The High-Latitudude Ionospehere

This page intentionally left blank
The high-latitude ionosphere and its effects on radio

propagation
The physical properties of the ionized layer in the Earths 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 Earths 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 SolarTerrestrial Relations (1979) and The
SolarTerrestrial 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 Earths plasmasphere
J. R. Garratt
The atmospheric boundary layer
D. Hastings and H. Garrett

Spacecraftenvironment interactions
J. K. Hargreaves
The solarterrestrial 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
Oceanatmosphere 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, So 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 dierent 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 prole 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 eects 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 eld 48
1.5.3
The eect of a magnetic eld 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 inuencing the high-latitude ionosphere 61

2.1
Introduction 61
2.2
The magnetosphere 61
2.2.1
The geomagnetic eld 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 elds 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
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 signicance 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 eects 124
3.3.3
Noise and interference 127
3.4
Ionospheric propagation 140
3.4.1
Magnetoionic theory 140
3.4.2
Reection 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 reection 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
Chapter 4 Radio techniques for probing the ionosphere 181

4.1
Introduction 181
4.2
Ground-based systems 181
ix
Contents
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 modication by HF transmitters 210
4.3
Space-based systems 215
4.3.1
A history of Earthsatellite 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
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 eect 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
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 eects 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 inuence 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
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 signicance 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 ares and radio emissions 389
7.3.4
Eects arising during the protons journey to Earth 390
7.3.5
Non-uniformity and the midday recovery 395
7.3.6
Eects in the terrestrial atmosphere 398
7.4
Coherent scatter and the summer mesospheric echo 406
7.5
Summary and implications 409
7.6
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 xed frequencies 439
8.4.2
Tests involving transmission between Alaska and the continental USA 448
8.4.3
Other trans-polar HF experiments on xed frequencies 450
8.4.4
CollegeKiruna absorption studies at xed frequencies 457
8.4.5
Eects 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. 19561969 479
8.4.8
Doppler and fading eects on HF high-latitude propagation paths 492
8.5
VHF/UHF and microwave propagation 529
8.6
Summary 531
8.7
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 ELPHF 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 eorts 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 solarterrestrial 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
Eects 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

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 Marconis famous radio transmission across the Atlantic
Ocean, an experiment closely followed by Kennelly and Heavisides suggestions
that an ionized layer in the Earths upper atmosphere had made it possible. From
the rst, the ionosphere has been put to use, supporting an increasing range of
applications from point-to-point communication and broadcasting, to directionnding, 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 scientic problems and technical diculties 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 inuence of the magnetosphere, to the principles of radio propagation, and to the major techniques of ionospheric observation. Chapters 57
describe the various phenomena of the ionosphere that are peculiar to the high
latitudes. The nal 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 ll 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 nd 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 eld 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
eld 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 solarterrestrial environment. After Synoptic Data for SolarTerrestrial 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 inuence 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
reections.
The early history of the ionosphere is very much bound up with the development of communications. The rst suggestions that there are electried layers
within the upper atmosphere go back to the nineteenth century, but the modern
developments really started with Marconis 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 Earths curvature, the waves could not have traveled directly across the
Atlantic but must have been reected 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 inuence its behavior.
Looked at most simply, the ionosphere acts as a mirror situated between 100
and 400 km above the Earths surface, as in Figure 1.1, which allows reected
Ionosphere
Ground
30
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 reection occurs depend on the radio frequency of the signal, but the most usual mechanism, which applies in the high-frequency (HF) band (330 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
reections between ionosphere and ground. Reection 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 reected from greater
heights, but if the frequency is too high there may be insucient bending and the
signal then penetrates the layer and is lost to space. This is the rst complication
of radio propagation.
The second complication is that the lower layers of the ionosphere tend to
absorb the signal. This eect 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 ecient radio communication.
Further details about radio propagation are given in Chapter 3, and our central
topic of how propagation at high latitudes is aected 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 dierent 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 aected 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 inuenced by electromagnetic forces that arise because the geomagnetic eld 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) ows in the E region, and the F region is subject
to electrodynamic lifting and a fountain eect that distorts the general form of
the ionosphere throughout the low-latitude zone.
At high latitudes we nd the opposite situation. Here the geomagnetic eld
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 eld-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
inuence. 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 aected 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 dierence 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 aecting the ionosphere in general and summarize the
more normal behavior at middle latitudes. In order to do that, we must rst
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
1500
Temperature (K)
10
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 SolarTerrestrial 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 dierent regions are called spheres and the boundaries between them
are pauses. The lowest region is the troposphere, in which the temperature falls
o with increasing height at a rate of 10 K km1 or less. Its upper boundary is the
tropopause at a height of 1012 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 8085 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
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 classication 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 diusion 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
Earths 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
eld controls the particle motions. The outer termination of the geomagnetic eld
(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 atmospheres 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 signicant 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 rst principles. The variation of pressure
with height is
PP0 exp(h/H ),
(1.2)
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 Boltzmanns 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 dene the thickness of an atmosphere. H is greater, and the atmosphere thicker,
if the gas is hotter or lighter. In the Earths atmosphere H varies from about 5 km
at height 80 km to 7080 km at 500 km.
Using equation (1.1), the hydrostatic equation may be written in dierential
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[(h h0)/H ]ez,
(1.5)
where PP0 at the height hh0, and z is the reduced height dened 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
occupy a column extending just one scale height. Note also that the total mass of
the atmosphere above unit area of the Earths 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 eect of changing gravity may be taken into account by dening 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 at 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 Daltons law.
1.2.3
The exosphere
In discussing the atmosphere in terms of the hydrostatic equation we are treating

the atmosphere as a compressible uid whose temperature, pressure, and density
are related by the gas law. This is valid only if there are sucient 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 dierent way, not as a uid but as an assembly of
individual molecules or atoms, each following its own trajectory in the Earths
gravitational eld. 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 15002000 km, at least as an approximation. However, this liberty may not be taken if there is signicant 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 aected most.
The rate at which gas molecules escape from the gravitational eld 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
v 2e 2gr,
(1.10)
where r is the distance of the particle from the center of the Earth. (At the Earths
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 dened an escape temperature (Te).
Te is 84000 K for atomic oxygen, 21 000 K for helium, but only 5200 K for atomic hydrogen. At 10002000 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 insignicant for O, slight for He, but
signicant for H. Detailed computations show that the resulting vertical distribution of H departs signicantly 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 prole of the neutral atmosphere
The atmospheres temperature prole 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 rst 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
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 prole (Figure 1.2) can be explained as follows. The maximum
at the stratopause is due to the absorption of 200300 nm (20003000 ) radiation by ozone (O3) over the height range 2050 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 95250 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 specic 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 115130
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 eect can be
heating instead of cooling.
Transport
The thermal balance and temperature prole of the upper atmosphere are also
aected by processes of heat transport. At various levels conduction, convection,
and radiation all come into play.
Radiation is the most ecient process at the lowest levels, and the atmosphere
is in radiative equilibrium between 30 and 90 km. Eddy diusion, 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 ow represents a
major loss of heat from the thermosphere but is a minor source for the mesosphere.
10
In the thermosphere (above 150 km) thermal conduction is ecient 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 aect 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 inuence 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 diusion, 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
Figure 1.3. Atmospheric composition to 1000 km for a typical temperature prole. (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 Earths crust. It diuses 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, ows 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 signicant for the ionosphere.
Water does not have the same dominating inuence in the upper atmosphere as
in the troposphere. It is important nevertheless, rst 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
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 aected 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 150160 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 diuses down to the mesosphere by molecular and then by eddy diusion. Loss by photodissociation and
recombination, aided by the eect of the low temperature at the mesopause, is
sucient to create a minimum at 8590 km. The diusion 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 34 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 NOO2 NO2
ONO2 O2 NO
O3 O2O2.
(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 aected by the
natural production of nitric oxide discussed above.
Metallic atoms are introduced into the atmosphere in meteors, whose ux 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 proles 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 signicant 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 proles of the ionosphere are shown in Figure 1.4. The identication of the regions was much inuenced by their signatures on ionograms (see
Section 4.2.1), which tend to emphasize inections in the prole, 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, 6090 km: electron density 1081010 m3 (102104 cm3);
E region, 105160 km: electron density of several times 1011 m3

(105 cm3);
13
14
F1 region, 160180 km: electron density of several times 1011 to about 1012
m3 (105106 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]
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 Chapmans 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 sucient 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 stratied: there are no variations in the horizontal

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

and
the absorption coecient is constant: this is equivalent to assuming that

we have monochromatic radiation.
The rate of production of ionelectron 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 rst absorb radiation, and the amount absorbed is expressed by
the absorption crossection, : if the ux 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
eciency,, 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 z sec 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
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 rst 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 prole. 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
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-o. 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 coecient 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 eectively 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.
18
1.3.3
Principles of chemical recombination
Working out the rate of electron production is just the rst 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 rst with the principles
of chemical recombination. The question of which individual reactions are most
important in dierent 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 coecient. 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 dierent 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 coecient. 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 signicant eect on their total number and [M] is eectively
constant.
At equilibrium,
q Ne
(1.31)
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 aect 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)
AXeAX
(1.34)
in which A2 is one of the common molecular species such as O2 and N2. The rst
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 signicant 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)
20
e and i being recombination coecients 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 eective recombination coecient. 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 nal 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 eect of
movements, and then the continuity equation becomes
dN
(wN )
,

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 diusion of the gas, and then
we can put
w
D N
,
N h
(1.41)
D being the diusion coecient. This equation simply states that the bulk drift of
a gas is proportional to its pressure gradient, and it eectively denes the diu-
21
sion coecient 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 diuses through a stationary majority gas) the diusion
coecient may be derived in its simplest form as DkT/(m). Here k is
Boltzmanns 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/dh N/HN)
(1.42)
for the upward speed instead of (1.41). Substitution into the continuity equation
then gives
dN
dN N

D
dt h
dh HN
(1.43)
This is the basic equation that has to be satised by the time and height variations
of those regions (specically the upper F region and the protonosphere) where ion
production and recombination are both suciently 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, dened as

1 dN
N dh
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 dierent masses. Initially the ions, being heavier, tend to settle away from the electrons,
but the resulting separation of electric charge produces an electric eld, E, and a
restoring force eE on each charged particle. This electrostatic force also aects 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)
22
and
Hp k(TeTi )/(mi g),
(1.46)
respectively known as the ambipolar or plasma diusion coecient and the plasma
scale height.
In that part of the ionosphere where plasma diusion is important, the electron
temperature usually exceeds the ion temperature. However, taking Te Ti by way
of illustration, we see that the plasma diusion coecient and scale height are
then just double those of the neutral gas at the same temperature. Eectively, the
light electrons have the eect 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 diusion coecient, 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 diusion coecient can be written as
DD0 exp(hh0)/H
(1.48)
where D0 is the value of D at a height h0. Thus, diusion 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
dN
1 1
DN
dt
Hp
1 1
.

H
(1.49)
If dN/dt0 this has two solutions. The rst, Hp, is diusive 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 diusion coecient). Here, dN/dt0 as before, but the drift speed is
1
1 1
1
wD
D
,
Hp
H Hp
(1.50)
which is not zero since Hp H. The upward ow 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 outow of plasma.
The effect of a neutral-air wind

Since the ow of ionospheric plasma is constrained by the geomagnetic eld, the
exact eect varies with latitude. One consequence is that, at middle latitudes, the
height distribution of ionization is aected by the neutral-air wind which ows 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 eld is U|| U cosI, and the plasma tends to move in the
same way. This motion, along the magnetic eld, has a vertical component
WU|| sin I 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 ow. The eect 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 signicant (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 coecients in Equations (1.29) and (1.31) (q N e2 and
q Ne), and, at the higher levels, consider the diusion coecient (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 uxes 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.
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 ngstrm units and in nanometers.
These values of max immediately identify the relevant parts of the solar spectrum as the X-ray (0.117 nm, 1170 ) and EUV, (17175 nm, 1701750 ),
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 eciency, , is such that, for atomic species, all the absorbed
energy goes into ion production at the rate of one ionelectron 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/ ().
(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
n H sec 1.
i
i i
Table 1.1. Ionization potentials

Maximum
wavelength max
Species
Ionization potential I (eV)
()
(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
25
Wavelength ()
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
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 eect of all gases at a given wavelength has to be considered and then,
since the ionosphere is in eect 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.
26
1.4.2
The E and F1 regions

Aeronomy
The E region which peaks at 105110 km, and the F1 region at 160180 km, are
both fairly well understood. The F1 region is attributed to that part of the solar
spectrum between about 200 and 900 , which is strongly absorbed in atomic
oxygen, whose ionization limit is at 911 . The optical depth reaches unity from
about 140 to 170 km. The band includes an intense solar emission line at 304 .
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 (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
10100 (110 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 signicant in the normal E and F
regions. Dissociative recombination, as
eXY XY,
(1.55)
is 105 times faster (with a reaction coecient 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 rst converted to a
molecular ion by a charge-exchange reaction
O O2 O2 O
or
O N2 NO N.
(1.57)
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 coecient () 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
eect 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 10131014 m3 s1 (107108 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 proles 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.
28
Figure 1.7. Speciman electron-density proles 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
6070 magnetic) and the polar cap (poleward of the auroral zone). A full classication of sporadic-E, particularly regarding its identication 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 sucient intensity to reect 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 eld,
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 coecients are typical of the radiative process (1018 m3 s1), which gives
them relatively long lifetimes. Temperate sporadic-E occurs at heights of 95135
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 110 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
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 dierent 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 signicant in radio propagation because it may reect signals
that would otherwise penetrate to the F region, though in some cases (for example
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 proles 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, rst, to the relatively high pressure, which causes minor as well as
major species to be important in the photochemical reactions, and, second,
because several dierent sources contribute to ion production.
The Lyman- line of the solar spectrum at 1215 penetrates below 95 km and
ionizes the minor species nitric oxide (NO), whose ionization limit is at 1340 .
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
, 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 28 ionize all constituents, the most eect 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 aects 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 o, 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 eects later in the book.
31
32
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 dierent sources vary with latitude,
time of day, and level of solar activity. By way of example, theoretical proles of
the production rate (for solar zenith angle 42 and a 10-cm solar ux of 165 units)
are given in Figure 1.10. Note that all the sources mentioned above are signicant
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 ux varies
strongly with solar activity (by a factor of a hundred to a thousand) and is probably not signicant in the D region at sunspot minimum.
These production-rate proles are consistent with measurements of D-region
electron densities (Figure 1.11). Friedrich and Torkar (1992) analyzed 164 electron-density proles 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 proles 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 rst occurs is a natural boundary within the D region.
33
Figure 1.11. Electron-density proles observed at Arecibo for two solar zenith angles. (J. D.
Mathews, private communication.)
1019
10 20
70
60
40
20
Figure 1.12. Electrondensity proles 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 coecient of about 5
1013 m3 s1, the
reaction of NO being somewhat faster than that of O2 . In total the situation is
much more complex, as illustrated in Figure 1.13. This scheme includes O
, NO,
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 10121011 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 Sodankyl Geophysical Observatory, Finland, includes 24 positive and 11 negative ions, 35 in all. Later versions include as many as 55 ions.
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 MO2 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
anity 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 eect 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 eective recombination coecient (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 rst approximation,
reected 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 reection 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 dierent pattern. The
36
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.)
reection 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 reection 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 0.71.0. 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.
1.4.4
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 rst aquaintance, can be quite puzzling. In the rst place it peaks at
200400 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 eNO
As discussed in Section 1.3.3, the second reaction controls the overall rate at low
altitude and the rst 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]
Ne q/ exp
h
h

H(O) H(N2 )
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
h
H(O)
1
H(O)
H(N2 )
0.75h
.
H(O)
Ne exp
exp
(1.64)
This is a layer whose electron density increases with height because the loss rate
falls o more quickly than does the production rate. It is often called a Bradbury
layer.
38
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 diusion, 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 diusion 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 eective
control.
The characteristic time for recombination is
1/,
(1.65)
and it may be shown that the corresponding time for diusion 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
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 eectively denes 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 prole
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 OHO
is k, then the lifetime of a proton is (k[O])1. Taking the time constant for diusion 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 dicult to predict. From the point of view of the Chapman theory the F2 regions
behavior is anomalous in several ways, and these are sometimes called the classical anomalies of the F2 layer. Briey, 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
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:
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 ux, and the magnitude of the interplanetary magnetic eld. (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.
(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 inuenced 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 rst 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 250300 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
owing equatorward and elevating it when it is owing poleward (Section 1.3.4),
exerts a major inuence both on electron densities and on electron content. At 300
km the neutral wind ows poleward by day and equatorward by night at speeds
ranging between tens and hundreds of m s1. Thus its eect 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 diusion coecient, 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
44
these are not dominant inuences 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 aect 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 ecient.
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 aect
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 ow up
along the eld lines to populate the protonosphere. In the evening the proton population ows 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
eect, since protonospheric plasma, coming mainly from the summer ionosphere,
is equally available to replenish the winter ionosphere. Computations show that
this is a signicant 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 benets
from the conjugate region in the summer hemisphere. At sunrise, when electron
densities are low, the ionosphere may be signicantly heated by photoelectrons
arriving from the conjugate hemisphere. The eect 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 ares occur, or the
45
intensity of the 10-cm radio ux, also aects 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 inuence 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 dened as the highest radio frequencies reected 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[(10.008R)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 gures. 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.
46
1.4.7
The F-region ionospheric storm
From time to time the ionosphere suers 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 dierent mechanisms must be involved. Three phases may be identied.
(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 eect 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 eect 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,
conrming 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 signicant parameter. Statistical studies, as
well as case histories of major storms, show that the magnitude and even the sign
of the eect 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 rst day of the storm and does not reappear on the second day.
Seasonal and hemispheric eects 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 dierence is such that Nm(F2) is actually increased during the
main phase of storms occurring in the southern hemisphere during winter. The
interhemispheric dierence arises from the larger separation between the geographic and the geomagnetic poles in the south.
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 eect of a greater proportion of
molecular species in the F region is to reduce the equilibrium electron density. This
mechanism has been veried 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
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 105102
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 (1071 1 m1) or a semiconductor
(101102 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 aected by geophysical disturbances. These are particularly important at the high latitudes. The
solargeophysical 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 eld
If no magnetic eld 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 eld) 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 eld
Unfortunately the Earths magnetic eld permeates the ionosphere, and this complicates the conductivity enormously. A charged particle moving through a magnetic eld 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
Hall current
Figure 1.19. Currents due to the electric-eld components parallel (E) and perpendicular
(E) to the magnetic eld (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 eld and to the direction of motion of the particle.
If the particle is moving directly along the magnetic eld, the Lorentz force is zero;
the magnetic eld has no eect and Equation (1.77) applies.
However, if the motion has a component at right angles to the magnetic eld,
the corresponding conductivity has two parts:
1
Ne
2e
2i
Ni

e2
mee ( 2e 2e ) mii ( 2i 2i )
(1.78)
2
Ne
e e
N
ii
i
e2.
mee ( 2e 2e ) mii ( 2i 2i )
(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 ux density). 1 is the Pedersen
conductivity, which gives the current in the same direction as the applied electric
eld, whereas 2 is the Hall conductivity giving the current at right angles to it it
being understood that the electric eld and the currents are all in the plane normal
to the magnetic eld. Figure 1.19 may clarify the geometry.
50
Figure 1.20. Conductivity proles calculated for middle latitude at noon. (S.-I. Akasofu
and S. Chapman (after K. Maeda and H. Matsumoto), SolarTerrestrial 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 dierence 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 eect of an
electric eld 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 eld) 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)
Height: 60 km
75 km
100 km
125 km
>150 km
e >> e
e e
i >> i
e << e
i i
i << 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 eld
(bottom) at selected key altitudes. The current is proportional to the vector dierence
between the ion and electron velocities. (After H. Rishbeth. J. Inst. Electronic Radio
Engineers 58, 207, 1988.)
the latter is not present, the ow of current is inhibited or modied by the electric
potentials created at the boundaries.
The geomagnetic equator is one interesting case. Here the magnetic eld runs
horizontally and therefore the current which would otherwise ow normal to the
eld is inhibited in the vertical direction. Charges are created at the upper and
lower boundaries, and the resulting electric eld acts to increase the current in the
horizontal plane. It can be shown that, in this special situation, the conductivity
across the magnetic eld 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 ow readily along the geomagnetic eld direction. The existence of eld-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 owing purely horizontally. It was not until
52
eld-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 stratied 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 classication 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 solarterrestrial
system: for example, perturbations in the solar wind can produce magnetospheric
Table 1.2. A classication scheme for AGWs
Nomenclature
Horizontal trace
velocity (m s1)
Period (min)
Wavelength
Large-scale
Medium-scale
Small-scale
2501000
90 to 250
300
70
1570
25
1000 km
Several hundred kilometers
1.6 Acoustic-gravity waves
53
eects, which couple to the high-latitude ionosphere as particleprecipitation

events and electric-eld disturbances, which in turn generate medium- and largescale AGWs. We shall meet other examples of solargeophysical 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 eld.
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 stratied, 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 specic 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 signicant frequencies are the acoustic cut-o frequency,
a g/(2s)
(1.82)
54
and the buoyancy or BruntVisala 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
kz2 1
2a 2
2
kx2 1 b2 .
2
2
s

(1.84)
Putting 2 b2 gives
kx2 kz2 1
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 eect of compressibility, gives
kz2 kx2
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,
Figure 1.22. The acoustic, evanescent, and gravity regimes of acoustic-gravity waves. The
dashed lines show the eects of neglecting gravity and compressibility, repectively. At ionospheric levels, waves with periods longer than 1015 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 ow 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 ows upward (as it must) but the phase propagation is
downward. Furthermore, the amplitude of the air displacement increases with
altitude so that the energy ux 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 xed values
of the ascent angle of the energy (i.e. the angle between the group velocity and the
horizontal). The energy ow 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 dened as the ratio between the speed
of sound and the phase velocity of the wave. (Compare with Section 3.2.3.) Then
55
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 ow.
1
b
a
= 0
= 0
= 20
= 5
= 40
GRAVITY
WAVES
0.5
= 60
= 10
BUOYANCY
WAVES
ACOUSTIC
WAVES
= 20
= 40
= 80
0
0.2
0.5
10
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.)
2 1
2a
2
1
57
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 aect AGWs in real life,
are energy loss through the viscosity of the air, non-linear eects if the amplitude
becomes too large at the higher altitudes, reection and ducting due to the change
of atmospheric properties with altitude, the curvature of the Earths 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 eect
is strongly modied by the geomagnetic eld which permits ion motion along the
eld 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 eects 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 Moett (1997, 1999), Huang et al. (1998) and
Hall et al. (1999).
58
Figure 1.25. A train of gravity waves observed by ionosonde over Missouri, USA, in
December 1966, identied from the virtual heights of echoes at frequencies between 1.6 and
3.6 MHz. (T. M. Georges, Ionospheric Eects 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 SolarTerrestrial Environment. Cambridge University

Press, Cambridge.
Richmond, A. D. (1983) Thermospheric dynamics and electrodynamics.
SolarTerrestrial 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.
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)

SolarTerrestrial Physics, Oxford University Press, Oxford.
Kelley, M. (1989) The Earths 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 Moett, R. J. (1997) A study of atmospheric gravity waves and

travelling ionospheric disturbances at equatorial latitudes. Ann. Geophysicae 15, 1048.
Balthazor, R. L. and Moett, 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: 19821995. 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 waveTID relationship
with the aid of theoretical simulations. J. Geophys. Res. 101, 13 353.
59
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) SolarTerrestrial 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) SolarTerrestrial 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,
Springeld, 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.
Ratclie, 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.
atmosphere. In Space Physics (eds. D. P. Le Galley and A. Rosen). Wiley, New York.
Whitten, R. C. and Poppo, I. G. (1965) Physics of the Lower Ionosphere. PrenticeHall, Englewood Clis, New Jersey.
Whitten, R. C. and Poppo, 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 inuencing the high-latitude
ionosphere
2.1
Introduction
Whereas the mid-latitude ionosphere is dominated by solar radiation and the

chemistry of the upper atmosphere, modied by dynamic eects, the high-latitude
ionosphere is, in addition, strongly aected by the nature of the geophysical environment and by various processes occurring within it. In particular, the form of
the geomagnetic eld 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 aected by the dynamics of the magnetosphere and
is thus subject to electric elds and currents generated by motions at high levels.
At the highest latitudes the ionosphere is connected, via the eld-lines, to the outer
magnetosphere, giving it a ready response to variations in the ow 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 eld
To a rst approximation the geomagnetic eld at and close to the planets surface
can be represented as a dipole eld. The poles of the dipole are at geographic
61
62
Geophysical phenomena
Figure 2.1. Dipolar eld-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 ux 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 eld-lines, in a dipole eld.
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 eld-line in the plane normal to
the axis of the dipole. There is a dierent value of r0 for each line of Figure 2.1,
but, since the eld 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 ux density at the magnetic equator on the
Earths surface. In these terms, the eld-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 eld-line
intersects the Earths 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 sucently 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
eld from the magnetic potential expressed as a series of spherical harmonics
the ux density being the gradient of the magnetic potential. The dipole form corresponds to the rst term of the expansion.
The coecients are derived by tting the expression to measurements of the
magnetic elements on the global scale, using magnetometers both on the ground
and on satellites. Because the geomagnetic eld changes with time the secular
variation a fresh set of coecients, relating to a specic 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 eld 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 connes the geomagnetic eld on the sunward
side and forms the geomagnetic cavity.
2.2.2
The solar wind
The solar wind was rst 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 outow 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 uctuations 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 Earths 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
(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 eld, 19881990. By is
eastwest and Bz is northsouth. (F. J. Rich and M. Hairston. J. Geophys. Res. 99, 3827,
standards, the energy being more directed than random. It carries an energy ux
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
solarterrestrial relations and in the behavior of the high-latitude ionosphere.
The interaction depends on a weak magnetic eld, the interplanetary magnetic
eld (IMF), which is carried along by the plasma. This eld 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 eld 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 eld.
Although the solar wind ows out almost radially from the Sun, the solar rotation gives the magnetic eld a spiral form, as in Figure 2.3. This is sometimes
known as the garden-hose eect 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 eld-lines run at about 45 to the radial direction: the radial and
the eastwest components of the IMF are therefore about equal in magnitude.
One of the most remarkable of the early results, and a fact of great signicance,
is that distinct sectors may be recognized within the solar wind, the eld 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 rst sight the form of the IMF appears anomalous. Although there may be
a northsouth 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 eld which we expect
to be essentially dipolar? The problem is that the early observations were conned
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 eectively divides the outward eld (above the plane) from the
inward eld (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
Figure 2.3. (a) The form of the interplanetary magnetic eld (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.)
67
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 Suns rotation axis. (E. J. Smith et al., J. Geophys. Res., 83, 717,
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 eld
has a single polarity all inward or all outward. This is an open magnetic eld
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 eld 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 eld 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 specic 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 are driving an interplanetary shock. The
ejected plasma contains an ordered magnetic eld, but between the shock and the ejecta the eld is turbulent. (Reprinted from L. F. Burlaga, Adv.
Space Res. 2, 51, copyright 1982, with permission from Elsevier Sceince.)
69
solar are, though in some instances a are 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 are 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 eld 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 eld is illustrated in Figure 2.6. Because it has very high electrical conductivity, the solar wind is not able to penetrate the geomagnetic eld but is swept
around it. Pressure is exerted against the magnetic eld, which is distorted and
conned 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 eld.)
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 rst approximation the form of the boundary between the geomagnetic eld
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 eld 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 ux 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 eect.
A simple calculation along these lines readily gives a realistic distance for the
position of the boundary (approximately 10RE) along the EarthSun line, and
allows one to estimate how it varies if the solar wind changes. We assume that
0, and the magnetic ux density B varies as (distance)3. Then the distance to
the magnetopause is
lm
B2E
40Nmv2
1/6
(2.6)
70
(a)
(b)
Figure 2.6. Two sketches of the geomagnetic cavity in northsouth cross-section.

