the asymmetry and limited cross section of the pickup ion wake, properties that are critical when evaluating global escape rates from limited orbital sampling. This is further emphasized by Figures 6 and 7 which show the four selected periapsis orbit segments together with the pickup ion trajectories. The ion trajectories have been rotated to three orthogonal positions to represent the effect of a range of interplanetary field clock angles on the orbital sampling

10 of 15

E00B04

LUHMANN ET AL.: VEX—OXYGEN ESCAPE AND CME

E00B04

Figure 7. Same as Figure 6 but for the ecliptic plane projection: (a) 0°; (b) 45°; (c) 90°; and (d) 270°.

geometry. These illustrate that one could infer a large range of ion escape fluxes, from negligible to representative of the maximum, from what is measured along the same VEX orbit, depending on the location of the orbital plane and the prevailing interplanetary field. Models provide a way to better understand the implications of the observations and also allow numerical experiments to determine which model assumptions produce results consistent what is observed. [23] To obtain the general character of the VEX heavy ion energy-time spectrograms, we found the following choices for the O+ starting grid were necessary: (1) starting points only in a narrow altitude range near the inner (spherical) model boundary; (2) a starting point grid including both dayside and nightside hemispheres, rather than just the dayside; (3) nonzero test particle starting velocities. The first of these was necessary to reproduce the narrow (in energy) pickup ion signatures; the nightside grid was necessary to obtain the observed deep wake fluxes of O+ near the base of the induced magnetotail current sheet separating the tail lobes; the nonzero starting velocities allowed the consistent reproduction, with the same model, of the low-altitude, low-energy ion signature and the sheath energetic ion signatures for the same pass. This last assumption also has the merit that PVO Retarding Potential Analyzer observations [Miller and Whitten, 1991] showed significant (5 – 10 km/s, up to nearly 10 eV) antisunward

thermal O+ flows in the upper ionosphere at solar zenith angles near and beyond the terminator plane, interpreted to result from day-to-night ionospheric plasma pressure gradients. [24] To illustrate the general characteristics of the modeling results we show in Figure 8 simulated O+ energy-time spectrograms for eight interplanetary field clock angles, spaced at 45° intervals, for the trajectory associated with the 18 July 2006 periapsis. This pass provided a particularly good trajectory for illustrating the asymmetries in the pickup ion distribution because it roughly samples the northern hemisphere in the terminator plane from dawn to dusk. We found it was not necessary to use many ions or a closely specified altitude of the starting point grid to establish the general patterns in the spectrograms, thus we started approximately 1200 O+ ions uniformly distributed over a spherical grid at 1.05 times the obstacle radius. The results are shown for two cases: the background MHD model without mass loading and an initial O+ velocity of a few 10s of km/s at the terminator, and an MHD model with some rudimentary ionospheric mass loading (see discussion of these two models by Kallio et al. [1998]), and a lower (by 0.1x) initial O+ velocity. The spectrograms were constructed by calculating the number of constant time step ion trajectory points falling within spherical volumes of 0.1 Rv radius, stepped along the VEX orbit through the

11 of 15

E00B04

LUHMANN ET AL.: VEX—OXYGEN ESCAPE AND CME

E00B04

Figure 8. Simulated ASPERA-4 IMA O+ spectrograms for the nominal (no mass loading) MHD field and flows case, for clock angles spaced 45° apart, obtained by ‘‘flying’’ the 18 July 2006 orbit through the modeled ion wake. As expected, the spectrograms look very different, although the ion wake is the same in all cases. periapsis pass. The velocity vectors of the particles entering the spherical volumes were filtered to accept only those within the ±45° elevation angles accepted by the IMA. The color scale and its range are analogous to the log (counts) scale in the VEX spectrograms. Here the model statistics provide an idea of the numbers of O+ trajectory points falling within the sampling spheres. These statistics are proportional to the IMA counts. As mentioned above, the assumption of the initial velocity of the ions affects the low-

energy near-periapsis O+ signatures, as does the use of mass loaded versus not loaded MHD field and flow models. [25] The general patterns in the modeled heavy ion spectrograms in Figure 8 resemble those in the observations in Figures 2a– 2d. In particular they reproduce the occasionally observed narrow bands of high intensity O+ in the sheath on inbound and/or outbound legs with velocities (energies) that smoothly increase (decrease) as the spacecraft altitude increases (decreases). In the model, these features do not represent a time-dependent acceleration

Figure 9. Energy-mass diagrams for the narrow layers (5 min duration) visible on (a) the outbound leg of case 1 and (b) the inbound leg of case 2 ICME pass spectrograms shown in the middle of Figures 2a and 2b. O+ appears to be present in these features. 12 of 15

E00B04

LUHMANN ET AL.: VEX—OXYGEN ESCAPE AND CME

Figure 10. (top) Views from the Sun of the O+ trajectories in the nominal model (red) and for a mass-loaded, increased (3) interplanetary field model (blue). (bottom) VEX ICME orbit sampling of the latter model. The VEX orbit intersects much less of the pickup ion wake owing to the reduced O+ gyroradius and the already confining effects of the mass loading on the O+ wake extent. signature, but rather sampling of a stationary spatial distribution of ions, with locally narrow energy spectra at each location. They are analogous to the features referred to as beams in the literature [e.g., Carlsson et al., 2006], although the ions are beam-like only in the sense that they have narrow energy spectra and are traveling in approximately the same directions at a given location. The term ring beam distribution, widely used in pickup ion studies, is the better descriptor. [26] A key result is that for some clock angles practically no model pickup ions are detected along the VEX orbit even though the global pickup ion population, including the escaping component, is identical. In these cases the nar-

E00B04

rowness and low altitude of the production layer, coupled with the magnetic field clock angle, minimize the orbit’s intersection with the pickup ion wake. Of special interest for the ICME observations are the spectrograms for 0° and 180° clock angles. One characteristic of ICME ejecta fields is rotation to large inclinations out of the ecliptic plane. The model results suggest that the prominent pickup ions features in the sheath should become observable as the ICME fields sweep through the large inclinations. However, the observations in Figures 2a– 2d show that this is generally not the case during the periods of the selected ICME passes, which are short on the ICME timescales (compare Figure 1). The approximate clock angles at the times of the IMA data are 0° for the 2006 5 July case, 180° for the 2006 18 July case, 225° for 2006 11 September case, and 45° for 2007 14 February. Only in the last case does the VEX heavy ion spectrogram show the O+ feature suggested by its model counterpart in Figure 8, although the orbital sampling differs from cases 2 and 3. [27] These model results also assume fixed typical solar wind B and V magnitudes of about 14 nT and 400 km/s, respectively, while the ICME often have higher B by a factor of 2 or more, and higher V by up to a factor of 4. The combined higher V and B effects on the ion trajectories will offset one another somewhat because the gyroradius is proportional to particle v, which is accelerated to up to twice the ambient flow speed, and inversely proportional to B. However, the higher V means that the energies attained by the picked up O+ can be much larger than is typical. Thus the O+ trajectories can be the same but the O+ energies may exceed the energy range of the ASPERA-4 IMA. In such cases the IMA would still observe the lower altitude, lower energy portions of the picked up O+ features, and just not detect the full energy-dispersed features in the sheath. However, this does not seem to be the case in our first three examples. In addition, the solar wind velocities for the four selected ICME, obtained by inspection of the IMA energy-time spectrograms in Figures 2a– 2d, seem to rule out an energy range limitation to the picked up O+ detection in the sheath. For the pass on 5 July 2006 the center of the solar wind spectral band prior to and after the broadening due to the shock suggests a nominal solar wind proton energy only slightly greater (1.3 keV) than the typical 1 keV. In fact many ICME travel only modestly faster than the ambient solar wind. On 18 July 2006, only the outbound shock is visible in the spectrogram, but it too suggests solar wind proton energies at or slightly less than the typical 1 keV. The 11 September 2006 case also shows only outbound solar wind, again near 1 keV, and the ICME segment passing on 14 February 2007 appears to have only 800 eV solar wind protons. However, while none of the cases under consideration here should be subject to the IMA energy range upper limit mentioned above, we expect the strongest events to be affected in this way. [28] A more likely explanation for the apparent absence of significant O+ in the IMA spectrograms may be the increased magnetic fields. Figure 9 shows energy-mass analyses of the very narrow features inbound on 5 July 2006 and outbound on 18 July 2006, where the spacecraft apparently enters and leaves the ionosphere. These reveal that O+ is indeed present for those short periods (5 min.). In addition, hints of O+ below the 10 eV lower threshold of

13 of 15

E00B04

LUHMANN ET AL.: VEX—OXYGEN ESCAPE AND CME

E00B04

Figure 11. Simulated ASPERA-4 IMA O+ spectrograms for the four selected ICME. The left column is the nominal (not-loaded) model; the right is for loaded plus high field. The most striking features consistent with the observations in Figures 2a– 2d are the energy-dispersed, spectrally narrow features occasionally seen in the sheath regions. The other notable features also present in the VEX data are the occasional low-energy ion enhancements observed throughout the lowest altitude portion of the pass. The origins of these features are discussed in the text. the IMA energy range in both cases suggest that intensive mass loading of the solar wind interaction boundary at these times reduces the initial velocities of the source O+. In addition, the larger magnetic field magnitudes during the ICME reduce the pickup ion gyroradii so that the pickup ion wake spatial distribution may be more closely confined to the optical wake. As an approximate model of these conditions we used the mass loaded MHD field and flow results with the magnetic field multiplied by 3, and the lower O+ starting velocities (2 km/s) used for the mass-loaded case as described earlier. The particle trajectories for this case are compared with the nominal case and with the spacecraft near-periapsis trajectories in Figure 10. The larger magnetic field reduces the O+ gyroradius as expected, altering the ion wake cross section and its orbital intersection. In Figure 11 the resulting O+ spectrograms for the four ICME cases are compared with the same orbital samples of the normal (standard magnetic field, no mass loading) model for the event pass trajectories. The high field, mass loaded case on the right shows attributes that tend toward the observed ICME heavy ion spectrogram behavior: an absence of the

long, energy-dispersed sheath ion features, and O+ concentrated in nearly vertical features at the boundaries of the source region entry and exit. A feature that remains puzzling, and for which we have no explanation, is the apparent opposite sides of these latter features from what would be expected for the 5 July and 18 July 2006 cases with 0° and 180° clock angles respectively. Clearly there is a need for carrying out more specific simulations using the observed ICME external parameters, with a detailed treatment of the ionospheric source, in order to obtain more realistic background fields and with them, more accurate models of the picked up O+ behavior.

4. Conclusion [29] So is there any evidence that the ICME selected for study in this paper produced an increase in the escaping pickup O+ fluxes? For the four selected events examined here, perhaps only in one case (case 4). Although the flux conversion for IMA counts is still under discussion, a quantitative estimate of the increase can be made from the

14 of 15

E00B04

LUHMANN ET AL.: VEX—OXYGEN ESCAPE AND CME

color contours in the bottom middle panel of Figure 3b. Assuming this 15 min sample in the mass-energy analysis contains the bulk of the escaping oxygen ions for the pass, there are at least 10 times more present on this occasion compared to their counterparts in the other panels. The fact that the bulk of the ions are at low energies may be an artifact of the sampling, or it may be key to the physical process underlying their escape. Models of the pickup ion spatial distribution discussed above suggest a possible relationship between the low-altitude, lower energy ions and the higher-altitude more energetic ions. They also provide an explanation for the nonobservation of O+ signatures for the three other events, related to the alteration of the pickup ion wake geometry by the enhanced ICME magnetic field and VEX orbital sampling. But a more detailed examination of this question requires further model development of the ICME-disturbed Venus interaction that is beyond the scope of the present report. We also expect an increase in the number of CME as solar activity rises to give us a picture of the responses of Venus picked up O+ to much stronger (faster) events. [30] Acknowledgments. Several of the authors (J.G.L., C.T.R., and D.A.B.) acknowledge support from the Venus Participating Scientist and Instrument Scientist Programs through NASA grants: NNG06GC61G (J.G.L.), NNG06GC62G (C.T.R.), NNX07AI62G S01 (D.A.B.). We are grateful to ESA and NASA for this opportunity to work with the Venus Express mission data.

References Barabash, S., et al. (2006), The Analyzer of Space Plasmas and Energetic Atoms (ASPERA-3) for the Mars Express Mission, Space Sci. Rev., 126, 113 – 164, doi:10.1007/s11214-006-9124-8. Barabash, S., et al. (2007a), The loss of ions from Venus through the plasma wake, Nature, 450, 650 – 653, doi:10.1038/nature06434. Barabash, S., et al. (2007b), The Analyser of Space Plasmas and Energetic Atoms (ASPERA-4) for the Venus Express mission, Planet. Space Sci., 55, 1772 – 1792, doi:10.1016/j.pss.2007.01.014. Brace, L. H., W. T. Kasprzak, H. A. Taylor, R. F. Theis, C. T. Russell, A. Barnes, J. D. Mihalov, and D. M. Hunten (1987), The ionotail of Venus: Its configuration and evidence for ion escape, J. Geophys. Res., 92, 15 – 26, doi:10.1029/JA092iA01p00015. Brace, L. H., R. E. Hartle, and R. F. Theis (1995), The nightward ion flow scenario at Venus revisited, Adv. Space Res., 16, 99 – 112, doi:10.1016/ 0273-1177(95)00255-D. Carlsson, E., et al. (2006), Mass composition of the escaping plasma at Mars, Icarus, 182, 320 – 328, doi:10.1016/j.icarus.2005.09.020. Chassefiere, E. (1997), Loss of water on the young Venus: The effect of a strong primitive solar wind, Icarus, 126, 229 – 232, doi:10.1006/ icar.1997.5677. Fedorov, A., et al. (2006), Structure of the Martian wake, Icarus, 182, 329 – 336, doi:10.1016/j.icarus.2005.09.021. Fox, J. L., and A. J. Kliore (1997), Ionosphere: Solar activity variations, in Venus II, edited by S. Bougher, D. L. Hunten, and R. Phillips, pp. 161 – 188, Univ. of Ariz. Press, Tucson. F u t a a n a , Y. , S . B a r a b a s h , M . Ya m a u c h i , R . L u n d i n , a n d S. McKenna-Lawlor (2007), Geo-effective solar flare events in December 2006: Space weather effect on Mars and Venus oxygen loss to space, Eos Trans. AGU, 88(52), Fall Meet. Suppl., Abstract P23A-1087. Grinspoon, D. H., and M. A. Bullock (2007), Searching for evidence of past oceans on Venus, paper presented at DPS Meeting 39, Am. Astron. Soc., Washington, D. C. Jian, L., C. T. Russell, J. G. Luhmann, and R. M. Skoug (2006), Properties of interplanetary coronal mass ejections at one AU during 1995 – 2004, Sol. Phys., 239, 393 – 436, doi:10.1007/s11207-006-0133-2. Jian, L., C. T. Russell, J. G. Luhmann, and R. M. Skoug (2008), Evolution of solar wind structures from 0.72 to 1 AU, Adv. Space Res., 41, 259 – 266, doi:10.1016/j.asr.2007.03.023. Kallio, E., J. G. Luhmann, and J. G. Lyon (1998), Magnetic field near Venus: A comparison between Pioneer Venus Orbiter magnetic field

E00B04

observations and an MHD simulation, J. Geophys. Res., 103, 4723 – 4737, doi:10.1029/97JA02862. Kallio, E., R. Jarvinen, and P. Janhunen (2006), Venus solar wind interaction: Asymmetries and the escape of O ions, Planet. Space Sci., 54, 1472 – 1481, doi:10.1016/j.pss.2006.04.030. Kulikov, Y. N., et al. (2006), Atmospheric and water loss from early Venus, Planet. Space Sci., 54, 1425 – 1444, doi:10.1016/j.pss.2006.04.021. Lammer, H., et al. (2006), Loss of hydrogen and oxygen from the upper atmosphere of Venus, Planet. Space Sci., 54, 1445 – 1456, doi:10.1016/ j.pss.2006.04.022. Lindsay, G. M., C. T. Russell, and J. G. Luhmann (1995), Coronal mass ejection and stream interaction region characteristics and their potential geomagnetic effectiveness, J. Geophys. Res., 100, 16,999 – 17,014, doi:10.1029/95JA00525. Luhmann, J. G., and T. E. Cravens (1991), Magnetic fields in the ionosphere of Venus, Space Sci. Rev., 55, 201 – 274, doi:10.1007/ BF00177138. Luhmann, J. G., and J. U. Kozyra (1991), Dayside pickup oxygen ion precipitation at Venus and Mars – Spatial distributions, energy deposition and consequences, J. Geophys. Res., 96, 5457 – 5467, doi:10.1029/ 90JA01753. Luhmann, J. G., S. A. Ledvina, J. G. Lyon, and C. T. Russell (2006), Venus O+ pickup ions: Collected PVO results and expectations for Venus Express, Planet. Space Sci., 54, 1457 – 1471, doi:10.1016/j.pss.2005.10.009. Luhmann, J. G., W. T. Kasprzak, and C. T. Russell (2007), Space weather at Venus and its potential consequences for atmosphere evolution, J. Geophys. Res., 112, E04S10, doi:10.1029/2006JE002820. Mihalov, J. D., and A. Barnes (1982), The distant interplanetary wake of Venus- Plasma observations from Pioneer Venus, J. Geophys. Res., 87, 9045 – 9053, doi:10.1029/JA087iA11p09045. Mihalov, J. D., C. T. Russell, W. T. Kasprzak, and W. C. Knudsen (1995), Observations of ionospheric escape on Venus’ nightside, J. Geophys. Res., 100, 19,579 – 19,584, doi:10.1029/95JA01342. Miller, K. L., and R. C. Whitten (1991), Ion dynamics in the Venus ionosphere, Space Sci. Rev., 55, 165 – 200, doi:10.1007/BF00177137. Moore, K. R., D. J. McComas, C. T. Russell, and J. D. Mihalov (1990), A statistical study of ions and magnetic fields in the Venus magnetotail, J. Geophys. Res., 95, 12,005 – 12,018, doi:10.1029/JA095iA08p12005. Moore, K. R., D. J. McComas, C. T. Russell, S. S. Stahara, and J. R. Spreiter (1991), Gasdynamic modeling of the Venus magnetotail, J. Geophys. Res., 96, 5667 – 5681, doi:10.1029/90JA02251. Mulligan, T., C. T. Russell, and J. G. Luhmann (1998), Solar cycle evolution of the structure of magnetic clouds in the inner heliosphere, Geophys. Res. Lett., 25, 2959 – 2962, doi:10.1029/98GL01302. Phillips, J. L., J. G. Luhmann, W. C. Knudsen, and L. H. Brace (1988), Asymmetries in the location of the Venus ionopause, J. Geophys. Res., 93, 3927 – 3941, doi:10.1029/JA093iA05p03927. Svedhem, H., et al. (2007), Venus Express-The first European mission to Venus, Planet. Space Sci., 55, 1636 – 1652, doi:10.1016/j.pss.2007. 01.013. Terada, N., S. Machida, and H. Shinagawa (2002), Global hybrid simulation of the Kelvin-Helmholtz instability at the Venus ionopause, J. Geophys. Res., 107(A12), 1471,doi:10.1029/2001JA009224. Vaisberg, O. L., et al. (1995), Ion populations in the tail of Venus, Adv. Space Res., 16, 105 – 118, doi:10.1016/0273-1177(95)00217-3. Zhang, T. L., et al. (2006), Magnetic field investigation of the Venus plasma environment: Expected new results from Venus Express, Planet. Space Sci., 54, 1336 – 1343, doi:10.1016/j.pss.2006.04.018. S. Barabash, E. Carlsson, and Y. Futaana, Swedish Institute of Space Physics, Box 812, SE-981, 28 Kiruna, Sweden. ([email protected]; [email protected]; [email protected]) D. A. Brain, S. A. Ledvina, and J. G. Luhmann, Space Sciences Laboratory, University of California, 7 Gauss Way, Berkeley, CA 94720-7450, USA. ([email protected]; [email protected]; [email protected]) A. Fedorov, Centre d’Etude Spatiale des Rayonnements, BP-4346, F-31028 Toulouse CEDEX 4, France. ([email protected]) J. G. Lyon, Department of Physics and Astronomy, Dartmouth College, 6127 Wilder Laboratory, Hanover, NH 03755-3528, USA. (john.lyon@ dartmouth.edu) C. T. Russell, Institute of Geophysics and Planetary Physics, University of California, Box 951567, Los Angeles, CA 90095, USA. (ctrussel@igpp. ucla.edu) T. L. Zhang, Space Research Institute, Austrian Academy of Sciences, 8042 Graz, Austria. ([email protected])

15 of 15

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, E00B15, doi:10.1029/2008JE003096, 2008

for

Full Article

Tailward flow of energetic neutral atoms observed at Venus A. Galli,1 M.-C. Fok,2 P. Wurz,1 S. Barabash,3 A. Grigoriev,3 Y. Futaana,3 M. Holmstro¨m,3 A. Ekenba¨ck,3 E. Kallio,4 and H. Gunell5 Received 31 January 2008; revised 11 April 2008; accepted 26 August 2008; published 2 December 2008.

[1] The Analyzer of Space Plasma and Energetic Atoms (ASPERA-4) experiment on

Venus Express provides the first measurements of energetic neutral atoms (ENAs) from Venus. The results improve our knowledge on the interaction of the solar wind with a nonmagnetized planet and they present an observational constraint to existing plasma models. We characterize the tailward flow of hydrogen ENAs observed on the nightside by providing global images of the ENA intensity. The images show a highly concentrated tailward flow of hydrogen ENAs tangential to the Venus limb around the Sun’s direction. No oxygen ENAs above the instrument threshold are detected. The observed ENA intensities are reproduced with a simple ENA model within a factor of 2, indicating that the observed hydrogen ENAs originate from shocked solar wind protons that charge exchange with the neutral hydrogen exosphere. Citation: Galli, A., M.-C. Fok, P. Wurz, S. Barabash, A. Grigoriev, Y. Futaana, M. Holmstro¨m, A. Ekenba¨ck, E. Kallio, and H. Gunell (2008), Tailward flow of energetic neutral atoms observed at Venus, J. Geophys. Res., 113, E00B15, doi:10.1029/2008JE003096.

1. Introduction [2] As a part of the Venus Express (VEX) scientific payload, the Analyzer of Space Plasma and Energetic Atoms (ASPERA-4) experiment, consisting of the Electron Spectrometer (ELS), the Ion Mass Analyzer (IMA), the Neutral Particle Detector (NPD) and the Neutral Particle Imager (NPI), is designed to study the plasma environment of Venus. It provides the first observations of energetic neutral atoms (ENA) from Venus, in a similar manner as the ASPERA-3 experiment on Mars Express (MEX) provided the first ENA observations of Mars [Futaana et al., 2006a, 2006b; Galli et al., 2006b]. Unlike the MEX spacecraft, VEX is also equipped with a magnetometer (MAG) allowing for a better characterization of the plasma environment. [3] The first results of IMA, ELS, and MAG data from Venus have been published by Barabash et al. [2007b], Coates et al. [2008], Martinecz et al. [2008], and by Zhang et al. [2007]. The location and temporal variation of the plasma boundaries are found to be similar to the results derived from Pioneer Venus Orbiter (PVO) data [Martinecz et al., 2008]. Figure 1 shows the location of plasma boundaries derived in that study: at the bow shock the solar wind is slowed down to subsonic velocities, the magnetosheath between the bow shock and the inner plasma boundary is dominated by slow, heated up solar wind. The inner plasma boundary can either be defined as the 1

Physikalisches Institut, Universita¨t Bern, Bern, Switzerland. Geospace Physics Laboratory, NASA Goddard Space Flight Center, Greenbelt, Maryland, USA. 3 Instituted fo¨r Rymdfysik, Kiruna, Sweden. 4 Finnish Meteorological Institute, Helsinki, Finland. 5 Department of Physics, West Virginia University, Morgantown, West Virginia, USA. 2

Copyright 2008 by the American Geophysical Union. 0148-0227/08/2008JE003096$09.00

boundary where the planetary ions start to dominate the plasma or as the boundary where the interplanetary magnetic field BIMF piles up around the ionosphere. The two different concepts are equivalent. In this paper we shall use the term ‘‘Induced Magnetosphere Boundary’’ (IMB) for the inner plasma boundary. It is close to the geometrical shadow behind Venus as is shown in Figure 1. [4] Charge exchange between energetic ions from the plasma and neutral atoms in the exosphere produces ENAs. For a flux of monoenergetic protons jp(r) that are neutralized along the line of sight (LOS) through atmospheric hydrogen with a density nH(r) the ENA production equation reads Z jENA ¼ s

dr nH ðrÞjp ðrÞ;

ð1Þ

LOS

with jENA the ENA intensity in cm2 sr1 s1, and s the charge exchange cross section. For a proton-hydrogen reaction at 1 keV energy s  2  1015 cm2, and s  1  1015 cm2 for O+-hydrogen charge exchange [Lindsay and Stebbings, 2005]. ENA measurements therefore reflect the plasma flux distribution as well as the neutral densities along the LOS of the sensor. The motivation of the ENA observations at Mars and Venus is to better understand the interaction of the solar wind with the neutral atmosphere of nonmagnetized planets. To interpret the observations, comparison with ENA models is needed. Such comparisons in turn constrain the uncertainties about the exospheric densities and the plasma populations implemented in the model. [5] For Venus, so far only the ENA model predictions by Fok et al. [2004] and by Gunell et al. [2005] are available. For Mars, the list is longer. The ENAs originating from solar wind and magnetosheath protons were modeled by Holmstro¨m et al. [2002]. ENAs originating from planetary

E00B15

1 of 17

E00B15

GALLI ET AL.: TAILWARD FLOW OF ENAS AT VENUS

E00B15

Figure 1. The two plasma boundaries around Venus, bow shock and Induced Magnetosphere Boundary (IMB), as measured with Analyzer of Space Plasma and Energetic Atoms/Ion Mass Analyzer (ASPERA-4/IMA) and electron spectrometer (ASPERA-4/ELS) during the first 5 months of observations [Martinecz et al., 2008]. The coordinates are given relative to the Venus Solar Orbital (VSO) referenceq frame in units of planetary radii RV. The x axis is the Sun-Venus line XVSO, and the y axis is the ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 2 to XVSO : distance YVSO þ ZVSO

hydrogen [Lichtenegger et al., 2002] and from planetary oxygen [Barabash et al., 2002] were also investigated, but both for Mars and Venus the total ENA production is expected to be dominated by ENAs originating from solar wind protons. Because charge exchange hardly affects the energy of the fast particle, the typical energy of ENAs around Mars and Venus is predicted to be comparable to the 1 keV of the solar wind. The ENA models also show that the ENA outflow originating from planetary ions is coaligned with the convective electric field E = vSW  BIMF. On the other hand, the ENA outflow caused by solar wind protons is homogeneous around the planetary limb with respect to the Sun’s direction, irrespective of the BIMF direction. Gunell et al. [2006] show that this prediction does not change when an MHD or a hybrid code instead of an empirical code is implemented. The observations of the Martian ENAs made with MEX/NPD agree with the basic characteristics of the ENA models. The ENA emission of the dayside atmosphere and the tailward flow of ENAs from shocked solar wind are easily recognizable [Futaana et al., 2006a, 2006b; Galli et al., 2006b]. By analogy, we expect to see above the dayside of Venus solar wind ENAs that are scattered back from the atmosphere and a narrow ENA stream of shocked solar wind right above the subsolar point in the magnetosheath. On the nightside we expect to

observe a tailward flow of ENAs, for which unperturbed solar wind, magnetosheath plasma, or accelerated planetary ions are possible parent ions. The latter parent population is the only one that also produces a measurable amount of oxygen ENAs. [6] Because of the UV sensitivity of the NPD sensor, ENA measurements outside the Venus shadow have to be avoided in most cases. We will therefore concentrate on the tailward ENA flow observed when the spacecraft is inside the shadow. First examples of such measurements were published by Galli et al. [2008]. Here, we will provide a complete account of the nightside ENA observations. We will briefly describe the NPD sensor in section 2, before presenting the database that underlies this work (section 3). In section 4 we will show the global intensity images of tailward-flowing hydrogen ENAs; they will be compared to model predictions in section 5. The search for oxygen ENAs at Venus will be summarized in section 6, followed by the conclusions in section 7.

2. NPD Instrument [7] NPD is one of the four sensors constituting the ASPERA-4 [Barabash et al., 2007a] experiment on ESA’s VEX spacecraft (orbit insertion in May 2006). NPD is designed to measure hydrogen ENAs at energies between

2 of 17

E00B15

GALLI ET AL.: TAILWARD FLOW OF ENAS AT VENUS

E00B15

Figure 2. Typical energetic neutral atom (ENA) energy spectrum, integrated over 8 min of observation time in Venus eclipse on 24 May 2006. Top shows the spacecraft centered view. The black cross is the Sun’s direction, the thick red line is the Venus limb. The red stars that cover the edge of the field of view (FOV) of sector NPD2_0 denote the limb of the solar panel. Middle and bottom show the time-of-flight (TOF) spectrum and the derived ENA intensity spectrum observed in channel NPD2_2 (see text for further explanations). 0.2 and 10 keV and oxygen ENAs between 0.4 and 10 keV, using the time-of-flight (TOF) technique. Angular resolution is provided by having two NPD sensors (NPD1 and NPD2), each with three angular channels with a field of view (FOV) of 30°  5° giving a total instantaneous FOV of 180°  5° [Barabash et al., 2007a]. We can distinguish between hydrogen and oxygen ENAs mainly because the velocities of hydrogen and oxygen ENAs of equal energy are very different. Only oxygen ENAs more energetic than several keV have TOF values that overlap with the ones of hydrogen ENAs between 0.2 keV and 0.5 keV.

[8] Beside ENAs, NPD is also sensitive to UV photons. Observations with the Sun in the FOV therefore have to be avoided. Unfortunately, the dayside hydrogen exosphere of Venus proves to be too bright (more than 20 kR Lyman-a [Bertaux et al., 1978]) for the NPD sensor as well. [9] The spectrum reconstruction from raw data follows the same method described by Galli et al. [2006a]. As improvement over the first report on tailward ENAs observed at Venus [Galli et al., 2008] we have completed the database for observations inside the IMB and we have now incorporated the final laboratory calibration informa-

3 of 17

E00B15

GALLI ET AL.: TAILWARD FLOW OF ENAS AT VENUS

Figure 3. Venus Express orbit trajectory on 24 May 2006, plotted in the same coordinate system as Figure 1. The green rectangle highlights the time period (30 min around 0200 UT) during which the spacecraft is inside the shadow of Venus. tion for NPD, which became available recently [Grigoriev, 2007]. The instrument response, efficiency, and geometrical factors have remained the same within the error bars since our earlier publication [Galli et al., 2008].

3. Database [10] The database of nightside observations includes all suitable ENA observations made inside the eclipse of Venus from orbit insertion until the end of 2007. We will not discuss observations for which the FOV was directed away from Venus, but up to now no signals above the detection threshold have been encountered under such circumstances. [11] Observations with poor counting statistics or with other technical problems were excluded from further analysis. For instance, several of the six sectors of NPD may be obstructed by the solar panel, as is illustrated in Figure 2 by the red stars adjacent to the two sectors NPD2_0 and NPD1_0. To complicate things further, the ASPERA-4 main unit is mounted on a rotating platform. This allows for a faster coverage of the sky but drastically blurs the ENA signal for a given pointing direction. By default, the NPD FOV is rotated by 180° every 32 seconds. The minimum integration time of two minutes for a FOV sector of 30° therefore necessitates an observation time longer than 6  2 minutes. In the meantime the spacecraft position relative to Venus may vary considerably. [12] The following VEX orbits belong to the database of nightside NPD observations: [13] 1. Eclipse season from 16 to 29 May 2006, 11 different suitable orbits. Detection threshold is 1  104 cm2 sr1 s1. [14] 2. Eclipse season from 19 August to 3 September 2006, 5 suitable orbits. For this period the raw ENA intensities have been multiplied by a factor 4 ± 2 to compensate for a decrease in detection efficiency caused by an overexposure to UV light. Detection threshold is 4  104 cm2 sr1 s1. [15] 3. Eclipse season from 27 November 2006 to 7 January 2007: never a reliable ENA signal above the detection threshold of several 104 cm2 sr1 s1. [16] 4. Eclipse season from 8 July to 16 August 2007. Detection threshold is several 104 cm2 sr1 s1. All observations are taken into account that have suitably high count rates to synthesize a TOF spectrum. The ENA intensity derived from these measurements is to be understood only as a relative number to be multiplied with a yet

E00B15

unknown correction factor because the NPD sensor settings have been adjusted to make up for the decrease in detection efficiency. This gives us 12 different orbits from 8 July to 4 August 2007, and four further suitable orbits from 5 to 16 August 2007 when the end of this eclipse season approached. [17] We have statistically evaluated (see section 5) only the data obtained during the first two eclipse seasons (16 different orbits), for which absolute intensities can be derived. All ENA intensities were integrated over the energy range from 0.2 to 10 keV, the uncertainty of the final value being 30% typically. For analysis, all measurements were divided into intervals of two to ten minutes integration time, two minutes being enough only for very intense ENA signals. Moreover, we restricted the data evaluation to sector NPD2_2 because of the six NPD channels only this one was directed at the tailward ENA flow. All observations correspond to low solar activity.

4. Results [18] Figure 2 shows a typical NPD observation of the tailward hydrogen ENA flow, obtained on 24 May 2006. The VEX orbit trajectory for that date is sketched in Figure 3. As for most observations in our database, the time window for NPD observations inside the Venus eclipse was roughly 30 min (green rectangle in Figure 3). Figure 2 (top) shows the spacecraft centered view for 0154:46 UT, when VEX was on the nightside of Venus at an altitude of 2900 km. The thick red line indicates the Venus limb, the cross marks the Sun’s position. Shown in black are the six NPD sectors, the red stars close to sectors NPD2_0 and NPD1_0 outline the solar panel. Figure 2 (middle) shows the original and the reconstructed TOF spectrum of the hydrogen ENAs (peak between TOF bin 10 and 100) measured in sector NPD2_0. The flat background in the TOF spectrum (dashed line) is caused by UV light. Figure 2 (bottom) shows the ENA intensity spectrum derived from the TOF spectrum in the energy range from 0.2 to 10 keV. The ENA intensity, integrated from 0.2 to 10 keV, calculates to 1.4  105 cm2 sr1 s1. [19] The intensity spectrum shown in Figure 2 (bottom) is typical for the hydrogen ENA signals observed with NPD. The intensity spectrum shown in Figure 2 may be described as a two-component power law with a low-energy slope of 1.8 ± 0.2, a rollover at 1.0 ± 0.1 keV, and a slope at energies above 1.0 keV of 3.3 ± 0.1. Taking into account all ENA measurements from 2006 with a high signalto-noise ratio we find that most intensity spectra can be characterized by a two-component power law with a rollover between 0.4 and 2.0 keV. The median rollover lies at 1.1 keV. Unfortunately, for many measurement intervals poor counting statistics avoid a reliable reconstruction of the spectral shape of the ENA signal. Therefore, we do not use the shape of the measured spectrum as observational constraint for the ENA model. In section 5 we will illustrate the problem of interpreting measured energy spectra by showing a modeled ENA spectrum (see Figure 8). To make a quantitative comparison between observations and model, we will only use the integral ENA intensities because they are more reliable. [20] We have synthesized images of the ENA intensities taking into account all measurement intervals that fulfill the

4 of 17

E00B15

GALLI ET AL.: TAILWARD FLOW OF ENAS AT VENUS

E00B15

Figure 4. Hydrogen ENA (H-ENA) intensity image 1 of Venus, including 54 measurement intervals from 16 to 29 May 2006. During the measurements the Venus center varies between 32° ± 10° longitude and 34° ± 5° latitude, the spacecraft altitude is 0.25 ± 0.05 RV. The average FOV footprint equals 38°  36°. criteria listed in section 3. The data from May to September 2006 resulted in four different images for different vantage points and altitudes of the spacecraft. To create the images, the single intensity measurements were averaged over a 10°  10° square mesh. The three most comprehensive images are shown in Figures 4, 5, and 6. The plots are shown in a cylindrical projection of the Venus Solar Orbital (VSO) reference frame, the x axis being the VSO longitude, the y axis being the VSO latitude in degrees. The VSO reference frame is defined as follows: XVSO points from Venus to the Sun, ZVSO points to the North pole of the Venusian orbital plane, and YVSO closes the right-handed reference frame. By this definition, the Sun’s direction is always at 0° longitude and 0° latitude. Pixels that are covered more than once by the NPD FOV are shown in color corresponding to the observed ENA intensity. In units of 104 cm2 sr1 s1, the colors represent the following:

black is below 1, purple is 1 to 2, blue is 2 to 4, light blue is 4 to 6, green is 6 to 8, yellow is 8 to 10, orange is 10 to 12, and red is above 12. The identical color scale is used for observations and for models. Pixels that have not been covered more than once are left white. Because of the fast spacecraft proper motion during the pericenter passage of VEX and because of the simultaneous scanner operations, single measurements are associated with a FOV footprint of typically 40°  40°. This has to be taken into account when comparing model images with a much higher resolution to the observations (see section 5). [21] The evaluation of the first and second eclipse season (see list in section 3) yields the following statistics: (1) In 15 of the 16 different orbits the hydrogen ENA signal exceeds the detection threshold. (2) In 14 out of 15 cases the ENA intensity reaches its maximum in the ecliptic plane (20° to +20°) within 40° around the Sun’s direction.

5 of 17

E00B15

GALLI ET AL.: TAILWARD FLOW OF ENAS AT VENUS

E00B15

Figure 5. H-ENA intensity image 2 of Venus, including 35 measurement intervals from 16 to 24 May 2006. The Venus center varies between 27° ± 6° longitude and 9° ± 5° latitude, the spacecraft altitude is 0.57 ± 0.05 RV. The average FOV footprint equals 30°  30°. [22] In Figure 4 the maximum ENA outflow seems to be shifted 30° away from the Sun’s direction. However, this shift should not be overinterpreted as it may also be caused by an observational bias: for the image shown in Figure 4 the orange region between 30° and 60° VSO longitude has been covered dozens of times. The region closest to the Sun between 0° and 30° VSO longitude, which is predicted to produce even higher ENA intensities, was covered only once. [23] The median of all 16 orbits calculates to 1.2  105 cm2 sr1 s1 for the direction of most intense ENA outflow; the highest intensity ever measured is (4.2 ± 1.6) 105 cm2 sr1 s1. From a direction more than 60° away from the Venus limb we never (taking into account all NPD observation inside and outside the eclipse) see an unambiguous signal >4  104 cm2 sr1 s1.

5. Comparison to an MHD Model [24] To predict the ENA production of a planet, one requires a plasma model for the ion flux distribution around the planet and a model of the neutral exosphere. For Venus only few ENA model calculations have been done so far.

The models presented by Fok et al. [2004] and by Gunell et al. [2005] are technically similar since both are based on MHD calculations of the plasma and PVO measurements of the neutral oxygen and hydrogen exosphere. The resulting maximum ENA intensities on the nightside and the total production rates are consistent with each other within a factor of 2. [25] To interpret the NPD observations, we use the Venus ENA model developed by Fok et al. [2004]. This model predicts hydrogen and oxygen ENA images based on the MHD plasma simulation by Tanaka and Murawski [1997] assuming the neutral exosphere parameters from the Venus International Reference Atmosphere (VIRA) model [Keating et al., 1985]. A plot of the ion densities, velocities and temperatures calculated by Tanaka and Murawski [1997] is shown in Figure 7. The bow shock is easily recognizable in all three images; the IMB corresponds to the location in the middle where the velocity of the solar wind protons sharply drops from 200 to 0 km s1. No planetary hydrogen ions are included in the MHD model, all predicted hydrogen ENAs (H-ENAs) originate from solar wind protons. The oxygen ENAs (O-ENAs) are caused by accelerated planetary oxygen from the ionosphere (not shown in Figure 7). The most

6 of 17

GALLI ET AL.: TAILWARD FLOW OF ENAS AT VENUS

E00B15

E00B15

Figure 6. H-ENA intensity image 3 of Venus, including 14 measurement intervals from 20 August to 3 September 2006. The Venus center varies between 3° ± 12° longitude and 24° ± 10° latitude, the spacecraft altitude is 0.64 ± 0.20 RV. The average FOV footprint equals 70°  45°. important model parameters are the solar wind strength and the density profile of the hydrogen exosphere. [26] The parameters of the exosphere model by Keating et al. [1985] are summarized in Table 1. We implement only the neutral hydrogen and oxygen components, all other species have very low scale heights and particle densities above the exobase. The exobase height is set to 250 km. The vertical density profiles at altitudes above 3500 km, where the VIRA model is not defined any more, are extrapolated using the Chamberlain model [Chamberlain and Hunten, 1987]. If T be the exospheric temperature, MV the planetary mass, and r the radial distance from the planet center, the scale height H of the density profile is given by H¼

kB Tr2 : GmMV

ð2Þ

[27] For the thermal hydrogen, equation 2 yields about 200 km given the temperatures in Table 1, whereas for the hot hydrogen component H  1000 km. The scale height of the thermal oxygen is only 20 km at most for the hot subsolar region. Therefore, the exospheric oxygen hardly affects the production of ENAs, and the NPD observations

will be sensitive only to the neutral hydrogen parameters listed in Table 1. The only exception to this rule applies to ENA observations at low altitudes above the subsolar region, but this region is never crossed by the LOS of NPD. Above the terminator and the nightside the neutral hydrogen density is higher than the one of oxygen even at the exobase, and the charge exchange cross section for H-ENA production is twice as large as for neutral oxygen [Lindsay and Stebbings, 2005]. Table 1 also indicates that it is the thermal hydrogen component that dominates the ENA production on the nightside. Because of the low exobase density, the particle density of the hot component exceeds the one of the thermal component only at altitudes above 2000 km according to equation 2. [28] As a starting point we will compare the NPD observations to the model predictions using the values in Table 1. Later on we can easily simulate the effects of a thinner or denser neutral exosphere by simply varying the ENA model output. Because we deal with an ENA thin medium around Venus, the predicted ENA outflow depends linearly on the exobase density. [29] As far as the solar wind is concerned, we have to stick to the parameters implemented in the plasma model by Tanaka and Murawski [1997]: vsw = 400 km s1, nsw =

7 of 17

GALLI ET AL.: TAILWARD FLOW OF ENAS AT VENUS

E00B15

E00B15

Figure 7. Plasma MHD model by Tanaka and Murawski [1997]. The left shows the proton density, the middle shows the ion velocity along the Sun-Venus line, and the right shows the temperature. The top half corresponds to values in the meridional plane, whereas the bottom half corresponds to values in the equatorial plane of the Venus Solar Orbital (VSO) reference frame. The color scales are linear. 14 cm3, and BIMF = (0, 0, 14) nT. As mentioned in the introduction, knowledge of BIMF becomes only important when interpreting ENAs from planetary ions. On the other hand, the intensity of the solar wind ENAs and magnetosheath ENAs directly scales with the flux of the unperturbed solar wind. Unfortunately, we do not know yet the solar wind flux at Venus during the NPD measurements because the absolute calibration of the IMA sensor (and therefore density and speed of the solar wind ions) has not been finished yet. In a recent statistical evaluation of the magnetic field data measured with PVO, Jarvinen et al. [2008] find that the particle flux of 0.6  109 cm2 s1 assumed by Tanaka and Murawski [1997] is close to the most probable flux and is thus a good choice for a single observation. But the median particle flux, observed during solar minimum at Venus, is close to 0.9  109 cm2 s1. The probability that the actual median particle flux encountered during the 16 VEX orbits in 2006 was lower than 0.7  109 cm2 s1 or higher than 1.1  109 cm2 s1 is 25% in both cases [Jarvinen et al., 2008]. At the end of this section we will discuss how the choice of a solar wind flux that is 1.5 times smaller than the expected median value affects the comparison between model prediction and measurements. [30] In Figure 8 we compare the ENA intensity spectrum derived from the ENA model to the NPD observation from

24 May 2006. To achieve a direct comparison, we have rebinned the NPD observation shown in Figure 2 (bottom) to the 12 bins between 0.2 and 10 keV used in the ENA model (red curve). Obviously, the modeled integral intensity is higher than the observed one (black curve), around 1 keV the differential intensity is four times higher. The spectral shapes of the NPD measurement and of the model agree in so far as the peak of the model spectrum at 0.8 keV corresponds to a rollover in the NPD spectrum at the same energy. This is the energy of ENAs originating from weakly shocked solar wind. The slopes of the two-component power law that characterizes the measured spectrum seem to be quite different from the model. Note, however, the error bars of the measured spectrum. We find in general that the slopes of NPD energy spectra strongly depend on the instrument efficiency [Grigoriev, 2007] and on the exact algorithm used to reconstruct differential intensities. The position of the rollover and the integral intensity are much more robust parameters. Figure 8 is a further indication that the position of the rollover may be the only trustworthy spectral parameter to be derived from NPD measurements. Moreover, the integration over several minutes blurs all spectral features. We caution the reader against overinterpreting Figure 8. It only indicates that the observed rollover between 0.4 and 2.0 keV (see section 4) is consistent with the hypothesis that the majority of the observed ENA

Table 1. Default Model Parameters of the Neutral Exosphere for Varying Solar Zenith Anglea SZA (deg)

nH (cm3)

TH (K)

16 34 61 90 119 146 164

5.6e4 7.0e4 2.0e5 1.1e6 2.9e7 1.7e7 7.5e6

284 289 292 230 141 124 127

nH,

hot

(cm3)

1.0e3 – – – – – 1.0e3

TH,

hot

(K)

1000 – – – – – 1500

nO (cm3)

TO (K)

1.1e7 1.1e7 8.8e6 1.1e6 1.2e4 2.7e3 3.0e3

284 289 292 230 141 124 127

nO,

hot

(cm3)

6.6e4 – – – – – 2.0e3

TO,

hot

(K)

4800 – – – – – 4800

a SZA, solar zenith angle; see Keating et al. [1985]. The most important parameters are the exobase density and the temperature of the thermal hydrogen component.

