新丰江水库流域相似洪水识别及洪量预报

Similar Flood Identification in Xinfengjiang Reservoir Basin and Flood Forecast

采用统计方法进行相似洪水的识别,挖掘流域历史洪水数据,可用于延长洪水预报预见期,从而为水库预泄调度工作提供指导。研究分析影响洪水形成的流域产流特性、气象成因和季节性因素等,经过指标聚类筛选出累积降雨量、洪水起涨时间、前期平均流量作为相似指标归类历史洪水,使用灰色关联度优选最相似洪水并进行洪量移用。以新丰江水库流域为对象,在212场洪水事件中选取100场洪水对所提方法进行验证,结果表明:除小量级和少数大量级的洪水,洪量预测误差整体偏小,有59场洪水的预报洪量误差小于30%,集中于"龙舟水"时期的第二类洪水,其整体误差在三类中最小。该方法有助于提前预判入库洪量,对华南地区多年调节水库的洪量预报调度有重要的借鉴意义。

The identification of similar floods by statistical methods and the mining of historical flood data in the river basin can be used to prolong the forecast period of floods and thus provide a guidance for the pre-discharge operation of reservoirs.In this paper,watershed runoff characteristics,meteorological causes and seasonal factors that affect flood formation are analyzed.Cumulative rainfall,flood rise time and early average flow are screened out through index clustering to classify historical floods as similarity indexes,and most similar floods are optimized with grey correlation degree and the flood volume is transferred.In Xinfengjiang Reservoir watershed as an object,games during the 212 flood events of 100 flood to validate the proposed method,the results show that in addition to the small scale and a handful of great magnitude of floods,the prediction error of the actual overall slants small,with 59 of flood forecast error is less than 30%of the actual,focuses on the“dragon boat water”period of the second category of floods,the least prediction error in three categories.This method is helpful to predicting flood inflows in advance and has important reference values for flood forecast and operation of multi-year regulated reservoirs in South China.