摘要
为具体考虑各水文模型适用条件、灵活利用模型预报特征对研究流域进行高精度耦合径流预报,将雅砻江流域雅江~吉居区间作为研究对象,以提升径流预报精度和稳定性为首要目的,构建涵盖新安江模型、水箱模型和TOPMODEL模型的多模型径流预报方法库,引入最小二乘法、岭回归法和极端梯度提升树法耦合各模型进行水文预报,并提出多评价指标体系对各耦合方法的预测性能进行对比分析。结论表明,极端梯度提升树法相较于其余两种方法有稳定的预测性能和强大的泛化能力,为该流域其他区间的预报工作提供了新的思路。
In order to flexibly use the prediction features of each independent model to perform high-precision coupled runoff forecasting in the study basin under the premise of considering the applicable conditions of each hydrological model,this paper constructs a multi-model runoff forecasting method library covering Xinanjiang model,Tank model and TOPMODEL model,and introduces least squares,ridge regression and extreme gradient boosting tree methods to couple each independent model for hydrological forecasting.A multi-evaluation met⁃ric system is proposed to analyze the prediction performance of each coupled method.The conclusions show that the extreme gradient boost⁃ing tree method has stable prediction performance and strong generalization ability compared with the other two methods in the runoff forecast⁃ing of the Yajiang-Jiju interval in the Yalong River Basin,which provides a new idea for the runoff forecasting of other areas in the basin.
作者
祝宾皓
周建中
方威
张勇传
ZHU Bin-hao;ZHOU Jian-zhong;FANG Wei;ZHANG Yong-chuan(School of Hydropower and Information Engineering,Huazhong University of Science and Technology,Wuhan 430074,Hubei Province,China;Hubei Key Laboratory of Digital Valley Science and Technology,Huazhong University of Science and Technology,Wuhan 430074,Hubei Province,China)
出处
《中国农村水利水电》
北大核心
2023年第5期119-123,128,共6页
China Rural Water and Hydropower
基金
国家自然科学基金雅砻江联合研究基金重点项目(U1865202)。
关键词
多模型预报
水文预报
极端梯度提升树
岭回归
multi-model forecasting
hydrological forecasting
extreme gradient boosting trees
ridge regression
