Ionosferic Radio Propagation - Article

Chapter 10
IONOSPHERICRADIO WAVE PROPAGATION

Section 10.1
Section 10.2
Section 10.3
Section 10.4
Section 10.5
Section 10.6
Section 10.7
Section 10.8
Section 10.9
Section 10.10
S. Basu, J. Buchau, F.J. Rich and E.J. Weber

E.C. Field, J.L. Heckscher, P.A. Kossey, and E.A. Lewis
B.S. Dandekar
L.F. McNamara
E.W. Cliver
G.H. Millman
J. Aarons and S. Basu
J.A. Klobuchar
J.A. Klobuchar
S. Basu, M.F. Mendillo
The series of reviews presented is an attempt to introduce

in HF communications is leading to a rejuvenation of the
ionospheric radio wave propagation of interest to system
global ionosonde network.
users. Although the attempt is made to summarize the field,
the individuals writing each section have oriented the work
10.1.1.1 Ionogram. Ionospheric sounders or ionosondes
in the direction judged to be most important.
are, in principle, HF radars that record the time of flight or
We cover areas such as HF and VLF propagation where
travel of a transmitted HF signal as a measure of its ionosthe ionosphere is essentially a "black box", that is, a vital
pheric reflection height. By sweeping in frequency, typically
part of the system. We also cover areas where the ionosphere
from 0.5 to 20 MHz, an ionosonde obtains a measis essentially a nuisance, such as the scintillations of transurement of the ionospheric reflection height as a function
ionospheric radio signals.
of frequency. A recording of this reflection height measFinally, we have included a summary of the main feaurement as a function of frequency is called an ionogram.
tures of the models being used at the time of writing these Ionograms
can be used to determine the electron density
reviews. [J. Aarons]
distribution as a function of height, Ne(h), from a height
that is approximately the bottom of the E layer to generally
the peak of the F2 layer, except under spread F conditions
10.1 MEASURING TECHNIQUES
or under conditions when the underlying ionization prevents
measurement of the F2 layer peak density. More directly,
ionosondes can be used to determine propagation conditions
10.1.1 Ionosonde
on HF communications links.
Two typical ionograms produced by a standard analog
For more than four decades, sounding the ionosphere
ionosphericsounder using film recording techniquesare shown
with ionospheric sounders or ionosondes has been the most
in Figure 10-1. The frequency range is 0.25 to 20 MHz
important technique developed for the investigation of the
(horizontal axis), and the displayed height range is 600 km,
global structure of the ionosphere, its diurnal, seasonal and
with 100km height markers. The bottom ionogram is typical
solar cycle changes, and its response to solar disturbances.
for daytime, showing the signatures of reflections from the
Even the advent of the extremely powerful incoherent scatter
E, F1 and F2 layers. The cusps, seen at various frequencies
radar technique [Evans, 1975], which permits measurement
(where the trace tends to become vertical) indicate the soof the complete electron density profile, electron and ion
called critical frequencies, foE, foF1, and foF2. The critical
temperatures, and ionospheric motions, has not made the
frequencies are those frequencies at which the ionospheric
relatively inexpensive and versatile ionosonde obsolete. On
sounder signals penetrate the respective layers. These frethe contrary, modern techniques of complex ionospheric
quencies are a measure of the maximum electron densities
parameter measurements and data processing [Bibl and
of the respective layers. Since the densities vary with time,
Reinisch, 1978a; Wright and Pitteway, 1979; Buchau et al.,
ionosphericsounding is used to obtain informationon changes
19781 have led to a resurgence of interest in ionospheric
in the critical frequency and other parameters of the electron
sounding as a basic research tool, while a renewed interest
density vs height profile.
10-1
CHAPTER 10
km
700
600
400
NIGHT
AM BAND
300
.25
2
DAY
6 7 8
9 10
20 MHz
Figure 10-1. Typical midlatitude day and nighttime ionograms, recorded by a C-4 ionosonde at Boulder, Colorado. The daytime ionogram shows reflections
from E, Es, F1 and F2 layers; the nighttime ionogram those from Es and F2 layers.
The ionogram (Figure 10-1) shows signatures of various

phenomena that complicate the process of ionospheric
sounding or the ionogram analysis. Superimposed on the
primary F layer echo trace is a similar but not identical
trace, shifted up in frequency: the so-called extraordinary
or X component. The primary trace is called the ordinary
or 0 component. The echo trace is split into two traces due
to effects of the earth's magnetic field. A second trace similar to the primary trace is seen at twice the range, a multiple
reflection. Only a small fraction of the wave energy is received by the antenna after it has returned from the ionosphere. Most of the returned energy is reflected back from
the ground and provides the first multiple (second order
echo) at twice the range. If the ionosphere is a good reflector,
and losses in the D region are low, additional reflections
can be observed. Figure 10-1 (Night) shows a second multiple (third order echo) for part of the Es trace. It is easy
to see that slopes increase by a factor that corresponds to
the order of the echo.
10-2
Finally, we see vertical bands in the frequency range

from 0.5 to 1.7 MHz, the signature of radio frequency
interference (RFI) in an ionogram, here from the AM band.
RFI can become severe enough to prevent the recording of
ionospheric echoes; for example, interference masks part or
all of the E layer trace below 1.7 MHz.
The top of Figure 10-1 shows a typical nighttime ionogram. The E and F1 traces have disappeared because these
layers dissipate after sunset. (Residual nighttime E region
ionization of low density can be observed in the absenceof
low sporadic E layers at stations with low RF1 and large
antennas.) Echoes from a sporadic E layer (Es) and F2 layer
echoes and their multiples are clearly visible in Figure
10-1. At times a brushlike spreading of the F2 layer cusps
is observed. It is called spread F and is caused by small
scale irregularities embedded in the ionosphere and ripples
in the equidensity contours on the order of hundreds of
meters to kilometers. For a detailed discussion of spread F
see Davies [1966] and Rawer and Suchy [19671; for a dis-
IONOSPHERIC RADIO WAVE PROPAGATION

cussion of the occurrence and global distribution see Herman
[19661. The nighttime ionogram also shows increased RFI
bands at higher frequencies. Because the D layer disappears
at night, HF propagation over large distances is possible.
This long distance propagation is heavily used for broadcasting by commercial users and for shortwave radio communications by government services and radio amateurs.
Fortunately the ionosonde's own echoes also increase in
amplitude due to the disappearance of the D layer, reducing
to some extent the effect of increased propagated noise on
the systems overall signal-to-noise ratio.
10.1.1.2 Principles of Ionospheric Sounding. The concept of ionospheric sounding was born as early as 1924,
when Breit and Tuve [1926] proved the existence of an
ionized layer with the reception of ionospheric echoes of
HF pulses transmitted at 4.3 MHz from a remote transmitter
(distance 13.8 km). This, during the next decade, led to the
development of monostatic ionospheric sounders by the National Bureau of Standards and the Carnegie Institution.
Even today the principles used by Breit and Tuves constitute
the principles on which most ionospheric sounders are based.
These are the transmission of HF pulses and the measurement of their time of flight to the reflection level. For a
short historical review of the development of ionospheric
sounders see Villard [1976].
Ionospheric sounding takes advantage of the refractive
properties of the ionosphere. A radio wave propagating into
the ionospheric plasma encounters a medium with the refractive index (in the absence of the earth's magnetic field
B, and ignoring collisions between electrons and the neutral u
atmosphere)
= 0.009
Ne = 1.24 x 104 f2n
(10.4)
(10.5)
where fN is in MHz and Ne in electrons/cm3 . The plasma

frequency is the natural frequency of oscillation for a slab
of neutral plasma with the density Ne after the electrons
have been displaced from the ions and are allowed to move
freely. For further discussions of the relation of u to the
wave propagation see Davies [1966].
Peak densities of the ionospheric layers vary between
l0 4 and > 106el/cm3 . Inserting these numbers into Equation
(10.4) gives a plasma frequency range from 1 to > 9 MHz;
this is the reason for the frequency range (0.5 MHz <f <
20 MHz) covered by a typical ionosonde. The low densities
of the D layer can only be probed with low frequencies
< 250 Hz, requiring large antennas and complex processing/
analysis techniques and are not directly measurable by the
standard ionosondes (for details see Kelso [1964] and references therein). Indirectly the D region ionization is measured by the integral absorption effects that it imposes on
the HF waves propagating through it to the E or F region
reflection levels (see discussion of fmin).
The inclusion of the magnetic field in the formula for
the refractive index leads to the well known Appleton dispersion formula (dispersion means that the refractive index
depends on the propagating frequency) for a magnetized
plasma, here given for the case of no collisions, generally
valid for frequencies > 2 MHz, in the E and F regions.
2
2X(1 - X)
(10.6)
where
with
(10.2)
rm f
(10.7)
and
e, Eo,
and m are natural constants, Ne is the electron density,
and f is the wave frequency. Below the ionosphere, Ne = 0,
and u = 1. Within the ionosphere, Nc > 0, and u < 1.
At a level where X = 1,
(10.3)
(10.3)
the refractive index u becomes zero. The wave cannot propagate any farther and is reflected. The quantity fN, which
relates the electron density to the frequency being reflected,
is called the plasma frequency. Inserting the natural constants into Equation (10.3) permits us to deduce the useful
relation between electrondensity and plasma frequency (which
is identical to the probing frequency being reflected)
eB
(10.8)
where fH is the gyrofrequency, the natural frequency at

which free electrons circle around the magnetic field lines.
BL,T are the components of the magnetic field in the direction
of (longitudinal) or perpendicular to (transverse) the wave
normal. Inserting the constants into Equation (10.8) leads
to the useful relation for the gyrofrequency
(10.9)
where fH is in MHz and B in gauss (1 gauss = 10 - 4 tesla).
The refractive index given in Equation (10.6) shows,
by the + solution to the square root, that in a magnetized
10-3
CHAPTER 10
plasma two and only two "characteristic" waves can propagate. These two characteristic waves are called the ordinary
or o-component and the extraordinary or x-component seen
in the ionogram shown in Figure 10-1. A radio wave with
arbitrary (often linear) polarization will split in the ionospheric medium into two characteristic or o-and x-components, which in general propagate independently.
The reflection condition u = 0 gives two solutions for
X; for the + sign (o-component)
detailed discussion see Davies [1966] and Chapter 10 of

Budden [1961].
As a result, the actual reflection height h is smaller than
the so-called virtual height h', which is derived, assuming
propagation in the medium with the speed of light from
ct
2
with t the round trip travel time of the pulse. Or since
X = 1
(10.10)
(10.16)
as in the no-field case, Equation (10.3); for the - sign (xcomponent)
X = I -
Y.
then
(10.11)
At the reflection level for the O-component the plasma frequency equals the probing frequency, fN = f. The x-component is reflected at a lower level that depends on the local
magnetic field strength. It can be shown that the critical
frequencies fo and fx, for fo > fH, are related by
h' > h.
As stated before, one of the main objectives of ionospheric

sounding is the determination of h(f), which through the
relation between f and Nc, Equations (10.3) and (10.4) represents the desired function NC(h). Since the group travel
time is
(10.12)
that is, the magneto-ionic splitting (due to the presence of

the magnetic field in the ionospheric plasma) depends on
the local magnetic field strength and therefore varies, from
station to station. For a typical midlatitudestation, B = 0.5G
and from Equation (10.9) we determine f,, = 1.4 MHz,
leading to the fo-fx separation of -0.7 MHz seen in Figure
10-1. A solution X = I + Y exists for frequencies below
the gyrofrequency fH. For details see Davies [1966].
Using ionograms to determine the true height electron
density profile Nc(h) is further complicated by the slowingdown effect that the ionization below the reflection level
has on the group velocity of the pulse. While the phase
velocity v of the wave is
(10.13)
it can be shown that the group velocity u, defined as the
propagation velocity of the pulse envelope, is given for the
no-magnetic field case by
c
(10.14)
f)
where u'(h,f) is the group refractive index. Therefore, while

the phase velocity increases above the speed of light in a
plasma , the group velocity, the velocity at which the energy
propagates, slows down (u < 1 in a plasma). For a more
10-4
(10.17)
dz,
(10.18)
the virtual height is related to the group refractive index by
f] dz.
(10.19)
If the electron density NC(h)is considered as a function of

the height h above the ground, u is also a function of h
and the problem is now to solve the integral equation (10.19),
for given values of h'(f) obtained from the ionogram. The
techniques used to solve this equation are known as true
height analysis for which in general numerical methods are
used; they are discussed in detail in a 1967 special issue of
Radio Science.
10.1.1.3 Analog Ionosonde. The general principle of an

ionospheric pulse sounder is shown in Figure 10-2 [Rawer
and Suchy, 1967]. A superheterodyne technique is used to
both generate the transmitted pulse of frequency fT and to
mix the received signals back to an intermediate frequency
or IF for further amplification. Tuning the receiver mixer
stage so that its output frequency is equal to the frequency
of the fixed frequency (pulsed) oscillator fc, and using a
common variable local oscillator fo, ensures that the receiver
and transmitter are automatically tuned for every value of
2I
Detectorand
I-f
Ii
Receiverm/)er
"
/T-Ampofer
V4?eo-Amrp/feer
,/ fi'
_CT
//morkersro
Figure
10-2. fo.Schematic
presentation
of the
components of
Ionospheric
major
Pulse
Sounder.
the
frequency
oscillator
The transmit
pulse
is amplified
inansynchronism
with
the transmission
for reception
ausedusing either a
and pulling a film
that shown in Figure 10-1. Sine soundersbased on the
is
matched with
bandwidth
in width
to of transmitted
width
the pulse
(overall
bandwitho
B=1/P,
P the
pulse).
the major
After
componeng
/Digital
Ionosonde
Whileand
verdetection
amplification,
thethe
video
signalof
modulates
the
amplified
incal
synchronismng
withtheofHybrid
transmissiontter
receiverng
theirm
in one orand
several
power stages
and transmitted,
using
a
slowly
in the direction
the X-axis inand
the focal plane
of
suitable wide-band antenna with a vertical radiation pattern.
an imaging optic results in an ionogram recording such as
The same antenna can be used for reception using either a
that shown in Figure 10-1. Since sounders based on the
polarization
or mode identification
[Biblup)
Reinisch,
The
are mixed
signals
down
to and
intermediate
frequencyreceived
in
an
and
IF(or
amplifier,
amplified
that
especially problematic
transistorized
for receivers. More
primary ionosonte for the

the International Geophysical Year
become commonplace.
This
therefore
less
costly
receiver
still
at many
ionospheric
observatories,
especially
theoperated
well-known
C3 and
C4
which
ionosondes,
were
devel-
permits
the use
antennas
phased
inof smaller
arrays and
for
angle-of-arrival
measurements
polarized
asprotectas
antennas
for
Transmit/Receive
or T/R during
switch, and
which
receiver
input
fromproblematic
overloading
transmission
ofthe
the
pulse,
especially
for transistorized
receivers.
More
re-
Deflecting the
for
a Y-axis
sawtooth
transmitter
with
sounders, in contrast
re-to the digital sounders
developed
in
oped
by NBS have
and which
distributed
were
as
the
respective
antennas
ollocated
generation,
synchr
reworldwide
eptioniz
of
cessing,
more
become
known
asation
analog
sounders,they
in contrast
to recently
the digital
sounders
developed
in
voltage
haveand reception
several places
relatively
easy, a much
sounders
moreanalog
are-
10-5
10-5
CHAPTER 10
manding task arose when investigators attempted to sound
the ionosphere over paths of varying distances to determine
the mode structure and the propagation conditions directly.
If the transmitted signal is to be received within the
receiver bandwidth, the systems must be started at a precise
time, and must have perfectly aligned frequency scans. This
was achieved using linear frequency scans and synchronous
motor drives, which derived their A/C voltage from crystal
oscillators IBibI, 1963]. A large step forward was the development of frequency stepping sounders such as the Granger Path Sounder IGowell and Whidden, 1968] which combined digital and analog techniques. Digital techniques
generated ionograms by stepping synthesizer/transmitter and
receiver through the desired frequency range, providing selectable frequency spacing (for example, 25, 50, or 100
kHz, linear or linear over octave bands). The frequency
synthesis itself and the data processing/recording however,
used the standard analog techniques. All digital and hybrid
pulse sounders currently available use these frequency stepping techniques.
10.1.1.5 Digital lonosonde. The rapid development of
integrated circuits, microprocessors and especially ReadOnly-Memories, and of inexpensive storage of large capacity, has led to the development of digital ionospheric
sounders. These systems have some analog components,
but use digital techniques for frequency synthesis, receiver
tuning, signal processing, recording, and displaying of the
ionograms. However, to the modern sounder, the digital
control of all sounder functions, the ability to digitally control the antenna configuration, and above all, the immense
power of digital real time processing of the data prior to
recording on magnetic tape or printing with digital printers
are of special importance.
A digital amplitude ionogram, recorded by a Digisonde
128 PS at the AFGL Goose Bay Ionospheric Observatory
is shown in Figure 10-3. This system developed at the University of Lowell [Bibl and Reinisch, 1978a,b] uses phase
coding, spectral integration, polarized receive antennas for
o/x component identification, and fixed angle beam steering
of the receive antenna array for coarse angle of arrival measurements to provide a rather complete description of the
properties and origin of the reflected echoes. Using a standard set of 128 range bins for each frequency, the sounder
integrates the sampled receiver output signals for a selectable number of integrations, improving the signal-to-noise
ratio and providing the samples for spectral analysis. Since
for each frequency-range-bin or FRB only one return is
recorded, a search algorithm determines from the set of
separate signals (o, x, several antenna directions, Doppler
lines) the signal with the largest amplitude and retains amplitude and STATUS, that is, special signal characteristics.
Using a special font [Patenaude et al., 1973], the resulting
digital amplitudes are printed out providing the analog presentation essential for the recognition of the detailed struc10-6
ture of an ionogram trace simultaneously with the digital

information. Preprocessing has largely eliminated the noise
background. The bottom part of Figure 10-3 shows a digital
amplitude ionogram, represented by all amplitudes above a
noise level determined automatically and separately for each
frequency. The noise level on each frequency can be estimated, since the unmodified signals of the lowest four height
bins are shown at the bottom of the ionogram. The displayed
range starts at 60 km and in 128 height increments with a
A = 5 km covers the range to 695 km. Each frequency
step is in 100 kHz, which covers the range from (nominally)
0 to 13 MHz in 130 frequency steps. Ionogramsof this type
can be produced in between 30 s and 2 min, depending on
the complexity of signal characterization selected. The number of integrations required to achieve an acceptable signalto-noise ratio, and the desired spectral resolution of the
Doppler measurements also affect the duration of the ionogram sweep. The ionogram is similar in structure to the
daytime ionogram in Figure 10-1, showing clearly an
E-trace (foE = 3.25 MHz), an F1-cusp (foF1 = 5.0 MHz),
and the F2 trace (foF2 = 8.2 MHz). The top part of the
figure was produced by printing only those amplitudes which
had a STATUS indicating o-polarization, vertical signals
only. The resulting suppression of the x-component and of
the (obliquely received) noise shows the effectiveness of
these techniques.
The digital "HF Radar System" developed at NOAA,
Boulder, Colorado [Grubb, 1979] is an ionospheric sounder,
built around a minicomputer. Appropriate software allows
freedom in generating the transmit signal phase coding and
sequence, and in processing procedures. However, instruction execution times of the minicomputer limit this flexibility. The sounder with its present software uses an echo
detection scheme rather than a fixed FRB grid to obtain the
information on the ionospheric returns. This scheme requires that the return has to be identified beforehand, generally using a selectable level above the noise, and an "a
hits out of b samples" criterion. This system, by the use of
a receive antenna array, then determines on-line for this
initially identified return, the echo amplitude, its polarization, Doppler shift, and reflector location [ Wright and Pitteway, 1979].
The spectral information available in digital ionograms
has been used to track moving irregularities in the equatorial
[Weber et al., 1982] and polar ionosphere [Buchau et al.,
1983]. An example of a Doppler ionogram recorded in the
polar cap is shown in Figure 10-4. The right lower panel
shows a heavily spread amplitude ionogram and superimposed two oblique backscatter traces. The right upper panel
shows the Status or Doppler ionogram: each FRB displays
the Doppler bin number instead of an amplitude. FRB's
with an amplitude below an automatically determined noise
level show neither amplitude nor status. Separating the ionogram into positive and negative Doppler ionograms permits
the identification and subsequent tracking of approaching

[KM]
:
600
O-SIGNALS
VERTICAL
400
300
200300
600
500-
10
11
[MHz]
1
Goose Bay Ionospheric Observatory 16 June 1980 1720
Figure 10-3. Digital daytime amplitude ionogram recorded by a Digisonde 128PS at the AFGL
AST. Coarse angle of arrival and polarization information is used to separate the vertical ordinary trace shown in the upper part of the figure.
300
200
CHAPTER 10
NEGATIVE
-
POSITIVE
STATUS
DOPPLER
IONOGRAMS
AMPLITUDE AND
STATUS
ONOGRAMS
AMPLITUDE
DOPPLER
......
MHZ
Figure 10-4. Amplitude/status ionogram taken by the AFGL airborne ionospheric observatory with a Digisonde 128PS at Thule, Greenland 9 December
2231 (UT). The lower right panel shows the amplitude ionogram after removal of radio noise. The Doppler ionogram shown in the upper
right panel is produced by replacing each amplitude in the ionogram below with a number representing the measured Doppler shift. The
separation into positive and negative Doppler traces (approaching and receding reflection regions) is shown in the two panels on the left.
(traces marked B and C) and receding (trace marked A)

reflecting or scattering centers. The overhead trace (very
low Doppler) is marked 0.
10.1.1.6 Digital Data Processing. The availability of
ionograms in digital form has finally provided the basis for
successful automatic processing of these complex data. Real
time monitoring [Buchau et al., 1978], survey of large data
bases [Reinisch et al., 1982a], real time analysis of ionospheric parameters [Reinisch et al., 1982b], automatic trace
identification and true height analysis [Reinisch and Xuequin, 1982, 1983] have been made possible by the availability of data in digital form. Analysis concepts for angleof-arrival determination and other parameters for the
NOAA/SEL digital sounder have been presented by Wright
and Pitteway [1982] and by Paul [1982].
An example of a data survey presentation using digital
ionosonde data from Goose Bay is shown in Figure 10-5.
The top row shows the integrated height characteristic,
obtained by collapsing each ionogram onto its height axis.
This characteristic provides the history of E and F layer
(minimum) height variations over the course of three days.
The middle panel shows the temporal changes of the F layer
returns, with the lower envelope determined by foE (daytime) or fmin (nighttime), while the upper envelope is determined generally by foF2. The bottom panel representing
10-8
the frequency extent of E and Es traces shows the typical

cos X (X = solar zenith angle) pattern of the solar E-layer,
maximizing at noon. Sporadic E events observed on all three
nights are typically observed at these high latitudes during
auroral storms [Buchau et al., 1978].
10.1.1.7 FM/CW or Chirp Sounder. The availability
of very linear sweep-frequency synthesizers resulted in the
development of FM/CW (frequency modulated continuous
wave) or Chirp Sounders, initially for oblique incidence and
in the 1970s also for monostatic vertical incidence sounding
[Barry, 19711. A linear waveform with the constant sweep
rate df/dt is transmitted. Receiving the waveform after propagation to the ionosphere and back and measuring the time
delay of each frequency component against the original
waveform permits the determination of the travel time as a
function of frequency. This is actually done by mixing the
received waveform with the original, resulting in a difference frequency which can be measured by spectrum analysis. The difference frequency as a function of frequency
(the "Chirp") is proportional to the travel time of the signal
as a function of frequency; therefore, a graph of the difference frequency as a function of time or transmitted frequency, through the known sweep rate df/dt, forms an ionogram. While initiallytransmitter and receiver were separated
by a substantial distance, to avoid overloading of the re-

Itegrate
KM
600
600
500
500-
-400
00
100
MHZ
MHZ
-13.0
Frequency Characteiistics
13.0
12.0-
11.0
11.0
10
10.0
9.0
9.0
70
8.0
60
....
4.0 -
12.011.0
10.09.09
M13.0-
I
a
"i
'H tir~
,..
ate
Frn4.0
t... fifi"
a
-10.0
I H 1.
aill II~Ia
MH
....- .....
4.0
'rEiit
ic
Ant~~~~~~~~~~~~~~~~~iilllsilii
lill:-iiil HI !IIli
A"Iglflij
,;;i;?i~i~li,)i
!-i" -tHHE13
a
9'I~~~~~
'.0~
-1~~~~'1~
10.
9.0 -j
3.-8.0
6.0-
3.0
9
ji
1.0
7.0
l,
.ta
3.3.
6.0
5.0
4.0
5..
main
linear waveform
theFM/CW
of
interruption
12
12AST
instanthe very narrow
system is transmission.
the
advantageof
120The
419
Figure
10-5.
Ionospheric
131.0
7.0fl
i ..i
6.0..
:4.
il[ilIi
:!l:liiii'?$iiiiiil$'''
~-~.........
iiiii:!sbriiii$iiiiii'iiiLi~ii?
13.0
E
1.0 ~
12.0
.
~11.0
~
~
7.0
6.0- : ~:['
13.011.0,......
10D-
2.aIai
7.01.0 :-
-12.0
9.0
characteristics
spanning
three
days
produced
of 1Hz.
Although the
121digital integration and spectral analysis
digital
ionograms
recorded
at Goose
Bay
28-30
83The
U0AR
1980.
April
integrated
Figure 10-5. Ionosphericcharacteristicsspanningthree days producedfromdigitalionogramsrecordedat Goose Bay 28-30 April 1980.The integrated
E layers.
layers.
changesof
dynamicchanges
sporadic E
the dynamic
and sporadic
solar and
showsthe
the solar
of the
characteristicshows
appearance of
heightcharacteristic
the appearance
and the
F layer
layer and
the F
of the
height of
minimum height
of the
the minimum
height
evidenceof
as evidence
well as
as well
of these
these layers
variabilityof
events.
diurnal variability
the diurnal
auroral events.
some auroral
showthe
of some
characteristicsshow
E frequency
frequencycharacteristics
and E
layers as
The FF and
The
system
ceivers with
monostaticsystem
signal, aa monostatic
directsignal,
unwanteddirect
the unwanted
with the
ceivers
and
quasi-random
and a quasi-random
switch
T/R switch
using aa T/R
developed, using
was developed,
was
interruption of the linear waveform transmission. The main
advantage of the FM/CW system is the very narrow instantaneous bandwidth of the transmitted signal, allowing a
similarly narrow receiver bandwidth (nominal 100 Hz at
sweeprates
reduced
is further
furtherreduced
bandwidthis
This bandwidth
kHz/s). This
20 kHz/s).
of 20
sweeprates of
order
the order
of the
ofR
of 1 Hz.
Although the digital integration and spectral analysis
used in the modern digital pulsed ionosondes decreases the
effective bandwidth of a pulse receiver significantly (by a
bandwidth
effective bandwidth
an effective
to an
analysis to
spectrum analysis
by spectrum
by
10-9
CHAPTER 10
-I
inreceiver
Maine.
I--
-J
site
have FM/CW
been obtained
with transmit
as low
as I1Wfor
(CW).
The
system
definitely
is power
a good
solution
the
MacLean
[1969].
Since groundbased
ionosondes obtain ionospheric echoes
the FM/CW system allows a substantial reduction of the
measuring geophysical
Re
rcie
site in Maine.
instruments (ISIS I, 1969 and ISIS

fzS
fB
fT
fN fxS 3fB
4fB
foF2
-...........
....
fxF2
I
200400-
1200i
it1200-
'
_u,,nt_ Si
T.e
PS.B,
'.
Ar
140005
Figure 10-7.
15 2025
45
55
65 70 75
85
95
FREQUENCY (Mc/s)
DAY 319 (15 NOVEMBER 1962) 0731/10 GMT (138 0 E, 30 0 S)
SATELLITE HEIGHT I011 Km
10.5 11.5
An Alouette I topside ionogram illustrating Z-, 0- and X-wave traces, cutoffs, resonance spikes, and earth echoes.
ionization of lower layers exceeds the maximum density of

the F2-layer). A typical topside ionogram is shown in Figure
10-7 from the URSI Handbook of Ionogram Interpretation
and Reduction [UAG-23, 1972]. A unique phenomenon
observed in topside ionograms are the ionospheric resonance
spikes due to the excitation of the ambient plasma by the
transmissions.The most frequently observed resonance spikes
occur at the (local) plasma frequency fN, at the local gyrofrequency fH (labeled fB in Figure 10-7), at the hybrid
frequency
(10.20)
therefore the final appearance of the h'f-trace. This trace

is sometimes further complicated by ionospheric irregularities and oblique returns. All these factors combined ensure an incredible variety of ionograms. To capture their geophysically significant parameters, a large number of rules
and definitions have evolved over the decades, which after
acceptance by the International Radio Science Union (URSI)
have been published as the URSI Handbook of Ionogram
Interpretation and Reduction, [UAG-23, 19721governing
the analysis of ionograms at all ionospheric stations.
This set of rules, resulting from the still continuing or terminated operation
of more than 300 ionosonde stations
10.1.1.9 Ionogram Interpretation. The behavior of the

ionosphere is often very dynamic. This fact and the large
range of electron densities, over which the ionospheric layers change from day to day, from day to night, with season
and with solar cycle result in a large variety of ionograms.
distributed over the whole globe, has produced a rather uniform analyzed data base which is archived at the World Data
Centers for Solar Terrestrial Research located at Boulder,
Colorado (WDC A), Izmiran, USSR (WDC B), Tokyo,
Japan (WDC C1) and Slough, UK (WDC C2). With some
exceptions, the individual world data centers store data originating in their respective regions. WDC A stores the data
from the western hemisphere and also data from France and
India.
To provide special instructions for the analysis of the
extremely complex ionograms from high latitude stations,
There are also extreme differences in ionospheric variations

and structures from the equators to the poles and in the
a High-Latitude Supplement to the URSI Handbook on Ionogram Interpretation and Reduction has been published
regular or sporadic appearance and disappearance of the

lower layers. Dynamic effects that shape the profile along
the ray path and specifically in the vicinity of the reflection
region also affect the group delay at each frequency and
[UAG-50, 19751.
For special research efforts, it is often essential to go
back to the source data, the ionospheric films of a specific
station(s). For the western hemisphere, these films are stored
and at certain harmonics of these frequencies [Hagg et al.,

1969].
Many of the references given here and a large amount
of further material can be found in the special issue on
topside sounding of the Proceedings of the IEEE 11969].
10-11
CHAPTER 10
at the World Data Center A for Solar Terrestrial Physics,
NOAA/NGSDC, Boulder, Colorado. A Catalogue of Ionosphere Soundings Data [UAG-85, 1982] provides access
to this data base, which spans the period from 1930 through
fmin
toE
foFI foF2fxF2
600
400
today. The longest and still continuing operation of an ionosonde station started at Slough, UK in January 1930.
Continuous operation starting before 1940 is still ongoing
at Canberra, Australia (1937); Heiss Island, USSR (1938);
Huancayo, Peru (1937); Leningrad, USSR (1939); Tomsk,
USSR (1937); and Tromso, Norway (1932).
To provide an overview of some of the more important
ionospheric parameters that can be derived from an ionogram and introduce their geophysical meaning, two ionograms are provided in the form of a sketch (Figure 10-8),
and the parameters are identified. Both ionograms depict
the same ionospheric conditions (taken from Figure 10-1)
with the exception of an Es layer that can suddenly appear,
possibly as the result of a windshear at E layer heights. This
Es layer can obscure parts of the trace from reflections at
higher regions of the ionosphere. A list of parameters and
their identification and interpretation is provided here as a
general reference and not as a guide for ionogram analysis.
300
For detailed instructions in the evaluation of ionograms please

refer to the URSI Handbooks UAG-23 and UAG-50.
Figure 10-8. Line sketchof daytimeionogramshows definitionof importantionogramparameters.
Parameter
h' F2
200
00
o0
h'F
fbEs
fxEs
h'E
transmitted
-1-17l
- -
14
16 IBMH
600
soo400
selected
300ves
200
h'Es
6 7 8 9 10
15
20 MHz
Meaning/Comments
a) Critical and characteristic frequencies

foF2
F2 layer ordinary wave critical frequency. A measure of the maximum density Nemax of this
layer [see Equation (10.5)].
fxF2
F2 layer extraordinary wave critical frequency. Can be used to infer foF2 using Equation (10.12)
if foF2 is obscured by interference.
foF1
F1 layer ordinary wave critical frequency. This layer is often smoothly merging with the F2 layer
resulting in the absence of a distinct cusp and in difficulties of determining the exact frequency
(L condition).
foE
solar produced E layer ordinary wave critical frequency.

Comment: Extraordinary wave returns exist for all layers. However, absorption of the
extraordinary component is stronger than that of the o-component and the x-trace of the E layer
is rarely, that of the F1 layers not always observed.
fbEs
Es layer blanketing frequency. Returns from higher layers are obscured by the Es layer up to this
frequency. This frequency corresponds closely to the maximum plasma density in the (thin) Eslayer [Reddy and Rao, 1968].
fxEs
Highest frequency at which a continuous Es trace is observed.
foEs
foEs can be inferred, applying Equation (10.12). If fbEs < foEs, the layer is semitransparent. Es
and higher layers are both observable. The determination of foEs and fxEs for all cases is subject
to a complex set of rules beyond the scope of this outline (see URSI Handbook on Ionogram
Interpretation). Modern Sounders, using polarized receive antennas, permit unambiguous foEs
determination.
10-12

Parameter
fmin
b) Virtual heights
h'F
Meaning/Comments
Minimum frequency at which returns are observed on the ionogram. Since radio wave energy is
absorbed in the D region according to an inverse square law (Absorption - l/f2 ), the variation of
fmin is often used as a coarse indicator of the variation of D region ionization. fmin is not an
absolute value (as for example foF2), but depends directly on the transmitted power and the
antenna gain. Comparison between stations, therefore, can be only qualitative.
The minimum virtual height of the ordinary wave F trace taken as a whole. Due to the effects of
underlying ionization and profile shape on the travel time of the pulse, these minimum virtual
heights are only useful as coarse and "relative" height classifiers (high, average, or low layer,
compared to a reference day). True height analysis must be made to give more meaningful height
parameters, such as the height of the layer maximum (hmaxF2).
h'F2
The minimum virtual height of the ordinary wave F2 layer trace during the daytime presence of
the F1 layer. When an Fl layer is absent, the minimum virtual height of the F2 layer is h'F,
defined above.
h'E
The minimum virtual height of the normal E layer, taken as a whole.
h'Es
The minimum virtual height of the trace, used to determine foEs.
hpF2
The virtual height of the ordinary wave mode F trace at the frequency 0.834 x foF2. For a
single parabolic layer with no underlying ionization this is equal to the height of the maximum of
the layer, hmax.In practice hpF2 is usually higher than the true height of the layer maximum.
Useful as a rough estimator of hmaxbut strongly affected by a low foF2/foF1 ratio (< 1.3).
MUF(3000)F
A set of "transmission curves" [Davies, 1966 and 1969] developed for a selected propagation link
distance (the URSI standard is 3000 km) permits the determination of the Maximum Usable
Frequency, which the overhead ionosphere will permit to propagate over the selected distance.
The MUF is determined from the estimated transmission curve tangential to the F-trace. For this
ionogram MUF(3000)F would be 17.0 MHz.
10.1.1.10 Ionosonde Network. Even though the routinely operating ionosondes forming the worldwide network
are independent, generally operated as subchains or as individual stations by national or private organizations, their
operation is coordinated by the "Ionospheric Network Advisory Group (INAG)", working under the auspices of Commission G (On the Ionosphere), a Working Group of the
International Union of Radio Science (URSI). INAG publishes the "Ionospheric Station Information Bulletin" at varying intervals. The Bulletin provides a means of exchanging
experiences gained at the various ionospheric stations, discusses in detail difficult ionograms for the benefit of all
participants, and disperses information on new systems, new
techniques, special events (for example, eclipses), relevant
meetings, ahd general network news. URSI's International
Digital Ionosonde Group (IDIG), which provides a forum
for the discussion of standardization proposals, for the exchange of software, and for the general exchange of experiences with these rather new and still maturing systems
has been incorporated into INAG as of September 1984.

The INAG bulletin can be obtained from the World Data
Center A, Boulder, Colorado, 80303.
With the advent of modern digital ionosondes and onsite automated processing, a carefully planned network of
remotely controllable ionosondes can provide ionospheric
data and electron density profiles to a control location for
real time monitoring of ionospheric and geophysical conditions. Automatic oblique propagation measurements between stations of the link can increase manyfold the number
of ionospheric points that can be monitored. Considerations
for the deployment of a modern ionosonde network have
recently been presented by Wright and Paul [ 1981]. Operational and technical information on the individual stations
of the world wide network of ionosondes, as well as their
respective affiliations and addresses, are available in the
Directory of Solar Terrestrial Physics Monitoring Stations
[Shea et al., 1984]. Figure 10-9, taken from the report in
preparation, shows the locations of all ionosondes reported
10-13
CHAPTER 10
180
150W 120W
90W
60W
30W
30E
60E
90E
120E
150E
180
690
-I ~~ <ad<>t -- a
_-
p30t
SASH
e- -X--- e
[;S4~
x60
90
Figure 10-9. Map of vertical incidence ionospheric sounder stations 1984.
as operational or operating in 1984. World Data Center A

Report, UAG-85, lists all past and present ionospheric observatories.
10.1.2 Incoherent Scatter

J.J. Thomson [1906] showed that single electrons can
scatter electromagnetic waves, and that the energy scattered
by an electron into unit solid angle per unit incident flux is
given by (rc sinU)2 where re is the classical electron radius
(= e2 /EomcC2 = 2.82 x 10-15m) and U is the polarization
angle, that is, the angle between the direction of the incident
electric field and the direction of the observer. Thus the
radar backscatter (U = ii/2) cross-section of a single electron will be Oe = 4iir2e . Gordon [1958] first proposed that
by the use of a powerful radar operating at a frequency
f > foF2 where foF2 is the plasma frequency at the peak
of the F2 layer, the backscattered power from the electrons
in the upper atmosphere should be detectable. The measurement of scattered power and its characteristics as a function of altitude was expected to provide a measurement of
the various geophysical parameters both in the bottomside
and the topside ionosphere. Gordon assumed that the electrons were in random thermal motion of the same type as
the motion executed by neutral particles so that the radar
would detect scattering from individual electrons that are
random in phase or incoherent. This is known as incoherent
scatter or Thomson scatter (for a comprehensive review,
see Evans [1969]). Gordon calculated the backscatteredpower
10-14
per unit volume to be NOc, where N is the electron number

density. He also predicted that the spectrum of the scattered
signal will be Doppler broadened by the random electron
thermal motion. The spectrum of the scattered signal was
expected to be Gaussian with center to half-power width of
0.71 Afe where Afe is the Doppler shift of an electron approaching the radar at mean thermal speed so that
1/2
where X is the radar wavelength (m), k is Boltzmann's

constant (= 1.38 x 1023 J/K), Te is the electron temperature, and me is the mass of an electron (= 9. 1x 10 31
kg). At a wavelength A = 1 m, and Te = 1600 K, 0.71 Afe
= 200 kHz. Soon after Gordon [1958] proposed the feasibility of the incoherent scatter radar experiment to study
the upper atmosphere, Bowles [1958] was able to detect
radar echoes from the ionosphere. The echoes resembled
the predicted ionospheric scatter signal except that the bandwidth of the signal was considerably less than the predicted
value. The decrease of the bandwidth of the scatter signals
contributing a larger signal power per unit bandwidth obviously made it easier to detect the signal. Bowles [1958]
correctly surmised that the presence of ions causes a reduction of the bandwidth of the scattered signal. Later theoretical work [Fejer, 1960; Dougherty and Farley, 1960;
Salpeter, 1960; Hagfors, 1961] showed that the spectral
form of the scattered signal is dictated by the radar wave-

length in relation to the Debye length in the upper atmosphere. The Debye length (D) for electrons is defined as
D = (,
2 m
kTe/4iiN2e)l/
COMPLETESPECTRAFOR VARIOUS
VALUES
OF a
(10.22)
is the permittivity of free space (= 8.85 x 10-12

where Eo,,
F/m), e is the charge on an electron (= 1.6 x 10 19 C), k
is the Boltzmann's constant, No is the electron density (m 3)
and Tc is the electron temperature (K). The Debye length
for the electrons in the ionosphere is typically of the order
of I cm or less below 1000 km and it is not possible to
sustain organized motion at scales smaller than these values.
It was shown that, in general, the spectrum of the scattered signal consists of two parts, one due to the ions and
the other to electrons. If the radar wavelength is much smaller
than the Debye length, the scattered energy is entirely due
to the electronic component and the initial predictions of
Gordon [1958] for the scattered power (Noe) and its spectrum (Afe) are valid. On the other hand, for radar wavelengths much larger than the Debye length, which represents
the experimental situation, the electronic component de-2
creases and appear as a single plasma line at a Doppler shift
approximately equal to the plasma frequency of the medium.
Under this condition,
the
largest
Under
this
thecondition,
largestpart
part of
of the
the scattered
scatteredenergy
energy
resides in the ionic componentand the spectral width is

controlledby the Dopplershift Afj for an ion approaching
the radar at the mean thermal speed of the ions, given by
00
a10.003
10
'
-e
: 0
:
L 10-
0
DOPPLERSHIFT(
Figure 10-10. The variation of the overall spectrum for different values
of the ratio
. The ion has been assumed 0 +.
These curves assume that collisions are negligible and that
Te = Ti [Hagfors, 1961].
1/2
(10.23)
where mi is the mass of the dominant positive ion and Tj is
the ion temperature. Considering Te Ti, and the dominant
ion to be O+, Afj 2 x 10- 2 Afe. The echo energy is,
therefore, mainly concentrated in a relatively narrow spectral window rendering the radar investigation feasible with
apparatus of much lower sensitivity than initially envisaged.
Figure
how
10-10
the
spectral
Figure 10-10 shows
how
theshows
spectral shape
shape depends
depends on
on aa
parameter a - 47iiD/Afor the case Te = Ti. For a > 10,
the scattered energy is entirely due to the electronic component, whereas for very small values of a, the electronic
component decreases and the energy appears mainly in the
ionic component with a much smaller bandwidth. The electronic component now appears as a single line, known as
the plasma line, at a Doppler shift approximately equal to
the plasma frequency of the medium.
In the ionosphere, the electrons and the ions are at different temperatures and the spectrum of the ionic component
changes for different values of the ratio Te/Ti at a given
value of a. This is shown in Figure 10-11 for the case
a = 0.1 for O + ions which illustrates the double-humped
formof the spectrum.By measuringthe heightof the hump

at the wing relative to the center of the spectrum, it is
possible to estimate Te/Tj and the total scattering cross-
sectiondue to the ionic componentis simplyobtainedfrom

the area under the curves. For smallvalues of Te/Ti,which
is encountered in the ionosphere, the total scattering crosssection (o) may approximately be given by
(10.24)
The incoherent scatter radar technique opened up the

possibility of in situ sampling of a wide range of upper
atmospheric parameters by the use of a powerful ground-
04
12
16
20
24
DOPPLERSHIFT (Af,)
Figure 10-11. Spectra of the ionic component for the case of
(=4iiD/A) = 0.1 [Evans. 19691. (Reprinted with per
mission from IEEE c 1969.)
10-15
CHAPTER 10
based radar system [Evans, 1969]. The most obvious measurement is the electron density (N) versus altitude (h) profile
made by recording the variation of echo power Ps as a
7000
functionof delayby usinga verticallydirectedpulsedradar.
4000
The echo power is given by

3000
(10.25)
1550 EST
2500
2000
where C is a constant. The constant C can be determined

either by a careful determination of the radar parameters or
by an absolute determination of N at an altitude by an
ionosonde or other techniques. However, as mentioned earlier (Equation 10.24), the scattering cross-section o(h) depends both on a and Te/T which are both functions of
altitude. From a measurement of the scattered energy spectrum, these corrections can be introduced and electron density profiles are determined. It has also been possible to
obviate this difficulty entirely by the use of Faraday rotation
technique. Figure 10-12 illustrates the electron density profile extending to almost one earth radius obtained at Jicamarca by this technique. In addition to the rather straightforward measurement of electron density profiles, electron
and ion temperatures, ion composition, and photoelectron
flux, the ionospheric electric field and a variety of other upper atmospheric parameters have been successfully measured at various locations extending from the magnetic equator to the auroral zone [Radio Science, special issue, 1974].
500
400
300
200
150
10
ELECTRON
DENSITY
N/m3)
Figure 10-12. An electron density profile obtained at Jicamarca that extends to almost one earth radius [Bowles, 1963].
Table 10-1. Incoherent scatter facilities.
Location
Frequency (MHz)
Jicamarca,
Peru
Arecibo,
Puerto Rico
St. Santin,
France
Millstone Hill,
USA
Sondrestrom,
Greenland
EISCAT*
Transmitter:
Tromso, Norway
Receiver:
Tromso, Norway
Kiruna, Sweden
Sodankyla, Finland
*European Incoherent SCATter facility
10-16
50
430
935
440
1300
1300
224
(monostatic)
933.5
(tristatic)
Antenna
Dip latitude (N)
6
Pulsed
2
Pulsed
0.15
Continuous
3
4
Pulsed
5
Pulsed
290 m x 290 m array
32 m parabola
71
30 m x 40 m
parabola cylinder
67
Power (MW)
300 m spherical
reflector
20 m x 100 m
reflector
68 m
25 m parabola
2
32 m parabola
32 m parabola
30
47
57
FLOATING
GUARD
GUARD
DRVEN
(DIA = 0.24m)
COLLECTCO
(DIA = 0.165cm)
(DIA = 0.058cm)
S/
probe is replaced with the entire exposure conducting surface of the rocket or satellite. If the exposed conducting
surface of the rocket or satellite is much greater than the
~2Tc
tC
2.3-
SPACECRFAFT
SURFACE
Figure 10-13. Cylindrical LangmuirProbe.
Table 10-1 adopted from Hargreaves [1979] gives a list of

the incoherent scatter facilities now in operation and the
characteristics of the radar system.
area of the probe, the potential of the surfaces will remain

fixed as the potential on the probe is swept. As a minimum,
the area of the conducting surfaces should be 100 times
greater than the area of the probe, and ideally the area should
be 10 000 times greater then the area of the probe. By setting
the potentialof the probe very positive( + 1.5v to + 20v),
all electrons within a few Debye lengths of the probe will
be drawn in and measured; this allows a direct measurement
of plasma density oscillations which are directly related to
plasma turbulence.
The two most common shapes for Langmuir probes are
the cylindrical probe (Figure 10-13) and the spherical probe
(Figure 10-14). Any shape probe is possible, but these shapes
are the easiest to analyze mathematically.
10.1.3 Langmuir Probes

One of the simplest devices used on rockets or satellites
to measure the ionospheric density in situ is the Langmuir
probe, named for Irving Langmuir, who pioneeredthe method
at General Electric in the 1920s. The density is determined
from a measurement of electric current passing between two
conducting surfaces in contact with the environment. A
varying electrostatic potential placed between the two surfaces causes the current to vary. The magnitude of the current indicates the density of the ionospheric plasma, and the
change in current with respect to changes in the potential
between the surfaces indicates the ion and electron temperatures. The double-floating-probe, which is the closest
version to an idealized Langmuir probe, usually consists of
a conductor at each end of a dipole antenna flown on a
rocket or satellite for other purposes. The major disadvantage of a double-floating-probe is that ion thermal velocity
is much lower than the electron thermal velocity and the
rocket or satellite speeds. Therefore, the usual Langmuir
probe is a single probe to measure only electrons; the other
10.1.4 Faraday Cups for Rockets

and Satellites
The most commonly used device for measuring the thermal ions is the Faraday cup (see Figure 10-15). It is usually
an aperture that is a section of a flat, infinite surface in
contact with a plasma. A screen across the aperture shields
electrostatic potentials inside the sensor from the outside
environment. The arrangement of grids or screens inside
the sensor is determined by the function of the sensor. Most
Faraday cups use a suppressor screen in front of the collector. This screen has a large negative potential (- 10V to
- 100V) to repel electrons from the environment away from
the collector and to drive secondary and photoelectrons from
the collector back to the collector.
.89
DMSP SSIE
07
DMSP
SSIE
ELECTRON
SENSOR
ION SENSOR
GOLD PLATED
ALUMINUM
COLLECTOR
1.75" DIAMETER
COLLECTOR
art,//~~\
/
//
ok
\ GUARDRING
//
<~~~~\
YMOUNTING BOOM
\
\
XSWEPT
B~APERTURE
GRID
SUPPRESSOR
\\ ELECTRON
///
2.25"
AMPLIF
ER
ALL GRDS GOLDPLATED TUNGSTEN,

0.92 TRANSPARENCY.
// /
\\ ELECTRON
TO
I _LL
F I
AMP
GOLD PLATED
TUNGSTEN
TRANSPARENCY0.80
ON
E R
COLL:CTOR
COLD PLATED ALUMINUM.
AAMPL
A
CONDUCTING SURFACES GOLD
PLATED.
OND.
26
41'
15
DIAMETER
Figure 10-14. The Spherical Langmuir Probe on the DMSP Satellite.
Figure 10-15.
The Faraday Cup used on the DMSP Satellite.
10-17
CHAPTER 10
10.1.5 Optical Measurements

Ground, airborne, satellite, and rocket based optical
measurements are commonly used to determine ionospheric
structure and dynamics. While a number of different instruments are employed, all analysis techniques must relate
spectral emission features to ionospheric structure and dynamic processes. This is done through a knowledge of the
atmosphere/ionosphere chemistry that leads to the measured
photon emission. Ionospheric domains are conveniently divided into regions that are produced or influenced by energetic particle precipitation (auroral regions) and those controlled mainly by solar ionizing radiation (equatorial and
midlatitude). Optical measurements have played important
roles in both regions in defining the spatial and temporal
characteristics of ionospheric plasma. Commonly used observing techniques will be discussed followed by a section
describing important results.
10.1.5.1 Observing Techniques. Optical instruments can
be classified according to spectral resolution as low, medium, and high.
Low Resolution Systems: The all sky camera has historically been used to measure auroral structure. This is
perhaps the lowest resolution system, measuring all wavelengths over the sensitivity range for the type of film used
(typically Kodak TRI X). The system uses a 160 field of
view lens to measure auroras over a circle of 1000 km
diameter in the lower ionosphere (110 km altitude). All sky
from 2-18 A. They can be scanned across the sky or operated

in the zenith. When properly calibrated, spectrometers provide the absolute intensity of auroral and airglow features
as well as some measure of spectral character.
High Resolution Systems: Fabry-Perot interferometers
use multiple path interference to achieve high spectral resolution. These instruments are primarily used to measure
spectral line broadening and Doppler shift. From these parameters, atmospheric temperatures and drift velocities can
be derived. Primary spectral features of interest are 6300 A
[OI], 5577 A [0I] for neutral winds, and 7320 A [O11]for
plasma drift.
Optical
from
Structure
10.1.5.2 Ionospheric
Measurements. Ionospheric structure at mid and equatorial latitudes is controlled by solar ionizing radiation, electric fields, and neutral atmosphere dynamics. Airglow observations of equatorial plasma depletions are one example
of optical measurements used to define ionospheric processes. A brief review of equatorial airglow chemical production mechanisms is presented to illustrate the techniques
used to infer ionospheric plasma density variations from
remote optical measurements.
Two primary airglow spectral emission features are of
interest for nighttime, F region phenomena 6300 A [OI] and
7774 AOI. The 6300A atomic oxygen emission results from
the following sequence of reaction:
K,
0 +
camerastypicallymeasureonly bright auroralfeatures, primarily at E region altitudes.

