基于多次波生成层自适应提取的层间多次波压制方法研究

基于地表数据驱动的层间多次波压制方法需要识别产生层间多次波的界面, 该过程需要较多的人工交互, 且一次只能对一个界面进行识别, 过程较为繁琐。为提高层间多次波生成层的提取质量以及层间多次波的预测效率, 提出了一种基于多次波生成层自适应提取的层间多次波压制方法。首先利用自适应算法对层间多次波生成层进行自适应性提取, 然后采用滑动窗口的预测方式实现对所有层间多次波的一次性预测, 最后使用基于2D卷积信号盲分离的多次波自适应相减方法将预测的层间多次波从原始地震数据中自适应减去。模型数据和实际数据的应用结果表明, 与基于地表数据驱动的方法相比, 基于多次波生成层自适应提取的层间多次波压制方法可以较好地自适应提取层间多次波生成层数据, 并一次性预测出所有层间多次波, 进一步提高了层间多次波的压制效率。

In seismic exploration, internal multiple attenuations can still be challenging due to poor discrimination between primaries and multiples.Internal multiple attenuations are important in processing seismic data, particularly field data.The traditional internal multiple attenuation method, based on surface data, requires significant manual intervention to identify and extract multiple internal generators, and the prediction process is cumbersome.In this study, we proposed a new internal multiple attenuation method based on the adaptive extraction of multiple internal generators.The proposed method first used an adaptive algorithm to extract the data of multiple internal generators and then combined the extracted generator data and original pre-stack seismic data to predict all internal multiples simultaneously using the top-down sliding window approach.Finally, we used an adaptive subtraction method based on the 2D blind separation of convolved mixtures to suppress the predicted internal multiples, which can protect the primaries more effectively.Compared with the traditional surface-based internal multiple elimination method, the proposed method is entirely data-driven and does not require subsurface information, which can more successfully adapt to internal multiple generators and improve the internal multiple prediction efficiency.Tests on synthetic and field data showed that the proposed method could reduce the errors caused by manual intervention and improve the efficiency of internal multiple suppression.