Académique Documents
Professionnel Documents
Culture Documents
Editors
Alexander J. Dessler
John T. Houghton
Michael J. Rycroft
R. Daley
Atmosphere data analysis
J. R. Garratt
The atmospheric boundary layer
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
M. Walt
Introduction to geomagnetically trapped radiation
T. I. Gombosi
Physics of space environment
T. I. Gombosi
Gaskinetic theory
B. A. Kagan
Oceanatmosphere interaction and climate
modelling
I. G. Enting
Inverse problems in atmospheric constituent
transport
I. N. James
Introduction to circulating atmospheres
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
Contents
Introduction 1
1.1.1
1.1.2
1.2
1.2.1
Nomenclature 4
1.2.2
1.2.3
The exosphere 7
1.2.4
1.2.5
Composition 10
1.3
Physical aeronomy 13
1.3.1
Introduction 13
1.3.2
1.3.3
1.3.4
Vertical transport 20
1.4
1.4.1
Introduction 23
1.4.2
1.4.3
The D region 31
1.4.4
1.4.5
1.4.6
1.4.7
vii
Contents
viii
1.5
1.5.1
Introduction 48
1.5.2
1.5.3
1.5.4
1.5.5
Currents 50
1.6
1.6.1
Introduction 52
1.6.2
Theory 53
1.6.3
1.6.4
The literature 57
1.7
Introduction 61
2.2
The magnetosphere 61
2.2.1
2.2.2
2.2.3
The magnetopause 69
2.2.4
2.2.5
2.2.6
The magnetotail 72
2.3
2.3.1
2.3.2
The plasmasphere 74
2.3.3
2.3.4
Trapped particles 78
2.3.5
2.3.6
Birkeland currents 85
2.4
2.4.1
Circulation patterns 86
2.4.2
Field merging 90
2.4.3
2.4.4
2.5
Magnetic storms 93
2.5.1
Introduction 93
2.5.2
2.5.3
2.5.4
Magnetic indices 96
Contents
2.5.5
2.5.6
2.6
2.6.1
Electrons 105
2.6.2
2.6.3
Protons 107
2.7
Introduction 113
3.2
3.2.1
3.2.2
3.2.3
3.2.4
3.3
3.3.1
3.3.2
3.3.3
3.4
3.4.1
3.4.2
3.4.3
3.4.4
3.4.5
3.4.6
3.4.7
Whistlers 167
3.5
3.5.1
3.5.2
3.5.3
3.6
3.7
Summary 175
3.8
Introduction 181
4.2
ix
Contents
4.2.1
Ionosondes 181
4.2.2
4.2.3
4.2.4
4.2.5
4.3
4.3.1
4.3.2
4.3.3
4.3.4
4.3.5
4.4
4.4.1
4.4.2
4.4.3
4.5
Summary 220
4.6
5.1.1
Introduction 227
5.1.2
5.2
5.2.1
5.2.2
5.2.3
5.2.4
5.3
5.3.1
Introduction 242
5.3.2
5.3.3
5.4
5.4.1
Introduction 260
5.4.2
5.4.3
5.4.4
5.4.5
5.5
Contents
5.6
Summary 280
5.7
Introduction 285
6.2
6.2.1
6.2.2
6.3
6.3.1
6.3.2
6.3.3
6.3.4
6.3.5
6.4
6.4.1
History 308
6.4.2
6.4.3
6.4.4
6.4.5
6.4.6
The inuence of the IMF and the question of substorm triggering 319
6.4.7
6.5
6.5.1
Introduction 322
6.5.2
6.5.3
6.5.4
6.5.5
6.5.6
6.5.7
6.6
6.7
Introduction 337
7.2
7.2.1
7.2.2
Typical auroral-absorption events and their temporal and spatial properties 340
7.2.3
xi
Contents
xii
7.2.4
Dynamics 354
7.2.5
7.2.6
7.3
7.3.1
Introduction 382
7.3.2
7.3.3
7.3.4
7.3.5
7.3.6
7.4
7.5
7.6
Introduction 417
8.2
8.3
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
8.4.4
8.4.5
8.4.6
8.4.7
8.4.8
8.5
8.6
Summary 531
8.7
Introduction 537
9.2
9.2.1
9.2.2
9.2.3
Contents
9.2.4
9.2.5
9.3
9.4
9.5
9.5.1
9.5.2
9.5.3
9.5.4
9.5.5
9.6
9.6.1
9.6.2
9.6.3
9.6.4
9.7
9.8
xiii
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 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
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
Composition
baropause
Ionosphere
Barosphere
Temperature
Heterosphere
1500
Temperature (K)
10
Electron density
(10 5 cm 3 )
consider the nature of the neutral upper atmosphere in which the ionosphere is
formed.
1.2
1.2.1
Nomenclature
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
(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)
(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
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 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
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)
Figure 1.3. Atmospheric composition to 1000 km for a typical temperature prole. (US
Standard Atmosphere, 1976.)
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
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:
13
14
F1 region, 160180 km: electron density of several times 1011 to about 1012
m3 (105106 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)
(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
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
(1.19)
(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)
(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
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)
(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)
(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)
(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)
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)
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)
(1.51)
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.
(1.52)
1.4
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
(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
()
(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
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.
26
1.4.2
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).
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.)
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,
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.)
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
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
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
(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
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
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)
(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)
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
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.)
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.
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 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)
(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)
46
1.4.7
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
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
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
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
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
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
51
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
1.6
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
53
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
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
(1.83)
kz2 1
2a 2
2
kx2 1 b2 .
2
2
s
(1.84)
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)
(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
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
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
1.2
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
1.5
1.6
59
60
Yeh, K-C. and Liu, C-H. (1974) Acoustic-gravity waves in the upper atmosphere. Rev.
Geophys. Space Phys. 12, 193.
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
2.2
The magnetosphere
2.2.1
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.)
(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)
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 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
Geophysical phenomena
(a)
(b)
(c)
Figure 2.2. (a) The speed of the solar wind: a histogram of measurements between 1962
and 1970. (J. T. Gosling, in Solar Activity Observations and Predictions (eds. McIntosh and
Dryer), by kind permission of The MIT Press, 1972.) In (b) (c) are shown the distributions
of the magnitude and components of the interplanetary magnetic eld, 19881990. By is
eastwest and Bz is northsouth. (F. J. Rich and M. Hairston. J. Geophys. Res. 99, 3827,
1994, copyright by the American Geophysical Union.)
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
Geophysical phenomena
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,
1978, copyright by the American Geophysical Union.)
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
Geophysical phenomena
(a)
(b)
A shock front is formed in the solar wind two or three RE upstream of the magnetopause (Figure 2.6). The region between the shock front and the magnetopause is the magnetosheath, and here the plasma, composed mainly of solar
material but in other respects not typical of the solar wind, is turbulent.
Tenuous though it may be by any ordinary standard, the magnetosphere is a
71
Geophysical phenomena
72
relatively solid object in comparison with the solar plasma. Furthermore, the solar
wind is supersonic at the orbit of Earth, meaning that its velocity exceeds that
of any waves that can propagate within it. In the solar wind the speed of hydromagnetic waves, that is, the 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 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,
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
2.3.1
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
Geophysical phenomena
74
Figure 2.7. Plasma populations and current systems of the magnetosphere in three dimensions. (T. A. Potemra, Johns Hopkins APL Tech. Digest 4, 276, 1983. The Johns Hopkins
University Applied Physics Laboratory, 1983. All rights reserved. Reproduced by permission.)
