Professional Documents
Culture Documents
Dec2008 Avo Crossplotting
Dec2008 Avo Crossplotting
Dec2008 Avo Crossplotting
A Practitioner’s Perspective
Heath Pelletier
Talisman Energy, Calgary, Alberta, Canada
The development of AVO crossplot analysis has been the AVO Crossplot Rotations
subject of much discussion over the past decade and has
provided interpreters with new tools for meeting exploration The key value of AVO crossplot interpretation is the ability to
objectives. Papers by Ross (2000) and Simm et al. (2000) differentiate population outliers relative to background trend
provide blueprints for performing AVO crossplot interpreta- points within crossplot space. Incorporating direct hydro-
tion. These articles refer to the Castagna and Swan (1997) carbon indicators (DHIs) via crossplotting can play a signifi-
paper which laid the foundation for AVO crossplotting. The cant role in minimizing the risk associated with an
AVO classification scheme presented by Castagna and Swan, exploration play. The stability of the background trend (fluid
which was expanded from the work of Rutherford and angle) can have an impact on what is being interpreted as
Williams (1989), has become the industry standard. Castagna anomalous, be it fluid or lithology induced outliers. Castagna
and Swan also investigated the behavior of constant Vp/Vs and Swan’s (1997) Interpreter’s Corner article concluded that
trends, concluding that for significant variations in Vp/Vs many background trends can be erroneously superimposed
many different trends may be superimposed within AVO in AVO crossplot space, particularly if too large a depth range
crossplot space, making it difficult to differentiate a single is brought into the interpretation window. This presents diffi-
background trend. A misapplication of this concept has been culties when performing crossplotting analysis, particularly
to infer a direct correlation between these changing back- as it pertains to non-uniqueness. For a given anomaly (popu-
ground Vp/Vs trends with rotating intercept/gradient cross- lation outliers) an interpretation of an increase in Vp/Vs (due
plot slopes (what Gidlow and Smith (2003) call the fluid to a superimposed background trend) can be just as feasible
factor angle, and Foster et al. (1997) the fluid line) observed in
as a decrease in Vp/Vs (due to a fluid response or coal)
seismic data. The central question of this paper is: when is
dependent upon the choice of background trend. The
this background trend (or fluid line) rotation a representation
Castagna and Swan crossplot template, seen in figure 1a, is
of real geology and when is it a processing-related phenom-
often used in the literature to explain crossplot behavior
enon? To answer this question I review the theoretical expec-
observed in seismic data. However, the constant Vp/Vs lines
tations regarding rotating AVO crossplot trends, the role of
seismic gather calibration (or lack thereof), and the value of used in the design of this crossplot template do not reflect
various compensating methods. Furthermore, I investigate most worldwide compaction trends and established empir-
the appropriateness of using constant Vp/Vs lines in a cross- ical mudrock line relationships. Figure 1b highlights how
plot template by examining both the mathematical and constant Vp/Vs lines cut across a given mudrock line only
modeled AVO crossplot responses which incorporate an within a limited range. Outside these overlap zones the
established compaction trend. I conclude that when constant Vp/Vs lines would be considered physically unreal-
exploring in a reasonably compacted environment (Vp/Vs istic. A follow-up Interpreter’s Corner article by Sams (1998)
ratios of 1.6-2.4) a relatively small range of fluid angles (back- also questions the appropriateness of using these constant
ground trend rotations) can be expected. Large variations of Vp/Vs trends in crossplot templates. Sams demonstrates that
fluid angle observed in seismic data can be attributed to the constant Vp/Vs lines approach a mudrock line relationship
difficulty in preconditioning gathers for AVO analysis. only at very high P-wave velocities.
Figure 1. AVO behavior in crossplot space. (a) Castagna and Swan’s (1997) background trend illustrates AVO crossplot rotation as Vp/Vs varies. (b) The empir-
ical relationship that exists between Vp and Vs cuts across constant Vp/Vs lines. Image modified from Scott Pickford poster (2000). (c) Castagna et al.’s (1998)
revised background trend rotations incorporating a linear Vp versus Vs trend. The amount of rotation present in the crossplot that incorporates a mudrock rela-
tionship is significantly less than for the constant Vp/Vs scenario (for all but extremely low velocity ranges), suggesting that large rotations in crossplot space should
not be considered normal in most seismic datasets. This also presents a case for utilizing larger windows in crossplot analysis than previously considered.
Continued on Page 41
Figure 2. AVO Model Example. (a) An AVO model was built using: Vp honouring a compaction trend; Density estimated using the Gardner et al. (1974) equation; Vs
estimated using the Castagna et al. (1985) equation. These logs and a seismic wavelet (b) were used to create an offset synthetic. (c) AVO crossplot of the modeled data high-
lighting: the background trend line; data points associated with wavelet side lobe effects; and rotated data points stemming from the lower velocity reflections. This model is
a confirmation of figure 1c where background trends lines (fluid lines) are closely spaced in compacted sediments (low Vp/Vs).
