Advances in Geophysics, Volume 16

ADVANCES IN

GEOPHYSICS

VOLUME 16

Contributors to This Volume JON BERGER

KTJLDIP P.CHOPRB D. DEIRMENDJIAN J. C.RIJCKLIDQE

S. M.S ~ V E B M A N D. G.W. SMITH T.F. TUAN

Advances in

GEOPHYSICS Edited by

H. E. LANDSBERG Institute for Fluid Dynamics ond Applied Mathematics University of Maryland, College Park, Maryland

J. VAN MIEGHEM Royal Belgian Meteorological Institute Uccle, Belgium

Editorial Advisory Committee BERNARD HAURWITZ ROGER REVELLE

R. STONELEY URHO A. UOTILA

VOLUME 16

I973

Academic Press

New York and London

A Subsidiary of Harcourt Brace Jovanovich, Publishers

COPYRIGHT 0 1973, BY ACADEMIC PRESS, INC. ALL RIGHTS RESERVED. NO PART OF THIS PUBLICATION MAY BE REPRODUCED OR TRANSMITTED IN ANY FORM OR BY ANY MEANS, ELECTRONIC OR MECHANICAL, INCLUDING PHOTOCOPY, RECORDING, OR A N Y INFORMATION STORAGE AND RETRIEVAL SYSTEM, WITHOUT PERMISSION IN WRITING FROM THE PUBLISHER.

ACADEMIC PRESS, INC.

111 Fifth Avenue, New York, New York 10003

United Kingdom Edition published by ACADEMIC PRESS, INC. (LONDON) LTD. 24/28 Oval Road, London N W l

L m w y OF CONGRESS CATALOO CARDNUMBER: 52-12266

PRINTED IN THE UNITED STATES OF AMERICA

CONTENTS LISTOF CONTRIBUTORS

............................................................

vii

Application of Laser Techniques to Geodesy and Geophysics

JONBERQER 1. Introduction ..................................................................... 2 Terrestrial Laser Ranging Devices .......................................... 3 . Extraterrestrial Laser Ranging Devices ................................. 4 Laser Strain Meters ............................................................ 5 . Results of Strain and Range Measurements .............................. 6. Miscellaneous Laser Applications .......................................... 7. Summary ........................................................................

. .

List of Symbols ..................................................................... References ..............................................................................

1 6 14 23 38 48 52 52 53

Electron Microprobe Analysis in the Earth Sciences

D. G. W . SMITHand J . C. RUCRLIDGE 1. 2. 3 4. 5.

Introduction ..................................................................... The Instrument and Samples ................................................ Quantitative Analysis .......................................................... Errors .............................................................................. Applications ..................................................................... List of Principal Symbols ......................................................... References ..............................................................................

.

58 60 76 103 125 142 143

Auroral Audibility

S . M. SILVERMAN and T. F. TUAN 1. 2. 3. 4. 5.

Introduction ..................................................................... Observational Results ......................................................... Characteristics and Analysis of Auroral Sound Events ............... Hypotheses of Auroral Audibility .......................................... The Case for Brush Discharge and Aurorally Induced Electric Field8 ...11......................................................................... 6. Conclusions........................................................................ List of Symbols........................................................................ Appendix: Auroral Sound Events ................................................ References .............................................................................. V

156 157 176 198 208 216 217 218 259

vi

CONTENTS

O n Volcanic and Other Particulate Turbidity Anomalies

D . DEIRMENDJIAN

..................................................................... ................................................ 3 . The Katmai Event of 1912 ................................................... 4 . The Agung Event of 1963 ................................................... 5. Climatic Effects of Volcanic Dust .......................................... 6. Summary and Conclusions ................................................... References .............................................................................. 1. Introduction

2. The Krakatoa Event of 1883

267 268 274 279 290 292 295

Atmospheric and Oceanic Flow Problems Introduced by Islands

KULDIPP. CHOPRA 1. Introduction and Summary................................................... 2 . Microscale Perturbations ...................................................... 3. Group of Small Islands as Mesometeorological Network ............ 4 . Mesoscale Atmospheric Vortices Leeward of Islands .................. 5 . Vortices Leeward of the Hawaiian Islands .............................. 6. Anomalous Oceanic Circulations Around Islands., ...................... 7 . Upwelling Due to Circulations Around Islands ........................ 8. Air Flow Over a Heated Island ............................................. 9. Recent Experiments in Tropical Island Meteorology .................. 10. Concluding Remarks ............................................................ References ..............................................................................

298 323 326 328 357 371 377 383 406 414 416

AUTHOR INDEX ........................................................................

423 432

SUBJECTINDEX ........................................................................

LIST OF CONTRIBUTORS Numbers in parentheses indicate the pages on which the authors’ contributions begin.

JON BEROER, Institute of Geophysics and Planetary Physics, University of California at Sun Uiego, La Jolla, California (1) KULDIPP. CHOPRA,Department of Physics, School of Sciences, Old Dominion University, Norfolk, Virginia and Division of Physical Science and Coastal Engineering, Virginia Institute of Marine Science, Cloucester Point, Virginia (297)

D. DEIRMENDJIAN, The Rand Corporation, Santa Monica, California (267) J . C. RUCKLIDOE, Department of Geology, University of Toronto, Toronto, Ontario, Canada (57)

S . M . SILVERMAN, Air Force Cambridge Research Laboratories, L. G. Hanscom Field, Bedford, Massachusetts (155)

D. G. W. SMITH,Department of Geology, University of Alberta, Edmonton, Alberta, Canada (57) T. F . TUAN,Physics Department, University of Cincinnati, Cincinnati, Ohio (155)

vii

This Page Intentionally Left Blank

APPLICATION OF LASER TECHNIQUES TO GEODESY AND GEOPHYSICS Jon Berger Institute of Geophysics and Planetary Physics University of California at San Diego. La Jolla. California

.

..........................................................

1 Introduction 1.1. Laser Strain Meters and Laaer Ranging Devices 1.2. Lesers 2 Terrestrial Laser Ranging Devices 2.1. Fundamental Considerations 2.2. Electro-optic Light Modulators 2.3. LRD Instrumental Description 2.4. LRD Accuraay 2.6. Two-Color Teohniques 2.6. Two-Color LRD Instrumental Description 3 Extraterrestrial Laser Ranging Devices 3.1. Lunar Laaer Ranging Experiment 3.2. Laser Ranging to Artificial Satellites 4 Laser Strain Meters 4.1. Linear Extensometers 4.2. Laaers Applied t o Linear Extensometers 4.3. The Laser's Wavelength

...................... .......................................................... . ........................................ ....................................... ..................................... ..................................... ................................................... ............................................. ........................... . ................................... ................................... ................................ . .................................................... ............................................. ............................. ........................................... 4.4. Methods of Leser Stabilization...................................... 4.6. A Michelson Interferometer ........................................ 4.6. A Fabry-Perot Interferometer ..................................... 6. Results of Strain and Range Meesurements ................................ 6.1. Seoular Strain Rates .............................................. 6.2. Earth Tides ...................................................... 6.3. Earthquakes ..................................................... 6. Miscellaneous Laser Applications ........................................ 6.1. Laser Interferometer for Absolute g Measurements .................... 6.2. The Laser Heterodyne Interferometer ............................... 7. S ~ m ............................................................. ~ y List of Symbols ....................................................... References ............................................................

Page 1 2 3 6 6 7 8 10 11 12 14 14 22 23 23 24 26 27 32 37

38 38 41 44 48 48 60 62 62

63

1. INTRODUCTION

It is now a decade since the laser became available to the scientist as a tool of investigation rather than just a subject of investigation. Its uses have been many and diversified [l] and new applications are continually being developed. The capabilities of the laser are undergoing change as the devices themselves become more and more sophisticated. One can now produce laser 1

2

JON BERUER

beams with an average power of over 60 kW and a peak power of lo0 W, while the discrete spectrum of the output covers the range from submillimeter to 1523 A with tunable lasers operating from 3410 A to 3500 A. Lasers with wavelength precision and long term stability of better than lo-" now far exceed the capabilities of the international length standard (the wavelength of a particular Kr line), and it is probable that a laser wavelength will soon become the international standard of length [2]. Further, techniques have been developed whereby one can measure both the wavelength and frequency independently and thus deduce the velocity of light [3]. In the past there was no direct way of connecting the length and time (frequency) standards. However, it is now possible to have only one standard with time and length connected by a highly accurate value of the velocity of light [4]. In the field of geophysics, lasers have been applied both to improve existing instrumentation and to develop totally new instruments. LIDAR, a sort of laser radar, has been developed to probe the structure of the atmosphere [6],and allied techniques of laser Raman spectroscopy have been used to investigate the composition of the atmosphere [6]. Both of these subjects have been dealt with elsewhere and will not be considered in this article. Two types of instrumentation will be discussed-laser strain meters and laser ranging devices-both of which are essentially results of applying laser technology to already existing instrumentation with a subsequent significant improvement of the capabilities of these devices. Since this instrumentation has only recently been developed and in some cases not yet deployed in the field, data of geophysical interest are scant. Concentration will be, therefore, mainly on descriptions of the various instrumental systems, their capabilities and limitations. However, in Section 5 some of the results that have been obtained to date will be outlined as well as what can be expected in the near future. Finally, the use of lasers in two new types of instrumentation will be mentioned: an instrument designed to measure g, the acceleration of gravity absolutely, and an instrument capable of measuring the earth's rotation rate.

1.1. h e r Strain Meters and h e r Ranging Devices The basic laser ranging device (LRD) measures the distance between two points in contrast to the laser strain meter (LSM)which measures only changes in this distance. Typically, the LRD will operate horizontally through the atmosphere for distances up to 50 km. Vertically, it has been used as an extremely sensitive altimeter from aircraft to earth and from the earth to satellites. More recently, on an impressive scale, measurements of the earthmoon distance (some 400,000 km) have been made using retroreflectors placed on the moon by Apollo astronauts and a Soviet rocket. Less sensitive

LASER TECHNIQUES

3

and sophisticated models find everyday use as a tool for engineers and surveyors, but the geophysicist is interested primarily in LRD's which can provide a basic accuracy of a part per million or better. The limitation on accuracy attainable with the LRD is set by the uncertainties involved in calculating the index of refraction of the atmospheric and better, the successful measurements to date path. For accuracies of have been made along evacuated paths, usually with an interferometric strain meter. The LSM is capable of accuracies of over ranges limited by practical considerations only (cost land, availability, etc.) to about 1 km on the earth. However, because of the need for a long vacuum system and, in general, a large instrument, the laser strain meters are at best only semiportable. Of course, in an evacuated environment such as that of the moon, the techniques would be applicable to much longer distances. Strain measurements with the LRD are made by making determinations of a line length at different times. A particular station may be rememured every 6 months or so. In contrast, the LSM makes continuous measurements of AL, the change in line length, without measuring L at all. Its accuracy, resolution, and in some cases stability far exceed the figures for the LRD. Typically, phenomena with strains of 10-lo are detectable (as opposed to a resolution of to for LRD). However, due to the physical size of the LSM, it is not as portable as the LRD and hence is used primarily to make continuous observations at one site for a long period (months or years) rather than spot measurements at a wide variety of sites. Generally speaking, then, these two classes of instruments do not compete but rather complement each other. The LRD is used for large area low resolution surveys in places where the earth motions give rise to strains on the order of per year and larger. The LSM gives, at a few select sites, an extremely accurate and continuous record of earth strain changes down to 10-lo per year. In fact, the LSM can be thought of as a wide band horizontal seismometer with continuous calibration, a response that is linear and flat down to dc, and an extremely large dynamic range.

1.2. Lasers For the instruments described in this paper, two distinct types of laser are employed. One type, used in the distance-measuring devices that employ a modulated light beam and in the interferometric strain meters, has a continuous fairly low power output and is usually a He-Ne gas laser. The other type, used for time-of-flight measurements, such as the Lunar Ranging Experiment, is a pulsed laser of extremely high power and short pulse length.

1.2.1. He-Ne h e r . The basic H e N e laser consists of a plasma tube placed within an optically resonant cavity. The plasma tube contains a

4

JON BERGER

mixture of helium and neon gasesthrough which an electrical dischargeis maintained. As shown in Fig. 1, helium atoms are excited by the discharge to the metastable 1s state at 20.61 eV. Nearly coincident in energy is the 3S2 state of Ne at 20.66 eV. Thermally caused collisions between the 1s He atoms and the ground state Ne atoms cause the latter to become excited to the 3S2 state. This action leads to a population inversion of this state over the 3p4 and 2p4states of the Ne. The system will then exhibit a net gain for radiation

I

HELIUM

NEON

i GROUND STATES

FIG.1. He-Ne leser energy levels.

a t the transition wavelengths of 3391 A and 6328 A. This gain is achieved by stimulated emission where one photon at, say, 6328 A will induce a transition between the 352 and 2p4 states which will then produce another photon at the same frequency and phase. The resonant cavity which encloses the plasma tube consists of two highly reflecting mirrors some distance apart. Its frequency response (or transfer function) is given by Born and Wolf [7]. T(v) =

1

1

+ F sin2+

where alrrliv00s e +=

and c is velocity of light ; is the refractive index of the material between the mirrors; F is a function of the mirror reflectivity and for mirrors of equal reflectivity R, F = 4R/(1 - R ) 2 ;1 is the mirror separation; 9 is the angle between the direction of propagation and the normal to the mirror surface; and v is the frequency of the light.

LASER TEOHNIQUES

5

The resonant frequency of the cavity is controlled principally by its length 1. The linewidth Avc is given by

The separation between adjacent modes is

and hence the cavity Q is (1.4)

Q, = .rrvqlF1I2/c

From a 15 cm cavity with reasonably high reflectivity mirrors (1.5)

Q, N 5 x

107.

The exact output frequency of the laser is a function of the cavity parameters and the atomic system. As will be discussed in Section 4.3,the linewidth of the atomic system is much wider than the linewidth of the cavity. Indeed, if the cavity is much longer than 15 cm, the linewidth of the atomic system is wider than the mode separation of the cavity and hence several axial modes can resonate simultaneously. This means that the output beam will not be monochromatic but rather composed of several different frequenciea. For interferometric purposes single frequency operation i s mandatory and so the cavity lengths must be short. Note that, since it is the cavity that provides the narrow bandwidth, the exact frequency of lasing is determined primarily by the distance between the two mirrors and hence is a function of the mechanical system. This point will be discussed further in Section 4.3. 1.2.2. Ruby Laser. The second type of laser is used for time-of-flight measurements such as those made in the Lunar Ranging Experiment (LURE). Here, the requirement is for extremely short bursts of very high energy radiation and so a giant pulse or &-switched ruby laser is usually employed. I n this device the lasing medium is in the solid state. A rod of ruby [sapphire (A1,03) and a small amount of chromium (Cr20,)] is pumped optically by a xenon flash tube. The ruby has two broad absorption bands which match approximately the spectral bands produced by the xenon flash tube. Hence, ground level ions are elevated to the absorption bands (see Fig. 2). These decay spontaneously to energy levels near 2 eV. Since the relaxation time from the two absorption bands to these energy levels is short compared with that from these energy levels to the ground state, population inversion occurs. A drop from these levels to ground level is accompanied by the emission of a photon of red light. Stimulated emission is accomplished in the same way as the He-Ne laser, normally by using an optical resonator to form a high Q cavity a t the lasing frequency. The discharge from the xenon flash tube lasts

6

JON BERGER

.t

m A

Absorption Band 1

’/////nL

Absorption Band 2

-

u Ground States

FIG.2. Ruby laser energy levels.

for a millisecond or so and lasing action will begin about 0.5 msec after the initiation of the discharge and continue until the discharge is complete. The output of the laser which is not switched will typically consist of spikes of random amplitude and spacing, each a fraction of a microsecond in duration. The &-switched laser has a shutter added to the optical circuit which switches the Q of the cavity such that during the flash tube discharge lasing action can occur for only an extremely short time. This allows the population of energy level 2 to build up to an optimum value before the cavity is “turned on.” The energy is dumped in one very high amplitude short duration pulse. The lasers used successfully in the LURE a t the McDonald Observatory have a pulse duration of 2 x sec and an output of 7 J. There are now commercially available lasers with pulses as short as 0.1 msec and enough power t o produce good returns from the lunar retroreflectors. 2. TERRESTRIAL LASERRANQINQDEVIOES 2.1, Fundamental Considerations

To make laser ranging devices useful t o geophysicists for ground-to-ground applications, a basic resolution of a millimeter or so is desirable. However, since the speed of light is 3 x 1O’O cm/sec, pure time-of-flight measurements would require timing to 3 psec to achieve this resolution. What is most commonly done is t o modulate the light beam a t a high frequency and measure the phase lag introduced by the atmospheric path into this modulated signal. It is preferable to use an optical carrier rather than a microwave carrier because the latter is subject to large fluctuations in the index of refraction of air due to water vapor content.

LASER TEUHNIQUES

7

In the past, light sources other than lasers have been used with some success [8] ; however, incorporating the laser into this device has greatly increased its range and usefulness. The spectral radiance (or power per unit bandwidth) available with a laser, coupled with its narrow bandwidth, make it practical to return significant light fluxes through the atmosphere from distant points with reasonable sized optics. A He-Ne laser with an output power of 100 pW has the equivalent spectral radiance of a 10 kW light bulb (if one supposes all the bulb’s energy were concentrated in a uniform band in the visible). Further, because of the narrow bandwidth of the laser, high quality interference filters can be used to reject all light outside a narrow band centered on the laser frequency and thus greatly improve the signal to noise ratio.

2.2. Electro-optic Light Modulators Common to most laser ranging devices is the electro-opticallight modulator (EOLM).Usually this device is an electro-optic crystal that introduces a phase lag into a polarized light beam which traverses it. This phase lag is proportional to the electric field applied to the device. Consider a light beam traveling in the x3 direction incident upon one of these devices. The index of refraction is in general different in the x, and x2 directions. Further, the index is proportional to the electric field applied to the device so that qzl = ql becomes ql + Aq, when the field is applied. It may be shown [9] that (2.1)

Aql

=

-ql3Ql1E1/2

and

AT, = --r)23Q21El/2

where Q,, are the linear electro-optic coefficients and El is the electric field applied in the x1 direction. A light beam incident in the x3 direction has electric field components given by El = A, exp i[wt - 2nq1x3/h] E, = A, exp i[wt - 2nq2x3/h].

After traversing a thickness of crystal L, the total phase lag will be

and

where E , is called the modulation index. If linearly polarized light is incident upon the modulator and (2.4)

E~

- c2 = (2N + l)m/4

8

JON BERBER

where

N

= 0,

f 1, J 2 ,i ...

the emergent beam will be circularly polarized. If

- €2

€1

=Nr/2

then the emergent beam will be again linearly polarized but rotated 90" with respect t o incident polarization. Hence an amplitude modulator may easily be constructed with two polarizers and an EOLM.

2.3. LRD Imtrumental Description The basic single frequency laser ranging device designed for use up to 100 km is illustrated in Fig. 3. The light from the laser (usually a He-Ne

t

1

fl

fl

I

I Modulated light I 1 Cta u

Eh

{ Beamexpander

,[ I I

Narrowband filter I

'"

Photomultiplier Phase

Analyzer

Phase shifter

I

I Amplifier

Amplifler Mixer

FIG.3. B8sio LRD blook diagr8m (afterthe design of the AOA Model 8 Oeodimeter).

laser operating at the 6328 A red line) passes through a n EOLM modulated at a microwave frequency jT and into the beam expanding telescope. Some distance d away a retroreflector sends the beam back upon itself. The returning light is collected by a telescope (which may or may not be the beam expanding telescope) and directed onto a photodetector. The signal from the photodetector is mixed with a frequency fR slightly below jT and the phase of the mixed return (fT - fR)R is compared with the phase of the transmitted signal (fT - fR)T. The distance d is measured in terms of the distance traveled

LASER TECHNIQUES

9

by the light in one modulation period. At a modulation frequency of 30 MHz, 180” of phase difference between (fT - fR)T and (fT - fR)R corresponds to a round trip distance change of (2.5)

c/2fT =

5m

where c is the velocity of light. Hence the change in d will be 2.5 m, the basic unit of distance measure. If the phase difference is measured to f4 min, the resolution will be f l mm. If the distance is not known beforehand to 2.5 m, then measurements at other modulation frequencies can be made to determine d uniquely. Figure 4 shows a Spectra Physics “Geodolite” which operates a t five different modulation frequencies around 50 MHz allowing a resolution of 1 mm with ambiguous distance determinations to 30 km.

FIG.4. Photograph of Spectra Physics Model 3G Geodolite.

10

JON BEROER

2.4. LRD Accuracy The fundamental limitation on the accuracy obtainable with the ranging devices, be they optical or microwave, is a result of the uncertainties in the index of refraction of the air along the path [lo]. Of course, what one is measuring is optical distance rather than geometric distance, the two being unequal because of nonuniformity and turbulence in the atmosphere. The index of refraction 77 depends upon pressure, temperature, and, to a lesser extent, water vapor content e. 77

= q ( P ,T ,

4

For small deviations about standard pressure and temperature conditions (15 c",1013 mb dry air) for optical wavelengths (2.6)

p = (7 - 1) = 2.76 x

(PTo/PoT)- 1.3 x

( T o / TR )

where P is pressure in millibars, Po is standard pressure (1013 mb), T is temperature in "K, T o is standard temperature (298°K)) and R is percent relative humidity, but for microwave wavelengths [1 11 (2.7)

p = 2.6 x

+

(To/T)[(P/Po) .5 (To/T)R]

For optical or microwave wavelengths

(Ap/AP)T=2880K = 3 x 10m7/mb and (2.8)

(Ap/AT)p=lolamb = 8 x 10-7/0K

however, the variation of p with water vapor content is roughly 100 greater for microwaves than it is for optical wavelengths. This is one of the principal reasons for preferring optical rather than microwave carriers for ranging devices. Because the fractional errors in distance measurement are

AL AA Ap, -= L A N

in order t o reduce the errors to 1 ppm, it is necessary to know the average pressure along the path to 3 mb and the temperature to nearly 1°C. This means that accurate temperature and pressure measurements must be made along the entire path while the ranging is in progress. I n most circumstances, this is so difficult as to make any greater accuracy impractical using this method. However, aircraft, flying the path while measurements are in progress, have been employed for this purpose.

11

LASER TECHNIQUES

2.5. Two-Color Techniques

A technique introduced by Bender and Owens [12] makes use of the dispersive properties of the atmosphere to improve the signal to noise ratio of ranging devices. For modulated light beams, the group refractive index of air rather than the phase refractive index is used. (2.10)

=c p

?f

where U is the group velocity. The quantity fractive index by

is related to the phase re-

(2.11)

If L is the geometric distance between two points, L with the extra contribution of the air being s.

s,

+ s is the optical path

L

(2.12)

s=

(qG -

I)&

Now suppose that measurements of L are made at two different optical wavelengths A, and A,. Then (2.13)

As =

loL(7)'"

- 'lG) (qlG- 1)dZ.

IllG- 1

It was shown by Erickson [13] that the quantity A = rlzG - rllG rllG- 1 is independent of atmospheric density and only moderately dependent upon water vapor content. The quantity A, the ratios of refradivities, can be approximated to a high degree of accuracy by its average value A. Then

AS = As, where s1 is the extra length at wavelength 1, and from laboratory measurements of the constant A and field measurements of As, the integral s can be calculated and the corrected geometric distance deduced. For a 15 km path using A, = 6328 A and A, = 3660 A, s is approximately 400 cm, while As will be about 40 cm. If the modulation frequency is 3 GHz and the relative phase is measured t o 1" then the accuracy of measurement of As will give ALIL = provided laboratory measurements of A plus the approximaand hence it seems that the twotion of A = A is good to better than color method offers promise of distance measurements through the atmosphere.

12

JON BEROER

2.6. T W O - C OLRD ~ T Instrumental Deecription

This ingenious device uses laser beams of two different wavelengths (6330 A from a He-Ne laser and 4417 tf from a He-Cd laser) to obtain an estimate of average group refractive index of the atmoBpheric path and hence make a correction as outlined above. The instrument has been developed at the Boulder Labs of National Oceanic and Atmospheric Administration (NOAA) over the past several years [14,16] and a similar device using different wavelengths has been constructed at North American Rockwell [161. The basic instrument design is illustrated in Figs. 6 and 6. Two laser beams are combined and polarized in a Wollaston polarizing prism before passing through a KDP (potassium diphosphate) EOLM. The phase of the beams are modulated at microwave frequencies (2.7 GHz) before they are expanded to a 20 om aperture and sent down range to a retroreflector. Upon return, the light is gathered in the beam-expanding telescope, passed through the EOLM,and analyzed in the Wollaston prism for the horizontally polarized component. If the EOLM is driven at a frequency such that the modulation adds, then there will be no horizontal component-the null condition. Null detecting techniques &re used to servo the EOLM microwave frequency so that there is always a zero component of returning light horizontally polarized. Hence the modulation wavelength is locked to the optical path length.

Extmctionof Red Sign01 from Noise Digital Servosystem

Converter

b i a b l e Frequency Mimwave Source

Fro. 6. Basic two-color LRD block diagram (after Earnshaw and Hernandez [16]).

LASER TECHNIQUES

13

FIG.6. Photo of two-color LRD.

This whole process is alternated for the two beams 500 times a second. The apparent length for the red light is (2.144

vRGL= ( N ~ h / 2) KR

and for the blue (2.14b)

vBGL= (NB42) - K ,

where L is the true length; vRG,vBGare the group refractive indices for red and blue light; N, , N, are the number of modulation wavelengths for red and blue light ; X is the modulation wavelength in vacuum; and KR, K , are corrections for optical components in the red and blue paths.

14

J O N BEROER

K R and K B are determined experimentally by measuring a straight line ABC in three segments AB, BC, and AC and attributing AC - (AB BC) to optical component corrections. The corrected length is calculated either as

+

(2.Ma)

L = qBGL- (qBGL- qRGL)/AB

or (2.15b)

L

= qRGL- (TBGL

- qRGL)/AFI

where z=

(VBG

- VRG)/(VRG

-

and = ( q B G - I)/(qBff- ?1RG).

Because of the sensitivity of A to water vapor pressure, however, measurements of temperature and relative humidity are needed. At 15°C and a relative humidity of 50 yo,temperature errors of 3°C and relative humidity errors of 5 % would lead to errors in the calculated distance of 1 x 10-I. Some results of field measurements over a 5.9 km path are shown in Fig. 7. For 10 sec averaging time, the rms in the fluctuations corrected distance are verynearly5 x 10-I.

3. EXTRATERRESTRIAL LASERRANGINGDEVICES

3.1. Lunar Laser Ranging Experiment On July 20, 1969, when the Apollo XI astronauts landed on the surface of the moon, they carried with them an array of retroreflectors designed to reflect laser beams from earth. Deploying the unit some 20 meters from their spacecraft, an astronaut aligned the array so that its optic surface was approximately perpendicular to a line from the landing site to the earth. The basic measurement to be made with this lunar benchmark is simply to observe very accurately the earth-moon distance. Measurements of the occultations of stars and optical parallax had yielded a measure of the earthmoon distance to an accuracy of f3.5 km. In the last decade, radar measurements reduced this uncertainty to f1 km. Before this experiment, groups in Russia and in the United States had succeeded in obtaining returns with laser pulses reflected from the moon’s surface without the aid of retroreflectors. Results of the Russian effort further refined the earth-moon distance f30 meters. However, little improvement on this could be expected with this method because of the irregularities of the moon’s surface, which spread out

5.9 km CI1.

24

. Blue Optical Path Length

Red Optical Path Length

T

C.25.842

p.Zb.027

r=.O76

8..Ia

1

10 sec Average

1000

1100

1200

FIQ.7. Results of 5.9 km line for two-color LRD (after Emmhaw and Hernandez [15]).

r-.SOI

I

__1 0

1400

16

JON BERGER

the return in time, and the problem of separating lunar topographic effects from errors in the lunar orbit.

3.1.1. The LUWT Retrore$ector. The retroreilector array that went to the moon was designed with several criteria in mind. First, of course, it had t o increase the return power by redirecting the light back to the earth in a more efficient manner than the moon’s surface could. Second, it had to survive the 300°C changes in the day and night temperature on the moon’s surface. Third, it must not be so efficient as to focus the return too well, since that would cause the return beam to miss the earth observatory entirely due to the observatory’s motion in the 2 t sec transit time. The effect of this velocity aberration is to displace the return beam by about 1.5 km from its origination point. Hence the return disk size must be larger than this. A study of these critera led to the choice of a 45 cm square array consisting of one hundred 4 om diameter fused quartz corner cubes (see Fig. 8 ) to be carried in Apollo 11

FIG.8. NASA photograph of luner retrorefleotor.

LASER TECHNIQUES

17

and 14 flights. The Apollo 15 retroreflector used 300 of the same size corner cubes in order to enhance the return signals. The three retroreflectors form a triangle some loo0 km on a side and permit accurate measurements of the lunar librations.

3.1.2. System EfJiciency. The experiment was designed to work with moderate sized telescopes at the earth stations but to date most data have come from the 2.7 meter telescope at McDonald Observatory [17,18]. The overall efficiency E of the whole system as originally conceived was [19]

where A is the effective area of the retroreflector array ( 4 5 0 cma),r is the range to the retroreflector ( ~ 3 . x7 1O1O om), 0 is the laserbeam divergence rad), DR is the diameter of the receiving telescope (-100 cm), DTis the diameter of the transmitting telescope ( 4 0 0 cm), D , is the diameter of the laser rod ( ~ cm), 2 T , is the transmission of the atmosphere ( ~ 8 yo), 5 T o is the transmission of the optical system ( ~ 3 yo), 0 and, A is the laser wavelength (-7 x 10- cm). For these parameters the efficiency would be 3.3 x and hence a laser pulse of l O l g photons would be needed to produce an observable pulse from a photo multiplier with a 3 % quantum efficiency using a receiver of about 1 meter diameter. This corresponds to a laser output power of 3 J. 3.1.3. The Earth Station. The optical system used at the McDonald Observatory is shown in Fig. 9 [20]. The laser, a four-stage ruby system, has a repetition rate of 20 pulses per minute with an energy of 3 J. The pulse width is 3 nsec and the bandwidth is 0.4 x cm (i.e., A h l h ~ 6x lod6). The beam diameter is 2 cm with a divergence of rad. This allows operation at the limits of atmospheric seeing, 1.5 sec, when the beam is expanded to the full 2.7 meter aperture of the telescope. The telescope is pointed by using the positions of a number of craters near the retroreflector site. A separate reticule has been drawn for each day showing the positions of these craters relative to the retroreflector). Extremely narrow band filters are placed in front of the detectors to reduce background illumination and improve the signal to noise ratio of the return beam. To achieve a basic accuracy of 1 nsec in the 24 sec round-trip transit time, the timing system shown in Fig. 10 is employed. The firing times are controlled by an on-line computer which computes from the orbital parameters an expected range figure which is accurate to nemly 1 nsec. The outgoing pulse from the laser triggers two timing systems, one which measures the time between laser firing and a clock pulse by means of a vernier (and thus

\ beam expander

1

-+-

X-Y guider /reticle holder

detector packoge --

I

FIG.9. LURE optics at McDonald Observatory (after Silverberg and Currie [ Z O ] ) . PREDICTION EPHEMERIS

STANDARD

WRITE TIME-OF-DAY CLOCK t.1 MSEC 1

RANGE DELAY GENERATOR

CONTROL PROCESSOR 1 8 K COMPUTER 1 ACCUMULATOR

FIRE LASER

U 12 BIT

CONVERTER MULTIPLEXER

VERNIER 1200 NSEC TPHC )* LASER OUTPUT DETECTOR

I

I

STOP

FINAL VERNIER 1200 NSEC TPHC )*

20 MHZ TIME START

1

LASERRETURN TIME-TO-PULSE HEIGHT CONVERTER

FIG.10. Timing schematic for LURE at McDonald Observatory (after Silverberg and Currie [ZO]).

LASER TECHNIQUES

19

gives the exact time of firing), and the other, which initiates a delay circuit already set by the computer with the expected range. Just before the return pulse is expected to arrive, a third timing system is initiated which serves as a vernier to clock accurately the pulse arrival time. Most of the 23 sec range interval is clocked by a 20 MHz counter while the laser firing time and return pulse arrival time are timed by the two verniers which subdivide the 50 nsec period of the 20 MHz counter.

3.1.4. Basic Measurements. The basic measurement made by the Lunar Ranging Experiment is simply the range p between the earth observatory E and the lunar retroreflector M (see Fig. 11). Expressed simply to a good approximation p =r

(3.2)

- X, - X ,

R

FIG.11. Basic LURE range measurement geometry.

where r is the distance from earth to lunar center, and X,, X, are the projections of the earth observatory-earth center line and lunar retrolunar center line onto the earth-moon center line. The earth observatory motions affect the range only through the term X, . Referring t o Fig. 12,

(3.3)

X, = R, COS a cos a = cos 0 sin 6 + sin 0 cos 6 cos h

where R, = distance from earth station to earth center, 0, h = co-latitude and longitude of earth station, h = X - L = the local hour .angle of the lunar mass center = local time - lunar right ascension, (T = R, sin 9 = the distance of the earth station from the axis of rotation, z = R, cos 0 = the height above the equatorial plane, 6 = the declination of the lunar mass center, so X , = z sin 6 (T cos 6 cos h. One of the simplest measurements made involves observing the changes in p as the earth rotates. Taking into account the moon's motion (approx. 3" of its orbit in the 6 hr it takes the earth to rotate 90") one can deduce u, the distance of the earth station from the axis of rotation, and the time of

+

20

JON BEROER

M

Fro. 12. Earth stetion geometry.

meridian paasage. The uncertainties in the lunar parameters (the orbital parameters, the libration parameters and the lunar retroreflector coordinates) give rise chiefly to errors in range with periods of 13 days or longer [MI. The COB h term has a 25 hr period and thus may easily be separated. Observations of the range p may be made near meridian passage and roughly 4 hr before and after. From these, CI cos 6 and the time of meridian passage can be determined. The meridian passage time t , is related to the ranges a t f 4 hours by [21] (3.4)

1

t, = - (t 2 1

Pi - Pa +

tz) = 2

~ sin 08

where t, is meridian passage time, t , , t, are timesof range measurements, p,, Pa which are made roughly &4 hr of t,, CI is the distance from station to earth’s rotation axis, R is earth’s rotation rate, and 28 is the angle of rotation between t , and ta . The accuracy with which t , can be determined is given by (3.5)

At,

=

- Pa) 2Qo sin 8

and for a A(p, - P a ) of 16 cm At, = 0.25 msec.

LASER TECHNIQUES

21

Similar measurements a t two stations would yield the difference in their geocentric longitude. When range measurements near t , have been fit to an improved lunar theory, the fluctuations in r - x, ,L, and 6 will be predictable with sufficient accuracy to investigate the variations in u and t (i.e., variations in the polar motion in the direction of the station and variations in local sidereal time). Combined observations from two stations would yield, in addition, the other component of polar motion. With the accuracy in t, quoted above, pole position could be determined t o 15 cm.

3.1.5. Atmospheric Index of Refraction Corrections. A significant development since the start of this experiment has resulted from the work of Hopfield [22] and Saastamoinen [23]. They have pointed out that since the atmosphere is usually in a state of nearly hydrostatic equilibrium (to roughly 1 part in a good predictor for the atmospheric correction to the range may be made with a measurement of the surface pressure Since (3.6) where g is the acceleration of gravity and p is the atmospheric density. Deviations from hydrostatic equilibrium are expected to introduce a range error of less than 3 mm for propagation angles up t o 70" from the zenith. However, further errors can be introduced by horizontal gradients in the atmosphere. Radiosonde measurements a t McDonald Observatory and nearby locations indicate that a reasonable estimate of the overall atmospheric correction accuracy is about 1 cm. This is considerably better than the 6 cm original estimate [21]. Fortunately, there are now available suitable lasers with pulse widths of .2 nsec and shorter and fast photomultipliers to receive the light. One cm accuracy in the range measurement corresponds to timing accuracy of& nsec which is indeed difficult. With short pulse lasers, however, the rms accuracy for a single shot is expected t o be 5 om, with the overall systematic error less than 2 cm.

3.1.6. Accuracy of Measurements. The accuracy of determinations of the dynamical earth parameters in the face of uncertainty in the lunar parameters have been calculated [24] using a fairly complete model involving 19 parameters. The uncertainties in the lunar range parameters were calculated assuming no earth polar motion in the direction toward the earth station. With a basic range accuracy of f 3 cm, the uncertainty in 17 of the 19 variables was found. (The other two variables, u and AE contain the dynamical earth information.) The lunar motion parameters which still have uncertainties on the same order of magnitude as the basic range accuracy have periods

22

JON BEROER

of 13 days or longer. Hence, it is only the lunar right ascension L and declination 8 that affect the dynamical earth parameters a t shorter periods. That is to say that during 8 hr of observation (&4 hr of meridian passage), the change in lunar distance will be accurately known. With the uncertainties calculated for L and 8, the errors in determining fluctuations in polar motion and earth rotation are not appreciably increased except for a possible 27 day term [25]. 3.2. Laser Ranging to ArtiJicial Satellites

At present there are seven artificial satellites in orbit equipped with laser retroreflectors. Initially most of these systems were designed to permit laser tracking of the satellites as a check on the results of the more conventional method. As the laser tracking systems became more sophisticated, the measurements of range began t o provide interesting geophysical information on their own. The basic information yielded by the laser tracking systems is similar to the lunar ranging information except, of course, measurements are relative to an artificial satellite rather than the moon. From a geophysical point of view then, it is not the satellite orbital parameters which are of interest but rather the discrepancies between predicted and observed values which yield information on the earth’s gravitational field [26]. The measurement of the observing station’s geocentric position can be interpreted in terms of polar motion and interstation chord distance [27]. 3.2.1. Laser Tracking Systems. The satellite laser tracking system is similar in principle with the lunar ranging system described above [28]. The light source is a pulsed ruby laser, Q switched with a 1 J pulse of 12 to 15 ns halfwidth and a 1 pulse per second repetition rate. The beam is expanded in a 10 power telescope so that the divergence is reduced to 1.2 mrad. The receiver is a 40 cm aperture telescope focusing the return light through an interference filter onto a photomultiplier tube. The whole device is mounted on a tracking pedestal which is computer controlled t o track the satelli under observation. The range measurements are made with a 100 MHz time interval unit which provides the time of flight measurement to an accuracy of f10 nsec or 5 1.5 meter for a single range measurement. Presently under construction are systems which will have a basic accuracy of f 2 5 cm and ultimately accuracies equal to those quoted in connection with the LURE should be attainable. 3.2.2. Geodetic Studies. An experiment carried out in 1970 to test the feasibility of measuring intersite distance by means of laser tracking showed consistent results with a repeatability of 25 cm over a 400 km path [26].

LASER TECHNIQUES

23

As a result of this experiment as well as advances in the art of laser ranging, the San Andreas Fault Experiment (SAFE)was initiated. The basic idea is to measure the accumulation of strain and the tectonic plate motions by measuring intersite distances between stations on either side of the San Andreas Fault in a manner similar to the LRD experiments (see Section 6.1),but on a much larger scale. An advantage of the increase in scale will be to allow the stations to be placed “deeper” into plates and hence obtain a different kind of average plate motion. The same range data may also be used to study the earth tides and the gravity tides as the orbit of the satellite is slightly perturbed by these tidal forces. The 1970 experiment [26]was able to detect perturbations in a satellite orbit due to the solid earth tide and from which an estimate of the Love number Ka was obtained. Preliminary analysis of data already obtained indicate that the orbital inclination of a spacecraft can be determined to 0.04 arclsec with 6 hr of data using existing tracking systems [28a].This means that the position of the pole of rotation of the earth in the meridian of the tracking station can be located to 1.2 meters in one quarter of a day. With the new improvement in laser ranging, it is estimated that a 10 cm polar position with a fine resolution of a few tenths of a day will be possible. 4. LASERSTRAINMETERS

4.1. Linear Extensometers A second major application of laser technology to the study of geophysics When one speaks of earth is the development of the laser strain meter (LSM). strain measurements one is usually referring to measurements of linear extension. Changes in the distance L between two fiducial points are compared with the length Lo of some length standard. The linear strain c is then calculated as E

=

[A(& - Lo)/L].

If it is assumed that there are no changes in the length standard Lo then (4.2)

=

AL/L

The linear strain E is simply one component of the strain tensor e. This tensor has six components in a homogeneous elastic medium. On a free surface (a surface where the normal components of stress are zero), it reduces to a three-component tensor. In Cartesian coordinates, the linear strain, measured at an angle 8 to the x axis, is related to the strain components e,, by (4.3)

E

= err cos2 B

+

COB

0 sin 8 + eyy sina 8.

24

JON BEROER

Hence, the measurement of the linear strain in three different directions will serve to determine the strain tensor. The concept of a linear strain meter is not new. I n fact, instruments were built before the turn of the century [29] but it was not until 1935 that a modern, sensitive strain meter 'was constructed by Benioff. Many instruments of his basic design are in use today. They consist basically of two piers fixed t o the ground with a quartz rod cemented into one pier extending to within a small distance of the other pier. The length standard, the quartz rod, is made nearly as long as L, the distance to be monitored. Changes in the small distance ( L - Lo) are measured with a variety of electromechanical transducers. To measure earth strain that is representative of the surrounding region, the instrument must be long enough to average out the small-scale inhomogeneities of the rocks on which it rests. Typically, strain meters have a length on the ordec loa meters. Unfortunately, this causes problems with the mechanical and thermal stability of the length standard. Quartz (fused silica), which has excellent mechanical properties, has a coefficient of thermal expansion of 6 x lO-'/'C. No suitable available material is much better. Hence, to obtain a stability of 10-lo, the temperature along the length of the strain meter must be known or be constant to 2 x 10-40C.By placing the instruments deep underground in mine shafts or tunnels, reasonably good temperature stability is obtained. However, as measurements extend to lower frequencies, the thermal stability becomes less reliable ; hence, the record becomes more noisy. Further, mine shafts and tunnels are not the most desirable sites a t which t o monitor earth strain. Usually, their construction has caused a great deal of fracturing of the rock, and, in many mines, the geology is by nature inhomogeneous.

4.2. h e r s Applied to Linear Extensometers With the development of the laser as a coherent light source, it has become possible to extend the techniques of optical interferometry to much greater distances. With conventional light sources, path lengths of only a few centimeters were possible. However, because of the laser's coherence, laser interferometers with paths of up to 1 km have been successfully operated [30]. The device, in a simple form, consists of a Michelson interferometer (although other interferometers such as a Fabry-Perot are also used) with the source, beamsplitter, and the fixed arm on one pier and the long arm mirror on another pier a distance L away. The light returning from the far end is mixed with the light from the local mirror t o produce the classic fringe pattern. The long arm path must be a constant pressure path (usually evacuated) to reduce the effects of wave front distortion so that a fringe

25

LASER TECHNIQUES

pattern is visible and to reduce the effects of refraction corrections. The intensity of the light at a point in the fringe pattern is related to the length of the long arm by (4.4)

I

=I,

cos (4TLlVh)

where h is the laser wavelength and 7 is the refractive index alongthe path L . Peaks in the intensity occur when L is an integral number of half wavelengths of the laser light. If L changes by 4 2 , a fringe will move past the field of view and hence counting the passage of fringes past a photodetector will provide cm and a measure of AL in terms of h/2. For the He-Ne laser h N 6 x hence for a 1 km interferometer the least count strain (LCS) is 3 x 10-lo. Note, however, that in this instrument the length s t a n d a d i s not nearly as long as L as in the case of the conventional strain meter.

4.3. The Laser's Wavelength Unfortunately, the wavelength of a laser is not good enough as length standard for these purposes. The laser output, while being highly monochromatic, does not have a particularly stable frequency. Or, to put it another way, the laser light looks monochromatic only on a short time scale. At any one instant the output spectrum is a very narrow line, but slowly the center of this narrow line can move so that for periods comparable to the laser's lifetime (months), Ah/h II The natural linewidth or Fourier spectrum of the atomic transition that gives rise to the lasing action is similar t o the Fourier spectrum of a damped classical harmonic oscillator with damping constant y and resonant frequency vo. That is, (4.5)

I(v)= 1 0 2277

- vo)S

9

+4

-l

where

and the linewidth Av,, = y. For a simple two-energy level system, y is just the transition probability related to the transition lifetime 7 by y = 117. The intrinsic spectral width for most gas lasers should be on the order of 1 Hz, that is, (Ah/X)N [31]. With extreme care, an instantaneous linewidth of has been reported [32]. However, the long term stability or resettability is another matter. Because the lasing action takes place in a hot excited plasma, there is considerable Doppler shift in the frequency of each elemental radiator (atom).

26

BEROER

JON

In thermal equilibrium, the gas atoms have a continuous distribution of velocities following the Maxwell-Boltzmann statistics.

where dn represents the fraction of atoms of total number n whose velocity lies between v, and vz + dv,. The Doppler broadened line shape is thus a Gaussian with a Doppler width of

[z In 2 g]

1/2

(4.7)

AvD = 2v0

= 7.17

x lo-'

yo

(T/N')1'2

where M' is the atomic weight. For the 6328 A line of Ne (M' AvD/v,-,

= 2.8

= 20) :

x lo-'

so the equivalent Q of the atomic transition is &A

= 3.6

x lo6.

The output frequency of the laser is a function of the atomic transition line shape and the transmission characteristics of the Fabry-Perot resonator which forms the laser cavity [32a].

t

1 = (QA

+

+

2)

where QA and Q, are the Q factors of the atomic transition and the cavity resonances and vA and v, are their resonant frequencies. For a laser with a I6 cm cavity with a finesse of 100, Q, = 5 x 10'. Contrasting with this is the Doppler broadened laser QA = 3.6 x lo6.Since QA < Q,, the expression for the laser output frequency reduces to v N v, , the resonant frequency of the cavity. Hence, the atomic transition only determines the laser output frequency to within the transition linewidth, a part in lo6, whereas the exact output frequency is determined by the laser's resonant cavity. The stability of the laser's output frequency is dependent upon changes in the optical length of the cavity. (4.9)

where r) is the index of refraction along the path. The first term All1 is controlled by the mechanical stability of the resonator structure which, with care, can be made quite good, and by its thermal expansion, which necessitates temperature control. Av/q depends upon the

LASER TECHNIQUES

27

plasma density in the lasing medium as well as density change in the residual air path of the laser cavity. The plasma density is dependent upon the degree of ionization of the gas and upon the ratio of partial pressures of He to Ne in the plasma tube. The degree of ionization can be controlled by regulation of the current passing through the tube, but control of the gas ratios is difficult since the diffusion rates of He and Ne through the tube walls are quite different. Stability of the density of the residual air path in the laser requires both thermal and barometric control.

4.4. Methods of h e r Stabilization To circumvent problems mentioned above, various forms of stabilization have been employed. These fall into two general categories: one locks the laser frequency to atomic lines which are particularly sharp or have particularly sharp features ; the other utilizes passive optically tesonant cavities. Before discussing these, however, a word on the measurement of frequency stability is in order. The most widely employed method of determining laser frequencystability has been to make a relative measurement between two similar but independently stabilized lasers by observingthe beat frequency spectrum. The results obtained this way are, however, ambiguous since there may be systematic perturbations which affect both lasers equally and hence do not appear in the beat spectrum. Some of the ambiguity has been resolved by measurements made with reference to a third laser and with reference to the 86Krwavelength standard 1331,although this standard has a short term stability that with [34]. The stability of a He-Ne laser for short extreme care is only 2 x term (short compared to a laser lifetime) has been quoted as f 5 x [35]. However, comparisons of two different lasers of the same manufacturer exhibited an offset of due to different operating parameters. As mentioned in the Introduction, the development of a wide range of lasers, masers, and microwave oscillators, coupled with a parallel development of wideband detectors, has made it possible to compare the optical frequency of a laser with the frequency of a cesium-beam clock. This source, which serves as the international time standard, has a frequency of 9 x loe Hz with long term stability of 2 x [36]. The comparison is made by beating the laser frequency down to the cesium-beamfrequency with a series of intermediate oscillators. This process will allow the establishment of one standard for both time (frequency) and length, connected by an extremely accurate value of the velocity of light [3]. 4.4.1. Lamb Dip Stabilization. There are a number of ways of stabilizing a laser with reference to an atomic standard [37]. The method of stabilization

28

JON BERGER

to an atomic line most widely used in commercial lasers is Lamb dip stabilization. The spectrum of the laser output has a characteristic dip (Lamb dip) at the center of the Doppler broadened atomic transition line. The laser is frequency modulated slightly by moving one of the cavity.mirrors, and the output intensity is observed. Phase-lock techniques are used to lock the average output frequency to vo , the center line, by detecting the position of the local minimum. This method has the disadvantage of requiring a frequency-modulated output. Typically, the frequency deviations are about 5 mHz and the rate of oscillation of frequency about 5 kHz, resulting in a maximum rate of frequency change of 2.5 x loe Hz/sec. For interferometric purposes, this means that the light returning from the remote retroreflector is not a t the same frequency as the light with which it must interfere (for 800 meters the difference is 126 kHz), Hence, the interference pattern will result primarily from beating of the two frequencies and the fringe pattern will oscillate rapidly. 4.4.2. Dispersion Stabilization. A second method suggested by Bennett et al. [38] is called dispersion stabilization. Since the output frequency of the laser is a function of the cavity resonant frequency, v c , and the atomic

transition line center frequency, it may be shown that if the gain of a laser is varied, the lasing frequency will be pulled by different amounts toward the cavity resonant frequency. (4.10)

AV = -

where H is twice the natural linewidth, v is the lasing frequency, vA is the atomic line center frequency, Avo is the Doppler broadened linewidth, and AG is the change in gain. The magnitude of the effect depends upon how far the lasing frequency is from the line center; the sign changes as lasing frequency crosses the line center frequency. Hence, a discriminant suitable for frequency stabilization is obtained by heterodyning the controlled laser with another local laser oscillator and observing changes in beat frequency as the gain is modulated. Quite good results have been reported using this method [38] with stabilities better than a part in loll over a period of 8 hr. 4.4.3. Zeeman Stabilization. It is possible to obtain a frequency discriminator using an external absorption cell. The Zeeman absorption cell splits the atomic transition by the application of an axial magnetic field. If circularly polarized light is passed through the cell, only right-hand polariza-

LASER TECHNIQUES

29

tion interacts with lower frequency (Am = -1) transition while only lefthand polarization interacts with the high frequency (Am = +1) transition. The difference in absorption for each polarization at frequency v is proportional to the difference between v and v, the line center frequency. If the magnetic field is held constant and the sense of circular polarization of the laser beam switched, a frequency discriminator is obtained. Signal-to-noise considerations for reasonable parameters show that this method has a limitation of detectability (i.e., one for which the signal-to-noise ratio =1) of Avlv = 5 x 10-l1. I n practice Avlv = has been attained [39]. Another quite similar method [40] applies an axial magnetic field directly to the plasma tube of a single mode laser. This induces circular birefringence and the mode degeneracy is removed. The laser then lases a t two different frequencies simultaneously-one frequency with a left-handed polarization and the other with right-handed polarization. These two are of equal amplitude only when the mode splitting is symmetrical about line center ; hence, by observing the amplitude of the two components, a frequency discriminator is obtained.

4.4.4. Saturated Absorption Stabilization. One of the most promising methods of frequency stabilization developed to date is the recent work of Hall and Barger [41]. In this method, an absorption cell filled with methane is placed inside the laser cavity in line with the He-Ne plasma tube. The He-Ne tube can provide gain over a wide band of frequencies, as discussed earlier, and the cavity can be constructed so that only one axial mode operates (i.e., a single frequency laser). The resulting standing wave pattern in the cavity may be thought of as the sum of two oppositely traveling waves. I n the methane cell, an elemental absorber (molecule) can interact with only one of these two traveling waves since the absorber’s velocity will have the effect of making the two traveling waves appear to have different frequencies. Only those molecules which have identical Doppler shift for both traveling waves can react with both waves. These are only the absorbers with zero Doppler shift, that is, stationary absorbers. If the laser frequency is tuned to the center of the resonance frequency of the methane, which is simply the frequency of resonance of those particles a t rest, both traveling waves react with the molecules. The result is a depletion of absorbers a t this particular frequency which leads to an increase in the lasers output. The frequency interval over which both waves can interact with the absorbers is simply the natural linewidth of the atomic resonance. Q’s in the excess of loe are obtainable this way. More important, however, is the anomalously small pressure shift of the line center, less than 10-l1/mTorr, negligible Stark shift, and small interaction with magnetic fields (the earth’s field causes a 500 Hz splitting of the line center). The stability of this system was measured by

30

JON BEROER

NCERTAINTY BARS ARElu LIMITS

.I

msec

I

I

I

10msec 100msec lsec 10wc SAMPLE TIME INTERVAL (11

I

1OOsec 100Osec

FIG.13. Results of methane stabilization scheme. The Allen variance is computed by measuring the average fluctuations of the beet frequency between two independently stabilized lasers as a function of averaging time. The variance is the ratio of these quantities to the output frequency ( Y lOI4 Hz) (after Levine and Hal1 [42]).

beating two independently stabilized lasers together producing the results shown in Fig. 13. Over the time period shown, the stability was impressively better than lo-',. A highly successful LSM, using this methane line as its length standard [42] will be described in Section 4.6. The methane system introduces some difficulty in long path length interferometry because its output is in the infrared. However, quite recently, Hanes and Baird [43] and Baird [44], developed a similar system using a hyperfine component of I, absorption line at A = 6330 A. They quote an accuracy for this system of 2 x 4.4.5 Stabilization with Passive Optical Resonators. The second class of stabilizers utilizes optically resonant cavities. High Q resonant cavities have been used extensively in microwave applications since their introduction by Pound in 1946. Indeed, the use of optically resonant cavities of high Q were essential to the development of lasers. Basically, this method transfers the stabilization problem from the laser cavity to an external cavity; there are advantages, however, in doing this. Environmental control is simplified by not having the plasma tube present. Modulation of the cavities' resonant frequency is possible without frequency modulating the laser output. The standard Fabry-Perot resonator is usually the cavity employed. In practice, a Q of lo8 is readily attainable. This corresponds to a resonance linewidth of 6 MHz and so, by locating the center of this peak to a part in lo4,a precision of 10-la is obtained.

31

LASER TECHNIQUES

-3+

rcrL Incident light

Optical defector

Reflecting flotr

FIG.14. Febry-Perot optiael resonator.

The Fabry-Perot cavity consists of two mirrors aligned parallel to each other a t some distance apart. (See Fig. 14.) The transfer function T ( v )(i.e., the ratio of transmitted to incident light intensity), given in Section 1.2.1 is

+ P sina 4)-l

T ( v )= (1

(4.11)

where

4 = (2777+v/c) 00s 8.

Peaks in the transmittance occur a t 40

=

where m is an integer, or vg

=

mc

271 cos 8

.

Then, in the vicinity of the reasonence peek (4.12)

To(v)= 1 - P

(--)

m7r

2

(v - V d 2 *

Thus, measuring the intensity of the transmitted light produces a frequency discriminant whose magnitude is proportional t o (v - vo)aand whose phase determines the sign of (v - y o ) . The chief drawback of this system is that the resonant frequency v,, is linearly proportional to the cavity length 1. (4.13)

32

JON BERQER

Because the cavity is essentially a passive element, its environment can be carefully controlled, but the largest length variations will still come from thermal effects on the spacer tube. The temperature coefficient of fused silica, the best suitable material, is 5 x lO-"/OC and so

Ah/h = 6 x lO-"/"C. To obtain wavelength stability of the cavity temperature must be controlled to better than a millidegree. This is not particularly difficult to do, but one is still left with unknown creep in the quartz tube. Quartz, annealed properly and housed in a closely controlled environment probably exhibits better mechanical stability than other materials but creep will be indistinguishable from a secular wavelength change. An upper limit on this is simply the observed geophysical secular strain rate measured with a laser strain meter controlled by such a device: 1.5 x lo-' per year [as]. 4.5. A Micheleon Interferometer A two-component 800 meter LSM that utilizes the quartz cavity length standard is currently operated by the University of California, San Diego, in the San Bernardino National Forest at a position between the San Andreas and San Jacinto fault systems. A similar instrument [46,47] has been operating at Camp Elliott near San Diego for the past three years. 4.5.1. Mechanical Design. The strain meter end points are 4 meter columns of black granite of approximately 1 square meter cross section (see Fig. 15). They are set into slightly oversized holes drilled some 3 meters into the msuloted box remote retroreflector \

-interferometer

ure controlled building

-k

4-

800m

FIQ.16. Leser strain meter: mechanical design (after Berger and Lovberg [46]),

LASER TECHNIQUES

33

ground and grouted along the bottom 1 meter with the rest of the column free standing. Insulation material fills the rest of the hole. It is hoped by this means to decouple the piers from the surface layers of the ground that undergo large temperature changes and may subsequently induce thermoelastic strains. The piers are not set in bedrock but rather in semi-consolidated decomposed granite which overlies and grades into the bedrock some 10 meters below the surface. The top 50 cm or so are quite unconsolidated, and it is probable that this layer affords considerable thermal insulation, while being too weak in shear to transmit much thermally induced strain to the material below. Heavily insulated buildings are built around each pier and the instrument pier is further enclosed in a heavy box which affords extra insulation, both thermal and acoustical. Heat pumps keep the building temperature constant to f3"C while the instrument pier fluctuations are held t o f.5"C. The 800 meter optical path between the end piers is provided by an evacuated aluminium tube. The tube, flanged a t 7.3 meter sections, is bolted together with " 0 " ring seals. It is tied t o the ground only a t its midpoint and its roller supports allow it to change length freely in response to the outdoor temperature (&25 cm a t each end). Servo-driven telescopic joints a t each end keep the tube end-pier distance constant to 5 5 x cm. The tube is evacuated t o a pressure of Torr by two 1500 liter per minute mechanical pumps which run continuously. Because of the concave elevation profile along one of the lines, it was necessary to bend the path at its midpoint through a n angle of some 3".This kept the tube within 7 feet of the ground along its entire length without any earth moving. The light beam was re-steered a t the midpoint by means of a matched set of counter-rotating prisms. 4.5.2. Optical Design. The light source for the LSM is a unimode single frequency He-Ne laser with a power output of some 100 pW at a wavelength of 6.328 x cm. The cold cathode, dc-excited plasma tube is mounted in a heavy invar resonator structure designed to be stable t o thermal and acoustical disturbances. The plasma tube can be changed without moving the resonator structure and hence without upsetting the instruments' optical alignment. Dielectric mirrors are mounted on either end of the resonator structure, a flat on one end and a hemispherical mirror mounted on a piezoelectric element a t the other end. This arrangement dlows for a variation of the resonator length and hence control of the output frequency of the laser (see Fig. 16). The laser must be isolated from the rest of the optical. circuit so that only a small amount of light is reflected back into it from the interferometer. If light returned from a retroreflector is allowed to enter the laser, one has, in

34

JON BERGER

P

FIG.16. Laser strain meter: optioal design (sfter Berger and Lovberg [46]).

effect, two coupled resonant cavities ; the outboard cavity formed by the retroreflector thus has an effect upon the frequency of the laser output. While the Q of the outboard cavity is low compared with the Q of the laser, its much greater length results in very close spacing of adjacent axial modes. I n fact, there will always be a resonance peak of the external cavity (the interferometer) within the linewidth of the laser cavity. Hence, the laser output frequency willbe slightly pulled toward the interferometer cavity peak [42,48]. In extreme cases of pulling, the laser will lock tightly to the external cavity and instead of the fringe pattern moving when the interferometer arm changes length, the fringes will remain stationary and the laser frequency will change. Strain measurements based on fringe counting in such cases would be meaningless. It has been calculated [48] that in order to keep this “pulling” less than 1 part in lolo, the isolation ratio (i.e., the ratio of light emitted by the laser to the light returned to it) must be a t least lo8. The isolator is a Glan Thompson prism (a form of Nicol prism) and a quarter-wave plate. The prism linearly polarizes the light from the laser and allows rotation of the direction of polarization. The quarter-wave plate circularly polarizes the light from the prism. Upon return from the remote (or local) retroreflector, the light will still be circularly polarized, but since its direction of propagation has been changed by 180°,the sense of the circular polarization will be opposite of the outgoing beam. The return beam, after

LASER TECHNIQUES

35

passing through the quarter-wave plate will again be linearly polarized, but with a direction that is perpendicular to the polarization direction of the outgoing beam. Hence, the polarizing prism will block passage of this light back into the laser cavity. Steering of the beam is accomplished by two separate systems. For coarse adjustment, the laser, isolator, and beam expanding assembly are mounted on a movable stage with two degrees of freedom-a rotation about a vertical and horizontal axis. This serves to align the beam initially so that light is returned from the remote retroreflector to the interferometer base. The second system, the fine control, is a gimbal mirror mount, fitted with differential micrometers, that steers the beam between the laser and the beam expander. This latter system is motor-driven so that the beam steering may be remotely controlled. After reflection from the fine steering mirror, the light enters the beam expanding assembly, which consists of an expanding lens, spatial filter, and collimating lens. It expands the beam from approximately 5 to 25 mm to reduce the effects of diffraction spreading. The spatial filter is a 50 p pinhole whose main function, as the source is a single mode laser, is to filter out light that has been scattered by small particles on the optics up to this point in the circuit. It is large enough to accommodate small transverse and vertical shifts of the focal spot produced by the fine steering mirror. The beam is then divided by the beam splitter and directed into the local retroreflector and down the tube to the remote retroreflector. These retroreflectors are cats-eye type employing a converging lens and a plane mirror which is mounted on a micrometer screw to allow focusing of the return beam. The outgoing beam is focused by the collimating lens of the beam expanding telescope so that it converges at the remote retroreflector. Here, a spatial filter accepts only the central spot of the diffraction pattern and the retroreflector is focused to diverge the return beam slightly and so reduce the effects of any beam steering drift. At the input to the local retroreflector, ti phase shifter is inserted in the optical path. This element retards the phase of one half of the beam by 90' with respect to the other half to allow determination of the direction of fringe motion. The phase shifter consists of an optical flat having two opposing quadrants coated with a dielectric that retards the beam phase by 45', while the other two quadrants are clear. The retroreflector rotates the beam by 180' upon reflection so that two quadrants of the beam pass through the coated sections twice and hence are phase shifted by 90" with respect to the other two. After recombination of the locally reflected beam with the remotely reflected beam, an arrangement of mirrors separates the four quadrants. The

36

JON BEROER

two phase-shifted quadrants are directed into one photomultiplier and the other two into a second photomultiplier. Spike filters, passing 6328 A light are placed over the entrance to the photomultipliers to reject stray light. 4.5.3. Electronic Readout. The 90" phase shift between the two photomultiplier signals is utilized in & digital circuit to tell whether the long arm is increasing or decreasing in length. The fringe pattern directed onto the photomultipliers produce signals

(4.14~3)

v, = I0 + I, 00s 47rLelA

and (4.14b)

v, = I, + I, 00s [(47rLa/h)+ $1.

When the optics are correctly aligned, $ signals are (4.16a)

=~

/ and 2 the ac components of the

v, = I, C08(47rLe/h)

and (4.16b)

V, = I, sin(47rLe/h).

Then (4.16)

V,2

+

Try2

= 112

and (4.17)

tan 6 = V,/V, = tan(47rLe/h).

If these two signals are applied to the x and y inputs of an oscilloscope, the resulting pattern will be a spot moving around the circumference of a circle of radius I,,with an angular position 0 which is related to the instantaneous value of the strain E by (4.18) Each revolution of the spot corresponds to the passage of one fringe, or a strain ohange of E = A/2L = 4 x 10-10. The digital electronics produces one oount for every 90" of revolution of the spot. Thus, there are four counts per fringe and the least count strain

LCS = 4 (h/2L)=

for L = 800 meters.

LASER TECHNIQUES

37

I n principle, the sensitivity of this instrument may be greatly increased by resolving the angle of the spot in the Lissajous pattern to higher precision. Fast digitization of the photomultiplier outputs and processing in a microcomputer is quite feasible. In this method, the strain is calculated as (4.19)

E =

h tan-I(?) 4aL

as before, but the digitization grid is much finer and hence the resolution is increased. An alternate system which dynamically tracks the fringe motion with a galvanometer has been used in the LSM of the University of Washington [30]. The fringe pattern is focused onto a galvanometer mirror which reflects it onto the apex of a right angle prism. Two photodetectors observe the split fringe pattern and a servo-system controls the galvanometer position t o lock the center of the fringe pattern onto the prism apex. The output is the analog signal which is fed back to the galvanometer. The system can be made to respond to fringe motions up to 300 Hz with a dynamic range of one fringe. Ultimately, of course, uncertainties in laser frequency will vitiate increased sensitivity by producing more noise than signal. With the quartz etalon stabilization scheme, long term stability is probably not much better than but short term stability may be much better and hence, increasing sensitivity in the seismic band ( j > 1 cycle per hour) to 10-la or so will probably be profitable.

4.6. A Fabry-Perot Interferometer

An entirely different approach was taken by Levine and Hall [42] a t the University of Colorado (see Fig. 17). This LSM, located in a mine near Boulder, Colorado, uses the methane stabilization discussed earlier and cm). The long arm in this case is a operates in the infrared (A = 3.39 x 30 meter Fabry-Perot interferometer. One laser is locked to this cavity by changing its wavelength so that the fringe pattern remains stationary. A second laser is stabilized by the methane system and the light from the two lasers is mixed. One of the components of this mixed signal has an amplitude which is proportional to the difference of the two laser frequencies.

(4.20)

vbeat = Vlongarm

- Vmethane

The resonant frequency of the long arm, vlongarm, is proportional to the arm length L. Assuming that Avmethaneis zero, the strain (4.21)

38

JON BERGER

112 INCH OSCILLATOR COUNTER

RECORDER

AUDIO OSCILLATOR

FIG.17. Methane LSM design (after Levine end Hall [42]).

A digital system is used t o count the beat frequency for some integration period. The least count strain in this system is

LCS N 10-14 for a 1 sec integration time. The stability of the methane stabilization system, plus systematic and instrumental noise, combine to give a noise level corresponding to a strain of 4 x 10-lafor 1 seo integration times [42]. 5. RESULTSOF STRAIN AND RANGEMEASUREMENTS

5.1. Secular Strain Rates The understanding of the processes that control tectonics, earthquakes, and fault movements requires an extremely accurate knowledge of the displacement field of the earth over a very broad spectrum. The cosmologies of Dirac and Brans-Dicke predict a secular decrease in the magnitude of the gravitational constant G, which would result in a strain rate of per year for Dirac’s theory and 3 x per year to 3 x 10-l4per year for the Brans-Dicke theory. From independent geological evidence, others have

39

LASER TEUHNIQUES

arrived at approximately per year. Long term strains associated with geologic processes will, of course, vary widely from place to place. Rates of per year in Japan, per year in California, and 10-7 per year in New Jersey have been reported and presumably reflect the degree of tectonic activity in these regions. One suspects that in old, stable areas such as the Canadian Shield, the secular strain rates would be even lower. Seasonal effects resulting from temperature variations, snow loading, and ground water effects will also be present in the strain signal. Measurements of secular strain rates by laser strainmeters are scarce as a consequence of their recent development. Some two years of “piecewise continuous’’ data have been gathered from the Camp Elliott Observatory and the Pifion Flat Geophysical Observatory, both of the University of California [as]. These data show an annual strain rate of less than a few parts in lo7 per year (see Fig. 18). This is somewhat lower than the usually quoted for California, but may simply be a result of the observatories’ locations. Measurement on the LSM operated by the University of Colorado, although limited to shorter time segments, also exhibits low strain rates. In England, the observatory operated by Cambridge University has reported secular rates of 2 x per day [50]. Long period phenomena are of particular interest in the study of earthquake

’

to-’

PlkoN TIDES M:12~!i4 TO 169:23:54,l971

...

EXPANSION

MAR 16,1971 to JAN l8,(972

D(&

Raw data low pass filtered at

1/40

cycle per how

i t APR

MAY

JUN

JUL

AUG

SEP

OCT

NOV

DEC

JAN (972

FEE

MAR

FIQ.18. Secular strain record from Piiion Flat Geophysical Observatory. T h e data are recorded once every 6 min and low passed with a cutoff frequenoy of 1/40 cycle per hour to produce the secular reoord. The insert in the upper left shows a section of the raw before filtering with the filtered line running through the tides.

40

JON BEXGER

fault movements. At the accuracies obtainable with LRD, some very interesting measurements have been made of fault creep and strain release associated with earthquakes. Extensive programs of fault motion monitoring have been carried out in California by the California Department of Water Resources and the U.S. Geological Survey over the past few years using optical ranging devices. Originally, these programs used devices with mercury arc lamps for light sources, but the advent of the LRD has not only improved themeasurement accuracy but also significantly increased the productivity. (Perhaps, an order of magnitude increase in both quantities can be expected with the introduction of a field-qualified two-color LRD.) Some of the most pertinent data relating to earthquake phenomena,are measurements of the motions of active faults where earthquakes occur. Typically, a setup as pictured in Fig. 19 is FAULT

-

FIG.19. LRD set up across active fault. The open circles indicate the end points of the LRD lines and the arrows show the motion of the stations. The line of short dashes is the trace of a line drawn straight between A and C before any fault movement. The long dashed lines are the LRD lines after fault movement with a or - sign beside them to indicate their change in magnitude relative to the solid lines, the LRD lines before fault motion.

+

LASER TECHNIQUES

41

used. From measurements of the three lines, one can distinguish between slippage along the fault, homogeneous strain, and elastic slippage. A net of such lines crisscrossing the Sen Andreas fault system from San Francisco t o the Mexican border has yielded much valuable information on the tectonics of this highly active fault system. The picture that has emerged from these and other studies serves to delineate the seismically dangerous areas from those that appear t o be relatively safe. It appears that the seismically quiescent areas are not necessarily the safe areas. Along the San Andreas fault in California, a t least, the contrary seems to be true. Where fault slippage occurs regularly, there is a high degree of seismicity, but of small-magnitude events. Where the seismicity is low and little or no motion is detected along the fault, one suspects that large earthquakes will occur. The two large (>8.0 mag) California earthquakes which occurred on the San Andreas fault (namely, the 1857 Fort Tehon earthquake and the disastrous 1906 San Francisco quake) both occurred in such regions. Current thinking on the San Andreas fault tectonics is to a great extent a result of the geodimeter surveys carried out in the state over the past decade. The basic ideas that have emerged are illustrated in Fig. 20 [HI. The overall tectonics of the area is controlled by the relative motions of the North American and Pacific plates which are separated by the San Andreas fault. The fault is a transform fault over most of its length accommodating the right lateral motion by fault creep amounting to some 5 cm/yr, on the average. However, in the San Bernardino area, the fault curves from its predominant NW-SE orientation to nearly East-West. The forces which elsewhere on a fault cause a slippage, here are normal to the fault-trace, locking the two sides together under compression. The rocks, which are much stronger under compression than under shear or tension, can store up large amounts of energy before failing in a large earthquake. Some evidence has emerged from the LRD work in California that there may be “strain events ’) associated with, and possibly preceding, earthquakes. The ‘ I strain events” are simply anomalous behavior of the strain rate and to observe these, if indeed they do occur, one must have continuous recordings of earth strain near active faults. 5.2. Earth Tides

The largest and most obvious form of earth motion recorded on ultra-long period instruments is a result of the earth’s response t o changes in the gravitational field owing to motions of the moon and sun. Fortunately, this ever-present signal is generated by well-understood forces. Newtonian mechanics, combined with well-determined astrophysical quantities, allow calculation of the perturbation in the gravitational field t o better than 0.1 %. Indeed, the internal structure of the earth is so well known, principally from

42

JON BERGER

FIG.20. Southern California teotonics (after Brune [51]).

seismology, that the displacements on a radially structured oceanless earth, caused by these gravitational perturbations, can be calculated to something like 1 yo [62]. This means, in effect, that over the period range of Q day t o a month, Nature has provided us with an ever present calibrating signal. A particularly useful method of examining the observations of earth tides is t o construct a tidal phasor diagram [53] of the various tidal components.

43

LASER TECHNIQUES

The time series of the theoretical earth tide at the station is calculated. The difference series is formed by subtracting the observed series point-by point from the theoretical. Identical Fourier analysis is performed on the theoretical, observed, and difference series. For each peak in the spectrum, the ratios of observed and difference amplitude to theoretical amplitude are calculated, and the phase differences between the theory and observed and difference peaks noted. The tidal phasor is then constructed by assigning to the theoretical peak a unit amplitude and zero phase and drawing the observed and difference vectors with amplitude and phase relative to the theoretical. The difference vector is the vector that must be added to the theoretical to produce the observed. It represents the sum‘bf all “noise” at that frequency. This “noise” can come from the loading of the earth’s surface by the ocean tides, from the loading by barometric pressure, and from thermal effects at tidal K

\ P ‘\

theoretical = 1 FIQ.21. Tidal phesor diagram. At Pifion Flat 172.6”.(Note: This figure is not drawn to sale.)

8 = 2.6’

and 8 = .76; hence 0

=

frequencies. Figure 21 shows an example of a tidal phasor for the M , component of the strain (2 cycles per day) at the Piiion Flat Geophysioal Observatory. The observed amplitude is .75 of the theoretical and the observed phase leads the theoretical by 2.5”. The phase of the difference vector is 172.5” with an amplitude of .25. From the offshore ocean tide model of Munk et al. [53a], the calculated phase angle for the ocean load perturbation is 170”. This angle is not dependent upon the earth model used. Assuming, from the good phase agreement, that the perturbation vector is totally the result of ocean loading, an amplitude may be calculated from the ocean tide input if an earth structure is assumed. It is easy t o see that these earth strain measurements can be used to measure an “integrated” ocean tide [53]. Fame11 [52] has computed Green’s functions for a spherical, radially layered earth which enables one to calculate the displacements from an arbitrary load. He has pointed out that it is much more useful to use the observations of earth tides to study ocean tides and other loading phenomena assuming an earth structure, rather than to assume certain loading inputs and calculate the earth structure. The point is, that from independent seismological evidence, the earth’s internal structure is known to about 1 yo,whereas the offshore ocean tides and the other loading inputs are scarcely known to an order of magnitude.

44

JON BEROER

5.3. Earthquakes 5.3.1. Normal Modes of the Earth. Continuous recordings of large teleseismic events are of great interest as they often excite the normal modes of the earth. Since 1952, when these modes were first observed from a magnitude 8.5 earthquake in Kamchatka, gravimeters (vertical accelerometers) and quartz rod strainmeters have made mode observations on the earth's largest quakes. Up to 1968, only seven events were of sufficient magnitude t o produce observable mode activity. Since then, however, improvements in instrumentation, particularly the development of the Block-Moore quartz fiber gravimeter and the laser strain meter has led to observations of normal modes with a good signal to noise ratio, from events as small as magnitude 6.3 [45]. Figure 22 shows the record obtained on a LSM from a magnitude 7 earthquake

VALPARAISO EARTHQUAKE JULY 9,1971 k L l D A NUCLEAR TEST

1

, , , ,

I

,

1

I

,i'Hrk, ,

, , ,

,

1

.

,

.

,

,

,

,

,

1

, ,

I

,

I

,

,

,

,

, , ,

,

1

,

, , ,

FIG.22. Valparaiso earthquake recorded at La Jolla. This is the wideband record from the Camp Elliott LSM.LCS refer to least count strain or the change in length of the line by 1 fringe (4 x strain).

near Valparaiso, Chili on July 9, 1971. The Fourier analysis of this record revealed clearly the line structure of the strain spectrum (Fig. 23). It is quite important for purposes of mode identification and subsequent source mechanism studies to be able to separate the toroidal modes (which are not seen on vertical instruments) and the spheroidal mode. Hence one needs to record both horizontal and vertical motion. (In this regard it might be noted that for normal mode studies the two-component LSM a t the University of California's Pifion Flat site combines with the LSM a t the Camp Elliott site 80 km away to constitute a three-component array. Thus, it is possible to specify the strain tensor completely for motions whose wavelength is long compared to 80 km.) 5.3.2. Local Earthquakes. Measurements of nearby earthquake-associated phenomena on the continuously recording LSM yield different kinds of information. Figure 24 is a record of the San Fernando earthquake of Feb. 9,

LOG FWRIER AMPLITWE

46

JON BERGER 750

600

SAN FERNANDO EARTHQUAKE FEB 9,1971

450

300

-q

I50

- -150o - 300 -450 -600

- 750 TIME (SECI

FIG.24. Ban Fernando earthquake reaord.

SAN FERNANDO EARTHQUAKE FEE 9,1971

20

-’ _II

-20

-

-30

-

LOW PASSED WITH f,

=~ x ~ O - ~ H Z

.

TIME (SEC)

FIG.26. Low-passAltered version of San Fernando earthquake reaord.

FIO.26. Secular strain before and aftar the 6en Fernando earthquake. The earth tides hmve been removed by a least squares fit of spherical harmonic components of the theoretical potential to the observed record.

Beven

48

JON BEROER

1971 (magnitude 6.3) obtained on the LSM at La Jolla, some 200 km from the epicenter. If the data are low-pass filtered, the record shown in Fig. 25 is obtained. A clear offset of 1.6 x occurs at the time of the earthquake. Note that this offset is only ,0025 of the maximum peak-to-peak strain. Whether this offset is “permanent ” or not is undetermined because of the change in the secular strain rate which occurred at the time of the earthquake *8 hr. Figure 26 shows the residual left after the earth tides have been removed in an optimal manner (in the least squares sense). The sudden change in the secular strain rate is interesting, but difficult to interpret with one strain-meter record some 200 km from the event, and may not be significant. (Other even larger changes occur that are not associated with earthquakes.) However, there is little doubt that this type of observations will eventually yield much information on tectonic processes in general and earthquakes in particular. 6.

MISCELLANEOUS LASERAPPLICATIONS

6.1. Laser Interferometerfor Absolute g Measurements An instrument has been developed which measures the absolute value of g, the acceleration of gravity, by utilizing a laser interferometer with a free-falling reflector [MI. Figure 27 shows the optical layout of the instrument. The distance the free-falling corner cube moves in a precise time interval is determined in terms of the number of fringes moving past the FALLING CORNER CUBE

REFERENCECORNERCUBE

MICROSCOPE

PHOTOMULTIPLIER

FIG.27. Sohemtio of absolute “8” apperatua (afterHemmond and Faller [64]).

49

LASER TECHNIQUES

photomultiplier. Two time intervals r1 and 7, are used in the measurement as the initial velocity of the corner cube is unknown. The value for g is then given by

where h is the wavelength of the laser light and N, and N, are the number of fringes counted in the two time intervals. These time intervals have the same initial time and r 2is usually twice 7,. To minimize the effects of ground motion the reference corner cube is mounted on a seismometer. Corrections must be made for the vertical position a t which the measurements are made, because of the gradient of the earth’s gravitational field. The measured value resulting from the above equation is used to calculate g a t the floor of the room where the apparatus sits using the measure value of y = &g/dz. Due t o the velocity of the falling corner cube, there is a correction to be made for the Doppler shift of the light. A9l9 = - [*9(Tl

(6.2)

+

d / C

+ 2Vo/Cl

where vo is the corner cube’s velocity a t the start of the measurement. This correction amounts t o some three parts in lo8. Errors that would be introduced by rotation of the falling corner cube are minimized by constructing the device t o have its center of mass accurately positioned a t the optical center. Other effects are considered such as air resistance, electrical and magnetic forces particularly from the seismometer coils, laser wavelength stability, and, of course, basic timing accuracy. Table I is a list of typical corrections and their estimated errors. TABLE I. Typical Corrections and systematic effects end their estimated errorsa

Source Laser wavelength Direction end collimation Time intervals Gravitation81 gredient Velocity of light Air resistsnce Eleotrostetic end megnetic Seismometer magnetic Net correction After Hemmond and Faller [85]. 10-3cm/seca.

* mga1=

Correction (mgab

Estimeted uncertainty (mgaUb

-0.94

f0.020

-0.006 0.00 0.432 -0.028 +0.010 0.00 0.00

&0.005

+

+0.316

fO.10

fO.O1O fO.001 f0.005 fO.O1O f0.030 f0.041

50

JON BERGER

This apparatus has been made portable and to date several measurements at various locations have been reported [55].

6.2. The h e r Heterodyne Interferometer A final device to be described, the laser heterodyne interferometer [56], can be utilized to measure rotation rates by observing the passage of “fixed” stars across the sky, However, the instrument is designed primarily to measure stellar diameters in a manner akin to Michelson’s stellar interferometer [57]. The basic idea behind his instrument was the following : If one observes the light from an extended source, such as a star, through two small apertures and lets these beams interfere, a minimum in the fringe visibility will occur when the separation d between the apertures satisfies the equation d = Ahole,

(6.3)

where 8 is the source angular dimension, ho , the mean wavelength, and A , a constant depending upon the source intensity distribution [7, p. 2741. For example, A = 0.5 for a point source and 1.22 for a uniformly illuminated circular disk. I n Michelson’s experiment, the 100-in. telescope a t Mt. Wilson was diaphragmed to produce two apertures and a symmetrical mirror arrangement used t o direct the light to the observer. Problems of mechanical stability limited the outer mirror separation to 6.1 meter so that the smallest diameter measured was about 0.02 arcsec. The laser heterodyne interferometer overcomes this resolution limitation in a manner similar to very long base line radio interferometry. As shown in Fig. 28, Lovberg’s instrument used two separate, modest telescopes with a single frequency laser connecting them. The device works much like a heterodyne radio receiver. The carrier frequency is the starlight passed through a narrowband optical filter centered at the laser frequency. This light is mixed with the laser light to produce a microwave beat. The intensity of the mixed signal a t the photodetector is given by (6.4)

I

= ELEL*

+ I lom E, etWtdwla + 2 lom ELE, cos (wo - w)tdw,

where E, and EL are the electric field strengths of the star and laser light and w and wo are the frequencies of the star and laser light. The photodetector will respond to all frequencies within its bandwidth which, with modern devices, can be easily lo9 Hz.This signal, the IF, will consist of the contributions of the second integral above. The coherence length for this bandwidth is 30 cm, and hence signals from the star, received at the two telescopes, will exhibit

61

LASER TEUHNIQUES

PHOTOMULTIPLIER

FIG.28. Heterodyne interferometer (after Lovberg [SS]).

some degree of coherence as long as the distances from the star t o the telescopes are within 30 cm of being equal (Michelson’sexperiment required the distance t o be within 5A!).When the two signals are coherent (or partially so) the microwave signals traveling in opposite directions down the coaxial line will interfere somewhere along its length. The region of interference will include a length of line equal to Az, the coherence length. A detector located in the region of the coaxial line will detect beats at an audio frequency as the fringes move by in response t o the rotation of the earth. For telescope separation of 100 meters and a star near the local zenith, the audio signal is near 11 kHz ( A = 6328 A). The zone of coherence in the coaxial line moves past the detector a t 7 mm/sec and hence “fringes” will be detected for 40 seconds at each position of the detector. Extended observations could be made by moving the detector along the line and thus tracking the star across the sky.

62

JON BEROER

7. SUMMARY

The applications of laser technology to geophysical instrumentation over the past decade have resulted in significant improvements in the instrumental capabilities. The data from geodetic surveys using LRDs have made great contributions t o the understanding of the workings of the Sen Andreas fault system in particular and to the understanding of tectonic processes in general. These data combined with the measurements now being made with LSMs will have great bearing on the problem of earthquake prediction-probably the outstanding problem facing geophysics today. The pioneering work by the LURE team offers new possibilities in very large area surveys and provides an opportunity to check the theories of continental drift and plate tectonics directly. Indeed, the variety of geophysical and geodetic instruments now available, from the ultra sensitive and highly stable LSM installations to the intercontinental experiments of the LURE apparatus allows one to make, for the first time, a truly integrated effort in unraveling the complex fabric of earth motions. The rotation sensitive devices that have been proposed offer the capability of extending our knowledge of such diverse phenomena as tidal friction and earthquake excitation. Hopefully, these ideas will be brought t o fruition in the near future. LIST OF SYMBOLS Effective area of retrorefleator array; ratio of refraativities Amplitude of electric Aeld in X1, Xp direction Diameter of receiving telescope Diameter of trensmitting telescope Diameter of h e r rod Electric field in X , . X2direction Electric field strength of laser and star light Water vapor content Microwave transmitted frequency Reference microwave frequency Function of mirror reflectivity Acceleration of gravity Twice the natural linewidth Local hour angle Light density Corrections for optical components in the red and blue paths Pressure in millibars Standard prns~ure(1013 mb) Resonance quality fector Percent relative humidity Distance from earth to lunar center Trensmission function; temperature in "K Standard temperature (298'K)

LASER TECHNIQUES

53

Transmission of the atmosphere Meridian passage time Times of range measurements Group velocity of light Photomultiplier voltages Velocity in x direction Projections of earth observatory-earth center line and lunar retro-lunar center line onto the earth-moon center line Transition probability Strain Modulation indices Index of refraction Boltzmesn constant Light wavelength and modulation wavelength

7-1

Light frequency Atomic line center frequency Atmospheric density and lunar range Radian frequency = 2 d Earth’s rotation rate Linear electro-optic ooefficients

ACKNOWLEDGMENTS This work was supported in part by the National Science Foundation under Grant NSF-GA-25700 and by the National Oceanic and Atmospheric Administration under Grant N22-17-72(G). In preparation of this review I have relied on the advice and assistance of Dr. P. L. Bender of the University of Colorado for the section on Lunar Ranging. I wish t o express my appreciation to Mrs. Elaine Blackmore who prepared the manuscript and to my co-workers, F. K. Wyatt and Dr. R. H. Lovberg.

REFERENCES 1. Emmett, J. L.(1971). Frontiers of laser development. Phyhys. Today 24, 24-34. 2. Baird, K. M. (1971). Length standards. Nat. Bur. Stand. (U.S.), Spec. Publ. 848, 39-48. 3. Evenson, K. M., Wells, J. S., and Matarrese, L. M. (1971). Defining the speed of light: A combination time, frequency, and length standard: Recent progress toward measuring the frequency of visible light. Nut. Bur. Stand. (U.S.), Spec. PubZ. 848, 67-7 1. 4. Evenson, K. M., Day, G. W., Wells, J. S., and Mullen, L. 0. (1972). Extension of absolute frequency measurements to the cw He-Ne laser at 88 THz ( 3 . 3 9 ~ )AppZ. . Phys. Lett. 20, 133-137. 6. Collis, R. T. H. (1969). Lidar. Advun. Qeophy.8. 18, 113-141. 6. Cooney, J. (1970). Laser Raman probing of the atmosphere. I n “Laser Applications in the Geosciences” (J. Gauger and F. F. Hall, eds.), pp. 51-69. Western Periodicals Co., North Hollywood, CalifOd8.

54

JON BEROER

7. Born, M., and Wolf, E. (1970). “Principles of Optics,” p. 327. Pergamon, Oxford. 8. Froome, K. D., and Bredsell, R. H. (1966). A new method for the measurement of distances up to 6000 ft by means of a modulated light beam. J. Sci. Inatrum. 48, 129-133. 9. Kaminow, I. P., and Turner, E. H. (1966). Electro-optic light modulators. Proc. IEEE 54, 1374-1390. 10. Owens, J. C. (1967).Optical refractive index of air: dependence on pressure, temperature and composition. AppZ. Opt. 6 , 61-69. 11. Kerr, D. E. (1966). “Propagation of Short Radio Waves,” MIT Radiat. Lab. Ser. No. 13. Dover, New York. 12. Bender, P. L., and Owens, J. C. (1966).Correction of optical distance measurements for the fluctuating atmospheric index of refraction. J. Ueophye. Rea. 70, 2461-2462. 13. Erickson, K. E. (1962). Investigation of the invariance of atmospheric dispersion with a long-path refrectometer. J . Opt. SOC.Amer. 52, 777-787. 14. Earnshaw, K. B., and Owens, J. C. (1967). A dual wavelength optical distance measuring instrument which corrects for the air density. IEEE J. Quant. Electron. a, 644-650. 16. Earnshew,K. B., and Hernandez, E.N. (1972).Atwo-laseropticaldistance measuring instrument that corrects for atmospheric index of refraction. AppZ. Opt. 11,749-764. 16. Fowler, R. A. (1968). Earthquake prediction from laser surveying. N AS A Spec. Publ. NASA SP-6042. 17. Alley, C. O., Chang, R. F., Currie, D. G., Poultney, 8. K., Bender, P. L., Dicke, R. H., Wilkinson, D. T.,Faller, J. E., Kaula, W. M., MaoDonald, 0.J. F., Mulholland,J. D., Plotkin, H. H., Carrion, W., and Wampler, E. J. (1970). Laser ranging retro-reflector : Continuing measurements and expected results. Science 167, 468-460. 18. Faller, J. E., Bender, P. L., Alley, C. O., Currie, D. G., Dicke, R. H., Kaula, W. M., MacDonald, G. J. F., Mulholland, J. D., Plotkin, H. H., Silverberg, E. C., and Wilkinson, D. T. (1972). Geodsey results obtainable using lunar retroreflectors. Proc. Symp. Uees Artif. Satellite8 Ueodeey, 1971 (in press). 19. Alley, C. O., Bender, P. L., Diake, R. H., Faller, J. E., Franken, P. A., Plotkin, H. H., and Wilkinson, D. T. (1966). Optical radar using a corner reflector on the moon. J. Ueophye. Rea. 70, 2267-2269. 20. Silverberg, E. C., and Currie, D. G. (1972).A description of the lunar ranging station at McDonald Observatory. Pap., 14th Meet. COSPAR, 1971 (in press). 21. Alley, C. O., and Bender, P. L. (1968). Information obtainable from laser range measurements to a luner corner reflector. I n “Continental Drift, Secular Motion of the Pole-Rotation of the Earth” (W. Markowitz and B. Guinot, eds.), pp. 86-90. Springer-Verlag,New York. 22. Hopfield, H. 8. (1970). Tropospheric effect on electromagnetically measured range : Prediction from surface weather data. Trane. Amer. Ueophye. Union 51,266 (abstr.). 23, Sesstamoinen, J. (1970). The atmospheric correction for laser ranging of satellites. Tram. Amer. Ueophye. Union 51, 266 (abstr.). 24. Bender, P. L., Dicke, R. H., Wilkinson, D. T., Alley, C. O., Currie, D. Q., Faller, J. E., Mulholland, J. D., Silverberg, E. C., Plotkin, H. H., Kaula, W. M., and MacDonald, G. J. F. (1971). The lunar laser ranging experiment. Proc. Conf. E z p . Tea% gravitation Thmriea, JPL Tech. Mem. 33-499, pp. 178-181. 25. Bender, P. L. (1972). Private communication. 26. Smith, D. E., Kolenkiewiez, R., and Dunn, P. J. (1971). “Geodetic Studies by Laser Ranging to Satellites,” Preprint x-663-71-361. Goddard Space Flight Center, Greenbelt. Meryltmd.

LASER TECHNIQUES

55

27. Smith, D.E.(1972).Private communication. 28. Johnson, T.S., Plotkin, H. H., and Spadin, P. L. (1967).A laser satellite ranging, system. Part 1. Equipment design. IEEE J . Quantum Electron. 8, 435-439. 28a. Smith, D. E. (1972).Poler motion from laser tracking of artificial satellites. Pap. 2nd Aatrodyn. Qeodyn. Meet., 1972. 29. Benioff, H. (1935). A linear strain seismograph. Bull. Sek. SOC. Amer. 26, 238-309. 30. Vali, V., and Bostrom, R. C. (1968).One thoueand meter laser interferometer. Rev. Sci. Inatrum. 89, 1304-1306. 31. Shalow, A. L., and Townes, C. H. (1958).Infrared and optical masers. Phya. Rev. 112, 1940. 32. Jaseja, T. S.,Javan, A., and Townes, C. H. (1963).Frequency stability of He-Ne masers and measurements of length. Phya. Rev. Lett. 10, 165. 32a. Shimoda, K. and Javan, A. (1965).Stabilization of the He-Ne maser on the atomic line center. J. AppZ. Phya. 16, 718-726. 33. Englehard, E. (1966).Wellenliingenstabilitilt ekes Neon-Helium lasers. 2. Angew. Phya. 20, 404-407. 34. Baird, K.M.,and Smith, D. S. (1962).Primary standard of length. J . Opt. Soc. Amer. 62, 607-614. 36. Mielenz, K.R., Nefflen, K. F., Rowley, W. R. C., Wilson, D. C., and Englehard, E. (1968).Reproducibility of Helium-Neon laser wavelength at 633 nm. J . AppZ. Opt. 7, 289-293. 36. Hellwig, H., and Halford, D. (1971).Accurate frequency measurements: Survey, signscance, and forecast. Nat. Bur. Stand. (U.S.), Spec. Publ. 343, 17-27. 37. Hall, J. L.(1968).Theleser absolute wavelength problem. IEEE J. Quantum Electron. 638-641. 38. Bennet, W. R., Jacobs, S. F., Latourette, J. T., and Rabinowitz, P. (1964).Dispersion characteristics and frequency stabilization of 8 gas 188er. Appz. Phya. Lett. 6, 66. 39. White, A. D., Gordon, E. I., and Labuda, E. F. (1964).Frequency stabilization of single mode gas lasers. Appl. Phya. Lett 6 , 97. 40. Tobias, I., Skolnick, M., Wallace, R. A., and Pohnyi, T. (1966).Derivation of 8 frequency-sensitivesignal from a gas 1-r in an axial megnetic field. Appl. Phya. Lett. 6 , 198. 41. Hall, J. L., and Berger, R. L. (1971).The implication of saturated moleculer absorption for the laser wavelength standard problem. Nut. Bur. Stand. (U.S.), Spec. PubZ. 343, 49-60. 42. Levine, J., and Hall, J. L. (1972).Design and operation of a methane absorption stabilization laser strainmeter. J . Qeophya. R w . (in press). 43. Hanes, G. R.,and Baird, K. M. (1969).I2controlled He-Ne laser at 633 nm preliminmy wavelength. Metrologia 6 , 32-33. 44. Baird, K.M. (1969).Personal communication. 45. Block, B., and Dratler, J., Jr. (1972).Improvements in the wide band quartz toreion accelerometer. J . Qeophya. Rea. 7 7 , 3678-3689. 46. Berger, J., and Lovberg, R. H. (1970).A laser earth strain meter. Rev. Sci. Inatwm. 40, 1569-1676. 47. Berger, J., and Lovberg, R. H. (1971). Earth strain measurements with a laser interferometer. Science 170, 298-303. 48. Berger, J. (1970).A laser earth strain meter. Ph.D. Dissertation, University of California, San Diego.

56

JON BERGER

49. Berger, J., Wyatt, F. K., end Lovberg, R. H. (1972). A year of strain measurements in Southern California. Nature (London) 288, 93. 50. King,G. C. P., Bilham, R. G., Gerard, J. B., Davies, D., and Sydenham, P. H. (1969). New strain meters for geophysics. Nature (London)288, 818-819. 61. Brune, J. 11971). Private ~ommunication. 52. Farrell, W. E. (1972). Deformation of the earth by surface loads. Rev. Ueophys. 10, 761-797. 63. Farrell, W. E. (1970). Gravity tides. Ph.D. Dissertation. University of California, San

Diego. 63a. Munk, W. H., Snodgrass, F. E., and Wimbush, M. (1970). Tides offshore: Transition from California coastal to deep-sea waters. Ueophys. Fluid Dyn. 1, 161-235. 64. Hammond, J. A., end Faller, J. E. (1967). Lsser-interferometer system for the determination of the aooeleration of gravity. IEEE J. Quantum Electron. 8, 597-602. 66. Hammond, J. A., end Faller, J. E.(1971). Results of absolute gravity determinations at a number of different sites. J. Ueophys. Ree. 7 6 , 7850-7854. 56. Lovberg, R. H. (1970). The optical heterodyne interferometer: A proposal for stellar diameter measurements. I n “Laser Applications in the Geosciences” (J.Gauger and F. F. Hall, eds.), pp. 249-267. Western Periodicals Co., North Hollywood, California. 67. Miohelson, A. A. (1920). On the application of interference methods to astronomical measurements. Aetrophye. J. 51, 267-262.

ELECTRON MICROPROBE ANALYSIS I N THE EARTH SCIENCES

. . .

D G W Smith Department of Geology University of Alberta. Edmonton. Alberta. Canada

and

. .

J C Rucklidge Department of Geology University of Toronto. Toronto. Ontario. Canada

.

.

Page

........................................................ ........................................... ......................................... ................................................... ................................... ........................................ .......................................... ............................................. ................................................ ........................ ........................................... .............................................................. ...................................... ................................

1. Introduction 2. The Instrument and Samples .......................................... 2.1. Basic Design Features 2.2. Electron Optical System 2.3. Light Optics 2.4. X-Ray and Electron Detection 2.5. Electron Beam Scanning 2.6. ReadoutandDisplay ........................................... 2.7. Surfaoe Contamination 2.8. Sample Preparation 3 Quantitative Analysis 3.1. X-Ray Emission and Absorption Processes 3.2. Correction Procedures 3.3. Computer Applications .......................................... 4 . Errors 4.1. Factors Affecting Precision 4.2. Accuracy and Instrumental Effects 4.3. Accuracy and Experimental Parameters ........................... 4.4. Accuracy and Matrix Effects 5 Applications ......................................................... 6.1. Applications to Qualitative Analysis and the Identification of Phases 6.2. Applications to Quantitative Analysis 8.3. Applications Utilizing Soft X-Ray Spectra .........................

58

.....................................

..............................

List ofPrincipa1 Symbols ............................................. References ..........................................................

67

..

60 60 62 63 64 71 72 73 73 76 76 93 100 103 104 110 111 118 125 125 130 139 142

143

58

D . G. W. SMITH AND J. C. RUCKLIDGE

1. INTRODUCTION

Of the thousand or so electron microprobes presently in laboratories around the world, a substantial proportion are used in connection with problems in the earth sciences. Since its early application t o mineralogy [l] in 1960, this technique has been responsible for opening up whole new realms of investigation in mineralogy and petrology. The situation has now arrived where the electron microprobe is a standard item of equipment in most earth science research institutions, and major advances in its development in recent years have been responsible for its displacing, to some extent, older analytical methods. The original design of Castaing [2] (Fig. 1, [3]) has remained basically unchanged. An electron beam is focused to a micronysized spot on a target, and the emitted X-ray spectrum analyzed in terms of intensity and wavelength to gain information regarding the target composition. An auxiliary light microscope to assist in positioning the electron beam is a necessity, and

T

Electron gun

Magnetic condenser

and illuminat

Magnetic objective

Reflecting objective

A

Specimen

FIQ.1. Schemetic design of electron probe microanalyzer. (After Castaing [3]. Reproduced by permission of Academic Press, Inc.)

ELECTRON MICROPROBE ANALYSIS

59

the whole system must be maintained under high vacuum conditions. Developments since the inception of the method have been centered on improving the designs of the electron, light, and X-ray optical systems, and on extending the range and quality of external electronic equipment such as power supplies and scanning, counting, and recording circuits. Only in very recent years have drastic changes been made with the appearance of solid state X-ray detectors and small computers. Whereas conventional spatially dispersive crystal X-ray spectrometers are restricted to monitoring a single wavelength or energy a t a given time, the solid state, electronically dispersive, Li-drifted Si detector allows a complete spectrum of X-rays to be accepted and resolved into its components simultaneously. The computer opens the way for unattended control of the instrument and rapid processing of results, and thus contributes to more efficient use of an instrument which represents considerable investment in terms of money (usually in excess of $100,000)and skilled personnel. An extension of the,technique into the realm of mass spectrometry has produced the ion microprobe [a], which will not only provide for isotope microanalysis through a wide range of the periodic table, but also has a much higher sensitivity. Although the sensitivity of the electron microprobe is excellent in terms of g the the mass that can be detected (Marton [5] considers to lower limit in the most favorable circumstances-a figure bettered in instrumental methods only by the ion microprobe a t g), the concentration sensitivity is not great. A figure of 10 ppm is quoted [5] as the lowest limit in optimal circumstances. This is not particularly good compared with other analytical techniques-X-ray fluorescence, 1 ppm and neutron activation, 1 ppb. Furthermore, detection limits vary with the design of the instrument and also with the wavelength of the radiation. Thus for F , say, the detection limit might be only lo00 ppm but that for Fe as low as 50 ppm. Table I compares sensitivities of various microanalytical methods. TABLEI. Sensitivities of microanalytical methodsa X-ray fluorescence Chemical Absorption spectroscopy Emission spectroscopy Mess spectroscopy Fluorescent microscopy Electron probe Ion probe Olfactory sense Sex attraction of bee Modified after Marton [ 5 ] .

10-7g 10-9g 10-9g 10-1=g 10-13g 10-14g 10-1L10-1Sg 10-18g 10-'*g 10-aog

60

D. 0. W. SMITH AND J.

(1.

RUOKLIDOE

An important advantage of electron microprobe analysis over most other quantitative methods is its nondestructive quality ; in general samples are unaltered by analysis, and remain available for subsequent examination. The method is unique in giving chemical analyses of minute mineral grains without the need for separation, and in demonstrating the distribution of elements within and between phases in rocks. The wide range of applications possible is discussed in Section 5. The purpose of this article is to review the impact that the electron microprobe has had on the earth sciences, and to describe in context the most recent instrumental, theoretical, and practical developments. The subjects of X-ray emission and absorption, and errors in analysis are treated at some length, as these have not received detailed attention before in an article of this type. For a more general introduction to the method and its applications, the reader is referred to the excellent review articles of Keil[6], Long [7], and to the text of Birks [8]. 2.

THE INSTRUMENT AND SAMPLES

2.1. Baeic Design Features

The principal design features of all available microprobes were recently tabulated [S]. The three essential components of any electron microprobe are (1) An electron optical system able to form a micron-sized electron beam at the sample surface. (2) An X-ray optical system to detect X rays emitted by the sample. (3) A light optical system to allow the sample to be viewed and to select points for analysis. Additional but less essential features are scanning circuitry to allow television-type display of element distribution, and secondary electron detection to give extra information on the nature of the target. The whole mechanical system must be maintained under a vacuum of Torr or better to permit the free passage of electrons and soft X rays. The design of an electron microprobe revolvbs around the three optical systems outlined above. Ideally they should all focus simultaneously on the same point on the sample surface. The h a 1 magnetio lens should be close to the sample in order to produce the smallest electron spot ; X rays should be observed from as high an angle to the sample surface as possible, to minimize absorption losses; the light microscope should allow viewing of the electron impact area in the center of the field. These three requirements impose considerable constraint on design, and the ways different manufacturers have solved the problem are shown in Fig. 2 [lo]. The major compromise is usually in the light microscope, which can only be made coaxial with the electron optics if a reflecting objective is used, having holes through the

ELECTRON BEAM

ELECTRON BEAM

I

I

LIGHTBEAM

M S 4 6 18JXA-3A 20" JXA-5 4 0 " AMR-3 15" XMA-5 PHILIPS 4 5 0 0

0 = 380 8 = 40'

(b) ELECTRON BEAM

ELEVATOR DEVICE

ELECTRON BEAM

M A C 400

A,R,L, AMX IL EMX

(C)

(d)

ELECTRON BEAM

ELECTRON BEAM

1

1

X RAVS

,TURNTABLE

45'

SLIDE FOR MOVING SPECIMEN UNDER

AEI SEM 2A

LIGHT OBJECTIVE

( e)

ELMISONDE

(f) ELECTRON BEAM XRAYS,

1

MOVABLE LIGHT OBJECTIVE AND PRISM

75'

MICROSCAN Y

(9)

FIG.2. Various mmngements of light, electron, end X-my optics1 systems in commercial microprobee. (After Brown end Thresh [lo]. Reproduced by permission of Maroe1 Dekker Inc.)

62

D. 0.W. SMITH AND J. C. RUCKLIDQE

mirrors for the electrons. Refracting objectives, as used in the MAC 400, Microscan V, and Elmisonde, demand that the microscope axis be inclined to the electron beam (thus resulting in non-normal electron incidence a t the surface), that the objective be placed in the path of the electrons, or that the sample be removed to a different position for viewing. I n both the latter systems, while excellent optical performance is possible, the sample cannot be viewed normally during electron bombardment, except in the case of thin sections fiom the back. The desirability of a high X-ray take-off angle 0 (not t o be confused with Bragg angle 0) is frustrated by the presence of the magnetic lens immediately above the sample. This is one reason that all early instruments settled for low take-off angles with X rays emerging below the lens. Under these conditions fluorescence corrections, often of importance in metallurgical applications, are minimized. More modern designs have increased the angle of lens polepieces, or observe X rays through the bore of the lens, as in the Microscan V and ARL instruments. The latter used an inverted lens design which restricted the space available for samples, and involved an elevator device in early models.

2.2. Electron Optical System A schematic diagram of a typical electron optical system is shown in Fig. 3. Electrons are emitted from the tip of a heated filament and accelerated through an aperture in the anode plate, maintained at ground potential relative to the filament, which latter is at a high negative voltage (1-50 kV). The gun grid cap, or Wehnelt cylinder, operates a t a few hundred volts negative with respect to the filament, serving to focus the electron beam as it leaves the gun. The first magnetic lens, the condenser lens, produces a n image of the filament tip, demagnified about 5 times. The purpose of this lens is not so much to reduce the diameter as to control the electron beam intensity in the lower part of the column. This is accomplished by placing an aperture (Al, Fig. 3) below the condenser lens, so that when the filament image lies in the plane of this aperture the current is maximized. Both raising and lowering this image, by varying the current through the lens, reduces the beam current, but it is desirable to exercise control with the crossover above the aperture only, in order to have a smaller image size for subsequent demagnification. Another aperture A, when placed slightly below A,, may intercept a proportion, fixed within limits, of the beam, so that a measure of the beam current may be made. This sample of beam current can be used to correct the instrument for drift, either through a feedback network to the condenser lens, or by letting it control the length of counting period when X-ray intensity data are being accumulated.

ELEUTRON MICROPROBE ANALYSIS

Y

63

Filament

I Guncap

A

’I

Apertures

A2

FIQ.3. Schematic diagram of electron microprobe eleotron optical system.

The final, objective, lens provides the most demagnification producing a spot of 1 pm diameter or less. This lens must be most carefully designed to achieve a truly circular spot, and astigmatism, chromatic, and spherical aberrations must be minimized. Often external stigmators are provided to make minor corrections and give optimum spot shape. Typically the sample current is 0.01-0.1 p A while the beam current, as measured at aperture A,, would be about an order of magnitude greater. The absolute beam current or probe current can be measured by placing a Faraday cage in the path of the beam somewhere below the aperture A,. The beam current when measured in this way is cut off from the sample, whereas the aperture A, gives continuous beam current monitoring, even during analysis. Less than 0.1 yoof the electrons emitted from the gun may reach the specimen t o produce X rays. 2.3. Light Optics

2.3.1. Microswpe. The polarizing microscope is the most useful tool available to petrologists and mineralogists. It is often a disappointment that microscopes on electron probes do not usually provide very satisfactory service in all the desired modes such as variable magnification, transmitted

64

D. 0. W. SMITH AND J. C. RUCKLIDOE

and reflected light, plane and crossed polarized viewing. Only where refracting objectives are used do quality and versatility approach that of a bench microscope. Reflecting objective systems are fixed at a high power, usually about 300x, resolution is about 1 pm, though definition and contrast are invariably disappointing. Sometimes a low power auxiliary lens provides aid in these systems, but the situation is never wholly satisfactory. Traditionally, electron probes have been designed primarily for metallurgical use, so transmitted light facilities have often been added as an afterthought. Only the Geoscan, now superseded by the Microscan V but preserving most of the former's qualities, was specifically intended for use with thin sections, and in this it has been successful. Although the electron beam is interrupted when the sample is viewed from above, a low power oblique observing system is also available for simultaneous observations. Further difficulties arise in the use of polarizing filters. The need for mirrors in the light path necessarily upsets the normal plane/crossed polarized light behavior, though the peculiarities can usually be tolerated. Attempts to use circularly polarized light have been quite successful in instruments of CAMECA design.

2.3.2. Stage Movements. Stage controls invariably include X and Y movements ;less common are rotation motions, particularly when centered about the optic axis of the instrument. Without this rotation the value of high polarizing performance is reduced, hence the petrologist quickly learns to make accommodations in using the microprobe microscope. The sample should be well studied beforehand to avoid time-wasting searches for areas of interest. It is a great help to have photographs of relevant features when actually working at the instrument. A system of sample positioning by X Y coordinates is valuable when moving between different areas of a sample, or to and h m standards. "he Microscan V with its servo-assistedstage mechanism can relocate positions accurately and quickly. Focus control of the microscope is accomplished by 2 movement of the sample. This is a very critical adjustment since optical focus is taken on a plane to which the electron and X-ray optics are referred. A small error in optical focal setting can result in large differences in X-ray count rate. 2.4. X-Ray and Electron Detection 2.4.1. Spatially Dispersive X-Ray Spectrometers. The fully focusing type of crystal spectrometer gives the highest X-ray intensity at the detector. Various possible geometries of crystal spectrometers are shown in Fig. 4, where it can be seen that complex mechanical systems must be designed to satisfy the Bragg requirement nX = 2d sin 0 and at the same time conform

65

ELECTRON MICROPROBE ANALYSIS

I

(a) Center

-

bearing, fully focusing

(b)’Linear‘ fully focusing showing successive positions of crystal and slit Path of slit on 28 a r m

Path of counter

Concentric 19 and

28 axes

X

-

X ray source ( c ) Semi

-

focusing

-

ray source

( d ) ‘Bending crystal’ radius of curvature of crystal varied continuously with 8 to maintain focusing condition ~

FIG.4. Different arrangements for o w e d crystal X-ray spectrometers. (After Long [7]. Reproduced by permission of Aaademic Press, Ino.)

to constraints imposed by the space available. The linear spectrometer, which maintains a constant angle to the sample surface, requires an intricate system of cables, gears, and springs to “roll” the Rowland circle as the wavelength is changed, and to keep the detector on the Rowland circle facing the crystal. The size of the Rowland circle controls the spectrometer resolution, the larger diameters having higher resolution but necessarily losing intensity. A radius of about 10 cm provides a good compromise between intensity, resolution, and space required, but up to 25 cm radius (Associated Electrical Industries, Ltd., Manchester England) has been used for high resolution purposes. A small radius design leaves more space for additional spectrometers, commonly three, occasionally more, to be included.

2.4.2. Spectrometer Crystals. Crystals used most frequently in microprobe spectrometers are listed in Table 11. It is often possible to exchange crystals

TABLEII. Crystals commonly used in microprobe X-ray spectrometers Nsme Pb - m e l h t e Pb-lignocerate Pb/Ba-stesrata Pb-lsureate Octadecyl hydrogen maleste

l6OA 130 100.6 70.0 63.35

40-147A 32.5-11 1 25.0-93.0 17.5-65.6 15.8-59

C-B C-B N-B 0-c F-C

28.4

7.10-26.6

Si-0

Rb-V

26.63 26.12 19.8 15.19

6.57-24.8 6.45-24.4 4.95-18.5 3.80-14.3

Si-0 Si-0 Ar-Na

Sr-V Sr-V Mo-Fe Cd-Cu

ADP EDDT PET

10.64 8.803 8.742

2.6610.0 2.20-8.23 2.18-8.16

Ti-Mg Cr-Si Cr-Si

Bs-As Pm-Kr Pm-Kr

SiO, Ge(ll1)

6.708

co-P

6.532

1.67-6.26 1.62-6.11

Ni-S

Tm-Zr Yb-Zr

LiF(200)

4.027

1.00-3.77

Br-K

Bi-Sn

PB/BA SD

OHM

Clinochlore

K scid phthalste Rb acid phthalate Mica

KAP R(b)AP

Gypsum

Ammonium dihydrogen phosphate

Ethylene diamine dextrotertrete Pentserythritol Quartz

Germanium Lithium fluoride ~

a

~

~~

hrange''(d) Karange Larsnge

Abbrevistion 2d(A)

~~~

~~~~

~

S-F

Ti-Ca

Comments

Poor resolution Pseudocrystal films

Fe-Ca COAL

Good resolution, good peak-tobackground rstio [113 High reflectivity 1st and 2nd order, fluorescence Medium-high reflectivity Medium-high reflectivity Medium-high reflectivity High reflectivity, Cs, S fluorescence, unstable in vaeu.o Medium reflectivity Medium reflectivity High reflectivity, temperature sensitive Good reflectivity, fluorescence High reflectivity 1st order, sbsent 2nd order High reflectivity, nq fluorescence

~

"Range" indicates the geometricslly sccessible region, but does not necessarily imply sstisfsctory performance st the extremes.

rrn i P

2 U

4 9 0

2E

U 4,

u

ELECTRON MICROPROBE M ~ Y S I S

67

on a single spectrometer, thus extending its useful wavelength range. However, the choice of crystals may depend on factors other than the available range of wavelengths, since X-ray reflection efficiency of various crystals varies greatly. Some crystals such as P E T are particularly thermally sensitive and most have coefficients of expansion which are large enough that a small change in temperature will substantially affect the correct 28 setting for a given wavelength (see Jenkins and de Vries [12] for the thermal characteristics of some common analyzing crystals). This is one of the reasons why all microprobes should be operated in temperature-controlled environments. Apart from such thermal effects, some analyzing crystals may be affected by the vacuum in the spectrometer. Gypsum crystals have a tendency to dehydrate in vacuo and thus are suitable for use only in spectrometers maintained at atmospheric pressure and isolated from the column by a window. It has been suggested that RbAP changes its characteristics during the initial few hours in vacuo after exposure t o the atmosphere. Ln spectrometers let down to air frequently to change crystals (asis the case for some instruments), this will be a serious disadvantage. The effect has yet to be substantiated by the published results of controlled experiments, but deserves thorough investigation in view of the growing popularity of RbAP, which otherwise performs very well in much of the soft X-ray region. The presence of certain elements in analyzing crystals may give rise t o increased background from fluorescent radiation, a case in point being P in ADP.

2.4.3. X-Ray Diffraction Gratings. The range of wavelengths accessible through crystals or pseudo-crystals extends to about 150 A, which allows elements as light as Be (Be Kcr = 113 A) to be detected. Crystal efficiency in the higher part of the range is not good, and the recent development of X-ray diffraction gratings may offer some relief. Franks [13] has reported gratings that can be used over the range from 0.05 to 200 A. The gratings consist of flat lands and flat-bottomed grooves, and for maximum diffraction efficiency the radiation is diffracted from both land and groove. The dependence of diffraction efficiency on wavelength, angles of incidence, and groove depth has been determined, and gratings can be tailored to suit a particular application. Development of blazed gratings [14,15] has permitted improvements in detection efficiency in the range 8-200 8. Uses of gratings in electron microprobe analysis include the study of spectral effects due to chemical combination. 2.4.4. Gas Detectors. Gas proportional detectors are used on spatially dispersive spectrometers, and these may be efficient over a range from 1 to 100 A [12]. In the shorter wavelength range, up to 10 A, sealed counters with Be or A1 windows and Ar or Ne gas are often used. For softer X rays the

68

D. 0. W. SMITH AND J. C. RUUKLIDQE

window material must be thin and light, for example, Mylar, polypropylene, nitrocellulose, and nitrolucid [16]. Since these windows are not completely vacuum tight, the gas (e.g., P10, a 90 yo Ar, 10 yo CH, mixture, although other mixtures may be used) must be passed continuously through the counter, hence referred to as a flow proportional counter. The detector itself consists of a gas-filled chamber with a central anode wire maintained at 2-2.5 kV positive. Incoming X-ray photons ionize gas atoms and the electrons released are accelerated toward the anode, producing further ionization on the way. Provided the anode voltage is constant, the h a 1 charge reaching the wire will be proportional to the energy of the original X-ray photons, as will the voltage pulses emerging from the counter. For a general discussion of detectors, see Jenkins and de Vries [12]. 2.4.5. Puke Height A d y s i s . The type of detector described above provides electronic dispersion of the X-ray spectrum. Although the resolution is not great, this property can be utilized with the assistance of pulse height analysis (PHA) to discriminate between different components of the X-ray spectrum which may be diffracted at the same Bragg angle. For a given Bragg angle 0 several values of X corresponding to different values of n will satisfy the equation nh = 2d sin 8. The use of PHA will effectively remove such unwanted signals, since the interfering radiation must differ by at least one order of diffraction. Errors which can arise from such interferences are discussed in Section 4.3.1, together with techniques which can be used to avoid them. PHA is widely used to improve peak-to-background ratios by removing pulses from higher orders of the continuum stray radiation and electronic noise.

