科研之友
微信
新浪微博
Facebook
分享链接
摘要
It is particularly important to quantify the effectiveness of the earthquake sequence short-term forecast model and investigate the time variation of the earthquake sequence parameters for construction of the regional earthquake forecasting experience. The epidemic type aftershock sequence (ETAS) model and the ETAS-based thinning algorithm were considered, which are the best seismicity description model and forecast method accepted widely in the world. The Ludian, Yunnan M(S)6. 5 earthquake sequence provides new scientific data for such retrospective case study. The ETAS model was fitted to the earthquake sequence of the Ludian, Yunnan M(S)6. 5 earthquake for examining the stability of early-estimation sequence parameters and relationships among those parameters, with cutoff magnitude M(L)2. 0, and the starting time fixed by C-0=0. 049 day, the ending time from 2. 00 days to 149. 20 days by sliding with 0. 05 day steps and a total of 2945 time periods in ETAS fitting. The ETAS-based thinning algorithm was used to forecast the future next-day and 3-days aftershock rate in all the time periods, and a statistical method N-test was used to test the efficiency of such a continuous forecasting. The parameters are mu=0. 0000, K=0. 0103, c=0. 0039, alpha=1. 8180 and p = 0. 8018 for the ETAS fitting in the Ludian, Yunnan M(S)6. 5 earthquake sequence, and related triggering ability in generating secondary aftershocks, and the decay rate of aftershocks show no significant differences compared to other earthquake sequences in China mainland. The alpha value reduced from 3. 5 similar to 3. 8 to 1. 6 similar to 2. 0 in the early stage after the mainshock occurred since the ending time t(2)=5. 05 similar to 5. 10 days; the p value from 1. 07. down to 0. 78 within 25. 00 days after the mainshock and remains steady at 0. 72 similar to 0. 85; and the b value increased from 0. 80 to 0. 95 gradually within 35. 00 days after the mainshock, and remains steady at 0. 93 similar to 0. 97. Furthermore, the p and alpha values show a reverse change, the p and b value also show a reverse change within 35. 00 days after the mainshock but no obvious correlation after this stage. There is no obvious correlation between the b and alpha values. The N-test result shows that the ETAS model-based thinning algorithm will fail to forecast the aftershock occurrence rate sometimes, for the different ending time selected and continues forecasting performed. For future 1-day forecasting, the failed forecasting is more dispersed when sliding the ending time of earthquake sequence, and the failure rate can reach to 12%. For the 3-day forecasting time window selected, the failure rate is about 6%, and failure forecasting exhibits relative concentration in the time axis, especially in the early stages after the mainshock. It can be inferred that for optimizing the strategy of aftershock forecasting, the shorter forecasting time window should be used in the early stage after the mainshock, while a relatively longer forecasting time window can be used in the subsequent stages, such as a future 3-day forecasting time
- 出版日期2015-11
- 单位中国地震局
