Vlad and Biondi (2002) and Vlad (2002) show that Wave-Equation Migration Velocity Analysis Biondi and Sava (1999) is a highly suitable method for finding the FEAVA-causing velocity lenses. Vlad et al. (2003) show that for a simple synthetic dataset under optimal conditions, this method generates velocity models which eliminate FEAVA through migration with an operator of the same accuracy as the one used for modeling. Vlad (2004) also demonstrates a FEAVA detector on the same synthetic dataset.
Simple controlled experiments, however, are only the first step in the testing of scientific hypotheses. In this paper I advance by testing the previously proposed methods on a significantly more complex synthetic dataset. I examine the behavior of FEAVA in the data domain and in the image domain, in a setting with and without heavy internal multiples contamination, in an image migrated just with the background velocity trend and with the correct velocity model (one and eight reference velocities), with high offset sampling and with sparse offset sampling. In particular, the purposes of the study are: 1. Determining the robustness of FEAVA detection with respect to multiples and imaging approximations; 2. Probing the degree of amplitude accuracy needed by a migration operator in order to eliminate FEAVA from the image; 3. Finding to what extent internal multiples can interfere with FEAVA removal through migration.