|
|
 |
|||||||||||||||||
|
|||||||||||||||||
 |
|||||||||||||||||
| JOURNAL HOME | HELP | CONTACT PUBLISHER | SUBSCRIBE | ARCHIVE | SEARCH | TABLE OF CONTENTS |
University of Calgary, Alberta, Canada
In January 1980, Milo Backus, then SEG President, published an article in GEOPHYSICS that summarized the progress in exploration seismology during SEG's first 50 years and listed 10 challenges for the future. Much has been achieved in the subsequent 25 years to answer his challenges—and some problems remain.
This article uses some insightful statements from Milo's article as a baseline for assessing technological progress in applied seismology since 1980 and for speculating about challenges that remain as we celebrate SEG's 75th Anniversary.
Milo's first three "challenges" were about the problems of seismic imaging:
In 2005, all of these statements need to be greatly amended. Although the seismic image is still not perfect, we have seen great strides in both temporal and spatial resolution. The problem of lateral heterogeneities has been effectively handled by prestack depth migration in which the wave equation has been effectively used to convert time to depth in cases where reliable velocity estimates are available. Nowhere are improvements in depth imaging more apparent than in recent subsalt seismic exploration in the Gulf of Mexico. The case history of Ratcliff et al. (1994) is an excellent example of how 3D prestack depth migration can greatly enhance our ability to pursue and even delineate subsalt petroleum traps. Figure 1 from this article shows a 3D prestack depth migration of a salt intrusion that is vastly superior to 2D poststack time migration results.
|
The next two statements by Backus about the 1980 state of seismology were:
The biggest advances in fluid detection (especially with gas sands) have been made in the areas of AVO (amplitude variation with offset) and rock physics. AVO was essentially unknown (no entry in the second edition of the Encyclopedic Dictionary of Exploration Geophysics, published in 1984) when Backus issued his challenges. Its debut into geophysics was a presentation by Ostrander at SEG's 1982 Annual Meeting. An avalanche of literature followed and SEG published a (quickly sold out and reprinted) book on the subject just a decade later. That book, incidentally, was co-edited by Backus, indicating that he was part of the solution to his own fourth challenge. AVO advanced so rapidly that the various effects were organized in 1989 in a now-standard classification scheme by Rutherford and Williams (which was extended by Castagna and Swan in 1997). An important paper by Hilterman in 1989 ("Is AVO the seismic signature of rock properties?") outlined the importance of AVO in determining lithology and fluid content. AVO advances were coupled with advances in rock physics models. The fluid substitution models of Batzle and Wang (1992) were excellent examples of this progress.
AVO is related to P-wave reflectivity, S-wave reflectivity, and rock density. Therefore our estimation of elastic properties have improved through AVO and through the use of multicomponent recording for the more complete evaluation of the wavefield (i.e., shear waves as well as P-waves). Stewart et al. (2003) showed several examples of multicomponent applications.
Some of the 1980 problems are still with us; one of these is Backus' sixth challenge:
This problem could be examined from two perspectives. First, there is the detection and quantification of high-frequency loss. Since 1980, major advances in seismic resolution have come from the use of vertical seismic profiles (almost always called VSPs and, like AVO, virtually unknown before the 1980s) and the in-situ measurement of high-frequency absorption of wavefields. VSPs represent our best chance of getting reliable Q (inverse-attenuation) measurements from the use of spectral ratios—a technique introduced by Spencer et al. (1982). The processing remedy for high-frequency loss has been deconvolution. VSPs have also allowed us to measure time-varying wavelets in order to deconvolve these from the VSP reflection record. (See discussions by Hardage, 1983). Recently, progress has been made in the use of time-varying deconvolution using Gabor deconvolution methods (Margrave et al., 2003).
|
Backus' seventh challenge is one that has, as he predicted, received immediate attention and new ideas rather quickly made their way into practice:
Almost as if spurred by this prediction, a new generation of "anisotropists" quickly developed a series of new earth velocity models. A very important paper in this area was Thomsen's 1986 article (in retrospect probably one of the most seminal papers published in the past 25 years) in which transversely isotropic layered media were described by a new velocity model with three parameters (
, 
, and 
) which were weighted by functions of propagation angle. A modification of Thomsen's model and its relation to NMO velocities was published by Tsvankin in 1997. With this new way of describing velocities, it has become apparent that anisotropic depth migration should include this velocity information. The research of Vestrum et al. (1999) showed that anisotropic prestack depth migration was essential in order to avoid mispositioning well targets below dipping shales.
