GEOPHYSICS, VOL. 67, NO. 1 (JANUARY-FEBRUARY 2002); P. 60–62, 2 FIGS.

10.1190/1.1451331
Downloaded 11/09/15 to 132.203.227.62. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Short Note

Elastic impedance normalization

David N. Whitcombe∗

INTRODUCTION [EI(θ )n+1 − EI(θ )n ]

R(θ ) = . (2)
The elastic impedance, or EI, function has enabled far-offset- [EI(θ )n+1 + EI(θ )n ]
angle stack data to be inverted using technology developed for He showed that EI could be approximated as a simple function
acoustic impedance inversion. An undesirable feature of the EI of α, β, and ρ,
function has been that its dimensionality varies with incidence
angle θ and provides numerical values that change significantly EI(θ ) = α a β b ρ c , (3)
with θ . These problems have been overcome by modifying the
where the exponents a, b, and c are functions of incidence
EI function with constants αo , βo , and ρo . These modifications
angle θ:
allow for a direct comparison between elastic impedance values
across a range of angles in a manner that was not available with a = (1 + sin2 θ ),
the previous formulation. The modifications neither improve
nor degrade the accuracy of the reflectivity that can be derived b = −8K sin2 θ,
from the EI function.
c = (1 − 4K sin2 θ ). (4)
ELASTIC IMPEDANCE

Elastic impedance was developed in the early 1990s to aid The variable K is taken to be a constant over the interval of
exploration and development in the new Atlantic Margin log of interest. Constant K is an approximation that limits the
province west of the Shetland Islands (Connolly, 1999). The accuracy of the EI equation. For an interface where
à ! à !
use of impedance data improved the understanding of the ge- βn2 βn+1
2
ology across the multidisciplinary team. However, much of the +
seismic data exhibited a Class 3 behavior, with hydrocarbons
αn2 αn+1
2
K = , (5)
being more visible at far offset. The use of elastic impedance 2
combined the benefits of working with inverted data and with
the reflectivity coefficient predicted from the EI equations will
far-offset data.
be the same as the reflectivity predicted by the two-term lin-
Elastic impedance aids the inversion of nonzero offset data
earized Zoeppritz form [equation (1)]. For a section of log over
because it provides a log trace derived from a set of P-wave
an interval of interest, the constant K value can be determined
velocity, S-wave velocity, and density logs (α, β, ρ), consistent
by averaging (βn2 /αn2 ) over the interval. Alternatively, the con-
with the reflectivity of a far-offset angular projection. The EI
stant K value can be determined by finding the value of K
log can be used to calibrate far-offset-angle stack data in the
that minimizes the errors in the reflection coefficients over the
same way that acoustic impedance logs are used to calibrate
interval (Connolly, 1999).
zero-offset seismic data.
Connolly (1999) also demonstrated that by starting with the
Connolly started with the two-term linearization of the
three-term linearization of the Zoeppritz equation (Aki and
Zoeppritz equation (Aki and Richards, 1980),
Richards, 1980),
R(θ) = A + B sin2 θ, (1)
R(θ ) = A + B sin2 θ + C sin2 θ tan2 θ, (6)
and defined elastic impedance as analogous to acoustic
impedance in terms of impedance changes from formation n the EI function [equation (3)] still applies, after modifying the
to formation n + 1: exponent a in equation (4) as

Manuscript received by the Editor July 18, 2000; revised manuscript received March 20, 2001.
∗
BP, Farburn Industrial Estate, Dyce, Aberdeen AB21 7PB, U.K. E-mail: whitcodn@bp.com.
°c 2002 Society of Exploration Geophysicists. All rights reserved.

