<p>Wind farms, as an important renewable energy source to combat climate change, have had explosive development in recent years. Assessing impacts of wind farms on atmospheric and marine environments requires an accurate parameterization of wind farms in atmospheric models. The current wind farm parameterization scheme (Fitch et al. 2012) in WRF plays an important role in the study of impacts of wind farms. The scheme, however, has some shortfalls, e.g., does not consider the wind wake behind turbines inside a grid cell. In this research, the Fitch scheme in WRF is modified by inclusion of the wake effect of wind turbines. Based on an engineering wake model of a turbine, a wake superposition coefficient and an angle correction coefficient are proposed. A solution model for the inflow wind speed is established to obtain the angle correction coefficient. Other coefficients in the engineering wake model are calculated based on the CFD results. These coefficients are added in the WRF to improve the wind farm parameterization, and sensitivity experiments are conducted. Model results show that the new improved scheme significantly increases wind energy, output power and turbulent kinetic energy in the wind farm area compared with the original scheme. Sensitivity experiments also reveal that, with enlarged model grid size and shortened turbine spacing, the subgrid wake effect becomes more significant, and the new scheme shows more advantages.</p>