Urban DEM generation,analysis and enhancements using TanDEM-X |
| |
Affiliation: | 1. Geodetic Institute, Karlsruhe Institute of Technology, Englerstrasse 7, 76131 Karlsruhe, Germany;2. Geology Agency, Ministry of Energy and Mineral Resources, Jalan Cendana 15, Yogyakarta 55166, Indonesia;3. Laboratory of Geophysics, Department of Physics, Faculty of Mathematics and Natural Science, Universitas Gadjah Mada (UGM), Yogyakarta, Indonesia |
| |
Abstract: | This paper analyzes the potential of the TanDEM-X mission for the generation of urban Digital Elevation Models (DEMs). The high resolution of the sensors and the absence of temporal decorrelation are exploited. The interferometric chain and the problems encountered for correct mapping of urban areas are analyzed first. The operational Integrated TanDEM-X Processor (ITP) algorithms are taken as reference. The ITP main product is called the raw DEM. Whereas the ITP coregistration stage is demonstrated to be robust enough, large improvements in the raw DEM such as fewer percentages of phase unwrapping errors, can be obtained by using adaptive fringe filters instead of the conventional ones in the interferogram generation stage. The shape of the raw DEM in the layover area is also shown and determined to be regular for buildings with vertical walls. Generally, in the presence of layover, the raw DEM exhibits a height ramp, resulting in a height underestimation for the affected structure. Examples provided confirm the theoretical background. The focus is centered on high resolution DEMs produced using spotlight acquisitions. In particular, a raw DEM over Berlin (Germany) with a 2.5 m raster is generated and validated. For this purpose, ITP is modified in its interferogram generation stage by adopting the Intensity Driven Adaptive Neighbourhood (IDAN) algorithm. The height Root Mean Square Error (RMSE) between the raw DEM and a reference is about 8 m for the two classes defining the urban DEM: structures and non-structures. The result can be further improved for the structure class using a DEM generated with Persistent Scatterer Interferometry. A DEM fusion is thus proposed and a drop of about 20% in the RMSE is reported. |
| |
Keywords: | TanDEM-X Urban DEM InSAR processing PSI-DEM DEM fusion |
本文献已被 ScienceDirect 等数据库收录! |
|