Composition of Reflection and Transmission Formulae: John Heading
Composition of Reflection and Transmission Formulae: John Heading
Composition of Reflection and Transmission Formulae: John Heading
Radio Propagation
Vol. 67D, No. 1, January- February 1963
1. Introduction
In a r ecen t paper , W a i L [19 62] has followed a n icl ea of Bl'ekhovs kikh II960] a nd has ouLain eci
appro ximate for111 s £or t he r eflect ion coeffLcien Ls or VLF a nd ELF radio waves when t h e
ionosphere is no t sh arply bounded. In Lhe isotropic case, Lhe lll eLhod is to obta in a fir st order
nonlinear d ifferen tial equ atio n for R (z), a funct ion of t h e h eigh t z t haL r educes to t he r eflection
coefficient below t h e ioni zed layer. T h is equ ation is solved b.v s uccess ive app roxim ation s,
Lhe fLrst approximation b ein g t he familiar Fl'es nell'efi ec Lio ll coeHicient. Ne it he r BrekllOvskikh
in h is comprehensive text (p . 220) nor 'W ait discusses a ny p hys ical in Lel'preLaLion of the cor-
respondin g in tegral equaLion . H er e, we S llOW LhaL b y m ea ns o r a d ifl'ere n L, yet u l LimaLely
equiv alent, formul ation of the problem , t he r es ul t in g equation s a r e s uscep Lible to ph:ysical
in terpreta tion . T h e t heory is applicable to all a ngles of incide nce a nd to a ni sotropi c iono-
sph er es. T h e inLel'p retfL Lion is made possible by t he recenL inves t iga tion s of vVes LcoLL [1962<1 ] in
wh ich h e h as derived various exp ress ion s y ield ing t he effect or p a rtic ular region s of (h e
ionosph e re on Lbe r efl ected wave .
2 . Isotropic Model
V erti cal prOpltglttion in an isotropic ionosphcre , j nclud in g Lh e efrecL of Lh e coll is ion fr e-
q u c ncy, is govcrn ed by th e diifer c n tilll equaLio ll
(1 )
I:
= (E' +iknjE) e- ikn j(z-r) -Fni i bE e-ikn j(z- Il dz . (2)
Similarly,
(3)
W"e now consider a wave incident below the ionosphere. \iVe therefore restrict ourselves
to tha t solu tion of (1) for which
and
The sum of the two integrals (2) and (3) is now given by
(4)
This constitutes a suitable in tegral equation for the field at any point .I in t he medium ;
the integration process is extended over those portions of the ionized medium where the r e-
fractive index differs from unity or from n t . In particular, if fr ee space exists above Z= b
(that is, if n = 1), the in tegral equation becom es
j
It sho uld b e pointed out that the right-hand sides of both eqs (4) and (5) are independent
of a, provided that a is any point in t he free space below the ionized region.
reducing t.o
( 6)
1 .
R=2 tk
[b1-X ·Z E( z)e - 'k-dz
'. . (7)
• a t
It is obvious that the limits may be replaced by - 00 and + 00 respectively, since the integr and
vanishes in these extended regions.
Similarly, we place 1;= b in eq (4), in which case
Hence,
reducing to
(8)
It should be pointed out that results (6), (7), (8), and (9) are independent of a and b provided
that a and b lie in the respective homogeneous r egions b elow and above the ioni zed m edium.
If there exists no homogeneous medium. of refractive index nl above z= b, then the varying
medium will extend to infinity. In particular, the medium becomes infinitely overdense
there if X --'7 OO as Z --'7 OO. Under these circumstances, we multiply eq (1) by eik (z-n and in-
tegrate from a to I; as before, but now we multiply eq (1) by e- ik(Z-n and integrate from I; to
ex:>. The same analysis as b efore leads to the consideration of
It should be observed that t his factor does not converge at the upper limit if a homogeneous
m edium extends to infinity . It was for this reason that previously we employed the factor
e-ikn1(z- n to secure a result that vanished when 8= b. For convergence in the present case,
we r eq uire E' +ikE--'70 as Z--'7 OO . This limit certainly exists if the region becom es infinitely
overden se, since the allowed W.K.B.J. solution for E then becomes exponentially small in
magni tude as Z--'7 OO. We obtain the r esult
where a occurs below the ionization. No formula for T exists in this case.
5. Physical Interpretation
There are three distinct contributions to the field Em in eq (4). These are:
(i) 2e- ikr / (1 + nl);
(ii) elements of the form 1 tnl . 1 ~iz E (z) oz eik(Z- r) originating below the height L
(iii) elements of the form 1 tnl C 3Z-1 ~z) E (z) oze- ikn1 (Z- rJ originating above the
height \.
