Documente Academic
Documente Profesional
Documente Cultură
, 2 TABLES
R. T. Shuey*
ABSTRACT
l
R(6) = R o + AoR o + - -6a
-2
(I - a)
1"
Sill 2 6
neering work of Koefoed (1955), His method was laborious
computation of reflection coefficient versus angle out to 30
degrees for 17 different sets of elastic properties, Koefoed took
16V the three elastic parameters for each medium to be longitudinal
+- -p (tan ' 8 - sin? 8),
2 Vp velocity Vp , density p, and Poisson's ratio a, He gave his con-
clusions as follows.
The first term gives the amplitude at normal incidence
(6 = 0), the second term characterizes R(6) at intermedi-
(a) 'When the underlying medium has the greater longi-
ate angles, and the third term describes the approach to
tudinal velocity and other relevant properties of the
critical angle, The coefficient of the second term is that
two strata arc equal to each other, an increase of
combination of elastic properties which can be deter-
Poisson's ratio for the underlying medium causes an
mined by analyzing the offset dependence of event am-
increase of the reflection coefficient at the larger
plitude in conventional multichannel reflection data, If
angles of incidence,
the event amplitude is normalized to its value for
(b] When, in the above case, Poisson's ratio for the inci-
normal incidence, then the quantity determined is
dent medium is increased, the reflection coefficient at
I 6a the larger angles of incidence is thereby decreased.
A = A o + (I _ a)2 R ' (c) When. in the above case, Poisson's ratios for both
o
media are increased and kept equal to each other, the
A o specifies the normal, gradual decrease of amplitude reflection coelTicient at the larger angles of incidence
with offset; its value is constrained well enough that the is thereby increased.
main information conveyed is 6a/R o. where 6a is the (d) The effect mentioned in (a) becomes more pro-
contrast in Poisson's ratio at the reflecting interface and nounced as the velocity contrast hecomes smaller.
R o is the amplitude at normal incidence, This simplified (e) Interchange of the incident and the underlying
formula for R(6) accounts for all of the relations be- medium affects the shape of the curves only slightly,
tween R(6) and elastic properties first described by Koe- at least up to values of the angle of incidence of about
foed in 1955, 30'
Manuscript received by the Editor May 29, 1984; revised manuscript received October 15. 1984.
*P. O. Box 37048, Houston, TX 77236
D 1985 Society of Exploration Geophysicists. All rights reserved.
609
610 Shuey
DERIVATION
A - A __1_ ~O'
- 0 + (1 -
0')1 R o ' (11 )
Provided the percentage changes in elastic properties are
small, the P-wave reflection amplitude R(8) is given approxi- 1 - 20'
A o=B-2(1 + B ) - - , (12)
mately by Aki and Richards (1980, p. 153)as 1-0'
V; . 1 ) ~P
1 (1 - 4 -sm sec! 8 ~v.p 4V' . 1 ~v.s
-s and
R(8) :::::;- 8 -+- -- --s m 8_
2 Vpl P 2 Vp Vpl Vs .
Downloaded 03/26/17 to 212.26.2.132. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
B= ~~/~ (13)
(1 )
~ Vp/Vp + ~p/p'
The elastic properties in equation (1) are related as follows to Equation (9) displays which combinations of elastic proper-
those on each side of the interface: ties are effective in successive ranges of angle O. The third term
~ Vp = (Vpl - Vptl,
vanishes as 04 , so it does not normally contribute for 0 < 30
(2) degrees. However, at large angles it dominates. For consider-
Vp = (Vpl + ~tl/2, ation of absolute instead of relative amplitude, equation (9)
should be multiplied through by R o' i.e.,
M~ = (v"l - v"tl,
v" = (v"l + v" 1)/2,
(3)
R(O) :::::; R o + [A o R o + ~O'
(I - 0')
lJ sin? 0
~P = (Pl - PI)'
1 ~V
and (4) + -2 -Vp (tan? 0 - sin? 0)
.
(14)
p
P = (Pl + PI)/2,
DISCUSSION
where the incident and reflected waves are on side 1 and the
transmitted wave is on side 2. The angle 8 is the average of
Equation (14) has some similarity to the approximation of
incidence and transmission angles,
R(O) given in Bortfeld (1961) and discussed recently in numer-
o= (81 + 8tl/2. (5) ous unpublished lectures by Hilterrnan (1983). The Bortfeld
approximation has about the same accuracy as equation (1) or
These two angles are related by Snell's law, equation (14), but the three expressions differ in the philosophy
sin 8 1 sin 81 of grouping terms. Equation (14) displays which combinations
p=--=-- (6) of elastic properties are effective in successive ranges of angle O.
