'se
al.
)O-
ch
nd
se
| is
ap
ed
ler
on
1S,
ed
ot
al
ed
or
1d
en
te
en
er
he
by
fic
38 ——
quu à (WS u
i CHUBUT
; SANTA
52 — m
feet 100 km
q
56
Fig. 3 Localization of Transects and Data
Processing Units
of the latest sensors (like ERS-1, for instance) may be
checked in these areas.
The final system for data acquisition in the transects
is displayed in Fig. 4.
There exist 4 marked data lines:
1. Interpretation of LANDSAT-MSS imagery
2. Digital multispectral classification
3. Geocoding
4. Digitizing of line maps
All four data streams flow finally into data bases (GIS
archives) after geocoding them to a common system.
A prerequisite for effective and correct interpretation
or classification is ecological knowledge of the
ecosystem of the area. The respective information is
drawn on maps (topographic work sheets - 1:250 000),
plotted on the base of digitized topographic maps.
The gaps between transects are filled by interpolation.
Here NOAA data are applied (see paragraph 6).
Beside the transect approach coupled. with interpre-
tation of NOAA imagery, another arrangement will
increase the efficiency of data processing in the large
area: the decentralized distribution of data processing
laboratories. This has been arranged due to the
conviction that:
1. Regional groups should perform the data
processing since they have a deeper knowledge of the
local desertification indicators
2. Summing up the potential capacity of a number of
separate units will be more efficient than burdening
the central unit with the whole work
3. A central laboratory is more exposed to possible
desasters than the separate ones together
4. The learning effect is higher in smaller individual
units as more operators are involved for the whole
project. Exchange of experience stimulates people.
In fact, 3 data processing laboratories have been
installed in Patagonia, as shown in Fig. 3. All of them
are with the local representation of INTA.
In Bariloche, two units are available; outside
Patagonia there exist two more, one at the University
of Buenos Aires/IFEVA (Prof. Soriano), another
in Mendoza with IADIZA. The basic hardware
configuration, nearly identical for all laboratories, is
shown in Fig. 5.
The PC-controlled system does not include a
precision recorder for image output but only an ink
jet plotter for quick look.
Software in ERDAS, as raster processing seems to
be more adequate than vector processing. This is
because the structure of ecological data corresponds
better to pixels than to lines since limits are never
,exact^ like for instance in cadaster. Moreover, many
. of the basic data - like satellite imagery - are in
153
raster format. Nevertheless, the many different data
involved require a vector system, too (ARC/INFO).
It is mainly welcome for digitizing conventional line
maps, but also for registering non-spatial data. The
final ultimate data base (,archives“ in Fig. 4) is in
raster format after vector-raster transformation of the
ARC/INFO controlled digitized line maps.
.6 Preliminary Results
Operational work is performed in Argentina,
principally at the INTA stations for the particular
transects. These activities are accompanied by special
research at the University of Karlsruhe.
An important question refers to the role of NOAA
data. High temporal resolution is a positive factor,
whereas the geometrical resolution of less than 1 km
represents a limiting factor. Nevertheless, NOAA data
may be used for
Interpolation between the transects
Detecting seasonal changes in Patagonia as a whole
As a first step, a poster in 1:1.5 million was produced
in order to give a general overview of Patagonia. The
result shows obvious regional differences especially
the degradation effect starting from the Andes
+ eastbound and the basaltic central meseta, starting
from 40 S southbound.