Four-Scale Linear Model for Anisotropic Reflectance (FLAIR) for plant canopies. I. Model description and partial validation

As optical remote sensing techniques provide increasingly detailed canopy reflectance data at a variety of illumination/view geometries, direct quantitative comparisons between data sets require a flexible model of the bidirectional reflectance distribution function (BRDF) suitable for inversion. Typically, such derivations rely on: 1) complex and computationally expensive empirical canopy descriptions, or 2) simplifications for specific canopy types, conditions, or view geometry. More practical would be one general model not requiring significant computing resources, but that provides information on canopy architecture when utilized as an inverse model. The Four-Scale Model, developed by Chen and Leblanc (1997), describes canopy reflectance considering four levels of architecture, distributions of tree crowns, branches, shoots, and leaves. A linear kernel-like model has been developed from this Four-Scale Linear Model for Anisotropic Reflectance (FLAIR). While simplifications are performed, an effort has been made not to limit FLAIR to specific canopy characteristics, while maintaining relationships between modeled coefficients and canopy architecture. Comparisons between Four-Scale and FLAIR, and use of FLAIR in the forward mode on multi-angular data sets obtained during BOREAS 1994, allow examination of the suitability, capabilities, and limitations of this model in describing canopy reflectance. As partial validation, this paper compares FLAIR functions to aspects of the Four-Scale Model from which they are developed. Examination of how this model reacts to inversion of simulated reflectance data sets demonstrates its ability to simulate and reproduce canopy reflectance. leading toward the retrieval of reasonable LAI.