Abstract: Quickly and accurately predicting the diffusion pattern of radioactive substances in the atmosphere after a nuclear power plant accident can provide important reference for relevant departments to make timely scientific decisions on radiation protection. The research takes the hypothetical accident situation of a nuclear power plant in Zhejiang Province, China as an example, and uses WRF and FLEXPART-WRF models for virtual simulation to explore the effects of three cloud microphysical parameterization schemes (WSM3, Thompson, CAM5.1) on the diffusion simulation of radioactive substance Cs-137 during typhoons, and analyze the transport and diffusion characteristics of Cs-137 concentration.. The results are as follows. The simulated wind patterns and precipitation from all three schemes are generally consistent with the empirical observations, with the CAM5.1 scheme achieving the highest overall simulation accuracy. The transport and diffusion profiles of Cs-137 concentrations are heavily influenced by the simulated meteorological conditions, exhibiting significant variations among the different microphysics schemes. The CAM5.1 scheme, in particular, displays the broadest dispersion of near-surface concentrations, corresponding to the widest spread of rainfall. The variations in dry deposition are relatively uniform across all three schemes. Cs-137 concentrations are primarily driven by wet deposition, which peaks sharply within the initial 12 hours of the simulation and then levels off. The settlement peak is highest in the Thompson scheme, followed by the WSM3 scheme, and lowest in the CAM5.1 scheme.