Abstract: The weather forecast produced by numerical weather prediction models has been an important means in the operational forecast, however there exists intrinsic uncertainty. This uncertainty is caused by both the ones in the initial conditions and the ones in the numerical model itself. In the process of the numerical prediction, it is important to consider these errors since the nature of the atmospheric dynamics is to amplify errors originating from either of them. At present, with the application of the massive non-conventional data, such as satellite and radar data in the assimilation system, it is possible to produce the relatively more accurate initial fields. Thus, the uncertainty of numerical model may be more important to forecast especially in the various moist physical process schemes. In recent years, with the ever-improving performance of numerical prediction models and ever-increasing computational resources, the explicit schemes are used to forecast the strong precipitation process in the high resolution mesoscale models gradually with hybrid approach. However, what is the role of explicit schemes and what extension they lead to the Meiyu front heavy rainfall forecast uncertainty? All these issues are worth being further studied. In terms of non-hydrostatic MM5, the impact of explicit moist schemes on the uncertainties in the numerical simulation of the "03·7" Nanjing heavy rainfall is investigated. According to the simulated 24-hour accumulated rain spatial pattern and Ts skill of the variance experiments, the uncertainty depends on the explicit and implicit schemes' relative coordination and sensitivity when the hybrid approach is used to describe the cloud and precipitation process. The implicit schemes determine the spatial pattern of the rain bands, while the explicit schemes adjust the local rain pattern and density. The extent to which the rain band adjusted is related to the convective parameterization. The difference in the rates of resolve scale rainfall to total rainfall in different experiments indicates the uncertainty due to explicit schemes. At the same time, in the center of the simulated strong precipitation region, for the convective parameterization of Grell and KF2, the resolve resolution precipitation of the explicit schemes accounts for the 1/3—2/3 of the total rainfall; for the BM, the unresolved rain dominates the total accumulated rainfall. It indicates that, the explicit schemes play an important role in the precipitation simulation for Grell and KF2 with the hybrid approach, on the other hand, it is inhibited for the BM schemes. Thus, the uncertainties of explicit schemes should be considered when implicit scheme Grell or KF2 is used to simulate the heavy rainfall.