A physical interpretation of the correlation between canopy albedo and nitrogen using hyperspectral data

Abstract

Recent studies have shown that there is a high correlation between canopy nitrogen and NIR reflectance and subsequently canopy albedo. We provide a physical explanation for the correlation using the spectral invariants of the radiative transfer. The spectral invariant approach allows for a very accurate parameterization of the canopy reflectance in terms of the wavelength dependant single scattering albedo and two spectrally invariant and structurally varying parameters-recollision and escape probabilities. The spectral invariant parameters depend on macro-scale structural features such as crown shape and size, the proportion of sunlit and shaded leaves and ground cover, as well as micro-scale information such as within crown foliage distribution. We retrieve the spectral invariant parameters from Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) hyperspectral data for 3 sites in New England and 2 sites in the southeastern United States for which ground data on mass-based foliar %N were available. Theoretical and statistical analyses showed that canopy structure is highly correlated to canopy albedo, R2=94, suggesting that canopy structure is a dominant factor causing observed variation in NIR albedo. We therefore hypothesize that the amount of canopy nitrogen may have an indirect impact on NIR albedo through the formation of macro-scale features. Finally we show that we can predict the amount of canopy nitrogen more accurately using the macro-scale features than canopy albedo indicating that competing factors at the leaf and canopy scales are imbued in the measured albedo signal.