FORWARD AND INVERSE MODELING OF CANOPY DIRECTIONAL REFLECTANCE USING A NEURAL-NETWORK

This article explores the use of artificial neural networks for both f orward and inverse canopy modelling. The forward neural modelling para digm involved training a network for predicting the bidirectional refl ectance distribution function (BRDF) of a canopy given the density of the trees, their height, crown shape, viewing, and illumination geomet ry. The neural network model was able to predict the BRDF of unseen ca nopy sites with 90% accuracy. Analysis of the signal captured by the m odel indicates that the canopy structural parameters, and illumination and viewing geometry, are essential for predicting the BRDF of vegeta ted surfaces. The inverse neural network model involved learning the u nderlying relationship between canopy structural parameters and their corresponding bidirectional reflectance. The inversion results show th at the R-2 between the network predicted canopy parameters and the act ual canopy parameters was 0.85 for density and 0.75 for both the crown shape and the height parameters. The results of both forward and inve rse modelling suggest that neural networks can model accurately the BR DF of vegetated canopies.