Denoising Algorithm For Microseismic Exploration Based On Joint Bivariate Shrinkage In Shearlet Transform

Microseismic monitoring technology use microseismic wave inversion to locate rock fractures,describe the properties of fractures,and monitor the fluid movement and lithology changes.It has very important value in the unconventional oil exploration.With the increasing depth of microseismic exploration and the complication of the environment,noise interference in microseismic data has become increasingly strong.The existence of noise greatly interferes the following research such as the pick of the first phase and microseismic source location.Therefore,attenuating microseismic random noise and effectively recovering microseismic signal to improve the signal-to-noise ratio(SNR)of microseismic data are important components of microseismic signal processing.Microseismic data has the characteristics of weak signal energy,high frequency,and strong random noise,thus seismic denoising methods are difficult to recover microseismic signals effectively under low SNR.In this situation,this paper adopts Shearlet transform,which has multi-scale and multi-directional characteristics,to solve the problem of denoising surface and downhole microseismic data.Based on Shearlet transform theory,the paper systematically analyzes the structural features of microseismic signal at different scales and different directions,and verifies the ability of the Shearlet transform to characterize the structure of microseismic signal.The researches show that the Shearlet transform can realize the sparse representation of microseismic data,surface microseismic signal are sparsely concentrated in the low and middle frequency bands,but downhole microseismic data have a lot of effective signals in the high frequency bands.In addition,the Shearlet coefficients of microseismic signal in adjacent scales have similar structure and correlation.The main work and innovative of this paper is as follows:A microseismic denoising method based on adaptive bivariate shrinkage in Shearlet domain is proposed.In order to avoid the loss of high-frequency microseismic components,the Shearlet denoising method that discards high-frequency coefficients can't be used.In low SNR case,the noisy coefficients in Shearlet domain are strong,it seriously affects the identification of signal coefficients and the low-frequency components are also affected by noise.Therefore,it is difficult to use the threshold method to suppress noise.Based on thecorrelation of Shearlet coefficients between scales,a microseismic denoising method based on adaptive bivariate shrinkage in Shearlet domain is proposed.The algorithm first decomposes the microseismic data by Shearlet transform,and calculates the correlation between the parent and child coefficients(the coefficient in the same direction and the adjacent scale)as a bivariate shrinkage function(BSF),and obtains the BSFs with adaptive estimation of the noise variance in each scale.Then the microseismic signal corresponding Shearlet coefficients are estimated by the BSFs,and the effective signal is reconstructed from the noisy microseismic data.Simulation denoised results show that Shearlet transform based adaptive bivariate shrinkage denoising method can effectively recover the surface microseismic signal.To solve the problem of high-frequency signal distortion and low-frequency coefficient exist noise in low SNR,the Shearlet denoising method based on joint bivariate shrinkage is proposed.Because of the very low SNR of high-frequency Shearlet coefficients,the finest-scale BSFs cannot accurately characterize microseismic signal.Further researches show that the structure of BSFs between adjacent scales are similar,and the coarser-scale BSFs can more accurately describe the structure information of microseismic signal.Therefore,we use the structural similarity of the BSFs between scales to construct the joint bivariate shrinkage by weighting the BSFs of the adjacent scale,and improve the ability to recover the high frequency component.In addition,the paper combines the BSFs in adjacent scale of low-frequency coefficient to estimate a low-frequency joint BSF,which improves the noise reduction performance.To verify the effectiveness of the proposed method,it was applied to denoise the surface and downhole microseismic data.The denoised results show that the joint bivariate shrinking Shearlet transform denoising method can effectively suppress random noise,enhance high frequency signal,and improve the conherence of the events.Compared with the wavelet transform based denoising method and the bivariate shrinking based Shearlet denoising method,the joint bivariate shrinking Shearlet transform has best performance in noise attenuation signal preservation,and improvement of the SNR of microseismic data.