Simultaneous Sparse Deconvolution

Increasing seismic resolution has been long pursued by the geophysical community, serving, among other applications, for more detailed interpretation of seismic clinoforms, better seismic inversion and quantitative interpretation. Sparse deconvolution methods play central role in this pursuit. However, although sparse methods have performed well for single seismic stack or acoustic inversion tasks, their application for multi-stack seismic volumes and consequent use in elastic inversion is still a challenging and ongoing research topic. The challenge is to obtain reflectivity volumes with high correlation between them. In this study, we present a new method to perform simultaneous sparse deconvolution (SSD) in a group of seismic volumes associated with different reflection angles. The proposed algorithm enforces co-localization of the spikes on the estimated reflectivity traces and additionally allows user control of the sparsity via hyperparameters. The method is validated in both synthetic and real datasets proving its co-localization capability and resulting in higher correlations between the reflectivity volumes, when compared to independent sparse deconvolution (ISD) of the seismic stacks. The resulting reflectivity volumes are, henceforth, better suited for downstream tasks such as high resolution amplitude versus angle (AVA) analysis, or input for high resolution elastic inversions.