Comparison of Lambertian and non-Lambertian topographic normalization algorithms: A case study in Gelibolu, Turkey

Abstract

Aim of this study is to compare the Lambertian and non Lambertian approaches and produce more accurate maps of land use and change in the region, which could be vital for land use decision makers of the region. In this study, both approaches were applied to TM images of Gelibolu located in the Western Turkey to reduce topographic effects of the rugged area extensively covered with forest. A digital elevation model (DEM) was compiled with a resolution of 30 meters in order to study the relationship between the reflectance and the local sun-surface-sensor geometry. The positive effect of topographic normalization methods performed in this study is tested through comparison of classification results carried out for uncorrected and corrected images. The results show that more robust approach, non-Lambertian, performs significantly better than the Lambertian model in terms of the classification accuracies estimated, and it is observed that topographically corrected satellite data provides more accurate assessment of species type than the uncorrected data.