13
dependent on the
/er and on the
and microrelief,
control the
Hence, auto-
i very rare,
thonous geo-eco-
ie area. So one
itions Arabidetum
sslerio-Caricetum
Lcate limestone
garnet muscovite
3 is due to the
nation of these
-bearing surface
Lon of carbonate-
3. The above
sales cause a
r of the stati-
3. Therefore the
lological units
.fications, which
-iations in the
leless, it has to
type and the
ihe random sample
influence the
lis understanding
;t limitations in
>te sensing for
lis application
itself as a test
theories about
■eas where rocks
ie classification
'ises up to some
tly the similar
if the various
prevent better
illy valid for
,e and rauhwacke.
icks is studied
show significant
ieir weathered
lly be classified
ie lithological
cover could be
y than others
‘ormed part of an
ons cannot be
at 27,3 % of the
i a 1" and 20,9 %
were classified
rhaeticite and
to some extent,
ial deformations
p, which only
by inaccuracies
In general, the
ping will rather
itional than by
o it could be
ering conditions
that in places
gement of the
and soil map
the formalized
ion 2) than from
of distribution
drawing of
more or less
ans, always be
In the present
boundaries in
out using an
an additional
The aspects of
n, which is
erformed during
Table 1.
Classification results of airborne multispectral
percents).
scanner data for various lithological classes (in
No.
Material
N
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
1
Gabbro-Amphibolite
61
100
2
Prasinite, Garnet-Bearing, partly Eclogitic
1.477
45,8
0,3
33,0
14,5
1,9
0,1
0,6
0,1
3,4
0,1
0,2
3
Peridotlte, Serpentinite
6.743
0,7
2,7
67,9
0,0
5,8
1,4
1,0
0,0
14,3
0,3
0,2
0,0
2,7
0,3
1.9
0,6
4
Chlorite-Talcum-Hornblonde-Schist
72
100
5
Quartzite, Rhaeticite, Graphitic Schist
837
91,2
0,8
1,4
0,5
1,0
2,3
0,4
2,5
6
Calcareous Mica-Schist
1.390
2,1
6,1
4,7
79,2
0,7
0,4
0,1
2,5
0,2
0,1
2,9
0,4
0,4
0,2
7
Garnet-Bearing Muscovite-Schist and Phyllite
4.847
0,3
15,2
3,2
1,8
35,9
28,8
0,9
1.3
4,4
0,4
0,2
0,1
1.3
5,5
0,1
0,1
0,7
8
Dark Mica-Schist and Phyllite
9.626
0,6
2,5
3,7
0,8
0,5
8,3
7,7
20,8
4,1
20,1
0,4
1,1
0,0
5,5
3,4
1,3
15,9
0,8
2,4
9
Quartz-Rich Breccia
177
4,0
1,1
93,2
1,7
10
Quartzite
4.200
0,1
0,4
4,3
0,0
5,7
3,4
2,7
0,5
68,9
0,0
3,4
5,7
0,1
1,3
1,9
0,6
0,9
11
Marble
4.706
0,3
0,6
4,6
0,9
2,0
1,0
0,2
3,3
64,8
4,0
11,0
2,8
2,3
0,1
2,1
12
Dolomite
3.426
0,0
1.8
3,8
1,1
0,5
2,5
0,3
12,0
21,3
17,4
14,5
10,0
0,7
0,3
12,2
0,8
0,7
13
Rauhwacke
1.250
0,2
0,1
1,0
0,3
17,7
5,6
69,0
4,6
1,6
14
Moraines: Fernau, Egesen, Daun, Older
40.150
1,6
0,1
5,1
0,6
9,9
3,0
0,3
4,2
0,1
25,0
0,5
3,6
0,4
22,6
2,2
7,3
6,4
2,3
4,8
15
Moraine of 1850
1.378
0,7
2,0
0,3
2,6
0,1
5,0
88,9
0,5
16
Peat and Bog
527
20,9
3,2
0,4
1,3
62,4
1,5
0,2
10,1
17
Ancient Rock Fall
3.844
0,1
2,3
1,8
0,8
5,3
0,1
4,4
1,2
2,4
0,0
4,8
1,1
0,1
70,6
1,3
3,7
18
Talus Fans
22.624
0,3
3,0
9,7
0,1
0,8
8,8
7,0
6,5
0,9
17,1
4,6
10,3
2,9
5,6
8,5
1,4
5,5
6,2
0,8
19
Recent Alluvial Material
2.024
0,3
27,3
0,0
0,6
1,4
0,7
1.9
0,5
13,4
2,7
0,4
50,6
field mapping, have also to be taken into
consideration. For the above mentioned
example e.g., the misclassification is due
to wrong geometric demarcation of the class
"quartzite, rhaeticite and graphitic
schists".
