论文部分内容阅读
在地震危险性分析中,有些由强震的分布控制的量值,一个典型的量是在一特定场地上500年复发周期的峰值地面加速度p(500).这个量值在工程上非常重要。由于地震目录中强震数量不多,可利用的数据通常不足以表现震级分布模型在统计上的有效性。在其他条件都相同的情况下,即使不同的模型在场地产生明显不同的a(500)值,它们也可以用相同水平的统计似然性解释可利用的数据。对特定场地的地震危险性分析,哪个模型更可靠、更符合实际呢?在大多数情况下,传统的拟合检验不能给出有意义的答案。基于特定场地评估a(500)值中期望误差的分布,本文提出了一个特殊的震级模型信度的定义。要在两个可比模型之间选择更合适的模型,信度的概念是一个有用的手段。
In seismic risk analyzes, some of the magnitude controlled by the distribution of strong earthquakes, a typical amount is the peak ground acceleration p (500) at a recurrence of 500 years over a particular site. This magnitude is very important in engineering. Due to the small number of strong earthquakes in the earthquake catalog, the available data are often not sufficient to show the statistical validity of the magnitude distribution model. All other things being equal, they can explain the available data with the same level of statistical likelihood, even though different models have significantly different a (500) values at the site. Which model is more reliable and practical for the seismic risk analysis of a particular site? In most cases, the traditional fitting test can not give a meaningful answer. Based on the distribution of expected error in a (500) value of a particular site assessment, this paper presents a definition of the reliability of a particular magnitude model. To choose a more appropriate model between two comparable models, the notion of reliability is a useful tool.