A one dimensional (1 D) heat diffusion lake model has been successfully coupled into the Weather Research and Forecasting (WRF) model version 3.2. The ability of simulating lake surface temperature of the coupled model (WRF Lake) was first tested in Taihu Lake using observations obtained from Meiliangwan Station from August 11 to 28, 2010. Three experiments were conducted to evaluate the performance of this coupled model. The three experiments include: (1) simulated by original WRF model without lake scheme (WRF); (2) simulated by coupled model without parameter optimization (WRF_Lake_Old); (3) simulated by coupled model with parameter optimization (WRF_Lake_New). Results show that the coupled model with parameter optimization (WRF_Lake_New) has greatly improved the lake surface temperature (LST) simulations when compared with that simulated by WRF. It is found that LSTs for the Meiliangwan Station simulated by WRF_Lake_New show good agreement with observations, WRF_Lake_New can generate realistically daily variation of LST, and the averaged LST bias range reduces to ±1℃. The averaged bias for 2 meter high air temperature is within the range of ±0.5℃. With the improvement of LST and 2 meter high air temperature simulations, the simulated sensible and latent heat flux are both much closer to observations than WRF simulations. These results show that the coupled model (WRF_Lake) has provided a good tool for the study of lake processes and lake atmosphere interactions in the future.