大渡河流域径流预报不确定性溯源及降低控制方法

Runoff forecasting uncertainty traceability analysis and control method research of Dadu River Basin

根据流域水文、气象、下垫面等资料及现有模型特点,从模型输入、模型结构及参数、人类活动影响等方面,开展大渡河流域径流预报不确定性溯源分析,识别影响径流预报不确定性的主要因素,并提出了预报不确定性降低控制方法。在模型输入方面,引入气象降雨数据,并结合协-克里金插值处理以提高降雨输入质量;在模型结构及参数方面,针对流域上游存在的融雪径流,构建了新安江-融雪径流预报耦合模型,并依据季节不同采用不同的模型参数;在预报误差校正方面,采用动态系统响应曲线修正方法进行终端预报误差校正。应用结果表明:无论是上游的丹巴流域还是中下游的毛头码流域,通过采用不确定性降低控制措施后,径流深预报合格率达100%,整体预报精度显著提高。

According to the hydrological and meteorological data,underlying surface and other data of the basin and the characteristics of existing models,the uncertainty traceability analysis of runoff forecast in Dadu River Basin is carried out from the aspects of model input,structure,parameters and human activities.The main factors influencing the uncertainty of runoff forecast are identified,and the control method for reducing the prediction uncertainty is proposed.In the aspect of model input,meteorological rainfall data is introduced and combined with co-Kriging interpolation to improve the quality of rainfall input.In terms of model structure and parameters,the Xin’anjiang-snowmelt runoff prediction coupling model is constructed for the snowmelt runoff in the upper reaches of the basin,and different model parameters are adopted according to different seasons.In terms of forecast error correction,dynamic system response curve is used to correct terminal prediction error.The application results show that the qualified rate of runoff forecast reaches 100% and the overall forecast accuracy is significantly improved after adopting uncertainty reduction control measures whether in the Danba Watershed in the upper reaches or in the Maotouma Watershed in the middle and lower reaches of Dadu River.