The locations of these particle populations are indicated in Figure 2.7. They are
not merely incidental to the magnetosphere, but are in fact essential to its properties and behavior. Except for the boundary layers, each of these populations is discussed in a following section. In most of the magnetosphere the ratio of the energy
density of the particles to that of the magnetic 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
Geophysical phenomena
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
Geophysical phenomena
78
the electron density in the plasmasphere inside the pause and its regular variations,
the thickness of the pause, and the distribution in the trough beyond it.
2.3.3
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
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).
(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.)
(a) Electrons
(b) Protons
83
longitudinal drift
B
gyration
Earth
e
p
M
bouncing
mirror
points
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.
Geophysical phenomena
84
Figure 2.14. The time taken for trapped electrons to make one circuit of the Earth at the
orbit of a geostationary satellite (6.6RE): (a) as a function of pitch angle for various energies; (b) as a function of energy for pitch angle 90. (a) Also gives the velocity of the footprint (at 67 latitude) 100 km above the 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
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.)
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
Geophysical phenomena
86
Figure 2.16. Distributions of Birkeland currents during (a) weak and (b) active disturbances. (T. Iijima and T. A. Potemra, J. Geophys. Res. 83, 599, 1978, copyright by the
American Geophysical Union.)
direction of the current. The intensity varies during the day in each region and is
generally up to 1 or 2 A m2. The total 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
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,
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,
Cambridge University Press, 1972.)
in which liquid is swept back at the surface and returns down the middle of the
drop.
The concept of a circulating magnetosphere driven by viscous interaction at the
surface was put forward by Axford and Hines in 1961. The nature of the viscous
interaction was not 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
Geophysical phenomena
Figure 2.18. Patterns of magnetospheric circulation in the equatorial plane: (a) if due to
friction of the solar wind at the magnetopause; and (b) if the 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
Geophysical phenomena
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.
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
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)
Geophysical phenomena
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
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)
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)
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
Geophysical phenomena
94
2.5.2
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)
95
2.5.3
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
Geophysical phenomena
96
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-
97
(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.
ap
0
1
2
3
4
5
6
7
8
9
0
3
7
15
27
48
80
140
240
400
98
Geophysical phenomena
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
99
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.)
Geophysical phenomena
100
The history of magnetic indices and their derivations, advantages, and disadvantages are discussed in detail by Maynaud (1980). Some of the older indices
continue to be of interest and are still produced to maintain continuity. The C
index, one of the 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
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
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.
101
102
Geophysical phenomena
have had to increase by a factor of 60. Some one RE of the movement might be
attributable to the southward component of the IMF (Section 2.2.3), but even so
one can say with reasonable certainty that the solar-wind pressure must have
increased by at least a factor of (10/5.7)6 30 on 13 March.
The ionosphere
There was also a severe ionospheric storm on that day. The mid-latitude electron
content in the night sector was unusually low immediately after the commencement. After sunrise it remained at essentially night-time values for most of the day
and then it returned rapidly to a daytime value just before sunset. The equatorial
ionosphere virtually disappeared, and HF radio communications were practically
impossible over many circuits, particularly those involving high latitudes.
However, VHF communication, which is normally restricted to line-of-sight propagation, was achieved over remarkably long distances due to high-latitude sporadic-E. An analysis of the remarkable world-wide ionospheric 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.
2.5.6
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
Geophysical phenomena
regular and persistent regular pulsations (Pc) prefer the morning and the daylight
hours.
The magnetospheric origin of micropulsations is demonstrated by their similarity in magnetically conjugate regions, but a variety of mechanisms is involved
in their generation. Pc pulsations are generated either at the surface of the magnetosphere or within it, and they propagate in a hydromagnetic mode. Pc1 are
attributed to bunches of protons (probably) traveling back and forth between
mirror points (Section 2.3.4). A resonance between protons and ion-cyclotron
waves, which rotate in the same sense in the geomagnetic 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
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
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
Geophysical phenomena
106
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.
107
Protons
Direct
126
108
94
84
Maximum
production rate
(ion pairs cm3 per electron)
X-ray
Direct
X-ray
88
70
48
37
2.5
10
1.4
104
5.6
104
1.9
103
5
5.9
1010
1.3
108
2.3
107
1.3
105
Geophysical phenomena
108
140
130
120
110
100
Altitude (km)
90
80
70
60
50
40
30
20
10
12
11
10
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
109
Geophysical phenomena
110
Wilcox, J. M. and Ness, N. F. (1965) Quasi-stationary corotating structure in the interplanetary medium. J. Geophys. Res. 70, 5793.
2.3
2.4
2.5
Magnetic storms
2.6
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.
111
112
Geophysical phenomena
Le Galley, D. P. and Rosen, A. (eds) (1964) Space Physics. Wiley, New York.
Lyons, L. R. and Williams, D. J. (1984) Quantitative Aspects of Magnetospheric
Physics. Reidel, Dordrecht.
Nishida, A. (ed.) (1982) Magnetospheric Plasma Physics. Reidel, Dordrecht.
Parks, G. K. (1991) Physics of Space Plasmas. Addison-Wesley Publishing Co.,
Redwood City, California.
Roederer, J. G. (1974) Dynamics of Geomagnetically Trapped Radiation. SpringerVerlag, Berlin.
Schulz, M. and Lanzerotti, L. J. (1974) Particle 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
113
(3.1)
114
Fundamentals of propagation
SE
H,
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
FM broadcasting,
television,
aeronautical
communication
Ultra-high-(UHF)
frequency
3003000 MHz
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.
115
(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)
(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)
(3.7)
and
(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.
(3.9)
which follows from Equations (3.6) and (3.8) assuming that both antennas are
isotropic radiators. The gain (sometimes called directivity) is given approximately
by
G30000/(),
(3.10)
Fundamentals of propagation
116
where and are the half-power beamwidths (in degrees) in the E and H planes,
respectively, assuming that there are no sidelobes. The formula applies only up to
20 beamwidth.
Although it is an important topic in radio propagation, a full discussion of
antennas would be outside the scope of this book. Some of the many treatments
are listed in the references. Information on radiation patterns and advice on siting
are contained in the publications of the American Radio Relay League (ARRL)
in the references. For detailed discussions of Fresnel-zone siting fundamentals,
bandwidth, and terrain 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
Description
Source
NEC
NEC/WIRES 1.5
Brian Beezley
3532 Linda Vista
Dr., San Marcos,
CA 92069, USA
NEC/Yagis 2.0
YO 6.0
AO 6.0
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
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)),
Fundamentals of propagation
118
(3.14a)
(3.14b)
(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
119
(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.
(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)
Fundamentals of propagation
120
decreases by a factor of e over a distance 1/!. It is convenient here to use the j notation, writing
EE0 exp[ j(t nz/c)],
(3.22)
where j 1, and it is understood that the real part is taken (since e j cos
j sin ). Taking a complex refractive index
n j
(3.23)
(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)
(3.26)
giving
EE0 exp[ j(tjz)
EE0 exp[ j(t z)]exp(z).