2.4.6. Solid State Detectors. Solid state detectors have enjoyed a spectacular rise in popularity since their energy resolution fell below 200 eV (measured at 5.9 keV). The best presently available have resolutions of about 160 eV [17], which is approaching the theoretical limit. The Ka radiation of all elements above 2 = 11 (Na) are separated by at least this energy, so most elements can be resolved by electronic dispersion. Figure 5 [18] compares the performance of electronically and spatially dispersive detection systems, where it can be seen that Li-drifted Si detectors have achieved resolution sufficient to distinguish between adjacent elements. In the light element range, they can never give as good resolution as the crystal diffraction systems, but in the heavy element range the performance is greatly superior. Sensitivityin the lighter elements does not compete with the crystal spectrometers because background tends to be much higher. To minimize this effect the detector and part of the preamplifier are maintained at liquid nitrogen temperature, but even so the peak-to-background ratio for Ti K a in pure Ti, for example, in a crystal spectrometer would be about IOOO: 1, but in a

69

ELECTRON MICROPROBE ANALYSIS

I

40 X-RAY

I00

ENERGY ( K e V )

FIQ.6. Comparison of resolutions of various x-ray detectors 8s a function of energy. Dashed curve indicates resolution required to distinguish between adjacent elements (dE/E = FWHM). Modified after Clayton [lS].

Li-drifted Si detector only about 70 : 1. The solid state detector can be placed close to the sample to subtend a large solid angle (although it must be appreciated that the take-off angle may range through as much as 10") thus giving a much greater efficiency. When used with a multichannel pulse height analyzer, it has a great advantage in that the electronic dispersion allows the whole X-ray spectrum to be recorded in a matter of minutes. Such a spectrum is illustrated in Fig. 6. The comparable time for a spatially dispersive system is at least 1 hr. Accurate quantitative analysis of geological materials using solid state devices is just beginning, and we may expect the design of electron probes to alter radically, probably in the direction of miniaturization to take advantage of this new technology.

70

D. 10000

I

a. W. I

SMITH AND J. C. RUCKLIDOE I

I

I

I

I

9000

8000

7000

-

o)

6000

C

c m s 0

k

n

6000

u)

w C 3

6

4000

Al 17.6%) 3000

2000

1000

0 ChannelNo 30 Energy keV 0 6

40 08

60 1.0

60 12

70 14

80 16

I

I

90 18

100

0

20

2 2

FIG.6. Electronically dispersed partial X-ray spectrum from kaersutite. Li-drifted Si detector, 163 eV resolution. 16 kV electrons, 1000 counts per second total counting rate, 200 sec counting time.

2.4.7. Back-Scattered Electrons. Electrons which are scattered back from the impact area are referred to as back-scattered electrons (BSE) if their energy exceeds 60 eV, and others are (arbitrarily) defined as secondary electrons (SE) [19]. Secondary electrons are used in scanning electron microscopy (SEM), a subject not of direct interest here, except that some probes have secondary electron detection facilities. This compromise fulfillment of a dual role usually results in less than optimal performance of the SEM, because the spot is not small enough, but quantitative analyses are being performed on SEMs which have X-ray detection facilities, albeit less efficiently than in microprobes designed specifically for the purpose. A com-

ELECTRON MICROPROBE ANALYSIS

71

bination transmission electron microscope and microprobe was described by Duncumb [20]; this design has been manufactured, in modification, as EMMA by AEI Ltd. The number of back-scattered electrons depends on the mean atomic number of the target, and so they may be used to display the variation in this parameter across a sample. Scintillation counters are used to detect electrons. This detector consists of a photomultiplier tube which amplifies light scintillations caused by electrons impinging on a T1-activated NaI crystal. The final amplifier output is dependent on mean atomic number and this can be very useful when seeking mineral grains which have high contrast with the matrix, for example, Pt metal minerals, Au or P b in sulfides, fluorite in silicates. 2.5. Electron Beam Scanning

The BSE signal, which is approximately complementary t o the current transmitted through the sample, is best displayed by television type scanning. The primary electron beam is moved by electrostatic deflection plates, or preferably magnetic deflection coils, contained in the bore of the objective lens. The spot on an oscilloscope is synchronized with the movement of the primary beam, and the intensity can be controlled by the BSE signal, or alternatively an X-ray detector signal. I n this way, a picture of the sample area in terms of mean atomic number (BSE) or element distribution (characteristic X ray) is built up, as illustrated in Fig. 7. The magnification is varied

FIQ.7. (A) Back-scattered electron and (B)C1 K a X-ray scanning images of a partially serpentinized dunite. The chlorine is seen to be concentrated in the serpentine veins between olivine grains. Bulk concentration of C1= 0.12 yo.

72

D. 0 . W.SMITH AND J. U. RUCKLIDOE

by altering the size of the area scanned, but pictures taken at very low magnification may give distorted information because of defocusing of the X-ray spectrometer. When the X-ray aource moves perpendicular to the plane of the spectrometer, the defocusing effect is small because there is little change in 29, but when the source moves parallel to the plane of the spectrometer the defocusing effect is large, since the change in 29 is relatively large. This effect will give rise to low X-ray intensities on two sides of the display, but exactly which sides depends on the relationship between the plane of the spectrometer and the direction of scan. Attempts to overcome this effect have involved mechanical movement of the sample in one scan direction, or, alternatively, widening of the detector slit which is then accompanied by increased background and reduced resolution. This defocusing effect is not observed with electron images or with X-ray images from solid state detectors. 2.6. Readout and Digplay Much of the electronic equipment associated with an electron probe is for “readout,” that is, presenting the X-ray intensity data in the most useful form for the purpose in hand. In recent years the introduction of nuclear instrumentation modules (NIM) has made an impact in this area, and many instruments now utilize these compact transistorized units which plug into standard bins containing power supplies. Scalers, timers, rate-meters, and pulse height analyzers are among the more common units available, although these are a small part of the wide range of electronic gear which several manufacturers offer in this line. For quantitative analysis pulses from the X-ray detectors are accumulated digitally in a scaler controlled by, or controlling, a timer. This is the best way of making a single point analysis. For studying element distributions the X-ray signal is converted to analog form through a ratemeter, the dc voltage output of this device controlling a pen recorder. The sample is driven beneath the electron beam at a steady speed which is synchronized with the recorder chart movement and an element profile obtained. Where more than one element is being studied on a multispectrometer instrument it may be desirable to use a multipen recorder. For determining the elements present in an unknown, a wavelength profile of a point may be obtained by scanning a spectrometer through its range in synchronization with a pen recorder. X-Y recorders are more convenient than conventional strip chart units, as the X movement may be controlled by sample or spectrometer position, and the Y by the X-ray signal. Profiles can easily be fitted to the size of the paper and can be re-run with a small Y displacement for comparative purposes. Records are then on easily filed sheets rather than endless strips of barely used chart paper.

ELECTRON MICROPROBE ANALYSIS

73

The oscilloscope display of element distribution over scanned areas is easily photographed and is an excellent means of displaying qualitative information for presentation in reports. It is unfortunate that many poor quality photographs are considered acceptable, since with a little careful thought about the interaction of X-ray count rate, camera f-stop, oscilloscope brightness, exposure time, and PHA and amplifier setting, photographic quality can be reliably improved. Devices for pulse rate discrimination are now available [21] which suppress pulses arriving below a certain threshold rate, and this can have the effect of greatly enhancing peak-to-background contrast in photographs where low element concentrations are involved. The scanning display serves a further useful purpose as an aid t o positioning an electron spot not visible in the microscope.

2.7. Surface Contamination Contamination of the sample surface is encountered after prolonged exposure to the electron beam. It is a deposit principally of carbon arising from the breakdown on hydrocarbon molecules from vacuum pump oils. It can be useful in indicating the exact location of the impact area on samples which do not luminesce, and it is often invaluable for this purpose on instruments which do not have simultaneous viewing and analyzing facilities. The problems associated with the building up of contamination together with techniques used to minimize or eliminate it are discussed in some detail in Section 4.1.3.

2.8. Sample Preparation 2.8.1. Polishing. For study in the electron probe, materials must have a highly polished relief-free surface if accurate quantitative data are sought. The importance of these requirements are discussed in Section 4.3.4. Sample sizes vary with the instrument, but standard circular mounts or 2 x 1 in, petrographic slides are usually acceptable. The purpose of using polished thin sections as opposed t o polished block mounts is to assist in microscopic examination of transparent phases, and it is becoming standard technique to prepare most samples in this form. Many methods for the preparation of geological samples for microprobe analysis have been described in the literature (e.g. [7,22-251) and doubtless many different but equally successful approaches remain unpublished. The following method used in the authors’ laboratories has been found to be entirely satisfactory and perhaps to have certain advantages. All materials are prepared initially as polished 1 in. diameter blocks by the following process, illustrated in Fig. 8. Rock fragments, if not friable or

74

D. 0. W. SMITH AND J. 0.RUCKLIDQE

--(d)

(el

(f)

FIG.8. Steps in the preparation of polished thin sections. For explanation see text.

porous, can be embedded directly in epoxy resin (e.g., Araldite) or bakelite, and cast in a mold (Fig. 8a). Porous or friable material must be impregnated in resin [26,27], preferably in a vacuum chamber, prior to embedding. Polishing technique is the standard process through different grades of diamond paste to end up on pm diamond on lead or cloth laps. This produces a final finish on the surface (Fig. 8b), and if thin sections are not required the mount is ready. For thin sections the polished surface is glued with Lakeside 70 cement to a glass slide (Pig. 80). The back is then sawed off and ground down as for a conventional thin section (Fig. 8d). It is quite permissible to grind it to the standard thickness for petrographic sections, as judged by optical interference colors. The section is now cemented with epoxy resin (Araldite) to the final glass slide (Fig. 8e), and when set, the Lakeside 70 cement may be melted on a hot plate and the temporary glass slide removed (Fig. 8f). The polished surface is easily cleaned with alcohol. The advantage of this method lies in doing the polishing in the early stages. It is not necessary to decide a t the outset if a thin section is needed, as a polished thin section can be made from any polished mount. Further, there is no difficulty in bringing the thin section to the standard thickness, as the last stage is grinding rather than polishing which, on a thin section, tends t o dome or pluck away individual grains. Standards are made in the same way, and it is easy to prepare several materials in a single mount by placing individual grains in holes in a blank epoxy mount. These can be covered with more resin, and then polished. Alternatively, small cylindrical standard mounts can be prepared and any desired combination inserted into a suitable holder.

2.8.2. Coating. Since most minerals are nonconductors they must be coated with a material which will conduct away electrons entering the sample from the beam. If it is a metal this layer may also serve to conduct away heat. It must be thin enough to avoid substantial X-ray absorption and also to have insignificant effect on the energy of the electrons passing into the sample; yet it must also be sufficiently conducting to provide a low resistance

ELECTRON MICROPROBE ANALYSIS

75

path to ground. Various substances have been used for this purpose including Be, C, Al, Cu, and Au. Carbon is most widely used for geological samples because of its low absorption coefficients for lines used in routine analysis, its relatively minor effect on optical properties in transmitted and reflected light, and because it is seldom sought during analysis. Sweatman and Long [28] have discussed the effects of differing thicknesses of carbon on observed X-ray intensities, and it is clear that variations of, say, 100 A between specimen and standard can produce significant errors (see Section 4.3.5). Problems arise from the difficulties of producing identical evaporation conditions during coating. Apparently, carbon deposited under somewhat different conditions varies in density, electrical conductivity, opacity, and color, etc., so that even if the same thickness of film is deposited on two samples there is no guarantee that the physical characteristics will be identical. Thus the use of thickness gauges based on quartz oscillators (which are expensive) does not really solve the problem. This may be’circumvented to a large extent by coating specimen and standard at the same time, but this is often inconvenient. The traditional method of carbon coating which utilizes two pointed, spring-loaded carbon rods suffers from the disadvantages that if the film is deposited very rapidly (“flash” coating) a thick “soft” coat is required t o give adequate conductivity ; attempts t o improve the electrical conductivity of the film by slower deposition often results in contact between the carbon tips being broken owing to the weakening of the springs at the high temperatures reached. Flash coating also may result in relatively large particles of carbon being sputtered onto the surface. A method developed by Tomlinson and Smith a t the University of Alberta obviates these problems and gives most satisfactory and reproducible results. It is therefore described below. A single carbon filament1 is held rigidly between two copper blocks (heat sinks) and a vacuum of loe4 Torr or better obtained. A current is passed through the rod to outgas i t by bringing it to red heat for several seconds. Once the vacuum has been restored, a sufficient current is passed through the rod to produce intense white heat (but no sputtering of fragments) a t the center point. Coating is continued for a very brief period and then the vacuum (which deteriorates rapidly during emission) is brought back to Tom. The procedure is repeated several times until a coat of sufficient conductivity is deposited. The endpoint is determined by measuring the conductivity across a dummy glass slide by means of an ohmmeter. Once a resistance of about 10 MSZ has been reached, the slide is sufficiently conducting. The film that is deposited in this manner is found to be very hard and not easily removed except by final-stage repolishing. However, qualitative interferometric measurements that have been made indicate that it is less than 100 A 1

Ringsdorf RW 1447 graphite rod 1.5 mm diameter is very satisfactory.

76

D. 0. W. SMITH AND J. C. RUOIUIDGE

thick, reproducible, and has only very slight effects on the optical properties in either transmitted or reflected light. In many respects Be is a more suitable coating material than carbon. Its absorption coefficientsfor analytical lines routinely used are even lower than those of carbon, and its electrical conductivity very much better, so that thinner films may be used. Unfortunately Be is extremely toxic and health hazards may attend its use unless adequate precautions are taken. As coating materials, Al, Cu, and Au suffer from the disadvantages of having higher absorption coeflcients for the wavelengths of most analytical lines and of affecting optical properties rather severely. Furthermore, they are much more likely to be present in the sample as elements of interest. For these reasons they have not been used widely by earth scientists.

3. QUANTITATIVE ANBLYSIS 3.1. X-Rcdpl Emiseion and Absorption Processes Two aspects of X-rays are of over-riding importance to the microprobe analyst-their energy and their intensity. A general understanding of these aspects is therefore essential, and this section reviews the controlling factors, recognizing that there are several comprehensiveand fundamental treatments of the whole subject for those requiring more detail (e.g. [29-311).

3.1.1. The Continuum. X rays make up that part of the electromagnetic spectrum between 0.1 A and 200 A being bounded on the long wavelength side by ultr iolet radiation and on the short wavelength side by gamma rays. The region f interest in microprobe analysis lies between about 0.5 and 100 A. The general features of the X-ray spectrum are illustrated in Fig. 9. Superimposedlon a broad asymmetric peak (thecontinuum) are a series of high intensity narrow peaks representing the characteristic radiation of the elements present. The continuum is a complete spectrum of energies produced by random deceleration of electrons by collisions in the sample; its most energetic component clearly cannot exceed the energy of the exciting electrons. This sets the Duane-Hunt limit, which is given by

t

where h is Planck’s constant, c the velocity of light, and Eo the operating voltage (i.e., electron energy) in kilovolts. The shape of the continuum depends upon the atomic number of the target (Z), the Duane-Hunt limit (Amin), and the absorbed electron current, i.e., the sample current, i. At any

77

ELEOTRON MICROPROBE ANALYSIS

“I CON

EM~SION

NUUM

\

I

I

I

1

LOG

WAVELENGTH

I

I

I

(1)

l l l l

10

I

I

I

1

50

FIG. 9. Log-log plot of generated continuum intensity v. wavelength for two substances of very different average atomic number (PbS and Si02)a t two excitation voltages (30and 15 kV).The wavelengths of the principal characteristic lines of the elements present are also shown. Note that Pb K lines cannot be excited at either operating voltage. The continuum intensity at long wavelengths falls rapidly. The obaerved continuum (and characteristic) intensity at any wavelength will be affected by absorption processes in the sample, crystal, and detector window. The intensity scale is in arbitrery units, the absolute intensity being controlled by the probe current. Note also that the intensity scale is based on power cmd that if the diagram were to be drawn with intensity in terms of quantalsec, the continuum would have 8 very different shape, the intensity increasing continuously with wavelength.

78

D. 0 . W. SMITH AND J. C. RUCKLIDOE

particular wavelength the generated intensity of radiation (where the units are of power and not X-ray quantalsec) is given by the following expression due to Kramers [32] 1

IA= K i Z -

A2

(G - -i),

where K is a constant and ithe target current. The sample in the microprobe is the target and in most instances is composed of more than one element. The shape and intensity of the continuum in such a case can be obtained by substituting (the weighted average value for 2 in the target) for 2 in Kramers’ formula. The peak continuum intensity is a t an energy close to two thirds of the Duane-Hunt limit. The continuum is important to the analyst inasmuch as it is responsible for a large part of the background that must be subtracted from measured peak intensities ;it also ultimately controls the extent t o which peak-to-background ratios may be improved under given operating conditions. Reed [33] derived theoretical formulas for the fundamental peak-to-background ratio and compared predicted values with some obtained experimentally. The continuum is also important in that it is capable of the fluorescent excitation of wavelengths of analytical interest. Under given operating conditions the intensity of the continuum at a certain wavelength is proportional to the average atomic number. This relationship may be useful to the analyst in that having measured the background intensity for one compound, that for others of known composition may be calculated. This procedure, although often utilized, suffers from some serious dangers. In particular, Kramers’ expression applies to generated intensity and before this is measured it may be significantly modified by absorption processes. That part of the continuum that lies close to but on the high energy side of a characteristic line of an element in the sample may be subject to particularly heavy absorption (as an absorption edge is reached) and hence measured will be appreciably lower than generated intensity. The problem becomes more serious the higher the operating voltage and the longer the wavelength of radiation being detected. Figure 10 illustrates the relationship between measured background and average atomic number a t the wavelengths normally used for correction of (a) Be Kal,,, (b) Al Kal,2and (c) Na Ka,,, peak measurements for background in some standard materials used in one of the authors’ (D.G.W.S.)laboratories. The data for the longer wavelength radiation shcw much more scatter from a smooth curve mainly because of the much greater absorption. Furthermore, the background measured in practice includes radiation not belonging to the continuum and which cannot, therefore, be expressed by Kramers’ equation. The origin of this extra background is varied, coming from sources such as

79

ELECTRON MICROPROBE ANALYSIS

INTENSITY (XU? 0

I

3

2

0

4

5

6

8

7

I

1 0 1 1

9

SILICON METAL FLUORITE ICoF21 ILMENITE

-

2

1

3

l

d

1

5

A l l 5 1 :INTENSITY X I 0

FeIBpl= INTENSITYXS

- H E M A T I r E ~ w cos ~~~~~ -SYMHEllC

\

NiS

-COPPER

-

I

METAL ZINC METAL

-

-SILVER '\

-

\\

?!%tMETALGALENA

i I

T

U

N

~

i IPbSl A

-,GOLD

,L,

METAL TIN METAL

TANTALUM METAL A METAL

"i

FIG. 10. Relationship between observed background intensity and average atomic number for various simple standard materials. The three sets of data were collected using an operating voltage of 15kV at wavelengths near the characteristic Kcr lines where the background intensities for the elements Na, Al, and Fe would normally be measured. Note that the relative scatter of the data points increases with wavelength.

scattered radiation of other wavelengths, secondary and back-scattered electrons, fluorescence excited radiation from the analyzing crystal, cosmic rays, and electronic noise. Non-continuum background can be largely eliminated by instrumental improvements and pulse height discrimination.

3.1.2. Characteristic Radiation. Characteristic radiation arises when atoms are ionized in an inner shell by the ejection of an electron to the first suitable vacant orbital or beyond. Radiation is emitted when an electron falls from a higher shell to fill the vacancy in the inner shell; its energy is equal to the difference in energy of the levels between which the transitions take place. Selection rules which govern permissible and forbidden transitions will not be covered here. Readers wishing to review the subject are referred to the concise and simple treatment of Jenkins and de Vries [12] or, for a more rigorous treatment, to any of many available textbooks which deal with the subject (e.g. [31,34,35]).

80

D. 0. W. SMITH AND J. C. RUCKLIDQE

The characteristic spectrum is divided into groups of lines according to the shells to which the transition giving rise to the X rays takes place-i.e., K, L, M, and N series lines. The most energetic line in a series cannot exceed the energy required to remove an electron from that shell to a suitable vacant orbital. Furthermore, clearly no line may appear which is more energetic than the incident (exciting) electrons. It will be observed in Fig. 9 that no Pb K lines will be generated since in either of the cases illustrated, the operating voltage is insufficient to eject electrons from the K shell. The minimum energy required to excite a series of lines is termed the critical excitation energy; a recent compilation of these energies for K, L, and M lines was made by Cork [36] and also included in White and Johnson [37]. The selection of an analysis line from the characteristic spectrum is governed by many factors. In the first place, the choice of operating voltage limits the lines that can be excited and therefore used in analysis. Also, it is generally advantageous from the point of view of intensities and peak-tobackground ratios to operate at two to three times the critical excitation energy of the line chosen. This is also desirable in that it allows certain matrix corrections to be made more accurately. Even though a line is excited it may not necessarily be obtainable with crystal spectrometers: the wavelength must not be greater than 2d, i.e., twice the lattice spacing of the analyzing crystal (since, in the Bragg equation, sin 8 < 1). Geometrical considerations impose further restrictions on wavelengths that can be used and these vary with the design of the spectrometer. When matrix corrections are to be applied it is essential that physical constants such as mass absorption coefficients,fluorescence yields are available for the analysis line chosen, Assuming that all of these constraints are met, the analyst normally selects the most intense line available. In the K spectrum, Kal (or Kal, where the two are not resolved) is always more intense than KP. In the L spectrum, La, is again the most intense. Reed [38] gives the following relative intensities of lines in the L series (values were essentially constant for the four elements Tb, Er,Lu,and W investigated) : a1 = 51 yo,a2 = 6 %, P1 = 22 yo, = 9 yo,8 3 = 4 %, 8 4 = 3 yo, and y l = 5 yo. Few comparable data are available for M lines, but these are seldom used for quantitative analysis for reasons discussed below. The terminology of X-ray lines is far from systematic. However, the designation a implies that transitions producing the lines are from the next electron shell, i.e., the principal quantum number changes by 1. This means that for all elements above neon (10) in the periodic table, Ka lines arise from inner atomic transitions and, therefore, are not drastically affected by chemical bonding (which alters energies and symmetries of electron orbitals in the valence shell). However, because of the relatively small differences between

ELECTRON MICROPROBE ANALYSIS

81

energies of the L shell and valence ( M ) shell in the next few elements, there is still a limited chemical effect on wavelengths and intensities and care must be exercised in the use of these lines. The causes of these effects are discussed by Arrhenius [39]. Above sulfur, chemical effects on Ku lines may be ignored for most practical purposes. Similar considerations apply to L and M lines: for La lines, elements above krypton and for Ma lines, those from hafnium onward will probably not be significantly affected. These observations are made on general considerations of energy levels of filled shells; very few direct observations have been made on effects of bonding on M lines. Below Na all the Ka lines are significantly affected by bonding. This must be considered very carefully in their use for quantitative analysis. Provided that transitions from the valence shell are not involved, the wavelength (A) of a particular line in a series is related to the atomic number (2)of the emitter by Moseley's law (3.3)

l / h = K (2- 0)'

where K is constant for a particular spectral series and 0, a screening term, varies with atomic number and allows for repulsive effects of other electrons in the atom. Kelly [40] derived the following empirical formulas for the average value of u doublets of the principal lines.

Bearden [41] re-evaluated many experimental measurements and tabulated emission lines and absorption edges by element and by wavelength. A later tabulation, designed especially for X-ray spectroscopists and therefore including 26' values for 23 different analyzing crystals or pseudocrystals, was made by White and Johnson [42]. A compilation of X-ray lines for wavelength-geared spectrometers was produced by White and Johnson. [37]. I n using such tabulations, the possibility of appreciable wavelength shifts must always be considered whenever soft radiations are involved. Although wavelengths of certain lines are given to several significant figures, these may be relevant only for the particular material on which the original measurements were made. I n general it is only when the analyst has to identify an element present in a substance that he is concerned with the precise wavelength of the X rays.

82

D. 0. W. SMITH AND J. C. RUCKLIDQE

I n most instances, he need only know the correct spectrometer setting for the line’s intensity to be measured accurately. Since microprobe spectrometers are not generally of sufficient resetability, t o allow direct wavelength dialing, and since appreciable uncertainties still exist in the exact wavelength of many emission lines, the position must be found experimentally. For hard radiation, the wavelength remains the same for both the specimen and the standard. As discussed above, soft radiation produced by transitions from the valence shell may introduce significant wavelength shifts from substance t o substance, according to the bonding character. Furthermore, the X-ray energy is spread into bands (often of the order of 10 eV wide) within which positions of peak intensity vary with the compound. Although general features of soft X-ray spectra have been appreciated for many years, only recent developments in instrumentation have permitted significant advances in the study of their detailed character. For example, distribution of energy and intensity within bands of the L spectra of the first transition series metals and their compounds was studied by Fischer and Baun [43,44] and Fischer [46,46]. Interpretation of such spectra has been based on satellite emission during double ionization, Auger transitions, self-absorption within the sample, band theory of metals, cross-over transitions, and finally molecular orbital theory. Only the last approach, combined with the effects of selfabsorption, shows much promise of explaining the overall character and most of the detail of the spectra from non-metallic materials. Thus the molecular orbital approach has been used [47,48] to account for the characteristics of absorption spectra of some first series transition metal compounds and the K-emission spectra of Mg, Al, and Si compounds. It was also adopted by Fischer [49] to re-interpret his data on emission and absorption spectra of Ti and V compounds. In a theoretical treatment, Urch [50] calculated the energies and approximate relative intensities of the peaks, within Si K/3 and 0 K a bands from SiO, . Smith and O’Nions [51] shows that 0 K a emission spectra from many simple oxides can be interpreted using molecular orbital theory. According to this theory valence electrons are delocalized into a series of molecular orbitals which are shared by the bonded atoms instead of remaining centered on individual atoms. The energies and symmetries of molecular orbitals depend upon the nature of the bonded elements (metal and ligand) and on the character of bonding between them. Transitions take place from these molecular orbitals to inner shells of both the bonded atoms, transition By resetability is meant the acouracy with which a Spectrometer may be reset to a pre-determinedBragg angle.

ELECTRON MICROPROBE ANALYSIS

83

probabilities being controlled by factors such as symmetry of the orbitals, distribution of electron density in the orbitals, and energy difference of initial and final states. Most molecular orbitals will be stabilized relative t o the atomic orbitals and consequently, in most cases, shifts will be observed in emission peaks toward longer wavelengths. While only those X rays which arise due to transitions from the valence shell will be significantly affected by bonding, all absorption spectra should show evidence of its influence. An absorption edge corresponds t o the energy required t o eject electrons to a vacancy in the first orbital of suitable symmetry, and the energy of this vacant or partially vacant level will be affected by bonding. Fine structure has been detected on many absorption edges. This is not to be confused with the multiple edges for the principal shells, such as LI, I.rII, LIII,etc., which arise from ejections from different subshells, but is a smaller scale structure; by the molecular orbital interpretation it arises from ejections to various symmetry-permissible vacant molecular levels. Any comprehensive listing of X-ray lines associated with the elements will show wavelengths not predicted by selection rules. These “non-diagram ” or “satellite” lines are of varied origin and generally of low intensity. Some undoubtedly belong t o band spectra, originating in transitions from valence shells as discussed above. Such transitions may produce rather strong emission lines. Other satellites, however, arise from different causes : there will always be a certain probability of emission during double ionization; two vacancies can exist simultaneously, often as the result of the emission of an Auger electron, and energy levels between which transitions take place are altered sufficiently to produce X-ray lines of distinct and characteristic wavelength. I n general, the probability of such double ionizations during the emission of the line of interest is low and hence intensities of these satellites are low. It may be anticipated that fine structure on absorption edges will be affected by multiple ionizations, since the energies of inner levels and vacant orbitals to which electrons are ejected will be somewhat altered by inner-shell ionizations.

3.1.3. Factors Controlling Generated Intensities. The fundamental basis of microprobe analysis is the measurement of X-ray intensities ; factors that control generated X-ray intensity and its relationship to the observed (measured) intensity will now be reviewed, before corrections that must be applied in quantitative analysis are discussed. Basically three factors govern generated X-ray intensity under given operating conditions : (1) the rate a t which ionization of the target atoms takes place, (2) the probability of various possible electron transitions actually occurring, and (3), the probability of these transitions resulting in the emission of an X-ray line of interest.

84

D. 0.W. SMITH AND J. C. RUCKLIDOH:

A certain proportion of electrons impinging on a sample are backscattered from the surface and take no further part in the production of X raysalthough they. may be monitored to give a backscattered electron image of the sample when the instrument is being used in the scanning mode. The remaining electrons penetrate the sample and interact with the atoms to produce X rays. Of these electrons some will be re-emitted from the sample before all their energy is spent while the remainder will lose progressively more and more energy and eventually flow away as sample current. The proportion of electrons backscattered varies smoothly with the average atomic number of the sample and is almost independent of incident electron energy in the range of operating voltages commonly employed. The electron back-scatter coeacient q incremes with average atomic number. For this reason, the generated X-ray intensity for a given concentration of an element in a matrix of high atomic number will be less than that generated for the same concentration in a matrix of lower atomic number: in the second case a higher proportion of the electrons penetrate the sample to excite the X rays. The stopping power of atoms for electrons that penetrate the sample also depends upon their atomic number. The ratio of atomic number to atomic weight progressively decreases through the periodic table, and hence the number of eleatrons per unit mass decreases. The higher electron density in lighter elements makes them more efficient stoppers of electrons. Thus the generated intensity for a given concentration of an element in a relatively light matrix will be less than that generated by the same concentration in a heavy matrix. Clearly, the two factors operate in opposite directions, a1though, unfortunately, their effects are usually of different size and do not normally canoe1 out. Knowledge of the distribution of X-ray production beneath the point of beam impact is essential for the formulation of corrections to allow for matrix effects. By using thin films of “tracer” materials at known depths beneath the surface of special samples, experimental evidence on this distribution has been obtained and semi-empirical expressions derived to fit the data (e.g. [62-541). A somewhat different experimental technique for obtaining similar data was described by Schmitz et al. [56]. The shape of the electron-excited volume varies with overvoltage (the operating voltage less the critical excitation voltage) and the average atomic number of the sample, and with the electron beam diameter on impact. The influence of these effects is illustrated in Fig. 11. A more fundamental physical approach to the problem utilizes a Monte Carlo method [56]. Electron retardation and scattering, the ionization process, and the absorption of excited X rays in the sample are simulated on a computer. A large number of individual electron trajectories are followed, each being divided into many steps. At each stage the X-ray

ELECTRON MICROPROBE ANALYSIS

85

MAM

(d)

(C)

FIQ.11. Schematic diagram showing the effects of (a) and (b) operating voltage and average atomic number of the sample, (c)diameter of the incident electron beam, and (d) the critical excitation energy of the analysis line, on the size and shape of the volumes of primary (electron)excitation.

generation is calculated. Figure 12 shows the results of such calculations published in a recent short review article [57]. In general, agreement between the calculations and experiment appears to be good. The probabilities of transitions from several higher filled shells to a vacancy in an inner shell are determined by the wave functions of the levels involved. Since these wave functions are dependent on energies, symmetries, and electron densities of the levels, each of these factors is important. Direct theoretical evaluation of the probabilities is impractical, but as experimental data on intensities are available, an empirical approach can be taken. The ionization function is a measure of the electrons ejected from a particular shell or level per unit path length by an electron moving a t uniform speed through a target material. Thus (3.7)

dildx

= O ( E ,Eq, nq)

where dildx is the number of ionizations per unit path length, E the energy of incident electrons, Eq the energy required to remove an electron from the

86

D. 0. W. SMITH AND J. C. RUCKLIDOE

+

ELECTRON BEAM (20 k V )

CU-METAL

FIG. 12. A computer plot of simulated trajectories and X-ray production for 100 electrons entering a copper target with an energy of 20 kV. Some of the electrons are backscsttered out of the sample (From Duncumb [67]. Reproduced with permission from Institute of Physics, London.)

shell or level, and nq the number of electrons per unit volume having an energy Eq. Von Bethe [58] showed that

where, additionally, bq and Bq are constants for particular shells or levels, and e the charge on the electron.

Although the ionization function describes the rate at which holes are created, it provides no information on the proportion of ionizations leading to emission of X rays. I n many instances an X-ray photon is no sooner emitted than it is reabsorbed by the atom itself. Its energy is used in the ejection of an electron from a higher shell. This new ejected electron will, in most cases, be removed completely from the atom and will have a kinetic energy equal to the difference between the energy of the original photon and that of the level from which it was ejected. This process is known as the Auger efSect and ejected electrons as Auger electrons. At depth within the sample such electrons will usually be reabsorbed in producing other low-energy ionizations before eventually leaking away as sample current. Near the surface a proportion of the Auger electrons will escape from the sample and may be monitored

ELECTRON MICROPROBE ANALYSIS

87

by a secondary electron d e t e ~ t o r .The ~ ratio of X-ray photons actually emitted from a shell to the number of ionizations produced in it is the fluorescence yield (usually designated w ) . w will take different values for each shell or subshell within the atom. Values are known quite well for the K shell and limited information is also available on values for the three L subshelh. Data for M shells are extremely sparse. The following empirical relationships have been established on the basis of the available data [61,62] : (3.9) where for 2 < 10 2 = 10 - 18 2 = 18 - 50

x 105 x 105 aK = 1.06 x lo6

aK

UK

= 0.75 = 0.99

2 4

and

%I1

=

ULIII

+ Z4

where ULIII =

1.02 x 10s

wLIIIis of particular importance because the La line which is most intense and therefore usually used for analytical purposes, arises from transitions t o the LII, shell. Burhop [63] proposed an alternative formula for wK,taking into consideration screening and relativistic effects. Constants were re-evaluated and modified by Hagedoorn and Wapstra [64] :

(3.10) where

a,

= ==

a3

=

- 0.064 f 0.021,

+ 0.0340 f 0.0008, - (1.03 f 0.14) x

Figure 13 summarizes data on fluorescence yields for K, L, and M shell [65,66] The yields fall rapidly with decreasing atomic number, reaching extremely low values for elements of atomic number less than 20; also L The analysis of the energy of such Auger electrons is a developing field that shows promise in the study of energy levels within the atom and also some potential in quantitative analysis in the very light element range. Surface impurity in a quantity as small as 0.1 monolayer can be detected and identified,but quantitative analysis is hindered by noise and lack of known surface condition for calibration [59,60].

88

D. G . W. SMITH AND J. U. RUCKLIDOE

Fig. 13. Vsriation of fluoresoenoe yields for K shell [MI, L shell [eel, and M shell [66] with atomio number. Widths of lines give some indioation of uncertainties in the data.

shell are always less than K shell yields for a given atomic number. The low fluorescence yields for low atomic numbers adversely affect sensitivity for these elements but conversely the Auger electron yields are greatly increased and hence the potentialities of approach to analysis from the latter direction are enhanced. Total generated intensities of K and L lines can be expressed as (Reed [38]): (3.11)

= Kk(f&k/A)(u - 1)1'67 and

1, = KL(WL/A)(u - 1)1'87,

where I = intensity, K k and K, are (approximately) constants which depend upon operating conditions, wk and wL are K and L shell fluorescence yields (actually each subshell will have its own fluorescence yield), A is the atomic weight of the element concerned and U the over-voltage ratio-the incident electron energy (i.e., operating voltage) divided by the critical excitation energy of the line series. Thus the generated intensities of lines in the characteristic spectrum of an element depend upon several factors, and it is difficult to make any useful generalized statement about relative intensities of lines in the various series. However, the total K and total L intensities from equivalent elements (i.e., those for which K and LIIIexcitation potentials are about the same) are in the approximate ratio 1.2 :1-the K intensities being the greater [38]. As a very broad generalization it may be said that at operating voltages well above the critical excitation energies of all the characteristic lines, K lines

ELECTRON MICROPROBE ANALYSIS

89

will be preferred for analysis, largely because they are harder and the detected intensities are likely to be appreciably higher than for softer L or M radiation. Furthermore, theoretical and observed peak-to-background ratios (which improve with increasing operating voltage) are in general significantly better for K than for L lines [33].

3.1.4. Factors Affecting Observed Intensities. So far only the generated X-ray intensities have been considered, i.e., the rate a t which X-ray quanta are actually emitted from atoms. These intensities may be considerably modified before detection, largely due t o absorption within the sample. The extent to which this reabsorption occurs depends mainly on the target composition, the energy of the X-rays and the take-off angle. For X rays passing through a substance the incident intensity (I,)is related to the transmitted intensity (I)by the expression :

I = I 0 exP[-(p/P)Ptl

(3.12)

where p is the linear absorption coefficient, p / p (also written pm) the mass absorption coefficient, p the target density, and t the thickness of material through which the X rays pass. Unlike p, plp is independent of the state of the target material (gaseous, liquid, or solid) because of the introduction of p. The mass absorption coefficient ( p / ~is )thus ~ ~a measure of the extent to which radiation of wavelength h is absorbed by unit mass of material of composition X. Many experimental measurements have been made to determine mass absorption coefficients of the elements for a range of wavelengths, and where no such data exist close approximations to true values may often be obtained from formulas fitted t o existing data. Recent tabulations of mass absorption coefficients, new measurements and generalized formulas are listed in Section 3.2.4. The value of (p/p)As of a compound X for an analytical wavelength h is found by taking

(3.13)

-

where c1 , c2 * * c, are the mass concentrations of elements 1 to n in a sample. Values of ( p / ~ )change ~ * with A, but not continuously. For any absorbing element there is a series of absorption edges corresponding to energies a t which the X rays have just sufficient energy to promote particular absorption transitions. The general features of such edges and the regions between them can be visualized from Fig. 14, which shows changes in p / p with atomic number of the absorber and wavelength of the emitted line. Formulas derived for mass absorption coefficients generally show rather good agreement with much of the experimental data in the regions between the edges. However, close to

90

D. 0. W. SMITH AND J. C. RUCKLIDQE

F I ~14. . A three-dimensional representation of the variation of the maw absorption coefficient (p/p)with atomic number of the absorber ( 2 )and wavelength of the emitted radiation (A). (From Kelly [40],reproduced with permission from the Institute of Mining & Metallurgy, London).

absorption edges there are some marked discrepancies between both experimentally determined and predicted values and the formulas of different authors. I n view of the influences of chemical bonding on emission and absorption spectra this is hardly surprising. The true value of (p/p)Lzclose to an absorption edge will depend on the nature of bonding in the compound X-and hence variations from substance to substance can be anticipated. No fixed values can be expected in such circumstances and the problems of calculating average p/p values will not be overcome easily. It will beobserved that there is a general decrease in plp with both decreasing wavelength and decreasing atomic number of the absorber. Uncertainties in the true values of plp close t o an absorption edge therefore become particularly significant in the analysis of light elements in a heavy matrix. The effect of take-off angle (8)results from the different path lengths in the sample for X rays seen at different angles. The effect is illustrated in Fig. 16, and it can be accommodated by multiplying p/p by cosec 8. Where substantial uncertainties in p/p exist, clearly the smaller the value of 0, the greater the error that may be introduced. The severe problems introduced by uncertainties in p/p in the analysis of light elements are illustrated by the case of the oxygen K u emission band of two simple compounds, SiOa and Fe,Oe. These spectra, shown in Fig. 16,

ELECTRON MICROPROBE ANALYSIS

nvtmr

ElKllON 8EAM

--+ :

91

TO

DETECTION

(lOcALLV FLAT) SCALE:

tpm+

/

VOLUME ff PRIMARY

EXCITUIDN

PATH LENGTHS THROUGH mat:

rn-

Z ~ S p m .52.30'. O W p a IJ' = 0.O p a l

FIG.16. The effect of X-ray take-off angle on the distance X rays must travel through the sample and thereby suffer absorption before reaching the detection system. The amount of absorption will increase with decreasing take-off angle and also with increasing average depth of X-ray generation.

FIG.10. Shapes of the 0 Ka emission bands for quartz and hematite at 16 kV operating voltage using a KAP analyzing crystal. Peaks due to transitions from various molecular and atomic orbitals in the valence shell clearly vary in both energy and intensity between the two compounds. Peak energies are given in eV. Inferred positions of 0 K absorption edges in the two compounds are shown as dotted lines. The peaks a t about 63leV are strongly enhanced due t o a reflectivity spike at this wavelength in KAP [07a].

92

D. G . W. SMITH AND J. 0.RUCKLIDGE

can be satisfactorily interpreted by use of molecular orbital theory [67]. Changes in the relative intensities of peaks in the emission band with operating voltage suggest partial overlap of the emission and absorption bands and the inferred positions of the oxygen K absorption edges are shown dotted in Fig. 16 [67]. Although direct evidence for this overlap has not yet been obtained in the case of oxygen (indeed, further evidence obtained recently argues strongly against it), the phenomenon of partial overlap has been observed in many soft X-ray band spectra. I n these instances, the high energy parts of the emission band will be much more strongly absorbed than the low energy parts. In such a situation, the extent to which the total energy in a band is absorbed depends upon the distribution of energy in the band, and this in turn upon the bonding, which differs from compound to compound. I n the cited case of oxygen Ku,the uncertainties are further increased by the fact that there is now strong evidenoe that the intensity of the component of the spectra a t about 531 eV is strongly enhanced by the use of certain acid phthalate analyzing crystals [67a]. Some X rays produced but reabsorbed within the sample may be energetic enough to excite the radiation whose intensity is being measured. Such jluorescent excitation leads to enhancement of an analysis line to a n intensity greater than that which would be found in the absence of the higher energy radiation. Some fluorescence excitation always occurs because any part of the continuum more energetic than the analysis line will be capable of some such excitation. This continuum fluorescence is seldom very important because continuum intensity is generally rather low and anyway, in many cases, the effect in specimen and standard will be of similar size. Nevertheless the effect has been investigated and corrections suggested for it (e.g. [62,68,69]). Normally more important is characteristic jluoreecence excitation which occurs when characteristic lines more energetic than the analysis line are generated in the sample. The magnitude of fluorescence effects is inversely dependent on differences between the energies of exciting and excited wavelengths and directly dependent on the intensities of exciting wavelengths. Fluorescence effects are normally much less important than absorption effects in materials of interest to earth scientists, but nevertheless corrections should normally be applied to allow a t least for any characteristic fluorescence when quantitative results are sought. Several other factors may affect measured intensities-the quality of the vacuum, the path length of X rays through this vacuum, the reflectivity of analyzing crystals for wavelengths being used, the perfection of the crystal, absorption in the detector window, and the detector response curve to different energies of incoming radiation. However, all of these factors are completely independent of the material in which X rays are generated and thus only of significance in their effects on sensitivity.

ELECTRON MICROPROBE ANALYSIS

93

3.2. Correction Procedures The characteristic X-ray intensity yield from any element in a matrix can be shown [70], under ideal conditions, to be proportional to the mass concentration of that element. However, the preceding discussion on X-ray emission has shown that the ideal situation will never in practice obtain. I n general, corrections must be applied to the measured data t o convert from apparent to true concentrations. It is not, as a rule, possible to make absolute intensity measurements, so the X-ray intensity measured on an unknown material is always compared with a measurement on a standard under the identical experimental conditions. The first approximation or apparent concentration c1 of an element in an unknown is given by

(3.14) where I refers to X-ray intensity, and 0 and 1 to standard and unknown respectively. co is the true concentration of the element in the standard. The intensity I is assumed to be corrected for background and dead time. The background level which is subtracted from the peak is usually measured by simply detuning the spectrometer to the side of the peak. Care must be taken to avoid neighboring peaks and to allow for background slope and changes in level caused by absorption edges (see Section 4.3.2). Dead time is normally understood to mean the interval after the arrival of one pulse during which the arrival of any other pulses will be ignored. Thus, because X-ray generation processes are random in time, two or more pulses may arrive either essentially a t the same time or very quickly one after another and therefore be recorded as a single pulse. Thus some X-ray quanta are lost to the record. The more quanta generated per second the greater will be the likelihood of two or more pulses being counted as one and therefore the larger the number of “coincidence” losses to the recorded total. While detectors once represented a major source of dead time, the modern proportional counter has a dead time of a fraction of one microsecond and can normally be discounted as a source of error. Unfortunately the problem remains because dead times of other electronic components further along the line, such as amplifiers and pulse height analyzers, are significant. Beaman et al. [71,72] discussed electronic systems in terms of self-prolonging (sp) or non self-prolonging (nsp) dead time. I n the first case (sp), two pulses closely separated in time may pile up in an amplifier to form a single pulse of increased amplitude and width; this may subsequently be discriminated against by a pulse height analyzer and thus both input pulses lost to the record. I n the second case (nsp) two pulses of normal amplitude and width enter the pulse height analyzer but their

94

D. 0.W. SMITH AND J. C. RUCKLIDGE

separation is so small that only the first triggers the standard output pulse, and only one of the initial pulses is recorded. The correction for sp dead time is given by the formula

(3.15)

I0

.

= IT exp( -IT DTC)

where I , is the observed intensity (in cps), I T the true intensity, and DTC the appropriate dead time constant. On the other hand, the correction for nsp dead time is given by the same formula applicable t o detectors, i.e.,

(3.16)

IT ==

I o / ( l - I0 DTC),

where the terms have the same meaning as above. The effective dead time of a microprobe channel is a combination of dead times from the various components and must be investigated and determined experimentally for any particular combination of commercial electronic units. The problems of combinations of two types of dead time were discussed by Beaman et al. [73] and a variety of methods for experimentally determining dead time constants reviewed by Short [74]. Multichannel analyzers used with solid state detectors have dead times an order of magnitude greater than single channel systems. This arises from the time taken to sort each pulse into its channel; dead time corrections could become very laborious, having to be made separately for each channel. Instead the corrections are made automatically by allowing counts t o accumulate for times slightly longer than the present limit, the extension depending on the count rates involved.

3.2.1. Empirical Methods. Three methods are available for conversion of apparent concentration to true concentration. Stimulated by uncertainties in theoretical approaches, various researchers have put forward methods for taking matrix effects into account by purely empirical means. The simplest is the calibration curve approach, which can be applied where it is known that the sample has a composition close t o that of the standard, thus rendering corrections small or altogether unnecessary. This procedure has been employed in the works of Smith and others [76-791 and is readily applicable where the unknowns belong t o simple solid solution series, e.g., olivines, orthopyroxenes, feldspars, and where a range of appropriate standards is available. This latter requirement is the main drawback of the method, however, since suitable sets of homogeneous standards are not always easy t o acquire, and assumptions regarding stoichiometry of the unknown may not always be valid. I n the later papers of the work inaugurated by Smith on the X-ray emission microanalysis of rock forming minerals, on garnets and amphiboles [80,81], it was found necessary to adopt a different approach. A more versatile empirical method, commonly referred to as the “a factor”

ELECTRON MICROPROBE ANALYSIS

95

method, is popular in many microprobe laboratories. This was developed by a number of workers [82-861 with specifically silicate and oxide applications being explored by Bence and Albee [87]. This method depends on the fact that calibration curves for electron probe microanalysis of binary oxide systems can be closely described by the linear expression (3.17)

c i E* K i E=

+ (1 -

*

ctB)

where c i B is the concentration of oxide A in oxide binary AB relative to pure oxide A, and K L the background corrected intensity of a characteristic line of the cation in oxide A in the oxide binary AB relative to that of pure oxide A. The binary correction parameter uiE can be extended to multicomponent systems by using the concentration-weighted average of the binary parameters, to give “/3 factors.” This method uses simple oxides as standards initially, and a number of minerals of carefully established composition and homogeneity are measured against the oxides. Thus for each complex mineral standard, empirical /3 correction factors are derived which allow measurements on unknowns to be referred indirectly back to simple oxides. The /3 factor for each radiation A in the unknown U is given by

where K refers to the first approximation of the weight fraction of each oxide in the unknown, and is obtained from the counting rate relative to that for the pure oxide. The first weight fraction approximation may be relined to a close estimate of the true value by a small number of iterations and often one is sufficient. Providing that appropriate a factors are available for all oxides under examination, this method may be applied very successfully over a wide range of compositions. It is essential, however, that experimental conditions, e.g., take-off angle and operating voltage, remain constant, and this may limit the usefulness of the method when comparing between laboratories. Albee and Ray [88]have expanded the available date by tabulating oxidenormalized a factors for 36 elements commonly occurring in silicates, oxides, phosphates, carbonates, and sulfates, at 15 and 20 keV for take-off angles of 52.5” (ARL microprobe) and 38.5” (MAC and Hitachi microprobes). The method lends itself to simple hand calculation, though computer programs exist to expedite the procedure. The most serious drawbacks are that (1)insufficient compilations of factors are available at present to cover all mineralogical applications on all microprobes, and (2) each complex standard must be calibrated empirically before use, and its chemical composition must be known accurately.

3.2.2. Theoretical Nethods. The third approach to converting X-ray intensities to weight concentrations is to calculate from theory the necessary

96

D. 0.W. SMITH AND J. C. RUCKLIDQE

correction factors. This is commonly known as the ZAF (Atomic Number, Absorption, Fluorescence) approach and is in fact a semi-empirical method since many formulas and constants are empirically fitted to existing data. If all necessary equations and parameters were known perfectly, this method would undoubtedly be the most versatile and accurate, since all variables such as instrumental characteristics, experimental conditions, and standard compositions could be taken into account. The possibility of a universally applicable system thus arises, and it is for this reason that many laboratories have wholeheartedly followed this line. The relationship between X-ray intensity and concentration was reviewed in detail by Martin and Poole [66] and Campbell and Brown [89]. The reader is referred to these works for details of the development of various correction formulas, but here it is sufficient to outline corrections that are necessary, and refer only to the most generally accepted solutions. Long [7] and Sweatman and Long [28] have presented a comprehensive and practical account of the electron probe analysis of minerals, and the system they describe is basically the one most popular in the many computer programs that have been written t o perform the corrections [go]. 3.2.3. Atomic Number Correction. The discussion of X-ray excitation and emission above has prepared the reader for consideration of the three processes which disturb the ideally linear relationship between X-ray intensity and concentration. The atomic number correction, also sometimes called the generationfactor, takes account of the behavior of electrons in the target up to the point where their energy falls below that required to excite characteristic X rays. The factor has two components, namely the backscatter loss R, and the electron stopping power S. The corrected concentration (c,') takes the form

(3.19)

c1'

= cdR0 &)/(Rl So)

where 0 and 1 refer to standard and sample, respectively. Stopping power may be expressed by Bethe's formula (3.20)

St = (constant)(l/E)(Z,/A,)In (1.66 EIJ,),

where Z and A refer to atomic number and atomic weight of element i , E is the energy of the electrons (ideally integrated for all electron energies between the incident beam energy Eo and the critical excitation energy E, , but usually simplified t o (Eo E,)/2), and J is the ionization potential of element i. In a compound specimen with n elements, the stopping power becomes the weighted mean of the individual values

+

ELECTRON MICROPROBE ANALYSIS

97

and this form of obtaining the mean value of parameters for compound targets applies wherever the need arises in subsequent expressions. I n all corrections i t is necessary to use initially uncorrected concentrations t o find a starting value for the factor, and continual adjustments are called for as iteration proceeds. The value for the backscattering loss R may be derived from tables relating R t o Z and U (=E,/E,, the overvoltage ratio) given by Duncumb and Reed [91].

3.2.4. Absorption Correction. The absorption correction takes into account the way in which X rays are attenuated in traveling t o the surface from the point of excitation a t depth. This factor depends on the path length which is proportional to the cosecant of the take-off angle 8 and also t o the mass absorption coefficient pIp of the material for the characteristic radiation of the element being determined. The factor is expressed as f(x) where x = ( p / p )cosec 8. f(x) has been determined by experiments on pure elements [52,53,92], but beyond allowing a comparison of experimental and theoretical data, these results are of limited application in the mineralogical field. The equation of Philibert [93] has become the most widely accepted theoretical expression of f(x). Originally containing an atomic number correction too, the absorption factor alone is (3.21) where h = 1.2 ( A / Z 2 ) ,a is the Lenard coefficient which gives the number of electrons at a given depth in the target, and in the original was a function of E, only. The parameter a has been modified to depend on E, also [94,95]. The expression of Heinrich [95] (3.22)

u = 4.5

x 106/(EA*65 - Ei*s5)

has been shown to give marginally better results than other values [28], but Duncumb et al. [96] have demonstrated that several combinations of numerator and power of E give very similar results. Several compilations of X-ray mass absorption coefficients exist. The most recent of these are listed in Table I11 where the appropriate wavelength and absorber element ranges are also compared [40,97-1011. I n addition more recent measurements of certain mass absorption coefficients are available [102-1071. While no attempt is made here to assess the accuracy of the different tabulations, it is disturbing t o note the fact that the coefficients often differ by considerably more than the error claimed, and this emphasizes the uncertainty existing in our knowledge of these parameters, particularly in the

98

D. 0.W. SMITH AND J. 0.RUCKLIDGE

TABLE111. Compilations of X-ray mass absorption coeffioients Reference Heinrioh [97]

Emitter range Na-Mo K a Na-Zr KP Ga-Am La

Kelly [40] Theisen [loo] Theisen and Vollath [ l o l l

Li-Pu

Yb-U

MP

Be-U Sr-U

La

Dewey et al. [98]

Frazer [99]

Absorber range

Kal

I

F-Mo Ka Fe-Ao La W-U M a 0.6-15 A continuous 0.1-11.9 A continuous Ne-Mo K a Ne-Y KB Ni-Am La Ni-U Lp La-Pu M a La-Pu Mp

I

Be-U

Li-Pu

B-U H-U

Li-Pu

soft X-ray region. Generalized formulas have been fitted to existing measurements of mass absorption coefficients by Kelly [40], Frazer [99], Heinrich [97], and Theisen et al. [108]. A discussion of the errors propagated by inaccuracies in these coefficients is given in Section 4.4. 3.2.5. Fluorescence Correction. As mentioned above, the fluorescence correction factor which is applied is usually restricted to that arising from characteristic radiation rather than from the continuum. However, since instances of continuous fluorescence correction factors [lo91 as high as 7 yo have been encountered, it may sometimes be worth applying this rather complex correction factor, and the work of Springer [78,109] gives a thorough account of it. The characteristic fluorescence factor y has been treated comprehensively by Reed [38] who has derived an expression which can be applied to all possibilities of K-K, K-L and L-L interaction. Reed's expression gives I f / I A= y , the ratio of fluorescent intensity Ifto primary intensity I,, from analyzed element A: (3.23)

ELECTRON MICROPROBE ANALYSIS

99

This expression is very bulky, and a simplified version is usually of sufficient accuracy, since fluorescence corrections, where required a t all, are often less important than other corrections, particularly in the lighter elements encountered in silicate work. The simplified expression is

The details of the parameters in these expressions are best obtained by reference t o Reed’s original work, since several quantities must be interpolated from tables and graphs. The complete correction formula for the measurement becomes

(3.25) where # is the continuum fluorescence factor. The ZAP correction procedure has become most popular among earth scientists because of its versatility in the variety of materials that can be analyzed and the standards which can be used, ranging from pure elements to complex compositions. The accuracy claimed for the results is in most instances at least as good as that for the empirical methods [28]. Frazer et al. [110] have proposed a novel approach t o quantitative analysis in a method which eliminates the need t o correct for absorption, fluorescence and electron back scatter. An element is analyzed a t two or more electron beam energies, and observed intensity ratios are extrapolated t o an intensity ratio a t the critical excitation potential, when absorption, fluorescence, and backscatter corrections all become zero. Only the stopping power correction remains, and the authors claim results on minerals and alloys are as accurate as those obtained by conventional methods. A critical evaluation of the variations in the basic correction procedures as applied to certain alloys has been given by Beaman [11 13. It should be noted that the correction procedures described above assume normal incidence of the electron beam on the sample surface. For inclined incidence there does not appear to be a fully developed theory, but introduction of a sine (angle of incidence) factor in thef(x) formula is often claimed to suffice [90,93,112]. However, Brown’s [113], Bishop’s [56], and Duncumb’s [57] work suggests this may not be reliable. The major drawbacks to ZAP are (1)the complexity of correction formulas requiring tedious hand calculation when computers are not available, and (2) deficiencies in the knowledge of the necessary parameters and expressions over parts of the range of elements where it is desirable t o use them. While these are well established for many common

100

D. 0.W. SMITH AND J. 0 . RUCRLIDQE

mineral compositions, occasional compounds appear which do not yield to the routine approach. These often contain heavy elements such as Te, Bi, U, Th, where M radiation may have to be used, and it is in these instances where shortcomings in the theory become most apparent and the empirical approach may be more satisfactory.

3.3. Computer Applications 3.3.1. Computers in Dada Processing. The complete correction formula given above lends itself readily to computer processing and consequently many programs have been written. A critical evaluation of all known programs (at least 40) has been performed by Beaman and Isasi [go], and their valuable work permits easy comparison of the features of the different programs. They range in core requirements from 75K t o 1.8K, and most are written in Fortran, but Algol and APL are represented; cost per element corrected varies from 1 cent to $3.60; many programs use the ZAF procedure as outlined above, but other variants appear ; some are specifically written for metallurgical use and others for geological or both; some provide comprehensive treatment of data all the way from raw intensities to corrected element or oxide concentrations, while others perform only parts of this task; some are restricted in the range of elements that can be used while others are universal; some make provision for unmeasured elements to be included in the analysis in fixed concentrations or in fixed ratios t o other elements-a facility very useful in geological applications ; some provide statistical evaluation of large amounts of raw input data, while others require input to be presented in an intensity ratio form ; some have rigid input data requirements while others are completely flexible in terms of the amount of data and order in which it is presented-the latter feature may be valuable when collecting large amounts of information from complex geological materials ;some permit standards of complex composition to be used while others restrict standards t o simple alloy and oxide compositions. Clearly, with such a variety of features for these programs it is not possible to grade them in order of merit. Each program was written to satisfy a particular need, and for the author his program may be ideal. It is interesting to note that although several of the programs are claimed to use identical correction schemes with identical parameters, there are still some significant differences between corrected results in a set of test data which was circulated t o all authors. The evaluation of the outstanding programs by Beaman and Isasi [go] has been based on a number of criteria, not the least being the performance of

ELECTRON MICROPROBE ANALYSIS

101

these programs on some metallurgical test data. For exclusively geological purposes the following six programs are recommended to satisfy the majority of needs. The Fortran program of Duncumb and Jones [114] requires only 8K storage and is suitable for very small computers. It will perform the complete ZAP correction scheme and present oxide concentrations. Its main drawback is that the raw X-ray intensity data must be converted to intensity ratios before input. The Fortran program ARFAN of Boyd, Finger, and Chayes [115] has modest core requirements of 15K and will perform all the corrections of Duncumb and Jones [114]. X-ray counts, as measured on the probe, are converted into intensity ratios and corrected for drift. It is designed exclusively for geological materials and performs comprehensive data handling, including statistical treatment. The Fortran program MK2 of Mason, Frost, and Reed [116] has similarly modest core requirements and is possibly a little more versatile than ABFAN. The input is rather rigidly controlled, but is excellent in applications where data are punched manually. The Fortran program EMPADR7 of Rucklidge and Gasparrini [117] has large core requirements of 75K, but it gives a data handling capability not available in the other programs. This program is well suited for laboratories where data are collected directly on punched tape or cards, its main strength being its ability t o unscramble large amounts of complexly presented data. It performs the normal ZAF corrections. The APL program PROBEDATA of Smith and Tomlinson [118] is ideally suited for a remote conversational terminal in a time-sharing system, and basically has small core needs. This program takes raw data through the normal ZAF method t o final element or oxide concentrations. The Algol program of Springer [119], is another which has small core requirements and operates in a time-sharing mode. The program is well suited to handle paper tape input as received directly from the probe, and its outstanding feature is that the continuous fluorescence correction can be performed as well as the standard ZAF correction. Programs are also available, for the a-factor method, and the one of Bence and Albee [87], Fortran, 15K, is standard. This program has good data handling facilities, but without the need for this it is probably easy t o create a program for a-factor correction for individual needs since the method is so simple. The use of a computer for processing probe data will be greatly facilitated if the data are read out in a computer compatible form, e.g., punched cards, paper or magnetic tape. Cards have often been found t o be the most useful of these forms, as editing of data is invariably required, and this is difficult with the sequential data set one has on tape. If data are addressed by a code number on each line of output, as required by [117] the degree of flexibility is increased, though a code number becomes a tedious repetition if added by hand.

102

D.

a. W.

SMITH AND J. C. RUCKLIDOE

A word of warning a t this point may be appropriate. It is often tempting to give more credence to computer-generated results than may be justified by the uncertainties and limitations of the correction procedures. A result where the constituents total within 1 yo of 100% makes it easy to believe that the results themselves are accurate to that figure, which may be entirely unjustified. The user should always have a thorough understanding of the generation of his computer program and its theory, and should be armed with a certain degree of skepticism, in order to make a realistic assessment of the results.

3.3.2. Computers in Control. The computer can be integrated with the instrument to a much greater extent than that required for mere data processing. Several sophisticated systems are now available commercially where, with a small computer of 4K or 8K capacity, and with stepping motor drives on spectrometers and sample translations, the computer can exercise control over the analysis. The computer may be part of the instrument or of a larger external time-sharing system. At the simplest level the computer may be used to set spectrometers to predetermined positions for peak and background intensity measurements for a variety of elements on a manually selected spot. If the computer is large enough (e.g., 8K) it can also process the data to give a complete 9-element analysis in the space of 10 minutes [120]. The results will then guide the operator in his subsequent action. The computer can be made to seek out peak positions on the spectrometers, so that it does not rely on instrumental reproducibility, but this involves a great deal of time. The selection of a point for analysis is usually best done by an operator, but attempts have been made [121] to preselect a series of points, instrumental settings, and element combinations, and leave the instrument to perform a sequence of varied analyses unattended. A variant of this is to select a combination of elements characteristic of a particular mineral and have the sample translated point by point until the computer decides that the desired mineral has been found. At this point a complete analysis is made and the search continues. Such an approach is most valuable in fine particle analysis, and also in looking for rare minerals, e.g., Pt metal minerals in sulfide ores. Where solid state detectors are employed the computer may be just as valuable if used t o analyze and strip the spectrum either directly or from the multichannel analyzer, and process the data accordingly. Once the hardware exists the range of applications depends only on the ingenuity of the programmer. Computer control systems for electron probes will undoubtedly become very popular, for with a relatively modest additional outlay, the return on the total investment can be increased enormously.

ELECTRON MICROPROBE A N U Y S I S

103

4. ERRORS

Disagreement has persisted over the last decade regarding precision and accuracy of electron probe analysis. Thus Smith [22], in recommending an empirical approach to matrix corrections by the construction of calibration curves, observed that : “The safest course in microprobe analysis of silicates is to prepare working curves for each mineral group, using correction formulas only when standards cannot be obtained or when the concentration of the element is so low that an accurate calibration is unnecessary.” This view was maintained by Bence and Albee [87] who, when advancing an alternative empirical approach to the correction of silicates and oxides analyses, wrote : ((Theuncertainties which shroud the calculation of the individual matrix effects are so great that an empirical determination of the correction parameters is clearly necessary.” On the other hand, after a detailed analysis of factors influencing precision and accuracy, Sweatman and Long [28] reached a very different conclusion : “It is shown that analyses with an accuracy of 1% or better can be obtained with recently developed correction methods and it is suggested that this approach is now a workable and possibly preferable alternative to methods employing empirical calibration.” This divergence of views undoubtedly reflects in part improvements made in the semi-empirical ZAF correction formulas in the last few years, but also indicates that sources of possible error are many and varied; it is only with great care and experimental expertise, and proper awareness of possible hazards that a high degree of precision and accuracy can be obtained and the eficacy of various correction procedures fairly evaluated. This section deals with sources of error, treating them according to whether they affect precision, or accuracy alone. Two particularly important articles in the literature concerning precision and accuracy in microprobe analysis are those by Heinrich and Yakowitz [122] and Sweatman and Long [28]. The first authors investigated propagation of errors by correction models for quantitative electron microprobe analysis, while the second authors examined the effects of many factors including those introduced analytically. Frequent reference to these excellent articles will be made here and they constitute essential reading for anyone practically concerned with the acquisition, assessment, and use of microprobe data. One frequently sees the statement that the limit of accuracy (or precision) of the microprobe method is about & l % of the amount present. Such observations have some general, order-of-magnitude validity, but little else to recommend them. In the first place, a figure can only be put on the error if the confidence level is stated-in a normal statistical distribution a *l.66 % error at the 90% confidence level becomes a &2.58% error at the 99%

104

D. Q. W. SMITH AND J. C. RUCKLIDQE

confidence level. Furthermore, in practice the size of likely errors changes not only with concentration but also with atomic number of the analyzed element. 4.1. Factors Affecting Precision

4.1.1. The X-Ray Generation Error. The most fundamental factor controlling precision is the random nature of X-ray generation processes, often referred to as the statistical error. Poisson statistics apply to X-ray generation [123] and, if all other sources of error can be discounted or eliminated, for a particular confidence level the precision of a measurement is simply related to the total counts recorded and may be easily calculated. However, the precision of measurement of a single intensity is in practice of little direct use to the analyst in that to determine the concentration in an unknown four measurements are normally required-the intensities of peaks and backgrounds for both specimen and standard. The precision ofthe final result depends upon counts recorded during each and all of the measurements. An approximation to the statistical limit of detectability is given by 0

< I , - I , - 2.6

+

ya

where I refers t o the intensity in counts per second, t to the counting time, p t o peak and b to background, and the confidence level is close to 99%. A useful discussion of counting statistics in X-ray spectroscopy is given by Jenkins and de Vries [12].

4.1.2. Changes in Instrumental Behavior. Ideally, the microprobe should be perfectly stable for maximum precision at a particular count rate. Unfortunately, no instrument quite conforms to the ideal and several possible errors from changes in instrumental characteristics during operation must be considered. The filament is the first source of potential trouble. Emission of electrons is associated with progressive filament thinning (and finally rupture) and consequently emission characteristics change. Depending on the conditions under which it is operated (particularly emission and filament currents), these changes are detectable in as little as an hour, or only over several days. The changes affect the electron flux distribution in the beam slightly and small and steady changes in the beam current to specimen current ratio can be observed. Since we wish to maintain the specimen current constant, the common procedure of maintaining constant beam current will result in slight differences between consecutive measurements and will introduce an error if corrections are not made for this “drift.” Such corrections are usually applied

105

ELECTRON MICROPROBE ANALYSIS

by bracketing specimen measurements by two standard measurements (or vice versa) and assuming linear drift with time between measurements. Fluctuations in beam current may arise from instability in gun and lens power supplies and by filament warpage. Major variations may arise from the last cause because of physical changes in the filament and the thermal regime of the gun; normally a manual correction t o filament osition will then be made. However, provided that they are not gross, all o these effects may be compensated for automatically by the 'beam-current integrating system mentioned in Section 2.2. This is a very effective way of compensating for short-term fluctuations. It can also be applied to the specimen current, in some ways even more satisfactorily, since the changes in electron flux distribution in the beam will not affect the measurements. However, for a fixed beam current, specimen current varies with the material bombarded and if specimen currents are to be integrated they must be calibrated for each sample against a standard and the ideal counting period adjusted accordingly. This may be a more accurate procedure, but it lacks the simplicityof continual monitoring of beam current. Other important sources of error may occur in the crystal spectrometers, and these have been discussed recently by Killingworth [124]. Depending upon the counting strategy used, in some operations it may be necessary to adjust the spectrometer setting between measurements of a particular peak. The spectrometer resetability then becomes important. I n this respect instruments vary considerably in their quality and also tend to deteriorate with age as moving parts become worn. This becomes particularly important if an instrument is automated for programmed work. Jefferies and Long [121], for example, specify a resetability of *lo seconds of arc for an automated electron probe of their design which uses a semifocusing spectrometer. This is intended t o cover all situations likely to be encountered during routine analysis. I n a particular case the necessary resetability will depend on factors such as peak width, crystal resolution, and accuracy required. I n fully focusing spectrometers provided in most microprobes now manufactured, sample, analyzing crystal, and detector should lie on the surface of the imaginary Rowland circle. Normally the sample is positioned correctly by focusing on it optically; if the sample is out of focus then it lies off the Rowland circle and errors may be introduced ; the latitude permitted depends on the particular instrument, but in general the larger the radius of the circle the less critical will be the exact focusing of the sample. The effects of errors in sample repositioning were investigated for one particular instrument by Smith and Pedigo [125]. This is another particularly important factor in automated instruments : samples changed automatically must be brought to the correct focal position within established limits of tolerance. I n certain operations when the material being analyzed is not clearly visible

P

106

D . 0.W. SMITH AND J. C. RUCKLIDGE

through the optical system its location is established by scanning images. The beam may then be positioned a t a point of interest by use of the beam deflection controls. However, if this is done, the defocusing effects described in Section 2.5 must be considered. The Bragg angle for a given wavelength is set by the d spacing of the analyzing crystal. Coefficients of expansion for most analyzing crystals are of an order that temperature changes of a few degrees alter d spacings and produce significant effects on the Bragg angle, and an error when a spectrometer is returned to a predetermined angular position to measure a peak (see also Section 2.4.2). The environment may also affect the performance of the gas-flow proportional counters. The flow rate is controlled by a pressure differential between the atmosphere and gas inlet, and hence barometric changes will result in a change of flow rate, which in turn will affect the counter efficiency. I n this respect sealed proportional counters, which are unaffected by such fluctuations, give much greater stability. However, they are not suitable for use with long wavelength X rays. Gas density compensators are now available and serve to keep counter gas pressure constant and independent of atmospheric fluctuations. Temperature changes, if permitted, by affecting countergas density will also affect the amplitude of output pulses. The energy of pulses emitted by a proportional counter in response t o X rays of a given wavelength depends upon the voltage applied to it. When tight pulse height analysis is used t o discriminate against unwanted radiation, it is essential that the size of output pulses corresponding to the radiation of interest remain constant with time. This requires that power supplies for detectors maintain a high degree of stability if errors are not to be introduced. Pulses from the detector undergo amplification in preamplifier and amplifier units. Again, stability in these circuits is important particularly when pulse height analysis is used, for the same reasons outlined above. All pulses accepted by the pulse height analyzer are reshaped and passed at a standard size to counting circuits. Here errors may sometimes occur due to electronic faults, the most frequent being the missing of a digit at a certain point in the counting sequence. Errors of this kind are, of course, correctable once noticed, but are not always immediately apparent from output data. Here again temperature control of the environment is important as much transistorized equipment is designed t o function properly in a rather limited temperature range. Continuous running of the instrument in a n air conditioned laboratory effectively eliminates thermal problems. A certain proportion of the X rays entering a proportional counter from all elements with Z > 18 are absorbed in producing K-shell ionizations of Ar atoms in the gas. The energy required for this is 3.200 kV and the remaining energy will be detected as output pulses with an amplitude proportional to

ELECTRON MICROPROBE ANALYSIS

107

FIG. 17. Oscilloscope trace of signals emerging from the pre-amplifier when pure Fe metal is bombarded by 30 kV electrons. Note the presence of two peaks, the more intense corresponding to the characteristic Fe Kcr radiation to which the spectrometer is tuned and the less intense to the escape peak, resulting from the partial absorption of Fe Ka quanta by the Ar gas of the counter. Possible threshold and window settings ere indicated.

the differencebetween the original X-ray photons and 3.200 kV. These pulses form the escape peak which will be distinct from other peaks corresponding to unaffected characteristic radiation. Figure 17 shows an oscilloscope picture of pulses emerging from the preamplifier : the higher energy, more intense peak corresponds t o Fe Ka radiation (6.398 kV) and the lower energy, weaker peak to the escape peak (3.198 kV). Since the escape peak intensityrepresents incoming quanta (that happened to be partially absorbed by Ar) there may be nothing t o be gained by using pulse height analysis to exclude the peak; its intensity can often be measured along with the characteristic intensity usefully to enhance the counting rate. However, where pulse height analysis is t o be used to eliminate lower energy, interfering background radiation, considerable care must be taken in setting the threshold voltage if errors are to be avoided. If the chosen setting encroaches on the energy band of the escape peak any slight drift in amplification will change the proportions of escape peak pulses accepted and rejected. It is thus essential that the threshhold setting lies either distinctly below that of the escape peak or else clearly in the energy gap between escape and characteristic peaks.

4.1.3. Sample Contamination. During bombardment samples have a tendency to collect a deposit, which gets progressively thicker with time, in the

108

D. 0. W. SMITH AND J. C. RUCKLIDOE

area of beam impact. Taylor et al. [126] examined the composition of this contamination for one microprobe and found C = 94 %, 0 = 3-4 yo, Sn = 0.7%-0.8%, Si = 0.6-0.6%, S = 0.3-0.4%, P = O.l%, and C1 = 0.06 %. The C, Si, S, P, and C1 are probably derived mainly from oils, lubricants, and greases, Sn from soldered internal vacuum components, and 0 from adhering gas molecules and the sample itself. S, P, and C1 may also be derived in part from O-ring seals. It is likely that the composition of this contamination will vary appreciably from instrument to instrument because of the different pump oils, lubricants, and greases used and because of differences in construction. It will also probably vary significantly with the composition of the sample being anlayzed. Although carbon contamination is undesirable, the presence of other elements which are commonly sought during analysis (e.g., Si and S) may be particularly serious. Vacuum pump oils, lubricants, and greases containing these latter elements should be avoided as far as possible. The effects of carbon buildup are probably very similar t o those produced by variations in conducting film thickness, discussed later. If substantial thicknesses accumulate, both the energy of electrons incident on the sample and also the intensity of emergent X rays may be affected. The thickness of deposit apparently depends on electron beam energy, bombardment time, the beam impact area, and extent of oil vapor contamination of the vacuum. Various means of avoiding or curtailing the problem can be adopted: a cold trap in the diffusion pump minimises vapors leaking into the probe. A cold r‘finger” (a device cooled to liquid nitrogen temperature), rigged to lie very close t o the point of beam impact, will also trap oil molecules that would otherwise be deposited on the specimens [127]. Cooling the objective lens with a freon or liquid nitrogen system is possible with some microprobes and is apparently very effective in reducing contamination. Bombarding the point of beam impact with a small jet of gas has also been found useful [128,129]. Another alternative in some cases is to keep the sample moving so that no point remains beneath the beam long enough to collect a significant deposit; clearly this alternative cannot be adopted with very small areas, but is usually practicable for large homogeneous standard materials. Evidence of contamination is usually clearly visible by means of the light optical system as a brownish discoloration. Ong [130] showed that the rate of buildup of carbon contamination increases with operating voltage, but also that when a l?t wire was maintained at 300°C during bombardment, carbon contamination was eliminated completely. One way he suggested for minimizing buildup is to use a defocused beam. This, however, is completely a t variance with observations of Duncumb and Melford [131] who found that with a fully focused beam, the actual point of beam impact remained free of contamination, although a halo of carbon was deposited around it; when a

ELECTRON MICROPROBE ANALYSIS

109

defocused beam was used, a large solid contamination mark developed. Although Duncumb and Melford did not attempt t o explain their observations, it seems possible that with a fully focused beam, the temperature in the immediate area of impact inhibits carbon deposition, as in the case of Ong’s heated F’t wire. Other forms of contamination can result from traces of oils and other materials used t o prepare a polished surface remaining in cracks, pores, etc. When a sample is held in a good vacuum such as that of a microprobe, oils are liable t o come to the surface and spread out to form serious contamination. The importance of thorough sample cleaning before analysis is obvious. Atmospheric oxidation, alteration and sample contamination during handling may also pose problems, which can, however, be largely avoided by proper storage and careful manipulation. Effects of contamination are most marked on soft X rays since these are most heavily absorbed by the contamination. Particular attention must be paid t o the problem in this context.

4.1.4. Sample Damage. Certain minerals (e.g., feldspars and carbonates) and most alkali-bearing glasses are easily damaged by the electron beam. The processes causing damage are incompletely understood. Many analysts have ascribed the effects to volatilization of relatively weakly bound alkali ions. However, it is doubtful whether, even locally, samples reach temperatures a t which this could occur, and it seems certain that the mechanisms are more complex. Furthermore, different mechanisms may well predominate under different operating conditions. Evidence t o support several theories has been put forward [132-1361. Whatever causes the damage, the result is that count rates for certain elements a t a given analytical point change with time. I n feldspars and glasses count rates for alkalis usually fall with time. On the other hand, in carbonates metal count rates increase suggesting that CO, is lost and metal concentrated as the oxide. Experiments by Lineweaver [132] suggested to him that incident electrons stop a t some finite depth below the surface, producing a field that causes weakly bound positively charged ions (e.g., alkalis) to migrate from the excited area t o an adjacent region, while oxygen is liberated a t the surface. However, McConnell [1361 observed directly the degradation of feldspar, nepheline, and Na glasses in an electron microscope beam. It seems that the loss of an electron produces a positive hole in the structure and migration of a weakly bound alkali ion. I n some (boundary) situations the alkali ion may be permanently lost t o the structure which, in the presence of a sufficient concentration of such defects, ultimately collapses and becomes amorphous. The degraded material may then act as a sink for other migrating alkali ions, the reaction increasing until an equilibrium steady rate is established and finally falling off and terminating as the whole region influenced by the

110

D. Q. W. SMITH AND J. C. RUCKLIDQE

electron beam is degraded. One feature noted by most investigators is that there is characteristically a time lag in the initiation of the degradation reactions; this has been referred t o as an incubation period and its length decreases with increasing sample temperature and electron flux and also changes with composition of the bombarded material. Metals in some semiconducting minerals may change concentration during exposure to the electron beam [I371 and may even re-equilibrate after analysis. Most techniques developed to overcome these difficulties have been directed towards preventing samples from reaching temperatures a t which volatilization would supposedly occur. That such approaches have been successful seems t o have been coincidental ; actually they have allowed analysis in the time lag before degradation became significant. The simplest procedure reduces the counting period a t any analysis point to a few seconds. This has the disadvantage of poor precision because of the low total counts accumulated. A similar drawback attends reduction of beam current. Another possibility is to defocus the beam (to say 20 pm) thereby decreasing electron flux per unit area and allowing longer counting times. This will be impossible near grain boundaries or if the grain being analyzed is too small. A third alternative is t o keep the sample moving continuously beneath the beam. With a speed of 1 to 2 pmlsec it seems that significant damage is avoided. Once again, however, it is often impracticable t o use this technique for the same reasons that a defocused beam cannot always be used. Almasi et al. [138] recommended coating samples with a thin A1 film t o prevent temperature rises in poor thermal conductors. However, A1 is an important constituent in many geological materials and hence cannot be used in these cases. Heavier elements produce undesirably heavy absorption. It is possible that Be metal coating may inhibit temperature rises in samples and thereby increase the incubation period, but no work has been done to show whether this is worthwhile.

4.2. Accuracy and Instrumental Effects 4.2.1. Dead Time. Unless corrections are made, high counting rates can introduce significant dead time errors, particularly when large differences exist between rates for specimen and standard. With nsp dead time, a dead time constant = 1 psec and a counting rate of 10,000 cps, 1 % of the counts are lost. This compares with only 0.2 % a t 2000 cps and a mere 0.05 % a t 600 cps. Clearly dead time effects become insignificant a t low counting rates, which for this reason are often favored at the expense of extra counting time. 4.2.2. Operating Voltage Measurement. As the size of matrix effects depends strongly on the energy of the incident electrons, an error in the value assumed for the operating voltage will produce inaccuracies in the final results. The

ELECTRON MICROPROBE ANALYSIS

111

propagation of such errors by ZAF corrections for matrix effects was discussed by Yakowitz and Heinrich [139] and illustrated for Fe in a n olivine by Sweatman and Long [28]. Errors increase with decreasing operating voltage and decreasing X-ray take-off angle. Although a slight drift in voltage may occur during a long run, its effect is usually insignificant. More usually, errors occur because the operating voltage meter on certain instruments reads the voltage from anode to gun cap rather than anode to filament, and it is the latter that gives the true electron energy. A simple method using critical excitation energies of K lines from various metals permits meter calibration t o be checked [28]. Taking this precaution, errors from this source will normally be negligible. 4.2.3. Pulse Height Depression. This effect only occurs if pulse height analysis is used. At high counting rates pulses leaving the detector tend t o pile up-the previous pulse has not decayed entirely before the next arrives. The effect is illustrated in Fig. 18. If the window is centered on the energy

FIQ.18. Effective pulse amplitudes at two different counting rates: (a) the rate is such that each pulse is fully resolved from its neighbor, whereas (b) with a higher rate 8 new pulse starts to grow before the previous one has decayed fully. The result is that an artificial baseline is set up at EB and measured pulses have an amplitude of E , - E B . (From Jenkins and de Vries [12]. Reproduced with permission from Philips Technical Library, Eindhoven.)

of the depressed peak, say, then a pulse with normal amplitude may be discriminated against and an error results. Therefore, if a very high counting rate is employed, broad windows must be used to allow plenty of latitude for the range of peak energies generated. Recently, an electronic stabilizer has been described [1401 which provides automatic compensation for shrinkage before pulses are fed t o the pulse height analyzer. 4.3. Accuracy and Experimental Parameters 4.3.1 Line Interferences. Two kinds of line interference commonly present difficulties. I n the first case, the first-order reflection of different lines (e.g., Ka and Kt9) of the two elements occur at very nearly the same wavelength.

112

D. 13.W. SMITH AND J. 0. RUCKLIDOE

When both elements are present either an alternative analysis line which is not subject to such interference is chosen, or else an experimentally determined correction is applied to the measured intensity. The correction factor can be found by measuring the intensity of the interfering line at the analysis position relative to that of another nearby line in the series, but in a substance free of the element whose concentration is sought. In this way the contribution of the interfering line to the measured intensity can be calculated. This kind of interference can occur, for example, between Ti K/?and V Ka lines and poses problems in the determination of V in titaniferous magnetites and ilmenites [141,1421. Such problems are minimized by good spectrometer resolution. This may often be achieved by using narrow collimating slits, but at the sacrifice of some intensity. Furthermore, extremely sharp resolution can be a positive hindrance at times: for example, very narrow peaks require more accurate setting of 28. Resetability of spectrometers becomes more and more important as peak width diminishes, if the analytical strategy of moving from peak to peak for a particular sample is used (rather than the alternative of measuring the same element at all analysis points on each and every sample). Also, effects on the count rate of wavelength shifts (such as may occur for Na, Mg, Al, and Si Ka lines) will be enhanced unless peak positions are redetermined for each sample analyzed. The second form of interference is caused by near coincidence of a second (or higher) order line of another element with the first order of the analysis line. The interfering line has an energy that is distinctly greater than that of the analysisline and detector pulses will be of two distinct energies. Unwanted pulses may be discriminated against by the pulse height analyzer and interference eliminated. An example of such an interference which may be encounteredin the analysisof P in any Ca-phosphate mineral is the secondorder Ca K/?line (AEFF = 6.179) and the first-order P Ka line (A = 6.168). The greater the energy difference between interfering lines the more complete and successful the discrimination. 4.3.2. Background Determination. Since background intensity at a particular spectrometer angle varies appreciably with the material, failure to correct peak measurements for it will lead to errors of variable and unknown magnitude. The source of background radiation is considered in Section 3.1.1. Much of the contribution that is not a result of the continuum and not of the same energy as the analysis line may be eliminated by careful pulse height analysis. The remaining background intensity is normally measured on either side of the analysis line and the average taken as that at the line. This procedure may lead to errors when the position chosen for the background measurement happens to correspond to the position of a line of another element present in the sample. When this has appreciable intensity, its

113

ELECTRON MICROPROBE ANALYSIS

presence will be obvious from large discrepancies between the two background readings. However, a relatively small contribution from a weak line may easily be missed. The cause of significant differences between intensities of the two background readings should always be sought before they are accepted and averaged. I n order to ensure freedom from such errors, an intensity profile from one background position across the analysis line to the other position can be obtained on a chart or X-Y recorder. The dangers of using the approximate relationship between average atomic number of a sample and the background intensity t o calculate the background intensities for other samples have been mentioned in Section 3.1.1 (see Fig. 10). The procedure is tempting in that it saves time, and when peak-to-background ratios are large, it is unlikely t o introduce significant errors. However, it is not recommended when maximal accuracy is sought. It will be seen from Section 4.1.1 that when peak-to-background ratios are very low, accurate determinations of background intensities are almost as important as those of the peaks; for adequate statistical accuracy in this situation, peak and background counting times must be approximately the same. When peak-to-background ratios are high, precision of background intensity measurements may be much poorer than that of the peak without affecting overall precision significantly. When X-ray scanning images are obtained of intergrowths of material varying markedly in average atomic number, variations in intensity within different areas may be observed a t any given spectrometer setting and may give rise to erroneous conclusions regarding element distributions. This occurs because of the profound effect of atomic number on continuum intensity. It is illustrated in Fig. 19 which shows the picture obtained with a spectrometer '

FIa. 19. X-ray scanning images of a silver grain set in a quartz matrix. (A) Ag La image, (B) Bi M a image. The apparent concentration of Bi in the silver grain is illusory and due to the much more intense continuum (background)radiation from the higher atomic number silver then from the relatively low average atomic number quartz.

114

D . U. W. SMITH AND J. C. RUCKLIDOE

tuned to a Bi emission line, from a grain of native silver set in a quartz matrix. Subsequent quantitative analysis showed that the silver contained no Bi, the relatively high count rates observed being due simply to the difference in continuum intensity from the two materials. Thus caution must be exercised in interpreting such pictures, particularly when minor element concentrations are being investigated.

4.3.3. Errors Caused by Microinhomogeneities. One of the least emphasized and probably more common sources of error stems from the analyst analyzing material other than that intended. Analytical points are normally selected optically or from X-ray scanning images. Reflected light examination while often allowing distinction between different phases gives information only on what is a t the surface. I n microprobe analysis, the primarily excited volume extends some depth below the surface (possibly more than 10 pm in some materials under certain operating conditions). Thus hidden beneath the surface may lie a whole range of hazards-gas, liquid or glass " bubbles ", micro-inclusions of other minerals, exsolution lamellae, etc., and even straightforward grain boundaries. The material actually analyzed is, of course, the average of everything within the excited volume. Furthermore, unless every element is determined on exactly the same spot, the average composition of the excited volume may vary. Also, the volume excited directly by the beam varies with the energy of X rays being measured (see Fig. 11). Thus, unless the area within the excited volume for the least energetic radiation is entirely homogeneous, analysis for different elements is carried out on material of different average composition. Furthermore, the volume that can be fluoresced by a hard radiation generated by the beam within the sample may be more than an order of magnitude greater than the primarily excited volume. For example, Fe Ku radiation generated within an orthopyroxene exsolution lamella could excite Ca K lines in the host clinopyroxene and give an erroneously high Ca content for the orthopyroxene. Figure 20 shows some of the problems which may arise. Similar effects can occur when small particles are analyzed against large homogeneous standard grains, even when both are of similar composition. Whereas in the standard the primarily and fluorescence-excited volumes all lie within grain boundaries, in the particle, although the primarily excited volumes all lie within it, the volume which could be excited by fluorescence may lie well within the mounting medium, which will contribute nothing to the observed counts. The size of such errors is always hard to evaluate although some attempt may have to be made to do so if the problem cannot be avoided. Some relatively simple situations have been treated theoretically [ 143,1441, but no

115

ELECTRON MICROPROBE ANALYSIS

VOLUMES OF PRIMARY EXCITATION FOR TWO ELEMENTS -FLUID

MINERAL

INCLUSIONS-

COMPOSITIONAL

INCLUSIONS

I

ZONING

FIQ.20. Indicates schematically some of the hazards which may be encountered during analysis. (a) Variation in composition of the mineral within the volume of primary excitation, because of zoning. The average composition here is likely to be different from t,hat at other randomly sampled points. Furthermore, it will be noted that the fluorescence exoited volume (dashed) is not wholly in the grain of interest and hence the average composition of material from which any fluorescence excited radiation is emitted is different from that elsewhere. (b) The excited volume includes part of an exsolution lamella which is not apparent at the surface at the point of beam impact and which is not present, for example, in (a).( 0 ) The excited volume includesgas and or liquid inclusions. (d)The excited volume for one element is significantly smaller than that for another. For one element it includes part of a mineral inclusion, whereas for the other it does not. Thus the concentrations of the two elements differ between the two volumes. (e) The volume contains an inclusion of another mineral. (f) The grain is very thin and the excited volume penetrates to the mineral beneath, and this may have B totally different composition. (9) An extreme case of ( f ) is seen combined with an extra long exit path for the X-rays which must pass through adjacent material (of different composition) end thus suffer absorption to a different extent to radiation simply passing along a normal path length through the mineral of interest. Relief effects shown here are common at the edges of grains when two minerals have very different hardness. Many combinations and variations of these hazards may be encountered in practice.

general approach can be taken and each case must be assessed in the light of information available on geometries, grain sizes, etc. Errors of this kind cease t o be so serious when their presence is recognized, since erroneous conclusions are then less likely t o be drawn from the data. With polished thin sections and transmitted light examination, many potential hazards may become apparent. Even in reflected light their presence beneath the surface may be suspected by their appearance elsewhere a t the surface.

4.3.4. Errors Resulting from Xample Preparation. Ideally a sample should be perfectly flat, highly polished, and free of irregularities such as pits and scratches. I n the probe it should be oriented t o make precisely the designed angles between specimen and incident electron beam and between specimen and analyzing crystal. The most difficult requirement t o meet in rocks is the

116

D. Q. W. SMITH AND J. C. RUCKLIDGE

freedom from irregularities because of their inhomogeneity ; the various minerals present have different polishing characteristics. Only much experience and expertise on the part of the sample preparator can obviate the problems. The principal error introduced by irregular specimen surfaces is one of enhanced absorption of the emitted S rays. Calculations show [28] that for a pyroxene of composition En,, Fs,, , a scratch 0.5 pm deep reduces the measured Mg concentration by about 10 yo for an instrument with a 20"take-off angle [using a (common) 15 kV operating voltage]. The origin of the extra absorption can be visualized from Fig. 15. Clearly the effect diminishes with increasing take-off angle. Another possible source of error is from specimen contamination during preparation. Unless a sample is very thoroughly cleaned after polishing, remnants of polishing materials such as y-alumina may remain in any pores, pits, or scratches. Also, with lead laps there is the risk of contamination due to P b being taken into the Beilby layer which may be formed at the surface of some minerals, e.g., certain sulfides [145]. The use of cloth laps will avoid the problem, but other difficulties, such as the development of relief, may then arise.

4.3.5. Errors Caused by Coating Procedures. The necessity for coating samples with a conducting film, and some ways of doing it, werediscussedin Section 2.8.2. Various problems can arise. For example, if there is an appreciable difference in thickness of the films on the specimen and standard, the emerging X rays will be absorbed to a different extent by the films. Also, if the conducting film is too thin its resistance is such that there is an appreciable potential difference between the point of beam impact and ground. This negative potential on the specimen repels electrons in the beam or, in other words, effectively reduces their incident energy, causing a decrease in generated X-ray intensities. Clearly, differences in size of the effect between specimen and standard will lead to errors. Furthermore, even if the effect is identical in specimen and standard, small errors may still be introduced from the use of incorrect effective beam energies in the correction formulas for matrix effects. Local charging effects may occur when a grain of interest is isolated by cracks, ruptures of the conducting film, etc. When grains are mounted in a material that is easily damaged by the beam (e.g., most resins) ringing a grain by burning a mark in the medium around it may destroy the continuity of the conducting path and may produce charging effects in the grain. Similar problems will arise if a grain is ringed for locational purposes by, for example, a diamond indentor, after coating. Therefore, marking should be carried out prior t o coating. Poor connection between sample and ground can also produce

ELECTRON MICROPROBE ANALYSIS

117

charging effects. This is normally avoided by painting the junction between sample and holder with a conducting material such as colloidal graphite or, better, the silver paint used for repairing printed circuits, etc. 4.3.6. Errors in Standard Concentrations. I n quantitative analysis all intensity measurements are made relative to standards of known composition. The closer the composition of standard and specimen the smaller the matrix corrections, as effects will be similar in both. This consideration has led many analysts to assemble a library of standards covering, as far as possible, the range of common minerals and geological materials; in any particular analysis a standard is chosen which is as similar as possible to the specimen. Unfortunately this procedure may incidentally introduce other errors, possibly larger than those that i t seeks to avoid. With the startling results of surveys of analytical work on samples G1 and W-1 [146,147], it became clear that classical methods of wet chemical analysis resulted in interlaboratory variations of a much greater magnitude than most earth scientists had assumed. Thus confidence in the accuracy of such determinations has been severely shaken. Uncertainties for any particular standard material will only be significantly reduced if it is analyzed many times in several different laboratories, by experts using a variety of techniques. Furthermore, in many instances material analyzed in bulk is of somewhat different composition from that analyzed by microprobe; it is virtually impossible to eliminate contaminating material entirely by standard mineral separation techniques and “purified” separates frequently contain as much as 1yoimpurity. Thisis, of course, included in the bulk analysis, but invariably rejected when encountered in the microprobe. Additionally, most natural minerals are t o some extent inhomogeneous as a result of compositional zoning, etc. The average composition is obtained in bulk analysis, but unless great care is taken, standard grains mounted for microprobe analysis are unlikely to have precisely this average composition. These problems were investigated in some detail by Sweatman and Long [28], who concluded that it is best to use a mixture of pure oxides and a few simple, stoichiometric, very carefully analyzed minerals. Larger matrix effects are probably more than offset by decreased uncertainty in standard compositions. Sweatman and Long [28] were concerned primarily with the analysis of so-called rock-forming minerals (mainly silicates) and hence their results are only indirectly applicable t o much of the other great group of minerals-the ore-forming minerals (sulfides, sulfosalts, native elements, etc.). Well analyzed, pure, homogeneous natural mineral standards from this group are very scarce and most researchers in this field have found i t expedient to produce synthetic standards. Great caution must be exercised with sulfides and sulfosalts in particular because deviations from ideal

118

D . 0.W. SMITH AND J. C. RUCKLIDQE

stoichiometry are ubiquitous. Furthermore, atomic number corrections are often much larger than those encountered for silicates. As yet it is still not possible to obtain quantitative microprobe analyses of oxygen, a major constituent of many minerals. Fluorine and carbon are in the same category but are not normally so important. Nor is it possible to analyze for hydrogen. Since corrections for matrix effects depend on the concentrations of all elements present, assumptions must be made about the concentration of elements not analyzed. These may be based on various criteria such as summation of the analysis to 100 yo,the structural formula of the mineral, the distribution of ions among available sites, and overall electrical neutrality. The last criterion, which is that used in wet chemical analysis, is more difficult to apply in the microprobe technique because the instrument cannot distinguish between valence states. Fe, for example, is present as both Fea and Fe3 in many minerals. The proportions of the two valence states will affect the amount of oxygen reqvired for electrical neutrality. The choice of criteria depends upon individual circumstances and no generalizations can be made. However, errors in the concentrations finally assumed for these elements will be reflected in the accuracy of final results. As correction procedures are iterative, assumed concentrations should be readjusted (on the basis of the chosen criteria) after each iteration. +

+

4.3.7. Electronically Dispersive Analysis. I n electronically dispersive analysis, many of the problems discussed previously remain and others appear or become more severe. I n particular, resolution of peaks is poor and interferences a major problem. Dead time losses may also be severe, at the high counting rates often encountered. Beaman [148] and Russ [17] have discussed the accuracy and precision of this method of analysis in its current state of development. 4.4. Accuracy and Matrix Eflects However accurate the data acquired, the final accuracy of results from microprobe analysis depends upon the size and accuracy of corrections for matrix effects. As shown in Section 3.2 these may be applied in several ways. We shall consider fist semi-empirical ZAP corrections, such as those based on the formulas of Duncumb and Reed [91], Philibert [93], and Reed [38]. Although there is now ample evidence that, properly applied, these corrections produce marked improvements in the accuracy of results (probably of about an order of magnitude in most cases), approximations in the formulas and inaccuracies in input parameters do lead to errors in the final results. Assessment of the accuracy of various models proposed is difficult because of

ELECTRON MICROPROBE ANALYSIS

119

uncertainties that exist (and which are discussed below) in input parameters. I n individual cases it is impossible t o decide whether inaccuracies in results (measured against those from an independent and reliable technique) arise from simplifying aasumptions that are made in the formulation of the correction models, from errors in empirically determined constants that are used, or from other, analytical, errors. One approach used rather successfully applies each of the various models t o a large set of data and then makes the reasonable assumption that the model giving the smallest standard deviation of errors and a mean error closest to zero is the most satisfactory. An example of such an approach to the atomic number correction was provided recently by Martin and Poole [66]. Heinrich [149] has pointed out, however, that unless all sources of error are recognized and only the more reliable data used, it is still not possible to choose confidently the best model from several that are similar. Another approach compares the ZAF models with the results of more fundamental physical treatments such as Monte Carlo calculations [56] or the electron tranAport model [150]. Comparisons made t o date offer encouragement that the ZAF approach is very satisfactory throughout most of the wavelength range ;for soft X rays, however, substantial inadequacies remain. Yakowitz and Heinrich [139] and Heinrich and Yakowitz [122] discussed in detail the propagation of errors in the input parameters by ZAP correction models. Their conclusions, which with minor modifications would seem likely to apply to all variants of the models, are discussed below. 4.4.1. The Absoqtion Correction. Errors are minimized for (i) low values of (pip), (ii) high take-off angles, and (iii) low operating voltages, since these minimize effects of errors in (p/p) and 8. Also, a given error in operating voltage becomes less important as voltage increases. Figure 21 summarizes the effects of errors in (p/p),8, and E (the operating voltage). The diagram was constructed on the assumption that E 9 E , (the critical excitation energy). When E, is relatively large, f(x),the reciprocal of the correction factor, becomes larger anyway. It will be noted that all of the effects increase rapidly with decreasing f(x). If errors from absorption corrections are to be kept below about 1 % of the amount present, values of f(x) should not fall below about 0.8, i.e., the corrections applied t o specimen and standard should amount t o no more than about 20 % in each case. This consideration imposes severe restrictions on the application of the corrections t o soft X rays. The problem is eased somewhat by using operating voltages that are as low as possible, compatible with other considerations. It might be assumed that errors in take-off angle would be negligible, the angle being instrumentally fixed. However, in many microprobes the specimen surface may easily be misaligned, &5' error being quite feasible if correct

120

D. 0.W. SMITH AND J. C. RUCXLIDGE

loo0 2ooo

1000

JL

eoO0

Q

FIQ.21. Shows the effeots of errors in the X-ray take-offangle. operating voltage, and maas absorption ooefflcient on the h a 1 error in the absorption oorreation faator. (From Heinrich and Yekowitz [1221. Reproduced with permission from Springer-Verlag.)

precautions are not taken. Analysis in areas of surface irregularities can lead to extreme errors in the value of 8. In either case, not only is the value of 8 incorrect, but also the angle of electron beam incidence. Propagation of the latter error was not dealt with by Heinrich and Yakowitz. 4.4.2. The Fluorescence Correction. Errors propagated by characteristic fluorescence corrections are generally rather small. The major uncertainties are in the input parameters w (fluorescenceyields), J , (absorption edge jump ratios, normally designated r), and E (operating voltage). Examples of effects of errors in p/p, w , J , , 8, and E ere shown in Fig. 22, while Fig. 23 shows the variation of these effects with changing 8 and E. Errors propagated by continuous fluorescence corrections have not yet been studied. However, they will be even less significant in the vast majority of cases of interest to the earth scientist than errors propagated by the characteristic fluorescence correction. 4.4.3. The Atomic Number Correction. Errors propagated by this correction result largely from uncertainties in the electron backscatter coefficient (77) and its variation with both average atomic number of the target and overvoltage

121

ELECTRON MICROPROBE ANALYSIS

I

1

I

I

1

I

1

I

,M'+ IVY.

I

0

I

I

.20

I

I

I

I

I

/

I

.60

40

I

I

I

80

I

1 1.00

Ch

FIG.22. Shows the effect of errors in various input parameters in Reed's [38] fluorescence correction on t)he corrected percentage of iron in a series of hypothetical simple Fe-Ni alloys. (From Heinrich and Yakowitz [122]. Reproduced with permission from Springer-Verlag.)

ratio ( U = E/E,). Although the general forms of curves relating 17 to atomic number and overvoltage are well established, Heinrich [149] suggests that in detail there is appreciable fine structure, for example in the range Z = 22-29. Duncumb and Reed [91] find that the largest uncertainties are for Z > 50. Uncertainties in the value of the mean ionization potential J are also important, becoming particularly pronounced a t low atomic numbers. Furthermore, there is some evidence that J varies with the chemical state of the element. Figure 24 [150a,150b] shows the effects of uncertainties in the value of J on hypothetical analyses of simple theoretical binary compounds of A1 with elements up to atomic number 80. Errors are minimized by (i) minimizing the difference in average atomic number of specimen and standard, and (ii) by avoiding extremely large or small overvoltage ratios. I n practice earth scientists normally find atomic number corrections second in size and importance only t o absorption corrections and thus they represent an important possible source of error. Duncumb and Reed [91] suggest that these errors are normally less than 1% of the amount present provided the correction factor is kept within the limits 0.9 to 1.1. Using their

122

D. 0.W. SMITH AND J. C. RUCKLIDUE

8

0.1

J

0

0 0 -ai

FIa. 23. Diagrams showing the effects of (a) changing take-off angle and (b) changing operating voltage on errors in the corrected Fe concentration in a Fe 10 yo-Ni 90 % alloy. Errors in the input parameters are as for Fig. 22. (From Heinrich and Yakowitz [122]. Reproduced with permission from Springer-Verlag.)

123

ELECTRON MICROPROBE ANALYSIS

I

1.05-

z =

0.85

1

I

I

1

I

I

--- - --------___ ______ __-__---__*_---

13

Cmc.(Z,)

0.90

I

-. 0.2

-

I

1

I

I

I

I

I

I

FIQ.24. The effect of various expressions for the mean ionization potential ( J )on the result of a series of hypothetical analyses of binary alloys of 20 yo A1 and 80 yoof another element with atomic number between 8 and 80. (2,) C, is the result using in turn the expression for J of Duncumb and Reed [Ql] (-), Caldwell [150a] (-*-), and Berger and Seltzer [150b] (---) while C11.8is the result using J = 11.52. The data are for a n operating voltage of 20 kV. (From Heinrich and Yakowitz [122]. Reproduced with permission from Springer-Verlag.)

formula and their fitted J values, errors are minimized a t an opergting voltage of 20 kV. Largest errors occur for heavy elements present in combination with light elements and may reach more than 20 % of the difference between the factor and unity in some cases. Although these effects are minimized by using operating voltages of 20 kV or more, absorption uncertainties are then increased and probably nothing gained in overall accuracy. Monte Carlo type calculations take account of matrix effects more accurately than ZAF corrections and offer some hope of making reasonably accurate corrections t o soft X-ray data. Unfortunately the method is time-consuming and expensive even for modern, fast computers. It is doubtful, therefore, whether in most cases the extra expense is warranted by increased accuracy in corrections, particularly in view of the probable magnitude of analytical errors which will, of course, persist. 4.4.4. Alpha .Factors. Like other empirical methods the a-factor approach suffers from several disadvantages, but the problems are not as severe as

124

D. Q. W. SMITH AND J. C. RUCKLIDOE

those encountered with calibration curves. The accuracy that can be achieved by the a-factor method cannot, however, be better than that of the original determination of correction factors. Errors in these factors may be compounded by the gamut of analytical errors outlined in this section when new analyses are made. On the other hand, because matrix effects are experimentally determined, the accuracy in situations where these are very large may be better than that resulting from application of full ZAP corrections. Hence it may be expected that this approach will give more reliable results for soft X-ray data, where (p/p) is commonly large. Its efficacy in dealing with intensity data for soft X rays affected severely by bonding has, however, yet t o be established. Furthermore, to date the approach has not been extended t o cover the important groups of ore-forming minerals, the sulfides and sulfosalts.

4.4.5. Errors Associated with Bonding Effects. Because energy levels in valence shells are affected by the nature and strength of bonding, X rays arising as a result of transitions from these shells are subject to two important effects-wavelength shift (usually understood to mean a shift in wavelength of peak emission intensity) and changes in the energylintensity distribution within the emission band. Causes of these effects are discussed in Sections 3.1.2. and 3.1.4, and their practical importance in quantitative microprobe analysis was dealt with in detail by Sweatman and Long [28]. Clearly, when such effects remain undetected or are not taken into account, and when they differ between specimen and standard, inaccuracies (which may be large) can be anticipated in the final result. Wavelength shift phenomena can be countered either by retuning the spectrometer to the correct 28 angle for each compound analyzed or else by applying correction factors [28]. The former alternative is usually quicker and simpler provided the spectrometers are accurately resetable. Wavelength shifts can significantly affect the KU lines of all elements with 2 < 17. For elements of 2 < 11, perhaps more important than wavelength shifts are changes in the distribution of intensity within emission bands. Some evidence for such ohanges in the Ku lines of Na, Mg, Al, and Si of various materials was presented by Sweatman and Long [28], but in most instances their measurements were not precise enough to be conclusive. However, for oxygen, changes become quite dramatic and easily detected with a pen recorder. One approach to the problem is to integrate the total energy beneath the emission band. This may be conveniently achieved by an analyzing crystal or pseudocrystal of such poor resolution that various peaks within the band are not resolved ; then the peak intensity will be approximately proportional to the total intensity in the emission band. Semiquantitative oxygen analyses

ELECTRON MICROPROBE ANALYSIS

125

were performed in this way by Shiraiwa and Fujino [151] using an 11 in. radius lead stearate pseudocrystal. More accurate integrated intensities can be obtained using a crystal of higher resolution by driving the spectrometers a t fixed speed through the wavelength range of the emission band while counting continuously. 5 . APPLICATIONS

Over the last decade a number of reviews of applications of the electron microprobe in the earth sciences have appeared [1,6,7,152,153]. The purpose of that which follows is t o indicate both the types of application and also the range of topics within the field that have been investigated using the instrument. Reference will be made t o examples culled from the literature on the strength of their importance, their diversity, and sometimes their ingenuity. Many applications of the electron microprobe envisaged by Castaing were metallurgical in nature and researchers in that field quickly grasped the importance of the new tool. It was applied t o a wide range of tasks, centered initially on problems of diffusion in particular. The spectrum of applications developed in the field of metallurgy was reviewed recently by Goldstein [154] and Poole and Martin [9]. Earth scientists were rather slower off the mark and not until the early sixties did any appreciable number of papers including results of microprobe studies appear. In succeeding years, however, realization of the potential of the instrument became much more general and now, perhaps because of the natural breadth of the subject, applications in the earth sciences have outstripped all others in their diversity.

5.1. Applications to Qualitative Analysis and the Identification of Phases There are many instances in the study of rocks, ore specimens, etc., where straightforward identifications are needed, but where these cannot be made by normal optical, X-ray, or chemical techniques because of the limited amount of material present and/or the fine grain size. This is often the case with inclusions, exsolved material, etc. Often possibilities may be limited t o a few minerals, and thus it is usually necessary only t o establish the presence of one or two elements to make the choice. Electronically dispersive spectrometers provide a particularly convenient approach to such simple problems. The X-ray spectrum from an unidentified substance can be displayed within a few seconds on an oscilloscope and the presence or absence of the elements concerned established very rapidly. I n the absence of electronically dispersive facilities, normal spatially dispersive spectrometers may be used by tuning t o

126

D.

a. W. SMITH A N D J.

C. RUCKLIDGE

elements of interest on pure metal or other convenient standard, and then observing intensities on the unknown. I n other situations, any of several minerals containing different proportions of the same group of elements may be present. I n such situations intensities of lines from these elements are compared with those obtained from simple standards (such as pure metals). The first approximation may then be used t o distinguish between various possibilities. This would be the case, for example, for the Cu-Fe sulfides cubanite, chalcopyrite, and bornite; or, in most instances, the 3 Fe-Ti oxides ulvospinel, ilmenite, and pseudobrookite ; or the Fe-Ni minerals kamacite and taenite. Such simple identifications can usually be made in a matter of a few minutes to half an hour, depending upon facilities available and instrumental adjustments required. Sometimes the identity of a substance cannot be limited to a few possibilities and then part or all of the emission spectrum must be obtained. Once again, this is conveniently accomplished by an electronically dispersive d etectionsystem, but it can also be done by scanning spectrometers through the wavelength range and recording the output by a pen recorder. Elements present in significant amounts will appear as peaks in the spectrum and can be identified by reference to tables of X-ray wavelengths. Alternatively, charts may be compared directly with traces obtained from standard minerals once the possibilities are reduced to a manageable number. A good example of this approach is seen in a paper by Schwander and Wenk [155] who showed that in certain gneisses of the Lepontine Alps the nuclei of pleochroic halos in biotite are in fact monazite and not, as would commonly be assumed, zircon; the latter mineral is nevertheless a common accessory in the rocks. It is often useful t o depict visually the chemical variation of constituent elements in a mineral, reaction zone, alteration product, etc. This can be done in several ways: for example, a pen recorder may be used t o trace the variation in X-ray intensity (i.e., concentration) with distance across an area of interest. A multiple pen recorder may be used to show the sympathetic variation of several elements. An example of the use of the instrument for this purpose is seen in Fig. 25 [155a]. Rucklidge [156,157] has made detailed profiles of elements in low concentration (<1%) by translating the sample a t 8 pmlmin while continuously accumulating and printing X-ray counts. Computer processing of the data allows profiles to be drawn automatically which have background corrections made and are free of ratemeter distortion, thus giving a more satisfactory presentation than a conventional pen recorder. Using this method, the variation of minor elements such as Ni and C1 in dunites and serpentinites has been investigated. X-ray scanning pictures can usually be acquired from areas ranging from about 250,000 t o 100 pma so that chemical variations on a relatively large or

ELECTRON MICROPROBE ANALYSIS

127

8

-

200p .

FIG.25. Element distributions in a lunar sample (No. 10058) as obtained using a chart recorder. The upper diagram shows the outline of the grain examined with the positions of the two recorder traverses marked “a” and “b”. Capital letters indicate position of discontinuous change of extinction angle observed during examination under the microscope using cross polerizers. (From Hargraves and Hollister [ 155a-J. Reproduced with permission of Pergamon Press.)

128

D. G . W. SMITH AND J. C. RUOKLJDGE

very small scale may be examined. These pictures give only a vague idea of relative concentrations. To obtain semiquantitative information on the concentration of a particular element in different parts of the picture, the beam may be traversed linearly across the specimen and a signal from one of the detectors applied to the oscilloscope Y deflection plate and the trace superimposed on the scanning picture. Color pictures may also be obtained: elements are selected to correspond to each of the three primary colors and X-ray scanning pictures obtained from each element in turn. In areas containing more than one of these elements distinctive colors are produced by combinations of red, green, and blue. The photographs are made either by inserting red, green, and then blue light filters between the oscilloscope screen and film (when Polaroid color film may be used) or alternatively by taking normal black and white negatives and then printing in color by means of filters. Baum and Lewis [158] described an automated system for obtaining colored X-ray images. Qualitative applications, despite being very common, appear rather infrequently in the literature. They often form the preliminary phase of investigations subsequently extended on the quantitative level, or which lead to other and different lines of experiment. However, there are interesting examples that can be cited. Stumpfl [159] identified and qualitatively or semiquantitatively analyzed 10new minerals in platinum concentrates from the Driekop mine, Traansvaal. Excepting the sub-sulfide (Fe,Ni),S, all are compounds of the platinoid Ir)As,, metals withoneormoreofAs,Sb,andBi: [PtSb,,PtSb,Pt(Sb,Bi),(Pt, Pt(Ir, Os),As, ,Pd, CuSb, Pd(Sb, Bi), Pd, CuSb, ,and Pt, Sn, Cu,]. I n another study of Pt minerals, Rucklidge [1601 determined semiquantitatively the composition of some minute grains from Ontario deposits. Roedder and Dwornik [1611 investigated color-banded sphalerite-colloform material from Pine Point, North West Territories, Canada. Although the iron content of the sphalerite (which is always less than 3 yo) varies somewhat between bands, the scale of this variation is very different from that of the color banding and the two effects cannot be correlated. The real cause of the color banding remains to be determined. Hollander [1621 investigated spinel exsolution lamellae and their alteration products in titaniferous magnetites from Taberg, Sweden. Using beam scanning and point-counting techniques, he showed that the spinel alteration involved introduction of Si and loss of A1 and that the alteration product approaches chlorite in composition. Beeson and Jackson [1631 found that certain primary cumulus chromites from the Stillwater complex developed very Fe-rich margins during alteration of accompanying primary silicates to chlorite. X-ray scanning pictures and step-scanning graphs were used to illustrate these effects. Treub and de Wys [I641 demonstrated that carbonados from Bahia, Brazil are porous aggregates

ELECTRON MICROPROBE ANALYSIS

129

of mainly xenomorphic diamond crystallites ranging in size from less than 1 to over 20 pm. However, u p to 3 % igneous, metamorphic, or secondary inclusions occur in the pores. Vinogradov et al. [165] used beam scanning and chart recording techniques extensively to investigate Fe, Ni, Cr, and S distributions in diamond-graphite intergrowths of ureilite meteorites. The data support their hypothesis that the intergrowths have an impactr origin, forming when carbonaceous chondrites are converted .to ureilites by collisions in the Asteroid belt. Richardson et al. [166] used the microprobe to locate phases containing U and Th in a study of u-particle activity in Apollo XI lunar samples. Burns and Fuerstenau [I671 provided an excellent example of the use of the microprobe in qualitative investigations when they studied Mn, Fe, Co, Mg, Al, K, Ca, Ti, and Si distribution in Mn nodules. Beam scanning and pen recorder techniques established distinct correlations in concentration of certain of these elements, and showed which are associated with the manganite and which the iron hydroxide phase. I n the field of carbonate petrography, Katz [l68], using X-ray scanning techniques t o study zoning of dolomites from the Mahmal Formation of southern Israel, drew conclusions regarding growth history of the crystals and petrogenesis of the rocks. Marine oolitic iron ores from the Dogger Beta in Oberfranken, Germany were examined by Halbach [I691 by X-ray scanning techniques. Photomicrographs illustrate the presence in the innermost layers of Ca-phosphate concentrations that are also enriched in rare earths. Weber [I701 investigated Mg incorporation into the skeletal calcite of fossil echinoderms. He found significant and consistent variations between different genera from the same environment. He concluded that both environmental and genetic factors influence concentrations in these calcites. As discussed in Section 1, trace and minor element concentrations can normally be determined only qualitatively, or at the best semiquantitatively, by means of the electron microprobe. Corlett and Ribbe [I711 in some respects pushed the method t o its limits in their determination of K, Ba, Sr, Fe, Ti, and P in a large collection of plagioclase feldspars. However, their work clearly showed that many determinations of these elements (particularly K and Ba) made by other methods on bulk samples of separated plagioclase have been in error due t o the fact that antiperthitic intergrowth is much more common than previously recognized. Minor elements present in coexisting Al, SiO, polymorphs (kyanite, sillimanite, and andalusite) were determined by Albee and Chodos [172]. It has been suggested that polymorphs of this compound may coexist in equilibrium over a range of physical conditions as a result of the presence of different amounts of minor elements. However, the microprobe investigations revealed that only Fe is present in significant quantities in the samples studied, and that Fe :A1 ratios, while differing from sample t o sample, are nearly identical in the polymorphs of any one sample. A

130

D.

a. W. SMITH AND

J. C. RUCKLIDQE

combination of optical absorption and emission spectroscopic techniques and microprobe analysis was used by White and White [173] to show that the characteristic blue color of kyanite results from the presence of Ti3+ in the ppm range. The bright red to violet cathodoluminescence of some specimens is apparently related to traces of Cr. Cathodoluminescence refers to luminescence observed when substances are bombarded by electrons. The phenomenon, which results from the diffusion of electrons and transfer of energy to “impurity centers,” is well known t o all microprobe analysts whose instruments permit sample observation during bombardment. Being common in minerals it is often used qualitatively for identification purposes, positioning and focusing the beam, and sometimes for studying textures. Long and Agrell [174] outlined some potential uses in mineralogy, and subsequently several papers have appeared detailing more specific applications. Using a combination of microprobe and emission spectrographic techniques, Goni and RBmond [1751 investigated the cathodoluminescence of eight sphalerites. They found that while significant amounts of Fe quickly extinguish the cathodoluminescence, traces of elements such as Cu, Cd, Mn, Ga, and Ge produce distinctive colors. These characteristic colors may appear in well-definedzones indicating growth stages, show u p the distribution of defects, and also indicate the presence of trace elements in concentrations below the detection limits using emitted X-ray intensity. The use of cathodoluminescence in the study of growth textures, while quite feasible with the microprobe, is better carried out with special (and relatively inexpensive) equipment that may be attached to petrographic research microscopes (see, for example, Sippel [1781). Optical characteristics will normally be much superior to those of the light optical systems provided for most microprobes. Davey [177] discussed the extent to which compositional information might be obtained from observation of cathodoluminescence and described apparatus whereby this might be accomplished.

