The High-latitude Ionosphere and its Effects on Radio Propagation

This page intentionally left blank

The high-latitude ionosphere and its effects on radio propagation The physical properties of the ionized layer in the Earth’s upper atmosphere enable us to use it to support an increasing range of communications applications. This book presents a modern treatment of the physics and phenomena of the high-latitude upper atmosphere and the morphology of radio propagation in the auroral and polar regions. Chapters cover the basics of radio propagation and the use of radio techniques in ionospheric studies, as well as descriptions of the behavior and physics of the ionosphere at high latitude. Many investigations of high-latitude radio propagation have previously been published only in conference proceedings and organizational reports. This book includes many examples of the behavior of quiet and disturbed high-latitude high-frequency propagation. Ample cross-referencing, chapter summaries, and reference lists make this book an invaluable aid for graduate students, ionospheric physicists, and radio engineers. Cambridge Atmospheric and Space Science Series Editors: J. T. Houghton, M. J. Rycroft, and A. J. Dessler This series of upper-level texts and research monographs covers the physics and chemistry of the various regions of the Earth’s atmosphere, from the troposphere and stratosphere, up through the ionosphere and magnetosphere, and out to the interplanetary medium.  . .  is Professor Emeritus at the University of Alaska, Fairbanks and is Senior Partner of RP Consultants in Klamath Falls, Oregon. His considerable research experience in high- and mid-latitude radio-wave propagation and ionospheric studies using radio techniques was gained at the Geophysical Institute and Electrical Engineering Department of the University of Alaska, the Institute for Telecommunication Science (Boulder, Colorado), the Bell Labs (Murray Hill, New Jersey) and as a consultant. He has published over 100 papers and one book: Radio Techniques for Probing the Terrestrial Ionosphere (1991). From 1995 through 2002 he was Editor-in-Chief of the journal Radio Science.  . .  is Senior Research Fellow in the Department of Communication Systems of the University of Lancaster, and Senior Visiting Fellow of the University of Central Lancashire. He was formerly Senior Lecturer in the Department of Environmental Sciences at the University of Lancaster. He studied at the (then) Jodrell Bank Experimental Station of the University of Manchester, and has worked at the Radio Research Station (Slough, England), and the Space Environment Laboratory (Boulder, Colorado). With over forty years of research experience, mainly on studies of the upper atmosphere and ionosphere by radio methods, he has published 98 papers and two books: The Upper Atmosphere and Solar–Terrestrial Relations (1979) and The Solar–Terrestrial Environment (Cambridge University Press, 1992).

Cambridge Atmospheric and Space Science Series

Editors Alexander J. Dessler John T. Houghton Michael J. Rycroft

Titles in print in this series M. H. Rees Physics and chemistry of the upper atmosphere

J. C. King and J. Turner Antarctic meteorology and climatology

R. Daley Atmosphere data analysis

J. F. Lemaire and K. I. Gringauz The Earth’s plasmasphere

J. R. Garratt The atmospheric boundary layer

D. Hastings and H. Garrett Spacecraft–environment interactions

J. K. Hargreaves The solar–terrestrial environment

T. E. Cravens Physics of solar system plasmas

S. Sazhin Whistler-mode waves in a hot plasma

J. Green Atmospheric dynamics

S. P. Gary Theory of space plasma microinstabilities

G. E. Thomas and K. Stamnes Radiative transfer in the atmosphere and ocean

M. Walt Introduction to geomagnetically trapped radiation

T. I. Gombosi Physics of space environment

T. I. Gombosi Gaskinetic theory

R. W. Schunk and A. F. Nagy Ionospheres: Physics, plasma physics, and chemistry

B. A. Kagan Ocean–atmosphere interaction and climate modelling

I. G. Enting Inverse problems in atmospheric constituent transport

I. N. James Introduction to circulating atmospheres

R. D. Hunsucker and J. K. Hargreaves The high-latitude ionosphere and its effects on radio propagation

The high-latitude ionosphere and its effects on radio propagation R. D. Hunsucker Geophysical Institute, University of Alaska, Fairbanks

J. K. Hargreaves University of Lancaster

   Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo Cambridge University Press The Edinburgh Building, Cambridge  , United Kingdom Published in the United States by Cambridge University Press, New York www.cambridge.org Information on this title: www.cambridge.org/9780521330831 © Cambridge University Press 2003 This book is in copyright. Subject to statutory exception and to the provision of relevant collective licensing agreements, no reproduction of any part may take place without the written permission of Cambridge University Press. First published in print format 2002 ISBN-13 978-0-511-06742-6 eBook (EBL) ISBN-10 0-511-06742-9 eBook (EBL) ISBN-13 978-0-521-33083-1 hardback ISBN-10 0-521-33083-1 hardback

Cambridge University Press has no responsibility for the persistence or accuracy of s for external or third-party internet websites referred to in this book, and does not guarantee that any content on such websites is, or will remain, accurate or appropriate.

Contents

From the Times of London xv Preface xvii

Chapter 1 Basic principles of the ionosphere 1 1.1

Introduction 1

1.1.1

The ionosphere and radio-wave propagation 1

1.1.2

Why the ionosphere is so different at high latitude 2

1.2

The vertical structure of the atmosphere 4

1.2.1

Nomenclature 4

1.2.2

Hydrostatic equilibrium in the atmosphere 5

1.2.3

The exosphere 7

1.2.4

The temperature profile of the neutral atmosphere 8

1.2.5

Composition 10

1.3

Physical aeronomy 13

1.3.1

Introduction 13

1.3.2

The Chapman production function 15

1.3.3

Principles of chemical recombination 18

1.3.4

Vertical transport 20

1.4

The main ionospheric layers 23

1.4.1

Introduction 23

1.4.2

The E and F1 regions 26

1.4.3

The D region 31

1.4.4

The F2 region and the protonosphere 37

1.4.5

Anomalies of the F2 region 39

1.4.6

The effects of the sunspot cycle 44

1.4.7

The F-region ionospheric storm 46

vii

Contents

viii

1.5

The electrical conductivity of the ionosphere 48

1.5.1

Introduction 48

1.5.2

Conductivity in the absence of a magnetic field 48

1.5.3

The effect of a magnetic field 48

1.5.4

The height variation of conductivity 50

1.5.5

Currents 50

1.6

Acoustic-gravity waves and traveling ionospheric disturbances 52

1.6.1

Introduction 52

1.6.2

Theory 53

1.6.3

Traveling ionospheric disturbances 57

1.6.4

The literature 57

1.7

References and bibliography 58

Chapter 2 Geophysical phenomena influencing the high-latitude ionosphere 61 2.1

Introduction 61

2.2

The magnetosphere 61

2.2.1

The geomagnetic field 61

2.2.2

The solar wind 63

2.2.3

The magnetopause 69

2.2.4

The magnetosheath and the shock 71

2.2.5

The polar cusps 72

2.2.6

The magnetotail 72

2.3

Particles in the magnetosphere 73

2.3.1

Principal particle populations 73

2.3.2

The plasmasphere 74

2.3.3

The plasma sheet 78

2.3.4

Trapped particles 78

2.3.5

The ring current 84

2.3.6

Birkeland currents 85

2.4

The dynamics of the magnetosphere 86

2.4.1

Circulation patterns 86

2.4.2

Field merging 90

2.4.3

Magnetospheric electric fields 91

2.4.4

The dynamics of the plasmasphere 92

2.5

Magnetic storms 93

2.5.1

Introduction 93

2.5.2

The classical magnetic storm and the Dst index 94

2.5.3

Magnetic bays at high latitude; the auroral electrojet 95

2.5.4

Magnetic indices 96

Contents

2.5.5

Great magnetic storms and a case history 100

2.5.6

Wave phenomena of the magnetosphere 103

2.6

Ionization by energetic particles 105

2.6.1

Electrons 105

2.6.2

Bremsstrahlung X-rays 106

2.6.3

Protons 107

2.7

References and bibliography 109

Chapter 3 Fundamentals of terrestrial radio propagation 113 3.1

Introduction 113

3.2

Electromagnetic radiation 113

3.2.1

Basics of line-of-sight propagation in vacuo 113

3.2.2

Principles of radar 116

3.2.3

The significance of the refractive index 118

3.2.4

Interactions between radio waves and matter 121

3.3

Propagation through the neutral atmosphere 122

3.3.1

The refractivity of the neutral atmosphere 122

3.3.2

Terrain effects 124

3.3.3

Noise and interference 127

3.4

Ionospheric propagation 140

3.4.1

Magnetoionic theory 140

3.4.2

Reflection of radio waves from an ionospheric layer 144

3.4.3

Relations between oblique and vertical incidence 149

3.4.4

Trans-ionospheric propagation 147

3.4.5

Principles of radio scintillation 152

3.4.6

Propagation involving reflection from a sharp boundary and full-wave solutions 159

3.4.7

Whistlers 167

3.5

Ionospheric scatter 169

3.5.1

Coherent scatter 169

3.5.2

Forward scatter 171

3.5.3

Incoherent scatter 171

3.6

HF-propagation-prediction programs 174

3.7

Summary 175

3.8

References and bibliography 176

Chapter 4 Radio techniques for probing the ionosphere 181 4.1

Introduction 181

4.2

Ground-based systems 181

ix

Contents

x

4.2.1

Ionosondes 181

4.2.2

Coherent oblique-incidence radio-sounding systems 187

4.2.3

Incoherent-scatter radars 203

4.2.4

D-region absorption measurements 203

4.2.5

Ionospheric modification by HF transmitters 210

4.3

Space-based systems 215

4.3.1

A history of Earth–satellite and radio-rocket probing 215

4.3.2

Basic principles of operation and current-deployment of radio-beacon experiments 215

4.3.3

Topside sounders 216

4.3.4

In situ techniques for satellites and rockets 217

4.3.5

Capabilities and limitations 217

4.4

Other techniques 217

4.4.1

HF spaced-receiver and Doppler systems 217

4.4.2

The HF Doppler technique 219

4.4.3

Ionospheric imaging 220

4.5

Summary 220

4.6

References and bibliography 221

Chapter 5 The high-latitude F region and the trough 227 5.1

Circulation of the high-latitude ionosphere 227

5.1.1

Introduction 227

5.1.2

Circulation patterns 228

5.2

The behavior of the F region at high latitude 234

5.2.1

The F region in the polar cap 234

5.2.2

The effect of the polar cusps 237

5.2.3

The polar wind 239

5.2.4

The F layer in and near the auroral oval 240

5.3

Irregularities of the F region at high latitude 242

5.3.1

Introduction 242

5.3.2

Enhancements: patches, and blobs 244

5.3.3

Scintillation-producing irregularities 249

5.4

The main trough 260

5.4.1

Introduction 260

5.4.2

Observed properties and behavior of the main trough 261

5.4.3

The poleward edge of the trough 269

5.4.4

Motions of individual troughs 271

5.4.5

Mechanisms and models 273

5.5

Troughs and holes at high latitude 276

Contents

5.6

Summary 280

5.7

References and bibliography 281

Chapter 6 The aurora, the substorm, and the E region 285 6.1

Introduction 285

6.2

Occurrence zones 286

6.2.1

The auroral zone and the auroral oval 286

6.2.2

Models of the oval 288

6.3

The auroral phenomena 291

6.3.1

The luminous aurora 291

6.3.2

The distribution and intensity of the luminous aurora 291

6.3.3

Auroral spectroscopy 302

6.3.4

Ionospheric effects 302

6.3.5

The outer precipitation zone 305

6.4

The substorm 308

6.4.1

History 308

6.4.2

The substorm in the aurora 308

6.4.3

Ionospheric aspects of the substorm 311

6.4.4

Substorm currents 312

6.4.5

The substorm in the magnetosphere 315

6.4.6

The influence of the IMF and the question of substorm triggering 319

6.4.7

Relations between the storm and the substorm 321

6.5

The E region at high latitude 322

6.5.1

Introduction 322

6.5.2

The polar E layer 323

6.5.3

The auroral E layer under quiet conditions 323

6.5.4

The disturbed auroral E layer 323

6.5.5

Auroral radar 326

6.5.6

Auroral infrasonic waves 330

6.5.7

The generation of acoustic gravity waves 331

6.6

Summary and implications 332

6.7

References and bibliography 333

Chapter 7 The high-latitude D region 337 7.1

Introduction 337

7.2

Auroral radio absorption 339

7.2.1

Introduction – history and technique 339

7.2.2

Typical auroral-absorption events and their temporal and spatial properties 340

7.2.3

General statistics in space and time 350

xi

Contents

xii

7.2.4

Dynamics 354

7.2.5

The relation to geophysical activity, and predictions of auroral absorption 365

7.2.6

The wider geophysical significance of auroral-absorption events 371

7.3

The polar-cap event 382

7.3.1

Introduction 382

7.3.2

Observed properties of PCA events 384

7.3.3

The relation to solar flares and radio emissions 389

7.3.4

Effects arising during the proton’s journey to Earth 390

7.3.5

Non-uniformity and the midday recovery 395

7.3.6

Effects in the terrestrial atmosphere 398

7.4

Coherent scatter and the summer mesospheric echo 406

7.5

Summary and implications 409

7.6

References and bibliography 411

Chapter 8 High-latitude radio propagation: part 1 – fundamentals and early results 417 8.1

Introduction 417

8.2

ELF and VLF propagation 419

8.3

LF and MF propagation 429

8.4

HF propagation 439

8.4.1

Tests carried out between Alaska and Scandinavia on fixed frequencies 439

8.4.2

Tests involving transmission between Alaska and the continental USA 448

8.4.3

Other trans-polar HF experiments on fixed frequencies 450

8.4.4

College–Kiruna absorption studies at fixed frequencies 457

8.4.5

Effects of auroral-zone-absorption events on HF propagation 473

8.4.6

Sweep-frequency experiments 473

8.4.7

Other results from HF high-latitude studies from c. 1956–1969 479

8.4.8

Doppler and fading effects on HF high-latitude propagation paths 492

8.5

VHF/UHF and microwave propagation 529

8.6

Summary 531

8.7

References and bibliography 532

Chapter 9 High-latitude radio propagation: part 2 – modeling, prediction, and mitigation of problem 537 9.1

Introduction 537

9.2

Ionospheric ray-tracing, modeling, and prediction of propagation 538

9.2.1

Ionospheric ray-tracing 538

9.2.2

Realistic high-latitude models 538

9.2.3

Validation of ionospheric models 545

Contents

9.2.4

The performance of ELP–HF predictions at high latitudes 546

9.2.5

Recent validation of selected ionospheric prediction models using HF propagation data 553

9.3

Predictions of VHF/UHF propagation 568

9.4

Recent efforts at validation of ionospheric models 568

9.5

Mitigation of disturbance of HF propagation 572

9.5.1

Early attempts 572

9.5.2

Mitigation using solar–terrestrial data 572

9.5.3

Adaptive HF techniques 574

9.5.4

Realtime channel evaluation 580

9.5.5

Recent advances in assessment of HF high-latitude propagation channels 586

9.6

Other high-latitude propagation phenomena and evaluations 591

9.6.1

Large bearing errors on HF high-latitude paths 591

9.6.2

Effects of substorm on auroral and subauroral paths 593

9.6.3

Use of GPS/TEC data to investigate HF auroral propagation 594

9.6.4

The performance of HF modems at high latitude using multiple frequencies 597

9.7

Summary and discussion 597

9.8

References and bibliography 607 Appendix: some books for general reading 612 Index 613

xiii

From the Times of London TRANS-ATLANTIC MESSAGE Monday, December 16, 1901 — From our correspondent, St. Johns, NF, Dec. 14; Signor Marconi authorizes me to announce that he received on Wednesday and Thursday electrical signals at his experimental station here from the station at Poldhu, Cornwall, thus solving the problem of telegraphing across the Atlantic without wire. He has informed the Governor, Sir Cavendish Boyle, requesting him to apprise the British Cabinet of the discovery, the importance of which it is impossible to overvalue.

To Phyllis and Sylvia. For forbearance.

Preface

It is over a century since Marconi’s famous radio transmission across the Atlantic Ocean, an experiment closely followed by Kennelly and Heaviside’s suggestions that an ionized layer in the Earth’s upper atmosphere had made it possible. From the first, the ionosphere has been put to use, supporting an increasing range of applications from point-to-point communication and broadcasting, to directionfinding, navigation, and over-the-horizon radar. After 75 years of active research, the ionosphere can hardly be considered one of the mysteries of the Universe, but in fact some scientific problems and technical difficulties do remain. Many of them concern the high-latitude regions, which are particularly subject to disturbances arising initially on the sun. Since radio propagation depends so strongly on the behavior of the ionosphere, we have tried to bring the two topics together into a single monograph about the polar regions. The early chapters (1–4) provide introductions to the ionosphere in general, to the influence of the magnetosphere, to the principles of radio propagation, and to the major techniques of ionospheric observation. Chapters 5–7 describe the various phenomena of the ionosphere that are peculiar to the high latitudes. The final chapters (8–9) present the results of high-latitude propagation experiments, many of which have been published only in reports that were not widely disseminated at the time or have indeed remained unpublished. Short summaries are included at the end of each chapter to aid readers in getting a quick overview of the material in the chapter. Some useful Internet references (URLs) are given within the text. This book will fill a gap for scientists, engineers and students both at the graduate and at the undergraduate level whose interest is in understanding and/or predicting the behavior of radio propagation at auroral and polar latitudes.

xvii

xviii

Preface

Advanced amateur radio operators and shortwave listeners should also find useful information in this monograph. The book contains interlinking references between chapters, which, it is hoped, will aid the reader when a deeper understanding of the phenomena is desired. Now a word or two about references: The book includes material ranging from the classical to the recently published. References to the newer material are given at the end of each chapter, there divided by section. They are there partly as the usual courtesy to the original authors, but also so that the more inquisitive reader, such as yourself, may follow up topics in more detail by going back to the original sources. These, moreover, will often cite further valuable references. It would be impractical to cite all the original authors of material that has become standard in the field through being re-digested and re-presented in numerous books and review papers. To support material of this kind (mainly in Chapters 1, 2 and 6), a selection of books and conference reports is listed at the end of the chapter, and readers will be able to use these to broaden their knowledge of the field in general and also to check our own presentation of it if they feel so inclined. (Needless to say, the present authors will appreciate being told of any errors discovered.) An appendix lists some books that discuss more broadly the highlatitude phenomena connected with disturbances of the magnetosphere. We thank the many authors and publishers who have granted permission to reproduce diagrams, including some previously unpublished ones. We are grateful in particular to M. Angling, D. H. Bliss, N. J. Flowers, N. Gerson, J. M. Goodman, M. S. Gussenhoven, C. H. Jackman, M. J. Jarvis, V. Jodalen, E. Johnson, L. Kersley, R. L. McPherron, T. I. Pulkinnen, M. H. Rees, J. Secan, P. N. Smith, E. Turunen, M. Walt, J. W. Wright, and M. Wild.

The high-latitude zones within the solar–terrestrial environment. After Synoptic Data for Solar–Terrestrial Physics, The Royal Society (September 1992). Wildlife by J. C. Hargreaves.

Chapter 1 Basic principles of the ionosphere

1.1

Introduction

1.1.1

The ionosphere and radio-wave propagation

The ionosphere is the ionized component of the atmosphere, comprising free electrons and positive ions, generally in equal numbers, in a medium that is electrically neutral. Though the charged particles are only a minority amongst the neutral ones, they nevertheless exert a great influence on the electrical properties of the medium, and it is their presence that brings about the possibility of radio communication over large distances by making use of one or more ionospheric reflections. The early history of the ionosphere is very much bound up with the development of communications. The first suggestions that there are electrified layers within the upper atmosphere go back to the nineteenth century, but the modern developments really started with Marconi’s well-known experiments in transAtlantic communication (from Cornwall to Newfoundland) in 1901. These led to the suggestions by Kennelly and by Heaviside (made independently) that, because of the Earth’s curvature, the waves could not have traveled directly across the Atlantic but must have been reflected from an ionized layer. The name ionosphere came into use about 1932, having been coined by Watson-Watt several years previously. Subsequent research has revealed a great deal of information about the ionosphere: its vertical structure, its temporal and spatial variations, and the physical processes by which it is formed and which influence its behavior. Looked at most simply, the ionosphere acts as a mirror situated between 100 and 400 km above the Earth’s surface, as in Figure 1.1, which allows reflected

1

Basic principles of the ionosphere

2

Ionosphere

Ground

30

0

km

Figure 1.1. Long distance propagation by multiple hops between the ionosphere and the ground.

signals to reach points around the bulge of the Earth. The details of how reflection occurs depend on the radio frequency of the signal, but the most usual mechanism, which applies in the high-frequency (HF) band (3–30 MHz), is actually a gradual bending of the ray towards the horizontal as the refractive index of the ionospheric medium decreases with altitude. Under good conditions, signals can be propagated in this way for several thousand kilometers by means of repeated reflections between ionosphere and ground. Reflection from a higher level (the F region) obviously gives a greater range per “hop” than does one from a lower level (the E region), but which mode is possible depends on the structure of the ionosphere at the time. Higher radio frequencies tend to be reflected from greater heights, but if the frequency is too high there may be insufficient bending and the signal then penetrates the layer and is lost to space. This is the first complication of radio propagation. The second complication is that the lower layers of the ionosphere tend to absorb the signal. This effect is greater for signals of lower frequency and greater obliquity. Hence, practical radio communication generally requires a compromise. The ionosphere is constantly changing, and the art of propagation prediction is to determine the best radio frequency for a given path for the current state of the ionosphere. Plainly, an understanding of ionospheric mechanisms is basic to efficient radio communication. Further details about radio propagation are given in Chapter 3, and our central topic of how propagation at high latitudes is affected by the vagaries of the highlatitude ionosphere is discussed later in the book. 1.1.2

Why the ionosphere is so different at high latitude

The terrestrial ionosphere may be divided broadly into three regions that have rather different properties according to their geomagnetic latitude. The midlatitude region has been explored the most completely and is the best understood. There, the ionization is produced almost entirely by energetic ultra-violet and Xray emissions from the Sun, and is removed again by chemical recombination processes that may involve the neutral atmosphere as well as the ionized species. The

1.1 Introduction

movement of ions, and the balance between production and loss, are affected by winds in the neutral air. The processes typical of the mid-latitude ionosphere also operate at high and low latitudes, but in those regions additional processes are also important. The low-latitude zone, spanning 20° or 30° either side of the magnetic equator, is strongly influenced by electromagnetic forces that arise because the geomagnetic field runs horizontally over the magnetic equator. The primary consequence is that the electrical conductivity is abnormally large over the equator. A strong electric current (an “electrojet”) flows in the E region, and the F region is subject to electrodynamic lifting and a “fountain effect” that distorts the general form of the ionosphere throughout the low-latitude zone. At high latitudes we find the opposite situation. Here the geomagnetic field runs nearly vertical, and this simple fact of nature leads to the existence of an ionosphere that is considerably more complex than that in either the middle or the low-latitude zones. This happens because the magnetic field-lines connect the high latitudes to the outer part of the magnetosphere which is driven by the solar wind, whereas the ionosphere at middle latitude is connected to the inner magnetosphere, which essentially rotates with the Earth and so is less sensitive to external influence. We can immediately identify four general consequences. (a). The high-latitude ionosphere is dynamic. It circulates in a pattern mainly controlled by the solar wind but which is also variable. (b). The region is generally more accessible to energetic particle emissions from the Sun that produce additional ionization. Thus it is affected by sporadic events, which can seriously degrade polar radio propagation. Over a limited range of latitudes the dayside ionosphere is directly accessible to material from the solar wind. (c). The auroral zones occur within the high-latitude region. Again, their location depends on the linkage with the magnetosphere, in this case into the distorted tail of the magnetosphere. The auroral phenomena include electrojets, which cause magnetic perturbations, and there are “substorms” in which the rate of ionization is greatly increased by the arrival of energetic electrons. The auroral regions are particularly complex for radio propagation. (d). A “trough” of lesser ionization may be formed between the auroral and the mid-latitude ionospheres. Although the mechanisms leading to the formation of the trough are not completely known, it is clear that one fundamental cause is the difference in circulation pattern between the inner and outer parts of the magnetosphere. This monograph is concerned mainly with the ionosphere at high latitudes, but before considering the special behavior which occurs in those regions we must review some processes affecting the ionosphere in general and summarize the more normal behavior at middle latitudes. In order to do that, we must first

3

Basic principles of the ionosphere

3000

Thermosphere

100 30 10 3 1

Exobase or Heliosphere Turbopause

300 Mesopause Mesosphere Stratopause Stratosphere Tropopause Troposphere 500

1000

Turbosphere or homosphere

Height (km)

1000

Protonosphere

Magnetosphere

10 000

Gaseous escape Ionization Exosphere

Composition

baropause Ionosphere Barosphere

Temperature

Heterosphere

4

1500

Temperature (K)

10

5

0

Electron density (10 5 cm –3 )

Figure 1.2. Nomenclature of the upper atmosphere based on temperature, composition, mixing, and ionization. (J. K. Hargreaves, The Solar–Terrestrial Environment. Cambridge University Press, 1992.)

consider the nature of the neutral upper atmosphere in which the ionosphere is formed.

1.2

The vertical structure of the atmosphere

1.2.1

Nomenclature

A static planetary atmosphere may be described by four properties: pressure (P), density (
), temperature (T ), and composition. Since these are not independent it is not necessary to specify all of them. The nomenclature of the atmosphere is based principally on the variation of temperature with height, as in Figure 1.2. Here, the different regions are called “spheres” and the boundaries between them are “pauses”. The lowest region is the troposphere, in which the temperature falls off with increasing height at a rate of 10 K km1 or less. Its upper boundary is the tropopause at a height of 10–12 km. The stratosphere which lies above it was once thought to be isothermal, but it is actually a region where the temperature increases with height. At about 50 km is a maximum due to the absorption of solar ultra-violet radiation in ozone; this is the stratopause. Above that the temperature again decreases in the mesosphere (or middle atmosphere) and passes through another minimum at the mesopause at 80–85 km. At about 180 K, this is the coldest part of the whole atmosphere. Above the mesopause, heating by solar ultra-violet radiation ensures that the temperature gradient remains positive, and this is the thermosphere. Eventually the temperature of the thermosphere becomes

1.2 Vertical structure

5

almost constant at a value that varies with time but is generally over 1000 K; this is the hottest part of the atmosphere. Though the classification by temperature is generally the most useful, others based on the state of mixing, the composition or the state of ionization are also useful. The lowest part of the atmosphere is well mixed, with a composition much like that at sea level except for minor components. This is the turbosphere or homosphere. In the upper region, essentially the thermosphere, mixing is inhibited by the positive temperature gradient, and here, in the heterosphere, the various components separate under gravity and as a result the composition varies with altitude. The boundary between the two regions, which occurs at about 100 km, is the turbopause. Above the turbopause the gases separate by gaseous diffusion more rapidly than they are mixed by turbulence. Within the heterosphere there are regions where helium or hydrogen may be the main component. These are the heliosphere and the protonosphere, respectively. From the higher levels, above about 600 km, individual atoms can escape from the Earth’s gravitational attraction; this region is called the exosphere. The base of the exosphere is the exobase or the baropause, and the region below the baropause is the barosphere. The terms ionosphere and magnetosphere apply, respectively, to the ionized regions of the atmosphere and to the outermost region where the geomagnetic field controls the particle motions. The outer termination of the geomagnetic field (at about ten Earth radii in the sunward direction) is the magnetopause.

1.2.2

Hydrostatic equilibrium in the atmosphere

Between them the properties temperature, pressure, density, and composition determine much of the atmosphere’s behaviour. They are not independent, being related by the universal gas law which may be written in various forms, but for our purposes the form PnkT,

(1.1)

where n is the number of molecules per unit volume, is the most useful. The quantity n is properly called the concentration or the number density, but “density” alone is often used when the sense is clear. Apart from its composition, the most significant feature of the atmosphere is that the pressure and density decrease with increasing altitude. This height variation is described by the hydrostatic equation, sometimes called the barometric equation, which is easily derived from first principles. The variation of pressure with height is PP0 exp(h/H ),

(1.2)

6

Basic principles of the ionosphere

where P is the pressure at height h, P0 is the pressure where h0, and H is the scale height given by HkT/(mg),

(1.3)

in which k is Boltzmann’s constant, T is the absolute temperature, m is the mass of a single molecule of the atmospheric gas, and g is the acceleration due to gravity. If T and m are constant (and any variation of g with height is neglected), H is the vertical distance over which n falls by a factor e (2.718), and thus it serves to define the thickness of an atmosphere. H is greater, and the atmosphere thicker, if the gas is hotter or lighter. In the Earth’s atmosphere H varies from about 5 km at height 80 km to 70–80 km at 500 km. Using equation (1.1), the hydrostatic equation may be written in differential form as dP/Pdn/ndT/Tdh/H.

(1.4)

From this, H can be ascribed a local value, even if it varies with height. Another useful form is P/P0 exp[(hh0)/H ]ez,

(1.5)

where PP0 at the height hh0, and z is the reduced height defined by z(hh0)/H.

(1.6)

The hydrostatic equation can also be written in terms of the density (
) and the number density (n). If T, g, and m are constant over at least one scale height, the equation is essentially the same in terms of P,
, and n, since n/n0 
/
0 P/P0. The ratio k/m can also be replaced by R/M, where R is the gas constant and M is the relative molecular mass. Whatever the height distribution of the atmospheric gas, its pressure P0 at height h0 is just the weight of gas above h0 in a column of unit cross-section. Hence P0 NT mgn0kT0,

(1.7)

where NT is the total number of molecules in the column above h0, and n0 and T0 are the concentration and the temperature at h0. Therefore we can write NT n0kT0/(mg)n0H0,

(1.8)

H0 being the scale height at h0. This equation says that, if all the atmosphere above h0 were compressed to density n0 (that already applying at h0), then it would

1.2 Vertical structure

7

occupy a column extending just one scale height. Note also that the total mass of the atmosphere above unit area of the Earth’s surface is equal to the surface pressure divided by g. Although we often assume that g, the acceleration due to gravity, is a constant, in fact it varies with altitude as g(h)  1/(RE h)2, where RE is the radius of the Earth. The effect of changing gravity may be taken into account by defining a geopotential height h*REh/(RE h).

(1.9)

A molecule at height h over the spherical Earth has the same potential energy as one at height h* over a hypothetical flat Earth having gravitational acceleration g(0). Within the homosphere, where the atmosphere is well mixed, the mean relative molecular mass determines the scale height and the variation of pressure with height. In the heterosphere, the partial pressure of each constituent is determined by the relative molecular mass of that species. Each species takes up its own distribution, and the total pressure of the atmosphere is the sum of the partial pressures in accordance with Dalton’s law. 1.2.3

The exosphere

In discussing the atmosphere in terms of the hydrostatic equation we are treating the atmosphere as a compressible fluid whose temperature, pressure, and density are related by the gas law. This is valid only if there are sufficient collisions between the gas molecules for a Maxwellian velocity distribution to be established. As the pressure decreases with increasing height so does the collision frequency, and at about 600 km the distance traveled by a typical molecule between collisions, the mean free path, becomes equal to the scale height. At this level and above we have to regard the atmosphere in a different way, not as a fluid but as an assembly of individual molecules or atoms, each following its own trajectory in the Earth’s gravitational field. This region is called the exosphere. While the hydrostatic equation is strictly valid only in the barosphere, it has been shown that the same form may still be used if the velocity distribution is Maxwellian. This is true to some degree in the exosphere, and the use of the hydrostatic equation is commonly extended to 1500–2000 km, at least as an approximation. However, this liberty may not be taken if there is significant loss of gas from the atmosphere, since more of the faster molecules will be lost and the velocity distribution of those remaining will be altered thereby. The lighter gases, helium and hydrogen, are affected most. The rate at which gas molecules escape from the gravitational field in the exosphere depends on their vertical speed. Equating the kinetic and potential energies of an upward-moving particle, its escape velocity (ve ) is given by

Basic principles of the ionosphere

8

v 2e 2gr,

(1.10)

where r is the distance of the particle from the center of the Earth. (At the Earth’s surface the escape velocity is 11.2 km s1, irrespective of the mass of the particle.) By kinetic theory the root mean square (r.m.s.) thermal speed of gas molecules (v2) depends on their mass and temperature, and, for speeds in one direction, i.e. vertical, mv2/23kT/2.

(1.11)

Thus, corresponding to an escape velocity (ve) there can be defined an escape temperature (Te). Te is 84000 K for atomic oxygen, 21000 K for helium, but only 5200 K for atomic hydrogen. At 1000–2000 K, exospheric temperatures are smaller than these escape temperatures, and loss of gas, if any, will be mainly at the high-speed end of the velocity distribution. In fact, the loss is insignificant for O, slight for He, but significant for H. Detailed computations show that the resulting vertical distribution of H departs significantly from the hydrostatic at distances more than one Earth radius above the surface, but for He the departure is small. 1.2.4

The temperature profile of the neutral atmosphere

The atmosphere’s temperature profile results from the balance amongst sources of heat, loss processes, and transport mechanisms. The total picture is complicated, but the main points are as follows.

Sources The troposphere is heated by convection from the hot ground, but in the upper atmosphere there are four sources of heat: (a). Absorption of solar ultra-violet and X-ray radiation, causing photodissociation, ionization, and consequent reactions that liberate heat; (b). Energetic charged particles entering the upper atmosphere from the magnetosphere; (c). Joule heating by ionospheric electric currents; and (d). Dissipation of tidal motions and gravity waves by turbulence and molecular viscosity. Generally speaking, the first source (a) is the most important, though (b) and (c) are also important at high latitude. Most solar radiation of wavelength less than 180 nm is absorbed by N2, O2 and O. Photons that dissociate or ionize molecules or atoms generally have more energy than that needed for the reaction, and the excess appears as kinetic energy of the reaction products. A newly created photoelectron, for example, may have between 1 and 100 eV of kinetic energy, which

1.2 Vertical structure

subsequently becomes distributed throughout the medium by interactions between the particles (optical, electronic, vibrational, or rotational excitation, or elastic collisions, depending on the energy.) Elastic collisions redistribute energy less than 2 eV, and, since this process operates mainly between electrons, these remain hotter than the ions. Some energy is reradiated, but on average about half goes into local heating. It can generally be assumed that in the ionosphere the rate of heating in a given region is proportional to the ionization rate. The temperature profile (Figure 1.2) can be explained as follows. The maximum at the stratopause is due to the absorption of 200–300 nm (2000–3000 Å) radiation by ozone (O3) over the height range 20–50 km. Some 18 W m2 is absorbed in the ozone layer. Molecular oxygen (O2), which is relatively abundant up to 95 km, absorbs radiation between 102.7 and 175 nm, much of this energy being used to dissociate O2 to atomic oxygen (O). This contribution amounts to some 30 mW m2. Radiation of wavelengths shorter than 102.7 nm, which is the ionization limit for O2 (See Table 1.1 of Section 1.4.1), is absorbed to ionize the major atmospheric gases O2, O, and N2 over the approximate height range 95–250 km, and this is what heats the thermosphere. Though the amount absorbed is only about 3 mW m2 at solar minimum (more at solar maximum), a small amount of heat may raise the temperature considerably at great height because the air density is small. Indeed, at the greater altitudes the heating rate and the specific heat are both proportional to the gas concentration, and then the rate of increase in temperature is actually independent of height. At high latitude, heating associated with the aurora – items (b) and (c) – is important during storms. Joule heating by electric currents is greatest at 115–130 km. Auroral electrons heat the atmosphere mainly between 100 and 130 km.

Losses The principal mechanism of heat loss from the upper atmosphere is radiation, particularly in the infra-red. Emission by oxygen at 63 m is important, as are spectral bands of the radical OH and the visible airglow from oxygen and nitrogen. The mesosphere is cooled by radiation from CO2 at 15 m and from ozone at 9.6 m, though during the long days of the polar summer the net effect can be heating instead of cooling.

Transport The thermal balance and temperature profile of the upper atmosphere are also affected by processes of heat transport. At various levels conduction, convection, and radiation all come into play. Radiation is the most efficient process at the lowest levels, and the atmosphere is in radiative equilibrium between 30 and 90 km. Eddy diffusion, or convection, also operates below the turbopause (at about 100 km), and allows heat to be carried down into the mesosphere from the thermosphere. This flow represents a major loss of heat from the thermosphere but is a minor source for the mesosphere.

9

Basic principles of the ionosphere

10

In the thermosphere (above 150 km) thermal conduction is efficient because of the low pressure and the presence of free electrons. The large thermal conductivity ensures that the thermosphere is isothermal above 300 or 400 km, though the thermospheric temperature varies greatly from time to time. Chemical transport of heat occurs when an ionized or dissociated species is created in one place and recombines in another. The mesosphere is heated in part by the recombination of atomic oxygen created at a higher level. There can also be horizontal heat transport by large-scale winds, which can affect the horizonal distribution of temperature in the thermosphere. The balance amongst these various processes produces an atmosphere with two hot regions, one at the stratopause and one in the thermosphere. The thermospheric temperature, in particular, undergoes strong variations daily and with the sunspot cycle, both due to the changing intensity of solar radiation. 1.2.5

Composition

The upper atmosphere is composed of various major and minor species. The former are the familiar oxygen and nitrogen in molecular or atomic forms, or helium and hydrogen at the greater heights. The minor constituents are other molecules that may be present as no more than mere traces, but in some cases they can exert an influence far beyond their numbers.

Major species The constant mixing within the turbosphere results in an almost constant proportion of major species up to 100 km, essentially the mixture as at ground-level called “air”, although complete uniformity cannot be maintained if there are sources and sinks for particular species. Molecular oxygen is dissociated to atomic oxygen by ultra-violet radiation between 102.7 and 175.9 nm: O2 h →OO,

(1.12)

where h is a quantum of radiation. An increasing amount of O appears above 90 km. The atomic and molecular forms are present in equal concentrations at about 125 km, and above that the atomic form increasingly dominates. Nitrogen is not directly dissociated to the atomic form in the atmosphere, though it does appear as a product of other reactions. Above the turbopause mixing is less important than diffusion, and then each component takes an individual scale height depending on its relative atomic or molecular mass (HkT/(mg)). Because the scale heights of the common gases vary over a wide range – H1, He4, O16, N2 28, O2 32 – the relative composition of the thermosphere is a marked function of height, the lighter gases becoming progressively more abundant as illustrated in Figure 1.3. Atomic oxygen dominates at a height of several hundred kilometers. Above that is the

1.2 Vertical structure

Figure 1.3. Atmospheric composition to 1000 km for a typical temperature profile. (US Standard Atmosphere, 1976.)

heliosphere, where helium is the most abundant, and eventually hydrogen becomes the major species in the protonosphere. Because the scale height also depends on the temperature, so do the details of the composition. The protonosphere starts much higher in a hot thermosphere, and the heliosphere may be absent from a cool one. Two of the important species of the upper atmosphere, helium and hydrogen, are no more than minor species in the troposphere. Helium comes from radioactive decay in the Earth’s crust. It diffuses up through the atmosphere, eventually escaping into space. The source of atomic hydrogen is the dissociation of water vapor near the turbopause from where it, also, flows constantly up through the atmosphere.

Minor species Water, carbon dioxide, oxides of nitrogen, ozone, and alkali metals are all minor species of the atmosphere, but not all of them are significant for the ionosphere. Water does not have the same dominating influence in the upper atmosphere as in the troposphere. It is important nevertheless, first as a source of hydrogen, and second because it causes ions to be hydrated below the mesopause. Carbon dioxide, also, plays a part in the chemistry of the D region.

11

12

Basic principles of the ionosphere

Nitric oxide (NO), on the other hand, makes an important contribution to the lower ionosphere since it is ionized by the intense Lyman- line of the solar spectrum and is thereby responsible for much of the ionospheric D region at middle latitudes (Section 1.4.3). The chemical story of NO is complicated because several production and loss mechanisms are at work and the distribution is affected by the dynamics of the mesosphere. Nitric oxide in the mesosphere comes from two sources. One source is in the stratosphere and involves the oxidation of nitrous oxide (N2O) by excited atomic oxygen. The second one peaks in the thermosphere, at 150–160 km, and involves a reaction with neutral or ionized atomic nitrogen, for example N*O2 →NOO,

(1.13)

where the * indicates an excited state. The resulting NO diffuses down to the mesosphere by molecular and then by eddy diffusion. Loss by photodissociation and recombination, aided by the effect of the low temperature at the mesopause, is sufficient to create a minimum at 85–90 km. The diffusion is weaker in the summer, and that is when the minimum is most marked. The depth of the minimum also varies with latitude. The production of these atomic-nitrogen species is closely linked to ionization processes, and it is estimated that 1.3 NO molecules are produced on average for each ion produced. The concentration of nitric oxide therefore varies with time of day, latitude, and season. It is 3–4 times greater at high latitude than it is at middle latitude, and more variable. The production rate increases dramatically during particle precipitation events, and this is plainly an important mechanism in the high-latitude ionosphere. The ozonosphere peaks between heights of 15 and 35 km, well below the ionosphere. The small amounts of ozone that occur in the mesosphere are involved in certain reacions in the D region, but we shall not be particularly concerned with them in this monograph. It is, however, of some general interest that there is a reaction between ozone and nitric oxide that tends to remove ozone at mesospheric levels. Thus, O3 NO→O2 NO2 ONO2 →O2 NO O3 O→2O2.

(1.14)

The net result, in the presence of atomic oxygen, is a catalytic conversion of ozone back to molecular oxygen. In this way the ozone concentration is affected by the natural production of nitric oxide discussed above. Metallic atoms are introduced into the atmosphere in meteors, whose flux over the whole Earth amounts to 44 metric tons per day. In the ionized state, metals

1.3 Physical aeronomy

Figure 1.4. Typical vertical profiles of electron density in the mid-latitude ionosphere: ——, sunspot maximum; and – – –, sunspot minimum. (After W. Swider, Wallchart Aerospace Environment, US Air Force Geophysics Laboratory.)

such as sodium, calcium, iron, and magnesium are significant to the aeronomy of the lower ionosphere in various ways, but they will not be of great concern to us at high latitudes.

1.3

Physical aeronomy

1.3.1

Introduction

The topic of physical aeronomy covers the physical considerations governing the formation and shape of an ionospheric layer. The detailed photochemical processes which are involved in a particular case are generally considered under chemical aeronomy; however, we shall include such chemical details as we require in Section 1.4 as part of our description of the actual terrestrial ionosphere. Typical vertical profiles of the ionosphere are shown in Figure 1.4. The identification of the regions was much influenced by their signatures on ionograms (see Section 4.2.1), which tend to emphasize inflections in the profile, and it is not necessarily the case that the various layers are separated by distinct minima. The main regions are designated D, E, F1, and F2, with the following daytime characteristics: ■

D region, 60–90 km: electron density 108–1010 m3 (102–104 cm3);

■

E region, 105–160 km: electron density of several times 1011 m3 (105 cm3);

13

Basic principles of the ionosphere

14

■

F1 region, 160–180 km: electron density of several times 1011 to about 1012 m3 (105–106 cm3);

■

F2 region, height of maximum variable around 300 km: electron density up to several times 1012 m3 (106 cm3).

All these ionospheric regions are highly variable, and in particular there is generally a large change between day and night. The D and F1 regions vanish at night, and the E region becomes much weaker. The F2 region, however, tends to persist, though at reduced intensity. The ionosphere is formed by the ionization of atmospheric gases such as N2, O2, and O. At middle and low latitude the required energy comes from solar radiation in the extreme ultra-violet (EUV) and X-ray parts of the spectrum. Once they have been formed, the ions and electrons tend to recombine and to react with other gaseous species to produce other ions. Thus there is a dynamic equilibrium in which the net concentration of free electrons (which, following standard practice, we call the electron density) depends on the relative speed of the production and loss processes. In general terms the rate of change of electron density is expressed by a continuity equation: N/ tqLdiv(Nv)

(1.15)

where q is the production rate (per unit volume), L is the rate of loss by recombination, and div(Nv) expresses the loss of electrons by movement, v being their mean drift velocity. If we consider a representative ionization and recombination reaction and neglect movements, Xh
 Xe.

(1.16)

The “law of mass action” tells us that, at equilibrium, [X][h]constant [X][e],

(1.17)

where the square brackets signify concentrations. Thus, since [e][X] for electrical neutrality, [e]2 constant [X][h]/[X]

(1.18)

During the day the intensity of ionizing radiation varies with the elevation of the Sun, and the electron density responds to the variation of [h]. At night the source of radiation is removed and so the electron density decays. From this simple model we can also see that the electron density must vary with altitude. The intensity of ionizing radiation increases with height but the concentration of ionizable gas [X]

1.3 Physical aeronomy

15

decreases. It is reasonable to expect from this that the electron density will pass through a maximum at some altitude. 1.3.2

The Chapman production function

In 1931, S. Chapman developed a formula that predicts the form of a simple ionospheric layer and how it varies during the day. Although it is only partly successful in explaining the observed behavior of the terrestrial ionosphere – and this because of phenomena that it does not include – Chapman’s formula is at the root of our modern understanding of the ionosphere and therefore it deserves a brief mention in this section. At this stage we deal only with the rate of production of ionization (q), and the formula expressing this is the Chapman production function. In the simple treatment, which is sufficient for our purposes, it is assumed that ■

the atmosphere is composed of a single species, exponentially distributed with constant scale height;

■

the atmosphere is plane stratified: there are no variations in the horizontal plane;

■

radiation is absorbed in proportion to the concentration of gas particles; and

■

the absorption coefficient is constant: this is equivalent to assuming that we have monochromatic radiation.

The rate of production of ion–electron pairs at some level of the atmosphere can be expressed as the product of four terms: q  nI.

(1.19)

Here, I is the intensity of ionizing radiation and n is the concentration of atoms or molecules capable of being ionized by that radiation. For an atom or molecule to be ionized it must first absorb radiation, and the amount absorbed is expressed by the absorption crossection, : if the flux of incident radiation is I (J m2 s1) then the total energy absorbed per unit volume of the atmosphere per unit time is nI. However, not all this energy will go into the ionization process, and the ionization efficiency,, takes that into account, being the fraction of the absorbed radiation that goes into producing ionization. The Chapman production function is usually written in a normalized form as qqm0 exp(1 zsec ez).

(1.20)

Here, z is the reduced height for the neutral gas, z(h  hm0)/H, H being the scale height. is the solar zenith angle, hm0 is the height of the maximum rate of production when the Sun is overhead (i.e. hm when 0), and qm0 is the production

16

Basic principles of the ionosphere

Figure 1.5. The Chapman production function. (After T. E. VanZandt and R. W. Knecht, in Space Physics (eds. LeGalley and Rosen). Wiley, 1964.)

rate at this altitude, also when the Sun is overhead. Derivations of equation (1.20) are given in many of the standard textbooks (see the list of further reading). Equation (1.20) can also be written q/qm0 eeze[sec .exp(z)],

(1.21)

where the first term is a constant, the second expresses the height variation of the density of ionizable atoms, and the third is proportional to the intensity of the ionizing radiation. Figure 1.5 illustrates some general properties of the production-rate profile. At a great height, where z is large and positive, q →qm0eez.

(1.22)

Thus the curves merge above the peak, becoming independent of and exhibiting an exponential decrease with height due to the decreasing density of the

1.3 Physical aeronomy

17

neutral atmosphere. In the region well below the peak, when z is large and negative, the shape becomes dominated by the last term of Equation (1.21), producing a rapid cut-off. Thus, as predicted in the previous section, the production rate is limited by a shortage of ionizable gas at the greater altitudes and by a lack of ionizing radiation low down. On a plot of ln(q) against z all the curves are the same shape, but they are displaced upwards and to the left as the zenith angle, , increases. The intensity of radiation in an absorbing atmosphere may be written as IIinf e

(1.23)

where  is the optical depth, which is equal to the absorption coefficient times the number of absorbing atoms down to the level considered:   NT;

(1.24)

and Iinf is the intensity at great height. This leads to an important theorem: The production rate is greatest at the level where the optical depth is unity. From this general result there follow some particularly useful rules. (1). The maximum production rate at a given value of is given by qm  Iinf /(eHsec ).

(1.25)

(2). The reduced height of the maximum depends on the solar zenith angle as zm ln(sec ).

(1.26)

(3). The rate of production at this maximum is qm qm0 cos .

(1.27)

These simple results are important in studies of the ionosphere because the maximum of a layer is the part most readily observed. From Equations (1.26) and (1.27) we see that a plot of ln(qm) against zm is effectively a plot of ln(cos ) against ln(sec ), which obviously gives a straight line of slope 1. This line is shown in Figure 1.5. The Chapman production function is important because it expresses fundamentals of ionospheric formation and of the absorption of radiation in any exponential atmosphere. Although real ionospheres may be more complicated, the Chapman theory provides an invaluable reference point for interpreting observations and a relatively simple starting point for ionospheric theory.

Basic principles of the ionosphere

18

1.3.3

Principles of chemical recombination

Working out the rate of electron production is just the first step in calculating the electron density in an ionized layer, and the next step is to reckon the rates at which electrons are removed from the volume under consideration. This is represented in the continuity equation (1.15) by two further terms, one for the recombination of ions and electrons to reform neutral particles, and the other to account for movement of plasma into or out of the volume. We deal first with the principles of chemical recombination. The question of which individual reactions are most important in different parts of the ionosphere will be addressed in Section 1.4. First we assume that the electrons recombine directly with positive ions and that no negative ions are present: X e→X. Then the rate of electron loss is L [X]Ne  N e2

(1.28)

where Ne is the electron density (equal to the ion density [ X]) and  is the recombination coefficient. At equilibrium, therefore, q N e2.

(1.29)

The equilibrium electron density is proportional to the square root of the production rate, which may be replaced by the Chapman production function (1.20) to get the variation of electron density with height and solar zenith angle. In particular, it is seen that the electron density at the peak of the layer varies as cos1/2 : Nm Nm0 cos1/2 .

(1.30)

A layer with these properties is called an -Chapman layer. If one is concerned particularly with electron loss, then attachment to neutral particles to form negative ions can itself be regarded as another type of electronloss process. In fact, as we shall see, this becomes the dominant type at somewhat higher levels of the ionosphere (though by a different process). Without at this stage specifying chemical details, we can see that the attachment type of reaction can be written Me→M, and the rate of electron loss is L N, where  is the attachment coefficient. The loss rate is now linear with N because the neutral species M is assumed to be by far the more numerous, in which case removing a few of them has no significant effect on their total number and [M] is effectively constant. At equilibrium, q Ne

(1.31)

1.3 Physical aeronomy

19

and taking q from the Chapman production function as before shows that the peak electron density now varies as Nm Nm0 cos .

(1.32)

Such a layer is a -Chapman layer. This simple formulation assumes that  does not vary with height, though this restriction does not affect the validity of Equation (1.31) at a given height. In fact  is expected to vary with height because it depends on the concentration of the neutral molecules (M), and this has important consequences for the form of the terrestrial ionosphere. It is known that electron loss in the F region occurs in a two-stage process: X A2 →AX A

(1.33)

AXe→AX

(1.34)

in which A2 is one of the common molecular species such as O2 and N2. The first step moves the positive charge from X to AX, and the second one dissociates the molecular ion through recombination with an electron, a dissociative-recombination reaction. The rate of Equation (1.33) is [X] and that of (1.34) is [AX]Ne. At low altitude  is large, (1.33) goes quickly and all X is rapidly converted to AX; the overall rate is then governed by the rate of (1.34), giving an -type process because [AX]Ne for neutrality. At a high altitude  is small, and (1.33) is slow and controls the overall rate. Then [X]Ne and the overall process appears to be of -type. As height increases, the reaction type therefore alters from -type to type. The reaction scheme represented by Equations (1.33) and (1.34) leads to equilibrium given by 1 1 1   , q  (h)Ne Ne2

(1.35)

where q is the production rate as before. The change from - to -type behaviour occurs at height ht where (ht ) Ne.

(1.36)

In the lower ionosphere there are also significant numbers of negative ions. Electrical neutrality then requires Ne N N, where Ne, N and N are, respectively, the concentrations of electrons, negative ions, and positive ions. Since the negative and positive ions may also recombine with each other, the overall balance between production and loss is now expressed by q eNeN  iNN,

(1.37)

Basic principles of the ionosphere

20

e and i being recombination coefficients for the reactions of positive ions with electrons and negative ions, respectively. The ratio between negative-ion and electron concentrations is traditionally represented by  – which has nothing to do with wavelength! In terms of , N  Ne and N (1 )Ne, and thus q(1 )(e  i)N e2,

(1.38)

which, in cases for which i  e, becomes q(1 )eN e2.

(1.39)

In the presence of negative ions the equilibrium electron density is still proportional to the square root of the production rate but its magnitude is changed. The term (1 )(e  i)

is often called the effective recombination coefficient. As we shall see in Section 1.4.3, the chemistry of the D region is complicated because of the presence of many kinds of positive and negative ions. 1.3.4

Vertical transport Diffusion

The final term of the continuity equation (1.15) represents changes of electron and ion density at a given location due to bulk movement of the plasma. Such movements can have various causes and can occur in the horizontal and the vertical planes in general, but since our present emphasis is on the overall vertical structure of the ionosphere, we shall concentrate here on the vertical movement of ionization, which, indeed, is very important in the F region. We assume now that photochemical production and loss are negligible in comparison with the effect of movements, and then the continuity equation becomes (wN ) dN ,  dt h

(1.40)

where w is the vertical drift speed and h is the height. We now suppose that this drift is entirely due to diffusion of the gas, and then we can put w

D N , N h

(1.41)

D being the diffusion coefficient. This equation simply states that the bulk drift of a gas is proportional to its pressure gradient, and it effectively defines the diffu-

1.3 Physical aeronomy

21

sion coefficient whose dimensions are (length)2/time. From kinetic theory (equating the driving force due to the pressure gradient to the drag force due to collisions as a minority gas diffuses through a stationary majority gas) the diffusion coefficient may be derived in its simplest form as DkT/(m). Here k is Boltzmann’s constant, T the temperature, m the particle mass and  the collision frequency. In the present case the minority gas is the plasma composed of ions and electrons, and the majority gas is the neutral air. However, for drift in the vertical direction the force of gravity also acts on each particle, adding to (or subtracting from) the drag force, and in this case we obtain w(D/N )(dN/dhN/HN)

(1.42)

for the upward speed instead of (1.41). Substitution into the continuity equation then gives

冤冢

dN dN N   D dt h dh HN

冣冥

.

(1.43)

This is the basic equation that has to be satisfied by the time and height variations of those regions (specifically the upper F region and the protonosphere) where ion production and recombination are both sufficiently small. In this equation the scale height HN merely represents the value of kT/(mg), and does not necessarily describe the actual height distribution. This is given by the distribution height, defined as

冢

 

1 dN N dh

冣

1

.

(1.44)

Using Equations (1.43) and (1.44) we can easily see that  is equal to the scale height at equilibrium. A complication is introduced by the fact that a plasma is composed of two minority species, ions and electrons, which have opposite charges and very different masses. Initially the ions, being heavier, tend to settle away from the electrons, but the resulting separation of electric charge produces an electric field, E, and a restoring force eE on each charged particle. This electrostatic force also affects the drift of the plasma. This problem is handled by writing separate equations for each species and including the electrostatic force on each. We assume (1). that the electron mass is small compared with the ion mass; (2). that ion and electron number densities are equal; and (3). that both species drift at the same speed; and then it can be shown that Equations (1.42) and (1.43) are still valid for a plasma if one replaces D and H by Dp k(TeTi )/(mii )

(1.45)

Basic principles of the ionosphere

22

and Hp k(TeTi )/(mi g),

(1.46)

respectively known as the ambipolar or plasma diffusion coefficient and the plasma scale height. In that part of the ionosphere where plasma diffusion is important, the electron temperature usually exceeds the ion temperature. However, taking Te Ti by way of illustration, we see that the plasma diffusion coefficient and scale height are then just double those of the neutral gas at the same temperature. Effectively, the light electrons have the effect of halving the ion mass since the two species cannot separate very far. At equilibrium dN/dhN/Hp and the plasma is exponentially distributed as N/N0 exp(h/Hp)

(1.47)

with scale height Hp. Note that this distribution has the same form as the upper part of a Chapman layer but with (about) twice the scale height. If the plasma is not in equilibrium the distribution changes with time at a rate depending on the value of the diffusion coefficient, which, since it depends on the relevant collision frequency, increases with altitude. If H is the scale height of the neutral gas, then the height variation of the diffusion coefficient can be written as DD0 exp(hh0)/H

(1.48)

where D0 is the value of D at a height h0. Thus, diffusion becomes ever more important at greater heights as the photochemistry becomes less important. Another consequence of the height variation of D is that it leads to a second solution of Equation (1.43) for the case dN/dt0. Substituting DD0 exp(h  h0)/H and NN0 exp  (h  h0)/ into (1.43) and rearranging, gives

冢

dN 1 1 DN  dt  Hp

冣冢

冣

1 1  .  H

(1.49)

If dN/dt0 this has two solutions. The first,  Hp, is diffusive equilibrium as has already been pointed out, and in this case the vertical drift speed (Equation (1.41)) is wD(1/ 1/Hp)0. The second solution is  H (H being the scale height of the neutral gas, governing the diffusion coefficient). Here, dN/dt0 as before, but the drift speed is

冢

冣 冢

冣

1 1 1 1 wD   D   ,  Hp H Hp

(1.50)

which is not zero since Hp H. The upward flow of plasma NwND(1/H1/Hp),

(1.51)

1.4 The main ionospheric layers

23

and in fact this is independent of height when  H because the height variations of D and of N cancel out. Thus, this second solution represents an unchanging distribution of electron density and a constant outflow of plasma.

The effect of a neutral-air wind Since the flow of ionospheric plasma is constrained by the geomagnetic field, the exact effect varies with latitude. One consequence is that, at middle latitudes, the height distribution of ionization is affected by the neutral-air wind which flows in the thermosphere. Suppose that the wind speed in the magnetic meridian is U and the magnetic dip angle is I. Then the component of the neutral wind along the direction of the magnetic field is U|| Ucos I, and the plasma tends to move in the same way. This motion, along the magnetic field, has a vertical component WU|| sinI 12 . Usin(2I).

(1.52)

Thus, a horizontal wind in the thermosphere tends to move the ionosphere up or down depending on its direction of flow. The effect is greatest where the magnetic dip angle is 45°. The consequences both for the height and for the magnitude of the peak of the F region can be significant (Section 1.4.5).

1.4

The main ionospheric layers

1.4.1

Introduction

The physical principles which govern the intensity and form of an ionospheric layer were outlined in Section 1.3. To work out what the actual ionosphere should be like on Earth or any other planet, we would have to consider the terms in Equation (1.19) (q  nI ) in detail to get the ion production rate, specify the ion chemistry to obtain values for the loss coefficients in Equations (1.29) and (1.31) (q N e2 and q Ne), and, at the higher levels, consider the diffusion coefficient (Equation (1.46)) and take movements into account. We should then require to know about the neutral atmosphere: its composition and physical parameters such as density and temperature. Then we should need full information on the solar spectrum and any fluxes of energetic particles able to ionize the constituents of the atmosphere. Knowing which gases could be ionized by the incident radiation, we could then determine the ionization rate of each species and sum over all wavelengths and all gases to get the total production rate in a given volume (q). If the loss processes indicated rapid attainment of equilibrium, the electron density (Ne) would be given by Equation (1.29) or (1.31). Otherwise a more complex computation would be required. (Mathematical modeling of the high-latitude ionosphere is discussed in Section 9.2.2.) There is no need to go into all these details here, but a few important points will be made.

Basic principles of the ionosphere

24

Table 1.1 lists the ionization potentials of various atmospheric gases. To be ionized a species must absorb a quantum of radiation whose energy exceeds the ionization potential. Since the energy of a quantum of wavelength  is Ehc/, there is a maximum wavelength of radiation that is able to ionize any particular gas. These values are included in Table 1.1. For easy reference the wavelengths are given both in ångström units and in nanometers. These values of max immediately identify the relevant parts of the solar spectrum as the X-ray (0.1–17 nm, 1–170 Å) and EUV, (17–175 nm, 170–1750 Å), emissions which come from the solar chromosphere and corona. The value of the absorption cross-section, , generally increases with increasing wavelength up to max and then falls rapidly to zero. There is no ionization at all by any radiation with wavelength exceeding max, regardless of its intensity. The ionization efficiency, , is such that, for atomic species, all the absorbed energy goes into ion production at the rate of one ion–electron pair for every 34 eV of energy. The energy is inversely proportional to the wavelength, and a convenient formula in terms of wavelength is  360/ (Å).

(1.53)

The Chapman theory (Section 1.3.2) shows that the production rate is a maximum at the level where the optical depth, nH sec , is unity. If the absorption at a given wavelength is due to several species, then the condition for maximum production is

兺 n H sec 1. i

i i

i

Table 1.1. Ionization potentials Maximum wavelength max Species

Ionization potential I (eV)

(Å)

(nm)

NO O2 H2O O3 H O CO2 N H2 N2 A Ne He

9.25 12.08 12.60 12.80 13.59 13.61 13.79 14.54 15.41 15.58 15.75 21.56 24.58

1340 1027 985 970 912 911 899 853 804 796 787 575 504

134.0 102.7 98.5 97.0 91.2 91.1 89.9 85.3 80.4 79.6 78.7 57.5 50.4

1.4 The main ionospheric layers

25

Wavelength (Å) 0 200

100

200

300

400

500

600

700

800

900 1000 1100 1200 1300 1400

Height (km)

150

100 He

N2

O, H

O2

NO

50 Lyman α F1 E D

Figure 1.6. The height at which the optical depth reaches unity for radiation vertically incident on the atmosphere. Ionization limits for common gases are marked. (J. D. Mathews, private communication.) The ranges responsible for the major ionospheric layers are indicated below.

The height of unit optical depth in a model terrestrial atmosphere is given as a function of wavelength in Figure 1.6 and this, not the intensity of the ionizing radiation, is what determines the height of the ionospheric layers. This is an important point. It means, simply, that strongly absorbed radiation produces ionization high up, and that low-level ionization must be due to radiation that is more weakly absorbed in the atmosphere. The simple theory of Section 1.3.2 deals with the shape and intensity of an ionosphere produced by monochromatic radiation acting on a single gas. On a real planet the effect of all gases at a given wavelength has to be considered and then, since the ionosphere is in effect a number of overlapping Chapman layers, the production rate due to all relevant wavelengths has to be summed at each height. The wavelength ranges giving the D, E, and F regions are summarized in Figure 1.6.

Basic principles of the ionosphere

26

1.4.2

The E and F1 regions Aeronomy

The E region which peaks at 105–110 km, and the F1 region at 160–180 km, are both fairly well understood. The F1 region is attributed to that part of the solar spectrum between about 200 and 900 Å, which is strongly absorbed in atomic oxygen, whose ionization limit is at 911 Å. The optical depth reaches unity from about 140 to 170 km. The band includes an intense solar emission line at 304 Å. The primary reaction products are O2 , N2 , O, He, and N, but subsequent reactions leave NO and O2 as the most abundant positive ions. The E region is formed by the less strongly absorbed, and therefore more penetrating, parts of the spectrum. EUV radiation between 800 and 1027 Å (the ionization limit of O2) is absorbed by molecular oxygen to form O2 . The band includes several important emission lines. At the short-wavelength end X-rays of 10–100 Å (1–10 nm) ionize all the atmospheric constituents. The main primary ions are N2 , O2 , and O, but the most numerous are again observed to be NO and O2 . The intensity of solar X-rays varies over the solar cycle and they probably make little contribution to the E region at solar minimum. Direct radiative recombination of the type eX →Xh

(1.54)

is slow relative to other reactions and is not significant in the normal E and F regions. Dissociative recombination, as eXY →XY,

(1.55)

is 105 times faster (with a reaction coefficient of 1013 m3 s1) and, both in the E region and in the F region, the electron and ion loss proceeds via molecular ions. The main recombination reactions of the E region are therefore eO2 →OO, eN2 →NN, eNO →NO.

(1.56)

In the F region the principal primary ion is O, which is first converted to a molecular ion by a charge-exchange reaction O O2 →O2 O

or O N2 →NO N.

(1.57)

1.4 The main ionospheric layers

27

The molecular ion then reacts with an electron as in Equation (1.56), to give as the net result eO O2 →OOO

or eO N2 →ONN.

(1.58)

In the F1 region the overall reaction is controlled by the rate of the dissociative recombination. Observations show that both the E and the F1 layers behave like, or almost like, -Chapman layers (Equation (1.30)). On average the critical frequency, fOE or f0F1 (Section 3.4.2), varies with the solar zenith angle, , as (cos )1/4, which means that the peak electron density, Nm, varies as (cos )1/2. The exponent is subject to some variation and ranges between about 0.1 and 0.4 for the E region. Given that the E region is an -Chapman layer, the Chapman theory can be applied to determine the recombination coefficient () from observations, and this may be done using Equation (1.29): (1). taking an observed electron density and an observed or computed production rate; (2). by observing the rate of decay of the layer after sunset and assuming that q0; or (3). by measuring the asymmetry of the diurnal variation about local noon, an effect sometimes called the sluggishness of the ionosphere, the time delay being given by  1/(2N).

(1.59)

Such methods give values of  in the range 1013–1014 m3 s1 (107–108 cm3 s1).

The night E layer The E layer does not quite vanish at night, but a weakly ionized layer remains with electron density about 5 109 m3 (against 1011 m3 by day). One possible cause is meteoric ionization, though other weak sources might also contribute. Figure 1.7 shows speciman electron-density profiles of the E region for day and night, measured by incoherent-scatter radar.

Sporadic-E The most remarkable anomaly of the E region is sporadic-E, often abbreviated to Es. On ionograms sporadic-E is seen as an echo at constant height that extends to a higher frequency than is usual for the E layer; for example to above 5 MHz. Rocket measurements, and more recently incoherent-scatter radar, show that, at mid-latitude, these layers are very thin, perhaps less than a kilometer across. Examples are shown in Figure 1.8.

Basic principles of the ionosphere

28

Figure 1.7. Speciman electron-density profiles of the E region for night and day, measured by the incoherent-scatter radar at Arecibo, Puerto Rico (18° N, 67° W), in January 1981. (J. D. Mathews, private communication.)

Figure 1.9 indicates the probability of occurrence of sporadic-E against time of day and season in three latitude zones: ■

the equatorial zone, within 20° of the magnetic equator;

■

the high-latitude zone, poleward of about 60° geomagnetic;

■

and the temperate zone in between.

The high-latitude zone may be sub-divided into the auroral zone (approximately 60°–70° magnetic) and the polar cap (poleward of the auroral zone). A full classification of sporadic-E, particularly regarding its identification on ionograms, is given by Piggott and Rawer (1972). In general, sporadic-E exhibits little direct relationship with the incidence of solar ionizing radiation. Sporadic-E tends to be particularly severe at low latitude. It occurs frequently during the daytime hours, often with sufficient intensity to reflect radio waves up to 10 MHz. A major cause is the occurrence of instabilities in the equatorial electrojet (Section 1.5.5). The principal cause of sporadic-E at middle latitude is a variation of wind speed with height, a wind shear, which, in the presence of the geomagnetic field,

1.4 The main ionospheric layers

Figure 1.8. Some sporadic-E layers observed at Arecibo by incoherent-scatter radar, January 1981. (J. D. Mathews, private communication.)

acts to compress the ionization by a mechanism similar to that which allows the neutral-air wind in the thermosphere to raise or lower the F region (Section 1.3.4). The time scale of the process needs ions of relatively long life, and it is thought that these are metallic ions of meteoric origin such as Fe, Mg, Ca, and Si. Being atomic, these cannot recombine dissociatively and therefore their recombination coefficients are typical of the radiative process (1018 m3 s1), which gives them relatively long lifetimes. Temperate sporadic-E occurs at heights of 95–135 km, and the most probable height is 110 km. It occurs most frequently in summer daytime, with maxima in mid-morning and near sunset. The seasonal variation is complex. Its character changes abruptly at about 60° magnetic latitude, the boundary of auroral Es. The sporadic-E which occurs at high latitude is attributed to ionization by incoming energetic particles in the energy range 1–10 keV. It is mainly a night-time phenomenon, correlating to magnetic activity (Section 2.5.3), but not to sunspot activity as such. Clouds of auroral Es drift at speeds between 200 and 3000 m s1, westward in the evening and eastward in the early morning, much like the aurora. The layer may be either “thick” or “thin”. Within the polar caps sporadic-E has

29

30

Basic principles of the ionosphere

Figure 1.9. Diurnal and seasonal occurrence patterns for three kinds of sporadic-E. (a) The auroral kind maximizes at night but exhibits no seasonal variation. (b) The temperate kind peaks near noon in summer. (c) The equatorial kind occurs mainly by day but has no seasonal preference. (After E. K. Smith, NBS Circular 582, US National Bureau of Standards, 1957.)

a different character. It is weaker, and exhibits a negative correlation to magnetic activity. It takes the form of bands or ribbons extending across the polar cap in a roughly sunward direction. The properties and causes of sporadic-E have been reviewed in detail by Whitehead (1970). The high-latitude E region is discussed further in Section 6.5. Sporadic-E is significant in radio propagation because it may reflect signals that would otherwise penetrate to the F region, though in some cases (for example

1.4 The main ionospheric layers

the equatorial type) it is partly transparent. The irregularities within a sporadicE layer can scatter radio waves if their dimensions are comparable to half a radio wavelength, and at times they may cause scintillation of trans-ionospheric signals, though F-layer irregularities are the more usual cause of this phenomenon.

The F1 ledge The strange thing about the F1 region is that it does not always appear! In fact, real-height profiles show that it seldom exists as a distinct peak and for this reason it is more correctly called the F1 ledge. The ledge is more pronounced in summer and at sunspot minimum, and it is never seen in winter at sunspot maximum. The explanation is to be found by comparing ht, the height at which transition between -type and -type recombination occurs, as discussed in Section 1.3.3, and hm, the height of maximum electron-production rate. The F1 ledge appears only if ht hm, and, since ht depends on the electron density (Equation (1.36)), the ledge vanishes when the electron density is greatest. 1.4.3

The D region Aeronomy

The D region of the ionosphere does not include a maximumum but is that part below about 95 km which is not accounted for by the processes of the E region. It is also the most complex part of the ionosphere from the chemical point of view. This is due, first, to the relatively high pressure, which causes minor as well as major species to be important in the photochemical reactions, and, second, because several different sources contribute to ion production. The Lyman- line of the solar spectrum at 1215 Å penetrates below 95 km and ionizes the minor species nitric oxide (NO), whose ionization limit is at 1340 Å. This is the main source at middle latitudes, though not necessarily at all heights. There is a smaller contribution from the EUV spectrum between 1027 and 1118 Å, which ionizes another minor constituent, molecular oxygen in an excited state. At the higher levels ionization of O2 and N2 by EUV, as in the E region, makes a contribution. Hard X-rays of 2–8 Å ionize all constituents, the most effect being therefore from the major species O2 and N2. Since the intensity of the solar X-ray emissions varies considerably from time to time, this source is sometimes a major one but at other times only minor. The lowest levels are dominated by cosmic-ray ionization, which continues by night as well as by day and affects the whole atmosphere down to the ground. The production rate due to cosmic rays increases downward in proportion to the total air density, and, since the production from other sources is falling off, it is inevitable that the cosmic rays must come to dominate at some level. At high latitudes particles from the Sun or of auroral origin ionize the D region and at times they form the main source. We shall be particularly concerned with those sources and their effects later in the book.

31

32

Basic principles of the ionosphere

Figure 1.10. Calculated production rates at 42° due to extreme ultra-violet (EUV), Lyman- and nitric oxide (NO), X-rays (X), excited oxygen (O*2), and galactic cosmic rays (GCR). (J. D. Mathews, private communication.)

Clearly, the relative contributions of these different sources vary with latitude, time of day, and level of solar activity. By way of example, theoretical profiles of the production rate (for solar zenith angle 42° and a 10-cm solar flux of 165 units) are given in Figure 1.10. Note that all the sources mentioned above are significant and that their relative importance depends on the altitude. At greater solar zenith angles the contributions from Lyman- and X-rays are reduced, and the cosmic rays become relatively more important below 70 km. The X-ray flux varies strongly with solar activity (by a factor of a hundred to a thousand) and is probably not significant in the D region at sunspot minimum. These production-rate profiles are consistent with measurements of D-region electron densities (Figure 1.11). Friedrich and Torkar (1992) analyzed 164 electron-density profiles of the D region measured by rocket-based wave-propagation techniques (as in Section 4.3.4), to derive an empirical model covering a range of solar zenith angles. Figure 1.12 shows a set of profiles corresponding to a sunspot number of 60. Following ionization, the primary ions in the D region are NO, O2 , and N2 , but the latter are rapidly converted to O2 by the charge-exchange reaction N2 O2 →O2 N2,

(1.60)

leaving NO and O2 as the major ions. However, below 80 or 85 km, apparently the level of the mesopause, are detected heavier ions that are hydrated species such as H.H2O, H3O.H2O, and hydrates of NO. These hydrates occur when the concentration of water vapor exceeds about 1015 m3. The level at which hydration first occurs is a natural boundary within the D region.

1.4 The main ionospheric layers

33

Figure 1.11. Electron-density profiles observed at Arecibo for two solar zenith angles. (J. D. Mathews, private communication.)

1019

10 20

70° 60° 40° 20°

Figure 1.12. Electrondensity profiles in the D region derived from rocket measurements for a range of solar zenith angles. The number density of the neutral air is also shown. (M. Friedrich and K. M. Torkar, Radio Sci. 27, 945, 1992. Copyright by the American Geophysical Union.)

100 ALTITUDE, km

NEUTRAL DENSITY, m –3

1018

80°

160° 90°

120

80 10 21

10 22

60 10 7

10 8

10 9

10 10

ELECTRON DENSITY,

1011

1012

m –3

Where simple ions dominate, the loss process is dissociative recombination as in the E region, with a recombination coefficient of about 5 1013 m3 s1, the reaction of NO being somewhat faster than that of O2 . In total the situation is , NO, much more complex, as illustrated in Figure 1.13. This scheme includes O 2 , hydrates and others, and has to be solved by means of a computer program. O 4 The hydrated ions, being larger molecules, have greater recombination rates than do the simple ions, of the order of 1012–1011 m3 s1, depending on their size. Thus the equilibrium electron density is relatively smaller in regions where hydrates dominate.

Figure 1.13. A scheme of positive-ion chemistry for the D region. (E. Turunen, private communication.) This model, developed at Sodankylä Geophysical Observatory, Finland, includes 24 positive and 11 negative ions, 35 in all. Later versions include as many as 55 ions.

1.4 The main ionospheric layers

35

Below about 70 km by day or 80 km by night much of the negative charge is in the form of negative ions. Their creation begins with the attachment of an electron to an oxygen molecule, forming O2 . This is a three-body reaction involving any other molecule, M, whose function is to remove excess kinetic energy from the reactants: eO2 M→O2 M.

(1.61)

This is followed by further reactions forming other and more complex negative ions such as CO3, NO2 , and NO3 (the most abundant negative ion in the D region) and clusters such as O2 .O2, O2 .CO2, and O2 .H2O. Because the electron affinity of O2 is small (0.45 eV), the electron may be removed by a photon of visible or near infra-red light: O2 h →O2 e.

(1.62)

It may also be detatched through chemical reactions, such as with atomic oxygen (forming ozone), and with excited molecular oxygen. The effect of negative ions on the balance between electron production and loss was included in Equations (1.37)–(1.39). Variations of electron density in the D region can be due to changes in the negative-ion/electron ratio, , as well as to changes in production rate. The complexity and uncertainty of D-region photochemistry is one reason why, when one is relating electron-production rates to electron densities, it is usual to work with an “effective recombination coefficient” (Equation (1.38)), which may be either theoretically or experimentally determined.

Diurnal behavior Although the mid-latitude D region is complex chemically, observationally its behavior may be deceptively simple. The region is under strong solar control and it vanishes at night. VLF ( f 30 kHz) radio waves are, to a first approximation, reflected as at a sharp boundary in the D region because the refractive index changes markedly within one wavelength (Section 3.4.6). For VLF waves incident on the ionosphere at steep incidence, the reflection height, h, appears to vary as hh0 Hln(sec ),

(1.63)

where is the solar zenith angle. h0 is about 72 km, and H is about 5 km, which happens to be the scale height of the neutral gas in the mesosphere. This form of height variation is just what is predicted for a level of constant electron density in the underside of a Chapman layer, and it is consistent with the ionization of NO by solar Lyman- radiation. At oblique incidence, when the transmitter and the receiver are more than about 300 km apart, the height variation follows a quite different pattern. The

36

Basic principles of the ionosphere

Figure 1.14. Two kinds of diurnal behavior of the D region inferred from VLF radio propagation at vertical and oblique incidence. The regions originally called D and D are now more usually called D and C. The evening recovery at oblique incidence tends to be more gradual than that in a simple D pattern and similar to the dashed curve. (After R. N. Bracewell and W. C. Bain, J. Atmos. Terr. Phys., 2, 216, Copyright 1952, with permission from Elsevier Science.)

reflection level now falls sharply before ground sunrise, remains almost constant during the day, and then recovers fairly rapidly following ground sunset. The reason has to do with the formation and detachment of negative ions at sunset and sunrise, coupled with electron production by cosmic-ray ionization – a source with no diurnal variation. This lower part of the D region is sometimes called a C layer. These patterns of height variation are illustrated in Figure 1.14.

Radio absorption The D region is the principal seat of radio absorption, and absorption measurements (Section 4.2.4) are one way of monitoring the region. The absorption per unit height depends both on the electron density and on the frequency of collisions between electrons and neutral particles, and the measurement gives the integrated absorption up to the reflection level. Multi-frequency absorption measurements can provide some information about the height distribution. Generally, the absorption varies with the solar zenith angle as (cos )n with n in the range 0.7–1.0. However, the seasonal variation contains an intriguing anomaly, which is that, during the winter months, the absorption exceeds by a factor of two or three the amount that would be expected by extrapolation from summer. Moreover, the absorption is much more variable from day to day in the winter. This phenomenon is the winter anomaly of ionospheric radio absorption.

1.4 The main ionospheric layers

1.4.4

37

The F2 region and the protonosphere The peak of the F2 layer

Compared with the good behavior of the lower layers of the ionosphere, the F2 region, on first aquaintance, can be quite puzzling. In the first place it peaks at 200–400 km, whereas Figure 1.6 shows no band of radiation producing a maximum ionization rate at any height above 180 km. The answer is to be found in the height variation of the recombination rate, which forms the F2 region as an upward extension of F1 even though the production rate is now decreasing with height. Taking O as the major ion, the two-stage recombination process is O N2 →NO N

with rate  [O]

followed by with rate  [NO]Ne.

NO e→NO

As discussed in Section 1.3.3, the second reaction controls the overall rate at low altitude and the first is the rate-determining step at high levels, the transition being where Ne  (ht). The transition height, ht, is generally between 160 and 200 km. The F1 ledge can appear if ht is above the height of the maximum production rate, hm: that is, if there is a production maximum within an -type region. To explain the F2 region we consider the upper part where the recombination is of  type, and where  depends on the concentration of N2. On the other hand, the production rate depends on the concentration of O.Thus, at equilibrium, Ne q/  [O]/[N2]

冢

Ne q/  exp 

h h  H(O) H(N2 )

冣

where H(O) and H(N2) are the scale heights for O and N2. Since the masses of N2 and O are in the ratio 1.75: 1, this rearranges to give

冢

冤

h H(O) 1 H(O) H(N2 )

冢

0.75h . H(O)

Ne  exp 

exp 

冣

冣冥 (1.64)

This is a layer whose electron density increases with height because the loss rate falls off more quickly than does the production rate. It is often called a Bradbury layer.

38

Basic principles of the ionosphere

The Bradbury layer explains why the electron density increases with height above the level of maximum ion production, but it does not explain why the F2 layer has a maximum. Here we have to invoke plasma transport. At the higher levels, in situ production and loss are less important than diffusion, which has become more important because of the decreasing air density. (That is, the righthand side of Equation (1.15) is now dominated by the third term.) The F2 layer peaks where chemical recombination and diffusion are equally important. To decide the level at which this will occur, we regard the two loss processes – -type recombination and transport – as being in competition, and compare their time constants for electron loss on the principle that the more rapid will be in effective control. The characteristic time for recombination is  1/,

(1.65)

and it may be shown that the corresponding time for diffusion is approximately D H 12/D,

(1.66)

where H1 is a typical scale height for the F2 region. Comparing these two equations places the F2 peak at the level where  ⬃D/H 12.

(1.67)

The electron density at the peak is given by Nm ⬃qm /m.

(1.68)

The protonosphere At some level in the topside the ionosphere dominated by O gives way to the protonosphere dominated by H. It so happens that the ionization potentials for these two ions are almost the same (Table 1.1), and therefore the reaction HO
 H O

(1.69)

goes rapidly in either direction, and, around the transition level, the equilibrium is given by [H][O](9/8)[H][O].

(1.70)

(The factor 9/8 arises for statistical reasons, and there is also a temperature dependence proportional to (Tn/Ti)1/2.) Through this reaction ionization can move

1.4 The main ionospheric layers

39

readily between the ionosphere (as O) and the protonosphere (H). This is a very important aspect of the behavior of the topside ionosphere. The transition effectively defines the base of the protonosphere. Below that level the H distribution is determined by (1.71), and is related to the distribution of O by [H]  [H][O]/[O] exp[h/H(H)]exp[h/H(O)]/exp[h/H(O)] exp[7h/H(H)].

(1.71)

There is a strong upward gradient in the H concentration below the transition level. Above the transition the concentration of O decreases rapidly, and in this region the protonosphere, when it is in equilibrium, takes an exponential profile with the appropriate scale height (Equation (1.47)). As for the F2 peak, the transition level between ionosphere and protonosphere can be estimated by comparing time constants. If the rate constant of the reaction H O→HO

is k, then the lifetime of a proton is (k[O])1. Taking the time constant for diffusion in the protonosphere as H 22/D, the boundary occurs where k[O]⬃D/H 22.

(1.72)

This occurs at 700 km or higher, which is always well above the peak of the F2 layer. 1.4.5

Anomalies of the F2 region The phenomena

The F2 region has the greatest concentration of electrons of any layer, and therefore it is the region of greatest interest in radio propagation. Unfortunately, it is also the region which is the most variable, the most anomalous, and the most difficult to predict. From the point of view of the Chapman theory the F2 region’s behavior is anomalous in several ways, and these are sometimes called the classical anomalies of the F2 layer. Briefly, they are as follows. (a). The diurnal variation may be asymmetrical about noon. There may be a rapid change at sunrise but little or no change in the evening until well after sunset or even until just before the next sunrise (Figure 1.15). The daily peak may occur either before or after local noon in the summer,

40

Basic principles of the ionosphere

Figure 1.15. The diurnal behavior of f0F2 on successive days in December 1959 at a lowlatitude station, Talara, Peru. Note, by contrast, the regularity of the E layer. (T. E. VanZandt and R. W. Knecht, in Space Physics (eds. Le Galley and Rosen), Wiley, 1964.)

though it is likely to be near noon in the winter (Figure 1.16). On some days a secondary minimum appears near noon between the morning and evening maxima (Figure 1.16(a)). (b). The daily pattern of variation often does not repeat from day to day. (If it did, the next day could at least be predicted from the previous one.) Figure 1.15 illustrates this point. (c). There are several anomalous features in the seasonal variation. The main one is that noon values of the F-layer critical frequency (see Equation (3.67)) are usually greater in winter than they are in summer, whereas the Chapman theory leads us to expect the opposite. This is the seasonal anomaly, which is clear in Figure 1.16. The summer electron content (the summation of electron density in a column through the ionosphere) is greater than the winter value at some stations, but at others it is smaller or about the same. The electron content is abnormally large at the equinoxes, giving the semi-annual anomaly. Some stations also show this anomaly in the F-region critical frequency (Figure 1.17). (d). The mid-latitude F2 region does not vanish at night, but remains through to the next sunrise at a substantial level. Although not all anomalies have yet been fully explained, it now appears that there are four main causes for this seemingly anomalous behaviour:

1.4 The main ionospheric layers

Figure 1.16. (a) The diurnal behavior of f0F2 in summer and winter at a high-latitude station in the northern hemisphere, Adak, Alaska. The F region is anomalous whereas the E layer behaves as expected according to the Chapman theory. (T. E. VanZandt and R. W. Knecht, in Space Physics (eds. Le Galley and Rosen), Wiley, 1964.) (b) Summer and winter electron contents measured at Fairbanks, Alaska. (R. D. Hunsucker and J. K. Hargreaves, private communication.)

41

Figure 1.17. Variations of critical frequencies over several sunspot cycles. The three top panels show the sunspot number, the 10.7-cm solar radio flux, and the magnitude of the interplanetary magnetic field. (Diagram provided by M. Wild, Rutherford Appleton Laboratory, Chilton, UK.) Note also the seasonal modulations at Slough and Port Stanley. The E and F1 regions peak in the summer whereas F2 peaks in the winter. The semi-annual anomaly is prominent at Port Stanley.

1.4 The main ionospheric layers

(a). reaction rates are sensitive to temperature; (b). the chemical composition varies; (c). there are winds in the neutral air that lift or depress the layer by the mechanism indicated in Section 1.3.4; and (d). the ionosphere is influenced by the protonosphere and by conditions in the conjugate hemisphere.

Reaction rates Reaction rates are generally temperature sensitive. The rate for the reaction O N2 →NO N,

the first step in an important two-stage loss process (Equations (1.57) and (1.56)), varies strongly with the temperature of neutral N2 and increases by a factor of 16 between 1000 and 4000 K. This property obviously contributes both to the persistence of the night F region and to the seasonal anomaly.

Composition Since the electron-production rate depends on the concentration of atomic oxygen, O, whereas the loss rate is controlled by the molecular species N2 and O2, increases in the ratios [O]/[O2] and [O]/[N2] will increase the equilibrium electron density. Satellite measurements have shown that such variations do occur. The ratio [O]/[N2] at 250–300 km is measured as about 6 in winter and about 2 in summer, a seasonal change amounting to a factor of three. The change of composition is attributed to the pattern of global circulation in the thermosphere. This is plainly a factor in the seasonal anomaly.

Winds Mathematical modeling has demonstrated how the meridional component of the thermospheric neutral wind, acting to depress the ionosphere when the wind is flowing equatorward and elevating it when it is flowing poleward (Section 1.3.4), exerts a major influence both on electron densities and on electron content. At 300 km the neutral wind flows poleward by day and equatorward by night at speeds ranging between tens and hundreds of m s1. Thus its effect is usually to depress the ionosphere and thereby increase the rate of loss by day, but to lift the region and reduce its rate of decay at night. It is estimated (taking H60 km for the neutral scale height, D2 106 m2 s1 for the diffusion coefficient, and W30 m s1 as a typical vertical drift due to the poleward wind), that by day the peak of the layer is lowered by about 50 km. The variability of the F region from one day to the next (e.g. Figure 1.15) is one of its most remarkable and puzzling features. This might not be surprising in the polar regions because of the sporadic nature of solar and auroral activity, but

43

Basic principles of the ionosphere

44

these are not dominant influences at middle latitudes. Presumably the origin must be a source in the terrestrial atmosphere or in the solar wind. Variations of the neutral-air wind in the thermosphere are one possible cause.

The plasma temperature and the protonosphere Variations in the temperature of the plasma affect its vertical distribution. The heating comes from the excess energy of absorbed photons above that needed for ionization. The excess energy is initially in the electrons and it is gradually shared with the positive ions, though transfer to the neutral species is less efficient. Consequently the plasma is hotter than the neutral air, and within the plasma the electrons are hotter than the ions (Te Ti). The electron temperature can be two or three times the ion temperature by day, though by night the electron and ion temperatures are more nearly equal. These changes in temperature strongly affect the distribution of F2-region plasma. When it is hotter, the plasma has a greater scale height (Equation (1.46)) and so spreads to greater altitudes, where it tends to persist for longer because the loss rate is smaller. At the greater altitudes the positive ions are protons, and, as discussed in Section 1.4.4, the ionosphere and the protonosphere are strongly coupled through the charge-exchange reaction between protons and atomic oxygen ions (Equation (1.69)). As the F region builds up and is also heated during the hours after sunrise, plasma moves to higher altitudes where protons are created. These then flow up along the field lines to populate the protonosphere. In the evening the proton population flows back to lower levels, where it undergoes charge exchange to give oxygen ions and so helps to maintain the F region at night. Via the protonosphere the magnetically conjugate ionosphere may also have an effect, since protonospheric plasma, coming mainly from the summer ionosphere, is equally available to replenish the winter ionosphere. Computations show that this is a significant source. Indeed, it is useful to treat the mid-latitude plasmasphere as consisting of winter and summer ionospheres linked by a common protonosphere; the ionospheres act as sources to the protonosphere, which in turn serves as a reservoir to the ionospheres. Overall, the winter ionosphere benefits from the conjugate region in the summer hemisphere. At sunrise, when electron densities are low, the ionosphere may be significantly heated by photoelectrons arriving from the conjugate hemisphere. The effect may show up as an increase of slab thickness (the ratio of electron content to maximum electron density) just before local sunrise. It appears likely that the various classical anomalies of the F2 region arise from combinations of the factors outlined above, though the details might not be clear in any particular case. 1.4.6

The effects of the sunspot cycle

The varying activity of the Sun over a period of about 11 years, measured in terms of the number of sunspots visible on the disk, the rate at which flares occur, or the

1.4 The main ionospheric layers

45

intensity of the 10-cm radio flux, also affects the ionosphere because of variations in the intensity of the ionizing radiations in the X-ray and EUV bands. The temperature of the upper atmosphere also varies with solar activity, approximately by a factor of two between sunspot minimum and maximum. Consequently, the gas density at a given height varies by a large factor. The maxima of the E, F1, and F2 layers all depend on the number of sunspots, R. This influence can be seen in Figure 1.17. (The critical frequencies plotted there, fOE, fOF1, and fOF2, are proportional to the square root of the maximum electron density, and are defined as the highest radio frequencies reflected from the layer at vertical incidence – see Section 3.4.2.) We have seen that the E and the F1 layers both behave as -Chapman layers. In such a layer (Equation (1.30)) the critical frequency varies with the solar zenith angle as (cos )1/4. Taking the number of sunspots into account as well gives two empirical relations: fOE3.3[(1 0.008R)cos ]1/4 MHz

(1.73)

fOF14.25[10.015R)cos ]1/4 MHz.

(1.74)

Note that the F1 layer is nearly twice as sensitive as the E layer to variations in the sunspot number. From the status of the E and F1 as -Chapman layers it follows that the ratios (fOE)4/cos and (fOF1)4/cos are proportional to the ionization rates (q) in the E and F1 layers, respectively. These ratios are called character figures. Taking R10 for a typical solar minimum and R150 for a maximum, we see from Equation (1.73) that the E-region production rate varies by a factor of two over a typical sunspot cycle. The F2 layer does not behave like a Chapman layer but it nevertheless varies with the sunspot number. The dependence may be seen by plotting the noon values of fOF2, and if these are smoothed over 12 months to remove the seasonal anomalies, a dependence such as fOF2  (10.02R)1/2 MHz

(1.75)

can be recognized. One measure of the strength of the D region is the radio absorption measured, for example, by the pulsed sounding technique (Section 4.2.4). Other parameters being constant, it is observed that the absorption increases by about 1% for each unit of sunspot number: A(dB)  (10.01R).

(1.76)

At mid-latitude the absorption is expected to vary over a sunspot cycle by about a factor of two.

Basic principles of the ionosphere

46

1.4.7

The F-region ionospheric storm

From time to time the ionosphere suffers major perturbations called storms. They last from a few hours to a few days and tend to occur during times of geophysical disturbance resulting from increases in solar activity communicated via the solar wind. There are, on the face of it, connections with magnetic storms (Section 2.5), though some different mechanisms must be involved. Three phases may be identified. (a). In the initial or positive phase, which lasts for a few hours, the electron density and the electron content are greater than normal. (b). Then follows the main or negative phase when these quantities are reduced below normal values. (c). Finally, the ionosphere gradually returns to normal over a period of one to several days in the recovery phase. The magnitude of the effect varies with latitude, being greatest at middle and high latitude, where the maximum electron density may be depressed by 30% in a strong storm. At latitudes below about 30° the effect is not likely to exceed a few percent. The beginning can be sudden or gradual, the term sudden commencement being used (as for magnetic storms) to describe the former. At middle latitudes ionosondes show the apparent height of the maximium, h(F2), to be increased, though real-height analysis attributes this mainly to greater group retardation (Section 3.4.2) below the peak rather than to a genuine lifting of the region. The slab thickness (the ratio of the electron content to Nmax) does increase, however, confirming that the F region broadens during the negative phase. Figure 1.18 compares electron content, electron density, and slab thickness in a typical midlatitude storm. The progress of the storm since its time of commencement is the storm-time variation, but the time of day is also a significant parameter. Statistical studies, as well as case histories of major storms, show that the magnitude and even the sign of the effect depend on the time of day. The negative phase tends to be weaker in the afternoon and evening, stronger in the night and morning. The positive phase is often missing altogether at stations that were in the night sector at commencement. It has been suggested (Hargreaves and Bagenal, 1977) that the positive phase co-rotates with the Earth on the first day of the storm and does not reappear on the second day. Seasonal and hemispheric effects are also marked. The negative phase is relatively stronger, and the positive phase relatively weaker, in the summer hemisphere. This holds both for the northern and for the southern hemisphere, though the interhemispheric difference is such that Nm(F2) is actually increased during the main phase of storms occurring in the southern hemisphere during winter. The interhemispheric difference arises from the larger separation between the geographic and the geomagnetic poles in the south.

1.4 The main ionospheric layers

Figure 1.18. The electron content, electron density, and slab thickness at a mid-latitude station during an F-region storm. SC marks the time of sudden commencement. The 7-day mean is shown to indicate normal behavior. (M. Mendillo and J. A. Klobuchar, Report AFGRL-TR-74-0065, US Air Force, 1974.)

The most likely cause of the main phase is abnormal heating at high latitude, which also alters the pattern of circulation of the thermospheric wind. The heating reduces the ratio [O]/[N2] at given height in the F region, and the molecularly enriched air is then convected down to the middle latitudes by the changed air circulation. As was pointed out in Section 1.4.5, the effect of a greater proportion of molecular species in the F region is to reduce the equilibrium electron density. This mechanism has been verified by computer modeling (Rishbeth, 1991), though some problems remain to be solved. There appears to be no generally agreed cause of the initial phase, though various mechanisms have been suggested.

47

Basic principles of the ionosphere

48

1.5

The electrical conductivity of the ionosphere

1.5.1

Introduction

The presence of free electrons and ions allows the ionospheric layers to carry electric currents. The conductivities of the ionosphere lie in the range 105–102 1 m1, a broad middle range between insulators (such as the tropospere, ⬃1014 1 m1) and good conductors like metals ( 6 107 1 m1 for copper), being akin to that of the ground (107–1 1 m1) or a semiconductor (101–102 1 m1). Radio propagation is generally considered in terms of the electron density of an ionospheric layer rather than its conductivity, and we shall not need to deal with conductivities very much, at least for propagation in the MF, HF, or VHF bands. However, the electric currents of the ionosphere and magnetosphere are a major factor in the behavior of the ionosphere and in the way it is affected by geophysical disturbances. These are particularly important at the high latitudes. The solar–geophysical environment, of which the ionosphere is a part, cannot be understood without including the several current systems that may exist within it. Hence, we give in this section the basis of ionospheric conductivity. 1.5.2

Conductivity in the absence of a magnetic field

If no magnetic field is present, the formula for the conductivity of an ionized gas is a simple one: 0 Ne2/(m),

(1.77)

where N is the number density of particles each with charge e and mass m, and  is the collision frequency for collisions of a charged particle with neutral species (which are assumed to be in the majority). The formula is easily proved, remembering that the mobility of a charged particle (its velocity in a unit electric field) is e/(m), and the total charge per unit volume is Ne. If more than one species of charge is present, for example electrons and positive ions, the total conductivity is the sum of the conductivities for each species separately. 1.5.3

The effect of a magnetic field

Unfortunately the Earth’s magnetic field permeates the ionosphere, and this complicates the conductivity enormously. A charged particle moving through a magnetic field experiences a force (the Lorentz force) that acts at right angles both to

1.5 Electrical conductivity

49

Pe E⊥ der sen cur ren t

E ||

Direct current

B

Hall current

Figure 1.19. Currents due to the electric-field components parallel (E
) and perpendicular (E⬜) to the magnetic field (B). The currents shown are those due to positive charges. The direct and Pedersen currents due to negative charges are the same as those shown, but the Hall current is opposite. The Hall current in the ionosphere is mainly due to electrons.

the direction of the magnetic field and to the direction of motion of the particle. If the particle is moving directly along the magnetic field, the Lorentz force is zero; the magnetic field has no effect and Equation (1.77) applies. However, if the motion has a component at right angles to the magnetic field, the corresponding conductivity has two parts: 1 

冢

Ne  2e  2i Ni  e2 mee (  2e  2e ) mii (  2i   2i )

冣

(1.78)

2 

冢

Ne e  e N  i i  i e2. mee (  2e  2e ) mii (  2i   2i )

冣

(1.79)

The subscript e here refers to electrons and i refers to positive ions.  is the relevent gyrofrequency (eB/m, where B is the magnetic flux density). 1 is the Pedersen conductivity, which gives the current in the same direction as the applied electric field, whereas 2 is the Hall conductivity giving the current at right angles to it – it being understood that the electric field and the currents are all in the plane normal to the magnetic field. Figure 1.19 may clarify the geometry.

Basic principles of the ionosphere

50

Figure 1.20. Conductivity profiles calculated for middle latitude at noon. (S.-I. Akasofu and S. Chapman (after K. Maeda and H. Matsumoto), Solar–Terrestrial Physics, Oxford University Press, 1972. By permission of Oxford University Press.). Multiply the conductivity values by 1011 to convert them to the SI unit 1 m1.

1.5.4

The height variation of conductivity

It is clear from Equations (1.78) and (1.79) that the conductivity due to a single species depends on the ratio /. Indeed, the ratio between the Hall and Pedersen conductivities for a given electron (or ion) density is just /, and is therefore strongly height-dependent. Note, also, that, in Equation (1.79) the electron and ion terms are of opposite sign, so the total Hall conductivity depends on the difference between the electron and ion conductivities, not on their sum. Figure 1.20 illustrates the height variations of the direct, Pederson, and Hall conductivities in a typical mid-latitude ionosphere. The Hall conductivity peaks in the E region, the Pedersen conductivity peaks somewhat higher, and the direct conductivity continues to increase with height. The Hall conductivity is very small in the F region because the electron and ion components almost cancel out there. Figure 1.21 indicates the motions of ions and electrons, and the resulting electric current, at various key altitudes. In the upper panel the driving force is a wind in the neutral air, which induces ion motion through collisions. The effect of an electric field is shown in the lower panel. 1.5.5

Currents

For there to be an electric current there must also be a driving force (either a wind or an electric field) and a path of conductivity providing a complete circuit. Where

1.5 Electrical conductivity

51

Ion and electron motions due to a wind U U

B (in)

B

Height: 60 km νe >> ω e

75 km νe ∼ ω e

100 km νi >> ω i νe << ω e

125 km νi ∼ ω i

>150 km νi << ω i

E Ion and electron motions due to an electric field E

Key:

Vi IONS

Ve ELECTRONS

j ∝ (V i – Ve) CURRENT

Figure 1.21. Ion and electron motions due to a neutral-air wind (top) and an electric field (bottom) at selected key altitudes. The current is proportional to the vector difference between the ion and electron velocities. (After H. Rishbeth. J. Inst. Electronic Radio Engineers 58, 207, 1988.)

the latter is not present, the flow of current is inhibited or modified by the electric potentials created at the boundaries. The geomagnetic equator is one interesting case. Here the magnetic field runs horizontally and therefore the current which would otherwise flow normal to the field is inhibited in the vertical direction. Charges are created at the upper and lower boundaries, and the resulting electric field acts to increase the current in the horizontal plane. It can be shown that, in this special situation, the conductivity across the magnetic field and in the horizontal direction is given by 3  1  22/ 1,

(1.80)

called the Cowling conductivity. The value of the Cowling conductivity is comparable to that of the direct conductivity (Equation (1.77)), and therefore the current over the magnetic equator is abnormally large. This is the equatorial electrojet. The large value of the direct conductivity suggests that current should be able to flow readily along the geomagnetic field direction. The existence of field-aligned currents was suggested by K. Birkeland in 1908, but the idea lay dormant for many years due to lack of evidence, and magnetic perturbations observed at the ground were interpreted in terms of currents flowing purely horizontally. It was not until

Basic principles of the ionosphere

52

field-aligned currents were detected by satellite-borne magnetometers in the early 1970s that the Birkeland current came into fashion and current systems became three-dimensional. Birkeland currents are particularly important in the auroral regions. A fuller treatment of conductivity and the current systems of the solar– geophysical environment is given in several of the standard textbooks.

1.6

Acoustic-gravity waves and traveling ionospheric disturbances

1.6.1

Introduction

The familiar acoustic wave, in which the compression of the gas provides the force restoring a displaced particle towards its original position, is actually the highfrequency limit of a more general class, the acoustic-gravity wave (AGW ). A parcel of air displaced vertically in a stratified atmosphere tends to be restored by buoyancy (due to gravity), and the AGW family results when both gravity and the compressional force are taken into account. We are here concerned mainly with atmospheric waves towards the low-frequency end of the AGW range, whose periods range from a few minutes to an hour or two. They have horizontal wavelengths from several hundred to about a thousand kilometers. Gravity waves in the atmosphere (which should not be confused with cosmological gravity waves, to which they have no connection whatsoever) are transverse waves, the displacement of the gas being normal to the direction of travel of the wave. Their properties, in fact, are complex and in many respects not at all obvious. Several sources of AGWs are known: the motion of the ground during an earthquake, man-made explosions, weather systems, and ionospheric disturbances at high latitude. Table 1.2 shows a classification based on period and wavelength. Waves of small scale come mainly from the troposphere; the medium-scale waves may be tropospheric or ionospheric in origin; and the large-scale events generally have their source in the high-latitude ionosphere – hence their appearance in this opus! Some AGWs are, no doubt, a consequence of events in the solar–terrestrial system: for example, perturbations in the solar wind can produce magnetospheric Table 1.2. A classification scheme for AGWs

Nomenclature

Horizontal trace velocity (m s1)

Period (min)

Wavelength

Large-scale Medium-scale Small-scale

⬃250–1000 90 to ⬃250 300

70 ⬃15–70 ⬃2–5

1000 km Several hundred kilometers —

1.6 Acoustic-gravity waves

53

effects, which couple to the high-latitude ionosphere as particle–precipitation events and electric-field disturbances, which in turn generate medium- and largescale AGWs. We shall meet other examples of solar–geophysical chains of events later in the book. The ionospheric manifestation of AGWs is the traveling ionospheric disturbance (TID), which is due to ion movement communicated from the motion of the neutral air through collisions. There are, however, some complications, the principle one being that, in the F region, the ion motion is constrained along the geomagnetic field. The generation of atmospheric waves at high latitude is discussed in Sections 6.5.6 and 6.5.7.

Theory

1.6.2

Wave motions in the upper atmosphere have been known for over 100 years and TIDs have been noted in ionospheric observations since the 1940s, but not until the 1950s did adequate explanations start to emerge, the key theory being developed by C. O. Hines (Hines, 1960). The underlying concepts of wave propagation are given in Sections 3.2.1 and 3.2.3 in the context of electromagnetic waves. We outline here some of the basic theory governing the properties and behavior of AGWs. In a planar, horizontally stratified, isothermal, single-species, windless, nonrotating atmosphere, the AGW obeys a dispersion relation 4  2s2(kx2 kz2)( 1)g2kx2  22g2/(4s2)0.

(1.81)

where ■

 is the angular frequency of the wave,

■

kx is the horizontal wave number (2/x),

■

x being the wavelength in the horizontal,

■

kz similarly is the vertical wave number,

■

 is the ratio of specific heats (constant pressure/constant volume),

■

s is the speed of sound, and

■

g is the acceleration due to gravity.

This equation states the relation between the frequency and the wavelength (or wave number) in the vertical and the horizontal directions for an AGW. ky does not appear in the equation because there is no asymmetry between the x and y directions. Two significant frequencies are the acoustic cut-off frequency, a  g/(2s)

(1.82)

54

Basic principles of the ionosphere

and the buoyancy or Brunt–Väisala frequency, b ( 1)1/2g/s.

(1.83)

a is the resonance frequency in the acoustic mode of a column of air extending through the whole atmosphere, whereas b is the natural frequency of oscillation of a displaced parcel of air when buoyancy is the restoring force. Substituting these frequencies into Equation (1.81) and rearranging gives

冢

kz2  1 

冣

冢

冣

2a 2 2 kx2 1  b2 . 2 2  s 

(1.84)

Putting 2  b2 gives

冢

kx2 kz2  1 

冣

2a 2 [(2/)2], 2 s 2

(1.85)

where  is the wavelength. There is now no distinction between the x and z coordinates, and this is the acoustic regime. If we go a stage further by putting 2  a2, we get s/(kx2 kz2)  /(2).

(1.86)

This represents a sound wave. In the acoustic regime the phase speed is independent of direction. Putting, now, 2  s2kx2, which removes the effect of compressibility, gives kz2 kx2

冢

冣

2b 1 , 2

(1.87)

which represents a pure gravity wave. Since kx and kz must both be positive in a propagating wave, the frequency  must be either larger than a or smaller than b. These, the acoustic and the gravity regimes, are illustrated in Figure 1.22, which plots the regimes of AGW in terms of the frequency () and the horizontal wave number (kx). Between the acoustic and the gravity regimes the waves are evanescent and do not propagate. The angle of propagation with respect to the horizontal is  tan1(kz /kx).

(1.88)

If 2 is small compared with b2, the ratio kz /kx is large and then the wave propagates almost vertically. This is for the propagation of phase. The energy, on the other hand, travels at the group velocity, given (Equation (3.21)) by u(dk/d)1,

1.6 Acoustic-gravity waves

Figure 1.22. The acoustic, evanescent, and gravity regimes of acoustic-gravity waves. The dashed lines show the effects of neglecting gravity and compressibility, repectively. At ionospheric levels, waves with periods longer than 10–15 min are likely to be gravity waves, and any with periods of only a few minutes are probably acoustic. (After J. C. Gille, in Winds and Waves in Stratosphere, Mesosphere and Ionosphere (ed. Rawer). North-Holland, 1968. Elsevier Science Publishers.)

and, in a gravity wave, the energy flow is at right angles to the direction of phase propagation. Figure 1.23 illustrates the relations amongst particle displacement, phase propagation, and group propagation in a gravity wave. Note that, if the source is below, the energy flows upward (as it must) but the phase propagation is downward. Furthermore, the amplitude of the air displacement increases with altitude so that the energy flux may be constant (provided that there are no losses). Figure 1.24 shows how the horizontal component of the group velocity varies with the wave period (normalized by the Brunt frequency as b/) at fixed values of the ascent angle of the energy (i.e. the angle between the group velocity and the horizontal). The energy flow approaches horizontal when the wavelength is very large. A distinction between the sections of the curves labeled “buoyancy” and “gravity” needs to be made when one is considering AGW propagation over large distances (Francis, 1975). For an AGW, the refractive index () is defined as the ratio between the speed of sound and the phase velocity of the wave. (Compare with Section 3.2.3.) Then

55

Basic principles of the ionosphere

56

( )

1 ∂ NORMALIZED HORIZONTAL GROUP VELOCITY — —– C ∂kx kz

Figure 1.23. A simple gravity wave, showing the essential relations amongst phase propagation, air displacement, and energy flow.

1 —–b a

Φ = 0° Φ = 0°

Φ = 20°

Φ = 5°

Φ = 40° GRAVITY WAVES

0.5 Φ = 60°

Φ = 10°

BUOYANCY WAVES ACOUSTIC WAVES

Φ = 20° Φ = 40°

Φ = 80° 0 0.2

0.5

1

2

5

10

a —– b

NORMALIZED PERIOD (

b/

)

Figure 1.24. Contours of constant , the ascent angle of the group velocity from the horizontal, against the wave period and the horizontal component of the group velocity. The details of the diagram depend on the values assumed for the acoustic (a) and Brunt (b) frequencies. (Reprinted from S. H. Francis, J. Atmos. Terr. Phys. 37, 1011, Copyright 1975, with permission from Elsevier Science.)

1.6 Acoustic-gravity waves

冢

2  1 

2a 2

冣 冤冢

/

1

57

冣

冥

2b cos2  . 2

1.89

If   a,b,  →1, and if   a,b, →

a 1 . b cos 

In general the particle motion is elliptical in an AGW, combining the longitudinal displacement of an acoustic wave with the transverse displacement of a gravity wave. There are alternate compressions and rarefactions at successive zerodisplacement points in Figure 1.23. At extremely low frequency, the air motion and the group velocity would be horizontal, the phase propagation vertical, and the compression and rarefaction zero. Complexities neglected by the simple theory, but which affect AGWs in real life, are energy loss through the viscosity of the air, non-linear effects if the amplitude becomes too large at the higher altitudes, reflection and ducting due to the change of atmospheric properties with altitude, the curvature of the Earth’s surface, and winds. 1.6.3

Traveling ionospheric disturbances

The mechanism by which AGWs produce ionospheric disturbances (TID) is collisional coupling between neutral and ionized particles. This force acts in the direction of motion of the neutral air, but, in the ionospheric F region, the effect is strongly modified by the geomagnetic field which permits ion motion along the field only. Thus, while there are several radio techniques able to measure properties of a TID, to interpret these data as properties of the AGW causing it may be less than straightforward. This, however, hardly matters if propagation effects are the principal concern. Figure 1.25 is an elegant example of a TID observation by ionosonde (Section 4.2.1). It shows the period of the wave and its wavelength, the latter derived using the velocity estimated from spaced observations. The downward phase propagation is clearly seen. 1.6.4

The literature

The literature of published research on the topics of AGW and TID is very large. Surveys of the earlier work have been published by Yeh and Liu (1974) and Francis (1975). Studies performed from the mid-1970s up to 1981 have been reviewed by Hunsucker (1982), and those between 1982 and 1995 by Hocke and Schlegel (1996). Work since then is addressed by Kirchengast (1996), Bristow and Greenwald (1997), Balthazor and Moffett (1997, 1999), Huang et al. (1998) and Hall et al. (1999).

Basic principles of the ionosphere

58

Figure 1.25. A train of gravity waves observed by ionosonde over Missouri, USA, in December 1966, identified from the virtual heights of echoes at frequencies between 1.6 and 3.6 MHz. (T. M. Georges, Ionospheric Effects of Atmospheric Waves. Institutes for Environmental Research, report IER 57-ITSA 54, 1967, Boulder, Colorado, USA.)