(3.27)
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)
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
Fundamentals of propagation
122
8.5). Also, a (usually weak) echo may be detected at the transmitter site, which is
utilized in the techniques of coherent and incoherent scatter radars described in
Section 3.5 and in Sections 4.2.2 and 4.2.3.
Polarization changes occur in an ionized medium in the presence of a geomagnetic 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
3.3.1
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)
123
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)
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
Fundamentals of propagation
124
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
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
Fundamentals of propagation
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)
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
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)
Fundamentals of propagation
130
Description
Source
MSITE,
TPATH, MCS
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
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 3.7. 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 3.14. 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).
Fundamentals of propagation
140
3.4
Ionospheric propagation
3.4.1
Magnetoionic theory
The Appleton equation
For an ionized medium the refractive index is expressed by the Appleton equation. In its complete form this is a complicated expression using the dimensionless quantities X, Y, and Z, each of which is 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;
141
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
Fundamentals of propagation
(3.51)
1
Ne2
2
0me
1/2
(3.52)
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.)
(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.
(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)
(1 2 ) .
2c
(3.59)
,
2c
(3.60)
Fundamentals of propagation
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
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)
2
c
dz
n
(3.65)
[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)
(3.68)
(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
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
Fundamentals of propagation
(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
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)
(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
Fundamentals of propagation
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)
(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).
(3.82)
149
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)
(c)
(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)
(e)
(3.84)
(3.85)
(3.86)
Fundamentals of propagation
150
and
E 2f [tnE l/c],
(3.87)
(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
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)
(3.93)
N
( ) dl
2
(3.94)
1.168
1018
[ f (MHz)]2
N dl
(3.95)
Fundamentals of propagation
152
3.4.5
153
E
a
Incident wave
Screen
E'
Emerging wave
O
P
Observing
O' plane
Phase
Amplitude
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)
(3.97)
and in this case the angular power spectrum would also be Gaussian:
P(sin )exp[sin2 /(2 sin2 0)],
(3.98)
Fundamentals of propagation
154
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)
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
(b) Signal at OO
Total
0 0
Figure 3.22. Development of (a) phase and (b) amplitude perturbations from initial perturbations.
(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
Fundamentals of propagation
Figure 3.23. Spectra of the amplitude and phase recorded at 40 MHz from a geosynchronous satellite transmission. Power spectra are plotted on a log scale of relative values in
decibels. The phase spectrum levels off due to detrending (at 3
103 Hz), but the turn in
the amplitude spectrum marks the Fresnel frequency. (After W. J. Myers et al., J. Geophys.
Res. 84, 2039 (1979), copyright by the American Geophysical Union.)
157
(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
Fundamentals of propagation
Figure 3.24. Examples of amplitude scintillation at three frequencies from a geosynchronous satellite, and their spectra. (R. Umeki et al., J. Geophys. Res., 82, 2752 (1997b).)
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
The treatment of propagation outlined in the foregoing sections, which are based
on the concept of the refractive index, assumes that the medium is uniform. Of
course this is seldom the case, but in practice the assumption may be used provided that any variations are not too large over a distance of several wavelengths.
Such a medium is said to be slowly varying. There are, however, situations in which
159
160
Fundamentals of propagation
Figure 3.26. Empirical amplitude distributions for a range of S4 values. (After R. K. Crane,
Technical Note 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)
(3.110)
Fundamentals of propagation
162
Figure 3.28. The continuity of electric and magnetic vectors at a sharp boundary.
When the wave is incident at an angle to the boundary a further condition must
be applied, which is that the normal components of the electric and magnetic 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,
(3.113)
(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
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).
(3.116)
The ionosphere now behaves as a metal rather than a dielectric, having conductivity
( 0N2 )/ Ne2/(me),
(3.117)
164
Fundamentals of propagation
Ionosphere
p
2-ho
p
ho
1-
ground
wave
Ground
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.)
166
Fundamentals of propagation
but now they alternate in sign, which is equivalent to a change in phase of , and
resonance occurs at
2hCn (n 12 ),
(3.119)
(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)
167
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
Fundamentals of propagation
Figure 3.32. A sketch of the basic manifestation of a whistler and its initiating disturbances: (a) illustrating the dispersion; (b) the 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).
169
(3.124)
1
2c
f
s
N L
(3.125)
Ionospheric scatter
Coherent scatter
Other important considerations in the case of coherent backscatter are the relation
between the size of the scattering irregularity relative to the free-space wavelength
170
Fundamentals of propagation
Figure 3.34. Heightfrequency regimes of various ionospheric radar probes (from Schlegel,
1984).
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
(3.126)
Forward scatter
Incoherent scatter
(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
Fundamentals of propagation
(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)
173
(3.132)
Fundamentals of propagation
174
In this, the collision-dominated region, the returned spectrum has the Lorentz
form,
S( f )A/[1f 2 /(f )2 ].
(3.133)
(3.134)
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
Description
AMBCOM
Includes some
eects of the highlatitude ionosphere
ASAPS 2
FTZMUF2
FTZ4
HFBC84
HFMUFES4
ICEPAC
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.
Fundamentals of propagation
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.
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
Fundamentals of propagation
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.
Geophys. Res. 65, 2635.
Hagen, J. B. and Behnke, R. A. (1976) Detection of the electron component of the
spectrum in incoherent scatter of radio waves by the ionosphere. J. Geophys. Res. 81,
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
Fundamentals of propagation
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
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
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.
183
184
Radio techniques
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.
Specications
Source
Digisonde Portable
Sounder (DPS)
University of
Massachusetts Center for
Atmospheric Research,
600 Suolk Street, 3rd
Floor, Lowell, MA 01854,
USA www.uml.edu
Canadian Advanced
Digital Ionosonde (CADI)
Scientic Instruments,
Ltd, 2233 Hanselman
Avenue, Saskatoon, CA
S7L6A7, USA
Advanced Digital
Ionosonde, IPS-71
Figure 4.5. A typical modern digital ionogram (compare with Figure 4.2).
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.
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
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
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
90
70
50
130
110
EO
6 12 18
TIME (GMT)
264 0000
1978
100
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
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
195
196
Radio techniques
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).
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
Name
Type
Reference/description
Finland
COSCAT/XMTRa
Auroral/pulsed/bistatic
Bistatic/pulsed and CW
U. K. and
Sweden
SABRE
Auroral/pulsed
Scandinavia
STARE
Auroral/pulsed/bistatic
Greenwald (1987)
Canadian
Arctic
SAPPHIRE
NE Canada
SHERPA
Polar
SUPERDARN
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
Antarctica
SYOWA
Auroral/pulsed
Peru
Jicamarca
Equatorial/pulsed/
monostatic
Radio techniques
202
Name
Type
Reference/description
Kwajalein
Altair
Equatorial/monostatic/
pulsed
Japan
MU Radar
mid-latitude, monostatic/
pulsed
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.
Figure 4.18. A map showing currently operational ISRs (courtesy of EISCAT Association).
4.2.3
Incoherent-scatter radars
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.