Photometers are low resolution systems. They rely on
narrow band interference filters to isolate spectral lines and
bands of interest. Meridian scanning photometers use a narrow (0.5 to 2.5 ) field of view and -2 A filters to measureD)
absolute intensity of auroral and airglow along a vertical
circle, commonly aligned along a magnetic meridian. Tilting filters use the change in transmitted wavelength versus
tilt angle to perform a limited wavelength scan. This allows
separation of non-spectral continuum from the line or band
emission.
More recently, all-sky imaging photometers have been
developed to perform all-sky (155 ) monochromatic measurements at high sensitivity (20 Rayleighs). These employ
slightly wider (-20 Ao) interference filters because of the
lack of convergence of the extreme optical rays at large
zenith angles. Image intensifiers are employed to achieve
the high sensitivity. Data are recorded either on a photographic image or by using a TV system to produce a video
signal. Typical system parameters are shown in Table
10-2.
Medium Resolution Systems: Ebert-Fastie type scanning
spectometers are used as medium resolution optical systems.
These are effectively used over the (visible) wavelength
range of 3800 to 7900 A with variable spectral resolution
10-18
02
(10.26)
K2
O2 + e -- O + O('D)
O(1 D) -- O (3P) + hv(6300)
OP)
(10.27)
(10.28)
(1029)
Since K2
K1, and in regions where O+ is the dominant
ion (0+ = N) the 6300A volume emission rate is given
by
dl(6300) = 0.75 K2E [N,][021/

IA
dh
IQ2N
(10.30)
where
K2E = 1.4 x 10-11 cm3 s

KQ = 7.0 X 10 -" cm3 s - l
-I
A
(See Weber et al., [1980] and Noxon and Johanson [1970]

for a more complete discussion).
The 7774 A O results from radiative recombination of
O+:

Table 10-2. Summary of system specifications.
Field of View
1550
Pass Band
25 A at f 1.4; 5 A at f 8
Resolution
1/2 zenith, 2 horizon
Spectral Response
S-20, exceeding 100 u.A/lumen
Picture Storage
No detectable degradation for up to 3 s
Tube Gain
Photon noise granularity visible above tube noise
Threshold Sensitivity
20 R at 2 s exposure
I kR at 30 frames/s
Dynamic Range
20 R to 10 kR covered by 3 preset HV settings
Flatness of Field
30 percent loss at edge of field
Repetition Rate
Typically 20 s for complete filter cycle
Temporary Storage
Video disc, three video tracks + one sync track
Permanent Storage
Video tape deck, time-lapse type (9 h recording time

on a single reel): 16 mm color movie camera
Process Controller
In-field programming capability
Display Systems
Four black and white monitors, 9 in. diagnal; Color

monitor, RGB and A-B input, 12 in. diagonal
Real-Time Display
Simultaneous fully registered display of three filter

channels. Capability of displaying difference of any
two pictures. Display of two or three filters as
pseudo-color on RGB monitor
Character Generators
Date/time display on each frame for frame

identification
Digital Encoding
Digital encoding of time and housekeeping data for

computer-controlled data handling
K3
O+ + e -m O(CP)
5
0( P)-->
O(5S) + hv(7774)
(10.31)
(10.32)
and the volume emission rate is given by
700
600 -
ANOMALY- 0131 UT 15 DEC 1979

OUTSIDEDEPLETION
1(7774):30R
\0(7774)
Ne
E 5001(6300} 4R
dl(7774) = K3 [O+l[NI dh.
(10.33)
400
300
To illustrate the altitude dependence, 6300 A and 7774 A

airglow volume emission rates were calculated for an electron density profile representative of the Appleton anomaly
region, and are shown in Figure 10-16. The bottomside

profile was obtained from true height analysis of a digital
ionogram. This was matched to a modified Chapman function [Tinsley et al. 1973] to represent the topside profile
from hmax F2 to 690 km. Because of the exponentially
20
I0-
I
I0
PHOTONS/cm
- ec
' 0'
0'
'
EL
Figure 10-16. Electron density profile derived from true height analysis
of bottomside ionogram matched to a modified Chapman
function for the topside. Also shown are calculated 6300
and 7774 A volume emission rates and column intensities
in Rayleighs.
10-19
CHAPTER 10
decreasing O2 concentration, the 6300 A volume emission
rate is confined to the bottomside and reaches a maximum
value at 300 km, below hmax F2 (360 km). The 7774 A
volume emission rate is proportional to [Ne]2 and attains its
maximum value at hmax F2. Because of the broad altitude
exatent of the equatorial electron density profile the 7774
A volume emission rate displays a similar broad extent and
falls to 50% of the maximum value at 300 km and 450 km.
Thus the 7774 A airglow measurements provide information
over a broad altitude range, with approximately one half of
the emission produced above the F layer peak.
All sky imaging photometer measurements conducted
within a few degrees of the magnetic equator and near the
Appleton Anomaly region (--18 ML) have established the
two-dimensional horizontal extent of equatorial plasma depletions. These are also the regions of post-sunset equatorial
spread F, VHF radar backscatter plumes, and amplitude and
phase scintillation on transionospheric radio propagation.
Figure 10-17 shows an example of all sky images at 6300 A
and 7774 A near the equatorial edge of the Appleton Anomaly. The bright region over the southern two-thirds of the
image is airglow from the high-density anomaly region. The
North-South aligned dark band is a region of decreased
airglow emission. Comparison with simultaneous in situ
measurements from the Atmosphere Explorer satellite (AEE) shows this airglow depletion to the collocated with a
region of significantly decreased ion density. Having established the relation between airglow emission processes and
F layer density, the all sky images provide a two-dimensional map of these depleted regions. In addition, the dynamics of these regions can be monitored over extended
15 DECEMBER
1979
6300ooA
7774A
MLm
GNORTH
periods in the absence of "snapshot" satellite measurements.

Optical measurements coordinated with VHF radar, ionosonde, in situ density, and satellite radio beacon scintillation
observations have helped to provide a detailed description
of the development, structure, drift, and decay of these
important equatorial ionospheric features.
Optical measurements have improved our understanding
of auroral zone and polar cap ionospheric structure and
magnetosphere-ionosphere coupling processes. In this region dominated by the effect of precipitating electrons and
ions over a wide energy range (few eV to 100's of keV),
optical measurements of impact excitation and chemical
recombination aid in understanding a wide variety of processes. In this section, several examples of all-sky monochromatic images are shown, primarily to demonstrate the
use of optical measurements, especially when coordinated
with other ionospheric diagnostics. Figure 10-18 shows a
montage of auroral images at 10-min intervals at 6300 A
[OI] and 4278 A N2+ . These images were recorded on an
aircraft which flew North-South legs along the Chatanika,
Alaska Incoherent Scatter Radar magnetic meridian. The
images provide a map of the instantaneous particle precipitation patterns separately for the E (4278 A) and F (6300
A) layers. Measurements with the radar mapped electron
density structure and satellite UHF radio beacon scintillation
measurements mapped regions of ionospheric irregularities
(from tens of meters to a few kilometers). In this experiment,
optical measurements provide a continuous map of particle
precipitation regions over a large area (1200 km diameter
at F-region altitudes) for interpretation of magnetosphereionosphere coupling and ionospheric dynamics.
Measurements in the polar cap have recently clarified
local particle precipitation effects from plasma transport (Efield) effects. Local precipitation of low energy (100's of
eV) electrons, during IMF Bz north conditions, leads to the

production of sun-aligned F region auroras. Some of these
MAG NORTH
auroras are characterized
ACPOSITIONX
Eplitude
GLON 7W
0o~"aj ,
L,\
1o
. E BIMS
E
(..
434kin)
particles,
'
but are convected from a source region at or equatorward of the dayside cusp. These patches are also subject
to structuring processes that lead to scintillation.
......
98F~iljV
V
l
//
l?
rO',
that these patches are not locally produced by precipitating
*IA
/-
SOME ASPECTSOF
1031/10.2
102L~__,_
0035
129
03
0033
-138 -144
~k
oo
0034
36
by F region plasma density en-
hancement and structuring within these auroras leds to amand phase scintillation.
During Bz south conditions, large patches of F region
plasma are observed to drift across the polar cap in the antisunward direction. Coordinated satellite measurements show
70-0 14
0005
LONG WAVE PROPAGATION
loi_--4
0031
0029
0027
002 5
147 -144 -139 -31
-122 -11 -100
27
36 17, 20o245-0624 23
2831
-7041 -8 73
2336
2306
2236
UT
MLAT
134
GLAI
3344GLON
(W
2208
L-(WTMV
It is convenient to refer to radio waves having frequencies below 3000 kHz as "long waves". These include Ex-
tremely Low Frequencies (ELF), Very Low Frequencies

Figure 10-17. All sky images at 6300 and 7774 A recorded near the
equatorial edge of the Appleton Anomaly.
10-20
(VLF), Low Frequencies (LF), and Medium Frequencies

(MF), as outlined in Table 10-3. ELF has had very little
NORTHW
0800
0900 UT
Figure 10-18. Auroral images taken at 10 minute intervals from 0620 to 0950 UT on 29 January 1979. The upper row under each hour shows the 6300
A images; the lower row, the 4278 A images.
10-21
CHAPTER 10
Table 10-3. Long wave frequency bands.
Designation
Abbreviation
Extremely Low Frequency

Very Low Frequency
Low Frequency
Medium Frequency
ELF
VLF
LF
MF
use, except for communications that require wave penetration beneath the surface of the ocean or earth. The VLF/LF
bands are used extensively for navigation and military communication. The standard AM broadcast systems utilize part
of the MF band (535-1606 kHz). Long radio waves are also
used in basic ionospheric research, meteorology and thunderstorm study and tracking, standard frequency and time
distribution, geological studies, and minerals exploration.
Long waves propagate by a number of different modes.
These include propagation over the surface of the earth by
diffraction modes, ELF propagation by transmission-line
type modes, propagation by ionospheric reflection (or earthionosphere waveguide modes) and propagation through the
ionosphere by so-called "whistler" modes. Each type of
mode requires a separate physical description and mathematical formulation.
10.2.1 Groundwave Propagation

The most general definition of the groundwave is the
wave that would be excited by an antenna at or near the airearth boundary if there were no wave reflections from the
upper atmosphere. At long wavelengths ionospheric reflections are important, and for continuous wave (CW) transmissions it is necessary to regard the total wave field as a
vector sum of the groundwave and skywaves. If an antenna
radiates a very short pulse, however, it may be possible for
a distant receiver to resolve the groundwave and skywaves
individually. The time interval between the onsets of the
groundwave and the first-hop skywave is given by
At = (2\/h 2 + 4a(a+h)sin 2(d/4a) - d) (l/c),
(10.34)
providing
d Us2a cos '{a/(a+h)},
(10.35)
where d is the distance between the transmitter and receiver,

h is the effective height of reflection in the ionosphere, a is
the earth's radius (-6370 km) and c is the wave velocity
(3 x 105 km/s). If the transmitted pulse is short enough,
At may be long enough (for example, At = 93 us for
d = 500 km and h = 80 km) to permit very accurate measurements of the groundwave, especially its arrival time or
10-22
Frequency Range
0.003-3
3-30
30-300
300-3000
kHz
kHz
kHz
kHz
Wavelengths
105-102 km
102-101 km
10-1 km
1-0. 1 km
phase. The latter is the basis of the long-range 100 kHz

groundwave navigation system, Loran-C.
If the transmitted signal is a continuous wave, the amplitude and phase of the composite signal received at a fixed
distance vary with time as the ionosphere changes. On the
other hand, at a given moment the signal amplitude is a
function of distance [Hollingworth, 1926], having maxima
and minima typical of an interference pattern. The groundwave component is stronger than the skywaves out to a
distance that depends on the wave frequency, among other
factors. This region of groundwave dominance is the most
stable, or primary, coverage area of MF broadcast transmitters.
As defined above, the groundwave exists at all radio
frequencies, but at wavelengths comparable to the height
of the ionosphere or greater, the usefulness of the concept
begins to fade. Also, for transmitters high above the ground,
or at high frequencies where quasi-optical propagation analysis is appropriate, the term groundwave is seldom used.
The earth often acts as a fairly good conductor for long
waves, in which case the electromagnetic boundary conditions permit electric fields perpendicular to the surface,
while tending to suppress electric fields tangential to the
surface. It follows that groundwave fields near the earth's
surface tend to have transverse magnetic (TM) polarization
rather than transverse electric (TE) polarization. In common
usage the unqualified term "groundwave" implies TM pousage the unqualified term "groundwave" implies TM polarization.
10.2.1.1 Idealized Flat-Earth Models. In a simple model
the earth's surface is regarded as a flat perfect conductor,
and the air is homogeneous with refractive index 1. The
most elementary source is a vertically-directed current
I(t) = 1,,exp (iwt), at frequency f = w/2Tr,of infinitesimal
length de, which has an electric dipole moment M(t) = I(t)
de (Note: complex antennas can be regarded as distributions
of such elementary currents). The fields of such a source
may be found readily by the method of images. When the
current element is just above the surface, the fields in air
are simply twice the homogeneous free-space fields. Because of symmetry the magnetic field is everywhere in the
azimuthal direction 0 (see Figure 10-19), while the electric
field on the surface is constrained by the boundary conditions
to be strictly in the vertical direction z. In mks units the
magnetic and electric fields at a distance, d, on the flat
perfectly-conducting surface are, respectively,

IGIcl
".
CURRENT
ELEMENT
10
1.0
0i0JC
EZ
IGI
2
E
/
-V
I
Figure 10-19. Vector field-components at a point P in a cylindrical coordinate system. The plane XOY represents the surface of
the earth.
H,= M(t')
i2dr 1A0.2
or G
0.5
argG. l=seconds
(10.36)
0.05
(10.37)
where
t' = (t -
d/c),
X is the
vacuum
of
space
is the
Eo
permittivity of free space
is the permittivity
108/f m),
m), Eo
(=3 xx 5108/f
(= 8.854 x 10-12 F/m), and u is the free space permeability (= 4ii x 10-7 H/m). The far-field components are
related by
(=3
(10.39)
Ez = -ZoHo
groundwave fields for a source on a plane earth. Values

are shown at distances of 30, 100, and 300 km, for prop3
agation over fresh water, o = 10- S/m, E/Eo= 80
[Heckscher
and Tichovolsky,1981]
wavelength
free
where ZO,is the impedance of free-space (= 120ii ohm).

In a more realistic model, the plane earth is allowed to
have finite conductivity o and permittivity E. The solution
2.0
1.0
Figure 10-20. Height variations in the amplitudes and phases of 100 kHz
(10.38)
Eo =0,
0.15
0.25
0.10
Zenneck wave, without any radiation field, requires an infinitely long source distribution [Hill and Wait, 1978].
For a finitely conducting earth, Equation (10.39) is still
true approximately, but the radial electric field component

EP has a finite value related to the loss of wave power into
12
r---
A-v-T-,
-r
1 -r
,.--
00
of this boundary-value problem was given by Sommerfeld

[1909] in terms of an infinite complex integral [see Stratton,
80s
1941]. A more complex problem, that of an elevated dipole,

was solved by Weyl [1919], who expressed the free-space
field as a sum of plane waves that reflected at the earth's
60\
surface
in
accordance
with
the
Fresnel
formulas.
Norton
[1941] and others have calculated numerical values from

the formal solutions. Height variations of the fields are shown
in Figure 10-20 for a source on a plane earth surface
[Heckscher and Tichovolsky, 1981], and curves illustrating
SPHERICAL EARTH
groundwave field amplitudes along the surface are given in

Figure 10-21 for both plane and spherical earth models.
A type of groundwave, the Zenneck surface wave, has
fields expressed exactly in simple closed forms. Although
the Zenneck wave is important historically and conceptually, it is generally difficult to excite because of its rather
slow decay with height. In fact, the excitation of a pure
1000
100
DISTANCE,
10000
km
Figure 10-21. Longwave groundwavefield amplitudesas a functionof

distance over plane and sphericalearths, for propagation
over good earth, (a 102 S/m, E/Eo = 20.
10-23
CHAPTER 10
Table 10-4. Ratio of groundwave radial and vertical electric fields for
variousearth surfacesand 100 kHz.
SURFACE TYPE
(S/m)
a/o
be sensed with a vertical monopole antenna, or a vertical

dipole. The radial component E,, may be sensed by a hor-
EP/E,
Perfect Conductor
80 0.00118 144.99
Sea Water
Good Soil
10 2
20 0.0236 144.660
Fresh Water
10- 3
80 0.0713
izontal dipole with its axis in the direction of propagation.

It follows from the principle of reciprocity that if a horizontal
dipole on the earth is driven with RF current, a TM-polarized
groundwave is radiated in directions along its axis. The
fields produced by horizontal and vertical current elements
have been discussed in mathematical detail by Wait [1954,
1957, 1961, 19711.
132.890
143.25
10.2.1.2 Idealized Spherical-Earth Models. A mathematical treatment of the groundwave on a smooth spherical
earth of homogeneous, isotropic material was undertaken
5 0.403 114.760
2 x 10 5
Thick Ice
by Watson [1919] to determine if an atmospheric reflecting
layer (ionosphere) was required to explain the large fields
produced by distant transmitters. Such early theoretical analthe ground. The ratio of the radial and vertical electric fields
yses were handicapped by the poor convergence of the inat the surface is given by
finite series contained in the solutions. That difficulty was
largely overcome by Van der Pol and Bremmer's 11938]
(10.40)
EP/E,
(1 - /p)/p,
"residue series" solution, which has become a basis for
modern numerical analysis of the groundwave.
where
Figure 10-21 shows examples of field-strength vs dis(10.41)
tance curves for waves of selected frequencies propagating
p - e/E, - ia/Weo.
over "good soil," assuming a vertical source on the surface.
The earth-curvature causes the wave amplitude to decrease
Starting just below the surface of the assumed uniform
with distance more rapidly than it would on a flat earth of
earth, Ep and Ho decrease exponentially with depth. The
the same material, but near the source the flat- and roundfields are 1/e of their values on the surface at the "skin
earth models give essentially the same fields.
depth" 8, which may be estimated from
One way to present both amplitude and phase data is in
terms of the complex factor W, which is the ratio of the
(10.42) actual field component to an idealized one calculated as if
the earth were flat and perfectly conducting. The flat earth
10
Poor Soil
= 503/of
10 0.0745
-ii/2.
when arg p
(10.43)
Sample values of Ep/E, are given in Table 10-4 for various

earth surfaces at a frequency of 100 kHz, and Figure
10-22 shows skin depths over the long wave spectrum.
The magnetic field of the groundwave Ho,may be sensed
with a loop antenna having its axis parallel to the surface
and perpendicular to the direction of propagation. E, may
o I
,FK
........
propagation distance is taken to be the same as the curvilinear distance on the sphere. Figure 10-23 shows curves
of W in the complex plane, for 100 kHz propagation over
sea water and good soil.
0.7
t100OOkHz
M 0.6
........
.''..'I
-_
go
-0.3
-0.
o;0,3
-0.1 O0S
0.1
0.2
0.3
0.4
.5
0.6
0.7
0.8
10
o1
o2
11
;1o
FREQUENCY,
H2
Figure 10-22. Long wave skin-depths for various earth surfaces.
10-24
Figure 10-23. Complex values of the ratio W of actual groundwave field

components
and ones obtained assuming a flat, perfectly-
conducting earth, for 100 kHz over sea water (o =4 S/m)

and good soil (o = 10-2 S/m). Derived from Wait and
Howe [1956].
.0

10.2.1.3 Models with Earth-Properties Gently Varying
Along the Propagation Path. Except for large bodies of
water, the earth's surface is too uneven, both in electrical
properties and in topology, to be represented well by the
idealized models discussed above. However, if the earth's
electrical properties and curvature do not vary much in a
wavelength, the groundwave amplitude and phase can be
approximated by the solution of an integral equation for W
[Hufford, 1952]. The two-dimensional integral equationthe most general form-is valid provided, first, impedance
boundary conditions can be applied; and second, terrain
irregularities are not too severe.
A much simpler, one-dimensional, integral equation is
more commonly used. It is derived by a stationary-phase
integration that reduces the dimensionality of the general
version; being an approximation, it is not valid for all terrain
types, particularly at low frequencies where wave lengths
and Fresnel zones have sizes comparable with terrain features. Field [1982a] compared solutions of the one- and
two-dimensional equations to quantify the errors incurred
by using the one-dimensional equation for terrain features
narrower than a Fresnel zone. For a frequency of 100 kHz
and on-path features narrower than about 10 km, the twodimensional equation is needed to properly account for the
signal's dependence on obstacle width, recovery at long
distances, and transverse diffraction patterns. The one- and
two-dimensional solutions approach one another far beyond
wide terrain features.
Considerable error can be incurred at low frequencies
by applying the one-dimensional equation, even for large
terrain features. For example, for a path-length of 500 km,
that equation overstates by a factor of about four the effect
of an obstacle 6 km in diameter. It cannot give accurate
results unless the diameter approaches a Fresnel zone width,
which for this example is several tens of kilometers. However, because numerical solution of the two-dimensional
integral equation is costly, its use has been limited to highly
idealized irregularities. An alternate approach given by King
and Wait [1976] obtains an equivalent one-dimensionalmodel
by averaging the terrain over the Fresnel zone.
If, instead of being homogeneous, the earth were comprised of layers of different conductivity and dielectric constant, it could still be characterized by a (frequency dependent) surface impedance, so that the integral equation for
W can be formulated. Some progress has been made in this
and other ways of estimating earth-constants for groundwave phase prediction at 100 kHz, but uncertainties in these
constants remain a major source of prediction error.
Special models of non-uniform terrain exist for which
solution of the integral equation is unnecessary. Groundwave propagation from land to sea, and vice versa, has been
modeled with an earth having a sharp discontinuity of conductivity along a horizontal straight line. A useful approximate solution was given by Millington [19491 based on
reciprocity arguments, and mathematical solutions have been
given by Wait [1956] and Wait and Walters [1963].
Groundwave perturbations caused by hills have been

studied with models having semielliptical bosses on otherwise smooth surfaces [Wait and Murphy, 1957, 1958].
In the electrostatic case, the vertical electric field at the
summit of a hemispherical boss is exactly 3 times that on
a flat plane, and for a semicylindrical ridge the factor is 2
times. This field-enhancement effect carries over (with modifications) to VLF groundwaves, and has been demonstrated
experimentally [Harrison et al., 1971].
10.2.2 ELF Propagation

ELF propagation has been the subject of theoretical study
for many years. The texts by Wait [1970], Galejs [1972]
and Burrows [1978], along with review papers by Bernstein
et al. [19741, Wait [1974 and 19771and Bannister, [1980],
provide comprehensive descriptions of the propagation characteristics of ELF waves. Much of the discussion that follows is based on a review paper by Bannister [1982].
ELF waves below a few hundred hertz propagate with little attenuation, penetrate well into lossy media and are very
stable compared with higher frequencies. Nevertheless, ELF
has limitations relative to conventional radio communication
bands. Its restrictedbandwidth allows only very low data rates,
and because of the great wavelengths, ELF transmitting antennas are very inefficient (less than 0.5% is typical).
The energy of an ELF wave is confined principally to
the waveguide that exists between the earth and the ionosphere. At ELF the effective height h of the waveguide is
much less than the wavelength Xof the wave and the waveguide is below cut-off for all but the lowest order mode,
the transverse electromagnetic (TEM) mode. On the other
hand, at VLF the waveguide height exceeds the wavelength
and several modes propagate. At LF the number of significant propagating modes may exceed 20.
The principal TEM fields are the vertical electric field
Ev, and the horizontal magnetic field, Hh. Secondary field
components arise because the surface impedance Ngof the
ground-albeit small-is not zero; hence the term "quasi
TEM" mode. The secondary fields are small compared with
the principal fields, but are important because horizontal
antennas would not radiate if the secondary fields were zero
[Burrows, 1978].
Attenuation of the ELF quasi-TEM mode in the earthionosphere waveguide is low, on the order of 1or 2 dB/Mm.
The attenuation is caused mainly by power absorption in
the ionosphere, since the surface impedance of the ionosphere is typically much larger than the surface impedance
of the ground. That effect is evidenced in the expression
for the waveguide attenuation rate a, which is inversely
proportional to h. That behavior indicates that the rate of
power leakage from the guide is proportional to the field
intensity at the surface, whereas the rate of power flow along
the guide is proportional to the guide's volume. Thus, as
10-25
CHAPTER 10
the guide's height decreases, the ratio of power leakage to
total power flow increases [Burrows, 1978].
ELF attenuation is low enough to support very longrange propagation, and a planar model of the earth is inadequate if the path length exceeds the earth's radius. The
most important effect of earth curvature is the closure of
the guide, which allows the field to return to the source
point after one complete encirclement. Therefore, the total
field is the sum of the field arising from propagation over
the shorter great-circle path from the source, and that arising
from propagation over the longer one. The local effect of
the curvature is small, however, and the wave propagates
in the curved guide with nearly the same parameters as it
would in a planar one.
The geomagnetic field interacts with the charged particles of the ionosphere to produce an anisotropic conducting
medium. However, the electrical mismatch between the atmosphere and the ionosphere is large at ELF, and the transition between them abrupt, so the ionosphere acts much
like a perfect reflector. The effect of the anisotropy is therefore small, and ELF attenuation and phase velocity depend
only slightly on the direction of propagation.
10.2.2.1 ELF Field-Strength Calculations. The expressions most often employed for calculating ELF fields in the
earth-ionosphere waveguide are based upon an idealized
model that assumes the earth and the ionosphereto be sharply
bounded and homogeneous. Experimental measurements of
the waveguide properties at ELF have shown consistently
that they can be represented accurately by formulas based
on such a simple model. Complicated calculations that account for vertical structure of the ionosphere [Field, 1969
and Pappert and Moler, 1978] also confirm that the simple
model is adequate for many purposes.
For the idealized model, the magnitude of the magnetic
field of the signal from a horizontal dipole is approximately
IH
IdA
2Xo
c/v - 0.985
C
cos
,
h1/ho
(10.46)
and
o.143f
(10.47)
(10.47)
where ho is the altitude where o1 = weo;h1 is the altitude

where 4wuo = 1;
and 1 are the conductivity scale
heights at altitudes ho and h1 , respectively; and oj is the
conductivity of the ionosphere, which varies with altitude.
Equations (10.46) and (10.47) show that the phase constant depends primarily on the two reflecting heights and is
nearly independent of the conductivity scale heights. On the
other hand, the attenuation rate depends on the scale heights
as well as the reflection heights.
A simple exponential fit to the ionospheric conductivity
profile is given by Wait [1970] for determining propagation
parameters:
(z)/Eo = 2.5 x 105exp [(z - H)/o],
(10.48)
where H is an (arbitrary) reference height. The corresponding values for ho and h1 are
2] ( p/a 1/2
J sin p/a
-7T
H -
Ln
2.5 x 10
2iTrf
(10.49)
and
o10 p/2x 107
(10.44)
where
2+ L (2.39 x
104)
(10.50
Note that all heights are in kilometers.

A
Xg(10.45)
riwpoXc/v
is an excitation factor, Ngis the surface impedance of the

ground, c is the speed of light, v is the phase velocity of
the TEM mode and a is the radius of the earth.
There are six distinct factors in Equation (10.44): (1)
the source strength Ide (2) the excitation factor A; (3) a
collection of free space parameters, all of which are known
once the frequency is specified; (4) a spherical focusing
factor; (5) the radial propagation loss factor, including both
10-26
the exponential decay due to absorption and the p 1/2 decay

due to spreading; and (6) the directional dependence of the
radiated field. Once the current moment Ide, angular frequency w, and coordinates p, o of the field point are specified, only two parameters are left undetermined, A and the
attenuation rate, a. Both depend on the ionosphere.
Greifinger and Greifinger [1978, 1979], derived simple
approximate expressions for ELF propagation constants that
agree well with full wave numerical calculations. For daytime propagation, the approximate expressions for c/v and
a are
10.2.2.2 Theoretical and Measured ELF Propagation

Constants. It can be shown that the effective waveguide
height of reflection is roughly ho, rather than the higher
reflecting height h1. This is in excellent agreement with the
effective reflection heights inferred from ELF propagation
measurements [Bannister, 1975]. The most common values
of H and io, employed in interpreting VLF daytime propagation measurements are H = 70 km and ,o= 3.33 km.
By using these values the values of h_, h,, c/v and a can
be determined readily at ELF.

Under nighttime propagation conditions, an E region
bottom where the electron density increases very sharply is
usually encountered below the altitude h1. For a simple
model that assumes the density above the E region bottom
to vary slowly on the scale of the local wavelength, the
propagation constant is [Greifinger and Greifinger, 1979]
230.0
2-CALCULATED
* MEASURED
E 100
eIn
e MEA
/URE~
-
2.0
I
c/v
XFX,/h,,
h
(10.51)
or 2.0
and the attenuation rate is
DAYTIME
I-I
0.5
0.143f
0.2
'5 7 10
20
50
100 200
500 1000 2000
FREQUENCY,
Hz
Figure 10-24. Calculatedand measuredELF daytime attenuationrates
[Bannister,1982].
hE/ho
+
hrrkonEhE)
(10.52)
where hE is the altitude of the E region bottom and konEis

the local wave number.
Nighttime ELF attenuation rates are plotted in Figure
10-26 for frequencies from 40 Hz to 1000 Hz. The calculated
values were obtained using Wait's nighttime ionospheric
conductivity model (with a reference height of 90 km and
scale height of 2.5 km) in conjunction with Equations (10.51)
Theoretical values of ELF daytime attenuation rates are

plotted in Figure 10-24 for frequencies from 5 Hz to 2000
Hz. Also plotted are values of cxdetermined from controlled
source measurements [Bannister, 1982], or inferred from
Schumann resonance measurements [Chapman et al., 1966].
Figure 10-24 shows that the agreement between the theoretical and measured values of ELF daytime attenuation rates
is excellent.
Theoretical values of ELF daytime relative phase velocity are plotted in Figure 10-25 for frequencies from 5 Hz
to 1000 Hz. Also plotted are various values of c/v determined from measurements of atmospherics. Figure 10-25
shows excellent agreement between the theoretical and measured values of c/v for frequencies above 50 Hz, and fair
agreement for frequencies below 50 Hz.
and (10.52), and assuming the height of the E region bottom

to be 90 km and its conductivity to be 8 x 106 S/m. Also
plotted are various measured values of a. Figure 10-26
shows that, for frequencies from 45 Hz to 800 Hz, there is
excellent agreement between the theoretical and the measured values.
In addition to their very low attenuation rates, ELF radio
waves below about 200 Hz can penetrate lossy media and
retain usable strengths to substantial depths. Those features
make them attractive for communicating over great distances to sub-surface locations. For example, even for sea
water, with a conductivity of 4 S/m, the skin depth is about
36 m at a frequency of 50 Hz.
10.0
-CALCULATED
E
m 5.0
MEASURED
1.8
1.6_
CALCULATED
DAYTIME
MEASURED
z 2.0
C)
0
NIGHTTIME
0 l
5'
20
50 100 200 500
FREQUENCY,Hz
Figure 10-25. Calculated and measured daytime ELF phase velocities
[Bannister, 1982].
50
FREQUENCY,Hz
Figure 10-26. Calculated and measured ELF nighttime attenuation rates
[Bannister, 19821.
10-27
CHAPTER 10
10.2.2.3 Anomalous ELF Propagation Occasionally the
nighttime field strengths measured at 40 Hz to 80 Hz have
decreased by 4 dB to 8 dB in the northeastern United States
[Bannister, 1975, 1980]. Those relatively severe nighttime
fades sometimes occurred during the several days following
magnetic storms, when similar (but less pronounced) behavior was found to coincide with phase disturbances on
VLF paths across the northern United States. These shortpath (- 1.6 Mm) field strength reductions might have been
caused by enhanced ionospheric ionization due to precipitating electrons from the radiation belts. However, attempts
to correlate the fades with geomagnetic indices have met
with limited success. Simutaneous measurements taken in
Connecticut and the North Atlantic area during the magnetically quiet period of early March 1977 have indicated
some of these anomalies might have been caused by a moving nocturnal sporadic E layer.
Calculationsby Barr [1977] and Pappert and Moler [1978]
show that nocturnal sporadic E can produce marked maxima
and minima in the propagation characteristics of ELF radio
waves. One physical explanation for the effect is interference between waves reflected from the normal E region and
from the sporadic E region. Pappert [1980] showed that a
sporadic E patch one square megameter in extent could
account for the 6-8 dB fades that have been observed. Similarly, patches 0.5 Mm2 in extent could account for the more
commonly observed 3-4 dB fades.
Many other ionospheric disturbances can cause ELF
agation in the earth-ionosphere waveguide where conditions

change over transverse distances comparable with a Fresnel
zone. They derived an expression for the relative errors
introduced by neglecting transverse ionospheric gradients
over the path and found that the WKB method is inaccurate
when the width of a disturbance is less than two thirds of
the width of the first Fresnel zone. Further, the WKB approximation significantly overestimates the propagation
anomaly when the disturbance is centered near the propagation path and underestimates the anomaly when the disturbance is centered far off path.
Strong localized disturbances behave like a cylindrical
lens filling a narrow aperture. Lateral diffraction, focusing,
and reflection cause the transverse electromagnetic (TEM)
mode to exhibit a transverse pattern of maxima and minima
beyond the disturbance and a standing-wave pattern in front
of it. The focusing and diffraction diminish when the transverse dimension of the disturbance approaches the width of
the first Fresnel zone, typically, several megameters. Reflection from widespread inhomogeneities can be important
in two situations: first, for great-circle propagation paths
that are nearly tangential to the boundary of the disturbed
polar cap; and second, when the TEM mode is obliquely
incident on the day/night terminator, in which case a phenomenon analogous to internal reflection can occur.
propagation anomalies, including those associated with solar
10.2.3
x-ray flares, energetic electrons, protons from solar particle

events (SPEs) in the polar cap, and high-altitude nuclear
bursts. The attenuation rate can increase easily by 1 or 2
dB/Mm relative to normal daytime conditions during such
disturbances, depending on the wave frequency and severity
of the disturbance, and certain moderate solar proton events
can cause the attenuation rate to approach 4 dB/Mm at 75
Hz [Field, 1982b].
10.2.2.4 Analysis of Laterally Non-Uniform Ionospheric Disturbances. Most predictions of ELF fields in
the earth-ionosphere waveguide have used a WKB method
described by Pappert and Molar [1978]. To find E-fields
along any particular radial from the transmitter, that method
assumes the properties of the guide to depend only on distance from the transmitter. However, even such large inhomogeneities as sporadic E patches, the polar cap boundary, and the day/night terminator can cause the properties
of the earth-ionosphere waveguide to change markedly over
the huge wavelength, or Fresnel zone, of an ELF signal.
Such inhomogeneities can cause lateral reflection, diffraction, and focusing of ELF modes. Those phenomena are
usually unimportant at higher frequencies where the earthionosphere waveguide can be assumed to be vary slowly in
the lateral directions.
To handle such situations, Field and Joiner [1979,19821
employed an integral equation approach for analyzing prop10-28
Long-Range VLF/LF Propagation
Very low and low frequency (VLF/LF) waves are reflected from the lowest regions of the ionosphere (the D
region during daylight and the lower E region at night), and
apart from the sunrise and sunset periods, exhibit propagation characteristics that are very stable in both phase and
amplitude. The LF band (30-300 kHz) is useful for communications to distances ranging from hundreds to several
thousand kilometers, shorter than the almost global ranges
achievable at VLF but much longer than the groundwave
distances normally associated with the MF band.
Beginning in 1911 with Austin, various empirical formulas, deduced from numerous measurements, have been
used to estimate the field strengths of these long waves. For
example, Pierce derived a semiempirical formula to describe
VLF propagation over water during the day. That formula,
which gives the vertical electric field strength at a distance
d from a transmitter having a radiated power P (kw) at
frequency f (kHz), is [Watt, 1967]
210 N/P
=
exp
da
(10.53)
where a is the radius of the earth and the absorption term

is the exponential.
Modern mathematical approaches for predicting the
propagation characteristics of VLF/LF waves are formulated

in terms of a conductingsphericalearth surroundedby a
TM -TRANSVERSE
MAGNETIC
concentric electron-ion plasma (the ionosphere) into which

waves are launched from a Hertzian dipole source. TheTM
application of Maxwell's equations and the appropriate
-DIRECTIN
boundary conditions allow the electromagnetic field to be
E'ND
calculated everywhere. In doing this the earth and the ionosphere can be regarded as forming a waveguide (without
side-walls) in which propagation may be viewed either as
ELTM
to the higher LF band, as described below. In addition to
WAVEPOLARIZATIONS
ONOPHERE
'
'-
"M
and Spies [1964], Watt [1967], Pappert [1968], Wait [1970],

Galejs [1972], Field et al. [1976], and the AGARD proceedings on long waves edited by Belrose [1982].
10.2.3.1 Waveguide Modes. In the waveguide-mode
formulation for VLF/LF propagation an arbitrary propagating field is regarded as being composed of a series of modal
patterns that propagate with characteristic phase velocities,
little change in pattern shape, and gradually decreasing amplitudes. Generally, the more complex (or higher-order)
modal patterns attenuate more rapidly: at great distances
only the simpler (lower-order) modes may be important.
The earth-ionosphere cavity exhibits such waveguide features as cut-off frequencies, and reflections from mismatched sections caused by abrupt changes in the electrical
properties along its boundaries.
The field at a point in the waveguide depends on how
strongly the various modes are excited, and upon their relative amplitudes and phases at the observation point. Although there is evidence of wave focusing in the vicinity
of the point antipodal to the transmitter, typical field strength
versus distance curves generally show a decreasing trend,
upon which are superimposed local variations due to modal
interference.
A transmitting antenna modeled by a vertical currentelement produces transverse magnetic (TM) polarized waves,
which have a magnetic field parallel with the earth's surface,
while the electric field is perpendicular to the magnetic field
and not quite vertical. For TM waves, the earth-ionosphere
waveguide has a quasi cutoff frequency during the daytime
at about 2 kHz. Figure 10-27 illustrates idealized TM wave
vectors and mode patterns.
Airborne VLF/LF transmitting antennas with horizontal
current elements excite transverse electric (TE) waves with
horizontal electric fields, and magnetic fields in the vertical
plane. Figure 10-27 illustrates idealized TE wave vectors
mode and waveguide patterns. In general, TE fields are very
small at the ground, and the modes are difficult to excite
with groundbased transmitters.
Because of the presence of the geomagnetic field in the
EXCITED
Y
EXCITEDBY
VERTICAL
CURRENTS HORIZONTAL
CURRENTS
the references cited under the specific topics that follow,

more detailed descriptions of the features of VLF/LF prop-
agationcan be found in the worksby Budden[1961],Wait
:.
a series of wave reflections (wave-hops), or by the math-
ematical equivalent-traveling wave modes. Generally, it

is more convenient to apply the waveguide mode approach
to the VLF and lower LF band, and the wave-hop approach
TE-TRANSVERSE
ELECTRIC
FAMILIES
OF PROPAGATION
MODES
Figure10-27. Characteristicsof transversemagnetic(TM)and transverse
electric(TE)waves [Kosseyet al., 1982].
ionosphere, the TM and TE modes are not entirely independent, but are coupled. In general, when an electron acquires a velocity from the electric field of the wave the
magnetic forces cause it to have a component of motion
perpendicular to the electric field, thus causing polarization
conversion. The interaction is described by the AppletonHartree equation (for example, see Ratcliffe [1959]), and
the polarization conversion has been demonstrated experimentally (as reported by Bracewell, et al., [1951] Lewis,
et al., [1973] among others). The coupling effect is most
pronounced at night when the waves reflect higher in the
ionosphere, where the electron-neutral collision frequency
is smaller than the electron gyrofrequency.
10.2.3.2 Waveguide Propagation Equations. The detailed equations for VLF/LF waveguide propagation are
described, for example, by Wait [1970], Pappert and Bickel
[1970], Galejs [1972], and Field et al., [1976]. To illustrate the key dependencies and to define the commonly
used notations, they are given here for the case when
geomagnetic anisotrophy (that is, polarization conversion) effects can be neglected. That approximation is fairly
accurate for long-range VLF/LF propagation under normal daytime conditions.
Transverse Magnetic (TM) Modes
Conventional ground-based VLF/LF transmitters are
vertical and their fields are composed of a superposition of
TM waveguide modes. Following Field [1982c] the vertical
electric field is given by
e
exp
/ d/a
ir
IL cos 4
exp
sin d/a
exp
e
2id)G(hT)Gj(h)
Vj
d
8.7/
(10.54)
V/m,
10-29
CHAPTER 10
10-4
where the subscript j denotes quantities associated with the

jth TM mode, IL is the effective electric dipole moment of
the transmitting antenna; X is the free-space wavelength; d
is the distance from the transmitter; a is the earth's radius;
and c is the speed of light. Included is a factor cos where U is the angle between the dipole orientation and the
vertical-to account for inclined transmitting antennas (cos
1 = 1 for a vertical electric dipole). Although most quantities are in mks units, all distances (L, A,d, a) are expressed
in megameters.
The quantity Sj is essentially the eigenvalue of the jth
TM mode and must be computed numerically. At VLF,
however, S has a magnitude close to unity, so the term
S3 2/ in Equation (10.54) does not appreciably influence the
field. The magnitude of the vertical electric field depends
on the state of the ionosphere through three parameters: Aj
the excitation factor for the TM mode; oj , the attenuation
rate in decibels per megameter of propagation (dB/Mm);
and Gj , the height gain function for transmitter and receiver
heights hT and hR, respectively. The phase of the jth mode
is governed by the relative phase velocity, c/vi . These
propagation parameters must all be computed numerically
for model ionospheres having arbitrary height profiles.
Transverse Electric (TE) Modes
Airborne VLF/LF transmitters use long trailing-wire antennas that radiate a complicated superposition of TM and
Figure 10-28. Excitation factors vs frequency for lowest three TM modes

(j = 1,2,3) and lowest two TE modes (m = 1,2): ambient
TE modes. Here much of that complexity is avoidedby
day, a = 10- S/m [Field et al., 1976].
considering broadside propagation, where the great-circle

path connecting transmitter and receiver is perpendicular to
the plane containing the inclined electric-dipole transmitting
antenna. The vertical electric field produced by the vertical
component of the inclined transmitting antenna is given by
Equation (10.55). The broadside horizontal electric field
produced by the horizontal component is given by
EH
d/a
sin d/a
exp
(-
iTr)\
120(riILexp
sin ti
4EH=-l2TiXd exp
Sm 12
V-4
AI exp
(10.55)
8.7
-d)
v,"
G,(hr)G,,(hR) V/m.
The symbols are the same as in Equation (10.54), except

that m denotes the mth TE mode.
10.2.3.3 TE/TM Mode Structure. Equations(10.54) and
(10.55) show that each mode's contribution to the field is
proportional to the product of four quantities: the excitation
factor A, the transmitter height-gain function G(hr), the
receiver height-gain function G(hR), and the propagation
factor exp ( - d/8.7). This section gives calculated values
of these four quantities for a nominal ambient daytime ionosphere and an assumed ground conductivity of 10-3 S/m.
10-30
11
10-9
'
E-
o
E 10-5-
lo-lo z
z
\
',
-\
m\\
\
E
106o
10
15
\\\
20
25
30
30
35
35
40
40
45,
45
FREQUENCY,kHz
TE/TM Excitation Factors

Figure 10-28 shows the frequency dependence of the
excitation-factor magnitudes of the first three TM modes
and the first two TE modes. The first three TM modes are
excited equally at the lower VLF frequencies, but above
about 30 kHz the higher-order TM modes are much more
effectively excited than the first one. The TE modes are
excited much more poorly than the TM modes, by four or
five orders of magnitude, as shown in Figure 10-28. The
efficiency of TE mode excitation relative to TM mode excitation improves as the ground conductivity is reduced. For
example, at 20 kHz the TE mode excitation factors are
nearly two orders of magnitude greater for a 10 5 S/m conductivity than for 10-3S/m. On the other hand, the excitation
factor for the lowest TM mode is less by almost an order
of magnitude if the conductivity is changed in the same
fashion.
The excitation factors also depend on the state of the
ionosphere. The excitation factors as defined here and shown
in Figure 10-28 are inversely proportional to a quantity that
becomes the "height of the ionosphere" in the limit of a
sharply bounded ionosphere. For the diffuse ionospheres
the excitation factors at the lower VLF frequenciesare roughly
proportional to the inverse of the average height at which
important reflections occur. Thus, one would expect these
factors to become somewhat larger under disturbed conditions.
\o~l
70
60
lo
',
,
\
10
MAGNITUDEOF HEIGHT-GAIN FACTOR
kHz,
10
cr = -
et3
S/m
1976].
[Field
al
kHz, o = 10-3 S/m [Field et al.,
.,
1976
'%
=2
0:2
TE/TM Height-Gain Factors
The height-gain factor of the waveguide mode accounts

for the effects of non-zero transmitter and receiver heights.
The transmitter and receiver height-gain factors for a given
mode are identical and, therefore, are equal when the transmitter and receiver are at the same altitude. Figure 10-29
shows computed height-gain factors for the first three TM
modes and first two TE modes for a frequency of 20 kHz.
These height-gain factors exhibit the classic height-depen-
10
15
20
25
30
35
45
40
dences for antennas over a highly conductive ground; the

TM mode height-gain factors are of the order of unity over
most of the waveguide, except for some rather sharp nulls;
Figure 10-30. Attenuation rates vs frequency for lowest three TM modes

(j = 1,2,3)and lowest two TE modes(m -1,2) ambient
day, (r = 10-3 S/m [Field et al., 1976].
and, above a few kilometers, the TE mode height-gain factors increase sharply to values well in excess of one-hundred.
For elevated antennas, the large TE mode height-gain factor
mitigates the effects of the small excitation factor, and these
modes can be excited about as effectively as TM modes.
At frequencies above about 30 kHz the first TM mode
develops a broad maximum in its height-gain factor in the
40-60 km altitude range. Such "whispering gallery" behavior is not important for ground-based or airborne terminals, but may be significant for very high, balloon-borne,
agation analysis than is mode theory. In intense disturbances, however, higher order modes are much more severely
attenuated than lower, and the mode sum can be used well
into the LF regime. Figure 10-30 shows that the first TE
mode is slightly less attenuated than the first TM mode,
although that result depends on the specific normal daytime
conditions and ground conductivity assumed.
Figure 10-31 illustrates the ground conductivity depen-
terminals operating at higher frequencies [Videberg and Sales,

1973].
dence of the attenuation rates of the first TM and TE modes

at 20 kHz. Results are given for normal daytime conditions
TE/TM Attenuation Rates

Figure 10-30 shows attenuation rates as a function of
frequency for the first three TM modes and the first two TE
modes. The higher order modes are more heavily attenuated
than the lower, which often allows them to be neglected at
VLF for long path-lengths, At the higher frequencies, the
attenuation of the higher order modes can be mitigated by
efficient excitation. Under normal conditions, therefore, it
is usually necessary to retain many terms in the mode sums
for frequencies throughout the LF band. It that case, geometric optics is often a more convenient approach to prop-
and a moderate ionospheric disturbance, such as a solar

