A land surface temperature data assimilation scheme was developed on weak-constraint viarational method and simple land surface model，which is mainly used to improve the estimation of the turbulent heat fluxes by assimilating geostationary operational environmental satellite (GOES) retrieved land surface temperature (LST). A variational data assimilation scheme was developed based on the weak-constraint concept. It can estimate both state variables and model unspecified parameters together, which is depend on the building of the cost function. The objective of the variational method is to minimize the cost function to seek the most optimal control variables and accurately estimate sensible heat and latent heat. The GOES LST is compared with the ground measured LST, and the root mean square error (RMSE) was taken as the observation error. The scheme was tested and validated based on measurements in two mainly observation sites of Ameriflux. Results indicate that data assimilation method improves the estimation of surface temperature, sensible heat flux and latent heat flux. The RMSE of estimated LST is around to 1K in both sites. Meantime, the average RMSE of estimated sensible heat and latent heat dropped to 22W/m2 and 26W/m2 respectively. It is a promising way to improve the estimation of turbulent heat fluxes by assimilating GOES LST into land surface model.