青藏高原广泛分布着季节性冻土和多年冻土，是典型的高山冻土区。受到全球气候变暖的影响，青藏高原冻土正在发生显著退化，进而引发了许多环境生态和水文水资源问题。为了保护青藏高原地区脆弱的生态系统、保障下游地区的用水安全，亟需掌握高原冻土的空间分布特征和时间变化规律。青藏高原地形地貌条件复杂，而且地面冻土和地面气象观测站点稀少。因此，本论文建立了完全由卫星遥感气象数据驱动的土壤冻融过程模型（GBEHM-RS），在此基础上模拟分析青藏高原地区冻土的时空变化及其影响因素。首先，在冻土-生态-水文模型GBEHM的基础上，采用遥感地表温度作为热量传输的上边界条件、基于熵增原理求解地表能量平衡，建立了GBEHM-RS模型。针对青藏高原的下垫面特点，改进了模型中土壤反照率的参数化方案。通过引入太阳天顶角和土壤含水量，以及采用依据土壤组成的参数赋值方法，减小了地表反照率的计算误差，提升了土壤冻融过程的模拟效果。此外，还根据地面观测，分析了地表反照率卫星遥感产品的精度。其次，详细分析和评价了GBEHM-RS模型对于青藏高原地区冻土模拟的适用性。与地面气象数据驱动的GBEHM模型相比，以遥感地表温度为驱动降低了原模型在地表热通量求解中的误差，同时有效避免了地面气象站点插值带来的误差，提高了青藏高原冻土模拟的精度，使模拟结果更好地反映空间分布特征。此外，与以往研究采用的遥感冻土模型相比，GBEHM-RS模型考虑了冻融条件下的土壤水分运动，更好地刻画了土壤冻融过程，因此冻土模拟结果的精度更高。最后，将GBEHM-RS模型用于整个青藏高原，模拟2002~2016年间高原冻土的时空变化，并分析冻土变化的环境影响因子。结果显示，多年冻土的面积为122万km2，季节性冻土的面积为186万km2。由于地表温度在融化季的升温尤为剧烈，青藏高原多年冻土的退化尤为明显，近15年间多年冻土的面积减少了6%，活动层厚度普遍增加；而季节性冻土年最大冻深的变化趋势呈现复杂的空间分布特征。地表温度是冻土变化的主要影响因子，80%以上面积的多年冻土变化和季节性冻土变化主要由该因子控制。此外，降雪和降雨也分别通过积雪和土壤水影响冻土变化，受它们控制的区域面积分别占整个青藏高原的11%和6%。其中，降雪的控制区域主要位于喀喇昆仑山脉、喜马拉雅山脉、念青唐古拉山脉和黄河源等地；降雨的控制区域主要位于柴达木盆地、长江源和黄河源等地。
The Qinghai-Tibetan Plateau is characterized by mountainous frozen region and is covered by permafrost and seasonally frozen ground. Under the global climate warming, the degradation of frozen soils over the plateau region is significant and has triggered a number of ecological and hydrological problems. In order to preserve the delicate eco-environment in the plateau and ensure the water safety in the lower regions, it is imperative to know the spatial pattern and temporal changes of frozen soils over the entire plateau. Since the Qinghai-Tibetan Plateau has a complex topography and landscape, and the ground observatory sites of frozen soil and meteorology are quite sparse, the dissertation aimed to develop a soil freeze-thaw process model fully driven by satellite remote sensing data, and used the model to simulate the spatiotemporal changes of frozen soil over the plateau region and to analyze the relevant influential factors.Firstly, this study developed a soil freeze-thaw process model fully driven by satellite remote sensing data based on a geomorphology-based eco-hydrological model (GBEHM), and the new model is referred to as GBEHM-RS. In GBEHM-RS, satellite remotely sensed land surface temperature was used as the upper boundary condition for thermal transfer equations and the principle of maximum enthropy prodctiuon was used to solve the surface energy balance. This study developed a physically-based parameterization for soil albedo estimation to make it more suitable for the Qinghai-Tibetan Plateau. By introducing the influences of solar zenith angle and near-surface liquid soil moisture in formulation and using soil compositions in parameter estimations, the newly developed parameterization of soil albedo reduced errors in calculated surface albedo. Additionally, the validation results indicated that satellite-based surface albedo obtained reliable accuracy, which was also suitable for frozen soil simulations in the plateau region.Next, this study carefully evaluated GBEHM-RS in simulations of frozen soils on the Qinghai-Tibetan Plateau. Compared with the tradiational simulations driven by ground observed meteorological data, using satellite-based land surface temperature could obviously reduce errors associated with the calculation of surface heat flux and avoid additional errors introduced by the spatial extrapolation of ground meteorological observations. The simulated frozen soils thus had higher accuracy and better reproduced the spatial patterns of ground observed frozen soils. Additionally, those previously developed satellite-based frozen soil models did not consider soil water movement; in comparison, GBEHM-RS fully coupled the heat and water transfers processes to ensure the soil freeze-thaw simulated results having higher accuracy. Finally, this study employed GBEHM-RS to simulate the spatiotemporal changes of frozen soil over the entire Qinghai-Tibetan Plateau during 2002~2016 and to analyze on the influential factors. Results showed that the permafrost covered an area of 1.22 million km2, and the seasonally frozen ground covered an area of 1.86 million km2. Due to rapid rising of land surface temperature during the thawing season, the permafrost degraded evidently during the study period with permafrost area decreasing by 6% and active layer thickness increasing in most places. Land surface temperature is the most important factor in influencing on the changes of frozen soils, dominating in over 80% of permafrost and seasonally frozen ground regions. The snowfall and rainfall influenced on the frozen soils through snow processes and soil water movement, and dominated 11% and 6% of permafrost and seasonally frozen ground regions, repestively. The snowfall dominated area were primarily located on the Karakoram Mountain ranges, the Himalayan Mounatin ranges, the Nyainqentanglha Mountain ranges, and the source region of Yellow River; the rainfall dominated area were primarily located on the Qaidam Basin and the source regions of Yellow and Yangtze Rivers.
