Autoconvolution is a technique which can predict all surface-related multiples given a 1-D earth and zero-offset data. Data prepared to resemble zero-offset data from a 1-D earth are suitable input. In practice this means that the multiples need to be periodic, the two-way time of the water-bottom primary needs to equal this period, and the record above this event needs to be clear of any other events (such as direct arrivals). Furthermore the wavelet shape needs to be constant down each trace.
No takes yet. Share an insight, caveat, or question.
Lamont et al. (1999) studied this question.