Assessing biodiversity of boreal forests with imaging spectrometry

Recently, imaging spectrometry employing high spectral resolution has been applied to vegetation diversity assessment. A foundation for these studies is the idea of plant optical type, which is affected by canopy structure, leaf traits, and phenology. This study proposes that the optical variation of a forest canopy is strongly correlated to the vegetation diversity. Using data collected with an imaging spectrometer (pixel size 1m, and usable spectral range 400-900nm), flown over the North Saskatchewan River Valley, Edmonton, Alberta, we show that metrics of biodiversity (species richness and Simpson's Index of 30x30m plots) are strongly correlated to metrics of optical diversity (Optical Diversity Indices). These optical diversity indices (ODIs) were built using linear models correlating diversity indices to a combination of variables, which included: the standard deviation of NDVI (NDVIsd), PRI (PRIsd), principal component 1 (PC1sd), and principal component 2 (PC2sd). Three variations of ODIs were built based upon vegetation indices, principal components, and a combination of vegetation indices and principal components. Correlation values were 0.70, 0.80, and 0.86 respectively. In further analyses, PC1 was correlated to the green and NIR wavelengths where PC2 was correlated with blue and red chlorophyll absorption wavelengths. Variation in optical diversity was also clearly visible in spectra and airborne imagery. These indices show that spectral regions used to track the photosynthetic activity (PC2, PRI, and NDVI) and green canopy structure of plants (NDVI, and PC1) can be used to differentiate optical types in a way that yields information on biodiversity. Further experimental studies involving diversity manipulation are needed to answer the mechanisms behind optical diversity. Further testing of optical diversity across many ecosystems is also needed to develop operational methods to survey large areas from and aircraft or satellite. Two contrasting vegetation diversity plots. The left column is a low diversity (SR3) plot while the right plot has middle range diversity (SR6). The top panel is the first principal component of variation (red=high values, and blue=low values). The middle panel shows the average spectra for the plot (black line) and one standard deviation above and below the average (green). The bottom panel shows the correlation between species richness and ODI#3.