When is anti-aliasing needed in Kirchhoff migration?

Migration with low velocity

In Figure 5 the data have been deliberately migrated with an unrealistic low velocity of 1000 m/s in order to confirm that operator aliasing will occur when the criteria of equation 3 are met. In these examples $f_{op} < f_s$ so that the operator is aliased. The diffraction from the salt flank is much better imaged in the anti-aliased migration (Figure 5a) than in the aliased migration (Figure 5b). All of the same operator aliasing artifacts that were visible at early times in Figure 2 are still present, but now the events at late times also suffer. The right hand diffraction in Figure 5b is weaker and less continuous than the anti-aliased diffraction in Figure 5a. Some of the diffractions on the left side of Figure 5b are completely lost in the aliased migration.

Miglow
Figure 5. Kirchhoff migration of Gulf of Mexico data with an unrealistically low velocity: (a) with anti-aliasing, and (b) without anti-aliasing.

When is anti-aliasing needed in Kirchhoff migration?