An Adaptive Damped Rank-Reduction Method for 3D Seismic Data Denoising

Random noise suppression is a significant step and plays an important role in seismic data processing. Researchers have developed many methods to attenuate random noise in the past decades and rank-reduction method is one of them. In recent years, with many benefits such as stable denoising performance, rank-reduction method is becoming prevalent in seismic denoising algorithms. In this paper, we put forward an adaptive damped rank-reduction method for seismic denoising. This method is an enhanced version of multichannel singular spectrum analysis (MSSA) algorithm. We first introduce a damping factor to suppress random noise, which operates the truncated singular value decomposition (TSVD) on Hankel matrices. Then, we derive a novel adaptive method of rank selection by finding the maximum of difference of energy intensities between two consecutive singular values. We verify the efficiency and competitiveness of the new method via applying it to a simple synthetic example with linear events and a complex synthetic example containing highly curved events. The proposed method is compared with other existing seismic denoising algorithms. By comparing the results on two synthetic examples, we come to the conclusion that our method acquires significant effects on attenuating random noise and recovering the useful signal structure.