Accounting for attenuation in the downward generator adaptive subtraction domain

Scott Keating, Andrew Iverson, Kristopher A. Innanen

Inverse-scattering based internal multiple prediction strategies suffer from inaccuracies in the amplitudes of predicted multiples. A notable source of these errors is the mistreatment of transmission losses by the conventional prediction algorithm. Interface-related transmission losses have been investigated previously, but artifacts from wave attenuation impair the predictions in much the same way, and potentially to a larger extent. Compensating for these errors with a conventional adaptive subtraction is problematic: the attenuation-related errors do not vary consistently as a function of the prediction dimensions. Here, we illustrate that attenuation errors are highly predictable in downward-generator space, and that adaptive subtraction in the downward-generator-prediction space is capable of compensating for attenuation errors.