5.2. Applications to Quantitative Awlyaia 5.2.1 Quantitative Analyaia of Minerals, Inclusions, and Lamellae. Such applications represent the bulk of microprobe work reported for the earth sciences, and the literatwe is now far too vast for more than reference to some typical examples. The instrument has been applied in all branches of petrology (igneous, metamorphic, sedimentary, and experimental), in mineralogy and crystallography, the study of ore deposits, and in the investigation of meteorites and lunar material. Studies of exsolution lamellae, inclusions, and any material of extremely fine grain size often present particular analytical problems, but in principle do not differ from studies of large, homogeneous phases.

ELECTRON MICROPROBE ANALYSIS

131

Many problems are encountered where possible chemical variations are far too complex for X-ray or optical measurements to give useful results or where insufficient material is available for other techniques t o be applied. I n other cases, one mineral may be intimately intergrown with another or riddled with inclusions or exsolved material so that separation by heavy liquids, etc., is impractical. I n all such situations the microprobe, because of its unique spatial analytical cqpabilities, offers the only practical method of quantitative analysis. Moreover, even without these constraints, when pure bulk samples are readily available, more and more investigators are turning to the microprobe simply because the technique is much more rapid than classical wet chemical analysis, and, in many instances, more accurate than other techniques. Furthermore, in inhomogeneous materials the results from numerous points will give both the average and the range of composition. The only other techniques that provide useful information on compositional zoning in minerals are optical. Such methods usually provide little information other than that the mineral is indeed zoned. I n a few cases, provided that zoning is not on too fine a scale, chemical information may be obtained from optical data, the best docurnanted example being that of the feldspars. Microprobe techniques can not only depict graphically chemical variation of several elements within minerals, but may also provide semiquantitative t o quantitative information if it is required. This applies equally t o opaque and non-opaque minerals. Furthermore, in many minerals, particularly those that are either allochromatic or opaque, zoning may not be readily apparent optically and thus may remain unsuspected if microprobe analyses are not made. Schorer [178] provided an excellent example of the elucidation of chemical controls of optical zoning in a study of the hour-glass structure commonly exhibited by titanaugites. He found that (111) sectors are much richer in Si and poorer in A1 and Ti than (100) sectors, and also that progressive compositional changes take place in these sectors from core to margin. Furthermore, compositional changes were related t o measurements of the optic axial dispersion. A rather similar type of investigation was made by Hollister [179] who in this case was concerned with sector zoning in staurolite and its interpretation. One of the best illustrations of the careful application of quantitative microprobe analysis t o a petrological problem is the paper by Evans and Moore [180] “Mineralogy as a function of depth in the prehistoric Makaopuhi tholeitic lava lake, Hawaii.” Compositional variations of groundmass minerals and glass were determined in 10 samples from a complete 75 m section of the lake. Using in addition petrographic techniques, changes in the modal percentages, the chemical composition and zoning of olivines, orthopyroxenes, clinopyroxenes, feldspars, and coexisting oxide pairs were traced from the chill margin t o the interior and floor of the lake. Groundmass glass shows

132

D. 0. W. SMITH A N D J. C. BUCKLIDOE

minor compositional variations and is clearly granitic (e.g., SiO,

= 75.5 %,

Also, = 12.5 yo,K,O = 5.7 yo,and Na,O = 3.1 yo). The experimental petrologist is frequently faced with the need t o get accurate information on the composition of synthetic phases and also on any glass formed on quenching. Synthetic crystals are usually very small, often set in glass, and, not infrequently, show zoning, a feature that may be very important in interpreting experimental results. Very often, because of the size limitations of the apparatus, experimental charges are small in volume and little material is available. Techniques used before the advent of the microprobe were normally optical microscopy combined with X-ray powder diffraction, X-ray fluorescence analysis, and in some cases wet chemical analysis, Where synthetic phases are zoned, X-ray diffraction data give little indication of i t except, perhaps, for slight broadening of diffraction lines. I n h e grained material zoning may not even be apparent optically. Although these well-established techniques remain today as important tools in the repertoire of the experimental petrologist, the microprobe is being used increasingly t o obtain the chemical data on crystalline phases and glasses, while X-ray diffraction provides information on structural state, polymorphism, etc. Not only are the chemical data obtained by microprobe undoubtedly more reliable in most cases, but also they can be obtained much more rapidly, allowing more comprehensive investigations to be performed in reasonable periods of time. It should be noted, however, that all special precautions required in analysis of natural glasses and inclusions should also be exercised in this work. I n particular, the excited volume under the experimental conditions selected must not exceed the dimensions of the phases being analyzed ; as discussed in Section 4.3.3,an analysis line may be subject to characteristic or continuum fluorescence excitation by harder radiation in a volume that is much larger than that due to primary electron excitation. Several recent examples of applications in experimental petrology appear in reports of investigations of Apollo XI lunar samples, Ringwood and Essene [181] studied quench products from synthetic melts with the composition of average Apollo X I rocks and used the data to draw conclusions about likely pressuretemperature regimes in which the lunar samples were formed, and thereby to support more general arguments concerning the origin of the Moon. Akimoto et al. [182] used the instrument in a similar capacity but in this case studied liquidus phases of an actual lunar rock remelted under controlled laboratory conditions. The study of olivine-liquid equilibria in basaltic liquids by Roeder and Emslie [183] is a good example of the application to terrestrial materials-the distribution coefficient relating the partitioning of Fe and Mg between olivine and liquid was found t o be 0.30 and independent of temperature in the range investigated (1150'-1300°C). In the area of metamorphic petrology, Brown [184], using the microprobe

ELECTRON MICROPROBE ANALYSIS

133

in a straightforward analytical r d e , determined the composition of albite, epidote, muscovite, chlorite, biotite, amphibole, stilpnomelane, and garnet from green schist facies rocks of eastern Otago, New Zealand. Investigation of the mineralogy of such rocks, which are commonly fine grained with many intergrowth textures, has long been inhibited by the extreme difficulty of separation of pure mineral samples for analysis by other techniques. Atherton [185] examined a suite of specimens from the garnet isograd t o the kyanite zone of the Dalradian rocks of central Perthshire (Scotland). He analyzed garnet, biotite, and host rock and showed that in the garnet Fe and Mg increase and Mn and Ca decrease with increasing grade. The microprobe revealed zoning in the garnet indicating that the crystals as a whole do not completely equilibrate with the matrix during growth. I n addition, analyses showed that the system from which the growing garnets abstracted material ranged from 0.1 to 2.0 grams in weight. Saxena [186] used the microprobe t o determine the distribution of elements between coexisting minerals in metamorphic rocks. He partially analyzed minerals from Caledonian epidoteamphibolites from Rissa, Norway, and discussed the distribution of Fe between biotite and hornblende, and Fe and Mg between garnet and coexisting orthopyroxene, hornblende, or biotite. Evans [187] studied C1 and F contents of coexisting muscovites and biotites from near t h e sillimaniteorthoclase isograd in Maine, confirming that the concentration of halogens in biotite is usually an order of magnitude greater than that in coexisting muscovite. He pointed out that experimental data on equilibria in halogen-free systems may not be applicable to natural occurrences of biotite. Smith [188] determined compositions in fine-grained aluminoua ferromagnesian pyrometamorphic assemblages which included minerals such as mullite, pseudobrookite, iron-corundum, etc. It was shown that the rocks formed adjacent to an ancient volcanic neck. Klein [189] analyzed 22 coexisting amphibole pairs. These included actinolite-hornblende, hornblende-glaucophone, actinolite-glaucophane, and hornblende-hornblende pairs and came mainly from the glaucophane, green schist, and epidote amphibolite facies. The evidence indicates substantial miscibility gaps in the actinolite-hornblendeglaucophane system a t low temperatures. An important observation which probably has a bearing on why coexisting amphibole pairs have not been recognized very frequently, is that only occasionally do the two amphiboles occur as separate grains; usually they show patchy or complex zonal intergrowths. However, the contact between the two amphiboles is nearly always sharp, both optically and chemically. Kisch and Warnaars [1901 analyzed eight cummingtonite-hornblende pairs from metamorphic rocks, mainly from the Tydal, Sor-Trondelag area of Norway, and used these and previous analyses t o determine the Mg :Fe distribution coefficients which they contrasted with those for cummingtonite-actinolite pairs. A relationship appears

134

D . 0.W. SMITH AND J. C . RUCKLIDOE

to exist between the Mg:Fe distribution and the A1 content of the calcic amphibole, possibly related to ordering of Aly in the octahedral sites. Common feldspar compositions can be obtained extremely quickly with the microprobe, particularly when elements such as Ba, Sr, Fe, may be ignored. Na, K, and Ca can be run together on three-channel instruments and An, Ab, and Or molecules calculated from concentrations of these elements. Evans [I911 determined An, Ab, and Or contents of sodic plagioclases from lowgrade regional metamorphic rocks, demonstrating the existence in several rocks of two distinctly different coexisting compositions lying, presumably, either side of the peristerite solvus. Although these plagioclases had been examined using universal stage techniques by an experienced investigator, in every case the presence of two plagioclases had been missed. Nissen [I921 suggested the presence of a second immiscibility gap in the An,,-An,, region of the low plagioclase series from a study of certain bytownites by electron probe and other techniques. Plagioclase feldspars with oscillatory zoning and zones of cloudy inclusions were studied by Bottinga et al. [193]. They measured concentration gradients a t plagioclase-volcanic glass interfaces and concluded that plagioclase crystallization rates are diffusion controlled during certain stages of growth; a modification of Harloff's theory was used t o explain the oscillatory zoning. Clouded areas may result from build-up of Fe and Mg in the boundary layer during certain growth stages. The differential solution of plagioclase in supercritical water was investigated by Adams [I941 who analyzed the starting and residual materials of the experiments. It was shown that a t 2 kb and between 500"-800"C,while albite is removed in solution the anorthite component remains relatively insoluble ; it retains the outline, cleavage, and twinning of original grains but is cut by intricate networks of interconnecting channelways. Anderson [1951 analyzed numerous feldspars from the Labrieville anorthosite complex for Ca, Na, K, Sr, Ba, Fe, and Ti and used them together with many analyses of these and other minerals obtained by classical techniques, t o document cryptic mineralogical variations in this differentiated body. Van Schmus and Ribbe [196] investigated the composition of 38 plagioclases from chondritic meteorites showing that in the H, L, LL, and enstatite groups there are small but distinct differences in the composition of this mineral. The microprobe has been particularly useful in the study of meteorites. Applications closely parallel those in petrology and mineralogy and were reviewed by Keil.[6] and Fredriksson and Reid [197]. Many further examples appear in subsequent issues of Geochimicaet CosmochimicaActa and elsewhere. Although one can foresee that with the advent of the ion microprobe the isotope geologist or geochemist will become increasingly interested in microanalysis, as yet the electron microprobe has not found extensive application in this field. Nevertheless, some interesting examples are available. O'Nions

135

ELECTRON MICROPROBE ANALYSIS

et al. [198] found that argon retentivities of certain calcic amphiboles from southern Norway were clearly related to chemical composition and in particular to the Mg:Fe ratios. Apparent ages of the rocks, as determined by K/Ar dating of these minerals, spread over some 150 million years. Amaral et al. [199] used microprobe analyses of pyroxene to show that the K contents often reported for this mineral are almost certainly due to inclusions, adhering grains, etc., possibly explaining the widely discordant ages often obtained when the mineral has been dated by the K/Ar method. Accessory minerals of the Malsburg granite were analyzed by Willgallis [ZOO]. Having considered the U-Th balance in the granite as a whole he concluded that the relatively abundant accessories such as zircon, monazite, and allanite, make only minor contributions t o the total U-Th content, the bulk coming from relatively rare but highly radioactive accessories such as uraninite and uranothorite. Ziihringer [201] outlined a novel application for the instrument when he used it t o liberate selectively primordial He4+ from the Fayetteville chondrite. Using a sensitive mass spectrometer mounted behind the diffusion pump, he showed that although some 60 % of the He4 releases occurred a t grain boundaries, the remainder were from the interiors of mineral grains. He concluded that i t is unlikely that grains have trapped solar wind particles. During the first half of this century many mineralogists hoped that optical measurements would eventually provide the key to obtaining compositional information on a microscopic scale. Unhappily, it became clear that in many common rock-forming minerals substitutional possibilities introduce far too many variables for the equation to be solved. It should, however, be remembered that optical measurements on simpler minerals may provide entirely satisfactory ways to determine composition semiquantitativelyand even quantitatively once adequate calibration curves are established. This applies to both non-opaque and opaque minerals. I n the latter case, recent advances in instrumentation and techniques permit greatly improved measurements of properties such as reflectance, bireflectance, microindentation hardness, etc. Here the microprobe can play an important r61e in construction of calibration curves detailing variations of these properties with composition. The presence of other, perhaps unsuspected, elements, and deviations from ideal stoichiometry that may affect physical properties of the minerals, may also be detected. The surface of such work has hardly been scratched. The presence of small quantities of minute black magnetic spherules in environments stretching from the deep oceans to polar ice caps and the upper atmosphere, and in geological deposits ranging in age from Recent to at least Lower Palaeozoic, has long been known. However, their nature and origin have remained controversial although it is now widely accepted that they are derived largely from meteoritic, volcanic, and industrial sources. Determination of the fraction of extraterrestrial particles is crucial t o +

136

D. Q. W. SM”H AND J. C. RUCKLIDQE

calculations of the rate of infall, which in the past have ranged between lo2 and lo6 metric tons per year. The microprobe is being used extensively to study these spherules, and it is now clear that most commonly they are composed of one or more of magnetite, maghemite, hematite, and nickel-iron. Schmidt and Keil [202] studied spherules from Atlantic Ocean sediments, Marvin and Einaudi [203] those from Pleistocene and Recent beach sands, and El Goresy [204], examples from the Greenland ice cap. More recently, Jedwab [205] described spherules of probable cosmic origin in Mn nodules. Doubtless further investigations will be made and perhaps the ion microprobe may be used t o search for isotopes of extraterrestrial origin in spherules whose cosmic origin is likely but not proven.

5.2.2. Applications to the Analysis of Natural Glasses. Keil [6], reviewing applications of the microprobe in the earth sciences, remarked that there had been very few examples of its use in investigation of natural glasses. These glasses, produced by agencies as diverse as volcanic activity, pyrometamorphism, meteorite impact, and lightning strikes, are often available only in small amounts and microprobe analysis often proves t o be the only suitable approach. Undoubtedly among the most important of the many examples now available of applications in the study of natural glasses are the analyses of the various kinds of lunar glass in samples brought back by Apollo missions. This is an excellent example of the usefulness of this nondestructive technique in examining precious materials only available in small quantities and which may be needed subsequently for work of either a similar or different nature. Lunar glasses analyzed range from basaltic t o granitic, and include examples of the distinctive Ti-rich compositions. Some of the glass occurs as (‘splash” coatings on otherwise well-crystallized material ; in other instances, it is the acid interstitial material in basic rocks. Analyses of the abundant glass spherules are also reported and particularly exciting is the evidence for acidbasic silicate liquid immiscibility observed by Roedder and Weiblen [206]. Certain chondritic meteorites occasionally contain small amounts of glass, Thus Van Schmus [207] analyzed “primary” glass in chondrules from the Mezo Madaras meteorite and showed it to be Na-Al rich; he also analyzed ((secondary” glass interstitial t o olivine, pyroxene, etc., in an inclusion in the meteorite. He concluded that the glass resulted from metamorphism in the 800”-1200”C range. Bunch and Fuchs [208] investigated plagioclase glass (“maskelynite”) in the Tucson meteorite and found that it deviated significantly from stoichiometric plagioclase-particularly in its high silica content. Walter and Cassidy [209] examined tektites from various strewn fields and found that most types are homogeneous. However, moldavites and especially Muong Nong type tektites were found t o be heterogeneous, the latter being

ELECTRON MICROPROBE ANALYSIS

137

composed of homogeneous lenticular grains with sharp compositional boundaries between individual grains. Smith and Westgate [210] developed a technique for characterizing pyroclastic materials by means of the compositions of the glass shards that they contain. It proved possible t o differentiate unequivocally pyroclastics of rather similar dacitic composition that are widespread in Quaternary deposits of the northwestern United States and southwestern Canada (e.g., the Mazama, Glacier Peak, and St. Helens “ Y ” deposits). This technique used in conjunction with radio-carbon dating offers the possibility of precise stratigraphic correlation over extensive areas even when lithological variations in the enclosing rocks are extreme. Another example of the application to volcanic materials was described by Hay and Iijima [211] who examined tuffs of the Koko Craters on Oahu in the Hawaiian chain. They showed that about a quarter of the Si, half of the A1 and Mg, and three quarters of the Ca, Na, and K must be lost in converting sideromelane to an equal volume of palagonite. Also it was demonstrated that changes from sideromelane to palagonite occur abruptly across a sharp interface. Switzer and Melsom [212] analyzed rock fulgurites (glass crusts and glasslined tubes) formed in andesites, diorites, and hornfelses, etc., and have found that the glass is very heterogeneous-individual mineral grains apparently having been isochemically melted in situ and not mixed to any significant extent with melt from adjacent grains. Thus it is clear that an extremely rapid increase to a high temperature was followed by very rapid cooling. Kleinmann [213] identified baddelyite (SrO,) in highly siliceous Libyan “desert glass,” suggesting a temperature of formation in excess of 1600°C and thus an origin by impact fusion of desert sand. Yang et al. [214] investigated the so-called desert varnish that appears on pebbles after long exposure to desert conditions. This material is amorphous and thus in some respects similar to glass. On the basis of the changing distribution of Fe, Ti, Mn (increasing towards the margin), and Si, Al, and K (decreasing towards the margin), they suggested that the coating is partially derived from an external source and partially from the material being “varnished.” The darker ( Z older) the varnish the higher the Fe, Mn, Al, and K contents that were found.

5.2.3. Application to the Analysis of Roch. The possibility has recently been demonstrated of obtaining microprobe analyses of rock samples of comparable accuracy to that obtained by wet chemical techniques and in a small fraction of the time. The main advantage over X-ray fluorescence spectroscopy, on the other hand, is that matrix effects are smaller and corrections that take account of them, quite well known. Because up t o four elements may be run a t one time (depending upon the microprobe and its facilities), only a few runs may be needed to analyze the common rock

138

D. Q. W. SMITH AND J. 0. RUCKLIDQE

forming elements. Sensitivity and peak-to-background ratios are generally appreciably better in X-ray fluorescence spectroscopy, but for the major and minor elements of most rocks the microprobe’s limits are adequate. Three basically different techniques have been used for whole rock microprobe analyses. I n the first the rock is powdered to less than 10 pm grain size, pressed into a tablet and coated with a carbon film. The sample may then be analyzed-preferably with a defocused beam and continuous sample movement to improve averaging of inhomogeneities. I n the second method [215], glass disks prepared by the method of Norrish and Chappell [216] for X-ray fluorescence analysis, are carbon-coated and analyzed directly by the microprobe. Because of the dilutant effect of fluxes (Li, B,07, LiaC03, La,03), count rates from a given element plotted against concentration for standards of known composition give a nearly perfect straight line and corrections for matrix effects need not be applied. Reed [217] pointed out that removal of the Laa03 “heavy element absorber” from the flux produces significantly better peak-to-background ratios and yet affects the linearity of calibration curves only marginally. The main drawback lies in the great dilution used to standardize matrix effects from rock t o rock; this greatly decreases sensitivity, and increases counting times for satisfactory statistical accuracy. The third method [218] relies on direct fusion of rock powder t o produce a glass bead. This is polished, carbon-coated, and analyzed in the normal way. The great advantage is that since the sample is not diluted, relatively high counting rates are obtained giving more rapid analysis and better sensitivity. Although matrix corrections must be applied to undiluted fused rock samples, these cutn be made reasonably accurately. The main problems may lie in homogenization of rock powders during fusion without at the same time losing material by volatilization. Reed [217] has critically discussed the methods of whole rock analysis by the microprobe. Point counting beneath the petrographic microscope is a well-established method of modal analysis. There, are, however, several difficulties which are not easily surmounted. Perhaps most important is the fact that every point requires a decision (identification) by the petrographer, who, depending on his experience and capabilities, will be more or less fallible. Second, opaque minerals or those of high relief appear more prominent in the analysis than they really are; when they are present as small grains they will often be counted as a t the surface when actually they lie some distance below it. Opaque phases are seldom differentiated during normal point-counting procedures, either because reflected light facilities are not available on the microscope or because the thin section is not polished (and probably, anyway, has a cover slip). Microprobe techniques for modal analysis overcome most of these problems. The instrument is tuned t o “key” wavelengths so that

ELECTRON MICROPROBE ANALYSIS

139

as many phases as possible contain one or more of the corresponding elements. The approximate counting rates for each element on each mineral phase is established by preliminary measurements and the microprobe programmed to count on a grid of chosen size, control being through a small on-line computer. The same computer may be used to identify the mineral a t each grid point on the basis of predetermined counting rates and a given error window. The technique has the obvious advantages of objectivity and accuracy and can be carried out during times of low work pressure. Surface irregularities in microprobe. mounts may pose problems and where these result from poor polishing characteristics of a particular phase, an inordinate number of “unidentifieds” will be recorded, depleting the count for the phase in question. Such difficulties will be surmounted only by high quality specimen preparation. Keil[219] outlined the fundamentals of the technique.

5.3. Applications Utilizing Soft X - R a y Spectra Technological advances in recent years, particularly in crystals and detectors, permit the present-day microprobe to be used as a rather versatile soft X-ray spectrometer. Although the term soft X ray lacks precise definition it is generally taken to include X rays of wavelength greater than about 5 A. I n the main, such X rays are due t o transitions from the valence orbitals of atoms or molecules, and it is becoming increasingly clear that they carry much useful information on bonding. Early investigations centered on the measurement of wavelength shifts. Arrhenius [39] explained the cause of wavelength shift in terms of the changing screening effect that occura when certain outer valence electrons, which normally spend part of their time close to the nucleus, are either removed or else the radial distribution in charge altered by a change in hybridization. I n particular, shifts of Al and Si K radiation have been investigated. Day [220] used X-ray fluorescence spectrometry to investigate the shift of the A1 KU line with coordination, and similar measurements can now be made with the microprobe. Thus the shift of position (A) was studied in 45 different silicates by White and Gibbs [221]. They found a strong correlation between A and the mean Si-0 bond length and that this length may be predicted with a n accuracy approaching that obtainable in modern crystal structure analysis. Because the bond length is closely related to structural type, the degree of polymerization of SiO, tetrahedra can also be predicted. Furthermore, there is a strong correlation obtained between A and the Al ;Si ratio in tetrahedral sites of tectosilicates. Thus the method provides a means of characterizing new silicates in terms oftheir structures and will be particularly useful in glasses where precise structural refinements cannot be made. White and Gibbs [222] followed this study with one of the A1 KP spectra of 27

140

D. 0.W. SMITH AND J. C. RUCKLIDOE

Al-bearing substances and again showed correlations between peak shift, on the one hand, and coordination number and A1-0 distances on the other. Lovering and Widdowson [223] investigated wavelength shift of S Ku from sulfate minerals relative to that from sulfides. The S Ku peak shifts by amounts up to 0.0032 A-i.e., about 1.4 eV at this wavelength. Using such measurements they demonstrated that meionite scapolites contain SO: ions while helvite contains S2-ions, as does the brittle mica anandite. While these studies clearly demonstrated the usefulness of such measurements, the underlying factors controlling the detailed character of soft X-ray emission bands, in terms of both intensity and energy, only started to become clear with the work of Dodd and Glen [47,48]. Very careful and accurate observations on the Si, Al, and Mg KP emission spectra showed that their characteristics could be very satisfactorily explained by a qualitative application of molecular orbital theory. The door has now been opened for observations on the nature of bonding in minerals, etc., by means of the microprobe. Although such studies are still in their infancy, even the initial work of Dodd and Glen [48] allowed them to infer a surprisingly large component of covalent bonding in the oompounds MgO and aAl,O, . O’Nions and Smith [224] have argued the presence of Ti4+ in ilmenite and pseudobrookite produces a more ionic Fe-0 bond compared with that in wustite and hematite, respectively. Dodd and Glen [225] applied a molecular orbital approach to the understanding of Si KP spectra from a wide range of silicates, and by combining i t with peak-shift measurements were able to determine in each case the approximate decrease in stability of the Si-0 bond relative to that for quartz. They further showed [226] that this approach provides a badly needed new tool for studying bonding in glasses. Among soft X-ray spectra, perhaps of greatest significance is that of oxygen. The majority of minerals are compounds of this element and thus the understanding of the detailed character of the spectrum, preferably together with the soft X-ray spectra of the elements combined with it, may contribute very greatly to the understanding of bonding in these compounds. Furthermore, there are at present no satisfactory simple methods for the direct and routine quantitative determination of oxygen in minerals, glasses, etc. It is, therefore, perhaps surprising that little effort has been devoted to unraveling the details of the 0 Ka spectrum. Progress has undoubtedly been greatly hindered by the fact that a rather distorted version of the spectrum is obtained with crystals used in most soft X-ray spectrometers. Smith and O’Nions [61,67] argued that a prominent peak in the 0 Ka spectrum, dismissed throughout most of the literature as anomalous, is probably real but heavily absorbed in most analyzing crystals and the sample. However, subsequent research (to be published) indicated that a reflectivity “spike” which +

ELECTRON MICROPROBE ANALYSIS

141

exists in KAP [67a] and other acid phthalate crystals a t or near the energy of the oxygen absorption edge, and which varies in iatensity from phthalate to phthalate, actually strongly enhances real but relatively weak emission intensity in this wavelength region, thereby producing an intense peak. Ideally, a diffracting crystal or grating free of oxygen is desirable in any investigation concerning oxygen spectra, but none is currently commercially available. Clearly, further work is necessary t o unravel completely the relationships between molecular orbital levels, energies, intensities, absorption, and instrumental effects. If this can be achieved, however, it may prove possible not only t o use the 0 Ku spectrum in the interpretation of bonding, but also to develop methods for the quantitative analysis of oxygen. The problems that remain to be solved in that latter area are, however, formidable. One of the biggest unresolved problems for the microprobe analyst investigating geological materials is that he cannot determine the valence state of elements present. This is a serious limitation in that elements such as Ti, Mn, and Fe may occur in one or more different valence states. Albee and Chodos [227], following up observations of Andersen [228] on the intensity characteristics of Fe L emission spectra from different substances, suggested that it is possible t o determine the valence state of first series transition metals (specifically Fe and Mn) by means of the microprobe. They recorded measurements on members of several simple oxide and silicate mineral series and showed an essentially linear relationship between Fe L,,/Fe L,,, intensity ratios and composition in each case. However, these simple relationships break down when the substances are chemically more complex. Smith and O’Nions [229,230] and O’Nions and Smith [224] have shown that the measured Fe LII/L,II ratio is partly a function of the extent of self-absorption of the L,, and L,,, emission bands in the sample, partly reflects differences in character of bonding, and also is severely affected by the nature of the other metal cations that are bonded to the shared oxygens. I n simple mineral series there will necessarily be a simple relationship between the L,,/L,,, ratio and composition, but in complex solid solutions this is lost. The problem remains. Recently a particularly dramatic illustration of the potential and power of the electron microprobe was provided by accounts of research into lunar samples collected by Apollo spacecraft. I n the report on the mineralogy and petrology of lunar rocks in the proceedings of the Apollo XI Lunar Science Conference, no less than 44 of the 68 papers presented included results obtained by the instrument, and in many cases relied heavily on the data. A similar pattern is obvious in the publications resulting from the subsequent Conference on Apollo XI1 Rocks [231]. Many important conclusions

142

D. C J . W. SMITH AND J. C. RUCRLIDCJE

regarding these samples could not or would not have been reached without the microprobe, and the scientific return on the invested space dollar would have been drastically diminished.

LISTOF PRINCIPAL SYMBOLS Atomic weight of a n element Mass concentration of an element in a standard Mass concentration of an element in a n unknown (may be approximate) Mass concentration of an element after a correction has been applied Dead time constant Accelerating potential (operating voltage) of the microprobe ( = electron energy) Critical excitation energy of an element for a particular characteristic radiation The absorption factor-the reciprocal of the absorption correction factor 1.2(A/Z2) (h is also used as the standard symbol for Planck’s constant) X-ray intensity Observed X-ray intensity True X-ray intensity Generated (continuum) X-ray intensity at a wavelength h Mean ionization potential of an atom Absorption edge jump ratios as used in Figs. 22 and 23 First approximate weight fraction of an element A in an unknown, in terminology of [87] Absorption edge jump ratio (defined as J R in Figs. 22 and 23) Electron back scatter factor Mass stopping power of target for electrons ( = d E / d ( p x ) ) Overvoltage ratio (=E,/E,) Atomic number of an element Alpha factor for element A in binary system AB (see [87]) Beta factor for element A in complex unknown (see [87]) Characteristic fluorescence factor Electron back scatter coefficient (i.e., fraction of incident electron current that is backscattered) Take-off angle (emergence angle) of X radiation-measured from surface of sample to beam Bragg angle of X-ray diffraction X-ray wavelength Linear absorption (attenuation) coefficient Mass absorption (attenuation) coefficient ( =pm) Mass absorption coefficient of an absorber of composition X for radiation of wavelength Target density Lenard coefficient (a modified linear absorption coefficient for electrons) (PIP) cosec 8 Continuous fluorescence factor Fluorescence yield for “ X ” shell of atom The use of a bar over a symbol (e.g., parameter.

2) indicates the weighted mean value of that

ELECTRON MICROPROBE ANALYSIS

143

ACKNOWLEDGMENTS We are grateful to many colleagues for helpful discussions in the preparation of this article, in particular to Dr. J. V. P. Long, Dr. S. J. B. Reed, Professor J. F. Lovering, and Mr. D. Sewell. D. G. W. S. is also pleased to acknowledge facilities kindly provided by the Department of Mineralogy and Petrology, Cambridge and a Nuffield Foundation Travel Award during the tenure of which the manuscript was prepared. J.C.R. gratefully acknowledges the facilities that were made available in the School of Geology, University of Melbourne. Both authors received financial support for the work from the National Research Council of Canada. We 8lSO thank publishers and authors for permission to reproduce various figures used in this article.

REFERENCES 1. Agrell, S. O., and Long, J. V. P. (1960). The application of the scanning X-ray

2. 3.

4. 5.

6. 7.

8.

microanalyser to mineralogy. Ifa “X-ray Microscopy and X-ray Microanalysis ” (A. Engstrom, V. E. Coslett, and H. H. Pattee, eds.), pp. 391-400. Elsevier, Amsterdam. Castaing, R. (1951). Ph.D. Thesis, Publ. O.N.E.R.A. No. 55. UniversityofParis. Castaing, R. (1963). X-ray microprobe techniques. I n “Third International Symposium on X-ray Optics and X-ray Microanalysis” (H. H. Pettee, V. E. Cosslett, and A. Engstrom, eds.), pp. 263-277. Academic Press, New York. Anderson, C. A., and Hinthorne, J. R. (1972). Ion microprobe mass analysis. Science 175, 853-860. Marton, L. (1969). Survey of microanalysis-interpolation and extrapolation. I n “Electron Probe Microanalysis” (A. J. Tousimis and L. Marton, eds.), pp. 1-14. Academic Press, New York. Keil, K. (1966). The electron microprobe X-ray analyzer and its application in mineralogy. Forlschr. Mineral. 44, 4-66. Long, J. V. P. (1967). Electron probe microanalysis. I n “Physical Methods in Determinative Mineralogy” (J. Zussman, ed.), pp. 216-260. Academic Press, New York. Birks, L. S. (1963). “Electron Probe Microanalysis.” Wiley (Interscience), New

York. 9. Poole, D. M., and Martin, P. M. (1969). Electron-probe microanalysis: Instrumental and experimental aspects. Metals Mater. 8 (Metall. Rev. 14, 61-84, Rev. No. 133). 10. Brown, L. C., and Thresh, H. (1970). Electron-probe microanlaysis. I n “Tools and Techniques in Physical Metallurgy” (F.Weinberg, ed.), Vol. 2, pp. 601-649. Dekker,

New York. 11. McFarlane, A. A. (1971). The use of octadecyl hydrogen maleate and rubidium

hydrogen phthalate as X-ray spectrometer crystals in the electron microprobe, together with evidence on the electron bonding in the K-spectra from carbon and carbon compounds using the OHM crystal. Proc. Nat. Conf. Electron Microprobe Anal., 6th. 1971. Pap. No. 49. 12. Jenkins, R., and de Vries, J. L. (1970). “Practical X-ray Spectrometry,” 2nd ed. Philips, Eindhoven. 13. Franks, A. (1972). NPL X-ray gratings and their application. Proc. 6th I n t . Cong. X - R a y Opti. Microanal., 1971, pp. 57-62.

144

D. Q. W. SMITH AND J. C. RUCKLIDQE

14. Nicholson, J. B.,and Hasler, M. F. (1966). A blazed-grating scanning speotrometer for ultrasoft X-rays suitable for use in a n electron mioroprobe. Advan. X-Ray Anal. 9, 420-429. 16. Davidson, E., Hartwick, A. J., and Taylor, J. M. (1971). Expanded wavelength coverage with digitally oontrolled X-ray spectrometers. Microatruclures 2, 16-1 9. 16. Tousimis, A. J., and Nioolino, J. A. (1971). Universal X-ray detector: A flow proportional counter for analysis of all elements from beryllium to uranium. Proc. Nat. Conj. Electron Microprobe Anal., 6th, 1971. Pap. No. 60. 17. Rum, J. C. (1971). Progress in the dasign and application of energy dispersion X-ray analysers for the SEM. I n “Soanning Electron Microscopy/l971 (Proc. 4th Annual Scanning Electron Microscope Symposium)” (0.Jahari and I. Corvin, eds.), pp. 66-72. Illinois Inst. Technology, Chicago, Illinois. 18. Clayton, c. G. (1969). Application of radio isotope x - r a y fluorescence. Analysis in geologioal essay, mining and mineral processing. I n “Nuclear Techniques and Mineral Resources,” Pap. SM 112/120, pp. 293-324. IAEA, Vienna. 19. Shinode, G. (1969). Behaviour of electrons in a specimen. In “Electron Probe Microanalysis” (A. J. Tousimis and L. Marton, eds.), pp. 16-43. Academic Press, New York. 20, Dunoumb, P. (1963). Mioroenalysis with electron microscopy. In “Third International symposium on x - r a y Optics and X-ray Microanalysis” (H. H. Pattee, V. E.Cosslett, and A. Engstrom, eds.), pp. 431-439. Academic Press, New York. 21. Reid, L. R., and Smith, J. P. (1970). A method of electron microprobe signal enhancement by pulse rate analysis. Rev. Sci. Znatrum. 41, 26-28. 22. Smith, J. V. (1966). X-ray emission microanalysis of rock-forming minerals. I. Experimental techniques. J. Ueol. 7 8 , 830-864. 23. Cameron, E.N., and van Rensburg, W. C. J. (1966). Chemica1-mechanicalpolishing of ores. Econ. Ueol. 60, 630-632. 24. Taylor, C. M., and Radtke, A. S. (1966). Preparation and polishing of ores and mill products for microsoopio examination and electron mioroprobe analysis. Econ. Ueol. 60, 1306-1319. 26. Cddwell, D. E.,and Weiblen, P. W. (1966). Diamond disc preparation of polished thin seotions for electron microprobe analysis. Econ. Ueol. 60, 1320-1326. 26. Sinkankas, J. (1968). High pressure epoxy impregnation of porous materials for thin-seotion and mioroprobe analysis. Amer. Mineral. 63, 339-342. 27. Innes, R. P., and Pluth, D. J. (1970). Thin motion preparation using epoxy impregnation for petrographic end eleotron microprobe analysis. Soil Sci. SOC.Arner. PTOC. 8, 483-486. 28. Sweatman, T . R., and Long, 5 . V. P. (1969). Quantitative electron-probe microanalysis of rook-forming minerals. J. Petrol. 10, 332-379. 29. Kaelble, E.F. (ed.) (1967). “Handbook of X-rays.” MoGraw-Hill, New York. 30. Compton, A. H., and Allison, S. K. (1936). “X-rays in Theory and Experiment.” Van Nostrand-Reinhold, Prinoeton, New Jersey. 31. Sproull, W. T. (1946). “X-rays in Praotioe.” McGraw-Hill, New York. 32. Kramers, H. A. (1923). On the theory of X-ray absorption and the continuous X-ray spectrum. Phil. Mag. [a] 46, 836-871. 33. Reed, 5. J. B. (1969). Peak to background ratio in microprobe analysis. Proc. Znt. Congr. X-Ray Opt. Microanal., 5th. 1968, pp. 80-83. 34. Bertin, E. P. (1970). “Principles and Practice of X-ray Spectrometric Analysis.” Plenum, New York. 36. Clark, C. L. (1966). “Applied X-rays.” MoGraw-Hill, New York.

ELECTRON MICROPROBE ANALYSIS

145

36. Cork, J. M. (1971). Critical X-ray absorption energies in KeV. In “Handbook of Chemistry and Physics,” (R. C. Wedst, ed.) 5lst ed. pp. E-196. Chem. Rubber Publ. Co., Cleveland, Ohio. 37. White, E. W., and Johnson, G. G. (1970). X-ray emission and absorption edge wavelengths and interchange settings for LiF geared curved crystal spectrometer. Pa. State Univ., Earth and Miner. Sci. Exp. Sta., Spec. Publ. 1-70. 38. Reed, S. J. B. (1965). Characteristic fluorescence corrections in electron-probe microanalysis. Brit. J . Appl. Phys. 16, 913-926. 39. Arrhenius, G. (1966). Chemical bond effect on electron transitions between inner levels. I n “X-ray Optics and Microanalysis” (R. Castaing, P. Deschamps, and T. Philibert, eds.), pp. 328-332. Hermaxixi, Paris. 40. Kelly, T. K. (1966). Mass absorption coefficients and their relevance in electronprobe microanalysis. Inst. Mining Met., Trans., 5’ect.B 7 6 , B59-B73. 41. Bearden, J. A. (1967). X-ray wavelengths. Rev. Mod. Phya. 39,78-124. 42. White, E. W., and Johnson, G. G., Jr. (1970). “X-ray Emission and Absorption Wavelengths and Two-theta Tables,” 2nd ed., ASTM Data Ser. DS 37A. Am. SOC. Test. Mater., Philadelphia, Pennsylvania. 43. Fischer, D. W., and Baun, W. L. (1967). The influence of chemical cornbination and sample self-absorption on some long wavelength X-ray emission spectra. Norelco Rep. 14, 92-98. 44. Fischer, D. W., and Baun, W. L. (1968). Band structure and the titanium LII, 111 X-ray emission and absorption speotra, from pure metal, oxides, nitride, carbide, and boride. J. Appl. Phys. 38, 4757-4776. 45. Fischer, D. W. (1965). Changes in the soft X-ray L emission spectra with oxidation of the first series transition metals. J. Appl. Phya. 36, 2048-2053. 46. Fischer, D. W. (1969). Vanadium LII, 111 X-ray emission and absorption spectra from metal, oxides, nitride, carbide and boride. J . Appl. Phya. 40, 4151-4163. 47. Glen, G. L., and Dodd, C. G. (1968). Use of molecular orbital theory to interpret X-ray K-absorption spectral data. J. AppZ. Phya. 39,5372-5377. 48. Dodd, C. G., and Glen, G. L. (1968). Chemical bonding studies of silicates and oxides by X-ray K-emission spectroscopy. J . Appl. Phya. 39, 5377-5384. 49. Fischer, D. W. (1970). Molecular-orbital interpretation of the soft X-ray LII, X-ray emission and absorption spectra from some titanium and vanadium compounds. J . Appl. Phys. 41, 3561-3569. 50. Urch, D. S. (1970). The origin and intensities of low energy satellite lines in X-ray emission spectra : A molecular orbital interpretation. Proc. Phys, SOC.,London (Solid State Phys.) [2] 3, 1276-1291. 51. Smith, D. G. W., and O’Nions, R. K. (1972). Investigations of bonding by oxygen KOL emission spectroscopy: further evidence concerning the true character of the oxygen Ka emission band. Chem. Qeol. 19, 145-146. 52. Castaing, R., and Descamps, J. (1955). Sur lea bases physiques de l’analyse ponctuelle par spectrographie X. J . Phya. Radium 16, 304-317. 53. Castaing, R., and Henoc, J. (1966). RBpartition en profondeur du rayonnement caract6ristique. In “X-ray Optics and Microanalysis” (R. Cestaing, P., Deschamps, and T. Philibert, eds.), pp. 120-126. H e r m n n , Paris. 64. Shimizu, R., Murata, K., and Shinoda, G. (1966). Depth distribution of characteristic X-ray in microanalyser target. I n “X-ray Optics and Microanalysis” (R. Castaing, P. Deschamps, and T.Philibert, eds.), pp. 127-138. Hermam, Paris. 55. Schmitz, U., Ryder, P. L., and Pitsch, W. (1969). A n experimental method for determining the depth distribution of characteristic X-rays in electron microprobe specimens. Proc. Int. Congr. X-Ray Opt. Microanal., 5th, 1968, pp. 104-113.

146

D. 0 . W. SMITH AND J. C. RUCKLIDOE

66. Bishop, H. E. (1968). The absorption and atomic number corrections in electron-

probe X-ray microenelysis. Brit. J . Appl. Phys. [2] 1, 673-684. 57. Duncumb, P. (1971). Quantitative electron probe microanalysis. In “Proceedings of the 26th Anniversary Meeting EMAG Institute of Physics” (W. C. Nixon, ed.), pp. 132-137. Institute of Physics, London and Bristol. 68. von Bethe, H. (1930). Zur Theorie des Durchgangs schneller Korpuskularstrahlen durch Materie. Ann. Phye. (Leiprig) [6], 325-400. 69. Harris, L. A. (1968). Analysis of materials by electron-excited Auger electrons. J . Appl, Phy8. 89, 1419-1427. 60. Weber, R. E., and Peria, W. T. (1967). Use of LEED apparatus for the detection and identification of surfece contaminants. J . Appl. Phys. 88,4366-4368. 61. Burhop, E. H. S. (1952). “The Auger effect.” Cambridge Univ. Press, London and New York. 62. Ureen, M. (1962). The efficiency of production of characteristic X-radiation. Ph.D. Thesis Cavendish Lab., Cambridge University, England. 63. Burhop, E . H. S. (1966). Le rendement de fluorescence. J. Phye. Radium 16, 625-629. 64. Hagedoorn, H. L., and Wapstre, A. H. (1960). Measurements of the fluorescent yield of the K-shell with a proportional counter. Nucl. Phya. 15, 146-151. 66. Martin, P. M., and Poole, D. M. (1971). Electron-probe microanalysis: The relation between intensity ratio and concentration. Metals Mater. 5 (Metall. Rev. 16, 19-46, Rev. No. 160). 66. Fink, R. W., Jopson, R. C., Mark, H., and Swift, C. D. (1966). Atomic fluorescence yields. Rev. Mod. Phys. 88, 513-640. 67. O’Nions, R. K., and Smith, D. U.W. (1971). Bonding in SiOa and Fez03 by oxygen K a X-ray emission spectre. Nature (London),Phye. Sci. 231, 130-133. 678. Liefeld, R. J., HanzBly, S., Kirby, T. B., and Mott, D. (1970). X-ray spectrometric properties of potassium acid phthalate crystals. Advan. X-Ray Anal. 13, 373-381. 68. Springer, 0.(1967). The correction for “continuous fluorescence’’ in electronprobe microanalysis. Neues Jahrb. Mineral., Abh. 106, 241-256. 69. HBnoc, J. (1968). Fluorescence excited by the continuum. Nat. Bur. Stand. (U.S.), Spec. Publ. 298, 197-214. 70. Casteing, R. (1960). Electron probe microanalysis. Advan. Electron. Electron Phy8. 18, 317-386. 7 1 . Beaman, D. R., Isasi, J. A., and Lewis, R. (1969). The source and nature of deadtime in electron microprobe analysis. Proc. Nat. Conj. Electron. Microprobe Anal., 4th, 1969. Pap. No. 13. 72. Beaman, D. R., Lewis, R., and Isasi, J. A. (1969). A study of the deadtime correction in electron probe microanalysis. Proc. Int. Congr. X-Ray Opt. Microanal., 5th, 1968, pp. 84-92. 73. Beaman, D. R., Isasi, J. A.. end Birnbaum, H. K. (1970). The interaction of selfprolonging and nonselfprolonging counting components. Proc. Nat. Conj. Electron Miwoprobe Anal., 5th. 1970. Pap. No. 46. 74. Short, M. A. (1966). Detection and correction of non-linearity in X-ray counting systems. Proc. Conf. X-Ray Anal. Methods, 5th, 1966, pp. 60-62. 76. Smith, J. V. (1966). X-ray emission microanalysis of rock-forming minerals. 11. Olivines. J . Ueol. 74 1-16. 76. Smith, J. V. (1966). X-ray emission microanalysis of rock-forming minerals. VI. Clinopyroxenes near the diopside-hedenbergite join. J. Ueol. 74, 463-477.

ELECTRON MICROPROBE ANALYSIS

147

77. Smith, J. V. and Ribbe, P. H. (1966). X-ray emission microanalysis of rock-forming minerals. 111.Alkali felspars. J . Geol. 74, 197-216. 78. Ribbe, P. H. and Smith, J. V. (1966). X-ray emission microanalysis of rockformingminerals. IV. Plagioclase felspars. J. Geol. 74, 217-233. 79. Howie, R. A., and Smith, J. V. (1966). X-ray emission microanalysis of rockforming minerals. V. Orthopyroxenes. J. Geol. 74, 443-462. 80. Knowles, C. R., Smith, J. V., Bence, A. E., and Albee, A. L. (1969). X-ray emission microanalysis of rock-forming minerals. VII. Garnets. J . Geol. 77, 439-451. 81. Rucklidge, J. C., Gasparrini, E., Smith, J. V., and Knowles, C. R. (1971). X-ray emission microanalysis of rock-forming minerals. VIII. Amphiboles. Can. J . Earth Sci. 8 1171-1183. 82. Ziebold, T. 0. and Ogilvie, R. E. (1963). Quantitative analysis with the electron microanalyser. Anal. Chem. 35, 621-627. 83. Ziebold, T. O., and Ogilvie, R. E. (1964). An empirical method for electron microanalysis. Anal Chem. 36, 322-327. 84. Ziebold, T. O., and Ogilvie, R. E. (1966). Correlations of empirical calibration for electron microanalysis. I n “The Electron Microprobe” (L. McKinley, K. F. J. Heinrich, and D. B. Wittry eds.) pp. 378-389. Wiley New York. 85. Lachence, G. R., and Traill, R. J. (1966). A practical solution to the matrix problem in X-ray analysis. Part 1. Method. Can. Spectroac. 11, 43-62. 86. Traill, R. J., and Lachance, G. R. (1966). A practical solution to the matrix problem in X-ray analysis. Part 2. Application to a multi-component alloy system. c a n . SpeCtTO8C. 11, 63-7 1. 87. Bence, A. E., and Albee, A. L. (1968). Empirical correction factors for the electron microanalysis of silicates and oxides. J. Geol. 76, 382-403. 88. Albee, A. L. and Ray, L. A. (1970). Correction factors for electron probe microanalysis of silicates, oxides, carbonates, phosphates, and sulphates. Anal. Chem. 42. 1408-1414. 89. Campbell, W. J., and Brown, J. D. (1968). X-ray absorption and emission. Anal. Chem. 40, 346R375R. 90. Beeman, D. R., and Isasi, J. A. (1970). A critical examination of computer programs used in quantitative electron probe microanalysis. Anal. Chem. 42, 15401568. 91. Dmcumb, P., and Reed, S. B. J. (1968). The calculation of stopping power and backscatter effects in electron probe microanalysis. Nat. Bur. Stand. (U.S.), Spec. Publ. 298, 133-154. 92. Green, M. (1964). The angular distribution of characteristic X-radiation and its origin within a solid target. Proc. Phy8. SOC.,London 85, 435-451. 93. Philibert, J. (1963). A method for calculating the absorption correction in electronprobe microanalysis. In “Third International Symposium on X-ray Optics and X-ray Microanalysis” (H. H. Pattee, V. E. Coslett, and A. Engstrom, eds.), pp. 379-392. Academic Press, New York. 94. Duncumb, P., and Shields, P. K. (1966). The effect of excitation potential on the absorption correction. In “The Electron Microprobe” (L. McKinley, K. T. J. Heinrich, and D. B. Wittry, eds.), pp. 284-295. Wiley, New York. 95. Heinrich, K. F. J. (1967). The absorption correction model for microanalysis. Proc. Nat. Conf. Electron Microprobe Anal., Znd, 1967 Pap. No. 7. 96. Duncumb, P., Shields-Mason, P. K., and da Case, C. (1969). Accuracy of atomic number and absorption corrections in electron probe microanalysis. Proc. Int. Coragr. X-Ray Opt. Microanal., 5th, 1968, pp. 146-160.

148

D. 0.W. SMITH AND J. C. RUCELIDOE

97. Heinrich, K. F. J. (1966). X-ray absorption uncertainty. I n “The Electron Microprobe” (L. McKinley, K. T. J. Heinrich, and B. B. Wittry, eds.), pp. 297-377.

Wiley, New York. 98. Dewey, R. D., Mapes, R. S., and Reynolds, T. W. (1969). Tables of X-ray data. Progr. Nucl. Energy, Ser. I X 9, 29-353. 99. Frazer, J. Z. (1967). “A Computer Fit to Mass Absorption Coefficient Data,” Publ. No. 67-29. Inst. for the Study of Matter, Univ. of Cplifornia, La Jolls. 100. Theisen, R. (1965). “ Quantitative Electron Microprobe Analysis.” Springer-Verlag,

Berlin and New York. 101. Theisen, R., and Vollath, D. (1967). “Tabellen der Massenschwiiohungskoeffizienten

von Rontgenstrahlen.” Verleg Stahleisen, Dusseldorf. 102. Bearden, A. J. (1966). X-ray photoeffect cross sections in low- and medium-2 absorbers for the energy range 852 eV t o 4OkeV. J . AppZ. Phys. 87, 1681-1692. 103. Hughes, 0. D., Woodhouse, J. B., and Bucklow, I. A. (1968). The determination of some X-ray mass absorption coefficients. Brit. J . Appl. Phya. [2] 1, 695-706. 104. Carter, R. W., Rohrer, R. H., and Dyer, G. R. (1967). Low energy X-ray mass absorption coefficients from 1.48 to 16.77 KeV for scandium, titanium, vanadium, iron, oobalt, nickel and zinc. Health Phys. 18, 593-599. 106. Ortner, B., Ebel, H., and Lihl, F. (1970). Rontgenabsorptionskoeffizienten von Lanthanoiden. Mikrochim. Acta, Suppl. 1,270-279. 106. Vollath, D. (1968). Massenschwilchungskoeffizientenfur langwellige Rontgenstrahlen. Mikrochim. Acta, Suppl. 8, 11-18. 107. Henke, B. L., and Elgin, R. L. (1970). X-ray absorption tables for the 2 to 200 A region. Adwan. X-Ray Anal. 18, 639-666. 108. Theisen, R., Togel, K., and Vollath, D. (1967). Massenschwilchungskoeffizienten von Rontgenlinien. Mikrochina. Acta, Suppl. 2 , 16-24. 109. Springer, G. (1972). Fluorescence by continuous radiation in multi-element targets. Proc. Int. Cong. X-Ray Opt. Microanal., 6th, 19Y1, pp. 141-146. 110. Frazer, J. 2.. Fujita, H., and Fitzgereld, R. W. (1971). A new procedure for quantitative X-ray analysis with electron excitation. Muter. Res. Bull. 6 , 71 1-724. 111. Beaman, D. R. (1967). Evaluation of correction procedures used in electron probe microanalysis with emphasis on atomic number interval 13-33. Anal. Chena. 89, 418-426. 112. Wolf, R. U., and Macres, V. C. (1969). Quantitative microprobe analysis: A basis

for universal atomic number correction tables. I n “Electron Probe Microanalysis” (A. J. Tousimis and L. Marton, eds.), pp. 73-116. Academic Press, New York. 113. Brown, J. D. (1966). Ph.D. Thesis, University of Maryland. 114. Duncomb, P., and Jones, E. M. (1969). Electron probe microanalysis: an easy-to-use computer program for correcting quantatitive data. Company report. Tube Investments Research Laboratories, Hinxton Hall, Nr. Saffron Wdden, Essex, United Kingdom. 116. Boyd, F. R., Finger, L. W., and Chayes, F. (1968). Computer reduction of electronprobe data. Carnegie Inst. Wuah., Yearb. 67, 210-216. 116. Mason, P. K., Frost, M. T., and Reed, S. J. B. (1969). Computer programs for calculating corrections in quantitative X-raymicroanalysis. IMS Report No. 2. National Physical Laboratory, Teddington, Middlesex, United Kingdom. 117. Rucklidge, J. C., and Gasparrini, E. L. (1969). EMPADR VII. Geology Dept., Univ. of Toronto, Toronto. 118. Smith,D.G.W.,andTomlinson,M.C. (1970).AnAPLlanguagecomputerprogram for use in electron microprobe analysis. Kana.State Qeol.Sum., Computer Contrib. No. 45.

ELECTRON MICROPROBE ANALYSIS

149

119. Springer, G. (1967). Die Berechnung von Korrecturen fur die quantitative Elektronenstrahl-Mikroanalyse Fortshr. Mineral. 45, 103-124. 120. Chodos, A. A., and Albee, A. L. (1972). Automated microprobe analysis and applications to analysis of exotic lunar phases. Proc. Int. Cong. X-Ray Opt. Microanal., 6th, 1971, pp. 779-784. 121. Jefferies, B., and Long, J. V. P. (1971). An automated electron-probe system for quantitative mineral analysis. I n “Proceedings of the 25th Anniversary Meeting EMAG Institute of Physics” (W. C. Nixon, ed.), pp. 150-153. Institute of Physics, London and Bristol. 122. Heinrich, K. F. J., and Yakowitz, H. (1969). Propagation of errors in correction models for quantitative electron probe microanalysis. Proc. Int. Congr. X - R a y Opt. Microanal., 5th. 1968 pp. 151-159. 123. Liebhafsky, H., Pfeiffer, H. G., and Zemany, P. D. (1955). Precision in X-ray emission spectography. A d . Chem. 27, 1257-1258. 124. Killingworth, P. J. (1971). The dependence of analytical accuracy upon X-ray spectrometer geometry. Proc. Nat. Conf. Electron. Microprobe Anal., 6th, 1971 Pap. No. 63. 125. Smith, J. P., and Pedigo, J. E . (1968). Effects of specimen repositioning on statistics of X-ray intensity measurements from a n electron microprobe analyzer. Anal. Chem. 40, 2028-2031. 126. Taylor, C. M., Tousimis, A. J., and Nicolino, J. A. (1971). Specimen contamination during electron probe analysis. Proc. Nat. Conf. Electron Microprobe Anal., 6th, 1971 Pap. No. 33. 127. Campbell, A. J., and Gibbons, R. (1966). Specimen contamination in the electron microprobe. I n “The Electron Microprobe” (L. McKinley, K. F. J. Heinrich, and D. B. Withry, eds.), pp. 75-82. Wiley, New York. 128. Castaing, R., and Deschamps, J. (1954). S u r la contamination des Qchantillons dans le microanalyseur B sonde Blectronique. C . R . Acad. Sci. 238, 1506-1508. 129. Mulvey, T. (1960). A new X-ray a n a l p r . I n “X-ray Microscopy and X-ray Microanalysis” (A. Engstrom, V. E. Cosslett, and H. H. Pattee, eds.), pp. 372-378. Elsevier, Amsterdam. 130. Ong, P. S. (1966). Reducing carbon contamination in an electron microprobe and measuring low energyback scatteredelectrons. I n “X-ray Opticsand Microanalysis” (R. Castaing, P. Deschamps, and T. Philibert, eds.), pp. 181-192. Hermann, Paris. 131. Duncumb, P., and Melford, D. A. (1966). Quantitative applications of ultra-soft X-ray microanalysis in metdlurgical problems. I n “ X-ray Optics and Microanalysis” (P. Castsing, P. Deschamps, and T. Philibert, eds.), pp. 240-253. Herm m n , Paris. 132. Lineweaver, J. L. (1963). Oxygen outgassing caused by electron bombardment of glass. J . Appl. Phys. 34, 1786-1791. 133. Borom, M. P., and Hanneman, R. E. (1967). Local compositional changes in alkali silicate glasses during electron microprobe analysis. J. Appl. Phys. 38, 2406-2407. 134. Vessamillet, L. F., and Caldwell, V. E. (1969). Electron-probe microanalysis of alkali metals in glasses. J . Appl. Phys. 40, 1637-1643. 135. Scholes, S., and Wilkinson, F. C. F. (1969). Specimen damage during microprobe analysis of silicate glasses. Proc. Int. Congr. X-Ray Opt. Microanal., 5th, 1968, pp. 438-442. 136. McConnell, J. D. C. (1969). Photochemical degradation of a silicate in the beam of the electron microscope. Phil. Mag. [8] 20, 1195-1202.

150

D. 0.W. SMITH AND J. C. RUCKLIDOE

137. Rucklidge, J. C., and Stumpfl, E. F. (1968). Changes in the composition of petzite (Ag, AuTe2) during analysis by electron probe. Neuea Jahrb. Mineral., Monatah. pp. 61-68. 138. Almasi, G. S., Blair, J., Ogilvie, R. E., and Schwartz, R. J. (1966). A heat-flow problem in electron-beam microprobe analysis. J . Appl. p h y s . 86, 1848-1864. 139. Yakowitz, H., and Heinrich, K. F. J. (1968). Quantitative electron probe microanalysis : absorption correction uncertainty. Mikrochim. Acta pp. 182-200. 140. Kelly, C. J., Short, M. A., and Reynolds, C. A. (1971). Pulse height stabilization in gas proportional X-ray detector systems. Proc. Nat. Conf. Electron Microprobe Anal., 6th. 1971 Pap. No. 63. 141. Carmichael, I. S. A. (1967). The iron titanium oxides of salic volcanic rocks and their associated ferromagnesian silicates. Contrib. Mineral. Petrol. 14, 36-64. 142. Snetsinger, K. G . , Bunch, T. E., and Keil, K. (1968). Electron microprobe analysis of vanadium in the presence of titanium. Amer. Mineral. 68, 1770-1773. 143. Reed, S. J. B., and Long, J. V. P. (1963). Electron-probe measurements near phase boundaries. I n “Third International Symposium on X-ray Optics and X-ray Microanalysis’’ (H. H. Pattee, V. E. Cosslett, and A. Engstrom, eds.), pp. 317-327 Academic Press, New York. 144. Dils, R. R., Zeitz, L.,and Huggins, R. A. (1963). A suggested secondary fluorescence correction technique for electron-probe analysis in the vicinity of a steep concentration gradient. I n “Third International Symposium on X-ray Optics and X-ray Microanalysis” (H. H. Pattee, V. E. Cosslett, and A. Engstrom, eds.), pp. 341-360. Academic Press, New York. 146. Samuels, L. E. (1967). “Metallographic Polishing by Mechanical Means.” Pitman, London. 146. Fleischer, M., and Stevens, R. E . (1962). Summary of new data on rock samples G-1 and W-1. Geochim. Cosmochim. Acta 26, 626-643. 147. Fleischer, M. (1966). Summary of new data on rock samples G-1 and W-1, 19621966. Geochim. Cosmochim. Acta 29, 1263-1283. 148. Beaman, D. R. (1971). The accuracy of quantitative microanalysis using energy dispersive spectrometers. Proc. Nat. Conf. Electron Microprobe Anal., 6th, 1971 Pap. No. 2. 149. Heinrich, K. F. J. (1968). Common sources of error in electron probe microanalysis. Advan. X - R a y Anal. 11, 40-66. 160. Brown, D. B., Wittry, D. B., and Kyser, D. F. (1969). Prediction of X-ray production and electron scattering in electron-probe analysis using a transport equation. J . Appl. Phya. 40, 1627-1636. 16Oa. Caldwell, D. 0. (1966). Range-energy relation and masses of the new particles. Phys. Rev. 100, 291-294. 16Ob. Berger, M. J., and Seltzer, S. M. (1964). Tables of energy losses and ranges of electrons and positions. Nat. Acad. Sci.-Nat. Rea. Counc., Publ. 1133, 206-268. 161. Shiraiwa, T., and Fujino, N. (1969). Quantitative analysis of oxygen in electron probe microanalysis. PTOC. Int. Congr. X - R a y Opt. Microanal., 5th, 1968 pp. 366-368. 162. Springer, G. ( 1969). Mineralogical applications of the electron probe microanalyser. Proc. Int. Congr. X - R a y Opt. Microanal., 5th, 1968 pp. 424-431. 163. Meed, C. W. (1969). Electron probe microanalysis in mineralogy. I n “Electron Probe Microanalysis” (A. J. Tousimis and L. Marton, eds.), pp. 227-244. Academic Press, New York. 164. Goldstein, J. I. (1969). Electron probe enalysis in metallurgy. I n “Electron Probe Microanalysis” (A. J. Tousimis and L. Marton, eds.), pp. 245-290. Academic Press, New York.

ELECTRON MICROPROBE ANdLYSIS

151

155. Schwander, H., and Wenk, 0. E. (1965). Monazit als Kern pleochroitischer Hofe in Biotiten der Tessiner Gneisse. Schweiz. Mineral Petrogr. Mitt. 45, 797-817. 155a. Hollister, L. S., and Hargraves, R. B. (1970). Compositional zoning and its significance in pyroxenes from two coarse grained Apollo 11 samples. Ueochim. Cosmochim. Acta Proc. Apollo 11 Lunar Sci. Conf. 1, 541-550. 156. Rucklidge, J. C. (1971). A study of the redistribution of nickel in the serpentinization of olivine. Proc. Int. Cong. X-Ray Opt. Microanal., 6th, 1971, pp. 743-747. 157. Rucklidge, J. C. (1972). Chlorine in partially serpentinised dunite. Econ. Ueol. 67, 38-40. 158. Baum, T., and Lewis, R. (1970). An automatic color recording technique for electron

probe X-ray images. Proc. Nat. Conf. Electron Microprobe Anal., 5th, 1970 Pap. No. 39. 159. Stumpfl, E. F. (1961). Some new pletinoid-rich minerals, identified with the electron microanalyser. Mineral. Mag. 32, 833-847. 160. Rucklidge, J. C. (1969). Electron microprobe investigations of platinum metal minerals from Ontario. Can. Mineral. 9, 617-628. 161. Roedder, E., and Dwornik, E. J. (1968). Sphalerite color banding: Lack of correla-

tion with iron content, Pine Point, Northwest Territories, Canada. Amer. Mineral. 58, 1523-1529. 162. Hollander, N. B. (1968). Electron microprobe analyses of spinels and their alteration product from M h a r p and Taberg, Sweden. Amer. Mineral. 53, 1918-1928. 163. Beeson, M. H., and Jackson, E. D. (1969). Chemical composition of altered chromites from the Stillwater complex, Montana. Amer. Mineral. 54,1084-1100. 164. Treub, L. F., and de Wys, E. C. (1969). Carbonado, natural polycrystalline diamond. Science 165, 799-802. 165. Vinogradov, A. P., Udovykin, G. P., I l k , N. P., and Loseva, L. Ye. (1968). Struc-

166.

167. 168. 169.

ture of diamond-graphite intergrowths in ureilites and the origin of ureilites. Ueochem. Int. 5 , 756-769. Richardson, K. A., McKay, D. S., Greenwood, W. R., and Foss, T. € (1970). I. Alpha particle activity of Apollo 11 samples. Ueochim. Cosmochim. Acta, Proc. Apollo 11 Lunar Sci. Conf. 1, 763-767. Burns, R. G., and Fuerstenau, D. W. (1966). Electron-probe determination of interelement relationships in manganese nodules. Amer. Mineral. 51, 896-902. Katz, A. (1971). Zoned dolomite crystals. J. CJeol. 79, 38-51. Halbach, P. (1968). Zum Gehalt von Phosphor und anderen Spurenelementen in Brauneisenerzooiden aus dem Friinkischen Dogger bete. Contrib. Mineral. Petrol.

18, 241-251. 170. Weber, J. N. (1969). The incorporation of Mg into the skeletal calcites of echinoderms. Amer. J. Sci. 267, 537-544. 171. Corlett, M., and Ribbe, P. H. (1967). Electron probe microanalysis of minor elements in plagioclase feldspars. Schweiz. Mineral. Petrogr. Mitt. 47, 317-332. 172. Albee, A. L., and Chodos, A. A. (1969). Minor element content of coexistent Al, Si05 polymorphs. Amer. J. Sci. 267, 310416. 173. White, E. W., and White, W. B. (1967). Electronmicroprobeandopticalabsorption study of coloured kyanites. Science 158, 916-917. 174. Long, J. V. P., and Agrell, S. 0. (1965). The cathodo-luminescence of minerals in thin section. Mineral. Mag. 34, 318-326. 176. Goni, J., and RBmond, G. (1969). Localization and distribution of impurities in blende by cethodo-luminescence. Mineral. Mag. 37, 153-155. 176. Sippel, R . F. (1968). Sandstone petrology, evidence from luminescence petrography. J. Sediment. Petrol. 38, 530-554.

162

D. 0.W. SMITH AND J. C. RUCKLIDOE

177. Davey, J. P. (1966). A scanning electron microprobe system for the study of cathodo-luminescence. I n “X-ray Optics and Microanalysis” (R. Castaing, P. Desohemps, and R. Philibert, eds.), pp. 566-571. Hermann, Paris. 178. Schorer, G. (1970). Senduhrbau und Optik von Titanaugiten alkalibasaltischer Gesteine des Vogelsberges. Neuea Jahrb. Mineral., Monatah. pp. 310-325. 179. Hollister, L. S. (1970). Origin, mechanism and consequences of compositional sector-zoning in staurolite. Amer. Mineral. 55, 742-766. 180. Evans, B. W., and Moore, J. 0. (1968). Mineralogy as a function of depth in the prehistoric Makaopuhi tholeiitic lava lake, Hawaii. Contrib. Mineral. Petrol. 17, 85-116. 181. Ringwood, A. E., and Essene, E.(1970). Petrogenesis of Apollo 11 basalts, internal constitution and origin of the moon. Geochim. Cosmochim. Acta, Proc. Apollo I 1 Lunar Sci. Conf. 1, 769-799. 182. Akimoto, S., Nishikawa, M., Nakamura, Y., Kushiro, I., and Katsura, T. (1970). Melting experiments of lunar crystalline rocks. Qeochim. Cosmochim. Acta, Proc. Apollo I1 Lunar Sci. Conf. 1, 129-133. 183. Roeder, P. L., and Emslie, R. F. (1970). Olivine liquid equilibrium. Contrib. Mineral. Petrol. 29, 275-289. 184. Brown, E. H. (1967). The greenschist fecies in part of eastern Otago, New Zealand. Contrib. Mineral. Petrol. 14, 269-292. 186. Atherton, M. P. (1968). The variation in garnet, biotite and chlorite composition in medium grade pelitic rocks from the Dalradian, Scotland, with pctrticular reference to the zonation in garnet. Contrib. Mineral. Petrol. 18, 347-371. 186. Saxena, S. K. (1968). Distribution of elements between coexisting minerals and the nature of solid solution in garnet. Amer. Mineral. 51 994-1014. 187. Evens, B. W. (1969). Chlorine and fluorine in micas of pelitic schists from the sillimanite-orthoclase isograd, Maine. Amer. Mineral. 64, 1209-121 1. 188. Smith, D. 0. W. (1965). The chemistry and mineralogy of some emery-like rocks from Sithean Sluaigh, Strachur, Argyllshire. Amer. Mineral. 50, 1982-2022. 189. Klein, C. J. (1968). Two-amphibole assemblages in the system ectinolite-hornblendeglaucophane. Amer. Mineral. 54, 212-237. 190. Kisch, H. J. and Warnsers, F. W. (1969). Distribution of Mg and Fe in cummingtonite-hornblende and cummingtonite-actinolitepairs from metamorphic assemblages. Contrib. Mineral. Petrol. 24, 245-267. 191. Evans, B. W. (1964). Coexisting albite and oligoclase in some schists from New Zealand. Amer. Mineral. 49 173-179. 192. Nissen, H.-U. (1968). A study of bytownites in amphibolites of the Ivrea-zone (Italian Alps) and in anorthosites: A new unmixing gap in the low plagioclases. Schweiz. Mineral. Petrogr. Mitt. 48, 53-55. 193. Bottinga, Y.,Kudo, A., and Weill, D. (1966). Some observations on oscillatory zoning and crystallization of magmatic plagioclase. Amer. Minerol. 61,792-806. 194. Adems, J. B. (1968). Differential solution of plagioclase in supercritical water. Amer. Mineral. 53, 1603-1613. 195. Anderson, A. T., Jr. (1966). Mineralogy of the Labrieville anorthosite, Quebec. Amer. Mineral. 51, 1671-1711. 196. Van Schmus, W. R., and Ribbe, P. H. (1968). The composition and structural state of feldspar from chondritic meteorites. Geochim. Comochim. Acta 82, 13271342. 197. Fredriksson, K., and Reid, A. M. (1967). Meteorite investigations by electron microprobe techniques. Ree. Qeochem. 2, 143-169.

ELECTRON MICROPROBE ANlLLYSIS

153

198. O’Nions, R. K., Smith, D. G. W., Baadsgaard, H., and Morton, R. D. (1969). Influence of chemical composition on argon retentivity in metamorphic calcic amphiboles from south Norway. Earth Planet. Sci. Lett. 5 , 339-345. 199. Amaral, G., Bushee, J., Cordeni, U. G., Kawashita, K., and Reynolds, J. H. (1967). Potassium-argon ages of alkaline rocks from southern Brazil. Geochim. Comochim. Acta 31, 117-142. 200. Willgallis, A. ( 1970). Zur Mikrosondenanalyse der U-Th-minerale rim Malsburger Granit. Neues Jahrb. Mineral., Abh. 114, 48-60. 201. Ziihringer, J. (1966). Primordial helium detection by microprobe technique. Earth Planet. Sci. Lett. 1, 20-22. 202. Schmidt, R. A., and Keil, K. (1966). Electron microprobe study of spherules from Atlantic Ocean sediments. Geochim. Cosmochim. Acta 30, 471-478. 203. Marvin, U. B., and Einaudi, M. T. (1967). Black, magnetic spherules from Pleistocene and Recent beach sands. Geochim. Cosmochim. Acta 31, 1871-1884. 204. E l Goresy, A. (1968). Electron microprobe analysis and ore microscopic study of magnetic spherules and grains collected from the Greenland ice. Contrib. Mineral. Petrol. 17, 331-346. 206. Jedwab, J. (1970). Les spherules cosmiques dans les nodules de manganbse. Beochim. Cosmochim. Acta 34, 447-457. 206. Roedder, E., and Wieblen, P. W. (1970).Lunar petrology of silicate melt inclusions. Beochim. Cosmochim. Acta, Proc. Apollo 11 Lunar Sci. Conf. 1, 801-837. 207. Van Schmus, W. R. (1967). Polymict structure of the Mezo-Madams chondrite. Beochim. Comochim. Acta 31, 2027-2042. 208. Bunch, T. E., and Fuchs, L. H. (1969). A new mineral: Brezinaite, Cr, S4, and the Tucson meteorite. Amer. Mineral. 54, 1509-1518. 209. Walter, L. S., end Cassidy, W. (1967). Compositional variations and trends in individual tektites. Abstr. Geol. SOC.Amer. Annu. Meet., New Orleans pp. 231. 210. Smith, D. G. W., and Westgate, J. A. (1969). Electron probe technique for characterizing pyroclastic deposits. Earth Planet. Sci. Lett. 5 , 313-319. 211. Hay, R. L., and Iijima, A. (1968). Petrology of palagonite tuffs of Koko Craters, Oahu, Hawaii. Contrib. Mineral. Petrol. 17, 141-164. 212. Switzer, G. S.,and Melsom, W. G. (1970). Origin and composition of rock fulgarite glass. Abstr. IMA-IAGOD Meet., Tokyo-Kyotop. 153. 213. Kleinmann, B. (1969). The breakdown of zircon observed in the Libyan desert glass as evidenoe of its impact origin. Earth Planet. Sci. Lett. S, 497-501. 214. Yang, H., Hooke, R. LeB., and Weiblen, P. W. (1967). Desert varnish. An electron probe study. Abatr. Geol. SOC.Amer. Annu. Meet., New Orleans p. 243. 215. Gulson, B. L., and Lovering, J. F. (1968). Rock analysis using the electron probe. Beochim. Coamochim. Acta 32, 119-122. 216. Norrish, K. N., and Chappell, B. W. (1967). X-ray fluorescence spectrography. I n “Physical Methods in Determinative Mineralogy” (J.Zussman, ed.), pp. 161-214. Academic Press, New York. 217. Reed, S. J. B. (1970). The analysis of rocks in the electron probe. Geochim. Coamochim. Acta 34, 416-421. 218. Ruoklidge, J. C., Gibb, F. G. F., Fawcett, J. J., and Gasparrini, E. L. (1970). Rapid rock analysis by electron probe. Geochim. Coamochim.Acta 84, 243-247. 219. Keil, K. (1966). Mineralogioal modal analysis with the electron microprobe X-ray analyzer. Amer. Mineral. 50, 2089-2092. 220. Day, D. E. (1963). Determining the coordination number of aluminium ions by X-ray spectrosoopy. Nature (London) 200, 649-651.

154

D. 0 . W. SMITH AND J. C. RUCKLIDOE

221. White, E. W., and Gibbs, G. V. (1967). Structural and chemical effects on the Si KP X-ray line for silioates. Amer. Mineral. 52,986-993. 222. White, E. W., and Gibbs, G . V. (1969). Structural and chemical effects on the A1 KP X-ray emission band among aluminium containing silicates and aluminium oxides. Amer. M i ~ ~ a54, l . 931-936. 223. Lovering, J. F., and Widdowson, J. R. (1968). Electron microprobe determination of sulphur coordination in miner8lS. Gthos 1, 264-267. 224. O’Niona, R. K., and Smith, D. G. W. (1971). Investigations of the LII, 111 X-ray emission speotra of Fe by electron mioroprobe. Part 2. The Fe LII, 111 spectra of Fe and Fe-Ti oxides. Amer. Minerd. 66, 1462-1463. 226. Dodd, C. G., and Glen, 0.L. (1969). A survey of chemical bonding in silicate minerals by X-ray emission spectroscopy. Amer. Mineral. 54, 1299-131 1. 226. Dodd, C. G., and Glen, a. L. (1970). Studies of chemical bonding in glasses by X-ray emission spectroscopy. J. Amer. Ceram. Soc. 58, 322-326. 227. Albee, A. L., and Chodos, A. A. (1970). Semiquantitative electron microprobe determination of Fea /Fes and Mna /Mn3 in oxides and silicates and its application t o petrologic problems. Amer. Mineral. 55, 491-601. 228. Andersen, C. A. (1967). The quality of X-ray mioroanalysis in the ultra-soft X-ray region. Brit. J. Appl. Phye. 18, 1033-1043. 229. Smith, D. G. W., and O’Nions, R. K. (1970). Some problem in the determination of valence state of iron using Fe LII, 111 X-ray emission spectra. Proc. Nat. Conf. Electron Microprobe A n d . , 5th, 1970 Pap. No. 16. 230. Smith, D. 0.W., and O’Nions, R. K. (1971). Investigations of the LII, 111 X-ray emission spectra of Fe by the electron microprobe. Part 1. Some aspects of the Fe LII, 111 spectra from metallic iron and hsematite. J . Phya. Sect. D . , Appl. Phya., 4, 147-169. 231. Levinson, A. A., Ed. (1971).“Proceedings of the Second Lunar Science Conference,” Vol. 1. MIT Press, Cambridge, Massechusetts. +

+

+

+

AU RORAL AU DIBI LlTY S. M. Silverman A i r Force Cambridge Research Laboratories. L G Hanscom Field. Bedford. Massachusetts

. .

and

..

T F Tuan Physics Department. University of Cincinnati. Cincinnati. Ohio

.

Page

........................................................ ................................................ ......................

1 Introduction 2. Observational Results 2.1. Methodology and the Use of Anecdotal Data 2.2. Observations by Professionally Trained Observers 2.3. Surveys 2.4. Individual Observations 3 Characteristics and Analysis of Auroral Sound Events 3.1. Frequency of Occurrence 3.2. Geographic Extent 3.3. Latitudinal Dependency 3.4. Semonal Dependency 3.6. Diurnal Variation 3.6. Sunspot Cycle Dependency 3.7. Correlation with Magnetic Activity 3.8. Auroral Characteristics 3.9. Localization of Effect 3.10. Weather 3.11. Effect of Altitude and Terrain 3.12. Association with Low Aurora 3.13. Association with Odor 3.14. Audibility While the Aurora Was Not Visible 3.16. Effectson Anknals 3.16. Auroral Sounds in Poetry 3.17. Possibly Related Phenomena-Sounds from Lightning and Meteors 3.18. Miscellaneous 4 Hypotheses of Auroral Audibility

.................. ........................................................ ........................................ .................... ...................................... .............................................. ........................................ ............................................ .............................................. ...................................... ................................ .......................................... .......................................... ...................................................... .................................. .................................... .......................................... .................... ..............................................

.

......................................

.

4.1. 4.2. 4.3. 4.4. 4.6. 4.6. 4.7. 4.8. 4.9. 4.10.

.................................................. ...................................... Psychological ................................................

..

Tinnitua ...................................................... Swishing Breath ................................................ Meteorological Theories .......................................... Direct Transmission of Audible Sound ............................ Infrasonic Waves .............................................. Direct Perception of Electromagnetic Radiation .................... The Electric Field Pressure Effect Direct Effects of Auroral Electric Fields .......................... Emission of Radio Waves in the Audio Region and Conversion to Pressure a t the Ear ....................................................

................................

166

166 167 167 169 169 161 176 176 177 178 180 180 181 186 187 191 191 192 192 193 194 196 196 197 198 198 198 199 200 200 201 202 203 204 207 208

156

S. M. SILVERMAN AND

T. F.

TUAN

6 . The Case for Brush Discharge and Aurorally Induced Electric Fields

......

Cornperison of the Behavior of Auroral Sounds and Brush Discharges. . Aurorally Associated Electric Fields .............................. Mechanism for Production of Aurorally Associated Electric Fields .... 5.4. Energetics of Sound Events Deriving from Aurorally Associated Electric Fields ........................................................ 6. Conclusions.......................................................... List of Symbols ...................................................... Appendix: Auroral Sound Events ...................................... References .......................................................... 5.1. 5.2. 5.3.

208 209 210 214 214 216 217 218 259

1. INTRODUCTION

Sounds which simultaneously accompany visual observations have been reported for auroral displays for a t least the past two centuries. Because of the absence of any clear physical mechanism for the production of sound from auroras whose heights cannot be much less than 60 km, these reports have met with considerable uneasiness and skepticism on the part of physical scientists. I n this review we present a comprehensive survey of the observational material available, and of the theories which have been proposed to account for the sounds. The explanation eventually adopted will depend on the elimination of other less likely theories. It is important, therefore, to develop to the maximum extent the connections between such objective facts as are available in the observations and geophysical parameters such as, for example, latitude or sunspot cycle. Such relationships as may exist can then be used to test proposed theories. These include such possibilities as psychological origin, physiological, meteorological, direct transmission of sound, conversion of electromagnetic waves to pressure a t the ear, infrasonic waves, and electric field effects, both directly as well as indirectly through brush discharges. Sounds whose descriptions are remarkably similar t o those reported for auroras have also been reported from lightning and meteors. The similarity leads t o the question of whether similar explanations can be used for the three phenomena. This question will not be considered in detail, but some comparative material- will be included, particularly in connection with the theoretical treatment. We anticipate the results of our discussion by stating that electric fields, a t least indirectly, can cause the observed sounds. It will be shown that for intense auroral displays and when weather conditions are right, brush discharges from trees and bushes can occur producing a hissing, rustling, and crackling sound compatible with the reported descriptions. Brush discharges can also result from lightning-induced fields, so that it is indeed possible to find a common explanation for these very diverse phenomena, although the origin of the fields are quite different.

AURORAL AUDIBILITY

157

2. OBSERVATIONAL RESULTS I n this section the observational results are summarized with particular emphasis on the correlation of these observations with geophysical parameters. Two general observations may be made beforehand. (1)Permanent residents of Arctic regions typically report hearing the aurora as a “ not uncommon ” occurrence, while explorers, even with some years of residence, often report never having heard the aurora. (2) The question has often been raised as to whether the sounds have been heard by professionally trained observers, the implication being that laymen or uneducated natives (including those whose survival depends on accurate perceptions of their environment) cannot be trusted to give accurate reports of their own sensual impressions. This question will therefore be considered first.

2.1. Methodology and the Use of Anecdotal Data Because our discussion of auroral audibility depends strongly on anecdotal data, consideration of the appropriate usage of these data is appropriate a t the start. The data are derived from anecdotal observations by both professional and lay people, with a good many of the observers having been professionally trained in various disciplines. Basically, what the observer is asked is whether he heard sounds and to recall as many of the related circumstances as possible. Our problem is then to determine whether the sounds are of subjective origin, and whether, if objective, a credible physical mechanism can be devised which will produce these sounds. Any observer, with or without training, if speaking in good faith, and if the discrepancy in cultures does not produce complete distortion in the transmission from speaker to recorder, can be relied on to state a simple “ yes ” or “no ” choice as to whether he heard a sound or not. The decision must then be made as to what aspects of anecdotal data can be useful, in addition to the simple “ yes ” or “ no ” statement of having heard sounds. The following data can be provided by even inexperienced and untrained people: the location at which the sounds were heard; the year; the month and day; the time of day; some statement about the weather, such as if it were windy or calm; some statements about the aurora, as, for example, whether it was in rapid motion and whether it was overhead; and a miscellaneous group of statements such as whether they could hear sounds with their eyes covered, or whether dogs were affected or not. The preceding, in essence, forms the raw material for the discussion. The problem of cultural distortion has been brought out by Gartlein (1947), in discussing Eskimo testimony. (It may be noted that even if all the Eskimo testimony were discarded the results would not be changed, since this evidence is largely in the general category while our discussion is based

158

S. M. SILVERMAN AND T. F. TUAN

mostly on the individual reports.) Gartlein quotes an experience of J. A. Easley, Jr., who had spent two winters (1943-45) at River Clyde on Baffin Island operating an ionosonde station. He had learned enough of the Eskimo language to speak with them. Easley noted that an aurora which they had told him was noisy turned out not to be noisy, but that the Eskimos then told him that a rapidly moving aurora was called noisy. That no noise was heard is not surprising, as will become apparent later, for the geomagnetic latitude of his station is about 80".His experience, however, does raise the question of whether all data coming from Eskimos must be discarded. Two objections may be made to the elimination of the Eskimo testimony. First, many traders and missionaries who have spent many years among the Eskimo and presumably spoke the language fluently have quoted the Eskimos as having heard sounds. Second, we may note that Davies and Currie (1933) have pointed out that the Eskimo word for aurora is " akshanik ') meaning that which moves rapidly, and that the same word is used by them for Chesterfield Inlet because of its rapid tides. Eskimo mental economy thus seems to call for the use of the same word for natural phenomena with similar dynamic characteristics, with the specific phenomena being defined by the context. It is thus likely that the great majority of reports can be accepted since the context would have been clearly defined by the questioner. I n any event, possible distortion would be expected to be less for most of our reports, the cultural backgrounds of questioner and observer being more closely similar. Finally, account must be taken of the bias t o be expected in the case of creative physical scientists, who are likely to be the group doing research on the subject. McClelland (1962) has summarized and discussed the findings relating to the psychodynamics of creative physical scientists. Of particular interest in connection with the use of anecdotal data are the characteristics of avoidance of interpersonal contact and the strong interest in the analysis and structure of Nature. The f i s t characteristic implies a strong suspicion of data derived subjectively, presumably because to the scientist human relations seem difficult,' and data arising from this source is therefore difficult to evaluate. The second characteristic leads to a diaculty in accepting already suspected data when the existing known theoretical structure seems to contradict the data. I n the present case, it is known that sounds 'Thus McClelland states: " Outstanding scientists are anything but normally gregarious. They like being self-sufficient and like being alone probably because people and human relations seem both difficult and uninteresting to them." Ferenczi (1912), in the Utopian days of psychoanalysis had pointed out the possibility of distortions being introduced into scientific work by unconscious affects of the scientist. He proposed that every scientist should be psychoanalyzed in order that the enormous effort I'. . which is now wasted on infantile controversies and priority disputes, could be put at the disposal of more serious aims."

.

AURORAL AUDIBILITY

159

presumed t o be originating at 100 km level cannot reach ground level. If the data are perceived t o be suspect, as being of subjective origin, then the search for an alternative mechanism seems wasteful. If the preceding factors are taken together, then the creative physical scientist introduces a bias whose primary aspect is that of ignoring the data and the subject altogether, even when, as will be noted below, the observers are themselves scientists or otherwise professionally trained.

2.2. Observations by Professionally Trained Observers The impression appears to be widespread that reports of auroral audibility come almost entirely from (‘natives” or other ignorant and easily misled individuals. This impression, however, is grossly in error, in addition t o being symptomatic of an unwarranted prejudice that people who spend most of their lives outdoors are incapable of observing nature accurately. I n any event, it is possible to note quite a large number of reports from individuals whose professional observing skills or good faith, or both, can in no way be doubted. Beals (1933a), in fact, compiled his questionnaire with special emphasis on reports from Canadian government officials, Hudson’s Bay Company employees, and members of the Roman Catholic and Anglican clergy in order to make use, as he put it, of the “large body of men whose intelligence and education is well above the average and who have had years of experience in regions where brilliant auroral displays are of frequent occurrence ”. Taking these, as well as the other reports, we find, at a minimum, fourteen government officials, eight explorers or military men, four physicians and clergy, and more than thirteen academically or scientifically trained men. The latter group includes a Dean of Applied Science, the director of an astronomical observatory, and several members of the Yale faculty. It is, of course, still possible to question the observational powers of this group of trained individuals, but the basis would have to be a more universal human failing. These observations by scientists, in fact, should be given additional weight as positive evidence because of the bias discussed in the preceding section.

2.3. Surveys Two questionnaire-type surveys have been carried out, one by Tromholt (1885b)in Norway and one by Beals (1933a) in Canada in 1931. Although the two questionnaires are not completely comparable i t is possible to compare the statistics of those replying to the question of audibility. Thus for Tromholt, of 113 reports on audibility 92, or 81 %, heard sounds. For Beals, of 184 reports 144, or 78 % heard sounds. These figures are, of course, not representative of the fraction of the population who might be expected to hear sound. Properly designed mail surveys in the contemporary United States, in which

160

S. M. SILVERMAN AND T. F. TUAN

it is made easy for people to respond, typically have no more than a 2 6 3 0 % response rate.a The response rate in Tromholt’s 1885 survey was approximately 15%, indicating a distribution skewed towards those most aware of the question and thus most highly motivated to answer. If we make the most extreme assumption, that the difference in response between the contemporary United States and 1885 Norway is entirely made up of people who have never heard auroral sound, then the fraction of the population hearing sound on some occasion would be about 40 yoin Norway. We should remember in this oonnection that Norway is almost entirely south of the auroral zone, with the frequency of aurorae in the southern part about 5-10y0 of that in the maximal frequency zone in the northern part. The majority of the population living in the South, however, would tend to bias the results towards not hearing sounds. This would be particularly true as the hearing of sounds appears to be a rare occurrence. Beals quotes one observer who estimated that only 5 yoof the brilliant displays he had witnessed gave rise to sounds, and mentions also numerous reports from people who had spent 20 or 30 years in the Arctic but had only heard the sounds on one or two occasions. Beals, on the assumption that the sounds would be more frequent in the auroral zone, sent his questionnaire to people who had lived or travelled in northern Canada. Unfortunately he does not give the number of questionnaires sent out so that it is impossible to estimate the fraction of the population in the auroral zone who have heard sounds. Davies and Currie (1933) report from Hudson’s Bay, Canada, that all Caucasians and Eskimos in and near the auroral zone had heard sounds. A strong latitudinal dependence (discussedin more detail below) is indicated by their data, particularly by the report that Eskimos from Southampton Island, approximately 5” north of the auroral zone, had never heard sounds. The statistical evidence given by Tromholt and Beals thus seems to support the reality of auroral sounds, though it appears that account would have to be taken of sampling bias, latitudinal effects, and different residence times of, say, government officials with limited tours of duty (of the order of 1-3 years) and permanent residents (of the order of 20-30 years). I n view of these differences it is remarkable that the percentage of those hearing sounds is almost the same in these two surveys taken almost a half century apart in two different parts of the world. Finally, we may note that, as Beals points out, the descriptions of the sounds are monotonously similar, with (in Beal’s survey) 95% of them classified as hissing, swishing, rustling, or crackling. The majority of the descriptions awe are indebted to Dr. Robert Weiss, Laboratory of Community Psychiatry, Harvard Medical School, for a discuseion of mail surveys and their limitations. See also Parten (1960).

AURORAL AUDIBILITY

161

in Tromholt’s survey also fit into these categories. The consistency of these descriptions seems to be strong presumptive evidence for the reality of auroral audibility. 2.4. Individual Observations

A great many observations of auroral sound can be found in the literature, particularly in the papers of Chant and of Beals (1933a) and, earlier, in the classic work of Fritz (1881) on the aurora. These observations are listed in Table I. All of the relevant information for all the observations has not been tracked down because of time limitations, but the table is nevertheless sufficiently complete for significant conclusions to be drawn. The literature from which these have been drawn, as well as all other papers of which we are aware relating to auroral audibility, are listed separately in the References. Reference to the table shows that there are no literature references to observations for the decades subsequent to 1940, and this raises the question as to whether auroral sounds are no longer being heard or merely no longer being reported. One of the authors (S.M.S.) has, over the past ten years, spoken or corresponded with several people who have heard the aurora, but neglected to obtain these testimonies in writing. The reporters have included a bus driver in Saskatoon, Saskatchewan; a technician at the Churchill Rocket Range, Manitoba; a government official resident for many years in Greenland; a resident of northern Maine; and a naturalist in the North of England. It is clear, then, that auroral sounds are being heard. The listing in Table I derives from a bibliographic search through the literature available t o the authors, a search of the references, and discussions with several people who have considered the subject over the years. Obviously a search through newspapers and nontechnical periodicals for casual references to auroral sounds is impossible because of time limitations. Thus the question resolves itself into that of the reasons for the absence of references since 1941 in the technical literature. A convincing explanation would involve excursions into the history of science and into the interaction of science and society-topics which are outside the scope of this paper. Briefly, however, we speculate that there are two major factors. The first is the change in the emphasis of science to instrumental methods and an increasing suspicion of the human as sensor and reporter. The second, and possibly the more important when we consider that the big gap is since 1940, is the change resulting from the close symbiotic relationship of science and government during end following the World War 11. The focus of research is then on what can attract financial support from government, and this does not include fringe phenomena of doubtful validity. The net result of both factors is then the absence of published reports reflected in Table I.

TAB-

I. Catalog of auroral sound evenha

Part 1: Individual ReportgDated Geographic

No.

Date

P h

Lat.

Long.

Geomagnetic Lat. Reference

l b

Dec. 1563

Bergen, Norway

60'24"

5"20'E

61"N

0-1

2=

Oct. 8, 1726

Little Chelsea, England

51'30"

4'40W

55.2'N

Derham (1726-1727)

3

1762

Peris, h c e

48'50"

2O20'E

51.2'N

Fritz (1881), pp. 282-283 Angot (1897), p. 49

4

1766

Norway

Fritz (1881), p. 281

5

1767or1768

Norway

Fritz (1881), p. 281

6d

Between 1769-1772

Great Slave Lake, Canada Churchill River, Canada.

7;

Jan. 18,1778

8

Nov. or Dec., 1781 or Jan. 1782

9'

N

N

N 68.7"N -67.3"N

62"N 57.5'N

(1914)

Chant (1915e);Beak (1933b), p. 71

Oxaal (1914)

Spydeberg Near New Haven, Conn.

41'18"

72'58'W

52.8'N

Holyoke (1828)

1781

Dover, New Hampshire

43'12"

70°52'W

54.7"N

Belknap (1786)

70"52'W

54.7'N

Fritz (1881), pp. 278-279

2'46117

57"N

Dalton (1834), p. 55

65.8"N

Beak (1933b),p. 71

10

Mar. 31, 1783

Dover, New Hampshire

43"12%

11

May 24, 1788

Kendal

54'17"

12

Winter 179&97

Reindeer Lake, Saskatchewan

N

57'N

-

102.5'W

N

13

Dee. 5, 1801

Orkney

59"N

3"W

61.9'"

Fritz (1881), p. 280

14

Winter 181P15

Reykjavik, Iceland

64'1 ON '

21'25'W

70.2"N

Force (1856), p. 23 Gartlein et at!. (1967), Table IV; Fritz (1881), p. 279

15

1818

Skien, Norway

59'11"

9'37'E

59.5"N

Fritz (1881), p. 281

kU

65'N

Fritz (1881), pp. 280-281

160'56'E

59.9'N

Gartlein et al. (1967), Table IV; a. Fritz (1881), p. 282, b. Force (1856), p. 80

64'28"

113"6W

71.1°N

Capron (1879); Hood (1823); Chant (1923a), pp. 276-277; Gartlein et al. (1967), Table IV; Force (1856), p. 38

43"7'N

75"13'W

54.5'"

Silliman (1828), p. 94; Fritz (1881), p. 279; Olmsted (1856), p. 7

16

Nov. 1818

Between Iceland and Shetland

17

1820-1821

Nijnei Kolymsk, Siberia

68'32"

18

Mar. 11, 1821

Fort Enterprise

19

Aug. 28, 1827

Utica, New York

20

Aug. 28, 1827

St. Lawrence County, N.Y

21

Aug. 28, 1827

New Haven, Conn.

22

1828-1829

Southern Greenland

23

1833-1834

Fort Reliance

24'

Nov. 18, 1835

N

Sillimmi (1828), p. 94; Fritz (1881), p. 279

41"18'N

75"58'W

52.8"N

(1828), p. 95 b. Olmsted (1856), p. 7

a. S i l l h a n

Force (1856), p. 21; Gartlein et d.(1967), Table IV 70.3ON

Force (1856), p. 19; Gartlein et al. (1967), Table IV

Dunse, N. Britain

Stevenson ( 1853) 72.1°N

Force (1856), pp. 77-78; Gartlein et d.(1967), Table IV

25"

Mar. 1839

Fort Confidence

261

Sep. 3, 1839

Dunse, N. Britain

27'

Sep. 4, 1839

Dunse, N. Britain

28'6

Sep. 20, 1839

Livland

62"46'N

66"54'N

109OOW

11S049W

Stevenson (1853) Stevenson (1853)

.- 56"N

Continued

Fritz: (1881) p. 282; Angot (1897), p. 49 Footnotea on pagw 174-175

Table I Contilzwd

c.

a I+ Geographic Geomagnetic Lat. Reference

Date

Place

29

Mar. 22, 1840

Kaafjord, Norway

69O55'N

22'50%

66.7"N

30

1839-1849

Alten, Norway

69"58'N

23'43%

66.6"N

No.

31'

Mar. 16, 1842

Lat.

Long.

Winter 1846

Near Barra Head, New Hebrides

33"

1849

Fort Franklin

34

Dec. 2, 1850

Oddie (1933) a. Lemstrom (1874), p. 232;

Lapland between Maunu and Lyngen

32

a. Oxad (1914); b. Fritz (1881), p. 281

N

56.78'N

N7.5OW

b. Oxaal (1914) McC. (1881) ~60.7~N

LeFroy (1852); Chant (1915~)

ki U

Sep. 7, 1851

Oh0

Oxaal(1914)

okso

37"

Apr. 21, 1852

O h

Oxaal(1914) 0 x 4 (1914)

38n

Feb. 24, 1854

Oh0

39

Nov. 1856

Alten, Norway

Oxaal(1914) 69"58'N

66.6"N

23'43'E

Oxaal (1914) Fritz (1881), pp. 279-280

Iceland

41

Nov. 1865

Aloukuk, Alaska

62'N

Bannister (1866)

42

Jan. 1866

St. Michaels, Alaska

63"29'N

162'1W

60.7'"

Bannister (1866)

Dumfriesshire

55O2'N

3'38W

582N

Shaw (1881)

Ingersoll

43"5'N

slow

54.3'N

Chant (1923a), p. 278

1866

43

'Y

440

Summer 1870

P

River near cham of Rocky Mtns. running westward of Peel's River

Feb. 19, 1852

1860

m

d E

69.7"N

123'13W

36"

'Y

F

Force (1856), pp. 61-62; Gartlein et al. (1967), Table I V

65"12'N

35"

40

m

N

9 F

3 2

a. Capron (1879); Fritz (l88l), p. 285; b. Oxaal (1914) Angot (1897), p. 50

45'

Nov. 24, 1870

Between Paris and Norway

46

"Aug. 15, 1882

20 miles east of Fort McLeod

47s

Nov. 18,1882

Canada

48

Winter 1882-83

Fort Rae, N.W.T.

49

NJuI. 31, 1883

Midway between Quebec and Montreal in St. Lawrence River

N

55'0"

N

123"OW

~ 6 0 . 4 ' N Chant (1923a), pp. 281-282

King (1907); Beals (1933b), p. 72 116"3'W

62'50%

N

47'N

-

73"W

N

69"N

a. Beals (1933b), p. 71; Chant (1923a), p. 274; b. Chree (1910)

58.5"N

Webber (1887) b-

d

50

Midwinter 1884 or 1885

Kingston, Ontario

44"14'N

76'30W

55.6'N

Chant (1923a), p. 274

51

1884

Durham Grey Co., Ontario

44'20"

80"49W

55.5'N

Griffin (1922)

52

Jan. 1888

3 miles E & 1 mile N of Saskatchewan

53

N

1892

Northern Minnesota

Oct. 11, 1893

At sea

55

1893

Brampton, Peel Co., Ontario

56

Winter 1894-1895

South of LaTuque, Quebec

57

Late Fall 1897

Rocky Mts. near Arctic Circle

58' 59

Jan. 1898 Winter 1898

Iditarod, Alaska Fort Graham

N

1898

53'N

-107OW

~ 6 1 . 3 " N Chant (1923a), p. 280

N

48'N

N95"w

-58'N

F

Chant (1923a), p. 276 Chapman (1931)

54

60.

N

T1

N

-

43'40%

79"46W

54.9"N

47'N

73"w

N

Griffin (1922)

58"N

Chant (19384

67'N

140"W

~ 6 8 ~ N chant (1923a), p. 278

62"30'N 56'35%

15So11W 124'38'W

60.5"N 61.6'N

Chant (1915b) Chant (1928a) Chant (19234, p. 278

Strathaven Continued

Footnotes (

pagea t 174-176

~

'

1

Table I Co7atinued

c

a

Q,

GeOgTaphiC Date

Place

61

Winter 1901

Eagle, Alaska

62

1901 or 1902

Vaaga Bay North of Hamtad, Norway

63L

1903

-

Canada

No.

Let.

Long. 141"14W

64'47"

-

69'N

Geomagnetic Lat. Reference

17OE

N

-

65.9ON

J o h n (1927); Chant (192Sa)

66.9"N

Oxaal(1914) Chant (1923a),p. 283

Sackville, New Brum.

45O54"

64"22W

57.3"N

Chant (1929b)

Jul. & Aug. 1905

Cartright, Labrador

53'42"

57OW

64.9"N

8.

66

Sep. 1907

St. Lawrence River

67

Jan. 1908

Discovery on Sulphur Creek, Yukon Terr.

$4

65

1904

68

N%p. 10, 1908

1Mituni, Manitoba

69

Jan. 26, 1911

Dawson. Yukon

70

Oct. 11, 1911

7 1"

1911

2 or 3 miles south of Lake Enare, Finland Antarctica

-47.5'N

-

70'W

-

59ON

G f i n (1922); b. Beah (1933b),p. 73 Chant (1923a),p. 275

Beak (19334, pp. 190-191 49O21'N

-

64'337 69ON

-

98O11W

59"N

Chant (1923a), p. 278

139'25W

65.5"N

chant (1911)

65"N

0 x 4 (1914)

28"E

-

Oxaal (1914)

7 2"

1911

Antarctica

Oxaal (1914)

73'

Winter 1911

Framheim, Antarctica

Griffin (1922)

74

Winter 1914-1915

Gold Run Creek, SE of Dawmn, Yukon

75

About 1916

Spencers Island, N.S.

76

Feb. 1919

Cumberland IIouse

-

64"N

-

45'N

53'5837

N

139"W

-

65"W

-

102"16W

65'N

Beah (1933a),p. 195

57ON

Chant (1932b)

62.9'N

Chant (1923a),pp. 278-279

50'20"

102O30W

59.4"N

Chant (1923a), pp. 280-281

77

Oct. 15, 1919

78

Autumn 1920

9 miles SE of Broadview, Saskatchewan Labrador

79

Oct. 1921

Fort Fitzgerald, Alberta

59'51"

111'41W

67.1"N

80

Winters 1924-25 and 1925-26

Norman & Simpson, N.W.T.

65'19% 61'58"

126O46'W 122"OW

69.2'N Beah (1933a), p. 191 67.1"N

81

Aug. 8,1924

82

Winter 1925-26

Head of Nuluk River

83

Oct. 15, 1926

Oslo, Norway

84

Jan. 10 or 12, 1927

Wellington Bay South of Victoria Island

85'

July 1928

Pittsburgh, Penna.

86

Jul. or Aug. 1928

Smiley, Sask.

87

Winter 1928-29

Langton Bay, N.W.T.

88

winter 1928-29

Magnetic Pole

70°N 63"20'N

89

Mar. 20, 1933

Chesterfield Inlet

90

Jan. 25-26, 1938

Njuke Mtn. Tuddal, NOrW8Y

91

Jan. 25-26, 1938

Aldringhsm, Lieston,

Chant (1923a),p. 273

-

Beals (19334, p. 192 66'N

N69°8'N

51O37'N 69'40"

93

Sep. 27, 1938

Sep. 18, 1941

-

-43'39"

North of Maniwaki, P.Q.

-46.5"N

106'5W

60'N

-76.5'"

80"OlW

-

52'12"

Hill north of Toronto

-62.1'N

10'43%

40'26"

England 92

-167"W

59O54'N

-

Beah (19334, p. 194

-

Jelatrup (1927); b. Shrmer (1927); Beak (1933e),p. 73; Chapman (1931); Chant (1931)

a.

Beah (1933a),p. 191 Chant (1928b)

59.6"N

Stumbles (1938)

73.2"N

Beah (1933a), p. 189

96"W

79"N

Eve (1936)

90'42W

73.5"N Davies and h i e (1933)

109'29W 125O3OW

1'30'E

-

51"N

Garber (1933); Chant (1933b)

79"23W 75'58"

-

-

6O"N

Sterner (1938)

54.6"N

Chant (193%)

54.9"N

Chant (1938d)

57.8"N

Chant (194313) Footnotes o n pages 174-175

+

Table I Continued

Q,

00

Part 2: Individual Reports-Undsted Geographic No.

Place

94u

Fort Henley

95 96

Northampton

Let.

Long.

Geomagnetic Lat.

Reference Blagden (1784) Blagden (1784)

Sweden

52'15"

0'54 W

55.1"N

a. Blagden (1784);

?

b. Capron (1879), p. 33

F

Holyoke (1828)

97

EesternU.S.A.

98

Toronto

43'39"

Aberdeenehire

57'8"

2OWW

59.9"N

Rouse (1881 )

55'57"

3'12W

59.1'"

Paukhurst (see Kemble, 1881)

79"23W

54.9"N

9W 100 101

Edinburgh

102'O

Newfoundland, Canada

Fritz (1881), p. 279

103bb St. Lawrence Bay, Canada 1 0 4 c c Iceland 105d6 Reykjavik, Iceland 106=' Hafnaford, Iceland 107"

Unst, Shetlands

108gg

Hebrides

Chant (1916) Capmn (1879), p. 33

N

64'10%' 64'4%'

21"25'W 21'57'W

60"50'N

O"52'W

57'N

7"w

N

I

E

$

U H

Fritz (1881), p. 279

?

Fritz (1881), p. 279

H

70.2"N 70.2'N

Fritz (1881), p. 279 Fritz (1881), p. 279

63.1°N

Fritz (188l),p. 280

61"N

Fritz (1881), p. 280 Fritz (1881), p. 280

109"h

!

m

110"

Shetlands

Fritz (1881), pp. 280-281; Angot (1897), pp. 49-50

111"

Northern Sweden

Fritz (1881).p. 281; Joannis (1886)

4

Ei

112"k

Fritz (1881), p. 281

113"

Fritz (1881),p. 281

114'"

Esthland

115""

Gera, Germany

1160°

Alsen

117 11S P P

Gaspe, Canada

Fritz (1881),p. 282 Fritz (1881),p. 282

50'52%

12'06'E

51°N

59'23"

2"52W

62.2"N

Rendall (1887)

60°N

King (1907)

Fritz (1881),p. 283

-

48'N

64"W

N

N

OX^ (1914),pp. 27-29

11904 120

Dawson, Yukon

64'3"

139'25W

65.5"N

Griffi (1922),p. 256

121

St. John, N.B. Haliiax, N.S.

45'16" 44'39"

66'3'W 63'36'W

56.75"N 56.1"N

Griffin (19%2),p. 256

122w

Wood's Harbour

43'30"

123"

Fort William, Ontario

48'23"

89'15'W

59'N

84OO'W

-57.2"N

124

St. Joseph's Island

125

Nova Scotia

126

Yukon

127" 128

Silver Mtn. near Arrow Lakes

-46'15"

131""

Winnipeg

N

Chant (1923a), p. 274 Chant (1923a),p. 274 Chant (1923a), p. 274 Chant (1923a),p. 274

-

129

Toronto 13OUy Kettle Fall on English River

Griffin (1922),p. 256

50°N 43'39"

N

50'30" 49'53"

-

Chant (1923a),p. 275 118"W

~56.6'N

Chant (1923a),p. 275

79'23'W

-

54.9"N

Chant (1923a),p. 275

60.7'N

Chant (1923a),p. 276

59.6'N

Chant (1923a), p. 277

93OW 97"YW

-.

132

Canadian Northland

Chant (1923a),p. 282

133'O"

Yukon and Manitoba

Chant (1923a),p. 283

134

Klondyke

Chant (1923a), p. 283 Continued

Footnotee on pagee

174-175

c Q, (0

Table I Continuut

Geographic No.

Place

135

Yukon

Lat.

Long.

Geomagnetic Lat.

Reference Chant (19234, p. 283

136==

Chant (1923b)

137yy Yukon

Beals (1933a),p. 189

138

Beals (193k), p. 189 Beals (1933e),p. 190

139 140

Beals (1933a),p. 190

Yukon Vslley & Western part of North West Ten.

Beak (1933a),p. 192

141

Beals (1933a),p. 192

142 143

Juneau, Alaaka

57"30'N

134O30W

60.6"N

Beals (1933a),pp. 193-194

144

St. Stephen, N.B.

45'12%

67'17W

56.7"N

Beals (1933a),pp. 194-195

145

Montreal, Canada

45'31"

73"34W

57'N

146

Province of Quebec

Connecticut 14P2 Central Saskatchewan 147

149

Saskatchewm Prairies

Chant (19334, p. 256 Eve (1936),p. 173 Chant (1928b),p. 398 Davies and Currie (1933), pp. 855-856 Letter to National Geographic, April 9, 1948.-CourteSy of Dr. Sprague, Cornell University

Table I Colctinued

Part 3: General Reports No.

Place

Reference

150

Beyond the northern borders of Germmy

a. Rouse (1881a);R.O.S. (1881), p. 484

151

Siberia

152

Canada

Blagden (1784), p. 228; Capron (1879), p. 33; Fritz (1881), p. 282 Chant (1923a), p. 277; Franklin (1823); Force (1856). p. 26

153

Canada

Chant (1923a), p. 277; Richardson (1823); Force (1856), pp. 40-41

154

Can&

C h t (1916), pp. 97-99

155

Canada

Chant (1916), pp. 97-99

156

Canada

Rae (1881), p. 605; Force (1856), p. 77

157

Greenland

Capron (1879), p. 33

158

Greenland

Capron (1879), p. 34

159

Shetlands

Fritz (1881), p. 280

160

Siberia

Fritz (1881), p. 282

161

Siberia-Ob River

Fritz (1881), p. 282

162

Germany

Fritz (1881), pp. 282-283

163

Canada

a. Chant (1923a), p. 273; b. Chrea (1910)

164

Lapland

Oxael (1914), pp. 27-29

165"'

Alaska

Griffin (1922), p. 256

166

Canada

Beah (19334, p. 192

167

Canada

Davies and Currie (1933)

168

Alaska

Elvey (1957)

16gbbb Canada

Chant (1929a) Fritz (1881), p. 280

170CCc 171

Shetlancls

Fritz (1881), pp. 280-281

172

Finmark

Fritz (1881), p. 281 COntiaZd

+ -a + Footnotes cn pagee 174-175

Table I Continued

Part 4: Some Negative Reports

No.

Place

Reference

173

Lapland

Capron (1879), p. 34

174

Iceland

Fritz (1881), pp. 279-280

176

Iceland

Tromholt ( 1884)

176ddd Canada

Chant (19158)

177Cee Labrador

Clerke (1887)

Part 5: Supplementary Catalog. These include reports gathered subsequently to the completion of the original analysis. Geographic No.

Date

Place

178 179

Fall, 1842 or 43

Strafford, N.H.

Winter, 1894

Amy, Norway

180

Late August, 1904

SW Minnesota

181

Montreal?

182

Faroe Shetland

183 184

Orkney December 1821

Winter Island

Lat.

Long.

43'15%

-

45'30"

N

N

71"lOW

54.7"N

Cavern0 (1911)

2Oo30'E

67.3"N

Johnson (1971)

54"N

Pel1 (1912)

57.O"N

Olmsted (1856)

65"N 63"N

Force (1856), p. 15; Amer. J. Sci. (1824), p. 392

62'N

Rae, quoted in Force (1856), p. 77

7O"lO'N

44'N

62'N 60"20'N 59'N 66"11'25"N

Geomagnetic Lat. Reference

N

-

96"W

N

73"36'W

N

7"W l"25W

N

3"w 83"lOW

N

N

77.1"N

Lyon, quoted in Force (1856), p. 65

185

August 28, 1859

Grafton, Canada

44'3"

78O5'W

55.4'N

Hubbert (1860)

186

October 31, 1838 January 10, 1839 March 10, 1840

Bossekop, Norway Bossekop, Norway Kmfiord, Norway

70°N

23"E

66.7'"

Angot (1897), pp. 47-48

187

January 25-26, 1938

British Isles

Dixon (1948), p. 15

188

Greenland, Finmark, Siberia; Koutokeino, Bossekop, Norway

Tromholt (1885 ), pp. 284-285

189

Orkneys, Finmark, Hudson Bay, Lapland

Angot (1897), p. 47

190

Early 1950's

Cleveland, England

54'23"

l"19'W

57'N

C. Simms (private communication)

191

1930's

46"42'N

68'1 W

58"N

Cogswell (private communication)

1950

Presque Isle and Easton, Maine Near Bangor, Maine

44'49"

68'47 W

56'N

1933-1938

Easton, Maine

46'42"

68'1W

58'N

192

Markham (private communication)

193

Alaska

Elvey and Rust (1962)

194

Alaska, Canada, Greenland

Ray (1958)

195

Summer, 1920-21

Alvarado, Minnesota

48'15"

97"W

58'N

Olson (1972)

196

Deo. 1937 or Jan. 1938 Kirkenes, Finnmark, Norway

69'41"

30"E

65"N

Olson (1972)

197

Winter, 1938

Alvarado, Minnesota

48'15"

97OW

58'N

Olson (1972)

198

November, 1971

Edmonton, Alberta, Canada

53'34"

113'25W

61"N

Olson (1972) Footnotes on pages 174-175 W

174

9.

