Bootstrap抽样下基于贝叶斯估计的地震人员死亡评估研究

(1.兰州财经大学 统计与数据科学学院,甘肃 兰州 730020; 2.中国地震局兰州地震研究所,甘肃 兰州 730020)

地震; 人员死亡; Bootstrap抽样; 贝叶斯估计; 死亡率

Estimation of Earthquake Deaths Based on the Bayesian Estimation by the Bootstrap Sampling
ZHAO Yu1,WEI Maomao1,SUN Yanping2,SHI Yitong2,CHEN Wenkai2

(1.College of Statistics and Data Science,Lanzhou University of Finance and Economics,Lanzhou 730020,Gansu,China;2.Lanzhou Institute of Seismology,China Earthquake Administration,Lanzhou 730020,Gansu,China)

earthquake; earthquake-caused death; the Bootstrap sampling; the Bayesian estimation; death rate

DOI: 10.20015/j.cnki.ISSN1000-0666.2025.0014

备注

选取1950—2022年中国大陆地区历史震例数据信息,以极震区烈度、发震时间及人口密度为主要影响因素,建立三级场景,运用Bootstrap抽样方法对各场景下的历史震例进行扩充,进而采用贝叶斯估计给出不同场景下地震人员死亡变化分布函数,再根据实际情况对此分布函数进行截断分析,以明确地震人员死亡的不确定性变化范围及死亡率区间概率值,最后随机选取历史地震灾害事件验证模型精度。结果表明:①各场景的死亡率均值及95%死亡率区间基本符合在极震区烈度相等并且人口密度条件下夜间死亡率大于白天死亡率的规律; ②模型估计死亡人数区间能够较好覆盖实际上报的死亡人数。
The rapid post-earthquake assessment of the earthquake-caused casualty is crucial for decision-making and deployment of emergency response in earthquake scenarios.In this study,historical seismic data in Chinese mainland from 1950 to 2022 were selected for assessing the impact factors including the seismic intensity in the meizoseismal area,the original time,and the population density.Three levels of scenarios were established,and the Bootstrap sampling method was applied to expanding the historical seismic data for each scenario.Tthe Bayesian estimation was then employed in obtaining the distribution functions of the earthquake-caused casualties for different scenarios.Furthermore,truncation analysis was conducted based on real situations to determine the uncertainty range and the probability of mortality rate intervals.Finally,historical earthquake disasters were randomly selected to validate the accuracy of the model.The results showed that:①The mean and 95% interval estimates of the death rate of each scene basically accord with that the death rate of night in higher than that of day under the condition of the same intensity and the same population density in the earthquake area. ②The range of expected casualties estimated by the model effectively covered the actual statistical deaths.