微地震信号到时自动拾取方法

本文讨论了用于微地震信号到时自动拾取的几种方法的原理及特点,包括长短时均值比(STA/LTA)方法、AIC方法、基于高阶统计量偏斜度和峰度的PAI-S/K方法等,提出了移动时窗峰度的快速算法和改进的峰度拾取初至方法.对我国西部某地观测到的13359个微地震记录,采用两种时窗进行了初至到时拾取,并与人工拾取的结果进行了对比.为使所研究的方法达到最佳效果,采用DE全局搜索方法,以人工拾取的初至作为参照,以时差在0.3 s以内的记录所占百分比作为目标函数,自动搜索最佳的拾取参数.结果显示,在拾取时窗选为P波初至前3 s至S波初至位置时,AIC方法的结果最佳,时差在0.3 s以内的记录占比达到93.6%;在拾取时窗选为包含S波到时的时窗时,改进的峰度法效果最佳,时差在0.3 s以内的记录占比83.8%.

On micro-seismic first arrival identification: A case study

Several approaches for automatic seismic first arrival identification are discussed in this paper, which include STA/LTA method, AIC method, and PAI-S/K method. A fast algorithm for computing kurtosis of seismic wave and a revised PAI-S/K method are proposed. The methods are applied to 13359 micro-seismic records from western China, and the results are compared to manual picking results. In order to achieve best picking results, the DE (differencial evolution) method is used to optimize picking parameters. The result shows that for the time window containing P signals only, the AIC method gives best results, about 93.6 percent of records are within 0.3 seconds time difference from manual picking. For the time window containing both P and S signals, the revised PAI-S/K method turns out to be the best one, about 83.8 percent of records are within 0.3 seconds time difference from manual picking.