We describe the implementation of a so-called object-space algorithm to use a digital elevation model (DEM) and a sensor flight path as input for the production of simulated SAR images. The method aims to achieve a high geometric accuracy and employs a parametric mapping model based on SAR range and Doppler equations. Information about SAR layover and shadow areas is stored in a so-called ‘layover and shadow map’. We simulate ERS-1, X-SAR, and JBRS-1 images of high-relief terrain, and compare them with corresponding real images. Manual measurements on the ERS-1 image show a simulation accuracy of better than 25 m, if sufficiently accurate sensor parameters are available. A cosine reflectance model delivers a good overall radiometric impression. However, this is being improved where ground truth exists, most notably on glaciers and areas covered with snow. As an application, we discuss the use of the simulated products for geometric rectification and radiometric calibration as a prerequisite for thematic analysis.
Gelautz, M., H. Frick, H. Raggam, J. Burgstaller, and F. Leberl, SAR Image Simulation and Analysis of Alpine Terrain, ISPRS Journal of Photogrammetry and Remote Sensing, vol. 53, no. 1, pp. 17-38, 1998.