In the previous paper in this report, Claerbout describes how velocity
is more accurately determined if a datum shift is determined
simultaneously. A Soviet paper shows that such an apparent datum shift can be caused by strong variations in velocity. These velocity variations, quantitatively described by the mean fourth power of velocity, can be determined from the parameters picked by Claerbout's
Overlay program. The true (vertical-ray-path) root-mean-square velocity
can also be determined from these parameters. This determination is accurate even when offsets two or three times greater than the reflector depth
are used.
The results of the Soviet paper may also have some application
to Stolt stretch.