Refined Evaluation of Numerical Model Forecasts for the 2023 Continuous Rain in the Wheat-planting Region of Henan Province

Ma Yunqi; Li Han; Lyu Linyi

doi:10.16765/j.cnki.1673-7148.2024.01.008

Abstract：Based on precipitation observations from 2706 automatic weather stations in Henan Prov-ince and forecast products of the European Centre for Medium-Range Weather Forecasts(ECMWF),the Meso-scale Weather Numerical Prediction System of China Meteorological Administration(CMA-MESO),and the Meso-scale Numerical Prediction Model of the East China Regional Meteorological Center,China Meteorological Administration(CMA-SH9),we test and evaluate the hourly rainfall forecast effect of the numerical models for the continuous rain process that occurred in the critical period of wheat harvest in Henan Province from May 25 to June 5,2023,from the aspects of TS score and BIAS score of precipitati-on with different magnitudes,the spatial distribution characteristics of errors,and the daily variation characteristics of typical regional forecast deviations.The results show that ECMWF performs well in pre-dicting rainfall Rh≥0.1 mm/h,CMA-SH9 performs better for Rh ≥ 2 mm/h and Rh ≥ 5 mm/h,but CMA-MESO performs poorly in the precipitation prediction.For Rh≥2 mm/h,CMA-SH9 makes good prediction for the morning and night precipitation,while ECMWF does better in predicting the noon to night precipitation.All the models have the significant biases in predicting large-value zones of precipita-tion.The average effective precipitation frequency forecast by CMA-SH9 in the first stage of rainfall is the closest to the observations,especially in the western mountainous areas.ECMWF's forecast for the sec-ond-stage precipitation is more realistic.Although the average deviation of hourly precipitation and hourly precipitation intensity by all models are large during the continuous rain process,there are no distinctly stronger or weaker characteristics.As for the proportion prediction of effective precipitation frequency,ECMWF and CMA-MESO show significant characteristics of more or less at most times,while the CMA-SH9 forecast is much closer to the observation.