Spectral behavior of species used in urban afforestation

Abstract

The aim of this research was to characterize and measure the spectral behavior of six tree species grown at the Federal University of Santa Maria/RS. The studied species were Eucalyptus sp., Ficus benjamina, Inga marginata, Schinus terebinthifolius, Psidium cattleyanum and Handroanthus chrysotrichus. Were randomly collected seven healthy leaves for each specie, andanalyzed with FieldSpec3 Spectroradiometer, in the wavelength range from 350 to 2.500 nm. Were calculated 61 vegetation indices based on the mean average reflectances. Was applied a principal component analysis based on array x species matrice. The differences between the spectral behavior of the species were tested with a variance analysis (Tukey) by each spectral band of the Worldview 2 satellite sensor. The statistical analyses were performed with R software. Hierarchical clustering analysis was performed with PC-ORD 6.0 taking into account the Euclidean distance as measure of dissimilarity. To obtain and process spectral data were used ASD ViewSpecPro 4.05 and SAMS 3.2 softwares. The first component comprehended 90.94% of the variance for the dataset. The lowest ratio between the group and the average differences per specie was in the Red band. The Coastal, Blue, Green, Yellow and Red bands obtained similar behaviors, with differences in the species F. benjamina (except Green's band), I. marginata (Green and Yellow), and P. cattleyanum (Green). The Red Edge band presented no distinction between species. Reflectance differences were more visible in the Near Infrared 1 and 2 bands. Optimized Vegetation Index 1 and Simple Ratio 8 were the most effective indices to study the species.