Research On Seismic Data Processing Methods Based On Shearlet Sparse Constraint

In recent years,geophysical exploration targets are gradually focusing towards concealed,complex and unconventional reservoirs,leading to further exacerbation of data irregularity and noise interference.The regularized data is the basis of many seismic data processing methods,the reconstruction for irregular wavefield will directly affect the subsequent process and and geology interpretation.In addition,the acquired data contains random noise and coherent noise may greatly reduced the signal-to-noise ratio of seismic data.Deep area and complex objects require high resolution methods,which prompted us to put forward more flexible data acquisition and higher resolution data processing methods.The theory of compressed sensing makes it possible to obtain an ideal reconstruction below the Nyquist sampling rate under certain conditions.Sparsifying transformation,sparse sampling strategy and optimization algorithm constitute the three elements of data reconstruction based on compressed sensing theory.The new sampling modes are more flexible than the conventional regular sampling method,which not only reduces the sampling rate,but also reduces the sampling time cost and the economic cost.Shearlet transform is an anisotropic wavelet with fairly good sparse representation of seismic signals and found the main line throughout the whole text.Its sparsity,multi-directionality,and multi-scales has pushed conventional seismic data processing methods to new heights in seismic de-noising,reconstruction and deconvolution methods.In this paper,we first establish an effective seismic data reconstruction method combined the compressive sensing theory and Shearlet transformation,which has acquired higher accuracy compared with conventional seismic reconstruction methods.Based on compressive sensing theory and Shearlet transformation.On the basis of this,it was extended to the reconstruction of marine streamer seismic data.The crossline direction of offshore seismic data is usually extremely sparse,and conventional data reconstruction methods are not applicable.Therefore,using multicomponent measurements techniques,we propose a multi-component crossline wavefield reconstruction method based on sparse Shearlet constraint inversion.By combining the pressure wavefield and its crossline gradient obtained through Vy measurements,the proposed method can effectively reduce the multiplicity caused by a limited number of samples,which lays a good foundation for high precision processing of OBC\OBN and other data.Secondly,based on the sparsity and directionality of Shearlet transform,an effective noise suppression method is established to achieve the fidelity de-noising.As to random noise,coherent noise and effective signals show different distribution characteristics in the shearlet domain.We established the random noise and effective signals on the Bayesian formula,and achieved effective suppression of the random noise.For the coherent noise suppression method,we take the surface wave as example.Surface waves can be recognized at different scales and different directions in the Shearlet domain.A multi-step method for separating surface waves is used to retain the most effective information.After effectively predicting the interference wavefield,the conventional direct-subtraction method may damage the effective wave field in the cross section of the effective wave field and the interference wave field.In addition,the inaccuracy predicted of the interference wavefield can also damage the effective wave field.By combination with Bayesian probability maximization theory,a sparse domain wavefield separation method is established.This method can better control the error between the predicted signal and the actual signal,and can correct the deviation of the interference wave field in phase and amplitude to achieve the purpose of accurate wave field separation.Finally,based on the sparsity and multiscale frequency division characteristics of Shearlet transform,we use the Shearlet coefficients to represent the non-spiky reflectivity and solve the deconvolution problem in multichannel way.Moreover,compared to single-trace deconvolution methods,the multichannel method can enhance the continuity of reflection events and suppress high-frequency noise in the deconvolved data.In addition,the Cauchy constraint is introduced to improve the antinoise of the proposed algorithm.Compensation for formation absorption is another important way to improve high-frequency energy.In this paper,we first establish the stratal absorption compensation(SAC)model,and then we combine the SAC model with the shearlet transform,and establish the new compensation method.Combining the compressed sensing and the sparse representation properties of Shearlet,this paper carries out wave field reconstruction,fidelity-based de-noising,and high-resolution seismic data processing methods in three aspects.Numerical experiments verify the effectiveness and applicability of the method.It provides new ideas for the application of sparse representation in seismic data processing.