Velocity analysis based on pre-stack imaging

Velocity analysis is usually based on common midpoint gathers from a 2D seismic line. For horizontally layered media, the velocities obtained are rms velocities. If the reflector is not horizontal but parallel to other reflectors, the velocities obtained are larger than the rms velocities by a factor equal to the cosine of the dip angle. Dip-Moveout (DMO) refers to a transformation of the raw data which corrects for the effect of dip. After DM0, velocity analysis yields rms velocities even when the reflector is not horizontal. While the problem of dip is handled by DM0, several other problems remain: for example, when the data are stacked after DM0, they still need to be migrated and appropriate velocities for migration may not be apparent. Pre-Stack Imaging (PSI) refers to a transformation applied after DMO which migrates the data before stack. After DMO and PSI, velocity analysis yields rms velocities and stacking yields migrated traces. This thesis presents a computer program which does velocity analysis after DMO and PSI and illustrates its operation with data from a physical model experiment. Both DMO and PSI are transformations which do not require any velocity information. It can be shown that DMO and PSI achieve migration before stack without using velocity if the medium has a constant velocity of propagation. When the velocity varies laterally by a large amount, the result may still be useful. The stacking velocity may become physically unreasonable but it still may be possible to produce a "stacked" section. Since this section approximates the correctly migrated result, it does not need any further processing. By delaying velocity analysis until after all the major processing, DMO plus PSI substantially reduce the chance of making errors in interpretation.