The mutual misclassifications of the classes
"peat and bog" and "recent alluvial
material" can probably be explained by a
certain generosity during field mapping as
well as through a change in spatial
extension of these two classes between the
time of field mapping (around 1930) and the
remote sensing data acquisition (1979).
These changes during time are also valid for
the Grossglockner-Hochalpenstrasse (pass
road). The deviations in its course given
in the lithological map are not only due to
cartographic generalization but mainly a
result of reconstructions performed in the
meantime and slight changes in position
related to these activities. If there had
only been a cartographic generalization that
effected all objects of the study area to
the same extent, the classes "gab-
bro-amphibolite, chlorite, talcum,
hornblende-schists" could not have been
located so precisely and reclassified with a
100 % accuracy.
An additional problem represents the
different empirical content of identical
classes in different thematic maps. The
class "talus fans" e.g., has been
distinguished in all four maps used. The
common areal coverage, however, only amounts
to some 10 % of the total of all individual
talus areas. It is a well known fact that
in classical (uncovered) geological maps the
term detritus (talus/scree) is differently
used from geomorphological maps.
Nevertheless it is surprising that in the
study area there also exist considerable
differences between the soil and the
vegetation map (80 % of the detritus of the
vegetation map are also distinguished as
this class in the soil map; however, only
16 % of the detritus of the soil map are
classified as such in the vegetation map).
It could be clearly demonstrated, that the
congelifraction zone in the soil map was
mainly asigned through the class detritus,
thus simply omitting the initial soil
formation, whereas in the vegetation map
these areas were asigned to various plant
associations with a low degree of density.
This example may point out, that prior to
such classifications the theoretical
background and the conceptual contents of
the used class names have to be clarified.
This aspect is at least as important as the
selection of an adequate classification
method.
As shown in Table 1, due to its heterogenous
spectral properties, the material described
by the term detritus as used in uncovered
geological maps can not or hardly be
reclassified. A splitting of this category
in at least three sub-classes (detritus
below soil and vegetation cover, detritus of
bright rocks, detritus of dark rocks) yields
an improvement of the classification
results. Rock detritus under soil and
vegetation cover can only be sufficiently
reclassified in those places, where it
causes a differenciating effect in the
ecosystem, e.g. a high rate of
subterraneous run-off and therefore a dry
habitat.
A similar problem arose in the
classification of genetically different soil
types. Many soils in the test area are in
the process of pseudogleying and are in the
soil map correspondingly distinguished as
pseudogley and pseudogleyed brown-earth
respectively. Due to the effects of the
macro- and microrelief this type which has
been distinguished according to the
scientific rules of the official Austrian
soil mapping, can be split into ecologically
different subtypes, which do not have an
official character and are therefore not
displayed in the soil map but can be
excellently determined from the airborne
scanner data. The attempt to reclassify the
superior main soil type failed due to the
heterogenous statistics of the training
areas.
The same problem with reversed premises
occured in the reclassification of the
vegetation associations. Here often
associations are distinguished, which are
only different from each other through the
