Joint wave-equation inversion of time-lapse seismic data |
Since different geometries were used for all surveys in the numerical test, deterioration of the time-lapse amplitudes is due to a combined effect of geometry and complex overburden. The illumination-ratio maps (Figure 5) show the variability of illumination between surveys. Disparities in point-spread functions (Figure 8) suggest that the a diagonal approximation to the Hessian is insufficient to remove the unwanted artifacts. Although separately inverted time-lapse images in Figure 11 show some improvement in resolution over migration results (Figure 10), the images are dominated by the large amplitude residual artifacts. Time-lapse images obtained from joint-inversion using the RJID (Figure 12), and RJMI (Figure 13) formulations are less noisy than those from the migration (Figure 10) and separate inversion (Figure 11) and are comparable to the true reflectivity change in Figure 9. Since one-way operators were used in this study, the Hessian contains no information regarding secondary events such as multiples. Residual effects of internal multiples due to the salt persist in the inversion results but are significantly suppressed in the jointly inverted images. In many cases of interest, multiple energy will be sufficiently attenuated or may not fall within the target region, and thus have little impact on the inversion.
Joint wave-equation inversion of time-lapse seismic data |