proton event (SPE). The disturbance increases the attenuation rate of both polarizations over normal values. However, the TE attenuation rate is virtually independent of
ground conductivity, whereas the TM rate exhibits a strong,
broad maximum for conductivities between 10-5 and 10- 4
S/m, where the TM eigenangle is near the Brewster's angle
of the ground. The TM mode propagates somewhat better
than the TE mode for most common ground conductivities,
but propagates much worse over low-conductivity ground,
such as occurs throughout Greenland and much of Canada.
10-31
CHAPTER 10
40
'
-TM
35
-- I
al0
\
TM MODE
i0~~~~~g~~
---
-- let TE MODE
\
~~~~\
a2
'
E0
. _\.MODERATE
--
w
5
104
1d73
o2
o
10-1
CONDUCTIVITY,
Siemens/m
Although these results pertain to a frequency of 20 kHz,

curves for other frequencies exhibit the same general behavior. The main difference is that the Brewster's angle
peak in the TM attenuation rate occurs at higher values of
conductivity for higher frequencies, and vice versa.
Except for propagationover very low-conductivityground,
TE modesare morevulnerablethanTM modeto degradation

in disturbedionosphericenvironments.In the VLF band,
for example, the TE mode attenuation becomes prohibitive
for intense disturbances, as illustrated in Figure 10-32.
TE/TM Signal Strengths Versus Distance
Figure 10-33 shows 20 kHz TE and TM signal strengths,
computed as a function of distance for ambient and intensely
disturbed ionospheric conditions. The curves provide a di-
1I o' /
,/30
/
30
t-
I< WEAK
MODERATE
INTENSE
SEVERITYOF DISTURBANCE
Figure 10-32. Attenuationratesof the lowest orderTM and TE modes,
for disturbedionosphericconditionsand a perfectlyconductingearth [Fieldet al., 1976].
10-32
INTENSE
DISTURBANCE
I
0I
1ll
DISTANCE,Mm
Figure 10-33. TE/TMsignal strengthsvs distancefor ambientday and
3
intenselydisturbedionospheric
conditions:
20kHz,o = 10S/m,
10atdegree
antenna
inclination
receiver
12.2km
[Field,
1982c].and transmitterand
enhancementsat rangesup to several megameters,caused
l/./
2 lo
O"H
,/
-I
IOkHz/
rect comparison of TE and TM polarized signals radiated

broadside from a trailing-wire antenna inclined 10 degrees
with respect to the horizontal. Both the transmitter and the
receiver are assumed to be at an altitude of 12.2 km, and
propagation path is over poorly conducting earth
(z = 10- 3 S/m).
In Figure 10-33 the ambient signals exhibit nulls and
E --TE
DAY
\Id\
0dB
al., 1976].
20
,the
,AMBIENT
Figure 10-31. Attenuation rates of the lowest order TM and TE models

vs ground conductivity, for 20 kHz under normal daytime
and moderately disturbed ionospheric conditions [Field et
-TM
A E
\DISTURBANCE
NORMALD6ATIhEI
TE
by interferenceamong several propagatingwaveguide modes.

At greater distances, the higher order modes, which are
more heavily attenuated than the lower order modes, diminish in importance and the signals fall off smoothly with
distance. The curves also reveal several differences between
signals in ambient and disturbed environments. First, and
most important, if at least 2 or 3 Mm of the path are disturbed the disturbed signals fall well below the ambient
signals. Such behavior is typical during strong solar proton
events in the polar regions, and results when such widespread ionospheric disturbances depress the effective height
of reflection in the ionosphere significantly. Second, mode
interference patterns are nearly absent in disturbed environ-
ments, indicatingthat heavyattenuationof the higherorder

modes leaves only the lowest order modes to contribute

significantly to the signal strengths. Third, at ranges under
approximately 1.5 Mm, the disturbed signals can be stronger
than the ambient signals, because the disturbed environment
increases the mode excitation factors but destroys interference nulls. However, if the disturbance covers most of the
path, but not the transmitter and receiver, the attenuation
rates increase but not the the excitations.
In Figure 10-33 the ambient TE signal is stronger than
the TM signal at all distances, owing primarily to the relatively high transmitter and receiver altitudes and the nearly
horizontal antenna orientation assumed in the calculations.
Under the disturbed condition, the TE signal is more adversely affected than the TM signal and falls below it at
most distances; however, as indicated by Figure 10-31, if
the surface conductivity was reduced sufficiently, the TE
signal would again become stronger than the TM signal.
10.2.3.4 Numerical Modeling of VLF/LF Waveguide

Propagation. Numerous sophisticated computer programs have been developed for making VLF/LF field strength
predictions. As described by Morfitt et al. [1982], the model
developed at the United States Naval Ocean Systems Center
(NOSC) is particularly attractive in that it incorporates (1)
arbitrary electron and ion density distributions and collision
frequency (with height), and (2) a lower boundary that is a
smooth homogeneous earth characterized by an adjustable
surface conductivity and dielectric constant. The model also
allows for earth curvature, ionospheric inhomogeneity, and
anisotropy resulting from the geomagnetic field. In addition,
air-to-air, ground-to-air, and air-to-ground TE/TM propagation predictions can be made involving a horizontally
inhomogeneous waveguide channel. The NOSC waveguide
model can be used for computing long wave fields at fre-
80
70
8C0
102897kHz
70
60
60
20
20
15 567 kHz
st00
0
l60
c
21 794 kHz
70
60
28021 kHz
DISTANCE,Mm
Figure 10-34. VLF signal calculations for a mostly seawater path from Hawaii to Sentinel, Arizona:-daytime,----nighttime
[Morfitt, 1977].
10-33
CHAPTER 10
\,
50
< 40
3 x~i~{'~
'E 20
i
k.
9\\
/
o0
-10,
0
.~---
\""<*~,
X-. ~
"o
250':.'
m0-
number of hops required to describe the signal strengths are

less than the number of waveguide modes that are required.
The most general wave-hop formulations describe the propagation of TE/TM waves excited by an inclined dipole over
a spherical earth with an anisotropic ionosphere [Lewis,
1970]. In addition to these sophisticated wave-hop techniques, other (simpler) approaches have been developed that
provide good, quick, estimates of VLF/LF signal strengths,
but without showing such propagation features as wave interferencephenomena or polarizationconversioneffects [e.g.,
see Lewis and Kossey, 1975].
\
x( o
10.2.3.6 VLF/LF Probing of the Ionosphere. The

of long radio waves to great distances is con-
_propagation
,
1
DISTANCE , Mm
Figure 10-35. LF daytimesignalcalculationsfor midlatitudepropagation

[Pappert, 1981].
quencies as high as 300 kHz for daytime propagation [Pappert, 1981] and as high as 60 kHz for nighttime propagation
[Morfitt et al., 1982]. Figures 10-34 and 10-35 show representative TM signal strengths computed by NOSC for
nominal daytime and nighttime models of the ionosphere
appropriate for midlatitudes, and for propagationover water.
At VLF, many of the propagation predictions have been
validated by NOSC airborne measurements, such as those
described by Bickel et al. [1970].
10.2.3.5 Other VLF/LF Propagation Prediction
Techniques. In addition to the waveguide mode formulations a number of other mathematical techniques have been
developed for describing the propagation of VLF/LF waves.
These include the zonal harmonic or spherical wave analysis
method [Johler, 1964, 1966]and the wave-hopmethod [Berry,
1964, and Berry and Chrisman, 1965]. The spherical wave
technique has the attractive feature that it can model variations in height in the earth-ionosphere cavity, but it requires
the use of large-scale, very fast, digital computers for its
implementation. For the higher VLF and LF frequencies the
wave-hop method requires the least computing time, but it
has not been formulated in a way that lends itself to modeling
of discontinuities in the earth-ionosphere duct [Jones and
Mowforth, 1982].
In the wave-hop approach the field strength at any point
is the sum of the groundwave (see Section 10.2.1) and a
series of "hops," which represent waves that have been
reflected from the ionosphere and/or the ground. The hops
are numbered according to the number of times they have
been reflected from the ionosphere. Each reflection results
in a reduction in hop amplitude so that usually a relatively
small number of hops are needed to provide good fieldstrength estimates. For VLF propagation at distances less
than about 1000 km and for LF propagation, in general, the
10-34
trolled by the lowest regions of the ionosphere (usually the

lower E region and the D region). As such, the variations
in the amplitudes and phases of propagating long waves are
very sensitive indicators of changes in the lower ionosphere.
It is not surprising, therefore, that in addition to their uses
It is not surprising, therefore, that in addition to their uses
for long-range communication and navigation long waves
are used to assess the state of the lower ionosphere and as
a tool for characterizing some of its properties.
Because of the extremely long wavelengths, ELF waves
are affected by the electron and ion densities that are present
over a very large range of altitudes. At night, for example,
that altitude range can extend from below 50 km, to well
up into the F region. Similarly, under disturbed conditions
the electrons and ions at altitudes appreciably below 50 km
can play an important (if not dominant) role in ELF propagation. However, owing to the difficulties involved in interpreting long-path ELF propagation data (which tend to
represent an "average" of the state of the ionosphere), ELF
has not been used extensively for ionospheric research.
The propagation of VLF/LF radio waves is controlled
by the region of the ionosphere below about 90 km at night
and below 75 km during the day. Unlike ELF, the observation of the signal characteristics of VLF/LF waves has
provided a relatively simple ground-based technique for exploring the state and nature of the lower ionosphere. The
technique has proven to be especially sensitive for monitoring ionospheric disturbances, such as those produced by
solar x ray flares [for example, see Reder, 1969 and Kossey
and Lewis, 1974], geomagneticstorms [Belroseand Thomas,
1968], electron precipitation events [Potemra and Rosenberg, 1973], ionospheric substorms [Svennesson, 1973],
polar cap absorption events [Oelbermann, 1970], and high
altitude nuclear bursts [Frisius et al., 1964 and Field and
Engle, 1965]. The observation of continuous-wave transmissions over very long propagation paths has often been
used for monitoring because such paths provide coverage
over very large geographical areas. As with ELF, however,
a disadvantage of such long path observations is that the
effects of relatively localized disturbances are integrated,
or smoothed-out, making it difficult to obtain information
on the severity, extent, and structure of the disturbed region
of the ionosphere.

Steep-incidence (that is, short-path) VLF/LF propagation techniques provide data on more localized regions of
the ionosphere. However, with continuous-waves the direct
and reflected components (groundwave and skywaves) overlap in space and time and can only be resolved indirectly
by observing the interference pattern on the ground [Hollingworth, 1926] or by direct interpretation of diurnal phase
and amplitude changes [Bracewell et al., 1951]. For example, the interference patterns produced by the groundwave and skywave from a 16 kHz transmitter were used by
Bracewell and Bain [1952] to first suggest the presence of
two ionized layers well below the ionospheric E-region.
Phase and amplitude observations can be used to characterize the steep-incidence VLF/LF reflection properties of
the lower ionosphere. Of particular interest is the use of the
data to determine effective heights of reflection and effective
plane wave reflection coefficients of the ionosphere [Bracewell et al., 1951]. Such experimental data can be compared
directly with that obtained theoretically, using full-wave
computational techniques in conjunction with electron density and collision frequency models of the ionosphere [for
example, see Budden, 1961, Pitteway, 1965 and Inoue and
Horowitz, 1968]. Thus, the experimental data can be used
to validate theoretical models of the ionosphere, such as
those obtained from the chemistry of the upper atmosphere.
In addition the data can be used to develop and validate
phenomological models of the lower ionosphere, important
for long wave propagation prediction [for example, see Bain,
1982].
More recently, with the advent of high-resolutionVLF/LF
pulse ionosounding [Lewis et al., 1973] it became possible
to observe ionospheric reflections free of the ambiguities
of the groundwave and skywave interference phenomena
characteristic of continuous-wave measurements. The technique has been used to obtain a variety of steep-incidence
reflection data at low-, mid-, and high geomagnetic latitudes
[Lewis et al., 1973 and Kossey et al., 1974], and to investigate features of the C-layer of the daytime ionosphere
[Rasmussen et al., 1980 and Rasmussen et al., 1982]. The
technique provides a relatively direct means for determining
VLF/LF ionospheric reflection heights and effective plane
wave reflection coefficients, which then can be used to develop electron density models of the lower ionosphere [Kossey et al., 1983].
The inversion of steep-incidence VLF/LF reflection data
to obtain electron density models of the lower ionosphere
is not an easy task. Under quiet ionospheric conditions it
is especially difficult, since usually the polarization conversion effects of the geomagnetic field cannot be ignored.
Nevertheless, mathematical approaches that employ fullwave and iterative computational techniques have been developed and applied with some success [for example, see
Shellman, 1970 and Field and Warber, 1984]. Under disturbed ionospheric conditions and certain daytime ambient
conditions, when the VLF/LF reflections are controlled
primarily by ionization below about 70 km, the effects of
the geomagnetic field are greatly diminished and the ionosphere can be assumed isotropic. Under such conditions the
mathematical inversion problem becomes somewhat simpler.
Field et al. [1983] have developed an inversion technique,
appropriate for isotropic propagation, which has been used
in conjunction with VLF/LF pulse reflection data to derive conductivity profiles of the severely disturbed polar
ionosphere.
A problem with profiles calculated by inversion is that
of nonuniqueness, which can be caused by either incompleteness of data or the nonlinear dependence of the reflected
signal on the propagation medium. In addition, the profiles
characterize narrow regions of the ionosphere, since the
propagation data contain information about only those altitudes where the ionosphere interacts appreciably with the
reflected wave.
The altitude constraints are even more severe if long
path propagation data are used, rather than steep-incidence
reflection data. Nevertheless, some effort has been devoted
to deducing the structure of the ionosphere from long path
data. As described by Crain [1970], the data in this case
are the attenuation rates and phase velocities of the propagating waveguide modes, and the analysis is a trial-anderror technique effectively to find an ionospheric conductivity profile which provides a waveguide mode or wave
hop structure that agrees with the observed distribution of
radio field strength.
In essence the long-path technique is similar to the steepincidence approach insomuch as ionospheric reflection coefficients are calculated as an intermediate step in obtaining
the mode constants. In order to synthesize the total field as
measured, one has to take care to add in as many modes as
contribute to the field. This can add a great deal of complexity to the application of the technique. Nevertheless,
the technique has been applied with much success to develop
phenomenological models of the lower ionosphere. Although such models may not be consistent in all respects
with those derived from detailed analyses of the aeronomy
of the upper atmosphere, they have found widespread application in long wave propagation prediction codes.
The results obtained by Bickel et al. [1970], Morfitt
[1977] and Ferguson [1980] are especially noteworthy in
that regard. They have performed detailed analyses of a
large volume and a wide variety of VLF/LF propagation
data and have derived analytic models of the lower ionosphere for propagation prediction. Those models are simply
exponential height-profiles of conductivity, which can be
specified by only two parameters, scale height and reference
height. Following Wait and Spies [1964] the conductivity
parameter, wr, depends on the ratio of electron density to
electron-neutral collison frequency, and is taken to be of
the form wr(z) = 2.5 x 105 exp (B(z-H')), where z (km)
is altitude, B is the inverse scale-height (km1) and H' (km)
is a reference height. The value of electron density N(z),
in electrons/cm3 , is calculated as a function of height by the
equation N(z) = 1.43 x 107 exp(B(z-H')-0.15z). The
10-35
CHAPTER 10
Table 10-5. Suggested exponential profiles for use in long wave propagation prediction codes. Frequencies, f, are in kHz [Morfitt et al., 1982].
Seasonal-Diurnal
Propagation
Condition
Summer day
Summer night
Winter day
Winter night
H' (km)
(km )
B
Magnetic Dip (0)
70
87
74
80
0.5
0.0077f + 0.31
0.3
0.035f - 0.025
(10 < f < 35)
Linear change between high and
middle latitude
0.0077f + 0.31
collision frequency v (collisions/s) is given by

v(z) = 1.82 x 1011exp(-0.15z). Table 10-5 gives exponential profiles, based on VLF/LF propagationdata, which
are suggested for use in long wave propagation prediction
codes [Morfitt et al., 1982].
10.2.4 MF Propagation
At night medium frequency skywaves can propagate to
considerable distances with relatively little attenuation, but
during the day the skywaves are severely attenuated in passing through the ionospheric D-region, so that only the
groundwave provides usable signals. Thus, during the daytime MF signals are very stable, while at night they are
much less so owing to the variability of the lower E-region
of the ionosphere, and to interferences between the groundwave and skywaves.
10.2.4.1 MF Groundwave Propagation. The propagation of MF groundwaves can be described using the techniques discussed in Section 10.2.1. Because of the shorter
wavelengths, however, such factors as the earth's atmosphere (and hence, the effective radius of curvature of the
earth), terrain elevation, conductivity changes, and trees and
buildings along the propagation path usually influence MF
90-75
(high latitudes)
(transition latitudes)
<70
(middle latitudes)
associated with ground-reflections on multi-hop paths (see

Figure 10-36). The problem is further compounded in that
the ionospheric absorptionlosses alone show significantshortperiod and day-to-day variations, as well as diurnal, seasonal, latitudinal, and solar-cycle effects [Knight, 1982].
A number of techniques have been developed to estimate
MF field strengths. A relatively simple technique, based on
an empirical formula has been adopted by the CCIR [CCIR,
1978]. A more complex wave-hop method has also been
developed by Knight 11975].Figure 10-37 shows computed
ionospheric losses over the 500-1500 kHz band for singlehop paths at mid- and low geomagnetic latitudes. In the
auroral zones the ionospheric losses are somewhat greater
than those shown in the figure, and the estimates do not
apply if the waves penetrate the E-layer and are reflected
by the F-layer. The latter is most likely to occur at frequencies above 1500 kHz [Knight, 1982].
10.2.4.3 Effect of MF Waves on the Ionosphere. Even
relatively small electromagnetic fields impart appreciable
energies to the electrons in the ionosphere causing their
temperatures, and consequently their thermal velocities, increase. This increases the effective electron-neutral collision
frequency; as a result, the complex dielectric constant of
groundwave characteristics to much larger extents than they

do for VLF/LF waves. Most of these effects can produce
INPHERIC
strong local interference patterns in the amplitudes of MF

groundwaves
[Knight, 1982 and Hizal and Fer, 1982].
POLARIZATION
/
COUPLINGLOSS
COUPLINGLOSS
10.2.4.2 MF Skywave Propagation. Because there are

so many factors that affect the characteristics of MF skywaves, it is difficult to draw a representative
set of propa-
gation curves. These factors include the losses associated

with an imperfectly conducting earth at the transmitter and
receiver, polarization coupling losses that depend on the
geomagnetic field, ionospheric absorption losses, and losses
10-36
POLARIZATION
RUN
o
LOSS
TRASMITTER
RECEIVER
Figure 10-36. Losses associated with MF skywave propagation [Knight,
19821.

6,
......... |
,,,,,
10.2.5 Long Wave Propagation Through
| W|
m
14
the Ionosphere
J0
Electromagnetic waves cannot propagate in an ideal

plasma unless the wave frequency is less than the plasma
_frequency, or approximately 8980 N, where N is the number
of electronsper cubic centimeter. Thus, a density only slightly
more than 1 el/cm3 would suffice to completely reflect a 9
kHz wave. The well-known "whistler" phenomenon, how-
0.7
i'
l.5
U),
Z
:t
4
'~T~i~-~
-es--------s---5
10
11
1213 15
'
20o
HOPLENGTH,kmx100
Figure 10-37. Computed MF ionospheric reflection losses:-east-west
propagation at all latitudes,---north-south propagation at
the magnetic equator [Knight, 1982].
the medium becomes appreciably dependent on the field.

Thus, the associated physical processes, and the differential
equations which describe the radio wave propagation in the
ionosphere, become non-linear. This gives rise to various
phenomena, including cross-modulation and de-modulation
which, in principle, can be observed experimentally. The
effects depend on wave frequency and the collision frequency and are such that they have been mainly observed
in the MF band, particularly at night.
The main influence of the non-linear effects on the properties of radio waves reflected from the ionosphere manifest
themselves through self-interference of the wave; that is, its
influence on itself, and the interaction between a number
of waves. These waves can be of the same frequency, or
two independent modulated or unmodulated waves of different frequencies. An early observation of such an effect,
the so-called Luxemburg effect, was reported by Tellegen
[1933] who noted that the signal received in Holland from
a 650 kHz Swiss station appeared to be modulated by the
signals from a powerful station at Luxemburg (252 kHz).
This phenomenon of cross-modulation has been found to be
quite common when the unwanted, or disturbing, station is
situated near the transmission path of the wanted wave [see
Davies, 1969 and Al'pert, 1960].
Although the phenomenon of ionospheric cross-modulation or Luxemburg effect was discovered accidentally, it
has been systematically investigated ever since its discovery. The motives of the earliest theoretical and experimental
investigations were interest in the effect itself and the possible use of the effect for probing the ionospheric regions
in which the cross-modulation occurs [Fejer, 1970]. In the
latter case techniques have been developed to obtain Dregion parameters such as electron density and collision
frequency profiles.
Another MF probing technique, the "partial reflection"
experiment, has proven to be a valuable method for obtaining quantitative measurements of electron distributions
in the lower ionosphere. It has been one of the most extensively employed techniques for synoptic studies of the ionosphere below 100 km [Belrose, 1970, 1972].
ever, demonstratesthat undercertainconditionslong waves

can penetrate even through the F-max region of the ionosphere, where the electron density is one hundred thousand
times larger than would produce complete opacity if the
ionosphere were a simple plasma. The long wave ionospheric transmission window is due to the geomagnetic field
of the earth, which constrains the electron motion produced
by electromagnetic waves incident on the ionosphere. As
such the magnetic field provides a propagation mechanism.
The term "whistler" refers to an audio-frequency phenomenon associated with lightning discharges in the lower
atmosphere. Electromagnetic energy at audio frequencies
emitted by such discharges propagate in the ionosphere in
a highly dispersive manner. The higher frequencies travel
faster than the lower ones with the result that the signal,
which was originally impulsive, is received over a relatively
long time interval with the frequency generally descending
with time (hence, a whistling sound).
Extensive studies have been made to determine the properties of whistlers [for example, see Storey, 1953, Ratcliffe,
1959, Pitteway, 1965 and Helliwell, 1965]. Their results
[Watt, 1967] show that from the complete expression of the
Appleton-Hartree equations for the refractive index, approximate expressions can be developed which give insight
into the nature of the whistler mode. For example, the energy transmitted through the ionosphere is well-coupled to
the whistler mode when the direction of propagation is in
the same direction as the earth's magnetic field. For transverse propagation, this coupling is very poor. Also, the
coupling is increased for sharp gradients of refractive index
at the ionosphere boundary. When collisions are included,
there is a finite range of angles between the wave normal
and the direction of the earth's magnetic field for coupling
to the whistler mode. In an anisotropic ionosphere, the wave
normal and the direction of energy flow along the field lines
are different depending upon the magnitude of the refractive
index and the static magnetic field. If ducts of ionization
that are aligned with the earth's magnetic field exist, the
waves can be guided in them.
VLF signals from terrestrial transmitters have been observed at satellite altitudes [Leiphart et al., 1962], and have
been tracked from ground level to altitudes of 500 km or
more [Orsak et al., 1965 and Harvey et al., 1973]. Such
probes showed that the polarization changes from linear to
circular as the wave penetrates the ionosphere, and that very
significant delays in the signals occur.
Such phenomena can be calculated using full-wave tech-
10-37
CHAPTER 10
.I.l_l
90c
--- ABIENT
,I
DAY
-NIT/
60
/
I
4,) 40
V 40 -
i/
I
/
30
F
20
lo
_-
3
lo
O3 lo-2 lo 10 s1 12
14
O15 i6
~
07~108 109
1010
FREQUENCY, Hz
Figure 10-38. Computed long wave transmission losses through ambient

day and night ionospheres [Booker et al., 1970].
niques [Pitteway, 1965] or WKB methods [Booker et al.,

1970], in conjunction with appropriate models of the ionosphere. Figure 10-38 gives computed long wave transmission losses for plane waves incident on the ionosphere
in a direction parallel to the geomagnetic field. The results
are representative of those expected for ambient ionospheric
conditions at mid- and high geomagnetic latitudes. Under
disturbed conditions or at very low geomagnetic latitudes
the penetration losses are much more severe, especially for
frequencies above a few kilohertz [Booker et al., 1970 and
Harvey et al., 1973].
10.3 IONOSPHERIC MODELING

For successful radio communication, it is essential to
predict the behavior of the ionospheric region that will affect
a given radio communication circuit. Such a prediction will
identify the time periods, the path regions and the sections
of high frequency bands that will allow or disrupt the use
of the selected high frequency communication circuit. This
need for prediction leads to modeling of the ionosphere.
A model is a numerical statistical description of the
ionosphere in terms of location (geographic or geomagnetic
latitudes and longitudes), time (solar zenith angle), seasons,
and other factors such as the solar activity (10.7 cm flux,
sunspot number). The empirical equations are derived from
the dependence of the observed phenomena on variables
mentioned above. These observed phenomena include: the
behavior of critical frequencies* foE, foFl, foF2, and foEs
for the E, Fl, F2, and sporadic E layers; the altitudes (hmE,
hmFl, and hmF2) for peak (maximum electron densities for
*The critical frequency is the limiting radio frequency below which a radio wave is reflected by, and above which
it penetrates and passes through, the ionized medium (an
ionospheric layer) at vertical incidence.
10-38
the layers, and the half-thickness widths** ymE, ymFl; and

ymF2. These models are called phenomenological models
[Barghausen et al., 1969;Bent et al., 1972;Ching and Chiu,
1973; Chiu, 1975; and Kohnlein, 1978].
The ITS-78 model [Barghausen et al., 1969] based on
the analysis by Jones et al. [1966] of world-wide, ground
based ionosonde data, predicts only the bottomside of the
ionosphere. The Bent model [Bent et al., 1972] predicts the
total electron content of the ionosphere in the altitude range
from 150 to 2000 km, without a direct consideration of the
bottomside E and Fl layers. The Ching and Chiu [1973]
model covers the altitude range from 110 to 1000 km. Instead of parabolic layers [Barghausen et al., 1969] they
assume Chapman functions for the electron density distributions in the E, F1, and F2 layers. Later Chiu [1975]
modifiedthe Ching-Chiu model to incorporate the polar ionosphere. Their models are useful only for studying the large
scale phenomena such as global thermospheric and ionospheric calculations.
Using ionospheric data from ESRO satellites, Kohnlein
[1978] extended the altitude range up to 3500 km. He suggested a "differential approach" for the ionospheric modeling. He separated small scale spatial structures such as
the equatorial trough, the midlatitude trough and the polar
ionosphere, from the large scale global structure. He modeled these individual structures and added them into the
global structure. His method reduces the number of coefficients otherwise needed to model the complicated ionospheric behavior.
The other approach for ionospheric predictions is to use
theoretical models [Stubbe, 1970; Strobel and McElroy,
1970; Nisbet, 1971; Oran et al., 1974; and Oran and Young,
1977].Theseare based on the physical processes responsible
for the production, maintenance and decay of the ionosphere. A theoretical model would thus rely on the process
of ionization of neutral atmospheric constituents by the incident solar extreme ultraviolet radiation, the transport processes such as diffusion and neutral winds, and also on the
effect of electric and magnetic fields on the transport processes. Essentially the theoretical model tries to explain the
experimental observations in terms of known physical processes. In addition, this approach seeks new physical processes to explain the differences between the observational
results and the predictions based on the theoretical considerations.
Radio communication can be divided in two main categories. Ground-to-ground radio communication is based
on the reflection and scattering characteristics of the ionospheric layers. On the other hand, ground to satellite, satellite
to ground, or satellite to satellite radio communications depend on the transmission and refraction characteristics of
**The half-thickness width ym of the ionospheric layer is
determined with the assumption that the layer has a parabolic
shape [Appleton and Beynon, 1940].

the ionosphere. The main goal of any modeling effort is to
predict the periods of good or poor radio communications
for the selected paths to enable a continuous undisrupted
communication through the ionosphere or by some other
means.
We will consider several ionospheric models that are
routinely used (or are available) for the prediction and specification of the ionosphere. The emphasis here is on acquainting the user with the modeling programs and their
limitations. We do not attempt to review the scientific literature for a determination of the state of the art of modeling
efforts. Therefore only the essential references will be cited.
First, we will consider the numerical-phenomenological
models. Then we will consider the theoretical models. This
will be followed by the modifications to models to take into
account high latitude phenomena such as the auroral E layer
and the midlatitude F region trough. In the concluding section, we will look at the limitations of these models and a
possible approach to overcome these limitations.
10.3.1 The Numerical-Phenomenological

Models
At present the three most widely used numerical models
for ionospheric predictions are (1) The ITS-78 model, (2)
The Bent model, and (3) the Ionospheric Communications
Analysis and Prediction Program (IONCAP). In addition,
the 4-D model of the Air Force Global Weather Central and
the Bradley model will be considered. We will also look at
the International Reference Ionosphere-IRI 79.
10.3.1.1 The ITS-78 Model. The main purpose of this
model [Barghausen et al., 1969] is to predict long term
performance of communication systems operating in the
2-30 MHz frequency range. The ITS-78 model and its computer program was developed by the Institute of Telecommunication Sciences, ESSA, Boulder, Colorado. The model
is based on the presentation of the ionospheric characteristics
in a form of synoptic numerical coefficients developed by
Jones and Gallet [1960] and improved by Jones et al. [1966].
The important features of the ITS-78 model are the parameters for the D, E, Es (sporadic), and F2 layers of the
ionosphere.
The model provides (output) circuit operational parameters such as the maximum usable frequency (MUF), optimum traffic frequency (FOT), and the lowest usable frequency (LUF). In addition to the regular E-layer propagation
mode, it takes into account propagation via the sporadic E
layer. The program computes all the probable modes. It
computes the system performance. For that purpose it calculates the antenna patterns and gains for 10 most commonly
used antennas. It also has a program to determine MUF as
a function of the magnetic activity index Kp,
The inputs for the ITS-78 model are the date, Universal
Time, geographical location (latitude and longitude) of the

transmitter and receiver, and sunspot number. To compute
the system performance the model needs the antenna parameters, the radiation power of the transmitter, and the
signal to noise ratio of the receiver.
For the D region the model considers only the absorption
losses. The non-deviative absorption is in the form of a
semi-empirical expression. It enables the user to compute
the losses for the HF frequencies penetrating the D layer.
The deviative absorption losses are included in the loss
calculations as uncertainty factors.
For the E region the model computes the parameter foE.
It assumes a constant height of 110 km for the maximum
(peak) electron density of the E layer, with a constant semithickness of 20 km. The numerical coefficients for foE are
based on the experimental ground ionosonde data during
high solar activity phase in 1958, and the low solar activity
phase in 1964.
For the F2 region the model computes the parameters
foF2, the height of maximum electron density hmF2and the
semi-thickness ymF2 of the F layer. These are in the form
of numerical coefficients for the high (1958) and low (1964)
phases of the solar activity. Both the E and F2 layers are
assumed to be parabolic in shape.
The sporadic E (Es) layer could be very helpful or harmful to radio communications depending on the nature of the
Es layer. A blanketing, totally reflecting Es layer extends
the frequency range of the E-mode communications. However a semi-transparent or partly reflecting Es layer would
cause serious multipath and mode interference and would
be detrimental to communication systems. Using numerical
coefficients, the ITS-78 model computes foEs only for the
ordinary ray. (The earth's magnetic field splits the incident
ray into the ordinary and the extraordinary rays.) The numerical coefficients are for both the high (1958) and low
(1964) phases of solar activity. As the model does not predict the occurrence of Es, the foEs maps are used only when
propagation via regular E layer is not possible. To compute
the system performance, the model incorporates three kinds
of noise: galactic, atmospheric, and manmade.
To determine the operational parameters such as the
maximum usable frequency (MUF), the model computes
the path geometry (between the transmitter and the receiver).
The parameters computed in the path geometry are the path
length, path bearing (azimuth), and the solar zenith angle
X of the sun. The model computes paths for reflections from
E, Es, and F2 layers. These are called the E, Es, and F2
modes. The paths could involvemore layers (multiple modes)
and more reflections (multihop).
To determine wave propagation the electron density distribution with altitude is needed. Both the E and F layers
are assumed to have parabolic shapes. The maximum usable
frequency (MUF) is obtained by multiplying the critical
frequency of the layer by the MUF factor M(3000). The
term in parentheses refers to the standard ground distance
10-39
CHAPTER 10
of 3000 km between the (hypothetical) transmitter and the
receiver. The experimental data for the numerical factor
M(3000) (in terms of coefficients) come from 13 ionosonde
stations covering the geomagnetic latitude range from 7 S
to 880N.
The stability and predictability of the E layer results in
a 99% probability (highest) of supporting radio propagation
and communication via the E layer. The next highest probability is via the regular F layer. When neither of the above
modes is possible, the Es mode is considered for communication.
For computing the system performance the program allows a selection from 10 antenna patterns. The program
takes into account the ground losses, ionospheric losses,
free space losses, and the excess losses. The program computes the radio communication circuit reliability, service
reliability, and the multipath evaluations.
The ITS-78 model has several limitations. The results
from the model are useful only when the operating frequency
is below the maximum usable frequency. The model assumes that transmission will be by reflection from the ionosphere. For this the transmitter and the receiver must be
on the same side of the ionospheric layer (for example,
ground-to-ground communication. The model does not take
into account the daytime Fl layer which usually develops
between the E and F2 layers. The model does not adequately
account for the electron density above the altitude of hmF2.
Finally, the model does not take into account the dependence
of absorption on the operating frequency in considering the
D layer absorption.
10.3.1.2 The Bent Model. The Bent Model [Bent et al.,
1972; Llewellyn and Bent, 1973] is basically for groundto-satellite communication but can be adapted for groundto-ground or satellite-to-satellite communication. The main
purpose of the Bent model is to determine the total electron
content (TEC) of the ionosphere as accurately as possible
in order to obtain high precision values of the delay and
directional changes of a wave due to refraction. Ground-tosatellite communicationdemands operatingfrequencies which
are higher than the MUF. Thus the mode involves the transmision refraction characteristics of the F2 layer and the
electron density distribution above the height of the F2 peak
must be known.
The model provides (output) the vertical total electron
content above the transmitter, the profile of vertical electron
density with altitude, and the total electron content along
the path between the satellite and the ground. It also provides
the refraction corrections to the elevation angle, the range,
and the range rate.
The input parameters to the model are the date, Universal
Time, locations of the transmitter and receiver (ground and
satellite), rate of change in elevation and altitude of the
satellite, operating frequency, the solar flux (10.7 cm flux),
and sunspot number.
10-40
The data base of the Bent model consists of 50 000

topside ionospheric soundings, 6000 satellite measurements
of electron density and 400 000 bottomside soundings of
the ionosphere. The data extend from 1962 to 1969 to cover
the maximum and the minimum of the solar cycle.
The bottomside data are foF2 hourly values from 14
stations in the American sector covering geographic latitudes 76 N to 12 S or geomagnetic latitudes from 85 to
0 . The topside soundings cover the period 1962 to 1966,
with geomagnetic latitude range 85 to - 750, and the electron density profiles are from 1000 km down to the altitudes
of the foF2 peak (h,,,F2). The satellite data are from the
Ariel 3 satellite covering the period May 1967 to April 1968
and are linked with real time foF2 from 13 ground stations.
Thus the data base of the Bent model refers to solar cycle
20, while the data base of the ITS-78 model is from solar
cycle 19.
The Bent model uses foF2 from the ITS-78 model. Instead of the monthly median values, the Bent model computes average values for every 10-day interval of the month
from the 10.7 cm solar flux input. For the height hmF2, it
uses M(3000) factors of NOAA (ITS-78) in terms of the
sunspot number. It uses an empirical polynominal for M in
computing hmF2, in place of the Shimazaki equation [1955]
used by ITS-78.
The distribution of electron density with altitude, assumed by the Bent model for the computation of the total
electron density is shown schematically in Figure 10-39.
Starting from the bottom, it divides the profile into five
sections; a bottom bi-parabolic F2 layer; a parabolic F2 layer
above the peak; and three exponential sections to cover the
altitude above ho (hmF2 to 1000 km). The construction of
the profile needs the parameters k1, k2 , k3, Yt, Ym, foF2,
and hmF2. The last two have already been explained. The
dependence of the other parameters on geomagnetic latitude,
solar flux, foF2, and the season is from work of Bent [Llewellyn and Bent, 1973]. The topside and the first adjoining
exponential section are matched at a height d (above hm)
by the equation
I
(10.56
where Y1 is the half thickness of the F2 layer and k1 is the

exponential constant.
The remaining profile above the F2 peak [of altitude
range (h,,, + d to 1012) km] is divided in three equal intervals of altitude.
The model can predict with an accuracy of 75%-80%.
If the model is updated with observed recent data within a
range of 2000 km radius (from the transmitters), the predictability is improved to 90%.
Though the model predicts total electron content (TEC)
with good accuracy, the model does not have separate E
and F1 layers. As the model was constructed for the TEC,

HEIGHTh
l000 km-
In the system performance options, 22 performance pa-
rameters are available. The program for the antenna output

option computes the elevation angles and the operating frequencies for optimum antenna geometry and its gain.
Inputs for the program are the date, Universal Time,
geographical locations of the transmitter and the receiver,
and sunspot number. The program can accept external ionospheric parameters as input to the program. For antenna
pattern, one can select the antenna from 17 antennas in the
program (7 antennas from ITS-78 have been modified). For
the system performance additional inputs such as radiation
aN N e' 3
h2 -
3;
N:
N=Ne
hi_
LtL__.
__ I
k202
=--N.t e l '\ 1'

I NNoe
N
N:
power of the transmitter,
ho.
h.L
-8i, L. 1-----
N N(
tI-
b. \2
-........
fib
2I
j\
Ym
I N Nm( N: I I;
I ~f,,F2
YmF1, and YmF2 respectively. For the Fl layer, hmF1/YmF1
if
f. F2
I
N2
iI
II
N1
NoNm
and the S/N ratio of the receiver
are needed.
The schematic for the electron density distribution with
altitude
-____~for the IONCAP program is shown in Figure
10-40. The model has 3 parabolic layers, E, Fl, and F2.
2) The altitudes for the peak electron densities are hmE, hmF1,
and hmF2. The half thickness widths for the layers are YmE,
is assumed to be 4. The E layer has fixed altitudeshmE = 110
km and YmE = 20 km. IONCAP improves on the ITS-78
model by incorporating D and Fl layers. The D layer con-
ELECTRON
DENSITY,
N
tribution is considered indirectly by adding an exponential

tail for the E layer down to the altitude of 70 km. In the
transition region between the E and Fl (or F2 if Fl is absent)
Figure 10-39. Schematic for exponential and bi-parabolic profiles for the
electron density distributionwith altitudefor the Bent model.
the E and Fl layers are included as the bi-parabolic bot-
HEIGHT,km
tomside of the F2 layer. Also, the program does not take

into account the non-deviative absorption in the underlying
D layer.
10.3.1.3 The Ionospheric Communications Analysis and

Prediction Program (IONCAP). The IONCAP [Lloyd
et al., 1978] is essentially the latest, improved, and more
versatile and flexible version of the ITS-78 model.
It provides 30 output options which can be divided into
four categories, (1) ionospheric description, (2) antenna patterns, (3) MUF predictions, and (4) system performance
VIRTUAL
hmFz.. ....
T
ymF2
PARABOLIC
NOSE
mFI
predictions.
For ionospheric predictions it provides monthly median

values for the parameters foE, foF1, foF2, hmE, hmFl, hmF2,
YmE, YmF1, and YmF2.It also provides the lower, median,
and upper decile values of the minimum foE or foEs. It can
also provide a prediction in the form of a plot of operating
c
LINEARVALLEY
frequency with virtual height and also with true height.
The MUF option of the output provides the minimum

radiation angle and the M factors for all four modes, E, Fl,
F2, and Es. The plots for the diurnal variation of the MUFs
are also available. The MUFs provide the description of the
state of the ionosphere and do not include any system parameters. The operating frequency for a given radio communication circuit is the critical frequency of the layer multiplied by the MUF factor.
LINEAR
(or PARABOLIC)
LEDGE
ymF
-
hmE
YE_
70
PARABOLC
NOSE
-EXPONENTIAL
/70m I
TAIL
I
fOE
foFI
FREQUENCY
foF2
Figure 10-40. Schematic for the electron density profile and virtual height
for the IONCAPmodel.
10-41
CHAPTER 10
layers, the electron density is assumed to be linear for the
frequency range fv to fu where fv = xv x foE and f,, = xu
x foE. Typical values for xv and xu are 0.85 and 0.98
respectively. Thus the electron density decreases above the
parabolic nose 0.85 foE (<foE) and continues upwards up
to 0.98 foE (<foE) producing a linear valley in the transition
region. When the x,, = xv = 1 the valley is absent in the
transition region and the curve is a vertical line starting at
the tip of the parabolic nose of the E layer. The F1 layer
forms a linear or parabolic ledge depending on the magnitudes of hmF1,hmF2, foF1, and foF2. In the ITS-78 model,
the Fl layer is assumed to be absent. In the IONCAP model
the numerical coefficients for foE are functions of geographic latitude for both solar maximum and minimum from
the work of Leftin [1976]. The model uses foF1 maps of
Rosich and Jones [1973]. It also takes into account the
retardation below the F2 layer.
For the MUF computations the model uses the corrected
form of Martyn's theorem. As the absorption equations using the secant law do not work for lower frequencies at
altitudes below 90 km, these equations have been modified
in the IONCAP program. The IONCAP provides two programs (1) the ITS-78 short path geometry and (2) the long
path (>10 000 km) geometry. In addition to the ITS-78
model, the path computations now include the Fl mode,
the over-the-MUF mode, D and E region absorption losses,
and sporadic E losses. A correction to frequency dependence
is added for low frequencies reflected from altitudes below
<90 km.
The improvements over ITS-78 can be summarized (see
IONCAP) as follows:
1. The description of the ionosphere is now more complete.
2. The loss equations have been supplemented. This
includes E mode adjustments, sporadicE effects, overthe-MUF losses, and losses for low reflection heights.
3. The ray path geometry calculations have been revised. This was an empirical adjustment to Martyn's
theorem.
4. The loss statistics were revised to include the effects
of the sporadic E layer and of over-the-MUF modes.
5. A separate long path model was developed.
6. The antenna gain package was revised.
All models predict only the quiet ionosphere, which
shows a large systematic dependence on latitude, longitude,
season, time, and sunspot activity.
10.3.1.4 The Bradley Model. The Bradley model contains two modifications to the existing models: (1) the filling
of the valley between the E and F layers (Fl, or F2 if Fl
is absent), by parabolic layers [Bradley and Dudeney, 1973]
and (2) a simple formulation of the prediction of the probability of the high-frequency propagation [Bradley and Bedford, 1976].
The assumption that the electron density distribution in
the E and F layers is parabolic in shape, results in a valley-
10-42
a reduction in electron density between the two peak values,

NmEand NmF-at altitudes hmEand hmF,respectively. The
rocket observations have shown that in reality the electron
density at any altitude between the altitudes hmEand hmF
is rarely smaller than the peak electron density NmE. Thus
the assumption of the parabolic shapes for E and F layers
underestimates the electron density in the altitude region
between hmEand hmF. To correct such an underestimation
Bradley and Dudeney [1973] suggested a linear distribution
of electron density from foE to 1.7 foE. The lower end is
at hmE. At the upper end, the F layer is parabolic in shape
down the altitude where the plasma frequency is 1.7 foE.
This linear interpolation has not yet been incorporated in
the IONCAP model (see Figure 10-40)of Lloyd et al. [1978].
In high frequency prediction it is essential to know the
probability of communication at any particular operating
frequency f. For convenience the observed data are expressed as follows: Fu is the upper decile, fmis the median,
and Fe is the lower decile of the ratios of f/fm. The distribution functions of Fu and Fe are not simple Gaussian distributions (Fu - MUF = MUF - Fe for Gaussian). The
distributions are x2-distributions of Fu and Fe. For probability determinations these two x2-distributions (of Fu and
Fe) have to be used. Two variables Fu and Fe with their
associated degrees of freedom, and the need to integrate the
x2 -distribution curve makes the process of determining the
probability distribution very cumbersome. Bradley and Bedford [1976] derived simple empirical equations for this probability distribution. The equations are
80
- (f/fm)
1 -F
(10.57)
or 100, whichever is smaller for f<fm
and
80
()
301
(10.58)
or 0, whichever is larger for f>fm,

where
Q - is
f - is
fm - is
Fe - is
Fu - is
the cumulative probability,

the operating frequency,
the predicted median frequency,
the lower decile (of f/fm),
the upper decile (of f/fm).
They note that the probability distribution from their

simple empirical equation is as good as, though not always
better than, that from the X2 distribution. Therefore they
highly recommend a replacement of X2distribution proce-

dure by these equations for a determination of the present
probability that signals will propagate at a given hour over
a given sky-wave path.
The latest computer model like the IONCAP has not
incorporated "Bradley Features" in its program.
10.3.1.5 The Air Force Global Weather Central 4-D
Model. The input data to the 4-D numerical model [Flattery et al., personal communication] are the critical frequencies for the layers and M(3000), real time or near real
time observations from 40 ground stations around the world,
and total electron content (TEC) from eleven stations. The
frequency of observations varies from hourly (best) at one
end to weekly (worst) at the other end. The desired purpose
of the 4-D model is to produce a consistent ionospheric
specification anywhere in the northern hemisphere for a 24h period. In that sense it is not a forecasting model like the
other models mentioned above.
This model has three ionospheric layers, E, Fl, and F2.
Each layer is represented by a Chapman distribution function
Nj(h) = Ne,,.x exp {a[l - z - exp (- z)}j ,
(10.59)
where a refers to the loss mechanism and z is given by

h - hm
z =
and hs is the scale height for the layer. At a
given altitude the total contribution to electron density is
the sum of contributions by all three layers.
For any height the electron density is approximated by
N
NJ(e) =
akWk(0),
E
k =
where a is the weighting factor and W(f) is an empirically

derived set of discrete orthogonal functions for the altitude
interval e. The 95 to 2000 km range is divided into 127
intervals. The widths of the intervals range from 5 km at
the lowest altitudes to 50 km at highest altitudes. The empirically derived function W(f) is in two parts, spherical
harmonic functions for spatial dependence and trigonometric
functions (sine, cosine) for temporal dependence.
With the help of these variables ak and Wk, the entire
data base for the ionosphere is reduced to a limited number
of coefficients. These can be used to construct the electron
density profile for any location in the Northern Hemisphere
valid for a 24-h specification period. The model is still being
developed. The specification accuracy of the model will
depend strongly on the frequency and reliability of the input
data-real-time experimentalobservationsfrom the 40 ground
stations of the northern hemisphere. Also the quality of
specifications interpolated for locations inside the network
will be better than those extrapolated outside the observational network.
10.3.1.6 International Reference Ionosphere-IRI 79.

The IRI 79 [Rawer, 1981] is the latest addition to the continued efforts of ionospheric modeling. The emphasis of IRI
79 is to summarize the experimental data from rockets and
satellites to provide true height profiles of the ionosphere.
The model serves as a standard reference for various purposes such as design of experiments, estimation of environmental and other effects, and testing theories. The model
gives the altitude dependence of four parameters: electron
density, electron and ion temperatures, and the composition
of positive ions. It computes the density for atomic ions
O', H', He' and for molecular ions O+ and NO+.
For the worldwide description of the peak electron density, the model uses foF2 from CCIR [1967]] coefficients,
with modified dip coordinates [Rawer, 1963]. As the foF2s
are from the ground based ionosonde stations, the model
really computes a relative distribution of electron density
with true height, with respect to that of the foF2 peak. For
a true peak height the model uses an empirical relation
[Bilitza et al., 19711 with M(3000)foF2 coefficients from
CCIR [1967]. This empirical relation is based on the incoherent scattering measurements which yield electron density with true height. The model has an alternate procedure
based on the results of Chiu [19751 to replace the foF2 and
M(3000)foF2 coefficients from CCIR. This procedure when
used limits the ability of reproducing the complex ionospheric features available from the CCIR coefficients. The
IRI 79 can also use direct data of the peak electron density
and the peak altitude for computing the profiles.
The schematic for the altitude dependence of the electron
density for the IRI 79 model is shown in Figure 10-41. The
altitude
80 to
altitude range
range from
from 80
to 1000
1000 km
km is
is divided
divided into
into six
six secsections: topside, F2 bottomside, F1, intermediate, valley, and
E/D regions respectively. The topside region is modeled
with the use of 'harmonized Bent' model [Ramakrishnan et
al., 1979]. The bottomside F2 is expressed as the sum of
Epstein Transition Functions [Rawer, 1981]. The F region
is based on the work of Eyfrig [1955] and Ducharme et al.
[1973]. The intermediate region fills the gap between the
valley region and the Fl layer. The rocket measurements
compiled by Maeda [1971] determine the shape and the
depth of the valley region. The foE is from Kouris and
Muggleton [1973a,b]. The model also takes into account
the contribution from the D layer. The model does not
account for the highly variable Es layer.
The IRI 79 is the only numerical model with information
on additional parameters such as the electron and ion temperatures and the composition of positive ions. The composition is determined with the assumption that the plasma
is electrically neutral above 84 km. The model also computes the distribution of cluster ions in the altitude range
80 to 90 km.
The inputs for the program are location (latitude and
longitude), sunspot number and time. The optional inputs
are the peak altitude and peak electron density. The output
consists of 11parameters: absolute electron density, relative
10-43
CHAPTER 10
The programs are in FORTRAN-4 and ALGOL-60 computer codes. These programs are available from the World
Data Center, Boulder, Colorado.
Topside ( 1 )
10.3.2 The Theoretical Models

HMF2
The theoretical models for the ionosphere are based on

F 2 121
F 1 (3o1
HST
Intermed. (
--
the physical processes responsible for the observed ionospheric phenomena. The processes responsible for the ionosphere are production, maintenance, and decay of the ionsphere. As the approach deals directly with the physical
u,
processes, and not with the observed phenomena, the emerg-
IHST'"ting
-/ Xntermed.
(4)
model is called a physical model.

Four models are summarized in Table 10-6 to show
several variations in the same processes considered by dif-
/2[HEF
~B
-E - Valley 5 1 --
I HBR
HABR
t
HME/
_.
_HgOX
/D
ferent workers. Strobel and McElroy [1970] considered only

the F2 region (200 to 700 km), whereas others took into
account the altitude range from 120 to 1200 km. Nisbet
[19711 constructed the first computer-based simple physical
model MK-I for the ionosphere. He considered only three
neutral constituents N2 , 02 and 0, whereas Stubbe [1970]
HA
NME
log N
and Oran and Young [1977] also considered the minor constituents He and H. For the dissociation and ionization of
the neutral species, the incident solar EUV radiation in the
NMF2
range 30 to 1912 A is used, along with the wavelength
Figure10-41. TheIRI79 modelprofile.FordetailsrefertoRawer[1981].
dependent absorption and ionization cross sections for the

neutral species. Nisbet considered three basic predominant
ionic species: O+ , NO + , and O2+ . Oran and Young [1977]
took into account the additional ionic species H2+ , Ne + ,
N + and H+ . One has to consider the chemical reactions
that produce ions by charge exchange processes. Nisbet
[1971] used 5 reactions whereas Oran and Young [1977]
used 24 chemical reactions [see Strobel and McElroy, 1970].
For maintenance of the ionosphere, the processes of diffusion and photoionization are assumed. The processes of
dissociative recombination and radiative recombination are
electron density, neutral temperature, electron temperature,

ion temperature, ratio of electron to ion temperature, percent
concentrations of O+ (and N+), H , He+ , 02 + and NO+
ions. The accuracy (ao)of predictions is as below:
F region
E region
Peak
Height Density
+ 15% 30%
5% 10%
Temperature
+ 30%
+ 10%
Table 10-6. Variations in the physical processes used in the theoretical models
Nisbet
[1971]
Processes
Stubbe
[1970]
In the Altitude Region (km)

Neutral Constituents for
Ionization
Chemical Reactions
(Charge-Exchange)
Ionized Constituents
120-1250
N2 , 02, 0
O +, NO ', 02 +
O +, NO +0, 02,
Hi
Neutral Winds
Horizontal Winds
Electric Fields
Magnetic Fields
Yes
Yes
Additional Features
10-44
5 Reactions
120-1500
0, He,
H
10 Reactions
N 2 , 02,
Strobel and McElroy

[1970]
Oran and Young

[1977]
200 700
N2, 02, O, He
120-1200
N2 , 02, O, He, H
4 Reactions
24 Reactions
Horizontal Winds
O , NO +, 02,
H+
N + , He + , N2 +
Horizontal Winds
Solar Flare Effects

responsible for the decay of the ionosphere. For his simple
model, Nisbet neglected the transport processes such as
neutral winds, electric fields, and magnetic fields. The procedure is further complicated because coupled simultaneous
equations must be solved for neutral winds, mass transport,
and energy transport. For determining electron density in
the ionosphere, the gas consisting of both ions and electrons,
is considered electrically neutral. Thus, in every elementary
volume, the number of electrons is equal to the number of
ions. All the models reproduce many of the observed features such as the diurnal variation, seasonal variation, and
solar cycle dependence of the midlatitude ionosphere under
quiet conditions. The accuracy of the theoretical models
depends upon the understanding of the physical processes
considered in the models. For accurate predictions from the
theoretical models, precise information on the large number
of variables used in the models is necessary. Also, the
models use several observed average boundary conditions
which could have a large variability dependent on other
geophysical parameters such as solar activity and magnetic
activity. The results from the models are adequate for long
term planningof science and engineeringapplications.Though
these models reproduce main observed average features of
the ionosphere, they are unable to specify the ionosphere
within an accuracy of -20% needed by the systems in
operation. At present, the main input information of solar
EUV radiation needed for the theoretical models is not routinely available for predicting the ionosphere.
The additional limitations of these models are

1. All the models are poor in predicting the high
latitude ionosphere.
2. None of the models take into account the effects
of particle precipitation in the auroral region which
enhance the E(Es) and F layers.
3. The mid-latitude trough which exhibits large horizontal gradients in electron density is not incorporated in these models.
4. These models are good for latitudes+20 to - 60 ,
and are poor predictors for the equatorial region
and the high latitude region.
Nonetheless these models serve two useful functions:
(1) to predict ionospheric parameters, and (2) to determine
physical phenomena and/or to modify existing coefficients
for explaining the deviations between the experimentally
observed value and the predictions from these models.
The computer programs for the ITS-78 [Barghausen et
al., 1969] and the IONCAP program [Lloyd et al., 1978]
are available from the Institute for Telecommunications Sciences, Boulder, Colorado 80303. The computer programs
for the Bent model [Llewellyn and Bent, 1973] are available
from the Atlantic Science Corporation, P.O. Box 3201,
Indialantic, Florida 32903. The computer programs for IRI
79 are available from World Data Center A, Boulder, Colorado 80303.
10.4 HIGH FREQUENCY RADIO

PROPAGATION
10.3.3 Comparison of the Phenomenological
Models, Their Limitations and Ability
In comparing the models one must note that IONCAP
is the modified and more flexible version which replaces
the ITS-78 model. As the ITS-78/IONCAP and the Bent
models serve entirely different purposes, it is essential to
understand the difference in their approaches and final output parameters computed by the models. These are summarized in Table 10-7. The left-hand column in Table 107 lists the parameter under consideration. The next four
columns summarize the features in each of the models, ITS78, IONCAP, the Bent, and the IRI 79 models, respectively.
From the table it is seen that the selection of a model will
depend more upon the information sought under the parameter headings, than on accuracy. The IONCAP model is
basically useful for wave propagation using operating frequencies which would be reflected by the E, Es, Fl, and
F2 layers. On the other hand, the Bent model relies on the
transmission, refraction, and absorption characteristics of
the ionosphere, with the operating frequency much larger
than the foF2 frequency. The IRI 79 model basically provides a distribution of electron density with altitude. All the
models predict quiet ionospheric conditions only. The models
do not hold for disturbed ionospheric conditions.
The high frequency (HF) band of the electromagnetic

spectrum extends from 3 to 30 MHz, corresponding to a
wavelength range of 100 to 10 m. Many services have
frequencies allocated in this band-Local/International
Broadcast, Amateurs, Standard Frequencies, Maritime and
Land Mobile, Point-to-point Communications, Industrial,
Scientific, Medical Diathermy, Aero Fixed, Citizens' Band,
and so on. The band is also used for ionospheric sounding
and over-the-horizon surveillance. Its use in most applications depends on the fact that HF waves are reflected by the
ionosphere.
HF is used for broadcasting because of its greater area
coverage relative to the bands on either side, which are
restricted to either ground wave or line-of-sight propagation.
Its use for communications stems mainly from the fact that
it is often the only means of communication. It is also very
often the simplest and least expensive form of communication.
With the advent of satellite communications, which use
signals of such high frequency that the normal ionosphere
has little effect on them, and improvements to submarine
cables, the proportion of traffic that goes by HF is significantly smaller than it used to be. However, the total use of
HF radio circuits is actually greater now than ever before,
and a substantial research effort is still being devoted to
10-45
CHAPTER 10
Table 10-7.
Intercomparison
of the empirical-computer
based ionospheric
models.
Ionospheric Models
Parameter
ITS-78
D Region
E Region
foE
hmE
YmE
F1 Region
foF1
IONCAP
Bent
IRI 79
Non-deviative and
deviative absorptions only
Same as ITS-78
+ E Layer
exponential
extension down
to 70 km
Not modeled
Modeled
Modeled by Leftin
et al. [1968]
110 km fixed
20 km parabolic
shape
Same as ITS-78
+ exponential
down to 70 km
Leftin [1976]
coefficients
Not modeled
Modeled
Kouris and
Muggleton [1973a,b]
Not modeled
Modeled
Eyfrig 11955]
Ducharme et
al. [1973]
Bi-parabolic
Modeled
Rawer [1981]
Not modeled
Rosich & Jones
coefficients 11973]
hmF 1
YmF I
hmF1/ymF1 = 4
(fixed)
F2 Region
Bottomside
foF2
hmF2
YmF2
Haydon-Lucas
coefficients [1968]
Shimazaki eq [1955]
+ E layer
retardation
Same as ITS-78
Bent coefficients
Kelso [1964]
F2 Region Topside
Not modeled
Not modeled
Up to 1000 km
Modeled
Rawer [1981]
E-F Transition
Region
Not modeled
Modeled
Not modeled
Maeda [1971]
Electron-Density
Profile
Not computed
Available up
to hmF2
Available up
to 3500 km
Available up
to 1000 km
Electron, Ion
Temperatures
Not modeled
Not modeled
Not modeled
Modeled
Ion Composition
Not modeled
Not modeled
Not modeled
Modeled
Total Electron
Content (TEC)
Not computed
Not computed
Computed
Not computed
MUF
For short path

only
Also for long

path (> 10,000 km)
Not modeled
Not modeled
Short-Term Prediction of MUF
Function of Kp
Not modeled
Not modeled
Not modeled
Input Parameters
required
Sunspot number
Sunspot number
Sunspot number
and 10.7 cm
solar flux
Sunspot number
10-46

Table 10-7.
(Continued)
Ionospheric Models
Parameter
Noise Parameters
Galactic
Atmospheric
Manmade
Modeled
Same
Modified
Same
Modeled
Not modeled
Not modeled
Not modeled
Not modeled
System Performance
Modeled for
short path
(<3000 km)
Also has a
long path
option >
10,000 km
Not modeled
Not modeled
Antenna Patterns
Uses ITSAPackage with

10 antenna
options
Modified
ITS-78
package with
17 antenna
options
Not modeled
Not modeled
Sporadic E
Modeled in terms
of occurrence
frequency
Same as ITS-78
Not modeled
Not modeled
Circuit Reliability
Service Probability
Multipath Evaluation
Modeled
Modeled as
ITS-78
Not modeled
Not modeled
MUF 50%
FOT 90%
HPF 10%
improving our knowledge of the ionosphere and HF propagation.

Some of the difficulties associated with using HF for
communications, broadcasting, or surveillance stem from
the ionosphere itself and success in any of these fields demands a good knowledge of the ionosphere and its vagaries.
Irreducible difficulties associated with HF propagation can
usually be traced to characteristics of the ionosphere or of
radio waves propagating through any lightly ionized medium. Thus it is essential for the professional user of HF
to have a good knowledge of both the ionosphere and radio
wave propagation.
Much has already been written about the ionosphere and
radio wave propagation and the reader should look elsewhere for details. See, for example, the books by Davies
[1966, 1969], David and Voge [1969], Rishbeth and Garriot
[1969], Hargreaves [1979], Ratcliffe [1970], Lied [1967],
and Picquenard [1974]. The four volume report "Solar Terrestrial Predictions Proceedings" [Donnelly, 1979, 1980] is
an excellent supplement to these books, providing more
recent reports on the general problem of forecasting the
solar-terrestrial environment. The reports of Study Group 6
of the Consultative Committee for InternationalRadio (CCIR)
are also a very useful source of information, and are particularly valuablebecause they are regularlyupdated. AGARD
Lecture Series No. 127 on "Modern HF Communications"

is also a valuable source of information [AGARD, 1983].
It is the intention of this section to provide a broad
overview of HF propagation, its relationship to the ionosphere, its problems, and to indicate those areas of current
interest to users of HF. A basic knowledge of the ionosphere
itself is assumed (see Chapter 9 of the present volume).
Emphasis will be placed on the use of HF for communications. The same concepts and problems also apply to the
use of HF for broadcasting and surveillance (over-the-horizon radar). Section 10.6 covers the effects of over-thehorizon radars in the HF band.
10.4.1
Morphology
of the Ionosphere
An understanding of the morphology of the ionosphere

is an essential prerequisite for its successful use as a communications medium. The basic theory of the ionosphere
and its variations has been outlined in Chapter 9 of this
volume-here we are concerned mainly with how the ionosphere varies, rather than why it does.
There are five main variations of the electron density of
the ionosphere that must be taken into account:
10-47
SOLARMAXIMUM
CHAPTER 10
14
14
SUMMER 1958
12 -
oI
12 -
T= 188
O-
06
12
WINTER 1958
F2
18
24O
T= 159
06 8
12
6z
SUNSET
SUNRISE
O
0
06
12
18
HOUR (LT)
24
150E 0
SUNRISE
II
06
SUNSET
12
18
HOUR (LT)
24
SOLAR MAXIMUM
10
10
SUMMER 1964
8- ~~~~~N
lo "
4
WINTER 1964
I~~8
T=lO
'1OE/
~
06
10
cr
SUNRISE
Figure 10-42.
10-48
II[z
T=O
06
12
18
24
JSUNSET
SUNRISEI
SUNSET,
06
12
18
24
Diurnal variations of the critical frequencies of the E, F1, and F2 layers for solar maximum (1958) and solar minimum (1964) and for
summer (January) and winter (June), at a typical midlatitude station (Canberra). The parameter T is an ionospheric index related to the
level of solar activity.

90.
90
60 E
180
120E
600 W
120W
J60
CD
0-
310090080070060*
a.
60070
300 40050
80
200
500400 30
~~~~~~~~~~~30,
-1
'~~~~~~~-J
:z ,,I
30
Crr
~~~5
\\\\\
SOLAR
60X/-ZENITH
e::~~~~~~~~~~0
ANGLE
700
::
-0.90SUN
"50
7~~~~~~~~~~~~00
80
901
0O
90
04
08
600 E
120*E
12
1800
16
20
1200 W
'"UW~~~~~~60
0.5
60 0 W
~.
1.
2.0
00
1.0
C,
1.0 1.5 2.0
303
3.0
3.0
C,,I
0 90104
90 o .
O'E
08
12
12O'E
180,
16
~~~~~2
.5
20
120OW
60'W
solar
(botton panel).
angle for the
W maximum
60 *
2.0 2.5The top
30panel shows the
E variation of the solar zenith
0~~~~~~~~~~~~~~
.5 same month.
~10-49
80
20E
W
1.5 60
3.0
O1.
it
3.5~~~~~~~~~~~.4.
o
0
a"30 - 9004
3C3
IzI"'
r-,
08
2~~
60
16
20
~0
2.5
30
90
12
~~~~~~.02.
-~~5
04
81
62
Figure
10-43The
Figure
10-43. The 00
geographical
frequency of the
solar minimum (center panel) and
gclvraino variation of the
h critical
rtclfeunyo
h E layer
ae as
a a function
ucino of local
oa time
iefrJnfor June attslrmnmm(etrpnl
n
J 90
60.5
O
B
60
12
soau
aiuhto
ae) 4 h oppnlsostevraino
h olrznt 16 nl
o h 20aemnh
"r'~~~~~~~~~~~~~~~~~~~04
CHAPTER 10
1. DIURNAL-variation throughout the day, which is
largely due to the variation of the solar zenith angle.
2. SEASONAL-throughout the year.
3. LOCATION-both geographic and geomagnetic.
:
4. SOLAR ACTIVITY-both long term and disturbances
5. HEIGHT-the different layers.
These variations have all been deduced experimentally,
by world-wide observations of the ionosphere over the past
few decades. The reader may refer to Davies 11966, Chapter
31and Hargreaves [ 1979, Chapter 5] for details. The diurnal,
seasonal, solar cycle and height variations of the ionosphere
may all be deduced by routine monitoring of the ionosphere
at one location. Figure 10-42 shows these four variations
for a typical midlatitude station. 1958 was a period of high
solar activity, as indicated by the high values of the ionospheric index, T (see Section 10.4.4). The figure also illustrates the mid-latitude seasonal anomaly, the name given to
the initially unexpected fact that foF2 is higher in the winter
than in the summer, in spite of the larger solar zenith angle.
This anomalyandothersare describedby Hargreaves[ 1979,
Chapter 5].
Once the diurnal, seasonal, solar cycle and height variations of the ionosphere at a given location have been deduced, the next step is to measure and understand the variation with location. This has been achieved through an
international effort of observations and data exchange, and
we now have reliable maps of the world-wide distribution
of the important ionospheric parameters. The accuracy of these mapsover the ocean areas, whereno observationsare
foF. -JUNE, 1954, SOLARMINIMUM

80'
60
Z 40
w 20
/
4.2
-J 20
WO
60
foE for a given zenith angle, X [for example, Hargreaves

1979; Muggleton, 1971]:
4
foE = 3.3 [(I + 0.008R) cos X]l" MHz
so
8
I
00
I
04
I
08
I
12
16
80
60
10-50
20
24
4.5
5.5
/40/
4\0
04 0
z 40.5/
w 20
o
-J 2 0
6.8
6.4
/
/ GROUND
DAY-NIGHT
LINE
_
04
08
I I I I
12
20
16
24
LOCAL TIME
Figure 10-44. The geographical variation of the critical frequency of the

F1 layeras a functionof local time for June at solarminimum and maximum.
10-44), except that the Fl layer tends to disappear in winter.

Hargreaves [1979] gives the following formula for foF1:
foF1 = 4.25 [(1 + 0.0015R) cos
Note that foE also varies linearly with sunspot number, R,

increasing by about 20% over a typical slow cycle for a
zenith angle of zero.
Sporadic E (Es) layers also occur in the height range of
the normal E region. These layers are patchy and only a
few kilometers thick at mid latitudes. They tend to appear
and disappear almost at random (hence the name), but have
well-defined gross seasonal and latitudinal variations. See,
for example, Hargreaves [1979, p.90].
The Fl layer is similar to the E layer (see Figure
/DAY- NIGHT
D
LINE
f, Fi - JUNE, 1958, SOLAR MAXIMUM
available, still remains somewhat limited [Rush et al., 1983]. "

The easiest part of the ionosphere to model on a worldo0 40
wide basis is the E layer. Figure 10-43 shows the variation
60 _/
of foE, the critical frequency for the E layer, for June at
solar minimum and solar maximum. The figure also shows
80
the variation of the solar zenith angle for the month, and it
00
can be seen that the variation of foE follows closely that of
the solar zenith angle. In fact, the variations are so close,
indicating that foE is very largely solar controlled, that it
is possible to use a simple empirical representation to deduce
4.0
3
X]11
MHz .
The variations of foE and foF1 with (cos X)1/4identify both

the E and F layers as well behaved Chapman layers. Other,
more accurate, world-wide representations of foFl have
been given by Rosich and Jones [1973] and Ducharme et
al. [1971].
Moving on to the F2 layer, due to its large height and
electron density the most important layer as far as HF propagation is concerned, we find that the simple situation that
holds for the E and F1 layers does not hold very well for

N 90
70
tric andmagneticfields.Theoreticalmodelingstudiesof the

equatorialionospherehave been performedby Anderson
-=5
.___
50
,)
1
0_
-)
\(
9>
8
_~ ,o ~30
{ 50
'" '~6~-'
~
2_3
5 -
10 </ C)
"-
"
\\
/ /
<much
7
._
70
s90o )
0
[1981], among others. Empirical maps of foF2 and other
30'E 60"E 90e120E

2
500E180
L===step
1
I'
s
150-W 120W 90W 60W 30W 0c
GEOGRAPHIC LONGITUDE
N90
Lw=_
8
7
a
30
10I
8
_
98=
N5\\\\\(_=--
7c
electron densities are always too low to reflect HF waves.
.1
However, the D region is very important from the point of

view of absorption of the energy of an HF wave, especially
2.
30
,o50
C
a
>
70 -
'oo[
7c
,~~)'o~
sgo9,_|-
=-<
5
o
the
\\\4\\> lower end
/at
of the HF band. A review of the D region
\'4,.
30E
ionosphericparameters have been publishedby CCIR [1966].

The morphology of the high latitude ionosphere is even
more complicated than that of the equatorial ionosphere and
remains unknown about it. Probably the most important feature of the high latitude ionosphere is the midlatitude ionosphere trough, which lies equatorwards of the
auroral oval. The trough is a narrow feature that moves in
with geomagnetic activity and thus fails to appear in
monthlymedianmaps of foF2. However,it can havevery
serious effects on HF communic
at high latitudes because of the strong horizontal gradients associated with it.
The morphology of the high latitude ionosphere has been
reviewed, for example, by Hunsucker [1979], Hunsucker
et al. [1979] and CCIR [1981a].
Lastly, we must consider the D region. This region is
of no direct concern for HF radio propagation since the
i~oo~
< 8'been
C
-,|I
60 L2~
'GEOGRAPHIC
ONGIT;;0UDE
and the prediction of its effects on radio propagation has

presented by Thrane [1979]. Synoptic models of the
D region electron density are unreliable because of the complexities of the D region and the difficulties encountered in
measurement of the electron density profiles.

Figure10-45.The geographical
variationof the criticalfrequency
of the
F2 layerforJuneat solarminimum
andmaximum,
for 00
UT.Worldmapssuchas theseare madeforeachhourof
eachmonth(576maps).
the F2 layer. Figure 10-45 shows, for example, how the F2
layer critical frequency, foF2, varies over the earth at 00
UT in June, for low and high solar activities. It can be seen
that the simple structure obtained for the E and Fl layers,
with the contours of foE and foF1 closely following the
contours of the solar zenith angle, no longer applies although
a clear zenith angle dependence can be seen around sunrise
(-90 to -120E). In fact, the departures of foF2 from a
simple R, cos 1/4xdependence are so great that it is necessary
to make world-wide observations to determine the actual
variations of foF2. Detailed studies of foF2 have shown that
as well as depending on R and cos X, foF2 also depends
on other factors such as electric fields, and neutral winds
to name a few, and its large scale morphology is controlled
by the geomagnetic field.
foF2 is also found to have variations with latitude which
are not seen in foE and foF1. For example, Figure 10-45
shows that foF2 exhibits two afternoon peaks (-12 MHz
in the solar maximum portion of the figure) situated on either
side of the equator. This feature is known as the equatorial
anomaly and is due to electrodynamic lifting of the layer at
the equator under the combined influence of horizontal elec-
Absorption of HF waves occurs mainly in the D and

lower E regions of the ionosphere. The free electrons absorb
energy from the incident wave and reradiate it in a continuous process. However, if an energetic electron collides
with a neutral particle before it can reradiate its energy, this
energy will be taken up by the neutral particle as kinetic
energy and will be lost to the HF wave, that is, energy will
be absorbed by the medium.
This type of absorption is known as non-deviative absorption and is roughly proportional to 1/f2 where f is the
wave frequency. Extra absorption, known as deviative absorption, also occurs near the reflection level. The nondeviative term usually dominates for oblique propagation.
If the operational frequency, f, becomes too low, the absorption will increase to the point where the signal disappears below the level of the noise at the receiver site and
becomes unusable. This frequency is known as the lowest
usable frequency, or LUF. A good treatment of absorption
is given by Davies [1969, Chapter 6] while the variations
of absorption are discussed by Davies [1966, Chapter 3].
10.4.2 Simple Ray Propagation

Many of the operational aspects of HF propagation may
be studied using simple ray concepts. Figure 10-46 illustrates the basic geometry of a one-hop HF circuit. Note that
10-51
CHAPTER
10
.. .:.'..: .. ::' :
';)~.-
2'
the diagram ignores the ground wave which is usable for

ranges up to about 50-100 km, depending on the frequency,
antenna, ground conductivity, etc. The reflected ray is conrefracted as it passes through the ionosphere and
if sufficient refraction occurs the ray will be bent down
sufficiently to reach the receiver. Figure 10-46 also shows
one of the most basic formulasof ray propagation:
'1.: .::::...::::-... .
IONOSPHERE:
.tinuously
''-
'. '-IONO.P..
'-~
.':"'::*
.OFINCIENcE
ELETION
ANGLE
A7..... __.._./~
MUF = fc x obliquity factor = fc x k x sec 4,
MUF= fck sec

that is, the maximum usable frequency (MUF) is equal to the
product of the critical frequency, fc, of the reflecting layer
and an obliquity factor related to the geometry of the circuit. For a flat earth/ionosphere approximation, this factor is
Figure 10-46. Simple geometry of an HF oblique circuit, illustrating the

essential elements of the circuit. The formula for the max-
imum usablefrequency(MUF)is all that is requiredin
sec 4, where 0 is the angle of incidence. For a curved earth

and ionosphere, the factor k is introduced to allow for the
different geometry. This factor is typically of the order 1.1.
In practice, the obliquity factor for a given circuit relying
on reflection from the F2 layer for example, is obtained from
manycalculations
for HFpropagation.
E- MODE
F-MODE
I- hop 2-hop I-hop
9.0
20.5 28.1 48.4
4.4
2.6
1.9
1.3
CIRCUIT:FORA 1OOOkm
12-hop
elevation angle
K* sec ()
E- MODES
90
90
80 -
8.0
8_0-
70
7.0
70 -
aW 60
,,\
-J
F-MODES
9.0
8.0
--
7.0
ELEVATIONANGLE
6.0
\2-HOP
5.
50
5.0
W 60o
L
6.0
I-HOP
ELEVATION
ANGLE 2-HOP
50
\
(9
9.0
.x
4
4.0
o
v-Y
w"J~
,30V
20C
>
- 3,0
seca~~~
~--_
101
0
I 1 1111111
II1111111
I 1
No
/\2.0
~~~~~K*
PATH LENGTH(km)
%
o
o0
0.0
40-
4.0
1~i
>
I-HOP
I~I-HOP
I-HOP
, 30-I
ISA
20C
)~K (PHIO~
e
2.0
sec (PHI)
1.0
01
3.0
\,
1
11
1 1 1111111
0.0
U)
PATHLENGTH (km)
Figure 10-47. Plots showing the elevation angle and obliquity factor (k sec 0) for propagation via the E and F layers. For a given circuit length and
number of hops, these plots show the required elevation angle (necessary for selecting an appropriate antenna) and obliquity factor for the
two layers (useful, among other things, for consideration of possible E layer screening).
10-52
.......
LOWER.
..L . O
.. IONOSPHERE
o o.MHEIGHT
.-
OF MAX ION
OENSITY
. . . . .
*-
Figure 10-49. Sample ray path for a fixed frequency but varying elevation
angle. If the operating frequency
is above the local critical
frequency, high elevation rays will penetrate and there will