(a) Showing the external ow of solar-wind plasma and the principal features of the distorted geomagnetic eld. Note the tapering of the inner eld-lines, suggesting that there is a
neutral line further down the tail. (Adapted from V. M. Vasyliunas, in SolarTerrestrial
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 Earths Magnetosphere (ed. R.
G. Johnson), Terra Scientic Publishing Co., Tokyo, 1983, p. 99. After Bahnsen 1978).)
where BE is the geomagnetic ux density at the Earths 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
elds, which aects the location of the boundary. Spacecraft nd 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 aected 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 Earths dipole, ME. The elds due
to these two dipoles are added and the distorted eld-lines associated with ME
the two elds do not interconnect are taken to represent the geomagnetic eld
within the magnetopause. Those associated with the image have no physical signicance. 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 eld 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 eld 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 ux density falls to zero as they reach it; here, neutral points are
formed.
The magnetopause has a nite 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
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 Alfvn speed, given by
vA
B
,
( 0 ) 1/2
(2.7)
where B is the magnetic ux density and the particle density (in kg m3), is about
50 km s1. For a solar wind speed of 400 km s1, therefore, the Alfvn 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 veried 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 510 times hotter.
2.2.5
The polar cusps
The simple models of the magnetosphere predict two neutral points on the magnetopause where the total eld is zero. These points connect along eld-lines to
places on the Earths surface near 78 magnetic latitude. These are in fact the
only points on the Earths surface which connect directly to the magnetopause,
and all the eld 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 eld-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 eld lled with magnetosheath plasma, and they have signicant eects on the high-latitude ionosphere.
The ionospheric eects 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 eld
beyond about 10RE on the night side of the Earth tends to run in the SunEarth
direction, and there is a central plane within which the eld reverses direction.
This is the neutral sheet. The eld 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 signicance for the high-latitude ionosphere is that it
maps into the polar caps at its earthward end, and thus the polar ionosphere can
be aected 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 ux density is about 20 (20 nT) in the tail lobes,
but the eld 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 eld 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 eld-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 eld 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 ows 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
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 eld () 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:
specically 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 ows up the
eld-lines to populate the protonosphere as far as the equatorial plane, provided
Figure 2.8. Electron density in the equatorial plane determined from whistlers. (J. A.
Ratclie (after D. L. Carpenter), An Introduction to the Ionosphere and Magnetosphere.
Cambridge University Press, 1972.)
that the eld-lines are closed. Some of this plasma ows 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 eld 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
eld-line the whistler has traveled along, and the time taken can be interpreted to
give the minimum electron density encountered along that eld-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 eld, it is found to correspond approximately to the
ionospheric main trough which eectively marks the poleward extent of the midlatitude ionosphere (Section 5.4). The plasmasphere thus occupies a doughnutshaped region of the inner magnetosphere where the eld-lines are not too
75
76
Figure 2.9. Plasma ow 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 (K p) by the empirical relation
Lpp 5.60.46K p.
(2.8)
For this purpose K p is derived over the previous 24 h, except that, for the hours
0609, 0912, and 1215, respectively, the Kp values for the preceeding one, two, and
three periods (of 3 h) are omitted. Carpenter and Anderson also give formulae for
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
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 dierent population of particles. The electron and
ion densities are each only about 0.5 cm3. Particle energies are generally 102104
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 eld reverses. To a rst
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 eld outside the plasma sheet. As indicated in Figure
2.6, the plasma sheet follows the magnetic eld 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 identiable, 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 Earths 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 eld 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 conguration 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 eld 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
rst 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
79
Figure 2.11. Van Allens

rst 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 scientic 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-simplication 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 eld, 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 ux enclosed by the drift orbit is constant.
The rst holds if the magnetic eld 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 rst invariant. Consider a
charged particle gyrating in the geomagnetic eld but also with a component of
motion along the eld. 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 ux falls with energy, these distributions are dominated by the particles just above the threshold
stated. The uxes are omnidirectional and in units of cm2 s1. They are diurnal averages at sunspot maximum. The inner zone is seen in the uxes 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
83
longitudinal drift
B
gyration
Earth
e
p
M
bouncing
mirror
points
Figure 2.13. The motions of a particle trapped in the geomagnetic eld.
moves into a region of stronger eld. A consequence of the rst invariant is that the
component of kinetic energy perpendicular to the eld is proportional to the magnetic ux density. Hence that component of the particles energy increases, and the
parallel component decreases by the same amount. Eventually, provided that the
particle does not run into the atmosphere rst, all the energy is transverse, forward
motion stops, and, at that point, the mirror point, the particle is reected 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 particles 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 denes the loss cone, which is generally a small angle of only a few degrees at
the equator but increases progressively towards 90 as the Earths 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 eld. 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 eld-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.
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 Earths surface. (P. N. Collis, private
communication.)
The Van Allen particles do not have signicant eects 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 dierent regions of the magnetosphere and then deposit them in the ionosphere, are important at high latitudes.
2.3.5
The ring current
One signicant 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 10100
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.
Figure 2.15. Radial proles 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 5315 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 ow along the magnetic eld-lines between the magnetosphere
and the ionosphere. Such currents had rst 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 eldaligned 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
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 eld-aligned current is 106107 A.
The concept of the Birkeland current has profoundly aected 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 owing horizontally at some (unspecied) 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 eld-lines. (J. A. Ratclie, An Introduction to the Ionosphere and Magnetosphere,
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 specied but was thought to be eectively 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 ows 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 eld in the dynamo region (whereas the
positive ions move with the neutral air), the Spq current ow can be interpreted as
a motion of magnetic eld-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
eld-lines circulate over the poles from the day to the night sectors of the Earth
with a return ow 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
Earths rotation which carries the inner part of the magnetosphere with it. Section
2.4.4 indicates how the combined eect of two circulation patterns may be handled.
87
88
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 Earths 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 sucient 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 eld
and the geomagnetic eld. The various eld congurations that arise when a
dipole is placed in a uniform magnetic eld are easily illustrated in simple laboratory experiments using a bar magnet situated in an external eld. When the elds
are parallel there are neutral points on the equator and connections between the
two elds. When they are antiparallel, the neutral points are over the poles and
there is no interconnection. Figure 2.19 depicts a distorted dipole eld representing the geomagnetic eld 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 rst 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 northsouth
component as well and it is this, when it is directed southward, which connects
with the geomagnetic eld.
The IMF is frozen into the solar wind and is therefore carried along with it.
When geomagnetic eld-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 eld-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 simplied 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
Figure 2.19. The interaction of terrestrial and interplanetary magnetic elds seen in polar
section: (a) northward IMF; (b) southward IMF; and (c) circulation due to the ow 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 eld-line. B: Interplanetary eld-line connecting to, or disconnecting
from, a geomagnetic eld-line. C: Open geomagnetic eld-line. D: Closed geomagnetic
eld-line. N: Neutral point. 07: Successive positions of a selected interplanetary eld-line.
is northward, and how the eastwest component aects 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
90
Figure 2.20. An X-type neutral line in the magnetotail. Plasma ows 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 ux 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 signicance, 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 eld-lines are broken and then reconnected in a dierent conguration. 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
conguration cannot be static because the tension in the eld-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 eld-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 eld
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
ares. In the magnetosphere the process is thought to be that of fast reconnection,
rst proposed by Petschek (1964) more than 30 years ago. This mechanism
invokes an Alfvn wave, which allows reconnection to proceed more rapidly than
would diusion only. The velocity of reconnected eld-lines towards the Earth is
estimated as about 100 km s1, and the drift towards the neutral sheet as about 10
91
km s1. Particles on the eld-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 eld-lines on the day
side which drives the circulation. Though various ideas have been suggested, the
details of this process have not been nally decided. Obviously, a geomagnetic ux
tube has to break and connect with an IMF tube, and this is the event which has
been identied from the magnetic signature recorded by a nearby spacecraft, as a
ux-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 ((0.51)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
eastwest component. Quasi-steady connection is also a possibility.
2.4.3
Magnetospheric electric elds
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 eld (the IMF) and an electric potential is developed by
dynamo action. The total potential dierence 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 ux density normal to the boundary. Its value is estimated as about
60 kV, equivalent to an electric eld of about 0.3 mV m1. The electric eld is
directed from the dawn to the dusk side of the magnetosphere. The same potential dierence appears across the open eld region of the high-latitude ionosphere,
the eld again being directed from dawn to dusk.
The motion of magnetoplasma within the magnetosphere can now be regarded
as the eect of this electric eld on the geomagnetic eld as in an electric motor,
according to
vE
B/|B|2,
(2.11)
where v is the velocity, E the electric eld, and B the magnetic eld. The magnitude is simply
vE/B.
(2.12)
92
The potential distribution across the magnetotail maps along the eld-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 dierence across the polar
cap is 60 kV, the eld-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 elds 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 eld-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 eld, and the plasmapause
marks the boundary between the inner and outer regions.
If we imagine that the plasmasphere is observed by a person xed 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 eld of magnitude
Er BLRE,
(2.13)
where B is the geomagnetic ux density, L is the observers geocentric distance in

Earth-radii, RE is the radius of the Earth and is the Earths angular velocity. The
plasmapause occurs approximately where the cross-tail and co-rotation elds are
equal:
ET
BE
LRE,
L3
(2.14)
where BE is the geomagnetic ux density at the surface of the Earth at the equator,
and we have used the radial variation of ux density in a dipole eld, 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 eld is 1 mV m1 we expect to nd the plasmapause at about 4RE.

A computation of plasma convection about the Earth was shown in Figure 2.9.
In general these ow 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 elds are in opposite directions on the evening side.
2.5 Magnetic storms
We now see that the principal dynamics of the plasmasphere are (1) lling 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 aected 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 satised
closer to the Earth and the plasmasphere must then be smaller. It is thought that
the change of circulation peels o 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
elds return to their previous state, but now the outer tubes of magnetic ux are
devoid of plasma. These gradually rell from the ionosphere over a period of
several days. The rate of lling is determined by the rate of diusion 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 ux
tube to be lled. Since the latter varies as L4 it takes much longer to rell 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 suers 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 eect 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 diculty 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 supercial 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, aecting the magnetosphere. The magnetospheric consequences then aect the upper atmosphere, and, in some cases, there
might even be contributions ascending from the troposphere or stratosphere.
93
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 eld 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 Ferraros 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 ux (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 ux 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 dierent guise at high latitude. By contrast with
records from low latitudes, where the eects 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, owing
not in the magnetosphere but in the auroral E region. To explain the sign of the
bay, the current ow must be eastward before midnight and westward afterwards:
i.e. converging on the midnight meridian. Obviously there must also be return currents for continuity. Chapmans original interpretation assumed that the currents
only owed 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
96
Figure 2.23. Chapmans

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,
Publishers.)
magnetic variations during the rst 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 dierent 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)(8.440.74)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 eld.
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 dierent proceedure. Magnetograms from observatories at several dierent
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
dierence 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
Figure 2.24. A Bartels diagram of Kp. (SolarGeophysical 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 aected
by the ring current. The advantage of AE, being their dierence, 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
rst 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 eort, 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.)
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 rst, 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 eects so dramatic that they attract both scientic and popular attention. Though such storms are rare, they serve to illustrate
how great the eects 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 eects. It was
related to the largest group of sunspots to be seen on the solar disk since 1982, and
the eects 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 eects 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 owing south of
Fredericksburg, Virginia, whose geomagnetic latitude is 49 N.
It was reported from Alaska that the ight of homing pigeons was aected.
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-eld 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
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 eects was
presented by Yeh et al. (1992).
Satellite drag
The main eect 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 identied 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 eect on
electric-power distribution. It is known that uctuations in the geomagnetic eld
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 Qubec
suered a power black-out lasting 9 h on 13 March. Users in the north-eastern
USA were also aected. 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 scientic explanation, and because they may have
serious eects 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 signicant, giving the acoustic-gravity wave (Section 1.6). In a magnetic eld another
restoring force comes into play, and that is the magnetic pressure across the eld
and the magnetic tension along the eld-lines.
The situation in the magnetosphere is that the ionization may not cross eldlines, and therefore in transverse motions they must move together. Ordinary
sound waves are allowed along the eld-lines because the gas displacement is longitudinal, but in waves transverse to the eld 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 Alfvn wave, which propagates along the
magnetic eld but whose displacement is transverse. The Alfvn 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 ux density, 0 the permeability of free space, and the
density of the plasma in kg m3.
The Alfvn wave and the sonic wave are independent when they are traveling
along the eld direction, but at other angles they interact to give mixed magnetosonic waves. There are two such waves in general, except that perpendicular to the
eld 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 suciently, small uctuations of
the geomagnetic eld 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 eld. Their connection with ionospheric radio is
somewhat indirect, but an introduction is in order because they comprise a signicant 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 classied according to period and duration, as in Table
2.3. The impulsive variety (Pi) occurs mainly in the evening, whereas the more
103
104
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 eld, is probably
involved. Pc25 are explained as various modes of oscillation within the magnetosphere, some propagating across and some along the eld lines. The period of
Pc3 and 4 may be interpreted in terms of Alfvn waves, whose speed depends on
the plasma density and the magnetic eld 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 waveparticle 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 FarleyBuneman,
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)
0.25
510
1045
45150
150600
Irregular
Type
Period (s)
Pi1
Pi2
140
40150
2.6 Ionization by energetic particles
streams of ions and electrons dier 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 KelvinHelmholz 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 ow 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 KelvinHelmholtz waves and vortices may be produced there.
A slightly less supercial 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 dierent 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
signicant sources at high latitude are electrons associated with the aurora, and
protons (plus some -particles) emitted from the Sun during some solar ares.
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 ionelectron 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 particles
105
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
velocity. To quantify this, laboratory measurements are again brought to bear in

the form of an eciency, 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 eciency 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 eld. 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 eciency peaks at a smaller value of s/sp such as 0.1 or 0.2. These factors clearly inuence 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 eect 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 3040 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 ux 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 dicult. The usual practice is to draw on a set of computations giving the Bremsstrahlung due to single
electrons of specied 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
Signicant ionization may also be caused by energetic protons, especially at high

latitudes during polar-cap events, which are due to uxes of protons released from
the Sun at the time of a solar are. A lesser ux of -particles will generally arrive
simultaneously. These particles, which are signicantly 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 10200 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
108
140
130
120
110
100
Altitude (km)
90
80
70
60
50
40
30
20
10
12
11
10
Log of Energy Deposition Rate (in keV
cm 3
s1)
Figure 2.27. A comparison of direct and X-ray energy-deposition rates from an incoming
electron ux 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 ionelectron 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 proles 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 eect would be calculated by appropriate
summing over these curves at each height.
There is a similar procedure for dealing with the ionization by -particles.
Figure 2.28. Production rates due to incident monoenergetic protons. The initial energies
are given in MeV, and in each case the ux 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
2.2
The magnetosphere
Burlaga, L. F. (1982) Magnetic elds 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 inuences 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 Earths Magnetosphere (ed. R. G. Johnson), p.
99. Terra Scientic 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.
SolarTerrestrial Physics (eds. R. L. Carovillano and J. M. Forbes), p. 243. Reidel,
Dordrecht.
109
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 inuence 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 eld-aligned currents associated with substorms. J. Geophys. Res. 83, 599.
Potemra, T. A. (1983) Magnetospheric currents. Johns Hopkins APL Tech. Digest 4, 276.
Ratclie, 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.
Walt, M. (1994) Introduction to Geomagnetically Trapped Radiation. Cambridge
Williams, D. J. (1985) Dynamics of the Earths 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 Earths 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 ow. 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.
Petschek, H. E. (1964) Magnetic eld annihilation. The Physics of Solar Flares (ed. W.
N. Hess), Report SP-50, p. 425. NASA, Washington DC.
Ratclie, J. A. (1972) An Introduction to the Ionosphere and Magnetosphere.
Cambridge University Press, Cambridge.
Russell, C. T. (1972) The conguration 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.
Press, Cambridge.
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. Gttingen 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
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) SolarTerrestrial 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,
Springeld, Virginia.
Kamide, Y. (1988) Electrodynamic Processes in the Earths Ionosphere and
Magnetosphere. Kyoto Sangyo University Press, Kyoto.
111
112
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 Diusion 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 Earths Magnetosphere.
Terra Scientic Publishing Co., Tokyo.
Kamide, Y. and Slavin, J. A. (eds.) (1986) Solar WindMagnetosphere Coupling. Terra
Scientic Publishing Co., Tokyo.
King, J. W. and Newman, W. S. (eds.) (1967) SolarTerrestrial 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 Earths Magnetic Field.
Reidel, Dordrecht.
McCormac, B. M. (ed.) (1968) Earths Particles and Fields. Reinhold, New York.
McCormac, B. M. (ed.) (1970) Particles and Fields in the Magnetosphere. Reidel,
Dordrecht.
McCormac, B. M. (ed.) (1972) Earths 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 denition, 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 ux, given by

(3.2)
SE
H,
Table 3.1. The radio spectrum (as dened by the International

Telecommunications Union (ITU)), primary modes of propagation, and eects of
the terrestrial ionosphere
Principal
modes of propagation
ITU designation
Frequency range
Principal uses
Extra-lowfrequency (ELF)
30300 Hz
Groundwave and
Earthionosphere
waveguide mode
Submarine
communication
Very-lowfrequency (VLF)
330 kHz
Same as above
Navigation,
standard frequency
and time
dissemination
Low-frequency
(LF)
30300 kHz
Same as above
Navigation
LORAN-Ca
Mediumfrequency (MF)
3003000 kHz
Primarily groundwave,
but skywaveb at night
AM broadcasting,
maritime,
aeronautical
communication
High-frequency
(HF)
330 MHz
Primarily skywave,
some groundwave
Shortwave
broadcasting,
amateur, xed
services
Very-highfrequency (VHF)
30300 MHz
Primarily LOS, some

skywave at lower VHF
FM broadcasting,
television,
aeronautical
communication
Ultra-high-(UHF)
frequency
3003000 MHz
Primarily LOS, some

refraction and
scattering by the
ionosphere
Television, radar,
navigationc,
aeronautical
communication
Super-high-(SHF)
frequency
330 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 EarthionosphereEarth-reected 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 eld
strengths,
SE 2/(120).
(3.4)
Therefore,
E 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 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 ux at a point on the axis divided by the ux 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 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 ecient dish,
but may be an eective area otherwise. An isotropic radiator (which is hypothetical in any case for an EM wave) has unity gain and eective 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)
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 eects see Appendix A7 of Hunsucker (1991), Freeman
(1997), or Wol (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 reected 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 YagiUda 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
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 eective 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)),
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 beamlling 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. 3839).
3.2.3
The signicance of the refractive index

A simple propagating wave
If a radio source generates an electric eld EE0 cos(t), which propagates at

speed v in the z direction, the eld at a distance z from the source is
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 denition. 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
eld at (t, z) and E0 is the amplitude of the electric eld. 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,
(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 coecient, and the amplitude
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
j sin ). 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 coecient
! /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. 2531) that
( j / ) .
(3.28)
Squaring, and equating real and imaginary parts, gives

2
2
1 2 2

1/2
1
1/2
(3.29)
121
and

2
2
1 2 2

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 reected 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 reection, refraction, dispersion, diraction, 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.
Reection 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 eects are discussed in the context of ionospheric reection in Section 3.4.3, and of the partial
reection technique in Section 4.2.4.
Dispersion, the variation of velocity with frequency, has consequences for the
transmission of information (Section 3.4.1).
Diraction 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
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 eld. 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
eects upon radio propagation, there are some tropospheric eects peculiar to high
latitudes that aect radio propagation in the line-of-sight (LOS) and earth-to-satellite modes. For that reason, we will briey discuss some of the fundamentals of these
modes. We will exclude the troposcatter propagation mode in which forward scatter
in the troposphere (38 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
aected by the refractive index, n which is a function of atmospheric pressure, temperature, and humidity, and, near the Earths surface at VHF/UHF, n is approximately 1.0003. It is convenient to dene 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 dierence 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 eective 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,

N Ns exp(ch)
Ns
N
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
r0
K 1 cNs
106
ns
1
(3.36)
where ns 1Ns
106, r0 is the Earth-radius6373.02 km, Ns 289, and
c0.136, so K1.333 241 0, or 4/3.
The basic exponential reference atmosphere is dened 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 rst 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 dierence between
Ns and N 1 km above the Earths 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 termwet 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
124
Table 3.4. Constants for the standard reference atmosphere

Ns
hs (ft)
a (miles)
N
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 eective Earth-radius and is equal to the product aK.
a ahs, where hs is the height of the Earths surface above sea level.
a3960 miles.
c
N1
1
ln
.
8 hs
105
refractivity as a function of its value at the surface, Ns, has been dened.. Near the
ground the following empirical relationship between Ns and the dierence in
refractivity, N, between Ns and N at 1 km above the Earths 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 dierent
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 eects will be described in Chapters 8 and 9.
3.3.2
Terrain effects
The most obvious feature of the Earth aecting terrestrial radiowave propagation
is its curvature. The troposphere of the Earth refracts radiowaves on LOS paths
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 modied 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 aect the propagation of radio waves especially if they block o low
takeo angles for HF paths or if the ground-reection 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 reecting plane, when in reality the conductivity and permittivity of the Earths 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 Earths
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.
Figure 3.4. Earth-curvature-correction curves for D1 from 0.5 to 7 miles (from Freeman,
1997).
pp. 498503; Hall and Barclay, 1989, pp. 3842 and Hunsucker, 1991, pp.
258266). Several computer programs which treat terrain eects 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 gure and noise temperature of receivers, and
cosmic, atmospheric, and manmade noise in Collin (1985), Kraus (1988),
127
128
Spaulding and Washburn (1985), and CCIR Report 322; and a shorter description in Hunsucker (1991, pp. 1520 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 lter, 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
0.010.02
0.002
0.0050.02
0.1
0.001
0.001
3.05.0
0.001
0.000025
0.0001
0.00050.001
Tundra underlain by permafrost surfacea
103 to 102
Permitivity ( )
(relative dielectric)
constant
10.0
14.030.0
10.015.0
30.0
30.0
15.0
80.081.0
80.081.0
14.0
13.0
11.0
2334 for silts
12 for dry sand
570
Note:
Acquired in 1988/1989 in Central Alaska from 230 MHz by G. Hagn of SRI
International.
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 diraction/terrain predictions
Name
Description
Source
Reference
EREPS
Engineers
Refractive Eects
Prediction System
http://trout.nosc.
mil/NraDMosaic
Home.html
Patterson
(1994), Proc. of
the BLOS
Conference
IFDG/GTD
*Finite Dierence.../
Generalized Theory
of Diraction
GELTI/ATLM
GTD Estimated
Loss due to Terrain
Interaction/
Automated Terrain
Linearization Model
HARPO
Hamiltonian
equations in
spherical
coordinates,
modied by using
Gaussian beams
EFEPE/SSP
IRT
Institut fr
Rundfunktechnik
propagation model
for digital broadcast
systems in urban
areas
Dr R. Grokopf
Institut fr
Rundfunktechnik
Mnchen
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)
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 modications to the development of the model
and using correction factors for certain locations. Figures 3.113.15 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
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 nalizing 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 gure from CCIR Report 322 (from Sailors, 1993).
Figure 3.13. Radio-noise-recording locations (from Sailors, 1993).
Figure 3.14. Determination of the 1-MHz Fam value for Moscow for June, July and August;
16002000 UT. (from Spaulding and Washburn, 1985).
137
Figure 3.15. In (a) and (b) are shown examples of Spaulding and Washburns corrections to the CCIR Report
322 (from Spaulding and Washburn, 1985, p. 18).
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 dened 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 ux
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 ux 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 ux 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
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-eld polarization in

the plane of the wavefront. Ox and Oy are the
principal directions and
the projection of the
imposed magnetic eld 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 Ratclie, 1959).
YL Y cos
(3.48a)
YT Ysin ,
(3.48b)
and
being the angle between the direction of propagation and the magnetic eld.
Note that the refractive index is complex, with real and imaginary parts:
n j .
Polarization
In order to calculate the eects of this anisotropic medium on the polarization of
a radio wave traversing the region, it is convenient to dene 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 Ratclie, 1959)
R
j
Y 2T
Y 4T
"
Y 2L
YL 2(1 X jZ)
4(1 X jZ) 2