1.7

References and bibliography

1.2

The Vertical structure of the atmosphere

Hargreaves, J. K. (1992) The Solar–Terrestrial Environment. Cambridge University Press, Cambridge. Richmond, A. D. (1983) Thermospheric dynamics and electrodynamics. Solar–Terrestrial Physics (eds. R. L. Carovillano and J. M. Forbes), p. 523. Reidel, Dordrecht.

1.3

Physical aeronomy

VanZandt, T. E. and Knecht, R. W. (1964) The structure and physics of the upper atmosphere. Space Physics (eds. D. P. LeGalley and A. Rosen), p. 166. Wiley, New York.

1.4

The main ionospheric layers

Bracewell, R. N. and Bain, W. C. (1952) An explanation of radio propagation at 16 kc/sec in terms of two layers below E layer. J. Atmos. Terr. Phys. 2, 216. Friedrich, M. and Torkar, K. M. (1992) An empirical model of the nonauroral D region. Radio Sci. 27, 945. Hargreaves, J. K. and Bagenal, F. (1977) The behavior of the electron content during ionospheric storms: a new method of presentation and comments on the positive phase. J. Geophys. Res. 82, 731.

1.7 References and bibliography

Piggott, W. R. and Rawer, K. (1972) URSI Handbook of Ionogram Interpretation and Reduction, Chapter 4. Report UAG-23A, World Data Center A, NOAA, Boulder, Colorado. Rishbeth, H. (1991) F-region storms and thermospheric dynamics. J. Geomag. Geoelectr. 43 suppl., 513. VanZandt, T. E. and Knecht, R. W. (1964) The structure and physics of the upper atmosphere. Space Physics (eds. D. P. LeGalley and A. Rosen), p. 166. Wiley, New York. Whitehead, J. D. (1970) Production and prediction of sporadic E. Rev. Geophys. Space Phys. 8, 65.

1.5

The electrical conductivity of the ionosphere

Akasofu, S.-I. and Chapman, S. (after K. Maeda and H. Matsumoto) (1972) Solar–Terrestrial Physics, Oxford University Press, Oxford. Kelley, M. (1989) The Earth’s Ionosphere. Academic Press, New York. Rishbeth, H. (1988) Basic physics of the ionosphere – a tutorial review. J. Inst. Electronic Radio Engineers 58, 207.

1.6

Acoustic-gravity waves and traveling ionospheric disturbances

Balthazor, R. L. and Moffett, R. J. (1997) A study of atmospheric gravity waves and travelling ionospheric disturbances at equatorial latitudes. Ann. Geophysicae 15, 1048. Balthazor, R. L. and Moffett, R. J. (1999) Morphology of large-scale traveling atmospheric disturbances in the polar thermosphere. J. Geophys. Res. 104, 15. Bristow, W. A. and Greenwald, R. A. (1997) On the spectrum of thermospheric gravity waves observed by the Super Dual Auroral Radar Network. J. Geophys. Res. 102, 11585. Francis, S. H. (1975) Global propagation of atmospheric gravity waves: a review. J. Atmos. Terr. Phys. 37, 1011. Gille, J. C. (1968) The general nature of acoustic-gravity waves. Winds and Turbulence in Stratosphere, Mesosphere and Ionosphere (ed. Rawer). Elsevier Science Publishers, Amsterdam. Hall, G. E., MacDougall, J. W., Cecile, J.-F., Moorcroft, D. R. and St.-Maurice, J. P. (1999) Finding gravity wave positions using the Super Dual Auroral Radar network. J. Geophys. Res. 104, 67. Hines, C.O. (1960) Internal atmospheric gravity waves at ionospheric heights. Can. J. Phys. 38, 1441. Hocke, K. and Schlegel, K. (1996) A review of atmospheric gravity waves and travelling ionospheric disturbances: 1982–1995. Ann. Geophysicae 14, 917. Huang, C.-S., Andre, D. A. and Sofko, G. (1998) High-latitude ionospheric perturbations and gravity waves: 1. Observational results. J. Geophys. Res. 103, 2131. Hunsucker, R. D. (1982) Atmospheric gravity waves and traveling ionospheric disturbances. Encyclopedia of Earth System Science, p. 217. Academic Press, New York. Kirchengast, G. (1996) Elucidation of the physics of the gravity wave–TID relationship with the aid of theoretical simulations. J. Geophys. Res. 101, 13 353.

59

Basic principles of the ionosphere

60

Yeh, K-C. and Liu, C-H. (1974) Acoustic-gravity waves in the upper atmosphere. Rev. Geophys. Space Phys. 12, 193.

General reading on the topics of Chapter 1 Books Akasofu, S.-I. and Chapman, S. (1972) Solar–Terrestrial Physics. Oxford University Press, Oxford. Banks, P. M. and Kockarts, G. (1973) Aeronomy. Academic Press, New York. Bauer, S. J. (1973) Physics of Planetary Atmospheres. Springer-Verlag, Berlin. Brasseur, G. and Solomon, S. (1984) Aeronomy of the Middle Atmosphere. Reidel, Dordrecht. Carovillano, R. L. and Forbes, J. M. (eds.) (1983) Solar–Terrestrial Physics. Reidel, Dordrecht. Dieminger, W., Hartmann, G. K. and Leitinger, R. (eds.) (1996) The Upper Atmosphere – Data Analysis and Interpretion. Springer-Verlag, Berlin. Hess, W. N. and Mead, G. D. (eds.) (1968) Introduction to Space Science. Gordon and Breach, New York. Jursa, A. S. (ed.) (1985) Handbook of Geophysics and the Space Environment. Air Force Geophysics Laboratory, US Air Force, National Technical Information Service, Springfield, Virginia. Kato, S. (1980) Dynamics of the Upper Atmosphere. Center for Academic Publication Japan, Tokyo. Matsushita, S. and Campbell, W. H. (eds.) (1967) Physics of Geomagnetic Phenomena. Academic Press, New York. Ratcliffe, J. A. (ed.) (1960) Physics of the Upper Atmosphere. Academic Press, New York. Rawer, K. (1956) The Ionosphere. Frederick Ungar Publishing Co., New York. Rees, H. M. (1989) Physics and Chemistry of the Upper Atmosphere. Cambridge University Press, Cambridge. Rishbeth, H. and Garriott, O. K. (1969) Introduction to Ionospheric Physics. Academic Press, New York. VanZandt, T. E. and Knecht, R. W. (1964) The structure and physics of the upper atmosphere. In Space Physics (eds. D. P. Le Galley and A. Rosen). Wiley, New York. Whitten, R. C. and Poppoff, I. G. (1965) Physics of the Lower Ionosphere. PrenticeHall, Englewood Cliffs, New Jersey. Whitten, R. C. and Poppoff, I. G. (1971) Fundamentals of Aeromony. Wiley, New York.

Conference reports McCormac, B. M. (ed.) (1973) Physics and Chemistry of Upper Atmosphere. Reidel, Dordrecht. McCormac, B. M. (ed.) (1975) Atmospheres of Earth and Planets. Reidel, Dordrecht.

Chapter 2 Geophysical phenomena influencing the high-latitude ionosphere

2.1

Introduction

Whereas the mid-latitude ionosphere is dominated by solar radiation and the chemistry of the upper atmosphere, modified by dynamic effects, the high-latitude ionosphere is, in addition, strongly affected by the nature of the geophysical environment and by various processes occurring within it. In particular, the form of the geomagnetic field connects the polar upper atmosphere to the magnetosphere. Thereby the polar ionosphere becomes accessible to particles that have been energized within the magnetosphere or have come from the Sun; these provide another source of ionization. It is also affected by the dynamics of the magnetosphere and is thus subject to electric fields and currents generated by motions at high levels. At the highest latitudes the ionosphere is connected, via the field-lines, to the outer magnetosphere, giving it a ready response to variations in the flow of the solar wind. The present chapter therefore summarizes the basic properties and behavior of the magnetosphere, which we must appreciate in order to understand the behavior of the ionosphere poleward of 60° latitude.

2.2

The magnetosphere

2.2.1

The geomagnetic field

To a first approximation the geomagnetic field at and close to the planet’s surface can be represented as a dipole field. The poles of the dipole are at geographic

61

62

Geophysical phenomena

Figure 2.1. Dipolar field-lines. (D. L. Carpenter and R. L. Smith, Rev. Geophys. 2, 415, 1964, copyright by the American Geophysical Union.)

latitudes and longitudes 79° N, 70° W, and 79° S, 70° E. The magnetic flux density is given by M B(r, ) 3 (1 3sin2 )1/2, r

(2.1)

where M is the dipole moment, r the geocentric radial distance, and  the magnetic latitude. This is accurate to within about 30% at points within two or three Earth-radii of the surface. Although it is not very accurate, the dipole form is useful for making approximate calculations. Figure 2.1 shows the lines of force, generally called field-lines, in a dipole field. Each line is the locus of the force on a single north pole and is represented by a simple equation, rr0 cos2 .

(2.2)

If  0, rr0; r0 is thus the radial distance to the field-line in the plane normal to the axis of the dipole. There is a different value of r0 for each line of Figure 2.1, but, since the field is three-dimensional, each r0 actually describes a shell. The other coordinate is provided by magnetic longitude. In the magnetosphere it is convenient to use the radius of the Earth, RE, as the unit of distance. Then, putting r/RE R, B(R, )

0.31 (1 3sin2 )1/2 G. R3

(2.3)

0.31 G (3.1 105 Wb m2) is the flux density at the magnetic equator on the Earth’s surface. In these terms, the field-line equation becomes RR0 cos2 ,

(2.4)

2.2 The magnetosphere

63

where both R and R0 are measured in Earth-radii. The latitude where the field-line intersects the Earth’s surface is given by cos E R1/2 . 0

(2.5)

The dipole form is convenient for its mathematical simplicity, but for many purposes it is not sufficently accurate. A closer approximation is the displaced-dipole model, in which the dipole is displaced by 400 km from the center of the Earth. However, for accurate work (not too far above the surface) it is usual to derive the field from the magnetic potential expressed as a series of spherical harmonics – the flux density being the gradient of the magnetic potential. The dipole form corresponds to the first term of the expansion. The coefficients are derived by fitting the expression to measurements of the magnetic elements on the global scale, using magnetometers both on the ground and on satellites. Because the geomagnetic field changes with time – the secular variation – a fresh set of coefficients, relating to a specific epoch, is published from time to time. Such representations are accurate to within about 0.5% at and near the surface. The terms of higher order become less important at greater distances and the field tends to become more dipolar. However, beyond three or four Earthradii the distortion due to the solar wind has to be taken increasingly into account. The pressure of the solar wind confines the geomagnetic field on the sunward side and forms the geomagnetic cavity. 2.2.2

The solar wind

The solar wind was first observed directly by space probes in the early 1960s, though its existence had previously been proposed in theoretical work and some of its properties had been deduced from studies of comets. There have been many observations of the solar wind since that time. It is basically an outflow driven by the continual expansion of the solar corona and it is therefore composed of solar material. Most of the ions are protons (H) but there is also an -particle (He2) component typically amounting to 5% though exceptionally up to 20%. Still heavier atoms amount to perhaps 0.5% in total, though, in contrast to the light ions, these are not fully ionized. The concentration of positive ions varies between 3 and 10 cm3 (3 106 to 107 m3), the most typical value being 5 cm3, and there is a similar number of electrons for bulk neutrality. The mean mass of solar-wind particles is therefore about half that of the proton, about 1027 kg. There are fluctuations as large as by a factor of ten over times of minutes and hours, implying irregularities within the solar wind over distances of 105 km and more. At the distance of the Earth’s orbit the speed of the solar wind is usually between 200 and 700 or 800 km s1 (Figure 2.2), on which is superimposed a random component of temperature 105 K. The solar wind is not very hot by solar

64

Geophysical phenomena

(a)

(b)

(c)

Figure 2.2. (a) The speed of the solar wind: a histogram of measurements between 1962 and 1970. (J. T. Gosling, in Solar Activity Observations and Predictions (eds. McIntosh and Dryer), by kind permission of The MIT Press, 1972.) In (b) (c) are shown the distributions of the magnitude and components of the interplanetary magnetic field, 1988–1990. By is east–west and Bz is north–south. (F. J. Rich and M. Hairston. J. Geophys. Res. 99, 3827, 1994, copyright by the American Geophysical Union.)

2.2 The magnetosphere

standards, the energy being more directed than random. It carries an energy flux of about 104 W m2, which is approximately a tenth of that in the EUV region of the solar spectrum. The solar wind is the principal medium by which the activity of the Sun is communicated to the vicinity of the Earth, and it is extremely important in solar–terrestrial relations and in the behavior of the high-latitude ionosphere. The interaction depends on a weak magnetic field, the interplanetary magnetic field (IMF), which is carried along by the plasma. This field amounts only to a few nanoteslas (a few ) and it is “frozen in” to the plasma because of the large electrical conductivity. The magnitude of the IMF varies slightly with the sunspot cycle (Figure 1.17). The kinetic energy of the solar-wind particles exceeds the energy density of the magnetic field by a factor of about eight, and therefore the motion of the total magnetoplasma is governed by the motion of the particles rather than by the magnetic field. Although the solar wind flows out almost radially from the Sun, the solar rotation gives the magnetic field a spiral form, as in Figure 2.3. This is sometimes known as the garden-hose effect since it may be simulated by turning round while watering the garden and noting that the jet of water follows a spiral path although the trajectories of individual drops are radial. It so happens that, at the orbit of Earth, the IMF field-lines run at about 45° to the radial direction: the radial and the east–west components of the IMF are therefore about equal in magnitude. One of the most remarkable of the early results, and a fact of great significance, is that distinct sectors may be recognized within the solar wind, the field being inward and outward in alternate sectors. Figure 2.3 shows some of the original measurements, in which four sectors – two inward and two outward – were present. However, this is not always the case because the sector structure evolves with time. Sometimes there are only two sectors, and sometimes the sectors are not all of the same width. The proton density can vary by more than a factor of ten and the speed of the solar wind by a factor of two during one solar rotation as the sectors go by, with a degree of anticorrelation. At first sight the form of the IMF appears anomalous. Although there may be a north–south component, it is equally likely to be northward or southward; thus it seems that a spiral in the ecliptic plane is indeed the basic form of the IMF. How is this to be reconciled with an origin in the solar magnetic field which we expect to be essentially dipolar? The problem is that the early observations were confined to the ecliptic plane and there is still not much knowledge of its form at higher solar latitudes. It is now thought that there is a current sheet in or near the equatorial plane that effectively divides the outward field (above the plane) from the inward field (below it) as in Figure 2.4. If the solar magnetic dipole is tilted from the rotation axis, the current sheet will be tilted from the ecliptic plane and a spacecraft near the Earth will observe a two-sector structure as the Sun rotates. When more than two sectors are seen, it is thought that the current sheet has developed

65

66

Geophysical phenomena

Figure 2.3. (a) The form of the interplanetary magnetic field (IMF) in the solar equatorial plane, corresponding to a solar-wind speed of 300 km s1. (T. E. Holzer, Solar System Plasma Physics, Vol. I, North-Holland, 1979, p. 103, Elsevier Science Publishers.) (b) The sector structure of the solar wind in late 1963, showing inward () and outward () IMF. (J. M. Wilcox and N. F. Ness, J. Geophys. Res. 70, 5793, 1965, copyright by the American Geophysical Union.)

2.2 The magnetosphere

67

M

Ω

Figure 2.4. The ballerina model of the current sheet in the solar wind. M is the axis of the current sheet and  is the Sun’s rotation axis. (E. J. Smith et al., J. Geophys. Res., 83, 717, 1978, copyright by the American Geophysical Union.)

undulations as in the skirt of a pirouetting ballerina; hence the concept of Figure 2.4 is often known as the ballerina model. Spacecraft venturing out of the ecliptic plane have observed that the sector structure disappears – which is consistent with the ballerina model. A link between the solar wind and a particular feature of the corona was discovered by the Skylab missions between May 1973 and February 1974. A so-called coronal hole emits less light at all wavelengths than do adjoining regions, but it is most marked in an X-ray photograph, on which it appears as a black area. Coronal holes are regions with abnormally low density where the magnetic field has a single polarity – all inward or all outward. This is an open magnetic field that goes out into interplanetary space rather than returning to the Sun. The hole is the source of fast solar-wind streams in which the speed exceeds 700 km s1. The speed is greater from a larger hole. Less than 20% of the solar surface is composed of coronal holes, and they are more numerous during the declining phase of the sunspot cycle. The fast streams interact with the slower solar wind as in Figure 2.5(a), compressing the magnetic field and the plasma ahead and sometimes, though not always, creating a shock front. The compressed plasma is heated, and a rarefaction follows. Within the stream the magnetic field maintains the same polarity (inward or outward) and is the same as in the corresponding coronal hole. The fast streams from coronal holes co-rotate with the Sun and can persist for several rotations. They are the probable cause of recurring geomagnetic storms (Section 2.5.4). Intermittant perturbations of the solar wind can be caused by specific solar events, particularly the coronal mass ejection (CME). This is not the same as a

Figure 2.5. (a) Interaction between a localized stream of high-speed plasma and the slower, ambient solar wind. (T. E. Holzer, in Solar System Plasma Physics, Vol I, North-Holland, 1979, p. 103, Elsevier Science Publishers.) (b) High-speed plasma from a solar flare driving an interplanetary shock. The ejected plasma contains an ordered magnetic field, but between the shock and the ejecta the field is turbulent. (Reprinted from L. F. Burlaga, Adv. Space Res. 2, 51, copyright 1982, with permission from Elsevier Sceince.)

2.2 The magnetosphere

69

solar flare, though in some instances a flare occurs at about the same time or shortly afterwards. The CME travels away from the Sun at a speed that may be less than 50 km s1 or greater than 1200 km s1, and the speedier examples produce a shock front in the solar wind. The typical structure of such a disturbance is illustrated by Figure 2.5(b), (except that “flare” should be replaced by “CME”). The IMF is compressed by the shock, and a turbulent region is formed between the shock and the ejected matter. Within the CME the magnetic field is strong and well ordered, possibly as a closed loop. These magnetic structures, sometimes called magnetic clouds, are about 0.25 AU across at the orbit of Earth. The form of the cavity formed by the interaction of the solar wind and the geomagnetic field is illustrated in Figure 2.6. Because it has very high electrical conductivity, the solar wind is not able to penetrate the geomagnetic field but is swept around it. Pressure is exerted against the magnetic field, which is distorted and confined within a large but nevertheless limited region around the Earth. This kind of behavior was foreseen by Chapman and Ferraro as long ago as 1930 in their pioneering study of the cause of magnetic storms (Section 2.5.2). (In modern terms the solar wind is said to be “frozen out” of the geomagnetic field.) The magnetosphere has a complex structure. In the rest of this section we will describe some of its main features: the magnetopause, the magnetosheath and the shock, the polar cusps, and the magnetotail. To begin with they will be treated as though they were essentially static. Dynamic aspects will be introduced in Section 2.4. 2.2.3

The magnetopause

To a first approximation the form of the boundary between the geomagnetic field and the incident solar wind can be deduced by considering the pressure balance across the boundary. We assume that, when the system is in equilibrium, the pressure of the solar wind outside is at every point of the surface equal to that of the magnetic field inside. If the solar wind contains N particles m3, each of mass m kg, traveling at velocity v m s1 and striking the surface at angle  from the normal, then it can be shown that the total rate of change of momentum due to the flux of solar wind particles is 2Nmv2 cos2  N m2. This has to be equated to the magnetic pressure B2/(20). All species within the solar wind contribute but the protons have greatest effect. A simple calculation along these lines readily gives a realistic distance for the position of the boundary (approximately 10RE) along the Earth–Sun line, and allows one to estimate how it varies if the solar wind changes. We assume that  0, and the magnetic flux density B varies as (distance)3. Then the distance to the magnetopause is lm 

冢

B2E 40Nmv2

冣

1/6

,

(2.6)

70

Geophysical phenomena

(a)

(b)

Figure 2.6. Two sketches of the geomagnetic cavity in north–south cross-section. (a) Showing the external flow of solar-wind plasma and the principal features of the distorted geomagnetic field. Note the tapering of the inner field-lines, suggesting that there is a neutral line further down the tail. (Adapted from V. M. Vasyliunas, in Solar–Terrestrial Physics, Reidel, 1983, p. 243, with kind permission from Kluwer Academic Publishers.) (b) Showing the main plasma regions in relation to the magnetic structure. Note the northward displacement of the plasma sheet during summer in the northern hemisphere. (W. J. Raitt and R. W. Schunk, in Energetic Ion Composition in the Earth’s Magnetosphere (ed. R. G. Johnson), Terra Scientific Publishing Co., Tokyo, 1983, p. 99. After Bahnsen 1978).)

2.2 The magnetosphere

where BE is the geomagnetic flux density at the Earth’s surface at the magnetic equator. A full computation is more complicated since the orientation of the boundary at each point is not known at the outset and an iteration proceedure is required. There is also a degree of coupling between the interplanetary and geomagnetic fields, which affects the location of the boundary. Spacecraft find the magnetopause about 0.5RE closer to the Earth when the IMF has a southward component rather than a northward one. According to Petrinec and Russell (1993) the boundary moves closer by one RE for every 7.4 nT of southward IMF component, but the distance is not much affected by varying amounts of northward component. Another approximate method, which is fairly successful over the sunward side of the magnetopause only, is the image-dipole method, which replaces the dynamic pressure of the solar wind by the magnetic pressure of an image dipole of moment MI placed parallel to and at distance d from the Earth’s dipole, ME. The fields due to these two dipoles are added and the distorted field-lines associated with ME – the two fields do not interconnect – are taken to represent the geomagnetic field within the magnetopause. Those associated with the image have no physical significance. A satisfactory model is given by MI 28ME; d40RE. This method is not valid down the sides of the magnetosphere and in the anti-sunward direction. The resulting boundary, the magnetopause, is indicated in Figure 2.6. The geomagnetic field is severely distorted within the magnetosphere. Note in particular the following points. (a). Field-lines originating at low latitude form closed loops between northern and southern hemispheres, though there can be some distortion from the dipole form. (b). Lines emerging from the polar regions are swept back, away from the Sun; in a dipole field some of these would have connected on the day side. (c). Intermediate between these regions are two lines, one in each hemisphere, that go out and meet the magnetopause on the day side, though in fact their flux density falls to zero as they reach it; here, neutral points are formed. The magnetopause has a finite thickness, though it is thin (approximately 1 km thick) in comparison with the size of the magnetosphere. Figure 2.7 (in Section 2.3.1) gives some idea of the form of the magnetosphere in three dimensions. 2.2.4

The magnetosheath and the shock

A shock front is formed in the solar wind two or three RE upstream of the magnetopause (Figure 2.6). The region between the shock front and the magnetopause is the magnetosheath, and here the plasma, composed mainly of solar material but in other respects not typical of the solar wind, is turbulent. Tenuous though it may be by any ordinary standard, the magnetosphere is a

71

Geophysical phenomena

72

relatively solid object in comparison with the solar plasma. Furthermore, the solar wind is “supersonic” at the orbit of Earth, meaning that its velocity exceeds that of any waves that can propagate within it. In the solar wind the speed of hydromagnetic waves, that is, the Alfvén speed, given by vA 

B , ( 0
) 1/2

(2.7)

where B is the magnetic flux density and
the particle density (in kg m3), is about 50 km s1. For a solar wind speed of 400 km s1, therefore, the Alfvén Mach number is 8. We therefore have the conditions for a shock front, a discontinuity created when information about an approaching obstruction is not transmitted ahead into the medium. The existence and location of the shock were predicted from theory in the early 1960s and subsequently verified by observation. On crossing the shock, solarwind plasma is slowed down to about 250 km s1 and the corresponding loss of directed kinetic energy is dissipated as thermal energy, increasing the temperature to 5 106 K. Magnetosheath plasma is therefore slower than the solar wind proper but 5–10 times hotter. 2.2.5

The polar cusps

The simple models of the magnetosphere predict two neutral points on the magnetopause where the total field is zero. These points connect along field-lines to places on the Earth’s surface near 78° magnetic latitude. These are in fact the only points on the Earth’s surface which connect directly to the magnetopause, and all the field from the magnetopause converges into those two points. They are therefore regions of great interest where solar-wind particles (from the magnetosheath) can enter the magnetosphere without having to cross field-lines. Measurements reveal regions that are more extended than points, and they are now called the polar cusps or clefts. Particles with energies typical of those in the sheath are observed over some 5° of latitude around 77°, and over 8 h of local time around noon. The cusps are funnels of weak magnetic field filled with magnetosheath plasma, and they have significant effects on the high-latitude ionosphere. The ionospheric effects of particle entry provide “signatures” of the cusp, indicating its location – see Section 5.2.2 and Figure 5.7. 2.2.6

The magnetotail

In the anti-sunward direction the magnetosphere is extended into a long tail, the magnetotail. As is shown by spacecraft magnetometers, the geomagnetic field beyond about 10RE on the night side of the Earth tends to run in the Sun–Earth direction, and there is a central plane within which the field reverses direction. This is the neutral sheet. The field points towards the Earth in the northern lobe,

2.3 Particles in the magnetosphere

and away from the Earth in the southern. The tail is roughly circular, some 30RE (2 105 km) across, and of uncertain length, though it has been detected downwind beyond 107 km. Its significance for the high-latitude ionosphere is that it maps into the polar caps at its earthward end, and thus the polar ionosphere can be affected directly by events in the tail. The basic form of the magnetotail in the plane containing the magnetic poles is shown in Figure 2.6. The flux density is about 20 (20 nT) in the tail lobes, but the field is much weaker in the neutral sheet where the reversal occurs. In this region the magnetic pressure of the tail lobes (BT2 /(20) is more or less balanced by an enhancement of the plasma density, the plasma sheet (to be considered further in Section 2.3.3). However, in fact the tail, like the whole magnetosphere, is dynamic and it forms an essential part of the magnetospheric circulation, to be considered in Section 2.4.

2.3

Particles in the magnetosphere

2.3.1

Principal particle populations

The geomagnetic field holds within it several distinct populations of charged particles. (a). Deep within the magnetosphere (in the region often known as the inner magnetosphere) is the plasmasphere, closely linked to the mid-latitude ionosphere and comprising electrons, protons, and some heavy ions, all having energies in the thermal range. (b). Also trapped on closed field-lines are the energetic particles generally known as the Van Allen particles after their discoverer. Apart from cosmic rays and solar protons, which are merely passing through, the Van Allen particles are the most energetic particles in the magnetosphere and they make some contribution to the ionization of the upper atmosphere when they are precipitated out of the trapping region. (c). The plasma sheet is associated with the magnetotail, essentially with the central region where the magnetic field reverses direction. Plasma-sheet particles are energized within the magnetotail and they are important in auroral activity and the behavior of the high-latitude ionosphere. Their energy is intermediate between those of the plasmasphere and the Van Allen belt. The inner edge of the plasma sheet supports the ring current that flows in the magnetosphere during magnetic storms. (d). At the edges of the magnetosphere, and obviously connected with the physics of the magnetopause, are boundary layers. Their composition and energy are governed by the solar wind and plasma in the magnetosheath.

73

Geophysical phenomena

74

Figure 2.7. Plasma populations and current systems of the magnetosphere in three dimensions. (T. A. Potemra, Johns Hopkins APL Tech. Digest 4, 276, 1983. © The Johns Hopkins University Applied Physics Laboratory, 1983. All rights reserved. Reproduced by permission.)

The locations of these particle populations are indicated in Figure 2.7. They are not merely incidental to the magnetosphere, but are in fact essential to its properties and behavior. Except for the boundary layers, each of these populations is discussed in a following section. In most of the magnetosphere the ratio of the energy density of the particles to that of the magnetic field () is less than unity, but there are exceptions. Originally it was thought that most of the particles in the magnetosphere come from the solar wind, but, on the evidence of heavy ions observed in the magnetosphere, it is now recognized that there are also major sources in the ionosphere: specifically the auroral zones, the clefts, and the polar caps. It is thought that the solar wind is the dominant source in the distant magnetotail, but the ionospheric sources are important during storms and are sometimes dominant. (See also the discussion of the polar wind, Section 5.2.3.) 2.3.2

The plasmasphere

Ionized particles in the upper ionosphere (F region and topside) have temperatures up to several thousand degrees Kelvin, and electron energies are therefore several tenths of an electron volt. The particle density is typically 1010 m3 at 1000 km altitude, decreasing with increasing height – though not very rapidly because of the large scale height when atomic hydrogen is the principal atom. The theory of the protonosphere (Section 1.4.4) shows how ionospheric plasma flows up the field-lines to populate the protonosphere as far as the equatorial plane, provided

2.3 Particles in the magnetosphere

Figure 2.8. Electron density in the equatorial plane determined from whistlers. (J. A. Ratcliffe (after D. L. Carpenter), An Introduction to the Ionosphere and Magnetosphere. Cambridge University Press, 1972.)

that the field-lines are closed. Some of this plasma flows back to lower levels at night, where it helps to maintain the ionosphere during hours of darkness, but the plasmasphere nevertheless persists as a permanent feature of the inner magnetosphere. The outer boundary of the plasmasphere is called the plasmapause. The plasmapause was discovered by a ground-based technique based on the reception of VLF whistlers. The whistler is a naturally occurring radio signal that propagates along the geomagnetic field between the northern and southern hemispheres. If the travel time of a whistler is displayed against frequency, it is seen that there is one frequency at which the travel time is a minimum. This is a characteristic of all whistlers. Not all show it clearly but those which do are called nose whistlers. The frequency corresponding to the minimum travel time indicates which field-line the whistler has traveled along, and the time taken can be interpreted to give the minimum electron density encountered along that field-line. A detailed discussion of whistlers and other magnetospheric noises would be beyond our scope; the reader who wishes to persue that interesting topic is referred to the book by Helliwell (1976). By means of this technique it is possible to determine the variation of electron density in the equatorial plane, as in Figure 2.8. The remarkable feature of such plots is that they often exhibit a sudden drop in the electron density near 4RE. The decrease may be of a factor of ten or more within a distance of 0.5RE or less. This edge is the plasmapause, sometimes also known as the knee. If it is traced inward along the geomagnetic field, it is found to correspond approximately to the ionospheric main trough which effectively marks the poleward extent of the midlatitude ionosphere (Section 5.4). The plasmasphere thus occupies a doughnutshaped region of the inner magnetosphere where the field-lines are not too

75

76

Geophysical phenomena

Figure 2.9. Plasma flow in the equatorial plane and the daily variation of the plasmapause. (After J. L. Burch, in The Upper atmosphere and Magnetosphere. National Academy of Sciences, Washington DC, 1977.)

distorted from the dipolar form, and the mid-latitude ionosphere is part of it. The plasmapause is dynamic and variable. Its position varies during the day, the most notable feature being a bulge in the evening sector (Figure 2.9). In addition, the whole region contracts when geomagnetic activity increases, and there is then a gradual recovery over the next few days. Figure 2.10 shows measurements of the plasmapause position as a function of the global magnetic activity index Kp (described in Section 2.5.4). For most of the time it occurs between three and six RE, though it has been detected as close to the Earth as 2RE (i.e. only one RE above the surface), and satellite data show occasions when no plasmapause was detected within seven or eight RE. According to whistler observations and in situ data (Carpenter and Anderson, 1992), the average geocentric distance to the plasmapause in Earth-radii (Lpp) is related to the greatest preceeding value of Kp (Kˆ p) by the empirical relation Lpp 5.60.46Kˆ p.

(2.8)

For this purpose Kˆ p is derived over the previous 24 h, except that, for the hours 06–09, 09–12, and 12–15, respectively, the Kp values for the preceeding one, two, and three periods (of 3 h) are omitted. Carpenter and Anderson also give formulae for

2.3 Particles in the magnetosphere

Figure 2.10. Variations of the plasmapause with the magnetic activity index, Kp. (a) Satellite observations of ion density, showing the plasmapause at several levels of Kp. (b) The relation between the plasmapause distance, Lpp, and Kp. (After C. R. Chappell et al., J. Geophys. Res. 75, 50, 1970, copyright by the American Geophysical Union.)

77

Geophysical phenomena

78

the electron density in the plasmasphere inside the pause and its regular variations, the thickness of the pause, and the distribution in the “trough” beyond it. 2.3.3

The plasma sheet

Beyond the plasmapause the electron density is much smaller and the temperature is much higher. Clearly this is a different population of particles. The electron and ion densities are each only about 0.5 cm3. Particle energies are generally 102–104 eV. The average energy of the electrons is about 0.6 keV and that of the protons about 5 keV. The total energy density of the particles is about 3 keV cm3. The particular importance of the plasma sheet lies in its association with the central plane of the magnetotail where the magnetic field reverses. To a first approximation, the pressure of the particles in the sheet balances the magnetic pressure in the tail lobes. Thus, nkTBT2 /(20),

(2.9)

where BT is the tail magnetic field outside the plasma sheet. As indicated in Figure 2.6, the plasma sheet follows the magnetic field down to lower altitudes in the vicinity of the auroral zone. It also continues round to the day side of the Earth. In the equatorial plane there is an identifiable, though variable, inner edge near 7RE at midnight. The sheet is several Earth-radii thick (also variable) and it extends across the tail between the dusk and dawn sides. As the Earth’s magnetic axis tilts seasonally and diurnally with respect to the Sun, the tail plasma sheet and neutral sheet oscillate north and south of the solar-ecliptic plane. The magnetic field runs in opposite directions in the two lobes of the magnetotail, and the existence of a sheet of plasma between them creates an unusual physical situation. The configuration being far from dipolar, it represents a store of energy that could be released in the right circumstances. There is good evidence that a neutral line forms some 50RE down the tail. Here the magnetic field is locally collapsed and plasma is accelerated both towards and away from the Earth. It is also known that events in the magnetotail are closely related to the phenomena of the aurora and the substorm, and it is thought that, at such times, a neutral line forms closer to the Earth, matter in the plasma sheet being then accelerated to higher energies. (This topic is discussed further in Section 6.4.) 2.3.4

Trapped particles

The discovery that there are energetic particles trapped in the magnetosphere was made early in the satellite era by Van Allen and colleagues at the University of Iowa. The information came from a Geiger counter, which they had built for the first successful satellite launched by the USA, Explorer 1, with the intention of studying cosmic rays. The cosmic rays were detected, but the high counting rates

2.3 Particles in the magnetosphere

79

Figure 2.11. Van Allen’s first map of the radiation belts, showing counting rates of the Geiger counter on Pioneer-3. (J. A. Van Allen and L. A. Frank, J. Geophys. Res. 64, 1683, 1959, copyright by the American Geophysical Union.)

which were recorded in parts of the orbit indicated something much more exciting. Apart from its scientific value, this discovery was important in a political sense since it showed that the “space” near the Earth was not an empty void but contained at least some matter and energy and would, very likely, repay closer investigation. Figure 2.11 reproduces what is probably the most famous illustration of that period, showing the double structure deduced from the passage of Pioneer 3, an eccentric-orbit spacecraft that went out 107 400 km in 1958. Thus were discovered the “inner” and “outer” Van Allen belts. The division into two belts is something of an over-simplification because the structure of the trapping region depends on the nature and energy of the particles. The original discovery concerned protons with energy exceeding 30 MeV. Figure 2.12 shows a modern version of the trapped-particle distribution. The mechanism of particle trapping is of interest. A trapped particle moves in three ways (Figure 2.13). It gyrates around a line of the geomagnetic field, bounces back and forth along the line of force between mirror points, and gradually drifts longitudinally around the Earth. The motions are based on a set of adiabatic invariants: (1).

the magnetic moment is constant;

(2). the integral of the parallel momentum over one bouce between mirror points is constant; and (3). the total geomagnetic flux enclosed by the drift orbit is constant. The first holds if the magnetic field does not change during a gyration period, the second if it does not change during a bounce period, and the third if it does not change during the time taken for the particle to encircle the Earth. Hence they are progressively less stringent. The basic trapping mechanism is determined by the first invariant. Consider a charged particle gyrating in the geomagnetic field but also with a component of motion along the field. As the particle spirals from the equator towards the pole it

(a) Electrons

(b) Protons

Figure 2.12. Spatial distributions of (a) trapped electrons of energy exceeding 40 keV, 1 MeV, and 5 MeV, and (b) trapped protons of energy exceeding 100 keV, 10 MeV, and 50 MeV. Since the particle flux falls with energy, these distributions are dominated by the particles just above the threshold stated. The fluxes are omnidirectional and in units of cm2 s1. They are diurnal averages at sunspot maximum. The inner zone is seen in the fluxes of electrons of lower energy, and the slot between the zones is visible in the 1-MeV electron distribution. Otherwise the maximum occurs closer to the Earth at higher energy. (M. Walt, Introduction to Geomagnetically Trapped Radiation. Cambridge University Press, 1994.)

Figure 2.12. (cont.)

(a) Electrons

(b) Protons

2.3 Particles in the magnetosphere

83

longitudinal drift

B

M

M

B gyration

Earth e

p M

M

bouncing

mirror points

Figure 2.13. The motions of a particle trapped in the geomagnetic field.

moves into a region of stronger field. A consequence of the first invariant is that the component of kinetic energy perpendicular to the field is proportional to the magnetic flux density. Hence that component of the particle’s energy increases, and the parallel component decreases by the same amount. Eventually, provided that the particle does not run into the atmosphere first, all the energy is transverse, forward motion stops, and, at that point, the mirror point, the particle is reflected back towards the equator and then into the other hemisphere. Since no energy is being lost or gained, the particle can continue thus for ever or until something else happens to it. The mirror point does not depend on the energy of the particle, but it is directly related to the particle’s direction of travel (the pitch angle) as it crosses the equator. If the mirror point is deep enough to be within the atmosphere, those particles will be lost. Correspondingly, there is, at any point along the path, some pitch angle within which all particles will be lost to the atmosphere at the next bounce. This defines the loss cone, which is generally a small angle of only a few degrees at the equator but increases progressively towards 90° as the Earth’s surface is approached. A trapped particle drifts longitudinally due to the forces which arise from the curvature and the radial change of intensity of the geomagnetic field. Electrons drift eastward and protons westward, and the rate of drift depends (more or less linearly) on the energy of the particle. The times which electrons of various energies and pitch angles take to orbit the Earth are illustrated in Figure 2.14. The longitudinal drift path is determined by the second invariant. The principal population of trapped particles lives in the region of the magnetosphere where the field-lines are closed and almost dipolar. To remove particles from such orbits it is necessary to infringe one of the invariants. These particles are stably trapped. However, owing to the distorted form of the magnetosphere, some drift paths starting in the outer zone take particles into the magnetotail or into the solar wind. These particles cannot complete a full circuit of the Earth, and are trapped only temporally, in pseudo-trapping regions.

Geophysical phenomena

84

Figure 2.14. The time taken for trapped electrons to make one circuit of the Earth at the orbit of a geostationary satellite (6.6RE): (a) as a function of pitch angle for various energies; (b) as a function of energy for pitch angle 90°. (a) Also gives the velocity of the footprint (at 67° latitude) 100 km above the Earth’s surface. (P. N. Collis, private communication.)

The Van Allen particles do not have significant effects on the high-latitude ionosphere while they remain stably trapped. However, it is almost certain that the processes of trapping and loss, because they may transport energetic particles between different regions of the magnetosphere and then deposit them in the ionosphere, are important at high latitudes. 2.3.5

The ring current

One significant consequence of particle trapping is the formation of a ring current in the magnetosphere at or near the inner edge of the plasma sheet. The westward drift of trapped protons and the eastward drift of trapped electrons both contribute to a ring current directed clockwise as viewed from over the north pole. This current increases under disturbed conditions and may be detected with a magnetomenter at the ground as the main phase of a magnetic storm (Section 2.5.2). These particles are not the energetic ones typical of the Van Allen belts, but have been shown by direct measurement to be mainly protons of energy 10–100 keV. The current is generally located at a distance between four and six RE (Figure 2.15) and its existence indicates that there is a concentration of charged particles in that region. Since the drift rate of a trapped particle is proportional to its energy, and all protons carry the same charge, it is easily shown that the total current in the ring is proportional to the total energy of the contributing protons.

2.3 Particles in the magnetosphere

Figure 2.15. Radial profiles of various heavy ions during an inbound pass of the AMPTE satellite on 5 September 1984. (a) The number density. (b) The energy density of particles of energy 5–315 keV per unit electronic charge. (D. J. Williams, Space Sci. Rev. 42, 375, 1985. With kind permission from Kluwer Academic Publishers; after G. Gloeckler et al., private communication.)

It is possible to have a ring current, or a component of a ring current, which goes only part way around the Earth: a partial ring current. 2.3.6

Birkeland currents

Since current has to be continuous, we may well ask how the circuit of a partial ring current is completed. Since the early 1970s it has become generally accepted that currents may flow along the magnetic field-lines between the magnetosphere and the ionosphere. Such currents had first been suggested in 1908 by K. Birkeland, but direct measurements in situ were required in order to prove their existence. Typical distributions of Birkeland currents, sometimes called fieldaligned currents, are illustrated in Figure 2.16. The currents fall into several distinct regions, and by convention the poleward one is “region 1” and the equatorward one is “region 2”, irrespective of the

85

Geophysical phenomena

86

Figure 2.16. Distributions of Birkeland currents during (a) weak and (b) active disturbances. (T. Iijima and T. A. Potemra, J. Geophys. Res. 83, 599, 1978, copyright by the American Geophysical Union.)

direction of the current. The intensity varies during the day in each region and is generally up to 1 or 2 A m2. The total field-aligned current is 106–107 A. The concept of the Birkeland current has profoundly affected ideas about current systems in the ionosphere and magnetosphere. Magnetic observations made at the surface of the Earth may always be interpreted as a two-dimensional current system flowing horizontally at some (unspecified) height in the ionosphere, and the earlier analyses were always performed in this way. These are actually equivalent current systems, which are mathematically correct but are not the only solutions if vertical current is also allowed. The inclusion of Birkeland current has led to distributions that include both ionospheric and magnetospheric parts, and are physically more instructive.

2.4

The dynamics of the magnetosphere

2.4.1

Circulation patterns

A static description of the magnetosphere is alright as far as it goes, but there are certain facts and phenomena that it cannot hope to explain. If we perform a simple calculation of the shape of the magnetopause from the pressure of the solar wind (as suggested in Section 2.2.3), we are using just the component of force normal to the boundary. However, if the solar wind also exerts a force tangential to the boundary, energy will be transfered into the magnetosphere from the solar wind, and we have the possibility that material within the magnetosphere will be set in motion. The situation is then somewhat akin to that of a falling raindrop,

2.4 The dynamics of the magnetosphere

Figure 2.17. The Spq current system in the polar regions due to the circulation of magnetospheric field-lines. (J. A. Ratcliffe, An Introduction to the Ionosphere and Magnetosphere, Cambridge University Press, 1972.)

in which liquid is swept back at the surface and returns down the middle of the drop. The concept of a circulating magnetosphere driven by viscous interaction at the surface was put forward by Axford and Hines in 1961. The nature of the viscous interaction was not specified but was thought to be effectively some kind of friction. Experimental support for circulation came from a study of the Spq current system, whose existence may be inferred from observations with magnetometers at medium and high latitudes. The current system causing Spq – the term means the polar part (p) of the magnetic variation related to the solar day (S) which is observed under magnetically quiet conditions (q) – is illustrated in Figure 2.17. The current flows over the poles from night to day, and there are return currents at lower latitudes. We show the pattern here because it is one that will prove basic in later considerations of the high-latitude ionosphere. Since electrons are tied to the magnetic field in the dynamo region (whereas the positive ions move with the neutral air), the Spq current flow can be interpreted as a motion of magnetic field-lines in the opposite direction; that is, over the pole from the day to the night side of the Earth. In the magnetosphere, therefore, the field-lines circulate over the poles from the day to the night sectors of the Earth with a return flow around the dawn and dusk sides. Figure 2.18(a) shows the basic pattern of magnetospheric circulation in a section through the equatorial plane. Figure 2.18(b) includes the distortion due to the Earth’s rotation which carries the inner part of the magnetosphere with it. Section 2.4.4 indicates how the combined effect of two circulation patterns may be handled.

87

88

Geophysical phenomena

Figure 2.18. Patterns of magnetospheric circulation in the equatorial plane: (a) if due to “friction” of the solar wind at the magnetopause; and (b) if the Earth’s rotation is included. (After A. Nishida, J. Geophys. Res. 71, 5669, 1966, copyright by the American Geophysical Union.)

Present evidence, however, is that, although viscous interaction plays some part in driving the magnetosphere, it is not the major cause. One reason is that the solar wind is so tenuous (having a mean free path between collisions of perhaps 109 km!) that it is hard to believe in a sufficient amount of friction at the magnetopause. Attention therefore moved to an alternative mechanism, based on the work of J. W. Dungey concerning interconnection between the interplanetary magnetic field and the geomagnetic field. The various field configurations that arise when a dipole is placed in a uniform magnetic field are easily illustrated in simple laboratory experiments using a bar magnet situated in an external field. When the fields are parallel there are neutral points on the equator and connections between the two fields. When they are antiparallel, the neutral points are over the poles and there is no interconnection. Figure 2.19 depicts a distorted dipole field representing the geomagnetic field in polar section, with the addition of (a) a northward IMF and (b) a southward IMF. In the second case, neutral points are formed in the equatorial plane and some lines of the IMF connect to geomagnetic lines. This is not so in the first case. We have seen that the IMF tends to lie in the solar-ecliptic plane, oriented at the “garden-hose” angle, but there is usually a north–south component as well and it is this, when it is directed southward, which connects with the geomagnetic field. The IMF is frozen into the solar wind and is therefore carried along with it. When geomagnetic field-lines are connected to those from the IMF they are dragged over the poles from the sunward neutral point, as in Figure 2.19(c), and thereby transported from the day to the night side. While they are over the polar caps the field-lines are open in the sense that they do not connect back to the other hemisphere in any simple or obvious manner. In the tail these lines reconnect and move back towards the Earth. The above picture is of course a simplified one. Detailed consideration taking account of the three-dimensional form of the magnetosphere shows how it is possible to have a degree of connection when the IMF

2.4 The dynamics of the magnetosphere

Figure 2.19. The interaction of terrestrial and interplanetary magnetic fields seen in polar section: (a) northward IMF; (b) southward IMF; and (c) circulation due to the flow of the solar wind. (After C. T. Russell, Critical Problems of Magnetospheric Physics, 1972, after J. W. Dungey, 1963, and R. H. Levy et al., Am. Inst. Aeronaut. Astronaut. J. 2, 2065, 1964.) A: Interplanetary field-line. B: Interplanetary field-line connecting to, or disconnecting from, a geomagnetic field-line. C: Open geomagnetic field-line. D: Closed geomagnetic field-line. N: Neutral point. 0–7: Successive positions of a selected interplanetary field-line.

is northward, and how the east–west component affects the connection point and the resulting circulation pattern. Also, it is thought that viscous interaction does make some contribution; a minor one when the IMF has a southward component, but perhaps the main one when the IMF is northward (and the circulation is then much reduced). The details and mechanisms of magnetic connection and magnetospheric circulation continue to be topics for research, but there is little doubt that the IMF plays an important role. Signatures indicating magnetic connection are observed

89

Geophysical phenomena

90

Figure 2.20. An X-type neutral line in the magnetotail. Plasma flows in from the north and south lobes, and leaves Earthward and tailward along the plasma sheet.

by spacecraft passing through the magnetopause, and the level of geophysical activity increases when the IMF has a southward component. At such times (a) substorms (Section 6.4) occur more frequently; (b) the magnetic flux in the tail lobes is increased; (c) the auroral zone and the dayside cusps are displaced equatorward; and (d) the dayside magnetopause moves inward – all suggesting that a southward IMF increases the rate of magnetospheric circulation. Magnetospheric circulation is a concept of great significance, not only in magnetospheric theory but also, as we shall see, for the high-latitude ionosphere. 2.4.2

Field merging

Magnetospheric circulation requires that, on both the day and the night sides of the Earth, magnetic field-lines are broken and then reconnected in a different configuration. The simplest model of such a process, the X-type neutral line, is illustrated by Figure 2.20. This shows a situation in the central region of the tail. The configuration cannot be static because the tension in the field-lines will produce net forces towards the Earth and into the tail. However, there can be dynamic equilibrium, in which the depletion is replaced by other field-lines moving in from the lobes. Those lines are, of course, replaced by others moving over the poles from the day side of the Earth. There can also be a Y-type neutral line, where the field continues to converge on the tailward side; in that case all the reconnected lines move towards the Earth. The theories of magnetic reconnection came originally from studies of solar flares. In the magnetosphere the process is thought to be that of fast reconnection, first proposed by Petschek (1964) more than 30 years ago. This mechanism invokes an Alfvén wave, which allows reconnection to proceed more rapidly than would diffusion only. The velocity of reconnected field-lines towards the Earth is estimated as about 100 km s1, and the drift towards the neutral sheet as about 10

2.4 The dynamics of the magnetosphere

91

km s1. Particles on the field-lines passing through the reconnection region are accelerated in the direction of the contraction. It is likely that reconnection in the tail occurs not steadily but intermittently in limited regions, and this is probably important in the causes of the aurora. While a pattern of circulation must plainly include reconnection on the night side, it is the connection between the IMF and geomagnetic field-lines on the day side which drives the circulation. Though various ideas have been suggested, the details of this process have not been finally decided. Obviously, a geomagnetic flux tube has to break and connect with an IMF tube, and this is the event which has been identified from the magnetic signature recorded by a nearby spacecraft, as a flux-transfer event (FTE). The newly connected tube of plasma then moves poleward into the boundary layer and joins the general circulation. FTEs are frequent, though individually of short duration and limited spatial extent ((0.5–1)RE). There are more FTEs when the IMF has a strong southward component, and none when it is northward. Presumably, details also vary with the direction of the east–west component. “Quasi-steady” connection is also a possibility. 2.4.3

Magnetospheric electric fields

It is sometimes helpful to regard the dynamic magnetosphere in terms of an electric dynamo and a motor. The magnetosphere may be treated as a magnetohydrodynamic generator, in which a jet of plasma (the magnetosphere) is forced through a static magnetic field (the IMF) and an electric potential is developed by dynamo action. The total potential difference across the magnetosphere is VT vLBn,

(2.10)

where v is the solar-wind speed, L is the width of the magnetosphere, and Bn is the magnetic flux density normal to the boundary. Its value is estimated as about 60 kV, equivalent to an electric field of about 0.3 mV m1. The electric field is directed from the dawn to the dusk side of the magnetosphere. The same potential difference appears across the open field region of the high-latitude ionosphere, the field again being directed from dawn to dusk. The motion of magnetoplasma within the magnetosphere can now be regarded as the effect of this electric field on the geomagnetic field as in an electric motor, according to vE B/|B|2,

(2.11)

where v is the velocity, E the electric field, and B the magnetic field. The magnitude is simply vE/B.

(2.12)

Geophysical phenomena

92

The potential distribution across the magnetotail maps along the field-lines into the polar caps, where it is more accessible to direct measurement, and relationships have been found with the speed of the solar wind and the magnitude and direction of the IMF (Section 5.1.2). If the potential difference across the polar cap is 60 kV, the field-line velocity is about 300 m s1. 2.4.4

The dynamics of the plasmasphere

A good example of treating the dynamics of the magnetosphere in terms of electric fields is the question of the location of the plasmapause, the boundary between the plasmasphere and the outer magnetosphere. The higher levels of the plasmasphere are created by ionospheric plasma moving up and down closed geomagnetic field-lines. However, to explain the dynamics of the plasmasphere as a whole, it is also necessary to take account of the circulation of the magnetosphere. The inner magnetosphere co-rotates with the Earth while the outer magnetosphere follows its own circulation pattern under the control of the solar wind. Generally speaking, the plasmasphere exists on the co-rotating field, and the plasmapause marks the boundary between the inner and outer regions. If we imagine that the plasmasphere is observed by a person fixed in space (i.e. not rotating with the Earth), we can show that its motion in the equatorial plane may be ascribed to a co-rotation electric field of magnitude Er BLRE,

(2.13)

where B is the geomagnetic flux density, L is the observer’s geocentric distance in Earth-radii, RE is the radius of the Earth and  is the Earth’s angular velocity. The plasmapause occurs approximately where the cross-tail and co-rotation fields are equal: ET 

BE LRE, L3

(2.14)

where BE is the geomagnetic flux density at the surface of the Earth at the equator, and we have used the radial variation of flux density in a dipole field, BBE/L3. The condition expressed by Equation (2.14) marks the transition between the circulation regimes of the inner and outer magnetosphere. Putting in numerical values gives: L2 14.4/ET (mV m1).

(2.15)

If the tail field is 1 mV m1 we expect to find the plasmapause at about 4RE. A computation of plasma convection about the Earth was shown in Figure 2.9. In general these flow lines are also equipotentials. The bulge in the plasmasphere in the evening sector, a well-established feature, occurs because the co-rotation and cross-tail fields are in opposite directions on the evening side.

2.5 Magnetic storms

We now see that the principal dynamics of the plasmasphere are (1) filling and emptying along the tubes of force from the ionosphere, which depends on the time of day; and (2) rotation about the Earth in a pattern that is also affected towards its outer edge by the circulation of the magnetosphere. The second factor explains why the location of the plasmapause varies with geomagnetic activity. An increase in the circulation of the magnetosphere implies that condition (2.15) is satisfied closer to the Earth and the plasmasphere must then be smaller. It is thought that the change of circulation peels off layers of plasma, which may exist as detached regions before becoming lost to the outer magnetosphere or into the solar wind. When activity returns to normal the magnetospheric circulation and electric fields return to their previous state, but now the outer tubes of magnetic flux are devoid of plasma. These gradually refill from the ionosphere over a period of several days. The rate of filling is determined by the rate of diffusion of protons from the upper ionosphere (where they are formed by charge exchange between hydrogen atoms and oxygen ions – Section 1.4.4), and by the volume of the flux tube to be filled. Since the latter varies as L4 it takes much longer to refill tubes originating at higher latitude, and, since active periods may recur every few days, there will be periods when the outer tubes are never full. It is probably safe to say that the plasmasphere always suffers from some degree of depletion.

2.5

Magnetic storms

2.5.1

Introduction

The ionospheric storm was introduced in Chapter 1, but the magnetic storm, the main part of which is due to the ring current, is probably the more fundamental. Like the ionospheric storm, it may last from a few hours to several days and it often exhibits three phases. It has been known – though not by its present name – since the eighteenth century from its effect on a compass needle, but progress in understanding any of the storm phenomena dates only from modern times. Because magnetic storms can be monitored without great difficulty using a magnetometer, and long runs of such measurements exist, the magnetic storm has come to be a common reference point in geophysical studies. Although there are superficial similarities between magnetic and F-region storms, the physical connections are not so obvious. These are phenomena that have not proved amenable to simple explanations, and some major questions remain. Part of the problem is that a chain of events is involved. The primary cause is almost certainly the solar wind, affecting the magnetosphere. The magnetospheric consequences then affect the upper atmosphere, and, in some cases, there might even be contributions ascending from the troposphere or stratosphere.

93

Geophysical phenomena

94

Figure 2.21. Typical magnetic storms registered by an equatorial magnetometer. (After M. Sugiura and S. Chapman, Abhandl. Akad. Wiss. Gottingen Math.-Phys. Kl., Special Issue 4, 1960.)

2.5.2

The classical magnetic storm and the Dst index

The typical magnetic storm is illustrated in Figure 2.21. Like the ionospheric storm, this classical magnetic storm consists of three phases: (a). an increase of magnetic field lasting a few hours only, called the initial phase; (b). a large decrease in the H component building up to a maximum in about a day, the main phase; and (c). a slow recovery to normal over the next few days, the recovery phase. The initial phase is caused by the compression of the front of the magnetosphere with the arrival of a burst of solar plasma, as in Chapman and Ferraro’s theory of 1930 (Section 2.2.2). The main phase is due to the ring current which was introduced in Section 2.3.5. The recovery phase is simply a recovery to the pre-storm condition as the ring current decays. The Dst index of magnetic storms is derived from low-latitude magnetograms. In units of nanoteslas ( ), it simply expresses the reduction of the magnetic H component at the equator due to the ring current, and it serves as a useful indicator of the intensity and duration of individual storms. If we assume a distance for the ring current, its magnitude may be derived from the equation B ⬇3I/(10r)⬇Ir,

(2.16)

where B is the change of magnetic flux (in nanoteslas), I is the current (in amperes) and r is the assumed distance (in kilometers). Equation (2.16) is derived from the standard formula for the flux density at the center of a current loop, but with a correction for currents induced in the ground.

2.5 Magnetic storms

95

Figure 2.22. Examples of positive and negative magnetic bays recorded at College, Alaska. The time zone is that of the 150° W meridian. (After S.-I. Akasofu, Polar and Magnetospheric Substorms, Reidel, 1968, with kind permission from Kluwer Academic Publishers.)

2.5.3

Magnetic bays at high latitude; the auroral electrojet

The magnetic storm appears in a different guise at high latitude. By contrast with records from low latitudes, where the effects are due to the growth and decay of the ring current in times measured in hours and days, magnetograms recorded in and near the auroral zone show more rapid changes. The typical pattern there is a series of bay events with typical durations of tens of minutes to an hour or two, such as those illustrated in Figure 2.22. The magnitude of the perturbation in the horizontal component (H ) can be as much as 1000 nT, and its sign tends to be positive before midnight and negative afterwards. Where the sign changes is called the Harang discontinuity. The magnetic bay is caused by an electric current, the auroral electrojet, flowing not in the magnetosphere but in the auroral E region. To explain the sign of the bay, the current flow must be eastward before midnight and westward afterwards: i.e. converging on the midnight meridian. Obviously there must also be return currents for continuity. Chapman’s original interpretation assumed that the currents only flowed only horizontally and this gives a pattern called the SD current system, which is composed of two electrojets with return currents at higher and lower latitudes as in Figure 2.23. This pattern was obtained by averaging the daily

Geophysical phenomena

96

Figure 2.23. Chapman’s original SD current system. The SD analysis takes magnetic disturbance vectors observed simultaneously at a number of stations and infers a current system that could give rise to them. (S.-I. Akasofu, Polar and Magnetospheric Substorms, Reidel, 1968, with kind permission from Kluwer Academic Publishers.)

magnetic variations during the first two days of a number of storms, a proceedure that concentrates the inferred current into the auroral zone. If a three-dimensional current system is allowed, other distributions become possible; modern interpretations include Birkeland currents (Sections 2.3.6 and 6.4.4). 2.5.4

Magnetic indices

The magnetic bays of the kind illustrated in Figure 2.22 are the basis of several magnetic indices, which are regularly compiled and published. The primary purpose of these indices is to quantify the intensity of geomagnetic disturbance and thereby provide a common reference point and a basis for comparison between different observations. The bays, and of course the electrojets causing them, are part of the substorm phenomenon – see Section 6.4 – and as such may be expected also to be related to the intensity of substorms and their frequency of occurrence. The most useful indices are probably those known as Kp, Ap, and AE.

Kp and Ap Kp is based on the range of variation within 3-h periods of the UT day observed in the records from about a dozen selected magnetic observatories. After local weighting, and averaging, the Kp value for each 3 h of the day is obtained on a scale from 0 (for “very quiet”) to 9 (for “very disturbed”). The scale is quasilogarithmic, and the integer values are sub-divided into thirds by use of the symbolsand : thus, 2, 2, 3, 3, 3, etc. Ap is a daily index, obtained from the same basic data, but converted to a linear scale (the 3-h ap) and then averaged over the U. T. day. The value of the intermediate ap is approximately half the range of variation of the most disturbed mag-

2.5 Magnetic storms

97

netic component measured in nanoteslas. The relationship between Kp and ap is given in Table 2.1. It is convenient to present Kp as a Bartels musical diagram, as in Figure 2.24, in which form it often shows how the activity tends to recur with the 27-day solar rotation. In such cases the diagram has some predictive value, but the 27-day recurrence is not always in evidence, being more apparent during the declining phase of the solar cycle. Early measurements of the solar wind led to an empirical relation between the speed of the solar wind and Kp: v (km s1)(8.44 0.74)Kp (330 17),

(2.17)

where Kp is the sum of the eight Kp values over a U. T. day. This was an important early result in that it demonstrated a relationship between the solar wind and disturbances of the geomagnetic field.

AU, AL, and AE The magnetic observatories which contribute to Kp and Ap are situated at various latitudes and longitudes, but with a preponderance in the higher middle latitudes, i.e. the equatorward side of the auroral zone. To achieve an index more tightly related to the auroral regions and to provide better time resolution, AU, AL, and AE were invented by Davis and Sugiura (1966). These indices are obtained by a rather different proceedure. Magnetograms from observatories at several different longitudes around the auroral zone are superimposed and the upper and lower envelopes are read. The upper envelope is AU, the lower envelope is AL and the difference between the envelopes is AE. AL indicates the greatest positive excursion in the auroral zone (probably a pre-midnight bay), AU the greatest negative

Table 2.1. The relationship between Kp and ap Kp

ap

0 1 2 3 4 5 6 7 8 9

0 3 7 15 27 48 80 140 240 400

98

Geophysical phenomena

Figure 2.24. A Bartels diagram of Kp. (Solar–Geophysical Data. National Geophysical Data Center, NOAA, Boulder, Colorado, February 1987.)

excursion (probably post-midnight), and AE, which is the most widely used of the three, is taken as a general indicator of auroral-zone activity irrespective of the local time. Figure 2.25 shows an example. The mean of AU and AL is plotted as A0. The values are published at an hourly interval in printed reports, and they are available at a 2-min interval by special request. In principle, AU is a measure of the eastward auroral electrojet and AL a measure of the westward electrojet. However, both of these indices may be affected by the ring current. The advantage of AE, being their difference, is that it depends

2.5 Magnetic storms

99

Figure 2.25. AU, AL, AE, and AO indices for 12 and 14 June 1988, the first a quiet day (Ap 4) and the second a disturbed one (Ap 20). (Data Book 23, September 1994, World Data Center-C2, Kyoto University, Kyoto.)

solely on the eastward and westward electrojets and should be independent of the ring current and any other zonal current. AE is particularly valuable for indicating when a magnetic substorm occurs. It is also well correlated to the energy coupled into the magnetosphere from the solar wind (Section 6.4.6). Indices such as AE are obviously more sophisticated than Kp, and their preparation requires a correspondingly greater effort, so that the values may not become generally available for a year or more. (Kp, on the other hand, can be obtained through the World Data Centers within a few days.)

Geophysical phenomena

100

The history of magnetic indices and their derivations, advantages, and disadvantages are discussed in detail by Maynaud (1980). Some of the older indices continue to be of interest and are still produced to maintain continuity. The C index, one of the first, is a simple character index in which 0, 1, and 2 mean simply “quiet”, “moderately disturbed” and “disturbed”, respectively. The R and Q indices are range indices like K, but are derived at hourly and 15-min intervals, respectively, instead of every 3 h. We shall refer frequently to magnetic indices of one variety or another when dealing in later chapters with the behavior of the disturbed ionosphere at high latitude. They are the common currency of geophysical disturbance, and are useful beyond the limited topic of magnetic disturbance because of their ease of measurement and the long runs of values accumulated over the years.

2.5.5

Great magnetic storms and a case history

Some storms are so intense and their effects so dramatic that they attract both scientific and popular attention. Though such storms are rare, they serve to illustrate how great the effects can be in extreme cases. Table 2.2 gives the top ten magnetic storms of modern times, ranked in order of the maximum Ap occurring during the storm. In terms of the equatorial index Dst, which measures the strength of the ring current (and comes out with a negative value, since the H component is reduced), the greatest storm was that of 13 March 1989. (Values are not available for the earliest storms of Table 2.2 because Dst has been derived only since 1957.) From more extended lists of great storms it has been noted that (1) most of them occurred after solar maximum rather than at the maximum or before it; and (2) more than half occurred during the four months of the year nearest the equinoxes: that is, during March, April, September, and October. Table 2.2. The top ten magnetic storms of modern times (after J. A. Joslyn, private communication) Order

Date

Maximum Dst

Ap

Solar cycle

1 2 3 4 5 6 7 8 9 10

13 Nov 1960 13 Mar 1989 1 Apr 1960 15 Jul 1959 18 Sep 1941 5 Jul 1941 28 Mar 1946 1 Mar 1941 6 Oct 1960 8 Feb 1986

301 599 327 429

280 246 241 236 230 220 215 205 203 202

19 22 19 19 17 17 18 17 19 21

287 (7th) 307

2.5 Magnetic storms

The storm of 13 March 1989 The storm of March 1989 (Joslyn, 1990), the largest or next to largest on record – depending on the criterion applied – had some quite remarkable effects. It was related to the largest group of sunspots to be seen on the solar disk since 1982, and the effects were not only magnetic but were also detected in the neutral atmosphere, in the ionosphere and on radio communications, as auroral displays in unusual places, and on electric-power transmission.

Magnetic effects The storm began with a sudden commencement at 0127 UT on 13 March, and later that day the magnetic deviation at one mid-latitude station (Boulder, Colorado) amounted to more than 1300 nT. This is nearly three times the typical deviation for a K index of 9, and clearly this storm went well beyond the normal range of measurement for magnetic storms. The Ap index for 13 March was 246 (which is the second largest value recorded during the 57 years since that index was commenced in 1932), and the Dst index determined from equatorial ionograms reached almost 600 nT at one time. The storm continued for about two days. The magnetic variations were large enough to have serious effects on magnetic prospecting. Whereas geophysical exploration techniques are concerned with variations of half a degree, in Alaska the magnetic declination varied by as much as 5°. Analysis of magnetometer data showed that, towards the end of 13 March, the electrojet (normally considered an auroral electrojet) was flowing south of Fredericksburg, Virginia, whose geomagnetic latitude is 49° N. It was reported from Alaska that the flight of homing pigeons was affected.

The aurora, magnetosphere, and solar wind Aurorae were reported at unusually low latitude in several countries. Over the western hemisphere, a red aurora was observed as far south as Florida, Mexico, and Grand Cayman Island. The key to the apparent displacement of the auroral zone, evidenced by the electrojet as well as by the luminosity, is provided by magnetic-field measurements on geosynchronous satellites. GOES-6 and GOES7, both at 6.6RE, left the magnetosphere and entered the solar wind between 0700 and 0800 LT on 13 March. Making a reasonable assumption about the shape of the magnetosphere, it was deduced that the magnetopause at noon was at 4.7RE instead of the usual 10RE. Clearly the magnetosphere was so compressed that its various internal as well as external boundaries moved to unusual places. Unfortunately, there were no direct measurements in the solar wind during this storm. Section 2.2.3 showed that the position of the magnetopause is related to the pressure of the solar wind (Equation (2.6)); in the simplest case the distance depends on the inverse sixth root of the pressure (2Nmv2). Therefore, if the subsolar magnetosphere moved in from 10RE to 4.7Re, the solar-wind pressure would

101

102

Geophysical phenomena

have had to increase by a factor of 60. Some one RE of the movement might be attributable to the southward component of the IMF (Section 2.2.3), but even so one can say with reasonable certainty that the solar-wind pressure must have increased by at least a factor of (10/5.7)6 30 on 13 March.

The ionosphere There was also a severe ionospheric storm on that day. The mid-latitude electron content in the night sector was unusually low immediately after the commencement. After sunrise it remained at essentially night-time values for most of the day and then it returned rapidly to a daytime value just before sunset. The equatorial ionosphere virtually disappeared, and HF radio communications were practically impossible over many circuits, particularly those involving high latitudes. However, VHF communication, which is normally restricted to line-of-sight propagation, was achieved over remarkably long distances due to high-latitude sporadic-E. An analysis of the remarkable world-wide ionospheric effects was presented by Yeh et al. (1992).

Satellite drag The main effect on the neutral upper atmosphere was an increase in air density (and thus an increase in satellite drag) resulting from the heating of the atmosphere. Those whose work it is to track satellites found themselves with many more examples than usual of objects that could not immediately be identified because they were not in the places where they were expected. (In general, magnetic storms increase the rate of decay of satellites in orbit and cause them to re-enter the atmosphere sooner than predicted.)

Electric-power distribution Perhaps the most serious consequence of this storm, however, was its effect on electric-power distribution. It is known that fluctuations in the geomagnetic field induce currents in long metallic lines (both power lines and oil pipelines). In power-distribution systems these may cause the voltage to surge, saturating transformers and tripping protective relays. The electric-power system of Québec suffered a power black-out lasting 9 h on 13 March. Users in the north-eastern USA were also affected. There was a loss of voltage on several power-distribution lines in Sweden. Great storms may be infrequent, but they attract interest as extreme cases of the storm phenomena requiring scientific explanation, and because they may have serious effects on a number of practical activities.

2.5 Magnetic storms

2.5.6

Wave phenomena of the magnetosphere Hydromagnetic and magnetosonic waves

In a sound wave the restoring force is due to the compressibility of the medium; in the case of a gas, its pressure. At the lowest frequencies gravity can also be significant, giving the acoustic-gravity wave (Section 1.6). In a magnetic field another restoring force comes into play, and that is the magnetic pressure across the field and the magnetic tension along the field-lines. The situation in the magnetosphere is that the ionization may not cross fieldlines, and therefore in transverse motions they must move together. Ordinary sound waves are allowed along the field-lines because the gas displacement is longitudinal, but in waves transverse to the field both the gas pressure and the magnetic pressure must be included. This combination makes possible a range of hydromagnetic waves. The basic hydromagnetic wave is the Alfvén wave, which propagates along the magnetic field but whose displacement is transverse. The Alfvén wave is analogous to the transverse wave on a taut string, the tension being the magnetic tension (B2/0), and the mass per unit length being simply the mass density of the plasma. The speed of the transverse wave is then given by Equation (2.7): vA 

B , ( 0
) 1/2

where B is the magnetic flux density, 0 the permeability of free space, and
the density of the plasma in kg m3. The Alfvén wave and the sonic wave are independent when they are traveling along the field direction, but at other angles they interact to give mixed magnetosonic waves. There are two such waves in general, except that perpendicular to the field there is only one, having speed (vA2 s2)1/2, s being the speed of sound.

Micropulsations If the sensitivity of a magnetometer is increased sufficiently, small fluctuations of the geomagnetic field with periodicites of minutes and seconds can be detected. These are micropulsations. They are due not to electromagnetic waves but to hydromagnetic waves in the magnetosphere, and in magnitude they are less than 104 of the total geomagnetic field. Their connection with ionospheric radio is somewhat indirect, but an introduction is in order because they comprise a significant phenomenon of the high-latitude ionosphere. In some cases they are connected with auroral activity, and there are also some diagnostic applications indicating conditions in the magnetosphere. Micropulsations are classified according to period and duration, as in Table 2.3. The impulsive variety (Pi) occurs mainly in the evening, whereas the more

103

104

Geophysical phenomena

regular and persistent regular pulsations (Pc) prefer the morning and the daylight hours. The magnetospheric origin of micropulsations is demonstrated by their similarity in magnetically conjugate regions, but a variety of mechanisms is involved in their generation. Pc pulsations are generated either at the surface of the magnetosphere or within it, and they propagate in a hydromagnetic mode. Pc1 are attributed to bunches of protons (probably) traveling back and forth between mirror points (Section 2.3.4). A resonance between protons and ion-cyclotron waves, which rotate in the same sense in the geomagnetic field, is probably involved. Pc2–5 are explained as various modes of oscillation within the magnetosphere, some propagating across and some along the field lines. The period of Pc3 and 4 may be interpreted in terms of Alfvén waves, whose speed depends on the plasma density and the magnetic field strength. The characteristic frequency changes across the plasmapause (Section 2.3.2) due to the sharp change of particle density. The topic of micropulsations was discussed in detail by Jacobs (1970).

Instabilities The interaction of magnetospheric waves with the particle population of the magnetosphere is a complex subject that we can no more than indicate here. For ionospheric physics its basic importance is that waves and particles may exchange energy, and that this process can become unstable. For example, trapped electrons generate whistler-mode waves in the VLF band (Section 3.4.7), which, under the right conditions, may then interact with the population of trapped electrons, scattering them into the loss cone (Section 2.3.4). This is a mechanism that, thereby, may cause the spontaneous precipitation of trapped electrons into the atmosphere. There is a large literature on wave–particle interactions in the magnetosphere. The interested reader might like to start with Lyons and Williams (1984), Chapter 5. One purely ionospheric instability is the two-stream, or Farley–Buneman, instability, which produces electrostatic waves in the E-region electrojet when Table 2.3. Micropulsations Continuous and regular Type Pc1 Pc2 Pc3 Pc4 Pc5

Period (s) 0.2–5 5–10 10–45 45–150 150–600

Irregular Type

Period (s)

Pi1 Pi2

1–40 40–150

2.6 Ionization by energetic particles

streams of ions and electrons differ in velocity by more than a critical amount. This is the mechanism causing the irregularities in ionization that make auroral radar possible. That topic is discussed in Sections 3.5.1, 4.2.2, and 6.5.5. The Kelvin–Helmholz instability is a commonplace phenonenon, being the cause of waves on the surface of a pond on a windy day. It works because any projection above the level surface alters the air flow in such a way as to increase the perturbation – a simple case of positive feedback. The magnetosphere also has interfaces, the most obvious being that with the solar wind at the magnetopause, and Kelvin–Helmholtz waves and vortices may be produced there. A slightly less superficial introduction to magnetospheric waves can be found in Hargreaves (1992), Chapter 9.

2.6

Ionization by energetic particles

The main source of ionization in the upper atmosphere is solar radiation in the Xray and EUV bands. There is, however, another source of a quite different kind, namely energetic particles. Although they are not entirely absent from middle latitudes, these are much more important at high latitudes, where they may at times become the main source of ionization. As we shall see in later chapters, two very significant sources at high latitude are electrons associated with the aurora, and protons (plus some -particles) emitted from the Sun during some solar flares. 2.6.1

Electrons

Various methods have been used to calculate the rate of ion production by a stream of energetic electrons arriving from some source above the atmosphere. The most generally useful one relies on laboratory measurements of the range of electrons in air. An electron loses energy to the neutral gas particles with which it collides, and the rate of loss obviously depends on the number of gas particles encountered. Thus, in a uniform atmosphere the distance traveled varies in inverse proportion to the gas pressure. The unit of range (r0) is therefore [pressure] [distance]: atm cm, for example. The energy goes into exciting and ionizing the neutral particles. In this instance we are interested in the ionization. An energetic particle entering the atmosphere from above travels into a medium of increasing density, and the altitude, hp, to which it penetrates is such that the product of pressure and distance, integrated above hp, is equal to the range r0. Obviously, this particle will ionize only above height hp, and the total number of ion–electron pairs produced will depend on E/E, where E is the initial energy of the particle and E is the energy required for each ionization (generally taken as 34 or 35 eV). The third fact which has to be taken into account is that the rate of energy loss, and therefore of ion production, along the path is a function of the particle’s

105

Geophysical phenomena

106

Figure 2.26. Production rates due to monoenergetic electrons of various initial energies. (After M. H. Rees, Planet. Space Sci. 11, 1209, Copyright 1963, with permission from Elsevier Science.)

velocity. To quantify this, laboratory measurements are again brought to bear in the form of an “efficiency”, which is a function of the atmospheric depth at a point along the track divided by that at the point where the particle eventually stops; i.e. s/sp. (The atmospheric depth is the total mass of gas in a column of unit crosssection along the path of the particle.) The efficiency is normalized to unity over the whole path (from s/sp 0 to s/sp 1), and it comes to a maximum at s/sp 0.4 for a monoenergetic electron beam traveling directly along the magnetic field. If the electrons arrive over a range of angles, as would be the case in nature, some particles travel in spiral paths and so cover a greater distance; in this case the efficiency peaks at a smaller value of s/sp such as 0.1 or 0.2. These factors clearly influence the distribution of ion production in the atmosphere, and the height of the maximum production rate in particular. Figure 2.26 shows calculated production rates in a model atmosphere due to monoenergetic electrons of various initial energies. Note that the production rate peaks at lower altitude and the distribution is narrower for higher initial enery. To get the effect of a more realistic spectrum, the production rate must be integrated over energy. 2.6.2

Bremsstrahlung X-rays

When energetic electrons collide with neutral gas particles a small amount of their energy is converted to X-rays through the Bremsstrahlung process – literally “braking radiation” – as they are rapidly decelerated. The X-rays penetrate deeper into the atmosphere than do the primary electrons and may be observed by balloon-borne detectors at heights of 30–40 km. Some X-rays are also scattered back out of the atmosphere and may be detected on satellites.

2.6 Ionization by energetic particles

107

The computation of the Bremsstrahlung X-ray flux is fairly complicated, since an electron of energy E can produce photons of energy E or less, and the X-rays are emitted over a wide range of angles. Inverting an observed X-ray spectrum to give the spectrum of the primary electrons is even more difficult. The usual practice is to draw on a set of computations giving the Bremsstrahlung due to single electrons of specified energy, but even so it is usually neccessary to assume a form for the spectrum (e.g. exponential). When the X-rays are stopped by the atmosphere they create ionization at that level. The ionization rate due to Bremsstrahlung is several factors of ten smaller than that due to the primary electrons higher up, but, at the height concerned, possibly 50 km or below, it is the major source of ionization at times of auroral electron precipitation. Table 2.4 compares the heights and maximum production rates due to direct and Bremsstrahlung ionization for several initial electron energies. Figure 2.27 illustrates the relative altitude and magnitude of direct and X-ray ionization (actually the rate of deposition of energy) due to a spectrum of incident electrons with characteristic energy 10 keV. 2.6.3

Protons

Significant ionization may also be caused by energetic protons, especially at high latitudes during polar-cap events, which are due to fluxes of protons released from the Sun at the time of a solar flare. A lesser flux of -particles will generally arrive simultaneously. These particles, which are significantly more energetic than the auroral electrons discussed above, lose energy in colliding with the atmospheric gas and leave ionized trails. The gas concerned is principally that of the mesosphere, whose composition is essentially like that of the troposphere, and therefore the rate of energy loss is well known from laboratory measurements. A graph showing the rate of energy loss against the distance traveled is called a Bragg curve. In the energy range of interest to us the loss rate increases as the proton slows down, and, over the range 10–200 MeV, the loss rate is almost inversely proportional to the energy, a typical value being 0.8 MeV per meter of path in air at standard temperature and Table 2.4. Direct and X-ray ion production Height of maximum production (km) E (keV) 3 10 30 100

Direct 126 108 94 84

Maximum production rate (ion pairs cm3 per electron)

X-ray

Direct

X-ray

88 70 48 37

2.5 10 1.4 104 5.6 104 1.9 103 5

5.9 1010 1.3 108 2.3 107 1.3 105

Geophysical phenomena

108

140 130 120 110 100

Altitude (km)

90 80 70 60 50 40 30 20 10 –12

–11

–10

–9

–8

–7

Log of Energy Deposition Rate (in keV

cm– 3

–6

–5

s–1)

Figure 2.27. A comparison of direct and X-ray energy-deposition rates from an incoming electron flux with an exponential spectrum of characteristic energy 10 keV. The solid line (J. G. Luhmann, J. Atmos. Terr. Phys. 38, 605, 1976) shows the deposition by electron impact. The dashed line (M. J. Berger et al., J. Atmos. Terr. Phys. 36, 591, 1974) and the circles (J. G. Luhmann, J. Atmos. Terr. Phys. 39, 595, 1977) include the energy deposited by Bremsstrahlung X-rays. (After Luhmann 1977, Copyright, with permission from Elsevier Science.)

pressure when the energy is 100 MeV. The energy may be assumed to be used entirely in creating ion–electron pairs, each requiring about 35 eV. The nature of the Bragg curve, combined with the density distribution of the atmosphere, means that the ionization due to a proton entering the atmosphere from space is very concentrated towards the end of the path. A vertically incident 50 MeV proton, for example, loses half its energy over the last 2.5 km of the path and the last 10% over only 100 m. One consequence is that the penetration level does not depend strongly on the angle of incidence except near 90°. Productionrate profiles for protons of various initial energies are given in Figure 2.28. Note the low altitudes which may be reached by the more energetic particles. For a spectrum of proton energies the total effect would be calculated by appropriate summing over these curves at each height. There is a similar procedure for dealing with the ionization by -particles.

2.7 References and bibliography

Figure 2.28. Production rates due to incident monoenergetic protons. The initial energies are given in MeV, and in each case the flux is 1 proton cm2 s1 sr1. (G. C. Reid, Fundamentals Cosmic Phys. 1, 167, 1974. Copyright OPA (Overseas Publishers Association) NV, with the permission of Gordon and Breach Publishers.)

2.7

References and bibliography

2.2

The magnetosphere

Burlaga, L. F. (1982) Magnetic fields in the interplanetary medium. Solar System Plasmas and Fields (eds. J. Lemaire and M. J. Rycroft), p. 51. Pergamon Press, Oxford. Carpenter, D. L. and Smith, R. L. (1964) Whistler measurements of electron density in the magnetosphere, Rev. Geophys. 2, 415. Gosling, J. T. (1972) Predicting the solar wind speed. Solar Activity Observations and Predictions (eds. P. S. McIntosh and M. Dryer), p. 231. MIT Press, Cambridge Massachusetts. Holzer, T. E. (1979) The solar wind and related astrophysical phenomena. Solar System Plasma Physics (eds. C. F. Kennel, L. J. Lanzerotti and E. N. Parker), Vol. I, p. 103. Elsevier Science Publishers, Amsterdam. Pertinec, S. M. and Russell, C. T. (1993) External and internal influences on the size of the dayside terrestrial magnetosphere. Geophys. Res. Lett. 20, 339. Raitt, W. J. and Schunk, R. W. (1983) Composition and characteristics of the polar wind. Energetic Ion Composition in the Earth’s Magnetosphere (ed. R. G. Johnson), p. 99. Terra Scientific Publishing Co., Tokyo. Rich, F. J. and Hairston, M. (1994) Large-scale convection patterns observed by DMSP. J. Geophys. Res. 99, 3827. Smith, E. J., Tsurutani, B. T. and Rosenberg, R. L. (1978) Observations of the interplanetary sector structure up to heliographic latitudes of 16°: Pioneer 11. J. Geophys. Res. 83, 717. Vasyliunas, V. M. (1983) Large-scale morphology of the magnetosphere. Solar–Terrestrial Physics (eds. R. L. Carovillano and J. M. Forbes), p. 243. Reidel, Dordrecht.

109

Geophysical phenomena

110

Wilcox, J. M. and Ness, N. F. (1965) Quasi-stationary corotating structure in the interplanetary medium. J. Geophys. Res. 70, 5793.

2.3

Particles in the magnetosphere

Burch, J. L. (1977) The magnetosphere. Upper Atmosphere and Magnetosphere, p. 42. National Academy of Sciences, Washington DC. Carpenter, D. L. and Anderson, R. R. (1992) An ISEE/whistler model of equatorial electron density in the magnetosphere. J. Geophys. Res. 97, 1097. Chappell, C. R., Harris, K. K. and Sharp, G. W. (1970) A study of the influence of magnetic activity on the location of the plasmapause as measured by OGO-5. J. Geophys. Res. 75, 50. Helliwell, R. A. (1976) Whistlers and Related Ionospheric Phenomena. Stanford University Press, Stanford, California. Iijima, T. and Potemra, T. A. (1978) Large-scale characteristics of field-aligned currents associated with substorms. J. Geophys. Res. 83, 599. Potemra, T. A. (1983) Magnetospheric currents. Johns Hopkins APL Tech. Digest 4, 276. Ratcliffe, J. A. (1972) An Introduction to the Ionosphere and Magnetosphere. Cambridge University Press, Cambridge. Van Allen, J. A. (1959) The geomagnetically trapped corpuscular radiation. J. Geophys. Res. 64, 1683. Walt, M. (1994) Introduction to Geomagnetically Trapped Radiation. Cambridge University Press, Cambridge. Williams, D. J. (1985) Dynamics of the Earth’s ring current: theory and observation. Space Sci. Rev. 42, 375.

2.4

Dynamics of the magnetosphere

Dungey, J. W. (1963) The structure of the exosphere or adventures in velocity space. In Geophysics, The Earth’s Environment (eds. C. De Witt, J. Hieblot, and A. Lebeau). Gordon and Breach, New York. Levy, R. H., Petschek, H. E., and Siscoe, G. L. (1964) Aerodynamic aspects of the magnetospheric flow. Am. Inst. Aeronaut. Astronaut. J. 2, 2065. Nishida, A. (1966) Formation of plasmapause, or magnetospheric plasma knee, by the combined action of magnetospheric convection and plasma escape from the tail. J. Geophys. Res. 71, 5669. Petschek, H. E. (1964) Magnetic field annihilation. The Physics of Solar Flares (ed. W. N. Hess), Report SP-50, p. 425. NASA, Washington DC. Ratcliffe, J. A. (1972) An Introduction to the Ionosphere and Magnetosphere. Cambridge University Press, Cambridge. Russell, C. T. (1972) The configuration of the magnetosphere. Critical Problems of Magnetospheric Physics, p.1. IUCSTP Secretariat, National Academy of Science, Washington D.C.

2.5

Magnetic storms

Akasofu, S.-I. (1968) Polar and Magnetospheric Substorms. Reidel, Dordrecht. Davis, T. N. and Sugiura, M. (1966) Auroral electrojet activity index AE and its universal time variations. J. Geophys. Res. 71, 785. Hargreaves, J. K. (1992) The Solar–Terrestrial Environment. Cambridge University Press, Cambridge.

2.7 References and bibliography

Jacobs, J. A. (1970) Geomagnetic Micropulsations. Springer-Verlag, Berlin. Joslyn, J. A. (1990) Case study of the great magnetic storm of 13 March 1989. Astrodynamics (eds. Thornton, Proulx, Prussing and Hoots). Lyons, L. R. and Williams, D. J. (1984) Quantitative Aspects of Magnetospheric Physics. Reidel, Dordrecht. Maynaud, P. N. (1980) Derivation, Meaning and Use of Geomagnetic Indices. American Geophysical Union, New York. Sugiura, M. and Chapman, S. (1960) The average morphology of geomagnetic storms with sudden commencement. Abhandl. Akad. Wiss. Göttingen Math.-Phys. Kl. Special Issue 4, 53. Yeh, K. C., Lin, K. H. and Conkright, R. O. (1992) The global behavior of the March 1989 ionospheric storm. Can. J. Phys. 70, 532.

2.6

Ionization by energetic particles

Berger, M. J., Seltzer, S. M. and Maeda, K. (1974) Some new results on electron transport in the atmosphere. J. Atmos. Terr. Phys. 36, 591. Luhmann, J. G. (1976) Auroral electron spectra in the atmosphere. J. Atmos. Terr. Phys. 38, 605. Luhmann, J. G. (1977) Auroral bremsstrahlung spectra in the atmosphere. J. Atmos. Terr. Phys. 39, 595. Rees, M. H. (1963) Auroral ionization and excitation by incident energetic electrons. Planet. Space Sci. 11, 1209. Reid, G. C. (1974) Polar-cap absorption – observations and theory. Fundamentals Cosmic Phys. 1, 167.

General reading on the topics of Chapter 2 Books Akasofu, S.-I. and Chapman, S. (1972) Solar–Terrestrial Physics. Oxford University Press, Oxford. Baumjohann, W. and Treumann, R. A. (1996) Basic Space Plasma Physics. Imperial College Press, London. Carovillano, R. L. and Forbes, J. M. (eds.) (1983) Solar–Terrestrial Physics. Reidel, Dordrecht. Carovillano, R. L., McClay, J. F. and Radoski, H. R. (1968) Physics of the Magnetosphere. Springer-Verlag, New York. Hess, W. N. (1968) The Radiation Belt and Magnetosphere. Blaisdell, Waltham, Massachusetts. Hess, W. N. and Mead, G. D. (eds.) (1968) Introduction to Space Science. Gordon and Breach, New York. Hundhausen, A. J. (1972) Coronal Expansion and Solar Wind. Springer-Verlag, New York. Jacobs, J. A. (1970) Geomagnetic Micropulsations. Springer-Verlag, New York. Jursa, A. S. (ed.). (1985) Handbook of Geophysics and the Space Environment. Air Force Geophysics Laboratory, US Air Force, National Technical Information Service, Springfield, Virginia. Kamide, Y. (1988) Electrodynamic Processes in the Earth’s Ionosphere and Magnetosphere. Kyoto Sangyo University Press, Kyoto.

111

112

Geophysical phenomena

Le Galley, D. P. and Rosen, A. (eds) (1964) Space Physics. Wiley, New York. Lyons, L. R. and Williams, D. J. (1984) Quantitative Aspects of Magnetospheric Physics. Reidel, Dordrecht. Nishida, A. (ed.) (1982) Magnetospheric Plasma Physics. Reidel, Dordrecht. Parks, G. K. (1991) Physics of Space Plasmas. Addison-Wesley Publishing Co., Redwood City, California. Roederer, J. G. (1974) Dynamics of Geomagnetically Trapped Radiation. SpringerVerlag, Berlin. Schulz, M. and Lanzerotti, L. J. (1974) Particle Diffusion in the Radiation Belts. Springer-Verlag, New York. Treumann, R. A. and Baumjohann, W. (1997) Advanced Space Plasma Physics. Imperial College Press, London.

Conference reports Akasofu, S.-I. (ed.) (1980) Dynamics of the Magnetosphere. Reidel, Dordrecht. Beynon, W. J. G., Boyd, R. L. F., Cowley, S. W. H. and Rycroft, M. J. (1989) The Magnetosphere, the High-Latitude Ionosphere, and their Interactions. The Royal Society, London. Johnson, R. G. (ed.) (1983) Energetic Ion Composition in the Earth’s Magnetosphere. Terra Scientific Publishing Co., Tokyo. Kamide, Y. and Slavin, J. A. (eds.) (1986) Solar Wind–Magnetosphere Coupling. Terra Scientific Publishing Co., Tokyo. King, J. W. and Newman, W. S. (eds.) (1967) Solar–Terrestrial Physics. Academic Press, London. Lemaire, J. F., Heynderickx, D. and Baker, D. N. (eds.) (1996) Radiation Belts: Models and Standards. American Geophysical Union, Washington DC. McCormac, B. M. (ed.) (1966) Radiation Trapped in the Earth’s Magnetic Field. Reidel, Dordrecht. McCormac, B. M. (ed.) (1968) Earth’s Particles and Fields. Reinhold, New York. McCormac, B. M. (ed.) (1970) Particles and Fields in the Magnetosphere. Reidel, Dordrecht. McCormac, B. M. (ed.) (1972) Earth’s Magnetospheric Processes. Reidel, Dordrecht. McCormac, B. M. (ed.) (1974) Magnetospheric Physics. Reidel, Dordrecht. McCormac, B. M. (ed.) (1976) Magnetospheric Particles and Fields. Reidel, Dordrecht. Olsen, W. P. (ed.) (1979) Quantitative Modeling of Magnetopheric Processes. American Geophysical Union, Washington DC. Potemra, T. A. (ed.) (1984) Magnetospheric Currents. American Geophysical Union, Washington DC. Song, P., Sonnerup, B. U. O. and Thomsen, M. F. (eds.) (1995) Physics of the Magnetopause. American Geophysical Union, Washington DC. Tsurutani, B. T., Gonzalez, W. D., Kamide Y. and Arballo, J. K. (1997) Magnetic Storms. American Geophysical Union, Washington DC.

Chapter 3 Fundamentals of terrestrial radio propagation

3.1

Introduction

Since we are concerned with the propagation of radio waves over the entire radio spectrum at high latitudes, it should be useful to review the basic physics and terminology of the propagation of radio waves in general. The radio spectrum extends from the extra-low-frequencies (ELF) band through microwaves and millimeter waves. Table 3.1 shows the radio spectrum from ⬃30 Hz to 30 GHz, along with the International Telecommunications Union (ITU) band designations.

3.2

Electromagnetic radiation

3.2.1

Basics of line-of-sight propagation in vacuo

An example of line-of-sight (LOS) propagation is that between two spacecraft in deep space where the medium is virtually a vacuum. The refractive index is unity and the speed of an electromagnetic (EM) wave is independent of its frequency and equal to the speed of light. By definition, an isotropic radiator is one that radiates equally in all directions. If power P is radiated, the power density S (the power crossing unit area) at a distance d from the source is SP/(4d 2),

113

(3.1)

114

Fundamentals of propagation

From EM theory, the vector S is the Poynting flux, given by (3.2)

SE H,

Table 3.1. The radio spectrum (as defined by the International Telecommunications Union (ITU)), primary modes of propagation, and effects of the terrestrial ionosphere Principal modes of propagation

ITU designation

Frequency range

Principal uses

Extra-lowfrequency (ELF)

30–300 Hz

Groundwave and Earth–ionosphere waveguide mode

Submarine communication

Very-lowfrequency (VLF)

3–30 kHz

Same as above

Navigation, standard frequency and time dissemination

Low-frequency (LF)

30–300 kHz

Same as above

Navigation LORAN-Ca

Mediumfrequency (MF)

300–3000 kHz

Primarily groundwave, but skywaveb at night

AM broadcasting, maritime, aeronautical communication

High-frequency (HF)

3–30 MHz

Primarily skywave, some groundwave

Shortwave broadcasting, amateur, fixed services

Very-highfrequency (VHF)

30–300 MHz

Primarily LOS, some skywave at lower VHF

FM broadcasting, television, aeronautical communication

Ultra-high-(UHF) frequency

300–3000 MHz

Primarily LOS, some refraction and scattering by the ionosphere

Television, radar, navigationc, aeronautical communication

Super-high-(SHF) frequency

3–30 GHz

Same as above

Radar, space communication

Notes: a The LORAN-C system will probably be superseded by the GPS system. b “Skywave” denotes the Earth–ionosphere–Earth-reflected mode. c Global Positioning System of satellite constellation.

3.2 Electromagnetic radiation

115

E being the electric vector and H the magnetic vector. Since E/H120

(3.3)

for an EM wave, E and H being, respectively, the electric and magnetic field strengths, SE 2/(120).

(3.4)

Therefore, E 兹30P/d .

(3.5a)

In SI units, P is in watts, d in meters, S in W m2, and E in V m1. It may be more practical to express d in kilometers, P in kilowatts, and E in mV m1, in which case E (mV m1)173 兹P (kW) /d (km).

(3.5b)

If the antenna does not radiate isotropically, it is said to have a gain (G), given by the ratio of the Poynting flux at a point on the axis divided by the flux that would be received at the same point if the same power were radiated instead from an isotropic radiator. If an antenna with gain Gt transmits power Pt and the receiving antenna has aperture Ar (m2 ), the power received is Pr Ar SAr Gt Pt /(4d 2)

(3.6)

Er  兹30Pt ArGt /d.

(3.7)

and

Antenna theory shows that gain and aperture are related by G4A/2,

(3.8)

in which A is the true aperture if the antenna has the form of an efficient dish, but may be an effective area otherwise. An isotropic radiator (which is hypothetical in any case for an EM wave) has unity gain and effective area 2/(4). For a half-wave dipole, which may be taken as the reference, G1.64 and A1.642/(4)0.13052.

In a point-to-point link it is often convenient to represent the reduction of signal due to the separation (d ) between transmitting and receiving antennas as the free-space attenuation, Lb 20 log(4d /),

(3.9)

which follows from Equations (3.6) and (3.8) assuming that both antennas are isotropic radiators. The gain (sometimes called directivity) is given approximately by G30000/(),

(3.10)

Fundamentals of propagation

116

where  and  are the half-power beamwidths (in degrees) in the E and H planes, respectively, assuming that there are no sidelobes. The formula applies only up to 20° beamwidth. Although it is an important topic in radio propagation, a full discussion of antennas would be outside the scope of this book. Some of the many treatments are listed in the references. Information on radiation patterns and advice on siting are contained in the publications of the American Radio Relay League (ARRL) in the references. For detailed discussions of Fresnel-zone siting fundamentals, bandwidth, and terrain effects see Appendix A7 of Hunsucker (1991), Freeman (1997), or Wolff (1988). Computer programs for antenna design and performance analysis are listed in Table 3.2 and in Balanis (1997). 3.2.2

Principles of radar

In radar a transmitted signal is reflected from a target and then detected by a receiver, which may but need not be co-located with the transmitter. These are monostatic and bistatic systems, respectively. The target may be a solid object (Figure 3.1) or a distributed scattering medium (as in coherent and incoherent Table 3.2. Antenna design and performance-analysis programs Name of software

Description

Source

NEC

Numerical electromagnetic code

NEC/WIRES 1.5

One version of NEC

Brian Beezley 3532 Linda Vista Dr., San Marcos, CA 92069, USA

NEC/Yagis 2.0

Uses NEC to model Yagis and arrays of Yagis

″

″

YO 6.0

Optimizes Yagi–Uda designs

″

″

AO 6.0

Optimizes antenna designs for any wire-or tubing-type antenna

″

″

ELNEC MININEC GAP, BIA, ACP, and General Antenna COMSAT antenna lab suite Phased Array Program, Beam Intermodulation Program Analyzer, Antenna Coverage http://www.comsat. Program, and Phased Array Com/Corp/lab/labs.html Program XFDTD 4.0

User-friendly electromagnetics software, covers more esoteric antennas, scattering, etc.

REMCOM Inc. http://www.remcominc.com

3.2 Electromagnetic radiation

117

R ∆R

Ae

Figure 3.1. A schematic diagram of the radar principle.

scatter radars – Sections 4.2.2 and 4.2.3). A treatment of radar begins with the radar equation. If power Pt is radiated by the transmitter using an antenna with gain G, the power density at a target at distance R is (Equation (3.6)) SGPt /(4R2).

(3.11)

If the target has cross-section , and the power intercepted is scattered equally in all directions, the power received back at the radar is Pr GPt /(4R2) Ae /(4R2) Pr GPt Ae /[(4)2 R4]

(3.12)

where Ae is the effective area of the radar antenna. (If the scattering is not omnidirectional, this is taken into account in the value of .) From Equation(3.8) we may also write the radar equation as Pr G 2 2  Pt 4 4R2 G 22 /(64 2 R4).

(3.13)

The distance beyond which the target cannot be detected is the maximum radar range, Rmax, and the limit is when the received echo power, Pr, just equals the minimum detectable signal, Smin. Hence (from Equation (3.12)),

Fundamentals of propagation

118

Rmax {PtGAe /[(4)2 Smin]}1/4,

(3.14a)

which is the most common form of the radar range equation. Using Equation (3.8) gives the alternative forms Rmax {Pt G2 2 /[(4)3 Smin]}1/4

(3.14b)

Rmax [Pt A2e /(42 Smin )]1/4.

(3.14c)

and

The foregoing discussion applies to situations like that in Figure 3.1, where the target is smaller than the transmitter beamwidth. The larger the target, the more power is returned. If the scattering region is larger than the beamwidth (a beamfilling target), as may happen when the ionosphere is the target, all the incident power is intercepted and then the expression for the echo power received back at the radar has the form Pr Pt Ae /(4 R2),

(3.15)

where represents the scattering property of the target medium. The echo power now varies as R2 instead of R4. If the ionosphere is the target, the return would probably come from a large number of individual scatterers, and the would include the number of scatterers within the radar pulse and beamwidth at any one time, as well as their directional properties. The physical length of the transmitted pulse (Figure 3.1) is Rc,

(3.16)

c being the speed of light in vacuo and  the pulse duration. The resolution in range is R/2. Discussions of the various forms of the radar equation and their implications, including theorems applicable to “soft targets,” are given by Skolnik (1980) and by Hunsucker (1991; pp. 38–39).

3.2.3

The significance of the refractive index A simple propagating wave

If a radio source generates an electric field EE0 cos(t), which propagates at speed v in the z direction, the field at a distance z from the source is

3.2 Electromagnetic radiation

119

EE0 cos[(t z/v)] E E0 cos[(t kz)] E E0 cos[2(t /T z/)]

(3.17)

since  2f, v f, and k2 / by definition. T is the period,  and f are the frequency in radians s1 and hertz, respectively, and k is the wave number, propagation constant, or phase-shift factor. E is the instantaneous value of the electric field at (t, z) and E0 is the amplitude of the electric field. Plainly, the same phase repeats itself every T (1/f ) in time and every  2 /k in distance. For the propagation of a plane wave in three dimensions, k can be regarded as a vector along the propagation direction, having components kx , ky, and kz that give the wavelengths in the x, y, and z directions and thus the phase velocities vx  x f, vy  y f, and vz  z f.

The refractive index We will use vp to denote the phase velocity, and for an EM wave its value depends on the nature of the medium; vp 1/( )1/2,

(3.18)

where  and are the permeability and permittivity of the medium. In free space this becomes c1/(0

)1/2 3 108 m s1,

0

(3.19)

where 0 and 0 are, respectively, the permeability and permittivity of free space. The ratio nc/v is the refractive index of the medium, and the propagating wave may then be written EE0 cos(t nz/c).

(3.20)

If the refractive index varies with the wave frequency, the medium is said to be dispersive. A modulated wave is not monochromatic, and in a dispersive medium the modulation travels not at the phase velocity but at the group velocity (u), which is related to the phase velocity by u( k/ )1.

(3.21)

Only if vp is independent of , so that k /vp, does uvp.

Propagation in a lossy medium If the medium absorbs energy from the wave, the amplitude decreases with distance as exp(!z), where ! is the absorption coefficient, and the amplitude

Fundamentals of propagation

120

decreases by a factor of e over a distance 1/!. It is convenient here to use the j notation, writing EE0 exp[ j(t nz/c)],

(3.22)

where j√ 1, and it is understood that the real part is taken (since e j cos   jsin  ). Taking a complex refractive index n j

(3.23)

( has been added to  to avoid confusion with the permeability), gives EE0 exp[j(t  z/c)] exp( z/c).

(3.24)

Hence, the real part of the refractive index determines the velocity of the wave, and the imaginary part gives the absorption coefficient !   /c2 /0,

(3.25)

0 being the free-space wavelength. Alternatively, we may introduce a complex propagation constant,  jk  j

(3.26)

giving EE0 exp[ j(tjz) EE0 exp[ j(t z)]exp(z).

(3.27)

Thus, comparing with Equations (3.24) and (3.25),   /c;    /c !.

Conductivity For a partial conductor the absorption is related to the conductivity, , and it can be shown (Hunsucker, 1991, pp. 25–31) that    兹 ( j /  ) .

(3.28)

Squaring, and equating real and imaginary parts, gives  兹  兹2

冤冢

2 1 2 2 

冣 冥 1/2

1

1/2

(3.29)

3.2 Electromagnetic radiation

121

and  兹  兹2

冤冢

2 1 2 2 

冣 冥 1/2

1

1/2

.

(3.30)

The units of  and  are nepers m1 and radians m1, respectively. If, in (3.28),  /, the medium approximates a pure dielectric. If  /, the medium approximates a conductor. There is a cross-over frequency given by   / .

(3.31)

Evanescent waves Going back to Equation (3.23), it is possible for the refractive index to be purely imaginary, so that nj , and then EE0 exp( jt)exp( z/c).

(3.32)

This is an evanescent wave, which extends into the medium by about c/( ) but does not propagate because its phase does not vary with distance. When a propagating wave is totally reflected at the interface between two media, an evanescent wave exists just inside the second medium.

3.2.4

Interactions between radio waves and matter

The basic interactions are reflection, refraction, dispersion, diffraction, scattering, change of polarization, and attenuation; and these – singly or in combination – are the processes which underlie the various phenomena of terrestrial radio propagation. They have also provided us with a number of well-proven techniques for the investigation of the propagation media and their behavior, knowledge of which is essential to the understanding of radio communication and its problems. Reflection occurs at the boundary between two media, returning energy back towards the source in the case of normal incidence, whereas refraction causes any transmitted ray to emerge at an angle to the incident ray. These effects are discussed in the context of ionospheric reflection in Section 3.4.3, and of the partial reflection technique in Section 4.2.4. Dispersion, the variation of velocity with frequency, has consequences for the transmission of information (Section 3.4.1). Diffraction phenomena occur when there are irregularities in the propagating wavefront, causing the wavefront to evolve as the wave travels on. It is the basis of radio scintillation (Section 3.4.5). Scattering from structures in the medium that are small relative to the wavelength of the incident wave diverts some fraction of the incident signal over a wide range of directions. It is the basis of communication over scatter links (Section

Fundamentals of propagation

122

8.5). Also, a (usually weak) echo may be detected at the transmitter site, which is utilized in the techniques of coherent and incoherent scatter radars described in Section 3.5 and in Sections 4.2.2 and 4.2.3. Polarization changes occur in an ionized medium in the presence of a geomagnetic field. There are consequences for the design of transmitting and receiving antennas and polarization may be exploited in ionospheric measurements (Sections 3.4.1 and 3.4.4). Attenuation is, of course, undesirable in communications, often setting the lower limit to the usable frequency band. Measurements of absorption may give useful information, particularly about the lower ionosphere (Sections 3.4.4 and 4.2.4).

3.3

Propagation through the neutral atmosphere

3.3.1

The refractivity of the neutral atmosphere

Although this book is primarily concerned with the high latitude ionosphere and its effects upon radio propagation, there are some tropospheric effects peculiar to high latitudes that affect radio propagation in the line-of-sight (LOS) and earth-to-satellite modes. For that reason, we will briefly discuss some of the fundamentals of these modes. We will exclude the troposcatter propagation mode in which forward scatter in the troposphere (3–8 km height) permits communication over path-lengths from ⬃300 to 600 km, using frequencies from 200 MHz to 10 GHz (see Norton and Wiesner, 1955; and Collin, 1985). Radio waves propagating in the troposphere are affected by the refractive index, n – which is a function of atmospheric pressure, temperature, and humidity, and, near the Earth’s surface at VHF/UHF, n is approximately 1.0003. It is convenient to define a radio refractive index, N, as N(n 1) 106.

(3.33)

Since the terrestrial atmosphere varies exponentially with height, we may express it as N(h)Ns exp(ch),

(3.34)

where Ns is the surface refractivity, h is the height above the surface in kilometers, Cln(N/Ns) N, and N is the difference between the values of N at a height of 1 km above the surface and at the surface. Ns may be estimated from N7.32exp(0.005577Ns).

(3.35)

A useful parameter, the effective Earth-radius (the actual Earth-radius corrected for “normal” atmospheric refraction) for radio propagation is given by

3.3 Through the neutral atmosphere

123

Table 3.3. CRPL exponential radiorefractivity atmospheres, NNs exp(ch) Ns

N

K

C

200 250 289 300 320 350 400 450

22.33177 29.33177 36.68483 39.00579 43.60342 51.55041 68.12950 90.010 56

1.17769 1.25016 1.33324 1.36280 1.42587 1.55105 1.90766 2.77761

0.118399 0.125626 0.135747 0.139284 0.146502 0.159332 0.186719 0.223256

冢

r0 K 1  cNs 106 ns

1

冣

(3.36)

where ns 1Ns 106, r0 is the Earth-radius6373.02 km, Ns 289, and c0.136, so K1.3332410, or 4/3. The basic exponential reference atmosphere is defined by the relation N(h)289exp(0.136h),

(3.37)

where h is the height above the surface in kilometers. Table 3.3 shows the CRPL (the old Central Radio Propagation Lab – now the Institute for Telecommunication Sciences (ITS) – in Boulder, Colorado) exponential radio refractivity atmosphere. The standard model of the atmosphere is obtained by assuming that N decreases linearly over the first kilometer above the surface: NNs N(hhs );

hs h(hs 1),

(3.38)

where N is from Equation (3.34), h is the height above sea level, hs is the height of the surface above mean sea level in kilometers and N is the difference between Ns and N 1 km above the Earth’s surface. The constants adopted for the standard atmosphere are given in Table 3.4. N can be calculated from radiosonde data: N77.6P/T3.73 105e/T 2 “dry term”“wet term”,

(3.39)

where P is the atmospheric pressure in millibars, e is the vapor pressure in millibars, and T is the temperature in kelvins. A set of “standard atmospheres” showing the height dependence of radio

Fundamentals of propagation

124

Table 3.4. Constants for the standard reference atmosphere Ns

hs (ft)

a (miles)

N

K

ae (miles)

c (km)

0 200 250 301 313 350 400 450

0 10000 5000 1000 900 0 0 0

3960.0000 3961.8939 2960.9470 3960.1894 3960.1324 3960.0000 3960.0000 3960.0000

0.3318 22.3318 29.5124 39.2320 41.9388 51.5530 68.1295 90.0406

1.00000 1.16599 1.23165 1.33327 1.36479 1.48905 1.76684 2.34506

3960.00 4619.53 4878.50 5280.00 5403.88 5896.66 6996.67 9286.44

0 0.106211 0.114559 0.118710 0.121796 0.130579 0.143848 0.154004

Note: ae is the effective Earth-radius and is equal to the product aK. a ahs, where hs is the height of the Earth’s surface above sea level. a3960 miles. c

冢 冣

N1 1 ln . 8  hs 105

refractivity as a function of its value at the surface, Ns, has been defined.. Near the ground the following empirical relationship between Ns and the difference in refractivity, N, between Ns and N at 1 km above the Earth’s surface is valid: N (km)7.32exp(0.005577Ns).

(3.40)

inverting Equation (3.37), we can obtain Ns as a function of the refractory gradient N: Ns 412.87log| N| 356.93.

(3.41)

Figures 3.2 and 3.3 show estimates of Ns for winter afternoons in the northern temperate zone and global variations. Charts similar to Figure 3.2 applicable for high latitudes may be obtained from the appropriate national meteorological departments. Radio-refractivity values at high latitudes are sometimes radically different from those in temperate zones. For example, Fairbanks, Alaska has some of the steepest temperature inversions in the world, causing anomalous refraction on some VHF/UHF radio paths. These effects will be described in Chapters 8 and 9. 3.3.2

Terrain effects

The most obvious feature of the Earth affecting terrestrial radiowave propagation is its curvature. The troposphere of the Earth refracts radiowaves on LOS paths

3.3 Through the neutral atmosphere

Figure 3.2. Minimum surface-refractivity values (Ns) referred to mean sea level for an average winter afternoon, continental U. S. A. (from Freeman, 1997).

in such a way that one can use a modified Earth-radius when planning these propagation paths and, from Section 3.3.1, we use the “4/3 Earth-curvature” as shown in Figure 3.4. Topographical features such as mountain ranges and deep valleys will, of course, also affect the propagation of radio waves – especially if they block off low takeoff angles for HF paths or if the ground-reflection areas of a skywave mode occur where are large topographic features. A “rule of thumb” is that the radio horizon should be no higher than about 5° in the desired direction of propagation for a long-haul HF skywave circuit. For LOS propagation, one usually takes advantage of mountains to site either active or passive repeaters for VHF through microwave frequencies. Theoretical calculations of antenna patterns usually assume that one has a perfectly conducting reflecting plane, when in reality the conductivity and permittivity of the Earth’s surface exhibit great variation – as illustrated in Table 3.5. The vertical radiation pattern of a practical antenna depends upon the electrical characteristics of the ground plane of the antenna. For antennas that use the Earth’s surface as their ground plane, in addition to the electrical properties of the Earth, the relative “smoothness” of the Earth is also important. The concept of the Fresnel zone is invaluable in calculating the relation of the propagation path to the terrain in the context of engineering the best path characteristics. Extensive treatments of Fresnel zones applied to radio propagation may be found in standard electrical engineering textbooks and Handbooks (see Jordan and Balmain, 1968,

125

Figure 3.3. Minimum monthly surface-refractivity values (Ns) referred to mean sea level.

