寒区土壤水文过程及其对气候变化的响应是水文学、土壤学和冰冻圈科学的前沿问题。寒区土壤水文过程作为寒区水文循环的重要部分，对地表水文过程和地-气之间的能量交换有着重要影响。尤其，由于寒区广泛分布着季节冻土，使得寒区土壤水文过程更加复杂多变，其内涵是冻融循环影响下的土壤水分运移和热量传输（即水热耦合过程）。因此，亟需深入研究冻融循环影响下的寒区土壤水热传输过程。然而，寒区自然环境恶劣、人迹罕至，野外观测研究在观测范围、分辨率和时序性等方面均存在较多困难。在获取一定的观测数据的基础上，构建基于寒区土壤水文物理机制的数值模型，并开展相应的数值模拟研究，已成为认识和理解寒区土壤水热传输过程，揭示其动力学机制和影响因素的有效工具。寒区土壤水热传输过程的物理机制及其复杂，兼具土壤水分多相流动（液态水、固态冰、空气和土壤颗粒四相共存）、冰水相变和土壤高度异质性等过程及特性，为建模和数值模拟带来了极大挑战。本论文针对寒区土壤水热传输过程的关键科学问题，基于寒区土壤水文物理过程，采用计算流体力学（Computational Fluid Dynamics, CFD）方法，自主研发了高分辨率、高计算效率的寒区土壤水热耦合模型darcyTHFOAM；在对模型进行了严格的基准性验证后，在黑河流域上游源区开展了数值模拟研究，在不同尺度（站点、坡面、流域）阐明了季节冻土内水分多相流动、冰水相变和热量传输的动力学过程，揭示了季节变化、坡向和微地形对寒区土壤水热传输过程的影响机制。研究成果为认识和理解寒区土壤水文过程提供了科学依据和科技支撑，主要结论包括：

（1）基于寒区土壤水文物理机制，耦合了季节冻土内水分多相流动、冰水相变和热量传输等关键过程，构建了寒区土壤水热耦合数值模型darcyTHFOAM；基于开源CFD研发环境OpenFOAM，自主研发了可用于模拟寒区土壤水热耦合过程的求解器；采用Qt设计器及Python编译器，研发了相应的具有可视化界面的模拟软件。

（2）采用了一系列从简单到复杂的基准测试算例，包括经典传热方程的解析解、基准测试算例和室内土壤冻结实验等，对darcyTHFOAM模型进行了系统的验证评估。结果表明，darcyTHFOAM模型对土壤温度和液态含水量的模拟结果，与解析解、同类模型模拟结果和实验数据相比，均具有高度的一致性，从而证明了darcyTHFOAM模型的准确性和较好的计算性能。

（3）在站点尺度上，在黑河流域上游的阿柔冻融观测站，采用自主研发的寒区土壤水热耦合模型darcyTHFOAM，模拟了站内观测点从2013至2017年的土壤水分和温度变化。结合站内观测点不同深度（10-120 cm）的土壤实地观测值（包括土壤温度和液态含水量），定量评价了darcyTHFOAM模型的模拟结果。结果表明，darcyTHFOAM模型针对土壤温度和水分的模拟结果与实地观测数据具有非常好的一致性，且精细刻画了土壤水分和温度在冻融循环影响下的季节性变化；尤其，土壤温度的变化呈现出正弦函数的趋势，而浅层土壤温度和水分对季节变化的响应更加敏感。

（4）在坡面尺度上，以黑河流域上游阿柔冻融观测站周边的阳坡和阴坡为研究对象，结合野外观测数据，采用darcyTHFOAM模型对阿柔阳坡和阴坡的土壤水热传输过程进行了数值模拟研究。研究结果不仅揭示了土壤温度、液态水储量的时空变化规律，还给出了土壤固态冰储量的季节性变化。基于研究结果，重点分析探讨了局部地形因素，即坡向，对土壤水热传输过程的影响。具体而言，阴坡比阳坡更寒冷、更潮湿、解冻时间更晚、冻结稳定期（CFP）持续的时间更长。

（5）在流域尺度上，从黑河流域上游的冰沟小流域中，选取了部分区域（冰沟小区域）为最终的研究对象，借助于数字高程模型（DEM）建立了基于真实地形的三维模型。首先验证了模型对于三维真实地形条件下水热传输过程数值模拟的适应性，其次研究了微地形对冻土区土壤水热传输过程的影响。研究结果表明，darcyTHFOAM模型在三维流域尺度上表现出了较高的准确性和较强的适应性，能够满足大区域、长时间数值计算需求。此外，通过将三维实际模型和理想模型的研究结果进行对比，重点分析了微地形对寒区土壤水热传输过程的影响。具体而言，忽略微地形的影响将无法准确描述计算区域侧向流动和传热过程，导致模拟的土壤温度最大低估8.17℃，土壤液态含水量最大低估15.03%。

