CHANGES oF THE FOREST iN WESTERN SUDETY MOUNTAINS iN POLAND IN 1976-1990 PERIOD 
DETECTED on LANDSAT MSS AnD TM IMAGES 
Zbigniew Bochenek, Andrzej Ciolkosz, Maria Iracka 
Institute of Geodesy and Cartography 
Remote Sensing and Spatial Information Centre - OPOLIS 
Warsaw, POLAND 
Commission VII, Working Group 2 
KEY WORDS: remote sensing, satellite images, forests, damage, pollution 
ABSTRACT 
Different methods of classification of satellite images were analysed, in order to derive information on the state of forests 
in southwestern Poland heavily affected by air pollution. Detailed information concerning type and quality of stands, 
derived from field inventories and aerial colour infrared photographs, was used for assessing accuracy of classification of 
satellite data. Landsat TM images enable characteristics of health conditions and structure of stands, permitting 
discrimination of three levels of damage and various phenomena, which accompany forest decline. Comparison of the 
classified images with forest maps revealed good consistence of classes obtained from satellite data with those derived 
from forest maps. Major changes were detected between 1976 and 1990. 
1. MATERIALS AND METHODS 
1.1 Study area 
The study site covers the western part of the Sudety 
Mountains, in the Karkonosze and lzerskie Ranges, 
containing a large forest complex of spruce (Picea 
excelsa). The Sudety Mountains, located on the borders 
of Poland, Germany and the Czech Republic, were 
chosen as a test site because forests at this region in 
Poland are seriously damaged (Ciolkosz and Zawila- 
Niedzwiecki, 1990). Forest degradation is caused by a 
concentration of brown coal open-pit mines and electic 
power stations emiting a large volume of polltants into the 
atmosphere. The damage is so severe that the region has 
been concidered an ecological disaster called the "Black 
Triangle". 
1.2 Satellite image systems 
The first stage of the study examined the usefulness of 
high-resolution satellite images for classifying health 
conditions of forest, species composition and other forest 
phenomena (clear-cuts, open canopy closure, etc.). In 
order to fulfil this task, Landsat TM data collected in July 
1984 and in August 1990, as well as MSS image 
collected in May 1976, have been selected. The analysis 
was done with the use of ERDAS image processing 
system. This system enables interactive analysis of 
satellite images, i.e. radiometric and geometric correction, 
location of training areas on Satellite images, 
determination of spectral signatures and data 
classification. 
2. FIELD WORKS AND AERIAL PHOTOGRAPHS 
In order to determine relationships between field 
characteristics of stands and their signatures on satellite 
images, field evaluation of spruce stands was done at 
preliminary stage of the study, using over 500 ground 
sites. Field appraisal was performed on the basis of 
methods prepared by the Forestry Research Institute 
(Zawila-Niedzwiecki, 1994). Each ground site covered 
homogeneous appraisal unit (1-10 ha), which was 
characterised through direct measurements and 
observations. The following information was collected in 
the field: 
® description of terrain conditions (type, relief, slope, 
aspect, type of cover and degree of coverage by 
rocks, stones, debris); 
€ description of stand (structure, layer, species 
composition, age, form of mixing, closure, average 
breast-diameter, mean height, height of crown base, 
number of trees per unit area, including dying and 
dead trees); 
€ evaluation of damage of stand (defoliation - Df, 
discoloration - Dc, quality of trees described by vigor 
of tree crown - Dm ). 
Defoliation and discoloration was evaluated in accordance 
with International Co-operation Programme on 
Assessment and Monitoring Pollution Effects on Forests 
(ICP) recommendations, while quality of trees was 
assessed on the basis of state of assimilatory apparatus, 
height increments and vigour of trees. 
The numerous relations were studied between 
stand/terrain characteristics and spatial signatures of 
stands, derived from Landsat TM satellite images, taking 
into account elevation/aspect information. The following 
conclusions were drawn from these studies (Zawila- 
Niedzwiecki, 1994): 
€ with increasing elevation tree height and the height of 
tree crown base lowered; trees of lower quality 
(moderately and heavily damaged stands) are more 
closely correlated with elevation, than less damaged 
trees. 
€ there are significant correlations between spectral 
response in Landsat TM bands and some indices of 
quality of stands (Dr, De, Dm). The strongest relations 
exist for defoliation and quality index (vigor of trees), 
while discoloration is less correlated. Usually the 
relations are defined through multiple regression, 
which involves use of 2-4 spectral TM bands (most 
commonly visible, near infrared and middle infrared 
bands). 
Intemational Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998 801 
A ee eS SRR : © mc cr