Continued on Page 42
Figure 3. AVO Crossplot interpretation. (a) Using the Castagna et al. (1985) equation for estimating shear velocity produced numerically unrealistic <Vp/Vs> values in
the low velocity (shallow) section. The higher <Vp/Vs> contrasts experience the most rotation in I/G crossplot space (identified by letters A-F). Compacted rocks (<Vp/Vs>
values of 2.42 or lower) were concentrated on a relatively tight background/fluid trend line. (b) Using the Han et al. (1986) equation to estimate shear velocity produced
similar results as the Castagna case for most of the reflectors G-U. The exception being that the <Vp/Vs> values in the shallow section reflectors A-F seemed more reason-
able and the fluid angle rotation was less dramatic as a result.
Continued on Page 43
Figure 4. The AVO Intercept and Gradient components. (a) The Intercept term (R0) shows an increase in reflection coefficient magnitude as velocity decreases. (b) The
Gradient term (R2) experiences the most dramatic change when velocities become low, eventually culminating in a sign reversal. After a low threshold velocity is reached,
as observed in shallow environments, large magnitude seismic amplitudes increasing with offset are still representing wet/brine rocks. (c) The previous I/G plots incorpo-
rate several ranges of impedances (reflection coefficients - RC) that reflect a typical compaction setting (modified from Castagna and Swan (1997)). For reference the AVO
crossplot quadrant is specified thereby imparting information regarding the polarity of both I and G. (d) The Vp/Vs versus Vp plot reflects where the crossover point in AVO
gradient will take place petrophysically, at Vp/Vs > 2.8*.
Continued on Page 44
Processing for AVO / Calibration necessary to crossplot the AVO attributes, before and after the
process to test the response on known population outliers.
Cambois (2000) defines a processing workflow meant to be AVO
friendly simply as “any sequence that makes the data compatible The goal of these true amplitude flows is to restore the amplitude
with Shuey’s equation” (or various other Zoeppritz appro x i m a- (and phase) response to a point where geologic meaning can be
tion methodologies). True Amplitude, Preserved Amplitude, and inferred. It is important to understand that the gradient behavior
Controlled Amplitude Controlled Phase (CACP) a re a few examples. (as a function of time) is dependent upon the processing applied
Many of these workflows include deterministic and/or statistical to that data. Revisiting figure 4c, we observe that all background
corrections calculated by incorporating well logs and/or geologic trend rocks have amplitudes that decay as a function of offset. In
models. These workflows may comprise larger amplitude corre c- other words, positive intercepts have negative gradients, while
tions, such as geometrical spreading and absorption (Q), or more negative intercepts have positive gradients (For the sake of
subtle ones, like angle of emergence and array corrections. simplicity I am generalizing to include only the consolidated
Typically these workflows apply time and offset variant corre c- rocks, not the rarer low velocity regimes discussed previously).
tions in an attempt to compensate for the earth filter. Several The rate at which these gradients decay can be very much
authors have discussed at length the difficulties inherent in this affected by the processing applied, especially when offset
process, including Cambois (2001) and Bachrach; Kozlov and varying applications are present. The first check when quality
Ivanova; Landro and Stavos (2006). controlling CDP gathers is to verify that the majority (i.e. non-
anomalous) of the reflection events do indeed decay. This decay
Processing workflows like the one described by Ramos (1998) should be noticeable over the angle of incidence range of 0-40
highlight the need to compare and contrast the AVO behavior degrees, and then the amplitudes can sharply increase again as
after every processing step. Qualitative and quantitative evalua- they approach the critical angle. Note: this means that for shallow
tion involving the amplitudes of our primaries is performed to data the amplitude decay will happen relatively quickly over a
determine if any gradient responses have been altered. Gather given offset range compared to the decay that will be more
difference plots are an interpreter’s best friend when main- gradual deeper in the section over the same offset range. Another
taining quality control. Some QC methods need to be more way of expressing this is: as time increases the angle of incidence
sophisticated in order to quantify any possible changes. For range decreases.
example, the diff e rence of a gather pre- and post-spectral
whitening cannot be used because the frequency content of the A common misunderstanding is to assign a correlation between
primaries has been altered too dramatically. In these cases it is the rotating intercept/gradient crossplot slope observed in
Figure 5. Vp/Vs plotted versus depth for 4 basins around the world. (a) The Gulf of Mexico (from Castagna et al. (1985)), (b) Deepwater Gulf of Mexico (from Smith and
Sondergeld (2001)), (c) Western Canada, Alaska, and Qatar. Except for the unconsolidated shallow GOM sediments the vast majority of the Vp/Vs ratios are less than 2.25
which suggest we would expect relatively small deviations in crossplot fluid angle behavior.
Continued on Page 45
Continued on Page 46
Conclusion difficult; however, simple and eff e c t i v e Downton, J., and Chaveste, A., 2004, Calibrated Three-
term AVO to Estimate Density and Water Saturation,
alternatives are available. For most 2004 CSEG National Convention, Technical Abstracts.