8 of 17

GALLI ET AL.: TAILWARD FLOW OF ENAS AT VENUS

E00B15

Figure 8. Observed (solid curve) versus modeled (dotted curve) ENA intensity spectrum for the maximum ENA outflow observed on 24 May 2006 (see Figure 2 for the pointing direction).

tailward flow originates from shocked solar wind. Because the derived spectral features are not very reliable except for the rollover, we based the evaluation of NPD observations on the images of the modeled ENA intensities, integrated from 0.2 to 10 keV. [31] We now compare the four ENA intensity images from the first two eclipse seasons (three of them shown in Figures 4, 5, and 6) to the ENA model predictions. The square sum over all image pixels of observed ENA intensity jobs,i minus predicted ENA intensity jmod,i, divided by the uncertainty of the observed ENA intensity sj,i, serves as our merit function: c2 ¼ Si



jobs;i  jmod;i sj;i

2 :

ð3Þ

The goodness of fit is measured with the probability     P c2 ; f ¼ g 0:5c2 ; 0:5f :

ð4Þ

It indicates if the square sum of deviations c2 between model and observations is tolerable or if the model prediction is very unlikely to be consistent with the observations. As probability threshold below which a model has to be rejected we choose P(c2, f) = 1%. The degree of freedom f is the number i of statistically reliable image pixels (approximately the number of independent measurement intervals) minus the number of model parameters. For the number of model parameters we set 12, appropriate to describe the solar wind strength and the exospheric densities of the thermal and hot hydrogen. Before evaluating equation 3, the modeled ENA images of 2.5°  2.5° resolution are convolved with the same FOV footprints (typically 40°  40°) from which the NPD images were composed. The effect of this smearing process is shown for instance in Figure 10. The typical uncertainties sj,i in equation 3 are as large as jobs,i: the same location in the image is only covered a few times by independent measurements, each of which with a typical uncertainty of

E00B15

30%, and one measurement corresponds to a FOV footprint that covers a dozen image pixels. [32] Figure 9 shows the ENA model result for the observer position of NPD image 3 (Figure 6). Assuming the original VIRA exobase densities listed in Table 1 the smeared ENA model images show up to 3.5  105 cm2 sr1 s1 (Figure 9, top). The NPD images, however, do not exceed 1.4  105 cm2 sr1 s1 (red pixels in Figure 6). The corresponding fit probabilities calculated from equation 4 yield P  1% for three of the four NPD images. The VIRA model thus is consistent with most NPD measurements but it seems to be an upper limit at low solar activity, at least for the thermal hydrogen component. [33] Because the default ENA model predicts too high ENA intensities and because we did not have the time to calculate plasma simulations with changed input parameters, we test how a reduction of exospheric densities and temperatures affects the ENA model predictions. We choose two temperature regimes, one as stated in Table 1, the other one with all exospheric temperatures reduced by 20%. This reduction is motivated by the 25% error bars of the default temperatures suggested by Keating et al. [1985]. That is, an exospheric model with a 20% lower exobase bulk temperature is a plausible lower limit compared to observations. We then search for the best fitting exospheric density with the crude approach c  nH with c one free parameter, identical for all solar zenith angles (SZA). The optimized c is defined by the product of the fit probabilities (equation 4) for the four NPD images. Since most hydrogen ENAs owe their existence to the thermal hydrogen component, the outcome is not changed much whether the correction factor c is applied only to the thermal hydrogen or to all components listed in Table 1. The resulting ENA intensities change linearly with c. [34] For the default temperatures we find that the agreement between the ENA model and the four NPD images is optimized for neutral hydrogen exobase densities two times lower than in Table 1. The confidence level of 1% implies for the range of uncertainty c = 0.5 ± 0.25. For c = 0.5 we find P(c21 = 43, 7212) = 96%, P(c22 = 45, 3912) = 1%, but only 2 of 39 pixels deviate from the measurement by more than 2 sj,i. P(c23 = 15, 4012) = 98%, P(c24 = 13, 4612) = 99.97%, whereby 0 of 46 pixels deviate from the measurement by more than 2 sj,i. The latter is not a very selective image; because of large error bars and poor spatial coverage it puts only loose constraints on the best fitting model. Obviously, only image 2 (Figure 5) has a really low probability of agreement. This is because of the row of purple pixels to the upper right of the Sun’s direction: two independent NPD measurements show for this direction an ENA intensity of (1 ± 1) 104 cm2 sr1 s1 whereas the ENA model, convolved with the same FOV footprints, predicts an order of magnitude higher ENA intensities (see Figure 10, bottom). [35] Two examples for the model with c = 0.5 are shown in Figures 10 and 11. These images were calculated for the same SZA and planetary distance as the NPD images 2 and 3 in Figures 5 and 6. While the model image with default exospheric densities in Figure 9 has still a fit probability larger than 1%, the fit probability of the ENA model shown in Figure 11 is significantly higher. The model with reduced hydrogen exobase densities is also consistent with the

9 of 17

E00B15

GALLI ET AL.: TAILWARD FLOW OF ENAS AT VENUS

Figure 9. Model prediction of the tailward flow of H-ENAs on the nightside of Venus for image 3 in Figure 6 with the original exospheric densities and temperatures as proposed by Keating et al. [1985]. The top shows the model image in a resolution of 2°  2°, the bottom is the same image, convolved by the same FOV footprints appropriate for the observed image 3. 10 of 17

E00B15

E00B15

GALLI ET AL.: TAILWARD FLOW OF ENAS AT VENUS

Figure 10. Model prediction of the tailward flow of H-ENAs on the nightside of Venus for image 2 in Figure 5. The hydrogen exobase densities have been reduced by a factor of 2. The format is the same as in Figure 9. 11 of 17

E00B15

E00B15

GALLI ET AL.: TAILWARD FLOW OF ENAS AT VENUS

Figure 11. Model prediction of the tailward flow of H-ENAs on the nightside of Venus for image 3 in Figure 6. The hydrogen exobase densities have been reduced by a factor of 2. The format is the same as in Figure 9. 12 of 17

E00B15

E00B15

GALLI ET AL.: TAILWARD FLOW OF ENAS AT VENUS

Figure 12. Model prediction of the tailward flow of H-ENAs on the nightside of Venus for image 3 in Figure 6. All exospheric temperatures have been reduced to 80% of the default values listed in Table 1. The format is the same as in Figure 9. 13 of 17

E00B15

E00B15

GALLI ET AL.: TAILWARD FLOW OF ENAS AT VENUS

Figure 13. Modeled oxygen ENA intensity spectrum for the pixel of maximum ENA outflow in Figure 14. observational upper limit on O-ENAs, whereas the original VIRA parameters lead to O-ENA intensities that should be detectable with NPD (see section 6). [36] Reducing the exospheric temperatures is an alternative possibility to reduce the modeled ENA intensities. Switching on and off the hot hydrogen component has no significant effect on the predicted ENA outflow. The maximum ENA intensities along the Venus limb (the red pixels in Figures 10 and 11) change by 20% at most. But if the bulk exobase temperatures are reduced by 20% (including the thermal hydrogen) we find that the predicted ENA intensities drop by a factor of 2. An analogous evaluation as for the models with default exobase temperatures yields a correction factor of c = 1 with a factor of 2 uncertainty. The range of uncertainty 0.5 < c < 2 again corresponds to the confidence level of 1%. [37] We find that the agreement between the ENA model prediction and the NPD observations is optimized if two times lower hydrogen densities, lower exobase temperatures or a combination thereof is assumed. The low number of independent pixels along the Venus limb and the error bars of the measured images make it impossible to determine which effect offers the better explanation. Comparing Figure 11 (default temperatures, reduced hydrogen densities) to Figure 12 (default densities, reduced temperatures) shows that the two parameter sets lead to very similar ENA intensity images. The only visible difference is that the cool exosphere model leads to intensities that decrease faster for directions away from the maximum ENA outflow. [38] The only discrepancy between NPD and model images that cannot be solved by adjusting the exospheric parameters is the fact that in two images the observed maximum of the ENA outflow is shifted along the Venus limb away from the Sun’s direction. One example of this behavior is shown in Figure 4. Probably it can only be reproduced with a more refined plasma simulation. In all other respects the model images, after smudging them by the NPD FOV footprints, reproduce the four NPD images from the Venus eclipse observations in 2006. [39] About the influence of the unknown solar wind strength on the model results we can make the following estimate: for low solar activity a median solar wind flux of (9 ± 2) 108 cm2 s1 is expected [Jarvinen et al.,

E00B15

2008] whereas the plasma simulation was run for jSW = 6  108 cm2 s1. In the case of jSW = 9  108 cm2 s1 the ENA model would predict hydrogen ENA intensities approximately 1.5 times higher than shown here. This would imply that the exospheric densities or temperatures should be reduced even further. Keep in mind, however, that a factor of 2 is usually considered the accuracy in ENA modeling. The model used in this work predicts an ENA production rate that is almost two times higher than the value predicted by Gunell et al. [2005] although the input parameters for the solar wind and the neutral exosphere are identical in both models. Therefore, we do not change the optimized fit parameters for the exospheric hydrogen. [40] We conclude that the MHD plasma model, including only solar wind protons, reproduces the observed hydrogen ENA intensities and can therefore be relied upon for further work. The observed ENA outflow can be explained as shocked solar wind protons that charge exchange with the neutral hydrogen in the magnetosheath. Hydrogen ENAs of planetary origin are not needed to reproduce the required intensities, and ENAs of unperturbed solar wind protons cannot be observed in the shadow. The high temperature of the shocked solar wind is the reason why NPD detects an ENA signal although the spacecraft is inside the Venus shadow. This can be illustrated with the following comparison: the aperture angle of the ENA outflow around the Sun’s direction is observed to be typically 30° (see Figure 6). We obtain the same result if we refer to Figure 7 for the modeled temperature and flow velocity of the protons inside the magnetosheath at the terminator (T = (2 ± 1) 106 K, v = (250 ± 50) km s1) and calculate a proton scatter angle of rffiffiffiffiffiffiffiffi 2kT ¼ 25 . . . 50 : a ¼ arctan mv2

ð5Þ

[41] To conclude this section, we now calculate the total HENA production rate of Venus. The ENA model with default exobase densities and temperatures yields 1.2  1025 s1. Since a hydrogen exosphere with reduced density or temperature fits better to the observations we recommend for low solar activity an H-ENA production rate of 0.6  1025 s1, with a factor of 2 uncertainty. Because the neutral hydrogen exosphere is the dominant neutralizing agent to produce these H-ENAs, the total hydrogen loss of Venus due to charge exchange has to be also at least QH;CX  0:6  1025 s1 :

ð6Þ

This is consistent with a preliminary evaluation of IMA data [Barabash et al., 2007b] that suggests a lower limit of 1025 s1 for the H+ loss through the tail region.

6. Upper Limit of Observed Oxygen ENAs at Venus [42] Observing oxygen ENAs at Venus would be interesting as it would directly reveal an atmospheric erosion process. If there is an observable O-ENA signal it is truly planetary since in the solar wind the oxygen abundance is 5000 times lower than hydrogen. If Venus produces intense O-ENA signals anywhere, they are expected to appear in the

14 of 17

E00B15

GALLI ET AL.: TAILWARD FLOW OF ENAS AT VENUS

Figure 14. MHD simulation for the tailward flow of O-ENAs on the nightside of Venus for image 2 in Figure 5. The neutral hydrogen exospheric densities have been reduced by a factor of 2 compared to Keating et al. [1985]. The top shows the model image in a resolution of 2°  2°, the bottom is the same image, convolved by the same FOV footprints appropriate for the observed image shown in Figure 5. 15 of 17

E00B15

E00B15

GALLI ET AL.: TAILWARD FLOW OF ENAS AT VENUS

tailward flow (see Figure 7 by Fok et al. [2004]). According to the calibration in laboratory already an O-ENA signal of more than 1  104 cm2 sr1 s1 can be observed even in the presence of UV noise and an H-ENA peak if the O-ENA energy lies above 1 keV. [43] The entire data set of Venus eclipse measurements (see list in section 3) has up to now not provided any distinct O-ENA signal. A thorough analysis of the first eclipse season in May 2006, with the laboratory instrument response and efficiency for O-ENAs [Grigoriev, 2007] yields as upper limit for O-ENAs in the tailward flow of Venus the following numbers: 1  104 cm2 sr1 s1 for energies between 1 and 10 keV, and 5  104 cm2 sr1 s1 for the interval 1.5 down to 0.4 keV. [44] For the default exospheric parameters in Table 1 the model predicts oxygen ENA intensities above the NPD detection threshold. When discussing ENAs originating from planetary ions the ignorance of the BIMF direction is troublesome (see section 1), but it is very unlikely that during each of the 16 different orbits the convective electric field pointed to a direction not covered with NPD2_2. The median solar wind flux was probably even stronger than assumed in the plasma model, which deepens the discrepancy between model predictions and observations. However, the model predicts oxygen ENA intensities that are consistent with the observed upper limits if two times lower hydrogen exobase densities are implemented (see previous section 5). Figure 13 shows the modeled O-ENA spectrum: only 1/3 of the integral intensity is attributed to energies above 0.4 keV. If we take that into account and if we reduce the exospheric density of the thermal hydrogen by a factor of 2, the modeled O-ENA outflow results in Figure 14. The predicted ENA intensities are just below or equal to the observational limit of 5  104 cm2 sr1 s1, corresponding to the dark blue pixels. Alternatively, reducing the exobase bulk temperature by 20% also leads to oxygen ENA intensities that are consistent with the observations. The shape of the O-ENA outflow is in general very similar to the H-ENA outflow; it is only slightly narrower. Both the O-ENA and the H-ENA outflow are predominantly caused by charge exchange with neutral hydrogen. [45] The NPD observations do not rule out that the oxygen exosphere in 2006 was also more tenuous than assumed in the VIRA model. All we can say with the NPD images is that the presented plasma model is consistent with the nonobservation of oxygen ENAs for energies larger than 0.4 keV if we assume hydrogen exobase densities two times lower or exobase temperatures 20% cooler than the VIRA values. The total O-ENA production rate predicted by the MHD model with reduced hydrogen densities (illustrated in Figure 14) calculates to 7  1024 s1 over the energy range from 0.2 to 10 keV. [46] The total loss of oxygen due to charge exchange reactions at Venus will be higher than the total O-ENA production rate since the loss of oxygen pickup ions that are not neutralized before leaving the exosphere has to be accounted for as well. For the sum of losses due to charge exchange processes Fok et al. [2004] found for the default ENA model 1.5  1025 s1 but they cautioned that this should be understood as a lower limit. The reason is that the MHD model does not take into account finite gyroradius pick up of O+, which will heat up the ionospheric

E00B15

oxygen. On the basis of our current knowledge we estimate the total oxygen loss rate due to charge exchange to be of the order of QO;CX  1025 s1 :

ð7Þ

7. Conclusions [47] 1. The first ENA images of Venus have been compared to an ENA model. The model, combining the canonical VIRA exosphere model with an MHD plasma code of solar wind ions and planetary oxygen ions, reproduces the observed tailward flow of ENAs within a factor of 2. The observed ENA tailward flow can be understood as neutralized magnetosheath plasma. Although, the uncertainties of the measurements and of the model results limit us to an accuracy of a factor of 2 some exospheric parameters can be constrained. [48] 2. The MHD approximation seems appropriate to predict the ENA outflow. The resolution of the observed ENA images is insufficient to motivate the use of kinetic models. The images in most cases do not even allow to statistically discriminate the shape of H-ENA outflow against the one of O-ENAs. [49] 3. The VIRA model by Keating et al. [1985], extrapolated to higher altitudes assuming a Chamberlain profile, is consistent with most NPD measurements. However, the agreement between model and observations is optimized if the exobase densities or temperatures of the thermal hydrogen proposed by Keating et al. [1985] are reduced. If the exobase temperatures are not changed, the NPD observations are reproduced by multiplying the VIRA densities by a factor of 0.5 ± 0.25 for all SZA. If the bulk temperatures are reduced by 20%, the possible range for the thermal hydrogen is 1.0+1.0 0.5 times the VIRA densities. The inherent uncertainties of the observed ENA images do not allow to determine whether the exobase density or the exospheric temperature should be reduced with respect to Keating et al. [1985] for low solar activity. The other neutral components of the Venusian exosphere, including the hot hydrogen component, hardly affect the ENA outflow and therefore cannot be constrained with NPD measurements. [50] 4. Unfortunately, we cannot say anything about a possible contribution of planetary hydrogen and the influence of the interplanetary magnetic field on the ENA outflow, lacking high spatial and temporal resolution. The present model, which assumes that all H-ENAs originate from solar wind or magnetosheath ions, produces already higher H-ENA intensities than actually observed. Therefore, the contribution of planetary H-ENAs should be of minor importance. [51] 5. From the model we derive a total hydrogen ENA production rate of 0.6  1025 s1 with a factor of 2 uncertainty. This agrees well with Gunell et al. [2005], who predicted, prior to VEX, an average of 0.7  1025 s1 from a different MHD model. The authors also noted that this number does not vary much over the solar cycle. The H-ENA production rate of Venus is similar to the one measured at Mars with MEX/NPD [Galli et al., 2006b]. For the total oxygen ENA production of Venus at low solar activity an upper limit of 0.7  1025 s1 between 0.2 and 10 keV can be derived.

16 of 17

E00B15

GALLI ET AL.: TAILWARD FLOW OF ENAS AT VENUS

[52] 6. We estimate the total loss rates of oxygen and hydrogen due to charge exchange reactions to be of the order of 1025 s1 for low solar activity. This seems to be the same order of magnitude as found in preliminary evaluations of IMA data [Barabash et al., 2007b]. This number will be better constrained in the near future when absolute calibration of the IMA sensor becomes available. Lammer et al. [2006] calculated similar loss rates from ion pickup of 1  1025 s1 for hydrogen and 2  1025 s1 for oxygen. For oxygen this is already close to the total loss rate derived by Lammer et al. [2006]; minor contributions from ion sputtering and detached plasma clouds are predicted to add up to a total of 3  1025 s1. For hydrogen, Lammer et al. [2006] find that the thermal escape of the photochemically produced hot hydrogen component is more important. For this erosion process they predict, on the basis of hydrogen exobase densities of Rodriguez et al. [1984], a loss rate of 4  1025 s1. Rodriguez et al. [1984] studied models with thermal hydrogen densities similar to the default values incorporated in our MHD model. We therefore recommend for the hydrogen loss caused by photochemical processes QH,UV = (2 ± 1) 1025 s1 for low solar activity. [53] 7. ENA production on its own is not important for atmospheric loss, neither for oxygen nor for hydrogen. But ENA imaging enables researchers to obtain within short time a global overview on the interaction of the solar wind with planetary atmospheres and ionospheres. [54] Acknowledgments. The ASPERA-4 experiment on the ESA Venus Express mission is a joint effort between 16 laboratories in 11 countries, all sponsored by their national agencies. We thank all these agencies as well as the various departments/institutes hosting these efforts. The lead author wishes to acknowledge the support of the Swiss National Science Foundation.

References Barabash, S., M. Holmstro¨m, A. Lukyanov, and E. Kallio (2002), Energetic neutral atoms at Mars: 4. Imaging of planetary oxygen, J. Geophys. Res., 107(A10), 1280, doi:10.1029/2001JA000326. Barabash, S., et al. (2007a), The analyzer of space plasmas and energetic atoms (ASPERA-4) for the Venus Express mission, Planet. Space Sci., 55, 1772. Barabash, S., et al. (2007b), The loss of ions from Venus through the plasma wake, Nature, 450, 650. Bertaux, J. L., J. Blamont, M. Marcelin, V. G. Kurt, N. N. Romanova, and A. S. Smirnov (1978), Lyman-Alpha observations of Venera-9 and 10 I. The non-thermal hydrogen population in the exosphere of Venus, Planet. Space Sci., 26, 817. Chamberlain, J. W., and D. M. Hunten (1987), Theory of Planetary Atmospheres, 2nd ed. Academic, Orlando, Fla. Coates, A. J., et al. (2008), Ionospheric photoelectrons at Venus: Initial observations by ASPERA-4 ELS, Planet. Space Sci., 56, 802.

E00B15

Fok, M.-C., T. E. Moore, M. R. Collier, and T. Tanaka (2004), Neutral atom imaging of solar wind interaction with the Earth and Venus, J. Geophys. Res., 109, A01206, doi:10.1029/2003JA010094. Futaana, Y., et al. (2006a), First ENA observations at Mars: Subsolar ENA jet at Mars, Icarus, 182, 413. Futaana, Y., et al. (2006b), First ENA observations at Mars: ENA emissions from the Martian upper atmosphere, Icarus, 182, 424. Galli, A., et al. (2006a), Direct measurements of energetic neutral hydrogen in the interplanetary medium, Astrophys. J., 644, 1317. Galli, A., P. Wurz, S. Barabash, A. Grigoriev, H. Gunell, R. Lundin, M. Holmstro¨m, and A. Fedorov (2006b), Energetic hydrogen and oxygen atoms observed on the nightside of Mars, Space Sci. Rev., 126, 267. Galli, A., et al. (2008), First observation of energetic neutral atoms in the Venus environment, Planet. Space Sci., 56, 807. Grigoriev, A. (2007), The neutral particle detector on the Mars and Venus Express missions, Ph.D. thesis, Inst. of Space Phys., Kiruna, Sweden. Gunell, H., M. Holmstro¨m, H. K. Biernat, and N. V. Erkaev (2005), Planetary ENA imaging: Venus and a comparison with Mars, Planet. Space Sci., 53, 433. Gunell, H., M. Holmstro¨m, S. Barabash, E. Kallio, P. Janhunen, A. F. Nagy, and Y. Ma (2006), Planetary ENA imaging: Effects of different interaction models for Mars, Planet. Space Sci., 54, 117. Holmstro¨m, M., S. Barabash, and E. Kallio (2002), Energetic neutral atoms at Mars 1. Imaging of solar wind protons, J. Geophys. Res., 107(A10), 1277, doi:10.1029/2001JA000325. Jarvinen, R., E. Kallio, I. Sillanpa¨a¨, and P. Janhunen (2008), Hybrid modelling the Pioneer Venus Orbiter magnetic field observations, Adv. Space Res., 41, 1361. Keating, G. M., et al. (1985), Models of Venus neutral upper atmosphere: Structure and composition, Adv. Space Res., 5, 117. Lammer, H., et al. (2006), Loss of hydrogen from the upper atmosphere of Venus, Planet. Space Sci., 54, 1445. Lichtenegger, H. I. M., H. Lammer, and W. Stumptner (2002), Energetic neutral atoms at Mars: 3. Flux and energy distributions of planetary energetic H atoms, J. Geophys. Res., 107(A10), 1279, doi:10.1029/ 2001JA000322. Lindsay, B. G., and R. F. Stebbings (2005), Charge transfer cross sections for energetic neutral atom data analysis, J. Geophys. Res., 110, A12213, doi:10.1029/2005JA011298. Martinecz, C., et al. (2008), Location of the bow shock and ion composition boundaries at Venus: Initial determinations from Venus Express ASPERA-4, Planet. Space Sci., 56, 780. Rodriguez, J. M., M. J. Prather, and M. B. McElroy (1984), Hydrogen on Venus: Exospheric distribution and escape, Planet. Space Sci., 32, 1235. Tanaka, T., and K. Murawski (1997), Three-dimensional MHD simulation of the solar wind interaction with the ionosphere of Venus: Results of two-component reacting plasma simulation, J. Geophys. Res., 102, 19,805. Zhang, T.-L., et al. (2007), Little or no solar wind enters Venus’ atmosphere at solar minimum, Nature, 450, 654. 

S. Barabash, A. Ekenba¨ck, Y. Futaana, A. Grigoriev, and M. Holmstro¨m, Instituted fo¨r Rymdfysik, P.O. Box 812, SE-98128 Kiruna, Sweden. M.-C. Fok, Geospace Physics Laboratory, NASA Goddard Space Flight Center, Code 673, Greenbelt, MD 20771, USA. A. Galli and P. Wurz, Physikalisches Institut, Universita¨t Bern, Sidlerstrasse 5, CH-3012 Bern, Switzerland. ([email protected]) H. Gunell, Department of Physics, West Virginia University, P.O. Box 6315, Morgantown, WV 26506, USA. E. Kallio, Finnish Meteorological Institute, P.O. Box 503, FIN-00101 Helsinki, Finland.

17 of 17

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, E00B20, doi:10.1029/2008JE003215, 2008

for

Full Article

Induced magnetosphere and its outer boundary at Venus T. L. Zhang,1,2 M. Delva,1 W. Baumjohann,1 M. Volwerk,1 C. T. Russell,3 H. Y. Wei,3 C. Wang,2 M. Balikhin,4 S. Barabash,5 H.-U. Auster,6 and K. Kudela7 Received 4 June 2008; revised 11 August 2008; accepted 29 September 2008; published 20 December 2008.

[1] The induced magnetosphere at Venus consists of regions near the planet and its wake

for which the magnetic pressure dominates all other pressure contributions. Initial Venus Express measurements indicate a well-defined outer boundary, the magnetopause, of the induced magnetosphere. This magnetopause acts as an obstacle to deflect the solar wind. Across this boundary, the magnetic field exhibits abrupt directional changes and pronounced draping. In this paper, we examine the structure of the magnetopause using Venus Express magnetic measurements. We find that the magnetopause is a directional discontinuity resembling either a tangential or a rotational discontinuity depending on the interplanetary magnetic field orientation. Citation: Zhang, T. L., et al. (2008), Induced magnetosphere and its outer boundary at Venus, J. Geophys. Res., 113, E00B20, doi:10.1029/2008JE003215.

1. Introduction [2] It is well known that the solar wind interaction with a planet produces a magnetosphere-like structure near the planet with common features such as bow shock, magnetosheath, magnetotail, and boundary layers. These magnetosphere-like structures are found at all planets in the solar system regardless if the planet has an intrinsic global magnetic field or not. In the case of planets like Mercury, Earth, Jupiter, Saturn, Uranus and Neptune, a magnetosphere is formed by the interaction of the solar wind with the planet’s large intrinsic magnetic field. For a planet like Venus or Mars, which has no global intrinsic magnetic field but with atmosphere, an induced magnetosphere is created by the solar wind interaction with the highly conducting ionosphere. The induced magnetosphere is therefore analogous to the magnetosphere of an intrinsically magnetized planet, but occupies a smaller volume. [3] The term ‘‘induced magnetosphere,’’ to our knowledge, was coined first by Podgorny et al. [1980] in describing the solar wind interaction with comets and Venus. Ness et al. [1982] used this term in studying the interaction between Titan and the Saturnian magnetospheric flow. The induced magnetosphere was used because although 1 Space Research Institute, Austrian Academy of Sciences, Graz, Austria. 2 State Key Laboratory of Space Weather, Center for Space Science and Applied Research, Chinese Academy of Sciences, Beijing, China. 3 Institute of Geophysics and Planetary Physics, University of California, Los Angeles, California, USA. 4 Department of Automatic Control and Systems Engineering, University of Sheffield, Sheffield, UK. 5 Swedish Institute of Space Physics, Kiruna, Sweden. 6 Institut fu¨r Geophysik und Extraterrestrische Physik, Technische Universita¨t, Braunschweig, Germany. 7 Institute of Experimental Physics, Slovak Academy of Sciences, Kosˇice, Slovakia.

Copyright 2008 by the American Geophysical Union. 0148-0227/08/2008JE003215$09.00

there is no intrinsic magnetic field at Titan, it possesses a magnetotail with similar configuration as the magnetotail produced from an intrinsic magnetic field. Further, Ness et al. [1982] called the outer boundary of the induced magnetosphere the magnetopause. [4] Recently, Luhmann et al. [2004] revitalized the terminduced magnetosphere by giving it a global picture. Luhmann et al. [2004] gave a working definition of an induced magnetosphere as ‘‘everything between an outer boundary outside of which the obstacle has no effect on the external medium, and an inner boundary inside of which there is no effect of the external conditions.’’ The terminduced magnetosphere has been widely used by the recent Venus Express publications [Barabash et al., 2007; Zhang et al., 2007; Kallio et al., 2008; Fedorov et al., 2008]. Zhang et al. [2007] further updated the definition of the induced magnetosphere to be the regions near the planet and its wake in which magnetic pressure dominates the other pressure contributions. Figure 1 illustrates the current understanding of Venus induced magnetosphere and its boundaries. The outer boundary of the induced magnetosphere is the magnetopause and the inner boundary is the ionopause. The dayside portion of the induced magnetosphere is often called the magnetic barrier and the nightside portion of the induced magnetosphere the magnetotail. [5] Russell et al. [1979] noticed that the magnetic field piles up to form a magnetic barrier in the inner magnetosheath on the dayside from the initial observations of Pioneer Venus Orbiter (PVO). Zhang et al. [1991] found that the magnetic barrier acts as an obstacle to the solar wind in analogy to the Earth’s magnetosphere. They found that the magnetic barrier is bounded by the ionopause as the lower boundary and a pressure balanced magnetic barrier upper boundary. [6] An initial survey of the Venus Express (VEX) magnetic field time series data indicates a well-defined boundary, the magnetopause, located at the outer edge of the magnetic barrier region [Zhang et al., 2007]. This magne-

E00B20

1 of 9

E00B20

ZHANG ET AL.: VENUS-INDUCED MAGNETOSPHERE

E00B20

Figure 2. Magnetic field measurements around periapsis (0153 UT, altitude is 302 km, SZA is 88.8°) on 27 June 2006. The shadowed area is the induce magnetosphere.

Figure 1. Schematic of Venus induced magnetosphere and its boundaries. topause is easily identified as a sudden cessation of magnetosheath wave activity and an abrupt start of pronounced field draping in the inner magnetosheath where the field magnitude is enhanced. The magnetic barrier, an induced magnetosphere on the dayside, is found with the same thickness at both solar minimum and solar maximum [Zhang et al., 2008a]. Finally, it was found that the magnetopause has no signature in the field magnitude and the thickness is about 200 km in vertical extent. It is the purpose of this paper to present a more detailed analysis of the magnetic field data on the induced magnetosphere and its outer boundary. A complete description of the induced magnetosphere will require a combination of all relevant magnetic and plasma data. However, at time of writing, the plasma moment data are still not available, thus we restrict our analysis to the magnetic field data only.

carefully designed to cope with this great challenge of a ‘‘dirty’’ spacecraft with a large dynamic range, large automatic compensation range, and high digital resolution. [8] We process and clean the data based on dual-sensor measurements. The detailed description of the data cleaning method can be found by Zhang et al. [2008b]. Basically, we examine the time series DB, of the differences of the inboard and outboard magnetic field measurements in

2. Instrumentation and Data [7] The Venus Express mission is the first European misson to Venus [Titov et al., 2006]. The spacecraft has a highly elliptical polar orbit with a period of 24 h. Among other instruments, it carries a magnetometer to investigate the Venus plasma environment [Zhang et al., 2006]. The Venus Express magnetometer (MAG) consists of two triaxial fluxgate sensors. Because of schedule and budget constraint, no magnetic cleanliness program was implemented. The outboard sensor is mounted to the tip of a one meter deployable boom whereas the inboard sensor is directly attached to the top panel of the spacecraft. The magnetometer is designed using dual sensor or gradiometer method [Ness et al., 1971]. Both sensors sample simultaneously, to enable separation of spacecraft originated stray field effects from the ambient space field. The instrument has been

Figure 3. The magnetic field in VSO coordinates plotted along the trajectory through the induced magnetosphere on 27 June 2006. The thick line marks the induced magnetosphere part of the trajectory. Magnetopause crossings are indicated at 0146 and 0200 UT.

2 of 9

E00B20

ZHANG ET AL.: VENUS-INDUCED MAGNETOSPHERE

E00B20

Figure 4. (a) Altitude profiles of the magnetic field components and magnitude in the inbound magnetopause crossing on 27 June 2006. MVA is applied on the shaded area. (b) Hodogram of the magnetopause crossing.

spacecraft coordinates for all three components. In the DB time series, all variations are caused by the spacecraft disturbances, no natural signal appears on these time series. This allows us to categorize the spacecraft disturbances using various mathematical approaches such as fuzzy logic, neural network etc. In general, there are three kinds of disturbances in the DB time series: jumps, regular drift variations, and irregular variations. Once all disturbances are categorized from the DB time series, we examine the effect of each kind of disturbance on the outboard sensor

measurements. Finally, automatic data cleaning software has been developed to remove all identified disturbances at the outboard sensor measurements and the offset determination algorithm is applied to remove the large strayfield [Leinweber et al., 2008]. [9] One of the advantages of the VEX magnetometer (MAG) measurements is the higher time resolution compared with that of PVO. During a nominal 24 h VEX orbit, the MAG data are sampled at 1, 32, and 128 Hz. Initially, the sampling rate was set at 128 Hz for 2 min and 32 Hz for

3 of 9

E00B20

ZHANG ET AL.: VENUS-INDUCED MAGNETOSPHERE

E00B20

Figure 5. (a) Altitude profiles of the magnetic field components and magnitude in the outbound magnetopause crossing on 27 June 2006. MVA is applied on the shaded area. (b) Hodogram of the magnetopause crossing. 120 min around the pericenter. Later, the fast modes have been increased to 10 min for 128 Hz and 4 h for 32 Hz. Furthermore, all 128 and 32 Hz data are merged into the 1 Hz data set by applying the exact same averaging algorithm as was done onboard. So far, all 1 Hz data are cleaned with an absolute field accuracy of
1 nT, and the variable field accuracy better than 0.1 nT. Thus the data are of a quality comparable to magnetic field measurements made onboard magnetically clean spacecraft.

3. Observations of Induced Magnetosphere [10] Figure 2 displays VEX MAG measurements during a typical induced magnetosphere encounter while the interplanetary magnetic field (IMF) was very steady indicated by the little variation in both orientation and magnitude between inbound and outbound IMF. The observations are 1 Hz averages in Venus Solar Orbital (VSO) coordinates where the x axis points from Venus to the Sun, the Y axis is

opposite to the Venus orbital motion and Z axis is northward. The spacecraft moves near the terminator region, defined as where the solar zenith angle (SZA) near 90°, from dusk (Y) to dawn (+Y), enters the magnetosheath at about 0128 UT and exits the magnetosheath at about 0232 UT. Around periapsis (0153 UT, altitude 302 km, SZA 88.8°) in the inner magnetosheath, the field is enhanced to form a magnetic barrier, the induced magnetosphere on the dayside (shaded region in the Figure 2). Apparently, the wave activity is suppressed in the magnetic barrier. Here we define the magnetopause as where the magnetosheath waves stop. Along with the cessation of wave activity at the magnetopause, the field direction exhibits an abrupt change as shown in Figure 2. Using also the dynamic power spectrum of the 32 Hz data (not shown in this paper [see Zhang et al., 2007]), we identify the magnetopause crossings at 0146 UT (altitude 823 km, SZA 87°) for inbound and 0200 UT (altitude 905 km, SZA 91°) for outbound. As expected for solar minimum conditions

4 of 9

E00B20

ZHANG ET AL.: VENUS-INDUCED MAGNETOSPHERE

E00B20

planet. We note that the inbound magnetopause crossing (0140 UT) is much better defined than the outbound magnetopause crossing (0200 UT). The reason for this difference is due to the location and nature of the magnetopause structure which we will examine in detail in the following section.

4. Magnetopause Structure [12] The minimum variance analysis (MVA), pioneered for space magnetic field applications by Sonnerup and Cahill [1967], has been widely used in the determination of the normal direction of the magnetopause. Since the divergence of B is zero, the variation of B along the normal to a thin current sheet is generally small, unless there is small-scale, patchy reconnection that is easily recognized. Thus MVA allows us to estimate the normal to simple planar current sheets from magnetic field data measured with a single spacecraft. Here we also test the stability of the normal orientation determined from MVA by performing the MVA with several nested time intervals around the discontinuity and checked for a consistent result. [13] Ideally, when the magnetic component along the current sheet normal, Bn, is zero, the current sheet can be categorized as tangential discontinuity; otherwise, it is a rotational discontinuity [Burlaga, 1969]. However, this definition is not valid when one applies MVA to determine the normal component of the current sheet using single spacecraft magnetic field measurements since the obtain Bn is never a zero number. Lepping and Behannon [1980] found that the ratio between Bn, obtained from MVA, and the averaged magnetic field magnitude B across the discontinuity can be used in distinguishing tangential from rotational discontinuity. They determined a threshold of jBn/Bj = 0.3 as the upper bound of a tangential discontinuity.

Figure 6. (a) Magnetic field observations of the induced magnetosphere and its outer boundary on 11 June 2006. (b) The magnetic field plotted along the spacecraft trajectory between 0132 and 0142 UT, 11 June 2006. The thick line marks the induced magnetosphere part of the trajectory. The magnetopause occurs at 0138 UT. [Zhang et al., 2006], no ionopause is observed for this periapsis passage since the ionopause is generally lower than the 250 – 350 km pericenter altitude of the spacecraft. In Figure 3 the magnetic field direction is plotted along the spacecraft trajectory. The reversal of the field in the X direction around the periapsis 0153 UT indicates the formation of the ‘‘magnetic pole’’ at terminator [Russell et al., 1982]. In describing the geometry of an induced magnetosphere, it is common to refer the magnetic pole and ‘‘magnetic equator’’ with respect to the orientation of IMF. The magnetic equator lies in plane which contains the IMF. The magnetic pole is in the direction perpendicular to the magnetic equatorial plane. [11] Clearly evident in the induced magnetosphere portion in the Figure 3 is the draping of the field lines about the

4.1. Case 1: 27 June 2006, Inbound [14] Figure 4 shows the magnetic field altitude profile of the inbound crossing on 27 June 2006. Between
700 and 900 km, the field direction changes dramatically with the tendency of By and Bz toward zero and increasing Bx, an indication of more draping. We apply MVA on the data between 0145:30 and 0147:00 UT, corresponding to the altitude between 898 and 685 km indicated by the shaded band in the plot. We find that the normal vector is [0.346 0.331 0.878]. The ratio of the intermediate to the minimum eigenvalues, l2/l3, is found to be 50.1, indicating a well-defined minimum variance direction. Here we have also tested the stability of the normal orientation determined from MVA by performing the MVAwith nested time intervals and the result is consistent. The magnetic field component along the minimum variance direction Bn = 6.01 nT and the average magnetic field magnitude across the magnetopause Bt = 37.07 nT. Since the ratio Bn/Bt is 0.16, which is much less the threshold of 0.3 [Lepping and Behannon, 1980], this magnetopause crossing is a tangential discontinuity. 4.2. Case 2: 27 June 2006, Outbound [15] Figure 5 shows the magnetic field altitude profile of the outbound magnetopause crossing on 27 June 2006. The crossing is not well identified as expected for a nightside magnetopause crossing of an induced magnetosphere. We

5 of 9

E00B20

ZHANG ET AL.: VENUS-INDUCED MAGNETOSPHERE

E00B20

Figure 7. (a) Altitude profiles of the magnetic field measurements on 11 June 2006. MVA is applied on the shaded area. (b) Hodogram of the magnetopause crossing. apply MVA on the data between 0158:35 and 0159:55 UT. We find that the normal vector is [0.478 0.571 0.668], the ratio of the intermediate to the minimum eigenvalues, l2/l3, is 11.5. Thus the minimum variance direction is well defined. Again the stability of the normal was tested with nested time interval and we found that the obtained minimum variance direction is stationary. For this magnetopause crossing, the magnetic field component along the minimum variance direction Bn = 13.9 nT and the average field magnitude Bt = 32.9 nT, the ratio is 0.42. Thus here we have a rotational discontinuity with a large field component along the normal direction.

4.3. Case 3: 11 June 2006, Inbound [16] Figure 6a shows magnetic field time series of a welldefined magnetopause crossing at 0138 UT, 11 June 2006 (altitude 862 km, SZA 76°). Similar as in the example of 27 June 2006, the IMF is mainly in Y direction. Figure 6b shows the spacecraft trajectory and field vector along it. The magnetopause crossing is evident by the field directional change with pronounced draping. Also the wave activity clearly stops at the magnetopause. Figure 7 shows the magnetic altitude profile. We apply MVA on the data between 0137:10 and 0139:00 UT, corresponding to the altitude between 994 and 718 km indicated by the shading in the

6 of 9

E00B20

ZHANG ET AL.: VENUS-INDUCED MAGNETOSPHERE

E00B20

We apply MVA on the data between 0134:15 and 0135:45 UT, corresponding to the altitude between 829 and 627 km indicated by the shading in Figure 9. We find that the normal vector is [0.347 0.158 0.925] and ratio l2/l3 of 3.79. Although the minimum variance direction is not well defined from MVA in this case, the normal direction is reasonably well determined since it is normal to the expected magnetopause at solar minimum. Furthermore, we have also tested the stability of the normal orientation determined from MVA by performing the MVA with nested time intervals and the result is consistent. The magnetic field component along the minimum variance direction Bn = 29.45 nT and the average field magnitude Bt = 41.5 nT, the ratio Bn/Bt is 0.71. Thus this magnetopause crossing resemble to a rotational discontinuity.

5. Discussion and Conclusions

Figure 8. (a) Magnetic field observations of the induced magnetosphere and its outer boundary on 1 June 2006. (b) The magnetic field plotted along the spacecraft trajectory between 0130 and 0140 UT, 1 June 2006. The thick line marks the induced magnetosphere part of the trajectory. The magnetopause crossing occurs at 0134 UT. plot. We find that the normal vector is [0.046 0.259 0.841]. The minimum variance direction is well defined with l2/l3 of 24.3. The magnetic field component along the minimum variance direction Bn = 6.04 nT and the average field magnitude Bt = 35.93 nT, the ratio is 0.17. Similar as case 1, this magnetopause crossing is a tangential discontinuity. 4.4. Case 4: 1 June 2006, Inbound [17] In Figure 8a we show the inbound magnetopause crossing on 1 June 2006. One main reason to select this event is that the IMF is mainly in the Z direction. From Figures 8a and 8b, the magnetopause crossing at 0134 UT (altitude 864 km, SZA 71°) is evident by the magnetosheath wave activity abrupt ending and pronounced field draping.

[18] The induced magnetosphere of Venus is a region within which the magnetic pressure dominates all other pressure contributions. It is strongest at the subsolar point and weakens with increasing solar zenith angle. Although the lower boundary of the induced magnetosphere can normally be defined by the ionopause, this is not always the case. For example, at solar minimum or at solar maximum but when the solar wind ram pressure is high, the ionosphere is fully magnetized, and the induced magnetosphere is found deeply inside the ionosphere. In such case, the lower boundary of the induced magnetosphere is below the nominal ionopause location. Furthermore, the nightside ionopause is ill-defined and so is the lower boundary of the induced magnetosphere at the nightside. [19] In contrast to the lower boundary of the induced magnetosphere at Venus, the upper boundary, a magnetopause separating the magnetosheath solar wind plasma from the induced magnetosphere, is well defined as shown by initial Venus Express measurements [Zhang et al., 2007]. While the VEX measurements characterize the magnetopause in the terminator region, previous studies show that the same magnetopause, although other names might have been given, extends both to the subsolar region and the distant magnetotail. In a study of the dimension and magnetic structure of the distant Venus magnetotail, Saunders and Russell [1986] identified the magnetopause as a directional discontinuity. By examining four unambiguous tail boundary crossings, they showed that Venus magnetopause is either a rotational or a tangential discontinuity depending on the location of the observation. They pointed out that the tangential discontinuity occurs at a plasma sheet/magnetosheath interface, wheras the rotational discontinuity occurs at a tail lobe/magnetosheath interface. Therefore, Saunders and Russell [1986] first revealed the IMF control of the magnetopause structure at the distant magnetotail. [20] A more recent study of the structure of the magnetopause, i.e., the magnetic pileup boundary, at Mars and Venus was performed by Bertucci et al. [2005]. Using six magnetopause crossings both at dayside Venus and Mars, they concluded that this boundary resembles a tangential discontinuity rather than a rotational discontinuity. However, no IMF orientation control of the structure of the magnetopause was considered in their investigation.

