Performance Evaluation of the Forecasting of Areal Rainfall for the Haihe River Basin during the Flood Season

Based on observations, forecasts for objective rainfall were obtained from NGWD (National Guidance Weather Forecast), ECMWF (European Centre for Medium-range Weather Forecasts), and CMA_BJ, while forecasts for subjective rainfall were obtained from the Haihe River Basin Meteorological Center (LYK), the forecasting performances of the three prediction models for objective rainfall and a subjective rainfall mentioned above were evaluated for areal rainfall in the Haihe River basin using mean absolute error, fuzzy grading, percentage correct, threat score, missing ratio, and false alarm ratio. The evaluation results are as follows: (1) The performances gradually declined with the extension of the period of validity (i.e., 24 h, 48 h, and 72 h). (2) ECMWF demonstrated an overall better performance than the other models, followed by LYK, NGWD, and CMA_BJ. (3) An analysis of typical rain events revealed that the grades of areal rainfall forecasts through ECMWF, LYK, and NGWD exhibited good predictive quality with a forecasting time of 24 h. However, the forecasting results generated by CMA_BJ exhibited remarkable advantages in predicting the level of heavy precipitation processes with lower forecast level errors, and more close proximity to the observation results to a greater extent..