Note:
ACFs, autocorrelation functions
Gain (dBi)
Polarization
Feed system
Cassegrain
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.
(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.
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
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
A2 method
Not as sensitive as
some other methods
A3a method
Quite sensitive
Interference, more
complex than A1
A3b method
Sensitive
Interference
PRE method
Interference,
complex system,
more expensive than
others
MF radar can be
used to probe the
middle atmosphere
fmin method
Very sensitive to
variations in
equipment
parameters
LOF
Interference, dicult
to interpret
Satellite HF
beacon
Global coverage
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
Parameters
NAIC; Arecibo,
Puerto Rico, USA
18 N/67 W;
300 MW/315 MHz
Operational in 1971;
Gordon et al. (1971)
69.6 N; 1200 MW
64.9 N/146.9 W;
50 MW/2.84.9 MHz
38 N; 68 MW/46 MHz
Operational in 1981
Erukhimov et al. (1985)
Kharkov
50 N; 612 MHz
Moscow
Schluyger (1974)
Sura Radiophysics
Research Institute
Nizhni Novgorod
56 N; 4.59 MHz
Zimenki
56 N; 20 MW/
4.65.7 MHz
Monchegorsk
68 N; 10 MW/3.3 MHz
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.
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
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
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
4.3.4
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
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
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
Radio beacons
for scintillation
studies
Topside sounders
More-complex instrumentation
than most beacons.
219
Multibeam
Reflections
Multiple
Scattering
0.1
(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
4.4.2
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
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)
225
226
Radio techniques
Chapter 5
The high-latitude F region and the trough
5.1
5.1.1
Introduction
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
228
5.1.2
Circulation patterns
if Bz 0 (northward),
or
180 tan1 |By /Bz |
if Bz 0 (southward).
(5.1)
(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)
(2)
(3)
229
230
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.
232
The northern and southern patterns of circulation are mirror images when
the IMF is southward.
(2)
(3)
If the IMF is strongly northward, the patterns in the summer and winter
polar caps are very dierent.
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
5.2.1
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.)
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
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)
(2)
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)
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 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
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
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
5.3 Irregularities
243
(a)
600
500
400
300
200
100
0
Yearly Average
250 300 km
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
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)
(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
American Geophysical Union.)
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.
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
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
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
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.
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,
copyright by the American Geophysical Union.)
Average values of S4 at 137 MHz over a range of latitude and at various times
of day (geomagnetic local time) are shown in Figure 5.17. All these measurements
were made between autumn and early summer (October to May) at Hornsund,
Svalbard (invariant latitude 73.4), whose position is marked on the plots with an
arrow. Since the magnetic 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
(a)
257
30
40
50
60
S4 = 0.226 q = 2.792 0.200
70
0.1
10
100
Frequency (Hz)
(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)
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)
(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
5.4.1
Introduction
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
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)
(2)
(3)
(4)
(5)
an auroral enhancement;
(6)
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
from Elsevier Science.)
H(MAX)
(km)
106
263
500
500
400
400
300
300
200
200
8
6
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.
Geophys. Res. 82, 5129, 1977, copyright by the American Geophysical Union.)
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.
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.)
265
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.
(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)
(5.5)
(5.6)
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
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)
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.)
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
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
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
permission from Elsevier Science.)
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
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
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.)
1010
O+ O2
O+ N2
1011
5
1012
5
1013
10
12
2.0
3.0
4.0
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
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.
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.
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
1983, with permission from Elsevier Science.)
The ion temperature (Ti ) and the electric eld are usually increased within
the trough but the electron temperature is not usually aected.
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
??
Low
Low
B. Polar hole
Te
High
Vi
Trough type
Presumably low,
although no
measurements have
been made
Normal
Ti
NO rich
Normal
Normal
Composition
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
5.7
5.1
Boyle, C. B., Rei, P. H., and Hairston, M. R. (1997). Empirical polar cap potentials.
J. Geophys. Res. 102, 111.
Cowley, S. W. and Lockwood, M. (1997) Excitation and decay of solar wind-driven
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
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
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.
Geophys. Res. 89, 1683.
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
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.
Terr. Phys. 45, 315.
Muldrew, D. B. (1965) F-layer ionization troughs deduced from Alouette data. J.
Geophys. Res. 70, 2635.
Pike, C. P., Whalen, J. A., and Buchau, J. (1977) A 12-hour case study of auroral phenomena in the midnight sector: F layer and 6300 measurements. J. Geophys. Res. 82,
3547.
Rodger, A. S., Brace, L. H., Hoegy, W. R., and Winningham. J. D. (1986) The poleward edge of the mid-latitude trough its formation, orientation and dynamics. J.
Atmos. Terr. Phys. 48, 715.
Rodger, A. S., 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.
Atmos. Terr. Phys. 54, 1.
Rycroft, M. J. and Burnell, S. J. (1970) Statistical analysis of movements of the ionospheric trough and the plasmapause. J. Geophys. Res. 75, 5600.
Spiro, R. W. (1978) A study of plasma 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
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.
Atmos. Terr. Phys. 54, 1.
Chapter 6
The aurora, the substorm, and the E region
6.1
Introduction
By aurora people usually mean the emission of light from the upper atmosphere, but in fact there are numerous related phenomena, each being a direct or
indirect consequence of energetic particles entering the atmosphere from the magnetosphere. They include
(a)
luminous aurora;
(b)
radar aurora, by which is meant the 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)
(f)
They all exhibit a general relationship with solar activity, though often
there is no specic association with any obvious solar event. From the
285
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)
(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
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
Figure 6.1. The northern auroral zone, showing the percentage of good observing nights
when aurorae may be seen. (After E. H. Vestine, Terr. Magn. Atmos. Electricity, 49, 77,
1944, copyright by the American Geophysical Union.)
to mark the division between open and closed 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
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
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
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.)
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.)
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
6.3.1
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
AURORAL OVAL
(Feldstein 1967)
Q=1
AURORAL OVAL
(Feldstein 1967)
Q=3
AURORAL OVAL
(Feldstein 1967)
Q=5
AURORAL OVAL
(Feldstein 1967)
Q=7
296
(a)
(c)
(b)
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
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.)
(a)
500
500
299
1000
(b)
500
80
Height (km)
400
400
300
300
200
200
100
100
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)
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
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
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.)
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
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,
diuse aurorae,
radio absorption,
continuous micropulsations,
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
309
(a)
(b)
(c)
(d)
(e)
(f)
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
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)
(4)
The onset region is very localized, being less than 500 km across in the
ionosphere.
(5)
(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
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
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.)
314
(a)
Electrojet
N
Field
Aligned
Currents
Near
Earth
Neutral
Line
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.)
(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
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
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)
(2)
convecting over the north and south poles as two separate halves;
(3)
(4)
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
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)
(b)
319
(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)
(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
(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 6.24. 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.
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
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.
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 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
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 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)
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)
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,
Cambridge University Press, 1989.)
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
American Geophysical Union.)
great magnetic storm of August 1972 the E-region electron density exceeded
1.8
106 cm3 (Figure 6.29), one of the higher values of electron density observed
in the E region.
The morphology, structure, and dynamics of the auroral-E layer have been
described in some detail by Hunsucker (1975) and others referred to therein.