3.3 Through the neutral atmosphere

Figure 3.4. Earth-curvature-correction curves for D1 from 0.5 to 7 miles (from Freeman, 1997).

pp. 498–503; Hall and Barclay, 1989, pp. 38–42 and Hunsucker, 1991, pp. 258–266). Several computer programs which treat terrain effects and LOS link performance have recently become available (Table 3.6). It should also be mentioned that certain atmospheric and ionospheric conditions could produce signals over the LOS distance. 3.3.3

Noise and interference

Electrical noise is one of the limiting factors in radio communication and must be considered in the design of communications circuits. The three components of electrical noise are cosmic noise, atmospheric noise, and manmade noise. There are extensive discussions of the noise figure and noise temperature of receivers, and cosmic, atmospheric, and manmade noise in Collin (1985), Kraus (1988),

127

Fundamentals of propagation

128

Spaulding and Washburn (1985), and CCIR Report 322; and a shorter description in Hunsucker (1991, pp. 15–20 and Appendix A5). Cosmic noise emanates from sources of extraterrestrial origin, such as our Sun, galactic radio sources, and the extragalactic sources, and its dependences on frequency and antenna-pointing direction are shown in Figure 3.5. From Figure 3.5 we see that cosmic radio noise decreases with increasing frequency and varies with the antenna-pointing direction. The terrestrial ionosphere acts as a “highpass” filter, attenuating or refracting cosmic noise in the ELF through low-HF bands. Solar radio noise varies in frequency, intensity, and time. An example of the behavior of the quiet Sun of large bursts, storms, and plages for frequencies from ⬃15 MHz to microwave frequencies is shown in Figure 3.6. Good representations of the dynamic behavior of solar radio burst frequency and intensity are shown in Figures 3.7 and 3.8. An example of radio noise from a galactic source is shown in Figure 3.9. Examples of the variation with frequency of some extragalactic radio sources are illustrated in Figure 3.10. Atmospheric noise originates in atmospheric electrical discharges like lightning and precipitation static, etc., and may reach the receiving antenna either by a LOS path or via propagation by the ionosphere. The most intense thunderstorms on

Table 3.5. Electrical conductivities and permitivities for various types of terrain

Type of surface

Conductivity, (1 m1)

Coastal dry sand Flat, wet coastal Rocky land (steep hills) Highly moist soil Marshy Hills (to ⬃1000 m) Freshwater Sea water Sea ice Polar ice (free) Polar ice (cap) Arctic land

0.002 0.01–0.02 0.002 0.005–0.02 0.1 0.001 0.001 3.0–5.0 0.001 0.000025 0.0001 0.0005–0.001

Tundra underlain by permafrost surfacea

⬃103 to 102

Permitivity ( ) (relative dielectric) constant ⬃10.0 ⬃14.0–30.0 ⬃10.0–15.0 ⬃30.0 ⬃30.0 ⬃15.0 ⬃80.0–81.0 ⬃80.0–81.0 ⬃14.0 ⬃13.0 ⬃11.0 ⬃23–34 for silts ⬃12 for dry sand ⬃5–70

Note: Acquired in 1988/1989 in Central Alaska from 2–30 MHz by G. Hagn of SRI International.

a

3.3 Through the neutral atmosphere

129

earth occur in the tropics and this HF noise is propagated by LOS modes and by the ionosphere to distances of thousands of kilometers. The areas of lowest propagated atmospheric noise are at high northern and southern latitudes (55° geographic latitude). Table 3.6. Computer programs for diffraction/terrain predictions Name

Description

Source

Reference

EREPS

Engineer’s Refractive Effects Prediction System

http://trout.nosc. mil/NraDMosaic Home.html

Patterson (1994), Proc. of the BLOS Conference

IFDG/GTD

*Finite Difference.../ Generalized Theory of Diffraction

GELTI/ATLM

GTD Estimated Loss due to Terrain Interaction/ Automated Terrain Linearization Model

HARPO

Hamiltonian equations in spherical coordinates, modified by using Gaussian beams

EFEPE/SSP IRT

Institut für Rundfunktechnik propagation model for digital broadcast systems in urban areas

Dr R. Großkopf Institut für Rundfunktechnik München

Ditto Grosskopf (1994)

VTRPE

Variable Terrain Radio Parabolic Equation microwave propagation in complex real-world environments

Dr Frank Ryan NCCOSC/RDT&E Division, San Diego, CA 92152-6435

Ryan (1991)

Anderson et al. (1993) Marcus (1994) Dr Kent Chamberlain Department of Electrical and Computer Engineering, University of New Hampshire, Durham, NH 03824-3591

Chamberlain and Luebbers (1992)

Brent and Ormsby (1994)

Fundamentals of propagation

130

Table 3.6. (cont.) Name

Description

Source

MSITE, TPATH, MCS

Two- and threedimensional plots of signal levels from multiple transmitters, microwave-link studies and interference prediction, raytracing for urban and indoor environments, wireless, etc.

EDX Engineering, Inc., P. O. Box 1547, Eugene, OR 97440 Ph. (541)345-0019 Fax (541)345-8145

Terrain Integrated Rough-Earth Model/ Ducting and Anomalous Propagation Environment

Dr Homer Riggins and Dr David Eppink, IIT Research Institute, 185 Admiral Cochrane Drive, Annapolis, MD

TIREM/ DUCTAPE

Reference

http://www.edx.com

Eppink and Kuebler (1994)

Several reports and papers deal with the global levels of atmospheric noise, the most cited being Spaulding and Washburn (1985) and the CCIR Report 322-3c (1988). Sailors (1993) has noted some major problems in CCIR Report 322-3, and concludes that “the model should be used with caution, especially in the northern and southern high latitudes, the Arabian Peninsula, northern Africa and the midAtlantic areas. In these areas, consider using the original CCIR Report 322 model.” He also suggests serious modifications to the development of the model and using correction factors for certain locations. Figures 3.11–3.15 give examples of atmospheric models and noise as a function of frequency. Manmade noise usually originates from rotating electrical machinery, highcurrent switching circuits, and arcing power-line components. It is obviously most intense in industrial areas and problems from this type of noise need to be resolved on a case-by-case basis as outlined in a report by Vincent and Munsch (1996). Interference from other transmitters sometimes dominates portions of the spectrum, such as the HF band – where frequency assignments seem to be largely ignored. Interference can be minimized by maintaining the frequency stability of

3.3 Through the neutral atmosphere

Figure 3.5. Variations of “antenna temperature” as a function of frequency from 10 MHz to 100 GHz (from Freeman, 1997).

the transmitter and maximizing the selectivity of the receiver and by making rather extensive interference measurements at the receiver site before finalizing the operating frequency and time slots. The basic theorems governing vertical and oblique HF propagation are given in the following section.

131

Figure 3.6. A typical radio spectrum from the Sun (after Hey, 1983).

Figure 3.7. Dynamic spectra of solar radio bursts (from Hey, 1983, p. 100).

Figure 3.8. The power variation of solar radio bursts (from Hey, 1983, p. 100).

Figure 3.9. The spectrum of radio sources in the Orion Nebula compared with a curve calculated for an electron temperature of 10 000 K (from Hey, 1983).

Figure 3.10. Spectra of radio galaxies Cygnus A, Virgo A, and Hercules A, compared with the supernova remnants in Cassiopeia (dashed curve) (from Hey, 1983).

Figure 3.11. Radio-noise-recording stations used to obtain data used to develop the original CCIR Report 322 (from Sailors, 1993).

Figure 3.12. A typical figure from CCIR Report 322 (from Sailors, 1993).

Figure 3.13. Radio-noise-recording locations (from Sailors, 1993).

3.3 Through the neutral atmosphere

Figure 3.14. Determination of the 1-MHz Fam value for Moscow for June, July and August; 1600–2000 UT. (from Spaulding and Washburn, 1985).

137

Figure 3.15. In (a) and (b) are shown examples of Spaulding and Washburn’s corrections to the CCIR Report 322 (from Spaulding and Washburn, 1985, p. 18).

Fundamentals of propagation

140

3.4

Ionospheric propagation

3.4.1

Magnetoionic theory The Appleton equation

For an ionized medium the refractive index is expressed by the Appleton equation. In its complete form this is a complicated expression using the dimensionless quantities X, Y, and Z, each of which is defined as a ratio between the wave frequency and a frequency characteristic of the medium. The latter are the plasma frequency, N [Ne2/ 0me ]1/2,

(3.42)

the gyrofrequency, B Be/me,

(3.43)

and the collision frequency, , where N is the electron concentration (usually called the electron density), e is the charge on the electron (taken to be positive), me is the mass of the electron, 0 is the permittivity of free space, and B is the magnetic flux density in the medium. The plasma frequency is the natural frequency of oscillation for electrostatic perturbations within the plasma, the gyrofrequency is the frequency of gyration of an electron in magnetic flux density B, and  is the rate of collision between a given electron and other particles. Then the dimensionless quantities are X N2 /2,

(3.44)

Y B /,

(3.45)

Z  /.

(3.46)

and

In these terms the Appleton equation for the refractive index (n) of an ionized medium with N electrons cm3, permeated by a magnetic flux density B (W m1) and in which the electron-collision frequency is  (s1) is given by n2 1

X Y T2 Y 4T 1  jZ   Y 2L  2(1  X  jZ) 4(1  X  jZ) 2

冢

冣

1/2

,

(3.47)

where  denotes the ordinary and  the extraordinary wave. In (3.47), Y has been divided into longitudinal and transverse components;

3.4 Ionospheric propagation

141

Figure 3.16. The electric-field polarization in the plane of the wavefront. Ox and Oy are the principal directions and the projection of the imposed magnetic field is along Oy. The positive wave-normal is directed into the paper, along positive Oz. The ordinary-wave ellipse is shown as a continuous line and the extraordinary-wave ellipse is shown as a dashed line (from Ratcliffe, 1959).

YL Y cos 

(3.48a)

YT Ysin ,

(3.48b)

and

 being the angle between the direction of propagation and the magnetic field. Note that the refractive index is complex, with real and imaginary parts: n  j .

Polarization In order to calculate the effects of this anisotropic medium on the polarization of a radio wave traversing the region, it is convenient to define the polarization ratio R as RHy /Hx Ex /Ey ,

(3.49)

where Hy and Ey are the y-components of E and H, and Hx and Ex are the xcomponents of E and H, respectively. Then we can obtain the magnetoionic polarization equation (see Kelso, 1964; and Ratcliffe, 1959) R

冤

冢

j Y 2T Y 4T " Y 2L YL 2(1  X  jZ) 4(1  X  jZ) 2

冣 冥 1/2

.

(3.50)

The polarization equation gives values of R that are complex. In general, this means an elliptical polarization. If R is purely real, the polarization is linear; if R is purely imaginary, the polarization is circular. See Figure 3.16.

142

Fundamentals of propagation

It is virtually impossible for an ordinary mortal to make much sense of Equations (3.47) and (3.50) in their full glory – see Ratcliffe (1959) or Budden (1985) for a full discussion – but when special cases are taken the picture begins to clarify. Luckily, many applications can be treated using these special cases.

Special case 1: Neglecting collisions and magnetic field If there are no collisions and the magnetic field is neglected, the refractive index, n, is real: n2 1X1  N2 /2 n2 1Ne2/( 0me2).

(3.51)

Then the phase velocity is vp c/nc

/冢

1

Ne2 2 0me

冣

1/2

.

(3.52)

The group velocity, using Equation 3.21, is

冢

Ne2 uc/ng c 1  2 0me

冣

1/2

,

(3.53)

where ng is the group refractive index. (Note that ng 1/n in this case.)

Special case 2: The effect of a magnetic field If the magnetic field is now included and the propagation is almost directly along the magnetic vector so that YT may be neglected, then n2 1X/(1 YL ) n2 1 N2 /[(  L )

(3.54)

and Rj. There are now two waves, circularly polarized in opposite directions, having different velocities. These are characteristic waves, termed ordinary and extraordinary (for the upper and lower signs, respectively) by analogy with birefringence in crystals. In general, where YT # 0, the characteristic waves are elliptically polarized.

Special case 3: The effect of collisions If collisions are significant (but in the absence of a magnetic field), then n2 1X/(1jZ) n2 1 N2 /[( j)].

(3.55)

3.4 Ionospheric propagation

143

Taking the imaginary part ( ) and applying Equation (3.25) gives the absorption coefficient !

 1 XZ c 2 1  Z 2

!

e2 1 N  . 2 0mec  2   2

(3.56)

The refractive index (n) is modified by collisions between the electrons and heavy particles, and the wave undergoes absorption – which physically is due to the conversion of ordered momentum into random motion of the particles after collision. At each collision, some energy is transferred from the wave to the neutral molecules and appears as thermal energy. Details of the microscopic processes involved in ionospheric absorption are discussed by Ratcliffe (1959, Ch. 5) and derivations of the equations describing macroscopic features of absorption are given by Davies (1969, Ch. 6). We can conveniently divide absorption into two limiting types, commonly called non-deviative absorption and deviative absorption. Non-deviative absorption occurs in regions where the product N is large and  ⬇1, and is characterized by the absorption of LF, MF, and HF waves in the D region. Deviative absorption, on the other hand, occurs near the top of the ray trajectory or anywhere else on the ray path where significant bending takes place (for small N and ). When the refractive index is ⬇1, we can write ! 4.6 102

N (dB km1). 2   2

(3.57)

We can further simplify Equation (3.57) for the VHF case, since 2 2 , as ! 4.6 102N/2 (dB km1).

(3.58)

In deviative absorption,   1, and !⬇

 (1  2 ) . 2c

(3.59)

Near a reflection level, 2  1, and then the preceding equation reduces to !⬇

 , 2c

(3.60)

where  is the group refractive index. One important case is for non-deviative absorption and the quasi-longitudinal (QL) approximation, when

Fundamentals of propagation

144

!⬇

N e2 . 2 0mc (   L ) 2   2

(3.61)

The absorption coefficient is therefore smaller for the ordinary than it is for the extraordinary wave. For a given value of the electron density, the absorption coefficient is a maximum at the level where     L.

(3.62)

The absorption of the extraordinary wave becomes very strong at the higher levels ( small) when the wave frequency is close to the gyrofrequency. 3.4.2

Reflection of radio waves from an ionospheric layer Reflection at vertical incidence

If a pulse of radio waves of frequency f /(2) enters an ionospheric layer at vertical incidence from below, it will travel at the group velocity (u). Neglecting the magnetic field, u is given by Equation (3.53) and u decreases as the electron density increases with altitude. Provided that the layer is sufficiently intense, a level where the group velocity is zero (and the phase velocity infinite) will eventually be reached, and here the energy is reflected. At this level the plasma frequency ( fN  N /(2)) equals the wave frequency ( f ) and N4 2 0me f N2 /e2.

(3.63)

Numerically, N (m3)1.24 1010[ f (MHz )]2.

(3.64)

Above this level the wave is evanescent (Equation (3.32)). The time required for the journey to the reflection point and back is t

2 c

h

冮

0

dz n

(3.65)

and the virtual height is ct h  2

h

冮 [1  ( f /f ) ] 0

dz n

2 1/2

.

(3.66)

The virtual height is the height calculated on the assumption that the signal traveled at the speed of light (in vacuo). In fact, since the pulse always travels more slowly in the layer, the virtual height is always greater than the true height. If the electron density at the layer maximum is Nmax, the greatest radio fre-

3.4 Ionospheric propagation

145

quency that may be reflected at vertical incidence is the critical frequency of the layer, fc, which is related to the maximum electron density by Nmax 1.24 1010f c2.

(3.67)

A good discussion of the solution of Abel’s equation (3.66) (Appleton, 1930) is given by Kelso (1964). In the general case (including the geomagnetic field), several numerical techniques have been employed successfully to invert the ionogram trace of the ordinary wave to give an equivalent monotonic electron-density profile (see the special issue of Radio Science, 1967). One of the most comprehensive of the numerical true height programs is the POLAN program developed by Titheridge (1985) and a discussion of this program is given by Davies (1990). In the real ionosphere, where the geomagnetic field has to be taken into account, there are two reflection conditions. The extraordinary wave is reflected where fN2 f( ffB )

(3.68)

and the ordinary wave where fN f.

(3.69)

The first reflection occurs according to the QL approximation, whereas the second relates to the quasi-transverse (QT) approximation. If fB fN, the difference between the two critical frequencies is fB /2, that for the extraordinary wave being the greater. 3.4.3

Relations between oblique and vertical incidence

When the signal is incident obliquely on the layer, the process by which it is returned to the ground can be appreciated as follows. Consider the ionospheric layer to be composed of a large number of thin, uniform slabs, whose electron density increases with altitude. If successive slabs have refractive indices n1 and n2, Snell’s law relates the angles of incidence (1) and refraction (2) by n1 sin 1 n2 sin 2.

(3.70)

Applying this law to each boundary in turn readily shows that, if a ray enters the ionosphere at incidence 0, its angle to the normal in a slab with refractive index nr is simply sin r sin 0/nr

(3.71)

146

Fundamentals of propagation

Figure 3.17. The geometry describing vertical and oblique ionospheric propagation (from Hunsucker, 1992).

(the refractive index below the layer being unity). The ray therefore travels horizontally when nr sin 0

(3.72)

and this is the reflection condition (magnetic field neglected) for an obliquely incident signal. The ray then returns to the ground by a similar path. The process is now one of bending rather than reflection at a boundary. Combining the two equations for fN in this section yields the secant law relating vertical and oblique propagation: fob fv sec 0

(3.73)

where fob and fv are the frequencies of signals reflected from the same true height when fob is incident at angle 0 and fv is incident vertically. In order to determine values of sec 0 and fob from vertical-incidence soundings (which measure the virtual height, h), we need the results of two more theorems. Breit and Tuve’s theorem states that the time taken to traverse the actual curved path TABCR in Figure 3.17 at the group velocity u equals the time necessary to travel over the straight-line path TER at the free-space velocity c. Referring to the geometry shown in Figure 3.17, we can write the expression t 

1 c

冮

dx TER sin  0

(3.74)

D c sin a0

t(TEER)/c.

(3.75)

3.4 Ionospheric propagation

147

Figure 3.18. Logarithmic ionospheric-transmission curves for a curved-Earth ionosphere (after Smith, 1939).

Martyn’s theorem states that, if fv and fob are the vertical and oblique frequencies respectively, reflected from the same true height (h), then the virtual height at which the frequency fob is reflected equals the height of the equivalent triangular path for the frequency fv. Referring to Figure 3.17 and defining the equivalent path at oblique incidence for frequency fob as Pob 2TE,

(3.76)

we obtain Pv cos 0 Pob 2DE

(3.77)

Martyn’s theorem may be written more concisely as hob hv.

(3.78)

Newbern Smith (1939) devised the set of logarithmic transmission curves parametric in range for curved Earth shown in Figure 3.18, which are sufficiently accurate for the distances shown. Details concerning the use of these curves to relate the parameters given in Equations (3.76)–(3.78) may be found in Davies (1969) and in the URSI Handbook of Ionogram Analysis (1972). 3.4.4

Trans-ionospheric propagation

If the radio frequency exceeds the critical frequency of the ionosphere, the signal is not reflected but continues out into space. Similarly, signals from beyond the

148

Fundamentals of propagation

ionosphere may be received at the ground if their frequencies are sufficiently high. However, these signals are not necessarily unaffected by the ionosphere: there can be significant and measurable effects on their phase, their polarization and their intensity. In each case the effect becomes weaker with increasing frequency, and in practice they are significant from the upper part of the HF band, through the VHF band, and into the lower part of the UHF band. Another common feature is that the effects are cumulative and the total depends on an integral along the propagation path.

Phase effects In the Appleton equation for the refractive index, let X1 (radio frequency large relative to the plasma frequency), YL YT 0 (geomagnetic field neglected), and Z0 (collisions neglected). Then the second term of Equation (3.47) is much less than unity, and we can write n1X/2 n1 Ne2/(2 0 me2).

(3.79)

Inserting values for the constants, and using f instead of , gives n1 40.30N (m3)/[ f (Hz)]2.

(3.80)

The refractive index is smaller than unity by an amount proportional to the electron density and inversely proportional to the square of the radio frequency. If a radio wave travels a distance dl in an ionized medium, i.e. dl/ wavelengths, its phase lags by 2 dl/ (2fndl/c) radians. Over a path l the advance of phase is therefore 

2 f 2 f l 2 40.30 n dl  N dl. c c cf

冮

冮

(3.81)

The first term is just the phase delay due to a wave of frequency f traveling a distance l at the speed of light. The second is a phase advance that arises because the refractive index is less than unity and the phase speed greater than c. This term is

冮

cumulative and simply proportional to the electron content, I Ndl, which is the number of electrons in a column of unit cross-section along the propagation path. Numerically, the phase advance due to the medium is  (8.45 107)I /f (radians).

f is in hertz and I in m2 .

(3.82)

3.4 Ionospheric propagation

149

Several significant applications follow. (a)

Since the phase advance depends upon the radio frequency, the electron content can be determined by comparing the effects on two frequencies transmitted coherently from, for example, a satellite.

(b)

Since the frequency is the rate of change of phase, another method is to observe the Doppler shift in the frequency of a signal received from a satellite passing overhead.

(c)

If a carrier of frequency fc is modulated at frequency fm , the phase of the modulation is changed by m 8.45 107( fm /f c2)I.

(3.83)

In this case the phase is retarded because the modulation travels at the group speed, which is less than the speed of light. (d)

Corresponding to this phase delay, the time delay of a pulse is t8.45 107I/(2f c2) (s).

(e)

(3.84)

If there is a gradient of electron content in a direction (x) perpendicular to the propagation direction, the ray is deviated. This is wedge refraction. The wave is deviated through an angle  [c/(2)](8.45 107/f 2 )I/x.

(3.85)

(In Equations (3.82)–(3.85) the constant 8.45 107 is given to three significant figures, therefore with an inaccuracy of 0.12%. To four figures, for more accurate work, the constant is 8.448 107. In Equation (3.81) the constant 40.30 is accurate to within 0.025%. To five figures this constant is 40.302).

The Faraday effect When the geomagnetic field is taken into account and propagation is almost along the field direction, there are two characteristic waves that travel at different speeds. These waves are circularly polarized in opposite directions, and their sum is a linear polarization. If the circularly polarized components make instantaneous angles O and E with respect to a reference direction, then the linear wave is at an angle (O  E )/2.

See Figure 3.19. Let O  E 0 at the source. Then, after a distance l in the medium, O 2f [tnOl/c]

(3.86)

Fundamentals of propagation

150

Figure 3.19. Addition of two circularly polarized waves to give a linear wave, as seen by a stationary observer looking along the geomagnetic field (from Hargreaves, 1992).

and E 2f [tnE l/c],

(3.87)

(f /c)(nO nE )l.

(3.88)

giving

At a sufficiently high frequency (e.g. 50 MHz) the gyrofrequency fB f, and then the ordinary and extraordinary refractive indices differ by nO nE XY( f 2N fB)/f 3,

(3.89)

giving 

1 f N2 fB l. 2c f 2

(3.90)

Therefore the polarization angle changes progressively as the wave travels through the ionized medium. On substituting values and allowing for varying electron density and magnetic field strength, we obtain 

8.448 107 f2



2.365 104 f2

冮 f Ndl L

冮 B Ndl, L

(3.91)

since fL 2.799 1010 BL, BL being in webers m2 . We have now moved to the QL approximation, to allow for propagation somewhat across the field. (In fact the QL approximation has wide application in the Faraday effect, being valid to a few degrees of normal to the field). As seen by an observer at the ground looking up, the polarization rotates anticlockwise in the northern hemisphere and clockwise

3.4 Ionospheric propagation

151

in the southern hemisphere, irrespective of the direction in which the wave is traveling. A recent extensive discussion of the Faraday effect is given by Yeh et al. (1999).

Absorption Equation (3.61) gives the absorption coefficient ! in the case of non-deviative absorption and the QL approximation. The signal amplitude falls by a factor of e over the distance 1/!. Provided that the radio frequency is considerably greater than the critical frequency of the layer, this formula applies to trans-ionospheric propagation through the whole of the ionosphere. Whereas ! is in units of nepers, it is usual to express signal loss in decibels (dB), defined by the ratio between initial (P1 ) and final (P2) powers: Absorption A (dB)10log10(P1/P2).

(3.92)

The neper and the decibel are related by 1 neper8.686 dB.

(3.93)

On putting in the appropriate values, Equation (3.92) gives A (dB)4.611 105

N

冮   (   ) dl 2

L

2

(3.94)

for the total absorption over the path. Since the collision frequency decreases sharply with altitude, most nondeviative absorption occurs in the lower ionosphere, and it is maximized when the terms in the denominator of Equation (3.94) are equal. If  is the larger term, the absorption varies as 1/ and therefore decreases at the lower levels. However, over most of the height range affected the second term dominates and then the total absorption is just proportional to the integral of N. Moreover, the gyrofrequency may be neglected if it is much smaller than the radio frequency, which is certainly the case at frequencies greater than about 30 MHz. Then Equation (3.94) simplifies to A (dB)

1.168 10 18 [ f (MHz)]2

冮 N dl

(3.95)

for the total absorption over the path. The limit to high-latitude communications is often set by the ionospheric absorption, and measuring the absorption of the cosmic radio noise is a valuable technique in high-latitude studies.

Fundamentals of propagation

152

3.4.5

Principles of radio scintillation Introduction

The phenomenon of scintillation, which appears principally in trans-ionospheric signals, is caused by relative phase shifts in the propagating wavefront and by subsequent diffraction. The phase shifts are a direct result of spatial irregularity in the medium and specifically in the electron content, to which they are related by Equation (3.82). It should be noted that, other things being equal, the irregular component of the electron content varies not linearly with the path length (slab thickness), but with its square root. According to Huygens’ principle, each part of a wavefront may be regarded as a source of secondary wavelets, whose superposition builds up the wavefront at a point further along. In diffraction theory this principle is applied to determine how the amplitude and phase of a received signal are affected by passage through a region of irregularities. Diffraction theory applies to “small” irregularities, the criterion for which is that there are at least several of them within the distance of the first Fresnel zone (see below).

Diffraction by a thin screen of weak irregularities and the concept of the angular spectrum The simplest case to treat is that of a thin, shallow, phase-changing screen. In this model the irregularities are assumed to lie in an infinitely thin layer, and to introduce small (1 radian) phase perturbations along the wavefront of a wave passing through it, as in Figure 3.20 The incident wave is planar (the source being located at infinity), but the emerging wavefront is irregular. To obtain the field at a point P in the observing plane OO, it is necessary to sum the contributions from each point of the emerging wavefront, EE. Since EE is irregular in phase, the field at OO will also be irregular, and in general both the phase and the amplitude are affected. Since the wavefield at the observing plane is made up from contributions from points all along the diffracting screen, it is clear that there is not necessarily a oneto-one relationship between the irregularities in the ionosphere and the wavefield at the ground. There are, nevertheless, some relationships of a statistical nature. The link between the properties of the screen and the variations observed at the ground is the angular spectrum of the waves leaving the screen. Just as a wave modulated in time may be expressed as a frequency spectrum that may be derived by a Fourier transformation, so a wave modulated in distance may be expressed by a spectrum in angle. The spectrum of periodicities in the screen, F(d ), is related by Fourier transformation to an angular spectrum of waves, f(sin ), where d is the spatial wavelength of irregularities and  is the angle of propagation measured from the normal. The same spectrum reaches the ground, though with the phase of each sine wave modified by the distance traveled, where it may be transformed back to a

3.4 Ionospheric propagation

153

E a

b

Incident wave Screen E' Emerging wave

x

O P

Observing O' plane

Phase Amplitude

Figure 3.20. Diffraction of a plane wave incident on a thin phase-changing screen.

wavefield in the observing plane. Both F and f are complex, and therefore full information may be obtained only by observing both the amplitude and the phase. The problem now is that, whereas it is easy to measure the amplitude of a received radio signal, it is much more difficult to measure the phase. However, it has been shown that, if the observations are made sufficiently far from the screen (and provided that the screen is shallow – i.e. the initial modulation is only small), the statistical properties of the amplitude and phase irregularities at the ground are the same as each other and the same as those at the screen. In that case, amplitude observations alone suffice to give the statistical properties of the ionospheric screen. It is unlikely that irregularities will be sinusoidal or have any other analytical form; they will more probably look like random noise. Such irregularities may be handled using the correlation function,
. If a(x) are the differences of a varying quantity A(x) from its mean A, and 2 is the variance of a [A(x)  A¯] 2 (the bars denoting averages over many values), the correlation function of A over the interval y is
(y) [a(x)a(x  y)]/ 2.

(3.96)

The correlation function may sometimes be assumed to have a Gaussian form,
(d )exp[d 2/(2d 20)

(3.97)

and in this case the angular power spectrum would also be Gaussian: P(sin )exp[sin2 /(2sin2 0)],

(3.98)

Fundamentals of propagation

154

(a) Correlation function 1

ρ

e –1/2

d0

d (b) Angular power spectrum 1

e –1/2

P

sin θ 0 sin θ

Figure 3.21. (a) Correlation function and (b) angular power spectrum for a random diffraction screen.

where sin 0  /(2d0).

(3.99)

Figure 3.21 illustrates the relationship between
and P in this case.

Fresnel-zone effects The distance between the screen and the observer is significant because the size of the Fresnel zones depends upon the distance as well as the wavelength. Recall that, by definition, the first Fresnel zone extends to the point where the distance to the observer exceeds the minimum distance by  /2, the resulting phase difference being 180°. Referring to Figure 3.20, if the overhead point is a, we can pick a point b such that Pb Pa   /4. If the screen alters the phase only, the signal at EE may be sketched as in Figure 3.22(a), where A is the unaffected signal and E is the perturbation due to the screen. At a point P on the observing plane, if the perturbation due to a alters the phase of the signal, that due to b will affect its amplitude because of the extra /4 traveled. The resulting signal might now look like Figure 3.22(b), with both phase and amplitude fluctuations involved. Since contributions may affect the amplitude only if they fall within the angular spectrum, it follows that (D/2)1/2  d0

(3.100)

3.4 Ionospheric propagation

155

(a) Signal at EE Total

αE

A (b) Signal at OO Total

α0 β0

A

Figure 3.22. Development of (a) phase and (b) amplitude perturbations from initial perturbations.

for amplitude scintillation to appear at an observing plane at distance D from a pure phase screen, the source being at infinity. This says that the signal received from a phase screen will contain both amplitude and phase perturbations if the observer is sufficiently far from the screen for the first Fresnel zone to contain several irregularities of typical size. At infinity, the fading power becomes equally divided: (A)/A ()  s ()/√2,

(3.101)

where (A) and () are the standard deviations of amplitude and phase. At a lesser distance there will be phase fluctuation, but the amplitude fluctuation will not be fully developed, and this is often the situation in practice. If the radio wavelength, , is 6 m and the irregular screen is 400 km away, the radius of the first Fresnel zone is 兹D 1.5 km. Many of the irregularities will be larger than that and therefore the amplitude fluctuations will not be fully developed. The properties of a phase screen are important because the ionosphere behaves as a phase screen in most cases, and the bulk motion of the irregularities causes the signal received at a fixed place to scintillate. If, by means of a specially devised experiment, it is possible to observe phase as well as amplitude scintillation, the Fresnel-zone effect can be investigated directly by comparing the spectra of phase and amplitude fluctuations. An example is shown in Figure 3.23 The irregularities in the ionosphere generally exhibit a power-law spectrum of form !P, where ! is the wave number (2 /d, in which d is the spatial wavelength of the irregularities). We may generally suppose that the phase screen in the ionosphere produces a pattern of amplitude and phase fluctuation over the ground that is related to the spectrum of the irregularities themselves, and that scintillations are observed because the pattern is moving across the observing point. It is by this means that the variation in distance is converted into a time variation. Since the conversion of phase to amplitude scintillation depends on the size of the irregularity, the low-frequency (arising from the large scale) end of the spectrum

156

Fundamentals of propagation

Figure 3.23. Spectra of the amplitude and phase recorded at 40 MHz from a geosynchronous satellite transmission. Power spectra are plotted on a log scale of relative values in decibels. The phase spectrum levels off due to detrending (at 3 103 Hz), but the turn in the amplitude spectrum marks the Fresnel frequency. (After W. J. Myers et al., J. Geophys. Res. 84, 2039 (1979), copyright by the American Geophysical Union.)

is attenuated. The attenuation operates at frequencies less than u/ 兹2D, where u is the velocity (it being assumed that all the irregularities move together) and 兹D is the radius of the first Fresnel zone. Since  is known and D may be assumed (to be approximately 350 km), u can be determined by this means. This effect is seen in the spectra of Figure 3.23. When spectra can be determined, one can therefore obtain further information about the irregularities and their motion, particularly if the “Fresnel frequency” can be identified. The above results are altered if the source is not at infinity (because the wavefront reaching the screen is then curved), and/or the phase screen introduces deep modulation, s() 1 radian (since that broadens the angular spectrum). There is an extensive body of literature on the theory of scintillations. Hargreaves (1992) gives further details at an introductory level. The basic theory and early work were reviewed by Ratcliffe (1956), and later developments by Yeh and Liu (1982).

3.4 Ionospheric propagation

157

Indices and simple statistics of scintillation The intensity of amplitude scintillation is usually expressed by using one of four indices (Briggs and Parkin, 1963). If A is the amplitude, A is the mean amplitude, P is the power, PA2, P is the mean power, aAA, and pPP, S1 |a|/A,

(3.102)

S2 (a2)1/2/A,

(3.103)

S3 | p|/P,

(3.104)

S4 ( p2 )1/2 /P.

(3.105)

These are all dimensionless. S1 is the mean deviation of the amplitude normalized by the mean amplitude, and S2 the root-mean-square deviation of the amplitude also divided by the mean amplitude. S3 is the mean deviation of the power normalized by the mean power, and S4 the root-mean-square deviation of the power, similarly normalized. Note that S3 and S4 are similar to S1 and S2 but are written in terms of power instead of amplitude. From this selection of indices, S4 is the most commonly used. It has been shown (Chytil, 1967) that the following approximate relations apply: S1 0.42S4, S2 0.52S4,

(3.106)

S3 0.78S4.

An example of weak scintillation is shown in the top three panels of Figure 3.24. The S4 values are 0.016, 0.076, and 0.54 at 360, 140, and 40 MHz, respectively, all of which are less than unity. The bottom panel of Figure 3.24 gives the amplitude spectra, normalized with respect to magnitude for easier comparison. The turnover points indicate Fresnel frequencies of 0.07, 0.045, and 0.025 Hz, respectively, varying approximately as the square root of the radio frequency. The fading spectrum varies as (fading frequency)3.5. For comparison, Figure 3.25 illustrates the appearance of records with deep scintillation. Here the S4 values are respectively 0.13, 0.54, and 1.42. The character of the record changes dramatically when the modulation becomes deep. In Figure 3.22b, the fading signal is represented as a steady component plus random in-phase and quadrature components. If the random components are small relative to the steady one, the amplitude of the total signal (A ) will fluctuate about the mean with a Gaussian distribution. At the other extreme, if the steady component is small relative to the random one, the amplitude distribution will be a “random walk” having the Rayleigh form. Between these extremes the family of Nakagami m-distributions (Nakagami, 1960) applies. Figure 3.26 illustrates

158

Fundamentals of propagation

Figure 3.24. Examples of amplitude scintillation at three frequencies from a geosynchronous satellite, and their spectra. (R. Umeki et al., J. Geophys. Res., 82, 2752 (1997b).)

3.4 Ionospheric propagation

Figure 3.25. Scintillations at 360, 140, and 40 MHz, showing the transition to deep fading: S4 0.13, 0.54, and 1.42. (R. Umeki et al., Radio Science, 12, 311 (1997a).)

amplitude distributions for various S4 values covering the range between the Gaussian (S4 0.1) and the Rayleigh. Figure 3.27 gives a range of phase distributions, all of which are, of course, symmetrical about zero, The m-distributions are characterized by a single parameter that can be related to S4 and to the standard deviation of the phase. 3.4.6

Propagation involving reflection from a sharp boundary and full-wave solutions Reflection at a boundary

The treatment of propagation outlined in the foregoing sections, which are based on the concept of the refractive index, assumes that the medium is uniform. Of course this is seldom the case, but in practice the assumption may be used provided that any variations are not too large over a distance of several wavelengths. Such a medium is said to be slowly varying. There are, however, situations in which

159

160

Fundamentals of propagation

Figure 3.26. Empirical amplitude distributions for a range of S4 values. (After R. K. Crane, Technical Note 1974–26, Lincoln Laboratory (1974).)

this is plainly not so, and then a different sort of treatment is required. If the medium changes significantly within a wavelength then we may use the physics of reflection at a sharp boundary, as at a partially reflecting mirror. If a wave is normally incident at a sharp boundary, the coefficients of reflection and transmission are determined by the condition that the tangential components of the E and H vectors must be continuous across the boundary Referring to Figure 3.28, where the subscripts i, t, and r mean incident, transmitted, and reflected, the wave being incident from below, Et EI Er

(3.107)

3.4 Ionospheric propagation

161

Figure 3.27. Empirical phase distributions for a range of S4 values. (After R. K. Crane, Technical Note 1974–26, Lincoln Laboratory (1974).)

and Ht Hi Hr,

(3.108)

the negative sign arising because the reflected wave propagates downward. In a non-magnetic medium, H/En /( 00)1/2

(3.109)

and, by substitution, the reflection coefficient
is given by
Er /EI (n2 n1)/(n2 n1).

The fraction of power reflected is (Er /Ei )2 .

(3.110)

Fundamentals of propagation

162

Figure 3.28. The continuity of electric and magnetic vectors at a sharp boundary.

When the wave is incident at an angle to the boundary a further condition must be applied, which is that the normal components of the electric and magnetic flux ( E and H ) are also continuous across the interface. One familiar result that follows is Snell’s Law: n1 sin I n2 sin t,

(3.111)

where i is the angle of incidence in the medium of refractive index n1, and t is the angle of the ray transmitted into medium n2 . We now consider two special cases. First, let the plane of polarization (by convention the direction of the electric field) be perpendicular to the plane of incidence. Then application of the continuity conditions gives
⬜ sin(i  t )/sin(i  t ),

(3.112)

or,
⬜ 

兹(n2 /n1 ) 2  sin2 i  cos i 兹(n2 /n1 ) 2  sin2 i  cos i

.

(3.113)

This is the first Fresnel equation for reflection. If the plane of polarization lies in the plane of incidence, the reflection coefficient is given by

tan(i  t )/tan(i  t ) 

(n2 /n1 ) 2cos i  兹(n2/n1 ) 2  sin2 i (n2 /n1 ) 2cos i  兹(n2/n1 ) 2  sin2 i

(3.114) .

(3.115)

This is the second Fresnel equation. When i  t 90°, tan(i  t )$, and then

0. This is the Brewster angle, given by tan B n2 /n1, where the reflection coefficient goes to zero if the E vector is in the plane of incidence – in practice, the wave is vertically polarized. The reflected wave is reversed in phase as the Brewster angle is crossed. There is no such effect if the wave is horizontally polarized.

3.4 Ionospheric propagation

163

At normal incidence Equations (3.113) and (3.115) both revert to (3.110). At grazing incidence, as i →90°,

→1, but
⬜→1, implying that there is a reversal of phase on reflection. These remarks apply to reflection at the interface between dielectrics, n1 and n2 being both real. If the reflector is a partial conductor, the Fresnel equations still apply but the refractive indices are now complex. In the general case reflection involves a change of phase as well as of amplitude. Provided that the conditions of a sharp boundary are satisfied and the appropriate refractive indices are used, the Fresnel formulae are of wide application throughout the electromagnetic spectrum.

Full-wave solutions There are (unfortunately) other cases in which the medium changes over a radio wavelength but the change is not sharp enough to count as a sharp boundary. In these cases the only approach is to develop a full-wave solution, which amounts to solving Maxwell’s equations at each step through the layer by a numerical method. Conditions are imposed above and below the spatially varying medium to correspond to incident waves, and then the transmitted and reflected waves may be deduced. Though the method is applicable generally, preference would obviously be given to the simpler ones where they are valid. For more information about this technique the reader is referred to Budden (1985).

Sub-ionospheric propagation at ELF and VLF At frequencies below about 30 kHz the base of the ionosphere is only a few wavelengths above the ground, and across the boundary the ionosphere alters greatly within a wavelength. The propagation may now be considered in terms of reflection at a sharp boundary. At oblique incidence the loss on reflection is relatively small, and in consequence these signals may propagate over great distances with an attenuation amounting to only 2–3 dB per 1000 km. They exhibit some interesting properties, one being that (except at high latitude) the diurnal variation is more predictable than it is at higher frequencies, which makes them particularly suitable for those applications, such as navigation and time transmission, which require high stability. In the lower ionosphere the collision frequency (2 106 s1 at 70 km height) is greater than the wave frequency at VLF. Neglecting the magnetic field, Equation (3.55) then gives the refractive index (n) as n2 |1jX/2 |1  N2 /( j).

(3.116)

The ionosphere now behaves as a metal rather than a dielectric, having conductivity ( 0N2 )/ Ne2/(me),

where  is the collision frequency.

(3.117)

164

Fundamentals of propagation

Ionosphere

p 2-ho

p ho 1-

ground wave

T

Ground

R

Figure 3.29. Propagation in terms of a ground wave and two skywaves.

Studies of the amplitude and phase of VLF signals received from transmitters at various distances indicate the effective reflection height (about 70 km by day) and the ionospheric conductivity. Reflection coefficients are typically 0.2–0.5 There are actually four reflection coefficients because the presence of the geomagnetic field causes changes of polarization on reflection as well as of amplitude and phase. Putting typical values into the criterion of Equation (3.31) confirms that, at VLF and ELF, the lower ionosphere behaves as a conductor. Then, inserting the condition [ /( )]2 1 into Equation (3.29) leads to a skin depth (at which the amplitude falls by a factor of 1/e) of 1/  兹 0c/(  ) .

(3.118)

The skin depth varies as the square root of the wavelength and inversely as the square root of the conductivity. The ground is also a partial conductor, and, even in sea water, the most highly conducting part of the Earth’s surface, there is sufficient penetration to permit VLF and ELF communication with submerged submarines. Over distances up to several hundred kilometers, VLF propagation can be treated by summing the ground wave and the first few hops (Figure 3.29). This is the basis of geometrical-optical, or ray, theory. For long-distance propagation, one must resort to waveguide theory as developed by Budden (1961) and Wait (1970) and illustrated in Figures 3.30 and 3.31. This waveguide treatment is applicable because both the Earth and the ionosphere are partial conductors separated by a few wavelengths. In Figure 3.30 one assumes that the signal at a point consists of component wavelets emanating from images of the source. For long-distance VLF propagation the ionosphere behaves approximately like a conductor with a reflection coefficient of 1 and the ground has a reflection coefficient of 1. As in Figure 3.30, the images are located at z 2h, 4h, . . . ,

3.4 Ionospheric propagation

165

Figure 3.30. Using the method of images to construct one of the pair of waves that will interfere to produce the field patterns in the waveguide, such as those shown in Figure 3.31. The second wave (not shown) comes from the negative side. (After Davies, 1990.)

Figure 3.31. An idealization of the E field in the Earth–ionosphere waveguide for waves polarized with their electric fields in the vertical plane and their magnetic fields transverse to the plane of propagation (TM02 mode) (from Davies, 1990).

166

Fundamentals of propagation

but now they alternate in sign, which is equivalent to a change in phase of , and resonance occurs at 2hCn  (n  21 ),

(3.119)

where Cn is n /(2h) and n1, 2, . . . . In Figures 3.30 and 3.31 there is no zeroth-order mode and the horizontal wavelength g   /Sn is given by (1/g )2 (1/)2 [n/(2h)]2,

(3.120)

where Sn is the Fresnel coefficient showing that, for  2h/n, g is imaginary and hence the mode is evanescent. Thus there is a minimum cutoff frequency, fn, below which waves will not propagate, where fn nc/(2h). The cutoff frequency for the first-order mode during daytime, when the height of the ionospheric D region is low, ⬇2 kHz. For the case of a conducting ionosphere, the cutoff frequency is given by fn  (n  12 )c/(2h).

(3.121)

So we see that the change of reflection coefficient R from 1 to 1 changes the cutoff frequency (for n1) from about 2 Hz to 1 kHz. When they are being compared with waveguide modes with perfectly conducting walls, the ideal Earth–ionosphere modes should be denoted by n  12, rather than by n. A more complete analysis of the VLF waveguide mode must include Earth–ionosphere irregularities, changes in the height of the ionosphere, the effect of the geomagnetic field and collisions of electrons. There is a voluminous literature on VLF propagation (see Budden, 1985, references; and Davies, 1990, pp. 371–379). The main natural sources at ELF are lightning discharges, and the actual use of ELF for propagating signals is quite limited because of the practical constraints on constructing antennas several thousand meters long. Another important feature of ELF propagation is that the distance between source and receiver may be comparable to the wavelength (for example, a 300-Hz ELF signal has a wavelength of  1000 km. At these extremely low frequencies, the ionosphere behaves more like a conductor than a conducting dielectric and the displacement current is small. Because of this large skin depth (see Ramo et al. 1965, pp. 249–299) in the D region for ELF, the reflection height is ⬇90 km. As an actual example, the U. S. Navy’s Wisconsin Test Facility (WTF) radiates frequencies in the 40–50 and 70–80 Hz ranges. At the WTF the antennas are two 22.5-km quasi-orthogonal antennas. At middle latitudes the attenuation rate at 75 Hz is about 1.2 dB Mm1 during the day and 0.8 dB Mm1 at night. ELF propagation is discussed in considerable detail in the June 1974 Proceedings of the IEEE. Anomalously strong ELF signals have also been received at antipodal regions (Fraser-Smith and Bannister, 1997).

3.4 Ionospheric propagation

167

Partial reflections at MF and HF Turbulence in the lower ionosphere, at heights up to about 100 km, produces spatial irregularities on a scale sufficiently fine that partial reflections may be detected from them in the band 2–6 MHz (wavelengths 1.5 km to 500 m). In this case the reflections are very weak, a mere 103 to 105 of the amplitude of a total reflection, but they may be observed using a transmitter of high power and a large antenna array for transmission and reception. Although they are not useful for communications, these partial reflections may be exploited in a technique for measuring the electron-density profile of the lower ionosphere. 3.4.7

Whistlers

Whistlers are bursts of electromagnetic radiation in the VLF range that are produced by lightning discharges. These bursts travel through the ionosphere and magnetosphere in ducts approximately parallel to lines of force in the geomagnetic field and can be detected using low noise amplifiers with short antennas. Since about 1951 these signals have been studied scientifically for the information they reveal about the ionospheric and magnetospheric plasma. Other natural VLF emissions (called dawn chorus, risers, hiss, etc.) which are thought to originate in the ionosphere can also be heard on whistler detection equipment. Some of the fascination with the whistler phenomenon is due to the fact that it is a remarkable sound in the audio range, resembling a human whistle, that can be heard on sensitive audio equipment and on telephone lines under certain circumstances. The history of the scientific study of whistlers is covered by Eckersley (1925, 1928, 1929, 1931, and 1932), Helliwell (1965, 1988), Davies (1990), Hunsucker (1991), and in reviews by Park and Carpenter (1978) and Carpenter (1988). The starting points for whistler theory are Appleton’s equations for dispersion and polarization and the QL approximation. Figure 3.32 is a simplified presentation of basic whistler signatures obtained near the source and near the conjugate area of the source (i.e. the other end of the field line). Another basic feature (not always present on a signature) is the “nose” (Helliwell, 1965) illustrated in Figure 3.33. Helliwell (1965) showed that energy flow in the whistler will be guided along ducts in the geophysical magnetoplasma according to the following relation: tan(  )⬇(0.5tan )/(1 12 tan2 ),

(3.122)

where  is the angle between the ray path of the whistler and the wave normal,  is the propagation angle limited by 0   max, and fH cos max f, where fH is the electron gyrofrequency. Another important characteristic of a whistler wave packet is that the group velocity, vg, is vg 2c[ f 1/2(| fL |f )]/| fL | fN

(3.123)

168

Fundamentals of propagation

Figure 3.32. A sketch of the basic manifestation of a whistler and its initiating disturbances: (a) illustrating the dispersion; (b) the frequency–time curve of a typical whistler; (c) the curve of √f with time, and (d) the initiating disturbance and multiple hops when the source and receiver are at the same end of a geomagnetic field-line (from Helliwell, 1965).

3.5 Ionospheric scatter

169

Figure 3.33. An idealized sketch of the frequency-versus-time characteristics of a “nose whistler” (from Helliwell, 1965).

where fL is the longitudinal component of fB and fL f, Equation (3.123) simplifies to vg 2c( f 1/2 fL1/2 )/fN.

(3.124)

The dispersion law for whistlers is T

1 2c

冮f s

f ds/[ f 1/2 ( fL f )3/2],

N L

(3.125)

which can be applied to determine 兰Ndl along the field line. 3.5

Ionospheric scatter

One can qualitatively describe ionospheric scattering as either strong or weak in terms of the received signal strength of the scattered wave at the receiving radar antenna. An example of the former is VHF/UHF backscatter echoes received from electron density gradients in the auroral or equatorial ionosphere, and an example of the latter is incoherent backscatter by a VHF/UHF radar from the undisturbed E or F layer. Another way of classifying scattered echoes is in terms of their backscatter cross-section ( , in m2 ) using pulsed radar systems, and their temporal stability. A coherent echo exhibits a statistical correlation of the amplitude and phase from one pulse to another, and emanates from quasi-deterministic gradients in electron density that have correlation times greater than 1 ms, which corresponds to a spectral width of the radar echo of less than 1000 Hz (sometimes less than 100 Hz). It also has a backscatter cross-section 104–109 times greater than that from an incoherent-scatter radar echo. 3.5.1

Coherent scatter

Other important considerations in the case of coherent backscatter are the relation between the size of the scattering irregularity relative to the free-space wavelength

170

Fundamentals of propagation

Figure 3.34. Height–frequency regimes of various ionospheric radar probes (from Schlegel, 1984).

Figure 3.35. The geometry for scatter from ionospheric irregularities.

of the backscatter sounder, the mean fractional deviation in electron density of the scatterer, and the aspect angle between the radar LOS and the major axis of the irregularity. Figure 3.34 shows the approximate height–frequency domains of typical ionospheric sounding systems. The first quantitative description of coherent scatter from ionospheric irregularities was published by Booker (1956) (an extension of the Booker– Gordon (1950) troposcatter theory), when he developed a theory that described backscatter from field-aligned irregularities in the auroral E region. The results are also applicable to backscatter from F-region irregularities. The geometry of scatter from an ionospheric irregularity is shown in Figure 3.35. From the geometry in Figure 3.35, we can obtain one form of the Booker

3.5 Ionospheric scatter

171

ionospheric-irregularity scatter equation expressed in terms related to ionospheric parameters as (, )(N/N)2(2L/N)2 sin2 /{N [1(4L/)2 sin2( /2)]},

(3.126)

where (, ) is the backscatter cross-section of the irregularity, (N/N)2 is the mean square fractional deviation in electron density, N is the wavelength of plasma oscillation, and L is the scale size of the irregularity along B. Relations for the backscatter cross-section in the cases of large and small irregularities are derived in Hunsucker (1991, p. 56). Walker et al. (1987) started with the Booker scattering equation and derived a more general expression for the backscattered power at the receiver (see also Hunsucker, 1991, pp. 56-58). 3.5.2

Forward scatter

Irregularities due to turbulence in the 75–90-km regions of the ionosphere permit one to design one-hop communication circuits at VHF (Bailey et al., 1955; Norton and Wiesner, 1955). The ionoscatter mode typically uses frequencies from 30 to 60 MHz, over distances of 1000–2000 km with system losses of 140–210 dB and a usable bandwidth of ⬇10 kHz. Because of the high system loss, very-high-power transmitters, large high-gain antennas and sensitive receiver front ends are required. Ionoscatter systems are also characterized by very high reliability and security, but use of this bandwidth probably involves the highest cost per system of all radio systems. In the late 1950s and early 1960s considerable use of the ionoscatter mode was made because of its 99.9% reliability and security, but, with the advent of satellite–Earth radio systems, use of the ionoscatter mode decreased drastically. 3.5.3

Incoherent scatter

The development of the incoherent-scatter radar technique has provided a very powerful method for investigating the ionosphere. Evans (1969 and 1972) summarised the essentials of incoherent-scatter theory and practice and rigorous derivations of the salient equations are given by Krall and Trivelpiece (1973). The basic theory of scattering of electromagnetic waves from free electrons was developed by the discoverer of the electron, J. J. Thomson, who in 1906 showed that the energy scattered by a single electron is W(re sin )2,

(3.127)

where W is the energy scattered by a single electron into unit solid angle per unit of incident electromagnetic flux (1 W m2 ); re is the classical electron radius, re e2/( 0mec2 )2.82 1015 m; and  is the angle between the direction of the incident electric field and the direction of the observer.

172

Fundamentals of propagation

The radar cross-section of an individual electron would then be e 4 (re sin )2 ⬇1028 sin2  (m2 )

and, for backscatter (   /2), e 4 r e2.

(3.128)

Fejer (1960) showed that the radar cross-section per unit volume is simply N e,

(3.129)

where N is the electron density, and Buneman (1962) showed that the incoherent scatter effective radar cross-section ( eff ) can be written as eff 1/[(1 2 )(1Te /TI  2 )]

(3.130)

for Te /TI 3.0, and Te is the electron temperature, TI is the ion temperature,  4D/, where D is the Debye length, D6.9(Te /Ne )1/2, in centimeters, and  is the free-space wavelength of the radar signal. Since the electrons are in random thermal motion, they will scatter signals whose phases are varying with time and are not related to one another. At the radar-receiving antenna the signal powers will add so that, on the average, the crosssection per unit volume is that given by Equation (3.129), giving use to the name “incoherent scatter”. The interesting history of the development of incoherent-scatter theory and practice starting shortly after the end of WWII has been described by Davies (1990, pp. 106–111) and by Hunsucker (1991, pp. 58–64). Dougherty and Farley (1960) explained the discrepancy between the predicted and measured Doppler broadening of the echo spectrum in terms of the radar wavelength, electron and ion temperatures, and the Debye length, D69(Te /Ne )1/2 (m),

(3.131)

Where T is in kelvins and N in m3. Incoherent scattering occurs from fluctuations in electron density having a scale of D. The backscatter, then, is actually due to local fluctuations in electron density, instead of purely scatter from free electrons, and more correctly should be called something like quasi-incoherent scatter, but the term incoherent scatter has persisted. Some authors continue to refer to the incoherent-scatter phenomena as Thomson scatter for a variety of reasons; however, the term Thomson scatter is normally reserved for situations of scattering from free electrons without influence from ions. In practice, this incoherent scatter is detected from the ionosphere principally

3.5 Ionospheric scatter

173

Figure 3.36. An idealized sketch of the ISR spectrum.

when  D, although experiments at Arecibo have detected incoherent scatter when  ⬇D (Hagen and Behnke, 1976). The spatial scale of the irregularities p is given by the Bragg formula, p   /[2sin( /2)]

(3.132)

with the geometry as shown in Figure 3.35. The spectrum of an incoherent-scatter echo is very rich in information about the magnetoplasma which it is probing. A few of these plasma properties are easy to obtain, most requiring only straightforward data-analysis techniques, but some require complex processing using specific models of ionospheric regions. Figure 3.36 is an idealized sketch of the spectrum of an incoherent-scatter echo from the ionosphere, showing the ion line on the left and the plasma line on the right. The ion line is centered on the operating frequency, f, and the energy backscattered by irregularities of scale characterized by the Debye length that are in random motion, whereas the plasma line is centered on the plasma frequency fN and is due to the thermal motions of electrons not under the influence of ions. The plasma line is a weak line, except when it is enhanced by “hot” photoelectrons; and when the line is enhanced, both the electron density and the characteristics of the photoelectron flux may be measured. In the lower ionosphere the motions of the ions (which in turn control those of the electrons) are increasingly affected by collisions with the neutral air. The spectrum now becomes single-peaked, with width proportional to T/(mII2 ), where T is the temperature, mi and i are the mass and collision frequency of the ions, and  is the wavelength of the radar. If  70 cm and T230 K, the line is 1000 Hz wide at the height of 100 km, but, due to the increase of collision frequency, only a few hertz wide at 75 km.

Fundamentals of propagation

174

In this, the collision-dominated region, the returned spectrum has the Lorentz form, S( f )A/[1f 2 /(f )2 ].

(3.133)

(A is just a constant.) It is obviously much simpler than the F-region spectrum of Figure 3.44, and is fully described by its half-width: f16kT/(mii2).

(3.134)

A Doppler shift is superimposed if the scattering volume is moving towards or away from the radar. The spectrum is somewhat broadened if negative ions are present. Equation (3.134) also assumes that the ion and electron temperatures are equal.

3.6

HF-propagation-prediction programs

In the last two decades, over a dozen HF-propagation programs have been developed for use on personal computers. Some representative examples are listed in Table 3.7. It should be emphasized that all these programs input median-value data and produce median values of MUF, LUF, signal strength, etc. as output and are basically intended for HF-circuit planning, not real-time prediction. Most of the programs above take transmitter and receiver locations, time, month, year, and usually the number of sunspots as input, and provide MUF, LUF mode structure, antenna headings, great-circle distance and root-meansquare median-field-strength values for mid-latitude HF paths. The calculation of signal strength is especially difficult, because the exact mode structure on a particular path is not accurately known and all the path losses (in the D region, in the transmission line of the antenna, and from mismatch, ground reflection, etc.) are difficult to accurately characterize (see Sailors and Rose, 1991; and AGARDograph No. 326, 1990). Also, HF-propagation mode structure and losses at high latitudes are almost impossible to describe, so predictions of paths that include ionospheric reflections and points of D-region penetration in the auroral and polar ionosphere are almost useless (see Hunsucker, 1992; and discussions in Chapters 8 and 9 in this book). There are several books covering the essentials of antennas, radio propagation at all frequencies, and related topics, such as those by Jordan and Balmain (1968), Sanders and Reed (1986), Rao (1977), Stutzman and Thiele (1981), Kraus (1988), Collin (1985), Hall and Barclay (1989), Freeman (1997), Balmain (1997), Hansen (1998), and Kildal (2000). There are also several recent books covering all aspects of ionospheric radio propagation and magnetoionic theory, such as those by Maslin (1987), Davies (1990), McNamara (1991), and Goodman (1992).

3.7 Summary

3.7

175

Summary

It is, of course, impossible to cover the entire topic of radio propagation in one chapter, but we have attempted to list the essential elements of pertinent terrestrial propagation modes and of antenna systems. It is fortunate that there are recent books available, which describe in considerable detail the particulars of these modes (Budden, 1985; Hall and Barclay, 1989; Davies, 1990; Goodman, 1992; Freeman, 1997). A very significant new development is the availability of PC or workstation-based software to analyze antennas, terrain and propagation prediction, as listed in the tables of this chapter. Another new development is the availability on the internet/www of URLs, which give near-real-time data for

Table 3.7. Representative PC-based HF-propagation-prediction programs Name of program

Description

AMBCOM

Includes some effects of the highlatitude ionosphere

ASAPS 2 FTZMUF2 FTZ4 HFBC84 HFMUFES4 ICEPAC

foF2 and M3000 MUF–LUF (?) Improved calculation of several parameters

Includes some effects of the highlatitude ionosphere

IONOSOND MINIFTZ4

Field strength

MINIMUF

MUF, LUF

PROPHET PROPMAN VOACAP

MUF/LUF, signal strength, User-friendly shell for IONCAP

Source

References and remarks Hatfield (1980)

IPS (1991) Dambolt and Sussman (1988a, b)

Barghausen et al. (1969) Stewart (1990), private communication W1FM (Lexington, MA) Dambolt and Sussman (1988a, b) Rose (1982) Rose (1982) Roesler (1990) Lane (1993)

Note: MUF, maximum usable frequency; LUF, lowest usable frequency.

Fundamentals of propagation

176

radio-prediction purposes. One excellent example is the “space–weather”, magnetospheric, and ionospheric data bases available from the U. S. NOAA Space Environment Center (http://www.sec.noaa.gov).

3.8

References and bibliography

Section 3.2 ARRL (1999) The ARRL Antenna Book. The American Radio Relay League, Newington, Connecticut. ARRL (2000) The ARRL Handbook, 77th edition. The American Radio Relay League, Newington, Connecticut. Balanis, C. A. (1997) Antenna Theory, Analysis and Design. Wiley, New York. Hunsucker, R. D. (1991) Radio Techniques for Probing The Terrestrial Ionosphere. Springer-Verlag, Heidelberg. Skolnik, M. F. (1980) Introduction to Radar Systems, 2nd edition. McGraw-Hill, New York. Wolf, E. A. (1988) Antenna Analysis. Artech House, Norwood, MA.

Section 3.3 AGARDograph No. 326 (1990) Radio Wave Propagation Modeling, Prediction and Assessment, pp. 69–72. AGARD/NATO. Andersen, J. B., Hvid, J. T., and Toftgard, J. (1993) Comparison between different path loss prediction models, COST 231-TD(93)-06, January, Barcelona. Brent, R. I. and Ormsby, J. F. A. (1994) Electromagnetic propagation modeling in 3D environments using the Gaussian beam method. Joint Electronic Warfare Center Technical Report JDR 3-94. CCIR Report 322-3c (1988) Characteristics and applications of atmospheric noise data. XVth Plenary Assembly, Dubrovnik. International Telecommunications Union, Geneva. Chamberlain, K. and Luebbers. R. (1992) GELTI Propagation Model: Theory of Operation and Users’ Manual. Available through the authors. Collin, R. E. (1985) Antennas and Radiowave Propagation. McGraw-Hill Book Co., New York. Eppink, D. and Kuebler, W. (1994) TIREM/SEM Handbook. DoD ECAC, Annapolis, Maryland. Freeman, R. L. (1997) Radio System Design for Telecommunications. Wiley, New York. Grosskopf, R. (1994) Propagation of urban propagation loss. IEEE Trans. Antennas Propagation 42, 1–7. Hansen, R. C. (1998) Phased Array Antennas. Wiley, New York. Hey, H. S. (1983) The Radio Universe, 3rd Edition. Pergamon Press, Oxford. Hunsucker, R. D. (1992) Auroral and polar cap ionospheric effects on radio propagation. IEEE Trans. Antennas Propagation 40, 818–828.

3.8 References and bibliography

Jordan, E. C. and Balmain, K. G. (1968) Electromagnetic Waves and Radiating Systems, 2nd Edition. Prentice-Hall, Inc., Englewood Cliffs, New Jersey. Kraus, J. D. (1988) Antennas, 2nd Edition. Cygnus-Quasar Books, Powell, Ohio. Marcus, S. (1994) Duct propagation over a wedge-shaped hill, BLOS Proc. Applied Research Laboratory, University of Texas, Austin, Texas. Patterson, W. (1994) EM propagation program at NCCOSC, BLOS Proc. Applied Research Laboratory, University of Texas, Austin, Texas. Rao, N. N. (1977) Elements of Engineering Electromagnetics. Prentice-Hall, Inc., Englewood Cliffs, New Jersey. Ryan, F. J. (1991) Analysis of Electromagnetic Propagation Over Variable Terrain Using the Parabolic Wave Equation. Naval Ocean Systems Center, San Diego, California. Sailors, D. B. (1993) A Discrepancy in the CCIR Report #22-3 Radio Noise Model. NCCOSC/NRaD, San Diego, California. Sanders, K. F. and Reed, G. A. L. (1986) Transmission and Propagation of Electromagnetic Waves. Cambridge University Press, Cambridge. Spaulding, A. D. and Washburn, J. S. (1985) Atmospheric Radio Noise: Worldwide Levels and Other Characteristics. ITS, Boulder, Colorado. Vincent, W. R. and Munsch, G. F. (1996) Power-line Noise Mitigation Handbook for Naval Receiving Sites, 3rd Edition. COMMNAVSECGRU, Meade, Maryland.

Section 3.4 Appleton, E. V. (1930) Some notes on wireless methods of investigating the electrical structure of the upper atmosphere. Proc. Phys. Soc. 42, 321. Budden, K. G. (1961) Radio Waves in the Ionosphere. Cambridge University Press, Cambridge. Budden, K. G. (1985) The Propagation Of Radio Waves: The Theory of Radio Waves of Low Power in the Ionosphere and Magnetosphere. Cambridge University Press, Cambridge. Carpenter, D. L. (1988) Remote sensing of the magnetospheric plasma by means of whistler mode signals. Rev. Geophys. 26, 535–549. Crane. R. K. (1974) Morphology of ionospheric scintillation. Technical Note 1974–26, Lincoln Laboratory, MIT. Davies, K. (1969) Ionospheric Radio Waves. Blaisdell Publishing Co., Waltham, Massachusetts. Eckersley, T. L. (1925) Note on musical atmospheric disturbances. Phil. Mag. 49: (5), 1250–1259. Eckersley, T. L. (1928) Letter to the editor. Nature 122,768–769. Eckersley, T. L. (1929) An investigation of short waves. J. Inst. Electr. Engineers 67, 992–1032. Eckersley, T. L. (1931) 1929–1930 developments in the study of radio wave propagation. Marconi Rev. 5: 1–8. Eckersley, T. L. (1932) Studies in radio transmission. J. Inst. Electr. Engineers. 71, 434–443.

177

178

Fundamentals of propagation

Fraser-Smith, A. C. and Bannister, P. R. (1997) Reception of ELF signals at antipodal distances. Radio Sci. 32. Hargreaves, J. K. (1992) The Solar–Terrestrial Environment. Cambridge University Press, Cambridge. Helliwell, R. A. (1965) Whistlers and Related Ionospheric Phenomena. Stanford University Press, Stanford, California. Helliwell, R. A. (1988) VLF wave stimulation experiments in the magnetosphere for Siple Station, Antarctica. Rev. Geophys. 26, 551–578. Hunsucker, R. D. (1999) Electromagnetic Waves in the Ionosphere. In Wiley Encyclopedia of Electrical and Electronics Engineering (ed. J. Webster), pp. 494–506. Wiley, New York. Kelso, J. M. (1964) Radio Ray Propagation in the Ionosphere. McGraw-Hill, New York. Park, D. and Carpenter, D. (1978) Very low frequency radio waves in the magnetosphere. In Upper Atmospheric Research in Antarctica (ed. L. J. Lanzerotti and C. G. Parr). American Geophysical Union, Washington, DC. Radio Science (1967). Special issue on analysis of ionograms for electron density profiles. Radio Sci., 2, 1119–1282. Ratcliffe, J. A. (1956) Some aspects of diffraction theory and their application to the ionosphere. Rep. Prog. Phys., 19, 188. Ratcliffe, J. A. (1959) The Magneto-ionic Theory and its Application to the Ionosphere. A Monograph. Cambridge University Press, Cambridge. Smith, N. (1939) The relation of radio sky-wave transmission to ionosphere measurements. Proc. IRE 27, 332–347. Titheridge, J. E. (1985) Ionogram Analysis with the Generalized Program POLAN. World Data Center-A, NOAA, Boulder, Colorado. Umeki, R., Liu, C. H. and Yeh, K. C. (1997a) Multifrequency studies of ionospheric scintillations. Radio Science 12, 311. Umeki, R., Liu, C. H. and Yeh, K. C. (1997b) Multifrequency spectra of ionospheric amplitude scintillations. J. Geophys. Res 82, 2752. URSI (1972) URSI handbook on ionogram interpretation and reduction, 2nd Ed., NOAA WDC-A, Rep. UAG-23, Boulder, Colorado. Wait, J. R. (1970) Electromagnetic Waves in Stratified Media, 2nd Edition. Pergamon Press, New York. Yeh, K.-C., Chao, H. Y. and Lin, K. H. (1999) A study of the generalized Faraday effect in several media. Radio Sci. 34, 139. Yeh, K.-C. and Liu, C.-H. (1982) Radio wave scintillation in the ionosphere. Proc. IEEE 70, 324–360.

Section 3.5 Bailey, D. K., Bateman, R. and Kirby, R. C. (1955) Radio transmission at VHF by scattering and other processes in the lower in the lower ionosphere. Proc. IRE 43, 1181.

3.8 References and bibliography

Booker, H. G. (1956) A theory of scattering by nonisotropic irregularities with application to radar reflection from the aurora. J. Atmos. Terr. Phys. 8, 204–221. Booker, H. G. and Gordon, W. E. (1950) A theory of radio scattering in the troposphere. Proc. Inst. Radio Engineers 38, 401–402. Buneman, O. (1962) Scattering of radiation by the fluctuations in a non-equilibrium plasma. J. Geophys. Res. 67, 2050–2053. Dougherty, J. P. and Farley, D. T. (1960) A theory of incoherent scatter of radio waves by a plasma. Proc. R. Soc. A 259, 79. Evans, J. V. (1969) Theory and practice of ionospheric study by Thomson scatter radar. Proc. IEEE 57, 496. Evans, J. V. (1972) Ionospheric movements measured by incoherent scatter: A review. J. Atmos. Terr. Phys. 34, 175. Fejer, J. A. (1960) Scattering of radiowaves by an ionized gas in thermal equilibrium. J. Geophys. Res. 65, 2635. Hagen, J. B. and Behnke, R. A. (1976) Detection of the electron component of the spectrum in incoherent scatter of radio waves by the ionosphere. J. Geophys. Res. 81, 3441–3443. Krall, N. A and Trivelpiece, A. W. (1973) Principles of Plasma Physics. McGraw-Hill, New York. Nakajima, M. (1960) The m-distribution – A general formulation of intensity distribution of rapid fading. In Statistical Methods in Radio Propagation (ed. W. C. Hoffman). Oxford, Pergamon. Norton, K. A. and Wiesner, J. B. (1955) The scatter propagation issue. Proc. IRE 43, 1174. Schlegel, K. (1984) HF and VHF Coherent Radars for Investigation of the High-latitude Ionosphere. Max Planck Institut für Aeronomie, Katlenburg-Lindau. Walker, A. D. M, Greenwald, R. A., and Baker, K. D. (1987) Determination of the fluctuation level of ionospheric irregularities from radar backscatter measurements. Rad. Sci. 22: 689–705.

Section 3.6 Barghausen, A. F., Finney, J. W., Proctor, L. L. and Schultz, L. D. (1969) Predicting Long-term Operational Parameters of High Frequency Skywave Telecommunications Systems. ESSA, Boulder, Colorado. Damboldt, T. and Suessmann, P. (1988a) FTZ High Frequency Sky-wave Field Strength Prediction Method for Use on Home Computers. Forschungsinstitut der DBP beim FTZ. Damboldt, T. and Suessmann P. (1988b) A Simple Method of Estimating foF2 and M3000 with the Aid of a Home Computer. Forschungsinstitut der DBP beim FTZ. Davies, K. (1990) Ionospheric Radio. Peter Peregrinus, London. Hatfield, V. E. (1980) HF communications predictions, 1978. (An economical up-todate computer code, AMBCOM). In Solar–Terrestrial Predictions Proc. (ed. R. F. Donnelly), Vol. 4, D2 1–15. US Government Printing Office, Washington DC.

179

180

Fundamentals of propagation

Jordan, E. C. and Balmain, K. G. (1968) Electromagnetic Waves and Radiating Systems, 2nd Edition, Prentice-Hall, Englewood Cliffs, New Jersey. Lane, G. (1993) Voice of America Coverage Analysis Program (VOACAP). US Information Agency, Bureau of Broadcasting, Washington DC. Maslin, N. M. (1987) HF Communications: A Systems Approach. Plenum Press, New York. McNamara, L. F. (1991) The Ionosphere: Communications, Surveillance, and Direction Finding. Krieger Publishing Co., Malabar, Florida. Roesler, D. P. (1990) HF/VHF Propagation resource management using expert systems. In The Effect of the Ionosphere on Radiowave Signals and Systems Performance (IES90) (ed. J. M. Goodman), pp. 313–321. USGPO, available through NTIS, Springfield, Virginia. Rose, R. (1982) An emerging propagation prediction technology. In Effects of the Ionosphere on Radiowave Systems (IES81) (ed. J. Goodman). US Government Printing Office, Washington, DC. Sailors, D. B. and Rose, R. B. (1991) HF Sky Wave Field Strength Predictions. NCCOSC/NRaD, San Diego, California.

Section 3.7 Briggs, B. H. and Parkin, J. A. (1963) On the variation of radio star and satellite scintillation with zenith angle. J. Atmos. Terrestr. Phys. 25, 339. Goodman, J. (1992) HF Communications – Science and Technology. Van Nostrand Reinhold, New York. Hall, M. P. M. and Barclay, L. W. (eds.) (1989) Radiowave Propagation. Peter Peregrinus Press for the IEE, London. Nakajima, M. (1960) The m-distribution – A general formulation of intensity distribution of rapid fading. In Statistical Methods in Radio Propagation (ed. W. C. Hoffman). Oxford, Pergamon.

General reading Kildal, P.-S. (2000) Foundations of Antennas – A Unified Approach. Studentlitteratur, Lund. Ramo, S., Whinnery, S., and van Duzer, T. (1965) Fields and Waves in Communication Electronics. Wiley, New York.

Chapter 4 Radio techniques for probing the ionosphere

4.1

Introduction

The purpose of this chapter is to review the basic techniques (and the newer modifications and adaptations of these techniques) for studying the terrestrial ionosphere, with particular emphasis on the capabilities and limitations of the techniques when they are used to probe the high-latitude ionosphere. We are fortunate to have several books and reports written since 1989 that have addressed the general topic of ionospheric investigations using radio techniques (Kelley, 1989; Liu, 1989; Davies, 1990; Hunsucker, 1991; Hargreaves, 1992; Hunsucker, 1993 and 1999; pp. 502–505), so in this chapter we will emphasize the limitations and capabilities of these techniques and update the information on deployment of ionospheric instrumentation at high latitudes. Figure 3.34 of Chapter 3 shows the frequency–height regimes which various selected radio techniques can probe.

4.2

Ground-based systems

4.2.1

Ionosondes

In its simplest form, an ionosonde consists of a transmitter and receiver with coupled tuning circuits, which is swept in frequency (usually in the frequency range of approximately 0.5–25 MHz). It can be either a pulsed or a CW-FM (chirp) system, and the transmitter and receiver can either be co-located (monostatic) or separated (bistatic). After the RF signals have been reflected by the ionosphere they are received and processed by the receiver to produce ionograms. The

181

182

Radio techniques

Figure 4.1. NBS Model C-3 ionosonde installation. The power supply is on the left and the actual ionosonde is on the right.

basic information in the received signal is the transit time for passage between ionospheric layers and the Earth, frequency, amplitude, phase, polarization, Doppler shift, and spectrum shape (see Section 3.2.4). From these quantities, we can obtain an ionogram, which is a plot of the virtual height of reflection versus frequency. We can also deduce the true height of ionospheric layers as a function of frequency, the line-of-sight (LOS) velocity, some communication parameters, and the vector velocity of ionospheric irregularities (with an array of several antennas). Historically, the ionosonde was the instrument used to confirm the existence of the ionosphere by Appleton and Barnett (1926) and by Breit and Tuve (1926). A brief account of the development of the primitive and first-generation ionosonde is given in Sections 3.1 and 3.2 of Hunsucker (1991) and by Bibl (1998). The so-called “standard” ionosondes used vacuum tubes and electromechanical tuning mechanisms and were very bulky and heavy, as shown in Figure 4.1. A typical ionogram from a “standard” ionosonde in Yamagawa, Japan is shown in Figure 4.2, whereas an idealized ionogram is shown in Figure 4.3. These standard ionosondes were produced in relatively large numbers, and were deployed globally from c. 1942 until 1975. The photographically recorded data provided by these sounders have contributed greatly to our state of knowledge of the ionosphere. The data, however, must be manually analyzed by trained “scalers” and the data film archived in controlled-climate storage facilities. A map

4.2 Ground-based systems

Figure 4.2. A “typical ionogram” from a “standard” ionosonde (frequency range 0.5–12 MHz , height range 1000 km, power 10 kW, sweep time 20 s, linear frequency scale. Note the heavy vertical lines – caused by MF and HF interference.

Figure 4.3. An idealized ionogram.

of the global distribution of ionosondes (mainly the standard models) as of 1982 is shown in Figure 4.4. With the advent of reasonably priced compact personal computers, digital signal processing, new modulation-coding techniques, and VLSI, a new generation of ionosondes was developed, starting in the mid-1960s and continuing into this century. Many of these ionosondes are portable and all have much smaller volume, weight, and power consumption than did the standard ionosondes, and they produce much better ionograms. The modern sounders also permit the deletion of discrete frequencies that are contaminated by interference, and the deletion of frequencies that may interfere with other services. Advances in antenna-array theory have also made it possible to deploy arrays of receiving antennas in such a way as to permit direction-of-arrival (DOA) determination for echoes, permitting the production of “skymaps” for selected heights.

183

184

Radio techniques

Figure 4.4. A map of all ionosondes known to have existed as of 1982.

4.2 Ground-based systems

185

186

Radio techniques

Representative examples of the new sounders available at the time of writing are shown in Table 4.1. An ionogram obtained from a typical modern ionosonde is illustrated in Figure 4.5. Most of the ionosondes which produce ionograms such as that shown in Figure 4.5 are of the “modern” type, since the “standard” ionosondes are obsolescent and extremely difficult to maintain. An up-to-date description of the modern sounders and their deployment is given by Wilkinson (1995). The modern ionosondes permit the study of a wide range of ionospheric irregularities as illustrated schematically in Figure 3.34.

Capabilities and limitations A limitation of all ionosondes is that they can yield information on the ionosphere only up to the height of maximum ionization of the F2 layer (the “bottomside” of the ionosphere). Also, unless one extends the low-frequency end of the sweep (to at least 250 kHz) by increasing the height of the transmitting antenna tower and using relatively high power, not much information can be obtained from the D Table 4.1. Typical available ionosondes Name of Sounder

Specifications

Source

Digisonde Portable Sounder (DPS)

Frequency range 1–32 MHz Power 300 W pulse Height range 90–1000 km Doppler sounding, etc. Realtime data transfer via the internet Automatic scaling available

University of Massachusetts Center for Atmospheric Research, 600 Suffolk Street, 3rd Floor, Lowell, MA 01854, USA www.uml.edu

Canadian Advanced Digital Ionosonde (CADI)

Frequency range 1–20 MHz Power 600 W pulse (13-bit Barker code) Height range 90–1024 km Doppler sounding, etc. Realtime data transfer via the internet

Scientific Instruments, Ltd, 2233 Hanselman Avenue, Saskatoon, CA S7L6A7, USA

Ionosonde: HF Diagnostics Frequency range 1–30 MHz Module, 01-2000 Power 1 kW CW and peak Doppler sounding, etc.

Center for Remote Sensing, Inc., 11350 Random Hills Road, Suite 710, Fairfax, VA 22030, USA

Advanced Digital Ionosonde, IPS-71

KEL Aerospace Pty Ltd, 231 High Street, Ashburton, Victoria 3147, Australia

For specifications contact www.kel.gov.au

4.2 Ground-based systems

Figure 4.5. A typical modern digital ionogram (compare with Figure 4.2).

region. This is in contrast with the incoherent-scatter-radar (ISR) technique, which, however, is much more expensive and definitely not as portable. Another limitation is that, during episodes of intense E-region ionization (“blanketing-E”), it is not possible to obtain much information on the F region. Approximate costs of new “modern” ionosondes currently vary from about $ 30 000 to over $ 250 000. At auroral latitudes all ionosondes are subject to several rather severe limitations – namely that, during some of the most “interesting” times, auroral-E ionization or D-region absorption precludes the gathering of any ionospheric information on the layers above! These “interesting” times include magnetic storms and substorms and associated auroral and polar-cap absorption, intense auroral events, and extreme spread-F conditions. 4.2.2

Coherent oblique-incidence radio-sounding systems

We shall refer to the systems which utilize coherent radars to obtain either direct backscatter or ground-reflected backscatter from ionospheric features as oblique backscatter sounders (OBSs). The systems may be either bistatic or monostatic in

187

Radio techniques

188

(a)

(b)

Figure 4.6. An idealized sketch of the ground backscatter mode (a) and a sketch of direct backscatter from field-aligned irregularities (FAIs) in the auroral oval (b). In reality, the HF ray path is usually refracted by the ionosphere, making it orthogonal to the FAI.

configuration. OBS systems are also referred to in the literature as ionospheric radars, coherent scatter radars (CSRs), backscatter sounders, and auroral radars. They are discussed in considerable detail in Greenwald et al. (1978), Liu (1989, Sections 11 and 12), Hunsucker (1991, pp. 94–109), and Hunsucker (1993, pp. 441–450). Specifically, the WITS Handbook, edited by Liu (1989) devotes Sections 11 and 12 (64 pages) to two types of OBS systems: auroral radars and HF groundscatter radars in Appendix A1.2, as well as fundamentals of plasma dynamics and electrodynamics of the equatorial, mid-latitude, and high-latitude ionosphere in Chapters 2, 3, 5, and 6.

Basic principles A coherent-scatter echo exhibits a statistical correlation of the amplitude and phase from one pulse to another, and emanates from quasi-deterministic gra-

4.2 Ground-based systems

dients in electron density, which have correlation times usually greater than 1 ms. One can also describe backscatter as “strong” compared with incoherent-scatter echoes (the “scattering cross-section” for coherent backscatter is 104–109 times greater than that for incoherent scatter). In general, coherent backscatter is obtained when the ray path from the transmitting antenna intersects large electron-density gradients or field-aligned irregularities, at near-perpendicular incidence. Thus, coherent backscatter is 40–90 dB stronger than incoherent scatter, and is qualitatively similar to specular reflection. However, for a full understanding of the ionospheric physics, considerable plasma theory must be employed. The essence of the plasma-theory description is that, when plasma instabilities are present in the ionosphere, the amplitude of fluctuations in the medium can grow to much higher levels than the thermal background. Coherent scatter occurs when the wave vector of the medium (km) equals twice the wave vector of the transmitted wave (kt). Rather complete descriptions of the history of the development of the OBS technique, and basics of the various systems, were given by Croft (1972), in Chapter 11 of the WITS Handbook, and in Hunsucker (1991, Sections 4.1.1, 4.2.1, and 4.3.1). It is interesting to note that the first observation of coherent backscatter (from the ground) was made by Mogel in 1926, but not really understood until 1951, when it was explained independently by Dieminger (1951) and Peterson (1951). There is another class of sounders known as oblique ionosondes or “synchronized-sounders,” which are used primarily for assessing propagation characteristics of the ionosphere for HF communication circuits (see Goodman, 1992, Chapter 6). There is also an important “subset” of OBSs, most often referred to as over-the-horizon (OTH) radars, which are used by military services and other government agencies primarily for the detection of airplanes, ships, and missiles. The hardware and software are quite sophisticated, and the subject had been highly classified until fairly recently, when some of the systems were made available for ionospheric and oceanographic research. Descriptions of some of the OTH radar systems and results are given by Barnum (1986), Brookner (1987), in a special issue of the IEEE Journal on Oceanic Engineering (1986), and in a special section of Radio Science (1998). Modern OBS systems typically operate in the HF and VHF bands and use continuous-wave (CW), pulse-coded, or FMCW modulation. They obtain ionospheric information either from direct backscatter from field-aligned irregularities, or by backscatter from irregularities via a ground-reflected mode, as illustrated in the idealized sketch in Figure 4.6. In the groundscatter mode (at the top of Figure 4.6), the echoes returned to the receiver will be affected by irregularities near the ionospheric-reflection point, by the Earth-surface characteristics, and by field-aligned irregularities (FAIs), where the second hop enters the ionosphere. It is necessary to analyze the Doppler velocity, the phase characteristics, and the spectral shape of the echo to identify the scattered echo of interest. The bottom part of Figure 4.6 illustrates the mode

189

190

Radio techniques

NATURAL PHENOMENA

ARTIFICIALLY CREATED IRREGULARITIES

COHERENT RADAR

SPECTRUM VELOCITY

INTENSITY

COMPARISON WITH ROCKETS SATELLITES INCLUDING SCATTER

COMPARISON WITH CONDUCTIVITIES ELECTRON DENSITIES TEMPERATURES

NATURE OF IRREGULARITIES

COMPARISON WITH THEORETICAL APPROACHES

Figure 4.7. A summary of coherent-scatter radar investigations from a plasma-physics point of view (after Schlegel, 1984).

involving direct backscatter from FAIs, which may be significantly influenced by ionospheric refraction (depending on the frequency of the sounder). Figure 4.7 summarizes the type of information from an OBS which may be of interest to plasma physicists.

Types of oblique sounders currently in use Having generically described the sounders in the previous section, we will proceed to classify and describe them by their operating frequency and describe several of the systems currently deployed globally. The lowest-frequency OBS systems are the VLF sounders described by Kossey et al. (1983), sweeping between 5 and 30 kHz using pulse widths 100 ms. Figure 4.8 illustrates the basic system configuration and Figure 4.9 shows data obtained during disturbed periods in the polar lower ionosphere. To the best of the authors’ knowledge, no VLF sounders are at present in operation. However, VLF sounding remains a practical technique for probing the D and E regions of the ionosphere in some detail, especially at high latitudes. In the HF region (3–30 MHz) of the radio spectrum, the OBS technique has been employed since the mid-1920s. See Hunsucker (1991, Chapter 4) for a description of the history and theory for OBS systems. Perhaps the best examples of the HF OBS technique is the SuperDARN (Dual Auroral Radar Network) system (Greenwald et al., 1995) and the PACE (Polar and Conjugate Network) system (Baker et al., 1989). These HF radars operate in the frequency range of

4.2 Ground-based systems

191

CONVERTED SKYWAVE PULSE

(a) NORMAL SKYWAVE PULSE GROUNDWAVE PULSE

TRANSMITTER

NORMAL SKYWAVE COMPONENT

CONVERTED SKYWAVE COMPONENT

0

100

200

300

TIME (MICROSECONDS)

400

500

0.40 AMPLITUDE (× 10–1)

GROUNDWAVE

RELATIVE AMPLITUDE

(b)

RECEIVER

(c)

0.30

0.20

0.10

0.0 0.0

20.0 40.0 60.0 80.0 100.0 FREQUENCY (kHZ)

Figure 4.8. (a) The VLF pulsed-ionosonde technique. (b) An example of typical observed waveforms. (c) The spectrum of a typical transmitted pulse. (After Kossey et al., 1983.)

⬃8–20 MHz with an azimuth coverage of 52° and extend in range from a few hundred kilometers to more than 3000 km. Backscatter from F-region ionospheric irregularities is typically observed from ⬃10% to 60% of this range interval. The first HF radar of this type is located in Goose Bay, Labrador (Greenwald et al., 1985) and has been in continuous operation since 1983. The present SuperDARN system covers over most of the northern polar ionosphere and part of the south polar ionosphere. The fields of view of the existing, funded, and proposed northern-hemisphere SuperDARN radars are shown in Figure 4.10 (and listed in Table 4.2) and the southern-hemisphere HF radar coverage is shown in Figure 4.11. The SuperDARN radars utilize ionospheric refraction to achieve orthogonality with the magnetic-field-aligned irregularities in the high-latitude F region, and their frequency range of ⬃8–20 MHz permits achieving orthogonality over a factor of more than six in electron density. They are also frequency-agile, permitting observations at two or more different frequencies to be interwoven. An example of a SuperDARN-derived polar plasma-convection pattern is shown in Figure 4.12. The SuperDARN antenna array consists of 16 log-periodic antennas (LPAs) in the primary array and four LPAs to form a small-scale interferometer array for elevation-angle determination, as shown in Figure 4.13. RF signals from or to these antennas are phased with electronically controlled time-delay phasing elements that allow the beam to be steered into 16 directions covering the 52° azimuth sector. The azimuthal resolution of the measurements is

SOLAR ZENITH ANGLE (DEG)

90

70

50

130

110

EO

0

6 12 18 TIME (GMT)

264 0000 1978

100

SOLAR ZENITH ANGLE (DEG) 130 6

12 18 24 TIME (GMT)

300

400

JUNE

344 0000 1978

100

130

110

90

70

50

6 12 18 TIME (GMT)

300

24

400

157 0000 1979

156 0000 1979

157 0000 1979

158 0000 1979

TIME –
SECONDS

200

0

159 0000 1979

160 0000 1979

161 0000 1979

162 0000 1979

163 0000 1979

164 0000 1979

165 0000 1979

166 0000 1979

167 0000 1979

168 0000 1979

169 0000 1979

345 0000 1978

346 0000 1978

347 0000 1978

348 0000 1978

349 0000 1978

0

TIME –
SECONDS

200

90

70

110

350 0000 1978

351 0000 1978

352 0000 1978

353 0000 1978

354 0000 1978

355 0000 1978

356 0000 1978

357 0000 1978

358 0000 1978

265 0000 1978

266 0000 1978

24

DECEMBER

50

SOLAR ZENITH ANGLE (DEG)

Figure 4.9. VLF pulse-reflection data for a disturbed polar period (after Kossey et al., 1983).

UT

267 0000 1978

268 0000 1978

269 0000 1978

270 0000 1978

271 0000 1978

272 0000 1978

273 0000 1978

274 0000 1978

275 0000 1978

276 0000 1978

SEPTEMBER

TIM

– AY FD

100

300 TIME –
SECONDS

200

400

4.2 Ground-based systems

Figure 4.10 Locations and fields of view of the eight operating northern-hemisphere SuperDARN HF radars, as well as the STARE radar in northern Scandinavia and the remaining SABRE radar in Wick, Scotland (after Greenwald et al., King Salmon (C), operated by the Communications Research Laboratory in Japan; Kodiak (A), operated by the Geophysical Institute UAF in the USA; Prince George (B), operated by the University of Saskatchewan in Canada; Saskatoon (T), operated by the University of Saskatchewan in Canada; Kapuskasing (K), operated by the JHU/APL in the USA; Goose Bay (G), operated by the JHU/APL in the USA; Stokkseyri (W), operated by the CNRS/LPCE in France; 2ykkvibær (E), operated by the Radio and Space Plasma Physics Group, University of Leicester in the UK (also known as Cutlass/Iceland); and Hankasalmi (F), operated by the Radio and Space Plasma Physics Group, University of Leicester in the UK (also known as Cutlass/Finland).

193

F E W G K T B A

CUTLASSa/Finland CUTLASSa/Iceland Iceland West Goose Bay Kapuskasing Saskatoon Prince George Kodiak

Hankasalmi, Finland Pykkvibær, Iceland Stokkseyri, Iceland Labrador, Canada Ontario, Canada Saskatchewan, Canada British Columbia, Canada Kodiak Island, Alsaska

Location University of Leicester University of Leicester CNRSb JHU/APLc JHU/APLc University of Saskatoon University of Saskatoon UAFd

Affiliation

a

Notes: Co-operative United Kingdom Twin Located Auroral Sounding System. b Centre National de la Recherche Scientifique. c Johns Hopkins University Applied Physics Labratory. d University of Alaska, Fairbanks.

ID

Radar

Table 4.2. SuperDARN radars operating in the northern hemisphere

62.32 63.77 63.86 53.32 49.39 52.16 53.98 57.62

Latitude (°N) 26.61 20.54 20.02 60.46 83.32 106.53 122.59 152.19

Longitude (°E)

April 1995 December 1995 October 1994 June 1983 September 1993 September 1993 March 2000 January 2000

Operational

4.2 Ground-based systems

Figure 4.11. Fields of view of southern-hemisphere SuperDARN HF radars (after Greenwald et al., 1995). Halley (H), operated by the British Antarctic Survey in the UK (also known as the Southern Hemisphere Auroral Radar Experiment (SHARE)); SANAE (D), operated jointly by the University of Natal and the PUCHE in the Republic of South Africa; Syowa South (J), operated by the National Institute of Polar Research in Japan; Syowa East (N), operated by the National Institute of Polar Research in Japan; Kerguelen (P), operated by the CNRS/LPCE in France; and TIGER (R), operated by the La Trobe University in Australia.

dependent on radar operating frequency and ranges from ⬃2.5° at 20 MHz to 6° at 8 MHz. Since most of the observations are made in the frequency range 12–14 MHz, the nominal azimuthal resolution of the radar is ⬃4°. At a range of 1500 km, this corresponds to a transverse spatial dimension of ⬃100 km. A secondary parallel antenna array of four LPAs located 100 m in front of the primary array is used to determine the vertical angle of arrival of the backscattered signal. This secondary array also uses a phasing matrix and functions as an interferometer to determine the relative phases of the backscattered signals arriving at the two arrays. The phase information is converted into an elevation angle,

195

196

Radio techniques

Figure 4.12. A typical polar plasma-convection pattern (courtesy of R. Greenwald).

which is used to determine the propagation modes of the backscattered signal as a function of range, as well as the approximate height of the scatterers. This secondary antenna array is also visible in Figure 4.13. The range resolution of the SuperDARN measurements is determined by the transmitted pulse length (200–300 ms) and is equivalent to 30–45 km. Electronic steering of the SuperDARN antenna array occurs on microsecond time scales, which allows the radar to be scanned rapidly through a number of beams or to dwell for an extended time on a single beam. Typically a radar will scan in a sequential manner with a dwell time of 6 s in each beam and a full-scan time of 96 s. Although very useful information has been obtained using single HF radars, it became apparent that bi-directional common-volume observations with radar separations greater than 500 km were the best approach to advancing the study of high-latitude convection with HF radars (Ruohoniemi et al., 1989). The common field of view of a pair of HF SuperDARN antennas covers 15–20° of invariant latitude and 3 h of magnetic local time. The fields of view of several pairs of HF

4.2 Ground-based systems

Figure 4.13. The SuperDARN HF antenna array at Kapuskasing, Ontario (after Greenwald et al., 1995).

radars extend the spatial coverage of the high-latitude auroral zone and the polarcap boundary over many hours of magnetic local time. If ionospheric irregularities were to fill this common viewing area, it would be possible to monitor the dynamics of plasma convection over a significant part of a convection cell. The rates of occurrence of HF scattering during a solar-cycle maximum are given by Ruohoniemi and Greenwald (1997). Figure 4.14 is a sketch of the manner in which VHF and HF radars intercept field-aligned irregularities in the high-latitude E and F regions and Figure 4.15 shows a comparison between F-region Doppler velocities obtained simultaneously with the Sondrestrom ISR and the Goose Bay HF radar. More details on the SuperDARN system may be found in the review paper by Greenwald et al., (1995) and on the SuperDARN homepage on the internet. At VHF/UHF frequencies, OBS systems are primarily used as auroral radars and sometimes, at near-equatorial latitudes, to investigate irregularity structures associated with the equatorial electrojet. See Kelley (1989) for the physics of auroral and equatorial VHF/UHF echoes. Examples of VHF/UHF radars used in research into auroral and equatorial ionospheric irregularities are the Cornell University Portable Interferometer (CUPRI) (Providakes, 1985), the Saskatchewan Auroral Polarimetric Phased Ionospheric Radar Experiment (SAPPHIRE), (Kustov et al., 1996 and (1997). Auroral radars are exemplified by the Scandinavian Twin Auroral Radar Experiment (STARE), which was first described by Greenwald et al., (1978). The STARE system consists of two pulsed

197

198

Radio techniques

Figure 4.14. Idealized ray paths for VHF and HF radars to E-region and F-region FAIs (after Greenwald et al., 1995).

Figure 4.15. A comparison of F-region Doppler velocities obtained with the Goose Bay HF radar and velocities obtained by the Sondrestrom ISR (after Greenwald et al., 1995).

4.2 Ground-based systems

Figure 4.16. A map of the eight overlapping beams of the STARE radar over northern Scandinavia (after Greenwald et al., 1978).

bistatic phased-array radars located at Malvik, Norway and Hankasalmi, Finland. Beams from the radars are directed over a large common-viewing area (approximately 16000 km2) centered on the auroral zone in northern Scandinavia – as illustrated in Figure 4.16. The Doppler data from the two radars are combined to determine the vector velocity of the irregularities with 20-km 20-km spatial and 20-s temporal resolution. An example of the data obtained with the STARE system and simultaneous all-sky-camera data is shown in Figure 4.17 illustrating a westward-traveling auroral surge. (See Section 6.4.) A list of OBSs (coherent radars) deployed globally as of 2000 is shown in Table 4.3. Some of the advanced OBS systems employ arrays of interferometer antennas (Farley et al., 1981) similar to those used in radio astronomy. The Fourier transform of the digitized signals from the respective antennas is taken, and the complex cross-correlation spectrum for each pair is determined in the time domain. Spaced-antenna analysis can also be carried out in the frequency domain

199

200

Radio techniques

Figure 4.17. Superimposed epoch analysis of the spatial distribution of auroral luminosity (upper panel) and equivalent currents (lower panel) during the passage of a westward traveling surge at approximately 1911 UT on 27 March 1977 (from Inhester et al., 1981).

(Briggs and Vincent, 1992), offering some advantages over time-domain analysis. Two new novel approaches in the design of OBS systems are the FrequencyAgile Radar (FAR) (Tsunoda et al., 1995) and the multi-use system described by Ganguli et al. (1999), which may be used in modes other than as an OBS, and the Manatash Ridge Radar (a passive bistatic radar for upper-atmospheric radio science) (Sahr and Lind, 1997), which utilizes transmissions from standard FM broadcast stations.

4.2 Ground-based systems

201

Table 4.3. Currently deployed OBS (HF/VHF/UHF) systems Radar Location

Name

Type

Reference/description

Finland

COSCAT/XMTRa

Auroral/pulsed/bistatic

McCrea et al. (1991); 929.5 MHz

Finland and COSCAT/RCVRSa Sweden

Bistatic/pulsed and CW

0.5 kW, 4° elevation, 2° azimuth

U. K. and Sweden

SABRE

Auroral/pulsed

Jones et al. (1981); 150 MHz; twin radars

Scandinavia

STARE

Auroral/pulsed/bistatic

Greenwald (1987) Kustov et al. (1996); 50 kW

a

Canadian Arctic

SAPPHIRE

Auroral and polar cap/ CW/bistatic

NE Canada

SHERPA

Auroral and polar/pulsed Hanuise et al. (1992)

Polar

SUPERDARN

Polar cap and auroral/ pulsed

Greenwald et al. (1995); 6–16 MHz; 1 kW each into 16 antennas, 52º azimuth sector

Crete

SESCAT

Mid-latitude, E region/ CW/bistatic

Haldoupis and Schlegel (1993); 50.52 MHz, 1 kW, four Yagi arrays

(Portable)

CUPRI

E region/monostatic

Providakes et al. (1985); 49.92 MHz, 25 kW, five antennas

(Portable)

FAR

D, E, and F regions/ pulsed

Tsunoda (1992); 2–50 MHz, various pulse widths

Halley Bay, Antarctica

PACE

Polar cap F region/pulsed Baker et al. (1989); 8–20 MHz, 1 kW each into 16 antennas, 52° azimuth sector

Antarctica

SYOWA

Auroral/pulsed

50 and 112 MHz, 15 kW, 3–14-element coaxial antennas

Peru

Jicamarca

Equatorial/pulsed/ monostatic

Kelley (1989, Chapter 4); 50 MHz (oblique and vertical incidence), 49.9 MHz

Radio techniques

202

Table 4.3. (cont.) Radar Location

Name

Type

Reference/description

Kwajalein

Altair

Equatorial/monostatic/ pulsed

Tsunoda (1981); 155.5 MHz

Japan

MU Radar

mid-latitude, monostatic/ pulsed

Kato et al. (1989); 46.5 MHz

Notes: Acronyms for radars: COSCAT: Coherent scatter, CUPRI: Cornell University Portable Radar Interferometer, CW: Continuous Wave, PACE: Polar Anglo-American Conjugate Experiment, SABRE: Scandinavian and British Radar Experiment, SAPPHIRE: Saskatchewan Auroral Polarimetric Phased Array Ionospheric Radio Experiment, SESCAT: Sporadic-E scatter, SHERPA: System HF d’Etude Radar Polaires Auroral, STARE: Scandinavian Twin Radar Experiment, DARN: originally was Dual Auroral Radar Network – now SUPERDARN refers to the network of HF backscatter sounders that mainly probes the polar F region, FAR: Frequency Agile Radar. The SABRE radar in Sweden has been decommissioned, but the radar in Wick, Scotland, is still operational.

Some advantages and disadvantages of auroral and HF radars Auroral radars The radio ray path from these radars must intercept the Earth’s magnetic field at near-normal incidence, so siting of the radars is of critical importance. This requires that transmitters and receivers be located at high-latitude sites, which are sometimes rather inhospitable and distant from “civilization,” which, in turn, complicates the logistics. Also, in order to achieve the narrow azimuthal beamwidths required, rather large antenna arrays are required, affecting the cost. HF radars HF radar systems require larger areas for the antenna array than do VHF/UHF systems. Siting of the radars, although it is not as critical as it is for auroral radars, is important. In order to cover the entire polar cap (as in the SuperDARN system), considerable international cooperation is required. Because of their remote location, some of the sites are quite expensive to maintain. During severe auroral or polar-cap absorption the lower HF frequencies used may be unusable.

4.2 Ground-based systems

Figure 4.18. A map showing currently operational ISRs (courtesy of EISCAT Association).

4.2.3

Incoherent-scatter radars

Incoherent-scatter radars (ISRs) are a relatively new development compared with coherent backscatter techniques – they were first developed and deployed during the early 1960s. The fundamentals of the theory of incoherent scatter from the ionosphere are covered by Evans (1969), in Section 4.7 of Davies (1990), in Section 2.3.2 of Hunsucker (1991) and in Section 3.5.3 of this book. ISR technique has matured and proven to be one of the most powerful Earth-based radio techniques for probing the ionosphere and thermosphere and even for probing into the mesosphere under certain conditions. At present there are nine functional ISRs (some operating only sporadically), as shown in Figure 4.18. Most of the ISRs in use today have been described in some detail in Chapter 7 of Hunsucker (1991) and in Section 5 of Hunsucker (1993). The newest addition to the global array of ISRs is the Longyearbyen, Svalbard installation (Figure 4.19) – which is part of the EISCAT system, whose parameters are listed in Table 4.4. The design features of the Svalbard ISR are described in detail by Wannberg et al., (1997). The other operational ISRs are shown in Figure 4.18 and current facility addresses and contact personnel are listed in the current version of the NCAR CEDAR Data Base. 4.2.4

D-region absorption measurements

The power density (or attenuation) of radio waves at a distance, d, from the transmitter is reduced by geometric effects, refraction, absorption in the atmosphere,

203

204

Radio techniques

Figure 4.19. A photo of the new EISCAT ISR in Longyearbyen Svalbard (Spitzbergen) (courtesy of EISCAT Association).

and scattering and diffraction by objects in the ray path. For frequencies used in ionospheric techniques (ELF/UHF), most of the absorption occurs in the D region and is characterized as either deviative or non-deviative absorption. The theory of ionospheric absorption is treated in Davies (1990, pp. 65–66 and 215–217), Hunsucker (1991, pp. 50–53), Hargreaves (1992, pp. 65–66 and 71–72), and Section 3.4.1 of this book.

Current status and global deployment Since there are several radio techniques for measuring ionospheric absorption, we employ the URSI designations for the most-used methods. See Rawer (1976), Davies (1990, pp. 217–219), and Hunsucker (1991, Chapter 7, pp. 165–183) for extensive descriptions of these techniques. Certain of these techniques are currently in use, whereas others have fallen into disuse for various reasons.

The URSI A1a and A1b methods The URSI A1a method is usually employed at mid-latitudes, since the frequencies used (2–5 MHz) would be highly absorbed at auroral and polar latitudes. Basically, this method uses a stable, constant-output pulsed transmitter, an antenna with a uniform, vertically directed main lobe (and low sidelobes), plus a stable, sensitive receiver to analyze a signal that traverses the D region twice, being reflected by the E region. This technique requires very careful, frequent calibration of the system, plus a measurement of the E-region reflection coefficient. A variant of this method is the URSI A1b method, which uses the same basic equipment and modified equations for oblique incidence at short distances. The URSI A1a and A1b techniques were used rather extensively from the

Note: ACFs, autocorrelation functions

Gain (dBi) Polarization

Feed system Cassegrain

128 crossed dipoles 46 48 Circular Circular

Dish 32 m Cassegrain

48 Any

Dish 32 m Cassegrain

Cylinder 120 m 40 m Line feed

Antenna

67°52 N 20°26 E 76°48 N 64°27 N

VHF UHF UHF 224 931 931 3 8 8 2 klystrons 1 klystron – 8 8 8 2 1.5 1.3 – 2 0.19 0.16 – 0.001–2.0 0.001–1.0 – Binary Binary Binary 1.0 1.0 – Analog Analog Analog 250–350 70–80 30–35 8-bit ADC, 32-bit complex, ACFs, parallel channels

69°35 N 19°14 E 77°30 N 66°12 N

Kiruna

Band Frequency (MHz) Maximum bandwith (MHz) Transmitter Channels Peak power (MW) Average power (MW) Pulse duration (ms) Phase coding Minimum interpulse time (ms) Receiver System temperature (K) Digital processing

Geomagnetic inclination Invariant latitude

Geographic coordinates

Tromsø

Location

Table 4.4. Parameters of the EISCAT radar system (courtesy of EISCAT Corp.)

48 Any

Dish 32 m Cassegrain

UHF 931 8 – 8 – – – Binary – Analog 30–35

67°22 N 26°38 E 76°43 N 63°34 N

Sodankylä

42.5 Circular

45 Circular

UHF 500 10 16 klystrons 6 1.0 0.25 0.001–2.0 Binary 0.1 Analog–digital 55–65 12-bit ADC, Lag profiles 32-bit complex Antenna 1 Antenna 2 Dish Dish 32 m 42 m fixed Cassegrain

78°09 N 16°02 E 82°06 N 75°18 N

Longyearbyen

206

Radio techniques

mid-1950s through the mid-1970s, but, to the best of the authors’ knowledge, only a few installations are still in operation. However, it remains a useful method – especially in view of advances in VLSI, DSP, antenna theory, and computer techniques.

The URSI A2 method Brief discussions of the URSI A2 cosmic-noise method of measuring absorption may be found in Davies (1990, pp. 218–219 and in Hargreaves (1992, pp. 71–72), and a rather extended discussion in Hunsucker (1991, pp. 169–178). The instrument used to make URSI A2 absorption measurements is called the riometer (Relative Ionospheric Opacity Meter, Extra-terrestrial Electromagnetic Radiation). It was designed and built at the Geophysical Institute of the University of Alaska (Little and Leinbach, 1959), and was first globally deployed during the International Geophysical Year (IGY), 1957–1959. It was based on work done in the early 1950s by several investigators, and the riometer was found to be ideally suited for measuring the strong D-region absorption at high latitudes. Indeed, both polar-cap and auroral-zone absorption were verified using this instrument. See also Hargreaves (1969). In essence the riometer is just a stable radio receiver, and, in its usual form, this stability is achieved by switching the receiver input rapidly between the signal and a stable local noise source, a principle first enunciated by Machin et al. (1952). The riometer operates at some frequency above the penetration frequency of the ionosphere so that it receives the signal coming from outer space – i.e. the cosmicradio noise. Since the intensity of the cosmic noise source does not vary, reductions of the received intensity are interpreted to mean that the signal has been absorbed somewhere in the ionosphere. The cosmic-noise absorption in decibels can be calculated by using A10log10(P0 /P),

(4.1)

where P0 is the power output in the absence of the ionosphere and P is the power output of the riometer. A plot of typical riometer results is shown in Figure 4.20. Although the cosmic noise may be assumed constant over time, it is not constant over the sky. The riometer antenna, which points in a fixed direction from the observing site – to the zenith, for example – is scanned around the radio sky as the Earth rotates, coming back to the same place every sidereal day (24 h, 4 min). In order to measure the absorption, we must know what the intensity would have been in the absence of the absorption. This is usually estimated by superimposing measurements over some period of time as a function of sidereal time, and taking a line along the top of the distribution to indicate the intensity when absorption is absent. The resulting curve is generally called the quiet-day curve (QDC), and, although the idea is simple, the accurate derivation of the QDC can be the most difficult part of absorption measurement by the riometer technique (Krishnaswamy et al., 1985).

4.2 Ground-based systems

207

Figure 4.20. Auroral radio absorption at 30 MHz over a 6-h period. On the riometer chart (lower panel) the “noise diode current” is proportional to the received cosmic noise power, and the straight line is the ‘quiet-day curve’ representing the power that would be received in the absence of absorption.

Most riometers have operated with a small antenna that has a wide beam – e.g. ⬃60° between half-power points. This has been done for practical reasons, but it does bring a disadvantage in that the antenna pattern projects to a region about 100 km across in the D region. Therefore a standard riometer installation does not have good spatial resolution. In recent years, however, there has been an increase in narrow-beam work and in the use of imaging riometers. The absorption depends on the radio frequency as the inverse square (see Section 3.4.1), and this is one factor that influences the choice of a frequency for the riometer. At higher VHF frequencies the antenna can be smaller (for a given beamwidth) but the instrument becomes less sensitive to weak absorption. At the lower VHF frequencies the antenna must be large and also there is more interference from ionospherically propagated signals. The compromise has generally led to using the 30–50-MHz band. When data are obtained at several frequencies, it is usual to reduce the results to 30 MHz for comparison purposes, A(30 MHz)A( f )(30)2/f 2

(4.2)

The first generation of riometers (from the IGY/IGC era) used vacuum tubes, and solid-state circuits were introduced into this type of instrument in the 1960s, which permitted the riometer to be packaged as a small unit with low power comsumption. A problem with the solid-state riometer, however, was a lack of discrimination against interference in the front end, but this has been remedied using ceramic filters and integrated circuits (Chivers, 1999, personal communication).

208

Radio techniques

Figure 4.21. Projection of the IRIS beams at 90 km altitude (Derrick and Rosenberg, 1990). The beam centers are marked as dots, and the 3-dB levels as solid lines. The dashed circle is the projection of a typical wide-beam riometer antenna.

Imaging riometry Technical developments have now made it possible to construct riometer systems that produce a large number of narrow beams simultaneously, sufficient to construct a picture of the absorbing region out to, say, 150 km (horizontal) from the installation. Several such systems are operating at the time of writing (2002), and several more are planned. These systems are called imaging riometers. The first Imaging Riometer for Ionospheric Studies (IRIS) was installed at the South Pole in 1988–1989 (Detrick and Rosenberg, 1990). It forms 49 beams and the best resolution is about 29 km at the 90-km level (Figure 4.21). In principle, one could use 49 riometers to record the signals, but, to reduce the number, this system switches the signals sequentially among seven riometers; and, although this implies some loss of sensitivity, it is nevertheless adequate for observations at a time resolution of 10 s. The operating frequency is 38.2 MHz. The imaging riometers have demonstrated that the absorption contains features of finer scale, whose motions may be also be observed. This type of system

4.2 Ground-based systems

209

Pole star cosmic noise (method A2) 27 MHz

F layer

E layer 2.6 MHz 200 MHz

R

200−400 km

D layer (Absorber)

6 MHz

CW propagation (method A3) T

Figure 4.22. The geometry for the A3 absorption-measurement method. The dashed line from R to the Pole Star is an idealized ray path for the A2 (riometer) method (after Rawer, 1976).

is expected to produce a lot of new information about the structure and dynamics of auroral radio absorption, and the occurrence of finer-scale absorption must have implications for the effect of auroral absorption on HF radio propagation related to high-resolution systems. Some results are given in Sections 7.2.2 and 7.2.4.

