Tree Species Typing in Temperate Forests using Hyperspectral data of Hyperion EO-1 in western Himalayas

A few dominant tree species in temperate forests of western Himalayas have been mapped using Hyperion data. The data was processed for removing atmospheric distortions using FLAASH before classification. The replicate spectral profiles were collected in ground using field spectrometer (ASD) during same season for classification. Reconnaissance survey was conducted to locate populations or sufficiently large patches of forest trees species such as Quercus leucotrichophora, Cryptomeria japonica, Pinus wallichii, Thuja orientalis , Viburnum coriaceum Trema politoria , Cedrus deodara, and Pogostemeon sp. Spectral Angle Mapper (SAM) classifier was applied using end-members obtained from the collected spectra of different species. Ground check was taken up after classification and classified output was checked on the ground for mapping accuracy. An accuracy assessment was done by means of confusion matrix and ground data collected from the field. All the species and land cover features could be mapped 100% accuracy except for settlements (66.67%) and overall accuracy was 96.88%.