Abstract: This study aims to evaluate the area precipitation forecasting performance of various numerical models during the2024flood in the Changjinag River Basin, and understand the forecasting capabilities of current operational models.Through TS score and mean absolute error(MAE),we analyze the precipitation forecasting performances of different numerical models(EC,GRAPES-GFS,NCEP,GERMAN,and the Japanese model) during the2024Changjinag River No.1flood and No.2flood.The results indicated that:(1) during the2024Changjinag River No.1flood, the EC model demonstrated the most stable performance in precipitation forecasting, exhibiting the best results, particularly in forecasting heavy rain and rainstorms.The GRAPES-GFS model ranked second, showing good performance, especially in rainstorm forecasts.In contrast, the Japanese model performed poorly across all forecast periods, with TS scores lower than other models, particularly in the48to96-hour forecasts.(2) Additionally, analysis during the2024Changjinag River No.2flood revealed that the GERMAN model performed well in forecasting light and middle rain, but was less effective than the EC and GRAPES-GFS models in forecasting heavy rain and rainstorm.Overall, there were large differences in the precipitation forecasting capabilities of various models facing different magnitudes of rain and regions.A deeper understanding of these inter-model differences can provide important references for future precipitation forecasting efforts.