Abstract: The first-generation rapid refresh assimilation numerical forecast system of North Meteorological Center (NMC-RAP) and the upgraded second generation system (NMC-HRRR, the High-Resolution Rapid Refresh assimilation numerical forecast system of North Meteorological Center) have been independently developed by the Meteorological Center of North China Regional Air Traffic Management Bureau of CAAC. The radar reflectivity forecasts of two generation systems are evaluated comprehensively for three large-scale thunderstorms in North China during June to August of 2020 and 2021, in terms of the spatial distribution, Aviation Thunderstorm Nowcasting System (ATNS) score, and severe convective warning of major airports. Results show that the spatial distribution characteristics are well captured by both forecast systems, and the intensity is also closer to the composite radar reflectivity observations. NMC-HRRR always performs better in critical forecasts, such as the coverage of strong reflectivity echo, which is more consistent with the observations. For the two generation systems, the ATNS scores of weak echoes are better than strong echoes. In general, the strong and weak echo scores of NMC-HRRR forecasts within 0−9 h are significantly better than NMC-RAP, especially for the strong echoes in 0−360 min and at the 540th minute. NMC-HRRR shows better forecasting in strong and weak echoes, and the timing when the echo that affects the flight. The total counts of occurrences of severe convective warning from five major airports quantitatively prove that NMC-HRRR shows effective improvements over NMC-RAP to forecast strong and weak echoes with better accuracy in time and space.