information and taking note of models for this
basic sorting category for review is applications.
report, emphasis was placed on the importance
This is because we are asking what features we
of successful model verification as a selection cri-
are able to model for inclusion in a generated
terion. In this context, verification means all test-
scene: fallen and falling snow, fresh- and salt-
ing and validation, including comparison of com-
water ice, bare soil, vegetated soil, vegetation
putations to common-knowledge expectations or
canopy, clouds and rain ...? While the availabil-
limits imposed by basic physics, comparison with
ity of codes may determine our options in the
"exact" numerical solutions, or qualitative and
end, it is the need to model particular media and
quantitative comparison with physical measure-
configurations that drives both our quest for codes
ments.
as well as our drive to develop new methods. The
The models of MMW propagation, scattering,
bulk of the Applications section is therefore orga-
and emission may be sorted according to three
nized according to applications.
considerations:
The model types provide the broadest and most
1. Model type: physical process and basis, ap-
far-reaching terms in which to consider the field.
proach, methods and techniques used.
Various general approaches are applied in mod-
2. Applications: media considered and valida-
els of substantially different media. The same gen-
tion.
eral types of models also appear in many differ-
3. Computer codes: packages and their origi-
ent computer codes. Therefore, we begin what
nators.
follows with a brief discussion of model types.
For the ultimate purpose of scene generation
In summary: we converge on computer codes
we must deal with computer codes, so we sought
by passing through model types and applications.
existing, facilitated, documented programs. Ide-
Some published models were eliminated at the
ally these should be in integrated, modular pack-
outset because they fail in some serious way to
ages, but single programs are also of interest if
meet one or more of these selection criteria. For
they pose no insurmountable problems to inte-
reasons explained above, the Applications section
gration into larger ensembles in standard lan-
and the next section of this report include items
guages. The codes must be directly accessible, in
that have distinct value but fail in one or another
the public domain or easily available for a mod-
way to meet our criteria ideally. Finally, in the
est cost, and should ideally have the prospect of
Available Computer Codes section, a subset of the
continuing support. Thus, our ultimate constraints
models considered first in terms of type and then
give a certain prominence to producers and ac-
application is presented in terms of specific, avail-
tive disseminators of finished software vehicles.
able computer programs or packages.
At the same time, it is recognized that the physi-
cal problems we are addressing are extremely
difficult. There is no well-established body of
METHODS AND APPROACHES
valid, recognized, elaborated computer packages
as in some other areas, such as structural analysis
Volume scattering
or heat transfer. Typically, the models we desire
In essentially all applications of interest here,
are being generated at the frontiers of research as
some type of randomness must be considered.
pursued at the most advanced research institu-
For volumes, the randomness has been included
tions. For the most part, these institutions have
by assuming that either
neither the resources nor the mandate to produce
1. Dielectric properties vary continuously, with
polished software for dissemination. We recog-
a zero mean fluctuating term added to the
nize also that when energy has been devoted sub-
average permittivity, or
stantially to software polish and dissemination,
2. Dielectric properties have a discrete distri-
scientific rigor may have been left a bit behind.
bution, i.e., certain step changes in value
Therefore, if our ultimate demand is for general-
occur across boundaries of discrete embed-
ity and fidelity to nature, we must also devote
ded elements.
attention to models that are in some intermediate
Idealized representative elements are often as-
or even early stage of development as far as com-
sumed, such as dielectric disks, spheroids, or
puter code is concerned. In the very least, doing
cylinders, possibly with statistical distributions
so provides valuable perspective on the more
of size, shape, or orientation. The simple sum of
available packages.
contributions from individual elements may suf-
Before considering the specific codes, our most
fice to compute the response when the medium is
2
Previous Page
