Many geophysicists argue that high technology two-way methods are much more sensitive to errors in the velocity model than any of the one-way approaches. Both mathematical theory and experience contradict this assumption.
Any impulse analysis shows multi-arrival energy is very prevalent below complex geometries defined by anomalously high velocities. Such structures include salt domes and granitic overthrusts but can also be carbonate based.
Since rays, as computed by most raytraces, represent a high frequency approximation, virtually any ray-based migration may be excessively sensitive to sharp velocity discontinuities. This, together with the single arrival assumption, is certainly a major reason that single arrival Kirchhoff methods are extremely poor at imaging below salt.
Most one-way implementations do a much better job of handling multiple arrivals, and so it is not surprising that images produced with these methods are usually better than single arrival Kirchhoff applications. But, because one-way methods have a built-in angle limit, they also have a built in instability.
Two-way methods, or at least those that are above the one-way line in Figure 1, have almost no inherent restrictions on velocity variation, angle, or amplitude response. Depending on the implementation, they can produce significant amounts of grid dispersion resulting in something akin to a grid-based anisotropy, but, if that is handled properly, we can be assured that the two-way method will have the least sensitivity to velocity errors.
