陕北黄土高原无资料地区洪水特征值分位数估算方法研究

Study on Estimation Method of Quantile of Flood Characteristic Value in Data Shortage Area on the Loess Plateau in Northern Shaanxi

数据短缺是制约陕北黄土高原区科学防御洪水灾害的关键性因素,探索数据短缺区域的洪水特征值分位数估算方法具有科学意义。以榆林市28个水文站点的实测最大一日流量(MDF)的分位数为响应变量,以集水面积、淤地坝控制面积、水文站点的经纬度等12个指标为解释变量,研究了线性CCA与非线性NLCCA两种同质区域识别方法与线性MLR和非线性GAM两种区域水文分位数估计方法交叉耦合所形成的4种区域洪水频率分析模型的估算效能,并评价了其在陕北黄土高原的适用性。结果显示,NLCCA-GAM凭借其非线性优势,相比其他耦合模型(CCA-MLR、NLCCA-MLR、CCA-GAM)具有更高的精度和鲁棒性,可作为陕北黄土高原无资料地区洪水特征值分位数的最优估计方法。

Data shortage is a key factor that restricts scientific flood prevention in the Loess Plateau area of northern Shaanxi.It is of scientific significance to explore the estimation method of quantile of flood characteristic values in data shortage areas.The quantile of the measured maximum daily flow(MDF)of 28 hydrological stations in Yulin is taken as the response variable,and 12 indicators such as catchment area,check dam control area,longitude and latitude of hydrological stations are taken as the explanatory variables.The estimation efficiency of four regional flood frequency analysis models formed by the cross coupling of linear CCA and nonlinear NLCCA homogeneous region identification methods and linear MLR and nonlinear GAM regional hydrological quantile estimation methods is studied,and its applicability in the Loess Plateau of northern Shaanxi is evaluated.The results show that NLCCA-GAM has higher precision and robustness than other coupled models(CCA-MLR,NLCCA-MLR,CCA-GAM)by virtue of its nonlinear advantages,and can be used as the optimal estimation method of quantile of flood characteristic values in data shortage area of the Loess Plateau in northern Shaanxi.