压制与自由表面相关多次波的独立分量分析方法

Surface-related multiple suppression by independent component analysis

摘要:

基于二阶统计量的多次波压制自适应相减法要求一次波与多次波正交才能实现完美的多次波消除，但是一般地震资料一次波和多次波不是正交的，所以多次波不能完全被压制。本文将独立分量分析（ICA）方法与逆散射级数多次波预测方法结合起来压制多次波。用逆散射级数法预测与自由表面相关各阶多次波，并与原始地震记录组成多个观测量，通过独立分量分析方法，将与自由表面相关多次波从原始地震记录中分离出来。理论合成数据试验表明，即使预测多次波与原始记录中的多次波能量不能很好匹配，ICA法分离多次波也能取得较好的效果，但是，原始资料低信噪比使得ICA方法分离多次波具有不确定性。

Abstract:

Multiples suppression by adaptive subtraction methods based on second order statistic can obtain a perfect result in case the primaries and multiples are orthogonal, but the assumption is always not valid. In this paper, we combine independent component analysis (ICA) with inverse scattering series (ISS) method to suppress surface-related multiples. Multiples predicted by ISS and raw seismic records make up several observation components, then multiples are separated from the seismic records by ICA. Tests on synthetic data generated by convolution and finite-difference modeling show the advantages of ICA in multiples suppression even mismatching in amplitude between the predicted multiples and raw seismic data. But there is uncertain in multiples separation by ICA in case of low signal-to-noise ratio.