全球气候变暖已导致多年冻土发生严重的退化, 现已威胁到许多区域的生态安全, 特别是作为五大畜牧业生产基地的天山地区。 当前区域内的研究多针对冰川进行分析, 较少对多年冻土深入挖掘。 通过39景ENVISAT ASAR影像（覆盖时间从2003年6月17日到2010年6月15日）, 利用SBAS-InSAR方法监测区域内多年冻土区的地表形变现状。 详细介绍算法原理, 生成ASAR数据集单视复数影像的连接图时, 以空间和时间基线距分别小于500 m和550 d为原则, 生成126幅差分干涉图。 其中因基线和多普勒质心差的原因, 6景ASAR影像没有进行配对。 生成连接图后, 配合STRM V4版本的DEM数据, 经过干涉图去平、 自适应滤波、 相干系数图生成、 相位解缠后, 再对这126对干涉图进行质量较差52对的移除。 利用地面控制点, 进行轨道精炼和重去平等操作, 估算形变速率和残余地形, 经过相关系数阈值控制和奇异值分解(SVD)方法, 再结合空间低通滤波和时间高通滤波方法, 计算得到研究区时间序列地表形变反演。 反演结果共包括33期形变影像, 涉及时间从2004年至2010年。 根据研究区形变结果可知, 虽然区域内有不同程度的沉降和抬升, 但整体的形变速率在±5 cm·yr-1之内, 平均形变速率为(-0.07±3.38) mm·yr-1, 说明研究区内地表存在轻微的沉降现象。 此外以3 000 m海拔为界, 分别探讨分布季节冻土和多年冻土的平原地区和山地区域形变规律发现, 平原地区的形变主要表现为抬升, 仅北方接近城市的区域表现出强烈的沉降现象。 而山区的形变结果则较为零散, 整体形变基调以沉降为主, 大体上在西部和东部地区分别呈现出沉降和抬升现象。 海拔高于3 000 m的形变点存在15 198个, 利用温度和降水数据, 探讨形变点不同形变间隔和总体样本的温度和降水年变化规律后得知, 总体和不同变形速率间隔的形变样本点基本都呈现出逐步变暖的趋势。 并且研究区的山区地表形变速率小于-2.0 cm·yr-1、 在-2.0～2.0 cm·yr-1之间和大于2.0 cm·yr-1的样本点数量依次为6 364, 6 449和2 385个, 山区地表负值的区域多于正值的区域, 体现出该区域的沉降位置多于抬升位置, 也同全球变暖, 冻土退化, 地表发生沉降的规律相一致。 利用主动微波波谱段的ASAR数据, 成功反演了2004年至2010年间的研究区地表形变结果, 并从地表形变反演结果的空间角度、 时间角度和冻土变化的时间滞后性进行了讨论和展望。 以上研究结果期望能为天山区域多年冻土区的地表形变监测提供新的途径和参考。

As global warming, permafrost has degraded seriously. The ecological security of many regions has faced a serious threat to their ecological environment, especially the Tianshan mountain regions, which is one of the five major animal husbandry production bases. At present, in these regions, most studies focus on glacier analysis and few pieces of research about permafrost measure. According to 39 ENVISAT ASAR imagines, covered form 2003 June 17th to 2010 June 15th, surface deformation in permafrost region was monitored by SBAS-InSAR method. In this paper, the principles of deformation algorithm were introduced first. When generating the connection graph of the single look complex image of ASAR dataset, there was 126 differential interferogram based on 500 m and 550 days for temporal and spatial baseline respectively. Because of Spatio-temporal baselines and the Doppler centroid difference, 6 ASAR imagines were not generated the connection graph. Then using STRM V4 DEM, 52 low-quality pair of interferogram were eliminated, after the processes of interferograms flattening, adaptive filter, coherence generation and unwrapping. The ground deformation results of the study area were calculated by external ground control points, refinement and re-flattening, estimation of displacement velocity and residual deformation, coherence threshold control, SVD, spatially low-path filtering and temporally high-path filtering. There were 33 results of ground deformation, which covered from 2004 to 2010. According to the deformation results, there were different subsidence and uplift phenomenon in study areas. The deformation rate of the overall study area was no more than ±5 cm·yr-1, and its average deformation rate was (-0.07±3.38) mm·yr-1. It is indicating that there is a slight subsidence phenomenon in the study area. With the altitude of 3 000 m, the deformation changing mechanism were excavated for the plains and mountain areas distributed by seasonal frozen ground and permafrost respectively. From the research results, deformations in the plain region were uplift except for deformations in the area near cities were subsidence largely. In the mountainous region, the deformations were very scattered than them in the plain region. The overall trend of deformations of the mountain was dominated by subsidence, and subsidence and uplift in the western and eastern regions respectively. There were 15 198 deformation points, which altitude were more than 3 000 m. The annual variation mechanisms of temperature and precipitation about overall deformation points and different deformation intervals points were demonstrated by temperature and precipitation dataset. The results showed that both trends of them have a gradual warming phenomenon. The numbers of deformation rate points about different intervals were 6 364, 6 449 and 2 385 for rates lower than -2.0 cm·yr-1, from -2.0 to 2.0 cm·yr-1 and higher than 2.0 cm·yr-1 in the study mountainous region respectively. Points with negative values were more than points with positive values in the mountainous region, which reflected that subsidence positions were more than uplift positions. This result was also consistent with that global warming cause permafrost degradation then ground subsided. In this paper, the ground deformation results of the study area were successfully calculated by ASAR dataset which was active microwave spectrum. Meanwhile, the deformation results were discussed and prospected in the respects of space, time and the time lag of the permafrost deformation. The study results could provide a new way and reference for the monitoring of permafrost deformation in the Tianshan mountain region.