M. SILVERMAN AND T. F. TUAN

FOOTNOTES TO TABLE I QThenumbers refer to quotations of the literature which are listed in the Appendix. Geomagnetic latitudes, when to the nearest degree, have been estimated from Fig. 11-2, in Chernosky et al., 1965; and when t o tenths of a degree have been calculated from a program of the Geomagnetism Branch, Air Force Cambridge Research Laboratory. bThe description is that of a n uneducated layman, but seems nevertheless to be that of a bright aurora, and has therefore been included here. OThis has been assumed t o be Chelsea, London. The coordinates given are those of London. dThe report is from Samuel Hearne's "Journey from Prince of Wales' Fort in Hudson's Bay to the Northern Ocean in the Years 1769, 1770-1 and 1772." We have not checked the original to see whether the dates can be more precisely defined. ePresumebly Norway. 'This has been taken to be Dover, New Hampshire, as for (lo), from the context of the letter. 'Probably Duns, Scotland, 55"47'N, 2'20'W. "Gartlein et al. give the date as 1838. Wee note on (24). 'See note on (24). kAssumed to be Livonia, a district in Latvia. 'The descriptions of Oxml and Lemstrom are sufficiently similar that they have been assumed to be from the same source. Lyngen, Norway, is at 69"36,'N, 20"lO'E. mFromthe fuller description in Force this does not appear to be a true sound event. It has not been used in the statistical analysis. "Presumably Norway. OAssumed to be Ontario, Canada. PThe sound appears to have been heard at the landing point of the balloon, a mountain in Lifjeld, Telemarken, Norway. The story itself is one of the romantic (or possibly romanticized) adventures associated with the siege of Paris, unfortunately outside the scope of this paper. QProbablyYukon Territory, since the reporter, Mr. Wm. Ogilvie, is noted as a wellknown surveyor and explorer of this territory. 'The place of the sounds has been taken as the residence of the reporter, a miner, though this is not explicitly stated in the account. *Probably in Ontario, Canada. 'Possibly Yukon Territory. "Reported by Mr. Trygve Gran, the only Norwegian member of Scott's expedition. "Report by Mr. Gran [see note on (71)] of sounds heard by the party of Lt. Campbell. WThisia the often-quoted report of Amundsen that the periods of the rustling noise coincided with his breathing. It is included here for completeness, but was not used in the Statistical treatment. 'Reference is made to sounds during two auroras in succeeding weeks. One was probably that of July 7-8. "Probably Henley House, on the Albany river, approximately 51°N, 84"W, Geomag. let. about 62'N. See D. McKay, "The Honourable Company" Toronto: McClelland and Stewart, 1966, p. 75, for this location. aA quotation from a Mr. Cavallo, "Elements of Nat. or Exper. Phil." Vol. iii, p. 449. QaAreport by de la Pilaye. No source is given.

AURORAL AUDIBILITY

175

bbAreport by Steward. No source is given. ccA report by Count Trampe. Reports by Mr. Siemens and Dr. Hjaltalin from about 1860 are also given but not listed here separately. No sources are given. ddAreport by Jorgensen. No source is given. eeA report by Hygom. No source is given. ”A report by James Hay, a guide. No source is given. ggA report by Dunbar. The source is given as Edinburgh J. Natur. and Qeogr. Sci., N . S ., LV. hhAreport by a Captain Abrahamson. The source is given aa Schweiger J., N . S . XV. “The reporter is a lighthouse keeper. No source is given. ”The observers are Gissler and Hellant, as reported by Wargentin. No source is given. kkThereporter is Pontoppidan. No source is given. “The reporter is N. Hertzberg, in Ullensvang in 1826, speaking of his youth, when aurora were more frequent. Thus the auroras referred to may be as early aa those of the maximum of 1788 or earlier. No source is given. mmProbablyEsthonia. The reporter is Petri. No source is given. ““The reporter is Winkler. No source is given. OOThe reporter is an astronomer, Brorsen. No source is given. PPThe reporters were four members of E prominent Ottawa family. QQThe reporter was Tromholt’s father, “ a skillful and reliable meteorological observer about whose trustworthiness there cannot be the slightest doubt.” The place is therefore probably Norway. “The observer is Mr. H. E. S. Asbury, President of the Montreal Centre of the Royal Astronomical Socieby of Canada. He is also the observer for Nos. 51 and 55. s*The observer is Dr. 0. C. J. Withrow, a physician from the context. “The observation is based on many years of daily meteorological observations. The observer is A. J. Woodward. “The position given is the approximate center of the English River. We have not determined the exact location of Kettle Fall. ““The observation or observations were during the winters of 1891-2, 1892-3, and 1904-5. WWThe observer was Mr. W. W. Cory, Deputy Minister of the Interior. IZThe poetic imagery of the report is such aa would normally disqualify it from serious consideration. We nevertheless note it here, though it has not been used in the statistical studies. YYThe reporter was F. H. Kitto, Director, National Development Bureau, Department of the Interior. ZaThe observation is unique in that it was from the observation platform of a moving train. It is one of the most questionable included here because of the masking n o k of the train and has not been used in the statistical study. 00aQuoted from H. Stuck, F.R.G.S., “Ten Thousand Miles by Dog Sled.” bbbQuotedfrom P. H. Gosse, “The Canadian Naturalist,” p. 47. London, 1840. CCCThe reporters are seamen at latitudes under 63”30‘, presumably Atlantic Ocean. The source is given as Edmonstone, Phil. Trana., 1784. dddQuoted from E. R. Young, “Stories from Indian Wigwams and Northern CampFires.” eeeIn a review of K. R. Koch, “ Resultate der Polarlicht-Beobachtungenangestellt in Winter 1882 und 1883 aufden Stationen Kingua Fjord und Nain’ Berlin: A. Asher, 1886.

+

176

8.

M. SILVERMAN AND T. F. TUAN

We have noted that criticisms of the reality of auroral sound have rested essentially on the basis that trained observers who have spent many years in the Arctic have never heard the sounds and on questions regarding the credibility of untrained observers. Fritz (1881), for example, has listed many observers with long residence in the North who have never heard the sounds, and comparing these with the appreciable number of reliable reporters, concludes that the balance must go against the reality of sounds from the aurora itself, and searches for alternate explanations. This is a common pattern in the literature, usually resulting in meteorological explanations. I n the discussion that follows all negative evidence, except that bearing on geographic extent, has been neglected and attention is focused on the conclusions that can be drawn from the positive reports. An observation tending to reduce credibility has been that of two observers in the same place, one of whom may hear sounds and the other not. In the records listed here only two [47,9213are instances where two observers were present and only one heard the sound. I n one of these [92] the comment is made by the hearer that her hearing is very acute, while her fellow observer's was less acute. The sound was heard while the auroral light was overhead and to the south but was not heard half an hour later when the aurora was to the north. I n eight other cases [12,21,57,77,83,84,93,118];with groups of two to a t least six and possibly more, all, or a t least many, members of the party heard the sound. Included in these groups was one composed of Yale faculty members observing the aurora of August 28, 1827 in New Haven. It thus appears likely that acuity of hearing can be a factor. 3.

CHARACTERISTICS AND

ANALYSIS O F AURORAL SOUND EVENTS

3.1. Frequency of Occurrence

This is almost certainly related to the geomagnetic latitude of the observer. For those individuals who have heard sounds many times [52,67,76,77,107, 108,135,139-1421 the sites are in geomagnetic latitudes from 59" to 68", with the majority being in or near the auroral zone. I n four instances the observers had only heard sounds rarely: (1) in the Yukon, Canada, once in eleven years; (2) in Norway, three times in many years; (3) in Finland, four times in 34 years; and (4) in Eastern Canada, the Central West and Northern Alberta, only once. I n one instance, a t a party in the Yukon in 1908, many people said they had heard the sounds for the first time. To these, of course, may be added the many instances of several years residence without having heard 3Throughout the text these numbers will refer to the catalog given in Table I and the Appendix.

AURORAL AUDIBILITY

177

the sounds. It is difficult to evaluate these negative reports. Many of the positive reports come from individuals who spent much time outdoors. I n the first negative instance above the reporter was a housewife and presumably was generally indoors. For the group of people at the party i t may be presumed that most of their time would normally be indoors because of the typical Arctic weather conditions. For the cases in Norway and Finland we would expect that most of the time the observers would have been much to the south of the auroral zone, though there is insufficient information to judge unequivocally. Beals (19334 has quoted an observer who estimated that only 5% of the brilliant displays he has witnessed gave rise to audible sounds. Stevenson (1853) kept a record of auroras a t Dunse, in north Britain, during the period 1838-1847. I n this period he heard sounds on only three occasions. He recorded a total of 238 auroras of which he considered 21 to be remarkable displays, including the three during which he heard sounds. Thus the frequency of sound events would be of the order of 1 % of all auroras and about 15% of remarkable displays. It will appear later that almost all auroras associated with sound events are remarkable displays which are overhead, and this would be a more useful parameter for our purposes. Unfortunately no data on this point are available. It is thus perhaps safest to say, on the basis of the positive reports, that the sounds will be heard most often in and near the auroral zone, and to leave in abeyance questions as to the fraction of auroral displays which produce sounds.

3.2. Geographic Extent Reports of auroral sound can be found in a band in and several degrees to the south of the auroral zone which extends around the entire northern hemisphere. Attention is restricted here to the more general reports, of the type, “Indians and Eskimos state . . . ” or “ a majority of the residents state . . . ”. Generally the groups involved are those who spend a great deal of time outdoors: hunters, trappers, herders, etc. In North America reports are found from Alaska [82,165,168], and from Canada, where they are attributed to natives [152,154,1551, Indians [57,153,163,141,156], Eskimos [80,153,156,163]. From Greenland there are two reports, one from whale fishers (157) and one from a n unspecified group, possibly the same [158]. Iceland, however, presents a negative report if we restrict ourselves, as is done here, to common belief [174,175]. Positive reports are available from the Shetland Islands, and specifically from the northernmost island, Unst [159], and from the Faroes and Orkney [182,13,183]. For Northern Europe, the first report. may very well be that of Tacitus, who describes sounds connected with what appears to be an auroral description, from a region beyond that of the Suiones, a tribe on the northern borders of Germany [150]. For

178

9. M. SILVERMAN

AND T. F. TUAN

Lapland there are two reports from the same period, one positive and one negative [173,164]. The positive report notes a connection of sounds with strong, energetic auroras a t low temperatures. One report gives a positive indication from Finmark [172]. Finally, to complete the circuit, we have reports from Siberia from the mouth of the Ob [161] and from between the Yenisei and Lena rivers [151,160]. Summarizing these general reports, we may state that positive reports can be found in a band around the northern hemisphere, with the exception of Iceland.

3.3. Latitudinal Dependency These results of the general testimony can be compared with the evidence presented by Davies and Currie (1933) for the Hudson's Bay region of Canada obtained from an extensive inquiry among traders, policemen, missionaries, and Eskimos. All the Caucasians had heard sounds, though more frequently in the auroral zone and more to the south of the zone than to the north. From the Eskimos it appeared that the sounds were infrequent a t the latitude of Chesterfield Inlet, corresponding to about 74" geomagnetic latitude, and were not heard a t all on Southampton Island, within a degree or two of Chesterfield Inlet. These results are consistent with the existence of a band within which the sounds are normally heard, and raise the possibility of a relatively sharp cutoff t o the north. It is possible to obtain a more complete picture of the latitudinal dependency by using the data of Table I. A plot of these data versus the geomagnetic latitude is shown in Fig. 1. The data fall almost entirely into two groups, one between 65" and 70" with a maximum a t 67", and the other a broader band between 55' and 62", with a maximum possibly a t 55". The relative maxima of the two groups are not significant, since these represent individual observations where the dates were noted, and many of the auroral zone reports state that the sounds were heard " many " times, while a t the lower latitudes they are more likely to be unique occurrences. Thus we may anticipate that if it were possible to note every occurrence, the frequency would be heavily weighted towards the auroral zone. The same argument applies to the higher latitudes of the low latitude group, but not to the lower latitudes of this group, so that the position of the maximum here remains uncertain. It appears also that there may be some fine structure in this group, but the data are insufficient to define this or to prove i t with certainty. There appears t o be a high latitude cutoff. Of the four points above 70"geomagnetic, three occurred between 1927 and 1929, straddling the year to sunspot maximum. The fourth point occurred on March 20, 1933 during a magnetic storm with C = 1.6 (compared to a maximum of 2.0) and CKp = 35. There appears also t o be a low latitude cutoff, though not as complete as the high latitude one, a t about

AURORAL AUDIBILITY

179

GEOMAGNETIC LATITUDE

FIQ.1. Number of auroral sound events as a function of geomagnetic latitude. 55". This is consistent with the drop in overhead auroras a t this latitude

reported by Gartlein and Moore (1951). It should be remembered that this drop off represents perhaps a typical situation, and that there will be cases, such as that of September 13,1957 discussed by Obayashi and Hakura (1960), in which the southernmost extent of the aurora will move to lower latitudes for some period during the night. Overall, however, the results are consistent with the statistically observed behavior of overhead auroras in North America. The presence of two belts is somewhat puzzling. The statistical auroral zone is apparently a function primarily of magnetic activity (Stringer and Belon, 1967a). For greater than moderate activity the maximum of the relative incidence, a term used to denote the time average of the number of auroral forms within a given segment of sky and a given time interval, and the southern extent, shift to the south with increasing magnetic activity, but only by one or two degrees. The auroral incidence itself increases and this would give an appreciable increase a t any given latitude. From this statistical result we would anticipate a single zone with greatest frequency near the maximum of the statistical auroral zone. Auroral sound events occur with intense displays on the day on which the storminess reaches a maximum (see below), and it is possible that a particular class of events is involved which may show a different behavior. That this type of behavior is possible is shown by the work of Akasofu and Chapman (1963) on quiet auroral arcs which occur during a magnetically calm period of the main phase. They found that the southernmost extent of these arcs was related to Dst. More interesting

180

9.

M. SILVERMAN AND T. F. TUAN

from our poinh of view is Fig. 4 of their paper, which shows these arcs occurring in a belt from 60” to 50’. Although these arcs are not of the sort which produce sound events, the observed behavior indicates that particular classes of auroras may exist within belts south of the auroral zone. The latitudinal dependency of the observed sound events would then indicate the existence of two classes of auroras which can produce sound, one in and near the auroral zone maximum, and the other in a belt to the south. Very few data are available for the southern hemisphere [71,72,73], and these are all from the Antarctic expeditions of 1911. They are clearly insufficient for data analysis but nonetheless serve to show the existence of auroral sound in both hemispheres. 3.4. Seasonal Dependency

The seasonal dependency derived from the individual observations is given in Table 11, which lists the number for each month of the year. The data, with TABLE 11. Seamnal dependency of auroral sound events Jan 7

Feb 3

Mar 6

Apr 1

May June July 0 0 3

Aug 6

Sept 7

Oct 6

Nov 6

Dec 3

the exception of January are similar to the dependency for overhead aurora for geomagnetic latitudes south of 58” (Gartlein and Moore, 1951). There is a strong fall maximum, and a less well marked spring maximum. The January results may be the results of a somewhat different winter behavior in and near the auroral zone, from where many of the data originate. Some evidence of this can be. found in the results for North America for 1850-1851 (Lefroy, 1852) where incomplete data indicate a winter maximum in February or January for the higher latitude stations. When this is taken into account the data are consistent with the data for overhead aurora. Similar results can be found in the various tables of Lovering (1868), though definitive results do not yet seem to be a t hand. 3.5. Diurnal Variation

A total of 28 reports give some information on the time of day a t which sounds were heard [l,10,18,28,32,34,~,46,52,58,67,69,70,77,70,81,82,83,84,86, 90,91,02,93,122,143,145].Some are merely statements, such as “ after sunset,” most are approximate times, such as “about 9 P.M.,” and a few are accurate to within a few minutes. We may make a simple before and after midnight breakdown utilizing all the statements. This gives 21 occurrences before, 7 after midnight. Table I11 gives the results for those events where times are

181

AURORAL AUDIBILITY

TABLE111. Diurnal variation of auroral sound events (hr., P.M.) ~~~

7-8 3

8-9 1

9-10 3

10-11

I

11-12 2

12-1 2

1-2 3

later than2 2

given, assuming that they are correct to at least the hour. A statement such as “about 9 P.M.” has been put into the 9-10 P.M. group, and similarly for other hours. Other than the peak a t 10-11 P.M. the data are fairly well scattered through the night. The peak a t 10-11 P.M. may be a result of observational selection, with many observers being on their way home from one place or another. It is, however, also consistent with the behavior, so far aa is known, of visual auroras. A definitive study of the diurnal variation, especially of overhead or intense auroras and its relationship to location and other geophysical parameters, has yet t o be made (Chamberlain, 1961, p. 114). To the extent that it is known, the maximum appears to be typically about 9 P.M. local time with a somewhat later maximum a t higher latitudes (Fritz, 1881).There appears t o be some evidence that the geomagnetic time may be a better parameter (Gartlein, Kimball and Sprague, undated; Stringer and Belon, 1967b) but we have not attempted to treat our data in this way. I n any event, it appears that the observations of auroral sounds are consistent in a t least a gross fashion with those of the visual aurora as far as the diurnal variation is concerned.

3.6. Sunspot Cycle Dependency The number of reports have been counted in two ways, in years starting from sunspot minimum and from sunspot maximum. These are tabulated in Table IV and shown graphically in Fig. 2. I n order to maximize the number of data points reports such as “winter 1884-1885” were counted as being in the first year. The sunspot numbers and related data were taken from Chernosky and Hagan (1958). A broad maximum occurs, as might be expected, a t and just after the maximum of the sunspot cycle. An almost linear increase occura on the ascending portion of the cycle, normally of the order of four to five years. Following the maximum the frequency decreases for the next three or four years. A surprising result is the presence of a secondary maximum about five years after the primary maximum. Typical statements regarding the sunspot cycle dependency of the frequency of visual auroras found in most textbooks speak only of a unimodal distribution, possibly lagging that of the sunspot cycle by about two years. Examination of published data, however, shows that the situation is much more complex [see, for example, Gartlein and Moore (1951)or many curves for different locations (Fritz, 1881; Lovering, 1868)],with a considerably more complicated structure even when a

182

5. M. SILVERMAN AND T. F. TUAN

Years from minimum

No. of reports

Years from maximum

2 4

0 1 2 3 4 5

0

6 8 13 10 11

7 8

0 7

9 10 11 12

8 8

2 0

-4 -3 -2 -1 0 1 2 3 4 6 0

3 3

No. of reports

8

7 13 8 10 7 6 11 0 0 2 1

7 8 9

(a 1

0 0

0

a

0

$ 6

8

0

4

0 0 0

0

b

4L

=

eo

d

2 4 6 8 10le YEARS FROM MINIMUM

I

O- 4 - 2

0

f!h 4 6 8 L10

YEARS FROM MbXIMUM

FIG.2. Number of auroral sound events for differentparts of the sunspot cycle starting from (a) sunspot minimum, and (b) sunspot maximum.

183

AURORAL AUDIBILITY

great many observations are averaged together. The existence of two maxima (at least) for some other geophysical parameters has been indicated, but not proven definitively, from work on polar cap absorption events and airglow (see, for example, the discussion by Silverman, 1970), and if the reality and mechanisms of these could be established then the connection with the auroral sound maxima should be of considerable interest. An unexpected and interesting result is obtained when the data are arranged by the different solar cycles. This was originally done to compare the frequency for those cycles with high sunspot number maxima with those of low sunspot cycle maxima. The data are given in Table V. If all of the data are taken together, then a preference, although not strong, is found for sound TABLE V. Auroral sound events for different sunspot cycles

Cycle

Year of minimum

No. of sound events in cycle

-3 -2 -1 1 2 3 4 5 6 7 8 9

1723.5 1734.0 1745.0 1765.2 (March) 1766.5 (May) 1775.5 (Jan) 1784.7 (Sep) 1798.3 (Apr) 1810.6 (Jul) 1823.3 (Apr) 1833.9 (Nov) 1843.5 (Jul)

1 0 0 1

10 11 12 13 14 16

1856.0 (Dec 55) 1867.2 (Mar) 1878.9 (Dec) 1889.6 (Feb 90) 1901.7 (Jan 02) 1913.6 (Aug) 1923.6 (Jul) 1933.8 (Sep) 1944.2 (Feb)

16

17 18

For all cycles C for R,,, > 100 = 33 (8 cycles) I: for R,,, < 100 = 50 (9 cycles) For cycles 8 to 17 C for R,,, > 100 = 24 C for Rm,,< 100= 42

R,,,

R,,, 11 6

4 4 1 1 4 2 7 6

113 106 83.4 86.9 106.1 164.4 132.0 47.6 45.8 71.0 138.3 124.3

5 2 7 8 11 6 11 3

95.7 139.1 63.7 84.9 63.6 103.9 77.8 114.4

(I

9.6 11.4 7.0 10.2 4.1 0.0 1.8 8.5 10.7 4.3 7.3 3.4 6.3 2.7 1.4 5.8 5.7 9.6

C for R,,, > 6 = 33 (8 cycles) Z for R,,, < 6 = 50 (9 cycles)

For cycles -3 t o 7 C for R,,, > 100 = 10 C for R,,, < 100 = 8

184

9. M. SILVERMAN AND T. F. TUAN

reports to occur in cycles with lower sunspot number maxima. Thus, for cycles 1 to 17, the sum for the eight cycles with either sunspot maximum greater than 100 or sunspot minimum greater than 6 is 34, while the sum for the other nine cycles is 50. If, however, we include the three earlier cycles and separate the data into the two groups, cycles -3 to 7, from 1723 to 1833, and cycles 8 to 17, from 1833 to 1944, then we find a very distinct difference in behavior. Thus, for cycles -3 to 7, the sum of the five cycles with maximum sunspot number greater than 100 is 10, while the sum for the other five cycles is 8. Thus, the numbers are approximately equal, with the preference, if any, for the more active cycles. For cycles 8 to 17, on the other hand, the sum for the five cycles with maximum sunspot number greater than 100 is 25, while the sum for the other five cycles is 42. Thus there is a very definite preference for audibility reports to come from the less active cycles and the total of these reports is much greater than for the other group. The results for the M e r e n t cycles are shown graphically in Fig. 3. It might be thought a t first that obser-

W

:9

0

CYCLES -3 to 7 CYCLES at017

> 8

W

0

9 40I *

0

i

60

hi*80

100

120

140

160

R

FIG.3. Number of auroral sound events as a function of the sunspot number at the maximum for different cycles.

vational selection related to the pattern of settlement might be the cause of the marked difference in the two groups. Examination of the data shows, however, that the reports of the earlier years come from Europe and North America, without strong preference for either, so that, while this factor perhaps cannot be completely ruled out, it does not appear to be dominant. A more intriguing connection is that with the sharp decline in the relative frequency of auroras in Europe, shown in Fig. 4, which started about 1788 and continued until about 1826 (Fritz, 1881). The decline is quite dramatic, and it would aeem likely that whatever was the cause of this decline would also be responsible for a change in the distribution of the frequency of audible auroras.

185

AURORAL AUDIBILTTY 350

300 3

2250

v)

ti a

.j 200 W

5> I50 A

W

a I00

50 0

1700

1720

1740

1760

I600 YEAR

I780

1820

1840

I860 1880

FIG.4. The relative yearly sum of auroral occurrences in Europe. The actual number of occurrences are weighted for the extent and magnitude of the auroras (taken from Fritz, 1881, pp. 124ff).

3.7. Correlation with Magnetic Activity Magnetic character figures, C , , are available on a daily basis from 1884 on. Definite dates are given in nine of the sound event reports collected here, and three additional reports give approximate dates. The character figures for the period from three days before to three days after the date on which a sound was reported are listed in Table VI. I n order t o obtain a clearer picture of the relationship with magnetic activity the data for the nine events were averaged. The results are also listed in Table VI and are shown graphically in Fig. 5. The overall picture that results is then that of magnetic storminess reaching a maximum on or shortly after the day on which auroral sounds are heard. The mean maximum storminess of C , = 1.43 may be compared to the maximum possible C, value of 2.0. An attempt was made to broaden the data base by comparing sound events for which the month, rather than just the day, was available with the average C , values for those months. This gave, for a total of nineteen events including those used above, a mean C , = 0.73, which may be compared with the average annual means for the years 1884-1943 of C , = 0.64. The months in which the sound events occurred were thus slightly more stormy than average, but it appears that no real additional information is added by this procedure. I n summary then, on the basis of a limited sample

186

9.M. SILVERMAN AND T. %. TUAN

TABLE VI. 0,Values for days surrounding known dates of auroral sounds

-3

-2

-1

-0

0.8 0.7 0.5 1.0 0.4 0.4 2.0 0.2 1.3

0.6 1.7 0.8 0.8 1.2 1.5 1.0 0.2 1.0

1.4 1.4 0.9 0.5 1.9 1.5 0.9 1.6 0.6

0.6 1.1 1.6 0.6 2.0 1.6 2.0 1.6 2.0

0.81

0.98

1.19

1.43

0.7 0.2 1.9 0.1

1.1 0.1

1.1 0.0 0.1 0.8

1.2 0.9 0.1 0.8

1893 Oct 11 1911 Jan 26 1911 Oct 10 1924 Aug 8 1926 Oct 15 1933 March 20 1938 Jan 25 1938 Sept 27 1941 Sept 18

8, 1908 Sept 10 1919 Oct 15 1927 Jttn 10 12

1.1 0.1

1

2

3

0.8 0.8 1.5 0.1 1.8 1.5 1.8 1.8 2.0

0.9 1.1 0.5 0.1 0.9 1.5 0.8 1.1 1.5

0.9 0.8 0.1 0.6 1.5 0.6 1.6 1.3

1.34

1.06

0.89

1.8 1.2 0.8 0.5

2.0 1.1 0.8 0.7

0.8 1.0 0.5 0.3

0.6

1.4 -

1.2 1.0 -

c

0.8 -

0.2

O’

-4 -1 -: A

I

1

A

SOUND EVENT DAY

FIG.5. Magnetic activity for days around that of auroral sound events.

AURORAL AUDIBILITY

187

of nine events whose dates are known and for which magnetic indices of storminess are available, we conclude that sound events are most likely to occur during days of magnetic storminess. 3.8. Auroral Characteristics

A number of the characteristics of the auroras which produce sound events can be derived from the data. Perhaps the most important of these is the question of whether the aurora was overhead or in some other part of the sky. That the presence of sound was related to aurora being in the magnetic zenith had been stated by Hansteen in 1827: If the observer sees the aurora reach beyond the magnetic zenith, he is surrounded on all sides by the substance of the polar light issuing from the earth: in this cam, if the development is rapid, and he stands in the open field far from any extraneous sounds or noises, he will frequently hear a noise resembling the buzzing caused by the effervescence of a mixture of acid and alkali; but if the aurora does not reach his zenith, i.e. if he stands beyond the region from which the emanation takes place, and sees it low in the north or south, he will not hear such a noise.

Stormer (1955),on the basis of his assistant's reports, has stated that the sounds occur sometimes when a fine auroral corona is present. The corona is, of course, an effect of perspective and is equivalent to a statement of an overhead aurora. In the data presented here it is possible to obtain some information on location or extent in 32 cases. Of these, 25 indicate at least a portion of the auroral display to be overhead. Nine are recognizable as coronas [10,27,39,50,83,90,118,128,129]; in five the display was over the whole sky [32,49,69,139,145];in ten a streamer or ray of the southern border of the display was in the zenith or passed overhead [28,34,43,65,68,76,77,82,89,92];and in one [18] the aurora had passed from NNW to E by S. In five other instances [27,46,47,66,147]it appears likely or possible that a portion of the aurora was overhead. In only two cases [70,88], does it seem that the aurora was unambiguously in a different portion of the sky. In one [MI, the observer was at the magnetic dip pole. In the other [70], the sounds were heard after supper, the aurora being like a bow in the north, and had been preceded, before supper, by a very active aurora which had streamed across the sky from the west through the zenith to the east. Overall, then, it seems that in the vast majority of cases the aurora must be overhead before sound events occur, but that in rare cases sounds may occur for auroras in other parts of the sky. We may close this paragraph with the description given in Chant (1923a) (see also [77] in the Appendix): We watched this display approaching from the north. At first there was no sound, but a8 it got nearer, we heard a subdued swishing sound, which grew more distinct as it approached, and was loudest when the ribbon or belt of light was

188

9. M . SILVERMAN AND T. F. TUAN

. ..

right overhead in a few minutes the whole sky was full of auror&l streamers which seemed to culminate at a point in the zenith. A few minutes after the first display had passed over our heads, however, we could not hear the sound. It appeared as if the display was too far distant for the sound to reach us, andan hour after our attention was first drawn to it the display had faded to quite an ordinary one.

An almost universal characteristic is the association with moving, usually rapidly moving, forms. Of more than 40 reports, only one [68] seems to indicate the absence of motion, with the statement that there were no streamers and no corona but only two well-defined arcs. The overwhelming mass of evidence is, however, in favor of an association with movement. I n two other instances [20,70], sounds were heard from arcs, and these may be presumed to be a t least relatively motionless. Occurrence of sound is associated with the activity of the aurora in most cases. Thus Indians and Eskimos in Canada state that sound occurs with flickering, rapidly moving auroras [156], and Alaskan miners and trappers state that it is associated with active displays [168]. Individual descriptions include statements such as “when the aurora was a t its greatest intensity and waving in the sky like a blanket or a sheet ” [134] or “ the sound lasted about ten minutes, rose to a maximum and fell down again, following the intensity of the aurora ” [go]. Many similar descriptions associate the sound with either the intensity variation or general movement of the aurora [32,44,64,65,77,78, 83,84,126,131,132].More often the sound is associated with streamers, flashes, or rays, that is, with the more localized aspects of an auroral display [4,8,8, 15,20,21,24,26,27,28,34,46,47,49,50,55,56,61,62,75,82,101,118,137,139,149]. In almost all cases the visual movements of the aurora and the sounds are said to occur simultaneously. I n one case [50] it followed the movement closely, within only one second or less, which is certainly much less than the time required for sound to travel from auroral heights. I n another case [89], the definite statement is made that it was not simultaneous with the greenishwhite flashes which were darting overhead from south to north. This instance is one of the few from north of the auroral zone, and this may have a bearing on the lack of simultaneity. I n one instance, however, at about 60’ geomagnetic latitude, the sound was heard when ‘‘ a broad yellowish splash of flame spread across from the west to the east, ascending from the horizon” [46]. It would be of interest to have more information on the connection, if any, with auroral surges. Some idea of the intensity of the auroras associated with sound events can be obtained from some 37 instances, if one includes such descriptions as “remarkable ” or “ brilliant ” displays as referring to a more than usually intense aurora. Of the 37 cases, only one [145] describes the aurora as not very intense. The general statements of Alaskan trappers and miners [168], Canadi-

AURORAL AUDIBILITY

189

an Eskimos, Indians, and Caucasians [155,156,167], Shetlanders [171] and Laplanders [1641 agree that sounds are heard with strong, intense, or brilliant auroras. Individual descriptions using such phrases as ‘‘ unusually brilliant,” “ greatest display ever witnessed,” or ‘‘ magnificent ” account for the great bulk of all reports [39,46,47,52,53,55,60,66,67,68,75,76,77,82,83,8~90,96,101, 118,124,129,131,139].In two cases it is reported that the aurora was so intense that it was either heard before darkness [93] or traces of it remained after daylight [82]. Five of the reports provide semiquantitative information which with sufficient effort could provide an estimate of brightness. These are (i) light enough to read the finest print [41]; (ii) both banks of a river of considerable width visible [79]; (iii)light enough to see that the leaves on the tree were motionless [92]; (iv) as bright as the moon in quarters [lo];and (v)the brighter stars visible and a t times none of the smaller ones [50]. No attempt was made to evaluate these various reports, but it is clear that the auroras involved were quite luminous. The report on star visibility, if completely reliable, would indicate a twilight condition corresponding t o approximately three or four degrees solar depression angle (Tousey and Koomen, 1953) since the brightest stars are greater than first magnitude (van de Kamp, 1953). Studies of star observations during total solar eclipses, however, indicate that these tend seriously to overestimate the brightness (S. M. Silverman and E. G. Mullen, unpublished results). Even when this is taken into account, however, the auroral brightness can be estimated to have been of the order of that for a solar depression angle of 6”, or quitrebright. When all evidence is considered, then, it appears that sound events are associated with only the more intense or most intense auroras. Some information on colors is available in 19 reports [32,34,43,44,46,50,56, 61,62,75,76,82,83,89,90,126,129,143).The descriptions range from the simple notation that the aurora was distinctly colored, through white to all the colors of the rainbow. We have not attempted to correlate the colors with any particular lines or bands. It appears clear, however, that a goodly degree of excitation is involved, consistent with the intensity discussed in the preceding paragraph. The duration of the sound is explicitly mentioned in 9 reports [30,76,77,8284,90,118,130]. These range from a few moments to more than an hour. No particular time seems to be favored, though minutes are most often mentioned. If we remember that the sound is often noticed in coincidence with rapid movements it would seem likely that the duration would be more heavily weighted towards shorter times. An interesting statement is made in one report [147], that the sounds emitted are a hissing followed by a crackling, with the interval from the first hissing to the crackling being about two seconds. On two occasions [83,90] heights of a portion of the auroral display a t times

190

S. M . SILVERMAN AND T. F. TUAN

near to those when sounds were heard were reported by Stcarmer. For the aurora of October 15, 1926, sounds were heard during the period from 1915 to 1925. The lower borders of three curtains photographed a t 19h05m46s were 103, 110, and 124 km. For the aurora of January 25, 1938, the heights were not lower than 95 km (Stcarmer, 1955). Currie (1955) has published heights which include the period March 21-27, 1933 immediately after the March 20, 1933 event [89] and these do not seem to be lower than normal. Other than these few, no height determinations are available. Rapidly moving forms are not suitable for height determinations, and since these seem to be the primary source of sound events we must be careful in applying the height determinations in discussions of the sound data. We close this section with the radio observations of Jelstrup and the visual observations of Rostad (quoted by Stcarmer, 1955) during the aurora of October 15, 1926. These are the most complete notes in any report. From 20h. lm. to 20h. 6m. (Greenwich Civil Time) we registered on our radioreceiving set the rhythmic time-signals from the LY station (Bordeaux). We secured the whole series of tops-but a t the same time the “aurora statics ” disturbed the pen of the registering instrument. The impulses thus registered are of varying strength, and each of them is of course exceptionally well determined in time, being ‘‘ received ” all a t the same time as the scientific time signals. I therefore think that they may be of some interest. The maximum impulses of ‘‘ aurora statics ” and their duration were:

No. 1 2 3 4

Greenwich Civil Time 20h. 20h. 20h. 20h.

4m. 4m. 4m. 4m.

28.60s. 29.49s. 39.90s. 40.50s.

Duration 0.085. 0.10s.

0.258. 0.25s.

Wavelength 18,900 metres (Jelstrup) As regards the intensity of these impulses, I find that in each case the vertical component was greater than 100 microvolt/metre. When, after, the reception of the time signals, we again went out of the observatory, the curious sound had absolutely ceamd, and later in the night, when the aurora had also vanished, we noticed that the atmosphere was as if swept clean from statics and disturbances of our wavelength. JELSTRUP

I9h. 10m. Dense masses of rays and curtains down to the horizon in E and SE. 19h. 12m. The same down t o the horizon in W, to the polar star in N, to the horizon in E, and down to 40’ over the horizon in S. 19h. 14m. The same to the Great Bear in N. 19h. 16m. The same in the N and S down to 15” over the horizon. Red in 8. 19h. 21m. Strong diffuse arc through Can. venat. from the horizon in NW to the horizon in NE. 1Qh.24m. Pulsating aurora begins. 19h. 26m. Strong pulsations over the whole heavens. Red in 8.

AURORAL AUDIBILITY

191

During the radio signals mentioned by M. Jelstrup the following observations are noted: 20h. 3m. The pulsations have ended. All over the heavens a diffuse light. I n the pocket spectroscope the yellow-green auroral line could be seen everywhere, from the entire heavens, from the snow, and from everything which was illuminated by the diffuse aurora. 20h. 9m. Some pulsating bundles of rays. 20h. 12m. The pulsations stronger and rays begin t o appear. R~STAD

3.9. Localization of Effect An important question from the point of view of the physical basis of the phenomenon is the question of the spatial extent. Unfortunately, there is an almost complete lack of information on the subject. One observer [77] who heard the sound notes that the effect ‘‘ seems t o be more or less local, as people living a few miles away, while noticing that there was a brilliant aurora, did not notice the accompanying sound.” Broad limits can be obtained from two reports. I n one report [go] the sound was heard on a mountain and in the valley below. Another report notes that the sound was not heard, by an observer who had previously heard such sounds, a t a point 60-65 km away. For the aurora of August 28, 1827 sounds were heard in Utica and St. Lawrence County, New York, and in New Haven, Connecticut [19-211, but since no time information is given it is impossible to decide whether they were simultaneous or a t different phases of the event. I n summary, then, the very sparse information available indicates that the effect occurs on a scale of kilometers or less, but that this estimate must be treated with great caution.

3.10. Weather Some information about the weather is given in 38 reports. The most common statements are that the night was still or calm (24 reports), with only one report mentioning a breeze. Another common statement (13 reports) is that the night was clear or bright, with one noting that the stars seemed to assume an unusual brilliancy. A certain clarity of sky must in any event be present before the aurora would be seen. The most common observation other than these, and almost the only one, was that the night was cold. this being noted some 15 times, with temperatures ranging to as low as -50°F. Despite the large number of mentions this does not appear to be a necessary condition, since there are a number of observations on summer nights or otherwise mild nights (see, for example [49,65,68,86,1 IS]). To summarize, the most common condition seems to be that of a still, clear night.

192

9. M. SILVERMAN AND T. F. TUAN

3.11. Eflect of Altitude and Terrain

A reading of the various reports leads to the impression that audibility may be more prevalent a t higher altitudes. I n one report [141] it is specifically stated t ha t Indiana who had heard sounds had heard them especially when camped a t higher altitudes. There seems to be no way of definitely proving this from the available data. I n one instance [go], sounds were heard both on a mountain and in the valley below. Many of the reports are from the prairies of Saskatchewan, and one [28] from a treeless plain. Reports have ranged from eighteenth century London and nineteenth century Connecticut, for example, to the Rocky mountains at the Arctic Circle. It would seem that any reliable indication of the effect of either altitude or terrain would necessitate a much more complete knowledge of location and contemporaneous topography than is available or is likely to be available. We therefore leave this question unresolved. 3.12. Association with Low Aurora

By low aurora we mean luminous effects from ground level to perhaps cloud heights. Reports of low aurora and of people walking in the midst of aurora recur with remarkable frequency. An extended, essentially popular, description of such an event is given by Stumbles (1938). In a t least twenty one of our reports of sound, a statement of low aurora was included [31,34,41, 45,49,50,58,61,62,65,74,79,81,82,84,86,130,143-145,147]. The association of sound with low aurora, though common, is by no means universal. In most accounts the auroral displays are in their accustomed places in the sky. Furthermore there are many accounts of low auroras with no accompanying sound. A good example is that observed at Fort Conger, near the magnetic pole, on November 17, 1882 (Greely, 1886) which in its intensity and movement was similar to many which have produced reports of auroral sounds. The aurora was such that two men unconsciously dodged to avoid it. The intensity waa judged equal to that of the full moon, and only first magnitude stars were visible. No sound, however, wm audible. A number of explanations have been proposed for the low aurora. Davies and Currie (1933) and others have proposed that an event of this sort that they witnessed was due to an intense aurora covering almost the whole sky coupled with reflected light from the snow covered ground. A number of observations have occurred during the summer months (see, for example, [49,65]). Simpson (1918) discussed some observations of aurora apparently in front of or on the side of Mt. Erebus in the Antarctic, and concluded that these were optical illusions. Observations have been reported, however, from other types

AURORBL AUDIBILITY

193

of terrain, such as open prairie land in Saskatchewan (Stumbles, 1938) where the conditions which Simpson noted for the illusions were not present. At this time no ready explanation is a t hand for all the observations. The phenomenon may be related to the feeble luminosity observed around mountain tops and forests in the polar regions by Lemstrom (1898). If an electric field were present then, on a very still night with low humidity, there would be nothing to remove the space charge surrounding any points and therefore the field would continue to build up to a value when a discharge sufficient to cause a sound would occur. But if there were wind and humidity, the electric field could be less and there could be a continuous leak from the electrodes causing some luminosity. Since our interest here is in audibility we have not attempted to collect all the observations of low auroras or to define their characteristics. We summarize our discussion merely by stating that while the association of low aurora with audibility is common, the two do not appear to be necessarily connected.

3.13. Association with Odor

A few reports have noted the presence of an odor at the time of an auroral display, described as either like a strong smell of sulfur or like ozone. Beals (1933~) mentions a number of reports of odor during displays when they were accompanied by sounds. During the aurora of October 19, 1726 (new style), sounds and odor were noted at London, though whether these were simultaneous or not is not noted [2]. I n 1870, Rollier, the balloonist, descended on a mountain in Norway 1300 meters high, saw auroral rays across a thin mist, and heard a muttering. When the sound ceased he perceived a very strong smell of sulfur [45]. This is the only account in which any sense of time sequence is given. I n one instance (Zanotti, 1739-1740) an odor was noted without any mention of accompanying sounds. It was noted, however, that this had also been observed during other auroral displays. Trevelyan [182] quotes the report of a person in the Faroes, “ who stated that, when the color of the Aurora Borealis is dark red, and extends from west t o east with a violent motion, he had experienced a smell similar to that which is perceived when an electric machine is in action.” It appears, therefore, that a n odor can accompany an auroral display, possibly even when sounds are not present. The simultaneous noting of both sound and odor would not be expected to be a common occurrence, even if both were always present, because of the masking effect which might be anticipated for either or both from the normal noise and odor environment. The few instances recorded are thus more significant than indicated by their numbers.

194

8. M. SILVERMAN AND T. F. TUAN

3.14. Audibility While the Aurora Was Not Visible Because of its importance in eliminating the theory of psychologicalguilt by association, we note here that in a number of instances the aurora has been heard before being seen. Thus, a lighthouse keeper in the Shetlands was able to tell when a n aurora was in progress even while the shutters were closed [110]; an employee of the Canadian Forest Service notes that he often “ heard it where the blanket was over my face and I did not know that ‘ lights were on , and, in a possibly suspicious case, a Sergeant of the Royal Canadian Mounted Police after going back into the snow house asked the Eskimo who was with him whether he had heard any noise and was told that “he had heard the northern lights move a short time ago ”.4 I n another case the reporter, lying awake in his tent, heard a mild, crackling noise, dashed outside to see whether the fire had been extinguished, and found the fire dead and an auroral display to the north greater than he had ever seen [46]. A less convincing case is that of the couple who were hurrying home with heads bent down into fur storm-collars, the temperature being about 50 below zero, and who only looked up when they heard unusual sounds above them [69]. In this case the possibility of reflected auroral light calling attention to their presence exists. A possibility of reflected light is also present in another instance [180] where the observer, sitting on a hill, heard sounds behind him and, turning, saw the aurora. A more interesting account is that of sounds being heard before dark and only being associated with the aurora as it grew darker [93]. This account is also unique as the only instance of daylight sounds. I n two instances men were blindfolded and asked whether they heard sounds. I n the first [47] Mr. William Ogilvie describes the result of the experiment thus “At nearly every brilliant rush of the auroral light he exclaimed, ‘Don’t you hear it?’ All the time I was unconscious of any sensation of sound.’’ The second case [12] is a good example of a pre-judged and biased experiment. Mr. David Thompson and his men all heard the aurora, but, for Mr. Thompson, “ reason told me that I did not but it was cool reason against sense,” and he was persuaded that it was a case of the eye deceiving the ear. He thereupon blindfolded the men by turns, with the result that “ they soon became sensible that they did not and yet so powerful was the illusion of the eye on the ear that they still believed that they heard the Aurora.’’ It is not necessary t o comment further on the validity of the experiment. Overall then it appears that there are sufficient cases to indicate that auroral sounds can be heard in the absence of visual stimulation. $.*,

40n close reading the report does not state explicitly that the Eskimo was in the snow house the entire time, though the context and use of pronouns indicates this. I therefore asked six people to read the account and then tell me where the Eskimo was. Five of the six felt he had been in the snow house the entire time.

AURORAL AUDIBILITY

195

3.15. Effects on Animals In five reports, effects on dogs are noted. These are of interest as representing a response from a species whose psychology and prejudices should be distinctly different from the human. The earliest such report is that of Gmelin in the eighteenth century, in a report on a Siberian trip [151]. He states that the hunter’s dogs lie obstinately on the ground till the noise had passed, A twentieth century reporter [142] notes that dogs whine and turn about in circles when the sound becomes noticeable, and that this is especially true of the white Siberian breed. The other reports state that the dogs ears stand up as if they were listening [141]; that “when thc swishing noise came so suddenly and apparently so close, most of them immediately jumped up and commenced to growl ” [84];and, finally, simply that the popping sound scared the dogs. No reports have noted effects on other animals and it is possible that differences in frequency range of audibility between species may be involved, but we have not attempted to check this out.

3.16. Auroral Sounds in Poetry Before I see another day, Oh let my body dio away! I n sleep I heard the northern gleams; The stars, they were among my dreams; In rustling conflict through the skies, I heard, I saw the flashes drive, And yet they are upon my eyes, And yet I am alive: Before I see another day, Oh let my body die away!

These are the opening lines of Wordsworth’s poem, ”The Complaint of a Forsaken Indian Woman ” composed in 1798. Lest it be considered fanciful to ascribe the description here to auroral sounds we quote here Wordsworth’s notes on the origin of the poem: Written at Alfoxclen, where I read Hearno’s Journey with deep interest. It was composed for the volrnne of Lyrical Rallatls. Il’lien a Northern Indian, from sickness, is unable to continue his journey with his companions, he is left behind, covered over with (leer-skins, and is supplied with water, food, and fuel, if the sitimtion of the place will afford it. He is informed of the t,ra.cakwhich his companions intend to pursue, and if he be unable to follow, or overtake them, he perishes alone in the desert; unless he should have the good fortune t o fall in with some other tribes of Indians. The females are equally, or still more, exposed to the same fate. See that very interesting work Hearne’s ,Journey from Hudson’s Ray to tho Northern Ocean. In the high northern latitudes, as the same writer informs us, when the northern lights vary their position in the air, they make a rustling and a crackling noise, as alluded to in the following poem.

196

9. M . SILVERMAN AND

T. F.

TUAN

It is interesting to note that Wordsworth was a close friend of John Gough, the blind botanist and a most unusual man, and had described him‘in one of his poems. Gough was a teacher of John Dalton, the founder of atomic theory and a student of the aurora for many years. (For these connections, see Watson, 1894.) Dalton’s interest in the aurora extended over many years, and is exemplified by the journal of auroras he kept between 1786 and 1793 (Dalton, 1834) and his calculations of the height of the aurora of March 29, 1826 (Dalton, 1828). In 1788, ten years before Wordsworth’s poem, Dalton noted an aurora which was said to have been heard, and it is tempting to speculate that Wordsworth’s interest may have been aroused not only by Hearne’s book but by Dalton’s observations. We have not, however, attempted to confirm this, and mention it here only because of its interest in connection with the environment of English science a t this period. Robert Burns, the Scottishpoet, hasgivenagooddescriptionin “AVision ” The cauld blue north was streaming forth Her lights, wi’ hissing eerie din; Athort the lift they start and shift, Like fortune’s favours tint as won.

A recent description from the North of England is (Simms, 1971): Beyond sky curtains were being drawn back on the heaviest and smoothest of brass ring%. The sound was as if rhythms unrealized within were the whole round of heaven. Not, as in thunder, being rolled about, a gill-brack about to descend: this was a levitation of things. From “Aura” by Colin Simms (1967)

Petrie (1963) has quoted the following extracts; They writhed like a brood of angry snakes, hissing and sulphur pale. They rolled around with a soundless sounds like softly bruised silk. From “The Ballad of the Northern Lights,” by Robert W. Service And close ranked spears of gold and blue, Thin scarlet and thin green, Hurtled and clashed across the sphere And hissed in sibilant whisperings, And died. From “The Iceberg,” by Sir Charles G. D. Robert

You tink you hear da skruffel 0 der lang green goons o sylk From “Nordern Lichta,” by T. A. Robertson

AURORAL AUDIBILITY

197

If you listen, when weirdly the lights are streaming, Perhaps you may hear a whisper low. From “Northern Lights,” by T. A. Robertson

These all give the sense of personal experience. It appears then that auroral sounds are not uncommon in Scotland and Canada, in accord with our other data.

3.17. Possibly Related Phenomena-Sounds from Lightning and Meteors Sounds similar to those reported for aurora have been reported as an instantaneous accompaniment of both lightning and meteors. These are distinct from the sounds which follow the visual observation after some length of time, that is, thunder and a similar sound apparently due to the rare fireball which breaks up a t relatively low altitudes. Brontophonics, that is, sounds associated instantaneously. with lightning discharges, have been reported on rare occasions (Hubbard, 1881; Cooke, quoted in Griffin, 1922; Astapowitsch, quoted in Romig and Lamar, 1963). On one occasion (Constable, 1881) a crackling and rustling was heard all around the observer in conjunction with a lightning storm near the horizon. Astapowitsch (1934) has stated that they are produced by the leader strokes, and can be explained by an induced charge which appears at more than one point in the surroundings. We would not a priori anticipate that brontophonics would be of common occurrence since the sounds would be expected to be relatively faint and would easily be masked by the rain and wind noises normally present a t such times. Romig and Lamar (1963) have cataloged observations of sounds and radio interference for 42 fireballs. They also note that an occasional smell of ozone as well as electromagnetic disturbances suggest that a local corona discharge is responsible for the hissing noises. We may note here that the odor of ozone is also occasionally reported in connection with auroral sound events. Astapowitsch (1934)has stated that the phenomenon has been observed for several score of fireballs. A popular discussion with a negative conclusion is given by Ley (1969). The same conclusion is also reached by the very experienced observer Denning (1902-1903). He quotes a number of accounts of hissing and rushing noises associated with meteors, including one by a n experienced observer, and notes that for the thirty year period of fireball reports to which he referred, the proportion of accounts where noises were heard was a small proportion ofthe whole. He also quotes one account in which a strong smell of sulfur was noticed “ immediately after the meteor burst.)’ He nevertheless concludes that the sounds are subjective in origin on the ‘basis of his own observations of some thousands of meteors on very quiet, calm nights without ever having heard anything. Let us note here that if the situation is analogous to that of the aurora he would have had to have been directly under a very

198

S. M. SILVERMAN AND T. F. TUAN

bright meteor and that all sky visibility would not be an appropriate criterion. No attempt is made here to make a collection of reports of sounds from lightning or meteors. The above remarks are of interest, however, in indicating a possible mechanistic connection between diverse physical phenomena.

3.18. Miscellaneous Astapowitsch (quoted in Romig and Lamar, 1963) has stated that auroral sounds have been recorded on tape. Attempts at the Geophysical Institute, University of Alaska, t o record such sounds using directive, high sensitivity, audio equipment covering the range 10-200 kHz have failed to provide conclusive results. This may, as the experimenters have recognized, be a result of the fact that the experiments were carried out from 1962-1964, too close to sunspot minimum (University of Alaska, Geophys. Inst. Ann. Repts. 19621963, 1963-1964). Reports by Eskimos from Western Alaska and across Canada and by Caucasian Alaskans indicate that when auroral sounds can be heard then the aurora can be manipulated by whistling (Ray, 1958). Mrs. Ray describes a fifteen minute demonstration by a friend in Alaska for her benefit in which the whistling drew the aurora nearer or further from the ground. Garber (1933) has also, skeptically, reported this as an Eskimo belief. Graah (in Force, 1856, p. 21) stated that the Greenlanders reported that the aurora drew nearer when they whistled. OF AURORALAUDIBILITY 4. HYPOTHESES

We present here discussions of the various hypotheses which have been proposed to explain or explain away the reports of auroral audibility. I n all of the past discussions no attempt seems to have been made to use the full amount of the data nor has any serious effort been made to correlate these data with other geophysical parameters. The reality of auroral audibility has therefore always remained questionable. It appears also that the observational evidence is in itself insufficient for proof, and that for this some theory must be presented to make the phenomenon physically plausible. 4.1. Psychological

This is based essentially on the premise that the mind tends to fill in elements which it expects to find in a situation even when these are in reality absent. There is no question but that affect-full real situations with which people cannot cope are quite often treated by a distortion or denial of reality (see, for example, Rank, 1927; Freud, 1946; Miller and Swanson, 1960). Also,

AURORAL AUDIBILITY

199

medical practitioners find that an appreciable fraction of their patients are susceptible to placebo, so that suggestibility is clearly important in some situations. It has been argued (for example, by Burder, 1881; Hubbard, 1881) that the close connection between the senses of hearing and seeing would lead to an expectation of sound in a vivid auroral display, and that the mind would then provide this. If this were so, however, one would expect a greater frequency of reports. Furthermore, most observers have noted sound only rarely, while if suggestibility were a factor one would expect two groups, one who never heard sounds and one who almost always heard sounds. I n and near the auroral zone, however, almost all natives are in this class, of those having heard sounds, though not always, and even allowing for cultural differences this would seem t o rule out a placebo effect. Perhaps the severest difficulties of a psychological theory are the connections with geophysical parameters brought out in Section 3. Maintaining a psychological origin in the face of these unexpected realities would require the introduction of strong influences on our psychology by the sunspot cycle, magnetic activity, etc., and there seems a t this time to be insuficient evidence for such connections (for examples of opposing views, see Presman, 1970; Pokorny and Mefferd, 1966). Above all, psychological reaction should be a two-edged explanation. Just as a nonprofessional observer may be expected t o hear something when he expects it, a trained professional who does not expect to hear anything may be expected not to hear it. But as shown in section 1 there are a significant number of professional observers who not only heard the sounds but some even heard them simultaneously with the visual observations for both aurora and lightning, a phenomenon hardly consistent with what a trained physicist would normally expect. A psychological origin for auroral audibility must therefore be ruled out.

4.2. Tinnitus Sexton (1886) proposed that the sound was simply the ringing in the ears which occurs normally in the absence of physical sound, and which is known as tinnitus. These sounds (Davis, 1951) seem to represent a spontaneous discharge of elements in the auditory system and are due usually to an irritation or hypersensitivity of the hair cells or their nerves. A small amount of activity seems to be normal, but is often present and can be symptomatic of several diseases involving the ear. The pitch is sometimes distinct and sometimes resembles a vague low-pitched roar or a wide band of noise. The hypothesis of tinnitus as a source of auroral audibility suffers from too many implausibilities to be taken seriously. There is first of all the fact that descriptions of the auroral sounds are quite different from those noted above for tinnitus. Second, and equally invalidating, is the fact that the occurrence

200

9.

M. SILVERMAN AND T. F. TUAN

of tinnitus is independent of auroral frequency, and should therefore be heard for all types of auroras and even in the total absence of aurora. With one possible exception, all reports of sounds have come from auroras in motion, and sounds have been listened for on non-auroral nights without result (see, for example, 139). This explanation must therefore be completely ruled out. 4.3. Swishing Breath

When the air temperature is below -40°F and the air is calm, a swishing or hissing sound can be heard (Sverdrup, 1931; Henry, 1886) which appears to come from a point either to the right or left of the observer when in motion (Henry). The sound has been attributed either t o the collision of ice crystals which are formed on exhalation on their way to the ground (Sverdrup) or the sudden volume reduction of the saturated air leaving the lungs as it encounters the very cold atmosphere (Henry). The attribution of auroral sounds to this phenomenon has been made by Amundsen (quoted, for example, in Griffin, 1922)) Davies (quoted in Beals, 1933a))and, possibly by implication Henry (1886). Beals has pointed out that the basic objections to this are that the sounds should then also be heard on numerous occasions in the absence of aurora, and, a more forceful argument, that they have been heard on many occasions in summer and fall when the temperature was much greater than -40". I n the reports collected here there are several [49,65,77,86,89,118] in which the temperature is above -30", and several others where this appears very likely [9,10,46,54,81], so that this explanation is insufficient. Furthermore one would anticipate that those who had spent some time in very cold weather conditions would have been sufficiently accustomed to the swishing breath to distinguish it from other sounds heard during a n auroral display as was true for a Mr. Ritch [25]. Thus, while the explanation has a certain appealing plausibility it appears nevertheless to be ruled out for most reports.

4.4. Meteorological Theories These are essentially statements that snow and ice in some fashion produce sounds. The simplest of these was that of Sir George Nares, the Arctic explorer, who had never heard the sounds and supposed that it was merely the breakingup of the ice and the grinding of the icebergs (quoted in Webber, 1887). Count Trampe (Fritz, 1881, p. 279) believed that it came from snow and ice because of the cold. Mr. Wentzel of Franklin's party assured them that the noise was caused by severe cold succeeding mild weather and acting on the surface of snow previously melted by the sun (Capron, 1879; Force, 1856, p. 38). Wentzel[18], however, was familiar with the sound of the aurora, and was able to distinguish it from those of meteorological origin. Finally,

AURORAL AUDIBILITY

20 1

Richardson (Chant, 1916) stated that the formation of minute icy spiculae in very cold clear nights was accompanied by a crackling in the air. All of these fail for the reason that in a good many reports the presence of snow and ice is not mentioned. In addition, Nares explanation suffers strongly from the fact that there are no icebergs in many of the localities, as, for example, the prairies of Saskatchewan, or London. Overall, then, theories involving snow or ice in some form or other must be rejected. 4.5. Direct Transmission of Audible Sound

Almost all auroras occur at heights greater than 80 km (Stmmer, 1955; Egeland and Omholt, 1967). I n one instance an aurora has been observed as low as 60-70 km (Stsrmer, 1955, p. 80).On a t least two occasions during which sound was heard from an aurora the height was 90-100 km (Stmmer, 1955, p. 139). The low auroras sometimes reported with sound events (see Section 3.12) seem to be fundamentally different and are not considered here. We must consider, then, whether audible sound can be produced from auroras a t heights greater than 60 km. Gartlein e l al. (1967) have pointed out that, since the wavelength of a sound wave must be greater than the mean free path of the molecules in the gas, an upper limit on the frequency must exist a t any altitude. This is given by c/h,(h),where c is the velocity of sound, and A&) is the mean free path a t height h. At 60 km, the mean free path of air molecules meters (Cole et al., 1965). This means that the upper is about 2.65 x limit on frequency transmitted from a height of 60 km is about 113 x lo4 Hz which is sufficiently high to include practically all the sound in the audible range. At 100 km the upper frequency limit is about 2000 Hz, so that much of the audible range is accessible. However, attenuation of high frequency sounds is in general much higher than a t lower frequencies. Even if we neglect inverse square attenuation, the highest frequency which will allow 0.1 % of the initial energy to be transmitted to the ground level from a height of about 60 km is about 40 Hz (Procunier and Sharp, 1971). The minimum pressure change for a n audible sound a t 40 Hz is about 0.1 dyn/cm2 (Fletcher, 1923). Assuming an averagesoundvelocity of about 300 m/s, the energy flux required to produce this pressure variation is about 3000 erg/cm2s a t the ground level. This means that we require an initial energy flux a t 60 km height of about 3 x lo6 erg/cm2s which exceeds the particle energy flux required to produce the brightest aurora by three orders of magnitude. If we go down in frequency to 2 Hz, then 99% of the initial energy a t 60 km is transmitted. However, a t this frequency the threshold of audibility is something like 200 dyn/cm2, which means an energy flux of the order of 6 x lo7 erg/cmas a t 60 km altitude since we have practically no attenuation.

202

S. M. SILVERMAN AND T. F. TUAN

This figure is again far too large when compared with incident particle flux for even the brightest aurora. Furthermore, the frequencies we are dealing with are hardly compatible with anything like a hissing or swishing sound. Hence on the basis of energy considerations alone, direct transmission of auroral sound must be ruled out. In addition, we have noted earlier that almost all observers report that the visual movements and the sound occur simultaneously. If the sound were in fact produced a t heights of 60 km or greater then a time lag of several minutes would occur between sight and sound. The argument is, however, not conclusive, since if the aurora remains in rapid motion for several minutes then sound would always be associated with movement, though not necessarily always simultaneous. The sense of the accounts is nevertheless strongly suggestive of a real, as opposed to perceived, simultaneity, so that an origin at auroral heights would be ruled out. 4.6. Infrasonic Waves

It is known both experimentally and theoretically that auroral activity can produce infrasonic waves which can propagate for long distances (see, for example, Maeda and Watanabe, 1964; Maeda and Young, 1966; Campbell and Young, 1963; Wilson 1967,1969). These waves have periods which are usually in the 20 to 80 second range, but are occasionally longer, and pressure amplitude ranges from about 1 to 10 dyn/cm2, averaging about 2 dyn/cma (Wilson, 1969). The propagation velocity is in general slower than the acoustic waves, and one would thus expect a rather long time lag between a visual observation and an infrasonic wave associated with it. I n view of the simultaneity noted in the observations, these would therefore appear to be ruled out as a source of auroral sounds. A more serious objectior, is the fact that these are well below the audible region of the ear, and the threshold of hearing at these frequencies involves quite large pressures. The minimum audible pressure threshold of hearing for the human ear has been experimentally measured down t o 1 Hz (see Yeowart et al., 1967, and the references cited therein). At 1 Hz acoustic pressures of up to 2000 dyn/cma are required, and this is well above the pressures available from infrasonics. It is interesting to note the descriptions given of the low frequency sound stimuli. Above 20 Hz these are described as smooth or tonal, between 5 and 15 Hz as rough, or as a popping effect, and below 5 Hz as chugging or whooshing. Below 5 Hz the subjects reported that they could ‘ifeel” the acoustic stimulus, and a t 1.5 Hz they found the stimulus somewhat unpleasant. These descriptions of sounds are inconsistent with the observations of auroral sounds, corroborating the negation of infrasonics as a source of auroral sounds.

AURORAL AUDIBILITY

203

Procunier (1971) has published some data on measurements with Helmholtz resonators in the 1-16 Hz range, including a figure of an impulsive event as recorded a t 4 Hz, with a pressure variation of the order of 10 dyn/cm2 (Procunier, private communication). This raises the possibility that a sufficiently low source might give pressures high enough to be audible. The objections of simultaneity and the nature of the sounds would nevertheless still be valid. I n addition the events described by Procunier are impulsive, while the sound events are continuous in character. While this hypothesis seems to be ruled out by these considerations additional work along these lines would still be of interest.

4.7. Direct Perception of Electromagnetic Radiation I n 1961 it was reported that the human auditory system could respond directly to electromagnetic radiation in the frequency range from a t least 425 to about 3000 MHz (Frey, 1961; see also Frey, 1962; for a more general discussion, see Frey, 1965). The rf sound was described as a buzz, clicking, hiss, or knocking, depending on transmitter parameters such as pulse width and pulse repetition rate. It is known that auroras occasionally produce radio noise in this region (see Ellyett, 1969, and the references cited therein), hence it is pertinent to ask whether direct perception of this noise is possible. Frey’s interpretation of his data as a direct excitation has been criticized. Sommer and von Gierke (1964) have assumed an electrostatic force resulting from the field in order to account for the transduction of electrical energy into mechanical vibrations, and have related these to the sensitivity of the ear for hearing by air and bone conduction. These modes would then be the conventional ones. Their calculations and experiments indicate that Frey’s data for free field rf stimulation is consistent with a pressure slightly above the free field air conduction threshold, though some of his observations indicate bone conduction, which would require a higher pressure. Whatever the mechanism, we may take Frey’s minimal average power density of 400 pW/cm2 and see whether this is consistent with the observations of radio auroral noise. Ellyett (1969) has published a figure, a 9 dB increase in noise a t 43 MHz, corresponding to an auroral sky temperature of the order of lo5 O K , and the resulting energy flux5 for unit bandwidth is of the order of 10-24W/cm2Hz-1. For a 6 kHz bandwidth the resulting energy flux is of the order of W/cm2, and even for an unrealistic bandwidth of 6 MHz, the energy density is only of the order of 10- l7 W/cm2. Since this is very considerably below the minimal values reported by Frey the direct perception of electromagnetic radiation is ruled out. We are indebted to Dr. D. Guidice, Radio Astronomy Branch, Air Force Cambridge Research Laboratories, for this and the following calculations.

204

9.M. SILVERMAN AND T. F. TUAN

4.8. The Electric Field Pressure Effect Next the possibility of air pressure changes produced directly by electric fields in the aurora or by the electric field component of electromagnetic radiation produced by the aurora is considered. As a prerequisite to this discussion the equations for such an effect are first derived. The sudden presence of an electric field in the atmosphere polarizes the air molecules. Work is therefore done on the molecules by the external field. This work produces an increase in pressure which can be detected, if the field is sufficiently strong. This interesting hypothesis was originally suggested by E. K. Franke (private communication) to explain the anomalous sounds associated with lightning. The formula for the increase in pressure as a function of electric field as used by Franke corresponds t o an isothermal model. Because in such a model the temperature is constant, the electric field has to be switched on slowly enough to allow the atmosphere to expand isothermally. The rather sudden change in the electric field associated with the stepped leader and first return stroke of the lightning should be more compatible with a n adiabatic model. There is also the question of whether we can assume that all the Fourier components of the pressure change lie in the audible range. I n this section a derivation of both the isothermal and the adiabatic model is presented. The adiabatic model appears to be more realistic and holds for arbitrary electric field change. Therefore, it is used to examine the anomalous sounds from lightning, meteors, and, in particular, the aurora. To show how an electric field increase can produce an increase in atmospheric pressure, the isothermal model is first given. In this model, no increase in internal energy exists and hence the work done on the gas by the electric field must be equal to work done by the gas. Thus, (4.1)

aw = -pdV,

where d W is the work done on the gas and pd V is the work done by the gas. The work done on a molecule d.z by an increase in the electric field dE is

where p is the induced average dipole moment of the molecule and u is the average molecular polarizability . If p is the molecular number density, the work done by the electric field on the atmosphere per unit volume is given by (4.3)

d W / V = paE dE = -p(dV/V ).

For an isothermal atmosphere, -d V / V = d p / p Hence, (4.4)

aE d E = k To(d p / p ) .

AURORAL AUDIBILITY

205

In equation (4.4),we have assumed a perfect gas p = pkTo, where Tois the constant atmospheric temperature and k is Boltzmann’s constant. By integrating Eq. (4.4)we obtain (4.5)

P=PoexP

a(E2- EO’) 2kTo

In general, even for electric fields approaching the breakdown value, a(E2- Eoa)/2kTo< 1, hence we may write ApI

(4.6)

-=

Po

a(E2- Eo2) 2kT0

where ApI =p -p o is the pressure increase for an isothermal model. I n the adiabatic model it is assumed that the energy supplied by the electric field goes solely into increasing the internal energy and that there is not time for heat exchange with the surrounding atmosphere to take place. Then,

O=dU-dW

(4.7)

where dU is the increase in internal energy. Thus,

dW dU = paE dE = -= pmoC, d T , V V where d U / V is the increase in internal energy per unit volume, mo is the average mass of a molecule, and C , is the specific heat a t constant volume. For adiabatic expansion, (4.9)

-= P

(.!$= ($.

Po

Using the perfect gas law, we obtain (4.10)

Hence (4.11)

Substituting Eq. (4.8)in (4.11), and integrating we obtain, (4.12)

206

S. M. SILVERMAN AND T. F. TUAN

where Eo is the initial electric field. Thus,

(4.13)

<

Again, i t can be shown that a ( E 2- Eo2)/2moCvTo 1. Hence,

(4.14) where Ap, = p - po is the increase in pressure for the adiabatic model. Dividing Eq. (4.14)by (4.6), the ratio for the increase in pressure for the adiabatic model to the isothermal model is obtained,

(4.15) This means that the adiabatic model gives a 40% greater pressure than the isothermal model. It is interesting to note that the result is independent of temperature or the electric field. The average molecular polarizability CL may be estimated from the Clausius-Mosotti equation. Using known values of the dielectric constant and the number density of air, we found that O! 14.4x cm3. Stergis (1966)has made a more rigorous quantum calculation for the oxygen atom and found its polarizability u to be given by u 7.37 x cm3 which is consistent with our result if we assume that one molecule is approximately equivalent to two atoms. Assuming y to be given by y = 1.4 and the initial electric field Eo to be negligible in comparison with E , and substituting known values for mo, C , , T o and the normal atmospheric pressure p, we obtain, N

N

(4.16)

ApA = 2.77 x 10-14(n2- no2)dyn/cm2

where n is the value of the electric field in number of volts per meter and n is the value of the initial field. Let us now consider Eq. (4.16).Fletcher (1923)has given 0.001dyn/cm2 as the minimum pressure required for audibility in the frequency range of 500 Hz to 5000 Hz. The electric field required, as calculated from Eq. (4.16)is approximately 1.9x lo5 V/m. Such a figure would seem quite incompatible with either the aurora or the meteor. For large thunderstorms, electric field measurements near the tip of a stepped leader have been estimated by Malan (1963)to be as high as 60 kV/cm which is certainly extremely audible.

AURORAL AUDIBILITY

207

Even the ionization channel itself which requires a field of 6 kV/cm t o maintain is well above the field required for audibility. However, near the ground level and a t a distance of a kilometer or so away from the lightning flash, the electric field is much less. To be sure, if one uses the dipole model, it is always possible to get very high fields. For instance, Pierce and Wormell (1953) have argued that a median value of 110 coulomb kilometer may be taken as the dipole moment for the production of lightning. With the dipole model, one can obtain values of 1 V/m and 1000 V/m a t 100 km and 10 km, respectively, which is in good agreement with experimental results. However, in extrapolating to 1 km distance, the dipole model would give a field of lo6 V/m which is certainly audible, except by then the distance becomes comparable to the dimensions of the thundercloud and the dipole model no longer works. It is known, in fact, that the field reverses in sign as one approaches the lightning source (Wilson, 1921). The quality of the sound has been described as either a click, a crackling, or a hiss (Griffin, 1922)-all of which have frequency components well above 5000 Hz. Indeed, a hiss is something like 14,000 to 16,000 Hz. According to the Fletcher audiograms (Fletcher, 1923), this would require a t least a pressure change of 0.1 dyn/cm2to produce anything audible.. The electric field change required is then something like 1.9 x lo6 V/m which is far too large when compared with most observed values a t distances of a few kilometers. In fact, the value begins to approach the fields a t the tip of the leader streamer and is more than a factor of 3 larger than the average fields a t the base of the cloud (Malan, 1963). It is obviously out of the question to apply this theory to the auroral or the meteor sounds which have been described in the first part of this paper as almost invariably a rustling, hissing, or swishing high frequency sound.

4.9. Direct Effects of Auroral Electric Fields A number of investigations of electric fields in auroras a t heights of 80 km and above have been carried out in recent years (see, for example, Wescott et al., 1969; Potter, 1970). These studies estimate fields of the order of 0.1 V/m, which is far too small to produce any significant pressure change, either a t auroral heights where the ambient pressure is small, or a t ground level. Small electric fields, however, may be expected inside auroral forms where the conductivity is high. This does not preclude high electric fields outside the aurora or near ground level, and several recent studies do in fact indicate a strong correlation between the electric field a t ground level and an overhead aurora. In the extreme these fields have been reported to reach as high as 10,000 V/m. Using this value, the direct pressure effect, as can be seen from Section 4.8, would be too small to produce audibility. This hypothesis must therefore be ruled out.

208

9. M. SILVERMAN AND T. F. TUAN

4.10. Emission of Radio Waves in the Audio Region and Conversion to Pressure at the Ear As has been discussed, an electric field can polarize air molecules, and this in turn can produce a change in pressure. Furthermore, the pressure change depends on the square of the electric field irrespective of the model used. Hence the pressure change is independent of the direction of the electric field. I n principle, one would expect that the electric field component of electromagnetic radiation can also produce pressure changes, and if the field contains components in the audio region, then sound would be heard. For the case of lightning the observer a t a few kilometers distance is so close to the lightning source in, say, a dipole model, that the dipole electric field predominates. During auroral displays a low frequency hiss within the audio range has sometimes been detected by a number of observers (Jerrgensen and Ungstrup, 1962; Martin et al., 1960; Gherzi, 1960). This hiss ranges from 800 Hz to 10 kHz, although the greater part of its energy is concentrated a t greater than 4 kHz, as compared to normal hiss bands which lie a t frequencies below 4 kHz. The hiss was found by Jargensen and Ungstrup to have variations in intensity which corresponded t o the auroral activity. The description and character of the auroral hiss seems rather similar to the auroral sounds which have been described, that is, rustling and hissing. Furthermore the auroral sounds have also been observed to coincide with the intensity variations and motion of aurora. It is tempting, therefore, to ascribe both phenomena t o the same common origin. Measurements of the average electric field intensity (Martin et al., 1960), however, have shown that it lies somewhere between 1 and 3 mV/m. Using an adiabatic model the pressure increase is given by -

APo

N

'

(.E2)Po y - 1 2MoC,To

-

2.5 x

dyn/cm2

This pressure change is too small to be audible, even though the frequency is within the audible range. This possibility must therefore be ruled out. 5 . THE CASEFOR BRUSHDISCHARGE AND AURORALLY INDUCED

ELECTRIC FIELDS One hypothesis remains to be discussed, that of brush discharges. This is presented as a separate section because, in our opinion, this hypothesis allows for the most satisfactory explanation of the observations, and it will therefore be discussed in more detail than the other hypotheses which have been proposed. The logical sequence which is followed here to show that this is indeed a plausible mechanism is (1) a comparison with the observations to see if these are consistent with the known facts about brush discharges; (2) the experi-

AURORAL AUDIBILITY

209

mental evidence for aurorally associated electric fields a t ground level; and (3) the possible origins of such a field. Finally, those additional experiments which are needed to conclusively confirm or refute this hypothesis will be indicated.

5.1. Comparison of the Behavior of Auroral Sounds and Brush Discharges The hypothesis that brush discharges may be the cause of anomalous sounds is not new (Chant, 1923a, 1931; Stclrmer, 1927; Beals, 1933a,b,c; Eve, 1936). However, electric field measurements during auroral displays available a t that time did not always show any marked deviation from normal values. Hence, it would be theoretically difficult to connect the aurora with brush discharge. Recent measurements of the electric field, to be discussed next, would seem to indicate that although the electric fields directly underneath a corona can rise to high values, they are still not sufficient to produce sounds directly by polarization of molecules and a resultant pressure change. An electric field of some 10,000 V/m, however, is well above the value required for brush discharge which usually takes place at around 1500 V/m (W. C. A. Hutchinson, private communication). Brush discharge is the occurrence of discharge from point electrodes where very sharp potential gradients can exist. The point electrodes may be trees, bushes, or sharp protruding objects on the ground. The normal average electric field on the earth’s surface over open ground is about 100 V/m which is insufficient to cause discharge. During an auroral corona, however, this field can rise to very high values which are sufficient to cause discharge. Some of the properties of brush discharges have been succinctly summarized by Shipley and Barnes (1940). The local potential gradient can be increased by a marked irregularity on the surface, such 8,s an isolated tree or a mountain, by lowered air density as the altitude is increased, by dust, by fog, and by induction from clouds with active internal circulation. The discharge can produce a rustling or hissing sound, light, generally purple, near the ground, and can make light slender objects such as hair stand on end. Very little heat is produced unless the point is very fine or the current comparatively large. A tingling of the skin and a feeling of apprehension can result. The discharge from pointed or sharp objects (electrodes) occurs in the following way under quiet conditions when there is no wind or other disturbing weather conditions which can remove space charges from the vicinity of the electrodes. The electric field immediately surrounding the pointed electrode is very strong and a slight discharge current will only produce sufficient space charge to weaken the field in its immediate vicinity. As the externally applied electric field increases, the current increases and a stable discharge occur8 until there is sufficient space charge t o cause an increase in the potential

210

9.M. SILVERMAN AND T. F. TUAN

gradient a small distance away from the electrode. Further increase in the externally applied field eventually causes a breakdown. Thus, if a steady wind exists which is sufficient to remove the space charge so that there is not enough left to produce the sharp increase in the potential gradient, then a steady but continuous small discharge would take place without breakdown. We would hence expect to hear crackling sounds only under very still weather conditions. This is in complete agreement with all the observations in which claims for auroral sounds have only been made in very clear and undisturbed weather. In one reported incident the occurrence of auroral sounds follows the visual motion by about a second or less. This can hardly be explained by anything except brush discharge which takes place a t some distance away, say 150 to 300 meters. A direct conversion of field change to pressure change a t the ear must always appear to be simultaneous, while direct transmission of sound from auroral heights must, of course, take much longer. The only report (147) on the sequence of the sounds was observed to be that of hissing followed by crackling. This is again easily explained in terms of brush discharge. We have a steady-state discharge from the electrodes at first which can produce a hissing sound. Then, as sufficient discharge is accumulated to produce the potential gradient sharp enough for the breakdown, we hear the crackling. In general, the quality of the sounds described does strongly suggest some kind of electrostatic discharge, since such discharges invariably produce hissing, swishing, and crackling sounds which have a frequency range very much above anything that can be transmitted from auroral heights directly. The presence of an odor similar to sulfur or ozone further reinforces the brush discharge theory. The ozone is formed from dissociation of O2molecules into oxygen atoms by the discharge and the atoms then recombine with O2to form 0,molecules. Ozone, of course, may be formed so long as we have a discharge. The accompaniment of sound is hence not required for its presence. In the one case [45] where the sequence was observed, the sound comes before the odor which is what we would expect since the sound arrives from the source through wave propagation while the odor arrives by either diffusion or convection, as already mentioned earlier, and hence E a s t take a longer time. Thus it appears that only brush discharge can fit all the facts unambiguously. Any other explanation would immediately run into one difficulty or another. Let us next consider whether there are in fact aurorally associated electric fields sufficiently high to produce brush discharges.

5.2. Aurorally Associated Electric Fields Brush discharge, or point discharge, can occur a t the tips of grounded pointed objects when the vertical electric potential gradient rises to some hundreds of volts per meter. These discharges are important in the natural environment as a major contributor in maintaining the electrical equilibrium.

AURORAL AUDIBILITY

211

Nevertheless only a few measurements for natural objects exist (Hutchinson and Stromberg, 1969). In one measurement (Hutchinson and Stromberg, 1969)a spruce tree discharged when the potential gradient reached 6000 V/m, compared to an onset value of 2500 V/m for a metal point. Assuming that gradients of the order of magnitude of lo00 V/m or greater are needed to produce auroral sounds it remains to be seen whether there is observational evidence for the existence of such gradients associated with auroras. A number of measurements of auroral effects on the earth’s electric field have been made (see the summary in Pakiam and Johnson, 1967), with varying results. Several of these (Freier, 1961; van der Schueren and Koenigsfeld, 1963; Kikuchi, 1970; Olson, 1971,and private communication) seem to indicate a correlation between the electric field at the ground level and the aurora. However, most of the results give only averaged electric fields. Freier obtained gradients greater than 1000 V/m during auroral activity on Oct. 7,Nov. 13, and Dec. 1, 1960. On all three occasions the aurora was overhead and the weather was perfectly clear with little or no wind. On October 7, 1970 the record of the electric field as a function of time showed large spikes in the field with 1500 V/m peaks and rise times of as short as 30 seconds. These took place between 0700 and 0800 P.M. It is our understanding that these sharp variations in the field strength took place during a corona,although no attempt was made to establish any correlation between the field changes and the streamer motions in the corona. It may be noted that the event of Nov. 13, 1960 was a major solar-terrestrial event with one of the largest cosmic ray increases recorded. Also, the conditions under which these high gradients occurred are those of the typical auroral sound event inferred from the observational material of Section 3.Pakiam and Johnson, on the other hand, reported no relationship between auroras and potential gradient in measurements made a t Saskatoon between October 1, 1964 and September 30, 1965. During a ten day period in March 1965 they operated their equipment a t Churchill, in the auroral zone, and found no field fluctuations during one aurora and large field fluctuations during another. They also noted that peculiar field effects were often found during periods of calm or light winds, and implied that the results of Freier could be interpreted among these peculiar meteorological effects. Their measurements were made during sunspot minimum, when our observational auroral sound data would indicate the least likelihood of success on the brush discharge model. The electric field measurements of Olson were taken with field mills a t Duluth and Minneapolis. The instruments were suspended some twenty meters above the ground level. The measurements a t Duluth showed very sharp peaks with peak values close to 10,000V/m and rise times of about 2 to 4 minutes. He has published figures showing auroral electric field effects from measurements in Minnesota on August 19, 1963, September 3, 1966, and Ootober26,1966,and from Churchill on September 9,1967.An example of his

212

9.M. SILVERMAN AND T. F. TUAN

data is reproduced in Fig. 6. He reports that the effects are observed only when an intensely active coronal form, in subauroral latitudes, and the westward surge form, in the auroral zone, are in the vicinity of the magnetic zenith. At subauroral latitudes he notes that K, is generally greater than 5 when effects are observed, and that the weather is generally clear and calm. No effect is noted in clear weather without aurora. Finally, he finds that under favorable conditions an event is observed in about one out of four cases. An interesting feature of his results is the reversal in field direction occurring over times of minutes or tens of minutes. These are reminiscent of some observations of auroral effects on telegraph lines. Muhleisen (1969) has noted that at the IUGG Conference in 1967, Olson’s results and objections thereto were discussed, and no errors in the measurement were found. ON

8:: 00

82

88

I

4.2

0

-4.2

-9.4 403vm-1

4.4

2.7

0

-2.7

Fro. 6. Atmospheric electric field near ground during an overhead aurora (from Olson, 1971).

AURORAL AUDIBILITY

213

The conditions noted as favorable for high electric fields here are comparable to those which, as indicated, favor the occurrence of auroral sound events. Taken all together, and making allowance for the correlations shown in Section 3, these results are consistent with the hypothesis that auroral sound events are produced by brush discharges produced by transient high potential gradients associated with intense overhead aurora. It is unfortunate that no concurrent reports of auroral sounds and high electric fields have been made, but the possibility of masking noises in the few electric field measurements could account for this. Nevertheless it would appear that future observations of high aurorally associated fields should take account of the possibility of sounds, so that this could be definitely confirmed or refuted. Finally, auroral effects on telegraph lines and electric power lines have also been noted. These may very well be related to the electric field effects pointed out above, but cannot be brought into closer conjunction since, because of the long lines involved, one cannot unambiguously separate out local effects from those occurring over a large spatial extent.6 ‘During the nineteenth century a number of papers appeared on the effects of auroras on telegraphic service (see for example, Prescott et al., 1860; Loomis, 1866; Sargent, 1872; Donati, 1873). These effects were particularly marked during the great auroras of 1859 and 1872. Two aspects are of interest to us. First, the induced currents along the lines were so great that on several lines the operators disconnected the batteries and used the aurorally induced currents alone for the transmission of messages. Second, the currents reversed their polarity in times of the order of minutes or tens of minutes. The phenomena are thus in some important respects similar to those we have discussed above for electric fields, and we may hypothesize that they represent a different manifestation of the same phenomenon. Unfortunately, however, the fact that the fields developed over long lines leaves open the possibility that the cause was the small horizontal field observed in auroras extended over long path lengths, rather than the more highly localized field required by the observations. For most cams, such as those used by Tromholt (1885a) in his Statistical study of perturbations on the telegraph, it is probably this small horizontal field which is responsible. For the most extreme cases where polarity reversals also occur it is possible that the more localized fields are involved. Separation of the effects on a long line is not obvious, however, and these phenomena are therefore merely noted here. In the same class is the observation by Joseph Henry (1881) of the deflection of a galvanometer needle, one end of which was connected with the water pipes of the city and the other end to the gas pipes, during the course of an aurora. Henry noted the effect was similar to that observed when a flash of lightning took place within the visible horizon of Washington. Finally, we note parenthetically that one of the dates of the auroral effect on telegraph lines in New England noted by Prescott et d.(1860), Feb. 19, 1852, coincides with a n auroral sound event in Okso (36), and that he notes a strong effect in September, 1851 which may coincide with the September 7, 1851 sound event in Okso (35). I n a recent study Albertson et al. (1970) have shown that significant auroral effects occur in the northern United States on electric power lines. Their published curves for one aurora show field reversals similar to those described for telegraphic lines and in Olson’s measurements.

214

9. M . SILVERMAN AND T. F. TUAN

5.3. Mechanism for the Production of Aurorally Associated Electric Fields There still remains the problem of defining a mechanism by which an aurora can produce an observable electric field effect a t ground level. The terrestrial electric system, on a global basis, consists of the earth with a net negative charge and the lower ionosphere with a corresponding positive charge (for a general discussion of atmospheric electricity and its characteristics see, for example, Chalmers, 1967). The charge is presumed to be maintained primarily by thunderstorms and is lost by conduction through the atmosphere between ground and lower ionosphere. Conduction through the atmosphere is determined largely by the columnar resistance below about 25 km. Cole and Pierce (1965), for example, calculate that 90% of the columnar resistance lies below 2.4 km, 99% below 10.8 km, and 99.9% below about 25 km. Hence, any auroral effect by way of changes in conductivity would have to be a t low altitudes. Direct effects of particle precipitation would thus be ruled out, inasmuch as there are insufficient fluxes of sufficiently high energy particles which can penetrate to sufficiently low levels [see, for example, the discussions by Dolezalek (1964) or Olson (1971)l. Changes in current flow resulting from such global changes as might possibly result, for example, from any magnetic field effects, can also be ruled out, as we are dealing with an essentially localized phenomenon. This leaves then the possibility of the auroral production of a space charge at stratospheric heights or below (Freier, 1961; Cole and Pierce, 1965; Olson, 1971). A space charge can be produced by a differential mobility of electrons and positive ions. The primary incident particles in an aurora, as noted above, cannot penetrate to sufficiently low levels to produce a ground level effect. X-Rays produced by these particles, however, can penetrate considerably lower, and these, in turn, can produce ionization. Then, as Cole and Pierce put it ". . . in the ionization produced the electrons have a net forward component of velocity relative to the positive ions, a dipole is thus created, and an associated electric field develops a t the ground ". X-Rays associated with aurora have been studied with balloon-borne instrumentation over the past few years (see summaries by Kremser, 1967, 1969), and i t is known that bursts of energetic X-rays do in fact occur. The field is in a state of rapid development at this time and a detailed comparison of either auroral sound event statistics or electric field measurements with the presently available corpus of auroral X-ray data seems premature, especially as we are looking for properties of occasional large events. The hypothesis of aurorally produced electric fields nevertheless develops a certain plausibility, if not conviction, as the origin of auroral sound events. 5.4. Energetics of Sound Events Deriving from Aurorally Associated Electric Fields As noted in our previous discussions seemingly plausible mechanisms can easily fall by the wayside when the energetics are considered. I n this section

AURORAL AUDIBILITY

215

the energetics of sound events will be considered, on the assumption that they derive from aurorally associated electric fields. We shall first make some approximate estimates of the average energy per second required to build up the auroral electric field from charged particle precipitation and compare the model calculations with the input particle energy flux. Using the model of Freier (1961) as a first approximation, it is simple to show that the energy per square centimeter column required to build up an electric field Eo a t the ground level is given by

where H is the scale height and S is the approximate height of the column. Everything here is in c.g.s. units. Our calculations are very approximate, hence one may drop the exponential term and rewrite Q as (5.2)

Q

N

ho2Ji2H,

where A. is the relaxation time a t the ground level and J , is the incident current density. As an example, the energy per square centimeter column for a ground level field of 10,000 V/m and a scale height of 70 km is 16,000 erg/cma. Using Eq. (5.2)and a relaxation time of 15 minutes, the incident current is about 30 x statamps/cma giving a flux of about 6 x lo4 per cma s. This value for the number flux is very much below the expected particle flux during an intense aurora. The approximate average rate of building up this field is about (5.3)

&/Ao

N

18 erg/cma s.

I n comparing this with the incident energy flux of an intense corona which can be 100 ergs/cm2 s, the fraction of the energy that goes to building this field is of the order of 20%. If the energy is available in the field it must then be considered whether it is reasonable to convert sufficient of this energy to produce audible sounds. Let us assume that the electrodes form a square lattice with the lattice points spaced 10 meters apart. The power required to produce an audible sound of .001 dyn/cma from an electrode 10 meters away from the observer is about 3.6 x lo8 erg/s. This power is produced from a local breakdown a t the electrode. The energy is supplied by the collapse of the electric field in the vicinity of the electrode. At the ground level the breakdown field is about 100 statvolts/cm. This corresponds to an energy density of about 400 erg/cm3. Assuming that the breakdown takes place in a small volume of about 1 cm3 and the breakdown time is about 1 ps, the power

216

S. M. SILVERMAN AND T. F; TUAN

liberated is about 4 x lo8 erg/s, which is more than adequate to produce an audible sound 10 meters away. A repetitive succession of crackling sounds would, a t some time interval, appear to be continuous to the observer. The effect is an auditory analog of the visual flicker experiment in which a t some critical pulse frequency fusion occurs and the light appears to be continuous. I n the auditory case the experiments, although somewhat in doubt because of leakage in the early work, indicate that for pure tones the critical rate for fusion will vary with the frequency, and, for tones over 1000 Hz, should be in the range of 150 t o 250 per second (Wever, 1949, pp. 408ff).Some roughness would remain even after this. In our case, the descriptions are those of a nonsteady sound, so this latter is no problem. A more important distinction is that the discharge should be of the nature of noise, and thus contain a wide spectrum of frequencies, and the pulse shape would not be square. It is likely that these factors, as well as a random timing between pulses, would tend to reduce the mean repetition rate necessary to produce a sensation of continuousness. For this discussion, a succession of sounds a t 11200 second intervals will be assumed, and the rate of energy loss through discharge is then about 80,000 ergls. Aseach electrode is allocated an area of 10 m2, the average energy loss is 8x erg/cm2 s. This is insignificant when compared to the energy required to build the electric field which is, as just computed, 18 erg/cm2 s. Thus the loss through brush discharge would be less than a half of a percent of the energy that goes into building the field.

6. CONCLUSIONS We have concluded, then, that the observational evidence supports the reality of auroral sounds and that the most likely source of these seems to be brush discharges, and that these are generated by aurorally associated electric fields. The latter are most likely produced by a space charge possibly resulting from ionization by hard X-rays which have themselves been produced at higher altitudes by the impact of energetic particles. A definitive proof of these assertions would involve the recording of auroral sounds simultaneously with the measurement of the electric field, photography of the aurora, and measurements of the flux and energy spectrum of aurorally associated X-rays, with notations of the meteorological conditions a t the time. As we are dealing with an occasional phenomenon which is dependent on latitude and sunspot cycle, among other parameters, it is clear that an experiment designed solely for the determination of auroral sounds would be unrealistic in terms of the time and money involved. Measurements of the parameters involved are, however, being made because of the inherent

AURORAL AUDIBILITY

217

interest in these parameters themselves. It would seem that the most sensible course would then be to consider auroral sounds as an interesting auxiliary problem for an observatory already carrying out most of these measurements. An auroral zone station would be the most favorable location a t the present time. Several stations are already in existence whicb measure the basic geophysical parameters, and the addition of equipment for the recording of sounds should be easily feasible. More sophisticated equipment similar to that of Procunier (1971) might also be useful. A similar approach a t the next sunspot cycle maximum would be useful a t a station a t geomagnetic latitude near 55", and this would also help to clarify differences in behavior between auroral and sub-auroral zones. Finally, a good theoretical treatment of the entire hypothesized process is needed. This would begin with the incident energetic particles and trace the various processes involved in the dissipation of the energy of these particles and their relative importance. The chemistry of the atmosphere for altitudes a t and below the D-region is complicated and the results are strongly dependent on the nature of the species present and the reaction rates, all of which are imperfectly known. The problem is not a simple one, but must still be done before a solid theoretical base can be presented. Additional work is also needed on some of the basic parameters of the atmosphere's electrical behavior, such as the behavior of point discharges for natural objects in different sorts of terrain and the variation of this behavior for different meteorological conditions. The electrical condition of the upper atmosphere under highly disturbed conditions is also poorly known. R. Sagalyn (private communication) has pointed out that the lifetime of a space charge produced by X-rays at altitudes of the order of 30 km or below is likely to be small because of the high recombination rates a t these altitudes. She points out, however, that the earth-ionosphere potential is irregular and that the high conductivity a t D-region heights produced by particle input can lead to a localized lowering of the ionospheric potential to much lower altitudes. I n any event, it is clear that measurements of the electrical state of the upper atmosphere during disturbed conditions are needed. Much of the work outlined above is already going on for different purposes. We may anticipate that over the coming years a more solid foundation for understanding the phenomenon of auroral sounds will be available.

LISTOF SYMBOLS C

6 c

Cy E

Magnetic character figure Average magnetic character figure Velocity of sound Specific heat at constant volume Electric field

S. M. SILVERMAN AND T. F. TUAN

Undisturbed original electric field Average scale height of atmosphere Incident current density Magnetic index Boltzmann’s constant Average mass of a n air molecule Value of electric field in number of volts per meter Value of undisturbed electric field in number of volts per meter Pressure of atmosphere Pressure of undisturbed atmosphere Average dipole moment (electric) of molecule Electric field energy per cma column Maximum sunspot number Minimum sunspot number Approximate height of atmosphere Temperature of atmosphere Temperature of undisturbed atmosphere Internal energy of atmosphere Volume of atmosphere Volume of undisturbed atmosphere Work Average molecular polarizability Ratio of specific heat at constant pressure to constant volume Increase in pressure under isothermal condition Increase in pressure under adiabatic condition Dielectric constant Mean free path of molecules at height h Number density of atmosphere Relaxation time at ground level

APPENDIX:AURORAL SOUNDEVENTS Direct quotations of the literature which were used in the preparation of Table I are presented here using the numbering system of the Table. Particulars of each event, a s well as the source reference, are also given in the Table. Where several references are available the original or the most complete has been quoted. 1. The earliest, and in some respects rather unreliable, account is that of Absalon Pedersson, dated December, 1663. This recital is so amusing that I will repeat it here: One evening a little before Christmas I witnessed the following occurrence which began about 7:30 and continued until fully 9 p.m. Christern Ulff and a goldsmith and both their wives and servants saw it also. At first the moon shone clear in the east. Then a dark cloud which reached high up into the sky came over it. Presently a bright cloud, which shone like a white flame, formed and both remained stationary for some time. After they dieappeared an unusually black cloud with scattered cloud wisps all about it approached from the south and overshadowed the moon so that it lost it6 light. After it had passed the sky grew red in the west and fire and flame darted back and forth so that a great noise was given off. I asked Christern Ulff what caused the sound, a s I thought perhaps it came from the Alreichstadselff.

AURORAL AUDIBILITY

2.

3. 4.

5.

6.

7. 8.

219

[A river near Bergen] He replied, “Don’t you see it is in the sky? It was the clouds that ran rapidly back and forth.” Afterwards other clouds, some black and some white, overcast the moon and then disappeared. Another reason is, that I am assured by an ingenious sensible gentleman of my acquaintance, that as he was viewing this appearance, on the top of his house at Little Chelsea, he plainly perceived a Sulphureous smell in the air; and that another person did the same, on the top of another house near him. Another thing which concurs with what hath been said, is, that I am assured from several persons, that an hissing, and in some places a crackling noise was heard in the time of the streaming, like to what is reported to be often heard in earthquakes. Nach Messier wiire es sogar 1762 in Paris horbar gewesen; Charles bestiitigt die Angabe. Ramm (koniglicher Forstinspector in Norwegen) schrieb 1825 an Hansteen, daas er in den Jahren 1766, 1767 oder vielleicht 1768 das Geriiusch gehort habe (“ Schweiger Journal,” Neue Reihe, XV). Da Ramm jedoch das Geriiusch gleichzeitig mit dem Aufschiessen der Strahlen bemerkt haben will, so ist die Angabe, da Licht und Schall betriichtlich verschiedene Geschwindigkeiten haben, etwas verdiichtig. Samuel Hearne’s “Journey from Prince of Wales’s Fort in Hudson’s Bay to the Northern Ocean in the Years 1769, 1770 and 1772,” p. 235 (Champlain Society’s Edition, Toronto, 1911) says: “ I do not remember to have met with any travellers into high northern latitudes, who remarked their having heard the Northern Lights make any noise in the air as they vary their colours or position; which may probably be owing to the want of perfect silence at the time they made their observations on those meteors. I can positively affirm, that in still nights I have frequently heard them make a rustling and crackling noise, like the waving of a large flag in a fresh gale of wind. This is not peculiar to the place of which I am now writing, (Great Slave Lake), as I have heard the same noise very plain at Churchill River; and in all probability it is only for want of attention that it has not been heard in every part of the Northern Hemisphere where they have been known to shine with any considerable degree of lustre. It is, however, very probable that these lights are sometimes much nearer the earth than they are at others, according to the state of the atmosphere, and this may have a great effect on the sound; but the truth or falsehood of this conjecture I leave to the determinations of those who are better skilled in natural philosophy than I can pretend to be.” On January 18, 1778, Dean Wilse of Spydeberg thought he heard the auroral sound. Since my last, of the 19th instant, I luckily fell in company with a friend, Ichabod Tucker, Esq. a clerk of our supreme court, and a respectable character, who informed me, that while he lived in Connecticut, near New Haven, in the month November or December, 1781, or in January, 1782, he saw an exhibition of an aurora borealis, exactly similar to that second, which I gave you a description of, only that he took no notice of the absence of the blue colored rays. This he saw while walking about two miles, in a N.E. direction, so that he had half an hour’s time to view it, which he did with great trepidation, being then about fourteen years old, and having never seen an Aurora. But by his account, he heard the noise attending the shootings and vibrations of the columns, more loud, and more frequent than I ever had, or had heard of, for he heard not only this rushing noise,

220

9.

10.

11.

12.

13.

14.

16.

S. M. SILVERMAN AND T. F. TUAN

which I as well as three or four of my other acquaintances have repeatedly done, but loud claps or snaps, of which he was so perfectly assured, that he told me, he should not hesitate t o confirm by his oath, if necessary. [He tells me in further conversation on the subject, that the claps, or snaps which he heard, were most like the flapping of a vessels mils, when thrown up in the wind.] Did you ever, in observing the Aurora Borealis, perceive a sound? I own I once looked on the idea as frivolous and chimerical, having heard it a t first from persona whose credulity, I supposed, exceeded their judgment; but, upon hearing it repeatedly, and from some others whom I thought judicious and curious, I began to entertain an opinion in favour of it. I was strengthened in this opinion about two years ago, by listening with attention to the flashing of a luminous arch which appeared in a calm frosty night, when I thought I heard a faint rustling noise like the brushing of silk. Last Saturday evening I had full auricular demonstration of the reality of this phenomenon. About ten o’clock the hemisphere was all in a glow; the vapours ascended from all points, and met in a central one in the zenith: All the difference between the south and north part of the heavens was, that the vapour did not begin to ascend so near the horizon in the south as in the north. There had been a small shower with a few thunder claps, and a bright rainbow in the afternoon; and there was a gentle western breeze in the evening which came in flaws, with intervals of two or three minutes; in these intervals I could plainly perceive the rustling noise, which was easily distinguishable from the sound of the wind, and could not be heard till the flaw had subsided. The slashing of the vapour was extremely quick; whether accelerated by the wind I cannot say; but from that quarter where the greatest quantity of the vapour seemed to be in motion, the sound was plainest; and this, during my observation, was the eastern. The scene lasted about half an hour, though the whole night was as light as when the moon is in the quarters. [May 24, 1788: very grand] From 10 to 11 P.M. uncommonly brilliant, active streamers over most of the hemisphere: they were said t o be heard-Not much inferior the next night. Another early reference is by David Thompson. He spent the winter of 1796-1797 at Reindeer Lake, Saskatchewan, and writes the following account of a brilliant aurora seen by himself and his men: “ I n the rapid motions of the Aurora we were all persuaded that we heard them, reason told me that I did not but it was cool reason against sense. My men were positive that they did hear the motions of the Aurora, this was the eye deceiving the ear; I had my men blindfolded by turns and then enquired of them if they still heard the motiona of the Aurora. They soon became sensible that they did not and yet so powerful was the illusion of the eye on the ear that they still believed that they heard the Aurora.” Brewster, Grant and Burness wollen auf Orkney am 6. December 1801 Gerausch gehort haben, welches an das Ueberspringen des elektrischen Funkens vom Gbscylinder zum Conductor erinnere. They were mostly of a dunnish yellow, yet often assuming mixtures of red and green. When they are particularly quick and vivid, a crackling noise is heard, resembling that which accompanies the escape of the sparks from an electric machine. Bei Skien sol1 1818 bei jedem Strahlenachiessen das Gerilusch wie bei dem Aufrollen von Seidenzeugen gehort worden sein.

AURORAL AUDIBILITY

221

16. Oder auch Biot (“ Lehrbuch der Experimentalphysik,” Leipzig 1825) im November 1818 in +Sl&” zwischen Island und Shetland das starke Rauschen oder Brausen des Nordlichtes wollen gehort haben. 17a. Wrangel, jedenfalls einer der zuverliissigsten Berichterstatter, da er aus mehrjahriger Erfahrung spricht und ohne Vorurtheil die Erscheinung beobachtete, sagt (“Reise liings der Nordkiiste von Sibirien in den Jahren 1820-1824”): “Krachen oder sonst bedeutendes Geriiusch horten wir bei den starksten Nordlichtern nicht, wohl aber in diesem Falle ein leichtes Zischen, wie wenn Wind in eine Flamme blast.” 17b. Even during the most brilliant Auroras, we could never perceive any considerable noise, but in such cases we did hear a slight hissing sound, as when the wind blows on a flame. 18. “On the 11th of March, at loh p.m., a body of Aurora rose NNW, and after a -8 of it had passed to E by S the remainder broke away, in portions consisting each of several beams, which crossed about 40” of the sky with great rapidity. We repeatedly heard a hissing noise, like that of a musket-bullet passing through the air, and which seemed to proceed from the Aurora; but Mr. Wentzel assures us, that this noise was occasioned by severe cold, succeeding mild weather, and acting upon the surface of the snow, previously melted in the sun’s rays. The temperature of the air was then - 35’, and on the two preceding days, it had been above zero. The next morning it was -42”, and we frequently heard a similar noise. Mr. Herne’s description of the noise of the Aurora agrees exactly with Mr. Wentzel’s, and with that of every other person who has heard it. It would be an absurd degree of scepticism to doubt the fact any longer; for our observations have rather increased than diminished the probability of it.” 19. A t Rochester and at Utica its appearance was the same; and at the latter place reports were distinctly heard, producing sharp, snapping noise, like the discharge of an electric battery. 20. An intelligent gentleman who was at the time in St. Lawrence county, informs me, that the reports were heard during the existence of the arch, but that afterwards, while the coruscations of the aurora were so splendid, the report was very distinct and loud, and of the character of those heard a t Utica. 21a. I n New Haven, Conn. it presented the same appearance, and was accompanied by reports, which increased in frequency, and distinctness, with the darting8 of the Aurora. These reports were noticed by some gentlemen of the Faculty of Yale College, who were making observations upon the phenomena at the time. 21b. Peculiar sounds referred t o the aurora were also mid t o have been heard at this place (Yale College). One of the students, indeed, was under the impression that he distinctly heard such sounds, but I was unable, either on this or on any subsequent occasion, t o detect any sound which in my judgment could be fairly attributed t o the aurora borealis. 22. When the Aurora displays itself in all its splendor, its light is brighter than that of the full moon. It has been asserted, that this phenomenon is sometimes accompanied by a low, hissing noise. I myself, in fact, have often heard the sound, but am satisfied it has nothing to do with the Aurora, but proceeds partly from the ice, partly from the wind sweeping over the snow and ling-clad hills. 23. It has always been an interesting question with those who attempt to ascribe this beautiful phenomenon to electrical c a w s , whether the Aurora be attended with any sound or noise; and although many accurate observers have paid particular

222

5. M . SILVERMAN AND T. F. TUAN

attention to this subject in various parts of the northern hemisphere, yet the point is far from being settled. On no occasion, during two winters, was any sound heard to accompany the motions of the Aurora by either Captain Back or myself. Once or twice I thought a sound was audible, but afterwards ascertained it to be the hissing noise produced by the sudden condensation of my breath into icy particles; and Captain Back several times positively declared he heard a whizzing noise during the rapidity of the motion, until he convinced himself it was the faint murmuring only of Anderson’s Fall that had deceived him. 24. 1836, Nov., 18 Streamers accompanied with a crackling noise, very distinctly audible. Most of the streamers on this occasion were in the southern part of the sky. 25. Fort Confidence-Lat. 66”53’36”N.Long. 118”48’45”W. April, 1838. Simpson. March 5, 1839. This season, as I have already remarked, was less severe than its predecessor; and, as if it were a consequence of the difference, the Aurora was more brilliant, displaying on several occasions the prismatic hues; but the same arched form from northwest to southeast predominated. Every clear night, when not eclipsed by the moon it was to be seen, but was brightest and most active in the mornings some time before daylight. At a quarter to four a.m., on the 5th of March, Ritch witnessed a most brilliant exhibition. It formed a quadrant issuing from W.N.W. and extending to the zenith. There it doubled on itself, and terminated in a semi-elliptical figure, apparently very near the earth, in rapid motion, and tinged with red, purple, and green. The half ellipse seemed to descend and ascend, accompanied by an audible sound, resembling the rustling of silk. This lasted for about ten minutes, when the whole phenomenon suddenly rose upwards and its splendor was gone. Ritch is an intelligent and credible person, and, on questioning him closely, he assured me that he had perfectly distinguished the sound of the Aurora from that produced by the congelation of his breath-for the temperature at the time was 44” below zero. I can, therefore, no longer entertain any doubt of a fact uniformly asserted by the natives, and insisted on by Hearne, by my friend Mr. Dease, and by many of the oldest residents in the fur countries; though I have not had the good fortune t o hear it myself. 26. 1839, Sept. 3. Sound of streamers like that produced by striking the air sharply with a switch. This aurora was seen in North America. 27. 1839, Sept. 4. Aurora more hazy than that of last night, and in the form of sheets flashing wildly up to the coronal point, accompanied with a ‘‘ whiffling” sound. N.B. Except in the above instances, I have never heard sounds given forth by the aurora; in these cases the streamers probably descended lower than usual. [This note also refers to 24 above.] 28. J. E. Schonfeldt (“Bulletin de la Societ6 imphriale des naturalistes de Moscou,” Bd. 49, 1875) versichert, am 20 September 1839 in Livland a n einem windstillen Abende, nach Sonnenuntergang, auf baumloser Ebene und zwischen Stoppelfeldern, fern von jedem Gewasser, das Nordlicht-Gerausch deutlich gehort zu haben, indem es periodisch die uber das Zenith hinschiessenden Strahlen zu begleiten schien. Es klang wie zeitweise schwach aneinandergewetzte Degenklingen 28e. However, in the large catalogue which contains reports of thousands of auroras, there are few instances of this sound recorded. On March 22, 1840, Ihle observed I

AURORAL AUDIBILITY

223

an aurora in Kaafjord which, according to his report, was accompanied by a peculiar noise. Ihle was a naturalist and has himself given a n account of hie observations which has been published. He is reported to have heard the sound on two other occasions during the same winter. 29b. Ihle will bei Windstille das raschelnde Nordlicht-Gerausch am Kaafjord bei Alten, 1840, dreimal vernommen haben. 30. It was a n unfortunate consequence of this assumption that the observers gave no special attention to the occasions (five are mentioned) when they believed that aurora occurred beneath clouds. The only entry of this type which exceeds a couple of lines in length is that on March 16, 1840. It runs: “Other beams were distinctly visible below the clouds and some of the clouds evidently became luminous and finally passed into phosphorescent auroral masses gradually becoming brighter and brighter until they became vivid streamers which shot very rapidly in various directions below the dense nimbi clouds.” This is scarcely convincing. For sounds accompanying aurora they listened attentively, and recorded hearing something on three occasions, though each time for a few moments only. Their descriptions resemble those of most others who claim t o have heard aurorae. ‘I Sounds . . as of folding dry paper or very stiff silk.” “Resembling the sound made by . . the shaking of straw.’’ 31a. M. A. W. Malin, intendant of the Museum of Gottenburg, relates, in a description of a journey made in 1842 in the Laplands of Sweden and Norway, that, during an excursion from Maunu to Lyngen on the night of the 16th of March, he observed, at a height of 3,000 feet, with the temperature at 40 degrees below zero, a polar light between himself and the neighboring mountains, and heard a crackling sound which accompanied it. 31b. The Finnish physicist, Lemstrom, has written a work entitled “The Polar Lights.” Here also are several accounts, one by a miner from Goteborg who was in Lapland in 1842. He describes an aurora observed at night in midwinter in a temperature of -45OC. A band of rays streamed up from the plateau between him and adjacent peaks and “ a rushing sound could at the same time be plainly heard.” 32. A correspondent (A. H. McC.) of the New Y w k Weekly Evening Post writes on the question of the Sound of the Aurora: in your edition of Saturday I noticed an abstract from “ Record of a Girlhood ” in regard to hearing the Aurora Borealis and therefore beg to give you my experience on the subject. I n the winter of 1846 I crossed the Atlantic from Newfoundland to Greenock in the brig Amanda. We had strong southerly winds the whole passage, without seeing the sun until after making land; three days previously a strong southerly gale carried away our only topsail, leaving us without sufficient after sail, and consequently we were driven far to the northward. The day before we made land the wind suddenly changed t o the north-west, and as night approached the sky became clear. At about 9 o’clock p.m. the captain called all passengers on deck, and a more magnificent spectacle was never contemplated-the whole heaven was a blaze of white light, the aurora darted and rushed from every point and reflected each colour of the rainbow. While it lasted we could distinctly hear the sound, as if the folds of heavy silk were shaken, sometimes sharp and quick, and then receding until the sound was lost, according to the intensity of the flash. During most of the time a book could easily be read on deck. The phenomenon lasted about four hours, during which time we all remained on deck. Next morning we made land, which proved t o be Barra Head, Southern Hebrides, and were able to lay our course.

. .

224

9. M. SILVERMAN AND T. F. TUAN

33. And now, too, a question long doubted is by me doubted no more. I have heard the Aurora; not once, nor twice, merely, but many times; not faint nor indistinct, but loud and unmistakable; now from this quarter, now from that; now from on high, and again from low down. At first it seemed to be like a field of ice cracking, then like the distant stroke of an axe; again it resembled the noise of pile driving by a monkey, and a t last like the whirring of a cannon-shot when heard from a short distance. Once, three like this followed in rapid succession, and I thought I could see the maas whence the sounds proceeded tumbling or vibrating.’ The night is intensely cold, the sky perfectly clear, the stars showing aa brilliantly through the illuminated fluid aa where the ‘lights’ are not; the wind is moderate from N.N.W. I have no doubt that we shall have heavy weather after this display. I have read that in other northern voyages, the sound of the Aurora resembles the cracking of a whip, but to-night I heard nothing like this, to my idea. 12th. Still blowing hard all day from the same quarter, N.W., as yesterday; the snow driving fast and furious. The Aurora at night was very fine, the wind having gradually decreased from sunset, and the night became very calm and fine. We again heard the cracking sounds, and our fisherman had a fine laugh at my sounding Aurora, saying that the noise is only that of the ice cracking on Bear Lake; but this solution of the question was not at all t o my taste, and I retired to rest perfectly satisfied that it waa caused by the Aurora, and not the ice. 13th. Fine and cold, with little wind. All my enthusiastic ideas respecting the Aurora’s sound are dispelled, and I find that I have, t o use a vulgar phrase ‘found 8 mare’s nest,’ for those noises which I before heard with so much rapture, as belonging to an exquisite and wondrous phenomenon, were this morning repeated in broad daylight, and are, I now see, unmistakably caused by the ice cracking. A moderate breeze in the evening from N.E.; weather cloudy. 34. I CaMOt close this letter without referring to the great value of such observations aa the following by Mr. Hardisty; which, probably, but for this attempt to follow up the phaenomenon to its fountain-head, would never have been added to the very few and much-disputed observations of the same nature which are on record. That gentleman writes,-“During a voyage in the beginning of the paat winter, I mw the most beautiful display of aurora borealis that I believe I ever witneeaed. On the 2nd of December 1860, I encamped on the banks of a moderate-sized river near the chain of the Rocky Mountains running westward of Peel’s River, the opposite banks rising precipitously several hundred feet until they joined the mountains beyond. Having no time-piece with me, I cannot speak positively as to time; but it would have been probably between 10 and 11 p.m. with a fresh breeze blowing from N.E. and very cold. The phaenomenon was evidently very near the earth, for it appeared between me and the trees on the opposite side of the river, which could not have been 40 feet above the level of the stream; the trees toward the top of the hill being high above it. Large compact masses were moving from E to W, and bright streamers passing in the same direction in quick and vivid flaahes; they returning to the zenith, would from thence spread out t o the N and S in beautiful waves or clouds, and sheets of light of the most beautiful colours, until they disappeared altogether and left the sky entirely clear. Every time the streamers passed over me from E to W they were accompanied by a rustling noise, such as ’This error respecting the Aurora’s sound affords a curious indication of the power of imagination in assisting delusion.

225

AURORAL AUDIBILITY

would proceed from the gentle waving of a silk flag; but in returning from W to E I am not conscious of having heard any sound proceed from them.” It is a confirmation of the very remarkable proximity of this display to the observer, that the following are the only other observations on the same evening, although it was clear a t four or five stations:“ Fort Confidence-A very faint band of aurora near horizon in the at 5 p.m.; 7b 30”’ it formed a pale arch across the zenith from NNW to SSE; at 8h 50”’ it exhibited a broad arch from N by W to SSE, altitude towards SW about 9 O at vertex (true bearings). Moose Factory, a t Eb 40m, a faint auroral light in N; gh 20m brighter, but partly obscured by clouds; loh still visible, never higher than 30”.” It is also reported to have been heard following marked displays in Okso on no less than four different occasions in the course of two and one-half years, viz. September 7, 1854; February 19 and April 21, 1852; and February 24, 1854. But Okso lies far out toward the sea, and even if the reports are true they cannot be given much weight since the observations were made in such exposed places. One of the most interesting of these is perhaps one from Dr. Follum, of Alten, who writes: Once, in November, 1856, on Beskades, a mountain ridge between Alten and Kautokeino nearly 1,500 feet above sea level on the occasion of an exceptionally brilliant aurora with gleaming rays of light shooting out from the crown I heard a peculiar, faint, crackling noise in the sky. My companion heard it also and I remember distinctly having stopped and remarked on the sound. K. F. Siemens &usHamburg und der Landessyndikus Dr. Hjaltalin, die vor und um 1860 auf Island beobachteten, geben an, wahrend des Nordlichts starkes Gerausch gehort zu haben. An aurora, observed by Messrs, Pease and Ketchum. at Aloukuk, a place about forty miles inland from the head of Norton Sound, in November, 1865, was described as appearing against the side of a mountain a few miles distant, so that the top of the hill waa seen above the display, and it was also accompanied with a hissing noise. The air was quite still, its temperature near -30” Fahrenheit. While the display lasted it was light enough to read the finest print. I have never myself heard these sounds with the aurora, though once, in the month of January, they were reported to me at St. Michael’s by Russian residents, at a time when, on account of indisposition, I was unable to verify the statement by personal observation, I have no doubt, however, of their occasional occurrence. I beg to assure Mr. Rouse that about fifteen years ago, early in the evening, in this very quiet locality, I listened, along with my father, to the sound of an aurora, pulsing above us, across the zenith, and appearing nearer to us, or lower, than most auroras I had seen. The sound was somewhat like the rustling or switching of silk, and we listened to it for some time with great curiosity. The aurora was not coloured, as more imposing ones have sometimes appeared, but white. It recalled to me the lines of Burns in a fragment entitled “A Vision.”

m

3538.

