Introduction

The difficulties of imaging below salt edges are compounded by the difficulty of generating an accurate velocity model in these areas. The majority of imaging techniques require an accurate velocity model in order to produce a well focused result. The artifacts seen in a migrated image caused by errors in the velocity model are well known Claerbout (1985). In fact, migration can be used as a tool for developing a velocity model Biondi and Sava (1999). However, in areas such as those around salt, where it is most difficult to obtain a good velocity model, poor illumination makes it impossible for migration alone to provide a satisfactory image.

In complex areas, imaging can be improved by using a migration operator in an iterative inversion scheme. Unfortunately, if the velocity model is inaccurate the artifacts that are seen in the migration result will affect the inversion result as well. Even more ominously, some iterative inversion techniques make the assumption of correct velocity a critical part of their theory. In particular, imaging using the regularized inversion described by Prucha et al. (2000) and Kuehl and Sacchi (2001) assumes that the correct velocity is being used to justify the choice of regularization operator. The regularization operators they use assume that there is no moveout along the offset ray parameter axis. In this paper, I will examine the sensitivity of this assumption for a variation of Prucha et al. (2000)'s implementation.

I will first explain the manner in which I carry out regularized inversion and the regularization operator used. Then I will perform migration and regularized inversion on a synthetic dataset using the correct velocity model and two velocity models that have been perturbed in different ways. I will compare these results to make a statement on the validity of the zero-moveout assumption.