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
It is virtually impossible for an ordinary mortal to make much sense of

Equations (3.47) and (3.50) in their full glory see Ratclie (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 eld

If there are no collisions and the magnetic eld 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
1
Ne2
2
0me
1/2
(3.52)
The group velocity, using Equation 3.21, is
Ne2
uc/ng c 1
2
0me
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 eld

If the magnetic eld is now included and the propagation is almost directly along
the magnetic vector so that YT may be neglected, then
n2 1X/(1YL )
n2 1 N2 /[( L )
(3.54)
and Rj. There are now two waves, circularly polarized in opposite directions,
having dierent 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 signicant (but in the absence of a magnetic eld), then
n2 1X/(1jZ)
n2 1 N2 /[( j)].
(3.55)
143
Taking the imaginary part ( ) and applying Equation (3.25) gives the absorption
coecient
!
1 XZ
c 2 1 Z 2
!
e 2 1 N
.
2 0mec 2 2
(3.56)
The refractive index (n) is modied 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 Ratclie (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 signicant 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 reection 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
144
N
e2
.
2 0mc ( L ) 2 2
(3.61)
The absorption coecient is therefore smaller for the ordinary than it is for the
extraordinary wave. For a given value of the electron density, the absorption
coecient 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
Reection of radio waves from an ionospheric layer

Reection 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 eld, u is given by Equation (3.53) and u decreases as the electron density
increases with altitude. Provided that the layer is suciently intense, a level where
the group velocity is zero (and the phase velocity innite) will eventually be
reached, and here the energy is reected. 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 reection point and back is
t
2
c
dz
n
(3.65)
and the virtual height is

ct
h
2
[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-
145
quency that may be reected 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 Abels equation (3.66) (Appleton, 1930) is

given by Kelso (1964). In the general case (including the geomagnetic eld),
several numerical techniques have been employed successfully to invert the ionogram trace of the ordinary wave to give an equivalent monotonic electron-density
prole (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 eld has to be taken into
account, there are two reection conditions. The extraordinary wave is reected
where
fN2 f( ffB )
(3.68)
and the ordinary wave where

fN f.
(3.69)
The rst reection occurs according to the QL approximation, whereas the second
relates to the quasi-transverse (QT) approximation. If fB fN, the dierence
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,
Snells 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
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 reection condition (magnetic eld 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 reection 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 reected 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 Tuves 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
dx
TER sin 0
(3.74)
D
c sin a0
t(TEER)/c.
(3.75)
147
Figure 3.18. Logarithmic ionospheric-transmission curves for a curved-Earth ionosphere

(after Smith, 1939).
Martyns theorem states that, if fv and fob are the vertical and oblique frequencies
respectively, reected from the same true height (h), then the virtual height at
which the frequency fob is reected equals the height of the equivalent triangular
path for the frequency fv. Referring to Figure 3.17 and dening the equivalent path
at oblique incidence for frequency fob as
Pob 2TE,
(3.76)
we obtain
Pv cos 0 Pob 2DE
(3.77)
Martyns 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 suciently 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 reected but continues out into space. Similarly, signals from beyond the
148
ionosphere may be received at the ground if their frequencies are suciently high.
However, these signals are not necessarily unaected by the ionosphere: there can
be signicant and measurable eects on their phase, their polarization and their
intensity. In each case the eect becomes weaker with increasing frequency, and
in practice they are signicant 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 eects are cumulative and the total depends on an integral along the
propagation path.
Phase effects
In the Appleton equation for the refractive index, let X 1 (radio frequency large
relative to the plasma frequency), YL YT 0 (geomagnetic eld neglected), and
Z0 (collisions neglected). Then the second term of Equation (3.47) is much less
than unity, and we can write
n1 X/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/ (2fn dl /c) radians. Over a path l the advance of phase
is therefore
2 f l 2
40.30
2 f
ndl

Ndl.
c
c
cf
(3.81)
The rst 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
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)
149
Several signicant applications follow.

(a)
Since the phase advance depends upon the radio frequency, the electron
content can be determined by comparing the eects 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 signicant
gures, therefore with an inaccuracy of 0.12%. To four gures, 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 ve gures this constant is 40.302).
The Faraday effect

When the geomagnetic eld is taken into account and propagation is almost along
the eld direction, there are two characteristic waves that travel at dierent 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)
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
eld (from Hargreaves,
1992).
and
E 2f [tnE l/c],
(3.87)
(f /c)(nO nE )l.
(3.88)
giving
At a suciently high frequency (e.g. 50 MHz) the gyrofrequency fB f, and
then the ordinary and extraordinary refractive indices dier 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 eld strength, we obtain

8.448
107
f2

2.365
104
f2
f N dl
L
B N dl,
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 eld. (In fact the
QL approximation has wide application in the Faraday eect, being valid to a few
degrees of normal to the eld). As seen by an observer at the ground looking up,
the polarization rotates anticlockwise in the northern hemisphere and clockwise
151
in the southern hemisphere, irrespective of the direction in which the wave is

traveling. A recent extensive discussion of the Faraday eect is given by Yeh et al.
(1999).
Absorption
Equation (3.61) gives the absorption coecient ! 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), dened by the ratio between initial
(P1 ) and nal (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
( ) dl
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 aected 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) simplies to
A (dB)
1.168
1018
[ f (MHz)]2
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.
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 diraction. The phase shifts are a direct result of spatial irregularity in the
medium and specically 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 diraction theory this principle is applied to determine
how the amplitude and phase of a received signal are aected by passage through
a region of irregularities. Diraction theory applies to small irregularities, the
criterion for which is that there are at least several of them within the distance of
the rst 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 innitely 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 innity), but the emerging wavefront is irregular. To obtain the eld 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 eld at OO will also be irregular, and in general both the phase and the amplitude are aected.
Since the waveeld at the observing plane is made up from contributions from
points all along the diracting screen, it is clear that there is not necessarily a oneto-one relationship between the irregularities in the ionosphere and the waveeld
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 modied by the distance traveled, where it may be transformed back to a
153
E
a
Incident wave
Screen
E'
Emerging wave
O
P
Observing
O' plane
Phase
Amplitude
Figure 3.20. Diraction of a plane wave incident on a thin phase-changing screen.
waveeld 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 dicult to measure the phase. However, it has been
shown that, if the observations are made suciently 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 suce 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 dierences 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 /(2 sin2 0)],
(3.98)
154
(a) Correlation function

1
e 1/2
d0
d
(b) Angular power spectrum
1
e 1/2
sin 0
sin
Figure 3.21. (a) Correlation function and (b) angular power spectrum for a random
diraction 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 signicant because the size of
the Fresnel zones depends upon the distance as well as the wavelength. Recall that,
by denition, the rst Fresnel zone extends to the point where the distance to the
observer exceeds the minimum distance by /2, the resulting phase dierence
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 unaected 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 aect its amplitude because of the extra /4 traveled. The resulting signal might now look like Figure 3.22(b), with both phase and
amplitude uctuations involved.
Since contributions may aect the amplitude only if they fall within the angular
spectrum, it follows that
(D/2)1/2 d0
(3.100)
155
(a) Signal at EE

Total
A
(b) Signal at OO
Total
0 0
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 innity. This says that the signal received
from a phase screen will contain both amplitude and phase perturbations if the
observer is suciently far from the screen for the rst Fresnel zone to contain
several irregularities of typical size. At innity, 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 uctuation, but the amplitude uctuation 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 rst Fresnel zone is D 1.5 km. Many of the irregularities will be larger
than that and therefore the amplitude uctuations 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 xed 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 eect can be investigated directly by comparing the spectra of phase
and amplitude uctuations. 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 uctuation 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
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 / 2D, where u

is the velocity (it being assumed that all the irregularities move together) and D
is the radius of the rst Fresnel zone. Since is known and D may be assumed (to
be approximately 350 km), u can be determined by this means. This eect 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 identied.
The above results are altered if the source is not at innity (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 Ratclie (1956), and later developments by Yeh
and Liu (1982).
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, aA A, and pP P,
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 uctuate
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
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).)
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 reection from a sharp

boundary and full-wave solutions
Reection 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
Figure 3.26. Empirical amplitude distributions for a range of S4 values. (After R. K. Crane,
Technical Note 197426, Lincoln Laboratory (1974).)
this is plainly not so, and then a dierent sort of treatment is required.
If the medium changes signicantly within a wavelength then we may use the
physics of reection at a sharp boundary, as at a partially reecting mirror. If a
wave is normally incident at a sharp boundary, the coecients of reection 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 reected, the wave being incident from below,
Et EI Er
(3.107)
161
Figure 3.27. Empirical phase distributions for a range of S4 values. (After R. K. Crane,
Technical Note 197426, Lincoln Laboratory (1974).)
and
Ht Hi Hr,
(3.108)
the negative sign arising because the reected wave propagates downward. In a
non-magnetic medium,
H/En /( 00)1/2
(3.109)
and, by substitution, the reection coecient is given by

Er /EI (n2 n1)/(n2 n1).
The fraction of power reected is (Er /Ei )2 .
(3.110)
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 ux
( E and H ) are also continuous across the interface. One familiar result that
follows is Snells 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 eld) be perpendicular to the plane of incidence. Then application of the continuity conditions gives
sin(i t )/sin(i t ),
(3.112)
or,

(n2 /n1 ) 2 sin2 i cos i

.
(n2 /n1 ) 2 sin2 i cos i
(3.113)
This is the rst Fresnel equation for reection.

If the plane of polarization lies in the plane of incidence, the reection coecient is given by
tan(i t )/tan(i t )
(n2 /n1 ) 2cos i (n2/n1 ) 2 sin2 i

(n2 /n1 ) 2cos i (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 reection coecient goes to zero if the E vector is in the plane of incidence in practice, the wave
is vertically polarized. The reected wave is reversed in phase as the Brewster angle
is crossed. There is no such eect if the wave is horizontally polarized.
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 reection.
These remarks apply to reection at the interface between dielectrics, n1 and n2
being both real. If the reector is a partial conductor, the Fresnel equations still
apply but the refractive indices are now complex. In the general case reection
involves a change of phase as well as of amplitude. Provided that the conditions of
a sharp boundary are satised 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 Maxwells 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 reected 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 reection at a sharp boundary. At oblique incidence the loss on reection is relatively
small, and in consequence these signals may propagate over great distances with
an attenuation amounting to only 23 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 eld, 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
Ionosphere
p
2-ho
p
ho
1-
ground
wave
Ground
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 eective reection height (about 70 km by day) and
the ionospheric conductivity. Reection coecients are typically 0.20.5 There are
actually four reection coecients because the presence of the geomagnetic eld
causes changes of polarization on reection as well as of amplitude and phase.
Putting typical values into the criterion of Equation (3.31) conrms 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/ 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 Earths surface, there is sucient 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 rst 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 reection coecient of 1 and the ground has a reection
coecient of 1. As in Figure 3.30, the images are located at z 2h, 4h, . . . ,
165
Figure 3.30. Using the method of images to construct one of the pair of waves that will
interfere to produce the eld 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 eld in the

Earthionosphere waveguide for waves polarized with their electric
elds in the vertical
plane and their magnetic
elds transverse to the
plane of propagation
(TM02 mode) (from
Davies, 1990).
166
but now they alternate in sign, which is equivalent to a change in phase of , and
resonance occurs at
2hCn (n 12 ),
(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 coecient showing that, for 2h/n, g is imaginary and
hence the mode is evanescent. Thus there is a minimum cuto frequency, fn, below
which waves will not propagate, where fn nc/(2h).
The cuto frequency for the rst-order mode during daytime, when the height
of the ionospheric D region is low, 2 kHz. For the case of a conducting ionosphere, the cuto frequency is given by
fn (n 12 )c/(2h).
(3.121)
So we see that the change of reection coecient R from 1 to 1 changes the

cuto frequency (for n1) from about 2 Hz to 1 kHz. When they are being
compared with waveguide modes with perfectly conducting walls, the ideal
Earthionosphere modes should be denoted by n 12, rather than by n. A more
complete analysis of the VLF waveguide mode must include Earthionosphere
irregularities, changes in the height of the ionosphere, the eect of the geomagnetic eld and collisions of electrons. There is a voluminous literature on VLF
propagation (see Budden, 1985, references; and Davies, 1990, pp. 371379).
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. 249299) in the
D region for ELF, the reection height is 90 km.
As an actual example, the U. S. Navys Wisconsin Test Facility (WTF) radiates
frequencies in the 4050 and 7080 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).
167
Partial reections at MF and HF

Turbulence in the lower ionosphere, at heights up to about 100 km, produces
spatial irregularities on a scale suciently ne that partial reections may be
detected from them in the band 26 MHz (wavelengths 1.5 km to 500 m). In this
case the reections are very weak, a mere 103 to 105 of the amplitude of a total
reection, 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 reections may be exploited in a technique for
measuring the electron-density prole 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
eld and can be detected using low noise ampliers with short antennas. Since
about 1951 these signals have been studied scientically 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 scientic 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 Appletons equations for dispersion
and polarization and the QL approximation. Figure 3.32 is a simplied presentation of basic whistler signatures obtained near the source and near the conjugate
area of the source (i.e. the other end of the eld 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 ow 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
Figure 3.32. A sketch of the basic manifestation of a whistler and its initiating disturbances: (a) illustrating the dispersion; (b) the frequencytime 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 eld-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) simplies

to
vg 2c( f 1/2 fL1/2 )/fN.
(3.124)
The dispersion law for whistlers is

T
1
2c
f
s
f ds/[ f 1/2 ( fL f )3/2],
N L
(3.125)
which can be applied to determine Ndl along the eld 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 104109 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
Figure 3.34. Heightfrequency 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 heightfrequency domains of
typical ionospheric sounding systems.
The rst 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 eld-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
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 7590-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 10002000 km with system losses of 140210 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 satelliteEarth 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 ux (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 eld and the direction of the observer.
172
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
Ne,
(3.129)
where N is the electron density, and Buneman (1962) showed that the incoherent
scatter eective radar cross-section (e ) can be written as
e 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. 106111) and by Hunsucker (1991, pp. 5864). 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 uctuations in electron density having a
scale of D. The backscatter, then, is actually due to local uctuations 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 inuence
from ions.
In practice, this incoherent scatter is detected from the ionosphere principally
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 specic 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 inuence 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 ux may be measured.
In the lower ionosphere the motions of the ions (which in turn control those of
the electrons) are increasingly aected 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.
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-eld-strength values for mid-latitude HF paths. The calculation of
signal strength is especially dicult, 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 reection, etc.) are
dicult 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 reections 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 signicant 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
eects of the highlatitude ionosphere
ASAPS 2
FTZMUF2
FTZ4
HFBC84
HFMUFES4
ICEPAC
foF2 and M3000

MUFLUF (?)
Improved
calculation of
several parameters
Includes some
eects 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
Hateld (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.
176
radio-prediction purposes. One excellent example is the spaceweather, magnetospheric, and ionospheric data bases available from the U. S. NOAA Space
Environment Center (http://www.sec.noaa.gov).
3.8
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. 6972. AGARD/NATO.
Andersen, J. B., Hvid, J. T., and Toftgard, J. (1993) Comparison between dierent 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, 17.
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 eects on radio propagation. IEEE Trans. Antennas Propagation 40, 818828.
Jordan, E. C. and Balmain, K. G. (1968) Electromagnetic Waves and Radiating

Systems, 2nd Edition. Prentice-Hall, Inc., Englewood Clis, 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 Clis, 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, 535549.
Crane. R. K. (1974) Morphology of ionospheric scintillation. Technical Note 197426,
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),
12501259.
Eckersley, T. L. (1928) Letter to the editor. Nature 122,768769.
Eckersley, T. L. (1929) An investigation of short waves. J. Inst. Electr. Engineers 67,
9921032.
Eckersley, T. L. (1931) 19291930 developments in the study of radio wave propagation. Marconi Rev. 5: 18.
Eckersley, T. L. (1932) Studies in radio transmission. J. Inst. Electr. Engineers. 71,
434443.
177
178
Fraser-Smith, A. C. and Bannister, P. R. (1997) Reception of ELF signals at antipodal

distances. Radio Sci. 32.
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, 551578.
Hunsucker, R. D. (1999) Electromagnetic Waves in the Ionosphere. In Wiley
Encyclopedia of Electrical and Electronics Engineering (ed. J. Webster), pp. 494506.
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 proles. Radio Sci., 2, 11191282.
Ratclie, J. A. (1956) Some aspects of diraction theory and their application to the
ionosphere. Rep. Prog. Phys., 19, 188.
Ratclie, 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, 332347.
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 Stratied Media, 2nd Edition. Pergamon
Press, New York.
Yeh, K.-C., Chao, H. Y. and Lin, K. H. (1999) A study of the generalized Faraday
eect in several media. Radio Sci. 34, 139.
Yeh, K.-C. and Liu, C.-H. (1982) Radio wave scintillation in the ionosphere. Proc.
IEEE 70, 324360.
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.
Booker, H. G. (1956) A theory of scattering by nonisotropic irregularities with application to radar reection from the aurora. J. Atmos. Terr. Phys. 8, 204221.
Booker, H. G. and Gordon, W. E. (1950) A theory of radio scattering in the troposphere. Proc. Inst. Radio Engineers 38, 401402.
Buneman, O. (1962) Scattering of radiation by the uctuations in a non-equilibrium
plasma. J. Geophys. Res. 67, 20502053.
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.
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,
34413443.
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. Homan).
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 fr Aeronomie, Katlenburg-Lindau.
Walker, A. D. M, Greenwald, R. A., and Baker, K. D. (1987) Determination of the
uctuation level of ionospheric irregularities from radar backscatter measurements.
Rad. Sci. 22: 689705.
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.
Hateld, V. E. (1980) HF communications predictions, 1978. (An economical up-todate computer code, AMBCOM). In SolarTerrestrial Predictions Proc. (ed. R. F.
Donnelly), Vol. 4, D2 115. US Government Printing Oce, Washington DC.
179
180
Jordan, E. C. and Balmain, K. G. (1968) Electromagnetic Waves and Radiating

Systems, 2nd Edition, Prentice-Hall, Englewood Clis, 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 Eect of the Ionosphere on Radiowave Signals and Systems
Performance (IES90) (ed. J. M. Goodman), pp. 313321. USGPO, available through
NTIS, Springeld, Virginia.
Rose, R. (1982) An emerging propagation prediction technology. In Eects of the
Ionosphere on Radiowave Systems (IES81) (ed. J. Goodman). US Government
Printing Oce, 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. Homan).
Oxford, Pergamon.
General reading
Kildal, P.-S. (2000) Foundations of Antennas A Unied 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 modications 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. 502505), 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
frequencyheight 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 0.525 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 reected 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 reection 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 conrm 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 rst-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 lm 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 0.512 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.
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 dicult 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
Specications
Source
Digisonde Portable
Sounder (DPS)
Frequency range 132 MHz

Power 300 W pulse
Height range 901000 km
Doppler sounding, etc.
Realtime data transfer via
the internet
Automatic scaling available
University of
Massachusetts Center for
Atmospheric Research,
600 Suolk Street, 3rd
Floor, Lowell, MA 01854,
USA www.uml.edu
Canadian Advanced
Digital Ionosonde (CADI)
Frequency range 120 MHz

Power 600 W pulse (13-bit
Barker code)
Height range 901024 km
Realtime data transfer via
the internet
Scientic Instruments,
Ltd, 2233 Hanselman
Avenue, Saskatoon, CA
S7L6A7, USA
Ionosonde: HF Diagnostics Frequency range 130 MHz

Module, 01-2000
Power 1 kW CW and
peak
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 specications contact

www.kel.gov.au
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 denitely 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-reected 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 eld-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.
conguration. 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. 94109), and Hunsucker (1993, pp.
441450). Specically, 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-
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 104109 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 eld-aligned irregularities, at near-perpendicular incidence. Thus, coherent backscatter is 4090 dB stronger than incoherent scatter,
and is qualitatively similar to specular reection. 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 uctuations 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 rst 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 classied 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 eld-aligned irregularities, or by backscatter from irregularities via a ground-reected 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 aected by irregularities near the ionospheric-reection point, by
the Earth-surface characteristics, and by eld-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 signicantly inuenced 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 conguration 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 (330 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
191
CONVERTED
SKYWAVE PULSE
(a)
NORMAL
SKYWAVE
PULSE
GROUNDWAVE
PULSE
TRANSMITTER
NORMAL SKYWAVE
COMPONENT
CONVERTED SKYWAVE
COMPONENT
100
200
300
TIME (MICROSECONDS)
400
500
0.40
AMPLITUDE ( 101)
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.)
820 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 rst 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 elds 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-eld-aligned irregularities in the high-latitude F region, and
their frequency range of 820 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 dierent 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
6 12 18
TIME (GMT)
264 0000
1978
100

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
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
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
Figure 4.9. VLF pulse-reection 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
Figure 4.10 Locations and elds 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; 2ykkvibr (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
Pykkvibr, 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
Aliation
Notes:
Co-operative United Kingdom Twin Located Auroral Sounding System.
b
Centre National de la Recherche Scientique.
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
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 1214
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
(200300 ms) and is equivalent to 3045 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
eld of view of a pair of HF SuperDARN antennas covers 1520 of invariant latitude and 3 h of magnetic local time. The elds of view of several pairs of HF
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 ll this common viewing area, it would be possible to monitor the
dynamics of plasma convection over a signicant 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
eld-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 exemplied by
the Scandinavian Twin Auroral Radar Experiment (STARE), which was rst
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).
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), oering 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.
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)
Auroral and polar cap/

CW/bistatic
Kustov et al. (1996);

50 kW
Canadian
Arctic
SAPPHIRE
NE Canada
SHERPA
Auroral and polar/pulsed Hanuise et al. (1992)
Polar
SUPERDARN
Polar cap and auroral/

pulsed
Greenwald et al.
(1995); 616 MHz;
1 kW each into
16 antennas,
52 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, ve antennas
(Portable)
FAR
D, E, and F regions/
pulsed
Tsunoda (1992);
250 MHz, various
pulse widths
Halley Bay,
Antarctica
PACE
Polar cap F region/pulsed Baker et al. (1989);

820 MHz, 1 kW each
into 16 antennas,
52 azimuth sector
Antarctica
SYOWA
Auroral/pulsed
50 and 112 MHz,

15 kW, 314-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
dEtude 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 Earths magnetic eld 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, aecting 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.
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 rst 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 eects, 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 diraction 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. 6566 and
215217), Hunsucker (1991, pp. 5053), Hargreaves (1992, pp. 6566 and 7172),
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. 217219), and Hunsucker (1991, Chapter 7, pp. 165183) 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 (25 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 reected by the E region. This technique requires very careful, frequent
calibration of the system, plus a measurement of the E-region reection coecient. A variant of this method is the URSI A1b method, which uses the same
basic equipment and modied 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
6752 N
2026 E
7648 N
6427 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
0.0012.0
0.0011.0
Binary
Binary
Binary
1.0
1.0
Analog
Analog
Analog
250350
7080
3035
8-bit ADC,
32-bit complex, ACFs, parallel channels
6935 N
1914 E
7730 N
6612 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
Troms
Location
Table 4.4. Parameters of the EISCAT radar system (courtesy of EISCAT Corp.)
48
Any
Dish
32 m
Cassegrain
UHF
931
8
Binary
Analog
3035
6722 N
2638 E
7643 N
6334 N
Sodankyl
42.5
Circular
45
Circular
UHF
500
10
16 klystrons
6
1.0
0.25
0.0012.0
Binary
0.1
Analogdigital
5565
12-bit ADC,
Lag proles 32-bit complex
Antenna 1
Antenna 2
Dish
Dish
32 m
42 m xed
Cassegrain
7809 N
1602 E
8206 N
7518 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. 218219 and in Hargreaves (1992, pp. 7172),
and a rather extended discussion in Hunsucker (1991, pp. 169178). 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 rst globally deployed
during the International Geophysical Year (IGY), 19571959. 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 veried 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 rst 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 xed 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 dicult part of absorption measurement by the riometer
technique (Krishnaswamy et al., 1985).
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 inuences 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 3050-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 rst 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
lters 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, sucient 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 rst Imaging Riometer for Ionospheric Studies (IRIS) was installed at the
South Pole in 19881989 (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 ner scale, whose motions may be also be observed. This type of system
209
Pole star
cosmic noise
(method A2)
27 MHz
F layer
E layer
2.6 MHz
200 MHz
200400 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 ner-scale absorption must
have implications for the eect 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 eld-strength measurements at oblique incidence
over ground distances of 200400 km, using frequencies from 23 and 6 MHz. The
vertical-plane antenna patterns must be very uniform, so that small changes in
reecting-layer height will not aect the system losses, and one dominant mode
must be used. Transmitter outputs and receiver sensitivities must be stable and calibrated, and no signicant 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 dierence 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-reection-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 dierential-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. 2829 and 7677) and in Hunsucker (1991, pp. 180182). Other
techniques that have been used to measure D-region absorption are described in
Hunsucker (1991, pp. 182183) and in Hunsucker (1993, pp. 459464). Table 4.5
summarizes most of the absorption-measurement techniques.
4.2.5
Ionospheric modication 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 eect (or the LuxembourgGorkii
eect). Bailey (1937) was apparently the rst 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
conrmed 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. 506537) to ionospheric modication, as did Hunsucker (1991, pp. 142164). The former stressed results of modication experiments, whereas the latter stressed the technique. Another
description (mainly theoretical) of ionospheric modication was Chapter 10 (pp.
267284) by Erukhimov and Mityakov in the WITS Handbook (Liu, 1989).
Radio-wave interaction and ionospheric heating were also discussed by
Hargreaves (1992, pp. 9394).
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. 507511)) 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.
146152); HF heating theory was given on pp. 152155; 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
proles
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, dicult
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.
213
Figure 4.24. Some of the eects 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
110 ms
ENHANCED
ELECTRON
TEMPERATURE
110 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 modication facilities in
operation from c. 1970 to the present time.
Capabilities and limitations of ionospheric-modication techniques

