Comprehensive Evaluation of Runoff Forecast Level Based on Forecasting Difficulty

ABSTRACT The current traditional forecasting level evaluation method only uses the forecast error value series for relevant analysis, such as dividing the number of qualified forecasts by the total number of forecasts to indicate the corresponding forecasting accuracy or level. This does not consider the different external environments during the forecast. When assessing the forecasting level of different forecasters, fairness cannot be reflected to a certain extent. So this paper puts forward the concept of forecasting difficulty and analyses the physical significance of forecasting difficulty. Based on the traditional concepts of forecasting accuracy and error, two methods calculating the forecasting difficulty are established to consider different forecasting situations, such as rainy, rainless, different foresight periods, and different inflow levels. Further, a new comprehensive forecasting level evaluation method of forecasters is proposed. Taking the Guandi reservoir as an example, the case study results show that compared with the traditional method, the proposed forecasting level evaluation method can effectively consider the difficulty factors of different forecasting situations. In addition, the methods can better reflect the contribution of difficult situations to the comprehensive forecasting level when considering the improvement of forecasting accuracy, which makes the obtained results more scientific and reasonable.