基于相似性聚类分析的半干旱流域水文模拟与预报研究  

Hydrological Modeling and Forecasting of Semi-arid Watersheds Based on Similarity Cluster Analysis

作　　者：蒋春源 张汉辰 徐小涵 孙袁媛 JIANG Chun-yuan;ZHANG Han-chen;XU Xiao-han;SUN Yuan-yuan(School of Geography and Planning,Ningxia University,Yinchuan 750021,China)

机构地区：[1]宁夏大学地理科学与规划学院,宁夏银川750021

出　　处：《水电能源科学》2024年第12期1-5,共5页Water Resources and Power

基　　金：宁夏回族自治区重点研发计划项目(2023BEG02054);国家自然科学基金项目(42161008)。

摘　　要：干旱半干旱地区的中小流域降雨特征复杂、下垫面特征多变,水文模拟精度较低。为提高分布式水文物理模型在旱区模拟预报精度,基于宁夏南部半干旱地区黄家河流域1981~2017年水沙数据,考虑前期雨量和最大雨量对超渗地面径流量的影响,以多个时段前期累积雨量特征值、最大雨量特征值和不同洪水重现期为变量进行聚类分析,对洪水进行分类,利用CASC2D-SED模型进行水沙模拟,预报场次通过直接移植各类别模型敏感参数进行模拟预报。结果表明,基于CASC2D-SED模型的黄家河流域水沙模拟精度较好,平均纳什效率系数洪量值为0.74,平均纳什效率系数沙量值为0.67,模型产流模拟精度要高于产沙模拟精度。基于聚类分析后,预报场次直接移植各类别模型参数预报结果较好,平均纳什效率系数洪量值、沙量值精度分别提高了0.07、0.01。Due to the complex rainfall characteristics and variable subsurface characteristics of small and medium-sized watersheds in arid and semi-arid regions,the accuracy of hydrological simulation is low.In order to improve the simulation and forecasting accuracy of distributed hydrophysical model in arid areas,based on the water and sand data of Huangjiahe in the semi-arid region of southern Ningxia from 1981 to 2017,considering the influence of the pre-rainfall and maximum rainfall on the over-permeable surface runoff,this paper carried out the clustering analysis to categorize flooding by taking the characteristic values of the pre-accumulated rainfall in multiple time periods,the characteristic val-ues of the maximum rainfall,and the different flood recurrence periods as the variables.The CASC2D-SED model was used to simulate water-sand.The forecasting field was simulated by directly transplanting the sensitive parameters of each category model.The results show that the water-sand simulation accuracy of Huangjiahe by CASC2D-SED model is bet-ter,with the average Nash-Sutcliffe efficiency coefficient flood value of 0.74 and the average Nash-Sutcliffe efficiency co-efficient sand value of 0.67;The model runoff producing simulation accuracy is higher than the sand-producing simulation accuracy.Based on the clustering analysis,the forecast field directly transplanted the model parameters of each category has better forecast results,and the average Nash-Sutcliffe efficiency coefficient flood volume value and sand volume value accuracy distribution improved by 0.07 and 0.01,respectively.

关 键 词：黄家河流域 CASC2D-SED模型 K-均值聚类 水沙模拟 

分 类 号：TV124[水利工程—水文学及水资源] P333[天文地球—水文科学]