URSI A3a and A3b methods The URSI A3 technique uses short, one-hop ionospheric modes at LF through HF frequencies at mid-latitudes, and the basic geometry is shown in Figure 4.22. The A3a method consists of CW field-strength measurements at oblique incidence over ground distances of 200–400 km, using frequencies from 2–3 and 6 MHz. The vertical-plane antenna patterns must be very uniform, so that small changes in reflecting-layer height will not affect the system losses, and one dominant mode must be used. Transmitter outputs and receiver sensitivities must be stable and calibrated, and no significant groundwave should be present to contaminate the results. This method is probably most applicable for long-term measurement of seasonal and sunspot variations of D-region absorption at mid-latitudes. The main difference with the URSI A3b mode is that it uses frequencies in and

Radio techniques

210

below the MF band, where the groundwave is quite strong. Therefore, a verticalloop antenna, with its plane perpendicular to the direction of the transmitter, is used to null out the groundwave, and another antenna is used to receive the skywave. The URSI A3a and A3b methods are described in considerable detail in the URSI Handbook, by Rawer (1976). Gardner and Pawsey (1953) and Belrose and Burke (1964) pioneered the development of the partial-reflection-experiment (PRE) technique. This involves a high-powered transmitter and a sensitive receiver, operating at frequencies not near the plasma frequency. The receiving antenna array has vertically directed lobes, which can distinguish between the downcoming x and o polarizations. So, by measuring the amplitudes of both magnetoionic components, one may obtain information on the D-region electron density, collision frequency, and absorption. The PRE technique has been further enhanced by measuring both the amplitude and the phase of the downcoming waves. This is a differential-phase measurement. Belrose (1970) and Meek and Manson (1987) have used MF radars in the interferometric mode to obtain more information on the middle atmosphere and the lower D region. PRE theory and experimental results were outlined in Hargreaves (1992, pp. 28–29 and 76–77) and in Hunsucker (1991, pp. 180–182). Other techniques that have been used to measure D-region absorption are described in Hunsucker (1991, pp. 182–183) and in Hunsucker (1993, pp. 459–464). Table 4.5 summarizes most of the absorption-measurement techniques. 4.2.5

Ionospheric modification by HF transmitters

During the early years of radio broadcasting Butt (1933) and Tellegin (1933) published papers describing observations of the transfer of modulation from one transmitted signal to another signal, and Tellegen correctly described the phenomenon as radio-wave interaction in the ionosphere. This was labeled in following publications as the “Luxembourg effect” (or the “Luxembourg–Gorkii effect”). Bailey (1937) was apparently the first to suggest that the ionosphere could be “heated” by a powerful HF transmitter and that this heating could produce new information about the ionosphere. “Ionospheric heating” was not experimentally confirmed until the 1960s, and results were not published until 1970, by Utlaut. Experimental and theoretical studies of “ionospheric cross-modulation,” however, were pursued from the 1940s until the 1970s, when funding for this research decreased, due to the high operating and maintenance costs of these facilities and the advent of other less expensive facilities. Davies (1990) devoted an entire chapter (pp. 506–537) to ionospheric modification, as did Hunsucker (1991, pp. 142–164). The former stressed results of modification experiments, whereas the latter stressed the technique. Another description (mainly theoretical) of ionospheric modification was Chapter 10 (pp. 267–284) by Erukhimov and Mityakov in the WITS Handbook (Liu, 1989). Radio-wave interaction and ionospheric heating were also discussed by Hargreaves (1992, pp. 93–94).

4.2 Ground-based systems

211

Basic principles It is possible to modify the ionosphere by heating it with a high-powered HF transmitter, releasing chemicals, using plasma-beam injection, explosions, and tropospheric (severe weather – Davies (1990, pp. 507–511)) and VLF wave injection. We will restrict our discussion to HF waves interacting with the ionosphere. A generic wave-interaction experiment is described in Figure 4.23 and the accompanying caption. Similarly, a generic HF heating experiment is described schematically in Figure 4.24 and the stages of the heating process are shown in Figure 4.25. An outline of cross-modulation theory was given by Hunsucker (1991, pp. 146–152); HF heating theory was given on pp. 152–155; and also by Erukhimov and Mityakov in the WITS Handbook (Liu, 1989). Some special theoretical considerations, which apply to HF heating of the high-latitude ionosphere, were Table 4.5. Capabilities and limitations of absorption-measurement techniques URSI designation or other name

Capabilities

Limitations

Remarks

A1 method

Quite sensitive

Interference, cannot measure high values

For mid- or low latitudes

A2 method

Large dynamic range

Not as sensitive as some other methods

Passive, low cost, can be used to measure polar-cap and auroral absorption

A3a method

Quite sensitive

Interference, more complex than A1

Mid- or low latitude

A3b method

Sensitive

Interference

Mid- or low latitude

PRE method

Can obtain electron and collision-frequency profiles

Interference, complex system, more expensive than others

MF radar can be used to probe the middle atmosphere

fmin method

Can give a qualitative indication of variation of absorption

Very sensitive to variations in equipment parameters

Really not too useful

LOF

Gives a value that can be applied fairly directly to HF circuits

Interference, difficult to interpret

Not used very much

Satellite HF beacon

Global coverage

Too many variables

Not too useful

212

Radio techniques

Figure 4.23. The geometry and nomenclature describing a generic ionospheric crossmodulation experiment (from Hunsucker, 1991). WT, “wanted” transmitter; DT, “disturbing” transmitter; R, receiver; A, WT keying sequence; B, DT keying sequence; C, detected echo amplitude of the wanted wave (for 50% cross modulation) at the receiver. The bottom panel shows the technique for measuring the height of attenuation. The upper trace is the received wanted echo; the lower trace is the DT pulse.

4.2 Ground-based systems

213

Figure 4.24. Some of the effects produced by high-power HF heating facilities (after Carlson and Duncan, 1977).

HEATER ANOMOLOUS ABSORPTION

ENHANCED ION LINE

HEATER ON

ENHANCED PLASMA LINE

ELECTRON DIFFUSION ALONG FIELD

PLASMA STRIATIONS 10 m

PARAMETRIC DECAY INSTABILITY

THERMAL PARAMETRIC INSTABILITY

PONDEROMOTIVE FORCE

NON-LINEAR THERMAL EFFECTS

1—10 ms

ENHANCED ELECTRON TEMPERATURE

1—10 s

10 s

1m

TIME

Figure 4.25. A schematic representation of the four stages of ionospheric heating (from Jones et al. 1986).

Radio techniques

214

presented by Stubbe et al. (1985), in a special edition of Radio Science, edited by Wong et al. (Wong, 1990). Table 4.6 lists the ionospheric modification facilities in operation from c. 1970 to the present time.

Capabilities and limitations of ionospheric-modification techniques The HF stimulation of the ionospheric plasma produces both linear and nonlinear effects, and a wide spectrum of scale sizes and lifetimes of irregularities, as well as modulating ionospheric-current systems to produce VLF and ELF propagation. This has proven to be an extremely important technique, stimulating many experimental and theoretical advances (see Carlson and Duncan, 1977; Hunsucker, 1991, pp. 162–163, and the Proceedings of the AGARD Conference on Ionospheric Modification, 1991.) It has also been demonstrated that the auroral electrojet can be modified by HF-modulated stimulation, to produce both VLF Table 4.6. Ionospheric-modification facilities in operation since 1970 HF heaters and their locations

Parameters

Remarks and references

NAIC; Arecibo, Puerto Rico, USA

18° N/67° W; 300 MW/3–15 MHz

Operational in 1971; Gordon et al. (1971)

EISCAT; Tromsø, Norway

69.6° N; 1200 MW

HIPAS UCLA, USA

64.9° N/146.9° W; 50 MW/2.8–4.9 MHz

Wong et al. (1983)

Established by the USSR Gissar (Dushanbe)

38° N; 6–8 MW/4–6 MHz

Operational in 1981 Erukhimov et al. (1985)

Khar’kov

50° N; 6–12 MHz

Bogdan et al. (1980)

Moscow

56° N; 1000 MW/1.35 MHz

Schluyger (1974)

Sura Radiophysics Research Institute Nizhni Novgorod

56° N; 4.5–9 MHz

Belov et al. (1981)

Zimenki

56° N; 20 MW/ 4.6–5.7 MHz

Getmatsey et al. (1973)

Monchegorsk

68° N; 10 MW/3.3 MHz

Kaputsin et al. (1977)

HAARP, Alaska, USA

63° N; 145.1° W; 2.8–10 MHz

www.haarp.alaska.edu

Notes: 1. Several diagnostic techniques are usually employed at the HF heater sites to detect ionospheric changes caused by the heater. Some typical diagnostics include ISRs, ionosondes, coherent radars, and spectro-photometers. 2. The description of the facilities in this table incorporates the latest information available to the author at the time of writing.

4.3 Spaced-based systems

and ELF radiation. This technique is quite expensive, both in terms of initial costs and in terms of operating and maintenance costs, which means that most operations are in the campaign mode. Also, because of the high levels of effective radiated power and the large area needed for high-gain antenna arrays, environmental-impact studies can drive up the capital costs, and require special measures to reduce possibly harmful radiation effects.

4.3

Space-based systems

4.3.1

A history of Earth-satellite and radio-rocket probing

Hey et al. (1946) were probably the first scientists to realize that extraterrestrial sources could be utilized to study the ionosphere. Subsequently, Smith et al. (1950), Little (1952), and Hewish (1952) showed that the radio-star emanations could be used to study the irregular nature of the ionosphere. Radar echoes from the moon resulted in the discovery of the ionospheric Faraday-rotation effect (Murray and Hargreaves, 1954; Browne et al., 1956; Evans, 1956). With the advent of the artificial-Earth-satellite era (Sputnik, October 1957), satellite radio beacons were utilized to study the ionosphere. As electronics technology and rocket-booster capabilities advanced, it became possible to actually place miniaturized ionosondes into orbit, starting with the Canadian–US Alouette I topside sounder in 1962. Actual in situ measurements of the ionospheric plasma from rockets and satellites have been made since the late 1940s, and a variety of radio-frequency (RF) probes has been utilized. The Langmuir probe, retarding-potential analyzers, plasma-drift meters, etc. are not really RF devices; they have been described by Kelley (1989, pp. 437–454), but will not be discussed in this book. During the last decade, several books that discuss Earth-satellite and rocketradio techniques for probing the Earth’s ionosphere have been published (Liu, 1989, pp. 44–147; Davies, 1990, pp. 260–296; Hunsucker, 1991, pp. 187–207; Hargreaves, 1992, pp. 64–65 and 67–71). 4.3.2

Basic principles of operation and current deployment of radio-beacon experiments

The first class of satellite experiments carries an onboard transmitter (a radio beacon) and utilizes a network (sometimes global in coverage) to receive the transmissions. The daily, seasonal, geographic, and magnetic-storm-time variations of the total electron content (TEC) of the ionosphere have been obtained for the global ionosphere from various radio-beacon-experiment (RBE) satellites since the early 1960s. These TEC studies have yielded information on the large-scale changes in the ionosphere, such as orders-ofmagnitude changes in TEC and medium-scale variations such as those caused

215

Radio techniques

216

by atmospheric gravity waves (AGWs). Another class of experiments measures the scintillations in phase and amplitude of a stable (usually multifrequency) beacon transmitter, thus providing information on the fine structure of ionospheric irregularities. The TEC can be determined from RBE satellites by measuring the differential Doppler effect between two signals (Bowhill, 1958), the Faraday rotation of the electric vector, the modulation phase (or group delay) between two different frequencies, or the carrier-phase difference between two widely spaced frequencies. Most of the TEC studies, from the early 1960s through the mid-1970s, simply monitored the transmissions of radio beacons aboard the satellite whose primary purpose was to track the satellite, and both near-polar-orbiting and geostationary satellites were used as “targets of opportunity.” The first results obtained using geostationary satellite RBEs were reported by (Garriott et al., 1965). Hargreaves developed the first proposal for a geosynchronous RBE specifically designed for ionospheric studies, which was described by Davies et al.,1975. More recent RBE studies involve the geostationary ETS-1 and ETS-2, the US Navy NNSS (TRANSIT) satellites, and the GPS constellation. Other RBE satellites, used for studies both of TEC and of scintillation, were WIDEBAND and POLAR BEAR. More recently, the constellation of GPS satellites has provided much new information on ionospheric morphology and the structure of irregularities from TEC and tomographic methods (Davies, 1990; Crain et al., (1993). The geometry and equations describing Faraday rotation, scintillation, and other TEC methodologies are described by Fremouw et al. (1978), Basu et al. (1988), Ho et al. (1996), and Pi et al. (1997), and in Sections 3.4.4 and 3.4.5 of this book. 4.3.3

Topside sounders

As mentioned in the introductory paragraph to this section, it became possible to place miniaturized sounders in satellites in the early 1960s, thus initiating the era of continuous global monitoring of the ionosphere using topside ionosondes. Several topside sounders have been launched and have performed beyond expectations in the last three decades: Alouette I (1962), Explorer (1964), Alouette II (1965), ISIS-I (1969), Cosmos-381 (1970), ISIS-B (1964), ISS-B (1978), EXOS-B (1978), Intercosmos-19 (1979), EXOS-C (1984), Cosmos 1809 (1984) and ISEE-1 and 2 (1979). Strictly speaking, EXOS-B, EXOS-C, and ISEE-1 are not topside sounders in the ionosonde sense, but they are “relaxation sounders” used to excite plasma waves in situ. Again, we are fortunate to have extended descriptions in the literature: the WITS Handbook, edited by Liu (1989), Davies (1990, pp. 261–273), Hunsucker (1991, pp. 200–203), and Hargreaves (1992, pp. 64–65). Vast quantities of data have been obtained using topside sounders, some of which have not been analyzed. As an example, the Alouette/ISIS series of sounders provided 50 satellite years of measurements, and has led to the publication of over 1000 papers (see Jackson, 1986, and Benson, 1997).

4.4 Other techniques

4.3.4

In situ techniques for satellites and rockets

In situ RF probes used aboard rockets and satellites were described in detail in the WITS Handbook, by Hunsucker (1991, pp. 205–207), and by Hargreaves (1992, pp. 52–53). These methods of trans-ionospheric propagation can be adapted to investigate the lower ionosphere. Since the signal need not penetrate the denser part of the ionosphere, its frequency can be reduced to make the observations more sensitive. The electron density and collision frequency can then be determined as functions of height as the rocket ascends and descends. One basic type of instrument is the RF impedance probe, which was first suggested by Jackson and Kane (1959)]. Its basic principle of operation is that the input impedance of an electrically short antenna is given by a capacitive reactance (1/(C0)) in free space, but the behavior departs from C0 when it is immersed in a plasma. Another basic in situ probe is the resonance probe, which is identical to the relaxation sounders mentioned in the previous subsection. It consists of a transmitter and a receiver immersed in the plasma, which excites the plasma in such a way as to make it oscillate at the various magnetoionic frequencies, as described by Benson and Vinas (1988). Other sensors include the Langmuir probe and its derivatives, mass spectrometers, particle detectors, and magnetic and electric-field instruments (see pp. 49–58 of Hargreaves, 1992). 4.3.5

Capabilities and limitations

Each of the three techniques (involving RBEs, topside sounders and in situ probes) discussed in this section does some things very well and other things not so well. However, when these three techniques are employed together in campaigns, they provide considerable information about the ionospheric plasma. Table 4.7 attempts to summarize the salient capabilities of these techniques.

4.4

Other techniques

The techniques discussed in this section are no less important than those discussed in previous sections. However, some of them are variants of certain basic methods, whereas others are quite new and in the process of being implemented. 4.4.1

HF spaced-receiver and Doppler systems

Unfortunately, there is some confusion between the spaced-receiver technique (SRT) and Doppler techniques for measuring the motion of ionospheric irregularities. This may be due in part to the fact that both techniques use multiple receiving antennas, although the antenna spacing for the Doppler method is usually much less than that in the SRT. The concept of the SRT was conceived by

217

218

Radio techniques

Ratcliffe and Pawsey (1933) and by Pawsey (1935), and was first applied experimentally by Mitra (1949). Discussions of these techniques were given by Kelley (1989, pp. 431–434), Davies (1990, p. 243–245), Hunsucker (1992, pp. 207–211), and Hargreaves (1992, pp. 300–302), and in some recent papers. The SRT usually involves one transmitter and several receivers, with the location of the receiving antennas optimally spaced in regard to the horizontal scalesize of the particular ionospheric irregularity to be investigated. Figure 4.26 illustrates the wide range of irregularities in the terrestrial ionosphere. An extended discussion of the SRT is given by Hargreaves (1992, pp. 300–302) and by Hunsucker (1993, pp. 470–473). Table 4.7. Advantages and limitations of radio beacons and topside sounders Technique

Advantages

Limitations

Radio beacons for TEC studies

Global coverage of the ionosphere from polar-orbiting satellites; constant beacon parameters; not-too-complex receiving system; continuous coverage for a large area from geostationary satellites; ability to study TIDS.

(1) Relatively expensive; (2) rather complex calibration problems; yields a vast quantity of data (sometimes overwhelming!); also, rather painstaking data analysis is required. At present, there are few RBEs suitable for ionospheric studies; (3) the polar-orbiting satellites are, of course, always moving in reference to the Earth station, thus convoluting spatial and temporal effects.

Radio beacons for scintillation studies

The averages of global and temporal coverage listed above also apply to the polar-orbiting and geostationary satellites, respectively, used in scintillation studies. Many earth stations can use the same beacon for TEC and scintillation studies.

Interpretation of these data in the context of extant theories is a non-trivial task. Remarks 1, 2, and 3 above also apply.

Topside sounders

Since all topside sounders use relatively high-inclination orbits, they have good global coverage. They are also free from D-layer absorption effects, and provide much information on the ionospheric above the F2 peak.

More-complex instrumentation than most beacons.

4.4 Other techniques

219

1

Multibeam Reflections

Multiple Scattering 0.1

Irregularity Level ∆N/N

(Anomalous Attenuation Effect, Lacuna Effect)

(Spread F; ∆f / f Method; Polarization Distortions; s 2 , s 3 Indices)

1e –2 Tilts, AGWs (Echolocation, Slow Phase Variations)

Weak Diffraction

1e –3

(Phase Structure Function Method) 1e –4

λ 1e –5 1e -2

LF 0.1

1

1e +1

1e +2

1e +3

Irregularity Scales, km

Figure 4.26. A composite spectrum summarizing the intensity of ionospheric irregularities as a function of wavenumber, over a large spatial scale (after Booker, 1979).

4.4.2

The HF Doppler technique

This technique is quite useful for monitoring small, transient changes in the ionosphere. It has been incorporated into several of the modern digital ionosondes and coherent radars, as well as being used as a “stand-alone” technique. Basically, in its first implementation, this technique used a very stable transmitter and one or more stable receivers and local oscillators. These heterodyned the received skywave signal and then the beat frequency was usually recorded on tape at slow speed. The data tapes were then speeded up by a factor of several thousand and the amplitude and phase of the Doppler variation with time were spectrum analyzed. This version of the stand-alone HF/CW Doppler sounder was pioneered in Boulder, Colorado in the early 1960s (Watts and Davies, 1960; Davies, 1962; Davies and Baker, 1966). Modern Doppler techniques utilize digital signal processing and computers instead of tape recorders. A thorough treatment of ionospheric phase and frequency variations and of the HFD technique was given by Davies (1969), and other descriptions may be found in Jones (1989, Chapter 4, pp.

Radio techniques

220

383–398); the WITS Handbook, edited by Liu; Hunsucker (1991, pp. 211–213); Hargreaves (1992, pp. 66–67); and Haldoupis and Schlegel (1993). 4.4.3

Ionospheric imaging

For over four decades now, ionospheric physicists and engineers have discussed and used radio methods to image the ionosphere. Rogers (1956) was probably the first to suggest using the wavelength-reconstruction method for this purpose. Schmidt (1972) proposed using VHF signals from a satellite to localize ionospheric irregularities, and a description of a two-dimensional technique was given by Parthasarathy (1975) and Schmidt and Taurianen (1975). Stone (1976) developed a more sophisticated holographic radio camera, using a 32-element antenna array oriented perpendicular to the path of the beacon satellite, with which he produced three-dimensional reconstructions from measured data. Additional details concerning the development of radio-imaging techniques (including computerized ionospheric tomography) from c. 1975 to the present may be found in Nygren et al. (1997), Pryse and Kersley (1992), and in reviews by Hunsucker (1993 and 1999) and Kunitsyn and Tereschenko (1992). 4.5

Summary

As we move into the twenty-first century, we see an extensive deployment of stateof-the-art, sophisticated ground- and space-based radio installations for probing the terrestrial ionosphere – probably surpassing the deployment during the IGY/IGC. There has also been a sea change in the availability of near-realtime and archived data from these radio installations on the internet. Ionospheric scientists thus have rapid access to an unprecedented assemblage of data as well as using email to rapidly communicate with the principal investigators of the various observatories. There is now a global distribution of modern ground-based instruments such as digital ionosondes, coherent VHF/UHF radars (CUPRI, COSCAT, STARE, SABRE CANOPUS, . . .), incoherent-scatter radars (EISCAT, Millstone Hill, Jicamarca, Arecibo, MU Radar, and Russian installations), imaging riometers (IRIS), and ionospheric HF heaters (HIPAS, HAARP, Arecibo, EISCAT, . . .). For the first time, we now have near-realtime access to solar, interplanetary, and magnetospheric data from a new generation of scientific satellites such as ACE, WIND, POLAR, and FAST. When one is analyzing data from these instruments located at high latitude, one must remember that there are some limitations – especially for those using HF. Under especially disturbed conditions (magnetic storms, etc.) ionosondes may be strongly affected by D-region absorption and intense E-region ionization, and HF radars may also be affected by these phenomena.

4.6 References and bilbliography

An invaluable compilation of details of and data from most of the radio techniques listed in this chapter is contained in the CEDAR (Coupling, Energetics and Dynamics of Atmospheric Regions) Data Base published by the National Center for Atmospheric Research (NCAR) in Boulder, Colorado. CEDAR is a program sponsored by the US National Research Foundation (NSF) which sponsors many research programs at institutions in the USA, holds an annual meeting in June in Boulder, Colorado, and updates its data catalog.

4.6

References and bibliography

Section 4.1 Hargreaves, J. K. (1992) The Solar–Terrestrial Environment. Ch. 3, Techniques for observing geospace. Cambridge University Press, Cambridge. Hunsucker, R. D. (1991) Radio Techniques for Probing the Terrestrial Ionosphere – Physics and Chemistry in Space, Vol. 22 – Planetology. Springer-Verlag, Berlin. Hunsucker, R. D. (1993) A review of ionospheric radio techniques: present status and recent innovations, Ch. 22. In The Review of Radio Science, 1990–1992 (ed. W. R. Stone). Published for the International Union of Radio Science (URSI) by Oxford University Press, Oxford. Hunsucker, R.D. (1999) Electromagnetic waves in the ionospheric. In Wiley Encyclopedia of Electrical and Electronic Engineering (ed. J. Webster), Vol. 6, pp. 494–506. Wiley, New York. Kelley, M. C. (1989) Appendix A – ionspheric measurement techniques. In The Earth’s Ionosphere – Plasma Physics and Electrodynamics. Academic Press, New York. Liu, C.-H. (1989) World Ionosphere/Thermosphere Study. WITS Handbook, Vol. 2, Instrumentation. SCOSTEP, University of Illinois Champaign-Urbana, Illinois

Section 4.2 Appleton, E. V. and Barnett, M. A. F. (1925) On some direct evidence for the downward atmospheric reflection of radio waves. Proc R. Soc. A 109, 621–641. Bailey, V. A. (1937) Interaction by resonance of radio waves. Nature 139, 68–69. Baker, K. B., Greenwald, R. A., and Ruohoniemi, J. M. (1989) PACE: Polar AngloAmerican conjugate experiment. Eos 22, 785–799. Barnum, J. R. (1986) Ship detection with a high-resolution HF skywave radar. IEEE J. Oceanic Eng. 11, 196. Belrose, J. S. and Burke, M. J. (1964) Study of the lower ionosphere using partial reflection. J. Geophys. Res. 69, 2779–2818. Belrose, J. S. (1970) Radio wave probing of the ionosphere by the partial reflection of radio waves (from heights below 100 km). J. Atmos. Terr. Phys. 32, 567–596. Booker, H. G. (1979) The role of acoustic gravity waves in the generation of spread-F and ionospheric scintillations. J. Atmos. Terr. Phys. 41, 501–515.

221

222

Radio techniques

Breit, G. and Tuve, M. A. (1926) A test of the existence of the conducting layer. Phys. Rev. 28, 554. Briggs, B. H. and Vincent, R. A. (1992) Spaced antenna analysis in the frequency domain. Radio Sci., 27, 117–129. Brookner, (1987) Array radars: an update. Microwave J., February, 117–137. Butt, A. G. (1933) World Radio, April, 28. Carlson, H. C. and Duncan, L. M. (1977) HF excited instabilities in space plasmas. Radio Sci., 12, 1001. Croft, T. A. (1972) Skywave backscatter: a means for observing our environment at great distance. Rev. Geophys. Space Phys. 10, 73–155. Davies, K. (1990), Ch. 4, Radio soundings of the ionosphere. In Ionospheric Radio, Peter Peregrinus on behalf of the IEE, London. Detrick, D. L. and Rosenberg, T. J. (1990) A phased-array radiowave imager for studies of cosmic noise absorption. Radio Sci. 25, 325. Dieminger, W. (1951) The scattering of radio waves. Proc. Phys. Soc. B 64, 142–158. Evans, J. V. (1969) Theory and practice of ionospheric study by Thomson scatter radar. Proc. IEEE 57, 496. Ganguli, S., Von Bavel, G., and Brown, A. (1999) Imaging of electron density and magnetic field distributions in the magnetosphere: a new technique. Proc. IES99, 563–574. Gardner, F. F and Pawsey, J.L. (1953) Study of the ionospheric D-region using partial reflections. J. Atmos. Terr. Phys. 3, 321. Goodman, J. (1992) HF Communication: Science and Technology. Van Nostrand Reinhold, New York. Greenwald, R. A., Weiss, W., Nielson, E., and Thomson, N. R. (1978) Stare: a new radar auroral backscatter experiment in Northern Scandinavia. Radio Sci. 13, 1021–1039. Greenwald, R. A., Baker, K. B., Hutchins, R.A., and Hanuise, C. (1985) An HF phased array radar for studying small-scale structure in the high-latitude ionosphere. Radio Sci. 20, 63–74. Greenwald, R.A., Baker, K. B., Dudeney, J. R. Pinnock, M. Jones, T. B., Thomas, E. C., Villain, J.-P., Cerisier, J. C., Senior, C., Hanuise, C., Hunsucker, R. D., Sofko, G., Koehler, J. Nielsen, E., Pellinen, R., Walker, A. D. M., Sato, N., and Yamagishi, H. (1995) The Sapphire North radar experiment: Observation of discrete and diffuse echoes. Space Sci. Rev. 71, 761–796. Haldoupis, C. and Schlegel, K. (1993) A 50 MHz radio Doppler experiment for midlatitude E-region coherent backscatter studies. Radio Sci. 28, 959. Hargreaves, J. K. (1969) Auroral Absorption of HF radio waves in the ionosphere: a review of results from the first decade of riometry. Proc. IEEE 57, 1348–1373. Kossey, P. A., Turtle, J. P., Pagliarulo, R. P., Klemetti, W. I., and Rasmussen, J. E. (1983) VLF reflection properties of the normal and disturbed polar ionosphere in northern Greenland. Rad. Sci., 18, 907–916.

4.6 References and bilbliography

Krishnaswamy, S., Detrick, D. L., and Rosenberg, T. J. (1985) The inflection point method of determining riometer quiet day curves. Radio Sci. 20, 123–130. Kustov, A.V., Koehler, J. A., Sofko, G. J., and Danskin, D. W. (1996) The SAPPHIRENorth radar experiment: Observations of discrete and diffuse echoes. J. Geophys. Res., 101, 7973–7986. Kustov, A.V., Koehler, J. A., Sofko, G. J., Danskin, D. W., and Schiffler, A. (1997) Relationship of the SAPPHIRE -North merged velocity and the plasma convection velocity derived from simultaneous SuperDARN radar measurements. J. Geophys. Res. 102, 2495–2501. Machin, K. E., Ryle, M., and Vomberg, D. D. (1952) The design of an equipment for measuring small radio-frequency noise powers, Proc.IEE 99, 127. McCrea, K., Schlegel, K., Nygren, T., and Jones, T. B. (1991) COSCAT, A new auroral radar facility on 930 MHz – system description and first results. Ann.Geophysicae 9, 461–469 Meek, C. H. and Manson, A. H. (1987) Medium frequency interferometry of Saskatchewan, Canada. Can J. Phys. A 35, 917–921. Peterson, A. M. (1951) The mechanism of F-layer propagated backscatter. J Geophys. Res. 56, 221–237. Rawer, K. (1976) Manual on Ionospheric Absorption Measurements. World Data Center A for Solar–Terrestrial Physics, Boulder, Colorado. Ruohoniemi, J. M., Greenwald, R. A., Baker, K. B., Villain, J.-P., Hanuise, C., and Kelly, J. (1989) J. Geophys. Res. 13, 463. Sahr, J. D. and Lind, F. D. (1997) The Manatash Ridge radar: passive bistatic radar for upper atmospheric radio science. Radio Sci. 32, 2345–2358. Schlegel, K. (1984) HF and VHF Coherent Radars for Investigation of the High Latitude Ionosphere. Max-Planck-Institut für Aeronomie, Katlenburg-Lindau. Stubbe, P., Kopka, H. and Rietveld, M. T. (1985) Ionospheric modification experiment with the Tromsø heating facility., J. Atmos. Terr. Phys. 47, 1151–1163. Tellegin, B. D. (1933) Interaction between radio waves? Nature 131, 840. Tsunoda, R. T., Livingston, R. C., Buoncore, J. J., and McKinley, A. V. (1995) The frequency-agile radar: a multi-functional approach to remote sensing of the ionosphere. Radio Sci. 30, 1623. Utlaut, W. F. (1970) An ionospheric modification experiment using very high power, high frequency transmission. J. Geophys. Res. 73, 6402–6405. Wannberg, G., Wolf, I., Vanhainen, L.-G., Koskenniemi, K., Rottger, J., Postila, M., Markkanen, J., Jacobsen, R., Stenberg, A., Larsen, R., Eliassen, S., Heck, S., and Huuskonen, A. (1997) The EISCAT Svalbard radar: A case study in modern incoherent scatter radar system design. Radio Sci. 32, 2283–2307. Wilkinson, P. (ed.) (1995) Ionosonde Networks and Stations. World Data Center A for Solar–Terrestrial Physics, National Geophysical Data Center, Boulder, Colorado. Wong, A. Y. (1990) Foreword: Ionospheric modification in the polar region (IMPR). Radio Sci. 25, 1249.

223

224

Radio techniques

Section 4.3 Basu, Su., Basu, Sa., Weber, E. J., and Coley, W. R. (1988) Case-study of polar cap scintillation modeling using DE2 irregularity measurements at 800 km. Radio Sci. 23, 545–553. Benson, R. F. and Vinas, F. (1988) Plasma instabilities stimulated by HF transmitters in space. Radio Sci. 23, 585–590. Benson, R. F. (1997) Evidence for the stimulation of field-aligned electron density irregularities on a short time scale by ionospheric topside sounders. J. Atmos. Solar–Terr. Phys. 59, 2281–2293. Bowhill, S. A. (1958) The Faraday-rotation rate of a satellite radio signal. J. Atmos. Terr. Phys. 13, 175–176. Browne, I. C., Evans, J.V., Hargreaves, J. K., and Murray, J. A. W. (1956) Radio echoes from the moon. Proc.Phys.Soc. B 69, 901–920. Burke, G. J. and Poggio, A. J. (1981) Numerical electromagnetics code (NEC) – Method of Moments. NOSC, San Diego, California. Crain, D. J., Sojka, J. J., Schunk, R. W., nd Klobuchar, J. A. (1993) A first-principle derivation of the high-latitude total election context distribution. Radio Sci. 28, 49. Davies, K. (1962) The measurement of ionospheric drifts by means of a Doppler shift technique. J. Geophys.Res. 67, 4909–4913. Davies, K. (1969) Ionospheric Radiowaves. Blaisdell, Waltham, Massachusetts. Davies, K. (1980) Recent studies in satellite radio beacon studies with particular emphasis on the ATS-6 radio beacon experiment. Space Sci. Rev. 25, 357–430. Davies, K. and Baker, D. M. (1966) On frequency variations of ionospherically propagated HF radio signals. Radio Sci. 1, 545–556. Davies, K., Fritz, R. B., Grubb, R. N., and Jones, J. E. (1975) Some early results from the ATS-6 Radio Beacon experiments. Radio Sci. 10, 785–799. Doherty, P. H., Decker, D. T., Sultan, P. J., Rich, F. J., Borer, W. S., and Daniell, R. E. (1999) Validation of PRISM: the climatology. Proc. IES99, pp. 330–339. Evans, J. V. (1956) The measurement of electron content of the ionosphere by the lunar radar method. Proc. Phys. Soc. B 69, 953–955. Farley, D. T., Ierkic, H. M., and Fejer, B. G. (1981) Radar interferometry. A new technique for studying plasma turbulence in the ionosphere. J. Geophys. Res. 86. Fremouw, E. J., Leadabrand, R. L., and Livingston, R. C. (1978) Early results from the DNA wideband satellite experiment – complex signal scintillation. Radio Sci. 13, 167–187. Garriott, O. K., Smith, F. L., and Yuen, P. C. (1965) Observations of ionospheric electron content using a geostationary satellite. Planet. Space Sci. 13, 829–838. Hewish, A. (1952) The diffraction of galactic radio waves as a method of investigating the irregular structure of the ionosphere. Proc. R. Soc. A 214, 494–514. Hey, J. S., .Parsons, S. J., and Phillips, J. W. (1946) Fluctuations in cosmic radiation at radio frequencies. Nature 158, 234. Ho, C. M., Mannucci, A. J., Lindqwister, U. J., Pi, X., and Tsuratani, B. T. (1996)

4.6 References and bilbliography

Global ionosphere perturbations monitored by the worldwide GPS network. Geophys. Res. Lett. 23, 3219–3222. Hunsucker, R. D. and Owren, L. (1962) Auroral sporadic-E ionization. J. Res. NBS Radio propagation D 66, 581–592. Inhester, W., Baumjohann, B., Greenwald, R. A., and Nielsen, E. (1981) J. Geophys. Res. 49, 155. Jackson, J. E. (1986) Alouette-ISIS Program Summary. NationaL Space Science Data Center/World Data Center A for Rockets and Satellites, NASA/GSFC, Greenbelt, Maryland. Jackson, J. E. and Kane, J. A. (1959) Measurements of ionospheric electron densities using a RF probe technique. J. Geophys. Res. 64, 8. Jones, T. B. (1989) The HF Doppler technique for monitoring transient ionospheric disturbances. WITS Handbook vol. 22, p. 383. Jones, T. B., Spracken, C. T., Stewart, C. P., and Thomas, E. C. (1981) SABRE, a UK/German auroral radar. IEE Conf. Proc., 195, 269–271. Kunitsyn, V. E., and Tereschenko, E. D. (1992) Radio tomography of the ionosphere. IEEE Antennas Propagation Mag. 34, 22–32. Little, C. G. (1952) The origin of the fluctuations on galactic radio noise. Ph. D. Thesis, University of Manchester, Manchester. Little, C. G. and Leinbach, H. (1959) The riometer – a device for the continuous measurement of ionospheric absorption. Proc. IRE 47, 315–320. Mitra, S. N. (1949) A radio method of measuring winds in the ionosphere. Proc. IEE 96, 441. Murray, W. A. S. and Hargreaves, J. K. (1954) Lunar radio echoes and the Faraday effect in the ionosphere. Nature 173, 944. Nygren, T., Markkanen, M., Lehtinen, M., Tereshehrnko, E. D., and Khudukon, B. Z. (1997) Stochastic inversion in ionospheric radiotomography. Radio Sci. 32, 2359–2372. Parthasarathy, R. (1975) Ionospheric Photography at Radio Wavelengths. Geophysical Institute, University of Alaska-Fairbanks, Fairbanks, Alaska. Pawsey, J. L. (1935) Further investigations of the amplitude variations of downcoming wireless waves. Proc. Camb. Phil. Soc. 31, 125. Pi, X, Mannucci, A. J., Lindqwister, U. J., and Ho, C. M. (1997) Monitoring of global ionospheric irregularities using the worldwide GPS network. Geophys. Res. Lett. 24, 2283–2286. Providakes, J. F. (1985) Radar interferometer observations and theory of plasma irregularities in the auroral ionosphere. Ph.D. Thesis, Cornell University. Pryse, S. E. and Kersley, L. (1992) A preliminary experimental test of ionospheric tomography. J. Atmos. Terr. Phys., 54, 1007–1012. Ratcliffe, J. A. and Pawsey, J. L. (1933) A study of the intensity variations of downcoming radio waves. Proc. Camb. Phil. Soc. 29, 301. Rodgers, G. L. and Ireland, W. (1980) Ionospheric holography I: the holographic interpretation of ionospheric data. J. Atmos. Terr. Phys. 42, 385–396.

225

226

Radio techniques

Rogers, G. L. (1956) A new method of analyzing ionospheric movement records. Nature 177, 613–614. Ruohoniemi, J. M. and Greenwald, R. A. (1997) Rates of scattering occurrence in routine HF radar observations during solar cycle maximum. Radio Sci. 32, 1051. Schmidt, G. (1972) Determination of the height of ionospheric irregularities with the holographic method. Z. Geophys. 38, 891. Schmidt, G. and Taurianen, A. (1975) The localization of ionospheric irregularities by the holographic method. J. Geophys. Res., 80, 4313–4324. Smith, F. G., Little, C. G., and Lovell, A. C. B. (1950) Origin of the fluctuations in the intensity of radio waves from galactic sources. Nature 165, 422–423. Stone, W. R. (1976) A holographic radio camera technique for the 3D reconstruction of ionospheric inhomogeneities. J. Atmos. Terr. Phys. 38, 583–592. Swider, W. (1996) E-region time-dependent chemical model. In STEP Handbook of Ionospheric Models (ed. R. Schunk). SCOSTEP Secretariat, NOAA/NGDC, Boulder, Colorado. Taurianen, A. (1982) Application of wave field reconstruction of VHF radio waves in investigating single, isolated ionospheric irregularities. Radio Sci. 17, 684–692. Watts, J. M. and Davies, K. (1960) Rapid frequency analysis of fading radio signals. J. Geophys.Res. 65 2295–2302.

Chapter 5 The high-latitude F region and the trough

5.1

Circulation of the high-latitude F region

5.1.1

Introduction

The high-latitude ionosphere is greatly influenced by the outer magnetosphere and the solar wind, the essential connection being via the geomagnetic field. Through this connection the high-latitude F region is exposed to the interplanetary medium and thence to disturbances originating in the Sun. The circulation of the magnetosphere (Section 2.4.1) establishes a corresponding circulation pattern in the high-latitude F region. Although production by solar EUV is still important, these added features lead to a more complex ionosphere, which exhibits some striking differences both from the middle- and from the low-latitude zones. In describing the F region at high latitude, therefore, we shall be particularly concerned with two underlying factors: (a)

the dynamic nature of the high-latitude ionosphere, the pattern of circulation of the F region being mainly controlled by the solar wind and its variations, and

(b)

the influence of energetic particles from the magnetosphere and the solar wind, to which the region is generally more accessible than is the ionosphere at lower latitudes.

The auroral zones, which occur within the high-latitude region, are particularly complex, and the trough of depleted ionization on its equatorward side has its own pattern of behavior. The present chapter deals with the behavior of the highlatitude F region, its patterns of circulation, and their consequences. The auroral phenomena are discussed in Chapter 6.

227

The high-latitude F region

228

5.1.2

Circulation patterns

In the F region the ion–neutral-species collision frequency is small relative to the gyrofrequency, and therefore the plasma moves with the magnetic-field lines rather than with the neutral wind. The motion may also be considered as the motor effect of the cross-polar electric field mapped down from the magnetosphere. The plasma speed v, the electric field E and the magnetic flux density B are related by vE/B (Equation (2.12)). Since the polar magnetic field is almost vertical, with a value of about 5 105 Wb m2, typical plasma speeds of 200–1000 m s1 correspond to electric fields of 10–50 mV m1. The integral of the electric field across the polar cap, which may be determined from satellite measurements, provides an estimate of the total electric potential difference across the magnetosphere between its dusk and dawn sides. Various formulae have been derived to express the polar-cap potential () in terms of the solar-wind speed (vsw), the total flux density of the interplanetary magnetic field (IMF) (B) and the “clock angle” of the IMF () as seen from the Earth. If Bz and By are the northward and westward components of the IMF, then  tan1 |By /Bz |

if Bz 0 (northward),

or  180° tan1 |By /Bz |

if Bz 0 (southward).

A recent analysis (Boyle et al., 1997) gives  104v 2sw 11.7Bsin3( /2),

(5.1)

where vsw is in km s1, B is in nanoteslas, and  is in kilovolts. If vsw 400 km s1 and B5 nT, the first term gives 16 kV and the second one between 0 and 58.5 kV, depending on the orientation of the IMF. In terms of the magnetic activity index Kp (Section 2.5.4), which is derived entirely from ground-based data,  16.515.5Kp.

(5.2)

The basic flow pattern caused by the polar-cap electric field is simple enough. The plasma flows from the noon sector to the midnight sector directly over the pole, and there is a return flow around the low-latitude edge of the polar cap, in the vicinity of the auroral oval, and so back to noon. See Figure 5.1(a). The speed of flow is typically several hundred m s1. The flow over the polar cap corresponds to the motion of open field-lines from the cusp to the tail (Section 2.4.1), and the return flow corresponds to the sunward flow of closed field-lines down the flanks of the magnetosphere. However the co-rotation effect, conveniently represented by the co-rotation electric field (Section 2.4.4), must also be included, and then the flow pattern becomes distorted as in Figure 5.1(b). The two

5.1 Circulation

Figure 5.1. Plasma convection at high latitude. (a) Polar convection pattern without corotation. (R. W. Spiro et al., J. Geophys. Res. 83, 4255, 1978.) (b) Examples of convection paths of plasma at 300 km altitude in the northern hemisphere under the combined electric fields due to the magnetosphere and co-rotation. The large dots indicate the starting points used in the calculations. The time between successive dots is 1 h, except for the return to the starting point. Each path is an equipotential, whose value is indicated. The boundary of the polar cap is a circle (not marked) of radius 15°, centered 5° towards midnight from the geomagnetic pole. (After S. Quegan et al., J. Atmos. Terr. Phys. 44, 619, Copyright 1982, with permission from Elsevier Science.)

circulation cells are now different, and the evening cell is particularly affected because here the return flow and the co-rotation act in opposite directions. There are some field-lines that follow long, complicated paths, while others may circulate endlessly in small vortices. All these features have ionospheric consequences. In addition the whole pattern is constantly changing in response to variations of the solar wind. The IMF exerts a major influence on the circulation pattern of the polar ionosphere. Because of the stronger coupling at the magnetopause, the magnetosphere circulates most strongly when the IMF has a southward component (Figure 2.19). This is the situation in Figure 5.1. The control exercised over the drift by the IMF has been proved by measurements of the drift at high latitude at times when satellites were situated in the solar wind just outside the bow shock (Willis et al., 1986; Todd et al., 1986). There is a more complicated pattern of circulation when the IMF is northward, but its nature has been more controversial. Various two-, three-, and four-cell patterns have been proposed. Several agreed features distinguish it from the pattern for southward IMF. (1)

It is more structured, but the speeds are lower.

(2)

The region of moving plasma is restricted to higher latitudes.

(3)

There is a region of sunward convection at the highest latitudes.

229

230

The high-latitude F region

MERGING REGIONS FOR By > 0 Bx > 0 Dominant Merging Region

AWAY SECTOR Bx < 0 Dominant Merging Region

MERGING REGIONS FOR By < 0 AWAY SECTOR Bx > 0 Dominant Merging Region

Bx < 0 Dominant Merging Region Figure 5.2. Geometry of the IMF and the geomagnetic field viewed from the Sun. Regions of preferred merging for various orientations of the IMF are indicated by shaded boxes. The principal merging region changes its location according to the “Sun–Earth” (Bx) and “east–west” (By) components of the IMF. (R. A. Heelis, J. Geophys. Res. 89, 2873, 1984, copyright by the American Geophysical Union.)

The east–west component of the IMF (usually called By ) also affects the circulation, presumably because of shifting connection regions at the magnetopause (Figure 5.2). Figure 5.3 shows versions of the circulation patterns of the northern polar region according to the directions of the north–south and the east–west components of the IMF. Versions (a) and (b) agree that the circulation is generally weaker when Bz is positive (i.e. northward), though the details differ. The influence of the east–west component over the form and size of the cells is clear in version (a) but less so in version (b). The latter shows two-cell patterns throughout, whereas in (a) three- or four-cell patterns appear when the IMF is northward. Rich and Hairston (1994) have made a comprehensive compilation of poten-

5.1 Circulation

(a)

231

By < 0

By = 0

By > 0

Bz % 1nT

 50°

Bz & 1nT

12

Auroral oval

06

18

LT = 0hrs

Figure 5.3. Patterns of the circulation of the high-latitude F region in the northern hemisphere for various orientations of the IMF. The viewpoint is that of an observer looking down on the polar region. In each diagram noon is at the top and the geomagnetic pole is in the center. (a) A conceptual picture based on various studies including European incoherent-scatter radar data. (After S. W. H. Cowley and M. Lockwood, Ann. Geophysicae 10, 103, 1992, copyright notice of Springer-Verlag.) The two-cell pattern for southward IMF (top row) gives way to three- or four-cell patterns when it is northward (bottom row). The columns are respectively for when the east–west component is directed towards the west, is zero, and is directed towards the east. (b) Results from a HF radar in North America, for a moderate level of disturbance (Kp from 2 to 3 inclusive). In these patterns the IMF is northward at the top, southward at the bottom, westward on the left and eastward on the right. (J. M. Ruohoniemi and R. A. Greenwald, J. Geophys. Res. 101, 21 743, 1996, copyright by the American Geophysical Union.) The two-cell pattern dominates throughout, though with differing magnitude.

tial distributions (equivalent to flow diagrams) from satellite measurements recorded between 1988 and 1990, divided according to season, and to magnitude and orientation of the IMF. Using advanced ionosondes at polar-cap stations in northern Canada, Jayachandran and MacDonald (1999) find a marked seasonal variation in the flow pattern. When the IMF is southward, the central region of the flow, that which passes over the pole, is observed to be towards magnetic midnight in winter but towards 2000 local magnetic time in the summer, with a gradual transition between. For northward IMF, those times are about 2 h earlier (i.e. 2200–1800 LT). There is (of course!) a considerable spread about these trends on individual days. The effect of the east–west IMF component should be in opposite directions in northern and southern hemispheres. Cases studied by Dudeney et al. (1991) using HF coherent radar (Section 4.2.2) appear to verify this. Lu et al. (1994), using

The high-latitude F region

232

Figure 5.3. (cont.)

magnetogram interpretation combined with incoherent-scatter radar, distinguish three cases. (1)

The northern and southern patterns of circulation are mirror images when the IMF is southward.

(2)

When the IMF has a northward component that is smaller in magnitude than the east–west component, the patterns are similar to each other but of different intensities.

(3)

If the IMF is strongly northward, the patterns in the summer and winter polar caps are very different.

Figure 5.4 illustrates these patterns. Cowley and Lockwood (1992) point out that the polar circulation is driven by two components, one the dayside coupling between the solar wind and the geomagnetic field, and the other the night-time ionosphere’s reaction to changes in

5.1 Circulation

Figure 5.4. Convection in northern and southern hemispheres under various conditions of the IMF: (a) southward component but east–west component larger, (b) northward component but east–west component larger, and (c) northward component with smaller east–west component. The plots show the equipotentials, which are also the streamlines of the polar flow. (G. Lu et al., J. Geophys. Res. 99, 6491, 1994, copyright by the American Geophysical Union.)

233

The high-latitude F region

234

the magnetotail. The latter are expected to be delayed 30–60 min behind a change in the IMF. It is not surprising, then, that it takes some time for the circulation to settle into a new pattern following a change in the IMF. According to Hairston and Heelis (1995), analyzing a limited number of cases, a new convection pattern appeared 17–25 min after the IMF turned from northward to southward. For a northward turning the lag was 28–44 min. Since the IMF is always changing to some extent, there will obviously be some times when the polar circulation is in a state of transition and will not conform to any particular model. A discussion of observations of high-latitude convection is given by Kelley (1989). There can be an abrupt change of plasma speed, or even a reversal of direction, across the boundaries of circulation cells, particularly when the IMF is northward. The plasma drift is equivalent to an electric field (as measured by a stationary observer), which is communicated along the field-lines to the E region. Therefore the Pedersen current (Section 1.5) in the E region also alters abruptly. To maintain continuity, current then flows up the field lines as a Birkeland current. See Figure 5.5. The corresponding downward flow of electrons is probably the cause of the sun-aligned arcs observed in the polar cap when the IMF is northward (Section 6.3.2).

5.2

The behavior of the F region at high latitude

5.2.1

The F region in the polar cap The tongue

Figure 5.6 shows F-region critical frequencies measured with ionosondes within the northern polar cap. As would be expected, the values are generally much larger in the sunlit region than they are in the dark. On this occasion there is also a tongue of ionization, drawn out from the day side, over the pole, and into the night sector. This disrupts the pattern that we might have expected, which ought to exhibit a sharp gradient between the sunlit and dark regions. The tongue is most pronounced near the equinoxes, when the terminator crosses the polar cap. The tongue may be broken into dynamic patches, which will be discussed in more detail later in this chapter. The polar F region is at its most variable when it is at its darkest – during winter, and when the magnetic pole is anti-sunward of the geographic pole. Critical frequencies can be very low: values of f0F of approximately 2–3 MHz (electron densities from several times 104 to 105) are common, and f0F1 MHz (implying a peak electron density as low as 1.4 104 cm3) has been reported from ionosonde data (Whitteker et al., 1978). The lowest values occur in the dark, anti-sunward, part of the polar cap and generally near local midnight. (Also see Section 5.5.)

5.2 The F region at high latitude

Figure 5.5. (a) Field-aligned current due to velocity shear in a magnetoplasma. B, magnetic field; v, velocity; E, electric field; IH, IP, and I
: Hall, Pedersen, and field-aligned currents. (b) The field-aligned current associated with the polar-cap aurora at the boundary between circulation cells. (Reprinted from H. C. Carlson et al., Adv. Space Res. 8, 49, copyright 1988, with permission from Elsevier Science.)

The UT effect A remarkable observation is that the daily variations of the F region depend on universal time as well as on local time. There is, for example, a daily variation at the South Pole, even though the solar zenith angle is virtually constant there over 24 h. The electron density there, as elsewhere in the Antarctic, peaks about 0600 UT, which happens to be near magnetic midnight.

235

236

The high-latitude F region

Figure 5.6. Maps of the F-region critical frequency (f0F2) showing the development of a “sporadic-F” event on 12 October 1957. (G. E. Hill, J. Atmos. Sci., 20, 492, 1963.) The plots are successively for 1700, 1800 and 1900 UT and the sunlit hemisphere is at the bottom of each plot. The contours range between 4 and 13 MHz.

5.2 The F region at high latitude

The explanation of the UT variation depends on the separation of the geographic and magnetic poles. The neutral-air wind in the thermosphere blows over the polar regions generally away from the Sun. At 0600 UT in the Antarctic and 1800 UT in the Arctic the geographic pole is on the midnight side of the magnetic pole, and the drag of the neutral particles against the ions therefore acts to lift the ionosphere (as described in Section 1.3.4). The rate of recombination of ions is thereby reduced and the net ion density is increased. This is also the time of day when the largest amount of the geomagnetic polar cap is sunlit, and it is therefore when the circulation pattern will be most effective at bringing solar-produced ionization over the pole. It is significant that, in the northern hemisphere, where the separation between the poles is smaller, the UT effect is less pronounced than it is in the south. In the polar cap the F1 layer can be almost as strong as the F2 layer, and on occasions it may be even stronger. This produces the so-called “G condition” on ionograms. 5.2.2

The effect of the polar cusps

On the day side of the Earth are two regions, one in each hemisphere, where the geomagnetic field-lines provide a direct connection between the ionosphere and the magnetosheath (Section 2.2.5). In the simplest models of the magnetosphere, in which there is no circulation, they correspond to the neutral points on the surface of the magnetosphere. Field-lines at lower latitude are closed, whereas those at higher latitude are “open”, connecting with the solar wind and the IMF or sweeping back into the magnetotail. In more realistic, dynamic, models (Sections 2.4.1–2) the cusps are where the dayside field-lines open before being swept over the poles (Figure 5.7(a)). The cusps are significant regions of the magnetosphere and also of the ionosphere. In the ionosphere the cusp regions have several signatures. (1)

Charged particles with energies similar to those in the magnetosheath may be detected. Whereas the cusps are typically located near 78° geomagnetic latitude, and are about 5° wide, the particle observations show the cusps extending over all daylight hours and merging into the auroral oval (Section 6.2.1). There is also a second, smaller, region extending only a few hours from local noon. The particle flux from the magnetosheath is highly variable over short times (or over small distances, since these observations come from orbiting satellites).

(2)

Luminous emissions at 630 nm are enhanced, indicating the occurrence of low-energy excitation of the upper atmosphere. Emissions typical of the aurora (Section 6.3.3) are actually reduced – a feature sometimes called the noon gap. These photometric observations reveal a considerable variation in the latitude of the cusp, from 84° under very quiet geomagnetic conditions to 61° under very disturbed conditions.

237

238

The high-latitude F region

(a)

(b)

Figure 5.7. Aspects of the polar cusp and its F-region effects. (a) Details of the polar cusp: MS, magnetosheath; LLBL, low-latitude boundary layer; EL, entry layer; and PM, Plasma mantle. (G. Haerendel et al., J. Geophys. Res. 83, 3295, 1978, copyright by the American Geophysical Union.) (b) A tomographic image of the F region on 14 December 1996 at 10:46 UT showing signatures arising from magnetic reconnection. The dashed line marks the boundary between closed and open field-lines, and other features are described in the text. (I. K. Walker et al., Geophys. Res. Lett. 25, 293, 1998, copyright by the American Geophysical Union.)

5.2 The F region at high latitude

(3)

Owing to the influx of particles from the magnetosheath the density and temperature of the ionosphere is increased and there is a greater degree of irregularity. Owing to the opening of the field-lines, ionospheric plasma may flow out into the magnetosphere, where its ionospheric origin has been recognized from its temperature and composition.

(4)

Magnetic pulsations of type Pi2 (of period approximately 30 s) are enhanced. (See Section 2.5.6.)

The image of Figure 5.7(b), which was obtained by the tomography technique (Section 4.4.3), shows features of the F region due to magnetic reconnection at the cusp. The boundary between closed and open field-lines is marked, and, from scanning-photometer observations, Walker et al. (1998) were able to identify ionospheric effects due to (1) precipitation of electrons from the ring current on the last closed field-lines, (2) a downward field-aligned current on the first open fieldlines, and (3) dispersion of precipitating soft ions on the flux tubes convecting poleward. The last effect shows up as the increasing height of the layer maximum. 5.2.3

The polar wind

The circulation of the magnetosphere carries field-lines from the closed region, through the cusp, and into the polar region where they are open to the solar wind or go deep into the tail of the magnetosphere. These tubes of force lack an effective outer boundary. Furthermore, the scale height is large for light ions at high temperature (Equations (1.3) and (1.46)). Therefore the ionospheric plasma may readily flow upward, and, in the absence of a boundary, the flow may continue as long as the tube remains open. A steady outward flow is one of the solutions of Equation (1.43), describing the motion of a minority gas under the forces due to a pressure gradient and gravity. As was pointed out in Section 1.3.4, the separation between the heavy ions (oxygen) and the electrons produces an electric field directed upward. When light ions (hydrogen and helium) are also present, they are accelerated by this electric field, which tends to drive them upward. Detailed consideration shows that gravitational attraction is able to bring about a state of hydrostatic equilibrium (Equation (1.47)) in the oxygen ions, but that H is light enough for the electric field to cause the dynamic equilibrium state having a steady outflow above some altitude. He may also flow out, though to a lesser extent. This continuous outflow of light-ion plasma is the polar wind. In theory the flow can even reach supersonic speeds, but the details depend on what is assumed about the flow speed at a great distance. The term “polar wind” is sometimes restricted to the supersonic regime, in which case subsonic flow would be a “polar breeze.” The flow is limited by collisions with stationary ions, and by the rate of production of H by the charge exchange between oxygen ions and neutral atomic hydrogen (Equation (1.69)) in the topside ionosphere. Since the concentration of

239

The high-latitude F region

240

O is far from uniform over the polar cap, the polar wind must be similarly variable. The lighter ions are the most affected by the outflow, and it is commonly observed in satellite measurements that the concentration of H is greatly reduced relative to the O in the topside ionosphere over the polar caps. The upward speed can be several km s1. The flux of H is heated by collision with the heavy ions, and its temperature is significantly raised. The theory of the polar wind has been reviewed by Raitt and Shunk (1983). Figure 5.8, from that paper, shows computations illustrating the reduction of topside ion density and the upward drift speed of H for various assumptions about the outer boundary. One important point established by satellite observations is that the polar wind is a significant source of the plasma in the magnetosphere. That material then convects with the magnetospheric circulation and eventually reaches the plasma sheet at a distance from the Earth that depends on the nature of the ion but is estimated generally to be within 50RE. Figure 5.9 illustrates some aspects of the interchange of plasma between ionosphere and magnetosphere. Plainly, the polar wind is a mechanism that removes ionization from the polar ionosphere from above. Typical loss rates are 3 108 cm2 s1 for H ions and 3 107 cm2 s1 for He. It is secondary to the loss by recombination acting most effectively in the lower F region, and for which electron-content observations lead to estimates in the range 109–1010 cm2 s1 at middle latitudes. 5.2.4

The F layer in and near the auroral oval

On a long-term view the F region in the auroral zone exhibits properties similar to those typical of middle latitudes. Figure 5.10 shows how the average electron density near the peak of the layer varies diurnally during summer and winter at sunspot maximum and minimum. These measurements are by incoherent-scatter radar at Tromsø, Norway (geographic latitude 69.6° N, invariant latitude 67°, L6.5). The winter anomaly (Section 1.4.5) is seen at sunspot maximum but not at sunspot minimum, which is also the case at mid-latitude. The electron density is larger in the months either side of the winter solstice, indicating the presence of a semi-annual anomaly as well (Farmer et al., 1990). In addition, there are additional factors that make the ionosphere more irregular in both time and distance. In the poleward part of the auroral oval and extending several degrees into the polar cap, the electron density may be enhanced by the precipitation of low-energy electrons (maintaining the F-region penetration frequency at at least 3 MHz). There may be large variations over short distances, probably due to irregularity in the intensity of the particle precipitation. The precipitation (of particles with energy 300 eV) is particularly strong in the cusp region (75°–80° magnetic), where the penetration frequency may be increased by several megahertz. This precipitation creates irregularities tens of kilometres across, which then break down into smaller structures (tens of metres

5.2 The F region at high latitude

241

4000

ALTITUDE (km)

3000

2000

1000 O

H 0 101

102

103 DENSITY

104

105

106

(cm3

)

4000

ALTITUDE (km)

3000

2000

1000

0 2

2

6

10

14

18

VELOCITY (KM.SEC1)

Figure 5.8. Theoretical properties of the polar wind, showing the density of H and the field-aligned drift speed. The various curves are for a range of H escape speeds between 0.06 and 20 km s1 at 3000 km. The range of O is given by the shaded region. (W. J. Raitt and R. W. Schunk, Energetic Ion Composition in the Earth’s Magnetosphere, Terra Scientific Publishing, Tokyo, 1983, p. 99.)

The high-latitude F region

242

Figure 5.9. Ionospheric sources of plasma for the magnetosphere. Ions leaving the high latitudes tend to separate according to mass. They may subsequently be trapped in the plasma sheet and drift towards the Earth, being energized by betatron acceleration. Computations indicate that the ionosphere is a significant source of magnetospheric plasma. (After C. R. Chappell, Rev. Geophys. 26, 229, 1988, copyright by the American Geophysical Union.)

across or less) as they drift in the general convection (Muldrew and Vickrey, 1982). No doubt the transport of plasma over the pole also contributes significantly to the ionization observed in the vicinity of the auroral oval near midnight. Structures moving over the pole, provided that they continue to drift in the convection pattern (Figure 5.1), are expected to become distorted on reaching the Harang discontinuity and be diverted eastward or westward along the oval (Robinson et al., 1985). As will be discussed in Section 5.3.2, it is clear from their properties that at least some of the structures in the oval are not of local origin. On the equatorward side of the auroral oval the F region tends to be depleted of ionization. This is the “main trough”, sometimes known by its older name of “mid-latitude trough.” The depletion in the trough can be as much as by a factor of ten, though it is often not so great. It is a complex feature, created by the combination of loss processes and the circulation pattern in the region where the highand mid-latitude ionospheres meet. The trough is considered in detail in Section 5.4.

5.3

Irregularities of the F region at high latitude

5.3.1

Introduction

Spatial irregularities are a common feature of the atmosphere and ionosphere, and their scales of variation cover a wide range in both time and distance. The existence of F-region irregularites has been known for at least 40 years from their

Electron density (10 3 cm –3)

5.3 Irregularities

243

(a) 600 500 400 300 200 100 0 Yearly Average 250 –300 km

Electron density (10 3 cm –3)

4

8

12 16 20 24

Summer 250 –300 km

4

8

12 16 20 24 Local time

Winter 250 –300 km

4

8

12 16 20 24

(b) 600 500 400 300 200 100 0

Yearly Average 250 –300 km

4

8

12 16 20 24

Summer 250 –300 km

4

8

12 16 20 24

Winter 250 –300 km

4

8

12 16 20 24

Local time

Figure 5.10. Yearly, summer, and winter diurnal variations of electron density near the peak of the F layer at Tromsø. (a) sunspot maximum (August 1981–August 1983), (b) sunspot minimum (April 1986–March 1987). (Reprinted from A. D. Farmer et al., J. Atmos. Terr. Phys. 52, 561, copyright 1990, with permission from Elsevier Science.)

The high-latitude F region

244

effects on trans-ionospheric radio propagation, originally in observations of radio stars, though our knowledge of them is still incomplete. We are here concerned with the two principal kinds affecting the F region: enhancements extending over tens and even hundreds of kilometers, which can be observed by incoherentscatter radar and other ionospheric techniques; and irregularities smaller than about 10 km, which, by a diffraction mechanism, produce in propagating radio waves the phenomenon of scintillation. 5.3.2

Enhancements: patches and blobs

We first consider enhancements of large size occurring in the polar cap and the auroral zone. They may be 50–1000 km across and are remarkable for their high plasma density. Even when they are observed during the polar winter night, their density can be more typical of that of the daylit mid-latitude ionosphere. There are several techniques by which they may be observed. Some of the first reports came from polar-cap ionosonde data, when they were described as “sporadic-F.” Figure 5.6 showed a good example in which the evolution of a patch may be seen. Speeds of 2000–5000 km h1 (500–1400 m s1) were reported. The cause of the motion was correctly interpreted as being due to an electric field, but it was (incorrectly) supposed to arise in the E region rather than the magnetosphere. While much of the information about patches has come from ionosondes, they can also be detected by virtue of the 630-nm airglow which they emit. Other techniques, such as incoherent-scatter radar and tomography, have been significant in the more recent studies of enhancements.

Patches Enhancements within the polar cap are generally called patches. They are seen during the winter night under disturbed conditions, and the F-region electron density may be increased by as much as a factor of ten above the background, which would typically be about 105 cm3. They tend to be stronger at times of high sunspot number. It seems clear that this type of enhancement is not produced locally, but was formed some distance away and has then drifted in the polar convection to the point of observation. Because the F region decays only slowly by recombination, the lifetime of the patches should be quite long enough for them to cross the polar cap at a speed of several times 100 m s1 (up to 1 km s1) from a source on the day side. This possibility has been verified by computations that have also demonstrated how a change of polar circulation, for instance due to an increase in the flow of the solar wind or a sudden change in the IMF (Anderson et al., 1988; Sojka et al., 1993), can detach plasma from the dayside cusp region and carry it over the pole into the midnight sector along a path such those shown in Figure 5.1(b). Lockwood and Carlson (1992) have attributed patch creation to the enhanced plasma flow during a flux-transfer event (Section 2.4.2). What happens when the

5.3 Irregularities

enhancement reaches the night sector is less clear, but it probably becomes stretched along the auroral zone in the return flow or merges into the mid-latitude ionosphere (Robinson et al., 1985). Computer modeling of the high-latitude F region (Sojka et al., 1994) suggests that, at the winter solstice, patch formation should be absent between 0800 and 1200 UT and at a maximum from 2000 to 2400 UT. From then until the equinox there should be strong patches all day. In the summer they should be considerably weaker. While much is still not understood about these larger structures of the polar ionosphere, a number of observational facts have been established about them. (a)

They are roughly circular, and between 200 km and 1000 km in size.

(b)

The patches are smaller than the gaps between them, suggesting that we should consider them as enhancements of ionization above a low background rather than as depletions within a higher background.

(c)

The degree of enhancement in a typical patch is 2–10 times the ion density of the background.

(d)

The gradients at the edges of patches are fairly sharp, on a scale of a few km to about 100 kilometers, and these gradients are the same in all horizontal directions.

(e)

The patches appear when the IMF is southward.

(f)

They move with the general plasma drift in the polar cap and at the same speed, neither overtaking the general flow nor lagging behind.

(g)

They occur during all seasons of the year but more frequently during the winter.

A different pattern is seen in the weaker circulation which occurs when the IMF has a northward component and conditions are less disturbed. At such times the airglow emissions form thin strips with noon–midnight alignment, and these drift slowly across the polar cap in a dusk-to-dawn direction. In these elongated structures the electron density is enhanced by a factor of 5–8 at times of high sunspot number, but by a smaller amount (a factor of two) near solar minimum (Buchau et al., 1983). Figure 5.11 compares the structures typically associated with northward and southward IMF.

Blobs In the auroral zone the enhancements are generally known as blobs. They are smaller than the patches in the horizontal, extending for tens of kilometers rather than hundreds. Some of them peak low in the ionosphere, in the E region or the lower F region. Figure 5.12 illustrates the structures of the ionosphere as seen by two different techniques: (a) was derived by the tomography technique (Section 4.4.3) from electron-content data in the Scandinavian sector, and (b) was obtained by a scanning incoherent-scatter radar (Section 4.2.3) in Alaska. The upper panel

245

246

The high-latitude F region

Figure 5.11. Typical irregular structures of the polar F region. (a) arcs with noon–midnight alignment and dusk–dawn drift during northward IMF (Bz 0), and (b) patches drifting towards midnight during southward IMF (Bz 0). The coordinates are corrected geomagnetic latitude (CGL) and local time (CGLT), and the heavy lines mark the auroral oval (Section 6.2). (After H. C. Carlson, private communication.)

gives an overall view showing the mid-latitude ionosphere on the left, the more structured auroral region on the right, and the main trough (Section 5.4) in between, while the two lower panels show similar features as contour diagrams emphasizing the irregularities. There is some uncertainty about the cause of blobs in the auroral zone. They seem to vary greatly in size. There is some evidence, though it is perhaps not yet definitive, that they move with the plasma drift of the auroral F region as a whole. It seems clear that more than one source is involved, since, as Figure 5.13 illustrates, they may occur over different altitude ranges. Moreover, those at the higher levels are generally cooler than their surroundings by about 10%, whereas those peaking in the lower F region tend to be hotter by about 20%. (According to the results of Burns and Hargreaves (1996), typical electron temperatures are about 1280 and 1540 K, repectively, for the two types, compared with about 1410 K for the plasma outside the blob – all these values being medians over a number of separate determinations.) It is generally assumed that the higher structures arrive as patches drifting from the polar cap (since they are also cooler than their surroundings), but the exact connection and the mechanism which breaks them up are unknown. Those blobs which are hotter and appear at lower altitudes are more likely to have been produced by particle precipitation nearer to the point of observation. Figure 5.12 shows one of these lower blobs, and also examples of the boundary blob which is situated just poleward of the main trough. The boundary blob is a long-lived feature that may continue for several hours.

5.3 Irregularities

(a)

(b)

Figure 5.12. (a) A tomographic image of the ionosphere in the Scandinavian sector, 15 October 1993, pre-midnight, showing the mid-latitude ionosphere, the main trough (Section 5.4), and the structured auroral ionosphere. (L. Kersley, private communication, 1998.) (b) Blobs and other features observed with the Chatanika incoherent-scatter radar on 11 November 1981. The time of each scan is marked, and the main trough, a boundary blob, an auroral blob, and the auroral E layer (Section 6.5.4) may be seen from south to north. A distance of 100 km is about 0.9° of latitude. Since Alaskan time is UT – 10 h, these are in the early evening. (C. L. Rino et al., Radio Sci. 18, 1167, 1983, copyright by the American Geophysical Union.)

247

The high-latitude F region

248

(a)

(b)

Figure 5.13. Three kinds of blob observed with the EISCAT incoherent-scatter radar. (a) F-region enhancement peaking at 250–400 km (cooler than the surroundings). (b) an intermediate type having an F-region peak and related E-region structure, and (c) a low-altitude blob peaking below 200 km (hotter than the surroundings). (Reprinted from C. J. Burns and J. K. Hargreaves, J. Atmos. Terr. Phys. 58, 217, copyright 1996, with permission from Elsevier Science.)

5.3 Irregularities

(c)

Figure 5.13. (cont.)

Table 5.1 compares the main properties of the various types of enhancement. A comprehensive review of high-latitude enhancements of the F region was made by Tsunoda (1988). A collection of relevant papers was published as a special section of Radio Science (1994). 5.3.3

Scintillation-producing irregularities

The irregularities of smaller scale produce scintillation phenomena in trans-ionospheric radio signals. The theory of scintillation was outlined in Section 3.4.5, where it was seen that the radius of the first Fresnel zone is an important parameter. For a radio frequency of 100 MHz the first Fresnel zone has a radius of about 1 km if the effective diffraction screen is at a height of 300 km; therefore irregularites smaller than about 1 km produce both amplitude and phase scintillation. Irregularities larger than that produce phase scintillation only.

Distribution and occurrence Scintillation occurs at all latitudes, including the polar region, but it tends to be particularly severe at and around the auroral zone (Aarons, 1982; Yeh and Liu, 1982). See Figure 5.14. (The other region of heavy scintillation is at the equator.) The auroral scintillation zone is offset from the magnetic pole and exhibits a general correspondence to the auroral oval (Sections 6.2.1 and 6.3.5), being nearer to the equator in the night sector. Both in the auroral and in the polar regions the

249

Location

Polar cap when Bz is south and KP 4.

Equatorward boundary of the auroral zone. In the midnight sector and extending to the morning and evening.

Auroral zone in the night sector

Type of irregularity

Polar cap patches Buchau et al. (1983; 1985) Weber et al. (1984; 1986) Weber and Buchau (1985)

Boundary blobs Kelley et al. (1980) Vickrey et al. (1980) Muldrew and Vickrey (1982) Rino et al. (1983) de la Beaujardiere and Heelis (1984) Robinson et al. (1985)

Auroral blobs Rino et al. (1983) Robinson et al. (1984)

Field-aligned 10–100 km north–south,

Extreme longitudinal extent but confined to 100 km in latitudinal width.

100s to 1000s of km horizontal extent. 500 km radius.

Typical size

Altitude F region

300 to 500 km

Between 200 and

Magnitude 106 cm3, about eight times the background F layer

4 105 cm3

3 105 cm3

Table 5.1. A summary of large-scale, irregular structures at high latitude

Intermittent, about 1 h

Very persistent, 12 h

2–3 h

Duration

Poleward auroral boundary.

Either reconfigured patches or semipermanent structures enhanced by soft particle precipitation

Sub-auroral latitudes equatorward of the dayside cusp. Plasma produced by solar EUV.

Origin

Zonal drift 250 m s1

Move equatorward with time, and sunward along the equatorward boundary of the auroral zone

Anti-sunward through the polar cap at 250–700 m s1

Motion

and postnoon sector.

Polar cap. Aligned with the noon– midnight meridian. Bz north.

Hargreaves et al. (1985a; 1985b)

Sun-aligned arcs Weber and Buchau (1981) Carlson et al. (1984)

a few times 100 km east– west. Wavelike structure of wavelength about 15 km.

2 105 cm3

350 km. Isolated blobs near 700 km.

‘Polar shower’ precipitation in the central polar cap.

Soft particle precipitation and possibly ‘spatial’ resonance. Source of wavelike structures is unknown. Slow dawn-todusk movement

252

The high-latitude F region

Figure 5.14. The principal regions of scintillation at L band (1.6 GHz). (S. Basu et al., Radio Sci. 23, 363, 1988, copyright by the American Geophysical Union.)

rate of occurrence and the intensity maximize at night, and there is also a daytime maximum in the auroral region only. The seasonal variation depends on the longitude. Figure 5.15 shows the seasonal and daily occurrence patterns for an auroral station in the European sector (Kiruna). The occurrence and the intensity of scintillation increase strongly with the sunspot number; the occurrence also increases with magnetic activity (Kp), but this effect is only slight in the polar cap.

The period and depth of fading The period of fading varies considerably, but is generally in the range of seconds to a few minutes. It depends on the apparent motion of the irregularities as well as on the depth of the fading. Figure 5.16 shows an example of amplitude scintillation. The intensity of amplitude fading is commonly measured using the index S4 (defined in Section 3.4.5). In these terms it depends on the radio frequency as f1.5 if the fading is not too severe, but less steeply for strong scintillation. The observed depth of scintillation also depends on the direction of propagation between the sender and the receiver (for instance from a satellite to a ground station). Increasing obliquity tends to make the fading more severe because the ray traverses a longer path through the ionosphere, thereby encountering more irregularity in total. Details depend on the form of the individual structures. It may be assumed that there will be considerable elongation along the geomagnetic field, and, according to Rino (1978), auroral irregularities are extended east–west, giving a sheet-like form. Since individual irregularities are strongly field-aligned, there is another maximum for rays traveling directly along the direction of the magnetic field (because rays traveling almost directly along the magnetic field tend to remain within a single irregularity). These effects may be seen in Figure 5.15.

5.3 Irregularities

Figure 5.15. The occurrence of scintillation over magnetic latitudes 55°–80° observed from Kiruna (64.3° N, 102.8° E CGM), September 1984–September 1986. The contours show the percentage of time for which the scintillation at 150 MHz exceeded S4 0.2. The contours 1–5 represent 25%, 35%, 45%, 55% and 65% respectively. (a) Variation with month. Note the summer maxima. (b) Variation with local time, low magnetic activity (Kp %1). (c) Variation with local time, moderately high magnetic activity (Kp &4). (L. Kersley et al., Radio Sci. 23, 320, 1988, copyright by the American Geophysical Union.)

253

254

The high-latitude F region

Figure 5.16. Examples of scintillation fading observed in Alaska in transmissions at (a) 140 MHz and (b) 360 MHz from a geosynchronous satellite (ATS-6), on 30 March 1979. The satellite was at low elevation to the south, and the raypath crossed the F region at about 60° geomagnetic latitude. The fading is considerably greater at the lower frequency, with a ratio of almost four between the scintillation indices (c).

5.3 Irregularities

Figure 5.17. The S4 scintillation index at Hornsund, Svalbard (invariant latitude 73.4° N) at various local magnetic times. The bars indicate the standard deviation. The latitude of the receiving station is marked with an arrow, and horizontal lines indicate the typical latitude of the auroral oval at that time of day. (A. W. Wernick et al., Radio Sci. 25, 883, 1990, copyright by the American Geophysical Union.)

Average values of S4 at 137 MHz over a range of latitude and at various times of day (geomagnetic local time) are shown in Figure 5.17. All these measurements were made between autumn and early summer (October to May) at Hornsund, Svalbard (invariant latitude 73.4°), whose position is marked on the plots with an arrow. Since the magnetic field is nearly vertical over the polar cap, the propagation path is closest to the magnetic-field direction when the satellite is at the same latitude as the receiving station. The maximum in S4 at the latitude of Hornsund is clearly present at night. A value of S4 equal to 0.25 corresponds to fading with about 1 dB standard

255

256

The high-latitude F region

deviation. However, the fading may be much more severe on occasion, and especially so in the auroral zone. Table 5.2 shows the incidence of intense scintillation at the auroral station Narssarssuaq (Greenland) at two radio frequencies. Note that severe fading is considerably more common under magnetically disturbed conditions and at night. At the highest latitudes (82° magnetic latitude) the scintillation is associated with polar arcs (Section 6.3.2), and fading of more than 28 db (peak–peak) has been observed at 250 MHz.

Spectrum The irregularities causing scintillation may be considered as an irregular, spatial distribution that is drifting but also evolving in time. The temporal variation observed at a single site includes the intrinsic time variation, but the main part of the variation is likely to be due to the relative motion between the irregularities and the probing signal. A satellite in low orbit converts the spatial spectrum along the orbit to a temporal spectrum according to the orbital speed. In the case of a geostationary satellite, the temporal change arises from the drift of the irregularities through the satellite-to-ground ray. Examples of the intensity spectrum of 137-MHz scintillations recorded at Hornsund are shown in Figure 5.18. Since the transmitting satellite, HiLat, was in orbit at an altitude of 800 km, we expect that the time variation will be due mainly to the motion of the satellite across the spatial irregularities – though exact conversion would require knowledge of the irregularity motion as well. The large maxima in Figure 5.18 are due to the effect of diffraction (Section 3.4.5), which prevents large-scale phase irregularities generating amplitude scintillation at the ground. The peak marks the Fresnel frequency. The falling part of the spectrum represents a range of spatial size from about 700 to 130 m (when the satellite is overhead). These are power-law spectra, as is commonly the case, and, in the Hornsund data set, the average spectral index, q, is generally between 2 and 3. That is, for a factor of ten in fading frequency the intensity declined by a factor between 100 and 1000 (20–30 dB). Amplitude fading tends to be dominated by the Fresnel frequency. Table 5.2. Depth of scintillation at Narssarssuaq Percentage occurrence &12 dB at 137 MHz

&10 dB at 254 HMz

Kp

Day

Night

Day

Night

0–3 3

2.9 19

18 45

0.1 0.9

2.6 8.4

5.3 Irregularities

Intensity Spectrum (dB)

(a)

257

30

40

50

60 S4 = 0.226 q = 2.792 ± 0.200 70 0.1

1

10

100

Frequency (Hz)

Intensity Spectrum (dB)

(b)

20 30 40 50 60 S4 = 0.854 q = 2.359 ± 0.140 70 0.1

1

10

100

Frequency (Hz) Figure 5.18. Typical spectra of amplitude scintillation in 137-MHz signals received from the HiLat satellite at Hornsund: (a) 24 April 1986 and (b) 28 October 1985. q is the spectral index. (A. W. Wernick et al., Radio Sci. 25, 883, 1990, copyright by the American Geophysical Union.)

Direct measurements Spatial fluctuations of electron density can be measured in situ using satelliteborne probes, though the high velocity of an orbiting satellite limits the structural detail that can be resolved in this manner. In Figure 5.19, which shows measurements of ion (and therefore electron) density made on an orbiting satellite, the fluctuations are as much as 20% of the mean. In some cases it has been observed that the small-scale irregularities which produce scintillation are located at the edges of large-scale enhancements, and Figure 5.19 is such an example. There are mechanisms (such as the gradient-drift and Kelvin–Helmholtz instabilities) that can cause a large patch to break up at the edges, thereby generating smaller ones, which may break up in turn. By this means the larger structures can progressively break down to give smaller ones in a

258

The high-latitude F region

Figure 5.19. (a) Relative irregularity and (b) ion density measured on a satellite crossing the polar cap. In (a) the variation N was taken with respect to a linear least-squares fit to the electron density measured for 3 s; the plotted N/N therefore refers to irregularities smaller than about 25 km. (After S. Basu et al., The Effect of the Ionosphere on Communication, Navigation, and Surveillance Systems (ed. Goodman), p. 599. IES’87, National Technical Information Service, US Government Printing Office, Springfield, Virginia, 1987.)

cascade process. The relationship between large structures in the polar cap and the incidence of radio scintillation has been discussed by Buchau et al. (1985).

Modeling For forecasting purposes, empirical models have been developed to represent the intensity of scintillation at high (and other) latitudes. Scintillation depends on the spatial variation of the electron content, rather than on its actual value (see Section 3.4.5), and varies with parameters such as magnetic latitude and longitude, time of day, season, magnetic activity, and sunspot number. The high-latitude model proposed by Secan et al. (1997) is derived from scintillations observed in the 137.67-MHz transmissions from several orbiting satellites (Wideband, HiLat and Polar BEAR) received at stations in Greenland, Norway, Canada, and the USA (Washington State) between 1976 and 1988. Figure 5.20 gives an example showing a quantity called the irregularity strength parameter (defined as the power-spectral density of the variation in electron density at the wave number for 1 km, multiplied by the thickness of the irregular region), which is propor-

5.3 Irregularities

Figure 5.20. Contours of the irregularity parameter CkL for 2300 UT on 21 July at solar maximum (sunspot number 175) and high geomagnetic activity (Kp 6), from model version 13.04. CkL is the height integrated power spectrum of irregularities for a periodicity of 1 km, here shown as the logarithm, and is proportional to the variance of the electron content. (J. A. Secan et al., Radio Sci. 32, 1567, 1997, copyright by the American Geophysical Union.)

tional to the variance of the vertical electron content. The calculation of the depth of scintillation is then based on the theory of a phase screen (Section 3.4.5) with an assumed power-law spectrum of intensity. The scintillation index derived depends also on the propagation direction, the radio frequency of the signal, the speed at which the propagation path crosses the plasma irregularities, and assumptions about the height of the effective phase screen and the form of the irregularities. The data compilation underlying this model revealed several significant features. (1)

The high-latitude scintillation region has a well defined boundary, across which the irregularity strength increases by more than a factor of ten.

259

The high-latitude F region

260

(2)

The peak of the enhancement associated with the auroral zone lies 2° poleward of the boundary of particle precipitation at midnight, but 14° poleward of it at noon. (The aforsaid precipitation boundary is the equatorward edge of the region where auroral electrons of energies 50 eV to 20 keV are precipitated, as determined by Gussenhoven et al. (1983). See also Figure 6.6.)

(3)

Equatorward of the scintillation boundary there is a transition region, most evident from 0800 to 1600 magnetic local time, before the irregularity strength assumes the lower values typical of middle latitudes.

(4)

The auroral enhancement has maxima near midnight and noon, both of which become more intense with increasing Kp. The night maximum occurs later as Kp increases, and the day maximum occurs later with increasing sunspot number.

(5)

The polar cap contains a strong enhancement after noon, and a minimum after midnight. The overall level of irregularity in the polar cap increases with the sunspot number and decreases with increasing Kp.

These features do not show up clearly in Figure 5.20, but are illustrated in Figure 1 of Secan et al. (1997), to which the reader is referred for further details of the model and its use.

5.4

The main trough

5.4.1

Introduction

An ionospheric trough is a region of depleted ionization, limited in width but extended in the east–west direction, with more intense regions to the north and south. We deal here only with depletions that are observed regularly and appear to be permanent or semi-permanent features of the F region, accepting that they vary in intensity and location. A depletion that is not elongated would be described as a hole. The reader should be warned that the terminology of troughs and holes has been subject to some ambiguity in the literature, as may well happen when phenomena have not yet been fully defined. Most investigations relate to an F-region trough that seems to mark the boundary between the mid-latitude and highlatitude ionospheres. Originally this was called the “mid-latitude” trough, a term that continues to be used. It has also been called the “main” trough, and that term will be prefered here, first to emphasize its importance as the principal trough-like feature of the F region, and second because its occurrence is by no means restricted to middle latitudes. Under a blanket definition of middle and high latitudes, the trough would appear sometimes in one and sometimes in the other, and it is probably more helpful to consider it as the variable boundary between the

5.4 The main trough

high- and middle-latitude regions of the ionosphere, at least on the night side of the Earth. Various other troughs and holes that are wholly within the highlatitude region are observed, and these will be called simply high-latitude troughs or holes, as the case may be. Ionospheric troughs are depletions of the heavy ions, principally O. They are related to, but not identical to, depletions of light ions (H and He) in the topside ionosphere and the protonosphere as far as the equatorial plane. (The inner edge of the main depletion in the plasmasphere is, of course, the plasmapause – Section 2.3.2.)

5.4.2

Observed properties and behavior of the main trough Observations

The main trough was first observed in the early 1960s by the topside sounder Alouette 1 (Muldrew, 1965; Thomas et al., 1966) as a local depletion of electron density when the satellite crossed the frontier between Canada and the USA. In those early days it was sometimes known as the Canadian-border effect. Since then it has been studied from the ground by a variety of techniques, particularly electroncontent measurement, incoherent-scatter radar, and by using ionosondes. An example from Dynamics Explorer 2, showing the variations of electron density and temperature across the northern high-latitude region at the height of the satellite (733–371 km), is shown in Figure 5.21. The main (mid-latitude) trough appears just after 0931 UT near 60° invariant latitude, and two other troughs are seen at higher latitudes. The electron temperature was enhanced in the main trough, and this is typical. The main trough is wider in the example of Figure 5.22, which was derived from ISIS-2 topside ionograms. Here the trough is more than 15° wide, and the complexity of detail in the trough region is indicated. The numbers 1–8 pick out a number of features, namely: (1)

a latitudinal variation in the mid-latitude ionosphere;

(2)

the equatorward wall of the trough;

(3)

the trough minimum;

(4)

the poleward wall of the trough (which is often sharp, as it is here);

(5)

an auroral enhancement;

(6)

a decline on the poleward side of the auroral oval; and

(7, 8) structure within the polar cap. Troughs are also observed in the electron content but generally they do not exhibit the sharp gradients or as much detail as those observed by satellite-borne probes or topside sounders. Some examples are given in Figure 5.23. The reason for the different appearance is probably that the electron content is an integral of the electron density rather than the value at one height. Figure 5.23 shows selected

261

Figure 5.21. Latitude profiles of electron density (left-hand scale) and electron temperature (right-hand scale) measured on the satellite DE-2, 22 November 1981, showing mid-latitude (main) and high-latitude troughs. (Reprinted from A. S. Rodger et al., J. Atmos. Terr. Phys. 54, 1, copyright 1992, with permission from Elsevier Science.)

5.4 The main trough

H(MAX) (km) 106

263

500

500

400

400

300

300

200

200

8 6

ELECTRON CONCENTRATION (EL/CM3)

4 5

2 1

105

8 6

6

2 4

2

5

1 4

4 3

8

6

3

7

8 7

N(550) N(650) N(750) N(850) N(950)

2

104

N(HMAX) N(450)

8 6 4

N(HSAT)

2

103

8 6 4 2

102 40

60 70 50 CORRECTED GEOMAGNETIC LATITUDE

80

Figure 5.22. Features of the main trough, recorded by the topside sounder ISIS-2 on 18 December 1971. The local time is near midnight. (M. Mendillo and C. C. Chacko, J. Geophys. Res. 82, 5129, 1977, copyright by the American Geophysical Union.)

examples in which the trough is clearly defined. Some troughs are more structured than these, and some have a second minimum. Figure 5.24 indicates by means of a schematic diagram the structure of the trough as it affects the electron isopleths on the bottom side of the ionosphere near 60° geomagnetic, and in Figure 5.25 we see them both on the topside and on the bottom side obtained by tomographic analysis of electron-content data.

A summary of principal properties (northern hemisphere) Following Moffett and Quegan (1983), the location and occurrence of the trough (in the northern hemisphere) may be summarized as follows. ■

The trough is primarily a night-time phenomenon, extending from dusk to dawn. It has on occasion been observed at all local times.

264

The high-latitude F region

Figure 5.23. Troughs in electron content on four separate occasions when the trough was narrow and well defined. The observations were made in Scandinavia and time is marked in UT. (Reprinted from L. Liszka, J. Atmos. Terr. Phys. 29, 1243, copyright 1967, with permission from Elsevier Science.)

Figure 5.24. A sketch of the trough as it often appears near Halley, Antarctica. (J. R. Dudeney et al., Radio Sci. 18, 927, 1983, copyright by the American Geophysical Union.)

5.4 The main trough

265

Figure 5.25. The trough as seen by tomography. Results are from the Scandinavian sector, early afternoon, 17 November 1995. Note the narrow upward extension on the poleward side. (L. Kersley, private communication, 1998.)

■

It is observed most regularly in the winter and equinoctial seasons. It occurs more rarely in summer, and then only near local midnight.

■

The poleward edge of the trough, which is usually sharp, is close to the equatorward edge of the region of diffuse aurora.

■

The trough moves to lower latitudes as the night proceeds. Under geomagnetically quiet conditions it can turn back to higher latitudes during the early morning.

■

It also moves to lower latitude with increasing geomagnetic activity; solar activity as such appears to have no effect.

■

There is no general agreement regarding the depth of the trough or its width, or on how these properties vary with the time of day.

Formulae for variations with time and magnetic activity Knowledge of the locations both of the trough minimum and of the poleward edge is important for radio communication at high latitudes and for trans-polar paths. The position of the trough minimum as a function of local time and Kp has been expressed by the linear relationship 'T '0 aKp bt,

(5.3)

where 'T is the invariant latitude of the trough minimum, '0 is its invariant latitude at midnight (t0) if Kp 0, t is the local time in hours reckoned from midnight (negative before, positive after), and a and b are coefficients. The values of '0, a, and b given in Table 5.3 were derived from independent sets of observations. These formulae have the merit of simplicity, but they cannot give the whole story because there is no provision for poleward motion in the morning. Halcrow and Nisbet (1977) and Spiro (1978) have derived equations of non-linear form. Equation (5.3) implies that, at a given latitude, the trough minimum appears

266

The high-latitude F region

earlier if Kp is higher. The dependence (in h per unit of Kp) is just a/b, or 2.0, 4.2, 3.8, and 1.2, respectively, for the coefficients of Table 5.3. The increase in the latitude of the trough after about 0700 LT is seen in the electron-content observations (Liszka, 1967) reproduced in Figure 5.26. However, these data are fitted quite well by the formula of Kohnlein and Raitt (shown superimposed) during the hours around midnight. (Liszka’s observations were mainly for times of low Kp.) The same data give a Kp dependence of about 2° of latitude for one unit of Kp within the range 0–3 (Figure 5.27), which again agrees with the formula of Kohnlein and Raitt. Note, however, that individual values are spread 2°–3° of latitude about the trend. Rodger et al. (1986) have commented that Kp is a poor predictor of the position of the poleward edge of the trough, and this is probably true for all its features except in the statistical sense. The incoherent-scatter results of Collis and Häggström (1988) were obtained from a review of observations made during a year at sunspot minimum. The troughs were observed during the afternoon and evening hours but none was recorded during the summer period between early April and late August. Their formula gives the strongest variation of latitude with time of day, and significantly higher latitudes during the afternoon than does that of Kohnlein and Raitt. Note that the trough in Figure 5.25 occurred at 72°–74° during the afternoon. In addition to their formula for the latitude of the minimum, Best et al. (1984) also produced expressions in terms of L and for the electron temperature: L(trough minimum)5.40.5Kp 0.13t,

(5.4)

L(Te maximum)5.2 0.4Kp 0.12t,

(5.5)

Te(maximum)3250 8.06 /Dst K.

(5.6)

In Equation (5.6), Dst is the magnetic-storm index (Section 2.5.2) in nanoteslas. Table 5.3. Coefficients for Equation (5.3)

Reference

Data source

'0

a (degrees per unit of KP )

b (degrees h1)

LT for which applicable

Rycroft and Burnell (1970)

Satellite Alouette-1

62.7

1.4

0.7

1900–0500

Kohnlein and Raitt (1977)

Satellite ESRO-4

65.2

2.1

0.5

2000–0700

Best et al. (1984)

Satellite Intercosmos 18

64.0

0.5

0.13

Not stated

Collis and Häggström (1988)

EISCAT

62.2

1.6

1.35

1300–0100

60

65

70

75

16

18

20

22

0 LMT

02

04

06

Kp = 2

Winter 1965–66

Spring 1965 Summer 1965

Autumn 1965

Winter 1964–65

08

Kp = 0

10

Figure 5.26. Latitude of the trough against local time, from a year’s electron-content observations at Kiruna, Sweden. The time is local. (Reprinted from L. Liszka, J. Atmos. Terr. Phys. 29, 1243, copyright 1967, with permission from Elsevier Science.) The trends from the Kohnlein and Raitt formula have been added.

Geographic latitude, °N

0

1

2

3

4

5 Kp

60

60

0

1

2

3

4

5 Kp

WINTER 65–66

65

AUTUMN 65

WINTER 65–66

65

SUMMER 65

AUTUMN 65 70

SPRING 65

WINTER 64–65

SUMMER 65

°N 75

GEOGRAPHIC LATITUDE

SPRING 65

WINTER 64–65

(b)

70

°N 75

GEOGRAPHIC LATITUDE

Figure 5.27. Latitude variation of the trough in electron content with the magnetic index Kp: (a) 1900–2000 LMT, and (b) 0300–0400 LMT. (Reprinted from L. Liszka, J. Atmos. Terr. Phys. 29, 1243, copyright 1967, with permission from Elsevier Science.)

(a)

5.4 The main trough

The southern hemisphere The known synoptics of the main trough have been derived mainly from observations in the northern hemisphere. Mallis and Essex (1993) studied the trough in the southern hemisphere as observed in the electron content, and conclude that there are some marked differences between the hemispheres. In the south, troughs are observable in all seasons and at all times of day. They occur less frequently in winter than they do at the equinoxes or in summer, with a relatively high incidence by day. Compared with the north, the southern hemisphere has more troughs by day but fewer by night. It is assumed that these differences are due to hemispheric differences in polar circulation. 5.4.3

The poleward edge of the trough Introduction

The results in the previous section refer mainly to the minimum of electron density in the trough, but the poleward edge is also a feature of special interest. Valid questions are why the electron density increases again to the poleward side, and why that increase is so sharp. The sharpness of the poleward edge may also be put to use, since it is often the trough feature which is the most easily detected and the most precisely located.

Orientation The orientation of the trough has been studied using the poleward edge. The equatorward drift of the trough during the night suggests that, at a given time, the trough should not lie exactly along a contour of constant invariant latitude but should be oriented at a small angle to it. This property was investigated in the Antarctic using the Advanced Ionospheric Sounder (AIS) at Halley (76° S, 27° W, L4.2) by Rodger et al. (1986). The AIS can measure the direction of arrival of an ionospheric echo as well as its range. Assuming that the reflection is specular, the position of the perpendicular from the sounder to the edge of the trough can be plotted, and thus the orientation observed. The results are illustrated in Figure 5.28, which plots the positions of the echoes from troughs observed on 16 occasions. From Halley the perpendicular to the contours of constant invariant latitude is east of south, and close to the direction determined for the period 0000–0159 LT (i.e. line number 4 in panel (c)). Before this time, therefore, the poleward edge is tilted towards lower latitude at later local time (i.e. to the east), and the reverse is true after 0200 LT. The sense of these tilts is consistent with a general equatorward motion during the earlier part of the night and a poleward motion later. In panel (d), the orientations are mapped into the equatorial plane at L4.2 and compared with the “teardrop” model of Kavanagh et al. (1968) representing

269

270

The high-latitude F region

(a)

(b)

(d) (c)

Figure 5.28. On the orientation of the main trough. (a) The location of the poleward edge with respect to Halley on 20–21 June 1982. The time is local. (b) Collection of the poleward edge locations for all observations during 2200–2359 LT, with the best-fitting straight line. (c) The best-fitting straight lines for six 2-h periods. (d) Perpendiculars to the lines in (c) projected to L4.2 in the equatorial plane and compared with the Kavanagh model of magnetospheric equipotentials. (Reprinted from A. S. Rodger et al., J. Atmos. Terr. Phys. 48, 715, copyright 1986, with permission from Elsevier Science.)

the equipotentials resulting from a simple magnetospheric electric field. It appears that the trough is aligned with the equipotential and, thus, with the direction of plasma drift.

Electron precipitation and the poleward edge Following the first observations by Bates et al. (1973) it has often been noted that the minimum of the trough lies some few degrees equatorward of the edge of the region of auroral precipitation, and it is therefore natural to postulate that auroral

5.4 The main trough

ionization is what causes the electron density to increase on the poleward side of the trough. Supporting evidence comes from particle measurements (Rodger et al., 1986) and from incoherent-scatter radar (Jones et al., 1997). Electron precipitation (at 1 keV) was present on nearly every trough overpass of Dynamics Explorer-2 (DE-2) before 2230 magnetic local time. The radar evidence is of an enhanced electron temperature on the poleward side of the trough. These observations confirm earlier results published by Pike et al. (1977). Later in the night, however, after the passage of the Harang discontinuity (Section 2.5.3), the association was less clear. Two other classes of event were seen, one in which the poleward edge was accompanied by softer electrons (50 eV), and one in which the level of electron precipitation did not alter over the trough. In the DE-2 study, all three types occurred with about the same frequency after midnight. The radar study, also, was unable to establish a clear association with electron precipitation during the second half of the night. The source of the ionization forming the poleward edge is therefore less clear in the post-midnight sector. It is supposed that transport of ionization in the polar circulation is important. 5.4.4

Motions of individual troughs

Most of the studies which produced formulae for the latitude of the trough as functions of the time of day and Kp (Equation (5.3)) were based on observations from (or on signals transmitted from) orbiting satellites. As such, the data consist of a sequence of snapshots taken on different occasions; there is no opportunity to observe any one trough continuously. Therefore these formulae do not neccessarily describe the instantanous motion of the trough. The trough shown in Figure 5.28(a), for example, moved equatorward at 1.3° h1, a faster drift than would be indicated by any of the formulae except the last of Table 5.3, which, indeed, was based on the tracking of individual examples (by incoherent-scatter radar). Results from the AIS, tracking the poleward edge from Halley station (L4), also tend to show relatively high speeds. In the examples in Figure 5.29(a), showing the change of invariant latitude with time, many of the slopes exceed 1°h1. If the higher speeds were maintained for several hours, these troughs would cover a greater range of latitude than is actually observed. However, it is also significant that, in some cases, the slope flattens out, indicating that the drift is not uniform. The drift speed also varies greatly from one example to another. The examples in Figure 5.29(b), also from Halley, cover the hours 2130–0800 LT overall, though every example extends into the period 0000–0400 LT. The drift speed varies by a factor of ten (from 60 to 600 km h1), with half the speeds between 100 and 300 km h1 and the median at 200 km h1 (1.8° h1). The examples in Figure 5.29(c) are from electron-content measurements from a site in the auroral zone. (The local time is UT  1 h.) At this higher latitude the trough is seen during the afternoon, but note that the locations and the speeds again agree with the formula of Collis and Häggström (C H) rather than with

271

272

The high-latitude F region

(b)

–5

200 km N

–4

–3

–2

1

2

3

4

Time (hr) after passing over Halley 100

200

300

400 km S

Figure 5.29. The latitudinal drift of the main trough. (a) The poleward edge observed by the Halley Advanced Ionospheric Sounder for five nights of 1982. In each case Kp 2. (Reprinted from A. S. Rodger et al., J. Atmos. Terr. Phys. 48, 715, copyright 1986, with permission from Elsevier Science.) (b) A collection of Halley troughs from 1982–1983, showing the variability of the speed of equatorward drift. (Time is counted from the appearance of a weak precipitation event associated with the poleward edge.) (W. G. Howarth and J. K. Hargreaves, private communication.) (c) Trough minima from electron-content measurements in the auroral zone in Scandinavia. Values of Kp are marked and the formulae of Kohnlein and Raitt, and of Collis and Haggestrom have been superimposed. (Reprinted from J. K. Hargreaves and C. J. Burns, J. Atmos. Terr. Phys. 58, 1449, Copyright 1996, with permission from Elsevier Science.)

5.4 The main trough

Figure 5.29. (cont.)

that of Kohnlein and Raitt (KR). These troughs would not link up with those shown in Figure 5.29(a) if they continued to move equatorward at the same speed. Thus, the evidence indicates that, although formulae based on satellite data may express the latitude at which the trough is likely to be seen, individual troughs move considerably faster than those formulae would indicate. One explanation (Rodger et al., 1986) is based on the effect of substorms, which on some occasions are seen to be related to a partial filling of the trough from the poleward side. Figure 5.30 illustrates the point, showing how the polar edge steepened between two successive orbits of the satellite DE-2, a substorm having commenced in the interim. This filling was most likely due to enhanced particle precipitation due to the substorm. This is a new factor, not included in the assumptions of Equation (5.3), but it is not clear whether this is the whole explanation. 5.4.5

Mechanisms and models The main trough caused by plasma decay

Since the main trough lies between the mid- and high-latitude ionospheres, one may reasonably expect that its cause has some connection with the different circulation patterns in those two regions. Various attempts to predict the position of the trough have been made by modeling the ionosphere mathematically (Moffett and Quegan, 1983). These models represent the high-latitude convection in a

273

274

The high-latitude F region

Figure 5.30. (a) Two consecutive passes of DE-2 near Halley on 14 August 1982, showing a steepening of the poleward edge. (b) The Halley magnetometer indicated that a substorm occurred between those two orbits. (Reprinted from A. S. Rodger et al., J. Atmos. Terr. Phys. 48, 715, copyright 1986, with permission from Elsevier Science.)

steady state and, although they might not include all the physical processes that could be relevant, they do predict a main trough in about the right place. The basic cause is that there are some convection paths (e.g. path 5 in Figure 5.1(b)) which do not encounter a production region for several hours, a time long enough for the plasma density to decay to a low value. Measurements by incoherent-scatter radar (Collis and Haggstrom, 1988) support this theory, showing that the trough minimum generally lies in a zone where the plasma flow (with respect to the Earth) is strongly westward. Such a flow tends to offest the Earth’s rotation and hence prolong the time for which the plasma remains in a dark region. One possible complication is that a steady-state pattern of convection is unlikely to continue for very long, due to the constant variations in the solar wind which drives the polar convection. While ionization decay is now accepted as the essential cause of the main trough, we are some way from being able to predict details of the trough for any given day.

Other mechanisms Rodger et al. (1992) reviewed all the mechanisms that could create, or help to create, ionization troughs, and concluded that the difference in velocity between

5.4 The main trough

275

109

Figure 5.31. Temperature dependences of recombination reactions in the F region. The relative velocity of ions and neutral species is shown on the second scale. (Reprinted from A. S. Rodger et al., J. Atmos. Terr. Phys. 54, 1, copyright 1992, with permission from Elsevier Science.)

5

RATE COEFFICIENTS (cm3 s1)

2

1010

O+  O2

5

O+  N2

2

1011 5

2

1012 5

2

1013

0

2

4

6

8

10

12

3

ION TEMPERATURE (10 K) 0 1.0

2.0

3.0

4.0

RELATIVE VELOCITY (km s1)

ions and neutral particles is likely to be an important factor. The rate coefficients k1 and k2 in the expression for the recombination coefficient  k1[N2]k2[O2]

(5.7)

are temperature dependent, as shown in Figure 5.31, and a relative drift between the ions and the neutral species heats the gas. Figure 5.31 shows the relative velocity as a second abscissa scale. The heating increases the rate of loss by recombination and causes an upward flow of plasma that also depletes the F region. It is argued, therefore, that plasma depletion is expected in regions heated by high differences in speed between ions and neutral particles.

The high-latitude F region

276

Figure 5.32. The average location of the high-latitude trough determined from passes of the satellite OGO-6 (dotted line), plotted over an electric-field pattern (solid lines). (Reprinted from A. S. Rodger et al., J. Atmos. Terr. Phys. 54, 1, copyright 1992, with permission from Elsevier Science.)

5.5

Troughs and holes at high latitude

Depletions occurring poleward of the main trough – that is, within the auroral oval and polar cap – have been observed by incoherent-scatter radar and in satellite passes, but in general they have not been so intensively studied as the main trough. Rodger et al. (1992) have summarized the principal features of these troughs as follows. ■

High-latitude troughs are between 5° and 9° wide, with a poleward edge between 67° and 71° magnetic latitude and an equatorward edge between 61° and 67°. (Note that this overlaps with the position of the main trough in the afternoon sector.)

■

They last for 4–8 h, moving to higher latitude towards the end of the period.

■

Their equatorward edge moves equatorward with increasing Kp, and there is some evidence that the trough forms earlier when Kp is larger.

■

They are often associated with a reversal of the convection (as a function of latitude) in the morning sector, but in the evening are on the equatorward side of the reversal. Figure 5.32 illustrates this point.

5.5 Troughs and holes

Figure 5.33. High-latitude troughs in O and H at 70°–75° north, from OGO-6, 18 March 1970, 1830–1854 Kp 2. There is an enhancement in concentration of the molecular species NO. (Reprinted from J. M. Grebowsky et al., Planet. Space Sci. 31, 99, copyright 1983, with permission from Elsevier Science.)

■

The ion temperature (Ti ) and the electric field are usually increased within the trough but the electron temperature is not usually affected.

■

There are relationships between Ti and the field-aligned plasma velocity.

■

The atomic ions (H, O, and N) are reduced in concentration, but concentrations of molecular species (NO and O2) are increased.

Figure 5.33 shows further examples of high-latitude troughs in terms of the ion densities; note the enhancements in concentration of NO. The polar hole is recognized as a distinct feature. It is a long-lived depletion observed in years of low solar activity during winter in the Antarctic polar cap (Brinton et al., 1978), occurring shortly after midnight at magnetic latitudes near 80º. The electron density (at 300 km) is as low as (1–3) 102 cm3, compared with up to 105 cm3 elsewhere in the polar cap. The hole appears sporadically at the

277

278

The high-latitude F region

12h00

6 5

C

4 3 2 1 C

A 18h00

06h00 B

A

7 80° B

70° A

A 60° 50° 24h00

Figure 5.34. A summary of F-region depletions under steady geophysical conditions when the cross-tail electric field is small. The solar terminator is along the line 1800–0600. A, main trough; B, polar hole. C, region of significant frictional heating of ions and neutral species. The features are superposed on the polar convection pattern of Figure 5.1(b). (Reprinted from A. S. Rodger et al., J. Atmos. Terr. Phys. 54, 1, copyright 1992, with permission from Elsevier Science.)

equinoxes and hardly ever in summer. The seasonal variation can be explained by invoking the movement of the solar terminator, which ensures that the relevant region is dark in winter but illuminated in summer. The electron temperature is reduced in the polar hole and the ion speeds are low. Concentrations of molecular species are not enhanced there. For reasons that remain unknown, the polar hole has not been observed in the Arctic. It must be appreciated that it is not unusual for the high-latitude ionosphere to be irregular where it is not illuminated by the Sun. We have drawn attention in Section 5.3.2 to the phenomena of patches and blobs in the high-latitude ionosphere, where the emphasis is on the enhancements. A study of the depletions should be complementary to this, and it might not always be clear whether it is the enhancement or the depletion which is abnormal. In some cases the structure may actually comprise both – that is, the mechanism may remove ionization from one place and concentrate it elsewhere. Figure 5.34 summarizes the location of high-latitude depletions (as well as the main trough). Table 5.4 describes the various features.

High 1 km s1

C. High electric field in rest frame of neutral species

??

Low

Low

B. Polar hole

Te

High

Vi

A. Stagnation trough Low

Trough type

High and anisotropic

Presumably low, although no measurements have been made

Normal

Ti

Table 5.4. Characteristics of features in Figure 5.34

NO rich

Normal

Normal

Composition

Joule heating and outflow of ions can be significant; often occurs in regions of high electron precipitation, and thus Te can be elevated

Enhancements of He and H concentrations in the topside

Protonospheric maintenance affects equatorward edge; high Te by virtue of conduction of heat from high altitudes

Comment

The high-latitude F region

280

The reader is referred to the review paper by Rodger et al. (1992) for further details and for discussion of other high-latitude depletions of the F region.

5.6

Summary and implications

In radio propagation the F region principally affects systems operating at the higher frequencies, specifically in the HF, VHF, and UHF bands. Effects can be major even when the F region is undisturbed, especially during the winter season when electron densities tend to become very small during the long polar night. During geomagnetic storms and substorms additional effects appear. The precipitation of energetic charged particles increases, whereupon circuits operating in these bands may be seriously degraded. The main ionospheric trough, a region of depleted electron density just equatorward of the auroral oval, depresses HF operating frequencies when the reflection or control points come within its boundaries. The main trough is a semi-permanent feature at the transition between the mid-latitude and the highlatitude ionospheres, occurring mainly at night and more strongly in winter than in summer in the northern hemisphere. The incidence is somewhat different in the southern hemisphere. When energetic electrons and protons precipitate into the auroral F region they produce field-aligned irregularities of various sizes, which may deviate and scatter HF to UHF signals incident on them. Backscatter may be produced by the component of irregularity having a wavelength equal to half the radio wavelength when the signal is propagating in a direction essentially normal to the geomagnetic field-lines. The geometry is such that HF scattering is most likely to occur when signals propagate from temperate latitudes towards and into the auroral oval. HF radars operated for research purposes make use of this backscatter to study the structure and dynamics of the polar ionosphere. Over-thehorizon HF radars experience system degradation by field-aligned irregularities, and satellite-to-earth VHF and UHF signals suffer scintillation phenomena causing a rapid and sometimes severe fading of amplitude and irregular fluctuations of phase. In the polar ionosphere, by which we mean that part poleward of the auroral zone, the particle precipitation is generally not as intense as that into the oval. Nevertheless, some major F-region irregularities do occur. The dominant features, known as arcs or patches, are enhancements of F-region plasma density that probably originate not locally but in the ionosphere at lower latitude, and then drift over the polar cap under the control of the electric field between the dusk and dawn sides of the polar cap. There is now a substantial body of knowledge about these structures, though it is not yet sufficient for prediction purposes. A detailed description of the effects of the high-latitude F region on the propagation of radio signals over the whole spectrum from MF to UHF is given in Chapters 8 and 9.