6.5.5
Auroral radar
Figure 6.28. The variation of electron density at three heights, showing a sharp spike in the
electron density at 0800 UT (2200 LT) on 2 March 1973. At the same time the ionosonde
registered sporadic-E above 7 MHz. Observations from Alaska, using the Chatanika ISR
and the College ionosonde. (H. F. Bates and R. D. Hunsucker, Radio Sci. 9, 455, 1974,
copyright by the American Geophysical Union.)
oval using radars operating in the VHF/UHF frequency range. The term radio
aurora is often used for the phenomena so observed, and radar aurora is equally
valid. There is now a voluminous literature on the subject.
There is an enormous 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
American Geophysical Union.)
Theory
The two most generally accepted theories to explain the observations are the twostream plasma instability and the gradient-drift mechanism. These plasma instabilities are generated in the auroral electrojet and produce electrostatic ion waves that
may scatter incident radio waves as discussed in Section 3.5.1. A necessary condition for the occurrence of these instabilities is a 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.
Figure 6.30. Four types of echo in auroral radar. The observations were with a 50-MHz
radar in March 1989, and each analysis was based on 20 s of data. Cs is the local speed of
sound. (J. D. Sahr and B. G. Fejer, J. Geophys. Res. 101, 26 893, 1996, copyright by the
American Geophysical Union.)
Broadly speaking, the diurnal and seasonal occurrence of the radio aurora is
similar to that of the visual aurora, except, of course, during daylight when the
visual aurora cannot be seen. The strongest echoes occur near 65 latitude, and
during disturbances the echoing region extends equatorward. The echoes can be
detected at any time of day, but are most pronounced near local magnetic midnight.
The phenomenon of E-region irregularities is closely related to the behavior of
the auroral electrojet. Sahr and Fejer (1996) draw attention to the problem of
modeling them globally, to their importance in radio propagation, and to gaps in
methods of data analysis.
329
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
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.)
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
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
6.7
6.2
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
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.
Akasofu, S.-I. and Chapman, S. (1972) SolarTerrestrial Physics. Oxford University
Press, Oxford.
Angelopoulos, V. (1996) The role of impulsive particle acceleration in magnetotail circulation. Third International Conference on Substorms, Versailles, France. Report ESA
SP-389, p. 17.
Birkeland, K. (1908) The Norwegian Aurora Polaris Expedition 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-
6.5
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
University Press, Cambridge.
Sahr, J. D. and Fejer, B. G. (1996)Auroral electrojet plasma irregularity theory and
experiment: a critical review of present understanding and future directions.
J. Geophys. Res. 101, 26 893.
Wilson, C. R. (1969) Auroral infrasonic waves. J. Geophys. Res. 74, 1812.
Wilson, C. R. and Hargreaves, J. K. (1974) The motions of peaks in ionospheric
absorption and infrasonic waves. J. Atmos. Terr. Phys. 36, 1555.
Wilson, C. R., Hunsucker, R. D., and Romick, G. J. (1976) An auroral substorm investigation using Chatanika radar and other geophysical sensors. Planet. Space Sci. 24, 1155.
Conference reports
Akasofu, S.-I. (ed.) (1980) Dynamics of the Magnetosphere. Reidel, Dordrecht.
McCormac, B. M. (ed.) (1967) Aurora and Airglow. Van Nostrand Reinhold Co., New
York.
McCormac, B. M. and Omholt, A. (eds.) (1969) Atmospheric Emissions. Van Nostrand
Reinhold Co., New York.
McCormac, B. M. (ed.) (1971) The Radiating Atmosphere. Reidel, Dordrecht.
Meng, C.-I., Rycroft, M. J., and Frank, L. A. (eds.) (1991) Auroral Physics. Cambridge
University Press, Cambridge.
Chapter 7
The high-latitude D region
7.1
Introduction
The 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)
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
7.2.1
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)
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.
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.
Hargreaves et al., J. Atmos. SolarTerr. Phys. 59, 872, copyright 1997, with permission from
Elsevier Science.)
more limited range of latitude (probably less than 200 km) than the onset itself,
which in some cases may be tracked over a wide range of L values if some time
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.
Hargreaves et al., J. Atmos. SolarTerr. Phys. 59, 872, copyright 1997, with permission from
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.
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.
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
1990, with permission from Elsevier Science.)
smooth, with little structure (Figure 7.8(a)). Some of these are modulated with
quasi-periodic pulsations having a period of several minutes (Figure 7.8b), in the
range of Pc4 and Pc5 (Section 2.5.6). The SVA is spatially more widespread than
the spike and the preceeding bay.
(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
0.05
0.10
0.15
Absorption (dB km 1)
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
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
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.
5.5
5.59
48
5.5
5.5
5.5
Author
Kavadas (1961)
Ansari (1965)
Sudden-onset events
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
Elliptical
Circular
Elliptical
0.5 at 380 km
Results
Hourly values
Hourly values
Hourly values
Values correlated
57
12 months
17 months,
AugustJanuary
Disturbed nights
Circular
Circular
Elliptical or
circular
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
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.
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
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.
(b)
The onset is earlier and at lower latitude when the level of magnetic disturbance is greater.
(c)
(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)
(f)
(g)
361
362
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.)
1938
Nothing
gap
Patches drifting
towards equator
17481805
Main event from north
1830
173639
Spike
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.
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
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.
365
100
(a)
80
60
40
20
0
0
10
20
30
40
Ap
100
(b)
80
60
40
20
0
0
10
20
30
20
30
Ap
5
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.)
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
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
369
(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.)
1
(log A log Am ) 2
exp
d(logA),
2
2 2
(7.4)
f (log A)d(logA).
(7.5)
log A
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
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
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 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)
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.
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
375
(7.7a)
105
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)
dt
E(mV m ) 5.88
10
d(1/L2 )
1
4
1
(7.10)
200
175
175
160320 keV
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
permission from Elsevier Science.)
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
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
383
2.0
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
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
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 )
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
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
20
Year (19 )
10
9
8
7
(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
387
(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.
(a)
1.0
Relative occurrence
388
0.5
M A M
S O
N D
(b)
10
Month
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
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
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
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
propagation from Sun to Earth i.e. from the Sun to the boundary of the
magnetosphere;
(b)
(c)
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
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
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.
393
(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)
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
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)
(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)
(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
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
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.
50
60
70
80
90
10 7
(4)
(3)
(2)
(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
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)
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
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.
Geophys. Res. 81, 4579, 1976, copyright by the American Geophysical Union.)
Figure 7.44. Twilight variations expressed as hysteresis curves. The reference station is South Pole (solar elevation 7), and the curves are described
counter-clockwise, implying larger absorption at sunset than at sunrise for the same solar elevation. (H. H. Sauer. J. Geophys. Res. 73, 3058, 1968,
copyright by the American Geophysical Union.)
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.
(7.16a)
(7.16b)
(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
(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
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
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
1996, with permission from Elsevier Science.)
Most strikingly, the echoes are clearly a summer phenomenon, occurring from
June to August only, with a maximum in July in the northern hemisphere (Palmer
et al., 1996). The occurrence also varies during the day. There are maxima near
noon and midnight, and minima in the morning and the evening hours. The percentage occurrence, though not very well established, is some 50%75% of days at
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
1990, with permission from Elsevier Science.)