The HF stimulation of the ionospheric plasma produces both linear and nonlinear eects, 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. 162163, and the Proceedings of the AGARD Conference on
Ionospheric Modication, 1991.) It has also been demonstrated that the auroral
electrojet can be modied by HF-modulated stimulation, to produce both VLF
Table 4.6. Ionospheric-modication 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/315 MHz
Operational in 1971;
Gordon et al. (1971)
EISCAT; Troms, Norway
69.6 N; 1200 MW
HIPAS UCLA, USA
64.9 N/146.9 W;
50 MW/2.84.9 MHz
Wong et al. (1983)
Established by the USSR

Gissar (Dushanbe)
38 N; 68 MW/46 MHz
Operational in 1981
Erukhimov et al. (1985)
Kharkov
50 N; 612 MHz
Bogdan et al. (1980)
Moscow
56 N; 1000 MW/1.35 MHz
Schluyger (1974)
Sura Radiophysics
Research Institute
Nizhni Novgorod
56 N; 4.59 MHz
Belov et al. (1981)
Zimenki
56 N; 20 MW/
4.65.7 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;
2.810 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 eective 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 eects.
4.3
Space-based systems
4.3.1
A history of Earth-satellite and radio-rocket probing
Hey et al. (1946) were probably the rst 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 eect (Murray and
Hargreaves, 1954; Browne et al., 1956; Evans, 1956). With the advent of the
articial-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 CanadianUS 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. 437454), but will not be discussed in this book.
During the last decade, several books that discuss Earth-satellite and rocketradio techniques for probing the Earths ionosphere have been published (Liu,
1989, pp. 44147; Davies, 1990, pp. 260296; Hunsucker, 1991, pp. 187207;
Hargreaves, 1992, pp. 6465 and 6771).
4.3.2
Basic principles of operation and current deployment

of radio-beacon experiments
The rst 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 ne
structure of ionospheric irregularities.
The TEC can be determined from RBE satellites by measuring the dierential
Doppler eect between two signals (Bowhill, 1958), the Faraday rotation of the
electric vector, the modulation phase (or group delay) between two dierent frequencies, or the carrier-phase dierence 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 rst results obtained using geostationary satellite RBEs were reported by
(Garriott et al., 1965). Hargreaves developed the rst proposal for a geosynchronous RBE specically 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. 261273),
Hunsucker (1991, pp. 200203), and Hargreaves (1992, pp. 6465). 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. 205207), and by Hargreaves (1992,
pp. 5253). 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 rst 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-eld
instruments (see pp. 4958 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
Ratclie and Pawsey (1933) and by Pawsey (1935), and was rst applied experimentally by Mitra (1949). Discussions of these techniques were given by Kelley
(1989, pp. 431434), Davies (1990, p. 243245), Hunsucker (1992, pp. 207211),
and Hargreaves (1992, pp. 300302), 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. 300302)
and by Hunsucker (1993, pp. 470473).
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 eects.
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 eects,
and provide much information
on the ionospheric above the
F2 peak.
More-complex instrumentation
than most beacons.
4.4 Other techniques
219
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
1e 5
1e -2
LF
0.1
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 rst 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
383398); the WITS Handbook, edited by Liu; Hunsucker (1991, pp. 211213);
Hargreaves (1992, pp. 6667); 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
rst 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-rst 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 rst time, we now have near-realtime access to solar, interplanetary, and
magnetospheric data from a new generation of scientic 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 aected by D-region absorption and intense E-region ionization, and HF
radars may also be aected 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
Section 4.1
Hargreaves, J. K. (1992) The SolarTerrestrial 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, 19901992 (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.
494506. Wiley, New York.
Kelley, M. C. (1989) Appendix A ionspheric measurement techniques. In The Earths
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 reection of radio waves. Proc R. Soc. A 109, 621641.
Bailey, V. A. (1937) Interaction by resonance of radio waves. Nature 139, 6869.
Baker, K. B., Greenwald, R. A., and Ruohoniemi, J. M. (1989) PACE: Polar AngloAmerican conjugate experiment. Eos 22, 785799.
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
reection. J. Geophys. Res. 69, 27792818.
Belrose, J. S. (1970) Radio wave probing of the ionosphere by the partial reection of
radio waves (from heights below 100 km). J. Atmos. Terr. Phys. 32, 567596.
Booker, H. G. (1979) The role of acoustic gravity waves in the generation of spread-F
and ionospheric scintillations. J. Atmos. Terr. Phys. 41, 501515.
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, 117129.
Brookner, (1987) Array radars: an update. Microwave J., February, 117137.
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, 73155.
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, 142158.
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 eld distributions in the magnetosphere: a new technique. Proc. IES99,
563574.
Gardner, F. F and Pawsey, J.L. (1953) Study of the ionospheric D-region using partial
reections. 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,
10211039.
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, 6374.
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, 761796.
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 rst decade of riometry. Proc. IEEE 57, 13481373.
Kossey, P. A., Turtle, J. P., Pagliarulo, R. P., Klemetti, W. I., and Rasmussen, J. E.
(1983) VLF reection properties of the normal and disturbed polar ionosphere in
northern Greenland. Rad. Sci., 18, 907916.
Krishnaswamy, S., Detrick, D. L., and Rosenberg, T. J. (1985) The inection point
method of determining riometer quiet day curves. Radio Sci. 20, 123130.
Kustov, A.V., Koehler, J. A., Sofko, G. J., and Danskin, D. W. (1996) The SAPPHIRENorth radar experiment: Observations of discrete and diuse echoes. J. Geophys. Res.,
101, 79737986.
Kustov, A.V., Koehler, J. A., Sofko, G. J., Danskin, D. W., and Schier, A. (1997)
Relationship of the SAPPHIRE -North merged velocity and the plasma convection
velocity derived from simultaneous SuperDARN radar measurements. J. Geophys.
Res. 102, 24952501.
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 rst results. Ann.Geophysicae 9,
461469
Meek, C. H. and Manson, A. H. (1987) Medium frequency interferometry of
Saskatchewan, Canada. Can J. Phys. A 35, 917921.
Peterson, A. M. (1951) The mechanism of F-layer propagated backscatter. J Geophys.
Res. 56, 221237.
Rawer, K. (1976) Manual on Ionospheric Absorption Measurements. World Data
Center A for SolarTerrestrial 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, 23452358.
Schlegel, K. (1984) HF and VHF Coherent Radars for Investigation of the High
Latitude Ionosphere. Max-Planck-Institut fr Aeronomie, Katlenburg-Lindau.
Stubbe, P., Kopka, H. and Rietveld, M. T. (1985) Ionospheric modication experiment
with the Troms heating facility., J. Atmos. Terr. Phys. 47, 11511163.
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 modication experiment using very high power,
high frequency transmission. J. Geophys. Res. 73, 64026405.
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, 22832307.
Wilkinson, P. (ed.) (1995) Ionosonde Networks and Stations. World Data Center A for
SolarTerrestrial Physics, National Geophysical Data Center, Boulder, Colorado.
Wong, A. Y. (1990) Foreword: Ionospheric modication 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,
545553.
Benson, R. F. and Vinas, F. (1988) Plasma instabilities stimulated by HF transmitters
in space. Radio Sci. 23, 585590.
Benson, R. F. (1997) Evidence for the stimulation of eld-aligned electron density
irregularities on a short time scale by ionospheric topside sounders. J. Atmos.
SolarTerr. Phys. 59, 22812293.
Bowhill, S. A. (1958) The Faraday-rotation rate of a satellite radio signal. J. Atmos.
Terr. Phys. 13, 175176.
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, 901920.
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 rst-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, 49094913.
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, 357430.
Davies, K. and Baker, D. M. (1966) On frequency variations of ionospherically propagated HF radio signals. Radio Sci. 1, 545556.
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, 785799.
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. 330339.
Evans, J. V. (1956) The measurement of electron content of the ionosphere by the
lunar radar method. Proc. Phys. Soc. B 69, 953955.
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,
167187.
Garriott, O. K., Smith, F. L., and Yuen, P. C. (1965) Observations of ionospheric electron content using a geostationary satellite. Planet. Space Sci. 13, 829838.
Hewish, A. (1952) The diraction of galactic radio waves as a method of investigating
the irregular structure of the ionosphere. Proc. R. Soc. A 214, 494514.
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)
Global ionosphere perturbations monitored by the worldwide GPS network. Geophys.

Res. Lett. 23, 32193222.
Hunsucker, R. D. and Owren, L. (1962) Auroral sporadic-E ionization. J. Res. NBS
Radio propagation D 66, 581592.
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, 269271.
Kunitsyn, V. E., and Tereschenko, E. D. (1992) Radio tomography of the ionosphere.
IEEE Antennas Propagation Mag. 34, 2232.
Little, C. G. (1952) The origin of the uctuations 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, 315320.
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
eect 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, 23592372.
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,
22832286.
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, 10071012.
Ratclie, 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, 385396.
225
226
Radio techniques
Rogers, G. L. (1956) A new method of analyzing ionospheric movement records.

Nature 177, 613614.
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, 43134324.
Smith, F. G., Little, C. G., and Lovell, A. C. B. (1950) Origin of the uctuations in the
intensity of radio waves from galactic sources. Nature 165, 422423.
Stone, W. R. (1976) A holographic radio camera technique for the 3D reconstruction
of ionospheric inhomogeneities. J. Atmos. Terr. Phys. 38, 583592.
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 eld reconstruction of VHF radio waves in
investigating single, isolated ionospheric irregularities. Radio Sci. 17, 684692.
Watts, J. M. and Davies, K. (1960) Rapid frequency analysis of fading radio signals. J.
Geophys.Res. 65 22952302.
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 inuenced by the outer magnetosphere

and the solar wind, the essential connection being via the geomagnetic eld.
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 dierences 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 inuence 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 ionneutral-species collision frequency is small relative to the

gyrofrequency, and therefore the plasma moves with the magnetic-eld lines
rather than with the neutral wind. The motion may also be considered as the
motor eect of the cross-polar electric eld mapped down from the magnetosphere. The plasma speed v, the electric eld E and the magnetic ux density B
are related by vE/B (Equation (2.12)). Since the polar magnetic eld is almost
vertical, with a value of about 5
105 Wb m2, typical plasma speeds of
2001000 m s1 correspond to electric elds of 1050 mV m1.
The integral of the electric eld across the polar cap, which may be determined
from satellite measurements, provides an estimate of the total electric potential
dierence 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 ux density of the interplanetary magnetic eld
(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 11.7B sin3( /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 rst 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 ow pattern caused by the polar-cap electric eld is simple enough.
The plasma ows from the noon sector to the midnight sector directly over the
pole, and there is a return ow 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 ow is typically several hundred m s1. The ow over the polar
cap corresponds to the motion of open eld-lines from the cusp to the tail (Section
2.4.1), and the return ow corresponds to the sunward ow of closed eld-lines
down the anks of the magnetosphere. However the co-rotation eect, conveniently represented by the co-rotation electric eld (Section 2.4.4), must also be
included, and then the ow 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
elds 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 dierent, and the evening cell is particularly aected
because here the return ow and the co-rotation act in opposite directions. There
are some eld-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 inuence 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
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 eld 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 SunEarth (Bx) and
eastwest (By) components of the IMF. (R. A. Heelis, J. Geophys. Res. 89, 2873, 1984,
copyright by the American Geophysical Union.)
The eastwest component of the IMF (usually called By ) also aects 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 northsouth and the eastwest 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 dier. The inuence of the
eastwest 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 eastwest 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 diering magnitude.
tial distributions (equivalent to ow 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) nd a marked seasonal
variation in the ow pattern. When the IMF is southward, the central region of
the ow, 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. 22001800 LT). There is (of course!) a considerable spread about these trends
on individual days.
The eect of the eastwest 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
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 eastwest component, the patterns are similar to each other but
of dierent intensities.
(3)
If the IMF is strongly northward, the patterns in the summer and winter
polar caps are very dierent.
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 eld, and the other the night-time ionospheres 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 eastwest component larger, (b) northward component but eastwest component larger, and (c) northward component with smaller eastwest
component. The plots show the equipotentials, which are also the streamlines of the polar
ow. (G. Lu et al., J. Geophys. Res. 99, 6491, 1994, copyright by the American Geophysical
Union.)
233
234
the magnetotail. The latter are expected to be delayed 3060 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 1725 min after the IMF turned from northward to southward. For a
northward turning the lag was 2844 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 eld (as measured by a stationary observer), which is communicated along the eld-lines to the E region.
Therefore the Pedersen current (Section 1.5) in the E region also alters abruptly.
To maintain continuity, current then ows up the eld lines as a Birkeland current.
See Figure 5.5. The corresponding downward ow 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 23 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
eld; v, velocity; E, electric eld; IH, IP, and I: Hall, Pedersen, and eld-aligned currents.
(b) The eld-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
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.
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 eective at bringing solar-produced ionization over the pole. It is signicant that, in the northern hemisphere, where the
separation between the poles is smaller, the UT eect 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 eld-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 2.4.12) the cusps are where the dayside eld-lines open before being
swept over the poles (Figure 5.7(a)). The cusps are signicant 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 ux 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
(a)
(b)
Figure 5.7. Aspects of the polar cusp and its F-region eects. (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 eld-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.)
(3)
Owing to the inux 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 eld-lines, ionospheric plasma
may ow 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 eld-lines is marked, and, from
scanning-photometer observations, Walker et al. (1998) were able to identify ionospheric eects due to (1) precipitation of electrons from the ring current on the
last closed eld-lines, (2) a downward eld-aligned current on the rst open eldlines, and (3) dispersion of precipitating soft ions on the ux tubes convecting
poleward. The last eect shows up as the increasing height of the layer maximum.
5.2.3
The polar wind
The circulation of the magnetosphere carries eld-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 eective 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 ow upward, and, in the absence of a boundary, the ow may continue as
long as the tube remains open.
A steady outward ow 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 eld directed upward. When light
ions (hydrogen and helium) are also present, they are accelerated by this electric
eld, 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
eld to cause the dynamic equilibrium state having a steady outow above some
altitude. He may also ow out, though to a lesser extent.
This continuous outow of light-ion plasma is the polar wind. In theory the
ow can even reach supersonic speeds, but the details depend on what is assumed
about the ow speed at a great distance. The term polar wind is sometimes
restricted to the supersonic regime, in which case subsonic ow would be a polar
breeze. The ow 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
240
O is far from uniform over the polar cap, the polar wind must be similarly variable.
The lighter ions are the most aected by the outow, 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 ux of H is heated by collision with the heavy ions,
and its temperature is signicantly 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 signicant 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
eectively in the lower F region, and for which electron-content observations lead
to estimates in the range 1091010 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 Troms, 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 (7580 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
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
10
14
18
VELOCITY (KM.SEC1)
Figure 5.8. Theoretical properties of the polar wind, showing the density of H and the
eld-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 Earths Magnetosphere, Terra Scientic
Publishing, Tokyo, 1983, p. 99.)
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 signicant 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 signicantly
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)
12 16 20 24
Summer
250 300 km
12 16 20 24
Local time
Winter
250 300 km
12 16 20 24
(b)
600
500
400
300
200
100
0
Yearly Average
250 300 km
12 16 20 24
Summer
250 300 km
12 16 20 24
Winter
250 300 km
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 Troms. (a) sunspot maximum (August 1981August 1983), (b)
sunspot minimum (April 1986March 1987). (Reprinted from A. D. Farmer et al., J.
Atmos. Terr. Phys. 52, 561, copyright 1990, with permission from Elsevier Science.)
244
eects 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 aecting 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 diraction mechanism, produce in propagating radio
waves the phenomenon of scintillation.
5.3.2
Enhancements: patches and blobs
We rst consider enhancements of large size occurring in the polar cap and the
auroral zone. They may be 501000 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 rst 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 20005000 km h1 (5001400 m s1) were reported. The cause of the
motion was correctly interpreted as being due to an electric eld, 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 signicant 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 veried by computations that have also demonstrated how a change of polar circulation, for instance due to an increase in the
ow 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 ow during a ux-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 ow 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 210 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 ow nor lagging behind.
(g)
They occur during all seasons of the year but more frequently during the
winter.
A dierent 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 noonmidnight 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 58 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 dierent 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
Figure 5.11. Typical irregular structures of the polar F region. (a) arcs with noonmidnight
alignment and duskdawn 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
denitive, 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 dierent 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
247
248
(a)
(b)
Figure 5.13. Three kinds of blob observed with the EISCAT incoherent-scatter radar. (a)
F-region enhancement peaking at 250400 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)
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 rst Fresnel zone is an important parameter. For a radio frequency of 100 MHz the rst Fresnel zone has a radius of about
1 km if the eective diraction 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 oset 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
10100 km
northsouth,
Extreme
longitudinal
extent but
conned 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
23 h
Duration
Poleward
auroral
boundary.
Either
recongured
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
250700 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
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 eect 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
(dened 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
eld, and, according to Rino (1978), auroral irregularities are extended eastwest,
giving a sheet-like form. Since individual irregularities are strongly eld-aligned,
there is another maximum for rays traveling directly along the direction of the
magnetic eld (because rays traveling almost directly along the magnetic eld tend
to remain within a single irregularity). These eects may be seen in Figure 5.15.
5.3 Irregularities
Figure 5.15. The occurrence of scintillation over magnetic latitudes 5580 observed from
Kiruna (64.3 N, 102.8 E CGM), September 1984September 1986. The contours show
the percentage of time for which the scintillation at 150 MHz exceeded S4 0.2. The contours 15 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
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,
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 eld is nearly vertical over the polar cap, the propagation path is closest to the magnetic-eld 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
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 (peakpeak) 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 eect of diraction (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 (2030 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
03
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
10
100
Frequency (Hz)
Intensity Spectrum (dB)
(b)
20
30
40
50
60
S4 = 0.854 q = 2.359 0.140
70
0.1
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 uctuations 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 uctuations 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 KelvinHelmholtz 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
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 t 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 Eect of the Ionosphere on
Communication, Navigation, and Surveillance Systems (ed. Goodman), p. 599. IES87,
National Technical Information Service, US Government Printing Oce, Springeld,
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 (dened 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 eective phase screen and the form of the
irregularities.
The data compilation underlying this model revealed several signicant features.
(1)
The high-latitude scintillation region has a well dened boundary, across

which the irregularity strength increases by more than a factor of ten.
259
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 eastwest 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 dened. 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, rst 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 denition 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 rst 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 eect. 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 (733371 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 18 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 dierent 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 proles 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
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
2
4
1
4
4
3
8
7
N(550)
N(650)
N(750)
N(850)
N(950)
104
N(HMAX)
N(450)
8
6
4
N(HSAT)
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.
examples in which the trough is clearly dened. 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 aects 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 Moett 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
Figure 5.23. Troughs in electron content on four separate occasions when the trough was
narrow and well dened. 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 diuse 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 eect.
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 coecients. 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
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 coecients 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 tted quite well by the formula of Kohnlein and Raitt (shown superimposed) during the hours around midnight. (Liszkas 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 03 (Figure 5.27), which again agrees with the
formula of Kohnlein and Raitt. Note, however, that individual values are spread
23 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 Hggstrm (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 signicantly
higher latitudes during the afternoon than does that of Kohnlein and Raitt. Note
that the trough in Figure 5.25 occurred at 7274 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)5.4 0.5Kp 0.13t,
(5.4)
L(Te maximum)5.20.4Kp 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. Coecients 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
19000500
Kohnlein and
Raitt (1977)
Satellite
ESRO-4
65.2
2.1
0.5
20000700
Best et al. (1984)
Satellite
Intercosmos
18
64.0
0.5
0.13
Not stated
Collis and
Hggstrm (1988)
EISCAT
62.2
1.6
1.35
13000100
60
65
70
75
16
18
20
22
0
LMT
02
04
06
Kp = 2
Winter
196566
Spring
1965
Summer
1965
Autumn
1965
Winter
196465
08
Kp = 0
10
Figure 5.26. Latitude of the trough against local time, from a years 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
5 Kp
60
60
5 Kp
WINTER 6566
65
AUTUMN 65
WINTER 6566
65
SUMMER 65
AUTUMN 65
70
SPRING 65
WINTER 6465
SUMMER 65
N
75
GEOGRAPHIC
LATITUDE
SPRING 65
WINTER 6465
(b)
70
N
75
GEOGRAPHIC
LATITUDE
Figure 5.27. Latitude variation of the trough in electron content with the magnetic index Kp: (a) 19002000 LMT, and (b) 03000400 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 dierences 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 dierences are due to hemispheric
dierences 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 reection 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 00000159 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
(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 2021 June 1982. The time is local. (b) Collection of the poleward
edge locations for all observations during 22002359 LT, with the best-tting straight line.
(c) The best-tting 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 eld. 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 rst 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 conrm 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 dierent 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 signicant that, in some cases, the slope attens 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 21300800 LT
overall, though every example extends into the period 00000400 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 Hggstrm (C H) rather than with
271
272
(b)
200 km N
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 ve 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 19821983, 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
5.4 The main trough
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 eect of substorms, which
on some occasions are seen to be related to a partial lling 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 lling 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 dierent circulation patterns in those two regions. Various attempts to predict the position of
the trough have been made by modeling the ionosphere mathematically (Moett
and Quegan, 1983). These models represent the high-latitude convection in a
273
274
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 ow (with respect to the Earth)
is strongly westward. Such a ow tends to oest the Earths 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 dierence 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.)
RATE COEFFICIENTS (cm3 s1)
1010
O+ O2
O+ N2
1011
5
1012
5
1013
10
12
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 coecients
k1 and k2 in the expression for the recombination coecient
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 ow of plasma that also depletes the F region. It is
argued, therefore, that plasma depletion is expected in regions heated by high
dierences in speed between ions and neutral particles.
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-eld 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 48 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 7075 north, from OGO-6, 18 March
1970, 18301854 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
The ion temperature (Ti ) and the electric eld are usually increased within
the trough but the electron temperature is not usually aected.
There are relationships between Ti and the eld-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
80. The electron density (at 300 km) is as low as (13)
102 cm3, compared with
up to 105 cm3 elsewhere in the polar cap. The hole appears sporadically at the
277
278
12h00
6
5
4
3
2
1
C
A
18h00
06h00
B
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 eld is small. The solar terminator is along the line 18000600. A,
main trough; B, polar hole. C, region of signicant 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 eld

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 outow of ions can be signicant;

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 aects equatorward

edge; high Te by virtue of conduction of heat from
high altitudes
Comment
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 aects systems operating at the

higher frequencies, specically in the HF, VHF, and UHF bands. Eects 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 eects 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 reection 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 dierent in the
southern hemisphere.
When energetic electrons and protons precipitate into the auroral F region
they produce eld-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 eld-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 eld-aligned irregularities,
and satellite-to-earth VHF and UHF signals suer scintillation phenomena
causing a rapid and sometimes severe fading of amplitude and irregular uctuations 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 eld 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 sucient for prediction purposes.
A detailed description of the eects 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
5.1
Circulation of the high-latitude ionosphere
Boyle, C. B., Rei, P. H., and Hairston, M. R. (1997). Empirical polar cap potentials.
Cowley, S. W. and Lockwood, M. (1997) Excitation and decay of solar wind-driven
ows 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 eect on the polar
cap convection. Geophys. Res. Lett. 26, 975.
Kelley, M. C. (1989) Section 6.2. In The Earths 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 ow 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 eld 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
Troms. 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 Earths Magnetosphere (ed. R. G. Johnson),
p. 99. Terra Scientic 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
282
Walker, I. K., Moen, J., Mitchell, C. N., Kersley, L., and Sandholt, P. E. (1998)
Ionospheric eects 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 Beaujardire, 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 diuse 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,
155315.
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.
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.
Weber, E. J., Klobuchar, J. A., Buchau, J., Carlson, H. C., Livingston, R. C., de la
Beaujardire, 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 Hggstrm, 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
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 eld above the EISCAT radar. Ann.
Geophysicae 15, 740.
Kavanagh, L. D., Freeman, L. W., and Chen, A. J. (1968) Plasma ow 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 dierential-phase measurements. J. Atmos.
Terr. Phys. 55, 1021.
Moett, 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.
Muldrew, D. B. (1965) F-layer ionization troughs deduced from Alouette data. J.
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 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.
Rodger, A. S., Moett, 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.
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 ow 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
Proles 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., Moett, 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.
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 reection 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 owing 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 waveparticle 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 specic 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 1025 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 3060 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 o 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 rst
dened 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 19571958, pointed out that at a xed 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 xed 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,
to mark the division between open and closed eld-lines. The regions poleward of
the ovals (one in each hemisphere) are generally taken to be the polar caps in
which the magnetic eld-lines connect to the IMF and circulate under the inuence 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 conguration called the -aurora. Sometimes the morning side of the oval is quiet while the
287
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 congurations and
dynamics emphasizes the complexity of the auroral distribution and suggests that
present classications 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 Whalens results: Q8 if
Kp &6; QKp 2 if 1Kp 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 (118165 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 inux in terms of magnetic latitude and local time, at
levels of geomagnetic activity quantied by Kp. Figure 6.6 shows the distribution
of electrons for Kp 3. Figure 6.6(a) is very like the auroral oval, being oset 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) dened 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 ux of 107 electrons cm2 s1 steradian1.
The transition latitude, where the uxes of low- and high-energy particles were
equal, was also noted.
289
290
Figure 6.4. Images of the northern auroral region observed by the Viking satellite. The
camera had a 20 by 25 eld of view and responded to UV of 134180 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 15
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 13 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 rst written
work devoted entirely to the polar aurora was published in France in 1733.
The rst 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 red into an aurora, that energetic electrons were
identied 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 ne 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 anities 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 rst 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

