利用偏振约束对最小信息准则方法自动拾取微震初至的改进

An Improved AIC Method for Automatic Micro-seismic Data Arrival Picking with the Constraint by Polarization

初至拾取是影响微震事件分析精度的重要因素之一。本文结合微震事件偏振特性和最小信息准则(akaike information criterion,AIC)函数特性,提出了一种利用偏振约束实现AIC初至拾取的改进方法:可以将偏振特征值拾取微震初至的应用扩展至单分量的微震数据,该方法将单分量微震数据视为三分量微震数据的一种特殊形式,利用三分量微震数据协方差矩阵的最大值序列对AIC方法进行约束,从而快速准确的拾取到微震数据的初至。文中应用该方法对不同信噪比的合成数据和实测数据进行了验证,同时与长短时平均(short time average/long time average,STA/LTA)、Maeda-AIC和偏振特征值方法进行了对比,结果显示该算法速度略低于上述3种方法,但精度和可靠性优于其他三种方法,同时与其他改进算法对比,不用设置阈值,并且选取时窗的长短对拾取结果几乎没有影响,可极大地提高算法的自动化程度。

Arrival-time picking is the basic and premise of micro-seismic data analysis. We have analyzed the function of polarization and AIC,combined with the method of polarization,presented an improved method of AIC.This approach includes 3 steps： First,calculate the eigenvalue of covariance matrix which is made up by the original seismic data; Second,find the maximal value of the maximal eigenvalue sequence,generally,the value is behind the first arrival time of micro-seismic data; In the end,find the minimum value of Maeda-AIC function which range is from the very beginning to the maximal value of the maximal eigenvalue sequence,the minimum value can be seen as the arrival time of the micro-seismic data. We verified the method with synthetic data and field data,and also compared with STA/LTA,Maeda-AIC and polarization method,method in this paper is a little slower than the other three method in calculation speed,but more accuracy than the other three method,compared with any other improved method,the method does not need to set a threshold value,just one time window need to be setting,and the length of time window almost have no effect on the arrival picking,greatly enhance the automation of the method,and can be easily added into an automatic processing workflow.