随着全球气候变暖，冻土大面积退化，由此诱发融冻泥流愈加频繁，将对三江源生态环境构成威胁。目前关于此类灾害多停留在定性描述，其监测数据不足、形变机理不清、前兆信息不明。本项目选择典型融冻泥流，利用时序InSAR技术监测融冻泥流形变，结合坡体内部温度、湿度，分析周期性冻胀对相位延迟影响，建立冻土形变模型，分离短时间季节形变，获得高精度、长时间序列坡体形变；采用卫星遥感、无人机、传感器、现场调查及室内试验等手段，从物理力学性质、微地貌变形迹象、内部水热变化以及外界环境等方面，选取形变影响因子，通过建立模糊聚类支持向量机混合学习模型，定量分析不同时间尺度下形变特征与影响因子关系，明确动态演化过程中关键因子及特征信息，揭示三江源典型融冻泥流形变机理，为该类灾害的前兆信息识别奠定基础，为三江源生态环境保护提供重要科学认识。

With the increasing of global climate, large areas of permafrost have degenerated, and will induce gelifluction occurrence even more frequently and threaten ecological environment in the source of three rivers region. At present, most studies about gelifluction are qualitative description, of which monitor data is deficient, deformation mechanism is not clear and precursor information is not distinct. So this program intends to select typical gelifluction as study area, explore deformation by long time series InSAR, analyze phase delay from the seasonal freeze and thawing by the internal temperature and moisture of the slope, establish deformation model of permafrost, extract seasonal deformation and obtain high precision and long time slope deformation. Meanwhile, we choose deformation influence factors from physical and mechanical properties, microtopographical deformation signs, internal hydrothermal and external environment using satellite remote sensing, unmanned aerial vehicle, sensors, field investigation and laboratory geotechnical test. Then, the fuzzy clustering support vector machine hybrid learning method is developed to analyze the relationship between deformation characteristics and its influence varies quantitatively. Furthermore, crucial factors and feature information during dynamic evolvement process are determined, which reveal deformation mechanism, lay solid foundation for recognition of precursory information and finally provide important scientific understanding for environmental protection in the source of three rivers region.