Retrieving chlorophyll content in conifer needles from hyperspectral measurements

Abstract

Spectrally continuous hyperspectral data can be used to detect subtle features in the leaf optical spectra, which correlate especially well with major leaf pigments such as the leaf chlorophyll content. Extensive field and laboratory measurements were carried out at 10 sites in black spruce (Picea mariana (Mill.)) forests near Sudbury, Canada, to collect leaf optical spectra, leaf pigment contents, and leaf biophysical parameters. It was found that black spruce needles sampled from different sites, age classes, and branch orientations demonstrated variability in both optical properties and chlorophyll contents. The variability in needle optical spectra showed a good correlation (R2 = 0.63) between the average visible absorptance and needle chlorophyll content. The leaf optical model PROSPECT was modified to incorporate the edge effects of needles on light transfer through them. Two leaf biophysical parameters, namely needle width and thickness, were introduced into the model to take into account the effects of leaf morphology on chlorophyll content retrieval. With the modifications to PROSPECT, the model can capture the variability of needle optical properties and chlorophyll content from the measurements. The retrieval of needle chlorophyll contents was improved with an accuracy of R2 = 0.59 and root mean squared error of RMSE = 6.32 µg/cm2 compared with the original PROSPECT model with an accuracy of R2 = 0.31 and RMSE = 9.51 µg/cm2.