A fast canopy reflectance model to simulate realistic remote sensing scenarios

A model for light interaction has been developed to compute spectral and bidirectional reflectance from discontinuous canopies approximated by an arbitrary configuration of plants. The model assumes certain principles of geometric models, e.g., that sensor integrates the radiance reflected from tree plants, shaded soil, and illuminated soil. However, the model attempts to compensate for errors due to multiple scattering in vegetation canopies that lead to nonlinear mixing. In contrast to geometric models, tree crowns are treated as porous (partially transmitting), geometric bodies. Reflectance of canopy and shadowed ground are nonlinear mixtures of leaves and background signatures, which are moreover influenced by parameters of canopy, such as leaf area index (LAI), coefficient of attenuation, and leaves transmittance. Optical parameters are not constant but stochastic variables are controlled by a certain texture in canopy envelopes, and roughness and relief in surface background. The model may also be run using backgrounds presenting variable topography and comprising different landscape features on imported real images. The model predicts the basic features of the Bidirectional Reflectance Factor (BRF), i.e., bowl shape and the hotspot, but unlike common models, it is well suited to address the spectral and spatial domains. For example, the model provides a fast and efficient strategy to derive hyperspectral images at appropriate spatial resolutions (e.g., regional scale) over a wide range of ecosystems.