融合DEM和植被覆盖指数的滑坡区InSAR形变测量值不确定度估计

Since the landslide hazards generally occur in the mountainous areas characterized by steep terrain and heavy vegetation, the displacement measurements of InSAR are quite vulnerable to the geometric distortion and decorrelation. Therefore, it is difficult to guarantee the reliability of the InSAR measurements. However, if the uncentainty measures of the InSAR measurements can be estimated before the acquirement of the SAR data, they can be used to select the optimal SAR data in the monitoring of landslides, and provide the precisions of the InSAR measurements in the landslide areas. Based on the integration of the DEM and NDVI data from the multiple satellites, we propose in the project to develop a novel method for evaluating the uncertainty measures of different tracks and different wavelength InSAR measurements in the landslide areas. First, the geometry distortion will be simulated for ascending and descending InSAR by considering the variant of the ground range resolution. Second, the decorrelation will be assessed for multi-wavelength InSAR based on the NDVI data. Third, the uncertainty measures of the InSAR measurements in the landslide areas will be estimated on the basis of the quantitative results from the integration of geometry distortion and decorrelation. The results from the project will help us to understand the reactions between InSAR and terrain landform, and thus make InSAR reaches its best applicability in the monitoring of landslides. In addition, the results can provide scientific evidences for selecting the best configuration and assessing the accuracy of SAR data in the monitoring of landslides, and thus accelerate the engineer and marketization process of the InSAR technique of the monitoring of landslides.

由于滑坡多发生于地形复杂、植被覆盖的山区，导致InSAR形变测量值非常容易受到几何畸变和失相关的影响，可靠性不太稳定。但如果在获取SAR数据之前就能准确估计InSAR形变测量值的不确定度，不仅能实现针对目标滑坡选择最为合适的数据，从而发挥InSAR技术的最佳性能，同时可以提供滑坡区InSAR结果的精密度。鉴于此，本项目提出融合多源卫星数据获取的DEM和植被覆盖指数，研究一种可以客观定量估计滑坡区InSAR形变测量值不确定度的理论与方法，包括1）顾及地距分辨率变化的升降轨InSAR几何畸变定量模拟；2）基于植被覆盖指数的多波段InSAR失相关定量评估；和3）融合几何畸变和失相关的滑坡区InSAR形变测量值不确定度估计。项目成果将有助于理解InSAR对不同地形地貌的作用规律，使其达到滑坡监测性能的上限；并为SAR数据的配置优化和精度评估提供依据，推动InSAR滑坡监测技术的工程化和市场化。

星载合成孔径雷达干涉测量（InSAR）技术凭借其全天时、全天候、大范围和高精度的独特优势，能够及时获取滑坡的静态分布和动态形变信息，被广泛的应用于滑坡形变监测领域。然而滑坡灾害一般发生于地形复杂、植被覆盖的山区，使得InSAR形变测量值非常容易受到几何畸变和失相关的影响，导致InSAR滑坡形变监测结果不可靠。此外，随着技术的发展，融合多维度(多时相、多轨道、多波段和多角度等)SAR数据进行滑坡形变监测即将成为常态，如何实现多维SAR数据的最优配置尚缺少指导依据。针对以上问题，本项目展开了三个方面的技术攻关，包括（1）提出了一套顾及地距分辨率变化的升降轨InSAR几何畸变定量模拟的理论和方法体系，评估了不同分辨率下的InSAR几何畸变模拟差异；（2）建立了基于植被覆盖指数（NDVI）的InSAR失相关定量估计模型，实现了植被区InSAR失相关程度的定量估计；（3）研究了融合几何畸变和失相关的滑坡区InSAR形变测量值不确定度估计，定量评估了不同来源DEM对InSAR形变测量精度的影响，并在三峡新浦、重庆金坪子等典型滑坡区域开展大量实验。结果表明，本项目提出的滑坡区InSAR形变测量值不确定度估计方法能够很好地评估SAR数据在滑坡形变监测中的表现性能，为SAR数据的优化配置提供参考依据，极大地推动了InSAR滑坡形变监测技术的工程化和市场化应用。基于项目成果，项目组共发表论文10篇，其中SCI论文6篇、CSCD论文4篇；申请并授权国家发明专利1项。