Latitude
Figure 6.5. (cont.)
Figure 6.5b (2)
AURORAL OVAL
(Feldstein 1967)
Q=5

Latitude
AURORAL OVAL
(Feldstein 1967)
Q=7

Latitude
296
(a)
(c)
(b)
Figure 6.6. Electron-precipitation zones for

Kp 3. (a) The total ux 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 ux, in units of keV cm2 s1 sr1,
due to the same ux 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 classication 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. Strmer made 12 000 height
determinations with spaced cameras and found that the lower borders of auroral
forms were usually at heights of 100110 km (Figure 6.8(a)). In some of the forms
the luminosity is concentrated into a band only 1020 km deep and the lower edge,
in particular, can be quite sharp. The brightness of a discrete arc typically falls o
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.
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)
8082
8082
7375
7677
High-to-lowenergy
transition
7375
7375
7072
7072
Equatorward
boundary
(high energy)
6971
6769
6466
6466
Table 6.2. Values of L0 and a for

Equation (6.1)
MLT (h)
L0
0001
0102
66.1
65.1
1.99
1.55
0405
0506
0607
0708
0809
0910
1011
1112
1213
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
1516
1617
1718
1819
1920
2021
2122
2223
2324
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
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. Classication of auroral forms

Forms without ray structure
Homogeneous ray structure: a luminous arch stretching across the sky in a
magnetically eastwest 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
Diuse surface: an amorphous glow without distinct boundary, or isolated patches
resembling clouds
Pulsating surface: a diuse 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
(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
500
1000
Light intensity
Number of observations
Figure 6.8. Observations of auroral luminosity. (a) The distribution of 12 330 height measurements made by Strmer and colleagues. The vast majority lie between 90 and 150 km.
(C. Strmer, The Polar Aurora. Oxford University Press, 1955. By kind permission of
Oxford University Press.) (b) Proles 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 dened 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 classication of brightness uses a scale of IIV, as in Table 6.4. This also shows the
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 xed 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 suciently sensitive to record the most rapid uctuations 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 scientic 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 signicant distinction that should be made is that between discrete and
diuse (or mantle) aurora. All the earlier studies concentrated on the discrete
Table 6.4. Intensity classication 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
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 ne and dynamic structure.
However, as was demonstrated in the early 1960s, the aurora may also take the
form of a diuse glow. This contributes at least as much total light as the discrete
forms, though it is more dicult to observe from the ground because of its low
intensity per unit area. The night-time discrete and diuse aurorae are thought to
map along the geomagnetic eld into dierent regions of the magnetotail (Section
2.2.6 and Figure 2.6). The diuse aurora is generally associated with the central
part of the plasma sheet, and the discrete forms, which tend to appear poleward
of the diuse 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 aects the ground-based techniques, have provided much new information about the distribution of the luminous emissions. The diuse aurora tends to dominate in these pictures, but
discrete forms are also seen within the diuse 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
302
present when the IMF is northward. It is believed that these arcs are on closed
eld-lines and that they may be magnetically conjugate (i.e. they occur simultaneously at opposite ends of eld-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 dierent 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 rst 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
303
(a)
15
90
68
394
7
77
74
49
811
989 878
1027
13
56
16
1304
922
41
1152
999
1218
10
79
436
8
84
46
12
4
55
77
Singlets
(b)
Triplets
D40
D40
5
P3,2,1
0
5
P3,2,1
S2
D3,2,1
D3,2,1
P2,1,0
P2,1,0
S1
1 0
F3
F3
D2
D2
P1
P1
297
2
6300
1
S0
S0
Energy above Ol ground state (eV)
14
Quintets
Excitation
energy
Quantum
state
4.17 eV
S0
0.74 s
2972
UV
1.96 eV
1D
110
s
6364
Red
0.00 eV
5577
Green
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 ngstrom unit.
radio propagation. They are briey 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
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 reect 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 reection 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 eld-aligned, the echo intensity is aspect-sensitive and the best observing geometry is when the radar lies in the plane normal to the magnetic eld-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 eect is usually monitored with a Riometer (Section
4.2.4), which typically operates in the range 3050 MHz, at which frequencies the
absorption rarely exceeds 10 dB; but the eect 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 proles 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 inuence 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.
Figure 6.12. The two zones of auroral particle precipitation in the northern hemisphere.
The density of symbols indicates the average ux, 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 ows 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 waveparticle 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 denition of auroral phenomena by recognizing that they actually fall into two groups having dierent patterns of occurrence (Figure 6.12). The inner one, corresponding to the luminous oval, is
characterized by
305
306
luminosity,
sporadic-E on ionograms,
spread-F on ionograms,
soft X-rays,
impulsive micropulsations,
negative bays on magnetometers,
soft but intense electron uxes 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 6070, with its center at about 65 geomagnetic latitude. This zone displays
diuse aurorae,
radio absorption,
sporadic-E at 8090 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 ux of particles (a) is distributed like the luminous oval.
However, the dayside particles being relatively soft at the higher latitudes, the total
energy ux (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 Brices 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
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 fr extraterrestrische Physik,
Garching bei Mnchen).
Figure provided by T. I.
Pulkkinen (Finnish
Meteorological
Institute).
308
camera on the POLAR satellite (averaged over 1 h), and the outer one is a composite of uxes 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 conrm 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 rst studied this tendency in
magnetic records and identied what he called the elementary polar magnetic
storm. However, Birkelands work in this area, which also involved eld-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) eld, 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 Chapmans 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 aects the auroral regions is best described by

Akasofus 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 (19561958), 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
05 min; (c) T510
min; (d) T1030 min;
(e) T30 min1 h; and
(f) T12 h. (S.-I.
Akasofu, Polar and
Magnetospheric
Substorms. Reidel, 1968,
Publishers.)
23 h between, and there is also a dynamic aspect. Akasofus 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 23 h later. By dening 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 rst studied in the magnetotail
(Section 2.2.6), which becomes gradually more tail-like for tens of minutes to an
310
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 conrmed
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 diuse
aurora is generally equatorward with a speed less than a few
hundred m s1.
(3)
The expansion phase is preceded by auroral intensications 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 eld-lines.
(6)
When the position of the onset is mapped along the geomagnetic eld to
the equatorial plane, it is consistent with the location of the inner boundary of energetic particle ux, 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 uxes 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 ux, and over a range of altitudes determined by the particle energies (Figure 2.26). Consequently, the substorm behav-
311
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
eects (Section 6.3.4). The E-region reections 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 owing during an
individual substorm. This is still an equivalent-current system, because it assumes
that the current ows 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
dierent 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 eld-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
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, SolarTerrestrial Physics. By permission of Oxford University
Press, 1972.)
within the magnetosphere as well as currents owing 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 inuential 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 eld-lines (as Birkeland currents) into the ionosphere. There the
circuit is completed in the E region, probably by an electrojet owing 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 ows
314
(a)
Electrojet
N
Field
Aligned
Currents
Near
Earth
Neutral
Line
(b) substorm electrojets
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 eld-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 eld 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 owing
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 ows in the late evening sector, before the Harang discontinuity, is dominated by a northward electric eld (in the northern hemisphere). An
electric eld, this time southward, also dominates the situation in the westward
electrojet later in the morning (Figure 6.20). These are the elds 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
RD
WA JET
T
S
O
WE CTR
E
EL
20
04
Ha
n
ra
g
Conductivity
Dominant
Di
sc
on
uit
tin
22
02
00
Magnetic Local Time
Figure 6.20. Regions of the electrojet dominated by conductivity and by electric elds. (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 aected 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 nal story has not yet emerged,
there are some aspects that seem well established.
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 eld-line at high latitude on the day side of the Earth and
follow its progress, we nd 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 eld 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 eld is produced
than removed, the total magnetic ux 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 ux coming from the polar regions,
the tail becomes more dipolar, ux is lost from the polar caps and the auroral oval
shrinks again. This sequence of events in the magnetosphere may be identied
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 eld 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 BiotSavart law. At the same time, satellites at geosynchronous distance observe an increase in the ux of energetic electrons, and the
geomagnetic eld 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 (20100)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 ux during
the expansion phase, indicating that the plasma sheet becomes thinner at that
time.
318
Figure 6.22. A Y-type

neutral point at the
Earthward edge of the
neutral sheet where the
high-speed ion ow is
stopped by the dipolelike eld 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 ows (BBFs) in the plasma sheet. Plasma in the sheet generally ows 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 ow 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 (90100)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-owing
plasma gets closer to the Earth, it is stopped by the stronger geomagnetic eld
which is dipolar in form. Figure 6.22 illustrates the stopping region. Note that the
magnetic eld is congured with a Y-type neutral line (compare with Figure 2.20).
In this treatment the ow stops at the boundary between the tail eld and the
dipole eld, and this region is also the inner edge of the plasma sheet.
Various theories
The exact conguration 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 eld, 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 (1020)RE,

which allows magnetic reconnection between the lobes of the magnetotail.
(b)
Generation of the KelvinHelmholtz 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 Alfvn waves and consequent sudden heating.
(d)
Intense eld-aligned currents due to an increase in the rate of eld reconnection on the day side of the magnetosphere, leading to the current
wedge of the substorm and a collapse of eld in the magnetotail.
(e)
A disruption of the cross-tail current due to a current instabilty.
(f)
A ballooning instability invoking a transition between eld congurations 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 inuence 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 ux of the solar wind and the eciency 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 dened 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 dierent combinations of solar-wind parameters also correlate to the occurrence of substorms, though ) is perhaps the best (Figure 6.24).
320
Figure 6.23. A selection of substorm-triggering ideas. (A. T. Y. Lui, in Magnetospheric

Substorms, copyright by the American Geophysical Union, 1991.)
The inuence 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 identiable 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 8001500 per annum, or about 24 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 dened in dierent ways, the former mainly from magnetic observations at low latitude where the greatest inuence is the ring current,
321
322
and the latter from observations at a higher latitude where the greatest contribution comes from the auroral electrojet. Their eects 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 signicant 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 ow. This view is supported by direct observations in the magnetotail, in which a signicant dierence between those substorms occurring
during a storm and those occurring at other times is found (Baumjohann, 1996).
In a storm-time substorm the magnetic eld moves from tail-like to dipole form
in a matter of 1530 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
dierent distances down the tail), one of which is the more eective 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 eld across the magnetosphere, showing that the solar
wind aects the magnetic storm directly, as well as indirectly via substorm activity (Clauer and McPherron, 1980).
Plainly, the nature of the stormsubstorm 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
diers 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 eects, 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 proles 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 aecting 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 110 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 proles of the rate of ionization due to a ux of 108 electrons cm2 s1
at several initial values of energy Ep (keV) precipitating along geomagnetic eld
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
100
20
Ti
2333 2247
Kp : 2
Temperature (K)
1000
3
4
5
Frequency (MHz)
100
200
300
400
500
200
F1
F2
(b)
200
300
400
500
600
700
0
1
100
200
300
400
500
2230 UT
20
40
Ti
1211 1219
Kp : 2
Temperature (K)
1000
3
4
5
Frequency (MHz)
7 8 9 10
2000
Te
3000
60
80 100 120 140 160 180 200
Electron Density (1010 m 3)
1215 UT
Figure 6.25. Comparable electron-density proles 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)
Figure 6.26. Proles of the rate of ionization due to monoenergetic uxes of 108 electrons
cm2 s1, incident from above, for the energy range (2100 keV) having greatest eect in the
E (and D) regions. (M. H. Rees, Physics and Chemistry of the Upper Atmosphere,
lines into the auroral ionosphere are shown in Figure 6.26. It is instructive to
compare these ionization proles with luminosity proles (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 ows, 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 elds 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
326
(a)
(b)
Figure 6.27. E-region electron density proles and ionograms associated with the passage
of an auroral arc across the eld 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
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 eld-aligned irregularities within the auroral
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,
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 dierence between the scattering cross-sections for
coherent and incoherent radar, the former being the stronger by 5080 dB.
Backscatter from auroral E-layer irregularities has been classied 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
Figure 6.29. An electron-density prole of the auroral E region during a great magnetic
storm in August 1972. (R. D. Hunsucker, Radio Sci. 10, 277, 1975, copyright by the
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 suciently large relative velocity
between the electrons drifting in an electric eld and the ions whose motion is
dominated by collisions. The waves travel nearly perpendicular to the geomagnetic
eld lines. The latter property implies that the backscatter cross-section is
maximum when the radar is pointing almost perpendicular to the eld 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 conrms that
the scattering process is a weak coherent one. There were, however, some signicant 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 eld-aligned irregularities in the auroral zone.
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
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
330
Figure 6.31. The geometry of coherent radar-scatter at high latitude. For signicant scattering the beam must be normal to the eld-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 ow
within auroral arcs. The motion produces an infrasonic bow wave, whose propagation is highly anisotropic and which reaches the Earths 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 10100 s a
frequency range of 0.010.1 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 AIWs
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 eect. 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-
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 eld, which
occurs in localized regions. The other is the Lorenz force (J
B), where B is the
ux density of the geomagnetic eld. 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
332
In an investigation of TIDs in electron content at L4, most of which were in

the medium-scale range with period 2060 min, Hunsucker and Hargreaves
(1988) noted that the waves were present almost continuously during daylight
hours at the level of 14% of the electron content. Although no specifc source was
identied, it must be signicant 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 510% of this redistribution is due
to AGWs.
6.6
Except for the very large seasonal variability, the polar E region is relatively
benign, compared with the auroral region. Precipitation of 110 keV electrons
along geomagnetic eld lines through the magnetospheric plasma sheet into the
auroral ionosphere produces several very important eects: 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 eld into the northern and southern auroral ovals,
which are stationary in space in SunEarth 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 proles 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
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 intensied
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 eect 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 eld, 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
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.
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 diuse 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
WindMagnetosphere Coupling (eds. Kamide and Slavin), p. 605. Terra Scientic, Tokyo.
Vestine, E. H. (1944) The geographic incidence of aurora and magnetic disturbance,
Northern Hemisphere. Terr. Magn. Atmos. Electricity 49, 77.
333
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 ow. 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 rst 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
Strmer, 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 ux 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.
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 19023, vol. 1, section
1. H. Aschehoug, Christiana.
Baumjohann, W. (1996) Stormsubstorm 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-
tary electric eld and magnetospheric substorms on partial current development. J.

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),
Hargreaves, J. K. (1996) Substorm eects 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 elds. 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.
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),
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
denitions and signatures. J. Geophys. Res. 85, 1663.
Shiokawa, K., Baumjohann, W., and Haerendel, G. (1997) Braking of high-speed
ows 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 proles
in the auroral zone ionosphere. Radio Sci. 9, 455.
335
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
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.
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) MagnetosphereIonosphere 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.
Strmer, 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
Chapter 7
The high-latitude D region
7.1
Introduction
The dierences 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) aecting 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 proles 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 regions 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 rst is auroral radio absorption (AA), which occurs only in the auroral
regions and is due to uxes 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 proles of the D and lower E regions due to electron precipitation. The incident electron ux 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 ux 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 prole of the eective recombination coecient 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 are.
These two kinds of phenomenon have rather dierent 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 conned 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 diculty of predicting the eects 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 Troms during the International Polar Year
(19321933), 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 dierent 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 19571958, 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 rst the determination of a quietday curve (QDC), an estimate of the signal level in the absence of absorption.
339
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 dierence 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 dicult 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 3050 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
(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) Kilpisjrvi (L5.9), 6 October 1994. (Reprinted from J. K.
Hargreaves et al., J. Atmos. SolarTerr. Phys. 59, 872, copyright 1997, with permission from
Elsevier Science.). The rst 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 classication
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
Figure 7.4. The occurrence of spike events (1 dB) at Abisko, 19801985. Magnetic midnight is about 2130 UT. The riometer frequency was 30 MHz. (Reprinted from J. K.
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
dierences are allowed.
Figure 7.5 illustrates the spatial connement 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 eastwest. 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 (L13) have been found to be remarkably similar in form
though they are generally smaller in magnitude. The spike event lasts for 12 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).
(a)
343
175.0
S
100 km
125.0
75.0
25.0
km
2
25.0
4
6
75.0
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) Kilpisjrvi, 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
Kilpisjrvi (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.
Elsevier Science.)
344
Figure 7.6. Absorption distributions during the main part of a night event at Kilpisjrvi 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 eld of view in a westward direction, namely the westward surge. It extends 7585 km northsouth but
its eastwest 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 73.7, L 13) they have
durations of less than 5 min, with the most probable value 12 min. The distribution of magnitude (at 38 MHz) peaks at 0.20.3 dB, and most occur between 1200
and 1800 local magnetic time with the mode at 15001600. Their spatial extent is
50100 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
Figure 7.7. An absorption arc observed at Kilpisjrvi (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 dened 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 11.5 h before the onset there may be observed a weak absorption event
lasting 4060 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 identied the form of the preceeding bay as an arc,
extended eastwest but only 60100 km wide northsouth (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
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)
Andya
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 Andya (L6.2), 23 August 1985
(Reprinted from J. K. Hargreaves and T. Devlin. J. Atmos. Terr. Phys. 52, 193, copyright
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 aecting
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 conned to a small area in the D region, less than 100 km
northsouth though more extended eastwest. Figure 7.9 is an example. In this
(a)
349
175
125
0.50
75
0.
75
0.50
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 prole of electron density, at 1-min resolution. (c) The vertical
prole 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.)
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
60
IC
70
LA
TIT
UD
E
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
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 dier
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 dierent 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 06001000 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 rst 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 16001700 magnetic time.
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 6768 during the morning, and a further
decrease after noon. The latitudinal distribution of absorption may be approximated by a Gaussian curve,
Am A0 exp
( 0 ) 2
2 2
(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 dened 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 scientically. Should they be very small,
their eect on HF propagation could be reduced by space-diversity reception. If,
on the other hand, they should blanket very large areas, it is dicult 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 eastwest 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 nding 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 eect on the general
5.5
5.59
48
5.5
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,

DecemberFebruary
49 days, JanuaryMarch
12 selected periods
3 days, summer
1 month, March
Data
Peak absorption
Peak absorption
Circular
Shape of
correlation
pattern
0.70 at 250 km NS
0.74 at 800 km EW
Fair agreement over 350 km NS
0.26 at 250 km NS (90 events)

0.41 at 800 km EW (35 events)
Elliptical
Circular
Elliptical
Can be similar over 380 km or dierent over

35 km, NS
High correlation over 10 km NS
0.5 at 380 km
Regions at least 200 km NS and

90 km EW
Day: 0.57 at 800 km NS
Night: 0.43 at 800 km NS
Results
2-min values if absorption 0.5 at 650 km NS and

&0.3 dB
700 km EW
Hourly values
Hourly values
Hourly values
Values correlated
Table 7.1. Spatial properties of Auroral Absorption (Hargreaves, 1969)
57
Ecklund and Hargreaves

(1968)
Bewersdor et al. (1968)
Hargreaves and Ecklund

(1968)
12 months
Slowly varying events,

four examples
17 months,
AugustJanuary
Disturbed nights
Night: 0.5 at 160 km

Day: 0.5 at 250 km
Hourly values if &0.3 dB
Circular
Circular
Day: 0.5 at 365 km EW and

170 or 300 kma NS
Little variation over 300400 km
EW
Elliptical or
circular
Night: 0.5 at 750 km EW and

155 or 465 kma NS
Hourly values if &0.3 dB
Elliptical
0.9 at 20 km NS
? &0.5 dB
Note:
Measurements poleward and equatorward of the central station, respectively.
5.5
Berkey (1968)
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 signicance in showing that, even when
ner structure is present, at least some part of the absorption extends for 200300
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 Siglufjordr, 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 signicance even if we are not
yet sure what that signicance 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 rst 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 Kilpisjrvi
6
9
Time (h)
12
30
20
10
Morning (0209 UT )
104 events
30
20
10
20
Night (1623 UT )
138 events
30
10
Kiruna
40
10
20
30
40
Q M
Morning (04 11 UT )
122 events
Night (1824 UT )
92 events
Siglufjordr
Figure 7.11. Durations of events starting in the night and morning sectors at Kiruna (L5.4) and Siglufjordr (L6.9).
Occurrence (%)
12
M: Median
Q: Quartile
Time (h)
356
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
Siglufjordr (L 6.9) (durations are given in hours)
Kiruna
Lower quartile
Median
Upper quartile
Siglufjordr
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
Figure 7.13. Maximum absorption, and distances in the X (westeast) and Y (southnorth)
directions of the location of the maximum from overhead at Kilpisjrvi, for the spike event
at 20462051 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 westeast, the poleward
progression is remarkably persistent. These features are typical of night-time spike events.
357
358
120
2050:13 UT
80
40
0
2047:45 UT
km
2049:13 UT
40
2047:13 UT
80
2047:06 UT
120
120
80
40
0
W
km
40
80
120
Figure 7.14. The spike event of Figure 7.13 at ve 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 o 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 Kilpisjrvi imaging riometer in Figure 7.13. This details the
spike event at 20462051 UT in Figure 7.3(b), and shows the value of the
maximum absorption and its location within the eld of view, all at 1-s resolution.
Note that the magnitude of the absorption varies with quasi-period 3060 s.
Eastwest 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
7.14 shows the position of the absorption patch at ve 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 eastwest extension is maintained.
Motions on the global scale

The onset also propagates eastward and westward from its rst 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 gure
of 4 min1 is for specic 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
2.74.5 km s1 for pre-midnight sudden onsets at L values of 5.26.1.
A simple model of the longitudinal movements at L7 is given in Figure 7.15.
From its rst appearance near midnight it takes (on average) about 20 min for an
onset to travel 5 h of local time and 3040 min to reach the morning sector. The
eastward and westward sections are not alternatives; observations have veried
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 eld (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 dierent 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 conrmed 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 0.77 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
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.
Figure 7.16. The progression of the onset of absorption projected onto the equatorial plane
(assuming a dipole eld). 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

A (dB)
DIXON ISLAND
NORILSK
1
3
1
3
KEVO
1
3
IVALO
1
4
2
SODANKYL
4
2
ROVANIEMI
3
OULU
JYVSKYL
BJRNYA
TROMS
ANDYA
ABISKO
1
1
1
1
1
3
1
3
THORSHAVN
GODTHB
1
1
NARSSARSSUAQ 1
SIGLUFJORDR
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 eld 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.
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 ve 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 rst 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 identied 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
17481805
Main event from north
1830
173639
Spike
Arc moving equatorward
1703
Level (dB)
2
4
6
16
17
18
19
20
Time (UT)
Figure 7.19. Night activity on 30 January 1995 observed with a wide-beam 38.2-MHz
riometer at Kilpisjrvi, 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 xed with respect to the rotating Earth for a long period. The
event shown in Figure 7.9 remained in the eld 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
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 825.
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 signicant, 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
K p.
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
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)
0
0
10
Ap
Kp
Figure 7.20. Eects 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
19901992 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.)
radio-communication circuit it also depends on the geometry (specically, 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 suer about 20 dB (Agy, 1970). Thus, absorption greater than 1 dB on a
30-MHz riometer is likely to be of practical signicance 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: rst, to specify from existing riometer data the statistics regarding the occurrence of absorption and the
eects 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 eects. 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 signicant 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 dierent 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 eect
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
Table 7.3. The FoppianoBradley formula for Q(1)

Total
Day component
Night component
Q(1) Q1d Q1s

Q1d Kd ddT dRddM
Q1s KsssT sRssM
magnetic latitude; T local 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/(22)]
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 (T27) 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)2/2T2]
dT exp[(TTm)2/(22T)]
Tm and T m local times of maxima for day and night components:
Tm 10(10.002R);
T m 0
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
dM 1 0.3 sin(3.86);
sm 1
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 Brices (1967)
drizzle and splash terminology.
369
the sunspot number. In a subsequent study of absorption over Finland during a

whole solar cycle (19721983), 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 signicance 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)
1
(log A log Am ) 2
exp
d(logA),
2
2 2
(7.4)
where A is the absorption in decibels, Am is the median absorption (logAm being

the mean of logA), and is the standard deviation of log A. The probabilility of
A being exceeded is then
Q(A)
f (log A)d(logA).
(7.5)
log A
Table 7.4. Parameters of Gaussian curves

tted to latitudinal variations of Q(1) in the
Finland sector
Ap
0 (degrees)
(degrees)
Q0 (%)
010
1015
1520
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
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-
371
olate the log-normal law below 0.1 or 0.2 dB. In most sets of riometer data these
small values are much aected by any error in the quiet-day curve, and also there
are theoretical reasons why the log-normal form cannot continue indenately
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 signicant 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 rst 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 signicance 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.
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 00000259 and
so forth.
Electron-density proles
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 own into absorption events
and thus calibrated the riometer in terms of the electron-density prole 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 prole during events,
and from one event to another. Figure 7.23 illustrates the changing electrondensity prole 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 proles
Since the electronneutral species collision frequency is known (for Equation
(3.95)), the absorption prole may be computed from the electron-density prole.
In most cases the computed and observed absorptions agree well enough to serve
as conrmation 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 proles 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 1520 km between points where the incremental absorption is half the
maximum. About 80% of the total absorption is produced in this slab. The specic absorption coecient increases downward, and the absorption peaks some
515 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 uxes

