Figure shows a Kirchhoff prestack time migration with the 1-D velocity field described before. Although the previous 1-D stack was not very coherent to the east, the prestack migration has done an amazing job. This seems to indicate that time migration is less sensitive to velocity error than NMO stacking. Figure is the Kirchhoff prestack time migration using the optimal 2-D Monte Carlo velocity field of Figure b. The faulted blocks are very well defined from the base right up to the surface. Figure is an overlay of the smoothed Monte Carlo interval velocity field on the prestack time migration of Figure . Above 3 seconds there is a fairly good low frequency correlation between migrated structure and interval velocity. However, below 3 seconds the velocities are rather dubious in reliability. There is definite bias due to structural dip, as seen at 12 km, 3.5 seconds, and at the easternmost edge of the section, kilometer 17. It would be interesting to compute migration velocity spectra for this data set, and repeat the process of making the automatic Monte Carlo velocity picks and remigrating the data.