7 of 9

ZHANG ET AL.: VENUS-INDUCED MAGNETOSPHERE

E00B20

E00B20

Figure 9. (a) Altitude profiles of the magnetic field measurements on 1 June 2006. MVA is applied on the shaded area. (b) Hodogram of the magnetopause crossing. [21] In the present study, we examined four magnetopause crossings of the induced magnetosphere of Venus. We found that Venus magnetopause exhibits properties resembling both tangential and rotational discontinuity, a conclusion consistent with the results of Saunders and Russell

[1986], but different from the results from Bertucci et al. [2005]. In order to resolve this difference, we further examine the possible controlling factor of the structure of the magnetopause. In Table 1, we list the magnetopause crossing location and the IMF conditions. Further we list the

Table 1. Magnetopause Crossing Locations and IMF Conditions Case

Magnetopause Crossing in VSO (km)

IMF (nT)

Altitude (km)

SZA

Clock Angle

Magnetopause Structure

1 2 3 4

390 2853 6245 177 4895 4940 1645 2535 6219 2284 2007 6213

0.18 7.94 0.97 0.11 8.69 2.05 0.73 5.20 0.18 0.48 3.84 6.62

823 905 862 864

87° 91° 76° 71°

108° 32° 114° 168°

tangential rotational tangential rotational

8 of 9

E00B20

ZHANG ET AL.: VENUS-INDUCED MAGNETOSPHERE

altitude, the SZA and the clock angle of the magnetopause crossing. The clock angle is defined as the angle between the radius vector to the spacecraft at the magnetopause and the IMF both projected in the YZ plane. At terminator region, it is usual to call a clock angle of 90° or 90° as the magnetic pole and a clock angle of 0° or 180° as the magnetic equator. Because of the VEX orbital geometry, our magnetopause examples are limited to the terminator region covering a SZA between 71° to 91°. From Table 1, we can see that the magnetopause in a magnetic pole region exhibits properties of a tangential discontinuity, whereas a rotational discontinuity is found for a magnetopause at the magnetic equator. Therefore, it is evident that the IMF orientation exerts strong control on the structure of the magnetopause of the induced magnetosphere at Venus. [22] Ideally, the magnetopause is defined as pressurebalanced boundary where the magnetic pressure in the induced magnetosphere equals the ram pressure of the magnetosheath solar wind. However, due the insufficient temporal resolution of the PVO plasma instrument, such a definition could not be applied to data analysis in practice. A modified pressure balance boundary definition was proposed by Zhang et al. [1991] to determine the outer boundary of the magnetic barrier as the altitude where the magnetosheath magnetic pressure is equal half of the upstream solar wind dynamic pressure. Indeed, initial analysis of the ASPERA data shows that the so-called ion composition boundary [Martinecz et al., 2008] is colocated with the magnetopause. When the combined plasma and magnetic field data set is available, we plan to investigate the pressure balance nature of the magnetopause. [23] Acknowledgments. The work at China was supported by CAS grant KJCX2-YW-T13 and NNSFC grants 40621003 and 40628003. The work at UCLA was supported by the National Aeronautics and Space Administration under research grant NNG06GC62G. The work in Slovakia was supported by Slovak Research and Development Agency under the contract APVV-51-053805.

References Barabash, S., et al. (2007), The loss of ions from Venus through the plasma wake, Nature, 450, 650 – 653, doi:10.1038/nature06434. Bertucci, C., C. Mazelle, M. H. Acuna, C. T. Russell, and J. A. Slavin (2005), Structure of the magnetic pileup boundary at Mars and Venus, J. Geophys. Res., 110, A01209, doi:10.1029/2004JA010592. Burlaga, L. F. (1969), Directional discontinuities in the interplanetary magnetic field, Sol. Phys., 7, 54 – 71, doi:10.1007/BF00148406. Fedorov, A., et al. (2008), Comparative analysis of Venus and Mars magnetotails, Planet. Space Sci., 56, 812 – 817, doi:10.1016/j.pss.2007.12.012. Kallio, E., et al. (2008), The Venusian induced magnetosphere: A case study of plasma and magnetic field measurements on the Venus Express mission, Planet. Space Sci., 56, 796 – 801, doi:10.1016/j.pss.2007.09.011. Leinweber, H. K., C. T. Russell, K. Torkar, T. L. Zhang, and V. Angelopulos (2008), An advanced approach to finding magnetometer zero levels in the interplanetary magnetic field, Meas. Sci. Technol., 19(5), 055104, doi:10.1088/0957-0233/19/5/055104.

E00B20

Lepping, R. P., and K. W. Behannon (1980), Magnetic field directional discontinuities: 1. Minimum variance errors, J. Geophys. Res., 85, 4695 – 4703, doi:10.1029/JA085iA09p04695. Luhmann, J. G., S. A. Ledvina, and C. T. Russell (2004), Induced magnetospheres, Adv. Space Res., 33, 1905 – 1912, doi:10.1016/j.asr.2003.03.031. Martinecz, C., et al. (2008), Location of the bow shock and ion composition boundaries at Venus: Initial determinations from Venus Express ASPERA-4, Planet. Space Sci., 56, 780 – 784, doi:10.1016/j.pss. 2007.07.007. Ness, N. F., K. W. Behannon, R. P. Lepping, and K. H. Schatten (1971), Use of two magnetometers for magnetic field measurements on a spacecraft, J. Geophys. Res., 76, 3565 – 3573. Ness, N. F., M. H. Acuna, K. W. Bahannon, and F. M. Neubauer (1982), The induced magnetosphere of Titan, J. Geophys. Res., 87, 1369 – 1381, doi:10.1029/JA087iA03p01369. Podgorny, I. M., E. M. Dubinin, and P. L. Israelevich (1980), Laboratory simulation of the induced magnetospheres of Comets and Venus, Earth Moon Planets, 23, 323 – 338, doi:10.1007/BF00902047. Russell, C. T., R. C. Elphic, and J. A. Slavin (1979), Initial Pioneer Venus magnetic field results: Dayside observations, Science, 203, 745 – 748, doi:10.1126/science.203.4382.745. Russell, C. T., J. G. Luhmann, R. C. Elphic, F. L. Scarf, and L. H. Brace (1982), Magnetic field and plasma wave observations in a plasma cloud at Venus, Geophys. Res. Lett., 9, 45 – 48, doi:10.1029/GL009i001p00045. Saunders, M. A., and C. T. Russell (1986), Average dimension and magnetic structure of the distant Venus magnetotail, J. Geophys. Res., 91, 5589 – 5604, doi:10.1029/JA091iA05p05589. ¨ ., and L. J. Cahill (1967), Magnetopause structure and Sonnerup, B. U. O attitude from Explorer 12 observations, J. Geophys. Res., 72, 171 – 183, doi:10.1029/JZ072i001p00171. Titov, D. V., et al. (2006), Venus Express science planning, Planet. Space Sci., 54, 1279 – 1297, doi:10.1016/j.pss.2006.04.017. Zhang, T. L., J. G. Luhmann, and C. T. Russell (1991), The magnetic barrier at Venus, J. Geophys. Res., 96, 11,145 – 11,153, doi:10.1029/ 91JA00088. Zhang, T. L., et al. (2006), Magnetic field investigation of the Venus plasma environment: Expected new results, Planet. Space Sci., 54, 1336 – 1343, doi:10.1016/j.pss.2006.04.018. Zhang, T. L., et al. (2007), Little or no solar wind enters Venus’ atmosphere at solar minimum, Nature, 450, 654 – 656, doi:10.1038/nature06026. Zhang, T. L., et al. (2008a), Initial Venus Express magnetic field observations of the magnetic barrier at solar minimum, Planet. Space Sci., 56, 790 – 795, doi:10.1016/j.pss.2007.10.013. Zhang, T. L., et al. (2008b), Initial Venus Express magnetic field observations of the Venus bow shock location at solar minimum, Planet. Space Sci., 56, 785 – 789, doi:10.1016/j.pss.2007.09.012. 

H.-U. Auster, Institut fu¨r Geophysik und Extraterrestrische Physik, Technische Universita¨t, Mendelssohnstrasse 3, D-38106 Braunschweig, Germany. M. Balikhin, Department of Automatic Control and Systems Engineering, University of Sheffield, Mappin Street, Sheffield S1 3JD, UK. S. Barabash, Swedish Institute of Space Physics, Box 812, SE-98128 Kiruna, Sweden. W. Baumjohann, M. Delva, M. Volwerk, and T. L. Zhang, Space Research Institute, Austrian Academy of Sciences, A-8042 Graz, Austria. ([email protected]) K. Kudela, Institute of Experimental Physics, Slovak Academy of Sciences, Watsonova 47, 040 01 Kosˇice, Slovakia. C. T. Russell and H. Y. Wei, Institute of Geophysics and Planetary Physics, University of California, Los Angeles, CA 90095, USA. C. Wang, State Key Laboratory of Space Weather, Center for Space Science and Applied Research, Chinese Academy of Sciences, Beijing 100080, China.

9 of 9

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, E00B06, doi:10.1029/2008JE003148, 2008

for

Full Article

Proton cyclotron waves in the solar wind at Venus M. Delva,1 T. L. Zhang,1 M. Volwerk,1 Z. Vo¨ro¨s,2 and S. A. Pope3 Received 20 March 2008; revised 30 May 2008; accepted 24 June 2008; published 18 September 2008.

[1] Magnetometer data from two Venus years of the Venus Express mission in orbit are

investigated for the occurrence of ion cyclotron waves. Proton cyclotron waves were recently detected in the upstream region of Venus by the spacecraft, indicating pickup of planetary protons from Venus’s exosphere by the solar wind and loss of hydrogen to interplanetary space. A study of representative cases illustrates the waveform, spectrum, duration, and higher-order resonances of the transverse waves with left-hand circular polarization and propagation nearly along the magnetic field; their properties in the magnetic field principal axes system are determined. A statistical approach studies the wave properties as a function of the angle between the solar wind and magnetic field direction, as a function of their occurrence in space, and with respect to the motional solar wind electric field. Proton cyclotron waves are found up to 9 Venus radii from the planet, for a large range of angles between the solar wind and magnetic field direction, independent from foreshock geometry and independent from the direction of the motional electric field. This reveals that cyclotron wave generation from local pickup of neutral hydrogen is efficient over a large volume of space upstream of the planet and imposes the existence of an extended reservoir of planetary neutral hydrogen at Venus. Citation: Delva, M., T. L. Zhang, M. Volwerk, Z. Vo¨ro¨s, and S. A. Pope (2008), Proton cyclotron waves in the solar wind at Venus, J. Geophys. Res., 113, E00B06, doi:10.1029/2008JE003148.

1. Introduction [2] The existence of waves at the proton cyclotron frequency in the solar wind at Venus has long been a puzzling question. The absence of a Venus-intrinsic magnetic field leads to an induced magnetosphere around a small impenetrable obstacle and a bow shock at close distance to the planet. This leaves a substantial part of the upper neutral exosphere outside and upstream of the bow shock and thus directly accessible to the solar wind [Luhmann and Bauer, 1992]. Several processes can ionize the neutrals, mainly photoionization, electron impact ionization and charge exchange [Zhang et al., 1993]. The newborn ions form a secondary ion population in the solar wind, its interaction with the background plasma enables wave generation through different mechanisms; mainly scattering from a ring distribution in velocity space [Huddleston and Johnstone, 1992] or interaction of the new ‘‘beam’’ through the ion/ion beam plasma microinstability [Gary, 1991] generate waves at the cyclotron frequency in the plasma system. The cyclotron waves can be expected at any location where pickup in adequate quantity from the neutral atmosphere is possible, especially upstream of the bow shock. In contrast to waves 1 Space Research Institute, Austrian Academy of Sciences, Graz, Austria. 2 Institute of Astro- and Particle Physics, University of Innsbruck, Innsbruck, Austria. 3 Department of Automatic Control and Systems Engineering, University of Sheffield, Sheffield, UK.

Copyright 2008 by the American Geophysical Union. 0148-0227/08/2008JE003148$09.00

generated by back streaming solar wind ions which are confined to the foreshock region defined by the tangent magnetic field line to the bow shock, cyclotron waves can occur everywhere upstream [Mazelle et al., 2004]. Therefore, the observation of cyclotron waves in the upstream region is proof that the specific ion is present in the exosphere and is being lost to the solar wind and into interplanetary space. This is especially important for hydrogen at Venus, whose very low water and hydrogen content in comparison to Earth is not yet fully understood. [3] At Mars the situation is similar to Venus with absence of an intrinsic magnetic field and the solar wind accessing the upper exospheric layers outside of the bow shock directly. There, upstream proton cyclotron waves were reported already from the Phobos-2 spacecraft data [Russell et al., 1990]. From the Mars Global Surveyor magnetometer data extensive confirmation was obtained and a thorough study of the wave properties was made [Brain et al., 2002]. The proton cyclotron waves are concentrated near the subsolar point and at the flanks of the solar wind interaction region, outside the bow shock in the unperturbed solar wind, and found up to a distance of at least 12 planetary radii on the sides of the planet; they are left-hand elliptically polarized and propagate at a small angle with respect to the ambient magnetic field direction. The waves originate from the pickup process of protons from the hydrogen corona of Mars, which extends well beyond the Martian bow shock [Barabash et al., 1991]. An overview of all observed cyclotron wave features at Mars is given by Mazelle et al. [2004]. [4] Ion cyclotron waves can also occur at planetary satellites, they were reported at Io and Europa within the Jovian

E00B06

1 of 12

E00B06

DELVA ET AL.: UPSTREAM PROTON CYCLOTRON WAVES AT VENUS

magnetosphere [Huddleston et al., 1997; Volwerk et al., 2001]. The same situation occurs for pickup in the environment of active comets, as was observed at, e.g., Comet Halley [Johnstone et al., 1987; Glassmeier et al., 1989]. [5] At Venus, ion cyclotron waves from pickup protons as the lightest species have been reported from Pioneer Venus Orbiter (PVO) data in the magnetosheath, but never in the upstream region [Russell et al., 2006], although they are expected there too from the existence of a Venus hydrogen corona [Nagy et al., 1990]. Recently, the first observations of upstream cyclotron waves were reported [Delva et al., 2008] from the magnetometer MAG [Zhang et al., 2006] aboard the Venus Express spacecraft. This proof of existence of upstream proton cyclotron waves at Venus resolves the question why they would not be there, and adds a point of similarity to the plasma environment at Mars and Venus. [6] In this paper, we present a detailed study of the proton cyclotron waves observed by the magnetometer MAG during the first two Venus years of the Venus Express mission. The typical wave properties are discussed for several representative examples. For the long-term observations a statistical approach was performed, with emphasis on the wave properties and on their spatial occurrence.

2. MAG Instrumentation and Analysis Methods [7] The magnetometer consists of two sensors, mounted at different distances from the spacecraft body and operating synchronously. The dual sensor configuration enables separation of the spacecraft generated time-dependent fields from the ambient space field [Zhang et al., 2008; Leinweber et al., 2008]. Data are sampled over the entire 24 h orbit, delivering 1 Hz data over 22 h and 32 Hz data over 2 h around pericenter. The polar orbit with its pericenter at 78° north at an altitude of
260 km leaves the spacecraft most of the time in the solar wind and therefore in optimal position to detect the upstream waves. [8] The magnetometer is operated continuously and longterm 1 Hz data are now available, mainly without data gaps. We used the data from 10 May 2006 through 10 August 2007 making up two Venus years, i.e., two full revolutions of the spacecraft orbital plane around Venus’s polar axis; these are two revolutions in the Venus solar orbital (VSO) coordinate system (centered at Venus, xVSO axis toward the Sun, zVSO axis perpendicular to Venus’s orbital plane and positive to ecliptic north, yVSO axis completing the righthand system), which provides rotationally uniform data coverage. [9] Theoretically, pickup ions can generate different ULF waves including resonant and nonresonant instabilities. The resonant instabilities are the most likely since they have usually the highest linear growth rates. From linear theory, the highest growth rates are predicted for propagation parallel to the magnetic field, either the right-hand or the left-hand circular polarized mode [Gary, 1991; Brinca, 1991]. [10] In the solar wind frame, a pickup ion with gyrofrequency Wi (Wi = q/m B; mass m, charge q, magnetic field strength B) and drift velocity V// along the magnetic field B will be in resonance with a solar wind wave of frequency w and wave vector k if [Brinca, 1991]: w  k  V == ¼  n Wi

ð1Þ

E00B06

where the prevailing case is the fundamental resonance for n = 1. Resonance occurs mainly with the right-hand mode (minus sign) in the plasma frame, through the ion/ion righthand resonant instability with dispersion. The resonant solar wind wave parallel to B and with frequency w will be observed in the spacecraft system with a Doppler-shifted frequency: wsc ¼ w þ k==  V SW

ð2Þ

Combination of both equations for the fundamental resonance and taking into account that the planetary ion’s injection velocity into the plasma system is VSW (neglecting any original velocity, being small with respect to the solar wind velocity) we obtain for the frequency observed at the spacecraft [Brinca, 1991; Gary, 1991]: wsc þ Wi

ð3Þ

which is independent from the pitch angle a (VSW, B); the wave generation mechanism is independent from the solar wind electric field E = Vsw B [Gary, 1991]. [11] Therefore, the waves will be observed at the local ion gyrofrequency in the spacecraft frame and with specific left-hand polarization due to the anomalous Doppler effect [Mazelle and Neubauer, 1993]. This fact immediately excludes confusion with ULF waves generated by solar wind protons back streaming from the bow shock. It is known from numerous studies of foreshock ULF waves at Earth, both observational [e.g., Paschmann et al., 1979] and theoretical [e.g., Gary et al., 1981] that these particles have streaming velocities very different from VSW; they generate waves at frequencies different from the gyrofrequency, which are observed at the spacecraft at frequencies much lower than the cyclotron frequency. Studies of the ion ring beam in velocity space generated by the pickup process showed that the increase of observed wave power is at a frequency just below the local ion gyrofrequency. [12] The MAG data are investigated for enhanced wave power at or just below the proton cyclotron frequency fp fp ¼ ð2pÞ1 q=mB ¼ kB:

ð4Þ

The magnetic field at Venus in the range 5 – 40 nT and the MAG data accuracy of ±DB = ±1 nT in the total field value lead to an accuracy in the proton cyclotron frequency of ±0.015 Hz in the frequency range 0.076 – 0.610 Hz. [13] After transferring the VSO data to the magnetic principal axes (PA) system, where the wave oscillations occur in the (PAx, PAy) plane, power spectra are calculated and analysis of the waves regarding ellipticity, polarization and direction of propagation with respect to the mean magnetic field is performed [McPherron et al., 1972]. [14] The position of the wave observations with respect to their possible connection along a field line to the bow shock is also investigated. Since no ion data are yet available, a static model bow shock from a fit of MAG observations of the bow shock crossings [Zhang et al., 2008] was used and the tangential field line determined to see if the spacecraft is

2 of 12

E00B06

DELVA ET AL.: UPSTREAM PROTON CYCLOTRON WAVES AT VENUS

E00B06

Figure 1. Proton cyclotron waves observed on 6 July 2006 (0040 to 0050 UT) at solar zenith angle
100°. (a) Waveform in Venus solar orbital coordinates showing three components and total field. These are measured data without any filtering. (b) Compressional power and left- and right-hand polarized transverse power as a function of frequency. Vertical dashed line indicates local proton cyclotron frequency (fp = 0.15 Hz), and horizontal bar denotes (fp ± Dfp). The strong peak in the left-hand power at left of this line is due to locally generated proton cyclotron waves. located inside or outside (upstream) of the model foreshock region, i.e., was magnetically connected with the model bow shock or not. Distinction of proton cyclotron waves from waves generated by back streaming particles from the

bow shock is possible from two characteristics. First, the latter occur only within the foreshock. Second, waves generated by back streaming particles will be observed at

3 of 12

E00B06

DELVA ET AL.: UPSTREAM PROTON CYCLOTRON WAVES AT VENUS

E00B06

Figure 2. Wave properties in principal axes coordinates for proton cyclotron waves on 21 January 2007. (a) Waveform in principal axes coordinates, (Bx and Bx)PA components for 40 s of time. (b – d) Hodogram in each plane of the principal axes coordinate system for the same time of approximately four cycles. Markers are every 20 s, starting with 1 and in increasing order with time. The wave is clearly plane and nearly circular in (x, y)PA plane. the spacecraft at frequencies much lower than the cyclotron frequency [Gary, 1991].

3. Representative Proton Cyclotron Wave Cases [15] Analysis of the data of
450 orbits of Venus Express shows enhanced power in the transverse, left-hand polarized waves at or just below the proton cyclotron frequency many times. To enable better insight in the characteristic properties of PCWs, we discuss representative examples which illustrate specific interesting aspects of the waves. 3.1. Waveform and Spectrum [16] As a typical example for the waveform and the spectrum, we discuss an observation on 6 July 2006 (0040 to 0050 UT) at a solar zenith angle (SZA) of 100°. The VSO data are displayed in Figure 1a; the mean field B
10.2 nT yields a calculated local proton frequency fp = 0.15 Hz, the waveform with period
9 s is clearly seen (fpobs
0.11 Hz), overlaid with a slow modulation of
5 min period. The properties in the magnetic principal

axes system are as follows: polarization
60%, ellipticity
0.55 (minus for left-hand polarization); propagation with small angle to the mean magnetic field q (k, B)
6°. The spacecraft was located inside the modeled foreshock on the side of the parallel shock, the cone angle was q (xvso, B)
149°. [17] The spectrum in Figure 1b shows the compressional power, and left- and right-hand polarized transverse components; the strong enhancement of the power just below the calculated proton cyclotron frequency is only seen in the left-hand transverse component. The timely evolution of the enhanced left hand power is shown as dynamical spectrum for another example further below. 3.2. Properties in the Principal Axes System [18] A second example on 21 January 2007 illustrates that the waves are generally plane in the transverse direction to the magnetic field and almost circular. In Figure 2 the two transverse components in the principle axes system are shown as a function of time (Figure 2a), for a time span of about 4 cycles (0624:40 to 0625:20 UT). Figures 2b– 2d

4 of 12

E00B06

DELVA ET AL.: UPSTREAM PROTON CYCLOTRON WAVES AT VENUS

display hodograms for the same time interval; the wave is plane in the (PAx, PAy) plane and clearly left-handed polarized. [19] The mean values over the 10 min time interval (0620 to 0630 UT) for the wave properties are as follows: ellipticity 0.69, polarization 78%, angle of propagation to the mean magnetic field direction q (k, B)
8°, ratio of eigenvalues l1/l2 = 1.34, l2/l3 = 16.3. This again illustrates the left-hand nearly circular polarization of the plane but transverse waves. 3.3. Duration [20] The proton cyclotron wave (PCW) duration spans from a few minutes (5– 10 min) to more than 1 h (maximum observed
1:30 h) in the considered period of two Venus years. The main part of the events (75%) is rather short (10 min), with a mean value of 16 min. An example of long and persistent wave occurrence of 1:20 h duration on 11 June 2006 (2240 to 2400 UT) is shown in Figure 3a as dynamic spectrum. Only part of the time interval is shown in VSO coordinates in Figure 3b, to make the waveform more visible. The spacecraft was at SZA
73° to 70° and approached Venus from altitude 6 to 4 Venus radii (RV) during the time interval, it was located in the (model) foreshock for cone angle q (xvso, B)
145°. 3.4. Observation of Higher-Order Resonances [21] On several occasions not only the fundamental resonance (n = 1) is observed, but also higher-order harmonics. Similar observations were reported from the Giotto spacecraft magnetometer data at Comet Giacobini-Zinner [Glassmeier et al., 1989]; these authors state that harmonics are seen less frequently for nearly perpendicular configurations or cone angle q (xvso, B)
90°. [22] In Figure 4a the spectrum is displayed for 30 July 2006 (2210 to 2220 UT); the fundamental resonance frequency is seen as peak in the left-hand polarized power at the proton cyclotron frequency (fp = 0.95 Hz) and a second peak occurs at the first higher-order resonance (fp1 = 1.90 Hz). In Figure 4b, in the dynamic spectrum the enhanced power is clearly visible at both frequencies in Figure 4b (top) for the left-hand power. The magnetic field had a nearly antiparallel configuration with the solar wind direction or cone angle q (xvso, B)
20°, the waves propagate at an angle q (k, B)
16° to the mean magnetic field direction; the spacecraft was at an altitude of
7 RV and SZA = 117°, not located in the modeled foreshock; the bow shock inbound crossing was observed on the next day at 31 July 2006 0043 UT.

4. Statistical Approach for Long-Term Observations [23] To gain an overview of the PCW occurrences for the long time span of two Venus years, a statistical approach was applied. For each 24 h orbit of the spacecraft, 4 h of data before and after the model bow shock crossing [Zhang et al., 2008], corresponding to a maximal distance of 9 RV from the Venus-Sun line, were investigated in intervals of 10 min. For each time interval the frequency fp and the error range Dfp = k (s (B) + DB) were determined from the mean total field m (B), its standard deviation s (B) and the data

E00B06

accuracy DB. In the magnetic principal axes system, power spectra were calculated and the power per component was integrated in the frequency interval [0.8 (fp  Dfp), fp + Dfp], in order to account for power maxima just below the calculated cyclotron frequency. Detection of the waves was done on basis of enhanced power in the given frequency interval. PCWs are characterized by a larger power transverse to the mean magnetic field and a strong left-hand circular polarization of this component; selection was performed on the criteria PTransverse =PCompressional > 1:5 and PLeft =PRight > 1:5 Ellipticity < 0:5

ð5Þ

where PCompname denotes the power in the respective spectrum component. Only observations lying far enough outside of the model bow shock crossing were considered, at least 10 min apart from the crossing time. These criteria and time selection ensure that only upstream waves and with the typical properties expected from PCWs are taken into account. [24] In the two Venus years of data (
450 orbits) from numerous wave occurrences, only 153 PCW occurrences from 10 min intervals fulfilled our criteria; this corresponds to 1% of the investigated time intervals. It has to be pointed out that the use of the automated approach with fixed time setting of 10 min intervals and selection according to (5) reduces the number of selected ‘‘good’’ samples. If a wave occurrence happens near the edge of the 10 min time interval or overlaps over two intervals, its integrated power components may not fulfill the criteria, although careful selection of the integration time would identify the event as PCW. Therefore, the automated approach will systematically underestimate the number of PCW occurrences taken into account. However, a precise selection of time intervals and generation of spectra by hand is not feasible for the time span of 450 8 h under investigation. [25] We now discuss the occurrence of the PCWs as a function of the most significant parameters for the generation mechanism of the waves, leaving other possible investigations beyond the scope of this paper. 4.1. Statistics of PCW Properties [26] Median values for the main characteristic wave parameters are listed in Table 1. The amplitudes are generally moderate (median amplitude
0.56 nT) and the ratio amplitude to B is small. The waves have much more power in the transverse component (PT/PC
4.34) and the lefthand component is severely stronger (PL/PR
3.9). The angle of propagation with the mean magnetic field is generally small q (k, B)
13°, and polarization and ellipticity are strong. [27] These values are characteristics of proton cyclotron waves [Mazelle et al., 2004; Bertucci et al., 2005] and prove that they are generated from pickup of local neutral hydrogen. From the properties listed in Table 1, only the amplitude and amplitude/B show a dependence on the solar zenith angle with larger values and more variability for larger SZA (see Figure 6), which is in accordance with the observations at Mars [Brain et al., 2002]. The other prop-

5 of 12

E00B06

DELVA ET AL.: UPSTREAM PROTON CYCLOTRON WAVES AT VENUS

Figure 3. (a) Dynamic spectrum for waves on 11 June 2006 with duration of 1:20 h. The spacecraft was at solar zenith angle
72° and approached Venus from 7 to 5 RV in the time interval (2240 to 2400 UT). White line indicates local proton cyclotron frequency fp. Enhanced power near fp is seen only in left-hand polarized component. (b) Waveform in VSO coordinates of the event on 11 June 2006. Only a limited part of the wave occurrence is shown (2300 to 2340 UT) to enable clear view to the waveform. 6 of 12

E00B06

E00B06

DELVA ET AL.: UPSTREAM PROTON CYCLOTRON WAVES AT VENUS

Figure 4. (a) Observation of cyclotron waves at the fundamental resonance and first harmonics on 30 July 2006, at distance of 8 RV and SZA
115°. Dashed line and horizontal bar are as in Figure 1b. The vertical arrows indicate the peak in the power at these frequencies. (b) Dynamic spectrum for waves on 30 July 2006. The white line indicates local proton cyclotron frequency fp; enhanced power near fp and the first harmonics fp1 is observed in the left-hand polarized component.

7 of 12

E00B06

DELVA ET AL.: UPSTREAM PROTON CYCLOTRON WAVES AT VENUS

E00B06

Table 1. Statistics of PCW Propertiesa Wave Property Amplitude (nT) Amplitude/B PT/PC PL/PR Ellipticity Polarization (%) Propagation angle q (k, B) (deg) Eigenvalue ratio (l1/l2) Eigenvalue ratio (l2/l3)

Median First Quartile Third Quartile 0.56 0.06 4.34 3.92 0.61 63.40 13.08 1.49 6.93

0.32 0.04 2.91 2.98 0.56 55.33 8.56 1.35 5.05

0.82 0.08 6.43 6.01 0.67 72.49 19.35 1.66 9.70

a

ns = 153.

erties are rather uniformly distributed (not shown) in the observed range of SZA (20° to 140°). 4.2. PCW Occurrence as a Function of Cone Angle [28] An important question is, if there is any dependence of PCW occurrence on the angular configuration of Vsw and B. Equation (3) for the local proton cyclotron frequency in the spacecraft frame is independent from the angle q (Vsw, B) and we principally can expect to observe PCWs at all cone angles q (xvso, B). From the theory, we know that PCWs are generated as soon as a large enough number of newborn ions form a secondary ion population in the solar wind which interacts with the background plasma, where two different mechanisms are effective with a gradual transition between them [Huddleston and Johnstone, 1992; Gary, 1991]. To investigate this, we consider the injection of the planetary ions into the solar wind flow in velocity space. Neglecting any (small) original velocity of the neutral particle, in the plasma frame the ion has an

E00B06

injection velocity Vinj = Vsw, which is in fact in direction of xVSO. Figure 5 shows Vinj in velocity space, where the x axis is parallel to the magnetic field direction, i.e., decomposition into components parallel and perpendicular to the local mean field B; the angle from the B direction to the injection velocity Vinj is equal to the cone angle q (xvso, B). [29] In rather parallel or antiparallel configurations of Vsw and B (cone angle < 40° or > 140°) the ion’s velocity component V? in the plane vertical to the B direction is small, the ion/ion beam plasma microinstability for (anti-) parallel beams is very efficient and strong waves can be expected [Gary, 1991]. Indeed, the largest amplitudes are observed for this range of cone angles (Figure 5). [30] In more oblique cases the ions gyrate with V? in the plane vertical to B and form a ring distribution. This ion ring is unstable and scattering from the ring to a shell distribution takes place, supplying the energy for cyclotron wave generation. However, the time scale for scattering and wave generation is longer than in case of the more effective (anti-) parallel beam mechanism. Furthermore, Huddleston and Johnstone [1992] have shown that for ring distributions and small ratio VA/VSW, where VA is the Alfve´n velocity, cyclotron wave generation is less efficient for approximately perpendicular configurations or pitch angles q (Vinj, B)
90°. At Venus, using nominal solar wind conditions VSW = 400 km s1, VA = 55 km s1 (for B = 8 nT and solar wind proton number density npSW = 10 cm3) we obtain VA/VSW
0.1375 and the condition for more efficient wave generation at smaller pitch angles is fulfilled. This means that we expect less and weaker waves for quasiperpendicular cases of the solar wind velocity and magnetic field and stronger waves for oblique configurations.

Figure 5. Injection velocity of the pickup ion into the solar wind flow for all PCWs observed from 10 May 2006 to 10 August 2007, in velocity space in the plasma frame. The x axis is in direction of the magnetic field B; colors are according to wave amplitude. The inner part shows the distribution of observations with cone angle q (xvso, B); no PCWs are observed for cone angle
90°, more and stronger waves for cone angles corresponding to the prevailing Parker spiral angles. 8 of 12

E00B06

DELVA ET AL.: UPSTREAM PROTON CYCLOTRON WAVES AT VENUS

E00B06

Figure 6. Positions of observations of PCWs from 10 May 2006 to 10 August 2007 in cylindrical VSO coordinates. Colors are according to wave amplitude, and black circles denote observations upstream of the foreshock region. Gray lines denote approximately the limiting orbits of the spacecraft; regions outside of these lines were not accessed. [31] The histogram in the center of Figure 5 displays the PCW observations as a function of the cone angle in bins of 20°; the lack of observations for cone angle q (xvso, B)
90° is obvious. In the intervals [20°, 40°] and [140°, 160°], for rather antiparallel or parallel configurations of the solar wind velocity and B, more and stronger PCWs are observed; this corresponds also to the prevailing magnetic field orientation of the Parker spiral at Venus. 4.3. Spatial Distribution of PCW Observations [32] The spatial occurrence of the PCWs is displayed in Figure 6 in cylindrical VSO coordinates: the ordinate is the distance from the Venus-Sun line, color coding is according to amplitude, black surrounding circles indicate PCWs observed outside the modeled foreshock region; the gray lines are the approximate limiting orbits of the spacecraft; regions outside of the gray lines were not scanned by the spacecraft. Several ‘‘trains’’ of observation points indicate observations during longer times, i.e., subsequent intervals of 10 min. [33] 1. From Figure 6, we see that more observations occur at larger SZA. This is as expected since the waves are swept downstream by the solar wind; these waves are not

observed at the position of their generation, which lays further upstream, but have propagated along the field lines in the flowing solar wind plasma. This is in accordance with observations at Mars [Brain et al., 2002]. [34] 2. Furthermore, larger amplitudes are observed closer to the bow shock and/or toward larger SZA. Again, this is in accordance with predictions: stronger waves occur for higher pickup ion density, i.e., closer to the planet, or if there was time for wave growth during propagation to larger SZA. Similar effects were reported from observations at Mars [Brain et al., 2002]. [35] 3. The observations in or outside of the modeled foreshock region show no difference in spatial distribution, proving that the waves are not a foreshock-related phenomenon. The discrepancy in sample size in and out of the foreshock (ns_in = 93; ns_out = 60) is an observational effect due to the orbit of the spacecraft. The pericenter of the orbit lays at 78° north in the VSO coordinate system and for zVSO > 1 RV the spacecraft is always close to the planet and therefore mostly in the foreshock region, because of the Parker spiral angle of the IMF (37° or 143°) at Venus. Moreover, this region is scanned during every orbit, which

9 of 12

E00B06

DELVA ET AL.: UPSTREAM PROTON CYCLOTRON WAVES AT VENUS

E00B06

Figure 7a. Distribution of observed wave energy as a function of altitude above the Venus surface. Colors are according to wave amplitude. Weak waves occur from low up to high altitudes. explains more observations within the foreshock. The region further away from the line Venus-Sun is scanned only for zVSO < 1 RV and here the spacecraft can be in or outside of the foreshock, depending on the cone angle q (xvso, B). [36] 4. In Figure 7a the wave energy is shown as a function of the altitude above the Venus surface. Waves with low energy occur at all altitudes from close to Venus up to 8 RV; more energetic waves are confined to lower altitudes, between 2 and 5 RV.

field direction for cone angle < 90°, more PCWs in negative E field regions for cone angle > 90°) is an observational effect due to the orbit of the spacecraft with pericenter around zVSO 1 RV and PCW observations mainly for zVSO < 0. We therefore conclude that no asymmetry effect is observed in the PCW distribution for positive or negative motional electric field directions; this can be explained by the small gyroradius of pickup hydrogen with respect to the size of the planet.

4.4. PCW Observations as a Function of Motional Electric Field [37] We now investigate the PCW observations as a function of the direction of the motional electric field E = Vsw B. We transform the positions of PCW observations from the VSO system to a local electromagnetic coordinate system, where xEM is positive in direction of Vsw, the (x, y)EM plane contains Vsw and B and zEM is positive in direction of the motional electric field. Figure 7b (top) shows the positions in zEM as a function of the cone angle: waves occur up to large positive and negative zEM values or large distances on either side of the (Vsw, B) plane; again, observations are lacking for cone angle q (xvso, B)
90°. In Figure 7b (bottom), the histogram of observations is separated into regions with positive or negative E field; an approximately equal number of observations is in negative (44%) as in positive direction (56%), so no preferential effect for positive E field direction is found. The unequal distribution (more PCWs in positive E

5. Discussion [38] The studied cases of observed PCWs give insight into the main characteristics of the waves. For the fundamental resonance, the observed properties are typical for what is known from cyclotron waves at other planets. In addition to the fundamental resonance, also the first harmonics is observed in several cases. The duration of the cyclotron wave occurrences ranges from some min (5 to 10) to more than 1 h, the mean value is 16 min. We expect that solar wind changes cause this variability, which needs further investigation but is beyond the scope of this paper. Similar and even longer durations have been reported for Mars from long-term Mars Global Surveyor (MGS) observations, with no dependence on the cone angle [Brain et al., 2002; Mazelle et al., 2004; Bertucci et al., 2005]. However, for some specific MGS orbits, more short duration cyclotron wave occurrences were reported by Wei and Russell [2006]. The authors explain this as result of an asymmetry effect in direction of positive electric field due to secondary ioniza-

10 of 12

E00B06

DELVA ET AL.: UPSTREAM PROTON CYCLOTRON WAVES AT VENUS

E00B06

Figure 7b. (top) Distribution of zEM component of PCW observation positions in electromagnetic coordinate system as a function of cone angle q (xvso, B). (bottom) Histogram of zEM component, separated for positive and negative direction of motional electric field. A comparable number of cases is observed for each direction. tion from a disk of fast reneutralized hydrogen near and parallel to the (Vsw, B) plane, both downstream and to the side of Mars. From the observations at Venus presented in this paper, with also long duration of the PCWs and no such asymmetry, the suggested mechanism is not confirmed; rather the here observed PCWs are generated at initial ionization of local planetary hydrogen at any position around Venus. [39] The statistical approach for the long time interval uses severe criteria, which leads to a selection of 153 cases of PCWs corresponding to wave occurrence during 1% of the investigated time, but with systematical underestimation of the number of PCW occurrences. The statistical properties derived from the data set fulfill the general requirements for cyclotron waves. PCWs are observed mainly for nearly parallel or antiparallel configurations of the magnetic field and solar wind direction, with no observations for perpendicular conditions; this is in accordance with theoretical predictions.

[40] The waves are observed up to high altitudes (8 RV) from the planet and also up to 4 RV in direction toward the Sun, as far as the spacecraft scanned the sunward side of the planet. No asymmetry is found with respect to positive or negative directions of the motional electric field. This is important, because PCWs propagate generally (anti-) parallel to the B field lines [Mazelle et al., 2004] and pickup ions can only move in direction of positive motional electric field; no mechanism is known to move ions across the magnetic field lines or against the electric field. Therefore the waves can be only generated by local pickup also for large distances on the negative side of the (Vsw, B) plane and a sufficient density of local neutral hydrogen must be available. A detailed study of the consequences of these findings for the neutral hydrogen exosphere at Venus is made by M. Delva et al. (An extended neutral hydrogen exosphere at Venus, submitted to Geophysical Research Letters, 2008), giving escape rates of 1.4 1023 s1 up to

11 of 12

E00B06

DELVA ET AL.: UPSTREAM PROTON CYCLOTRON WAVES AT VENUS

5.6 1023 s1 and claiming that an extended neutral hydrogen exosphere must exist at Venus.

6. Conclusions [41] The MAG data from the two first Venus years of the Venus Express mission in orbit (May 2006 to August 2007) were analyzed for the occurrence of proton cyclotron waves in the solar wind. Proton cyclotron waves are important as precursor of the approaching planet and indicator of pickup from local neutral planetary hydrogen and loss of the ions to the solar wind. The general properties of the waves were found to fulfill all the criteria as known from upstream cyclotron waves at other planets or comets. The observations indicate that wave generation takes place at initial ionization of planetary hydrogen, by the ion/ion beam plasma microinstability for (anti-) parallel beams or by scattering from a ring distribution for more oblique configurations of the solar wind velocity and magnetic field direction, and a gradual transition between both mechanisms. A statistical approach showed that the waves are present up to large distances (9 RV) from the planet and exhibit no asymmetry with respect to the motional electric field direction, proving that pickup of planetary hydrogen takes place everywhere upstream of the bow shock in a large volume of space. This leads to a permanent escape of hydrogen from the extended environment of the planet, which may have a significant impact on the overall evolution of the water content of Venus over the age of the Solar System. [42] Acknowledgments. The authors thank all members of the MAG team for development of the magnetometer with its outstanding performance and accuracy, the ASPERA-4 team for valuable discussions, and K. Kudela (University of Kosicˇe, Slovakia) and W. Zambelli (IWF Graz, Austria) for assistance in preparation of the VSO data. The work by Z.V. is supported by the Austrian Wissenschaftsfonds under grant P20131-N16.

References Barabash, S., et al. (1991), Picked up protons near Mars: Phobos observations, Geophys. Res. Lett., 18, 1805 – 1808, doi:10.1029/91GL02082. Bertucci, C., C. Mazelle, and M. Acuna (2005), Interaction of the solar wind with Mars from Mars Global Surveyor MAG/ER observations, J. Atmos. Sol. Terr. Phys., 67, 1797 – 1808, doi:10.1016/j.jastp.2005.04.007. Brain, D. A., et al. (2002), Observations of low-frequency electromagnetic plasma waves upstream from the Martian shock, J. Geophys. Res., 107(A6), 1076, doi:10.1029/2000JA000416. Brinca, A. L. (1991), Cometary linear instabilities: From profusion to perspective, in Cometary Plasma Processes, Geophys. Monogr. Ser., vol. 61, edited by A. P. Johnstone, pp. 211 – 221, AGU, Washington, D. C. Delva, M., et al. (2008), First upstream proton cyclotron wave observations at Venus, Geophys. Res. Lett., 35, L03105, doi:10.1029/2007GL032594. Gary, P. (1991), Electromagnetic ion/ion instabilities and their consequences in space plasmas: A review, Space Sci. Rev., 56, 373 – 415, doi:10.1007/BF00196632. Gary, S. P., J. T. Gosling, and D. W. Forslund (1981), The electromagnetic ion beam instability upstream of the Earth’s bow shock, J. Geophys. Res., 86, 6691 – 6696, doi:10.1029/JA086iA08p06691.

E00B06

Glassmeier, K. H., et al. (1989), Spectral characteristics of low-frequency plasma turbulence upstream of Comet Halley, J. Geophys. Res., 94, 37 – 48, doi:10.1029/JA094iA01p00037. Huddleston, D. E., and A. D. Johnstone (1992), Relationship between wave energy and free energy from pickup ions in the Comet Halley environment, J. Geophys. Res., 97, 12,217 – 12,230, doi:10.1029/92JA00726. Huddleston, D. E., et al. (1997), Ion cyclotron waves in the Io torus during the Galileo encounter: Warm plasma dispersion analysis, Geophys. Res. Lett., 24, 2143 – 2146, doi:10.1029/97GL01203. Johnstone, A. D., et al. (1987), Waves in the magnetic field and solar wind flow outside the bow shock at Comet Halley, Astron. Astrophys., 187, 47 – 54. Leinweber, H. K., C. T. Russell, K. Torkar, T. L. Zhang, and V. Angelopoulos (2008), An advanced approach to finding magnetometer zero levels in the interplanetary magnetic field, Meas. Sci. Technol., 19, 055104, doi:10.1088/0957-0233/19/5/055104. Luhmann, J. G., and S. Bauer (1992), Solar wind effects on atmosphere evolution at Venus and Mars, in Venus and Mars: Atmospheres, Ionospheres and Solar Wind Interactions, Geophys. Monogr. Ser., vol. 66, edited by J. G. Luhmann, M. Tatrallyay, and R. O. Pepin, pp. 417 – 430, AGU, Washington, D. C. Mazelle, C., and F. M. Neubauer (1993), Discrete wave packets at the proton cyclotron frequency at comet P/Halley, Geophys. Res. Lett., 20, 153 – 156, doi:10.1029/92GL02613. Mazelle, C., et al. (2004), Bow shock and upstream phenomena at Mars, S p a c e S c i . R e v . , 11 1 ( 1 – 2 ) , 1 1 5 – 1 8 1 , d o i : 1 0 . 1 0 2 3 / B:SPAC.0000032717.98679.d0. McPherron, R. L., C. T. Russell, and P. J. Coleman (1972), Fluctuating magnetic fields in the magnetosphere, II, ULF waves, Space Sci. Rev., 13, 411 – 454, doi:10.1007/BF00219165. Nagy, A. F., J. Kim, and T. E. Cravens (1990), Hot hydrogen and oxygen atoms in the upper atmospheres of Venus and Mars, Ann. Geophys., 8, 251 – 256. Paschmann, G., N. Sckopke, S. J. Bame, J. R. Asbridge, J. T. Gosling, C. T. Russell, and E. W. Greenstadt (1979), Association of low frequency waves with suprathermal ions in the upstream solar wind, Geophys. Res. Lett., 6, 209 – 212, doi:10.1029/GL006i003p00209. Russell, C. T., et al. (1990), Upstream waves at Mars: Phobos observations, Geophys. Res. Lett., 17, 897 – 900, doi:10.1029/GL017i006p00897. Russell, C. T., S. S. Mayerberger, and X. Blanco-Cano (2006), Proton cyclotron waves at Mars and Venus, Adv. Space Res., 38, 745 – 751, doi:10.1016/j.asr.2005.02.091. Volwerk, M., M. G. Kivelson, and K. K. Khurana (2001), Wave activity in Europa’s wake: Implications for ion pickup, J. Geophys. Res., 106, 26,033 – 26,048, doi:10.1029/2000JA000347. Wei, H. Y., and C. T. Russell (2006), Proton cyclotron waves at Mars: Exosphere structure and evidence for a fast neutral disk, Geophys. Res. Lett., 33, L23103, doi:10.1029/2006GL026244. Zhang, M. H. G., J. G. Luhmann, A. F. Nagy, J. R. Spreiter, and S. S. Stahara (1993), Oxygen ionization rates at Mars and Venus: Relative contributions of impact ionization and charge exchange, J. Geophys. Res., 98(E2), 3311 – 3318, doi:10.1029/92JE02229. Zhang, T. L., et al. (2006), Magnetic field investigation of the Venus plasma environment: Expected new results from Venus Express, Planet. Space Sci., 54, 1336 – 1343, doi:10.1016/j.pss.2006.04.018. Zhang, T. L., et al. (2008), Initial Venus Express magnetic field observations of the Venus bow shock location at solar minimum, Planet. Space Sci., 56, 785 – 789, doi:10.1016/j.pss.2007.09.012. 

