Model-Based Subtraction Of Multiple Events

Attenuation (if not complete elimination) of multiple reflections is one of the important tasks to be performed when<br>processing seismic data. Interpretation of the data, or of extracted attributes such as AVO, requires the amplitude of the<br>signal to be preserved by the attenuation process. I present in this study a method that achieves this aim. The method<br>requires a model of the multiples present in the dataset. The algorithm is based on the spatial predictability of the events<br>that form the signal and the multiple noise. Under this predictability assumption interferences between primaries and<br>multiples can be resolved, irrespective of the possible spatial aliasing of some events in the dataset.