Gulf of Mexico example

Figure  shows panels from two very different marine datasets shot over a salt diapir in the Gulf of Mexico. The left-side of Figure  is very early 3-D, shot by Chevron in 1979. The right-side is contractor (spec) data from 1991/1992.

 

before
Figure 1 Before cross-equalization: panels from the 1979 survey (left), panels from the 1991 survey (right).

Classic Gulf of Mexico rock physics means the reservoir would score well in a technical risk assessment (). However, questions of seismic repeatability cast doubt over the success of a 4-D project.

Improvements in 3-D technology over the 12 years between surveys have resulted in large differences in quality between them. For example, as well as having very different geometries, it is apparent that they have been processed very differently too. Most noticeably the 1991 dataset was migrated with a turning-ray alogorithm, to image the over-hanging salt-flank. The 1979 dataset contains a much lower dip range.

Ideally we would like to go back to the pre-stack seismic data, and perform some kind of inversion to estimate a common Earth model. If this is not possible, we should at least reprocess one of the datasets to be consistent with the other. However, for these datasets (and many others) finding the pre-stack tapes is impossible, or at least prohibitively expensive. Even if the pre-stack tapes were available, the cost of reprocessing them will be high. As a result non-production related differences between these surveys have to be removed by cross-equalization in the post-stack domain.

In this paper, we demonstrate a powerful post-stack cross-equalization flow with the aim of determining whether any useful information can be obtained from these off-the-shelf processed datacubes. The processing flow is robust, and addresses non-stationarity through the use of spatially-variable operators.