Abstract: In order to investigate influence of different atmospheric forcing data sets on land surface temperature(LST)simulation,we conducted three simulations for the upper and middle reaches of Yellow River Basin by using the CLM 5.0 driven by CMFD,GSWP and CRU-NCEP.The results showed that:(1) CLM 5.0 had good performance for simulation.(2) Simulated summer LST driven by CMFD had the best performance with a low bias of 0.54 ℃,a root mean square error(RMSE)of 0.63 ℃,and a correlation coefficient(R)of 0.95.(3) Winter LST driven by GSWP performed best with a bias of 0.87 ℃,RMSE of 1.24 ℃,and R up to 0.95.Therefore,different atmospheric forcing data sets had a greater impact on LST simulation based on land surface model.It is necessary to select the optimal atmospheric forcing data set to drive the land surface model in the prediction of future climate and hydrological processes.