基于时变增益水文模型的实时预报研究

Real-time Prediction of Hydrological Model Based on Time-varying Gains

近年来,洪水灾害事件频发,洪水预报是有效预防和抵御山洪灾害的重要非工程手段,为流域防洪预警提供重要决策依据。实时校正可以有效地改善洪水预报精度。以湖北省3个典型中小流域(高家堰、西河驿、渔洋关)为研究区,基于时变增益水文模型,采用具有可变遗忘因子的递推最小二乘法作为校正方法进行洪水预报实时校正研究。结果表明,未校正前,3个流域洪水预报所得纳什效率系数总体均值为0.78。实时校正后,3个流域各自的纳什效率系数均值均达0.90以上,峰现时间和洪峰流量预报合格率均达乙级及以上,预报精度得到明显提升。

In recent years,flood disasters have occurred frequently,and flood forecasting is an important non-engineering way to effectively prevent and resist mountain torrent disasters,and provide vital decision-making basis for watershed flood warning. Real-time correction method can effectively improve the accuracy of flood forecasting. In this paper,three typical small and medium-sized watersheds in Hubei Province( Gaojiayan,Xiheyi,Yuyangguan) are used as research areas. Based on the Time Variant Gain Model,the model prediction results are corrected in real-time by recursive least square method with variable forgetting factor. The results show that the average Nash efficiency coefficient of flood forecasting in three watersheds is 0.78 before being corrected. After a real-time correction,the Nash efficiency coefficients of the three basins are all above 0.90,and the peak time forecast qualification rates and the peak flow forecast qualification rates are all up to Grade B or above. The forecast accuracy has been significantly improved.