M. Delva, M. Volwerk, and T. L. Zhang, Space Research Institute, Austrian Academy of Sciences, A-8042 Graz, Austria. S. A. Pope, Department of Automatic Control and Systems Engineering, University of Sheffield, Sheffield S10 2TN, UK. Z. Vo¨ro¨s, Institute of Astro- and Particle Physics, University of Innsbruck, A-6020, Innsbruck, Austria.

12 of 12

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, E00B16, doi:10.1029/2008JE003154, 2008

for

Full Article

Mirror-mode-like structures in Venus’ induced magnetosphere M. Volwerk,1 T. L. Zhang,1 M. Delva,1 Z. Vo¨ro¨s,2 W. Baumjohann,1 and K.-H. Glassmeier3 Received 26 March 2008; revised 1 July 2008; accepted 5 September 2008; published 10 December 2008.

[1] The solar wind interaction with Venus creates an induced magnetosphere around the

planet. It is shown that within the space bound by Venus’ bow shock and ionopause, there is a rich occurrence of mirror-mode-like structures in the magnetic field data. The dayside magnetosheath and nightside magnetosheath/wake regions are investigated separately. It is shown that the probability to observe mirror mode structures is much higher at the dayside, where it is also strongly dependent on the angle between the solar wind magnetic field and the bow shock normal. In Venus’ wake the chance to observe these structures is low, most likely because of the fully developed turbulence in this region, which will decrease temperature anisotropies. The results stand in contrast to the very low occurrence rate claimed from data taken by the Pioneer Venus Orbiter mission. Citation: Volwerk, M., T. L. Zhang, M. Delva, Z. Vo¨ro¨s, W. Baumjohann, and K.-H. Glassmeier (2008), Mirror-mode-like structures in Venus’ induced magnetosphere, J. Geophys. Res., 113, E00B16, doi:10.1029/2008JE003154.

1. Introduction [2] The mirror instability occurs in plasmas with strong temperature anisotropies, i.e., plasmas in which the perpendicular temperature is higher than the parallel temperature [Hasegawa, 1969; Gary et al., 1993; Southwood and Kivelson, 1993]. These waves have been found in various objects, e.g., the Earth’s magnetotail [Rae et al., 2007], the magnetosheath of the Earth [e.g., Tsurutani et al., 1982; Baumjohann et al., 1999; Lucek et al., 1999a; Constantinescu et al., 2003], Mar’s [Bertucci et al., 2004] and of Jupiter [Erdo¨s and Balogh, 1993; Joy et al., 2006], in Saturn’s middle magnetosphere [Russell et al., 2006a], in cometary tails [e.g., Russell et al., 1987] and magnetic pile up boundaries [Glassmeier et al., 1993] and in the ion pickup region near Io [e.g., Huddleston et al., 1999]. Lately, Volwerk et al. [2008] have shown that mirror-mode-like structures appear at different locations of Venus’ magnetosheath, and Zhang et al. [2008] have shown that mirror mode waves also appear in the solar wind at Venus’ orbit. [3] The instability criterion for mirror mode (MM) waves, i.e., high-b plasma and T? > Tk, however, is shared with another wave mode, namely the ion cyclotron (IC) instability (see for IC waves at Venus Delva et al. [2008a]). Gary et al. [1993] have shown that the growth rate for IC is usually greater than for MM. Southwood and Kivelson [1993] have 1 Space Research Institute, Austrian Academy of Sciences, Graz, Austria. 2 Institute for Astro- and Particle Physics, University of Innsbruck, Innsbruck, Austria. 3 Institut fu¨r Geophysik und Extraterrestrische Physik, Technische Universita¨t, Braunschweig, Germany.

Copyright 2008 by the American Geophysical Union. 0148-0227/08/2008JE003154$09.00

proposed that the inhomogeneities in the background plasma, created by the MM, will inhibit the IC growth in planetary magnetosheaths. Also, the presence of He ions suppresses the growth rate for proton IC waves [Gary et al., 1993]. The MM instability generates compressional waves that grow preferentially in the direction perpendicular to the ambient magnetic field [see Treumannn and Baumjohann, 1996, section 3.5]. [4] At frequencies below the IC frequency in the plasma frame, the MM behaves in such a way that the perpendicular pressure p? of the plasma will be in antiphase with compressional variations in the magnetic field [Hasegawa, 1969]. In a bi-Maxwellian plasma with perpendicular temperature T? and parallel temperature Tk this means that
 T? dB dp? ¼ 2p? 1  ; Tk B

ð1Þ

which leads to the instability criterion
 T? < 0: 1 þ b? 1  Tk

ð2Þ

[5] The temperature asymmetry in the (Earth’s) magnetosheath can be created by two mechanisms [Lucek et al., 1999b]: (1) enhanced gyratory motion of the ions caused by (multiple) reflection(s) at the bow shock under quasiperpendicular conditions before entering the magnetosheath [shock drift acceleration [see, e.g., Kirk et al., 1994, chapter 2.2] or (2) compression of the magnetosheath close to the magnetopause, where the increased magnetic field strength leads to a gyration velocity increase through the first adiabatic invariant [see, e.g., Baumjohann and

E00B16

1 of 13

E00B16

VOLWERK ET AL.: VENUS’ MIRROR MODE WAVES

Treumannn, 1996, chapter 2.5]. Volwerk et al. [2008] presented two events in Venus’ magnetosheath which show that these two mechanisms create MM there. [6] From the case studies presented by Volwerk et al. [2008] it was found that the MM waves in Venus’ magnetosheath have a large amplitude, with variations in magnetic field strength sometimes greater than 50%. The waves were shown to be highly compressional (i.e., the maximum variance direction of the waves was almost aligned with the background field) and the propagation direction of the waves was perpendicular to the background field (i.e., the minimum variance direction of the waves was almost perpendicular to the background field). The duration of the events is only up to several minutes, instead of up to hours at Earth. At the same time the wave period of the MM waves is shorter, e.g., near the bowshock 5 s at Venus and 30 s at Earth. These two facts combined with the fact that Venus’ magnetosheath is approximately a factor 10 smaller than the Earth’s led to the empirical assumption that Venus’ magnetosheath may be similar to the Earth’s magnetosheat, only scaled down by a factor of 10. When the MM waves are present in the data they are highly compressional and propagating perpendicular to the ambient field direction. [7] In this paper, the case study paper presented by Volwerk et al. [2008] is extended to a full statistical study of the MM-like structures in Venus’ magnetosheath and magnetotail for 1 Venus year of data. The occurrence rate for these structures in Venus’ induced magnetosphere will be determined on the basis of the location of the spacecraft and it will be shown that also under quasi-parallel bow shock there is a chance for MM like structures to appear. The occurrence rate as a function of the angle between solar wind magnetic field and bow shock normal will be shown.

2. Venus Express [8] Magnetometer data from the Venus Express mission (VEX) [Svedhem et al., 2007] are used; the spacecraft is in a polar orbit around Venus with periapsis at 300 km and therefore will enter deeply into Venus’ induced magnetosphere, as shown by Zhang et al. [2007a]. [9] Magnetic field data from VEXMAG [Zhang et al., 2006] are used with a sampling rate of 1 Hz. During the nominal mission of VEX data are also available at 32 Hz sampling rate (and for short intervals also at 128 Hz). The plasma data from ASPERA [Barabash et al., 2007] have a resolution of 3 min for ions and 4 s for electrons. The MM waves were shown to have periods 4 T 15 s [Volwerk et al., 2008] which means that the ion plasma data cannot be used for identification of the waves on the basis of equilibrium between magnetic and plasma pressure. Yet, the plasma data could give information on the temperature asymmetry of the magnetospheric plasma. However, as this paper is written, plasma data with the required accuracy are not yet available.

3. Mirror Mode Identification [10] In the absence of high-resolution ion data, MM waves need to be identified from the magnetic field data only. This situation is similar to that by Lucek et al. [1999a, 1999b] for Equator-S, and in this paper the same identifi-

E00B16

cation method will be adopted and enhanced. MM waves are identified as having strengths DB/B 10%, and having small angles qBmv between the maximum variance and the magnetic field direction qBmv 30° [Price et al., 1986]. These two quantities are determined for sliding windows of 30 s width and 1 s shift. [11] To investigate the wave activity the mean magnetic field is determined first through low-pass filtering of the data, with a shortest period of 1.5 min. After this the data and the mean magnetic field are subtracted such that only the amplitude of the field oscillation is left. Within each 30 s window the amplitude of the waves DB in the data is then determined as twice the maximum difference between the data and mean field to get an upper estimate of the peak-topeak value of the waves. This means that the value for the strength of the MM-like structures, DB/B, can reach a maximum value of 2, and DB/B = 0.2  10% variation around the mean value. This method is used as the MM-like structures in Venus’ magnetosheath are not as nicely ordered as in the Earth’s magnetosheath. In the latter, it is clear what the base value of the magnetic field is from which the MM waves depart [see, e.g., Lucek et al., 1999a]. This is dependent on whether the MM waves are ‘‘dips’’ or ‘‘peaks’’ with respect to the background field, which are two different appearances for this instability. Califano et al. [2008] investigate this dip-peak phenomenon numerically and show that it depends on the plasma b and the distance of the system from the instability threshold. Joy et al. [2006] have shown that, in the Jovian magnetosheath, the dips occur mainly in the low plasma-b regions near the magnetopause and on the flanks, whereas the peaks show up mainly in the dayside high plasma-b regions in the middle magnetosheath. [12] For each window a minimum variance analysis [Sonnerup and Scheible, 1998] is performed and the angle qBmv between the maximum variance direction of the waves and the mean magnetic field is determined. Here we use a more strict directional requirement qBmv 20° for the selection of the compressional waves. Additionally, the angle fBmv between the minimum variance direction and the magnetic field is determined, which is expected to be nearly perpendicular for MM waves. As there is no possibility for the determination of pressure balance in the MM waves, because of the insufficient data resolution of the plasma instrument, an extra requirement on the structures is introduced, to make sure only MM-like structures are selected, here a limit is set at fBmv 80°.

4. Data [13] Before a statistical study of the MM waves is performed, it is useful to discuss one event in detail, observed on 5 May 2006, which is Volwerk et al.’s [2008] event 1. 4.1. One Quasi-Perpendicular Bow Shock Event [14] On 5 May 2006 VEX entered from the solar wind (SW) through the bow shock (BS) into the magnetosheath (MS). At the beginning of the event the spacecraft was located near (1.69, 0.05, 0.48) RV in Venus-Sun Orbital (VSO) coordinates implying a local time of 1200 (noon) and the solar wind magnetic field was BSW = (2.04, 4.73, 1.37) nT. The angle qBn between the solar wind

2 of 13

VOLWERK ET AL.: VENUS’ MIRROR MODE WAVES

E00B16

E00B16

Figure 1. The magnetic field data for 5 May 2006. (a –d) The magnetic field components Bx,y,z in VSO coordinates and magnitude Bm. The smooth curve shows the low-pass filtered (periods longer that 1.5 min) magnetic field, which is used as the baseline to determine the amplitude of the magnetic fluctuations. The spacecraft moves from the solar wind (SW), through the BS into the magnetosheath (MS). (e and f) The fluctuation of the magnetic field DB/B and the angle qBmv between the maximum variance direction and the mean magnetic field (dots), and the angle fBmv between the minimum variance direction and the mean magnetic field (pluses). The two grey shaded areas I and II show the intervals for which the MM wave condition is met (i.e., qBmv < 20, fBmv > 80 and DB/B large). magnetic field and the local shock normal is calculated by using the best fit for the BS, given by Zhang et al. [2007a] RBS ¼

2:169505 ; 1 þ 0:617330 cos g SZA

ð3Þ

where g SZA is the solar zenith angle. One can calculate the normal direction in the X - R plane through 0 @

nX nR

1

0

A¼B @

g Þ sin g  @RBS ð@g @RBS ðg Þ cos g @g

1

0 1 2:17 cos g þ 2:17 0:62 C @ A: A 2:17 sin g ð4Þ

[15] The angle g SZA is easily determined from the location of the spacecraft in the solar wind. The solar wind magnetic field is determined 30 min before VEX enters the bow shock. The R component of the normal is easily split into two components ny and nz through multiplication with cosd and sind respectively, where d is the angle between the positive y axis and the direction of the spacecraft in the YZ plane. [16] For this event it is found that qBn 106°, implying a quasi-perpendicular BS. Immediately after crossing the BS large amplitude compressional waves occur as shown in shaded region I in Figure 1. In Figures 1e and 1f DB/B and the angles qBmv and fBmv are shown. In Figure 1f, the angle qBmv drops well below 20°, indicating that the waves are

3 of 13

E00B16

VOLWERK ET AL.: VENUS’ MIRROR MODE WAVES

E00B16

Figure 2. All MM-like events plotted in cylindrical VSO coordinates. The thick solid line depicts Venus’ bow shock, and the dotted line depicts Venus’ magnetopause/ionopause, as determined by Zhang et al. [2007a]. mainly compressional. At the same time fBmv increases to well above 80°. This region shows compressional waves at a period of 5 s propagating perpendicular to the field. [17] Sightly later, 0117– 0118:30 UT, with the spacecraft deeper into the MS, there is another region (shaded region II in Figure 1) displaying compressional waves propagating almost perpendicular to the ambient magnetic field. These waves, in contrast to shaded region I, now have a period of 15 s. 4.2. Statistical Analysis [18] To perform a statistical analysis of MM-like structures in Venus’ magnetosphere, 1 Venus year of data (224 Earth days, from 24 April to 31 December 2006) is used from the VEX magnetometer. For this study which identifies the MM-like structures automatically, an extra limitation to the data has been put into the analysis software. Only MM-like structures that are observed for Bm 10 nT are included in the database, to avoid similar structures observed while the spacecraft is still in the solar wind [see, e.g., Zhang et al., 2008]. In Figure 2 all the events are shown in a cylindrical coordinate system (i.e., (Y2VSO + Z2VSO)1/2 versus XVSO, where VSO is the Venus-centered Venus-Sun Orbital coordinate system, where X is in the direction to the Sun, Y opposite to the orbital velocity of Venus and Z along the rotational axis of Venus). [19] In order to find how many separate events happen in this 1 Venus year a limit distance in time is set between separate events of 30 s, also ‘‘events’’ of one point are discarded from the database. This leaves 1637 events in the database varying strongly in strength in the range 0.2 DB/ B 2. The total distribution of the MM-like structures as a function of DB/B is shown in Figure 3. of VEX is sampled in cylindrical coordi[20] The orbit pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi nates into X Y 2 þ Z 2 = 0.25 0.25 RV boxes to find the time the spacecraft spends in each box over 1 Venus year.

This is used to find the chance of measuring MM waves in each box. For each box one should calculate the ratio number of events in box total time in MS

) number of total events time spent in box number of events in box P0 ¼ : time spent in box

P¼

[21] As in the calculation of the chance P to measure MM-like structures in a box the values for ‘‘number of total events’’ and ‘‘total time in MS’’ are constant, these values are taken out of the equation, which leads to the determination of P0. The chance P0 of events in each box is shown in Figure 4 on a log scale. Clearly, the greatest chance to observe MM is on the dayside, where close to Venus the maximum chance seems to follow the ionopause and further outward it follows the bow shock, with little chance in between. Down Venus’ magnetotail is another region of enhanced MM-like wave signatures, in the wake of the planet, between 1.75RV XVSO 2.75RV. Interestingly, the wake is also the region of developed turbulence, where the power spectra showed individual peaks at low frequencies, which may be correlated to MM-like structures [Vo¨ro¨s et al., 2008]. Two regions of enhanced chance P0 were identified, the dayside magnetosphere and the wake region, which are discussed separately below. 4.3. Dayside Activity [22] The dayside activity is assumed to be at XVSO 0. Volwerk et al. [2008] have shown two case studies for MMlike structures in this region, one near the bow shock for quasi-perpendicular conditions (reproduced in this paper) and one near the ionopause/magnetopause for an assumed compressed magnetosphere. In this paper a third event is shown of MM-like waves just behind the bow shock, for 11 June 2006. In this case the angle between solar wind magnetic field and local bow shock normal is 148°, which

4 of 13

E00B16

VOLWERK ET AL.: VENUS’ MIRROR MODE WAVES

E00B16

Figure 3. A histogram showing the distribution of the MM-like events as a function of their strength DB/B. means quasi-parallel bow shock conditions. Although the MM instability favours a quasi-perpendicular BS, at a moderately quasi-parallel BS, as in this case, the chance of MM waves is not zero. [23] Figure 5 shows the magnetic field data from VEX and the low-pass filtered data, with respect to which the strength DB/B is calculated. The spacecraft moves from the solar wind (SW), through the bow shock (BS) into the magnetosheath (MS). The location of the bow shock could be argued to be slightly later than the left boundary of shaded box I. The shaded boxes labelled ‘‘I’’ and ‘‘II’’ are two locations where the maximum variance direction of the waves makes an angle qBmv 20° with the background magnetic field (circles in Figure 5f), which is one of the requirements for the identification of MM-like structures being compressional waves. However, only in box II the minimum variance direction and the background magnetic field have an angle fBmv 80°, whereas in box I the value is slightly below 80°. In this case, the waves measured in box II are (automatically) identified as MM-like structures, whereas the waves in box I are not. [24] The strength of the MM-like structures in box II is rather small, DB/B 0.7 with a median 0.3. Therefore, these are not well developed waves, i.e., the driving mechanism is not efficient. This is to be expected, as this case represents a quasi-parallel shock, for which the shock drift acceleration of the particles, perpendicular to the magnetic field, by the bow shock is less efficient. [25] Interesting, too, are the waves between 0129 and 0130 UT, which have a minimum variance direction (pluses) along the magnetic field and a maximum variance direction (dots) perpendicular to the magnetic field. These are transverse waves at a period of 4 s, which would agree with

proton cyclotron waves at that location just below the local gyro frequency [see, e.g., Delva et al. 2008a], Bm 19 nT ! fp 0.29 Hz. These waves will be discussed in a separate paper. [26] Now, a closer look will be taken to all events on the dayside of Venus. It is expected that the MM-like structures near the bow shock occur mainly during quasi-perpendicular shock conditions [Lucek et al., 1999a, 1999b]. Therefore, as stated above, the angle between the solar wind magnetic field direction and the bow shock normal, qBn, is determined for each orbit of VEX. The dayside events are projected onto the YZVSO plane, and are sorted into five bins on the basis of qBn. In Figure 6 all dayside events are shown in Figure 6a and the cases of quasi-parallel shock (0° < qBn < 45° and 135° < qBn < 180°) are shown below it, whereas in Figure 6 (right) the events for the quasi-perpendicular shock for the angular bins 45° < qBn < 75°, 75° < qBn < 105° and 105° < qBn < 135° are shown. In Table 1 the number of events per bin is listed, with a total number of 1071 events on the dayside of Venus, covering 190 days of the whole data set (compared to 224 days for 1 Venus year). [27] Table 1 shows that quasi-parallel bow shock conditions, as defined by the bins in this paper, occur for approximately 30% of the time in the current data set. Figure 6 displays the chance to observe MM-like structures as a function of qBn, with all cases and the quasi-parallel cases in the left column and the quasi-perpendicular cases in the right column. The angle qBn between the local normal and the solar wind magnetic field has a clear influence on the chance of observing MM waves. Interestingly, the greatest chance to measure MM-like structures is for 75° < qBn < 135° (Figures 6d and 6f).

5 of 13

E00B16

VOLWERK ET AL.: VENUS’ MIRROR MODE WAVES

E00B16

Figure 4. The ‘‘normalized’’ occurrence rate of MM-like structures around Venus. Shown is the number of events in boxes 0.25 0.25RV divided by the number of seconds that VEX is located in each box on a log scale. (top) The occurrence rate for events with DB/B 0.2, and (bottom) the occurrence rate for events with DB/B 0.6. The white areas within the BS mean that there were no events matching the criteria of the figure. [28] In Figure 6 there is a preference for the MM-like structures to appear at positive ZVSO. This effect is created by the highly elliptical polar orbit of the spacecraft, with periapsis near the north polar region. Indeed, on the flanks (large YVSO), where the MS is wider, there are also data for negative ZVSO. [29] The nominal angle of the Parker spiral of the solar wind magnetic field qP 36°, which means that at the nose of the BS qBn 36° or 144°, depending on the direction of the magnetic field. From the current data set the direction of

the solar wind magnetic field can be determined, to obtain an estimate of the ‘‘Parker spiral direction’’ 0qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1 B2y þ B2z A: qP ¼ atan@ Bx

ð5Þ

[30] A histogram of the solar wind magnetic field direction, measured 30 min before VEX enters the bow

6 of 13

E00B16

VOLWERK ET AL.: VENUS’ MIRROR MODE WAVES

E00B16

Figure 5. Magnetic field data for 11 June 2006 in the same format as Figure 1. shock, see Figure 7, shows that the distribution peaks near the nominal values of the Parker spiral (indicated by the two vertical white lines), however, at large angles there is a broad variation. This distribution is comparable to the one found by Delva et al. [2008b]. The broad spread in directions of the solar wind magnetic field, with main occurrence between 90° qP 160°, this gives enough variation to create the observed distribution shown in Figure 6. 4.4. Wake Activity [31] The wake [Lundin and Barabash, 2004] will be qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 2 1. assumed to be at XVSO 1.5 and YVSO þ ZVSO

This region coincides with the optical shadow of the planet. From Figure 4 it is clear that at the night side the events occur mainly in the wake of Venus. This may complicate the investigation, as Vo¨ro¨s et al. [2008] have shown that as soon as VEX enters Venus’ shadow (or wake) the plasma turbulence spectral density (P( f ) / f a) changes from little developed (scaling index a  0.8) to almost fully developed (scaling index a  2.5). Two events of MM-like structures in the wake are discussed in detail below. [32] As first example, the data from 15 August 2006 are shown in Figure 8 (left), when the spacecraft was located near (2.1, 0.5, 0.1) RV, in the Venusian wake. The solar wind magnetic field was B (7, 6, 9) nT. In Figure 8 the low-pass

7 of 13

E00B16

VOLWERK ET AL.: VENUS’ MIRROR MODE WAVES

E00B16

Figure 6. Normalized occurrence rate of MM-like structures at the dayside of Venus in the YZ plane. (a) All events and (b – f) individual events sorted with respect to the angle between the solar wind magnetic field and the bow shock normal. The white areas mean that there were no events matching the criteria of the figure. filtered magnetic field has been subtracted from the nonfiltered data, to make the waves of interest better visible because of the strong variation in the amplitudes of the magnetic field components as can be seen from the variation in Bm. [33] Two intervals are shaded, where the angle of B to the maximum variance direction is below 20° and to the minimum variance direction larger than 80°. Box I shows weak waves with a mean DB/B 0.2, being small amplitude compressional waves, with the correct characteristics for MM waves. In box II is a superposition of ‘‘high-’’ frequency waves onto a large ‘‘low-’’ frequency oscillation of magnetic field in the X and Z component. The strength of the structure (as defined by DB/B 2) in this case clearly does not describe the situation correctly. The mean magnetic field magnitude approaches zero at the centre of the structure, which on such a large scale may indicate the crossing of a current sheet. In the case of box II, the difference

between filtered and nonfiltered data is misleading and the total magnetic field data needs to be looked at (shown in Figure 9). [34] From the above considerations it is clear that in reality we have a current sheet/neutral sheet crossing in Venus’ magnetotail, and not a MM-like structure. This is the only case where the automated approach led to misidenti-

Table 1. Number of Events in Each qBn Bin on the Dayside of Venusa qBn Bin

0 – 45°

45 – 75°

75 – 105°

105 – 135°

135 – 180°

Number of events Number of days

122 21

151 31

287 52

330 51

181 36

a In total there are 1071 events on the dayside covering 190 days of the whole data set.

8 of 13

E00B16

VOLWERK ET AL.: VENUS’ MIRROR MODE WAVES

E00B16

Figure 7. Histogram of the solar wind magnetic field ‘‘Parker angle’’ qP for 1 Venus year. The two vertical white lines indicate the nominal Parker spiral direction of 36° and 144°. fication. The structure is very interesting because it shows that on that day the Venusian current sheet was similar to the Earth’s current sheet in the YZ plane, as often observed by Cluster [e.g., Baumjohann et al. 2007]. Depending on the solar wind magnetic field and its draping around the planet, the current sheet at Venus can rotate between the XZ plane and the YX plane [Luhmann et al., 1991]. [35] As a second example in the wake, the data from 1 September 2006 are shown in Figure 8 (right), when the spacecraft was located near (2.3, 0.5, 0.3) RV and the solar wind magnetic field B (10, 1, 5) nT. There are two intervals, shaded boxes I and II, which satisfy the demands for MM-like structures. In box I is a very nice MM structure, with a strong decrease in Bm of 50%. Such a structure is reminiscent of so-called magnetic holes in the solar wind [see, e.g., Zhang et al., 2008, and references therein] and a clear indication of MM. In box II, again there is a very weak structure with DB/B 0.15. [36] Most of the MM-like structures in Venus’ wake are rather weak in strength. Indeed, limiting the data set to only stronger MM-like structures with DB/B 0.6 (corresponding to a 30% decrease in field magnitude) drastically reduces the occurence rate in the wake region by 1–1.5 orders of magnitude (Figure 4 (bottom)), whereas for the dayside this limit increase barely changes the occurrence rate. Most likely, the turbulence in the wake region disturbs the plasma too much to allow full development of MM-like structures

because of isotropization of any ion temperature asymmetry [see, e.g., Zimbardo et al., 2003].

5. Summary and Discussion [37] Venus’ magnetosheath has been investigated by a slew of spacecraft and the results from these missions have been well summarized by Phillips and McComas [1991]. Interestingly, most of the investigated wave activity therein was for regions under quasi-parallel bow shock conditions. This means that the mirror mode instability has not been investigated in any of the papers mentioned in the review. Up to the arrival of VEX there has been little discussion of MM waves in Venus’ magnetosheath. The reason could be that they either have not been observed, or they were not looked for in the various data sets. Indeed, identification of MM-like structures can be difficult, but not impossible, without high enough resolution plasma data to accompany the magnetic field data. Rae et al. [2007] have confirmed the existence of MM waves by using both magnetic field and plasma data from Equator-S, for cases that Lucek et al. [1999b] had identified first from magnetic field data only. [38] The (non)occurrence of MM waves in Venus’ magnetosheath is discussed by Luhmann [1995]. Using Pioneer Venus Orbiter (PVO) data for selected orbits between numbers 625 and 650, which probed the subsolar point of the magnetosheath, the authors found that the waves in the magnetic field were mainly linearly polarized and trans-

9 of 13

E00B16

VOLWERK ET AL.: VENUS’ MIRROR MODE WAVES

E00B16

Figure 8. (left) Waves in the Venusian wake for 15 August 2006. (right) Magnetic field data for 1 September 2006. (a – c) The difference between the measured magnetic field and the low-pass filtered data describing the mean field (DBx,y,z. (d) The magnitude of the magnetic field Bm is plotted with the smooth curve showing the low-pass filtered data. (e and f) The fluctuation of the magnetic field DB/B and the angle qBmv between the maximum variance direction and the mean magnetic field (circles) and the angle fBmv between the minimum variance direction and the mean magnetic field (pluses). The two grey shaded areas I and II show the intervals for which the MM wave condition is met (i.e., qBmv < 20, fBmv > 80, and DB/B large). verse, with no evidence for regularly occurring slow mode like waves structures (to which the MM waves belong) standing in the flow of the magnetosheath. However, this was for low-altitude regions in the magnetosheath. It could well be argued that some of the peaks that appear in the compressional component of the power spectra, presented by Luhmann [1995], are related to MM like structures. There seem to be spectral peaks between 0.07 and 0.1 Hz, which would translate to 10– 14 s period, which are slightly longer than were shown by Volwerk et al. [2008] for MMlike structures near the magnetopause/ionopause. However, without reevaluating the PVO data, no definite conclusions can be drawn on these waves. [39] In the present study, 1 Venus year of VEX data have been investigated on mirror-mode-like structures in Venus’ induced magnetosphere. The structures appear when in a high-b plasma a temperature asymmetry perpendicular and parallel to the ambient magnetic field exists. Such asymmetry can either occur through ion drift shock acceleration at the bow shock under quasi-perpendicular conditions or by magnetic pumping through the first adiabatic invariant

during compression of the induced magnetosphere. Also the motion of the plasma through the magnetosheath is creating the conditions favoring the MM instability. Treumannn and Baumjohann [1996, p. 59] write about the global phenomenon in the magnetosheath: ‘‘The shocked solar wind, on its path from the bow shock to the magnetopause, is adiabatically heated in the perpendicular direction, while at the same time field-aligned outflow toward the flanks of the magnetopause cools the plasma adiabatically in the parallel direction. The main effect is parallel cooling, leading to pressure anisotropy of the plasma which increases toward the magnetopause.’’ These statements are in agreement with the results from the current paper, with respect to the occurrence of the MM-like structures. However, to truly check this phenomenon, one has to wait for the plasma data to become available. [40] The occurrence rate of the MM-like structures is much higher at the dayside of Venus’ magnetosheath, in comparison to the night side wake region, when looking at structures of reasonably large strength (DB/B 30%). The lack of strong events in the wake is most likely caused

10 of 13

E00B16

VOLWERK ET AL.: VENUS’ MIRROR MODE WAVES

E00B16

Figure 9. The magnetic field for 15 August 2006 in VSO coordinates. (a – d) Shown are Bx, y, z, m and the smooth curve shows the low-pass filtered magnetic field. Grey shaded interval I shows the interval where there is MM activity (as determined in Figure 8) and grey shaded interval II shows the interval where erroneously MM was inferred but is shown to be a current sheet crossing. by the fully developed turbulence in that region, which will reduce temperature anisotropies. Another reason for the lower chance to observe these waves could be the stretching of the magnetic flux tubes that are ‘‘hung-up’’ at the magnetopause/ionopause. Russell et al. [2006b] describes how in such a case the length of the flux tubes becomes long relative to the velocity of the plasma along the magnetic field, leading to a decrease in the plasma b. If the plasma b becomes too low, then the MM instability does not occur anymore. [41] On the dayside, the occurrence rate is maximum near the bow shock and along the magnetopause/ionopause boundary, which is expected, following the above comment by Treumannn and Baumjohann [1996]. Near the bow shock the drift shock acceleration will generate temperature asymmetries under quasi-perpendicular situations. Near the magnetopause/ionopause the asymmetry will be created

through compression of the induced magnetosphere and by adiabatic parallel cooling. Unfortunately, because of the lack of plasma data at time of preparation of this paper, it is not possible to find out when the solar wind compresses the magnetosphere or how the temperature anisotropy develops through the magnetosheath. However, it is possible to obtain some indication about whether the magnetosheath is compressed or not by looking at the magnetometer data. For one, the location of the bow shock crossing can be compared with the best fit model of Zhang et al. [2007b], or the magnetic field strength near periapsis and can be compared with a long time average of the magnetic field strength at that location, e.g., Volwerk et al.’s [2008] second event shows a Bm 50 nT, whereas the orbit (day) before VEX measured Bm 40 nT. [42] The direction of the solar wind magnetic field has a strong influence on the occurrence rate and location of the

11 of 13

E00B16

VOLWERK ET AL.: VENUS’ MIRROR MODE WAVES

MM-like structures. As expected, the occurrence rate drops for quasi-parallel bow shock conditions. [43] Comparing the MM-like structures studied here with those observed in the Earth’s magnetosheath by Lucek et al. [1999b] on the dawnside (their Figure 1) shows that there the MM waves appear mainly around the magnetopause. Most likely, this is an orbital effect as the apogee of Equator-S is 11 RE and is thus not expected to reach the bow shock. Also, the duration of the events in the current paper is much shorter up to minutes instead of up to hours there. [44] Joy et al. [2006] discussed the MM waves in the Jovian magnetosheath. The authors find that ‘‘between the bow shock and the magnetopause, the fluctuations appear to increase in amplitude and decrease in frequency.’’ From the first event presented in this current paper (see Figure 1) it is seen that close to the BS, in interval I, the period of the wave is 5 s, whereas further away from the BS, in interval II, the period of the wave is 15 s, which is in agreement with what is observed at Jupiter. This effect can also be seen in the second event of this current paper (see Figure 5). However, one event presented by Volwerk et al. [2008], with VEX close to the magnetopause/ionopause shows that the period is 10 s, which is longer that near the BS, but shorter than in the mentioned interval II above. The fact that the period is shorter is not a contradiction with the results from Joy et al. [2006], as the plasma parameters in Venus’ magnetosheath may be different in the two events and therefore the frequency of the MM-like structures. [45] Interestingly, Bavassano Cattaneo et al. [1998] find at Saturn that the MM waves directly behind the BS, in the outer magnetosheath, have a ‘‘significant transverse component’’, and that they become almost purely compressional first in the inner magnetosheath. This is different from what is found here at Venus, where the MM-like structures are almost purely compressional also in the outer magnetosheath (near the BS). This could be an effect of the specific identification criteria used here. [46] Bertucci et al. [2004] studied the magnetic pile up region at Mars, where they find mirror mode waves at the upstream side of the pile up region, whereas on the downstream side the waves are fast magnetosonic waves. In our study at Venus it was found that the chance to observe MM-like structures follows the magnetopause/ionopause, which can be identified with the magnetic pile up boundary. Volwerk et al. [2008] showed that MM-like structures occur near the magnetopause/ionopause (their Figure 3). Near periapsis there is still evidence of wave activity at a similar period as the MM-like structures. However, the polarization of these waves changes over the magnetopause/ionopause from compressional, through intermediate propagating perpendicular to the field, to transverse with propagation direction more aligned with the background field. This could imply fast magnetosonic waves, but need further investigation. Bertucci et al. [2004] also find that MM waves are always present, independent of the BS nature. However, the MM waves are always ‘‘attached’’ to the magnetic pile up boundary.

6. Conclusions [47] A statistical study of MM waves for 1 Venus year of magnetometer data showed that at the dayside mirror-mode-

E00B16

like structures in Venus’ magnetosheath occur frequently, preferably under quasi-perpendicular bow shock conditions. This is in contrast to the findings during the PVO era, where no such waves were detected. In the wake the occurrence rate is much lower, caused by turbulence (reducing temperature asymmetries) and/or low plasma-b (suppressing the mirror mode instability). An explanation for the difference with the PVO era could be found in that previous research was concentrating on quasi-parallel bow shock conditions, which are unfavorable for mirror mode generation. [48] Acknowledgments. The authors thank Simon Pope at the University of Sheffield for preparing the data. The work by Z.V. is supported by the Austrian Wissenschaftsfonds under grant P20131-N16.

References Barabash, S., et al. (2007), The analyser of space plasmas and energetic atoms (ASPERA-4) for the Venus Express mission, Planet. Space Sci., 55, 1772 – 1792, doi:10.1016/j.pss.2007.01.014, 2007. Baumjohann, W., and R. A. Treumannn (1996), Basic Space Plasma Physics, Imperial Coll. Press, London. Baumjohann, W., R. A. Treumann, E. Georgescu, G. Haerendel, K.-H. Fornac¸on, and U. Auster (1999), Waveform and packet structure of lion roars, Ann. Geophys., 17, 1528 – 1534. Baumjohann, W., et al. (2007), Dynamics of thin current sheets: Cluster observation, Ann. Geophys., 25, 1365 – 1389. Bavassano Cattaneo, M. B., C. Basile, G. Moreno, and J. D. Richardson (1998), Evolution of mirror structures in the magnetosheath of Saturn from the bow shock to the magnetopause, J. Geophys. Res., 103, 11,961 – 11,972. Bertucci, C., et al. (2004), MGS MAG/ER observations at the magnetic pileup boundary of Mars: Draping enhancement and low frequency waves, Adv. Space Res., 33, 1938 – 1944. Califano, F., P. Hellinger, E. Kuznetsov, T. Passot, P. L. Sulem, and P. Tra´vnı´
cek (2008), Nonlinear mirror mode dynamics: Simulations and modeling, J. Geophys. Res., 113, A08219, doi:10.1029/2007JA012898. Constantinescu, O. D., K.-H. Glassmeier, R. Treumann, and K.-H. Fornac¸on (2003), Magnetic mirror structures observed by Cluster in the magnetosheath, Geophys. Res. Lett., 30(15), 1802, doi:10.1029/2003GL017313. Delva, M., T. L. Zhang, M. Volwerk, W. Magnes, C. T. Russell, and H. Y. Wei (2008a), First upstream proton cyclotron wave observations at Venus, Geophys. Res. Lett., 35, L03105, doi:10.1029/2007GL032594. Delva, M., T. L. Zhang, M. Volwerk, Z. Vo¨ro¨s, and S. A. Pope (2008b), Proton cyclotron waves in the solar wind at Venus, J. Geophys. Res., 113, E00B06, doi:10.1029/2008JE003148. Erdo¨s, G., and A. Balogh (1993), Statistical properties of mirror mode structures observed by ulysses in the magnetosheath of Jupiter, J. Geophys. Res., 101, 1 – 12. Gary, S. P., S. A. Fuselier, and B. J. Anderson (1993), Ion anisotropy instabilities in the magnetosheath, J. Geophys. Res., 98, 1481 – 1488. Glassmeier, K. H., U. Motschmann, C. Mazelle, F. M. Neubauer, K. Sauer, S. A. Fuselier, and M. Acun˜a (1993), Mirror modes and fast magnetoacoustic waves near the magnetic pileup boundary of comet P/Halley, J. Geophys. Res., 98, 20,955 – 20,964. Hasegawa, A. (1969), Drift mirror instability in the magnetosphere, Phys. Fluids, 12, 2642 – 2650. Huddleston, D. E., R. J. Strangeway, X. Blanco-Cano, C. T. Russell, M. G. Kivelson, and K. K. Khurana (1999), mirror mode structures at the Galileo-Io flyby: Instability criterion and dispersion analysis, J. Geophys. Res., 104, 17,479 – 17,489. Joy, S. P., M. G. Kivelson, R. J. Walker, K. K. Khurana, C. T. Russell, and W. R. Paterson (2006), Mirror mode structures in the Jovian magnetosheath, J. Geophys. Res., 111, A12212, doi:10.1029/2006JA011985. Kirk, J. G., D. B. Melrose, and E. R. Priest (1994), Plasma Astrophysics: Saas-Fee Advanced Course 24, Springer Verlag, Berlin, Germany. Lucek, E. A., M. W. Dunlop, A. Balogh, P. Cargill, W. Baumjohann, E. Georgescu, G. Haerendel, and K.-H. Fornac¸on (1999a), Mirror mode structures observed in the dawn-side magnetosheath by Equator-S, Geophys. Res. Lett., 26, 2159 – 2162. Lucek, E. A., M. W. Dunlop, A. Balogh, P. Cargill, W. Baumjohann, E. Georgescu, G. Haerendel, and K.-H. Fornac¸on (1999b), Identification of magnetosheath mirror modes in Equator-S magnetic field data, Ann. Geophys., 17, 1560 – 1573. Luhmann, J. G. (1995), The innner magnetosheath of Venus: An analog for Earth?, J. Geophys. Res., 100, 12,035 – 12,045.

12 of 13

E00B16

VOLWERK ET AL.: VENUS’ MIRROR MODE WAVES

Luhmann, J. G., C. T. Russell, K. Schwingenschuh, and Y. Yeroshenko (1991), A comparison of induced magnetotails of planetary bodies: Venus, Mars, and Titan, J. Geophys. Res., 96, 11,199 – 11,208. Lundin, R., and S. Barabash (2004), The wakes and magnetotails of Mars and Venus, Adv. Space Res., 33, 1945 – 1955, doi:10.1016/j.asr.2003. 07.054. Phillips, J. L., and D. L. McComas (1991), The magnetosheath and magnetotail of Venus, Space Sci. Rev., 55, 1 – 80. Price, C. P., D. W. Swift, and L.-C. Lee (1986), Numerical simulation of nonoscillatory mirror waves at the Earth’s magnetosheath, J. Geophys. Res., 91, 101 – 112. Rae, I., I. Mann, C. Watt, L. Kistler, and W. Baumjohann (2007), Equator-S observations of drift mirror mode waves in the dawnside magnetosphere, J. Geophys. Res., A11203, doi:10.1029/2006JA012064. Russell, C. T., W. Riedler, K. Schwingenschuh, and Y. Yeroshenko (1987), Mirror instability in the magnetoshpere of Comet Halley, Geophys. Res. Lett., 14, 644 – 647. Russell, C. T., J. S. Leisner, C. S. Arridge, M. K. Dougherty, and X. BlanoCano (2006a), Nature of magnetic fluctuations in saturn’s middle magnetosphere, J. Geophys. Res., 111, A12205, doi:10.1029/2006JA011921. Russell, C. T., J. G. Luhmann, and R. J. Strangeway (2006b), The solar wind interaction with Venus through the eyes of the Pioneer Venus orbiter, Planet. Space Sci., 54, 1482 – 1495, doi:10.1016/j.pss.2006.04.025. ¨ ., and M. Scheible (1998), Minimum and maximum Sonnerup, B. U. O variance analysis, in Analysis Methods for Multi-Spacecraft Data, edited by G. Paschmann and P. Daly, pp. 185 – 220, Eur. Space Agency, Noordwijk, Netherlands. Southwood, D. J., and M. G. Kivelson (1993), Mirror instability: 1. Physical mechanism of linear instability, J. Geophys. Res., 98, 9181 – 9187. Svedhem, H., D. V. Titov, F. W. Taylor, and O. Witasse (2007), Venus as a more Earth-like planet, Nature, 450, 629 – 632, doi:10.1038/nature06432. Treumannn, R., and W. Baumjohann (1996), Advanced Space Plasma Physics, Imperial Coll. Press, London. Tsurutani, B. T., E. J. Smith, R. R. Anderson, K. W. Ogilvie, J. D. Scudder, D. N. Baker, and S. J. Bame (1982), Lion roars and nonoscillatory drift

E00B16

mode mirror waves in the magnetosheath, J. Geophys. Res., 87, 6060 – 6072. Volwerk, M., T. L. Zhang, M. Delva, Z. Vo¨ro¨s, W. Baumjohann, and K.-H. Glassmeier (2008), First identification of mirror mode waves in Venus’ magnetosheath, Geophys. Res. Lett., 35, L12204, doi:10.1029/ 2008GL033621. Vo¨ro¨s, Z., T. L. Zhang, M. P. Leubner, M. Volwerk, M. Delva, W. Baumjohann, and K. Kudela (2008), Magnetic fluctuations and turbulence in the Venus magnetosheath and wake, Geophys. Res. Lett., 35, L11102, doi:10.1029/2008GL033879. Zhang, T. L., et al. (2006), Magnetic field investigation of the venus plasma environment: expected new results, Planet. Space Sci., 54, 1336 – 1343. Zhang, T. L., et al. (2007a), Little or no solar wind enters venus atmosphere at solar minimum, Nature, 450, 654 – 656, doi:10.1038/nature06026. Zhang, T. L., et al. (2007b), Initial Venus express magnetic field observations of the Venus bowshock location at solar minimum, Planet. Space Sci., 56, 785 – 789, doi:10.1016/j.pss.2007.09.012. Zhang, T. L., et al. (2008), Characteristic size and shape of the mirror mode structures in the solar wind at 0.72 AU, Geophys. Res. Lett., 35, L10106, doi:10.1029/2008GL033793. Zimbardo, G., A. Greco, A. L. Taktakishvili, P. Veltri, and L. M. Zelenyi (2003), Magnetic turbulence and particle dynamics in the earth’s magnetotail, Ann. Geophys., 21, 1947 – 1953. 

W. Baumjohann, M. Delva, M. Volwerk, and T. L. Zhang, Space Research Institute, Austrian Academy of Sciences, Schmiedlstraße 6, A-8042 Graz, Austria. ([email protected]) K.-H. Glassmeier, Institut fu¨r Geophysik und extraterrestrische Physik, Technische Universita¨t, D-38106 Braunschweig, Germany. Z. Vo¨ro¨s, Institute for Astro- and Particle Physics, University of Innsbruck, Technikerstraße 25/8, A-6020 Innsbruck, Austria.