39.

40.

41.

42.

43.

“ The cauld blue north was streaming forth Her lights, wi’hissing eerie din; Athort the lift they start and shift, Like fortune’s favours tint as won.”

44.

Dumfriesshire, March 20. Mrs. John Myers, Brantford: I n 1870 the writer, then a young woman of 22, lived in Ingersoll. It was the

226

46a.

46b.

46.

47.

9. M. SILVERMAN AND T. F. TUAN

time of the Franco-Prussian war and all that summer there were the most wonderful displays of the aurora. Often I had to be going home very late, probably 11 or 12 o’clock and the nights were quiet. The colour was all a beautiful crimson. Generally there were two broad bands right across the sky, with shoots of crimson darting from one band to the other, and certainly there was a faint crackling sound whenever the lights moved. It was a very solemn and impressive scene and has remained with me all these years. Accounts are given by travellers in Norway of their being enveloped in the Aurora, and perceiving a strong smell of sulphur, which was attributed to the presence of ozone. M. Paul Rollier, the aeronaut, descended on a mountain in Norway 1300 metres high, and saw brilliant rays of the Aurora across a thin mist which glowed with a remarkable light. To his astonishment, a n incomprehensible muttering caught his ear; when this ceased he perceived a very strong smell of sulphur, almost Suffocating him. (Arctic Manual, p. 726). The well-known French balloonist, Rollier, who with an attendant ascended from Paris in 1870 during the siege and came down in Lifjeld, Telemarken, reports that he observed polar light rays through the light fog and, “presently a peculiar rushing sound was heard. A short time after, strong, almost suffocating fumes of sulphur were encountered.” About the 16th August, 1882, after a day’s journey, I with three cowboys, camped on the prairie some 20 miles east of the N.W.M.P. headquarters at Fort McLeod. We built our camp f i e and had supper and soon after retired to rest. The night was calm and bright. Lying awake in the tent, I heard a mild crackling noise which brought me outside quickly, fearing that our fire had not been thoroughly extinguished. The fire was dead, but the heavens to the north were showing a greater display than I had ever seen. The aurora was shooting upwards and receding with almost lightning rapidity and with varying colours. A broad yellowish splash of flame spread across from the west to the east, ascending from the horizon and proceeding with what I can best describe as a swishing noise, while at the same time a crackling noise accompanied the darting and shooting of the aurora. The whole display seemed near. Its immanency impressed me and together with the very clear audibility inspired something bordering on fear. I have been near to pine forest fires and the flames running through the branches made a crackling noise which impressed me aa similar to that accompanying the aurora which I am endeavouring to describe. I was “ brought up ” on a farm in Toronto Township and often witnessed and admired the aurora, but there never waa anything approaching the display I saw and heard in Alberta on or about the 16th August, 1882. There was no settlement. Lethbridge was not born and Calgary had three log buildings. Our camp was about 3,000 feet above sea-level. There was no one in the vicinity but ourselves, and our horses were picketed at a distance. The sound from the aurora was clear, distinct impressive and so indelible that the forty years which have elapsed have left the audibility of this grand display fresh and clear. There is no exaggeration but on the other hand my description is weak in comparison with the reality. It is by William Ogilvie, who is described as an exceptionally careful observer:“As to the Aurora making an audible sound, although I often listened when there was a brilliant display and despite the profound stillness which is favourable to hearing the sound, if any sound occurs, I cannot say that I ever even fancied I heard anything. I have often met people who said they could hear a slight rustling noise whenever the Aurora made a sudden rush. One man, a member of my party

AURORAL AUDIBILITY

227

in 1882 was so positive of this that on the 18th of November, when there was an unusually brilliant and extensive display, I took him beyond all noise of the camp, blindfolded him and told him to let me know when he heard anything while I watched the play of the streamers. At nearly every brilliant rush of the auroral light he exclaimed: ‘ Don’t you hear it?’ All the time I was unconscious of any sensation of sound.” 48a. There is a reference in Nature to observations of Major Dawson in charge of the Britiph Polar Station a t Fort Rae, N.W.T., in the winter of 1882-3. He described the sound of the aurora as “ like the swishing of a whip or the noise produced by a sharp squall of wind in the rigging of a ship.” 48b. On the one occasion when Captain Dawson says he heard it himself, “ the sound was like the swishing of a whip or the noise produced by a sharp squall of wind in the upper rigging of a ship, and as the aurora brightened and faded so did the sound which accompanied it.” If under these conditions the sound was really due to the aurora, the latter, as Captain Dawson himself remarks, must have been pretty close. 49. I have made several voyages to Canada, and I have had the good fortune to see some brilliant displays in the Gulf and River St. Lawence. One in particular was very remarkable and perhaps phenomenal. We lay a t anchor midway between Quebec and Montreal. It was a lovely summer’s night, very calm and remarkably quiet, and we witnessed a n auroral display in its full splendour. Nearly the whole of the sky, from the horizon to the zenith, was covered by it-enormous streamers were being continuously shot out, and were accompanied by a sharp crackling sound, very similar to that which is caused by certain powerful electric experiments. The sound was by no means subdued; but I cannot say it was absolutely loud, for it might give one a n exaggerated idea of it. The display lasted several hours, and was at a very short distance from the earth. 50. Professor Will C. Baker, Queen’s University, Kingston: I distinctly recollect a great auroral display seen here in Kingston in 1884 or 1885. I well remember, that it was mid-winter (I think February) for I cannot forget the three-mile drive to the farm, with its short-cut over the ice in Little Cataraqui Bay, while we observed and wondered at the marvellous sight. From sundown until 10 o’clock-and I do not know for how long afterwards-the whole sky was full of bright auroral light so that the stars were paled by it. It seemed to issue from a radiant point a t or near our zenith and to mark out the whole dome of the sky, south, east, west, and north, into wedge shaped bands of light. It looked like a striped jockey cap with its segments of coloured cloth. The wedges all centred in the zenith and reached to the horizon in all directions in about equal intensities, as far as one could judge, around the whole horizon. The colours were red, yellow and green; and they shifted with the motion peculiar to the aurora. I remember noticing here and there the brighter stars shining through it, but at times none of the smaller ones could be seen through the screen of light. One of the men in the sleigh with me called attention t o the curious crackling noise that often, though not always, accompanied the flash of a new band of light across the sky. The sound seemed to follow very closely (within a second or so a t most) of the appearance, and was like the crumpling of stiff paper. I distinctly remember a light that spread over the snow-covered ground that was noticed and discussed at the time. We put it down to a reflection of the light from the snow surface. 51, Mr. H. E. S. Asbury, President of the Montreal Centre of the Royal Astro56, nomical Society of Canada (to the Journal of which Society I am indebted for

228

9. M. SILVERMAN AND T. F. TUAN

122. the previous as well as the next quotation) states that he heard the Aurora craclfle about 11 o’clock at a place called Wood’s Harbour (about 43’30”) when a wonderful auroral display occurred, and on two remarkable other displays, one a t Durham, Grey Co. (about 44’20’N), in 1884, and another at Brampton, Peel Co. (about 43’40”) in 1893, both in Ontario, Canada, whenthe rays seemed to merge into one another and shift about, I distinctly heard sounds which I would liken more to the click made by a comb drawn through the hair, than a swish. This sound I heard on both occasions.” 52. Charles Harvey, Hamilton, Ont: I have on several occasions heard quite distinctly a sound (which may be described aa a subdued swishing sound) accompanying very brilliant displays of the aurora, The first time I noticed this phenomenon was in January, 1888. The place was three miles eaat and one mile north of Saskatchewan. I was fourteen years old and was on my way home from a neighbour’s place. I was very much impressed and considerably frightened and have never forgotten it. A few times since then I have observed the same phenomenon, always on the prairie where I lived from 1883 to 1888. 63. “Observer,” Arnprior, Ont. (Globe, March 31): I know for a fact that a sound sometimes accompanies the northern lights, having heard it on at least two occasions in northern Minnesota about the year 1892. During the fall very brilliant displays were seen, and on at least two occasions a faint but unmistakable crisp, rustling sound was heard by myself and others. 64. One man at ma, in an open boat with four natives, on Oct. 11, 1893, heard “ the most fearful whizzling and crackling sounds, sounding at times as if thousands of firearms were fired within short distance”: at the time there waa “ n o wind and no clouds.” Another writer mentions “ loud reports similar to rifle cracks,” “ the air wm still and the aurora was just above the tops of the birches;” the few loud reports were followed by much crackling. 55. See 61. 56. In the Toronto Globe and Mail for September 17, 1938, is a letter signed Robert R. Racey, Paris, Ont., recounting an experience he had over forty years ago. I a m pleased to add it to my collection of personal testimonies. Omitting the first paragraph, the letter is aa follows: During the winter of 1894 and 1895 I happened to be employed as a culler in the Upper St. Maurice River country, and on one particular occasion 1 drove all night southward from La Tuque. The ice was firm, there was little snow, the road was smooth, and my sturdy little French horse made good headway. Temperature was probably 30 degrees below zero. The atmosphere appeared to be quite clear, the heavens being brilliantly illuminated with stars. My back was turned to the north, and somewhere below the R a t River I became conscious of much light behind. I then heard prolonged, regular swishing sounds, again and again, which somewhat resembled music produced when the strings of a harp are lightly touched. On looking back, it wm observed the aurora presented a periphery whose interior surface was black, while the edge was golden colored. Rays of “light” intermittently shot skyward from the edge, and each time a ray shot upward it was accompanied by the “swishuig ” sound, and only then waa the sound audible. This evidence is similar to that supplied by several other observers and cannot be simply brushed aside as impossible. L‘

A U R O R U AUDIBILITY

57.

58.

59.

60.

61.

229

Glenn A. Green. Hamilton: In 1897 I crossed the Rocky Mountains near the Arctic circle and was near the summit for three or four months. Everything was still and the air clear. Late in the fall the northern lights were certainly quite audible. The sound is like one heard in a n electric power house, and by the Indians it is mid to be the Great Spirit whispering. I waa within the Arctic circle for four years and never tired going out of the tent to see and to listen to the lights. There were six in our party and they all said they distinctly heard the sound. I n the Dawaon Daily News for January 16, 1915, (which is a four-page paper and sells at 25 cents a single copy) is a despatch from Los Angeles, California, as follows: “Writers have tried to describe the beauty of the northern lights, but they are indescribable,” said A. B. Ferguson, a mining man of Iditarod, Alaska, who is registered at the Stowell hotel. ‘‘ Drawing a whisk-broom across silk would more closely produce the effect of the lights than anything else,” continued Mr. Ferguson. “In all the colors of the rainbow, the lights fall like a large drapery. Usually they make a swishing sound and sometimes when the lights are near the earth a weird, popping sound is heard. Back in January, 1898, I had a n experience while driving a dog team up in the interior. It was just before daylight when the lights fell so close to earth that the popping sound scared the dogs-and I guess they scared me worse.” I am afraid the above cannot be accepted at its full face value, but it seems hardly fair to rule it out entirely. I would thank the thoughtful person who sent the paper and hope to receive further reports from high latitudes. Mr. C. Carlyle Davidson, of Sorrento, B.C., writes as follows: “ I spent the winter of 1898 at Fort Graham, on the Finley River, B.C. and during the appearance of the aurora, which frequently occurred, I have distinctly heard a ‘swishing’ sound. The Indians there paid no attention to this sound, seemingly having heard it before. At Fort Graham the source of the aurora appeared to be just behind the mountain not far distant, and not as a reflection appearing in the sky, as I see it now, at times, from where I live.” John F. Henry, Owen Sound: About twenty-five years ago I was farming in the Township of Holland near the village of Strathaven. Ae I wm coming home from the village one night I noticed that the sky was bright with northern lights, very bright and changing every few minutes. I could hear a t times a long drawn out noise, faint but perfectly distinct, and a few high musical notes very sweet and clear. The night was perfectly still. I lived for many years near, and beyond, the Arctic circle and witnessed many aurorae. In the winter of 1901 I resided in Eagle, Alaska, situate on the Yukon river, latitude 64”47‘longitude 141’10’. Just below the town of Eagle, at a distance of one-half mile, or less, there is a precipitous bluff facing towards Eagle. The altitude of this bluff is about 1200 feet above the level of Eagle (Checked up with the U.S.Geological Survey.) One cold night, with temperature about 45 degrees below zero, (F.),I witnessed a singular aurora-an array of dancing streamers, having prismatic colors-which aurora was in line between me and the bluff. The streamers were about one-quarter

230

S. M . SILVERMAN AND T. F. TUAN

of the way down below the summit of the bluff, and no part thereof reached to that summit. There was a swishing sound a t times as they moved. I n fact I think that it was this sound that first directed my attention. Faint sounds oft accompany aurorae and are heard when not too distant. The coloring did not cover the full range of a continuous spectrum, but shaded off into darkness towards the red and the violet. I marvelled at the fact of nearness, as all other aurorae that I had seen appeared to be in the distant sky. I stood intently studying the phenomenon for the several minutes that it lasted. It seemed to be nearer to the bluff than to me. Amongst other thoughts I speculated as to whether the bluff had anything to do with it. At that time I had read but little with reference to the Aurora Borealis, and was entirely ignorant of the fact that the aurora was fixed by the scientists in the rare upper atmosphere. My wonder as to nearness was entirely due to the contrast between other recalled apparent distances and this certain one. There is no room for mistake or illusion in what I saw. The situation was so simple and defined. It is conclusive as evidence, and I hope that others will so regard it. A t any rate, since I cannot indulge the thought that a special display was staged for my sole delectation, it will at sometime occur to the view of others under like circumstances, and confirm my observations. 62. The following contribution from Pilot 0. J. Dahle (pilot of the Haakon Adelsten) dated March 30, 1910, was furnished me through the courtesy of Prof. Stermer: Eight or nine years ago I witnessed from the steamship Erlipg JurZ an extraordinarily interesting aurora. While our ship was crossing Vaags Bay, a little north of Harstad, a brilliant aurora in rapid motion, was seen so low down in the air that it barely cleared the tops of the masts. It flamed forth in all the colors of the rainbow and was followed by a peculiar sound, precisely such a sound as would be produced by rubbing together a well-dried skin in the hands. It was neither imagination nor the mistaking of any sound on board, but undoubtedly the result of the movement of the aurora. I have noticed also on other occasions that auroras in rapid motion hanging low in the atmosphere have emitted sounds similar to those mentioned above. In the above-named display it seemed that the auroral rays had a horizontal position and appeared as separate layers, one above the other. But the probability is that it was a vertical ray which, by reason of its nearness and its position directly over the ship, appeared to us to be horizontal and that the higher layers were the movements of the same ray seen through this identical ray from below. 63. C. B. Burns, Ottawa: While I was in the Yukon I observed the aurora many times with great interest. We used to have “aurora parties,” a t which times we would sit up for hours paying great attention t o the skies. There certainly was a distinct crackling sound, varying in sharpness apparently according to the state of the atmosphere. On my drive in 1903 with you [Mr. Cory], we heard the same sound, I remember. 64. I n the Journal for November, I read with much interest your quotation from the book, “The Canadian Naturalist” in connection with the remarks made by P. H. Gosse on the possibility of hearing the Aurora Borealis. I have seen many very brilliant displays of the Aurora Borealis in Eastern Canada, the Central West and Northern Alberta, but only on one occasion did I hear any sound. About twentyfive years ago, in Sackville, New Brunswick, on a still frosty night, as I watched the lights running across the sky, I heard very distinctly a rustling, or crackling, sound. I listened very intently as up to that time I did not know that such a sound was

AURORAL AUDIBILITY

231

ever heard. On looking the matter up in my encyclopedia I found that on rare occasions the Aurora Borealis makes a sound such as I have described, and shortly afterwards I found reference to the same matter in Wordsworth’s Poem entitled “The Complaint of a Forsaken Indian Woman.”-W. H. Harrison, 1042 Linden Avenue, Victoria, B.C. 65a. In Science for September, 1921, Dr. H. D. Curtis, Director of the Allegheny Observatory, who is well known as a highly competent and skilful observer, when writing on this subject, states: “ I desire to place on record certain early experiences under almost perfect conditions of isolation and quiet. While in charge of the Labrador Station of the Lick Observatory-Crocker Eclipse Expeditions of 1905-much of the work of adjusting the instruments was necessarily done at night. The station was located at Cartright (latitude 53”42’N), and auroral displays were frequent and bright during July and August. On several nights I heard faint swishing, crackling sounds, which I could atribute only to the Aurora. There were times when large, faintly luminous patches or ‘Curtains’ passed rapidly over our camp; these seemed to be close, and not more than a few hundred feet above the ground, though doubtless much higher. The faint hissing and crackling sounds were more in evidence as such luminous patches swept over us.” 65b. It doesn’t fit at all with the conditions at Cartright. We had light frosts several nights in the summer, but I believe they were not on the nights of greatest auroral display; from recollection, without the records of the expedition available, I should say that the temperature on these nights was around 40 degrees F. The proximity of Labrador to the north magnetic pole makes the aurora a frequent and wonderful phenomenon there; I have never elsewhere seen such bright and magnificent displays. I tried in vain to assign the sounds heard t o some reasonable source other than the aurora, but was forced t o exclude them as possible sources; besides, what I heard didn’t sound, like anything from anything I could postulate. The conditions of quiet and isolation were ideal; around midnight and no one stirring except myself within a distance of many miles; an occasional Eskimo dog howling near the post a quarter of a mile away; landlocked quiet harbour with no perceptible breeze; no waves. Any surf sounds from the open sea were four or five miles off. I doubt if distant surf, if audible at such distances, could have given the peculiar swishing sounds, and certainly not the faint crackling. I n short, I feel certain that the sounds I heard were caused by t,he aurora and nothing else. There was, moreover, a certain synchronism between the maxima of these sounds and the sweeping of auroral curtains across the sky. 66. E. A. Collyer, Toronto (Globe, April 4): Coming up the St. Lawrence in September, 1907, I witnessed a great display. The whole northern sky was aglow with shifting lights-the greatest auroral exhibit in my experience. I heard unquestionably a distinct crackling, rustling sound. The vessel was the 8.8.Parisian. There was an entertainment in the saloon below, and the deck was almost deserted. Critics will, of course, say it was sounds from the saloon which I mistook for the aurora. But they will be wrong-it came from the heavens and nowhere else. 67a. One night in January, 1908, between one and two a.m., when returning to our house at five above Discovery on Sulphur Creek, Yukon Territory, from fifty below Discovery, where a dance had been held, Mrs. Baird and I saw a wonderful display of northern lights.

232

9. M. SILVERMAN AND T. F. TUAN

They were close at hand, exceedingly vivid and a noise like the swishing of a silken garment with a n occasional crack, was distinctly audible. We had seen many similar exhibitions, but had never before known them accompanied by the peculiar rustling noise. Andrew L. Baird, Manager, Yukon Consolidated Gold Co. Ltd., Nov. 30, 1931. 67b. One night in January, 1908, the exact date I do not remember, approximately fifty people had gathered a t a dance held in the hall at fifty below Discovery on Sulphur Creek, Yukon Territory. Between one and two o’clock in the morning someone came into the hall and said that the northern lights were unusually brilliant and suggested that we go out and see the display. Everyone in the hall went immediately and we saw the most beautiful panarama I have ever witnessed. The lights were quite close and the colours unusually bright. A sound like the swishing of silk was distinctly heard. Everyone present heard it and commented on it. Many said it was the first time they had heard the sound clearly. R. L. Allen, Land Registrar, Dawson, Y.T., Nov. 30, 1931. 68. Dr. Geo. A. Fraser, Park Hill: About September 10, 1908, Mrs. Fraser and I were visiting in Miami, Manitoba, and one evening were coming home at about 10 o’clock in company with Mr. and Mrs. Robert Munro, when we saw one of the grandest displays of the aurora we ever witnessed. The whole northern sky, clear to the zenith, was one mass of shooting rays. The evening was beautiful and mild. Our attention was drawn to the “sound ” by Mr. Munro, and on listening one could hear light swishing sound mentioned by various correspondents. The Manitoba people spoke as if i t was a matter of course to hear the sound. 69. Recently the Secretary of the Society received a communication from Mrs. Consuelo Craig, wife of Mr. George Craig, of the Department of Justice, dated at Dawson, Yukon Territory, March 21, 1911, giving a clear account of auroral sounds, which I a m pleased to present below: “About 1:30 a.m., on January 26, as Mr. Craig and I were returning home after spending en evening out, we were startled by unusual sounds, seemingly from above us. We were not looking for auroral displays as the temperature was about fifty below zero, (much too cold for indulging in astronomical observation), and were skurrying home with heads bent down into our fur storm-collars, and with no thought of anything except the desirability of reaching our own fireside as quickly as possible. We were arrested by strange sounds, like the swishing and brushing together of particles of finely-broken glass. The sound came in great waves, passing slowly backwards and forward over the auroral arc. Sometimes the wave, with its musical tinkling, would almost seem to surround us; then it would recede so far aa to be almost inaudible. Then again it would come nearer, and then drop down quite near to us and then recede again up high over head. For the most part, however, the wave travelled back and forth regularly over the auroral arcs nearest to us. Our “fifty below zero” mist enveloped the heavens and the earth, but, as far aa we could me, there were no “streamers ” and no “ corona,” merely two pretty well-defined arcs. I have for many years taken a deep interest in aurorae, both light and dark, and have during my eleven years’ residence in the Yukon seen some splendid dis-

AURORAL AUDIBILITY

233

plays; but with the exception of the present instance I have never heard anything approaching a sound coming even from the brightest of them. I may mention that we were crossing the Park at the time we first heard the sounds, and everything was as near to absolute silence as could well be. As you, perhaps, are aware, there are no horses and rigs moving about at fifty below zero at midnight in the Yukon. Nothing noisier than the soft-footed Malemiut dog.” 70. On a trip to the most northerly part of Finland, in the autumn of 1911, I remained for some time 2 or 3 miles (Norwegian miles. One Norwegian mile eqyals 7 English miles.) south of Lake Enare. On returning to my hut after the day’s work on October 10, while my Finnish guides were preparing supper I witnessed the most beautiful auroral display I have ever seen. Several parallel bands, now two, now three, which alternately united and divided, streamed across the sky from the west through the zenith to the east. They were in constant wave motion, one instant slow, and deliberate and the next swift and impetuous, while rays of light darted out from them and disappeared again. On the northerly side the bands were of a reddish-violet hue while on the southerly, now bluish, now yellowish-green. The colors were repeated in each band. Little by little the aurora lost its strength and I sat down to supper. Some time I can not tell the exact instant-I heard in the north a after-unfortunately peculiar, even insistent, rumbling noise not unlike distant thunder. It was so characteristic that I jumped up to see what was going on. The aurora appeared like a bow in the north. I t struck me a t once that this must be the much-talked-of mysterious auroral sound, and in order to make sure of it I asked my two attendants if it proceeded from the aurora. They replied in the affirmative and continued their work as if it were a well-known and common occurrence. We may attribute the sound to other causes, but it will be difficult to find a satisfactory one. The air was calm; so it could not have been the soughing of the wind in the forest nor the sound of falling trees. It w&s 5 miles to the nearest inhabited place toward the north and south, 2 toward the northeast, and several to the post road. It is not likely that any travelers were abroad a t this time of the night in the winter’s cold. A river flowed past our camp on the south. Its noise could be heard constantly, but it was even, continuous, and of a different character. I wrote Mr. Waenerberg, superintendent of mines in Thule, west of Lake Enare, to ascertain whether or not the sound had been heard there. He has a long record of meteorological observations, a t least 30 years in length, made for the Finnish Meteorological Institute, and is a very careful observer. He replied as follows: On October 10, 1911, we had a very beautiful, flaming aurora over the whole dome of the sky, but no sound was heard here. It is when the aurora sinks down low over field and forest that it is accompanied by a noise similar to that of a roaring and rushing stream. Four times in 34 years have I observed this sound and reported it to various observatories, of which Mr. Tromholt’s is one. It is 60 to 65 kilometers from my camp to Thule; so it is not at all unlikely that the sound might be heard at one place and not at the other. Again, it is specially noteworthy that this reliable observer also states he had heard auroral sound, but according to his experience it is seldom so pronounced as to be heard generally. I might again remark that while the aurora flamed and played in the most brilliant hues no sound was heard. a letter from the author, dated Christiania, November 27, 71, Editor’s Note-In

234 72.

9, M. SILVERMAN AND T. F. TUAN

1913, he adds:

It will perhaps interest you t o hear that the only Norwegian member of the Scott Antarctic Expedition, Mr. Trygve Gran, once heard a peculiar noise attending a n Aurora Borealis (i.e. Aurora Australis) and Mr. Gran also told me that the party of Lieut. Campbell had repeatedly heard such a noise. 73. “Amundsen told me that when he was at Framheim, at the east end of the Barrier Reef, just before his famous dash to the South Pole, he was called out of his winter quarters one very cold night by Johannsen, in order to hear what Johannsen described as the crackling of the Aurora Australis. It was a very cold and very still night. Amundsen distinctly heard a very faint rhythmically repeated rustling noise in the air. After a time he discovered that this was due to the rapid freezing of the moisture from his breath, and the tiny tinkle made by the minute crystals as they slowly descended under gravity close to his face, sufficiently close for the ear just to catch the faint sound. He said there was no doubt about it that the rustling noises exactly coincided with the periods when he exhaled air from his lungs. He said that he was now confident that this was the true explanation of what the poets call the ‘Crackling of the Northern Lights ’.” 74. Testimony made under oath. 9 a.m. a miner by occupation. In the winter of 1914-15 I was residing on Gold Run Creek, south-east of Dawson, Yukon. My cabin was situate on a hillside, approximately 2,000 feet above the creek bottom. I left to visit a fellow miner residing a distance from the creek. When at half a mile from my cabin I suddenly became enveloped by streamers of the aurora, and to tell the truth I thought “my time had come. ” I didn’t know what t o do, but made the dash of my life for my cabin. I could see the cabin plainly through and by the light of the streamers. The noise occasioned was a shu, shu, shu, as near as I can describe it. There were also sharp quick cracks, something like a rifle report in the distant bush would sound. That’s the best description I can give of the cracking, which was quite distinct. Signed, James Lloyd. 75. Among those tendering evidence was Dr. H. E. Bigelow, Dean of Applied Science, Mount Allison University, Sackville, N.B., and at the request of the present writer he has supplied the following statement: “The experience of which I spoke to you in May occurred about 1916 during a visit to my old home a t Spencers Island, Cumberland County, N.S., on the shore of the Bay of Fundy. The time, if I remember correctly, although I cannot be sure of that, waa in the fall. The aurora was exceedingly bright and at times quite distinctly coloured, and coincident with these unusually bright flashes I heard a very low sound something like that made when you sound S repeatedly. It occurred in intensity and decreased as these long streamers shot to the sky. As I have earlier said, I waa under the impression, I don’t know why, that these displays of aurora had often been heard before, and so I attached no great significance to the fact that I heard them myself. My home is within 100 yards of the shore, but I would rule out without any question the thought that it was any sound of the waves on the beach. From my bedroom for years I listened t o sounds of that sort and they are entirely unlike any such sound that I heard on the night in question. Moreover, they were always coincident with these long streamers, and I am perfectly satisfied they were caused directly or indirectly by them.” 76. Luta Munday, Niagara Falls, Ont: I resided many years in the vicinity of Cumberland House, northern Saskatchew-

AURORAL AUDIBILITY

77.

78.

235

an, and while there I heard the northern lights many, many times. Words of mine are inadequate t o describe their marvellous beauty and the sound of them was always audible when they were vivid. I would describe the sound as a soft swishing and crackling sound, similar to that of a woman walking in a stiff silk dress. The last display I saw before leaving that truly wonderful country was in February, 1919, and it lasted over an hour. The heavens seemed to be divided into four sections: on one side, pure white rays; on the opposite side, rays of all colours; and joining the two, a vivid flame of the very brightest; and all meeting. I lay for an hour on my back on the snow watching the display and all that time I heard the same low swishing sound. In the year 1919 while surveying for the Dominion Government, in company with Mr. G. A. Bennett, another Dominion Land Surveyor, while we were camped about nine miles southeast of Broadview, Sask., we noticed the same phenomenon. The atmospheric and other conditions were almost exactly the same in every instance, and so a description of the one which took place about October 15, 1919, will serve for all; and I might here state that these displays seem to be more or less local, as people living a few miles away, while noticing that there was a brilliant aurora, did not notice the accompanying sound. The hour was about 9 o’clock p.m., not a cloud in the sky, not a breath of air stirring, and the temperature, I should judge, not lower than zero Fahrenheit. We were called out of our tents by one of our men to see the unusually brilliant display of the aurora. It certainly was a sight well worth watching, but would require a much more able pen than mine to do it justice. Some distance to the north appeared a long, wavy, brilliantly luminous belt stretching roughly in a horizontal line from east to west seemingly at a height of only a few hundred feet above the ground, and moving southward. This brilliant belt, while seeming t o fold and unfold on itself, like a bright ribbon which is continually doubling up and straightening out, seemed to keep about the same distance above the ground in its progress southward. Behind this oncoming wave of brilliance, streamers of light, more or less intense but never so brilliant as the lowest belt, kept shooting up towards the zenith, and the display seemed to fill the whole northern sky. We watched this display approaching from the north. At f i s t there was no sound, but as it got nearer, we heard a subdued swishing sound, which grew more distinct as it approached, and was loudest when the ribbon or belt of light was right overhead. The sound was at no time loud but was quite distinct for several minutes and seemed t o vary in intensity with the brilliance and the wavy motion of the luminous belt. It passed on to the south, and in a few minutes the whole sky was full of auroral streamers which seemed to culminate at a point in the zenith. A few minutes after the first display had passed over our heads, however, we could not hear the sound. It appeared as if the display was too distant for the sound to reach us, and an hour after our attention was first drawn t o it the display had faded to quite a n ordinary one. On March 24, 1922, there appeared in the Toronto a b b e a n article on the editorial page, which discussed an address before the Royal Geographical Society by Mr. G . M. Gathorne-Hardy, who claimed that he had heard a characteristic sound from the aurora while he was in Labrador in the autumn of 1920. The speaker, it was noted was a man of standing; he had been a distinguished student a t Oxford and was a wide traveller. H e was quoted as saying: Two points occur to me aa worthy of mention in this connection. The fist is that

236

79.

80.

81.

82.

9. M. SILVERMAN AND T. F. TUAN

I have occasionally seen what appears to be the aurora by day, in the form of faint clouds having the characteristic appearance of the bands and streamers. The second point I raise with some hesitation, as I believe the balance of scientific opinion is against its possibility. That is, that, judging merely by the evidence of my senses, I should say that I had sometimes heard the aurora,when in rapid movement, making a faint, crisp, rustling noise. If this is a hallucination, it is a very strange one. Answer to Question 2: Yes, at the same time as the audibility mentioned in question one. T h e approximate date was in the month of October, 1921. I waa in a canoe on the Slave river, probably half a mile above Fort Fitegerald, Alberta, returning from a short patrol up river. The time would be about 10 p.m. Weather very clear and a cloudless sky. Suddenly the whole surface of the river waa lighted up and the northern lights appeared to be right on the surface of the water, accompanied by the sound referred to in question one. Both banks of the river were discernible, the light being so strong, and at this point the river is of considerable width. It might be of interest to mention that this phenomenon was noticed at a point two miles above the head of a 10-milerapids, and the greatest display of surface contact appeared t o be nearest the head of the rapids. Corporal R. A. G. Baker, R.C.M.P., Jasper, Alberta, Jan. 26, 1931. Yes. On several occasions in the winters of 1924-26 and 1926-26 at Norman and Simpson, N.W.T. The sound is similar to the long drawn articulation of the word SHOE through pursed lips. The nativee (Eskimos)along the arctic coast, that I have spoken to, all said that they had heard the aurora borealis. Signed A. N. Bames. Aklavik, N.W.T., Inspector, Royal Canadian Mounted Police, April 20, 1931. I n reply to your request re aurora borealis: On the night of August 8, 1924, fifteen minutes after twelve, I saw the northern lights, as we call it, waving close to the ground and among the poplar trees, with clear sky above. I went out into a field of wheat close to the house and the light played around me and among the wheat like whirlwings, with a sound like silk rustling, or tissue paper. Signed, S. G. Squires, Valparaiso, Sask., Feb. 19, 1931. I n Science for September 8, 1933, Clark M. Garber gives the following evidence regarding the much debated subject of the audibility of the aurora: The proposition of the audibility of the aurora borealis has been the subject of considerable speculation and much doubt. Some scientists have claimed with much positiveness that the aurora emits no audible sounds and that the beams of light or electrical waves, such as they may choose to call them, do not come close enough to the earth’s surface t o be audible, even if any sound were emitted. I n my own mind there can be no doubt left as to the audibility of certain types of aurora, for I have heard them under conditions when no other sound could have been interpreted as such, for no other sounds were present. From the Eskimos I fist learned that the aurora could be heard and, like most people, was rather skeptical about it, believing that their statements were baaed to a great extent on their superstitions. I was told by some of the older Eskimos that when the aurora displays become audible they are able to imitate the sound by whistling in such manner that the beams of light will be attracted or drawn down to them. This, of course, is purely superstition. However, it does bring out the fact that the Eskimos were frequently able t o hear the aurora. The following is my o w n personal experience which convinced me that the aurora borealis waa actually audible. In the winter of 1925-1926, I was engaged in

AURORAL AUDIBILITY

237

making a drive of reindeer across the mountain range bordering the Arctic coast north of Cape Prince of Wales on Bering Strait. One night during this drive found me travelling by starlight &crowthe divide at the head of Nuluk River. This divide haa a n elevation of approximately two thousand feet. It was two o’clock in the morning when my native driver and I h o k e camp in order to overtake the reindeer herd ahead of us. As we climbed with our dog team to the summit of the divide we were both spellbound and astounded by the magnificent display of aurora, the most wonderful display I have ever witnessed during my eight years of life among the Eskimos. Great beams of light shot up from the northern horizon as if a battery of gigantic searchlights were searching the arctic landscape. I n front of these beams and throughout the whole length of the northern horizon great waves of iridescent light travelled from west t o east like gigantic draperies before the stage of nature’s amphitheatre. Great folds or waves, ever changing in colour, travelled one after another across the horizon and from behind them streamed the powerful beams of white light. These beams of light could be seen passing directly over our heada, and when one chanced to come over the divide it appeared to be not more than a hundred feet above the surface. The spectacle was so awe inspiring that a dog team was stopped and I sat upon the sled for more than a n hour absorbing the marvellous beauty of this most unusual display. As we sat upon the sled and the great beams passed directly over our heads they emitted a distinctly audible sound which resembled the crackling of steam escaping from a small jet. Possibly the sound would bear a closer resemblance to tho cracking sound produced by spraying fine jets of water on a very hot surface of metal. Each streamer or beam of light passed overhead with a rather accurate uniformity of duration. By count it was estimated to require six to eight seconds for a projected beam t o pass, while the continuous beam would often emit the sound for a minute or more. This particular display was so brilliant that traces could easily be seen long after daylight. 83a. The Aurora of October 15, 1926, in Norway and Sounds Associated with it. Some curious phenomena accompanying the splendid aurora of Oct. 15, 1928, were observed by me. On the night in ,question I was working aa observer of international determinations of longitude a t the top of a hill named Voxeneasen in the neighbourhood of Oslo (approximate altitude, 470 metres). I was at work in a field observatory with a transit instrument registering star transits and chronometer beats for time determinations, when an initial aurora attracted my attention. My assistant was Mr. G. Jelstrup, electrotechnical student. I was able, during intervals between my observations of time and polar stars, to observe the aurora, which was certainly one of the most splendid I had ever seen. But what is of preponderant interest is the following fact: When, with my assistant, at lgh 15m Greenwich Civil Time, I went out of the observatory to observe the aurora, the latter seemed to be at its maximum; Yellow-green and fan-shaped, it undulated above, from zenith downwards-and at the same time both of us noticed a very curious faint whistling sound distinctly undulatory, which seemed to follow exactly the vibrations of the aurora. The sound wm first noticed by me. and upon asking my assistant if he could hear anything, he answered that he noticed a curious increasing and decreasing whistling sound. We heard the sound during the ten minutes we were able to stay outside the observatory, before continuing our observations. From 20h lmt o 20h gm (Greenwich Civil Time) we registered on our radio-receiving set the rhythmic time-signals from the LY station (Bordeaux). W e secured the

238

9. M. SILVERMAN AND T. F. TUAN

whole series of tops-but a t the same time the 'aurora statics' disturbed the pen of the registering instrument. The impulses thus registered are of varying strength, and each of them is of course exceptionally well determined in time, being ' received ' at the same time as the scientific time signals. I therefore think that they may be of some interest. The maximum impulses of 'aurora statics' and their duration were: No.

Greenwich Civil Time

Duration

1 2 3 4

20" 4m 28' 60 20" 4'" 29' 49 20" 4m 39' 90 20" 4m 40" 50

0' 08 0 s 10 0' 25 0' 25

As regards the intensity of these impulses, I find that in each case the vertical component was greater than 100 microvolt/metre. When, after the reception of the time signals, we again went out of the observatory, the curious sound had absolutely ceased, and later in the night, when also the aurora had vanished, we noticed that the atmosphere was as if swept clean from statics and disturbances of our wave-length. Concerning the curious sound, I would only remark that the weather was absolutely calm when it was heard. As regards our antennae system, it may be said that it consists of 5 strands of 40 metres each. Our receiver set is a n aggregate, consisting of a three-circuits tuner, two high-frequency valves, one modulator, one heterodyne, four low-frequency valves, relay and chronograph. Hans S. Jelstrup, Astronomer to the Norwegian Geographical Survey, Oslo (Norway), Dec. 1, 1926. 83b. To the above most interesting communication from M. Jelstrup I may add the following: On Oct. 15 I had the aurora stations a t Bygdo, Oslo, Oscarsborg, Tomte, and Kongsberg in action from about 18h Greenwich Civil Time to about 2" on the following morning, and photographs were taken all the time from single stations, and from two or three stations simultaneously. About 70 successful photograms for the determination of height were secured. Only a few of these have a8 yet been measured and calculated, and they show the ordinary heights of the aurora, from 90 kilometers to more than 400 kilometres Fig. 1 [We do not reproduce Fig. 1. See the original article for this figure.] shows one of the best photographs of curtains at 19" 6'" 46', taken from Oslo to the west. The simultaneous photograph from Oscarsborg proves that the lower border of the left curtain was from 124 to 131 kilometres above the earth, the middle curtain from 103 to 114 kilometres, and of the right from 110 to 132 km. During the period from 19" 15'" to lgh 25'" when M. Jelstrup heard the whistling of the aurora, I regret that no successful photograms were taken. From the visual observations made simultaneously by Bygdo by the meteorologist Rostad, who helped me during the work, I quote the following: 19b lom. Dense masses of rays and curtains down to the horizon in E and SE. 19" 12m.The same down to the horizon in W., to the polar star in N, to the horizon in E, and down to 40' over the horizon in S. lgb 14m.The same to the Great Bear in N. lgb lBm. The same in the N and S down to 15' over the horizon. Red in S.

AURORAL AUDIBILITY

239

lgb 21m. Strong diffuse are through Can. venat. from the horizon in NW t o the horizon in NE. lgh 24m. Pulsating aurora begins. lgh 26m. Strong pulsations over the whole heavens. Red in S. During the radio signals mentioned by M. Jelstrup the following observations are noted: 20” 3”. The pulsations have ended. All over the heavens a diffuse light. In the pocket spectroscope the yellow-green auroral line could be seen everywhere, from the entire heavens, from the snow, and from everything which was illuminated by the diffuse aurora. 20h grn. Some pulsating bundles of rays. 2 P 12m. The pulsations stronger and rays being to appear.

It seems to me probable that the sound which, as M. Jelstrup says, exactly followed the vibrations of the aurora, could not come directly from the latter but from the surroundings, trees, antennae, and so on, and were caused by electrostatic discharges, which in their turn were caused by influence from the varying electrostatic charges of the aurora overhead. Carl Stermer, Bygdo, Oslo, Norway. 84. I have several times in the past heard a distinct swishing noise accompanying the aurora, but only on very cold clear nights. The occasion that is very clear in my mind is that on about the loth or 12th of January, 1927, I was crossing Wellington Bay on the south side of Victoria Island and had camped about midway between the north end of Finlayson Islands and Cape Peel. It was a bitterly cold night and upon coming out of the Snow house about 10:30 pm the moon was full and the aurora borealis was very clear, close to the ground and active. By active I mean it was very vibratory and quick moving in form. I then heard a very loud swishing noise and the aurora seemed to take a downward sweep when the noise occurred, also there were two distinct streamers that seemed to come in contact with the tops of the range of hills about 15 miles north of me. After a few seconds the streamers disappeared and the noise ceased, the aurora was again natural. This W&B so unusual that I watched for some time, but there was no recurrence of noise or streamers. At this time my dogs were unharnessed and lying curled up in the snow. When the swishing noise came so suddenly and apparently so close, most of them immediately jumped up and commenced to growl. On going into the snow house again I asked the native who was with me if he had heard any noise, and he stated that he had heard the northern lights move a short time ago. F. Anderton, Sergeant, RCMP, Tree River, W. T, May 25, 1931. 85. Recently there appeared in a newspaper a note on “Musical Auroras ” by Garrett P. Serviss, the able American writer on astronomy. He first prints a letter, as follows: Will you kindly explain the origin of the faintly audible sound that comes with the aurora borealis? This was a little more noticeable during the display lest week than of that of this week, neither being as loud as the one several years ago. The first aurora was much lighter in colour and more fluctuating. The one lest week was darker, with more colour, while the one this week was of a deeper red colour, with less fluctuations. It would seem that the brighter shades give a louder and more musical tone. Or, expressed in radio terms, higher pitch and greater volume.

240

9. M. SILVERMAN AND T. F. TUAN

When I ask any of my friends they think I am “spoofing ”, but there I S a sound to the aurora, end you may be familiar with it. X.Pittsburgh, PA., July 11, 1928. 86. When close the ionisation kaa audible in the form of a faint “crickling” or “apittering” sound, similar to a weak electrical sparking such as obtained by stroking a fur animal, and was too faint t o be heard at a distance. On one occasion a feeling of warmth t o my windward cheek waa suspected. Following this, whilst standing in the dark of my stable watching the display through the open door, I heard very distinctly, a crickling sound overhead, my attention was then directed to a vent about half an inch wide and a foot long in the board roof overhead. Here luminosity in the form of strings of sparks waa entering, end penetrated to a depth of a n inch or two down to about a foot, the depth changing erratically and at a speed that waa readily followed by the eye. 87. Q. Have you ever heard sound accompanying the Aurora? If so, what was the nature of the sound? A. Yes, quite distinctly, and it was of a hissing or swishing noise, aa of someone shaking a wisp of strew and in some cases it was so pronounced that one almost instinctively turned around to see if anything waa following. The observations giving rise to the above answer were made whilst at Langston Bay, NWT, during the winter of 1928-29. A. F. C. Tudor, Constable, Royal Canadian Mounted Police. Winnipeg, Man., J a n 13, 1931. 88. The following letter was written to me by a McGill engineering student who was fortunate enough to belong to a n Arctic expedition that, after so many years of search by so many navigators, did actually discover the North-West Passage. Dear Sir, Westmount, February 27, 1931. Yesterday evening you were kind enough to request me t o write concerning the aurora. I waa fortunate enough to be a member of a Survey Expedition to find the channel through the North-West Passage. I wintered in the Arctic a t the magnetic pole. 70 N, 96 W, during 1928-29. During the period of darkness there were intense displays of aurora, most frequently seen from the SW to SE in dead calm weather, temperature SOOF. below zero.I n the Arctic silence, it can be definitely said that whistling crackling sounds accompanied the aurora display. They were also seen several times in the NE. Their effect on radio communication was also marked. Most frequently, north and south communication predominated, and only seldom did the east and west predominate. “he communication never showed freak reception in all directions a t once, but faded on one and increased on the other. It waa difficult to get consistent data of the effects of aurora on radio communication, even though I tested and communicated daily. The same observations were noted from King William Island during 1929-30. H. Ross Smyth 89. The members of the party listened occasionally for auroral sounds during brilliant displays, but were unsuccessful except on the night of March 20; when J. Rea, assistant observer, heard sounds with a brilliant display. We were occupied with double station photographs at the time and listened carefully but unsuccessfully for the sounds. Our failure to hear the sounds may have been due to less sensitive hearing, since conditions for detecting objective sounds were extremely good.

AURORAL AUDIBILITY

241

During the day, a pilot balloon had been followed for 138 minutes, and the same clear, calm conditions prevailed during the night with a minimum temperature of -17’F. During the interval in which the sounds were heard, brilliant greenishwhite flashes were darting overhead from south to north, and some of these may have come momentarily lower than usual. Height determinations have yet to be made from the double station photographs taken at this time, but the plates show no unusual displacements. Rea is of the opinion that the sounds did not occur simultaneously with the flashes. 90. At Oh 14mGMT the whole heavens was filled with a gorgeous corona, and it was difficult to find parts of it characteristic enough to be measured. Only the base points of some rays towards the east against the dark sky behind could be measured, and a height of about 95 km was found. During this corona, which may be characterized as the maximum activity of the aurora in Norway that night, my assistant, Mr. Tjonn, a t my station on Njuke Mountain in Tuddal, 733 metres above sea-level, reported that he heard sounds with the aurora. He says in a letter to me: “During the imposing display of this big corona, where the whole heavens was like an ocean of flames, my assistant and I heard a curious sound which came from above, first from the south-west, then from the zenith and at last from the northeast. The sound lasted about ten minutes, rose to a maximum and fell down again, following the intensity of the aurora. I had the impression that it had something to do with the white rays. At that time I had taken off the earphones from my ears (when I had them on, I heard nothing of this sound) and went some steps aside to hear it better; that the sound came from the telephone is quite excluded. The sound is difficult to describe, it was similar to the sound from burning grass and spray. On the mountain it was absolutely quiet, no sound coming from wind, waterfalls, telegraph lines or motors. Both my assistant and I heard it and are quite convinced that the sound was real. On the summit of the mountain where we had our station, only a few fir trees grow. Under us on all sides we had the forest.” The same sound was heard independently of Mr. Tjonn, down in the valley Tuddal, by Oystein Reisjaa, who in a letter has confirmed in all details the description given by Mr. Tjonn. When Mr. Tjonn heard the sound, he telephoned to me; but I was so occupied by securing height determinations of the big corona that I did not pay sufficient attention to his call. On account of the great height of the aurora, it is clear that the sound could not come from the aurora itself, but from lower parts of the atmosphere; but where its origin was future observations may decide. 91. The Great Aurora of January, 1938-Its Audibility From Mr. David J. Howell, of Berrynarbor, Devon, England, has been received a bundle of newspaper clippings regarding the brilliant auroral display of January 25-26 last. One writer states that the display was more vivid than any he had seen since 1870, when, as a little boy living near Canterbury, he was taken out to 888 the brilliantly illuminated sky, which, at the time, was attributed to the burning of Paris by German troops. That display is well recorded in works on meteorology, for example in the volume by Angot in the International Scientific Series published fifty years ago. Another correspondent, C. H. Lay, F.R.I.B.A., of Aldringham, Leiston, Suffolk, under date January 27, writes: I n the descriptions I have read of the display of the Aurora Borealis I have not

242

S. M. SILVERMAN AND T. F. TUAN

seen any mention of the curious noise which accompanied the phenomenon at about 9 o’clock. The occurrence of m y noise from the Aurora Borealis is disputed by experts but that night it was clearly evident to me when I was not expecting it. As I saw this phenomenon from open country the sound could not have proceeded from any other source. 92. Further evidence regarding sound heard at the time of a n auroral display‘has been received. The writer of the following note is Miss E. M. Monro and the place of observation was on Lauder Avenue, near Regal Road, at the top of a hill north of Toronto. Miss Monro’s statement: On Tuesday, September 27, at about 10:30 p.m., I was asked to come out to see the remarkable auroral display. We stood on the lawn at the rear of the house, the position being shielded from street noises by houses, shrubbery and trees. The night waa very still and at once my attention waa drawn by a peculiar crackling noise. It was faint, but I heard it very distinctly-a sound such as is made by the leaves of trees when moved by a light breeze. I asked what the sound was and looked up t o the nearby trees and to my amazement not a leaf was stirring. Against the strong light in the sky, I could see that the leaves were motionless. No air movement could be felt when I held up my hand. My hearing is rather acute. My fellow observer, whose hearing is less acute, could hear no sound, but I heard it distinctly. It may be of interest to note that at this time the sky overhead and farther south than the position of Jupiter was covered with auroral light, but there waa none in the northern part of the sky. When more than half an hour later we went out again, the appearance was much changed. There were much more vivid colours and a brilliant display in the northern part of the sky. At this time I could hear no sound such as I had heard before. The above statement was forwarded by Rev. Dr. D. W. Best, secretary of the Toronto Centre of the RASC, and he adds these remarks: I should mention that her Statement seems to me to have value in view of the fact that on the first occasion, when she heard the sound, it did not occur to her to attribute it t o the aurora. She was in fact looking for any source-chiefly the movement of leaves in nearby trees; and, finding them motionless, waa much puzzled and enquired of me what the sound could be. I then mentioned to her that some observers believed they had heard the aurora make such a sound aa she had described. On the second occasion, perhaps three-quarters of an hour later, with this information fresh in her mind, she waa not able to hear the sound. This would SBem to rule out the usual explanation-that her experience was purely subjective. She was impressed by the absence of light in the north part of the sky on the first occasion, expecting that “Northern Lights ” should be found in the north. 93. Referring to the quotation from Humphreys’ “Ways of the Weather ” on page 168 of the April issue, Mr. David W. Rosebrugh, of Waterburg, Conn., writes: After the astounding aurora of September 18, 1941, I learned that some good friends of mine had heard sounds which they associated with the aurora. I secured statements from them which I sent to Dr. C. W. Gartlein of Cornell who is heading up the Cornell-National Geographic Society study of aurorae which was started about the end of 1938. The following is a summary of my recollections of the remarks of Messrs. Edwin K. Ellis and Kenneth Pearce of Poughkeepsie, N.Y. These gentlemen established a camp for the night of September 18, 1941, somewhere north of Maniwaki, P.Q., at about latitude 464’. Before the normal hour of darkness they were walking over farm lands hunting

243

AURORAL AUDIBILITY

for a spring of water, when Mr. Ellis remarked to Mr. Pearce, ‘I Do you hear what I hear? ” Both had heard an unusual noise and they then listened more carefully. They both agreed that they had never heard any noise quite like it before. They described it as being moderately loud and resembling the rustling of a silk dress or the continuous crushing of aluminum foil of the type once used around chocolate bars. They were not able to determine the source of the noise and could not account for it at the time in any other way than by associating it with the aurora which they then noticed for the first time as it had grown darker. 94. There is now working with Mr. Nairne F.R.S. a person of the name of Arnold, who resided seven years at Hudson’s Bay, the last three at Fort Henley. He confirms M. Gmelin’s account of the h e appearance and brilliant colours of the northern lights, and particularly of their rushing noise, which he affirms he has very frequently heard, and compares it to the sound produced by whirling round a stick swiftly at the end of a string. 95. He adds, that on conversing about this matter with a Swedish watch-maker of the name of LIND, that person assured him, that he had heard a similar noise in his own country. 96. Mr. Nairne too, one time, at Northampton, when the northern lights were remarkably bright, is confident he perceived a hissing or whizzing sound. This hissing or rushing noise, as well as that attending meteors in their passage, supposing it in both cases to be real, I would attribute to small streams of electric matter, running off to the earth from the great masses or accumulations of electricity, by which I suppose both meteors and the northern lights to be produced. Compare M. de Mairan’s Trait6 de 1’Aurore Borbale, p. 126. 97. The light of this aurora was so bright as to obscure that of the moon, then in the SE about two or three hours high, and about as many days after the full. I viewed this phenomenon for half an hour or more, which left such an impression on my mind, as I think I never shall forget. This exhibition, as I may call it, occurred about the year 1754 or 1755. The second time I saw the like appearance was about the years * It very much resembled that above described, though the second was I think, a little more south, than the first, and abounded more in red rays. I have some faint notion, that during the first exhibition, I heard the noise attending the rapid motions of the columns, though I cannot speak with any certainty, but I certainly have more than once, heard such rushings, during the vibrations of the aurora, as plainly, though not so audibly, as ever I did that of a rocket, which though fainter, i t very much resembled. 98. Similar testimony has been borne very positively by the assistants a t the Observatory at Toronto, upon one or two occasions of great display. 99. Mr. Cavallo declares he “ has repeatedly heard a crackling sound proceeding from the Aurora Borealis ” (Elements of Nat. or Exper. Phil.vol. iii, p. 449). 100. As to the question of sounds being heard, the din of carts and factories in our city, and the roar of trains in our suburbs make a n observation here for determining it impossible; while the rarity of the phenomenon in England generally keeps spectators from being on the watch. But I have heard an intelligent old man who has often gazed on the bright streamers during the clear still nights of Aberdeenshire declare that he has plainly observed sharp switching sounds to proceed from them. It seems to me probable, since electricity can change into sound and takes part in producing the aurora, that the spectacle is attended by audible vibrations. 101. With reference to the question mooted in lest week’s Nature (p. 459) by M. L. ~

244

9. M. SILVERMAN AND T. F. TUAN

Rouse as to the sounds emitted by aurorae, perhaps the accompanying extracts may be of interest. Brighton, March 20 Edwd. Alloway Paukhurst ‘‘ Record of a Girlhood,” F. A. Kemble. Vol. 1. “ Standing on that balcony (at Edinburgh) late one cold, clear night, I saw for the first time the sky illuminated with the aurora borealis. It was a magnificent display of the phenomenon, and I feel certain that my attention was first attracted to it by the crackling sound which appeared to accompany the motion of the pale flames as they streamed across the sky; indeed crackling is not the word that properly describes the sound I heard, which was precisely that made by the flickering of blazing fire; and as I have often since read and heard discussions upon the question whether the motion of the aurora is or is not accompanied by a n audible sound, I can only say that on this occasion it was the sound that first induced me to observe the sheets of white light that were leaping up the sky. At this time I knew nothing of such phenomena or the debates among scientific men to which they had given rise, and can therefore trust the impression made on my senses.” 102, De la Pilaye will das Geriiusch auf New-Foundland nur einmal gegen das Ende 103. eines Nordlichtes gehort haben; wahrend nach Steward dasselbe in der SanctLorenzbucht, Nordamerika, in stillen Nachten stets gehort werde. 104, Und widersprechen dem Grafen Trampe, danischem Stiftsamtmanne, der das 105, Knistern wahrend des Nordlichts zugab, aber glaubte annehmen zu miissen, dass 106. dasselbe von dem Schnee und Eis herriihre und in den hellen Nachten infolge der K@lteentstehe. Jorgenson in Reykjavik und Kaufmann Hygom in Hafnafjordur wollen das Gerausch unzahligemal gehort haben. 107. Diesen widersprach wieder der Begleiter der beiden Reisenden, James Hay, ein Shetlander, der behauptete, dass auf der Insel Unst, wo er zu Hause sei, stets mit dem Northernlight ein Gerausch verbunden sei, als ob jemand Kaffeebohnen in einem Siebe aushiilse. 108. Auf den Hebriden will es Dunbar (“Edinburgh Journal of natur. and geogr. Sciences,” Neue Serie, IV) wahrend sechsjahrigen Aufenthalts mehr als fiinfzigmal gehort haben. 109. Kapitiln Abrahamson (“ Schweiger Journal,” Neue Reihe XV) bringt mehrfache Bestiitigung der Thatsache durch Ohrenzeugen bei. 110. Ein Wiiohter des 250 pariser Fuss (81 m) hohen Leuchtthurms zu Sumburgh-Head, Siidende der Shetlandsinsel Mainland (+60°), wollte dasselbe jedesmal gehort haben, wenn ein Nordlicht aufleuchtete, und sogar bei verschlossenen Fensterladen daraus auf eine Nordlichterscheinung geschlossen haben. l l l a . Gisler, qui a longtemps habit6 le N. de la Suede, dit: “La matiere des aurores bor6ales descend quelquefois si bas qu’elle touche le sol; au sommet des hautes montagnes, elle produit sur la figure des voyageurs un effet analogue celui du vent.” E t ailleurs: “J’ai souvent entendu le bruit des aurores, il ressemble ir. celui d’un fort vent ou au bruissement que font quelques matibres chimiques dens l’acte de leur d6composition. . . .J’ai cru souvent trouver que le nuage avant l’odeur de fum6e ou de sel bhl6.” l l l b . Nach Gissler und Hellant wird (wie Wargentin berichtet) im nordlichen 112. Schweden daa Gerausch und Brausen wie bei Wind oder chemischen Zersetzungen in der Luft gehort. Pontoppidan spricht ebenfalls von dem Geriiusche. 113. Prediger N. Hertzberg in Ullensvang sagt 1826, dass aus seiner Jugend, als das Nordlicht badger gewesen sei, er sich erinnere, wahrend desselben ein Gerausch wahrgenommen zu haben.

AURORlPL AUDIBILITY

245

114. Nach Petri hort man das Nordlicht-Gerausch in Esthland. 115. Winkler will es in Gera gehort haben und ebenso sol1 es dort von vielen Leuten haufig( 9 ) wahrgenommen worden sein. 116. In der neuern Zeit behauptet der Astronom Brorsen, auf Alsen, sich unzweideutig von dem Gerausch iiberzeugt zu haben. 117. I read with much interest an article in the April number of the ‘ Observatory ’ on the Aurora by Mr. Henry J. Webber, in which he mentions a brilliant display seen by him in Canada which was accompanied by a sharp crackling sound. I have much pleasure in bearing testimony to the truth of his statement, having heard similar sounds in a display I witnessed in Lt. 59”23’N, and long, 2’52‘W. June 22, 1887, R.Rendall, Master R.M.S. ‘Cou~land.’ 118. It is interesting to note that at the reading of Mr. McDiarmid’s paper there were present four persons, members of a prominent Ottawa family, who have heard the sound of the aurora. It was in Gasp6 some years ago, in the month of August. The aurora waa a very brilliant one, forming a corona overhead with diverging streamers covering almost the entire sky. The movements of the streamers were very rapid, and accompanied by a sound, described as of a hissing or whizzing sound, and very distinct. The sound recurred at every brilliant flash for more than a n hour, and was heard by all the members of the family (six). The night waa perfectly calm, and the place far enough from the sea to preclude the supposition that the sound could be confounded with that of the waves, to which indeed it had little resemblance. 119. A more extended report by Tromholt’s father, “ a skillful and reliable meteorological observer about whose trustworthiness there cannot be the slightest doubt,” is of great interest. He purports to have heard the sound three times in all and eays that he is sure it could not reasonably be attributed to other muses. 120, A Mrs. Craig gives an instance of these sounds being heard at Dawson in the 121. Yukon, and Mr. R. Flaherty also frequently heard them at St. John, N.B., and Halifax, N.S., Canada. 122. See 51. 123. Dr. 0. C. J. Withrow, Toronto (Globe, April 7): For many years I lived in Fort William, Ont., and there one sees the aurora in all its brilliancy and grandeur. I n the silent midnight hours I have been frequently called from my bed in the practice of my profession, and have many times seen the heavens swept by the majesty of this manifestation of nature’s handiwork. Many a time I have heard a swishing sound, which I have always felt came from the aurora. 124. Miss Rose Duncan, Forest, Ont. (Globe, April 7 ) : I lived on St. Joseph’s Island a great many years, in a very quiet place, where I could account for every sound, especially on a winter’s night, and have on several occasions heard sounds I could attribute t o nothing else but the northern lights which were very bright. Sometimes the sounds would be quite plain. I do not recall hearing sounds unless the lights were plain. 125. Mary D. Kennedy, Toronto ( a b b e , April 8): I was brought up in the country, in Nova Scotia, and have known all my life that sometimes during certain displays a soft, slithering whisper can be heard. My father used to tell us t o “ listen and you will hear them change.” At such times they seem very near, and move quickly. 126. F. G. Horner, Bracebridge, Ont. ((%be, April 13): Having spent several years in the Yukon, three winters being dogteam work that kept me in the winds, where a tent was our chief habitation, I have had B very

246

127.

128.

129.

130.

131.

132.

9. M . SILVERMAN AND T. F. TUAN

good opportunity to observe the aurora in all its grandeur near the Arctic circle. The sounds were very plainly heard in the very cold, dry atmosphere when the aurora was most active and showed the colours of the rainbow. A. J. Woodward, Mimico, Ont. (based on many years of daily meteorological observations (Globe, April 22): There is a distinctly audible noise from the aurora, but only from general displays that produce running waves from horizon to zenith in about one second. If these waves are wide, and consequently nearer, a rustling sound is produced, and the narrower the wave, the sharper the noise, almost to a crackling sound. Mrs. H. F. Nash, Toronto (Globe, April 25): We were staying at the Kootenay Hotel, on the Arrow Lakes, B.C. A party of us took a trip to the Minnie Mack Mine and the summit of Silver Mountain, 11,000 feet above sea level. While there we had a wonderful view of the northern lights. Great long shafts of light spread out in the shape of huge fans, waving from side to side like powerful searchlights. We heard a gentle swish, faint but still audible. We watched them for a long time. Mrs. Isabel Davids, Toronto (Globe, April 10): One very cold night many years ago I was walking along College St., when I was attracted to a magnificent display of northern lights, which took on different shapes, at once resembling a huge fan spreading wide, then part way, then closing, and other designs after the manner of a kaleidoscope. The colours were delicate blue, pink and corn colour. While gazing upward I distinctly heard coming from the sky a sound resembling the rustling of silk. One of the partners of the North West Company related to me the following singular story: He was travelling in a canoe in the English River, and had landed near the Kettle Fall, when the coruscations of the Aurora Borealis were so vivid and low that the Canadians fell on their faces, and began praying and crying, fearing they should be killed; he himself threw away his gun and knife that they might not attract the flashes, for they were within two feet of the earth, flitting along with incredible swiftness and moving parallel to its surface. They continued for upwards of five minutes, as near as he could judge, and made a loud rustling noise, like the waving of a flag in a strong breeze. Miss Marietta L. Dingle, Toronto: While living in Winnipeg with others of my family, and while enjoying some very brilliant displays of the aurora, more than once I have distinctly heard a n accompanying sound like a rustle of silk or tissue paper, which certainly was allied to the aurora very closely, and appeared to follow its waves of light as it travelled across the heavens. Many times during three winters I spent in Winnipeg, 1891-2,1892-3, and 19046, and occasionally in the autumn I saw wonderfully beautiful displays spread out upon the sky softly, like colour on a map, but having no audible accompaniment. When the sound was noticeable the aurora seemed to travel rapidly, in waves of light. The sound was not a t all loud, but was decidedly arresting to music-loving ears, and is difficult to describe as it is so distinctive-neither rustling, nor crackling, nor swishing, but a mingling of all three sounds, faint and distant. Major L. T. Burwash, Ottawa: During the years I have spent in the Canadian Northland I have, on many occasions, been greatly interested in the aurora borealis and have observed it closely. There is absolutely no doubt as to its audibility. I have heard it most distinctly

AURORAL AUDIBILITY

133.

134.

135. 136.

137.

247

and have discussed this phenomenon with many persons at the time of its occurrence when all noted identically similar hissing or crackling sounds. That these were directly caused by the aurora, there can be no doubt as they coincided with it as to time, commencing as the light of the aurora began to intensify, following it in its course across the sky and dying as the light faded. These sounds are very similar to those known to wireless operators as ‘‘ static disturbances,” which so frequently interfere with the receiving of aerial messages. The effect of these disturbances when intensified or concentrated by the ordinary wireless instruments may be divided into several divisions. (1) The more or less constant sounds, much like those made by a n effervescing fluid. (2) The sharp staccato sounds which come a t more or less regular intervals. (3) The heavier and duller sound coming a t somewhat irregular intervals and resembling the “knock ” which is heard in a defective engine. The sounds accompanying the aurora are identical with the first two of the above divisions, the first not being as intense as I have heard from a wireless instrument, and the second, while distinctly present, being much subdued. The third and heaviest disturbing sound I have not heard from the aurora. Mr. W. W. Cory, Deputy Minister of the Interior, states that his experience in the Yukon and in Manitoba corroborates Major Burwash’s comparison of the auroral sounds to the first two kinds of wireless disturbance mentioned above. 0. S. Finnie, Department of the Interior, Ottawa: My recollection of the auroral displays in the Klondyke is that the “crackling ” sound was not present, but I distinctly remember hearing a “swishing” sound when the aurora was at its greatest intensity and waving in the sky like a blanket or a sheet. See 63. Later, about twenty years ago, being again at my old home, I was awakened by the distress in my spine which, I have learned to know, means that the morning papers will tell of some disturbance of the telegraph and telephone lines. I went out to investigate, and was rewarded by one of the most wonderful of the many memorable experiences of a long life. I n the north-eastern sky was a small but magnificent aurora, in form resembling the photographs of a sunspot, but coloured, reds predominating. Accompanying, was music which made one recall the poetic Old Testament imagery, “When the morning stars sang together and the sons of God shouted for joy.” It was unmistakable, and was like the music produced by violin strings when the violin is on or near a piano. With reference to your letter of the 22nd instant relative to the Aurora Borealis, I am rather at a loss to know just what to say. I n reply to your first question, I feel quite convinced that I have heard, and quite distinctly heard, a sound accompanying the Aurora. On the other hand, I realize that scientists may prove conclusively that the Aurora makes no sound. I n such case it would be apparent that I must have been mistaken. However, I do not think I was mistaken. The sound I heard on more than one occasion when living in the Yukon was, I believe, unmistakably caused by simultaneous flashes of the Aurora. The nature of the sound might be described as swishing. To be more precise, it was like that produced by a handful of birdseed being thrown in the air to fall on hardwood floor. F. H. Kitto, Director, National Development Bureau, Department of the Interior, Canada, January 31, 1931.

248

9. M. SILVERMAN AND T. F. TUAN

I n reply to question No. 1, I might state that I have personally heard sounds accompanying a display. They were of a cracking nature not unlike the crack of a whip, and also a swishing sound. They were distinctly audible on a clear calm night of extremely low temperature, and the stars also appeared t o assume an unusual brilliancy. G. I. MacLean, Gold Commissioner, Yukon Territory, March 21, 1931. 139. Certainly where there is an unusual display of Northern Lights and where they are not only very active, but cover the whole sky, there is a peculiar faint noise, so faint that unless there is absolute silence it cannot be heard. Often I have heard it where the blanket was over my face and I did not know that “ the lights were on.” But whether it comes from the Aurora or not I was not able t o determine. It sounds like two pieces of paper being softly drawn over one another. This description, of course, I had heard many times before I had seen the lights. But it fits very well, I made many experiments of stopping my earn and then suddenly exposing them, and always the same sound was again heard. I wondered at the time whether it might not be the action of the intense cold on the eardrums, as it is very akin to the singing in one’s ears caused by coming down rapidly from a high to a low level. Moreover, the noise is not steady, but seems to wax as the streamers form and re-form into various patterns. On clear cold nights when there was no display of lights, I have tried t o distinguish the same sound, but could not do so. Geo. A. Mulloy, Forest Service, Dept. of Interior, Ottawa, J a n 27, 1931. 140. I have certainly heard the Aurora many times, both in the Yukon Valley and in the Western part of the North-West Territories. The sound resembles that produced by a fluid effervescing, or like the sound made by drawing silk over silk. I may say that at times the aurora is active enough to produce sound, I have observed it in company with a number of other persons, all of whom agreed as to the sound and described it in a similar way. L. T.Burwash, NWT and Yukon Branch, Dept. of the Interior, Ottawa, J a n 27,

138.

1931.

Both Reg. No. 10621, Constable May and myself are positive we have heard the aurora many times. I have also questioned a number of Indians, and they definitely aasert that they have heard them, especially when camped in the higher altitudes. The aurora streamers have a sound like the faint swishing of pure silk. I have noticed when the aurora is brilliant, the dogs’ ears stand up as if they were listening. Corporal A. B. Thornthwaite, RCMP Old Crow Detachment, Y.T. 142. I believe that sound does accompany the displays, and on many occasions while on hunting trips, sleeping in the open, I have heard this sound which might be compared to the distant rustling of silken drapes, and on these occasions I have noticed distinct uneasiness manifested by dogs in camp. This uneasiness I have noticed particularly in the white Siberian breed, and the dogs whine and turn about in circles when this sound becomes noticeable. When sheep hunting in the mountains of the Alaska Range, I have seen brilliant displays and have seen the streamers come down in front of the mountain at night when camped in the valleys. The brilliant nights in the mountains make the snow covered mountains stand out very clearly, and at times the auroral rays distinctly come between one and the background. Douglas 0. Preston, Fairbanks, Alaska, Sept. 23, 1931. 143. In Juneau it was between 7 and 8 o’clock, if my memory is right. I was to ferry the

141.

AURORAL AUDIBILITY

144.

146.

146.

147.

249

channel from Juneau to Douglas. Quiet, silent, clear sky. The aurora consisted of big waves of white rolling from behind the mainland mountain walls, (to the east), and tumbling below some few hundred feet, over the channel whose entire width they covered in the shape of a sheet. There was audible now and then a light sound as the rustle of a stiff silk robe of a person walking. The waves had the shape of heavy cylindrical masses rolling down on the channel, and after unrolling themselves into a sheet, they would ascend in spreading over the hills (low mountains) on each side. Joseph R . Crimont, S.J., Bishop of Alaska, February 11, 1931. One evening several years ago in St. Stephen, N.B., I was about two hundred yards from a building blocking the end of a dead end street when I noticed the streamers of the aurora apparently almost touching the building and extending very close to the ground in the vacant lots adjoining to the south, then apparently almost touching an old chapel, the next building to the south. The streamers appeared much below the eaves of the two buildings. There was a slight rustling sound or crackling noise accompanying the streamers. H. P. Moulton, Geodetic Survey of Canada, Dept. of the Interior, Ottawa, Feb. 6, 1931. Mr. A. L. Roberts, of 629 Belmont St., Montreal, under date May 29 writes: “The aurora as seen by me a t St. Denis St., in Montreal, which I described in a letter to Dr. A. Vibert Douglas of McGill University shortly after, agrees with the accounts given in the May-June issue of the Journal. “It was the sound that first drew my attention to the phenomenon. It was a quiet evening, not a soul in sight, not a sound of a car of any kind and no perceptible breeze. The aurora appeared to be of great extent in all directions but not very intense in luminosity; I could not say that it touched the ground, though it appeared to do so, but I could see no ray between myself and the buildings. I appeared to be in the centre of the zone, in the midst of a mighty hall, draped with luminous gossamer curtains which seemed to waft with some mysterious breeze. “ The sound was similar to a person moving about dressed in voluminous folds of taffeta-silk, faint but unmistakable. It had nothing to do with sound in the head. I am not subject t o influence by fancy, besides I had already observed, on previous occasions, a number of distant effects. There is no doubt about the audibility of the aurora, but it is evident that one must be well within the affected zone to hear it.” Here is a letter, exactly M written, sent to me by a trapper in the Province of Quebec. Such men live much in the open both by day and by night, and they see much of Nature: I1 a QtQ donne ici une conference par un de vos professeurs dont je ne puis me rapeler le nom, le sujet 6tais les Aurores-BorQalesil disait donc que c’est impossible d’entendre le bruit des dit aurores. J e suis trapeur et peut predire la temperature deux 8. trois jours d’avance par bien des signes quel vous ne connaisser pas et quand des Aurores-Borales se produisent sa agit sur la temperature c’est par la place ici de dire sur quel cote du temps. Mais soyez certain aussi que certaine aurore produisent du bruit crepitement faible c’est vrai e t aussi comme un bruit de soie qu’une personne froisserait bien a vous. Un trapeur. This brought forth a letter from Frank R. Porter, of 228 Main St., Biloxi, Miss., dated September 19, which has been sent to me by the Director of the Harvard

260

9. M. SILVERMAN AND T. F. TUAN

Observatory and is given below. I shall be pleased to receive reports of observations at any time: The aurora borealis does emit sounds-a sort of swishing or hissing sound, followed by a crackling sound like breaking a handful of small dry twigs, the time consumed from the first hissing to the crackling being about two seconds. I observed this phenomenon in Connecticut and the aurora was in small patches of the curtain variety and floated from the north very low down, seemingly not over a hundred feet distant. I heard no musical sounds. This description, I believe, will be verified by a friend who was with me at the time and who is, I think, still living in New Haven, Conn. It took place many years ago, however. 148. One account which we heard of audible aurora indicates a subjective origin for the sound. In this case the aurora was heard in central Saskatchewan from the observation platform of a moving train. Here the noise of the train would have drowned all sounds coming from the aurora, and having the low intensity commonly reported. 149. Definitely the noise can be heard on certain rare occasions. For I have heard it. As a child I lived on the Saskatchewan prairies, in the early days. And on a number of occasions had, with my parents, brothers, and sisters, witnessed amazing displays of the Northern Lights. Sometimes they were almost overhead. And on one or two occasions we definitely heard them. It was at such times, as when the bright dancing “curtains ” would suddenly dash across the sky at tremendous speed that the soft swish could be heard, and the sound and movement were absolutely together. What I mean is that the sound did not follow the dis. play as thunder follows lightning. It went with it. And the sound was like a whispered “Sh! ” And at no other time-except when this rush across the sky was made-was there any sound made by the dancing Aurora. Peggy Perm Harvey (author and artist). Letter to National Geographic, April 9, 1948. Courtesy of Gale Sprague, Cornell. 16Oa. There is a passage in the “Germania” of Tacitus (chapter xlv), which I do not think can have ever been examined by the historians of natural science, or it would have created a considerable stir amongst them. Side by side with a plain accountprobably the earliest written o n e - o f an arctic twilight, there lurks in i t a description of the aurora boreelis, which moreover lends countenance to the still prevailing notion that the northern lights are accompanied by sound. Speaking of the Suiones, a tribe on the northern borders of Germany, the great writer says:-“ Beyond them is another sea, calm even to stagnation, by which the circle of the earth is believed t o be surrounded and confined; because the last gleam of the setting sun lingers till he rises again, and so brightly that it dims the stars. It is believed too that a sound is heard, that the forms of gods and raysfromahead are seen (persuasio adjicit sonum audiri insuper formas deorum e t radios capitis adspici). Up to that point[however]-and the report [I have given] is true-everything is natural.’’ 16Ob. With reference to the paeaage of Tacitus, “Germ.” 45, quoted in Nature, vol. xxiii, p. 459, I would suggest that the reading equorurn, proposed by some commentators, is far happier than decorum. “It is believed that a sound is heard, that the forms of the horses and rays from a head are seen.” 161. Gmelin, Reise durch Siberien, vol. 111. p. 136. As the whole passrtge is very remarkable, and has never, that I know, appeared in English, I thought the following translation of it might be acceptable. “For however fine the illumination may be, it is attended, as I have learned from the relation of many persons, with such a hissing, cracking, and rushing noise

...

AURORAL AUDIBILITY

152.

153.

154.

165.

261

throughout the air, as if the largest h - w o r k s were playing off. To describe what they then hear, they make use of the expression, Spolochi chodjat, that is, the raging host is passing. The hunters who pursue the white and blue foxes in the confines of the Icy Sea, are often overtaken in their course by these northern lights. Their dogs are then so much frightened, that they will not move, but lie obstinately on the ground till the noise has passed. Commonly clear and calm weather follows this kind of northern lights. I have heard this account, not from one person only, but confirmed by the uniform testimony of many, who have spent part of several years in these very northern regions, and inhabited different countries from the Yenisei to the Lena; so that no doubt of its truth can remain. This seems indeed t o be the real birth-place of the aurora borealis.” It is here to be observed, that Gmelin did not collect the account himself, but extracted it from letters or papers of M. de 1’Isle de la Croyere’s, who was himself far to the northward of Yakutsk, without hearing these noises; probably, therefore, it is much exaggerated, though one can scarcely suppose the whole to be fabulous. I have not heard the noise ascribed t o the Aurora, but the uniform testimony of the natives and of the residents in this country, induces me to believe that it is occasionally audible. The circumstance, however, must be of rare occurrence, as is evidenced by our having witnessed the Aurora upwards of two hundred times without being able to attest the fact. I was almost inclined, last year, to suppose that unusu~1agitationsof the Aurora were followed by storms ofwind; but the more extended opportunities I enjoyed of observing it in 1821, at Fort Entreprise, have convinced me that no such inference ought to have been drawn. Dr. Richardson states: “ Ihave never heard any sound that could be unequivocally considered aa originating in the aurora, but the uniform testimony of the natives, both Crees, Copper Indians and Esquimaux, and that of older residents of the country, induces me to believe that its motions are sometimes audible. These circumstances are very rare, as will appear when I state that I have now had a n opportunity of observing that meteor for upwards of two hundred different nights.” Captain (afterwards Sir Henry) Lefroy was in charge of the magnetic observatory a t Toronto from 1841 to 1853. From September 23, 1843, to February 29, 1844 he was at Fort Chipewyan on Lake Athabasca (W. Long, 7h 3Sm 16. = 123’49’ N. Let. 6E043’),afterwhich he went to Fort Simpson (W.Long. Eb 6m 40’ = 121’36’. N. Lat. 61’51’7). With regard t o the much disputed question of sound, neither the writer nor his assistant, Serjeant Henry, was ever positive of hearing any, but the latter thought he did so upon one or two occasions. The result of inquiries upon the subject was, that opinions were nearly equally divided among the educated residents in the country: a small majority of those the writer consulted, agreed that a sound sometimes accompanied the phenomenon, but among the uneducated and native inhabitants, whose acuteness of sense is probably much superior t o that of the other class, a belief in the sound is almost universal, and many individuals assured the writer they had heard it. Sir John Richardson and Dr. Rae when in search of Sir John Franklin and were at Fort Confidence on Great Bear Lake (W. Long, 7” 5Sm 16’ = 118’49’. N. Lat. 66’ 54’) from October, 1848, to March, 1849. With respect t o sounds of the aurora, the belief prevails in the Arctic regions that it is occasionally audible when very bright and active, at which times it is believed by the natives to be near the earth. Having witnessed the phenomenon some thousands of times without hearing it, I have become sceptical of its ever

262

9. M. SILVERMAN AND T. F. TUAN

producing sounds audible on the surface of the earth. The sounds it is said to came are liked by many to the rustling of silk; and I may observe that the curtainlike appearances and motions of the brightest aurorae are likely to be associated with the remembrance of such sounds, and also that the formation of minute icy spiculae in very cold clear nights is accompanied by a crackling in the air. 156. All Indians, both on the shores of Hudson’s Bay and near Bear Lake, and the Eskimos on many parts of the coast, assert positively that the bright, varying, flickering, and rapidly-moving aurorae do produce sound. The senses of hearing and smelling in the Indian and Eskimo are far more acute than in the civilised man; and both sounds and smells which to the latter are not perceptible are perfectly 80 to the more sensitive auditory and olfactory organs of the savage. 157. And Musschenbroek mentions that the Greenland whale-fishers assured him they had frequently heard the noise of the Aurora Borealis, but adds that “ no person in Holland had ever experienced this phenomenon.” 158. In the article “Aurora Polaris,” Encyc. Brit., edition ix., the writer admits the evidence of scientific Arctic voyagers having listened in vain for such noises; but, referring t o the statements of Greenlanders and others on the subject, concludes there is no a priori improbability of such sounds being occasionally heard, since a somewhat similar sound accompanies the brush-discharge of the electric machine. 159. Auch Biot erhielt von den Bewohnern der Insel Unst die Versicherung der Horbarkeit des Gerausches; er selbst vernahm es jedoch dort niemals. 160, Billings spricht ebenfalls von dem in Sibirien horbaren Nordlicht-Gerhusch. ,161. Nach Finsch berichteten die Ostjaken in der Nahe der Obmundung, 1876, nur von einem schwachen Gerliusch. 162. See 115. 163a. The article also gives the evidence of Captain H. P. Dawson, who was in charge of the British Polar Station at Fort Rae in 1882-83 and who wrote: The Indians and the voyageurs of the Hudson’s Bay Company, who often spend their nights in the open, say that it (the sound) is not uncommon. 163b. According to Captain H. P. Dawson, in charge of the British Polar Station at Fort Rae in 1882-1883, “ The Indians and voyageurs of the Hudson Bay Company, who often pass their nights in the open, say that it [sound] is not uncommon . . there can be no doubt that distinct sound does occasionally accompany certain displays of aurora.” 164. Lemstrom himself, according t o his o w n report, has never heard the auroral sound, but he specifically states that he is convinced of its existence, and this assertion ia repeated several times. He speaks of the Laplanders’ firm belief in the sound in the following language: They say that a rumbling sound can often be heard and since it had frequently been observed by skilled observers, their belief that it actually exists in connection with strong, energetic auroras at low temperatures is absolute. 165. According t o Mr. Hudson Stuck, FRGS, in “Ten Thousand Miles by Dog Sled,” many residents and travellers in Alaska, men whose work and statements are worthy of belief, claim that they have heard the Aurora swish across the sky.” 166. See 141. 167. The Eskimos were questioned only after we found that they were calling US “ foolish white men,” because we had said that the aurora was inaudible. Natives living in the region extending from Repulse Bay in the north to Eskimo Point in the south, and from Baker Lake in the west to Southampton Island in the east were included in the inquiry.

.

AURORAL AUDIBILITY

263

All the white people insisted that they had heard rustling or swishing sounds accompanying brilliant auroral displays. These sounds were heard more frequently by them at Burrel on Hudson Straits, at Harrison on the east coast of Hudson Bay, and in the region between Chesterfield Inlet and the Churchill River than farther north. Bishop Turquetil, who has had more than twelve years’ experience at both Reindeer Lake in northern Saskatchewan and Chesterfield Inlet, stated that sounds with aurora were more frequent at the former place. Nothing in their description of aurora a t such times would suggest unusually low aurora, because forms common to normal displays were always mentioned. The natives in this region call the aurora, abhanik, meaning that which moves rapidly. The same expression is used also by them for Chesterfield Inlet, because of the rapid flow of the tides in the Inlet. Stories involving the audibility of the aurora have been handed down from one to another for many generations. One that is repeated frequently explains that the sounds are made by spirits playing a game. A small group of the natives have also a story of the aurora coming low enough to kill some of their people. In this particular case the story may have originated from the effects of a destructive lightning flash, although lightning is so rare in this region that the natives often seek shelter with white men during even the mildest storms. Very few natives from Baker Lake, Chesterfield Inlet, and north of these places, had heard the aurora, although all knew people who had heard it. Natives from Southampton Island knew of no one who had heard it there. Practically all natives from south of Chesterfield Inlet had heard the aurora, and described the sounds by blowing through rounded lips. According to them the aurora was heard more frequently during some winters than others. None had heard it during the winter of 1932-33. Nothing of a definite nature concerning low aurora could be found from them. The collected testimony of both whites and natives indicates that the region of maximum audibility lies in the region of maximum auroral frequency. Unfortunately, this does not distinguish between an objective or a subjective explanation of the sounds, since the greater the number of aurora seen the more likely it is that conditions favourable to either effect may occur. The extent to which the testimony of natives can be relied upon is debatable. As observers of unusual occurrences in their native habitat they are superior to the average white man. However, traditional accounts of the sounds may be faulty and have induced a greater susceptibility to a subjective effect. 168. Audio Noise from Aurora In any discuasion of auroras with persons who have lived in Alaska and have been in isolated places during the winter months, the question always arises concerning the noise of the aurora. Almost t o a man, these trappers and miners are convinced that a noise can be heard during extremely intense and active displays. Undoubtedly they have heard something, but what is it? Scientists who have attempted t o detect audio noise from the aurora almost invariably have encountered negative results. There are, nevertheless, a few very reliable observations that make it advisable t o reserve judgment on this problem. For example, Carl Stsrmer on one occasion recorded noise accompanying a n auroral display. In a separate report another observer also revealed hearing the noise, thus providing independent confirmation of Stsrmer’s observation. 169. P. H. Gosse, in his book, “The Canadian Naturalist,” p. 47 (London, 1840) has the following remarks on the possibility of hearing the Aurora Borealis: “ I have

264

170.

171.

172. 173.

174.

176.

176.

177.

178.

9. M. SILVERMAN AND T. F. TUAN

never heard any sound accompanying the Aurora Borealis, though I have seen very many, and some very splendid ones; and though I have often eagerly and intently listened; yet I cannot doubt the fact, for I have been assured by persons of undoubted veracity that they have distinctly heard an accompanying sound, though exceeding rarely. Some of these individuals could not be suspected of having taken the idea from books, yet the character of the sound attributed to the Aurora exactly agrees in all the recorded instances in which it has been heard. It was described to me as being like the rustling of a silk flag in a small breeze. These were all heard in Newfoundland, where it is much more common than in this country (Quebec Province).” Edmonstone (‘I Philosoph. Transactions ”, 1784) bezieht sich auf Aussagen von Seeleuten, welche unter +63”30’ . . das starke Rauschen oder Brausen des Nordlichtes wollen gehort haben. dagegen behaupteten Leute aus verschiedenen Standen auf den Shetlandsinseln einstimmig, dass bei starken Nordlichtern Gerausch vorkomme. In Finmarken wollen, nach Keilhau, die Bewohner das Cferausch gehort haben; er selbst vernahm dasselbe niemals. Herr Carl Bock, who accompanied the Laplanders visiting this country (at the Westminster Aquarium) in 1877-78, and who witnessed many brilliant auroral displays in Lapland, assured me he could trace no noise, except on one occasion, when he heard a sort of rustling, which he attributed to the wind. The Laplanders themselves did not associate any special noise with the Aurora. . wahrend der Student des Collegiums zu Reykjavik, Oddur Gisalon, behauptete, dam nie ein Islander das Gerausch vernommen habe, und sich erbot, jeden Bewohner Reykjavik8 zum Zeugen aufzurufen. Joseph Skaptwen, Districtsarzt in Hnauser (Hdnavatnsylsla), Oddur Thorarensen zu Akureyri am Eismeere, friiher zu Reykjavik, und Svein SkGlason, Redacteur der Zeitung “Norari” zu Akureyri, wollen niemals ein Gerausch gehort haben. It may perhaps be superfluous to state that neither here nor in any other place have I heard the mystic auroral sound. Neither has it ever been heard by any of the Icelanders I have aa yet met with. But it is a little disconcerting when trained explorers have carefully listened for the sound, under apparently favorable conditions, but have heard nothing. Recently I came across an instance of this kind in a book by a missionary-explorer of our Canadian North-West, Egerton R. Young, entitled, “ Stories from Indian Wigwams and Northern Camp-Fires.” Referring to some fine aurorae, he says: “ Often have I seen a cloud of light flit swiftly acroea these ever-changing bars with a resemblance aa natural to that across the strings of a harp that I have suddenly stopped and listened for that rustling sound which some arctic travellers have affirmed they have heard from these auroral displays; but although I have often watched and listened amid the death-like stillness of this dreary land, no sound have I ever heard. Amid all their flashing, changing glories they seemed as voiceless as the stars above them.” Resultate der Polarlicht-Beobachtungen angestellt im Winter 1882 und 1883 auf den Stationen Kingua Fjord und Nain, Von Dr. K. R. Kock (Berlin: A. Asher and Co., 1886) These [the auroras] were, at both stations, completely mute; not a suspicion of audibility attended their movements. Audibility of Aurora Borealis.-The above question is up I submit my testimony. Somewhere about 1842 or 3 when I was ten or eleven years of age, one night in the

.

. ..

..

A U R O R a AUDIBILITY

255

fall of the year after the frost season had set in, my mother bade me get on the horse and ride two miles to our country store in Stafford, N.H. on a n errand. It was fairly dark when I started and there was a bright aurora in the northern sky. The road was straight north so I faced the aurora for the whole distance. I walked the horse every step of the way as I was not then, and have never since been, able to ride faster, because of some abdominal sensitiveness. The horse knew the road and I had nothing to do but observe the aurora. Everything was quiet along the road. It ran past farms 1/4 or 1/2 miles apart. I was no more afraid than if I had been riding in sunlight. I have never seen an aurora whose movement was more brilliant. The general movement waa from west to east and consisted of repeated quick rushes of light, wide above, and coming to a point lower down-roughly one might say it was an exhibition of triangles of light very acute at the lower point. These would be thrust down and then withdrawn and then the exhibition would advance again from west to east. Now I as distinctly heard what I have always called a “ swish ” attend these movements as I saw the light. The exhibition with various intensity was kept up during my slow return home. Once the noise was more marked and seemed nearer than a t other times. It sounded like the “crackle” or the burning of a hemlock broom. Now I trust the hearing of my boyhood’s ears on that night as implicitly as I do the vision of my eyes. The one waa no more a pseudotion than the other waa a pseudopion. I heard the “ swish ” and “ crackle ” and aaw the light for nearly two hours. I have never heard a n Aurora since though I have seen many. But that a thing is of rare occurrence is not proof that it does not occur at all. Charles Caverno. Lombard, Ill. July 5, 1911. 179. On Hearing The Aurora I heard it frequently in Norway, as far back in childhood as I can remember. We were all very used to hearing it, as well as seeing it, and it wasn’t much talked about, unless it was unusually wild. I lived on the Island of Arney, just above 70 degrees north. I left this place at age 18, and although I lived at other places in Norway, and aaw the Aurora, I heard it only at Amey. As children, we heard stories about how the aurora would take you if you teased it. I recall vividly one night when I was eight years old, the year 1894. It waa winter, cold and clear, and night. The lights were very wild, and the roar of it frightened me so that I ran inside. I thought they would take me. They were not within reach, but still icy, and wild. The sound waa like shaking paper. Usually the lights were up too high to hear. You could hear them only when they were low (still way out of reach, of course, but low) and wild. This happened often in winter, on cold, clear nights. The lights have many colors, like the rainbow, but they are not that bright. We thought that a wild display meant that stormy weather was coming. Mrs. Albertine Johnson 180. The Audibility of the Aurora has been mentioned in Popular Astronomy, I would be glad to qualify and testify aa witness if such testimony is admissable. Few Indians, I believe, who live out of doors in the old northwest or in the prairies of western Canada would question the occasional audibility of the aurora. Scientists and explorers do. First they ascribe sounds described by the Indians to ice in cold weather. Second they think the aurora to be too high above the earth to be audible.

266

9. M. SILVERMAN AND T. F. TUAN

First. It Beems t o me that although the sounds have similarities more difficult t o describe than to differentiate in experience, it would be easy to prove that the vibrations of one ought to be much more rapid than the other and therefore of a different pitch. Second. Accredited scientists have seen aurora so low as to appear between the observer and a house at a little distance. When living in houses or surroundings common to congested communities or where mind and ears, in any locality, are occupied with immediate employment or surroundinga I have never heard the gentle sound of the aurora but have heard lower pitched sounds of miles of ice on Lake Shetack, Minnesota, sing under temperatures of 34 degrees below zero and lower. I n 1888 studying surveying at Carleton College, we did field work with the theodolite. One day the needle would touch the glaas when our fingers touched the glass. The variations of the needle were not constant. etc. We reported to you, and after explaining the cause, an electric storm, and that probably the electric storm was due to a sunspot appearing in the sun, we were shown the sunspot on the limb of the sun through the telescope in Goodsell Observatory. The memory of your kindness to us that day has always been a plmsure. I n the years since at one time for a tonic I took static electricity, seated in a n isolated chair under an inverted cone or helmet, from the latter of which electricity descended when the generator waa started. In late August 1904 on an isolated hill in S.W. Minnesota, far from people and moving things, in absolute quiet perfectly alone with nature, on a still and lovely but cool evening, when enjoying the great out of doors, the gentle rustling, crackling, snapping sound aa if a shower of static or other electricity had descended behind me, caught my ear. I turned quickly and looked N.E. and behind me low, with compressed subarc cloud and depressed arc, there glowed not brightly a n aurora. It might have rebounded from the earth. When I had walked 400 feet, the aurora wm high, are usual height and streamers becoming more bright, shooting higher and higher toward the apparent focus. There wm no sound except in the possible descent. Then however the aurora attracted my attention to itself by its audibility. Oak Park, Ill. Oct. 10, 1911. WM. J. PELL. 181. Judge M’Cord, of Montreal, a n attentive and accurate observer, informed me that he had, on one occasion, and only one, heard a sound-a “ rustling noise,’’ which he attributed to an aurora then prevailing. The possibility, however, that it might have proceeded from some other source is obvious; and perhaps such a conclusion is more probable than that this, among hundreds which he had observed, should alone afford such a token of its presence. Trevelyan observed, that the 182. Aurora borealis in Faroe and Shetland.-Mr. Aurora borealis in Faroe and Shetland was often seen very low, not more than 40 or 60 feet above the level of the sea; and he learned that in both countries it is frequently heard. In Faroe Mr. T. met one person who stated, that when the colour of the Aurora borealis is dark red, and extends from west to east with a violent motion, he had experienced a smell similar to that which is perceived when a n electric machine is in action. 183. The Esquimau, like the Indians, assert that the Aurora produces a distinctly audible sound, and the generality of Orkney men and Zetlanders maintain the same opinion, although, for my own part, I cannot say that I ever heard any sound from it.

AURORAL AUDIBILITY

257

184. The sudden glare and rapid bursts of these wondrous showers of fire render it impossible to observe them without fancying that they produce a rushing sound; but I am confident there is no actual noise attending the changes, and that the idea is erroneous. I frequently stood for hours together on the ice, t o ascertain this fact, at a distance from any noise but my own breathing, and thus I formed my opinion. 185. I listened with great earnestness, and once or twice thought I heard a rustling noise, but I think it must have been the wind. When the wind was hushed, as i t was at intervals in the latter part of the night, not a sound could be heard. 186. The members of the French Commission of Bossekop, from whose work we have already borrowed so much, did not neglect this question. Once, for instance, on January 10, 1839, during a very brilliant aurora, Bravais notes: ‘ I listened with much care for possible sounds; the circumstances were favourable, the air and sea calm, yet I only heard a very faint whistling sound, coming I knew not whence, but without doubt independent of the aurora, since it was continued after the latter had faded.’ At another time, October 31, 1838, the sound appears to have been more distinct, for Lottin notes as follows: ‘At one moment I, by an instinctive movement, took my cap off to listen better, fancying that the rays darting above our heads made a sort of crackling sound; perhaps the sound was due to the distant steps of some one, perhaps of some animal on the hardened snow. M. Silijestroem, at some distance from me, close to his house, had the same illusion at exactly the same hour.’ Finally, Thomas, a mining engineer at Kaafiord, near Bossekop, who made a t the same time as the French Commission very complete observations of the aurora borealis, notes on March 10, 1840: ‘A noise resembling the rustling of straw was distinctly audible, and seemed to coincide with the darting of the rays of the aurora. This sound was only heard when the rays were near the zenith.’ I n spite of these observations, which appear to be very definite, the members of the French Commission pronounced rather against the existence of any sound proper to the aurora. ‘Although I dare not,’ says Bravais, ‘ question the validity of the testimony in regard to it, we must yet conclude that this sound is very rare. Moreover, during these observations the ear may be deceived by more than one source of error against which it is impossible to be too much on one’s guard; such as the whistling of the wind, the drifting of the dry snow, the distant murmur of the sea, the crackling of snow which begins to freeze again after a temporary thaw, &c.’ Siljestroem, one of Bravais’ companions, concludes in almost identical terms. 187. Although the belief that the aurora can be heard is widespread, reliable observations of the sound are very rare. The sound is usually described as resembling the rustle of silk, and is said to be noticeable more frequently in some years than others. On this occasion only two writers stationed in the British Isles reported sound; one heard faint clicks although his hearing is “ b y no means perfect” and another heard (at about 19h. 45m.) a sound like that made by deer crashing through undergrowth, though presumably less loud. Unfortunately neither said what type of aurora was present at the time, but it is worth noting that the electric current estimated for the magnetic displacements was probably almost immediately over the second observer at 19h. 45m., when the display was at its brightest. 188. There is no point relating to the Aurora Borealis which is more disputed than the sound which some say accompanies the phenomenon, at all events at certain times. It is described as of various natures, viz., crackling, whizzing, and hissing, from nearly every part of the world where the Aurora Borealis is visible, and the faith

268

9. M. SILVERMAN AND T. F. TUAN

in the “ sound ” is as orthodox among the Eskimo of Greenland and the Lapps of Finmarken aa the Tchuctches of New Siberia. . Without absolutely refusing to believe in the possible existence of such a sound, I fancy that there must be some acoustic deception or misunderstanding which haa created this belief in a n auroral sound. During my stay at Koutokaeino I was daily surrounded by people who believed as firmly in the sound as in the Holy Gospel, yea, at Bossekop they even told me that they did not tliink there was any Aurora Borealis at all until it whizzed, and I still maintain, that of all the intense aurorae I have observed in various pasts of the Arctic regions, and which I am sure I have watched with more attention than is generally bestowed on them, every one hap been perfectly silent. It is a very general belief in certain countries-for instance, in the Orkneys, in Finmark, and among the Indians of the territories round Hudson Bay-that the aurora is accompanied by a particular sound, somewhat resembling the rustling of silk. The Lapps, who also believe in the existence of this sound, compare it to the ‘cracking’ which may be heard in the joints of the reindeer when in movement. A greet number of trustworthy observers maintain that they have distinctly heard this sound during very vivid auroras. Others, on the contrary, have never remarked any sound which in their opinion could reasonably be attributed to the aurora; we must note, however, that purely negative results cannot be set against a single positive and certain fact. . but I cannot put a nearer date on the event than the early 1960’s. The sound, for vibratory quality compares to some output of the electronic Onde Mertinote (c.f. Pys-Colpix NPL 28023 side one tracks 3/4 and closing bar) but I heard a more ‘swishing ’ element; an ebbing-and-flowing. I n Cleveland the north sky at night (of the Tees-side industrial complex) carries the glow of iron- and steel-works, and (particularly since my boyhood) a very heavy chemical industry including oil-cracking, amongst other lights and fumes. I differentiated the aurorae from all this by the colors (silver-yellows and green aa well as reds) and the features that they did not reach the earth or burn SO steady. Letters from Colin Simms, York, England, 4 May 1971 and 27 November 1971. On many occasions in Presque Isle, Maine and Easton, Maine I heard noises on very clear and very cold nights when there was very active and extensive aurora present. These occurred in the winter in the 1930’s. The most recent occurrence was in 1960 near Bangor, Maine. Private communication from A. G. Cogswell When I was a little girl living in Easton, Maine (1933-38) with my sister and brothers we often heard noises on very clear and very cold nights that appeared to be related to the movement of the aurora in the early evening hours when we would be out sliding. Private communication from Mrs. Lorna Markham, December, 1971. Aurora borealis, the scientists call them, but to US old sourdoughs they will always be the northern lights. Many nights I have stood outdoors almost freezing (and getting a kink in my neck), gazing upward almost by the hour when the display has been active and colorful. To me it is always a source of wonder and delight, and at such times I wish I were a poet or had the magic of words t o express my feelings. Often I just hurt from the awe and beauty and wondrous mystery of the lights. First there would be a few wisps of wavering silvery movement here and there across the sky. Soon they would all gather like a large, wide ribbon waving in a

. .

189.

190.

..

...

191.

192.

193.

AURORAL AUDIBILITY

259

fan’s breeze, faster and faster, gathering more loose wisps of light. Then, before my very eyes would come a gorgeous rainbow of color-greens, purple, orange-with a sound just like the crackling of a whip. Then in seconds the whole sky would be clear. Over on the horizon, to the left or right, they would start to gather again, and come waving back over my head. Sometimes the whole sky would be aglow with fiery red or gorgeous rose. On very cold nights they came so low it seemed as if I could reach my hand into them. They would hiss and crack and go on their mysterious way, back into oblivion. Scientists a t our University of Alaska say the northern lights do not make any sound, nor do they come down low. Many of us old-timers do not agree. I am no scientist. I can only try t o describe what I have actually seen and heard many times during my fifty-two years in Alaska. 194. For instance, one of the most common current beliefs is that the Aurora Borealis makes crackling or swishing noises, and that by applying the proper formula it can be engaged in conversation. . . Reports abound that the Aurora whistles, crackles, swishes, snaps or howls. Early Eskimos took for granted that the northern lights made noises, usually ascribing them to the spirits that resided in the lights. When the Eskimos of eastern Greenland heard the lights swishing, they said they were the spirits of children whirling and twisting in their games and dances. Eskimos of northern Canada believed that the whistling and crackling sounds were the footsteps of departed souls tramping about on the snows of heaven. The Eskimos around Ungava Bay in Canada could hear the spirits speaking to them in a whistling kind of noise, which they took pains to answer in a similar voice. These particular spirits they thought were intermediaries between the living and the dead. The Eskimos of western Alaska say today that “things are not the same as they used to be,” because in the early days the northern lights howled a great deal more than they do now. 195- These are noted from a questionnaire put out by Professor Donald E. Olson, 198. Department of Physics, University of Minnesota, Duluth, Minnesota. Much additional information is given in the questionnaire which is not included here. The observers were Mrs. Gust Bohlin (197);Gust Bohlin (195);Solveig Johnson (196);Arnold Janz (198).

.

...

REFERENCES Papers which include some mention or discussion of auroral sounds ere designated by an mterisk. With the exception of Johnson (1930)and Foghammar (1960)all of these have been consulted during the preparation of this paper. Akasofu, S-I., and Chapman, S. (1963).The lower limit of latitude (U.S. sector) of northern quiet auroral arcs and its relation to Dst(H).J . Atmos. Few. Phys. %, 9-12. Albertson, V. D., Tripathy, S. C., and Clayton, R. E. (1970)Electric power systems and geomagnetic disturbances. Presented at the 1970 Midwest Power Symposium, Ames, Iowa, October 1970. *Amer. J . Sci. (1824)8, 392. *Angot, A. (1897)“The Aurora Borealis,” p. 46ff.Appleton, New York.

260

9. M. SILVERMAN AND T. F. TUAN

*Anonymous. (1933). Low auroras. Nature (London) 181, 828-829. Astapowitsch, I. S. (1934). Quart. J. Roy. Meteorol. SOC.60, 493-504 (see p. 501). See also (1957) Meteorilca, No. 9, 71-101. *Bannister, H. M. (1866). Letter. Smithson. Inat., Annu. Rep. p. 411 [Publ. 18721. *Barnard, F. A. P. (1838). On the aurora borealis of November 14,1837. Amer. J.Sci. 34, 267-290 (SW pp. 289-290). *Beds, C. 5. (1933a). The audibility of the aurora and its appearance at low atmospheric Can. 27, 184-200. Ievels. J. Roy. Astron. SOC. *Beds, C. S. (193313). Audibility of the aurora and its appearance a t low atmospheric levels. Quart. J. Roy. Meteorol. SOC.69, 71-78. Low auroraa and terrestrial discharges. Nature (London)182, 245. *Beds, C. S. (1933~). *Belknap, J. (1786). Letter containing observations on the aurora borealis. Trans.Amer. Phil. SOC.2, 196-197. *Blagden, C. (1784).Account of some late fiery meteors. Phil. Trans.Roy. SOC.London 74, 201-232 (see particularly pp. 227ff). *Burder, G. F. (1881). Sound of the aurora. Nature (London)23, 529. *Campbell, W. H., and Young, J. M. (1963). Auroral-zone observations of infrasonic pressure waves related to ionospheric disturbances and geomagnetic activity. J. Qeophys. Res. 68, 5909-5916. *Capron, J. R. (1879). “Aurorae: Their Characters and Spectra,” pp. 33-34 and 44-45. Spon, London. *Caverno, C. (1911). Audibility of aurora borealis. Pop. Astron. 19, 453-454. Chalmers, J. A. (1967). “Atmospheric Electricity,” 2nd ed. Pergamon, Oxford. 1967. “Chamberlain, J. W. (1961). “Physics of the Aurora and Airglow,” p. 147. Academic Press, New York. *Chant, C. A. (1911). J. Roy. Astron. Soc. Can. 6, 154-155. *Chant, C. A. (1915a). J. Roy. Astron. SOC.Can. 9, 35-36. *Chant, C. A. (1915b). J. Roy. Aatron. SOC.Can. 9, 100. *Chant, C. A. (19150).J. Roy. Astron. SOC.Can. 9, 268-269. *Chant, C. A. (1915d). J. ROY.A8h’On. SOC.Can. 9, 369-370. *Chant, C. A. (19150).J. Roy. Astron. SOC.can. 9, 457-468. *Chant, C. A. (1916). J. Roy. Astron. SOC.Can. 10, 97-99. *Chant, C. A. (19234. The audibility of the aurora. J. Roy. Astron.Soc. Can. 17,273-284. *Chant, C. A. (192313). J. Roy. Astron. SOC.Can. 17, 305-306. *Chant, C. A. (1928a). J. Roy. Astron. SOC.Can. 22, 28-29. *Chant, C. A. (1928b). J. Roy. Astron. SOC.Can. 22, 396-398. *Chant, C. A. (1929a). J. Roy. Astron. SOC.Can. 25, 418. ‘Chant, C. A. (1929b). J. Roy. Astron. SOC.Can. 23, 464-465. *Chant, C. A. (1931). J. Roy. Astron. SOC.Can. 25, 43-44. *Chant, C. A. (1932a). J. Roy. Astron. SOC. Can. 26, 36-37. *Chant. C. A. (1932b). J. Roy. Astron. SOC.Can. 26, 448-449. *Chant, C. A. (1933a). J. Roy. Aatron. SOC.Can. 27, 256257. *Chant, C. A. (193313).J. Roy. Astron. SOC.Can. 27, 394-395. *Chant, C. A. (103th). J. Roy. Astron. SOC.Can. 82, 157-158. *Chant, C. A. (193813).J. Roy. Astron. SOC.Cam. 32, 313-314. *Chant, C. A. (1938~). J. Roy. Ashon. SOC.Can. 82, 396-397. *Chant, C. A. (1938d). J. Roy. Aatron. SOC.Can. 435-436. *Chant, C. A. (1943a). J. Roy. Aetron. SOC.Can. 37, 168. *Chant, C. A. (1943b). J. Roy. Astron. SOC.Can. 87, 254-265. *Chapman, 5. (1931). The audibility and lowermost altitude of the aurora polaris. Nature (London) 127, 341-342.

a,

AURORAL AUDIBILITY

261

*Chapman, S., and Bartels, J. (1940). “Geomagnetism,” Vol. 1, pp. 466-467. Oxford Univ. Press (Clarendon), London and New York. Chernosky, E. J., and Hagan, M. P. (1958). The Zurich SUnSpOt number and its variations for 1700-1957. J . Ueophys. Res. 83, 776-788. Chernosky, E. J., Fougere, P. F., and Hutchinson, R. 0. (1965). The geomagnetic field. In “Handbook of Geophysics and Space Environments” (5. Valley, ed.), Chapter 11. McGraw-Hill, New York. *Chree, C. (1910). Article on aurora. In “Encyclopedia Brihnnica,” 11th ed. *Clerke, A. M. (1887). Review of books on aurora by Lemstrom and Koch. Nature (London) 85, 433-436. Cole, A. E., Court, A.. and Kantor, A. J. (1965). Model atmospheres. I n “Handbook of Geophysics ” (S. Valley, ed.), Chapter 2. McGraw-Hill, New York. Cole, R. K., Jr., and Pierce, E. T. (1965). Electrification in the earth’s atmosphere for altitudes between 0 and 100 kilometers. J . Ueophys. Reg. 70, 2735-2749. Constable, F. C. (1881). Sound of the aurora. Nature (London)24, 53. *Cresswell, G. (1968). Fire in the sky. AZaskaSportsmn 84, No. 1, 27. Currie, B. W. (1955). Auroral heights over west-central Canada. Can. J. Phys. 88, 773779.

Dalton, J. (1828). On the height of the aurora borealis above the surface of the earth; particularly one seen on the 29th of March, 1826. Phil. Trans. Roy. 800. London 118, 291-302.

*Dalton, J. (1834). “Meteorological Observations and Essays,” 2nd ed. Harrison & Crosfield. Manchester. *Davies, F. T., and Currie, B. W. (1933). Audibility of the aurora and low aurora.Nature (London) 132, 855-866. Davis, H. (1951). Psychophysiology of hearing and deafness. I n “Handbook of Experimental Psychology” (S. S. Stevens, ed.), Chapter 28, pp. 1137-1138. Wiley, New York. Denning, W. B. (1902-1903). On the sounds alleged to precede or accompany the flights of meteors. J. Brit. Astron. Ass. 13, 277-279. *Derham, W. (1726-1727). Observations on the Lumen Boreale, or streaming on Oct. 8, 1726. Phil. Trans. Roy. SOC.London 34, 245-252 (see, particularly, p. 261). *Dixon, F. E. (1948). The Aurora of January 26 t o 26, 1938 and associated magnetic storm. Ueqphys. Mem., London 10, No. 81, 1-16. Dolezalek, H. (1964). ‘‘ Study of Atmospheric Electric Tracing of High-altitude Irregularities,” Final Report, Contract A F 19 (628)-3289. AVCO Corp., Wilmington, Massachusetts. Donati, G. B. (1872). Phenomena manifested in telegraphic lines during the great aurora borealis of February 4, 1872; and the origin of northern 1ights.Smithaon. Inst., Annu. Rep. pp. 299-309. [Publ. 18731. Egeland, A., and Omholt, A. (1967). A statistical study of Carl Stormer’s height measurements of aurora borealis. In “Aurora and Airglow” (B. McCormac, ed.). van Nostrand-Reinhold, Princeton, New Jersey. Ellyett, C. D. (1969). Radio noise of auroral origin. J. Atmos. Terr. Phys. 51, 671-682. *Elvey, C. (1957). Aurora borealis. Aduan. Electron. Electron Phya. 9, 1-42 (for comments on sound, BIW pp. 20-21). *EIvey, C., and Rust, C. (1962). Can You hear the Northern Lights? Alaska Sportsman 28, No. 6, 1%19. *Encyclopedia Britannica. (1810). Article on Aurora. *Eve, A. S. (1936). Northern Lights. Proc. Roy. Inst. Ut. Brit. 29, 166-188 (see particularly, pp. 172-175).

262

9. M. SILVERMAN AND T. F. TUAN

Ferenczi, 8. (1912). “Symbolism ” (translated in “Sex in Psycho-analysis,” p. 217. Dover, New York, 1956). Fletcher, H. (1923). Physical measurements of audition and their bearing on the theory of hearing. J. Franklin Inst. 196, 289-326. *Foghammar, R. (1960). 1st das Nordlicht horbar? Koenlos 66, 183-184. ‘Force, P. (1856). Record of auroral phenomena observed in the higher northern latitudes. Smithson. Contrib. Knowl. 8, Appendix, 1-1 10. *Franklin, J. (1823). Excerpt. Ann. Phil. [2] 6, 57. *Freier, G. D. (1961). Auroral effects on the earth’s electrical field. J. Geophya. Res. 66, 2695-2702. Freud, A. (1946). “ The Ego and the Mechanisms of Defence.” International Universities Press, New York. Frey, A. H. (1961). Auditory system response to radio frequency energy. Aerosp. Med. 83, 1140-1142. Frey, A. H. (1962). Human auditory system response to modulated electromagnetic energy. J. Appl. Physiol. 17,689-692. Frey, A. H. (1965). Behavorial biophysics. Psychol. Bull. 83, 322-337. *Fritz, H. (1881). “Daa Polarlicht.” F. A. Brockhaus, Leipzig (see Chapter 12 on sounds). *amber, C. M. (1933). On the audibility of the aurora borealis. Science 78, 213-214 (reprinted as Chant, 1933b). *Gartlein. C. W. (1947). Unlocking secrets of the northern lights. Nat. Geogr. Mag. 92, 673-704 (comments on sound on pp. 693-695). Gartlein, C. W., and Moore, R. K. (1951). Southern extent of aurora borealis in North America. J. Geophya. Res. 66, 85-96. *Gartlein, C. W., Kimball, D. S., and Sprague, G. (1967). “Visual Aurora,” Vol. 1, unpublished report. Cornell University, Ithaca, New York (see, particularly, pp. 66-67). Gartlein, C. W., Kimball, D. S., and Sprague, G. (no date). “The Origin and Morphology of the Aurora,” unpublished report. Physics Dept., Cornell University, Ithaca, New York. Gherzi, E. (1960). Atmospherics on 20 kc/s a t the time of local aurora. Nature (London) 187, 225-226. *Grand Dictionnaire Universe1 (Larousse). (1866). Article on Aurore. Greely, A. W. (1886). “Three Years of Arctic Service,” pp. 413ff. Scribner’s, New York. *Griffin, J. G. (1922). Audibility of the aurora. J. Roy. Astron. SOC.Can. 16,255-260. *Gromnica, E., “An unbelievable story about Northern Lights ” North 18 No. 4, 44-46 (July-August 1971). *Hansteen, C. (1827). On the polar lights, or aurora borealis and Australis. Phil. Mag. [2] 2, 334-344 (see paticularly, p. 340). *Harang, L. (1951). “The Aurorae.” Wiley, New York (see, particularly, pp. 60-61). *Harvey, P. P. (1948). Letter to National Geographic. Henry, J. (1881). Smithson. Misc. Collect. 80, Art. 1, 46. *Henry, J. (1866). Comment on letter of Bannister. Smithson. Znet., Annu. Rep. p. 412 [Publ. 18721. *Holyoke, E. A. (1828). Letters. Amer. J. Sci. 14, 98-101. *Hood, R. (1823). Excerpt. Ann. Phil. [2] 6, 68. *Hubbard, E. (1881). Sound of the aurora. Nature (London) 24, 5. *Hubbert, J. (1860). Observations [of the aurora of 18691 at Grafton, Canada West (lat. 44”3’N, long. 78”s‘). Amer. J. Sci. 79, 252-253.

AURORAL AUDIBILITY

263

*Hunsucker, R. D. (1963). The Northern Lights. Alaska Sportsman 29, No. 3, 8. Hutchinson, W. C. A,, and Stromberg, I. M. (1989). New pulse technique for measuring point discharge in the atmosphere. Nature (London) 225, 654-655. *International Cyclopedia, New York. (1892). Article on Aurora. ‘Jelstrup, H. S. (1927). The aurora of October 15, 1926, in Norway and sounds associated with it. Nature (London)119, 45. *Joannis, A. (1886). Article on Aurore. I n “La Grande Encyclopbdie.” Paris. *Johnson, A. (1971). “On hearing the aurora’’ Letter to D. Olson. *Johnson, J. H. (1927). On the altitude of the aurora. PubE.Astron.Soc. Pac. 39,347-350. *Johnson, J. H. (1930). “Concerning the Aurora Borealis.” Gazette Press, Berkeley, California (privately printed). Jergensen, T. S., and Ungstrup, E. (1962). Direct observation of correlation between aurorae and hiss in Greenland. Nature (London) 194, 462-463. *Kemble, F. A. (1881). Quoted by E. A. Paukhurst, “Sound of the aurora.” Nature (London)28, 484. Kikuchi, G. (1970). Observations of the atmospheric electric field at Syowa Station, Antarctica. J. Met. SOC.Jap. 48, 452-460. *King, W. F. (1907). Audibility of the aurora. J. Roy. Astron. SOC.Can. 1, 193-194. Kremser, G. (1967). Balloon measurements of auroral X-rays. I n “Aurora and Airglow ” (B. McCormac, ed.), pp. 477-488. Van Nostrand-Reinhold. Princeton, New Jersey. Kremser, G. (1969). Some characteristics of auroral zone X-rays. I n “Atmospheric Emissions” (B. McCormac and A. Omholt, eds.), pp. 181-191. Van NostrandReinhold, Princeton, New Jersey. *Lefroy, Capt. (1862). Second report on observations of the Aurora Borealis, 185051, made by the non-commissioned officers of the Royal Artillery, at the various guard-rooms in Canada. Phil. Mag. [4] 4, 59-68. *Lemstrom, S. (1874). Observations upon the electricity of the atmosphereandtheaurora borealis, made during the Swedish expedition of 1868 to the North Pole. Smithson. Inst., Annu. Rep. pp. 227-238. Lemstrom, S. (1898). On the earth-currents and the electrical currents in the atmosphere and their relations to the earth-magnetism. Forh. Ofversigt Fimka Ve t e m k q sSOC.41, 60-104. Ley, W. (1969). The sound of meteors. I n “Another Look a t Atlantis,” Chapter 13. Ace Publ. Co., New York. *Loomis, E. (1866). The aurora borealis, or polar light. Smithson. Inst., Annu. Rep. pp. 208-248 (see p. 222). Lovering, J. (1868). On the periodicity of the aurora borealis. Mem. Amer. A d . ArtsSci.