AVO crossplotting techniques and appli- datasets, sticking to an AVO friendly Foster, D. J., Keys, R.G., and Schmitt, D. P., 1997,
cations have evolved over the last decade. p rocessing flow and applying a time- Detecting subsurface hydrocarbons with elastic wavefields,
Deciding which AVO crossplot template variant fluid factor correction after AVO in Chavent, G., Papanicolaou, G., Sacks, P., and Symes,
to apply in a given play requires a petro- W., Eds., Inverse problems in wave propagation:
extraction is a practical workaround to a Springer-Verlag.
physical understanding of the enviro n- r i g o rous calibration process. Finally, Gardner, G.H. F., Gardner, L.W., and Gregory, A. R.,
ment. For instance, when logs are although it may be tempting to skip the 1974, Formation velocity and density – The diagnostic
indicating a well compacted environment AVO crossplot analysis altogether to basics for stratigraphic traps, Geophysics, Society of
(Vp/Vs ratios of 1.6-2.4) a relatively small Exploration Geophysicists, 39, 770–780.
perform an inversion (the ultimate form
range of fluid angles (background trend Gidlow, P. M., and Smith, G. C., 2003, The fluid factor
of calibration) it is important to be angle, 65th Annual Conference and Exhibition, EAGE,
rotations) can be expected in the seismic cautious and not trust the inversion Extended Abstracts, E27.
data during the course of an AVO investi- implicitly. Any anomalies identified by an Gidlow, P.M., Smith, G.C., and Vail, P.J., 1992,
gation. This can allow for more latitude in inversion analysis should always be Hydrocarbon detection using fluid factor traces: A case
crossplot analysis strategies. For example, history, SEG/EAEG Summer Workshop, 78-79.
confirmed in AVO crossplot space as well
larger time windows can be brought into Han, D-H., Nur, A., and Morgan, D.,1986, Effects of
as reviewed on the input gathers them- porosity and clay content on wave velocities in sandstones,
crossplot space than previously consid- selves to rule out other, less geologically Geophysics, Society of Exploration Geophysicists, 51,
ered. It is important to understand how driven, causes. R 2093-2107.
calibrated the seismic data is beforehand Jin, S., Cambois, G., and Vuillermoz, C., 2000, Shear-
when performing crossplot analysis. wave velocity and density estimation from PS-wave AVO
Acknowledgements analysis: Application to an OBS dataset from the
Background trend rotations observed in North Sea, Geophysics, Society of Exploration
the seismic, as a function of time/depth, Special thanks to David D’Amico, Hugh Geophysicists, 65, 1446-1454.
are typically not re p resentative of the Geiger, and Brian Russell for helping this Ramos, A., 1998, AVO processing calibration, The
local geology. The AVO fluid angle experi- Leading Edge, Society of Exploration Geophysicists,
article to evolve through the various 8, 1075-155.
ences dramatic variations only in stages of development. Ross, C. P., 2000, Effective AVO crossplot modeling: A
e x t remely low velocity enviro n m e n t s , tutorial, Geophysics, Society of Exploration
often producing polarity shifts in the AVO Geophysicists, 65, 700-711.
References
gradient. The calibration process can be R u t h e r f o rd, S. R., and Williams, R. H., 1989,
Bachrach, R.; Kozlov, E., Ivanova, N.; Landro, M., and Amplitude-versus-offset variations in gas sands,
Stavos, A., 2006, Expert Answers, CSEG RECORDER, Geophysics, Society of Exploration Geophysicists, 54,
October 2006, 9-15. 680-688.
Cambois, G., 1998, AVO attributes and noise: pitfalls of Sams, M., 1998, Yet another perspective on AVO cross-
crossplotting, EAGE, Abstract 2-14. plotting, The Leading Edge, Society of Exploration
Cambois, G., 2000, Can P-wave AVO be quantitative?, Geophysicists, 17, 911–917.
The Leading Edge, Society of Exploration Shuey, R.T., 1985, A simplification of Zoeppritz equations,
Geophysicists, 19, 1246-1251. Geophysics, Society of Exploration Geophysicists, 50,
Cambois, G., 2001, AVO processing: Myths and Reality, 609–614.
CSEG RECORDER, March 2001, 30-33. Simm, R., White, R., and Uden, R., 2000, The anatomy
Castagna, J. P., Batzle, M. L., and Eastwood, R. L., of AVO cro s s p l o t , The Leading Edge, Society of
1985, Relationships between compressional and shear- Exploration Geophysicists, 19, 150-155.
wave velocities in clastic silicate rocks, Geophysics, Smith, G., and Gidlow, P. M., 1987, Weighted stacking
Society of Exploration Geophysicists, 50, 551–570. for rock property estimation and detection of gas,
Castagna, J. P., and Swan, H.W., 1997, Principles of Geophysical Prospecting, EAGE, 35, 993–1014.
AVO crossplotting, The Leading Edge, Society of Smith, T., Sondergeld, C., 2001, Examination of AVO
Exploration Geophysicists, 17, 337–342. responses in the eastern deepwater Gulf of Mexico,
Castagna, J. P., and Swan, H.W., and Foster, D. J., Geophysics, Society of Exploration Geophysicists, 66,
1998, Framework for AVO gradient and intercept inter- 1864-1876.
pretation, Geophysics, Society of Exploration Thomsen L., SEG DISC 2002 manual.
Geophysicists, 63, 948–956.