Research On The Separation Of Simultaneous Sources Data Based On Sparse Inversion Method

Simultaneous sources acquisition can greatly boost economic saving and improve the equality of imaging,however,which can not be processed by conventional seismic data processing flow directly,because of serious interference noise generated by it.Therefore,we must separate the simultaneous source data.In this paper,we mainly focus on the deblending of simultaneous source data using sparse inversion method.Firstly,the problem of deblending can be treated as an inversion method,solved by three inversion iteration frameworks.We apply theory and modeling tests to compare and analyze these frameworks,in order to choose a suitable one.Secondly,the inversion method based on low rank is selected as a comparison method to highlight the advantage of inversion method based on curvelet transform.Thirdly,because of the shortcoming of inversion method based on curvelet transform,especially for data with large curvature,we find the reason why this method can not process the data with large curvature well,and the virtue of data with small curvature,by analyzing the difference of curvelet coefficients between these two kinds of data.Then we provide an approximately flattened operator to transform the data with large curvature into one with small curvature to improve the original method.In the process of inversion,the approximately flattened operator and curvelet transform operator can be integrated into coherency-filter operator,to greatly improve the deblending results.Finally,in order to prove the advantage of improved method,we test the complex modeling and simulated real data,and obtain favorite results.