论文部分内容阅读
利用M8和MSc算法(即:the Mendoeino Scenario),对环太平洋地震带的最大地震进行了实时中期预测的研究。首先用M8算法进行预测,而后在第二次逼近预测中,以漏报部分地震为代价用MSc算法缩小了发布警报的预测范围。1992～1997年间,有5次8级和8级以上地震发生在试验区内。M8算法对这些强震全部实现了预测,MSc算法准确地判定出其中4次地震的位置。当用震中和均匀时间经验分布的归一化乘积量度估计时,发布警报的空间-时间体积分别为36％和18％。M8和MSc算法两者得到的预测结果的统计显著性均超过了99％。对于震级为7.5+的地震,利用M8算法,19次中预测了10次,占总研究时空体积的40％;用M8-MSc,预测出其中的5次,占总研究时空体积的13％。这就意味着用M8算法预测的显著性水平为81％,用M8—MSc为92％。显著性水平偏低可能是由于1993～1996年期间全球地震活动水平的变化引起的。在此期间,最大地震的发生率增加了1倍,而且又都全部发生在正断层或逆断层上。这些预测都是可重复的。我们在启动预测检验之前,已发表了介绍M8和MSc算法的文章,其中给出了这两种算法的完整正式定义(Keilis-Borok and Kossobokov,1990;Kossobokov,et al,1990;Healy,et al,1992)。IASPEI软件库中存有M8算法的介绍材料(Healy,et al,1997)。
Using the M8 and MSc algorithms (ie, the Mendoeino Scenario), real-time, medium-term prediction of the maximum earthquakes in the Pacific Rim has been carried out. First of all, M8 algorithm is used to predict. Then in the second approximation prediction, MSc algorithm is used to reduce the prediction range of alarm release at the expense of missing part of the earthquake. Between 1992 and 1997, there were five earthquakes of magnitude 8 and above occurred in the experimental area. M8 algorithm to predict all of these strong earthquakes, MSc algorithm accurately determine the location of 4 earthquakes. When estimated using a normalized product measure of empirical distribution of epicentral and uniform time, the space-time volumes of alerts were 36% and 18%, respectively. The statistical significance of both M8 and MSc predictions was over 99%. For the earthquake of magnitude 7.5+, M8 algorithm was used to forecast 10 out of 19 times, accounting for 40% of the total space-time volume. Five of them were predicted by M8-MSc, accounting for 13% of the total time-space volume. This means that the level of significance predicted by the M8 algorithm was 81% and the M8-MSc was 92%. The low level of significance may be due to the change of the global seismic activity level from 1993 to 1996. During this period, the incidence of the largest earthquakes doubled, and all of them occurred on normal faults or reverse faults. These predictions are repeatable. We have published an article introducing the M8 and MSc algorithms prior to starting the predictive test, which gives a complete formal definition of the two algorithms (Keilis-Borok and Kossobokov, 1990; Kossobokov, et al, 1990; Healy, et al , 1992). The IASPEI repository contains introductory material for the M8 algorithm (Healy, et al., 1997).