论文部分内容阅读
区域地震似然模型(RELM)工作组的5年实验是设计用来比较预测加州及附近各纬度-经度-震级单元地震发生率的若干有希望的方法。这种预测模式被作为世界范围内其他地震可预测性试验的蓝本,因此考虑如何评估这种预测的性能是很重要的。最近采用的两个试验都基于给定预测情况下观测到的地震分布的概率,一个测试比较了空间-发生率-震级单元的观测值和预测值,另一个测试仅仅比较了预测的发生率和观测地震的数目。在本文中,我们讨论了目前关于发生率预测的微小的缺陷,我们建议采用另外两个测试分别进行空间-发生率-震级预测的空间和震级分量的预测。为了更好地说明问题,我们考虑了区域地震似然模型预测和进行区域地震似然模型实验的前半期观测到的地震分布。我们得出空间—发生率—震级预测好像是和观测地震的分布相一致,尽管空间预测和观测地震的空间分布是不一致的,我们建议这些新的测试应该被用于提供更详细的地震预测评估。我们也讨论了每个基于似然测试的统计学功效以及基于似然测试的结果的稳定性(相对于地震目录的不确定性)。
The 5-year experiment of the Regional Seismicity Likelihood Model (RELM) Working Group is a promising method for the comparison of predictions of earthquake occurrences for latitude-longitude-magnitude units in and around California and nearby. This predictive model is used as a blueprint for other earthquakes in predictability testing worldwide, so it is important to consider how to evaluate the performance of such predictions. The two most recent tests are based on the probability of the observed seismic distribution for a given prediction case. One test compares observations and predicted values of space-occurrence-magnitude units. The other test compares only the predicted incidence and Observed the number of earthquakes. In this paper, we discuss the current minor pitfalls in incidence forecasting. We propose to use two other tests to predict the spatial-occurrence-magnitude prediction of space and magnitude, respectively. In order to illustrate the problem better, we consider the observed seismic distribution in the first half of the regional earthquake likelihood model prediction and the regional earthquake likelihood model experiment. We conclude that space-incidence-magnitude predictions seem to be consistent with the distribution of observed earthquakes and that although the spatial distribution of spatial and observed earthquakes is inconsistent, we suggest that these new tests should be used to provide more detailed estimates of earthquake prediction . We also discuss the statistical power of each likelihood-based test and the stability (based on the uncertainty of the earthquake catalog) of results based on likelihood tests.