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
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.
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
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.
J. Atmos. Terr. Phys. 55, 751.
Reagan, J. B. and Watt, T. M. (1976) Simultaneous satellite and radar studies of the Dregion ionosphere during the intense solar particle events of August 1972. J. Geophys.
Res. 81, 4579.
Reid, G. C. (1961) A study of the enhanced ionisation produced by solar protons
during a polar cap absorption event. J. Geophys. Res. 66, 4071.
Reid, G. C. (1967) Ionospheric disturbances. In Physics of Geomagnetic Phenomena
(eds. Matsushita and Campbell), p. 627. Academic Press, New York.
Reid, G. C. (1986) Solar energetic particles and their 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.
J. Geophys. Res. 73, 3058.
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.
Terr. Phys. 57, 905.
Weeks, L. H., CuiKay, R. S., and Corbin, J. R. (1972) Ozone measurements in the
mesosphere during the solar proton event of 2 November 1969. J. Atmos. Sci. 29,
1138.
7.4
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.
J. Geophys. Res. 94, 5199.
Ecklund, W. L. and Balsley, B. B. (1981) Long-term observations of the Arctic mesosphere with the MST radar at Poker Flat, Alaska. J. Geophys. Res. 86, 7775.
Hoppe, U.-P., Hall, C., and 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
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
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
High-latitude propagation: 1
Figure 8.1. The variation of phase-speed with VLF frequency for a perfectly conducting
Earth, sg $, r 2
105 for (a) mode number 1 and (mode number 2) (from Davies, 1990).
Figure 8.2. The diurnal variation of WWVL at 20 kHz over the 113-km path from Fort
Collins to Wiggins, CO (from Davies, 1990).
421
422
High-latitude propagation: 1
Figure 8.3. The average amplitude spectrum of the lower ELF band. Note the Russian
ELF transmission at 82 Hz and the and Bannister power-line frequencies (50 and 60 Hz)
and their harmonics (from Fraser-Smith, 1998).
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
High-latitude propagation: 1
Figure 8.5. A map showing the geometry of the ELF path from WTF to the Gulf of
Alaska (from Field et al., 1985).
D-region height. Systems such as the US 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
Figure 8.11. VLF propagation paths to College, Alaska during the 19611964 study (after Bates and Albee, 1966).
HYPERBOLIC
NORWAYHAWAII)
2
429
RANGING
(NORWAY)
30s = 3 nm HYPERBOLIC
6 nm RANGING
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
High-latitude propagation: 1
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
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
High-latitude propagation: 1
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 8.15. The location of the auroral oval at 1000 UT on 4 February 1986 Q0. (after
Hunsucker, 1988).
433
434
High-latitude propagation: 1
Figure 8.16. Signal behavior during a moderately disturbed equinoctial day (Ak 20), 8
September 1986 (from Hunsucker, 1988).
Figure 8.17. The location of the auroral oval for a moderately disturbed day, 8 September
1985, at 1000 UT, 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
High-latitude propagation: 1
436
Frequency
(kHz)
Power
output
(KW)
Call letters
Location
KSKO
McGrath,
Alaska
870
KTWO
Casper,
Wyoming
1030
50
KFAX
San Francisco,
California
1100
50
KGA
Spokane,
Washington
1510
50
Short northsouth
auroral path.
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
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)
(3)
(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.
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
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
High-latitude propagation: 1
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).
441
442
High-latitude propagation: 1
Figure 8.21. A comparison of the College, Alaska signal received at Kiruna and College
groundscatter for 18 MHz on 4 December 1958 (after Owren et al., 1959).
Figure 8.22. The average signal strength at Kiruna, Sweden for the month of December
1958 (from Owren et al., 1958).
8.4 HF propagation
443
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
High-latitude propagation: 1
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).
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).
High-latitude propagation: 1
448
Figure 8.25. A map showing three great-circle HF-propagation paths and the normal and
expanded auroral zones (after Owren et al., 1963).
circuits shown in Figure 8.25 during the magnetic storm of 1617 August 1959 is
illustrative of near-sunspot-maximum conditions.
8.4.2
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
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.
449
High-latitude propagation: 1
450
24
20
16
12
08
04
00
1
11
13
15
17
19
21
23
25
27
29
31
AUGUST
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
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
High-latitude propagation: 1
452
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
High-latitude propagation: 1
signals present, hours with blocking interference, and outages. The heights of the
black and dotted columns measure periods of reception and interference, respectively. Transmitter or receiver 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
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
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
e e e e e e e e e e e e e e e e
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
e e e e e e e e e e e e e e e e
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
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
e e e e e e e e e e e
e e e e e e e e e e e
(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
e e e e e e e e e e e e e e e e
noisy period
DD
10
e e e e e e e e e e e e e e e e
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
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
High-latitude propagation: 1
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
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
High-latitude propagation: 1
Figure 8.42. The solar-cycle variation of the Thule 18-MHz signal. Hourly average signal
levels are plotted on the ordinate (from Owren et al., 1963).
8.4 HF propagation
469
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
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
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
High-latitude propagation: 1
8.4 HF propagation
8.4.5
Sweep-frequency experiments
Forward oblique sounding investigations near sunspot minimum
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
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
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
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
High-latitude propagation: 1
Figure 8.48. The lower plot shows the diurnal variation of the auroral-E maximum
observed frequency MOF for the same period (after Bates and Hunsucker, 1964).
is indicated between 0730 and 1230 UT (21300230, 150 WMT) in the average
MOF curve for Thule, which corresponds to the diurnal auroral peak at
College.
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
High-latitude propagation: 1
Figure 8.50. Typical winter records from Thule, Greenland. The long-pulse record is shown
above and the short-pulse record is shown below (after Bates and Hunsucker, 1964).
path are shown in Figure 8.52, with the direct path (probably auroral-E) at the
proper time delay, plus several distinct NGC modes. The time delay and lack of
retardation of the mode structure do not allow a multiple-hop interpretation,
so we postulate a NGC mode.
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
Most of these results are taken from a survey paper by Hunsucker and Bates
(1969).
479
480
High-latitude propagation: 1
Figure 8.52. Direct and 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).
8.4 HF propagation
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
High-latitude propagation: 1
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
483
Figure 8.54. Forward oblique ionograms recorded on the Palo Alto-to-College path: (a) and (b) were recorded during October
(SSN19.7), and (c) and (d) during July 1965 (SSN15.5). Time marks are 1.0 ms apart. One microsecond pulse per channel
was transmitted per 1000 frequency channels between 4 and 24 MHz (from Bates and Albee, 1966).
(b)
(a)
(d)
(c)
Figure 8.55. A long-tailed trace sequence recorded on 18 July 1964 (SSN10.3) on the Thule-to-College path. Times are UT.
The pulse width is 1 ms (after Bates and Albee, 1996).
486
High-latitude propagation: 1
Figure 8.56. A long-tailed trace sequence recorded on 18 July 1964 on the Palo Alto-toCollege path. Times are UT. The pulse width used was 1 ms (after Bates and Albee, 1966).
to the decrease of the F1-layer critical frequency with latitude relative to that of
the F2 layer.
