Abstract: When a radioactive material leakage accident occurs in a nuclear power plant, the timely forecast of the distribution and concentration of pollutant will provide important technical support for the formulation of emergency plan. While solving the transport and diffusion simulation of pollutant in the atmosphere, the boundary conditions, source terms and other physical parameters may change, but the mathematical form of the control equation is constant. The reduced basis finite element method was used to regard these changing conditions as parameters, and the basis functions of solution space were constructed by using a few representative classical finite element solutions. The system was divided into parameter-related part and parameter-independent part by affine decomposition, which greatly reduced the difficulty of assembling system matrix and load vector. The experimental results show that the solution speed of the online stage is increased by about three orders of magnitude. Therefore, this method can be applied to near real-time simulation of pollutant diffusion.