Identification Leaves of Ulmus pumila L., Tilia cordata Mill. and Acer campestre L. Using Vegetation Indices

Abstract

The aim of the research was to identify a group of vegetation indices (VIs) for species identification. The objects of research were leaves of Ulmus pumila L., Tilia cordata Mill. and Acer campestre L. Hyperspectral imaging (HSI) was carried out under artificial lighting in laboratory conditions using a Cubert UHD-185 hyperspectral camera. A technique was developed for the automated selection of pure spectral profiles from hyperspectral images by setting a double barrier specified by intervals of PSSR and NDVI VIs. A total of 80 VIs was calculated. A statistical analysis of the data was carried out for their representativeness. To determine VIs, the value of which is most dependent on the species characteristics of trees, analysis of variance (ANOVA) and principal component analysis (PCA) methods were used. Research has shown that the PCA method is effective and sufficient to identify the group of VIs characterized by the greatest dispersion in relation to tree species. The PCA carried out for pairs of tree species made it possible to identify a group of vegetation indices, the value of which to the greatest extent depends on species characteristics. These are Carter2, CI2, CRI4, GMI2, mSR2, NDVI2, OSAVI2, SR1, Carter4, Datt2, SR6, Datt, DD, Maccioni, MTC.