We must insure that our definition of the image perturbation is mainly a product of the slowness model perturbation. Much care needs to be taken to eliminate all acquisition differences between the repeat surveys and all processing differences of the different datasets. An ideal case consists of fixed acquisition (permanent water-bottom receivers, for example) and identical seismic processing.
Correct handling of amplitude data in migration is as important as in any method addressing reservoir-related properties. However, in this method we are mainly concerned with the differences between repeat images and not as much interested in their absolute magnitude. Therefore, this method is likely to be robust with respect to the accuracy of the more or less accurate migration amplitudes. This particular subject, however, requires careful further analysis.