The vector sum of all these waves yields the total electric field at the height r. (i) is a
kind of transmitted wave complete with the appropriate Fresnel transmission coefficient.
(ii) represents a plane wave propagated vertically upwards as in free space (according to the
factor e- ikr ), originating from the elementary layer of thickness oZ situated at z< 5' All such
layers below 5" yield waves of similar forms . (iii) represents a plane wave propagated vertically
downwards as in a homogeneous medium of refractive index nl (according to t h e factor e iknll) ,
originating from the elementary layer of thickness oZ situated at z> 5". All such layers above
5" yield waves of similar forms.
The work of Westcott [1962a] shows th at this is t he correct interpretation of the elements
involved. Each individual layer of thickness oz in an ionized medium gives rise to reradiated
waves, their originating strength depending on the exact valu e of the electric field within the
lay er oz. In Westcott's series of papers he considered these reradiated waves as propagated
in free space, thereby excluding the possibility of a homogeneous medium existing above z= b
of refractive index other than unity. His formulae would not converge if such a medium
existed there. We have modified the theory to allow for this, at the same time providing
the reflection and transmission formulae derived in section 3. It should be observed that
our method has not proved the interpretation placed upon the elements involved; to accomplish
this would r equire the more elaborate analysis of Westcott, considering from first principles
the Hertz vectors involved in the reradiation processes.
6. Approximation Methods
In order to calculate the reflection coefficient (6), it is necessary to know E( z) throughout
the medium beforehand. If E( z) were known, R would also be known, implying that there
would be no need to evaluate R from form ula (6), which after all is but an identity. But an
iterative method of solution may be adopted, using approximate solutions derived from eq (4).
A series of approximations may form ally be written down based on successive substitution
into eq (4). We write
68
L_
-I
Now I
then
that is, provided the layer is sufficiently thin reb-a) small enough] 01' provided M is small
enough (the ionization density small throughout).
If frce space exists above z= b, the formulae simplify. The successive approximations are
·1 r r
Ein (t ) -_12 UC X E () ik(z- rJl +1 ·1 f ~ .X E n 1 () - ikl'- rld
J a l - iZ " n _ 1 Z e GZ 2~/C r l - iZ f _ Z e Z
(11)
from eq (7). Under these circumstances, eq (9) yields for the first term in T,
This value of B arises from the contributions to the reflected wave produced by the incident
fi eld e- ikz being regarded as the sole cause of the reradiation process at each level. It corre-
sponds, in fact, to the use of the Born approximation.
For this case, we have from eq (5 ),
658514- 63- 6 69
yielding the second term in the series for R:
If \ve place
v=iE' /n1kE,
then differentiation and the use of eq (1) show immediately that
dv
-- _ - -ik (2 2 2)
n - nv
dz nl I,
where v= 1 when z> b and
e-ikZ_ R eikZ
v nl(e ikz+ ReikZ)
when z<a. This is the equation derived by Wait [1962] and which was solved by successive
approximations. The corresponding integral equation would be
(13)
no doub t simpler than our eq (5), but certainly less comprehensive in physical content. M ore-
over, our eq (6) yields R directly as a series expansion, but Wait has shown how eq (13) may
be solved approximately, yielding R as a series expansion only indirectly through the expansion
of v.
7. Examples
As the first example, we may consider the model for which free space exists above and
below the slab with bounding surfaces z= O and z= h; homogeneous medium exists in thp, slab
such that n 2 = 1-X. It ma~- easily be shown that the reflection coefficient is given by
. (n 2 - 1) (1 - e2iknh)
R = (n+ 1) 2e2iknh_ (n - ] )2'
In order to apply the second-order reflection formula (12), X is assumed to be small within
the slab. EA,])anding R to the second order in X, we easily obtain
R = iX(l- e -2ikh) + iX2(1- e -zikh-2ikh e -Zikh). (14)
We may also use eq (12). Since X is constant throughout t h e range of integration, we
obtain
These integrals are trivial to evaluate, the result without any approximation being equal to (14).
Secondly, we may consider a symmetrical Epstein [1930] profile, tending to free-space
conditions above and below the layer. Let
ae~ Z
X = (e 8z + 1)2'
70
The work of Epstein (or of Budden [1961]) shows that the reflection coefficient has the valu e
(15)
{32 sinh (27rk/{3)
This result should also be recovered by the direct applicaLion of formu la (11 ). This yields
This may be evaluated by considerin g the correspondin g complex contour integral Laken
around the rectangle with vertices ( ± L, 0), ( ± L , 2i7r/(3) , the contour enclosing one double
pole at z= i7r/{3. The result tur ns out to be identical with (15).
