tally and up/down through profiles of clear
tively distinct edges. On the other hand, the sheer
air, turbulence, dust, clouds, and precipi-
variety of features and their distinctive forms is
tation.
so great as to prevent much generalization.
Vegetation
For well-defined structures and their environ-
Low: Grass, fields, agricultural plantings,
ments it may sometimes be possible to use vari-
shrubs, mostly single layer with diminished
ous programs in the suite available from the Geor-
importance of branches, greater importance
gia Tech Research Institute (Peifer et al. 1988, Faust
of leaves, blades, stalks.
et al. 1989, Davis et al. 1992). This graphically
Tree canopy: Usually but not always two-
facilitated package offers capabilities for defining
layer, continuous/intermittent crown with
structure geometry and then calculating returns
branches and seasonal or evergreen leaves/
in terms of combined PO responses of an assem-
needles, lower layer of trunks, horizontal
blage of simple shapes (scattering centers). In less
and nonhorizontal propagation; as for low
deterministic situations we may consider the pos-
vegetation, possible significance of under-
sible presence of highly distinct structures at the
lying ground/water surface.
subpixel level, that is, a terrain category of
Snow and ice
"patches," each of which may be relatively easy
Snow: Shallow/deep, uniform/layered,
to model in itself but that must somehow be inte-
wet/dry, smooth/rough.
grated statistically into an image effect. Rational,
Ice: Fresh/saltwater, smooth/jumbled,
or at least rationalized, schemes exist for con-
first-year/multiyear.
structing "fading" (speckle) in images from this
Exposed terrain and water bodies
sort of process. Such schemes can be based on
Rock: Rock-strewn terrains, or more mono-
reasonable or at least specific physical assump-
lithic formations that may be smooth ex-
tions and procedures (e.g., Ulaby et al. 1982, 1986;
cept at MM scale, or may be steep, jagged
Yueh et al. 1991). While they come into play dur-
outcroppings.
ing the final steps of the scene generation se-
Soil: Wet/dry, rough/smoothly undulat-
quence, we consider them to fall within the fringes
ing, multiscale roughness.
of the domain of modeling that we are pursuing
Water: Smooth/rough, moving/still, ex-
here. For this reason, no specific review of ap-
pansive/narrow.
proaches and programs and no recommendations
Subpixel patches: Mud puddles, ice patch-
in this area are included in this report.
es, thaw/wet patches in snow ...
Some expansive cultural features such as lawns,
This list is something of a litany of objectives,
bare tilled soil, or crop fields may be addressed
as context. What follows is a summary of what
under other headings below.
has been done toward addressing this list.
Atmosphere
Many of the radiative transfer approaches listed
Cultural features
Clearly one can attempt to analyze specific con-
below can in principle be applied to atmospheric
structed elements such as wires, stairs, or win-
dust and precipitation. Ishimaru and his co-work-
dows. Here, however, we seek more generic ca-
ers have developed models that are in principle
pabilities applicable in relatively indiscriminate
applicable to rain or ice precipitation (Ishimaru
remote sensing of a large-scale environment. Rela-
and Cheung 1980; Ishimaru et al. 1984a,b), and
tively little has been done to model lower-fre-
others are directed at turbulence effects (Honig et
quency microwave sensing of large-scale con-
al. 1977, Ishimaru and Painter 1980). McMillan et
structed or cultivated environments, and even less
al. (1982) review applications of theory to mea-
is available in the public domain at MMW fre-
surement for MMW propagation. They conclude
quencies. A few measurements have been reported
that while in most cases agreement is "plausible,"
(e.g., Violette et al. 1988), but without substantial
data have been inadequate to determine the ex-
attempts at modeling. Beyond the obvious (such
tent to which this is fortuitous. Weinman (1989)
as considering buildings or curbs as corner reflec-
applies a RayleighGans model of MMW scatter-
tors) we draw no inferences here for modeling
ing from falling snow, with flakes considered to
purposes. We note that the constructed or con-
be a polydispersion of equivalent randomly ori-
trolled nature of cultural features offers greater
ented circular disks. Disk dielectric constants are
regularity to the analyst on the one hand: swim-
obtained from the Bruggemann mixing rule with
ming pools, walkways, and even lawns have rela-
estimated ice fractions in the equivalent flakes.
9
Previous Page