5.7 References and bibliography

5.7

References and bibliography

5.1

Circulation of the high-latitude ionosphere

Boyle, C. B., Reiff, P. H., and Hairston, M. R. (1997). Empirical polar cap potentials. J. Geophys. Res. 102, 111. Cowley, S. W. and Lockwood, M. (1997) Excitation and decay of solar wind-driven flows in the magnetosphere-ionosphere system. Ann. Geophysicae. 10, 103. Dudeney, J. R., Rodger, A. S., Pinnock, M., Ruohoniemi, J. M., Baker K. B., and Greenwald, R. A. (1991) Studies of conjugate plasma convection in the vicinity of the Harang discontinuity. J. Atmos. Terr. Phys. 53, 249. Hairston, M. R. and Heelis, R. A. (1995) Response time of the polar ionospheric convection pattern to changes in the north-south direction of the IMF. Geophys. Res. Lett. 22, 631. Jayachandran, P. T. and MacDougall, J. W. (1999) Seasonal and By effect on the polar cap convection. Geophys. Res. Lett. 26, 975. Kelley, M. C. (1989) Section 6.2. In The Earth’s Ionosphere. Academic Press, New York. Lu, G. and 20 others. (1994) Interhemispheric asymmetry of the high-latitude ionospheric convection pattern. J. Geophys. Res. 99, 6491. Rich, F. J. and Hairston, M. (1994) Large-scale convection patterns observed by DMSP. J. Geophys. Res. 99, 3827. Ruohoniemi, J. M. and Greenwald, R. A. (1996) Statistical patterns of high-latitude convection obtained from Goose Bay HF radar observations. J. Geophys. Res. 101, 21 743. Todd, H., Bromage, B. J. I., Cowley, S. W. H., Lockwood, M., van Eyken, A. P., and Willis, D. M. (1986) EISCAT observations of rapid flow in the high latitude dayside ionosphere. Geophys. Res. Lett. 13, 909. Willis, D. M., Lockwood, M., Cowley, S. W. H., van Eyken, A. P., Bromage, B. J. I., Rishbeth, H., Smith, P. R., and Crothers, S. R. (1986) A survey of simultaneous observations of the high-latitude ionosphere and interplanetary magnetic field with EISCAT and AMPTE UKS. J. Atmos. Terr. Phys. 48, 987.

5.2

Behaviour of the F region at high latitude

Farmer, A. D., Crothers, S. R., and Davda, V. N. (1990) The winter anomaly at Tromsø. J. Atmos. Terr. Phys. 52, 561. Muldrew, D. B. and Vickrey, J. F. (1982) High-latitude F region enhancements observed simultaneously with ISIS 1 and the Chatanika radar. J. Geophys. Res. 87, 8263. Raitt, W. J. and Schunk, R. W. (1983) Composition and characteristics of the polar wind. In Energetic Ion Composition in the Earth’s Magnetosphere (ed. R. G. Johnson), p. 99. Terra Scientific Publishing, Tokyo. Robinson, R. M., Tsunoda, R. T., Vickrey, J. F., and Guerin, L. (1985) Sources of Fregion ionization enhancements in the night-time auroral zone. J. Geophys. Res. 90, 7533.

281

The high-latitude F region

282

Walker, I. K., Moen, J., Mitchell, C. N., Kersley, L., and Sandholt, P. E. (1998) Ionospheric effects of magnetopause reconnection observed by ionospheric tomography. Geophys. Res. Lett. 25, 293. Whitteker, J. H., Shepherd, G. G., Anger, C. D., Burrows, J. R., Wallis, D. D., Klumpar, D. M., and Walker, J. R. (1978) The winter polar ionosphere. J. Geophys. Res. 83, 1503.

5.3

Irregularities of the F region at high latitude

Aarons, J. (1982) Global morphology of ionospheric scintillations. Proc IEEE 70, 360. Anderson, D. N., Buchau, J., and Heelis, R. A. (1988) Origin of density enhancements in the winter polar-cap ionosphere. Radio Sci. 23, 513. Buchau, J., Reinish, B. W., Weber, E. J., and Moore, J. F. (1983) Structure and dynamics of the winter polar cap F region. Radio Sci. 18, 995. Buchau, J., Weber, E. J., Anderson, D. N., Carlson, H. C., Moore, J. G., Reinisch, B. W., and Livingston, R. C. (1985) Ionospheric structures in the polar cap: their origin and relation to 250 MHz scintillation. Radio Sci. 20, 325. Burns, C. J. and Hargreaves, J. K. (1996) The occurrence and properties of large-scale electron-density structures in the auroral F region. J. Atmos. Terr. Phys. 58, 217. Carlson, H. C., Wickwar, V. B., Weber, E. J., Buchau, J., Moore, J. G., and Whiting, W (1984). Plasma characteristics of polar cap F-layer arcs. Geophys. Res. Lett. 11, 895. de la Beaujardière, O. and Heelis, R. A. (1984) Velocity spike at the poleward edge of the auroral zone. J. Geophys. Res. 89, 1627. Gussenhoven, M. S., Hardy, D. A., and Heinemann, N. (1983) Systematics of the equatorward diffuse auroral boundary. J. Geophys. Res. 88, 5692. Hargreaves, J. K., Burns, C. J., and Kirkwood, S. C. (1985a) EISCAT studies of Fregion irregularities using beam scanning. Radio Sci. 20, 745. Hargreaves, J. K., Burns, C. J., and Kirkwood, S. C. (1985b) Irregular structures in the high-latitude F-region observed using the EISCAT incoherent scatter radar. Proc. AGARD Conference 382 (Fairbanks, Alaska) p. 6.2-1. Kelley, M. C., Baker, K. D., Ulwick, J. C., Rino, C. L., and Baron, M. J. (1980) Simultaneous rocket probe, scintillation and incoherent scatter observations of irregularities in the auroral zone ionosphere. Radio Sci. 15, 491. Lockwood, M. and Carlson, H. C. (1992) Production of polar cap electron density patches by transient magnetopause reconnection. Geophys. Res. Lett. 19, 1731. Muldrew, D. B. and Vickrey, J. F. (1982) High-latitude F region irregularities observed simultaneously with ISIS 1 and the Chatanika radar. J. Geophys. Res. 87, 8263. Radio Science (1994) Special section on high-latitude structures. Radio Sci. 29, 155–315. Rino, C. L. (1978) Evidence for sheetlike auroral ionospheric irregularities. Geophys. Res. Lett. 5, 1039. Rino, C. L., Livingston, R. C., Tsunoda, R. T., Robinson, R. M., Vickrey, J. F., Senior, C., Cousins, M. D., and Owen, J. (1983) Recent studies of the structure and morphology of auroral-zone F-region irregularities. Radio Sci. 18, 1167.

5.7 References and bibliography

Robinson, R. M., Tsunoda, R. T., Vickrey, J. F., and Guerin, L. (1985) Sources of Fregion ionization enhancements in the night-time auroral zone. J. Geophys. Res. 90, 7533. Secan, J. A., Bussey, R. M., Fremouw, E. J., and Basu, S. (1997) High-latitude upgrade to the Wideband ionospheric scintillation model. Radio Sci. 32, 1567. Sojka, J. J., Bowline, M. D., Schunk, R. W., Decker, D. T., Valladares, C. E., Sheehan, R., Anderson, D. N., and Heelis, R. A. (1993) Modelling polar cap F region patches using time varying convection. Geophys. Res. Lett. 20, 1783. Sojka, J. J., Bowline, M. D., and Schunk, R. W. (1994) Patches in the polar ionosphere: UT and seasonal dependence. J. Geophys. Res. 99, 14959. Tsunoda, R. T. (1988) High-latitude F region irregularities: a review and synthesis. Rev. Geophys. 26, 719. Vickrey, J. F., Rino, L. C. and Potemra, T. A. (1980) Chatanika/TRIAD observations of unstable ionization enhancements in the auroral F-region. Geophys. Res. Lett. 7, 789. Weber, E. J. and Buchau, J. (1981) Polar cap F layer auroras. Geophys. Res. Lett. 8, 125. Weber, E. J. and Buchau, J. (1985) Observations of plasma structure and transport at high latitudes. The Polar Cusp (eds. Holtet and Egeland) p. 279. Reidel, Hingham, Massachusetts. Weber, E. J., Buchau, J., Moore, J. G., Sharber, J. R., Livingston, R. C. ,Winningham, J. D., and Reinisch, B. W. (1984) F layer ionization patches in the polar cap. J. Geophys. Res. 89, 1683. Weber, E. J., Klobuchar, J. A., Buchau, J., Carlson, H. C., Livingston, R. C., de la Beaujardière, O., McCready, M., Moore, J. G., and Bishop, G. J. (1986) Polar cap Flayer patches: structure and dynamics. J. Geophys. Res. 91, 12121. Yeh, K. C. and Liu, C. H. (1982) Radio wave scintillation in the ionosphere. Proc. IEEE 70, 324.

5.4

The main trough

Bates, H. F., Belon, A. E., and Hunsucker, R. D. (1973) Aurora and the poleward edge of the main ionospheric trough. J. Geophys. Res. 78, 648. Best, A., Best, I., Lehmann, H.-R., Johanning, D., Seifert, W., and Wagner, C.-U. (1984) Results of the Langmuir probe experiment on board Intercosmos-18. Proc. Conference on Achievements of the IMS, Graz, Austria (June 1984). ESA report SP217, p. 349. Collis, P. N. and Häggström, I. (1988) Plasma convection and auroral precipitation processes associated with the main ionospheric trough at high latitudes. J. Atmos. Terr. Phys. 50, 389. Halcrow, B. W. and Nisbet, J. S. (1977) A model of F2 peak electron densities in the main trough region of the ionosphere. Radio Sci. 12, 825. Hargreaves, J. K. and Burns., C. J. (1996) Electron content measurement in the auroral zone using GPS: observations of the main trough and a survey of the degree of irregularity in summer. J. Atmos. Terr. Phys. 58, 1449.

283

The high-latitude F region

284

Jones, D. G., Walker, I. K., and Kersley, L. (1997) Structure of the poleward wall of the trough and the inclination of the geomagnetic field above the EISCAT radar. Ann. Geophysicae 15, 740. Kavanagh, L. D., Freeman, L. W., and Chen, A. J. (1968) Plasma flow in the magnetosphere. J. Geophys. Res. 73, 5511. Kohnlein, W., and Raitt, W. J. (1977) Position of the mid-latitude trough in the topside ionosphere as deduced from ESRO 4 observations. Planet. Space Sci. 25, 600. Liszka, L. (1967) The high-latitude trough in ionospheric electron content. J. Atmos. Terr. Phys. 29, 1243. Mallis, M. and Essex, E. A. (1993) Diurnal and seasonal variability of the southernhemisphere main ionospheric trough from differential-phase measurements. J. Atmos. Terr. Phys. 55, 1021. Moffett, R. J. and Quegan, S. (1983) The mid-latitude trough in the electron concentration of the ionospheric F-layer: a review of observations and modelling. J. Atmos. Terr. Phys. 45, 315. Muldrew, D. B. (1965) F-layer ionization troughs deduced from Alouette data. J. Geophys. Res. 70, 2635. Pike, C. P., Whalen, J. A., and Buchau, J. (1977) A 12-hour case study of auroral phenomena in the midnight sector: F layer and 6300 Å measurements. J. Geophys. Res. 82, 3547. Rodger, A. S., Brace, L. H., Hoegy, W. R., and Winningham. J. D. (1986) The poleward edge of the mid-latitude trough – its formation, orientation and dynamics. J. Atmos. Terr. Phys. 48, 715. Rodger, A. S., Moffett, R. J., and Quegan, S. (1992) The role of ion drift in the formation of ionisation troughs in the mid- and high-latitude ionosphere – a review. J. Atmos. Terr. Phys. 54, 1. Rycroft, M. J. and Burnell, S. J. (1970) Statistical analysis of movements of the ionospheric trough and the plasmapause. J. Geophys. Res. 75, 5600. Spiro, R. W. (1978) A study of plasma flow in the mid-latitude ionization trough. Ph.D. thesis, University of Texas at Dallas, Richardson, Texas. Thomas, J. O., Rycroft, M. J., Colin, L., and Chan, K. L. (1966) The topside ionosphere. 2. Experimental results from the Alouette 1 satellite. In Electron Density Profiles in Ionosphere and Exosphere, p. 322. Amsterdam, North-Holland.

5.5

Troughs and holes at high latitude

Brinton, H. C., Grebowsky, J. M., and Brace, L. H. (1978) The high-latitude winter F region at 300 km: thermal plasma observations from AE-C. J. Geophys. Res. 83, 4767. Rodger, A. S., Moffett, R. J., and Quegan, S. (1992) The role of ion drift in the formation of ionisation troughs in the mid- and high-latitude ionosphere – a review. J. Atmos. Terr. Phys. 54, 1.

Chapter 6 The aurora, the substorm, and the E region

6.1

Introduction

By “aurora” people usually mean the emission of light from the upper atmosphere, but in fact there are numerous related phenomena, each being a direct or indirect consequence of energetic particles entering the atmosphere from the magnetosphere. They include (a)

luminous aurora;

(b)

radar aurora, by which is meant the reflection of radio signals from ionization in the auroral region;

(c)

auroral radio absorption, the absorption of radio waves in the auroral ionization;

(d)

auroral X-rays, which are generated by the incoming particles and may be detected on high-altitude balloons;

(e)

magnetic disturbances, due to enhanced electric currents flowing in the auroral ionization, which may be detected by magnetometers;

(f)

electromagnetic emissions in the very-low- and ultra-low-frequency bands, which are generated in the magnetosphere by wave–particle interactions (Section 2.5.6), and which then propagate to the ground where they may be detected with a radio receiver or a sensitive magnetometer.

Arising as they do from a common cause, the auroral phenomena display several common properties. (1)

They all exhibit a general relationship with solar activity, though often there is no specific association with any obvious solar event. From the

285

The aurora, substorm, and E region

286

1930s the term M region was used to signify a hypothetical and unseen solar region causing aurora and magnetic storms, and this served as a unifying hypothesis for some 40 years. It is now well appreciated, of course, that the unseen agent is the solar wind. (2)

They are essentially zonal in occurrence, their occurrence and intensity coming to a maximum some 10°–25° from the magnetic poles. This property is treated in Section 6.2.

(3)

All the auroral phenomena exhibit substorm behavior. They are greatly enhanced during bursts of activity lasting perhaps 30–60 min, which are separated by quieter intervals of several hours. It is now clear that the substorm is caused by processes in the magnetosphere. This aspect is discussed in Section 6.4.

The auroral luminosity originates within the ionospheric E region. The particles which excite the emission of light also create additional ionization and thereby enhance the electron density. This in turn increases the ionospheric current at those heights, which has further consequences. The behavior of the auroral E region is therefore closely related to that of the aurora. The high-latitude E region is considered in Section 6.5.

6.2

Occurrence zones

6.2.1

The auroral zone and the auroral oval

In general the auroral phenomena are highly structured in both space and time, with essentially zonal patterns of occurrence. The classical picture of the occurrence of aurorae (Figure 6.1) shows a zone centered about 23° from the geomagnetic pole (i.e. about 67° geomagnetic latitude) and about 10° wide in latitude. The isochasms show the occurrence of discrete aurorae, which is 100% at the maximum and falls off both to the equatorward and to the poleward sides. (“100%” here means that some aurora was seen every clear night, not that it was visible all the time.) This pattern, which is a geographic distribution, was first defined by Vestine (1944) and is based on reports of visual sightings of the aurora over several decades. However, in 1963 Y. I. Feldstein, using all-sky camera data from the International Geophysical Year of 1957–1958, pointed out that at a fixed time the locus of the aurora is not circular but oval (Figure 6.2). The maximum is near 67° latitude at midnight, but this increases to about 77° (the latitude of the cusp – Section 2.2.5) at noon. The auroral oval, as it is generally known, is widest at midnight and narrowest at noon. It is essentially fixed with respect to the Sun, and the classical auroral zone is the locus of the midnight sector of the oval as the Earth rotates underneath it. The auroral oval is one of the important boundaries of geospace. In relation to the structure of the magnetosphere it is generally considered

6.2 Occurrence zones

Figure 6.1. The northern auroral zone, showing the percentage of good observing nights when aurorae may be seen. (After E. H. Vestine, Terr. Magn. Atmos. Electricity, 49, 77, 1944, copyright by the American Geophysical Union.)

to mark the division between open and closed field-lines. The regions poleward of the ovals (one in each hemisphere) are generally taken to be the polar caps in which the magnetic field-lines connect to the IMF and circulate under the influence of the solar wind (Section 2.4.1). Although it was originally just a statistical concept, later work, both groundbased (Feldstein and Starkov, 1967) and using photography from space (Akasofu, 1974; Frank and Craven, 1988), has shown that the oval exists virtually continuously as a permanent ring of light around the magnetic pole, and also as a ring of particle precipitation (Fuller-Rowell and Evans, 1987; Hardy et al., 1985). In the pictures from space the general form of the oval is clearly observed (Figure 6.3) as a continuous band of light around the pole that is nearly always present, though its intensity varies greatly from time to time. The latest auroral pictures from space are adding much detail and have shown that the oval form is not by any means the whole story. The detailed spatial distribution varies considerably from time to time. Sometimes an arc is seen to extend across the polar cap, connecting the day and night sides of the oval – a configuration called the -aurora. Sometimes the morning side of the oval is quiet while the

287

The aurora, substorm, and E region

288

SUN

FLUX = 104 cm2 s1 TRAPPED ELECTRONS (E ( 40 KeV) (FRANK, VAN ALLEN, & CRAVEN)

AURORAL OVAL (75%–90%) (FELDSTEIN)

Figure 6.2. The auroral oval in relation to the 40-keV trapping boundary. (S.-I. Akasofu, Polar and Magnetospheric Substorms, Reidel, 1968, with kind permission from Kluwer Academic Publishers.)

evening is active, and sometimes the morning side is the more active. The behavior of more localized brightenings within the oval can also be observed from space. Some examples are shown in Figure 6.4. The variety of configurations and dynamics emphasizes the complexity of the auroral distribution and suggests that present classifications are incomplete. 6.2.2

Models of the oval

Without doubt the auroral oval is a special region of the ionosphere. That being so, it is often convenient to refer observed phenomena to the location (or probable location) of the oval at the time of the observation, and thus it is helpful to have models giving the typical position of the oval under stated conditions. Figure 6.5(a) indicates, for typical conditions, the geographic location of the oval every 2 h of the UT day. It is usual to quantify the level of disturbance by using one of the magnetic activity indices (see Section 2.5.4), Q being a popular one for this purpose since it is derived at 15-min intervals. Figure 6.5(b) gives the locations of the oval for several levels of Q taken from a set of diagrams developed by Whalen (1970). (Kp being a more common index, the following relations may be used to obtain the appropriate value of Q when one is using Whalen’s results: Q8 if Kp &6; QKp 2 if 1 Kp 6; and Q3Kp if Kp %1.) Since the oval is closer to the magnetic pole at noon than it is at midnight, it is quite possible for an observer on the Earth to be poleward of the oval at midnight and equatorward of it at noon. Distributions based not on luminosity but on measurements of particles of

6.2 Occurrence zones

Figure 6.3. The auroral oval from space, observed in the ultra-violet (118–165 nm) by the Dynamics Explorer I spacecraft on 16 February 1982. The aurora is plainly visible around the northern pole. Airglow bands north and south of the equator, dayglow above the morning (right) limb of the Earth, and resonant Lyman- scattering in the protonosphere are also to be seen. (L. A. Frank and J. D. Craven, University of Iowa, private communication.)

energy 30 or 50 eV to 20 keV on DMSP satellites (Hardy et al., 1985; 1989) show zones of electron and ion influx in terms of magnetic latitude and local time, at levels of geomagnetic activity quantified by Kp. Figure 6.6 shows the distribution of electrons for Kp 3. Figure 6.6(a) is very like the auroral oval, being offset from the magnetic pole towards midnight. The particles forming the dayside maximum are relatively soft (i.e. of low energy). Using data from the same source, Meng and Makita (1986) defined the boundaries of the precipitation zones for “low-energy” (500 eV) and “high-energy” (500 eV) electrons, for magnetically quiet (AE%150 nT) and disturbed (AE 400 nT) conditions, and for the evening and morning sectors – see Table 6.1. The criterion for the boundary was a flux of 107 electrons cm2 s1 steradian1. The transition latitude, where the fluxes of low- and high-energy particles were equal, was also noted.

289

290

The aurora, substorm, and E region

Figure 6.4. Images of the northern auroral region observed by the Viking satellite. The camera had a 20° by 25° field of view and responded to UV of 134–180 nm, mainly emissions from nitrogen. Each exposure lasted 1.2 s. The top left-hand image shows the whole auroral oval including the day side. The one below it shows a substorm in the midnight sector, with activity also around noon and faint arcs in the morning. The top right-hand image is from the last stage of a substorm, when regularly spaced bright spots, lasting 1–5 min, may appear along the poleward edge of the oval near midnight. The fourth image shows a sun-aligned arc extending across the polar region from midnight (at the bottom) to noon. (Pictures and commentary from G. Enno, private communication. The Viking project was managed by the Swedish Space Corporation for the Swedish Board for Space Activites. The UV imager was a project of the National Research Council of Canada, and was operated by the Institute for Space Research, University of Calgary, with support from the Natural Sciences and Engineering Research Council of Canada.)

It is generally agreed that the oval expands equatorward, to lower latitudes, as magnetic activity increases. The region of low-energy precipitation becomes narrower and the high-energy region broadens. According to Chubb and Hicks (1970), the equatorward boundary of the luminous oval moves about 1.7° equatorward per unit of Kp on the day side of the Earth, and 1.3° on the night side; it moves by 1°–3° of latitude in individual substorms. (See also Section 6.4.2.) The oval also varies with the IMF, increasing in size by about 0.5° for each one  increase in the southward component of the IMF. According to Meng (1984), the polar cap can be as small as 12° side to side under quiet conditions and as large as 50° when it is disturbed, which is not inconsistent with the results in Table 6.1. Gussenhoven et al. (1983) express the variation of the equatorward boundary of the oval in terms of the Kp index, as

6.3 The auroral phenomena

LL0 aKp,

291

(6.1)

where L0 and a depend on the local magnetic time (MLT) as in Table 6.2. Figure 6.7 illustrates the position of the oval at three levels of disturbance, and its magnetic latitude and thickness against Kp. The foregoing results may be expected to apply also to those propagation phenomena which are typical of the auroral oval.

6.3

The auroral phenomena

6.3.1

The luminous aurora

The luminous aurora is a well-known phenomenon of the high-latitude regions, and in fact the most readily observed consequence of the dynamic magnetosphere. Although it is only in the present century that there has been any kind of understanding of the aurora, it must surely rank amongst the oldest of the known geophysical phenomena. There are accounts of lights in the night sky going back to Greek and Roman times, when they were frequently given a mystical or prophetic interpretation. The term aurora borealis dates from 1621, and the southern lights, observed by Captain James Cook in 1773, were subsequently called the aurora australis. Detailed reports of auroral displays date from 1716 and the first written work devoted entirely to the polar aurora was published in France in 1733. The first proof that the auroral light is a consequence of the excitation of atmospheric gas by energetic particles did not come until the early 1950s, and it was not until 1958, when rockets were fired into an aurora, that energetic electrons were identified as the primary source. Where those electrons come from and how they are energized are questions that have not yet been answered in full, but their magnetospheric origin is beyond doubt and much has been learned about them in recent years. 6.3.2

The distribution and intensity of the luminous aurora

Auroral investigations before about 1950 tended to fall into one of two areas. Morphological studies were intended to map the occurrence of the aurora in space and time and to determine the details of the fine structure of individual auroral forms. Auroral spectroscopy was virtually a separate discipline, concerned with the emitted light, in particular with its spectrum and its origin in photochemical processes – a topic having strong affinities with airglow. The luminous aurora is highly structured and dynamic. Some features are as thin as 100 m, and the time changes can be as rapid as 10 s1. The basic recording intrument is the all-sky camera which was first used during the 1950s and is particularly valuable for surveying the occurrence of aurorae. It uses a convex

Figure 6.5a (1) Figure 6.5a (2)

Figure 6.5a (3) Figure 6.5. Representations of the auroral oval. (a) The geographic position of the auroral oval under typical conditions for each 2 h (UT) of the day. (S.-I. Akasofu, Polar and Magnetospheric Substorms, Reidel, 1968, with kind permission from Kluwer Academic Publishers.) (b) The position of the oval in geomagnetic coordinates at disturbance levels Q1, 3, 5 and 7. (J. A. Whalen. Report AFCRL-70-0422, 1970.)

Figure 6.5b (1)

AURORAL OVAL (Feldstein 1967)

Q=1

Corrected Geomagnetic Local Time Latitude

AURORAL OVAL (Feldstein 1967)

Q=3

Corrected Geomagnetic Local Time Latitude

Figure 6.5. (cont.)

Figure 6.5b (2)

AURORAL OVAL (Feldstein 1967)

Q=5

Corrected Geomagnetic Local Time Latitude

AURORAL OVAL (Feldstein 1967)

Q=7

Corrected Geomagnetic Local Time Latitude

The aurora, substorm, and E region

296

(a)

(c)

(b)

Figure 6.6. Electron-precipitation zones for Kp 3. (a) The total flux of precipitating electrons in the energy range 30 eV to 30 keV in units of cm2 s1 sr1. Numbers in brackets are powers of ten. (b) The total energy flux, in units of keV cm2 s1 sr1, due to the same flux of electrons. (c) The average energy (keV) of electrons in the band 30 eV to 30 keV. The data are from the DMSP satellites F6 and F7, and the maps are in corrected geomagnetic latitude (marked every 10° from 50° to 80°) and magnetic LT. (Private communication from M. S. Gussenhoven and D. H. Brautigan, Space Hazards Branch, Air Force Research Laboratory. Further details are given by D. A. Hardy et al., J. Geophys. Res. 90, 4229 (1985) and J. Geophys. Res. 94, 370 (1989).)

mirror to obtain a picture of the night sky from horizon to horizon, and it would typically be operated automatically at regular intervals on every clear night during the winter viewing season. There is a classification of auroral structure based on its general appearence, as in Table 6.3. When structure is present the height of the luminosity may be determined by triangulation. Between 1911 and 1943, C. Störmer made 12 000 height determinations with spaced cameras and found that the lower borders of auroral forms were usually at heights of 100–110 km (Figure 6.8(a)). In some of the forms the luminosity is concentrated into a band only 10–20 km deep and the lower edge, in particular, can be quite sharp. The brightness of a discrete arc typically falls off by a factor of ten within a few kilometers below the maximum, and by a further factor of ten only 1 or 2 km below that. The vertical distribution of auroral luminosity is illustrated in Figure 6.8(b) for several types of aurora.

6.3 The auroral phenomena

297

Table 6.1. The magnetic latitude of the poleward boundary of “low”-energy, and the equatorward boundary of “high”-energy electron precipitation (after Meng and Makita, 1986) Quiet conditions

Disturbed conditions

Evening

Morning

Evening

Morning

Poleward boundary (low energy)

80°–82°

80°–82°

73°–75°

76°–77°

High-to-lowenergy transition

73°–75°

73°–75°

70°–72°

70°–72°

Equatorward boundary (high energy)

69°–71°

67°–69°

64°–66°

64°–66°

Table 6.2. Values of L0 and a for Equation (6.1) MLT (h)

L0

a

00–01 01–02

66.1 65.1

1.99 1.55

04–05 05–06 06–07 07–08 08–09 09–10 10–11 11–12 12–13

67.7 67.8 68.2 68.9 69.3 69.5 69.5 70.1 69.4

1.48 1.87 1.90 1.91 1.87 1.69 1.41 1.25 0.84

15–16 16–17 17–18 18–19 19–20 20–21 21–22 22–23 23–24

70.9 71.6 71.1 71.2 70.4 69.4 68.6 67.9 67.8

0.81 1.28 1.31 1.74 1.83 1.89 1.86 1.78 2.07

298

The aurora, substorm, and E region

Figure 6.7. (a) Positions of the auroral oval under three levels of activity. (b) The magnetic latitude and thickness of the oval as functions of Kp. (J. M. Goodman, HF Communications. Van Nostrand Reinhold, 1992.)

Table 6.3. Classification of auroral forms Forms without ray structure Homogeneous ray structure: a luminous arch stretching across the sky in a magnetically east–west direction; the lower edge is sharper than the upper, and there is no perceptible ray structure Homogeneous band: somewhat like an arc but less uniform, and generally exhibiting motions along its length: the band may be twisted into horseshoe bends Pulsating arc: part or all of the arc pulsates Diffuse surface: an amorphous glow without distinct boundary, or isolated patches resembling clouds Pulsating surface: a diffuse surface that pulsates Feeble glow: auroral light seen near the horizon, so that the actual form is not observed Forms with ray structure Rayed arc: a homogeneous arc broken up into vertical striations Rayed band: a band made up of numerous vertical striations Drapery: a band made up of long rays, giving the appearance of a curtain; the curtain may be folded Rays: ray-like structures, appearing singly or in bundles separated from other forms Corona: a rayed aurora seen near the magnetic zenith, giving the appearance of a fan or a dome with the rays converging on one point Flaming aurora: waves of light moving rapidly upward over an auroral form

6.3 The auroral phenomena

(a)

500

500

299

1000

(b)

500

80

Height (km)

400

400

300

300

200

200

100

100

Height above lower border (km)

70

60

50 Rays

40

Draperies 30 Arcs with ray-structure

20

10 Arcs

0

500

1000

Light intensity

Number of observations

Figure 6.8. Observations of auroral luminosity. (a) The distribution of 12 330 height measurements made by Störmer and colleagues. The vast majority lie between 90 and 150 km. (C. Störmer, The Polar Aurora. Oxford University Press, 1955. By kind permission of Oxford University Press.) (b) Profiles of auroral luminosity along various forms. (After L. Harang, The Aurorae. Wiley, 1951.)

The intensity of an aurora as seen from the ground is measured in units of the Rayleigh, named in honour of R. J. Strutt (the fourth Baron Rayleigh), who was a notable amateur scientist of his time and the leading pioneer of airglow studies. The unit (R) is defined as 1 R106 photons cm2 s1.

(6.2)

It is a measure of the height-integrated rate of emission, as would be observed by an instrument on the ground looking vertically upward. A more general classification of brightness uses a scale of I–IV, as in Table 6.4. This also shows the

The aurora, substorm, and E region

300

Figure 6.9. A keogram from auroral TV in Scandinavia. North (poleward) is at the top. This example shows the main features of auroral activity during a 2-h period on 18 February 1993, giving some idea of the complexity of the aurora on an active day (Kp 4). Note the dominance of equatorward movements. There are also several poleward

standard a visual observer might use for comparison, the equivalent in kilorayleighs, and the approximate rate of deposition of energy into the atmosphere. A photometer is needed for exact intensity measurements, and this can either be pointed in a fixed direction, for example to the zenith, or scanned across the sky to record the spatial distribution of intensity as well. A diagram of the latitudinal variations with time is sometimes called a keogram. Neither scanning photometers nor cameras are sufficiently sensitive to record the most rapid fluctuations in the auroral emissions, but TV techniques, both monochrome and color, are more sensitive and have been applied very successfully to dynamic auroral photography in recent years. In addition to their scientific value, some of these auroral “videos” are possessed of no little esthetic interest (particularly if they are set to music). Figure 6.9 shows an example of a keogram composed from TV data. One significant distinction that should be made is that between discrete and diffuse (or mantle) aurora. All the earlier studies concentrated on the discrete Table 6.4. Intensity classification of the aurora Intensity

Equivalent to

Kilorayleighs

Energy deposition (erg cm2 s1)

I II III IV

Milky Way Thin moonlit cirrus Moonlit cumulus Full moonlight

1 10 100 1000

3 30 300 3000

6.3 The auroral phenomena

301

U.T.

expansions, either substorm onsets or pseudo-break-ups (Section 6.4.2). The distance scale assumes that the emission comes from a height of 110 km. A distance of 600 km is equivalent to approximately 5.5° of magnetic latitude. (Data from P. N. Smith, Space Physics Group, University of Sussex, via the Auroral TV Database.)

auroral forms (Table 6.3) which are the more readily observed against the background light of the night sky because of their fine and dynamic structure. However, as was demonstrated in the early 1960s, the aurora may also take the form of a diffuse glow. This contributes at least as much total light as the discrete forms, though it is more difficult to observe from the ground because of its low intensity per unit area. The night-time discrete and diffuse aurorae are thought to map along the geomagnetic field into different regions of the magnetotail (Section 2.2.6 and Figure 2.6). The diffuse aurora is generally associated with the central part of the plasma sheet, and the discrete forms, which tend to appear poleward of the diffuse aurora, are thought to map onto the edge of the plasma sheet or to an X-type neutral line (Section 2.4.2 and Figure 2.20). Downward-looking satellites, by virtue of their ability to observe a large part or even the whole of the auroral oval at the same time, and which avoid the problem of poor seeing conditions which so often affects the ground-based techniques, have provided much new information about the distribution of the luminous emissions. The diffuse aurora tends to dominate in these pictures, but discrete forms are also seen within the diffuse glow or poleward of it; they are not seen on the equatorward side, however. When the IMF is northward, luminous arcs extending for thousands of kilometers and aligned towards the Sun are observed in the polar caps. They are not bright (emitting only tens of rayleighs, against thousands for a normal aurora) but they can be detected with modern equipment and at that low intensity are observed about half the time. It appears, therefore, that they are almost always

The aurora, substorm, and E region

302

present when the IMF is northward. It is believed that these arcs are on closed field-lines and that they may be magnetically conjugate (i.e. they occur simultaneously at opposite ends of field-lines in northern and southern hemispheres). The Sun-aligned arcs are associated with velocity shears in the polar-cap convection (Section 5.1.2 and Figure 5.5). (The -aurora, mentioned above, is also associated with a velocity shear, but it is much brighter and also much rarer than the common Sun-aligned arcs. It is not at present clear whether it is a different phenomenon.) 6.3.3

Auroral spectroscopy

Aurorae and airglow have similar causes, both being the emission of quanta of radiation from common atmospheric gases, particularly O and N2. In the first case the excitation is by energetic particles entering the upper atmosphere from the magnetosphere, and in the second by electromagnetic radiation from the Sun. The emission lines represent transitions between energy states of the emitting species, but these may be complex and the task of interpreting the auroral spectrum was far from trivial. In spectroscopists’ terms the lines are in general “forbidden,” which means in practice that they are generated by transitions that are relatively improbable. Most aurorae are too faint to be seen in color by the naked eye, but a bright aurora appears green or red, the colours being due to atomic-oxygen lines at 557.7 nm (the green line) and 630.0 nm (the red line), respectively. The 391.4-nm line from ionized molecular nitrogen (N2 ) is also present in the violet. Some aurorae have red lower borders, and, when this occurs, the red light is due to emissions from molecular oxygen. Such aurorae result from unusually energetic particles that penetrate deeper into the atmosphere. The important group of emissions from atomic oxygen and the transitions which cause them are illustrated in Figure 6.10. Some of the UV emissions, particularly the O emissions near 130 nm, have proved particularly valuable for mapping the aurora from space vehicles because, at those wavelengths, the aurora may be seen in sunlight. In addition to their obvious applications to the detection and mapping of the aurora, some of the emission lines can be exploited to provide information helpful to other branches of upper-atmospheric science. The intensities of the emissions from N2 at 427.8 and 391.4 nm are proportional to the rate of ionization by the incoming electrons. The neutral-air wind in the thermosphere may be determined by measuring the Doppler shift of the 630-nm line of oxygen. 6.3.4

Ionospheric effects

The auroral phenomena are all associated with the precipitation of energetic electrons into the atmosphere. Although the best known of them, the luminosity is actually a byproduct of ionization by energetic particles and of the subsequent recombination processes. Other phenomena, more directly related to the enhanced electron density of the auroral region, are of greater direct concern in

6.3 The auroral phenomena

303

(a) 15

90 68

394 7 77 74

49

811 989 878 1027 13

56

16

6

1304

922 41

1152

8

999

1218

10

79

436 8 84 46

12

4 55

77

Singlets

(b)

Triplets

D4—0

D4—0 5

0

P3,2,1

5

5

0 5

5

0

P3,2,1

S2

D3,2,1

D3,2,1

3

0

P2,1,0

3

3

0

P2,1,0

S1

3

3

1 0 F3

F3

1

0

D2

1

0

D2

1

P1

P1

297 2 6300 1

0

S0

1

1

0

S0

2

1

Energy above Ol ground state (eV)

14

Quintets

Excitation energy

Quantum state

4.17 eV

1

S0

0.74 s

2972 UV

1.96 eV

1D

110 s 6364 Red

0.00 eV

5577 Green

2

6300 Red 0 1 3p 2

Figure 6.10. Energy levels and transitions in atomic oxygen. (a) Transitions that have been observed in airglow or aurorae. (M. H. Rees, Physics and Chemistry of the Upper Atmosphere. Cambridge University Press, 1989.) (b) Details of the most important lines. (After S. J. Bauer, Physics of Planetary Ionospheres. Springer-Verlag, 1973, copyright notice of Springer-Verlag.) In each case the unit of wavelength is the ångstrom unit.

radio propagation. They are briefly reviewed here, and some will be treated in detail in later chapters. Figure 6.11 illustrates the connections in a schematic, and admittedly simplistic, manner.

The E region In the E region, for example, it is not unusual for the electron density to be increased to several times 1012 m3 by electron precipitation. Electron densities of

304

The aurora, substorm, and E region

Figure 6.11. Some links between auroral phenomena. Techniques are shown in brackets. (J. K. Hargreaves, Proc. Inst. Electr. Electron. Engineers. 57, 1348, © 1969 IEEE.)

this magnitude may reflect vertically incident waves of radio frequency up to 20 MHz (Section 3.4.2, Equation (3.64)), and those of higher frequency if they are obliquely incident (Section 3.4.3, Equation (3.73)). The ionization may therefore be detected by radar as a total reflection if the frequency is not too high. If the observing geometry is suitable, echoes may also be received at higher frequencies, and these echoes come from electron-density irregularities that are produced by instabilities arising in the auroral electrojet (Section 2.5.3). Since the irregularities tend to be field-aligned, the echo intensity is aspect-sensitive and the best observing geometry is when the radar lies in the plane normal to the magnetic field-lines. The radar aurora is described in detail in Section 6.5.5.

The D region The more energetic electrons penetrate into the D region (See Figure 2.26), and the ionization they create there acts to absorb radio waves by an amount depending on their frequency. The effect is usually monitored with a Riometer (Section 4.2.4), which typically operates in the range 30–50 MHz, at which frequencies the absorption rarely exceeds 10 dB; but the effect will generally be considerably greater in the HF band. (The absorption varies approximately with the inverse square of the frequency – Section 3.4.4, Equation (3.95)). The properties of auroral radio absorption are detailed in Section 7.2. Figures 7.23 and 7.24 illustrate E- and D-region electron-density profiles observed by incoherent-scatter radar during electron-precipitation events.

X-rays Auroral X-rays are generated by the Bremsstrahlung process outlined in Section 2.6.2. They have no direct influence on radio propagation, but, because of their greater penetrating power, they produce ionization at a lower altitude than do their parent electrons. The incidence and morphology of auroral X-rays are in many ways similar to those of auroral radio absorption, both being due to the harder end of the electron spectrum.

6.3 The auroral phenomena

Figure 6.12. The two zones of auroral particle precipitation in the northern hemisphere. The density of symbols indicates the average flux, and the coordinates are geomagnetic latitude and time. (Reprinted from T. R. Hartz and N. M. Brice, Planet. Space Sci. 15, 301, copyright 1967, with permission from Elsevier Science.)

Magnetic effects Magnetic bays (Section 2.5.3) are essentially a phenomenon of the auroral zone, though, as a magnetic perturbation, they are also detected by magnetometers a considerable distance away. The bays are primarily due to the ionospheric current which flows in enhancements of the E-region electron density. VLF and ULF emissions also increase when the auroral zone is active. They have various causes, some involving wave–particle interactions, but they are basically magnetospheric in origin and are not a factor in radio propagation. 6.3.5

The outer precipitation zone

Hartz and Brice (1967) generalized the definition of auroral phenomena by recognizing that they actually fall into two groups having different patterns of occurrence (Figure 6.12). The inner one, corresponding to the luminous oval, is characterized by

305

The aurora, substorm, and E region

306

■

luminosity,

■

sporadic-E on ionograms,

■

spread-F on ionograms,

■

soft X-rays,

■

impulsive micropulsations,

■

negative bays on magnetometers,

■

soft but intense electron fluxes detected by satellites,

■

high frequency (4 kHz) VLF hiss, and

■

rapid fading of VHF scatter signals.

In addition there is a second zone at a lower latitude, which is almost circular and covers approximately 60°–70°, with its center at about 65° geomagnetic latitude. This zone displays ■

diffuse aurorae,

■

radio absorption,

■

sporadic-E at 80–90 km altitude,

■

continuous micropulsations,

■

hard X-rays of long duration,

■

harder (40 keV) electrons detected by satellites,

■

VLF emissions at 2 kHz, and

■

slow fading of VHF scatter signals.

This second zone of precipitation is generally thought to be connected with the outer Van Allen zone of trapped particles (Section 2.3.4); considering that it is also the outer of the two zones when plotted on a polar map, we shall call it the outer precipitation zone. The ionospheric phenomena in the outer zone are related to electron precipitation more energetic than that typical of the oval. The phenomena tend to be of longer duration in the outer zone. The rate of occurrence is greatest by day, whereas it is greatest by night in the oval. In both zones the phenomena are sporadic and dynamic, and both exhibit substorm behavior (Section 6.4.2). They occupy much the same latitude at midnight but become increasingly separated towards noon. In Figure 6.6, which shows properties of electron precipitation (30 eV to 20 keV), the total flux of particles (a) is distributed like the luminous oval. However, the dayside particles being relatively soft at the higher latitudes, the total energy flux (b) has a night maximum near and just before midnight, as in the inner oval of Figure 6.12. The average energy of the particles (c) is a maximum between 60° and 70° in the morning, in the vicinity of the peak of Hartz and Brice’s outer zone. It is also interesting to compare the Hartz-and-Brice picture, made 30 years ago and based mainly on ground-based observations, with the new satellite-based results in Figure 6.13. The inner zone is the luminous intensity recorded by a

6.3 The auroral phenomena

307

Figure 6.13. Comparison between the inner and outer precipitation zones. The auroral images were taken by the VIS camera on board POLAR, 7 May 1996 and 13 May 1996 (data courtesy of L. A. Frank, University of Iowa, USA). The radiationbelt data were obtained by the HILT electron detector on board SAMPEX (data courtesy of B. Klecker, MaxPlanck-Institut für extraterrestrische Physik, Garching bei München). Figure provided by T. I. Pulkkinen (Finnish Meteorological Institute).

The aurora, substorm, and E region

308

camera on the POLAR satellite (averaged over 1 h), and the outer one is a composite of fluxes of 1 MeV electrons taken during 15 orbits of the SAMPEX satellite over the course of one day. Quiet (Ap 4) and more active (Ap 14) conditions are represented. On 7 May the oval is contracted and the outer zone inactive; on 13 May the oval is expanded and the outer zone intense. The plots do not confirm the morning maximum of the Hartz-and-Brice picture, but this may be because SAMPEX samples at only two local times. Recall that the distribution of occurrence of scintillation (Section 5.3.3) also exhibits a latitudinal separation from the edge of the precipitation zone that is considerably greater by day than it is by night. This would seem to identify the region of F-region irregularity with the auroral oval rather than with the outer zone.

6.4

The substorm

6.4.1

History

As early as 1837, auroral observers had noted that during a single night there were times when the aurora was at its most intense, the activity being weaker during the periods in between (Stern, 1996). The same was true of the related magnetic signature, and it was Birkeland (1908) who first studied this tendency in magnetic records and identified what he called the “elementary polar magnetic storm.” However, Birkeland’s work in this area, which also involved field-aligned currents, fell into disfavor, and the topic made no further progress until the early 1960s. It was then that Akasofu and Chapman (1961), in a study of the polar disturbance (DP) field, coined the term “DP substorm” for the short periods of enhanced magnetic disturbance that Birkeland had noted more than 50 years before. Shortly thereafter, Akasofu noted that these events were often accompanied by bursts of auroral activity, which (at Chapman’s insistence, it is said) were named “auroral substorms” (Akasofu, 1970). Akasofu subsequently introduced the term “magnetospheric substorm” to indicate the generality of the phenomenon and to make it clear that, although the consequences of the substorm are most apparent in the polar regions, its cause lies in the magnetosphere (Rostoker et al., 1980). 6.4.2

The substorm in the aurora

The essence of a substorm as it affects the auroral regions is best described by Akasofu’s analysis of the “auroral substorm” which he developed in the 1960s (Akasofu, 1968; 1977). Akasofu used all-sky camera pictures of aurorae recorded during the International Geophysical Year (1956–1958), and applied them to derive a convincing description on the global scale of the typical behavior of the luminous aurora during a substorm. The aurora tends to be active for about an hour at a time, with quiet periods of

6.4 The substorm

309

(a)

(b)

(c)

(d)

(e)

(f)

Figure 6.14. The substorm in the luminous aurora: (a) T0; (b) T 0–5 min; (c) T5–10 min; (d) T10–30 min; (e) T30 min–1 h; and (f) T1–2 h. (S.-I. Akasofu, Polar and Magnetospheric Substorms. Reidel, 1968, with kind permission from Kluwer Academic Publishers.)

2–3 h between, and there is also a dynamic aspect. Akasofu’s representation is illustrated in Figure 6.14. The sequence begins as a quiet arc brightens and moves poleward, forming a bulge. If several arcs are present, it is often the equatorward one which brightens. Active auroral forms then appear in the bulge, equatorward of the original arc. This is called break-up or the expansion phase, and the instant when it begins is usually called the onset. Near midnight the oval is now broader than before, while the polar cap contained within the oval is smaller. At the same time active auroral patches move eastward towards the morning sector and other forms travel westward towards the evening. The westward movement is called the westward-traveling surge. After 30 min to 1 h the night sector recovers and the substorm as a whole dies away (the recovery phase). The sequence is likely to be repeated 2–3 h later. By defining a repeating pattern in auroral behavior it was this analysis which really established the substorm as the central concept in studies of the auroral phenomena. The period before the break-up is now recognized as a growth phase, which is not so spectacular in the aurora but which was first studied in the magnetotail (Section 2.2.6), which becomes gradually more tail-like for tens of minutes to an

310

The aurora, substorm, and E region

Figure 6.15. An aurora observed from space by a DMSP satellite at the maximum of a substorm, 9 January 1973 at 2024: (a) a photograph, and (b) interpretation over a map including the magnetic latitude. (S.-I. Akasofu, Space Sci. Rev. 16, 617, 1974, with kind permission from Kluwer Academic Publishers.)

hour before the onset. During the growth phase the arcs of the auroral oval move equatorward and the area of the polar cap contained within the oval grows larger. The equatorward motion of the oval during the growth phase is typically several hundred m s1 (Elphinstone et al., 1991). Arcs form again during the recovery phase, and these also drift equatorward. Some brightenings of the aurora do not develop into full substorms. They remain limited to a few hundred kilometers (Akasofu, 1964), and are relatively short-lived. Such events are called pseudo-breakups. The distinction between substorms and pseudo-breakups has been discussed by Pulkkinen (1996). Satellite observations using downward-pointing photometers have confirmed this general picture. Figure 6.15 shows an example of a substorm breakup observed from a DMSP satellite. The auroral satellites (such as DMSP, Viking, Akebono, and POLAR) have also added much detail to the original concepts, both of the auroral oval and of the substorm – and, as so often happens, the topic

6.4 The substorm

turns out to be more complicated than had originally been thought! For example, it now appears that the westward-traveling surge is made up of a number of localized brightenings or surges that do not move far as individuals. Each surge lasts for just a few minutes, and then a new surge appears to its west. Thus the aurora as a whole does indeed move westward toward the evening sector, but it goes in a series of jumps. Murphree et al. (1991) have summarized the following details of the optical substorm observed by the VIKING satellite. (1)

The latitudinal width of the auroral activity does not vary systematically during the growth phase.

(2)

During this phase the motion of the equatorward boundary of the diffuse aurora is generally equatorward with a speed less than a few hundred m s1.

(3)

The expansion phase is preceded by auroral intensifications lasting up to several hours of local time, which fade shortly before onset.

(4)

The onset region is very localized, being less than 500 km across in the ionosphere.

(5)

Auroral observations under moderately active conditions indicate that auroral emissions can extend several degrees of latitude poleward of the location of the onset. This suggests that the onset region can be well away from the boundary between open and closed field-lines.

(6)

When the position of the onset is mapped along the geomagnetic field to the equatorial plane, it is consistent with the location of the inner boundary of energetic particle flux, the so-called “injection boundary,” observed in other studies.

In Section 6.3.5 it was pointed out that the auroral phenomena occupy not one “zone” but two. Both zones exhibit substorm behavior, and Figure 6.16 shows an overall picture illustrating how the substorm develops in each zone, as represented by the fluxes of softer (5 keV) and harder (50 keV) electrons, respectively. The oval and the outer zone are obviously related to each other in some way. The most likely mechanism is that, when the auroral oval is active in a substorm, the outer zone becomes populated with energetic particles that drift in longitude and are subsequently precipitated. However, not all the physical connections between the two zones have been explained fully. 6.4.3

Ionospheric aspects of the substorm

The enhanced precipitation of energetic electrons during a substorm increases the rate of ionization of the ionosphere, and of the lower ionosphere in particular, by an amount depending on the particle flux, and over a range of altitudes determined by the particle energies (Figure 2.26). Consequently, the substorm behav-

311

The aurora, substorm, and E region

312

60°

60°

70°

70°

80°

80°

T = 0 – 5 min

Electrons 5 keV 50 keV

T = 5 – 10 min

60°

60°

70°

70°

80°

80°

T = 10 – 30 min

T = 30 – 1 hr

Figure 6.16. The typical development of electron precipitation in a substorm. Note that the two zones are distinct on the day side. (S.-I. Akasofu, Polar and Magnetospheric Substorms. Reidel, 1968, with kind permission from Kluwer Academic Publishers.)

ior observed in the luminous aurora carries over into the various ionospheric effects (Section 6.3.4). The E-region reflections called radar aurora, radiowave absorption in the D region, X-ray generation, and the occurrence of magnetic bays are all, therefore, substorm phenomena. Some of these will be described in more detail in Section 6.5 and Chapter 7. 6.4.4

Substorm currents

Figure 6.17 shows one of the earlier descriptions of the current flowing during an individual substorm. This is still an equivalent-current system, because it assumes that the current flows only horizontally. Note that the intensity is relatively greater on the morning side of midnight. It will be seen that Figure 6.17 is considerably different from Figure 2.23, which showed the auroral electrojets converging on the Harang discontinuity at midnight. The patterns of Figures 6.17 and 2.23 are related rather as is the auroral oval to the auroral zone. However, the acceptance of field-aligned, or Birkeland, currents (Section 2.3.6) has fundamentally changed the approach to current modeling, because currents

6.4 The substorm

313

Sun 12 50°

60°

70°

80°

6

18

0 Local magnetic time Figure 6.17. The equivalent current system of a magnetic substorm. The concentrations of current lines in the early morning and near 1800 LT would appear as electrojets. (S.-I. Akasofu and S. Chapman, Solar–Terrestrial Physics. By permission of Oxford University Press, 1972.)

within the magnetosphere as well as currents flowing between the magnetosphere and the ionosphere may be included in the circuit. Despite general agreement on this point, the form of the current system during a substorm is still a topic of investigation. One influential concept in present-day modeling of substorm currents is that of the current wedge. As we shall see, the magnetotail collapses in a limited region when a substorm begins, and the cross-tail current from that region becomes diverted along field-lines (as Birkeland currents) into the ionosphere. There the circuit is completed in the E region, probably by an electrojet flowing along an arc or through some other form whose conductivity is enhanced by particle precipitation. Figure 6.18(a) shows the magnetospheric part of the circuit, and Figure 6.18(b) the “substorm electrojet” in the ionosphere. The electrojet, which flows

The aurora, substorm, and E region

314

(a)

Electrojet N

Field Aligned Currents

Near Earth Neutral Line

(b) substorm electrojets

S

Tail Axis Tail Current

Figure 6.18. (a) The substorm current wedge due to the diversion of tail current to the ionosphere. (Y. Kamide, Report ESA SP-389, 1996, after McPherron et al., 1973.) (b) The substorm electrojet in the auroral zone. (G. Rostoker, in Magnetospheric Substorms, copyright by the American Geophysical Union, 1991.)

westward in the midnight sector, connects to an upward field-aligned current at its western end, which also concides with a bright auroral feature. However, this is only part of the picture. The local collapse in the magnetotail at the onset of a substorm accelerates particles towards the Earth, and some become trapped to form a partial ring current (Section 2.3.5) that is completed by Birkeland currents to the ionosphere and currents within the ionosphere. These are driven, at least in part, by the electric field due to the general polar convection (Sections 2.4.1 and 2.4.3), which is likely to be enhanced during substorm activity. Various suggestions about the form and relationship of these currents flowing during a substorm have been put forward. Figure 6.19 indicates one possibility; Figure 6.19(b) shows the “convection electrojet” in the ionosphere. Kamide (1996) has pointed out that, whereas the enhanced conductivity of the ionosphere is the main factor controling the westward electrojet near midnight, the eastward electrojet which flows in the late evening sector, before the Harang discontinuity, is dominated by a northward electric field (in the northern hemisphere). An electric field, this time southward, also dominates the situation in the westward electrojet later in the morning (Figure 6.20). These are the fields generated by plasma convection. The components due to the wedge of current and convection do not vary in the same way during a substorm, however – their typical time constants are 15 min and 2 h, respectively (Rostoker, 1991) – so that the total behavior cannot be expected to be simple, or even the same in all cases. The total electrojet should be a combination of Figures 18(b) and 19(b), but in varying amounts. Although the electrojet is usually conceived in terms of a single wedge of current, more recent studies (Rostoker, 1991) have shown that it is composed of a sequence of short bursts (lasting about 12 min) of westward current (christened wedgelets!), following one after the other and often appearing sequentially further to the west. Thus there is a gradual westward progression of the current, as there is of the luminosity (Section 6.4.2).

6.4 The substorm

(a)

315

(b)

TO SUN

convection electrojets

DAWN

DUSK

Figure 6.19. (a) Magnetospheric currents showing the ring current and associated Birkeland currents. (Y. I. Feldstein, in Magnetospheric Substorms, copyright by the American Geophysical Union, 1991.) (b) The substorm electrojet in the auroral zone. (G. Rostoker, in Magnetospheric Substorms, copyright by the American Geophysical Union, 1991.)

Electric Field Dominant

06 E

ARD T TW EAS TROJE C ELE

18

Electric Field Dominant

E

RD WA JET T S O WE CTR E EL

20

04

Ha n ra g

Conductivity Dominant

Di sc on

E

uit tin

22

y

02 00

Magnetic Local Time

Figure 6.20. Regions of the electrojet dominated by conductivity and by electric fields. (Y. Kamide, in Auroral Physics, Cambridge University Press, 1991, p. 385.)

6.4.5

The substorm in the magnetosphere

If the substorm is the unit of auroral activity, then it is important to discover the details of the phenomena through which the substorm is revealed, and this includes its appearence in the magnetosphere. Moreover, if we are to predict when the high-latitude ionosphere is likely to be affected by substorms we have to understand the essential nature of the substorm and the factors that make it happen; these factors concern the dynamics of the magnetosphere and its interaction with the solar wind. It is another topic that is still being actively researched, both experimentally and theoretically. Although the final story has not yet emerged, there are some aspects that seem well established.

The aurora, substorm, and E region

316

Field-line circulation The circulation of the magnetosphere, discussed in Section 2.4, is mainly driven by magnetic merging on the sunward side of the magnetopause, but its continuity depends on reconnection in the plasmasheet which lies along the central plane of the tail. If we select a field-line at high latitude on the day side of the Earth and follow its progress, we find that it goes through a sequence: (1)

connecting with the IMF which divides it in two;

(2)

convecting over the north and south poles as two separate halves;

(3)

reconnecting in the tail; and

(4)

returning to a more dipolar form and returning to the day side.

In a steady state these stages would be in balance. Substorms occur because neither the dayside connection nor the nightside merging are continuous processes. Thus, energy accumulates in the tail and the substorm marks its sudden release. The Interplanetary Magnetic Field (IMF) reaching the Earth lies principally in the ecliptic plane, but it generally has some northward or southward component as well, and it couples most strongly with the geomagnetic field when that component is southward (Section 2.4.2). When the IMF turns from northward to southward, the connection rate increases. For a while more open field is produced than removed, the total magnetic flux in the polar cap increases, the auroral oval moves equatorward, and the tail of the magnetosphere grows fatter, representing a store of energy. This energy is released in the substorm, when the reconnection rate in the tail exceeds the supply of magnetic flux coming from the polar regions, the tail becomes more dipolar, flux is lost from the polar caps and the auroral oval shrinks again. This sequence of events in the magnetosphere may be identified with the phases of the auroral substorm observed from the ground: the growth phase, the expansion phase, and the recovery phase.

Behavior in the tail In the magnetosphere the growth phase corresponds to an increase in erosion from the front of the magnetosphere, and the plasma sheet and the current sheet become thinner (though not necessarily at the same time) as illustrated in Figure 6.21(a). One concept of the expansion phase begins with the formation of a neutral line nearer to the Earth than that which exists during quiet times (Figure 6.21(b)). Here the magnetotail collapses because the magnetic field has gone to zero in a localized region, and the cross-tail electric current is diverted into the current wedge as described above. In fact the collapse and the diversion of current must go together because of the Biot–Savart law. At the same time, satellites at geosynchronous distance observe an increase in the flux of energetic electrons, and the geomagnetic field becomes more dipolar.

6.4 The substorm

317

(a) EQUATORWARD MOTION ENLARGEMENT REDUCTION THINNING

INWARD MOTION EROSION

(b)

PLASMOID

N1

N2 N3

Figure 6.21. (a) Changes in the magnetosphere during the growth phase of a substorm. (R. L. McPherron et al., J. Geophys. Res. 78, 3131, 1973, copyright by the American Geophysical Union.) (b) The near-Earth neutral line (N3) and the plasmoid formed in a substorm. N1 is the merging point at the front of the magnetosphere and N2 the merging region in the distant tail. (D. P. Stern, Rev. Geophys. 34, 1, 1996, copyright by the American Geophysical Union.)

The region of the tail between the two neutral lines forms a plasmoid, which is ejected along the magnetotail as the recovery phase begins; this may be detected by satellites (20–100)RE down the tail as a burst of energetic particles moving away from the Earth. Satellites near the neutral sheet detect a loss of particle flux during the expansion phase, indicating that the plasma sheet becomes thinner at that time.

The aurora, substorm, and E region

318

Figure 6.22. A Y-type neutral point at the Earthward edge of the neutral sheet where the high-speed ion flow is stopped by the dipolelike field of the inner magnetosphere. (K. Shiokawa et al., Geophys. Res. Lett. 24, 1179, 1997, copyright by the American Geophysical Union.)

Quite recently, spacecraft in the magnetotail (AMPTE and GEOTAIL) have observed bursty bulk flows (BBFs) in the plasma sheet. Plasma in the sheet generally flows at speeds of less than 100 km s1, but during a BBF, which typically lasts for about 10 min, the speed exceeds 400 km s1 and the direction of flow is Earthward. BBFs occur at all observed distances beyond 15RE, and appear to be associated with the occurrence of substorms (Angelopoulos, 1996). If they are observed a long way down the tail (e.g. at (90–100)RE) the event appears some 90 min after the substorm, suggesting that there is a centre of acceleration that retreats progressively down the tail away from the Earth. When the fast-flowing plasma gets closer to the Earth, it is stopped by the stronger geomagnetic field which is dipolar in form. Figure 6.22 illustrates the stopping region. Note that the magnetic field is configured with a Y-type neutral line (compare with Figure 2.20). In this treatment the flow stops at the boundary between the tail field and the dipole field, and this region is also the inner edge of the plasma sheet.

Various theories The exact configuration of the magnetotail during the phases of a substorm has not been established fully, and several models – among which the observations have not yet been able to distinguish – have been put forward. Some models involve a local reversal of the tail field, and others include multiple neutral lines to correspond to the multiple arcs seen in aurorae. Liu (1992) has summarized the contending theories as six models. (a)

Formation of a neutral line in the magnetotail at a distance of (10–20)RE, which allows magnetic reconnection between the lobes of the magnetotail.

(b)

Generation of the Kelvin–Helmholtz instability (Section 2.5.6) in the mag-

6.4 The substorm

319

netospheric boundary layer, by enhanced reconnection at a neutral line some 100RE distant down the tail. (c)

A “thermal catastrophe” in the plasma sheet, due to the sheet becoming opaque to Alfvén waves and consequent sudden heating.

(d)

Intense field-aligned currents due to an increase in the rate of field reconnection on the day side of the magnetosphere, leading to the “current wedge” of the substorm and a collapse of field in the magnetotail.

(e)

A disruption of the cross-tail current due to a current instabilty.

(f)

A “ballooning instability” invoking a transition between field configurations that are essentially dipolar and essentially tail-like, which again diverts the cross-tail current.

These are illustrated in Figure 6.23, but nothing would be gained by going into all their details here. The task of making a synthesis of all the various substorm observations and theories has been considered by Elphinstone et al. (1996). 6.4.6

The influence of the IMF and the question of substorm triggering The magnetic power of the solar wind

It is clear that the term “substorm” includes a considerable range of phenomena, but the central idea is of a sudden and sporadic episode in the magnetosphere in which a large amount of stored energy is released. The energy comes initially from the solar wind, and important factors in the occurrence of substorms, therefore, are the energy flux of the solar wind and the efficiency with which the energy is coupled into the magnetosphere. It is found that the index AE, which indicates the level of geomagnetic activity in the northern auroral zone, is well correlated to a quantity vB2 sin4( /2)l 20,

(6.2)

where v is the solar wind speed, B is the magnitude of the IMF, l0 is a length related to the cross-section of the magnetosphere (7RE), and  is the “clock angle” of the IMF seen from the Earth (as defined in Section 5.1.2.). The magnetic energy reaching the magnetosphere per unit time is proportional to vB2l02: this is the “magnetic power” of the solar wind. The expression sin4( /2) is intended to represent the fraction of this power coupled into the magnetosphere. Its form gives a gradual transition between full coupling when the IMF is fully southward (sin4( /2)1) and zero coupling when it is fully northward (sin4( /2)0). If Bz By,  /245°, and the coupling factor is 0.25. Some other expressions based on different combinations of solar-wind parameters also correlate to the occurrence of substorms, though ) is perhaps the best (Figure 6.24).

320

The aurora, substorm, and E region

Figure 6.23. A selection of substorm-triggering ideas. (A. T. Y. Lui, in Magnetospheric Substorms, copyright by the American Geophysical Union, 1991.)

The influence of Bz on triggering Although much is known about the sequence of events in a substorm, it is not clear what causes the event to begin. Clearly, a store of energy must have built up awaiting release, but then we still have to ask whether the substorm is triggered by some other identifiable event, for instance in the solar wind, or whether it might be a spontaneous phenomenon without apparent cause. This point is clearly a vital element in any substorm theory, and (at the time of writing) it is not yet settled.

6.4 The substorm

Figure 6.24. Correlation between the AE index of magnetic activity and the parameter during a storm in July 1974. (Reprinted from S.-I. Akasofu, Planet. Space Sci. 27, 425, copyright 1979, with permission from Elsevier Science.)

Some things can be said, however. That substorms occur most frequently when Bz is southward has been known for many years, and it is found that the beginning of the substorm often coincides with a southward turning of the IMF. However, there are also cases when the substorm begins as the IMF turns northward, having previously been southward for an hour or two. In such a case it appears that a southward IMF puts energy into the magnetosphere and then the shock of the northward turning in some way triggers its release. Indeed, in many cases the growth phase begins as the IMF turns south. It is possible that there may be more than one kind of trigger.

The substorm rate If the speed of the solar wind is 440 km s1 (Figure 2.2(a)), the substorm rate is in the range 800–1500 per annum, or about 2–4 a day on the average (Borovsky et al., 1993; N. Flowers, private communication). Also, the rate of occurrence of substorms increases with the speed of the solar wind. Observations of substorms in auroral absorption give a median rate of 3.8 per day at 450 km s1 increasing to 7.2 per day at 700 km s1, approximately a v 2sw dependence (Hargreaves, 1996). 6.4.7

Relations between the storm and the substorm

Storms and substorms are defined in different ways, the former mainly from magnetic observations at low latitude where the greatest influence is the ring current,

321

The aurora, substorm, and E region

322

and the latter from observations at a higher latitude where the greatest contribution comes from the auroral electrojet. Their effects at the ground are usually represented by the magnetic indices Dst and AE (Sections 2.5.2 and 2.5.4). It is well known that during a significant storm there will almost certainly be one or more substorms. On the other hand, substorms may very well occur when there is no storm. Because it is the more frequent occurrence, the substorm has usually been regarded as the fundamental element, which, it is commonly supposed, leads to an increase in the population of trapped particles in which an increased ring current may then flow. This view is supported by direct observations in the magnetotail, in which a significant difference between those substorms occurring during a storm and those occurring at other times is found (Baumjohann, 1996). In a “storm-time substorm” the magnetic field moves from tail-like to dipole form in a matter of 15–30 min, whereas in “non-storm substorms” the change is both slower and less complete. There is a greater reduction of magnetic pressure in the storm-time substorm and the ion temperature in the tail is larger throughout. These results suggest that there are two kinds of substorm (perhaps arising at different distances down the tail), one of which is the more effective at populating the ring current (and thus promoting the signature of the classical magnetic storm). There are also contrary arguments, notably from studies showing that the ring current is as likely to grow before the auroral activity begins as it is to follow it. Furthermore, there is a good correlation between the magnitude of the ring current and the electric field across the magnetosphere, showing that the solar wind affects the magnetic storm directly, as well as indirectly via substorm activity (Clauer and McPherron, 1980). Plainly, the nature of the storm–substorm relationship is not yet fully understood.

6.5

The E region at high latitude

6.5.1

Introduction

At middle latitude the E region is easily the most boring part of the ionosphere. It behaves as an -Chapman layer (Section 1.3.3) and supports the (Sq) current generated by atmospheric tides. Sporadic-E (Section 1.4.2) adds some interest, but beyond that there is little more to be said. The same is true at high latitude while geophysical conditions are quiet, but, when the Sun is active, the high-latitude E region becomes arguably one of the most exciting parts of the ionosphere. It then differs markedly from the mid-latitude and equatorial E regions – in terms of ionization sources, plasma processes, and radio-propagation characteristics. It is often the case that precipitating particles are the dominant source of ionization.

6.5 The E region at high latitude

When thus enhanced the E region supports the auroral electrojet, in which instabilities may arise. Ionograms exhibit sporadic-E of the auroral variety. 6.5.2

The polar E layer

The most benign part of the high-latitude E layer is over the polar cap – that is, poleward of the auroral zone. Here it is essentially under solar control; it varies with the solar zenith angle and exhibits strong seasonal effects, as does the midlatitude E region. 6.5.3

The auroral E layer under quiet conditions

When Kp is small the auroral oval retreats poleward (see Figure 6.7), and, under quiet conditions, there is relatively little disturbance by precipitating particles in the nominal auroral zone, from, say, 60° to 70° magnetic latitude. The normal ionospheric layers occur much as at middle latitude and are subject to the same diurnal, seasonal, and sunspot-cycle variations. Figure 6.25 shows typical electron-density profiles from the Chatanika incoherent-scatter radar (ISR) compared with ionograms from College, Alaska, sites both near 65° magnetic latitude. These results are for magnetically quiet conditions near sunspot minimum. Because of its high latitude the site was illuminated by the Sun throughout the day. Thus, even the ionogram at 0215 LT shows a strong E layer that masks the F region, giving the G condition (Section 5.2.1). Observing by ionosonde alone would suggest, falsely, that the F layer was absent. Note that Te Ti in the F region, which is usual. 6.5.4

The disturbed auroral E layer

The main disturbances affecting the auroral E layer are geomagnetic storms and substorms (Section 6.4). With increasing activity, and particularly if Kp 3, the auroral oval expands both poleward and equatorward, and the auroral structure and motion become much more dynamic. The precipitating electrons of energy 1–10 keV which create the visual aurora also create the auroral-E ionization. As pointed out in Section 2.6, fast electrons (and protons) entering the atmosphere produce one ion pair (an ion plus an electron) for each 36 eV of energy lost, most of which is deposited towards the end of the path. Since the average ionization potential of the ionospheric atoms and molecules is about 15 eV, approximately 40% of the energy goes into ionization and 60% goes into the motion of the electron produced, which subsequently thermalizes. In the E region the neutral air is dense relative to the higher levels, and the recombination between electrons and ions proceeds rapidly. Altitude profiles of the rate of ionization due to a flux of 108 electrons cm2 s1 at several initial values of energy Ep (keV) precipitating along geomagnetic field

323

(a)

Virtual Height (km)

60 80 100 120 140 160 180 200 Electron Density (1010 m —3)

300

400

500

600

700

0 1

00

0 0

100

40

3000

Te

7 8 9 10

2000

6

100

20

Ti

2333 — 2247 Kp : 2—

Temperature (K) 1000

3 4 5 Frequency (MHz)

100

200

300

400

500

200

N

0

2

E

F1

F2

(b)

200

300

400

500

600

700

0 1

100

200

300

400

500

2230 UT

N

20

40

0

Ti

1211 — 1219 Kp : 2—

Temperature (K) 1000

3 4 5 Frequency (MHz)

7 8 9 10

2000

6

Te

3000

60 80 100 120 140 160 180 200 Electron Density (1010 m —3)

2

1215 UT

Figure 6.25. Comparable electron-density profiles from the Chatanika incoherent-scatter radar and ionograms from the College ionosonde, both in Alaska, 16 July 1971 Ion and electron temperatures are also given: (a) afternoon (Alaskan time being UT – 10 h), and (b) night. Note that, by night, the electron density is reduced, the valley between the E and F regions is more marked, and region F is masked by E, giving a simpler ionogram. (H. F. Bates and R. D. Hunsucker, Radio Sci. 9, 455, 1974, copyright by the American Geophysical Union.)

Height (km)

Virtual Height (km) Height (km)

6.5 The E region at high latitude

Figure 6.26. Profiles of the rate of ionization due to monoenergetic fluxes of 108 electrons cm2 s1, incident from above, for the energy range (2–100 keV) having greatest effect in the E (and D) regions. (M. H. Rees, Physics and Chemistry of the Upper Atmosphere, Cambridge University Press, 1989.)

lines into the auroral ionosphere are shown in Figure 6.26. It is instructive to compare these ionization profiles with luminosity profiles (for example Figure 6.8), as a general indication of the energies of the particles which cause the auroral luminosity and the disturbed E region. The regions of enhanced electron density are also regions of high conductivity and this is where the auroral electrojet flows, its magnitude increasing with the auroral luminosity. The form of these current systems is discussed in Sections 2.5.3 and 6.4.4. Plasma waves generated in the electrojet produce the various types of radar-backscatter signature discussed in Section 6.5.5. The relation between the visual aurora and auroral-E ionization has been studied in a semi-quantitative fashion by Hunsucker (1975). Figure 6.27 shows simultaneous all-sky camera, ionosonde, and incoherent-scatter radar data during the passage of an auroral arc through the fields of view of the instruments. The Eregion electron density is greatly increased within the arc. The rapid changes of electron density that may be observed by incoherentscatter-radar over an hour are illustrated in Figure 6.28. Note that, during the electron-density spike, the ionogram shows intense sporadic-E. The enhancement of the E region may be surprisingly large during major disturbances. During the

325

The aurora, substorm, and E region

326

(a)

(b)

Figure 6.27. E-region electron density profiles and ionograms associated with the passage of an auroral arc across the field of view, 2 March 1973. The incoherent-scatter radar was at Chatanika, Alaska, and the all-sky camera and ionosonde were at College nearby. The radar beam was just south of the arc in (a) but in the centre of the arc in (b). The maximum electron density and the penetration frequency of the E layer were greatly increased within the arc. It was late evening. (R. D. Hunsucker, Radio Sci. 10, 277, 1975, copyright by the American Geophysical Union.)

great magnetic storm of August 1972 the E-region electron density exceeded 1.8 106 cm3 (Figure 6.29), one of the higher values of electron density observed in the E region. The morphology, structure, and dynamics of the auroral-E layer have been described in some detail by Hunsucker (1975) and others referred to therein. 6.5.5

Auroral radar

Most of our knowledge of irregularities in the auroral E layer is based on data obtained by direct backscatter from field-aligned irregularities within the auroral

6.5 The E region at high latitude

Figure 6.28. The variation of electron density at three heights, showing a sharp spike in the electron density at 0800 UT (2200 LT) on 2 March 1973. At the same time the ionosonde registered sporadic-E above 7 MHz. Observations from Alaska, using the Chatanika ISR and the College ionosonde. (H. F. Bates and R. D. Hunsucker, Radio Sci. 9, 455, 1974, copyright by the American Geophysical Union.)

oval using radars operating in the VHF/UHF frequency range. The term radio aurora is often used for the phenomena so observed, and radar aurora is equally valid. There is now a voluminous literature on the subject. There is an enormous difference between the scattering cross-sections for coherent and incoherent radar, the former being the stronger by 50–80 dB. Backscatter from auroral E-layer irregularities has been classified into four groups in terms of their line-of-sight Doppler velocities, as shown in Figure 6.30, which also summarizes their essential properties.

327

328

The aurora, substorm, and E region

Figure 6.29. An electron-density profile of the auroral E region during a great magnetic storm in August 1972. (R. D. Hunsucker, Radio Sci. 10, 277, 1975, copyright by the American Geophysical Union.)

Theory The two most generally accepted theories to explain the observations are the twostream plasma instability and the gradient-drift mechanism. These plasma instabilities are generated in the auroral electrojet and produce electrostatic ion waves that may scatter incident radio waves as discussed in Section 3.5.1. A necessary condition for the occurrence of these instabilities is a sufficiently large relative velocity between the electrons drifting in an electric field and the ions whose motion is dominated by collisions. The waves travel nearly perpendicular to the geomagnetic field lines. The latter property implies that the backscatter cross-section is maximum when the radar is pointing almost perpendicular to the field line, although there have been several instances of auroral backscatter occurring at large aspect angles. Other physical mechanisms for producing the auroral irregularities have also been proposed. A critical review of plasma irregularities in the auroral electrojet, covering both theory and experiment, has been given by Sahr and Fejer (1996).

Polarization Investigations of the polarization of backscatter from auroral E-region irregularities have concluded that coherent scatter of spectral classes 1, 2, and 3 has similar polarization characteristics. For most of the observations, the scattering of a linearly polarized incident wave produced an essentially linear and highly polarized scattered wave, implying that there was a small scattering volume and/or a small number of discrete scatterers located close to one another. This also confirms that the scattering process is a weak coherent one. There were, however, some significant exceptions.

Observing geometry and occurrence The region which may be observed is restricted by the aspect sensitivity. As Figure 6.31 illustrates, the radar must be situated at middle latitude if the beam is to intersect field-aligned irregularities in the auroral zone.

6.5 The E region at high latitude

Figure 6.30. Four types of echo in auroral radar. The observations were with a 50-MHz radar in March 1989, and each analysis was based on 20 s of data. Cs is the local speed of sound. (J. D. Sahr and B. G. Fejer, J. Geophys. Res. 101, 26 893, 1996, copyright by the American Geophysical Union.)

Broadly speaking, the diurnal and seasonal occurrence of the radio aurora is similar to that of the visual aurora, except, of course, during daylight when the visual aurora cannot be seen. The strongest echoes occur near 65° latitude, and during disturbances the echoing region extends equatorward. The echoes can be detected at any time of day, but are most pronounced near local magnetic midnight. The phenomenon of E-region irregularities is closely related to the behavior of the auroral electrojet. Sahr and Fejer (1996) draw attention to the problem of modeling them globally, to their importance in radio propagation, and to gaps in methods of data analysis.

329

The aurora, substorm, and E region

330

Figure 6.31. The geometry of coherent radar-scatter at high latitude. For significant scattering the beam must be normal to the field-aligned irregularites to within about 2°. In the absence of refraction (at VHF and UHF) the range will be between 400 and 1200 km. (J. D. Sahr and B. G. Fejer, J. Geophys. Res. 101, 26 893, 1996, copyright by the American Geophysical Union.)

6.5.6

Auroral infrasonic waves

Coming within the acoustic-wave domain of atmospheric waves (Section 1.6), auroral infrasonic waves (AIWs) are an interesting, though not very well known, feature in the spectrum of high-latitude phenomena (Wilson, 1969). They originate in the supersonic horizontal motion of the large-scale electrojets that flow within auroral arcs. The motion produces an infrasonic “bow wave,” whose propagation is highly anisotropic and which reaches the Earth’s surface about 6 min after the zenith passage of the arc. AIWs can sometimes be detected at the ground as much as 1000 km from the source. The period will be in the range 10–100 s – a frequency range of 0.01–0.1 Hz – with maximum spectral power density at a period of about 70 s, the pressure wave having amplitude from 0.5 to 20 bars. They are detected with microphone arrays and are found to be highly coherent across arrays whose sensors are spaced by up to 6 km. The speed at which an AIW crosses the array will be between about 300 and 1000 m s1, the average being about 500 m s1 (Wilson et al., 1976). An example observed with a microphone array at Fairbanks, Alaska, is shown in Figure 6.32. AIWs occur on the night side of the Earth, having a diurnal occurrence maximum near local midnight and a seasonal maximum around the equinoxes. Episodes of AIW activity, AIW substorms, are highly correlated to the occurrence of negative magnetic bays (Section 2.5.3). The horizontal component of the AIW’s velocity is parallel to the motion of the supersonic auroral arc and also parallel to the horizontal magnetic-disturbance vector associated with the westward electrojet within the arc. It is an interesting fact that AIWs are generated only by arcs moving equatorward; those moving poleward have no such effect. It has also been noted (Wilson and Hargreaves, 1974) that, statistically speaking, their direction of movement is similar to that of peaks in auroral radio absorption (Section 7.2.4) at a similar lat-

6.5 The E region at high latitude

Figure 6.32. An auroral infrasonic wave observed with a detector array at Fairbanks, Alaska. The upper panel shows the power spectrum of the event, and the lower one shows the waveform. The wave traveled at 502 m s1, arriving from azimuth 27.6°. (R. D. Hunsucker, private communication.)

itude, having, in addition to the equatorward motion, a westward component before midnight and an eastward one after midnight. AIWs should be considered as part of the total energy budget of terrestrial auroral phenomena and as a sensitive sensor of dynamic auroral arcs. 6.5.7

The generation of acoustic gravity waves

Another consequence of the electron-precipitation events, which create large electron densities supporting electrojets, is that the auroral E region is a source of acoustic-gravity waves (AGWs). Strictly, there are two mechanisms. One is intense Joule heating (J• E), where J is the current density and E is the electric field, which occurs in localized regions. The other is the Lorenz force (J B), where B is the flux density of the geomagnetic field. AGWs were introduced in Section 1.6 and it has been shown that those in the “large-scale” category originate in the auroral regions, probably from one of these sources. In the ionosphere the AGW is recognized as a traveling ionospheric disturbance (TID), which propagates in the F region, primarily equatorward, for distances that may exceed 10 000 km.

331

The aurora, substorm, and E region

332

In an investigation of TIDs in electron content at L4, most of which were in the “medium-scale” range with period 20–60 min, Hunsucker and Hargreaves (1988) noted that the waves were present almost continuously during daylight hours at the level of 1–4% of the electron content. Although no specifc source was identified, it must be significant that the incidence of these waves was far greater at L4 than it is at middle latitudes. Some of the energy deposited in the auroral ionosphere from the magnetosphere may be transported to other latitudes by the action of AGWs, as well as by neutral-air winds and tides. It is estimated that 5–10% of this redistribution is due to AGWs.

6.6

Summary and implications

Except for the very large seasonal variability, the polar E region is relatively benign, compared with the auroral region. Precipitation of 1–10 keV electrons along geomagnetic field lines through the magnetospheric plasma sheet into the auroral ionosphere produces several very important effects: the luminous aurora, anomalously high E-region electron densities (conductivity), and localized regions of intense Joule heating and Lorenz forces. These phenomena are organized by the geomagnetic field into the northern and southern auroral ovals, which are stationary in space in Sun–Earth coordinates, with the Earth rotating underneath. The ovals are centered at approximately 67° geomagnetic latitude near magnetic midnight and 77° geomagnetic latitude near geomagnetic noon under “quiet conditions,” and the latitudinal “thickness” of the oval increases as Kp increases. The most used auroral oval models are those derived from visual auroral observations, which give a reasonable estimate of auroral-E ionization for Kp values up through Kp 7. Other ovals based on the TIROS and DMSP satellite particle measurements, which may give a more accurate mapping both of the electron precipitation which produces the auroral-E ionization and of the particle precipitation which produces D-region absorption and F-region irregularities, are also available. The altitude profiles of auroral luminosity and of electron density in the E region are almost identical in shape. Electron densities as high as 4.4 106 electrons cm3 have been measured by the Chatanika ISR during a large geomagnetic storm, and densities from ⬃5 105 to 106 electrons cm3 are quite common around magnetic midnight near 65° north geomagnetic latitude (Fairbanks, Alaska). The College (Fairbanks) ionosonde has also measured an auroral-E top frequency of 13 MHz, which corresponds to an oblique frequency of 57 MHz on a 1000-km long, curved-Earth-limited one-hop propagation path. There is some question, however, regarding whether the auroral-E top frequency measured by an ionosonde is really a true plasma frequency. The temporal and spatial behavior of the auroral-E layer is dynamic and is probably best demonstrated by observing the visual aurora and realizing that the

6.7 References and bibliography

regions of highest intensity (brightness) are, indeed, regions of high auroral-E electron density The regions of high intensity in the visual aurora are also regions of intensified conductivity (hence electric current) of the auroral electrojet in the E region. These enhanced currents can produce intense Joule heating and Lorenz forces, which in turn generate atmospheric gravity waves (AGWs) that couple with the electron plasma to produce traveling ionospheric disturbances (TIDs). The largescale TIDs may travel equatorward in the F region for over 10 000 km, creating anomalous propagation in mid-latitude HF systems. Another effect of auroral-oval E-region dynamics is the generation of auroral infrasonic waves (AIWs), which occur when an auroral arc travels supersonically towards the equator. For certain conditions of electron density, auroral brightness, particle-precipitation energies, Mach number, and orientation of the arc with respect to the geomagnetic field, a “bow shock wave” is formed, launching AIWs that may then be detected by a suitable array of acoustic sensors on the ground.

6.7

References and bibliography

6.2

Statistical distribution of the aurora

Akasofu, S.-I. (1968) Polar and Magnetospheric Substorms, Reidel, Dordrecht. Akasofu, S.-I. (1974) The aurora and the magnetosphere; the Chapman memorial lecture. Planet. Space Sci., 22, 885. Chubb, T. A. and Hicks, G. T. (1970) Observations of the aurora in the far ultraviolet from OGO 4. J. Geophys. Res. 75, 1290. Feldstein, Y. I. and Starkov, G. V. (1967) Dynamics of auroral belt and polar geomagnetic disturbances. Planet. Space Sci. 15, 209. Frank, L. A. and Craven, J. D. (1988) Imaging results from Dynamics Explorer 1. J. Geophys. Res. 26, 246. Fuller-Rowell, T. J. and Evans, D. S. (1987) Height-integrated Pedersen and Hall conductivity patterns inferred from TIROS-NOAA satellite data. J. Geophys. Res. 92, 7606. Goodman, J. M. (1992) HF Communications – Science and Technology. Van Nostrand Reinhold, New York. Gussenhoven, M. S., Hardy, D. A., and Heinemann, N. (1983) Systematics of the equatorward diffuse auroral boundary. J. Geophys. Res. 88, 5692. Hardy, D. A., Gussenhoven, M. S., and Holeman, E. (1985) A statistical model of auroral electron precipitation. J. Geophys. Res. 90, 4229. Hardy, D.A., Gussenhoven, M. S., and Brautigan, D. (1989) A statistical model of auroral ion precipitation. J. Geophys. Res. 94, 370. Meng, C.-I. (1984) Dynamic variation of the auroral oval during intense magnetic storms. J. Geophys. Res. 89, 227. Meng, C.-I. and Makita. K. (1986) Dynamic variations of the polar cap. Solar Wind–Magnetosphere Coupling (eds. Kamide and Slavin), p. 605. Terra Scientific, Tokyo. Vestine, E. H. (1944) The geographic incidence of aurora and magnetic disturbance, Northern Hemisphere. Terr. Magn. Atmos. Electricity 49, 77.

333

The aurora, substorm, and E region

334

Whalen, J. A. (1970) Auroral oval plotter and nomograph for determining geomagnetic local time, latitude and longitude in the Northern Hemisphere. Report AFCRL70-0422, Environmental Research Paper 327. (From Defense Technical Information Center, Cameron Station, Alexandria, VA 22314, USA)

6.3

The auroral phenomena

Bauer, S. J. (1973) Physics of Planetary Ionospheres. Springer-Verlag, Berlin. Gazey, N. G. J., Smith, P. N., Rijnbeek, R. P., Buchan, M., and Lockwood, M. (1996) The motion of auroral arcs within the convective plasma flow. Third International Conference on Substorms, Versailles, France. Report ESA SP-389, p. 11. Harang, L. (1951) The Aurorae. Wiley, New York. Hargreaves, J. K. (1969) Auroral absorption of HF radio waves in the ionosphere – a review of results from the first decade of riometry. Proc. Inst. Elect. Electronics Engineer 57, 1348. Hartz, T. R. and Brice, N. M. (1967) The general pattern of auroral particle precipitation. Planet. Space Sci. 15, 301. Rees, M. H. (1989) Physics and Chemistry of the Upper Atmosphere. Cambridge University Press, Cambridge. Störmer, C. (1955) The Polar Aurora. Oxford University Press, Oxford.

6.4

The substorm

Akasofu, S.-I. (1964) The development of the auroral substorm. Planet. Space Sci. 12, 273. Akasofu, S.-I. (1968) Polar and Magnetospheric Substorms. Springer-Verlag, New York. Akasofu, S.-I. (1970) In memoriam Sydney Chapman. Space Sci. Rev. 11, 599. Akasofu, S.-I. (1974) A study of auroral displays photographed from the DMSP-2 satellite and from the Alaska meridian chain of stations. Space Sci. Rev. 16, 617. Akasofu, S.-I. (1977) Physics of Magnetospheric Substorms. Reidel, Dordrecht. Akasofu, S.-I. (1979) Interplanetary energy flux associated with magnetospheric storms. Planet. Space Sci. 27, 425. Akasofu, S.-I. and Chapman, S. (1961) The ring current, geomagnetic disturbance and the Van Allen radiation belts. J. Geophys. Res. 66, 1321. Akasofu, S.-I. and Chapman, S. (1972) Solar–Terrestrial Physics. Oxford University Press, Oxford. Angelopoulos, V. (1996) The role of impulsive particle acceleration in magnetotail circulation. Third International Conference on Substorms, Versailles, France. Report ESA SP-389, p. 17. Birkeland, K. (1908) The Norwegian Aurora Polaris Expedition 1902–3, vol. 1, section 1. H. Aschehoug, Christiana. Baumjohann, W. (1996) Storm–substorm relationship. Third International Conference on Substorms, Versailles, France. Report ESA SP-389, p. 627. Borovsky, J. E., Nemzek, R. J., and Belian, R. D. (1993) The occurrence rate of magnetospheric-substorm onsets. J. Geophys. Res. 98, 3807. Clauer, C. R. and McPherron, R. L. (1980) The relative importance of the interplane-

6.7 References and bibliography

tary electric field and magnetospheric substorms on partial current development. J. Geophys. Res. 85, 6747. Elphinstone, R. D., Murphree, J. S., Cogger, L. L., Hearn, D., and Henderson, M. G. (1991) Observations of changes to the auroral distribution prior to substorm onset. Magnetospheric Substorms (eds. J. R. Kan, T. A. Potemra, S. Kokubun, and T. Iijima), p. 257. American Geophysical Union, Washington DC. Elphinstone, R. D., Murphree, J. S., and Cogger, L. L. (1996) What is a global auroral substorm? Rev. Geophys. 34, 169. Feldstein, Y. I. (1991) Substorm current systems and auroral dynamics. Magnetospheric Substorms (eds. J. R. Kan, T. A. Potemra, S. Kokubun and T. Iijima), p. 29. American Geophysical Union, Washington DC. Hargreaves, J. K. (1996) Substorm effects in the D region. Third International Conference on Substorms, Versailles, France. Report ESA SP-389, p. 663. Kamide, Y. (1991) The auroral electrojets: relative importance of ionospheric conductivities and electric fields. Auroral Physics (eds. C.-I. Meng, M. J. Rycroft, and L. A. Frank), p. 385. Cambridge University Press, Cambridge. Lui, A. T. Y. (1991) Extended consideration of a synthesis model for magnetospheric substorms. Magnetospheric Substorms (eds. J. R. Kan, T. A. Potemra, S. Kokubun, and T. Iijima), p. 43. American Geophysical Union, Washington DC. Lui, A. T. Y. (1992) Magnetospheric substorms. Phys. Fluids B, 4, 2257. McPherron, R. L., Russell, C. T., and Aubry, M. P. (1973) Satellite studies of magnetospheric substorms on August 15, 1968: 9. Phenomenological model for substorms. J. Geophys. Res. 78, 3131. Murphree, J. S., Elphinstone, R. D., Cogger, L. L., and Hearn, D. (1991) Viking optical substorm signatures. Magnetospheric Substorms (eds. J. R. Kan, T. A. Potemra, S. Kokubun, and T. Iijima), p. 241. American Geophysical Union, Washington DC. Pulkkinen, T. I. (1996) Pseudobreakup or substorm? Third International Conference on Substorms, Versailles, France. Report ESA SP-389, p. 285. Rostocker, G. (1991) Some observational constraints for substorm models. Magnetospheric Substorms (eds. J. R. Kan, T. A. Potemra, S. Kokubun, and T. Iijima), p. 61. American Geophysical Union, Washington DC. Rostoker, G., Akasofu, S.-I., Foster, J., Greenwald, R. A., Kamide, Y., Kawasaki, K., Liu, A. T. Y,. McPherron, R. L., and Russell, C. T. (1980) Magnetospheric substorms – definitions and signatures. J. Geophys. Res. 85, 1663. Shiokawa, K., Baumjohann, W., and Haerendel, G. (1997) Braking of high-speed flows in the near-Earth tail. Geophys. Res. Lett. 24, 1179. Stern, D. P. (1991) The beginning of substorm research. Magnetospheric Substorms (eds. J. R. Kan, T. A. Potemra, S. Kokubun, and T. Iijima), p. 11. American Geophysical Union, Washington DC. Stern, D. P. (1996) A brief history of magnetospheric physics during the space age. Rev. Geophysics 34, 1.

6.5

The E region at high latitude

Bates, H. F. and Hunsucker, R. D. (1974) Quiet and disturbed electron density profiles in the auroral zone ionosphere. Radio Sci. 9, 455.

335

The aurora, substorm, and E region

336

Hunsucker, R. D. (1975) Chatanika radar investigation of high-latitude E-region ionization structure and dynamics. Radio Sci. 10, 277. Hunsucker, R. D. and Hargreaves, J. K. (1988) A study of gravity waves in ionospheric electron content at L4. J. Atmos. Terr. Phys. 50, 167. Rees, M. H. (1989) Physics and Chemistry of the Upper Atmosphere. Cambridge University Press, Cambridge. Sahr, J. D. and Fejer, B. G. (1996)Auroral electrojet plasma irregularity theory and experiment: a critical review of present understanding and future directions. J. Geophys. Res. 101, 26 893. Wilson, C. R. (1969) Auroral infrasonic waves. J. Geophys. Res. 74, 1812. Wilson, C. R. and Hargreaves, J. K. (1974) The motions of peaks in ionospheric absorption and infrasonic waves. J. Atmos. Terr. Phys. 36, 1555. Wilson, C. R., Hunsucker, R. D., and Romick, G. J. (1976) An auroral substorm investigation using Chatanika radar and other geophysical sensors. Planet. Space Sci. 24, 1155.

General reading on the aurora and related topics Books Akasofu, S.-I. (1968) Polar and Magnetospheric Substorms. Reidel, Dordrecht. Akasofu, S.-I. (1977) Physics of Magnetospheric Substorms. Reidel, Dordrecht. Akasofu, S.-I. and Chapman, S. (1972) Solar–Terrestrial Physics. Oxford University Press, Oxford. Brekke, A. (1997) Physics of the Upper Polar Atmosphere. Wiley, Chichester. Brekke, A. and Egeland, A. (1983) The Northern Lights – From Mythology to Space Research. Springer-Verlag, Berlin. Chamberlain, J. W. (1961) Physics of the Aurora and Airglow. Academic Press, New York. Eather, R. H. (1980) Majestic Lights. American Geophysical Union, Washington, DC. Kamide, Y. and Baumjohann, W. (1993) Magnetosphere–Ionosphere Coupling. Springer-Verlag, Berlin. Kan, J. R., Potemra, T. A., Kokubun, S., and Iijima, T. (eds.) (1991) Magnetospheric Substorms. American Geophysical Union, Washington,DC. Kennel. C. F. (1995) Convection and Substorms. Oxford University Press, Oxford. Omholt, A. (1971) The Optical Aurora. Springer-Verlag, Berlin. Störmer, C. (1955) The Polar Aurora. Oxford University Press, Oxford. Vallance Jones, A. (1974) Aurora. Reidel, Dordrecht.

Conference reports Akasofu, S.-I. (ed.) (1980) Dynamics of the Magnetosphere. Reidel, Dordrecht. McCormac, B. M. (ed.) (1967) Aurora and Airglow. Van Nostrand Reinhold Co., New York. McCormac, B. M. and Omholt, A. (eds.) (1969) Atmospheric Emissions. Van Nostrand Reinhold Co., New York. McCormac, B. M. (ed.) (1971) The Radiating Atmosphere. Reidel, Dordrecht. Meng, C.-I., Rycroft, M. J., and Frank, L. A. (eds.) (1991) Auroral Physics. Cambridge University Press, Cambridge.

Chapter 7 The high-latitude D region

7.1

Introduction

The differences between the E and D regions in middle latitudes hold also at high latitude. The E region is characterized by relatively simple photochemistry and high electrical conductivity, whereas the D region below it has a complex and less well-known chemistry, the electric currents and plasma motions being inhibited by the higher atmospheric pressure. What they have in common at high latitude is the importance of ionization by energetic particles. Typical spectra include particles with energies such that they are stopped and ionize in both regions, the lower energies (for example, electrons of a few kilo-electron volts) affecting the E region and the higher ones (e.g. electrons with energies of tens of kilo-electron volts) penetrating into region D. Figure 7.1 shows electron-density profiles between 65 and 110 km due to representative spectra of ionizing electrons incident on the atmosphere from above. Increasing the characteristic energy of the spectrum lowers the peak of the layer, increasing the electron density in the D region but reducing it in the E region. At middle latitude the D region’s role in radio propagation is a secondary one. The main parameters of HF propagation are determined by the E and the F regions, and the D region acts mainly as an absorbing layer, reducing the strength of the signal but seldom preventing communications for any long period. At high latitude the D region may be much enhanced and then absorption becomes a considerable problem. There are two principal phenomena, each peculiar to high latitude. The first is auroral radio absorption (AA), which occurs only in the auroral regions and is due to fluxes of energetic electrons precipitated from the magneto-

337

10 2

70

80

90

100

110

10 3

80

40

20

10

E 0 (keV)

Electron density (cm –3)

10 4

10 5

Figure 7.1. Specimen profiles of the D and lower E regions due to electron precipitation. The incident electron flux is assumed to be of the form exp(E/E0), E being the energy in kilo-electron volts and E0 the characteristic energy, the total energy flux being 4 107 keV s1 cm2 sr1 in each case. Production rates were worked out using the method outlined in Section 2.6.1, and a profile of the effective recombination coefficient was assumed.

Height (km)

7.2 Auroral radio absorption

sphere sporadically during periods of auroral activity. The second is polar-cap absorption (PCA), which is caused by energetic protons emitted from the Sun, usually at the time of a major solar flare. These two kinds of phenomenon have rather different properties. The PCA is relatively infrequent, there being only about one event per month on the average in a year of high solar activity; many less when the Sun is quiet. However, when an event does occur, the absorption may be very strong. The absorbing region is relatively uniform over the whole polar cap, leading to HF black-out over a wide area. Auroral absorption is more common, but it is confined to the auroral zones and is generally more structured. Though the amount of absorption does not rise to the intensity sometimes seen in PCA, the spatial structure, which is generally not known in detail, adds to the difficulty of predicting the effects on HF propagation. AA and PCA are discussed in Sections 7.2 and 7.3, respectively. The chapter concludes with an introduction to a phenomenon that is still not well understood, the polar mesosphere summer echo.

7.2

Auroral radio absorption

7.2.1

Introduction – history and technique

Auroral radio absorption was discovered by Appleton and colleagues (Appleton et al., 1933) during an expedition to Tromsø during the International Polar Year (1932–1933), when it was observed that ionosondes were blacked out during periods of auroral and magnetic activity. The earliest studies of the phenomenon were performed using ionosondes, but this method is not entirely satisfactory because all that can be measured is the incidence of black-out, which in any case might not always be due to absorption; furthermore, ionosondes are not all equally sensitive. The absorption due to proton events, which has different properties, could also have confused the early results since the proton event was not at that time recognized as a separate entity. Since the International Geophysical Year of 1957–1958, auroral absorption (AA) has generally been studied with a riometer – and preferably with a group of riometers covering a range of latitude and/or longitude. Since it uses transionospheric propagation, the riometer does not say at what height the absorption occurs, but various studies have left little doubt that most of the absorption detected in the auroral regions arises in the D region of the ionosphere and is caused by energetic electrons arriving from the magnetosphere. Riometer technique is outlined in Section 4.2.4. We should recall here that the absorption is not measured directly but requires first the determination of a quietday curve (QDC), an estimate of the signal level in the absence of absorption.

339

The high-latitude D region

340

Figure 7.2. Auroral radio absorption observed with a 30-MHz riometer on 15 October 1963 at Byrd Station, Antarctica. The descriptions below the axis refer to the typical behavior; the evening minimum was not respected on the day shown! Note the difference of structure between the night and day activity. (After J. K. Hargreaves, Proc. Inst. Electr. Electron. Engineers 57, 1348, 1969, © 1969 IEEE.)

Although the idea is simple enough, the accurate derivation of the QDC can be the most difficult part of absorption measurement by the riometer technique. Most riometer-based absorption data come from instruments using a simple antenna, which, therefore, has a wide beam – e.g. 60° between half-power points – projecting onto a region about 100 km across in the D region. Therefore a standard riometer installation has limited spatial resolution. In recent years, however, there has been an increase in narrow-beam work and the use of imaging riometers. We shall quote results from both wide-beam and narrowbeam instruments. Since the absorption is strongly frequency-dependent (an inverse square law in most circumstances), the observing frequency must also be stated. The reduction of absorption with increasing frequency is one factor determining the optimum observing frequncy. At the lower frequencies the antenna is larger and also there is more interference from ionospherically propagated signals. The compromise has generally led to use of the band 30–50 MHz. When data are obtained at several frequencies, it is usual to reduce them to 30 MHz for comparison purposes: A(30 MHz)A( f )(30)2/f 2. 7.2.2

(7.1)

Typical auroral-absorption events and their temporal and spatial properties

One notable fact about auroral absorption is its temporal structure, distinguishing it from other major varieties of radio absorption, which generally vary more gradually. In the example of Figure 7.2 it is seen that the absorption tends to occur in bursts (or events); these show preferences for certain times of day, and they

7.2 Auroral radio absorption

(a)

(b)

Figure 7.3. Examples of sharp-onset night events. (a) Skibotn (L6.0), 4 November 1975 (J. K. Hargreaves et al., J. Geophys. Res. 84, 4225, 1979, copyright by the American Geophysical Union.) (b) Kilpisjärvi (L5.9), 6 October 1994. (Reprinted from J. K. Hargreaves et al., J. Atmos. Solar–Terr. Phys. 59, 872, copyright 1997, with permission from Elsevier Science.). The first is in the form of a traditional riometer chart, with the level proportional to the received power. The lower one was reconstructed from digital data and the signal power is plotted on a scale of decibels. The absorption is reckoned from the marked quiet-day curve.

change character between day and night. While there is no general classification of auroral absorption that covers all events, there are some recognized types that occur frequently.

Sharp-onset and spike events at night Occurring near magnetic midnight (and more before than after) are sharp-onset events. Here the event rises in a couple of minutes or less (Figure 7.3). The duration is tens of minutes to an hour. Some of these events appear isolated, but others are followed by continuing activity lasting for several hours. The continuing activity tends to be less prominent at the higher latitudes. At some other latitudes what appears to be the same event may begin with a more gradual onset. Many sharp onsets, though not all, coincide with the beginning of a substorm. At the beginning of the event there may be a “spike,” as in the examples of Figure 7.3, in which case that feature is a spike event. The occurrence of spike events at L5.6 is shown in Figure 7.4. At that location, half occurred in a 3-h period up to local magnetic midnight. Usually, an individual spike is seen over a

341

342

The high-latitude D region

Figure 7.4. The occurrence of spike events (1 dB) at Abisko, 1980–1985. Magnetic midnight is about 2130 UT. The riometer frequency was 30 MHz. (Reprinted from J. K. Hargreaves et al., J. Atmos. Solar–Terr. Phys. 59, 872, copyright 1997, with permission from Elsevier Science.)

more limited range of latitude (probably less than 200 km) than the onset itself, which in some cases may be tracked over a wide range of L values if some time differences are allowed. Figure 7.5 illustrates the spatial confinement of the spike event determined with an imaging riometer. At L5.9 the typical spike event is elliptical in shape, the major axis being generally east–west. Typical dimensions are 190 km by 80 km, and the axial ratio is about 2.5 (Hargreaves et al., 1997). The properties of spike events at the South Pole (L⬃13) have been found to be remarkably similar in form though they are generally smaller in magnitude. The spike event lasts for 1–2 min only, and is dynamic (Section 7.2.4). The main part of the night event is considerably more widespread than the spike. As an example, Figure 7.6 shows the distribution before, at, and after the intense peak (2132 UT) in the event of Figure 7.3(b). During the main part of the event (which peaked at about 9 dB at 38.2 MHz) the absorption covers the whole area at a substantial level, although spatial structure is also present. The properties of these events have not yet been worked out in full, but they are clearly distinct from those of spikes and arcs (see below).

7.2 Auroral radio absorption

(a)

343

175.0 S 100 km 125.0

W

75.0

25.0 km 2

– 25.0 4 6 – 75.0

6

8

2

4

2

4

2 2

– 125.0

– 175.0 – 175.0 – 125.0

– 75.0

– 25.0

25.0

75.0

125.0

175.0

km

(b)

175.0

0.5

125.0

0.5

1.0 1.5

1.0

75.0 0.5

1.5 1.0

0.5

25.0 km – 25.0

0.0

– 75.0 N E

– 125.0

– 175.0 – 175.0 – 125.0

– 75.0

– 25.0

25.0

75.0

125.0

175.0

km

Figure 7.5. Examples of spike events at (a) the South Pole, 22 July 1988 at 2042:50 UT, and (b) Kilpisjärvi, 14 November 1994 at 2015:10 UT showing typical dimensions. A height of 90 km is assumed. Contours are of absorption in decibels at 38.2 MHz, and the time resolution was 10 s. This South-Pole (L⬃13) event was an exceptionally intense one, but that at Kilpisjärvi (L5.9) is more typical for that latitude. ((a) J. K. Hargreaves et al., Radio Sci. 26, 925, 1991, copyright by the American Geophysical Union; (b) reprinted from J. K. Hargreaves et al., J. Atmos. Solar–Terr. Phys. 59, 872, copyright 1997, with permission from Elsevier Science.)

344

The high-latitude D region

✖

✖

✖

7.2 Auroral radio absorption

✖

Figure 7.6. Absorption distributions during the main part of a night event at Kilpisjärvi on 6 October 1994. The highest contours are 2.7, 8.2, and 3.1 dB at 2128:20, 2132:20, and 2146:30 UT, respectively. The frequency is 38.2 MHz. The maxima are marked x and the contours at half the maximum are dotted. These are 10-s averages. The event came into view from the north-west, peaked overhead, and then drifted westward.

Apparently distinct from the spike is another short-duration feature, which appears for only a short time because it rapidly crosses the field of view in a westward direction, namely the westward surge. It extends 75–85 km north–south but its east–west dimension is not known. It may be related to the westward-traveling surge in the luminous aurora (Section 6.4.2).

Daytime spike events The larger spike events are never observed in the day sector, but smaller ones have been observed by day at high latitude in the northern hemisphere (Stauning and Rosenberg, 1996). At Sondrestrom (invariant latitude 73.7º, L⬃13) they have durations of less than 5 min, with the most probable value 1–2 min. The distribution of magnitude (at 38 MHz) peaks at 0.2–0.3 dB, and most occur between 1200 and 1800 local magnetic time with the mode at 1500–1600. Their spatial extent is 50–100 km. On present evidence these events are distinct from the larger spikes typical of the night sector both at this site and at lower latitudes.

345

346

The high-latitude D region

7.2 Auroral radio absorption

Figure 7.7. An absorption arc observed at Kilpisjärvi (L5.9) on 11 April 1995, at 1745, 1817, and 1826 UT. Each picture is a 1-min average and the dotted lines show the absorption arc defined by absorption half that at the peak. There is marked spatial structure along the arc. Each picture is 240 km on the side, and a height of 90 km is assumed. The feature was visible for an hour, and it drifted equatorward at less than 10 m s1 initially but then more rapidly at 130 m s1. At 1845 UT it was followed by a sharp-onset event.

Preceeding bays Starting 1–1.5 h before the onset there may be observed a weak absorption event lasting 40–60 min (pre-onset absorption or the preceding bay), and there can be little doubt that this feature is in some way related to the main event which follows. Imaging riometers have identified the form of the preceeding bay as an arc, extended east–west but only 60–100 km wide north–south (Figure 7.7). The whole feature tends to be weak, and it contains embedded structure. The arc normally undergoes a slow equatorward drift, and, as we shall see (Section 7.2.4), there are cases in which a sharp-onset event appears to grow from it. It is possible that this absorption feature is connected with the luminous auroral arc.

Slowly varying events and pulsations Then, in the morning sector, between about 0600 and 1200 local magnetic time, there occur slowly varying events (SVAs). These last for an hour or two and are

347

The high-latitude D region

Absorption (dB)

348

3.0 (a) 2.5 Abisko 2.0 1.5 1.0 0.5 0 0200

0300

0400

0500

0600

Absorption (dB)

U.T. (b)

3

Andøya 2 1 0 0200

0300

0400

0500

0600

0700

U.T.

Figure 7.8. Events in the morning sector: (a) a slowly varying event at Abisko (L5.6), 23 March 1985 and (b) an event with pulsations at Andøya (L6.2), 23 August 1985 (Reprinted from J. K. Hargreaves and T. Devlin. J. Atmos. Terr. Phys. 52, 193, copyright 1990, with permission from Elsevier Science.)

smooth, with little structure (Figure 7.8(a)). Some of these are modulated with quasi-periodic pulsations having a period of several minutes (Figure 7.8b), in the range of Pc4 and Pc5 (Section 2.5.6). The SVA is spatially more widespread than the spike and the preceeding bay.

Relativistic electron-precipitation events As long ago as 1965 it was realized that some of the absorption events affecting radio circuits (particularly in VHF forward-scatter communications) were due to electrons of unusually high energy and relativistic speed (Bailey, 1968). Relativistic electron precipitation (REP) is a daytime phenomenon, and more events are observed at the equinoxes than at the solstices. The events may be intense, and they are geographically widespread according to the reports of the 1960s. Since a riometer does not determine the height of the absorption, it is not immediately apparent which of the events detected are in the REP category, but simultaneous incoherent-scatter measurements have shown that, in some cases, the absorption was indeed at unusually low altitude (Collis et al., 1996), and these are almost certainly due to relativistic electrons. (An electron of energy 100 keV travels at just over half the speed of light and one of energy 500 keV travels at 0.86c. Electrons more energetic than 250 keV penetrate below 67 km and produce maximum ionization at heights below 75 km – Figure 2.26.) In at least some of these cases the event is confined to a small area in the D region, less than 100 km north–south though more extended east–west. Figure 7.9 is an example. In this

7.2 Auroral radio absorption

(a)

349

175

125 0.50 75

0. 75

0.50

R

km

25

0.50

–25

–75 0.5 0

–125 –175 –175

–75

25

125

km

(c)

110

110

100

100 Altitude (km)

Altitude (km)

(b)

120

90

80

70

90

80

70

60 0

0.5

1.0

1.5

60

Electron density (1011 m –3 ) 50 0

0.05

0.10

0.15

Absorption (dB km –1)

Figure 7.9. Properties of a co-rotating daytime event observed at 1204 UT on 1 March 1995 by an imaging riometer and the EISCAT radar simultaneously. (a) The spatial structure, assuming an altitude of 90 km. The contours are in decibels and the time resolution is 2 min. The radar beam intersected the event at R, somewhat away from the maximum of 3.5 dB. (b) The vertical profile of electron density, at 1-min resolution. (c) The vertical profile of the incremental absorption computed from (b). The electron-density peak below 90 km and the absorption peak below 70 km identify this as an event due to electron precipitation of unusually high energy. (P. N. Collis et al., Ann. Geophysicae 14, 1305, 1996, copyright notice of Springer-Verlag.)

The high-latitude D region

350

(a)

(b)

12 14

° 50

16 INV AR

AT IT

GE OM AG NE T

° 80 ° 90

18

10

° 50

16

1.0 0.5

° 70 UD E

14

08

1.5

° 60 IAN TL

12

10

4 ° 60

IC

° 70 LA TIT

UD E

0

° 80 ° 90

06 18

0.5 2.

08

8 4

06

0.5

1. 1.0 5

4 04

20

22

20

02

00 GEOMAGNETIC TIME

04

4

22

02

00 GEOMAGNETIC TIME

Figure 7.10. (a) The median intensity of AA events in decibels at 30 MHz. (Reprinted from J. K. Hargreaves and F. C. Cowley, Planet. Space Sci. 15, 1571, copyright 1967, with permission from Elsevier Science.) (b) the percentage occurrence of 30-MHz absorption exceeding 1 dB. (After T. R. Hartz et al., Can. J. Phys. 41, 581, 1963.) The diagrams differ because the night events are shorter than the day events.

event the electron density due to electron precipitation peaked below 90 km, and the absorption peaked at 67 km. These are unusually low altitudes for an auroral absorption event (see Section 7.2.6). It is interesting to note the similarity in the size and shape of some of these types of event, even when they are seen in different circumstances. This suggests that they have a common underlying physical cause. 7.2.3

General statistics in space and time Latitude and longitude distributions

Two versions of the overall global occurrence of AA are shown in Figure 7.10 with respect to magnetic latitude and time. Most obvious in the second diagram is the morning peak around 0600–1000 magnetic time, where 1 dB (at 30 MHz) is exceeded for 8% of the time. This does not mean that AA is only a daytime phenomenon, however. We have already described some night-time events, and there is just as much absorption activity in the night as there is in the day sector, a fact that shows up more clearly in the first diagram which plots the median intensity of those events which happen to peak at a given time and latitude. The daytime events dominate in the other kind of statistics because they tend to be of longer duration. However, the night events can be just as intense. There is a deep minimum in the pattern of occurrence at around 1600–1700 magnetic time.

