×

You are using an outdated browser that does not fully support the intranda viewer.
As a result, some pages may not be displayed correctly.

We recommend you use one of the following browsers:

Full text

Title
Remote sensing for resources development and environmental management
Author
Damen, M. C. J.

10
SOLAR
PROCESSES
SOLAR ¿L'Wr PROCESSES
>►
H
LATERAL
PROCESSES
ENDOGENOUS
PROCESSES
ENDOGENOUS
PROCESSES
TIME
/ TIME
F-
f-
SPATIAL REFERENCE UNIT J
SPATIAL REFERENCE UNIT J+1
Figure 2. Simplified scheme of the approaches in geosciences (horizontal structure) and in ecosystems research
(vertical structure). Compiled from various sources. Explanation see text.
intradisciplinary interest but also of
unquestionable practical value.
In the classic geoscientific way of thinking
in the "horizontal structure" (see Figure
2), out of the horizontally, mainly con
tinuously varying study objects,
quasi-homogeneous units are distinguished in
a typological manner. The type-limitation
of these units, which is in part based on
the processes occuring inside them, can be
relativated with regard to the respective
problem, and usually have a very strong
relation to theory. Nevertheless, they also
ought to be examinable empirically. The
spatial delineation is mostly problematic
and, moreover, subject to geometric
inaccuracies in mapping and to cartographic
generalization. Conclusivly, these units do
not have the same empiric evidence and are
by far not as clearly defined as the common
categories of "ground truth" for the
classification of the earth's surface.
Using operational methods of remote sensing,
some selective aspects of the sensor-exposed
geo-components are also described on the
basis of physical theories on elec
tro-magnetic waves and can be combined to
classes. Taking this information , we can
e.g. make statements on the water and
pigment contents of plants, or the water
content, the texture, and the percentage of
organic matter and ironoxyde in soils.
Therefore it is at hand, to put the spatial
patterns in relation in order to widen the
horizon of information in a very general way
and, if possible, to increase the certainty
of the above mentioned indirect deduction.
It will not be possible to substitute one
method by the other; but it seems that for
some questions they can complement each
other quite well.
The present paper mainly discusses the
automatic classification of remote sensing
data according to categories of thematic
maps. Results of visual interpretations are
to a great deal determinate by the level of
field knowledge of the interpreter. If the
latter is reasonably high, there is
principally no argument against. For the
clearing of the above mentioned questions,
however, the visual interpretation is
completely unsuitable, as in particular this
"pre-comprehension" - the "theoretical back
ground" - which to a great deal contributes
to the geoscientific typification, and so
has to be subject of the investigation. It
is easier to estimate the influence of
insufficiency of the computer-assisted
classification than the amount of
subjectivity of field expertise. The latter
is only taken into account for the
classification process if - like with the
catena principle (Opp 1984) by means of the
digital terrain model - it is formally
describable.
3 GENERATION OF REGISTERED DATA SETS
Within the scope of the Austrian Unesco
Program "Man and the Biosphere (MaB)"
various available data have been transformed
into a unique reference system, using the
software available at the Institute for
Image Processing and Computer Graphics in
Graz. This made it possible to compare the
different information contents of the
individual data sets in a raster on a pixel
by pixel basis.
3.1 Basic
The follow
Airbor
the 1
German
coveri
2x2
- Landsa
- Landsa
- two co
(film
a topo
1982) ;
- a ve
cover!
1980) ;
a soil
a geol
photog
map 1
an eco
prep.)
3.2 Data pi
3.2.1 Digit
photc
The two CIF
the study
cm2 were di
f ilmwriter.
quantized i
a magnetic
3.2.2 Geom
imag
In order to
the map d
the airborn
into the sy
case was th
the differe
sets had to
Whereas th
central pr
data were
earth surfa
was done
GAMSAD (K,
interpolati^
neighbour p:
falsificatii
The identif:
points in tl
the clearly
exclusively
which more
vertical ai
very unfavoL
points extr
the images v
slight devia
the rectifie
3.2.3 Digiti
The digitizi