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 prole (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 prole
373
(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 proles 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).)
375
Figure 7.23. Electron-density proles 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
matches the observed one. An essential element is the eective recombination

coecient as a function of altitude (Section 1.3.3), which relates the rate of
electronion 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 proles 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 prole peaks at a lower altitude
than does the night-time spectrum.
Some direct comparisons have also been made. Using particle uxes 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)
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
(cm3)
7
6
5
Log (Flux)
376
4
3
2
1
(a)
50
100
150
(b)
200
250
Energy (keV)
Figure 7.24. Electron-density and absorption proles 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 proles 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 ux of the night event between about 40 and 80 keV, whereas the night event has
a greater ux of softer (25 keV) particles. The daytime absorption peaks at 8788 km, the
night absorption some 5 km higher.
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 ux 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 signicantly correlated to the energy ux over some
energy ranges of electrons detected at geosynchronous orbit (Figure 7.25). Here
the ux 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 inux in the bands 4080 keV and 80160 keV. As might be
expected, the rate of production calculated from the particle ux also correlated
well at some heights. Schematic production-rate proles 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 reects 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 eld, in which case the value of the eld can be estimated from the
relation (Ranta et al., 1981)
dt
E(mV m ) 5.88
10
d(1/L2 )
1
4
1
(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 eld gives a median value of
1.3 mV m1. That the deduced eld is independent of L supports the hypothesis,
but the procedure has yet to be veried by direct comparisons with the electric
eld measured by other means. A later study using data from the Scandinavian
sector revealed speeds mostly in the range 0300 m s1 with a peak at 100200 m
200
175
175
160320 keV
ENERGY FLUX (10 keV cm
2
125
100
075
050
025
0
125
100
075
050
025
0
3
4
5
ABSORPTION (dB)
1
sr )
1
sr )
3
4
5
ABSORPTION (dB)
4080 keV
2040 keV
2
2
1
1
80160 keV
150
2
150
1
1
sr )
200
1
1
sr )
378
0
0
3
4
5
ABSORPTION (dB)
3
4
5
ABSORPTION (dB)
3
4
5
ABSORPTION (dB)
20320 keV
2
1
1
sr )
10
0
0
Figure 7.25. Relations between energy ux 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 40160 keV. (Reprinted from P. N. Collis et al.,
J. Atmos Terr. Phys. 46, 21, copyright 1984, with permission from Elsevier Science.)
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 proles 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 eld 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 signicantly
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 eld-line. In the rst 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
380
(a)
BI
70
(degrees)
FY
65
C
P
SM
A
60
0.5
1.0
V (km s1)
1.5
2.0
(b)
9
BI
8
L-value
7
FY
6
C
5
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 eld on the
assumption that the motion is E
B drift. The estimated eld is independent of L.
(Reprinted from J. K. Hargreaves et al., Planet. Space Sci. 23, 905, copyright 1975, with
the other. The night-time events, particularly at the higher latitudes, exhibit time
dierences between the peaks of events in conjugate regions, the event which
appears rst 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
eld carried by the solar wind (Figure 7.29).
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 dierence 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 eld deduced from
the equatorward drift of absorption arcs (Ranta et al., 1918)
Date
UT
E(mV m1)
27 March 1975
2 May 1975
14301530
13001500
17201900
17001800
16001730
19302030
19002000
18001900
20002100
19002000
18001930
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
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 eld 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 northsouth 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 proles observed
during pulsations, it appears that the modulation involves the energy of the particles, not just their ux (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 are 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 eects 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 ux of energetic
protons released from the Sun at the time of the are.
As with auroral absorption, studies based on the occurrence of the blackout
condition have their limitations. Most scientic studies of the phenomenon have
therefore made use of riometers, which give a quantitative measure of the absorption. These showed that the eects were conned 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 eects, 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 rst recognized GLE occurred on 28 February
1942; it was identied in retrospect, of course, since the PCA was not yet a known
phenomenon. The are responsible for that event has another claim to fame as the
source of the rst solar radio noise to be recorded at the Earth.
Since the early 1960s it has been possible to observe solar protons in space, and
384
the monitoring of energetic protons from satellites has now become a matter of
routine. As might be expected, the satellite monitors nd some events that are not
seen by ground-based methods. In fact, most solar ares emit protons at the lower
energies that is, up to 10 MeV. At energies of several tens of mega-electron volts
the ux of protons reaching the Earths 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 (11000 MeV) protons
emitted from the Sun, usually during a solar are. 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 19621972, which included the end of solar cycle 19 and the
rst 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 196265.
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 12108 h.
Those occurring within the narrow range 120132 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 14
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 ux of protons with energy at least 10 MeV. The plot covers the years
19761989, which included solar cycle 21 and the beginning of cycle 22. The
general inuence of the sunspot cycle is seen again, except that there is a remarkable dearth of events near the peak of the cycle in 19791980. This looks like an
extreme case of a well-reported eect. 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.
(a)
385
16
Number of events per year
12
120
10
100
80
60
40
20
Sunspot number
Absorption 5 dB
14
0
62
63
64
65
66
67
68
69
70
71
72
Year (19 )
Duration of 1 dB events, 196272

(b)
7
May exceed
stated value
Number of events
0
0
24
48
72
96
Duration (hrs)
120
144
168
>168
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 19621972. (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, SolarTerrestrial Physics and Meteorology: SCOSTEP Working Document II, 1997;
SCOSTEP Working Document III, 1979.)
(a)
24
Flux >10 MeV (cm 2 s1 str 1)

<100
1001000
>1000
22
18
16
160
14
140
12
120
10
100
80
60
40
20
76
77
78
79
80
81
82
83
84
85
86
87
88
89
Sunspot number
Number of events per year
20
Year (19 )
10
9
8
7
30 MHz absorption at Kilpisjrvi 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 s1 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 ux at synchronous orbit. The sunspot numbers are shown as the 12month running mean. The data cover the years 19761989. (b) The relation between 30MHz absorption at Kilpisjrvi and the ux 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).)
387
The direct connection between proton ux and PCA is conrmed by Figure

7.32(b), which plots the absorption at Kilpisjrvi, Finland, against the proton ux
detected on a satellite in synchronous orbit. The straight lines indicate the law
absorption (ux)1/2,
(7.11)
which is to be expected if the electron-production rate is proportional to the particle ux. Note that uxes greater than 100 cm2 s1 sr1 are likely to produce a
signicant PCA. Since Kilpisjrvi 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 ux from the
radio absorption (Smart and Shea, 1989) is
J10A2,
(7.12)
where J is the ux (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
ux 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 ares 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 signicantly aected 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 dicult 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 eect: 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.
(a)
1.0
Relative occurrence
388
0.5
M A M
S O
N D
(b)
10
Delay time (h)
Month
Figure 7.33. Seasonal

eect in PCA: (a) the
fraction of ares 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
A S O N D
Weak events
It has also been argued that the eect may be articial 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 eects 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 signicant seasonal
eect. 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 eects 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
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 dened 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 19621972.
Note that, of the events of magnitude &0.5 dB, about half reach 1 dB, about one
fth 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 ares and radio emissions
In fact, not all large ares give rise to proton events and there are some proton
events that have not been associated with any known are. 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
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
390
events are associated with the larger solar ares. Those ares that produce protons
are often called proton ares and they are recognized as a distinct class in the are
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
ares emit protons. The type-IV emission is a radio burst of long duration that
follows some ares and covers a wide band of radio frequency. (It is attributed to
synchrotron radiation from high-energy particles gyrating in the solar magnetic
eld.) 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 ux 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 are responsible and for timing the ight 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 protons journey to Earth
The production and release of energetic protons appears to be a normal part of

the solar-are phenomenon, and ares causing PCA and GLE at the Earth probably dier 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 aected by the form of the interplanetary magnetic eld. As illustrated in Figure
2.3, the eld has a spiral form due to the rotation of the Sun, and this aects the
propagation of solar protons despite the weakness of the eld and the high energy
of the protons. The gyroradius of a 1-GeV proton in a eld 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
Absorption (dB)
(a)
Thule, Greenland
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)
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 are
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 are 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 are 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 ares that are out of sight
around the western limb. This is illustrated by Figure 7.35, which gives the
positions of solar ares associated with those proton events energetic
enough to be detected at ground level (i.e. GLEs). (Note that the western
392
Figure 7.35. Solar longitudes of ares 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 ares with
solar longitude is signicantly biased towards one side of the central
meridian. The heliolongitudinal distribution of the source ares broadens
for protons of lower energy (Smart and Shea, 1995).
(b)
The time delay between a are and the related PCA increases with the
eastern longitude of the are.
(c)
The delay between are 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 ares 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 eld to reach the atmosphere. To a rst approximation this problem may be
treated by Strmer theory. The theory describing the trajectories of charged particles in a dipolar magnetic eld, which C. Strmer 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 eld a charged particle tends to follow a spiral path whose radius
393
of curvature (rB mv/(Be)) is directly proportional to its velocity and inversely

proportional to the magnetic ux density. Because solar protons are of relatively
high energy, the magnetic eld changes signicantly 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 eld undergo the
least deviation. To reach the equator, which is the least accessible region, a proton
has to cross eld-lines all the way down to the atmosphere. Charged particles may
do this only if they are suciently energetic, and the equatorial region is eectively
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 eld 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 eld.
Although the trajectory of a proton in the geomagnetic eld can be very complicated, Strmers analysis simplied matters by dening allowed and forbidden regions that could and could not, respectively, be reached by a charged
particle approaching the Earth from innity. To reach magnetic latitude c in a
dipole eld, the rigidity of the particle must exceed a cuto 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 Strmer
cuto latitude against energy both for protons and for electrons.
To perform an exact calculation of the trajectory of a proton through the geomagnetic eld, 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 conrm 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
394
(a)
(b)
Figure 7.36. (a) The

Strmer cuto latitude
for protons and electrons. (S.-I. Akasofu and
S. Chapman,
SolarTerrestrial
Physics. Oxford
University Press, 1972,
by permission of Oxford
University Press. After T.
Obayashi, Rep.
Ionosphere Space Res.
Japan, 13, 201, 1959.) (b)
Cuto latitudes for
dipole and realistic geomagnetic elds (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 Strmer theory these protons should have energies exceeding
400 MeV, but direct observations of the particles have shown that the cuto rigidity at the edge of the polar cap is signicantly less than the Strmer 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 cuto reverts fairly abruptly to the Strmer value. Much of this eect (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
route even for protons of low energy. Figure 7.36(b) shows the dierence in cuto
energy between dipolar and more realistic geomagnetic elds. The cuto 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 aected 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 rst
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 eect 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 rst 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 610 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 rst two stations
on the rst 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 dierent and will
be considered in Section 7.3.6.)
395
396
Figure 7.37. The polar areas normally aected 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 aected 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.)
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 eect was found slightly before
local noon, covering a range of magnetic latitude approximately from 60 to 70
(Figure 7.39). The eect seems to occur simultaneously and with the same magnitude in magnetically conjugate regions, and it is conrmed that the eect 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 cuto, and the development of anisotropy in the pitch-angle distribution of the
incoming protons. More recent studies have suggested that both eects may occur.
There is evidence that a change of cuto 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 6570.
397
398
Figure 7.39. The region aected 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 eective recombination coecient allows one to calculate the resulting
(a)
(b)
399
Figure 7.40. (a) Proton

uxes measured by the
geosynchronous satellite
GOES-5 on 16 February
1984, tted by spectra of
form E. (Data from F.
C. Cowley, NOAA,
Boulder, Colorado,
private communication.)
(b) Electron-density proles 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
electron density. If an event contains particles of energy 1100 MeV, the eects
should appear within the height range 3590 km (Figure 2.28). Eects due to the
higher energies tend to be smaller because the ux 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 coefcient

In fact, the recombination coecient in the lower ionosphere is not a wellestablished quantity, and one use of PCA events is to measure the recombination
coecient 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 proles 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 eective recombination coecient 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 dierence between day and
night, and also between dierent 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 dierences will be considered in the next section.
Daynight variation and twilight effects

Because the proton inux during a PCA decays relatively slowly, the eects of
daily variations in the complex chemistry of the region may also be detected. The
most obvious eect is a large diurnal variation in the absorption, which is typically about ve 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 eect 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 uxes 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 daynight modulation is without doubt a variation in the ratio
of the concentrations of electrons and negative ions, (dened 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. Eective recombination coecients 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)
402
10
Absorption (dB)
Mirnyy
Spitsbergen
Magnetic disturbance
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 1216 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 eect is conned 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 eect 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 eect is present in the proles of Figure 7.43, where
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. Eective recombination coecients 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.
the smaller recombination coecients at given zenith angle at sunset imply

greater electron densities. Collis and Rietveld (1990) showed that the
timing of the daynight transition depends also on the altitude, and suggested by way of explanation that dierent 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 eective recombination coecient 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
eective recombination coecient 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,
405
Assuming that the daynight 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 eective recombination
coecient e q/N e2 by denition 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

Inuxes of energetic particles have another important eect 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 5467 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)
OH O3 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 rst reaction of the pair but regenerated in the second. These processes require a sucient
concentration of water vapor and therefore they are conned to the region below
the mesopause. They are thought to be important over the height range 5090 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
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 eect 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:
NOO3 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 eect on the ozone concentration
in the stratosphere, which fell by 15%20% at latitudes 7580. In the event of
July 1982 ozone was depleted between 55 and 85 km. This was a relatively soft
event, which explains why the eects were higher up. A series of PCA events that
occurred in 1989 is also thought to have aected the ozone content. A computation of the eect 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 eects continued for a year or more. Signicant though these
eects are, they have no known eect 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 dierent 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 rst 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 (Rttger et al., 1990). Other observations cover the range 2.27 MHz to 1.29 GHz (Rttger, 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 dierent 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 8486 km (Figure 7.47), an altitude close to the mesopause. When the echoes are present their height uctuates (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 sucient proof that quite
dierent mechanisms are responsible.
407
408
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. Rttger, 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
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 uctuations 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 diusion coecient 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 Rttger (1994).
7.5
At middle and equatorial latitudes D-region absorption has only a minor eect
on HF propagation, but at high latitude it can aect the signal strength profoundly. There are two basic types at high latitude, each having a separate cause
and morphology. In its eect on radiowave propagation, auroral absorption (AA)
409
410
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 tted 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. Rttger. J. Atmos. Terr. Phys. 52, 569, copyright
is of rst-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 eastwest. 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 suciently 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 nders)
require information on the ner 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-
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 signicant 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 11000-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 aect the lower frequencies more than they do the higher
ones because the absorption varies (to a rst approximation) as f 2. At ELF and
VLF, propagating in the waveguide mode, an increase in precipitation causes signicant variation in the dimensions of the waveguide and thereby produces both
amplitude and phase changes in the received signals.
7.6
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.
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 (196465) and IASY (1969). Planet. Space Sci.
22, 255.
Bewersdor, 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
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 rst 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. Dierences between conjugate regions. Planet. Space
Sci. 15, 1585.
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 19761979 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 Kilpisjrvi. J. Atmos. SolarTerrestrial
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.
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., Schimacher, 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
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 ux
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

Press, Oxford.
Bailey, D. K. (1959) Abnormal ionization in the lower ionosphere associated with
cosmic-ray ux 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 ares and nonproton ares. 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 1921 March 1990: recombination coecients, the twilight transition and
the midday recovery. J. Atmos. Terr. Phys. 55, 857.
Hultqvist, B. (1969) Polar cap absorption and ground level eects. 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) Inuence of extremely large
proton events in a changing stratosphere. J. Geophys. Res. 105, 11659.
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 19811989.
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 eects 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 inuence of the geomagnetic tail on lowenergy cosmic-ray cutos. J. Geophys. Res. 72, 197.
Sauer, H. H. (1968) Nonconjugate aspects of recent polar cap absorption events.
Shea, M. A. and Smart, D. F. (1977) Signicant solar proton events, 19551969. In
SolarTerrestrial Physics and Meterology: Working Document II, p. 119. SCOSTEP.
Shea, M. A. and Smart, D. F. (1979) Signicant solar proton events, 19701972. In
SolarTerrestrial Physics and Meterology: Working Document III, p. 109. SCOSTEP.
Shea, M. A. and Smart, D. F. (1995) Solar proton uxes 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-ares
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 1921, 1990: a case study. J. Atmos.
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 Rttger, 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
observed in the summer and winter polar mesosphere over Andya, Norway.
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 Rttger, 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 Rttger, J. (1988) The eect 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.
Rttger, J. (1994) Polar mesosphere summer echoes: dynamics and aeronomy of the
mesosphere. Adv. Space Res. 14, 123.
Rttger, 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 proles. 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 dierent from propagation at middle and low
latitudes. This is primarily due to the fact that the magnetic eld-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 eld 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
ux), time of day, season, path-length, and orientation. Further renement 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 eld strength has not turned out to be quantitatively accurate because
of the diculty 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 eects, and
they are either qualitative or have not been adequately validated to inspire users
condence.
The eects of the polar cap and the auroral-oval ionosphere on signals of
various frequencies dier 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 eects 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 19301940 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 Eects 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) (19571959), 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 eects
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 eects 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. 19481991), 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 classied,
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 dier 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 eects of these variations on specic 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 Earthionosphere waveguide mode (Watt, 1967; Wait, 1970;
Davies, 1970; Davies, 1990, Ch. 10) or the wave-hop (Berry, 1964) mode. The
eectiveness of the waveguide mode depends upon the long- and short-term
variations in conductivity of the Earths surface and the lower ionosphere (D
region).
VLF propagation, in general, is characterized by relatively low path attenuation (23 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 20 000 km are realized; however, atmospheric
noise levels are high thus decreasing the signal-to-noise ratio (SNR), and the
signal bandwidths are low (20150 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 dicult to erect adequate
antennas, so practical antennas have radiation eciencies 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 (1014 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 ELFVLFLF 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
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 inuence VLF propagation (Wait, 1991).
The lowest frequencies used for communication purposes are the ELF band
(3300 Hz), which is used primarily for very-low-data-rate communication to
submerged submarines. One operational transmitter is located at the US Navys
Wisconsin Test Facility and was described in the military literature as the
Project Sanguine/Seafarer/ELF. The eective radiated power (ERP) is 0.25 W
in the 4050-Hz band and 0.5 W in the 7080-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 eectively change the waveguide characteristics (Wait, 1970; Hunsucker, 1992). Fraser-Smith and Bannister (1998)
recently measured ELF transmissions from a heretofore-unknown source, which
they identied 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.
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
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 eld 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 eects 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
Figure 8.4. Measured and theoretical values of the KPTF 82-Hz signal strength versus
range ( 0). CO, Connecticut; KB, Kings Bay, Georgia; SS, Sndrestrmfjord; 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 eect on VLF polar transmissions, and one of the
rst documentations of these events was presented by Bates (1962), who described
the eects 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 are 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 eective height of
the VLF waveguide over the polar cap dropped to about 5 km below the normal
423
424
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 Navys OMEGA VLF network depend
upon phase dierences 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 ares 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 aected by the boundary of the PCA, as was the ELF transmission noted previously.
Some typical navigation-location errors measured during the SPE of 69
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-
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).
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 1961Dec 1963

October 19611964
November 19621964
April 19621964
April 1963December 1963
January 1964December 1964
Figure 8.11. VLF propagation paths to College, Alaska during the 19611964 study (after Bates and Albee, 1966).
8.3 LF and MF propagation
HYPERBOLIC
NORWAYHAWAII)
2
429
RANGING
(NORWAY)
30s = 3 nm HYPERBOLIC
6 nm RANGING
LOCATION ERROR (nm)
0
CORRECTED
1
2
3
4
5
6
ACTUAL
7
8
9
0
12 18 24
6 MAR
12 18 24
7 MAR
12 18 24
8 MAR
6 12 18 24
9 MAR
12 18 24
10 MAR
1970
Figure 8.12. OMEGA location errors in nautical miles during the SPE of 69 March 1970
in Norway (from Larsen, 1979).
tion 10 dB) is dicult to nd, 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 eects. 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
line-of-position system, which was primarily groundwave propagation but also

was sometimes aected by skywave. Another primarily groundwave navigational
system (which is currently being phased out) was the Non-Directional-Beacon
(NDB) system operating in the 250450-kHz band, but again there do not seem
to be documented examples of high-latitude ionospheric eects . . .
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 xed services operate using the skywave mode but reliable propagation is dicult because of D-region absorption, which is particularly high at
auroral and polar latitudes. As an example, most of the xed-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 ve-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 modied 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 identication of individual stations was achieved. The three dierent 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
(89 February, 1986).
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 820
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) eects. To illustrate the
eects 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).
Specically, Figure 8.14 shows typical variations in signal for quiet magnetic conditions. The equinoctial recovery from the summer low eld 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 12 equatorward of the
432
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
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 aects all paths monitored. The channel-3
(Anchorage, Alaska) path lies entirely inside the auroral oval and its extreme variability is probably due to intense patchesof 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 eects. The KFAX, San Francisco path (channel 6) is roughly
perpendicular to the auroral oval and its ionospheric reection points are mainly
equatorward of the oval, so it is aected less than are channels 35. 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
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 Q0. (after
Hunsucker, 1988).
433
434
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, Q 4 (from Hunsucker, 1988).
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 o the air (08301330 UT), signals from an unknown
AM station were intermittently received.
The sunspot cycle also exerts profound eects 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 eects 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.
Eects 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
eects. The magnetic storm of 89 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
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
KTWO
Casper,
Wyoming
1030
50
D3553 km, long path.

One ionospheric
reection point in the
auroral oval during
moderately disturbed
conditions.
KFAX
San Francisco,
California
1100
50
D 3464 km, long path.

One ionospheric
reection point in the
auroral oval during
disturbed conditions.
KGA
Spokane,
Washington
1510
50
D 2640 km. Similar to

the KTWO path.
Short northsouth
auroral path.
Table 8.4. Sunspot-cycle eects 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
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 19811985
437
most pronounced eects 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 EdmontonFairbanks 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 eld-strength measurements collected at Fairbanks have been
compared with eld 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 ve 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 eld strengths for 1987

Median eld 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)
Modied 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 89
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 (330 MHz) is basically a skywave band day and night and is
used for broadcasting, point-to-point, and surveillance (actually, the range of
230 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

xed 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 rst 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
test progressed, modications 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 identied 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 0.522 MHz was put into eect
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 fth 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 JanuaryFebruary 1958 tests, with encouraging
results.
Specically, 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 modied 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 CollegeKiruna 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 10001900-km range
was interpreted as the rst hop of a three-hop mode. Similarly, groundscatter in
this direction in the 20003000-km range was considered as the rst hop of a twohop mode.
441
442
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 0718 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 1215 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 dened 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 groundscattersignal-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 eects 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 eorts
of the University of Alaska Geophysical Institute, the Kiruna Geophysical
444
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 rst 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
45 kW, using the rotating Yagi-antenna system.
The CollegeKiruna 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 reection points. The CollegeBoston great circle path is 5300 km long
and crosses the auroral zone tangentially. The CollegeStanford 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 rst 13 days of August 1959 were characterized by low solar activity and
magnetically quiet conditions, in particular the days of 1114 August. On 14
August and again on 18 August there occurred major solar ares in an active
region that crossed the solar central meridian on the 16th. Many lesser ares were
observed in this region during its passage over the Suns disk. The two major ares
were both followed by geomagnetic storms that together account for the 5 days of
the month designated as magnetically disturbed. The second are was also accompanied by cosmic-ray emission, causing a weak (1-dB) PCA.
The rst major are, 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 are,
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 conrmed the ndings of Basler, namely
that AA peaks a few degrees equatorward of the visual auroral oval, which nding
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 Vestines isochasms (after Owren et al., 1963).
The second major are, 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 solarterrestrial 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. Solarterrestrial conditions and transpolar HF radio propagation for near-sunspotmaximum conditions on 15 and 18 August 1959 (from Owren et al., 1963).
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 1617 August 1959 is
illustrative of near-sunspot-maximum conditions.
8.4.2
Tests involving transmission between Alaska and the

continental USA
The CollegeStanford 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 dierence were made. The nal recovery took
place on 18 August at 0900 UT. There were no 18-MHz data due to interference
during this period.
The CollegeBoston 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 715 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 CollegeStanford 12-MHz blackout, as the ionospheric disturbances
spread to the normal auroral-zone regions.
449
450
24
EASTERN STANFORD TIME
20
16
12
08
04
00
1
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 1315 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 suered a complete black-out during
the 1617 August storm, as might have been expected.
8.4.3
Other trans-polar HF experiments on xed

frequencies
The CollegeKiruna circuit (trans-polar, D5300 km)
Both the 12- and the 18-MHz signals were received at Kiruna most of the time
during 112 August, with very little outage during 1315 August. This great-circle
path passes over west Spitzbergen and the ionosonde observations at
Longyearbyen show that, during 115 August, the 12-MHz signal was nearly the
optimum trans-polar trac 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 1617 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 19581959) exhibited
unusually favorable propagation conditions on the propagation path between
College and Kiruna. The HF pulse reception at Kiruna showed, as a rule, three
dierent propagation modes on 18 MHz, with intervals of 3 and 67 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
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.
CollegeKjeller and ThuleKjeller propagation-path analysis

