Spectral sensitivity to carbon and nitrogen content in diverse temperate pastures of Australia

This paper describes an analysis of spectral data transformations for selection of absorption features and spectral patterns that are sensitive to carbon and nitrogen content of diverse temperate pastures in southeastern Australia. This work used a database of spectra acquired from field campaigns together with wet chemistry estimates of biochemicals including lignin and cellulose, and near infrared spectroscopy estimates of crude protein from coincident field samples. The spectra were grouped in a matrix based on species composition, phenology, greenness and time of acquisition. Absorption features derived from first derivative transforms and continuum removal for known spectral bands showed some predictive potential for crude protein and cellulose. Predictive potential was enhanced if analysis was restricted to green grass-clover pasture samples. The methods require further refinement and testing before they are ported to analysis of airborne and satellite hyperspectral imagery.