用最大似然法进行波场分解和震相识别

本文以介质中弹性波场的传播特征和多元统计分析为基础,从数学上进行严格的推导,得到了模型与资料拟合的最大似然判据,以检验地震波列中是否存在某种特殊波型的能量。利用概率滤波可得分解后的地震图,从图中可以直接得到震相到时。用此方法处理P波资料,效果较好。对于剪切波资料,有时处理结果不十分理想。这可能是由于径向分量和横向分量之间的线性相关性造成的。对于因介质各向异性而形成的分裂剪切波,用此方法得到的S波到时为快S波到时。此方法对于单台三分向记录资料的处理是十分有效的。

SEISMIC PHASE RECOGNITION AND WAVEFIELD DECOMPOSITION WITH MAXIMUM LIKELIHOOD METHOD

In this paper, according to the characters of wavefield propagation in medium and the multivariate statistical analysis, a ML criterion, which fits the data with the model is reached by strict mathematical derivation, is used to test whether the energy of specific Wave type is present. The decomposed seismogram is obtained by using probability filtering, the arrived time can be find out from it directly. The technique is satisfactory for P waves, but is sometimes a little poor for S waves, which is probably due to the correlation between radial and transverse components. Especially for split shear waves, the arrival time obtained by this method, should be that of fast shear waves. Practical examples shown that the method is very available for the proccessing of three component recordings.