         Under the seasonal and climate change, the thermo-hydrologic (TH) processes in frozen soils is a frontier research topic in hydrology, soil science, and cryosphere science. As a critical component of the hydrologic cycle in cold regions, the TH processes in frozen soils significantly influences the surface hydrologic process and energy exchange in the land-atmosphere continuum. Therefore, it is urgent to understand the complex TH processes in frozen soils under the freeze-thaw cycles. Due to the harsh natural environment, it is difficult to obtain long-term and high-resolution datasets in field observations. Based on the limited observation datasets, numerical modeling and simulation could serve as a powerful tool to improve mechanistic understanding of the coupled TH processes in frozen soils. However, the governing equations of numerical models are highly complex and nonlinear due to the complex TH processes in frozen soils, which include multiphase interactions, phase change and frost heave. It is no doubt that previous problems pose a grand challenge in developing cold region hydrologic models. In this study, an in-house developed physically-based cryo-hydrogeological model for simulating coupled TH processes in frozen soils was implemented into an open-source, massively-parallel computational fluid dynamics (CFD) software with a user-friendly interface. As validations, a series of benchmarking cases and laboratory freezing experiment were applied to validate the accuracy of the current model. The proposed model was then applied to simulate flow and heat transfer in a 1D column sampled from the upper reaches of the Heihe River Basin (uHRB), 2D sunny and shady slopes extended from two automatic meteorological stations in the uHRB, and further a 3D small watershed selected from the Binggou basin. This study is expected to fundamentally understand the dynamic multiphase flow, ice-water phase transition and heat transfer processes in seasonally frozen soils, and reveal the effects of seasonal change, slope aspect and micro-topography on the coupled TH processes in cold regions. Those insights would provide fundamental evidence for subsurface hydrology in cold regions and the conclusions are summarized as follows:

          Firstly, according to the conservation laws of viscous fluid flow and heat balance, the cryo-hydrogeological model was established and coupled with several vital processes (i.e., dynamic multiphase flow, ice-water phase transition and heat transfer processes). Based on the open-source CFD environment OpenFOAM, the cryo-hydrogeological model (named darcyTHFOAM) was in-house developed by C++ language in Linux environment, which was suitable for simulating the 1D, 2D and 3D coupled TH processes in frozen soils (or other porous media under FTCs) with adaptive spatio-temporal resolutions (from mm to km, from μs to h). A corresponding interface-based software was also developed using Qt Designer and Python 3.5.

         Secondly, a series of benchmark examples ranging from simple to complex were used systematically evaluate the model, including classical analytical solutions, benchmarking cases and laboratory experiments. The soil temperature and liquid water content simulated by the current model were highly consistent with the analytical solution, those simulated by the other cryo-hydrogeological models and experimental data, which verified the accuracy of the current model.

Thirdly, the current model was used to simulate the coupled TH processed in the 1D column sampled from the Arou superstation at the uHRB. By comparing with 5-year in situ frozen soil observation of temperature and liquid content, it was demonstrated that the current model was capable of predicting fine-resolution TH processes under FTCs with both high accuracy and efficiency. The computed temperature and liquid water evolutions clearly showed the entire freeze-thaw process in seasonally frozen soils. In addition, the soil temperature evolution was similar with the sine function and the shallow soil was more sensitive to seasonal change than deep aquifers.

Fourthly, in order to understand the variability for different slope in a small valley in the uHRB, two transects (i.e. south-facing slope, north-facing slope) were configured for numerical simulations based on two automatic meteorological stations. Simulations between south-facing and north-facing slope revealed that the aspect of a slope had significant impact on the soil temperature and moisture distribution in frozen soils. The north-facing slope was colder, wetter, thaw later than south-facing slope. Meanwhile, the north-facing slope had longer freezing duration and more water storage than those in the south-facing slope.

 Finally, according to the data of the digital terrain elevation model, a 3D actual watershed model selected from the Binggou basin in the uHRB was established successfully as well as for numerical simulation. The current model not only exhibited better accuracy at the 3D watershed scale, but also satisfied the high computational needs for larger areas. Moreover, by comparing the results of the simplified model and the three-dimensional actual model, this study analyzed the impact of micro-topography on the coupled TH processes in cold regions. Specifically, if ignoring the micro-topography, the lateral flow and heat transfer process would not be described exactly. The simulated soil temperature and the liquid water content would reach 8.17 °C and 15.03%, respectively.