论文部分内容阅读
采用一种新的统计检验方法对 VAN 方法在1987～1989年间预报成功率和随机采样地震目录获得的成功率之间进行了对比(成功率是指成功预报的次数与所发出的预报总次数的比值)。与以往的检验不同,这次检验所采样的分布是从1960～1985年间的历史地震空间和震级的分布所得出的。VAN 方法预报成功率以延时22天为计达67.9%,而在 NEIC 全部目录上随机采样中仅为37.2%,可望达到上述指标(67.9%)的概率比0.06%还要小(P≤0.06%)。如果从目录和预报清单中删除余震,则其成功率还可达66.7%,而随机实现的概率则小于0.1%(P≤0.1%),故仍有足够的显著性。用一种替代的余震模型,成功率降到48.0%,此时仅对 P 值小于7.96%才是显著的。从这些检验结果,我们的结论是 VAN 方法在延时22天的条件可正式通过检验。
A new statistical test was used to compare the success rate of VAN methods from 1987 to 1989 with the success rate of randomly sampled earthquake catalogs (success rate is the ratio between the number of successful forecasts and the total number of forecasts issued ratio). Unlike previous tests, the distribution of the samples tested in this test was derived from the distribution of historical seismic volumes and magnitude between 1960 and 1985. The VAN predictive success rate was 67.9% for a 22-day delay, compared to 37.2% for random sampling on the NEIC total list, and the probability of reaching the above target (67.9%) is less than 0.06% (P 0.06%). If aftershocks are deleted from the catalog and forecast list, the success rate is still up to 66.7%, while the probability of random realization is less than 0.1% (P≤0.1%), so there is still enough significance. With an alternative aftershock model, the success rate dropped to 48.0%, which is only significant for P values less than 7.96%. From these test results, we conclude that the VAN method can pass the test on condition of a delay of 22 days.