高益辉 钟平安 徐斌 朱非林 曹瀚翔

摘要：实时校正是改善洪水实时预报精度的重要手段。河流系统中多个站点之间具有高度的水力联系，各个站点之间的误差也具有空间关联性。依据河道水流演进的基本方程和动态自适应的自回归方法，建立了考虑误差空间演化的河流系统实时洪水预报误差多点联合校正方法。利用洪峰段洪量误差、洪峰流量误差、纳什效率系数（NSE）和峰滞时间等指标分别对多点联合校正和不考虑误差空间关联性的单点校正开展校正效果评估。以淮河王家坝断面以上为背景开展实证研究，结果表明：考虑河流系统误差空间关联性的多点联合校正效果优于单点校正，洪峰段效果更为显著，能够更有效地提高河流系统洪水预报的精度。

关键词：洪水预报误差；实时校正；马斯京根法；多点联合校正；单点校正；自回归模型

中图分类号：P338 文献标志码：A 文章编号：

16721683（2018）05002106

Study on multipoint joint correction method for realtime flood forecasting errors of river systems

GAO Yihui1，ZHONG Pingan1，XU Bin1，ZHU Feilin1，CAO Hanxiang2

（

1.College of Hydrology and Water Resources，Hohai University，Nanjing 210098，China；2.Taizhou Tian Qin Geographic Information Engineering Co.，Ltd.，Taizhou 318000，China

）

Abstract：

Realtime correction is an important way to improve the accuracy of realtime flood forecasting.There are highlevel hydraulic connections among the multiple stations in a river system，and the errors between the multiple stations are spatially interconnected as well.Based on the basic equations of river flood routing and the dynamic adaptive autoregressive method，this study proposed a multipoint joint correction method for realtime flood forecasting errors with consideration to the spatial evolution of errors.The correction effect was evaluated in terms of the flood volume error during flood peak period，flood peak discharge error，NashSutcliffe efficiency coefficient，and lag time of peak.This paper conducted a case study on the region up the Wangjiaba section of Huaihe River to demonstrate the proposed methodology.The results indicated that the effect of multipoint joint correction was better than that of singlepoint correction，and the effect was even more prominent for the flood peak period.The proposed method can effectively improve the accuracy of flood forecasting of the river system.

Key words：

flood forecasting errors；realtime correction；Muskingum method；multipoint joint correction；singlepoint correction；autoregressive model

由于水文物理過程的复杂性和人类认识水平的局限性，实时洪水预报中不可避免地存在由于数据输入、模型结构和模型参数等不确定性带来的误差[12]。实时校正是减轻不确定性影响，改善实时洪水预报精度的重要措施[34]。

从20世纪60年代开始，国内外学者对洪水预报实时校正方法进行了广泛的研究。Koren V.I.等[5]将广义差分ARMA模型引入预报模型，在多瑙河布达佩斯至巴加河段采取自我校正预报器算法进行水位预报；Wood E.F.[6]在大流域上使用卡尔曼滤波器技术，建议采用分区子系统的处理方式，采用增补噪声过程的技术补偿预报误差，来处理各个子系统之间互相影响的问题；……