INTRODUCTION In conventional utilization of the seismic reflection method, it has traditionally been assumed that seismic signals can be viewed as a band-limited normal incidence reflection coefficient series with appropriate traveltime and amplitude variation due to propagation through an overburden. Ostrander (1982) demonstrated that gas sand reflection coefficients vary in an anomalous fashion with increasing offset and showed how to utilize this anomalous behavior as a direct hydrocarbon indicator on real data. This work popularized the methodology which has come to be known as amplitude variation with offset analysis (AVO). In addition to being a “seismic lithology” tool, AVO provides an improved model of the reflection seismogram which allows us to better estimate both normal incidence reflection coefficients and “background” velocities. Exploration geophysics is, to a large extent, a science of anomalies. It is probably safe to assume that most hydrocarbons found in the past fifty years have been associated with some kind of geophysical anomaly. Explorationists routinely utilize deviations from expected gravity, seismic traveltime, and seismic amplitude, without recovering uniquely determined absolute density, depth, or reflectivity. Consequently, significant risk is associated with drilling a well, even when all the technology and analysis available have been thrown at the problem. Experience has shown that geophysical anomalies can be used to reduce risk, and consequently, to identify new prospects. Thus, one would expect the exploration geophysicist, who is concerned primarily with finding hydrocarbons, to have somewhat different expectations from AVO analysis, than say a research physicist, who may be primarily concerned with getting the right answer. Taking the latter approach, if one makes very simplistic assumptions, methods can be developed which work well on synthetic data. However, the more one appreciates the complexities of the real world with real geology, the more it becomes apparent that AVO analysis doesn't work. The only fault with this conclusion is that explorationists are successfully using AVO anomalies to find hydrocarbons throughout the world. The explorationist does not require answers that are correct in absolute terms. The presence of a deviation from some background trend may be sufficient; the magnitude of the deviation in absolute units may not even be required. An appropriate analogy is the spontaneous potential (SP) log. If I am told that an SP log reading is ?30 millivolts at some depth, I cannot interpret the rock properties, even with the most sophisticated modeling and inversion schemes. Of course, the SP log is an anomaly log without a meaningful absolute scale. When I see that ?30 millivolts is a significant deviation from the background trend produced by surrounding shales, I realize that the SP log is very valuable indeed in the context of my particular problem-that of identifying sands.

When seismic waves hit an interface between two layers in the subsurface, some of the energy may be reflected. The amplitude of the reflection depends on the reflection coefficient. The reflection coefficient is the ratio of the reflected amplitude to the incident amplitude.