Abstract: At present, the one-dimensional Freshwater Lake Model (FLake) is widely used in various climate models and reanalysis data algorithms around the world, but it is not effective in the simulation of lake temperature of Ngoring Lake, a typical lake in Qinghai-Tibet Plateau, mainly due to the friction velocity parameters in the calculation of sensible heat and latent heat flux in the model, and it has been found that after replacing the friction velocity parameter calculated by Community Land Model version 4.5 (CLM4.5) into FLake, FLake simulation effect improved. Therefore, according to the friction velocity parameterization scheme of CLM4.5 lake mode, the FLake friction velocity parameterization scheme was improved from three aspects: Richardson number calculation formula, layer stability condition and friction velocity calculation formula, and then the data of Ngoring Lake observatory was used as the driving field, the MODIS surface temperature data and the actual observation data of the station were used to evaluate the improved simulation effect of FLake. The results show that the improved FLake could better simulate the thermal status of the lake, especially in the cooling stage after the lake temperature reached the maximum each year. The simulated lake surface temperature, lake temperature at different depths of lake and net radiation were closer to the actual observations. Compared with the heating period, the simulated values of sensible heat and latent heat flux increased in the cooling period, the corresponding temperature simulation value was lower than that before the improvement, and closer to the actual observation temperature. The simulation accuracy of lake temperature and net radiation during the cooling period was improved significantly.