洪水超定量模型的广义Pareto分布及其应用

Generalized Pareto Distribution of Partial Duration Flood Model and Its Application

为了扩大洪水信息,提高洪水模拟精度,研究超定量洪水频率分析模型。介绍了洪水超定量模型的基本理论,假设超定量洪水年发生次数服从Poisson分布,超定量洪水系列服从广义Pareto(GP)分布,给出年最大超定量洪水分布和超定量洪水重现期的计算方法,提出通过模型拟合优度检验来综合确定超定量系列阈值的方法。将超定量模型应用在海河流域小觉站洪峰频率分析中,结果表明:通过模型拟合优度检验确定超定量系列阈值的方法有效且可靠,洪水超定量系列年平均发生次数服从Poisson分布,GP分布洪峰设计值略大于P-Ⅲ分布洪峰设计值,应用在水利工程设计及风险分析中是偏安全的。

Partial duration flood （PDF） frequency analysis model was studied for expanding the flood information and enhancing the simulation accuracy of the flood.The basic theory of the PDF was introduced and the hypothesis was given that the annual occurrence number of the partial duration flood series is estimated by the Poisson distribution,the partial duration flood series are estimated by the generalized Pareto （GP） distribution,then the distribution of the annual maxi mum partial duration flood series and the calculation formula of the PDF return period could be obtained.A model goodness of fit test has been proposed for determining the threshold of the partial duration flood.The PDF model was applied in the flood peak frequency calculation for the Xiaojue Station in the Haihe River Basin.The results show that the goodness of fit test is an efficient and reliable method for determining threshold,and the annual occurrence number of the partial duration flood series follow the Poisson distribution.The design value of the flood peak based on the GP distribution is slightly larger than that on the P-Ⅲ distribution,which will be more secure when used in the design and the risk analysis of the hydraulic engineering.