December, 1964
T O T A L E N E R G Y A N D E N E R G Y S P E C T R A L D E N S I T Y OF E L A S T I C
WAVE R A D I A T I O N F R O M P R O P A G A T I N G FAULTS
BY N . A. HASKELL
ABSTRACT
Starting with a Green's function representation of the solution of the elastic field equations
for the case of a prescribed displacement discontinuity on a fault surface, it is shown that a
shear fault (relative displacement parallel to the fault plane) is rigorously equivalent to a
distribution of double-couple point sources over the fault plane. In the case of a tensile fault
(relative displacement normal to the fault plane) the equivalent point source distribution is
composed of force dipoles normal to the fault plane with a superimposed purely compressional
component. Assuming that the fault break propagates in one direction along the long axis of
the fault plane and that the relative displacement at a given point has the form of a ramp time
function of finite duration, T, the total radiated P and S wave energies and the total energy
spectral densities are evaluated in closed form in terms of the fault plane dimensions, final
fault displacement, the time constant T, and the f~ult propagation velocity. Using fault pa-
rameters derived principally from the work of Ben-Menahem and ToksSz on the K a m c h a t k a
earthquake of November 4, 1952, the calculated total energy appears to be somewhat low and
the calculated energy spectrum appears to be deficient at short periods. It is suggested that
these discrepancies are due to over-simplification of the assumed model, and that they may be
corrected by (1) assuming a somewhat roughened ramp for the fault displacement time func-
tion to correspond to a stick-slip type of motion, and (2) assuming that the short period com-
ponents of the fault displacement wave are coherent only over distances considerably smaller
than the total fault length.
INTRODUCTION
Knopoff and Gilbert (1960) have used a Green's funct!on integral representation
of the solution of the elastic wave equations in an infinite medium to discuss the
radiation pattern of first motions from a moving fault. In the present paper we
shall employ the same approach to calculate the total radiation from a moving
fault of finite length.
In the compact tensor notation of de Hoop (1958) the elastic displacements,
ui(xl, x~, x3, t), in a volume, V, bounded by a surface, S, are given by
u ~ ( x ~ , x2 , x3 , t )
(1)
where
o = density
Gij is an operator which transforms a given function, @(~1 , ~2, ~3, t) into
G~j[~o(~, t)] = ( 74 t o ) - 1
f (3~,i'yj - 6 ~j)r - 3
fr/~
r[a
~(f., t -- t ' ) t ' dt
(2)
2 --1
q- ~'i%'(a r) ~(~, t -- r / a ) - - (~'i 3'~ -- ~i3) (~2r)-~(~, t - r/~)}
where
3"i = ( x i - ~i)/r.
then write
cjk,pq(Oup/OQ) = p( a 2 -- 2¢/2)6jkEpp + 2p132ejk= T ~k (3)
where eCk is the strain tensor and Tjk is the stress tensor, evaluated at the point
( ~ , ~2, ~.~). The unit normal vector nk is also constant with respect to the operator
G~ so that the first surface integral in equation ( 1 ) may be written as
where F~. = T~knkis the force per unit area that the material in the non-linear region
inside S exerts on the material in the elastic region external to S. The integral in (4)
can be written as
[,l'-
I'n's_i-i" ,I
°
FIG. 1. Cross section of source region.
where F~+ is the force per unit area exerted across the S + surface and F f f is similarly
referred to the S - surface. If the separation is small, it is difficult to imagine how
any process such as fracture, plastic flow, or phase transition occurring in the non-
linear region could produce forces, Fj + and F f f that are not equal in magnitude and
opposite in sign. In the initial state of static equilibrium they are necessarily equaI
and opposite, and even during the process of faulting it would be expected that
departures from this condition could exist only for times comparable to those re-
quired for P or S waves to travel the distance between S + and S-. As stated pre-
viously, we regard the energy associated with the corresponding frequencies as
negligible. Therefore we set F f f = - F j +, and the integrals of equation (5) cancel.
