ORTHOGONAL DATA DIMENSIONALITY REDUCTION FOR SPECTRAL DISTINCTION OF AMAZONIAN TREES SPECIES

Sampling trees in natural environment can be used in several studies. Relying on hyperspectral approach, this study aimed to differentiate spectral libraries of four Amazon tree species. We prepared the spectroradiometer data on representative bands on foliar biochemistry, followed by reflectance inflection difference. There was a higher contribution in distinguishing between species based on specific RID (Reflectance Inflection Difference) heights, such as seen on specific representative bands, where RID approach reached 99.87% of data variability related to principal component 1 (PC1) and 99.72% for leaf structure-based bands in PC1. These results can improve further research by using remote sensing techniques, as create brand-new data for Amazonian tree species policymaking, conservation and research.