5, and Gravel
r barren land
to be detected
activity was
ing to its
the more humid
s more difficult
ealed by the
fects of
al. Small
r imagery and
d use maps of
by J. R.
Resource Areas
(1965). In
could be made
bility were
s requisite to
ults obtained
is created from
hods. Eight
ren of eleven
and use divi-
'adar land use
in that derived
ould be termed
regions.
11ified endorse-
imagery could
iting borders
je employed in
le and/or
1ished detect-
»e fragmented
combination with
nore homogeneous
landscape.
ks in land use
ive in area nor
ccurate, if
attern could
s not possible
;itwasa
Summary and Observations
This paper has examined some of the effects of the environmental
modulation transfer function (EMTF) in mapping land use with radar
imagery. Although some shift was expected the degree of variation was
surprising. No precise mathematical function can be generated at this
time, but it is quite evident the Northeast is not as easily mapped as
the more extensive, open land use activities of the West and Midwest.
Level I land use categories could be detected in both study areas,but
more detailed analysis quickly precipitated a disparity in the data sets.
It is believed the paramount factor affecting detectability and identifi-
cation was the omnipresence of forest vegetation abetted by a strikingly
different settlement pattern. In the Northeast, trees and forest canopy
concealed drainage features, topographic variations, transportation
arteries, urban and rural settlement patterns, and field borders.
Regrowth and scrub vegetation reduced the clarity and heterogeneity of
cultivated fields, pastures, wetland, and idle land so apparent in the
Midwest and West. In cases where the landscape element was not entirely
masked the combination of the element and vegetation within the resolu-
tion cell tended to produce a scrambled or "averaged" signature. Larger
urban areas were visible in the Northeast but the extent of built-up
and urban sub-categories proved difficult if not impossible to determine:
the result of the continual presence of trees within and abutting developed
areas. Whereas medium sized towns and small villages were visible in
Study Area II, as were most transportation arteries, it was not possible
to consistently detect similar features in Study Area I. In the latter
area Level II detail is visible only on a random basis.
Results of this study provide empirical evidence supporting the
above cited EMTF formula by Everett and Simonett. Specifically, as land
use activity becomes more complex or intensive the level of difficulty
in identifying features rises rapidly. However, it is also apparent from
the effects of forest vegetation in the Northeast that a simplistic en-
vironment of few categories can present an equally formidable condition.
Predicated on these observations it is critical that much caution be
exercised in selecting the scale of imagery to be used, categories to be
inventoried, and cell or polygon unit employed to record land use data.
Obviously, the disparity noted in this study is not universal.
Highly satisfactory results have been obtained over capacious tracts as
is manifest by the success of project RADAM in Brazil and similar efforts
in Venezuela. Note also the urban detail extant in the X-band image of
Indianapolis, Indiana (Figure 9). Nevertheless, this study does point to
the need to carefully consider and examine the environment and its
composition before attempting to map land use.
While the list of environmental factors could be almost infinite
it is believed the dominant influences (as attested to in the Everett and
Simonett formalization) are a function of: (1) topography and surface
configuration, (2) vegetation pattern and general morphology (e.g.,
riparian, density, etc.), (3) settlement pattern and history, (4) types
of agriculture and crops present along with land ownership, (5) range of
field sizes and shapes expected in conjunction with their frequency and
periodicity, (6) the economic condition--that is, is the area now a major
