论文详情
基于稀疏反演理论,提出了采用L1正则化和L0正则化混合迭代算法对高效采集地震记录的干涉炮噪声进行压制的方法。将含有干涉炮噪声的地震记录重排为共中心点道集,并将干涉炮噪声去除问题转化为一个稀疏反演问题。使用软阈值迭代算法求解L1正则化问题,获得良好的迭代初值;然后采用硬阈值迭代算法求解L0正则化问题,获得最终干涉炮噪声的压制结果。利用该方法对实际独立同步扫描高效采集的地震资料中干涉炮噪声进行了噪声压制测试,获得了良好的效果。对比测试结果与中值滤波方法的去噪结果可以看出,基于稀疏反演的干涉炮噪声压制方法对干涉炮噪声的压制效果更好。
Based on sparse inversion theory,a mixed iterative algorithm that includes L1 and L0 regularization is proposed to attenuate the crosstalk noise in high-efficiency seismic acquisition data.First,shot gathers with crosstalk are rearranged into CMP gathers.Next,the removal of the crosstalk noise is transformed into a sparse inversion problem.Then,the L1 regularization problem is solved with a soft threshold iterative algorithm to obtain a good initial solution to the inverse problem.Finally,the L0 regularization problem is solved using a hard threshold iterative algorithm,which can effectively suppress the crosstalk noise.This method was applied to the actual seismic data acquired with independent simultaneous sweeping.The results showed that the method could effectively suppress the crosstalk noise,and obtain better denoising results than using the median filtering method.
国家自然科学基金面上项目(41574134)和中国科学院百人计划共同资助。