复合极值分布在台风风暴潮增水重现期计算中的应用研究  

Application study of compound extreme value distribution in calculating the return period of typhoon storm surge

作　　者：付翔[1] 梁森栋 李涛[1] 吴少华[1] FU Xiang;LIANG Sendong;LI Tao;WU Shaohua(National Marine Environmental Forecasting Center,Beijing 100081,China)

机构地区：[1]国家海洋环境预报中心,北京100081

出　　处：《海洋预报》2022年第4期1-8,共8页Marine Forecasts

基　　金：国家重点研发计划(2017YFA0604903、2017YFA0604901)。

摘　　要：考虑小样本或无规律极值样本的情况下,引入描述随机事件发生次数的离散型分布,以三沙站为例,研究了Poisson-Gumbel复合极值分布在台风风暴潮重现期计算中的应用效果。结果表明:复合极值分布计算结果对样本数据较为敏感,标准差小的样本数据其复合分布的低频值显著偏低;对具有一定离散度的短时间序列数据,复合分布的计算结果稳定,且对低频值的估算也更准;对于较长时间序列的数据,阈值取值偏低则高频次发生值更准,阈值取值偏高则低频次发生值更准。Taking Sansha station as an example,this paper studies the application effect of Poisson-Gumbel compound extreme value distribution in calculating the return period of typhoon storm surge,which considers the conditions of small samples or irregular extreme value samples and introduces the discrete distribution describing the occurrence of random events.The results show that the calculation results of compound extreme value distribution is relatively sensitive to samples,and the low frequency value of the compound distribution of the sample data with small standard deviation is significantly lower.For the short time series observation data with a certain degree of dispersion,the calculation result of the compound distribution is stable,and the estimation of the low-frequency value is also more accurate.For longer time series data,the high frequency occurrence value is more accurate if the threshold value is a lower,and the low frequency occurrence value is more accurate if the threshold value is a higher.

关 键 词：复合极值分布 重现期 台风风暴潮 阈值 

分 类 号：P731.23[天文地球—海洋科学]