We are then left with only the last surface integral in equation (1), which may be
written in the form
Since the limits of integration do not depend on the variables xl, the differential
operator in equation (6) may be taken inside the integral sign and the indicated
differentiation carried out explicitly. The result is
u , ( x , t) = - 232)r~+.]/Iiq,q[nj]-~-pl~2(nq+Miq,q[Dj]
(7)
+ n~+Miv.q[Dq])} ds
(
Mo,q[~(~, t)] = (4~-p) l~[15~i~s~q -- 3(6~j3'q + &q'rj + ~q'r~)]r -4
rr/~
• J ~ / ¢ ( ~ , t -- t')t' dt' + [6"Yi'Yj'Yq -- (~ij "Yq "-~ ~iq~Yj "7- ~jq~Yi)l(olr)--2~(~, t -- r/a)
(s)
We shall assume that the fault displacement vector, D~(~, t), has the form of a
wave propagating in one dimension with velocity v. Three linearly independent cases
m a y be distinguished, (1) a longitudinal shear fault in which the displacement is
parallel to the direction of fault propagation, (2) a transverse shear fault in which
the displacement lies in the fault plane and normal to the direction of propagation,
(3) a tensile fault in which the displacement is normal to the fault plane. The
first case would be represented by pure strike-slip motion on a horizontally propagat-
ing fault, and the second by the horizontal propagation of pure dip-slip motion. The
ENERGY OF ELASTIC WAVE RADIATION 1815
third case, which does not appear to have been considered in recent fault-plane
studies, might be realized in nature by the opening of a fissure by the injection of
magma or magmatic gases. Also, since real fault surfaces are not perfectly plane, a
relative displacement parallel to the fault surface must result in a small component
of displacement normal to the fault surface. An analogous artificial source could be
the opening of radial tension cracks about an explosion cavity (Kisslinger, Meteker,
McEvilly 1961 ). In anticipation of results to be obtained later, it may be mentioned
that cases (1) and (2) are equivalent to moving point sources of the double couple
type, while case (3) is equivalent to a moving point source composed of a superposi-
tion of a pure dilatation and a force dipole normal to the fault plane.
Let the (xl, x2) plane be the fault plane, with the positive xa axis normal to the
fault plane on the S + side, and let the positive x~ axis be parallel to the common
direction of fault displacement and propagation. Then n~+ = -~ia and D~ = &ID.
In this case equation (9) becomes
2( lw f + (ar) _1 ",'~'hyaO(~, t -- r / a ) ds
..
(10)
The integrands in equation (10) are exactly the same as those that one would ob-
tain by considering the fault plane to be covered with a distribution of double couple
sources, the common moment of the component couples having an areal density
M(~, t) = pB~D(~, t).
In accordance with the definition of radiated energy previously given, we may
consider the point of observation, x to be sufficiently remote that the radial distance,
r, and the direction cosines, % are constant over the area of the fault to any desired
degree of approximation. We further assume that the fault plane is rectangular
with length L in the direction of propagation and width w in the transverse direction.
We also suppose that the dependence of D on the transverse coordinate does not
vary with time. Equation (10) may then be written as
where
L
(~2)
L
/) is the displacement averaged over the width of the fault and ~ = &. Introducing a
1816 BULLETIN OF THE SEISMOLOGICAL SOCIETY OF AMERICA
spherical coordinate system (r, 0, ~) with polar axis in the direction of fault propaga-
tion,
"Y1 = COS 0
TOTAL RADIATED E N E R G Y
The total radiated energy, E, may be written as the sum of the P-wave energy,
E . , and shear wave energy, Ea, where
2~r ~rf+~
Ea=Pafo fo ~_~ ~tr2dtr2sinOdOd~
(15)
= (pw2/167r~)(f~/a)5 f0 sin2 20 sin 0 _
f7 Ia 2 dt dO
2*r 7rf +oo
E~ = p~ fo fo -~ (it°2 + its2) dtr2 sin O dO cl~
(16)
= (pw2/16~-~) fo (c°s2 20 + cos 2 O) sin 0 I~ 2 dt dO
If we write "h(o~) for the Fourier transform of a component of the particle velocity
~(t)
¢~(w) = ~ e-i'~tie dt (17)
the spectral energy density of the total P and S wave energies may be correspond-
ingly expressed as
e~ = 2pc~ [ 6~ sin 0 dO d~
fo2~fo ~ ]2r2
(18)
= (Pw"/8~a)(~/°~)~fo sin" 2o sin oIL 12 do
ENERGY OF ELASTIC WAVE RADIATION 1817
21r ~v
where the function G(t) is assumed to satisfy the initial and final conditions
G=O, t<0
(22)
G-)l, t-+
and F ( t ) expresses the variation (if any) of the final fault displacement along the
length of the fault. In the present paper we shall consider only the case F ( t ) =
const. = Do. The integral 14 then becomes
L
where
"~ = t -- ro/a
q, = a -- cos 0
1818 BULLETIN OF THE SEISMOLOGICAL SOCIETY OF AMERICA
a -- a/v
Since
and
£" 02(r -- q . C / a ) dr =
F" ¢o
G2(x) dx
we have
A similar expression also holds, of course, for the integral of I~2 with ¢~replacing a.