10, 1-361. *McC., A. H. (1881). Nature (London) 24, 109. McClelland, D. C. (1962). The psychodynamics of creative physical scientists. I n “Contemporary Approaches to Creative Thinking” (H. E. Gruber et al., eds.). Atherton Press, New York. (Reprinted in D. C. McClelland, “The Roots of Consciousness,” Van Nostrand-Reinhold, Princeton, New Jersey, 1964). Maeda, K., and Watanabe, T. (1964). Pulsating aurorae and infrasonic waves in the polar atmosphere. J. Atmos. Sci. $21,15-29. Maeda, K. and Young, J. M. (1966). Propagation of the pressure waves produced by auroras. J . Geomagn. Gwelec. 18,275-299. Malan, D. J. (1963). “Physics of Lightning.” English Univ. Press, London. Martin, L. H., Helliwell, R. A. and Marks, K. R. (1960). Association between aurorae and

264

S. M. SILVERMAN AND T. F. TUAN

very low-frequency hiss observed at Byrd Station, Antarctica. Nature (London) 187, 751-753. Miller, D. R., and Swanson, G. E. (1960). “Inner Conflict and Defense.” Holt, New York. *Moody, J. P., with W. de Groot van Embden, “Arctic Doctor,” pp. 47-48. Dodd, Mead, New York, 1955. Miihleisen, R. (1969). Zusammenhang zwischen luftelektrischen Parametern und Sonnenaktivitiit bzw. Nordlichtern. Kleinheubacher Ber. 15, 129-133. *Newcomb, S. (1911). Article on aurora. I n “The Americana Encyclopedia.” Obayashi, T., and Hakura, Y. (1960).Enhanced ionization in the polar ionosphere caused by solar corpuscular emissions. Rep. Ionos. Res. Jap. 14, 1-40. *Oddie, B. C. V. (1933). Some observations of aurora made a t the Alten Copper Mines, Lapland, in the Years 1839-1849. Quart.J. Roy. Meteorol. SOC.59, 399-400. *Olmsted, D. (1856). On the recent secular period of the aurora borealis. Smithson. Contrib. Knowl. 8, Art. 111, 1-52 (see pp. 7 and 30). Olson, D. (1971). The evidence for auroral effects on atmospheric electricity. Pure Appl. Geophys. 84, 118-138. *Oxaal, J. (1914). Is there an auroral sound? Mon. Weather Rev. 42, 27-29 [translated from Naturen (1913)l. Pakiam, J. E., and Johnson, D. W. (1967). The effects of particle radiation from a disturbed sun on atmospheric electricity. Can. J . Phy8. 46, 1337-1351. Parten, M. (1950). “ Surveys, Polls, and Samples: Practical Procedures.” Harper, New York (see, especially, Chapter XI). *Pell, W. J. (1912). The audibility of the aurora. Pop. Astron. 20, 54-55. *Petrie, W. (1963). “Keoeeit, The Story of the Aurora Borealis,” pp. 89-91. Pergamon, Oxford. Pierce, E. T., and Wormell, T. W. (1953). Field changes due to lightning discharges. In “Thunderstorm Electricity,” pp. 251-275. Univ. of Chicago Press, Chicago, Illinois. Pokorny, A., and Mefferd, R . B., Jr. (1966). Geomagnetic fluctuations and disturbed behavior. J. Ne w. Ment. Dia. 148, 140-151. Potter, W. E. (1970). Rocket measurement of auroral electric and magnetic fields. J. aeophys. Res. 76, 5415-5431. Prescott, G. B. et al. (1860). The great auroral exhibition of August 28th, September 4th, 1859. Amer. J. Sci. 79, 92-97. Presman, A. S. (1970). “Electromagnetic Fields and Life” (F. L. Sinclair and F. A. Brown, Jr., translators). Plenum, New York. Procunier, R. W. (1971). Observations of acoustic aurora in the 1-16 Hz range. Geophys. J . (U.B.) 26 (1/4), 183-189. Procunier, R. W., and Sharp, G. W. (1971). The optimum frequency for detection of acoustic sources in the upper atmosphere. J . Acouat. Soo. Amer. 49, 622-626. *Quetelet, A. (1835). Notice of the meetings of the British Aasociation for the Advancement of Science, in 1833, at Cambridge, and in 1834, at Edinburgh. Amer. J. Sci. 28, 55-84 (see p. 58). The report in the official minutes of the Association mentions only a public meeting devoted to the aurora and does not state what was discussed or who participated in the discussion. *Rae, J. (1881). Sound of the aurora. Nature (London) 28, 605. Rank, 0. (1927). “ Urundzuge einer genetischen Psychologie.” Deuticke, Vienna [see, especially, Chapter 3 of Part 1; an English translation of this chapter can be found in J . Otto Rank Asan. 5, 29-61 (1970)l. *Ray, D. J. (1958). Legends of the northern lights. A Z m b Sportsmn $34, No. 4, 20-21.

AURORAL AUDIBILITY

265

*Rendall, R. (1887). Noises accompanying aurorae. Observatory 10, 303. *Richardson, J. (1823). Excerpt. Ann. Phil. [2] 6, 61. Romig, M. F., and Lamar, D. L. (1963). “Anomalous Sounds and Electromagnetic Effects Associated with Fireball Entry,” Memo. RM-3724-ARPA. Rand Corp. *R.O.S. (1881). Tacitus on the aurora. Nature (London) 23, 484. *Rouse, M. L. (1881a). Tacitus on the aurora. Nature (London) 23, 459. *Rouse, M. L. (1881b). Sound of the aurora. Nature (London) 23. 556. Sargent, W. D. (1870). Influence of the aurora on the telegraph. Smithson. Inst., Annu. Rep. pp. 430-431. (Publ. 1872). *Sexton, S. (1885). The value of the testimony to the aurora-sound. Nature 32 626-626. *Shew, J. (1881). Sound of the aurora. Nature (London) 23, 484. Shipley, J. F., and Barnes, P. E. (1940). Injuries caused by lightning. Quurt. J . Roy. Meteorol. SOC. 66, 389-394. *Silliman, B. D. (1828). Notice of the late aurora borealis. Arner. J. Sci. 14, 91-98. Silverman, S. M. (1970). Night airglow phenomenology. Space Sci. Rev. 11, 341-379 (see, especially, pp. 360ff). Simpson, G. C. (1918). Auroral observations in the Antarctic. Nature (London)102,24-25 (followed by comment by C. Chree). *Simpson, G . C. (1933). Low auroras. Quart. J. Roy. Meteorol. SOC.59, 185-190. Sommer, H. C., and von Gierke, H. E. (1964). Hearing sensations in electric fields. Aerosp. Med. 35, 834-839. Stargis, C. G. (1966). Rayleigh scattering in the upper atmosphere. J . Atmos. Terr. Phys. 28, 273-284. *Stevenson, W. (1853). Abstract of observation on the Aurora, Cirri, etc. made at Dunse. Phil. Mag. [a] 6, 20-46. *Stermer, C. (1927). Comment on Jelstrup report. Nature (London) 110, 45-46. *Stermer, C. (1938). Photographic measurements of the great aurora of January 25-26, 1938. Nature (London) 141, 955-957. *Stermer, C. (1955). “The Polar Aurora.” Oxford Univ. Press (Clarendon), London and New York (see, particularly, pp. 10-11, and 137-139). Stringer, W. J., and Belon, A. (1967a). The statistical auroral zone during IQSY and its relationship to magnetic activity. J . Ueophys. Res. 72, 245-250. Stringer, W. J., and Belon, A. E. (1967b). The morphology of the IQSY auroral oval. J. Ueophys. Res. 72, 4415-4421. *Stumbles, H. E . (1938). An account of auroral phenomena observed in western Canada. J . Roy. Astron. SOC.Can. 32, 451-453. *Sverdrup, H. U.(1931). Audibility of the aurora polaris. Nature (London)128, 457. Tousey, R., and Koomen, M. J. (1953).The visibility of stars and planets during twilight. J . Opt. SOC. Amer. 43, 177-183. *Tromholt, S. (1884). On the aurora borealis in Iceland. Nature (London) 20, 537-538. Tromholt, S. (1885a). A yearly and daily period in telegraphic perturbations. Nature ( L o d o n ) 82, 88-89. *Tromholt, S. (1885b). Norwegian testimony to the aurora sounds. Nature (London) 82, 499-500. *Tromholt, S . (1885c), “Under the Rays of the Aurora Borealis,” Vol. 1, pp. 284-285. Houghton, Mifflin, Boston. van de Kamp, P. (1953). The twenty brightest stars. Publ. Astron. Soc. P m . 85, 30-31. van der Schueren, A., and Koenigsfeld, L. (1963). Electricit6 atmosph6rique a la base roi Bandoin. Inst. Roy. Mdtdorol. Belg. Publ., Ser. A NO. 40. Watson, J. (1894). “The Annals of a Quiet Valley.” J. M. Dent, London.

266

9. M. SILVERMAN AND T. F. TUAN

*Webber, H. J. (1887). Noises accompanying aurorae. Observatory 10, 161-162. Wescott, E. M., Stolarik, J. D., and Heppner, J. P. (1969). Electric fields in the vicinity of auroral forms from motions of barium vapor releases. J . &0phy8. Re8. 74, 34693487.

Wever, E. G. (1949). “Theory of Hearing.” Wiley, New York. Wilson, C. R. (1967). Infrasonic pressure waves from the aurora: A shock wave model. Nature (London) 816, 131-133. Wilson, C. R. (1969). Auroral infrasonic waves. J. Qeophys. Res. 74, 1812-1836. Wilson C. T. R. (1921). Investigations on lightning discharges and on the electric field of thunderstorms. Phil. Tram. Roy. SOC.London, Ser. A m1, 73-115. *Wordsworth, W. (1932). “ The Complete Poetical Works.” Houghton, Boston, Massachusetts (see, partioularly, p. 84). Wormell, T. W. (1940). The effects of thunderstorms and lightning discharges on the earth’s electric field. Phil. Trans. Roy. SOC.London, Ser. A 888, 249-303. Yeowart, N. S., Bryan, M.E., and Tempest, W. (1967). The monaural M. A. P. threshold of hearing at frequencies from 1.5 to 100 c/s. J . Sound Vib. 6, 335-342. Zanotti, E. (1739-1740). Description of an aurora borealis observed at the Observatory of the Institute of Bononia, the night of the 5/16 of December 1737. Phil. Tran8. ROY.SOC.London 41, 593-601.

ON VOLCANIC AND OTHER PARTICULATE TU RBI DlTY ANOMALIES' D. Deirmendjian The Rand Corporation, Santa Monica, California

.................................... ..........................................

entof1883 2.1. The Optical Phenomena .......................................... 2.2. Interpretation of Optical Phenomena ..............................

.................... .................... ........................

....................

...............................

Page 267 268 268 27 1 274 279 290 292 295

1. INTRODUCTION In trying to evaluate the effects of global atmospheric turbidity on weather and climate, it would appear useful to review critically the available information on anomalous, long-lived charlges of turbidity over large portions of the earth in order to estimate the amount and nature of the turbid components in each case. This information in turn may be correlated with the nature and mag?itude of the climatic effects, if any, in an effort to understand the responsible mechanism so that eventually one may predict future effects with some confidence. One type of large turbidity anomaly, unmistakably identified as such, is that produced by recorded extraordinary volcanic explosions capable of injecting massive quantities of so-called volcanic dust into the atmosphere and thereby of altering considerably its normal optical properties. Ideally one should monitor such optical parameters as the magnitude and the wavelength dependence of the scattering and absorption coefficients of a sample of turbid air, as well as the angular distribution of the intensity and polarization parameters of the scattered light before, during, and after volcanic dust periodti. Such a complete set of data, when analyzed in terms of an adequate scattering theory, would yield the desired information on the nature and amount of volcanic dust. As we shall see, these requirements were not completely met during the 'Based on work sponsored by the Advanced Research Projects Agency. Dedicated to the memory of Zdenek Sekera (1906-1973). 267

268

D. DEIRMENDJIAN

most recent onsets of volcanic dust. Consequently, possible climatic effects observed during the presence of such dust cannot be attributed to the anomalous turbidity with any degree of confidence. Specific examples of this situation are three volcanic events that occurred within the last one hundred years and produced widespread turbidity anomalies, viz., Krakatoa (1883), Katmai (1912),and Agung (1963).The first was the subject of a very detailed record and analysis of all reported geophysical phenomena, published only five years after the main event (Symons, 1888).Among these, the pertinent optical phenomena were recorded by means of qualitative and subjective visual observations, and their interpretation was based on an incomplete understanding of the scattering process. The second event, Katmai in Alaska, was not the subject of special study. However, some quantitative evidence of anomalous turbidity is available in the records of the Smithsonian Astrophysical Observatory, which was monitoring the solar constant on a routine basis at that time. The most recent event, the 1963 eruption of Agung volcano in Bali, did not receive the scientific attention it merited despite the wide spread and persistence of its dust. For example, no national or international effort was made t o organize systematic observations and measurements from strategic locations around the globe of the anomalies in atmospheric spectral transmission, and in sky brightness and polarization that followed this well-recorded event for a t least two years. The main purpose of this article is to summarize the findings of a recent extensive critical review of the pertinent literature (Deirmendjian, 1971) and to analyze the available data in terms of valid scattering criteria in order to estimate the amount and type of turbidity in each case. 2. THE KRAKATOA EVENTOF 1883

2.1. The Optical Phenomena By far the most comprehensive study, up to that time, of any single volcanic event was undertaken by the Krakatoa Committee of the Royal Society (Symons, 1888).This document, composed with outstanding clarity and style, still makes instructive reading even in the light of up-to-date knowledge (Deirmendjian, 1971).The paroxysmal explosion of Krakatoa volcano, on the island of the same name, located in the Sunda Strait a t about 6'05'5 105'30'E, occurred over the 26th and 27th August 1883. This is the explosion mainly credited with the injection of the large mass of particulates into the lower stratosphere, thus producing the optical phenomena later observed worldwide. These skylight phenomena, discussed in Part IV of the Report (Symons, 1888, pp. 151-426), are the main concern here.

VOLCANIC TURBIDITY ANOMALIES

269

The optical phenomena related to Krakatoa may be classified into three significant types: blue or green color of the sun or moon, Bishop’s rings, and unusual twilights. These phenomena did not necessarily appear simultaneously or seriatim a t one spot, nor did they follow any particular order in time or space; but all three were undoubtedly connected with the Krakatoa dust.

2.1.1. Blue or areen Sun and Moon. The phenomenon was visual and subjective but seen by many reliable observers. Undoubtedly on such occasions the broad spectral continuum of the directly transmitted sunlight differed from that under ‘(normal” conditions, giving the visual impression of a greenish or bluish hue; but data on the spectral transmission ofthe atmosphere were not available. The few solar spectra mentioned are of little help and difficult to interpret as they are not properly calibrated as to wavelength and absolute intensities. The hoped for discovery of some exotic volcanic gas responsible for the phenomenon did not materialize and it was agreed that the general “ cutoff” observed toward the red portion of the spectrum must be attributed to solid particulate material. The phenomenon was seen mostly in the tropical zone around the equator during the first few weeks immediately following the main eruption, and only rarely and with less reliability outside the tropics. Also the blue coloration was observed mostly when the sun reached an elevation of a t least 10” or so above the horizon and even near culmination; whereas, at low elevation near sunset or sunrise, the disk’s color tended to be green, “yellowish-green,” or “ yellowish-white.” When the sun set green, the rising moon was also greenish and so were bright stars and planets near the horizon. Other significant circumstances mentioned are that blue and green suns were observed together with unusually red twilight skies; that a large sunspot was seen by naked eye on the green sun just before sunset; and that the fully eclipsed moon was observed t o lack the copper tint usually produced by earthshine. It is noted that the blue and green sun phenomenon is not uniquely attributable to volcanic dust, since it has been observed also through other aerosol layers such as Sahara dust; conversely, not all volcanic dusts have produced blue and green sun effects. 2.1.2. Bishop’s Ring. This is one of the more significant corona phenomena associated with the Krakatoa dust, named after the Rev. S. E. Bishop, of Honolulu, who first observed it just ten days after the Krakatoa explosion. It was subsequently reported by several observers from various geographical locations for a t least two and a half years. The following significant features may be noted.

270

D. DEIRMENDJIAN

The Bishop’s ring phenomenon appears to have been a n aureole-corona complex within a circular region around the sun (or moon) of a visual radius 20” to 30”. Specifically it consisted of a system of more or less colored coronas such that a reddish outer ring of an average measured radius 22” t o 23” surrounded a bluish white inner ring whose border had an average radius of about 10.5”. The “ inner radius ” of Bishop’s ring is necessarily imprecise, depending on the definition of the “inner border” mentioned in the Report. However, the phenomenon clearly differed from the usual 22” halo produced by hexagonal ice crystals; rather it conformed to the ordinary coronas around the sun and moon produced by thin water-droplet clouds (there was even a report of the appearance of a common halo with sundogs concurrently with a Bishop’s ring). Furthermore, Bishop’s rings were definitely connected with matter in the higher atmosphere, since they appeared to be more brilliant when observed from high mountains in otherwise clear air, away from large cities. It was also noted that no change in the overall size of Bishop’s ring was observed during the first 12 months of its appearance, but it was most prominent about 8 months after the eruption, and there was an apparent eccentricity of the sun toward the horizon when the ring was observed near sunset. In general, Bishop’s ring appears to have been a phenomenon unique in the annals of optical meteorology for size, brilliancy, universality, and protracted duration.

2.1.3. Unusual Twilights and Crepuscular Phenomena. The descriptions of this aspect of the post-Krakatoa skylight effects, although among the most widespread and noticed, are perforce neither as explicit nor as uniform as those of the blue sun and Bishop’s ring. This is owing to the close dependence of twilight features on the optical characteristics of the local and trans-horizon troposphere, and their variation not only along the vertical but along the horizontal direction as well, and to the highly subjective nature of individual impressions of twilight colors, relative brightness, and their changes with time. The post-Krakatoa twilights were distinguished by their unusually long duration and vivid coloration, said to be markedly different from “ normal ” twilights; the presence of an unusual “ diffuse illumination ” of the whole sky so that the earth shadow could not be distinguished; and the existence of a secondary twilight glow clearly separated from and a t a higher elevation than the primary or “ordinary” glow usually seen above the sunset horizon on clear days. This secondary glow could be distinguished by a characteristic delicate purple hue rather difficult to describe verbally. (Anyone who has not witnessed the post-Agung twilights of 1963-1966 will be quite unable to visualize the delicate coloration that these observers were trying to describe.) Other phenomena observed in the cloudless daytime sky included unusual

VOLCANIC TURBIDITY ANOMALIES

271

skylight colors away from the sun, a considerable diminution in the polarization of skylight, and the appearance of several new neutral (zero-polarization) points outside the sun’s vertical plane. All these phenomena (except those of polarization) were observed by numerous people a t various widely separated locations and for long periods of time after the main eruption. The veracity of the descriptions must not be underrated, especially when made by sailing-ship officers who were keen observers of atmospheric optical phenomena for their weather-forecasting value. The bulk of the evidence presented in the Krakatoa Report supports the assumption that most of these phenomena were definitely connected with the volcanic dust. In fact, the systematic logging of the progressive appearance of the optical phenomena over different parts of the globe provided the first evidence of a high-level equatorial ring of strong winds that distributed the volcanic matter a t such a fast rate (the Report estimates that about 26 days after the paroxysmal eruption of 26 August 1883, the dust had made two complete circuits around the globe traveling from East to West). The discovery of these “ Krakatoa winds ” near the 30-km level is one of the main contributions of the Report.

2.2. Interpretation of the Optical Phenomena

It is clear that any interpretation of the post-Krakatoa optical effects to deduce the nature of the responsible volcanic dust cannot be very specific and quantitative. We outline here some of our own conclusions which do not necessarily reflect those reached in the Krakatoa Report (Symons, 1888). First, one must recognize that (a) the Krakatoa dust particles cannot have been strictly monodisperse and uniform in composition and shape, otherwise the related phenomena should have been much more spectacular than reported; and (b) the chemical composition of the particles, after a considerable residence in the lower stratosphere, may have changed from that of the original volcanic material, as a result of photochemical or other local reactions. As to the specific phenomena, we have the following: 2.2.1. Blue Sun. The simplest hypothesis is that the Krakatoa dust’s spectral extinction is responsible for the reported visual color impressions of the sun’s image. By extinction we mean the depletion produced by primary scattering and absorption within the dust particles. In principle, from the “ observed ” atmospheric spectral transmission, one may derive the extinction optical thickness T ( A ) of the volcanic dust-laden atmosphere as a function of wavelength in the visual range. From this, in turn, one may deduce the volcanic dust component, rD(A),by subtracting the parts contributed by the molecular or Rayleigh atmosphere T ~A),( and an appropriate background

272

D. D E J R M E N D W

aerosol component T&). It is then a question of modeling for optical constants (i.e., composition), size range, and size distribution of a dust cloud such that its optical thickness closely approximates ~ ~ (in hmagnitude ) and spectral variation. The trouble is that visual color impressions of the sun, such as those reported in relation to the Krakatoa event, are rather unreliable indicators of true spectral radiance, especially if recorded against an abnormal skylight background. Similarly, in the case of the Moon and Venus, one has to consider subjective impressionsrecorded against the memory of the normal appearance of these bodies. Nevertheless, i t is clear that the Krakatoa dust must have depleted the red end of the solar spectrum by an amount greater than the normal atmospheric aerosols. Taking into account all the recorded optical phenomena, it should be possible to make an educated conjecture on the general characteristics of the Krakatoa dust and to set up a synthetic model involving particulates of some hypothetical composition (or optical properties), not necessarily corresponding to any known substance but in fair agreement with observations as to extinction and other scattering properties. The analysis must be limited to a consideration of idealized, homogeneous, spherical particles-the only type whose scattering and absorption properties are well known, either for individual particles (van de Hulst, 1957), or in polydisperse aggregates (Deirmendjian, 1969a). This situation is not as bad as it might appear, however, because it is known that in the case of polydispersions of relatively small irregular particles, and for such properties as forward scattering and extinction cross section, the shape factor is not important provided the particle's volume and optical constants are specified (Holland and Gagne, 1970). Let the optical thickness of the Krakatoa dust be such that, when combined with that of the normal atmosphere, it should yield an overall value that is practically independent of wavelength in the visual range. This may indeed produce a blue green visual impression of the sun when observed through a sufficient amount of volcanic dust as the dusty atmosphere will merely act as a neutral filter on the extraterrestrial spectral radiance of the sun. (The reported reddening of the skylight would further enhance the color contrast between sky and sun contributing to the visual impression.) The value of T around h0.5 pm must have been several times the normal value to enable the reported naked eye observation of a sunspot. The conditions must have been similar to those for the " filtered sun " effect (Deirmendjian, 1969a,b), equivalent to a slant-path transmission of lo-' or about e-16. Assuming that the sun was about 3" above the horizon when the sunspot was observed, and that the air mass may be approximated by sec ( = zenith distance of the sun), yields

c0 co

16 21 .t sec 87"

VOLCANIC TURBIDITY ANOMALIES

273

or an optical thickness of 0.84 a t X0.5 pm. This is a t least four times a typical value of the normal optical thickness likely to be found a t X0.5 pm in clear air away from urban pollution. Thus, the Krakatoa dust contribution to the total optical thickness a t this wavelength may be estimated as (2.1)

~ ~ ( X 0pm) . 5 = 9 0.84 = 0.63.

This is not really excessive considering that the initial density of the dust within the volcano's latitude zone, related with sightings of the blue sun, must have been high compared to that of the long-lived dust responsible for the more widespread phenomena. The '' greening " of the blue sun when very near the horizon is not difficult to explain qualitatively in terms of the earth's sphericity and the high level of the volcanic-dust layer (the Chapman effect), as proposed in the Krakatoa Report. However, in the absence of actual measurements of spectral transmission of the setting blue and green suns such a conjecture cannot be substantiated.

2.2.2. Bishop's Ring. Accurate data on the angular size and order of colors for the various rings, as well as on the brightness gradient in the inner aureole region, would have been most useful in estimating the size and size distribution of the dust particles. Assuming monodisperse of quasimonodisperse particles, the Krakatoa Report estimated their radius to be about 0.8 pm, on the basis of Jirst-order diffraction theory for opaque disks. For spherical particles (rather than thin disks), this estimate would have to be revised downwards by about 25 percent (van de Hulst, 1957; Deirmendjian, 1969a), yielding about 0.6 pm for the dominant radius of such idealized Krakatoa particles. In fact, these particles could neither have been strictly monodisperse nor conformed to the wide size-distribution law followed by the aerosols responsible for the normal '' white and featureless aureole. Furthermore, the modal (or predominant) size in the Krakatoa particles must have been different from those in the common shallow tropospheric clouds which produce coronas too small in radius (1" for the inner to 12" for the outer border) to be considered Bishop's rings, and which involve particles too large (2 pm to 4 pm modal radius) for long stratospheric residence (Deirmendjian, 1969s). We must therefore assume that the Krakatoa dust in its initial stages conformed to an unusual type of relatively narrow size-distribution law with a modal radius near 0.6 pm, which, when combined with a tropospheric type of aerosol distribution, produced the Bishop's ring phenomenon as reported. ))

2.2.3. Unusual Twilight Phenomena. Because these phenomena are greatly affected by secondary mechanisms (e.g., second- and higher-order scattering and indirect illumination), and they depend on a number of

274

D. DEIRMENDJIAN

parameters (e.g., local extinction coefficient as well as directional scattering) and their variation both in the vertical and in the horizontal direction, the post-Krakatoa twilights cannot, in general, be interpreted in a straightforward manner. Similar remarks apply to the considerable anomalies in skylight polarization phenomena observed in the presence of volcanic dust (Sekera, 1957).As pointed out by Rozenberg (1963),even if accurate quantitative dat,a were available, the information deduced from their analysis could hardly be unique or unambiguous. The main significance of these unusual twilights lies in their indication of the great height of the responsible dust layer, inferred from their long duration and the presence of secondary glows; in their unusually high brightness, indicating that even if the dust were composed of light-absorbing material, it nevertheless was characterized by a high albedo of single scattering; and in their visibility from locations distributed over a wide latitude zone (where the other optical phenomena were rarely, if ever, observed), indicating the wide spread of the volcanic dust over higher latitudes within both hemispheres. The initial mass of material contained in the airborne dust was estimated t o be equivalent to some 4 km3 of " solid matter expelled from the volcano " out of a total of 18 km3 for all the ejecta, with a possibility that this might be 4 or 5 times larger under certain assumptions (Symons, 1888, pp. 440 and 448). Interestingly it was also suggested that the main body of the stratospheric dust might have been composed of" condensed gaseous products of the eruption (other than water) such as sulphurous acid or hydrochloric acid'' (Symons, 1888,p. 445).I n view of present ideas about the origin of stratospheric aerosols (of. Junge et al., 1961) this conjecture may have been very close to the truth. As for other meteorological effects related to the Krakatoa event such as unusual deviations from the norm of surface temperature, pressure, winds, cloudiness, precipitation, etc., there is very little information to be found in the Krakatoa Report. 3. THE KATMAI EVENT OF 1912 On 6 June 1912,a major convulsion apparently culminated in the blowingoff of the top of Mt. Katmai, on the Alaska Peninsula.near the Shelikoff 155"W. Although perhaps comparable in intensity Strait, a t about 58'16" to Krakatoa, the Katmai event was not the subject of a systematic collection and analysis of observations of the directly related geophysical phenomena. Captain K. W. Perry, of the U.S. Revenue Cutter Manning reports that, while bunkering a t Kodiak, about 100 miles from the volcano, on the day of the explosion there was a copious ash-fall on the town and a reduction of

VOLCANIC TURBIDITY ANOMALIES

275

visibility to practically zero for two days. When some visibility returned after 40 hours, “ t h e skies (at 2:30 pm) assumed a reddish color” (Anonymous,

1912). George C. Martin, a geologist with the U.S. Geological Survey, visited the site just four weeks after the eruption (Martin, 1913) and collected some eyewitness descriptions. We learn that the mail steamer Dora, which was actually steaming up the Shelikoff Strait on the evening of 6 June, was unable t o make Kodiak harbor when visibility dropped to zero two hours before sunset, forcing him to run out to the open sea for 12 hours. The Bertha reported seeing a red sun in “ clear skies ” about noon on 7 June when steaming about 375 miles northeast of the volcano. I n comparing the Krakatoa and Katmai volcanic events, Martin (1913) conjectures that Krakatoa’s must have been by far the greater of the two because its explosions were heard from as far as 3000 miles whereas Katmai’s were not recorded farther than 750 miles. (This of course is not a valid criterion, in view of differences in atmospheric structure between equatorial and subarctic regions.) However, he also believes that the two events were about equal in the quantity of material ejected. I n the reports of the later, semiscientific expeditions organized by the National Geographic Society (cf. Griggs, 1917, 1918, 1921) there is little additional information about the amount and nature of the material introduced by Katmai into the atmosphere. However, from estimates of the shape and height of Mt. Katmai before the eruption and later surveys of the actual crater and caldera, Griggs arrives a t an estimate of the total volume of ejecta of about 8 km3. (We note, however, that according to Gruening (1963) the total volume of the Katmai ejecta may have been of the order of 29 km3 with some material initially injected as high as 40 km into the stratosphere. This represents an uncertainty factor similar to that for Krakatoa’s ejecta mentioned in the previous section.) Thus, assuming that the two volcanoes introduced the same type of dust into the atmosphere, and roughly by a similar mechanism, other things remaining equal, one may conclude that Katmai’s contribution to long-lived atmospheric particulates must have been just under one half of Krakatoa’s. This circumstance, together with the high-latitude location of Katmai within the prevailing westerlies, must explain the limited spatial and temporal extent of its atmospheric optical effects. Interestingly, the effects of the Katmai dust on atmospheric turbidity were first reported from the state of Virginia, 3700 miles away, then from Bassour, Algeria, a t 6000 miles, and finally from Mt. Wilson, California, only 2500 miles from Ketmai (Abbot, 1913). The magnitude of the Katmai turbidity anomaly may be derived from the records of the then-operating Smithsonian Astrophysical Observatory. Those covering the Mt. Wilson data for the period following Katmai’s eruption

276

D. DEIRMENDJIAN

(Abbot et at., 1922) clearly show a general increase in turbidity that can very well be related to the volcanic dust. One may determine the total optical thickness from the normal transmissivities listed in the tables, then subtract the permanent Rayleigh component and, in certain spectral regions, the estimated absorption due to ozone (Deirmendjian, 1955a). The residual optical thickness thus obtained may be attributed to turbid components such as natural and artificial aerosols in the form of particles larger than molecules and atoms. I n particular, setting where q(h) is the recorded narrowband transmissivity and ~ ( his) considered the “ observed ” optical thickness, one obtains

+

TM(x) T D ( x ) = d X) 7R(h) -T03(h) (3.2) where ~ ~ (is hthe) standard atmosphere Rayleigh component for the height of the particular station (Deirmendjian, 1955b); and T ~ ~ (isA the ) component due to ozone absorption. Evidently, the residual optical thickness, given by the left-hand side of Eq. (3.2) as the sum of two terms, can be a highly variable quantity, as it depends on many factors a t any given locality including the amount of ozone above the station. Unless there are optical-climatological records, properly analyzed and classified, it is difficult to know the normal background turbidity T&) needed t o evaluate the anomalous component T&) attributable to volcanic dust or other unusual sources. We shall try to estimate the Katmai h ) presumably there was dust component by using likely values for ~ ~ (when no volcanic dust over the Southern California coast. I n Fig. 1, the straight line marked T~ (1737 m) indicates the Rayleigh optical thickness corresponding to the station elevation. The solid curve ) by marked (a) represents the residual optical thickness ( T - T ~ obtained combining the June 1912 (23 pre-Katmai-dust days) average transmission, recorded at Mt. Wilson, with that given for October 1913 (25 post-Katmaidust days). Both these months were characterized by exceptionally high recorded transmission values. Their average was used as a norm for clearest conditions a t Mt. Wilson in those years. The data points correspond to the seven standard wavelengths 0.35,0.40,0.45,0.50,0.60,0.70,0.80, and 1.O pm given in the Smithsonian’s tables. Curve (b) corresponds to similar data but for 3 August 1913, a day with moderately low turbidity over Mt. Wilson. The dashed straight-line portions joining the data points a t h0.45 and 0.80 pm on curves (a) and (b) roughly indicate the effect of subtracting the ozone contribution in the Chappuis continuum, which has a maximum absorption near h0.6 pm.Disregarding the values for h 5 0.35 pm, where the Smithsonian data could have been unreliable (Dunkelman and Scolnik, 1959),the single

277

I VOLCANIC TURBIDITY ANOMALIES 100

I

7

I

I

I

I

l

l

I

I

I

I

KATMAI

-

I

0.3

0.4

0.5 0.6 0.7 0.80.9 1.0

l

l

1.25 4.5 1.752.0 2.25

x FIG. 1. The turbidity anomaly produced by the Katmai (1912) eruption, evaluated from contemporary data of the Smithsonian Astrophysical Observatory obtained at Mt. Wilson, California. See text for explanation of curves.

clear-day data represented by curve (b) and the 48-day average given in curve (a) are remarkably similar in shape and slope except for a factor that is almost independent of wavelength. This suggests that the nature and size distribution of ‘‘ normal,” clear-day aerosols over Mt. Wilson was constant and only the total concentration varied from day to day during the period in question. We could consider curves (a) and (b) in Fig. 1 as typifying the norm for very low turbidity over Mt. Wilson, and use them to evaluate turbidity anomalies during abnormal conditions. Alternately we could obtain the anomalous component T&) by subtracting the total observed optical thickness for clear-day conditions similar to those represented by curves (a) and (b), from that for abnormally turbid days. I n the process, corrections for the Rayleigh component, ozone absorption, and for possible instrumental and other errors, would automatically be taken into account. This procedure was used to obtain the rest of the solid curves in Fig. 1, labeled (c) to (f), from data (Abbot et al., 1922) for days when Katmai dust was presumably present over the site, as follows. Curve (c): Mean optical thickness for 27 days in August 1912 minus that for 23 days in June 1912.

278

D . DEIRMENDJIAN

Curve (d): Mean for 17 and 19 August 1912 minus that used in curve (a). Curve (e): Value for 2 August 1912 minus 19 days in August 1913 with low turbidity. Curve (f): Value for 2 August 1912 minus that used in curve (a). An examination of the monthly mean transmissions for 1912 (Abbot et al., 1922) reveals that the local turbidity was a t a minimum in June but rose sharply during July and August, probably as a result of the Katmai dust, reaching a maximum sometime during September or October (for which two months no average values are given). Our above choice of August days as representative of Katmai dust conditions was governed by this circumstance as well as by the high quality attributed to the raw data by the authors of the Annals. Thus the August 2 data, reduced as in curves (e) and (f ), may be associated with the highest Katmai dust content; the mean for 17 and 19 August, as in curve (d), amounts to about two thirds of this maximum; and curve (c), the mean for all 27 days of August 1912 on which observations were made, though less than one half the maximum, still shows a value almost four times the clear-day turbidity as in curve (a). This gradual decrease in turbidity anomaly as more August days are considered suggests that the Katmai dust layer arrived over Southern California in waves, as might be expected. The family of four curves labeled (c), (d), (e), ( f ) in Fig. 1 can be bracketed very nicely by curves based on one of our own theoretical aerosol models, devised to represent light tropospheric turbidity rather than any volcanic effects. In Fig. 1 the two continuous dash-dot curves labeled “2.5 k m ” and “ 9 km water haze L,” respectively, show the effect of corresponding thicknesses of homogeneous aerosol layer for this model (Deirmendjian, 1969a, pp. 77-79, and Tables T.16 to T.19). These two curves thus represent the optical thickness produced by totals of 2.5 x 10‘ and 9 x lo7 particles per square centimeter column, respectively, following a haze-L size-distribution law, n ( r ) ,given by (3.3) n ( r )oc r2 exp{-15.12&}. This model has a mode radius r, of 0.07 pm and a volume mixing ratio Y , of 1.167 x lo-” per 100 particles, equivalent to a volume of 2.92 x and 1.05 ?c: cm3 per square centimeter column, respectively, for the lower and upper bracketing curves in Fig. 1. Using the upper estimate and assuming such a dust veil spread above the entire earth’s surface or an area of 6.1 x 1018 cm2, the total volume of the dust would be 5.36 x 1013 cm3 or 5.36 x 10-2 km3. Assuming that the Katmai dust spread only over the spherical cap above 30”N latitude or one quarter of the earth’s surface, the same model would km3 of solid material. This may be a reasonable assumprequire 1.34 x tion for the maximum initial spread of Katmai’s dust, and it represents only 11600 of Griggs’ estimate of 8 km3 for all ejecta in the 1912 eruption.

VOLCANIC TURBIDITY ANOMALIES

279

To test this, if we assume that the same fraction of Krakatoa’s total ejecta, km3 of material) was initially spread as estimated a t 18 km3 (i.e., 3 x volcanic dust over the intertropical zone between 23’5 North and South or 0.4 of the earth’s total surface, one gets about 1.47 x om3 of particles per square centimeter column. On the basis of the same model as for the Katmai dust, this is equivalent to 12.6 km haze L, which, in the case of a water-like substance, would result in a turbidity anomaly of rD = 0.60 a t h0.45 pm. This agrees rather well with the estimate of rD = 0.63 a t h0.50 pm for Krakatoa’s initial turbidity (see Section 2.2.1) by an entirely different reasoning. We note in passing that if we accept the Krakatoa Committee’s own estimate of 4 km3 of material injected into the atmosphere as dust, proceeding as above, we arrive a t the equivalent of 1680 km of haze L, yielding a dust veil with the fantastic optical thickness rD = 80 over the entire intertropical zone! The above given estimates of the magnitude and broad spectrum of the volcanic dust turbidity anomaly r D ( h )do not change radically if the duet composition is changed, say, from water to some other nonabsorbing dielectric substance of higher density, such as quartz, provided the size distribution and total number of particles are unchanged. This has been demonstrated by the theoretical models investigated by this author (Deirmendjian, 1969a) and others. Thus the total mass of either the Katmai or the Krakatoa atmospheric dust could hardly have exceeded 10-8 of that of the entire atmosphere (5.14 x loa1 g) or the large portions thereof that were affected, whereas the corresponding total optical thickness was little more than doubled with respect to that found under cloudless and very clear conditions away from urban pollution centers. 4. THE AGUNCI EVENTOF 1963

As mentioned earlier, whatever useful data are available on the Agung turbidity effects were randomly and accidentally obtained, mostly as a byproduct of unrelated programs and facilities, rather than by special design. However, in this case and for the first time, a direct in situ collection and identification of stratospheric particles, undoubtedly connected to the volcanic event, was accomplished (Mossop, 1964). Mount Agung, a dormant volcano on the southeast shore of Bali, at about 8’25’5, 115’30’E, “blew its top”on March 17,1963, resulting in considerable loss of life and damage in its immediate vicinity (Booth et al., 1963). Meinel and Meinel (1963) were among the first t o draw attention to subsequent skylight effects in the northern hemisphere. They not only related the unusual twilights seen from Tucson, Arizona with the incursion of Agung’s dust but also estimated the height of the main stratum as 22 km from simple considerations of the earth’s shadow.

280

D. DEIRMENDJIAN

The unusual duration and delicate coloration of evening twilights were observed also locally (southern California coast), notably during the fall and winter of 1963-1964. Under very clear sky conditions produced by so called " Santa Ana )' weather (dry, off-shore winds) even a secondary glow, similar to the corresponding Krakatoa phenomenon, was often observed by this writer. The asymmetry mentioned by the Meinels (1963)-i.e., the fact that the evening twilight glows, during the above period, were centered some 10" to 20' south of the (subhorizon) sun's azimuth-was also noticeable locally, as well as the nonuniformity of the glow, suggesting patchiness in the dust stratum, mentioned by Volz (1964).The latter, whose observations were made near Stuttgart, appears to be the first author to relate northwestern European twilights with the Agung volcanic dust. The unusual darkness of the totally eclipsed lunar disk, as observed on 30 December 1963 (Brooks, 1964),has also been attributed to the wide spread of Agung's dust. This phenomenon, however, cannot be used as a measure of the increase in overall turbidity (nor for estimates of changes in the planetary albedo) since it can be shown that small changes in stratospheric turbidity will cause large changes in the illumination of the eclipsed moon (Brooks, 1964). As already pointed out, the twilight method is the least reliable one for quantitative estimates of volcanic turbidity anomalies. For this reason we have not considered Volz's (1964) derivation and interpretation of the vertical profile of the atmospheric turbidity for the winter of 1963-1964 in our present derivation of the Agung turbidity anomaly. His deductions were apparently based merely on the analysis of brightness ratios a t a fixed point in the sky as a function of solar depression angle (Volz, 1970a). The underlying theory and the reliability of results thus obtained have been seriously questioned (cf. Rozenberg, 1963). However, one cannot deny the value of this particular technique in the detection of stratospheric aerosol layers (Shah, 1969). At least two other methods do exist, which are rather more reliable, though less sophisticated, than any twilight method, inasmuch as both are based on fewer assumptions and a simpler theory. As already noted, one is the old Smithsonian method of obtaining narrow-band transmissivities using the sun as a source, from which turbidity values for the entire atmospheric layer may be deduced. The other, based on recent technology, uses the pulsed-laser technique and is capable of indicating the vertical profile of the turbidity (Grams and Fiocco, 1967). To our knowledge there exist no atmospheric trammisaion data, similar to those after the Katmai eruption, that could be used in the case of the Agung 1963 event. However, some useful information is provided by occasional stellar extinction data. An attempt will be made to use these to obtain some idea about the nature of the Agung turbidity anomaly.

281

VOLCANIC TURBIDITY ANOMALIES

The most detailed such data seem to be those published by Irvine and Peterson (1970), particularly those from Boyden Observatory located in South Africa a t about 29’15‘5 26’E. These authors list stellar extinction coefficients for various wavelengths, obtained by filters, in “ magnitudes per unit air mass.” To convert these awkward units to those of the more rational optical thickness, one has simply to multiply their listed values by 0.917. Selected 1963 and 1965 data, presumably reflecting turbidity conditions during and after the passage of Agung dust over Boyden Observatory, are ( plotted on Fig. 2 after conversion as above. The straight line marked T ~1350

4 00

I

I

I

I

l

l

1

I

I

I

1

I

1

AGUNG

7

.-,

51

3c

2 ’

-

.” .I-

X, tn-1 t-

\ /

7 ’

5.

40-2

I

I

1

0.3

0.4

I

I

I

l

\

l

0.5 0.6 0.7 0.8 0.9 4.0

I

4.25

1

4.5

1.75 2.02.25

FIQ.2. The turbidity anomaly produced by the Agung (1963) eruption, evaluated on the basis of stellar extinction values obtained from a South African and Australian observatory (see text).

m) represents the Rayleigh optical thickness corresponding t o the Observatory’s height. The other curves were adapted from Irvine and Peterson’s (1970) original data as follows. Curve (a): Average optical thickness for seven “good” nights of May 1965 (consistently low extinction period) less the Rayleigh value. Curve (b): Average for five consecutive “fair” nights in August 1963 less the Rayleigh value. The increased extinction observed during this period

282

D. DEIRMENDJIAN

may be attributed to Agung’s dust, according to the authors.2 The dashed portion of this curve indicates the absence of original data a t h0.634 pm. Curve (c): Observed valucs on night of 10 September 1963, marked “ good ” and apparently showing the highest extinction observed in the blue to green filters, less the Rayleigh value. Curve (d): Obtained by simply subtracting the extinction data corresponding to curve (a) from those for curve (b). The May 1965 values were chosen as a standard for clear nights, assuming that the Agung dust effects had practically disappeared at Boyden by that time (data for only two pre-Agung nights are given by Irvine and Peterson). Thus curve (a) in Fig. 2 may be compared with curves (a) and (b) of Fig. 1, typical of clear days a t Mt. Wilson. The two sets seem to have little in common (there is no reason why they should) except for their magnitude and the ozone absorption feature a t A0.6 pm. However, after processing the Agung data as in curve (d) of Fig. 2, a smoother wavelength dependence is obtained similar to that of the Katmai data. We may reasonably assume that the processing has eliminated the ozone absorption in the 0.50 < h < 0.73 pm interval and that the dashed line in curve (d)joining these two points is a good interpolation for the missing data. (We have no explanation for the dip between hO.40 and 0.46 pm appearing in both curves (b) and (d), and shall assume that it is not related to the volcanic dust component.) A search for an independent set of data of this type on Agung turbidity waa largely unsuccessful. Przybylski’s (1964) description of some stellar extinction measurements from a South Australian site during 1963 are of little help here because there is no wavelength resolution given and the analysis ia not clear. He does mention that the extinction in ‘‘ visual light,” during the August 1963 peak turbidity period, reached a value three times the pre-Agung norm, which is roughly borne out by the ratio between curves (b) and (a) in Fig. 2. Hogg (1963)of the same (Mt. Stromlo) observatory reports the appearance of an unusually large and persistent “ meteorological corona ” around the sun after Agung, which we interpret as an aureole [rather than a Bishop’s ring, as interpreted by Volz (1970b)l in the absence of any mention of colored rings of the proper radius. Hogg also quotes stellar extinctiqn values at five wavelengths obtained variously in June and July 1963, together with a set of “ normal ” values a t the same wavelengths. Curve (e) in Fig. 2 represents the difference between these two sets without further processing except for the conversion from magnitudes to optical thickness. As may be seen, all things ‘The reference to a nonexistent volcmo in northern Australia (Irvine and Peterson, 1970, p. 66) should be corrected to reed “Agung volcano in Indonesia,” moording to a private communication from one of the authors.

283

VOLCANIC TURBIDITY ANOMALIES

considered, the Australian data are remarkably similar to the South African ones, represented in curve (d), both in magnitude and slope. We may now compare the turbidity anomaly curves in Fig. 2 directly with those of Fig. 1. Clearly, the Agung dust extinction curves, though of the same magnitude as those for Katmai, show a steeper slope than the latter, a t least in the available range 0.4 < A < 1.O pm. This indicates that there must have been a smaller proportion of large particles in the Agung dust than in that of Katmai (and most probably of Krakatoa). Thus the best choice for Agung is a “haze H ” model (Deirmendjian, 1969a, p. 78, and Tables T.102 and T.103), which follows a size distribution law of the form n(r)cc r2 exp{-20r}.

(4.1)

As may be seen from Fig. 3, where the logarithm of the size distribution function n(r) (normalized to 100 is plotted as a function of the radius r, the H model has a significantly larger concentration of small particles in I

I

I

I

I

I

FIG.3. Comparison of proposed volcanic dust size-distributionmodels (solid curves) with actual stratospheric sampling of Agung’s dust (dots) and the distribution proposed by Grams and Fiocco (dotted curve).

284

D. DEIRMENDJIAN

the range 0.6 6 r 0.25 pm than the L model; but for r > 0.25 pm, the concentration in the latter model considerably exceeds that in model H. Note that the verification of either model would entirely preclude the formation of any Bishop’s or other corona-type rings around the sun in the presence of such volcanic dust (Deirmendjian, 1969a). The equivalent depths of homogeneous model H haze needed to bracket the Agung rD(h) data are 7 and 14 km, respectively, as shown by the labeled dot-dash curves in Fig. 2. The designation “silicate” merely indicates a dielectric substance with real index of refraction between 1.54 and 1.56 in the visible range. Water substance with the same size distribution of particles would result in a steeper slope than given by the data. (The geometrical thickness cited should not be taken literally but merely serves to indicate the total amount of material needed t o produce the required turbidity anomaly regardless of concentration.) I n this case, taking the upper figure of 14-km haze H and using the published model parameters (Deirmendjian, 1969a, p. 78), we get (14 x 106)(3.142x 10-12) = 4.4 x

cm3(cm2column)-1

for the bulk volume of the dust. (This is less than one half of the comparable figure of 1.05 x cm3(cma column)-l derived for Katmai’s dust.) Assuming that the above amount of Agung dust was distributed over the same intertropical zone as used for Krakatoa, one arrives at a total of 9 x 10l2 om3 of dust material. From this point of view, and comparing this figure with our estimates for the other volcanoes, the Agung eruption of 1963 injected an amount of dust equivalent to 0.70 of that injected by Katmai in 1912 and 0.31 of that injected by Krakatoa. Finally, on the basis of the ratio 1/600 of dust to total, deduced earlier for Katmai and Krakatoa, the total volume of all ejecta from Agung would be 5.4 x 10l6 om3 or 5.4 km3 of material, assuming the same type of volcanic event. It would be interesting to see whether independent volcanological surveys arrive a t such an estimate for Agung. Curve (0) in Fig. 2, which represents the highest non-Rayleigh extinction observed a t Boyden Observatory (Irvine and Peterson, 1970), covers too small a wavelength range for detailed analysis. Interestingly, its magnitude of about ten times the clear-day values in curve (a) is quite close to the value of 0.63 we estimated for Krakatoa’s initial turbidity a t h0.5 pm. However, as neither blue and green suns nor Bishop’s rings were definitely reported (to our knowledge) in the case of Agung, we conclude that the latter’s dust, even in its initial stages, did not contain nearly as many large particles as Krakatoa’s but conformed t o a rather different size-distribution law. Turning to the optical radar method, Grams and Fiocco (1967) have discussed their use of the pulsed laser t o determine stratospheric turbidity during 1964 and 1965. In particular, assuming that the often detected stratospheric

VOLCANIC TURBIDITY ANOMALIES

285

aerosol maximum near 20 km (Junge and Manson, 1961) is a permanent feature between about 60"sand 70"N latitudes, they looked for anomalous increases in the northern hemisphere that might be attributed to Agung's dust. They used the well-known pulsed, ruby laser system (h0.6943pm) with a nominal spatial resolution of 0.015 km reduced to an effective resolution between 0.5 and 1.0 km after the necessary smoothing of the raw data. To minimize the ambiguities inherent in the interpretation of such data (Deirmendjian, 1965, 1969a), Grams and Fiocco (1967) used the ratio of the actual signal to a hypothetical norm based on the return from the region of minimum turbidity between 25 and 30 km, as an index of the turbidity. The average of all the ratios thus obtained on a number of nights during 1964 and 1965 from Lexington, Massachusetts, and College, Alaska, was found to vary between 1.5 and 2.3 a t its maximum, centered near an altitude of 16 km. On the basis of a Junge type size-distribution model, and assuming Mie scattering with refractive index 1.5, these authors then deduce a local extinction coefficient a t 16 km equivalent to 2 x km-l and a particle concentration of 0.9 From this and their average profiles between 12 and 24 km, they arrive a t a total of 6.8 x lo5 particles per square centimeter column in the intervening layer, equivalent to an optical thickness rD = 0.015 presumably at the laser wavelength near h0.7 pm. They also give a mass of 6 x lo-' g per square centimeter column for particles of an assumed density 2, equivalent to a bulk volume of 3 x lo-' om3 per square centimeter column in the units used here. This may be compared with the 2.2 to 4.4 x units estimated above for Agung's dust on the basis of southern hemisphere stellar extinction data. Thus Grams and Fiocco's (1967)own analysis of their laser data from northern latitudes yields about 1/10 of the bulk amount of Agung related aerosols we independently deduced for southern latitudes from a different set of data. This is consistent with their estimate of 0.015 for the optical thickness of the dust layer compared with values exceeding 0.15 near h0.7 pm we deduced as shown in Fig. 2. I n fact, quite likely the laser-derived amount may have been over-estimated, considering the ambiguity inherent in the reduction and analysis of monostatic laser-radar data in terms of the size-distribution model of the form

n(r)cc r-3.5,

0.275

r 5 3.3 pm

used by Grams and Fiocco (1967). The shape of such a function (multiplied by an arbitrary constant) is indicated by the dotted curve in Fig. 3 for comparison with the continuous distribution models we used (solid curves) in our own interpretation of the Katmai and Agung extinction data. The distribution (4.2) excludes the smaller particles, in the range 0.01 < r < 0.275 pm,

286

D. DEIRMENDJIAN

where the concentration of stratospheric particles should be high (Junge and Manson, 1961).On the other hand, the upper size limit in (4.2)seems excessive considering that the extinction coefficient for such a distribution is highly dependent on the range of integration at the upper end (Deirmendjian, 1969a, p. 80). The size range indicated in (4.2) seems to disagree with those accepted for stratospheric aerosols in general and with actual counts of Agung particles (see below and dashed curve in Fig. 3) in particular. The theory of polydisperse Mie scattering has shown that the mass extinction coefficient of a polydispersion increases with increasing proportion of smaller particles in the distribution. Hence the above mentioned laser data could have been interpreted in terms of smaller particles, say within the range 0.10 5 r 5 1.5 pm, in which case Grams and Fiocco's (1967) mass estimates would be reduced by a factor of 2 or 3. This is further reinforced by the fact that, for a given mass, the smaller particles in a polydispersion contribute most to the back-scattering cross section (Deirmendjian, 1969a, p. 90). A t any rate, the laser data suggest that, at least in the northern hemisphere, by the end of 1965 the Agung stratospheric turbidity anomaly was hardly distinguishable from the normal background. [Volz (1970b) seems to arrive a t a similar conclusion on the basis of (in our opinion) less reliable data.] Finally, let us consider Mossop's (1964) actual sampling of stratospheric aerosols from a U-2 aircraft before and after the Agung eruption. At 30 km over Australia, he did find a threefold increase in the concentration, as well as a larger average size in the irregular particles he collected. The size tended to diminish'again with the passage of time. The dashed line in Fig. 3 shows only one of Mossop's distributions (multiplied by an arbitrary constant), as derived from his counts and microscopic measurements, collected between 15" and 35" south latitude on 2 April 1964, one year after the Agung event. Although the size resolution of the data is very gross, the indicated maximum concentration around 0.1 pm and slope in the 0.3 5 r 5 0.5 pm region are rather well reproduced by our model H distribution. We have seen that this model provides a fairly good fit to the stellar extinction data obtained in August 1963 a t Boyden Observatory (curve d of Fig. 2). If we accept Mossop's (1964) deductions regarding the time variation of the dominant size and size range of Agung particles (his data for the same period indicate greater values than our model), we must conclude that the Agung dust layer must have lost some of the larger particles of its size distribution somewhere between Australia and South Africa. One of the most interesting aspects of the Agung evenCindeed of any volcanic activity that affects the optical weather-is the rate of spreading and the extent of its dust veil as a function of geographic location, time of year, and elapsed time from the main eruption. Despite the obvious importance of such information to the meteorological sciences, no coordinated observational

VOLCANIC TURBIDITY ANOMALIES

287

program seems to have been organized even after the magnitude of the explosion became known. The best attempt to collect and analyze available data is probably that of Dyer and Hicks (1965, 1968). This effort is mainly based on a rather crude dust index defined in terms of the reduction of the (unresolved) direct solar flux, as measured routinely a t a number of meteorological stations over the globe. Notwithstanding the gross simplifications involved in their analysis and the sparsity of data, these authors have been able to present a coherent and plausible picture of the spreading of Agung's dust. The most reliable and interesting of their conclusions seem to be: (a) that the initial injection height was about 22 t o 23 km, creating an equatorial reservoir of stratospheric dust; (b) that most-but not a l l - o f this dust remained in the southern hemisphere with a gradual decrease in total amount over subsequent years; (c) that there was a winter dust maximum in each hemisphere, uniformly in phase between 30" and 90" latitude, with an apparent poleward progression with time; (d) that the principal spreading agent could have been the mean large-scale motion of the atmosphere rather than eddy diffusion. [The last two inferences have been further discussed by Clemesha (1971) and by Dyer (1971.1 Another interesting phenomenon, which may be indirectly related to Agung, was suggested by Pittock (1966). He detected a rather sharply defined ozone-deficient layer between 20 and 21 km over Boulder, Colorado, in the routine ozonesonde records for MarchIApril 1964. He then observed that this coincided with an aerosol layer a t the same height, apparently detected by means of a single observation of abnormal attenuation of h0.44 pm light from the setting sun, as reflected on a large meteorological balloon floating a t 33 km. Although the reliability of the dust detection may be questioned on several grounds, Pittock's suggestion that it originated over Agung and that the entire layer was transported by horizontal winds to Boulder without much change is intriguing. Interestingly, the author does not attribute the ozone deficiency to the local action of the volcanic particles themselves, but rather to a conservation of the initial properties of the tropical stratospheric air. These conclusions were essentially confirmed by Grams and Fiocco (1967)who, from an independent analysis of ozone soundings over Bedford, Massachusetts, found a statistically significant anticorrelation between total ozone and stratospheric dust amounts. These authors also refrain from attributing the above to a direct causal connection between the presence of dust and the reduction of ozone amounts. Another possible effect has been suggested by Newel1 (1970a)b) who attributes a 5" positive anomaly in stratospheric temperatures, observed over northwest Australia late in 1963, to absorption of sunlight by Agung particles. Sparrow (1971) has pointed out that the anomaly might just as plausibly be attributed t o a breakdown of the so-called quasibiennial oscillation found in

288

D. DEJRMENDJIAN

recent meteorological soundings over the tropics. A partial confirmation of this may be found in a recent analysis of pertinent data by McInturff et al. (1971). I n any case, the effect of absorbing and emitting aerosols on the ambient atmospheric temperature is not well understood nor experimentally verified. Until the possible existence of such an effect can be proved, it should not be used to explain certain temperature anomalies before considering other likely possibilities. The unique-by location-measurements of direct sunlight and global radiation at the South Pole (about 3200 m above mean sea level) described by Viebrock and Flowers (1968) also deserve mention. The data for normally incident direct sunlight, measured with a n Eppley pyrheliometer on cloudless days, clearly show the effect of an attenuating layer after November 1963 (data for September and October 1963 were unavailable). Unfortunately, simultaneous transmission measurements in narrow bands were not conducted and hence there is a lack of information on the wavelength dependence of the turbidity anomaly. This would enable us to compare the Antarctic data with the stellar extinction data, discussed above, in order to check the likelihood of Agung dust as the attenuating layer over the South Pole, as suggested by these authors. As i t is, from their tabulated data for unresolved radiation on 10 February 1964, a rather turbid day (Viebrock and Flowers, 1968, Table 3), one may deduce an “ average ” optical thickness T~ over the spectral range of the instrument, by equating air mass with the secant of the solar zenith distance and by putting 0.68 (due to the elevation) for the normal air mass. We thus have 2.064 - 0.884 = 0.570 2.054

eXp{-o.68?~ SeC 75’26’}

where 2.064 is the reported extraterrestrial energy and 0.884 is the amount depleted by aerosol scattering. This yields a value of 0.208 for T ~the, average optical thickness of the dust layer. The same type of reduction applied to the 0.723 of “normal intensity,” observed on December 18, 1963 (also a highly turbid day), yields TD = 0.191. Comparing these values with the estimates shown in Fig. 2 for Agung’s dust turbidity anomaly, we see that they fall just below the values for the middle of the visible range, obtained from data closer to the volcanic source in latitude and time. Thus it is quite conceivable that the anomalous attenuation of direct sunlight, reported over the South Pole in late 1963 and early 1964, was entirely caused by Agung’s dust layer, transported essentially unmodified-whether by advection or otherwise-from the source region. Viebrock and Flowers (1968)also list values of the global radiation a, or the sum of direct sunlight S , and entire sky radiation H,incident on a horizontal

VOLCANIC TURBIDITY ANOMALIES

289

surface, a t Amundsen-Scott Station. These are given in terms of the ratio of the measured flux to the equivalent theoretical value G R for a plane-parallel Rayleigh scattering atmosphere over a Lambert surface with 0.80 reflectivity, determined some years ago by Deirmendjian and Sekera (1954). The most interesting result here is the rather high values of 0.89 to 0.97 for the ratio GIGR observed during the antarctic summer months of 1961 and 1962, despite the idealized conditions implied by the Rayleigh model. However, as pointed out in that early study, changes in the global radiation are not reliable indicators of the presence, amount, or type of turbidity, simply because such radiation represents an integral over too many variables that may be mutually compensating. Thus, it is not surprising that, after the presumed onset of Agung dust over the South Pole, there was only a modest 0.07 decrease in global radiation (Viebrock and Flowers, 1968). On the other hand, under cloudless skies and moderate turbidity conditions, an increase in turbidity produced by predominantly scattering aerosols will result in a lowering of S together with an increase in H . Thus, the ratio SIH will drop even faster with increasing turbidity, an effect that is intensified over a highly reflecting surface and a t low solar elevations. Both these conditions, of course, are precisely those found in Antarctica, and the ratio SIH, easily obtainable by means of the instrumentation used there, would be a very sensitive general index of turbidity. Nevertheless, Viebrock and Flowers (1968, Table 1) do not tabulate and analyze this parameter in detail, preferring the somewhat less sensitive ratio Hl(S H ) as a turbidity index. From their limited published data, we note that the value of SIH would have been as low as 0.875 and 1.66 in February 1964 and 1965, respectively, compared to a high of 4.42 in February 1960. These may be compared with SIH values derived from old Smithsonian Observatory data (Deirmendjian and Sekera, 1954, Fig. 4). The high South Pole value mentioned above falls well within the 1917 observations (ranging between about 2.7 and 6.5 for a solar zenith distance 75O.5) a t Hump Mountain, North Carolina, under clear conditions. The low South Pole value of 0.875 is well below that of about 2.0, observed a t Mt. Wilson, California, in September 1913, when the effects of Katmai’s dust were presumably still present. If we assume this t o be a true difference-and not the result of instrumental and other extraneous discrepancies-it must be concluded that the turbidity over the South Pole in February 1964 was higher than that over Mt. Wilson in 1913. Such a conclusion, however, is a t variance with our own derivation of the overall magnitude of the turbidity introduced by Katmai and Agung, respectively, on the basis of the more direct evidence provided by attenuation of sunlight and starlight. If corroborated, therefore, the high concentration of Agung dust over the South Pole must be explained in terms of atmospheric transport mechanisms capable of producing such concentrations.

+

290

D. DEIRMENDJIAN

6. CLIMATICEFFECTS OF VOLCANICDUST

Certain glacial and climatic effects have been loosely related in the literature to periods of increased volcanic activity (cf. SMIC, 1971, pp. 39 and 4143).Lamb (1970)recently collected and tabulated all the available information on past volcanic activity, and attempted their classification in terms of a quantitative dust veil index (d.v.i.) ” which might provide a better understanding of possible climatological effects related to volcanic dust. I n our opinion this was not quite successful, essentially because of that author’s failure to provide a clear and unambiguous definition of his proposed index, of the parameters involved, and their relation to various meteorological concepts and processes. Three separate, purely empirical formulas are used to determine the d.v.i., each based on a different criterion and chosen on the basis of the availability of information-rather than its reliability-in each case: reduction in insolation, reduction in “ mean temperature,” and volume of dust material, respectively, all weighted by the extent and duration of the “dust veil.” I n view of the unreliability of most of the source material and the diversity of these criteria, the ‘‘ d.v.i.” numbers thus obtained can hardly be expected to be mutually comparable and uniformly significant. Whereas two of the above criteria, as we have seen, may be related to the magnitude of the dust effect, a supposed reduction in “mean (surface) temperature,” despite Lamb’s (1970,pp. 460469) unsubstantiated claims, may by no means be related to the presence of volcanic dust until considerably more evidence than we now possess becomes available. Potentially the most meaningful d.v.i. criterion, the optical thickness anomaly rD(h),which depends on the amount, type, and size distribution of the volcanic dust particles, is not considered except in the most simplistic terms (Lamb, 1970, pp. 460-461). We compare below Lamb’s d.v.i. numbers for the three volcanic events we have been discussing here with our own estimates of relative amounts of injected dust, adjusted to Lamb’s own scale of lo00 units for Krakatoa. ‘I

d.v.i. (Lamb): This work:

Krakatoa

Katmai

Agung

1000 1000

500 443

800 310

The values in the last line are adapted from the ratios we arrived a t in the previous section on the basis of the optical thickness anomaly, definite aerosol models, and an assumed spread area for the dust. Comparison of the two evaluations shows that, whereas they agree for Katmai, Lamb’s d.v.i. for Agung is, in our view, considerably overestimated.

VOLCANIC TURBIDITY ANOMALIES

291

Because of the reservations mentioned above, the validity of Lamb’s remaining conclusions cannot be assessed. However, his detailed and annotated tabulation (Lamb, 1970, App. I) of most known volcanic events in chronological order is a useful contribution to the interested student of climatology. (We note that the important explosions of Mexico’s Paricutin volcano in 1943 and 1945 are not cited in Lamb’s list.) The recent report (SCEP, 1970),prepared by an ad hoc summer study group on problems of man-made modification of the global environment, is of some interest because it includes an attempt to evaluate the magnitude and effects of the particulate loading of the lower stratosphere as a result of the expected large-scale, world-wide operation of supersonic aircraft (SSTs). Such effects if any, would mainly arise through the scattering and absorptive properties of the particles by possible changes in the radiation balance of the stratosphere and perhaps the earth’s surface. It is well known that present knowledge in these areas is incomplete and subject to speculation. The report’s conclusions, therefore-at least on this subject-may not carry much weight, as they seem to be based mostly on conjectural articles and the preferences of participants rather than reliable scientific evidence. Nevertheless, since these conclusions were mostly arrived a t on the basis of Agung’s dust and its presumed meteorological effects, it is worthwhile to examine the SCEP’s estimates of SST particulates in the light of our present estimates and models of volcanic turbidity. The amount of particulates was based on the exhaust products of an estimated 500 SST craft operating a t about the 20-km level for 2500 hours yearly each (SCEP, 1970, pp. 71-74). Taking the “peak N. hemisphere” value of 3.46 parts per billion (by mass) given by the report, one obtains a total of (3.46 x 10-O)(3.85 x 1020)= 1.33 x 10l2 grams for the SST particulate loading, assuming a two-year residence, where 3.85 x 1020is the mass of one half of the stratosphere, as quoted in the report. This may be considered as a worst-case estimate, since the above peak value was taken as ten times the estimated global average. Assuming further that the SST particulates have unit density (again a worst case) this represents condensed particulate material with a total volume of 1.33 x l O l 2 om3. This means that the SSTs at most could add a particulate load equivalent to some 0.15 of that estimated for the Agung volcano (9 x l O l 2 cm3), or 47 of Lamb’s d.v.i. units. If, for the sake of comparison, we confine all the SST particulates to the area above 30”N or one fourth of the earth’s surface, we have 1.33 x 1012 = 1.04 x 1.275 x 10le

cm3 (cma column)-l

292

D. DEIRMENDJIAN

of equivalent condensed material. I n terms of our polydisperse moaels for stratospheric aerosols already mentioned and shown in Fig. 3, this amounts to 0.89 km of water haze L resulting in a turbidity anomaly, ~ ~ ( A 0 . 4= 5 )0.043 (see Fig. 2); or 3.32 km of water haze H with ~,(A0,45) = 0.068. On this basis, the expected turbidity anomaly from future SST jlights, at its worst, will hardly exceed the background turbidity found above mountains on clear days away from city pollution. This is rather less than the SCEP’s (1970, p. 16) estimate which implies SST particulate loadings comparable to Agung’s contribution. By this token, climatologic~1disturbances directly attributable to SST particulates should be negligible. This is not to say that the constant, more massive injection of SST-type exhaust products into the lower stratosphere presents no human environmental problem. The just-published report of yet another ad hoc study group (SMIC, 1971) essentially concurs with the SCEP (1970) evaluation of estimated SST particulates but introduces a note of caution regarding their possible climatic effects. I n general, the 1971 report is subject to the same criticism as the earlier one as regards the reliability of some of its sources. 6. SUMMARY AND CONCLUSIONS I n the light of the increasing attention being given recently by atmospheric scientists and others to the presumed role of particulate turbidity anomalies in past and future climatic changes, a critical review of the subject should be useful. To this end we have attempted to examine and compare the effects of the three major and best documented volcanic events, known to have introduced considerable amounts of particulates: Krakatoa (1883), Katmai (1912), and Agung (1963). Our survey shows that, whereas all three eruptions resulted in clearly recognizable turbidity anomalies over large areas of the earth for periods of a few years, little solid evidence exists of climatological (or weather) effects in terms of anomalies in the conventional meteorological parameters. In comparing the volcanically produced turbidities with that to be expected from the hypothetical operation of a nominal 500 commercial supersonic transports (SSTs), we find that the latter would be but a fraction of the turbidity introduced by a single “low yield” volcanic event such as Agung’s over a two year period. In more specific terms, our principal findings may be summarized as follows (with our ‘‘ worst case ” SST evaluations shown in parentheses). (i) Typical turbidity anomalies, in terms of absolute increments in optical thickness in the visible region (A = 0.5 pm), for the particulate layer as observed away from urban pollution sources, say within a few months a,fter the volcanic events, may be given as Krakatoa, 0.55; Katmai, 0.35; Agung, 0.25; (500 SSTs, 0.05).

VOLCANIC TURBIDITY ANOMALIES

293

(ii) Mainly on the basis of the above optical criterion and additional considerations of the nature and extent of the disturbances, the overall magnitude of the turbidity anomalies produced in each case may be rated as follows, in relative units: Krakatoa, 1.00; Katmai, 0.44; Agung, 0.31 (500 SSTs, 0.047). (iii) The physical and chemical characteristics of the volcanic particles have not, so far, been completely and accurately determined for any of the three events. However, well-defined polydisperse (optical) scattering models may be easily fitted to the observed, wavelength-dependent turbidity. These indicate that the particles may have been composed of nonabsorbing (or very weakly absorbing) dielectric material with refractive index close to that of water (1.33)or silicate (1.55).Whereas the shape of the particles may have been amorphous rather than spherical or crystalline, their size and size-distribution appear to conform to that of other natural aerosols normally found in the upper troposphere and lower stratosphere. Assuming a substance of unit density for the particles, typical local masses of the volcanic dust content for conditions as in (i) may be estimated as Krakatoa, 1.5 x Katmai, 9.0 x Agung, 2.9 x (500 SSTs, 1.0 x gram per square centimeter column, respectively. [These figures are not exactly in the same ratio as the turbidity estimates in (i) as they depend on the scattering model and the spread of the layer assumed in each case.] Likely values for the total mass of material (of unit density) in each case may be Krakatoa, 3 x 1013g; Katmai, 1.34 x 1013g; Agung 9 x 10l2 g; (500 SSTs, 1.33 x 10l2 g). The most massive injection, that of Krakatoa, could hardly have exceeded of the mass of the entire atmosphere (5.14 x loz1g). Among other qualitative conclusions worth emphasizing are the following. No significant anomalies in the global radiation-i.e., the total downward flux of direct and diffusely transmitted solar energy through a horizontal surface-clearly attributable to the volcanic dust layers have been demonstrated, although definite diminutions in the direct (unscattered) sunlight are evident in the records. This does not contradict theoretical expectations from present knowledge of the effects of moderate turbidity on the global radiation. Likewise, the planatary albedo of the earth may not have been altered by the dust. It follows that the radiation balance a t the boundaries of the atmosphere may not have been significantly disturbed, so far as we may judge, by any of the three volcanic events here considered. However, it is quite likely that the broad spectrum (and polarization) of the daylight sky was variously altered in each case. This may have affected the photosynthetic process, for example, which is known to be sensitive to the skylight spectrum; but we are unable to estimate the magnitude of such effects, if any, in the absence of data. As to the specific particle size and size-distribution characteristic of each volcanic event, our present analysis shows that the Katmai and Agung

294

D. DEIRMENDJIAN

particles must have been rather small with a relatively wide, continuous distribution, since no unusual blue sun or Bishop’s ring phenomena were widely evident. We may assume that in both cases the particles were mainly generated in situ by the same processes that are responsible for the creation of normal )’stratospheric aerosols, except for the higher number density resulting from the volcanic gaseous emissions. I n the case of Krakatoa, on the other hand, we believe that the size distribution may have been bimodal, i.e., it may have contained a second, narrow distribution around a large characteristic size, in addition to a long-lived, fine-particle component similar to that of the other volcanoes in size distribution. The larger component, needed to explain the unusual optical phenomena noted, may well have consisted of true particulates injected by the volcano directly into the stratosphere where they may have resided, with diminishing concentration relative to that of the finer component, for as long as two years before settling. Although Krakatoa and Agung are both situated in the same latitude and longitude zone, the extent and mode of spreading of their respective dust material seem to have been quite different. This may be partly attributed to the different time of year in which the events occurred (August and March, respectively), and hence to possible differences in the pertinent large-scale motions; and perhaps also to the 80-year lapse between the two events, so that secular changes in the so-called general circulation may not be entirely discounted. Clearly there is a need for careful and continuous monitoring of atmospheric turbidity. This could be best accomplished by permanent stations in strategic surface locations around the earth to record the narrowband atmospheric transmission of sunlight covering the visible and near infrared region (Deirmendjian, 1971, Appendix). Such data could be supplemented by routine measurements, also a t various wavelengths, of the aureole around the sun a t fixed angular distances in order to better evaluate the size distribution of the particulates. The use of satellites to detect turbidity changes from ‘‘ albedo ” measurements, proposed by participants in the SCEP (1970, pp. 200-202) seems of doubtful value to us, due to difficulties in the interpretation of such data. A most important objective of a good monitoring and data reduction system is, of course, the unambiguous identification of the type and origin of the various components of the observed turbidity, such as volcanic, surface, man-made, etc., as a function of geographical location and time of year. I n addition, a network of turbidity monitoring stations designed for global coverage would serve another important purpose by providing reliable information on the extent and rate of spreading from point sources, such as major volcanic events, to allow the identification of the responsible transport mechanisms. ((

VOLCANIC TURBIDITY ANOMALIES

295

Although this review has revealed little evidence of volcanically induced climatic effects, it is conceivable that the existence of such effects may be established by more refined observation and analysis of pertinent variables than hitherto available. In that event, it is hoped that the present estimates of volcanic dust turbidity anomalies and suggested polydisperse scattering models will be helpful in simulating such effects by incorporating them in radiative transfer calculations and eventually in numerical models of the general circulation.

REFERENCES Abbot, C. G. (1913). Do volcanic explosions affect our climate? Nut. Oeogr. Mag. 24 No. 2, 181-197. Abbot, C. G., Fowle, F. E., and Aldrich, L. B. (1922). Ann. Astrophys. Observ. Smitheon. Inst. 4. Anonymous. (1912). Volcanoes of Alaska. Nut. Qeogr. Mug. 25, No. 8, 824-832. Booth, P. W., Matthews, S. W., and Sisson, R. E. (1963). Disaster in Paradise. Nut. Cfeogr. Mug. 124,No. 3, 436458. Brooks, E. M. (1964). Why was last December’s lunar eclipse so dark? Sky & Telescope 27, NO. 6, 346-348. Clemesha, B. R. (1971). Comments on a paper by A. B. Dyer. J . Geophys. Res. 76, 755756. Deirmendjian, D. (1955a). In “Investigation of Polarization of Skylight ” (Z. Sekera, ed.), Final Report, Conttact AFl9( 122)-239.Univ. of Calif. Press, Los Angeles, California. Deirmendjian, D. (1955b). The optical thickness of the molecular atmosphere. Arch. Meteorol., Qeophys. Bioklimatol., Ser. B 6, 452-461. Deirmendjian, D. (1965). Note on laser detection of atmospheric dust layers. J . Oeophya. Res. 70, 743-745. Deirmendjian, D. (1969a). “Electromagnetic Scattering on Spherical Polydispersions.” h e r . Elsevier, New York. Deirmendjian, D. (1969b). Sunspot observation through water clouds. A p p l . Opt. 8, 833. Deirmendjian, D. (1971). “Global Turbidity Studies. I. Volcanic Dust Effects-A Critical Survey,” R-886-ARPA. Rand Corporation, Santa Monica, California. Deirmendjian, D., and Sekera, S. (1954). Global radiation resulting from multiple scattering in a Rayleigh atmosphere. Tellus 6, 382-398. Dunkelman, L., and Scolnik, R. (1959). Solar spectral irradiance and vertical atmospheric attenuation in the visible and ultraviolet. J . Opt. SOC.Amer. 40, 356-367. Dyer, A. J. (1971). Reply. J. Oeophys. Res. 76, 757. Dyer, A. J., and Hicks, B. B. (1965). Stratospheric transport of volcanic dust inferred from solar radiation measurements. Nature ( L o d o n )208, 131-133. Dyer, A. J., and Hicks, B. B. (1968). Global spread of volcanic dust from the Bali eruption of 1963. Quart. J. Roy. Meteorol. Soc. 04, 545-554. Grams, G., and Fiocco, G. (1967). Stratospheric aerosol layer during 1964 and 1965. J. G‘eophys. Res. 73, 3532-3542. Griggs, R. F. (1917). The valley of ten thoumnd smokes. Our greatest national monument. Nut. Oeogr. Mug. 31,No. 1, 12-68. Griggs, R. F. (1918). Nut. Oeogr. Mug. 33,No. 2, 115-169. Griggs, R. F. (1921). Nat. Oeogr. Mag. 40, No. 3, 219-292. Gruening, E. (1963). Lonely wonders of Katmai. Nut. Oeogr. Mug. 125, No. 6, 800-831.

296

D. DEIRMENDJIAN

Hogg. A. R. (1963). The Mount Agung eruption and atmospheric turbidity. Aust. J. Sci. 86,119-120. Holland, A. C., and Gagne, G. (1970). The scattering of polarized light by polydisperse system of irregular particles. Appl. Opt. 9, 1113-1121. Irvke, W. M., and Peterson, F. W. (1970). Observations of atmospheric extinction from 0.315 t o 1.06 microns. J. Atmos. Sci. 27, 62-69. Junge, C. E., and Menson, J. E. (1961). Stratospheric aerosol studies. J. Beophya. Res. 66,2163-2182. Junge, C. E., Chagnon, C. W., and Manson, J. E. (1961). Stratospheric aerosols. J. Metewol. 18,81-108. Lamb, H. H. (1970). Volcanic dust in the atmosphere; with a chronology and assessment of its meteorological significance. Phil. Trans. Roy. SOC.London 266, 425-533. McInturff, R. M., Miller, A. J., Angell, J. K., and Korshover, J. K. (1971). Possible effects on the stratosphere of the 1963Mt. Agung volcanic eruption. J. Atmos. Sci. 28, 1304-1 307. Martin, G. C. (1913). The recent eruption of Katmai volcano in Alaska. Nut. aeogr. Mag. W,NO. 2, 131-181. Meinel, M. P., and Meinel. A. B. (1963). Late twilight glow of the ash stratum from the eruption of Agung volcano. Science 142,582. Mossop, 8. C. (1964). Volcanic dust collected at a n altitude of 20 km. Nature (London) SOS, 824-827. Newell, R. E. (1970a). Stratospheric temperature change from the Mt. Agung volcanic eruption of 1963. J. Atmos. Sci. 27, 977-978. Newell, R. E. (1970b). Modification of stratospheric properties by trace constituent changes. Nature (London) M7, 697-699. Pittock, A. B. (1966). A thin stable layer of anomalous ozone and dust content. J. Atnwa. Sci. $38, 638-542. Przybylski, A. (1964). The reduction of photometric observations affected by variable extinction. Acta Astron. 14, 285-296. Rozenberg, G. V. (1963). “Twilight.” Plenum, New York (R. B. Rodman, transl., 1966). SCEP. (1970). “Man’s Impact on the Environment,” Report of the Study of Critical Environmental Problems. MIT Press, Cambridge, Masemhusetts. Sekera, A. (1957). Polarization of skylight. I n “Handbuch der Physik” (S. Flugge, ed.), Vol. 48, pp. 288-328. Springer-Verlag, Berlin and New York. Shah, C. M. (1969). Enhanced twilight glow caused by the volcanic eruption on Bali island in March and September 1963. Tellus 21, 636-640. SMIC. (1971). “Inadvertent Climate Modification,” Report of the Study of Man’s Impact on the Climate (SMIC). MIT Press, Cambridge, Massachusetts. Sparrow, J. G. (1971). Stratospheric properties and Bali dust. Nature (London)229, 107. Symons, 0. J., ed. (1888). “The Eruption of Krakatoa and Subsequent Phenomena.” Krakatoe Committee, Royal Society, London. van de Hulst, H. C. (1957). “ Light Scattering by Small Particles.” Wiley, New York. Viebrock, H. J., and Flowers, E. C. (1968). Comments on the recent decrease in solar radiation at the South Pole. Tellw, w),400-41 1. Volz, F. E. (1964). Twilight phenomena caused by the eruption of Agung volcano. Science 144. 1121-1122. Volz, F. E. (1970a). On dust in the tropical and midlatitude stratosphere from recent twilight memrements. J. Ueophys. Res. 76, 1641-1646. Volz, F. E. (1970b). Atmospheric turbidity after the Agung eruption of 1963 and size distribution of the volcanic aerosol. J. Ueophya. Res. 76, 5186-5183.

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS INTRODUCED BY ISLANDS Kuldip P. Chopra Department of Physics. School of Sciences Old Dominion University. Norfolk. Virginia and Division of Physical Science and Coastal Engineering Virginia Institute of Marine Science. Gloucester Point. Virginia

.

1 Introduction and Summary ............................................ 1.1. Atmospheric Flow Patterns Introduced by Islands (Nonconvective Effects) ......................................................... 1.2. Atmospheric Circulations Generated by Heated Islands . . . . . . . . . . . . . . . 1.3. Oceanic Circulations Introduced by Islands .......................... 1.4. Significance of Studies of Atmospheric and Oceanic Flow Problems Introducedby Islands ............................................ 1.5. Historical Background and Prospectus .............................. 2 . Microscale Perturbations .............................................. 2.1. Prevailing Wind Field Near Argus Island ........................... 2.2. Disturbance of Ambient Wind Field Caused by Argus Island .......... 3 . Group of Small Islands as Mesometeorological Network . . . . . . . . . . . . . . . . . . 3.1. Kwajalein Atoll ................................................. 3.2. Analysis of Rawinsonde Data at Kwajalein and Roi-Namur Islands .... 4. Mesoscale Atmospheric Vortices Leeward of Islands ...................... 4.1. Atmospheric Vortex Streets ....................................... 4.2. The Vortex Street Phenomenon: General Considerations .............. 4.3. Role of Viscosity in Vortex Street Phenomenon ...................... 4.4. Analysis of Atmospheric Vortex Streets Leeward of Islands . . . . . . . . . . . 5 . Vortices Leeward of the Hawaiian Island8 .............................. 6.1. The Hawaiian Islands ............................................ 5.2. Atmospheric Flow and Climatological Properties of the Region . . . . . . . . 5.3. Oceanic Currents and Circulations .................................. 6.4. Properties of Oceanic Eddies ...................................... 5.5. Generation Mechanisms for Hawaiian Eddies ........................ 5.6. Concluding Remarks ............................................. 6 Anomalous Oceanic Circulations Around Islands ......................... 7 . Upwelling Due to Circulations Around Islands. .......................... 7.1. Upwelling in Oceanic Eddies Leeward of Hawaii ...................... 7.2. Upwelling Caused by Winds Parallel to Long Islands . . . . . . . . . . . . . . . . . 8 . Air Flow Over a Heated Island ........................................ 8.1. Land and Sea Breezes ............................................ 8.2. Air Flow over Typical Islands ..................................... 8.3. Urban Heated Islands ............................................

.

VIMS Contribution No . 489. 297

Page 298 298 310 316 320 323 323 324 324 326 326 327 328 328 329 332 339 367 367 367 358 358 367 371 371 377 377 378 383 383 384 402

298

KULDIP P. CHOPRA

9. Recent Experiments in Tropical Island Meteorology ..................... 9.1. The Line Islands Experiment (LIE) 9.2. The Barbados Oceanographic and Meteorological Experiment (BOMEX) 10. Concluding Remarks ................................................. References ..........................................................

................................

406 406 409 414 418

1. INTRODUCTION AND SUMMARY

Islands act as obstacles to the prevailing winds and ocean currents. Because of differential heating of the land and the surrounding water masses, islands generate local circulations known as the land and sea breezes. Bounded by the contours of islands from within and the contours of continents from without, oceans form multiply connected regions. The interchange of energy (heat), momentum, and moisture between air and sea combines with the above-mentionedfeatures of islands to produce a variety of mesoscale atmospheric and oceanic circulations. The nature of flow around an island is also influenced by several other factors such as the island’s shape and size, its geographic location and relief, proximity of other land masses (continents and/or other islands), stratification of the atmosphere, and nature of the prevailing winds and ocean currents.

1 .I. Atmospheric Plow Patterns Introduced by Islands (Nonconvective Effects) An island acts as a barrier to the prevailing winds, and, determined by its size, it causes microscale or macroscale perturbations in the prevailing wind field in its vicinity. The terrain of the island and the atmospheric stratification determine the detailed structure of these perturbations. When encountering an obstacle, unstable air continues to rise, the stable air goes around, and the neutral air goes around, over, and under the obstacle if an underpass is available. Various kinds of atmospheric flow conditions generated by islands are briefly listed below, but will be discussed in more detail in later sections.

1.1.1. Microscab Perturbations. Many islands of very small size exist on our planet. For example, the numerous islands that form the Kwajalein Atoll (Fig. 1)occupy a total land mass of 15 kma. Ships and meteorological towers in oceans may-also be regarded as very small islands, which cause microscale changes in the prevailing wind field around them. The flow field around the Argus Island Tower may be treated as a typical illustration of microscale modifications. From the aerodynamic point of view, Argus Island (Fig. 2) is a complex-structure platform in the open Atlantic. The flow field introduced by it under atmospheric conditions

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

299

FIG.1. Locations of the Islands of Roi-Namur and Kwajalein in the Kwajalein Atoll, Marshall Islands.

of neutral (adiabatic) stratification consists of two jets of accelerated air just above and below the platform, and a spike upwind and a wake of retarded air flow (Thornthwaite et al., 1965). Simultaneous rawinsonde measurements at Kwajalein and Roi-Namur Islands, only 82 km apart, reveal significant differences in meteorological parameters at all levels, from ground up to an altitude of about 33.5 km. Therefore, meteorological measurements a t one island do not necessarily represent the atmospheric conditions prevailing all over the atoll (Billions, 1967). A striking microscale modification of winds is caused by the Isthmus of Catalina Island, located about 40 km off the coast of Southern California. Santa Catalina is hilly, 300 t o 600 m high, and about 32 km long. A stable layer of air coming in from the Pacific is suddenly accelerated to high speeds as i t pushes its way through the Isthmus as a result of the Venturi effect. The path of the strong winds is well marked, with all the branches of trees trailing off on the downwind side of their trunks from one side of the island to the other (Edinger, 1967).

300

KULDIP P. CHOPRA

I1.00

44 4 1 $140

WEST SIDE

FIG 2. Southerly ambient flow modified by the Argys Island Tower under conditions of neutral atmospheric stratification. The numbers marked on the streamlines identify the ratio of the observed to ambient wind speeds (after Thornthwaite et al., 1966).

1.1.2. Mesoscale Vortices Leeward of Islands. Tiros V and VI weather satellites were the first to reveal the mesoscale vortex patterns shown in Figs. 3 and 4 in the lee of the Canary Islands. These patterns, consisting of vortices with radii of 10 to 20 km in wakes 40 to 50 km wide and 400 to 600 km long, were observed t o persist for 18 to 30 hours. Gemini VI and IX missions photographed similar patterns in the same geographic area. Observations have also been made (Friday and Wilkins, 1967) of the vortex streets associated with Guadalupe Island, Mexico (Gemini V and VII missions), the Cape Verde Islands (Gemini V and VI missions), and the islands of La Reunion and Mauritius (Gemini VI mission). ESSA 7 AVCS and ESSA 8 APT pictures have provided the desired opportunity (Tsuchiya, 1969) to observe the time variation of vortex wakes leeward of the Cheju Island, Korea. Some of these vortex patterns are displayed in Figs. 3 through 11. Chopra and Hubert (1964, 1965b) identified these eddy patterns as the atmospheric analogs of the von KBrm4n vortex street phenomenon, in which vortices arrange in two rows (Fig. 12) in such a way that (i) the vortices in one row have similar circulations, but of a sense opposite to that of the vortices in the other row; and (ii) each vortex in one row is located across the

-

ATMOSPHERIC A N D OCEANIC FLOW PROBLEMS

301

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

303

FIG.6. Eddy patterns west of Cape Blenc, July 2, 1962 (TirosV).

FIG.4 (left) Eddy pattern superimposed on surface analysis for 1200 hours GMT, July 2, 1962. (right) Complex eddy pattern downstream from the Gran Canaria and Tenerife Islands at 1600 hours CMT,July 2, 1962 (TirosV pass 187).

304

KULDIP P. CHOPRA

Fro. 6. Vortioes associated with the Tenerife and Cornera Islands at 1046 hours OMT, December 16, 1966 (&mini VI).

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

305

FIQ.7. Terrain-induced vortices over Canary Islands (Nimbus I11 IDCS orbit 996, June 27, 1969).

306

KULDIP P. CHOPRA

FIU.8. Vortex wake of Guadalupe Island, Mexico at 1861 hours GMT,September 13, 1964 (Nimbus 1 AVCS).

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

307

FIG.9. Vortices leeward of the Guadalupe Island, Mexico at 2039 hours GMT, November 13, 1966 (Gemini XII).

308

RITLDIP P. CROPRA

Fig, 10. Vortex street associated with the Guctdalupe Island, Mexico at 1451 hours OMT, March 12, 1969 (Apollo 9).

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

309

FIG.11. Vortex street leeward of the Cheju Island, Korea at 05h l l m 13s GMT, March 6, 1969 (ESSA 7 AVCS).

.

MI

W

, w-----

y*

-.-# ’ ‘-A

FIG.12. Schematic diagram of a von Khrmin vortex street.

310

KULDIP P. CHOPRA

midpoint of the two adjacent vortices in the other row. After a vortex is formed and shed, it is carried downstream at a translational speed somewhat lower than the ambient mean wind speed. The Chopra-Hubert interpretation of these vortex patterns has been extensively explored by Chopra (1966, l968,1972c), Chopra and Hubert (1965a),Hunt and Wickins (1967),Tsuchiya (1969), Wilkins (1968), and Zimmerman (1969). These vortex patterns are generally associated with steep-sided islands which extend well above the low-lying inversion. Although the atmospheric vortex streets were revealed in the background of the stratocumulus clouds below strong inversions, they are formed in the atmospheric layers which are stable against convective motions. The island of Hawaii, centered a t 19" N latitude, acts as a barrier to the most constant northeasterly winds in the world, the trade winds, and generates cyclonic eddies to its north and anticyclonic eddies to its south. However, there is no evidence of formation of the vortex streets discussed above. Kona coist (west) of Hawaii, lies in the convergence zone of these vortices and receives more rainfall than any other leeward area in the chain of the Hawaiian Islands (Fig. 13).

I NW

I

I

I

1

I

- 5000 - 4000 - 3000 f - 2000 y - 1000 -0

P X

- 1000 ; -2000

n

-3000 g

- 4000

- 5000

FIQ. 13. Topography of the Hawaiian Islands showing the southeast to northwest cross section of the Hawaiian archipelago, below and above sea level (after Patzert, 1970).

I .2. Atmospheric Circulations Generated by Heated Islurbds Land and sea breezes are characteristic climatological features of most island8 and coastal areas. They blow from land to sea a t night and from sea

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

311

to land,during the day. However, there are some exceptions and a wide range of variations in the typical properties of these breezes.

1.2.1.Land and Sea Breezes. These breezes form a natural phenomenon in which solar radiation energy is converted into atmospheric convective motion. Land absorbs the incident solar radiation in a very narrow surface layer and warms the thin layer of air in contact with the surface layer of land causing i t to rise due to buoyancy. The sea, on the other hand, absorbs and stores the solar energy in a larger and deeper volume of water through turbulent mixing due t o winds, waves, currents, and vertical circulations. Furthermore, the thermal capacity of land mass is much lower than that of water. These two properties lead t o significant differences in temperature over land and sea surfaces, and in the larger magnitude of the diurnal variation of temperature over land. Defant (1951) estimates that heating over land during the day can be five times as much as that over the adjacent water surface. The resulting thermal gradient along the land-sea surface gives rise t o land and sea breezes. As warm air rises over land during the day, cooler air from the sea flows on land to replace it. As the cold air over the sea is denser than that over the land, i t sinks while the warm air over land rises. This causes pressure to decrease more rapidly with height in cold air than i t does in warm air. Over heated land, isobars are expanded upward and cause a pressuregradient flow seaward a t higher altitudes, while higher surface pressure over the sea causes the landward pressure-gradient flow. The direction, instants of onset and retreat, intensity, and vertical depth of sea-breeze circulation depend on several factors such as prevailing winds, atmospheric stratification, and size of island, its topography, and geographic location. Figure 148 illustrates the typical sea-breeze pattern for the simplest case of a uniformly heated island in the absence of any prevailing winds. Land and sea breezes are best developed in places where daily temperature variation is largest and sea-land temperature contrasts are maximum. These conditions prevail in the tropics where breezes are a year-round phenomenon. I n midlatitudes, breezes are milder and appear in spring or summer. Depending on local conditions and weather, the sea breeze sets in between 0800 and 1100 hours LST, reaches a peak intensity (wind speed of 6 t o 15 m/s) around 1300 to I600 hours, and subsides around 2000 hours. It penetrates inland to typical distances of 20 to 30 km,sometimes as far as 150 km inland in continental coastal areas, and extends to a height of about 1 km. The maximum sea-breeze speed occurs a t a height of 200 to 400 m above ground. The land breeze is limited to a shallower layer, about 100 to 200 m high. The presence of mountains enhances the sea-breeze effect. Neumann (1951) found that a concave coast would cause a divergent sea breeze and convergent

312

KULDIP P. CHOPRA

-

-

ISLAND ISOBARS

-- -

SEA SURFACE SEA B R E E Z E CIRCULATION PATTERN

4-

P FIU.14. (a) Typical sea-breeze pattern over a uniformly heated island in the absence of prevailing winds. (b). Typical influence of strong prevailing winds on sea breeze of a uniformly heated island (length of arrow indicative of magnitude of the horizontal wind speed).

land breeze. Sommerville (1958) points out that a low-level inversion induces a strong breeze beneath the inversion due to the Venturi effect. The sea breeze brings moist air from the sea and should, therefore, enhance moisture content and precipitation over land. Prevailing winds would accelerate an onshore breeze on the upwind side of the island and reduce or completely annihilate a sea breeze on the downwind side of the island (Fig. 14b). Even with a land-sea temperature difference of 8" t o 16"C, a sea breeze may not appear if the prevailing wind is strong enough in the opposite direction. I n essence, these breezes are established through the balance of the pressure (thermal) gradient and the frictional forces. Once developed, however, they are subjected to slow and slight action of the Coriolis force induced by rotation of the Earth. The Coriolis force tends to veer the sea breeze clockwise with time in the northern hemisphere and counterclockwise in the southern hemisphere. For example, Frizzola and Fisher (1963)found a 16"veer of the lower 200 m of the sea-breeze layer over Long Island, New York, 4 hours after the maximum land-sea temperature difference was established. In general, the influence of the Coriolis force is weak because of shortness of distance and duration over which it is effective in the sea-breezephenomenon. Linear theoretical models developed by Haurwitz (1947), Schmidt (1947), Pierson (1950), and Defant (1951) predict the above-mentioned general characteristics of sea breezes. Nonlinear models developed by Fisher (1961)

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

313

and Estoque (1961, 1962) introduce additional features of (i) advection and (ii) variable vertical eddy exchange coefficient. To a greater or lesser degree, most island-induced sea breezes satisfy the above mentioned general characteristics. Cuba (Fig. 15) is perhaps the model island with sea-breeze patterns possessing the characteristic properties mentioned above. There are, however, other islands that, are considered examples of noteworthy exceptions to the above rules.

FIQ.15. Topography of the Island of Cuba.

1.2.2. Paradoxical Sea-Breeze Effects.The following atypical features of sea breezes associated with certain islands are significant and provide food for further thought. (a) Palmer (1967) points out that the Island of Niue (19OS, 170"W) in the South Pacific is 24 km across, completely fiat, and displays no sea-breeze activity (Fig. 16a).Therearesmaller islands (Fig. 16b)withmountains, particularly toward the trade-wind side, which display sea-breeze effects.Theislandof New Caledonia, centered a t 21" S and 166" E, is only 40 km across along the path of the trades, but its west coast is able t o develop a strong sea breeze which reaches a speed of 30 to 35 knots (15 to 18 m/s) in the afternoon. New

314

KULDIP P. CHOPRA

ISLAND OF NlUE (a 1

0

lOkm

ISLANDS OF FIJI

u

(b)

2 oo

220

I

I64O

166.

168O

- O

(C)

FIG.10. (a)Island of Niue. (b) Islands of Fiji.

(a)

100 km

Island of New Caledonia.

ATMOSPHERIC A N D OCEANIC FLOW PROBLEMS

315

Caledonia has mountains 500 to lo00 m high along its length, acting as barrier to trade winds (Fig. 16c). (b) A mesoscale effect of significance frequently occurs in the region west of Hainan Island (Pig. 17) during lull periods when low-level flow is generally easterly to southeasterly in the northern part of the South China Sea. Bands of showers develop during the night and dissipate during early forenoon along a zone about 150 km long and about 30 km wide, extending toward the west or southwest from Hainan.

FIG.17. Convergence zone with bands of showers west and southwest of the Islend of Heinan. (c) Rows of cumuli, with periodic spacings 1 to 1.5 km and extending to great distances downwind of several islands in the Woods Hole arm (Fig. 18), are frequently observed on sunny summer days (Malkus and Stern, 1953). Observational features of the phenomenon are suggestive of the possibility of wave-like vertical oscillations in the streamlines before condensation, similar to those in the lee of mountains in which stability is the predominant restoring force. (d) Garstang (1967) points out the following discrepancies between observations and the familiar sea-breeze model discussed in the preceding section. (i) There are significant variations in time of onset, magnitude, and even direction of the sea breeze. These variations are observed over tropical islands embedded in basic currents which are characteristically steady.

316

KULDIP P. CHOPRA

142 KM

WIND DIRECTION

FIG.18. Cloud street leeward of Nantuaket Island (after Malkus and Stern, 1963).

(ii) Observations from Barbados, Bermuda, and some Pacific islands show that surface wind speeds decrease on the windward side and increase on the leeward side of certain islands as heating progresses. (iii) Rainfall over Barbados occurs mainly at night when organized atmospheric disturbances are not expected. (iv) Moisture content is frequently observed to decrease during the day over an island or a coastline.

1.3. Oceanic Circulations Introduced by Islands The nature of oceanic circulations introduced by an island depends on the island's location, the nature of prevailing ocean currents and winds, and atmospheric stratification.

1.3.1. Vortices Leeward of the Hawaiian Islands. The Hawaiian Islands are embedded in trade winds. Based on data gathered in 20 cruises since 1949, Patzert (1970) finds that the ocean circulation pattern in the region extending from 40 to 360 km on the downwind side of these islands is complex and variable. It consists of a largo number of eddies, mostly cyclonic, with radii of

317

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

25 to 75 km and surface wind speeds of 1 m/s or more in them (Fig. 19). These eddies are shallow with circulations limited to depths of 200 m. A cyclonic eddy is characterized by doming of isothermal layers and upwelling near it5 center. An anticyclonic eddy displays just the opposite characteristics and is, in general, less intense in this region than a cyclonic eddy. Variations in salinity, surface temperature, and dynamic height are also associated with these eddies. Eddies take about 2 to 4 weeks to form and about 2 months or more to dissipate as they move away from islands with an average translational speed of about 6 cmls. The volume transport is observed to be as large as 8 x lo6 m3/s and the phenomenon is observed year around. Patzert (1970) attributes this oceanic circulation pattern to strong local winds blowing through the restricted passage between the Islands of Maui and Hawaii. 160°

2201610

. ..

I

.

21O'

I9O

159O W

I

158' I

157O I

I56O I

KAUAI

220

..A.OAHU

.

t

I80 161O

160°

I59O

158'

I57O

I56O

FIG.19. Location and drift of eddies in the lee of Hawaiian Islands. Cyclonic; 0 anticyclonic; --f observededdydrift; - --f inferred eddydrift (UH-12,NEL 36 UH-14), (after Patzert, 1970).

1.3.2.Anomalous Oceanic Circulations. Makarov (1950) and Shtokman (1954, 1966) have pointed out a peculiar oceanic flow near certain islands. Under certain circumstances, circulation around an island follows an opposite direction to overall circulation in the region of the ocean containing the island. I n most cases, this anomalous circulation is anticyclonic (clockwise) in the northern hemisphere. Typical examples of this curious phenomenon are found in the Islands of Hainan, Taiwan (Fig. 20)) Iceland (Fig. 21), and

FIQ.20. Schematic diagram of the anomalous circulation around the Island of Taiwan (after Shtokman, 1966).

FIQ.21. Anomalous circulation around Iceland (after Shtokman, 1966).

ATMOSPHERIC A N D OCEANIC FLOW PROBLEMS

319

the Kuril Chain in the Sea of Okhotsk (Fig. 22). Whereas oceanic circulation around Iceland is observed during the year, that associated with Taiwan is seasonal and is observed only during the winter. The anomalous circulations in all three cases appear t o be associated with two additional conditions: (a) their asymmetric location with respect t o continental boundaries, and (b) existence of transverse cyclonic homogeneities in prevailing winds of the respective vicinity. Based on these features, Shtokman (1966) has proposed a semiquantitative explanation of observed phenomena.

Fro. 22. Circulations around the chain of Kuril Islands (after Shtokman, 1966).

1.3.3. Upwelling Effects. Two types of upwelling effects are of interest in this study: (a) Upwelling associated with cyclonic and anticyclonic eddies as described in Section 1.3.1.,and (b) upwelling which may be caused by ambient wind flow parallel to the length of a very long island. This, in turn, would result in separation of warm and cold surface waters on opposite sides of the island.

320

KULDIP P. CROPRA

1.4. Significance of Studies of Atmospheric and Oceanic Flow Problems Introduced by IslancEB Investigations of island-induced atmospheric and oceanic circulations are of importance for the following reasons. (a) As discussed in preceding sections, islands generate a great variety of air and oceanic flow problems and, hence, provide unlimited supply of challenge in imagination and thought. (b) Each island provides a unique and naturally controlled physical environment for study of modification of an air or water mass as it flows over or across the island. (c) Problems of islandic circulations have an unusual advantage over many other meteorological problems. Circulation problems are well defined for each island, and can be studied independently and simultaneously, therefore, by techniques involving direct observations (field studies), laboratory and numerical modeling, and purely theoretical considerations. (d) Each island provides a relatively simple and well-defined situation for inquiry into the relationship between certain types of convective motions and their initial energy sources in organized unsaturated motions. (e) Islands are Nature’s own weather modifiers. Therefore, efforts to understand the islandic environment have direct bearing on practical methods for weather modification. ( f ) A group of small islands can serve as a subsynoptic or mesoscale network without disturbing the large-scale (synoptic or planetary scale) flow pattern. Both the synoptic scale and mesoscale motions in the atmosphere are important and there are complicated feedbacks between them. The subsynoptic scale network, when embedded in a larger scale network with less finely spaced stations to monitor appreciable fraction of the globe, should yield important information on interactions among all scales of atmospheric motions. (g) Our understanding of the complex air-sea interactions would be enhanced by studies of the environment of individual or group(s) of islands. For example, the explanation of anomalous oceanic circulations (Section 1-3.2) may clarify the mechanism of interaction between inhomogeneous winds and ocean currents in multiply bounded oceans. The explanation of the paradoxical heated klud effects in Section 1.2.2 may shed light on the complex processes of interchange of mass (moisture), momentum (through shear stress) and energy (sensible and latent heat) between ocean and marine atmospheres. An analysis of how winds generate oceanic eddies leeward of the Hawaiian Islands and why this interaction results in more cyclonic and very few anticyclonic eddies (Section 1.3.2.) would explain another aspect of the air-sea interaction. The upwelling phemomens discussed in Section 1.1.3 are wind

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

321

driven in one form or another and provide additional phenomena worthy of further exploration. (h) Local changes in density, moisture, and temperature are caused by atmospheric eddies and, thereby, affect visibility and propagation of sound and light. However, the effect of greater fundamental and applied significance is the one related to local turbulence caused by vortices on the premise that larger eddies disintegrate into smaller eddies. (i) Oceanic eddies cause changes in the density structure through the associated upwelling effect and affect underwater sound propagation. Also the upwelling associated with eddies affects the biological environment. Therefore, study of these aspects is of importance from the point of view of local fishing. ( j ) Problems concerning islandic circulations bear similarities to other apparently different meteorological or physical problems. For example, the atmospheric air flow associated with Nantucket Island has some similarity to air flow over mountains; whereas, the mesoscale vortex pattern leeward of the Canary Islands is similar t o stable fluid flow around obstacles. Also,

FIG.23. Heat island effect of the Ward’s Corner shopping center, Norfolk, Virginia on April 16,1971 (1600 hours t o 1600 hours EDT).

322

KULDIP P. CHOPRA

certain characteristics of the Canary vortices bear similarities to problems in aviation (wake turbulence) created by jumbo jets. Although vortex wakes leeward of islands were revealed in the background of low-lying stratocumulus clouds, their formation depends only on interaction of ambient wind with islands. Wake regions of these islands may serve as natural laboratories for planned studies of dear air turbulence (Chopra, 1971). (k) Uneven heating of land and water masses in a variety of islandic situetions offers opportunities to explore and understand the working of the atmospheric heat engine operating on a smaller scale and to examine the role of water vapor in the energetics of a local (islandic)environment. (1) The problems of heated islandshave similaritiesto differentialheating of urban and suburban areas and to the acoompanying problem of quality of the urban environment. The intensity of the urban heat island effect depends on large temperature gradients, or sharp contrasts between urban and suburban or rural areas. On two clear sunny spring afternoona, Chopra and Pritchard (1971) obtained surface thermal variations around two shopping centers

T

FIQ.24. Heat island effect of the Ward's Corner and Southern Shopping Center in

Norfolk, Virginia on May 6, 1971 (1600 hours to 1600 hours EDT).

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

323

having well-defined residential areas with sharply contrasting surface features (Fig. 23 and 24). 1.5. Historical Background and Prospectw

Since ancient times, people have been interested in ocean currents and land-sea breezes from points of view of fishing, navigation, and desirability of a mild climate. Baralt and Brown (1965) list many early sea-breeze observations. One of the earliest descriptions is given by Herodotus, a 5th century B.C. Greek historian. It seems that the Athenian General Themistocles captured the Persian fleet in the Bay of Salamis by launching his attack with the onset of sea-breeze circulation. One of the first sea-breeze observations associated with islands was carried out by Dampier during 1701-1706. As early as 1873, WaUngford was the first to observe sea-breeze rotation with time due to the Coriolis force. Traditionally, however, islands have held appeal because of pleasant climatic conditions, therapeutic benefits of coastal breezes, and tourism. It is only during the past two decades that three principal scientificinvestigations of islandic environment were initiated. Malkus and Stern (1953) started a program of observational and theoretical studies of convective motions when a stable air stream flows over a localized heat source (a small flat island). In addition, related studies of urban heat islands received their impetus from Sundborg (1950). About the same time, Shtokman (1954) began observations on anomalous oceanic circulations around islands. Chopra and Hubert (1964) proposed an explanation for certain weathersatellite revealed mesoscale atmospheric circulations observed leeward of islands. These studies of specific islandic circulations have been subjects for further detailed analyses in recent years. During the last five years, two major research investigations, involving a large number of experiments and experimenters, have been launched as part of the Tropical Meteorological Experiment (TROMEX): (a) Line Islands Experiment (LIE), and (b) Barbados Oceanographic and Meteorological Experiment (BOMEX).Scientific results of LIE and BOMEX should be of significance in the later phases of TROMEX and in the international Global Atmospheric Research Program (GARP). The text to follow describes phenomena summarily introduced in this section and some thoughts for further work. 2. MICROSCALEPERTURBATIONS

Micrometeorologicalmeasurements over the open sea are usually made with instruments mounted on ships or fixed platforms like Texas towers. Both the mobile and stationary platforms distort the ambient wind field. Ships in

324

RULDIP P. CIIOPRA

particular act as sources of convective and radiative heat. By generation of sea spray, waves, and wakes, these platforms add further to various complications in the measurements of marine atmospheric parameters. This section is devoted to the study of undisturbed wind field modifications caused by a fixed platform, the Argus Island. Studies of this nature are valuable in providing insight into the choice of locations for instrument installation for measurement of wind and temperature fields and moisture and heat fluxes. Argus Island (Fig. 2) is essentially a steel box platform of complex superstructures standing on four legs in 17.5 m of water. It is located at 31"57'N and 65"ll'W in the open Atlantic, about 45 km southwest of Tudor Hill Laboratory in Bermuda. The microwave tower rises 15 m above the deck and is based on Plantagenet Bank whose surface lies 60 m below mean sea level.

2.1. Prevailing Wind Field Near Argus Island Perrone et al. (1965) analyzed wind data gathered during the three year period 1962-64 to establish diurnal and monthly variations of prevailing winds in the vicinity of Argus Island. Their analysis is limited to this period and excludes the durations of hurricanes. However, the data show large dispersion which is attributed to frequent storms of less in!ensity. Their findings are summarized as follows: (a) The wind modifying influence of Bermuda is not observable a t the location of Argus Island. (b) No definite pattern of diurnal variation in wind speed is discernible. (c) The cumulative pattern of wind speed for each month is identical for each of the three years during which observatio,ns were made. Wind speeds, with the exception of those associated with tropical storms, range from 0 t o 28 m/s with the most frequent wind speed of 7.5 m/s. Winds are stronger during the winter months, with the highest monthly median wind speed of 10 m/s during February. May through September is the period of significantly lower wind speeds with 5 m/s being the monthly median wind speed associated with August. (d) Prevailing winds are predominantly southerly during May through August, northeasterly during October and November, and northwesterly during the remaining months of the year.

2.2. Disturbance of Ambient Wind Field Caused by Argus Island Thornthwaite and associates (1965)observed winds during the late summer and early fall of 1962 and 1963 to establish the characteristic perturbations caused in the ambient wind field by Argus Island. They used cup anemometer rotors, weighing 7 g each, with accuracy of 0.6 yo.Measurements were made during periods of light t o moderate southerly (125"t o 240"with north) winds

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

325

under conditions of neutral (adiabatic) stratification. This condition justifies the use of the logarithmic wind profile

(2.1)

u = (u*/4log,(z/Zo)

where u is the ambient wind speed a t height z, K = 0.4 is the von KBrmBn constant. zo is the friction parameter, u* = ( r / p ) l I 2 is the friction velocity with r and p being shear stress and density, respectively. Using subscripts i and T to identify ambient wind values a t a n observation point and Tower reference anemometer, respectively, it is easy to write for the undisturbed wind field (2.2)

%lUT

= [lO~~(zl/zO)l/[lO~~(zT/~O)l

In terms of the parameter k = uTtIuTwith the primed quantity representing the disturbed (measured) wind speed, the disturbed wind field is given by (2.3)

ui'/ut= ~ ( u ~ ' / u T ' ) / ( u ~ )

Values of k and zo were determined by trial and error to fit the observations. Eddy correlation analysis of the 1962 data led to the range of 0.001
k = 1.00 = 1.05

= 1.06 = 1.09

(for 125" < 8 < 147") (for 170" < e < 1920) (for 192" < 6 < 215") (for 2150 < e < 237")

The values of k = 1.05 and zo = 0.078 appeared to provide the best quantitative description of observed wind fields above and below the deck. For lighter winds (3.7 m/s < uT < 5.8 m/s), zo = 0.02 cm appeared to describe the appropriate logarithmic wind profile. The observed (disturbed) wind field around the Argus Island is sketched in Fig. 2. The logarithmic wind profile, expressed as a fraction of the undisturbed wind speed a t the Tower reference anemometer is indicated along the left ordinate. The corresponding heights are shown on the right-hand side. The numbers within a solid line identify the corresponding wind speed of the constant wind speed contour. The following are the results of the data analysis. (a) The disturbance field consists of two jets of accelerated air: one above the platform extending upward to 15-25 m above the deck, and the other below the platform extending down to the sea surface. There is also a region of decelerated flow immediately above the platform and a larger region of

326

KULDIP P. CHOPRA

reduced winds directly below the deck. For light winds, the accelerated jet above the deck is more compact, and the winds are more retarded near the deck. The regions of decelerated wind are due t o the friction of the deck's upper surface and the complex structural features of the platform underneath. The accelerated jets are apparently due to the airfoil and Venturi effects. (b) The flow of air on the windward side is sharply retarded as far as 15 m upwind of the deck with the probability of a stagnant region immediately ahead of the platform. The wind field is almost undisturbed beyond 18 m. The upwind nose of the retarded air is very asymmetrical about the horizontal. (c) A turbulent wake of greatly reduced wind speeds extends far downwind of Argus Island. Wind speeds lower than one half the corresponding ambient values are observed as far as 12 m leeward of the platform. (d) The ratio uf/uT upwind and above the platform is independent of general wind speed, but depends on the wind direction.

3. GROUPOF SMALL ISLANDS AS MESOMETEOROLOGICAL NETWORK

A very small flat island is not likely to modify the general atmospheric conditions by an appreciable amount; therefore, it may act as a probe for measurement of local meteorological conditions. However, meteorological measurements a t one island may not necessarily represent the entire mesoscale atmospheric flow conditions prevailing over the local geographical region. On the other hand, small islands forming an atoll may provide a network for meteorological measurements to study the detailed structure of the local mesoscale wind field. Analysis of simultaneous rawinsonde data (Billions, 1967) gathered at the Kwajalein and Roi-Namur islands is presented in this section to emphasize these points. 3.1, Kwajalein Atoll Kwajalein Atoll consists of 88 small, flat islands of sandy soil and coral formation. It extends over a geographical area of about lo4 kma with total land mass of less than 16 kma. The two largest islands are Roi-Namur (9"24'N, 167'27'E) and Kwajalein (8"44'N, 167"44'E), about 82 km apart (Fig. 1). Kwajalein Atoll has typical tropical climate with average temperature of 27.7"Cin the range of 23.4"Cto 31.3"C,and mean relative humidity of 77 yo with average annual precipitation of 260 cm (102 in.). Moderate easterly trade winds prevail with occasional speeds of 20-26 m/s. February and November are the driest and wettest months, respectively, and the skies are usually partly cloudy.

