利用曲波变换预测多次波模型

针对SRME方法在多次波模型预测阶段存在的缺陷,利用曲波变换多方向的特性,自动选取遵循斯奈尔定律(即走时平稳区域)的多次波贡献道集,对多次波模型道进行构建。该策略减小了非平稳走时区域多次波贡献道集可能产生的假频对多次波模型的影响。对于原始含有空间假频的数据,借助曲波变换多尺度的特性,利用地震数据中不含假频的低频分量对假频分量进行约束,构建了一个抑制假频滤波器,可以减小空间假频对多次波模型的影响,提高多次波模型的预测精度,为后续的相减阶段提供更好的模型数据,有利于最终多次波压制。模型数据和实际资料验证了本文方法的有效性。

Multiple model prediction based on Curvelet transform

Considering the defects of multiple model prediction by SRME, we propose in this paper an approach to reconstruct multiple model trace based on Curvelet transform. The directional characteristics of Curvelet transform allow Snell's law at the free surface to automatically construct the multiple contribution gathers(MCG) which obey Snell's law or exist in stationary zones. This strategy reduces the artifacts caused by the non-constructive multiple contributions. Moreover, based on multi-scale property of Curvelet transform, an anti-aliasing weighting filter is designed using the low frequency dataset without space aliasing to reduce the adverse influence to the multiple model when the input original dataset have space aliasing. In this way, the accuracy of multiple prediction models can be improved for the latter subtraction stage. Finally, tests on synthetic and field datasets have demonstrate the effectiveness of the proposed approach.