VI Vl '
Equation (I) is arranged to separate the effects of the three
I modified equation (1) by eliminating the properties v" ~v, variables ~P, ~ Vp , and ~ v,. The Bortfeld arrangement is de-
in favor of 0', ~O'. The latter are defined as above by signed to contrast elastic and acoustic reflection coefficients. It
separates two terms, the first of which (fluid factor) involves
only velocity and density and is the same as R(O) for a fluid-
and (7) fluid contact. The linearization of this fluid factor is
Vl = V l 1 - 20' (8) which corresponds to the first term and part of the third term in
S p 2(1 - 0')
equation (14).
and also by the differential of this equation. This change of Equations (9) and (14) diagonalize the relationship between
variable was motivated by the general perception [embodied in elastic properties and R(O) in the sense that certain features are
Koefoed's rules (a), (b), and (c)] that Poisson's ratio is the elastic related to certain combinations of elastic properties without
property most directly related to angular dependence of reflec- significant coupling between the variables. As is apparent from
tion coefficient. Figure I, the dimensionless parameter A controls whether the
A further modification was to factor out R o, the amplitude at amplitude initially increases (A > 0) or decreases (A < 0), while
normal incidence. Because the practical problems in recovering the dimensionless parameter B controls the sign at large angles.
absolute reflection amplitude seem more severe than the prob- From equations (9) through (14) I perceive three quasi-
lems of recovering the relative variation of reflection amplitude independent connections between R(O) and elastic properties:
with offset, it is appropriate to consider the information content
of the relative curve R(O)/Ro. The result of these manipulations (1) Normal incidence.-The magnitude R o is the average of
is fractional changes in Vp and P [cf., equation (10)]. Alternatively,
R o is half the change in natural logarithm of impedance PVp
R(O)/R o :::::; 1 + A sin? 0 + B (tan? 0 - sin? 0), (9) since the approximation
where 1 ~pVp 1
Ro ~ - - = - ~ In (pV) (16)
2 pVp 2 p
R :::::;~ (~~ + ~p) (10)
o 2 Vp P ' is also valid to first order in change of elastic properties.
Simplified Zoeppritz Equations 611
U. -0.5 -0.5
!1O"/R o. l n Appendix B I consider the degenerate case Rn = O. U.
UJ
(3) Wide angles.-The reflection amplitude at wide angles
relates only to the change in velocity. At sufficiently large
o
o
EX ACT
EQUATION 14
\
angles where the third term in equation (14) dominates the first z
o
EQ. 14 WITH 8=8
1
\
two, it becomes I-
o
EQUATION 18 \
-1.0
R(B) ~
1 !1Vp
- -
2 '
(tan B - sin' 0). (17)
W - 1.0
..J \
U.
2 Vp W
a: \
Analysis of wide-angle seismic records for velocity is a highly
developed subject to which equation (17) probably contributes \
nothing. However, its derivation does indicate the relation to
- 1. 5 L-L-.L--L--L---l.---.L--L-l........J........J...---l.--l----'_L..J.L-L-.L..J - 1 . 5
analysis of amplitude at intermediate angles. 80
0
90
0
The basis of the derivation of equations (I) and (9) from the
exact Zoeppritz equations is that the percentage change in
elastic properties is small, i.e., !1 VpIVp, tJ.I.~ v" and !1p/P are all FIG. 2. Approximations in an instance with !1V < O. The elastic
small compared to unity. For the vast majority of exploration properties correspond to an actual gas sand in the Gulf of
Mexico: Vp l = 7 570 ft/s, PI = 2.15, and 0"1 = 0.40 for the
overlying shale and V;'2 = 6 400 ft/s, P2 = 1.95, and 0"2 = 0.10
for the gas sand.
+5 ---------------r--""T'"--...,
+4
00 30
0
60 0 90 0
+3 +1.5
+2
I-
Z
EXACT
EQUATION 14
I
+1 t-_c::::::-
UJ
0 EQ14 WITH 8=8
1
I
+ 1.0 - + 1.0
U.
U.
UJ
EQUATION 18
I
0
u ~
-1
z
0 I
l-
o
W
+0.5 I +0.5
-2
..J
U. /
UJ
a: /
-3 /
'" >
0 0
0 0 0
-4 0 30 60 90
0
ANGLE FROM NORMAL lating transmitted angle e2 by equation (6) and then averaging
by equation (5).The resulting curve has the right shape but rises
10' 20' 30' 40' too soon in the case of a velocity decrease (I'. V < 0, Figure 2)
and too late in the case of a velocity increase (I'. V > 0, Figure
2 Ro 2 Ro 3). The error is not significant in the intermediate angles 0 <
8 1 < 30 degrees.