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

327

3.2. Analysis of Rawinsonde Data at Kwajalein and Roi-Namur Islunds Billions (1967) has analyzed the meteorological data gathered in 49 almost simultaneous rawinsonde releases from Kwajalein and Roi-Namur islands conducted during January 3 through March, 3, 1967. Analysis of the data points to significant variations in winds, temperature, pressure, density, and relative humidity at all levels above the two islands. The general characteristics of the differences in various meteorological parameters are listed in Table I. TABLEI. Differences in meteorological elements a t Kwajalein and Roi-Namur Islands

Meteorological elements Wind speed

Wind direction

Atmospheric layer

(km) Surface 0.0 to 6.1 6.1 to 30.5 Surface 0.6 to 6.1 6.1 to 30.5

Difference between observed values over Kwajalein end Roi-Namur

f 34 s f 9 mls f13 m/s -40" to 20" f60' f200"

Short period gusthess

21.3 to 24.4

Temperature

Surface Troposhere Tropopause Stratosphere

-5°C to f3"C -1°C to +3"C - 3 " C t O +11"c -4°C to +8OC

Density

Troposphere Tropopause Stratosphere

1% 6%

Pressure

Surface Troposphere Tropopause Stratosphere

-1.2 mb to +1.3 mb (2 % 2 Yo >2 Yo

Relative humidity

0.0 to 1.0 km

50 to 60 Yo

3 Yo

The general wind pattern above Kwajalein Atoll may be characterized in three layers. Easterly winds prevail in the lowest layer from the surface to an altitude of 18.3 km. The winds change to westerlies in the region above the easterly region with the top of the westerly layer varying from 25.6 to 28 km. Strong easterly winds blow in the region above the layer characterized by the westerlies.

328

KULDIP P. CHOPRA

With a few exceptional observationsto the contrary, pressure at Kwajalein is consistently higher than that at Roi-Namur. The difference in pressure at the two islands increases to about 2 yoin the tropopause, and to higher values in the region above. The corresponding difference in density varies from 1 % at the surface to about 6 % in the tropopause region and to about 3 % in the stratosphere. The differences in relative humidity (50 to 60 Yo),temperature (-lo to +11"C), wind speed ( k 1 3 m/s) and wind direction (f200") are of larger magnitude than those expected from the influence of small-sized islands. Whereas these large differences in meteorological parameters may partly be attributed to advection from separate sources such as convergence and divergence zones or perturbations in the prevailing winds in the vicinity of islands (Billions, 1967),they are apparently due to the fine structure in the mesoscale or to synoptic scale flows at various levels. Measurements conducted from severd islands in an atoll or from larger islands several tens of kilometers apart, and their respective vicinities, would shed valuable information on the detailed structure of global and mesoscale circulations that would be helpful in our understanding of the detailed mechanism which drives these flows. 4. MESOSCALEATMOSPHERIC VORTICES LEEWARD OF ISLANDS

The production of planetary scale eddies (cyclones) and of microscale (turbulent) eddies has been extensively studied. Intermediate scale (mesoscale) eddies of 10 to 50 km are too small to be delineated by the standard weather network. The average station to station distance in North America is about 150 km. These vortices are too large to be seen by an observer on the ground or even pictured from a high flying aircraft. The Tiros weather satellites, equipped with narrow angle cameras, were the first to reveal the existence of mesoscale vortex patterns. These patterns were found leeward of the Canary Islands. Tiros cameras, pointing vertically downward from an altitude of 600 km, could view an area 130 km in diameter. 4.1. Atmospheric Vortex Streets Atmospheric eddies leeward of islands (Figs. 3 to 6 and 8 to 11) are made visible by patterns in stratocumulus clouds lying beneath a strong inversion, about 0.5 to 1.6 km above the ocean surface. The eddies range in size from 10 to 30 km in wakes 40 to 60 km wide and 400 to 600 km long. The vortex patterns resemble the classical von K6rmhn vortex street (Fig. 12) consisting of two parallel rows of vortices such that a vortex in one row is situated across the midpoint of the two adjacent vortices in the other row. All vortices in one row have similar circulation, but of the sense opposite to that of the

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

329

vortices in the other row. These vortex patterns have been observed t o persist for several (18 to 30) hours. The atmospheric analogs of von Karman vortex streets are associated with steep sided islands that extend far above the low-lying inversion 1ayer. These patterns have been observed leeward of the Canary Islands (Tiros V and VI, Gemini VI and IX, Nimbus I11 missions), Cape Blanc (Tiros V mission), Cape Verde Island (Gemini V and VI missions), Guadalupe Island, Mexico (Gemini V, VII and XII, and Apollo 9 missions), and Cheju island (ESSA 7 and 8 missions). Except for the islands of La Reunion and Mauritius, all other islands are located in the northern hemisphere. Since Chopra-Hubert (1964) identified the Canary-Madeira vortices as the atmospheric analogs of the vortex street phenomenon, the literature reflects some confusion among meteorologists and fluid dynamicists regarding the understanding of the vortex street phenomenon and its possible existence in the Earth’s atmosphere. An attempt to clarify this situation is made in the following sections.

4.2. The Vortex Street Phenomenon : General Considerations The vortex street of Fig. 12 consists of a double row of vortices, each of strength K . I n most laboratory experiments, vortices are caused by flow past a bluff body under such conditions that they are shed alternately from each side of the body with regular frequency N . The rate at which vorticity is discharged t o the wake is

K=NK Tho vortex street geometry is characterized by the lateral street width h (called the bandwidth) and longitudinal spacing a (called the wavelength) between two successive vortices in each row. The ratio hla, called the spacing ratio, has an empirical range 0.28 < h/a < 0.50. Experiments indicate that h increases with distance downstream while a remains constant, thereby causing h/a to increase with distance downstream. The channel walls tend to reduce the spacing ratio by compressional effect. Once shed, the vortices move downstream with propagation speed u, which is somewhat lower than the ambient flow speed u.The velocity ratio E = u,/u has an empirical range 0.7 < E < 0.9. The difference u - u, in the two flow speeds accounts for the generation of vorticity. The resulting flow is periodic with frequency

(4.2)

N = u,/a

Vortex streets are also observed to have been caused by oscillating cylinders or plates, in jet and cavity flows, natural convection flows along vertical

330

KULDIF P. CHOPRA

plates, and in the atmosphere. Many natural phenomena are attributed to periodic vortex shedding (Birkhoff and Zarantonello, 1957). Audible notes emitted by wires in a wind or the whirring sound produced by a stalk or a thin rod swished rapidly through the air are connected with the frequency of the eddies discharged behind them. Smokestack vibrations, the phenomenon of “galloping transmission lines,” and the failure of the Takoma Narrows bridge have also been explained as related to the periodic vortex shedding. Resonance caused by vortex shedding can produce waves of significant amplitude in fuel storage tanks. Reciprocally, waves of an appropriate frequency propagating in an atmosphere or a medium can lead to the formation of a vortex street. A recent study by Lee (1972) shows that foreat fires may also generate a vortex street. Thus a vortex street may be regarded as a special kind of periodic or oscillatory flow phenomenon which results from the existence of an appropriate velocity profile in the general flow. 4.2.1. Von Karmdn Theory, Von Kh m h n (1911, 1912) proposed a now classic theory t o explain the vortex street phenomenon in which flow is assumed irrotational except for the double row of point vortices of strength K extending to infinity in a nonviscous fluid. The eddy propagation speed u, with respect t o the obstacle is then given by

(4.3)

u, = u - (~/2a)tanh(nh/a)

The vorticity discharge rate in the wake is (4.4)

K = N K = ~ u , ( u- u,)coth(nh/a)

The fluid dynamical drag D, being equal to the rate a t which forward momentum is generated in the wake, is given by (4.5)

D = phK = 2phu,(u - u,)coth(d/a)

where p is the density of the fluid. Von Khrmhn (1911) took into account the difference between the upstream and downstream mean pressures, and introduced an important concept by relating vortex street configuration to fluid dynamical drag D, obtaining (4.6)

+

D = (Kp/s?ra)[K 2 ~ h ( 2 ~ ,U)]

The coefficient of drag C, on unit length of a bluff body, defined by D/(ipu2d) has an empirical relation (cf. Birkhoff and Zarantonello, 1957) (4.7)

C,

2:

h/d 2: 1.26

for a circular cylinder of diameter d. For objects of different shapes, it has different values which also vary with the nature of flow.

ATMOSPHERIC AND OCEAN10 FLOW PROBLEMS

331

Equations (4.5) and (4.6) are equivalent only if (4.8)

coth(wh/a)= wh/a = 1.2

or

h/a = 0.38

This value of h/a is close to the median value of the empirical range for the spacing ratio. On combining Eqs. (4.4), (4.7)and (4.8), Chopra and Hubert (1964) obtained (4.9)

1.5nfue/u)( 1 - u,/u) = 1

or

u, = 0 . 7 ~

for this very special case of vortex street. Von KArmhn (1912) found that the double-row of vortices is unstable for two-dimensional infinitesimal displacements parallel to itself unless (4.10)

cosh(nh/a) = 42

or

h/u = 0.28

The value 0.28 for h/a has acquired the name “stable spacing ratio.” Substitution of this value in Eqs. (4.3) and (4.6) yields (4.11)

K

= 242u(u - u,)

and (4.12)

- u,)’/u’]

C D = (a/d)[1.59(~- u,)/u - 0 . 6 3 ( ~

The last two equations are useful for estimating vorticity and drag coefficient if N, u,and a are known, because the eddy propagation speed u,is determined by N and a. Conversely, knowledge of u,, u, and a would also provide an estimate of K and C,, . The vortex street studied by von K h m h is unstable for two-dimensional disturbances if the ambient flow velocity is not uniform and is greatest along the street axis. It is also unstable for three-dimensional disturbances with high frequency components dong the axis of the street resulting in increasing distortion of vortices downstream. Assuming conservation of vorticity and that the vortices are rolled up periodically by Helmholtz instability in a nonviscous fluid, Heisenberg deduced for the parameter (4.13)

h = K/u’

an a priori value of 0.5. The same value for h is obtained in the very special case of h/a = 0.38 and u,/u = 0.7 when these values are substituted in Eq. (4.4). The Heisenberg parameter determines the fraction of vorticity of each sign which escapes annihilation by mixing with opposite vorticity. Its empirical values are somewhat lower than 0.5. The above discussion provides a fair comparison between the von K h m h theory and the observed vortex street properties. The formulas are not rigorously correct for application to actual vortex streets. For example, the

332

RULDIP P. CHOPRA

calculation of drag [Eq. (4.6)] assumes a doubly infinite, rather than a semiinfinite, vortex street. Also, von K&rm&nregards the fluid to be ideal (nonviscous), and therefore, the vortex street consisting of point vortices extending t o infinity could be justified. I n contrast, real fluids are viscous, vorticity is not concentrated in points, and the actual vortex street wake is limited in extent. The role of viscosity in the vortex street phenomenon is discussed in the next section. Other complications are introduced by the channel walls and surface tension (in experiments involving observations of liquid surfaces).

4.3. Role of Viscosity in Vortex Street Phenomenon Viscosity plays a very important role in the formation and decay of a vortex street. Also, the characteristic properties of viscous vortices are quite different from those of ideal (point) vortices.

4.3.1. Formation of Vortex Street Behind a Bluff Body. Streamlines behind a bluff body form a region called the wake which consists of two confluent shear layers. The vortex street wake may be divided into three regions: the formation region, the vortex-street region, and the decay or turbulent region. Vortex street formation depends critically on the entrainment of the fluid from outside into the shear boundary layer region immediately behind the obstacle. There is a flow in the general direction of fluid motion outside this region, whereas within it there is an inflow along the axis of the wake. This circulatory motion constitutes a pair of vortices of opposite circulation located symmetrically about the wake axis. Being fed by vorticity generated in the boundary layer by the inflow of the fluid from outside, the vortex pair continues t o grow in strength and size. At the same time, vorticity is lost by diffusion across the vortex layers into the main body of the fluid. The equilibrium size and shape of the two vortices, which are of opposite circulation, is determined by the balance between the rates of generation and dissipation of vorticity. The accompanying pressure changes start waviness in the wake leading to concentration of vorticity in a periodic pattern. At very low values of Reynolds number (4.14)

R=~d/v

where v is the kinematic fluid viscosity, the prtir of vortices are circularcylindrical, symmetrical about the wake axis, and stable. However, a t somewhat higher values of R, approach of the oppositely directed vorticity in sufficient strength from across the wake axis causes the vortices to become elongated in the direction of general flow and take up asymmetrical positions. This assymmetrical arrangement alters the pressure distribution around the

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

333

body, and for values of R above a certain critical value, the symmetrical vortex pair does not form. Instead, the vortex layers which spring from the bluff body become unstable and roll up in such a way that vortices are formed and discharged alternately from each side of the body with a definite frequency N for each Reynolds number. Once shed, these vortices assume a regular pattern of a double row as in Fig. 12, beginning a t a distance of a few (four or five) diameters downstream of the body. Gerrard (1966)introduced two characteristic lengths which simultaneously determine the frequency N of vortex shedding: the formation length LF, which is the characteristic scale of the formation region in which fluid from one side mixes with fluid from the other side in close proximity to the obstacle, and the diffusion length L, , which is characteristic of the width to which the free shear layers diffuse. At still higher values of R, vortices are shed alternately, but diffuse so rapidly that the double row never forms. Whereas vortices continue t o be shed until R = 4 x lo5, no vortex street has been observed for R > 2500. Incidentally, the ideal vortex street studied by von K6rmBn would correspond t o R = co. The Reynolds number R emerges from the principles of dynamical similarity as the single dimensionlees parameter needed for the specification of the dynamical state of the flow field with geometrically similar boundary and initial conditions. Its magnitude provides an estimate of the relative importance of the inertial and viscous forces acting on a unit volume of the fluid. To account for the additional feature of the frequency N in oscillating wake phenomena, the dynamical similarity provides another dimensionless parameter, the Strouhal number

(4.15)

S = Nd/u

Hence, a vortex street behind a bluff body is characterized by the Reynolds number R and the Strouhal number 8. One can then write

(4.16)

N =( u / W ( R )

where f (R) is a function of R. The vortex street is most marked in the range

40 < R

< lo00

with

(4.17)

S = Nd/u=O.21(1 -20/R)

The corresponding range for the Strouhal number is 0.1 <S <0.21.

334

KULDIP P. CHOPRA

Following Birkhoff and Zarantonello (1957), this empirical information leads t o the periodic ranges 6 < ud

< 50

with

1.2da< N

< 30/d2

in air, and 0.4 < ud

< 10

with

0.08d2< N < 2/d2

in water. Here, all parameters are expressed in the cgs units. This explains why audible notes (100 < N < 1000) are heard in air when d N 0.6 to 3 mm and u N 1 to 10 mls, and why vortex streets are seen in water when d N 0.1 to 1 cm and u 1: 0.5 t o 6 mls.

4.3.2. The Vortex Street Region. Real vortex streets display the following properties in the laboratory: (a) increase in size, lateral spacing h, spacing ratio hla, and distortion of vortices as they propagate downstream; (b) disappearance of the vortex street pattern a t a certain location in the wake; (c) compressional effects of the channel walls; (d) wide range of hla in different experiments; and (e) decrease in h with increase in S and R . The wide range of hla suggests that the spacing ratio is determined by initial conditions rather than by the tendency t o a stable spacing ratio 0.28. Also, the breakup of the real vortices is not due t o the random wandering of point vortices as in the von K&rrn&nanalysis. The size increase, distortion, and ultimate disappearance of vortices and the apparent vortex street widening with downstream distance are caused by the diffusive spread of vorticity due to viscosity. a. DiJusive spread of vorticity: Vorticity

tion

r as

c = curl v is related to the circula-

where ds and dA are the elements of length and area, respectively. The vorticity vector satisfies the equation

c

(4.18)

dcldt = Div v

+ vVa<

If the first term on the right is neglected for application t o Q rectilinear vortex or to a layer with negligible divergence. Eq. (4.18) reduces to (4.19)

d(/dt = vVa(

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

336

Equation (4.19)is the well-known heat conduction equation or the diffusion equation. Following Schaefer and Eskinazi (1959), it has a solution

5 = loexp(-r2/4vt)

(4.20)

where C0 = ~/4.rrvtis the vorticity at the center ( r = 0) of the vortex at any given instant of time t after shedding. Equation (4.20) describes the diffusive spread of an isolated vortex of strength K as a function of time t and radial distance r from its center. If the instantaneous size of the vortex is defined as the radial distance rewhere 5 falls to l/e times its value 5,, at the center, then (4.21)

re= 2(vt)lI2

Equation (4.20) reveals that (a) the vorticity at the vortex center decreases inversely with time t , and (b) it spreads over a radius re= ( 4 ~ t ) lin ’ ~an interval of time t.

b. Tangential velocity projile: Tangential velocity v and circulation I? in a circle of radius r are related by r

(4.22)

which on substitution for (4.23)

2%

I?=2~k=2mv[ [ ird8dr JoJo

5 from Eq. (4.20)yields

v = (~/27rr)[l- exp(-ra/4vt)].

It may be noted that as v or t approach zero or r approaches infinity, the expression for the tangential velocity reduces to that for a potential vortex. The velocity v has a maximum value (4.24)

v,,,

= 0.1145~/r,

at a radial distance r, , called the vortex core radius, given by (4.25)

r, = ( 5 . 0 4 ~ t ) 2: l ’ ~l.lre

It follows from Eq. (4.25)that the vortex radius defined by the exponential diffusive spread of vorticity is slightly smaller than that defined by the peak tangential velocity within the vortex. Figure 25 shows the plots of (i) the tangential velocity field, given by Eq. (4.23), of an isolated viscous vortex of core radius r , (solid curve); (ii) the velocity v = ~ / 2 m of a rectilinear vortex (dashed curve); and (iii)the potential velocity field of a hard-core vortex of radius r,. It follows from this diagram that the velocity field of an isolated axisymmetric viscous vortex resembles that of a rectilinear potential vortex for r > r, . For small values of

336

KULDIP P. UROPRA

I

0

I

I -(

,

e

I

I

3

,

,

4

I

,

5

r/rc)+

FIG.26. Circumferential velocity field of an isolated vortex: viscous vortex of core radius r o , solid line; potential rectilinear vortex, dashed line; and potential cylindrical vortex of radius r e , dotted line (after Chopra, 19728).

r, v N ~](2r),and the viscous vortex approximates that of a hard-core axisymmetric potential vortex of core radius r, , within which the fluid may be regarded to rotate as a rigid body with uniform angular velocity equal to 42. c. Vortices in a Street: The above discussion on diffusive spread of vorticity and the profde of tangential velocity applies to an isolated vortex in a viscous fluid. This situation is slightly different for a double row of vortices. Vortices normally display cylindrical symmetry in the range 40 < R < 60. Outside of this range, and particularly further downstream in the wake, they appear slanted or distorted. Also, the street appears to widen along the wake. Hooker (1936) attributed the widening of the vortex street to the effect of viscosity. In an isolated viscous vortex, velocity at its center falls to zero as soon as diffusion starts. However, due to the effect of other vortices, it is zero at a point further away than the vortex center from the axis of the vortex street. As usual measurements identify the vortex center by point of zero velocity, the lateral width h appears to increase with distance downstream. After allowance was made for this effect, Hooker found h approximately constant in the wake closer to the obstacle, but an increase in h further downstream could still not be explained. The clue to the puzzle may perhaps be found in the fact that the vorticity is also not distributed symmetrically in any vortex. It is concentrated in that portion of the vortex nearer to the axis of the street than at the point of zero velocity. This is because the vortex street axis acts as a virtual barrier,

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

337

compressing the streamlines on either side of the wake axis and inducing a small transverse velocity in each vortex. For these reasons, measurements, especially those of h, are difficult in a real fluid.

4.3.3. Disappearance of a Vortex Street. I n view of the foregoing discussion, it follows that after the vortices have been shed, their size increases and strength decreases as they propagate downstream. When the radius of a n eddy equals one half of the bandwidth of the vortex street, i.e., (4.26)

r = h/2,

it begins t o overlap the region of opposite circulation. This provides the criterion (Lin, 1959; Chopra and Hubert, 1964) for the onset of a process in which the vortex street disappears through mutual annihilation by mixing of opposite vorticities. This marks the beginning of the turbule,nt region of the wake.

4.3.4. Vortex Street as a n Oscillating Wake; Lin Parameter. Perhaps the true explanation of the vortex street phenomenon lies in its being an oscillator with frequency determined by local effects. Two alternative treatments (Birkhoff and Zarantonello, 1957) predict S = 0.2 and h/a = 0.35 within the corresponding empirical ranges. I n his discussion on oscillating wakes, Lin (1959) brought out a parameter (4.27)

/3 = vN/U2

as an inherent property of the vortex street. This parameter turns out to be the ratio of the Strouhal number S and the Reynolds number R, and is independent of the cross-stream diameter of the obstacle. Lin argued that, if one made an experimental observation of the vortex street without paying attention to its source, the only independent physical parameters are v , N, and u, and the only non-dimensional parameter which can be formed of these is /3. It is fortuitous that the cameras aboard the Tiros satellites photographed vortex patterns of Figs. 3 and 4 under just these conditions when the generating causes (islands) of the atmospheric vortex streets were out of the cameras’ field of view. Chopra (1972~)lists the following reasons in support of being a better characteristic of the vortex street phenomenon than S and R: (a) Precise range of /3 corresponding to vortex s t r e e t s S t a b l e vortex streets behind circular cylinders correspond to the range (Lin, 1959) (4.28)


<2.5 x

lod3

Values of R corresponding to the observed vortex streets extend over a much broader range.

338

KULDIP P. CHOPRA

(b) Nondependence of /3 on d-Vortex streets have been observed to form behind obstacles of all sizes and shapes, ranging from wires through large structures to islands. Therefore, the existence of the vortex street phenomenon should not depend critically on obstacle size. Whereas S and R depend on the size of the object generating the vortex street, Lin parameter /3 is independent of this linear scale. (c) Vortex streets are associated with appropriate velocity profile in the fluid flow-It is the existence of the appropriate velocity field rather than the presence of a n obstacle to ambient flow which is essential for vortex street formation. The significance of the Lin parameter lies in exactly rendering the description of vortex streets independent of the linear scale of the producing source. (d) Length of vortex street limited by P-The ratio of the vortex radius r [at a distance na (where n is a number) downstream of the obstacle] to the vortex street halfwidth h/2 is given by (4.29)

./(hl2) = 4(a/h)(4Ue)(n/3)1’2

The vortex street would begin t o disappear when r > h/2. Since the spacing ratio h/a determines the velocity ratio [of. Eq. (4.3)],it follows from Eq. (4.29) that for a given value of hla, the vortex street length, characterized by na, is determined by /IIf .we adopt the median values /3 = 1.7 x and hla = 0.38, then r begins to exceed h/2 when n z 6, and the vortex street would disappear further downstream.

4.3.5. Concluding Remarks. The above description of the vortex street phenomenon is limited to two-dimensional flow. The existence of a n appropriate velocity profile, rather than the size of a n obstacle, is essential for vortex street formation. The vortex street phenomenon may be regarded as a special case of oscillatory wakes in which the frequency N, and not the spacing ratio h/a, has an instability. The frequency N is determined by local conditions. There is an empirical range of values for the spacing ratio associated with vortex streets, but the actual value of hla is determined by initial conditions. Vortex street geometry is closely related t o fluid dynamical drag. Viscosity plays a very important role in all aspects of real vortex streets, particularly in the birth and death of vortices. Finally, the Lin parameter /3 provides a more appropriate criterion for vortex street formation than the Reynolds number R and the Strouhal number S. For further discussion of vortex street phenomenon, the readers may refer t o the extensive literature on the subject, particularly t o descriptions by Birkhoff and Zarantonello (1957), Goldstein (1965), Lamb (1945), Milne-Thompson (1950)) and Berger and Wille (1972).

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

339

4.4. Analysis of Atmospheric Vortex Streets Leeward of Islands

Two aspects need to be considered before the empirical-theoretical results of the laboratory-scale vortex street analysis can be applied to islandinduced atmospheric vortex streets. One of these concerns dynamical similitude, and the other deals with stratification of the atmosphere and whether the atmospheric phenomenon is predominantly two dimensional in nature.

4.4.1. Dynamical Similarity. It follows from the preceding sections that a vortex street is characterized by the dimensionless parameters hla, u,Iu, CD E h/d, h, S , R, and @.The empirical values of these parameters are 0.28 < h/a < 0.50 0.15 < S ( 0 . 2 1 10-3 < p <2.5 x 10-3

0.7
and

CD = 1.25

(circular cylinder)

For similar geometrical configurations, initial boundary conditions, and flow characteristics,these parameters must have identical values in the laboratoryscale and atmospheric-scalevortex street phenomena. All these parameters are not completely independent. For example, CD is related to h/a and u,/u by Eqs. (4.5)-(4.7).Similarly, ji? is the ratio of S and R, and viscosity enters in R and p. The spacing ratio hla, Lin parameter p, and length L of the vortex street are related by [of. Eq. (4.29)for r = h/2] (4.30)

L = (l / 1 6 ~ N ) ( h / a ) z ( ~ , / ~ ) z ~ ,

If the ambient flow speed u, one characteristic length, and one of the dimensionless parameters are known, all the others can be estimated with the aid of the formulas described in the preceding sections. Caution is needed in the choice of parameters which are measured directly or treated as known. For example, the drag coeficient CD depends on obstacle shape and is known only for certain well-defined regular shapes. Islands in general have irregular shapes. Except in the case of vortices associated with Cheju Island, no direct measurements of u, and N were possible, hence any value assigned to S would be arbitrary. Adoption of a specific value for R would require knowledge of atmospheric friction which in itself poses a special problem to be discussed shortly. Choice of R or S is further complicated by the determination of the value of the cross-wind diameter of the island. Lastly, the vortex patterns are made visible in the background of low-lying stratocumulus clouds. Although these clouds are not essential to vortex street formation, they act as tracers for the vortices. Therefore, the

340

KULDIP P. CHOPRA

degree of cloud cover in the vortex street region would have a bearing on the detection and accuracy of measurements on vortices in the Tiros and Gemini photographs. Also, the intensity of a real viscous vortex decreases with distance from its center, and this would introduce an element of error into vortex size or location measurement. Various investigators have chosen different parameters as known. Chopra and Hubert (1964, 1965b) assumed the theoretical spacing ratio h/a = 0.38, [of. Eq. (4.8)] as the known value, because the approximate estimates of h and a from Tiros photographs of the Canary and Madeira eddies yielded values close to 0.4. Hunt and Wickins (1967) deduced u,/u from two different expressions for drag. Zimmerman (1969) assumed hla = 0.38 and /3 = in his analysis of the Canary and Madeira vortices. Tsuchiya (1969) was able to measure u, and a from ESSA 7 AVCS and ESSA 8 APT photographs, and hence, he treated the Strouhal number 8 as the known dimensionless parameter. It is well known (e.g., see Long, 1959; Sutton, 1960) that among the numerous diEculties which arise in applying results of laboratory model experiments t o meteorological problems, an important one concerns viscosity. It is not possible to obtain equality of Reynolds numbers in meteorological phenomena and corresponding laboratory scale studies if one uses molecular viscosity. Typical values are R = 101O-lO1ain the atmospheric flow problems as compared t o R = 10-104 in the laboratory. It is possible to obtain comparable Reynolds numbers by introducing the concept of eddy viscosity to replace molecular viscosity in the equations of motion. Since the fluid-dynamical phenomena of interest are laminar in the laboratory and turbulent in the atmosphere, the mean motions in the laboratory must correspond t o the mean-mean motions in the atmosphere. The Navier-Stokes equations describe laminar fluid motion in the laboratory model studies, whereas atmospheric fluid dynamics is governed by the Reynolds equations with both molecular friction and Reynolds stress terms. Except in close proximity to solid boundaries, atmospheric molecular viscosity is usually negligible compared to the Reynolds stress terms. The latter take the same form 6s molecular viscosity in the Navier-Stokes equations. Viscosity due to Reynolds stress terms is called the eddy, virtual, or dynumical viscosity. However, this viewpoint involves the assumption bf the mixing length theory of turbulence with constant coetlicient of eddy viscosity. There is some analogy in the role played by molecules in laminar flow and by eddies in turbulent flow. An eddy, regarded as a mass of fluid, retains some measure of individuality and of motion as an entity, and plays the role of a molecule in laminar flow. This analogy, however, is crude, and the difference between a molecule and an eddy is much more than one of scale. For

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

341

example, there is a spectrum of intensity and size of eddies for any given degree of turbulence. Also, there is a distinct contrast between the natures of eddy and molecular viscosities. Molecular viscosity is a property of the medium and has a precise value. Eddy viscosity, on the other hand, is a property of the flow conditions, and is independent of the nature of the medium. It has no precise value. For example, Heffter (1965) found experimental values of eddy viscosity in the atmosphere to range from 10 ma/sto lo6 m2/s. 4.4.2. Two Dimensional Nature of Atmospheric Vortex Streets. Because of the manner in which the satellites photographed the vortices behind the islands, the observed features are two dimensional. Hubert and Krueger (1962) had attributed the long life of vortices to geostrophic wind instability with the vertical component of flow in the,neighborhood of the island serving as an obstacle. On the other hand, Chopra and Hubert (1964) explained the phenomenon as a result of the horizontal transport of momentum with implications concerning the vertical momentum transport. They emphasized the importance of a low inversion layer and the accompanying stable stratification of the atmosphere in the formation of these vortex patterns. Chopra and Hubert (196.h)) Zimmerman (1969), and Tsuchiya (1969) have provided meteorological data in support of the existence of low inversion layer and stable stratification, contrary to the remarks by Berger and Wille (1972). Two-dimensionalfeatures of the vortex street would be augmented if the air above the island would act as a Taylor column. No particle of fluid crosses the boundary of this column, which so far as the fluid outside the column is concerned, behaves as a solid body. According to Hide (1961), a Taylor column can form as a consequence of the Proudman-Taylor instability resulting from disturbances caused by slow and steady streaming motion past objects immersed in otherwise uniformly rotating fluids. Vortex streets are made visible by stratocumulus clouds beneath the inversion layer. To generate these vortex patterns, an island must extend well above the inversion layer. Subsequent data tend to associate the formation of vortices with steep-sided islands extending well above the inversion layer. Table I1 provides the geographical data on islands with associated vortex streets analyzed in this article. Horizontal momentum transfer in the vortex wake is predominant in the stable atmosphere above the inversion layer, and the vertical transfer of momentum may be more significant in the unstable air below the inversion. Whereas vortices may be regularly formed and shed behind islands under conditions of unstable stratification below the inversion layer, the vortices may dissipate their energy, momentum, and vorticity to the ocean surface, and a vortex street may not result at all in that region. A striking example of such a case will be presented in Section 5.

342

KULDIP P. CHOPRA

TABLE11. Geographical data on islands with associated vortex streets Island

Peak elevation (km)

Longitude

Canary Islands: Gran Canaria La Palma Madeira Tenerife

1.95 2.42 1.88 3.72

15'30'W 17" W 17' W 16"30'W

Cape Verde Islands: Fog0 Santo Antao Cheju Island, Korea Guadalupe, Mexico La Reunion

2.8 2.0 2.0 1.4 3.1

24'18'W 25'6' W 126"30'E 118"30'W 55'30'E

Latitude

28" 29"

N N

37'45" 28"30'N

14'48" 17'1 2" 33"20'N 29" N 2 1'1 2's

4.4.3. Tiros-Revealed Vortex Patterm Associated with Canary Idand8. Figure 26 illuBtrates the geographical complex of the Canary Islands. The contours inside the sea-surface boundaries correspond to an altitude of 0.30 km. The dot inside each island indicates the location of the peak with numbers corresponding to the peak altitude. The island of Madeira (not shown in Fig. 26) is located north of La Palma Island. The initial analyses by Chopra and Hubert (1964, 1965b) of the vortex patterns of Figs. 3 and 4 associated with these islands were directed a t identifying these patterns as the atmospheric analogs of the vortex street

LO k l m o

!

-0Gmn Conarla

- i a1 3091

Seala = I;ZpoOpoo

USAF-

JN Chart

~

~~

~~

FIQ.26. Geographical area comprising the Canary Islands.

343

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

phenomenon. No meteorological data on the vertical atmospheric stratification was readily available a t the time of that analysis. The vortex pattern of Fig. 3 was interpreted to have been caused by Madeira Island, and that of Fig. 4 was explained as consisting of two vortex streets, generated by the islands of Gran Canaria and Tenerife, respectively. For reasons given in the preceding section and in Chopra and Hubert (1964),they assumed h/a = 0.39 and adopted the measured value of the ambient wind speed u, the surface value of the cross-stream diameter d of the islands, the diagonal distance 1 between two adjacent vortices of opposite circulation as measured from the Tiros V photograph for each of the streets in the complex pattern of Fig. 4 as known parameters. The value of ,5 was determined from the location L = nu of the discernible end of the vortex street. With the assumption of h/a= 0.38and hence of u,/u = 0.7,Eq.(4.30)becomes

(4.31)

,5 = (1/16n)(h/~)'(~,/u)' = (4.8/n)10-3

TABLE111. Values of the vortex street parameters (Tiros V and VI data)

Parameter

Madeira Island 0.38" 0.7 10.0b 7.0 1.12 0.68 1.8 3.9 6.6 3.0 22 1.6 4.4 0.46" 0.17 1.05 12.6 0.49 24.0 6.0 40.0 3.6

" Assumed. Measured. urn-, youngest eddy. re, L , urnax,L , oldest eddy.

O r e . 1,

Vortex streets leeward of Gran Canaria Island 0.38" 0.7 7.6 6.4 0.99b 0.61 1.6 3.4 6.6 4.0 34 1.2 2.1 0.63" 0.18

1.6 8.6 0.6

32.0 3.2 34.7 2.9

Tenerife Island 0.38" 0.7 7.Bb 6.4 1.04 0.64 1.7 3.2 7.0 4.0 36 1.2 2.2 0.45" 0.22 1.83 9.0 0.6 29.2 3.7 36.7 2.9

344

KULDIP P. CHOPRA

With this basic information, all other parameters can be readily calculated. The Chopra-Hubert (1966b) analysis of the vortex wake associated with Madeira Island proceeds on exactly similar lines, except for the measurement of lateral street width h rather than diagonal separation 1 from Fig. 3. The results are listed in Table 111. It may be noted that the values of the various parameters compare well with the corresponding empirical values. Figure 27

X

3 0.24

0.22 L 3

0

A

X

3

2

s

FIG.27. The Roshko curve relating Strouhal number S and Reynolds number R with data points corresponding to vortex streets leeward of islands: A Chopra-Hubert (Tiroe), 0 Chopra (Gemini), x Hunt-Wickins (Tiros), 0 Zimmerman (Gemini), 1. Madeira Wake, 2. Oran Canaria Wake, 3. Tenerife Wake.

is a graph, called Roshko (1954) curve, of Eq. (4.17) which provides a satisfactory empirical relation between the Strouhal number S and the Reynolds number R correeponding to vortex street appearance. The computed values of S and R for the vortices associated with the Gran Canaria and Madeira Islands appear to agree with this.

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

345

TEMPERATURE

Fro. 28. Radiosonde and surface obervations for 1200 GCT, May 19, 1963 made at Funchal, Madeira (after Chopra and Hubert, 196th).

A subsequent article by Chopra and Hubert (1965a) elaborated on the implicationsof the various physical parameters involved in the analysis. They estimated average values of h and a from the photograph and obtained h/ii=0.43. I n view of the comments in the preceding section, this result should be accepted with some reservations. From the surface and radiosonde observations shown in Fig. 28, they determined the height of the inversion layer, and adopted the value of cross-stream diameter of the island at that height in their analysis. Corresponding to the range
(u,/u)[l - (u,/u)] = 0.25 tanh(nh/u)

which yields u,Iu E 0.65-0.69, values outside (somewhat lower than) the empirical velocity ratio range. Also, their computed values of S and R do not

346

KULDIP P. OHOPRA

conform to the Roshko curve (Fig. 27). Their estimate of the Heisenberg parameter X is slightly higher than = 0.6 corresponding to the nonviscous vortex streets. If similar calculations are made by adopting appropriate values of N,K , and u from Table I11 based on the Chopra-Hubert analysis, the results (also listed in Table 111)are more realistic, A 7 0.5. Hunt and Wickins provide support for the viscous-diffusive process as against the role of instability in spacing ratio leading to the decay of atmospheric vortex streets. Following the first-order perturbation analysis, the displacement 8(t)at a time t (in hours) of a vortex in an infinite vortex street may be approximately written as (4.33)

8(t)= S(O)exp(O.O4t)

for the vortex street associated with the Gran Canaria Island. If the stability criterion hlu = 0.28 applies, then at t = 0,

S(0)= B(h - 0 . 2 8 ~21) 15 km and a t t = ~ =34 h, S ( t ) N 58 km. Under these circumstances, a vortex street would break up much sooner than it does in fact. However, if one considers the motion of a double row of about eight eddies, the resulting lateral displacement of an eddy, determined from the complex potential

n 3

(4.34)

W ( Z ) = (iK/zT)lOg

(Z

-nu)[% - ih - ( n f *)a]-’

n=O

is initially linear, and at t = T = 34 h, 8(7)N- 36 km. Applying the results of Hunt (1961), Hunt and Wickins estimate that the coastline of Africa would influence u, and K by less than one part in one hundred. This effect is insignificant by itself and would be further reduced at the height of the inversion layer or above. The Chopra-Hubert (1965a) estimate of 88 knots peripheral wind speed in a Madeira eddy was admittedly excessive. The wind speeds associated with other younger eddies in the Madeira trail would be alarmingly high on the premise of the Chopra-Hubert calculations. Chopra (1966, 1972~)discussed the inadequacies of the Chopra-Hubert (1966a) calculation of wind speed in the Madeira eddy, attributing the discrepancy to two causes. (1) There is an omission of a factor of (1/2n) in Eq. (12) of Chopra and Hubert (1965a)which is an expression for circulation I’=27r~.Accordingly, the vorticity 5 and the corresponding value of circumferential wind speed v would have to be reduced by a factor of 1121~. Wilkins (1968) also points to this discrepancy which when corrected would reduce the estimate of v for the Madeira eddy to 14 knots. (2) The ChopraHubert calculation of peripheral wind speed is based on the eddy being treated as a potential rectilinear vortex. Experiments of Schaefer and

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

347

Eskinazi (1959) reveal that the vortices of 8 real viscous vortex street resemble closely the solution of the Navier-Stokes equation for an axisymmetric vortex. The results of this theory are contained in Eqs. (4.22)-(4.25). Using these equations, Chopra (1972~)obtained estimates for the range 2.9 < v,, < 6 for the peak tangential wind speed, in knots, in vortex wakes of Figs. 3 and 4. These values are also listed in Table 111. In summary, the above analysis of atmospheric eddy patterns leeward of the Canary Islands revealed by Tiros photographs leads to the following conclusions or predictions. (a) Observed mesoscale eddy patterns are the atmospheric analogs of the vortex street phenomenon observed behind bluff bodies in the laboratory. These patterns are formed by the interaction of the wind with the islands under conditions of atmospheric stratification. (b) Eddies are formed and shed alternately from each side of the island at the approximate rate of one every 8 h. (c) Once shed, eddies propagate downstream with speed about seventenths of the ambient wind speed. (d) Eddy life time ranges from 20 to 34 h. (e) The estimate of eddy viscosity, 2.1 x lo3 < v(m2/s)<4.4 x lo3 falls within the acceptable range of 10 to 106ma/sfor the atmosphere. (f) If conclusions (a) and (e) are correct, then the loss of momentum to the ocean surface from eddy patterns above the inversion layer is negligible. (g) Values of dimensionless parameters u&, A, 8,R, and fi are within their respective empirical ranges. However, h and fi appear to show more satisfactory agreement. (h) If the Canary eddies persist for more than 20 h, then meteorological conditions must remain stable for at least that interval. This would suggest that if an Earth-synchronous satellite observed such en area continuously, vortex-pattern time changes would provide a very sensitive indicator of the changing meteorological conditions.

4.4.4. Qemini Data on Vortex Patterns Associated with Canary I s l a d . The Gemini 6 spacecraft crew took several photographs of cloud formations in the vicinity of the Canary Islands. Figure 6 is a typical photograph taken on December 16, 1965. Zimmerman (1969) analyzed a mosaic of some of these photographs which revealed vortex trails similar to those of Figs. 3 and 4, discussed in the preceding section. Zimmerman paid considerable attention to the vertical structure of the atmosphere. The analysis of synoptic reports confirmed the presence of cumulus and stratocumulus clouds over Tenerife Island with the cloud base at 0.6-1.0 km altitude. The radiosonde observations showed a fairly strong inversion at approximately 1.6-2.0 km height above the sea surface. The

348

KULDIP P. CHOPRA

temperature increase through the layer was about 3°C providing rr vertical temperature gradient of +7.5”C/km. I n view of this information, Zimmerman assumed the crosswind diameter d of Gran Canaria and Tenerife Islands to be about 75 % of the corresponding sea-surface values. For Madeira Island, he assumed the value adopted by Chopra and Hubert (1965a,b). His analysis proceeds on lines very similar to those followed by Chopra and Hubert, in addition to h/a = 0.39. As pointed except for his assumption of p = out by Chopra (1972a,c), these two simultaneous assumptions render the analysis very restrictive and force the results to a very special situation. The results obtained by Zimmerman are given in Table IV. Also shown in this in Zimmerman’s table are the results omitting the restriction of p = analysis. are obtained if the Significantly different values for v, R, S, r,, and ,v is dropped. The revised (Chopra) values of /3, v, and rc restriction /3 = are higher than those obtained by Zimmerman, and the corresponding values are smaller. The new values of p are obtained on the basis of the of v, estimated wake length L. Also, the values of r under columns marked “ Chopra ” refer to the characteristic lifetime of an eddy in a wake, whereas, those under columns marked “Zimmerman” refer to the age of the oldest and discernible eddy. These differences arise because /3 K v , r, cc v, DC r,- l . The values of AS and R are also plotted on the Roshko curve (Fig. 27). The values of AS, R, and v fall in the ranges 0.15
TABLE IV. Values of vortex street parameters (Gemini 6 data) Vortex streets leeward of Gran Canaria Island

Madeira Island Parameter

Zimmerman

h/a a u./u

u(m/s) u.(m/s) E( lo4m) h( lo4 m) a(lo4 m) N ( 10-5 s-1) L = na(105m) 12

T ( 104 5)

p( 10- 3, v( lo3 m2/s) d(104m) a

R S ~ ( 1 m2/s) 0 ~

h rc,d 1 O 4 m)

l(m/s) pc. A104 m) ~ m * x dm/s) . %lax.

a

Assumed. Observed or estimated.

0.39 0.71 7.0 5.0 13.7 8.5 21.7 2.3

-

-

12.3 1.O" 2.2 4.6 150.0 0.15 10.3

-

3.6 3.3

Chopra

0.39 0.71 7.0 5.0 13.7 8.4 21.7 2.3 6.5 3.0 13.2 1.6 3.4 4.6 98.0 0.16 10.6 0.5

-

-

4.5 2.6

Zimmerman

0.39 0.71 7.0 5.0 8.4 5.2 13.2 3.8

-

5.4 1 .O" 1.3 3.0 170.0 0.17 6.3

-

1.1 11.0 1.8 4.0

Chopra

0.39 0.71 7.0 5.0 8.4 5.1 12.9 3.8 3.9 3.0 7.8 1.6 2.1 3.0 100.0 0.16 6.3 0.49 1.4 5.3 2.3 3.2

Tenerife Island Zimmerman

0.39 0.71 7.0 5.0 7.7 4.8 12.2 4.1

-

5.6 1.oa 1.2 3.6 210. 0.21 5.8

-

1.2 5.5 1.8 3.6

Chopra

0.39 0.71 7.0 5.0 7.7 4.8 12.3 4.1 3.7 3.0 7.4 1.6 1.9 3.6 133.0 0.21 6.0 0.50 1.4 4.4 2.4 2.4

360

KULDIP P. CHOPRA

man wind Flow 4

--

_____RIDGE AREAS A

0

PEAKS (height i n km) STATIONS

FIG.29. Tenerife Island with general direction of prevailing wind.

c

@

c-

_ _ _ - - - - --__.

.

STATIONS

I

/

Y'

\

J-

FIU. 30. Mean-wind vectors (800-mb) showing changing trajectory. 0 0000 GMT December 15, 1965, 0 1200 CMT December 15, 1965, x 0000 GMT December 16, 1965, - - -1200 CMT December 16, 1965 (after Zimmerman, 1969).

35 1

ATMOSPHERIU A N D OCEANIC FLOW PROBLEMS

path along the wind direction, because, otherwise these would have passed directly over the Canary Islands. This is in agreement with remark (h) of the preceding section, repeatedly mentioned by Chopra and Hubert (1964, 1965a,b). I n short, the Zimmerman analysis of the Gemini photographs provides support for the basic Chopra-Hubert hypothesis that the Canary Islands generate vortex streets in stable layers of the atmosphere. 4.4.5. Vortex Street Leeward of Cheju Island. ESSA 7 and 8 satellites have provided an unusual opportunity to study the vortex wake downwind of Cheju Island, to confirm the Chopra-Hubert hypothesis regarding the existence of the atmospheric analogs of the vortex street phenomenon, and to verify some of their predictions. Cheju Island (Fig. 31) has an approximately elliptical cross section in the horizontal plane. It is centered a t 33"N and 126'30'E with a peak elevation of about 2 km (Fig. 32). A study of daily ESSA 8 APT and ESSA 7 AVCS

r34N

30'

I

1

3 4 N-I

I

b

10

20

30

40

I

5OLm

-

CHEJU ISLAND A 1950 m 47184 (CHEJU) -30'

-47187 ( MOSULPO) UPPER AIR 0BS.ST.

I27E

l26E L33N

-

I

1

1

3-0'

1

3 3 4

FIQ.31. Relief map of Cheju Island with locations of radiosonde and surface observation stations (after Tsuohiya, 1969).

362

KULDIP P. CHOPRA

Km

S H A R P INVERSION

\

MODERATE INVERSION WEAK INVERSION

CHEJU ISLAND 10

20

30

40

50

60

70

80

FIQ. 32. Cross section of Cheju Island normal to NNW wind (after Tsuchiya, 1969).

pictures revealed that the vortex street associated with Cheju Island is a seasonal phenomenon. Under favorable atmospheric conditions, the vortex street appears from autumn to spring. Figure 11 is a typical AVCS photograph taken on March 6, 1969. Tsuchiya (1969) analyzed several of these photographs and directly measured several of the wake parameters indirectly determined in earlier studies of vortex trails associated with the Canary Islands. The ESSA weather satellite photographs reveal clear and well-defined vortex streets extending to a distance of approximately 600 km downwind of the island. The regional winds are generally NNW or N, and the general air flow curves anticyclonically near Okinawa Islands. This change in wind direction is reflected in the orientation of the vortex street. A great advantage in the combined use of ESSA 7 and 8 satellites has been in the periodic surveillance of the area. The two meteorological satellites were only 4 h apart. With the help of these observations, Tsuchiys obtained averaged values for the geometric parameters of the vortex street h = 36.8 km, 6 = 110.8 km, and h/a = 0.332 and for the eddy propagation speed u, = 7 m/s. Combined with the observed ambient wind speed u = 9 m/s, this yields E = u,/u = 0.78. With these parameters determined, it readily follows that the vortex shedding has a period of 4.4 h corresponding to the frequency

N = uo/6= 6.3 x

8-l

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

353

and the vortex strength, determined from Eq. (4.3),has a value K

= 5.67 x lo5 m2/s

Such an eddy, 51 km across, would have a peak tangential wind speed of 2.54 m/s. Basic information on 6,d, u, , and u provided by Tsuchiya (1969) has been used in arriving a t the above numerical results for u,/u, K , and v, . The corresponding values obtained by Tsuchiya are u,/u = 0.76, K = 5.12 x lo5 m2/s, and v , = 3.2 m/s. Tsuchiya has reported the following meteorological features of the area based on the analysis of the surface and the radiosonde data of March 5, 1972. (a) There is a thin isothermal layer above the lowest layer characterized by dry adiabatic lapse rate, whereas in the upper part there are three inversion layers. As shown in Fig. 32, there is a weak inversion a t an altitude of about 0.9 km, a moderate inversion a t about 1.2 km height, and a very sharp inversion a t about 1.5 km above seal level. (b) Existence of a n inversion a t about half the height of the mountain is one of the necessary conditions for the formation of vortex streets. (c) The clouds revealing the vortex wake are stratocumulus. Vertical distribution of dew point suggests 200 m thickness of this cloud layer. To determine S, R, /3, v, and C,,, Tsuchiya (1969) assumed d = 28.3 km, the average cross-wind diameter of the island between the sea surface and the height (1510 m) of the uppermost inversion. The following discussion tends to show that one must exercise extreme caution in the choice of the cross-wind diameter in analysis of the atmospheric vortex streets. s-l, d = 28.3 km, and u = 9 m s - l in Substitution of N = 6.3 x Eq. (4.15) yields S=Nd/u=0.198 compared to S=O.194 obtained by Tsuchiya. This difference, even in the third decimal place, in values of S is quite significant as far as the determination of R, b, and v from the Roshko curve is concerned. A reference to this curve, Fig. 27, leads t o the corresponddence between S and R, and the accompanying influence on t!? and v, shown in Table V. Tsuchiya’s values of S = 0.194 and R = 250 give him v = 1.05 x lo3 m2/s. Also, one should be careful in drawing conclusions regarding S and R from mere similarity between the atmospheric vortex street pattern and the laboratory photographs of vortex streets given in Homann (1936). A more serious question concerns the averaging of d from sea surface to the level of inversion. As explained in previous sections, the atmospheric air displays vertical instability below and stability above the inversion layer, and it is the stable stratification which is more conducive to the formation of vortex streets extending to great distances downwind of the island. Vortices formed or shed below the inversion layer are likely t o transfer their strength to the ocean by vertical transfer of momentum. Contrary t o remarks by

364

KULDIP P. OHOPRA

TABLEV. Influence of Strouhal number 6' on Reynolds number R, Lin parameter

B, and eddy viscosity v, using Roshko curve

Strouhal number Reynolds number Lin parameter Eddy viscosity

S R

0.190 210 9.0 x 10-4

B v (male)

1.10 x 103

0.194 260 7.5 x 1 0 - 4 0.98 x 103

0.198 350 6.7 x lo-*

0.92 x 103

Tsuchiya (1969) and Berger and Wille (1972), a more appropriate value of d would correspond to the height of inversion or above. Since, the observed vortices are made visible by the stratocumulus clouds just below the weak inversion layer, it would appear more appropriate to adopt d = 20 km, corresponding t o this level. Then

8 = 0.14, fi = 2.3 x

R = 60

(from Roshko curve) v = 2.94 x lo3m21s and

which are in better agreement with the earlier discussion on vortex streets in general, and with Lin criterion and estimates of atmospheric eddy viscosity leewards of islands in particular.

4.4.6. Formation of a Taylor Column. Hide (1961)invoked the application of the Proudman-Taylor theorem to explain the Great Red Spot of Jupiter as a fluid dynamical phenomenon. Proudman (1916) and Taylor (1923) have shown that slow, steady, and incompressible motion in a rotating system is two dimensional, i.e., independent of the direction in which hydrostatic equilibrium is maintained. Should the flow produced by a solid object moving through the fluid satisfy this theorem, the object would carry a column of fluid with it. The cross section of the column corresponds to the projection of the object on the plane perpendicular to the axis of rotation. The fluid motion within the column moves with the object while the fluid outside the body flows around it. Hide (1961) gave this column of fluid the name Taylor column. More recently, Stone and Baker (1968), Ingersoll (1969), Hide (1971), Moll2 (1971),Chopra (1972b), and Berger and Wille (1972) have argued the pros and cons of the existence of Taylor columns in planetary atmospheres. Whether a Taylor column will form above an obstacle to a given fluid motion is determined by whether the assumptions of the Proudman-Taylor theorem are satisfied. The criteria for applicability of the theorem to fluid dynamics a Moll's work is not available at the time of this writing. Reference and information are taken from Berger and Wille (1972).

ATMOSPHERIC AND OUEANIC FLOW PROBLEMS

355

of planetary atmospheres involve considerations of slow, steady, hydrodynamic motion past an obstacle in a rotating system, stratification, and baroclinicity. The principal controversy has centered on the question of Jupiter’s Great Red Spot being a Taylor column effect. Conflicting opinions on whether Taylor columns exist on Jupiter and Earth are in part due t o the use of widely different values ’for the physical parameters, suoh as the wind speed u and the linear scales of the phenomenon by various authors. For example, u = 1 m/s, 2 mls, and 50 m/s were adopted in analyses concerning the Great Red Spot of Jupiter by Hide (1961), Stone and Baker (1968), and Ingersoll (1969), respectively. Whereas, Ingersoll adopts the value 8 km for the atmospheric scale height of Earth as well as Jupiter, Stone and Baker assume 18 km for the scale height and 700 km for the depth of Jupiter’s atmosphere. Furthermore, Ingersoll assumes L 2: lo3 km while Stone and Baker adopt L = 500 km and lo4 km, respectively, for the linear horizontal scale in criteria concerning stratification and baroclinicity in the Earth’s atmosphere. The following analysis by Chopra (197213) tends t o establish that certain islands can set up Taylor columns if the islands extend well above the inversion layer. Hide and Ibbetson (1966) have shown that Taylor columns can form if the motion is slow enough such that (4.35)

h N h, = uH/f.L,

where h is the height of the obstacle, L is the horizontal scale of motion, H is the depth of the fluid, u is the magnitude of the ambient horizontal velocity, and f is the Coriolis parameter. For vortex trails leeward of certain islands discussedin the preceding section, u II 7 - 10mls, f 21 7 x 10- 8 - I , L 21 6001000 km, and if one assumes H N 8 km, corresponding to the atmospheric scale height, then h, N 1.3 km. A reference t o Table I1 shows that the islands which have been identified with atmospheric vortex streets have heights greater than this critical value. The criterion contained in Eq. (4.35) by itself is not suficient t o determine whether a Taylor column is formed. Based on considerations of density variations along the vertical (stratification) and along the horizon (baroclinicity), Stone and Baker (1968) point to two additional limitations: (4.36)

uH2/kL< 1

for stratification. and (4.37)

AT 4 ufLTlgz,

for baroclinicity. Here k , T , g, and z, are the coeficients of kinematical eddy conductivity, temperature (on the Kelvin scale), acceleration due to gravity,

356

RULDIP P. CHOPRA

and the effective height over which horizontal temperature gradient extehds. Numerically, the coeficients of eddy viscosity and conductivity are equal. Also, the height of the inversion layer may be taken as the typical value for zo for the mesoscale vortex trails behind islands. Then,

T

g N 9.8 m/s2, zo

N

1 km,

N

kN

end

300°K v N 3 x lo3m2/s

and hence,

uH2/kLN 0.35 < 1 and

AT < (AT),,.it= ufLT/gxo N 24°K which yields for the critical horizontal temperature gradient of

(AT/L)crit = O.O3"K/km. The average horizontal temperature gradient a t or above the level of the inversion layer over a horizontal scale L < 1000 km above water in low or m d-latitudes is lower than the critical value given above. The Stone-Baker application of the baroclinic criterion on the semiglobal (Equator-Pole) scale is not appropriate because the spherical shape of the globe would violate the concept of two-dimensional rotation about a vertical axis. On a suitable smaller scale, however, the criterion can be applied without serious error. Hide and Ibbetson (1966) have shown that Taylor columns form when the Rossby number Ro and the Ekman number E , defined by (4.38)

Ro = u/fL

and

E,

= vIfL2

are small. Substitution of values for u, f, L, and v appropriate t o the islandcaused vortex-wakes yields

R,

11 0.15

and

E,

2:

5x

I n summary, it may be concluded that vortex streets leeward of certain islands provide indirect evidence for the existence of Taylor columns on Earth. These vortex streets are formed behind islands with h >h, N 1.3 km, located in mid-latitudes and under conditions of low level inversion. Taylor columns do not result from flow over mountain ranges because of a n entirely different 00w regime in which gravity waves play en important role. The conclusion regarding the existence of Taylor columns on Earth differs from that reached by Stone and Baker, Ingersoll, and Moll primarily because of the different scale of the specific geophysical phenomenon and use of proper parametric values.

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

357

5. VORTICESLEEWARDOF THE HAWAIIANISLANDS

The Hawaiian Islands are from 1 to 4 km in height, and yet, there is no evidence of the existence of associated atmospheric vortex street wake patterns similar to those discussed in Section 4. However, the oceanic eddy patterns on the downwind side of these islands are very complex. Patzert (1970) has analyzed the data gathered in 20 cruises over a period of 30 years since 1949. He attributes the origin of these complex ocean eddy circulations to strong local winds introduced by the islands, acting as obstacles t o the incident trade winds. If this hypothesis is correct, then the atmospheric eddies generated by the Hawaiian Islands transfer their energy, momentum, and vorticity to the ocean long before these eddies could be carried sufficiently far downwind t o form any discernible wake pattern in the atmosphere.

5.1. The Hawaiian I s l a d Figure 13 describes the topographic features of the Hawaiian Islands which act a8 a series of vertical obstacles to the general atmospheric and oceanic currents. The islands of Hawaii and Maui are the tall islands in the group with their respective peaks a t heights of 4 km and 3 km above the sea surface. The islands of Kauai and Oahu have peaks a t respective altitudes between 1.5 km and 2 km. There are four channels, two of which are deep and the other two rather shallow. The Alenuihaha Channel between the islands of Hawaii and Maui is 2 km deep and the Kauai Channel between the islands of Oahu and Kauai is 3 km deep. The channels of Kaiwi and Pailolo are only 614 m and 80 m deep, respectively.

5.2. Atmospheric Flow and Climatological Properties of the Region Trade winds, incident from the NE or the SE a t the Hawaiian Archipelago form a regular atmospheric environmental feature of the region. The winds with average speeds of 10 t o 20 knots (5 to 10 m/s) are dominant in a region confined between the sea surface and an altitude of 1.8 km. However, there are frequent occurrences of wind speeds in excess of 10 m/s, lasting more than a week. The monthly averages of wind speeds display a yearly periodicity with a maximum in July and a minimum during December through March. Islands, acting as barriers to the trade winds, have a marked influence on the local atmospheric circulations. Hawaii offers a solid barrier, 120 km in the cross-wind direction. Both the islands of Hawaii and Maui tower more than 1 km above the top of the trade wind layer. There is a region of minimum winds in the lee of each island, and strong winds blow through the channels due to the Venturi effect. The shear boundary layer generates cyclonic eddies in air to the north of Hawaii and anticyclonic eddies t o its south during trade

358

'

KULDIP P. CHOPRA

wind conditions. Although winds get deflected by the islands of Oahu and Kauai, the atmospheric eddies in the lee of these islands are not evident. Indirect support for the existence of atmospheric eddies in the wake of Hawaii is provided by the rainfall pattern on the Kona coast of Hawaii. Other islands in the chain have showers on the trade wind side and less than 50 om yearly precipitation on the downwind side. Contrary to this typical pattern, the island of Hawaii does not receive trade wind showers on its East Coast, but its West coast (Kona) receives more than 150 cm rainfall each year. This is due to the blocking effect of the tall island. The region of maximum rainfall coincides with the convergence zone between the cyclonic and anticyclonic eddies to the west of Hawaii. Also, the rainfall activity coincides with the trade wind conditions; more rainfall occurs during the summer months when trade winds are at their peak strength. 5.3. Oceanic Currents an& Circulations

Description of the oceanic circulations and currents in the vicinity of the Hawaiian Islands may be divided into three parts: large-scale flow, nearshore currents, and the oceanic eddy pattern downwind (wdst) of the island chain. Recent observational studies by Seckel, Charnell, and Au (1967) and by Wyrtki, Graef, and Patzert (1969) indicate a net flow to the west in some areas with mean flow speed less than 20 cm/s. This leads them to conclude, contrary to the earlier belief, that the north-Equatorial current does not control or dominate the ocean circulations through the Hawaiian Islands. The ocean circulation pattern downwind of the Hawaiian Islands is complex and variable. It consists of eddies most of which are cyclonic. Figure 19 illustrates the location and drift of various eddies observed in several cruises discussed by Patzert (1970). The characteristic properties of these eddies are discussed in the next section. The near-shore currents are strong with flow speeds ranging from 20 to 75 cmls. The current speed decreases with depth except in the convergence zone between a nearby pair of eddies of opposite circulations. The flow speeds in the convergence zone exceed 60 cm/s even at a depth of 200 m, resulting in a toward-shore current which splits into north- and south-flowing components as it approaches the coast of Hawaii. The relative position of the eddy pair appears to control the intensity and divergence of the current along the Kona coast. 5.4. Properties of Oceanic Eddies

The eddies forming the space-time pattern of Fig. 19 downwind of the Hawaiian Islands exhibit the following features concerning their shape, size,

359

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

location, and drift, about their thermal, salinity, and dynamic structure, and in regard to their velocity profile.

5.4.1. Location, shape, and size of the Hawaiian eddies. These eddies are observed year round in a region from 40 km to 350 km downwind from the islands whenever trade winds are strong. Those observed in the wake of Hawaii are more intense and frequent than those leeward of the other islands. Most eddies are elliptical in shape with eccentricity ranging up to 0.79.Eddies closer to the islands are smaller in size, more eccentric, and with their axis of rotation slanted eastward. As these eddies move away from the islands, their size increases due to viscous diffusion of vorticity, and they become more symmetrical about a vertical axis. The observed eddy radius re varies from 28 km to 80 km with a mean value of 59 km. Figure 33 illustrates the variation of the eddy radius with distance from the island. Whereas cyclonic eddies extend to 300-360 m below the sea surface, the anticyclonic eddies are much shallower, and are barely detectable below a depth of 200 m.

I I

/

cn 65-

-

3

;

I

I

80.

50-

'

0

/

0

0

0

0

e 0

t

55.

457

e

I'@

'4 0

0

0-

0

'0'

0

r

DISTANCE (km)

FIG.33. Variation of observed eddy radius rewith down-wind location of the eddy oenter from the island of Hawaii (after Patzert, 1970).

5.4.2. Thermal and Salinity Features of the Hawaiian Eddies. The thermal structure of eddies in the Hawaiian waters is their most characteristic feature. The cyclonic eddies are characterized by a thermal dome, whereas a thermal depression is associated with the anticyclonic eddies. Therefore, cooler and heavier water accumulates around the axis of a cyclonic eddy and is carried

360

KULDIP P. CHOPRA

by a swirling outward flow to the periphery. Just the opposite occurs in an anticyclonic eddy in which the surface water swirls inward and accumulates around the axis of rotation. Typical difference in surface water temperature at the center and periphery of a cyclonic eddy near Hawaii is 1°C or slightly larger. Eddies display less or no surface thermal contrast as they move away from the islands. These features are illustrated in Fig. 34, which shows thermal structure of a pair of eddies in the lee of Hawaii during Cruise UH-3, July 19-23, 1966. The surface water at the center of the cyclonic eddy, closer to Hawaii, is 2°C cooler then the surface water at its periphery. In contrast, the difference between the central and peripheral surface water temperatures of the anticyclonic eddy is only 1°C. 15.0 I

1560 I

I579 I

0 0 0

0 0 0 0

\

/

/

0

fl

270>

a-

26.5

I;P

FIG.34. Variation of surfece temperature in Hawaiian eddies during cruise UH-3, July 19-23, 1966 (after Patzert, 1970).

The magnitudes of the cyclonic thermal doming and anticyclonic depression are equal only in a shallow layer of water, about 100 m deep. At greater depths, the doming is more intense, and it covers a smaller horizontal area than the depression. Also, a cyclonic doming extends to a greeter depth. For example, the mean depth of the 20°C isotherm in Hawaiian waters is 170 m below the sea surface. The average height of the 20°C isothermal dome is 35 m above this mean level whereas the correspondingdepression is, on the average,

361

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

12 m below the same mean level. The observed average cyclonic dome of the 15°C isotherm is 19 m above the mean level of 240 m, and the 10°C isothermal dome at a mean depth of 340 m is still well defined. However, the observed average depressions of the 15°C or lower isotherms are too weak t o be distinguished from the influence of the local random vertical fluctuations in temperature. I

O W

21' N

0

El'

KAHOOLAWE

190

,17O

I7

90

I d7'

I

io

Fro. 36. Depth of the 20°C isotherm in Hawaiian eddies observed during U.S.S. Marysville cruise 36, Phase 11, July 6-13, 1966 (after Petzert, 1970).

362

KULDIP P. CHOPRA

Figure 36 describes the dome-depression structural features of the 20°C isotherm in a typical eddy pair in Hawaiian waters. The cyclonic dome is more pronounced and extends to a smaller horizontal extent than the anticylonic depressions. Also, the 20°C isotherm at the center of the cyclonic eddy is 70 m above its mean depth of 170 m, whereas it dips 60 m below this depth at the center of the anticyclonic eddy. The undisturbed vertical variation of salinity in Hawaiian waters is characterized by a peak at a depth of 160 m below the sea surface. The salinity at this depth is 36% higher than the mean value. During Cruise UH-3 (July 19-23, 1966) the salinity maximum rose to the surface near the center of the cyclonic eddy where it was about 0.6 yogreater than that of the surrounding surface waters. This feature of high surface salinity has been observed to be associated with other cyclonic eddies in the lee of Hawaii. Thermal doming and salinity features of cyclonic eddies point to the associated phenomenon of upwelling. These features become less pronounced and disappear as the eddies move away from the islands. 5.4.3. Drift of Hawaiian Eo%es. The measurement and analysis of the transport (drift) properties of the Hawaiian eddies is rendered difficult because of the necessity of long period (a few months) needed to monitor an eddy. The BCF (Bureau of Commercial Fisheries) cruises 64, 68, and 72 aboard U.S.S. Charles H.Gilbert and the NEL (NavalElectronicsLaboratory) cruise 29 aboard U.S.S. Maysville were planned to study this feature. In these observations, the movement of an eddy was identified by the shift in the thermal structure of the ocean surface. These observations also provide information on the birth and death of these oceanic vortices. The survey of Hawaiian water during Gilbert cruise 64 (April 9-14, 1963) indicated the drift of an eddy to the northwest at a net eddy-propagation speed of 10 km/day. About ten days later, a series of small shallow cyclonic eddies had replaced the larger eddy. Thermal surveillance for 34 days of an eddy during Gilbert cruise 68 (August 21-29, 1963) showed it displaced by 16 km to the northwest at an average translational speed of 4.6 km/day. Gilbert cruise 72 monitored transport by 16 km of an intense eddy in a period of three days during April 1966. A month later, it was found 80 km away to the northwest. Observations during Marysville cruise 29 (August 17-24 and August 28-September 3,1964) located one strong eddy and one weak eddy at respective distances of 140 km and 270 km, west of Hawaii. Ten days later the strong eddy had moved 37 km to the west. Observations made during this period also showed the formation of an eddy pair. These and other observations indicate that the eddies move away from Hawaii in a westerly or northwesterly direction with an average speed of 6 km/day (or 6 cmls). Their formation time is estimated at approximately

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

363

one week for a weak eddy and at a month or longer for an intense eddy. On certain occasions, the prevailing wind and oceanic flow conditions are favorable for the eddy to be dynamically stable, neither growing nor weakening, and staying at a certain location. As an ocean eddy moves away from Hawaii, its horizontal size increases and vorticity decreases, and ultimately i t is replaced by smaller eddies. Since the eddies are observed over a downwind distance of 40 km t o 350 km from Hawaii, and their average propagation speed is about 5 km/day, their average lifetime is on the order of two months.

5.4.4 Velocity Field of the Hawaiian Eddies. The surface flow in an eddy can be measured by a variety of methods including direct observations with the aid of moored current meters, geomagnetic kinetograph (GEK) fixes, and by tracking paths of parachute drogues. These measurements indicate surface flow speeds around and in excess of 50 cmls. Studies of surface flow, salinity, and thermal features of the Hawaiian eddies are indicative of certain dynamic topographical structure of these eddies. Figure 36 illustrates the dynamic topography of Hawaiian eddies monitored during cruise UH-12, April 26-May 5, 1966. The dynamic topography of an eddy may be described in terms of the dynamic-height difference Ah, measured in centimeters and defined as the difference in dynamic heights at the center and periphery of an eddy, or in terms of the dynamic-height anomaly AD = g Ah,. Here g = 9.8 m/sa is the acceleration due t o gravity. Patzert (1970) notes that the dynamic-height difference in cyclonic eddiea is approximately twice that in anticyclonic eddies. It varies from 10 cm in a weak cyclonic eddy to 30 cm in a very intense one. Also, approximately 75 % of the observed dynamic-height difference in a typical eddy is concentrated in the upper 200 m. Therefore, typical depth 6 of a Hawaiian eddy may be regarded as 200 m. It follows from Patzert (1970) that the observations of the dynamicheight difference Ah a t any radial distance r from the eddy axis fits the exponential formula (5.1)

Ah = (Ah,)exp( -r/re)2

where re is the eddy radius. For an eddy in geostrophic balance, the tangential velocity field is given by (5.2)

2)

= (glf)(a/ar)(Ah)

which, on substitution for Ah from Eq.(5.1))yields (5.3)

w

= -2(g

Ah0lfTe2)rexp( -r/r$

364

RULDIP P. CHOPRA

1580

l5d0

I57O I

I

53.: .......

.. ....

lS8O

............* .. .:A

:

1570

158'

FIQ.36. Dynamio topography of Hawaiian eddies (dyn cm) O/SOO db during Cruise UH-12, April 26-May 6, 1966 (after Patzert, 1970).

At any depth z below the sea surface, the tangential velocity field v is given by (5.4) v = -2g (Ah0/fr2)rexp( exp(- Z / S ) ~ The Coriolis parameter f = 5 x s - l at 20" latitude and 8 is the eddy depth. For known values off and g at any specific value of z, the tangential velocity v has a maximum value (5.5)

=WAhO)/re

occurring at a radial distance r=0.7 re from the eddy axis. Here Ah, and reare expressed in centimeters and kilometres respectively.

365

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

The velocity field of a geostrophic model eddy of radius re = 60 km and Aho = 20 om is shown in Fig. 37. Also plotted are the velocity profile of a Rankine (solid-core) potential vortex of core radius rc = 0.7 r e , and the observed average drogue speeds from cruises Gilbert 64 and UH-17. These observations appear in good agreement with those calculated from the geostrophic model eddy which bears certain similarity to the velocity profile of a viscous vortex in Fig. 25. I n particular, the peak tangential velocity v,, = 65 cm/s of the geostrophic model in Fig. 37 compares well with the average GEK-measured speed of 48 cm/s and the average drogue speed of 57 cm/s. Patzert (1970)also points out that the rotational period of a Hawaiian eddy is inversely proportional t o the doming of its 20°C isotherm. Calculations based on the geostrophic eddy model led Wyrtki et at. (1967) to estimate the volume transport in the range of 2 x log to 8 x lo6 m3/s for various eddies observed during the UH cruises.

-

a0

a m

\

E

3

60

0

40

0 W

>

20

0 RADIUS

(km)

FIG.37. Tangential velocity profile of a Hawaiian eddy during Charles H . Gilbert cruise 64, Phase I ( O),and Cruise UH-17 Also shown are the velocity profile of a solid-core (Rankine) vortex (--) and a geostrophic model eddy (-) corresponding to Ah = 16 cm and re = 65 km (after Patzert, 1970).

(a).

5.4.5. Forces Governing Dynamics of Hawaiian Eddies. To determine the relative magnitude of the various forces contributing to the vortex dynamics in Hawaiian waters, it is necessary to consider the equations governing the horizontal motion. For an incompressible fluid of density p, these equations are (5.6) auiat

+V ( U V + ) w aqaz -jv

= -(i/p)(ap/ax)

+

v

~

+ (i/p)(aT1/az) 2

~

366

KULDIP P. CHOPRA

and

Here u and v are the eastward ( 2 ) and the northward (y) components of the horizontal velocity V, w is the vertical speed of motion, V and V 2 are the gradient and the Laplacian operators, u is the lateral coefficient of eddy viscosity, and r , , ruare the x- and y-components of the wind stress, respectively. The pressure-gradient term can be expressed in terms of the dynamic topohy,

(l/PNaP/~”) = 9 a(Ah)lax

(5.8)

Furthermore, if the eddies are in hydrostatic equilibrium

ap/az= -gp

(5.9)

Equations (5.6) and (5.7)are similar in form for symmetrical eddies i f f is assumed constant. Let us introduce, following Patzert (1970), the scaling factors L, 6, T , V , and Ah, representing the typical horizontal extent, depth, formation time, velocity, and dynamic-height difference in an eddy. Also, let w, r,,, and v be the typical vertical velocity, lateral shear stress and eddy viscosity. These parameters have values

LN

lo6 m,

T 1: lo6 - lo6 8, Ah, 21 0.2 m, w ~ l O - ~ m / s ,f ~ 5 x l O - ~ s - l

6 N 2 x lo2 m,

V~0.5m/s,

and corresponding to 20-37 knots wind speed, r , 2: 0.4 to 1.0 dyn cmW2

and

uN

10 t o lo3 m2/s.

With these values for the characteristic scaling parameters, various acceleretion terms in Eq. (5.6) and (5.7), taken in order, have the magnitudes (in units of ,Is2): Time-dependent inertial term Rotational inertial term Vertical inertial term Coriolis term Pressure-gradient term Viscous term Wind-stress term

t o 5 x lo-’ V/TN 5 x V 2 / L11 2.5 x wV/6 N 2.5 x 10-7 f V 2: 2.5 x g Ah,/L N 2.0 x VV/L2N 5 x 10- lo to 5 x 10-8 r o / p 6 N 2 x 10-6 to 5 x 10-6

It appears from the above estimates that the Coriolis and the pressuregradient forces being the strongest govern the development of eddies in the Hawaiian waters. The wind stress term may become important when winds

367

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

are very strong, and the inertial terms may become significant when the characteristic period is shorter than 10 days, as in case of inertial oscillations, or in the case of very intense eddies with large V . Three non-dimensional flow parameters may be formed from the inertial, viscous, gravitational (pressure-gradient), and Coriolis forces. These are Reynolds number R =

Inertial term = V L / v N 50 to 5000 Viscous term

Froude number F, =

Inertial term = Va/gAh0 Gravitational term

Rossby number R,

Inertial term Wind stress term

N 0.12

and =

=

V l f L N 0.1.

Range of values for R indicates that the inertial forces are much stronger than the viscous forces. However, low values of Ro end F, demonstrate that the wind-stress and the pressure-gradient forces exceed the inertial forces by a factor of 10. The almost equivalence of F, and R, signify the almost equivalence of the Coriolis and pressure-gradient forces. A balance of these forces results i n the geostrophic nature of the Hawaiian eddies. Following Patzert (1970), the value of v, taken from Deacon and Webb (1962) is used in the estimation of the viscous force term.

5.5. Generation Mechanisms for Hawaiian Eddies Two mechanisms have been proposed t o explain the formation of oceanic vortices downwind of Hawaii. The first mechanism is based on the modification of oceanic flow by islands, and the other invokes a wind-ocean interaction.

5.5.1. Vortex-Generation by Flaw around Hawaii. McGary (1955) and Barkley (in Manar, 1967) have explained the vortices in Hawaiian waters as a wake phenomenon, similar to the vortex street wake discussed in Section 4. Patzert (1970) subjected this hypothesis to the criteria related to the values of (a) the Reynolds number R, (b) the Strouhal number 8,(c) the spacing ratio hla, and ( d ) the velocity ratio u,/uo. We apply the same criteria in view of the discussion in Sections 4 and 5. Deacon and Webb (1962) find that the lateral oceanic eddy viscosity v varies from 10 to 1000 m2/s. With cross-stream diameter d 1: 1.5 x lo5 m for the island of Hawaii, ambient stream velocity uo 21 2 m/s, and the

368

KULDIP P. CROPRA

intermediate values of v = 200 and 400 ma/s, R has values of 150 and 75, respectively. These values for the Reynolds number correspond to the appearance of stable vortex streets in the laboratory and in the atmosphere. Application of the criterion S N 0.2yieldsforperiodicity T N 5d/uo 21 40days. There are no observations of eddy separation to verify this feature. However, this periodicity of 40 days is comparable to a 10- to 25-day intervalneededfor the formation of eddies and to 60 days for the average eddy lifetime. Patzert (1970) estimates lateral spacing h between the various cyclonic and anticyclonic eddies at 80 km.There are fewer observations which show the existence of more than one eddy of the same type. On the basis of these few observations, Barkley (in Manar, 1967) estimates a N 220 km. These iveraged values of h and a yield hla N 0.36 which is within the observed range for real vortex streets in the atmosphere and laboratory. The Hawaiian eddies have been observed to propagate downwind of Hawaii with speeds varying from 3.5 to 11.6 cm/s. To produce these translational speeds, the ambient flow must be steady with speed uo N 5 to 16 cm/s. The averaged eddy propagation speed u, 2: 5 cm/s corresponds to u o N 7 cm/s. The observations show that the north-Equatorial current is not steady in Hawaiian waters. However, the shallow trade wind layer may introduce the necessary oceanic flow through the islands. Hydrographic survey of the Hawaiian waters during Cruise UH-16 suggests a vortex wake, west of Hawaii, and the surface flow with speed uo H 30 cmls through the Alenuihaha Channel. With the channel width W N 40 km and 6 N 200 m for the thickness of the layer in which flow speed u has decreased by 62.5 yo of the surface value u o , the rate of volume transport through the channel is uo W 6 N 2.106 m3/s The kinetic energy carried through the Alenuihaha Channel is given by (Patzert, 1970) (5.10)

(K.E.)ch= ( ~ / 4 8 ) ~ % WGAt ,~

If At N lo6 a, corresponding to the average formation time of the Hawaiian eddy, is substituted in Eq. (&lo),and it is assumed that the entire energy of the oceanic flow through the channel is used in the generation of an eddy, one obtains (K.E.) N 6 x loaoergs for an average eddy. Whereas Patzert (1970) finds the estimate for voIume transport acceptable, he regards the estimate for eddy energy low by at least two orders of magnitude.

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

369

5.5.2. Wind-Driven Mechanism for Generation of Hawaiian Eddies. It is noted in Section 5.4.5. that the Coriolis and pressure-gradient forces are almost equal and dominate the dynamics of ocean eddies downwind of Hawaii. It is also pointed out in Section 5.4.4. that the velocity profile of a geostrophic model eddy bears resemblance t o the Rankine vorteg, and is in good agreement with observational data on Hawaiian eddies. Using the geostrophic velocity profile and considerations of dynamic topography, Patzert (1970) obtained the following expressions for the kinetic and potential energies of an eddy:

(5.11)

K.E. =

(5.12)

P.E. = I v p A D dV

J" (+)pv2dV = (7r3/32)"'pg S(Ah,/f)' = ( ~ ~ / ~ / 2&rez ) p gAh,

and (5.13)

P.E./K.E. = 2d/2(frJ2/g Ah,

It is assumed in the derivation of Eqs. (5.11)-(5.13) that the dynamic-height anomaly A D has a Gaussian distribution, A D = D o exp(--2/6)2, with depth, D o = g Aho, and Eq. (5.4) is valid for the geostrophic velocity profile of a n eddy. There are a few contrasting features of the parametric dependence of the kinetic and potential energies of a geostrophic eddy. The kinetic energy is independent of the eddy size, but it is directly proportional to (Ah)' and inversely proportional to f 2. On the other hand, its potential energy is directly proportional t o the eddy size (area) and (Ah),but it is not affected by the Coriolis parameter. Both kinetic and potential energies are directly proportional to 6, the eddy depth, and their ratio is a function of the eddy radius r e , surface topography Ah, and the Coriolis parameter f. Eddies were surveyed principally for their thermal structure, and data for many eddies included dynamic topography. I n case of weak or moderate eddies (Ah <20 cm), there is a linear relationship between the thermal doming A20" of the 20°C isotherm and the dynamic-height difference Ah, 3.35 Ah(cm)= A20(m)

I n such cases, this linear relation is used. Patzert (1970) adopted 6 = 170 m corresponding to the mean depth of the 20°C isotherm. At this level, the flow speed is 62.5 yo of the surface flow speed. The Coriolis parameter f = 5 x s - l a t 20" latitude. With this information Patzert calculated the

370

KULDIP P. CROPRA

following ranges of the potential, kinetic, and total energies of Hawaiian eddies: 0.7 x 1O2I < Potential energy (ergs) < 6 x 1021 4.2 x loa1
where T.E., w o , T ~ and , A are, respectively, the total energy of an eddy, oceanic flow speed, wind shear stress at the sea surface, and the surface area on which the wind stress acts. If w o and r o act in the same direction for an interval of time At, then (6.14)

(T.E.)eddy = v0 T~ A At.

This equation defines At as the time needed to transfer the energy (T.E.)eddy to an eddy by a wind stress T~ exerted over an area A with surface speed w o . The wind stress is assumed to act as a torque in a direction parallel to eddy corresponding to rotation, and has values in the range of 5 to 10 dyn 26-37 knots winds. The surface flow speed varies from 30 to 60 cm/s, and the ergs. Typical area of an eddy total energy varies from 2 x loz2to 5 x over which T~ acts is about 6.7 x 1013 cmz. Using this information, Patzert (1970) estimates the time At for eddy formation. Three values of At under different conditions are listed in Table VI. The longest time of 58 days corresponds t o the formation of an intense eddy generated by weakest wind stress. To generate this eddy, 28 knot winds must be steady for 58 days. The shortest interval of 6 days is needed for the formation of.a weak eddy generated by strong (37 knots) winds. Patzert (1970) cites evidence for existence of similar local wind conditions. Once oceanic eddies have been formed through the action of wind shear stress, these could be transported by the local oceanic flow. Even in a stationary medium, the translational motion of eddies could be caused by the

37 1

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

TABLEVI. Calculated formation times for wind-driven eddies@

Wind speed (knots)

Wind stress T~ (dyn/cm2)

Surface flow speed vo (4s)

Eddy energy (T.E.),dd, (ergs)

Eddy formation time At (days)

26 31 37

6.0 7.6 10.0

30 46 60

6.0 x loaa 3.6 x loaa 2.0 x loas

68

18 6

After Patzert (1970).

meridional variation in the Coriolis parameter. Warren (1967) finds that a n eddy located a t 20" latitude would have a tendency t o move westward a t a translational speed given by ( i)&re2,where E is the variation of Coriolis parameter with latitude. For vortices in the Hawaiian waters, E II 2 x 10- 13/cm/s and re II 50 to 100 km. Therefore, the net rate of westward drift of these eddies due to this cause would vary from 2.5 to 10 cmls. The observed westward drift is from 3.5 t o 11.6 cmls. 5.6. Concluding Remarks

The pattern of oceanic vortices west of Hawaii is complex and dominated by cyclonic eddies. Occasionally, there is a n evidence of vortex formation in a fashion similar to the one which leads t o vortex street. There is no evidence of steady oceanic flow with suficient energy to satisfactorily account for the eddy-energy spectrum. Therefore, the origin of these eddies lies in a n air-sea interaction mechanism in which strong winds and atmospheric eddies formed in the lee of Hawaii transfer their energy t o the oceanic eddies. Once formed, these eddies are carried westward by a cause attributable to the meridional variation of the Coriolis parameter. The predominance of intense cyclonic eddies is perhaps due to the orientation of the various Hawaiian Islands with respect to the trade winds. Patzert attributes the feature of cyclonic eddies being more intense than the anticyclonic eddies to the action ofthe otherwise weak rotational inertial term. 6. ANOMALOUSOCEANICCIRCULATIONS AROUNDISLANDS Considerations of the effects of wind vorticity, meridional variation of the Coriolis parameter, and the external oceanic boundaries in the hydrodynamic theory of oceans have led to the explanation of the intensification of currents like the Gulf Stream and the Kuroshio current near the western boundaries of oceans, the Cromwell deep equatorial counter-current and the equatorial

372

KULDIP P. CROPRA

inter-trade wind counter-current. Shtokman (1966) has proposed an intuitive theory, which says that the fact that oceans are also bounded from the inside by the contours of islands may explain the anomalous circulations around islands. Figures 20-22 exhibit the anomalous circulations around Taiwan, Iceland, and the islands in the Kuril chain. Here, thick arrows indicate the magnitude and direction of the local wind, and thin arrows describe the oceanic flow. Circulations around islands in these figures are anticyclonic, and are in a direction opposite t o that of the general cyclonic flow in the northern hemisphere in the oceanic regions containing the islands. Hence, these closed-loop type flows around certain islands are named anomalous. There appears a strong correlation between the anomalous circulations and the regional monsoons. For example, the anomalous flow around Taiwan (Fig. 20) is seasonal and occurs during winter months when the Chinese northeastern monsoons blow (Makarov, 1950). On the other hand, the anomalous flow around Iceland (Fig. 21) is year round and the local region is under the influence of the Greenland monsoons throughout the year as well. Makarov was the first to note the phenomenon around Taiwan, and he attributed it to the effect of the Coriolis force, abundant water supply, and strong winds. Shtokman (1966) pointed out two necessary conditions associated with the occurence of the observedphenomenon: the asymmetric location of the islands with respect to the continental boundaries and the wind vorticity (transverse inhomogeneity). Furthermore, since the phenomenon is a perturbation effect caused by the presence of an island, a large island would generate an intense effect. Following this logic, he proposed a qualitative explanation based on a n air-sea interaction which takes into account the wind vorticity, the Coriolis effect, viscosity, the boundedness of the oceans from within (by the islands) and from outside (by the continents), and the locations and sizes of the islands. Shtokman’s reasoning is easily understood with the aid of Fig. 38a-k, which exhibit various flow configurations in a rectangular ocean caused by the respective internal boundary conditions and by the prevailing rectilinear winds with transverse inhomogeneities. Here, the thin arrows describe the direction of ocean currents. The solid and hollow large arrows indicate the magnitude and direction of the prevailing wind. The wind velocity decreases from left to right when observed facing the wind. The curved arrows in the wind field indicate the vertical circulation in the prevailing wind. Wind vorticity in the northern hemisphere is cyclonic and i t is the cause of predominantly cyclonic circulations in closed seas such as the Black Sea or the Caspian Sea (Shtokman, 1945). Figure 38a exhibits this type of oceanic flow in an ocean of rectangular (square) boundaries. If this ocean is divided into two regions A and B by means of a strip as shown in Fig. 38b, two separate

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

373

FIQ.38. Anomalous ocean currents generated by inhomogeneous linear winds.

cyclonic circulations result, one in each region. However, if this strip is replaced by a long island and two straits C end D as in Fig. 38c, two additional cyclonic circulations appear in the straits if C and D are narrow but wide enough to permit partial exchange of water between the regions A and B. This results in an anticyclonic flow along the contour of the elongated island. The dots represent the singular points corresponding t o the condition of stagnant flow. Here, the breakdown in the main cyclonic circulation by an elongated island of fairly large dimension leads t o the anomalous (anticyclonic) circulation around the island. However, if the straits C and D

374

KULDIP P. CHOPRA

become wide enough to permit free exchange of water between regions A and B, there is no breakdown of vorticity, and therefore no anomalous circulation appears. This feature is shown in Fig. 38d-f. The last two figures deal with the cases of a small and a large island located symmetrically within the continental boundaries. Figure 38g,h illustrates two situations of islands located asymmetrically. I n each case there is a splitting of vorticity resulting in anticyclonic (clockwise)flow along the island’s contour. The breakup of the cyclonic vorticity can also be caused by flow through a strait, resulting in an anomalous loop current. This is shown in Fig. 3% and follows the generalized rule of Zubov (1947): “ I f you stand astride a strait and stretch your right arm forward and your left arm backward, the direction of your extended arms indicates the direction of the currents off the respective shores.” Finally, the preceding considerations can be generalized to apply to an island chain as in Fig. 38j,k. The closer the islands are to each other, the greater is the probability of anomalous circulations around each island. A very wide strait would permit a free exchange of water between regions across it. In summary, the anticyclonic circulations around an island (or through certain straits) in the northern hemisphere are wind-driven, and result from the breakup of cyclonic vorticity in the local wind. These circulations appear as perturbation effects caused by an island (or strait) or by a chain of islands. Hence, the magnitude of the effect depends on the size of the island (or strait) and the degree of asymmetry of the island’s contour with respect to the continental boundaries. To estimate the effect of island’s geometry and wind shear, Rzheplinskiy and Shtokman (1968)used the Navier-Stokes equations governing the motion of viscous fluid

and the equation of continuity (6.2)

aulax

+ avlay + awlaz = o

Here u, v, and w are the velocity components along the rectangular axes x, y in the horizontal plane and z-axis along the vertical; p is the pressure; f is the Coriolis parameter, and v is the kinematic coefficient of vertical momentum transfer. At the sea surface (6.3)

vaulaz = -7t,

vavlaz = T v ,

p

= const.

where r , and ru are the x- and y-components of the wind stress. Also (6.4)

w = uaglax + vaglay

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

375

where f is the rise of the sea level above a fixed surface end is a measure of the dynamic topography of the ocean. Furthermore, a t the bottom of the sea, z = 6, there is no flow

u=q)=w=o

(6.5)

and the volume transport is given by d

(6.6)

S2 -I0 udz

and

S,=Jbdvdr

These equations combined together take the form, given by Felzenbaum (1960) (6.7)

+

atlax = - M T ~ NT, - a(a+/ax) - b(a+/ay) a&ay = NT, - M ~ , b(a+/as) - a(a+/ay)

+

Here $(x, y) is the total flow stream function defined by (6.8)

s,=

-a+lay

and

s,=a+/aX

M, N, a, and b are coefficients which depend on j,v, and 6. These coefficients are constants for oceanic regions of small horizontal extent (f = const.) and constant depth 6. The total flow potential is then given by b v2+ = M[(aT,,/aZ)- (aT,/ay)l

(6.9)

+“(aT,/aY) + (aT,/a41

where V2 is the Laplacian operator. The transverse variation of wind is determined by the relations (6.10)

and

T,=O

T~=c+~x

The additive constant c has no effect on the total flow distribution, but it affects the velocity a t different levels and in particular a t the surface. Wind stress T~ is maximum when c = O . With the introduction of Eq. (6.10), Eq. (6.9) becomes (6.11)

bV2$

= MI

Lastly, there is no flow normal to any external or internal boundaries. Following Kamenkovich (1961), Rzephlinskiy and Shtokman (1968) sought a solution of the form (6.12)

where $o is the inhomogeneous solution of the Eq. (6.11) with zero values on all contours, and &,,(a= 1, . . . , a)are the homogeneous solutions of Eq. (6.11)

376

KULDIF' P. CHOPRA

vanishing a t all contours except the nth. To consider the surface flow or flow in shallow waters, the influence of Earth's rotation may be neglected so that

Rzheplinskiy and Shtokman (1968) obtained numerical solutions to these equations and their results are summarized below. A linear wind with cyclonic vorticity acting in a region containing islands generates one or several integral cyclonic circulations. There is no closed circulation in narrow channels between islands or immediately around them. I n some cases anticyclonic (anomalous) water transfer is observed, but only along certain segments of the island shores. For example, the total flow in an ocean containing a rectangular island may consist of a cyclonic external circulation enclosing two cyclonic eddies, one above and the other below the length of the island. The critical length of the island corresponding to the appearance of the two cyclonic eddies is one half the extent of the continental boundary. The smaller islands do not generate these eddies. On the other hand, these eddies become intense and the external cyclonic flow becomes weak for island length greater than this critical length. The flow near the island boundaries due to these eddies is in a reverse dwection t o the general (external) flow. I n general, the anomalous flow is due either to the generation of local closed cyclonic circulations or to the flow about an island or a group of islands being cyclonic from the right as well as from the left when an island or an island group is asymmetrically located with respect to the external boundary of the ocean. Numerical calculations do not predict any closed anomalous flow around the island. The circulation on the ocean surface or on shallow oceans differs considerably in some cases from the total flow described above. I n regions between islands and the external ocean boundary on the left of the island, or in straits or channels between islands, cyclonic circulations can develop giving rise t o anomalous flows along the left (while facing wind) contours of islands. The generation of such currents depends on wind shear stress which is maximum when c = 0. The constant c does not affect the total flow potential or the horizontal component u of velocity. However, it exerts a strong influence on v, and particularly on flows through straits and near the island boundaries. The reverse (anomalous) current along the left coasts of islands a&ears when 0.3 < c / l L < 0.9. The numerical analysis does not predict anomalouscurrents along the right coasts of islands nor does it indicate any closed anomalous currents around islands. I n short, the numerical analysis falls short of fully explaining the observed phenomenon..

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

377

7. UPWELLINQ DUETO CIRCULATIONS AROUND ISLANDS Wyrtki (1963) defines upwelling as an ascending motion by which water from subsurface layers, usually from depths not exceeding 200 m, is brought to the surface and is removed by horizontal flow from the area,of upwelling. It is always accompanied by downwelling, the two forming a macroscale vertical solenoidal current. Surface wind stress is regarded as the principal cause of the surface layer divergence which in turn produces upwelling. However, the exact nature of the interaction between wind stress, vertical and lateral mixing, internal waves, and the various boundary-layer processes is not clearly understood. Changes in density, dynamic topography, salinity, and thermal structure of the local oceanic region always accompany the phenomenon of upwelling. Surface temperature is lower, and the density and salinity of surface water are higher in regions of upwelling than their corresponding values in the surroundings or in regions of downwelling. Of these properties, spatial variation of temperature is the most pronounced feature. Studies of these physical properties is of importance for the following reasons. (a) Changes in dynamic topography, salinity, and thermal structure of ocean surface are helpful in identifying local macroscale oceanic flow characteristics. Most oceanic eddies in Hawaiian waters, discussed in Section 5, were identified by their thermal structure. (b) Observations of density, dynamic topographical, thermal and salinity variations associated with macroscale oceanic flows would lead t o our better understanding of the upwelling phenomenon in particular and the role played by wind shear stress in air-sea interaction phenomena in general. (c) Underwater propagation of sound depends on density and temperature structure of the medium. Spatial and temporal changes in density and temperature render any long-range prediction of underwater propagation of sound unreliable. (d) Surface layer divergence and upwelling cause enrichment of the surface layers by lifting more nutritious water from lower layers. This increases the productivity rate of organic matter in these areas. Greater abundance of fish off the coast of Peru and in certain equatorial regions is attributed to intense coastal and equatorial upwelling (Wooster and Reid, 1963; Cromwell, 1953).

7.1. Upwelling i n Oceanic Eddies Leeward of Hawaii The phenomenon of upwelling is closely associated with, and it may be identified by, the appearance of thermal doming, increase in salinity and the enrichment of surface layers with nutrient-rich subsurface wat,ers, and entrainment of organisms such as littoral larvae into circulation. The thermal and salinity features are strong characteristics of eddies in Hawaiian waters.

378

KULDIP P. CHOPRA

These properties are discussed in Section 5.4.2. The temperature a t the center of a cyclonic eddy is usually 1" to 2°C lower than a t the periphery of the eddy. Increase of salinity in surface waters is noticeable. I n one instance during the cruise UH-3, the level of peak intensity rose to sea surface from its normal depth of 150 m. The average vertical velocity due to upwelling is estimated at cm/s (Patzert, 1970). This estimate suggests strong upwelling. However, McGary (1955) studied the influence of upwelling in eddies on concentration of dissolved oxygen and inorganic phosphate, and found very little enrichment of the euphotic zone. To enrich the euphotic zone, the nutrientrich water has t o be raised from a depth of 300 m to a depth of 100 m. Insignificant enrichment of the euphotic zone is another indication of the shallowness of the Hawaiian eddies. Eddies which move away from the islands do not exhibit high salinity and thermal doming features. Absence of these features suggets that the upwelling in Hawaiian eddies occurs primarily during their formation phase. Brand (1970) has examined the effect of upwelling in one tropical storm on the intensity and direction of motion of another following the wake of the former. He concludes that (a) a hurricane approaching the wake of another tends to either move away from the wake or follow a course parallel to the wake; and (b) a hurricane which does follow the wake of an earlier hurricane is usually of much less intensity. It appears that these general remarks have some bearing on the dynamical properties of the Hawaiian eddies.

7.2. Upwelling Cawed by W i d Parallel to

Long Idand8

Upwelling is caused by surface wind shear stress. Winds parallel t o the length of a very long island may result in separation of cold surface water off one coast of the island and warm water off the other coast. This prediction is based on the application of the current theories explaining the phenomenon of coastal upwelling to the coastal regions of long islands.

7.2.1.Coaatal Upwelling. Most noted areas of coastal upwelling are associated with the California and Peru currents in the Pacific, Canary and Benguella currents in the Atlantic, and oceanic regions'off Somalia and the west coast of Australia. The region of upwelling extends to a lateral width of about 30 t o 40 km and a depth of about 80 m in mid-latitudes. The rising cooler water sets up anomalies in the zonal distribution of surface temperature. A coastal shelf or shallow sea modifies upwelling such that either the region of most upwelling is located further from the coast or the width of the upwelling zone is increased. Smith (1968) observed that the water temperature a t 10 m depth four nautical miles off Newport, Oregon was 8°C cooler than the water a t the same depth just west of the shelf edge. Compared to this is

ATMOSPHERIC A N D OCEANIC FLOW PROBLEMS

379

the 1°C normal difference, due to the geostrophic effect, observed during the winter when upwelling is absent. The average upwelling speed is on the order of cm/s, and the region of upwelling extends to about 50 km in lateral width and 200 m in depth. The mixing layer is rather shallow. It is 80 m deep and consists of two sublayers, each 40 m deep. The 00w is away from the coast in the upper sublayer and it is coastward in the lower layers off the California coast. Sverdrup (1938) noted that the surface flow away from the coast of California occurs when the winds are northerly. Thus, this condition should meet one of the criteria set by any satisfactory theory of coastal upwelling.

7.2.2. Theories of Coastal Upwelling. Following the basic concept of Ekman's (1905) wind drift theory, Sverdrup (1938), Hidaka (1954), and Yoshida (1955) have developed theories to explain several aspects of coastal upwelling. To understand the basic features of these theories, let us adopt a rectangular coordinate system with the x- and y-axes along the west and south directions, and the z-axis along the vertical, respectively. Let us also assume that the sea is to the west of the coastline which runs north-south and that the winds are northerly. This frame of reference would then represent situation off the coast of California. Ekman's wind-driven transport of surface waters normal t o the wind represents a steady state situation resulting from a balance of the Coriolis and the wind shear stress forces in the free ocean where coastal effects may be negligible. The governing equations are (7.1)

pfu = aTy/az,

and

pfv = -aT,/az

Application of the boundary conditions u=o,

v=o

at z=-w

and

u=O,

at z = O

r,=O

yields for the horizontal volume transport

s, = pu dz = T y / f

(7.2)

I n other words, the net horizontal water transport is 90" to the right of the wind direction in the northern hemisphere. Hence, northerly winds in the eastern Pacific would induce water transport t o the west. Sverdrup (1938) introduced the effect of coastline and finite width L of the wind stream on the upwelling phenomenon considering a region bounded by x = 0 to x = L and z = - H to z = 0. Applying the boundary conditions w=O

at z = O

and

w = w - ~

at z=-H

380

KULDIP P. CHOPRA

on the velocity of upwelling in the equation of continuity

+

aulax a w p z = o (7.3) with negligible divergence av/ay along the wind, he obtained L

0

S , = I _ H p u d z = $ pw-,dx

(7.4)

0

Here, z = -H is the bottom of the mixed layer. According t o Sverdrup, Eqs. (7.2) and (7.4) are equivalent. Therefore, northerly winds along the coast off northern California, Oregon, and Washington, which runs approximately north-south, would cause transport of surface water offshore, necessitating replacement inshore by upwelling. He estimated that the influx of water t o sustain cellular circulation pattern comes from a depth of about 200 m. However, he did not make any estimate of upwelling velocities. Hidaka (1954) improved Sverdrup's steady state model of upwelling by including effects of sea-surface topography and lateral mixing. He assumed constant horizontal and vertical mixing coefficients A , and A , in a sea of constant density p and negligible variation of wind in the y-direction. At the sea surface, he assumed the wind shear stress to vanish except for T,,= A , av/az L . Also, he introduced two friction lengths in the region 0 x

(7.5)

Dh= ?r(pf/Ah)"a,

and

D, = n(pf/Av)'/2

Application of the Hidaka model t o upwelling off the coast of California yielded the following estimates for some of the upwelling parameters:

8 D,

N3

x

cm/s,

N 160 km

L N 80 km Dh 2: 1.6 D,

!? 240

km

The estimate for the average upwelling velocity 8 is rather large, that for the width L is reasonable a t 30"N latitude, and the estimate for depth D from which the upwelled water comes is in agreement with the Sverdrup model. Yoshida (1955) developed a transient phase upwelling model in a two-layer ocean, and obtained a n expression

(7.6)

~ - , = 7 ~ ( p f l ) exp(-x/Z) -~

for the vertical velocity at the base z = -H of the upper layer in the coastal region for the early stages of upwelling. Here 1 = - (gHAp/p,)1/2f- l ; pu is the density of the upper layer, and Ap is the density difference between the upper and lower layers. The width of the upwelling region is given by

(7.7)

L=nl

For the coastal waters off California, 1- 10 km and L ment with observations.

21 30

km, in close agree-

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

381

Smith (1967) has shown that under the assumption of uniform wind over the California coastal waters, Eqs. (7.3) and (7.6) yield (7.8)

8,= (Tglf" - exp(~IO1

With L N ~ 1 the ) exponential term in Eq. (7.8) can be neglected, and

8,= %If as in the Ekman transport equation (7.2). It is noteworthy that two entirely different approaches lead to the same result for offshore volume transport of water. The SverdrupEkman drift results from an equilibrium state of the ocean under the influence of wind stress in a region away from the effect of coastal boundaries. Yoshida's model applies only to inshore coastal regions and is applicable during the transition stage of developing upwelling prior to the attainment of equilibrium state. Yoshida and Mao (1957) point out that it is necessary t o introduce more refined mechanism to explain all aspects of coastal upwelling, particularly in offshore regions away from coastal boundaries. They assumed a two-layer ocean model with 50 m thick upper layer separated from 600 m thick lower layer by a relatively thin stable transition layer. They combined the continuity equation (7.9)

aulax + aqay + awlaz = o

with the vorticity equation

they obtained

for the lower layer, and (7.12)

for the upper layer. Here,

W-H

=f-'(V x 7

5 is the vertical

) ~

component of relative vorticity

V and Vx are the divergence and curl operators, and Vh2 is the two-dimensional (horizontal) Laplacian operator. Equation (7.11) states that & poleward subsurface current is associated with regions of upwelling and an equatorward current is associated with areas of descending water motion. Equation (7.12) tells us that the sign and magnitude of vorticity in wind stress determines the sign and magnitude of vertical velocity. For example, positive

382

KULDIP P. CHOPRA

vorticity in inshore areas off the coast of California leads to upwelling, whereas downwelling occurs in offshore areas where wind stress vorticity is negative.

7.2.3.Coastal Upwelling on Sides of Islands. The theoretical models discussed above predict upwelling caused by northerly winds along the continental west coasts in the northern hemisphere. Similar winds along the eastern continental coasts would produce descending water motion. Application of these basic concepts to the case of a long ibland with prevailing winds parallel to the length of the island would lead us to expect different surface water temperatures on the two sides of the island.

I

N

1

FIG.39. Upwelling in coastal waters of long islands.

Let us consider a hypothetical island (Fig. 39), of rectangular shape with its length parallel to the north-south direction located in the northern hemisphere. A northerly wind would produce upwelling in the west coastal waters and downwelling in the east coastal waters. The surface waters on the two sides of the island would be a t different temperatures, the upwelling bringing the colder, biologically nutritious waters to surface off the west coast. Also, if Yoshide's upwelling model is applicable, a subsurface poleward current along the western boundary of the island and an equator-ward subsurface current along the eastern boundary would be necessary to sustain the ascending and descending motions on the two sides of the island. These coastal subsurface currents are clockwise (anticyclonic),similar to the ones discussed in Section 6, in the northern hemisphere.

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS OVER 8. AIR FLOW

A

383

HEATED ISLAND

Land mass of an island is usually warmer during day and colder at night than the surrounding water mass. This differential heating of land and sea alters the nature of air flow over a natural island. Heat from the underlying surface is transferred to the surface layer of air by conduction and small scale turbulence, and it is transported to upper layers by penetrative convection and by mesoscale circulation. Inland surface flow of cooler air from the sea side during the day is called the sea breeze. At night, the land breeze blows in the reverse direction. There are variations in the time of onset, magnitude, and even direction of sea breeze. Under suitable conditions cloud formation may occur. If this formation of clouds results in precipitation, the accompanying release of latent heat of vaporization may lead to the development of vigorous local disturbances. The heated island phenomenon has certain similarities to the effects related to the lake-induced storms and to cities as localized heat sources. Studies of this phenomenon also have a bearing on experimental efforts to artificially produce rain by asphalt coatings of land. 8.1. Land and Sea Breezes

Land and sea breezes are the mesoscale air circulations caused by heated islands and coastal land masses. Typical extent of their inland or offshore penetration is 30 km. The land breeze is usually shallower and weaker than the sea breeze. The former is limited to a surface layer of 200 to 400 m thickness, whereas the latter may extend to a height of 1 km. Mountains cause forced convection and hence strengthen the sea breeze effect. Furthermore, sea breeze carries moisture with it as it moves inland, and therefore it would increase the moisture content and precipitation over land, particularly over mountainous regions. The shape of a coastline has an influence on the direction of local breezes. In general, land and sea breezes are normal t o the coastline. Therefore, a convex coast would cause a convergent sea breeze and a divergent land breeze. Also, the earth’s rotation causes a slow clockwise veering of the sea breeze in the northern hemisphere. A prevailing wind is expected to cause acceleration of the sea breeze on the upwind side and a deceleration on the downwind side of an island. Land and sea breezes are strongest on islands characterized by large diurnal variation of temperature and strong land-sea thermal contrast. The land and sea breezes are usually strong in the tropics and appear year round. I n contrast, these are mild and seasonal in mid-latitudes. Haurwitz (1947), Schmidt (1947), Pierson (1960), and Defant (1961) developed linear models of sea breezes to explain these and other features of

384

KULDIP P. CHOPRA

land and sea breezes. Nonlinear models developed by Fisher (1961) and Estoque (1961, 1962) introduce additional features of advection and penetrative convection. The reader may find a recent review of the land- and seabreeze phenomena by Pannuto (1969) and a bibliography by Baralt and Brown (1965) useful. 8.2. Air F b w over Typical Islands

Several observational studies during the past decade have revealed some unusual features of air flow associated with certain islands. These observations have provided food for further thought in understanding the air-landsea interaction. 8.2.1. Air Flow over the Island of Cuba. The island of Cuba, centered a t 22"N and 69"W is 137 to 161 km wide and 966 km long, oriented in an ESEWNW direction. Its topographical features are shown in Fig. 15. Most of its terrain has an elevation below 250 m with a few peaks rising above 300 m. The southeastern Cuba is mountainous with peaks rising to 2 km along the south coast and to 1.2 km on the eastern tip. A small mountain range rises to 1.16 km in central Cuba. The Azores high pressure belt is generally located to the east and north of Cuba. The belt moves southward in winter and northward in summer. As a result, the island receives easterly or northeasterly winds which tend to be steady and persistent for several days. These winds bring plenty of moisture with them. This steady synoptic pattern is occasionally interrupted by the arrival of tropical storms during September and October and by cold fronts and northerly winds from the continental United States during winter. But for these abnormal features, neutral stability conditions exist in the marine atmosphere around Cuba to a height of 450 m. Slight stability occurs in the upper air layer up to 1.8 km. Diurnal heating or cooling of land produces slight instability or stability in the layer below 450 m, resulting in local convection and cloud formation. The mountains have considerable influence on the local patterns of clouds, rainfall, and winds. The land and sea breeze features of the island of Cuba are summarized below (Smith, 1968). (a) The northeasterly trades represent the onshore winds along the northeastern coastal region. The trade winds are reinforced by the sea breeze during the day and the weaker land breeze is overwhelmed by the trades a t night. Therefore, the northeasterly flow persists through most of the night. (b) Mountains in the southeastern region shield the southern coast from the trades and enhance the sea breeze which is strong enough to produce an onshore flow during late afternoon and early evening. The onshore flow along the south coast ends about 2100 LST.

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

385

(c) The north coast of Cuba runs parallel t o the trade wind flow. The onshore component (sea breeze) appears in the afternoon, decreases significantly during the early evening, and disappears around midnight during winter and about 2100-2200 LST during summer. (d) The onshore flow from the north and south meet in a zone, 30-46 km wide, which is closer to Cuba's south coast. Afternoon clouds and precipitation tend to appear along this zone. Therefore, it appears that Cuba is a model island which exhibits usually well-understood features of the sea breeze phenomenon.

8.2.2. The Islandsof Fiji, New Caledonia,and Niue. Influence of the abtlence or presence of mountains on the sea breeze phenomenon is best exemplified by air flow over the islands of Niue, New Caledonia, Fiji, and Java (Fig. 16). The islands of Niue, Fiji, and Java are small islands of comparable sizes. Niue centered a t 19"N and 120"Eis a flat island about 24 km across and it displays no sea breeze a t all. Apparently, the thermally induced pressuregradient forces from all sides cancel out and no sea breeze results. However, strong sea breezes are observed on the islands of Fiji (16"S,179"E)and Java (7"S, 1lO"E).There appears t o be a direct relationship between the depth of the sea breeze circulation and heights of the local mountains. The island of New Caledonia, like Cuba, is a long island. Centered a t 21'20'5 and 165"15'E, it is about 50 km wide and 400 km long. A mountain ridge 500-1000 m high runs parallel to its length along the WNW-ESE direction. This ridge shields the island's west coast from the southeasterly trade winds and reinforces the sea breeze effect which reaches a peak of 30-35 knots in the afternoon. 8.2.3. Shower Bands Produced by Air Flow around the Island of Hainan. An interesting diurnal mesoscale effect is produced by the interaction of the synoptic scale flow with the island of Hainan centered a t 19"lO'Nand 1lO"E (Fig. 17). The low-lying flow in the northern part of the South China Sea is easterly to northeasterly during lull periods. Atmospheric layer above the island becomes increasingly unstable due to heating of the land mass, and permits the prevailing winds to carry the marine air directly across the island. During the night, air above the island becomes very stable and forces the marine air to the north and south around Hainan. The two flows meet in the lee of the island and cause a convergence zone t o be formed. As a result bands of showers are formed in a region 30 km wide and 160 km long in the west or southwest direction from Hainan. Two to four parallel bands of showers are commonly observed. A line of showers is also formed along the coastline of North Vietnam. Winds t o the north or northwest of the line of showers are from E t o ENE with 5 t o 10 knots speeds whereas 12 to 20 knots

386

KULDIP P. CHOPRA

southerly winds prevail in the region south or southwest of the line. The line and bands of showers dissipate during the mid-morning hours.

8.2.4. Cloud Rows Leeward ofthe Nantucket Island. Periodically spaced rows of small cumulus clouds are frequently observed leeward of several islands in the Woods Hole area on sunny summer days. The appearance of these cloud streets is suggestive of island-induced convective motions leading t o vertical oscillations in the atmosphere. The island of Nantucket is the first island subjected t o a planned experimental and theoretical study as part of a broader program to investigate atmospheric convection and the relation between convective motions, their energy sources, and the atmospheric structure prior to their onset. Nantucket was chosen because of its small size and smooth flat surface. Located at about 41'15" and 70°5'W, the island is 5 to 10 km wide in the N-S direction and about 20 km long in the E-W direction. Its small size suggests that the scale of convective motions should be small enough for the influence of the Coriolis force to be negligible. Nowhere is the terrain more than 15 m above sea level, and it is free from all obstructions (tall trees, buildings, etc.). Therefore, heating of air from below is rather uniform. These features are helpful in introducing simplifying assumptions in theoretical analysis . Data gathered on August 8, 9, 14, 15, 26, 28, and September 5, 1960 were reported by Malkus and Bunker (1962). Summary of their data is presented in Table VII. No cumulus appeared on August 9 and 25. Varying degrees of clouds appeared on the other five days. Very well-defined cloud rows were formed on August 8 and 14. These two cases represent exactly opposite directions of basic wind flow over the island so that the cloud formation cannot be attributed to topography, warmer downwind waters, or any other local peculiarity. More specific data for these two days, appropriate for application to or for comparison with the Malkus-Stern (1953) heated island model are presented in Table VIII. It is interesting to note that the upperlevel wind Strengthened and remained in the same direction as the lower-level wind on 8 August, 1960. I n the other case (August 14), upper-level wind strengthened but completely reversed. The observed spacings of cloud rows were 1.6 km and 1.0 km on August 8 and 14, respectively. Figure 18 is a schematic drawing of the Nantucket cloud row observed on August 14, 1960. It is quite apparent that solar heating of the island is a necessary, though not sufficient, condition for convective motions leading t o the formation of cloud rows. It also appears that the formation of a well-mixed layer from the ground up to cloud base normally precedes the development of cloud-scale convective motions. The surface layer over the island was very shallow and poorly developed on the two days cumulus clouds did not appear.

