Discrimination of remnant tree species and regeneration stages in Queensland, Australia using hyperspectral imagery

This study assessed the utility of hyperspectral imagery in discriminating remnant tree species and stand regeneration stages in Southeast Queensland, Australia. Reflectance data of three species of woody vegetation (i.e. Eucalyptus populnea, Acacia pendula and Eucalyptus orgadophila), acquired using a HyMaptrade airborne system, were analysed using partial least squares (PLS) regression. Three groups of E. orgadophila species, representing stand regeneration status, were also evaluated. For discriminating such tree species, the PLS results showed high prediction accuracy ranging from 83-88%. The most significant spectral bands span from the visible region (peak at 558 nm and 689 nm), near-infrared region (peak at 987 nm), and shortwave infrared region (peak at 1788 nm). Hyperspectral data was able to discriminate the old stand of E. orgadophila from the young stand, with a moderate accuracy of 72%. Results such as these confirmed the potential utility of hyperspectral data in vegetation mapping and stand characterisation.