基于稀疏表示的非零井源距VSP波场分离方法

Offset VSP wave field separation based on sparse representation

随着垂直地震剖面(VSP)技术的不断发展,高保真的波场分离成为VSP资料处理与应用的关键之一。引入欠定盲源分离思想,利用VSP波场的极化特征及多分量资料在时频域的稀疏性,提出一种基于稀疏表示的波场分离方法。该方法分为两步:第一步将地震信号变换到时频域,并估计极化矩阵;第二步在时频域分离各单一波场,再反变换回时间域。在第二步波场分离过程中,针对时频域波场稀疏性不足的情况,提出一种改进的最短路径分解法,在一定程度上解决了时频域局部信号混叠的问题。模型试算与实际资料处理结果表明,与传统方法相比该方法能够更准确地分离上、下行P-P波、P-SV波,分离后的波场保幅性较好,不存在混波、假频以及边界问题。

With the development of vertical seismic profile (VSP),high fidelity wave separation becomes one of key issues in VSP data processing and applications.According to underdetermined blind signal separation theory,a method based on sparse representation is presented in this paper.It depends on polarization characteristics of VSP wave field and sparsity of multi-component data in the time-frequency domain.The method has two stages: first,transforming seismic data from the time domain into the time-frequency domain and estimating polarization matrix; second,separating pure wave in the time-frequency domain and transforming it back into the time domain.It is difficult to acquire pure wave if it is not sparse enough in the time-frequency domain.In this paper,a modified shortest path decomposition method is presented,which can solve the problem of pure wave overlap in the local time-frequency area to a certain degree.Synthetic and real data examples show that with the proposed method every pure wave is accurately separated and amplitudes are well maintained without flaws such as wave mixing,aliasing,and boundary problem compared with conventional methods. © 2017, Editorial Department OIL GEOPHYSICAL PROSPECTING. All right reserved.