摘要
为优化流域径流预报方案,引入数值降雨预报信息,对比分析了不同预见期下新安江-融雪径流预报模型(XAJ-DDF)、基于总误差分析途径的概率预报模型(HUP)、多因子最近邻抽样回归模型(NNBR)在大渡河上游丹巴断面的日径流预报精度。结果表明,汛期,HUP模型表现最优且稳定,3、7 d预见期NS分别大于0.9、0.7,且伴随推荐值Q_(50)的90%置信区间信息为决策人员明晰洪水风险提供依据;枯期,多因子NNBR模型预报精度最高,对退水特征把握更准确。通过模型对比,可进一步提升流域预报水平,为同类研究提供应用参考。
In order to optimize the runoff forecasting scheme in Dadu River Basin above Danba section,the forecasting accuracy of Xin anjiang-snowmelt runoff forecast(XAJ-DDF)model,runoff probability forecast model based on the total error analysis approach(HUP)and multi-factor nearest neighbor bootstrapping regressive(NNBR)model on daily runoff are compared by introducing numerical rainfall forecast information.The results indicate that,(a)in flood season,the HUP model has the best and stable performance,the Nash coefficients of forecasting the third day and seventh day are greater than 0.9 and 0.7 respectively,and the information contained in the 90%confidence interval provides the basis for decision-makers to clarify the flood risk;and(b)in dry season,the prediction accuracy of multi-factor NNBR model is the highest,and the characteristics of runoff recession can be better simulated.By model comparison,the runoff forecasting skill can be further improved,which can provide reference for similar researches.
作者
牟时宇
朱艳军
杨冬梅
曲田
MOU Shiyu;ZHU Yanjun;YANG Dongmei;QU Tian(Dadu River Hydropower Development Co.,Ltd.,Chengdu 610041,Sichuan,China)
出处
《水力发电》
CAS
2022年第5期27-32,共6页
Water Power
基金
国家自然科学基金资助项目(41730750)
国能大渡河流域水电开发有限公司科技项目(PDP-KY-2019-001、DDH-KY-2019-002)。
关键词
径流预报
数值降雨预报信息
新安江模型
概率预报模型
多因子最近邻抽样回归模型
大渡河上游
runoff forecasting
numerical rainfall forecast information
Xin anjing model
runoff probability forecast
multi-factor nearest neighbor bootstrapping regressive model
upper reaches of Dadu River Basin
