2. Hybrid Approach for Land Use and Forest Cover Classification in Sikkim Himalaya

Land use and forest cover (LUFC) classification from satellite data in mountainous terrain offers challenge due to varied topography and complexities owing to different illumination conditions. Digital classification following supervised and/or unsupervised techniques in combination with/without ancillary information often does not provide acceptable level of accuracy. This chapter formulates and applies a hybrid approach for LUFC classification using moderate resolution satellite data. Both ‘elimination’ and ‘fishing’ approaches were used to classify the state of Sikkim into seventeen categories. The classification accuracy was estimated at 94.87% at 1:50,000 scale, which is suitable for utilization in further studies such as surface hydrological and energy fluxes. Further, the digital elevation model was utilized to derive the topographic units at 1000 m elevation steps, slope and aspect and their distribution across the seventeen LUFC classes. The distribution of various LUFC classes across different elevation, slope and aspects offers useful information for ecosystem planning and management.