60
 Elastic Impedance Normalization
a = (1 + tan2 θ).
One of the criticisms of the EI function [equation (3)] is that
its dimensionality varies with θ . Additionally, or because of
this, the EI values vary significantly with θ. This is illustrated
in Figure 1, which shows the average EI for the 204/24a-2 well,
Downloaded 11/09/15 to 132.203.227.62. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
west of Shetland, plotted as a function of θ. This property makes
displaying the AI and EI logs together inconvenient. To over-
come this, the EI logs are usually scaled by a factor appropriate
to the particular angular projection being used. Note that the
EI function [equation (3)] always predicts the AI log at θ = 0
without any scaling.
This note gives a simple way of normalizing the EI equations
so the dimensionality remains constant with θ and EI values in
the normal AI range are returned for all angles θ.
VerWest et al. (2000) describe an alternative approach to
the problem. They use a different approximation for elas-
tic impedance, derived from a different expansion of the
Zoeppritz equations. This approach appears to give highly ac-
curate reflection coefficients, particularly at very high angles of
incidence where the three-term linearization begins to break
down.
NORMALIZATION
To remove the dimensionality as a function of θ, I introduce
constants αo , βo , and ρo and modify the EI function [equa-
tion (3)] to be
µ ¶a µ ¶b µ ¶c
α β ρ
EI(θ) = . (8)
αo βo ρo
If the values of these constants are chosen to be averages
of the α, β, and ρ logs, then EI(θ) will vary around unity. This
modification removes the dimensionality dependence and sta-
bilizes the function. If we further scale this function by a factor
FIG. 1. The average value for the EI logs for the 204/24a-2 well,
west of Shetland, plotted as a function of incidence angle θ.
The shape of this curve depends on the units in which the data
have been specified. In this example, velocities were measured
in meters per second and density was measured in grams per
cubic centimeter.
61
(7) αo ρo , the dimensionality of EI becomes the same as AI and we
find that EI(θ ) predicts the correct value of acoustic impedance,
αρ, at θ = 0:
"µ ¶a µ ¶b µ ¶c #
α β ρ
EI(θ ) = αo ρo . (9)
αo βo ρo
To achieve the new normalized form of EI, we have effec-
tively scaled the original definition of EI [equation (3)] by the
factor αo1−a βo−b ρo1−c . We can deduce from the reflectivity defini-
tion of elastic impedance [equation (2)] that applying a scaling
factor to EI does not alter the calculated reflectivity, since the
scaling factor appears in both the numerator and denomina-
tor of that equation. The modifications have therefore neither
degraded nor improved the accuracy of the EI function.
We can deduce from the new normalized form of EI [equa-
tion (9)] that for a formation with α, β, ρ values equal to
αo , βo , ρo , respectively, the elastic impedance in that forma-
tion will remain constant with increasing angle θ with a value
of αo ρo , the acoustic impedance of the formation. We could
therefore design the EI function to remain constant for the
reservoir cap rock by setting αo , βo , ρo equal to the properties
for that layer.
Figure 2 shows AI and EI (30◦ ) logs for the 204/24a-2 well.
In the upper example, αo , βo , ρo values have been used that
are appropriate to the start of the log. In the second example,
FIG. 2. AI and EI (30◦ ) traces for the 204/24a-2 well, west of
Shetland. (a) The αo , βo , ρo values have been used appropriate
to the start of the log. The AI and EI logs intersect at this point.
(b) The αo , βo , ρo values have been used appropriate to a time
of 2150 ms, appropriate to the shales between the pay zones at
approximately 2130 and 2180 ms.
 62
αo , βo , ρo values have been used that are appropriate to shales
close to the two pay zones in the well.
CONCLUSIONS
The new normalized form of elastic impedance [equa-
Downloaded 11/09/15 to 132.203.227.62. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
tion (9)] allows values to be derived that do not vary rapidly
with incident angle θ. The new constants, αo , βo , ρo , used in the
modified equation allow the EI function to be normalized so
that the elastic impedance of a particular layer does not change
with θ . These modifications allow for a direct comparison be-
tween elastic impedance values across a range of angles in a
manner that was not available with the previous formulation.
Whitcombe
ACKNOWLEDGMENTS
My thanks to Roger Reagan, Pat Connolly, and Terry
Redshaw for comments on improving the manuscript. I also
thank the board of BP for permission to publish this paper.
REFERENCES
Aki, K. I., and Richards, P. G., 1980, Quantitative seismology: W.H.
Freeman & Co.
Connolly, P., 1999, Elastic impedance: The Leading Edge, 18, 438–
452.
VerWest, B., Masters, R., and Sena, A., 2000, Elastic impedance in-
version: 70th Ann. Internat. Mtg., Soc. Expl. Geophys., Expanded
Abstracts, 1580–1582.