Equation (27) expresses the radiated energy in terms of the autocorrelation of
O(t).
It is now necessary to choose an explicit form for the displacement time function
G(t). For the sake of simplicity in this initial treatment of the problem we shall
take it to be a ramp function of finite duration, T, that is
,<0
G(t) = T 0 <- t <- T (28)
t>T
From their determination of the initial phase versus frequency of the long-period
surface waves from the K a m c h a t k a earthquake of November 4, 1952, Ben-Menahem
and Toks6z (1963) deduced a G(t) for this earthquake of the form
with the exponential time constant, T, having a value of about 22 sec. The Fourier
transform of the ramp function (28) is
~(~) = (1/~2T)(e - i ~ T - 1)
= - - (~r/2) - - tan-l~T
I I I I I I i I i
-0.1
e = G2('E-W)
x x = Rz { E - W )
- 0.2 x x
x
G* x X xe
OxoXx xXX X ® %
i~- O-3 _ ox 2x :
.J
(J
Q:
"e-
O,4
x GO
0
0 ~ . A M P T= 3 0 s e c
0.5
-0.6
I I I I I I I I
2 4 6 B I0 12 14 16 18
FREOUENCY {MILLICYCLES)
the ramp source function with T = 30 sec. and the exponential function with
T = 22 sec. Aki (1962) has estimated that present knowledge of phase velocities
limits the accuracy of determinations of the phase of the source time function to
about ± ~ circle. With this lack of accuracy it seems clear that the available long-
period phase data is inadequate to determine the time function with any degree
of precision and there is as much justification for choosing the finite ramp function
as any other function that satisfies the general requirement of being step-like if
viewed at sufficiently low frequencies. The discontinuities in G(t) at t = 0 and T
are, of course, physically unrealistic. However, if we imagine the corners of the
ramp function to be rounded off in a time interval <<T, only the very high frequency
components of the energy spectrum will be affected. As we shall see presently,
propagation of the fault has the effect of applying an attenuation factor propor-
1820 B U L L E T I N OF T H E S E I S M O L O G I C A L S O C I E T Y OF A M E R I C A
tional to 1/o~2 at high frequencies, so that the effect of the sharp corners is suppressed
to a large extent even without the rounding-off assumption•
Adopting the ramp function form for G(t) and substituting in equation (27), we
have
f~= 02(r) dr = l I T
and
f~¢ 1. 2 d t = \2(aDo)2/Vq2;
f2°LLD°2/T2[qa]; ]qaL//°~[ < TTI
[q. L / a [ >
(30)
Substituting this and the equivalent integral with f~ replacing a in equations (15)
and (16),
where
~01
01 (a, a T / L ) = J0 sin2ta
--1 2cos
sin
0 OO dO q- (aT~L) fe, ~---eosO)-)
sni2 20 sin O2do (33)
In these expressions
01 = {°1 s- (a -- o~T/L);
a T / L < a -- 1
a - - 1 < ceT/L < a--k 1
ceT/L > a q- 1
(35)
(0 ; flT/L < b - 1
02 = cos l ( b - b - 1 < < b + 1 (36)
; ~T/L > b + 1
Similar expressions have also been derived for the exponential time function, but
will not be considered here, since the results do not differ significantly from those
to be obtained with the ramp time function.
ENERGY OF ELASTIC WAVE RADIATION 1821
O~(b,~T/L) = fb -1
c2 y-l[4(b -- y ) 4 _ 3(b -- y ) 2 + 1] dy
b-i-1
(38)
+ (fiT/L) f y-214(b _ y)4 _ 3(5 -- y)2 ~_ 1] dy
~v 2
where
Further evaluation of these and similar integrals that occur in connection with the
other types of fault displacement will be found in the Appendix. The total radiated
energy, E = E~ + E~ is
E = (pw2Do~L/8~rT~)O(a, b, L, T ) (39)
where
I,:nopoff (1958) has given an expression for the total energy released by the forma-
tion of a long longitudinal shear fault in a homogeneous shear strain field by calcu-
lating the difference in the static elastic strain energy before and after the formation
182~ BULLETIN OF THE SEISMOLOGICAL SOCIETY OF AMERICA
J = ~rp~2D,~2L/8
where D~ is the maxinmm displacement at the center line of the fault. Knopoff's
static theory leads to an expression for the variation of the displacelnent across the
width, w, of the fault
where y is the lateral distance from the center line. The Do in equation (39) was de-
fined as the average displacement across the width and we therefore have
Do = 7rDm/4. In terms of Do the available strain energy is then
J = 2pfl2Do2L/~ - (41)
Since the total energy radiated dynamically cannot be greater than the available
strain energy, comparison of equations (39) and (41) gives a minimum value of the
time constant T.
