Internal Multiple Suppression Based On Seismic Interferometry In Linear Radon Domain

In seismic data processing,multiple is often the most common type of interfering signal,and its existence seriously affects subsequent offset imaging and construction interpretation.The research methods for free surface-related multiple are relatively mature and methods are available.More than that,including the more commonly used free surface-related multiple suppression method(SRME).For some strata,there is not only surface-related multiple but also strong internal multiple,which will also seriously affect subsequent processing and imaging.Therefore,Study of the method to suppress multiple,especially internal multiple is meaningful for producing.On the basis of the predecessors,this article has conducted some research on the prediction and subtraction methods of internal multiple,hoping to further promote the development and progress of internal multiple suppression methods.Firstly,the concept of surface-related multiple and internal multiple is introduced from the classification of multiple.Then,the characteristics of surface-related multiple and its conventional suppression methods are introduced,including the method based on the difference of time and dip,predictive deconvolution,wave field extrapolation and SRME method.Then introduce the characteristics and properties of internal multiple and the common strata which internal multiple exists,and introduce more mature internal multiple suppression methods today,including the common-focus-point(CFP)method and scattering series(ISS)method.In this paper,the basic formulae of convolution and correlation seismic interferometry are deduced through the acoustic wave reciprocity theorem.Since seismic interferometry does not use all seismic records,only the seismic traces near the “stationary phase point” have major contributions.Therefore,we deduce basic formula of the linear Radon domain seismic interference based on the principle of “stationary phase point” and linear Radon transform.Through prolongation and cancellation of ray paths by seismic interferometry,the seismic interferometry can predict internal multiples.Then we transform the space-time domain to the linear Radon domain,to demonstrate the feasibility of seismic interferometry in linear Radon domain to predict internal multiples.Because of the characteristics of sparse domains,it is possible to suppress some artifacts produced in the space-time domain.Due to the "stationary phase point" principle,we calculate only contributive trace to seismic interferometry,and it will increase the computational efficiency.We verify the method by simple layer model and pluto model numerical simulation.After predicting more accurate internal multiple,how to suppress internal multiple in the original seismic data is also a difficult problem.The traditional method of least-squares subtraction cannot match and subtract internal multiple very well,and some multiple event still have partial residuals.This paper uses the multi-scale transformation method-Shearlet transform as sparse transform domain.Combined with Bayes algorithm,transforms the problem of suppression of internal multiple into a conditional-probability maximization problem.Then we obtain Shearlet transform coefficients,and passes the optimal transform coefficients to the original seismic data and predicted internal multiple transform to Shearlet domain.It will have better results when we suppress the internal multiple in Shearlet domain,and the effectiveness of the method is validated by using the internal multiple data predicted by this paper.