7.2 Auroral radio absorption

351

These distributions reveal a zonal phenomenon, having a maximum at about 67° geomagnetic latitude (corresponding to L7), though the details vary somewhat. Hartz et al. (1963) found the maximum at 67° in Canada, but Holt et al. (1961) found it at 62° in Norway. Using data from Canada and conjugate stations in the Antarctic, Hargreaves and Cowley (1967a) found small daily variations in the latitude of the maximum, in particular a decrease of 2° or 3° over the few hours up to about midnight, a recovery to 67°–68° during the morning, and a further decrease after noon. The latitudinal distribution of absorption may be approximated by a Gaussian curve,

冢

Am A0 exp 

(   0 ) 2 2 2

冣

(7.2)

where Am is the median absorption,  the invariant latitude, and the half width (or “standard deviation”) of the absorption zone. The half width is several degrees: for example 4.5° (Hartz et al., 1963) or 3.7° (Holt et al., 1961). Despite its daily variation, it is clear that the absorption zone is not the same as the auroral oval defined by the occurrence of luminous aurorae, but corresponds to the more circular zone – the “outer zone” – discussed in Section 6.3.5. The luminous oval and the absorption zone coincide (or at least are very close together) near midnight, but the absorption zone is more circular than the oval and lies at lower latitude on the day side. It is instructive to compare the incidence of AA in Figure 7.10 with the distribution of energetic electron precipitation observed from satellites in Figure 6.6.

The spatial extent The horizontal extent of individual absorption events in kilometers is obviously an important matter practically as well as scientifically. Should they be very small, their effect on HF propagation could be reduced by space-diversity reception. If, on the other hand, they should blanket very large areas, it is difficult to see what could be done. (The same is the case with the polar-cap absorption due to protons – Section 7.3). Various measurements using groups of wide-beam riometers (Table 7.1) do not agree with each other completely, but the general indication is that the events cover a few hundred kilometers. Some reports suggest that there is a degree of elongation in the east–west direction, but other studies have come out in favor of more or less circular patches of absorption. The results of Table 7.1 come from the earlier period of observations, and we must remember that the broad beam of the antenna prevents structure smaller than about 100 km being detected. As pointed out in Section 7.2.2, work using narrower beams is finding events narrower than 100 km. Spike events, which are only some tens of kilometers across, tend to appear in isolation and their magnitude is considerably underestimated in broad-beam measurements. However, since they occur for such a short time, they will not have much effect on the general

L

5.5 5.5–9

6

4

4–8

5.5

4

5.5

5.5

Author

Little and Leinbach (1958)

Holt et al. (1961)

Kavadas (1961)

Jelly et al. (1961)

Leinbach and Basler (1963)

Little et al. (1965)

Parthasarathy and Berkey (1965)

Ansari (1965)

Slowly varying events, about 60 examples

Sudden-onset events

Daytime events, 54 days, December–February

49 days, January–March

?

?

12 selected periods

3 days, summer 1 month, March

Data

Peak absorption

Peak absorption

Circular

Shape of correlation pattern


0.70 at 250 km N–S
0.74 at 800 km E–W

Fair agreement over 350 km N–S


0.26 at 250 km N–S (90 events)
0.41 at 800 km E–W (35 events)

Elliptical

Circular

Elliptical

Can be similar over 380 km or different over 35 km, N–S

High correlation over 10 km N–S


0.5 at 380 km

Regions at least 200 km N–S and 90 km E–W Day:
0.57 at 800 km N–S Night:
0.43 at 800 km N–S

Results

2-min values if absorption
0.5 at 650 km N–S and
700 km E–W &0.3 dB

Hourly values

—

?

?

Hourly values Hourly values

Values correlated

Table 7.1. Spatial properties of Auroral Absorption (Hargreaves, 1969)

4

5–7

7

Ecklund and Hargreaves (1968)

Bewersdorff et al. (1968)

Hargreaves and Ecklund (1968)

12 months

Slowly varying events, four examples

17 months, August–January

Disturbed nights

Night:
0.5 at 160 km Day:
0.5 at 250 km

Hourly values if &0.3 dB

a

Circular Circular

Day:
0.5 at 365 km E–W and 170 or 300 kma N–S Little variation over 300–400 km E–W

Elliptical or circular

Night:
0.5 at 750 km E–W and 155 or 465 kma N–S

Hourly values if &0.3 dB

–

Elliptical


0.9 at 20 km N–S

? &0.5 dB

Note: Measurements poleward and equatorward of the central station, respectively.

5.5

Berkey (1968)

The high-latitude D region

354

statistics. At other times, particularly during the longer-lived activity, whether by night or by day, the smaller structures are only a component of the total distribution, and the results of Table 7.1 still have significance in showing that, even when finer structure is present, at least some part of the absorption extends for 200–300 km.

Durations It is not as easy as one might imagine to determine the durations of individual absorption events. Some events are isolated and thus easily recognized, whereas others run into each other and it might not be obvious whether any such case should be described as one long event or several short ones. Furthermore, many events fade away rather gradually, so the end is not always clear. (Onsets tend to be sharper.) Some values and the variation of duration from event to event are indicated by Figure 7.11, which shows the relative distribution of duration in the day and night groups at one station near the occurrence maximum and one just equatorward of it. Note that the durations are shorter at the higher L value, and are shorter by night than they are by day at both sites (Table 7.2). The groups of day and night events are specifed in UT. Add 2.5 and 0.5 h to get local magnetic time at Kiruna and Siglufjordür, respectively. 7.2.4

Dynamics

The dynamic nature of AA events is a property that is often not appreciated. The movements have been investigated using chains or groups of geographically separated riometer stations, supplemented more recently by imaging riometers. The results reveal a good degree of consistency in the movement of events of a given type, implying that the motion has some physical significance – even if we are not yet sure what that significance is! Some examples are given below.

The onset and main event in the night sector The sharp-onset event occurs in the pre-midnight sector. As pointed out above, it may but need not include a spike, and it may appear as a more gradual onset at some latitudes. The onset of this type of event usually appears first at an L value between 5 and 6, which is somewhat equatorward of the statistical absorption maximum (at L ⬃7). From that latitude it spreads both poleward and equatorward. The poleward section is the more often observed; the velocity is between 0.5 and 3 km s1 in most cases. There is a clear demarkation between the poleward and the equatorward sections, as may be seen in Figure 7.12(b). However, this is not the case for absorption peaks subsequent to the onset (Figure 7.12(a)): they can move in either direction, rather more than half moving equatorward. In the example of Figure 7.6 the event arrives over Kilpisjärvi

Q

6 9 Time (h)

12

30

20

10

Q

Morning (02–09 UT ) 104 events

30

20

10

0

3

20

0

M

Q

Night (16–23 UT ) 138 events

30

10

Q

M

Kiruna

40

10

20

30

40

Q

Q M

M

Q

Q

3

9

Morning (04 –11 UT ) 122 events

6

Night (18–24 UT ) 92 events

Siglufjordür

Figure 7.11. Durations of events starting in the night and morning sectors at Kiruna (L5.4) and Siglufjordür (L6.9).

Occurrence (%)

12

M: Median Q: Quartile

Time (h)

356

The high-latitude D region

Figure 7.12. Relative frequencies of poleward and equatorward movements along a meridian through Alaska: (a) peaks, and (b) onsets. At all latitudes most of the peaks move equatorward, whereas the onset tends to move equatorward at L 5 but poleward at L 6. (Reprinted from J. K. Hargreaves. Planet. Space Sci. 22, 1427, copyright 1974, with permission from Elsevier Science.)

Table 7.2. Medians and quartiles of the distributions of durations of events at Kiruna (L5.4) and Siglufjordür (L  6.9) (durations are given in hours) Kiruna

Lower quartile Median Upper quartile

Siglufjordür

Day

Night

Day

Night

1.5 2.2 3.5

0.7 1.5 2.9

0.8 1.5 2.9

0.3 0.9 1.3

7.2 Auroral radio absorption

Figure 7.13. Maximum absorption, and distances in the X (west–east) and Y (south–north) directions of the location of the maximum from overhead at Kilpisjärvi, for the spike event at 2046–2051 UT on 6 October 1994 (see Figure 7.3). The time resolution is 1 s. Despite the time structure revealed at this resolution and erratic movement west–east, the poleward progression is remarkably persistent. These features are typical of night-time spike events.

357

The high-latitude D region

358

120

2050:13 UT 80

40

0 2047:45 UT

S

km

N

2049:13 UT

–40

2047:13 UT –80

2047:06 UT –120 –120

–80

–40

0 W

km

40

80

120

E

Figure 7.14. The spike event of Figure 7.13 at five selected times. The absorption was maximum at the points marked by black circles, and the shaded areas show where the absorption was greater than half the maximum. Note the tendency towards an elliptical shape.

(L5.9) from the north and peaks almost overhead, but then moves off to the west. It is not clear whether this is typical, but previous observations (Hargreaves, 1970) using wide-beam riometers over a 250-km baseline at L7 have revealed a westward component in the night sector up to about 2 h after midnight. When an event begins with a spike, this invariably moves poleward, as in the example from the Kilpisjärvi imaging riometer in Figure 7.13. This details the spike event at 2046–2051 UT in Figure 7.3(b), and shows the value of the maximum absorption and its location within the field of view, all at 1-s resolution. Note that the magnitude of the absorption varies with quasi-period 30–60 s. East–west motions are rather irregular, but there is a poleward progression overall. This is typical of spikes occurring at the beginning of a night event. Figure

7.2 Auroral radio absorption

7.14 shows the position of the absorption patch at five selected times (which are also marked on Figure 7.13). The maximum moved by 200 km in just over 3 min, an average speed of 1 km s1, though the speed was greater to begin with. The dimensions of the patch are changing, but the tendency towards east–west extension is maintained.

Motions on the global scale The onset also propagates eastward and westward from its first appearence in the night sector. There is some variation among published results, partly due, no doubt, to actual variability in the movements from one instance to another, and perhaps also to observational selection. For stations near the centre of the absorption zone, observations between stations separated by some thousands of kilometers (for example over 90° longitude) indicate median speeds of about 4° longitude min1 (or 2.8 km s1), eastward between midnight and 1400 LT, westward otherwise (Hargreaves, 1967; Pudovkin et al., 1968; Jelly, 1970). The figure of 4° min1 is for specific features recognized at both the stations. If absorption events are compared without consideration of form, the median speed comes out smaller by a factor of three. Hajkowicz (1990) found westward speeds of 2.7–4.5 km s1 for pre-midnight sudden onsets at L values of 5.2–6.1. A simple model of the longitudinal movements at L⬃7 is given in Figure 7.15. From its first appearance near midnight it takes (on average) about 20 min for an onset to travel 5 h of local time and 30–40 min to reach the morning sector. The eastward and westward sections are not alternatives; observations have verified that they occur simultaneously. It is generally thought that the energetic electrons which are precipitated by day actually originated in the night sector and then drifted eastward as particles trapped in the geomagnetic field (Section 2.3.4 and Figure 2.14). This cannot explain the westward motion before midnight, which is presumably governed by other factors in the magnetospheric tail. Combining results from studies of latitudinal and longitudinal movements gives the overall global picture of Figure 7.16. This makes no attempt to distinguish among different types of event. The most comprehensive investigation of absorption movements on the global scale was performed by Berkey et al. (1974), who analyzed 60 substorm events at 40 riometer stations. Some of the main points, which confirmed the earlier work and added some new results, are as follows. (a)

The activity most frequently begins near midnight.

(b)

The onset is earlier and at lower latitude when the level of magnetic disturbance is greater.

(c)

The longitudinal velocity is in the range 0.7–7 km s1.

(d)

The westward part of the expansion (usually seen before midnight) sometimes follows the auroral zone (i.e. the outer zone) and sometimes follows the auroral oval.

359

360

The high-latitude D region

Figure 7.15. Longitudinal time delays. (J. K. Hargreaves. Proc. Inst. Electr. Electron. Engineers 57, 1348, 1969a, © 1969 IEEE.) (a) A simple model of delay with respect to an onset at midnight. (b) The delay over 5 h of local time, compared with observations from Hargreaves (1967), the basis of the model. (c) Time delays between events at College and Murmansk (Pudovkin et al., 1968) compared with predictions from the model.

7.2 Auroral radio absorption

Figure 7.16. The progression of the onset of absorption projected onto the equatorial plane (assuming a dipole field). The wavefronts are drawn at 10-min intervals. (Reprinted from J. K. Hargreaves. J. Atmos. Terr. Phys. 30, 1461, copyright 1968, with permission from Elsevier Science.)

(e)

The speed of the westward expansion is about 1 km s1 when it expands along the oval and about twice that when it expands along the zone.

(f)

There is much variability among individual substorms.

(g)

Some maps show morning activity (pre-noon) following a midnight onset, but with little or no activity between the midnight and day regions.

The drift of the pre-onset bay The weak absorption bay that may precede an onset moves equatorward at a typical speed of a few hundred m s1. (See also Figure 7.27 later.) Many of these are so weak that they may be detected only by the practiced eye, but Figure 7.17 shows one that was unusually strong. That bay was clearly seen in the sectors of Finland, Sweden/Norway, and

361

362

The high-latitude D region A (dB) DIXON ISLAND

NORILSK

1 3 1 3

KEVO

1 3

IVALO

1 4 2

SODANKYLÄ

4 2 ROVANIEMI

3 OULU JYVÄSKYLÄ BJÖRNÖYA TROMSØ ANDØYA

ABISKO

1 1 1 1 1 3 1 3

THORSHAVN GODTHÅB

1 1

NARSSARSSUAQ 1 SIGLUFJORDÜR LEIVORGUR

FAGURHOLSMYRI

1 1 3 1 17

18

19

UT

Figure 7.17. Equatorward motion of a bay preceeding an onset in the European sector on 4 May 1977. The diagram includes chains of stations at various longitudes, and the motion is clearly seen in the data from Finland, Norway/Sweden and Iceland. The related onset which followed is marked by arrows. (Reprinted from H. Ranta et al., Planet. Space Sci. 29, 1287, copyright 1981, with permission from Elsevier Science.)

Iceland. It turns out that the imaging riometer is able to observe these moving arcs in greater detail (as in Figure 7.7). The speed is not always uniform, and the arc may fade and strengthen during its passage across the field of view. It is tempting to relate the movement to that of auroral arcs (Section 6.4.2), which is also equatorward in most cases.

7.2 Auroral radio absorption

363

Figure 7.18. The connection between preceeding bay and sharp onset idealized as a “reversedy” event. (Reprinted from J. K. Hargreaves et al. Planet. Space Sci. 23, 905, copyright 1975, with permission from Elsevier Science.)

The relation between the bay and the onset Ranta et al. (1981) have studied the incidence of these bays in relation to the sharponset events which follow. Most of the bays occur between L values of 4 and 9, and individual examples cover between one and five or six L units. They have not been reported from the South Pole (at L13). In longitude they can extend more than 90°. The onset may be seen over a larger range of L, from 4 to 16 or more, and individuals have been observed to cover ten units of L. It can exceed 150° in longitude. The onset often appears first at or near the eastern end of the preceeding bay, which means that, statistically, the bays occur earlier in the day (in the afternoon and evening sectors) than do the sharp onsets whose preference is for the hours just before and up to midnight. The event following the bay of Figure 7.17 was observed in the sectors of Finland and Sweden/Norway, and exhibited poleward motion. The relationship between bay and sharp-onset event may be summarized as the “reversed-y” event of Figure 7.18. Figure 7.19 illustrates what appear to be the typical dynamics of a night-time event at L5.9, somewhat equatorward of the zone maximum. The record is from a wide-beam 38.2-MHz riometer but the movements have been identified using an imaging riometer at the same site. The spike had a rapid poleward movement, but an arc preceeding it and patches following drifted equatorward. The main event, which was widespread, came into view from the poleward side, but then drifted out of view to the west.

1938

Nothing

gap

Patches drifting towards equator

1748–1805 Main event from north

1830

1736–39 Spike

Arc moving equatorward

1703

2

Level (dB)

0

2 4 6 16

17

S

18

19

20

Time (UT)

Figure 7.19. Night activity on 30 January 1995 observed with a wide-beam 38.2-MHz riometer at Kilpisjärvi, noting the main features and their movement. Excepting the spike, the dominant movement was equatorward.

The slowly varying event The slowly varying event in the morning sector typically exhibits an eastward motion when it is observed by riometers 250 km apart (Hargreaves, 1970; Hargreaves and Berry, 1976). Each of these studies gave a median eastward speed just below 40 km min1 (about 620 m s1), but with a large variation in individual cases. (Half the eastward speeds were between 20 and 80 km min1.) It will be noticed that these speeds are considerably below those determined from widely spaced observations.

Co-rotation A tendency towards co-rotation has been noted in some events of the morning and dayside (Hargreaves et al., 1994; Collis et al., 1996). In particular, it appears that the spatially restricted, very energetic type of event described in Section 7.2.2. can remain virtually fixed with respect to the rotating Earth for a long period. The event shown in Figure 7.9 remained in the field of view for more than 1.5 h. According to radar measurements, the meridional F-region ion drift during the event varied from 100 m s1 westward to approximately zero – in agreement with the motion of the absorption event. This clear example, taken with the evidence

7.2 Auroral radio absorption

from the previous paragraph, indicates that more than trapped-particle drift is involved in the longitudinal motion of auroral absorption. 7.2.5

The relation to geophysical activity, and predictions of auroral absorption A relation to Ap

Auroral absorption (AA) is more frequent and more intense when geomagnetic activity is high. One might also expect that AA would be stronger and occur more often at times of high sunspot number, but that is not necessarily the case. In some solar cycles the magnetic activity does not rise and fall in the same way as the number of sunspots, and in such a case the AA is seen to go with the magnetic activity rather than with the sunspot number. We shall return to this point. In the shorter term, it is possible to show relations between the absorption on single days and the Ap index (Section 2.5.4). For example, the probability of there being at least one event of at least 1 dB during a period of 24 h rises almost linearly with Ap, becoming virtually unity at Ap 15 for a station at L5.6 (Figure 7.20(b)). The rate of occurrence is smaller at higher latitude (Figure 7.20(a)) but still increases approximately linearly with Ap. The average number of events per day also increases with Ap (Figure 7.20(c)), rising from one to three over the Ap range 8–25. One observation that this association explains is the tendency for AA to be intense for several days at a time, often then followed by a week or more when it is very low. The pattern tends to repeat from one month to the next. This behavior just mirrors that of magnetic disturbance, and is due to the rotation of the Sun which carries active regions out of view after a few days and tends to bring them round again a month later. The latitude of the absorption zone (Section 7.2.3) also shifts with the intensity of magnetic activity (Hargreaves, 1966). During that period of the day when AA is most significant, the latitude of the maximum (0) decreases from approximately 70° to 66° as Kp increases from 0 to 5, and at values of 6 or 7 it may be as low as 60°. At the same time the half width of the absorption zone ( ) increases somewhat (from about 4° to 5.5°), with even greater broadening at the largest values of Kp.

A relation to HF radio propagation This brings us to the topic of predictions. First, though, to get some idea of the importance of AA in HF propagation, we must consider the magnitude of the absorption involved. The intensity of the absorption, which is usually measured in decibels, depends inversely on the square of the radio frequency. In a

365

The high-latitude D region

Probability of at least one event (%)

100

(a)

80

60

40

20

0 0

10

20

30

40

Ap

Probability of at least one event (%)

100

(b)

80

60

40

20

0 0

10

20

30

20

30

Ap 5

Number of events per day

366

(c)

4

3

2

1

0 0

10 Ap

0

1

2

3

4

5

Kp

Figure 7.20. Effects of magnetic activity on the incidence of AA events. (a) The probability that at least one event of at least 0.3 dB at 51.4 MHz occurs at the South Pole (L13) within 24 h. This level is equivalent to 0.88 dB at 30 MHz. The data covered the three years 1990–1992 inclusive. (b) Similarly for 1-dB events at 30 MHz at Abisko (L5.6). The data covered the two years 1976 and 1977. (c) The average number of 30-MHz, 1-dB events per day at Abisko. (South Pole data from T. J. Rosenberg, private communication, and Abisko data compiled from Hargreaves et al., Report UAG-84, 1982.)

7.2 Auroral radio absorption

radio-communication circuit it also depends on the geometry (specifically, the angle at which the ray passes through the D region). Nonetheless, in round numbers, if a 30-MHz zenithal riometer detects 1 dB absorption, an oblique HF path will suffer about 20 dB (Agy, 1970). Thus, absorption greater than 1 dB on a 30-MHz riometer is likely to be of practical significance to HF propagation (especially between 3 and 15 MHz), and the statistics regarding the occurrence of absorption are often presented in terms of the 1-dB level. At some latitudes there are many days per month when this level is reached or exceeded at least once.

AA predictions Clearly, predictions of AA are going to be statistical in nature because the phenomenon is essentially sporadic; but at least there is a good base for the statistics because large quantities of riometer data are available. If the predictions are required for radio propagation, the task has two parts: first, to specify from existing riometer data the statistics regarding the occurrence of absorption and the effects of independent variables such as latitude, season, time of day, and solar and magnetic activity; and second, from propagation experiments to observe how the events detected by the riometer are related to circuit effects. Here we consider the matter of absorption statistics, in which there have been some useful developments. Relations to communications circuits are considered in Sections 8.2 and 8.4. The representation of absorption statistics can be taken in two stages. First, having decided our significant absorption level – for instance, 1 dB at 30 MHz – we can then inspect the data from various riometer stations and count up the probability of 1 dB being exceeded as a function of the various external parameters. The probability that A dB will be exceeded is generally called Q(A).

Calculation of Q(1) Foppiano and Bradley (1985) published a formula (Table 7.3) for calculating Q(1), based on an extensive study involving many sources and taking in data from several longitude sectors and different years. The formula is written as the sum of day and night contributions, each comprising the product of terms for the variation with magnetic latitude, time of day, solar activity, longitude, and season. The latitude variations are of Gaussian form (similar to Equation (7.2)) with the night peak at 67° and the day peak at 68° at low sunspot number. The time-of-day terms are also Gaussian, the night activity peaking at midnight and the day activity at 1000 local time – compare with Figure 7.10. The dependence on solar activity is expressed in a table, and there are empirical formulae for the dependences on longitude and season. Some of these terms are better established than others. Some seasonal variation probably occurs (see Section 7.2.6), but the question of a longitudinal effect has not been investigated thoroughly. However, the greatest problem with the formula of Table 7.3 is in the assumed dependence on solar activity expressed by

367

368

The high-latitude D region

Table 7.3. The Foppiano–Bradley formula for Q(1) Total Day component Night component

Q(1) Q1d Q1s Q1d Kd ddT dRddM Q1s KsssT sRssM

 magnetic latitude; Tlocal time in hours; R sunspot number;  geomagnetic longitude; Mmonth; Kd and Ks are constants Latitude terms d exp[(  m)2/(2 2)] s exp[(  m)2/(2 2)] m and m geomagnetic latitudes of maxima for day and night components: m 68(10.0004R) for R %100; m 65.28 for R&100; m 67(10.0006R) 0.3(10.012R) | t | where t (T 3) for 0 %T%15; t (T 27) for 15T24  and  widths of latitude distributions for day and night components:    3(1 0.004R) for R %100;    4.2 for R &100. Time-of-day terms sT exp[(T Tm)2/2 T2] dT exp[(TTm)2/(2 2T)] Tm and T m local times of maxima for day and night components: T m 0 Tm 10(10.002R); T and T widths of time distributions for day and night components: T  T 2.8 Solar-activity terms dR sR (1 aR), the values of a being from the following table: T (h) 00 02 04 06 08 10 12 14 16 18 20 22 a 0.0032 0.0025 0.0141 0.0048 0.0149 0.0146 0.0142 0.0090 0.0037 0.0156 0.0206 0.0092 Longitude terms d s0.580.42 sin[0.947( 85)] 0.16 0.580.42 sin[1.80( 130)] 0.580.42 sin[0.947( 275)] where  corrected geomagnetic longitude in °E

for 0°%  10° for 10°%  80° for 80°%  180° for 180°%  360°

Seasonal terms sm 1 dM 1 0.3 sin(3.86); where  solar declination angle in degrees (positive in summer, negative in winter). Constants Kd 21;

Ks 12

These give Q(1) values in percent Notes: The use of d and s for the day and night components derives from Hartz and Brice’s (1967) “drizzle” and “splash” terminology.

7.2 Auroral radio absorption

369

the sunspot number. In a subsequent study of absorption over Finland during a whole solar cycle (1972–1983), Hargreaves et al. (1987) concluded that the sunspot term of the formula was not very accurate, and they proposed an alternative based on the monthly mean value of Ap (Ap): Q(1)(Ap 30cos2 )exp[( 65)2/25].

(7.3)

This formula gives the average Q(1) over all times of day. The significance of Equation (7.3) is that, over the long term, the absorption probability is proportional to the mean Ap (Ap) above a threshold. The results may also be represented by a Gaussian variation with latitude in which the peak value (Q0), the latitude of the peak (0), and its width ( ) depend on Ap as in Table 7.4. Note in particular that, with increasing Ap, the maximum probability increases linearly and the position of the maximum moves equatorward. (The foregoing analysis is based on observations covering only the equatorward side of the absorption zone.)

The log-normal distribution The second stage is to consider the form of Q(A). That is, if we can predict Q(1), can we say what Q(2) or Q(0.5) will be? This means knowing the probability distribution for the occurrence of absorption. Foppiano and Bradley (1984) assumed a log-normal distribution for the occurrence of absorption; that is, that the logarithm of the absorption follows a normal distribution: f (logA)d(logA)

冢

冣

1 (log A  log Am ) 2 exp  d(logA), 兹2 2 2

(7.4)

where A is the absorption in decibels, Am is the median absorption (log Am being the mean of log A), and is the standard deviation of logA. The probabilility of A being exceeded is then Q(A)

冮

$

f (logA)d(log A).

(7.5)

log A

Table 7.4. Parameters of Gaussian curves fitted to latitudinal variations of Q(1) in the Finland sector Ap

0 (degrees)

(degrees)

Q0 (%)

0–10 10–15 15–20 20

68.1 67.8 66.9 65.6

3.8 3.9 3.6 3.6

5.7 9.3 13.3 17.4

The high-latitude D region

99.99

99.90

99.00

% Probability of absorption > A dB

370

90.00

50.00

10.00

1.00

0.10

0.01 0.01

0.02

0.05

0.1

0.2

0.5

1.0

2.0

5.0

10.0

30 MHz Absorption, A (dB)

Figure 7.21. The log-normal distribution of Q(A), South Pole, March 1982. (Data from T. J. Rosenberg, private communication.)

The cumulative distribution Q(A) should come out as a straight line on log-probability paper. In most cases this appears to be so (Figure 7.21), at least if the range of absorption is restricted to values between a few tenths of a decibel and a few decibels – that is, to the range for which riometer data are accurate and most plentiful. (Whether very large or very small values obey the same distribution is not really known.) The log-normal distribution is described by just two parameters, one of which can be Q(1). There is something to be said, also, for using Am and , the median and the standard deviation (the second giving the slope on the log-probability plot), since both of these appear explicitly in the formula. It is not wise to extrap-

7.2 Auroral radio absorption

371

olate the log-normal law below 0.1 or 0.2 dB. In most sets of riometer data these small values are much affected by any error in the quiet-day curve, and also there are theoretical reasons why the log-normal form cannot continue indefinately towards ever smaller values.

Predicting events The foregoing approach aims to predict the likehood of a certain level of absorption at a given site if the level of geomagnetic disturbance (or solar activity) is known – the latter, of course, also being a quantity requiring prediction. No account is taken of the event aspect: the fact that the absorption occurs in bursts; that, once started, it is likely to continue for some time but that there are also long periods with no significant absorption at all. Relatively little appears to have been done on prediction of AA using an event approach, though elements are implicit in some of the foregoing account. A comprehensive event description would specify the magnitude, duration, structure, etc., of which (apart from magnitude) the statistical approach takes no account. An event description would also include an element of short-term forecasting. One example, based on the data of Berkey et al. (1974), is shown in Table 7.5. Medians and deciles are given for the absorption at various local times for every 15 min after the onset of a substorm (Elkins, 1972). These are interesting for showing how the distribution of absorption develops in a substorm as a function of the local time, and also for what seems to be the first use of the log-normal distribution to describe absorption statistics. The actual magnitudes depend, of course, on the original selection of substorms. In this set the maximum absorption was found to be related to the AE index by the empirical formula (Absorption)max ⬃0.008(AE)max,

(7.6)

but this is not of much help in a prediction because AE is not a predicted quantity. It is better to relate absorption to the daily index Ap, predictions of which are published a month in advance. 7.2.6

The wider geophysical significance of auroral absorption events

The immediate implication of auroral radio absorption for high-latitude propagation is simply the resulting loss of signal. However, since we know that the absorption is due to additional ionization in the lower ionosphere, which in turn is produced by energetic electrons entering the atmosphere from above, these events clearly have deeper implications. In this section we review some contributions of riometer studies to geophysical topics.

The high-latitude D region

372

Table 7.5. Medians and deciles of absorption at various local times during a substorm (Elkins, 1972) Substorm time

LT

Median (dB)

Upper decile (dB)

Lower decile (dB)

T15

00 03 06 09 12 15 18 21 00 03 06 09 12 15 18 21 00 03 06 09 12 15 18 21 00 03 06 09 12 15 18 21

0.75 0.70 0.54 0.28 0.10 (0.14) 0.10 0.37 0.94 1.1 1.1 0.64 0.42 (0.20) (0.17) 0.50 1.1 1.3 1.6 1.4 0.67 0.28 (0.20) 0.44 1.0 1.3 1.6 1.6 1.1 0.38 (0.25) 0.50

2.6 2.2 1.8 1.25 0.56 (0.38) 0.56 1.5 3.7 4.0 4.5 2.8 1.7 (0.78) (0.58) 1.9 3.5 4.0 4.5 6.0 2.8 1.3 (0.84) 1.8 3.2 3.8 4.5 4.5 3.6 1.5 (1.0) 2.0

0.22 0.22 1.17 0.066 0.019 (0.052) 0.019 0.090 0.24 0.32 0.30 0.14 0.10 (0.052) (0.050) 0.13 0.34 0.43 0.54 0.32 0.16 0.064 (0.049) 0.11 0.30 0.40 0.56 0.56 0.35 0.096 (0.063) 0.070

T30

T45

T60

Notes: 1. Parentheses ( ) indicate values with large uncertainties due to the small sizes of statistical samples. 2. Time is to be interpreted as follows: local time “00” means the hours 0000–0259 and so forth.

7.2 Auroral radio absorption

Electron-density profiles That there are, indeed, relationships between total absorption and electron density at various heights has been shown by direct measurements. Friedrich and Torkar (1983) collected electron-density data from rockets flown into absorption events and thus “calibrated” the riometer in terms of the electron-density profile from 70 to 110 km. This has recently been extended (Friedrich and Kirkwood, 2000) using electron densities from the EISCAT incoherent-scatter radar in Scandinavia (Figure 7.22). This comparison provides an estimate of the electron density at given height for a given intensity of AA, though with considerable scatter about the average. Some 50% of values lie within a factor of two of the average plotted. Some of the scatter is no doubt due to real changes in the profile during events, and from one event to another. Figure 7.23 illustrates the changing electrondensity profile during a morning event of the slowly varying type in which the height of the peak lifts as the event decays. (The growth was more complicated.)

Absorption profiles Since the electron–neutral species collision frequency is known (for Equation (3.95)), the absorption profile may be computed from the electron-density profile. In most cases the computed and observed absorptions agree well enough to serve as confirmation that the reductions in signal recorded by the riometer are indeed due to non-deviative absorption. The height of the absorbing layer and its thickness are of direct interest in HF propagation. The heights of absorption maxima computed from rocket profiles of electron density range over 90 to 95 km at night, but may be lower (75 km) by day (Hargreaves, 1969a). The calculated absorption peaked between 88 and 95 km in the event of Figure 7.23 which occurred during the early morning, and at 67 km in the hard, daytime event of Figure 7.9. The absorbing layer is quite thick: generally 15–20 km between points where the incremental absorption is half the maximum. About 80% of the total absorption is produced in this slab. The specific absorption coefficient increases downward, and the absorption peaks some 5–15 km (depending on the spectrum) lower than the electron density; as a rough guide it would be true to say that the absorption layer occurs in the underside of the electron-density layer, starting just below the peak.

Incoming electron fluxes Since the AA event is due to precipitating electrons, the calibration can in principle extend back a stage further so that we may infer something about the intensity and spectra of the electrons precipitated during AA events. Proceedures for inverting the electron-density profile (Kirkwood, 1988; Hargreaves and Devlin, 1990; Osepian et al., 1993) involve routines that give the rate of production from an assumed incoming electron spectrum (Section 2.6.1 and Figure 2.26), and by some means adjust the spectrum until the computed electron-density profile

373

The high-latitude D region

(a) 150 140 130

Altitude (km)

120 110 100 90 80 70 60 108

109

1010 Electron Density

(b)

1011

1012

(m3)

150 140 130 120

Altitude (km)

374

110 100 90 80 70 60 108

109

1010

1011

1012

Electron Density (m3)

Figure 7.22. A riometer “calibration” against electron-density profiles measured by rockets and incoherent-scatter radar. (a) with the Sun below the horizon and (b) solar zenith angles 90° (dashed line) and 60° (solid line). In each case the curves are given for every 0.5 dB from 0 to 2.5 dB at 27.6 MHz. All seasons and times of day are included. (M. Friedrich and S. Kirkwood, Advances in Space Research, 25, 15 (2000).)

7.2 Auroral radio absorption

375

Figure 7.23. Electron-density profiles measured by the EISCAT incoherent-scatter radar during a slowly varying event in the morning of 23 March 1985. The heights of maximum ion production for electrons of the stated initial energy are marked on the right-hand axis. (Reprinted from J. K. Hargreaves and T. Devlin, J. Atmos. Terr. Phys. 52, 193, copyright 1990, with permission from Elsevier Science.)

matches the observed one. An essential element is the effective recombination coefficient as a function of altitude (Section 1.3.3), which relates the rate of electron–ion production to the resulting electron density. This may be taken from other experimental results or computed from the known chemistry of the D region (Section 1.4.3) – neither approach being entirely satisfactory. Figure 7.24 shows electron-density profiles measured by incoherent-scatter radar and the corresponding spectra of incoming electrons computed from them. Note that the daytime spectrum is “harder” (contains a greater proportion of more energetic particles) and that the resulting electron-density profile peaks at a lower altitude than does the night-time spectrum. Some direct comparisons have also been made. Using particle fluxes measured on low-orbit satellites, Jelly et al. (1964), Hargreaves and Sharp (1965), and Parthasarathy et al. (1966) obtained, respectively, the following empirical relations: A4 103J 1/2,

(7.7a)

The high-latitude D region

ELECTRON DENSITY AND ABSORPTION PROFILES DURING PRECIPITATION EVENTS Input values Matched profile Incremental absorption at 30 MHz

105

(a) 1984 Dec 14, 2056:10 UT (b) 1985 Mar 23, 0310 UT

Height (km)

100

95

90 (a)

(b)

(a)

(b)

85

80

75 10 –2

10 –1 Incr. abs. (dB km –1 ) 10 4

10 5 Ne

10 6

(cm–3)

7 6 5 Log (Flux)

376

4 3 2 1

(a)

0

50

100

150

(b)

200

250

Energy (keV)

Figure 7.24. Electron-density and absorption profiles for typical night and morning events, and the estimated spectra of incoming energetic electrons: (a) 14 December 1994 at 2056:10 UT; (b) 23 March 1985 at 0310 UT. In the upper panel the solid lines show the electrondensity profiles computed from the spectra in the lower panel, the black circles being observed values. (Flux is in units of cm2 sr1 s1 keV1.) The morning event has some ten times the flux of the night event between about 40 and 80 keV, whereas the night event has a greater flux of softer (25 keV) particles. The daytime absorption peaks at 87–88 km, the night absorption some 5 km higher.

7.2 Auroral radio absorption

377

A2 103J 1/2,

(7.7b)

A0.40Q1/2,

(7.8)

A3.3 103J 1/2

(7.9)

where A is the 30-MHz absorption in decibels, J is the flux of electrons of energies above 40 keV in cm2 s1 sr1, and Q is the total energy (above 80 eV) in erg cm2 s1. Equation (7.7a) is for day and (7.7b) is for night. The absorption is also significantly correlated to the energy flux over some energy ranges of electrons detected at geosynchronous orbit (Figure 7.25). Here the flux was taken only when the detector pointed into the loss cone; at other angles the electrons would mirror before reaching the D region. These, and other (Penman et al., 1979), comparisons have indicated that the absorption correlated best to the energy influx in the bands 40–80 keV and 80–160 keV. As might be expected, the rate of production calculated from the particle flux also correlated well at some heights. Schematic production-rate profiles derived from that comparison are shown in Figure 7.26. It should be stressed that these results and those of Figures 7.25 and 7.22 and Equations (7.7)–(7.9) make no distinction between types of event and should be taken as no more than indicative in any single instance.

The onset and dynamics of the substorm The night event which often begins with a sharp onset, and probably a spike too, is a consequence of the substorm in the magnetosphere. The riometer is therefore a useful monitor of the occurrence of substorms at the site of the riometer. This aspect was referred to in Section 6.4.6. Furthermore, the dynamics of absorption events which may be observed using a network (Section 7.2.4) relate to the development of particle precipitation in the substorm. The equatorward movement of the absorption arc preceeding an onset probably reflects the inward drift of an active region in the magnetosphere. Hargreaves et al. (1975) suggested that the motion may be E B drift due to the magnetospheric electric field, in which case the value of the field can be estimated from the relation (Ranta et al., 1981)

冢

dt E(mV m ) 5.88 10 d(1/L2 ) 1

4

1

冣

.

(7.10)

Equatorward drifts measured using a chain of riometers in Alaska (Figure 7.27) had speeds of several hundred m s1, greatest at the highest latitudes. Interpretation in terms of a magnetospheric electric field gives a median value of 1.3 mV m1. That the deduced field is independent of L supports the hypothesis, but the procedure has yet to be verified by direct comparisons with the electric field measured by other means. A later study using data from the Scandinavian sector revealed speeds mostly in the range 0–300 m s1 with a peak at 100–200 m

The high-latitude D region

200

200

175

175

ENERGY FLUX (109 keV cm2 s1 sr1)

ENERGY FLUX (109 keV cm2 s1 sr1)

378

160–320 keV

150 125 100 075 050 025 0

125 100 075 050 025 0

0

1

2

3 4 5 ABSORPTION (dB)

6

7

0

1

2

3 4 5 ABSORPTION (dB)

6

7

5 ENERGY FLUX (109 keV cm2 s1 sr1)

5 ENERGY FLUX (109 keV cm2 s1 sr1)

80–160 keV

150

40–80 keV

4

3

2

1

20–40 keV

4

3

2

1

0

0 0

1

2

3 4 5 ABSORPTION (dB)

6

7

3 4 5 ABSORPTION (dB)

6

7

0

1

2

3 4 5 ABSORPTION (dB)

6

7

ENERGY FLUX (109 keV cm2 s1 sr1)

10

20–320 keV

8

6

4

2

0 0

1

2

Figure 7.25. Relations between energy flux in selected bands measured on the GEOS-2 geosynchronous satellite, and radio absorption at 30 MHz observed in the auroral zone. The correlation is best in the two middle bands, indicating that the greatest contribution to the absorption comes from electrons of energy 40–160 keV. (Reprinted from P. N. Collis et al., J. Atmos Terr. Phys. 46, 21, copyright 1984, with permission from Elsevier Science.)

7.2 Auroral radio absorption

Altitude (km)

100

379

0.5 dB

1.0

2.0

3.0

5.0

90

80

70

102

103 104 3 Production rate (cm s1)

105

Figure 7.26. Schematic production-rate profiles for a range of values of 30-MHz radio absorption. The error bars are for one standard deviation. (Reprinted from P. N. Collis et al., J. Atmos Terr. Phys. 46, 21, copyright 1984, with permission from Elsevier Science.)

s1. Table 7.6 lists some values of the cross-tail electric field derived from this set of data. In this case the median value is 0.63 mV m1. On the other hand, the poleward motion of the spike event is not an E B drift (Nielsen, 1980). The morning events referred to in Section 7.2.2 typically move eastward, from the night towards the day side of the Earth. Velocities measured over wide baselines are not inconsistent with the concept that the electrons precipitated in the morning sector were originally injected into the closed magnetosphere near midnight and then moved eastward by gradient-curvature drift: an 80-keV electron would drift eastward at 2.6° min1 and cover 90° of longitude in 35 min. (Compare with Figure 7.16.) However, motions over smaller baselines tend to be significantly slower (even to the point of mere co-rotation) and some other mechanism is plainly at work as well. The westward movement before local midnight also requires some other explanation.

Conjugate behavior Auroral radio absorption is particularly well suited to studies of the relative behaviors of auroral phenomena at magnetically conjugate points: that is, at the northern and southern ends of the same field-line. In the first instance, one would expect to see the same intensity of absorption and the same patterns of variation. In fact these expectations are rarely met. For instance, the absorption tends to be stronger in the winter hemisphere (Figure 7.28), and there is considerable variation in individual cases, even to the extent that an event is seen at one station but not at all at

The high-latitude D region

380

(a)

BI

70

 (degrees)

FY

65

C P

SM A

60

W

0

0.5

1.0 V (km s1)

1.5

2.0

(b) 9 BI 8

L-value

7 FY 6 C 5

P

SM A 4 W 0.3

0.5

0.7

1.0

1.5 2.0 E (mV m1)

3.0 4.0

6.0 8.0

Figure 7.27. Equatorward drift of absorption bays preceeding onsets in Alaska. The locations of the riometer stations are shown by letters. (a) The speeds determined between pairs of stations (increasing with latitude). (b) The deduced magnetospheric electric field on the assumption that the motion is E B drift. The estimated field is independent of L. (Reprinted from J. K. Hargreaves et al., Planet. Space Sci. 23, 905, copyright 1975, with permission from Elsevier Science.)

the other. The night-time events, particularly at the higher latitudes, exhibit time differences between the peaks of events in conjugate regions, the event which appears first being of greater intensity than its counterpart in the conjugate region. One particularly interesting result, which so far lacks an explanation, is that the interhemispheric ratio depends on the direction of the interplanetary magnetic field carried by the solar wind (Figure 7.29).

7.2 Auroral radio absorption

381

Figure 7.28. The seasonal variation of the ratio of absorption in northern and southern conjugate regions at L values of 14, 7, and 4. At both of the higher latitudes the absorption tends to be larger in the winter hemisphere. There is some difference between the day and night events. (Reprinted from J. K. Hargreaves and F. C. Cowley (1967b), Planet. Space Sci. 15, 1585, copyright 1967, with permission from Elsevier Science.)

Table 7.6. Values of the cross-tail magnetotail electric field deduced from the equatorward drift of absorption arcs (Ranta et al., 1918) Date

UT

E(mV m1)

27 March 1975 2 May 1975

1430–1530 1300–1500 1720–1900 1700–1800 1600–1730 1930–2030 1900–2000 1800–1900 2000–2100 1900–2000 1800–1930

2.4 0.74 0.45 1.1 0.43 0.94 0.94 0.44 0.62 0.58 0.63

3 November 1975 2 March 1976 2 May 1976 22 May 1976 29 May 1977 4 May 1977

The high-latitude D region

382

Figure 7.29. The variation of the north-to-south absorption ratio between the conjugate stations Frobisher bay and South Pole (expressed as the “ratio function” (r1)/(r1)). In addition to the seasonal variation, note that the ratio is greater when the interplanetary magnetic field is pointed away from the Sun. (Reprinted from J. K. Hargreaves and F. C. Cowley (1976b), Planet. Space Sci. 15, 1585, copyright 1967, with permission from Elsevier Science.)

During the 1960s, spaced riometers were deployed around Byrd station in the Antarctic and its computed conjugate point (Great Whale River) in the Canadian Arctic, and these produced evidence that the conjugate point may be displaced by up to 85 km north–south with respect to the computed conjugate point, depending on the season and time of day (Hargreaves, 1969b). The absorption pulsations in the Pc4 and Pc5 bands, which appear as a modulation of slowly varying events in the morning sector, are observed to be in phase in magnetically conjugate regions (Chivers and Hargreaves, 1964). See Figure 7.30. This indicates that the modulation is symmetrical between hemispheres and is imposed in the magnetosphere. From the electron-density profiles observed during pulsations, it appears that the modulation involves the energy of the particles, not just their flux (Hargreaves and Devlin, 1990).

7.3

The polar-cap event

7.3.1

Introduction

In the history of ionospheric studies the polar-cap event is a relatively recent discovery. On 23 February 1956 there occurred a major solar flare that was followed by polar radio blackouts lasting for several days. At the same time, cosmic-ray monitors detected a large increase in the intensity of cosmic rays at ground level. The effects on VHF forward-scatter circuits operating at that time were studied by D. K. Bailey, who showed that the cause of the blackout was an enhancement of ionization in the D region of the polar ionosphere (Bailey, 1959). He deduced

7.3 The polar-cap event

383

2.0

Absorption difference (dB)

1.6 1.2 0.8 0.4 0 20.4

1840

1830

1820

1810

1800

1750

1740

1730

1720

1710

1700

1650

1640

20.8

UT

Figure 7.30. Conjugate pulsations at Great Whale River (– – –) and Byrd (——) (L7). The mean trend has been removed. The pulsations are essentially in phase at the conjugate stations. (Reprinted with permission, from J. K. Hargreaves and H. J. A. Chivers, Nature 203, 963, copyright 1964, Macmillan Magazines Limited.)

further that the most likely cause of this added ionization was a flux of energetic protons released from the Sun at the time of the flare. As with auroral absorption, studies based on the occurrence of the blackout condition have their limitations. Most scientific studies of the phenomenon have therefore made use of riometers, which give a quantitative measure of the absorption. These showed that the effects were confined to high magnetic latitudes, but, unlike AA, covered the whole polar cap. Thus they became known as polar-cap absorption (PCA) events. PCA events are much less frequent than auroral events, there being several each year on average. However, their effects, when they do occur, are more severe both because they blanket a large region of the Earth and because of the magnitude of the absorption. The most energetic events are also detected at the ground by cosmic-ray counters, and there is about one of these events each year on average. (In total 34 were noted in the 30 years between 1955 and 1985 – Smart and Shea, 1989.) A PCA that is also recorded by a cosmic-ray detector at the ground is called a ground-level event (GLE). The first recognized GLE occurred on 28 February 1942; it was identified in retrospect, of course, since the PCA was not yet a known phenomenon. The flare responsible for that event has another claim to fame as the source of the first solar radio noise to be recorded at the Earth. Since the early 1960s it has been possible to observe solar protons in space, and

The high-latitude D region

384

the monitoring of energetic protons from satellites has now become a matter of routine. As might be expected, the satellite monitors find some events that are not seen by ground-based methods. In fact, most solar flares emit protons at the lower energies – that is, up to 10 MeV. At energies of several tens of mega-electron volts the flux of protons reaching the Earth’s vicinity far exceeds that from galactic cosmic rays, though at the highest energies, greater than 1 GeV, the galactic particles dominate. 7.3.2

Observed properties of PCA events Occurrence and duration

There is no doubt that the PCA event is due to energetic (1–1000 MeV) protons emitted from the Sun, usually during a solar flare. The occurrence of PCA therefore depends strongly on the sunspot cycle. There can be more than ten events in an active year, and none at all – though more often one to three – near solar minimum. The long-term average is about six events per year. The numbers detected depend, of course, on what detection threshold has been selected. As an example, the occurrence of PCA events that reached at least 1 dB on a 30-MHz riometer situated within the polar cap is shown in Figure 7.31(a). This covers the period 1962–1972, which included the end of solar cycle 19 and the first eight years of cycle 20. Some of these events were much larger than 1 dB; 12% of them reached 10 dB or more. Those events recorded as &5 dB are indicated. It will be noted that none of these larger events occurred during the quiet years 1962–65. The durations of the events of magnitude &1 dB are shown in Figure 7.31(b). The median was about 2.5 days. The main group in the histogram spans 12–108 h. Those occurring within the narrow range 120–132 h appear as a separate group, but an alternative explanation may be that these long events were actually composed of several shorter ones. Be that as it may, we can summarize the distribution by saying that, once a PCA event has started, it is most likely to last for 1–4 days but in some cases may continue for a week or more. Figure 7.32(a) shows the occurrence of proton events in relation to the maximum flux of protons with energy at least 10 MeV. The plot covers the years 1976–1989, which included solar cycle 21 and the beginning of cycle 22. The general influence of the sunspot cycle is seen again, except that there is a remarkable dearth of events near the peak of the cycle in 1979–1980. This looks like an extreme case of a well-reported effect. The correlation between the solar cycle and the occurrence of PCA events is imperfect, but it has often been noticed that there are fewer events than might be expected at the maximum of the solar cycle. (Alternatively, there might be too many as the cycle begins and during its decline.) The pattern of occurrence varies from cycle to cycle, but this may be due in part to the statistics of small numbers. Although the number of sunspots is a guide, it is not safe, therefore, to try to predict from previous cycles too exactly.

7.3 The polar-cap event

(a)

385

16

Number of events per year

12

120

10

100

8

80

6

60

4

40

2

20

Sunspot number

Absorption ≥5 dB

14

0

0 62

63

64

65

66

67

68

69

70

71

72

Year (19 – –)

Duration of ≥1 dB events, 1962–72 (b)

7 May exceed stated value

6

Number of events

5

4

3

2

0 0

24

48

72

96 Duration (hrs)

120

144

168

>168

1

Figure 7.31. Occurrence and duration of PCA events producing at least 1 dB of absorption on a 30-MHz riometer in the polar cap. The period covered is 1962–1972. (a) The annual rate of occurrence, related to the 12-month running mean sunspot number. The incidence of events exceeding 5 dB is indicated. (b) Durations of&1-dB events. In some cases it was only recorded that the duration exceeded some value, and these are indicated by shading. The median duration was 62 h (about 2.5 days). It is possible that some of the longer events were composed of several shorter but overlapping events. (After M. A. Shea and D. F. Smart, Solar–Terrestrial Physics and Meteorology: SCOSTEP Working Document II, 1997; SCOSTEP Working Document III, 1979.)

The high-latitude D region

(a)

24

Flux >10 MeV (cm –2 s–1 str –1) <100 100–1000 >1000

22

18 16

160

14

140

12

120

10

100

8

80

6

60

4

40

2

20

0

76

77

78

79

80

81

82

83

84

85

86

87

88

89

Sunspot number

Number of events per year

20

0

Year (19 – –)

10 9 8 7

30 MHz absorption at Kilpisjärvi and flux of ≥10 MeV protons at geosynchronous orbit

(b)

6 5 4 Absorption (dB)

386

A ∝ J1/2

3 2.5 2 1.5

1.0

0.7 ≤0.5 10

20

50

100

200

500

1000

2000

5000 ≥10,000

Proton flux (cm –2 s–1 str –1)

Figure 7.32. Some properties of solar-proton events recorded by a geosynchronous satellite and a riometer in the auroral zone. (a) The incidence of proton events according to the maximum proton flux at synchronous orbit. The sunspot numbers are shown as the 12month running mean. The data cover the years 1976–1989. (b) The relation between 30MHz absorption at Kilpisjärvi and the flux of protons of energy&10 MeV at geosynchromous orbit. 60% of the points lie between the straight lines marked, representing J37A2 and J200A2 – compare with equation (7.12). (Data from H. Ranta et al., J. Atmos. Terr. Phys. 55, 751 (1993).)

7.3 The polar-cap event

387

The direct connection between proton flux and PCA is confirmed by Figure 7.32(b), which plots the absorption at Kilpisjärvi, Finland, against the proton flux detected on a satellite in synchronous orbit. The straight lines indicate the law absorption  (flux)1/2,

(7.11)

which is to be expected if the electron-production rate is proportional to the particle flux. Note that fluxes greater than 100 cm2 s1 sr1 are likely to produce a significant PCA. Since Kilpisjärvi is in the auroral zone (at L5.9) rather than the polar cap, the absorption recorded there may at times be reduced by the proximity of the edge of the polar cap. An approximate rule that is sometimes used to deduce the proton flux from the radio absorption (Smart and Shea, 1989) is J10A2,

(7.12)

where J is the flux (in cm2 s1 sr1) of protons with energy exceeding 10 MeV, and A is the absorption (in decibels) measured with a 30-MHz riometer in the sunlit polar cap. The statistics of the occurrence of PCA is complicated by episodic behavior. An individual proton event is generally recognized by noting an increase in proton flux or by virtue of radio absorption having the established PCA characteristics (i.e. a smooth event of long duration). However, an active solar region may well persist long enough to produce two or more proton flares and it is not unusual, therefore, for two or more PCAs to occur within a few days of each other. Since an event may last for several days, some events run into each other. The data set shown in Figure 7.31 contained 63 events. Of these, 25 occurred within one of ten groups of events, the criterion for a group being that events occurred within 5 days of each other. The count of groups is of course less than the count of individual events. To take an example, 1968 had 11 PCA events, but eight of them occurred in three groups and only three events were isolated. Perhaps 1968 should be credited with six PCA-producing regions, therefore, instead of with 11 PCAs. 1969 was also significantly affected in this way: in February of that year four events occurred on four successive days! Beyond a general impression that more events fall within groups in the more active years, it is difficult to draw general conclusions because of the small numbers involved.

Variation from month to month One of the puzzles regarding the occurrence of PCA which came to light early was what appeared to be a seasonal effect: it was observed that fewer events occurred during the northern hemisphere winter than at other times of year. There is no reason to suppose that the Sun becomes less active in December and January, but, taking type-IV radio bursts (see Section 7.3.3) as a reference, there is evidence that the protons were taking longer to reach the Earth at those times – see Figure 7.33.

The high-latitude D region

(a)

1.0

Relative occurrence

388

0.5

0

J

F

M A M

J

J

A

S O

N D

(b)

10

Delay time (h)

Month

5

Figure 7.33. Seasonal effect in PCA: (a) the fraction of flares having Type-IV radio bursts which also produce PCA, and (b) the seasonal variation of the time delay between a radio burst and commencement of PCA. (After B. Hultqvist, Solar Flares and Space Research (eds. de Jager and Svestka), p. 215, North-Holland, 1969.)

Delay time (h)

0 J

F M A

Strong events

M J

J

A S O N D

Weak events

It has also been argued that the effect may be artificial and due to some observational bias. The most likely cause of bias is that, since the absorption is weaker in a dark ionosphere (Section 7.3.6), the ionosphere is dark for more of the time in winter, and more of the early riometer stations were in the northern hemisphere, then the detection of PCAs by radio would be less sensitive overall in the northern winter. Supporting this view (which probably holds sway at present) is the fact that the anomaly in the seasonal occurrence seems to be one of those effects which becomes less convincing the more intensively they are studied. It has tended to vanish as the data base has grown with the passing of the years! Thus the sets of data used for Figures 7.31 and 7.32 both show the incidence varying considerably from month to month, but they contain no evidence for any significant seasonal effect. Indeed, the monthly distribution of proton events measured on a satellite appears to show some preference for the equinoxes (Smart and Shea, 1989). Since the question of seasonal effects remains in doubt, it is probably best to assume for prediction purposes that the incidence of PCA has no seasonal dependence beyond ordinary statistical variations. That assumption being made, the probability that a stated number of events will occur in one month may be calculated from the Poisson distribution which

7.3 The polar-cap event

389

describes the frequency of occurrence of independent events within a given period of time. We have to know (or assume) the average rate of occurrence. Table 7.7 gives the monthly statistics for annual rates of occurrence of two, six, and ten, corresponding approximately to low, average, and high PCA activity. Since events occurring in a group (as defined above) are probably not independent, a group should be counted as one event for this purpose.

Magnitude Not surprisingly, there are more small PCA events than large ones. Table 7.8, taken from the data of Shea and Smart (1977; 1979), shows how many events exceeded various absorption thresholds during the 11-year period 1962–1972. Note that, of the events of magnitude &0.5 dB, about half reach 1 dB, about one fifth of those reach 5 dB, and about one third of those reach 15 dB. An approximate rule that appears to satisfy the limited information available is that the number of events exceeding a stated threshold varies in inverse proportion to that threshold value. The review by Smart and Shea (1989) discusses the incidence of proton events in some detail. 7.3.3

The relation to solar flares and radio emissions

In fact, not all large flares give rise to proton events and there are some proton events that have not been associated with any known flare. However, although the correlation might not be 100%, there is no doubt that, as a general rule, proton Table 7.7. The probability that the stated number of PCA events will occur in one month, given the annual rate Expected annual rate

0

1

2

3

2 6 10

0.846 0.607 0.435

0.141 0.303 0.362

0.012 0.076 0.151

0.001 0.013 0.042

Probability of the stated number occurring in one month

Table 7.8. The distribution of magnitude of PCA events Threshold (dB) (30-MHz riometer) Total number of events exceeding threshold Percentage of total

0.5 113 100

1.0 63 56

2.0 36 32

5.0

10.0

15.0

13 11.5

8 7.1

3 2.7

The high-latitude D region

390

events are associated with the larger solar flares. Those flares that produce protons are often called proton flares and they are recognized as a distinct class in the flare predictions which are issued regularly by various national and international warning agencies. The solar radio emissions known as type IV are useful for predicting which flares emit protons. The type-IV emission is a radio burst of long duration that follows some flares and covers a wide band of radio frequency. (It is attributed to synchrotron radiation from high-energy particles gyrating in the solar magnetic field.) The bursts associated with proton emission are characterized by a U-shaped spectrum in which the intensity is smaller in the middle than it is at the ends. For example, if the spectrum covers the range from a few hundred megahertz to 10 GHz, it will be relatively strong at the high- and low-frequency ends but weaker at the middle frequencies around 1 GHz. From the spectral characteristics of the radio burst it is possible to predict the flux of protons with energy exceeding 10 MeV (Castelli et al., 1967) and also the proton spectrum (Bakshi and Barron, 1979). Since the radio burst is received at the Earth some time before the protons are due to arrive, the association obviously has some practical importance. The association between proton ejection and the radio burst is also useful for identifying the flare responsible and for timing the flight of the proton cloud to Earth. This time appears to be shorter (about 1 h) for strong events and longer (about 6 h) for weak ones.

7.3.4

Effects arising during the proton’s journey to Earth

The production and release of energetic protons appears to be a normal part of the solar-flare phenomenon, and flares causing PCA and GLE at the Earth probably differ from others more in degree than in nature. The journey from Sun to Earth may be considered in three parts: (a)

propagation from Sun to Earth – i.e. from the Sun to the boundary of the magnetosphere;

(b)

motions within the magnetosphere; and

(c)

the interaction between protons and the atmosphere.

Effects in interplanetary space The propagation of charged particles in the space between the Sun and the Earth is affected by the form of the interplanetary magnetic field. As illustrated in Figure 2.3, the field has a spiral form due to the rotation of the Sun, and this affects the propagation of solar protons despite the weakness of the field and the high energy of the protons. The gyroradius of a 1-GeV proton in a field of 5 nT is less than a hundredth of the distance between the Sun and the Earth; hence there is time for the IMF to act on even an energetic proton. Those with less energy gyrate in

7.3 The polar-cap event

Absorption (dB)

(a)

0

Thule, Greenland

5

10

15

20

(b) 0

10 11 12 13 14 15 16 17 18 19 20 21 22 May 1959

391

Figure 7.34. A PCA recorded by riometers at (a) Thule, Greenland and (b) College, Alaska. Thule is in the polar cap and College in the auroral zone. The event lasted for a week at both sites, but was modulated at the lower latitude. (G. C. Reid, in Physics of Geomagnetic Phenomena (eds. Matsushita and Campbell), Academic Press, 1967.)

College, Alaska

Absorption (dB)

5

10

15

20 10 11 12 13 14 15 16 17 18 19 20 21 22 May 1959

smaller loops (gyroradius  energy1/2) and are more tightly controlled, responding to irregularities in the IMF as well as to its general form. Hence there is scattering, and the protons appear to be coming from all directions by the time they reach the Earth. Scattering in the interplanetary medium, since it also provides a mechanism for storing particles in space, can account for the observed time delay between a flare and the beginning of the PCA, and for the duration of PCA events. A proton of energy 10 MeV would reach the Earth in only 1 h if it traveled in a straight line, and the duration of a flare is typically some tens of minutes only. In fact the delay before an event begins is typically several hours, and the event due to one flare may last for several days (Figure 7.34). Further evidence for the role of the IMF is as follows. (a)

Flares near the eastern limb of the Sun rarely give rise to PCA events, whereas some events seem to be associated with flares that are out of sight around the western limb. This is illustrated by Figure 7.35, which gives the positions of solar flares associated with those proton events energetic enough to be detected at ground level (i.e. GLEs). (Note that the western

The high-latitude D region

392

Figure 7.35. Solar longitudes of flares associated with ground-level events. The Sun is happy about this. (D. F. Smart and M. A. Shea. J. Spacecraft Rockets, 26, 403, 1989.)

limb of the Sun is on the right-hand side as seen from the terrestrial northern hemisphere.) It is obvious that the distribution of these flares with solar longitude is significantly biased towards one side of the central meridian. The heliolongitudinal distribution of the source flares broadens for protons of lower energy (Smart and Shea, 1995). (b)

The time delay between a flare and the related PCA increases with the eastern longitude of the flare.

(c)

The delay between flare and PCA is greatest at times of high solar activity, and this is also when the IMF is most irregular.

Recent improvements in detecting structures in the interplanetary medium have focused attention on the role of coronal mass ejections (Section 2.2.2). It is found that some PCAs may be attributed not directly to flares but to the shock wave related to a coronal ejection of mass from the Sun (Shea and Smart, 1995).

Effects in the magnetosphere On reaching the magnetopause the protons must then pass through the geomagnetic field to reach the atmosphere. To a first approximation this problem may be treated by Störmer theory. The theory describing the trajectories of charged particles in a dipolar magnetic field, which C. Störmer worked out in connection with his studies of the aurora, does not actually apply to auroral particles (because their energies are too low)! However, the theory is valid for cosmic rays and for solar protons. In a magnetic field a charged particle tends to follow a spiral path whose radius

7.3 The polar-cap event

393

of curvature (rB mv/(Be)) is directly proportional to its velocity and inversely proportional to the magnetic flux density. Because solar protons are of relatively high energy, the magnetic field changes significantly over one gyration, and therefore trapping theory (as outlined in Section 2.3.4) does not apply. Nonetheless, it is still true that particles traveling almost along the magnetic field undergo the least deviation. To reach the equator, which is the least accessible region, a proton has to cross field-lines all the way down to the atmosphere. Charged particles may do this only if they are sufficiently energetic, and the equatorial region is effectively forbidden to typical protons of solar origin. However, most of the particles in a proton event can penetrate into the atmosphere over a polar cap extending down to about 60° magnetic latitude. Since the radius of gyration in a given magnetic field depends on the momentum per unit charge (mv/e), it is convenient to discuss particle orbits in general in terms of a parameter called rigidity: RPc/(ze),

(7.13)

where P is the momentum, c the speed of light, z the atomic number, and e the electronic charge taken positive. The advantage of this parameter is that all particles with the same value of R will follow the same path in a given magnetic field. Although the trajectory of a proton in the geomagnetic field can be very complicated, Störmer’s analysis simplified matters by defining “allowed” and “forbidden” regions that could and could not, respectively, be reached by a charged particle approaching the Earth from infinity. To reach magnetic latitude c in a dipole field, the rigidity of the particle must exceed a cutoff rigidity, Rc: Rc 14.9cos4 c,

(7.14)

where Rc is measured in gigavolts (109 V). That is, particles of rigidity Rc reach latitudes c and above. Conversely, a place at latitude c would receive only those particles with rigidities equal to and greater than Rc. Figure 7.36(a) plots the Störmer cutoff latitude against energy both for protons and for electrons. To perform an exact calculation of the trajectory of a proton through the geomagnetic field, the procedure is to imagine that a proton with negative charge is projected upward from the point of impact, since the trajectory of such a particle is exactly the reverse of that of an incoming positively charged particle having the same rigidity. From a set of computations of this kind it is possible to work out the directions in space from which the particles reaching a given place at a given time must have come. Results confirm other evidence that, whereas most protons are isotropic near the Earth, the more energetic ones (those exceeding 1 GeV which are responsible for ground-level events) originate from the western side of the Sun. (See Figure 7.35.) During the main part of a typical PCA event the absorption region is essentially

The high-latitude D region

394

(a)

(b)

Figure 7.36. (a) The Störmer cutoff latitude for protons and electrons. (S.-I. Akasofu and S. Chapman, Solar–Terrestrial Physics. Oxford University Press, 1972, by permission of Oxford University Press. After T. Obayashi, Rep. Ionosphere Space Res. Japan, 13, 201, 1959.) (b) Cutoff latitudes for dipole and realistic geomagnetic fields (G. C. Reid and H. H. Sauer, J. Geophys. Res. 72, 197, 1967, copyright by the American Geophysical Union.)

uniform and symmetrical over the polar caps down to about 60° geomagnetic latitude. According to Störmer theory these protons should have energies exceeding 400 MeV, but direct observations of the particles have shown that the cutoff rigidity at the edge of the polar cap is significantly less than the Störmer value. The situation appears to be that there is a main polar cap surrounding the geomagnetic pole that is open to solar protons of all energies, and then at slightly lower latitude the cutoff reverts fairly abruptly to the Störmer value. Much of this effect (though perhaps not all) may be explained by taking account of the tail of the magnetosphere which connects directly to the polar caps and presumably provides an easy

7.3 The polar-cap event

route even for protons of low energy. Figure 7.36(b) shows the difference in cutoff energy between dipolar and more realistic geomagnetic fields. The cutoff is reduced still further if a magnetic storm (Section 2.2.3), which enhances the ring current (Section 2.3.5) and moves the magnetopause inward, occurs while a PCA event is in progress. The geographic regions most affected by PCA are illustrated in Figure 7.37 in general terms. The boundaries may be several degrees nearer the equator during a magnetic storm. 7.3.5

Non-uniformity and the midday recovery Non-uniformity

The spatial distribution of radio absorption is not always uniform, particularly during the early and late phases of a PCA. The absorption usually appears first near the geomagnetic poles and then spreads to cover the polar caps some hours later. Towards the end of the event there is likely to be contamination by auroral electrons related to a magnetic storm, and a concentration of absorption into the auroral zone is then to be expected. In addition, the polar cap expands during the storm, moving the PCA equatorward.

Midday recovery Some events exhibit a reduction in the absorption for several hours near local noon. This effect is known as the midday recovery (MDR), and its main properties are as follows (Leinbach, 1967). (a)

They occur during about 20% of PCA events.

(b)

They are usually pronounced on the first day of the event only.

(c)

They peak between 0800 and 1500 LT, most of them between 1000 and 1200.

(d)

They may last as long as 6–10 h, most being remarkably symmetrical about the peak.

(e)

They are strongest near the equatorward boundary of the polar cap, and are not evident at locations well within the polar cap.

(f)

When the polar cap expands during a magnetic storm, the recovery region remains at its equatorward edge.

Figure 7.38 illustrates some of these features during a PCA observed at the Alaskan stations College (L5.5), Farewell (L4.3) and King Salmon (L3.3). The time scale is given in UT, from which 10 h should be subtracted to obtain Alaskan time. MDRs occurred between 0800 and 1000 LT at the first two stations on the first day of the event. On the second day a magnetic storm extended the polar cap to lower latitude and a MDR was observed at King Salmon, but was not (College) or was barely (Farewell) seen at the higher latitudes. (The horizontal bars on Figure 7.38 indicate night-time recoveries; these are different and will be considered in Section 7.3.6.)

395

396

The high-latitude D region

Figure 7.37. The polar areas normally affected by polar-cap absorption. The regions inside the inner curves may be considered as “polar plateaux”, whereas regions outside the outer curves are usually not affected except during severe geomagnetic disturbance. The outer edges of the diagrams are at latitude 45°. (G. C. Reid, Physics of the Sun (ed. P. A. Sturrock), 3, 251, Reidel, 1986, with kind permission from Kluwer Academic Publishers.)

7.3 The polar-cap event

Figure 7.38. The PCA event of 7 July 1958, seen at College (L5.5), Farewell (L4.3), and King Salmon (L3.3). The horizontal bars indicate night recoveries and MDR marks midday recoveries. All observations were at 27.6 MHz. (H. Leinbach. J. Geophys. Res. 72, 5473, 1967, copyright by the American Geophysical Union.)

In a recent case study using data from 25 stations including some in the southern hemisphere (Uljev et al., 1995) the maximum effect was found slightly before local noon, covering a range of magnetic latitude approximately from 60° to 70° (Figure 7.39). The effect seems to occur simultaneously and with the same magnitude in magnetically conjugate regions, and it is confirmed that the effect is not seen at stations well inside the polar cap (at latitudes greater than 70°). Two possible explanations were put forward by Leinbach (1967): a local change of cutoff, and the development of anisotropy in the pitch-angle distribution of the incoming protons. More recent studies have suggested that both effects may occur. There is evidence that a change of cutoff near noon is indeed one factor (Hargreaves et al., 1993), and modeling studies (Uljev et al., 1995) suggest that anisotropy of the pitch-angle distribution also occurs but only over the latitude range 65°–70°.

397

The high-latitude D region

398

Figure 7.39. The region affected by midday recovery during the event of 20 March 1990. The coordinates are invariant latitude and magnetic LT. The broken line between 10 and 12 h marks the times of minimum absorption at each station. (Reprinted from V. A. Uljev et al., J. Atmos. Terr. Phys. 57, 905, copyright 1995, with permission from Elsevier Science.).

7.3.6

Effects in the terrestrial atmosphere Upper-atmosphere ionization during a proton event

Energetic protons entering the terrestrial atmosphere lose energy in collisions with the neutral molecules and leave behind an ionized trail. In order to reach an altitude of 50 km, a proton must have an initial energy of 30 MeV, and to reach the ground (to cause a GLE) the energy must be over 1 GeV. (Refer to Figure 2.28.) An example of proton spectra observed at geosynchronous orbit during a proton event in 1984 is shown in Figure 7.40(a). Despite the name solar proton event, it should be appreciated that other particles, -particles and heavier nuclei, also arrive (in proportions typical of the solar atmosphere). However, their contribution to the ionization is small relative to that of the protons. The computation of ionization by protons and -particles was discussed in Section 2.6.3. Having computed the rate of production of electrons at a given height, knowledge of the effective recombination coefficient allows one to calculate the resulting

7.3 The polar-cap event

(a)

(b)

399

Figure 7.40. (a) Proton fluxes measured by the geosynchronous satellite GOES-5 on 16 February 1984, fitted by spectra of form E. (Data from F. C. Cowley, NOAA, Boulder, Colorado, private communication.) (b) Electron-density profiles measured by incoherent scatter radar during the same event. (Reprinted with permission from J. K. Hargreaves et al., Planet. Space Sci. 35, 947, copyright 1987, with permission from Elsevier Science.)

400

The high-latitude D region

electron density. If an event contains particles of energy 1–100 MeV, the effects should appear within the height range 35–90 km (Figure 2.28). Effects due to the higher energies tend to be smaller because the flux is smaller and the rate of recombination is greater at lower height. Nevertheless, in some events substantial ionization is created down to 50 km.

Determination of the recombination coefficient In fact, the recombination coefficient in the lower ionosphere is not a wellestablished quantity, and one use of PCA events is to measure the recombination coefficient and its variations over a range of heights in the mesosphere. The proton spectrum may be measured from a geosynchronous satellite, and from it the production rate can be computed over a range of heights using a model of the neutral atmosphere. Electron-density profiles can be determined from rocket measurements or by incoherent-scatter radar. An example of the latter is shown in Figure 7.40(b). Some studies have used riometer data, which are more readily available, though in that case only the integrated absorption can be compared. Values of the effective recombination coefficient obtained using electron densities from incoherent-scatter radar are shown in Figure 7.41. Most striking about these values is their large spread. There is a major difference between day and night, and also between different determinations of daytime values. It is possible that there are seasonal variations caused by seasonal changes in the concentrations of minor species (Reagan and Watt, 1976). These differences will be considered in the next section.

Day–night variation and twilight effects Because the proton influx during a PCA decays relatively slowly, the effects of daily variations in the complex chemistry of the region may also be detected. The most obvious effect is a large diurnal variation in the absorption, which is typically about five times as large by day as it is by night, though the ratio can be as small as three or as large as ten. The critical factor in this is whether the lower ionosphere is sunlit. Night-time recoveries were marked on Figure 7.38, and they also account for the daily absorption recoveries in Figure 7.34(b). (Thule, Figure 7.34(a), was illuminated continuously and the recoveries did not occur there.) The effect is perhaps seen most clearly by comparing the absorption at magnetically conjugate stations, one in the summer and the other in winter, as in Figure 7.42. Over the Spitzbergen station the ionosphere was illuminated continuously, whereas at Mirnyy the Sun was above the horizon for only a few hours of the day. We expect the proton fluxes at each place to be similar, and, indeed, the absorption was almost the same when both stations were sunlit. However, the absorption fell to a considerably smaller value at Mirnyy during each night period. The cause of the day–night modulation is without doubt a variation in the ratio of the concentrations of electrons and negative ions,  (defined in Section 1.3.3). In a dark ionosphere, electrons become attached to oxygen molecules to form

50

60

70

80

90

10 –7

(4)

(3)

(2)

(1)

Effective recombination coefficient (cm 3 s –1)

10 –6

10 –5

10 –4

Figure 7.41. Effective recombination coefficients determined from PCA observations, using electron densities measured by incoherent-scatter radar. Key: (1) Daytime (range of values over several days). Summer (August). (Data from J. B. Reagan and T. M. Watt. J. Geophys. Res. 81, 4579, 1976.) (2) Daytime (range of values over 3 hours near noon). Winter (February). (Data from J. K. Hargreaves, H. Ranta, A. Ranta, E. Turunen, and T. Turunen. Planet. Space Sci. 35, 947, 1987.) (3) Daytime (Afternoon). Spring (March). (Data from J. K. Hargreaves, A. V. Shirochkov, and A. D. Farmer. J. Atmos. Terr. Phys. 55, 857, 1993). (4) Night. (Same source as (3).)

Height (km)

The high-latitude D region

402

10 Absorption (dB)

Mirnyy Spitsbergen Magnetic disturbance

5

Solar elevation (degrees) at Mirnyy

0 12 July12

12 July 13

12 July 14

12 July 15

12 UT July 16

10 0 10 20 30 40 50 Solar elevation at Spitsbergen: within 10° and 34° throughout

Figure 7.42. Polar-cap absorption at magnetically conjugate stations 12–16 July 1966, Spitsbergen in the northern hemisphere and Mirnyy in the Antarctic. (Reprinted from C. S. Gillmor, J. Atmos. Terr. Phys. 25, 263, copyright (1963), with permission from Elsevier Science.)

negative oxygen ions (O2 ), as Equation (1.61), but in sunlight the electrons are detached again by visible light (Equation (1.62)) or through other chemical reactions. (See Section 1.4.3.) Since only the ionospheric electrons contribute to the absorption, a variation of  leads to a variation of absorption even though the production rate, q, remains constant. The changes between night and day take place over the twilight periods at sunrise and sunset, and the details are of particular interest. The timing of the change in relation to the elevation angle of the Sun indicates the presence of a screening layer, probably ozone. Since ozone does not absorb in the visible, the solar radiation that detatches electrons from negative ions must be in the ultraviolet rather than the visible region of the spectrum (Reid, 1961). The effect is confined to altitudes below 80 km (Figure 7.43), which explains why it does not appear in AA (most of which occurs at a higher level). When the details are examined it becomes apparent that some other factors are also at work. (a)

There is an asymmetry between the sunrise and sunset changes. The increase of absorption over sunrise is slower than the decrease over sunset (Chivers and Hargreaves, 1965). This means that the absorption is larger at sunset than it is at sunrise for the same solar zenith angle. The effect may be seen by plotting the absorption at a station passing through twilight periods as a ratio to that at one that is constantly illuminated. The result is a hysteresis curve like Figure 7.44, in which the curve is described counterclockwise. The same effect is present in the profiles of Figure 7.43, where

7.3 The polar-cap event

403

90

(a) 80

70 98 96 60 ALTITUDE (km)

 = 90 91 92 93 94 95

(b) 80

70

 = 90

60

106

91

92 94 96 98 93 95 97

105

104

EFFECTIVE RECOMBINATION COEFFICIENT,  t (cm3 s1)

Figure 7.43. Effective recombination coefficients at various solar zenith angles over sunrise and sunset during the major proton event of August 1972. (J. B. Reagan and T. M. Watt. J. Geophys. Res. 81, 4579, 1976, copyright by the American Geophysical Union.)

the smaller recombination coefficients at given zenith angle at sunset imply greater electron densities. Collis and Rietveld (1990) showed that the timing of the day–night transition depends also on the altitude, and suggested by way of explanation that different processes control the electron density above 70 km (photodetachment from O2 ) and below 66 km (collisional detachment due to O2(1g)) during the twilight period. (b)

There is evidence that the effective recombination coefficient varies with time even within the day and night periods. Reagan and Watt (1976) found that its value declined gradually during the sunlit period (i.e. between sunrise and sunset) by as much as a factor of two (at some heights). On the other hand, Hargreaves et al. (1993) reported a gradual increase of the effective recombination coefficient throughout the night. The reasons for these slow changes are not known, though they presumably lie in the chemistry of the mesosphere.

Figure 7.44. Twilight variations expressed as hysteresis curves. The reference station is South Pole (solar elevation 7°), and the curves are described counter-clockwise, implying larger absorption at sunset than at sunrise for the same solar elevation. (H. H. Sauer. J. Geophys. Res. 73, 3058, 1968, copyright by the American Geophysical Union.)

7.3 The polar-cap event

405

Assuming that the day–night change which occurs rapidly over twilight is indeed due entirely to a variation in the ratio of negative ions to electrons (), and that recombination of positive and negative ions is negligible, then a simple application of Equation (1.39) – remembering also that the effective recombination coefficient eff q/N e2 by definition – gives a relation between night and day values of  at a given height: 1   (night) eff (night)  . 1   (day) eff (day)

(7.15)

If we take typical estimates of (day) of 1, 0.25, and 0.68 at 80, 75, and 70 km, respectively, the results of Hargreaves et al. (1993), to take an example, give (night) values of 1.7, 20, and 100 at the same heights. There is, however, no generally agreed set of values for this quantity.

Effects on the neutral-species composition Influxes of energetic particles have another important effect in that they may produce changes in the chemical composition of the atmosphere. As long ago as 1969 it was observed in rocket measurements that the ozone in the mesosphere (at heights of 54–67 km) was depleted during a PCA event by a factor between two and four depending on the height (Weeks et al., 1972). The mechanism is as follows. One .H2O), consequence of the ionization processes is the formation of hydrated ions (O 2 which then undergo further reactions leading to “odd hydrogen” species such as H and OH. These radicals then react with ozone to produce molecular oxygen: HO3 →OHO2 OH O3 →HO2 O3 O3 →2O2

(7.16a)

OHO3 →HO2 O2 HO2 O3 →OH2O2 2O3 →3O2

(7.16b)

OHO3 →HO2 O2 HO2 O3 →OHO2 2

HO3 O3 →2O2

(7.16c)

In each case the odd-hydrogen radical is a catalyst; it is destroyed in the first reaction of the pair but regenerated in the second. These processes require a sufficient concentration of water vapor and therefore they are confined to the region below the mesopause. They are thought to be important over the height range 50–90 km. Several hours to a day after the precipitation event, the odd-hydrogen species reform into stable molecules; then the above reactions cease and the concentration

The high-latitude D region

406

of ozone recovers. However, since the H atoms tend to recombine to form H2 rather than H2O, the water vapor may remain depleted for some time. There may be an increase in the concentration of ozone during this period. More serious from the point of view of ozone is the effect of “odd-nitrogen” species. These have a much longer lifetime (amounting to several years) in the stratosphere, which is also the site of most of the ozone. The ionization processes produce secondary electrons with energies of tens and hundreds of electron volts, and these can dissociate and ionize molecular nitrogen to produce atoms and ions of atomic nitrogen. The N and N then react with O2 to give nitric oxide, NO, which in turn acts to destroy ozone as follows: NO O3 →NO2 O2 NO2 O → NOO2 O3 O → 2O2

(7.17)

Here the NO is the catalyst. This reaction is important up to 45 km, and the long lifetime of NO at those levels means that a given molecule may pass through the reaction many times, converting one O3 at each pass. The above reactions do not depend on the nature of the primary ionizing radiation, but they are of particular importance in PCA because the more energetic protons ionize at particularly low altitudes and down into the stratosphere. The processes actually go on continuously with the arrival of galactic cosmic rays, but it has been estimated that the total production of NO during one major PCA event can be very great, even exceeding the annual production by cosmic rays. The great proton event of August 1972 had a measurable effect on the ozone concentration in the stratosphere, which fell by 15%–20% at latitudes 75°–80°. In the event of July 1982 ozone was depleted between 55 and 85 km. This was a relatively “soft” event, which explains why the effects were higher up. A series of PCA events that occurred in 1989 is also thought to have affected the ozone content. A computation of the effect of the events during that year is shown in Figure 7.45. The O3 was depleted by more than 10% over a limited height range for several months in 1989, and small effects continued for a year or more. Significant though these effects are, they have no known effect on high-latitude radio propagation. Further information is given in papers by Reid (1986) and Jackman (2000). 7.4

Coherent scatter and the summer mesopheric echo

Incoherent and coherent scattering of radio waves in the ionosphere exploit different phenomena, as a result of which the second process is much the stronger (see Section 4.2.2). Given the utility of incoherent-scatter radar in ionospheric studies at high latitude, it would be a great pity if the weak signals which it uses were to be swamped by coherent echoes from the same region. Yet this is just what may occur. Coherent echoes from the high-latitude D region were first detected in the VHF

7.4 Coherent scatter

Figure 7.45. Computed variation of NOy and O3 concentrations at 75° north due to the solar proton events of 1989. The contours for NOy are 0, 1, 2, 10, 20, 100 and 200%. For O3 they are 2, 1, 0.2, 0, 0.2, 1, 2, 10 and 20%. The concentration of NOy is increased but that of O3 is decreased. Note the long duration of the effects. (C. H. Jackman et al., J. Geophys. Res. 105, 11659, 2000, copyright by the American Geophysical Union.)

band (at 50 MHz) in Alaska (Ecklund and Balsley, 1981), and subsequently in Norway at 53.5 MHz (Czechowsky et al., 1989) and with the EISCAT 224-MHz radar (Hoppe et al., 1988). They have also been observed, though less frequently, with the EISCAT UHF system at 933 MHz (Röttger et al., 1990). Other observations cover the range 2.27 MHz to 1.29 GHz (Röttger, 1994). These strong echoes occur only in summer and are now usually called polar mesosphere summer echoes (PMSEs). They are a nuisance to IS radar but constitute an interesting topic in their own right, particularly since they have proved to be something of a mystery. Their characteristics are very different from those of the incoherent echoes received from the D region during particle precipitation. Not only are they much more intense, but also they are much narrower, usually less than 1.5 km deep, though there can also be multiple layers (Figure 7.46). The height range is more restricted, too, peaking at 84–86 km (Figure 7.47), an altitude close to the mesopause. When the echoes are present their height fluctuates (Figure 7.48), which is thought to indicate the passage of acoustic-gravity waves (Section 1.6). The spectrum of PMSE is considerably narrower than that of IS returns (Figure 7.49); even without the other evidence this point alone would be sufficient proof that quite different mechanisms are responsible.

407

408

The high-latitude D region

Figure 7.46. An example of PMSE observed at 224 MHz on 29 June 1988 using the EISCAT VHF radar. The density of blob suggests the strength of the echo. Note the height variations and the multiple layers. (Reprinted from P. N. Collis and J. Röttger, J. Atmos. Terr. Phys. 52, 569, copyright 1990, with permission from Elsevier Science.)

Figure 7.47. A histogram of the height distribution of PMSE observed with the EISCAT VHF radar. (Reprinted from J. R. Palmer et al., J. Atmos. Terr. Phys. 58, 307, copyright 1996, with permission from Elsevier Science.)

Most strikingly, the echoes are clearly a summer phenomenon, occurring from June to August only, with a maximum in July in the northern hemisphere (Palmer et al., 1996). The occurrence also varies during the day. There are maxima near noon and midnight, and minima in the morning and the evening hours. The percentage occurrence, though not very well established, is some 50%–75% of days at

7.6 Summary and implications

Figure 7.48. Rapid height fluctuations in PMSE, consistent with acoustic-gravity waves, on various dates in 1988 and 1991. The dashed curves show the rate of change of altitude, and the solid curves the vertical velocity derived from the Doppler shift of the echoes. (Reprinted from J. R. Palmer et al., J. Atmos. Terr. Phys. 58, 307, copyright 1996, with permission from Elsevier Science.)

the maxima and 10%–50% at the minima. The daily variations are most marked in June and August. The polar mesosphere is particularly cold in the summer, and this may be the key to the mechanism. It has been proposed (Kelley et al., 1987) that water-cluster ions, whose formation is favored by low temperature, reduce the diffusion coefficient of electrons and so extend the scale of turbulence, allowing coherent scatter to occur at shorter wavelengths. However, other mechanisms have also been proposed. The development of PMSE studies and the relevent theories have been reviewed by Cho and Kelley (1993) and by Röttger (1994).

7.5

Summary and implications

At middle and equatorial latitudes D-region absorption has only a minor effect on HF propagation, but at high latitude it can affect the signal strength profoundly. There are two basic types at high latitude, each having a separate cause and morphology. In its effect on radiowave propagation, auroral absorption (AA)

409

410

The high-latitude D region

Figure 7.49. Spectra of incoherent scatter (IS) and PMSE obtained with the 224-MHz EISCAT VHF radar. The left-hand panels show typical IS spectra fitted by Lorentzian curves, and the centre and right-hand panels show broad and narrow PMSE spectra. Even the broadest PMSE spectra are considerably narrower than the IS spectra from the same height. (Reprinted from P. N. Collis and J. Röttger. J. Atmos. Terr. Phys. 52, 569, copyright 1990, with permission from Elsevier Science.)

is of first-order importance. It may occur over a range of geomagnetic latitude from below 60° to above 75°, with a statistical maximum near 67°, and is patchy in its horizontal extent. The patches are tens to hundreds of kilometers in extent, and any elongation tends to be east–west. The diurnal occurrence peaks just before magnetic midnight and again in the morning sector between 0700 and 1000 magnetic LT. Most of our knowledge about high-latitude absorption has come from several decades of observation by standard riometers (having beams about 60° between half-power points), though the earliest studies were based on ionosonde data. This information probably describes AA sufficiently well for the purposes of those HF communication systems which also use relatively broad beams. However, some modern HF systems (such as over-the-horizon radars and direction finders) require information on the finer structure of D-region absorption. The imaging riometers developed during the 1980s (and further deployed in the 1990s) have improved the spatial resolution considerably, and have the potential to provide information relevent to the high-resolution HF systems. AA is a dynamic phenomenon, related, at least in part, to the auroral substorm; though almost certainly involving particle precipitation from the outer Van Allen belt in the day sector. The particles are electrons with energies from tens to hun-

7.7 References and bibliography

dreds of kilo-electron volts – generally greater than those which produce the visual aurora. As with the aurora, there is probably some measurable AA somewhere in the auroral zone in any given period of 24 h. AA is essentially conjugate, occurring almost simultaneously (though not necessarily with the same intensity) in magnetically conjugate regions. The other significant D-region absorption event at high latitude is polar-cap absorption (PCA), which may produce higher overall values of absorption than does AA but occurs much less frequently, only several times a year on the longterm average. PCA events are caused by the precipitation of 1–1000-MeV protons of solar origin into the polar D region. The occurrence and severity of PCA increases from solar minimum to maximum, and there may be ten or a dozen events in an active year. They produce a fairly uniform blanketing of the polar cap down to about 60° geomagnetic, and have been known to black out trans-polar HF propagation for 10 days at a time. Both AA and PCA affect the lower frequencies more than they do the higher ones because the absorption varies (to a first approximation) as f 2. At ELF and VLF, propagating in the waveguide mode, an increase in precipitation causes significant variation in the dimensions of the waveguide and thereby produces both amplitude and phase changes in the received signals.

7.6

References and bibliography

7.2

Auroral radio absorption

Agy, V. (1970) HF radar and auroral absorption. Radio Sci. 5, 1317. Ansari, Z. A. (1965) A peculiar type of daytime absorption in the auroral zone. J. Geophys. Res. 70, 3117. Appleton, E. V., Naismith, R., and Builder, G. (1933) Ionospheric investigations in high latitudes. Nature 132, 340. Bailey, D. K. (1968) Some quantitative aspects of electron precipitation in and near the auroral zone. Rev. Geophys. 6, 289. Berkey, F. T. (1968) Coordinated measurements of auroral absorption and luminosity using the narrow beam technique. J. Geophys. Res. 73, 319. Berkey, F. T., Driatskiy, V. M., Henriksen, K., Hultqvist, B., Jelly, D. H., Schuka, T. I., Theander, A., and Yliniemi, J. (1974) A synoptic investigation of particle precipitation dynamics for 60 substorms in IQSY (1964–65) and IASY (1969). Planet. Space Sci. 22, 255. Bewersdorff, A., Kremser, G., Stadnes, J., Trefall, H., and Ullaland, S. (1968) Simultaneous balloon measurements of auroral X-rays during slowly varying ionospheric absorption events. J. Atmos. Terr. Phys. 30, 591. Collis, P. N., Hargreaves, J. K., and Korth, A. (1984) Auroral radio absorption as an indicator of magnetospheric electrons and of conditions in the disturbed auroral Dregion. J. Atmos. Terr. Phys. 46, 21.

411

412

The high-latitude D region

Collis, P. N., Hargreaves, J. K., and White, G. P. (1996) A localised co-rotating auroral absorption event observed near noon using imaging riometer and EISCAT. Ann. Geophysicae 14, 1305. Ecklund, W. L. and Hargreaves, J. K. (1968) Some measurements of auroral absorption structure over distances of about 300 km and of absorption correlation between conjugate regions. J. Atmos. Terr. Phys. 30, 265. Elkins, T. J. (1972) A Model of Auroral Substorm Absorption. Report AFCRL-72-0413. Air Force Cambridge Research Laboratories, Bedford, Massachusetts. Foppiano, A. J. and Bradley, P. A. (1984) Day-to-day variability of riometer absorption. J. Atmos. Terr. Phys. 46, 689. Foppiano, A. J. and Bradley, P. A. (1985) Morphology of background auroral absorption. J. Atmos. Terr. Phys. 47, 663. Friedrich, M. and Torkar, K. M. (1983) High-latitude plasma densities and their relation to riometer absorption. J. Atmos. Terr. Phys. 45, 127. Friedrich, M. and Kirkwood, S. (2000) The D-region background at high latitudes. Adv. Space Res. 25, 15. Hajkovicz, L. A. (1990) The dynamics of a steep onset in the conjugate auroral riometer absorption. Planet. Space Sci. 38, 127. Hargreaves, J. K. (1966) On the variation of auroral radio absorption with geomagnetic activity. Planet. Space Sci. 14, 991. Hargreaves, J. K. (1967)Auroral motions observed with riometers: movements between stations widely separated in longitude. J. Atmos. Terr. Phys. 29, 1159. Hargreaves, J. K. (1968) Auroral motions observed with riometers: latitudinal movements and a median global pattern. J. Atmos. Terr. Phys. 30, 1461. Hargreaves, J. K. (1969a) Auroral absorption of HF radio waves in the ionosphere: a review of results from the first decade of riometry. Proc. Inst. Elect. Electronics Engineers 57, 1348 Hargreaves, J. K. (1969b) Conjugate and closely-spaced observations of auroral radio absorption – I. Seasonal and diurnal behaviour. Planet. Space Sci. 17, 1459. Hargreaves, J. K. (1970) Conjugate and closely-spaced observations of auroral radio absorption – IV. The movement of simple features. Planet. Space Sci. 18, 1691. Hargreaves, J. K. (1974) Dynamics of auroral absorption in the midnight sector – the movement of absorption peaks in relation to the substorm onset. Planet. Space Sci. 22, 1427. Hargreaves, J. K. and Chivers, H. J. A. (1964) Fluctuations in ionospheric absorption events at conjugate stations. Nature 203, 963. Hargreaves, J. K. and Sharp, R. D. (1965) Electron precipitation and ionospheric radio absorption in the auroral zones. Planet. Space Sci. 13, 1171. Hargreaves, J. K. and Cowley, F. C. (1967a) Studies of auroral radio absorption events at three magnetic latitudes. 1. Occurrence and statistical properties of the events. Planet. Space Sci. 15, 1571. Hargreaves, J. K. and Cowley, F. C. (1967b) Studies of auroral radio absorption events at three magnetic latitudes. 2. Differences between conjugate regions. Planet. Space Sci. 15, 1585.

7.7 References and bibliography

Hargreaves, J. K. and Ecklund, W. L. (1968) Correlation of auroral radio absorption between conjugate points. Radio Sci. 3, 698. Hargreaves, J. K., Chivers, H. J. A., and Axford, W. I. (1975) The development of the substorm in auroral radio absorption. Planet. Space Sci. 23, 905. Hargreaves, J. K. and Berry, M. G. (1976) The eastward movement of the structure of auroral radio absorption events in the morning sector. Ann. Geophysicae 32, 401. Hargreaves, J. K., Taylor, C. M., and Penman, J. M. (1982) Catalogue of Auroral Radio Absorption During 1976–1979 at Abisko, Sweden. World Data Center A, US Department of Commerce, Boulder, Colorado. Hargreaves, J. K., Feeney, M. T., Ranta, H. and Ranta, A. (1987) On the prediction of auroral radio absorption on the equatorial side of the absorption zone. J. Atmos. Terr. Phys. 49, 259. Hargreaves, J. K. and Devlin, T. (1990) Morning sector precipitation events observed by incoherent scatter radar. J. Atmos. Terr. Phys. 52, 193. Hargreaves, J. K., Detrick, D. L., and Rosenberg, T. J. (1991) Space-time structure of auroral radio absorption events observed with the imaging riometer at South Pole. Radio Sci. 26, 925. Hargreaves, J. K., Browne, S., Ranta, H., Ranta, A. Rosenberg, T. J., and Detrick, D. L. (1997) A study of substorm-associated nightside spike events in auroral absorption using imaging riometers at South Pole and Kilpisjärvi. J. Atmos. Solar–Terrestrial Phys. 59, 853. Hartz, T. R., Montbriand, L. E. and Vogan, E. L. (1963) A study of auroral absorption at 30 Mc/s. Can. J. Phys. 41, 581. Hartz, T. R. and Brice, N. M. (1967) The general pattern of auroral particle precipitation. Planet. Space Sci. 15, 301. Holt, O., Landmark, B., and Lied, F. (1961) Analysis of riometer observations obtained during polar radio blackouts. J. Atmos. Terr. Phys. 23, 229. Jelly, D. H., Matthews, A. G., and Collins, C. (1961) Study of polar cap and auroral absorption. J. Atmos. Terr. Phys. 23, 206. Jelly, D. H., McDiarmid, I. B., and Burrows, J. R. (1964) Correlation between intensities of auroral absorption and precipitated electrons. Can. J. Phys. 42, 2411. Jelly, D. H. (1970) On the morphology of auroral absorption during substorms. Can. J. Phys. 48, 335. Kavadas, A. W. (1961) Absorption measurements near the auroral zone. J. Atmos. Terr. Phys. 23, 170. Leinbach, H. and Basler, R. P. (1963) Ionospheric absorption of cosmic radio noise at magnetically conjugate auroral zone stations. J. Geophys. Res. 68, 3375. Little, C. G. and Leinbach, H. (1958) Some measurements of high-latitude ionospheric absorption using extraterrestrial radio waves. Proc. IRE 46, 334. Little, C. G., Schiffmacher, E. R., Chivers, H. J. A., and Sullivan, K. W. (1965) Cosmic noise absorption events at geomagnetically conjugate stations. J. Geophys. Res. 70, 639. Nielsen, E. (1980) Dynamics and spatial scale of auroral absorption spikes associated with the substorm expansion phase. J. Geophys. Res. 85, 2092.

413

The high-latitude D region

414

Parthasarathy, R. and Berkey, F. T. (1965) Auroral zone studies of sudden onset radio wave absorption events using multiple station and multiple frequency data. J. Geophys. Res. 70, 89. Parthasarathy, R., Berkey, F. T., and Venkatesan, D. (1966)Auroral zone electron flux and its relation to broadbeam radiowave absorption. Planet. Space Sci. 14, 65. Penman, J. M., Hargreaves, J. K., and McIlwain, C. E. (1979) The relation between 10 to 80 keV electron precipitation observed at geosynchronous orbit and auroral radio absorption observed with riometers. Planet. Space Sci. 27, 445. Pudovkin, M. I., Shumilov, O. I., and Zaitseva, S. A. (1968) Dynamics of the zone of corpuscular precipitations. Planet. Space Sci. 16, 881. Ranta, H., Ranta, A., Collis, P. N., and Hargreaves, J. K. (1981) Development of the auroral absorption substorm: studies of the pre-onset phase and sharp onset using an extensive riometer network. Planet. Space Sci. 29, 1287. Stauning, P. and Rosenberg, T. J. (1996) High-latitude daytime absorption spike events. J. Geophys. Res. 101, 2377.

7.3

The polar cap event

Akasofu, S.-I. and Chapman, S. (1972) Solar–Terrestrial Physics. Oxford University Press, Oxford. Bailey, D. K. (1959) Abnormal ionization in the lower ionosphere associated with cosmic-ray flux enhancements. Proc. IRE 47, 255. Bakshi, P. and Barron, W. (1979) Prediction of solar proton spectral slope from radio burst data. J. Geophys. Res. 84, 131. Castelli, J. P., Aarons, J., and Michael, G. A. (1967) Flux density measurements of radio bursts of proton-producing flares and nonproton flares. J. Geophys. Res. 72, 5491. Chivers, H. J. A. and Hargreaves, J. K. (1965) Conjugate observations of solar proton events: delayed ionospheric changes during twilight. Planet. Space Sci. 13, 583. Collis, P. N. and Rietveld, M. T. (1990) Mesospheric observations with the EISCAT UHF radar during polar cap absorption events: 1. Electron densities and negative ions. Ann. Geophys. 8, 809. Gillmor, C. S. (1963) The day-to-night ratio of cosmic noise absorption during polar cap absorption events. J. Atmos. Terr. Phys. 25, 263. Hargreaves, J. K., Ranta, H., Ranta, A., Turunen, E., and Turunen, T. (1987) Observation of the polar cap absorption event of February 1984 by the EISCAT incoherent scatter radar. Planet. Space Sci. 35, 947. Hargreaves, J. K., Shirochkov, A. V., and Farmer, A. D. (1993) The polar cap absorption event of 19–21 March 1990: recombination coefficients, the twilight transition and the midday recovery. J. Atmos. Terr. Phys. 55, 857. Hultqvist, B. (1969) Polar cap absorption and ground level effects. Solar Flares and Space Research (eds. C. de Jager and Z. Svestka), p. 215. North-Holland, Amsterdam. Jackman, C. H., Fleming, E. L., and Vitt, F. M. (2000) Influence of extremely large proton events in a changing stratosphere. J. Geophys. Res. 105, 11659.

7.7 References and bibliography

Leinbach, H. (1967) Midday recoveries of polar cap absorption. J. Geophys. Res. 72, 5473. Obayashi, T. (1959) Entry of high energy particles into the polar ionosphere. Rep. Ionosphere Space Res. Japan 13, 201. Ranta, H., Ranta, A., Yousef, S. M., Burns, J., and Stauning, P. (1993) D-region observations of polar cap absorption events during the EISCAT operation in 1981–1989. J. Atmos. Terr. Phys. 55, 751. Reagan, J. B. and Watt, T. M. (1976) Simultaneous satellite and radar studies of the Dregion ionosphere during the intense solar particle events of August 1972. J. Geophys. Res. 81, 4579. Reid, G. C. (1961) A study of the enhanced ionisation produced by solar protons during a polar cap absorption event. J. Geophys. Res. 66, 4071. Reid, G. C. (1967) Ionospheric disturbances. In Physics of Geomagnetic Phenomena (eds. Matsushita and Campbell), p. 627. Academic Press, New York. Reid, G. C. (1986) Solar energetic particles and their effects on the terrestrial environment. In Physics of the Sun (ed. P. A. Sturrock), vol. 3, p. 251. Reidel, Dordrecht. Reid, G. C. and Sauer, H. H. (1967) The influence of the geomagnetic tail on lowenergy cosmic-ray cutoffs. J. Geophys. Res. 72, 197. Sauer, H. H. (1968) Nonconjugate aspects of recent polar cap absorption events. J. Geophys. Res. 73, 3058. Shea, M. A. and Smart, D. F. (1977) Significant solar proton events, 1955–1969. In Solar–Terrestrial Physics and Meterology: Working Document II, p. 119. SCOSTEP. Shea, M. A. and Smart, D. F. (1979) Significant solar proton events, 1970–1972. In Solar–Terrestrial Physics and Meterology: Working Document III, p. 109. SCOSTEP. Shea, M. A. and Smart, D. F. (1995) Solar proton fluxes as a function of the observation location with respect to the parent solar-activity. Adv. Space Res. 17, 225. Smart, D. F. and Shea, M. A. (1989) Solar proton events during the past three solar cycles. Spacecraft and Rockets 26, 403. Smart, D. F. and Shea, M. A. (1995) The heliolongitudinal distribution of solar-flares associated with solar proton events. Adv. Space Res. 17, 113. Uljev, V. A., Shirochkov, A. V., Moskvin, I. V., and Hargreaves, J. K. (1995) Midday recovery of the polar cap absorption of March 19–21, 1990: a case study. J. Atmos. Terr. Phys. 57, 905. Weeks, L. H., CuiKay, R. S., and Corbin, J. R. (1972) Ozone measurements in the mesosphere during the solar proton event of 2 November 1969. J. Atmos. Sci. 29, 1138.

7.4

Coherent scatter and the polar mesosphere summer echo

Cho, J. Y. N. and Kelley, M. C. (1993) Polar mesosphere summer radar echoes: observations and current theories. Rev. Geophys. 31, 243. Collis, P. N. and Röttger, J. (1990) Mesospheric studies using EISCAT UHF and VHF radars: a review of principles and experimental results. J. Atmos. Terr. Phys. 52, 569. Czechowsky, P., Reid, I. M., Ruster, R., and Schmidt, S. (1989) VHF radar echoes

415

416

The high-latitude D region

observed in the summer and winter polar mesosphere over Andøya, Norway. J. Geophys. Res. 94, 5199. Ecklund, W. L. and Balsley, B. B. (1981) Long-term observations of the Arctic mesosphere with the MST radar at Poker Flat, Alaska. J. Geophys. Res. 86, 7775. Hoppe, U.-P., Hall, C., and Röttger, J. (1988) First observations of summer polar mesospheric back-scatter with a 224 MHz radar. Geophys. Res. Lett. 15, 28. Kelley, M. C., Farley D. T., and Röttger, J. (1988) The effect of cluster ions on anomalous VHF back-scatter from the summer polar mesosphere. Geophys. Res. Lett. 14, 1031. Palmer, J. R., Rishbeth, H., Jones, G. O. L., and Williams, P. J. S. (1996) A statistical study of polar mesosphere summer echoes observed by EISCAT. J. Atmos. Terr. Phys. 58, 307. Röttger, J. (1994) Polar mesosphere summer echoes: dynamics and aeronomy of the mesosphere. Adv. Space Res. 14, 123. Röttger, J., Rietveld, M. T., La Hoz, C., Hall, T., Kelley, M. C., and Swartz, W. E. (1990) Polar mesosphere summer echoes observed with the EISCAT 993-MHz radar and the CUPRI 46.4-MHz radar, their similarity to 224-MHz radar echoes, and their relation to turbulence and electron density profiles. Radio Sci. 25, 671.

Chapter 8 High-latitude radio propagation: part 1 – fundamentals and experimental results There cannot be a greater mistake than that of looking superciliously upon practical applications of science. The life and soul of science is its practical application Lord Kelvin

8.1

Introduction

Propagation of radio waves from ELF to UHF frequencies via the high latitude ionosphere is sometimes radically different from propagation at middle and low latitudes. This is primarily due to the fact that the magnetic field-lines at “corrected geomagnetic latitudes” greater than ⬃60° allow solar and magnetospheric particles and plasma to penetrate into the ionosphere. This results in the creation of many large-magnitude irregularities with scale sizes from meters to kilometers, most of which are aligned with the geomagnetic field in the auroral E and F regions. There are also sun-aligned arcs plus patches and blobs of ionization in the polar F region. Because of the extremely wide variation in ionospheric characteristics at high latitudes, this chapter contains many examples of actual propagation behavior. In contrast, it should also be mentioned that there is a wide spectrum of lessintense ionospheric irregularities in the mid-latitude ionosphere. Since most antennas used for communication and ionospheric sounding up until the 1960s had rather large antenna half-power beamwidths (typically 50° 50° in azimuth and elevation), these small irregularities were not observed. Starting in the early 1960s, several very-high-resolution HF backscatter sounders were constructed and employed in ionospheric research (see descriptions of the systems and results by Croft, 1968, and Hunsucker, 1991, Ch. 4). These systems revealed a plethora of echoes from irregularities, mostly of meter wavelengths. Hunsucker (1971), using a high-resolution HF sounder, found that irregularities of varying scale size and apparent motion were present in about 90% of the observations made during almost half a sunspot cycle in the mid-latitude ionosphere.

417

418

High-latitude propagation: 1

Starting in the late 1960s, several programs for prediction of HF ionospheric propagation were developed for “main-frame” computers, followed in the mid1970s by PC-based programs. These programs were intended to provide HFcommunications-circuit planners with median values of maximum, minimum, and optimum working frequencies as a function of the number of sunspots (or solar flux), time of day, season, path-length, and orientation. Further refinement of these programs made it possible to specify the type of antenna, transmitter power, receiver sensitivity, and receiver-location noise level. The actual prediction of “skywave” field strength has not turned out to be quantitatively accurate because of the difficulty of specifying mode structure, polarization loss, and non-deviative (and deviative) absorption loss (Hunsucker, 1992). Only two extant HFpropagation-prediction programs include high-latitude ionospheric effects, and they are either qualitative or have not been adequately validated to inspire users’ confidence. The effects of the polar cap and the auroral-oval ionosphere on signals of various frequencies differ substantially, and the morphology, phenomenology, and physics of these regions have been described in considerable detail in previous chapters of this book. (The fundamentals of the propagation of EM waves are described in Chapter 3, and the radio techniques for studying the high-latitude ionosphere have been described in Chapter 4.) The most profound effects of the high-latitude ionosphere on radio-wave propagation occur during geomagnetic storms and substorms (see Chapter 6). The emergence and proliferation of shortwave (SW) international broadcasting stations during the period 1930–1940 brought to the attention of some broadcasters the high unreliability of polar HF paths. Much of the history of high-latitude radio-propagation research has been summarized in several books and review papers (Rawer, 1976; Hunsucker, 1967; Hunsucker and Bates, 1969; Davies, 1990; Hunsucker, 1992). Other sources of research results on high-latitude HF propagation are in the Proceedings of the Ionospheric Effects Symposium (IES), held every three years in Alexandria, Vancouver, USA, and in the books of abstracts published at national meetings and general assemblies of the URSI. Research into high-latitude propagation began in earnest during the IGY and International Geophysical Cooperation (IGC) (1957–1959), which (fortuitously) coincided with the record-breaking sunspot maximum of cycle 19. During the period from the end of the Second World War through about 1975, it was thought by some that the best way to avoid the sometimes disastrous effects of auroral and polar ionospheric disturbances on high-latitude paths was simply to avoid these paths most of the time. Starting about 1975 there was a renewal of interest in studying high-latitude ionospheric effects because of the deployment of some sophisticated ELF/VLF and VHF/UHF satellite navigation and “over-thehorizon (OTH)” HF radar systems. During the “Cold War” (c. 1948–1991), both the USA and the USSR extensively deployed very sophisticated radio communi-

8.2 ELF and VLF propagation

cations and navigation systems in the Arctic regions, so considerable research was carried out in these areas of technology. Some of the research remains classified, but much was published in NATO and AGARD (Advisory Group for Aerospace Research and Development) conference reports (Landmark, 1964; Lied, 1967; Folkestad, 1968; Deehr and Holtet, 1981; Soicher, 1985). The use of rather sophisticated modulation techniques like frequency-shiftkeying (FSK), coded-pulses, frequency-hopping, and spread-spectrum on HF polar circuits has also prompted recent research on devising realistic atmospheric models, propagation-prediction techniques and rapid circuit sounding and switching (Goodman, 1992). The atmospheric density, temperature, composition, and dynamics from ground level up to ionospheric heights differ at high latitudes (sometimes drastically) from the values for mid-latitude and equatorial regions (see Chapters 1, 2, 5, 6, and 7). We will address the effects of these variations on specific frequency bands from ELF through UHF.

