洪水预报系统中实时校正模型的优选方法

针对实时洪水预报系统中实时校正模型存在一定的适用范围却又难以在线优选的问题,首先分析了自适应滤波模型、马斯京根矩阵模型和反馈模拟模型3种传统实时校正模型的原理与适用性,并编制算法流程图,为代码编程提供依据。其次,提出了一种实时校正模型的优选方法,适用于大流域多校正断面的实时洪水预报系统,可以为各断面在线优选出最优的实时校正模型。在WDS9002三峡实时洪水预报系统中的应用结果表明：优选方法在校正成功率和各项精度评价指标统计上出现最优的频次高于传统实时校正模型,且很少出现指标统计最差的情况;另外,优选方法有可能减小过校正现象的概率。

Optimal Selection Method of Real-time Correction Model in Flood Forecasting System

The real-time correction model （RTCM） has its applicable range in real-time flood forecasting system（RTFFS）, but the model is hard to do on-line optimal selection. For this problem, first the theory and applicability of three traditional RTCMs, including adaptive filtering model, Muskingum matrix model and feedback simulation model are analyzed in the paper. And algorithm flowchart is completed and base for code compiling of each model is provided. Secondary, optimal selection method （OSM） of RTCM is proposed , which is suitable to use RTFFS in large river basin with several correction profiles and to have capability of on-line selecting the optimal RTCM for each correction profile. Application results shows in the WDSg002 Three Gorges RTFFS, the OSM has higher best frequency than other traditional RTCM in correction successful rate and the best accuracy assessment indexes, and rarely emerges the worst accuracy assessment indexes. In addition, OSM can reduce probability of overcorreetion.