星载多基线与升降轨InSAR提取DEM方法研究

Spaceborne SAR interferometry is an advanced technology for global high-resolution DEM generation with high-accuracy. In rough terrain areas, multi-baseline and ascending/descending InSAR can avoid phase unwrapping and remove the voids caused by layover/shadow. However, for multi-baseline InSAR estimation existing stochastic models in analytical form are not accurate enough due to the assumption of point targets scattering. Also, most methods for ascending/descending InSAR data fusion do not consider local positioning errors between the two datasets. These issues reduce the accuracy and reliability of height estimation results. Therefore, in this project we are going to explore new approaches of multi-baseline and ascending/descending InSAR for DEM reconstruction. The multi-baseline observation model will be improved by using covariance matrix and Markov Random Field (MRF); meanwhile the local positioning errors between ascending and descending datasets are corrected by 3D surface matching. Based on these innovative ideas, new models and approaches for multi-baseline estimation and ascending/descending fusion will be built. Key problems to be solved include multi-baseline covariance estimation, MAP(Maximum a posteriori)-MRF solution, ascending/descending point clouds coregistration, optimal estimation for ascending/descending fusion, etc. The new approaches will be applied on spaceborne InSAR data, and corresponding algorithms and processing chain will be developed. This project will provide efficient technical means for topographic mapping in Western China.

星载InSAR是获取全球高分辨率高精度DEM的先进技术手段。在地形复杂地区，多基线和升降轨InSAR能够克服相位解缠困难、叠掩与阴影等问题，但现有实现方法仍存在多基线观测随机过程建模不准确和升降轨高程融合未深入考虑局部几何定位误差等问题,降低了高程估计与融合的精度和可靠性。针对这些问题，本项目将开展多基线与升降轨InSAR提取DEM的新方法研究：在多基线InSAR方面，拟使用多基线干涉协方差矩阵和马尔科夫随机场建立观测随机模型，提高建模的准确性；在升降轨InSAR方面，拟使用点云数据三维表面匹配校正升降轨高程存在的局部几何定位误差，并建立融合模型。在以上理论方法研究的基础上，重点突破协方差矩阵估计、多基线最大后验估计、升轨和降轨高程点云配准、升降轨高程融合最优估计等关键技术，并开展星载SAR数据实验，建立相应的数据处理流程，为我国困难地区地形测绘提供实用可靠的技术手段。

本项目针对星载InSAR地形测量在地形复杂区域存在的相位解缠困难、叠掩和阴影等问题，开展了多基线和升降轨InSAR提取DEM的方法研究，并通过星载实测数据实验进行了验证。项目研究内容和取得的主要进展成果包括三个方面：（1）通过理论分析和仿真实验，评估了多基线和升降轨InSAR性能，统计分类了四种地形，提出了推荐观测模式和相应的基线与入射角参数，可为星载InSAR测绘系统设计和观测策略提供指导；（2）使用干涉图协方差矩阵和Total Variation模型建立了多基线InSAR最大后验估计模型，提出了针对TanDEM-X双站干涉数据的处理方法，在普若冈日冰原地区开展了实验，提取了10 m分辨率的DEM，间接验证了高程估计的随机误差可达亚米级；（3）使用高程点云的三维表面匹配建立了升降轨InSAR DEM的融合方法，在夏威夷Kilauea火山口区域开展了实验，使用TanDEM-X聚束双站干涉数据提取了5 m分辨率的DEM，评估的高程中误差优于1 m。本项目的研究成果预计可满足我国1:25000 DEM标准，有望为国土测绘等领域提供新技术支持。