位于青藏高原腹地的多年冻土地带，其冻融过程中的土壤含水量和土壤冻结深度的变化对气候强烈响应并产生显著的陆面能-水平衡变化，进而又对全球气候产生较大的反馈作用。为了能准确模拟这种变化，选取青藏高原多年冻土分布区的风火山左冒孔流域（长江源）进行了相关的野外数据采集和试验，以考虑土壤冻融影响的水-热耦合陆面过程模型——SHAW为动力学约束框架，验证集合卡尔曼滤波算法在改进模型对土壤冻融过程中土壤水分和冻土深度的计算效果。基于试验点的数据同化计算结果表明：数据同化方法可以融合观测信息显著提高水-热耦合模型对土壤冻融过程中状态变量（土壤水分和冻深）的模拟，并进而改善模型对其它相关能量-水分变量的计算，为在高寒冻土地区利用多源信息进行融合监测提供了理论依据。

The frozen soil region of many years located in the Qinghai-Tibet Plain hinterland, Changing of the Water content of soil during freeze-thaw process, intensely responds to climate change and brings remarkable change on land energy-water balance, then has the large feedback function to the global climate. In order to reveal this kind of change, we used the observation data from the wind volcano testing field system, selected water - heat coupling model SHAW as dynamics restraint frame, which consider the influence of snow cover ＆ the vegetation cover ＆ the forest flooring to the soil freezing and thawing, and improved SHAW forecast capacity to the soil moisture content and the frozen soil depth by Ensemble Kalman Filter of data simulation method. The analysis based on data assimilation theory indicate that the data assimilation method may remarkably enhance the forecasting ability of water-heat coupling model to state variables, and provide theory basis for monitor utilizing multiple source information in frozen soil area.