USING GIS AND REMOTE SENSING IN LAND COVER MONITORING 
Kiira Aaviksoo, Andrus Meiner 
Environment Information Centre, Estonia 
KEY WORDS: GIS, satellite remote sensing, classification scheme, cover class, change detection, change map 
ISPRS Commission VII, Working Group 2 
ABSTRACT 
Results of the first year in using remote sensing for-land cover monitoring will be presented in this paper using one of the landscape 
monitoring test areas - Alam-Pedja Nature Reserve. 
Input data involved satellite images from Landsat 5 Thematic Mapper with dates 06.06.1988 and 12.06.1995, and digital data bases - GIS 
coverage’s in scale 1:50,000, stored in PC ARC/INFO format (layers of rivers, roads, settlement, nature reserve areas). 
Methods used in land cover change detection with integration of two technologies - remote sensing and GIS - involved the following 
steps: I. Pre-processing of satellite data (rectifying images Baltic Map System 1993; normalization of two dates to the similar radiometric 
range; principal component analysis based on TM3, TM4, and TMS bands. II. Classification scheme of 34 land cover classes elaborated 
for land cover change detection in Estonia was adapted for the test area, rich in mires. Computer aided classification was grounded on 
hybrid classification and maximum likelihood algorithms. III. Change detection based on post-classification comparison method. 
Multiple-date satellite images combined with spatial databases and ground truth data allowed us to study landscape change in Alam-Pedja 
Nature Reserve. For the area of 468 km? change map was composed and transitions between main classes: water, natural, seminatural 
and anthropogenic cover types were presented in transition (probability) matrix form. 
Changes of land cover classes reveal main trends in nature reserve and its neighborhood: Afforestation - increasing forest and shrub areas: 
tightening of trees canopy cover in mires; overgrowing of grasslands with shrubs; increasing of fallow land, 
paludification - grasslands and/or fields on lower places where drainage network is not functioning. 
INTRODUCTION 
In the framework of Estonian Environmental Monitoring Program 
the project Monitoring of Landscapes has been initiated in 1996 
(Meiner, 1998). Remote Sensing of Landscapes as one of its 
subprojects has the main goals: 
1) introduction satellite remote sensing technology into 
environmental monitoring, 2) developing a classification scheme 
for land cover types in Estonia, 3) creating a land cover database 
for change detection analysis, 4) carrying out change detection 
analysis on test areas. 
In Estonia there are four main types of nature protection areas: 
national parks, nature reserves, landscape reserves and pro- 
gramme areas. Remote Sensing of Landscapes subproject 
foresees investigation of 7 test areas in Estonia, 6 of which are 
under protection (Aaviksoo, 1998). In 1997 changes in land cover 
from 1988 to 1995 on two test areas - Soomaa National Park and 
Central Estonia (Alam-Pedja Nature Reserve and Saarejärve 
Integrated Monitoring Area) were investigated (Annual Report, 
1997). 
To assess the capabilities of satellite remote sensing for the 
analyses of land cover changes, information on the land cover of 
the study area is of critical importance. The use of satellite data for 
mapping purposes needs the elaboration of classification scheme 
of land cover types. This scheme, based on biophysical charac- 
teristics of plant cover and land use, shall consist of classification 
units separable from the satellite image. The classification scheme 
elaborated for Central-Estonian Landscapes (Aaviksoo, 1995b) 
has been modified in order to use it on the whole territory of 
Estonia (Annual Report, 1996). In the present study, this scheme 
was adapted to the specific territory, where mire types are pre- 
vailing. For the identification of different patterns in landscapes, 
all available materials were used. 
Landsat TM satellite data is the best available material for 
investigation of Earth resources. Estonian Basemap, composed on 
the bases of SPOT images from 1993 in scale 1:50,000 offers good 
material for rectification of the satellite images and for the final 
design of the land cover map - overlaying roads, rivers and the 
other vector data on classified image gave us satellite-based land 
cover map. 
Satellite remote sensing has a short history in Estonia - only some 
papers are published in this field concerning environmental 
mapping of North-East open-cast mining areas (Oberg et al., 
1992), Matsalu bay watershead area (Nisell et a/., 1994) and land 
cover mapping in Vilsandi (Nisell et a/., 1995) and Endla Nature 
Reserve (Aaviksoo, 1995b; Aaviksoo et al, 1997). 
Basing on the experience in using remote sensing technologies to 
investigate change in Estonian landscapes (Aaviksoo, 1993, 
1995a, Aaviksoo et al., 1996, Mander et al., 1996) we used 
Estonian Basemap GIS coverages and /DRISI GIS modules to 
map land cover change. The main objectives of this study are: 
1) application of classification scheme elaborated for land cover 
change detection in Estonia for the test area; 
2) quantitative and qualitative characterization of the test area in _ 
different years; 
3) integrating GIS and remote sensing environment for com- 
posing change map and calculating transition areas. 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998 53