Development of hyperspectral biochemistry through the use of statistical modeling and spectral unmixing

Prior attempts at mapping the biochemical characteristics of the forest canopy have met with mixed results. The use of simple regression or stepwise multiple regression has resulted in ambiguous or inconsistent correlations. The current project attempted to integrate two promising techniques: partial least squares (PLS) regression and spectral mixture analysis (SMA). The analysis demonstrate results that are consistent with other published results using the PLS approach. An incremental increase in the explanatory power of the model (to a maximum r² of 0.877 for foliar nitrogen) was observed with the inclusion of the SMA results.