Extraction of ground control points (GCPs) from synthetic aperture radar image using DEM

Abstract

Synthetic aperture radar (SAR) data have serious geometric distortions caused by inherent side looking geometry and are usually corrupted by speckle noises so that it is difficult to identify ground control points (GCPs). The location of the pixel in a given SAR image can be derived from the knowledge of sensor position and velocity. We propose a GCP extraction algorithm that has an improved capability. The algorithm is based on the geocoding process of the radar image and requires precise orbit information. Whereas conventional geolocation algorithms depend on the knowledge of the target height relative to an assumed Earth ellipsoid model, more precise positioning can be achieved if utilizing an available digital elevation model (e.g. USGS GTOPO-30 DEM, SRTM DEM, etc.). The ERS SAR data with precise Delft orbit information and the national digital elevation model (DEM) were used. We analyze the accuracy of the results from our algorithm by using digital map. In case of using SRTM DEM, the mean values of the planimetric distance error are 9.1 m and 14.7 m with standard deviations of 6.4 m and 9.3 m in northing and easting direction, respectively. The errors in northing direction depend mainly on the imaging pulse time, and those in easting direction are related to the accuracy of used digital elevation model. The GCPs extracted were used for georectification of optical images as well as SAR data.