For vegetative objects some theoretical model is needed which 
correlates bidirectional reflectance of a plant canopy or field, as 
measurable by airborne scanners, and the laboratory reflectance spectra 
of the constituent plant components. For geological objects a model is 
needed which will permit assessment and/or correlation for variations 
in texture of rock surfaces across the scene as they affect discrimina- 
tion among certain rock types. Suits [11] in an important development 
has devised a canopy model of the type needed which allows for more than 
one canopy layer and which relates laboratory spectra of plant components 
and approximate plant geometry (Planting density and average horizontal 
and vertical component cross sections) to the bidirectional reflectance 
of a crop field which it predicts without atmospheric effects. The 
model has been verified [12, 13] for the case of mature corn by field 
reflectance measurements at different angles over the visible and near 
infrared spectral bands. 
The problems encountered in geological remote sensing are different 
in several aspects from vegetative problems. First, temporal effects 
are much slower for geological than for vegetative objects. Secondly, 
for rock-type identification the spectral emittance or reflectance 
variations are much more important than geometrical variations (shapes, 
shadows, observation angle, etc.) across the scene, whereas both are 
relatively important for the identification of vegetative classes. 
Thirdly, the thermal IR spectral region contains more chemically 
diagnostic information about rocks than the visible-reflective IR wave- 
length regions, whereas the converse is true for vegetative classes. 
Rocks and minerals present a special problem in the thermal IR 
region not normally encountered in the shorter wavelength regions. 
Thermal IR wavelengths (A & 10 um) approach in size the particle 
diameters of some of the grains in fine-textured rock surfaces, which 
produces some complex optical phenomena related to surface roughness. 
For instance, it has been noted by Lyon [14] and other investigators 
that the spectral emittance within the major reststrahlen bands 
(intramolecular vibration modes in this case) of silicate rocks and 
minerals tend to increase with decreasing particle size. Later work 
[15] has shown that in spectral regions of moderate to small complex 
refractive index, outside the reststrahlen bands (high complex refractive 
index), the emittance can either increase or decrease with decreasing 
particle size. 
The spectral emittance dependence on particle size is an important 
factor in geological remote sensing, because textural variations from 
rock to rock may mask differences in chemical composition and vice 
versa. It is necessary, therefore, to separate textural and chemical 
effects on the spectral emittance, as much as possible. To do this, a 
model of rough rock and mineral surfaces is required which can at least 
qualitatively explain the effect of textural variations on the IR spectrum 
of those surfaces. The resulting calculated spectral emittances are 
specifically needed to assess what affects textural variations have on 
discrimination techniques. 
10