be a "skip zone" around the transmitter which cannot be
reached by an ionospherically
propagated ray.
Figure 10-48. Sample ray paths for a fixed distance but different fre-
quencies.As the frequencyincreases,the ray mustpenetrate further into the ionosphere.If the frequencyis too
The area surrounding the transmitter, which is defined by

k sec 4E
f/foF2, is known as the skip zone for that fre-
high, that is, above the MUF for the circuit, the ray pen-
etrates.
values of the obliquity factor for a 3000 km hop, M(3000)F2,

which is scaled routinely from vertical incidence ionograms
[Piggott and Rawer, 1972]. The obliquity factor for a distance D is related in an empirical fashion to M(3000)F2.
Figure 10-47 shows how the value of k sec 0 varies with
circuit length for one and two hop E and F layer modes.
Typical values for the E and F one-hop modes are -5 and
-3. Figure 10-47 also shows the elevation angles corresponding to the different propagation modes.
Figure 10-48 illustrates ray propagation for different
frequencies on a fixed circuit. As the frequency increases,
quency (and location and time). Signals at the frequency f

cannot penetrate into the skip zone, although the ground
wave would propagate out to about 50-80 km. This phenomenon can also be used to advantage by ensuring that an
unwanted receiver lies in the skip zone of the transmitter.
When the value of foF2 above the transmitter exceeds the
operating frequency there is, of course, no skip zone.
HF-communication via ground wave is important in
many areas, particularly over sea and flat land with high conductivities, where reliable circuits may be established up
to distances of several hundred kilometers. The conductivity
of the surface is strongly frequency dependent with rapid
attenuation at the higher frequencies. In the past CCIR has
published a set of curves of ground wave field strength
the ray must penetrate further into the ionosphere before it

is refracted to the horizontal and thence back to the ground.
The highest frequency that can be reflected back to the
ground is the MUF for that circuit. Note that the ray corresponding to the MUF does not reach the altitude of the
peak density of the layer, hmax It is only for vertical incidence that the ray actually reaches hmax.
versus distance. CCIR is in the course of implementing a

computer program to estimate ground wave field strengths.
Ground wave propagation may be quite complex, particularly over rough terrain and over mixed land-sea paths.
There is a need for better charts of ground conductivity, and
in some cases terrain modeling may be useful and important. Large topographical features such as mountain ranges
Figure 10-49 illustrates the concept of the skip zone,

which is a zone into which an ionospherically reflected
signal cannot propagate. The figure illustrates the effect of
different elevation angles for a fixed frequency. As the elevation angle increases, corresponding to a shorter circuit
length, the ray must penetrate deeper into the ionosphere in
order to be reflected. However, as the elevation angle increases the obliquity factor k sec 0 decreases. While the
product foF2 .k sec 0 remains greater than the operating
frequency f, the signal will be reflected. When f exceeds
foF2 k sec 0, the signal will penetrate the ionosphere.
and glaciers may cause reflections and strong attenuation,

and vegetation, soil humidity and snow cover also influence the propagation characteristics.
We have seen that the MUF for a given circuit is set by
the density of the ionosphere at the reflection point and the
geometry of the circuit. For a multi-hop circuit, the MUF
is set by the lowest of the MUFs for the individual hops.
The lowest usable frequency (LUF), on the other hand, is
set by absorption of the signal by the ionosphere and by the
generally poor performance (low gain) of most HF antennas
at low frequencies.
10-53
CHAPTER 10
CHOICEOF MODE
10.4.3 Requirements for

Successful Communications
::::::::.:.:...............................................................
Under
normal
conditions,
there
are threecomfactors
that must
be operating
considered
to achieve
successful
are
These
munications.
1. Choosing a suitable operating frequency.
2. Choosing a suitable antenna system.
3. Ensuring that the wanted signal is at a level above
that of the local radio noise at the operating frequency.
The choice of a suitable operating frequency is the subject of main concern in the present context since this is
where a knowledge of ionospheric physics and radio wave
m..,.....................\
.............................................
.. .......
................
. . ............
/
Tx
Rx
propagation is required. This choice will be discussed in

Section 10.4.4. The choice of a suitable antenna system
Figure 10-50. Sketchillustratingthe fact that the antennapatternshould

match the required propagation mode for a given circuit.
will require matching the antenna pattern to the propagation

angles of the HF signals-these can be deduced using the
ionospheric models developed for choosing a suitable operating frequency. However in many cases practical considerations will intervene and a far from ideal antenna will
be used. The ideal antennas tend to be very expensive and
mobile operators especially will often be forced to use a
random wire hung over a tree. As a general rule of thumb,
horizontal antennas are required for short circuits and vertical antennas are required for long circuits. The ubiquitous
whip antenna is absolutely the worst (but probably the most
common!) choice for short sky-wave circuits since its antenna pattern has a null in the vertical direction. Antenna
patterns for the common antennas can be obtained from
Lloyd et al. [1978]. Table 10-8 gives a brief summary of
the situation. Note that the radiation pattern of an antenna

is a function of frequency so that an antenna appropriate
for a low HF frequency may have a very poor performance
at a higher frequency. Selection of the correct antenna will
not only ensure that the bulk of the transmitted power goes
at the required elevation angle, but can also be used to select
a particular propagation mode and thus avoid multipath interference. Multipath interference arises when the transmitted signal arrives at the receiver over two or more separate propagation paths with different time delays.
Figure 10-50 illustrates propagation via 1F and 2F propagation modes, with an antenna pattern whose lobe favors
the higher angle 2F mode and almost completely prohibits
Table 10-8. Suitablesimpleantennasfor use on paths from 0 to more than 3000km.
Path Length (km)
Suitable Simple Antenna
0-200
600-90
Horizontal dipole: broadside

is required azimuth. 0.25
wavelength (X) above ground.
200-500
400-70
Horizontal dipole: broadside

to required azimuth. 0.3 X
above ground.
250-50
and 10-20
0.25 X vertical monopole, or

horizontal dipole, broadside to
required azimuth 0.5 X above
ground.
1000-2000
10-30
and low angles
Vertical monopole up to 0.3 X

long with ground screen.
2000-3000
5-15
and 20-30

500-1000
> 3000
10-54
Required Radiation
(Elevation) Angles
low angles


CHOICEOF ANTENNA
:-:--:-:--:-:--:-:--:-:: :-:-:::::::::::::::::::::::
.-'
:-:-:':-:-:-:- i:- :-:-::::::::::::
i:-!i:::.::::
::::
-i l.. ~.~.~
: :::: :............_............
' -
If an adequate signal-to-noise ratio cannot be obtained

by lowering the noise level, it is necessary to increase the
signal level. Some increase in signal level can be achieved
by choosing an antenna that has more gain at the given
:operating frequency and elevation angle. This is one approved solution. An alternate solution is to increase the
transmitter power.
10.4.4 Predictionsfor HF Communications

Tx
300
km
IOOO
Figure10-51.Sketchillustrating
thefactthattheantenna
shouldbechosen
tomatchthecircuitlength.Theantennapatternillustrated
is notappropriate
for shortcircuits(R - 300km).
propagation via the IF mode. The impact of the choice of
an antenna for a given circuit is illustrated in Figure 10-51
for hop lengths of 300 and 1000 km. In this case, the antenna
heavily favors the longer circuit and is quite inadequate for
short-haul circuits.
Given a frequency that the ionosphere will support and
an antenna which emits sufficient power in the direction
taken by the signal that arrives at the receiving site, the
third thing to ensure is that the signal strength is above the
strength of the local radio noise. This noise can be natural
or manmade. Below about 20 MHz, natural noise is caused
by either distant thunderstorms, which cause a general increase in background noise level, or local thunderstorms
which are usually much more obvious causes of poor signalto-noise ratio. Galactic cosmic noise becomes the dominating natural noise above -20 MHz when it penetrates the
ionospherefrom above at frequenciesabove foF2. The worldwide distribution and characteristics of atmospheric radio
noise can be obtained from the CCIR Report 332 [CCIR,
19631. Manmade noise includes such things as industrial
noise due to welders, diathermy machines, car ignition,
power lines and so on. Interference from other communicators using the same operating frequency can also be regarded as noise.
There are many techniques available to ensure an adequate signal-to-noise ratio. The more environmentally acceptable method is to aim at lowering the noise level. This
can be achieved by choosing as a receiver site some location
that is remote from the major sources of manmade noise,
which usually entails being away from major cities. Some
noise rejection is also possible using horizontally polarized
antennas which de-select the local noise that tends to be
vertically polarized. Careful attention to the azimuth of the
main lobe of the receiving antenna can also result in the
beam not being aimed at a nearby source of noise. In heavily
urbanized areas (Europe, East and West Coast of the United
States), the noise level is often set by the large number of
other communicators.
The first step in predicting HF communication conditions is to set up an appropriate model of the ionosphere.
To have practical application, this model must include all
five main variations of the electron density distribution of
the ionosphere (altitude, location, diurnal, seasonal, solar
cycle), must include some measure of the remaining variability of the ionosphere after these main variations have
been accounted for, must exist as a reasonably efficient and
fast computer code, and must possess some method for
projection forward in time. Ideally, it should also be capable
of a modelled response to short time scale events such as
shortwave fadeouts and ionospheric storms. (These are discussed in Section 10.4.6.)
Most of the ionospheric models used for communication
predictions are empirical models based on world-wide observations of the ionosphere over the past four decades.
Observations of the main parameters of the ionosphere, foE,
foF1, foF2, M(3000)F2, and Es have been used to produce
worldwide contour maps of monthly median values of these
parameters for each hour of each month and for low and
high levels of solar activity. An example of such maps is
given in Figure 10-45. To calculate the maximum usable
frequency (MUF) for a given circuit, hour and month, the
values of foF2 and M(3000)F2, for example, are determined
by interpolation in the appropriate world map for the expected point(s) of reflection of the signal. For a level of
solar activity other than the low or high levels for which
the maps are drawn, values of the ionospheric parameters
are obtained by interpolating in each of the low and high
activity level maps and then interpolating again in these
results to find the values appropriate to the given level of
activity.
In practice, the only prediction actually performed in
making predictions of suitable HF operating frequencies is
that of the general level of solar activity that can be expected
to pertain at the epoch for which the predictions have been
made. This prediction is usually made in the form of some
ionospheric or solar index, which is a single parameter describing the gross behavior of the ionosphere, and which is
used to drive the computer program. The most common
index used for prediction programs such as the U.S. programs IONCAP and ITS-78, and by the CCIR, is the average
sunspot number, R12. However, Wilkinson [1982] has shown
10-55
CHAPTER 10
A PLOTFORBRtSBANE
JUNEOOUT
200
______________________________
/r57
48
150
55
47
/59
a:
/79
49
69/70
X 60
w
Z
m
W
/67
46
71/
o
56
50
61
73
77/ 5
74years
6
43
0
53
54d756
6f/1
6Dudeney
10
12
14
FREQUENCY(MHz)
Figure 10-52. The variation of the monthly median value of foF2 for
June, 00 UT, at Brisbane, Australia, as a function of the
an appropriate signal-to-noise level at a receiver site is a

much more complicated procedure than the calculation of
the MUF. The required transmitter power is of no small
matter since the cost of the transmitter increases dramatically
with the power. The noise level may be either measured at
the site or estimated from empirical data bases such as
provided by CCIR. See for example Davies [1966, Chapter
7] and CCIR [1963].
There are two other main concerns here, the antenna
and the transmitter, but let us assume that an appropriate
antenna with a known gain has already been selected. To
calculate the required transmitter power, we need to calculate the losses. These include deviative and non-deviative
absorption losses, basic free-space transmission loss, multihop ground reflection loss, polarization coupling loss, sporadic E obscuration loss and horizon focus gain (a negative
loss). The largest of them is the free-space loss, and this
increases with the length of the propagation path.
Because the circuit losses depend on the propagation
path, all possible paths must be considered and compared.
Reliable determination of the actual propagation paths requires a reliable model of how the electron density varies
with altitude up to the peak of the F2 layer. Suitable models
of this N(h) variation have been developed by Bradley and
[1973], among others. As well as being an accurate
representation of the real N(h) distribution, such a model
must also be computationally simple because it is used many
times in the determination of the virtual reflection heights,
h', and thence the possible propagation modes. The International Reference Ionosphere [Lincoln and Conkright, 1981]
ionosphericindex,T. Eachdatapoint representsthe value

for June for years 1943-1980.
may also be used to deducean N(h) profile,but this model

has not beendesignedforthe speedthat is requiredin routine
that any index based on the ionosphere itself, such as IF2

[Minnis and Bazzard, 1960] is usually preferable to a purely
solar index such as R12. Figure 10-52 illustrates the relationship between the ionospheric index, T [Turner, 1968;
Turner and Wilkinson, 1979] and the monthly median values
of foF2 at Brisbane, Australia for 00 UT in June (10 LT).
Each data point represents data for June in the years 1943
to 1980. The low dispersion of the data points about the
regression line indicates the usefulness of the T index for
describing ionospheric conditions. The selection of the correct index for some future epoch will become more uncertain
as the lead time increases because of our general inability
to predict the detailed behavior of the sun.
Predictions of the main critical frequencies and heights
of the ionosphere, together with some simple geometry (Section 10.4.2), are adequate for the calculation of the appropriate frequencies to use for communication on a given
circuit. It is also possible to determine elevation angles of
the possible propagation modes (and thence choose a good
antenna) and to determine the conditions under which some
adverse propagation conditions will exist.
The calculation of the transmitter power required to yield
field strength calculations.

The ionospheric models currently available for use in
studies of HF propagation have been reviewed by Dandekar [1982] and Goodman [1982] (see also Section 10.3).
Full descriptions of the techniques used to calculate field
strengths at a receiving site are given by Davies [1966,
Chapter 7], Lloyd et al. [1978] and CCIR 11970, 1978].
A straight forward approach to obtaining MUF's for various modes and their respective path losses which also
permits the inclusion of realistic antenna patterns for
transmit and receive sites is available in the IONCAP
program [Lloyd et al., 1978]. The model computations
are based on a three layer (F2, Fl, E) representation of
the electron density profile with ITS78 coefficients being
used to determine the (Rz-dependent) foF2 value. The ionospheric parameters are evaluated at the reflection points
and two dimensional ray tracing is applied, assuming local horizontal stratification. For the most reliable results,
simplistic approaches to propagation mode determination
must be replaced by ray tracing techniques such as described by Davies [1969, Chapter 7]. These techniques
are essential when large horizontal gradients exist in the
ionosphere, but are rarely used for routine calculations
because they are very time consuming.
10-56

12
/l
'x,,
8_
'_-X
,-'
_ -,
N
I \
E
\,
>6 -
-.
.. '\-..
'
..
/-|/
a circuit by choosing a frequency that is supported by only

one propagation mode (for example, use a frequency above
the 2F MUF but below the 1F MUF) or by choosing an
antenna that heavily favors one propagation mode. Diversity
techniques, in which the same information is sent in two or
more ways and recombined at the receiver, can also be used
to overcome the effects of fading. Example are space (spaced
antennas), frequency, angle of arrival, and polarization diversity.
z x
0o
.
x
4 -
x
x
....
UPPER
DECILE
MEDIAN
_-.-.
LOWERDECILE
XXXXXX MEASUREDVALUE
FOR A MAGNETICALLY
DISTURBED
PERIOD
00
00
04
08
Il
12
16
2
20
I
24
I50'EMT
Figure 10-53. Plot of the observed values of foF2 a midlatitude station,

December 1980, as a function of local time. At each hour
there are 31 data points. 80% of observations lie within
the dashed upper and lower decile curves.
10.4.5 Problems with HF Communications

The use of the ionosphere for communication, broadcasting, and surveillance is fraught with difficulties, some
of which have already been mentioned. Predictions of the
MUF, LUF, and field strengths rely on median models of
the ionosphere and can therefore specify only mean propagation conditions. A knowledge of the spread of values
about the median is required for successful communication
for more than 50% of the time, and such statistics are in
fact usually provided by the better prediction programs. For
quiet, undisturbed conditions, this spread about the median
is typically 15%-20%. See for example Figure 10-53.
Even when conditions are undisturbed and the communicator has chosen an appropriate operating frequency,
antenna, and transmitter power, propagation problems are
still encountered. Some of these are associated with events
on the sun and will be discussed in Section 10.4.6. Here
we wish to consider some problems that can occur even
when the ionosphere is being cooperative.
10.4.5.1 Multipath Propagation. We have already seen
in Section 10.4.3 that propagation will normally occur by
several paths, for example, the 1F and 2F modes (see Figure
10-50). The received signal will be the vector sum of all
waves arriving at the receiver. If the different signal paths
change with time in different ways, deep and rapid fading
may occur, sometimes causing the signal level to drop below
the local noise level. Different modulation techniques are
affected in different ways by this multipath fading.
Multipath propagationmay sometimesbe eliminated from
10.4.5.2 SporadicE. Sporadic Eisatwo-facedphenomenon having both advantages and disadvantages. There is
no doubt that when a strong sporadic E layer is present, it
presents an ideal propagation mode suitable even for medium speed digital data transfer. However, a dense Es layer
can screen the F layer, preventing signals from ever reaching
that layer and causing Es-F combination propagation modes
that may have none of the desired characteristics-the signals may miss the target or may arrive at an angle not
specified in the original receiving antenna design. Many
prediction schemes for HF propagation include statistical
occurrence rates for Es propagation modes, but the Es layer
remains essentially unpredictable.
10.4.5.3 Problems in High Latitudes. After major solar
flares with particular characteristics, high energy protons
penetrate into the lower ionosphere at the poles and causes
wide-spread and long-lasting disruptions to HF communications. The polar cap absorption event is discussed briefly
in Section 10.4.6. The penetration of electrons with energies
in excess of 10 keV into the D region leads to increased
absorption in the auroral zone. This auroral absorption may
have severe consequences for circuits crossing the auroral
zone, but strong absorption is usually limited geographically
to patches a few hundred kilometers in extent and the duration is typically half an hour to a few hours.
An important feature of the undisturbed high latitude
ionosphere is the great variability in space and time. This
variability severely limits the usefulness of a median model
of the ionosphere. The F region trough, which marks a
transition between the midlatitude and high-latitude ionospheres, can have severe and detrimental effects on signals
propagating through it. The sharp gradients at the walls can
cause reflections and result in off-great-circle propagation.
Propagation in the auroral region may introduce rapid
and severe fading of HF signals. Diversity techniques or
some sort of real time channel evaluation technique (see
Section 10.4.9) therefore become almost mandatory.
The polar cap ionosphere in darkness, of importance to
long distance HF communication between higher latitude
stations such as in Canada and northern Europe, is essentially unpredictable for HF purposes. Enhanced F ionization
regions resulting from particle precipitation are randomly
distributed within a low density background ionization. Especially during slightly disturbed conditions polar cap plasma
convection moves high density particles anti-sunward at
10-57
CHAPTER 10
high velocities (Buchau et al, 1983]. These changes will
result in rapid MUF and mode variations and will in general
lead to poor channel performance. MUF variations of typically several MHz from hour to hour and day to day were
observed on an arctic link [Petrie and Warren, 1968]. In
summertime the Fl mode tends to dominate the mid-morning MUF, and during winter/solar minimum conditions Es
modes over transpolar circuits (for example, Andoya, Norway to College, Alaska) have a greater than 50% occurrence
[Hunsucker and Bates, 1969].
10.4.5.4 Problems in Low Latitudes. HF communication problems at low latitudes due to steep spatial and temporal gradients have been discussed by Lakshmi et al. [1980].
The very steep gradients in foF2 during sunrise hours give
rise to several difficulties [Lakshmi et al., 1980]:
1. HF link operators are expected to get their frequencies
cleared from the appropriate governmental authority well in
advance and it is usual practice to fix one frequency for the
daytime and another for the nighttime. The use of the night
frequency during sunrise will require much more power than
is normally permitted while the frequency allocated for the
daytime will be higher than the MUF during the transition
period.
2. Point-to-point links normally use inexpensive tuned
directional antennas, and frequent change of operational
frequency is deleterious from the point of view of antenna
efficiency.
3. In the case of long distance circuits in the east-west
direction involving multi-hop F region propagation, the
problem of the sunrise period will extend to a large number
of hours, because the different F region reflection points
will fall in the sunrise transition location at different periods.
The steep gradients associated with the equatorial anomaly cause problems with north-south circuits. For example,
if we consider the anomaly peak in the northern hemisphere
to be at 15N geomagnetic latitude, if a north-south
HF circuit is operating such that the reflection point is on
either of the sides of the peak, and if the frequency of the
link is very close to the MUF, a peculiar situation arises.
If the point of reflection is equatorward of this anomaly
peak, the radio waves incident on the ionosphere for the
northern circuit will continuously encounter increasing levels of electron density due both to the vertical gradient as
the radiowave penetrates higher into the ionosphere and the
horizontal gradient as the wave progresses in the direction
of increasing electron density. On the other hand for the
same link in the return direction, the horizontal gradient is
reversed. Thus the real MUF values for the two opposite
directions in the same circuit can vary by a large margin
depending on the angle of incidence and on the magnitude
of the horizontal gradient. In fact, rather frequently, especially when the operating frequency is close to the MUF
(calculated ignoring horizontal gradients), only one way
communication would be possible. This has been one of the
10-58
unusual complaints in the Indian Subcontinent [Lakshmi

et al., 1980].
10.4.5.5 Spread F and Irregularities. Small scale irregularities of ionization seem to exist at every level of the
ionosphere superimposed on the background of ionization
discussed in previous sections. They affect the propagation
of radio waves and their characteristics may, therefore, be
determined by radio techniques (see for example, Rishbeth
and Garriott, [1969], Section 6.4). When seen on ionograms, the presence of irregularities is described as "spread
F", the spreading being either in the range or frequency
domain. The corresponding descriptions are "range spreading" and "frequency spreading", and parameters describing
this spreading are scaled routinely from ionograms [Piggott
and Rawer, 1972]. Many morphological studies of spread
F, showing its diurnal, seasonal, solar cycle and latitudinal
variations, have been published (see for example Herman,
[1966] or Singleton, [1980], and references cited therein).
An earlier review of spread F and some of the effects on
radio propagation has been given by Newman [1966]. Most
morphological studies of spread F have been concerned with
frequency spreading. The characteristics of range spreading
are, however, quite different and only a few studies of this
type of variation have been made (see for example Cole and
McNamara, [1974]). At equatorial regions, range spreading
is often associated with the large scale electron density depletions known as bubbles or plumes [Basu et al., 1980c;
Tsunoda, 1980; and Tsunoda et al., 1982].
The effects of irregularities on radio propagation are
most important on paths that cross the equator when propagation actually relies on the presence of the irregularities
[Nielson and Crochet, 1974]. Most recent efforts on describing the effects of irregularities on communications
have concentrated on the effects of irregularities on transionospheric propagation, where the problem of scintillation
is encountered. This subject is covered in Section 10.7, this
Chapter.
10.4.6 Disturbancesto Normal

Communications
We have previously seen that HF communications are
subject to sudden and often large disturbances due to events
that originate on the sun. Solar flares can cause immediate
and complete absorption of HF radio waves by greatly enhancing the absorption suffered as the waves pass through
the D region. Theseevents, called short-wavefadeouts(SWF),
are due to increased ultraviolet and x-ray fluxes, and can
therefore occur on only the sunlit side of the earth. Major
flares can also eject a stream of protons which can penetrate
the ionosphere near the magnetic poles and give rise to
complete blackout of HF communications in polar regions.
Such events are called polar cap absorption events (PCAs)
and can last for days depending on the size of the flare and

how well it is connected magnetically to the earth. The
SWFs, on the other hand, usually last from a few minutes
to a few hours at the greatest and are most severe at low
latitudes (see Section 10.5).
Many large flares also eject a cloud of plasma, which
if geometrical and interplanetary magnetic field conditions
are favorable, can intersect the earth's magnetosphere and
cause both geomagnetic and ionospheric storms. The geomagnetic storm, in which the earth's magnetic field is usually depressed below its normal quiet day value, is a result
of a strong enhancement of the ring current and of no intrinsic interest to the HF communicator, but is a useful and
readily available indicator of an accompanying disturbance
to the ionosphere and thus to communicators. The ionospheric storms are of concern, however, especially when
the effect of the storm is to decrease the MUF on a circuit
to well below the predicted levels.
The physics of ionospheric storms is not yet completely
understood, but it is well established that both electric fields
and thermospheric winds (see Chapter 17) play a role. See
for example Hargreaves [1979, Chapter 11].
The effect of a storm at a given location can be either
to increase the value of foF2 (a so-called positive phase or
enhancement) or to decrease the value of foF2 (a negative
phase or depression), or to do both, a long depression following short enhancement. What actually happens depends
on such things as local time of onset of the storm, station
latitude, and season. During winter, most storm effects tend
to be enhancements, whereas in summer and equinox
depressions often follow short-lived enhancements. It is the
depression that causes the major communications problems-the enhancements often go unnoticed by communicators. Major depressions last typically for a day and can
decrease the MUF by a factor of two. Storm effects are
much more marked at higher latitudes (see for example
Rishbeth and Garriott, [1969], Chapter 8).
The lower regions of the ionosphere are not usually
affected during ionospheric storms and the lower frequency
limit for HF communication remains unchanged.
Ionospheric storms due to solar flares are a high solar
activity phenomenon. However, ionospheric storms also occur away from the peak of the solar cycle. These are attributed to enhanced solar wind streams emanating from
magnetically open features in the corona known as coronal
holes. As the stream sweeps over the earth, the electric
currents flowing in the magnetosphere and ionosphere are
modified, yielding both geomagnetic and ionospheric storms.
These storms are called recurrent storms because they tend
to recur every 27 days (solar rotation period as seen from
the earth) as the solar wind stream passes over the earth
again. The effects on the ionosphere are usually less marked
than those of a flare-induced storm but can last longer (a
few days) because of the time taken for the stream to pass
over the earth.
The scientific community keeps a 24-h watch on the sun
through an internationaleffortcoordinatedthroughthe IUWDS

(International Ursigram and World Days Service) and many
national agencies issue forecasts of solar and geophysical
conditions. This effort has been described, for example, by
several authors in Volume I of Donnelly [1979], including
Heckman [1979]. CCIR Report 727 [CCIR, 1981c] gives a
review of the subject and refers to many of the latest available papers.
The USAF Air Weather Service (AWS), through its
operational centers of the Air Force Global Weather Central
(AFGWC), provides space environmental support to the
entire Department of Defense. The overall driving requirement is to minimize system effects caused by impulsive
solar/geophysical activity and ionospheric variations. The
techniques for geophysicalforecasting used at AFGWC have
been described by Thompson and Secan [1979]. A major
advance in observational equipment during the last decade
was the deployment of the new solar observing network
(SOON) and radio solar telescope network (RSTN). With
the data from SOON, RSTN and x-ray data from satellites,
AFGWC can provide a real-time comprehensive analysis of
a flare and its effects on the space environment.
The action taken by an HF user to overcome the effects
of these disturbances depends on the nature of the disturbance. During a SWF, a move to higher operating frequencies
is appropriate since the absorption decreases with increasing
frequency. During the negative phase of an ionospheric
storm, lower frequencies must be used. The effect of a PCA,
on the other hand, is normally so severe that it becomes
necessary to reroute the traffic around the disturbed area.
This usually requires avoiding the whole high latitude ionosphere.
10.4.7 UnusualPropagationModes
The usual monthly-median HF predictions normally assume simple propagation modes such as the IF and 2F
modes and in general these predictions are quite successful.
Some prediction systems also include Es and such propagation modes as lEs and Es-F. However, it is found in
practice that other unusual propagation modes can also exist.
In general these unusual modes have one feature in common-they rely for their support on some particular feature
of the ionosphere that is restricted in latitude. We shall
consider here a few examples of such modes.
Possibly the most useful propagation mode not normally
predicted is the F2 super-mode encountered on transequatorial circuits during periods when the equatorial anomaly
is well developed, that is, during the afternoon and early
evenings, during equinoxes, at high levels of solar activity.
This mode involves two F region reflections without an
intervening ground reflection and is characterized by high
signal strength, low fading rates and an MUF 10-15 MHz
higher than for the normal (2F) mode. It is often described
as the afternoon-typeTEP (transequatorialpropagation)mode.
10-59
CHAPTER 10
A second mode associated with the equatorial ionosphere, and which can coexist with the afternoon-type TEP
mode in the early evening, is called the evening-type TEP
mode. It is characterized by strong signals, rapid flutter
fading and frequencies well above the normal 2F MUF. A
propagation frequency of 102 MHz has been regularly observed on a Japan-Australia circuit. The propagation mode
in this case is probably a ducted mode, the signals traveling
within the walls of equatorial "bubbles" [Heron and
McNamara, 1979 and Winkler, 1981]. This mode is also
most likely to occur during the equinoxes at solar maximum.
A review of TEP has been published by Nielson and Crochet
[1974].
Enhancements of the MUF on transequatorial circuits
have also been attributed to combination modes in which
one hop is via a reflection from the equatorial Es belt which
stretches a few degrees either side of the magnetic equator
[McNamara, 1974a]. In an Es-F combination mode, the
MUF is enhanced because the F layer hop is longer than
the usual hop length on a 2F mode. Similar MUF enhancements on nighttime circuits have been attributed to a combination mode in which one hop relies on scatter from F
layer irregularities [McNamara, 1974b]. Propagation above
the MUF due to scatter by small scale irregularities can
routinely be observed [Rawer, 1975].
The high latitude also produces its share of unusual
propagation modes, mostly because of the presence of the
midlatitude trough. Strictly speaking, the modes are not
different from those expected-they are just heavily affected
by the presence of the trough.
Another interesting propagation mode is the round-theworld (RTW) propagation mode. This mode relies on aspects of the vertical distribution of electron density rather
than on latitudinally localized features of the ionosphere.
Consideration of observed elevation angles, signal losses
and time delays have led to the conclusion that RTW signals
propagate via a chordal mode of propagation, rather than
by a uniform multihop mode.
10.4.8 Short-TermForecasting
of HF Conditions
Long-term predictions of monthly median parameters of
the ionosphere such as described in Section 10.4.4 are the
traditional approach to frequency management. Frequencies
are selected which should ensure at least a 90% success rate
for communications at all times of day, during any season,
and at all parts of the solar cycle. Engineering decisions
such as transmitter power and antenna configuration are also
made at this stage. Optimum working frequencies (OWF)
are derived for each hour of the day, usually for a month
at a time, using empirically derived frequencies and signalto-noise ratios. Use of predictions in this manner continues
to provide acceptable, though not high quality, communications for many purposes such as voice and telegraph con-
10-60
ditions. For a survey of the historical development of forecasting methods for HF propagation see Rawer [1975].
The effects on the ionosphere and communications of
solar disturbances which cause ionospheric storms have been
discussed in Section 10.4.6. However daily values of foF2
and MUF are known to vary by about 15% to 20% from
the monthly median values during quiet times as well as
during ionospheric/magnetic storms. See, for example, Figure 10-53. These variations may be superimposed on slower
upward or downward drifts in values over several days. It
is desirable to predict all of these variations for the purpose
of efficient radio communications.
There are several possible approaches to the short-term
forecasting of ionospheric parameters.
1. Using associated geophysical parameters.
2. Using the ionospheric parameters themselves.
3. Using ionospheric indices.
The day-to-day variability of foE and foF1 is so small that
monthly median values of these parameters can be used to
represent the daily variation in the E and Fl regions [Rush
and Gibbs, 1973] and therefore offer little difficulty.
Short-term forecasts of foF2 have been made by relating
changes in foF2 to corresponding changes in selected solar
and geophysical variables such as the 10.7 cm solar flux
and the geomagnetic index Kp [Bennett and Friedland, 1970;
Ichinose et al, 1980]. The disadvantage of this type of approach is that the independent variables upon which changes
in foF2 are assumed to depend must themselves first be
predicted. Even if this is done successfully, only limited
success is possible because the correlations of these parameters with parameters of the ionosphere are not very high.
A more successful approach to the short-term forecasting
of foF2, or alternatively of an operational MUF on a given
circuit, is a prediction scheme based directly upon immediate past observations of foF2 or an operational MUF. Such
prediction schemes for foF2 are described, for example, by
Rush and Gibbs [1973], Lyakhova and Kostina [1973],
McNamara [1976], and Wilkinson [1979].
Rush and Gibbs 11973] used a five-day weighted mean
value of foF2 to predict daily and hourly values of that
parameter. The method of Lyakhova and Kostina [1973] is
based on the observation that correlation coefficients between the deviations of foF2 from median values remain
greater than 0.5 for up to four hours. The high correlation
between hour-to-hour variations of foF2 has been discussed
by Lyakhova [1960], Radinov [1963], Gautier and Zacharisen [1965] and Rush [ 1972].
McNamara [1976] made predictions of foF2 at a particular location up to 3 hours ahead by projecting forward
the trend in the departures of the last few hours' observations
from a 15-day running median. Wilkinson [1979] on the
other hand, simply projected forward in time the deviation
of an observed foF2 value from the predicted monthly median value of foF2. He found this technique to be effective
for lead times of up to about 3 hours.
Similar techniques have been applied to oblique circuits

by Ames and Egan [1967], Ames et al. [1970], Krause et
al. [1970, 1973a,b], D'Accardi [1978], and Uffelman and
Harnish [1982].
The success of any of these forecasting schemes will
depend on the particular circumstances of its intended use,
especially as regards the requried accuracy of the forecast
and the lead time required. Most schemes are reasonably
successful in forecasting an operational MUF that is closer
to the actual value than is the predicted monthly median
value, but only for lead times of the order of an hour or
less.
Forecasts can be made with different lead times, depending on how closely the variations in the ionosphere
need to be tracked, and the sampling interval for actual
observations must match that lead time. In general, the error
in the forecast values will increase with the lead time and
if this error becomes too great, the forecasting scheme would
offer no advantages over the use of a monthly median value.
To make short-term forecasts for circuits for which no
real-time observations exist, the behavior of the ionosphere
must be inferred from such data as are available, using these
observations to infer the values along the required circuit.
Numerous studies have reported correlation coefficients that
illustrate the degree to which hourly deviations from median
values of ionospheric parameters at two or more locations
are related [Gautier and Zacharisen, 1965; Zacharisen, 1965;
Zevakina et al., 1967; Rush, 1972]. These correlations may
be used to infer values of foF2 at locations remote from the
observing site, with an error depending on the separation
of the two locations in latitude and longitude, and on the
local time and season. Rush [1976] has used the observed
correlations to establish the usefulness of, and requirements
for, a network of ground-based and satellite-borne ionospheric observations whose measurements are to be used
for short-term forecasting of radio propagation conditions.
However, it should be noted that the relatively high correlations used by Rush originate from the very disturbed
days (that is, ionospheric storm days) and these same high
correlations are not always obtained for days when the deviations from the median values are relatively small.
A third approach to short-term forecasting is to use observations at all available ionospheric stations to determine
an effective ionospheric index, which can then be used in
conjunction with synoptic monthly median maps of foF2
and other parameters (see Section 10.4.4) to predict the
values of these parameters at the reflection points of the
given circuit. The limitations of this technique are set by
the accuracy with which an appropriate index can be determined from a restricted subset of observations and the
accuracy with which the ionospheric model, driven by this
index, can reproduce the actual ionosphere at the reflection
points. An example of the use of ionospheric indices to
update models of the ionosphere has been given by McNamara [1979]. Other, more complicated, methods have
been described by Thompson and Secan [1979] and Tascione et al. 11979]. Actual forecasts of propagation con-
ditions requires the projection forward in time of the effective ionospheric index. In the limit of zero lead-time, the
"forecast" in fact becomes a real-time assessment of the
ionosphere.
A brief review of short-term forecasting for HF propagation has been given by CCIR [1981b]. The special case
of high latitude propagation is considered in CCIR [1983].
10.4.9 Real Time Channel Evaluation

To take full advantage of the HF communications potential of the ionosphere and to overcome its inherent variability, frequency management should be implemented in
three stages, namely long-term predictions (Section 10.4.4),
short-term forecasting (Section 10.4.8) and real-time channel evaluation (RTCE). It is at the first stage that engineering
decisions such as site location, antenna configuration, and
transmitter power are made, and suitable operating frequencies applied for from the appropriate regulating body.
The second stage, that of short-term forecasting, determines
which of the allocated frequencies will actually propagate
now or a short time ahead. These two stages are generally
all that would be required for voice and low-speed telegraphy circuits.
This two-stage approach has several limitations:
1. The signal-to-noise data are not always reliable and
this ratio is not necessarily a useful criterion for choosing
frequencies for some forms of communications.
2. Reliable long-term predictions of sporadic E are not
available. (Es modes are often the best to use.)
3. No account is taken of interference from other users.
4. The forecast available frequency range will be uncertain to some extent because of inevitable errors in the
forecasting models.
5. The approach does not indicate which of the assigned
frequencies propagating at a given time would be the best
to use for a given form of communication.
RTCE is the third stage of a frequency management
system required to maintain reliable high quality HF communications even under the most adverse conditions. It becomes especially important for medium and high speed digital data transfer. With an RTCE system, all channels are
sounded in turn to determine which is actually the best to
use at a given time for the particular type of communication/modulation system.
Darnell [1978] has given the following definition of
RTCE: "Real-time channel evaluation is the term used to
describe the processes of measuring appropriate parameters
of a set of communication channels in real time and of
employing the data thus obtained to describe quantitatively
the states of those channels and hence the relative capabilities for passing a given class, or classes, of communication
traffic."
A review of RTCE has been prepared by Study Group
6 of CCIR [CCIR, 1981d], and the interested reader is
10-61
CHAPTER 10
referred to that report. One of the more recently developed
techniques of RTCE, which can serve as an example of the
technique, is that of pilot tone sounding.
In pilot tone sounding [Betts and Darnell, 1975], low
level CW tones are either inserted into the data spectrum
or transmitted in potentially available alternative channels.
At the receiver, simple measurements on the tones enable
the relative states of the channels to be specified in terms
of predicted (digital) data error rates. In a situation where
a multicomponent broadcast is being used to radiate identical
data simultaneously on each of several frequencies, this type
of RTCE system allows the best component to be selected
automatically at the receiver.
One of the main advantages of the pilot-tone RTCE
technique is the extreme simplicity of the concept and the
implementation when compared with other techniques for
RTCE, for example, oblique incidence ionosondes. A further advantage of the technique is that it permits RTCE and
data signals to be combined in a simple format, rather than
requiring a separate stimulus. The technique also lends itself
readily to automation. With an automated system, there will
be little difficulty in principle in including into the algorithm
for selecting the optimum working frequency any measurements of other pertinent properties of the pilot tone, for
example, amplitude, Doppler shift, and spectral width. The
noise level in the channel could also be included.
The pilot-tone sounder technique does not permit the
determination of the operational MUF or identification of
propagation modes. The former may cause difficulties when
the MUF is decreasing (for example, due to normal diurnal
variation) if the working frequency is just below the MUF.
Difficulties could be severe if the MUF changes significantly
during the RTCE time (typically 3 min per channel).
The signal optimization problem at HF is potentially
very complex and to date RTCE has been used almost exclusively as a frequency selection aid. Darnell [1975b] lists
the various parameters on an HF link which could be under
the control of the communicator and hence be made adaptive
in response to the RTCE data:
Transmission frequency
Transmitter power level
Bandwidth
Data rate
Modulation type
Time at which the transmission is made
Signal processing algorithm at the receiver
Elevation angles for antenna array beams
Diversity type
To this list may be added antenna polarization for near
vertical incidence propagation.
The potential advantages accruing from the use of RTCE
techniques have been summarized by Darnell [1975a]:
1. The effect of man made noise and interference can
be measured and specified quantitively [Darnell, 19781.
2. The facility for real-time, on line measurement of
propagation and interference allows the use of relatively
10-62
transient propagation modes, for example, sporadic E layer

propagation.
3. RTCE evaluation allows a more efficient use of the
frequency spectrum by tending to select frequency channels
higher than those which would be chosen via prediction
techniques. Thus spectrum congestion is reduced.
4. RTCE will provide a means of automatically selecting
the best frequency and simultaneously indicating preferred
stand-by channels.
5. Transmitter power can be minimized, consistent with
providing an acceptable quality of received traffic.
6. In the longer term, RTCE data can be used to adapt
other parameters of a communication system, apart from
frequency, optimally for the prevailing path conditions.
10.4.10 Conclusion
It has not been possible within the present limitations
of space to present more than a cursory overview of the
subject of HF propagation. However, it is hoped that most
readers will be able to follow up on topics of special interest
by going to the references cited. In this regard, the value
of the proceedings of the "Boulder Workshop" (Donnelly,
1979; 1980) cannot be stressed too highly. The CCIR documents and the AGARD Lecture Series No. 127 should also
prove very useful, but possibly somewhat harder to obtain.
10.5 IONOSPHERIC DISTURBANCES