8. Integral Identities
The refl ection formul ae (7 ) and (10) ma~T be used to obtain certain interesting inflllite
integrals, when the valu es of R , X , and E are known from other considerations.
Our first illu stration involves the exact solution of a tractable model. Consider the
exponential profile X =e az/k2 with no collision frequency. The work of Budden [1961] shows
that
r (- 2i lc/a) -4ikla
R r (2ilc/a) a ,
where v is real and positive and arg t = O along t he positive real axis. The integra.l reduces to
the result containing the single parameter v:
where Ai denotes the standard Airy integral and C is a constant. This result is a generalization
of the simpler W.K.B.J. solutions
n- I / 2 exp ( ±ikiZndz}
solutions that cease to be valid near the transition point z= O, and that require connection
formulae to effect a suitable connection across z= O. Equation (16) , however, suffers from
none of th ese disadvantages ; this solu tion if' valid even at the transition point.
To be specific and consistent, let
arg n=O for z< O,
arg n =-}f7r for z>O,
arg z= O for z> O,
arg Z= 7r for z<' ().
, Vhen z> O, we use the standard asymptotic expression
J
Since n 2 is negative, we must write n=e- i1r /2m, where m > 0; hence
1 . (3 (r Z )
"'ZC7r - 1/2 e'1r/6 Zk ) -1/6 m - 1/2 exp -kjomclz,
an exponentially evanescent solution as z-,;>co. This solution therefore satisfies the boundary
condition there.
When z<O, let z=e i1r t, where t > O, yielding Ii'
f 'ndt
=Cn-1/2ei1r/6 ( Jo rrndt)2/ 3J.
)1/6Ai [ - (3zkJo
We now quote the standard asymptotic e)..rpression
yielding
72
In order to introduce the in cident and r eflec ted waves, we note that if rl is large and
positive, then
if o
n(- r)dr= i f!
0
n(- r)dr+ i-fI
z
dr,
rfn(-
Jo r~
g-) dg-=-Z- rl + Jo f [n(- g-) - l]dg-,
n(- r)dg-=- z+ Jo
oo
where the upp er limit is replaced by DO since the integrand mu st vanish at t hese additional low
heigh ts. It follows t h at as n -71,
l.
] . (3
('- I = 2" 7r- 1 /2e '~/6 2 k ) - 1/6csp (. ~kJ o
1 .)
r oo [n ( - r)- 1]d r-4 7r~ '
It follow s t hat
R = i esp ( - 2ik.Fo oo [n (- r) - I]d .\}
X = 1- n 2 ,
l / n - 1/2
rznd z)1/6At.[(3
(Jo . rzo ndz )J2/3
2 ei7r/2lcJ
r E 12" . (3
7r- 1 / 2e' ~/ 6
2 'C
) 1/6exp ( . 1 .)'
r oo [n ( - n - l]cl .\-4~
~IcJ o
Final sub stit ution in Lo cq (10) yields after cross multipli caLion
1 (3
27r- 1/2e5irr/ 12 2lc ) - 1/6 exp (-.~lcJro oo [n( - g-)-- l ]dr
.)
=2 1. f oo O - n
~lc -00
.
2)e-'kZn- 1/2 rzndz )116 At.
( Jo [(3
2 e'.7r Ic Jrz
/2
o nd z
)2/3J dz.
This formula embraces t h e buildup of t he r eflection coefficient from every la:ver of a slowly
varying medium of the type postulated.
73
Here, X, Y, Z are the usual dimensionless parameters describing the ionosphere, U = I-iZ,
and - (l, m, n) are the direction cosines of the earth 's magnetic field . The vector ME is pro-
portional to the electric polarization vector.
I n evalu ating curl E, we choose t h e customary coordinate system in which o jox= -ik
sin 0, o joy= O, and ojo z is replaced by a prime. In component form, eq (1 7) may be written as
( bEll e-ikz cos 0clz= (E' +ik cos OE )e - ikZ cos 0lb- k 2 cos 2 8 ( bE e-ikZ cos 0d z,
Ja a Ja
we obtain from eq (18 )
(E~+ik cos 8Ey)e- ikZ cos 01"= -1c2 ( "(ME )v e - ikZ cos odz, (19)
a .J a
Multiplying the third of these equation s by tan 8 and adding it to the first , we obtain the
simpler equation
(E ;+ik SQC 8Ex + ilc Sill 8Ez)e- ikZ cos 01"= -F ( b[(ME)x+ Lan 8(ME )z]e - ikZ cos °dz. (20)
a Ja
When the incident field is horizontally polarized, let
.Ex= cos 8.LR neikZ cos 0,
E y=e-ikZ cos 0+ .L R.LCikz co s 0,
the x- and t-factors b eing suppressed. 'iVhen the incident fi eld is vertically polarized, let
E y= I R J.eikz co s 0,
1
- 2ile cos O(A .l.R .l. + 13nR .l.)=-lc 2 a(ME )ye - ikZ cos odz,
- 2ilc(A 1-R II+ B I R I ) =- 1c2. f[(ME)x+ tan 8(ME )z]e- ikZCOB adz.
