Apart from viewing geometry and atmospheric conditions, the topography of the land surface substantially affects reflectance measurements from multispectral scanners in the optical domain. For this reason, topographically induced illumination effects must be accounted for using suitable preprocessing methods. Besides object properties and topographic attributes, reflectance characteristics are also influenced by the degree of surface anisotropy. Therefore, the terrain normalization methodology should ideally be tailored to the variability of structural surface properties within the area under investigation. This study evaluated different algorithms (both lambertian and non-lambertian) for topographic normalization of optical satellite imagery using statistical analysis. A modification of existing methods was carried out by the inclusion of land surface parameters which were derived from spaceborne interferometric SAR- and optical data. The results show that a stratified terrain correction based on structural surface characteristics strongly reduces the topographic impact and leads to an improved class homogeneity and spectral separability of landscape units, potentially minimizing misclassification due to spectral overlap.

The tool will be available for IDL (.sav) shortly. If you are interested, please contact Dr. Stefan Erasmi or André Twele.