Surface-multiple attenuation through sparse inversion: results for complex synthetics and real data

Abstract

We present new results of the recently introduced approach to surface-multiple attenuation through sparse inversion. This method aims to attenuate surface multiples of all periodicities without the need for an adaptive subtraction or a near offset extrapolation. The aim of our paper is twofold. First, we subject the sparse inversion method to data beyond previously published proof-of-principle models and present results on highly complex 2D synthetics and real data, demonstrating that sparse inversion may, in principle, serve as a powerful tool to attenuate short and long period multiples in complex settings. We illustrate the competitive nature of the sparse inversion predictions by benchmarking the method’s results to standard processing flows. In addition, we demonstrate the correspondence of the estimated primaries to the underlying impedance models. Next, we show results of a 2D sparse inversion approach to real 3D deep-marine data and demonstrate, through a further synthetic study, that the presence of cross-line dip leads to a time shift of the predicted multiples, resulting in reduced multiple attenuation. Our results confirm that future developments of this technique will require a full 3D approach in order for the sparse inversion method to be widely effective in practice.