8.2

ELF and VLF propagation

Propagation in this part of the radio spectrum is best described and understood by invoking the Earth–ionosphere waveguide mode (Watt, 1967; Wait, 1970; Davies, 1970; Davies, 1990, Ch. 10) or the wave-hop (Berry, 1964) mode. The effectiveness of the “waveguide” mode depends upon the long- and short-term variations in conductivity of the Earth’s surface and the lower ionosphere (D region). VLF propagation, in general, is characterized by relatively low path attenuation (2–3 dB per megameter, where 1 Mm1000 km), is relatively stable with time, and the phase delay during propagation follows a predictable diurnal pattern. Propagation distances from 5000 to 20000 km are realized; however, atmospheric noise levels are high – thus decreasing the signal-to-noise ratio (SNR), and the signal bandwidths are low (20–150 kHz). Large antennas and high power transmitters are required to achieve a usable SNR at long distances. Because of the large wavelengths, it is economically and physically difficult to erect adequate antennas, so practical antennas have radiation efficiencies of 10%–20% – thus requiring high transmitter power. The OMEGA VLF navigation system is deployed globally and has for many years been an important and much utilized navigation aid. OMEGA operates at the low end of the VLF band (10–14 kHz) and is still used as a backup navigational aid, even with the advent of the GPS satellite navigational system. (See Ch. 10 of Davies, 1990 for further details of ELF–VLF–LF propagation.) The variation of phase speed for a perfectly conducting Earth is shown in Figure 8.1, and the actual phase variation of signals from WWVL at f20 kHz,

419

420

High-latitude propagation: 1

Figure 8.1. The variation of phase-speed with VLF frequency for a perfectly conducting Earth, sg $, r 2 105 for (a) mode number 1 and (mode number 2) (from Davies, 1990).

on a 113-km path is shown in Figure 8.2. At high latitudes, some irregularities in the lower ionosphere can influence VLF propagation (Wait, 1991). The lowest frequencies used for communication purposes are the ELF band (3–300 Hz), which is used primarily for very-low-data-rate communication to submerged submarines. One operational transmitter is located at the US Navy’s Wisconsin Test Facility and was described in the military literature as the “Project Sanguine/Seafarer/ELF.” The effective radiated power (ERP) is 0.25 W in the 40–50-Hz band and 0.5 W in the 70–80-Hz band. In practice, the ELF system is a “bell-ringer” that signals the submerged submarine to ascend to an appropriate depth to receive communication on VLF. For more details on ELF communication, see Bannister (1993) and Davies (1990, Ch. 10). At high latitudes, lower-ionospheric irregularities can effectively change the waveguide characteristics (Wait, 1970; Hunsucker, 1992). Fraser-Smith and Bannister (1998) recently measured ELF transmissions from a heretofore-unknown source, which they identified as a Russian ELF transmitter operating on 82 Hz, located on the Kola Peninsula at 69° N, 33° E. This signal was received as far away as Dunedin, New Zealand, which was the antipodal point (D16.5 Mm) and at Arrival Heights, Antarctica (D18 Mm). Figure 8.3 shows the average amplitude spectrum of lower ELF radio noise at the Sondrestrom receiving site in Greenland during January 1990.

8.2 ELF and VLF propagation

Figure 8.2. The diurnal variation of WWVL at 20 kHz over the 113-km path from Fort Collins to Wiggins, CO (from Davies, 1990).

421

422

High-latitude propagation: 1

Figure 8.3. The average amplitude spectrum of the lower ELF band. Note the Russian ELF transmission at 82 Hz and the and Bannister power-line frequencies (50 and 60 Hz) and their harmonics (from Fraser-Smith, 1998).

Theoretical and measured values of the Kola-Peninsula transmitter facility 82-Hz field strength versus distance are shown in Figure 8.4. Note the predicted and measured increases at antipodal distances. Another recent paper (Chrissan and Fraser-Smith, 1996) presents some new information on the noise-envelope amplitude-probability-distribution models of radio noise at VLF/ELF frequencies. Three noise models are used for comparison of data and the two which most closely describe the data are the “Hall” and the “-stable” models and the authors conclude that the -stable should be used in the polar regions, except at the peak of the diurnal and seasonal storm cycle. The effects of the high-latitude ionosphere on ELF signals during PCAs (SPEs) have been described using full-wave theory for the TEM mode with measurements made in the Gulf of Alaska. During the 23 November 1982 SPE event, a submarineborne receiver measured an unusually severe reduction in signal, which was attributed to lateral refraction bending the signal path away from the polar-cap boundary and into the central cap – where the phase velocity of the TEM mode is slowest. Figure 8.5 shows the geometry of the ELF path from the Western Test Facility to the

8.2 ELF and VLF propagation

Figure 8.4. Measured and theoretical values of the KPTF 82-Hz signal strength versus range ( 0°). CO, Connecticut; KB, King’s Bay, Georgia; SS, Søndrestrømfjord; HA, Hawai; DU, Dunedin; and AH, Arrival Heights.

Gulf of Alaska and Figure 8.6 illustrates the variation of the 76-Hz signal. ELF ray trajectories for weak, moderate, and strong SPEs on the path from the Western Test Facility to the Gulf of Alaska are shown in Figure 8.7. SPEs also have a profound effect on VLF polar transmissions, and one of the first documentations of these events was presented by Bates (1962), who described the effects of a relatively weak SPE on the VLF signal from England to Alaska. The 16.0-kHz signal from the GBR VLF station in Rugby, England was monitored at College, Alaska during the SPE event of 10 November 1961. Twenty minutes after the solar flare believed responsible, the GBR signal shifted phase by approximately 250° and the amplitude decreased by 20 dB over a 1-h period. During this event the diurnal variations of phase and amplitude increased in magnitude and changed markedly from normal patterns, and the effective height of the VLF waveguide over the polar cap dropped to about 5 km below the normal

423

424

High-latitude propagation: 1

Figure 8.5. A map showing the geometry of the ELF path from WTF to the Gulf of Alaska (from Field et al., 1985).

D-region height. Systems such as the US Navy’s OMEGA VLF network depend upon phase differences for their navigational positional accuracy, so polar ionospheric events can cause serious errors. Figure 8.8 is a map of the Rugby, England to College, Alaska VLF propagation path, the cosmic-noise absorption from the College, Alaska riometer for 10 November 1961 is shown in Figure 8.9, and the amplitude and corrected phase of the GBR transmissions are shown in Figure 8.10. The results of a three-year study of VLF propagation (during sunspot minimum) monitored at College, Alaska have been reported by Bates and Albee (1965) and Albee and Bates (1965). During that period, 1846 optically detected solar flares were observed on sunlit paths, of which 66 produced phase anomalies on the NBA (non-polar) path. Table 8.1 lists the frequencies of the VLF stations monitored during this study and Figure 8.11 is a map showing the propagation paths. It can be seen that only the paths from GBR and NAA can truly be called high-latitude paths, but, during major SPEs, small portions of the other paths may be affected by the boundary of the PCA, as was the ELF transmission noted previously. Some typical navigation-location errors measured during the SPE of 6–9 March 1970 (3.8 dB maximum), are shown in Figure 8.12. Documentation on the behavior of ELF/VLF signals on polar paths during major SPEs (30 MHz absorp-

8.2 ELF and VLF propagation

Figure 8.6. The 76-Hz signal received in the Gulf of Alaska (from Field et al., 1985).

Figure 8.7. Ray trajectories for the three SPE strengths (r2 Mm) (from Field et al., 1985).

425

Figure 8.8. Map showing the 16.0-kHz propagation path from Rugby, UK to College, Alaska (after Bates, 1961).

8.2 ELF and VLF propagation

427

ATTENUATION (dB)

1.5

1.0

0.5

0 14

16

18

20

22 00 UNIVERSAL TIME

02

04

06

Figure 8.9. Cosmic-noise absorption from the 27.6-MHz College riometer (after Bates, 1961).

Figure 8.10. The amplitude and corrected phase of the GBR 16.0-kHz signal received at College, Alaska on 10 November 1961 (after Bates and Albee, 1966).

Table 8.1. A list of VLF stations monitored at College, Alaska from 1961 to 1964 Station

Frequency (kHz)

Location

Period recorded

NBA GBR NAA NPM NPG WWVL

18.0 16.0 Various 19.8 Various 20.0

Balboa, Panama Rugby, England Cutler, Maine Hawaii Jim Creek Fort Collins, Colorado

August 1961–Dec 1963 October 1961–1964 November 1962–1964 April 1962–1964 April 1963–December 1963 January 1964–December 1964

Figure 8.11. VLF propagation paths to College, Alaska during the 1961–1964 study (after Bates and Albee, 1966).

8.3 LF and MF propagation

HYPERBOLIC NORWAY–HAWAII) 2

429

RANGING (NORWAY)

30µs = 3 nm HYPERBOLIC 6 nm RANGING

1

LOCATION ERROR (nm)

0 CORRECTED

1 2 3 4 5 6

ACTUAL

7 8 9 0

6

12 18 24 6 MAR

6

12 18 24 7 MAR

6

12 18 24 8 MAR

6 12 18 24 9 MAR

6

12 18 24 10 MAR

1970

Figure 8.12. OMEGA location errors in nautical miles during the SPE of 6–9 March 1970 in Norway (from Larsen, 1979).

tion 10 dB) is difficult to find, but they should produce profound phase and amplitude variations. The DECCA navigational system, which utilizes frequencies from 70 to 100 kHz, is designed for high accuracy over medium ranges and depends upon the groundwave for its accuracy, so we are not very concerned about high-latitude ionospheric effects. Another hyperbolic radio navigation system is the LORAN-C global network, operating on 100 kHz, which also depends upon the groundwave for its accuracy. Some LORAN-C systems developed in the 1980s augmented the receiver by skywave signals in addition to groundwave and there were some indications that errors of up to 20 km occurred during geomagnetic disturbances (Hunsucker, 1992)

8.3

LF and MF propagation

The basic propagation modes for LF through MF (⬃300 kHz to 3 MHz) are groundwave at all hours, augmented by skywave modes at night. Groundwave propagation covers ranges of ⬃1 km to several hundreds of kilometers from the transmitter, with extended ranges over sea water and erratic results over mountainous terrain. The discontinued LORAN-A navigation system was a hyperbolic

430

High-latitude propagation: 1

line-of-position system, which was primarily groundwave propagation but also was sometimes affected by skywave. Another primarily groundwave navigational system (which is currently being phased out) was the Non-Directional-Beacon (NDB) system operating in the 250–450-kHz band, but again there do not seem to be documented examples of high-latitude ionospheric effects . . . In the MF band (300 kHz to 3 MHz) – the US standard AM broadcast band is 550 kHz to 1570 kHz – station frequencies (channels) are assigned on a groundwave and skywave non-interference basis for each 10 kHz channel. In the continental USA one is rarely out of groundwave range of several nearby commercial broadcast stations and at night-time additional stations from distances of 1000 km and greater are heard. From ⬃1.6 to 3.0 MHz various land and maritime navigational and fixed services operate using the skywave mode but reliable propagation is difficult because of D-region absorption, which is particularly high at auroral and polar latitudes. As an example, most of the fixed-service communication services in Alaska at these frequencies have been discontinued. In the northern tier of the USA it was necessary to modify the FCC mid-latitude frequency assignment procedure to allow for auroral E-layer anomalouspropagation modes. A five-and-a-half-year investigation (for one half of solar cycle 21) of skywave transmissions from clear-channel 50-kW standard broadcast stations in the USA, including Alaska, and Canada received at Fairbanks, Alaska was reported by Hunsucker et al. (1988). Some results of this investigation are the following. The site was located at the Ace Lake Field Site of the Geophysical Institute of the University of Alaska-Fairbanks at geographic coordinates of 64° 52 N latitude, 147° 56 W longitude at a north geomagnetic latitude of 64° 45 and a dip of 76° 54. The receiving/recording system was built around a commercial generalpurpose receiver modified for analog automatic-gain-control output and the receiver frequency was automatically stepped through 16 channels every 5 min by the system programmer. Digital tape-cassette recordings of signal amplitude were continuously made on ten or more standard broadcast stations, then the data were transferred to standard-format computer tape for analysis on a VAX 11/780-785 computer. A noise source was also recorded continuously for regular system calibration and occasional aural checks were made to insure that the proper identification of individual stations was achieved. The three different antennas used during this program were carefully calibrated against each other on standard broadcast-band groundwave and skywave transmissions. Other details of the instrumentation are given in Hunsucker et al. (1987; 1988). Figure 8.13 is a plot of sunspot-cycle variation during the course of the Alaska MF experiment, with the average monthly number of sunspots varying from 140 to 20. The range of geomagnetic activity was from Kp 0 to Kp 9, including the largest geomagnetic storm previously recorded at the College, Alaska observatory (8–9 February, 1986).

8.3 LF and MF propagation

431

200

180

Solar Cycle 21 Beginning June 1976

LEGEND = Observed Smoothed = Predicted Smoothed

160

Smoothed Rz

140 Solar Cycle 22 Beginning September 1986

120

100

80

60

Mean of Cycles 8–20

40

20

June 1976

June 1977

June 1978

June 1979

June 1980

June 1981

June 1982

June 1983

June 1984

June 1985

June 1986

June 1987

June 1988

June 1989

June 1990

Figure 8.13. Solar-cycle variation during the period of the Alaska MF experiment.

Because of the comprehensive nature of the Alaska MF data set (MF skywave signal strengths on paths inside, tangential to, and transverse to the auroral oval for a wide range of numbers of sunspots), we will present some of the salient results. Table 8.2 lists the Fairbanks MF channel assignments for 1985. There was no such thing as a “typical daily variation” for any of the MF signals received at Fairbanks during this experiment because of the pronounced seasonal, sunspot-cycle, and ionospheric storm (auroral) effects. To illustrate the effects of auroral disturbances on the daily variations in MF skywave signal strength, Figures 8.19(a)–(c) show the variations in signal strength and auroraloval locations for selected days near the Fall equinox of 1985 (see page 000). Specifically, Figure 8.14 shows typical variations in signal for quiet magnetic conditions. The equinoctial recovery from the summer low field strength is quite apparent. Figure 8.15 is a plot of the auroral oval at 1000 UT on 4 September, 1985, coinciding with the peak diurnal signal strength in Figure 8.14. Note that the auroral oval is well poleward of any of the propagation paths monitored at Fairbanks. The greater variability in signal associated with higher local magnetic activity (College Ak 20) is shown in Figure 8.16 and the auroral oval for 1000 UT is presented in Figure 8.17, showing its equatorward expansion south of Fairbanks. It should also be remembered that the AA region extends 1°–2° equatorward of the

432

High-latitude propagation: 1

Table 8.2. Fairbanks MF receiver channel assignments for 1985a (from Hunsucker, 1988) Channel Frequency (kHz) Station 0 1 2 3 4 5 6 7 8 9 10 11 12 13c 14 15

450 1000 450 750 1260 1030 1100 450 1260 450 450 870 750 1170 720 1510

b

Noise-diode calibrator b

KFQD, Anchorage CFRN, Edmonton KTWO, Casper KFAX, San Francisco b

CFRN, Edmonton b b

KSKO, McGrath KFQD, Anchorage KJNP, North Pole, Alaska KOTZ, Kotzebue KGA, Spokane

Notes: a The top-loaded vertical antenna (TLVA) was utilized for the entire year. b Channel programmed to a quiet frequency, not active at this time. c Channel assignment changed to KJNP from CHU Ottawa on 3 July 1985.

visual auroral oval. The variability of the MF signal strengths is most probably due to the increase in AA and sporadic-E ionization associated with the auroral oval. Figure 8.18 shows the MF signal behavior for a quite disturbed day (College, Ak 46) with extreme signal variation. The auroral oval extends well equatorward of Fairbanks and probably affects all paths monitored. The channel-3 (Anchorage, Alaska) path lies entirely inside the auroral oval and its extreme variability is probably due to intense “patches”of sporadic-E ionization. Signals on channels 4 and 5, Edmonton, Alberta (Canada) and Casper, Wyoming, respectively, are from paths passing obliquely through the auroral oval and show profound absorption effects. The KFAX, San Francisco path (channel 6) is roughly perpendicular to the auroral oval and its ionospheric reflection points are mainly equatorward of the oval, so it is affected less than are channels 3–5. Channels 11 and 14 (McGrath and Kotzebue, both in Alaska) behave similarly to Anchorage because the paths lie entirely inside the auroral oval. Channel 13 was

8.3 LF and MF propagation

Figure 8.14. Variations in MF-signal strength for broadcast stations monitored in Fairbanks for a quiet equinoctial day in 1986. Fairbanks local time (150° Western Meridian Time) is 10 h less than UT (after Hunsucker, 1988).

Figure 8.15. The location of the auroral oval at 1000 UT on 4 February 1986 Q 0. (after Hunsucker, 1988).

433

434

High-latitude propagation: 1

Figure 8.16. Signal behavior during a moderately disturbed equinoctial day (Ak 20), 8 September 1986 (from Hunsucker, 1988).

Figure 8.17. The location of the auroral oval for a moderately disturbed day, 8 September 1985, at 1000 UT, Q4 (from Hunsucker, 1988).

8.3 LF and MF propagation

Figure 8.18. MF-signal behavior during a disturbed day (after Hunsucker, 1988).

programmed to receive the groundwave signal from a local 50-kW station, KJNP, but, when the station was “off the air” (0830–1330 UT), signals from an unknown AM station were intermittently received. The sunspot cycle also exerts profound effects on the MF skywave-signal strengths measured at Fairbanks, depending, of course, on the frequencies, the path-lengths, and the orientations relative to the auroral oval. Tables 8.3 and 8.4 show the sunspot-cycle effects on four selected paths. The seasonal behavior of MF skywave signals received at Fairbanks for 1985 (sunspot-minimum year) is shown in Table 8.5, from which it may be seen that, except for one or two exceptions, the highest signal strengths occurred in the winter and the lowest signal levels occurred in the summer, with intermediate values during the equinoxes. Effects of the great geomagnetic storm of February 1986 on MF skywave reception at Fairbanks were documented by Hunsucker et al. (1987) and, since this was probably the most systematic investigation, we will present some of the salient effects. The magnetic storm of 8–9 February, 1986 was one of the largest for the previous 40 years and especially dynamic at high latitudes. The College USGS Observatory measured an H-component maximum excursion of 6110 nT and the local and planetary K indices were 9 for several hours on 8 February, 1986. The

435

High-latitude propagation: 1

436

Table 8.3. Characteristics of four selected MF propagation paths (from Hunsucker, 1988)

Frequency (kHz)

Power output (KW)

Call letters

Location

Path-length and remarks

KSKO

McGrath, Alaska

870

5

KTWO

Casper, Wyoming

1030

50

D3553 km, long path. One ionospheric reflection point in the auroral oval during moderately disturbed conditions.

KFAX

San Francisco, California

1100

50

D 3464 km, long path. One ionospheric reflection point in the auroral oval during disturbed conditions.

KGA

Spokane, Washington

1510

50

D 2640 km. Similar to the KTWO path.

Short north–south auroral path.

Table 8.4. Sunspot-cycle effects – a comparison of changes in signal strength on four paths from 1981 to 1985 (midwinter) (from Hunsucker, 1988) 1981 (Average relative international sunspot number 147)

1985 (Average relative international sunspot number 12)

Station

Signal present (%)a

Signal maximum (V)b

Signal present (%)a

Signal maximum (V)b,c

Signal present (%)a

Signal maximum (dB)b,c

KSKO KTWO KFAX KGA

53 30 62 50

7 8 9 7

75 54 71 46

60 8 70 8

22 24 8 4

18.7 0 17.8 1.2

a

Percentage of operating period when signal was present. All signal levels are referred to receiver input. c See the text for a discussion of the increase in signal. b

Increase in signal strength 1981–1985

8.3 LF and MF propagation

437

most pronounced effects were on the 1984-km path from the 50-kW station, CERN, in Edmonton, Alberta at 1260 kHz. Figures 8.19(a), (b), and (c) are maps showing the Edmonton–Fairbanks propagation path in relation to the auroral oval for periods before, during, and after the storm and the amplitude of the signal is displayed just below each map. Relatively normal night-time skywave propagation is seen in Figure 8.19(a), two days before the storm when the auroral oval was poleward of the path. During the maximum phase of the storm – shown in Figure 8.19(b) – there was almost complete absorption on the path. Three days after the storm, the skywave signal had almost returned to its pre-storm level (Figure 8.19(c)). Some of the field-strength measurements collected at Fairbanks have been compared with field strengths predicted by various methods and the full results have been published in FCC Rule Change Docket 20 642. Table 8.5 shows some selected examples of comparisons between measured and modeled values. Some conclusions of the Alaska MF study are as follows. (1)

The “high-end” commercial electronically scanned receiver, noise calibration, and digital data-recording systems worked exceptionally well during the five and a half years of the experiment.

(2)

An absolutely necessary requirement for a program of this sort is regular careful aural monitoring in order to positively identify the transmitters.

(3)

The selection of a “radiofrequency-interference-quiet” remote receiving site in Alaska produced excellent high-SNR data.

(4)

When the MF skywave propagation paths traversed the auroral oval there were profound variations in signal as a function of frequency, geomagnetic activity, time of day, and season.

Table 8.5. Measured and predicted field strengths for 1987 Median field strength for 1987 dB (1 V m1) Method of prediction

Path 1

Path 2

Measured FCC curve (Also used by region 2) Cairo curve CCIR method (Recommendation 435) Modified FCC method

26.8 33.2 40.2 16.2 27.7

34.7 54.8 55.0 54.4 44.1

Notes: Measured values are for the sixth hour after sunset at the mid-point of the path. Path 1 – San Francisco to Fairbanks, 3464 km, KFAX, 1100 KHz, 50 kW. Path 2 – Anchorage to Fairbanks, 431 km, KFQD, 750 kHz, 10 kW.

(b)

(c)

Figure 8.19. Behavior of the MF signal from Edmonton, Alberta and Fairbanks, Alaska before, during and after the great geomagnetic storm of 8–9 February 1986 (after Hunsucker et al., 1987).

(a)

8.4 HF propagation

(5)

8.4

These results prompted the FCC to issue new engineering skywave curves describing possible skywave interference between standard AM broadcasting stations in the northern tier of the USA, including Alaska, and Canada, thus making channel assignments more realistic.

HF propagation

The ITU HF band (3–30 MHz) is basically a skywave band day and night and is used for broadcasting, point-to-point, and surveillance (actually, the range of 2–30 MHz is primarily propagated by skywave). At mid-latitudes the average characteristics of HF propagation are reasonably predictable, except during geomagnetic storms. Fortunately, there are several books describing basic HF propagation (Maslin, 1987; McNamara, 1991; Davies, 1990, Ch. 6; Goodman, 1992) for those wanting more detailed accounts. 8.4.1

Tests carried out between Alaska and Scandinavia on fixed frequencies

Serious and methodical investigations of the behavior of trans-polar HF propagation on paths between Scandinavia and Alaska were initiated in the mid-1950s. Although most of the early CW transmissions were degraded by SW interference, subsequent transmissions utilized pulses, which were much more resistant to the SW interference. Up until about 1969, the results of most of the trans-polar HF propagation experiments were published in institutional reports, not in the “open literature,” and, because of the importance of these data, we will present selected extracts from these experiments starting in 1956. It was fortuitous that the calibrated pulsed HF trans-polar transmissions began just before the maximum of sunspot cycle 19 – the highest maximum on record! The following results are extracted from a report by the Geophysical Institute of the University of Alaska (Owren et al. 1959) and represent HF propagation conditions near the maximum of sunspot cycle 19. Early in 1956 (sunspot number (SSN) ⬵50) the Norwegian Research Establishment (NDRE) and the UAF Geophysical Institute agreed to cooperate in a program of test transmissions across the north polar region in order to investigate the propagation conditions. The first propagation test was made using a 3-kW CW transmission and a FSK teletype signal on 3.3 and 7.7 MHz from Fairbanks, Alaska and a 5-min h1 transmission of a 5-kW CW signal on 5.9 MHz from Harstad in northern Norway. In addition, the receiver stations in Alaska were to monitor the 100-kW broadcast transmissions on 629 kHz from Vigra in southern Norway. Receiver stations were set up at College and Barrow in Alaska and at Harstad, Norway, as well as on west Spitzbergen, Svalbard. As the

439

440

High-latitude propagation: 1

test progressed, modifications to the original plan had to be made. The 3.3-MHz transmission from Fairbanks had to be canceled because of interference with other services. The Norwegian receiver stations were unable to pick up the 7.7MHz transmission and on 12 July this was replaced by a pulsed transmission on 12.3 MHz beamed from College to northern Norway. This signal was immediately picked up and identified by the Spitzbergen station, illustrating the advantage of pulsed transmissions. The College receiver station was unable to identify the 5.9-MHz transmission from Harstad, but Barrow succeeded after coming into operation on 13 July. The signal was never received well, even at Barrow. Completely negative results were obtained regarding the Vigra MF transmissions, both at Barrow and College A supplementary program for monitoring Norwegian and Russian MF and HF broadcast transmitters in the frequency range 0.5–22 MHz was put into effect at College on 6 July and at Barrow on 13 July. Good results were obtained for the Norwegian SW transmissions at 17.825 MHz from Frederickstad in southern Norway. The July 1956 test showed clearly the superiority of pulsed signals over FSK and CW types of transmission and further indicated that future tests should be concentrated on frequencies in the HF band. Several other monitoring tests were carried out during 1956 and 1957, with rather inconclusive results, but the fourth and fifth tests in January and February 1958 (SSN200.9) proved to be more successful. During this part of the IGY a three-frequency HF backscatter sounder operating on 12, 18, and 30 MHz was located at College, Alaska (Peterson et al., 1959). This backscatter sounder had three three-element Yagi antennas mounted on a single rotating mast with transmitter pulse outputs of 4 kW (the antenna rotated at 1 RPM). Another pulse transmitter at College also operated on 6 MHz using a halfwave dipole antenna. The Geophysical Observatory at Kiruna, Sweden participated in the January–February 1958 tests, with encouraging results. Specifically, it was found that the 12-MHz signals could be picked up even when the antenna was rotating, in fact, the pulse emission from College could be received throughout the rotation cycle. Later it was found that the 18-MHz pulsed transmission could similarly be received over half the rotation cycle. The 30-MHz signals were also found to be detectable at Kiruna, but intermittently rather than regularly. The Kiruna Geophysical Observatory thereafter set up a program of continuous monitoring of the College pulsed transmissions on 12, 18, and 30 MHz starting in May 1958. The College transmissions on 12, 18, 24, and 30 MHz were recorded at Kiruna utilizing a rhombic antenna connected to a communication receiver modified for pulse reception. The receiver output was displayed on an oscilloscope and recorded photographically. We will include many examples of HF signal behavior over paths of various lengths, at various frequencies with various orientations with the auroral oval and

8.4 HF propagation

Figure 8.20. The College–Kiruna propagation path (D5300 km) (after Owren et al., 1959).

polar-cap ionosphere over a wide range of geophysical activity in order to fully illustrate the extreme variability of high-latitude HF propagation. Analysis of simultaneous College HF backscatter and the signal received in Kiruna revealed that the three-hop mode (not the two-hop mode) was predominant on the 5300-km path illustrated in Figure 8.20.

HF trans-polar propagation data for the maximum of sunspot cycle 19 In total 672 h of simultaneous recordings of received signal strengths at Kiruna and groundscatter echoes observed at College for the month of December 1958 were analyzed. Approximately 30% of the 672 h of data was lost due to SW interference at the frequencies being used and the usual equipment failures. It was hoped that groundscatter observed on the HF propagation paths would be a good indicator of forward-propagation conditions, so groundscatter observed from College appearing within 30° of the Kiruna azimuth in the 1000–1900-km range was interpreted as the first hop of a three-hop mode. Similarly, groundscatter in this direction in the 2000–3000-km range was considered as the first hop of a twohop mode.

441

442

High-latitude propagation: 1

Figure 8.21. A comparison of the College, Alaska signal received at Kiruna and College groundscatter for 18 MHz on 4 December 1958 (after Owren et al., 1959).

Figure 8.22. The average signal strength at Kiruna, Sweden for the month of December 1958 (from Owren et al., 1958).

8.4 HF propagation

443

Table 8.6. The relative occurrence of propagation modes and groundscatter

Propagation mode Three-hop Two-hop No indication (polar groundscatter absent)

Percentage of time during which mode occurred 12 MHz

18 MHz

65 11 24

61 15 24

Periods of high signal strength at Kiruna were sometimes observed during the interval 07–18 UT when there were no groundscatter echoes in the direction of propagation. This could indicate that the echoes were present but were below the sensitivity threshold of the receiver or that a one-hop Pedersen propagation mode was operative. The histogram for 18 MHz in Figure 8.21 illustrates this condition during the interval 12–15 UT (for SSN180.5). When the College transmissions were “readable” at Kiruna (the photographic records of the signals were scaled in arbitrary units from zero to three and a “readable” signal is defined as one of strength &0.5), groundscatter echoes from the polar region indicated the relative occurrence of the following propagation modes (Table 8.6). As a result of this and other groundscatter–signal-strength comparisons, it was concluded that groundscatter was not a very good indicator for the propagation of HF signals at high latitudes. The histogram in Figure 8.22 shows the average signal strengths at Kiruna for the month of December 1958 for 12 and 18 MHz. The pronounced dip at 1500 UT in the 12- and 18-MHz histograms occurs during the period of maximum interference at both Kiruna and College. The diurnal maximum of D-region absorption in the region north of College also occurs during this interval. Favorable circumstances made the month of August 1959 (SSN151.3) particularly suitable for a detailed study of the effects of solar-particle precipitation and radiation on high-latitude HF propagation. First, the solar events occurred after a quiet period with an unusual distinctiveness and included both low- and high-energy particle precipitation. Secondly, comprehensive geophysical observations obtained during the IGY were available, including radiation measurements made by Explorer VI, absorption measurements from an extended chain of stations, and good coverage of the arctic by ionosonde. Thirdly, a network of arctic and subarctic experimental HF circuits was in operation through the joint efforts of the University of Alaska Geophysical Institute, the Kiruna Geophysical

444

High-latitude propagation: 1

Observatory, Sweden, and the Radioscience Laboratory of Stanford University. The entire program was sponsored by the Electronics Research Directorate of the US Air Force Cambridge Research Laboratories, Massachusetts. Historically then, this was probably the first extensive, coordinated, well-instrumented HF propagation experiment at high latitudes during a disturbed period. The experimental circuits utilized the Stanford IGY backscatter sounders for 12, 18, and 30 MHz located at College, Alaska, Thule, Greenland, and Stanford, California. The receiving stations were in operation at Kiruna, Sweden on the distant side of the arctic region and at Boston, College, and Stanford on the North-American continent. We consider here only the 12- and 18-MHz pulse transmissions from College, Alaska – since the College receiver station for the Thule and Stanford transmissions was not operational in August 1959. The College sounder transmitted 1-ms pulses at 18.75 pulses s1 with a peak power of 4–5 kW, using the rotating Yagi-antenna system. The College–Kiruna great-circle path is a trans-polar circuit of length 5300 km passing within a few degrees of the north geographic pole and essentially inside the auroral zone – considering the low radiation angle (about 10°) and the ionospheric reflection points. The College–Boston great circle path is 5300 km long and crosses the auroral zone tangentially. The College–Stanford path is 3500 km long and lies outside the auroral zone under normal conditions. Figure 8.23 is a map showing the propagation paths and supporting ionospheric observations. The first 13 days of August 1959 were characterized by low solar activity and magnetically quiet conditions, in particular the days of 11–14 August. On 14 August and again on 18 August there occurred major solar flares in an active region that crossed the solar central meridian on the 16th. Many lesser flares were observed in this region during its passage over the Sun’s disk. The two major flares were both followed by geomagnetic storms that together account for the 5 days of the month designated as magnetically disturbed. The second flare was also accompanied by cosmic-ray emission, causing a weak (⬃1-dB) PCA. The first major flare, of importance 2, occurred on 14 August at 0040 UT and was accompanied by a sudden cosmic-noise absorption and a gradual SW fadeout (SWF). A severe sudden-commencement magnetic storm, which must be assumed to have been due to the low-energy particle emission associated with the 2 flare, started on 16 August at 0404 UT and lasted about 40 h, until the evening of the 17th. Riometer observations indicated that emissions of energetic particles occurred between geomagnetic latitudes of 52° and 62° with peak intensity around 57° or 58°. Thus, the initial impact of the auroral particles was in subauroral latitudes and the severe magnetic storm typically caused a southward shift and expansion of the zone of activity. Following the initial phase of the magnetic storm, AA was consistently greater at College than it was at King Salmon, Alaska during the remainder of the storm. This confirmed the findings of Basler, namely that AA peaks a few degrees equatorward of the visual auroral oval, which finding is of considerable importance to studies of auroral HF propagation.

8.4 HF propagation

Figure 8.23. A map of propagation paths showing the auroral zone based on Vestine’s isochasms (after Owren et al., 1963).

The second major flare, of importance 3, started on 18 August at 1015 UT and was accompanied by a SWF as well as by solar-noise emission, as is evident from the Thule riometer record. The Thule measurements show that a weak PCA event started at about 1200 UT. The PCA reached a maximum of 1.5 dB at 27.5 MHz on 19 August at about 2000 UT, and the recovery was completed by 21 August at 1200 UT. A sudden-commencement magnetic storm, of moderately severe intensity, started on 20 August at 0412 UT and lasted into the forenoon of the 24th. The solar–terrestrial events of August 1959 and the radio propagation behavior on several trans-polar paths are shown in Figure 8.24. In general, it is seen that peaks in AA were the most important factor explaining decreasing signal strength on most of the HF paths. It should be emphasized that this was a very small PCA. The behavior of HF propagation on the three

445

(a)

(b)

Figure 8.24. Solar–terrestrial conditions and transpolar HF radio propagation for near-sunspotmaximum conditions on 15 and 18 August 1959 (from Owren et al., 1963).

High-latitude propagation: 1

448

Figure 8.25. A map showing three great-circle HF-propagation paths and the normal and expanded auroral zones (after Owren et al., 1963).

circuits shown in Figure 8.25 during the magnetic storm of 16–17 August 1959 is illustrative of near-sunspot-maximum conditions. 8.4.2

Tests involving transmission between Alaska and the continental USA The College–Stanford circuit (basically a mid-latitude path, D3500 km)

During the undisturbed periods of August, the 12-MHz transmission was received with consistently high signal strength at all hours of the day at Stanford. The behavior of the circuit is illustrated in Figure 8.26, a contour plot of signal outage

8.4 HF propagation

Figure 8.26. A contour plot of College-to-Stanford (11.634-MHz) signal outage in August 1959 (from Owren et al., 1963).

in which the cross-hatched area indicates when the signal strength was less than 6 dB above 1 V. The plot shows that normally there were no diurnal outage periods. The groundscatter observed simultaneously at College indicated a two-hop F-layer mode. This implies that the propagation path traverses the D layer at about geomagnetic latitudes of 64°, 57°, 55°, and 45° N. A sudden circuit black-out started at the time of onset of auroral absorption at King Salmon, Alaska (geomagnetic latitude 57.4°). There were some temporary recoveries during the storm, which appear to be reasonably well related to the decreases in absorption at King Salmon if some allowance for longitude difference were made. The final recovery took place on 18 August at 0900 UT. There were no 18-MHz data due to interference during this period.

The College–Boston circuit (tangential to the auroral oval, D5300 km) The 12-MHz circuit had a diurnal black-out of duration about 5 h during the undisturbed period 7–15 August. The ionospheric data indicated that the signal outage was controled by AA of the outgoing signal near College. Consistently, the major black-out during the storm period started on 16 August at 1000 UT, 4 h after the College–Stanford 12-MHz blackout, as the ionospheric disturbances spread to the normal auroral-zone regions.

449

High-latitude propagation: 1

450

24

EASTERN STANFORD TIME

20

16

12

08

04

00 1

3

5

7

9

11

13

15

17

19

21

23

25

27

29

31

AUGUST

Figure 8.27. A contour plot of College-to-Boston (17.900-MHz) signal outage in August 1959 (after Owren et al., 1963).

The 18-MHz signal at Boston was generally sub-marginal except during the very quiet pre-storm period 13–15 August. The signal-outage contour plot (Figure 8.27) and the College absorption-contour plot (Figure 8.28) show a striking similarity during the pre-storm period. It is probable that this circuit was controled more by AA near Churchill, Canada than by MUF factors. The circuit suffered a complete black-out during the 16–17 August storm, as might have been expected. 8.4.3

Other trans-polar HF experiments on fixed frequencies The College–Kiruna circuit (trans-polar, D5300 km)

Both the 12- and the 18-MHz signals were received at Kiruna most of the time during 1–12 August, with very little outage during 13–15 August. This great-circle path passes over west Spitzbergen and the ionosonde observations at Longyearbyen show that, during 1–15 August, the 12-MHz signal was nearly the optimum trans-polar traffic frequency, whereas the arctic ionosphere could not at any time support a conventional, multihop propagation mode at 18 MHz.

8.4 HF propagation

Figure 8.28. A contour plot of absorption at College. The crosshatched sections denote absorption values exceeding 1 dB and the solid black areas indicate PCA events (from Owren et al., 1963).

During the 16–17 August storm the 12-MHz signal weakened somewhat but remained essentially receivable at Kiruna. The 18-MHz signal blacked out early, at 2240 UT on 15 August as the increasing magnetic activity reached a Kp of 5, and remained essentially out until midday on the 21st. The previous winter (the solar-cycle-19 maximum of 1958–1959) exhibited unusually favorable propagation conditions on the propagation path between College and Kiruna. The HF pulse reception at Kiruna showed, as a rule, three different propagation modes on 18 MHz, with intervals of 3 and 6–7 ms, respectively, which could hardly be explained as alternate modes on the great-circle path. Thus, there was some indication that non-great-circle propagation modes might be available at certain times – which might arise from sharp horizontal gradients in electron density from the polar cap equatorward though the auroral-oval ionosphere. In order to test this hypothesis, a so-called “pinwheel” experiment was employed at Kiruna, featuring a specially designed rotating three-element Yagi antenna for 18 MHz. The antenna was moved step-wise from 60° NEE over geographic north to 60° NWW and back, stopping in each indicated position for 1 min (the horizontal beamwidth was estimated to be 60°). The backscatter sounder

451

High-latitude propagation: 1

452

in Alaska transmitted with a higher pulse-repetition frequency (PRF) during the minute the transmitter was pointing towards Kiruna, so a direction mark was thereby registered on the reception records. This experiment was performed during the winter of 1961 (near the sunspot minimum), when only one propagation mode was operative. Results of this experiment indicated that the signal from the NE was sometimes stronger than the signal from the NW. The opposite should be expected for the great-circle path.

College–Kjeller and Thule–Kjeller propagation-path analysis (SSN38.3–80.2) From January 1961 through June 1962 backscatter transmissions from College, Alaska and from Thule, Greenland were monitored at Kjeller (near Oslo), Norway. A limited amount of data was also obtained at receiving stations at Isfjord on Spitzbergen island. The locations of the transmitting and receiving stations are shown in Figure 8.25. The Thule–Isfjord path is entirely within the auroral zone, all other paths traverse the oval at approximately right angles. The great-circle distances for the various paths are ■

College–Kjeller,

■

Thule–Kjeller,

■

College–Isfjord,

■

Thule–Isfjord,

6000 km; 3350 km; 4050 km; and 1250 km.

Of particular interest with these trans-polar circuits is the orientation of the raypaths relative to the D-region absorption regions in the auroral oval. The vertical plane geometry, assuming a symmetrical mode structure and an F-region reflection height of 300 km, is shown in Figures 8.29–8.31, with the hatched areas depicting the approximate position of the auroral oval and the shaded areas denoting maximum absorption. Under quiet conditions the three- and four-hop College–Kjeller transmissions were found to be vulnerable to AA at the transmitter end of the circuit, whereas the auroral–oval absorption region at the receiver end should leave all conventionally propagated signals virtually unaffected. Of course, during disturbed conditions the auroral oval expands considerably and could seriously influence circuits otherwise not exposed to AA. On the College–Kjeller path, the most likely modes are those of one, two or three hops; the two-hop mode is, however, strongly discriminated against by the radiation pattern of the antenna. Furthermore, the conventional propagation mode for 18 MHz from College during the winter sunspot-minimum conditions is highly improbable, so unconventional modes are most likely. The gross behavior of these circuits is displayed on the following selected histogram plots. The first type gives, on a daily basis, the total number of hours with

Figure 8.29. College–Kjeller idealized mode geometry (from Owren et al., 1963).

Figure 8.30. Kjeller–College idealized mode geometry (from Owren et al., 1963).

Figure 8.31. Kjeller–Thule idealized mode geometry (from Owren et al., 1963).

456

High-latitude propagation: 1

signals present, hours with blocking interference, and outages. The heights of the black and dotted columns measure periods of reception and interference, respectively. Transmitter or receiver off periods are represented by empty columns with a lower-case e inserted. Whenever the e sign appears above a black or dotted column, this implies for that particular day an outage period corresponding to the distance from the boundary of the upper column to the top line (the 24-h line). Magnetically quiet and disturbed days are denoted by the capital letters Q and D. Filled triangles below the bottom line serve to indicate times for the occurrence of sudden-commencement magnetic storms. Figures 8.32 and 8.33 represent conditions for summer (July–August 1961) for SSN ⬵50 and winter (October– December, 1961) on the cis-polar Thule–Kjeller circuit for 12 and 18 MHz. Similarly, Figures 8.34 and 8.35 illustrate the behavior of the trans-polar College–Kjeller HF circuit for SSN ⬵50 in summer (July–September 1961) and winter (October–December 1961) for 12 and 18 MHz. The seasonal variation of signals on these two circuits was studied by plotting signal strengths on selected quiet days, using riometer and K indices as indicators of disturbance. Each curve represents average values for the days chosen. If possible, 8–10 days were picked out for each of the months selected for displaying characteristic seasonal quiet-day trends. In some cases, interference and outages tended to seriously constrain the amount of data available: therefore the curves shown are not equally reliable for defining quantitatively the seasonal properties pertaining to the transmission in case. It should also be noted that these data were obtained for relatively low sunspot activity (SSN⬵50). These seasonal statistics are for the 18-MHz circuit and the College 12- and 18-MHz circuit. Figures 8.36–8.38 show the seasonal variations on these circuits. From April 1961 until June 1962 (SSN38.3–64.3) the Geophysical Institute monitored HF pulse transmissions from the 12-, 18-, and 30-MHz backscatter sounders located at Thule, Greenland (D2900 km). The equipment parameters were given by Peterson et al. (1959). Because of severe SW interference, the 12-MHz transmissions were 80%–90% unusable and the 30-MHz signals from Thule were blanked out by the College 30-MHz backscatter sounder, so only the 18-MHz Thule data were usable. The Thule–College path crosses the auroral oval at near normal incidence only once (compared with the trans-polar paths) and a vertical ionosonde at Resolute, Canada provided data near the midpoint of the path. Figure 8.39 shows the vertical-plane geometry of this path. The seasonal variation is illustrated by plots of the hourly average values of the 12- and 18-MHz Thule pulsed-signal strength for winter, summer, and equinox shown in Figure 8.40. The signal strengths shown in Figure 8.40 are scaled in arbitrary units and are based on 2–4-week periods centered on the dates of the winter and summer solstices and the autumnal equinox of 1961. The highest signal levels are in the winter period and the lowest for the summer solstice, with the equinoctial values falling

8.4 HF propagation

in between – which is similar to the case of mid-latitude HF propagation. The average signal strength of the Thule 18-MHz pulsed transmissions as a function of season is shown in Table 8.7. The high average winter values are probably due to low absorption and high critical frequencies. The high occurrence of auroral-E ionization during the night and normal F-layer propagation modes during the day qualitatively explain the relatively high average signal strength during the equinoctial period. The greatest variation in signal amplitude occurred during the equinoctial period and the least variation during the summer/winter periods. The diurnal variation of the average 18 MHz signal strength (in dB above 1 V) is illustrated in Figure 8.41 for a geomagnetically quiet day (11 June, 1961) for three polar paths. The solid line below the signal plot indicates that very strong interference was present at College. The open lines denote intervals when the vertical ionosonde located at Resolute, Canada indicated that the MUF (3000) F2 was less than 18 MHz. During these periods, communication with the Thule– College circuit should have been impossible, but actually quite high signal levels were recorded at College. This illustrates once again that predictions based on vertical-incidence data are usually very unreliable for polar propagation paths at frequencies near 18 MHz. The reliability of these 18-MHz cis-polar and trans-polar circuits is quite well illustrated in Figure 8.41 It should be remembered, however, that this was a geomagnetically quiet day about three years after solar maximum. In general, it was found that the Thule signal was present practically 24 h per day, except during disturbed periods. This illustrates the importance of auroral-E ionization in supporting 18-MHz propagation during periods when F-region ionization is low in the polar ionosphere. Hourly average signal levels for the months of June 1961 (SSN55.8) and June 1962 (SSN38.3) are shown in Figure 8.42 for UT morning and evening periods. The period 07–21 UT is not presented because many data were lost due to excessive interference on this circuit. Because of the paucity of data, it was not possible to draw definite conclusions about sunspot-cycle effects on this propagation path 8.4.4

College–Kiruna absorption studies at fixed frequencies

Fixed-frequency pulse transmissions from the IGY sounder at College, Alaska monitored in Kiruna, Sweden were compared with absorption measurements from a vertical riometer at Thule, Greenland (geomagnetic latitude 88° N) and an oblique-incidence riometer located at College. The oblique riometer at College utilized a three-element Yagi antenna 0.5 above ground, directed towards a geographic bearing of 015°. Simultaneous data on absorption and signal strength were obtained for two PCA events, a strong event in July 1959 (SSN155.8) and a weak event in May 1960 (SSN117.0).

457

Figure 8.32. Thule–Kjeller 18-MHz (a) and 12-MHz (b) propagation conditions for summer 1961 (from Owren et al., 1963).

HOURS

6

8

12

16

20

24

4

8

12

16

20

24

(b)

10 Q

5

10

e e e e e e e e e e e

D

5

e e e e e e e e e e e

(a)

20

15

20

25

25

signals present

15 QQQQ

OCTOBER

e

e

e

e

e

DDDD

e

30

e

30

e

e

DD

10

5

e

DD

20

10

20

NOVEMBER

15

e

e

e

30 Q DD

5

25

30

5

e

equipment failure

25

e

QQQ

e e e e e e e e e e e e e e e e

15 Q

e e e e e e e e e e e e e e e e

noisy period

5 D

e

10

10

20 QQQ

25 Q

30 D

20

25

30

e e e e e e e e e e e e e e

DECEMBER

15

D

e e e e e e e e e e e e e e e e

sudden commencement

Q

15

e

Figure 8.33. Thule–Kjeller 18-MHz (a) and 12-MHz (b) propagation conditions for winter 1961 (from Owren et al., 1963).

HOURS

HOURS

4

8

12

16

20

4

8

12

16

20

(b)

(a)

e

e

D

20

JULY

20

15

5

e

signals present

QQDD

15

e e e e

e e e e e

10

e

10

e

e e e e e

5 D

e e e e e

e

25

25

e

e

D

e

e

30

e

30 QQQ

e

e e

e

D

10 Q Q D

5

e

Q

15

15

AUGUST

10

noisy period

D

5

e

e

20

e

20

25

e e e

30 DDD

e e

25

e

e

e

30

e

equipment failure

QQ

e

QQ

D

15 Q

20 Q

25 QDD

sudden commencement

10

30 D

5

10

20

SEPTEMBER

15

25

30

e e e e e e e e e e e e e e e e e e e e e e e e e e e e e

5

e e e e e e e e e e e e e e e e e e e e e e e e e e e e e

Figure 8.34. College–Kjeller 18-MHz (a) and 12-MHz (b) propagation conditions for late summer 1961 (from Owren et al., 1963).

HOURS

HOURS

4

8

12

16

20

24

4

8

12

16

20

24

(b)

10 Q

5

10

e e e e e e e e e e e

D

5

e e e e e e e e e e e

(a)

e

e

15

25

20

25

signals present

15 20 QQQQ

e

OCTOBER

e

e

e

e

DDDD

e e

30

30

e

e

e

e

5

5 D

e

e

15 Q DD

20

10

20

NOVEMBER

15

e

e

30 Q DD

e

5

25

e

30

5

e

equipment failure

25 QQQ

e e e e e e e e e e e e e e e e

noisy period

DD

10

e e e e e e e e e e e e e e e e

Q

10

10

20 QQQ

25 Q

D

30 D

e e e e e e e e e e e e e e e

15

e

20

25

30

e e e e e e e e e e e e e e

sudden commencement

15

DECEMBER

e e

e e

Figure 8.35. College–Kjeller 18-MHz (a) and 12-MHz (b) propagation conditions for fall-winter 1961 (from Owren et al., 1963).

HOURS

8.4 HF propagation

463

60 JUNE 1961 OCT. 1961 JAN.–FEB. 1962 MAR. 1962

SIGNAL STRENGTH (dB)

50

40

30

20

10

0

00

03

06

09

12

15

18

21

24

HOURS, UT

Figure 8.36. Seasonal signal-strength variation on the College–Kjeller 18-MHz propagation path (from Owren et al., 1961).

60 JAN.–FEB. 1961 MAR.–APR. 1961 JUNE 1961 OCT. 1961

SIGNAL STRENGTH (dB)

50

40

30

20

10

0

00

03

06

09

12

15

18

21

24

HOURS, UT

Figure 8.37. Seasonal signal-strength variation on the Thule–Kjeller 18-MHz propagation path (from Owren et al., 1963).

High-latitude propagation: 1

464

JAN. 1961 MAR.–APR. 1961 JUNE 1961 OCT.–NOV. 1961

60

SIGNAL STRENGTH (dB)

50

40

30

20

10

0 00

03

06

09

12

15

18

21

24

HOURS, UT

Figure 8.38. Seasonal signal-strength variation on the College–Kjeller12-MHz propagation path (from Owren et al., 1963).

Table 8.7. The average signal strength of the Thule 18-MHz pulsed transmissions Signal strength (dB above 1 V)

UT

Time 150° WMT

Summer

Winter

Equinox

00 08 17 20

1400 2200 0700 1000

35 31 41 27

49 38 42 49

43 37 39 40

Figure 8.39. Most probable F-layer modes for the Thule–College propagation paths (from Owren et al., 1963).

Figure 8.40. Seasonal variations of the 12- and 18-MHz (MC) Thule signals received at College, Alaska in 1961. The ordinate shows hourly average values of signal strength. The horizontal bars labeled QRM denote periods of high loss of data due to severe interference (from Owren et al., 1963).

Figure 8.41. Quiet-day variations of 18-MHz signal strength for three propagation paths (SSN55.8 for June 1961) (from Owren et al., 1963).

468

High-latitude propagation: 1

Figure 8.42. The solar-cycle variation of the Thule 18-MHz signal. Hourly average signal levels are plotted on the ordinate (from Owren et al., 1963).

The strong PCA event (SSN155.8) In the period 1937 UT 9 July 1959 through 2115 UT 16 July 1959, four solar flares of importance 2 to 3 were observed. Absorption events associated with these flares are listed in Table 8.8. During this period data on the signal strength of pulse transmissions originating at College and received at Kiruna were available. The system parameters are given in Table 8.9. The length of the transpolar path is 5300 km. Figure 8.43 shows the details of the PCAs and HF “black-outs” (no signal received). The times of occurrence of the flares are indicated by the letter F, whereas the symbol SC denotes the start of a sudden-commencement geomagnetic storm. The periods of total black-outs of the 12- and 18-MHz College–Kiruna pulsed circuit are shown at the top of Figure 8.43. The 12-MHz black-out lasted almost three days longer than the 18-MHz black-out, which qualitatively illustrates the frequency dependence of signal attenuation in the D region. The sharp flattening off of the absorption on 11 July is mostly due to absorption exceeding the useful dynamic range of the riometer. It should be emphasized that this was a particularly complex event, with low-energy cosmic rays bombarding the polar cap almost continuously for about 14 days.

8.4 HF propagation

469

Table 8.8. Absorption events

Date

Starting time (UT) Duration (h)

10 July 14 July 16 July

0700 (Thule) 0700 2250

90 (College) 51 34

Maximum absorption at 27.6 MHz (dB) 20 23.7 21.2

Table 8.9. System parameters

Frequency (MHz)

Power output (kW peak)

Pulse-repetition frequency (pulses s1)

Pulse length (s)

11.634 17.900

5.0 5.0

18.75 18.75

1200 1200

The weak PCA event of 13 May 1960 (SSN117.0) The PCA event of 13 May 1960 was a small, typical event with maximum absorption values of 4.5 and 3.5 dB on the vertical riometers at Thule and College, respectively. Two solar flares of importances 2 and 3 occurred sometime before 0522 UT on 13 May. Simultaneous absorption data from the College oblique riometer and the 18-MHz signal strength on the College–Kiruna propagation path are shown in Figure 8.44. The first absorption peak at approximately 0600 UT corresponds to a dip in the 18-MHz signal from 3 to 1 on the arbitrary scale. The path was completely blacked out from 1045 to 2110 UT, corresponding to another absorption peak of approximately the same amplitude (⬃6 dB). This illustrates the relatively poor peak-to-peak correlation of attenuation and absorption of the 18-MHz signal for all except strong PCA events. It should be emphasized that the oblique riometer measures the absorption encountered by the signal on only the last “hop” of the propagation mode and not the attenuation on the other hops. During a strong PCA event the region of absorption includes all of the College–Kiruna propagation path; consequently one might expect better peak-to-peak correlation for the strong events.

Figure 8.43. Effects of large PCAs of 9–23 July 1959 on 12- and 18-MHz transpolar transmissions (from Owren et al., 1961).

8.4 HF propagation

Figure 8.44. Weak PCA effects on the College–Kiruna 18-MHz circuit, measured using the College oblique riometer (from Owren et al., 1963).

471

472

High-latitude propagation: 1

Figure 8.45. Effects of a strong AA event on 11 September 1961 on the 18-MHz Thule–College path (from Owren et al., 1963).

8.4 HF propagation

8.4.5

Effects of auroral-zone-absorption events on HF propagation

Simultaneous absorption and signal-attenuation data for the Thule–College 18-MHz path for a very strong auroral-zone-absorption (AA) event on 11 September 1961 (SSN52.3) are shown in Figure 8.45. Continuous data were available except during the periods 0330–0730 and 0830–1100 UT, when strong interference made identification of the Thule pulsed signal doubtful. A complete black-out of the 18-MHz signal occurred between 1140–2400 UT and lasted until the absorption level returned to approximately 1.5 dB at 0730 UT on 12 September. This was the strongest AA event recorded during this investigation and should not be regarded as a typical event. It was the only AA event studied which produced a black-out on this path. 8.4.6

Sweep-frequency experiments Forward oblique sounding investigations near sunspot minimum

During 1963 and 1964 the GI/UAF operated a combination HF step-frequency backscatter and synchronized forward sounding system (Davies, 1990) at College utilizing commercial pulse sounders and antennas. Figure 8.46 shows the five propagation paths studied in the course of this experiment and some of the pertinent parameters of the system are listed in Table 8.10. The data described in this section were obtained during the period November 1963 (SSN23.8) to February 1964 (SSN17.8) and represent winter, sunspot minimum conditions.

Thule–College path The Thule–College great-circle path is 2900 km long and the most probable propagation modes are one-, two-, and three-hop F-modes; two-, three-, and four-hop

Table 8.10. Parameters of the forward sounding system Power output

30 kW (rated), 15 kW (measured)

Frequency range

4–64 MHz

Short-pulse mode

PRF 50 pulses s1, pulse length 100 s, bandwidth 16 kHz, four pulses per channel

Long-pulse mode

PRF 20 pulses s1, pulse length 1000 s, bandwidth 4 kHz, two pulses per channel

Antennas

Granger Model 726-4/64 log periodic vertical monopole LPAs directed at true azimuths of 015°, 105°, 210°, 270°, and 325°

473

Figure 8.46. Great-circle HF-propagation paths studied during 1963 and 1964, along with an approximate auroral oval for Kp 4 (from Bates and Hunsucker, 1964).

8.4 HF propagation

475

Figure 8.47. Winter, sunspot-minimum mode structure on the Thule–College HF propagation path. The longpulse record is shown above and the shortpulse record is shown below (after Bates and Hunsucker, 1964).

modes or combination E–F modes. Figure 8.47 shows typical winter long- and short-pulse records from Thule. The upper long-pulse record displays a common winter auroral-E with a LOF of 5 MHz and a MOF of 17 MHz. The short-pulse record below illustrates another common mode with an auroral-E LOF of ⬇4 MHz and a MOF of 20 MHz. Very spread-F modes are present between 5 and 10 MHz.

Auroral-E modes The relative occurrence of auroral-E modes on the Thule–College circuit is shown in Figure 8.48 (lower plot). The histogram peaks around 2000–2400 UT (10–14, 150° WMT and has a minimum around 1100–1400 UT (01–04, 150° WMT). The histogram gives the fraction of time that auroral-E is present on this path, and shows that the dominant winter mode for this path is, in fact, supported by auroral-E. The histogram in Figure 8.49 shows the diurnal variation of the average MOF for each hour and the upper plot gives the highest MOF observed during each hour for the period 27 November 1963 to 12 February 1964. A maximum

476

High-latitude propagation: 1

Figure 8.48. The lower plot shows the diurnal variation of the auroral-E maximum observed frequency MOF for the same period (after Bates and Hunsucker, 1964).

is indicated between 0730 and 1230 UT (2130–0230, 150° WMT) in the average MOF curve for Thule, which corresponds to the diurnal auroral peak at College.

Other winter modes In addition to the predominant auroral-E modes during the winter on the Thule–College path, various other modes are observed. Figure 8.50 (lower record) shows the typical auroral-E along with what appears to be a very spread-F mode

8.4 HF propagation

Figure 8.49. The diurnal variation of the auroral-E maximum observed frequency (MOF) (after Bates and Hunsucker, 1964).

at lower frequencies. Short-pulse (100 s) records of the Thule signal taken in late February and March (daytime) show the normally expected F modes (the upper record in Figure 8.51) with a subsequent decrease in occurrence of the auroral-E mode. That this is to be expected is discussed in following sections.

Off-path modes One of the most interesting high-latitude HF modes is the “Off-path” or “nongreat-circle” (NGC) mode. Two examples of these modes on the Thule–College

477

478

High-latitude propagation: 1

Figure 8.50. Typical winter records from Thule, Greenland. The long-pulse record is shown above and the short-pulse record is shown below (after Bates and Hunsucker, 1964).

path are shown in Figure 8.52, with the direct path (probably auroral-E) at the proper time delay, plus several distinct NGC modes. The time delay and lack of retardation of the mode structure do not allow a “multiple-hop” interpretation, so we postulate a NGC mode.

The Andøya–College path (D5000 km) The most prominent winter-night mode on the Andøya–College trans-polar path appears to be at least partially supported by auroral-E, in that the signal exhibits the constant-range discrete-signal characteristics. Occasionally some retardation is present at the upper frequency end of the trace, but in most cases the signal appears similar to that shown in Figure 8.47. The relative occurrence of auroral-E on the Andøya–College path has two diurnal peaks, as shown in Figure 8.48

8.4 HF propagation

Figure 8.51. Normal F-modes from Thule (above) and Andøya (below). (after Bates and Hunsucker, 1964).

(upper trace), one peak is centered around 0400 UT (1800 local time). Auroral-E MOFs as high as 32 MHz and an average MOF of 18 MHz are observed on this circuit, as shown in Figure 8.49 (upper plot). There are two diurnal peaks in average MOF on the Andøya circuit as opposed to the single maximum on the Thule path. Average MOF maxima occur at the times 0600–0900 and 1900–2100 UT (2000–2300 and 2000–2300, 150° WMT). 8.4.7

Other results from HF high-latitude studies from c. 1956–1969

Most of these results are taken from a survey paper by Hunsucker and Bates (1969).

479

480

High-latitude propagation: 1

Figure 8.52. Direct and off-path modes from Thule long-pulse records. In the lower ionogram, the maximum off-path delay is 11 ms (after Bates and Hunsucker, 1964).

Auroral-E ionization effects It appears that, during the winter months, high-latitude HF propagation is predominantly supported by auroral-E (AE) ionization, even during moderate sunspot activity (SSN⬃35). The importance of AE in high-latitude HF propagation during winter night-time conditions was reported by Hunsucker and Stark (1959). Results obtained on trans-polar paths monitoring HF pulsed transmissions and fixed-frequency backscatter soundings to the north revealed that AE activity peaks during the period 1800–0600 150° WMT. Folkestad (1963, personal communication) reported that signal strengths of

8.4 HF propagation

18-MHz pulsed transmissions on the trans-polar College to Kjeller path during January and February 1961 peaked during 2100–0600 150° WMT, further illustrating the role of AE in trans-polar propagation during the winter night. The high percentage occurrence of AE of maximum frequency 5 MHz in the polar regions during the winter night was also emphasized by Leighton et al. (1962) using IGY results. See Figures 8.50–8.52. In an early study of the relationship between visual aurora and vertical ionosonde observations, Heppner et al. (1952) found a high correlation between certain zenithal auroral forms and values of fEs (the AE cutoff frequency). In particular, rayed bands at the zenith gave the highest correlation with fEs, whereas complete absorption was indicated 100% of the time when pulsating auroral forms were overhead. Another study of the relation between visual aurora and verticalionosonde fEs data was performed at an auroral-zone station by Hunsucker and Owren (1962). Using all-sky-camera photos, they found that the motion of an auroral arc or band from a low elevation angle to a position near the zenith was accompanied by an increase in the value of fEs by a factor of two or greater. With discrete auroral forms near the zenith, values of fEs from 8 to 11 MHz were common, with a maximum value of 13 MHz (also see Hunsucker, 1965). The results of this investigation are in good agreement with the foregoing findings concerning the occurrence and behavior of auroral-oval E-region ionization during winter-night sunspot-minimum conditions. The MOF peak on the Thule–College circuit coincides with the period of maximum auroral activity near the College end of the circuit (2130–0230, 150° WMT). AE propagation on the Andøya–College path is a much more complicated phenomenon, displaying several diurnal peaks in activity. This is to be expected, since the 5000-km propagation path traverses the auroral oval twice and hence exhibits sunrise/sunset effects twice, compared with once on the Thule–College path. The transition from the winter “night modes” (AE propagation) to the Fpropagated “day modes” for the polar paths investigated takes place at about the middle of February. As the reflection points on the Thule–College and Andøya– College paths become sunlit, the normal multihop F modes propagate – as shown in the records in Figure 8.51.

NGC modes Very strong evidence of HF/NGC propagation modes associated with the auroral oval was presented by Egan and Peterson (1962). Monitoring of the 12- and 18-MHz pulsed signals from Thule and College at Stanford revealed very strong delayed modes with time delays of up to 12 ms between the direct mode and the “sidescatter” modes. Ortner and Owren (1961) also presented evidence for the existence of such modes on the 18-MHz trans-polar path between College and Kiruna, Sweden. Additional evidence for the existence of NGC modes was given by Hunsucker (1964a; 1964b) for a synchronized step-frequency circuit between College and Oya, Norway and for the 18-MHz College–Thule circuit.

481

482

High-latitude propagation: 1

Figure 8.53. A schematic representation of the requirement for overlap between the elliptical locus of possible off-path sidescatter points on the Palo Alto-to-College path and the scattering belt, as determined from the College backscatter data (from Bates et al., 1966).

Bates et al. (1966) presented a detailed study on the relationship of the aurora to NGC HF propagation on HF forward-sounding records received at College from various sites during 1963 and 1964. No direction-finding equipment was available, so a statistical analysis was performed in order to determine the type of sidescatter involved. The number of occurrences during several periods was maximum at night. The excess propagation time on the Palo Alto to College path varied inversely with magnetic activity. A comparison of simultaneous College backscatter and Palo Alto to College off-path data showed that the locus of offpath sidescatter extended north of the ionospheric backscattering belts. These results were interpreted as showing that the deviated modes were produced by sidescatter from the auroral belt as shown for the Palo Alto-College path in Figure 8.53.

8.4 HF propagation

The HF forward-sounding experiments in the early 1960s emphasized (among other things) the importance of the AE mode for winter-night near-sunspotminimum conditions on polar paths. During the period 27 November 1963–12 February 1964, the average percentage occurrence of AE on the Thule–College and Andøya–College circuits was over 50%. AE MOFs as high as 46 MHz with typical values of 18 MHz were observed on the Thule–College path and MOFs as high as 32 MHz with typical values of 16 MHz were observed on the Andøya–College path. This suggests that long-haul HF traffic might be routed over polar paths during winter-night near-sunspot-minimum periods when midlatitude MOFs are quite low. The importance of NGC modes in carrying the MOF was also illustrated in these early-sixties studies during the midwinter period. Bates and Albee (1966) also pointed out the importance of the F1-layer effects on long-distance, high-latitude (and even some sub-polar) HF circuits. During the 1964 sunspot-minimum period the F2 critical frequency at College was not appreciably greater than that of the F1 layer. This condition resulted in considerably modified conditions, and frequently the F1 layer carried the maximum propagating frequency. The terminology of Bates and Albee (1966) is illustrated by an example, the 4F2 mode. In this case there are four reflections from the F2 layer, with ground reflection in between and the end-points are the ground. The 4F2 mode is also called the fourth-order F2-layer forward-propagation mode. In general, the first number gives the mode order and the rest of the symbols denote the ionospheric layer involved. Propagation modes involving successive reflections both from the E layer and from the F layer are termed combination E–F modes. Figures 8.54(a) and (b), recorded on the 3500-km Palo Alto to College path, illustrate normal F-region oblique ionograms. The mode structure is well defined, the low ray traces showing a slight retardation at the low-frequency end, the high rays are relatively short, and the magnetoionic splitting can be seen. Figures 8.54(c) and (d) illustrate ionograms that are representative of the summer 1964 and 1965 records obtained on all paths to College. The downward curving portion of each record is composed of several discrete traces produced by signals in the first four F layer modes. Traces such as those in Figures 8.54(c) and (d) are termed “long-tailed” traces. Figures 8.55–8.58 show long-tailed traces on the signals recorded at College from Thule (2900 km), Palo Alto (3500 km), Fort Monmouth (5200 km), and Okinawa (7600 km) for the morning of 18 July 1964. This was one of the few instances when the long-tailed traces were observed on the four paths more or less simultaneously. A noteworthy feature is the sudden appearance and disappearance of the longtailed traces. Long-tailed traces were never as outstandingly developed on the Thule to College signal as they were on the other paths; this is undoubtedly due

483

Figure 8.54. Forward oblique ionograms recorded on the Palo Alto-to-College path: (a) and (b) were recorded during October (SSN19.7), and (c) and (d) during July 1965 (SSN15.5). Time marks are 1.0 ms apart. One microsecond pulse per channel was transmitted per 1000 frequency channels between 4 and 24 MHz (from Bates and Albee, 1966).

(b)

(a)

(d)

(c)

Figure 8.55. A long-tailed trace sequence recorded on 18 July 1964 (SSN10.3) on the Thule-to-College path. Times are UT. The pulse width is 1 ms (after Bates and Albee, 1996).

486

High-latitude propagation: 1

Figure 8.56. A long-tailed trace sequence recorded on 18 July 1964 on the Palo Alto-toCollege path. Times are UT. The pulse width used was 1 ms (after Bates and Albee, 1966).

to the decrease of the F1-layer critical frequency with latitude relative to that of the F2 layer. Figure 8.59 contains an oblique ionogram that was derived from the vertical ionogram in Figure 8.59. For simplicity only E and F modes are shown; combination E–F modes are ignored. Signals from the E layer produce the constant time traces shown in Figure 8.59. The downward-curving portion of the oblique ionogram is primarily composed of two lines, which correspond to the F1 and F2 critical frequencies. Each down-curving line is approximately the vertical-tooblique transformation of the critical frequency. Figure 8.59 shows that the gap between the lines was produced by the relative closeness of the E and F1 critical frequencies, while the great decrease in travel time with increasing frequency was produced by the nearness of the F1 and F2 critical frequencies. The complete signal-trace structure for the first four modes is

Figure 8.57. A long-tailed trace sequence recorded on 18 July 1964 on the Fort Monmouth-to-College path. Times are UT. Pulse width 1 ms (after Bates and Albee, 1966).

Figure 8.58. A long-tailed trace sequence recorded on 18 July 1964 on the Okinawa-to-College path. Times are in UT. Pulse width 1 ms (after Bates and Albee, 1966).

8.4 HF propagation

Figure 8.59. The forward ionogram for a 3500-km path derived from the vertical-incidence ionogram in the inset (from Bates and Albee, 1966).

contained in Figure 8.61; no consideration was given to possible shielding effects of lower layers on the very-oblique modes. The F1 layer controls the maximum frequency for the path in the calculated example; experimentally the F1 layer appeared to carry the MOF during much of the 1964 summer daytime on the Palo Alto and Thule to College paths. Tveten (1961) found that the F1 layer frequently carried the MOF on the Barrow (Alaska) to Boulder (Colorado) path, and Maliphant (1969) noted the importance of F1 propagation on the long transAtlantic Slough to Ottawa path.

Possible ducted modes In the previous section it was shown that the long-tailed traces were produced by signals in the first three or four F2-layer modes. Figure 8.62 illustrates traces, however, that cannot be explained in that manner; these traces will be referred to as “delayed-long-tailed” (DLT) traces. We will first assume that these DLT traces were produced by higher-order F2-mode signals. Rough estimates of the travel time, and hence reflection heights, can be made by assuming that the first-arriving signals propagated via the minimum-order E or F1 modes possible for the paths in question. The records shown in Figures 8.60(a), (b), and (c) will be analyzed in this fashion; the minimum time delays between the first-arriving and the DLT signals were approx-

489

490

High-latitude propagation: 1

(a)

(c)

(b)

(d)

Figure 8.60. Records showing highly delayed signals. Traces in (a) and (b) were obtained on the Palo Alto-to-College path with four 100-s pulses per channel, per 100 frequency channels between 4 and 24 MHz. Records (c) and (d) were obtained using two 1.0-ms pulses per channel on the Fort Monmouth- and Palo Alto-to-College paths, respectively (from Bates and Albee, 1966).

Table 8.11. Virtual-reflection heights for each hop mode Virtual-reflection height (km) Palo Alto Mode

1.75 ms

2.3 ms

1F2 2F2 3F2 4F2

850 500 350 250

560 390 300

Fort Monmouth 3.0 ms 700 525 400

8.4 HF propagation

imately 1.75, 2.3, and 3.0 ms, respectively. Applying the Martyn and Breit–Tuve theorems (and accepting the small errors due to the spherical geometry), these data correspond to virtual reflection heights for each hop mode (Table 8.11) In each case F2 vertical-incidence heights near 500 km were observed at College, so that the most probable modes would be the third and second, and fourth- and third-order F2 modes in that order for the Palo Alto and the Fort Monmouth to College paths. This leads to the conclusion that some, if not all, of the lowest-possible F2-mode signals were absent if the DLT signals were in reality higher-mode signals. The horizon-cutoff height for the one-hop mode on the Fort Monmouth to College path was 600 km, so 1F2-mode signals would not be expected (although they might be possible). The absence of the lowest-possible F2 modes can be explained qualitatively by considering the rate of change of the secant of the angle of incidence with respect to the angle of incidence at the transmitter as a function of the reflection height. When such a computation is performed, it is evident that the secant of the angle of incidence increases more rapidly for F1-layer heights than it does for F2-layer heights. The secant factor is proportional to the maximum frequency reflected by the layer; hence, if the F1 and F2 critical frequencies differed by only ⬃10%, the F1 layer would prevent all but relatively high-angle rays from penetrating to the F2 layer at frequencies capable of undergoing F2 reflection. A relatively thick, dense F1 layer can therefore act as a shield for oblique F2 propagation to prevent the low-angle F2 modes from propagating. It may be constructive to consider the possibility that the DLT traces were not produced by a normal Earth–ionosphere hop-mode signal, but that the signal was ducted in the ionosphere. One type of ducted mode, usually termed an elevated or tilted mode, has been proposed to explain long-distance backscatter echoes and trans-equatorial propagation . Tilted modes do not appear to be the explanation in this case because the 3500-km Palo Alto path, for example, is so short that an excessive tilt would be required, and, furthermore, the required upward tilt to the south is in the wrong direction to that observed. An examination of the monthly median ionospheric heights and critical frequencies observed at various Alaskan, Canadian, and US sites during the summer of 1964 when the DLT signals were most likely to be observed indicated that the virtual height of the F2 layer decreased and the F1 and F2 critical frequencies increased to the south (CRPL F series, part A), thus producing a strong downward tilt to the south from College. Actual ducting within the ionosphere is, however, another matter. For the case at hand, ducting between the F1- and F2-layer maxima seems the most probable explanation for the single long-tailed traces. Such ducting could occur only if several relatively special ionospheric conditions applied. An electron-density valley must exist in order to provide the necessary velocity minimum around which the guided wave propagates. The wave could enter the duct at the beginning of a valley, or where a strong-enough horizontal gradient occurs in the F1 layer

491

High-latitude propagation: 1

492

to allow penetration at one point but not at a point further along the path. A tilted F2 layer might not be necessary, but a tilt would help by gradually changing the angle of incidence of the propagating wave. The proposed model is speculative and might not be necessary to explain the observed records. The high-order-mode model, though, will not satisfactorily explain the existence of the single DLT trace on a record, whereas ducting will, because only one ducted mode would generally be expected. A further observational point in favor of the idea of a ducted mode is the extreme variability of the single long-tailed trace. Within the span of several soundings (20 min apart) the trace appeared and disappeared. The high-ray trace was well defined and exhibited extreme retardation; this behavior is not a characteristic of normal Earth– ionosphere hop-mode signals. These observations are not readily explainable by the idea of a higher-order mode.

A summary of summer 1964 data The F1 layer considerably modified HF propagation conditions on high latitude paths during daytime in the summer of 1964. This period was characterized by thick F1 and F2 layers in the 4- and 5-MHz ranges, respectively. Conditions on the Palo Alto to College path were relatively predictable, in that frequencies in the 10–18MHz range propagated during most of the day with no great variation in flight time. On the Fort Monmouth to College path, however, the minimum-flight-time traces were in roughly the same frequency range as those on the Palo Alto paths, but they occasionally disappeared when the lower-frequency DLT traces appeared. This may have been partly an equipmental effect because the antennas used had relatively poor radiation characteristics at low elevation angles, but the records clearly indicate that there was an appreciable drop in signal strength. Whether these signals would have been received with better antennas is unknown. In any case, the records clearly show that signals propagated in relatively highangle F2 modes. Antennas for paths in the 5000-km range that discriminate against such high-angle radiation may cause communication outages, while none would be noted with less-directive arrays. Thus, less-directive arrays have their place, alongside very directive antennas in communications-antenna applications at high latitudes. 8.4.8

Doppler and fading effects on HF high-latitude propagation paths

A survey of polar and auroral-region effects on HF propagation (Hunsucker and Bates, 1969) lists some of the results given in Chapter 8, along with results from other investigations. Two important results in the survey paper are the Doppler spreading and fading of HF signals which traverse the auroral oval. Lomax (1967) presented an example of typical power spectra observed at the Palo Alto receiving site for transmissions from Thule, Greenland and Fort Monmouth, New

8.4 HF propagation

Figure 8.61. Typical power-spectra transmissions from Thule (a) and Fort Monmouth (b) monitored at Palo Alto (from Lomax, 1967).

Jersey, as displayed in Figure 8.61. Some more recent data on Doppler shifts and spreading will be presented in Chapter 9. Anyone who has monitored HF transmissions that have traversed the auroral ionosphere will probably have encountered “auroral flutter,” which results from the reception of multiple signal components from auroral ionospheric irregularities. Koch and Petrie (1962) studied fading characteristics on a long path and found that fading rates higher than 20 Hz were present for a small percentage of the time on 10, 15, and 20 MHz. These fading rates exhibited only a minor diurnal trend, with the maximum occurring during morning hours. A study of the fading correlation bandwidth and

493

494

High-latitude propagation: 1

Figure 8.62. A map showing the Sodankylä–Lindau HF path in relation to the auroral zone, Arctic and other instrumented stations (after Rose, 1967).

short-term frequency stability on the same path was performed by Auterman (1962). He found that the mean fading correlation bandwidth was 4.3 kHz, the value exceeded 90% of the time was 1.0 kHz, and the bandwidth generally was smaller during periods of high geomagnetic activity. One of the best-instrumented medium-range, high-latitude HF forward-

8.4 HF propagation

Figure 8.63. A vertical section through the Lindau–Sodankylä HF propagation path (after Moller, 1964).

Figure 8.64. A partial cross-section through the Sodankylä–Lindau path, showing a quarter of the 2F mode (after Rose, 1967).

sounding circuits was the 1965-km path between Sodankylä, Finland and Lindau, Germany described by Moller (1964) and Rose (1964). Stanford Research Institute provided English translations of these important reports in 1964 and 1967, respectively. Figure 8.62 is a map showing the Sodankylä–Lindau propagation path, the location of the Arctic circle, the auroral “zone,” and the location of supporting instrumentation at Kemi, Luleå, Uppsala, Kiruna, and Lycksele. A vertical section through the Lindau–Sodankylä path is shown in Figure 8.63, with a possible mode structure indicated. It is obvious that the Sodankylä end of the path was the most affected by the auroral ionosphere and the geometry of the Sodankylä end of the 2F-layer mode is shown in Figure 8.64. Table 8.12 lists the parameters of the HF pulsetransmission/reception system used on the Sodankylä–Lindau circuit. The system at Sodankylä was also operated as a backscatter sounder at selected intervals. Simultaneous vertical soundings from the Uppsala ionosonde (located near the mid-point of the path) and forward transmission on the

495

496

High-latitude propagation: 1

Table 8.12. HF-system parameters Transmitter output power 50 kW Pulse duration 100 s Pulse-repetition frequency 50 Hz Frequency range 1.4–22.6 MHz Sweep duration 8 min Antennas: three rhombic antennas each for the transmitter and the receiver Nominal gain of each antenna 10 dB Receiver bandwidth 16 kHz

Figure 8.65. Oblique and vertical ionograms for the Sodankylä–Lindau HF circuit for a summer evening (30 June 1958) during sunspot cycle 19 (SSN87; Kp 4) (after Moller, 1964).

Sodankylä–Lindau HF circuit for a summer evening in 1958 (SSN187) are shown in Figure 8.65 along with calculations of the vertical and obliquely derived critical frequencies. Figure 8.66 shows similar plots for a winter early morning. It should be noted that the Lycksele ionosonde was located approximately 500 km from the Sodankylä end of the circuit and thus indicated the presence of AE ionization – which is reflected in the complex mode structure of Figure 8.68. This is further illustrated in Figure 8.67, in which the Lycksele ionogram clearly shows strong AE ionization. Many examples of the seasonal and diurnal behavior on this HF path during the maximum of solar cycle 19 are shown in Section C of Moller’s (1964) report and representative forward ionograms are shown in Figures 8.68–8.72. The effects

8.4 HF propagation

Figure 8.66. An oblique ionogram for the Sodankylä–Lindau HF circuit for a winter early morning (13 November 1958) near sunspot maximum (SSN181; Kp 3). The ionograms are from Uppsala and Lycksele (after Moller, 1964).

Figure 8.67. Effects of auroral-E ionization on mode structure on the Sodankylä–Lindau HF circuit at local midnight, midwinter (8 November 1958), sunspot-cycle maximum (after Moller, 1964).

497

Figure 8.68. Normal summer daytime Sodankylä–Lindau ionograms for 14 June 1958 (recorded once per hour from 0000 to 1200 MEZ) (after Moller, 1964).

8.4 HF propagation

Figure 8.69. Normal fall daytime Sodankylä–Lindau ionograms for 17–18 September 1958 (recorded once per hour from 1900 to 1600 MEZ) (after Moller, 1964).

499

Figure 8.70. Simultaneous vertical and oblique ionograms showing strong spread-F effects near midnight during sunspot maximum, 7–8 November 1958 (after Moller, 1964).

8.4 HF propagation

Figure 8.71. Effects of dense auroral-E ionization on 4 September 1958 on the Sodankylä–Lindau HF circuit (after Moller, 1964).

Figure 8.72. Some effects of a moderate geomagnetic disturbance on 7 October 1958 on the Sodankylä–Lindau HF circuit (after Moller, 1964).

501

502

High-latitude propagation: 1

Figure 8.73. The Andøya–College HF average monthly propagation spectrum for summer 1964 (D denotes disturbed and Q denotes quiet periods) (after Bartholomew, 1969).

of strong spread-F on the mode structure of the Sodankylä–Lindau HF circuit are graphically illustrated in Figure 8.70, effects of AE on the oblique circuit are shown in Figure 8.71 and moderate geomagnetic effects in Figure 8.72. Another analysis of data from the 4–64 MHz forward sounding path from Andøya, Norway to College, Alaska for the year 1964 (D5000 km; sunspot minimum) was conducted by the SRI and presented by Bartholomew (1969). Figures 8.73, 8.74, and 8.75 display the propagation spectra for summer, equinoxes, and winter, respectively. As pointed out by Bartholomew, the width of the frequency spectrum that occurred 50% of the time was fairly constant for all seasons. The median MOF and LOF propagating at least 50% of the time present relatively small diurnal variations in any season. The median MOF and LOF increased by about 3 MHz as the season changed from winter to summer. This behavior is fairly typical of high-latitude propagation, since solar illumination in this region exhibits more seasonal than diurnal variation. The median LOF on disturbed days was generally higher than that on quiet days, but the median MOF on disturbed days was not consistently different from the quiet-day median MOF. Periods of increased LOF at least qualitatively are related to periods of increased auroral absorption – especially during equinoctial and winter months. It is also interesting to note that, during equinoctial and

8.4 HF propagation

Figure 8.74. The Andøya–College HF propagation spectrum for 1964 equinoxes (after Bartholomew, 1969).

Figure 8.75. Andøya–College HH propagation for Winter 1964. (after Bartholomew, 1969).

503

504

High-latitude propagation: 1

Figure 8.76. Andøya–College HF multipath and propagation outage for summer 1964 (after Bartholomew, 1969).

Figure 8.77. Andøya–College HF multipath and propagation outage for 1964 equinoxes (after Bartholomew, 1969).

summer periods, there was significant propagation on frequencies greater than 20 MHz even during this sunspot minimum period. This, again, illustrates the importance of AE ionization in supporting propagation on high-latitude HF paths. Multipath propagation was a quite significant factor on the Andøya–College path and the seasonal dependences of multipath and propagation outages are displayed in Figures 8.76, 8.77, and 8.78 (for summer, equinoxes, and winter 1964, respectively). These figures also display the percentages of actual

8.4 HF propagation

Figure 8.78. Andøya–College HF multipath and propagation outage for winter 1964 (after Bartholomew, 1969).

propagation-outage time. Outage was highest at night, especially in winter, and disturbed days generally had greater outage than did quiet days. Short-pulse data were used to determine the Andøya–College mode structure for January through March 1964. Although the short-pulse data were more sparse than the long-pulse data (with less reliable statistics implied), it is instructive to note the complex mode structure during the month of March 1964 shown in Figure 8.79. Some of the morphology of off-path (NGC) propagation on the Andøya–College HF path is displayed in Figure 8.80, showing seasonal, quiet/ disturbed conditions and diurnal effects. A good example of how actual LUF and MUF values differ from predicted values on this 5000-km trans-polar path is shown in Figures 8.81, 8.82, 8.83, and 8.84 for January, April, June, and September 1964, respectively. The prediction program utilized was the predecessor to the IONCAP program. The MUF used here is the highest frequency expected to propagate at least 50% of the time, while the LUF, used is the lowest usable frequency with 50% reliability. The FOT is defined as the optimum traffic frequency – an estimate of the frequency that will propagate at least 90% of the time. The wide divergence between predicted and observed values is apparent. The Norwegian Defense Research Establishment (NDRE) also conducted an analysis of data on paths from Andøya to College and to Fort Monmouth, New Jersey in 1964 as shown on the map in Figure 8.85. Plots of the diurnal and seasonal behavior of MOFs on the Andøya–College and Andøya–Ft Monmouth trans-polar paths for winter (January 1964) and spring (March 1964) are shown in Figure 8.86. Plots for May and July 1964 are shown in Figure 8.87.

505

High-latitude propagation: 1

506

(a)

(b)

Figure 8.79. The Andøya–College HF mode configuration for March 1964; (a) quiet days and (b) disturbed days (Bartholomew, 1969).

8.4 HF propagation

Figure 8.80. Diurnal and seasonal behavior of Andøya–College HF NGC propagation for quiet and disturbed conditions (after Bartholomew, 1969).

Figure 8.81. Predicted MUF, LUF, and percentage occurrence of observed signal for January 1964 on Andøya–College HF circuit (after Bartholomew, 1969).

507

508

High-latitude propagation: 1

Figure 8.82. Predicted MUF, LUF, and percentage occurrence of observed signal for April 1964 on Andøya–College HF circuit (after Bartholomew, 1969).

Figure 8.83. Predicted MUF, LUF, and percentage occurrence of observedsignal for June 1964 on Andøya–College HF circuit (after Bartholomew, 1969).

8.4 HF propagation

Figure 8.84. Predicted MUF, LUF, and percentage occurrence of observed signal for September 1964 on Andøya–College circuit (after Bartholomew, 1969).

Figure 8.85. Location of trans-polar paths investigated by the NDRE (from Folkestad, 1968).

509

510

High-latitude propagation: 1

Figure 8.86. Plots of observed MOF distribution (shaded area), median frequencies (broken line), MUFs predicted from vertical ionosonde data (solid line), and periods with predicted screening by the E-layer (the heavy solid line near the bottom of the MOF plot) for College–Andøya (right-hand plots) and Fort Monmouth Andøya (left-hand plots). At the bottom of the figure, the vertical lines represent the number of detectable signals as a percentage of the total number of readings. Data from January 1964 are shown in (a) and data from March 1964 are shown in (b).

8.4 HF propagation

Figure 8.87. Plots in the same format as Figures 8.86(a) and (b) for May and July 1964.

511

512

High-latitude propagation: 1

Figure 8.88. Circuit behavior during disturbed periods in April 1964 for the Andøya transmissions received in College and Fort Monmouth, along with riometer absorption values from Longyearbyen, Tromsø, and Andøya.

As pointed out by Folkestad, (1) the observed median and maximum values of the MOFs were substantially above the predicted MUFs most of the time; (2) for the spring and summer months the predicted MUFs were about 5 MHz below the corresponding observed medians; and (3) during the early morning hours during the winter, the transmissions on the Andøya–College circuit (approximately normal to the auroral oval) were more reliable than were those on the Andøya– Fort Monmouth path (tangential to the auroral oval). This is qualitatively explainable by invoking the greater amount of time the second path spends in the

8.4 HF propagation

Figure 8.89. The Boulder–Barrow HF propagation path with the auroral oval for low Kp (from Tveten, 1962).

AA regions. Examples of circuit behavior during selected disturbed periods in April 1964 are shown in Figure 8.88. Another early HF high-latitude propagation experiment was conducted in April–June 1960 (SSN113–120) on the Barrow, Alaska to Boulder, Colorado path (D4495 km) by the US National Bureau of Standards (NBS) and reported by Tveten (1961). The experiment utilized two modified C-3 ionosondes in a synchronized sweep-frequency sounder system. The ionosondes transmitted 100-s pulses at a PRF of 25 pulse s1 with an output power of ⬃10 kW using terminated horizontal V antennas approximately 400 ft long on each leg with the apex 70 ft high. The records were taken at the rate of one 7.5-min sweep every hour and ended at 7.5 min past the hour. Figure 8.89 is a map of the Boulder–Barrow path with an estimated auroral oval for low Kp. It is obvious that only the northern half of the path will be affected by auroral phenomena.

513

Figure 8.90. Examples of oblique ionograms obtained for summer, 2207 MST, 1 June 1960 (upper), and equinox. 2200 MST, 20 April 1960 (lower), near sunspot maximum on the Boulder–Barrow propagation path (from Tveten, 1962).

8.4 HF propagation

Figure 8.91. A comparison of observed and predicted MUFs for the Boulder–Barrow path for April 1960 (from Tveten, 1962).

Typical oblique ionograms from the Barrow–Boulder path for summer and equinoctial periods are shown in Figures 8.90(a) and (b), respectively. The missing low-angle ray on the 1F2 (one-hop F2-layer mode) in the ionogram in Figure 8.90(b) is probably due to shielding by the AE layer on the northern end of the path. The NBS/CRPL “two-control-point” method for computing the 4000-km MUF was employed to compare produced values with values observed on this path and the results are shown in Figures 8.91 and 8.92 for April and June 1960, showing the large discrepancies typical of this type of path. Some limited data on a very long path from McMurdo (Antarctica) to Thule (Greenland) for sunspot-maximum conditions were presented by Gerson (1964). This path was 18730 km long and was possibly affected both by the northern and by the southern auroral ovals. Frequencies of 13 and 17 MHz and output powers of 0.5–1.0 kW into delta antennas were used. Figure 8.93 is a plot of periods when the McMurdo transmissions were received at Thule during the period 15–17 May 1958 (SSN191). A minor SWF was observed on 17 May. Results from a well-instrumented and documented high-latitude HFpropagation experiment were reported by Jull (1964) shortly after the maximum of solar cycle 19 (1960 and 1961). Five propagation paths in the polar, auroral, and subauroral regions were studied using synchronized oblique-sounding systems and a network of six vertical 30 MHz riometers. Figure 8.94 shows the

515

516

High-latitude propagation: 1

Figure 8.92. The same as Figure 8.91, but for June 1960 (from Tveten, 1962)

Figure 8.93. Periods of reception of 13- and 17-MHz transmissions from McMurdo to Thule during 15–19 May 1958 (from Gerson, 1964).

location of the HF sounding paths and the riometers and the path characteristics are tabulated in Table 8.13. Except for PCAs, the attenuation of HF signal on these circuits is due to AA, and the relative occurrence of absorption at the six riometer stations from July 1959 to June 1961 is shown in Figure 8.95 and the percentages of the AA time occurred for various values of Kp are shown in Figure 8.96. Although they were obtained some 38 years ago, these two figures remain quite useful for estimating effects of AA on HF circuits. The statistical distributions of AA are described in detail in Section 7.2.

8.4 HF propagation

517

Figure 8.94. A map showing the location of five HF forward-sounding circuits and supporting riometers in relation to an idealized polar-cap absorption area for the Canadian DRTE propagation experiment in 1960–1961 (after Jull, 1964).

Table 8.13. Riometers and path characteristics

HF forward-sounding circuits

Path-length (km)

N–S subauroral: OT–Ch N–S transauroral: OT–RB N–S innerauroral: Ch–RB Trans-Atlantic: OT–HA Ground–air: HAL–A/C

1900 3400 1830 5640 0–2520

Relevant riometer stations North

South

Ch CH RB

VD CJ Ch

Notes: OT, Ottawa; Ch, Coral Harbour; RB, Resolute Bay; HA, The Hague; HAL, Halifax.

Table 8.14 gives estimates of attenuation on three paths due to AA extrapolated from the riometer data. The absorption is calculated from the 30-MHz riometer vertical-absorption values using the inverse frequency-squared relation for the one-hop F-layer mode. The behavior of the change in the lowest-usable frequency (LUF) – which is closely related to absorption – on four of these paths during an intense PCA is shown in Figure 8.97.

518

High-latitude propagation: 1

Figure 8.95. The percentage of time (solid line) and percentage of half-hour periods (dashed line) for which auroral absorption equaled or exceeded 1.0 dB, as functions of geomagnetic latitude. Locations of the stations are indicated by the two-letter abbreviations on the abscissa (after T. R. Hartz, L. E. Montbriand and E. L. Vogan. A study of auroral absorption at 30 Mc/s. Can. J. Phys., 41, 581 (1963).)

Figure 8.96. The time-percentage occurrence of auroral absorption as a function of Kp for 1959–1961 (after T. R. Hartz, L. E. Montbriand and E. L. Vogan. A study of auroral absorption at 30 Mc/s. Can. J. Phys., 41, 581 (1963).)

8.4 HF propagation

519

Table 8.14. Extrapolation of 1-dB cosmic-noise absorption to attenuation of 10 MHz one-hop F-layer transmissions

Circuit

Absorption on the north side of the path only (dB)

Absorption on the south side of the path only (dB)

Absorption on north and south sides of the path (dB)

OT–Ch OT–RB Ch–RB

29 46 32

25 43 29

54 89 61

Note: OT, Ottawa; Ch, Coral Harbour; RB, Resolute Bay.

Figure 8.97. July 1961 PCA effects on four HF circuits: (a) for the 30-MHz riometer at Resolute Bay; and (b) LUFs.

520

High-latitude propagation: 1 Figure 8.98. The flight plan for ground–air trials of 15–16 December 1960 (from Jull, 1962)

A unique part of Jull’s (1962) HF high-latitude-propagation experiment was the monitoring of 3–23-MHz transmissions from Halifax, Nova Scotia by an aircraft flying in the subauroral and auroral regions during disturbed periods. In particular, one ground–air trial was flown on 15–16 December 1960 (SSN83) during a minor geomagnetic storm (Kp 4). The flight plan is shown in Figure 8.98 and the observed MUFs and LUFs as functions of time and location are shown in Figure 8.99. During the flight it was found that taking soundings every 5 min provided sufficient information to select proper operating frequencies for ground–air communication. It was further found from this study that selection of communications frequencies once every hour instead of selecting frequency changes on the basis of sounding data would have resulted in the aircraft being out of contact with the ground station for 30% of the flight period. Some conclusions of Jull et al. (1964)

8.4 HF propagation

(a)

(b)

Figure 8.99. The MUF and LUF observed during the ground–air trials of 15–16 December 1960: (a) the outgoing leg of the flight; and (b) the incoming leg of the flight (from Jull, 1964).

521

522

High-latitude propagation: 1

Figure 8.100. A map showing the Wales-to-Fairbanks, Alaska propagation path in relation to the equatorward edge of the auroral oval as a function of Kp.

were that, during PCA events of low or moderate intensity, the optimum traffic routing is via AE in the oval and the optimum routing is through intermediate relay stations. The characteristics of 25.5-MHz one-hop propagation on a 950-km Alaskan path over a 14-month period shortly after the maximum of solar cycle 22 were presented as a function of Kp by Hunsucker et al. (1996). The location of the E-region reflection point was within the auroral oval for 3Kp 5 and the specific behavior of the signal was related to auroral-oval phenomena such as substorms, geomagnetic storms, and the Harang discontinuity. The location of the auroral electrojet with respect to the mid-point of the path was also found to be of considerable importance. A map showing the Wales to Fairbanks, Alaska path in relation to the equatorward edge of the auroral oval as a function of Kp is shown in Figure 8.100 and an example of the typical behavior of signal amplitude is shown in Figure 8.101. It is reasonable to assume that the AE mode (defined on p. 480) is uncontaminated by F-layer propagation, because, during the period of maximum occurrence

Figure 8.101. An example of the variation in amplitude of the 25.5-MHz signal, along with the amplitude of the auroral electrojet (derived from the Earth-current recorder at Fairbanks) for 23 November 1991 (from Hunsucker, et al., 1996).

524

High-latitude propagation: 1

of AE (⬃2100–0400 LT) – especially September through March – data from an ionosonde and an incoherent-scatter radar located near Fairbanks have shown that there is not enough F-region ionization present to support an F-layer mode; also the antenna takeoff angles, path distance, and operating frequency tend to exclude the possibility of an F-mode. An AE “burst” was defined as a signal received for 2 min or more. Burst duration, date/time of start of the burst, signal strength in decibels, and Earth-current amplitude and direction data were scaled from strip-charts, and Ap and Kp values were added and tabulated on spreadsheets. Figures 8.102, 8.103, and 8.104 show the diurnal and seasonal behaviors for winter, spring, and summer of 1992, respectively and Figure 8.105 shows the occurrence of AE as a function of Kp. The seasonal characteristics of the AE bursts are illustrated in Table 8.15. A schematic representation of the Harang discontinuity showing the eastward and westward electrojets is shown in Figure 8.106 and the responses of the AE signal and Earth current are shown in Figure 8.107. The high-latitude current systems are discussed in Sections 2.5.3 and 6.4.4. From analysis of the 14 months of data obtained during 1991–1992, it was found that the AE signal was very “bursty” in character, with bursts lasting from 1 min to over 3 h, with an average duration of 11 min and an average signal amplitude 20–30 dB above the detection threshold of 115 dB m for the receiver. Out of 1445 observations, 981 events (68%) lasted 10 min, 234 (16%) had durations between 11 and 20 min, 90 (6%) had durations between 21 and 30 min, and 11 had durations greater than 90 min. One of these “long” events occurred in the Fall, and the rest occurred in the Spring or Summer. Although the signal characteristics are quite poorly correlated to Kp, they are qualitatively correlated to the local magnetic indices and to Earth-current data at the receiving site in Fairbanks. The behavior of the Wales AE signal on 25.5 MHz received at Fairbanks very closely resembled the occurrence statistics of the visual aurora and VHF/UHF auroral-radar results. It is believed that this is the first quantitative demonstration of the “forward-propagation behavior” of a one-hop “auroral-E” path near the upper end of the HF band and it was suggested that it might be possible to utilize the “auroral-E-burst” mode for data transmission and/or communication over path-lengths on the order of 1000 km inside and parallel to the auroral oval or to enhance the meteor-burst-communication (MBC) mode at high latitudes. For 2 weeks during the period of observations of the Wales–Fairbanks AE experiment, Wagner et al. (1995) conducted a “HF-channel-probe” experiment on a 1294-km path between Sondrestrom, Geenland and Keflavík, Iceland. The channel probe measured delay, Doppler, and amplitude characteristics on this path. The equipment parameters, paths and other characteristics of these two experiments are compared in Table 8.16. Results from the Wagner et al. (1995) HF-channel-probe observations are summarized as: (1) strong, specularly reflected ionospheric returns characteristic of a

8.4 HF propagation

Figure 8.102. The occurrence of AE for Winter 1991–1992 (from Hunsucker et al., 1996).

Figure 8.103. The occurrence of AE for Spring 1992 (from Hunsucker et al., 1996).

525

526

High-latitude propagation: 1

Figure 8.104. The occurrence of AE for Summer 1992 (from Hunsucker et al., 1996).

Figure 8.105. The occurrence of AE as a function of Kp (from Hunsucker et al., 1996).

8.4 HF propagation

527

SUN 12h 50° 60° 70°

18h

6h

t

je

W es

tro

tw

ec

ar d

El

El

rd

ec

wa

tro

je

st

t

Ea

Harang Disc ontinuity

Figure 8.106. A schematic representation of the Harang discontinuity, illustrating the direction of flow of equivalent current in the electrojet.

Table 8.15. Seasonal characteristics of AE signals recorded in Fairbanks, Alaska (from Hunsucker et al., 1996)

Season

Average duration (min)

Average amplitude (dB m)

No. of events of duration exceeding 60 min

Longest duration observed (min)

No. of events

Autumn (August–October 1991)

9.9

17.4

7

120

403

Winter (November–December 1991, January 1992)

8.3

19.0

2

84

383

Spring (February–April 1992)

8.6

18.4

1

65

272

21.0

19.2

21

192

388

Summer (May–July 1992)

528

High-latitude propagation: 1

Figure 8.107. An example of simultaneous recording of Earth-current measured at Fairbanks, Alaska and Wales AE signal amplitude recorded at Fairbanks, on 9 October 1991 (Kp 7; Ap 101) (from Hunsucker et al., 1996).

quiescent daytime ionospheric channel during magnetically quiet conditions; (2) strong specular multipath signals reflected from horizontal gradients of electron density – which are regularly encountered at night; (3) weak scatter returns, also persistent at night; and (4) maximum Doppler shifts of ⬃16 Hz at 7.5 MHz near midnight (E layer ) and a maximum Doppler shift of ⬃22 Hz at 14.5 MHz near midnight. They infer that the drift speed of irregularities is ⬃1200 m s1 parallel to the great-circle propagation path for the midday “disturbed” ionosphere. It appears that most of the observations discussed were for subauroral ionosphere conditions. Warrington et al. (1997) have analyzed data on two paths: one within the polar cap (Clyde River on Baffin Island to Alert, Canadian NWT, 1345 km) and Clyde River to Prudhoe Bay, Alaska, 2955 km). Measurements of several parameters, including Doppler spreading, were made during July and August 1989 and during January and February 1989 (near the maximum of sunspot cycle 22).

8.4 HF propagation

529

Table 8.16. A comparison of pertinent parameters for the Wales–Fairbanks and the Sondrestrom–Keflavík HF experiments Wales–Fairbanks HF experiment

Sondrestom–Keflavík HF experiment

Path location

In auroral oval for 5Kp 3 Subauroral, except for Kp 5

Mid-point of path Path length Dominant mode Frequency

66° CGL 950 km Auroral-E 25.545 MHz

72° CGL 1294 km E and some F 3–11 MHz

Transmission mode

CW

Complex pulse simultaneous with Chirpsounder

Transmitter power

100 W CW

170 W pulse

Duration of experiment 14 months (July 1991–September 1992)

⬃2 weeks (13 March–2 April 1992)

Transmissions were made using 2-min sequences once per hour on each of 14 frequencies from 3 to 23 MHz; each sequence included a 30-s period of CW transmission during which the Doppler spectrum of the received signal was measured. The Doppler spreading was quantified in terms of the area under the normalized signal-amplitude spectrum, minus the area estimated to be due to noise – the resulting area was multiplied by 20 and the product, referred to as the Doppler spread index (DSI), employed as a measure of the Doppler spread. Within the polar cap (Alert to Clyde River) the DSI varied between ⬃10 and 30 Hz in the summer and between ⬃30 to 75 Hz during the winter, whereas the long path (Prudhoe–Clyde River) had a mean DSI of 60 Hz in the summer and 90 Hz during winter months. Specific details of the Doppler spread as a function of frequency, path, and magnetic activity are presented in their paper. It was not possible to accurately specify the time when the principal ionospheric reflection points were actually within the auroral oval.

8.5

VHF/UHF and microwave propagation

The international frequency-band delineations (LF/MF/HF/VHF, etc.) are somewhat arbitrary but still useful. Instead of defining the bands in decades, it would be better to define the bands in terms of their propagation behaviors in terrestrial atmospheric regions. Propagation in the VHF through UHF bands (30 MHz to 3 GHz) and microwaves (⬃1–10 GHz) is either LOS in the troposphere, with typical path lengths of 30–50 km, or via satellite–Earth links. Until the advent of satellite communications

530

High-latitude propagation: 1

in the 1960s, the “backbone” of the USA’s trans-continental communications system was the Bell microwave relay system. More rarely used modes are the beyond-line-of-sight modes, ionospheric scatter (ionoscatter) using frequencies from 30 to 150 MHz over path-lengths of 1000–2000 km, and troposcatter using frequencies from 200 MHz up to 19 GHz on path-lengths from 300 to 600 km. Since the forward-scattered energy for ionoscatter and troposcatter links is extremely weak (compared with that for LOS paths), scatter systems must utilize high transmitter power, very large antenna apertures, high-gain receiver front-ends and multiple diversity. These systems provided very high circuit reliability (99.9%) for high-security communications, but were very costly to install and maintain. The “White Alice” system supplied the communications between the “Dewline” (Distant Early Warning) radar system in the northern USA and Canada from the late 1950s until satellite communication came into use. Another scatter mode that is still in use is meteor scatter (from meteor ionization trails in the E region) using frequencies from ⬃40 to 150 MHz, because such systems offer very secure and survivable communications. It is a signal burst system with typical bandwidths of 100 kHz, Doppler spread of 5 Hz, and an information duty cycle of ⬃5% (see Davies, 1990; Section 13.4; and Weitzen, 1988). Millimeter waves ( f ⬃1.50–13 GHz) have also been used for terrestrial LOS communication links, but such use is limited by rather severe atmospheric absorption and high-latitude effects are not very well documented. Advances (in the last 20 years or so) in our understanding of the phenomena of VHF terrestrial propagation at high latitudes stem primarily from studies of the effects of AE ionization on MBC, and studies of trans-ionospheric propagation in the development of morphological models of scintillation effects. The latter subject will be covered in Section 9.2.2 and we will list salient effects of the former subject herewith. Meteor scatter was developed as a relatively inexpensive, high-data-rate, secure communication system primarily for the military, using frequencies typically in the 40–104-MHz region. Cannon et al. (1985) described results from a MBC propagation experiment involving transmission between Bodo, Norway and Wick, Scotland at 40 and 70 MHz. It was found that excess D-region ionization produced a rotation of polarization, which caused some deterioration of normal system performance and they also concluded that frequencies close to 40 MHz may be too low for use at high latitudes. In another investigation, Ostergaard et al. (1985) reported the advantages of using adaptive techniques to improve system performance on a 1200-km path in northern Greenland and qualitatively described some of the effects of AE ionization, irregularities, and D-region absorption on MBC systems. The applicability of adaptive antenna beam steering to the prediction model for MBC systems, including high-latitude effects, was reported by Akram and Cannon (1994). Specifically, the prediction models gave good results during the winter and

8.6 Summary

equinoctial months but poor agreement during the summer on the Sondrestrom– Narsarsuaq, Geenland path. Cannon et al. (1996) found that, on the Greenland paths at 35 and 45 MHz, MBC is sustained by E-region ionization and at 65 and 85 MHz the path is dominated by meteor-scatter modes. Although it is not due to the high-latitude ionosphere, VHF/UHF propagation in the Arctic and Antarctic regions frequently displays anomalous behavior. Kennedy and Rupar (1994) describe the Arctic Unattended Propagation Experiment (ARUPEX) on the north slope of Alaska, which operates on a pathlength of 50.9 km at 142.875 and 420.5 MHz. Many instances of VHF/UHF ducting and diffraction anomalies were observed.

8.6

Summary

A considerable amount of useful data on the phenomena of HF polar and auroral propagation was obtained during the period c. 1956–1997 and presented in this chapter, providing at least some qualitative indications for ameliorating the most deleterious effects. An extreme range of solar activity occurred during this period, from the maximum of solar cycle 19 in March 1958 (SSN201.3) to a minimum value of SSN9.6 in October 1964, providing “worst-case scenarios” for highlatitude HF propagation. At ELF/VLF frequencies, the most profound effects are probably those associated with polar-cap absorption (PCA) – also known as solar-proton events (SPEs). Refer to Section 7.3. These events result in a lowering of the reflection height of the D region, which changes the dimensions of the Earth–ionosphere waveguide, which then produces changes in the phase and amplitude of the ELF/VLF signals. There seems to be no firm quantitative evidence that these changes produce serious effects on ELF transmissions received in submerged submarines, but there is some evidence that polar effects on VLF transmissions might degrade the accuracy of certain navigational systems. An investigation in Alaska and the northern tier of the continental USA has revealed that E-region ionization in the auroral oval called auroral-E (AE) can strongly influence MF skywave propagation at night. A 5-year monitoring program revealed the diurnal, seasonal, and sunspot-cycle behavior of MF skywaves at high latitudes and resulted in the US Federal Communications Commission issuing new skywave curves describing possible skywave interference between standard AM broadcasting stations in the northern tier of the USA, Alaska and Canada, thus making channel assignments more realisitc. Since skywave propagation dominates at high frequencies, the polar and auroral-oval ionosphere profoundly affect HF propagation at high latitudes, and, during the period 1956–1969, there were many studies of the behavior of HF polar and auroral circuits. These studies revealed that the primary disturbance parameters were PCAs, AA events, AE ionization and F1-layer effects. The behavior of

531

High-latitude propagation: 1

532

HF propagation at high latitudes, then, is determined by the location of the propagation path in relation to the intersections of the path with the auroral and polar D region and the E-region and F-region reflection points. Because of the complexity of HF high-latitude mode structure and ionosopheric intersections, it is not possible to do accurate three-dimensional ray-tracing unless one has an accurate three-dimensional realtime description of the irregular structure of the highlatitude ionosphere – thus making it very difficult to devise reliable HFpropagation-prediction programs. Other important propagation phenomena on high-latitude HF paths (compared with mid-latitude paths) are increases in Doppler shift and spread, fading and non-great-circle (NGC) propagation. At certain times and on certain paths, the maximum operating frequency (MOF) may be carried by AE ionization, F1-layer effects, the NGC propagation mode, or possibly ducted modes. Studies during this period also revealed that VHF frequencies as high as 32 MHz on trans-polar paths and 46 MHz on a path from Thule, Greenland to College, Alaska were possible during periods of sunspot maximum. The effects of the polar and auroral ionosphere on trans-ionospheric signals will be described in the next chapter.

8.7

References and bibliography

Section 8.1 Croft, T. A. (1968) Skywave backscatter: a means for observing our environment at great distance. Rev. Geophys. Space Phys. 10, 73–155. Davies, K. (1990) Ionospheric Radio. Peter Peregrinus Press, on behalf of the Institute of Electrical Engineers, London. Deehr, C. S. and Holtet J. A. (eds.) (1981) Exploration of the polar upper atmosphere. Proc. NATO Advanced Study Institute held at Lillehammer, Norway; 5–16 May 1980. Reidel, Dordrecht. Folkestad, K. (1968) Ionospheric Radio Communications. Plenum Press, New York. Goodman, J. (1992) HF Communication – Science and Technology. Van Nostrand Reinhold, New York. Hunsucker, R. D. (1967) HF propagation at high latitudes, QST Mag. February, 16–19 and 132. Hunsucker, R. D. (1971) Characteristic signatures of the midlatitude ionosphere observed with a narrow-beam HF backscatter sounder. Radio Sci. 6535–548. Hunsucker, R. D. (1991) Radio Techniques for Probing the Terrestrial Ionosphere. Springer-Verlag, Heidelberg. Hunsucker, R. D. (1992) Auroral and polar-cap ionospheric effects on radio propagation. IEEE Trans. Antennas Propagation 40, 818–828. Hunsucker, R. D. and Bates, H. F. (1969) Survey of polar and auroral region effects on HF propagation. Radio Sci. 4 347–365.

8.7 References and bibliography

Landmark, R. (ed.) (1964) Arctic Communications. Published on behalf of NATA/AGARD; Pergamon Press, New York. Lied, F. (1967) Arctic Communications, with Emphasis on Polar Problems. AGARDograph 104; Technivision; Maidenhead. Rawer, K. (1976) Manual on Ionospheric Absorption Measurements. World Data Center A Solar–Terrestrial Physics, Boulder, Colorado. Soicher, H. (ed.) (1985) Propagation effects on military systems in the high-latitude region. In Proc. AGARD Conference, CP-382.

Section 8.2 Albee, P. R. and Bates, H. F. (1965) VLF observations at College, Alaska of various D-region disturbance phenomena. Planet. Space Sci. 13, 175–206. Bannister, P. (1993) ELF propagation highlights. In AGARD Conference Proc. 529, pp. 2-1–2-15. Bates, H. F. (1961) An HF Sweep-frequency Study of the Arctic Ionosphere. Geophysical Institute,University of Alaska, College, Alaska. Bates, H. F. (1962) VLF effects from the Nov. 10, 1961 polar-cap absorption event, J. Geophys. Res., 67, 2745–2751. Bates, H. F and Albee, P. R. (1965) General VLF phase variations observed at College, Alaska. J. Geophys. Res. 70, 2187–2208. Berry, L. A. (1964) Wave hop theory of long distance propagation of low-frequency radio waves. Radio Sci. D 68, 12. Chrissan, D. A. and Fraser-Smith, A. C. (1996) Seasonal variations of globally measured ELF/VLF radio noise. Radio Sci. 31, 1141–1152. Davies, K. (ed.) (1970) Phase and frequency instabilities in electromagnetic wave propagation. AGARD Conference Proc. 33. Technivision Services, Slough. Fraser-Smith, A. C. and Bannister, P. R. (1998) Reception of ELF signals at antipodal distances. Radio Sci. 33, 83–88. Wait, J. R. (1970) Electromagnetic Waves in Stratified Media. Pergamon Press, Oxford. Wait, J. R. (1991) EM scattering from a vertical column of ionization in the earth– ionosphere waveguide. IEEE Trans. Antennas Propagation 39, 1051–1054. Watt, A. D. (1967) VLF Radio Engineering. Pergamon Press, Oxford. Weitzen, J. A. (1988) Meteor scatter propagation. IEEE Trans. Antennas Propagation 37, 1813.

