Research On Water-Layer Demultiple Algorithm For 3D OBC Seismic Data

In marine seismic exploration,due to the existence of strong reflection interfaces such as sea level and seabed,water-layer multiples are widely developed in marine seismic exploration data.The existence of water-layer multiples severely reduces the signal-to-noise ratio and resolution of seismic data,resulting in structural artifacts and the lack of certain frequency components in seismic data,which directly affects the processing and interpretation of seismic data.Therefore,in marine broadband seismic exploration technology,water-layer demultiple is the key to broadening the frequency band of seismic data and improving the signal-to-noise ratio and resolution of seismic data.This paper mainly focuses on the research on demultiple methods of ocean bottom cable acquisition.Firstly,the development of marine broadband seismic exploration acquisition technology and the corresponding demultiple technology are analyzed and summarized.The types of multiple waves of ocean bottom cable acquisition,the generation mechanism and frequency characteristics of water-layer multiples are analyzed.The wave field characteristics of ocean bottom cable seismic data and the influence of water-layer multiples on the primary reflection waves and stack profile are analyzed by forward simulation.Cosidering that the conventional SRME method cannot achieve ideal results in the shallow water environment of OBC acquisition,a multiple iterations Green’s function multiple waves prediction method is proposed.Based on the 3D water velocity model,the Green’s function is used to simulate the propagation of seismic waves in the water layer,and the original seismic data is convolved with the constructed Green’s function to predict the water-layer multiples.Based on the actual data processing flow,multiple iterations are used to gradually improve the prediction accuracy of water-layer multiples.Finally,the curvelet transform theory is introduced,and a window by window matching and curvelet domain subtraction demultiple method is proposed.The original seismic data and the predicted multiple waves data are matched by least square matching method window by window,and then the matched data are transformed into the curvelet domain.Using the difference between the primary waves and the multiple waves in the curvelet domain,decompose it into the curvelet coefficients of different scales and directions,and then using the curvelet domain threshold processing method can effectively separate the primary waves and multiple waves.This method can effectively demultiple while retaining effective signals to the utmost extent.Finally,the method in this paper is applied to model data and actual 3D seismic data.The results show that the method in this paper can effectively demultiple,improve the signal-tonoise ratio and resolution of seismic data,compensate for frequency traps,and broaden the frequency band of seismic data.