10.5.1 Sudden Ionospheric Disturbance (SID)
The SID is caused by x-ray and ultraviolet emissions
from solar flares. These emissions produce increased ionization of the sunlit ionosphere. This excess ionization typically lasts for a period of 10 to 60 minutes, depending on
the intensity and duration of the responsible flare. SIDs are
observed by monitoring manmade and natural High Frequency (HF) or Very Low Frequency (VLF) radio signals
that propagate via the ionosphere. The various types of SIDs
are named according to the method of their observation:
1. SWF-short wave fadeout: a decrease (either sudden
or gradual) in the signal received from a distant HF
(2-32 MHz) transmitter,
2. SCNA-sudden cosmic noise absorption: a decrease
in the intensity of the constant galactic radio noise as
measured by riometers (relative ionospheric opacity
meters) operating between 15-60 MHz,
3. SPA-sudden phase anomaly: a change in phase of
a received VLF (10-150 kHz) signal relative to a
frequency standard at the receiving site,
4. SES-sudden enhancement of signal: an increase in
the strength of an incident VLF signal occurring at
the same time as an SPA,

5. SEA-sudden enhancement of atmospherics: an increase in the background VLF noise from distant
thunderstorms,
6. SFD-sudden frequency deviation: a short-lived increase in the frequency of the signal from a distant
HF transmitter,
7. SFE-solar flare effect or geomagnetic crochet: sudden variation in the H component of the earth's magnetic field.
The first five of these effects are all attributable to increased
ionization in the D layer, primarily by soft (1-8 A) flare xrays. The absorptive effects (SWF and SCNA) are caused
by a thickening of the absorbing layer through which the
HF waves pass, while the effects on VLF signals are due
to the lowering (SPA) and strengthening (SES and SEA) of
the D layer from which these lower frequency waves are
reflected. In contrast, SFDs arise from the flare associated
increase in ionization in the E and F regions produced by
extreme ultraviolet (EUV) radiation in the 10-1030 A range.
Finally, the geomagnetic crochet (SFE) appears to be a
hybrid effect, having an impulsive component associated
with the flare "flash phase" EUV emission, and a more
gradual component associated with the flare soft (1-8 A) xrays [Richmond and Venkateswaran, 1971]. These ionizing
emissions temporarily increase the D and E layer conductivity and alter the normal ionospheric currents to give rise
to the SFE. For a comprehensive review of the SID phenomenon, see Mitra [1974].
10.5.2 Polar Cap Absorption (PCA)

While SIDs are caused by flare electromagnetic emission, PCAs result from bombardment of the ionosphere by
flare-accelerated protons with energies < 10 MeV. These
particles stream into the earth's polar regions along geomagnetic field lines and produce increased ionization in the
D layer (as well as aurora at E layer heights). PCAs are
normally observed by means of riometers. In contrast to the
SID, the PCA is a long-lived effect, with durations ranging
from tens of hours to several days. In general, PCAs follow
only the most intense solar flares and begin within a few
hours after flare maximum, dependent on the flare longitude.
A review of the PCA phenomenon has been published by
Hultqvist [1969].
10.6 HF RADAR IONOSPHERIC

CLUTTER
Backscatter reflections from E layer auroral ionization
and F layer electron density irregularities can be an important source of clutter for a radar operating in the high latitudes. The ionospheric clutter can degrade the performance
of an over-the-horizon (OTH) radar designed for the surveillance of aircraft at ranges of approximately 1000 to 3500
km.
The radar reflections are the result of scattering from
electron density irregularities aligned along the lines of force
of the earth's magnetic field. The characteristics of the fieldaligned scatterers are such that the radar echoes originate
in a small range of angles about perpendicular incidence to
the magnetic field lines and that the amplitude of the auroral
echo is aspect-angle sensitive. In addition to the orthogonality condition, it is necessary that this geometric configuration take place at ionospheric altitudes, that is, 80 km
and above.
The probability of observing radar reflections from ionization irregularities is also dependent upon the frequency
of occurrence of E layer auroral or spread-F irregularity
activity in the region of interest. Reflections can be expected
to occur in regions where both the conditions of near-perpendicularityat ionsophericheightsand high auroral or spreadF irregularity activity are satisfied. The fulfillment of only
one requirement is not sufficient to warrant a radar reflection.
An appropriate model to use to define the condition of
auroral activity is the Feldstein-Starkov [1967] auroral belt
(oval) model. Since the location and extent of the auroral
oval are a function of time and geomagnetic activity, E layer
auroral echoes can be expected to appear over a wide area
during magnetically disturbed periods for an HF backscatter
radar located at high latitudes.
The F layer irregularities can be described in terms of
the probability of occurrence of spread-F derived by Penndorf [1962] or in terms of Aarons' [1973]irregularity scintillation region.
In this section, an estimate is made of the characteristics
of ionospheric clutter that could be observed by an HF
backscatter radar operating in the midlatitude with the antenna beam oriented towards the polar region. The topics
to be discussed are the amplitude, the backscatter crosssectional area, the angular extent, the Doppler frequency
variation, the frequency of occurrence, the diurnal and seasonal variation and the correlation with solar-geophysical
conditions.
10.6.1 Signal Amplitude

The amplitude of E layer auroral clutter that could be
encountered by an HF backscatter radar is deduced by extrapolation of radar-auroral data recorded by SRI International at Fraserburgh, Scotland, during 1959 and 1960
[Leadabrand et al., 1965]. The radar measurements were
made simultaneously at frequencies of 30, 401 and 800
MHz.
To predict the SRI International auroral data effects on
an HF backscatter radar in terms of the corresponding radar,
it is necessary to determine the relative sensitivities of the
10-63
CHAPTER 10
For aurora which fills the antenna beam, o is replaced
the volume scattering coefficient, ov,that is, radar cross
section per unit volume, such that, for a rectangular antenna,
30
AURORAL
20
x1
-by
~2o0~~~~
- a d
rREGIONOF
&
,B
200
300 40o0
500
FFREOQUENCY
(MHz)
o100
uJ
/
_
1> 15
x15
600
-
700
800
800-MHz ORTHOGONAL
POLARIZATION
---.
-10AURORA
800-MHz TRANSMIT.
POLARIZATION
-20
c = crvV = (vR 3.H3v02 ,
(10.61)
where V is the volume filled by the scatterers, c is the free

space velocity, T is the pulse length, and BHand Bv are the
horizontal and vertical antenna beamwidth, respectively.
Figure 10-54 is a plot of the relative signal strengths of
auroral echoes recorded by SRI International at the three
-30
Figure 10-54. Relativesignalstrengthof auroralechoesat 30, 401, and

800 MHz obtained by the SRI International-Scotland radars
[Leadabrand et al., 1965].
radars to the aurora. This is accomplished by comparing the

various parameters that enter into the radar equation.
According to the radar equation, the signal-to-noise ratio
of the received signal is given by
Pr
N
PtGtGrG r
(47r)R KTB NF LsLp
(10.60)
3 4
where Pt is the transmitted power, G1 and Gr are the gains

of the transmitting and receiving antennas, respectively, X
is the wavelength, o is the radar-target cross section, R is
the radar range, K is Boltzmann's constant, T is the ambient
temperature, B is the receiver noise bandwidth, NF is the
receiver noise figure, Ls is the system loss and Lp is the
two-way loss due to the propagation medium. This relationship is applicable to a point target, that is, the antenna
beamwidth is much larger than the dimensions of the target.
frequencies in April 1960. The region of uncertainty between 30 and 401 MHz results from the fact that there was
no way to specify to what degree the auroral scatterers filled
the 30-MHz antenna beam. At 401 and 80 MHz, the antenna
beamwidth was 1.2 , while at 30 MHz it was 15 . Thus, it
was assumed that the scatterers completely filled the beam
at the two higher frequencies. However, it was doubtful
that the beam was completely filled at 30 MHz. The 1 x 10
aurora is assumed to be a point target for the 30-MHz antenna.
The data in Figure 10-54 are normalized with respect to
401 MHz. For a filled beam antenna, the relative signal
strength of the 30-MHz echoes was 6-dB greater than those
at 401 MHz, while, for the point target case, there was a
28-dB difference between the 30- and 401-MHz data.
The absolute magnitude of the 30-MHz auroral echoes
can be deduced by comparing the system sensitivities of the
30- and 401-MHz radars. According to Table 10-9, the 401MHz radar was 48.9-dB more sensitive than the 30-MHz
radar assuming that the aurora was a point target and 26.9dB more sensitive, assuming that the 30-MHz antenna beam
was completely filled with scatterers. Thus, it follows that
Table 10-9. Comparison of SRI International-Scotland 30-MHz and 401-MHz radar sensitivities utilized in the April 1960 radar-auroral observations.
Point Target
(401 MHz/30 MHz)
(dB)
Scatterers Fill
Antenna Beam
(401 MHz/30 MHz)
(dB)
30-MHz
Radar
401-MHz
Radar
x2 (m2 )
Pt (Mw)
G,Gr(dB)
(10.0)2
(0.748)2
-22.5
-22.5
0.0015
40
0.12
84
+ 19
+44
+ 19
+44
15
1.2
- 11
15
1.2
- 11
T(IJS)
300
B (kHz)
NF (dB)
L, (dB)
6
9
7
300
6
4.5
3.1
Parameter
(DEG)
3
(DEG)
10-64
0
0
+4.5
+3.9
+48.9
0
0
+4.5
+3.9
+26.9

I
50
45
SATURATION
1
Z
.-
40-
79
35
Z
LEVEL-2 30 -NOISE
c
-133 dBm
u25
u. 20
0
70
50
30 MHz-
NORMAL
-115
-105
-95
-85
AMPLITUDE
-75
1dBm)
-65
-55 .
0.1
-45
_
_
Figure 10-55. Amplitude histogram of SRI International 401-MHz Scotland radar auroral data [Leadabrand et al., 1965].
0.1
0.01
-120
the magnitude of the 30-MHz radar-auroral data was 20.9dB less than that of the 401-MHz data.
The amplitude distribution of the 401 -MHz radar-auroral
__
eSR XPRIMENTAL DATA

STANDARD DEVIATION = 14.8dB
I
I , I
I I
-88
-72
-104
AMPLITUDE (dBm)
-56
Figure 10-56. Cumulativedistributionfunctionof SRI International-Scotland radar auroraldata.
echoes is presented as a histogram form in Figure 10-55.

The system noise level was approximately - 133 dBm and
that the system saturated at - 60 dBm, which accounts for
the large number of echoes at that amplitude.
The histogram data, when replotted as a cumulative
distribution on probability paper, shown in Figure 10-56,
closely fit a straight line. This characteristic is indicative
that the 401-MHz amplitudes are log-normally distributed.
The theoretical normal distribution function plotted in Fig-
ure 10-55 was computed utilizing the value of -77.2 dBm

for the mean as obtained from Figure 10-56. The 30-MHz
amplitude data, also shown in Figure 10-56, was derived
on the basis that the 30-MHz auroral echoes followed a
Gaussian distribution and were 20.9-dB less than the 401MHz results.
30-MHzradar.
Table 10-10. Comparisonof the sensitivityof an HF backscatterradar to that of the SRI International-Scotland
Point Target
(HF Radar/SRI)
(dB)
Scatterers Fill
Antenna Beam
(HF Radar/SRI)
(dB)
HF
Backscatter
Radar
SRIScotland
Radar
X2 (m2 )
Pt (MW)
G, Gr
(dB)
(10.0)2
0.8
45
(10.0)2
0.0015
40
13 H
(DEG)
13v
(DEG)
T (pLS)
15
-4.8
15
15
400
10
2.5
100
12
3
300
Parameter
B (kHz)
NF (dB)
L, (dB)
6
9
7
0
+ 27.3
+5
+ 3.5
- 12.2
-3
+4
(r = 400 Ls) + 37.1 dB
(T = 10 ps) +21.1 dB
0
+ 27.3
+5
+ 1.2
- 14.8
+3.8
- 12.2
-3
+4
+ 33.5 dB
+ 1.5 dB
10-65
CHAPTER
10
Table 10-11. Estimate of radar-auroral clutter levels of a hypothetical HF backscatter radar at 30 MHz based on SRI International data.
Scatterers Fill
Antenna Beam
Point Target
= 4 0 0 ps
T = 10 ps
Parameter
(dBm)
(dBm)
(dBm)
T = 1]0RS
(dBm)
Upper Decile
Upper Quartile
Median
-42.0
- 51.0
-61.0
- 58.0
- 67.0
- 77.0
- 45.6
- 54.6
- 64.6
- 77.6
- 86.6
- 96.6
Lower Quartile
Lower Decile
Standard
Deviation
-71.0
- 80.0
14.8
- 87.0
- 96.0
14.8
-74.6
- 83.6
14.8
- 106.6
- 115.6
14.8
Statistical
To estimate the magnitude of the auroral echoes that

could be observed by an HF backscatter radar, it is necessary
to compare the sensitivity of the HF backscatter radar to
that of the SRI International 30-MHz radar. The parameters
of an hypothetical HF radar, considered in this analysis, are
given in Table 10-10. For an assumed pulse length of 400
us, the HF radar is 37.1-dB more sensitive than the SRI
International 30-MHz radar for the point target case and
33.5-dB more sensitive for the filled beam case. When the
pulse length is decreased to 10 u.s, the difference in sensitivities decreases to + 21. 1dB and + 1.5 dB for the point
target and filled beam case, respectively.
The predicted HF backscatter radar-auroral clutter levels
at 30 MHz are presented in Table 10-11. The SRI International 30-MHz radar-auroral data given in Figure 10-56
and radar sensitivities in Table 10-10 are used in the calculations.
In deducing the radar-auroral signal levels at other frequencies in the HF band, it is assumed that the auroral echo
power is frequency dependent according to the law
P = k
f-",
T = 400ps
imately the 10- to 20-MHz range. From simultaneous auroral echo measurements at frequencies of 49.7, 143.5 and
226 MHz, Flood [1960] has deduced a value of n = 3.5
between 49.7 and 143.5 MHz and n = 6.5 between 143.5
and 226 MHz.
Measurements of E region radar aurora at frequencies
in the HF band and at VHF (143.8 MHz) from locations in
Finland and Germany indicated a frequency dependency of
n = 3 [Oksman et al., 1979].
Since there is a wide discrepancy in the experimental
measurements of the frequency dependence of auroral backscatter, which could be due to the characteristics of the
auroral ionization, that is, inhomogeneous distribution of
auroral electrons, varying scale sizes of ionization irregularities and different scattering altitudes, a value of 4 for
the exponent in Equation (10.62) is assumed in this analysis.
The estimated median auroral clutter amplitudes that
could confront the hypothetical HF backscatter radar in the
5- to 30-MHz band are shown in Figure 10-57. The external
(10.62)
-20-T = 400/ts
where k and n are constants.
Utilizing the data in Figure
. -0
10-54, it can be shown that for this power law n = 2.5 for
a point target and n = 0.5 for the filled beam case. However, an examination of other radar-auroral data taken at
Scotland revealed a value of n = 6.8 for a point target and
n = 2.7 for a filled beam [Leadabrand et al., 1965]. An
analysis of auroral echoes from simultaneous multiple frequency observations in Alaska by Leadabrand et al., [1967]
revealed than n = 2 for frequencies between 50 and 850
MHz and n = 5 between
-60
80
E
between approximately 20 and 50 MHz and -4 between 5

and 20 MHz. At times,
however,
it
and 20 MHz.
however,
At times,
it was
was found
found that
that the
the
amplitude returns were insensitive to frequency in approx-
10
i
'
POINTTARET
PFILLNTA
PULSE LENGTH
<-120
--
850 and 3000 MHz.
Radar backscatter measurements of artificial electron

clouds in the E region of the ionosphere by Gallagher and
Barnes [1963] yielded a constant, n, of 4 for frequencies
10-66
EXTERNAL NOISE LEVEL--___
-160
100
6 8 101
2
FREQUENCY
(MHz)
,
4
8 102
Figure 10-57. Estimated

median auroral
clutter
based on
SRI
International-Scotland
radar
data,amplitude
f4 frequency
dependency.
100
--0
_100
s
90
70
g901
TARGET
o -POINT
--FILLED BEAM
= PULSE LENGTH
70
T= 10sS
.....
POINT TARGET
---FILLED BEAM
T= PULSE LENGTH
os
60
0- 6[
C-
--
05o
404
10
2 4
6
8102
100
8 101
102
FREQUENCY (MHz)
FREQUENCY (MHz)
Figure 10-58. Estimated median radar-auroral clutter-to-noise ratio based

on SRI International-Scotland radar data, f-4 frequency dependency.
Figure 10-59. Estimated median radar-auroral clutter-to-noise ratio based

on SRI International-Scotland radar data, f-2 frequency dependency.
noise level of a rural environment also shown in Figure

10-57 is the average median value as predicted for a location
in Maine at an azimuth of 60 for all seasons, all times of
day and a sunspot number of 70. It is seen that the auroral
clutter amplitude increases with decreasing frequency and
bandwidth and that the amplitude of the point target case
is greater than that of the filled beam case.
Figure 10-58 is a replot of the data in Figure 10-57 in
terms of the auroral clutter-to-noise ratio. It is of interest to
note that, for a f-4 frequency dependency, the clutter-tonoise ratio maximizes in the vicinity of 10 MHz. This is
due to the fact that the external noise level increases with
decreasing frequency.
When a frequency dependency of f 2, that is, n = 2 in
Equation (10.62), is assumed, the ionospheric clutter levels
over the HF band are decreased, with respect to the f-4
estimates, by an amount of 15.6, 9.5, 6.0, 3.5 and 1.6 dB
at 5, 10, 15, 20 and 25 MHz, respectively. As shown in
Figure 10-59, the clutter-to-noise ratio for the f-2 case, monotonically increases with increasing frequency. The clutterto-noise ratio at 30 MHz is 12.2 dB greater than that at 5
MHz. The upper and lower decile values of the data presented in Figures 10-58 and 10-59 are + 19 dB and - 19
dB with respect to the median level.
The height distribution of the auroral echoes observed
by SRI International in Scotland was peaked at about 100
to 120 km, that is, in the E region, although heights as great
as 200 km were observed. Thus, the data presented in Figures 10-58 and 10-59 can be considered to apply only to E
region reflections.
It should also be mentioned that the amplitude distribution of the 401-MHz auroral echoes shown in Figure 1055, which was used as a basis of extrapolation to the HF
backscatter radar, was applicable to data obtained when the
direction of propagation at the reflection point was perpendicular to the magnetic field lines. When directions of prop-
agation and magnetic field lines are not orthogonal, the

auroral clutter amplitude should decrease.
Bates and Albee [1969] report that, assuming a simple
model of auroral backscatter from the E region that includes
ionospheric refraction effects, a lower limit of 6 dB/deg in
the 15- to 50-MHz range was observed for the aspect-sensitive decrease in backscatter cross section with off-perpendicular angle from the geomagnetic field. The aspect-sensitive decrease of the cross section of F layer backscatter
echoes was found by Bates and Albee [1970] to be on the
order of 5 dB/deg of off-perpendicularity from the magnetic
field in the 6- to 15-MHz region. An aspect angle decay of
2 dB/deg at 30 MHz was deduced from the Scotland-auroral
measurements by Leadabrand et al. [1965].
10.6.2 Cross Section

HF radar observations conducted at Caribou, Maine,
under the Polar Fox II program, at frequencies of 8.125 and
14.875 MHz, during the period between December 1971
and November 1972 are used to estimate the cross section
of ionospheric clutter.
The radar cross section given by Equation (10.61) can
also be expressed by
(r = cr,RPH
cT
,2
(10.63)
where Ah is the thickness of the aurora in the vertical direction. Since Ah was an unknown factor in the Polar Fox
it was assumed to be unity (1m). Thus, Equaexperiment,
II
tion (10-63) can be written as
=
or =
0`AA
('AR[
(10.64)
(10.64)
10-67
CHAPTER 10
Table 10-12. Average area scattering coefficient of E layer and F layer
irregularities based on Polar Fox II data, January 1972.
Area Scattering
Coefficient (dBsm/m2 )
Frequency
(MHz)
Statistical
Parameter
E Layer
F Layer
8.125
Upper Quartile
Median
Lower Quartile
Upper Quartile
Median
Lower Quartile
- 34.2
-47.6
-58.6
-56.5
-63.9
- 74.0
- 27.2
-33.6
-42.0
-37.1
-47.2
- 54.3
14.875
where OAis the area scattering coefficient and A is the area

containing the scatterers.
The area scattering coefficients of E layer and F layer
irregularities as deduced from the Polar Fox II data are given
in Table 10-12. The data show the averages over all times
in the month of January 1972, and over all azimuths in the
90 sector between - 30 and + 60 with respect to true
North. It is seen that, for all the statistical parameters, the
8.125-MHz data are from 12- to 21-dB greater than the
14.875-MHz results. The most interesting feature of the
basic data, from which Table 10-12 was derived, was the
lack of evidence of the influence of the geographic locations
of the ionospheric scatterers on the magnitude of the area
scattering coefficient. It was hypothesized that the scattering
coefficient would be greater in the northerly directions since
this is the region encompassed by the auroral oval [Feldstein
and Starkov, 1967].
The cross section of ionospheric clutter that could be

observed by the HF backscatter radar is presented in Table
10-13. The calculations are based on an assumed radar range
of 700 km for the E layer irregularities and 1500 km for
the F layer irregularities. These ranges correspond to a ray
path oriented at an elevation angle of 5 and intersecting
the ionosphere at altitudes of approximately 100 and 300
km, respectively. For this configuration, the E layer scattering area evaluates to 95.6 dBsm and 79.6 dBsm for a
pulse length of 400 us and 10 us, respectively. For the
corresponding pulse lengths, the F layer scattering area is
99.0 dBsm and 82.9 dBsm. According to Table 10-13, it
is evident that the F layer clutter cross section is about 10
to 20 dB greater than the E layer cross section at 8.125
MHz and 20 to 23 dB at 14.875 MHz. The differences
between the F layer and E layer cross sections are most
likely due to the fact that the F layer ionization level is
many times that of the E layer.
It should be noted that the Polar Fox II estimates of the
E layer and F layer area scattering cross section contain all
the propagation loss terms, that is, the ionospheric propagation losses had not been removed from the calculations.
Because of this, it is not possible to use the Polar Fox II
data for extrapolation to other frequencies in the HF band.
10.6.3 Angular Extent

The angular extent of HF ionospheric clutter was determined from Polar Fox II radar data recorded in January
1972 at six frequencies ranging between 8 and 23 MHz.
Figure 10-60 contains plots of the cumulative distribution
of the angular extent of E layer, F layer, and the combined
E and F layer clutter. According to Table 10-14 which
Table 10-13. HF backscatter radar-estimated average cross-section of E layer and F layer irregularities based on Polar Fox II data, January 1972.
Cross Section
Ratio
(dB)
Average
Cross Section
(dBsm)
Frequency
(MHz)
Pulse Length
(Ais)
Statistical
Parameter
E Layer
F Layer
F Layer/E Layer
8.125
400
Upper Quartile
Median
Lower Quartile
Upper Quartile
Median
Lower Quartile
Upper Quartile
Median
Lower Quartile
Upper Quartile
Median
Lower Quartile
61.2
48.0
37.0
45.4
32.0
21.0
39.1
31.7
21.6
23.1
15.7
5.6
71.8
65.4
57.0
55.7
49.3
40.9
61.9
51.8
44.7
45.8
35.7
28.6
10.6
17.4
20.0
10.3
17.3
19.9
22.8
20.1
23.1
22.7
20.0
23.0
10
14.875
400
10
10-68

1.0
120
__-___-r_
DOPPLER
[T
I
I
SPREAD (LEFT SCALE)
60
DOPPLERSHIFT (RIGHT SCALE)

50
100
0.6
o.2
<0.4
~::~-
='
,0.2
16
24
2 DE-AND F-LAYER-
g~~~~~~~~od
__I
32
40
48
56
64
72
80
80
40
60
<~c
88
nr
ANGULAR EXTENT (DEG)
.j
-
/+20
40 .
a.
0
Figure 10-60. Cumulative distributionof the angular extent of ionospheric
clutter based on Polar Fox II data.
summarizes the statistical characteristics of the three curves

in Figure 10-60, the median values evaluate to 60.20, 18.8
and 27.6 , respectively. The angular dimensions characterized in Table 10-14 were deduced from ionospheric clutter
present only in the 1-hoppropagation mode.
10.6.4 Doppler Frequency Spectrum

The Doppler spectrum of radar pulses reflected from the
aurora differs from that of the original transmitted pulse.
The changes that can take place are (1) the center frequency
can be shifted, and (2) the spectral width can be increased.
The shift in the center frequency corresponds to a drift
motion of the auroral ionization. Radar-auroral data indicate
an east-west drift before magnetic midnight and a west-east
drift after magnetic midnight [Leadabrand et al., 1965].
Drift velocities of the order of 500 m/s are typical. According to the analysis of E layer radar-auroral echoes recorded in Scotland at a frequency of 401 MHz, the maximum Doppler shift is normally 2.15 kHz [Larson and
Hodges, 1967].
The spread in the Doppler spectrum is due to the random,
turbulent motion of the irregularities of electron density in
the auroral ionization. According to Larson and Hodges
_+300
-/z-6 dB
ca
20
+10
0
FREQUENCY
(MHz)
Figure 10-61. Estimate of E layer HF radar-auroral doppler frequency
measpreadandshiftbased on SRI International-Scotland

surementsat 401 MHz.
[1967], the maximum Doppler frequency spread at 401 MHz

is on the order of 1.90 kHz at - 6 dB below the peak, 4.00
kHz at - 12 dB, and 4.75 kHz at - 18 dB.
The expected (one standard) deviation of Doppler frequency shift and spread at HF is presented in Figure 10-61.
The estimates are based on the fact that the Doppler frequency variations are directly proportional to frequency and
on the assumption that shifts and spreads are normally distributed. Thus, the maximum shift and spread are equivalent
to the 3-sigma (standard deviation) value of the distribution.
120
N
100
-18 d
_+80
+60
Table 10-14. Statistical distribution of the angular extent of ionospheric

clutter based on Polar Fox II data.
Angular Extent (Deg)40
-6 dB
+40
Statistical
Parameter
E Layer
Clutter
F Layer
Clutter
E and F Layer
Clutter
20
Upper Decile
Upper Quartile
Median
87.7
83.8
60.2
76.0
42.4
18.8
87.0
74.4
27.6
Lower Quartile
Lower Decile
19.0
11.6
10.0
2.8
13.6
3.4
10
15
20
25
30
FREQUENCY (MHz)
Figure 10-62. Estimate of E layer HF radar-auroral total doppler frequency
deviation based on SRI International-Scotland

measurements at 401 MHz.
10-69
CHAPTER
10
At 10 MHz, the Doppler frequency shift evaluates to + 18

Hz and the Doppler frequency spread becomes 16 Hz at the
-6 dB level, 33 Hz at - 12 dB and 40 Hz at - 18 dB.
The estimates of E layer HF radar-auroral total Doppler
frequency deviation are plotted in Figure 10-62. At 10MHz,
it is possible that the 1-sigma value of the total Doppler
deviation could be on the order of 26 Hz at the - 6 dB
level, + 35 Hz at - 12 dB, and + 38 Hz at - 18 dB.
E layer radar-auroral echoes, having radial velocities as
high as 1450 m/s, have been observed at 17 MHz by Brooks
[1966]. This corresponds to a Doppler frequency shift of
164.3 Hz. The maximum radial velocity of F region fieldaligned irregularitiesdetected at 17.3 MHz [Baggaley, 1970]
was found to be on the order of 165 m/s with a mean of 65
m/s. In terms of the Doppler frequency shift, these velocities
evaluate to 19.0 and 7.5 Hz, respectively.
The extrapolation to the HF band of the Doppler measurements of E region irregularities conducted by Hofstee
and Forsyth [ 1969] at a frequency of approximately 40 MHz,
Balsley and Ecklund [1972] at 50 MHz, Balsley et al. [1973]
at 50 MHz, Greenwald et al. [1975] at 50 MHz and Haldoupis and Sofko [1976] at 42 MHz are in agreement with
the results presented in Figures 10-61 and 10-62.
10.6.5 Frequency of Occurrence
and Correlation with

Solar-Geophysical Conditions
The experimental observations of field-aligned echoes
at 19.4 MHz conducted over a three-year period (1961-1963)
at a site located in the vicinity of Boston disclosed that the
echoes were present for as long as 11 h/day [Malik and
Aarons, 1964]. Although a seasonal pattern was clearly
defined, there appeared to be a tendency for the reflections
to occur on a greater number of days during the summer
than during the winter. The correlation of echo activity with
sunspot number was found to exist. For example, in 1961
when the sunspot number was 53, echoes appeared on 73%
of the days at an average of 3.0 h/day. In 1963 when the
sunspot number was 28, the echoes appeared on 55% of the
days at an average of 1.5 h/day. Maximum echo activity
occurred between 1800 and 2000 h local time.
A more thorough analysis of the 19.4 MHz field-aligned
echoes recorded from 1961 through 1965 had been performed by the Air Force Cambridge Research Laboratories
[Basu et al., 1973, 1974]. It was found that under quiet
magnetic conditions field-aligned E layer echoes showed a
summer evening maximum and appeared to be associated
with the ground-backscatter echoes from sporadic E [Basu
et al., 1973]. A weak secondary maximum existed in the
winter with no detectable field-aligned echoes during the
daytime. During disturbed magnetic conditions, the fieldaligned E layer echoes increased with magnetic activity and
appeared during the daytime with a corresponding decrease
10-70
in sporadic E. The data did not reveal a seasonal dependence.

Under quiet magnetic conditions, the field-aligned F
layer echoes were found to be a sunset phenomenon and
correlated with the sunspot cycle [Basu et al., 1974]. In
general, the echoes occurred in the range from 1050 to 1500
km. However, during the daytime, long range echoes, at
times, were observed at ranges of 3000 to 3300 km. An
interesting disclosure was the fact that the occurrence of the
F layer field-aligned echoes increased directly with the increase in magnetic activity until the level KFr (Fredericksburg, Maryland, three-hour K index) attained a value of
four. Beyond this level, there was a decrease in the occurrence of the F layer echoes with no echoes being observed
when KF.r> 7. Radar auroral measurements made at Stanford University at a frequency of 17.3 MHz reveal that
maximum auroral activity occurred between 1900 and 0400
hours local time [Peterson et al., 1955].
According to the HF radar echo observations conducted
at 12 and 18 MHz in the state of Washington, there was a
pronouncedpeak in the frequencyof occurrenceof the echoes
from field-aligned irregularities near the time of local sunset
[Hower et al., 1966]. The characteristics of the echoes detected at sunset were different from those occurring at night.
It was found that the local sunset echoes appeared at ranges
of about 2000 km and were generally spread in range (diffuse). The occurrence of the sunset echoes was relatively
independent of magnetic activity. The nighttime echoes, on
the other hand, appeared at ranges of about 1000 km and
tended to display little spread in range (discrete). The occurrence of nighttime echoes was highly dependent on magnetic activity.
Hower et al., [1966] noted a decreasing echo activity
with solar cycle, the maximum percentage occurrences during the nighttime being approximately 80% in December
1958, 37% in December 1963, and 35% in March 1964.
The (Zurich) relative sunspot number during the corresponding period was 185.2, 11.8, and 14.5, respectively. Hower
and Makhijani [1969] have concluded that HF F layer backscatter echoes arise from the same general irregularity regions in which spread F is detected.
Sprenger and Glode [1964] have reported that the diurnal
variation of the frequency of occurrence of radar-auroral
echoes recorded from late 1958 to mid 1962 at 33 MHz at
Kuhlungsborn (geographic coordinates: 54N, 12E; geomagnetic latitude: 54N) showed a double maximum at approximately 0100 and 1700hours local time. The probability
of occurrence of E layer auroral echoes was found to be a
function of magnetic activity. No auroral echoes were observed at a magnetic index, Kp < 5, and 100% occurrence
at Kp = 9. With regard to the correlation of auroral echo
activity with sunspot cycle, Sprenger and Glode [1964] have
concluded that the maximum of auroral activity occurred in
1960 about two years after the sunspot maximum and that
the activity decreased to zero within a period of only one
and a half years.

Brooks' [1965, 1966]investigationof radar-auroralechoes
at 17 MHz conducted in 1959 and 1960 indicated that, for
the discrete echoes, the maximum number appeared between
0000-0300 hours local time, and for the diffuse echoes,
between 1900-2100 hours local time.
A maximum number of backscatter echoes at 13.866
MHz from the F region on 75% of the nights for March
1958 was reported by Weaver [1965]. A minimum number
of echoes was observed in June, the data being collected
during 1957 and 1958 at Ithaca, New York, with the antenna
beam oriented in a northward direction.
Backscatter observations of F region field-aligned irregularities made at 17 MHz near Sheffield, England (geographic coordinates: 53.43N, 1.58W; geomagnetic latitude: 65.4N) between mid-October, 1964, and mid-January,
1966, [Baggaley, 1970] which was a period of low sunspot
activity, showed that, on the average, the echoes were present on 23.3% of the days while only 1.5% of the total
observing time was occupied by the echoes. A summer
maximum and a winter minimum were found to exist. This
seasonal variation correlates with Malik and Aarons' [1964]
results.
Baggaley [1970] noted that there was a correlation between the onset of F layer echoes and the solar zenith angle
at the reflection point in the F layer (assuming a height of
300 km). That is, a maximum number of echoes occurred
at a solar zenith angle of 90-95 , and a minimum at angles
less than 85 and greater than 125. In addition, no correlation was found to exist between F layer echo occurrence
and magnetic activity. Of the total time for which F layer
echo activity was present, only 11% was associated with
magnetic index Kp - 4.
Field-aligned echoes from the F region observed at Brisbane, Australia, at 16 MHz occurred preferentially during
geomagneticdisturbancesand correlated stronglywith spreadF and radio star scintillations occurring in the same region
of the ionosphere [Swenson, 1972].
An analysis of 29-MHz backscatter measurements carried out in Northern (West) Germany and Scandanavia (geomagnetic latitudes of 55-77N) by Czechowskyet al., [1974]
confirmed the close correlation between radio-auroral occurrence and geomagnetic activity. Most of the radio auroras
appeared in the afternoon and evening hours.
10.6.6 Conclusions
The radar-auroral clutter-to-noise ratio in the 5-30MHz
frequency range, as predicted from the SRI International
radar data acquired in Scotland, is dependent on the frequencydependencelaw of the auroral backscatterecho power.
Assuming a frequency variation of f-4 , the clutter level maximizes at approximately 10 MHz. When a f 2 law is assumed, the maximum clutter level is shifted to a frequency
of 30 MHz.
For an assumed f-2 dependence, the radar-auroral clutterto-noise ratio at 5 MHz is 15.6-dB less than that derived
on the basis of an f-4 law. Assuming that the radar aurora
is a point target, the clutter-to-noise ratio for a 400 us pulse
is estimated to be 3.6-dB greater than that deduced for
auroral scatterers that completely fill the antenna beam. For
a 10-us pulse, the corresponding difference is 19.6 dB.
An analysis of the HF ionospheric backscatter data obtained in Maine under the Polar Fox II program reveals that
the F layer cross section could be 10 to 20 dB greater than
the E layer cross section at 8.125 MHz and 20 to 23 dB at
14.875 MHz.
The median value of the angular extent of E layer, F
layer and combined E and F layer ionospheric clutter appears
to be on the order of 60.2 , 18.8 and 27.6 , respectively.
Extrapolating from the SRI International-Scotland UHF
radar-auroral data, it is estimated that for 5-MHz radar transmissions reflected from E layer auroral ionization, the
1-sigma value of the total Doppler frequency deviation, that
is, Doppler frequency shift and spread, could be on the
order of 13 Hz at the -6 dB normalized signal level,
+ 17 Hz at -12 dB, and + 19 Hz at -18 dB. At 30 MHz,
the 1-sigma Doppler deviations should increase to approximately + 78 Hz, + 105 Hz, and + 114 Hz, respectively.
HF field-aligned backscatter echoes usually occurs most
often near local sunset. However, experimental observations
have shown, at times, peak activity near the midnight hours.
There is a strong correlation of backscatter echoes with
the solar cycle. That is, the percentage of days displaying
backscatter reflections decreases with decreasing sunspot
number.
The characteristics of sunset echoes are found to be
different from those occurring during the night. The sunset
echoes are generally of the diffuse type, and are independent
of magnetic activity while the nighttime echoes are discrete
and correlate with magnetic activity.
10.7 SCINTILLATION ON TRANSIONOSPHERIC RADIO SIGNALS

A radio wave traversing the upper and lower atmosphere
of the earth suffers a distortion of phase and amplitude.
When it traverses ionospheric irregularities, the radio wave
experiences fading, phase deviations, and angle of arrival
variations. These signal fluctuations, known as ionospheric
scintillation, vary widely with frequency, magnetic and solar activity, time of day, season, and latitude.
The irregularities producing scintillations are predominantly in the F layer at altitudes ranging from 200 to 1000
km with the primary disturbance region for high and equatorial latitude irregularities between 250 and 400 km. There
are times when E layer irregularities in the 90 to 100 km
region produce scintillation, particularly sporadic E and auroral E.
10-71
CHAPTER 10
oo0
SCINTILLATION AND FADE DURATION ANALYSIS

60'
LOCALTIME
-I
GEOMAGNETIC
\
y\
9/ 95
753
TO
) No
S
PERCEUDENT-DWELL
Figure
le
SUNSET
6IM
j.I
10-64.
S
ample
fading
SECONDS
MET
ORDINATE
SGNXCEEDED
LEEL
are also shown.
trbution
-'3
intensity
Of
FAOEOdRT1o
TIME
9 SS
produc
WA
SUNRISE
-6
20"
----- -----IY)~50
216
DENSITY
OF
LOW
CROSSHATCHING
Moderate
solar During
activityEXCED.D
70
90,95
763
99
PERCENTOF TIME SIGNAPROPORTIONAL

TO
ORDINATTE
FADE DORATION
Io
SECONDS
Figure10-64. Sampleof intensityfadingproducedby signalpassingthrough

80-ib
DEPTHOF SCINTILLATIONFADING( PROPORTIONALTO
irregularities. Fade duration and cumulative probability distribution are also shown.
DENSITY OF CROSSHATCHING ) DURING

LOW
AND
MODERATE
SOLAR
ACTIVIIY
Figure 10-63. Globaldepthof scintillationfading(proportionalto density

of crosshatching) during low and moderate solar activity.
10.7.1 Global Morphology

From the global point of view there are three major
sectors of scintillation activity (Figure 10-63). The equatorial region comprises an area within 20 of the magnetic
equator.The high latitude region, for the purposes of the

scintillation description, comprises the area from the high
latitude edge of the trapped charged particle boundary into
the polar region. We shall term all other regions "middle

latitudes".
In all sectors, there is a pronounced nighttime maximum.
At the equator, activity begins only after sunset. Even in
10.7.3 Signal Characteristics

The amplitude, phase and angle of arrival of a signal
will fluctuate during periods of scintillation. The intensity
of the scintillation is characterized by the variance in received power with the S4 index commonly used for intensity
scintillation and defined as the square root of the variance
of received power divided by the mean value of the received
(I2) (1)2
power that is, S
()2
[Briggs and Parkin, 1963].
An alternative, less rigorous but simple measure of scintilation index has been adopted by many workers in the field
[Whitneyet al., 1969]for scalinglong-termchart records.

The definitionis
Px
SI
the polar region, there appears to be greater scintillation
occurrence during the winter months than during the months
(10.65)
mama
of continuoussolar visibility.
where Pmaxis the power level of the 3rd peak down from
10.7.2 Scintillation Examples
indicatedbelow.
The intensity fading can best be characterized by an

idealized example such as in Figure 10-64. The mean signal
level is modulated by the passage through the irregularities
so that the signal level very rapidly increases and decreases.
In Figure 10-64 the mean signal level at times fades below
the 3 dB level and below the 6 dB level. The number of
fades and the fade duration for a typical 15 minute length
of signal from a synchronous satellite is shown in Figure
10-64 along with the cumulative probability distribution
function (pdf). In this example the signal was above the 6
dB fade level 91.7% of time.
10-72
the maximum excursion of the scintillations and Pminis the

level of the 3rd peak up from the minimum excursion,
measured in dB [Whitney et al., 1969].
The equivalence of selected values of these indices is
S4
dB
0.075
0.17
0.3
3
6
0.45
10
Scaling of the chart records is facilitated by simply measuring the decibel change between the Pmaxand Pminlevels.

The phase variations are characterized by the standard deviation of phase,
over a given interval of fluctuation
frequency.
Attempts have been made to model the observed amplitude pdf. Whitney et al. [1972] and Crane [1977] have
constructed model distribution functions based upon the use
of the Nakagami-mdistribution(m = (S4)-2) and have shown
that the empirical models provide a reasonable approximation to the calculated distribution functions. Fremouw et
al. [1980] showed that the Nakagami distribution for intensity and the normal distribution for phase may be used to
characterize the statistics of the scintillation signal. In addition, the Rayleigh pdf provides a good fit to the data under
conditions of very strong scintillation (S4 - 1.0).
cartesian coordinate system (x,y,z), and a plane wave of

wavelength X traveling in the z-direction is incident on the
layer, then for weak scattering (rms phase 0o < 1) the spatial spectrum, 0)(Kx,Ky),
of phase (s) and log amplitude (X)
(Yeh and Liu, 1982) on the ground (z = z) is given by
2k sin K2 LA
K
k
Frequency
Dependence
quencies between 138 MHz and 2.9 GHz transmitted from

the same satellite, show a consistent wavelength (X) de5
pendence of the form A1.
of S4 for S4 less than about 0.6.
The frequency dependence becomes less steep for stronger
scintillation, as S4 approaches a maximum value near unity
with a few rare exceptions. When S4 exceeds 0.6 (peak to
peak values > 10 dB) the frequency dependence exponent
decreases. (If two frequencies are being compared and both
experience strong scattering to the extent that each displays
Rayleigh fading (S4 - 1.0), then there is effectively no
wavelength dependence over the frequency interval.) When
strong scattering occurs but is not constant over the frequency interval, the wavelength dependence is difficult to
determine. The observations [Fremouw et al., 1978] also
show that the phase scintillation index, , varies as X for
both weak and strong scattering. However, in extremely
strong scattering environment, the frequency dependence of
phase scintillation is also weakened.
K2
power spectrum of irregularity electron
the wavenumber of the radio wave
re
- the classical electron radius

(= 2.818 x 10-15 m)
- irregularity layer thickness.
(DN(K
KYK,)
In the presence of a relative motion between the propagation path and the irregularitiesthe spatialspectrum0(Kx,Ky)
is convected past the observer and a temporal variation of
signal phase and amplitude is observed. If the irregularity
structure does not change during the convection ('frozenin' hypothesis) and the irregularities have a uniform velocity
u in the x-direction then the power spectra of log amplitude
and phase in the frequency domain are given by
1 +
.U
10.7.5 Fading Spectra

During their passage through the ionospheric irregularities of electron density, radio waves from satellites undergo
spatial phase fluctuations. As the wave emerges from the
irregularity layer and propagates towards the ground, these
phase fluctuations cause interference to occur and a diffraction pattern in both intensity and phase develops on the
ground. In the presence of a relative motion between the
diffraction pattern and the observer, a temporal variation of
intensity and phase results. By transforming the temporal
pattern to the frequency domain, the frequency spectra of
intensity and phase fluctuations are obtained.
If a thin irregularity layer lies in the plane z = 0 of a
spatial wave numbers in x, y and
z directions
K:
Observations [Fremouw et al., 1978] employing ten fre-
(10.66)
the upper and lower signs within the bracket referring to X

and s respectively. In the above equation,
Kx,Ky,Kz
10.7.4
\ 2k /
KL
DX.
s ((K
f
x cosF-(z
KY)dKY
i(
)2k
(10.67)
(PAN,(,KOdK
Lkk
2/J
where
w-angular frequency of phase and amplitude fluctuation
2
q2 =
+ K.
u
The term within the bracket with the upper sign is known
10-73
CHAPTER 10
as the amplitude or Fresnel filter function and that with the
lower sign as the phase filter function. The Fresnel filter
function oscillates with the variation of frequency f and attains its first maximum at the frequency fF = u/ 2Xz .
The behavior of the phase filter function is very different
from the amplitude filter function as it fails to attenuate the
low frequency regime. Equation (10.67) provides a relationship between the irregularity power spectrum in the ionosphere and the amplitude or phase scintillation spectrum
obtained on the ground. Since the irregularity power spectrum has a power law variation [Dyson et al., 1974; Phelps
and Sagalyn, 1976] of the form K P, the power spectrum of
amplitude scintillation shows a maximum at the frequency
fF due to the Fresnel filter function. On the other hand, the
phase fluctuations are dominated by the low frequency regime. At f > fF, both amplitudeand phase scintillationspectra
show an asymptotic variation f-p when the three-dimensional irregularity spectrum is of the form K P. Thus from
a study of weak scintillation spectra the spectral form of the
irregularities causing scintillations may be deduced.
Figures 10-65a and 10-65b show two samples of weak
phase and amplitude spectra obtained from 244 MHz scintillation observations made at Goose Bay, Labrador by the
use of the Fleetsat geostationary satellite [Basu et al., 1982].
The phase and amplitude scintillation data were detrended
by a high pass detrend filter with a cut-off frequency of
0.0067 Hz and the data sample yields an rms phase deviation
of 2.4 radians and amplitude scintillation index S4 = 0.51
conforming to weak scintillation criterion. Both spectra show
an asymptotic variation in the high frequency region, the
amplitude spectrum showing f-2.35 variation, and the phase
spectrum showing f 2 .6 variation. If we consider that the
scintillation spectra have an average f-2.5variation, the corresponding three-dimensional irregularity power spectrum
is expected to have a power law wavenumber spectrum of
the form K-3. 5 . The decrease of power spectral density at
the low frequency end of the phase spectrum is caused by
the detrend filter. It may be noted that in the amplitude
scintillation spectrum the high frequency roll-off starts at
the Fresnel frequency of about 40 mHz. For an irregularity
layer height of 350 km, the observed Fresnel frequency
yields the irregularity drift velocity as 37 m/sec.
When scintillations become intense, the theory outlined
above does not hold, and strong single scattering as well as
multiple scatter theories appropriate to such cases have been
developed [Yeh and Liu, 1982 and references therein]. In
such cases it becomes difficult to relate in a straightforward
manner, the scintillation spectra to the irregularity structures
in the ionosphere. Figures 10-65cand 10-65d show a sample
of intense amplitude scintillation data and its spectrum
(S4 = 0.88) analyzed from 257 Hz scintillation data obtained at Ascension Island near the crest of the equatorial
anomaly by the use of transmission from the geostationary
satellite, Marisat [Basu and Whitney, 1983]. In contrast to
the weak amplitude scintillation spectrum shown earlier,
these spectra show a flat low frequency portion and increased
10-74
bandwidth of the frequency spectra, implying the development of shorter scales in the diffraction pattern. It is found
that the correlation lengths get progressively smaller with
increased strength of scattering [Rino and Owen, 1981; Basu
and Whitney, 1983]. Under conditions of strong scattering,
the phase scintillation spectra are also believed to suffer
from the refractive scattering from very large scale irregularities [Booker and MajidiAhi, 1981]. Since wave propagation through a strong irregularity environment has considerable systems applications, intensive work in this area
is in progress.
10.7.6 Geometrical Considerations

The intensity at which scintillations are observed depends upon the position of the observer relative to the irregularities in the ionosphere that cause the scintillation.
Keeping both the thickness of the irregularity region and
AN, the electron density deviation of the irregularity, constant, geometrical factors have to be considered to evaluate
data and to predict scintillation effects at a particular location. Among these are:
(a) Zenith distance of the irregularity at the ionospheric
layer. One study [Wand and Evans, 1975] found the intensity of scintillation related approximately to the secant of
the zenith angle up to 700; at angles >70 the dependence
ranges between 1/2and the first power of the secant of the
zenith angle.
(b) Propagation angle relative to the earth's magnetic
field. Performing this calculation demands the use of an
irregularity configuration and the consideration of Gaussian
or a power law model for the irregularities. At high latitudes,
irregularities in one study were elongated along the earth's
magnetic field with a cylindrical form of axial ratio of 5
along the lines of force. Sheet-like irregularities with forms
of 10 : 10 : 1 have also been found in recent auroral studies
[Rino et al., 1978]. For equatorial latitudes, this elongation
along the lines of force may be of the order of 50 to 100
[Koster, 1963].
(c) The distance from the irregularity region to the source
and to the observer (near the irregularities, only phase fluctuations are developed). As noted in Mikkelsen et al., [1978]
and Crane, [1977] the theoretical scintillation index can be
expressed in terms of the above factors when dealing with
ionospheric irregularities represented by a Gaussian power
spectrum.
Mikkelsen et al. [1978] have attempted to determine the
theoretical scintillation index, S4, when the irregularities are
described by a power-law power spectrum.
Mikkelsen assumed the approximate dividing line between weak and strong scintillation is -9 dB, with SI < 9
dB denoting the weak case.
The geometric variation of S4 is provided in Mikkelsen
et al. [1978].

i Ill
ll
[ IV 11 Ill
11
1 1111[
1 1I 1ll~ll
I I I1
I1
-20.
-26.8
P=
-2.60
II 1
20
T - 276
P
-235
10
10.
0.
-~~~~~~10
-10
257DAY 90 04549 -050059 UT

FS 50Hz
FC0.0066 Hz
T
0.00lO
lllI
IO.1
0.01
ll
DAY 90 045419 050059 UT

FS 50Hz
FC = 00-6.0+0H
-50
rl]
II[
TI
I I
[lllmI
Tl
0001
2158UT
0.1
lII
10
FREQUENCY-HZ
MIslandR.
257 MHZ
I lo
IOTi
(SE-HC)
FRE
122079MAR.
S410.88
UHF
10.88
20
257 MHz
TheASC.
I
2158 UT
I0
122079
S4=
n-6002established
-20
-50
-60
-20o
00rence
spread
andF
1 99Rastogi,
980]
that equatorial
159.is9817997
range
TIME (SEC)
c
that
nighttime
ionospheric
equatorial
irregularity
FREQUENCY- Hz
regions
Figure 10-65. Spectra of (a) phase scintillation and (b) intensity scintillation under weak scatter conditions at 244 MHz observed at Goose Bay, Labrador
on 31 March, 1979. (c) Data of 257 MHz intensity scintillation and (d) its spectrum under strong scatter conditions observed at Ascension
Island on 20 December, 1979.
10.7.7 Spread F and Scintillation
10.7.8 Equatorial Scintillations
The evidence from the corelation of scintillation occurrence and spread F [Rastogi, 1980] is that equatorial range
10.7.8.1 Patch Characteristics.

