Radiometric Terrain Correction Incorporating Local Antenna Gain

Radiometric terrain correction consists of normalising a SAR image for well-understood backscatter contributions in order to amplify less easily apparent influences (e.g. thematic land cover variance). A rigorous modelling of the SAR image formation process includes consideration of how foreshortening and layover create ambi guity when connecting map geometry grid points to and from counterparts in radar geometry (slant or ground range vs. azimuth). A radar amplitude image simulation is formed by iterating through a facetted DEM, calcu lating the accumulated illuminated area at every range and azimuth coordinate in radar geometry. We show how DEM-based image simulations gain further realism by incorporating knowledge of the SAR antenna's elevation an-tenna gain pattern (AGP). Although typical AGP corrections assume an ellipsoidal Earth, the AGP is actually draped upon the Earth's terrain. We quantify differences between estimates of local antenna gain and illuminated area performed using (a) the typical ellipsoid assumption, (b) a DEM. We demonstrate application of local antenna gain knowledge within the image simulation process using ENVISAT ASAR images acquired over Switzerland. We introduce a weighted resolution approach for robust combination of multiple radiometrically normalised terrain geocoded backscatter maps.