In regions with significant terrain variations, the modulation of SAR backscatter by mountain slopes can dominate interpretation of the radar imagery unless effective countermeasures are first applied. We first demonstrate deficiencies in conventional radiometric treatments. Geocoded-terrain-corrected (GTC) products assume an ellipsoid-model for the radiometry, even if they improve upon geocoded-ellipsoid-corrected (GEC) imagery by properly compensating for the effects of terrain variations on image geometry. Both the sigma nought and gamma nought radiometric normalisation conventions as applied to distributed targets have an ellipsoidal Earth assumption at their core.
Simply using a local-incidence-angle-mask (LIM) to normalise image radiometry fails to adequately model the image formation process. We prefer to use instead a product that we refer to as terrain-corrected gamma for backscatter analysis. The product makes use of SAR image simulation, incorporating shadow checks and proper accounting of local illuminated area in foreshortened and even layover areas: the result is a substantially improved sensor model in comparison to LIM-based backscatter retrieval. Use of terrain- corrected gamma in a radiometrically terrain-corrected (RTC) product enables multi-track and even multi- sensor image overlays, as terrain-induced backscatter variations are normalised using the available DEM. By properly normalising the hills and mountains, the growing availability of SAR images from diverse sensors can be compared on a “level playing field”.
Time series analysis of hundreds of multi-track ASAR wide swath images covering Switzerland is shown to benefit when comparisons are made using terrain-corrected gamma rather than GTC or LIM- normalised SAR backscatter retrievals. We show how the spring snow melt period can be followed well using multi-track ASAR WS data only if terrain-corrected gamma backscatter values are used as the basis for comparison. Finally, we recommend improved standard backscatter retrieval from land surfaces in future SAR missions such as Biomass or CoReH2O, and Sentinel-1.