基于标准协方差极化滤波方法（SCM），由于其物理意义明确、易于实现、计算效率较高，在多分量地震处理中发挥着重要的作用。但该方法的时窗长度选择完全依赖于经验判断，不可避免地会出现解释上的人为影响。基于此，对协方差矩阵的分析时窗进行改进，时窗的长度自适应于三分量地震记录的瞬时频率，实现了自适应协方差极化滤波方法（ACM）。模型数据及实际三分量台站地震数据处理结果表明，ACM 对局部变化比较剧烈的信号更加敏感，极大提高了滤波精度。%Polarization properties differ among various types of seismic waves.The seismic wave actually collected is the result of interference and is superimposed by vibrations with different types and different polarization properties.Polarization analysis is a signal processing method based on polarization characteristic of seismic waves and can simplify the extraction of information by measuring the polarization properties of the various types of seismic waves.It has a good effect on the identification and separation of specific wave type,the suppression of noise,shear-wave splitting analysis,multi-wave seismic phase identification,and wave arrival time determination. The polarization filtering method based on the covariance matrix plays an important role in multi-component seismogram processing due to its explicit physical meaning,easy implementation,and high efficiency.This type of polarization filtering calculates the polarization parameters in a given time window;thus,the choice of time is very critical.The window length in polarization analysis method based on the standard covariance matrix(SCM)is fixed in the time domain;the computed polarization attribution is an average value in the time window.In practical application,the selec-tion of time window length of the SCM is entirely dependent on the experience,and the polariza-tion attributions in a given length window do not have time-varying characteristics.The filtering results of the SCM are relatively stable,insensitive to disturbance,and unable to determine the polarization parameters in the beginning and end of the seismic record.Thus,the filtering effect is not ideal and will inevitably appear glossy in interpretation.For this reason,the present study in-troduces a new polarization method based on the adaptive covariance matrix(ACM).We use an approximate formula to compute the adaptive window function,in which the length is adapted to the instantaneous frequency of three-component seismic data.In particular,the window length of the covariance matrix adjusts to the minimum cycle of the desired signal,which reduces the facti-tious impact of the window length selection.In addition,there is no need to interpolate processing because the polarization parameters are computed in every time sampling point of the three-com-ponent seismic records,except for one-half of the time window of the start and end points.Due to the above advantages,ACM greatly improves the filtering accuracy.The processing results of the model and actual three-component station seismic data show that the SCM represents a smoothed version of the instantaneous attributes from the ACM.Furthermore,because the time window is fixed for the standard method,it is not possible to characterize the polarization attributes of a seis-mic event with a period lower than that of the time window used for the analysis.The polarization curves computed by SCM and ACM agree quite well in the region in which the period of the domi-nant signal is close to the time window selected for the covariance analysis,which greatly reduces the effective signal waveform differences before and after filtering.With ACM,a comparison of the original signal and that after filtering reveals that almost no high-frequency interference exists near the effective signal.In general,the significantly different regions in which the two curves from the SCM and ACM indicate that the selected window for the ACM is optimal.
展开▼
