Separabilitas Spektral Beberapa Jenis Pohon Menggunakan Citra Compact Airborne Spectograph Imager (CASI): Studi Kasus di Kebun Raya Bogor

Abstract

CASI (Compact Airborne Spectrometer Imager) data was examined to classrfi 20 tree species. The numerical taxonomy using nearest neighbor hierarchical classification method was applied to cluster the spectral reflectance of those species of interest. The study shows promising results expressing the abiliry of CASI image to discriminate 20 tree species. To get a better result of discriminating 20 species, the number of bands used should be more than eight bands. Using combination of less than eight bands caused some class pairs "inseparable". From the cluster analysis, the study also found that there is no relationship between botanical taxonomy of the species and their spectral reflectance. The tree species that belong to the same genus or family could not have similar spectral reflectance.