In addition to these improvements in prestack depth migration, there has been progress in improving the assumptions of deconvolution. We have already mentioned time-variant deconvolution, but Backus addressed another deconvolution assumption in his eighth challenge:
Although the assumption of impedance "whiteness" is still found in many deconvolution algorithms when we substitute the trace's autocorrelation for the wavelet's autocorrelation, there are many geologic situations where this assumption is invalid. Some researchers, including Ulrych (1999), have designed tests for the validity of this assumption and have suggested methods for dealing with this problem.
There are some problems that will always be with us due to the inherent ambiguity of our data, and Backus reminded us that we need to be cognizant of this fact with his ninth challenge:
An awareness of this problem and possible means of quantifying it have been constant concerns throughout the last 25 years. The topic could be labeled "inversion with a grain of salt," which was indeed the title of a 1985 paper by Lines and Treitel that produced ranges of models whose responses were consistent with given data sets. Application of sensitivity analysis led to advancements in Bayesian inference by Scales and Snieder (1997) and Ulrych (2001). In Bayesian analysis, probabilities are assigned to inversion solutions and this leads to the question (raised by Scales and Snieder with apologies to Shakespeare): "To Bayes or not to Bayes." Although geophysical inverters may choose to ignore formal Bayesian analysis of their solutions, the inherent ambiguity of geophysical inverse problems should always cause one to question the "reasonableness" of their inversion solutions, in physical or statistical terms.
The tenth and final Backus challenge is one that hints at the need for integration of geology and geophysics:
Here Backus is essentially defining the science (or art) of geophysical interpretation. There has always been a need to integrate geology and geophysics in our description of the earth but the interdisciplinary teams that are now common are a recent development. It is somewhat surprising to recall, in 2005, that at about the time of Backus' article, seismic stratigraphy was just beginning to develop (well summarized by Sheriff in 1980). Advances in interpretation also developed through the use of the coherency cube, introduced in a TLE article by Bahorich and Farmer in 1995, for fault detection. Marfurt et al. (1998) showed several examples of fault mapping using different coherency measures. Another advance was the introduction of elastic impedance by Connolly (1999) which provided a quantitative method for using seismic data to distinguish lithologic from pore-fluid effects.
Time-lapse seismology, another term that didn't exist when Backus wrote in 1980, made major strides as a tool for reservoir management around the turn of the millennium and this is one of the advances in "production" geophysics that has led to a "beyond Backus" challenge: To improve in this area, our integration efforts must go one step further and incorporate reservoir engineering data.
This review of past events in light of the Backus challenges leads to a prediction that reservoir geophysics will be the major direction in the next 25 years. This is stated for at least three reasons:
My personal prediction regarding this conclusion is that, in the immediate future, our efforts will concentrate on developing "joint inversion" of engineering, geologic, and geophysical data. This reservoir inversion process is illustrated by Figure 2 from Zou et al. (2003). In the past, geophysicists have spent their efforts in searching for geologic models that match geophysical data—basically between the seismic data analysis and modeling bubbles in Figure 2. In reservoir characterization, we will seek models that match geophysical data and that also match reservoir production history. This will be particularly important in production of the vast amounts of expensive heavy oil reserves in Canada and Venezuela that are still largely untapped. The relatively new field of time-lapse seismology is already playing a significant role in enhanced oil recovery. However, there is a great need for synergy and multidisciplinary interaction between geoscientists and engineers. Further progress in our science is likely to be based on "coordinated advances" involving geoscience and petroleum engineering in the development of oil fields.
|
 |
Suggested reading |
|---|
 TopAbstract Suggested reading |
|---|
|
Footnotes |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| JOURNAL HOME | HELP | CONTACT PUBLISHER | SUBSCRIBE | ARCHIVE | SEARCH | TABLE OF CONTENTS |
