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