Using ERA5 reanalysis data as the initial field, the WRF model was used to conduct a numerical simulation study of a large-scale snowstorm event from April 19–20, 2020. This work adopted different microphysical parameterization schemes for sensitivity experiments, and the capability of the model for simulation was evaluated based on observation data (precipitation data collected at automatic weather stations, radar base data). Temporal and spatial evolution characteristics of precipitation, radar reflectance, dynamic thermodynamics, and water condensate in heavy snow weather were also analyzed. Results reveal that the Morrison scheme can better simulate the snowstorm weather event, which shows that the simulated radar echo intensity, range, and shape are more consistent with the observation data, and the correlation coefficient and root mean square error of the simulated precipitation are better than other schemes. The detailed microphysical structure of the proposed scheme is characterized by a strong ascending motion and long-term maintenance of positive vorticity in the lower layer, more ice crystals in the upper layer above 7 km, and less graupel and rain particles in the middle and lower layers. From the perspective of the thermal dynamic field, there is an obvious vorticity wave train below the height of 600 hPa in the bin scheme. This is mainly because the bin scheme grades the particle swarm, does not bind different particle types to move, and can describe the sinking and dragging effect of different particles in a more detailed way. Cloud microphysical characteristics show that the specific masses of snow, graupel, cloud water, and rain particles simulated by different schemes are close to each other. However, the simulation of the specific mass of ice crystals has great differences in both magnitude and distribution range, which determine the difference in the magnitude and the phase state simulation of the radar echo and precipitation by different microphysics schemes.