13 of 13

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114, E00B26, doi:10.1029/2008JE003170, 2009

for

Full Article

O+ ion flow below the magnetic barrier at Venus post terminator K. Szego,1 Z. Bebesi,1 Z. Dobe,1 M. Fra¨nz,2 A. Fedorov,3 S. Barabash,4 A. J. Coates,5 and T. L. Zhang6 Received 20 April 2008; revised 21 August 2008; accepted 14 October 2008; published 1 January 2009.

[1] Venus forms an obstacle in the streaming solar wind; inside the obstacle boundary

(that is below the magnetic barrier) the ions of planetary origin dominate the plasma. The objective of this study is to investigate the properties of the O+ ions inside the obstacle boundary of Venus in the wake; we are especially interested in the characterization of the different plasma regions the O+ ions occupy. The study is based on the data collected by the ASPERA 4 plasma analyzer flying onboard of the Venus Express mission in a region never explored before experimentally. The obstacle boundary was approximately identified from the dropout of magnetospheric electrons and the sharp decrease of the proton speed; the entry point correlated well with the location of the magnetic barrier derived by eyes from magnetometer data. The most characteristic structures seen during the various flybys were (1) the tailward continuation of the mantle was evident; (2) in the mantle near Venus the O+ ion flow was significantly intense in low-energy counts; (3) the inbound and outbound crossings of the tailward boundary were sharp, characterized by less intense but higher-energy O+ beams; (4) the crossing of the central tail region (current sheet) was marked by the change of the sign of Bx and by an intense low-energy O+ ion flux; (5) it is remarkable that the O+ ion outflow was not confined to the central tail region; the intensity elsewhere was highly variable, resulting in a ray-like outflow pattern in most of the cases. Citation: Szego, K., Z. Bebesi, Z. Dobe, M. Fra¨nz, A. Fedorov, S. Barabash, A. J. Coates, and T. L. Zhang (2009), O+ ion flow below the magnetic barrier at Venus post terminator, J. Geophys. Res., 114, E00B26, doi:10.1029/2008JE003170.

1. Introduction [2] It is well known that Venus forms an obstacle in the streaming solar wind. Above the obstacle the solar wind dominates the plasma flow, below the obstacle boundary the ions of planetary origin play the major role. Zhang et al. [1991] identified the magnetic barrier as the obstacle boundary at Venus on the dayside, and Spenner et al. [1980] proved that below the magnetic barrier there is a region, the mantle, where solar wind and planetary populations coexist. The mantle is bounded below by the ionopause, inside the mantle; in the lower regions of the mantle, the planetary ions become dominant. Intuitively, and according to models, it is clear that the mantle continues tailward, but experimentally this was not verified. The objectives of this study are to investigate whether the mantle exists in the wake, to investigate the properties of the O+ 1 KFKI Research Institute for Particle and Nuclear Physics, Budapest, Hungary. 2 MPI fu¨r Sonnensystemforschung, Katlenburg-Lindau, Germany. 3 Centre d’Etude Spatiale des Rayonnements, Toulouse, France. 4 Swedish Institute of Space Physics, Kiruna, Sweden. 5 Mullard Space Science Laboratory, University College London, London, UK. 6 Space Research Institute, Austrian Academy of Sciences, Graz, Austria.

Copyright 2009 by the American Geophysical Union. 0148-0227/09/2008JE003170$09.00

ions below the obstacle boundary of Venus in the wake; especially we are interested in the characterization of the different plasma regions the O+ ions occupy, their boundaries, the motion of these ions and the process of tail formation (this latter question will be discussed elsewhere). Concerning the magnetic field, Vo¨ro¨s et al. [2008] pointed out recently that in the magnetic data the boundaries in question are evident in most of the cases, but with different signatures than on the dayside: they appear as enhanced perturbations rather than jumps in magnitudes. [3] The analysis presented here for O+ ions is based on the data collected by the Analyzer of Space Plasmas and Energetic Atoms (ASPERA-4) carried onboard of the Venus Express mission [Barabash et al., 2007a]. The ASPERA-4 sensor is unique relative to all previous instruments flown around Venus because it measures 3D plasma distributions above 10 eV, making possible to derive the moments of the major ion components, for the first time since the exploration of Venus started. The Venus Express (VEX) spacecraft has polar orbit (the latitude of the pericenter is at 76° North relative to Venus orbit, the period is 24 h) with pericenter altitude 220 – 350 km and apocenter altitude 66,000 km; this orbit is inertially fixed. The VEX spacecraft is 3-axes stabilized, the attitude varies with observation modes, in most cases it is nadir pointing. ASPERA-4 onboard VEX is exploring the near Venus plasma regions. The orbit constraints of earlier missions did not allow the detailed exploration of this region. We review the previous

E00B26

1 of 12

E00B26

SZEGO ET AL.: O+ ION FLOW AT VENUS

Figure 1. The plasma boundaries at Venus, in Venus coordinate system (VSO) cylindrical coordinates, as observed on Venera 9 and 10 [Vaisberg et al., 1995]. space mission results therefore only in the region relevant for this study, that is in the post terminator mantle and in the near plasma tail. [4] The first two spacecraft that studied the solar wind interaction with Venus were Venera 9 and 10 [Vaisberg et al., 1995]. They started orbiting Venus on 20 and 23 October 1975, respectively. Their main task was to support the lander missions, but they also studied the tail of Venus. The satellites had two plasma spectrometers: a 2-D ion spectrometer consisting of 6 narrow-angle cylindrical electrostatic analyzers and a combination of sunward-looking differential ion Faraday cup and antisunward oriented integral electron Faraday cup. The energy ranges were 50 eV/Q – 20 keV/Q, and 0 – 4 keV/Q for ions and 0– 400 eV for electrons. The time resolution of both spectrometers was 160 s. [5] The main regions identified downstream, based on the data collected by the onboard electron and ion spectrometers, in the close vicinity of the planet are shown in Figure 1. The investigators found a boundary layer separating the Venus magnetosheath and the tail; within this boundary layer low-energy plasma streamed tailward. This boundary layer seems to be the continuation of the plasma mantle identified during the Pioneer Venus mission by Spenner et al. [1980]. The tailward boundary of this layer (tail boundary) was sharp. Having crossed the tail boundary the Veneras measured higher-energy, E > 50 eV ion flow; ions were permanently seen within the tail. In the innermost region, near the central part of the tail no ions were visible above 50 eV, this region was called cavity. The Venera investigators reported that the tail thickness depends on the magnetic latitude, i.e., from the angular distance of the plane containing the component of B perpendicular to the SW velocity vector and the center of Venus. The tail at distances 4 –5 RV was studied only during two passes of

E00B26

Venera 10, and no boundary crossing was observed. The tail plasma is of planetary origin, as the investigators assessed it. During the two passes at distances 4– 5 RV in the tail plasma ions were permanently observed; no hot ions were seen at Bx reversals. [6] The Pioneer Venus Orbiter (PVO) mission started to operate in December 1978 and functioned till the fall of 1992. At the beginning of the mission (which coincided with low solar activity) the periapsis of the spacecraft was lowered to 150 km, and maintained there for about 19 months. Later the periapsis was raised till the reentry before the end of the mission. The orbit plane of the Pioneer Venus spacecraft (sc) was nearly polar (with a periapsis at 17°N latitude); the spacecraft was spinning with its spin axis normal to the plane of the Venus orbit. The analog experiment to ASPERA-4 onboard PVO, concerning low-energy ions, was the Orbiter Plasma Analyzer (OPA). It was a quadrispherical electrostatic analyzer; the energy/charge range was 50– 8000 eV (ions) in 32 steps and 1 – 500 eV (electrons) in 16 steps. The angular range covered was ±85° elevation. However, OPA was primarily designed for solar wind monitoring with a low time resolution of about 10 min [Intriligator et al., 1980]. [7] In the tail the PVO plasma instruments either explored the nightside ionosphere and the very near tail region, or the far tail; that is 150 km to 2500 km altitude, or 5 – 11 Venus radii altitude. Here we focus primarily on the near tail region. The most important findings were summarized by Phillips and McComas [1991] and Brace and Kliore [1991]. The first important fact is that the ionotail exhibits large solar cycle and orbit to orbit variations. The solar cycle variation of the tail is connected to the variation of the electron density on the dayside, which is also controlled by the solar EUV flux, and hence by the solar cycle. The height of the nightside ionopause correlates with solar cycles. The PVO spacecraft between July 1984 and November 1986, during solar cycle minimum conditions found that the ionopause was at low height; between December 1978 and October 1981 during solar cycle maximum conditions it found a more extended ionosphere. The mechanism of the orbit-to-orbit variation is yet unexplained. The VEX data, in 2006, were taken in solar minimum, that is high solar wind pressure conditions. It was argued by Slavin et al. [1989] and others that the tail structure observed in the ion spectra correlates with the magnetic field variations; the low-energy plasma in the tail is controlled by the IMF orientation. The current sheet in the tail is parallel to the vSW  BSW direction. In the central tail, at the current sheet, whenever Bx changed sign, the energy of ions increased. Ions, measured within the tail contained very cold ions (and moving at half solar wind speed if they were assumed to be protons). PVO observations confirmed that the plasma flow speed is characteristically slowest in the center of the tail, and faster outside the tail. It has been established that the origin of the O+ ions in the tail is the day-to-night O+ flow, and the escape is due to the polarization field that develops in the ionosphere [Hartle and Grebowski, 1995]. The curvature force of the magnetic field and the magnetic pressure force both act to accelerate the plasma down the tail [Russell, 1999]. McComas et al. [1986] determined the average structure of the Venus magnetotail from PVO magnetometer observations. They have found that in the tail the tailward velocity varied from 250 km/s at 8 RV

2 of 12

E00B26

SZEGO ET AL.: O+ ION FLOW AT VENUS

Figure 2a. The field of view (FOV) of the Ion Mass Analyzer (IMA) is 90°  360°. This can be modeled as a cylinder; its y axis is the direction about which the FOV is 360°, and the height of the cylinder models the ±45° FOV in the perpendicular direction. The horizontal arrow is the x direction in the spacecraft frame of reference. Owing to location of IMA on the spacecraft, the spacecraft body and the solar panels partially obscure the full FOV. The elevated surfaces illustrate the blocked portion of the sensor. to 470 km/s at 12 RV, the current sheet density was 0.06 O+/cc, while the lobe density was approximately 15% of that. A strong flapping was identified; at 12 RV it might reach the average cross section of the tail. [8] Recently, Barabash et al. [2007b] published the first results on ion escape through the Venusian tail. In that paper the fluxes of the escaping ions were collected from 33 orbits (basically from the same orbits we use in the present study), and integrated for the near tail. These integrated fluxes revealed structures in the tail: in the central part of the tail, called plasma sheet, the fluxes were the highest, and the plasma sheet separated the tail lobes inside the magnetosphere boundary, that is inside the magnetic barrier boundary layer. In the current paper, using mostly the same orbits as in the work by Barabash et al. [2007b], we give a detailed analysis of the O+ ions along the individual orbits in the tail plasma, pointing out also certain difficulties to have a full picture at this point of data analysis. [9] The plasma data provided by the onboard instrumentation of VEX differ significantly from the data we received from PVO or from other missions. The most important difference is that ASPERA-4 does not measure cold ions since the lowest energy limit of the instrument is 10 eV. This means that the ion spectrometer does not give direct information on the ionospheric plasma, and it is difficult to resolve the location of the ionopause in the ion data owing to our time resolution. On the other hand, as discussed above, on PVO the ion moments were measured sparsely, and the mass resolution was low. Hence, the ASPERA data significantly enrich our knowledge on the ion environment of Venus relative to previous missions in a region never explored before.

E00B26

[10] The Ion Mass Analyzer (IMA) instrument is part of the ASPERA-4 experiment package, the general scientific objective of ASPERA-4 is to study the solar windatmosphere interaction and characterize the plasma and neutral gas environment in the near-Venus space. The IMA instrument is a separate part attached to the spacecraft body, it provides ion measurements in the energy range 0.01– 36 keV/q for the main ion components with M/q > 40 amu/q. Its energy resolution is DE/E = 0.06, the field of view (FOV) is 90°  360°, the angular resolution is 10°  22.5°; and the time resolution for a full 3-D data acquisition is 192 s. The IMA FOV can be modeled as a cylinder; its axis is the direction about which the FOV is 360°, and the height of the cylinder models the ±45°FOV in the perpendicular direction (compare Figure 2a). However, owing to location of IMA on the spacecraft, the spacecraft body and the solar panels partially obscure the full FOV. The elevated surfaces in Figure 2a illustrate the blocked portion of the sensor in first approximation (Fedorov, internal team report, 2008). We do not need a refined obscuration model in this paper, though work is in progress to this end. [11] The Electron Spectrometer (ELS) on VEX [Barabash et al., 2007a] is an axially symmetric quadrispherical analyzer. It has a field of view of 360°  4°, with the 360° measurement plane divided into 16 sectors, each 22.5° wide; the energy resolution (DE/E) is 7%. The ELS operation varies, for the observations referred in this paper ELS covered the spectral range from 0.8 to 30 keV in 127 logarithmically spaced energy steps. ELS is located on a scanning platform providing a 4p coverage. Electron spectra can be measured about every 4 s. The magnetometer of VEX is described by Zhang et al. [2006], in this paper we use the 1-s resolution data product of the magnetometer.

2. Data Analysis [12] We use two coordinate systems for analyzing our results. The first is the standard Venus coordinate system (VSO), the second one is a Venus magnetic coordinate system (VMSO), where the x’ axis is parallel to the actual vSW velocity vector (hence aberrated to the actual flow B plane, direction), and the y’ axis is in the vSW perpendicular to vSW. The current sheet is parallel to the vSW  B direction. This coordinate system was introduced by Saunders and Russell [1986]. (In the VSE system the x axis points toward the Sun, the z axis is the same.) [13] In this paper we analyze data of the Ion Mass Analyzer (IMA) part of VEX, taken between 20 November and 31 December 2006. We also use data from orbits half a Venus year earlier and later. The analysis presented here is preliminary owing to several reasons: (1) the IMA data we are using are partially calibrated, the effects due to the saturation of the sensor at 104 counts per energy step has not yet been fully eliminated; (2) the characterization of the plasma requires the comparison of the data to other VEX sensors like ELS, and to other instruments such as the magnetometer [Zhang et al., 2006, 2007]. Though those data are available, and we are using them in this paper, a full joint analysis is a task ahead of us. The spacecraft potential is yet unknown at this point outside the ionosphere (in the ionosphere it is about 11 eV), whereas it has a significant influence on low-energy ion detection. As we stated, our

3 of 12

E00B26

SZEGO ET AL.: O+ ION FLOW AT VENUS

E00B26

Figure 2b. The change of the view geometry of IMA on 14 December 2006 between 0646 and 0752 UT. The frame is parallel to the VSO frame of reference (the Sun-Venus line is the horizontal axis); the arrow at the corner of the obscured region is parallel to the spacecraft x axis; the axis of the cylinder is the y axis. Above each plot the time of the actual orientation is shown. The azimuth angles (the vertical axis in Figure 4b) is in the x-z plane, measured from x toward z, whereas the elevation angle is measured relative to the x-z plane, its maximal value in Figure 4b is toward the y axis of the spacecraft. objective is to identify different plasma regions, and the signatures we shall rely on in this analysis will be the sudden changes in the counts of ASPERA-4 in neighboring regions. The uncalibrated nature of the data, in our opinion, will not mask these sudden changes; even if refinements might be required later. [14] Within the ASPERA-4 team the data flow from the spacecraft to the user is well organized through different Web sites where supporting data are also available, and quick look is made possible. Fra¨nz et al. [2006] have developed moment calculation for ASPERA-3 flying onboard of Mars Express, and also a code that has been upgraded for ASPERA-4, and made available for the team for data analysis. This is the code we have been using for this paper as well. [15] The VEX spacecraft has a polar orbit (the latitude of the pericenter is at 76°N relative to Venus orbit) and since Venus orbits the Sun in 224.7 days, the inclination of the spacecraft orbit plane to the Sun-Venus line changes daily by about 1.6°. For the current analysis we selected 16 orbits around a special one (orbit 237, pericenter at 0729:34 UT on

14 December 2006, the CA altitude was about 375 km) which is more or less parallel to the Sun-Venus line, and VEX was flying tailward; and the others were selected from a time interval +17/ 24 days around it. This means that for these orbits the smallest distance between the Sun-Venus line and the actual orbits changed from 0 to 1 Venus radius; and this set of orbits provides reasonable tail coverage. [16] After the pericenter passage the spacecraft attitude varies strongly; in Figure 2b we illustrate the attitude variation on 14 December 2006, between 0745 and 0752 UT. From these plots it can be seen that neither the attitude variation, nor the obscuration blocks the FOV for particles arriving from the ram or from planetary directions. Hence, we are convinced that if we measure low counts, it really means that in the appropriate tail region the ions are of very low density (we assume here that in the tail the sc potential is near zero, the data of the other space missions support this assumption). [17] There are a number of orbits appropriate fully or partially for the analysis of the plasma regions below the obstacle boundary. In Figures 3a and 3b we provide an overview of the electron spectra (collected by ELS), the

4 of 12

E00B26

SZEGO ET AL.: O+ ION FLOW AT VENUS

Figure 3a. Summary plot of electron and all-ion spectra, total magnetic field, and sc position on 14 December 2006 between 0700 and 0830 UT. The top and middle are dynamic spectrums (vertical axis is energy in eV units and the horizontal is time in minutes), with color coded count rated for the electrons and ions, respectively. The magnetic field values were measured by the magnetometer onboard Venus Express, the time resolution of the data is 1 s.

E00B26

total ion spectra (IMA), the total magnetic field, and the sc position along the flyby on 14 December 2006, to illustrate our text below. An important consequence of this plot is that whenever the ion counts are low, the electron counts are low as well. Since the electron motion is not directional, low electron density means low plasma density, substantiating our claim that the measured low ion densities are real. [18] The spacecraft speed near pericenter is close to 9 km/s. The O+ ions moving with 10 km/s relative speed to the spacecraft would impact with 8.35 eV. Therefore the ion speed and consequently the measured ion energies should have been corrected with the spacecraft motion. Since the ion velocity directions are not well known, these corrections could not be implemented for the data presented in this paper; we have to bear in mind the uncertainties due to that. [19] At the beginning of the flyby the spacecraft first was in the unperturbed solar wind, after it crossed the bow shock as witnessed by an increase of particle intensity and the increase of the magnetic field. Then Btot piles up; the magnetic barrier (that is the obstacle boundary) is identifiable on this and usually on all the other orbits we investigate. Further down the warm sheath electrons (E > 30eV) drop out (the region where the electrons are not detected is called the electron biteout region), the solar wind ions also drop out where the electrons do, clear signatures that the obstacle was crossed and the solar wind particles do not dominate the flow any longer, and the spacecraft reached the inner edge of the magnetic barrier behind terminator. Zhang et al. [1991] showed that the magnetic barrier is in effect the obstacle that diverts the bulk of the solar wind plasma from the inner regions of Venus. On PVO, below the magnetosheath of Venus on the dayside the Orbiter Retarding Potential Analyzer [Knudsen et al., 1980] identified a region called mantle where both sheath electrons and ionospheric electrons were present [Spenner et al., 1980]. A more detailed analysis of the ELS data of ASPERA-4 on 18 May 2006 [Coates et al., 2007] identifies the transition region, between 700 and 900 km in altitude, in the same way as Spenner et al. [1980] introduced the mantle, that is by the presence of both ionospheric photoelectrons (between 21 and 24 eV energy range) and sheath electrons. On that orbit VEX crossed the transition region in less than 3 min, that is in less time than the duration of a full IMA cycle (192 s). This illustrates well the difficulty we are confronted if we want to

Figure 3b. The spacecraft orbit is plotted in cylindrical coordinates, with tick marks for every 10 min. The distance unit is 1 Venus radius (RV = 6051 km). The other curves are the theoretical bow shock and magnetic pileup boundary locations, respectively. The circle represents Venus. 5 of 12

E00B26

SZEGO ET AL.: O+ ION FLOW AT VENUS

Figure 4a. Energy versus time spectrograms with color coded H+ and O+ energy fluxes derived from ASPERA-4 IMA data for the 14 December flyby, between 0710 and 0830, below the obstacle boundary. The O+ ions that appear at solar wind energies are due to crosstalk. At low proton densities the crosstalk is negligible. For the calculation we used the calibration of A. Fedorov (internal team report, the IMA bible, 2007) and also a weak noise reduction. study ions in the mantle along these orbits. We are also aware of a crosstalk between the proton and O+ channel, to be cleaned at a later phase of data processing. This crosstalk is not strong, and it is not very disturbing in the tail plasma where the proton density is low. On 14 December 2006, according to the plasma regions exhibited in Figure 1, the spacecraft, after having crossed the mantle, possibly dipped briefly into the ionosphere, than it entered into the tail region, and for a short period of time it crossed the central tail. When exiting the tail, the mantle in the tail was crossed again before returning to the magnetosheath. In Figures 4a – 4c we present further details of the data collected along this orbit. We note here that in the literature the term ‘‘plasma sheet’’ and ‘‘current sheet’’ are not used coherently; and because of lack of data, there is no experimental distinction for Venus. In order to avoid misunderstanding, in this paper we frequently

E00B26

use the term ‘‘central part of the tail’’, since our resolution certainly does not allow the ion sensor to discriminate between them. The current sheet is the surface that separates the two different polarities of the magnetic field in the tail (lobes). The plasma sheet is a broader region including the current sheet as seen from the analysis of Barabash et al. [2007b], but for individual flybys the plasma sheet boundary is not properly defined yet. [20] Below first we analyze in detail the flyby on 14 December 2006; then four more flybys will be discussed. On 14 December 2006 the pericenter altitude was 370 km, reached at 0731 UT (7.51 UT). On that day the vSW  BSW convective electric field pointed to the (0.46, 0.84, 0.3) direction in VSO, and the y’ axis of VMSO pointed to ( 0.175, 0.235, 0.96) in VSO. The projection of the orbit to the y z plane was parallel to the z axis in VSO; that is the magnetic latitude is almost identical to the geometric one. We shall use different plots to exhibit the data. The most conventional ones are probably the time-energy fluxes for the different plasma species, and we shall display counts, integrated pixel by pixel for all pixels of the IMA sensor in its attitude-elevation frame (compare Figure 2a). We shall plot also the variation of the magnetic field vectors along the orbit. From the time-energy fluxes shown in Figure 4a and from the angular distribution of the counts (Figure 4b), it can be concluded that the craft was below the obstacle boundary between 0725 and 0810 UT. Before that, as seen from the first attitude-elevation mosaic of Figure 4b collected at 0721:50, the beam was observed about the middle of the sensor, where the footprint of the shocked solar wind is expected. Later, below the boundary, as the next mosaic of Figure 4b shows, though the beam reached the detector, it was only partially in our field of view. The energy of the O+ ions can be obtained quite correctly, but the direction of the flow cannot be deduced in a simple way. What is evident, however, that the flow direction sensed onboard changed almost to the opposite direction right after entering into the boundary layer at 0725:02. Between 0721 and 0731 the sc attitude was stable; hence, the variation of the location of the peak counts on the detector plane is due to a change in the flow direction. We interpret this directional change by that the spacecraft was overpassing the ions that arrived from the ram direction. In this region the spacecraft was in the sunlight, hence it is expected that the overall sc potential was positive in the low-density plasma, and in addition to the ram effect, the ions must have been repelled from the spacecraft. Below the magnetic barrier we encountered low-energy O+ ions. The flow direction changed again at 0731:26, as well as the ion energy increased sharply, we interpret this by that the sc crossed the tail boundary. We could not associate any magnetic signature with this crossing. [21] The magnetic field vectors along the orbit, as viewed from the direction perpendicular to the Sun-Venus line is shown in Figure 4c. The field vector changes sign at 0750 UT, this coincides with the low-energy peak flux in Figures 4a and 4b. [22] We do not exhibit the mass-energy spectrum for this orbit. For count rates above 104 on these plots we have a saturation effect similar to the ‘‘blooming’’ of the CCD’s, peaks with such high count rates overflow the neighboring energy channels. Though we have already a fix to this

6 of 12

E00B26

SZEGO ET AL.: O+ ION FLOW AT VENUS

E00B26

Figure 4b. Azimuth-elevation plot for the 14 December flyby. The vertical axis is azimuth from 0° to 360° (corresponding to the equator of the cylinder in Figure 3), the horizontal axis is elevation (corresponding to the height of the cylinder in Figure 3), see the legend of Figure 2 for further details on the geometry. The azimuth is measured from the x axis of the spacecraft; the elevation (Theta) is measured from the y axis of the spacecraft. problem, this is not yet applied for mass production of data. Nevertheless, from such plots we are confirmed that the identification of the major tail ion species are correct. [23] As seen from Figure 4c, around 0750 UT Bx changed sign, corresponding to the geometric location of the crossing of the current sheet, and at that time the flux composed

of low-energy ions had a sharp maximum; we believe that we crossed the current sheet. There was also a peak for higher-energy ions as well. The angular plot at 0750:38 is consistent with a low-energy flow from about the ram direction (though in this case as the craft is in the shadow, the spacecraft potential is negative), but there are peaks in the

7 of 12

E00B26

SZEGO ET AL.: O+ ION FLOW AT VENUS

E00B26

Figure 4c. The magnetic field vectors projected on the VSO y-z plane. The length of the vectors is proportional to Btot. The blue lines show the bow shock and magnetic pileup boundary. Along the orbit the O+ ion counts are shown in color. azimuth elevation plots from more than one direction as well. (This, however, might be due to ions reflected from the sc body or from the solar arrays.) The mantle, outbound, is different from the inbound mantle (the magnetic field never piles up for that transition), the ions are more energetic and the tail boundary overlaps with a distant mantle. Outbound the sc crossed the boundary almost perpendicularly. [24] Next we present the analysis of four more flybys, 20 November, 26 November, 20 December, and 30 December 2006. The altitudes of the pericenter were 26.5 km at 0632 (6.54) UT, 335 km at 0647 (6.79) UT, 381 km at 0748 (7.8) UT, and 263 km at 0745 (7.75) UT, respectively. We selected these, because they are at the flank of the tail, and this way we can compare the flybys through the flank regions with a central one. Whereas geometrically the first two and the last two orbits were close, magnetically they were not, because the vSW  B directions are ( 0.08, 0.195, 0.98), (0.40, 0.91, 0.06), (0.37, 0.78, 0.5), and ( 0.10, 0.46, 0.88), respectively. In Figure 5a we exhibit four more time-energy plots for O+ ions collected along these orbits, the corresponding trajectories in VSO and in VMSO are shown in Figure 5b, and the variations of the magnetic fields along the orbits are shown in Figure 5c (the black lines show Bx, whereas the gray lines exhibit the variations of Btot). On these orbits the inbound/outbound crossings of the tail boundaries occurred at about 0625/ 0652 UT, 0640/0720 UT, 0740/0825 UT, and at about 0738/ 0820 UT. As it can be seen, on 20 November the current sheet was crossed right at the inbound crossing of the tail, and no further crossing occurred in the tail. On 26 November the current sheet was crossed three times around the entrance point, on 20 and 26 December the current sheet crossings took place in the middle of the tail flyby. [25] On 20 November we observe a strong O+ signal near 0635 UT. We analyze this further using the 3d velocity distribution plots collected in each 192 s shown in Figure 5d. The 3-D velocity distributions are projected onto the x-y, x-z,

and y-z planes, respectively (from the non obscured part of the sensor); however, these are not yet corrected for spacecraft velocity. These plots exhibit the velocity interval between 50 and 50 km/s, in the VSO frame of reference. We can conclude that in the sheet (0630 UT) the ion velocities are low, in the 10 km/s range. The asymmetry seen is due to the sensor obscuration. At 0635 UT when we observe the second ion peak in the time-energy plot, the velocity distribution show ions of higher velocity; the origin of these ions is definitely outside of the sheet, but not yet clarified. Similar peaks outside the sheet are observed on other days as well. [26] On 20 December the current sheet was crossed at 0811 UT (8.18 UT); and on 30 December it was crossed twice, at 0813 (8.22) and 0826 (8.43) UT, indicating that the current sheet can be wavy. Intense cold O+ ion beams are associated with these crossings. On all the orbits the outbound magnetic barrier is distinguishable, but characteristically different from the inbound one. The tail boundaries are distinguishable as sharp, higher-energy signals. There is no easy pattern to describe the higher-intensity fluxes which frequently are not correlated with current sheet crossing. This indicate, that the current sheet is not the only ion outflow location, though long-term averages of the ion outflow certainly preferential at current sheet crossings, as shown by Barabash et al. [2007b]. [27] The variability of the O+ spectra, in correlation with variations in the electron data indicate that the tail plasma might have ‘‘ray-like’’ structures, these will be analyzed later. It is certainly to be investigated what is the flow directions of the various ion species in the tail.

3. Discussion and Conclusions [28] In this paper we analyzed the plasma properties in the tail of Venus, using the IMA sensor data of the ASPERA-4 instrument, focusing on the O+ ions. We have used data taken between 20 November and 31 December

8 of 12

E00B26

SZEGO ET AL.: O+ ION FLOW AT VENUS

E00B26

Figure 5a. Energy versus time spectrograms with color coded O+ energy fluxes derived from ASPERA-4 IMA data for 20 and 26 November and 20 and 30 December below the obstacle boundary. We used the calibration method of Fedorov (internal team report, the IMA bible, 2007) and also a weak noise reduction. Since the data were not corrected for the effects of crosstalk, some O+ counts appeared at solar wind energies: we excluded these from further calculations. 2006, but we also looked into data taken half a Venus year before and after; those data show basically the same features reported here. We have investigated that portion of the orbits when the craft was below the obstacle boundary nightside, as witnessed by the biteout of magnetosheath electrons. The objective of the investigation was to identify the characteristic plasma structure of the tail that is identifiable despite the high orbit-to-orbit variations. Because of the limitations of the data processing mentioned above, we confine the discussion to the most basic properties of the tail plasma to its boundary structure. [29] It is somewhat surprising that though the pericenter altitude was at about 300– 400 km, we did not see any clear signature in the ion data that the craft entered into the ionosphere. In the electron data the ionosphere is seen by the characteristic electron excitation of the ionospheric O+ ions (S. M. E. Tsang et al., Ionospheric photoelectrons at Venus: Case, studies and first observations in the tail, submitted to Journal of Geophysical Research, 2008). The ionopause crossing could be derived from the drop of Btot. Very likely the spacecraft dipped into the ionosphere for brief periods shorter than our data collection cycle, 192 s. During the PVO flybys, from the analysis of the median

height profiles of the Orbiter Retarding Potential Analyzer it was concluded that the ion density between 90° and 150° solar zenith angle (SZA) was about 1000 ion/cc at 800 km altitude, and the ionopause altitude near 90° SZA was about 800– 1000 km [Brace and Kliore, 1991; Zhang et al., 2007]. Even though we are in a different solar activity phase, we would have expected larger count rates than those observed by IMA, unless the sensor saturation causes these lower rates; this could be substantiated only at a later stage of data processing. It is also known from the PVO data that the relative ion composition for the antisolar ionosphere is dominantly O+; hence, the ions we observe there must be dominantly of planetary origin. We need a better understanding of the spacecraft potential to clarify this discrepancy. Using only the ASPERA-4 data, in this orbit we cannot resolve the difference between the mantle and the ionosphere. [30] All of our data confirm that the tail plasma is very variable. The electron density profiles measured by PVO in the ionotail at 2000 – 3000 km range typically showed large enhancements separated by deep troughs. The time resolution of ASPERA-4 does not allow comparable spatial ion resolution, but the large variations in the O+ ion counts are

9 of 12

E00B26

SZEGO ET AL.: O+ ION FLOW AT VENUS

Figure 5b. The orbits of the four flybys in the Pioneer Venus Orbiter frame of reference as viewed toward the Sun. The gray sphere is Venus; the large dots indicate the entry/ exit points into the tail. The black arrows show the nominal current sheet directions.

E00B26

consistent with the PVO observations; indicating a ray-like ion structure in the region we investigate. [31] The most characteristic structures seen during the various flybys are the following: [32] 1. The continuation of the mantle in the wake; the near-Venus and far-Venus mantle crossings; the inbound being significantly more intense in low-energy counts, outbound the ions are more energetic than inbound. This is very likely connected with how ions in the tail mantle develop with increasing distance. [33] 2. The inbound and outbound crossings of the tail boundary were sharp, characterized by less intense but higher-energy ion beams. [34] 3. In most cases we observed the crossing of the current sheet within the central tail region where the lowenergy ion counts peak. [35] 4. In several cases the change of the sign of Bx is correlated with high-energy O+ ion peaks. [36] 5. A further important feature we have observed is the ray-like structure of the tail, regions with higher flux are separated by regions of very low flux; and the ion outflow outside of the sheet probably not negligible. These characteristic features together do appear roughly in most of the cases we have analyzed, the rest is more variable. [37] The central tail was characterized by the Venera investigators as a cavity. As opposed to this, in the lowenergy ion spectra we encountered relatively high-flux plasma what the Venera investigators could not detect

Figure 5c. Magnetic fields are shown for the four flybys. Btot is the gray line; Bx is the black line. On these orbits the inbound/outbound crossings of the tail boundaries occurred at about 0625/0652 (6.42/6.87) UT, 0640/0720 (6.66/7.33) UT, 0740/0825 (7.66/8.42) UT, and at about 0738/0820 (7.63/8.33) UT, respectively. 10 of 12

E00B26

SZEGO ET AL.: O+ ION FLOW AT VENUS

E00B26

Figure 5d. 3-D velocity distributions projected on to the x-y, x-z, and y-z planes, respectively (from the nonobscured part of the sensor), for 20 November 2006. It was evaluated for the velocity interval between 50 and 50 km/s, in the VSO frame of reference. The 16  8 directions are all the anodes and sectors of IMA; the individual points are the energy channels of the instrument (projected onto the respective planes). because of the energy limit of their sensors. Though these findings are not completely surprising, ASPERA-4 not only enriches but firmly establishes some of the previous results in a region never explored before; and our data have been derived in a much broader geometrical volume since the

VEX orbits scan through the whole tail. It is also important that the tail region is scanned from two opposite directions half a Venus year apart, because this allows us to use the change in the spacecraft velocity direction to sort out properties of the low-energy plasma.

11 of 12

E00B26

SZEGO ET AL.: O+ ION FLOW AT VENUS

[38] In the tail there are signatures of pickup ions, the discussion of those is deferred to later work. The pickup ions either originate from the neutral O corona of Venus; the hot O density being 3.4 1010/m3 at 400 km with a scale height of 400 km [Nagy et al., 1981], or they are ions leaked out of the ionosphere. [39] There are questions we shall answer in subsequent papers. We crossed the mantle twice; this allows clarifying how the mantle plasma develops in the tail as we move farther out. The physical nature of the mantle-tail boundary also requires further investigation, as well as how the planetary ions in the tail reach the observed energies. The location of the central tail region is also variable. Some of the reasons must be connected to the temporal variations in the solar wind, but the other participating partner in the interaction, Venus, might have some share as well.

References Barabash, S., et al. (2007a), The Analyser of Space Plasmas and Energetic Atoms (ASPERA-4) for the Venus Express mission, Planet. Space Sci., 55, 1772 – 1792, doi:10.1016/j.pss.2007.01.014. Barabash, S., et al. (2007b), The loss of ions from Venus through the plasma wake, Nature, 450, doi:10.1038/nature06434. Brace, L. H., and A. J. Kliore (1991), The structure of Venus ionosphere, Space Sci. Rev., 55, 81 – 164, doi:10.1007/BF00177136. Coates, A. J., et al. (2007), Ionospheric photoelectrons at Venus: Initial observations by ASPERA-4 ELS, Planet. Space Sci., 56, 802 – 806. Fra¨nz, M., E. Dubinin, E. Roussos, J. Woch, J. D. Winningham, R. Frahm, A. J. Coates, A. Fedorov, S. Barabash, and R. Lundin (2006), Plasma moments in the environment of Mars, Space Sci. Rev., 126, 165 – 207, doi:10.1007/s11214-006-9115-9. Hartle, R., and J. Grebowski (1995), Planetary loss from light ion escape on Venus, Adv. Space Res., 15, 117 – 122, doi:10.1016/0273-1177(94) 00073-A. Intriligator, D. S., J. H. Wolfe, and J. D. Mihalov (1980), The Pioneer Venus Orbiter plasma analyzer experiment, IEEE Trans. Geosci. Remote Sens., GE-18, 39 – 43, doi:10.1109/TGRS.1980.350258. Knudsen, W. C., K. Spenner, J. Bakke, and V. Novak (1980), Pioneer Venus Orbiter planar Retarding Potential Analyzer plasma experiment, IEEE Trans. Geosci. Remote Sens., GE-18, 54 – 59. McComas, D. J., H. E. Spence, C. T. Russell, and M. A. Saunders (1986), The average magnetic field draping and consistent plasma properties of the Venus magnetotail, J. Geophys. Res., 91, 7939 – 7953, doi:10.1029/ JA091iA07p07939.

E00B26

Nagy, A. F., et al. (1981), Hot oxygen atoms in the upper atmosphere of Venus, Geophys. Res. Lett., 8, 629 – 632, doi:10.1029/GL008i006p00629. Phillips, J. L., and D. McComas (1991), The magnetosheath and magnetotail of Venus, Space Sci. Rev., 55, 1 – 80, doi:10.1007/BF00177135. Russell, C. T. (1999), Magnetic stress in solar system plasma, Aust. J. Phys., 52, 733 – 751. Saunders, M. A., and C. T. Russell (1986), Average dimension and magnetic structure of the distant Venus magnetotail, J. Geophys. Res., 91, 5589 – 5604. Slavin, J. A., D. S. Intriligator, and E. J. Smith (1989), PVO magnetic field and plasma observations in the Venus magnetotail, J. Geophys. Res., 94, 2383 – 2398, doi:10.1029/JA094iA03p02383. Slavin, J. A., et al. (2008), MESSENGER observations of the solar wind interaction with Mercury, paper A-04431 presented at General Assembly, Eur. Geosci. Union, Vienna, 14 – 18 April. Spenner, K., W. C. Knudsen, K. L. Miller, V. Novak, C. T. Russell, and R. C. Elphic (1980), Observation of Venus mantle, the boundary region between solar wind and ionosphere, J. Geophys. Res., 85, 7655 – 7663, doi:10.1029/JA085iA13p07655. Vaisberg, O., A. Fedorov, F. Dunjushkin, A. Kozhukhovsky, V. Smirnov, L. Avanov, C. T. Russell, and J. G. Luhmann (1995), Ion populations in the tail of Venus, Adv. Space Res., 16, 105 – 118, doi:10.1016/02731177(95)00217-3. Vo¨ro¨s, Z., T. L. Zhang, M. P. Leubner, M. Volwerk, M. Delva, W. Baumjohann, and K. Kudela (2008), Magnetic fluctuations and turbulence in the Venus magnetosheath and wake, Geophys. Res. Lett., 35, L11102, doi:10.1029/ 2008GL033879. Zhang, T. L., J. G. Luhmann, and C. T. Russell (1991), The magnetic barrier at Venus, J. Geophys. Res., 96, 11,145 – 11,153, doi:10.1029/ 91JA00088. Zhang, T. L., et al. (2006), Magnetic field investigation of the Venus plasma environment: Expected new results, Planet. Space Sci., 54, 1336 – 1343, doi:10.1016/j.pss.2006.04.018. Zhang, T. L., et al. (2007), Little or no solar wind enters Venus atmosphere at solar minimum, Nature, 450, 654 – 656, doi:10.1038/nature06026. S. Barabash, Swedish Institute of Space Physics, Box 812, SE-98128 Kiruna, Sweden. Z. Bebesi, Z. Dobe, and K. Szego, KFKI Research Institute for Particle and Nuclear Physics, Konkoly Thege Miklo´s u´t 29-33, H-1121 Budapest, Hungary. ([email protected]) A. J. Coates, Mullard Space Science Laboratory, University College London, London RH5 6NT, UK. ([email protected]) A. Fedorov, Centre d’Etude Spatiale des Rayonnements, BP-4346, F-31028 Toulouse, France. ([email protected]) M. Fra¨nz, MPI fu¨r Sonnensystemforschung, Max-Planck-Strasse 2, D-37191 Katlenburg-Lindau, Germany. ([email protected]; [email protected]) T. L. Zhang, Space Research Institute, Austrian Academy of Sciences, Schmiedlstrasse 6, Graz, Austria. ([email protected])

12 of 12

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114, E00B30, doi:10.1029/2008JE003174, 2009

for

Full Article

Plasma environment of Venus: Comparison of Venus Express ASPERA-4 measurements with 3-D hybrid simulations C. Martinecz,1 A. Boesswetter,2 M. Fra¨nz,1 E. Roussos,1 J. Woch,1 N. Krupp,1 E. Dubinin,1 U. Motschmann,2,3 S. Wiehle,2 S. Simon,2 S. Barabash,4 R. Lundin,4 T. L. Zhang,5 H. Lammer,5 H. Lichtenegger,5 and Y. Kulikov6 Received 23 April 2008; revised 15 September 2008; accepted 30 December 2008; published 7 March 2009.

[1] We use data of the ASPERA-4 ion and electron spectrometers onboard Venus Express

to determine the locations and shapes of the plasma boundaries (bow shock, ion composition boundary, and mantle) at Venus. We also investigate the variation of the terminator bow shock position as a function of the solar wind dynamic pressure and solar EUV flux. We compare the results with a 3-D hybrid simulation. In the hybrid model, ions are treated as individual particles moving in self-consistently generated electromagnetic fields and electrons are modeled as a massless charge neutralizing fluid. The planetary heavy ion plasma is generated by an oxygen ionosphere and exosphere adapted to a profile, which depends on the solar zenith angle (Chapman layer). A comparison between observations and simulations indicates that the hybrid model is able to produce an adequate picture of the global plasma environment at Venus. The positions of the plasma boundaries are well reproduced by the model but a significant disagreement appears in the absolute values of the considered parameters. Citation: Martinecz, C., et al. (2009), Plasma environment of Venus: Comparison of Venus Express ASPERA-4 measurements with 3-D hybrid simulations, J. Geophys. Res., 114, E00B30, doi:10.1029/2008JE003174.

1. Introduction

1.1. Knowledge Based on Observations [2] For almost 50 years, planet Venus is an attractive target in planetary space science and thus, was the focus of several space missions (Mariner, Venera, and Vega) and ground-based observations. Although we owe most of our current knowledge of the solar wind interaction with Venus to the long-lasting Pioneer Venus Orbiter (PVO) mission (1978 – 1992), providing observations over more than a complete solar cycle, a lot of questions remain unanswered. Venus Express (VEX), which is the first European mission to planet Venus [Titov et al., 2006], aims at a comprehensive investigation of the Venusian atmosphere and its plasma environment in order to fill the gaps left by PVO. One of the central questions, addressed by the plasma instrument 1 Max Planck Institut fu¨r Sonnensystemforschung, Katlenburg-Lindau, Germany. 2 Institut fu¨r Theoretische Physik, Technische Universita¨t Braunschweig, Braunschweig, Germany. 3 Deutsches Zentrum fu¨r Luft- und Raumfahrt, Institut fu¨r Planetenforschung, Berlin, Germany. 4 Swedish Institute of Space Physics, Kiruna, Sweden. 5 Space Research Institute, Austrian Academy of Sciences, Graz, Austria. 6 Polar Geophysical Institute, Russian Academy of Sciences, Murmansk, Russia.