It has been established
that nighttime ionospheric equatorial irregularity regions
spread is associated with scintillation activity and frequency

spreadis not. Thus the available spreadF maps cannot be
used for scintillationobservationsin these regions; they are
dramatically misleading in many cases. In the high latitude
region no statistical study has been made to correlate types
of spread F with scintillation activity. It might be noted that
even range spreadoccurrence and scintillation have important differences.
emerging after sunset develop from bottomside instabilities,

probably of the Rayleigh-Taylor type. The depleted density
bubble rises into the region above the peak of the F2 layer.
Steep gradients on the edges of the hole help to generate
the smaller-scale irregularities within the patch which produces intense scintillation effects [Basu and Kelley, 1979].
A plume-like irregularity region develops, finally forming a patch of irregularities that has been likened to a banana
10-75
,o
> 000
km --
20
THREE
*C ,NS IOND
,o N.D
..
'IASCENSION
km
/(
DIMENSIONALE
ONC
GITU
O
PATCH MODEL
Figure 10-67.
Map of equatorial
regions using the 1975 epoch of the
DMA magnetic inclination map. X marks sub-ionospheric

intersection.
Three dimensional model of an equatorial irregularity patch

in the form of a banana or orange segment.
Figure 10-66.
HUANCAYO, PERU
80
0 ---60 -
NOV-JAN
20
NOV-JAN
20
12
18
24
12 LT
60m
12
24
18
12 LT
60
',
o
GUAM
GUAM
K=O-I
40
FEB-APR
FEB-APR
40-
20 -
20
I
LZ
40
12
18
I~
12
18
20 [Z -
12
(Kp
Z_7N
iI
_,_~
18
12 LT
~1_1
24
1'addtb60=3
-
z.LI
24
____I
12
40
12 LT
18
24
12
12 LT
MAY-JUL
18
24
12 LT
24
12 LT
9mantcodios60
20
.eI_.___I
24
and disturbed (Kp
6
3'
12 LT
- 9) magnetic conditions.U
12
18

or an orange segment. A cut through the center of the "banana" is shown in Figure 10-66.
The characteristics of the patch development, motion
and decay can be summarized as follows:
1. A new patch forms after sunset by expanding westward in the direction of the solar terminator with velocities
probably similar to that of the terminator. It comes to an
abrupt halt after typically expanding to an east-west dimension of 100 to several hundred kilometers. It appears
PERCENTOCCURRENCE
GREATER
THAN 2d8 (S4:SUNSET
(350 km)
SUNRISE
JUL
AUG
SEP
0CT
NOV
TEE
-0C
to have a minimumsize of -100 km.

2. It is composed of field-aligned elongated rod or sheetlike irregularities. The vertical thickness of the patch is 50
to several hundred kilometers. The patch has maximum
intensity irregularities in a height region from 225 to 450
km, with irregularities extending to over 1000 km.
3. Its north-south dimensions are of the order of 2000
km or greater.
4. Once formed, the patch drifts eastward with velocities
ranging from 100 to 200 m/s.
5. The patch duration as measured by scintillation techniques is known to be greater than 21/2h; individual patches
have been tracked by airglow techniques up to 3 hours where
they have maintained their integrity [Weber et al., 1978].
Effects have been seen over 8 h.
Studies of the variation in electron content in the patches
have been made by measuring the change in relative phase
between the two characteristic waves (ordinary and extraordinarymodes) with polarimeters. It is found that the patches
are regions of depletion in electron content. While the electron content depletions are found to be only of the order of
20%, the satellite in situ data may indicate density depletions
as large as two or three orders of magnitude at one fixed
altitude. In a strong irregularity environment, however, fast
fluctuations in polarization are often obtained. Lee et al.
[1982] have shown that scattering suffered by each characteristic wave may induce fast polarization fluctuations and
obtained expressions for the variance of these fluctuations
for irregularities with Gaussian and power law type spectra.
10.7.8.2 Variation of Scintillation Activity. A variety
of observatories used data taken over the same time period
to compare scintillation activity at 250 MHz [Aarons et al.,
1980b]. One set of data was taken at Huancayo, Peru, Natal,
Brazil, and Accra, Ghana with all observations made at
elevation angles greater than 20 and with distance between
the most separated stations about 70 of longitude; a map
of both geographical and magnetic coordinates is shown on
the right side of Figure 10-67.
A second comparison of data at 250 MHz was made
between observations from Huancayo and from Guam. The
data are shown in Figure 10-68; activity minima occur from
May-Julyin Huancayoand from November-Januaryin Guam.
The conclusion is that the occurrence patterns are longitudinally controlled.
It should be noted that in general maximum intensity
occurs in the equinoctial months. This can best be illustrated
,,
JAN
~0
FEB
FEB
,0\ ,o
MAR
APR
.MA
-
JUN
,
15
L
21
03
09
HUANCAYO
APRIL
76
15
LT
54 GH,
OCT 77
10-69
igurePercentage
occurrenceof 1.5 GHz scintillation 2 dB
by the occurrence of L band, 1500 MHz activity, at Huancayo, Peru. That evidence is shown in Figure 10-69 [Basu
et al., 1980b] L-band activity at Huancayo does not suffer
from strong scattering or from saturation (as does 136 MHz
and 250 MHz data on occasion); the data show clear equinoctial maxima.
From available data it appears as if geomagnetic control
of the occurrence of scintillation varies with longitude. The
generalization can be made that increased magnetic activity
inhibits scintillation activity before midnight--except duringthose monthswith very low scintillationactivity(May-July
for the region (00-70W) and November-January in the
Pacific longitudes (135-180E). After midnight the scintillation activity in general increases slightly with the presence of magnetic storms. The data shown in Figures 10-70
and 10-71 are for a year's observation in each case. The
complexities of the magnetic control of scintillation occurrence are illustrated by the variations in the curves of occurrence at each station in each season. For further details
see Mullen [1973].
10.7.8.3 In-Situ Data. Basu and Basu [1980] have developed a model from in situ, theoretical, and scintillation
studies. In their morphological model of scintillations, in
situ measurements of irregularity amplitude, AN/N, as computed from T sees of data are utilized in conjunction with
simultaneous measurement of electron density N. A combination of AN/N and N data provides the required AN
parameter as a function of position and time.
10.7.8.4 Sunspot Cycle Dependence. From the viewpoint of electron density variations, the equatorial region
around the magnetic equator displays a complex pattern.
During the day an increase in maximum electron density
occurs away from the equator. The electron density contours
10-77
CHAPTER 10
---
K= O- I
K= 3+-9
FEB. - APR.
---
.....
0- I
of
60-
in,
'40
257 MHz
60
20
24
12 LT
<'/\2
12
18
NATAL, BRAZIL
'
I
60
40
,-
20
lB
24
12 LT
NATAL, BRAZIL
12
18
24
12 LT
80 _
80
60
60
60
HUANCAYO,PERU
~40
_
20
2
12
18
24
20
12 LT
'A
HUANCAYO,PERU
~40
12
18
24
display a distinct trough of electron density in the bottomside

and topside ionosphere at the magnetic dip equator with
crests of ionization at 15-20 north and south dip latitudes; this is the Appleton anomaly with the region within
5 dip latitude of the magnetic equator termed the electrojet region.
From the solar cycle minimum in 1974 and maximum
in 1969-1970, Aarons [19771found that there was a higher
occurrence of deep scintillations during a year of high solar
flux than during a year with low solar flux for observations
at both Accra, Ghana and Huancayo, Peru.
Recent observations of L band scintillations from both
MARISAT and GPS during the period of maximum solar
flux (1979-1981) [DasGupta et al., 1981] have revealed
that scintillation intensities maximize in the Appleton anomaly region rather than near the magnetic equator.
The conclusion in the study [Aarons et al., 1981a] was
that the intense scintillation activity during years of high
solar flux are due to two factors:
12 LT
Figure 10-70. Seasonal patterns of occurrence of scintillation activity > 6 dB (S4 = 0.3) at 250 MHz for very quiet (Kp = 0- 1+)
+
(Kp = 3 - 9) magnetic conditions for Nov-Apr.
10-78
12 LT
20 -
12
.
_I
wu
24
'
60 40
0o
ACCRA, GHANA
ACCRA, GHANA
20 -
18
257
'
"
40-
80
80
40 -
12
^A
NOV. - JAN.
and for disturbed
1. The equatorial anomaly has considerably higher electron density values in high sunspot number years than in
years of low solar activity,
2. The occurrence of maximum electron density for
anomaly latitudes is near sunset in the years of high sunspot
number and in the afternoon in years of low solar activity.
Thus the post sunset irregularity patches attain high AN
levels in the years of high solar flux.
10.7.9 Middle Latitude Scintillation

The middle latitude scintillation activity is not as intense
as that encountered at equatorial, auroral or polar latitudes.
However, activity may reach levels, primarily at VHF and
UHF, that will increase error rates of systems with low fade
margins. The reader is referred to Aarons [1982] and Bramley [1974].

---K=3t 9
K =OIK 3t-9
AUG. - OCT.
MAY - JULY
60
60-
257 MHz
257MHz
40
40 -
ARAGHANA
ACCRA,
12
24
18
12 LT
12
18
24
12 LT
60
60
60
NATAL, BRAZIL
z 40
40
NATAL, BRAZIL
40
20 -
20-
ACCRA, GHANA
20 -
20
\
GHANA
60
40 12
40.
20
12
24
18
..
18
PERU
HUANCAYO,
12 LT 40
6
HUANCAYO,PERU
18
24
20 -
24
40 12
12 LT
12
18
Figure 10-71. Seasonal patterns ofoccurrence of scintillationactivity > 6dB (S4 = 0.3) for very quiet (Kp
magnetic conditions for May-Oct.
10.7.9.1 Effect of Magnetic Index on Midlatitude

Scintillation. At latitudes below the auroral oval, various
sets of data have yielded behavior indicating little correlation
with magnetic conditions. Evans [1973] found no correlation of their 400 MHz radar scintillations with magnetic
index south of their station at 560 invariant latitude. Aarons
and Martin [1975] found that during the August 4-10, 1972
magnetic storms there was a negative correlation of scintillation and magnetic index for Athens, Greece and Camp
12 LT
24
0=- 1+ ) andfordisturbed (Kp
12 LT
3+
9)
10.7.10 The High Latitude Region

Figure 10-72 depicts the intensity of scintillation at high
latitudes in a very broad manner for the period of time
around midnight.
10.7.10.1 The Plasmapause and the Trough. The present evidence is that there is a boundary at the high latitudes
where weak irregularities commence. It is probably a few
Parks, California and little correlation at the 450 invariant
degrees equatorwards of the plasmapause, between 45-55
latitude intersection for Aberystwyth, Wales. Bramley [1974]

found that except for the December 1971 magnetic storm
(when the irregularity region probably encompassed the intersection point of -45), there was no correlation between
magnetic activity and scintillations.
This type of data essentially corroborates the early radio
star obvservations in the U.K. which found little correlation
with magnetic index except in paths to the north (with the
exception of some intense magnetic storms).
Corrected Geomagnetic Latitude (CGL), the system used at

high latitudes in this review.
At night a trough or region of low electron density and
total electron content exists between the end of normal ionospheric plasma behavior and the auroral region where energetic electron precipitation and current systems are dominant factors in producing both the normal ionospheric layers
and the irregularities.
All observers of irregularities see a dramatic change in
10-79
CHAPTER 10
6O0km
60mSUNS
AURORAL
OVAL
PLASMAPA-USE
400km
pOLAR
HIGH SOLAR FLUX
'
JAN
FES.
/'2
wSOL A
LOW '# -"-,"
*6b0
800
700
SURISE
9)
NOV
DEC
MAR APR
MAY
JUN
CORRECTED GEOMAGNETIC
LATITUDE
JUL
Figure 10-72. Depiction of high latitude irregularities -22-02 LT. Sheetlike irregularities are seen in the auroral oval, rod irregularities at higher and lower latitudes.
SEP _
AUG
r-
OCT -
21
15
CGM
09
irregularities in the auroral oval at the poleward edge of the
MEAN
SCINTILLATION
INDEX
(dB)
trough. In the auroral oval, the intensity of scintillations is
NARSSARSSUAQ
1968-1974Kpz4-9
15
LT
a function of local magnetic activity. Poleward of the aurora

Figure 10-74. Contoursof monthlymean scintillationindexin dB at 137
MHz as a functionof local time for disturbed (Kp =
there may again be a lowering of scintillation activity until

the observing path transits the polar region [Aarons et al.,
1981b]
10.7.10.2 Auroral Scintillations. From studies of radio

star and low altitude satellite scintillations, a series of height
measurements have pointed to F layer heights as the primary
seat of the irregularities producing the signal fading.
Maximum irregularity intensity appears above the region
showing maximum intensity aurora [Martin and Aarons,
1977]. Vickrey et al. [1980] have shown that there is a
collocation of scintillation patches in the auroral oval and
F region ionization enhancements.
SUNS
JAN
Perhaps the most consistent studies of long term behavior of scintillations have been made in the auroral zone, at
Alaskan longitudes and along the 70oW meridian.
Both the diurnal pattern of scintillation activity and the
seasonal behavior as observed from one site can be noted
in Figures 10-73 and 10-74. The data used for this long
term study [Basu and Aarons, 1980] were taken over a
period of 6 years from Narssarssuaq by observing 137 MHz
scintillations of the ATS-3 beacon; the propagation path
traversed the ionosphere at -63o CGL.
The long term study used for Figure 10-75 incorporated
data from three observatories (Narssarssuaq, Greenland; Goose
SUNRISE
DEC_NOV
-Bay,
\
DEC
9) magnetic conditions obtained at Narssarssuaq dur-
ing 1968-1974.
Labrador; and Sagamore Hill, Massachusetts). The

contours are of reduced data for one season (May-July) for
magneticallyactive periods of time [Basu and Aarons, 1980].
The boundary of active scintillation is pushed equatorwards
FEB -
extending into what was the quiet trough and plasmapause

latitudes. Thus during magnetic storms scintillations and
MAR
APR
0MAY -I
JUN -
|D65' MAY
JUL
JUL
_
_
/g
/.
THRUdULY
....
.
<
S~~p~l(~~
I ""-
60<-
>/d
6001~//
NARSSARSSUAO
\V60'-
AOOSEBA
OCT
55,
-15
21-121
9
+GM
09
03
MEAN SC NTILLATION
15---L_6
INDEX (dB)
SAGAmORE
HILL
oIo
00
2
MEAN Sl(dB)
-8 2A
27 PTS
24 LT
Kp 4 9
NARSSARSSUAO1968-1974 Kp= 0-3
Figure 10-73. Contours of monthly mean scintillation index in dB at 137

MHz as a function of local time for quiet conditions (Kp =
0 - 3) obtained at Narssarssuaq during 1968-1974.
10-80
Figure 10-75. Variation of mean scintillation index during the northern

solstice in dB at 137 MHz with local time and invariant
latitude derived from hourly data at the 3 stations under
disturbed magnetic conditions (Kp = 4 - 9).

GEOMETRICAL
ENHANCEMENTFACTORFOR rms PHASE(8:8:1)
~~0
~ ~ ~ ~ 1~
1~|~
I~Ii
_
|
l
POKERFLAT
S~
[1~~~~
I7
3 -
1977- 1978
DIP LAT - deg

YUKON
Figure 10-78. S4 at 50sti exceedance level vs magnetic latitude for daytime

data during 1977-1978.
-F
6
G
_ 1~~~~~
4-_
\I /
55
60
2/ J.'
\\
ture-at least as observed from Alaska. Figure 10-78 illustrates the daytime increase with increasing latitude.
65
10.7.10.3 Polar Scintillations. A long term consistent

series of measurements has been taken at Thule, Greenland
with observations at 250 MHz [Aarons et al., 1981b]. The
/11
70
DIPLAT-
75
80
scintillations for this study ranged from very low values of
deg
3-6 dB peak to peak on occasion during a period of low

sunspot number to saturation fading of 28 dB peak to peak
for hours during winter months of years of high sunspot
Figure 10-76. Model computations of phase geometrical enhancement
factorfor sheet-likestructureswithan 8 : 8 : 1 anisotropy.
Becauseof the meridionalpass trajectory,the locationof
number.
One set of measurements was taken between April and

October 1975. During this period of low solar activity, there
the enhancement is independent of the pass elevation.
was an absence of strong scintillation activity to such an

extent that only the occurrence of scintillation greater than
[Rinoand Owen 19801
6 dB could be plotted. Figure 10-79 shows the contrast

optical aurora can be noted farther south than 55 . In the
between the 1975 period when solar flux was low (10.7 cm
70W longitude region this extends below the latitude of
flux was -75) and the same months in 1979 when the solar
Boston.
flux was high (150-225).
Geometry and Enhancement. Sheet-like irregularities
produce strong enhancements when observations are made
A contour plot of the percent occurrence of scintillation
index greater than 10 dB is shown in Figure 10-80. The
in specific directions. For two sites in Alaska, Rino and
Owen
1980]
haveconstructed
plot
thetheoretical
was developed
geometrical
from hourly average values of the 15
Owen [1980] have constructed the theoretical geometrical
for
each
month
for low magnetic activity (Kp = 0minute
SI
enhancement factor for rms phase fluctuations for an 8:8:1
3). Two patterns emerge: (1) Maximum occurrence of acirregularity (Figure 10-76) [Rino and Matthews, 19801.They
found this enhancement in phase fluctuations as can be seen
tivity takes place in the months of little or no sunlight at F
by the data in Figure 10-77. The amplitude enhancement,
-80 ------.
less dramatic but present, is also shown in Figure 10-77.
Daytime scintillation does not show the sheet-like strucTHULE
70
_~_
L ------_
760-_
l,
i 2o
____197850
lo_
|=
APR OCT 1979
5.
/\1977
\_r
--
K 3 -9
430
1020
7
20-
o7
iso
/iiJ
5J--
L___
"2400
)
40b.t-
/-,
'
APR-OCT
oI,r
975
200
Figure 10-77. RMS phase and S4 at 50% exceedance level vs magnetic

latitude for nighttime data during 1977-1978.
Figure 10-79. Percentage occurrence of scintillation greater than 6 dB for

low solar fluxperiod April-October 1975is contrasted with
that for high solar flux period April-October 1979 for both
quiet and disturbed magnetic conditions.
10-81
CHAPTER
10
PERCENTOCCURRENCE
GREATERTHAN lOdB
PERENTOCURRNC
GREATERTHAN1
of subvisual F layer (X = 6300
(saturated) amplitude scintillation at 250 MHz as the arcs

drifted through a satellite to ground ray path. Outside the
arcs, scintillation frequently persisted at a lower level (SI
- 6 dB).
A pictorial representation of both the small scale antisunward irregularity drift and the patch motion (predomidawn to dusk) is shown in Figure 10-81 (E. Weber,
-NO
i\DEC W NTER
h -2--0----
-JAN
e-
EABFA
a:2
HA
9 /L \)~20
~JUNSUMMER
ti;)2A
TA
MAR APRlNG
APR
MAY p20
R
cL
Bnantly
SAMER
40N
'"--'" ~~
5 NoOL
LCG
03
THULE 1979
K =0O 3
AJUL
_eoarcrate
SC_
......
24.0. ,
..CG 200
9
15
UT
Figure10-80. Contourplot ofdiurnalpatternof monthlypercentoccurrenceof scintillation

greaterthan10dB forlowmagnetic
activity(Kp - 0 - 3). Observations
weretakenduring
region heights. Much lower scintillation occurrence takes

place in the sunlit months. (2) The diurnal variation is weak,
and apparent only during the winter months.
Auroral arcs in the polar cap are approximately aligned
with the noon-midnight magnetic meridian [Davis, 1962].
These arcs generally drift in the dawn to dusk direction
[Danielson, 1969]; however, reversals have been noted
[Akasofu, 1972; Weber and Buchau, 1981]. Recently Weber and Buchau [1981] described the orientation and motion
12
75
/?
:>g80nM
A\
p
/j K 3/Q' Ad
X E \power-law
/// Xj a,, ,\>,/,
\
S
W
1~ ~~8
>1
//
8, ,,
3," a
855
f;5
//
,,,
\ 2\./ t
\5, c80/
private communication). Results point to two irregularity

components in the polar cap; antisunwarddrifting irregularities which produce a background level of weak to modeAL scintillation and intense irregularities within F layer
polar cap arcs that produce more discrete (-1 h duration)
intense scintillation events as the arcs drift through the ray
path.
10.7.11
Empirical Model Of Global

Scintillation
Mar 1979-Feb 1980.
/?
'
The descriptive irregularity model is based on numerous

observations[Fremouw and Bates, 1971;Fremouw and Rino,
1978], but most particularly on observations of phase scintillation performed in the DNA Wideband Satellite Experiment [Sagalyn et al., 1974]. The most significant caveat
about use of WBMOD, however, is that it has been cali-
10.7.11.1 WBMOD. Over a period of years, starting from

available data and from weak scintillation theory, a model
of scintillation termed WBMOD has been developed [Fremouw and Bates, 1971; Fremouw and Lansinger, 1981;
Fremouw and Rino, 1978, 1976;Fremouw, 1980; Fremouw
et al., 1977]. The program provides for phase and amplitude
information. Input user parameters include frequency, location, local time, sunspot number, and planetary magnetic
index, Kp. The user also must specify the longest time the
system needs phase stability. Scintillation indices are the
output. A model of the irregularity drift velocity is contained
in the program.
Program WBMOD permits a user to specify his operating scenario. The code returns the spectral index p for
phase scintillation, the spectral strength parameter T, the standard deviation v<1, of phase, and the intensity
scintillation index, S4, as functions of a changing indepen-
dent variablechosen by the user.
C,, ,3
,'
Behavior
.
-6
'vz
A O) polar cap arcs. Kil-
irregularitieswithin the arcs producedintense
:ometer-size
SUN-FISE
SUNSET
LAT.
brated quantitativelyagainst Widebanddata from only a
~~~~~~~75
o~
~The
DAWN-DUSK
ARC DRiFT
Figure 10-81. Schematicof smallscaleanti-sunwardirregularitydrift and
the patch motion.
10-82
single station in the northernauroral zone (Poker Flat, Alaska).

descriptive model was developed by iterative comparison with most of the Wideband data population from Poker
Flat, with a portion of the population reserved for final
comparative tests.
The basic calculations are made of two central quantities
T and p. T is the spectral strength of phase at a fluctuation
frequency of 1 Hz, p is the power-law spectral index of

phase; T is highly variable, unlike p. The program calculates
T and p and the two commonly used indices of scintillation
activity based on them, one for phase, oo, and one for
intensity, S4.
In order to calculate T, p, oo, and S4, one must have
values for eight parameters describing ionospheric irregularities. They are (1) the height, h; (2) vector drift velocity,
Vd, of the irregularities; (3) an outer scale, a; (4,5,6,7) four
"shape" parameters describing the irregularities' three-dimensional configuration and spatial "sharpness", a, b, ,
and v; and (8) the height integrated strength of turbulence,
CsL. Program WBMOD contains models for the foregoing
eight parameters, but the degree of detail is very much less
for some than for others.
The most variable and the most important of the eight
is the height-integrated strength of turbulence, CsL. The
irregularity strength is modeled by
more limited than WBMOD as (1) they are applicable only

for the frequency of the data base, 137 MHz, (2) there is
an equipment-biased limited excursion of the scintillations
and, (3) these data have an implicit dependence on the
geometry of the observations, namely, observing ATS-3
from the stations detailed above. This does not permit other
viewing geometries or taking into consideration the configuration of the irregularities unless correcting factors are
included.
With these caveats, the equations for each station are
Narssarssuaq (63 CGL intersection)
SI (dB) = -6.4
+ 9.2(1 - 0.2FD)1 + 0.23(1 - 0.3FD)
x cos (HL + 2.0 + 0.34Kp) + 0.03

xcos [2(HL - 0.6)] + 0.02 cos [3(HL
+ 3.0)]}2ro 14Kp(
CsL
E(Xm,XgT,D,R) + M(X,,,T)
4. H(Xm,Tm,Kp,R) (10.68)
+0.12FD)+0.09As(l
+ 176FD)]
FD = cos (DA + 15.6) + 0.56 cos [2(DA - 22.4)]

(10.69a)
where X,, = geomagnetic invariant latitude,

Ag = geographic latitude,
T = local meridian time,
Goose Bay (60 CGL intersection)

SI(dB) = -1.3
+ 1.1(1 - 0.77FD){I + 0.5(1 - 0.2FD)
D = day of the year,
x cos(HL + 2.1 - 0.6Kp) + 0.06
R = smoothed Zurich sunspot number,
x cos [2(HL - 2.1)] + 0.02 cos [3(HL
Tm = geomagnetic time,
Kp = planetary geomagnetic activity index.
The three terms in Equation (10.68) respectively describe
the strength of equatorial, midlatitude, and high-latitude
irregularities. The first two have not been tested extensively
against Wideband data but the high latitude term H has been.
The high-latitude term is based on the observation that
there often is a more-or-less abrupt boundary [Aarons et al.,
1969] between the midlatitude region of relatively smooth
ionosphere and the highlatitude scintillation region. It is
located, typically, equatorward of discrete-arc auroras in
the general vicinity of the diffuse auroral boundary.
+ 5. 2)]}2 10o3Kp(l
+0.FD)+O8As(l+ 2FD)]
FD = cos (DA + 0.5) + 0.2 cos [2(DA - 99)]
(10.69b)
Sagamore Hill (53 CGL intersection)
SI(dB) = 0.33 + 0.02(1 + 0.2FD) {1 + 1.2(1 - 0.01FD)
x cos (HL - 0.4 - 0.15 Kp) - 0.3
x cos [2(HL - 0.8)] - 0.1 cos [3(HL
+ 6. )]}[038Kp(1+03FD)+3
IAs(l-02FD)]
2
FD = cos (DA + 56) + 0.7 cos [2(DA - 143)].
(10.69c)
10.7.11.2 Formulas In Atlantic Sector. Since WBMOD

has been developed and calibrated against data from only
one longitude sector (Alaska), it is appropriate to note empirical formulas that, though not as complex, have been
developed for another longitude sector, along the 70W
meridian. These formulations have been made [Aarons et
al., 1980a] for Narssarssuaq, Greenland, Goose Bay, Labrador and Sagamore Hill, Massachusetts based on 3-7
years data base of 15-min scintillation indices. The forcing
functions are time of day, day of the year, magnetic index
and solar flux. However, these individual models are much
DA is day number, As = Sf/100, HL is local time (hours)

at subionospheric point (350 km), and Sf is solar flux at
2695 MHz in solar flux units; all angles are in radians.
Arguments of the cosines with diurnal and yearly terms
should be converted by factors of 2ii/24 and 2ii/365, respectively.
In Aarons et al. [1980a] corrections for frequency dependence are given thus allowing higher frequency scintillations to be estimated. In addition, corrections for geometry
are also given similar to those cited in Section 10.7.3.1.
10-83
CHAPTER 10
10.8 IONOSPHERIC TIME DELAY
EFFECTS ON EARTH-SPACE
PROPAGATION
10-5
One of the most important effects of the ionosphere on

radio waves that traverse it is a retardation, or group delay,
on the modulation or information carried on the radio wave,:
due to its encounter with the free, thermal electrons in the
earth's ionosphere. Other effects the ionosphere has on radio
waves include (1) RF carrier phase advance (2) Doppler
shift of the RF carrier of the radio wave (3) Faraday rotation
of the plane of polarization of linearly polarized waves (4)
angular refraction or bending of the radio wave path as it
travels through the ionosphere (5) distortion of the waveform
of transmitted pulses, and (6) amplitude and phase scintillation. With the exception of scintillation effects (see Sec-
o
1
10-7
0-
Z 10-9
i
,
tion 10.7), all the other effects listedhere are proportional,

at least to first order, to the total number of electrons encountered by the wave on its passage through the ionosphere
or to their time rate of change. In fact, phase scintillation
also is merely the short term, time rate of change of total
electron content (TEC) after the longer term variations have
been removed.
In this section a short description is given of each ionospheric TEC effect upon radio waves, along with a representative value of the magnitude of each of the these effects
under normal ionospheric conditions. This is followed by a
discussion of the important characteristics of average ionospheric TEC behavior and the temporal and spatial variability of TEC.
\/o
io0MHz
200
300 400 500
IGHz
4 5
FREQUENCY
Figure 10-82. Time delay vs frequency for various values of TEC.
km! Obviously, the TEC parameter is of potentially great

importance to precision satellite ranging systems.
10.8.1.1 Two-Frequency Ionospheric Time Delay
Corrections. If the navigation or ranging system bandwidth is large enough so that two, fairly widely spaced bands
can be used for ranging, the ionospheric time delay error
can be reduced to an acceptable level automatically and can
be made transparent to the system user. Because the ionospheric time delay is a function of frequency we can write:
10.8.1 Group Path Delay

k
The additional time delay, over the free space transit
time, of a signal transmitted from above the ionosphere to
a user on, or near, the earth's surface is given by
At = 40.3
cf2
TEC (s),
(10.70)
where the TEC is the total number of electrons along the

path from the transmitter to the receiver, c is the velocity
of light in m/sec., and f is the system operating frequency
in hertz. The TEC is generally expressed as the number of
electrons in a unit cross section column of one square meter
area along this path.
A plot of time delay versus system operating frequency
for TEC values from 1016 to 1019el/m2 is given in Figure
10-82. These two values represent the extremes of observed
TEC in the earth's ionosphere. Note that, at a system operating frequency of I GHZ, for example, a TEC of 1018,
a value frequently exceeded in many parts of the world,
would produce a time delay of 134 ns or 40.2 m of range
error. At a system operating frequency of 100 MHz this
same TEC values would produce a range error of over 4
10-84
At
k
x TEC, t2
= x TEC,
(10.71)
where At1 is the ionospheric error on frequency f1, and At2

is the ionospheric error on the frequency f2. If the normal
system operational frequency is f1 and we choose f2 at a
lower frequency for ionospheric correction purposes, we
obtain:
or:
b(At) = - x TEC (I/f2 - 1/f,)

c
= At, (f2 - f2)/ f2
(10.72)
At, = f2/(f, - f2) x 8(At)
The value (At) is obtained from the difference of the simultaneous measurements of the total range, including ionospheric time delay, at the two frequencies, f1 and f2, since
the geometric distance is, of course, the same at all frequencies. The quantity f2/ (f, - f2) is called the ionospheric
scaling factor. For ratios of f2/f1 near unity, the required
precision of the differential measurement may be unreasonably large. A plot of this quantity, normalized by fl, is given
in Figure 10-83. In this derivation the contribution of re-

assuming the satellite transmitted modulation phase at L1
and L2 is known and the receiving system frequency dispersive characteristics can be independently measured and
corrected
e
for.
For a typical daytime high solar activity TEC value of
/10 el/m2 column the 8(At) measured by a GPS receiver
would be 35 ns or 10.5 m of ionospheric error. For a direct
measure of absolute TEC from the modulation phase delay
at L2 minus L1 we have
6!
5-
I/
24t/
<3
---
03
~0.4
TEC = 2.852 x 1016x 6(At),
05
06
07
08
RATIOOF SECONDARY
FREOUENCY f2
0.9
f1.0
PRIMARY
scalingfactorvs ratioof primary(higher)to

Figure10-83.Ionospheric
secondary
(lower)frequency.
ceiver noise to the differential measurement accuracy has

not been considered.
10.8.1.2 An Example of a Two-Frequency Ionospheric
Time Delay System. The Department of Defense is currently testing an advanced navigation system, called the
NAVSTAR-Global Positioning System (GPS), [Demaro,
1981; Milliken and Zaller, 1978] which uses coherently
derived, identical modulation on two carrier frequencies,
called L1 and L2, to measure the ionospheric group path
delay directly and thereby correct for ionospheric time delay. The ratio of frequencies used in the GPS system is
exactly 154/120, with the higher frequency (L1) at 1575
MHz. The two carrier frequencies transmitted by the GPS
system are the 154th and 120th harmonics of 10.23 MHz.
This 10 MHz frequency is bi-phase modulated on both carriers with a psuedo random code resulting in a [(sin x)/x ]2
shaped spectrum of width 20 MHz to the first nulls. A user
with knowledge of the transmitted code collapses the received spectrum to equivalent carriers with 10 MHz modulation. The 10 MHz modulation is transmittedwith a known
phase difference on the two carriers, and the received modulation phase difference is a direct measure of the ionospheric group path delay.
For the GPS carrier and modulation frequencies the ionospheric group path delay at frequency L1, as obtained from
Equation 10.73 is
At = - 1.5457 8(At),
(10.73)
where 8(At) is the difference between the ionospheric time

delay measured at the two frequencies. This difference in
range is directly related to absolute ionospheric time delay
as, of course, the satellite is at the same range at both
frequencies. The only frequency dependent parameter in
range measurements is the ionospheric time delay effect,
(10.74)
where 8(At) is measured in nanoseconds (ns). Since, at

10.23 MHz, one complete cycle of modulation phase of
3600 is 97.75 ns, we obtain TEC = 0.7745
x 1016 el/m 2
per degree of 10.23 MHz modulation phase difference, or:

TEC = 278.8 x 10 ( 16
el/m2 per cycle of modulation phase
difference. Thus, the cycle ambiguity in absolute values of
TEC is trivial to resolve using the GPS system as a means
of determining ionospheric time delay.
Absolute ionospheric time delay measurements can be
made with an accuracy approaching I to 2 ns, depending
upon the received signal to noise ratio on both frequencies.
For the power levels transmitted by the GPS satellites, an
omnidirectional receiving antenna, and a receiver with a
modulation tracking bandwidth of approximately 15 Hz, the
differential modulation phase has been measured to within
approximately +-2 ns. The contribution of receiver noise
for the two-frequency ionospheric time delay corrections on
the GPS system has been considered by Cretcher [1975].
10.8.2 RF Carrier Phase Advance

In addition to group path delay, or modulation time
delay, over the free space delay, the phase of the carrier of
radio frequency transmissions is changed by the ionosphere.
The RF phase is advanced with respect to its phase in the
absence of an ionosphere. This effect is extremely important
in determining space object velocities by means of range
rate measurements. The amount of phase increase or phase
path decrease can be expressed as
1.34x 10
f
TEC (cycles),
where f is the system operating frequency in hertz, and TEC

is in el/m2 column. In practice, the amount of this phase
advance cannot readily be measured on a single frequency
and two, coherently derived, frequencies are required for
this measurement.
10.8.2.1 Differential Carrier Phase. In addition to the
dual frequency identical modulation transmitted from the
GPS satellites for ionospheric group path correction, these
satellites also transmit two, coherently-derived carrier fre10-85
CHAPTER 10
quenciesfor ionosphericdifferentialcarrierphase measure-
A2
ments. For the pair of frequencies used by GPS, approximately 1.2 and 1.6 GHz, the differential carrier phase shift,
referenced to the lower frequency, is
from Equation (10.75)
hA4 =
1.34
L- x
10 - 7 X
4=
(4(u-
(m2 - 1)
x TEC (cycles),
0)
- (c-
1.34
k4L)=
10 - 7 X
3 x
,, +
2 x k,
TEC (cycles)
(10.76)
thus
wherem = f1/f2 = 1.2833, Ao = 4.31 x 10 17 X TEC

or 2.32 x 1016el/m2 per complete 2ii cycle of differential
carrier phase between Ll and L2, measured at L2. The
differential carrier phase [Equation (10.76)] is related to the
differential modulation phase, [Equation (10.72)] simply by
the ratio of carrier to modulation frequencies. With a reasonable carrier signal to noise ratio, this differential carrier
phase can be measured to within a few degrees, or less than
approximately 0.04 x 1016el/m2. Since the TEC is generally much greater than 2.32 x 1016, corresponding to 2ii
of differential carrier phase, there is a 2nrr ambiguity in the
differential phase measurement.
The differential carrier method of measuring TEC cannot, in practice, be used to measure absolute values of TEC
by itself due to the large 2nii phase ambiguity in the measurement, but this is not important for navigation systems
which require a correction only for range rate errors due to
the ionosphere between two measurement times,
The US Navy Navigation SatelliteSystem, NNSS [Black,
1980; Kouba, 1983], determines position for stationary and
slowly moving vehicles by measuring satellite transmitted
RF carrier phase changes as a function of low-orbit satellite
motion across the sky. This method of positioning requires
only range rate information. The primary NNSS frequency
is 400 MHz. A second RF carrier at 150 MHz is used only
for ionospheric range rate corrections. While various techniques have been proposed for determining the absolute TEC
from the differential carrier phase information received from
the NNSS satellites, they all involve assumptions concerning some a priori knowledge of the ionosphere, and they
cannot be used in the general case.
As an ionospheric monitoring tool the combination of
differential carrier phase and differential modulation phase
provides an excellent means of determining ionospheric
electron content along the ray path to the satellite. The
absolute value of TEC can be determined by the group delay
technique and relative TEC changes can be measured with
great accuracy by the differential carrier phase technique.
10.8.2.2 Second Difference of Carrier Phase. The second difference in phase between an RF carrier and that of
its upper and lower sidebands can be used to measure absolute values of TEC, as described by Burns and Fremouw
[1970]. If three coherently derived frequencies, f- fm,, f,
and f+ f,,,are transmitted the second difference of phase is
given by
10-86
2.68 x
=2.
f(f2
TEC (cycles)
TEC (cycles).
0
f2 )
When
f2
>>
f2
(10.77)
2.68 x 10-7f2m
A2 4
TEC (cycles).
For a carrier frequency of 100 MHz a modulation frequency of 1.93 MHz would be required to give 2ii of second
differential phase for a TEC value of 1018el/m.2 A value
of A2 of 2ii for 1018el/m2 is a reasonable compromise
between the requirement for minimizing chances of an ambiguity in absolute TEC and accuracy in measuring TEC
relative changes. The second difference of carrier phase has
been used with the DNA-002 satellite to make estimates of
the absolute value of TEC [Freemouw et al., 1978].
10.8.3 Doppler Shift

Since frequency is simply the time derivative of phase,
an additional contribution to geometric Doppler shift results
due to changing TEC. This additional frequency shift is
generally small compared to the normal geometric Doppler
shift, but can be computed by
Af
TEC (Hz).
dt
(10.78)
dt
For high orbit satellites where the diurnal changes in

TEC are greater than geometric ones, an upper limit to the
rate of change of TEC is approximately0.1 x 1016el m-2s - 1.
This value yields an additional frequency shift of less than
0. 1 Hz at 1.6 GHz which would not be significant compared
with a typical required receiver loop bandwidth of at least
a few hertz. At 400 MHz a similiar rate of change of TEC
would produce a frequency shift of approximately 0.3 Hz,
probably still not significant.
During times of severe phase scintillation, which can
occur even at GHz frequencies, the TEC likely does not
change in a consistent, rapid manner to yield greater ion-

ospheric Doppler shifts, but the phase of the incoming RF
signal can have a large random fluctuation superimposed
upon the changes associated with the normal rate of change
in TEC. This large, random component may actually spread
out the spectrum of the received signal sufficiently to cause
the receiver to lose phase lock, as the receiver signal phase
may have little energy remaining in the carrier, and instead
may be spread over several Hz, with little recognizable
carrier remaining. A knowledge of phase scintillation rates
is required to determine the spread of received signal phase.
looo
a 100
z
l
/0 '
0
<
I.0
\-
10.8.4 Faraday Polarization RotationWhen a linearly polarized radio wave traverses the ionosphere the wave undergoes rotation of the plane of polar-00
ization. At frequencies of approximately100 MHz and higher
the amount of this polarization rotation can be described by:
- s
2.36 x 10
= 2
f B cos ONdl.
500
IGHz
Figure 10-84. Faradaypolarizationrotationvs frequencyfor various values of TEC.
(10.79a)
where the quantity inside the integral is the product of electron density times the longitudinal component of the earth's
magnetic field, integrated along the radio wave path. Many
ionospheric workers have used this effect, named for Michael Faraday who first observed polarization changes in an
optical experiment, to make measurements of the TEC of
the ionosphere. Since the longitudinal magnetic field intensity changes much slower with height than the electron
density of the ionosphere, the equation can be rewritten as
( = K Bin x TEC,
.ol
(10.79b)
where BL = B cos 0 is taken at a mean ionospheric height,

usually near 400 km, K = 2.36 x 10 5 and TEC is fNdl.
Typical values of polarization rotation for northern midlatitude stations viewing a geostationary satellite near their
station meridian are given in Figure 10-84 as a function of
system frequency and total electron content. In fact, the
largest portion of TEC data available today from stations
throughout the world have come from Faraday rotation measurements from geostationary satellite VHF signals of opportunity.
For satellite navigation and communication designers,
however, the Faraday polarization rotation effect is a nuisance. If a linearly polarized wave is transmitted from a
satellite to an observer on or near the surface of the earth,
the amount of polarization rotation may be nearly an odd
integral multiple of 90 degrees, thereby giving no signal on
the receiver's linearly polarized antenna, unless the user is
careful to realign his antenna polarization for maximum
received signal.
As shown in Figure 10.84 at 4 GHz, a commercial
satellite transponder frequency band, the amount of Faraday

rotation can be a tenth of a radian, well in excess of that
required for dual, linear orthogonal channel separation.
The Faraday rotation problem is overcome by the use
of circular polarization of the correct sense at both the satellite and at the user's receiver. Generally the mobile user
finds it difficult to utilize circular polarization due to the
continual vehicle directional changes; thus he settles for a
received linear polarization. The 3 dB loss between transmitted circular polarization and received linear polarization
is a necessary price to pay for user antenna maneuverability
and simplicity.
10.8.5 AngularRefraction
The refractive index of the earth's ionosphere is responsible for the bending of radio waves from a straight
line geometric path between satellite and ground. This angular refraction or bending produces an apparent higher
elevation angle than the geometric elevation. Millman and
Reinsmith [1974] have derived expressions relating the refraction to the resultant angular bending. Perhaps the easiest
expressions to use, as given by Millman and Reinsmith
[1974] relate the ionospheric range error to angular refraction:
AE
R + r. sin E. (r, cos E.)

hi (2ro + hi) + r 2 sin Eo
x-
AR
R
(10.80)
where Eo is the apparent elevation angle, R is the apparent

range, AR is computed from AR = (40.3/f2) x TEC, ro is
the earth's radius, and hi is the height of the centroid of the
TEC distribution, generally between 300 and 400 km.
10-87
CHAPTER 10
.2
20
_=18
o1
VERTICAL
15
15
NO
IN
C3
~f.
oo
Z
N
L
=
<
10
15
20
25
IE
U
<
cos E1
dispersion, or differential time delay due to the normal ionaL

W
:0
10
25
20
15
ELEVATION ANGLE AT SURFACE

Figure 10-85. Refraction in elevationangle vs elevation angle for indicated frequencies and values of TEC.
For low
elevation angles and satellites well above most
of theionization, R>rosinEo, and the angular refraction can
be
expressed
as:
AE=
cos
2hi Eo AR.
(10.81)
for a TEC
Typical values of elevation refraction error
of 1018el/m2 column are shown in Figure 10-85 for several
frequencies. Notethat, atthe lowest frequency, 100 MHz,
near the horizon the refraction is well over 1.5 degrees! The
curves shown
in Figure 10.85 have been constructed using
the approximationderived by Millman and Reinsmith [1974]
for low elevation angles given in Equation (10.81).
Generally, the range error itself is the main ionospheric
navigation systems,and elevation
for
problemadvanced
angle errors are insignificant. Satellite detection radar systems, on the other hand, do have the requirement to know
accurate pointing elevation angles for their large aperture
arrays, though generally the accurate tracking is done by
using range rate information, and elevation angle is of secondary importance as long as the beamwidth of the antenna
is large enough to see the target.
Errors in the azimuth of radio waves transmitted through
the ionosphere can also occur; they depend upon azimuthal
gradients in TEC which are generally small and which can
usually be neglected in practical cases.
10-88
10.8.6 Distortion of Pulse Waveforms

Two characteristics of the ionosphere can produce distortion of pulses of RF energy propagatedthrough it. The
time delay due to thenormaliondifferential
dispersion,
osphere,
asor derivedby Millman [1965] is proportional to
2
1/f , and produces a difference in pulse arrival time across
a bandwidth Af of
At =
Af x TEC,
(10.82)
where c is the velocity of light in m/s, f and Af are expressed

in Hertz, and TEC is in el/m2 column. The dispersive term
the
distortion
pulse is thus proportional to TEC. When
for
difference in group delay time across the bandwidth of the
pulse is the same magnitude as the width of the pulse it will
be significantly disturbed by the ionosphere. Millman and
Bell [1971] also derived mathematical relationships for ionospheric dispersive effects on an FM Gaussian shaped pulse.
In addition to pulse distortion by the dispersive effects
due to the TEC of the normal background ionosphere, radio
pulses are also modified by scattering from ionospheric irregularities. Yeh and Liu [1979] have computed pulse mean
arrival time and mean pulsewidth due to both dispersion
and scattering.

1700T
1700
1600-[
1600
1500-
1500
1400--\
1400-
13001200
1300
\
1100\
1000
900
900
800-
800
700
700
600-
600
500-
500
400-
400
300--
300
200-
200
100 I
100 -
,o
O-I'
104
MARCH 1980
(Fio. 7 167)
1200
1100 1000
BOULDER,COLORADO
105
el /cm
106
3
.2
.4
0 I I
.6
el/cm
.8
1.0
1.2
1.4
1.6
units of 106
Figure 10-86. Typicalprofileof electrondensityvs height.In (a) log Ne is plotted;in (b) Ne is plottedon a linear scale.
10.9 IONOSPHERIC TOTAL ELECTRON

CONTENT (TEC)
ionosondes have been used since the 1930s to make continuous, routine measurements of the density at the peak of
the F2 region, measured by ionosondes as foF2, and equated
to Nmaxby
10.9.1 Average TEC Behavior

(foF2)2 = 80.6 N,
The ionospheric parameter responsible for the effects
described in section 10.8 is the total number of free electrons, TEC, or its rate of change, along the path from a
satellite to a ground station. The greatest contribution to
TEC comes from the F2 region of the ionosphere. A typical
daytime midlatitude, high solar maximum electron density
profile is illustrated in Figure 10-86. The curve on the left
side of Figure 10-86 is the log of Ne plotted versus height
as normally shown by ionospheric workers. Since the TEC
is represented by the area under the curve of a linear plot
of Ne versus height, the right hand plot of Figure 10-86
illustrates the actual linear plot. Note that most of the contribution to TEC occurs near the peak of the F2 region. The
reason for making this point is as follows: ground-based
(10.83)
where foF2 is in MHz, and N is in units of 106 el/cc.

In the 1950s and 1960s, continuing to a more limited
extent even today, upwards of 150 ionosondes were operated
to provide improved prediction capability for long distance
high frequency propagation by means of ionospheric refraction. Various models of foF2 were developed for this
purpose, one of the more popular ones being commonly
known as ITS-78 [Barghausen et al., 1969] after the report
number which described the model. This model, among
other things, characterized the 10 day average worldwide
behavior of foF2 by Fourier temporal components and Legendre polynomial geographic coefficients ordered by magnetic, rather than geographic, latitude. The success of this
10-89
CHAPTER
10
experimental, data based, or empirical model, in representing the actual worldwide foF2 is due to the large amount
of data available from ionosondes in many regions of the
world. Other characteristics of this model are discussed by
Dandekar 11982]and in Section 10.3 of this chapter.
For the TEC parameter, data availability have been, and
will likely continue to be, much more sparse. First, TEC
measurements have generally been calculated from measurements of Faraday polarization rotation using VHF signals of opportunity transmitted from geostationary satellite
telemetry transmitters. A few lunar reflected Faraday rotation measurements in the late 1950s and early 1960s and
the TEC obtained from a few low orbit satellites did not
contribute significantlyto our knowledge of world-wideTEC
behavior, at least not for modeling average ionospheric conditions. Only since the early to mid-1960s have TEC values
been obtained on a more-or-less regular basis. Even today
fewer than one dozen stations regularly contribute TEC data,
which can be used in TEC modeling purposes, to a world
data center.
Fortunately, most of the contribution toTEC comes from
near the F2 region density peak where models of foF2 are
available. These foF2 models can be combined with some
limited knowledge of topside ionospheric thickness obtained
from topside sounders, along with topside in situ density
measurements, to produce a complete ionospheric height
profile model. The most well known of these models is the
one by Bent [Llewellyn and Bent, 1973] which uses ITS78 coefficients for foF2, and topside exponentials for com-
puting TEC. A representation of world-wide average behavior of TEC is illustrated in Figure 10-87 for 2000 hours
UT. To first order the TEC contours shown in Figure 1087 move westward along magnetic, rather than geographic,
latitude lines, at the earth's rotation rate. The Bent model
was constructed using solar maximum data from the 19681969 period and had to be adjusted upward somewhat to
account for the much higher 1979-1980 solar maximum than
that of 1968-1969. This adjustment was necessary to adequately represent the actual TEC values from stations making observations in March 1980, which was near the maximum of the second highest solar cycle ever recorded in the
more than 200 year history of solar cycle observations. The
Bent model, appropriately adjusted for high solar cycle values, does however, represent fairly well the average behavior of TEC for many locations tested. Other worldwide
ionospheric electron density profile models from which average TEC can be obtained include ones by Ching and Chiu
[1973], and Chiu [1975], Kohnlein [1978], the 4-D model
[VonFlotow, 1978], and the International Reference Ionosphere (IRI)[Rawer, 1981]. The characteristics of some of
these models are described in Dandekar [1982] and in Section 10.3.
Other empirical models of TEC have been developed
directly from TEC data alone, though these have necessarily
been limited in temporal and geographic extent by the available data base. These include models of TEC over Europe
and the Mediterranean [Klobuchar, 1973] for low and medium activity portions of the 11 year solar cycle, and a
406"~~~~~~~~
CONTOURS
OF TOTAL ELECTRON
CONTENTUNITS
MARCH 1980
167)OF 10- el/r
0LATITUDE-400
I___
___
30
010-080
60
9090
-
120
150
L 180
LONGITUDE
COLUMN
Il
Il
I09
120
100
90
60
2000 UT
Figure 10-87.
10-90
Contours of vertical TEC in units of 10 16el/m
column for 2000 UT, March 1980.
30

model of TEC over the Indian subcontinent for both solar
minimum and for an average solar maximum [Klobuchar,
et al.1977]. Models of the slab thickness parameter, the
ratio of TEC/Nmaxhave been developed for specific regions
such as the one for northern Europe by Kersley [1980], and
one for the eastern USA by Klobuchar and Allen [1970],
from which TEC can be obtained from a model of foF2.
An algorithm designed for an approximate 50% correction
to world-wide TEC, for use in an advanced navigation system, has been developed by Klobuchar [1975].
All of the models listed here, and the list is by no means
complete, are empirical models that attempt to correct for
average TEC behavior only. However, the variability from
average TEC behavior can be large and may be important
to some radio wave systems that must propagate through
the ionosphere.
10.9.2 Temporal Variability of TEC

10.9.2.1 Variability from Monthly Mean TEC Values.
The ionosphere is a weakly ionized plasma and the resultant
TEC is a function of many variables including solar ionizing
radiation, neutral wind and electric field effects, neutral
composition, and temperature changes. A monthly overplot
of curves of diurnal changes in TEC for a northern midlatitude station for twelve months during a solar maximum
period is shown in Figure 10-88. The standard deviation
from monthly mean diurnal behavior is approximately 20%25%, during the daytime hours when the absolute TEC
values are greatest. Figure 10-89 shows the standard deviation from monthly average TEC behavior for the midday hours for a number of stations during the solar maximum
period 1968-1969. Again 20%-25% is a good value for the
standard deviation from the monthly average behavior. The
standard deviation is somewhat higher during the nighttime
hours, but the absolute TEC values are much lower during
these periods.
If a satellite ranging system has error requirements such
that it must correct for monthly average ionospheric time
delay, but still can tolerate the approximate 20%-25% variability of TEC from monthly average conditions, approximately 70%-80% of the ionospheric effect on the system
can be eliminated by the use of an average TEC model such
as the one constructed by Bent [Llewellyn and Bent, 1973].
If the system only requires an approximate 50% rms correction of the ionospheric time delay, the algorithm developed by Klobuchar [ 1975 can be used. On the other hand,
if corrections for some portion of the remainder of the ionospheric time delay are required, after a state of the art TEC
20
JANUARY1979
MARCH
FEBRUARY
APRIL
60-
20
'MAY
JULY
AUGUST
E
1
JUNE
60-
20-
1 20
SEPTEMBER
8g
AIo
i58o
L- +2
OCTOBER
NOVEMBER
14
-t
'
18
1 ,J4 4
DECEMBER
:.
11
TOTALEQUIVALENTVERTICALELECTRONCONTENT
FROMHAMILTON(ATS-5)
(TIME UT)
Figure 10-88
Monthly overplots of TEC diurnal curves for Hamilton, Mass. for 1979.
10-91
CHAPTER 10
80
PERCENTAGE
STANDARD
12 - 16 LT
DEVIATIONS
TEC
,
EDMONTON 1969
RBERYSTWYTH1969
HRMILTON 1969
60 -
STRNFORD 1969
HONOLULU 1969
50 ,
L
HONG KONG
40 -
1968
30
20
10
JRN
FEB
MAR
RPR
MAY
JUN
JUL
RUG
SEP
OCT
NOV
DEC
Figure 10-89. Percentage standard deviations for daytime TEC from the stations indicated.
model, such as the Bent one, has been used to take out the
monthly mean TEC, then the short term (a few hours) temporal variability as well as the geographic variability, of
TEC must be considered.
direction. Their results are shown in Figures 10-90a and b

for the east-west and the north-south station alignments,
respectively. No significant difference in correlation distance was found with season.
The percent improvement, P.I., in TEC from the average
10.9.2.2 Short Term Temporal Variability of TEC.
value is related to the correlation coefficient r by
The correlation time of departures of TEC from monthly

average curves has been studied by Donatelli and Allen
[1981]. They concluded, for the midlatitude station they
studied, that the useful prediction time was a function of
local time, season, and long term sunspot activity. However,
in most cases they gained no significant improvement over
the use of monthly mean predictions when they used actual
data more than 3 hours old. The longest useful prediction
interval occurred, fortunately, during solar maximum daytime hours when absolute TEC values are highest. During
solar minimum periods their useful prediction time interval
was often as short as one hour.
was ofen au.
In their study, Donatelli and Allen [ 1981] predicted TEC
data for the same geographic location and direction in the
sky as their measurements. If the prediction is for a different
location, the temporal correlation will be lower.
P.I. = 100 x [1-(1-r
)0-5]
(see Gautier and Zacharisen, [1965])
(10.84)
Note that a correlation coefficient of 0.7 explains only 29%

of the variance between the data at station pairs; hence a
measurement at one station, with a correlation coefficient
of 0.7 between data sets with a second station would result
in an improvement at the second station over the average
predicted value of only 29%.
10.9.3 TEC in the Near-Equatorial Region

All of the preceding sections have concentrated on the
behavior of TEC in the midlatitude regions of the world,
mainly because most of the available data are from that
mainly because most of the available data are from that
region. The near-equatorial region deserves special mention

10.9.2.3 Geographic Variability of TEC. The variabildue to the fact that the highest TEC values in the world
ity of TEC at the same local time, but as a function of
occur in this region, as shown in Figure 10-87. This region
distance has been studied by Klobuchar and Johanson [1977].
extends to approximately + 20-25 either side of the magThey utilized TEC data from two sets of stations, one aligned
netic equator, with the highest TEC values not at the equaapproximately along an east-west direction, with the other
tor, but rather at the so called "equatorial anomaly" regions
set of stations aligned along an approximate north-south located
at approximately + 15 from the magnetic equator.
10-92

a
',
Z8
.
LL
zM
.
00
-_
.8
LE
r 0.7
SE
W
.6
SEASONAL MEAN DAYTIME

CORRELATION COEFFICIENT
.2
t
1000
2000
3000
4000
DISTANCE
IONOSPHERIC
LATITUDE
5000
(km)
SEPARATION
A
70N W LONGITUDE
<
ZH
a)20
r~~Z
0
Z
bJ
UI
00
~~~~1.0
U.
.._
:t
bJ3Z
cc
.4
wH0
n
U-
z0
>
0 'l~~~~~~0
.8
.0
0
Er
:
0
Zz
SEASONAL MEAN DAYTIME

