Estimation of canopy structure parameters from multiangular measurements of scattering components

Structure parameters for characterization of vegetation canopies are often estimated from remote optical measurements. Existing methods include those based on measurements of gap fraction, spectral vegetation indices, or the inversion of spectral canopy reflectance models. This paper proposes a novel method based on inversion of multiangular measurements of the abundances of light scattering components, which may be estimated using spectral unmixing. An algorithm is described for predicting the abundances of various scattering components using Monte Carlo simulation with a Poisson canopy model and an ellipsoidal leaf angle distribution. The method was tested using simulated data from ray-traced images in a ground-based measurement scenario. Model fit surfaces were calculated for 20 different values of leaf area index (LAI) and mean leaf angle (MLA). The experiments generally showed good correspondances between observations and predictions, except for high values of LAI and low values of MLA. Future work should include experiments on real data and robust unmixing of scattering components.