Research Of The Data-driven Suppression Of Internal Multiples

Multiple is a kind of common coherent noise in seismic data processing.Its existence will cause false reflection events in the seismic profile.In addition,multiple can overlap with the effective wave and conceal its energy.This will reduce the signal-to-noise ratio and resolution of seismic data,and then affect the accuracy of the migration of primaries.According to the relationship with free surface,multiples can be divided into the surface-related multiples and internal multiples.At present,the suppression method of surface-related multiples has been relatively mature,but with the increasing complexity of oil and gas exploration targets and the increasing requirements for exploration accuracy,the influence of internal multiples has become increasingly prominent.How to effectively suppress internal multiples has become a key problem in seismic exploration.Based on the existing research results,this article has conducted some research on the data-driven internal multiple suppression method.Firstly,this paper introduces the related properties of multiples,summarizes the conventional methods of multiples suppression,and analyses the kinematics and dynamics characteristics of internal multiples.This introduction lays a theoretical foundation for the suppression of internal multiples.Then,we derive the prediction formula of internal multiples according to the inverse scattering series theory,and simplify it by introducing the step function,the finite length of wavelet and the finite integral interval.Thus,the computational efficiency is further improved while retaining the data-driven advantages of the inverse scattering series method.Then,according to the analysis of the characteristics of internal multiples and primaries,we take the non-Gaussian maximization of primaries as the optimization objective,and adopt the adaptive subtraction method based on 2D blind separation of convolved mixtures(BSCM)to suppress internal multiples.The adaptive subtraction method utilizes the continuity of adjacent events,and does not need the orthogonality assumption of primaries and multiples.It can effectively protect primaries while subtracting internal multiples.The processing results of the simple horizontal layered model,SMARRT model and actual data show that the inverse scattering series method in this paper can predict internal multiple directly from seismic records without relying on velocity model.At the same time,the processing results of different subtraction methods prove that the 2D BSCM method in this paper can effectively subtract the predicted internal multiples and protect the energy of primaries.