JOURNAL C F RESEARCH of the National Bureau of Standards- D.

Radio Propagation
Vol. 670, No.5, September-October 1963

Reflection of VLF Radio Waves From an Inhomogeneous

Ionosphere. Part II. Perturbed Exponential Model
1

James R. Wait and Lillie C. Walters

C ontribution From Central Radio P ropagation Laboratory, National Bureau of Stan dard s, Boulder, C olo.
(R eceived April 10, 1963)

Th e obli qu e refl ection of radio waves from a continuously stratified ionized medium
is cons id ered . In th is p ap er t he m ed iu m is assumed to be isot rop ic. T he h eight profile of
the effect ive cond uctivity is a Gaussian curve superimposed on th e (u nd istu rbed ) expone ntia l
form. The refl ection coefficient is shown to be influenced by the verti cal location of the
Ga uss ian perturbat ion . In some cases the magnitude of the reflection coefficient is increased
whi le, in other situat ions, it is d ecrea cd. In nearly all cases, insofa r as p hase is concerned,
t he pre ence of t he p ert ur bation correspon ds to a lowering of the refl ection h eig ht .

1. Introduction in Lerms of Lbe a ngular freq uency w, collis ion fre-

In a previous communication from the present
auth ors [W ait and W alters , 1963], obliqu e reflection
of (VLF) radio waves from a continuously stratified
ionized medium was considered. The profile of the
effective conductivity was taken to be exponen tial
in form. Actu ally, this is a fairly good representation
of the actu al D layer of the ionosphere under day-
time conditions. Henceforth, t,hat paper will be
referr ed to simply as (J) .
It is the pm'pose of the present paper to consider
profiles whi ch are no longer exponential in form.
Since t he objPctive is to gain insight into the mecha-
nism of r efl ection from per turb ed layers, a number
of idealizations are ffittde . First, it is assumed,
under quiescent condi tions, that the ionosph eric
conductiv ity varies exponen tially with height. Then
t h e idealized perturbation is assumed to have a
Gaussian form . Again , f oj' sake of simplicity, the
earth's magnetic field is neglected as in (1 ) . This is
well justified wh en considerin g effects which result
from ionization in the lowest ionosph ere.
2. Description of the Profile
The notation follows that used in (1) as closely as
possible. Thus the undisturbed profil e, as a func-
tion of h eight z, is defined by the conductivity
parameter 1/1-(z) where
1
L( z) =z1 exp ({3 z) ,
and L is a constant, (3 is a gradient paramete r and z
is th e h eight above th e r eference level z= O. Und er
th e iso tr opic assumptio n, it is known Lhat [WaiL,
1962]
w(v +i w)
L (2)
2
Wo
' levels where t1 N( z) d rops to IlNO /L
'rho work in this paper was suppor ted by the Advanced Research Projects
1
Agency, W ashington 25, D.C ., u nde r AHP A Order No . 183-62.
quency v, and plas ma frequen cy woo At VLF,
v>>w, and th er efore
w w~
L~- wh ere W,=-' (3)
Wr v
to within a very good approxim ation .
In general , it is seen t h at L~z) is proportional to
~~~) where N (z) and v( z) ar e the elect ron den sity
a nd collis ion frequency r egard ed as a fun ction of
hei gh t. The constant {3, in t he exponent, is a mea,s-
ure of th e sh arpness of the gmdient. For example,
(3 = 1 km- 1 means th at th e ratio of w,(z) or N (z)/v(z)
increases by 2.71 for each km of ve rtical height.
F rom th e re cent work of Darrin gto n et al. [1962],
Kane [1962], and B elrose [1963], it appears that {3
for an undistul'bed ionosph ere m ay b e in th e range
from 0.2 to 0.8. If t he level from ab out 60 km to
70 km is considered, it appeal' t hat {3 = 0.3 typifies
many of t hese daytime D-Iaye r profiles. A detailed
s tudy of th e influence of cha ngin g {3 is to be fo und
in (1). For this p aper, (3 is chosen Lo be 0.3.
H aving specified our undisturb ed profil e, we now
