Reservoir characterization via time-lapse prestack seismic inversion

Diierencing of time-dependent seismic monitor surveys requires that datasets are collected and processed consistently. Inconsistencies can result in data diierence noise great enough to obscure any reservoir changes. Inversion applied to each dataset may account for some of these inconsistencies. As an example, if the shape, temporal location, or radiation pattern of the seismic energy source is not known exactly, data diierencing would produce only noise. However, we have shown in previous work that both the source and reeectivity can be accurately recovered via inversion. In this paper we describe a synthetic experiment in which the reservoir contains only oil and water initially. A single production well induces a pore pressure drop and related saturation changes including the formation of a gas cap. This work provides a baseline (best-case) scenario for testing the eeectiveness of applying inversion to seismic time-lapse data | we assume that the source is known. Prestack inversion recovers both the subtle increase in S-wave impedance and the larger decrease in the P-wave impedance at the well location in the reservoir.