Quality checking of DEM derived from satellite data (SPOT and SRTM)

摘要

The widespread availability of good quality digital elevation data opens the door to systematic and improved automation of orthoimage production, in the context of the Common Agricultural Policy and the checks on aid applications with remote sensing activity. Best practice for production of VHR ( < 1 m pixel) imagery meeting these requirements usually states a quality of 5 m RMSE in Z is required. During 2004, two sources have emerged: the "open source" SRTM C-band (3 arc second, approx. 90 m grid size) data, released to the general public via the internet and requiring careful processing, and SPOT Image's commercial Reference 3D® product, which is created using the stereo HRS sensor on the SPOT5 platform. Both data sources have official general specifications somewhat lower than the 5 m RMSE_Z, but given the usually limited relief in agricultural areas, an investigation into the quality of these dataset was considered an important task. This paper will report on the test carried out by the FOMI and JRC to determine the suitability of the SPOT Image Reference 3D® product, and the corresponding SRTM 3 arc second data, covering a single production tile (46-47° North/18°-19° East, approximately 10,000 km~2) in Hungary. The quality assessment methodology was developed at the JRC, and executed using reference data available in Hungary by FOEMI. The data set used for comparison consisted of high resolution (5 m x 5 m pixel size, RMSE_Z 0.7 m) digital elevation model derived from topographic maps at scale 1:10 000 and points of Ⅳth order triangulation network (accuracy in X,Y - 0.05 m, in Z - 0.1 m; approximately 5,000 points). For the transformation between the Hungarian EOV projection system and the international Word Geodetic System (WGS 84, EGM96 geoid) the model elaborated at FOEMI has been used. The results show that both datasets performed better than their standard specifications, and are suitable for rectification of most VHR imagery in the context of EU Common Agricultural Policy without further processing, besides projection and datum transformations.