Research On Noise Suppression Of Microseismic Data By Shearlet Transform Based On Spectral Multimanifold Clustering

Microseismic monitoring technology has been a kind of advanced monitoring technology of fracture created hydraulic fracturing,which has become an important method to improve the structure of low permeability reservoirs.Downhole monitoring and surface monitoring are two different ways to realize the microseismic monitoring.Among them,the surface microseismic monitoring technology has more advantages,such as low cost,strong practicability,no monitoring well and large monitoring range.So,it has a wider application prospect.However,the low signal-to-noise ratio(SNR)and weak energy of microseismic events greatly hinder the realization of the processing and interpretation of the microseismic data.Hence,noise suppression is a critical step in surface microseismic monitoring to enhance SNR and resolution,which has great significant for identifying and locating microseismic events.Shearlet transform is a new multiscale geometric analysis tool,and it is also a near optimal sparse representation of multidimensional data.The Shearlet transform is associated to multiresolution analysis and this leads to a unified treatment of both the continuous and discrete world.So,we adopt Shearlet transform to suppress microseismic noise.The microseismic data are decomposed into different scale layers by shearlet transform with the increase of the scale,the Shearlet coefficients gradually changes from the Coarse scale layer to the Fine scale layer.The coarser corresponds to the low frequency profile information of the original data,while the high frequency detail information is mainly concentrated in the Finer,and the random noise is mainly concentrated in the Finer.However,when the SNR is rather low,the coefficients related with random noise are so close to the coefficients associated with signals in the shearlet domain that the threshold method can not completely remove the random noise.In order to better separate the coefficients,we propose Shearlet transform based on spectral multimanifold clustering.The propose method takes full account of the microseismic events manifolds in the Shearlet transform domain,and use spectral multimanifold clustering to detect or group the microseismic events coefficients,so as to achieve the purpose of separating the coefficients.The microseismic events and noise have different low dimensional manifold structural information respectively.The effective microseismic events have the obvious manifold structural information,but the random noise has no obvious manifold structural information,and after the Shearlet transform,their manifold structural characteristics will not change.Based on this difference,the spectral multimanifold clustering can easily detect the coefficients related with effective events,thus avoiding the defects of the threshold scheme.In order to verify the effectiveness and feasibility of denoising method proposed in the paper.The proposed algorithm is applied to the Synthetic and real microseismic data processing.Synthetic and real microseismic data example demonstrate that our proposed method can more effectively eliminate noise and preserve microseismic events under low SNR.