Copyright 2009 by the American Geophysical Union. 0148-0227/09/2008JE003174$09.00

Analyzer of Space Plasmas and Energetic Atoms (ASPERA-4) onboard VEX, is the evolution of the Venus atmosphere under the combined effects of escape and interaction with the solid planet [Barabash et al., 2007a]. The basic morphology of the ionosphere and the plasma environment of Venus have been studied by PVO and a detailed discussion of the results can be found in the work by Luhmann and Cravens [1991] and Phillips and McComas [1991]. [3] Venus does not possess an intrinsic magnetic field to shield the upper atmosphere against the incoming solar wind flow. The lack of such a magnetic cavity results in an highly structured plasma environment, similar observed at Mars and comets, characterized by the direct interaction of the solar wind with the top of the ionosphere. This leads to atmospheric escape processes, solar wind induced current systems and a complex nightside ionosphere consisting of tail rays, filaments, ionospheric holes and plasma clouds. [4] Owing to the supersonic and super-Alfve´nic solar wind a bow shock is formed upstream of the planet in order to heat, slow down and deflect the solar plasma flow around the obstacle. Solar radiation ionizes the neutral atmosphere of Venus creating a substantial ionosphere. The upper boundary of the ionosphere, known as ionopause, is characterized by a sharp gradient in the electron density and forms at an altitude where the thermal plasma pressure of the ionosphere equals incident solar wind pressure. The ionopause was found to be located about 300 km above the surface in the subsolar region and to have an altitude about 1000 km

E00B30

1 of 15

E00B30

MARTINECZ ET AL.: PLASMA ENVIRONMENT OF VENUS

near the terminator at solar maximum [Phillips and McComas, 1991]. The region between the fast magnetosonic shock wave and the ionopause is called the magnetosheath, where an increased and turbulent magnetic field behavior is observed. Since the interplanetary magnetic field (IMF) is frozen into the magnetosheath flow, it piles up at the dayside of the obstacle producing a so-called magnetic barrier [Zhang et al., 2008b; Bertucci et al., 2003] as it is carried around the planet. The magnetic barrier is also called plasma depletion layer. The solar wind plasma is significantly excluded from this region and the magnetic pressure dominates over the incident ram pressure. The IMF lines drape around the obstacle and as a result, create an induced magnetotail consisting of two lobes of opposite polarity separated by a plasma sheet [Luhmann and Cravens, 1991]. Observations revealed another boundary layer, located between the magnetosheath and the ionopause, the so-called mantle region or transition zone, characterized by the presence of solar wind protons and planetary ions [Phillips and McComas, 1991]. Above the upper boundary of the mantle the shocked solar wind plasma is found whereas beyond the lower boundary of the mantle planetary ions become the main particle population (ion composition boundary). 1.2. Numerical Models [5] The global Venus-solar wind interaction can be studied by different numerical investigations such as gasdynamic, magnetohydrodynamic (MHD) and kinetic models. [6] In general, a gas dynamic model [Spreiter and Stahara, 1992] is able to reproduce some of the global characteristics of the interaction (bow shock, magnetosheath region) but it neglects the individual particle behavior as well as the effects of the magnetic field on the plasma dynamics. Thus, this fluid approximation does not include the magnetic barrier in a self-consistent manner and, additionally, does not consider the exospheric ion pickup process owing to the convective electric field [Zhang et al., 1993]. [7] More sophisticated treatments can be done using MHD models because here the magnetic field is taken into account in a self-consistent manner [Wu, 1992; Tanaka, 1993; Murawski and Steinolfson, 1996a, 1996b; Tanaka and Murawski, 1997; Kallio et al., 1998; Bauske et al., 1998]. The position and shape of the bow shock, the formation of the magnetic barrier, the magnetosheath and the magnetotail could be reproduced in these studies and are in a good agreement with observational data (e.g., from PVO). While MHD simulation provide an adequate picture of the large-scale solar wind interaction with Venus, including the mass loading of the solar wind with heavy oxygen ions, several important kinetic effects are not considered in these studies owing to the single fluid description. Thus, asymmetries in the form of the bow shock, the magnetic barrier intensity and the planetary pickup distribution as well as the formation of the ion composition boundary (separation between solar wind plasma and planetary plasma) and multiple shocklets (explained in section 4) cannot be reproduced by MHD models. [8] The gyroradii of the solar wind protons are in the range of several hundred kilometers, and therefore comparable with the characteristic scales of the subsolar interaction region. The gyroradii of the heavy pickup ions are in

E00B30

the range of several thousand kilometers, and therefore comparable with the radius of the planet. Thus, a kinetic treatment seems to be mandatory and is used here. A full particle code would be the most suitable numerical investigation but with the current computational resources, a fully kinetic approach is not feasible. Hence, a hybrid model is a very useful tool for studying the kinetic effects because it treats ions as gyrating particles. [9] Several hybrid simulations were carried out over the past years and, in general, demonstrated a good qualitative agreement between the simulation results and observational data. [10] The work by Moore et al. [1991] is one of the first important studies investigating the solar wind interaction with the dayside of Venus by means of a three-dimensional hybrid model. Their simulation results provided an asymmetrical bow shock depending on the upstream IMF orientation as well as asymmetries in the magnetic barrier location and field intensity. Furthermore, they found that the finite gyroradius of the planetary pickup ions causes an asymmetry in number density along the direction of the motional electric field. [11] The results of a three-dimensional hybrid simulation from Brecht and Ferrante [1991], including the ionosphere by assuming the planet as a conducting sphere, showed that the bow shock and magnetic barrier are asymmetrical at Venus. [12] In the hybrid model of Shimazu [2001] the Venusian planet was treated as an ionized gaseous body with uniform and constant supply of plasma. The simulation results provided an asymmetric bow shock with a multiple-shock structure, a magnetic barrier in front of the planet with asymmetries along the solar wind electric field, a magnetotail, tail rays and a plasma sheet. They found that the calculated asymmetry in the magnetic barrier intensity was consistent with observations, but the direction of the calculated asymmetry in the shock size did not concur with the observed asymmetry. Additionally, they demonstrated that ions escape to the magnetotail through the tail rays and that the tail rays were connected with the plasma sheet. [13] Other hybrid approaches deal with the investigation of the ion escape processes with a particular emphasis placed on the processes occurring at the ionopause associated with the Kelvin-Helmholtz instability [Terada et al., 2002, 2004]. Their 2-D simulation model results yielded an asymmetrical convection pattern of the ionosphere due to the asymmetrical momentum transport across the ionopause and showed the dynamic nature of the interaction. Moreover they concluded that most probably, the dynamic ion removal process associated with the Kelvin-Helmholtz instability plays a significantly role in the ion escape from the planet. [14] Kallio et al. [2006] studied the solar wind interaction with Venus using a global three-dimensional self-consistent quasi-neutral hybrid model (QNH) and focused on the asymmetries in the direction of the convection electric field caused by kinetic effects, the role of the interplanetary magnetic field (IMF) x component and the properties of the escaping planetary O+ ions. On the one hand, they found a notable north-south asymmetry of the magnetic field and plasma due to ion finite gyroradius effects and escaping O+ ions. On the other hand they showed that an asymmetry between the quasi-perpendicular bow shock hemisphere and the quasi-parallel bow shock hemisphere, which results

2 of 15

E00B30

MARTINECZ ET AL.: PLASMA ENVIRONMENT OF VENUS

E00B30

Figure 1. ASPERA-4 data recorded on 15 July 2006, about an hour before and after the pericenter of that concerned orbit (1RV = 6051.8 km). The top shows the total counts of energetic electrons measured by the Electron Spectrometer (ELS) sensor and the middle and bottom present the total counts of the proton and heavy ion channels of the Ion Mass Analyzer (IMA) sensor, respectively. The heavy ion channel contains proton counts whenever the proton channel saturates. The black vertical arrows mark the plasma boundaries separating the different interaction regions (solar wind, magnetosheath, mantle, and ionosphere). from the IMF x component, exists. In general, the QNH model is able to reproduce the main observed plasma and magnetic field regions such as the bow shock, the magnetosheath, the magnetic barrier and the magnetotail. [15] In this study we investigate the plasma environment of Venus using plasma moments obtained by ASPERA-4 instrument onboard VEX in order to study the solar windatmosphere interaction of an unmagnetized planet and the results of this analysis is put in context of a numerical model. The paper is organized as follows. Section 1 deals with our identification of the plasma boundaries based on ASPERA-4 measurements and the dependency of the terminator shock distance on the solar wind ram pressure as well as on the solar EUV flux. Section 2 describes in short our 3-D hybrid model which we use to study the global plasma environment of Venus. In section 3 we present the simulation results for a specific orbit, compared with our observations along the VEX trajectory in the following section and finally, the paper closes with a conclusion.

2. Observations [16] For this paper we used data from the ASPERA-4 experiment onboard of Venus Express which consist of an

Electron Spectrometer (ELS), an Ion Mass Analyzer (IMA), a Neutral Particle Imager (NPI), and two Neutral Particle Detectors (NPD1 and NPD2). We refer the reader to Barabash et al. [2007a] for a detailed description of the plasma instrument. In this study we present data from the electron and ion spectrometers, exclusively. Figure 1 displays data obtained on 15 July 2006 showing the main plasma features of the solar wind interaction with Venus about 1 h before and after the closest approach of orbit 85. [17] Figure 1 (top) shows an energy spectrogram of measured electrons in the energy range of 0.1 eV – 20 keV obtained by ELS. But electrons below 5 eV are reflected to avoid saturation of the counters. The sensor has 16 anodes covering a total field of view of 4
360
. Shown are counts obtained during 4s sampling intervals integrated over anodes 5 – 15 of the sensor because anodes 0 – 4 are more noisy. The data shown in Figure 1 (middle and bottom) represent protons and heavy ions, respectively, measured by IMA, integrated over all 16 anodes and 16 spatial sectors covering a total field of view of 90
360
. Note that signatures above 50 eV energy in Figure 1 (bottom) in the solar wind and magnetosheath regions are not caused by heavy ions but by saturation of the proton channels. A spatial scan during 192 s by electrostatic deflection produces the

3 of 15

E00B30

MARTINECZ ET AL.: PLASMA ENVIRONMENT OF VENUS

Figure 2. This plot displays axisymmetric bow shock (BS), upper mantle boundary (UMB), and lower mantle boundary (LMB) fits derived using the first 19 months of ASPERA-4 observations in an aberrated Venus Solar Orbital coordinate system. The BS crossings (red circles) were fitted to a conic function. The UMB (green diamonds) and LMB (blue triangles) crossings were fitted by a circle on the dayside and by linear regression on the nightside.

repeatable pattern visible in the spectrogram. The x axis shows the distance, position and time of the spacecraft along the orbit. [18] First, VEX is located inside the solar wind before crossing the bow shock (BS) at 0115 UT, identified by an increase in counts of energetic electrons (E > 35eV) in the magnetosheath with respect to the solar wind. Passing the BS, the spacecraft enters the magnetosheath, characterized by the shocked, slowed down and heated solar wind. At 0148 UT, VEX crosses the upper mantle boundary (UMB), identified by a strong decrease in electron counts (E > 35eV), and is located in a so-called mantle region or transition zone, where we observe a mixture of solar wind protons and planetary ions. The lower boundary of the mantle (LMB), crossed at 0157 UT, we also call the ion composition boundary (ICB), because at this boundary the solar wind protons disappear and the planetary ions become the main population. LMB is identified in ELS by the appearance of ionospheric photoelectrons (E > 10eV). Passing the LMB, the spacecraft is located in the ionosphere between 0157 UT and 0201 UT. On the outbound pass, VEX crosses again all the mentioned plasma regions and boundaries, but in reverse order, i.e., at 0201 the LMB, at 0208 UT the UMB and at 0222 the BS, and finally, is back in the solar wind. [19] From Martinecz et al. [2008] we obtained initial determinations of the mean plasma boundary locations (BS and ICB) based on the first 5 months of observations. After VEX being almost 2 years in orbit around Venus, we now analyzed the first 19 months (14 May 2006 to 31 December 2007) of electron and ion data in order to obtain better fits based on better statistics (see Figure 2). Furthermore, we defined a third plasma boundary (UMB) because a

E00B30

so-called mantle region (or transition zone) is visible in the data, especially on the nightside, where we observe a mixture of solar wind and planetary plasma. The lower boundary of the mantle coincides with the former ICB. The upper boundary of the mantle marks the transition from the magnetosheath to the mantle region. [20] The fitting techniques used is described by Martinecz et al. [2008]. The new fit parameters obtained by this analysis is presented in this paper. For the BS we applied a conic function in polar coordinates, assuming cylindrical symmetry along the x axis (solar wind flow direction) which is least-squares fitted to the observed BS positions [r = L/(1 + cosJ)]. Accordingly, we got for the semilatus rectum L = 1.515 RV (s ± 0.024), the eccentricity = 1.018 (s ± 0.008), the conic focus x0 = 0.664 RV (s ± 0.039), and the terminator shock distance rtsd = 2.088 RV. The standard deviations are a result of the fitting procedure assuming that there is no error in the individual position determinations. Since the radial coverage in the subsolar region is not homogeneously sampled, we also corrected the BS fit by omitting crossings below 30
solar zenith angle. The new three-parameter BS fit is in good agreement with the previous fit by Martinecz et al. [2008] and with the BS model of Slavin et al. [1984] rather than with the twoparameter BS fit from Zhang et al. [2008a] based on magnetic field observations. [21] For modeling the positions of LMB and UMB we used circular fits (rLMB = 1.076 RV, rUMB = 1.130 RV) for the dayside observations (x > 0) and linear regressions [ y = k x + d] (kLMB = 0.122 RV, dLMB = 1.076 RV; kUMB = 0.101 RV, dUMB = 1.130 RV) for the nightside measurements (x < 0). Note that, currently we do not have data for the mantle region below about 50
solar zenith angle and thus, both fits provide boundaries which are too far away from the planet on the dayside. In order to get more realistic mantle fits it is necessary to include crossings in the subsolar region, expected later during the VEX mission. [22] Dubinin et al. [2008] determined the magnetospheric boundary (MB) and photoelectron boundary (PEB) at Mars based on ASPERA-3 and MARSIS observations on board Mars Express. MB is identified by a strong decrease (inbound pass) or increase (outbound pass) in energetic electron density. PEB marks the outer boundary of the ionosphere which is identified by the appearance of energy peaks in the range between 20 and 30 eV in the electron spectrograms. Consequently, the physics of UMB and LMB can be compared with MB and PEB, respectively. However, the upper and lower boundary positions of the magnetic barrier determined by Zhang et al. [2008b] on the basis of the magnetometer observations on board VEX are partly in agreement with the UMB and LMB locations. At the terminator the average thickness of the mantle region is around 500 km and hence, 300 km less thick than the magnetic barrier. The thickness of the magnetic barrier in the subsolar region is around 200 km [Zhang et al., 2008b] but currently we cannot estimate this thickness for the mantle region owing to the lack of mantle crossings below around 50 solar zenith angle. [23] Additionally, Martinecz et al. [2008] presented the variation of the BS position at the terminator as a function of the solar wind dynamic pressure. Owing to improved

4 of 15

E00B30

MARTINECZ ET AL.: PLASMA ENVIRONMENT OF VENUS

E00B30

strength since the BS position should depend on the upstream Mach number [Zhang et al., 2008a].

3. Hybrid Model

Figure 3. The dependence of the BS position at the terminator on the dynamic pressure of the solar wind derived from ASPERA-4 measurements. All BS crossings (plus signs) were extrapolated to the terminator plane using a conic section curve with a fixed eccentricity ( = 1.018) and a fixed focus (x0 = 0.664) and with a variable semilatus rectum. No normalization has been applied to the data. The circles represent median values over pressure bins.

plasma moment calculations as well as a larger data set under consideration in this study, we decided to carry out the investigation again. Figure 3 demonstrates even better the independence of the BS position from the ram pressure during solar minimum conditions where the plus signs are the observed BS crossings extrapolated to the terminator plane using the conic section and the circles represent the median values. [24] From PVO observations we learned that the solar wind interaction with Venus is very dependent on the phase of the solar cycle [Russell et al., 2006]. During solar minimum the BS is found to be closer to the planet than during solar maximum owing to lower ionization and ion pickup rates caused by EUV flux changes over the 11-year solar cycle. Figure 4 shows the terminator BS position as a function of the solar EUV flux. Since the VEX spacecraft does not carry a solar UV monitor we here use the integrated flux of photons/cm 2 s 1 (F50) observed in the 0.1– 50 nm wavelength range by the Solar Extreme Ultraviolet Monitor (SEM) on board the Solar Heliospheric Observatory (SOHO) spacecraft (data from http://www. usc.edu/dept/space science/sem data) and shift the observation time according to the difference in solar longitude between the SOHO and Venus. This monitor is a highly stable photodiode spectrometer that continuously measures the full solar disk absolute photon flux at the He II 30.4 nm line as well as the absolute integral flux between 17 and 70 nm. Although this index is an excellent indicator of overall solar activity levels, we do not observe yet an effect on the terminator BS location in our data set, because the EUV flux variation is small over the period of observation as expected for solar minimum. [25] We attribute the large variations visible in the bow shock locations to the variations in the magnetic field

[26] The results of our data analysis is put in context of numerical investigations by means of a 3-D hybrid model which has been originally developed by Bagdonat and Motschmann [2002a, 2002b] to study the solar wind interaction with weak comets. Furthermore, the present version of this hybrid code has already been successfully applied for modeling the solar wind interaction with magnetized asteroids [Simon et al., 2006a], for studying the plasma environment of Mars [Bo¨ßwetter et al., 2004, 2007] and Titan [Simon et al., 2006b, 2007], as well as for the simulation of Rhea’s magnetospheric interaction [Roussos et al., 2008]. [27] In the hybrid model the ions are treated as individual particles moving in self-consistently generated electromagnetic fields, whereas the electrons are modeled as a massless charge-neutralizing fluid. For a detailed description of the code we refer the reader to the earlier studies just mentioned above. In the following we will only explain the main basic features of the model for completeness of this paper. [28] As already mentioned, the ions are treated as individual particles in order to cover ion dynamics. Hence, the code solves the equation of motion for the ions: and

d~ xs ¼~ vs dt d~ vs qs ~ E þ~ vs ¼ dt ms

~ B

ð1Þ

kD nn ð~ vs

~ un Þ

ð2Þ

xs and ~ vs are the charge, mass, position and where qs, ms, ~ velocity of an individual particle of species s, respectively;

Figure 4. This plot shows the terminator BS position as a function of solar EUV flux (F50 index, 0.1– 50 nm integrated photons cm 2 s 1) derived from Solar Heliospheric Observatory and Solar Extreme Ultraviolet Monitor observations and shifted to Venus. All BS crossings (asterisk signs) were extrapolated to the terminator plane by means of the conic function.

5 of 15

E00B30

MARTINECZ ET AL.: PLASMA ENVIRONMENT OF VENUS

E00B30

[30] We are using two different electron pressure terms in order to consider the different electron temperatures in the solar wind and ionosphere, respectively. Both electron populations are assumed to be adiabatic: Pe;SW ¼ be;SW

ne;SW n0

Pe;HI ¼ be;HI

ne;HI n0

k

k

ð5Þ

ð6Þ

where k is the adiabatic exponent which has a value of k = 2 instead of 5/3, because the thermodynamic coupling is only effective in the two dimensions perpendicular to the field [Bo¨ßwetter et al., 2004]. Finally, the set of hybrid equations is completed by Faraday’s law @~ B ¼r @t

Figure 5. The neutral number density (green line) and the ion production rate (red line in subsolar direction and blue line at the terminator) of oxygen as a function of altitude at solar minimum (ionospheric photoionization frequency n = 4.55 10 7 s 1 [Torr and Torr, 1985]). whereas ~ E and ~ B are the electromagnetic field quantities. The first term on the right in equation (2) expresses the Lorentz term and the second the drag force, which takes into account collisions of ions and neutrals (kD is a constant un are the taken from Israelevich et al. [1999]). nn and ~ number density and bulk velocity of the neutrals. We use ~ un = 0. [29] The electron fluid description is based on the momentum equation 0 ¼ ne me

d~ ue ¼ dt

E þ~ ue ene ~

~ B

rPe;sw

rPe;hi

ð3Þ

from which the electric field can be derived ~ E¼ ~ ui

~ Bþ

r

~ ~ B B m0 ene

rPe;sw þ rPe;hi ene

ð4Þ

where ~ ui denotes the mean ion velocity. The mean ion density (ni) is equal to the electron density (ne) owing to the assumption of quasi-neutrality.

" ~ ui

~ B

r

r

# ~ ~ B B m0 ene

ð7Þ

expressing the time evolution of the magnetic field. [31] The simulation is applied to a curvilinear grid in three spatial dimensions which allows high spatial resolution in the vicinity of the planetary atmosphere. This so-called Fisheye Grid is obtained from an equidistant Cartesian grid by means of a nonlinear coordinate transformation. For the simulations presented in this study the box size was chosen 6 RV with 101 grid cells in each spatial direction. By using this grid one obtains a spatial resolution of about 100 – 150 km per grid cell in the vicinity of the obstacle. Beyond the distorted grid (equidistant cartesian grid) a resolution of 250– 300 km per grid cell is achieved. A detailed description of the grid generation is given by Bo¨ßwetter et al. [2004] and Simon et al. [2006b]. [32] The Venusian atmosphere is modeled as a spherical symmetric gas cloud around the planet consisting of atomic oxygen which is the dominant neutral in the upper atmosphere. Radial density distribution includes atmospheric and exospheric exponential profiles. The profile for the cold oxygens is based on the model from Kulikov et al. [2006] and the hot oxygen profile is based on PVO measurements compared with model calculations obtained by Nagy et al. [1990]. The heavy planetary oxygen ions are incorporated into the simulation by means of a so-called Chapman production function (see Figure 5), that means, the local ion production rate q depends on the altitude above the surface as well as on the solar zenith angle. This production function results from the ionization of the neutral profile by means of solar EUV radiation (photoionization frequency for atomic oxygen, measured at Earth and scaled to the heliocentric distance of Venus). Other ion production processes, like electron impact ionization and charge exchange are not taken into account in our model. On the nightside of the planet, a constant ion production function was assumed, yielding about 10% of the dayside production rate. Consequently, one obtains a dense ionosphere and a less dense exosphere. [33] Any ion hitting the so-called inner boundary at an altitude of 150 km above the planetary surface is removed from the simulation. No boundary conditions are imposed

6 of 15

E00B30

MARTINECZ ET AL.: PLASMA ENVIRONMENT OF VENUS

E00B30

Table 1. Input Parameters for Simulation Run for VEX Orbit 85 on 15 July 2006 Parameter

Values

Comments

Solar wind density, nsw Solar wind velocity, vsw Magnetic field magnitude, Bsw IMF orientation, 8sw Electron beta, b e,sw Proton beta, b i,sw Ionospheric electron beta, b e,hi Alfe´n veloctiy, vA Alfe´nic mach number, MA Ionospheric photoionization frequency, n smin Total ion production, QO+ Lower boundary of atmosphere Upper boundary of atmosphere

5.0 cm 3 360.0 km s 1 5.0 nT 22.5 deg 0.81 1.61 0.02 48.77 km s 1 7.38 4.55 10 7 s 1 1.037 1026 s 1 1 RV + 150.0 km 2 RV

— in x direction in y direction in VSO Te,sw = 10.0 eV Ti,sw = 20.0 eV Te,hi = 0.3 eV [Miller et al., 1980] — dimensionless [Torr and Torr, 1985] Chapman production function RV = 6051.8 km —

on the electromagnetic fields, which means, the equations for ~ E and ~ B are solved outside as well as inside of the obstacle. An artificial inner density is assumed and increased while the simulation proceeds in order to match the surrounding ionospheric heavy ion density at the subsolar point as well as to avoid electric fields arising from density gradients [Bo¨ßwetter et al., 2007].

4. Simulation Results for the VEX Orbit 85 [34] Table 1 shows a list of input parameters mainly measured on orbit 85 (15 July 2006) by ASPERA-4 and Magnetometer [Zhang et al., 2006] onboard VEX and the rest of the parameters are taken from literature. The plasma moments are estimated either by integrating over a given range in the phase space (integration method), or by assuming a Boltzmann distribution for the phase space density (fitting method) [Fra¨nz et al., 2007]. The ionospheric electron temperature is the only parameter which was taken from the literature [Miller et al., 1980] (based on PVO observations) owing to the cutoff energy in the ELS sensor. A deflection voltage of about 5 eV prevents the measurement of low-energy electrons (spacecraft photoelectrons) in order to avoid saturation of the counters. [35] The simulation box has three spatial dimensions x, y, z 3 RV) in which the undisturbed solar ( 3 RV wind flows in the positive x direction, the convection electric field (~ E= ~ v ~ B) points in the negative z direction and thus, the background magnetic field completes the righthanded system pointing in the positive y direction (tilted 157.5
in the equatorial plane), as shown in Figure 6. Note that the positive y axis in the simulation coordinate system points into the direction of orbital motion while in the Venus Solar Orbital (VSO) system it is orientated opposite to the orbital motion. For practical reasons in discussing the simulation results we denote the hemisphere to which the solar wind electric field is pointing as the E+ hemisphere (southern hemisphere) and the other as E hemisphere (northern hemisphere). [36] Figures 7 and 8 represent global 3-D views of the simulation results for the magnetic field configuration and heavy ion density, respectively. The cutting planes through the simulation box are taken at x = 0 (terminator plane), y = 0 (polar plane) and z = 0 (equatorial plane). [37] In Figure 7, showing the magnetic field configuration at Venus, the solar wind enters the simulation box from the left forming a magnetosonic shock wave upstream of

the planet. On the dayside the IMF piles up in front of the obstacle, drapes around Venus and creates an induced magnetotail consisting of two lobes on the nightside. In the equatorial plane one can see the dusk magnetic field lobe where the IMF points toward the planet (visible in Figure 11). [38] A global picture of the planetary plasma is shown in Figure 8. In the northern hemisphere (E ) one can see a very sharp ion composition boundary (ICB) whereas in the southern hemisphere (E+) this boundary is less pronounced because the heavy ions are accelerated along the convective electric field and dragged away from the planet (pickup process). The formation of the ICB is discussed in more detail below. Figures 9, 10, and 11, represent 2-D cuts of the 3-D simulation results at the cutting planes x = 0 (terminator

Figure 6. Cross sections of the simulation box at the plane x = 0 (terminator plane), y = 0 (polar plane), and z = 0 (equatorial plane) presented in section 4. The undisturbed solar wind flows in the positive x direction, the convection electric field (~ E= ~ v ~ B) points in the negative z direction, and the background magnetic field completes the righthanded system pointing in the positive y direction (tilted 157.5
in the equatorial plane).

7 of 15

E00B30

MARTINECZ ET AL.: PLASMA ENVIRONMENT OF VENUS

E00B30

Figure 7. Global 3-D view of simulation results showing the magnetic field strength (nT). The cutting planes through the simulation box are taken at x = 0 (terminator plane), y = 0 (polar plane), and z = 0 (equatorial plane). plane), y = 0 (polar plane) and z = 0 (equatorial plane), respectively. [39] Figure 9 shows from left to right the solar wind density nsw (cm 3), heavy ion density nhi (cm 3) and the magnetic field jBj (nT) along the terminator plane. The increased solar wind density indicates a bow shock formation in front of the obstacle due to the supersonic solar wind

Figure 8. Global 3-D view of simulation results showing the heavy ion density (O+) (cm 3). The cutting planes through the simulation box are taken at x = 0 (terminator plane), y = 0 (polar plane), and z = 0 (equatorial plane).

Figure 12. Projections of the trajectory of the Venus Express orbit 85 on the equatorial, polar, and terminator plane of our 3-D hybrid simulation. The undisturbed plasma flow is directed along the (+x) axis, the convection electric field ~ (~ E = ~ v B) is orientated antiparallel to the z axis, and the background magnetic field completes the right-handed system pointing in the positive y direction (tilted 157.5
in the equatorial plane).

*The numbering of Figures 9 – 12 has been corrected here. The article as originally published appears online.

8 of 15

E00B30

MARTINECZ ET AL.: PLASMA ENVIRONMENT OF VENUS

Figure 9. Simulation results in the terminator plane, i.e., looking from behind on Venus. The solar wind is orientated out of the plane (+x axis), the convective electric field is antiparallel to the (+z) axis and points away from the planet; thus, the background magnetic field completes the right-handed system (+y axis). Figure 10 displays (left) the solar wind density (nsw), (middle) the heavy ion density (nhi), and (right) the background magnetic field. An asymmetric bow shock, exhibiting a shocklet structure, is formed in front of the obstacle owing to the supersonic solar wind (left plot). The interplanetary magnetic field is draping around Venus generating an induced magnetotail on the nightside of the planet (right plot). See text for details.

Figure 10. Cut through the simulation box along the polar plane, i.e., looking from the western (dusk) side (in the sense of orbital motion) on Venus. The plasma flow comes in from the left (+x axis), the solar wind electric field points away from the planet ( z axis), and the interplanetary magnetic field completes the right-handed system pointing into the plane (+y axis). Figure 11 illustrates from left to right in the first row (a) the solar wind density (nsw), (b) the heavy ion density (nhi), (c) the background magnetic field, (d) the solar wind bulk velocity (jvswj), (e) heavy ion bulk velocity (jvhij), and (f) the convective electric field. In the density plots a clear separation between the solar wind and ionospheric plasma takes place forming the so-called ion composition boundary (ICB) in the northern hemisphere. In the southern hemisphere there is no clear ICB owing to the mass loading effect; that is, the planetary ions are accelerated in the direction of the convective electric field and are picked up by the solar wind flow. The solar wind electric field ~ E= ~ v ~ B vanishes where the heavy ion plasma dominates. See text for details. 9 of 15

E00B30

E00B30

MARTINECZ ET AL.: PLASMA ENVIRONMENT OF VENUS

flow (left). The fast magnetosonic shock wave exhibits several asymmetries. On the one hand, the shock geometry offers a small asymmetry with respect to the north – south direction due to the pickup of oxygen ions in the direction of the convective electric field. As a consequence, the massloading effect decelerates the plasma flow in the E+ hemisphere. On the other hand, the bow shock is also asymmetric with respect to the dawn-dusk direction which results from the chosen interplanetary magnetic field (IMF) angle of 22.5
in VSO (for symmetric case, 90
). Furthermore, the bow shock is not circular at the terminator plane because the propagation velocity of the fast magnetosonic wave is greatest in the direction of 90
to the magnetic field (positive z axis). Thus, the plasma is less compressed and requires a greater volume in order to flow around the obstacle in this direction [Russell et al., 1988]. Besides these asymmetries there are two more features visible in the solar wind density. Behind the bow shock, so-called shocklets [Omidi and Winske, 1990] or multiple shocks [Shimazu, 2001] occur which are kinetic shock substructures due to the finite gyroradius of the solar wind protons where the proton density locally increases (at the reversal point) and the proton velocity decreases at the same time. Last, the shock is very weak (or almost not present) on the left side of the obstacle owing to a quasi-parallel shock scenario. The shock normal is parallel to the IMF orientation (see jBj in Figure 11) so that solar wind protons can be reflected there and are gyrating back into the plasma flow. In the heavy ion density (Figure 9) one can see clearly an exospheric region (hot oxygen corona) which is in direct interaction with the oncoming plasma flow, as well as a weak ionosphere (cold oxygens) because of the low solar activity condition. The global configuration of the magnetic field is illustrated in the last plot in Figure 9. As already mentioned in section 1, Venus does not possess an intrinsic magnetic field and thus, the IMF lines are draping around Venus leading to an induced magnetotail on the nightside of the planet. [40] Figure 10 displays the simulation results of the solar wind density nsw (cm 3) and velocity jvswj (km s 1), heavy ion density nhi (cm 3) and velocity jvhij (km s 1), the magnetic field jBj (nT) and the solar wind electric field jEj (V km 1) in the polar plane. In the polar plane the simulation produces an almost symmetric flow of the shocked and slowed down solar wind around the obstacle. Additionally, a plasma wake is formed behind the planet where the proton density vanishes (proton cavity). [41] The solar wind bulk velocity is accelerated downstream above the north pole (E hemisphere) whereas downstream below the south pole (E+ hemisphere) it is significantly decelerated. The same features appeared in the 3-D hybrid simulation of the plasma environment of Mars and Bo¨ßwetter et al. [2004] found the following explanation for that picture. The magnetic tension and the pressure force, which are caused by the curvature and magnetic field gradients, are operating on the solar wind protons and heavy ions. These forces are strongest in the vicinity of the poles where the draped magnetic field lines can unwind. As a result, it leads to an acceleration of the solar wind plasma in the E hemisphere while the same acceleration force is compensated by the decelerated solar plasma flow stemming from the mass loading of heavy ions in the E+ hemisphere.

E00B30

[42] By comparing the solar wind and heavy ion densities one observes a clear separation of the different plasmas in the E hemisphere which is divided by the so-called ion composition boundary (ICB), as also already mentioned in section 2 in the course of the determinations of the plasma boundaries based on spacecraft measurements. The formation of the ICB has already been extensively studied by Simon et al. [2007] in the framework of global 3-D hybrid simulations of the plasma environments of Mars and Titan. They concluded that the underlying physical mechanism which is giving rise to the ICB is associated with the Lorentz forces that act on protons and heavy ions and can be explained in terms of kinetic models. The decisive role for the development of the ICB plays the combination of the convective electric field and the electron pressure forces. That means, in the E hemisphere where both forces are antiparallel it leads to the formation of a sharply pronounced boundary layer inhibiting the mixing of the plasmas. In contrast, in the E+ hemisphere no ICB emerges owing to the parallelism of both forces. As a result heavy ions are dragged away from the planet in the direction of the convective electric field and thus, causing a significant extension of the ionospheric tail. In other words, on the one hand, at the ICB the solar wind ions are reflected because the electron pressure gradient points away from the planet owing to the high electron density in the ionosphere and on the other hand, the heavy ion particles are accelerated toward the ICB in the direction of the convective electric field which points to the planet. [43] Inside the ionospheric tail the particles gyrate with a small gyroradius owing to the vanishing solar wind electric field while the picked up oxygen ions move away from Venus on large cyclodial paths as can be seen in the bulk velocity of the heavy ions. Owing to the acceleration of the planetary particles in the direction of the convective electric field, heavy ions are picked up by the solar wind, mass is added to the plasma flow (mass loading effect) and atmospheric material leaves from the planet. [44] The intensity of the magnetic field shows an asymmetric behavior which also has been confirmed by observations [Zhang et al., 1991] as well as by previous simulations [Shimazu, 2001; Bo¨ßwetter et al., 2004]. The piling up of the magnetic field in the E+ hemisphere is a consequence of the mass loading by planetary ions which leads to a deceleration of the shocked solar wind. The convective electric field, carried by the solar wind, vanishes in the plasma wake because it is inhomogeneously filled with ionospheric plasma as can be seen in Figures 10 and 11. In the E hemisphere the electric field is orientated perpendicular to the ICB and thus, pointing toward the planet while in the E+ hemisphere it is pointing away from Venus, as already mentioned above in the course of explaining the ICB formation. [45] Figure 11 illustrates the simulation results of the solar wind density nsw (cm 3) and velocity jvswj (km s 1), heavy ion density nhi (cm 3) and velocity jvhij (km s 1), the magnetic field jBj (nT)and the solar wind electric field jEj (V km 1) in the equatorial plane. The bow shock is asymmetrical also in the equatorial plane, as mentioned above when describing the simulation results of the terminator plane. Again, one finds the multiple shocklet structure as well as the quasi-parallel shock scenario, on the western

10 of 15

E00B30

MARTINECZ ET AL.: PLASMA ENVIRONMENT OF VENUS

E00B30

Figure 11. Simulation results in the equatorial plane, i.e., looking from ecliptic North onto Venus. The undisturbed solar wind flow is parallel to the positive x axis, the convective electric field points into the plane ( z axis), and the background magnetic field completes the right-handed system (+y axis) and is tilted 157.5
in the equatorial plane. Figure 12 shows (a) the solar wind density (nsw), (b) the heavy ion density (nhi), (c) the background magnetic field, (d) the solar wind bulk velocity (jvswj), (e) heavy ion bulk velocity (jvhij), and (f) the convective electric field. Behind the bow shock the solar wind density is increased and the solar wind bulk velocity is decelerated, characterizing the magnetosheath region. Atmospheric material is lost through the plasma wake, thereby forming tail rays and filaments on the nightside ionosphere. The interplanetary magnetic field piles up on the dayside of the planet, producing a so-called magnetic barrier region wherein the magnetic pressure exceeds the thermal pressure of the solar wind (plasma beta below unity). See text for details. (dusk) side (in the sense of orbital motion). In addition, the solar wind bulk velocity is significantly decreased in the downstream region which is a characteristic of the magnetosheath. The geometry of the magnetic field on the nightside is related to the draping of the interplanetary magnetic field over the obstacle on the dayside [Luhmann and Cravens, 1991]. As a result, an induced magnetotail is formed behind the planet, consisting of 2 lobes of opposite polarity separated by a plasma sheet. Additionally, the IMF piles up on the dayside of the planet, producing a so-called magnetic barrier region, wherein the magnetic pressure exceeds the thermal pressure of the solar wind. This is also consistent with spacecraft measurements [Zhang et al., 1991] as well as global hybrid simulations of weakly magnetized bodies like, e.g., Mars [Bo¨ßwetter et al., 2004; Modolo et al., 2006], Venus [Jarvinen et al., 2008], and comets [Bagdonat and Motschmann, 2002b].

5. Comparison of the Simulation Results With VEX Orbit [46] Figure 12 shows that the spacecraft trajectory on orbit 85 lies in the terminator plane. We take a cut through

the simulation box along the VEX trajectory in order to discuss the parameters as a function of spacecraft time. A direct comparison between observations and simulation results are shown in Figures 13, 14, and 15. We take a cut through the simulation box along the VEX trajectory in order to discuss the parameters as a function of spacecraft time. All data have been resampled with 1 min resolution. Additionally, the positions of the plasma boundaries are marked by dashed vertical lines, namely the bow shock and the upper and lower boundary of the transition zone. [47] Figure 13 compares the magnetic field intensity (nT) extracted from the simulation with the measured value obtained by the Magnetometer along the spacecraft orbit 85. Most of the features are at least in qualitative agreement; that is, plasma boundary positions are well reproduced by the simulation, especially on the outbound pass. However, there are some disagreements regarding the absolute value where the real physics is still more complex than our model. During the inbound pass the simulation does not exhibit the jump in the magnetic field, indicating the BS crossing which is visible in the data. By comparing the VEX trajectory in Figure 9 with the simulation results in the terminator plane one can see clearly that this originates

11 of 15

E00B30

MARTINECZ ET AL.: PLASMA ENVIRONMENT OF VENUS

E00B30

Figure 13. Direct comparison of the magnetic field parameter derived from the simulation (black line) with the magnetometer data (green line) along the VEX orbit 85 on 15 July 2006 from inbound to outbound. In addition, the positions of the plasma boundaries (red, bow shock; dark blue, upper mantle boundary; light blue, lower mantle boundary (ICB)) are marked by dashed vertical lines.

Figure 15. Direct comparison of the solar wind density from the simulation (black line) with ELS data (green line, fitted density) along the VEX orbit 85 on 15 July 2006 from inbound to outbound. In addition, the positions of the plasma boundaries (red, bow shock; dark blue, upper mantle boundary; light blue, lower mantle boundary (ICB)) are marked by dashed vertical lines.

from the quasi-parallel shock situation. In the simulation we assumed a constant IMF angle of 22.5
(in VSO) but in reality the magnetic field orientation is strongly fluctuating. Our constant magnetic field boundary is surely too raw. [48] Figure 14 displays the comparison of the solar wind velocity derived from the simulation and observed by the IMA sensor. The simulation is more or less in agreement with the data, except for the transition zone and ionosphere where IMA does not resolve the double peak feature which is visible in the simulation. The source of this disagreement

could be the location of accelerated shocked solar wind plasma after the terminator (see Figure 10, jvswj as well as the polar projection in Figure 12). That means, in the simulation the spacecraft would transit this region of accelerated plasma during the inbound pass, traverse the ionosphere (sharp drop in bulk velocity) and then, cross the fast plasma flow once again during the outbound pass. Though we never see proton bulk speeds accelerated above solar wind speed in the transition zone. Probably the solar wind electric field is much weaker in reality because of the fluctuating magnetic field. The positions of the plasma boundaries are well reproduced by the hybrid model, however, again there are significant discrepancies regarding the absolute values. [49] Figure 15 compares the solar wind density extracted from the simulation with the fitted density measured by the ELS sensor which shows for the most part a good agreement between data and model (from 0148 to 0320 UT) regarding the plasma boundaries as well as the absolute value, while on the outbound pass the agreement is almost perfect. During the inbound pass, in the solar wind and magnetosheath regions, there is a large difference between the simulated and measured densities due to the quasiparallel shock. Although the features are similar, the absolute values do not match the observations, maybe also due to the absolute calibration of the instruments. [50] Figure 16 shows the result of the convective electric field extracted from the simulation along the VEX trajectory which demonstrates that the hybrid model is able to reproduce the positions of the plasma boundaries.

Figure 14. Direct comparison of the solar wind bulk velocity from the simulation (black line) with IMA data (green line; integrated velocity) along the VEX orbit 85 on 15 July 2006 from inbound to outbound. Additionally, the positions of the plasma boundaries (red, bow shock; dark blue, upper mantle boundary; light blue, lower mantle boundary (ICB)) are marked by dashed vertical lines.

6. Conclusions [51] In this paper we investigated the global plasma environment of Venus using a 3-D hybrid model, treating the electrons as a massless, charge-neutralizing fluid, whereas

12 of 15

E00B30

MARTINECZ ET AL.: PLASMA ENVIRONMENT OF VENUS

Figure 16. Convective electric field resulting from simulation. The positions of the plasma boundaries (red, bow shock; dark blue, upper mantle boundary; light blue, lower mantle boundary (ICB)) are marked by dashed vertical lines. a completely kinetic approach is used to cover ion dynamics. The Venusian atmosphere is modeled under the assumption of an effective ionization rate mainly caused by solar EUV radiation. The production rate is a function of the altitude above the surface and of the solar zenith angle on the dayside, but depends only on the altitude on the nightside. Two different electron populations are incorporated in the hybrid code, in order to take into account the significantly different temperatures of the solar wind and ionospheric electrons. Using a curvilinear grid in the vicinity of the planet enables a high resolution of the plasma structures near the planetary surface. [52] Emphasis has been placed on the direct comparison of the plasma and magnetic field observations made onboard the Venus Express spacecraft with the results of the hybrid simulation. A first comparison between measurements and simulations indicates that the hybrid approach is capable of providing an adequate picture of the global plasma processes at Venus. [53] The common features of the solar wind-atmosphere interaction of an unmagnetized planet are fairly good reproducible by the hybrid code. The simulated bow shock, exhibiting shocklets structuring (kinetic nature), is formed in front of the planet and is equal in shape, size and position with the observed bow shock. Behind the obstacle a plasma wake is developed, filled with planetary ions and split into a central tail (plasma sheet) with high density and several rays with smaller density. Owing to the lack of an intrinsic magnetic field the IMF lines are draping around the planet, generating a magnetic barrier on the dayside as well as an induced magnetotail, consisting of two lobes of opposite polarity separated by a plasma sheet, on the nightside. The most pronounced characteristics are asymmetries with respect to the convection electric field affecting the heavy ion pickup region, magnetic field intensity and the shock geometry. A first determination of the tail boundaries (mantle region) indicates a very broad transition zone from solar wind to planetary ions, which also is confirmed by the simulation results.

E00B30

[54] Additionally, we determined the positions of the plasma boundaries at Venus at solar minimum based on the ASPERA-4 observations. We conclude from our measurements that the bow shock position is relatively stable whereas the upper and lower boundary of the mantle are highly variable on the nightside. In comparison to the large variations observed for the UMB, the variations of the BS position is smaller (about 15% of the terminator distance). The variations visible in the BS locations can be attributed to the variations in the magnetic field strength, since the BS position should depend on the upstream Mach number [Zhang et al., 2008a]. Previous studies based on PVO data [Russell et al., 1988] showed that the main factors influencing the BS position are the solar cycle, the upstream magnetosonic Mach number and the IMF clock angle (arctan(By/Bz)). But to detect the relatively small influences on magnetosonic Mach number and IMF clock angle, Russell et al. [1988] use only small subset of the large PVO data set (3300 orbits) with crossings within 0.5 RV close to the terminator. If we would apply the same selection to our data set, our statistics would be insufficient in order to analyze the dependency of the terminator BS position on the magnetosonic Mach number and IMF clock angle. For that reason we leave these issues for a later investigation. On the other hand, if we cannot decrease the number of data points we introduce a larger error in the dependency analysis caused by the projection of all BS crossings onto the terminator plane as well as by the assumption of a constant aberration angle. [55] Moreover, the main nonthermal loss processes of oxygen, namely the ion pickup, plasma clouds and momentum transport, are self-consistently included in the hybrid model. Thus, we can also compare the escape rates resulting from our simulation with measurements and estimations based on other models. Lammer et al. [2006] provide a theoretical escape rate for averaged solar activity of about 1.7 1026 s 1 (Monte Carlo particle simulations and gas dynamic test particle model). Recently, Barabash et al. [2007b] gave an initial estimation of the oxygen escape rate at Venus based on the ASPERA-4 measurements and concluded that the lower limit is about 1025 s 1. The simulation run (orbit 85) for solar minimum conditions yields an escape rate of about 0.9 1026 s 1 which establishes only a lower limit because the dayside ionospheric densities produced in the simulation ( 102 cm 3) are about a factor of 100 less than the observed values ( 104 cm 3) [Pa¨tzold et al., 2007]. In the simulated ionosphere the radial density gradient is extremely high and the spatial resolution of the simulation grid yields always mean values over 100 km in radial distance reducing the maximum values of the heavy ion density observed in the simulation. However in summary, our 3-D hybrid model yields an escape rate estimation within the limits of observations and theoretical calculations, and thus, seems to be able to provide an insight into the loss processes of oxygen at Venus induced by the solar wind. [56] To summarize, the positions of the simulated plasma boundaries are in good agreement with the observations but we found significant discrepancies between the observed compression of the plasma and the magnetic field on the western (dusk) side (in the sense of orbital motion) of the orbit. Possible reasons are on the one hand, the calculated input parameters because there are still uncertainties in the

13 of 15

E00B30

MARTINECZ ET AL.: PLASMA ENVIRONMENT OF VENUS

estimation of the plasma moments and on the other hand, the atmospheric model has also to be refined. Another reason for the disagreements between simulations and measurements can be ascribed to fluctuations in the upstream solar wind parameters which are assumed to be completely homogeneous in the simulation. [57] Kallio et al. [2006] investigated the global plasma environment of Venus for solar minimum conditions (based on PVO measurements) with a quasi-neutral hybrid (QNH) model which is similar to our hybrid approach. The QNH model is also able to reproduce the basic observed plasma and magnetic field regions and boundaries near Venus. However, their grid resolution is not fine enough to model the position and shape of the bow shock, multiple shocklets, the inner structure of the magnetic barrier, tail rays and filaments. Nevertheless, they also find a north – south asymmetry in the direction of the convective electric field due to mass loading of the solar wind by heavy ions as stated in section 4 explaining the simulation results. Additionally, they also observe a dawn-dusk asymmetry and conclude that this asymmetry is associated with the IMF x component because they use a Parker spiral IMF where the IMF x component is larger in magnitude than the magnetic field components perpendicular to the flow. They also yield an asymmetry in the magnetic field strength along the solar wind electric field because of the mass loading effect but it is less pronounced than in our simulation results. As in our case study, Kallio et al. [2006] also use homogeneous upstream solar wind parameters. But since Venus shows a strong dependence on the phase of the solar cycle it is necessary to study the solar wind-ionosphere interaction with varying plasma and magnetic field parameters in order to carry out more realistic investigations. In addition, they obtain a total escape rate of oxygen ions from the simulation box of about 2.2 1024 s 1. This value lies beyond the recently estimated lower limit based on ASPERA-4 observations [Barabash et al., 2007b] and is much lower compared to our simulated escape rate. The reason is the coarse grid resolution used in the QNH approach which inhibits the corporation of a self-consistent ionosphere. [58] A further technical limitation is associated with the spatial resolution. Although we use a curvilinear grid to enhance the spatial resolution near the planetary surface, it is not high enough to resolve the ionopause since the ionosphere is just a few bins in size. In the last years our code was parallelized by Josef Schu¨le from the Institute of Scientific Computation at the Technical University of Braunschweig (Germany). By means of this MPI (messagepassing interface) parallelization one can improve the spatial grid resolution in order to resolve small-scale structures of the solar wind interaction region, such as for example the ionopause. [59] Acknowledgments. We thank all members of the ASPERA-4 team for the big effort which led to the successful operation of the instrument and the calibration of the data. For this paper we are especially grateful to Emmanuel Penou and Andrei Fedorov at CESR for data preparation and IMA calibration and Andrew Coates at MSSL for calibration of the ELS sensor data. We thank the CELIAS/SEM experiment on the Solar Heliospheric Observatory (SOHO) spacecraft for providing solar EUV data. The ASPERA ELS was constructed under NASA contract NASW-0003. This work was supported by DFG grant MO539/17-1.

