Conclusions

This paper shows that prestack Stolt residual migration, in spite of its constant velocity origin, can be successfully applied to depth images obtained with an arbitrary velocity. Several examples, both on synthetic and on real data, prove the effectiveness of the method. The key to improving the seismic images lies in correctly picking the variable surface of the parameter $\gamma$, rather than in selecting only one image at a particular value.

The method also shows that the blurred regions of the seismic image of the North Sea dataset are not so much an imaging problem as a velocity problem: a better velocity map could enable a good algorithm to image the overhang of the salt, at least partially. Prestack Stolt residual migration is able to improve the migrated image, at a fraction of the cost of the original migration.