sured and documented for only a few leaves of a few species. Even leaf area index is often lacking from sets of 
measurements, or is estimated only indirectly from regression relationships. Clearly, the modeling community 
and the experimentalists need to embark on joint ventures to collect appropriate validation data. The 
BOREAS experiment is an example of such a venture involving both modelers and measurers. More such 
opportunities are needed. 
The validation of vegetation surface radiance models serves mainly to validate the physical abstrac 
tions behind them. Carefully structured repeated comparisons between models and measurements can 
improve our confidence in those abstractions. But once we understand the physics, what then? 
Proper characterization and understanding of the BRDF of vegetated surfaces is an important build 
ing block in remote sensing of the earth. Without an understanding of the anisotropic reflectance behavior of 
the surface, our ability to infer the biophysical state of the surface is obviously limited. Inversion of physical 
models of surface scattering plays a key role here. With an appropriate array of remote measurements from 
spaceborne platforms, the potential exists to obtain the driving parameters that condition reflectance 
anisotropy for significant regions of the earth’s surface. Some of these parameters (e. g., leaf area index), will 
be of direct use in other fields, such as ecosystem modeling or global climate modeling. Other parameters 
(e. g., leaf angle distribution) are not especially useful. The challenge to the BRDF modeling community is to 
provide invertible models that are robust, reasonably accurate, and yield useful information at the broad spa 
tial scales that characterize the important applications of remote sensing. 
3.2. Global Change Agenda 
What are the important future applications that will rely on accurate characterization of angular surface 
reflectance? Elere we may look to the global change agenda. Global climate modeling is probably the most 
important component of that agenda. For global climate modeling, surface characteristics can provide two key 
pieces of information—albedo and surface roughness. These parameters have heretofore only been character 
ized at coarse spatial scales, and as the spatial resolution of global climate models increases, climate modelers 
will look to remote sensing for increasingly finer-scale information. Note that climate modelers are now inter 
ested in obtaining albedo as two quantities, broken into shortwave and longwave at about 0.7pm, due to the 
abrupt change in absorption by vegetation at that wavelength (J.-P. Muller, personal communication). The 
significance is that albedo is readily retrievable from the BRDF, provided that sufficient atmospheric informa 
tion is available to model the angular distribution of downwelling irradiance. Moreover, because albedo varies 
with sun and sky conditions, climate modelers may eventually need simple empirical BRDF descriptions so 
that albedo may be treated as a time-variant quantity. 
Another important global change application is ecosystem modeling, especially the modeling of car 
bon fluxes. BRDF models have shown that some important carbon-balance parameters, such as FAPAR. are 
well estimated by empirical measures such as NDVI. However, NDVI varies with look angle, so that a single 
look is not likely to characterize properly the behavior of the plant cover (Myneni et al., 1992c). Yet if the 
BRDF can be derived from multiangle measurements, the biophysical parameters can be properly summa 
rized. Further, calibration of carbon balance models is dependent on the plant community. For example, a 
thin, continuous canopy of annual or perennial grasses will photosynthesize with quite a different efficiency 
than a perennial desert shrub community of the same leaf area (S. Running, personal communication). Inas 
much as the structures of these community types are differentiable by directional reflectance, so will the iden 
tification of community types be facilitated. 
3.3. Once and Future Sensing 
Global inference of BRDF, albedo, and related biophysical surface parameters at fine spatial and temporal 
scales will be available in the twenty-first century with the advent of sensors aboard the EOS-AM and -PM 
platforms, notably the MODIS (Moderate Resolution Imaging Spectroradiometer) and MISR (Multiangle 
Imaging Spectroradiometer) instruments. These instruments will provide sufficient angular radiance measure 
ments to characterize surface BRDF and albedo at fine spatial and temporal resolutions, and inversion strate 
gies to be used with MODIS and MISR data are under development (Running et al., 1994; Strahler et al., 1994; 
Diner et al., 1994). In addition, directional surface measurements will be available from the spaceborne 
POLDER (Polarization and Directionality of Earth's Reflectances) and ATSR-2 (Along-Track Scanning 
Radiometer) instruments prior to the launch of the EOS platforms in 1998 and 1999. For the present, existing 
airborne instruments such as ASAS (Advanced Solid-state Array Spectrometer; Irons, 1991) and the aircraft 
version of POLDER (Douze et al., 1993) can simulate many of the characteristics of these future spaceborne 
sensors. They will be critical tools in the development phase of information-extraction algorithms that utilize 
directional radiance measurements. 
4 - CONCLUSION 
Significant advances in modeling the directional reflectance of vegetated land surfaces have been made in the 
last decade, and, notably, within the last two to three years. These advances have laid the foundation for the