562 
understanding the vegetation phenomena (e.g.,the notion of "spectral libraries"). Instead, there has been a 
recent proliferation of publications related to measuring and understanding the bidirectional reflectance 
distribution function, or BRDF, of Earth surfaces and certain features of the BRDF, such as the so called "hot 
spot," as they may be closely related to important vegetation parameters, such as leaf size, plant canopy 
architecture, LAI, and plant density (Suits, 1972; Hapke and Wells, 1981; Hapke, 1986; Gerstl and Simmer, 
1986; Goel, 1988; Ross and Marshak, 1989; Pinty, et al., 1990; Jupp and Strahler, 1991). 
This paper is primarily intended to review the reasons that multidirectional radiances or reflectances 
are needed and are being studied and to discuss some of the important problems that are encountered in trying 
to obtain accurate measurements of directional radiances. The emphasis in this paper is on vegetation. These 
considerations lead to developing some suggestions concerning what priorities should be given to field 
measurements of directional surface reflectances in future research. 
2 - DIRECTIONAL RADIANCE MEASUREMENT USES 
Measurements of surface radiances at off-nadir viewing directions from satellites are either currently useful or 
considered to be potentially useful primarily for the following reasons: 1) to improve spatial and/or temporal 
land surface area coverage; 2) to provide "corrections" for angular reflectance effects on first-generation remote 
sensing products commonly used for Earth resources applications (i.e, "adjustments" to nadir values); 3) to 
improve estimates of basic physical parameters of Earth landscape elements (Examples: a) Direct: single 
scattering albedo; phase function; b) Inferred: biomass, canopy architecture, vegetation structure); and 4) to 
determine atmospheric optical characteristics and conditions. For remote sensing science research purposes 
there is also a need to obtain off-nadir viewing measurements in order to develop general characterizations of 
the anisotropy and to otherwise estimate the BRDF for land surface cover types. Such measurements are 
required for model validations. 
Variations in satellite-measurable reflected or emitted energy must be related to variations in the 
biological and physical parameters of interest, either directly, or more often, through surrogate measures. Some 
direct spectral relationships with Earth surfaces features of interest have been demonstrated, such as the 
mineralogical composition of some minerals (Goetz, 1989). In such cases, the identification and potential 
quantification of the surface material may be realized with nadir spectral data alone. In a somewhat similar 
way, the abundance (density) of chlorophyll is well known to be closely related to the absorption of blue and 
red wavelength reflectance. However, significant complications usually exist in trying to quantify the vegetation 
parameters of interest from such relationships . 
Vegetation is a complex amalgamation living and dead matter, as well as organic and inorganic 
materials (complex plant constituents and soil organics and minerals and water) with various forms of 
presentation of these materials as different species and a variety of expressions of health and maturity. Thus, 
it is indeed a challenge to know to which useful parameters of the vegetation/land surface is the radiant energy 
that is interacting with the vegetated surface really responding, either as individual wavelength band reflectances 
or as spectral indices. For example, a dilemma has developed in the spectral assessment of green vegetation. 
Once we thought that spectral vegetation index variations resulted primarily from the variations in green 
vegetation biomass. But it seems that now were not sure whether Vis, such as the normalized difference 
vegetation index, or NDVI, are really responding to the biomass, the LAI, the fraction of the absorbed 
photosynthetically active radiation, the chlorophyll density, the photosynthesis or some other green plant canopy 
characteristic. Probably it is some complex combination of all of the above. 
In colloquial terms, the "bottom line" is simply that we need more information about how the radiant 
energy interacts with the Earth’s primary surface materials and its dynamic biological coverings in order to 
improve our ability to extract the biophysical parameters of interest from remote sensing measurements. The 
strong implication of recent studies is that it may be possible to extract considerably more detailed information 
about a surface, over and above that available from multispectral and temporal analyses, through a knowledge 
of its angular reflectance properties; and angular reflectance properties are determined through the acquisitions 
of directional radiance measurements. 
3 - SAMPLING THE BRDF 
The BRDF is an intrinsic property of a surface that describes the angular distribution of radiation reflected by 
it, for all angles of exitance and under any given illumination geometry (Nicodemus et al., 1977). However, 
the BRDF cannot be measured directly, as it is a ratio of infinitesimal elements of solid angle and wavelength; 
thus it does not include measurable quantities of radiant flux. Also, it is neither practical nor possible to record 
reflectance data for all angles of incidence and exitance. Therefore, the BRDF must be estimated from a limited 
number of angular reflectance measurements over finite intervals of solid angle and wavelength, using sensors 
capable of viewing a target at a variety of different angles. The term BRDF will be used rather loosely in this