AVO and AVA tuning and stretch eects

Bjrn Ursin

and Martin Tygel

Norwegian University of Science and Technology (NTNU),

Department of Petroleum Engineering and Applied Geophysics,
S.P. Andersenvei 15A, NO-7491 Trondheim, Norway

State University of Campinas (UNICAMP),

Applied Mathematics Department, IMECC,
CP 6065, Campinas SP, Brazil
(October 30, 2007)
ABSTRACT
The primary PP, SS or PS reection response from a thin layer embedded in a homogeneous
medium is approximately given by the delayed derivative of the wavelet times the reection
coecient times a tuning amplitude factor. This results in an amplitude decrease for a
common-oset image gather and an amplitude increase for a common-angle image gather.
The alignment of a common-image gather to the normal-incidence reection results in a
wavelet stretch for a common-oset image gather and a wavelet squeeze for a common-
angle image gather.
For a reection from a single interface there are the same stretch and squeeze eects,
but there is no amplitude eect. The common-angle image gathers have higher resolution
than the common-oset gathers, and they can be used directly in an inversion scheme for
elastic parameters.
1
INTRODUCTION
A number of applications in reservoir characterization and production is based on the anal-
ysis of the amplitudes of a primary reection as a function of oset, a technique referred
as AVO analysis. After suitable correction for the oset-dependent geometrical spreading,
and also disregarding transmission eects, the so-called true amplitudes provide a measure
of the reection coecient as a function of oset. A further transformation of oset into
incidence angle provides, then, the reection coecient as a function of angle (AVA). A
simple oset-to-angle transformation is described in Appendix A. Under the use of avail-
able approximations of the angle-dependent reection coecient (as a function of velocity
and density contrasts or appropriate combinations of these parameters), useful lithology
and uid information can be estimated.
In the case the single target interface is replaced by a composite of one or more thin
layers, a similar procedure can be performed once the seismic response of the layer composite
is adequately formulated. For a single thin acoustic layer embedded in a homogeneous
acoustic background of lower impedance and in the zero-oset situation, Widess (1973)
described the behavior of the reection amplitudes as a function of the layer thickness and
the dominant wavelength the wavelet. Considering a zero-phase cosine wavelet, Widess
(1973) reports that maximum constructive interference occurs when the layer thickness is
one-quarter of the dominant wavelength. When the layer thickness reduces to one-eight of
that wavelength (called by Widess the theoretical threshold of resolution), the composite
response has the shape of the derivative of the incident wavelet. Corresponding results for
the case of a thin layer between two dierent homogeneous media were given in Chung and
Lawton (1995). Lin and Phair (1993) extended the results of Widess (1973) to the oset
2
domain, showing also examples of improved images obtained under the proposed tuning
corrections. Finally, Bakke and Ursin (1998) provided unied tuning corrections that are
valid for general osets and arbitrary seismic wavelets beyond the dominant-wavelength
approximation. These expressions agree with the ones of Widess (1973) and Lin and Phair
(1993) in the case of a cosine wavelet. A discussion on the literature on tuning eects in
AVO is also provided in Bakke and Ursin (1998).
Signicant advantages can be gained when amplitudes (namely, reection coecients)
are extracted from the data as a function of incidence angle (AVA) instead of oset (AVO).
Ray-based imaging in the angle domain tend to be more reliable as they suer less from
multi-pathing and other artifacts, that are present in the oset domain (Brandsberg-Dahl
et al., 2003). Migration directly into angle and depth coordinates with 3-D corrections for
geometrical spreading results in gathers which can be used directly for AVA analysis and
inversion for elastic parameters. As shown, for example in Ursin et al. (2005a,b), angle-
domain common-image gathers can be used very eectively for imaging and tomography
applications. In this way, there is a motivation to extend present results on tuning amplitude
and wavelet stretch eects from oset (AVO) to angle domain (AVA).
Here we show that simple modications in the AVO expressions of Bakke and Ursin
(1998) provide corresponding expressions for AVA. The thin-layer tuning eect results in
reduced amplitude in a common-oset image gather and increased amplitude in a common-
angle gather. The alignment of traces to the zero-oset or normal-incidence trace results
in a wavelet stretch for a common-oset gather and a wavelet squeeze for a common-angle
gather.
The obtained results are in accordance with the behavior of common-image gathers from
3
the Jotun data example presented by Sollid and Ursin (2003), which show that common-
angle image gathers have higher resolution than common-oset image gathers.
TUNING EFFECTS FOR A COMMON-OFFSET GATHER
We consider a thin layer imbedded in a homogeneous background. We assume isotropic
elastic media both in the thin layer and the background. In principle, both media can
be transversely isotropic (VTI). In this situation, the reection coecients at the top and
bottom layer interfaces satisfy (Stovas and Ursin, 2003)
R
t
() = R
b
() = R(). (1)
where t denotes top, and b denotes bottom. We assume straight rays for the bottom
reector, and also that the reection angles at the top and bottom reectors are the same,
denoted by . Both these assumptions are justied by the thin-layer condition. In the
following, we shall use, besides R(), also the notation R(x), when oset, x, instead of
angle, = (x), is being considered.
Referring to Figure 1, we consider PP- or SS-primary reections from top and bottom
layer interfaces for a source-receiver pair at a given xed oset, x. We assume that their
amplitudes have been corrected for oset-dependent geometrical spreading (Ursin, 1990)
and also that transmission losses and intra-bed multiples can be neglected. In the zero-
order ray approximation, the top and bottom layer primaries, p
t
(x, t) and p
b
(x, t), where x
is oset and t is time, can be written
p
t
(x, t) = R(x)s(t T(x)) (2)
and
p
b
(x, t) = R(x)s(t T(x) T(x)), (3)
4
where T(x) is the traveltime to the top interface, T(x) the traveltime across the layer and
s(t) denotes the source wavelet. The thin layer assumption implies that T(x) is small, as
compared with T(x). Following Bakke and Ursin (1998), the composite response, p
c
(x, t),
of the two primaries can be conveniently approximated as
p
c
(x, t) = p
t
(x, t) + p
b
(x, t) = R(x)[s(t T(x)) s(t T(x) T(x)]
R(x) T(x)s