2
Tmia = w 2 0 / 1 6 ~ 2 (42)
,~ aL
where~(o~) is the Fourier transform of ~(t). Carrying out the integration over
we have
B2(~) =
(1)( (cos ~ 20 -4- cos 2 0) sin 0X~-2 sin 2 X~ dO (48}
- - .,~.,-
-!
I0
tO-2
-ENVELOPE
io-3
B:(~o) = ( 1 / 2 p ] ) fb+t [4(b - y)4 _ 3(b - y)2 -4- 1]y-2(1 - cos p~y) dy (51)
,]b--1
ENERGY OF ELASTIC WAVE RADIATION 1825
where p , = ~L/a and p~ = wL/~. The evaluation of these and related integrals is
given in the Appendix.
The factors B1 and B2 are plotted as functions of p~ and p a respectively in figures 3
a n d 4. With / ~ I ~ given by equation (49), and choosing flT/2L = 0.1, aT/2L =
~\,
i 0 -I _
\\\
\
IO "2 _
B2Ko) -
i 0 ~;
NVELOPE
jo-~
be much too low a frequency for the maximum energy density, even for a very large
earthquake. I n fact, if the spectrum were actually of the form shown in figure 6, it
would be difficult to understand how DeNoyer's integration of short-period Wood-
Anderson records could have yielded more than a minute fraction of the total energy.
One possible way out of this difficulty immediately suggesss itself. The actual fault
displacement-time function must be much more irregular than the straight ramp
function that we have assumed. Presumably in the real case a stick-slip, or chatter-
,o-
I I I I I III I I I I I IIII I I IIII~
i 0- 2
io- 3 _-
PB
FI~ ' 5. Total energy spectral densities for P and S waves; ramp time function,
~ T / L = 0.2.
ing type of motion occurs, and a more realistic representation would be a somewhat
roughened ramp function. As an approximation to such a function we shall consider
a ramp modulated by a sine wave
[0 t <0
G(t) = ~T-l[t - (T/2n~r)sin (2nTrt/T)] 0 < t < T (52)
[1 t> T
The amplitude of the modulation has been so chosen that G never becomes negative,
which seems a reasonable restriction to place on a stick-slip type of mechanism.
With this type of time function the integrals f I=2 dt and f 1~2 dt as given by equation
20
I I I I i i
18
16
14
12
%
-- 10
0 I0 20 30 40 50 60 ?0
FIG. 6. T o t a l e n e r g y s p e c t r a l d e n s i t y , r a m p t i m e f u n c t i o n , f ~ T / L = 0.2.
(27) can be evaluated in closed form, but the expressions are cumbersome and it
does not appear to be possible to carry out the next step of integration over the
angular variable 0 in any simple analytic form. We shall therefore onfit the calcu-
lation of the total energy by integration in the time domain and go directly to the
spectral energy density as given by equations (45) and (46). The Fourier transform
of G(t) is now
I0
I I I I I I i I I
8 --
4 --
30 40 50 60 70 80 90 I O0 I I0 120 130
FIG. 7. Total energy spectral density, modulated ramp time function, f~T/L = 0.2, n = 3.
n = 3 is i l l u s t r a t e d in F i g u r e 7, in w h i c h all o t h e r p a r a m e t e r s a r e t h e s a m e as t h o s e
a s s u m e d for t h e r a m p f u n c t i o n used in c o m p u t i n g figure 6. A t ~=, ] G 12 has t h e
value
2 = 2 = 2 (55)
F r o m e q u a t i o n ( A 2 8 ) , t h e high f r e q u e n c y a s y m p t o t e s of t h e f u n c t i o n s B I ( ~ ) a n d
B2(~) are
BI(~) ~ 2K,(a)/pa 2
* The fact that the two G(t) functions that we have used here have exactly the same phase
vs frequency relationship is in apparent contradiction with the use by Ben-Menahem and
ToksSz (1963) of the Kramers-Kronig dispersion relations to obtain the time function from the
phase function. The Kramers-Kronig theorem is applicable only if it is assumed that G(t) is an
analytic function for all t > 0. The functions that we have considered are not analytic at t = T
and the theorem is therefore not applicable in this case. Since we cannot give a complete causal
description of what happens in the non-linear source region, there seems to be no reason for
restricting G(t) to be analytic.