Figure 8.59 contains an oblique ionogram that was derived from the vertical
ionogram in Figure 8.59. For simplicity only E and F modes are shown; combination 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).
489
490
High-latitude propagation: 1
(a)
(c)
(b)
(d)
Figure 8.60. Records showing highly delayed signals. Traces in (a) and (b) were obtained on
the Palo Alto-to-College path with four 100-s pulses per channel, per 100 frequency channels between 4 and 24 MHz. Records (c) and (d) were obtained using two 1.0-ms pulses per
channel on the Fort Monmouth- and Palo Alto-to-College paths, respectively (from Bates
and Albee, 1966).
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
High-latitude propagation: 1
492
to allow penetration at one point but not at a point further along the path. A tilted
F2 layer might not be necessary, but a tilt would help by gradually changing the
angle of incidence of the propagating wave.
The proposed model is speculative and might not be necessary to explain the
observed records. The high-order-mode model, though, will not satisfactorily
explain the existence of the single DLT trace on a record, whereas ducting will,
because only one ducted mode would generally be expected. A further observational point in favor of the idea of a ducted mode is the extreme variability of the
single long-tailed trace. Within the span of several soundings (20 min apart) the
trace appeared and disappeared. The high-ray trace was well 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.
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
High-latitude propagation: 1
Figure 8.62. A map showing the SodankylLindau HF path in relation to the auroral
zone, Arctic and other instrumented stations (after Rose, 1967).
8.4 HF propagation
Figure 8.63. A vertical section through the LindauSodankyl HF propagation path (after
Moller, 1964).
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
High-latitude propagation: 1
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).
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).
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 8.72. Some eects of a moderate geomagnetic disturbance on 7 October 1958 on the
SodankylLindau HF circuit (after Moller, 1964).
501
502
High-latitude propagation: 1
Figure 8.73. The AndyaCollege HF average monthly propagation spectrum for summer
1964 (D denotes disturbed and Q denotes quiet periods) (after Bartholomew, 1969).
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
High-latitude propagation: 1
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).
8.4 HF propagation
Figure 8.78. AndyaCollege HF multipath and propagation outage for winter 1964 (after
Bartholomew, 1969).
505
High-latitude propagation: 1
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
High-latitude propagation: 1
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
High-latitude propagation: 1
Figure 8.86. Plots of observed MOF distribution (shaded area), median frequencies
(broken line), MUFs predicted from vertical ionosonde data (solid line), and periods with
predicted screening by the E-layer (the heavy solid line near the bottom of the MOF plot)
for 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
High-latitude propagation: 1
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).
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
High-latitude propagation: 1
Figure 8.92. The same as Figure 8.91, but for June 1960 (from Tveten, 1962)
Figure 8.93. Periods of reception of 13- and 17-MHz transmissions from McMurdo to
Thule during 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).
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.
518
High-latitude propagation: 1
Figure 8.95. The percentage of time (solid line) and percentage of half-hour periods
(dashed line) for which auroral absorption equaled or exceeded 1.0 dB, as functions of geomagnetic latitude. Locations of the stations are indicated by the two-letter abbreviations on
the abscissa (after T. R. Hartz, L. E. Montbriand and E. L. Vogan. A study of auroral
absorption at 30 Mc/s. Can. J. Phys., 41, 581 (1963).)
8.4 HF propagation
519
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
High-latitude propagation: 1
Figure 8.98. The ight
plan for groundair
trials of 1516 December
1960 (from Jull, 1962)
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
High-latitude propagation: 1
Figure 8.100. A map showing the Wales-to-Fairbanks, Alaska propagation path in relation
to the equatorward edge of the auroral oval as a function of Kp.
were that, during PCA events of low or moderate intensity, the optimum 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
High-latitude propagation: 1
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
High-latitude propagation: 1
Figure 8.104. The occurrence of AE for Summer 1992 (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
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
High-latitude propagation: 1
8.4 HF propagation
529
SondrestomKeavk HF
experiment
Path location
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
Transmitter power
100 W CW
170 W pulse
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
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
High-latitude propagation: 1
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
531
High-latitude propagation: 1
532
HF propagation at high latitudes, then, is determined by the location of the propagation path in relation to the intersections of the path with the auroral and polar
D region and the E-region and F-region 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.
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
High-latitude propagation: 1
Section 8.4
Auterman, J. L. (1962) Fading correlation bandwidth and short-term frequency stability measurements on a high-frequency transauroral path. NBS Tech. Note 165.
Bartholomew, R. R. (1966) Results of a High-latitude HF Backscatter Study. Stanford
Research Institute, Menlo Park, California.
Bates, H. F. and Hunsucker, R. D. (1964) HF/VHF Auroral and Polar Zone Forward
Sounding. Geophysical Institute, University of Alaska, Fairbanks, Alaska.
Bates, H. F and Albee, P. R. (1966) On the Strong 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.
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
High-latitude propagation: 1
Cannon, P. S., Weitzen, J. A., and Ostergaard, J. (1996) The relative impact of meteor
scatter and other long distance high latitude propagation modes on VHF communication systems. Radio Sci. 31.
Ostergaard, J. C., Rasmussen, J. E., Sowa, M. J., McQuinn, J. M., and Kossey, P. A.
(1985) Characteristics of high-latitude meteor scatter propagation parameters over the
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
537
High-latitude propagation: 2
538
9.2
9.2.1
Ionospheric ray-tracing
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).
Semi-empirical
Model
High-latitude F region
h3002000 km
High-latitude F region
(h150600 km)
High-latitude features
Table 9.1. Applicability of ionospheric models to the high-latitude ionosphere (from Schunk, 1996)
No longer available
Coupled Thermosphere
Ionosphere Model (CTIM)
FullerRowell et al.)
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
Available in FORTRAN on
supercomputer; the authors
welcome collaboration
Empirical model
Model
NCAR Thermosphere
IonosphereMesosphere
Electrodynamics General
Circulation Model (TIGCM)
(Roble)
USU Electrodynamic
Ionospheric Model (Zhu)
International Reference
Ionospheric (IRI) (Bilitza, 1997)
High-latitude enhancements
(Rawer and Bilitza, 1995)
High-latitude features
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
High-latitude propagation: 2
546
Figure 9.4. The daytime average absolute dierence between LWPC and measurements
parametric in frequency and distance interval (from Ferguson, 1995).
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
High-latitude propagation: 2
Figure 9.5. The night-time dierence in average absolute absorption as in Figure 9.4 (from
Ferguson, 1995).
549
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
High-latitude propagation: 2
Figure 9.7. The HF propagation path, between Clyde River, Canada and Leicester, UK
(after Gikas, 1990).
the ICED electron density 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).
Propagation path
Results
Reference
Gikas (1990)
IONCAP-PC 2.5
AMBCOM
ICEPAC
Absolute errors
from 0.3 to 26.4 dB
Burtch 1991
552
High-latitude propagation: 2
Figure 9.9. The average for Site D, winter 1989 of AMBCOM predictions from measured
values: total average error 13.5 dB, standard deviation 28.9 dB and total number of samples
2919 (after Wilson, 1991).