(SSN38.380.2)
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 ThuleIsfjord 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
CollegeKjeller,
ThuleKjeller,
CollegeIsfjord,
ThuleIsfjord,
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 reection height of 300 km, is shown in Figures 8.298.31, 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 CollegeKjeller transmissions
were found to be vulnerable to AA at the transmitter end of the circuit, whereas
the auroraloval absorption region at the receiver end should leave all conventionally propagated signals virtually unaected. Of course, during disturbed conditions the auroral oval expands considerably and could seriously inuence circuits
otherwise not exposed to AA. On the CollegeKjeller 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 rst type gives, on a daily basis, the total number of hours with
Figure 8.29. CollegeKjeller idealized mode geometry (from Owren et al., 1963).
Figure 8.30. KjellerCollege idealized mode geometry (from Owren et al., 1963).
Figure 8.31. KjellerThule idealized mode geometry (from Owren et al., 1963).
456
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 o 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 (JulyAugust 1961) for SSN 50 and winter (October
December, 1961) on the cis-polar ThuleKjeller circuit for 12 and 18 MHz.
Similarly, Figures 8.34 and 8.35 illustrate the behavior of the trans-polar
CollegeKjeller HF circuit for SSN 50 in summer (JulySeptember 1961) and
winter (OctoberDecember 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, 810 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 dening 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 (SSN50). These seasonal statistics
are for the 18-MHz circuit and the College 12- and 18-MHz circuit. Figures
8.368.38 show the seasonal variations on these circuits.
From April 1961 until June 1962 (SSN38.364.3) 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 ThuleCollege 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 24-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 0721 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 denite conclusions about sunspot-cycle eects on this propagation path
8.4.4
CollegeKiruna absorption studies at xed 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. ThuleKjeller 18-MHz (a) and 12-MHz (b) propagation conditions for summer 1961 (from Owren et al., 1963).
HOURS
12
16
20
24
12
16
20
24
(b)
10
Q
10
e e e e e e e e e e e
(a)
20
15
20
25
25
signals present
15
QQQQ
OCTOBER
DDDD
30
30
DD
10
DD
20
10
20
NOVEMBER
15
30
Q DD
25
30
equipment failure
25
QQQ
e e e e e e e e e e e e e e e e
15
Q
noisy period
5
D
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
sudden commencement
15
Figure 8.33. ThuleKjeller 18-MHz (a) and 12-MHz (b) propagation conditions for winter 1961 (from Owren et al., 1963).
HOURS
HOURS
12
16
20
12
16
20
(b)
(a)
20
JULY
20
15
signals present
QQDD
15
e e e e
e e e e e
10
10
e e e e e
5
D
e e e e e
25
25
30
30
QQQ
e e
10
Q Q D
15
15
AUGUST
10
noisy period
20
20
25
e e e
30
DDD
e e
25
30
equipment failure
QQ
QQ
15
Q
20
Q
25
QDD
sudden commencement
10
30
D
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
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. CollegeKjeller 18-MHz (a) and 12-MHz (b) propagation conditions for late summer 1961 (from Owren et al., 1963).
HOURS
HOURS
12
16
20
24
12
16
20
24
(b)
10
Q
10
(a)
15
25
20
25
signals present
15
20
QQQQ
OCTOBER
DDDD
e e
30
30
5
D
15
Q
DD
20
10
20
NOVEMBER
15
30
Q
DD
25
30
equipment failure
25
QQQ
noisy period
DD
10
10
10
20
QQQ
25
Q
30
D
e e e e e e e e e e e e e e e
15
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. CollegeKjeller 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
00
03
06
09
12
15
18
21
24
HOURS, UT
Figure 8.36. Seasonal signal-strength variation on the CollegeKjeller 18-MHz propagation path (from Owren et al., 1961).
60
JAN.FEB. 1961
MAR.APR. 1961
JUNE 1961
OCT. 1961
50
40
30
20
10
00
03
06
09
12
15
18
21
24
HOURS, UT
Figure 8.37. Seasonal signal-strength variation on the ThuleKjeller 18-MHz propagation

path (from Owren et al., 1963).
464
JAN. 1961
MAR.APR. 1961
JUNE 1961
OCT.NOV. 1961
60
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 CollegeKjeller12-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 ThuleCollege 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
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 ares
of importance 2 to 3 were observed. Absorption events associated with these
ares 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
ares 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 CollegeKiruna 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 attening o 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 ares 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 CollegeKiruna propagation path are shown in
Figure 8.44. The rst 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 CollegeKiruna
propagation path; consequently one might expect better peak-to-peak correlation
for the strong events.
Figure 8.43. Eects of large PCAs of 923 July 1959 on 12- and 18-MHz transpolar transmissions (from Owren et al., 1961).
8.4 HF propagation
Figure 8.44. Weak PCA eects on the CollegeKiruna 18-MHz circuit, measured using the
College oblique riometer (from Owren et al., 1963).
471
472
Figure 8.45. Eects of a strong AA event on 11 September 1961 on the 18-MHz

ThuleCollege 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 ThuleCollege

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 03300730 and 08301100 UT, when strong
interference made identication of the Thule pulsed signal doubtful. A complete
black-out of the 18-MHz signal occurred between 11402400 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 ve
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.
ThuleCollege path
The ThuleCollege 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
464 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
ThuleCollege HF propagation path. The longpulse record is shown
above and the shortpulse record is shown
below (after Bates and
Hunsucker, 1964).
modes or combination EF 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 ThuleCollege circuit is shown
in Figure 8.48 (lower plot). The histogram peaks around 20002400 UT (1014,
150 WMT and has a minimum around 11001400 UT (0104, 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
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 (21300230, 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
ThuleCollege 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 O-path or nongreat-circle (NGC) mode. Two examples of these modes on the ThuleCollege
477
478
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 AndyaCollege path (D5000 km)

The most prominent winter-night mode on the AndyaCollege 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 AndyaCollege 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 Andya (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 Andya circuit as opposed to the single maximum on the
Thule path. Average MOF maxima occur at the times 06000900 and 19002100
UT (20002300 and 20002300, 150 WMT).
8.4.7
Other results from HF high-latitude studies from

c. 19561969
Most of these results are taken from a survey paper by Hunsucker and Bates
(1969).
479
480
Figure 8.52. Direct and o-path modes from Thule long-pulse records. In the lower ionogram, the maximum o-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 xed-frequency backscatter soundings to the north revealed that AE
activity peaks during the period 18000600 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 21000600 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 8.508.52.
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 cuto 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 ndings concerning the occurrence and behavior of auroral-oval E-region ionization
during winter-night sunspot-minimum conditions. The MOF peak on the
ThuleCollege circuit coincides with the period of maximum auroral activity near
the College end of the circuit (21300230, 150 WMT). AE propagation on the
AndyaCollege 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
eects twice, compared with once on the ThuleCollege 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 reection points on the ThuleCollege and Andya
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 CollegeThule circuit.
481
482
Figure 8.53. A schematic representation of the requirement for overlap between the elliptical locus of possible o-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-nding 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 o-path data showed that the locus of opath 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 196312
February 1964, the average percentage occurrence of AE on the ThuleCollege
and AndyaCollege circuits was over 50%. AE MOFs as high as 46 MHz with
typical values of 18 MHz were observed on the ThuleCollege path and MOFs as
high as 32 MHz with typical values of 16 MHz were observed on the
AndyaCollege path. This suggests that long-haul HF trac 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 eects
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
modied 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 reections from the F2 layer, with ground reection 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 rst number
gives the mode order and the rest of the symbols denote the ionospheric layer
involved. Propagation modes involving successive reections both from the E
layer and from the F layer are termed combination EF 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 dened,
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
rst four F layer modes. Traces such as those in Figures 8.54(c) and (d) are termed
long-tailed traces.
Figures 8.558.58 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
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 EF 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 rst 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 eects

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 rst 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 rst assume that these DLT traces were produced by higher-order
F2-mode signals. Rough estimates of the travel time, and hence reection heights,
can be made by assuming that the rst-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 rst-arriving and the DLT signals were approx-
489
490
(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-reection heights for each hop

mode
Virtual-reection 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 BreitTuve
theorems (and accepting the small errors due to the spherical geometry), these
data correspond to virtual reection 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-cuto 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 reection 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 reected by
the layer; hence, if the F1 and F2 critical frequencies diered 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 reection. 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 Earthionosphere 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
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 dened 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 modied 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 1018MHz range propagated during most of the day with no great variation in ight time.
On the Fort Monmouth to College path, however, the minimum-ight-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 eect 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 eects 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 utter, 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
Figure 8.62. A map showing the SodankylLindau 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 LindauSodankyl HF propagation path (after
Moller, 1964).
Figure 8.64. A partial cross-section through the SodankylLindau path, showing a

quarter of the 2F mode (after Rose, 1967).
sounding circuits was the 1965-km path between Sodankyl, 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 SodankylLindau propagation path, the location of the Arctic circle, the auroral zone, and the location of
supporting instrumentation at Kemi, Lule, Uppsala, Kiruna, and Lycksele. A
vertical section through the LindauSodankyl path is shown in Figure 8.63, with
a possible mode structure indicated.
It is obvious that the Sodankyl end of the path was the most aected by
the auroral ionosphere and the geometry of the Sodankyl 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 SodankylLindau circuit.
The system at Sodankyl 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
Table 8.12. HF-system parameters

Transmitter output power 50 kW
Pulse duration 100 s
Pulse-repetition frequency 50 Hz
Frequency range 1.422.6 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 SodankylLindau HF circuit for a
summer evening (30 June 1958) during sunspot cycle 19 (SSN87; Kp 4) (after Moller,
1964).
SodankylLindau 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 Sodankyl end of the circuit and thus indicated the presence of
AE ionization which is reected 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 Mollers (1964) report
and representative forward ionograms are shown in Figures 8.688.72. The eects
8.4 HF propagation
Figure 8.66. An oblique ionogram for the SodankylLindau 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. Eects of auroral-E ionization on mode structure on the SodankylLindau

HF circuit at local midnight, midwinter (8 November 1958), sunspot-cycle maximum (after
Moller, 1964).
497
Figure 8.68. Normal summer daytime SodankylLindau 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 SodankylLindau ionograms for 1718 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 eects near midnight during sunspot maximum, 78 November
1958 (after Moller, 1964).
8.4 HF propagation
Figure 8.71. Eects of dense auroral-E ionization on 4 September 1958 on the

SodankylLindau HF circuit (after Moller, 1964).
Figure 8.72. Some eects of a moderate geomagnetic disturbance on 7 October 1958 on the
SodankylLindau HF circuit (after Moller, 1964).
501
502
Figure 8.73. The AndyaCollege 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 SodankylLindau HF circuit are

graphically illustrated in Figure 8.70, eects of AE on the oblique circuit are
shown in Figure 8.71 and moderate geomagnetic eects in Figure 8.72.
Another analysis of data from the 464 MHz forward sounding path from
Andya, 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 dierent 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 AndyaCollege HF propagation spectrum for 1964 equinoxes (after
Bartholomew, 1969).
Figure 8.75. AndyaCollege HH propagation for Winter 1964. (after Bartholomew, 1969).
503
504
Figure 8.76. AndyaCollege HF multipath and propagation outage for summer 1964
(after Bartholomew, 1969).
Figure 8.77. AndyaCollege HF multipath and propagation outage for 1964 equinoxes
(after Bartholomew, 1969).
summer periods, there was signicant 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 signicant factor on the
AndyaCollege 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 gures also display the percentages of actual
8.4 HF propagation
Figure 8.78. AndyaCollege 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 AndyaCollege 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 o-path (NGC) propagation on the
AndyaCollege HF path is displayed in Figure 8.80, showing seasonal, quiet/
disturbed conditions and diurnal eects.
A good example of how actual LUF and MUF values dier 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
dened as the optimum trac 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 Andya 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 AndyaCollege
and AndyaFt 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
505
506
(a)
(b)
Figure 8.79. The AndyaCollege HF mode conguration for March 1964; (a) quiet days
and (b) disturbed days (Bartholomew, 1969).
8.4 HF propagation
Figure 8.80. Diurnal and seasonal behavior of AndyaCollege 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 AndyaCollege HF circuit (after Bartholomew, 1969).
507
508
Figure 8.82. Predicted MUF, LUF, and percentage occurrence of observed signal for April
1964 on AndyaCollege HF circuit (after Bartholomew, 1969).
Figure 8.83. Predicted MUF, LUF, and percentage occurrence of observedsignal for June
1964 on AndyaCollege 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 AndyaCollege circuit (after Bartholomew, 1969).
Figure 8.85. Location of trans-polar paths investigated by the NDRE (from Folkestad,
1968).
509
510
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 CollegeAndya (right-hand plots) and Fort Monmouth Andya (left-hand plots). At
the bottom of the gure, 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
Figure 8.88. Circuit behavior during disturbed periods in April 1964 for the Andya transmissions received in College and Fort Monmouth, along with riometer absorption values
from Longyearbyen, Troms, and Andya.
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 AndyaCollege circuit (approximately
normal to the auroral oval) were more reliable than were those on the Andya
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 BoulderBarrow 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
AprilJune 1960 (SSN113120) 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 modied 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 BoulderBarrow path
with an estimated auroral oval for low Kp. It is obvious that only the northern half
of the path will be aected 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 BoulderBarrow propagation path (from Tveten, 1962).
8.4 HF propagation
Figure 8.91. A comparison of observed and predicted MUFs for the BoulderBarrow path
for April 1960 (from Tveten, 1962).
Typical oblique ionograms from the BarrowBoulder 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 aected both by the northern and
by the southern auroral ovals. Frequencies of 13 and 17 MHz and output powers
of 0.51.0 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 1517 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
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 1519 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 gures remain quite useful for estimating
eects 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 ve HF forward-sounding circuits and supporting riometers in relation to an idealized polar-cap absorption area for the Canadian
DRTE propagation experiment in 19601961 (after Jull, 1964).
Table 8.13. Riometers and path characteristics
HF forward-sounding
circuits
Path-length
(km)
NS subauroral: OTCh
NS transauroral: OTRB
NS innerauroral: ChRB
Trans-Atlantic: OTHA
Groundair: HALA/C
1900
3400
1830
5640
02520
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
518
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

19591961 (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)
OTCh
OTRB
ChRB
29
46
32
25
43
29
54
89
61
Note:
OT, Ottawa; Ch, Coral Harbour; RB, Resolute Bay.
Figure 8.97. July 1961 PCA eects on four HF circuits: (a) for the 30-MHz riometer at
Resolute Bay; and (b) LUFs.
520
Figure 8.98. The ight
plan for groundair
trials of 1516 December
1960 (from Jull, 1962)
A unique part of Julls (1962) HF high-latitude-propagation experiment was

the monitoring of 323-MHz transmissions from Halifax, Nova Scotia by an aircraft ying in the subauroral and auroral regions during disturbed periods. In particular, one groundair trial was own on 1516 December 1960 (SSN83)
during a minor geomagnetic storm (Kp 4). The ight plan is shown in Figure
8.98 and the observed MUFs and LUFs as functions of time and location are
During the ight it was found that taking soundings every 5 min provided sucient information to select proper operating frequencies for groundair 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 ight period. Some conclusions of Jull et al. (1964)
8.4 HF propagation
(a)
(b)
Figure 8.99. The MUF and LUF observed during the groundair trials of 1516 December
1960: (a) the outgoing leg of the ight; and (b) the incoming leg of the ight (from Jull,
1964).
521
522
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 trac
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
reection point was within the auroral oval for 3Kp 5 and the specic 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 (dened 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
of AE (21000400 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 takeo angles, path distance, and operating frequency tend to
exclude the possibility of an F-mode. An AE burst was dened 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 19911992, 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 2030 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 rst
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 WalesFairbanks AE
experiment, Wagner et al. (1995) conducted a HF-channel-probe experiment on
a 1294-km path between Sondrestrom, Geenland and Keavk, 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 reected ionospheric returns characteristic of a
8.4 HF propagation
Figure 8.102. The occurrence of AE for Winter 19911992 (from Hunsucker et al., 1996).
Figure 8.103. The occurrence of AE for Spring 1992 (from Hunsucker et al., 1996).
525
526
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
je
W
es
tro
tw
ec
ar
d
El
El
rd
ec
wa
tro
je
st
Ea
Harang Disc
ontinuity
Figure 8.106. A schematic representation of the Harang discontinuity, illustrating the

direction of ow 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
(AugustOctober 1991)
9.9
17.4
120
403
Winter
(NovemberDecember
1991, January 1992)
8.3
19.0
84
383
Spring
(FebruaryApril 1992)
8.6
18.4
65
272
21.0
19.2
21
192
388
Summer
(MayJuly 1992)
528
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 reected 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 Ban 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 WalesFairbanks and

the SondrestromKeavk HF experiments
WalesFairbanks HF
experiment
SondrestomKeavk 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
311 MHz
Transmission mode
CW
Complex pulse simultaneous

with Chirpsounder
Transmitter power
100 W CW
170 W pulse
Duration of experiment 14 months

(July 1991September 1992)
2 weeks (13 March2 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 quantied 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 (PrudhoeClyde River) had a mean DSI of 60 Hz in the summer and 90
Hz during winter months. Specic 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 reection
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 dening the bands in decades, it would
be better to dene 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 (110 GHz) is either LOS in the troposphere, with typical path lengths of
3050 km, or via satelliteEarth links. Until the advent of satellite communications
530
in the 1960s, the backbone of the USAs 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
10002000 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 oer 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 1.5013 GHz) have also been used for terrestrial LOS
communication links, but such use is limited by rather severe atmospheric absorption and high-latitude eects 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
eects of AE ionization on MBC, and studies of trans-ionospheric propagation
in the development of morphological models of scintillation eects. The latter
subject will be covered in Section 9.2.2 and we will list salient eects 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 40104-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 eects 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 eects, was reported by Akram and Cannon (1994).
Specically, 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 diraction 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. 19561997 and presented in this
chapter, providing at least some qualitative indications for ameliorating the most
deleterious eects. 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 eects 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 reection
height of the D region, which changes the dimensions of the Earthionosphere
waveguide, which then produces changes in the phase and amplitude of the
ELF/VLF signals. There seems to be no rm quantitative evidence that these
changes produce serious eects on ELF transmissions received in submerged submarines, but there is some evidence that polar eects 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 inuence 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 aect HF propagation at high latitudes, and,
during the period 19561969, 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 eects. The behavior of
531
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 reection 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 dicult 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 eects, 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 eects of
the polar and auroral ionosphere on trans-ionospheric signals will be described in
the next chapter.
8.7
Section 8.1
Croft, T. A. (1968) Skywave backscatter: a means for observing our environment at
great distance. Rev. Geophys. Space Phys. 10, 73155.
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; 516 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, 1619
and 132.
Hunsucker, R. D. (1971) Characteristic signatures of the midlatitude ionosphere
observed with a narrow-beam HF backscatter sounder. Radio Sci. 6535548.
Hunsucker, R. D. (1991) Radio Techniques for Probing the Terrestrial Ionosphere.
Springer-Verlag, Heidelberg.
Hunsucker, R. D. (1992) Auroral and polar-cap ionospheric eects on radio propagation. IEEE Trans. Antennas Propagation 40, 818828.
Hunsucker, R. D. and Bates, H. F. (1969) Survey of polar and auroral region eects on
HF propagation. Radio Sci. 4 347365.
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 SolarTerrestrial Physics, Boulder, Colorado.
Soicher, H. (ed.) (1985) Propagation eects 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, 175206.
Bannister, P. (1993) ELF propagation highlights. In AGARD Conference Proc. 529, pp.
2-12-15.
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 eects from the Nov. 10, 1961 polar-cap absorption event, J.
Geophys. Res., 67, 27452751.
Bates, H. F and Albee, P. R. (1965) General VLF phase variations observed at College,
Alaska. J. Geophys. Res. 70, 21872208.
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, 11411152.
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, 8388.
Wait, J. R. (1970) Electromagnetic Waves in Stratied 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, 10511054.
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) Eects of the February
1986 magnetic storm on medium frequency skywave signal received at Fairbanks,
Alaska. Proc. IES87, 197204.
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
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 Inuence 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, 14131420.
Bates, H. F. and Hunsucker, R. D. (1974) Quiet and disturbed electron density proles
in the auroral zone ionosphere. Radio Sci. 9, 455467.
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, 121134.
Hunsucker, R. D. and Stark, R. (1959) Oblique xed-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, 581592.
Hunsucker, R. D. (1964a) Auroral absorption eects on a transpolar synchronized
step-frequency circuit. Proc. IEEE, 52, March.
Hunsucker, R. D. (1964b) Auroral-zone absorption eects on an HF arctic propagation path. Radio Sci. D 68, 717721.
Hunsucker, R. D. (1965) On the determination of the electron density within discrete
auroral forms in the E-region. J. Geophys. Res. 70, 37913792.
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, 383388.
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. 157176. 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, 159166.
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.
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.
177188.
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
45104 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) Inuence of solar particle radiations on Arctic HF propagation, presented at the AGARD Ionospheric Research Communication meeting, Naples, 1520
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, 300314.
Rose, G. (1964) Field strength measurements over a 2000 km subauroral path
(SodankylLindau) 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, 115127.
Tveten, L. H. (1961) Long-distance one-hop F1 propagation through the auroral zone.
J. Geophys. Res. 66, 16831684.
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. 119123.
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, 213222.
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. 343347.
Cannon, P. S., Dickson, A. H., and Armstrong, M. H. (1985) Meteor scatter communication at high latitudes. In Proc. AGARD (NATO) Conference.
535
536
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
45104 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 2530-MHz frequency
range on 10002000-km paths tangential and normal to the auroral oval has
been documented by Hunsucker et al. (1996) and Nishino et al. (1999).
537
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 proles at
a sucient 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 dene 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
rst-order eect 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 9.19.3 The JonesStephenson (1975) three-dimensional ray-tracing
program was used with a model based on parabolic ts to the Fairbanks verticalincidence E- and F-region parameters to produce backscatter ray traces at three
dierent 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
prole 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 suciently 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 Specications 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 eective 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,

solarterrestrial 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
ux
Model
D-region ion-chemistry model

(Sodankyl ion data from
Chemistry Model SIC
Turunen et al.
Steady-state D-region model

(Swider)
Eight-moment uid 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
h3002000 km
Covers geographic latitudes

5090, h80500 km
High-latitude F region
(h150600 km)
High-latitude model Height

range 1003000 km, includes
convection, electric-eld and
Jouleheating eects
Good agreement with PCA data
Height range 70100 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)

through the NTIS
Applicable to ISR and riometer

absorption data; for PCs in
MATLAB 4.2
References and comments
Includes auroral regions;

h80600 km
First principles, threedimensional EulerLagrange

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 ux
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)
FullerRowell et al.)
Global model, h90600 km
Includes auroral and polar

regions, h901000 km
Auroral and polar ionosphere,

h100500 km
Semi-empirical parameterized
model, uses some GTIM
equations to generate a global set
of electron-density proles and
the PIM (Daniell et al., 1995)
data base, then uses near-realtime
data from sensors
Parameterized Realtime
Ionospheric Specications Model
(PRISM) (Anderson et al.)
F region only
First principles, threedimensional EulerLagrangian
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
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, Ohms 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
IonosphereMesosphere
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)
Specically for high latitudes

5090 MLAT. Spatial
resolution in tens of kilometers
both in MLT and in MLAT
Global model, h30500 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 ux. 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 ux (1985) is enormously reduced
compared with the occurrence for the same magnetic conditions during a year of
high solar ux (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, specically, 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
3050 in the North Pacic 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 dierences between
the model and median soundings are small enough to have negligible eect on predictions.
9.2.3
Validation of ionospheric models
In the last decade more eort 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 eorts, as
reported by Schunk (1996), Anderson et al. (1998), and Szuszczewicz et al. (1998)
Anderson et al. (1998) compared ve 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 ve models displayed
diurnal variations that, in general, agreed with measurements, but each one of the
545
546
ve models exhibited a clear deciency 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 ELFHF 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
Earthionosphere 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
Figure 9.4. The daytime average absolute dierence between LWPC and measurements
parametric in frequency and distance interval (from Ferguson, 1995).
Capability (LWPC) prediction program, using measured in-ight signal levels

from various transmitters from 10 to 60 kHz from 0 to 80 CGL. The average
absolute dierences 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 (230 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 Goodmans book, but only two of these programs (AMBCOM
and ICEPAC) include some high-latitude ionospheric eects. The widely used
mid-latitude HF propagation program IONCAP was modied to include AE ionization plus polar and AA eects by Hunsucker (1971) to make predictions for US
Coast Guard communications to aid search-and-rescue missions in the north
547
548
Figure 9.5. The night-time dierence in average absolute absorption as in Figure 9.4 (from
Ferguson, 1995).
Pacic, but there is no information on the reliability of these predictions. Dav

(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 rst-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
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 eld 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 signicantly 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
Figure 9.7. The HF propagation path, between Clyde River, Canada and Leicester, UK
(after Gikas, 1990).
the ICED electron density proles 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 masters 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 19881989
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 signicant dierences between the average
error and/or standard deviation of error as a function of frequency for dierent
prediction programs and data obtained during the 1989 winter campaign, as
shown in Figures 9.89.10.
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 RiverLeicester 70% of

4.0
predictions were
between 20 dB
and 20 dB error
Reference
Gikas (1990)
IONCAP-PC 2.5
Clyde RiverLeicester Predicts error with

Tsolekas (1990)
an error less than 10
dB, with signicant
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 711 dB
ICEPAC
Clyde River to four

polar receiver sites
Absolute errors
from 0.3 to 26.4 dB
Burtch 1991
552
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).
9.2.5
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
ux, 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 eorts.
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 specic propagation modes.
The basic modulation scheme selected was direct-sequencespread-spectrum
(DSSS), in which a digital code sequence modulates the carrier at a much higher
rate than the information and produces a (sinx/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 DSSS 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 DSSS
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 identication 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
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 DSSS 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 eort of PENEX concentrated on the DSSS 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
hearno 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 ux, system
parameters, frequencies,
Qe

coordinates, date, SSN,
solar ux or T-index,
system parameters,
usable frequencies
Inputs
ALFabsorptionlimited frequency,
BUF best usable
frequency, EMUFthe
E-layer maximum
frequency,
MUFmaximum
usable frequency via the
F region,
OWFoptimum
working frequency;
Caruana (1993)
Qeective Q index;
Stewart (1990, private
communication)
MUF, LUF, median eld

strength (dB)
ALF, BUF, EMUF,

MUF, OWF, takeo
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

coordinates, date, SSN
solar ux, Kp (an index

of solar-proton and x-ray
ux)

coordinates, date, SSN or
Inputs
MUF, median Field

strength
LUF, MUF, and OWF,

median eld strength
Outputs
Lane (1996)
(1981)
OWFoptimum
working frequency;
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
Figure 9.13. The PENEX signal amplitude for 27 September 1993, f 11.0 MHz.
Specic 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 DSSS 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 dene 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 dened 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
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 nestructure 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 1250 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 WalesRock Springs, Pennsylvania path

