The Study On Dip-constrained Seismic Data Regularization

In the actual data acquisition process of seismic exploration,d ue to the surface conditions(such as rivers,lakes and valleys)and,the position of geophones is usually irregular in real data acquisition of seismic prospecting.In addition,bad coupling effect,environment disturbance or instrument factors will generate some useless traces and shots,which need to be removed in the processing procedure.In marine seismic prospecting,cables deviate from designed position because of feathering.With the in-depth development of seismic prospecting,the acquired data becomes larger and larger.On one hand we hope to get dense sampled data in order to obtain more accurate subsurface structure.On the other hand,in consideration of cost,data should be sampled sparsely in special direction.In consequence,the data we obtain is usually sparse and irregular.In subsequent processing steps,sparsely and irregularly sampled data will bring in noise,as well as influence some mufti-traces processing technique,such as stacking,wave equation migration,multiple removal and time-lapse seismic.Therefore,reconstruction of seismic data is very crucial and necessary.In this paper,a data regularization method based on iteratively re-weighted least-squares inversion is proposed,which can eliminate the effect of burst noise on the interpolated data.The weighting operator is introduced to weight the data fit residual under Cauchy norm Local plane wave model constraint is used as prior information to make the process of inversion stable and interpolate the aliased data correctly.The inverse problem is solved by the preconditioning conjugate gradient method with fast convergence.Parallel processing along time slices for three dimensional data cube can promote the efficiency of three dimensional data regularization further.Experimental results on theoretical model and real seismic data show that the proposed method is fast,efficient and applicable.