Research On 3D Surface Related Multiple Elimination Based On Integral Path Optimization

Multiples are seismic responses received by receivers after multiple reflections at interfaces with large impedance differences.The existence of multiples reduces the signal-to-noise ratio of seismic data,interferes with the identification of primaries and affects the subsequent imaging and interpretation process.Therefore,multiples are often removed as noises.Especially in marine seismic exploration,the existence of the strong reflection interface on the sea surface develops lots of surface-related multiples,which make the multiple elimination be more important.In addition,since the characteristics of the surface-related multiples are similar to those of the primaries,it is difficult to eliminate multiples without damaging primaries.As a result,multiple elimination has been a classic problem in seismic data processing in recent years.This paper firstly introduces the theory of multiples and the principles of various multiple elimination methods.The kinematics and dynamic characteristics of multiples can be further understood by introducing the basis theory of multiples.The multiple elimination methods can be divided into filtering methods based on feature differences and prediction subtraction methods based on wave equation,including prediction deconvolution,Radon transform method,wavefield extrapolation,feedback loop,backscattering progression method and so on.Among them,surface-related multiple elimination(SRME)is the mainstream multiple elimination method in the industry.It is divided into two steps: prediction and subtraction.Compared with other methods,it has higher accuracy in suppressing multiples in the case of complex seabed structures.Therefore,SRME is the focus of this paper.Considering that the propagation path of the real multiples is a 3D spatial function,2D SRME cannot accurately predict complex 3D geological structures.Therefore,this paper mainly studies 3D SRME.However,by introducing the principles of the 3D marine seismic streamer acquisition system and the 3D SRME algorithm in detail,it is found that since the 3D acquisition system cannot meet the requirements of 3D SRME for dense sampling in the crossline direction,the direct summation of the sparse crossline multiple contribution gathers produces serious spatial aliasing.In the research of improved 3D SRME methods for solving the spatial aliasing caused by crossline sparse sampling,it is found that the 3D surface-related multiple prediction based on sparse inversion(3D MPSI)is an important 3D multiple elimination method in the industry.It assumed that the events of the sparse crossline multiple contribution gathers are hyperbolic,and the Cauchy-constrained inversion is used to sum the events along the hyperbolic direction instead of the direct summation process of 3D SRME.However,the single hyperbolic integral path selection,multi-parameter limitation in inversion,and large computational cost limit the application of 3D MPSI.Based on the assumption of the hyperbolic events in 3D MPSI,considering the optimization of the hyperbolic integral path,the Apex-shifted Radon transform is modified into 3D SRME.In this paper,an integral optimization multiple elimination method is proposed,which is 3D SRME algorithm for integral path optimization based on Apex-shifted Radon transform.This paper modify the apex-shifted hyperbolic Radon transform to implement the summation of crossline multiple contribution gathers with variable apexes along adaptive integration paths by traversing the apex positions and curvatures.The effect of the algorithm is verified by synthetic data,where the spatial aliasing and the damage of primaries are reduced.In the integral optimization multiple elimination method,locating apex position of hyperbolas accurately is the key to obtain good effect of multiple elimination.Locating the apex position by Apex-shifted Radon transform is by traversing the apex positions and curvatures and then filtering the maximum value of the transform result.It not only requires a lot of computational cost,but also the location accuracy is not stable.By analyzing the similarity of apexes between input data and the Radon transform results,this paper uses local similarity to locate the apex positions.Besides,using it in the integral optimization multiple elimination method,an improved integral optimization multiple suppression strategy is proposed.The effectiveness and applicability of the improved integral optimization multiple elimination strategy is demonstrated by synthetic data and field data.