基于互相关改进的DTW在角道集叠加优化中的应用

The application of improved Dynamic Time Warping in angle-domain gather stacking

逆时偏移对速度十分敏感,当速度存在误差时,角道集同相轴会发生弯曲,直接叠加会削弱成像结构信息.在无从改善速度的前提下,将角道集拉平再叠加是提升叠加效果的有效措施.借助动态时间规整算法(DTW),选取误差小的地震道作为参考道,通过欧式距离计算其他地震道与参考道的相似性,即可求取两者之间的匹配关系,拉平同相轴.但该方法仅考虑了地震道间幅值的相似性,当幅值差异较大时易造成错误匹配,导致同相轴波形畸变、振幅异常.为此,本文提出了一种基于互相关的动态时间规整算法(CDTW),采用一维互相关代替欧氏距离作为地震道间的相似性度量,并以利用二维互相关DTW计算的约束窗限制规整路径.测试表明,CDTW能将角道集同相轴拉平且对波形畸变具有良好的鲁棒性;经处理后叠加剖面中的同相轴较之前更加聚焦.

Reverse time migration is sensitive to the velocity.Curving events can be observed in the Angle-Domain Common-Image Gathers(ADCIG)when the velocity model has an error.The imaging structure information will be weakened if stacking directly.This work can be done with the Dynamic Time Warping(DTW)algorithm.In the algorithm,the seismic trace with small error is selected as the reference track and the similarity between other traces and the reference track is calculated by the Euclidean distance,so that the matching relationship between them can be determined by the similarity.Finally events can be flattened by the matching relationship.However,this method only considers the similarity of the amplitude between the seismic traces.When the amplitude difference is large,it is easy to cause mismatch,resulting in distortion of the waveform of events and abnormal amplitude.Aiming at this problem,an improved algorithm of Dynamic Time Warping based on Cross-correlation(CDTW)was proposed in the paper.This algorithm uses 1 D cross-correlation instead of Euclidean distance as the similarity measure between seismic traces,and uses the constraint window calculated by 2 D cross-correlation DTW to limit the warping path.The test results show that CDTW can flatten events in ADCIGs and has good robustness to waveform distortion.After processing by CDTW,events in the stacking image are more focused than before.