Section 8.3 Hunsucker, R. D., Delana, B. S., and Wang, J. C. H. (1987) Effects of the February 1986 magnetic storm on medium frequency skywave signal received at Fairbanks, Alaska. Proc. IES87, 197–204. Hunsucker, R. D. and Delana, B. S. (1988) High Latitude Field-strength Measurements of Standard Broadcast Band Skywave Transmissions Monitored at Fairbanks, Alaska. Geophysical Institute, University of Alaska, Fairbanks, Alaska.

533

534

High-latitude propagation: 1

Section 8.4 Auterman, J. L. (1962) Fading correlation bandwidth and short-term frequency stability measurements on a high-frequency transauroral path. NBS Tech. Note 165. Bartholomew, R. R. (1966) Results of a High-latitude HF Backscatter Study. Stanford Research Institute, Menlo Park, California. Bates, H. F. and Hunsucker, R. D. (1964) HF/VHF Auroral and Polar Zone Forward Sounding. Geophysical Institute, University of Alaska, Fairbanks, Alaska. Bates, H. F and Albee, P. R. (1966) On the Strong Influence of the F1 Layer on Medium to High Latitude HF Propagation. Geophysical Institute, University of Alaska, Fairbanks, Alaska. Bates, H. F., Albee, P. R., and Hunsucker, R. D. (1966) On the relationship of the aurora to non-great-circle HF propagation. J. Geophys. Res. 71, 1413–1420. Bates, H. F. and Hunsucker, R. D. (1974) Quiet and disturbed electron density profiles in the auroral zone ionosphere. Radio Sci. 9, 455–467. Egan, R. D. and Peterson, A. M. (1962) Backscatter observations of sporadic-E. In Ionospheric Sporadic-E (ed. E. K. Smith), p. 9. Gerson, N. C. (1964) Polar communications. In Arctic Communications (ed. B. Landmark), p. 83. Pergamon Press, Oxford. Goodman, J. M. (1992) HF Communication – Science and Technology. Van Nostrand Reinhold, New York. Hartz, T. R., Montbriand, L. E., and Vogan, E. L. (1963) Can. J. Phys. 41, 581. Heppner, J. P., Byrne, E. C., and Belon, A. E. (1952) The association of absorption and Es ionization with aurora at high latitudes. J.Geophys. Res. 57, 121–134. Hunsucker, R. D. and Stark, R. (1959) Oblique fixed-frequency soundings. In Final Report on Contract No. AF 19(604)–1859 (ed. L. Owren). Hunsucker, R. D. and Owren, L. (1962) Auroral sporadic-E ionization. J. Res. NBS Radio Propagation D 66, 581–592. Hunsucker, R. D. (1964a) Auroral absorption effects on a transpolar synchronized step-frequency circuit. Proc. IEEE, 52, March. Hunsucker, R. D. (1964b) Auroral-zone absorption effects on an HF arctic propagation path. Radio Sci. D 68, 717–721. Hunsucker, R. D. (1965) On the determination of the electron density within discrete auroral forms in the E-region. J. Geophys. Res. 70, 3791–3792. Hunsucker, R. D., Rose, R. B., Adler, R., and Lott, G. K. (1996) Auroral-E mode oblique HF propagation and its dependence on auroral oval position. IEEE Trans. Antennas Propagation 44, 383–388. Jelly, D. H. (1963) J. Geophys. Res. 68, 1705. Jull, G. W. (1964) HF propagation in the Arctic. In Arctic Communications (ed. B. Landmark), pp. 157–176. Pergamon Press, Oxford. Koch, J. W. and Petrie, L. E. (1962) Fading characteristics observed on a high frequency auroral radio path. J. Res. NBS Radio Propagation D 66, 159–166. Leighton, H. I., Shapley, A. H., and Smith, E. K. (1962) The occurrence of sporadic-E during the IGY. In Ionospheric Sporadic-E (ed. S. Matsushita and E. K. Smith), p. 166. MacMillan, London.

8.7 References and bibliography

Lomax, J. B. (1967) High-frequency Propagation Dispersion. Stanford Research Institute, Menlo Park, California. McNamara, L. (1991) The Ionosphere: Communications, Surveillance and Direction Finding. Krieger Publishing Co., Malabar, Florida. Maslin, N. (1987) HF Communications – A System Approach. Plenum Press, New York. Moller, H. G. (1964) Backscatter observations at Lindau-Hartz with variable frequency directed to the auroral zone. In Arctic Communications (ed. B. Landmark), pp. 177–188. Ortner, L. and Owren, L. (1961) Multipath Propagation on Transarctic HF Circuits. Kiruna Geophysical Observatory, Kiruna. Ostergaard, J. C., Rasmussen, J. E., Sowa, M. J., McQuinn, J. M., and Kossey, P. A. (1985) Characteristics of high-latitude meteor scatter propagation parameters over the 45–104 Mhz band. In Proc. AGARD (NATO) Conference. Owren, L., et al. (1959) Arctic Propagation Studies at Tropospheric And Ionospheric Modes of Propagation. Geophysical Institute, University of Alaska, College, Alaska. Owren, L. (1961) Influence of solar particle radiations on Arctic HF propagation, presented at the AGARD Ionospheric Research Communication meeting, Naples, 15–20 May. Owren, L., Ortner, J., Folkestad, K., and Hunsucker, R. D. (1963) Arctic Propagation at Ionospheric Modes of Propagation. Geophysical Institute, University of Alaska, College, Alaska. Peterson, A. M., Egan, R. D., and Pratt, D. S. (1959) The IGY three-frequency backscatter sounder. Proc. IRE 47, 300–314. Rose, G. (1964) Field strength measurements over a 2000 km subauroral path (Sodankylä–Lindau) compared with the absorption observed at the terminals. In Arctic Communications (ed. B. Landmark). Pergamon Press, Oxford. Tveten, L. H. (1961) Ionospheric motions observed with high-frequency backscatter sounders. J. Res. NBS D 65, 115–127. Tveten, L. H. (1961) Long-distance one-hop F1 propagation through the auroral zone. J. Geophys. Res. 66, 1683–1684. Warrington, E. M., Dhanda, B. S. and Jones, T. B. (1997) Observations of Doppler spreading and FSK signaling errors on HF signals propagating over a high-latitude path. Proc. IEE, 6th International Conference on HF Radio Systems and Techniques, pp. 119–123. Weitzen, J. A., Cannon, P. S., Ostergaard, J. C., and Rasmussen, J. E. (1993) Highlatitude seasonal variation of meteoric and nonmeteoric oblique propagation at a frequency of 45 MHz. Radio Sci. 28, 213–222.

Section 8.5 Akrun and Cannon, P. S. (1994) A meteor scatter communication system data throughput model. IEE HF Radio Systems and Techniques Conference, University of York, Vol. 392, pp. 343–347. Cannon, P. S., Dickson, A. H., and Armstrong, M. H. (1985) Meteor scatter communication at high latitudes. In Proc. AGARD (NATO) Conference.

535

536

High-latitude propagation: 1

Cannon, P. S., Weitzen, J. A., and Ostergaard, J. (1996) The relative impact of meteor scatter and other long distance high latitude propagation modes on VHF communication systems. Radio Sci. 31. Ostergaard, J. C., Rasmussen, J. E., Sowa, M. J., McQuinn, J. M., and Kossey, P. A. (1985) Characteristics of high-latitude meteor scatter propagation parameters over the 45–104 Mhz band. In Proc. AGARD (NATO) Conference. Weitzen, J. A. (1988) Meteor scatter propagation. IEEE Trans. Antennas Propagation 37, 1813.

Chapter 9 High-latitude radio propagation: part 2 – modeling, predictions, and mitigation of problems There are no such things as applied sciences, only applications of science. Louis Pasteur

9.1

Introduction

In Chapter 8 we reviewed the progress of our understanding of high-latitude radio propagation starting about 1956 when it was deemed to be a problem worth investigating, and continuing through the IGY, IGC, and IQSY international study periods until the present time. In the last 20 years we have made considerable progress in our level of understanding of the phenomena both of auroral and of polar radio propagation and there has been a “sea-change” in communications and computer technology. This forward leap in technology includes the availability of powerful, inexpensive computers and prediction/modeling/ray-tracing software, sophisticated modulation schemes, advanced antenna theory and practice, electronic-circuit VLSI, advanced ground-based and satellite-borne geophysical sensors, and active-circuit sounding systems (see Chapter 4). This chapter will concentrate on experimental results obtained starting in the late 1980s, ionospheric modeling, ray-tracing, prediction techniques, mitigation techniques and the impact of space-weather data on ionospheric propagation. The morphology of auroral-E (AE) propagation in the 25–30-MHz frequency range on ⬃1000–2000-km paths tangential and normal to the auroral oval has been documented by Hunsucker et al. (1996) and Nishino et al. (1999).

537

High-latitude propagation: 2

538

9.2

Ionospheric ray-tracing, modeling, and prediction of propagation

9.2.1

Ionospheric ray-tracing

In order to make useful predictions by applying ionospheric ray-tracing programs at high latitudes, one must have an accurate model of electron-density profiles at a sufficient number of points along the propagation path. Since most of the sophisticated ionospheric models available produce basically climatology (not “weather”) outputs on a fairly sparse data grid, they are at present not adequate to define the high-latitude ionosphere for the level of ray-tracing needed for prediction purposes. Additionally, none of these models includes D-region absorption in the polar or auroral ionosphere (see Sections 7.2 and 7.3), which is a first-order effect on HF propagation. Not all radio-propagation-prediction programs utilize a ray-tracing algorithm; some use a “virtual-geometry” technique whereas others base their predictions on a data base of actual forward-sounding circuits. The ray-tracing and virtual-geometry algorithms, of course, are very dependent on accurate ionospheric models and most of the data-based algorithms are data-sparse for high-latitude regions. Examples of the type of ray trace obtained in the auroral ionosphere are given in Figures 9.1–9.3 The Jones–Stephenson (1975) three-dimensional ray-tracing program was used with a model based on parabolic fits to the Fairbanks verticalincidence E- and F-region parameters to produce backscatter ray traces at three different azimuths. These ray traces illustrate the complex three-dimensional structure of the auroral ionosphere. 9.2.2

Current high-latitude models

Criteria for deciding the applicability of the numerous ionospheric models to adequately describe the high-latitude ionosphere include prediction of the ionization profile from the lower D-region up to the upper F-region (⬃500 km), polar plasma convection and the behavior of ionospheric currents, latitudinal coverage from ⬃55° to 90° CGL, a sufficiently dense grid of observations (cell dimension no larger than ⬃100 km), and realtime “space-weather” data input. Of the 16 ionospheric models listed in the STEP Handbook by Schunk (1996), 11 include some high-latitude ionospheric parameters. Table 9.1 lists the high-latitude models from Schunk (1996) plus the International Reference Ionosphere (IRI) model and the Parameterized Realtime Ionospheric Specifications Model (PRISM). Three other earlier ionospheric models have been used rather extensively in HF-propagation-prediction programs (Bent et al., 1975; Chiu, 1975; Rush et al. 1984) and, although they are “global models,” they are seriously lacking in effective high-latitude data. Apropos the scintillation models (Section 5.3.3), Aarons et al. (1995) emphasized

Figure 9.1. A backscatter ray-tracing from Fairbanks in fall 1988 for a backscatter sounding at a frequency of 11.3 MHz on a true azimuth of 10°, based on a parabolic layer model using a vertical ionosonde (from Hunsucker and Delana, 1988).

Figure 9.2. A backscatter ray-tracing as shown in Figure 9.1, but at an azimuth of 16.0° (from Hunsucker and Delana, 1988).

Figure 9.3. A backscatter ray-tracing as in Figure 9.2, but at an azimuth of 31.0° (from Hunsucker and Delana, 1988).

Type of model and inputs

First principles plus MSIS-90, “solar–terrestrial activity”

Semi-empirical

First principles, solar EUV, energy from ISR

First principles, MSIS-90, auroral precipitation solar EUV

First principles, VSH/MSIS, magnetic indices, IMF By and Bz; auroral imager data

First principles, MSIS, magnetic indices, ISR data, daily 10.7-cm flux

Model

D-region ion-chemistry model (Sodankylä ion data from Chemistry Model – SIC – Turunen et al.

Steady-state D-region model (Swider)

Eight-moment fluid Models (TRANSCAR) (Blelly)

Graz Ionospheric Flux Tube Simulation (GIFTS) model (Kirchengast)

UAF Eulerian model of polar ionosphere (Maurits and Watkins)

“High-latitude model” (Wu and Taieb)

High-latitude F region h300–2000 km

Covers geographic latitudes 50°–90°, h80–500 km

High-latitude F region (h150–600 km)

“High-latitude model” Height range 100–3000 km, includes convection, electric-field and Joule–heating effects

Good agreement with PCA data

Height range 70–100 km, includes particle precipitation

High-latitude features

Table 9.1. Applicability of ionospheric models to the high-latitude ionosphere (from Schunk, 1996)

No longer available

Available from the authors

Available from the authors in FORTRAN for DEC/VAX platforms

Available from the authors for workstations

Swider and Foley (1978) Available from the authors through the NTIS

Applicable to ISR and riometer absorption data; for PC’s in MATLAB 4.2

References and comments

Includes auroral regions; h80–600 km

First principles, threedimensional Euler–Lagrange hybrid, with some data from empirical models; the inputs are global, time-dependent data from models

First principles, one-dimensional

model, MSIS-86, solar EUV, magnetic indices, solar flux

First principles, statistical models for most inputs, threedimensional Eulerian

USU Global Ionospheric Model (Schunk and Sojka)

Field Line Inter-hemispherical on Model (FLIP) (Richards and Torr)

Coupled Thermosphere Ionosphere Model (CTIM) Fuller–Rowell et al.)

Global model, h90–600 km

Includes auroral and polar regions, h90–1000 km

Auroral and polar ionosphere, h100–500 km

Semi-empirical parameterized model, uses some GTIM equations to generate a global set of electron-density profiles and the PIM (Daniell et al., 1995) data base, then uses near-realtime data from sensors

Parameterized Realtime Ionospheric Specifications Model (PRISM) (Anderson et al.)

F region only

First principles, threedimensional Euler–Lagrangian

Global Theoretical Ionospheric Model (GTIM) (Anderson et al.)

Contact the authors for details of availability

request for installation on DEC/VAX platforms

Available from P. G. Richards

Available in FORTRAN on supercomputer; the authors welcome collaboration

Available from the authors, supercomputer platform

Available from the authors

Type of model and inputs

First principles, threedimensional, time-dependent, driven by a time-dependent species of solar EUV and UV spectral irradiance and magnetic conjugate auroral particle precipitation and convection patterns

First principles, Ohm’s law and current-continuity equations; inputs from models or from data

Empirical model

Empirical data base, climatological model

Empirical data base from Spitzbergen, 7000 satellite passes

Model

NCAR Thermosphere– Ionosphere–Mesosphere– Electrodynamics General Circulation Model (TIGCM) (Roble)

USU Electrodynamic Ionospheric Model (Zhu)

International Reference Ionospheric (IRI) (Bilitza, 1997)

WIDEBAND scintillation model (see http://www.com/nwra/ scintpred)

Polar Phase Scintillation

Table 9.1. (cont.)

North Polar region north of Scandinavia

Global model (WBMOD) VHF through L band

“High-latitude enhancements” (Rawer and Bilitza, 1995)

Specifically for high latitudes 50°–90° MLAT. Spatial resolution in tens of kilometers both in MLT and in MLAT

Global model, h30–500 km

High-latitude features

Kersley et al. (1995)

Also see Fremouw and Secan (1984) and Secan et al. (1997)

Contact authors for details of collaboration

Contact the author for details of availability; resides on NCAR CRAY YMP-8-64

References and comments

9.2 Ray-tracing, modeling, and prediction

the importance of the variation of solar activity on F-layer irregularities at the equatorward edge of the auroral region over a solar cycle. Their data indicate that high-latitude F-layer irregularities occur less often in the auroral region and are of lower intensity during periods of low solar flux. Data from Goose Bay, Labrador, observing from 67° to 70° CGL, indicate that the occurrence of scintillation at 250 MHz during a year of low solar flux (1985) is enormously reduced compared with the occurrence for the same magnetic conditions during a year of high solar flux (1980). Since the “bottom line” in the usefulness of communication and navigation system is the signal-to-noise ratio (SNR – see Section 3.2), it is obvious that radionoise models are also required. Since 1988, CCIR Report 322-3 has been the accepted global model of radio noise, but discrepancies have been noted and it has been found that this model, specifically, does not yield very accurate data at high latitudes, according to Sailors (1995). Warber and Field (1995) have also provided a long-wave transverse-electrictransverse-magnetic noise-prediction model for the range of 10 Hz to 60 kHz. This model predicts the global distribution of r.m.s. noise, standard deviation, voltage deviation, and amplitude probability distribution for both polarizations. Another investigation characterizing radio noise at geographic latitudes of ⬃30°–50° in the North Pacific using the US Navy HF Relocatable Over The Horizon Radar (ROTHR) at Amchitka, Alaska was reported by McNeal (1995). The area probed was the subauroral ionosphere near sunspot maximum using frequencies from 5 to 28 MHz, and no attempt to characterize the degree of geomagnetic disturbance was made. McNeal found that, on the basis of a relatively small sample, the ROTHR noise data were within 2.5 dB of the predictions of CCIR Report 322 for that latitudinal region. Also, vertical ionograms, oblique backscatter soundings, radar ground-backscatter amplitudes, and noise levels were compared with model predictions and the authors state that “the differences between the model and median soundings are small enough to have negligible effect on predictions.” 9.2.3

Validation of ionospheric models

In the last decade more effort has been devoted to attempts to verify and validate the various ionospheric models, especially through PRIMO (Problems Related to Ionospheric Modeling and Observations) workshops, the CEDAR (Coupling, Energetics and Dynamics of Atmospheric Regions) program, and other efforts, as reported by Schunk (1996), Anderson et al. (1998), and Szuszczewicz et al. (1998) Anderson et al. (1998) compared five of the physical models listed in Table 9.1 (TIGCM, TDIM, FLIP, GTIM, and CTIM) with each other and with data obtained at the Millstone Hill ISR for four geophysical cases, thus this was basically a mid-latitude evaluation. According to this study, the five models displayed diurnal variations that, in general, agreed with measurements, but each one of the

545

High-latitude propagation: 2

546

five models exhibited “a clear deficiency” in at least one of the four geophysical cases that was not common to the other models. In a related study, Szuszczewicz et al. (1998) compared f0F2 and hmF2 outputs of four models (IRI, TIEGCM, FLIP, and CTIP) at mid-latitudes during magnetically quiet conditions (0Kp 3) and found accuracies “generally better than 5%.” Since both of these studies were restricted to mid-latitudes, we cannot draw conclusions about comparisons between modeled and real data at high latitudes. Doherty et al. (1999), Decker et al. (1999), Bishop et al. (1999), Bilitza (1999), Bust and Coco (1999), and Ganguly et al. (1999) have investigated the validity of the PIM, the PRISM, the IRI and the GPS/NNSS ionospheric models and data bases, and found that moderate success is achieved for predictions at middle latitudes. There have been few validations of these models at high latitudes. 9.2.4

The performance of ELF–HF predictions at high latitudes

It should be re-emphasized that all of the radio-propagation-prediction programs are principally climatological models that produce median-value predictions; therefore, they cannot be expected to produce weather-type results. (In spite of this caveat, however, some HF communicators persist in attempting to use these models for “weather-type” propagation forecasting.) The programs should, correctly used, produce the type of predictive data that will allow a mid-latituderadio-circuit planner to design radio-communication or navigation-link behavior as a function of time, season, sunspot cycle, and equipment parameters, as well as specify “worst-case scenarios.” At high latitudes, we must conclude that the extant predictive systems are inadequate – even for MUF and LUF predictions. Proponents of these programs are understandably reluctant to adequately test their programs at high latitudes, and, for whatever reasons, funding agencies also seem to be rather hesitant to adequately validate the programs. Part of this may be due to the de-emphasis of the use of HF communications in the USA because of the predominance of communication satellites, cables, and LOS UHF links, and the advent of certain adaptive HF propagation techniques (see Section 9.6.4).

Validation of ELF/VLF prediction A representative software package for predicting and assessing long-wave propagation is described by Ferguson and Snyder (1989). They describe a collection of programs developed by the US Navy Labs in San Diego, California for the Earth–ionosphere waveguide mode for VLF through LF (10 kHz through 100 kHz) that predict signal strength and SNR on individual propagation paths over wide geographic areas. The model includes some high-latitude phenomena and is available in VAX/VMS language (Ferguson and Snyder, 1986). Ferguson (1995) also described a validation campaign for the Long Wave Propagation

9.2 Ray-tracing, modeling, and prediction

Figure 9.4. The daytime average absolute difference between LWPC and measurements parametric in frequency and distance interval (from Ferguson, 1995).

Capability (LWPC) prediction program, using measured in-flight signal levels from various transmitters from 10 to 60 kHz from 0° to 80° CGL. The average absolute differences between LWPC and measurements parametric in frequency and distance interval are shown in Figures 9.4 (daytime) and 9.5 (night-time).

Validation of HF prediction Compared with the ELF/MF portion of the spectrum, the HF (⬃2–30 MHz) band has a plethora of prediction programs available, as described by Goodman (1992, Ch. 5) and by Sailors and Rose (1993) – the latter report also addresses the prediction of skywave signal strengths. Thirteen of the extant programs are listed in Section 3.3.5 of Goodman’s book, but only two of these programs (AMBCOM and ICEPAC) include some high-latitude ionospheric effects. The widely used mid-latitude HF propagation program IONCAP was modified to include AE ionization plus polar and AA effects by Hunsucker (1971) to make predictions for US Coast Guard communications to aid search-and-rescue missions in the north

547

548

High-latitude propagation: 2

Figure 9.5. The night-time difference in average absolute absorption as in Figure 9.4 (from Ferguson, 1995).

Pacific, but there is no information on the reliability of these predictions. Davé (1990) illustrated the importance of mode-plot diagrams derived from the raytracing component of the AMBCOM program for determining the optimum paths and ground ranges at high latitudes. There is very little documentation in the literature concerning validation of predictions of these programs with good-quality high-latitude ionospheric data, although several such comparisons are currently in progress. One candidate for providing high-latitude HF propagation predictions is the PRISM/VOACAP program listed in Table 9.1, and other candidates include driving the ICEPAC or AMBCOM prediction programs with PRISM or with the PIM data base. The best HF high-latitude propagation-prediction program should ideally include a realistic quantitative first-principles model, a data base that accurately portrays the polar and auroral D-, E-, and F-region parameters, a realistic radio-noise data base, accurate equipmental and antenna parameters, an analytic ray-tracing program, and near-realtime “space weather” data inputs. These requirements are

9.2 Ray-tracing, modeling, and prediction

549

Figure 9.6. Locations of transmitters and receivers for short and long paths (after Thrane et al., 1994).

especially true for programs that purport to produce field strength predictions. The report by Sailors and Rose (1993) compares seven HF-propagation prediction programs (three empirical programs – PROPHET, FTZ, and FTZ4, and four analytic programs – HFTDA, IONCAP, ASAPS, and AMBCOM) in terms of their abilities to predict signal strength. Of these programs, only AMBCOM had an analytic ray-tracing routine and included high-latitude data. One of the attempts at validation of a high-latitude HF propagation program using real data was presented by Thrane et al. (1994) using the ICEPAC program for predicting performance on two propagation paths within Norway. The geometries of the two paths in relation to the auroral zone are shown in Figure 9.6. The results of this investigation indicated “that ICEPAC represents an improvement over IONCAP as far as the structure of the E and F regions is concerned . . . but that transmission losses are not properly included.” They also concluded that (1) the prediction code reproduced the main features of the observed diurnal variation of channel reliability, but significantly overestimated both the reliabilities and MUFs (the discrepancies are particularly pronounced for magnetically disturbed conditions and for the short path within the auroral oval) and (2)

550

High-latitude propagation: 2

Figure 9.7. The HF propagation path, between Clyde River, Canada and Leicester, UK (after Gikas, 1990).

the ICED electron density profiles used in ICEPAC depend upon the location of the control points of the paths relative to the auroral oval, and therefore change with the level of geomagnetic disturbance. Another attempt to validate certain HF-propagation-prediction programs was published in four reports based on master’s theses at the US Naval Postgraduate School at Monterey, California (Gikas, 1990; Tsolekas, 1990; Wilson, 1991; Burtch, 1991) analyzing PROPHET 4.0, IONCAP-PC 2.5, AMBCOM, and ICEPAC, respectively. HF SNR data obtained during the trans-polar 1988–1989 NONCENTRIC HF propagation experiment (Rogers et al. 1997) was used to test the prediction models. The long path between Clyde River, Canada and Leicester, UK is shown with an auroral oval for Kp of 5 in Figure 9.7. Some of the SNR results from these four studies are summarized in Table 9.2. It is quite interesting that there were significant differences between the average error and/or standard deviation of error as a function of frequency for different prediction programs and data obtained during the 1989 winter campaign, as shown in Figures 9.8–9.10.

9.2 Ray-tracing, modeling, and prediction

551

Figure 9.8. The average error for Site D, Winter 1989, of ICEPAC predictions from measured values (after Burtch, 1991).

Table 9.2. Selected SNR errors for four HF propagation programs HF-propagationprediction program

Propagation path

Results

Advanced PROPHET Clyde River–Leicester 70% of 4.0 predictions were between 20 dB and 20 dB error

Reference Gikas (1990)

IONCAP-PC 2.5

Clyde River–Leicester Predicts error with Tsolekas (1990) an error less than 10 dB, with significant errors during disturbed periods

AMBCOM

Clyde River to three polar receiver sites

Average error was Wilson (1991) typically distributed between 20 and 20 dB,absolute value of average error 7–11 dB

ICEPAC

Clyde River to four polar receiver sites

Absolute errors from 0.3 to 26.4 dB

Burtch 1991

552

High-latitude propagation: 2

Figure 9.9. The average for Site D, winter 1989 of AMBCOM predictions from measured values: total average error 13.5 dB, standard deviation 28.9 dB and total number of samples 2919 (after Wilson, 1991).

Figure 9.10. The standard deviation of IONCAP-PC 2.5 prediction errors versus frequency for the Winter 1989 campaign (after Tsolekas, 1990).

9.2 Ray-tracing, modeling, and prediction

9.2.5

Recent validation of selected ionospheric prediction models using HF propagation data

A study by Hunsucker (1999) illustrates the use of practical data on HF signal reception, along with space-weather parameters (such as the solar 10.7-cm radio flux, Kp, etc) to validate several HF propagation prediction (hereafter referred to as HFP) programs and one ionospheric model. All of the HFP programs were designed with the intent of providing information for planning HF circuits, not for short-term forecasting. The propagation data consisted of HF signal amplitudes obtained on auroral, subauroral, and mid-latitude propagation paths during July through December 1993 on 5.6, 11.0, and 16.8 MHz at 6-min intervals. These data were obtained during the PENEX (Polar, Equatorial and Near-Equatorial Experiment) sponsored by the US Navy. Some results on the equatorial parts of the experiment have been published by Smith (1998) and the “polar” data are used in the present analysis. Only limited space-weather data were available in 1993 and not all of the available HFP software is structured to utilize space weather data as input, but the concept is valid for future validation efforts.

A description of the PENEX The PENEX program utilized a HF transmitter located at Cape Prince of Wales, Alaska and receivers at Fairbanks, Alaska, Seattle, Washington and Rock Springs, Pennsylvania, as shown in the map of Figure 9.11 The transmitting antennas for each frequency were halfwave dipoles, one half wavelength above ground, and the receiving antennas were HF log-periodic antennas (LPAs) at a height of ⬃20 m. The elevation radiation patterns were modeled using the NEC analysis program (Burke, 1981) and no sharp nulls were found for the dominant modes. The funding for this project did not permit ray-tracing analysis of specific propagation modes. The basic modulation scheme selected was direct-sequence–spread-spectrum (DS–SS), in which a digital code sequence modulates the carrier at a much higher rate than the information and produces a (sin x/x)2 power envelope. One of the “Gold Code” pseudo-random noise sequences was selected, producing a signal bandwidth of 40 kHz (Rose, 1993; Omura et al. 1985). This DS–SS technique produced good HF signal levels, high rejection of interference, multipath rejection, and high-resolution range measurement over the planned paths using only 100 W of transmitter power. (On the basis of preliminary measurements, this DS–SS system produced an estimated 40 dB gain over a “conventional” system such as single-sideband transmission (Rose, 1993)). Transmissions were also made using continuous-wave (CW) Morse code for station identification and frequency-shiftkeying (FSK) for “housekeeping” data. These modulation schemes were used sequentially on 5.604, 11.004, and 16.804 MHz from July through December 1993. Only the 4 KB spread-spectrum sequence was analyzed, since it was observed to produce a higher SNR than did the other sequences, with reasonable

553

554

High-latitude propagation: 2

Figure 9.11. A map of the PENEX.

processing time. These three frequencies represent a reasonable sampling of typical HF frequencies used during this part of the solar cycle. The Fairbanks and Seattle receiving stations employed receivers that recorded the DS–SS transmissions and the Rock Springs receiver received only the CW and FSK transmissions – resulting in quite low signal levels being received at Rock Springs. Therefore, the principal analysis effort of PENEX concentrated on the DS–SS data received in Fairbanks and Seattle. Approximately 900 h of signal-amplitude data for the three frequencies – representing diurnal, seasonal, and geomagnetic activity – were used as the present data base. Several days were also available from the Rock Springs station on a “hear–no– hear” basis. Table 9.3 describes the salient features of the four HFP programs used in this investigation. Extensive discussions of the properties of HFP software are given by Goodman (1992, Ch. 5), Davies (1990, Ch. 12), and Sailors and Rose (1993). An example of the propagation predictions generated by one of the HFP programs that we used (VOACAP – see Lane, 1993) is shown in Figure 9.12 for the auroral oval path (Wales to Fairbanks) for a quiet day in November 1993. The horizontal lines below the prediction plot indicate the intervals when VOACAP predicted that propagation should occur on that frequency.

Virtual geometry, based on the analytical IONCAP program, with added high-latitude, CCIR and URSI data bases

ICEPAC (Ionospheric Conductivity and Electrodynamics, Prediction, etc.)

Table 9.3. (cont.)

Virtual geometry, Australian ionosonde data plus CCIR models of ionospheric parameters and noise (an empirical model)

ASAPS-4 (Advanced Stand-Alone Prediction System)

Characteristics of the model

Transmitter and receiver coordinates, date, SSN or solar flux, system parameters, frequencies, Qeff

Transmitter and receiver coordinates, date, SSN, solar flux or T-index, system parameters, usable frequencies

Inputs

ALFabsorptionlimited frequency, BUFbest usable frequency, EMUFthe E-layer maximum frequency, MUF maximum usable frequency via the F region, OWFoptimum working frequency; Caruana (1993) Q effective Q index; Stewart (1990, private communication)

MUF, LUF, median field strength (dB)

Remarks and references

ALF, BUF, EMUF, MUF, OWF, takeoff angle

Outputs

Table 9.3. Characteristics of HF-propagation-prediction programs used in the analysis

Virtual geometry, based on the IONCAP program (in a user-friendly shell), CCIR and URSI data models

sounder data, has an auroral oval and a raytracing module

Virtual geometry, data base of oblique HF

Transmitter and receiver coordinates, date, SSN

solar flux, Kp (an index of solar-proton and x-ray flux)

Transmitter and receiver coordinates, date, SSN or

Inputs

MUF, median Field strength

LUF, MUF, and OWF, median field strength

Outputs

Lane (1996)

(1981)

OWFoptimum working frequency;

Remarks and references

Notes: CCIR, the International Radio Consultative Committee of the International Telecommunication Union; SSN, sunspot number; URSI, International Union of Radio Science.

VOACAP (Voice-ofAmerica version of IONCAP)

Rose

PROPHET 4.3.2

Characteristics of the model

Figure 9.12. An example of VOACAP prediction and PENEX measurements.

558

High-latitude propagation: 2

Figure 9.13. The PENEX signal amplitude for 27 September 1993, f 11.0 MHz.

Specific results of the PENEX Over 900 h of PENEX signal reception at Fairbanks, Seattle, and Rock Springs from the period July through December 1993 were analyzed and discussed by Hunsucker (1999), and examples of the signal amplitudes received in Fairbanks are given in Figures 9.13 and 9.14. The abscissae in Figures 9.13 and 9.14 are equivalent to the received signal amplitude in the DS–SS system. The horizontal broken line near the bottom of the plot represents ⬃10% of the peak value and is approximately equivalent to the required signal level for HF single-sideband communications or short-wave (SW) broadcasting, on the basis of limited comparisons of propagated signals. We define the “correct prediction percentage” for each HFP for a 24-h period as “the number of hours for which the program predicted that propagation will occur on that frequency and path, compared with the number of hours that the propagation actually occurred, plus the number of hours that the program predicted no propagation on the path at that frequency. This is compared with the number of hours that no propagation occurred, all expressed as percentages. We believe that the “correct-propagation percentages” thus defined are at least semiquantitative and should be valid and understandable both to the ionosphericresearch community and to the HF-propagation/communications community. Tables 9.4 and 9.5 give the results of the comparison between predicted and

9.2 Ray-tracing, modeling, and prediction

Figure 9.14. The PENEX signal amplitude for 24 August 1993, f 11.0 MHz.

observed HF reception on the auroral-oval circuit and the subauroral (usually mid-latitude) paths, respectively. The unique feature of this HF dataset is that data are plotted approximately every 6 min, whereas most comparisons between predictions and HF data use hourly average HF values. Thus, the present data present more HF “finestructure” behavior. It has been known for some time that some variations in HF signal are produced by the gravity-wave-induced traveling ionospheric disturbances (TIDs). As noted by Hunsucker (1982), medium-scale TIDs typically have periods from ⬃12–50 min (also see Section 1.6). If hourly values of HF data had been used in comparison with the monthly hourly medians used by the predictions, the “correct-prediction percentages” would probably have been higher.

The Wales–Rock Springs, Pennsylvania path McDowell et al. (1993) described the equipment and results of a “hear–no-hear” program monitoring the PENEX transmissions from Wales, Alaska at the Rock Springs, Pennsylvania HF receiving site (latitude40.8° N, longitude 77.9° W). Because no complete PENEX receiver was available for this site, only the FSK and CW signals were recorded.

559

High-latitude propagation: 2

560

This was an interesting multihop 5925-km path, which, under quiet geomagnetic conditions (Kp 3), can be considered to be a mid-latitude path. Measurements were made only during August and September 1993, and there was usable data for 70% of the time. As geomagnetic activity increases, however, this increasingly becomes an auroral path, with the entire path lying within the auroral oval when Kp 8 (magnetic-storm conditions) and the first 70% of the path lying within the oval for Kp 5. Specific examples of PENEX signal reception are given in Figure 9.15. The statistical auroral oval (Feldstein and Galperin, 1985) was utilized for this comparison. The 5.6-MHz signal, K indices, and predictions of the ASAPS-4 program (see Table 9.4) during the geomagnetic storm of 13 September 1993 are shown as functions of UT in Figure 9.15(a). On this day, the entire path lay in the auroral oval from ⬃0300 to 1030 UT, the first 70% of the path was in the oval from ⬃1030 Table 9.4. Percentages of correct predictions on Wales-to-Fairbanks HF circuit Equinox (September) Quiet

Disturbed

Winter (November) Quiet

Frequency (MHz) A I P V A I P V A I P V

Disturbed A I P V Remarks

5.6

36 58 18 47 30 42 33 33 47 53 29 76 51 40 33 40 60 42% for 5.6 MHz

11.0

15 12 40 12 15 18 10 18 70 90 77 58

35 40 15 30 35% for 11.0 MHz

16.8

77 77 67 77 33 33 21 33 48 48 98 48

Program average

43 49 42 45 45 26 31 21 55 64 68 61

20 30 20 30 47% for 16.8 MHz 32 34 25 40

Notes: A ASAPS-4; I  ICEPAC; P  PROPHET; V VOACAP. (1) There were no “summer” data for this path. (2) The average corrrect predictions for 5.6, 11.0, and 16.8 MHz for all seasons and levels of disturbance are 42%, 35%, and 47%, respectively, for this auroral-oval path. (3) On “quiet” days the four programs predicted approximately equal percentages for equinox and winter. (4) On “disturbed” days the percentage of “corrrect” predictions decreased by a factor of two from the quiet-day predictions. (5) There is no significant advantage of one program over another on this path, except that, on quiet fall days, all programs gave high percentages on 16.8 MHz. (6) For all programs, seasons, and frequencies, the aggregate correctly predicted percentage is 44%.

29 45 35 36

5.6 11.0 16.8 Program average

57 70 50 59

I

78 55 35 64

P

78 90 25 64

V

39 43 53 45

A

23 55 40 39

I 15 50 53 39

P

Disturbed

84 17 26 42

V 44 13 50 36

A 28 53 40 40

I 20 73 80 41

P

Quiet

90 73 40 67

V

Fall

50 30 49 43

A 55 49 37 47

I 27 63 87 59

P

Disturbed

50 31 49 43

V 50 30 40 40

A 33 40 40 37

I 25 40 30 32

P

Quiet

42 30 40 37

V

40 20 40 37

A

Winter

93 40 67 66

I

6 40 33 26

P

Disturbed

50 40 67 52

V

Notes: A ASAPS-4; I  ICEPAC; P  PROPHET V.4; V VOACAP. (1) The average percentages of correct predictions for 5.6, 11.0, and 16.8 MHz for all seasons and levels of disturbance are 45%, 45%, and 46%, respectively, for this mid-latitude path. (2) On quiet days, it appears that the four programs averaged somewhat higher in the summer than they did in fall and winter (56%, 46%, and 45%, respectively). (3) On disturbed days, there is no significant difference among the accuracies of prediction for summer, fall, and winter (41%, 48%, and 45%, respectively). (4) There is no significant difference between accuracies of prediction for quiet days (46%) and disturbed days (45%). (5) For all programs, seasons, and frequencies, the aggregate correctly predicted percentage is 45%.

A

Frequency (MHz)

Quiet

Summer

Table 9.5. Percentages of correct predictions on Wales-to-Seattle HF Circuit

(a)

Figure 9.15. PENEX signal amplitudes on (a) 5.604, (b) 11.000, and (c) 16.804 MHz on a disturbed day, 13 September 1993; SSN11 and 10.7-cm flux 80.

(b)

Figure 9.15. (cont.)

(c)

9.2 Ray-tracing, modeling, and prediction

to 1700 UT and the path was tangential to the equatorward edge of the oval at ⬃2300 UT, so this nominally mid-latitude path became an auroral path this day. ASAPS predicted only ⬃33% propagation for the day, with moderately accurate signal-strength predictions from ⬃0400 to 1100 UT. On 11.0 MHz (Figure 9.15(b)) ASAPS-4 did a good qualitative propagation prediction, but was almost anticorrelated in the signal-strength predictions. Figure 9.15(c) illustrates that the 16.8-MHz signal was barely detectable and the qualitative and quantitative predictions produced by ASAPS-4 were poor (50%).

Space-weather data applied to the PENEX Some of the limited (1993) available space-weather data were utilized in this investigation of a small sample of the PENEX data and it was found that there appeared to be no connection between the measured signal strengths and the AE, the cross-polar-cap potential, and the local (Fairbanks) K index on the Wales-toSeattle path. (One might expect the first hop of this path to be influenced by the auroral ionosphere). However, the peak values of Ap (the linear planetary magnetic index) coincided with the signal-amplitude peaks on all three frequencies. The IMF Bz southward turnings do not seem to be closely related to variations in the signal strength, as might be expected, since there is a time delay between the southward turning and the ionospheric response. There seems to be some relation between the GOES X-ray flux and the first peaks of signal strength on this subauroral circuit, which is probably related to increases in F-region ionization. There are not many investigations reported in the refereed literature on the relationship of specific geophysical indices or other space-weather data to HF signal amplitudes (see discussions in Davies, 1965; Mather, et al., 1972; and Milan et al., 1998).

Auroral ovals and DMSP images applied to the PENEX PENEX data were analyzed during the “National Space Weather Event” of 3–11 November 1993 covering a large geomagnetic storm. During the disturbed day (4 November) predicted auroral ovals obtained from the PROPHET (see Rose, 1982) software and based upon the Feldstein and Galperin (1985) ovals were compared with DMSP optical-line-scanner auroral images (Figure 9.16). Figure 9.16 also shows the portions of the Wales–Fairbanks propagation path (straight line) which lay within the oval. It is seen that the auroral forms from DMSP lie well within the predicted ovals and in the propagation path. The discrete auroral form shown in Figure 9.16 lay over the mid-point of the Wales–Fairbanks path, which has been shown to be very closely related to high electron density in the E region (Hunsucker and Owren, 1962; Hunsucker, 1965; Hunsucker et al., 1996). An ionogram from the Fairbanks ionosonde recorded near the time when the auroral forms were in the propagation path which exhibits values of f0Es and fEs

565

566

High-latitude propagation: 2

Figure 9.16. “Space-weather data (a DMSP image in relation to the Wales–Fairbanks propagation path).

Figure 9.17. The HF propagation “metric” on 4 November 1993. The amplitude of the signal is 16.8 MHz.

9.2 Ray-tracing, modeling, and prediction

Figure 9.18. “Space-weather data” (Fairbanks ionosonde).

of 7.4 and 8.2 MHz, respectively, is shown in Figure 9.18, and Figure 9.19 displays the 16.8-MHz signal amplitude at Fairbanks on this disturbed day. As may be seen in Figure 9.17, the PENEX 16.8-MHz amplitude peak occurred when an auroral form lay over the mid-point and AE ionization was very intense, which is consistent with many previous examples given by Hunsucker. This is one illustration of the relation between the ground-based (ionosonde) and satellite (DMSP) spaceseather data and the HF propagation metric. We have given an example of how HF-propagation data can provide direct validation of space-weather effects upon actual operating systems. Analysis of over 900 h of data on three frequencies on auroral, subauroral, and mid-latitude paths was utilized for validation of four HF-propagation-prediction programs. The aggregate correct prediction from these programs was only ⬃45% for a wide variation of geomagnetic activity in 1993. This should serve as another caution to HF communicators not to use HFP programs for forecasting propagation. Limited space-weather data were available as inputs to prediction programs and models and were qualitatively useful in interpreting anomalies in HF propagation, but it was not possible to determine quantitative relationships between specific indices and variations in signal for this small sample. Another feature of

567

High-latitude propagation: 2

568

this investigation is that amplitudes of HF signals, which respond to observed variations in signal on high-latitude circuits, were recorded every 6 min. An auroral oval from the PROPHET prediction program was in very good agreement with a DMSP visual auroral image, with an ionogram obtained at Fairbanks, Alaska, and with the received amplitude of the 16.8-MHz signal on the Wales–Fairbanks path for the disturbed day of 4 November 1993. It is hoped to continue this type of investigation utilizing the now plentiful space-weather data with new HF propagation data and with several of the available large ionosphere models. The PENEX research was supported by the US Naval Security Group Command, the Office of Naval Research, and the Naval Postgraduate School-Monterey, and one must acknowledge the data-analysis contributions made at the Applied Research Laboratories of Pennsylvania State University and the antenna pattern analyses performed by J. K. Breakall.

9.3

Predictions of VHF/UHF propagation

As illustrated in Section 8.5, there are times (during sunspot-maximum periods) when VHF signal propagation (up to 46 MHz) has been observed on 2000–5000-km polar paths, but, as yet, there is no reliable method of predicting these openings. Qualitatively, one can, however, expect these modes to occur during geomagnetically disturbed periods in years with high numbers of sunspots when intense AE is present on part of the path.

9.4

Recent efforts at validation of ionospheric models

The Space Environment Corporation of Logan, Utah has under development an “Assimilating Ionosphere Model” (AIM) (Schunk and Sojka, 1999, private communication), and this program has been used to generate a database for 4 and 12 November 1993 for comparison with the PENEX data. Days 308 and 316 of 1993 (4 and 12 November) have been analyzed by displaying NmF2 over the region of interest at hourly intervals of UT. The approximate great-circle paths from Wales to Seattle and Fairbanks are shown as two sloping lines in the bottom-left-hand panel of Figure 9.19, which shows the plot for day 316 – corresponding to the geomagnetically quiet day. Each panel represents the F-region peak density (NmF2) color coded from 1011 to 1012 electrons m3 . The coordinates are mixed – the longitude is in geographic coordinates, while the latitude axis is in geomagnetic coordinates. The important dusk transition occurs from 0200 to 0600 UT and involves a relatively gradual decrease with UT. In contrast, the dawn–sunrise region has sharp density gradients in UT (see from 1700 to 1900 UT). The night-time region has densities dropping below 1011 m3(105 cm3) with a relatively narrow, deep-blue trough

9.4 Validation of models

Figure 9.19. The USU “Assimilating Ionospheric Model” images taken once per hour for a geomagnetically quiet day (12 November 1993). The approximate great-circle paths from Wales to Seattle and to Fairbanks are shown as two sloping lines in the bottom left-hand panel (after Schunk and Sojka, 1999).

569

High-latitude propagation: 2

570

feature. A caveat of importance at this point is that AIM is currently running in a climatology mode driven by only the Kp and 10.7-cm indices, since no ionospheric data are available for assimilation. Furthermore, the model is being run in a default mode with zero topside flux. Therefore, night-time maintenance and hence densities are probably too low. AIM is being modified to correct this shortcoming. In analyzing the PENEX data for comparison with AIM, it should be mentioned that AIM models the O density and molecular-ion density from ⬃100 to 1000 km and includes the Hardy et al. (1987) empirical electron-precipitation model. D-region absorption effects (which can be profound on disturbed days) are, however, not included. Therefore, we shall examine the behavior of the Wales–Seattle circuit (a mid-latitude path) in the framework of the AIM model. The peak amplitudes of the 5.6-MHz signal occurred at ⬃1300 UT (0300 AST) and ⬃1530 UT (0530 AST), when the path appears to be roughly aligned with and a few degrees equatorward of the trough feature. The propagation from ⬃1600 to 2400 UT on 11.0 and 16.8 MHz is, in general, related to increasing electron densities at F2max height. Thus the Wales–Seattle (“mid-latitude”) propagation path on this quiet day seems to be in at least qualitative agreement with AIM. In contrast with Figure 9.19, Figure 9.20 represents a disturbed day (day 308) and the most noteworthy features are (1)

the very clear night-sector auroral region,

(2)

the very marked “deep” night-sector trough,

(3)

the extremely sharp equatorward trough boundary in the afternoon sector (0200–0600 UT),

(4)

higher noon, sunlit densities, and

(5)

Storm-enhanced densities at 0200 and 0300 UT. These are the high densities prior to the precipitous drop into the trough.

On the disturbed day, propagation occurred on all three frequencies from ⬃0000 to 1000 UT, with no propagation from ⬃1100 to 2400 UT. The AIM plots from ⬃0000 to 0600 UT are consistent with the observed propagation, but from ⬃0700 to 1000 UT the path lies in a deep Ne trough. It is reasonable to assume that AE ionization could augment the propagation from ⬃1000 to 1100 UT (0000–0100 AST), since this path lies well within the auroral oval at this time. After ⬃1100 UT, auroral absorption could contribute to the loss of the PENEX signal. So, to a qualitative first approximation, the observed behavior of the PENEX HF propagation on these quiet and disturbed days is in agreement with the AIM outputs, as shown in Figures 9.19 and 9.20 and as outlined in the review of high-latitude radio propagation by Hunsucker (1992). It is to be hoped that future efforts will include ray-tracing through the AIM outputs and then validation by using HF-propagation-mode structure.

9.4 Validation of models

Figure 9.20. The same as in Figure 9.19, but for a geomagnetically disturbed day (after Schunk and Sojka, 1999).

571

High-latitude propagation: 2

572

9.5

Mitigation of disturbance of HF propagation

9.5.1

Early attempts

It has been known since early-1950s studies that the reliability and predictability of HF high-latitude propagation were lamentable, and Gerson (1962a, 1962b; 1964) presented an interesting qualitative evaluation of various communications modes as shown in Table 9.6. It should be emphasized that the evaluations and cost estimates in Table 9.6 were Gerson’s in the early 1960s and are subject to other reasonable estimates, and that the communications-satellite mode was not available for comparison at that time. Nevertheless, it is interesting to note that the VHF scatter mode has since been abandoned because of high costs and that VLF/LF was not really a communications mode. The row of totals indicated that the submarine cable and UHF tropospheric propagation modes rated the best in this evaluation, but, because of the high cost and difficulty in establishing a long-range UHF relay system, the latter was not considered. Some serious consideration was given to laying submarine cables, as evidenced by the routes indicated in Figure 9.21. Communications and navigational satellites have greatly reduced the use of VLF/LF and HF systems at high latitudes even with the high cost of such systems and their vulnerability to some high-latitude ionospheric effects. The use of forward-sounding circuits and link switching to ameliorate problems with high-latitude HF propagation was discussed in the 1960s (e.g. Fenwick and Villard, 1963; and Hunsucker and Bates, 1969) and later Fenwick and Table 9.6. Gerson’s (1964) comparison of various communications modes on highlatitude paths on a scale of 1–9 (1excellent, cheapest cost, most reliable, least problems, etc.)

Parameter Reliability Bandwidth Potential for Interference “Jammable”? Problems at solar maximum? Initial cost Operating cost Total

Submarine cable

VHF VLF/LF

HF

Scatter

Meteor

UHF Tropospheric

2 2 1

1 6 2

7 4 4

2 4 3

5 4 2

1 1 1

1 3

6 2

8 6

3 2

2 2

1 1

6 2 17

4 1 22

3 3 35

5 3 24

5 3 23

6 6 17

9.5 Mitigation of disturbance

Figure 9.21. A polar map indicating locations of proposed submarine cable routs from Scotland and Norway to Alert and thence to Moosonee and Barrow (from Gerson, 1964).

Woodhouse (1979) described the extensive US Navy HF frequency-management system using a world-wide network of chirp sounders. 9.5.2

Mitigation using solar–terrestrial data

All of the HF-propagation-prediction programs listed in Chapter 3 and in Table 9.3 provide for use of the sunspot number (or solar flux) and a geomagnetic-activity index (usually Kp) in addition to time of day, month, and year as inputs. The

573

High-latitude propagation: 2

574

programs which predict field strength also require the antenna gain, transmitter power, sensitivity of the receiver, noise levels of the receiver area, etc. as input information and, as shown in Table 9.4, these programs do not yield sufficiently accurate predictions – especially at high latitudes. See Sailors and Rose (1993) for a discussion of how seven of these programs calculate the field strength. At present, only the PRISM/VOACAP and the PROPMAN® (Hu et al., 1998) prediction programs utilize additional solar–terrestrial inputs, and possible improvements due to these additional parameters and the use of improved algorithms have not yet been evaluated (to the best of the authors’ knowledge). Use of the PRISM model with either the ICECAP or AMBCOM prediction programs and realtime solar–terrestrial inputs would probably be more effective at high latitudes than PRISM/VOACAP and such a combination should, of course, be validated. There is some question, however, regarding whether even these models can adequately describe in sufficient detail the near-realtime high-latitude ionosphere to permit ray-tracing or virtualgeometry calculations sufficient for calculations of HF propagation – especially field-strength predictions. Another approach to mitigation in predicting reliability of communication for HF through UHF propagation is utilized in the US Navy Radio Frequency Mission Planner (RFMP) described by Brant et al. (1994). RFMP is a suite of radio-propagation and terrain-modeling programs in an object-oriented interface on a work station that allows the user to translate communication-mission objectives into user-understandable results. Features include the integration of visualization tools, digital mapping, rule-based selection of propagation models, presentation of a model’s results as stochastic values, and estimation of the success of a mission presented in the geographic context. Real-time inputs to RFMP include measurements of tropospheric moisture density, GPS-derived TEC, electron-density profiles from ionospheric tomography and vertical-incidence sounders, plus satellite-measured solar–terrestrial parameters. RFMP is currently deployed on several platforms and is in the process of being validated. 9.5.3

Adaptive HF techniques

Some adaptive HF techniques are briefly described in Chapter 3 of this book and extended descriptions are given in Goodman (1992, Ch. 7) and in the book by Johnson et al. (1997). An essential part of any adaptive HF system is the automatic link-evaluation (ALE) scheme which is seen in the hierarchical diagram in Figure 9.22. The basic ALE operation of establishing a link between two stations proceeds as follows: (1) the calling station addresses and sends a call frame to the called station; (2) if the station “hears” the call, it sends a response frame addressed to the calling station; and (3) if the calling station receives the response, it now knows that a bilateral link has been established with the called station. The “polled” station does not yet know this, however, so the calling station sends an

9.5 Mitigation of disturbance

Figure 9.22. Hierarchical layers of a HF radio system (from Johnson et al., 1997).

575

High-latitude propagation: 2

576

acknowledgement frame addressed to the called station. At the conclusion of this three-way “handshake,” a link has been established, and the stations may commence transmission of voice or data traffic, or simply note that communication is possible, and then drop the link. The following protocol is taken from Johnson et al. (1997, pp. 9–10). The ALE standard also describes net and group calls and sounds as follows. (1)

A net call is addressed to a single address that implicitly names all members of a prearranged collection of stations (a net). All stations belonging to the net that hear the net call send their response frames in prearranged time slots. The calling station then completes the handshake by sending an acknowledgement frame as usual.

(2)

A group call works similarly, except that an arbitrary collection of stations is named in the call. Because no prearranged net address has been set up, each station must be individually named. Called stations respond in time slots, determining their slot positions by reversing the order in which stations were named in the call. The calling station sends an acknowledgement as usual.

(3)

A sound is a unidirectional broadcast of ALE signaling by a station to assist other stations in measuring channel quality. The broadcast is not addressed to any station or collection of stations, but merely carries the identification of the station sending the sound.

An example of the utilization of ALE techniques on a trans-auroral HF path was presented by Bliss et al. (1987), and a fairly detailed description of this experiment follows (in order to document this technique). The Trans-Auroral-HF Experiment (TAHFE) was conducted in 1986–1987 on a 4765-km path from Barrow, Alaska to Cedar Rapids, Iowa, as shown in Figure 9.23 in relation to a “disturbed” auroral oval (Q7). The TAHFE utilized a remote terminal at Barrow including a Collins HF-8070A transceiver, a 1-kW power amplifier, a selective calling and scanning unit (Collins HF-8096 SELSCAN® with test-signal-generation capability), FSK modems, a control microcomputer (PC), interface equipment, and a telephone modem. The similarly equipped receiving station was located at Cedar Rapids. The SELSCAN® unit was used as a control device to allow transmission of internally generated advanced-link-quality-analyzer (ALQA) tones and 300- and 75-bps binary FSK data (via auxiliary ports) and for automatic connectivity tests. The ALQA is a patented Rockwell developmental three-tone generation-andanalysis subsystem for measurement of HF channel parameters, utilizing narrowband signals within assigned radio channels.The SELSCAN® unit was controlled by a PC for sequencing purposes (time, frequency, and duration), which is programmed by remote control. At the receiver end of the circuit at Cedar Rapids, the HF-channel characteristics quantified are the several parameters measured directly by the ALQA (listed

9.5 Mitigation of disturbance

Figure 9.23. The Great-circle path (D 4765 km) from Barrow, Alaska to Cedar Rapids, Iowa on 14 October 1986 at 0500 UT in relation to an auroral oval for Q 7:1, BRW (71.30, 156.80); MP, mid-point (60.26, 109.78); and 2, CDR (42.02, 91.38) (after Bliss et al., 1987).

in Table 9.7). The objectives of the TAHFE are given in Table 9.8 and the experimental data base is listed in Table 9.9. The configuration of the TAHFE equipment is sketched in Figure 9.24 and the TAHFE test procedure is shown in Figure 9.25 and Table 9.10. We present selected data from one of the disturbed days during the 80-day duration of the TAHFE program to illustrate the types of data which were obtained. Figure 9.26 shows the frequencies from 6 to 21 MHz propagated during the day 12 November 1986 (SSN15.2; Kp 3) with the corresponding threefrequency sounding cycles shown in Figure 9.27. Some of the important signal parameters obtained during 12 November 1986 (SNR, Doppler spread, FSK biterror rate and multipath spread) are shown in Figures 9.28–9.30. It was planned

577

High-latitude propagation: 2

578

Table 9.7. Selected HF-channel parameters recorded during the TAHFE (after Bliss et al., 1987) Parameter

Units

Description

DOY TOD FRQ SNoR DS MP BER75 BER300

Days Hours Megahertz dB Hz Hertz Milliseconds Ratio Ratio

Day of year Time of day (universal time) in decimal hours Operating frequency Signal-to-noise power-density ratioa Doppler frequency spreada Multipath time-delay spreada Bit error rate for 75-bps data (10000 bits) Bit error rate for 300-bps data (10000 bits)

Notes: ALQA (advanced link-quality analyzer) measurement

Table 9.8. Objectives of the Trans-Auroral HF Experiment (TAHFE) (after Bliss et al., 1987) To collect data for a trans-auroral HF-channel data base To investigate (i) the correlation between TAHFE data and solar/geophysical data (ii) the correlation of oval and propagation predictions to TAHFE data To characterize (i) HF-channel characteristics by stepsounding with advanced linkquality analyzer (ALQA) (ii) FSK data bit-error rates with ALQA (iii) deduced ionospheric states with solar geophysical data

Table 9.9. The TAHFE data base (after Bliss et al., 1987) Identifier

Collection interval

TAHFE 1A TAHFE 1B TAHFE 2

16 September 1986 to 29 October 1986 5 November 1986 to 12 December 1986 11 March 1987 to 11 April 1987

Note: a Preliminary analysis on TAHFE 1B.

9.5 Mitigation of disturbance

Figure 9.24. The TAHFE equipment configuration.

Figure 9.25. The TAHFE test procedure.

579

High-latitude propagation: 2

580

22.00 20.00

FREQUENCY (MHz)

18.00 16.00 14.00 12.00 10.00 8.00 6.00 0.00

2.00

4.00

6.00

8.00 10.00 12.00 14.00 16.00 18.00 20.00 22.00 24.00 Time of Day (Hours UT)

Figure 9.26. The diurnal frequency variation for 12 November 1986 (from Bliss et al., 1986).

Table 9.10. TAHFE test procedures (after Bliss et al., 1987) Typical data file sent to remote (approximately 100 bytes) XXXX 015 050 150 N CDR 0000 LO 03260

START TIME ALQA MEASUREMENT TIME (SEC) 300 BAUD FSK (SEC) 75 BAUD FSK (SEC) CALL AT END OF CYCLE? CALL STATION ADDRESS OR SCAN (NCL) SCAN OR CALL TIME (SEC) POWER SETTING F1 FREQUENCY (KHz) : :

29700 Fn

to convolve a large solar–terrestrial data base into the results of the TAHFE-measured data, but no funding was available for this effort. 9.5.4

Realtime channel evaluation

The most promising technique for ameliorating deleterious effects on HF highlatitude communication circuits is realtime channel evaluation (RTCE), which is

9.5 Mitigation of disturbance

Figure 9.27. Three-frequency step-sounding cycles to be characterized (after Bliss et al., 1986).

Figure 9.28. Typical SNR, frequency, and multipath spread versus time of day for 12 November 1986 (from Bliss et al., 1986).

described in detail in Chapter 7 (122 pages) of Goodman (1992) and in CCIR Report 889-1 (1966). The technique basically consists of three stages for HF frequency management: long-term forecasting, short-term forecasting, and nowcasting. Specific classes of RTCE are oblique-incidence sounding (OIS), channel evaluation and calling (CHEC), vertical-incidence sounding (VIS), backscatter

581

582

High-latitude propagation: 2

Figure 9.29. Doppler spread (RMS) characterization for 12 November 1986 (from Bliss et al. 1987).

Figure 9.30. A frequency-shift-keying (FSK) error-rate comparison for 75 and 300 BPS for 12 November 1986 (from Bliss et al., 1986).

9.5 Mitigation of disturbance

sounding (BSS), frequency monitoring (FMON), pilot-tone sounding (PTS), and an error-counting system (ECS) – the acronyms are those used by Goodman (1992). The CCIR definition of RTCE is realtime channel evaluation is the term used to describe the processes of measuring appropriate parameters of a set of communications channels in real time and employing the data thus obtained to describe quantitatively the states of those channels and hence the capabilities for passing a given class, or classes, of communication traffic.

The CCIR classes of RTCE are listed in Table 9.11, along with some examples. A relatively long-term (December 1994–summer 1996) investigation of HF communication channels (some at high latitudes) that utilized a FMCW sounding network was reported by Goodman et al. (1997). Propagation parameters including ionospheric-mode information, MOFs, SNR, and availabilities of channels for digital data communications were derived and archived. Figure 9.31 is a map showing the HF propagation paths used during this experiment. One of the ultimate aims of the RTCE effort, according to Goodman et al. (1997), is to explore the potential for development of a practical HF data link (HFDL), even for high latitudes. The frequencies used were in the aeronautical mobile band (3.0, 3.5, 4.6, 6.6, 9.0, 10.1, 11.4, 13.3, 18.0, and 22.0 MHz) during a period when the number of sunspots was generally below 50. Data were compared with the minimum values of SNR required to pass traffic at 300–1800 bits s1. Figure 9.32 illustrates the percentage availability of signals received at Iceland and transmitted from four stations (Iqaluit and Jan Mayens being the most “auroral” of the paths). Figure 9.33 shows the percentage availability of HFDL service for each path and for frequency groups of 11, eight, six and four frequencies, respectively, illustrating the advantage of combining paths. Table 9.11. The CCIR classes of RTCE Class one:

Remote transmitted signal preprocessing a. Oblique-incidence sounder (OIS) 1. Pulse type 2. Chirp type b. Channel evaluation and calling (CHEC)

Class two:

Base transmitter signal preprocessing a. Vertical-incidence sounding (VIS) b. Backscatter sounding (BSS) c. Frequency monitoring (FMON)

Class three: Remote received signal processing a. Pilot-tone sounding (PTS) b. Error-counting system (ECS)

583

584

High-latitude propagation: 2

Figure 9.31. The geometry of HF propagation paths in the Northern Experiment (from Goodman et al., 1997).

Figure 9.32. The percentage availability of signals in the Aeronautical-Mobile bands received at Iceland and transmitted from four stations shwon in Figure 9.29 from 13 December 1994 to February 1995, SNR3dB (from Goodman et al., 1997).

Figure 9.33. Percentage availabilities of signals at Iceland for selected frequency groups and transmitter-station combinations. (For each group of four, the ordering from left to right is 11, eight, six, and four frequencies, respectively (from Goodman et al., 1997).

High-latitude propagation: 2

586

In conclusion, this study illustrated the advantages of the availability of a wide spectrum of HF frequencies, oblique frequency-sounding, spatial and frequency diversity, and dynamic frequency and link switching, even at high latitudes. Caveats include that the data were obtained during low sunspot activity and moderate geomagnetic activity, and that an essentially “mid-latitude” ionospheric climatological model (IONCAP/VOACAP) was used for prediction. 9.5.5

Recent advances in assessment of HF high-latitude propagation channel

Angling et al. (1998) presented results of measurements of Doppler and multipath spread on oblique high-latitude HF paths and their use in characterizing data- modem performance on the basis of four high-latitude HF communications paths. The data were analyzed in a manner pertinent to the design of robust HF data modems. The channel sounder utilized was the Doppler and Multipath Sounding Network (DAMSON) (Davies and Cannon, 1993). The DAMSON system operates from remote sites on preselected frequencies from 2 to 30 MHz. It is based on commercially available equipment (HF transceivers, PCs, etc.) and makes extensive use of DSP techniques and uses GPS for system timing – providing reception and transmission synchronized to within better than 10 s and to allow time-of-flight (TOF) measurements to be made. DAMSON uses several sounding waveforms, such as delay-Doppler, a Barker-13 sequence modulated at 2400 bps onto a biphase carrier, and passive noise measurements, as well as other modes. Figure 9.34 shows the geometry of the DAMSON paths studied in this investigation in relation to the auroral oval for low and high magnetic activity and the path-lengths are listed in Table 9.12. Rhombic and sloping-V antennas and power levels of about 250 W were utilized for this DAMSON investigation and the data were displayed in the format shown in Figure 9.35. Sample results from this DAMSON experiment are given in Figures 9.36–9.38 and Tables 9.13 and 9.14 The authors state that, in addition to measuring the multipath and Doppler spread and SNR conditions on HF paths, DAMSON data may also be used to Table 9.12. DAMSON HF propagation paths Path

Length (km)

Svalbard–Tuentangen Svalbard–Kiruna Harstad–Tuentangen Harstad–Kiruna

2019 1158 1019 194

Figure 9.34. Maps showing positions of Doppler and Multipath Sounding Network (DAMSON) sites (from Angling et al., 1998).

588

High-latitude propagation: 2

Figure 9.35. A schematic illustration of DAMSON analysis program display (from Angling et al., 1998).

Figure 9.36. The biterror-rate (BER) response of a MIL-STD188-110A 75-bps modem to Doppler spread (80% power region) and multipath spread measured at 0 dB SNR (from Angling et al., 1998).

Figure 9.37. An example of the analysis-routine display showing a single spread mode (from Angling et al., 1990).

Figure 9.38. An example of the analysis-routine display with an isometric plot (from Angling et al., 1998).

9.6 Other phenomena

591

evaluate the reliability of a circuit by using different modems on the same paths. The modems tested in this DAMSON experiment appeared to be rather robust, with availabilities of up to 95% on subauroral paths, dropping to 64% on the auroral paths. It was claimed that, using the lowest-frequency sub-band and proper frequency-selection, an auroral-path “availability” of 92.5% would be possible.

9.6

Other high-latitude propagation phenomena and evaluations

9.6.1

Large bearing errors on HF high-latitude paths

Warrington et al. (1997a) and Rogers et al. (1997) have presented results of an HF direction-finder (HFDF) experiment conducted at high latitudes, in which they Table 9.13. A summary of Doppler/SNR plots for multipath spreads of 0–5 ms (from Angling et al., 1998) Path Time

Frequency

S–T

S–K

H–T

H–K

00–24 UT

All frequencies 2.8–4.7 MHz 6.7–11.2 MHz 14.4–21.9 MHz

14.0 15.5 5.0 17.0

17.0 17.5 8.0 20.0

14.8 11.5 8.0 17.5

15.0 9.5 15.0 23.0

19–01 UT

All frequencies 2.8–4.7 MHz 6.7–11.2 MHz 14.4–21.9 MHz

11.5 8.0 6.5 15.0

13.0 12.0 9.5 16.0

13.0 4.0 10.0 16.0

15.5 4.5 11.0 20.0

00–24 UT

All frequencies 2.8–4.7 MHz 6.7–11.2 MHz 14.4–21.9 MHz

8.5 11.3 9.0 5.7

9.8 7.7 8.9 15.2

2.7 1.7 1.9 3.9

19.5 5.3 27.7 50.9

19–01 UT

All frequencies 2.8–4.7 MHz 6.7–11.2 MHz 14.4–21.9 MHz

9.7 13.0 9.2 7.6

12.3 8.9 11.2 18.9

3.8 2.6 4.0 4.7

30.9 8.1 36.3 50.0

CNR (3 kHz) (dB)

Doppler spread (Hz)

Notes: S–TSvalbard–Tuentangen S–KSvalbard–Kiruna H–THarstad–Tuentangen H–KHarstad–Kiruna

592

High-latitude propagation: 2

Table 9.14. A summary of Doppler/multipath plots for SNR of 5 to 5 dB (from Angling et al., 1998) Path Time

Frequency

S–T

S–K

H–T

H–K

00–24 UT

All frequencies 2.8–4.7 MHz 6.7–11.2 MHz 14.4–21.9 MHz

4.0 5.2 2.6 0.6

4.6 5.5 2.5 1.1

3.8 4.1 2.5 0.6

9.8 3.8 10.7 5.1

19–01 UT

All frequencies 2.8–4.7 MHz 6.7–11.2 MHz 14.4–21.9 MHz

3.2 4.2 1.7 0.6

4.1 4.2 1.9 1.1

3.2 3.1 2.6 0.6

7.5 1.9 9.2 6.3

00–24 UT

All frequencies 2.8–4.7 MHz 6.7–11.2 MHz 14.4–21.9 MHz

4.9 5.3 3.1 0.6

6.1 6.1 7.4 3.1

5.4 4.6 9.1 0.7

10.7 5.1 11.2 5.2

4.2 4.3 2.9 0.6

4.6 4.6 4.1 4.1

5.1

8.2

19–01 UT

All frequencies 2.8–4.7 MHz 6.7–11.2 MHz 14.4–21.9 MHz

00–24 UT

All frequencies 2.8–4.7 MHz 6.7–11.2 MHz 14.4–21.9 MHz

11.2 13.5 12.0 7.2

16.0 11.5 12.5 22.2

2.9 1.8 4.3 3.0

31.6 4.5 30.3 54.6

19–01 UT

All frequencies 2.8–4.7 MHz 6.7–11.2 MHz 14.4–21.9 MHz

11.3 12.8 8.8 9.3

15.5 7.0 10.5 25.8

4.8 2.5 6.0 3.9

44.7 4.9 32.9 53.0

00–24 UT

All frequencies 2.8–4.7 MHz 6.7–11.2 MHz 14.4–21.9 MHz

16.4 15.8 17.9 12.4

24.2 14.9 23.3 30.0

9.7 2.8 22.0 11.4

36.0 8.3 31.0 55.0

16.0 15.5 11.2 17.7

25.3 11.2 14.8 31.9

13.5

46.5

19–01 UT

All frequencies 2.8–4.7 MHz 6.7–11.2 MHz 14.4–21.9 MHz

No guard

Composite multipath spread (ms) Guard 0–1.67 ms

No guard

Composite Doppler spread (Hz) Guard 0–1.25 Hz

Notes: S–TSvalbard–Tuentangen S–KSvalbard–Kiruna H–THarstad–Tuentangen H–KHarstad–Kiruna

9.3 6.4

34.1 53.4

9.6 Other phenomena

report finding azimuthal deviations on paths tangential to the auroral oval up to 100° from the great-circle path (GCP), as predicted by Bates et al. (1966). The measurements were made using a wide-aperture goniometric direction-finding system with a dual-band antenna array, a single receiver, and a computer-based data-collection and -processing system on frequencies from 3 to 30 MHz. The VOACAP program (basically a mid-latitude data base) was used to predict mode structure on the high-latitude propagation paths. Rogers et al. (1997) conclude that ⬃50° bearing deviations and deviations as large as 100° from the GCP are primarily due to lateral reflection from the walls of the mid-latitude ionospheric trough – this was also confirmed by Warrington et al. (1997a). In another paper, Warrington et al. (1997b) also report bearing deviations of up to 100° from the GCP on paths contained within the polar cap. Reception of transmissions on frequencies near 8 MHz from Iqaluit, Canada (D2100 km) and Thule, Greenland (D670 km) were analyzed for the period from January through April 1994 (near the minimum of solar cycle 22). The authors attribute these large bearing deviations to lateral reflections from large, drifting electrondensity structures such as dense plasma and sun-aligned arcs. It is also interesting to speculate that these large bearing deviations might be caused by reflection from large-scale TIDs, which have been observed propagating in the polar-cap F region (see Rice et al., 1988; and Williams, 1989). Smaller bearing deviations from the GCP were reported by Warrington (1997) using the DAMSON HF experimental system on circuits between Svalbard and Cricklade, UK (D3073 km) and a 1383-km polar-cap path between Isfjord and Alert, Canada. Measurements on the trans-auroral Isfjord–Cricklade path were made on a frequency of 14.4 MHz for 7 days in late 1995 and early 1996 during the interval 1100–1600 UT and measurements on the Isfjord–Alert circuit were made from ⬃0145 to 1342 UT on 22 January 1996. Indicated bearing deviations up to 2.5° were found, while, for signals on the polar-cap path from Isfjord to Alert, standard deviations of bearing deviation of up to 35° were observed. A variation in bearing with Doppler shift was frequently evident and interpreted as evidence that the signal was scattered from ionospheric irregularities drifting across the reflection points. The 1997 HFDF measurements of large bearing deviations from the GCP are a verification of results based on time-delay measurements reported by Bates et al. (1966). 9.6.2

Effects of substorms on auroral and subauroral HF paths

Effects of an auroral substorm and ionospheric modification on HF signals propagated in February 1996 were reported by Blagoveschchenskaya et al. (1998). HF transmissions from London on 9.410 and 12.095 MHz were received directly at St Petersburg along with a signal reflected from the heated ionosphere over

593

High-latitude propagation: 2

594

Tromsø. Dynamic Doppler spectra on these received signals showed the presence of well-defined field-aligned scattered signal components that peaked during the maximum substorm phase. The proposed scattering mode is illustrated in the ionospheric ray-tracings in Figures 9.39 and 9.40. Substorm effects on HF propagation on four paths (transmissions from Quito, Havana, Ottawa and London received at St Petersburg) were also reported by Blagoveschchensky and Borisova (1998). The principal substorm effects are a substantial growth in strength of the signal several hours before the expansion phase of the substorm and a more significant influence of the ionospheric irregularities inside the poleward edge of the main ionospheric trough on the structure of the signal. 9.6.3

Use of GPS/TEC data to investigate HF auroral propagation

Hunsucker et al. (1995) presented results of an investigation utilizing GPS TEC “signatures” to forecast AE ionization on a 950-km east–west 25.5-MHz propagation path inside the auroral oval as shown in Figure 9.41. The strength and duration of the signal from the AE experiment (Hunsucker et al., 1996) were correlated to “signatures” obtained when the propagation path from GPS through the satellite was recorded at Fairbanks. Figure 9.41 shows the TEC signature types, data showing the structure over the mid-point of the path, an illustration of one particular signature, and the result of “filtering” of TEC signatures. During the period from December 1993 through January 1995, 58 passes of the GPS prn 28 satellite whose LOS path to Fairbanks was near the mid-point of E-region propagation path for the 25.5-MHz Wales–Fairbanks propagation path were analyzed. The GPS/TEC indications of AE, along with the strength and duration of the AE signal are shown in Figure 9.42. From analysis of these GPS passes in winter 1993–1994, it appears that it may be feasible to predict propagation of high-HF-to-low-VHF signals on a near-realtime basis if the mid-points of the E-region propagation paths are within the auroral oval. An extension of the technique described above to detect auroral activity within the oval was reported by Coker et al. (1995). An example of the latitudinal distribution of 1-min GPS satellite LOS tracks through the E region for 1 December 1993, detections of AE, and the College, Alaska magnetometer H trace are shown in Figure 9.43. Verifications of the relation between GPS/TEC detection of AE and TIROS precipitating particle auroral energy flux for Kp values of 1, 2, 3 and 4 are shown in Figure 9.44. Figure 9.45 illustrates how the GPS/TEC data can define the equatorward boundary of the auroral oval.

9.6 Other phenomena

Figure 9.39. Simulated ray-tracing of the field-aligned scattered HF signals from the Es region on the London–St Petersburg propagation path for 12.095 MHz for the geophysical conditions of 17 February 1996 at 2030 UT (Kp 3); (a) height of Es 110 km; and (b) height 130 km (after Blagoveschenskaya et al., 1998).

595

596

High-latitude propagation: 2

Figure 9.40. The same as Figure 9.38, but for f 9.410 MHz (after Blagoveschenskaya et al., 1998).

To quote the authors, Tremendous potential exists for monitoring the effects of auroral substorms (space weather) in real-time. A single GPS station or network of stations could track the motion of the equatorward edge of the oval, which is an important boundary for understanding magnetospheric processes.

9.7 Summary and discussion

9.6.4

The performance of HF modems at high latitude using multiple frequencies

In a recent paper, Jodalen et al. (2001) evaluated the performance of two of the “robust-Waveform” modems at 75 bps (STANAG 4415/4285) and 2400 bps (STANAG 4285) along with Morse and voice transmissions on short and midrange high-latitude paths. Data were acquired from April to December 1995 for smoothed numbers of sunspots 35–70, with average K indices of 0–3 at Kiruna, but including one disturbed period (K5). The paths are shown in Figure 9.46 and the method of comparisons between DAMSON measurements and simulated performance of modems is shown in Figure 9.46. Figures 9.47 and 9.48 show the overall availability of modems when the frequency set consists of 1, 2, . . . , 10 frequencies for the Isfjord–Tuentangen path (2019 km) and the Harstad–Kiruna path (194 km), respectively. The authors concluded from this investigation that the data rate must be sacrificed if high availability is required. Specifically, when there is mode support on the 2019-km path, the availability of two robust modems was 60%–70% higher than that achievable from the 2400-bps modem. On the 194-km path the availability was typically 75% higher. Also, the 75-bps modem benefits from being better able to cope with scattered and off-great-circle modes, therefore providing frequency availabilities above the MUF. The maximum overall availability achieved with a certain number of frequencies of the robust modems was high for both paths for all seasons, but a degradation of 5%–10% was observed on the short path during a geomagnetic disturbance. Using the robust modem (75 bps), the overall availability needed only one frequency during summer and winter and four frequencies during the disturbed period on both paths. The 2400 bps modem needed three or four frequencies on the short path and four to six frequencies on the long path for all periods. Roesler and Carmichael (2000) have reported that error-free transmissions of data approaching 9600 bps in a 2-kHz bandwidth and 19 200 bps in a 6-kHz intersymbol-interference-bandwidth mode have been achieved using a quadratureamplitude-modulation waveform and the STANAG 5066 modem in an automatic-request-for-repeat system (see Section 9.5.4). These data rates were measured on HF paths from Cedar Rapids, Iowa to Ottawa, Canada (1336 km) and Cedar Rapids to San Diego, California (2467 km).

9.7

Summary and discussion

It is obvious that a large amount of research has been carried out in the last four decades on radio propagation at high latitudes (mostly HF ionospheric propagation), but much of it is to be found in relatively obscure reports and conference proceedings. For that reason, we feel justified in including a considerable number

597

TECU

100

200

300

Distance from closest approach to midpoint (km)

0

Type 4 – Structure off of Midpoint

Type 3 – Structure over Midpoint

Type 2 – Structured Slant

Type 1 – Smooth Slant

TEC Signature Types

0 –500 –400 –300 –200 –100

5

10

15

20

25

400

500

TECU

day 344

day 020

day 028

100

200

300

400

day 002 – medium AEI

day 027 – strong AEI

day 349 – medium AEI

Distance from closest approach to midpoint (km)

0

day 365

TEC Data for pm 28 – Type 3: Structure over Midpoint

0 –500 –400 –300 –200 –100

5

10

15

20

25

30

35

40

500

day 342 – week AEI

day 336

day 351

day 001 – strong AEI

100

200

300

Distance from closest approach to midpoint (km)

0

day 335 – strong AEI

0 –500 –400 –300 –200 –100

5

10

15

20

25

30

35

40

400

500

0

100

200

day 330 – week AEI

day 330 – week AEI

100

200

300 Distance from closest approach to midpoint (km)

0

day 350

day 335 – strong AEI

300

day 345 – strong AEI

day 350

TEC

Filtered TEC (400 km cutoff)

0 –500 –400 –300 –200 –100

–2

0

2

4

20 –500 –400 –300 –200 –100

25

30

35

40

Figure 9.41. Examples of GPS/TEC “signatures” and auroral-E propagation (see legends below each figure) (from Hunsucker et al., 1996).

TECU

TEC Data for pm 28 – Type 2: Structured Slant

TECU TECU

400

400

500

500

High-latitude propagation: 2

600

AEI strength

AEI duration

Figure 9.42. A comparison of predicted and measured AEI for 60 GPS satellite passes (from Hunsucker et al., 1996).

Passes

Prediction

no maybe

yes

0

21 dB

0

34 min

of examples of data obtained by these research programs, especially since circuit behavior displays such a profound variation with solar-terrestrial conditions, path orientation, frequency, and even type of modulation. Specific examples have been included to illustrate these variations. Ionospheric modeling and propagation-prediction techniques have been improved significantly since the early 1970s for mid-latitudes, but most of the extant models are still inadequate for realistic portrayal of the auroral and polar ionosphere. It is important to note that practically all of the models and prediction programs are “climatological” in nature and do not really predict the “weather” aspects of propagation. In spite of this caveat, attempts to compare predictions with realtime data continue to be made. Fortunately, recent advances in availability of “space-weather” data, improvements in ionospheric data bases and new modeling theory have somewhat improved the situation. The advent of “realtime channel evaluation (RTCE), automatic link-quality (ALQ) evaluation, robust modems, and computer-controlled frequency management provide order-of-magnitude improvements in reliability of HF high-latitude propagation. If sufficient resources are available, probably the best approach is to use the above techniques to achieve high reliability on HF high-latitude circuits, instead of expending resources on improving ionospheric data bases, modeling, and prediction techniques.

9.7 Summary and discussion

Figure 9.43. The latitudinal distribution of 1-min GPS satellite LOS tracks through the auroral-E region for 1 December 1993 (from Coker et al., 1995).

Recent accurate measurements of HF bearing deviation have revealed that, as demonstrated by time-delay analysis in 1966, deviations as great as 100° from the great-circle-path (GCP) often occur on HF high latitude paths. Since most practical HF communication systems utilize antennas with ⬃50°–70° azimuthal beamwidths, it may be useful at times to have the capability to rotate the antenna beam to take advantage of this non-great-circle (NGC) mode. This could further improve the reliability of HF high-latitude circuits, since research in the 1960s indicated that the MOF on the circuit was sometimes carried by the NGC mode and the NGC mode is not an uncommon occurrence. Promising areas of research include validation of ionospheric models and prediction programs using quantitative HF circuit data appropriate to the outputs of the models (relatively little has been accomplished so far); near realtime availability of space-weather data needed for specific radio-propagation characterization,

601

Figure 9.44. The latitudinal variation in 1-h TEC data compared with TIROS particle precipitation for Kp 1–4 (from Coker et al., 1995).

9.7 Summary and discussion

Figure 9.45. The latitudinal distribution of 1-min GPS satellite LOS tracks through the E region for 1 December 1993 (top). Oval detection compared with a model of the equatorward boundary and individual TIROS passes for 1 December 1993 (bottom) (from Coker et al., 1995).

603

604

High-latitude propagation: 2 Figure 9.46. The method of comparison between DAMSON measurements and the simulated performance of modems (from Jodalen et al., 2001).

improvement of the spatial and temporal resolution of ionospheric models, and utilization of existing three-dimensional ionospheric ray-tracing techniques to verify high-latitude propagation modes by understanding magnetospheric processes. Similarly, GPS stations could monitor areas of intense AE ionization due to particle precipitation, which support high-HF and low-VHF propagation.

604

9.7 Summary and discussion

Figure 9.47. The overall availability of modems when the frequency set consists of 1, 2, . . ., 10 frequencies for the Isfjord–Tuentangen path (2019 km) (from Jodalen et al., 2001).

605

606

High-latitude propagation: 2

Figure 9.48. The same as Figure 9.45, but for the Harstad–Kiruna path (194 km) (from Jodalen et al., 2001).

9.8 References and bibliography

9.8

References and bibliography

Section 9.1 Hunsucker, R. D., Rose, R. D., Adler, R. W., and Lott, G. K. (1996) Auroral-E mode oblique HF propagation and its dependence on auroral oval position. IEEE Trans. Antennas Propagation 44, 383–388. Nishino, M., Gorokhov, N., Tanaka, Y., Yamagishi, H., and Hansen, T. (1999) Probe experiment characterizing 30 MHz radio wave scatter in the high-latitude ionosphere. Radio Sci. 34, 833–898.

Section 9.2 Aarons, J., Kersley, L., and Rodger, A. S. (1995) The sunspot cycle and “auroral” F-layer irregularities. Radio Sci., 30, 631–638. Anderson, D. N., Buonsanto, M. J., Codrescu, M., Decker, D., Fesen, G. G., Fuller-Rowell, T. J., Reinisch, B. W., Richards, P. G., Schunk, R. W., and Sojka, J. J. (1998) Intercomparison of physical models and observations of the ionosphere. J. Geophys. Res. 103, 2179–2192. Bent, R. B., Llewellen, S. K., Nesterczuk, G., and Schmid, P. E. (1975) The development of a highly successful worldwide empirical ionosphere model and its use in certain aspects of space communication and in worldwide total electron content investigations. In Proc. IES75 (ed. J. Goodman). US Government Printing Office, Washington DC. Bibl, K. (1998) Evolution of the ionosonde. Annal: de Geofisica 41. Bilitza, D. (1999) IRI 2000. In Proc. IES99, pp. 348–351. Bishop, G. J. et al. (1999) The effect of the protonosphere on the estimation of GPS total electron content: validation using model simulations. Radio Sci. 34, 1261. Burtch (1991) A comparison of high-latitude ionospheric propagation predictions from ICEPAC with measured data. M. S. Thesis. Naval Postgraduate School, Monterey, California. Bust, G. S. and Coco, D. (1999) CIT analysis of the combined ionospheric campaign (CIC).Proc. IES99, pp. 508–518. Chiu, Y. T. (1975) An improved phenomenological model of ionospheric density. J. Atmos. Terr. Phys. 37, 1563–1570. Davé, N. (1990) The use of mode structure diagrams in the prediction of high-latitude HF propagation. Radio Sci. 30, 309–323. Davies, K. (1965) Ionospheric Radio Propagation. National Bureau of Standards, Washington DC. Decker, D. T. et al. (1999) Longitude structure of ionospheric total electron content at low latitudes measured by the TOPEX/Poseidon satellite. Radio Sci. 34, 1239. Feldstein, Y. I. and Galperin, Yu. I., (1985) The auroral luminosity structure in the high-latitude upper atmosphere: its dynamics and relationship to the large-scale structure of the Earth’s magnetosphere. Rev. Geophys. 23, 217. Ferguson, J. and Snyder, F. P. (1986) The segmented waveguide program for long wavelength propagation calculations. NAVOCEANSYSTEM Report TD-1071.

607

608

High-latitude propagation: 2

Ferguson, J. A. (1995) Ionospheric model validation at VLF and LF. Radio Sci. 30, 775–782. Ferguson, J. and Snyder, F. P. (1989) Long wave propagation assessment. In Operational Decision Aids for Exploiting or Mitigating Electromagnetic Propagation Effects (eds. Albrecht and Richter). AGARD-CP-453. Ganguly, S. and Brown, A. (1999) Real time characterization of the ionosphere using diversity data and models. Proc. IES99, pp. 365–376. Gikas, S. S. (1990) A comparison of high-latitude ionospheric propagation predictions from advanced PROPHET 4.0 with measured data. M. S. Thesis. Naval Postgraduate School, Monterey, Calfornia. Goodman, J. M. (1992) HF Communication – Science and Technology. Van Nostrand Reinhold, New York. Hunsucker, R. D. and Owren, L. (1962) Auroral sporadic-E ionization. J. Res. NBS D 66, 581–592. Hunsucker, R. D. (1965) On the determination of the electron density within discrete auroral forms in the E-region. J. Geophys. Res. 70, 3791–3792. Hunsucker, R. D. (1971) High-frequency propagation predictions and analysis for circuits from the USCG San Francisco radio station to ships and aircraft operating in the North Pacific area. OT/TRER 15. Boulder, Colorado. Hunsucker, R. D. (1982) Atmospheric gravity waves generated in the high latitude ionosphere: a review. Rev. Geophys. Space Phys. 20, 293–315. Hunsucker, R. D. and Delana, B. S. (1988) High Latitude Field-strength Measurements of Standard Broadcast Band Skywave Transmissions Monitored at Fairbanks, Alaska. Geophysical Institute, University of Alaska, Fairbanks, Alaska. Hunsucker, R. D. (1992) Auroral and polar-cap ionospheric effects on radio propagation. IEEE Trans. Antennas Propagation 7, 818–828. Hunsucker, R. D. (1999) Final Report on PENEX Data Analysis Project for the Naval Postgraduate School. Naval Postgraduate School, Monterey, California. Jones, R. M. and Stephenson, J. J. (1975) A Versatile Three-Dimensional Ray Tracing Computer Program for Radio Waves in the Ionosphere. USGPO, Washington DC. Lane, G. (1993) Voice of America coverage analysis program (VOACAP). US Information Agency, Bureau of Broadcasting Engineering Report 01-93, p. 203. McNeal, G. D. (1995) The high frequency environment at the ROTHR Amchitka radar site. Radio Sci. 30, 739–746. Mather, R. A., Holtzclaw, B. L., and Swanson, R. W. (1972) High-latitude HF signal transmission characteristics. In Radio Propagation in the Arctic Conference Proc. CP-97. McDowell, A. I., Breakall, J. K., and Lunnen, R. (1993) Project PENEX Interim Report – 1993, High-frequency Receiving Site Research Center at Rock Springs. Applied Research Laboratory, Pennsylvania State University State College, Philadelphia. Milan, S. E., Lester, M., Jones, T. B. and Warrington, E. M. (1998) Observations of the reduction in the available HF band on four high latitude paths during periods of geomagnetic disturbance. J. Atmos. Terr. Phys. 60, 617–629.

9.8 References and bibliography

Omura, J. K., Schultz, R. A., and Levitt, B. K. (1985) Spread Spectrum Communications, volumes I–III. Computer Science Press, Rockville, Maryland. Rose, R. B. (1982) An emerging propagation prediction technology. In Effects of the Ionosphere on Radiowave Systems (IES81) (ed. J. Goodman). US Government Printing Office, Washington DC. Rose, R. B. (1993) Project PENEX: Polar, Equatorial, Near Vertical Incidence Experiment – Methodology Document, Rev. 1. Naval Command, Control and Ocean Surveillance Center, RDT&E Division, San Diego, California. Rush, C. M. et al. (1984) Maps of fôF2 derived from observations and theoretical data. Radio Sci. 19, 1083. Sailors, D. B. and Rose, R. B. (1993) HF Skywave Field Strength Predictions. NraD/NOSC, RDT&E, San Diego, California. Sailors, D. B. (1995) A discrepancy in the international radio consultative committee report 322-–3 radio noise model: the probable cause. Radio Sci. :30, 713–728. Schunk, R. W. (1996) Solar–Terrestrial Energy Program: Handbook of Ionospheric Models. STEP Report Center for Atmospheric and Space Science, Utah State University, Logan, Utah. Smith, R. W. (1988) Low latitude ionospheric effects on radiowave propagation. Dissertation. Naval Postgraduate School, Monterey, California. Szuszczewicz, E. P., Blanchard, P., Wilkinson, P., Crowley, G., Fuller-Rowell, T., Richards, P., Abdu, M., Bullet, T., Hanbaba, R., Lebreton, J. P., Lester, M., Lockwood, M., Millward, G., Wild, M., Pulinets, S., Reddy, B. M., Stanislawska, I., Vannorini, G., and Zoleski, B. (1998) The first real-time worldwide ionospheric predictions network: an advance in support of space borne experimentation, on-line model validation and space weather. Geophys. Res. Lett. 25, 449–452. Thrane, E. V., Jodalen, V., Stewart, E., Saleem, D., and Katan, J. (1994) Study of measured and predicted reliability of the ionospheric HF communication channel at high latitudes. Radio Sci. 29, 1293–1309. Tsolekas, M. D. (1990) A comparison of high latitude ionospheric propagation predictions from IONCAP-PC 2.5. M. S. Thesis. Naval Postgraduate School, Monterey, California. Warber, C. R. and Field, E. C. Jr (1995) A long wave transverse electric–transverse magnetic noise prediction model. Radio Sci. 30, 783–797. Wilson, D. J. (1991) A comparison of high-latitude ionosphere propagation predictions from AMBCOM with measured data. M. S. Thesis. Naval Postgraduate School, Monterey, California.

Section 9.4 Hardy, D. A., Gussenhoven, M. S., and Brautigan, D. (1987) A statistical model of auroral ion precipitation 2. Functional representation model of the average patterns. J. Geophys. Res. 96, 5539–5547. Hunsucker, R. D. (1992) Auroral and polar-cap ionospheric effects on radio propagation. IEEE Trans. Antennas Propagation 7, 818–828.

609

610

High-latitude propagation: 2

Section 9.5 Angling, M. J., Cannon, P. S., Davies, N. C., Willink, T. J., Jodalen, V., and Jundborg, B. (1998) Measurements of Doppler and multipath spread on oblique high-latitude HF paths and their use in characterizing data modem performance. Radio Sci. 33, 97–107. Bliss, D. H., Roessler, D. P., and Hunsucker, R. D. (1987) Preliminary results from a trans-auroral HF experiment. Proc. MILCOM87. Brant, D., Lott, G. K., Paluszek, S. E., and Skimmons, B. E. (1994) Modern HF mission planning combining propagation modeling and real-time environmental monitoring. Proc. IEE94. Davies, N. C. and Cannon, P. S. (1993) DAMSON – a system to measure multipath dispersion, Doppler spread and Doppler shift on multi-mechanism communications channels. Presented at AGARD Electromagnetic Wave Propagation Paths: Their Characteristics and Influences on System Design, Rotterdam. Fenwick, R. B. and Villard, O. G. (1963) A test of the importance of ionosphere– ionosphere reflections in long distance and around-the-world HF propagation. J. Geophys Res. 68, 5659–5666. Fenwick, R. B. and Woodhouse, T. J. (1979) Real-time adaptive HF frequency management. In Special Topics in HF Propagation, AGARD Conference Proc. No. 263 (ed. V. J. Coyne). Gerson, N. C. (1962a) Radio Wave Absorption in the Ionosphere, p. 113. Pergamon Press, London. Gerson, N. C. (1962b) Polar radio noise. In Arctic Communications (ed. B. Landmark). Pergamon Press, New York. Gerson, N. C. (1964) Polar communications. In Arctic Communications (ed. B. Landmark). Pergamon Press, New York. Goodman, J. M. (1992) HF Communication – Science and Technology. Van Nostrand Reinhold, New York. Goodman, J. M., Ballard, J. and Sharp, E. (1997) A long-term investigation of the HF communication channel over middle and high latitude paths. Radio Sci. 32, 1705–1715. Hu, S., Bhattacharjee, A., Hou, J., Sun, B., Roesler, D., Frierdich, S., Gibbs, N., and Whited, J. (1998) Ionospheric storm forecast for high-frequency communications. Radio Sci. 33, 1413–1428. Hunsucker, R. D. and Bates, H. F. (1969) Survey of polar and auroral region effects on HF propagation. Radio Sci. 4, 347–375. Jodalen, V., Bergsvik, T., Cannon, P. S., and Arthur, P. C. (2001) The performance of HF modems on high latitude paths using multiple frequencies. Radio Sci. 36, 1687. Johnson, E. E., Desourdis, R. I., Jr, Earle, G. D., Cook, S. C., and Ostergaard, J. C. (1997) Advanced High-frequency Radio Communications. Artech House, Boston.

Section 9.6 Bates, H. F., Albee, P. R., and Hunsucker, R. D. (1966) On the relationship of the aurora to non-great-circle HF propagation. J. Geophys. Res. 71, 1413–1420.

9.8 References and bibliography

Blagoveshchenskaya, N. F., Korienko, V. A., Brekke, A., Rietveld, M. T., Kosch, M., Borisova, T. D., and Krylosov, M. V. (2000) Phenomena observed by HF longdistance diagnostic tools in the HF modified auroral ionosphere during a magnetospheric substorm. Radio Sci. 34, 715–724. Blagoveshchensky, D. V. and Borisova, T. D. (2000) Substorm effects of ionosphere and HF propagation. Radio Sci. 35, 1165. Coker, C., Hunsucker, R. and Lott, G. (1995) Detection of auroral activity using GPS satellites. Geophys. Res. Lett. 22, 3259–3262. Hunsucker, R. D., Coker, C., Cook, J., and Lott, G. (1995) An investigation of the feasibility of utilizing GPS/TEC “Signatures” for near-real-time forecasting of auroral-E propagation at high-HF and low-VHF frequencies. IEEE Trans. Antennas Propagation 43, 1313–1318. Hunsucker, R. D., Rose, R. D., Adler, R. W., and Lott, G. K. (1996) Auroral-E mode oblique HF propagation and its dependence on auroral oval position. IEEE Trans. Antennas Propagation 44, 383–388. Rice, D. D., Hunsucker, R. D., Lanzerotti, L. J., Crowley, G., Williams, P. J. S., Craven, J. D., and Frank, L. (1988) An observation of atmospheric gravity wave cause and effect during the October 1995 WAGS campaign. Radio Sci. 23, 919–930. Roesler, D. P. and Carmichael, W. R. (2000) The implications and applicability of the QAM high data rate modem. IEE (in press). Rogers, A. S., Warrington, N. C., Jones, E. M., and Jones, T. B. (1997) Large HF bearing errors for propagation paths tangential to the auroral oval. IEE Proc. Microwaves, Antennas and Propagation 144, 91–96. Warrington, E. M. (1997) Observations of the directional characteristics of ionospherically propagated HF radio channel sounding signals over two high latitude paths. Proc. 2nd Symp. on Radiolocation and Direction Finding. SwRI, San Antonio, Texas. Warrington, E. M., Jones, T. B., and Dhanda, B. S. (1997a) Observations of Doppler spreading on HF signals propagating over high latitude paths. IEE Proc. Microwaves Antennas Propagation 144, 215–220. Warrington, E. M., Rogers, N. C., and Jones, T. B. (1997b) Large HF bearing errors for propagation paths contained within the polar cap. IEE Proc. Microwaves Antennas Propagation 144, 241–249. Williams, P. J. S. (1989) Observations of atmospheric gravity waves with incoherent scatter radar. Adv. Space Res. 9, 65–72.

611

Appendix: some books for general reading

Each of the following titles addresses a good range of the geophysical topics that have concerned us, including, in particular, chapters or articles on the upper atmosphere, the ionosphere and magnetosphere, and the aurora and substorms. They are therefore especially useful as works of general reference. Obviously, each will reflect the state of knowledge at the time it was written. While the more recent should be the most up to date, the older ones should not be neglected for they are closer to the development of the basic ideas and knowledge upon which the field stands today. Mitra’s famous book of 1952 is well worth re-reading. The auroral classics by Harang (1951) and Stormer (1955) are cited in Chapter 6. Brekke, A. Physics of the Upper Polar Atmosphere. Wiley, Chichester, New York, Brisbane, Toronto and Singapore (1997). Deehr, C. S. and Holtet, J. A. (eds.) Exploration of the Polar Upper Atmosphere. Reidel, Dordrecht (1981). Hargreaves, J. K. The Solar–Terrestrial Environment. Cambridge University Press, Cambridge (1992). Hines, C. O., Paghis, I., Hartz T. R., and Fejer, J. A. (eds.) Physics of the Earth’s Upper Atmosphere. Prentice-Hall, Englewood Cliffs, NJ (1965). Jacobs, J. A. (ed.) Geomagnetism; volume 3 and 4. Academic Press, London (1989, 1991). Mitra, S. K. The Upper Atmosphere. The Asiatic Society, Calcutta (1952). Scovli, G. (ed.) The Polar Ionosphere and Magnetospheric Processes. Gordon and Breach, New York (1970).

612

Index

absorption cross-section 15, 24 acoustic gravity waves and the aurora 331–332 acoustic gravity waves, theory 52–57 adiabatic invariants 79 aeronomy, physical 13–23 airglow and atmospheric cooling 9 Alfvén Mach number 72 Alfvén wave 103 all-sky camera 291 alpha-Chapman layer 18 antennas 115–116 basic principles 115 design 116 atmospheric composition 10–13 atmospheric heating 8–9 attachment coefficient 18 attenuation 115 atmospheric absorption 122 ionospheric absorption 151 measurement techniques 203–210 deviative techniques 151 non-deviative techniques 144 aurora australis 291 aurora borealis 291 aurora altitude of 296 diffuse 300–301 discrete 300–301 intensity of 299–300 luminous 285, 291–302 mantle 300 radar 285, 326–329 theta 288, 302 auroral activity predictions 367–371 electrojet 312–314 forms 296 green line 302 infrasonic waves 300–331 red line 302

613

auroral oval 286–288 and radio scintillation 308 boundaries of 289–290 models of 288–291 auroral radar 326–329 auroral radar echoes, occurrence of 328–329 polarization of 328 auroral radio absorption 285, 304, 339–382 and geomagnetic activity 365 and HF propagation 365–367 conjugacy of 379–382 co-rotation in 363–365 duration of 351–354 dynamics of 354–365 global movement of 358–359 preceeding bay in 345–347, 359–361 profiles of 373 pulsations in 347, 382 sharp onset in 341–342 substorm onset in 377–379 substorm dynamics in 377–379 slowly varying 347, 363 spatial extent of 351 spike event in 341–345 statistics of 350–354 zone 350–351 auroral spectroscopy 302 auroral substorm 285, 308–311 and ionospheric effects 311–312 break-up of 309 expansion phase of 309 growth phase of 309–310 pseudo-breakup of 310 recovery phase of 309 auroral X-rays 285–304 auroral zone 286 magnetic bays in 305 magnetic disturbances in 285 barometric equation 5 Bartels musical diagram 97–98 beta-Chapman layer 19

Index

614

Bragg curve 107 Bremsstrahlung X-rays 106–107 Brunt–Väisala frequency 54 bursty bulk flow 318 C layer 36 Canadian-border effect 261 Chapman production function 15 character figure 45 charge-exchange reaction 26 chemical transport (of heat) 9 conductivity Hall 49 height variation of 50 of the ionosphere 48–50 of the ground 48 Pedersen 49 with magnetic field 48–49 with no magnetic field 48 continuity equation 14 coronal hole 67 coronal mass ejection (CME) 67–69 current, Birkeland 85–86, 96, 312–314 ring 84–85 currents in substorm 312–315 current system, equivalent 86, 312 Spq 87 SD 95–96 current wedge, wedgelet 313–314 D region at high latitude 337–339 electron flux 373–377 production of 31 profiles in the auroral zone 373 recombination coefficient 400 summer mesospheric echo in 406–409

electric current, Birkland 51 Cowling 51 electric currents in the ionosphere 50 electric field and co-rotation 92 electrojet, auroral 75–76 equatorial 51 EM noise and interference atmospheric noise 127–139 galactic noise 133 solar noise 134–139 emissions, electromagnetic 285 escape temperature 8 EUV (extreme ultra-violet) 14 exobase, definition of 5 exosphere 7–8 definition of 5 F region blobs 245–249 in the auroral oval 240–242 in polar cap: U.T. effect 235–237 in polar cap: the tongue 234–236 patches 244–245 scintillation production in 249–260 storm-time variation of 46–47 F1 layer, aeronomy of 26 F1 ledge 31 F2 region 37–38 alpha–beta transition level in 37 and composition changes 43 and conjugate ionosphere 44 and effect of neutral wind 43 anomalies of 39–44 Bradbury layer in 37 seasonal anomaly of 40 semi-annual anomaly of 40 field-line circulation 316 field-lines 62

Dalton’s law 7 diffusion coefficient 20 ambipolar 22

flux-transfer event 91 frozen-in field 65

diffusion in the ionosphere 20–23 dipole field 61–63

geomagnetic cavity 63 geomagnetic field 61–63

distribution height 21

geopotential height 7

E layer at night 27 E region at high latitude 322–332 aeronomy of 26–31 disturbed auroral 323–326 polar 323 quiet auroral 323 sporadic-E phenomenon in 27–31

Harang discontinuity 95, 242 heliosphere, definition of 5

eddy diffusion in the atmosphere 9

heterosphere, definition of 5 HF propagation at high latitudes 440–530 adaptive HF techniques 574–580 assessment of HF channels 586–590 fixed-frequency tests 440–474 absorption effects 467–474

Index

Alaska/Scandinavia 440–446 other high-latitude paths 450–471, 503–511 Alaska (Wales–Fairbanks) 523–530 Alaska–continental USA 447–450 Alaska–Greenland 471–474 large bearing errors 591–593 mitigation of disturbances 572–574 use of GPS/TEC 594–603 modem use 597–606 PENEX 553–568 realtime HF channel evaluation 580–586 substorm effects 593–594 swept-frequency tests 474–493 auroral-E effects 480–482 Andøya–College 478–480, 503–512 Barrow–Boulder tests 506–516 Canadian tests 517–522 McMurdo–Thule path 514, 517 Sodankylä–Lindau tests 494–502 Sondrestrom–Keflavik tests 528–530 Thule–College tests 474–478 models of high-latitude ionosphere 538–546, 568–572 non-great-circle modes 477–480, 482–483 “ducted” modes 490–493 Doppler and fading characteristics 493–494 ionospheric ray-tracing 538–541 homosphere, definition of 5 hydrated ions 32–35 hydromagnetic wave 103 hydrostatic equation 5 hydrostatic equilibrium 5–7 instability ballooning 319 gradient-drift 328 Kelvin–Helmholtz 105, 318–319 two-stream 104–105, 328 interplanetary magnetic field (IMF) 65 ionization by alpha-particles 107–108 by energetic electrons 105–106 by energetic protons 107–108 ionization efficiency 15, 24 ionization potential 24 ionosphere definition of 4 naming of 1 sluggishness of 27 ionospheric effects of the aurora 302–305 of the sunspot cycle 44–45 of the thermospheric wind 23 ionospheric layers critical frequency of 27 definitions of 13–14 heights of 25 ionospheric storm 46–47 with sudden commencement 46 irregularity strength parameter 258

615

keogram 300 log-normal distribution 369–371 luminous aurora and the E region 325 Lyman-alpha line of solar spectrum 31 M region 285–286 magnetic bay 95–96 magnetic cloud 69 magnetic-field merging 90–91 magnetic field, reconnection of 90–91, 316 magnetic index Ap 96–97 Dst 94 Kp 96–97 magnetic indices 96–100 AU, AL, AE 97–99 magnetic longitude 62 magnetic micropulsation 103–104 magnetic storm 93–102 classical 94 phases of 94 practical effects of 100–102 magnetopause 69–71 and image-dipole method 71 definition of 5 magnetosheath 71–72 magnetosonic wave 103 magnetosphere, circulation of 86–90, 228–234, 316 electric fields in 91–92 definition of 5 shock front of 71–72 magnetosphere boundary layer 73–74 magnetospheric substorm 315–319 magnetotail 70, 72–73 behavior in substorm 316–318 lobes 78 plasma sheet 70, 73, 78 magnetotail, electric potential across 228 ionospheric sources to 240 neutral line 318 mean free path 7 mesopause, definition of 4 mesosphere, definition of 4 metallic ions in the atmosphere 12–13 negative-ion/electron ratio 20, 405 neutral line 90, 318 nitric oxide 31 in the atmosphere 12

616

Index

optical depth 17, 25 oxygen, dissociation of in the atmosphere 10 ozone and the ozonosphere 12 plasmapause 75 plasmasphere 73–78 depletion of 93 dynamics of 92–93 plasmoid 317 polar arc 301–302 polar cap circulation patterns in 228–234 electric field in 228 potential across 91–92 polar-cap absorption 382–406 and solar radio emissions 389–390 and solar flares 389–390 and proton flux 387 day–night effects in 400–405 duration of 384 magnitude of 389 midday recovery in 395–397 occurrence of 384–387 seasonal variation of 387–389 uniformity of 395 polar-cap edge (in PCA) 393–395 polar cusps (clefts) 237–239 and charged particles 237 and luminosity 237 and ionospheric heating 239 polar hole 260, 276–280 polar wind 74, 239–240 propagation conductivity 121 Faraday rotation 149–151 forward scatter 171 HF at high latitudes 440–530 LF and MF at high latitudes 430–439 ionospheric 140–169 ionospheric scatter 171–174 line-of-sight 113–116 lossy medium 120–121 magnetoionic theory 140–145 phase effects 148–149 prediction programs 174 predictions and validations 546–565 scintillations 152–163 terrain effects 125–127 transionospheric 147–159 VHF–microwave 530–531 VLF/UHF at high latitudes 530–531 VLF/ELF principles 163–167 VLF/ELF at high latitudes 419–430 whistlers 167–169 HF-propagation-prediction programs 174–176 noise and interference 127–139 proton effects in the neutral atmosphere 398–406

protonosphere 38–39 base of 39 definition of 5 protons IMF effects on 390–392 magnetospheric effects on 392–395 quiet-day curve 339 radar aurora 303–304 radar, basics of 116–119 radiation in the atmosphere 9 radio absorption in the D region 36 radio absorption, winter anomaly of 36 radio waves interaction with matter 122 terrain effects 125–127 reaction rates, temperature dependence of 43, 274–275 recombination dissociative 19, 26, 33 radiative 26 recombination coefficient 18 effective 20 recombination processes, types of 18 reflection from the ionosphere 144–145 relation between oblique and vertical 145–147 reflection at a boundary 159–163 refractive effects tropospheric 119–120 neutral atmosphere 122–125 relativistic electron precipitation 348 rigidity 393 ring current 73, 84–85 riometer 339–340 scale height, definition of 6 scale height of plasma 22 scatter from the ionosphere 169–174 coherent scatter 169–171 forward scatter 171 incoherent scatter 171–174 scintillation and Fresnel zones 256 modeling 258–260 scintillation, properties of 249–256 S4 index of 255 spectrum of 256 small irregularites, in situ measurement of 257–258 solar wind 63–69 and Kp 97 ballerina model of 67

Index

composition of 63 fast stream in 67 garden-hose effect in 65 sectors 65–66 space-weather-data use 565–574, 600 sporadic-E and metallic ions 29 and scintillation 31 and wind shear 28–29 at high latitude 29–30 Störmer theory 392–395 storm–substorm relations 321–322 stratopause, definition of 4 substorm 308–322 current wedge 313 rate 321 theories of 318–319 triggering 319–321 techniques D-region absorption 203–210 ground-based 181–214 HF Doppler and spaced receiver 217–219 incoherent scatter radars 203–205 in situ measurements 216–217 ionosondes 181–187 ionospheric imaging 219–220 modification by HF transmitters 210–214 oblique-incidence HF/VHF sounders 187–202 riometers (URSI A2 method) 206–208 satellite beacons 215–216 space-based measurements 214–217 topside ionospheric sounders 216–217 URSI A1a and A1b (HF) absorption methods 204–206 URSI A3a and A3b (LF) absorption methods 208–210 temperature of neutral atmosphere 8–10 thermosphere, definition of 4

617

three-body reaction 35 trapped (Van Allen) particles 78–84 longitude drift 83 loss cone of 83 mirror point of 83 pitch angle of 83 pseudo-trapping of 83 traveling ionospheric disturbance 57 tropopause, definition of 4 troposphere, definition of 4 trough and electron precipitation 270–271 trough in electron content 261, 266 in the southern hemisphere 269 main 260–275 motion of 271–273 orientation of 269–270 poleward edge of 269–271 principal properties of 263–265 time and activity variations of 265–269 causes of: heating 274–275 causes of: plasma decay 273–274 troughs at high latitude 276–280 turbopause, definition of 5 turbosphere, definition of 5 Van Allen belts 78–84 vertical transport in the atmosphere 20–23 viscous interaction 86–88 VLF-wave reflection in the D layer 35–36 VLF whistlers and the plasmasphere 75 VLF whistlers, nose 75 wave–particle interaction 104 X-rays 14, 106–107