A further approximation is to omit entirely the third term in
Downloaded 03/26/17 to 212.26.2.132. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
equation (9) or equation (14). I argued above that the third term
Ro 1------1f---::====-+----f--- Ro
was unnecessary for incident angles less than 30 degrees. One
caution is that in the case A < 0, B > 0 omission of the third
term may lead to false prediction of a zero crossing at wide
oI----I------l---'~"'<_~----"'''''" -----'''f.<-----j 0 angles. Figures I, 3, 5, and 6 all illustrate this situation.
A final approximation is to replace sin 8 1 by e1 , which is
accurate for e1 < 30 degrees. The concatenation of all these
approximations is
-Ro
R = R o(1 + Aei). (18)
Figure 4 shows this parabolic form of variation in more detail
for a range of values of parameter A. Figures 5 and 6 compare
the parabolic approximation to the exact solution of the Zoep-
FIG. 4. Parabolic approximation to R(e) for different values of pritz equations for a range of !'.(J and both signs of I'.V. These
the dimensionless parameter A [eq uation (18)].
figures do not include an instance with B < 0, but that is rare as
discussed in Appendix A.
o.~ r----.----,------r--,----.,-----,---,----,-----,
DERIVATION OF KOEFOED'S RULES
Vz /V, =aS45
Pz /P, =a938 The five observations made numerically by Koefoed can be
tT,=a3 established analytically using equations (9) through (14). Equa-
...z _tTz=O.3 tion (14) states that an increase (decrease) of Poisson's ratio for
~ 0.0~------_::_!::.::-::::::::::;;:2~;",..".----- the underlying medium produces an increase (decrease) in the
j;:
...o
IL
--- reflection coefficient at larger angles of incidence. This agrees
with Koefoed's rules (a) and (b) but without the qualification
o
that "the underlying medium has the greater longitudinal ve-
z
...Q locity and other relevant properties of the two strata are equal
lil to each other."
..J
l:J- M When the Poisson's ratios for the two media are equal
0:
(!'.(J = 0), then this Poisson's ratio enters into R(e) only through
A o. Koefoed's rule (e) is equivalent to saying A o increases as (J
increases. The derivative of equation (12) is
2(1 + B)
(19)
10 30 "10 00 60 70 80 (1 - (J)2'
INCIDENT ANGLE (DEGREES)
so we see rule (c) is true except for B < -1. Figure 8 illustrates
the increase of A o with (J for the normal range of B. Koefoed
FIG. 5. Comparison of exact reflection coefficients (solid line) only considered the case B = 1.
with the parabolic approximation [equation (18)J in a case of Koefoed's rule (d) is derived from equation (11). The smaller
velocity decrease. is Ro ; the larger is the effect of a given !'.(J upon A. This is
apparently what Koefoed meant by rule (d).
Rule (e) follows from the observation that equation (14) [or
equation (9)J is linear in the three differences!'. Vp , !'.(J, and I'.p.
situations, in which the reflection coefficient R o does not exceed Therefore R(8) simply changes sign when the two sets of elastic
0.2, this is no problem. However, the derivation also assumes properties "I" and "2" are interchanged. Rule (e) breaks down
that both 8 1 and e2 are real and less than 90 degrees. This at large angles when the difference between 8 1 and 8 [equation
means that the approximation is good only to about e1 = 80 (5)J cannot be neglected.
degrees in the case of a velocity decrease (Figure 2) and to e1
about 10 degrees less than critical in the case of a velocity INVERSION FOR ELASTIC PROPERTIES
increase (Figure 3). It is not necessary that !'.(J/(J be small;
indeed, it has the value l.2 for Figure 2. A current problem is inversion of the reflection coefficient
At intermediate angles further simplifications are possible R(e) to obtain information about elastic properties (Gassaway
(Figures 2 and 3). The incident angle 8 1 can be used instead of and Richgels, 1983; Rosa, 1976). The diagonalization accom-
average angle e in equation (14). This saves the work of calcu- plished in equation (9) or equation (14) can simplify this inver-
Simplified Zoeppritz Equations 613
sion and clarify which combinations of elastic properties are approach to critical angle. Thus I have approximately diago-
well determined and which are poorly determined. nalized the multivariate relationship between elastic properties
Because the practical problems in recovering absolute reflec- and curve features. The coefficient for intermediate angles also
tion amplitude differ from the problems of recovering the rela- has two terms: one term is proportional to L'1cr, the contrast in
tive variation of reflection amplitude with offset, it is appropri- Poisson's ratio: and the other term is A o, which describes the
ate to consider the information content of the relative curve bland decrease of R(6) in the absence of contrast in Poisson's
R(B)!Ro. If reflections are not recorded much past 30 degrees ratio. When angles approaching critical are not included, R(B)
Downloaded 03/26/17 to 212.26.2.132. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
then equation (18) or Figure 4 could be used to extract a value may be adequately approximated by a parabola [equation
of A from the input R(B)!Ro. If wider angles are involved. then (18)]. The approximation provides an analytic basis for the
R(O)! Ro could be fit to the trigonometric series in equation (9). systematics first described by Koefoed, and also for a simple
In either case, the dimensionless parameter A is the only infor- inversion of R(B) to L'1cr.
mation to be extracted other than critical-angle behavior.
