- 1734 -
Summary of Application to:
Forest Exploitation Along the Haiti and Dominican Republic Border
Darrel L. Williams Mark L. Stauffer Belden G. Bly Peter Van Wie
Background and Objective of the Test: There has been considerable interest in utilizing
Landsat digital imagery as a source of reliable, up-to-date information for conducting re-
gional or national-scale land use assessments. The dollar and informational value of the
Landsat imagery for this purpose is directly dependent upon the quality and availability of
alternate data sources. Land use information extracted from Landsat imagery has the great-
est potential value in those areas of the world where data describing various physical, eco-
nomic and cultural characteristics is not currently available. This situation is particularly
true in many of the lessor developing countries. Before these countries can initiate programs
to stimulate their economic and cultural growth, accurate assessments of their current re-
sources and land uses must be conducted on a national scale. In many instances Landsat
imagery is the only potential up-to-date source of this information. Thus, as a test applica-
tion, digital analyses of Landsat imagery for the delineation of current land use in two
provinces of the Dominican Republic was conducted at NASA/GSFC in cooperation with
the U.S. Agency for International Development (AID). The specific objectives of the NASA/
GSFC study were: a) to perform digital analyses of Landsat data for two test sites within
the Dominican Republic and derive land cover classification maps for each; b) to provide
the classification results and ancillary data in a format compatible with a data base manage-
ment system being developed by AID; and c) to apply various sub-sampling schemes to the
Landsat digital data to assess how such sampling would affect the validity of subsequent land
cover maps. A discussion of the analysis and results for one of the test sites, the Dajabon
Province, is presented here relative to its value in detecting forest exploitation.
Study Site: The Dominican Republic occupies the eastern part of the Caribbean Island of
Hispaniola, covering an area of 48,279 km2. Dajabon Province is located in the northwest
corner of the Republic and its entire western perimeter forms part of the border with Haiti.
The physiography, land use and climate within the 2,000 km? province is quite varied. The
northern section of the province is relatively flat and dry, quickly giving way to rolling hills
dominated by mixed agricultural corps and pastureland. The southern portion of the pro-
vince is dominated by steep, forest-covered mountains which rise to nearly 3,000 meters.
The economy of the area is depressed and an ever increasing amount of the remaining forest
land within the province is being cleared to plant agricultural crops and to provide charcoal
for cooking and heating. This phenomenon has occurred in an even more accelerated
fashion just across the border in neighboring Haiti, with the result that the Haitian mountain
slopes are practically devoid of forest vegetation and are severely eroded.
Approach and Results: The distinct contrast in vegetative cover characteristics along the
international border was apparent on Landsat imagery, especially on a geometrically recti-
fied and contrast stretched MSS band 5 sub-image of the study area (Fig. 3.1a). Thus sub-
image was extracted from a relatively cloud-free 3 November 1973 Landsat image for the
frame designated as path 8, row 47. This data were resampled to produce 67 by 67 meter
cells which would readily interface the subsequent classification results with AID's data base
management system. A euclidean distance algorithm was utilized to classify the resampled
Landsat digital data using spectral signatures and associated statistics obtained via training
areas a
and m:
tailed,
virtual
with t]
Howe
brushl
discre;
vision:
broad
spectr
3.1b).
3.1b)
of the
est co
tones
inanc«
occur
The fi
into 1
cellw
tional
sampl
Alk
was g
noise
forest
The s
2% 0!
an ap
large-
Furtl
Re