E00B30

References Bagdonat, T., and U. Motschmann (2002a), 3D hybrid simulation code using curvilinear coordinates, J. Comput. Phys., 183, 470 – 485. Bagdonat, T., and U. Motschmann (2002b), From a weak to a strong comet: 3D global hybrid simulation studies, Earth Moon Planets, 90, 305 – 321. Barabash, S., et al. (2007a), The Analyser of Space Plasmas and Energetic Atoms (ASPERA-4) for the Venus Express mission, Planet. Space Sci., 55, 1772 – 1792, doi:10.1016/j.pss.2007.01.014. Barabash, S., et al. (2007b), The loss of ions from Venus through the plasma wake, Nature, 450, 650 – 653, doi:10.1038/nature06,434. Bauske, R., A. F. Nagy, T. I. Gombosi, D. L. De Zeeuw, K. G. Powell, and J. G. Luhmann (1998), A three-dimensional MHD study of solar wind mass loading processes at Venus: Effects of photoionization, electron impact ionization, and charge exchange, J. Geophys. Res., 103, 23,625 – 23,638, doi:10.1029/98JA01791. Bertucci, C., C. Mazelle, J. A. Slavin, C. T. Russell, and M. H. Acun˜a (2003), Magnetic field draping enhancement at Venus: Evidence for a magnetic pileup boundary, Geophys. Res. Lett., 30(17), 1876, doi:10.1029/2003GL017271. Bo¨ßwetter, A., T. Bagdonat, U. Motschmann, and K. Sauer (2004), Plasma boundaries at Mars: A 3-D simulation study, Ann. Geophys., 22, 4363 – 4379. Bo¨ßwetter, A., et al. (2007), Comparison of plasma data from ASPERA-3/ Mars-Express with a 3-D hybrid simulation, Ann. Geophys., 25, 1851 – 1864. Brecht, S. H., and J. R. Ferrante (1991), Global hybrid simulation of unmagnetized planets: Comparison of Venus and Mars, J. Geophys. Res., 96, 11,209 – 11,220. Dubinin, E. (2008), Structure and dynamics of the solar wind/ionosphere interface on Mars: MEX-ASPERA-3 and MEX-MARSIS observations, Geophys. Res. Lett., 35, L11103, doi:10.1029/2008GL033730. Fra¨nz, M., et al. (2007), Plasma moments in the environment of Mars: Mars Express ASPERA-3 observations, Space Sci. Rev., 126, 165 – 207, doi:10.1007/s11214-006-9115-9. Israelevich, P. L., T. I. Gombosi, A. I. Ershkovich, D. L. DeZeeuw, F. M. Neubauer, and K. G. Powell (1999), The induced magnetosphere of comet Halley: 4. Comparison of in situ observations and numerical simulations, J. Geophys. Res., 104, 28,309 – 28,319. Jarvinen, R., E. Kallio, I. Sillanpa¨a¨, and P. Janhunen (2008), Hybrid modelling the Pioneer Venus Orbiter magnetic field observations, Adv. Space Res., 41, 1361 – 1374, doi:10.1016/j.asr.2007.10.003. Kallio, E., J. G. Luhmann, and J. G. Lyon (1998), Magnetic field near Venus: A comparison between Pioneer Venus Orbiter magnetic field observations and an MHD simulation, J. Geophys. Res., 103, 4723 – 4737. Kallio, E., R. Jarvinen, and P. Janhunen (2006), Venus solar wind interaction: Asymmetries and the escape of O+ ions, Planet. Space Sci., 54, 1472 – 1481, doi:10.1016/j.pss.2006.04.030. Kulikov, Y. N., et al. (2006), Atmospheric and water loss from early Venus, Planet. Space Sci., 54, 1425 – 1444, doi:10.1016/j.pss.2006.04.021. Lammer, H., et al. (2006), Loss of hydrogen and oxygen from the upper atmosphere of Venus, Planet. Space Sci., 54, 1445 – 1456, doi:10.1016/ j.pss.2006.04.022. Luhmann, J. G., and T. E. Cravens (1991), Magnetic fields in the ionosphere of Venus, Space Sci. Rev., 55, 201 – 274. Martinecz, C., et al. (2008), Location of the bow shock and ion composition boundaries at Venus: Initial determinations from Venus Express ASPERA-4, Planet. Space Sci., doi:10.1016/j.pss.2007.01.014. Miller, K. L., C. W. Knudsen, K. Spenner, R. C. Whitten, and V. Novak (1980), Solar zenith angle dependence of ionospheric ion and electron temperatures and density on Venus, J. Geophys. Res., 85, 7759 – 7764. Modolo, R., G. M. Chanteur, E. Dubinin, and A. P. Matthews (2006), Simulated solar wind plasma interaction with the Martian exosphere: Influence of the solar EUV flux on the bow shock and the magnetic pile-up boundary, Ann. Geophys., 24, 3403 – 3410. Moore, K. R., V. A. Thomas, and D. J. McComas (1991), Global hybrid simulation of the solar wind interaction with the dayside of Venus, J. Geophys. Res., 96, 7779 – 7791. Murawski, K., and R. S. Steinolfson (1996a), Numerical modeling of the solar wind interaction with Venus, Planet. Space Sci., 44, 243 – 252. Murawski, K., and R. S. Steinolfson (1996b), Numerical simulations of mass loading in the solar wind interaction with Venus, J. Geophys. Res., 101, 2547 – 2560, doi:10.1029/95JA02433. Nagy, A. F., J. Kim, and T. E. Cravens (1990), Hot hydrogen and oxygen atoms in the upper atmospheres of Venus and Mars, Ann. Geophys., 8, 251 – 256. Omidi, N., and D. Winske (1990), Steepening of kinetic magnetosonic waves into shocklets: Simulations and consequences for planetary shocks and comets, J. Geophys. Res., 95, 2281 – 2300. Pa¨tzold, M., et al. (2007), The structure of Venus’ middle atmosphere and ionosphere, Nature, 450, 657 – 660, doi:10.1038/nature06239.

14 of 15

E00B30

MARTINECZ ET AL.: PLASMA ENVIRONMENT OF VENUS

Phillips, J. L., and D. J. McComas (1991), The magnetosheath and magnetotail of Venus, Space Sci. Rev., 55, 1 – 80. Roussos, E., et al. (2008), Plasma and fields in the wake of Rhea: 3D hybrid simulation and comparison with Cassini data, Ann. Geophys., in press. Russell, C. T., E. Chou, J. G. Luhmann, P. Gazis, L. H. Brace, and W. R. Hoegy (1988), Solar and interplanetary control of the location of the Venus bow shock, J. Geophys. Res., 93, 5461 – 5469. Russell, C. T., J. G. Luhmann, and R. J. Strangeway (2006), The solar wind interaction with Venus through the eyes of the Pioneer Venus Orbiter, Planet. Space Sci., 54, 1482 – 1495, doi:10.1016/j.pss.2006.04.025. Shimazu, H. (2001), Three-dimensional hybrid simulation of solar wind interaction with unmagnetized planets, J. Geophys. Res., 106, 8333 – 8342. Simon, S., T. Bagdonat, U. Motschmann, and K.-H. Glassmeier (2006a), Plasma environment of magnetized asteroids: A 3D hybrid simulation study, Ann. Geophys., 24, 407 – 414. Simon, S., A. Bo¨ßwetter, T. Bagdonat, U. Motschmann, and K.-H. Glassmeier (2006b), Plasma environment of Titan: A 3-D hybrid simulation study, Ann. Geophys., 24, 1113– 1135. Simon, S., A. Boesswetter, T. Bagdonat, and U. Motschmann (2007), Physics of the ion composition boundary: A comparative 3-D hybrid simulation study of Mars and Titan, Ann. Geophys., 25, 99 – 115. Slavin, J. A., R. E. Holzer, J. R. Spreiter, and S. S. Stahara (1984), Planetary mach cones: Theory and observation, J. Geophys. Res., 89, 2708 – 2714. Spreiter, J. R., and S. S. Stahara (1992), Computer modeling of solar wind interaction with Venus and Mars, in Venus and Mars: Atmospheres, Ionospheres, and Solar Wind Interactions, Geophys. Monogr. Ser., vol. 66, edited by J. G. Luhman, M. Tatrallyay, and R. O. Pepin, pp. 345 – 383, AGU, Washington D. C. Tanaka, T. (1993), Configurations of the solar wind flow and magnetic field around the planets with no magnetic field: Calculation by a new MHD simulation scheme, J. Geophys. Res., 98, 17,251 – 17,262, doi:10.1029/ 93JA01516. Tanaka, T., and K. Murawski (1997), Three-dimensional MHD simulation of the solar wind interaction with the ionosphere of Venus: Results of two-component reacting plasma simulation, J. Geophys. Res., 102, 19,805 – 19,822, doi:10.1029/97JA01474. Terada, N., S. Machida, and H. Shinagawa (2002), Global hybrid simulation of the Kelvin-Helmholtz instability at the Venus ionopause, J. Geophys. Res., 107(A12), 1471, doi:10.1029/2001JA009224.

E00B30

Terada, N., H. Shinagawa, and S. Machida (2004), Global hybrid model of the solar wind interaction with the Venus ionosphere: Ion escape processes, Adv. Space Res., 33, 161 – 166, doi:10.1016/j.asr.2003.05.011. Titov, D. V., et al. (2006), Venus Express science planning, Planet. Space Sci., 54, 1279 – 1297, doi:10.1016/j.pss.2006.04.017. Torr, M. R., and D. G. Torr (1985), Ionization frequencies for solar cycle 21: Revised, J. Geophys. Res., 90, 6675 – 6678. Wu, C. C. (1992), MHD flow past an obstacle: Large-scale flow in the magnetosheath, Geophys. Res. Lett., 19, 87 – 90. Zhang, T. L., J. G. Luhmann, and C. T. Russell (1991), The magnetic barrier at Venus, J. Geophys. Res., 96, 11,145 – 11,153. Zhang, T. L., C. T. Russell, J. G. Luhmann, J. R. Spreiter, and S. S. Stahara (1993), On the spatial range of validity of the gas dynamic model in the magnetosheath of Venus, Geophys. Res. Lett., 20, 751 – 754. Zhang, T. L., et al. (2006), Magnetic field investigation of the Venus plasma environment: Expected new results from Venus Express, Planet. Space Sci., 54, 1336 – 1343. Zhang, T. L., et al. (2008a), Initial Venus Express magnetic field observations of the Venus bow shock location at solar minimum, Planet. Space Sci., 56, 785 – 789, doi:10.1016/j.pss.2007.09.012. Zhang, T. L., et al. (2008b), Initial Venus Express magnetic field observations of the magnetic barrier at solar minimum, Planet. Space Sci., 56, 790 – 795, doi:10.1016/j.pss.2007.10.013. S. Barabash and R. Lundin, Swedish Institute of Space Physics, P.O. Box 812, SE-981 28, Kiruna, Sweden. A. Boesswetter, U. Motschmann, S. Simon, and S. Wiehle, Institut fu¨r Theoretische Physik, Technische Universita¨t Braunschweig, Mendelssohnstrasse 3, D-38106 Braunschweig, Germany. E. Dubinin, M. Fra¨nz, N. Krupp, C. Martinecz, E. Roussos, and J. Woch, Max Planck Institut fu¨r Sonnensystemforschung, Max-Planck-Strasse 2, D-37191 Katlenburg-Lindau, Germany. ([email protected]) Y. Kulikov, Polar Geophysical Institute, Russian Academy of Sciences, Khalturina Street 15, 183010 Murmansk, Russia. H. Lammer, H. Lichtenegger, and T. L. Zhang, Space Research Institute, Austrian Academy of Sciences, Schmiedlstrasse 6, A-8042 Graz, Austria.

15 of 15

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114, E00B98, doi:10.1029/2009JE003377, 2009

Click Here for

Full Article

Correction to ‘‘Plasma environment of Venus: Comparison of Venus Express ASPERA-4 measurements with 3-D hybrid simulations’’ C. Martinecz, A. Boesswetter, M. Fra¨nz, E. Roussos, J. Woch, N. Krupp, E. Dubinin, U. Motschmann, S. Wiehle, S. Simon, S. Barabash, R. Lundin, T. L. Zhang, H. Lammer, H. Lichtenegger, and Y. Kulikov Received 12 March 2009; published 29 April 2009.

Citation: Martinecz, C., et al. (2009), Correction to ‘‘Plasma environment of Venus: Comparison of Venus Express ASPERA-4 measurements with 3-D hybrid simulations,’’ J. Geophys. Res., 114, E00B98, doi:10.1029/2009JE003377.

[1] In the paper ‘‘Plasma environment of Venus: Comparison of Venus Express ASPERA-4 measurements with 3-D hybrid simulations’’ by C. Martinecz et al. (Journal of Geo-

physical Research, 114, E00B30, doi:10.1029/2008JE003174, 2009), Figures 9–12 were numbered incorrectly. The correct numbered figures and their captions are shown here.

Figure 9. Simulation results in the terminator plane, i.e., looking from behind on Venus. The solar wind is orientated out of the plane (+x axis), the convective electric field is antiparallel to the (+z) axis and points away from the planet; thus, the background magnetic field completes the right-handed system (+y axis). Figure 10 displays (left) the solar wind density (nsw), (middle) the heavy ion density (nhi), and (right) the background magnetic field. An asymmetric bow shock, exhibiting a shocklet structure, is formed in front of the obstacle owing to the supersonic solar wind (left plot). The interplanetary magnetic field is draping around Venus generating an induced magnetotail on the nightside of the planet (right plot). See text for details.

Copyright 2009 by the American Geophysical Union. 0148-0227/09/2009JE003377$09.00

E00B98

1 of 4

E00B98

MARTINECZ ET AL.: CORRECTION

Figure 10. Cut through the simulation box along the polar plane, i.e., looking from the western (dusk) side (in the sense of orbital motion) on Venus. The plasma flow comes in from the left (+x axis), the solar wind electric field points away from the planet (
z axis), and the interplanetary magnetic field completes the right-handed system pointing into the plane (+y axis). Figure 11 illustrates from left to right in the first row (a) the solar wind density (nsw), (b) the heavy ion density (nhi), (c) the background magnetic field, (d) the solar wind bulk velocity (jn swj), (e) heavy ion bulk velocity (jn hij), and (f) the convective electric field. In the density plots a clear separation between the solar wind and ionospheric plasma takes place forming the so-called ion composition boundary (ICB) in the northern hemisphere. In the southern hemisphere there is no clear ICB owing to the mass loading effect; that is, the planetary ions are accelerated in the direction of the convective electric field and are picked up by the solar wind flow. The solar wind electric field ~ E =
~ v~ B vanishes where the heavy ion plasma dominates. See text for details.

2 of 4

E00B98

E00B98

MARTINECZ ET AL.: CORRECTION

Figure 11. Simulation results in the equatorial plane, i.e., looking from ecliptic North onto Venus. The undisturbed solar wind flow is parallel to the positive x axis, the convective electric field points into the plane (
z axis), and the background magnetic field completes the right-handed system (+y axis) and is tilted 157.5° in the equatorial plane. Figure 12 shows (a) the solar wind density (nsw), (b) the heavy ion density (nhi), (c) the background magnetic field, (d) the solar wind bulk velocity (jn swj), (e) heavy ion bulk velocity (jn hij), and (f) the convective electric field. Behind the bow shock the solar wind density is increased and the solar wind bulk velocity is decelerated, characterizing the magnetosheath region. Atmospheric material is lost through the plasma wake, thereby forming tail rays and filaments on the nightside ionosphere. The interplanetary magnetic field piles up on the dayside of the planet, producing a so-called magnetic barrier region wherein the magnetic pressure exceeds the thermal pressure of the solar wind (plasma beta below unity). See text for details.

3 of 4

E00B98

E00B98

MARTINECZ ET AL.: CORRECTION

Figure 12. Projections of the trajectory of the Venus Express orbit 85 on the equatorial, polar, and terminator plane of our 3-D hybrid simulation. The undisturbed plasma flow is directed along the (+x) axis, the convection electric field ~ E =
~ v~ B is orientated antiparallel to the z axis, and the background magnetic field completes the right-handed system pointing in the positive y direction (tilted 157.5° in the equatorial plane).

4 of 4

E00B98

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, E00B21, doi:10.1029/2008JE003159, 2008

for

Full Article

Intermittent turbulence, noisy fluctuations, and wavy structures in the Venusian magnetosheath and wake Z. Vo¨ro¨s,1 T. L. Zhang,2 M. P. Leubner,1 M. Volwerk,2 M. Delva,2 and W. Baumjohann2 Received 2 April 2008; revised 5 August 2008; accepted 18 September 2008; published 25 December 2008.

[1] Recent research has shown that distinct physical regions in the Venusian induced

magnetosphere are recognizable from the variations of strength of the magnetic field and its wave/fluctuation activity. In this paper the statistical properties of magnetic fluctuations are investigated in the Venusian magnetosheath and wake regions. The main goal is to identify the characteristic scaling features of fluctuations along Venus Express (VEX) trajectory and to understand the specific circumstances of the occurrence of different types of scalings. For the latter task we also use the results of measurements from the previous missions to Venus. Our main result is that the changing character of physical interactions between the solar wind and the planetary obstacle is leading to different types of spectral scaling in the near-Venusian space. Noisy fluctuations are observed in the magnetosheath, wavy structures near the terminator, and in the nightside near-planet wake. Multiscale turbulence is observed at the magnetosheath boundary layer and near the quasi-parallel bow shock. Magnetosheath boundary layer turbulence is associated with an average magnetic field which is nearly aligned with the Sun-Venus line. Noisy magnetic fluctuations are well described with the Gaussian statistics. Both magnetosheath boundary layer and near-shock turbulence statistics exhibit non-Gaussian features and intermittency over small spatiotemporal scales. The occurrence of turbulence near magnetosheath boundaries can be responsible for the local heating of plasma observed by previous missions. Citation: Vo¨ro¨s, Z., T. L. Zhang, M. P. Leubner, M. Volwerk, M. Delva, and W. Baumjohann (2008), Intermittent turbulence, noisy fluctuations, and wavy structures in the Venusian magnetosheath and wake, J. Geophys. Res., 113, E00B21, doi:10.1029/ 2008JE003159.

1. Introduction [2] The structured plasma environment of Venus is a natural plasma laboratory where the planetary ionosphere acts as an obstacle to the supersonic solar wind flow carrying a magnetic field. The absence of an intrinsic magnetic field ensures that this interaction is more cometlike than Earth-like. It is known from the previous missions to Venus [e.g., Russell and Vaisberg, 1983] that the most prominent features include the direct interaction of ionized magnetosheath flow with the ionosphere/quasi-neutral atmosphere, mass loading of the magnetosheath flux tubes, and the transport/convection of magnetic flux to the wake region, representing also the dominant source for the magnetotail fluxes. However, for most of the time, mainly during time intervals of low solar wind dynamic pressure, 1 Institute of Astro- and Particle Physics, University of Innsbruck, Innsbruck, Austria. 2 Space Research Institute, Austrian Academy of Sciences, Graz, Austria.

Copyright 2008 by the American Geophysical Union. 0148-0227/08/2008JE003159$09.00

the induced and piled up magnetic field around the planetary obstacle represents an effective magnetic barrier, preventing free entrance of solar wind plasma to the Venusian ionosphere [Zhang et al., 2007]. The draping of the interplanetary magnetic field (IMF), the accretion of magnetic flux by the planet, the characteristic spatial scales of physical processes as well as the location and specific features of boundaries, all depend on dynamical processes in the solar wind and make this environment unique. The results of Venus Express spacecraft (VEX) provide an altitude for the induced magnetopause of about 300 km at the subsolar point, while the subsolar bow shock distance from the surface of the planet is about 1900 km [Zhang et al., 2007]. These values exhibit a solar cycle dependence; that is, the thickness of the Venusian magnetosheath varies around 1500 km at the subsolar point and widens up to 5000– 7000 km at the terminator. During time intervals of high solar wind dynamic pressure, the ionopause moves to low altitude ( 250 km) and mainly the ionosphere forms the obstacle responsible for deflecting the solar wind flow. In this case the thickness of the ionopause increases and more direct interaction between the ionosphere and solar wind is possible [Elphic et al., 1981; Russell and Vaisberg, 1983].

*Author names correctly given here. The article as originally published appears online.

E00B21

1 of 11

E00B21

¨ RO ¨ S ET AL.: TURBULENCE AND FLUCTUATIONS NEAR VENUS VO

[3] In the dayside magnetosheath strong magnetic fluctuations and waves are present [e.g., Luhmann et al., 1983]. Mirror mode waves were observed here in case studies [Volwerk et al., 2008a], and also investigated statistically [Volwerk et al., 2008b]. On the other hand, the limited spatial scale of the magnetosheath on the dayside might not support a development of a turbulent cascade, the fluctuations rather resemble 1/f noise (f is the frequency). 1/fa noise is a signature of the presence of independent physical mechanisms driving fluctuations in the magnetosheath [Vo¨ro¨s et al., 2008]. Keeping in mind that the ion inertial length is of the order of 100 km, the magnetohydrodynamic (MHD) spatial scales in the Venusian magnetosheath are limited to one or two decades in wave number space. [4] The terminator and, further downstream, the nightside region is of particular interest [Spreiter and Stahara, 1992]. Here plasma instabilities, vortices, and turbulence can develop near boundaries. For example, Wolff et al. [1980] have shown that the distortion of the ionopause by KelvinHelmholtz instability might lead to the formation of magnetic flux ropes inside the ionosphere as well as ionospheric bubbles embedded in the solar wind. Numerical simulations indicate that the Kelvin-Helmholtz instability can occur at the terminator ionopause of Venus [Terada et al., 2002], capable of producing wave structures over 1000 km in size [Amerstorfer et al., 2007; Biernat et al., 2007]. In fact, the initial VEX observation detected these wavy structures (M. A. Balikhin et al., Giant vortices lead to ion escape from Venus, submitted to Geophysical Research Letters, 2008.). A recent study shows that MHD turbulence near and immediately after the terminator is not fully developed because of the rapid decrease of spectral power toward higher frequencies, resulting in spectral scaling indices a > 2 [Vo¨ro¨s et al., 2008]. Further downstream, in the magnetotail/ magnetosheath region, the spectral analysis indicates the presence of developed inertial range turbulence, with a spectral scaling index a 1.6, close to the values expected for hydrodynamic or magnetohydrodynamic turbulent flows. However, in the same region, non-Gaussian probability density functions (PDF) with typical long tails, corresponding to intermittent turbulence, were not observed [Vo¨ro¨s et al., 2008]. This can be explained by the shortness of turbulent time series during a tail crossing, in particular not allowing a full reconstruction of PDF tails. [5] The near-Venus tail is highly structured. There is no significant plasma inflow into the near-Venusian wake. The wake magnetic field is known to be stronger than the IMF and from the cavity hot plasma is excluded [Bauer et al., 1977]. The solar wind-ionosphere interaction in the presence of the draped interplanetary magnetic field, however, produces an extended boundary tailward from the terminator, at the inner edge of the magnetopause, or outer edge of the wake downstream, where boundary layer turbulence can develop and heat locally the plasma. The width of this boundary layer is limited, does not include the whole plasma sheet. In order to understand the energy content and energy dissipation of underlying processes, it is necessary to investigate systematically the statistical features of fluctuations in the structured near-Venusian plasma environment. In this paper we statistically analyze the spectral scaling features of fluctuations in the magnetosheath, wake and near-boundary regions using VEX magnetometer data

E00B21

during the first 20 days in May 2006. The time resolution of the magnetic data is 1 s. Because of rapid crossing of different structures, this time resolution is necessary for obtaining statistically reliable scaling results. The main emphasis is on the analysis of scalings, i.e., on the evaluation of the continuous part of magnetic power spectra. The analysis of waves (when the continuous parts of the spectra are not considered) is equally important, but out of the scope of this paper. In addition to spectral estimations, we will compare the probability distribution functions (PDFs) and evaluate the scale dependency of their shapes, associated with noisy and turbulent fluctuations. This helps to obtain a more reliable differentiation between turbulence and noise, occurring along the VEX trajectory. The nearpolar orbit of VEX with a periapsis altitude of 250– 350 km allows for the first time observations at terminator and midmagnetotail regions [Zhang et al., 2006]. These two important regions were not covered by the previous missions, e.g., Pioneer Venus Orbiter (PVO) [Russell, 1992].

2. Near-Venus Plasma Regions With Varying Spectral Scaling Properties [6] It was reported by Vo¨ro¨s et al. [2008] that the value of spectral scaling index a, describing the self-similarity of the power spectrum of the magnetic field data in the frequency range 0.03 – 0.5 Hz, exhibits different values in different regions near Venus. The magnetic fluctuations are nonstationary, e.g., in the magnetosheath the magnetic field strength is increasing toward the induced magnetopause, which introduces a trend into magnetic field data. In order to estimate a robustly, we used a wavelet method proposed by Abry et al. [2000] and applied it successfully to the description of magnetic fluctuations in the Earth’s plasma sheet [Vo¨ro¨s et al., 2004]. In this paper we use the Daubechies wavelets, for which finite data size effects are minimized and the number of vanishing moments can be changed. The latter feature of Daubechies analyzing wavelets is essential to cancel the influence of polynomial trends or periodic structures in the data on the estimation of the scaling index. [7] Let us demonstrate first how the spectral scaling features vary along the VEX trajectory during its journey from the dayside magnetosheath through the terminator region and wake to the post-terminator magnetosheath. As an example, we show the variation of total magnetic field strength B on 19 May 2006 (Figure 1, top) together with the corresponding power spectral densities (PSD) calculated during equally long time intervals (a, b, c, d and e, Figure 1, bottom subplots) along the spacecraft trajectory. Zhang et al. [2007] have already demonstrated that a crossing of each physical region in the near-Venus plasma environment is recognizable from the variations of strength and wave/fluctuation activity of the magnetic field. Indeed, B is not disturbed before t1 = 0115 UT, and after t2 = 0310 UT, the spacecraft is in the solar wind (Figure 1). The planetary obstacle perturbs the magnetic field only between t1 and t2. During the interval a, VEX enters the dayside magnetosheath from the solar wind (VEX trajectories and the intervals a – e are shown in Figure 2, left bottom subplot). Here the magnetic field strength is strongly fluctuating and its value increases up to 50 nT. The

2 of 11

E00B21

¨ RO ¨ S ET AL.: TURBULENCE AND FLUCTUATIONS NEAR VENUS VO

E00B21

Figure 1. (top) Magnetic field strength (B) on 19 May 2006. The horizontal black lines correspond to the time intervals a– e of equal length in the dayside magnetosheath (interval a), nightside near-planet wake (interval b), magnetosheath boundary layer (interval c), tailward magnetosheath (interval d), and in the vicinity of the bow shock. (bottom) Power spectra and spectral scalings estimated within the intervals a – e. Pronounced wavy structures are present mainly in the wake (interval b). The vertical arrow points to the spectral peak at 15 s. estimated spectral scaling index a = 1 ± 0.2 (the first bottom subplot in Figure 1) is low, indicating that fully developed turbulence is absent in this region [Vo¨ro¨s et al., 2008]. Higher values a = 5/3 or 3/2 are expected for hydrodynamic or MHD inertial range turbulence, respectively. After 0135 UT B decreases, which corresponds to the closest approach to Venus, VEX is close to or below the induced magnetopause [Zhang et al., 2007]. [8] Between 0140 and 0205 UT high-frequency fluctuations are absent, only low-frequency wavy fluctuations are seen. These wavy structures might be associated with Kelvin-Helmholtz instability at the terminator ionopause or detached plasma clouds near the terminator, observed already during Pioneer Venus orbits [Brace et al., 1982]. During interval b the corresponding spectrum exhibits significant wave power only near 0.07 Hz ( 14 s) then the spectral power rapidly decreases with a scaling index a = 2.5 ± 0.2 (the second bottom subplot in Figure 1). Further downstream (after 0205 UT in Figure 1) broadband fluctuations occur again. The spectral indices within the intervals c and e are a = 1.5 ± 0.2 and 1.6 ± 0.2 respectively, indicating the presence of developed turbulence. In contrary, the interval d, in between c and e, shows again 1/f noise-like scaling behavior (the last three bottom subplots in Figure 1). [9] The physical difference between turbulence and noise is clear. Turbulence in the wake is a consequence of nonlinear multiscale interactions and it is strongly dissipative, heating the background plasma at the small scales. Noisy fluctuations, exhibiting 1/fa scaling behavior with a

around 1, may have multiple physical sources not connected with nonlinear interactions, typical for turbulence. 0 < a < 1 over higher frequencies (around 1Hz) can also be associated with the noise of the magnetometer [Vo¨ro¨s et al., 2004]. The low values of a can also indicate that the spacecraft is not in a physical region where strong nonlinear interactions and turbulence can exist, e.g., because of low plasma density or lack of plasma flows. In the following, the notation ‘‘1/f noise’’ refers to 1/fa noise with spectral scaling index in a range of a2 (0.6, 1.4) (find below the definition of the spectral index used in this paper). [10] In this paper we put the emphasis on statistical examination of the circumstances under which developed turbulence occurs in the post-terminator wake and magnetosheath regions.

3. Statistical Analysis of Magnetic Fluctuations [11] We investigated the time series of magnetic field strength statistically, obtained in the near-Venusian space during the first 20 days in May 2006. During 1 day one orbit is performed, therefore, during the 20 day interval we have observed twenty crossings. Because of the nonhomogeneity and dynamical nature of the near-Venus physical regions the occurrence and the length of at least quasistationary intervals slightly differs each day. We identified the approximate beginning and end of the steady intervals computing the scaling indices over timescales 2 – 30 s, using the wavelet method within sliding overlapping windows. After that, changing slightly the starting point and the length

3 of 11

E00B21

¨ RO ¨ S ET AL.: TURBULENCE AND FLUCTUATIONS NEAR VENUS VO

E00B21

Figure 2. (top left) Magnetic field strength (B, gray line) and BX magnetic component (black line) on 19 May 2006; the horizontal lines show the intervals a – e depicted in Figure 1. (bottom left) Venus Express (VEX) crossings (thin black lines) of the Venusian plasma environment in Venus-Sun-Orbital (VSO) coordinates. The intervals a – e (black lines) are shown alongside the VEX trajectories. The filled circles show the bow shock, and the open circles show the magnetopause during multiple crossings (after Zhang et al. [2007]). The large triangles along the lowest trajectory correspond to the region of wavy structures and spectral scaling index a 2.5. The family of thick black lines along the trajectories (intervals c and e) correspond to turbulent regions with a 1.6. The gray ‘‘+’’ signs are intervals of 1/f noise (interval d). (right) Enlarged magnetosheath crossings in VSO coordinates. The same notations are used as in Figure 2, left. The bottom right subplot shows the approximate spatial size of the turbulent (intervals c and e) and noisy (interval d) regions in space. of turbulent/noisy data, optimized data intervals with smallest errors in a were found. We rejected time intervals where the error of the estimation of a was larger than ±0.3. The available steady intervals in the dayside magnetosheath are shorter, so we selected 9 min long intervals there. Further downstream 12 min long intervals were selected. Selecting longer data periods would significantly reduce the number of available intervals, while shorter data sets would decrease the statistical reliability of results. Data intervals with fluctuating magnetic field shorter than 12 min were not included into our analysis. Because of occasional data gaps or shortness of steady fluctuations in the time series, the number of events is not the same in different physical regions. There were crossings with no steady fluctuations along the VEX trajectory. There were also crossings where the time interval with steady fluctuations was longer than 24 min. [12] Because of data gaps or shortness of statistically stationary time series, only 17 out of 20 crossings are shown in Figure 2 (left bottom and right subplots). A cylindrical coordinatepsystem is used in Figure 2, where the events are shown in (Y2 + Z2) VSO versus X VSO (bottom left) or in VSO coordinate pairs (right subplots). VSO is the Venuscentered Venus-Sun-Orbital coordinate system, where X is in the direction of the Sun, Y opposite to the orbital

direction of Venus and Z perpendicular to the orbital plane, positive to ecliptic north. [13] The near-Venusian space is physically nonhomogeneous and the spatial and temporal variations cannot be straightforwardly separated from single-spacecraft measurements. Nevertheless, during the investigated time period in May 2006, the typical sequence of physical regions visited by VEX remained approximately the same as in Figure 1: crossing of the dayside magnetosheath (increasing B, interval a), low-frequency wavy structures after the terminator (interval b) and entering into the region of broadband fluctuations further downstream (intervals c – e). The intervals a –e were introduced for the event on 19 May 2006 (Figure 1, top, and Figure 2, top left). The wavy structures are not always present along the whole near-terminator region and wake. In the absence of the wavy structures only low-amplitude magnetometer noise is observed. Possibly, the occurrence/absence of wavy structures can be associated with the upstream conditions. For example, during times when the solar wind dynamic pressure is high the ionopause moves to low altitudes ( 250 km), and direct interactions between the ionosphere and the solar wind can occur (plasma-plasma interactions or solar wind electric field – ionospheric currents interactions). When the solar

4 of 11

E00B21

¨ RO ¨ S ET AL.: TURBULENCE AND FLUCTUATIONS NEAR VENUS VO

E00B21

Figure 3. Histograms of the scaling indices in different regions of near-Venusian space (approximately obtained during the intervals a – e in Figure 2.). (top to bottom) Dayside magnetosheath (interval a); nearterminator and post-terminator wake (interval b); magnetosheath boundary and near bow shock region (intervals c – e); post-terminator tailward magnetosheath (interval d). These spatial regions are depicted more clearly in Figure 4. wind dynamic pressure is low the ionospheric width increases [Elphic et al., 1981] and the draped IMF above 300 km can stop the solar wind more efficiently. In this paper we investigate only the features of magnetic fluctuations, without considering the changes in the upstream conditions. The open triangles (Figure 2, bottom left) indicate the whole region (marks are only on the lowest trajectory) where the wavy structures appear during the first 20 days in May 2006. [14] Because of the similarity of crossings, the depicted intervals (lines) along the trajectories (Figure 2, bottom left), in terms of spectral properties, refer generally to similar physical regions along the VEX trajectory. For example, the thick black lines (Figure 2, bottom left) indicate turbulent processes identified through a spectral scaling index near a = 1.6. The location of turbulence in space approximately coincides with time intervals c and e. The time interval d coincides roughly with the position of gray ‘‘+’’ signs along the trajectory, where scaling indices corresponding to 1/fa noise were observed. For a better visibility, the trajectories together with turbulent and noisy time intervals (c, e and d) are depicted in the VSO coordinates (Figure 2, right subplots). From Figure 2 (bottom right) it is visible that the spatial regions exhibiting the same a along multiple trajectories are partially overlapping, indicating that these regions and the corresponding boundaries are moving or the fluctuations are patchy or intermittent. Moving boundaries can appear in connection with changing conditions in the solar wind. [15] Switching between temporal and spatial coordinates helps us to identify spatial regions with typical scalings. Occasionally we will use plural indicating the change between temporal and spatial coordinates. For example,

the notation ‘‘intervals c’’ means the set of all crossings in space near the time interval/space region c on 19 May 2006. [16] Figure 3 shows histograms with the number of events per unit interval of the scaling index a, estimated during the time periods a – e. In each subplot, the horizontal lines over the histograms represent the average of 95% confidence limits for the mean in each value of a. The largest average uncertainty corresponds to the dayside magnetosheath region, where the length of analyzed data was shorter. Figure 3 indicates that the specific scaling features characterize the statistical properties of different regions in space rather well. [17] During the magnetosheath periods (intervals a and d: top and bottom subplots in Figure 3) only two events display scaling indices >1.3, for the other events a 2 (0.6 1.25). This indicates that the fluctuations or waves present in the magnetosheath, except of a few cases, usually do not evolve to a fully developed turbulence state. The observed range of scaling indices suggests that the continuous part of the magnetosheath spectra is associated with 1/f noise rather than turbulence. Noisy fluctuations can be convected from the dayside magnetosheath (interval a) to the post-terminator magnetosheath downstream (interval d), where the spectral power is smaller, possibly the driving sources are more distant than at the dayside (compare the spectra corresponding to the intervals a and d in Figure 1). [18] In the post-terminator wake (intervals b in Figure 3) wavy structures dominate with periods from 5 to 50 s (similar to event b in Figure 1). Since toward higher frequencies the power rapidly decreases (a 2.5)-independent driving sources associated with broadband noise or nonlinear multiscale turbulence are absent. [19] Finally, a well discernible turbulence scaling index a 1.6 was observed during the intervals c and e (Figure 3,

5 of 11

E00B21

¨ RO ¨ S ET AL.: TURBULENCE AND FLUCTUATIONS NEAR VENUS VO

Figure 4. A cartoon showing the type of physical processes in the spatial regions of near-Venusian space; the magnetopause and the bow shock (dashed lines); a VEX crossing (solid black line); the optical shadow (shaded region); and draped IMF (thin black line). The time intervals/spatial regions a – e are marked with squares, triangles, circle, squares, and circle, respectively. third subplot from top). The events from both intervals are plotted together because the corresponding scaling indices are similar. [20] The cartoon in Figure 4 helps to interpret the occurrence of typical scaling regimes, summarizing the results of the statistical analysis along VEX trajectory in the cylindrical VSO coordinate system: [21] 1. 1/f noise appears under rather different conditions in the magnetosheath: in front of the planetary obstacle and downstream in the post-terminator magnetosheath (regions a, d: squares in Figure 4). Scaling indices which can be associated with turbulence occurred in 5% of cases (see Figure 3). [22] 2. Coherent wavy fluctuations occur between the dayside magnetopause and post-terminator boundary of the wake (largely within the optical shadow, region b: triangles in Figure 4), where noise and turbulence are absent (only magnetometer noise is present when the wave activity is absent). Closer to the terminator, the wavy structures can be associated with the Kelvin-Helmholtz instability occurring at the terminator ionopause, resulting probably in detached plasma clouds, first observed during the Pioneer Venus mission by Brace et al. [1982]. The PVO spacecraft observed filamentary structures (rays), density holes and radially aligned draped magnetic field lines in the near-planet nightside wake [Luhmann and Russell, 1983; Marubashi et al., 1985]. [23] 3. Developed turbulence is present at or near boundaries: at the wake/magnetopause/magnetosheath boundary and near the bow shock (regions c, e: circles in Figure 4).

