基于压缩感知的快速最小二乘逆时偏移

为提高最小二乘逆时偏移成像的计算效率,推动其实用化,将压缩感知理论应用于最小二乘逆时偏移中,提出了一种基于压缩感知理论的快速最小二乘偏移方法。通过对激发震源和观测数据同时进行降采样,可以大幅度降低最小二乘逆时偏移中波动方程的计算。总结了利用压缩感知理论进行快速最小二乘逆时偏移的基本理论,以及在这个框架之下,利用变量投影技术解决未知的震源子波问题,以实现对包含表面多次波的海洋地震数据进行准确的成像。SEG/EAGE盐丘模型和Sigsbee 2B模型数据以及英国北海海域某工区实际数据实验结果表明快速最小二乘逆时偏移方法可行且有效。

Compressive sensing provides a novel approach to address the prohibitively expensive computational cost in the least-squares imaging of seismic data.By down-sampling the source and receiver wavefields,the proposed method greatly reduces the number of wave equations solved in the imaging process.However,down-sampling also leads to ill-posedness of the imaging system.By introducing sparsity constraints,compressive sensing stabilizes the inversion process.In this article,we review the principles of least-squares imaging using compressive sensing and discuss methods for extending this technique to address the issue of unknown source wavelets and to eliminate interferences from strong surface-related multiples.In addition,we review synthetic and field data examples that demonstrate the efficacy of the method.