McDowell et al. (1993) described the equipment and results of a hearno-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
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 rst 70% of the path lying within the oval for
Kp 5. Specic 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 rst 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 signicant 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
78
55
35
64
78
90
25
64
39
43
53
45
23
55
40
39
I
15
50
53
39
Disturbed
84
17
26
42
V
44
13
50
36
A
28
53
40
40
I
20
73
80
41
Quiet
90
73
40
67
Fall
50
30
49
43
A
55
49
37
47
I
27
63
87
59
Disturbed
50
31
49
43
V
50
30
40
40
A
33
40
40
37
I
25
40
30
32
Quiet
42
30
40
37
40
20
40
37
Winter
93
40
67
66
6
40
33
26
Disturbed
50
40
67
52
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 signicant dierence among the accuracies of prediction for summer, fall, and winter (41%, 48%, and 45%,
respectively).
(4) There is no signicant dierence 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%.
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 ux80.
(b)
(c)
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 rst hop of this path to be inuenced 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 ux and the rst 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 specic 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 311
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 WalesFairbanks 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
WalesFairbanks 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
Figure 9.16. Space-weather data (a DMSP image in relation to the WalesFairbanks

propagation path).
Figure 9.17. The HF propagation metric on 4 November 1993. The amplitude of the
signal is 16.8 MHz.
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 eects 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
specic indices and variations in signal for this small sample. Another feature of
567
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
WalesFairbanks 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 Oce 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
20005000-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 dawnsunrise 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
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 ux. Therefore, night-time maintenance and hence
densities are probably too low.
AIM is being modied 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 eects
(which can be profound on disturbed days) are, however, not included. Therefore,
we shall examine the behavior of the WalesSeattle 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
WalesSeattle (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

(02000600 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 (00000100
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 rst 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
eorts 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
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 Gersons 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 diculty 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 eects.
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. Gersons (1964) comparison of various communications modes on highlatitude paths on a scale of 19 (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 solarterrestrial 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 ux) and a geomagnetic-activity
index (usually Kp) in addition to time of day, month, and year as inputs. The
573
574
programs which predict eld 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 suciently accurate
predictions especially at high latitudes. See Sailors and Rose (1993) for a discussion of how seven of these programs calculate the eld strength. At present, only
the PRISM/VOACAP and the PROPMAN (Hu et al., 1998) prediction programs
utilize additional solarterrestrial 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 solarterrestrial
inputs would probably be more eective 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 sucient
detail the near-realtime high-latitude ionosphere to permit ray-tracing or virtualgeometry calculations sucient for calculations of HF propagation especially
eld-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 models 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 proles from ionospheric tomography and vertical-incidence sounders, plus satellite-measured solarterrestrial 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 briey 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
Figure 9.22. Hierarchical layers of a HF radio system (from Johnson et al., 1997).
575
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 trac, or simply note that communication
is possible, and then drop the link. The following protocol is taken from Johnson
et al. (1997, pp. 910).
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
identication 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 19861987 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 amplier, 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 quantied are the several parameters measured directly by the ALQA (listed
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 Q7:1, 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 conguration 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 9.289.30. It was planned
577
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)
Identier
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.
Figure 9.24. The TAHFE equipment conguration.
Figure 9.25. The TAHFE test procedure.
579
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 le 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 solarterrestrial data base into the results of the TAHFE-measured data, but no funding was available for this eort.
9.5.4
Realtime channel evaluation
The most promising technique for ameliorating deleterious eects on HF highlatitude communication circuits is realtime channel evaluation (RTCE), which is
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. Specic classes of RTCE are oblique-incidence sounding (OIS), channel
evaluation and calling (CHEC), vertical-incidence sounding (VIS), backscatter
581
582
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).
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 denition 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 trac.
The CCIR classes of RTCE are listed in Table 9.11, along with some examples.
A relatively long-term (December 1994summer 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 eort, 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 trac at 3001800 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
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, SNR 3dB (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).
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-ight (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 9.369.38 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)
SvalbardTuentangen
SvalbardKiruna
HarstadTuentangen
HarstadKiruna
2019
1158
1019
194
Figure 9.34. Maps showing positions of Doppler and Multipath Sounding Network (DAMSON) sites (from Angling et al., 1998).
588
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 dierent 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-nder (HFDF) experiment conducted at high latitudes, in which they
Table 9.13. A summary of Doppler/SNR plots for multipath spreads of 05 ms
(from Angling et al., 1998)
Path
Time
Frequency
ST
SK
HT
HK
0024 UT
All frequencies
2.84.7 MHz
6.711.2 MHz
14.421.9 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
1901 UT
All frequencies
2.84.7 MHz
6.711.2 MHz
14.421.9 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
0024 UT
All frequencies
2.84.7 MHz
6.711.2 MHz
14.421.9 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
1901 UT
All frequencies
2.84.7 MHz
6.711.2 MHz
14.421.9 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:
STSvalbardTuentangen
SKSvalbardKiruna
HTHarstadTuentangen
HKHarstadKiruna
592
Table 9.14. A summary of Doppler/multipath plots for SNR of 5 to 5 dB (from

Angling et al., 1998)
Path
Time
Frequency
ST
SK
HT
HK
0024 UT
All frequencies
2.84.7 MHz
6.711.2 MHz
14.421.9 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
1901 UT
All frequencies
2.84.7 MHz
6.711.2 MHz
14.421.9 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
0024 UT
All frequencies
2.84.7 MHz
6.711.2 MHz
14.421.9 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
1901 UT
All frequencies
2.84.7 MHz
6.711.2 MHz
14.421.9 MHz
0024 UT
All frequencies
2.84.7 MHz
6.711.2 MHz
14.421.9 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
1901 UT
All frequencies
2.84.7 MHz
6.711.2 MHz
14.421.9 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
0024 UT
All frequencies
2.84.7 MHz
6.711.2 MHz
14.421.9 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
1901 UT
All frequencies
2.84.7 MHz
6.711.2 MHz
14.421.9 MHz
No guard
Composite
multipath
spread
(ms)
Guard
01.67 ms
No guard
Composite
Doppler
spread
(Hz)
Guard
01.25 Hz
Notes:
ST SvalbardTuentangen
SK SvalbardKiruna
HTHarstadTuentangen
HKHarstadKiruna
9.3
6.4
34.1
53.4
9.6 Other phenomena
report nding 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-nding
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 reection from the walls of the mid-latitude ionospheric
trough this was also conrmed 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 reections 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 reection 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 IsfjordCricklade path were
made on a frequency of 14.4 MHz for 7 days in late 1995 and early 1996 during
the interval 11001600 UT and measurements on the IsfjordAlert 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 reection points. The 1997 HFDF measurements of large bearing deviations
from the GCP are a verication 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
Eects of an auroral substorm and ionospheric modication 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 reected from the heated ionosphere over
593
594
Troms. Dynamic Doppler spectra on these received signals showed the presence
of well-dened eld-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 eects 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 eects are a substantial growth in strength of the signal several hours before the expansion phase
of the substorm and a more signicant inuence 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 eastwest 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 ltering 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 WalesFairbanks 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 19931994, 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.
Verications of the relation between GPS/TEC detection of AE and TIROS
precipitating particle auroral energy ux 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 dene the equatorward boundary of the auroral oval.
9.6 Other phenomena
Figure 9.39. Simulated ray-tracing of the eld-aligned scattered HF signals from the Es
region on the LondonSt 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
Figure 9.40. The same as Figure 9.38, but for f9.410 MHz (after Blagoveschenskaya et
al., 1998).
To quote the authors,

Tremendous potential exists for monitoring the eects 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 3570, with average K indices of 03 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 IsfjordTuentangen path (2019 km) and the HarstadKiruna
path (194 km), respectively.
The authors concluded from this investigation that the data rate must be sacriced if high availability is required. Specically, 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 benets from being better
able to cope with scattered and o-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 justied in including a considerable number
597
TECU
100
200
300
Distance from closest approach to midpoint (km)
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
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
day 365
TEC Data for pm 28 Type 3: Structure over Midpoint
0
500 400 300 200 100
10
15
20
25
30
35
40
500
day 342 week AEI
day 336
day 351
day 001 strong AEI
100
200
300
day 335 strong AEI
0
500 400 300 200 100
10
15
20
25
30
35
40
400
500
100
200
day 330 week AEI
day 330 week AEI
100
200
300
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
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 gure) (from Hunsucker et al., 1996).
TECU
TEC Data for pm 28 Type 2: Structured Slant
TECU
TECU
400
400
500
500
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
21
dB
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. Specic examples have been
included to illustrate these variations.
Ionospheric modeling and propagation-prediction techniques have been
improved signicantly 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 sucient 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.
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 5070 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 specic radio-propagation characterization,
601
Figure 9.44. The latitudinal variation in 1-h TEC data compared with TIROS particle precipitation for Kp 14
(from Coker et al., 1995).
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
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
Figure 9.47. The overall availability of modems when the frequency set consists of 1, 2, . . .,
10 frequencies for the IsfjordTuentangen path (2019 km) (from Jodalen et al., 2001).
605
606
Figure 9.48. The same as Figure 9.45, but for the HarstadKiruna path (194 km) (from
Jodalen et al., 2001).
9.8
Section 9.1
Hunsucker, R. D., Rose, R. D., Adler, R. W., and Lott, G. K. (1996) Auroral-E mode
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, 833898.
Section 9.2
Aarons, J., Kersley, L., and Rodger, A. S. (1995) The sunspot cycle and auroral
F-layer irregularities. Radio Sci., 30, 631638.
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, 21792192.
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 Oce,
Washington DC.
Bibl, K. (1998) Evolution of the ionosonde. Annal: de Geosica 41.
Bilitza, D. (1999) IRI 2000. In Proc. IES99, pp. 348351.
Bishop, G. J. et al. (1999) The eect 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. 508518.
Chiu, Y. T. (1975) An improved phenomenological model of ionospheric density.
Dav, N. (1990) The use of mode structure diagrams in the prediction of high-latitude
HF propagation. Radio Sci. 30, 309323.
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 Earths 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
Ferguson, J. A. (1995) Ionospheric model validation at VLF and LF. Radio Sci. 30,
775782.
Ferguson, J. and Snyder, F. P. (1989) Long wave propagation assessment. In
Operational Decision Aids for Exploiting or Mitigating Electromagnetic Propagation
Eects (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. 365376.
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.
Reinhold, New York.
Hunsucker, R. D. and Owren, L. (1962) Auroral sporadic-E ionization. J. Res. NBS D
66, 581592.
Hunsucker, R. D. (1965) On the determination of the electron density within discrete
auroral forms in the E-region. J. Geophys. Res. 70, 37913792.
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 Pacic 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, 293315.
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. (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, 739746.
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, 617629.
Omura, J. K., Schultz, R. A., and Levitt, B. K. (1985) Spread Spectrum

Communications, volumes IIII. Computer Science Press, Rockville, Maryland.
Rose, R. B. (1982) An emerging propagation prediction technology. In Eects of the
Ionosphere on Radiowave Systems (IES81) (ed. J. Goodman). US Government
Printing Oce, 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 fF2 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, 713728.
Schunk, R. W. (1996) SolarTerrestrial 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 eects 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 rst 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, 449452.
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, 12931309.
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 electrictransverse
magnetic noise prediction model. Radio Sci. 30, 783797.
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, 55395547.
609
610
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,
97107.
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 Inuences on System Design, Rotterdam.
Fenwick, R. B. and Villard, O. G. (1963) A test of the importance of ionosphere
ionosphere reections in long distance and around-the-world HF propagation.
J. Geophys Res. 68, 56595666.
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.
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,
17051715.
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, 14131428.
Hunsucker, R. D. and Bates, H. F. (1969) Survey of polar and auroral region eects on
HF propagation. Radio Sci. 4, 347375.
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, 14131420.
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 modied auroral ionosphere during a magnetospheric substorm. Radio Sci. 34, 715724.
Blagoveshchensky, D. V. and Borisova, T. D. (2000) Substorm eects 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, 32593262.
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, 13131318.
Hunsucker, R. D., Rose, R. D., Adler, R. W., and Lott, G. K. (1996) Auroral-E mode
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
eect during the October 1995 WAGS campaign. Radio Sci. 23, 919930.
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, 9196.
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
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, 241249.
Williams, P. J. S. (1989) Observations of atmospheric gravity waves with incoherent
scatter radar. Adv. Space Res. 9, 6572.
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 reect 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 eld
stands today. Mitras 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 SolarTerrestrial Environment. Cambridge University Press,
Cambridge (1992).
Hines, C. O., Paghis, I., Hartz T. R., and Fejer, J. A. (eds.) Physics of the Earths Upper
Atmosphere. Prentice-Hall, Englewood Clis, 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 331332
acoustic gravity waves, theory 5257
adiabatic invariants 79
aeronomy, physical 1323
airglow and atmospheric cooling 9
Alfvn Mach number 72
Alfvn wave 103
all-sky camera 291
alpha-Chapman layer 18
antennas 115116
basic principles 115
design 116
atmospheric composition 1013
atmospheric heating 89
attachment coefcient 18
attenuation 115
atmospheric absorption 122
ionospheric absorption 151
measurement techniques 203210
deviative techniques 151
non-deviative techniques 144
aurora australis 291
aurora borealis 291
aurora
altitude of 296
diffuse 300301
discrete 300301
intensity of 299300
luminous 285, 291302
mantle 300
radar 285, 326329
theta 288, 302
auroral
activity predictions 367371
electrojet 312314
forms 296
green line 302
infrasonic waves 300331
red line 302
613
auroral oval 286288

and radio scintillation 308
boundaries of 289290
models of 288291
auroral radar 326329
auroral radar echoes,
occurrence of 328329
polarization of 328
auroral radio absorption 285, 304, 339382
and geomagnetic activity 365
and HF propagation 365367
conjugacy of 379382
co-rotation in 363365
duration of 351354
dynamics of 354365
global movement of 358359
preceeding bay in 345347, 359361
proles of 373
pulsations in 347, 382
sharp onset in 341342
substorm onset in 377379
substorm dynamics in 377379
slowly varying 347, 363
spatial extent of 351
spike event in 341345
statistics of 350354
zone 350351
auroral spectroscopy 302
auroral substorm 285, 308311
and ionospheric effects 311312
break-up of 309
expansion phase of 309
growth phase of 309310
pseudo-breakup of 310
recovery phase of 309
auroral X-rays 285304
auroral zone 286
magnetic bays in 305
magnetic disturbances in 285
barometric equation 5
Bartels musical diagram 9798
beta-Chapman layer 19
Index
614
Bragg curve 107

Bremsstrahlung X-rays 106107
BruntVisala frequency 54
bursty bulk ow 318
C layer 36
Canadian-border effect 261
Chapman production function 15
character gure 45
charge-exchange reaction 26
chemical transport (of heat) 9
conductivity
Hall 49
height variation of 50
of the ionosphere 4850
of the ground 48
Pedersen 49
with magnetic eld 4849
with no magnetic eld 48
continuity equation 14
coronal hole 67
coronal mass ejection (CME) 6769
current,
Birkeland 8586, 96, 312314
ring 8485
currents in substorm 312315
current system,
equivalent 86, 312
Spq 87
SD 9596
current wedge, wedgelet 313314
D region
at high latitude 337339
electron ux 373377
production of 31
proles in the auroral zone 373
recombination coefcient 400
summer mesospheric echo in 406409
electric current,
Birkland 51
Cowling 51
electric currents in the ionosphere 50
electric eld and co-rotation 92
electrojet,
auroral 7576
equatorial 51
EM noise and interference
atmospheric noise 127139
galactic noise 133
solar noise 134139
emissions, electromagnetic 285
escape temperature 8
EUV (extreme ultra-violet) 14
exobase, denition of 5
exosphere 78
denition of 5
F region
blobs 245249
in the auroral oval 240242
in polar cap: U.T. effect 235237
in polar cap: the tongue 234236
patches 244245
scintillation production in 249260
storm-time variation of 4647
F1 layer, aeronomy of 26
F1 ledge 31
F2 region 3738
alphabeta transition level in 37
and composition changes 43
and conjugate ionosphere 44
and effect of neutral wind 43
anomalies of 3944
Bradbury layer in 37
seasonal anomaly of 40
semi-annual anomaly of 40
eld-line circulation 316
eld-lines 62
Daltons law 7
diffusion coefcient 20
ambipolar 22
ux-transfer event 91
frozen-in eld 65
diffusion in the ionosphere 2023

dipole eld 6163
geomagnetic cavity 63
geomagnetic eld 6163
geopotential height 7
distribution height 21
E layer at night 27
E region
aeronomy of 2631
disturbed auroral 323326
polar 323
quiet auroral 323
sporadic-E phenomenon in 2731
eddy diffusion in the atmosphere 9
Harang discontinuity 95, 242

heliosphere, denition of 5
heterosphere, denition of 5
HF propagation at high latitudes 440530
adaptive HF techniques 574580
assessment of HF channels 586590
xed-frequency tests 440474
absorption effects 467474
Index
Alaska/Scandinavia 440446
other high-latitude paths 450471, 503511
Alaska (WalesFairbanks) 523530
Alaskacontinental USA 447450
AlaskaGreenland 471474
large bearing errors 591593
mitigation of disturbances 572574
use of GPS/TEC 594603
modem use 597606
PENEX 553568
realtime HF channel evaluation 580586
substorm effects 593594
swept-frequency tests 474493
auroral-E effects 480482
AndyaCollege 478480, 503512
BarrowBoulder tests 506516
Canadian tests 517522
McMurdoThule path 514, 517
SodankylLindau tests 494502
SondrestromKeavik tests 528530
ThuleCollege tests 474478
models of high-latitude ionosphere 538546,
568572
non-great-circle modes 477480, 482483
ducted modes 490493
Doppler and fading characteristics 493494
ionospheric ray-tracing 538541
homosphere, denition of 5
hydrated ions 3235
hydromagnetic wave 103
hydrostatic equation 5
hydrostatic equilibrium 57
instability
ballooning 319
gradient-drift 328
KelvinHelmholtz 105, 318319
two-stream 104105, 328
interplanetary magnetic eld (IMF) 65
ionization
by alpha-particles 107108
by energetic electrons 105106
by energetic protons 107108
ionization efciency 15, 24
ionization potential 24
ionosphere
denition of 4
naming of 1
sluggishness of 27
ionospheric effects
of the aurora 302305
of the sunspot cycle 4445
of the thermospheric wind 23
ionospheric layers
critical frequency of 27
denitions of 1314
heights of 25
ionospheric storm 4647
with sudden commencement 46
irregularity strength parameter 258
615
keogram 300
log-normal distribution 369371
luminous aurora and the E region 325
Lyman-alpha line of solar spectrum 31
M region 285286
magnetic bay 9596
magnetic cloud 69
magnetic-eld merging 9091
magnetic eld, reconnection of 9091, 316
magnetic index
Ap 9697
Dst 94
Kp 9697
magnetic indices 96100
AU, AL, AE 9799
magnetic longitude 62
magnetic micropulsation 103104
magnetic storm 93102
classical 94
phases of 94
practical effects of 100102
magnetopause 6971
and image-dipole method 71
denition of 5
magnetosheath 7172
magnetosonic wave 103
magnetosphere,
circulation of 8690, 228234, 316
electric elds in 9192
denition of 5
shock front of 7172
magnetosphere boundary layer 7374
magnetospheric substorm 315319
magnetotail 70, 7273
behavior in substorm 316318
lobes 78
plasma sheet 70, 73, 78
magnetotail,
electric potential across 228
ionospheric sources to 240
neutral line 318
mean free path 7
mesopause, denition of 4
mesosphere, denition of 4
metallic ions in the atmosphere 1213
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 7378
depletion of 93
dynamics of 9293
plasmoid 317
polar arc 301302
polar cap
circulation patterns in 228234
electric eld in 228
potential across 9192
polar-cap absorption 382406
and solar radio emissions 389390
and solar ares 389390
and proton ux 387
daynight effects in 400405
duration of 384
magnitude of 389
midday recovery in 395397
occurrence of 384387
seasonal variation of 387389
uniformity of 395
polar-cap edge (in PCA) 393395
polar cusps (clefts) 237239
and charged particles 237
and luminosity 237
and ionospheric heating 239
polar hole 260, 276280
polar wind 74, 239240
propagation
conductivity 121
Faraday rotation 149151
forward scatter 171
HF at high latitudes 440530
LF and MF at high latitudes 430439
ionospheric 140169
ionospheric scatter 171174
line-of-sight 113116
lossy medium 120121
magnetoionic theory 140145
phase effects 148149
prediction programs 174
predictions and validations 546565
scintillations 152163
terrain effects 125127
transionospheric 147159
VHFmicrowave 530531
VLF/UHF at high latitudes 530531
VLF/ELF principles 163167
VLF/ELF at high latitudes 419430
whistlers 167169
HF-propagation-prediction programs
174176
noise and interference 127139
proton effects in the neutral atmosphere
398406
protonosphere 3839
base of 39
denition of 5
protons
IMF effects on 390392
magnetospheric effects on 392395
quiet-day curve 339
radar aurora 303304
radar, basics of 116119
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 125127
reaction rates, temperature dependence of 43,
274275
recombination
dissociative 19, 26, 33
radiative 26
recombination coefcient 18
effective 20
recombination processes, types of 18
reection from the ionosphere 144145
relation between oblique and vertical
145147
reection at a boundary 159163
refractive effects
tropospheric 119120
neutral atmosphere 122125
relativistic electron precipitation 348
rigidity 393
ring current 73, 8485
riometer 339340
scale height, denition of 6
scale height of plasma 22
scatter from the ionosphere 169174
coherent scatter 169171
forward scatter 171
incoherent scatter 171174
scintillation
and Fresnel zones 256
modeling 258260
scintillation,
properties of 249256
S4 index of 255
spectrum of 256
small irregularites, in situ measurement of
257258
solar wind 6369
and Kp 97
ballerina model of 67
Index
composition of 63
fast stream in 67
garden-hose effect in 65
sectors 6566
space-weather-data use 565574, 600
sporadic-E
and metallic ions 29
and scintillation 31
and wind shear 2829
Strmer theory 392395
stormsubstorm relations 321322
stratopause, denition of 4
substorm 308322
current wedge 313
rate 321
theories of 318319
triggering 319321
techniques
D-region absorption 203210
ground-based 181214
HF Doppler and spaced receiver 217219
incoherent scatter radars 203205
in situ measurements 216217
ionosondes 181187
ionospheric imaging 219220
modication by HF transmitters 210214
oblique-incidence HF/VHF sounders
187202
riometers (URSI A2 method) 206208
satellite beacons 215216
space-based measurements 214217
topside ionospheric sounders 216217
URSI A1a and A1b (HF) absorption methods
204206
URSI A3a and A3b (LF) absorption methods
208210
temperature of neutral atmosphere 810
thermosphere, denition of 4
617
three-body reaction 35
trapped (Van Allen) particles 7884
longitude drift 83
loss cone of 83
mirror point of 83
pitch angle of 83
pseudo-trapping of 83
traveling ionospheric disturbance 57
tropopause, denition of 4
troposphere, denition of 4
trough and electron precipitation 270271
trough
in electron content 261, 266
in the southern hemisphere 269
main 260275
motion of 271273
orientation of 269270
poleward edge of 269271
principal properties of 263265
time and activity variations of 265269
causes of: heating 274275
causes of: plasma decay 273274
troughs at high latitude 276280
turbopause, denition of 5
turbosphere, denition of 5
Van Allen belts 7884
vertical transport in the atmosphere 2023
viscous interaction 8688
VLF-wave reection in the D layer 3536
VLF whistlers and the plasmasphere 75
VLF whistlers, nose 75
waveparticle interaction 104
X-rays 14, 106107

The High-Latitudude Ionospehere

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The High-Latitudude Ionospehere

Uploaded by

Copyright:

Available Formats

This page intentionally left blank

The high-latitude ionosphere and its effects on radio

Cambridge Atmospheric and Space Science Series

Titles in print in this series

J. C. King and J. Turner

J. F. Lemaire and K. I. Gringauz

D. Hastings and H. Garrett

G. E. Thomas and K. Stamnes

R. W. Schunk and A. F. Nagy

R. D. Hunsucker and J. K. Hargreaves

Cambridge University Press has no responsibility for the persistence or accuracy of

From the Times of London xv

Chapter 1 Basic principles of the ionosphere 1

The ionosphere and radio-wave propagation 1

Why the ionosphere is so dierent at high latitude 2

The vertical structure of the atmosphere 4

Hydrostatic equilibrium in the atmosphere 5

The temperature prole of the neutral atmosphere 8

The Chapman production function 15

Principles of chemical recombination 18

The main ionospheric layers 23

The E and F1 regions 26

The F2 region and the protonosphere 37

Anomalies of the F2 region 39

The eects of the sunspot cycle 44

The F-region ionospheric storm 46

The electrical conductivity of the ionosphere 48

Conductivity in the absence of a magnetic eld 48

The eect of a magnetic eld 48

The height variation of conductivity 50

Acoustic-gravity waves and traveling ionospheric disturbances 52

Traveling ionospheric disturbances 57

References and bibliography 58

Chapter 2 Geophysical phenomena inuencing the high-latitude ionosphere 61

The geomagnetic eld 61

The solar wind 63

The magnetosheath and the shock 71

The polar cusps 72

Particles in the magnetosphere 73

Principal particle populations 73

The plasma sheet 78

The ring current 84

The dynamics of the magnetosphere 86

Magnetospheric electric elds 91

The dynamics of the plasmasphere 92

The classical magnetic storm and the Dst index 94

Magnetic bays at high latitude; the auroral electrojet 95

Great magnetic storms and a case history 100

Wave phenomena of the magnetosphere 103

Ionization by energetic particles 105

Bremsstrahlung X-rays 106

References and bibliography 109

Chapter 3 Fundamentals of terrestrial radio propagation 113

Electromagnetic radiation 113

Basics of line-of-sight propagation in vacuo 113

Principles of radar 116

The signicance of the refractive index 118

Interactions between radio waves and matter 121

Propagation through the neutral atmosphere 122

The refractivity of the neutral atmosphere 122

Terrain eects 124

Noise and interference 127

Ionospheric propagation 140