Data-driven multiple suppression for laterally varying overburden with thin beds

Marchenko multiple elimination methods retrieve wavefields which can be used to remove all orders of overburden-borne multiples in a data-driven way. Although powerful, the method only accurately handles media with thin beds if the underlying Marchenko equations are constrained by the information that is typically unavailable a priori. The so-called augmented Marchenko scheme aims at addressing this problem, however, it requires a minimum-phase reconstruction step, which is only well-defined for 1D signals. As a result, the method can only be applied to media with laterally invariant (1.5-D) overburdens. Applications to 2-D are restricted by the fact that a generally-applicable minimum-phase reconstruction algorithm does not exist for arbitrarily-complex 2- and 3-D overburdens. Here, we show that accurate solutions to the augmented Marchenko-type equations can still be found in 2D by including additional processing steps in the 1.5-D approach. Furthermore, the approach presented here opens the possibility of calculating band-limited multidimensional minimum phase operators.