赣江流域非一致性条件下设计洪水估计

赣江流域年最大日径流量在过去60a呈现出显著的变异特征,以此为样本序列,选用广义极值(GEV)分布来对其进行拟合,并基于五种气候模式输出结果对未来设计洪水的变化情况进行分析。所选协变量包括时间t、洪水前N日降水量和下垫面信息。相关关系分析表明,N=10时降水与洪水关系最为密切,但流域内土地利用类型仅呈现出微弱的变化趋势。模拟结果表明,洪水序列服从Weibull分布,其中形状参数ξ=-0.169、尺度参数σ=2 078.9、位置参数μ=65.1c+437.1。非一致性条件下的设计洪水值与传统方法计算的设计值存在很大区别,设计洪水值在1994年要偏大4 000m~3/s以上,其面临的洪水风险要远超普通年份。未来的设计洪水在低、中、高三种排放情景下均呈现出上升趋势,峰值出现于2040年前后,较1994年估计值偏大1 000～2500 m~3/s。

Evaluation and uncertainty analysis of design flood under unstationary conditions in Ganjiang River Basin

In the past 60 years,the annual maximum daily discharged suffered obvious variations in Ganjang River Basin.The GEV distribution isused to fit the runoff series of it.Based on the five general circulation models(GCMs),the future trends of design floods areanalyzed.The selected covariates includetime t,N-day rainfall before a flood,and the underlying surface information.Pearson correlation coefficients showsthat,the N-day rainfall series arehighly correlated with the flood series when N=10.However,the land use variation tendency is tiny.The simulation result shows that the flood series following the Weibull distribution and has a shape parameterξ=-0.169,scale parameterσ=2078.9,and location parameterμ=65.1c+437.1.The design flood value obtains from the unstationary conditions is significantly different from thestationary model,and the design value iseven 4 000 m^3/s higher in 1994,suggesting a much higher flood risk than regular years.The design floods show increasing trends in low,medium,and highscenarios.The peak value willoccur around 2040,and is 1 000~2 500 m^3/s higher than the design value in 1994.