(t T(x)), (4)
where s

(t) is the derivative of s(t).

In accordance with Figure 1, the traveltime dierence for reection primaries at top and
bottom interfaces (also called dierential traveltime), T(x), is given by (Lin and Phair,
1993; Bakke and Ursin, 1998)
T(x) =
2z
v
cos (x) = T(0) cos (x), (5)
in which z denotes the layer thickness, T(0) is the vertical traveltime and v is the
velocity within the layer. Substitution into equation (4) yields the approximation
p
c
(x, t) = [R(x)T(0) cos (x)] s

(t T(x)). (6)
We see that the impact of tuning on the single-layer situation can be as described by Bakke
and Ursin (1998):
(a) The tuning amplitude is R(x)(x) = R(x)T(0) cos (x) instead of simply R(x) for
a single interface;
(b) The tuning wavelet is s

(t) instead of simply s(t) for a single interface.

The tuning behavior of amplitude has the important implication for AVO, namely that
under tuning conditions, there is a decrease in amplitude with oset as compared with the
single interface situation.
5
TUNING EFFECTS FOR A COMMON-ANGLE GATHER
We now refer to Figure 2, which depicts the common-angle gather situation, as encountered
in common-angle migration (see, e.g., Sollid and Ursin, 2003). We again assume straight
rays for the reection from the bottom layer, and now the dierential traveltime, T(), is
given by
T() =
2z
v cos
= T(0) sec . (7)
Substituting into equation (4) we obtain (compare with the expression (6) for the case of a
common-oset gather)
p
c
(, t) = [R()T(0) sec ] s