E N E R G Y OF E L A S T I C W A V E R A D I A T I O N 1829
B:(¢o) ~ 2K~(b)/p~ 2
Thus
and
1963a, 1963b)) indicate that this representation is applicable to the very low fre-
quency components of the smoothed mean motion, it seems very unlikely that the
superimposed irregularities can be treated as coherent over very large distances. It
is more probable that they are due to localized blockages, or stick-slip mechanisms,
that result in motions that can be correlated only over limited segments of the fault.
We are thus led to consider the case in which the fault displacement is represented
as a coherent wave only over segments of the fault of length l~, and the radiation
from adjacent segments is assumed to be statistically independent, or incoherent,
so that the energies are additive.
Assuming the same ramp time function for each individual segment, the total
energy radiated by the ith segment is given by equation (39)
E = 1.8245pw2Do:L/8~rT 2 (59)
On the assumption that the motion is coherent over the full length, L, the cor-
responding expression is
E = 3.0136([3T/L) (pw2Do2L/8~rT 2)
(60)
= .60272 ( pw2Do2L/8~T ~)
for the case f l T / L = 0.2, which was used in computing the spectra shown in Figures
5 and 6. The assumption of partial incoherence has thus increased the total radi-
ated energy by a factor of about 3.
The contribution of a coherent segment of length li to the energy spectrum is,
by equations (45) and (46),
The energy spectrum computed from equation (62) for the case IlL --- 0.2 is shown
in figure 8, all other parameters being the same as those used in computing figure 6.
It is seen that the assumption of a reduced coherence length has shifted the main
low frequency energy peak to a somewhat higher frequency and has considerably
increased the energy content of the next higher lobe in the spectrum.
ENERGY OF ELASTIC WAVE RADIATION 1831
I I I I I I I I I
50
90
O
m
¢.O
I0
0 I0 20 30 40 50 60 70 80 90 I00
%
FIG. 8. Total energy spectral density, reduced coherence length, i/L = 0.2.
the case assumed for illustrative purposes the height of the maximum shown in
figure 7 would be increased by a factor of 5. I t then appears that, by a combination
of superimposed roughness in the fault motion and a reduced coherence length, it
m a y be possible to construct theoretical models that will radiate realistic amounts of
short period energy as well as the very long periods associated with the smoothed
mean motion. Since no quantitative total energy spectra of large earthquakes are
available, we cannot, at present, be more specific about what ranges of the param-
eters L / l and n would be realistic.
The "coherence length", l, introduced here in a somewhat ad hoc manner, bears
some resemblance to the "mixing length" that has been used in the treatment of the
turbulent flow of fluids. As in the hydrodynamic case, a more fundamental approach
would be to recognize from the start that the details of the fault displacement func-
1832 BULLETIN OF THE SEISMOLOGICAL SOCIETY OF AMERICA
tion, D(}, t), may be too complex to be given a deterministic description, and there-
fore to treat D (~, t) as a semi-random function whose spacial and temporal correla-
tion properties are subject to constraints such that at very low frequencies it has the
characteristics of a step-like wave propagating at a definite velocity, while at higher
frequencies it becomes increasingly incoherent. An attempt to find an appropriate
statistical description is, however, beyond the scope of the present paper.
this would correspond to an earthquake of Richter magnitude about 2.1. The maxi-
nmm in the energy spectrum shown in figure 6 would fall at a frequency of about
3.7 cps. Some total seismogram energy spectra for very small earthquakes presented
by Frantti (1963) suggest that the frequency of maximum energy density for earth-
quakes of this magnitude might tend to be somewhat higher than this (by a factor
of about 2 or 3), but this difference is well within the uncertainty of the assumed
dimensions and, in any case, could easily be explained by the likely assumption that
very small earthquake faults are more nearly equidimensional than large ones, so
that a typical L for a magnitude 2.1 earthquake would be smaller and the w larger
than the values we obtained by assuming geometrical similarity.
where
sin s 0 dO q- ( a T ~ L ) sin 5 0
Os(a, s T / L ) = Jo l a - cos o I fo lr~ (a 2 c ~ s 0) 5 dO (68)
where
Explicit expressions for these integrals are also given in the Appendix.
