Improving Adaptive Subtraction of SRME Using Local Orthogonalization Filter

Surface-related multiple elimination (SRME) has been known as the most successful multiple elimination technique in the current seismic data processing industry. However, the early SRME may be influenced by the performance of the matching filter used in the adaptive subtraction process. To overcome this limitation, inversion-based SRME techniques such as estimation of primaries by sparse inversion (EPSI) and closed-loop SRME are developed. Inversion-based techniques can improve SRME result effectively but, they require complex algorithms and large computational time. In this paper, we introduce a simple method that can improve the adaptive subtraction process of SRME. To achieve this, we add a subsidiary filtering process based on local orthogonalization weight (LOW) to the conventional SRME procedures. Although the LOW filter was developed to separate random noise from seismic data, we apply it to the removal of remaining multiple energy after SRME. We verify the proposed algorithm through the reverse-time migration (RTM) using demultipled reflection data. Numerical examples show that the proposed algorithm can help reduce multiple-related artifacts in migration images.