A .l.R 1- + 13II R .l. = -~ i lc sec 8 I" (1V12IB cos 8+1\1£22A -11123B sin 8) e- 2ikz cos od z,
A.l.RII+ BII RII = -~ ilc I" (Mll B cos 0+ 1I1I zA -.M I3B sin 0+1\1£3113 sin 8
+ 1V13zA tan 8- M 33B tan 0 sin O)e-Z fkICOS adz.
H ence, if Z is constan t, we obtain
.L
Rn= -.!.2 i le Jfaa 01,([I ?+
.
M 32 tan 8)e - 2ikZ cOSOdz
_.!. . j (- inYU - lmY2)+ Lan 0(i lrU- mny2) ( " Y -2ik' COBO l
- 2 ~/C U (U2 _ y 2) .L .L1. e G z,
1I R II= -~ ileJ:b (l\111 cos 0- 11113 sin 0+1\131 sin 8- 1V133 tan 8 sin 8)e - 2ikZ cosod z
_.!. .j U2 cos 20 sec 8- (l2 cos 0- 112 tan 8 sin 8) Y 2- 2im sin oYU ( " Y - 2ikz COB 0d
- 2 ~/C U(U2 _ Y 2) Ja.L1. e z.
75
Moreover, we easily deduce the formulae for the isotropic case with Y = Q at oblique incidence,
The transmission coefficients may be found by multiplying eq (18) by e ikz co s °and inte-
grating over the height range in which ionization occurs. We then multiply the third of the
resulting equations by tan e and subtract it from the first, yielding
(E~ - ik cos eEy) eikz cos 0lb= _p { b (ME )ye -ikz cos 0dz ,
a Ja
and
I, {b
(E~-ik sec eEx+ ik sin eEz)e ikZ co s aIa = -P J a [(ME )x+ tan e(ME)zle ikZ cos °dz. {
Inserting the values (21) and (22) of the field above and below the range of ionization ,
we obtain
1 {b
A 1-T 1- +B uT 1- = A -2 ik sec eJ a (M E)ye i k Z cos ° dz,
Finally, we deduce the values of the transmission coefficients to the first order, when the
field throughout the medium is replaced by the incident field:
I TD= l-~ ik { b (M1l cos e-1\113 sin e+ M 31sin (J - )YI33 tan (J sin e)dz .
.L
In the isotropic case, these formulae reduce to
(23)
for zero collision frequency. On the other hand, the first approximation to the solution of
the integral eq (13) is obtained by taking V= 1, yielding
76
Hence, when t=a,
or
(24)
when the left hand sid e is expanded by the binomial since IRI is small.
It is obvious that this result is only valid for a thin layer, in which case the phase factor
e- 2 i k z in the integrand of (23) is replaced by e- 2 i ka • H ence, for the homogeneous slab di scussed
in section 7, eq (24) yields (when a= O, b= h) ,
11 . References
Brekhovskikh, L. lVI., Waves in layered media (Academic Press, London and I ew York, 1960).
Budden, K G., Radio waves in the ionosphere (Cambridge Univers ity Press, England, 1961).
Epstein , P . S., Refl ection of waves in an in homogeneous absorbing medium, Proc. Nat. Acad. Sci . (U.S.A.)
16, 627 (1930) .
H eadi ng, J. , An introduction to phase-integr al methods (M ethu en, London, 1962).
Wait, J . ll., On the propagation of VLF and ELF radio Wfl,ves when the ionosphere is not sharply bounded,
J . Resea rch NBS 66D (Radio Prop .) No.1, 53- 61 (J an.- Feb . 1962) .
Westcott, B. S., Ionospheric r efl ection processes for long radio-wavcs-I, J . Atmospheric and T errest. Phys.
24, 385- 399 (1962a) .
West cott, B. S., Ionospheric r efl ection processes for long radio-waves-II, J. Atmospheric and T erres t . Phys.
24, 619- 631 (1962b).
'Westcott, B. S., Ion ospher ic r efl ection processes for long radio-wav es-III, J. Atmospheric and T errest. Phys.
24,701- 71 3 (1962e) .
658514- 63--7 77