Towards the prediction of internal multiples from thin layering by Marchenko

The presence of internal multiples in our data due to strong reflectors could inhibit or complicate the correct (quantitative) interpretation in particular when, after imaging, they cross reflectors or areas of interest. Commonly used internal multiple prediction and subtraction workflows deal with these kind of situations reasonably well. Unfortunately, situations exist whereby internal multiples are not ‘simply’ crossing events. For instance, a package of thin layers can produce a ‘cloud’ of internal multiples, which completely obscures deeper primaries as observed in a dataset acquired at or near the Arabian Peninsula. So far, all available workflows fail to handle this particular case. Note that when just the subtraction is not feasible, the separate imaging of the predicted internal multiples could still be of benefit to the interpretation process. With this case in mind, a collaboration project was initiated to evaluate Marchenko based internal multiple prediction for thin layering using a series of datasets from synthetic models in, mostly, blind tests. This abstract presents the first observations from the initial results.