CORRELATION COEFFICIENT
w
E
.2
1000
3000
2000
IONOSPHERIC DISTANCE (km)
4000
5000
LONGITUDE SEPARATION AT NORTHERN MID-LATITUDES

Figure 10-90. Correlation coefficient vs station separation in (a) latitude and (b) longitude.
10-93
CHAPTER10
MAY 1980
SEPTEMBER
JUNE
JULY
AUGUST
OCTOBER
NOVEMBER
DECEMBER
FEBRUARY
MARCH
40 i
1801
JANUARY191
.4
APRIL
TOTAL EQUIVALENT
VERTICALELECTRON
CONTENT
FROM ASCENSION
ISLAND(SIRIO)
TIME(UT)
Figure 10-91. Monthly overplots of TEC diurnal curves for Ascension Island, May 1980-April 1981.
The regions of highest TEC values at 2000 hours UT are

clearly seen in Figure 10-87 near 1000west longitude.
Most of the day-to-day geographic variability of TEC
in the equatorial anomaly region during solar minimum conditions can be explained by the variability of equatorial
electrojet strength. Unfortunately, no such similar TEC data
are available for solar maximum.
An example of the high temporal variability of TEC for
solar maximum conditions for Ascension Island, a station
located near the peak of the southern TEC equatorial crest
region, is shown in Figure 10-91. Note the extremely large
day-to-day TEC variability in the afternoon and evening
hours in some months. Any satellite ranging system requiring ionospheric TEC corrections in the near-equatorial
region should not use the midlatitude standard deviation
values of approximately 20%-25% to represent the variability of the near-equatorial region.
10-94
10.9.4 TEC in the Auroral and

Polar Cap Regions
Since most available TEC values have been measured
using radio signals transmitted from geostationary satellites,
which can be viewed only at low elevation angles from high
latitudes, knowledge of the variability of TEC in the auroral
and polar cap regions is sparse. In the American longitude
sector, where the magnetic latitudes are lowest for a given
geographic latitude, there is considerable TEC data from
Goose Bay, Labrador, over which the aurora passes southward, even during moderately magnetically disturbed periods. The behavior of TEC during those periods can be
highly irregular, especially during the nighttime hours. TEC
values often exhibit rapid changes and occasionally even
exceed the daytime maximum values briefly. While the occurrence of general auroral activity may be predictable, the

6
SEPTEMBER
AUGUST
JULY
OCTOBER
xXXxx
XX
NO DATA
'""'-~ j
xXF
X
xxx KIRUNA, SWEDEN
s ~ ~ ~~J-~
_ O2
JANUARY
DECEMBER
6-1 NOVEMBER
-3
,FEBRUARY
CL
uI
o
SAO XXX
PAUL,
0
0
12
18
24
I2
18
24
BRAZIL
I0
6
12
lB 24
xxxSAD PAULO,BRAZIL
12
18
24
LOCAL TIME
Figure 10-92. Monthly average plasmaspheric electron content vs. local time for Aberystwyth, Wales (dashed line) and for Hamilton, Mass. (solid line).
Also plotted are values from Kiruna, Sweden for October 1975 and from Sao Paulo, Brazil for May 1975.
specific large increases in TEC, likely due to auroral precipitation, are not individually predictable, but may be statistically characterized as a function of magnetic activity.
In the polar cap region a negligible amount of TEC data
exists. The absolute TEC values are probably lower in this
region than in the midlatitudes, and the variability of the
polar cap TEC is probably very high.
10.9.5 ProtonosphericElectronContent
Most of the available TEC data has been taken by measurements of Faraday rotation of single frequency radio waves
transmitted from geostationary satellites to ground observers. The electron content obtained from Faraday rotation
observations, while made from radio waves transmitted from
satellites at geostationary satellite height above the earth's
surface, only includes the contribution of electrons up to
heights of approximately 2000 km. This is because the integrated product of the longitudinal component of the earth's
magnetic field times the electron density, above approximately 2000 km, is negligible. The only measurements of
the additional contribution of electrons above the Faraday
maximum height have been made using signals from an
ionospheric beacon on the geostationary satellite ATS-6.
Davies [19801 has reviewed the overall results of the ATS6 experiment. A summary of typical protonospheric electron
content data is shown in Figure 10-92 taken from Klobuchar
et. al. [1978]. Note that the protonospheric values are fairly
low in absolute value.
During the nighttime hours when the ionospheric TEC
is low, the protonospheric contribution may become a fairly
large percentage of the total number of electrons between
a satellite at geostationary height and an observer on, or
near the earth's surface. Unfortunately, no protonospheric
electron content data are available during solar maximum
conditions.
10.9.6 Short Term Variationsin TEC

The time rate of change of TEC, in addition to the normal
diurnal variations, also has periodic variations due to perturbations of the ionospheric F region from various potential
sources as geomagnetic substorms, meteorlogical sources
such as weather fronts, shock waves from supersonic aircraft, volcanic explosions, rocket launches, and other miscellaneous sources. While these short term variations in TEC
cover a large range of periods and amplitudes, common
periods range from 20 to over 100 minutes with amplitudes
10-95
CHAPTER 10
of a few percent of the background TEC. A 10% ionospheric
disturbance with respect to the background TEC is uncommon, while a 1% TEC perturbation is common. Titheridge
[1968] and Yeh [1972] have made studies of the statistics
of traveling ionospheric disturbances (TIDs), in TEC for
midlatitude regions.
A system that requires correction for the rate of change
of TEC cannot rely on models of TEC to provide reliable
information on short term rate of change of TEC information, and can use available TID information only in a statistical manner. The only recourse for a system significantly
affected by rate of change of TEC is to use a dual frequency
measurement technique to directly measure the ionospheric
contribution to range rate.
Experiments [Russell and Rycroft, 1980] dealt with a broad

spectrum of experiments: energetic particle injections, plasma
wave (VLF) injections, mass (neutral gas) injections, as well
as with laboratory and computer simulation experiments.
The most recent summary of Active Experiments in space
treated particle beams, neutral gas injections, wave injections and high power heating experiments [Burke, 1983].
The common thread that binds all of these methods is the
use of well-defined input/output experiments to probe the
system response functions for specific atmospheric and space
plasma systems.
In terms of purely ionospheric phenomena, the modification of ambient electrons and ions are most often achieved
by chemical injections or by radiowave heating experiments.
Each of these areas is treated in the following sections.
10.9.7 Conclusions
10.10.1 Chemical Releases
There are at least three categories of systems potentially
affected by ionospheric time delay. For the first category
of user the potential systems effects may be small, at least
under any naturally occurring worst case ionospheric conditions. In the second category, a user may require a nominal
correction for average ionospheric time delay, but is able
to tolerate the 20%-25% standard deviation from average
conditions. He should expect at least a 50% correction for
ionospheric time delay effects using a relatively simple time
delay algorithm, and up to 70%-80% for a state of the art,
fairly complex model. These model corrections can be improved by the use of actual ionospheric measurements within
a reasonable temporal and spatial frame. For the third category of user ionospheric model corrections, even updated
with near-real-time measurements, may not be sufficient to
correct for ionospheric time delay, and the system must then
make its own ionospheric correction. Fortunately, the ionosphere is a dispersive medium and the use of identical
modulation on two, widely-spaced frequencies will allow a
direct measurement to be made of ionospheric range delay.
Two coherently-derived carrier frequencies may be used to
obtain accurate time rate of change information for TEC.
Details of measuring ionospheric effects directly by a system's use of multiple frequencies are available in Burns and
Fremouw [1970].
10.10 ARTIFICIAL MODIFICATION

The field of ionospheric modifications is a subset of a
more general class of research today called "Active Experiments" in space plasmas. This field was initiated early in
the space program by using rocketborne chemical releases
as tracers and/or modifiers of upper atmospheric processes.
The physical basis for such experiments was reviewed in
some detail by Haerendel [1976] during the first internationalmeeting devoted entirely to artificialmodificationstudies
[Albrecht, 1976]. A second major symposium on Active
10-96
The history of chemical release experiments dates from

the the earliest days of space exploration when, shortly after
Sputnik-1 in 1957, rocket-borne payloads of highly reactive
chemicals were injected into the upper atmosphere in attempts to use artificial perturbation techniques as a way of
investigating the structure and dynamics of the neutral and
ionized components of the upper atmosphere. The Air Force
Cambridge Research Laboratories (ARCRL) carried out the
initial work with plasma cloud injections [Marmo et al.,
1959], and later a pioneering and comprehensive series of
chemical injection experiments under PROJECT FIREFLY
[Rosenberg, 1964]. Experiments using barium releases (or
similar, easily ionized species) have formed the major activity in this field, tracing and/or modifying ionospheric
processes from auroral locations [Holmgren et al., 1980] to
the equator [Kelly et al. 1979]. The symposium proceedings
referencedabove [Albrecht, 1976;Russell and Rycroft, 1980;
Burke, 1983] offer comprehensive summaries of these experiments.
The field of neutral mass injections was first concerned
with the environmental impacts that might result from the
larger and more powerful rockets being developed for space
exploration [Kellogg, 1964]. In 1973, when the last Saturn
V rocket to be used in the U.S. Space Program launched
NASA's Skylab Workshop, the resultant deposition of approximately 1000 kg/s of H2 and H2 0 exhaust molecules
into the 200-440 km altitude region initiated a rapid and
large-scale depletion of the ionosphere to an extent never
seen before (see Figure 10-93). The artificially-created "ionospheric hole" amounted to nearly a 50% decrease in the
total electron content (TEC) of the ionosphere over an area
of approximately a million square kilometers. Mendillo et
al. [1975 a,b] attributed the effect to the explosive expansion
of an exhaust cloud of highly reactive molecules that initiated a rapid recombination of the ionospheric plasma.
The introduction of such typical rocket exhaust products
as H2 , H20 and CO2 into the upper atmosphere causes the

OBSERVATIONS
FROMSAGAMORE
HILL, HAMILTON,MASS.
ATS-3
E 20
..
SKYLAB
LAUNCH
16 -
o
,,12
s,
<
a
~-F~~~~~~~~~~~
/
W
~-
of/
aJ
a | 1. l,,,,,
O0
03
I,,,,,,,,
06
09
I,,
12
5I
14 MAY1973
18
Figure 10-93. TotalElectronContent(TEC)datausedto detectthe"SKYLAB effect" on 14 May 1973. The dashed curve gives the
anticipated diurnal TEC behavior based upon a monthly
median predictionupdatedfor geomagneticstorm effects

[Mendilloet al., 1975b].
atomic ion F region plasma to be transformed to a molecular

ion plasma at rates 100 to 1000 times faster than occur with
the naturally present molecules of nitrogen (N2) and oxygen
(02).
These important reactions are

,,,
0 + + H20 -- H2O+ + O
K, = 2.4 x 10
O+ + H2
K2 = 2.0 x 10-9 cm3/s
--
OH
+ H
cm3/s
(10.85)
(10.86)
O + + CO 2 -O2
+ CO
Once a molecular ion is formed, its dissociative recombination with an ambient electron occurs rapidly,
d,
H2 O' + +
e -OH
+ H a, = 3.0 x 10-7 cm3/s
(10.88)
,,
-- H + O
c0, = 1.0 x 10 7 cm3 /s
(10.89)
2+ 0 + +
e-,0
modification effects upon
scheduled for the late 1980's. Some of these concepts have

been tested using rocketborne chemical payloads during projects LAGOPEDO [Pongratz and Smith, 1978], WATERHOLE [Whalen et al., 1981], BIME [Narcisi, 1983] and
COLOURED BUBBLES [Haerendelet al., 1983]. In the

AFGL Ionospheric Modification Study [Narcisi, 1983], attempts were made to study effects associated with SF6 induced negative ion plasmas [Mendillo and Forbes, 1978].
During so-called "experiments of opportunity," where
scheduled rocket launches are monitored by a variety of
techniques, satellite radio beacon observations have been
reported by Mendillo, et al. [1980] incoherent scatter measurements by Wand and Mendillo [1984], and optical diagnostics by Kofsky [1981] and Mendillo and Baumgardner
[1982]. Figure 10-95 offers an example of the artificial
airglow clouds associated with F region hole-making ex-
periments.
K3 = 1.2 x 10 - 9cm 3 /s
(10.87)
OH + + e
the lack of large-scale/long-lived
the lower regions of the ionosphere is due primarily to the

high neutral densities and molecular ion chemistry already
dominant at D and E region heights, as discussed in detail
by Forbes [1980].
Computer simulation models for the F region effects
have been constructed by Bernhardt et al. 1975], Mendillo
and Forbes [1978], Anderson and Bernhardt [1978] and Zinn
\z \v \ / and
\ Sutherland [19801. The emphasis in these studies has
\
\ranged
from environmental impacts of proposed in-space
construction scenarios [Rote, 1979], to laboratory-in-space
Is ,
experiments using "dedicated engine-burns" of the space
21
24EST
shuttle as part of the Spacelab-2 mission in 1985, to a series
of chemical modification experiments planned for the Combined Release and Radiation Effects Satellite (CRRES)
a,=
2.0 x
10-7
cm3 /s
(10.90)
and hence an "ionospheric hole" is formed. A review of
rocket induced ionospheric disturbances has been given by
Mendillo [1981].
Figure 10-94 contains a schematic showing the many
physical and chemical processes associated with artificiallyinduced depletions in the F region. It should be noted that
10.10.2 High Power HF Transmissions

Ground based high power high frequency transmitters
operating below the critical frequency of the ionosphere
have been used to artificially modify the ionospheric electron
thermal budget and plasma characteristics [Utlaut, 1970;
Gordon et al., 1971; Shlyger, 1974; for comprehensive reviews, see Carlson and Duncan, 1977 and Gurevich and
Fejer,
1979]. The power aperture product of these high
power transmitters have been typically of the order of 104

