Velocity can be estimated in isolated parts of the model. By resynthesizing the data at a lower datum, I remove the complicating effects of the overlying velocity structure. For instance, the imaging problem in the lower part of the section may be simpler since compaction tends to result in velocities that are generally better behaved at depth. This means that a simpler imaging algorithm can be used at depth once the complicated overburden has been datumed through. Care should be taken to make sure that the velocity of the overlying layers is accurately determined because any error in the datuming velocity will affect the resynthesized gathers.
Even if the overlying velocity structure is not very complicated, it can be advantageous to strip it off, and perform velocity estimation and imaging from a datum that is closer to the target or the more complicated structure of interest.
A related issue is imaging and velocity estimation under anisotropic media. By means of anisotropic traveltimes, it is possible to downward continue through anisotropic layers and remove their deleterious effects. This could be very useful because in some parts of the world, there are thick anisotropic layers overlying isotropic but complex structures. The stripping off of the anisotropic layers could allow proper velocity estimation and depth imaging of the deeper portions of the data.