FINAL SPECULATION

This interesting and challenging data set deserves much further study. I believe the supplemental code that is required is a prestack finite difference upward and downward continuation. Shots and geophones need to be upward and downward continued through the top 600ms with various experimental lateral velocity variations. This might seem like a ``solved problem'' but I see many difficulties. It must be done rapidly and without excessive artifacts. I think this means it needs to be done after NMO, again something which seems theoretically realistic, but which we have little experience since early days of SEP.

Given that we can master the technology of rapid upward and downward continuation, suppose we estimate static shifts by traditional delay of each trace from the CDP stack. Then we can attribute half the delay to the shot path and half to the geophone path. To ascertain a better depth for the anomaly than the zero depth we can make a Kjartansson-style amplitude tomograph. Now we could downward continue shots only treating the shot static as a velocity anomaly to be distributed over the range indicated by the amplitude tomograph. From this shot-downward continued data, we could redo the amplitude tomograph, presumably seeing a revised and improved model for geophones. Hopefully an orderly bootstrapping procedure can be found.