Abstract: The observation data of Maduo station from June 2015 to August 2018 were used as the forcing field data to drive CLM5.0 (Community Land Model) model, and the soil temperature and humidity changes of Maduo station from June 2017 to August 2018 were simulated by using different soil stratification schemes in the CLM5.0 model, and compared with the observations. The results show: (1) For the simulation of soil temperature, the four soil stratification schemes of CLM5.0 model could well simulate the seasonal variation trend of soil temperature at different depths of Maduo Station in a year, the shallow simulation value was more correlated with the observed value, the temperature change amplitude of the deep soil was relatively small, the curve was smoother, and the 20-layer scheme of the four stratification schemes had the best simulation effect on soil temperature, with an average correlation coefficient of 0.942. (2) For the simulation of soil moisture, the four soil stratification schemes could better simulate the seasonal changes and diurnal trends of soil moisture in each layer, but there were different degrees of deviation from the observed values. The 20-layer scheme had a better simulation effect on soil moisture, and the average correlation coefficient was 0.730.