2017年8月9日新疆精河6.6级、9月16日库车5.7级地震前“库米什地震窗”均出现异常。为了对该异常指标作出评价，对新疆库米什地震台2008年1月～2017年9月微震观测记录进行普查，分析在S-P≤10s范围内M_L≥1.0小震的月频次，定义小震月频次≥54次为“库米什地震窗”的异常标准，统计“库米什地震窗”异常与附近中强地震间的对应关系，并对其进行预测效能评价。结果显示：①“库米什地震窗”分别在精河6.6级、库车5.7级地震前3.3个月、0.5个月出现异常；②“库米什地震窗”2008年以来出现7组异常，分别对应了天山中部地区6次5.5级以上地震，优势发震时间为6个月；③“库米什地震窗”的异常形态表现为高值—低值—发震或高值—发震，异常的对应率为6/7，R值评分为0.45，预测效能评价较好。

The M_S6.6 earthquake in Jinghe on August 9, 2017 and the M_S5.7 earthquake in Kuche on September 16, 2017 occurred successively since August and there were anomalies in the Kümüx seismic window before the two earthquakes. To evaluate the abnormal indicator of the Kümüx seismic window, we investigated the digit seismograph records of the Kümüx seismic station from January 2008 to September 2017, and the monthly frequency change of small earthquakes with M_L ≥ 1.0 and the S-P value of 10s, and defined the monthly frequency of small earthquakes ≥ 54 times(i.e.54 times/month)as the abnormal index of the Kümüx seismic window, as well as their relationship with nearby strong earthquakes. We made an evaluation on its prediction effectiveness by analyzing the abnormal characteristics of the seismic window and the corresponding relation of the moderate strong earthquakes. The main results showed that:(1)The anomaly of the Kümüx seismic window appeared 3.3 and 0.5 months respectively before the M_S6.6 Jinghe earthquake on August 9, 2017 and the M_S5.7 Kuche earthquake on September 16, 2017. (2)There have been 7 groups of anomalies for the Kümüx seismic window since 2008, which were followed by 6 earthquakes of M_S ≥ 5.5 in the central Tianshan area. The advantage time for strong earthquake was within 6 months. (3)The anomaly patterns of monthly frequency for small earthquakes were characterized by high value-low-earthquake or high value-earthquake. The corresponding rate of abnormality was 6/7 and the score of R value was 0.45. The evaluation result of the prediction efficiency for the Kümüx seismic window is better.

P315

中国地震局震情跟踪定向工作任务（2017010127）、中国地震局地震预测研究所新疆精河6.6级地震深入科学研究项目（1747074512）共同资助