Evaluation of DEMs Derived by TerraSAR-X InSAR Data

As it is well known, a Digital Elevation Model (DEM) is the simplest form of digital representation of topography and the most common. DEMs are necessary for several applications like generation of orthoimages, terrain determination (such as point elevation, slope, distance, aspect), environmental analysis, generation of contour lines, modeling of hydrologic functions, cartography, civil applications, geographical information system (GIS), urban planning, disaster management (forest fires, erosion control, flood management, earthquake analysis), agriculture, visibility check and also for 3D-views. Because of this large application field, in order to satisfy the need, various DEM generation techniques have been developed up to this time. These techniques can be grouped under two headings according to data collection methods as ground survey and remote sensing. Remote sensing contains four sub-techniques for DEM generation as traditional photogrammetry based on aerial photos, stereo-optical satellite imagery, air-borne laser scanning (LIDAR) and interferometric synthetic aperture radar (InSAR). DEM generation using InSAR technique has several advantages. In this method, data collection is independent from weather conditions except heavy rainfall, moreover clouds can be penetrated and so large coverage up to global can be obtained rapidly at specific short times. Besides, it permits imaging at very shallow look angles thus results in remarkable different perspectives than common vertical photographs. Additionally, it enables resolution to be independent of distance to the target objects, polarization effects can be used and so on. But also there are some disadvantages in comparison with other techniques that, object recognition is poor in SAR images that’s why the selection of Ground Control Points (GCPs) are more difficult and because of imaging geometry and related distorsions as foreshortening, layover and shadow accuracies of DEMs are lower partially in steep, mountanious and forestry topographies.

By this time, several space-borne SAR sensors have been used for Interferometry. Mains of these sensors are SRTM, ENVISAT ASAR, ERS 1/2, RADARSAT 1/2,JERS and ALOS PALSAR. And the most actual space-borne SAR sensor is TerraSAR-X (TSX) which is the main topic of this paper. TSX is the most modern and technological InSAR satellite which has been launched on 15 June 2007 from oldest Russia’s space launch facility, Baikonur Cosmodrome in Kazakhstan. It is a German made satellite and its lifetime will be at least 5 years on the space. It is the most advanced SAR satellite using interferometry up to this day and offers the highest quality spatial data that were not available from space before using high frequency X-band SAR sensor which can be operated in different imaging and polarization modes. TSX uses 3 different operation modes as high resolution Spotlight (~1m), Stripmap (~3m) and ScanSAR (~16m).

The aim of this study is the evaluation of the DEMs derived by TSX InSAR image-pairs that’s why the elevation models have been generated using interferometric processing and afterwards different kinds of analysis have been performed on them. High resolution Spotlight (HRS) mode TSX and pancromatic (PAN) IKONOS image-pairs were used in this study at Istanbul test field and the generated DEMs which have 3m grid spacing were compared using a more accurate reference photogrammetric DEM which has 1m grid spacing and 10cm up to 2m accuracy.

At the evaluation of produced HRS mode TSX and PAN IKONOS DEMs, the absolute and relative accuracies were determined, frequency distributions of Z-differences and dependency up on the aspects were identified, influence of interpolation analysis and morphologic detail analysis were performed, differential DEMs (DIFFDEMs) and shadings were generated. After the evaluation analysis of the DEMs, it has been seen that HRS mode TSX model has nearly same visual quality and absolute accuracy with IKONOS DEM and has better relative accuracy.