基于泊松-对数正态复合极值模型的洪水灾害损失分析  

Flood disaster losses analysis based on the Poisson-lognormal compound extreme model

作　　者：刘家福[1,2] 吴锦[2] 蒋卫国[3] 占文凤[2] 

机构地区：[1]吉林师范大学旅游与地理科学学院,吉林四平136000 [2]北京师范大学减灾与应急管理研究院,北京100875 [3]民政部/教育部减灾与应急管理研究院(北京师范大学),北京100875

出　　处：《自然灾害学报》2010年第6期61-66,共6页Journal of Natural Disasters

基　　金：国际科技合作计划专项项目(2007DFA20640);国家科技支撑计划课题(2006BAJ05A01;2006BAJ09B06;2006BAJ09B03;2008BAC44B03);国家自然科学基金(4070117240771155);国家863计划(2009AA12Z124);水文水资源与水利工程科学国家重点实验室开放基金(2008490611);吉林师范大学硕博启动项目(吉师博2010022)

摘　　要：在国际紧急灾害数据库的支持下,得到了中国在1980-2008年间发生的年洪水灾害发生次数、年单次洪水灾害经济损失极大值和年洪水灾害经济总损失3个统计指标;根据复合极值理论,利用泊松-对数正态复合极值模型对洪水灾害经济损失进行了分析。研究结果表明:(1)洪水灾害经济损失遵从对数正态分布;(2)单次极值经济损失与年经济总损失具有高度相关性;(3)复合极值方法可有效地用于洪水极值重现期的重建;(4)与传统的经验频率计算方法相比,该方法能克服因资料年限短、数据不足而造成的洪灾重现周期估算困难。Based on the EM-DAT（OFDA/CRED） emergency disaster database,three samples of statistical data,including the number of flood disasters,the maximum economic loss in a single flood disaster and the total economic loss in floods were obtained annually from 1980 to 2008 in China.And then,according to compound extreme distribution,Poisson-lognormal compound extreme model was utilized to analyze economic losses of flood disaster.The results indicate that：（1）economic losses of flood disaster is subject to log-normal distribution;（2） a high correlation exists between single extreme economic losses and the total annual economic loss;（3） compound extreme value method is highly effective in reconstruction of return period of extreme flood.（4） compared with the traditional experience method of calculating the frequency,the method can overcome the difficulties of estimating return period due to insufficient information on data and year number.

关 键 词：复合极值模型 泊松分布 对数正态分布 灾害损失 

分 类 号：P426.6[天文地球—大气科学及气象学]