CONCLUSIONS

Although the residual migration operators come from kinematic considerations, they work despite troubles near their triplications. It is possible to build the operators more closely based on the wave equation, but the resulting algorithm would resemble a phase shift migration and would be costly to apply for a broad range of residual slownesses which is needed for velocity analysis. My application for residual NMO+DMO is velocity analysis, and it is ideally suited to this task. Residual NMO+DMO can find the best residual prestack time migration slowness for fixed events which is different from finding the best migration slowness for a fixed location in the Earth. It is not confused by events moving around the image as migration slowness changes. When a high-quality final wave-equation-migration output is needed, it is probably easier to migrate the data from scratch with a new interval velocity model or use wave-theoretic residual NMO+DMO.