METHODOLOGY

Creating a travel time error from residual migration involves several steps. My method is to first perform residual Stolt migration Sava (1999b); Stolt (1996) on offset domain image gathers with a slight twist: minimize reflector movement among the various velocity ratios. I convert the different images to the angle domain Sava and Fomel (2000) and calculate the coherence of the zero moveout. I then set up a simple inversion problem to obtain a smooth estimate of $\bf \gamma$ for each model location. Finally, I convert this $\bf \gamma$ to an approximate traveltime error and back project it.