wish to introduce th e perturbation. It is assumed
that the collision frequency profile is unchan ged
whereas the ionization is Lo b e increased by an
amoun t Il N (z) wh ere
(1 )
IlN(z) = IlNo exp [ _(2DF)] (4)
and t1No, F and Dare co nstanLs. Clearly, the
maximum value of t1N(z) is t1No which is located
at z= F. Furthermore, th e thi clmess of t his layer
is 2D which is the vertical dis tance between the
In order to estimate co rrectly the influence of this
Ga,u ssian sh aped layer , it is necessary to assume
519
something about the collision frequency profile. A The complete profile, under these idealized con-
careful study of the recent li terature indicates that ditions, is given by
an exponential variation of v( z) with height z is not
unreasonable. The form chosen here is

v( z) = Vo exp ( -~ z ). (5)

where 13= 0.3 lml- 1 • Therefore, the resulting con-

ductivity perturbation has the form
f>N(z)=tlNo exp
v( z) Vo
(f!.2 z) exp [_(Z-F)2J.
D
(6) where the right-hand side is propor tional to the
effective conductivity of the m.edium as a function
of height above (or below) the reference level at
z= O. The coefficient A defines the strength of the
_I§ 7 perturbation. In fact,
a:: 6
w
I-
A=f>No,
w
::;: 5 No
<!
a::
ct 4 ). , 15km where No is the electron density of the undisturbed
>- (OR f' 20 kciS)
l- 0= 2km profile at the reference level z= O. In this paper,
S> ·3
i=
A'
p.
2
O,3km"
as in (I ), L (0 )=7.5/'A where A is the wavelength in
u
6z 2 kilometers.
8 It is admitted that other ways to define a pertur-
bation in the profile may be preferable. Here the
electron density anomaly, for a given value of A,
2 6 does not change with its vertical location F. Con-
sequently, we may anticipate that the influence of
FIGURE 1. The undisturbed and disturbed conductivity profiles this type of perturbation will be diminished at
used in this paper. sufficiently low heights because of the increasing

,,
,,
(0)
\ = 15 km
(b) \
0.8 A =2
C = 0..0.5
160
I
,, \
c_
o = 2 km
, \s
- OOS I
I
f3 = 0.,3 km- I
0.7 120
1
10
I UJ
\~
0.1 \ = 15 km
I
I \
g:;
0,6
0/
80
A =2
0= 2 km
\
\
\
I
\
I-
Z
~
f3 = 0.3 km- I \
\
\
\.
u
LL 0.5
0.15 0..15
0>
<lJ
"0 40
\
\
\
\
LL
W
a::
\
\ \
0 \
u
0.2 0.2 LL
\
\ \
z 0.4
Q
l- I
I --,,, 0
W
(f)
«
0 \
\
\
\
\
u
W
..J I
I
I
I
\
\
,
I
0..
\
\
\
\
LL
w 0.3 I \ -40 \
\ '-...-.
a:: ____ <2:3
_-_ ' t "-~'
I 0.3
. '--~--
\
,,
.... ...... /
/

I , \ ____ .9~__ ,
0,2 / / .
-80
c= 0..4
- - ~-----~ 'f-'
// \c- =.0..4 0.,20. 0.20.

0, I
" '/
I

-120
0..15 0.15

0..1 0.,1
(, = o..o.'j

10 10
F 1 km
FIGURES 2a and b . The reflection coefficient as a function of the vertical location, F, of the Gaussian perturbation, fOl' various
angles of incidence.

520
collision frequency. However, we shall see the
problem is not quite this simple as other factors
come into play.
A sketch of the profiles used is given in figure 1.
The undisturbed profile is the ex.'Ponential form,
while the disturbed profiles have the superimposed
Gaussian "bump." The location of the "bump"
for five typical profiles is specified by the appropriate
value of F.
3. Results of the Calculations
The method used to calculate the reflection co-
efficient R has been described in detail in (I). The
quantities considered are the amplitude IRI and the
phase of R for a vertically polarized plane wave
incident at an angle whose cosine is C. The reflec-
tion coefficient is evaluated in the free space region
corresponding to Z----7 - co . However, it is important
to remember that the phase is 1'ejel'l'ed to the level
z=O.
The plan of the calculations is to vary the value
of one parameter while keeping the others constant.
To obtain a complete understanding of the various
phenomena, an enormous number of calculations is
needed . In order to keep the problem within reason-
able bounds and to reduce the expense of the compu-
