If we do three non-linear iterations of tomography using fitting goals (8) each time, we get Figure 14 as our velocity estimate. The velocity estimate does a good job recovering the anticline shape. However, it doesn't do a good job recovering the low velocity layer. Migrating with Figure 14 we get Figure 15. Overlaid on top of the migrated image are the correct reflector positions. Overall we did a good job positioning the reflectors. In addition, we can see that we have little residual moveout in our CRP gathers (Figure 15) but they are relatively flat.
Figure 14 Velocity model after three non-linear iterations of non-stationary filter regularization.