4. Discrimination Between Boundary Turbulence and Magnetosheath Noise [24] Figure 3 shows that, in terms of spectral scaling characteristics, turbulence and noisy intervals are well

E00B21

separable. In the following we will further investigate some other differences between turbulence (intervals c and e) and noisy magnetosheath fluctuations (intervals d). We first recall some similarities between the observations obtained from previous missions and VEX near the regions with developed turbulence. [25] Early observations from Mariner 5 [Bridge et al., 1967] already showed the existence of a plasma boundary located at the inner edge of the post-terminator magnetosheath (VEX is crossing this region where the occurrence of turbulence is depicted by the bottom circle in Figure 4). During a flyby, Mariner 5 found the location of the boundary layer from 3.2 to 4.5 RV behind Venus and from 2 (magnetopause location) to 3.5 RV from the center of the wake. Within the boundary layer Mariner 5 observed a strongly fluctuating lower-intensity magnetic field, a decrease of the density and velocity, but an increase of the temperature. The magnetic field strength was larger when the spacecraft entered from the inner magnetosheath, across the magnetopause, to the wake. VEX observations show that the magnetopause location is rather variable over distances X VSO < 1 RV (open circles in Figure 2, bottom left). Therefore, multipoint turbulence fluctuations indicating the crossing of the magnetopause start in different distances from the center of the wake. For example, on 19 May 2006 (Figure 2, top left) the strongly fluctuating magnetic field strength reaches a local maximum at about 0205 UT, then it slowly decreases. It indicates that VEX is entering to the boundary layer from the wake (from intervals b to c) and at the boundary the character of fluctuations is suddenly changing, from scaling index a 2.5 to a 1.6 (see also Figure 1). [26] Another interesting feature includes the change of the sign of BX. During the interval c on 19 May it happens several times (Figure 2, top left). It can indicate that VEX is crossing the neutral sheet behind the planet. The VEX trajectory for this day (marked by an arrow in Figure 2, top right subplot) is roughly along Y VSO 0.9 RV. In fact the Venusian wake/magnetotail can be formed by flux tubes convected around or slipping over the planet and filled with the plasma from the dayside. At the same time, both ends of the flux tubes are comoving with the solar wind. In a simple case, the resulting magnetic field in the tail is stretched along the X axis, because the central part of the flux tubes is slowed down by the planet while both ends travel faster with the solar wind. Since the IMF is usually close to the ecliptic plane, the tail current sheet, in ideal case, should be formed by the stretched magnetic field mainly in the northsouth plane [Russell and Vaisberg, 1983]. Depending on the upstream IMF and its draping around the planet, the current sheet plane in the Venusian magnetotail can also rotate [Luhmann et al., 1991]. Since turbulence can drive local mixing and vorticity, the change of the magnetic field direction might be also associated with turbulence, not necessarily with neutral sheet crossing. For example, BX is also changing sign several times during the shock associated turbulence interval e (Figure 2, top right), but it has nothing common with a neutral sheet crossing. [27] The same magnetic field and plasma signatures of the inner magnetosheath boundary layer were observed by other missions, too. For example, Venera 10 observed the boundary layer tailward from the Mariner 5 flyby [Romanov

6 of 11

E00B21

¨ RO ¨ S ET AL.: TURBULENCE AND FLUCTUATIONS NEAR VENUS VO

E00B21

Figure 5. (top to bottom) Comparison of 12 min averages of hbX2i, hbY2i, and hbZ2i with the observed spectral scaling indices a. The subscripts indicate VSO magnetic field components; and the horizontal lines correspond to averages of points (open and filled circles) over the noisy a2 (0.6 1.4) and turbulent a = 1.6 ± 0.2 scaling index ranges in each subplot. The turbulent intervals include boundary layer (filled circles) and near-shock (open circles) events. For each 12 min interval, bX2(t) + bY2(t) + bZ2(t) = 1. et al., 1978]. The near-terminator part of the boundary layer was investigated during the Pioneer Venus Orbiter mission. Perez-de-Tejada et al. [1991, 1993] have seen the boundary between the bow shock and magnetopause (they are using the term ionopause instead) in the vicinity of and downstream from the terminator. The boundary layer observed in different distances along the inner edge of magnetosheath confirms the persistent presence of this boundary as a rarefaction wave emerging from the terminator magnetopause and extending downstream. The high level of magnetic fluctuations and occurrence of magnetohydrodynamic waves near a quasi-parallel Venusian bow shock is wellknown from the PVO mission [Luhmann et al., 1983]. In this paper we emphasize the detection of intermittency near the Venusian bow shock and its differentiation from magnetosheath noisy fluctuations. Also, we investigate only the magnetic field signatures of the boundary layers. 4.1. Magnetic Field Orientation [28] First, we investigate the local magnetic conditions under which turbulence (or noise) can appear. For this purpose we calculate the instantaneous contribution of magnetic field VSO components to the total magnetic field: bX(t) = BX(t)/B(t), bY(t) = BY(t)/B(t) and bZ(t) = BZ(t)/B(t). In each time step t the sum b2X(t) + b2Y(t) + b2Z(t) = 1. Our goal is to compare the average magnetic field direction during turbulent (c, e) and noisy (d) intervals, when the scaling index is well defined (see Figure 2). Since a is a statistical descriptor over 12 min long intervals the mean values hb2Xi,

hb2Yi, and hb2Zi, respectively, were computed over the same time periods; their sum is again 1. Each average defines the relative contribution of a magnetic VSO component to the average magnetic strength during the considered intervals. [29] Figure 5 compares the local mean magnetic field components with the values of scaling indices estimated during the same 12 min long intervals (open and filled circles). The horizontal lines are averages of points over the noisy a < 1.4 and turbulent a = 1.6 ± 0.2 scaling index ranges. These distinct ranges of a correspond to the statistical results in Figure 3. The filled circles correspond to turbulent intervals within the magnetosheath boundary layer, while, over the same range of as, the open circles correspond to bow shock associated turbulent intervals. [30] The horizontal lines in Figure 5 show that, over the turbulence range, the average of hb2Xi and hb2Zi; is larger while the average of hb2Yi is smaller than over the noise range of scaling indices. Therefore, when turbulence is observed hb2Xi dominates over hb2Yi and the latter is still stronger than the increased hb2Zi. Magnetosheath boundary layer associated turbulence intervals (filled circles – intervals c in Figure 2) show even larger hb2Xi values than the nearshock turbulence values (open circles – intervals e in Figure 2, over the same a range). In accordance with our findings, the analysis of PVO magnetic, electric and plasma data has shown that, near the terminator region but within the magnetosheath boundary layer, the magnetic field is nearly aligned with the Sun-Venus line (X VSO direction). It was interpreted in terms of a friction-like viscous inter-

7 of 11

E00B21

¨ RO ¨ S ET AL.: TURBULENCE AND FLUCTUATIONS NEAR VENUS VO

E00B21

Figure 6. Probability density functions (PDFs) constructed from magnetic time series (two-point differences defined through dB = B(t + t) B(t)) of (left) turbulent and (right) noisy intervals. Gaussian fits are shown as dashed gray lines. At small scales (e.g., t = 3 s) the PDF is a non-Gaussian for turbulent time series and Gaussian for noisy time series. At large scales (e.g., t = 30 s) the PDF is a Gaussian in both cases. The error bars represent 95% confidence limits for the mean in each point of dBT. action between the shocked solar wind and the ionospheric plasma over the magnetic polar regions where the draped interplanetary magnetic field lines slip over the planet [Perez-de-Tejada et al., 1993]. The viscous plasma-plasma interaction and frictional heating can explain the enhanced temperatures inside the magnetosheath boundary layer, observed tailward by Venera 10 [Romanov et al., 1978]. An alternative explanation is represented by local turbulent heating of the boundary layer plasma. We note that near the magnetopause, where VEX crosses the turbulent boundary layer, the draped interplanetary magnetic field lines are more stretched having a large X VSO component (Figure 4). Although the number of events is rather limited, the near-shock turbulent regions have smaller X VSO but larger perpendicular Y VSO and Z VSO magnetic components than the magnetosheath boundary layer turbulence (Figure 5). [31] Finally, noise (open circles – intervals d in Figure 2) is associated with a large hb2Yi, smaller hb2Xi; and almost negligible hb2Zi. Since the IMF is mostly in the X-Y plane, the large-scale magnetosheath magnetic field is not affected by the boundaries, but exhibiting only noisy broadband fluctuations and it has the strongest average components in the X-Y plane. [32] Let us investigate deeper now the intermittent nature of turbulence. Besides the characteristic scaling exponents (a 1.6) and the expected average magnetic field directions, intermittency represents another key feature of fully developed turbulence. Therefore, the occurrence of intermittency represents a further evidence that we are dealing

with a real turbulence, capable of heating the background plasma. 4.2. Turbulent Intermittency Versus Gaussianly Distributed Noise [33] Higher-order statistics is needed to fully describe the nature of nonlinear fluctuations. In turbulent, nonhomogeneous plasma flows the shape of the probability density functions (PDFs) is scale-dependent and peaked with long tails [e.g., Frisch, 1995]. Because of the shortness of time series during one crossing the shapes of non-Gaussian PDFs or their scale dependency cannot be evaluated [Vo¨ro¨s et al., 2008]. Instead, we construct PDFs from the 15 magnetic time series (realizations) of turbulence for which spectral scaling near a 1.6 was observed (events from both c and e intervals in Figure 2). Splitting the data into magnetosheath boundary layer and near-shock turbulence regions would not change the shape of turbulent PDFs significantly. PDFs corresponding to the noisy magnetosheath (intervals d in Figure 2) will be also reconstructed. [34] PDFs of two-point differences of magnetic field strength were estimated from dB = B(t + t) B(t), for t = 2. . .30 s. Figure 6 shows the PDFs for t = 3, 30 s only. Boundary turbulence associated PDFs are shown on the left, 1/f noise related PDFs on the right-hand side. The error bars represent 95% confidence limits for the mean in each point of dB. The dashed gray points are least squares Gaussian fits to the experimental two-point PDFs. The ð x mÞ2 , where m Gaussian PDF is given by p1ffiffiffiffiffiffiffi exp 2s2 s 2ðpÞ 2 is the mean and s is the variance. For the smaller timescale t = 3 s the tails of the experimental turbulent PDFs are

8 of 11

¨ RO ¨ S ET AL.: TURBULENCE AND FLUCTUATIONS NEAR VENUS VO

E00B21

E00B21

Figure 7. The evolution of (K, top) kurtosis and (S, bottom) skewness with the timescale t, computed from two-point differences defined through dB = B(t + t) B(t)) of (left) turbulent and (right) noisy time intervals. K is peakedness of the PDF; and S is asymmetry around the mean, both relative to a Gaussian distribution, for which S = K = 0; The error bars represent 95% confidence limits for the mean in each value of t. higher than the Gaussian tails. The departure from the Gaussian indicates that nonhomogeneously distributed fluctuations become more probable as the scale decreases because of turbulent structures and long-range interactions [Leubner and Vo¨ ro¨ s, 2005]. Gradual decorrelation is obtained by enhancing the two-point separation scale and a Gaussian is approached for large enough t even in a turbulent field. This is because the typical correlations for turbulent structures are lost if the separation is large, and only Gaussianly distributed noise is observed. For t = 30 s, the PDF is a Gaussian in Figure 6, left. Noise shows Gaussianly distributed PDFs over both timescales. [35] Let us further investigate the shape of PDFs in terms of statistical moments. Using standard procedures [Press et al., 1992], the skewness (S) or the third moment, S ðt Þ ¼

N xj x 1 X N j¼1 s

3

ð1Þ

and the kurtosis (K) or the fourth moment, K ðt Þ ¼

N xj x 1 X N j¼1 s

4

3

ð2Þ

are computed for turbulent and noisy intervals, as above. Here xj = d B(tj, t), s is the standard deviation, x is the mean value of the elements and N is the number of the data points. In this way the dimensionless S(t) and K(t) characterize the scale and time evolution of the shape of a distribution (S is

asymmetry around the mean; K is peakedness or flatness; both relative to a Gaussian distribution, for which S = K = 0). K increases toward small scales in intermittent turbulence [Frisch, 1995]. [36] Figure 7 shows the timescale (t) evolution of kurtosis K(t) (top) and skewness S(t) (bottom) for turbulent (left) and noisy (right) intervals. The error bars represent 95% confidence limits for the mean in each value of t. As is expected from turbulent PDFs in Figure 6, K is increasing as t decreases, which proves that the underlying magnetic fluctuations are non-Gaussian, peaked and intermittent. K practically does not depend on t in the case of 1/f noise, indicating a Gaussianly distributed process. [37] The skewness remains close to zero (Figure 7, bottom) in both cases, showing symmetric distributions around the mean value.

5. Discussion and Conclusions [38] In this paper the unique data from the VEX spacecraft were used to investigate magnetic fluctuation statistics from Venusian magnetosheath and wake regions. The interaction of the solar wind with the planet drives magnetic fluctuations exhibiting different scaling regimes in different regions of near-Venusian space. To identify spectral scaling ranges and indices, we used a wavelet technique, successfully applied for studying the continuous spectra in the Earth’s plasma sheet turbulence [Vo¨ro¨s et al., 2004]. The technique and the data intervals were not optimized for finding waves (peaks in power spectra).

9 of 11

E00B21

¨ RO ¨ S ET AL.: TURBULENCE AND FLUCTUATIONS NEAR VENUS VO

[39] Three types of scaling were observed. Inside the dayside/tailward magnetosheath, far from boundaries, 1/f noise is present in the prevailing majority of cases indicating the contribution of multiple independent driving sources. It also means, that this type of spectral scaling is not formed by an isolated single physical process. In other words, there might exist multiple sources of fluctuations, but no one of them is close enough in space to dominate in the spectral power. This changes when VEX enters to a region where multiple sources are missing, or where a single physical process dominates with a scaling feature other than that of the noise. In fact, these are distinct regions of nearVenusian space where the ‘‘solar wind – planetary obstacle’’ interactions are enhanced or the multiple sources of noise are shielded. [40] The interaction is enhanced at the terminator ionopause, probably due to the Kelvin-Helmholtz instability, at the magnetosheath boundary layer due to the magnetosheath boundary shear flows and near the quasi-parallel bow shock. The near-planet wake represents a region which is shielded from the plasma of solar wind origin. Filamentary structures, detached plasma clouds, depleted density holes and radially aligned draped magnetic field lines were observed by PVO spacecraft in the near-planet nightside wake [Luhmann and Russell, 1983; Marubashi et al., 1985]. The magnetosheath flow is expected to converge into the wake only near 5 RV behind the planet [Intriligator et al., 1979]. The observed wavy structures near the terminator and in the nightside near-planet wake can be associated with the detached coherent structures or holes. The occurrence/ absence of these structures can be controlled by the direct interaction between the solar wind and ionosphere, e.g., by the high/low solar wind dynamic pressure. In our interpretation, the spectral index a 2.5 indicates, that the coherent wavy structures represent the dominating physical process in this region. Because of the shielding of the near-planet wake, turbulence or noise are absent in this region. Because of the converging flows the shielding can disappear at distances close to or larger than X VSO 5 RV, where the character of fluctuations would change. [41] The magnetosheath regions with distinct scaling indices (turbulence or noise) partially overlap (see Figure 2). This can be explained through a movement of boundaries under the influence of changing upstream IMF conditions. Spatial intermittency, typical for turbulence with scale-dependent non-Gaussian distributions, can lead also to interwoven scaling structures. The turbulent regions are formed near the ‘‘supersonic solar wind flow – planetary obstacle’’ boundaries in the presence of draped IMF. The outer boundary is the bow shock, where the solar wind slows down and heated for the first time. Here, the local turbulence is associated with the quasi-parallel shock geometry. Second time the solar wind flow decelerates at the inner magnetosheath producing a velocity shear near the magnetosheath boundary layer. The near-terminator ionosphere/magnetopause and its interaction with the solar wind plays an important role. The rarefaction wave (the boundary) and the observed plasma conditions in the tailward boundary layer can emerge from the magnetopause near the terminator and extend downstream [Perez-de-Tejada et al., 1991]. Local heating of the plasma within the boundary layers is also possible through the shear flow associated

E00B21

turbulence. The spatial size of boundary layer turbulence (see Figure 2) is roughly between 0.5 and 1 RV, shorter turbulent intervals were observed near the quasi-parallel bow shock. The width of the noisy magnetosheath in between the turbulent boundaries is of the same order. The estimation of the spatial sizes of these regions is rather rough. The data are available only from single point measurements (the boundaries can move during the measurements) and the 12 min long data intervals represent a pure resolution. The estimation of scaling indices within shorter time intervals, however, was not possible due to the large statistical errors in these cases. Multipoint Cluster observations show, that turbulence downstream of the quasiparallel terrestrial bow shock is intermittent and that the level of intermittency increases over the spacecraft separation, reaching larger values than 8000 km [Yordanova et al., 2008]. Our results show, that the spatial scale of intermittent turbulence is less than 1 RV near the Venusian bow shock. In between the near-shock region and the magnetosheath boundary layer fluctuations are noisy (in Figure 4, squares between triangles). [42] It was shown by Perez-de-Tejada et al. [1993] that the magnetic field is nearly aligned with the Sun-Venus line (X VSO direction) within the near-terminator magnetosheath boundary layer. We found a similar alignment within the boundary layer at a distance of X VSO 2.2 RV (Figures 2 and 5). It indicates, that the magnetic field geometry detected during the viscous plasma-plasma interactions near the terminator ionosphere is conserved and observed further downstream along the VEX trajectory, where the draped magnetic field is stretched (Figure 4). [43] Data intervals that were found to be turbulent or noisy in the spectral analysis were further investigated using the two-point (time delayed) probability density functions (PDFs). Over large two-point separations (e.g., t = 30 s), both turbulent and noisy PDFs are well fitted by the Gaussian distributions, indicating the occurrence of uncorrelated fluctuations. Over small separations (e.g., t = 3 s), large deviations from the Gaussian distribution are observed only for turbulent intervals, noise remains Gaussianly distributed (Figure 6). The scale dependency of kurtosis (Figure 7) shows that turbulent structures are intermittently distributed, noise is more homogeneous. Skewness remains close to zero in both cases which corresponds to symmetrical distributions. A simultaneous increase of S and K toward small scales would indicate that the multiscale fluctuations in turbulence might be affected by strong large-scale gradients or close boundaries [Vo¨ro¨s et al., 2007]. Therefore, S(t) 0 for t 2 (2 – 30) s is a signature of intermittency, not affected by boundaries along the VEX trajectory over the scale of seconds. However, asymmetries or nonzero skewness in magnetic field statistics can appear below the 1 s timescale. Anyhow, besides the expected spectral index (a 1.6), the observation of intermittency represents a further evidence for the occurrence of real turbulence in the near-Venusian space. The key feature of turbulence is its strong dissipative nature and a capability for the local heating of plasma. Multiscale turbulence can channel the large-scale energy of the flow to kinetic scales, where dissipation processes are strong. We will investigate this point in a different paper.

10 of 11

E00B21

¨ RO ¨ S ET AL.: TURBULENCE AND FLUCTUATIONS NEAR VENUS VO

[44] Acknowledgments. The work of Z.V. and M.P.L. was supported by the Austrian Wissenschaftsfonds under grant P20131-N16.

References Abry, P., P. Flandrin, M. S. Taqqu, and D. Veitch (2000), Wavelets for the analysis, estimation and synthesis of scaling data, in Self-Similar Network Traffic and Performance Evaluation, edited by K. Park and W. Willinger p. 39, Wiley-Intersci., Hoboken, N. J. Amerstorfer, U. V., N. V. Erkaev, D. Langmayr, and H. K. Biernat (2007), On Kelvin-Helmholtz instability due to the solar wind interaction with unmagnetized planets, Planet. Space Sci., 55, 1811 – 1816, doi:10.1016/ j.pss.2007.01.015. Bauer, S. J., L. H. Brace, D. M. Hunten, D. S. Intriligator, W. C. Knudsen, A. F. Nagy, C. T. Russell, F. L. Scarf, and J. H. Wolfe (1977), The Venus ionosphere and solar wind interaction, Space Sci. Rev., 20, 413 – 430. Biernat, H. K., N. V. Erkaev, U. V. Amerstorfer, T. Penz, and H. I. M. Lichtenegger (2007), Solar wind flow past Venus and its implications for the occurrence of the Kelvin-Helmholtz instability, Planet. Space Sci., 55, 1793 – 1803, doi:10.1016/j.pss.2007.01.015. Brace, L. H., R. F. Theis, and W. R. Hoegy (1982), Plasma clouds above the ionopause of Venus and their implications, Planet. Space Sci., 30, 29 – 37. Bridge, H. S., A. J. Lazarus, C. W. Snyder, E. J. Smith, L. Davis Jr., P. J. Coleman Jr., and D. E. Jones (1967), Mariner V: Plasma and magnetic fields observed near Venus, Science, 158, 1669 – 1673. Elphic, R. C., C. T. Russell, J. G. Luhmann, F. L. Scarf, and L. H. Brace (1981), The Venus ionopause current sheet: Thickness, length, scale and controlling factors, J. Geophys. Res., 86, 11,430 – 11,438. Frisch, U. (1995), Turbulence, the Legacy of A.N. Kolmogorov, 296 pp. Cambridge Univ. Press, New York. Intriligator, D. S., H. R. Collard, J. D. Mihalov, R. C. Whitten, and J. H. Wolfe (1979), Electron observations and ion flow from the Pioneer Venus Orbiter Plasma Analyzer Experiment, Science, 205, 116 – 119. Leubner, M. P., and Z. Vo¨ro¨s (2005), A nonextensive entropy approach to solar wind intermittency, Astrophys. J., 618, 547 – 555. Luhmann, J. G., and C. T. Russell (1983), Magnetic fields in the ionospheric holes of Venus: Evidence for an intrinsic field?, Geophys. Res. Lett., 10, 409 – 412. Luhmann, J. G., M. Tatrallyay, C. T. Russell, and D. Winterhalter (1983), Magnetic field fluctuations in the Venus magnetosheath, Geophys. Res. Lett., 10, 655 – 658. Luhmann, J. G., C. T. Russell, K. Schwingenschuh, and Y. Yeroshenko (1991), A comparison of induced magnetotails of planetary bodies: Venus, Mars and Titan, J. Geophys. Res., 96, 11,199 – 11,208. Marubashi, K., J. M. Grebowsky, H. A. Taylor, J. G. Luhmann, and C. T. Russell (1985), Ionosheath plasma flow in the wake of Venus and the formation of ionospheric holes, J. Geophys. Res., 90, 1385 – 1398. Pe´rez-de-Tejada, H., D. S. Intriligator, and R. J. Strangeway (1991), Steadystate plasma transition in the Venus ionosheath, Geophys. Res. Lett., 18, 131 – 134. Pe´rez-de-Tejada, H., D. S. Intriligator, and R. J. Strangeway (1993), Magnetic field properties of the intermediate transition of the Venus ionosheath, Geophys. Res. Lett., 20, 991 – 994. Press, W. H., B. P. Flannery, S. A. Teukolsky, and W. T. Vetterling (1992), Numerical Recipes in C: The Art of Scientific Computing, 994 pp., Cambridge Univ. Press.

E00B21

Romanov, S., V. Smirnov, and O. Vaisberg (1978), Interaction of the solar wind with Venus, Kosmich. Issled., 16, 746. Russell, C. T. (1992), The Pioneer Venus mission, in Venus and Mars: Atmospheres, Ionospheres and Solar Wind Interactions, Geophys Monogr. Ser., vol. 66, edited by J. G. Luhmann, M. Tatrallyay, and R. O. Repin, pp. 225 – 236, AGU, Washington, D. C Russell, C. T., and O. Vaisberg (1983), The interaction of the solar wind with Venus, in Venus, edited by D. M. Hunton et al., pp. 873 – 940, Univ. of Ariz. Press, Tucson. Spreiter, J. R., and S. S. Stahara (1992), Computer modeling of solar wind interaction with Venus and Mars, in Venus and Mars: Atmospheres, Ionospheres and Solar Wind Interactions, Geophys Monogr. Ser., vol. 66, edited by J. G. Luhmann, M. Tatrallyay, and R. O. Repin pp. 345 – 383, AGU, Washington, D. C Terada, N., S. Machida, and H. Shinagawa (2002), Global hybrid simulation of the Kelvin-Helmholtz instability at the Venus ionopause, J. Geophys. Res., 107(A12), 1471, doi:10.1029/2001JA009224. Volwerk, M., T. L. Zhang, M. Delva, Z. Vo¨ro¨s, W. Baumjohann, and K.-H. Glassmeier (2008a), First identification of mirror mode waves in Venus’ magnetosheath?, Geophys. Res. Lett., 35, L12204, doi:10.1029/ 2008GL033621. Volwerk, M., T. L. Zhang, M. Delva, Z. Vo¨ro¨s, W. Baumjohann, and K.-H. Glassmeier (2008b), Mirror-mode-like structures in Venus’ induced magnetosphere, J. Geophys. Res., 113, E00B16, doi:10.1029/2008JE003154. Vo¨ro¨s, Z., et al. (2004), Magnetic turbulence in the plasma sheet, J. Geophys. Res., 109, A11215, doi:10.1029/2004JA010404. Vo¨ro¨s, Z., W. Baumjohann, R. Nakamura, A. Runov, M. Volwerk, T. Takada, E. A. Lucek, and H. Re`me (2007), Spatial structure of plasma flow associated turbulence in the Earth’s plasma sheet, Ann. Geophys., 25, 13 – 17. Vo¨ro¨s, Z., T. L. Zhang, M. P. Leubner, M. Volwerk, M. Delva, W. Baumjohann, and K. Kudela (2008), Magnetic fluctuations and turbulence in the Venus magnetosheath and wake, Geophys. Res. Lett., 35, L11102, doi:10.1029/2008GL033879. Wolff, R., B. Goldstein, and C. Yeates (1980), The onset and development of Kelvin-Helmholtz instability at the Venus ionopause, J. Geophys. Res., 85, 7697 – 7707. Yordanova, E., A. Vaivads, M. Andr, S. C. Buchert, and Z. Vo¨ro¨s (2008), Magnetosheath plasma turbulence and its spatiotemporal evolution as observed by the Cluster spacecraft Phys, Rev. Lett., 100, 205003-1 – 205003-4. Zhang, T. L., et al. (2006), Magnetic field investigation of the Venus plasma environment: Expected new results from Venus Express, Planet. Space Sci., 54, 1336 – 1343. Zhang, T. L., et al. (2007), Little or no solar wind enters Venus’ atmosphere at solar minimum, Nature, 450, 654 – 656, doi:10.1038/nature06026. W. Baumjohann, M. Delva, M. Volwerk, and T. L. Zhang, Space Research Institute, Austrian Academy of Sciences, Schmiedlstrasse 6, A-8042 Graz, Austria. M. P. Leubner and Z. Vo¨ro¨s, Institute of Astro- and Particle Physics, University of Innsbruck, Technikerstrasse 25, A-6020 Innsbruck, Austria. ([email protected])

11 of 11

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, E00B12, doi:10.1029/2008JE003128, 2008

for

Full Article

Venus Express observations of an atypically distant bow shock during the passage of an interplanetary coronal mass ejection T. L. Zhang,1,2 S. Pope,3 M. Balikhin,3 C. T. Russell,4 L. K. Jian,4 M. Volwerk,1 M. Delva,1 W. Baumjohann,1 C. Wang,2 J. B. Cao,2 M. Gedalin,5 K.-H. Glassmeier,6 and K. Kudela7 Received 11 March 2008; revised 4 July 2008; accepted 4 August 2008; published 22 October 2008.

[1] On 10–11 September 2006 the Venus Express magnetometer detected a very

strong Interplanetary Coronal Mass Ejection (ICME) event with an average field about 2 times higher than that of a typical ICME at 0.72 AU. While the effective obstacle to the solar wind is compressed to a smaller dimension during this ICME event, the bow shock is located far upstream of its nominal location. The observed shocks are weak and appear very dynamic. The location of the shock crossing can be found all along the Venus Express trajectory, which has an apocenter of 12 RV. We attribute the atypical distant bow shock location as an effect of the extremely low Mach number during the ICME. Citation: Zhang, T. L., et al. (2008), Venus Express observations of an atypically distant bow shock during the passage of an interplanetary coronal mass ejection, J. Geophys. Res., 113, E00B12, doi:10.1029/2008JE003128.

1. Introduction [2] Understanding the response of the planetary environment to extreme solar conditions is crucial not only in space weather studies but also in reconstructing the evolution of planetary atmospheres. Recent studies suggest that extreme solar events might play an important role in the evolution of Venus and Mars atmospheres and their water loss processes [Kulikov et al., 2006; Luhmann et al., 2007; Terada et al., 2008]. In searching for clues to the nature and strength of atmospheric evolution, one has to develop models to simulate the solar wind interaction in ancient times. These models of necessity must be tested under present solar wind conditions. Since extreme solar events could provide conditions which might more resemble the ancient solar wind, it is important to study and gather information on these events which presently rarely occur. [3] Here we report observations of such an extreme solar event encountered by Venus Express. While the event itself is worth studying from several different aspects, we limit our effort in this paper to the atypical distant bow shock location. It is well known that the standoff distance of a 1 Space Research Institute, Austrian Academy of Sciences, Graz, Austria. 2 State Key Laboratory of Space Weather, Chinese Academy of Sciences, Beijing, China. 3 Department of Automatic Control and Systems Engineering, University of Sheffield, Sheffield, UK. 4 Institute of Geophysics and Planetary Physics, University of California, Los Angeles, California, USA. 5 Department of Physics, Ben Gurion University, Beer-Sheva, Israel. 6 Institut fu¨r Geophysik und Extraterrestrische Physik, Technische Universita¨t Braunschweig, Braunschweig, Germany. 7 Institute of Experimental Physics, Slovak Academy of Sciences, Kosice, Slovakia.

Copyright 2008 by the American Geophysical Union. 0148-0227/08/2008JE003128$09.00

shock in front of an obstacle depends on the size and the shape of the obstacle and the Mach number. In the case of Venus which has no intrinsic magnetic field, the obstacle is formed by the highly conducting ionosphere plus magnetic barrier, which is more stable to the solar wind dynamic pressure variations than the Earth’s magnetosphere. In fact, the location of the Venus bow shock does not vary with solar wind pressure [Zhang et al., 2004]. [ 4 ] Many previous researchers have examined the response of the Venus bow shock position to various external solar wind conditions [e.g., Slavin et al., 1980; Tatrallyay et al., 1983; Russell et al., 1988; Zhang et al., 1990]. It was found that Venus bow shock is controlled by solar EUV flux, solar wind Mach number, and interplanetary magnetic field (IMF) orientation. However, the data for the above studies are mainly from normal solar conditions. For example, no extremely low Mach number cases were included in these studies. [5] Russell and Zhang [1992] first presented two cases of an atypical distant Venus bow shock which were found at 12 RV above terminator. This motivated the modification of the theoretical models for bow shock [e.g., Farris and Russell, 1994; Verigin et al., 2004]. Several cases of distant bow shocks have also been observed at Mars and Earth [Slavin et al., 1994; Fairfield et al., 2001]. Nevertheless, the catalog of the low Mach number bow shock cases is still very small, making it difficult to choose between the existing bow shock models. In this paper we present Venus Express observations of unusual distant bow shock during an Interplanetary Coronal Mass Ejection (ICME) passage. We find that the dynamics of the shock location is controlled by magnetosonic Mach number.

2. Observations [6] Venus Express spacecraft has a polar orbit of 90° inclination, a low periapsis altitude of 250– 350 km at 80°N,

E00B12

1 of 5

E00B12

ZHANG ET AL.: VENUS DISTANT BOW SHOCK

Figure 1. Time series of interplanetary magnetic field behavior during an Interplanetary Coronal Mass Ejection (ICME) passage. The darker shade is the sheath, and the lighter shade is the flux rope of ICME. and a 24-h period orbit [Titov et al., 2006]. The magnetometer aboard measures the magnetic field continuously along the orbit in various modes [Zhang et al., 2006, 2007]. The data used in this study are at 1 Hz resolution in Venus Solar Orbital (VSO) coordinates where the x axis points from Venus to the Sun, the y axis is opposite to the Venus orbital motion, and the z axis is northward. [7] Figure 1 shows the magnetic field data on 10– 11 September 2006. During the 2 days, Venus Express observed an ICME with a leading shock at 1413 UT on 10 September and a 4-h sheath region thereafter. There was a well-defined left-handed flux rope (1817 UT 10 September-1640 UT 11 September). Within the flux rope, Bx stayed close to zero; By rotated from westward to eastward, while Bz rotated from northward to southward and then remained southward for about 19 h. [8] During a typical orbit, the duration between inbound and outbound bow shock crossing is about 2 h and multiple crossings rarely occur because of the favorable orbital geometry. [9] In Figure 2, we plot only the total field magnitude for part of the flux rope to illustrate the multiple shock crossings during this event. The first inbound bow shock was encountered at 2209 UT on 10 September 2006, and the last outbound shock was encountered at 1220 UT on 11 September 2006. Essentially, the shock crossing is rather ‘‘randomly’’ distributed along the Venus Express orbital trajectory. Altogether, we observe 18 shock crossings. In Table 1 we list for each shock crossing the timing, solar zenith angle (SZA), the shock distance, the upstream magnetic field magnitude Bu, and the shock strength Bd/Bu which is the ratio between upstream and down field magnitudes. We use the coplanarity assumption to determine the shock normal and to obtain qBn angle which is the angle between the shock normal and the upstream magnetic field. The magnetosonic Mach number listed in Table 1 is determined from the shock strength and qBn based on the Rankine-Hugoniot equations by assuming the plasma is isotropic and Maxwellian upstream and downstream, b as 1,

E00B12

and the ratio of specific heats of the plasma as 5/3 [Kivelson and Russell, 1995]. [10] Figure 3 displays nine distant bow shocks observed on 10– 11 September. Figure 3a, 3d, and 3g display the magnitude of magnetic field as measured by Venus Express magnetometer for three relatively oblique shocks (qBn angle around 50°) with their Mach number increasing from top to bottom. The shocks in Figures 3b, 3e, and 3h have a qBn angle of around 60° with Mach number increasing from 1.05 to 1.20. Figures 3c, 3f, and 3i display crossings of almost perpendicular shocks with qBn angle around 80° and mach number increasing from 1.18 to 1.40. All nine shocks display a front profile typical for subcritical shocks as the change in field magnitude that takes place in the ramp is 0.15– 0.70 Bu. All of the shocks except that in Figure 3d exhibit a whistler wave precursor typical of dispersive shocks, together with oscillations downstream of the ramp indicating the kinematic nature of ion relaxation [Balikhin et al., 2008]. Typical crossings of interplanetary shocks with such a high relative spacecraft shock speed do not allow the observation of the structure of the shock front and whistler wave precursor. On the contrary, all nine plots displayed in Figure 3 show well defined precursors providing another argument in favor of bow shock crossings as opposed to interplanetary shocks. [11] Figure 4 shows the locations of the shock crossings during this ICME passage. Also depicted is the average shock position model derived from the initial Venus Express observations at solar minimum [Zhang et al., 2008a]. As shown in Figure 4, on 10– 11 September 2006, Venus’ bow shock expanded far upstream and its extension was only limited by the Venus Express apocenter distance of 12 RV. It is notable that scaling this distance to the Earth’s magnetospheric obstacle, this would be equivalent to finding the bow shock at a distance of 180 RE above the terminator. Thus all the unusual distant bow shock cases summarized by Fairfield et al. [2001] are far less distant compared to the case presented here. In fact, such a very atypical distant bow shock as recorded by Venus Express on 10– 11 September, occurred only twice, namely, on 6 July 1981 and 4 August 1987, both at Venus [Russell and Zhang, 1992].

3. Discussion [12] As mentioned in the introduction, it was found that Venus bow shock is controlled by EUV flux, solar wind Mach number and IMF orientation. The effect of the EUV flux is mainly due to its control of the obstacle size, i.e., the upper atmosphere neutral and ionospheric ions scale

Figure 2. Total magnitude of the magnetic field observed during ICME with 18 multiple bow shock crossings indicated by the short vertical lines.

2 of 5

ZHANG ET AL.: VENUS DISTANT BOW SHOCK

E00B12

E00B12

Table 1. Parameters of Distant Bow Shock Crossings on 10 – 11 September 2006 Day

Shock Crossing Time (UT)

SZAa (deg)

Shock Distance (RV)

2208:57 2211:19 2214:47 2219:57 2221:49 2301:03 2303:31 0422:07 0434:12 0443:49 0921:19 0927:50 0944:22 0957:17 1158:19 1210:05 1216:18 1220:24

112 113 113 113 113 116 116 40 45 49 81 82 82 83 88 89 89 89

9.43 9.40 9.35 9.28 9.25 8.64 8.59 2.80 3.28 3.64 9.72 9.81 10.01 10.16 11.27 11.35 11.39 11.41

10 September 2006

11 September 2006

Shock Normal 0.49, 0.15, 0.28, 0.32, 0.49, 0.28, 0.81, 0.04, 0.46, 0.33, 0.60, 0.20, 0.73, 0.18, 0.80, 0.16, 0.95, 0.26, 0.63, 0.77, 0.79, 0.12, 0.10, 0.94, 0.81, 0.27, 0.65, 0.61, 0.67, 0.02, 0.71, 0.07, 0.87, 0.06, 0.91, 0.41,

0.86 0.91 0.83 0.58 0.82 0.78 0.66 0.57 0.17 0.13 0.60 0.31 0.52 0.45 0.74 0.71 0.48 0.04

Bu (nT)

qBn (deg)

Bd/Bu

Magnetosonic Mach Number

31.49 31.97 32.15 31.99 31.69 32.24 32.28 32.97 31.54 32.06 33.47 32.98 32.93 31.47 28.85 28.26 28.46 27.55

64 81 71 81 69 80 78 72 79 52 49 62 61 52 45 53 59 60

1.29 1.27 1.26 1.34 1.30 1.31 1.30 1.52 1.35 1.33 1.10 1.10 1.09 1.12 1.08 1.09 1.07 1.10

1.20 1.18 1.19 1.23 1.21 1.20 1.22 1.40 1.24 1.28 1.06 1.08 1.05 1.07 1.07 1.06 1.05 1.05

a

SZA, solar zenith angle.

heights. While EUV variation is mainly responsible for the Venus bow shock solar cycle variation [cf. Russell et al., 1988], it probably has lesser effect on the daily shock location variation. Examining the magnetic field data on 10– 11 September 2006, we find that the effective obstacle,

the magnetopause of the induced magnetosphere, is located at 391 km altitude with a SZA of 71°, which is much lower than its averaged location of 738 km [Zhang et al., 2008b]. This is probably due to the enhancement of solar wind dynamic pressure within the ICME. Typically, the maxi-

Figure 3. Nine distant bow shock crossings grouped in columns according to qBn angle and sorted in rows according to Mach number. 3 of 5

E00B12

ZHANG ET AL.: VENUS DISTANT BOW SHOCK

E00B12

Figure 4. Venus Express trajectories during 10– 11 September with mean bow shock from Zhang et al. [2008a]. Small dots display the distant bow shock events during ICME passage. Short lines are the shock normal determined from coplanarity. mum solar wind dynamic pressure within ICMEs is about 20.7 nPa which is about 3 times higher than average solar wind condition at solar minimum [Jian et al., 2007]. [13] Thus, the dynamic of the bow shock location in this event can only be explained by low Mach number and IMF orientation. Russell and Zhang [1992] found that when the Mach number approaches unity, the bow shock could be very dynamic. The models by Farris and Russell [1994] and Verigin et al. [2004] both indicate that the bow shock location is a very sensitive function of Mach number at low Mach numbers. [14] Another indirect Mach number effect is the dependence of the bow shock on IMF orientation. Strictly speaking, Mach number effect includes the IMF orientation since the IMF orientation comes into play when one considers the magnetosonic Mach number. However, in common practice, one use the perpendicular magnetosonic Mach number in the solar wind to refer as the Mach number and calculates only the nose location. The field direction effect is omitted from the Mach number effect as are effects due to the flaring angle. Since qBn, the angle between the shock normal and the upstream magnetic field, varies along the surface of the bow shock, the magnetosonic Mach number varies. This qBn effect can move the quasi-parallel

shock to be closer to the obstacle than the quasi-perpendicular shock as observed by Zhang et al. [1991]. The resulting qBn effect on the bow shock location was determined to be 0.1 RV. This is a rather small value. However, it is worth to point out that the 0.1 RV qBn effect was determined for normal solar wind conditions with an average Mach number of 5. This should not be true when the Mach number is very low. [15] To combine the IMF orientation effect into the Mach number effect, we determine the real magnetosonic Mach number using the shock strength and qBn based on the Rankine-Hugoniot equations in this study. The Mach number moderately varies between 1.05 and 1.4; thus, the shock remains in the subcritical regime. The extreme low Mach number is responsible for both the unusually distant and dynamics of the shock location during the ICME passage. [16] While ICMEs at Venus are often observed, extremely distant bow shock observations are very rare. Since the beginning of Venus Express magnetic field measurements in April 2006, many ICMEs were identified [Luhmann et al., 2007], but distant bow shocks were only identified during the ICME presented here. Recently, Jian et al. [2007] surveyed PVO data during 1979–1988, and found 124 ICMEs. However, distant bow shock crossings were only reported

4 of 5

E00B12

ZHANG ET AL.: VENUS DISTANT BOW SHOCK

for the two events of 6 July 1981 and 4 August 1987 by Russell and Zhang [1992], who also surveyed all the high field data. [17] It is interesting to examine why the majority of ICMEs do not move the bow shock to an atypical distant location; that is, it is interesting to compare the properties of 10– 11 September ICME case to typical ICME properties. Figure 1 shows the whole ICME lasted about 26.45 h, with a maximum magnetic field (Bmax) of 35 nT. This is near values for an average ICME at 0.72 AU, which has a median duration of 23.2 h and a Bmax of 26 nT [Jian et al., 2007]. In addition, the ratio of sheath duration to the whole ICME duration is about 15%, close to the median value (16%) of ICMEs at 0.72 AU. [18] The only notable difference of this ICMEs from the average ICMEs is the average magnetic field magnitude. As we can see in Figure 1, the ICME field magnitude is rather steady and stays above 30 nT most of the time within the flux rope. This is about 2 times higher than the average field value of 16 nT within a typical ICME [Forsyth et al., 2006]. Reexmaining the distant bow shock events from Russell and Zhang [1992], we find that those events also occur when the magnetic field magnitude is about 30 –40 nT. The magnetosonic Mach number consists of both sonic and Alfvenic Mach numbers. Since the Alfvenic Mach is inversely proportional to magnetic field magnitude, a twofold increases of the field can lead to a factor of 2 decrease of the Alfvenic Mach number. In this ICME event, the persistent abnormal enhanced magnetic field within flux rope might be the reason of low Mach number which resulted in the rare distant bow shock observations. [19] In summary, Venus plasma environment provides a better natural laboratory to study the fundamental plasma physics since the obstacle to solar wind is rather stable and less variable in comparison with the Earth’s magnetosphere. This is particularly true during the ICME passages when the solar wind dynamic pressure is about 3 times higher so that the ionosphere is magnetized and the obstacle is less compressive. In addition, the slowly varying IMF direction, strength and other solar wind parameters provide a quasisteady background condition, making it easier to distinguish between spatial and temporal variations. Finally, such extreme solar events provide rare vehicles for testing simulation and modeling to understand planetary atmosphere evolution. [20] Acknowledgments. The work at UCLA was supported by NASA under research grant NNG06GC62G. The work in China was supported by NNSFC grants 40621003, 40628003, and by 973 Program 2006CB806305; M.G. was partially supported by ISF grant 275/07; The work in Slovakia was supported by Slovak Research and Development Agency under the contract APVV-51-053805.

References Balikhin, M. A., T. L. Zhang, M. Gedalin, N. Y. Ganushkina, and S. A. Pope (2008), Venus Express observes a new type of shock with pure kinematic relaxation, Geophys. Res. Lett., 35, L01103, doi:10.1029/ 2007GL032495. Fairfield, D. H., I. H. Cairns, M. D. Desch, A. Szabo, A. J. Lazarus, and M. R. Aellig (2001), The location of low Mach number bow shocks at Earth, J. Geophys. Res., 106, 25,361 – 25,376, doi:10.1029/ 2000JA000252. Farris, M. H., and C. T. Russell (1994), Determining the standoff distance of the bow shock: Mach number dependence and use of models, J. Geophys. Res., 99, 17,681 – 17,689, doi:10.1029/94JA01020.

E00B12

Forsyth, R. J., et al. (2006), ICMEs in the inner heliosphere: Origin, evolution and propagation effects, Space Sci. Rev., 123, 383 – 416, doi:10.1007/ s11214-006-9022-0. Jian, L. K., C. T. Russell, J. G. Luhmann, R. M. Skoug, and J. T. Steinberg (2007), Stream interactions and interplanetary coronal mass ejections at 0.72 AU, Sol. Phys., 249, 85 – 101. Kivelson, M. G., and C. T. Russell (Eds.) (1995), Introduction to Space Physics, 568 pp., Cambridge Univ. Press, Cambridge, U. K. Kulikov, Y. N., H. Lammer, H. I. M. Lichtenegger, N. Terada, I. Ribas, H. K. Biernat, C. Kolb, E. F. Guinan, S. Barabash, and R. Lundin (2006), Atmospheric and water loss from early Venus, Planet. Space Sci., 54, 1425 – 1444, doi:10.1016/j.pss.2006.04.021. Luhmann, J. G., W. T. Kasprzak, and C. T. Russell (2007), Space weather at Venus and its potential consequences for atmosphere evolution, J. Geophys. Res., 112, E04S10, doi:10.1029/2006JE002820. Russell, C. T., and T. L. Zhang (1992), Unusually distant bow shock encounters at Venus, Geophys. Res. Lett., 19, 833 – 836, doi:10.1029/ 92GL00634. Russell, C. T., E. Chou, J. G. Luhmann, P. Gazis, L. H. Brace, and W. R. Hoegy (1988), Solar and interplanetary control of the location of the Venus bow shock, J. Geophys. Res., 93, 5461 – 5469, doi:10.1029/ JA093iA06p05461. Slavin, J. A., R. C. Elphic, C. T. Russell, F. L. Scarf, J. H. Wolfe, J. D. Mihalov, D. S. Intriligator, L. H. Brace, H. A. Taylor Jr., and R. E. Daniell Jr. (1980), The solar wind interaction with Venus: Pioneer Venus observations of bow shock location and structure, J. Geophys. Res., 85, 7625 – 7641, doi:10.1029/JA085iA13p07625. Slavin, J. A., et al. (1994), The solar wind interaction with Mars: Phobos-2 bow shock observations on 24 March 1989, in COSPAR Colloquia, vol. 4, edited by T. I. Gombosi, pp. 279 – 283, Pergamon, Amsterdam. Tatrallyay, M., C. T. Russell, J. D. Mihalov, and A. Barnes (1983), Factors controlling the location of the Venus bow shock, J. Geophys. Res., 88, 5613 – 5621, doi:10.1029/JA088iA07p05613. Terada, N., et al. (2008), The role of the induced magnetic field and ion loss of atmosphere and water from early Mars under extreme solar wind and XUV conditions, Astrobiology, in press. Titov, D. V., et al. (2006), Venus Express science planning, Planet. Space Sci., 54, 1279 – 1297, doi:10.1016/j.pss.2006.04.017. Verigin, M. I., et al. (2004), Unusually distant bow shock encounters at Mars: Analysis of March 24, 1989 event, Space Sci. Rev., 111, 233 – 243, doi:10.1023/B:SPAC.0000032713.86796.d1. Zhang, T. L., J. G. Luhmann, and C. T. Russell (1990), The solar cycle dependence of the location and shape of the Venus bow shock, J. Geophys. Res., 95, 14,961 – 14,967, doi:10.1029/JA095iA09p14961. Zhang, T. L., K. Schwingenschuh, C. T. Russell, and J. G. Luhmann (1991), Asymmetries in the location of the Venus and Mars bow shock, Geophys. Res. Lett., 18, 127 – 129, doi:10.1029/90GL02723. Zhang, T. L., K. K. Khurana, C. T. Russell, M. G. Kivelson, R. Nakamura, and W. Baumjohann (2004), On the Venus bow shock compressibility, Adv. Space Res., 33, 1920 – 1923, doi:10.1016/j.asr.2003.05.038. Zhang, T. L., et al. (2006), Magnetic field investigation of the Venus plasma environment: Expected new results, Planet. Space Sci., 54, 1336 – 1343, doi:10.1016/j.pss.2006.04.018. Zhang, T. L., et al. (2007), Little or no solar wind enters Venus’ atmosphere at solar minimum, Nature, 450, 654 – 656, doi:10.1038/nature06026. Zhang, T. L., et al. (2008a), Initial Venus Express magnetic field observations of the Venus bow shock location at solar minimum, Planet. Space Sci., 56, 785 – 789, doi:10.1016/j.pss.2007.09.012. Zhang, T. L., et al. (2008b), Initial Venus Express magnetic field observations of the magnetic barrier at solar minimum, Planet. Space Sci., 56, 790 – 795, doi:10.1016/j.pss.2007.10.013. M. Balikhin and S. Pope, Department of Automatic Control and Systems Engineering, University of Sheffield, Western Bank, Sheffield S10 2TN, UK. W. Baumjohann, M. Delva, M. Volwerk and T. L. Zhang, Space Research Institute, Austrian Academy of Sciences, Schmiedlstrasse 6, A-8042 Graz, Austria. J. B. Cao and C. Wang, State Key Laboratory of Space Weather, Chinese Academy of Sciences, 52 Sanlihe Road, Beijing 100864, China. M. Gedalin, Department of Physics, Ben Gurion University, Beer-Sheva 84105, Israel. K.-H. Glassmeier, Institut fu¨r Geophysik und Extraterrestrische Physik, Technische Universita¨t Braunschweig, Pockelsstrasse 14, D-38106 Braunschweig, Germany. L. K. Jian and C. T. Russell, Institute of Geophysics and Planetary Physics, University of California, Los Angeles, CA 90095, USA. K. Kudela, Institute of Experimental Physics, Slovak Academy of Sciences, Watsonova 47, 040 01 Kosice, Slovakia.

5 of 5