tation, only a limited number of cases was considered.
These results are shown in graphical form in figures
2 to 6. In all cases 13= 0.3 km- l .
In figure 20, the amplitude of the reflection co-
efficient is plotted as a function of F for A= 15 km
(j= 20 kc/s), A = 2, D = 2 km, and C values varying
from 0.05 to 0.4. Small values of Chel'e correspond
to angles near grazing. For lon o' distance propaga-
tion of VLF radio waves, value of C near 0.1 are
most important. For this case, it is interesting to
note, when F is near or above zero, tha t IRI takes
the same value as for the undisturbed profile. As
the "bump" or perturbed layer is lowered, the reflec-
tion coefficient first increases then decreases. Even-
tually, as the "bump" is brought down to very low
heights, IHI returns to its undisturbed value. The
other curves for highly oblique incidence have a
similar behavior. Thus, the "bump" may either im-
prove or degrade the reflection. Presumably, at the
lower heights the Gaussian layer is acting as an
absorber whereas, at greater heights, it enhances the
reflection. At the steeper angles of incidence, the
situation becomes more complicated. It is probable
that this results from interference between mul tiple
reflected rays between the upper side of the "bump"

180

IQO
0

.I>
0.6 r -- . - - , - - , - - - , - - - r - , - - - - r --,.----,- - . - - r - ----, A ; l5 km
100
c ; 0.2
A ; 2
0'
B ; 0.3 km- I
0.5
~ 60
a:: 0::
I- LL
~ 0.4 o 20
U w
(j)
LL
LL «
W I
8 03
0...
- 20 '
z
o
f= A; 15 km
~ 0.2 -60
-.l C ; 0 .2
LL A ; 2
W
0:: (3 ; 0 .3 km-I
0. 1 -100

~ 50 -10 0 10
-1 40_50
-40 -30 -20 -10 10
F, km

FI GU RI,S 3a a nd b . Th e refl ecti on coeffi cient, as a f 1tnction of F, f or va"ious w idths of the Gaussian pertu bation when C = 0. 2 (i .e.,
angle of inciden ce is 78°).

521
 0.7

~ -80
-0
LL
LL a:
W
8 0.5 ~ -100
A = 15 km w
z if)
o C = 0.1
.<1:

E 04 f3 = 03 km- I iE cl20
w (b)
--.J A = 2
LL
W
a:: 10
03
(0) F. km

FIGURES 4a and b. The Tefleetion eoeifieient, as a f 1!nction of F, fOT variou s widths of the Gaussian perttlrbation when C = 0.1 (i.e.,
angle of incidence is 84°) .

- 20

0.6 -40
ii
f- C = 0.1
~ 0.5 -60 A =2 (b)
~ D = 2 km
LL
LL f3 = 0.3 km- I
W en
8 04 OJ
'0 -80
z a::
0
~
w
03 C = 0..1 LL
0- 100
A = 2
-.J w
LL (0) D = 2 km (f)
w <{
et:: {3 = 0..3 km- I I
0.2 lL -120

0.1 -140

-160
Q60 -50 10 -60 -50 -40 -30 -20 -10 10
F, km

FIGURE S Sa and b. Th e Teflection coefficient, as a function of F , for vaTious wavelengths (from 10 to SO km).

and the eXPQnentially varying layer. Such an inter-
ference phenomenQn becomes more pronounced at
steeper incidence because the vertical compQnent Qf
the wavelength is becoming comparable with typical
values of F.
The phase o.f R is shown in figure 2b for the same
conditions as in figure 2a. Again, it is apparent that,
when the Gaussian "bump" is such that F is near 0
or above, the phase of R attains its undisturbed value.
When the angle of incidence is highly oblique the
phase undergoes an increase (i.e., decrease of lag)
522
as the "bump" comes down to lQwer heights. Suffi-
ciently far below the reference level, the phase o.f R
returns to. its undisturbed value. It is well to. note
that as 0 becomes small (i.e., appro.aching grazing
incidence), the phase of R is approaching - 180°.
FQr highly o.blique incidence the influence of the
"bump" is to lower the effective height of reflection
fQr the who.le range o.f F. Ho.wever, a very interest-
ing phenQmenQn o.ccurs at steeper incidence. As can
be seen in figure 2b, when 0 = 0.2 the phase undergoes
a rather rapid change as F varies fro.m about - 22
 0.6

(0)

0.7

A ~ 0.5
-60
c;:
A~4
f- (b)
z
~ A ~ 15 km
~ C ~ 0.1
LL
LL
-80
W D ~ 2 km
0
U 0- f3 ~ 0.3 km- I
Z
0 A~ 15 km '"
'0

i= C ~ 0. 1
u 0: -100
w D ~ 2 km
--' 0.4 LL
LL
W j3 ~ 0.3 km- I 0
