400 
FOREST MAPPING IN BAVARIA, USING SATELLITE IMAGERY AND INTEGRATING ADDITIONAL MAP IN 
FORMATION: RESULTS IN THE THREE MAP SHEETS 1:200 000 
REGENSBURG, NUREMBERG AND AUGSBURG 
M. Keil, M. Schardt, A. Schurek, R. Winter 
German Aerospace Research Establishment, Oberpfaffenhofen, FRG 
L. Ernesti 
Institute for Geography, University of Göttingen, FRG 
H. Sagischewski 
Institute for Geography, University of Marburg, FRG 
ABSTRACT 
A large-area forest mapping was carried out in Bavaria/FRG, based on Landsat Thematic 
mapper data. Besides the main classes - deciduous, coniferous, and mixed forest - , 
the separation of spruce and pine was included. In order to get reliable results for 
an additional class of unstocked forest areas and not yet closed cultures, information 
of topographic maps was integrated. 
This forest mapping will support conventional forms of large-area forest inventory and 
covers the three map sheets 1:200 000 of Regensburg, Nuremberg and Augsburg, each 
showing an area of about 90 km by 95 km. 
There are discussed the methods and results of forest/non-forest separation, signature 
analysis of training areas, classification, and verification. In order to take regio 
nal changes of reflection values because of influences of geological site conditions 
and atmospheric conditions into account, additional information of forest growth maps 
was integrated in the classification process. 
KEY WORDS: Forest Mapping, Landsat Thematic Mapper, Bavaria, Tree Species, Maximum 
Likelihood Classification, Verification 
1 INTRODUCTION AND OBJECTIVE 
Forest mapping using satellite imagery 
within Bavaria was part of the research 
project 'Investigation and Mapping of 
Forest Disease Using Remote Sensing 
Methods', which the German Aerospace 
Research Establishment (DLR) carried out 
together with several forestry 
institutions. The main objective of the 
entire project was to evaluate 
multispectral digital data from aircraft 
scanners and satellite systems to study 
forest disease (Landauer, 1989 a, Land- 
auer and Vo£, 1989 b). 
The objective of the present project was 
a large-area forest mapping which was to 
cover large parts of Bavaria in a scale 
of 1:200 000. The classification by 
Landsat TM data was to include the main 
classes of coniferous, deciduous, and 
mixed forest, and, in addition, the 
discrimination of spruce and pine 
stands. 
On the one hand, the studies were meant 
to show the possibilities of satellite 
imagery for supporting conventional 
forms of large-area forest inventory. On 
the other hand, the studies were also to 
serve as a basis for future forest 
disease mapping. Regarding the long-term 
objective of using satellite imagery for 
large-area forest disease mapping, it is 
first of all necessary to register the 
influences of various forest stand para 
meters on spectral signatures over ex 
tended areas. Such stand parameters, 
like tree species distribution, types of 
mixtures, natural age classes, canopy 
density and ground cover were included 
in the investigation. 
2 TEST SITES 
The Regensburg map sheet 1:200 000 was 
chosen as the first map for large-area 
mapping. The area represents a 
diversified geological structure which 
is expressed by the different growth 
zones and growth districts. 
The area includes extended stands of 
mixed deciduous and coniferous forest 
(esp. in the growth district 'Franconian 
Jura' on the Jurassic limestones) and 
also pine-rich stands (esp. on the sandy 
soils of the 'Upper Palatinate Basin') 
besides spruce as the main tree species 
(esp. in the 'Upper Palatinate Forest' 
and the 'Western Bavarian Forest'). 
Therefore the area was well situated for 
the investigation of stand type 
discrimination.