ABSTRACT: 
The modeling of the directional reflectance of vegetation canopies and vegetation-covered surfaces has been a 
highly active field in remote sensing within the past decade. Recent developments have refined physical mod 
els of directional reflectance; added coupled atmospheric models; included invertible models; and produced 
new, semiempirical models. Although models are well-formulated, more validation datasets are needed. For 
the EOS era, invertible models that provide useful information for global change studies are most desirable. 
KEY WORDS: BRDF Models, Vegetation Reflectance, Global Change, Earth Observing System 
1- INTRODUCTION 
1.1. Anisotropic Reflectance 
The earth's surface scatters radiation anisotropically, especially at the shorter wavelengths that characterize 
solar irradiance. Surface scattering is described by the bidirectional reflectance distribution function (BRDF) 
(Nicodemus, 1977), which specifies the behavior of surface scattering at a particular wavelength as a function 
of illumination and viewing positions within the hemisphere. The scattering behavior of a surface also 
determines its spectral albedo—the ratio of radiant energy flux within a particular waveband that is scattered 
upward and away from the surface in all directions to the downwelling irradiance in that waveband incident 
upon the surface. If the BRDF is known, the albedo can be derived given knowledge of the angular distribu 
tion of incoming irradiance. 
The anisotropic reflectance of the earth's surface provides an opportunity to infer information about 
the physical parameters of the surface cover that produce the anisotropic effect. In the case of vegetation-cov 
ered surfaces, this anisotropy derives largely from such factors as the scattering behavior of leaf surfaces; the 
distribution of leaf surface orientations; the size and spacing of leaves; the clumping of leaf area into individual 
plant crowns; the size and shape of plant crowns; the arrangement of plant crowns on the surface; and the 
anisotropic reflectance of the underlying layer of soil or ground cover. 
Physical models of the scattering behavior of vegetated surfaces that include such factors can, under 
proper circumstances, be inverted from measurements of surface radiance to infer parameters that are of value 
in studies of global change and global ecosystem dynamics. Such parameters include the amount of leaf scat 
tering material (leaf area index), which is useful in global ecological monitoring of ecosystem piimary produc 
tivity and also governs the transfer of latent heat through transpiration in surface energy balance studies; hemi 
spherical albedo, which describes the amount of radiation reflected (and therefore absorbed) by the 
soil/vegetation layer and is useful in surface energy balance studies and global climate modeling; and plant size 
and spacing, which condition surface roughness length and therefore turbulent energy flux and mass transfer 
as specified within global climate models. 
Inference of these types of physical parameters requires inversion of a physical model of canopy 
reflectance, which in turn requires a suite of reflectance measurements of the vegetated surface obtained from 
different viewing positions. Since the acquisition of such measurements over large areas is possible by remote 
sensing with airborne or spacebome instruments, this possibility has inspired the development during the past 
decade of the field of plant canopy directional reflectance modeling. 
Note also that the anisotropic reflectance of earth surfaces presents a problem for inference from 
remotely-sensed images. Reflectance anisotropy means that radiance measurements of the same surface cover 
will vary with viewing position, which can lead to incorrect scene inference, especially in the case of multidate 
satellite imagery in which views are acquired under different geometries. Remote determination of the surface 
BRDF allows correction for these view angle effects through such methods as normalizing individual mea 
surements to a standard viewing position, or integrating the BRDF to yield a hemispherical spectral albedo 
f hat is not dependent on viewing direction.