多变量洪水非平稳频率分析方法研究——以淮河流域蚌埠水文站为例  

Research on Nonstationary Multivariate Flood Frequecny Analysis:A Case Study of Bengbu Hydrological Station in the Huaihe River Basin

作　　者：徐鹏程 刘楠楠 李帆 仇建春 张昌盛 戴笠 XU Peng-cheng;LIU Nan-nan;LI Fan;QIU Jian-chun;ZHANG Chang-sheng;DAI li(College of Hydraulic Science and Engineering,Yangzhou University,Yangzhou 225009,Jiangsu Province,China;Jiangsu Hydraulic Engineering Construction Co.,Ltd,Yangzhou 225009,Jiangsu Province,China;Taizhou Port and Shipping Development Center,Taizhou 225300,Jiangsu Province,China)

机构地区：[1]扬州大学水利科学与工程学院,江苏扬州225009 [2]江苏省水利建设工程有限公司,江苏扬州225009 [3]泰州市港航事业发展中心,江苏泰州225300

出　　处：《中国农村水利水电》2023年第6期12-20,共9页China Rural Water and Hydropower

基　　金：江苏省青年基金项目(BK20220589)。

摘　　要：气候变化背景下淮河流域水文气象序列的趋势性引发的非平稳特征势必会对洪水多变量风险产生一定的影响。基于LR-AICc非平稳模型优选准则构建动态Copula函数为核心的联合分布模型,以淮河流域蚌埠水文站的洪峰、洪量序列为研究对象,将与洪峰、洪量序列相关的城市化因子、气候变化因子作为协变量纳入到时变边缘分布模型、Copula模型的参数结构中。结果表明:提出的LR-AICc非平稳检测方法相比于非参数的趋势检测法更能够捕捉到极值序列的非单调性和参数的反向趋势性;由于边缘分布参数的时变特性,50年一遇联合重现期水平下的洪峰-洪量分位数对呈现一定的时变特性:①1960-1970年间,洪峰、洪量的设计值呈增大趋势;②1970-1990年间呈现下降趋势;③1990-2010年呈现逐渐增大趋势。单一考虑气候变化或者城市化因素都会出现一定程度上的高估或低估洪峰-洪量的设计分位数。相比于城市化因素,气候变化因素是蚌埠水文站控制的子流域内影响洪水单变量、多变量风险的主导因素。基于动态Copula模型为核心的非平稳频率分析思路有利于定量剥离气候变化因子、人类活动因子(城市化)对于洪水多变量风险的影响。Under the background of climate change,the trend-caused nonstationarity of the hydrometeorological series in Huaihe River Basin are bound to impose a potential impact on the multivariable flood risk.Based on the LR-AIC criteria,this paper constructs a joint distribution model with dynamic Copula function as the core,takes the flood peak and flood volume series of Bengbu Hydrological Station in the Huaihe River Basin as the research object,and incorporates the urbanization factors and climate change factors related to the flood peak and flood volume series as covariates into the parameter structure of time-varying marginal distribution model and Copula model.The analytic results show that the proposed LR-AIC criteria for nonstationary detection can better capture the nonmonotonicity of extreme value series and the reverse trend of parameters than the nonparametric trend detection method.Due to the time-varying characteristics of maginal distribution parameters,the quantile pairs of peak flood volume at the level of 50-year joint return period show certain time-varying characteristics:①From 1960 to 1970,the design values of the peak flood volume showed an increasing trend;②From 1970 to 1990,it showed a downward trend;③From 1990 to 2010,it showed a gradually increasing trend.Only considering climate change or urbanization factors will lead to overestimation or underestimation of the design quantile of flood peak flood volume to a certain extent.Compared with urbanization factors,climate change factors are the dominant factors affecting flood univariate and multivariable risks in the sub-watershed controlled by Bengbu Hydrological Station.The nonstationary frequency analysis idea based on the dynamic Copula model is conducive to quantitatively isolating the impact of climate change factors and human activity factors(urbanization)on flood multivariate risk.

关 键 词：COPULA 非平稳 多变量频率分析 洪水 淮河 

分 类 号：TV213.9[水利工程—水文学及水资源]