Research On Desert Seismic Signal Denoising Algorithm Based On VMD-GreBsmo

In the process of global modernization,oil and natural gas are extremely precious non-renewable resources.The continuous depletion of conventional oil and gas reserves is pushing the development of resource exploration to unconventional oil and gas fields.There are abundant oil and gas resources in the desert region of Tarim Basin,so it has great value to ascertain its geological structure and resource storage status.Seismic exploration is an effective exploration method for solving the problem of oil and gas exploration,which can provide a set of accurate stratigraphic structure data.However,because of the complex and special environment,the seismic exploration data collected in the desert area are seriously disturbed by the random noise,which can not meet the “Three High”requirements of high signal-to-noise ratio,high resolution and high fidelity.Therefore,it is very important to find an effective method to suppression desert seismic random noise for oil and gas resource exploration.With the development of noise suppression techniques for seismic exploration data,many effective and feasible methods have been proposed and put into practice,such as f-x deconvolution,wavelet transform and band-pass filtering.These classical noise suppression algorithms have achieved some results in dealing with the random noise of the middle and shallow seismic exploration and improving the signal-to-noise ratio of seismic exploration records,it has also been well applied in oil and gas resources exploration engineering.However,in the face of desert seismic exploration data which have serious spectrum overlap between effective signal and low frequency random noise,the traditional seismic data denoising methods mentioned above have some limitations,such as incomplete noise suppression and difficult to identify effective signal,which can not meet the needs of high-precision exploration.In recent years,time-frequency analysis and low-rank matrix factorization have been developed rapidly.According to the sparse and low-rank characteristics of desert seismic signal in the time-frequency decomposition domain are obvious,and the corresponding rank of the effective signal and low-frequency noise in the decomposition domain are different,this paper presents a desert seismic noise suppression method based on low-rank matrix factorization in the time-frequency domain.Firstly,the desert seismic data are decomposed by Variational Mode Decomposition(VMD)into several modes.All the decomposed modalities are rearranged into a new signal matrix without discarding any modes,and then the matrix is subjected to low-rank factorization by Greedy Semi-Soft Go Decomposition(GreBsmo)to obtain the efficient low-rank component.Finally,superimpose all the modalities of each channel signal in the low-rank matrix and substract from the original seismic data to achieve denoising.This method avoids choosing the modes of VMD and the iterative convergence of the GreBsmo algorithm is used to select the rank adaptively in order to simplify the rank number selection process.In the processing of desert seismic exploration data,the effectiveness and feasibility of the desert seismic signal denoising algorithm based on VMD-GreBsmo proposed in this paper are verified by the simulation experiments.Both the overall denoising effect and the single channel comparison results in the time domain show that the algorithm of this paper is more effective than traditional denoising methods in suppressing the low-frequency noise from desert seismic.Based on the quantitative analysis of data,the method in this paper can improve the signal-tonoise ratio by about 13 d B in the simulation experiments.Applying VMD-GreBsmo to process real desert seismic data,the algorithm can effectively suppress the noise and recover the same phase axis clearly and continuously compared with the traditional methods.