Research On Medium-and Long-term Ensemble Streamflow Forecasts Based On ECMWF System 4 Product

Ensemble forecasts depict the statistical distribution of the forecast quantity and describe the uncertainty of the forecast information.With the development of meteorological ensemble forecasting,using meteorological forecasting information for hydrological ensemble forecasting has become the development direction of hydrological prediction research and application.According to the results of the ground verification of the weather forecast,the ECMWF(European Centre for Medium-Range Weather Forecasts)meteorological ensemble forecast data has a better forecast effect in China,but its effect in the application of hydrological forecasting is still not systematically evaluated.The paper evaluates the streamflow forecast performance of ECMWF System 4 meteorological products,and conducts research on the correction method of medium and long-term streamflow forecast.The main research and results are as follows:Based on the ground measured precipitation data,the ECMWF System 4 precipitation forecast data were corrected by the quantile mapping method.Futhermore,based on the biascorrection ECMWF System 4 precipitation forecast data and the VIC distributed hydrological model,the medium-and long-term ensemble streamflow forecasts is carried out in China.Using the simulated value of the VIC model as a benchmark,the skill of ensemble streamflow forecast was evaluated.The forecasts accuracy has a significant relationship with lead time and the size of the basin area.The results showed that the prediction accuracy changed slightly from one months to two months lead times.However,when the forecast period was extended to three months,the prediction accuracy decreased significantly.On the other hand,with the increase of the basin area,the forecast accuracy showed an increasing trend.The skill of the forecasts in different hydroclimatic regions in China was evaluated.The results show that the forecast skill is higher in the Songhua River basin,the upper Yellow River basin,the middle and lower Yellow River basin,the upper Yangtze River basin(I),the upper Yangtze River basin(II),the middle Yangtze River basin,and the lower Yangtze River basin.In addition,the forecast skill is lower in the Haihe basin,the northeast part of Pearl basin,the southwest of the Southeastern River basin,the southwest rivers basin(I),and the southwest rivers basin(II).A medium-and long-term ensemble forecast model at the basin scale was established which depended on the THREW hydrological model and ECMWF System 4 meteorological ensemble forecast data.The investigation is conducted through the case study of the Upper Hanjiang River Basin and the Middle Yalong River Basin.Comparing the ensemble streamflow forecasts which is based on historical meteorological data and ECMWF System 4 meteorological forecast data.The result showed that the accuracy of streamflow forecast based on ECMWF system 4 is higher than that of streamflow forecast based on historical meteorological data.The effects of bias-corrected precipitation and bias-corrected streamflow on ensemble forecasting are analyzed.The results show that the bias-corrected precipitation forecast data can reduce the uncertainty of ensemble forecast better,and biascorrected streamflow is more effective for ensemble forecast accuracy improvement.Futhermore,a medium-and long-term streamflow forecast model is proposed,which is based bias-corrected precipitation and streamflow forecasts.This model is carried out in the Upper Hanjiang River Basin and the Middle Yalong River Basin.The results show that the Nash efficiency coefficient of more than 0.71 in the Upper Hanjiang River Basin and more than 0.83 in the Middle Yalong River Basin,the average relative error within 0.03 in the Upper Hanjiang River Basin and 0.01 in the Middle Yalong River Basin in one month lead time.