TENSILE FAULT
where
~0~
~)6(b, ~T/L) = Jo (3 c°selb--0+cosl)o/Sin3 0 dO
(79)
+ (~T/L) fo '~ (3 cos e 0 -t- 1) sin 3
~-- cos 0) 2 OdO
ENERGY OF ELASTIC WAVE RADIATION 1835
where
[2(1 -- 2c) e + 4c(1 --2 c) sin s 0 + 3ce sin t0]
• [1 -- cos p~(a -- cos 0)] sin 0 dO (82)
Bs(~) = (1/2p~ 2) i ~ (a -- COS 0) e
Tensile fault
where the numerical values given correspond to the limiting cases given in the
Appendix, with the assumption that (a/fi) 2 = 3.
Nearly all estimates of P to S-wave energy ratios at large distances that have been
published (e.g. DeNoyer (1959)) have lead to considerably larger values of the
ratio t h a n those given here for either type of shear fault. Part of the difference,
perhaps most of it, may be explained in terms of S to P energy conversion at crustal
boundaries and the free surface, and a higher absorption coefficient for S waves than
for P waves. However, since a major shear fault is unlikely to be a perfect plane, it
is probable that shear faulting is accompanied by at least a small component of
displacement normal to the fault plane due to imperfect fit between the opposite
faces of the fracture. As indicated above, this normal component of displacement
produces an initial P to S energy ratio about 10 times as large as the ratio for a
1836 BULLETIN OF THE SEISMOLOGICAL SOCIETY OF AMERICA
A P P E N D I X : E V A L U A T I O N OF ~) AND B INTEGRALS
k
-1
O(b, tc) -- -1 y ( a o + a~y--p a2y 2 + asy 3-~ a4y 4) dy
b÷1 (A1)
+ k f y-2(a0 + aly + a2y 2 + asy 3 .-P a4y 4) dy
,,'b
O(b, k) -- a0 In ( k / ( b -- 1)) ~ - a l k l n ( ( b + 1 ) / k ) ~ - k [ ( a 4 / 3 ) b 3
F o r t h e case ]c < b - 1 only the second integral occurs, with lower limit -- b - 1
a n d 0 becomes
a~ = 16a, a4 = --4.
ENERGY OF ELASTIC WAVE RADIATION 1837
T h e l i m i t i n g cases t h e n r e d u c e to
F o r t h e f u n c t i o n O2 t h e c o r r e s p o n d i n g coefficients a r e
ao = 4 b 4 - 3b 2 + 1, a~ = -2b(8b 2 - - 3), a2 = 3 ( 8 b 2 - - 1)
aa = -16b, a4 = 4
a n d t h e l i m i t i n g cases are
where the numerical values given have been calculated assuming Mansinha's
m i n i m u m v a l u e s , a = 2.234 a n d b = 1.290.
I n t h e ease of t r a n s v e r s e s h e a r f a u l t i n g t h e a~.'s for t h e f u n c t i o n O~ a r e :
a n d t h e l i m i t i n g cases are
F o r Ok we h a v e
a n d t h e l i m i t i n g cases a r e
L i m i t i n g cases a r e
05(a, k)k>a+l
05(a, ]~)k.<a-1
= k[4ac(2 - - c - - 3a2c) In [(a -[- 1 ) / ( a - - 1)] - - 16c(1 - - c) (A14)
F o r a m e d i u m w i t h a P o i s s o n ' s r a t i o of 1/4, c = 1 / 3 a n d t h e s e r e d u c e to
= .65508
-- .324961c
E N E R G Y OF :ELASTIC W A V E R A D I A T I O N 1839
a3 = 12b, a4 = --3
a n d the limiting cases are
Ao = (1/2)[ao -~- bai + ( 1 / 3 ) ( 3 b 2 -~- 1)a2 -/- b(b 2 -Jr- 1)a3 + ( 1 / 5 ) ( 5 b 4 + 10b s + 1)a4]
F o r large values of p the following a s y m p t o t i c expression for the C~ and S,. functions
are convenient.
S i ( x ) --~ (~r/2) - x - l I P ( x ) cos x ~- x - I Q ( x ) sin x] (A24)