Abstract

The use of remote sensing techniques in identifying probable rattan (cane) habitats in the natural forests of Kerala has been successfully demonstrated. Rattans occur as climbing palms in scattered, isolated pockets in different forest types. The heterogenous nature of the overstorey vegetation, as well as the lack of information on their tonal differentiability, characterizes the problems in delineating such areas from the rest of the forest. A new approach has been adopted to overcome this difficulty. Ground truth information was collected from as many locations as possible, to include all habitat conditions. A few of these data were statistically selected as training sets for map preparation through digital image processing. The rest were used to cross-check the map, and a method involving deductive and inductive analysis based on the colour signature of overstorey vegetation was applied as a key feature of this approach