Two-dimensional variational mode decomposition for seismic record denoising
Xingli Zhang, Yan Chen, Rui‐Sheng Jia, Xinming Lu
Abstract
Abstract Seismic signal denoising is the main task of seismic data processing. This study proposes a novel method for the denoising seismic record on the basis of a two-dimensional variational mode decomposition (2D-VMD) algorithm and permutation entropy (PE). 2D-VMD is a recently introduced adaptive signal decomposition method in which $K$ and $\alpha $ are important decomposing parameters to determine the number of modes, and have a predictable effect on the nature of detected modes. We present a novel method to address the problems of selecting appropriate $K$ and $\alpha $ values and apply these values to the proposed method. First, for a 2D seismic signal, the 2D-VMD method can decompose it into $K$ modes with specific direction and vibration characteristics. Next, the PE value of each mode is calculated. Random noise components are eliminated according to the PE value. Finally, the signal components are reconstructed to acquire the denoised seismic signal. Experimental and simulation results indicate that the proposed method has remarkable denoising effect on synthetic and real seismic signals. We hope that this new method can inspire and help evaluate new ideas in this field.