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

387

I n their linearized steady state model for a heated island, Malkus and Stern (1953) considered a flat island of 5 km width along the x-axis, parallel to the ambient wind u, and infinite in extent along the y-axis t o explain the above phenomenon. This reduced the problem to two dimensions and the influence of the Coriolis parameter could be neglected. They further assumed u to be independent of x and z [along the vertical, the flow to be incompressible (except for heating an4 buoyancy)], inviscid, turbulence-free, and the atmosphere in hydrostatic equilibrium. I n terms of the undisturbed stability

and the displacement $ of the streamlines, due to the heating function V , and related t o the vertical velocity = d+/at = u(a+/aX)

(8.2) they obtained

+

Here, V2 = aa/ax2 a2/az2is the two-dimensional Laplacian operator, g is the acceleration due to gravity, c, is the constant-pressure specific heat, T, is the mean undisturbed absolute temperature of the air layer, 0 is the potential temperature, is the dry adiabatic lapse rate, and CL is the actual lapse rate. Equation (8.3) is similar to that studied by Scorer (1949, 1956) and Queney (1948) for air flow over mountains except for the neglect of the wind-shear term in Eq. (8.3) and the mountain problem being adiabatic ( V = 0). Malkus and Stern (1953) assumed a heating function V of the form

r

(8.4) where Q is the surface heating rate of air in calories per gram per second, h is the height a t which V reduces to l / e of its surface value, and q(x) describes its dependence on x. They chose a rectangular pulse of unit amplitude and width d (island width), repeating after an interval of 2L, to represent q(x). If d 4 2L, q(x) can be expressed as the sum of a Fourier series. Adopting L = 5d and the origin of x a t the center of the island, they considered three cases: (a) the infinite vertical extent model, (b) the rigid lid model, and (c) the two-layer model for the atmosphere. The lower boundary condition in all three cases requires that each harmonic in $ be zero for all values of x a t z = 0. It is also necessary that the amplitude of each harmonic of the streamline displacement $n must remain finite irrespective of the height z. For the infinite vertical extent model, the streamline displacement is given by ni

(8.5)

$ = 1 P,[{exp( -z/h) n= 1

+

- cos(Bz))sin(rwx/l) sin(Bz)cos(mx/l)]

W

00 00

TDLE VII. Summary of Nantucket cloud row observations, Summer of 1950" CaSe Date

Occurrence of cumulus

Lower layer Depth (m) Upwind Lapse rate (OC/lOO m) Upper layer Depth (m) Upwind Lapse rate (OC/lOOm) Maximum insolation (cal/cm2min)

2 8/8/50

3 8/9/50

4 8/14/50

Many

None

Many

5 8/15/50 1st period

None 2nd period Slight

6 8/25/50

7 8/28/50

8 9/5/50

Non-convective fractocumulus

Very small lenticular

Many

500

500

2100

- 0.36

+0.25

+0.65

w

1000

150

935

+0.85

-2.66

+0.70

500

1050

320

+0.30

+0.62

+0.59

300 200 +0.01 +0.36

450 400 +0.77

500

745

+0.42

* 0.0

1.5

1.5

3600 +0.42

+0.83

1.3

1.2

1.4

1.6

Missing (est. 1.2)

$

E'di

Mean wind Layer (m) U

0-1000 3.7 m/s from N

0-1200 lomls from WSW slight turning toward W

0-900 900-2100 1.2 m/s southerly 3.0 m/s northerly

0-1000

6.3 m/s southwest turning to west with height

0-900 900-1800 3.3 m/s northeasterly 2.1 mps south-

0-1800 5.2 m/s SW

0-1500 8.2 m/s from 360" No change in direction b

southwesterly

Evidence of vertical oscillation

Preferred locations of cloud development 1.5 km apart

Downdraft indicated by dry region at lee shore

Aglkm

N

Lifting condensation level based on lowest point on upwind sounding ( 35m)

~ 8 5m 0

N

Definite lee waves

323 m

438 m

Downdraft indicated at lee shore Possible indication of lee waves by small cumulus

313 m

updraft possibly indicated by rise in isentropes 2 km inside windward shore

340 m

Updraft Long, large amplitude a t 600 m at lee lee waves shore suggested indicated by forby small mation of clouds. cumuloThose down- nimbus stream 17km spaced downwind closer than 1 km

-

220 m

690 m (a parcel at 440 m needs only 160 m lifting for saturation)

2 8 #

d

c

2 U

8 F?

3 cd

8

2

0 m i 5 M

E

W 00

CD

After Malkua and Bunker (1952).

TABLE VIII. Pmdicted vs. observed wavelengths of the lee waves essociated with the Nantucket Island"

81

(10' em-l)

u1 (m/s)

Date August 8, 1950

1.25

August 14, 1950

1.00

(ga/u% Sharpest (1O'O cgs) transition level (m)

3.0 from N

13.9

1.5 from S

44.0

After Malkus and Stern (1953). From surface to 3 km height.

1600 900

82

(10' em-l)

1.65b 1.30D

u, (m/s)

(ga/ua)z

(lolocgs)

6.5 from N

4.0

4.0 from S

8.Ob

Wavelength Effective (,n2~2/Z2)1010 lid height n=1 hPrSd AObs (km) (km) (km)

2.0

2.5

1.8

1.5

?!

2

' 0

d!

0

1 .o

10.0

1.1

1.0

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

391

B = [(gs/u2)- ( ~ ~ n ~ / L ~ ) ] l ’ ~ and nl is the highest harmonic for which gs/u2 > .rr2n2/L2.Normally, n, H 20, but only the first few terms are significant. For a sample calculation with cal/g/s and h = 1 km, they noted the following unusual data, Q = 2 x features revealed by the model streamline pattern: (1) The predicted maximum vertical displacement $I of the streamlines is 215 m a t z = 800 m, just t o the lee of the island center. ( 2 ) The predicted maximum vertical velocity of 218 m/s (averaged over a horizontal distance of 1.5 km) occurs a t z = 1 km. (3) The surface wind decreases as the island is approached from the windward side and increases as the leeshore is approached. This backward sea breeze reverses a t z = 600 - 800 m. (4) The repeated lee oscillations are absent in the predicted streamline pattern. I n the rigid lid model, the layer of air above z = 1, the lid height, may be assumed t o have infinite wind speed so that $In = 0 a t z 5 1. The solution for the streamline displacement becomes ni

(8.7)

t,4

=

C P,,[exp( -z/h)

- COB Bz +C sin Bz]sin(mx/L)

n= 1

with (8.8)

C =[COBBZ - exp( -l/h)]/sin B1

As before, the harmonics n > nl contribute negligibly. There is a fundamental difference between the solutions for $I given by Eqs. (8.5) and (8.7). The third term in Eq. (8.7) may become very large, positive or negative, for one or more harmonics of the solution up to n = n,. For typical values of wind speed and stability, and for 1.5 < Z(km) < 3.5, the harmonics of the solution are amplified relative to one another in such a way as to contribute updrafts over the island and produce a regular sea breeze circulation. The amplified harmonics appear as lee waves with wavelength given by (8.9)

A, = 2 L / n

Since the lid must act as a node and energy reflector, waves with integral half wavelengths in z contained between the ground and the lid are amplified. Therefore, (8.10)

1= NAJ2

392

KULDIP P. CHOPRA

where N is the largest integer which gives the largest number of amplified harmonics. The sin Bz term in these harmonics is dominant if (8.11)

A,

= 2x/B = 2x[(gs/u2)- (nn/L)a]-1’2

Combination of Eqs. (8.9)-(8.11)yields for the wavelengths of the lee waves (8.12)

A,,

= 2n[(gs/u2) - (NT/z)a]-1’2

The integer N goes through values 1, 2, etc., as long as gs/u2> ( N x / l ) 2 . For typical values of gs/u2 and 1, N assumes values 1 and 2 before the wavelengths become complex. No harmonics are amplified for 1 = 0 to 1.22 km, only the first harmonic is amplified for 1 = 1.22 to 2.44 km, and two harmonics are etrengthened for 1 = 2.44 to 3.66 km. Also, for a prescribed 1, amplified wavelength increases as gs/u2 decreases. I n a sample calculation with 1 = 2.6 km, the two amplified harmonics have wavelengths of 2.7 km and 7.0 km, respectively. The lower wavelength should correspond to observed spacing of cumulus clouds downwind of the island on days with conditions similar to those of the model. The larger wavelength may explain the observation that every second or third cloud is taller than its neighbors. Cumulus cloud rows are formed only if the air is sufficiently moist and conditionally unstable. Eyebrow-like lenticular8 appeared downwind of Nantucket Island on August 15 and 28 when the latter condition was not met. The model strictly applies to unsaturated air but indirectly accounts for the condensation process. Release of latent heat by each cloud row may be regarded as a small heat source placed at a point of maximum lifting. Such heat sources, spaced periodically a t points of maximum lifting, would intensify the perturbation motions existing prior to each local saturation. For 1 = 2.6 km, the model also predicts (i) a regular sea breeze with 2 m/s speed; (ii) the 400 m vertical displacement of streamline at 600-800 m height; (iii)the maximum updraft speed of 72 cm/s; and (iv) the side by side appearance of the largest updraft and downdraft a t the leeshore of the island. In the two-layer model, a fast-current layer is superposed on top of a slow-current layer and replaces the rigid lid of the previous model. This renders the case analogous to the previous one with a weak (quasirigid) lid separating the two layers, and partial reflection of wave components in z-direction occurring a t the level of separation of the two layers. Here, gs/ua is smaller in the upper layer than its value in the lower layer. In actual cases, there would be a continuous change in gs/u2 within a transition layer. Malkus and Stern (1953) found that it is possible to use Eq. (8.12) t o predict the wavelengths of lee waves by defining an equivalent lid height in real cases where the mean value of gs/u2 in the upper layer is about onehalf or less than that in the lower layer, and the transition is rather rapid. The equivalent lid would then be placed somewhat above the transition level by

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

393

an amount roughly inversely proportional to the decrease in gs/u2. They applied these considerations to the data on cloud rows of August 8 and 14, 1950 in the lee of Nantucket. The data in Table VIII reveal that on these two days there was a rapid decrease of gs/u2 a t some height between 1 and 2 km due in each case to very sharp increase in wind speed. They estimated the equivalent lid height on August 8 and 14 a t 2 km and 1 km,respectively. The corresponding predicted spacings between cloud rows are 1.8 km and 1.1 km as compared to 1.5 km and 1.0 km on these two days, respectively. Whereas the two-layer model correctly explains the cloud row spacings on August 8 and 14, 1950, i t failed to explain why any indications of cloud rows or lee oscillations failed to appear on August 9 and 25 when similar conditions prevailed. To understand the cause of this failure, Stern and Malkus (1953) examined in detail the mechanism for heating of air above the island. Observations indicate that the heating in real cases increases far more sharply as the air first crosses the upwind shore of the island. The temperature increase is very rapid a t first and slows up as the air temperature approaches that of the warmed ground. Furthermore, the heating is rarely evident above 1 km. It also appears from observations that a well-mixed surface layer extending up to the cloud base (300-800 m) must precede the formation of cloud rows. Hence, Stern and Malkus (1953) found that a suitable heating function V would obey an equation of the form (8.13)

v/cp = aept = K a2T’/ax= = u(aT’/az) w(r - a)

+

where T‘ is the perturbation in temperature. Equation (8.13) states that the heat supplied t o air produces two effects: (i) it raises the internal energy of air, producing a horizontal temperature gradient along the wind direction, and (ii) it increases the potential energy of the stratified air by doing work on the upper air by lifting. Observations on Nantucket Island had shown that the two terms in Eq. (8.13) are equal at an elevation of 500 m. Most of the heat supplied t o air is furnished well below this height. Near the ground, large scale convective motions are not important. Therefore, the second term in Eq. (8.13) can be neglected, and the temperature field in this region is set up and maintained by small scale turbulence so that (8.14)

Ka2v/aZ2 = u(av/az)

Well above the ground, V is very small, and nearly adiabatic convective motions prevail which produce horizontal temperature gradient. The distribution of the heating function throughout the ground layer can be determined from the surface temperature profile, and may be written as (8.15)

7 = c,(gs)1’2(ik)p(k,O)exp[-(ik)1’2(uz/K)]

394

KULDIP P. CHOPRA

v

where and are the amplitude of the z-dependence of each harmonic, characterized by k, of the heating function end temperature, respectively. The rate of heating decreases exponentially with height a t a rate determined by the scale of turbulent mixing. The greater the wind speed u and the eddy wavenumber k, and the smaller the eddy conductivity K, the greater is the rate of decrease with height of the heating function. The streamline displacement may be divided into two components and $2 such that (8.16)

$=$i-$a

The first component is a solution of the heat conduction equation and is important only near the ground. The other component $2 obeys the equation of air flow over a mountain ridge. Since $2 is subtracted from the equivalent mountain is inverted. The streamline component i,h2 oscillates between near positive and negative. Thus the downdrafts in $2 cancel updrafts in the ground, and the total solution produces updrafts which increase upwards. The equivalent mountain corresponding to a heated island can be specified analytically in terms of the ambient wind speed u, eddy conductivity K in the ground layer, the undisturbed stability 8, and the temperature profile along the surface. The equivalent mountain has the following features: (8.17)

Maximum amplitude

A = ./(r-

(8.18)

Effective height

M = A ( l - l/e)

(8.19)

Location of peak

xo = u3/gsK (measured from the upwind side)

(8.20)

Level of maximum updraft

h =~u(4g.9)-”~

(8.21)

Mean updraft at level h

w = Mu/x,

Here r is the effective temperature excess of the island, and xd is the island width traversed by air. The mountain is geometrically similar to the surface temperature profile, and it is about one fifth times as high in kilometers as ground temperature at x o exceeds the water temperature in absolute degrees. The amplitude A of the equivalent mountain is related to the maximum streamline displacement. It is approached exponentially with distance from the windward aide of the island. The elevation of the mountain becomes proportional to the excess ground temperature at xo = 3 km downwind of the upwind shore for u = 2 mls, 8 = lo-’ cm-’, a’ = 10 km, and K = 26.7 ma/s. Changes in u and K change xo , but not A. For broad islands (x,,< a’), shape and height of the equivalent mountain are determined entirely from the surface temperature profile, undisturbed lapse rate u,and are independent of the detailed features of the mixed surface layer. When xo N d, windward and

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

395

leeward slopes of the mountain are effected by u and K, leaving the maximum height unchanged. When xo >d, M < A, and xo is pronounced. Thus, in addition to the island heating and change in gs/u2 a t 1-3 km height, formation and extent of the equivalent mountain is a very critical condition for the appearance of cloud rows. Vertical transport of heat due to eddying motion lifts the streamlines over the island upward in a mountainlike shape causing the overlying air flow to be deformed just as if a physical mountain were present. The height of this equivalent mountain depends on the formation and extent of the ground mixing layer, and it may vary by an order of magnitude from one sunny day to the next. The surface layer is well mixed and developed on days with low s and small u. Light winds and deep mixed layer in the Cape Cod area usually occurs two or three days following the passage of a cold front. On these days, a change in gs/u2 occurs at 1-2 km level. August 14, 1950 was one such ideal day for cumulus cloud rows. The island was in the weak-wind area of the cold high following a polar-front passage three days earlier, and an equivalent mountain of 700 m height developed. During the interval directly following the cold front, the northerly winds are too strong and dry to produce cumuli. Equally unfavorable for the cumulus clouds are the days when stable, southwesterly winds prevail. August 9, 1950 was such a day when only 25 m high equivalent mountain developed. The equivalent mountain model suggests a relationship between the islandproduced cumulus clouds and their initial energy sources: Only those clouds directly over the island derive significant amounts of energy from the upward transport of sensible heat and moisture through the mixing layer, and those downstream, formed in forced convective updrafts derive their energy from the air stream. The theory also predicts a normal sea breeze with increase in horizontal wind on the windward shore and a corresponding decrease in horizontal wind on the leeward shore. The reversal in sea breeze occurs at z = ~ u ( 4 g s ) - ' / ~ on the upwind side of the island. The horizontal wind difference across the island is (8.22)

AU = Mu/h

with sea breeze at each shore being roughly Au/2. Sea breeze is strong for large K and for small s and u.

8.2.5. Rain Induced by the Anegada Island. Based on the equivalent mountain theory of heated islands put forward by Stern and Malkus (1953), Black and Tarmy (1963) proposed that precipitation over dry areas may be increased by asphalt coatings of ground. Intrigued by this suggestion,Malkus (1963) analyzed rainfall over a small flat island comparable in size to the

396

KULDIP P. CHOPRA

contemplated asphalt coatings. The island of Anegada (18"45'N, 62"ZO'W)is 3 km wide and 16 km long with its length oriented in almost EW direction. The island is so flat that nowhere does the elevation exceed 10 m. The data analyzed by Malkus (1963) were collected on Mardh 26,1953 in an independent mission to study the open ocean trade wind flow, its wind structure, dynamics, and clouds. Anegada was chosen because of its upwind location in the West Indies and its small size. It was hoped that the island's effect on the ambient air flow would be negligible. Results of this investigation proved t o the contrary.

SHOWER CLOUDS

FIG.40. Cloud row formed by Anegada Island observed at noon on March 26, 1953 (after Malkus, 1963).

Clouds over Anegada arranged in a single row, 24 km long, parallel to the air flow (Fig. 40). The surface wind was from the ESE a t 5 m/s. The significant features of the observations are as follows. (1) Anegada produces showering clouds in a weak disturbance in dry season. Island heating is a necessary condition for the appearance of cloud row which grows and disappears with sun. Of the ten other cases analyzed, cloud row was absent only once when the sky was overcast. (2) Only island-produced clouds produced precipitation. Clouds produced by Anegada are about 1.8 km thick whereas oceanic cumuli are about 100 m thick. (3) No rain fell on Anegada on March 26, 1953. Rain fell from downwind clouds, 5.2 km apart, along the wind. The island is desertlike, and it appears that not enough rain falls on it.

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

397

Applying the data:

K

u = 5 m/s, 7

= 3"C,

CI = 6.7'C

km-l,

= 1.2 x

loe cm2/s = 1.12 x 10-7 cm-1

xd = 9 km, and 1 = 5 km

to the Stern-Malkus (1953) theory, Malkus (1963) obtained the following values for the various parameters related to the equivalent mountain :

A = 900 m h = 784 m

M

= 575 m G = 32 cmls

Sea breeze = Au/2 = 1.8 m/s

xo = 9.3 km Au = 3.6 m/s

X = 5.4 km,

kbs= 5.2 km

There is an excellent agreement between the predicted and observed values of A, G,and nu. The effective height of the mountain is attained a t a distance xo N 9 km from the upwind shore. It is fortuitous that x,,11x,, N 9 km, the effective island width traversed by air. The cloud pattern reveals that the greatest disturbance (largest cloud) is produced a t the leeward edge of the island. However, this tallest cloud does not produce any rain. Precipitation fell from three equally spaced slightly lower clouds downstream of the tallest cloud. I n short, the small flat island of Anegada produced a large enough " equivalent mountain " to cause a disturbance in the ambient air flow and produce showering cumuli.

8.2.6. Showering Clouds Leeward of the Grand Bahama Island. Following Malkus' (1963) lead in the study of the shower-producing clouds leeward of the island of Anegada, Bhiimralkar (1972) carried out a very extensive experimental and theoretical investigation of a similar phenomenon associated with Grand Bahama Island (GBI), (Fig. 41). Grand Bahama is a narrow, long, and flat island. Centered a t approximately 26'50" and 78"20'W, the island is 10 km wide and 130 km long, and it is oriented in the approximate E-W direction. The terrain is nearly flat, vegetated by pine trees of 10-12 m height. Sea to the north of the island is much shallower than that to the south. Studies of disturbances caused by heated islands are of particular significance because of their persistence over tropical oceans. Occurrence and properties of these disturbances are related t o the size, shape, topogaphy, and orientation of the island with respect to the prevailing wind. The structure of the ambient wind and the temperature excess of the island over surrounding water also plays a very crucial role. Observations have shown that when the prevailing flow is along the longer axie of the heated island, a single convective cell tends t o develop more or less equidistant from the coastlines.

398

KULDIP P. CHOPRA

LITTLE ABACO

\

MIAMI /

/ 2S0N

'

ANDROGQ

FIQ.41. Grand Bahama Island and its vicinity.

For wind at an inclination to the island, the updraft moves toward the leeward coast, sometimes several kilometers downwind of the island. The rainfall associated with the heated island is usually most pronounced at and just downwind from the leeward edge. Field program carried out by Bhumralkar (1972) during August 20-29, 1970 included measurements by an instrumented aircraft and a special surface network over the island. Panoramic cloud photographs and time lapse movies of the clouds were taken. To identify the heated island effect, data were gathered on days free of any large scale regional disturbances. Results of the data analysis are summarized below. (a) The heated island (GBI) can produce significant disturbance in the ambient flow t o produce rainfall. (b) The leeward edge of the island is the preferred region for the production of this disturbance. (c) Cooling of the environment produced by evaporation of falling rain causes the disturbance to propagate in the upwind direction. (d) Direction, speed, and structure (thermal, moisture) of the prevailing tiow have significant influence on the formation of this disturbance. The stronger flow is not likely to produce as large a disturbance as the weaker flow from the same direction.

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

399

(e) The life span of 8 disturbance produced by a heated island is 3 to 34 hours. To explain these observed features, Bhumralkar (1972) has developed a general nonlinear theoretical model of a heated island. Developments in this direction started with the linear, steady state model developed by Malkus and Stern (1953). Smith (1955, 1957) developed a time-dependent linearized model of air flow over an infinitely long flat island of 200 km width. The extent of the land mass considered by Smith is an order of magnitude greater than in Malkus-Stern model, so that cumulus size bubbles of air would not play an important role. Smith found that as heating progresses over an otherwise calm island, subsidence with associated inflow of air appears over land, and wavelike disturbance in vertical velocity propagates outward from the center of the land mass. A prevailing wind C8UBeS the vertical velocity to fall off less rapidly with height than in the zero wind case, the perturbations being carried to greater heights in the moving air stream. Estoque and Bhumralkar (1969) developed a two-dimensional dry model. They performed integration of the nonlinear conventional meteorological equations by incorporating effects of the surface boundary layer as well as the penetrative convective processes. Their results showed that the maximum acceleration occurs over leeward edge of the island. This finding has been supported by observations over the island of Barbados (Garstang, 1967). Estoque and Bhumralkar attribute this leeside effect more t o the pressure-gradient force and advective processes than to eddy transfer of momentum as suggested by Garstang. The Bhumralkar (1972) model also applies to an infinitely long island and includes microphysical processes of accretion, condensation, evaporation and diffusion. The growth and evaporation of precipitation accompanied by changes of phase provides sources and sinks of heat which influence air motion. Release of latent heat during condensation provides additional buoyancy which may sustain an updraft and promote its subsequent growth. Also, mixing with cloud environment and evaporation of precipitation can lead to cooling and the formation of a downdraft. The model atmosphere assumes hydrostatic equilibrium, and it incorporates features of the rigid-lid and two-layered models. The Bhumralkar model of the heated island includes horizontal and vertical diffusion of momentum, heat, and moisture. Dynamic and thermodynamic influences of water substance are also considered. The ice phase is excluded from these considerations. Account is taken of the convective mixing of temperature and moisture on the subgrid scale through parameterization, and lateral boundaries are placed far away from the region of large disturbances. The model was applied t o GBI conditions on August 27 and 28, 1970. The moisture, temperature, and wind direction were identice1 on these two

400

KULDIP P. CHOPRA

days. On August 27,a significant heated island effect appeared in the form of mesoscale dieturbances. Bhumralkar simulated the life span of the induced disturbance by integration of the model equations by finite difference techniques over a cartesian grid of minimum mesh size of 1 km in the horizontal and 200 m in the vertical. The model satisfactorily reproduced the observed disturbance, particularly the cloud and rainfall patterns, and predicted a life span of 2 t o 2+ hours for the disturbance as compared with the observed 3 to 34 hours. On August 28, the wind speed more than doubled that on the preceding day, and the model predicted no significant disturbance in agreement with observations. The model defines a parameter

(8.23) to characterize the intensity of the disturbance. Here u and v are the parallel and normal components of wind and the subscript c refers t o the control experiment. The large temperature excess r of the island causes an intense disturbance. The strength of the parallel and normal wind components have opposite influence on the magnitude of the disturbance. A stronger parallel component intensifies the disturbance and a stronger normal component weakens it. I n general, greater the value of B, more intense is the disturbance. The model shows a dependence of the heat island effect on the island width d. The small widths (2 and 4 km) in the numerical modeling experiment predicted one peak in precipitation, and the larger width (10 km) produced two peaks. I n all three cases, the predicted location of the maximum disturbance is 1 to 2 km downstream of the leeward edge of the island.

8.2.7. Air Flour over the Island of Barbados. To explain certain anomalous features of the observed air flow over some tropical islands, Garstang (1967) proposed a momentum exchange model and carried out observations on air flow over the island of Barbados, W.I. He noted the following discrepancies between the typical t3ea breeze models and some observations over the tropical islands. (a) There are significant variations in time of onset, magnitude, and direction of the sea breeze observed on tropical islands embedded in characistically steady current, (b) Surface wind speeds decrease on the windward side and increase on the downwind side of Barbados and certain islands in the Pacific as the heating advances. (c) Rainfall over Barbados occurs mainly at night in the absence of any organized atmospheric disturbances in the region.

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

401

(d) Moisture content is observed t o decrease during the day over a n island or a coastline. To explain these occasional discrepancies, Garstang ( 1967) proposed an eddy momentum flux model in which he assumed an infinite island of width 2a and a coastal transition zone of 26 a t each coast. The modre1 considers an initially steady, two-dimensional flow in which influences of the pressuregradient and the Coriolis forces are neglected. The eddy exchange coefficient A,, is assumed constant with z initially, irrespective of its dependence on x. With the assumption of the horizontal wind profiles (8.24)

u = u0

+AZ

over the sea, and (8.25)

u=u,+Az+C(Z-z)

over the land, the simple model yields for the average vertical velocity (8.26)

G = -(C/46)L(2Z-

L)

Here A and C are the vertical gradients of the horizontal wind speed in the turbulent mixing layer of characteristic lengths L and 1 over sea and land, respectively. Nature and intensity of the vertical motion @ is governed by Eq. (8.26)and is determined by the halfwidth 6 of the transition zone, the vertical shear component au/az of the horizontal wind field and the characteristic heights 1 and L of the turbulent mixing layers over land and sea. Depending on the sign of C = aulaz, four cases of interest arise: (1) C > 0, 21 > L: downward motion on the upwind side and upward motion on the downwind side of the island. (2) C < 0 , 21 > L : upward motion on the upwind side and downward motion on the downwind side of the island. (3) C > 0 , 21 < L: upward motion on the upwind side and downward motion on the downwind side. (4) C < 0 , 21 < L : downward motion on the upwind side and upward motion on the downwind side. Garstang regards situations (1) and (3) typical of daytime and nighttime, respectively, on a tropical island. He considers cases (2) and (4) possible and occasional on tropical days and nights, respectively. To substantiate his model, Garstang (1967) reports planned observations he carried out over the island of Barbados. Located a t 13"N and 60"W, Barbados is a small island with smooth relief. It is 24 km wide in the E-W direction and 40 km long in the N-S direction. The highest point has an elevation of 240 m. Measurements over the ocean display maximum speed, instability, and turbulent exchange a t night and

402

KULDIP P. CHOPRA

minimum during the early afternoon. Measurements of wind components indicate an average horizontal speed of 7 m/s over land and 5.75 m/s over the ocean. Ascending motions have speeds between 0.3 and 1.0 m/s, larger values applying over the sea. Descending motions have speeds in the range of 0.2 to 1.1 m/s, both over land and ma. Computed values of shear stress increase progressively from 0.4 to 1.6 g/ cm/s2 in the downwind direction over the island. At the leeward coast, shear stress suddenly drops to a low value and abruptly jumps to a value comparable t o the one observed over land. From the diurnal variations of wind on the east and west coasts, Garstang estimated the maximum value of the low level convergence at 9.3 x 8 - l a t 0200-0300 LST. The estimated maximum value of divergence during the day is6.44 x 10a t 1500-1600 LST. The change of sign of convergence occurs a t sunrise and sunset. His estimates for the parameters needed for Eq. (8.25) are

C = 1.4 x L=300m

s-l

6 = 2.5 km 2=600m

which yield a vertical velocity 6 = 38 cm/s, which is of the correct order of magnitude for regions of maximum vertical motions.

8.3. Urban Heated Islands Urban and suburban areas provide surfaces and activities which produce thermal contrasts and simulate a heated island effect. Because of its bearing on the quality of city air, the urban heat island effect has received increasing attention during recent years. Bornstein (1968), Chandler (1962)) Clarke (1969), Davidson (1967), Findlay and Hirt (1969), Landsberg (1956), Lowry (1967), Mitchell (1961), Myrup (1969), Preston-Whyte (1970), Sundborg (1950), and Woollum (1964) have reported the existence of the heat island effect associated with several cities. For the most part, the cities studied are industrial centers with comparatively large populations. However, it has been found in studies conducted a t Pa10 Alto, California (Duckworth and Sandberg, 1954), Corvallis, Oregon (Hutcheon et al., 1967) and Chapel Hill, North Carolina (Kopec, 1970) that small nonindustrialized cities also exhibit pronounced heat island effects. These cities are small university centers. It has recently been shown by Chopra and Pritchard (1971,1972)and Chopra, Pritchard, and Thompson (1972) that shopping centers located within urban areas may produce significant heat island effects on a microscale. The ,shopping center heat island effect is similar to, and perhaps more intense than, the small city effect. The existence oft) heat island effect for an urban shopping center can be attributed to essentially the same causes which are responsible for the

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

403

overall city heat island effect. These include uneven solar heating of the shopping center and the surrounding area, greater evaporation of water from plants and trees in residential areas, and greater concentration in the shopping center of heating and air conditioning systems, automobiles and people. Figures 23 and 24 show the thermal features of the Ward's Corner shopping center area on April, 16, 1971 and the Ward's Corner and the Southern shopping centers on May 6, 1971. Similar patterns were observed over these areas during the summers of 1971 and 1972. These shopping centers are located in Norfolk, Virginia, and were chosen for a number of reasons. The Ward's Corner is a small symmetrical shopping center, and there is a very sharp distinction between the commercial and residential areas. A major traffic intersection is located at the center of the Ward's Corner, and the sea breeze is usually strong along one of the streets in the afternoon. It was anticipated that the magnitude AT, of the heat island effect would be substantial under favorable circumstances, that the influence of the sea breeze could be studied, and that the symmetry would facilitate the development of a mathematical heat island model. Observations have provided support for these expectations. AT, was 7°F on April 16, 1971, and the sea breeze produced strong mixing (not shown in Fig. 23) along Granby Street. The Southern shopping center was chosen €or reasons of contrast and complexity. It is close to the Ward's Corner and field measurements could be made conveniently for both areas on a single field trip. Commercial buildings are arranged in two groups on opposite sides of a major street and are further separated by extensive parking areas. As a result of an underpass, there is no major trafic intersection, the sea breeze is weak and has no strongly preferred direction. Also, the center is asymmetrical and the contrast between commercial and residential areas is not as sharp as around the Ward's Corner. A numerical model for the spatial distribution of temperature is set up using Fourier series

T = P + 1C , sin[( m x /L) ++,,I

(8.27)

along the x-direction. Here T is the mean temperature along a traverse line, C , and 4, are the amplitude and phase angle of the nth harmonic, x is the distance from the origin, and L is a characteristic length. Figure 42 a and b shows several traverse lines drawn on the isotherm maps. With L = O.Gmiles, Chopra and Pritchard (1972) found that the Ward's Corner heat island effect can be described by a single equation (for all traverse lines) (8.28)

T=

+ AT,[0.36

sin(300z + 87")

+ 0.06 sin(900x + 92")]

+0.13 sin(600x + 273")

404

KULDIP P. CHOPRA

FIG. 42(a) Isothermal map of the Ward’s Corner shopping center heat island on April 18, 1971 with traverse lines suporimposed. (b) Thermal pattern of the combined Ward’s Corner and Southern shopping centers with traverse lines superimposed.

405

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

where is the space mean temperature of the area, and AT,,, is the maximum temperature difference. Only three harmonics are necessary to describe the phenomenon associated with Ward's Corner and the second and third harmonics accounting for the asymmetry. The combined influence of the two shopping centers cannot be described by one equation, and it needs inclusion of larger number of harmonics. The vertical circulation patterns generated by this effect have a significant bearing on the quality of air and hence should be considered in city planning. For example, two neighboring shopping centers could generate a circulation pattern shown in Fig. 43 with more concentration of the undesirable gaseous elements in the central residential regions and the heavier particulate materials drifting towards the shopping areas. r

Ir - - ---- -1 ty It r;rjl

--------'i

I

-

ii I

L---2-----I WSIOENTIAL

i

I I

SPOPPING

:-----1 1 1I---

1 1

L-------I ENTER

I I

L--3--

R f S l D f NIIAL

I

I

tl t i f 1 J

1 L -,

SWOCPING CENTER

lESlDfYTlAL

Fro. 43. Complex convective circulation patterns caused by two neighboring shopping centers.

More recently, numerical transport and circulation models pertaining t o urban heat islands have been proposed by Egan and Mahoney (1972) and by Olfe and Lee (1971). The advection-diffusion model developed by Egan and Mahoney is useful in making estimates of air pollution concentrations under conditions of spatial and time varying emissions, transport velocities, and diffusion rates. The Olfe-Lee study of the convective effects associated with urban heat islands follow the pattern set by Estoque and Bhumralkar (1969). They consider an initially planar flow in a stable atmosphere with constant stability and eddy diffusivity. Their model predicts the effects observed over urban areas: (1) positive surface perturbation in temperature which tends to cancel the early morning inversion, and (2) the negative temperature disturbance at upper levels which tends t o produce one or more weak inversions a few hundred meters above the city. The computed flow field shows a descending motion directly over the upwind portion of the urban heat island. This is in agreement with observations on certain natural heated islands and with nighttime observations over a city. Their model includes (a) different planar surface temperature distributions, (b) the three-dimensional case of a circular heat island, (c) a two-layer atmosphere having a change in stability a t a suitable altitude, (d)constant eddy viscosity for the perturbed flow, and (e)the Coriolis force.

406

KULDIP P. CHOPRA

I n a recent note, Rao (1972) demonstrates the possibility of detection of the urban heat island features by infrared radiometers aboard present operational weather satellites. In particular, he analyzed the digitized I R data for October 19, 1970 obtained over the east coast of the United States a t 0300 local time. General locations of Baltimore, New York, Philadelphia, and Washington, D. C., were indicated and thermal contrasts of 2" -5°C were discernible. 9. RECENTEXPERIMENTS IN TROPICAL ISLAND METEOROLOGY Two extensive field programs in tropical island meteorology have been carried out in recent years. These programs included meteorological and oceanographic observations over two island regions in the tropics using conventional surface techniques and observational systems aboard aircraft, earth satellites, and ships. These experiments are called the Line Islands Experiment (LIE) and the Barbados Oceanographic and Meteorological Experiment (BOMEX). 9.1. The Line Island Experiment ( L I E ) The Line Islands consist of eleven atolls, straddled about the Equator between 6"N and ll"S, and south of the Hawaiian Islands. The three islands chosen for the experiment are the islands of Christmas (2"N, 157"30'W), Fanning (4"N, 159"30'W), and Palmyra (6"N, 162"W) shown in Fig. 44. The island of Christmas is about 1930 km south of Honolulu, Fanning is 400 km northwest of Christmas, and Palmyra is about 290 km northwest of Fanning. The three islands lie on a straight line, hence the name "Line Islands Experiment. '' These islands are circles of coral reefs with lagoons in the center and are embedded in the predominantly easterly trade winds. I n spite of their small size, lack of topographical features, and closeness to each other, the three islands have characteristically different climates. The island of Christmas receives yearly average rainfall of only 18 inches and is predominantly arid. The island of Fanning receives about 80 inches of rainfall yearly, and has a dry season during August to November. The island of Palmyra is the wettest island, it receives an average of 180 inches of rain each year and has no dry season. The sharp contrast in climates of the three small neighboring islands is due to the influence of the meteorologically significant local low pressure, light wind zone known t o mariners as the doldrums and to the climatologists as the intertropical convergence ( I T C )zone. The ITC zone is somewhat north of the equator, and it swings seasonally a little t o the north and south of its mean position. The island of Christmas is outside of the ITC zone and is dry.

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

407

TROPIC

FIU.44. Location map of the Line Islands Experiment (LIE).

The island of Fanning is in it except during the dry season, and the island of Palmyra is in it most of the time. The Line Islands Experiment conducted on the islands of Christmas, Fanning, and Palmyra during February-April 1967 has resulted in unique and comprehensive aircraft, satellite, and surface-based data for studies in meteorology of the here t o fore data-void region of the equatorial Pacific. The data include carefully planned and well-coordinated radiometric surveys of sea surface temperature; rawinsonde observations with good vertical resolution and free of orographic effects; good research aircraft data on winds a t 150 m altitude; 50 days of continuous weather radar coverage, cloud photography, surface, and reconnaisance observations; and photographs from several satellites including 40 daily pictures from ATS-1. These data are available in catalogs prepared by Chaffee and Bunker (1968), Estoque (1970), Lanterman et al. (1967), Madden and Robitaille (1970), Madden and Zipser (1970), Wyrtki (1967), Yonker (1967), and Zipser and Taylor (1968). The Line Islands Experiment was planned for early spring because the islands serving as observation sites would extend from inside the ITC t o well outside it. Main objectives of the LIE were (a) To provide a data sample for basic observational studies of meteorological phenomena in the oceanic portion of the ITC.

408

KULDIP P. CHOPRA

(b) To provide a meteorological data sample with which to comprehensively evaluate cloud photography performed by the ATS-1 satellite, geosynchronous a t 150°W, and to evaluate the limits of satellite technology. (c) To serve as a pilot program for more extensive Tropical Oceanographic and Meteorological Experiment (TROMEX) and the Global Atmospheric Research Program (GARP) in the future. Toward this end, it provided an opportunity for simultaneous use of the conventional and satellite data to learn as much as possible about meteorological processes on sub-synoptic scale in the vicinity of the ITC. Zipser (1970) has reviewed some of the scientific results and their significance. A summary is presented below. (a) Careful processing of the rawinsonde data and filtering out the vertical wavelengths smaller than 700 m reveal a multilayered wind structure, especially in the meridional component (Madden and Zipser, 1970). At least seven distinct layers were observed between surface and 20 km. The layering is strongest above 14 km and below 9 km, and it is better marked a t Christmas than at Palmyra. Very large vertical wind shears are observed occurring most frequently near the tropopause. Considerable turbulence is usually associated with large shear, and therefore these regions may represent significant sinks of kinetic energy. The layered winds above 14 km and 9 km are probably associated with vertically propagating waves of lengths on the order of 2 km. The wind variations in the 9-14 kni region appear t o be linked to changes in position and intensity of circulation features in either hemisphere rather than a regular progression of wave disturbances. (b) Estoque (1970) has analyzed the double theodolite pilot balloon observations a t Christmas Island to study the planetary boundary layer wind distribution. He finds the winds predominantly easterly with a small backing with height. His analysis suggests that (i) the E-W component of the wind is approximately geostrophic, (ii) the N-S component is highly nongeostrophic, (iii) the change in wind direction with height is controlled mainly by the altitude variation of the horizontal pressure gradient, and (iv) the eddy components cannot be computed with sufficient reliability for determining stress. (c) Analysis of 150 m altitude wind data shows strong backing of the wind with height at Christmas, decreasing to negligible magnitude at Palmyra. This suggests a connection with the cold surface water at the equator which is verified by direct observations during the LIE of upwelling and cold water along the equator, strong easterlies along the equator, and maximum development of the equatorial dry zone. These observations have strong implications concerning meridional circulations near the equator. (d) Infrared radiation measurements above Christmas and Palmyra show significant differences which may form the basis for models of radiative

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

409

cooling in the tropics. At Christmas Island, the clouds rarely reach the 850 mb level and therefore subsidence must be the main process balancing the cooling. On the other hand, cooling a t Palmyra is for the most part compensated by the convective heat flux. Thus, radiative cooling is linked to the maintenance of the equatorial dry zone. (e) Records of surface wind speed a t Palmyra show that (i) even a small atoll can have important influences on the surface wind, and (ii) the energy in the power spectrum is mostly concentrated in the 5-7 day period range. (f ) Power spectra of the meridional and zonal wind components a t heights up to 28 km derived from the rawinsonde data provide evidence of wave motions in the lower stratosphere with downward phase propagation and horizontal wavelength of 10,000 km. The coherence between levels is significantly high over a depth of about 5 km. The 4-5 day peak of the meridional wind, characteristic of these waves, was most marked a t 22 km during the LIE, and the westerlies change to easterlies a t 26-27 km level. (8) The situation in regard to wave motions in the lower troposphere is puzzling. No spectral peak appears at Fanning and Palmyra, and a weak 4-5 day spectral peak, corresponding t o 6-7 km long waves with upward phase propagation, shows up a t Christmas. A well-defined disturbance was observed on March, 22-23, 1967 as a wave trough, eventually developing double-vortex structure with cyclones on both sides of the equator. This disturbance corresponded to an intermediate wavelength of 3500 km. (h) Zipser (1969) studied a disturbance which passed through the LIE observational area. This disturbance intensified rapidly just east of the Fanning Island during the night of March 31-April 1, 1967 and dissipated rapidly during the daylight hours of April 1. The convergence-divergence patterns associated with the growth and decay of the disturbance were most intense in the lowest 500 m. Rawinsonde and aircraft data showed the organization of convection-produced unsaturated downdrafts which in turn organized into squall systems on the mesoscale. The squall systems were important agents of energy transfer in the disturbance and their feedback t o the larger scale (500-1000 km) disturbance was so powerful that the entire active cloud system collapsed within a few hours. There is evidence that mesoscale squall systems modify the structure of these disturbances so profoundly that they cannot be regarded as being steady state even for a period of one day. Furthermore, a relatively cool dry squall downdrift acts as a strong constraint on tropical cyclone development.

9.2. The Barbados Oceanographic and Meterological Experiment ( B O M E X ) The large and complex BOMEX is designed to study just one phase of Earth as a heat engine. Air-sea interaction plays an important role in the operation of this engine. Short wave solar radiation received by Earth is

410

KULDIP P. CHOPRA

stored primarily in the upper layers of the tropical oceans. Earth radiates in the infrared rather uniformly at all latitudes. The resulting net gain of energy in tropical latitudes and net energy loss in the polar regions maintains an energy flux from low t o high latitudes via atmospheric, and quite remarkably not by oceanic, circulations. According to Kuettner and Holland (1969) this energy transfer occurs in three steps: (i) transfer of energy from the sea surface t o an atmospheric boundary layer, 1.5-1.8 km deep; (ii) convective transfer of energy from the boundary layer into deeper layers of the troposphere; and (iii) cross-latitudinal transport of energy from equatorial t o polar regions. Although responsible for atmospheric circulations on the largest scale, energy transfer from sea surface is governed by turbulent processes on the microscale. The primary component of the turbulent energy flux is the latent heat of evaporation. Simultaneous exchange of sensible heat, radiative and kinetic energy occurs in many directions. Wind, sea state, and air-sea humidity and temperature differences further complicate this process. Transport of energy through the deeper layers of the tropical atmosphere occurs in spurts of short-lived convective processes on the meso and synoptic scales. The primary objective of the BOMEX was to concentrate on the air-sea interaction and to gain some insight in the convective energy transfer process. The project was divided into five major subprograms: (i) Sea-air interaction program, called the Core program; (ii) Oceanographic program; (iii) Radiation program; (iv) Satellite program; and (v) Tropical exploration program. The observational phase of BOMEX was carried out in a 500 x 500 km land-free region of the tropical Atlantic to the east of the island of Barbados (Fig. 46) MayJuly 1969. Observations were made from satellites, 12 oceanographic ships including Navy’s FLIP (floating instrument platform) and Florida State University’s buoy Triton, and 28 research aircraft. The observations provided data for 100 investigations in which 1500 persons participated. The observations included aircraft measurements of radiation surface temperatures and heat flux; lidar and radar measurements of cloud and haze layers, mean and eddy velocities, temperature and humidity; satellite measurements of cloud and temperature distribution; buoy measurements of turbulent flux, dissipation, and other statistics; ship measurements of ocean structure, turbulence statistics in the ocean and atmosphere, radionuclide concentrations, cloud distribution from radar, boundary layer and upper air velocity, temperature and humidity; land-based radar measurements of clouds and hf radio backscatter from the sea. Bulletins, reports, and cloud photograph atlases on the progress and data on BOMEX are periodically released by BOMAP office of the National Oceanic and Atmospheric Administration (NOAA). A temporary archives description of available data has been compiled by de la Moriniere (1972).

ATMOSPHERI~ AND OCEANIC FLOW PROBLEMS

41 1

Fro. 45. BOMEX ship array during May 3-July 2, 1969.

Summarizing the BOMEX Symposium, Pleagle (1970) notes that the raw data reveal a number of generalizations including the following: Both the atmospheric and oceanic data show greater inhomogeneity in time and space than is expected for the BOMEX area. Surface temperature varies significantly over distances less than the width of the array. Near-surface currents, revealed by current meters and by the drift of FLIP were highly variable and appear t o be somewhat erratic. Radionuclide concentrations experienced marked coherent changes a t widely separated stations. Large changes in cloud structure occur as systems move within the easterly trade winds. Turbulence statistics based on observations made 30 m above the surface indicate that the temperature spectra exhibit maximum energy a t a horizontal

412

KULDIP P. CHOPRA

scale of about 14 m, whereas humidity spectra exhibit peak energy at or possibly beyond a horizontal scale of 500 m, suggesting dissimilarity in the processes of vertical heat and vapor flux. Intermittancy was characteristic of the turbulence observations with the result that the covariance calculated from 30 to 60 minute periods may underestimate the vertical flux over the area. Wisner (1971) analyzed the representativeness of the air-sea interaction data obtained from a variety of sensors and concludes that a ship does influence its temperature environment to a considerable extent. The air temperature over the ship is perturbed a t all levels even a t nighttime when i t is 0.1-0.2"C lower over the deck than a t boom height. The analysis also indicates that the boom measurements are more reliable than the deck observations when the relative wind blows off the sea past the boom and then over the ship. Fernandez-Partagas and Estoque (1970) present a preliminary analysis of the conditions over the BOMEX observation network during July 11-28, 1969. The average circulation for the period was found close to normal, the ITC zone was located near its mean July position, and the surface pressure observations indicated a large scale oscillation with an amplitude of 1 to 2 mb and a period of about 10 days. Six significant disturbances occurred during this period, one of which was a well-developed tropical depression, a warm core system. The remaining five disturbances showed characteristics of cyclonic circulations a t either high- or mid-troposhpere. At lower levels, these circulations are reflected as wave-type disturbances. Maximum amplitude in three of the disturbances lies in the middle or upper troposphere. The passage of each of these disturbances was preceded by a decrease in the easterlies and followed by an increase in the easterlies in the middle troposphere. The amplitude of each disturbance is maximum near 3 km height and is accompanied by a perturbation of the opposite sign with maximum amplitude a t about 12.2 km altitude. These disturbances have minima a t ground and 7.6 km levels. The analysis also indicates that the disturbances in the trades usually originate along the ITC zone and in the upper troposphere or as a result of an interaction between them. The upper level events form an important component of these disturbances which are dynamically and structurally more complex than the current models on tropical storma can explain. Reeves (1971) obtained horizontal divergences of the order of s- at 300 m altitude from a sample of nighttime aircraft wind measurements made on May 31-June 2, 1969. If these measurements represent approximately average divergence in the lowest 600 m thick layer above the BOMEX area, they imply a sinking rate of 60 m/day a t the cloud base level. The core Program of BOMEX was designed to provide data on sea-air flux of energy by three methods: (i) measurement of atmospheric budget

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

413

terms over a 500 km square; (ii) direct measurement of surface-layer vertical eddy fluxes a t various times and places within the square; and (iii) measurement of the major terms of the heat budget of the upper ocean a t each of the ship stations. The momentum flux was also t o be evaluated by the first two methods. Holland (1970, 1972) reports preliminary estimates of the evaporation rate, stress, sensible heat flux, and kinetic energy flux. The evaporation rate during undisturbed tradewind conditions is between 5-6 mm/day and the Bowen ratio is between 0.09 and 0.10. Evaporation, driven by wind speed and dry air supply, controls the sensible heat flux which is upward a t the surface but downward in the boundary layer as a whole. Water vapor plays three important roles. (i) Because of its large latent heat, it is the principal cooling agent a t the ocean surface, transferring energy in a nearly isothermal situation. (ii) It contributes to buoyancy on the microscale and thereby promotes convection when ordinary thermal convection is weak. (iii) Because of its modest saturation density, it is readily condensed in the troposphere, reconditioning the air to receive a fresh supply of vapor and energy. In the latter process, it releases heat in the troposphere and thus promotes stabilization of the lower atmosphere. There are negative feedback mechanisms to regulate each of these roles. For example, surface evaporation cools and destablizes the upper mixed layer of the sea, bringing warm water t o the surface and thus maintaining a nearly constant surface water temperature, even with varying evaporation rates. Similarly, the moist-buoyant convection on the boundary layer scale transfers sensible heat downward dissipating kinetic energy. Finally, the large-scale convective condensation and recycling of dry air is self-regulating by building up the stability of the troposphere. Based on BOMEX observations, Fleagle (1972) describes a conceptual model of air flow in the planetary boundary layer over the tropical oceans. The model consists of a series of helical rolls, one t o several kilometers in lateral size and 0.5-2 km in height, filling the planetary boundary layer. Horizontal velocities close to the sea surface below the center of the roll are about 1 m/s, and the maximum vertical velocities are 0.5 m/s. Water vapor is picked up by the air moving across the sea surface and is concentrated in areas of convergence. The higher local humidity makes these regions slightly unstable, and the resulting buoyant energy, organized by the boundary layer circulations is added t o the helical circulation along the line of convergence. In addition, higher frequency turbulence occurs preferentially in these regions of higher humidity. Appraising success of BOMEX, Fleagle (1972) identifies the following results of major scientific significance. (i) Vertical vapor flux on the synoptic scale can be evaluated with reasonable accuracy in undisturbed periods by budget equations applied t o both the atmosphere and the ocean.

414

KULDIP P. CHOPRA

(ii) Vertical flux of water vapor, heat, momentum, and other turbulence statistics can be determined reliably from aircraft observations as well as from fixed platforms. (iii) Temperature spectra in the surface layer exhibit maximum energy a t about 1 Hz (10 m wavelength) while humidity spectra exhibit peak energy a t 10-2-10-3 Hz (1-5 m wavelength). This striking difference, not observed over land or water in pre-BOMEX observations, indicates dissimilarity in the transfer mechanisms of water vapor and temperature. (iv) The planetary boundary layer is characterized by mesoscale structures of 1-5 km dimension. 10. CONCLUDINGREMARKS

The idea for this contribution to these Advances originated in an invited paper (Chopra, 1971) presented a t the 1971 spring meeting of the American Physical Society. Properties of the various atmospheric and oceanic flow configurations introduced by several islands have been described. Both the microscale and mesoscale perturbations are considered, and their significance to phenomena on the global (planetary) scale were discussed The results are summarized helow 1. Observational platforms, ships, and very small islands may generate microscale perturbations in wind and temperature. Observations on the Argus Island Tower and aboard ships during BOMEX testify to this conclusion. 2. Observations from an island do not necessarily represent the general environmental conditions in a region. However, observations from an atoll or a group of islands may serve as an observational network to describe the structural properties of the prevailing winds. Studies conducted at Kwajalein Atoll had revealed the layered structure of the Trades. The Line Islands Experiment provided similar and more contrasting feature in the equatorial region. More recently, the observational network has produced data which show significant time and space inhomogeneities and multilayered structure of the tropical atmosphere. 3. Most islands produce heated island effects in the form of land and sea breezes. Many islands produce an anomalous sea breeze effect in which the sea breeze is retarded on the upwind side and accelerated on the downwind side. Several islands produce mesoscale disturbances in the form of clouds, precipitation, gravity waves, and cloud rows which are formed on the leeward side or further downwind of the island. Whereas the island heating is necessary for the production of these disturbances, i t is not sufficient t o produce precipitation. Appropriate stability structure of the ambient wind

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

416

and the mechanism of surface-layer heating play an important role. A wellmixed boundary layer reaching up to the cloud-base level must form before cloud rows can appear. 4. The heated island phenomenon bears resemblance to the lake storm effect (Peace and Sykes, 1966). Local mesoscale weather disturbances frequently occur as cold continental polar air moves across the unfrozen Great Lakes during late fall and early winter. The heated island effect finds similarities in the urban heat island problem where it has a strong bearing on the quality of urban air. It also finds application in practical problems related to artificial production of rainfall in arid areas by coating surfaces with heatabsorbing substances like asphalt. 5 . Under conditions of stable stratification, an island may produce mesoscale vortex street in its wake. Several islands produce this phenomenon, but i t never develops in the wake of Hawaii. Vortices formed in the convergence zone in the lee of Hawaii are transported to the ocean surface by downdrafts and give birth to the longer lasting ocean eddies which form a very complex pattern in the Hawaiian waters. 6. Phenomena of wind driven counter-currents and upwelling along coastal waters are explored. It is predicted that the upwelling induced by winds along certain long islands may separate cold and warm surface waters astride the island, and generate subsurface counter current along the coastal boundaries. 7. The two recent major projects LIE and BOMEX are the forerunners of the more sophisticated project TROMEX and GARP planned for the future. The LIE and BOMEX data have yielded useful information on the layered wind structure in the tropical atmosphere and the trade winds, on the energetics and cyclogenesis of the tropical storms, and the role played by tropical mesoscale phenomena in the global atmospheric heat engine. I n particular, the BOMEX data will provide quantitative information on the energy sources and sinks over the tropical oceans. This information is needed to extend the numerical weather forecast period t o one or two weeks. Besides, LIE and BOMEX have provided a test for the limits of the present day technology and of coordination of varied and complex measurements and measuring techniques.

ACKNOWLEDGMENTS The manuscript was prepared under the most trying circumstances, and but for the constant encouragement by Drs. H. E. Landsberg and L. W. Webb, i t would not have made the press. Drs. W. J. Hargis and L. W. Webb arranged the logistio help for the preparation of this article. Donna McClellan expertly typed the manuscript, the artwork was prepared by Russell Bradley, Jim Cory, and Jane Davis, and Kenneth Thorneberry prepared the photographic art. Dr. D. Walsh made helpful editorial suggestions, and Drs. R. Ash, R. Byme, E. Kindle, W. Pritchard, L. Webb and J. Zeigler provided

416

RULDIP P. CHOPRA

opportunities for useful discussions. A summer appointment as Visiting Professor at the Virginia Institute of Marine Sciences provided an uninterrupted two months period devoted to this manuscript and the use of VIMS’ extensive facilities. All these inputs were timely and useful. The patient cooperation of Dr. Landsberg and his editorial staff is gratefully acknowledged.

REFERENCES Baralt, G. L., and Brown, R. A. (1965). “The Land and Sea Breezes: An Annotated Bibliography.” Department of Geophysical Sciences, University of Chicago, Chicago, Illinois. Berger, E., and Wille, R. (1972). Periodic flow phenomena. Ann. Rev. Fluid Mech. 4. 313-340.

Bhumralkar, C. M. (1972). An observational and theoretical study of atmospheric flow over a heated island. Ph.D. Thesis, University of Miami, Coral Gables, Florida. Billions, N. S. (1967). “Mesoscale Structure in the Tropics as Revealed by Comparative Rawinsonde Data for Kwajalein and Roi-Namur Islands, KTS,” Tech. Rep. No. 7277. Redstone Arsenal. Birkhoff, G., and Zarantonello, E. H. (1957). “Jets, Wakes, and Cavities,” pp. 280-293. Academic Press, New York. Black, J. F., and Tarmy, B. L. (1963). “The use of asphalt coatings to increase rainfall.” J. Appl. Meteorol. 2, 557-564. Bornstein, R. D. (1968). Observations of the urban heat island effect in New York City. J. Appl. Meteorol. I , 675-582. Brand, S. (1970). “The Effects on a Tropical Cyclone of Cooler Surface Waters Due to Upwelling and Mixing Produced by a Prior Tropical Cyclone, Tech. Rep. No. 18-70. Navy Weather Research Facility. Chaffee, M. A., and Bunker, A. F. (1968). Aircraft meteorological data from the Line Islands experiment. Woods Hole Ocean. Inst. Ref. No. 68-25. Chandler, T. J. (1962). London’s urban climate. Geogr. J. 128, 279-298. Chopra, K. P. (1966). Vortex streets in wake of Madeira Island. Bull. Amer. Phye. SOC. [2] 11, 708.

Chopra, K. P. (1968). Vortex streets leeward of islands. Trans. Anaer. aeophys. Union 49, 180.

Chopra, K. P. (1971). Island-induced atmospheric and oceanic circulations. Bull. Anaer. P h y ~SOC. . [2] 16, 487-488. Chopra, K. P. (19724. Velocity field in vortices leeward of islands. J . Atmoe. Sci. 29, 396-399.

Chopra, K. P. (1972b). On the existence of Taylor columns in planetary atmospheres. Bull. Amer. Phya. SOC.[2] 27, 523. Chopra, K. P. (19720). Lin parameter as characteristic of atmospheric vortex streets A I A A J . 10, 345-347. Chopra, K. P., and Hubert, L. F. (1964). Khrmhn vortex streets in earth’s atmosphere. Nature (London) 208, 1341-1344. Chopra, K. P., and Hubert, L. F. (1965a). Mesoscale eddies in wakes of islands. J. Atmoe. S C ~22, . 652-657.

Chopra, K. P., and Hubert, L. F. (196513). KOrmhn vortex streets in wakes of islands. A I A A J . 8, 1941-1943. Chopra, K. P., and Pritchard, W. M. (1971). Norfolk as an urban heat island. Va. J. s c i . 22, 91.

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

417

Chopra, K. P., and Pritchard, W. M. (1972). Urban shopping centers a s heated islands. Proc. Conf. Urban Climatol. let, pp. 310-317. Amer. Meteorol. SOC.,Boston, Massachusetts. Chopra, K. P., Pritchard, W. M., and Thompson, K. E . (1972). Heat, island studies of urban shopping centers. Trans. Amer. Geophys. Union 53, 388. Clarke, J. F. (1969). Nocturnal urban boundary layer over Cincinnati, Ohio. Mon. Weather Rev. 97, 582-589. Cromwell, J. (1953). Circulation in a meridional plane in the central equatorial Pacific. J . M a r . Rea. 12, 196-213. Davidson, B. (1967). A summary of the New York urban air pollution dynamics research program. J . Air Pollut. Con&. Ass. 17, 154-158. Deacon, E. L., and Webb. E . K. (1962). Drag coefficients of the sea surface. I n “ the Sea” (M. N. Hill, ed.), Vol. 1, pp. 57-66. Wiley, New York. Defant, F. (1951). Local winds. I n “Compendium of Meteorology” (F. Malone, ed.), pp. 655-672. Amer. Meteorol. SOC.,Boston, Massachusetts. de la Moriniere, T. C. (1972). BOMEX temporary archive description of available data. N O A A Tech. Rep. EDS 10. Duckworth, F. S., and Sandberg, J. S. (1954). The effect of cities upon horizontal and vertical temperature gradients. Bull. Amer. Meteorol. SOC.35, 198-207. Edinger, J. G. (1967). “Watching for the Wind,” p. 45. Doubleday, New York. Egan, B. A., and Mahoney, J. R. (1972). Numerical modeling of advection and diffusion of urban area source pollutants. J . A p p l . Meteorol. 11, 312-322. Ekman, V. W. (1905). On the influence of the earth’s rotation on ocean currents. Ark. Mat. Astron. Pya. 17, No. 26. Estoque, M. A. (1961). A theoretical investigation of the sea breeze. Quart. J. Roy. Meteorol. SOC.93, 136-146. Estoque, M. A. (1962). The sea breeze as a function of the prevailing synoptic situation. J . Atmos. Sci. 19, 244-250. Estoque, M. A. (1970). “The Planetary Boundary Layer Wind Near Christmas Island.” Report prepared under NSF Grant No. GA-1021. Inst. Mar. Atmos. Sci., University of Miami, Coral Gables, Florida. Estoque, M. A., and Bhumralkar, C. M. (1969). Flow over a localized heat source. Mon. Weather Rev. 97, 850-859. Felzenbaum, A. I. (1960). The theoretical basis for and methods of computation of steady oceanic currents. See citation in Rzheplinskiy and Shtokman (1968). Fernandez-Partagas, J. J., and Estoque, M. A. (1970).A Preliminary Report on Meteorological Conditions During BOMEX, Fourth Phase (July 11-28, 1969). Report prepared for NSF under Grant NSF-GA-10201. Inst. Mar. Atmos. Sci., University of Miami, Coral Gables, Florida. Findlay, B. F., and Hirt, M. S. (1969). An urban-induced meso-circulation. Atmos. Enwiron. 3, 537-542. Fisher, E. L. (1961). “Further Studies of Theoretical Sea-breeze Models,” Rep. No DA-36-039-SC-84939. Department of Meteorology, New York University, New York.. Fleagle, R G. (1970). Summary of symposium on “early results from BOMEX.” Bull. Amer. Meteorol. SOC.51, 319-321. Fleagle, R. G. (1972). BOMEX: an appraisal of results. Science 176, 1079-1084. Friday, E. W., and Wilkins, E. M. (1967). Experimental Investigations of Atmospheric Wake Trails,” Res. Inst. Rep. ARL-1576-3. University of Oklahoma, Norman. Frizzola, J., and Fisher, E. L. (1963). A series of sea breeze observations in the New York area. J . Appl. Meteorol. 2, 722-739.

418

KULDIP P. CHOPRA

Garstang, M. (1967). The role of momentum exchange in flow over a heated island. I n “Proceedings of the 1967 Army Conference on Tropical Meteorol,” pp. 178-195. University of Miami, Coral Gables, Florida. Gerrard, J. H. (1966). The mechanics of the formation region of vortices behind bluff bodies. J. Fluid Mech. 25, Part 11, 401-413. Goldstein, S. (1965). “Modern Developments in Fluid Dynamics, “Vol. 11, pp. 556-571. Dover, New York. Haurwitz, B. (1947). Comments on the sea breeze circulation. J . Meteorol. 4, 1-8. Heffter, G. L. (1965). The variation of horizontal diffusion parameters and travel periods of one hour or longer. J. Appl. Meteorol. 4, 163-156. Hidaka, K. (1964). A contribution to the theory of upwelling and coastal currents. Trans. Amer. Geophya. Union 35, 431-444. Hide, R. (1961). Origin of Jupiter’s Great Red Spot. Nature (London) 190, 895-896. Hide, R . (1971). On geostrophic motion of a non-homogeneous fluid. J . Fluid Mech. 49, Part IV, 745-775. Hide, R., and Ibbetson, A. (1966). An experimental study of Taylor columns. Icarus 5, 279-290. Holland, J. (1970). Preliminary report on the BOMEX sea-air interaction program. Bull. Amer. Meteorol. SOC.51, 809-820. Holland, J. (1972). The BOMEX sea-air interaction program: Background and results to date. N O A A Tech. Memo. E R L BOMAP-9. Homann, F. (1936). Einfluss grosser Zahigkeit bei Stromung um Zylinder. Forsch. Geb. Ingenieurw. 7 , 1-10. Hooker, S. G. (1936). On the action of viscosity in increasing the spacing ratio of a vortex street. Proc. Roy.SOC.,Ser. A 154, 67-89. Hubert, L. F., and Krueger, A. F. (1962). Satellite pictures of mesoscele eddies. Mon. Weather. Rev. 90, 457-463. Hunt, J. N. (1961). Stable vortex wakes near rigid boundaries. J . Math. Phye. (Gambridge, Maae.) 40, 33-40. Hunt, J. N., and Wickins, P. E. (1967). Vortex streets in the earth’s atmosphere. J. Pure Appl. Geophye. 67, 179-185. Hutcheon, R. J., Johnson, R. H., Lowry, W. P., Black, C. H., and Hadley, D. (1967). Observations of the urban heat island in a small city. Bull. Amer. Meteorol. SOC. 48, 7-9. Ingersoll, A. P. (1969). Inertial Taylor columns and Jupiter’s Great Red Spot. J. Atmos. Sci. 26, 744-752. Jenkins, G. R. (1945). Diurnal variation of the meteorological elements. In “Handbook of Meteorology,” pp.746-753. McGraw-Hill, New York. Kamenkovich, V. M. (1961). Integration of equations of the theory of ocean currents in multipli-connected regions. Dokl. Akad. Nauk. SSSR 138, No, 6. Kopec, R. J. (1970). Further observations of the urban heat island in a small city. Bull. Amer. Meteorol. SOC.61, 602-606. Kuettner, J. P., and Holland, J. (1969). The BOMEX project. Bull. Amer. Meteorol. SOC.SO, 394-402. Lamb, H. (1945). “Hydrodynamics,” pp. 228-229 and 680-684. Dover, New York. Landsberg, H. E. (1956). The climate of towns. I n “ Man’s Role in Changing the Face of the Earth.” (William L. Thomas, Jr. et al., eds.). pp. 584-606. Univ. of Chicago Press, Chicago, Illinois. Lanterman, W. S., Masterson, J., and Boogaard, H. V. (1967). The line Islands experiment. I n “ Proceedings of the 1967 Army Conference on Tropical Meteorology,” pp. 99-106. University of Miami, Coral Gables, Florida.

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

419

Lee, S. L. (1972). Fire research. Appl. Mech. Rev. 25, 503-509. Lin, C. C. (1959). On periodically oscillating wakes in Oseen approximation. I n ‘‘ Studies in Fluid Mechanics Presented to R. von Mises,” pp. 170-176. Academic Press, New York. Long, R. R. (1959). A laboratory model of air flow over the Sierra, Nevada Mountains. I n “The Rossby Memorial Volume,” pp. 372-380. Amer. Meteorol. SOC.,Boston, Massachusetts. Lowry, W. P. (1967). The climate of cities. Sci. Amer. 217, 15-23. McGary, J. W. (1955). Mid-Pacific oceanography. Part VI. Hawaiian off-shore waters, Dec. 1949-Nov. 1951. U.S., Fish. Wildl. Serv., Spec. Sci. Rep-Fish, 152, 1-138. Madden, R. A., and Robitaille, F. E. (1970). A comparison of equivalent potential temperature and static energy. J . Atmos. Sci. 27, 327-329. Madden, R. A., and Zipser, E. J. (1970). Multi-layered structure of the wind over the equatorial Pacific during the LineIs1andsexperiment.J. Atmos.Sci. 27,336-342. Makarov, S. 0. (1950). Oceanographic studies. Geografgiz pp. 207-208 (citation from Shtokman, 1966). Malkus, J. S. (1963). Tropical rain induced by a small natural heat source. J . Appl. Meteorol. 2, 547-556. Malkus, J. S., and Bunker, A. F. (1952). Observational Studies of the Air Flow over Nantucket Island During the Summer of 1950.” Pap. Phys. Oceanogr. Meteorol., M.I.T., Cambridge, Massachusetts, and Woods Hole Ocean Inst. 12, No. 2, 1-50. Malkus, J. S., and Stern, M. E. (1963). The flow of a stable atmosphere over a heated island. Part I. J . Meteorol. 10, 30-41. Manar, T. A. (1967). Progress in 1965-66 at the Bureau of Commercial Fisheries Biological Lab., Honolulu. U . S . , Fish. Wildl. Serv., Circ. 274, 1-51. Milne-Thompson, L. M. (1950). “ Theoretical Hydrodynamics, ”Chapter 13. Macmillan, New York. Mitchell, J. M. (1961). The temperature of cities. Weatherwise 14, 224-229. Moll, H. G. (1971). Die atmospharische Umstromung Madeiras. Beitr. Phys. Atmos. 44, 227-244. Myrup, L. 0. (1969). A numerical model of the urban heat island. J . Appl. Meteorol. 8, 908-91 8. Neumann, J. (1951). Land breezes and nocturnal thunderstorms. J . Meteorol. 8 , 60-67. Olfe, D. B., and Lee, R. L. (1971). Linearized calculations of urban heat island convection effects. J . Atmos. Sci. 28, 1374-1388. Palmer, C. E. (1967). Why have dynamic meteorologists been unsuccessful between the Tropics of Capricorn and Cancer? Keynote address in “Proceedings of the Working Panel in Tropical Meteorology,” Rep. No. 12-1167-132, pp. 1-2. Navy Weather Research Facility. Pannuto, R. J. (1969). A review of the sea breeze phenomenon. M. S. Thesis, New York University. Patzert, W. C. (1970). “Eddies in Hawaiian Waters,” Rep. No. HIG-69-8. Institute of Geophysics, University of Hawaii. Peace, R. L., and Sykes, R. B. (1966). Mesoscale study of a lake-effect storm. Mon. Weather Rev. 94, 495-507. Perrone, A. J., Gorman, J. M., and Weigle, F. G. (1965). “ Wind-speed and Wind Direction Distribution at Argus Island off Bermuda,” Rep. No. 683. U.S. Navy Underwater Sound Laboratory (USL). Pierson, W., Jr. (1950). “The Effects of Eddy Viscosity, Coriolis Deflection, and Temperature Fluctuation of the Sea Breeze as a Function of Time and Height,” Meteorol. Pap. 1, 1-20. Department of Meteorology, New York University, New York. I‘

420

KULDIP P. CHOPRA

Preston-Whyte, R. A. (1970). A spatial model of a n urban heat island. J. Appl. Meteorol. 9, 571-673. Proudman, J. (1916). On the motions of solids in a liquid possessing vorticity. Proc. Roy.SOC.,Ser. A 92, 408-424. Queney, P. (1948). The problem of air flow over mountains: A summary of theoretical studies. Bull. Amer. Meteorol. SOC.29, 16-25. Rao, P. K. (1972). Remote sensing of urban “heat islands” from an environmental satellite. Bull. Amer. Metwrol. SOC.53, 647-648. Reeves, R. W. (1971). Preliminary velocity divergence computations for BOMEX volume based on aircraft winds. NOAA Tech. Memo. ERL BOMAP-5. Roshko, A. (1964). On the development of turbulent wakes from vortex streets. NACA (Nat. Adv. Comm. Aeronaut.), 1191, 1-25. Rzheplinskiy, D. G., and Shtokman, V. B. (1968). An investigation of circulation about oceanic islands using numerical experimentation. Bul. Izv. Akad. Nauk SSSR. Fiz. Atmoa. Okeana 4, 1261-1274. Schaefer, J. W., and Eskinazi, S. (1959). An analysis of a vortex street generated in a viscous fluid. J. Fluid Mech. 6 , 241-260. Schmidt, F. H. (1947). An elementary theory of the land and sea breeze circulation. J . Meteorol. 14, 9-16. Scorer, R. S. (1949). Theory of waves in the lee of mountains. Quart. J. Roy. Meteorol. SOC.75, 41-56. Scorer, R. S . (1956). Air flow over an isolated hill. Quart. J . Roy. Meteorol. SOC.82, 75-81. Seckel, G. R., Charnell, R. L., and Au, D. W. K. (1967). The trade wind zone oceanography pilot study. Parts I-IV. U.S., Fish Wildl. Serv., Spec. Sci. Rep. 552-557, ‘1-422. Shotkman, V. 13. (1945). A transverse inhomogeneity of a surge wind as one of the important causes of horizontal circulation in the sea. Dokl. Akad. Sci. USSR 19, No. 2. Shtokman, V. B. (1966). A qualitative analysis of the causes of the anomalous circulation around oceanic islands. Izv. Akad. Nauk SSSR, Fiz. Atmoa. Okeuna 2 , 11751185. Smith, R. C. (1965). Theory of air flow over a heated land mass. Quart. J. Roy. Meteorol. SOC.81, 382-396. Smith, R. C. (1957). Air motion over a heated land mass. 11. Quart. J. Roy. Meteorol. SOC.88, 248-256. Smith, R. L. (1967). Note on Yoshida’s (1956) theory of coastal upwelling J. Geophya. Rea. 72, 1396-1397. Smith. R. L. (1968). Upwelling. Ann. Rev. Oceanogr. Mar. Biol. 6, 298-302. Smith, T. B. (1968), “Island Meteorology Study,” Rep. NWC-TP-4631. Naval Weapons Center. Sommerville, W. L., J r . (1958). The Sea Breeze at Norfolk Naval Air Station,” Rep. No. 12-1167-132. Navy Weather Research Facility, Norfolk, Virginia, Stern, M. E., and Malkus, J. S. (1953). The flow of a stable atmosphere over a heated island. Pt 11.J. Meteorol. 10, 106-120. Stone, P. H., and Baker, D. J., Jr. (1968). Concerning the existence of Taylor columns in atmospheres. Quart. J. Roy. ME&OTOZ. SOC.94, 576-580. Sundborg, A. (1950). Locd ~limatologicalstudies of the temperature conditions in an urban area. Tellus 2, 221-231. Sutton, 0. a. (1960). “Atmospheric Turbulence,” Chapter I. Wiley, New York.

ATMOSPHERIC AND OCEANIC FLOW PROBLEMS

42 1

Sverdrup, H. U. (1938). On the progress of upwelling. J. Mar. Res. 1, 155-164. Sverdrup, H. U., Johnson, M. W., and Fleming, R. H. (1942). “The Oceans,” pp. 486489. Prentice-Hall, Englewood Cliffs, New Jersey. Taylor, G. I. (1923). Experiments on the motion of solid bodies in rotating fluids. Proc. Roy. SOC.,Ser. A 104, 203-218. Thornthwaite, C. W., Superior, W. J., and Field, R. T. (1965). “Disturbance of Airflow Around Argus Island Tower near Bermuda,” Tech. Rep. No. 6295. U.S. Naval Oceanographic Office. Tsuchiya, K. (1969). The cloud8 with the shape of KArmBn vortex street in the wake of Cheju Island, Korea. J. Meteorol. SOC.J a p . [ 2 ] 47, 457-465. von KBrmBn, T. (1911). Uber den Mechanismus des Fliissigkeits- und Luftwiderstendes. 2. Phy8. la, 49-69. von KArmAn, T. (1912). tiber den Mechanismus den ein bewegter Korper in eines Fliissigkeit erfahrt. Goett. Nachr. (Math.-Phya. K l . ) pp. 547-556. Warren; B. A. (1967). Notes on translatory movement of rings of current with application to Gulf Stream eddies. Deep-sea Rea. 14, 505-524. Wilkins, E. M. (1968). Energy dissipated by atmospheric eddies in the wake of islands. J. Geophya. Res. 73, 1877-1881. Wisner, W. M. (1971). “Ship’s influence on surface and rawinsonde temperatures during BOMEX,” NOAA Tech. Memo. ERL BOMAP-6. Woollum, C. A. (1964). Notes from a study of the microclimatology of the Washington, D. C. area for the winter and spring seasons. Weatherwise 17, 263-271. Wooster, W. S., and Reid, J. L. (1963). Eastern boundary currents. I n “The Sea” (M. N. Hill, ed.,) Vol. 2, pp. 253-280. Wiley, New York. Wyrtki, K. (1963). The horizontal and vertical field of motion in the Peru current. Bull. Scrippa Inst. Oceanogr. 8, 313-346. Wyrtki, K. (1967). “ Oceanographic Observations During the Line Island Expedition,” Rep. HIG-67-17. Institute of Geophysics, University of Hawaii. Wyrtki, K., Burke, J. B., Latham, R. C., and Patzert, W. (1967). “Oceanographic Observations During 1965-1967 in the Hawaiian Archipelago,” Rep. No. HIG-6715. Institute of Geophysics, University of Hawaii. Wyrtki, K., Graefe, V., and Patzert, W. (1969). “Current Observations in the Hawaiian Archipelago, Rep. No. HIG-69-15. Institute of Geophysics, University of Hawaii. Yonker, T. L. (1967). The Line Islands experiment and ATS I data guide. I n “Studies in Atmospheric Energetics based on Aerospace Probings,” Annual report. Dept. of Meteorology, University of Wisconsin, Madison. Yoshida, K. (1955). Coastal upwelling off the California coast. Rec. Oceanogr. Work8 J a p . 2 , 1-13. Yoshida, K., and Mao, H. L. (1957). A theory of upwelling of large horizontal extent. J. Mar. Res. 16, 40-54. Zimmerman, L. I. (1969). Atmospheric wake phenomena near the Canary Islands. J. Appl. Meteorol. 8, 896-907. Zipser, E. J. (1969). The role of organized unsaturated downdrafts in the structure and rapid decay of an equatorial disturbance. J. Appl. Meteorol. 8, 799-814. Zipser, E. J. (1970). The line islands experiment, its place in tropical meteorology and the rise of the fourth school of thought. B d l . Amer. Meteorol. SOC.51, 1136-1146. Zipser, E. J., and Taylor, R. C. (1968). “A Catalogue of Meteorological Data Obtained During the Line Islands Experiment Feb.-April 1967,” Rep. HIG-67-19. Institute of Geophysics, University of Hawaii. Zubov, N. N. (1947). “ Dynamic Oceanography,” p. 321 (see citation in Rzheplinskiy and Shtokman, 1968).

This Page Intentionally Left Blank

AUTHOR INDEX Numbers in parenthesis are reference numbers and indicate that an author’s work is referred to although his name is not cited in the text. Numbers in italics show the page on which the complete reference is listed.

A

100, 118(148) 146, 147, 148, 150 Bearden, A. J., 97(102), 148 Bearden, J. A., 81, 145 Beeson, M. H., 128, 151 Belknap, J., 162, 260 Belon, A. E., 179, 181. 265 Bence, A. E., 94(80), 95, 101, 103, 147 Bender, P. L., 10, 17(18,19), 20(18,21), 21(21,24), 22(25), 54 Bender, P. L., 17(17), 54 Benioff, H., 24(29), 55 Bennet, W. R., 28, 55 Berger, E., 338, 341, 345, 354, 416 Berger, J., 32, 34, 39(49), 55, 56 Berger, M. J., 121(150b), 123, 150 Bertin, E. P., 79(34), 144 Bhumralkar, C. M., 397, 398, 399, 405, 412, 416, 417 Bilham, R. G., 39(50), 56 Billions, N. S., 299, 326, 327, 328, 416 Birkhoff, G., 330, 334, 337, 338, 416 Birks, L. S., 60, 143 Birnbaum, H. K., 94(73), 146 Bishop, H. E., 84(56), 99, 119(56), 146 Black, C. H., 402, 418 Black, J. F., 395, 418 Blagden, C., 168, 171, 260 Blair, J., 110(138), 150 Block, B., 44(45), 55 Boogamd, H. V., 407, 418 Booth, P. W., 279, 295 Borom, M. P., 109(133), 149 Born, M.,4, 50(7), 54 Bornstein, R. D., 402, 416 Bostrom, R. C., 24(30), 37(30), 55 Bottinga, Y., 134, 152 Boyd, F. R., 101, 148 Bradsell, R. H., 7(8), 54 Brand, S., 378, 416 Brooks, E. M., 280, 295 Brown, D. B., 119(150), 150 Brown, E. H., 132, 152

Abbot, C. G., 275, 276, 277, 278. 295 Adams, J. B., 134, 152 Agrell, S. O., 58(1), 125(1), 130, 143, 151 Akasofu, S. I., 179, 259 Akimoto, S., 132, 152 Albee, A. L., 94(80), 95, 101, 102(120), 103, 129, 141,14Y,149,151,154 Albertson, V. D., 213, 259 Aldrich, L. B., 276, 277, 278, 295 Alley, C. O., 17(17,18,19,21),20(18,21), 21(21,24), 54 Allison, S. K., 76(30), 144 Almasi, G. S., 110(138),150 Amaral, G., 135, 153 Anderson, A. T., Jr., 134, 152 Anderson, C. A., 59(4), 141, 143, 154 Angell, J. K., 288, 296 Angot, A., 162, 163, 165, 168, 173, 259 Arrhenius, G., 81, 139, 145 Astapowitsoh, I. S., 197, 260 Atherton, M. P., 133, 152 Au, D. W. K., 358, 420

B Baedsgaard, H., 134(198), 135(198), 153 Baird, K. M., 2(2), 27(34), 30, 53, 55 Baker, D. J., Jr., 354, 355, 420 Bannister, H. M., 164, 260 Baralt, G. L., 323, 384, 416 Barger, R. L., 29, 55 Barnard, F. A. P., 260 Barnes, P. E., 209, 265 Bartels, J., 261 Baum, T., 128, 151 Baun, W. L., 82, 145 Beals, C. S., 159, 161, 162, 165, 166, 167, 170, 171, 177, 193, 200, 209, 260 Beaman, D. R., 93, 94, 96(90), 99(90), 423

424

AUTHOR INDEX

Brown, J. D., 96, 99, 147, 148 Brown, L. C., 60(10), 61, 143 Brown, R. A., 323, 384, 416 Brune, J., 41(51), 42, 56 .Bryan, M. E., 202, 266 Bucklow, I. A., 97(103), 148 Bunch, T. E., 112(142), 136, 150, 153 Bunker, A. F., 386, 389, 407, 416, 417 Burder, G . F., 199, 260 Burhop, E. H. S., 87(61), 146 Burke, J. B., 365, 421 Burns, R. O., 129, 151 Bushee, J., 135(199), 153

C Caldwell, D. E., 73(25), 144 Caldwell, D. O., 121(150a), 123, 150 Caldwell, V . E., 109(134), 149 Cameron, E. N., 73(23), 144 Campbell, A. J., 108(127), 149 Campbell, W. H., 202, 260 Campbell, W. J., 96, 147 Capron, J. R., 163, 165, 168, 171, 172, 200, 260 Carmiohmel, I. S. A., 112(141), 150 Carrion, W., 17(17), 54 Carter, R. W., 97(104), 148 Cassidy, W . , 136, 153 Castaing, R., 68(2,3), 84(52,53), 93(70), 97(52,53), 108(128), 143, 145, 146, 149 Caverno, J. R., 172, 260 Chaffee, M. A., 407, 416 Chagon, C. W . , 274, 296 Chalmers, J. A., 214, 260 Chamberlain, J. W., 181, 260 Chandler, T. J., 402. 416 Chang, R. F., 17(17), 54 Chant, C. A., 161, 162, 164, 165, 166, 167, 168, 169, 170, 171, 172, 187, 201, 209, 260 Chapman, S.,165, 167, 179, 259, 260,

261 Chappell, B. W., 138, 153 Charell, R. L., 358, 420 Chayes, F., 101, 148 Chernosky, E. J., 174, 181, 261 Chodos, A. A., 102(120), 129, 141, 149,

151, 154 Chopra, K. P., 300, 310, 322, 323, 328, 331, 336, 337, 340, 341, 342, 343, 344, 345, 346, 347, 348, 351, 354, 402, 403, 414, 416, 417 Chree, C., 165, 171, 261 Clark, G. L., 79(35), 144 Clarke, J. F., 402, 417 Clayton, C. G., 68(18), 69, 144 Clayton, R. E., 213, 259 Clemesha, B. R., 287, 295 Clerke, A. M., 172, 261 Cole, A. E., 201, 261 Cole, R. K., Jr., 214, 261 Collis, R. T. H., 2(5) 53 Compton, A. H., 76(30), 144 Constable, F. C., 197, 261 Cooney, J., 2(6), 53 Cordani, U. G., 135(199), 153 Cork, J. M., 80, 145 Corlett, M., 129, 151 Court, A., 201, 261 Cresswell, G., 261 Cromwell, J., 377, 417 Currie, B. W., 158, 160, 167, 170, 171, 178, 190, 192, 261 Currie, D. G., 17(17,18,20), 18, 20(18), 21(24), 54

D da Casa, C., 97(96), 147 Dalton, J., 162, 196, 261 Davey, J. P., 130, 152 Davidson, B., 402, 417 Davidson, E., 67(15), 144 Davies, D., 39(50), 56 Davies, F. T., 158, 160, 167, 170, 171, 178, 192, 261 Davis, H., 199, 261 Day, D. E., 139, 153 Day, G. W., 2(4), 53 Deacon, E. L., 367, 417 Defant, F., 310, 312, 383, 417 Deirmendjian, D., 268, 272, 276, 278, 279, 283, 284, 285, 286, 289, 294,

295 de la Moriniere, T. C., 410, 417 Denning, W . F., 197, 261

425

AUTHOR INDEX

Derham, W., 162, 261 Descamps, J., 84(52), 97(52), 108(128), 145, 149 Dewey, R. D., 97(98), 98, 148 Dicke, R. H . , 17(18,19,20), 20(18), 21(24), 54 Dils, R. R., 114(144), 150 Dixon, F. E., 173, 261 Dodd, C. G., 82(47,48), 140, 145, 154 Dolezalek, H., 214, 261 Donati, G. B., 213, 261 Dratler, J., Jr., 44(45), 55 Duncumb, P., 71, 85(57), 86, 97, 99, 101, 108, 118(9l), 121, 123(91), 144, 146, 147, 148, 149 Dunkelman, L., 276, 295 Duckworth, F. S., 402, 417 Dunn, P. J., 22(26), 23(26), 54 Dwornik, E. J . , 128, 151 Dyer, A. J . , 287, 295 Dyer, G. R., 97(104), 148

E Earnshaw, K. B., 12(14), 15, 54 Ebel, H., 97(105), 148 Edinger, J. G., 299, 417 Egan, B. A., 405, 417 Egeland, A.. 201, 261 Einaudi, M. T., 136, 153 Ekman, V. W . , 379, 417 Elgin, R. L., 97(107), 148 El Goresy, A., 136, 153 Ellyett, C. D., 203, 261 Elvey, C., 171, 173, 261 Emmett, J. L., 1(1),53 Emslie, R. F., 132, 152 Englehard, E., 27(33,35), 55 Erickson, K. E., 54 Eskinszi, S., 336, 346, 347, 420 Essene, E., 132, 152 Estoque, M. A., 313, 384, 399,405, 407, 408, 417 Evans, B. W., 131, 133, 134, 152 Eve, A. S., 167, 170, 209, 261 Evenson, K. M., 2(3), 27(3), 53

Farrell, W. E., 42(52), 43, 56 Fawcett, J. J., 138(218), 153 Felzenbaum, A. I., 375, 417 Ferenczi, S., 158, 262 Fernandez-Partagas, J. J., 412, 417 Field, R. T., 299, 300, ,324, 421 Findlay, B. F., 402, 417 Finger,'L. W., 101, 148 Fink, R. W., 87(66), 88(66), 1 4 6 Fiocco, G., 280, 284, 286, 287, 295 Fischer, D. W . , 82, 145 Fisher, E. L., 312, 384, 417 Fitzgerald, R. W . , 99(110), 148 Fleagle, R. G., 411, 413, 417 Fleischer, M., 117(146,147), 150 Fleming, R. H . , 370, 421 Fletcher, H., 201, 206, 207, 262 Flowers, E. C., 288, 289; 296 Foghammar, R., 262 Force, P., 162, 164, 171, 172, 198, 200, 262 Boss, T. H., 129(166), 151 Fougere, P. F., 174, 261 Fowle, F. E., 276, 277, 278, 295 Fowler, R. A., 12(16), 54 Franken, P. A., 17(19), 54 Franklin, J., 171, 262 Franks, A., 67,143 Frazer, J. Z., 97(99), 98, 99, 148 Fredriksson, K., 134, 152 Freier, G. D., 211, 214, 215, 262 Freud, A., 198, 262 Prey, A. H . , 203, 262 Friday, E. W . , 300, 417 Fritz, H., 161, 162, 163, 164, 165, 168, 169, 171, 172, 176, 181, 184, 185, 200, 262 Frizzola, J., 312, 417 Froome, K. D., 7(8), 54 Frost, M. T., 101, 148 Fuch, L. H., 136, 153 Fuerstenau, D. W., 129, 151 Fujino, N., 126, 150 Fujita, H., 99(110), 148

G F Faller, J. E., 17(17,18,19), 20(18), 21(24), 48, 49, 50(56), 54, 56

/f-

G a p e , G., 272, 273, 296 Garber, C. M., 167, 198, 262 Garstang, M., 315, 399, 400, 401, 418

426

AUTHOR INDEX

Gartlein, C. W., 157, 162, 163, 164, 179, 180, 181, 201, 262 Gasparrini, E. L., 94(81), 101, 138(218), 147,148, 153 Gerard, J. B., 39(50), 56 Uerrard, J. H., 333, 418 Gherzi, E., 208, 262 Gibb, F. G. F., 138(218), 153 Gibbons, R., 108(127), 149 Gibbs, G. V., 139, 154 Glen, G. L.,82(47,48), 140, 145, 154 Goldstein, J. I., 126, 150 Goldstein, S., 338, 418 Goni, J., 130, 151 Gordon, E. I., 29(39), 55 Gorman, J. M., 324, 419 Graefe, V., 358, 421 Grams, G., 280, 284, 286, 287, 295 Greely, A. W., 192, 262 Green, M., 87(62), 92(62,92), 146, 147 Greenwood, W. R., 129(166), 151 Griffin, J. G., 166, 166, 169, 171, 197, 200, 207, 262 Griggs, R. F., 275, 295 Gromnica, E., 262 Gruening, E., 275, 295 Gulson, B. L.,138(215), 153

H Hadley, D., 402, 418 Hagan, M. P., 181, 261 Hagedoorn, H. L., 87, 146 Hekura, Y.,179, 264 Halbach, P., 129, 151 Hslford, D., 27(36), 55 Hall, J. L.,27(37), 29, 30, 34(42), 37, 38, 55 Hammond, J. A., 48, 49, 50(65), 56 Hanes, G. R., 30, 55 Hanneman, R. E., 109(133), 149 Hanzbly, S.,91(67a), 92(67a), 141(67a), 146 Harang, L.,262 firgraves, R.B., 126(155a), 127, 151 Hariis, L. A., 87(59), 146 Hartwick, A. J., 67(16), 144 Harvey, P. P., 262 Hasler, M. F., 67(14), 144 Haurwitz, B., 312, 383, 418

Hay, R. L., 137, 153 Heffter, G. L., 341, 418 Heinrich, K. F. J., 97, 98, 103, 111, 119, 120, 121, 122, 123, 147, 148, 149, 150 Helliwell, R. A., 208, 263 Hellwig, H., 27(36), 55 Henke, B. L., 97(107), 148 HBnoc, J., 84(53), 92(69), 97(53), 145, 146 Henry, J., 200, 213, 262 Heppner, J. P., 207, 266 Hernandez, E. M., 12, 15, 54 Hicks, B. B., 287, 295 Hidaka, K., 379, 380, 418 Hide, R., 341, 354, 355, 356, 418 Hinthorne, J. R., 59(4), 143 Hirt, M. S., 402, 417 Hogg, A. R., 282, 296 Holland, A. C., 272, 273, 296 Holland, J., 410, 413, 418 Hollander, N. B., 128, 151 Hollister, L. S., 126(155a), 127, 131, 151, 152 Holyoke, E. A., 162, 168, 262 Homan, F., 353, 418 Hood, R., 163, 262 Hooke, R. LeB.,137, 153 Hooker, S. G., 336, 418 Hopfield, H. S., 21, 54 Howie, R. A,, 94(79), 147 Hubbard, E., 197, 199, 262 Hubbert, J., 173, 262 Hubert, L. F., 300, 310, 323, 328, 331, 337, 340, 341, 342, 343, 344, 345, 346, 348, 351, 318, 416 Huggins, R. A., 114(144), 150 Hughes, G. D., 97(103), 148 Hunsucker, R. D., 263 Hunt, J. N., 310, 340, 345, 346, 418 Hutcheon, R. J., 402, 418 Hutchinson, R. O., 174, 261 Hutchinson, W. C. A., 211, 263

I Ibbetson, A., 355, 356, 418 Ilin, N. P., 129(165), 151 Iijima, A., 137, 153 Ingersoll, A. P., 354, 355, 418

42 7

AUTHOR INDEX

Innes, R. P., 74(27), 144 Irvine, W. M., 281, 282, 284, 296 Isasi, J. A,, 93(71,72), 94(73), 96(90), 99(90), 100, 146, 147

J Jackson, E. D., 128, 151 Jacobs, S. F., 28(38), Ei5 Javan, A., 26(32), 55 Jedwab, J., 136,153 Jefferies, B., 102(121), 105, 149 Jelstrup, H. S., 167, 263 Jenkins, G. R., 348, 418 Jenkins, R., 67, 68, 69, 104, 111, 143 Joannis, A., 168, 263 Johnson, A., 172, 263 Johnson, D. W., 211, 264 Johnson, G. G., Jr., 80, 81, 145 Johnson, J. H., 166, 263 Johnson, M. W., 370, 421 Johnson, R. H., 402, 418 Johnson, T. S., 22(28), 55 Jones, E. M., 101, 148 Jopson, R. C., 87(66), 88(66), 146 Jorgensen, T. S., 208, 263 Junge, C. E., 274, 285, 286, 296 K Kaelble, E. F., 76(29), 144 Kamenkovich, J. P., 375, 418 Kaminov, I. P., 54 Kantor, A. J., 201, 261 Katsura, T., 132(182), 152 Katz, A., 129, 151 Kaula, W. M., 17(17,18), 20(18), 21(24), 54

Kawashita, K., 135(199), 153 Keil, K., 60, 112(142), 126(6), 134, 136, 139, 143, 150, 153

Kelly, C. J., 111(140), 150 Kelly, T. K., 81, 90, 97(40), 98, 145 Kemble, F. A., 168, 263 Kerr, D. E., 10(11), 54 Kikuchi, G., 211, 263 Killingworth, P. J., 105, 149 Kimball, D. S., 162, 163, 164, 181, 201, 262

King, G. C. P., 39(50), 56

King, W. F., 165, 169, 263 Kirby, T. B., 91(67a), 92(67a), 141(67a), 146

Kisch, H. J., 133, 152 Klein, C. J., 133, 152 Kleinmann, B., 137, 153 Knowles, C. R., 94(80,81), 147 Koenigsfeld, L., 211, 265 Kolenkiewiez, R., 22(26), 23(26), 54 Koomen, M. J., 189, 265 Kopec, R. J., 402, 418 Korshover, J. K., 288, 296 Kramers, H. A., 78, 144 Kremser, G., 214, 263 Krueger, A. F., 341, 418 Kudo, A., 134(193), 152 Kuettner, J. P., 410, 418 Kushiro, I., 132( 182), 152 Kyser, D. F., 119(150), 150 L

Labuda, E. F., 29(39), 55 Lachance, G. R., 95(85,86), 147 Lamar, D. L., 197, 198, 265 Lamb, H. H., 290, 291, 296, 338, 418 Landsberg, H. E., 402, 418 Lanterman, W. S., 407, 418 Latham, R. C., 365, 421 Latourette, J. T., 28(38), 55 Lee, R. L., 406, 419 Lee, S. L., 330, 418 Lefroy, Capt., 164, 180, 263 Lemstrom, S., 164, 193, 263 Levine, J., 30, 34,(42), 37, 38, 55 Levinson, A. A., 141(231), 154 Lewis, R., 93(71,72), 128, 146, 151 Ley, W., 197, 263 Liebhafsky, H., 104( 123), 149 Liefeld, R. J., 91(67a), 92(67a), 141(67a), 146

Lihl, F., 97(106), 148 Lin, C. C., 337, 419 Lineweaver, J. L., 109,149 Long, J. V. P., 68(1), 60, 65, 73(7), 76, 96, 97(28), 99(28), 102(121), 103, 105, 111, 114(143), 116(28), 117, 124, 125(1,7), 130, 143, 144, 149, 150, 151 Long, R. R., 340, 419 Loomis, E., 213, 263

428

AUTHOR INDEX

Loseva, L. Ye., 129(165), 151 Lovberg, R. H., 32, 34, 39(49), 50(56), 61, 55, 56 Lovering, J. F., 138(215), 140, 153, 180, 181, 263 Lowry, W . P., 402, 418, 419

M McC., A. H., 164, 263 McClelland, D. C., 158, 263 McConnell, J . D. C., 109, 149 MacDonald, U. J. F., 17(17,18), 20(18), 21(24), 54 McFarlane, A. A., 66(11), 143 MoInturff, R. M., 288, 296 Mokay, D. S., 129( 166), 151 Macres, V . C., 99(112), 148 Madden, R. A., 407, 409, 419 Mmda, K., 202, 263 Mahoney, J. R., 406, 417 Makarov, S. O., 317, 372, 419 Malan, D. J., 206, 207, 263 Malkus, J . S., 315, 316, 323, 386, 387, 389, 390, 392, 395, 397, 399, 419, 420 Manar, T. A,, 367, 368, 419 Mao, H . L., 381, 421 Mapes, R. S., 97(98), 98(98), 148 Mark, H., 87(66), 88(66), 146 Marks, K . R., 208, 263 Martin, G. C., 275, 296 Martin, L. H., 206, 207, 265 Martin, P. M., 60(9), 87(66), 88(66), 96, 119, 125, 143, 146 Marton, L., 59, 143 Marvin, U . B., 136, 153 Mason, J . E., 273, 285, 286, 295 Mason, P . K., 101, I48 Masterson, J., 407, 418 Matarrese, L. M., 2(3,4), 27(3), 53 Matthews, S. W., 279, 295 Mead, C. W . , 125(153), 150 Mefferd, R. B., Jr., 199, 264 Meinel, A. B., 279, 280, 296 Meinel, M. P., 279, 280, 296 Melford, D . A., 108, 149 Melson, W . G., 137, 153 Michelson, A. A., 50(57), 56 Mielenz, K . R., 27(35), 55

Miller, A. J., 288, 296 Miller, D . R., 198, 264 Milne-Thompson, L. M., 338, 419 Mitchell, J . M.,402, 419 Moll, H . G., 354, 419 Moody, J. P., 264 Moore, J . G., 131, 152 Moore, R. K., 179, 180, 181, 262 Morton, R. D., 134(198), 135(198), 153 Mossop, S. C., 279, 286, 296 Mott, D., 91(67a), 92(67a), 141(67a), 146 Muhleisen, R., 212, 264 Mulholland, J . D., 17(17,18), 20(18), 21(24), 54 Mullen, L. D., 2(4), 53 Mulvey, T.,108(129), 149 Munk, W . H., 43(63a), 56 Murata, K., 84(54), 145 Myrup, L. O., 402, 419

N Nakamura, Y., 132(182), 152 Nefflen, K F., 27(35), 55 Neumann, J., 311, 419 Newcomb, S., 264 Newell, R. E., 287, 296 Nicholson, J . B., 67(14), 144 Nicolino, J. A., 68(16), 108(126), 144, 149 Nishikawa, M., 132(182), 152 Nissen, H . U., 134, 152 Norrish, K . N., 138, 153

0 Obayashi, T . , 179, 264 Oddie, B . C. V., 164, 264 Ogilvie, R. E., 95(82-84), 110(138), 147, 149 Olfe, D. B., 405, 419 Olmsted, D., 163, 172, 264 Olson, D., 173, 211, 212, 214, 264 Omholt, A., 201, 261 Ong, P. S., 108(130), 149 O’Nions, R. K., 82, 92(67), 134, 135, 140, 141, 145, 146, 1.53, 154 Ortner, B., 97(105), 148

AUTHOR INDEX

Owens, J. C., 10(10), 12(14), 54 Oxaal, J., 162, 164, 166, 166, 169, 171, 264

P Pakiam, J. E., 211, 264 Palmer, C . E., 313, 419 Pannuto, R. J., 384, 419 Parten, M., 160, 264 Patzert, W . , 368, 366, 421 Patzert, W. C., 310, 316, 317, 367, 368, 369, 360, 361, 363, 364, 366, 366, 367, 368, 369, 370, 371, 378, 419 Peace, R. L., 416, 419 Pedigo, J . E., 106, 149 Pell, W . J . , 172, 264 Peria, W . T., 87(60), 146 Perrone, A. J . , 324, 419 Peterson, F. W., 281, 282, 284, 296 Petrie, W., 196, 264 Pfeiffer, H . G., 104(123), 149 Philibert, J., 97, 99(93), 118(93), 147 Pierce, E . T., 207, 214, 261, 264 Pierson, W., Jr., 312, 383, 419 Pitsch, W., 84(66), 145 Pittock, A. B., 287, 296 Plotkin, H . H . , 17(17,18,19), 20(18), 21(24), 22(28), 54, 55 Pluth, D . J . , 74(27), 144 Pokorny, A., 199, 264 Polanyi, T., 29(40), 55 P o o h D. M., 60(9), 126, 87(66), 88(66), 90, 119, 126, 143, 146 Potter, W . E., 207, 264 Poultney, S. K . , 17(17), 54 Prescott, G. B., 213, 264 Presman, A. S., 199, 264 Preston-Whyte, R. A., 402, 420 Pritchard, W . M . , 324, 402, 403, 416,

41 7 Procunier, R. W., 201, 203, 217, 264 Proudman, J., 364, 420 Przybylski, A., 282, 296

429 R

Rabinowitz, P., 28(38), 55 Radtke, A. S., 73(24), 144 Rae, J . , 171, 264 Rank, O., 198, 264 Rao, P. K., 406, 420 Ray, D. J . , 173, 198, 264 Ray, L. A., 96, 147 Reed, S. J . B., 78, 80, 88, 89(33), 97, 98, 101, 114(43), 118(91), 121, 123(91), 138, 144, 145, 147, 148, 150,

153 Reeves, R. W., 412, 420 Reid, A. M., 134, 152 Reid, J . L., 377, 421 Reid, L. R., 73(21), 144 Remond, G., 130, 151 Rendall, R., 169, 265 Reynolds, C. A,, 111(140), 150 Reynolds, J . H., 136(199), 153 Reynolds, T . W., 97(98), 98(98), 148 Ribbe, P. H . , 94(77,78), 98(78), 129, 134, 147, 151, 152 Richardson, J., 171, 265 Richardson, K . A., 129, 151 Ringwood, A. E., 132, 152 Robitaille, F . E., 407, 419 Roedder, E., 128, 136, 151, 153 Roeder, P. L., 132, 152 Rohrer, R. H., 97(104), 148 Romig, M . F., 197, 198, 265 Roshko, A., 346, 420 Rouse, M . L., 168, 171, 265 Rowley, W. R. C., 27(36), 55 Rozenberg, G. V., 274, 280, 296 Rucklidge, J . C., 94(81), 101, 110(137), 126(166,167), 128, 138(218), 147, 148, 150, 151, 153 Russ, J. C., 68(17), 118(17), 144 Rust, C., 173, 261 Ryder, P. L., 84(66), 145 Rzheplinskiy, D. G., 374, 376, 376, 420

S

Q Queney, P., 387, 420 Quetelet, A., 264

Saastamoinen, J., 21, 54 Samuels, L. E., 116(146), 150 Sandberg, J . S., 402, 417

430

AUTHOR INDEX

Sargent, W. D., 213, 265 Saxena, S. K., 133, 152 Schaefer, J. W., 335, 346, 347, 420 Schmidt, F. H . , 312, 383, 420 Schmidt, R. A., 136, 153 Schmitz, U., 84, 145 Scholes, S., 109(135), 149 Schorer, G., 131, 152 Schwander, H., 126, 151 Schwartz, R. J., 110(138), 150 Scolnik, R., 276, 295 Scorer, R. S., 387, 420 Seckel, G. R., 358, 420 Sekera, A., 274, 296 Sekra, S., 289, 295 Seltzer, S. M., 121(150b), 123, 150 Sexton, S., 199, 265 Shah, G. M., 280, 296 Shalow, A. L., 25(31), 55 Sharp, G. W., 201, 264 Shaw, J., 164, 265 Shields-Mason, P. K., 97(96), 147 Shields, P. K., 97(94), 147 Shimizu, R., 84(54), 145 Shimoda, K., 26(32), 55 Shinoda, G., 70(19), 84(54), 144, 145 Shipley, J . F., 209, 265 Shiraiwa, T., 125, 150 Short, M. A., 94, 111(140), 146, 150 Shtokman, V. B., 317, 318, 319, 323, 372, 374, 375, 376, 420 Silliman, B. D., 163, 265 Silverberg, E. C., 17(18,20), 18, 20(18), 21(24), 54 Silverman, S. M.,183, 265 Simpson, G. C., 192, 265 Sinkankas, J., 74(26), 144 Sippel, R. F., 130, 151 Sisson, R. E., 279, 295 Skolnick, M., 29(40), 55 Smith, D. E., 22(26,27), 23(26,28a), 54, 55 Smith, D. G. W., 82, 92(67), 101, 133, 134(198), 135(198), 137, 140, 141, 145, 146, 148, 152, 153, 154 Smith, D . S., 27(34), 55 Smith, J. P., 73(21), 144 Smith, J. V . , 73(22), 94(80,81), 98, 103, 144, 146, 147 Smith, P. J., 105, 149

Smith, R. C., 420 Smith, R. L., 381, 420 Smith, T. B., 378, 384, 420 Snetsinger, K. G., 112(142), 150 Snodgrass, F. E., 43(53a), 56 Sommer, H. C., 203, 265 Sommerville, W. L., Jr., 312, 420 Spadin, P. L., 22(28), 55 Sparrow, J. G., 287, 296 Sprague, G., 162, 163, 164, 181, 201, 262 Springer, G., 92(68), 98, 101, 125(152), 146, 148, 149, 150 Sproull, W . T., 76(31), 79(31), 144 Stergis, C. G., 206, 265 Stern, M. S., 315, 316, 323, 386, 387, 390, 392, 395, 397, 399, 419, 420 Stevens, R. E., 117(146),150 Stevenson, W., 163, 177, 265 Stolarik, J. D., 207, 266 Stone, P. H.. 354, 365, 420 Stermer, C., 167, 187, 190, 201, 209, 265 Stringer, W. J . , 179, 181, 265 Stromberg, I. M.,211, 263 Stumbles, H. E., 167, 192, 193, 265 Stumpfl, E. F., 110(137), 128, 150, 151 Sundborg, A., 323, 402, 420 Superior, W. J . , 299, 300, 324, 421 Sutton, 0. G., 340, 379, 420 Sverdrup, H. U., 200, 265, 370, 420, 421 Swanson, G. E., 198, 264 Sweatman, T. R., 75, 96, 97(28), 99(28), 103, 111, 116(28), 117, 124, 144 Swift, C. D., 87(66), 88(66), 146 Switzer, G. S., 137, 153 Sydenham, P. H., 39(50), 56 Sykes, R. B., 415, 419 Symons, G. J., 268, 271, 274, 296

T Tarmy, B. L., 395, 416 Taylor, C. M., 73(24), 108, 144, 149 Taylor, G. I., 354, 421 Taylor, J. M., 67(15), 144 Taylor, R. C., 407, 421 Tempest, W., 202, 266 Thiesen, R., 97(100,101), 98, 148 Thompson, K . E., 402, 41 7

43 1

AUTHOR INDEX

Thornthwaite, C. W., 299, 300, 324, 421 Thresh, H., 60(10), 61, 143 Tobias, I., 29(40), 55 Togel, K., 98(108), 148 Tomlinson, M. C., 101, 148 Tousey, R., 189, 265 Tousimis, A. J., 68(16), 108(126), 144, 149 Townes, C. H., 26(31), 55 Traill, R. J., 95(86,86), 147 Treub, L. F., 128, 151 Tripathy, S. C., 213, 259 Tromholt, S., 159, 172, 173, 213, 265 Tsuchiya, K., 310, 340, 341, 361, 362, 353, 354, 421 Turner, E. H . , 54

U Udovykin, G. P., 129(165), 151 Ungstrup, E., 208, 263 Urch, D. S., 82, 145

v Vali, V., 24(30), 37(30), 55 Van de Hulst, H. C., 272, 273, 296 Van de Kamp, P., 189, 265 Van der Schueren, A., 211, 265 Van Rensburg, W. C. J., 73(23), 144 Van Schmus, W. R., 134, 136, 152, 153 Vassamillet, L. F., 109(134), 149 Viebrock, H. J., 288, 289, 296 Vinogradov, A. P., 129, 151 Vollath, D., 97(101,106), 98(108), 148 Volz, F. E., 280, 282, 286, 296 Von Bethe, H., 86, 146 Von Gierke, H . E., 203, 265 von KBrmBn, T., 330, 331, 421 Vries, J. L., 67, 68, 69, 104, 111, 143 W W. de Groot van Embden, 264 Wallace, R. A., 29(40), 55 Walter, L. S., 136, 153 Wampler, E. J., 17(17), 54 Wapstra, A. H., 87, 146

Warnaars, F. W., 133, 152 Warren, B. A., 371, 421 Watanabe, T., 202, 263 Watson, J., 196, 265 Webb, E. K., 367, 417 Webber, H. J., 166, 200, 266 Weber, J. N., 129, 151 Weber, R. E., 87(60), 146 Weiblen, P. W., 73(26), 137, 144, 153 Weigle, F . G., 324, 419 Weill, D., 134(193), 152 Wells, J. S., 2(3,4), 2(27), 53 Wenk, 0. E., 126, 155 Wescott, E. M., 207, 266 Westgate, J. A., 137, 153 Wever, E. G., 216, 266 White, A. D., 29(39), 55 Whittry, D. B., 119(150), 150 White, E. W., 80, 81, 130, 139, 145, 151, 154 White, W. B., 130, 151 Wickins, P. E., 310, 340, 346, 418 Widdowson, J . R., 140, 153 Wieblen, P. W., 136, 153 Wilkins, E. M., 300, 310, 346, 348, 417, 421 Wilkinson, D. T., 17(17,18,19), 20(18), 21(24), 54 Wilkinson, F. C. F., 109(135), 149 Wille, R., 338, 341, 346, 364, 416 Willgallis, A., 135, 153 Wilson, C. R., 202, 207, 266 Wilson, D. C., 27(36), 55 Wimbush, M., 43(66), 56 Wisner, W. M., 412, 421 Wolf, E., 4, 60(7), 54 Wolf, R. C., 99(112), 148 Woodhouse, J. B., 97(103), 148 Woollum, C. A., 402, 421 Wooster, W. S., 377, 421 Wordsworth, W., 266 Wormell, T. W., 207, 264, 266 Wyatt, F . K., 32(49), 39(49), 56 Wyrtki, K., 368, 366, 373, 407, 421 Wys, E. L., 128, 151

Y Yakowitz, H., 103, 111, 119, 120, 121, 122, 123, 149, 150

432

AUTHOR INDEX

Yang, H., 138,153 Yeowert, N. S., 202, 266 Yonker, T. L., 407, 421 Yoshida, K., 379, 380, 407, 421 Young, J. M., 202, 260, 263

Z Ziihringer, J., 136, 153

Zanotti, E., 193, 266 Zarantonello, E. H., 330, 334, 337, 338, 416

Zeitz, L., 114(144), 150 Ziemany, P. D., 104(123), 149 Ziebold, T. O., 96(82-84), 147 Zimmerman, L. I., 310, 340, 341, 347, 360, 421

Zipser, E. J., 407, 408, 409, 419, 421 Zubov, N. N., 374, 421

SUBJECT INDEX A

meteorological experiment, 409-414

Air flow, over islands, 383-406 Anegada Island, rain induced by,

Bishop’s ring, from volcanic events, 269-270, 273

396-397

Breezes, from heated islands, 311-316 Brush discharges, auroral audibility and, 208-216

Atmosphere, flow problems of, from islands, 297-421 Argus Island, wind flow studies on, 324-326

Auroral audibility, 165-266 altitude and terrain factors in, 192 animals affected by, 196 auroral characteristics in, 187-191 brush discharges end, 208-216 charecteristics of, 176-198 diurnal variation in, 180-181 electric field8 and, 214-216 frequency of occurrence, 176-177 geographic extent of, 177-178 hypotheses of, 198-208 electric field premure effect,

C Canary Islands, vortex patterns in, 342-361

Cheju Island, vortex street of, 361-364 Climate, volcanic dust effects on, 29&296

Computer, in electron probe analysis, 100-102

Crystals, for spectrometers, 66-67 Cuba, air flow over, 384-386

204-207

psychological, 188-199 radio waves, 208 tinnitus, 199-200 individual observations of, 161-176 latitudinal dependency of, 178-180 lightning and meteor sounds’ relation to, 197-198 localization of effect, 191 low auroras and, 192-193 magnetic activity and, 186-187 from nonvisible aurora, 194 observational results of, 167-176,

D Dispersion stabilization, for laser strain meters, 28 E Earthquakes, laser studies of, 44-48 Earth station, for laser ranging device, 17-19

Earth tides, laser measurements of, 41-43

218-269

Electromagnetic radiation, auroral audibility and, 203 Electron beam scanning, in electron probe analysis, 71-72 Electron probe analysis, 67-164 applications of, 126-142 t o qualitative analysis, 126-130 to quantitative analysis, 130-139 using soft X-ray spectra, 139-142 basic design features of, 60-62 electron optic system of, 62-63 errom in, 103-126

odor associated with, 193 in poetry, 196-197 seasonal dependency of, 180 sunspots and, 181-186 surveys on, 169-101 weather factors in, 191 B Barbados, air flow over, 400-401 Barbadoa oceanographic and 433

434

SUBJECT INDEX

accuracy and instrumental effects, 110-111

affecting precision, 104-1 10 alpha factors, 123 instrument for, 60-76 light optics of, 63-64 quantitative, 76-102 computer applications, 100-102 correction procedures, 93-100 X-ray emission and absorption, 76-92

Infrasonic waves, auroral audibility and, 202-203 Islands, atmospheric and ocean flow problems from, 297-421 heated islands, 310-316, 383-406 historical aspects, 323 island groups, 326-328 microscale perturbations, 323-326 significance of, 320-323 urban heated type, 402 vortex streets, 338-356

sample preparation for, 73-76 errors in, 115-116 surface contamination in, 73

F Fabry-Perot interferometer, for laser strain meters, 37 Fabry-Perot optical resonator, for laser stabilization, 30-31 Fiji, air flow over, 385

K Katmai event, turbidity anomalies from, 274-279 Krakatoa event, 268-274 optical phenomena from, 268-269 Kwajalein Atoll, as mesometeorological network, 326-328

L G Gas detectors, for electron probe analysis, 67-68 Geodosy, laser use in, 1-56 Geophysics, laser use in, 1-56 Glasses, electron probe analysis of, 136-137

Grand Bahama Island, showering Clouds Of, 397-400 Gravity, laser measurements of, 48-51

H Hainan, air flow over, 385-386 Hawaiian islands oceanic circulations from, 316-317 vortices leeward of, 357-371, 377 He-Ne laser, techniques using, 3-5 Heterodyne interferometer, laser type, 50-61

I Inclusions, electron probe analysis of, 130-136

Lamb dip stabilization, for laser strain meters, 27-28 Lamellae, electron probe analysis of, 130-136

Laser heterodyne interferometer, applications of, 50-51 Laser interferometer, gmvity measurements using, 48-50 Laser ranging devices accuracy of, 10 for artificial satellites, 22-23 electro-optic light modulation for, 7-8 extraterrestrial type, 14-23 results from, 38-48 techniques using, 2-3 terrestrial type, 6-14 tracking systems for, 22 two-color techniques for, 11-14 Laser strain meters linear extensometers for, 23-25 results from, 38-48 stabilization of, 27 techniques using, 2-3, 23-38 wavelength in, 25-27 Lasers, use in geodosy and geophysics, 1-56

436

SUBJEOT INDEX

Lightning auroral audibility and, 197-198 Line Islanda experiment, 4 0 M 0 9

M Meteors, auroral audibility and, 197-198 Michelson interferometers, for laser strain meters, 32-37 Microscope, for electron probe analysis, 62-64 Minerals, electron probe analysis of, 130-136 Moon, laser ranging experiment on, 1P 2 3

N Nantucket Island, cloud rows over, 386-395 New Caledonia, air flow over, 385 Niue, air flow over, 385

Solid state detectors, in eleotron probe analysis. 68-71 Sounds, auroral, 166-266 Sunspots, auroral audibility and, 181-186

T Taylor column, formation of, 364-367 Tinnitus, auroral audibility and, 199-200 Tropical islands, meteorology of, 406-414 Turbidity anomalies from Agumg event, 279-289 climatic effects of, 290-296 from Katmai event, 274-279 from Krakatoa. event, 268-274 volcanic type, 267-296 Twilights, unusual, following volcanic events, 270-271, 273-274

U 0

Upwelling, from islands, 377-378 Ocean, flow problems of, from islands, 297-421 Odor, auroral audibility and, 193 Optical resonator, for laser stabilization, 30-32

P Poetry, auroral audibility in, 196-197 Pulse height analysis, in electron probe analysis, 68

V Volcanic turbidity anomalies, 267-296 Von Kdrmtin theory of vortex street phenomenon, 330-331 Vortex street phenomenon, 329-330 viscosity role in, 332-338

W Weather, auroral audibility and, 191

R Retroreflector, lunar, 16-17 Rocks, electron probe analysis of, 137-139 Ruby laser, techniques using, 6-6

X X-ray diffraction gratings, for electron probe analysis, 67 X-ray spectrometers, for electron probe analysis, 64-66

S

Satellites, artificial, laser ranging to, 2 2 Secular strain rates, laser measurements of, 38-41

Zeeman stabilization, for laser strain meters, 28-29

This Page Intentionally Left Blank