Mwm2 providing power densities of about 10-100 uwm -2
at ionospheric heights. The ionospheric "modification"
or so-called "heating" experiments have been observed to
cause not only the initially intended enhancements of electron gas temperature with associated plasma redistribution
but give rise to a variety of nonlinear plasma phenomena.
Figure 10-96 [after Carlson and Duncan, 1977] summarizes,
in a schematic form, the striking variety of observed effects of ionospheric heating. The enhancements of electron
gas temperature have been observed to be a few hundred
degrees K [Gordon et al., 1971] caused by the deviative
10-97
CHAPTER 10
SPACELAB-2 PLASMA DEPLETION EXPERIMENTS
EXPANUSION
SF
by
IVPQH|
CHEMISTRYRRECELC
CHEICAL
XCITAF
ELECTRONS
sl#
ION-MOLECUL
IEXCINION
Of
LL
, -.ItCP
_ NEUTRAL
RISR
IE
EC!.s,
RCOE.
RECOMBINATION
|
RAPID INITIAL
EXPANS
HANCED
|O
PHOTO LELCIRN
*SCAPL 10
CON)UIGAT
IONOFSPHElLRE
l
JTC
C R IRONA
PLASMA
FLO
INTODEPLTE
OFC
|TRIGCERING
i~P(EQUATORIAL)
PLASMA
~~~~ACOUSTIC
~~GRA~VlI~~TY
BEACONSATELHLIESS_
FLUX TURN
IPROTONOSPHRICI
-i
THEORNE-
-A.~
\7/
BEACONSATELLITES
S4nv
C
PLASIA
HOLES
AT\~
v////
POIINCOHERENT
IN- SITU
PROBES
~/'PLASMA
DIAGNOST
SDIAGNOSTICS
INCOHERENT
SCATTER
DIAGNOSTICS
GROUND-BASED -- AIRBORNE -Figure 10-94.
Schematic summary of possible rocket effluent effects upon the upper atmosphere
1985.
mission scheduled for Spring-Summer
absorption of the heater wave near the altitude of HF reflection. The following manifestations ofplasma instabilities
have been observed: (1) artificially created spread F; (2)
strongly enhanced radio wave absorption; (3) the creation
of field-aligned density irregularities which scatter (this phenomenon is sometimes called field-aligned scatter of FAS)
an incident HF, VHF, or UHF wave with virtually no frequency change and make certain types of scatter communication circuits possible; (4) scattering process in which
the frequency of the scattered wave differs from the frequency of the incident wave by roughly the frequency of
10-98
IN-SPACE
(h > 200 km), associated
with the NASA Spacelab-2
the high-power HF transmissions; and (5) strongly enhanced

airglow at 6300 A; some enhancement at 5577 A is also
observed (see the Special Issue of Radio Science, [1974]).
The short-wavelength (I cm - 10 m) field-aligned irregularities produced by ionospheric heating are a result of
parametric decay instability or wave interaction between the
high power radio wave (pump) and the ion-acoustic and
Langmuir waves. This was predicted from theory [Perkins
and Kaw, 1971; Perkins et al., 1974;DuBois and Goldman,
1972] and experimentally confirmed at Arecibo [Carlson et
al., 1972].

is now attributed to either a thermal self-focusing mechanism [Perkins and Valeo, 1974; Thome and Perkins, 1974]
or the alternative mechanisms of stimulated Brillouin scattering [Cragin and Fejer, 1974] and stimulated diffusion
scattering [Goldman, 1974].
A02:57:37
B2:56:05
The aboverange of irregularityscalesizes has sufficient
C02:58:36
power spectral intensity to cause scintillation of radio signals

received from radio stars and artificial satellites. This was
demonstrated when VHF/UHF signals transmitted through
the artificially heated ionospheric F region were found to
D.2.so:05
E.2:59:35
F-3:00:5
G.:03:42
Figure 10-95. The growth of an ionospheric hole is shown in this sequence
of image-intensified,wide-anglephotographsof the ex-
panding6300A airglowcloudproduced
byexcitedoxygen
atomscreatedfromtherecombination
of freeelectrons
and
molecularions (021 , OHt, H2 0z ) produced
by exhaust
molecules
(CO2, H2, H20) andambientatomicions(0O).
Times are a.m., PST [Mendilloand Baumgardner,1982].
The long-wavelength (-1 km) field-aligned irregularities giving rise to artificial spread F [Utlaut et al., 1970;
Utlaut and Violette, 1972; Wright, 1973] could not, however, be explained in terms of the above instability process.
The causative mechanism for the generation of long wavelength irregularities remained obscure for quite a while and
exhibitscintillations[Rufenach,1973;PopeandFritz, 1974;
Bowhill, 1974]. Radio star scintillation measurements at 26
MHz during ionospheric modification indicated the presence
of either rapid and random or deep long-period (-5 mins)
fluctuations. In order to avoid some of the difficulties of
radio star observations,Bowhill [1974]performedscintil-
lation measurements with both geostationary and orbiting

satellites and established the field-aligned nature of the irregularities causing VHF and UHF scintillations, their transverse scale and drift speed. One feature common to all the
above studies was the fact that the heater frequency was
below the plasma frequency of the F region. The magnitude
of scintillations observed on transionospheric communication channels is found to be of the order of 5 dB at 250
MHz when the nighttime F region is heated by an incident
power density of about 50 uwm -2 [Basu et al., 1980a].
ACCELERATED ELECTRONS
(SOME ESCAPE TO CONJUGATE
HEMISPHERE)
at
HF REFLECTION ALTITUDE
ELECOeV S
TO
-J Wt
300 -
(-km's THICK)
<lOOm THICKSLABOF
... :'{'.::!
4'.'
.t-
*lr.:
r,4F.gAf/
I.
/~
AIRGLOW EXCITATION:
O :'-N2(,029
/
(H10's km THICK)
~"l e.gO.7m WAVELENGTH

PLASMA INSTABILITIES
HF-EXCITED
PLASMA INSTABILITIES
/.
'
5
fHF
IONOSPHERIC PLASMA FREQUENCY fp(MHz)
Figure 10-96. Effects produced by ground-based transmitter of power aperture of the order of 104Mwm2 in the 4-12 MHz frequency range. Energy
deposited in the ionospheric plasma alters both the thermal and nonthermal properties of its charged particle population. Controlled
experiments have applications to aeronomy, chemical rates, atomic cross sections, communications, and a number of areas of plasma
physics[Carlsonand Duncan, 1977].
10-99
CHAPTER 10
REFERENCES
Aarons, J., "A Descriptive Model of F-Layer High-Latitude
Irregularities as Shown by Scintillation Observations,"
J. Geophys. Res., 78: 7441-7450, 1973.
Aarons, J., "Equatorial Scintillations: A Review," IEEE
Trans. Antennas Propagat., AP-25, 729, 1977.
Aarons, J., "Global Morphology of Ionospheric Scintillations," Proc. IEEE, 70: 360-378, April 1982.
Aarons, J. and R.S. Allen, "Scintillation Boundaries During
Quiet and Disturbed Magnetic Conditions," J. Geophys.
Res., 76: 170, 1971.
Aarons, J. and E. Martin, "The Effects of the August 1972
Magnetic Storms on Ionospheric Scintillations," Radio
Sci., 10: 547-554, 1975.
Aarons, J., J.P. Mullen, and H.E. Whitney, "The Scintillation Boundary," J. Geophys. Res., 74: 884-889, 1969.
Aarons, J., E. MacKenzie, and K. Bhavnani, "High-Latitude Analytical Formulas for Scintillation Levels," Radio Sci., 15: 115-127, 1980a.
Aarons, J., J.P. Mullen, J.P. Koster, R.F. daSilva, J.R.
Medeiros, R.T. Medeiros, A. Bushby, J. Pantoja, J.
Lanat, and M.R. Paulson, "Seasonal and Geomagnetic Control of Equatorial Scintillations in Two Longitudinal Sectors," J. Atmos. Terr. Phys., 42: 861-866,
1980b.
Aarons, J., E. MacKenzie, and S. Basu, "Microwave Equatorial Scintillation Intensity During Solar Maximum,"
Radio Sci., 16: 939-945, 1981a.
Aarons, J., J.P. Mullen, H. Whitney, A. Johnson, and E.
Weber, "VHF Scintillation Activity Over Polar Latitudes," Geophys. Res., Lett., 8: 277-280, 1981b.
AGARD, "Modern HF Communications," AGARD Lecture
Series 127, (avail. through NTIS, Springfield, Va) 1983.
Akasofu, S.-I., "Midday Auroras and Polar Cap Auroras,"
Geofysiske Publikasjoner (Norske Videnskaps-Akademi
i Oslo), 29: 73-85, 1972.
Albrecht, H.J., (ed.), Artificial Modification of Propagation
Media, AGARD Conf. Proc. 192, Brussels, 26-29 April
1976.
Al'pert, Ia. L., "Ionospheric Propagation of Radio Waves,"
Acad. of Sci. of USSR, Moscow, Section on Long Waves
(avail) as Translation T5-60, National Bureau of Standards, Boulder, 1960.
Ames, J.W. and R.D. Egan, "Digital Recording and ShortTerm Prediction of Oblique Ionospheric Propagation,"
IEEE Trans. Antennas Propag. AP-15: 382, 1967.
Ames, J.W., R.D. Egan and G.F. McGintie, "Short-Term
Prediction of HF Communication Circuit Performance"
in Ionospheric Forecasting, AGARD Conf. Proc., 49,
AD700896, 1970.
Anderson, D.N., "Modelling the Ambient Low Latitude FRegion Ionosphere-A Review,"J. Atmos. Terr. Phys.,
43: 753, 1981.
Anderson, D.N. and P.A. Bernhardt, "Modeling the Effects
of an H Gas Release the Equatorial Ionosphere," J.
Geophys. Res., 83: 4777, 1978.
Appleton, E.V. and W.J.G. Beynon, "The Application of
Ionospheric Data to Radio-Communication Problems:
Part 1," Proc. Phys. Soc., 52: 518, 1940.
10-100
Baggaley, W.J., "BackscatterObservationsof F-RegionFieldAligned Irregularities During the I.Q.S.Y.," J. Geophys. Res., 75: 152-158, 1970.
Bain, W.C., "Models of the Mid Latitude D Region at
Noon" in Medium, Long and Very Long Wave Propagation, AGARD Conf. Proc., 305: 34-1-34-8
ADA 113969, 1982.
Balsley, B.B. and W.L. Ecklund, "VHF Power Spectra of
the Radar Aurora," J. Geophys. Res., 77: 4746-4760,
1972.
Balsley, B.B., W.L. Ecklund, and R.A. Greenwald, "VHF
Doppler Spectra of Radar Echoes Associated with a
Visual Auroral Form: Observations and Implications,"
J. Geophys. Res., 78: 1681-1687, 1973.
Bannister, P.R., "Variations in Extremely Low Frequency
Propagation Parameters," J. Atmos. Terr. Phys., 37:
1203-1210, 1975.
Bannister, P.R., "Extremely Low Frequency (ELF) Propagation," NUSC Scientific and Engineering Studies, Naval Underwater Systems Center, New London, Conn.,
February 1980.
Bannister, P.R., "Overview of ELF Propagation," in Medium, Long, and VeryLong Wave Propagation, AGARD
Conf. Proc., 305: 10-1-10-13 ADA113969, 1982.
Barghausen, A.L., J.W. Finney, L.L. Proctor, and L.D.
Schultz, "Predicting Long-term Operational Parameters
of High-Frequency Sky-Wave Telecommunication Systems," ESSA Tech. Rept. ERL 110-ITS 78, 1969.
Barr, R., "The Effect of Sporadic E on the Nocturnal Propagation of ELF Radio Waves," J. Atmos. Terr. Phys.,
39: 1379-1387, 1977.
Barry, G.H., "A Low Power Vertical Incidence lonosonde"
IEEE Trans. Geosci. Elect., GE-9(2): 86-89, 1971.
Basu, S. and Su. Basu, "Modeling of Equatorial Phase and
Amplitude Scintillations from OGO-6 and AE Irregularity Data, Low Latitude Aeronomical Processes,"
COSPAR Symposium Series, 8: 187-199, Pergamon Press
1980.
Basu, S. and H.E. Whitney, "The Temporal Structure of
Intensity Scintillation Near the Magnetic Equator," Radio Sci., 18: 263-271, 1983.
Basu, S., Su. Basu, A.L. Johnson, J.A. Klobuchar, and
C.M. Rush, "Preliminary Results of Scintillation Measurements Associated with Ionosphere Heating and Possible Implications for the Solar Power Satellite," Geophys. Res. Lett., 7: 609, 1980a.
Basu, S., Su. Basu, J.P. Mullen, and A. Bushby, "Longterm 1.5 GHz Amplitude Scintillation Measurements at
the MagneticEquator," Geophys. Res. Lett., 7: 259-262,
1980b.
Basu, S., J.P. McClure, Su. Basu, W.B. Hanson, and J.
Aarons, "Coordinated Study of Equatorial Scintillation
and in-situ and Radar Observations of Nighttime F Region Irregularities," J. Geophys. Res., 85: 5119-5130,
1980c.
Basu, Su. and J. Aarons, "The Morphology of High-Latitude VHF Scintillation Near 70W," Radio Sci., 15:
59-70, 1980.

Basu, Su. and M.C. Kelley, "A Review of Recent Observations of Equatorial Scintillations and their Relationship to Current Theories of F-Region Irregularity Generation," Radio Sci., 14:471, 1979.
Basu, Su., R.L. Vesprini, and J. Aarons, "Field-Aligned
Ionospheric E-Region Irregularities and Sporadic E,"
Radio Sci., 8: 235-246, 1973.
Basu, Su., R.L. Vesprini, and J. Aarons, "F-Layer Irregularities as Determined by Backscatter Studies at 19
MHz over Halfof a Solar Cycle," Radio Sci., 9: 355-371,
1974.
Basu, Su., S. Basu, R.C. Livingston, E. MacKenzie, and
H.E. Whitney, "Phase and Amplitude Scintillation Statistics at 244 MHz from Goose Bay using a Geostationary Satellite," AFGL TR-82-0222, ADA124291,
1982.
Bates, H.F. and P.R. Albee, "Aspect Sensitivity of HF
Auroral Echoes," J. Geophys. Res., 74: 1164-1168,
1969.
Bates, H.F. and P.R. Albee, "Aspect Sensitivity of F-Layer
HF Backscatter Echoes," J. Geophys. Res., 75: 165-170,
1970.
Belrose, J.S., "Radio Wave Probing of the Ionosphere by
the Partial Reflection of Radio Waves (From Heights
below 100 km)," J. Atmos. Terr. Phys., 32: 567-596,
1970.
Belrose, J.S., "COSPAR Symposium Record on D- and ERegion Ion Chemistry," Aeron. Rept. No. 48, Univ. of
Illinois, Urbana, III., 1972.
Belrose, J.S., "LF Propagation: An Overview" in Medium,
Long, & Very Long Wave Propagation, AGARD Conf.
Proc., 305: 22-1-22-10 ADA 113969, 1982.
Belrose, J.S. and L. Thomas, "Ionization Changes in the
Middle Latitude D-Region Associated with Geomagnetic Storms," J. Geophys. Res., 30: 1397-1413,
1968.
Bennett, S.M. and A.B. Friedland, "Prediction of Daily
Fluctuations of the F-Region Plasma Frequency," Ionospheric Forecasting, AGARD Conf. Proc., 49
AD700896, 1970.
Bent, R.B., S.K. Llewellyn, and M.K. Walloch, "Description and Evaluation of the Bent Ionospheric Model," I,
SAMSQ-TR-72-239, 1972.
Bernhardt, P.A., C.G. Park, and P.M. Banks, "Depletion
of the F Region Ionosphere and Protonosphere by the
Release of Molecular Hydrogen," Geophys. Res. Let.,
2: 341, 1975.
Bernstein, S.L., M.L. Burrows, J.E. Evans, A.S. Griffiths,
D.A. McNeill, C.W. Neissen, I. Richer, D.P. White,
and D.K. William, "Long Range Communication at Extremely Low Frequencies,"Proc. IEEE, 62(3), 292-312,
1974.
Berry, L.A., "Wave Hop Theory of Long Distance Propagation of LF Radio Waves," Radio Sci., 68D:
1275-1282, 1964.
Berry, L.A. and M.E. Chrisman, "The Path Integrals of
LF/VLF Wave Hop Theory," Radio Sci., 64D:
1469-1480, 1965.
Betts, J.A. and M. Darnell, "Real Time HF Channel Estimation by Phase Measurements on Low-Level Pilot
Tones" in "Radar Systems and the Ionosphere," AGARD

Conf. Proc., 173 ADA022439, 1975.
Bibl, K., "Fixed and Variable Frequency Pulse Transmission at Oblique Ionospheric Incidence," AFCRL 64392
AD603045, 1963.
Bibl, K. and B.W. Reinisch, "Digisonde 128 PS," University of Lowell, Center for Atmospheric Research,
Lowell, Mass., 1978a.
Bibl, K. and B.W. Reinisch, "The Universal Digital Ionosondes," Radio Sci., 13: 519-530, 1978b.
Bickel, J.E., J.A. Ferguson, and G.V. Stanley, "Experimental Observations of Magnetic Field Effects on VLF
Propagation at Night," Radio Sci., 5: 19-25, 1970.
Bilitza, D., M. Noor Sheikh, and R. Eyfrig, "A Global
Model for the Height of the F2 Peak using M3000 Values
from the CCIR Numerical Map," Telecomm. J., 46:
549-553, 1971.
Black, H.D. "The Transit System, 1977; Performance, Plans
and Potential," Philos. Trans. R. Soc. Lond. A294:
217-236. 1980.
Booker, H.G., C.M. Crain, and E.C. Field, "Transmission
of Electromagnetic Waves Through Normal and Disturbed Ionospheres," RAND Rept. R-558-PR, RAND
Corp., Santa Monica, Calif., 1970.
Booker, H.G. and G. MajidiAhi, "Theory of Refractive
Scattering in Scintillation Phenomena," J. Atmos. Terr.
Phys., 43: 1199, 1981.
Bowhill, S.A., "Satellite Transmission Studies of SpreadF Produced by Artificial Heating of the Ionosphere,"
Radio Sci., 9: 975, 1974.
Bowles, K.L., "Observations of Vertical Incidence Scatter
from the Ionosphere at 41 Mc/sec," Phys. Rev. Lett.,
1:454, 1958.
Bowles, K.L. and Staff of Jicamarca Radio Observatory
Institute, "Profiles of Electron Density Over the Magnetic Equator Obtained Using the Incoherent Scatter
Technique," Tech. Rept. Note 169, National Bureau of
Standards, Boulder, Colo., 1963.
Bracewell, R.N. and W.C. Bain, "An Explanation of
Radio Propagation at 16 kc/sec in Terms of Two Layers
Below E Layer," J. Atmos. Terr. Phys., 2: 216-225,
1952.
Bracewell, R.N., K.G. Budden, J.A. Ratcliff, T.W. Straker,
and K. Weekes, "The Ionospheric Propagation of Low
and Very Low Frequency Radio Waves Over Distances
Less than 1000 km," Proc. IEE, 98: 221, 1951.
Bradley, P.A. and C. Bedford, "Prediction of HF Circuit
Availability," Elec. Lett., 12: 32, 1976.
Bradley, P.A. and J.R. Dudeney, "A Simple Model of the
Vertical Distribution of Electron Concentration in the
Ionosphere," J. Atmos. Terr. Phys., 35: 2131, 1973.
Bramley, E.N., "Fluctuations in Direction and Amplitude
of 136 MHz Signals from a Geostationary Satellite," J.
Atmos. Terr. Phys., 36: 1503-1513, 1974.
Breit, G. and M.A. Tuve, "A Test of the Existence of the
Conducting Layer," Phys. Rev., 28: 554-575, 1926.
Briggs, B.H. and I.A. Parkin, "On the Variation of Radio
Star and Satellite Scintillation with Zenith Angle," J.
Atmos. Terr. Phys., 25: 339-365, 1963.
Brooks, D., "Observations of Radio Auroral Echoes and
10-101
CHAPTER 10
Simultaneous Magnetic Disturbances," J. Atmos. Terr.
Phys., 27: 1151-1158, 1965.
Brooks, D., "Observations of Auroral Echoes Following
Worldwide Storm Sudden Commencements," J. Atmos.
Terr. Phys., 28: 103-109, 1966.
Buchau, J., W.N. Hall, B.W. Reinisch, and S. Smith,
"Remote ionospheric Monitoring from Effects of the
Ionosphere on Space and Terrestrial Systems," Proc.
IES (Ionospheric Effects Symposium), edited by John
H. Goodman, U.S. Government Printing Office, Washington, D.C., 1978.
Buchau, J., B.W. Reinisch, E.J. Weber, and J.G. Moore,
"Structure and Dynamics of the Winter Polar Cap F
Region," Radio Sci., 19: 995-1010, 1983.
Budden, K.B., Radio Waves in the Ionosphere, Cambridge
University Press, Cambridge, UK, 1961.
Buneman, O., "Scattering of Radiation by the Fluctuations
in a Non-Equilibrium Plasma," J. Geophys. Res., 67:
2050, 1962.
Burke, W. (ed.), Proceedings of International Symposium
on Active Experiments in Space, 23-29 May 1983, Alpbach, Austria, European Space Agency Report ESASP-195, Paris, 1983.
Burns, A.A. and E.J. Fremouw, "A Real-Time Correction
Technique for Transionospheric Ranging Error," IEEE
Trans. Antennas Propag., AP-18:(6), Nov. 1970.
Burrows, M.L., ELF Communications Antennas, Peter Peregrinus Ltd., Stevenage, England, 1978.
Carlson, H.C. and L.M. Duncan, "HF Excited Instabilities
in Space Plasmas," Radio Sci., 12: 1001, 1977.
Carlson, H.C., W.E. Gordon, and R.L. Showen, "High Frequency Induced Enhancements of the Incoherent Scatter
Spectrum at Arecibo,"J. Geophys.Res., 77: 1242, 1972.
CCIR (Comite Cosultatif International des Radiocommunications = International Radio Consultative Committee), "World Distribution and Characteristics of Atmospheric Radio Noise," Report 332, Doc. Xth Plenary
Ass., International Telecommunications Union (ITU),
Geneva, 1963.
CCIR, "CCIR Atlas of Ionospheric Characteristics," Report
340 and 340-2, Doc. XIth Plenary Ass., ITU, Geneva,
1966.
CCIR, "Interim Method for Investigating Sky-Wave Field
Strength and Transmission Loss at Frequencies Between
the Approximate Limits of 2 and 30 MHz," Report 252-2,
Doc. XIlth Plenary Ass., ITU, Geneva, 1970.
CCIR, "Second CCIR Computer-Based Interim Method for
Estimating Sky-Wave Field Strength and Transmission
Loss Between 2 and 30 MHz," Suppl. Rept. 252-2, Doc.
XIVth Plenary Ass., ITU, Geneva, 1978.
CCIR, "Prediction of Sky-Wave Field Strength Between
150 and 1500 kHz," CCIR Recommendation453-3, Doc.
XIV Plenary Ass., ITU, Geneva, 1978.
CCIR, "Special Properties of the High Latitude Ionosphere
Affecting Radio-communications,"Study Program 25B/6,
Doc. 6/1005-E, 14 Oct 81, XVth Plenary Ass., ITU,
Geneva, 1981a.
CCIR,"Short-Term Forecasting of Operational Maximum
Usable Frequencies and Total Electron Content," Study
Program 27C/6, Doc. 6/1017-E, 14 Oct 81, XVth Plenary Ass., ITU, Geneva, 1981b.
10-102
CCIR, "Short-Term Prediction of Solar-Induced Variations

of Operational Parameters for Ionospheric Propagation,"
Study Program 27A/6, Doc. 6/1016-E, 14 Oct 81, XV
Plenary Ass., ITU, Geneva, 1981c.
CCIR, "Real-Time Channel Evaluation of Ionospheric Radio Circuits," Study Program 27B/6, Doc. 6/1018-E of
XVth Plenary Ass., Geneva, 1981d.
CCIR, "Prediction and Short-Term Forecasting of HF Radio
Propagation Conditions at High Latitudes," Draft Report
submitted ITU, Geneva, August 1983.
Chan, K.L. and L. Colin, "Global Electron Density Distribution from Topside Soundings," Proc. IEEE, 57: 6,
1969.
Chapman, F.W., D.L. Jones, J.D.W. Todd, and R.A. Challinor, "Observations on the Propagation Constant of the
Earth-Ionosphere Waveguide in the Frequency Band 8
c/s to 16 kc/s," Radio Sci., 1(11), 1273-1282, 1966.
Ching, B.K. and Chiu, Y.T., "A Phenomenological Model
of Global Ionospheric Electron Density in the E, Fl,
and F2 Regions," J. Atmos. Terr. Phys., 35: 1615, 1973.
Chiu, Y.T., "An Improved Phenomenological Model of IonosphericDensity,"J. Atmos. Terr. Phys., 37: 1563, 1975.
Cole, D.G. and L.F. McNamara, "Variations of Spread-F
Occurrence Rates at Near-Equatorial Stations in the Australian Zone," Aust. J. Phys., 27: 249, 1974.
Cragin, B.L. and J.A. Fejer, "Generation of Large-Scale
Field-Aligned Irregularities in Ionospheric Modification
Experiments," Radio Sci., 9: 1071, 1974.
Crain, C.M., "Ionospheric Probing with Long Wavelength
Radio Waves," J. Atmos. Terr. Phys., 32: 551-566,
1970.
Crane, R. K., "Spectra of Ionospheric Scintillation," J. Geophys. Res., 81: 2041-2050, 1976.
Crane, R.K., "Ionospheric Scintillation," Proc. IEEE, 65:
180, 1977.
Cretcher, C.K., "Ionospheric Effects in NAVSTAR-GPS,"
in Effects of the Ionosphere on Space Systems and Communications, edited by J.M. Goodman, Naval Research
Lab., Washington, D.C., U.S. Government Printing Office, Stock No. 08-051-00064-0, 1975.
Czechowsky, P., W. Dieminger, and H. Kochan, "Backscatter Results from Lindau-l. Observations of Radio
Auroras," J. Atmos. Terr. Phys., 36: 955-966, 1974.
D'Accardi, R.J., "Time Series Modelling and Analysis of
HF Vertical and Short-Path Oblique Incidence Ionospheric Soundings," U.S. Army Communications R &
D Command Tech. Rept. CORADCOM 78-7, NTIS
ADA058630, 1978.
Dandekar, B.S., "Ionospheric Modeling," AFGL TR-820024, ADA 115243, 1982.
Danielson, C., "Auroral Observations at Thule 1961-65,"
Danish Met. Inst., Geophys. Papers R-9, 1969.
Darnell, M., "Channel Evaluation Techniques for HF Communication" in Radio Systems and the Ionosphere,
AGARD Conf Proc., 173, 1975a.
Darnell, M., "Adaptive Signal Selection for Dispersive
Channels and its Practical Implications in Communications System Design" in Radio Systems and the Ionosphere, AGARD Conf. Proc., 173, 1975b.
Darnell, M., "Channel Evaluation Techniques for Dispersive Communications Paths," Communication System

and Random Process Theory, edited by J.K. Skwirzynski, Sijthoff and Noordhoff, Alphen aan den Rijn, Holland, 1978.
DasGupta, A., A. Maitra, and S. Basu, "Occurrence of
Nighttime VHF Scintillations near the Equatorial Anomaly Crest in the Indian Sector," Radio Sci., 16: 1455-1458,
1981.
David, P. and J. Voge, Propagation of Waves, Pergamon
Press, London, U.K., 1969.
Davies, K., Ionospheric Radio Propagation, National Bureau of Standards Monograph 80, U.S. Government
Printing Office, Washington, D.C., 1966.
Davies, K., Ionospheric Radio Waves, Blaisdell, Waltham,
Mass., 1969.
Davies, K., "Recent Progress in Satellite Radio Beacon
Studies with Particular Emphasis on the ATS-6 Radio
Beacon Experiment," Space Sci. Rev., 25: 357-430,
1980.
Davis, T.N., "The Morphology of the Auroral Displays of
1957-1958. (2) Detailed Analysis of Alaska Data and
Analysis of High Latitude Data," J. Geophys. Res 67:
75-110, 1962.
Demaro, R.P., "NAVSTAR: The All-Purpose Satellite,"
IEEE Spectrum, 18: 35-40, May 1981.
Donatelli, D.C. and R.S. Allen, "Time Cells for Adaptive
Predictions of Total Electron Content," Radio Sci., 16:
261-269, 1981.
Donnelly, R.F., ed., Solar Terrestrial Prediction Proceedings, U.S. Dept. of Commerce, U.S. Government Printing Office, Washington, D.C. 20402
Vol. 1 Prediction Group Reports, No. 003-023-0041-9,
1979.
Vol. 2 Working Group Reports and Reviews, No. 003017-00471-6, 1979.
Vol. 3 Solar Activity Predictions, No. 003-017-004732, 1979.
Vol. 4 Predictions of Terrestrial Effects of Solar Activity, No. 003-017-00479-1, 1980.
Dougherty, J.P. and D.T. Farley, "A Theory of Incoherent
Scattering of Radio Waves by a Plasma," Proc. Roy.
Soc. (London), A259: 79, 1960.
Dubois, D.F. and M.V. Goldman, "Spectrum and anomalous Resistivity for the Saturated Parametric Instability," Phys. Rev. Lett., 28: 218, 1972.
Ducharme, E.D., L.E. Petrie, and R. Eyfrig, "A Method
for Predicting the F1 Critical Frequency," Radio Sci.,
6: 369, 1971.
Ducharme, E.D., L.E. Petrie, and R. Eyfrig, "A Method
of Predicting the F1 Layer Critical Frequency Based on
Zurich Smoothed Sunspot Number," Radio Sci., 8:
837-839, 1973.
Dyson, P.L., J.P. McClure, and W.B. Hanson, "In situ
Measurements of the Spectral Characteristics of Ionospheric Irregularities,"J. Geophys. Res., 79: 1497, 1974.
Eccles, D. and J.W. King, "A Review of Topside Sounder
Studies of the Equatorial Ionosphere" Proc. IEEE 57(6):
1012-1018, 1969.
Evans, J.V., "Theory and Practice of Ionosphere Study by
Thomson Scatter Radar," Proc. IEEE, 57: 496, 1969.
Evans, J.V., "Millstone Hill Radar Propagation Study" Scientific Results," Tech. Rept. 509; Joint Radar Propa-
gation Study, Lincoln Laboratory, MIT, Lexington,

Mass., 1973.
Evans, J.V., "High-Power Radar Studies of the Ionosphere," Proc. IEEE, 63: 12, 1636-1650, 1975.
Eyfrig, R., "Contribution a l'Etude de l'Effect du Geomagnetisme sur le Couche F1," Comptes Rendus Academie des Sciences, 241: 759-761, 1955.
Fejer, J.A., "Scattering of Radiowaves by an Ionized Gas
in Thermal Equilibrium," Can. J. Phys., 38: 1114, 1960;
also J. Geophys. Res., 65: 2635, 1960.
Fejer, J.A., "Radio Wave Probing of the Lower Ionosphere
by Cross-ModulationTechniques," J. Atmos. Terr. Phys.,
32: 597-607, 1970.
Fejer, J.A., "Ionospheric Modification and Parametric Instabilities," Rev. Geophys. Space Phys., 17: 135, 1979.
Feldstein, Y.I. and G.V. Starkov, "Dynamics of Auroral
Belt and Polar Geomagnetic Disturbances," Plan. Space
Sci., 15: 209-229, 1967.
Ferguson, J.A., "Ionospheric Profiles for Predicting Nighttime VLF/LF Propagation,"Tech. Rep. 530, Naval Ocean
Systems Center, San Diego, Calif., 1980.
Field, E.C., "Propagation of ELF Waves Under Normal and
Naturally Disturbed Conditions," J. Geophys. Res., 74:
3639-3650, 1969.
Field, E.C., "Low-Frequency Ground Wave Propagation
Over Narrow Terrain Features," IEEE Trans. Antennas
Propag., 30: 831 836, 1982a.
Field, E.C., "ELF Propagation in Disturbed Environments"
in Medium, Long and Very Long Wave Propagation
AGARD Conf. Proc., 305: 11-1-11-10, ADA113969
1982b.
Field, E.C., "VLF Propagation in Disturbed Environments"
in Medium, Long and Very Long Wave Propagation
AGARD Conf. Proc., 305: 17-1-17-10, ADA113969,
1982c.
Field, E.C. and R.D. Engel, "The Detection of Daytime
Nuclear Bursts Below 150 km by Prompt VLF Phase
Anomalies," Proc. IEEE, 53: 2009-2017, 1965.
Field, E.C. and R.G. Joiner, "An Integral Equation Approach to Long Wave Propagation in a Non-Stratified
Earth-Ionosphere Waveguide," Radio Sci., 14:
1057-1068, 1979.
Field, E.C. and R.G. Joiner, "Effects of Lateral Ionospheric
Gradientson ELF Propagation,"Radio Sci., 17: 693-700,
1982.
Field, E.C. and C.R. Warber, "Calculation of Ionospheric
Electron Density Profilesby Inverting VLF/LF Reflection
Data, 2. Anisotropic Propagation" RADC-TR-84-242,
Rome AirDevelopmentCenter, GriffissAFB,N.Y., 1984.
Field, E.C., M. Lewinstein, and M.A. Dore, "Effects of
Antenna Elevation and Inclination on VLF/LF Signal
Structure RADC-TR-76-375 ADA035510, 1976.
Field, E.C., R.E. Warren, and C.R. Warber, "Calculation
of IonosphericConductivity Profilesby Inverting VLF/LF
Reflection Data, I. Isotropic Propagation," Radio Sci.,
18: 452-460, 1983.
Flattery, T.W., T.F. Tascione, J.A. Secan, and J.W. Taylor, Jr., A Four-Dimensional IonosphericModel (private
communication).
Flood, W.A. "Simultaneous VHF Auroral Backscatter Measurements," J. Geophys. Res., 65: 2261-2268, 1960.
10-103
CHAPTER 10
Forbes, J.M., "Upper Atmosphere Modifications Due to
Chronic Discharges of Water Vapor from Space Vehicle
Exhausts," in Space Systems and their Interactions with
Earth's Space Environment edited by H.B. Garrett and
C.P. Pike, American Institute of Aeronautics and Astronautics, New York, 1980.
Franklin, C.A. and M.A. MacLean, "The Design of Swept
Frequency Topside Sounders," Proc. IEEE, 57(6):
897-929, 1969.
Fremouw, E.J., "Geometrical Control of the Ratio of Intensity and Phase Scintillation Indices," J. Atmos. Terr.
Phys., 42: 775-782, 1980.
Fremouw, E.J. and J.F. Bates, "Worldwide Behavior of
Average VHF-UHF Scintillation," Radio Sci., 6:
863-869, 1971.
Fremouw, E.J. and C.L. Rino, "Modelling of Transionospheric Radio Propagation," RADC-TR-76-35, Rome
Air Development Center, Griffiss AFB, NY, 1976.
Fremouw, E.J., C.L. Rino, A.R. Hessing, and V.E. Hatfield, "A Transionospheric Communication Channel
Model," Quart. Tech. Rep. 7, SRI International (Stanford Research Institute), Menlo Park, Calif., 1977.
Fremouw, E.J. and C.L. Rino, "A Signal-Statistical and
MorphologicalModel of Ionospheric Scintillation," Proc.
AGARD Conf. Operational Modelling of the Aerospace Propagation Environment, Ottawa, Canada,
1978.
Fremouw, E.J., R.L. Leadabrand, R.C. Livingston, M.D.
Cousins, C.L. Rino, B.C. Fair, and R.A. Long, "Early
Results from the DNA Wideband Satellite ExperimentComplex Signal Scintillation," Radio Sci., 13: 167-187,
1978.
Fremouw, E.J., R.C. Livingston, and D.A. Miller, "On the
Statisticsof ScintillatingSignals," J. Atmos. Terr. Phys.,
43: 717-731, 1980.
Fremouw, E.J. and J.M. Lansinger, "Recent High Latitude
Improvement in a Computer Based Scintillation Model,"
Proc. Ionospheric Effects Symposium, 3rd, 1981, Alexandria, Va., edited by J.M. Goodman, Naval Research Lab., Washington, D.C., 1982.
Frisius, J., A. Ehmert, and D. Stratmann, "Effects of High
Altitude Nuclear Tests on VLF-Propagation," J. Atmos.
Terr. Phys., 26: 251-262, 1964.
Galejs, J., Terrestrial Propagation of Long Electromagnetic
Waves, Pergamon Press, Oxford, UK, 1972.
Gallagher, P.B. and R.A. Barnes, "Radio-Frequency Backscatter of Artificial Electron Clouds," J. Geophys. Res.,
68: 2987-3010, 1963.
Gautier, T.N. and D.H. Zacharisen, "Use of Space and
Time Correlations in Short-Term Ionospheric Predictions," Conference Record, 1st Annual IEEE Communications Convention, 671-676, 1965.
Goldman, M.V., "Field Aligned Instability Due to Stimulated Scattering of Intense Radio Waves from Diffusion
Quasimodes," Radio Sci., 9: 1077, 1974.
Goodman, J.M., "A Survey of Ionospheric Models. A Preliminary Report on the Development of an Ionospheric
Model Thesaurus and Users Guide," NRL Memorandum
Report 4830, Naval Research Lab., Washington, D.C.,
1982.
Gordon, W.E., "Incoherent Scattering of Radio Waves by
10-104
Free Electrons with Applications to Space Exploration

by Radar," Proc. IRE, 46: 1824, 1958.
Gordon, W.E., H.C. Carlson, and R.L. Showen, "Ionospheric Heating at Arecibo: First Tests," J. Geophys.
Res., 76: 7808, 1971.
Gowell, R.W. and R.W. Whidden, "Ionospheric Sounders
in Aircraft," AFCRL-68-0369, AD678047, 1968.
Greenwald, R.A., W.L. Ecklund, and B.B. Balsley, "Diffuse Radar Aurora: Spectral Observations of Non-TwoStream Irregularities," J. Geophys. Res., 80: 131-139,
1975.
Griefinger, C. and P. Greifinger, "Approximate Method for
Determinging ELF Eigenvalues in the Earth-Ionosphere
Waveguide," Radio Sci., 13: 831-837, 1978.
Griefinger, "C. and P. Griefinger, "On the Ionospheric Parameters which Govern High Latitude ELF Propagation
in the Earth-Ionosphere Waveguide," Radio Sci., 14:
889-895, 1979.
Grubb, R.N., "The NOAA SEL HF Radar Systems (Ionospheric Sounder)," NOAA Tech. Memo ERL SEL-55
NOAA Environmental Research Laboratories, Boulder,
CO, 1979.
Haerendel, G., "Modification of Ionized Media by Chemical Substances-A Review of Physical Processes," in
Artificial Modification of Propagation Media, edited
by H.J. Albrecht, AGARD Conf. Proc., 192, Brussels,
1976.
Haerendel, G., O.H. Bauer, S. Cakir, H. Foppl, E. Reiger,
and A. Valenzuela, "Coloured Bubbles-An Experiment for Triggering Equatorial Spread F," in Active
Experiments in Space edited by W. Burke, European
Space Agency Report SP-195: 295, 1983.
Hagfors, T., "Density Fluctuations in a Plasma in a Magnetic Field with Applications to the Ionosphere," J. Geophys. Res., 66: 1699, 1961.
Hagg, E.L., E.J. Hewens, and G.L. Nelms, "The Interpretation of Topside Sounder Ionograms," Proc. IEEE,
57: 949-959, 1969.
Haldoupis, C. and G. Sofko, "Doppler Spectrum of 42 MHz
CW Auroral Backscatter," Can. J. Phys., 54: 1571,
1976.
Hargreaves, J.K., The Upper Atmosphere and Solar-Terrestrial Relations, Van Nostrand Reinhold, New York,
1979.
Harrison, R.P., J.L. Heckscher, and E.A. Lewis, "Helicopter Observationsof Very Low Frequency Radio Waves
Over Certain Mountains and Shorelines,"J. Atmos. Terr.
Phys., 33: 101-110, 1971.
Harvey, R.P., R.B. Harrison, V.C. Fields, G.C. Hirst, P.A.
Kossey, and E.A. Lewis, "Rocket Investigations of the
VLF Ionospheric Transmission Window," AFCRL-TR73-0293, AD767185, 1973.
Haydon, G.W. and D.L. Lucas, "Predicting Ionosphere
Electron Density Profiles," Radio Sci., 3: 111, 1968.
Headrick, J.M. and M.I. Skolnik, "Over-the-Horizon Radar
in the HF Band," Proc. IEEE, 62: 664, 1974.
Heckman, G., "Predictions of the Space Environment Services Center," STPP, (Solar Terrestrial Predictions
Workshop) 1: 322 (see R.F. Donnelly, ed.) 1979.
Heckscher, J.L. and E.J. Tichovolsky, "Height-Gain Atlas
for an Elemental Vertical Dipole Above a Flat Earth,"

RADC-TR-81-39, Rome Air Development Center, Griffiss AFB, N.Y., 1981.
Helliwell, R.A., Whistlers and Related Ionospheric Phenomena, Stanford University Press, Stanford, Calif.,
1965.
Herman, J.R., "Spread F and Ionospheric F-Region Irregularities," Rev. Geophys., 4: 244-299, 1966. (Includes
important bibliography.)
Heron, M.L. and L.F. McNamara, "Transequatorial VHF
Propagation through Equatorial Plasma Bubbles," Radio
Sci., 14, 897, 1979.
Hill, D.A., and J.R. Wait, "Excitation of the Zenneck Surface Wave by a Vertical Aperture," Radio Sci., 13:
969-977, 1978.
Hizal, A. and A.F. Fer, "Medium Wave Multi-Section
Groundwave Propagation in Marmara and West BlackSea Region-Numerical Results" in Medium, Long &
Very Long Wave Propagation AGARD Conf. Proc., 305:
31-1-31-7, ADA113969 1982.
Hofstee, J. and P.A. Forsyth, "Ion Acoustic Waves in the
Auroral Plasma," Can. J. Phys., 47: 2797, 1969.
Hollingworth, J., "The Propagation of Radio Waves," J.
IEE, 64: 579, 1926.
Holmgren, G., R. Bostrom, M.C. Kelley, P.M. Kintner,
R. Lundin, U.V. Fahleson, E.A. Bering, and W.R.
Sheldon, "Trigger and Active Release Experiment that
Stimulated Auroral Particle Precipitation and Wave
Emissions," J. Geophys. Res., 85: 5043, 1980.
Hower, G.L. and A.B. Makhijani, "Further Comparison of
Spread-F and Backscatter Sounder Measurements," J.
Geophys. Res., 74: 3723-3725, 1969.
Hower, G.L., D.M. Ranz, and C.L. Allison, "Comparison of HF Radar Echoes and High-Lattitude Spread-F
Measurements," J. Geophys. Res., 71: 3215-3221,
1966.
Hufford, G.A., "An Integral Equation Approach to the
Problem of Wave Propagation Over an Irregular Surface," Quart. Appl. Math., 9: 391-403, 1952.
Hultqvist, B., "Polar Cap Absorption and Ground Level
Effects," in Solar Flare and Research, edited by G. de
Jager and Z. Svestka, p. 216, North Holland, Amsterdam, 1969.
Hunsucker, R.D., "Morphology and Phenomonology of the
High Latitude E and F Regions," STPP, 2:543 (see R.F.
Donnelly, ed.), 1979.
Hunsucker, R.D. (Chairman), "High Latitude E- and FRegion Ionospheric Predictions," STPP, 2: 513 (see R.F.
Donnelly, ed.), 1979.
Hunsucker, R.D. and R.F. Bates, "Survey of Polar and
Auroral Region Effects on HF Propagation," Radio Sci.,
4: 347-365, 1969.
Ichinose, M., R. Maeda and S. Ito, "MUF Short-Term
Prediction for HF Radio Propagation Circuits over Japan
Area," J. Rad. Res. Labs. (Japan), 27: 179, 1980.
Inoue, Y. and S. Horowitz, "Numerical Solution of FullWave Equation with Mode Coupling," Radio Sci., 1:
957, 1968.
Jelly, D.H. and L.E. Petrie, "The High-Latitude Ionosphere," Proc. IEEE, 57: 1005-1018, 1969.
Johler, J.R., "Concerning Limitations and Further Corrections to Geometric Optic Theory for LF, VLF Propa-
gation Between the Ionosphere and the Ground," Radio

Sci., 68D: 67-78, 1964.
Johler, J.R., "Zonal Harmonics in Low Frequency Terrestrial Radio Wave Propagation," NBS Tech. Note TN335,
U.S. Government Printing Office, Washington, D.C.,
1966.
Jones, T.B. and K. Mowforth, "A Review of the Analytical Techniques for Determining the Phase and Amplitude of a VLF Radio Wave Propagating in the EarthIonosphere Waveguide," in Medium, Long, and Very
Long Wave Propagation, AGARD Conf. Proc., 305:
16-1-16-16 ADA 113969, 1982.
Jones, W.B. and R.M. Gallet, "Ionospheric Mapping by
Numerical Methods," ITU Telecomm. J., 12: 260, 1960.
Jones, W.B., R.P. Graham, and M. Leftin, Advances in
Ionospheric Mapping by Numerical Methods, NBS Tech.
Note 337, U.S. Government Printing Office, Washington, D.C., 1966.
Kelley, M.C., K.D. Baker, and J.C. Ulwick, "Late Time
Barium Cloud Striations and Their Possible Relationship
toEquatorial SpreadF,"J. Geophys. Res., 84: 1898, 1979.
Kellog, W.W. "Pollution of the Upper Atmosphere by
Rockets," Space Sci. Rev., 3: 275, 1964.
Kelso, J.M., Radio Propagation in the Ionosphere, McGraw Hill, New York, 1964.
Kersley, L., "An Empirical Model of Ionospheric Slab
Thickness," AGARD-CPP-284, May 1980.
King, R.J. and J.R. Wait, "Electromagnetic Groundwave
Theory and Experiment," Symposia Mathematica (Instituto Nazionale di Alta Mathematica, Bologna, Italy),
18: 107-208, 1976.
Klobuchar, J.A., "Numerical Models of Total Electron Content over Europe and the Mediterranean and Multi-Station ScintillationComparisons," AGARDOGRAPH:AG166A, Nov. 1973.
Klobuchar, J.A., "A First-Order, Worldwide, Ionospheric
Time Delay Algorithm,"
AFCRL-TR-75-0502,
ADA018862, Sep 1975.
Klobuchar, J.A. and R.S. Allen, "A First-Order Prediction
Model of Total-Electron-Content Group Path Delay for
a Midlatitude Ionosphere," AFCRL-70-0403, AD711365
1970.
Klobuchar, J.A., M.J. Buonsanto, M.J. Mendillo, and J.M.
Johanson, "The Contribution of the Plasmasphere to
Total Time Delay" in Effects of the Ionosphere on Space
and Terrestrial Systems, edited by J.M. Goodman, U.S.
Government Printing Office stock no. 008-051-000691, 1978.
Klobuchar, J.A. and J.M. Johanson, "Correlation Distance
for Mean Daytime Electron Content," AFGL TR-770185, ADA048117, 22 Aug 1977.
Klobuchar, J.A., K.N. Iyer, H.O. Vats, and R.G. Rastogi,
"A Numerical Model of Equatorial and Low Latitude
Total Electron Content for use by Satellite Tracking
Systems for Ionospheric Corrections," Indian J. Radio
Space Phys., 6, 1977.
Knight, P., "MF Propagation: A Wave-Hop Method for
Ionospheric Field-Strength Prediction," BBC Engineering, 100: 22-34, 1975.
Knight, P., "Medium Frequency Propagation: A Survey,"
in Medium, Long, and Very Long Wave Propagation,
10-105
CHAPTER 10
AGARD Conf. Proc., 305: 28-1-28-17, ADA 113969
1982.
Kofsky, I.,"Further Evaluationsof Infrared SimulationData,"
Tech. Rept. DNA Contract 001-81-C-003, Photo Metrics, Inc., Woburn, Mass., 1981.
Kohnlein, W., "Electron Density Models of the Ionosphere," Rev. Geophys. Space Phys., 16: 341, 1978.
Kossey, P.A. and E.A. Lewis, "Detection and Classification
of Solar X-Ray Flares Using VLF Phase and Amplitude
Information," AFCRL-TR-74-0468, ADA003383 1974.
Kossey, P.A., E.A. Lewis, and E.C. Field, "Relative Characteristicsof TE/TM Waves Excited by Airborne VLF/LF
Transmitters," in Medium, Long, and Very Long Wave
Propagation, AGARD Conf. Proc., 305: 19-1-19-10
ADA 113969 1982.
Kossey, P.A., J.E. Rasmussen, and E.A. Lewis, "VLF
Pulse Ionosounder Measurements of the Reflection
Properties of the Lower Ionosphere," COSPAR, Lower
Ionosphere Structure, 133-138, Akademie Verlag, Berlin, 1974.
Kossey, P.A., J.P. Turtle, R.P. Pagliarulo, W.I. Klemetti,
and J.E. Rasmussen, "VLF Reflection Properties of the
Normal and Disturbed Polar Ionosphere in Northern
Greenland," Radio Sci., 18: 907-916, 1983.
Koster, J.R., "Some Measurements of the Irregularities Giving Rise to Radio-Star Scintillations at the Equator," J.
Geophys. Res., 68: 2579-2590, 1963.
Kouba, J. "A Review of Geodetic and Geodynamic Satellite
Doppler Positioning, Rev. Geophys. and Space Phys.
21:27-40, 1983.
Kouris, S.S. and L.M. Muggleton, "Diurnal Variation in
the E Layer Ionization," .. Atmos. Terr. Phys., 35:
133-139, 1973a.
Kouris, S.S. and L.M. Muggleton, "A Proposed Prediction
Method for Monthly Median foE," CCIR Report 2522, Contrib. No. 6/3/07 to Interim Working Party 6/3,
1973b.
Krause, G.E., R.J. D'Accardi, and E.L. Roswell III, "Field
Test of a Near-Real-TimeIonosphericForecasting Scheme
(60 km)," U.S. Army ECOM Rept. 3345, 1970.
Krause, G.E., R.J. D'Accardi, and E.L. Roswell 111,"Field
(200 km)," U.S. Army ECOM Rept. 4144, 1973a.
Krause, G.E., R.J. D'Accardi, and E.L. Roswell III, "Field
(500 km)," U.S. Army ECOM Rept. 4145, 1973b.
Lakshmi, D.R., S. Aggarwal, P.K. Pasricha, and B.M.
Reddy, "HF Communication Problems at Low Latitudes
due to Steep Spatial and Temporal Gradients," STPP,
4: D2-58 (see R.F. Donnelly, ed.), 1980.
Larson, A.C. and J.C. Hodges, "VHF/UHF Auroral Radar
Measurements," Stanford Research Institute International, Calif., June 1967.
Leadabrand, R.L., A.G. Larson, and J.C. Hodges, "Preliminary Results on the Wavelength Dependence and
Aspect Sensitivity of Radar Auroral Echoes Between 50
and 3000 MHz," .1. Geophys. Res., 72: 3877-3887,
1967.
Leadabrand, R.L., J.C. Schlobohm, and M.J. Baron, "Simultaneous Very High Frequency and Ultra High Fre-
10-106
quency Observations of the Aurora at Fraserburgh, Scotland," J. Geophys. Res., 70: 4235-4284, 1965.
Lee, M.C., A. DasGupta, J.A. Klobuchar, S. Basu, and
S. Basu, "Depolarization of VHF Geostationary Satellite
Signals near the Equatorial Anomaly Crest," Radio Sci.,
17: 399, 1982.
Leftin, M., "Numerical Representation of Monthly Median
Critical Frequencies of the Regular E Region (FOE),"
OT Rept. 76-88k, U.S. Office of Telecommunications,
Boulder, Colo., 1976.
Leftin, M., S.M. Ostrow, and C. Preston, "Numerical Maps
of foEs for Solar Cycle Minimum and Maximum,"
ESSA Tech Rept. ERL 73-ITS-63, Boulder, Colo.,
1968.
Leiphart, J.P., R.W. Zeek, L.S. Bearce, and E. Toth, "Penetration of the Ionosphere by Very-Low-Frequency Radio Signals-Interim Results of the LOFTI I Experiment,"
Proc. IRE, 50: 6-17, 1962.
Lewis, E.A. and P.A. Kossey, "POWERFLUX I: A Method
of Estimating Wave Intensities at Large Distances from
Ground-Based Low Frequency Transmitters," AFCRLTR-75-0338, ADA014787, 1975.
Lewis, E.A., J.E. Rasmussen, and P.A. Kossey, "Measurements of Ionospheric Reflectivity from 6 to 35 kHz,"
J. Geophys. Res., 78: 3903-3911, 1973.
Lewis, R.L., "The Wave-Hop fields for an Inclined Dipole
Over a Spherical Earth with an Anisotropic Ionosphere,"
Tech Rept. OT/ITS RR 5, Institute for Telecommunication Sciences, Boulder, Colo., 1970.
Lied, F., (ed.), H.F. Radio Communications, Technivision,
Maidenhead, UK, 1967.
Lincoln, J.V. and R.O. Conkright (eds.), "International
Reference Ionosphere-IRI 79," Report UAG-82, World
Data Center A for Solar-Terrestrial Physics, Boulder,
Colo., 1981.
Llewellyn, S.K. and R.B. Bent, "Documentation and Description of the Bent Ionospheric Model," AFCRL-TR73-0657, AD 772733, 1973.
Lloyd, J.S., G.W. Haydon, D.L. Lucas, and L.R. Teters,
Estimating the Performance of Telecommunication Systems Using the Ionospheric Transmission Channel, National Telecommunications and Information Administration, Boulder, Colo., 1978.
Lyakhova, L.N., "Short-Range Forecasting of the State of
the Ionosphere," Trudy IZMIRAN, 17: 240, 1960.
Lyakhova, L.N. and L.I. Kostina, "On Quantitative Ionospheric Forecasting," Geomag. Aeron., 13: 50, 1973.
Maeda, K.I., "Study on Electron Density in the Lower
Ionosphere,"J. Geomag. Geoelect., 23: 133-159, 1971.
Malik, C. and J. Aarons, "A Study of Auroral Echoes at
19.4 Megacycles per Second," J. Geophys. Res., 69:
2731-2736, 1964.
Marmo, F.F., L.M. Ashcenbrand, and J. Pressman, "Artificial Electron Clouds-1, Planet," Space Sci., 1: 277,
1959.
Martin, E. and J. Aarons, "F-Layer Scintillation and the
Aurora," J. Geophys. Res., 82: 2717, 1977.
McNamara, L.F., "Ionospheric Predictions on Transequatorial Circuits," Proc. IREE (Aust.), 35: 17, 1974a.
McNamara, L.F., "Unusual Nighttime Propagation Modes

on the Guam and Manila-N.W.
Cape Circuits,"
Iono-
spheric Prediction Service Series R Reports. IPS-R27,

Sydney, Australia, 1974b.
McNamara, L.F., "Short-Term Forecasting of foF2," lonospheric Prediction Service Series R Reports, IPS-R33,
Sydney, Australia, 1976.
McNamara, L.F., "The Use of Ionospheric Indices to Make
Real and Near Real Time Predictions of foF2 Around
Australia," STPP, 1: 249 (see R.F. Donnelly, ed.), 1979.
Mendillo, M., "The Effect of Rocket Launches on the lonosphere," Adv. Space Res., 1: 275, 1981.
Mendillo, M. and J. Baumgardner, "Optical Signature of
an IonosphericHole," Geophys. Res. Lett., 9: 215, 1982.
Mendillo, M. and J.M. Forbes, "Artificially Created Holes
in the Ionosphere," J. Geophys. Res., 83: 151, 1978.
Mendillo, M. and J. Forbes, "Theory and Observation of a
Dynamically Evolving Negative Ion Plasma," J. Geophys. Res. 87: 8273, 1982.
Mendillo, M., G.S. Hawkins, and J.A. Klobuchar, "A LargeScale Hole in the Ionosphere Caused by the Launch of
Skylab," Science, 187: 343, 1975a.
Mendillo, M., G.S. Hawkins, and J.A. Klobuchar, "A Sudden Vanishing of the Ionospheric F-Region Due to the
Launch of Skylab," J. Geophys. Res., 80: 2217, 1975b.
Mendillo, M. D. Rote, and P. Bernhardt, "Preliminary Report on the HEAO-Hole in the Ionosphere," Eos, Trans.
AGU, 61: 529, 1980.
Mikkelsen, I.S., J. Aarons, and E. Martin, "Geometrical
Considerations of 136 MHz Amplitude Scintillation in
the Auroral Oval," J. Atmos. Terr. Phys., 40: 479-483,
1978.
Milliken, R.J. and C.J. Zaller, "Principle of Operation of
NAVSTAR and System Characteristics," Navigation. J.
Institute of Navigation,
25(2): 95-106 Summer 1978.
Millington, G., "Ground-Wave Propagation Over an Inhomogeneous Smooth Earth," Proc. IEE, 96, Part III:
53-64, 1949.
Millman, G.H. Modern Radar, ed. R.S. Berkowitz, Wiley,
New York, 1965.
Millman, G.H., "An Evaluation of HF Ionospheric Backscatter Echoes," G.E. Tech. Information Series Rept.
No. R75EMH19, Nov 1975.
Millman, G.H. and C.D. Bell, "Ionospheric Dispersion of
an FM Electromagnetic
Pulse," IEEE Trans. Antennas
Propag., AP-19(1), Jan 1971.

Millman, G.H. and G.M. Reinsmith, "An Analysis of the
Incoherent Scatter-Faraday Rotation Technique for Ionospheric Propagation Error Correction," Rept. No.
R74EMH2, G.E., New York, Feb 1974.
Minnis, C.M. and G.H. Bazzard, "A Monthly Ionospheric
Index of Solar Activity Based on F2-Layer Ionization
at Eleven Stations," J. Atmos. Terr. Phys., 18: 297,
1960.
Mitra, A.P., Ionospheric Effects of Solar Flares, D. Reidel,
Dordrecht, Holland, 1974.
Morfitt, D.G., "Effective Electron Density Distributions
Describing VLF/LF PropagationData," Tech. Rept. 141,
Naval Ocean Systems Center, San Diego, Calif., 1977.
Morfitt, D.G., J.A. Ferguson, and F.P. Snyder, "Numerical
Modeling of the Propagation Medium at ELF/VLF/LF,"
in Medium, Long and Very Long Wave Propagation,
AGARD Conf. Proc., 305: ADA113969 32-1-32-14,

1982.
Muggleton, L.M., "Solar Cycle Control of Nm(E)," J. Atmos. Terr. Phys., 33: 1307, 1971.
Mullen, J.P., "Sensitivity of Equatorial Scintillation to
Magnetic Activity," J. Atmos. Terr. Phys., 35: 1187,
1973.
Muldrew, D.B., "F-Layer Ionization Troughs Deduced from
Alouette Data," J. Geophys. Res., 70(11): 2635-2650,
1965.
Narcisi, R.S., "Overview of Project BIME (Brazil Ionospheric Modification Experiments)," in Active Experiments in Space edited by W. Burke, European Space
Agency Report SP-195, p. 295, 1983.
Newman, P. (ed.), "Spread F and its Effects upon Radiowave Propagation and Communication," Technivision,
Maidenhead, UK, 1966. (Also AGARDOGRAPH 90).
Nielsen, E. and J. Aarons, "Satellite Scintillation Observations over the Northern High Latitude Regions," J.
Atmos. Terr. Phys., 36: 159-165, 1974.
Nielson, D.L. and M. Crochet, "Ionospheric Propagation of
HF and VHF Radio Waves Across the Geomagnetic
Equator," Rev. Geophys. Space Phys., 12: 688, 1974.
Nisbet, J.S., "On the Construction and use of a Simple
Ionospheric Model," Radio Sci., 6: 437, 1971.
Norton, K.A., "The Calculation of Ground-Wave Field Intensity Over a Finitely Conducting Spherical Earth,"
Proc. IRE, 29: 623-639, 1941.
Noxon, J.F. and A.F. Johanson, "Effects of Magnetically
Conjugate Photoelectronson OI (6300 A)," Planet. Space
Sci., 18: 1367, 1970.
Oelbermann, E., "Solar Particle Effects on Polar Cap VLF
Propagation,"
J. Franklin Inst., Special Issue on the
Polar Cap, 290: 281 296, 1970.

Oksman, J., H.G. Moller, and R. Greenwald, "Comparisons between Strong HF Backscatter and VHF Radar
Aurora," Radio Sci., 14: 1121-1133, 1979.
Oran, E.S. and T.R. Young, "Numerical Modeling of Ionospheric Chemistry and Transport Processes," J. Phys.
Chem., 81: 2463, 1977.
Oran, E.S., T.R. Young, D.V. Anderson, T.P. Coffey,
P.C. Kepple, A.W. Ali., and D.F. Strobel, A Numerical
Model of the Mid-Latitude Ionosphere,
Naval Research
Laboratory Memorandum Rept. No. 2839, 1974.

Orsak, L.E., L.H. Rorden, G.B. Carpenter, and B.P. Ficklin, "VLF Propagation and Noise in the Ionosphere Observed by Sounding Rockets," Final Report on Contract
NASr-49(01), Stanford Research Institute, Menlo Park,
Calif., 1965.
Pappert, R.A., "A Numerical Study of VLF Mode Structure
and Polarization Below an Anisotropic Ionosphere," Radio Sci., 3: 219-233, 1968.
Pappert, R.A., "Effects of a Large Patch of Sporadic E on
Nighttime Propagation at Lower ELF," J. Atmos. Terr.
Phys., 42: 417-425, 1980.
Pappert, R.A., "LF Daytime Earth Ionosphere Waveguide
Calculations," Tech. Rept. 647, Naval Ocean Systems
Center, San Diego, Calif., 1981.
Pappert, R.A. and J.E. Bickel, "Vertical and Horizontal
10-107
CHAPTER 10
VLF Fields Excited by Dipoles of Arbitrary Orientation
and Elevation," Radio Sci., 35: 1445-1452,
1970.
Pappert, R.A. and W.F. Moler, "A Theoretical Study of

ELF Normal Mode Reflection and Absorption Produced
by Nighttime Ionospheres," J. Atmos. Terr. Phys., 40:
1031-1045, 1978.
Patenaude, J., K. Bibl, and B.W. Reinisch, "Direct Digital
Graphics," American Lab., 9:5-101, 1973.
Paul, A.K., "Processing of Digital Ionograms," NOSCTD-529, Naval Ocean Systems Center, San Diego,
1982.
Penndorf, R., "Geographic Distribution of Spread-F in the
Arctic," J. Geophys. Res., 67: 2274 2288, 1962.
Perkins, F.W. and P.K. Kaw, "On the Role of Plasma
Instabilities in Ionospheric Heating by Radio Waves,"
Perkins, F.W., C. Oberman, and E.J. Valeo, "Parametric
Instabilities and Ionospheric Modification," J. Geophys.

Res., 79: 1478, 1974.
Peterson, A.M., O.G. Villard, R.L. Leadabrand, and P.B.
Gallagher, "Regularly Observable Aspect-Sensitive Radio Reflections from Ionization Aligned with the Earth's
Magnetic Field and Located within the Ionosphere Layer
at Middle Latitudes," J. Geophys. Res., 60: 497-512,

1955.
Petrie, L.E. and E.S. Warren, "The Propagation of High
Frequency Waves on the Winnipeg-ResoluteBay Oblique
in Ionospheric
Radio Communica-
tions, Plenum, New York, 1968.

Picquenard, A., Radio Wave Propagation, MacMillan, 1974.
Piggot, R.W. and K. Rawer, "URSI Handbook of Ionogram
Interpretationand Reduction," Report UAG-23and UAG23A, World Data Center A for Solar-Terrestrial Physics,
NOAA, Boulder, Colo., 1972.
Pike, C.P., "An Analytical Model of the Main F-Layer
Trough," AFGL-TR-76-0098, ADA026031, 1976.
Pitteway, M.L.V., "The Numerical Calculation of Wave
Fields, Reflection Coefficientsand Polarizationsfor Long
Radio Waves in the Lower Ionosphere, I," Phil. Trans.
Roy. Soc., A275: 219-241, 1965.
Phelps, A.D.R. and Sagalyn R.C. "Plasma Density Irregularities in the High Latitude Topside Ionosphere," J.
Geophys. Res., 81: 515 1976.
Pongratz, M.B. and G.M. Smith, "The LAGOPEDO Experiments-An Overview," Eos, Trans. AGU, 59: 334,
1978.
Pope, J.H. and R.B. Fritz, "Observations of Artificially
Produced Scintillation using Satellite Transmissions,"
J. Geophys. Res., 79: 1074, 1974.
Potemra, T.A. and T.J. Rosenberg, "VLF Propagation Disturbances and Electron Precipitation at Mid-Latitudes,"
J. Geophys. Res., 78: 1572-1580, 1973.
Proceedings IEEE, 57(6), 1969.
Radinov, Ya. S., "Autocorrelative Characteristics of Critical Frequency and Ionization Density Fluctuations,"
Geomag. Aeron., 3: 985, 1963.
Radio Science, 2(10), 1967.
Radio Science, 9(11), 1974.

10-108
tions of the IRI-78 Model, presented at COSPAR Gen-
eral Assembly, Bangalore, India, 1979.

Rasmussen, J.E., P.A. Kossey, and E.A. Lewis, "Evidence
of an Ionospheric Reflecting Layer Below the Classical
D region," J. Geophys. Res., 85: 3037-3044, 1980.
Rasmussen, J.E., P.A. Kossey, and J.P. Turtle, "VLF/LF
Pulse Reflections from Layers Below the Ionospheric
D-Region," AGARD Conf. Proc., 305: 5-1-5-10 ADA
113969, 1982.
Rastogi, R.G., "Seasonal and Solar Cycle Variations of
Equatorial Spread-F in the American Zone," J. Atmos.
Terr. Phys., 42: 593-597, 1980.
Ratcliffe, J.A., "The Magneto-Ionic Theory and its Application to the Ionosphere," Cambridge University Press,
London, 1959.
J. Geophys. Res., 76: 282, 1971.
Perkins, F.W. and E.J. Valeo, "Thermal Self-Focusing of

Electromagnetic Waves in Plasma," Phys. Res. Lett.,
32: 1234, 1974.
Sounder Circuit,"
Ramakrishnan, S., D. Bilitza, and H. Thiemann, Limita-
Ratcliffe, J.A., Sun, Earth and Radio, World University

Library, London, 1970.
Rawer, K., "Propagation of Decameter Waves (HF-Bands)"
in Dynamical and Chemical Coupling between the Neu-
tral and Ionized Atmosphere, edited by B. Grandal and

J.A. Holtlet, Reidel, Dordrecht, Holland, 1963.
Rawer, K., "The Historical Development of Forecasting
Weather for Ionospheric Propagation of HF Waves,"
Radio Sci., 10(7): 669-679, 1975.
Rawer,
K., International
Reference
Ionosphere-IRI-79,
edited by J.V. Lincoln and R.O. Conkright, World Data

Center A, for Solar Terrestrial Physics (STP), NOAA,
Boulder, Colo., 1981.
Rawer, K. and K. Suchy, "Radio Observations of the Ionosphere," in Encyclopedia of Physics, edited by S. Flugge,
Springer-Verlag, New York, 1967.

Reddy, C.A. and M.M. Rao, "On the Physical Significance
of the Es Parameters fbEs, fEs and foEs," J. Geophys.
Res., 73: 215, 1968.
Reder, F.H., "VLF Propagation Phenomena Observed During Low and High Solar Activity," XVI General Assembly of URSI, Ottawa, Canada, 1969.
Reilly, M.H., L.O. Harnish, and J.M. Goodman, "Polarimetry Studies of Ionospheric Modifications by Rocket
Boosters," NRL Mem. Rpt. No. 4517, Naval Research
Laboratory, Washington, D.C., 1981.
Reinisch, B.W. and H. Xuequin, "Automatic Calculations
of Electron Density Profiles from Digital Ionograms. 1.
Automatic O and X Trace Identification for Topside
Ionograms," Radio Sci., 17: 421-434, 1982.
Reinisch, B.W. and H. Xuequin, "Automatic Calculation
of Electron Density Profiles from Digital Ionograms. 3.
Processing of Bottomside Ionograms," Radio Sci., 18:
477-492, 1983.
Reinisch, B.S., J.S. Tang, W.N. Hall, and J. Buchau,
"Automatic lonogram Processing Shows Seasonal Variations in the Auroral Ionosphere," AFGL-TR-82-0122,
ADA 119911, 1982a.
Reinisch, B.W., J.S. Tang, and R.R. Gamache, "Automatic
Scaling of Digisonde lonograms," Test and Evaluation
Report, AFGL TR-82-0324, ADA 121666, 1982b.
Richmond, A.D. and S.V. Venkateswaran, "Geomagnetic
Crochets and Associated Ionospheric Current Systems,"
Radio Sci., 6: 139, 1971.
Rino, C.L. and S.J. Matthews, "On the Morphology of

Auroral Zone Radio Wave Scintillation," J. Geophys.
Res., 85: 4139, 1980.
Rino, C.L. and J. Owen, "The Structure of Localized Nighttime Auroral Zone Enhancements," J. Geophys. Res.,
85: 2941-2948, 1980.
Rino, C.L. and J. Owen, "On the Temporal Coherence Loss
of Strongly Scintillating Signals," Radio Sci., 16: 31,
1981.
Rino, C.L., R.C. Livingston, and S.J. Matthews, "Evidence for Sheet-Like Auroral Ionospheric Irregularities," Geophys. Res. Lett., 5: 1039, 1978.
Rishbeth, H. and O.K. Garriott, Introduction to Ionospheric
Physics, Academic Press, 1969.
Rosenberg, N.W. (ed.) PROJECT FIREFLY 1962-1963,
AFCRL-64-364, 1964.
Rosich, R.K. and W.B. Jones, "The Numerical Representation of the Critical Frequency of the F1 Region
of the Ionosphere," OT Rept. 73-22, Boulder, Colo.,
1973.
Rote, D.M., "Proceedings of the Workshop on Ionospheric
and Magnetospheric Effects of Satellite Power Systems
(SPS)," Conf. Rept. No. 7807115, U.S. Department of
Energy, December 1979.
Rufenach, C.L., "Radio Scintillation of Stellar Signals during Artificial Ionospheric Modification," J. Geophys.
Res., 78: 5611, 1973.
Rush, C.M., "Improvements in Ionospheric Forecasting Capability," AFCRL-72-0138, AD742258 1972.
Rush, C.M., "An Ionospheric Observation Network for use
in Short-Term Propagation Predictions," Telecomm. J.,
43: 544, 1976.
Rush, C.M. and J. Gibbs, "Predicting the Day-to-Day Variability of the Mid-Latitude Ionosphere for Application
to HF Propagation Predictions," AFCRL-TR-73-0335,
AD764711 1973.
Rush, C.M., M. Pokempner, D.N. Anderson, F.G. Steward, and J. Perry, "Improving Ionospheric Maps using
Theoretically Derived Values of foF2," Radio Sci., 18:
95, 1983.
Russell, C.T. and M.J. Rycroft (eds.), Active Experiments
in Space Plasmas, Proc. Sym. 9, COSPAR XXIII, Budapest, 2-14 June 1980, Pergamon Press, 1980.
Sagalyn, R.C., M. Smiddy, and J. Wisnia, "Measurements
and Interpretation of Ion Density Distribution in the
Daytime F-Region," J. Geophys. Res., 68: 199, 1963.
Sagalyn, R.C., M. Smiddy, and M. Ahmed, "High Latitude Irregularities in the Topside ionosphere based on
ISIS-ThermalProbeData,"J. Geophys. Res., 79:4252,
1974.
Saltpeter, E.E., "Electron Density Fluctuations in a Plasma,"
Phys. Rev., 120: 1528, 1960; also J. Geophys. Res.,
65: 1851, 1960.
Shea, M.A., S.A. Militello, and H.E. Coffey, "Directory
of Solar-Terrestrial physics Monitoring Stations,"
MONSEE Special Publication No. 2, AFGL TR 840237,1984.
Shellman, C.H., "Electron Density Distributionsin the Lower
Ionosphere with Associated Error Limits Derived from
VLF and LF Sounder Data," Radio Sci., 5: 1127-1135,
1970.
Shimazaki, T., "World Wide Variations in the Height of
the Maximum Electron Density of the Ionospheric F2

Layer," J. Radio Res. Labs. (Japan), 2: 86, 1955.
Shlyger, I.S., "Self-Modulation of a Powerful Electromagnetic Pulse Reflection from the Upper Layers of the
Ionosphere," JETP (Engl. Transl.), 19: 162, 1974.
Singh, M., E.P. Szuszczewicz, and J.C. Holmes, "The
STP/S3-4 Satellite Experiment," Equatorial F-Region
Irregularities," Proc-Ionospheric Effects Symposium
1981, Alexandria, Va. edited by J.M. Goodman, Naval
Research Lab., Washington D.C., i982.
Singleton, D.G., "The Morphology of Spread F Occurrence
over Half a Sunspot Cycle," J. Geophys. Res., 73:
295-308, 1968.
Singleton, D.G., "Dependence of Satellite Scintillations on
Zenith Angle and Azimuth," J. Atmos. Terr. Phys., 32:
789-803, 1970.
Singleton, D.G., "Transionsopheric Propagation Prediction. An ImprovedIonosphericIrregularityModel," STPP,
4: D1-l (see R.F. Donnelly, ed.), 1980.
Sinno, K. and M. Kan, "Ionospheric Scintillation and Fluctuations of Faraday Rotation Caused by Spread-F and
Sporadic-E over Kokobunji, Japan," J. Radio Res. Lab.,
27: 53-77, 1980.
Sommerfeld, A., "The Propagation of Waves in Wireless
Telegraphy," Ann. Physik, 28: 665, 1909.
Sprenger, K. and K. Glode, "Some Properties of Radio
Aurorae in Medium Latitudes," J. Atmos. Terr. Phys.,
26: 193-198, 1964.
Storey, L.R.O., "An Investigation of Whistling Atmospherics," Phil. Trans. Roy. Soc., London, A246:
113-141, 1953.
Stratton, J.A., Electromagnetic Theory, McGraw-Hill, New
York, 1941.
Strobel, D.F. and M.B. McElroy, "The F2 Layer at Middle
Latitudes," Planet. Space Sci., 18: 1181, 1970.
Stubbe, P., "Simultaneous Solution of the Time Dependent
Coupled Continuity Equations, and Equations of Motion
for a System Consisting of a Neutral Gas, and Electron
Gas, and a four Component Ion Gas," J. Atmos. Terr.
Phys., 32: 865, 1970.
Svennesson, J., "Effects on VLF Propagation During Ionospheric Substorms," J. Atmos. Terr. Phys., 35: 761-773,
1973.
Swenson, E.M., "Aspect-Sensitive Reflections from Ionization Irregularities in the F-Region," J. Atmos. Terr.
Phys., 34: 1469-1476, 1972.
Tascione, T.F., T.W. Flattery, V.G. Patterson, J.A. Secan,
and J.W. Taylor, Jr., "Ionospheric Modelling at Air
Force Global Weather Central," STPP, 1: 367 (see R.F.
Donnelly, Ed.) 1979.
Tellegen, B.D.H., "Interaction Between Radio Waves,"
Nature, 131: 840, 1933.
Thomas, J.O. and M.K. Andrews, "Transpolar Exospheric
Plasma," J. Geophys. Res,, 73: 7407-7417, 1968.
Thomas, J.O. and M.H. Rycroft, "The Exospheric Plasma
During the InternationalYears of the Quiet Sun," Planet.
Space Sci., 18: 41-63, 1970.
Thome, G.D. and F.W. Perkins, "Production of Ionospheric
Striations by Self-Focusing of Intense Radio Waves,"
Phys. Rev. Lett., 32: 1238, 1974.
Thompson, R.L. and J.A. Secan, "Geophysical Forecasting
10-109
CHAPTER 10
at AFGWC," STPP, 1: 355 (see R.F. Donnelly, ed.),
1979.
Thomson, J.J., Conduction of Electricity Through Gases,
p. 321, Cambridge University Press, London, 1906.
Thrane, E.V. (Chairman), "D-Region Predictions," STPP,
2: 573 (see R.F. Donnelly, ed.), 1979.
Tinsley, B.A., A.B. Christensen, J.A. Bittencourt, P.D.
Angreji, and H. Takahashi, "Excitation of Oxygen Permitted Emissions in the Tropical Nightglow," J. Geophys.Res., 78: 1174, 1973.
Titheridge, J.E., "Periodic Disturbances in the Ionosphere,"
J. Geophys. Res., 73: 243-252, 1968.
Trizna, D.B. and J.M. Headrick, "Ionospheric Effects on
HF Over-the Horizon Radar," in Effects of the Ionosphere on Radiowave Systems, Proc. Ionospheric Effects
Electromagnetic Waves from an Electrical Point Source

Round a Finitely Conducting Sphere," Phil. Mag. Series
7,24: 141-176, 1937;24: 825-864, 1937;25:817-834,
1938; and 26: 262-275, 1938.
Vickery, J.F., C.L. Rino, and T.A. Potemra, "Chatanika
Triad Observations of Unstable Ionization Enhancements in the Auroral F-Region," Geophys. Res. Lett.,
7: 789-792, 1980.
Videberg, J.I. and G.S. Sales, "Long Range Survivable MF
Radio Communication Study Using High Altitude Whispering
Gallery
Modes,"
AFCRL-TR-73-0552,
AD774989, 1973.
Villard, O.G. Jr., "The Ionospheric Sounder and Its Place
in the Historyof Radio Science," Radio Sci., 11: 847-860,
1976.
Symposium 1981,Alexandria, Va. edited by J.M. Goodman, Naval Research Lab., Washington, D.C., 1982.
Tsunoda, R.T., "On the Spatial Relationship of 1-mEquatorial Spread F Irregularities and Plasma Bubbles," J.
Geophys. Res. 85: 185, 1980.
Tsunoda, R.T., R.C. Livingston, J.P. McClure, and W.B.
Hanson, "Equatorial Plasma Bubbles" Vertically Elongated Wedges from the Bottomside F-Layer," J. Geophys. Res., 387: 9171, 1982.
Turner, J.F., "The Development of the Ionospheric Index
T," Ionospheric Prediction Science Series R Reports,
IPS-R12, Sydney, Australia, 1968.
Turner, J.F. and P.J. Wilkinson, "A Weekly Ionospheric
Index," STPP, 1: 279 (edited by R.F. Donnelly), 1979.
UAG Report 10, Atlas of Ionograms, (avail. World Data
Center A, U.S. Dept. of Commerce, NOAA/EDIS,
Boulder, Colo.), 1970.
von Flotow, C.S., "Ionospheric Forecasting at Air Force

Global Weather Central," in Effects of the Ionosphere
on Space and Terrestrial Systems, edited by J.M. Goodman, U.S. Government Printing Office, Stock No. 008051-00069-1, Jan 1978.
Wait, J.R., "Radiation from a Ground Antenna," Can. J.
Tech., 32: 1-9, 1954.
Wait, J.R., "Mixed-Path Ground Wave Propagation, I. Short
Distances," J. Res. NBS, 57: 1-15, 1956.
Wait, J.R., "Excitation of Surface Waves on Conducting,
Stratified, Dielectric-Clad, and Corrugated Surfaces,"
J. Res. NBS, 59: 365-377, 1957.
Wait, J.R., "The Electromagnetic Fields of a Horizontal
Dipole in the Presence of a Conducting Half-Space,"
Can. J. Phys., 39: 1017-1028, 1961.
Wait, J.R., Electromagnetic Waves in Stratified Media, Pergamon Press, New York, 1970.
UAG Report 23, URSI Handbook of Ionogram Interpre-
Wait, J.R., (ed.), Electromagnetic
tation andReduction, (avail. World Data Center A, U.S.

Dept. of Commerce, NOAA/EDIS, Boulder, Colo.),
1972.
Probing in Geophysics,
The Golem Press, Boulder, Colo., 163-207, 1971.

Wait, J.R., (ed.) "Special Issue on Extremely Low Frequency (ELF) Communications," IEEE Trans. Com-
UAG Report 23a, URSI Handook of Ionogram Interpretation and Reduction, Revision of Chapters 1-4, (avail.
mun., COM-22, 1974.

Wait, J.R., "Propagation of ELF Electromagnetic
World Data Center A, U.S. Dept. of Commerce, NOAA/

EDIS, Boulder, Colo.), 1978.
UAG Report 50, High Latitude Supplement to the URSI
and Project Sanguine/Seafarer," IEEE J. Ocean. Eng.,

OE-2: 161-172, 1977.
Wait, J.R. and H.H. Howe, "Amplitude and Phase Curves
Handbook of Ionogram Interpretation
and Reduction,
(avail. World Data Center A, U.S. Dept. of Commerce,

NOAA/EDIS, Boulder, Colo.), 1975.
UAG-85 Catalogue of Vertical Ionosphere Soundings Data,
(avail. World Data Center A, U.S. Dept. of Commerce,

NOAA/EDIS, Boulder, Colo.), 1982.
Uffelman, D.R. and L.O. Harnish, "Initial Results from
HF Propagation Studies During Solid Shield," Naval
Research Lab. Memorandum Report 4849, 1982.
Utlaut, W.F., "An Ionospheric Modification Experiment
using Very High Power, High-Frequency Transmission," J. Geophys. Res., 75: 6402, 1970.
Utlaut, W.F. and E.J. Violette, "Further Observations of
Ionospheric Modification by a High-Powered HF Transmitter," J. Geophys. Res., 77: 6804, 1972.
Utlaut, W.F., E.J. Violette, and A.K. Paul, "Some Ionosonde Observations of Ionospheric Modification by Very
High Power High-Frequency Ground-Based Transmission," J. Geophys. Res., 25: 6429, 1970.
Van der Pol, B. and H. Bremmer, "The Diffraction of
10-110
Waves
for Ground Wave Propagation in the Band 200 Cycles
per Second to 500 Kilocycles," NBS Circular 574, National Bureau of Standards, Washington, D.C., U.S.
Government Printing Office, 1956.
Wait, J.R. and A. Murphy, "Influence of a Ridge on the

Low Frequency Ground Wave," J. Res. NBS, 58: 1-5,
1957.
Wait, J.R. and A. Murphy, "Further Studies of the Influence
of a Ridge on the Low Frequency Ground Wave," J.
Res. NBS, 61: 57-60, 1958.
Wait, J.R. and K.P. Spies, "Characteristics of the EarthIonosphere Waveguide for VLF Radio Waves," NBS
Tech. Note 300, National Bureau of Standards, Washington, D.C., U.S. Government Printing Office, 1964.
Wait, J.R. and L.C. Walters, "Curves for Ground Wave
Propagation Over Mixed Land and Sea Paths," IEEE
Trans. Antennas Propag., 11: 38-45, 1963.
Wand, R.H. and J.V. Evans, "Morphology of Ionospheric
Scintillation in the Auroral Zone," in Effect of the Ionosphere on Space systems and Communications, edited

by J. Goodman, NTIS CSCL 04/1 N75-30714, Naval
Research Laboratory, Washington, D.C., 1975.
Wand, R.H. and M. Mendillo, "Incoherent Scatter Observations of an Artificially Modified Ionosphere," J. Geophys. Res., 89: 203, 1984.
Watson, G.N., "The Transmission of Electric Waves Round
the Earth," Proc. Roy. Soc., 95: 546-563, 1919.
Watt, A.D., VLF Radio Engineering, Pergamon Press, New
York, 1967.
Weaver, P.F., "Backscatter Echoes from Field-Aligned Irregularities in the F-Region," J. Geophys. Res., 70:
5425-5432, 1965.
Weber, E.J. and J. Buchau, "Polar Cap F-Layer Auroras,"
Geophys. Res. Lett. 8:125, 1981.
Weber, E.J., J. Buchau, R.H. Eather, and S.B. Mende,
"North-South Aligned Equatorial Airglow Depletions,"
J. Geophys. Res., 83: 712, 1978.
Weber, E.J., J. Buchau, and J.G. Moore, "Airborne Studies
of Equatorial F-Layer Ionospheric Irregularities," J.
Geophys. Res., 85: 4631-4641, 1980.
Weber, E.J., H.C. Brinton, J. Buchau, and J.G. Moore,
"Coordinated Airborne and Satellite Measurements of
Equatorial Plasma Depletions," J. Geophys. Res., 87:
503-510, 513, 1982.
Wernik, A.W., C.H. Liu, and K.C. Yeh, "Model Computations of Radio Wave Scintillation caused by Equatorial Ionospheric Bubbles," Radio Sci., 15: 559, 1980.
Weyl, H., "Propagation of Electromagnetic Waves Above
a Plane Conductor," Ann. Physik, 60: 481-500, 1919.
Whalen, B.A., et al., "Preliminary Results from Project
Waterhole-An Auroral Modification Experiment," Cam.
J. Phys., 59: 1175,1981.
Whitney, H.E., C. Malik, and J. Aarons, "A Proposed
Index for Measuring Ionospheric Scintillation," Planet.
Space Sci., 17: 1069-1073, 1969.
Whitney, H.E., J. Aarons, R.S. Allen, and D.R. Seemann,
"Estimation of the Cumulative Amplitude Probability
Distribution Function of Ionospheric Scintillations," Radio Sci., 7: 1095-1104, 1972.

Wilkinson, P.J., "Prediction Limits for foF2," STPP, 1:
259 (see R.F. Donnelly ed.), 1979.
Wilkinson, P.J., "A Comparison of Monthly Indices of the
Ionospheric F-Region," Ionospheric Prediction Service
Series R Reports, IPS-R41, Sydney, Australia, 1982.
Winkler, C., "Radio Wave Guidance at VHF Through Equatorial Plasma Bubbles," J. Atmos. Terr. Phys., 43: 307,
1981.
Wright, J.W., "Kinesonde Observations of Ionosphere Modification by Intense Electromagnetic Fields from Platteville, Colorado," J. Geophys. Res., 78: 5622, 1973.
Wright, J.W. and A.K. Paul, "Toward Global Monitoring
of the Ionosphere in Real Time by a Modern Ionosonde
Network: The Geophysical Requirements and Technological Opportunity," NOAA Special Report, Space Environment Lab., July 1981.
Wright, J.W. and M.L.V. Pitteway, "Real-Time Data Acquisition and Interpretation Capabilities of the Dynasonde, 1. Data Acquisition and Real Time Display, 2.
Determination of Magnetosonic Mode and Echolocation
using a Small Spaced Receiving Array," Radio Sci., 14,
815-835, 1979.
Wright, J.W. and M.L.V. Pitteway, "Data Processing for
the Dynasonde: The Dopplionogram,"J. Geophys. Res.,
87: 1589, 1982.
Yeh, K.C., "Travelling Ionospheric Disturbances as a Diagnostic Tool for Thermospheric Dynamics," J. Geophys. Res., 77: 709-719, 1972.
Yeh, K.C. and C.H. Liu, "Ionospheric Effects on Radio
Communication and Ranging Pulses," IEEE Trans. Antennas Propag., AP-27(6), Nov 1979.
Yeh, K.C. and C.H. Liu, "Radio Wave Scintillations in the
Ionosphere," Proc. IEEE, 70, 1982.
Yeh, K.C., H. Soicher, and C.H. Liu, "Observations of
Equatorial Ionospheric Bubbles by the Radio Propagation Method," J. Geophys. Res., 84: 6589, 1979.
Zacharisen, D.M., "Space-Time Correlation Coefficients
for use in Short-Term Ionospheric Predictions," NBS
Report 8811, 1965.
Zevakina, R.A. E.V., Lavrova, and L.N. Lyakhova, Manual on Short-Term Predictions of Ionospheric Geomagnetic Storms and Radio-Propagation Forecasting, (text
in Russian) Nauka, Moscow, 1967.

Zinn, J. and C.D. Sutherland, "Effects of Rocket Exhaust
Products in the Thermosphere and Ionosphere," Space
Solar Power Rev., 1: 109, 1980.
Zinn, J., C.D. Sutherland, S. Stove, and L. Duncan, "Ionospheric Effect of Rocket Exhaust Products: HEAO-C
and Skylab," J. Atmos. Terr. Phys., 44: 1143, 1982.
10-111

Ionosferic Radio Propagation - Article

Uploaded by

Copyright:

Available Formats

Ionosferic Radio Propagation - Article

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ionosferic Radio Propagation - Article

Uploaded by

Copyright:

Available Formats

Chapter 10

IONOSPHERICRADIO WAVE PROPAGATION

S. Basu, J. Buchau, F.J. Rich and E.J. Weber

The series of reviews presented is an attempt to introduce

The ionogram (Figure 10-1) shows signatures of various

Finally, we see vertical bands in the frequency range

IONOSPHERIC RADIO WAVE PROPAGATION

Ne = 1.24 x 104 f2n

where fN is in MHz and Ne in electrons/cm3 . The plasma

where fH is the gyrofrequency, the natural frequency at

detailed discussion see Davies [1966] and Chapter 10 of

As stated before, one of the main objectives of ionospheric

that is, the magneto-ionic splitting (due to the presence of

where u'(h,f) is the group refractive index. Therefore, while

the virtual height is related to the group refractive index by

If the electron density NC(h)is considered as a function of

10.1.1.3 Analog Ionosonde. The general principle of an

IONOSPHERIC RADIO WAVE PROPAGATION

and pulling a film

that shown in Figure 10-1. Sine soundersbased on the

primary ionosonte for the

ture of an ionogram trace simultaneously with the digital

IONOSPHERIC RADIO WAVE PROPAGATION

(traces marked B and C) and receding (trace marked A)

the frequency extent of E and Es traces shows the typical

IONOSPHERIC RADIO WAVE PROPAGATION

the FM/CW system allows a substantial reduction of the

instruments (ISIS I, 1969 and ISIS

IONOSPHERIC RADIO WAVE PROPAGATION

ionization of lower layers exceeds the maximum density of

therefore the final appearance of the h'f-trace. This trace

of more than 300 ionosonde stations

10.1.1.9 Ionogram Interpretation. The behavior of the

There are also extreme differences in ionospheric variations

regular or sporadic appearance and disappearance of the

and at certain harmonics of these frequencies [Hagg et al.,

For detailed instructions in the evaluation of ionograms please

Figure 10-8. Line sketchof daytimeionogramshows definitionof importantionogramparameters.

a) Critical and characteristic frequencies

solar produced E layer ordinary wave critical frequency.

Highest frequency at which a continuous Es trace is observed.

IONOSPHERIC RADIO WAVE PROPAGATION

The minimum virtual height of the normal E layer, taken as a whole.

The minimum virtual height of the trace, used to determine foEs.

has been incorporated into INAG as of September 1984.

as operational or operating in 1984. World Data Center A

10.1.2 Incoherent Scatter

per unit volume to be NOc, where N is the electron number

where X is the radar wavelength (m), k is Boltzmann's

IONOSPHERIC RADIO WAVE PROPAGATION

is the permittivity of free space (= 8.85 x 10-12

resides in the ionic componentand the spectral width is

formof the spectrum.By measuringthe heightof the hump

sectiondue to the ionic componentis simplyobtainedfrom

The incoherent scatter radar technique opened up the

functionof delayby usinga verticallydirectedpulsedradar.

The echo power is given by

where C is a constant. The constant C can be determined

Table 10-1. Incoherent scatter facilities.