n: UJ
(f)
<t
0.3 it -110

01 50 10 -14~50 10
F, km

FJGUHJeS Ga and b. Th e reflection coe.fJicient, as
km to - 27 km. As 0 is increased fur t her t here is
an /l,pparent ciisco nti nui ty when the phase cba.nges by
360°. S uch a cha nge of 271' i'<tdi,Ul s is quite perm is-
sible sin ce the ord illitte is arbitmry to within [my
integral number of 271' radians. T lms, the pbase
CUlYes for 0 = 0.3 a nd 0.4 co ulcllttwe been dmwn
I in t he ntnge below - 160°.
The CUI.TeS in figure 2b, eve n if t hey s how nothi ng
) else, demonstmte that phase s hifts in reflection
p henome na may h,tve sO lll e unu sual cycle am-
biguities.
The influen ce of the widLit of Lhe Gnu ss i,LI1 pertUl'-
bation 01' "hump" is s how n in figUl'e 3tL ror t be
amplitude IHI and in figure :3b for the p hase of R.
Here A = 2, A= 15 km, and 0 = 0.2. The ampliLud e
curves show t hat, whell D is increased, th e overall
influence of the layer becomes somewhat greftter.
, There is some tendency for th e thinner byers (i .e.,
smaller D ) to be more efl'ect ive at greater heights.
The correspon ding phase curves s how that the
thicker layers always produce a larger pbase change.
Furthermore, as D exceeds 2 km, il, point is reached
whcre th e "360° jump" takes place.
The curves in figures 4a and. b are for the Salne
conditions as figures 3a and b except that now 0 =
0.1, corresponding to nearer grazing incidence.
The amplitude curves have a very similar shape.
The phase curves are also similar except that the
"360° jump" is no longer present. . submittrc\ to Proe. lEE, Londo n).
The wavelen gth dependence or the reflectIOn
coefficient is shown in figures 5a and b. For these
0 = 0.1, A = 2, and D = '2 . The wavelen gths chosen
(10, 15, 20, 25. 30lem) correspond to frequencies of
30, 20, 15, 12 , and 10 kc/s. Qualitatively, the
curves have a very similar shape. There is some
tendency for the shorter wavelengths to be accom-
pfwied by more pronounced changes. In all eases
the "bump" acts as an absorber at low heights
while it enhances thc reflection at greater heights.
687-416- 63- -5
[L jllnction oj 11, ./'01' VCLI' ioIL8 amplitudes A oj the pel'tlll'bat ion .
Finftlly, in figures 6a and b , t.he influence of A ,
the relative magnitude of the anomalous rlectron
density, is showll. As expected , the indi vidual
CLll'ves are similar in sllape wit.h t llC lftrger values of
A cOl'l'esponding to an increased change over the
llndisLurb ed values. J t is importftnt to note that
tbe phase anom aly is alm ost directly proporLional
to A.
4 . Final Remarks
The results gi ven here constitute a small portion
or extensive cOll1put,ttio ns dealing with reflection
or waves from inhomogeneous media. In subse-
quent parts to t.hi s series other types or profiles will
be considered. Al 0, the app lications to the mode
theory o[ VLF propftgation are to be desc]'ibed in
some detail.
The authors wish to thank A. G. Jean and D. D.
Crombi e for t heir helpful suggestions during t ho
course of this work.
5. References
Barrington, R ., B. Landmark, O. Hol t, a nd E . Thrane
(Jun e 1962), Experimental studies of the ionospheric
D-regiolJ, Repor t No. 44, Norwegian Defence R esea rch
Establishment, Kj eller, Norway.
Belrose, J .S. (1 9G3, Low-frequency propagation (to be
Kan e J . A. (1962), Re-evaluation of ionosphrric electron
den sit ies and collision frrqurnc irs drrived from rock et
measurements, Chap. 29 in Radio Wave Absorption in
the I onosphere (P ergamon Press , Oxford) .
'Vait, J. R . (1962) , Elect ro magnet ic ",aves in strat ifi ed
media (Pergamon Prrss, Oxford).
W a it, J . R., and L. C. Walters (May- Jun e 1963), Reflection
of VLF radio wa ves from an inhomogeneous ionosphere.
Part 1. Exponentiall y varyi ng iso tropic mod el, J . R esearch
NBS 67D (Radio Prop.) No.3, :361- 367.
(Paper 67D5- 283)
523