Major forests and plant species discrimination in Mudumalai forest region using airborne hyperspectral sensing

Abstract

The present study focused on forest type classification and major plant species assemblages in Mudumalai forest region using Airborne Visible/Infrared Imaging Spectrometer Next Generation. The phytosociological analysis exhibited a total of 657 individuals (1095 individuals ha-1) of 24 tree species belonging to 22 genera and 18 families. The highest tree density was contributed by Tectona grandis (132 individuals ha-1 and 12.05% of total tree density) followed by Anogeissus latifolia (105 individuals ha-1 and 9.59% of total tree density). The support vector machine study showed the dominance of Southern Tropical Semi-Evergreen forests (31%) followed by Southern Tropical moist deciduous forests (26.7%) and Southern Tropical dry deciduous forests (24.8%) with a very high accuracy (92.37%). The comparative analysis of the existing forest types with Champion and Seth’s (1968) classification of forests exhibited a change of 30% in forest types in terms of their structure, composition, and extent over a period of 50 years. The spectral angle mapper–based study emphasized the defining role of elevation, rainfall, and temperature in species distribution, and physiognomy with dominance of A. latifolia (∼19.22%). The study implies the high spectral fidelity of airborne images for forest type mapping and plant species discrimination in tropical forests.

Keywords

• AVIRIS-NG
• Mudumalai forest region
• Plant species characterization
• Spectral angle mapper
• Spectroradiometer measurement
• Support vector machine