地震预警震级估算方法的不确定性评估模型——以τmax^p法为例

UNCERTAINTY EVALUATION MODEL FOR EARTHQUAKE MAGNITUDE ESTIMATION METHOD TAKING τmax^p AS AN EXAMPLE

地震预警震级估算结果的不确定性是导致地震预警信息不确定的主要因素之一,震级估算方法的不确定性概率模型也是地震预警系统信息发布决策以及重大工程地震紧急处置决策的关键依据参数。在实时概率危险性分析方法(RTPSHA)的基础上,采取震级均匀先验分布模型及离散震级分档策略。以τmax^p方法为例,选取日本K-Net台网的352次地震的24673条竖向强震加速度记录,建立了卓越周期与震级的统计规律。在此基础上,依据贝叶斯公式建立了地震预警震级估算方法的不确定性评估模型,并利用2次模拟算例与6次实际地震数据验证了方法的合理性。结果表明,震级先验分布采用均匀分布结合离散化处理策略进行震级后验概率计算能够更好地体现τmax^p法估算震级的不确定性特征,解决了RTPSHA方法中的大震概率估计偏低的问题。

The uncertainty of magnitude estimation results is one of the main factors leading to the uncertainty of earthquake early warning information. The uncertainty probability model of magnitude estimation method is also the key basis parameter of information release decision of earthquake early warning system and earthquake emergency disposal decision of major projects. Based on the Real-time probability hazard analysis(RTPSHA), the uniform prior distribution model of magnitude and the strategy of discrete magnitude grading are adopted in this paper. Taking the method as an example, 24 673 vertical acceleration records of 352 earthquakes in Japan’s K-Net network are selected, and the statistical rules of predominant period and magnitude are established. On this basis, the uncertainty evaluation model of the method of earthquake early warning magnitude estimation is established based on Bayesian formula, and the rationality of the method is verified by two actual earthquake data. The results show that using the strategy of uniform distribution and discretization to calculate the posterior probability of magnitude can better reflect the uncertainty of magnitude estimation, which solves the problem of low probability estimation of large earthquakes in RTPSHA method, and can reasonably consider the phenomenon of under estimation of large earthquakes and over estimation of small earthquakes.