an adequate representation of the target. The target surface to be characterized is commonly a mixture of 
background materials, such as soil, litter, with sub-dominant plant species, and dominant plant canopy elements 
including leaves, stems or twigs, and branches. This forms a vegetation canopy, which when illuminated, 
typically becomes a mixture of directly illuminated (sunlit) and shadowed components of all of the plant and 
background elements. Therefore, each IFOV representing the target vegetation canopy (plant + background) 
should generally contain both sunlit and shadowed canopy and background elements in their proper proportions, 
although other analysis approaches could be developed for sub-canopy element measurements with a sizable 
angular density. 
For acquiring field measurements of directional reflectance today, an IFOV of 15° is typically used. 
This provides a nadiir-view pixel with an area of 1.35 m 2 (1.31 m dia.) at a 5 m height above the target surface. 
Reducing the IFOV to 5° decreases the sample area to 0.15 m 2 (0.43 m dia.). Very high spatial resolution 
measurements (e.g., a T IFOV) may be needed, however, to adequately sample such phenomena as the "hot 
spot.” Determining the angular shape and amplitude of this directional radiance feature may enable the 
development of new algorithms for structural biophysical parameter assessments, as noted above. 
It is also important to consider that the typical multidirectional viewing sensor, whether it be ground- 
based, aircraft or satellite, views off-nadir angles with progressively larger footprints than the nadir footprint. 
For a 15° IFOV instrument at a fixed 5 m height, as discussed above, a 30° viewing angle pixel will sample an 
elliptical surface area of 2.1 m 2 (1.7 m major axis diameter), and a 60° off-nadir view will encompass an 
surface area of 11.7 m 2 (5.5 m major axis diameter) . This leads to other problematic considerations, such as 
how one should express the surface spectral reflectance on an area basis. This issue, and the derived quantity 
called the "reflectance fraction," is discussed by Middleton (1992). 
Another instrument related BRDF sampling issue is the angular sampling "density". In other words, 
how many different angles are measured and over what view and azimuth angle range. In the "goniometric" 
sampling mode discussed in section 3.1 a typical time interval for acquiring a set of BRDF measurements at 
increments of 10° in view zenith angle and 30° in view azimuth angle will be 20 to 30 minutes. The assumption 
must be made that there are insignificant changes in illumination conditions or target conditions during this 
period. An instrument like the PARABOLA, on the other hand, which requires making assumptions about 
spatial homogeneity of the surface, can sample the full BRDF (15° IFOV, 15° angular increments in view zenith 
and view azimuth; 3 spectral wavebands) in only 11s. A commercial version of the PARABOLA, which is 
being dubbed the PARABOLA-II, will sample at approximately 2.5°, 5° or 15° IFOV in approximately 15 s 
(with 7spectral wavebands). 
Among the other instrument issues that must be considered is calibration. As multiple instruments have 
been used in several joint experiments in recent years, including international collaborations, the importance 
of this issue has becomes more apparent. Even with good intercalibrations, the variety of instrumentation, 
sampling techniques, sample periods and other factors can present considerable variability in directional 
reflectances measured over the same target surfaces (Deering, et al., 1992) 
3.3.2 Illumination Conditions. Illumination conditions can be broadly grouped into two categories. The first 
is sun-target geometry effects, including both diurnal and seasonal variations, and the second is atmospheric 
effects. The essential points to make relative to the sampling of the BRDF for a given target surface are 1) that 
the measurement of the directional reflectance properties must be performed for a full range of solar zenith 
angles in order to fully characterize the target BRDF, and 2) that both the surface and atmosphere can be 
expected to change almost constantly. Therefore, both minimizing the unwanted changes in the surface and 
atmosphere during the period of sampling the BRDF and quantifying the changes that do occur in both the 
surface and atmosphere are not only desirable but often critically necessary. Other sun-target-geometry issues 
that will only be mentioned here are the latitudinal constraints on the angular ranges radiometric amplitudes 
available and the topography. 
For remote sensing field measurement campaigns the number one operational problem is unquestionably 
the sky conditions. We almost invariably want perfectly clear skies, and we want them to last all day long. 
More often than not this is not the case. Even in arid environments it is not uncommon to get cirrus cloud 
cover or pollution or dust storm effects. Too often we are also relegated to trying to "shoot the holes" when 
some types of cloud cover pop up during a critical mission. The atmosphere can be quite dynamic during the 
course of a field campaign and monitoring those changes that do occur should be routine. The effects are not 
only on the more obvious spectral dependencies and intensity changes in the irradiance that impinge upon the 
target surface, but also on the nature of the electromagnetic energy interactions with the surface components 
as well. For example, it has been shown that increasing the proportion of diffuse to direct irradiance can 
increase the percentage of the red spectral band energy that is absorbed by a plant canopy as more of the 
incident radiation is scattered deeper into the canopy (Deering and Eck, 1987). Documentation of cloud cover 
and measurements of the atmospheric optical properties throughout the BRDF sample period should be the rule 
rather than the exception.