Adjoint migration

To obtain the estimates of the primaries and multiples in data space we need to apply the inverse of the migration operator to the muted SODCIGs of the primaries and the multiples. Here I used the adjoint of the migration in lieu of the inverse. To assess just how good the adjoint migration is in recovering the kinematics of the data, I first applied the adjoint migration to the unmuted migrated SODCIGs (with the correct velocity) and I show the comparison between a CMP gather of the original dataset and the migration-adjoint migration result in Figure . Clearly, the kinematics of the reflections have been recovered fairly well, except at the large offsets of the water-bottom primary for which the subsurface offset sampling in the SODCIGs was a little coarse given its steep moveout as seen in Figure  (below -400 m). In the next subsection I show that adaptive subtraction recovers these amplitudes very well.

 

cmp_inv
Figure 6 Estimated multiples (a) and estimated primaries (b). Some multiples leaked into the estimate of the primaries and some primaries leaked into the estimate of the multiples.

Figure  shows the result of applying adjoint migration to the muted SODCIGs. The left panel corresponds to the estimated multiples whereas the right panel corresponds to the estimated primaries. Obviously, some primary energy remains in the estimated multiples and some energy from the water-bottom primary remain in the estimated primaries.