基于洪水分级的青山殿水库洪水预报  

Flood forecast of the Qingshandian Reservoir based on flood-based classification

作　　者：徐志刚[1] 刘风雨 陈瑞刚 杨丽丽 康爱卿[5] XU Zhigang;LIU Fengyu;CHEN Ruigang;YANG Lili;KANG Aiqing(Hangzhou Forestry Water Conservancy Bureau,Hangzhou 310014,China;Yunli Intelligent Technology Co.,Ltd.,Beijing 100037,China;Zhejiang Qiantang River Basin Center,Hangzhou 310020,China;Beijing Heyuan Technology Co.,Ltd.,Beijing 100086,China;China Institute of Water Resources and Hydropower Research,Beijing 100038,China)

机构地区：[1]杭州市林业水利局,杭州310014 [2]云粒智慧科技有限公司,北京100037 [3]浙江省钱塘江流域中心,杭州310020 [4]北京合源科技有限公司,北京100086 [5]中国水利水电科学研究院,北京100038

出　　处：《南水北调与水利科技（中英文）》2024年第4期661-671,共11页South-to-North Water Transfers and Water Science & Technology

基　　金：国家重点研发计划项目(2023YFC3006503);浙江省水利科技计划项目(RA2204)。

摘　　要：为提高洪水预报精度,减小全流域使用单一参数集而产生的预报误差,以青山殿水库为研究对象,将历史洪水分为大、中、小3个等级,采用新安江模型和SCE-UA参数优化算法,基于皮尔逊相关系数分析累计降雨量和实测洪峰流量的相关性,选取相关性最优的最大6 h累积降雨量作为参数应用指标,分级后的大中小洪水选用对应的参数集进行洪水验证及预报。结果表明:未对洪水进行分级时,38场洪水总体合格率为92.1%,平均确定性系数为0.82;按照洪水等级划分大中小洪水后进行洪水过程模拟,大中小洪水场次合格率100%,确定性系数平均为0.92、0.88和0.87,平均确定性系数为0.88,分级后各场次的合格率与确定性系数提高,拟合效果更好;基于皮尔逊相关系数分析洪峰流量与最大1、3、6、24 h降雨量的相关关系,最大6 h的降雨量与洪峰流量的皮尔逊相关系数最高,取预见期内最大6 h累积降雨量作为大中小参数判定条件,在2021年汛期实际预报中,洪水预报结果全部合格,4场大中型洪水拟合程度好,确定性系数高,基于洪水分级的洪水预报精度较高、合理可行,可为水库防洪提供参考。To enhance the accuracy of flood forecasting and mitigate prediction errors resulting from the utilization of a single parameter set across the entire hydrological model basin,various corresponding parameters were employed in distinct precipitation scenarios.The research focused on Qingshandian Reservoir,where historical floods were categorized into three levels:large,medium,and small.The Xin'anjiang model and SCE-UA parameter optimization algorithm were employed to investigate the correlation between cumulative rainfall and measured flood peak flow,utilizing the Pearson correlation coefficient.The parameter application index was determined as the maximum 6 h cumulative rainfall exhibiting the strongest correlation.Flood verification and prediction were subsequently performed utilizing the corresponding parameter set.The findings indicated that when floods were not classified,the overall pass rate for 38 floods stood at 92.1%,with an average deterministic coefficient of 0.82.Following the classification of floods into large,medium,and small categories,flood process simulation was conducted.Consequently,the pass rate for large,medium,and small floods reached 100%,accompanied by average deterministic coefficients of 0.92,0.88,and 0.87,respectively,resulting in an overall average deterministic coefficient of 0.88.The classification demonstrated enhancements in both the pass rate and deterministic coefficient for each individual flood,contributing to an improved fitting effect.Furthermore,an analysis was conducted on the correlation between flood peak flow and maximum rainfall within 1 h,3 h,6 h,and 24 h,based on the Pearson correlation coefficient.Results revealed that the maximum 6 h rainfall exhibited the highest Pearson correlation coefficient with flood peak flow.Accordingly,the maximum 6-hour cumulative rainfall during the forecast period was employed as the criterion for determining large,medium,and small parameters.In the actual flood forecasting during the 2021 flood season,all flood forecast results were de

关 键 词：青山殿水库 新安江模型 洪水等级划分 相关分析 皮尔逊相关系数 

分 类 号：TV697.1[水利工程—水利水电工程]