Figure 9.10. The standard deviation of IONCAP-PC 2.5 prediction errors versus frequency
for the Winter 1989 campaign (after Tsolekas, 1990).
9.2.5
553
554
High-latitude propagation: 2
ICEPAC (Ionospheric
Conductivity and
Electrodynamics,
Prediction, etc.)
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
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)
Outputs
Inputs
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.
Rose
PROPHET 4.3.2
Characteristics of the
model
558
High-latitude propagation: 2
Figure 9.13. The PENEX signal amplitude for 27 September 1993, f 11.0 MHz.
Figure 9.14. The PENEX signal amplitude for 24 August 1993, f 11.0 MHz.
559
High-latitude propagation: 2
560
This was an interesting multihop 5925-km path, which, under quiet geomagnetic conditions (Kp 3), can be considered to be a mid-latitude path.
Measurements were made only during August and September 1993, and there was
usable data for 70% of the time.
As geomagnetic activity increases, however, this increasingly becomes an
auroral path, with the entire path lying within the auroral oval when Kp 8
(magnetic-storm conditions) and the 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
(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%).
565
566
High-latitude propagation: 2
Figure 9.17. The HF propagation metric on 4 November 1993. The amplitude of the
signal is 16.8 MHz.
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
High-latitude propagation: 2
568
9.3
9.4
Figure 9.19. The USU Assimilating Ionospheric Model images taken once per hour for a
geomagnetically quiet day (12 November 1993). The approximate great-circle paths from
Wales to Seattle and to Fairbanks are shown as two sloping lines in the bottom left-hand
panel (after Schunk and Sojka, 1999).
569
High-latitude propagation: 2
570
(2)
(3)
(4)
(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.
Figure 9.20. The same as in Figure 9.19, but for a geomagnetically disturbed day (after
Schunk and Sojka, 1999).
571
High-latitude propagation: 2
572
9.5
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
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).
573
High-latitude propagation: 2
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
High-latitude propagation: 2
576
(2)
(3)
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
High-latitude propagation: 2
578
Table 9.7. Selected HF-channel parameters recorded during the TAHFE (after
Bliss et al., 1987)
Parameter
Units
Description
DOY
TOD
FRQ
SNoR
DS
MP
BER75
BER300
Days
Hours
Megahertz
dB Hz
Hertz
Milliseconds
Ratio
Ratio
Day of year
Time of day (universal time) in decimal hours
Operating frequency
Signal-to-noise power-density ratioa
Doppler frequency spreada
Multipath time-delay spreada
Bit error rate for 75-bps data (10000 bits)
Bit error rate for 300-bps data (10000 bits)
Notes:
ALQA (advanced link-quality analyzer) measurement
Collection interval
TAHFE 1A
TAHFE 1B
TAHFE 2
Note:
a
Preliminary analysis on TAHFE 1B.
579
High-latitude propagation: 2
580
22.00
20.00
FREQUENCY (MHz)
18.00
16.00
14.00
12.00
10.00
8.00
6.00
0.00
2.00
4.00
6.00
Figure 9.26. The diurnal frequency variation for 12 November 1986 (from Bliss et al.,
1986).
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
The most promising technique for ameliorating deleterious eects on HF highlatitude communication circuits is realtime channel evaluation (RTCE), which is
Figure 9.28. Typical SNR, frequency, and multipath spread versus time of day for 12
November 1986 (from Bliss et al., 1986).
581
582
High-latitude propagation: 2
Figure 9.29. Doppler spread (RMS) characterization for 12 November 1986 (from Bliss et
al. 1987).
Figure 9.30. A frequency-shift-keying (FSK) error-rate comparison for 75 and 300 BPS for
12 November 1986 (from Bliss et al., 1986).
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:
Class two:
583
584
High-latitude propagation: 2
Figure 9.31. The geometry of HF propagation paths in the Northern Experiment (from
Goodman et al., 1997).
Figure 9.33. Percentage availabilities of signals at Iceland for selected frequency groups and transmitter-station combinations. (For each group of four,
the ordering from left to right is 11, eight, six, and four frequencies, respectively (from Goodman et al., 1997).
High-latitude propagation: 2
586
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
High-latitude propagation: 2
Figure 9.35. A schematic illustration of DAMSON analysis program display (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).
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
9.6.1
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
High-latitude propagation: 2
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
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
High-latitude propagation: 2
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
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
High-latitude propagation: 2
Figure 9.40. The same as Figure 9.38, but for f9.410 MHz (after Blagoveschenskaya et
al., 1998).
9.6.4
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
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
0
500 400 300 200 100
10
15
20
25
400
500
TECU
day 344
day 020
day 028
100
200
300
400
day 365
0
500 400 300 200 100
10
15
20
25
30
35
40
500
day 336
day 351
100
200
300
0
500 400 300 200 100
10
15
20
25
30
35
40
400
500
100
200
100
200
300
Distance from closest approach to midpoint (km)
day 350
300
day 350
TEC
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
TECU
TECU
400
400
500
500
High-latitude propagation: 2
600
AEI strength
AEI duration
Passes
Prediction
no maybe
yes
21
dB
34
min
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).
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
High-latitude propagation: 2
Figure 9.46. The method
of comparison between
DAMSON measurements and the simulated
performance of modems
(from Jodalen et al.,
2001).
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
High-latitude propagation: 2
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
oblique HF propagation and its dependence on auroral oval position. IEEE Trans.
Antennas Propagation 44, 383388.
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.
J. Atmos. Terr. Phys. 37, 15631570.
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
High-latitude propagation: 2
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.
Goodman, J. M. (1992) HF Communication Science and Technology. Van Nostrand
Reinhold, New York.
Hunsucker, R. D. and Owren, L. (1962) Auroral sporadic-E ionization. J. Res. NBS D
66, 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. (1992) Auroral and polar-cap ionospheric eects on radio propagation. IEEE Trans. Antennas Propagation 7, 818828.
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.
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.
Hunsucker, R. D. (1992) Auroral and polar-cap ionospheric eects on radio propagation. IEEE Trans. Antennas Propagation 7, 818828.
609
610
High-latitude propagation: 2
Section 9.5
Angling, M. J., Cannon, P. S., Davies, N. C., Willink, T. J., Jodalen, V., and Jundborg,
B. (1998) Measurements of Doppler and multipath spread on oblique high-latitude
HF paths and their use in characterizing data modem performance. Radio Sci. 33,
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.
Goodman, J. M. (1992) HF Communication Science and Technology. Van Nostrand
Reinhold, New York.
Goodman, J. M., Ballard, J. and Sharp, E. (1997) A long-term investigation of the HF
communication channel over middle and high latitude paths. Radio Sci. 32,
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
oblique HF propagation and its dependence on auroral oval position. IEEE Trans.
Antennas Propagation 44, 383388.
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
Antennas Propagation 144, 215220.
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
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
613
Index
614
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
geomagnetic cavity 63
geomagnetic eld 6163
geopotential height 7
distribution height 21
E layer at night 27
E region
at high latitude 322332
aeronomy of 2631
disturbed auroral 323326
polar 323
quiet auroral 323
sporadic-E phenomenon in 2731
eddy diffusion in the atmosphere 9
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
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
at high latitude 2930
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