TerraSAR-X/TanDEM-X获取高精度数字高程模型技术研究

以双星系统(TerraSAR-X/TanDEM-X)下的bistatic数据模式为例研究了差分干涉获取高精度DEM产品的融合算法和技术流程.针对不同观测几何条件下(升降轨,不同入射角)观测数据的畸变和缺失,提出一种迭代的顾及垂直基线、阴影和叠影的数据融合新方法重建高分辨率高精度的数字高程模型,并对TanDEM-X融合的DEM在不同地物属性特征下(山区及高楼林立的城区)的精度进行定量分析.本文采用了两对覆盖珠海、澳门区域的升降轨TerraSAR-X/TanDEM-X干涉对进行融合处理,并通过收集的高精度Lidar数据进行精度比较分析.此外,本文还定量分析了TanDEM-X的DEM对常规DInSAR技术的改进,并与SRTM、ASTER的结果进行对比.结果表明:提出的升降轨融合方法较单一轨道平台能够较好地改正或减弱由于几何畸变引起的高程信息缺失或错误,通过与Lidar数据的对比发现TanDEM-X的融合DEM在山区的精度较高,其残差的标准差为3.5 m,较单一轨道(升、降轨)分别降低8%和22%,能够通过迭代的方法获取高分辨率(可达2~5 m)、高精度的地形信息;而针对城区密集建筑物的复杂地形来说,融合的DEM的精度稍低,残差的标准差为11.8 m,但较单一轨道(升、降轨)来说有较大改进,其残差标准差分别降低了28%和22%;而在分布较为稀疏的居民区,融合的DEM能够较Lidar数据获取更好的建筑物高度及外形信息,此时的残差标准差可达5 m.同时,TanDEM-X的融合DEM作为外部DEM能够较SRTM和ASTER来说更好地去除地形信息,尤其在山区及高程建筑密集分布的城区,从而利于后续的相位解缠和形变信息的精确获取和解译,为更高精度的时序InSAR形变监测提供有利条件.

Generation of high precision DEM from TerraSAR-X/TanDEM-X

Digital Elevation Model (DEM) derived from TerraSAR-X/TanDEM-X, can overcome the influence of atmospheric delay and surface deformation, and has a much higher resolution and precision compared with traditional InSAR. However, the DEM generated from TSX/TDX is vulnerable to geometric distortion such as layover, shadow, foreshortening etc. Moreover, the precision of TSX/TDX DEM, especially for different surface characteristic, such as mountainous areas, urban buildings and flat areas is still far from clear in application. Therefore, it has a practical significance on the guidance for DEM production derived from TSX/TDX.In this paper, we proposed a procedure which combines both ascending and descending measurements to obtain a high precision and resolution DEM product from the bistatic mode of TSX/TDX. We also use an iterative fusion method considering the perpendicular baseline and the location of layover and shadow as weighing. Because the ascending and descending SAR images have different geometrical conditions, such as incident angle and perpendicular baseline, it can regain the DEM affected by geometric distortion. We also adopted a quality assessment of the fusion DEM (5 meters resolution) for two terrain configurations to demonstrate the precision pattern. The fusion DEM in different kinds of terrain configurations: moderate topography and urban areas with dense high-rising buildings, is compared with DEM collected by Lidar technique in the study area. In addition, we qualitatively and quantitatively analyzed the improvement of fusion DEM in the processing of DInSAR technology compared with the existing external DEM.Two pairs of TSX/TDX data, acquired on 21 November 2011 in the ascending mode and 24 October 2013 in the descending mode, respectively, are used to map a fusion DEM in Zhuhai city and Macau. The results indicate that the fusion DEM with the proposed method can recover more information in the distorted areas compared with the DEM derived from a single platform. In order to validate the precision of the fusion DEM, a Lidar DEM with high resolution and precision of the research area are collected and compared. We find that the standard deviation (RMS) in mountainous areas is 3.57 m with a decrease of 8% and 22% for ascending and descending data, respectively. The standard deviation value in urban areas is 11.85 m which also has a drop of 28% and 22% compared with two single platform. Moreover, the precision of the fusion DEM is high in residential areas with sparse buildings with a standard deviation 5 m. In addition, the TerraSAR-X pairs whose temporal baseline is 11days and perpendicular baseline is 135 m are used to conduct comparison of different kinds of external DEM in DInSAR processing. Moreover, statistical analysis of differential phase (pixel whose coherence is higher than 0.8) in mountainous and urban areas are performed separately for a quantitative analysis. The results show that the fusion DEM used as an external DEM can remove topography phase better compared with SRTM and ASTER data especially in the mountainous area. The standard deviations of differential phases is 1.4 rad, 1.6 rad and 1.4 rad for SRTM, ASTER and TanDEM, respectively. In urban areas with dense high buildings, the standard deviations are 1.7 rad 1.3 rad and 1.1 rad respectively, which means higher resolution and precision of external DEM can improve the quality of DInSAR with high resolution SAR images.This work adopted an iterative fusion method to obtain high resolution and precision DEM. Locations of geometric distortion and perpendicular baseline are regarded as weighting factors to improve the efficiency of fusion procedure by comparing with traditional DEM. Besides, we also qualitatively and quantitatively analyzed the improvement of TSX/TDX DEM in traditional DInSAR, which can help us interpret deformation information and obtain a higher precision of time- series of InSAR.