Based on Subjective and Objective Circulation Classification, a framework of method for object-based diagnostic evaluation (MODE) is developed for numerical forecast of heavy rainfall, and this framework is used to verify the heavy rainfall forecast by the global forecast model of the European Center for Medium-Range Weather Forecasts (ECMWF) and the regional mesoscale forecast model of the China Meteorological Administration (CMA_MESO) in Northeast China during the warm season of 2019. The results show that 54 heavy rainfall days in Northeast China in the warm season of 2019 can be divided into trough pattern (P1), Western Pacific Subtropical High (WPSH) pattern (P2), jet pattern (P3), western Northeast China Cold Vortex (NCCV) pattern (P4), and Eastern NCCV pattern. Among the 5 synoptic patterns above, P1 and P3 are dominated by regional heavy rainfall, and the numerical model has strong predictability for the occurrence of heavy rainfall with higher Threshold Score (TS). The heavy rainfall in P4 and P5 is locality, and the numerical model has poor predictability with lower TS. The P2 is also dominated by regional heavy rainfall. However, the forecast deviation for the location, intensity, area of the heavy rainfall is relatively larger, and the TS is also lower. In addition, from the comparison of CMA_MESO and ECMWF, CMA_MESO is generally stronger than the actually intensity and area of heavy rainfall. The TS and false alarm rate (FAR) of CMA _ MESO for heavy rainfall are both generally higher than that of ECMWF. The forecast deviation of CMA_MESO has less consistent , and the predictability of CMA_MESO is generally lower.