Current land surface models still have difficulties with producing reliable surface heat fluxes and skin temperature (Tsfc) estimates for high-altitude regions, which may be addressed via adequate parameterization of the roughness lengths for momentum (z0m) and heat (z0h) transfer. In this study, the performance of various z0h and z0m schemes developed for the Noah land surface model is assessed for a high-altitude site (3430 m) on the northeastern part of the Tibetan Plateau. Based on the in situ surface heat fluxes and profile measurements of wind and temperature, monthly variations of z0m and diurnal variations of z0h are derived through application of the Monin–Obukhov similarity theory. These derived values together with the measured heat fluxes are utilized to assess the performance of those z0m and z0h schemes for different seasons. The analyses show that the z0m dynamics are related to vegetation dynamics and soil water freeze–thaw state, which are reproduced satisfactorily with current z0m schemes. Further, it is demonstrated that the heat flux simulations are very sensitive to the diurnal variations of z0h. The newly developed z0h schemes all capture, at least over the sparse vegetated surfaces during the winter season, the observed diurnal variability much better than the original one. It should, however, be noted that for the dense vegetated surfaces during the spring and monsoon seasons, not all newly developed schemes perform consistently better than the original one. With the most promising schemes, the Noah simulated sensible heat flux, latent heat flux, Tsfc, and soil temperature improved for the monsoon season by about 29%, 79%, 75%, and 81%, respectively. In addition, the impact of Tsfc calculation and energy balance closure associated with measurement uncertainties on the above findings are discussed, and the selection of the appropriate z0h scheme for applications is addressed.