Accurate simulation of thermal and hydrological soil processes is highly important for studying cold region climates. In this paper, two thermal conductivity parameterization schemes for soil proposed by Johansen and Luo were combined with a groundwater parameterization scheme (JN and LN schemes) to modify the soil water and heat parameterization schemes in the Community Land Model (CLM3.5). Land surface processes in the Central Tibetan Plateau were simulated based on the original scheme (Farouki scheme), JN scheme, and LN scheme using the atmospheric forces from meteorological observations at Nameqie, and they were compared with observed data in the same period. We found that all three schemes underestimated the upward shortwave radiation and overestimated the upward longwave radiation and net radiation, but the predictions from the modified scheme better agreed with the observations. The three schemes simulated the average soil temperatures higher than the observation data. The modified schemes, particularly the LN scheme, resulted in better predictions of the soil temperature compared with the original CLM3.5. More thawing days were simulated by three schemes. The observed freezing and thawing proceeded from the upper layers to the lower layers; however, the frozen soil in the Farouki and LN schemes began to thaw in the upper layers and below ~160 cm. The differences in the simulated soil thermal conductivity among the three schemes increased with depth. The soil thermal conductivity simulated by the Farouki scheme was the highest, while the smallest values were generated by the JN scheme. The soil volumetric water content of the Farouki scheme below a 100 cm depth was obviously larger than the observed data. After considering water exchange between the soil column and its underlying aquifer, the JN scheme and, particularly, the LN scheme, simulated the soil volumetric water content slightly higher than but similarly to the observed data. [ABSTRACT FROM AUTHOR]