Figure 7 shows that A can be transformed to L'1cr R o provided ACKNOWLEDGMENTS
very approximate information is available or hypothesized
about average Poisson's ratio c and velocity.impedance ratio This work benefited from collaboration with many col-
B. In brief, the single piece of information about elastic proper- leagues: N. C. Banik and P. L. Love helped with the original
ties available from analysis of relative change of reflection algebra, R. 1. Meyer pointed out the need for a higher approxi-
amplitude with offset is L'1cr! R o. The multidimensional. nonlin- mation than equation (18), E. M. Himsworth and M. T. Kramer
ear inversion problem discussed previously (Rosa. 1976; Gas- demonstrated the applicability to problems of data interpreta-
saway and Richgels, 1982) is reduced to a one-dimensional tion, and 1. R. Resnick helped with the final revision.
linear problem.
As I pointed out, R o is half the change in natural logarithm of
impedance [equation (16)]. Thus the information !\crjR o is REFERENCES
equivalent to slope on a crossplot of Poisson's ratio versus Aki. K. L. and Richards, P. G., 1980, Quantitative seismology: W. H.
impedance p Vp on a logarithmic scale. This could be the basis of Freeman and Co.
Bortfeld, R.. 196I. Approximation to the reflection and transmission
a graphic procedure to relate information about R(B) to infor- coefficients of plane longitudinal and transverse waves: Geophys,
mation about lithology. Prosp., 9, 485-502.
Gardner, G. H. F.. Gardner, L. W., and Gregory, A. R., 1974, Forma-
tion velocity and density-The diagnostic basis for stratigraphic
SUMMARY traps: Geophysics, 39, 770-780.
Gassaway. G. S.. and Richgels. H. L, 1983, SAMPLE: Seismic ampli-
tude measurement for primary lithology estimation: Presented at the
I took the known linearization of the P-wave reflection coef- 53rd Annual International SEG Meeting, September, Las Vegas:
ficient R(B) [equation (l)J and transformed variables from ~ to abstr. book,61Q-6lJ
Gregory. A. R., 1976, Fluid saturation effects on dynamic elastic
c to display analytically the effect of L'1cr, the contrast in Pois- properties of sedimentary rocks: Geophysics, 41,895-921.
son's ratio. The result [equations (9) or (l4)J was arranged into Hamilton. E. L. 1979, ~.V, and Poisson's ratios in marine sediments
three terms which contribute to three distinct features of the and rocks: J. Acoust. Soc. Amer.,66,1093 1101.
Hilterman. F. J.. 1983. Seismic lithology: Presented as a continuing
R(B) curve: (1) the normal-incidence magnitude. (2) the behav- education course at the 53rd Annual International SEG Meeting,
ior at intermediate angles of about 30 degrees, and (3) the September. Las Vegas.
A
6
Ve/V, = 1.183
Pe/P, = t. 067
e
Ue =0.3
f-
Z
w
U 0.5
...iLw
o -+---+---+---,.c,4--,4."L----+---I.- Ao
o
z Ro
o
E -- -17, =0.2
w
~ o.of----------=~~======~--e::_------_i
w
II: <,
..... <,
<,
<,
<,
'u, =0.4
Koefoed, 0., 1955, On the effect of Poisson's ratio of rock strata on the
reflection coefficients of plane waves:Geophys.Prosp., 3,381-387. lD
Muskat, M., and Meres, M. W., 1940, Reflection and transmission o
coefficients for plane waves in elasticmedia: Geophysics, 5,115-155. t=
Ostrander, W. 1., 1984, Plane-wave reflection coefficients for gas sands <
C[ -+1.0 + 1.0
at nonnormal anglesof incidence: Geophysics, 49, 1637-1648. W
o
Rosa, A. L. R., 1976, Extraction of elastic parameters using seismic Z
reflection amplitude with offset variations: M.Sc. thesis, Univ. of <
Cl
Houston. W
Q.
Sherwood, 1. W. C, Hilterman, F. 1., Neale, R. N., and Chen, K. C., :::;
Downloaded 03/26/17 to 212.26.2.132. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
T
1983, Synthetic seismograms with offset for a layered elastic >
I-
medium: Presented at the 1983 OffshoreTechnology Conference. oo
..J
W
>
.20 .30 .40