Land Cover Classification Using Landsat TM Imagery in the Tropical Highlands: The Influence of Anisotropic Reflectance

Despite the tremendous attention given to conservation projects in the Neotropics, few published studies have documented remote sensing studies in tropical highland areas. Even fewer publications have addressed the use of topographic normalization methods in these regions. This article discusses the influence of anisotropic reflectance patterns on land cover classification for two study areas characterized by very rugged terrain and high relief. Landsat Thematic Mapper subscenes for sites in both Costa Rica and Ecuador were corrected using both Lambertian and nonLambertian models. While use of the Lambertian model proved inappropriate for these mountainous areas, application of the non-Lambertian model enhanced classification accuracies. Colby, J.D., and Keating, P.L. (1998). Land Cover Classification Using Landsat TM Imagery in the Tropical Highlands: The Influence of Anisotropic Reflectance. International Journal of Remote Sensing, 19(8):1479-1500. (May 20, 1998) Published by Taylor & Francis (ISSN: 1366-590). DOI: 10.1080/014311698215306 Land cover classi® cation using Landsat TM imagery in the tropical highlands: the in ̄ uence of anisotropic re ̄ ectance (Received 6 May 1997; in ® nal form 10 November 1997 ) Abstract. Despite the tremendous attention given to conservation projects in the Neotropics, few published studies have documented remote sensing studies in tropical highland areas. Even fewer publications have addressed the use of topographic normalization methods in these regions. This article discusses the in ̄ uence of anisotropic re ̄ ectance patterns on land cover classi® cation for two study areas characterized by very rugged terrain and high relief. Landsat Thematic Mapper subscenes for sites in both Costa Rica and Ecuador were corrected using both Lambertian and non-Lambertian models. While use of the Lambertian model proved inappropriate for these mountainous areas, application of the non-Lambertian model enhanced classi® cation accuracies. Despite the tremendous attention given to conservation projects in the Neotropics, few published studies have documented remote sensing studies in tropical highland areas. Even fewer publications have addressed the use of topographic normalization methods in these regions. This article discusses the in ̄ uence of anisotropic re ̄ ectance patterns on land cover classi® cation for two study areas characterized by very rugged terrain and high relief. Landsat Thematic Mapper subscenes for sites in both Costa Rica and Ecuador were corrected using both Lambertian and non-Lambertian models. While use of the Lambertian model proved inappropriate for these mountainous areas, application of the non-Lambertian model enhanced classi® cation accuracies.