Object Oriented Land Cover Mapping of the Kanchenjunga Conservation Area (KCA) in Nepal, for Sustainable Development and Use of Natural Resources

Abstract

As a part of the collaboration with the WWF Nepal Program in strengthening the sustainability of natural resource use, the authors aimed at detecting the changes in forest and land cover in the Kanchenjunga Conservation Area (KCA) in eastern Nepal during the last 30 years. This goal was achieved through multitemporal land cover classification of Landsat TM and ETM+ data, based on an object oriented image analysis approach by eCognition software. The resulting land cover maps showed an overall accuracy of nearly 0.90 and a Kappa coefficient of 0.87 for the classification of the 2000 scene, and a value of 0.94 for both these accuracy measures for the 1989 classification. Forest cover in the KCA increased about 1% between the years 1989 and 2000. The study provides for the first time a reliable database including comparable maps on land cover and forest area in the study area. The availability of such data so far has been highly limited for the KCA. Consequently, the results provide a clear gain of information for decision makers as they can now be used for different ecological or political purposes by WWF Nepal as well as in University research groups or by local institutions in the KCA. Landsat, eCognition, object oriented classification, forest change, sustainable development, Kanchenjunga Conservation Area, Nepal