(t T()). (8)
Equation (8) has the following important consequence for AVA purposes: For AVA
analysis derived from common-angle gathers, there is an increase in amplitude with incident
angle as compared with the single interface situation.
WAVELET STRETCH EFFECTS
We consider stretch eects that are observed in p
c
(x, t) (common-oset gathers) and p
c
(, t)
(common-angle gathers) due to traveltime corrections. For common-oset gathers, the
stretch factor is
S(x) =
T
new
(x)
T
old
(x)
=
T(0)
T(x)
=
1
cos (x)
> 1, (9)
which means a wavelet stretch increase. For common-angle gathers, the stretch factor is
S() =
T
new
()
T
old
()
=
T(0)
T()
= cos < 1 , (10)
which means a wavelet squeeze.
6
These stretch eects will inuence the wavelet for a simple reection, and also the
composite wavelet from a thin layer. In Appendix B it is shown that the PS-reected
wave shows a similar eect. The wavelet stretch eects are clearly seen in the Jotum data
example from Sollid and Ursin (2003), shown in Figure . It is evident that the wavelets
in the common-oset image gathers to the left, are much broader than the ones for the
common-angle image gathers to the right, both for the PP and PS reections.
CONCLUSIONS
The composite reection from a thin layer can be approximated by the reection coecient
times the derivative of the wavelet together with an amplitude change. For a common-oset
image gather, there is an amplitude decrease and a wavelet stretch. For a common-angle
image gather, there is an amplitude increase and a wavelet squeeze.
For a simple reection, the stretch and squeeze eects are the same, but there is no
amplitude change. These eects have been observed on pre-stack depth migrated data,
and they explain why the common-angle image gathers have better resolution then the
common-oset image gathers.
ACKNOWLEDGMENTS
This work was kindly supported by the National Council of Scientic and Technological
Development (CNPq), Brazil, the Research Foundation of the State of Sao Paulo (FAPESP),
Brazil, the Norwegian Council of Research via the ROSE project and the sponsors of the
Wave Inversion Technology (WIT) Consortium, Germany.
7
REFERENCES
Bakke, N. E. and Ursin, B. (1998). Thin-bed AVO eects. Geophysical Prospecting,
46(6):571587.
Brandsberg-Dahl, S., De Hoop, M. V., and Ursin, B. (2003). Focusing in dip and AVA com-
pensation on scattering-angle/azimuth common image gathers. Geophysics, 68(1):232
254.
Chung, H. and Lawton, D. (1995). Amplitude responses for thin beds: Sinusoidal approxi-
mation versus Ricker approximation. Geophysics, 60(1):223230.
Lin, T. L. and Phair, R. (1993). AVO tuning. In 63rd Annual International Meeting, pages
727730. Soc. of Expl. Geophys.
Sollid, A. and Ursin, B. (2003). Scttering-angle migration of ocean-bottom seismic data in
weakly anisotropic media. Geophysics, 68(14):641655.
Stovas, A. and Ursin, B. (2003). Reection and transmission reponses of layered transversely
isotropic viscoelastic media. Geophysical Prospecting, 51(5):447477.
Ursin, B. (1990). Oset-dependent geometrical spreading in a layered medium. Geophysics,
55(4):492496.
Ursin, B., De Hoop, M. V., Brandsberg-Dahl, S., and Sollid, A. (2005a). Seismic angle
migration. The Leading Edge, 24(8):637640.
Ursin, B., De Hoop, M. V., Foss, S. K., and Brandsberg-Dahl, S. (2005b). Seismic angle
tomography. The Leading Edge, 24(6):628634.
Widess, M. B. (1973). How thin is a thin bed? Geophysics, 38(6):11761180.
8
APPENDIX A
OFFSET TO ANGLE TRANSFORMATIONS IN 1-D MEDIA
For the sake of completeness, we reproduce here the main expressions that approximate the
reection angle, = (x), as a function of oset in 1D media. In this situation, we start
with the observation that the horizontal slowness,
p =
dT
dx
=
sin(z)
v(z)
(A-1)
is constant for a given ray and v(z) and (z) are the velocity and ray angle at depth z,
respectively. For a xed reector, we consider the hyperbolic traveltime approximation
T(x) =

T(0)
2
+
x
2
v
2
rms
, (A-2)
where T(0) is the vertical traveltime, x is oset and v
rms
is the rms-velocity, corresponding
to that reector. From the above two equations, it follows that
dT
dx
=
x
T(x)v
2
rms
. (A-3)
from which the reection angle from the target reector at the given oset, = (x),
satises
sin(x) =
v x
T(x)v
2
rms
, (A-4)
where v is the interval velocity at the reecting layer. As a consequence, we also nd that
cos (x) =

1
v
2
x
2
T(x)
2
v
2
rms
=
T(0)
T(x)

1 + (v
2
rms
v
2
)

x
T(0) v
2
rms

2
. (A-5)
9
APPENDIX B
TUNING AND STRETCH EFFECTS FOR A CONVERTED WAVE
For a converted PS-wave, the normal-incidence traveltime dierential is
T(0) = T
P
(0) + T
S
(0) =
z
v
P
+
z
v
S
, (B-1)
where v
P
and v
S
are the P- and S-wave velocities in the thin layer, respectively. For a
common-oset gather, the thin-layer tuning eect is derived as for a non-converted wave,
but now with (see Figure 1)
T(x) = T
P
(x) + T
S
(x) = T
P
(0) cos + T
S
(0) cos , (B-2)
where is the ray angle for the reected S-wave. It can be computed from Snells law
sin
v
P
=
sin
v
S
. (B-3)
It is easily seen that T(x) < T(0), which results in a wavelet stretch, and an amplitude
decrease for a thin layer reection.
For a common-angle gather, the dierential traveltime is (see Figure 2)
T(x) =
T
P
(0)
cos
+
T
S
(0)
cos
. (B-4)
We see that T(x) > T(0), so that for a common-angle gather there is a wavelet squeeze.
The thin-layer tuning eect results in an increase in reection amplitude.
10
LIST OF FIGURES
1 Geometry of a common-oset image gather.
2 Geometry for a common-angle image gather.
3 The Top Heimdal event is at about 2.1 km depth. This is Figure 14 of Sollid and
Ursin (2003).
11
z

z
c
o
s

Figure 1: Geometry of a common-oset image gather.Ursin & Tygel Figure1B

12
z

z
s
e
c

Figure 2: Geometry for a common-angle image gather.Ursin & Tygel Figure2B

13
Figure 3: The Top Heimdal event is at about 2.1 km depth. This is Figure 14 of Sollid and
Ursin (2003).
Ursin & Tygel Sollid1.ps
14