基于洪水事件取样的两变量分布分析  

Analysis of Bivariate Distribution Based on Flood Event sampling

作　　者：赵萌[1] 胡铁松[1] 石涛[1] 吴凤燕 

机构地区：[1]武汉大学水资源与水电工程科学国家重点实验室,湖北武汉430072 [2]湖北省水利水电科学研究院,湖北武汉430070

出　　处：《水力发电》2015年第4期8-11,共4页Water Power

基　　金：国家自然科学基金资助项目(51339004;51479142);湖北省水利科技项目(HBSLKY201401)

摘　　要：针对洪水频率分析通常以年最大值选取样本进行单变量或者两变量分析存在的不足,采用Gumbel逻辑模型建立基于洪水事件取样的两变量联合分布,利用标准Gumbel分布拟合各特征变量边缘分布,并推求了两变量联合分布、重现期和条件概率。以三峡水库坝址下游宜昌站洪水研究为例的计算结果表明,该模型可以获得给定概率或者重现期洪水事件相应的洪量和洪峰组合情况,反之亦然;同时可以在给定洪峰出现的条件下推求洪量的分布情况,从而更全面、有效地评估洪水事件。In view of the defects of univariate or bivariate analysis existing in flood frequency analysis which often selecting samples based on annual maximum, the Gumbel logistic model is proposed to be used for establishing a bivariate extreme value distribution model based on flood event sampling. Then the standard Gumbel distribution is used to fit the marginal distribution of each feature variable and finally the joint distributions, the return periods and the conditional probability functions are derived. Taking the flood data of Yichang Hydrological Station downstream Three Gorges Reservoir as study case, the calculation results indicate that： （a） the model can provide various occurrence combinations of flood peaks and volumes when a flood event probability or return period is given, and vice versa, and （b） at the same time, the flood volume distribution can also be derived when the volume peak is given for a more comprehensive and the effective assessment of flood events can be given.

关 键 词：洪水事件 Gumbel逻辑模型 联合分布 

分 类 号：TV697[水利工程—水利水电工程]