长江源多年冻土区土壤水热传输过程机理与模拟，是广泛关系到高原生态环境保护恢复和区域水文过程的关键科学问题。因此，以GEOtop模型为研究平台，以长江源不同植被盖度下（裸地、30％、65％和92％）高寒草甸的观测数据为基础，检验模型对土壤水热迁移过程的描述与模拟精度。总体而言，GEOtop模型需要率定的参数较少，从而减少模型模拟的不确定性，提高了模拟精度。对不同植被盖度下土壤温度、水分和实际蒸散发模拟的NSE系数基本达到0．8，表明模型能准确模拟高寒生态系统土壤的水热传输过程，可以作为长江源区土壤水热过程的有效模拟工具。
 

The water and energy balance transmission mechanism and process simulation of Yangtze River Headwater region （YRHR）, are widely related to land surface process, ecological environment protection, frozen soil engineering, and regional hydrological processes. Due to the great sensitivity to climate change in source area, YRHR is regarded as a hot climate change research area. GEOtop model is a distributed hydrological model with coupled water and energy budget. The model domain consists of a soil profile of specified depth to the ground surface, the heat and subsurface water flow equations are then solved with finite differences schemes. The heat equation is solved as a one-dimensional form, while the equation describing the water flow in the soil is solved with a fully three-dimensional way. GEOtop model allows a complete coupled description of vertical and lateral flow. The determination of heat flux exchanged from the atmosphere to the ground surface is very important, which constitutes the upper boundary condition of the heat equation. Therefore, GEOtop model is used to model the water and heat transfer processes under bare land, 30°, 65° and 92°vegetation coverage. The model was calibrated using data in situ observations at local scale for. soil temperature profiles, soil moisture profiles, and evapotranspiration under different vegetation coverage. The results show that, the simulation of soil temperature under different vegetation coverage at different depth was ideal, especially for ground surface soil temperature of bare land. The Nash-Sutcliffe （NSE） coefficients of soil temperature were all larger than 0.9. Compared with soil completely melted periods, the modeling deviation of soil temperature in soil completely frozen periods was larger. Long-term field monitoring found that snow cover in the study area was thinner and melt in fast time, therefore snow module was not contained in our modeling. However, the snow melt and cover for a short time will cause soil temperature fluctuation. In addition, the NSE coefficients of soil moisture were around 0.8, indicating that model can preferably simulate water transfer processes in the active layer soil. The simulation deviation of soil moisture at soil completely melted periods was big. This is because of model not including plant growth module and assuming that the vegetation parameters fixed. Moreover, the modelling of actual evapotranspiration under different vegetation coverage also got desired result. The actual evapotranspiration was overestimated from March to June and from August to September. They were due to freeze-thaw cyle of daily temperature flucuation and vegetation growing, respectively. All of them will change the albedo and roughness of groundsurface, thus affecting the water and heat transfer processes. In a word, the model is able to predict soil freezing/thawing process, soil temperature, unfrozen soil moisture and actual evapotranspiraion under different vegetation coverage well. Because of fewer parameters needing to rate, thus it reduced the uncertainty of model simulation and improved the simulation accuracy. The GEOtop model shows promising performance in simulating hydrological processes in cold regions. This model can be applied to improve our understanding of the water and heat transfer processes in cold regions.