抗差修正在白水坑水库实时洪水预报中的应用  

Application of Robust Correction Method for Real-time Flood Forecasting in Baishuikeng Reservoir

作　　者：王旭滢 包为民[1] 王玉丽 游洋[3] 钟华 

机构地区：[1]河海大学水文水资源学院,南京210098 [2]中国电建集团中南勘测设计研究院有限公司,长沙410014 [3]江苏省水利厅水资源处,南京210029 [4]南京水利科学研究院,南京210029

出　　处：《三峡大学学报（自然科学版）》2017年第6期1-4,共4页Journal of China Three Gorges University：Natural Sciences

基　　金：国家自然科学基金面上基金(41371048/51479062);中央高校基本科研业务费专项资金资助(2015B14314);国家重点研发计划专项经费资助项目(2016YFC0402703)

摘　　要：在水库洪水预报中,由实测库水位、出库流量、水位库容曲线反推后得到的水库入库流量曲线多呈现"锯齿"状,带有波动误差,这种是不服从正态分布的粗差,因此无法通过分析其分布情况对预报值进行修正,给水库预报系统造成了不良影响.为此,尝试将抗差理论引入水库入库洪水预报实时校正中.本文以浙江白水坑水库为例,选择三水源新安江模型、马斯京根河道汇流模型,对白水坑水库入库洪水进行模拟,对实测入库洪水抗差修正后再进行实时校正.结果表明,抗差能够有效阻止非正常信息进入预报系统,从而提高洪水预报精度.The input flow curve of a reservoir calculated by observed water stage, outflow and stage-storage relation curve is usually in a serration shape. In this process, there are fluctuating errors which have no nor- mal distribution characteristics. They are called coarse errors and cannot amended by analyzing their distribu- tion which could be brought into the flood forecasting system. In this paper, the robust correction method is introduced to combine with real-time correction method. Based on the Xin＇anjiang model and Muskingum Riv- er Confluence model, Baishuikeng reservoir flood forecasting system is set up. Eight floods between 2010 and 2012 are selected to do the real-time correction on the basis of the robust correction method. The results show that the robust correction can improve the accuracy of flood forecasting.

关 键 词：抗差 洪水预报 实时校正 白水坑水库 

分 类 号：TV124[水利工程—水文学及水资源]