Hyperspectral remote sensing of boreal forest tree diversity at multiple scales

Abstract

This research compared the variability/diversity of spectral information captured with spectrometers at the airborne, field, and leaf level to tree species diversity. Airborne measurements were made over the North Saskatchewan River Valley while field and leaf measurements were done with synthetic tree plots on the roof of the Biological Sciences building, University of Alberta. Measures of optical diversity (spectral variables), such as the standard deviation in vegetation indices, principal components, and slope analysis, showed significant correlation to species diversity indices. The strongest correlations (R2: ODI#3 = 0.90, ODI#6 = 0.86) were achieved with linear models using three to five spectral variables, called Optical Diversity Indices (ODIs). Experimental methods found that this correlation was based primarily on variation in leaf optical properties. Additionally, rough canopies increased optical diversity and greater spectral range improved correlations slightly. These findings can help design operational methods for remote assessment of biodiversity based on optical diversity.