LAND COVER CHANGE DETECTION BY POST-CLASSIFYING FOR MULTI-TEMPORAL LANDSAT TM DATA 
CHEN Xiuwan 
Associate Professor, Institute of Remote Sensing and GIS, Peking University, Beijing 100871, China 
Shintaro GOTO 
Associate Professor, Environmental Information Research Lab., Kanazawa Institute of Technology, Japan 
Chung-Hyun AHN 
Senior Researcher, Systems Engineering Research Institute, Taejon 305-600, Korea 
Commission VII, Working Group 5 
KEY WORDS: Land Cover Change, Detection, Post-Classification, Maximum Likelihood, Satellite Data, Landsat TM, Multi- 
temporal. 
ABSTRACT: ; 
Land cover change information are very useful in many applications such as flooding monitoring, deforestation assessment, 
urbanization mapping, etc., and in decision making of sustainable development management. Detection of land cover changes is 
one of the most interesting aspects of the analysis of multitemporal remote sensing images. There are two main approaches to 
detect land cover changes from remotely sensed data: (1) by comparing the spectral reflectances of multitemporal raw satellite 
images, and (2) by using post-classification methods. The later approach was used in this study to detect yearly land cover changes 
from four dates of Landsat Thematic Mapper (TM) data acquired from 1985 to 1993 in Ansan City, Korea. 
Replacing to develop a new algorithm for land cover classification, this study pursues to select a suitable algorithm from the most 
common and reputable algorithms: (1) Parallelepiped (PAR), (2) Minimum Distance (MID), (3) Mahalanobis Distance (MAD), (4) 
Maximum Likelihood (MAL), and (5) IsoData unsupervised classification (ISD). According to the test study, the MAL was used to 
create land cover classification maps from TM data in the study area. Change information were derived by comparing the 
classification maps. The results showed that the land cover in the study area has changed very serious in the past decade, especially 
urban and built-up land has been increasing very fast. It is necessary to pay special attention to various environmental issues 
occurred in this area, such as deforestation, industrialization and urbanization, changes of water resources in quantity and quality, 
ecosystem and environment changes, etc. 
1. INTRODUCTION 
Land cover is the combination of vegetation, soil, rock, water 
and human-made structures, which make up the earth's 
landscape, influencing the exchange of energy and matter in 
the climatic system and biogeochemical cycles. Land cover 
change is significant to a range of themes and issues central to 
the study of global environmental change, has an important 
influence on hydrology, climate, global biogeochemical cycles 
and human activities (Flamm et al, 1991; Flamm, 1992; 
Kirchhof, 1995). These diverse roles have been recognized in a 
large number of research publications and international 
conferences, symposia, and workshops devoted to the subject 
over the past few years (e.g., American Institute of Biological 
Sciences, 1994; Thom, 1994; Meyer and Turner 1994), as well 
as the United Nations’ Agenda 21 (UNCED, 1992). 
Scientists, researchers and planners over the world have paid 
much attention to land cover change in the past decades. It has 
been recognized that it is unable to satisfactorily get land cover 
dynamic: information for development planning and 
management decision-making only by traditional approaches. 
Assisted by the analysis for real-time remotely sensed data, one 
could avoid the shortage from a lack of temporally and 
spatially detailed information on land cover change. A number 
of research works have been carried out by using various 
methodologies and algorithms to derive land cover and change 
information from different remotely sensed data (e.g., Bach et 
al, 1994; Berry et al, 1995; Carl and Roland, 1994; 
Lichtenegger, 1992; Stolz and Wolfram, 1995; Wismann, 
1994). But the complex component of geographic terrestrial 
make it difficult to develop a general method for all 
applications in different regions in the world (Chen, 1997). 
The land cover change in the west coast area of Korea has 
occurred seriously in the past decade, effected by both natural 
driving forces and human activities. It is very important for 
governmental decision-making at local, national and regional 
scales to obtain real-time information of land cover and its 
changes temporally and spatially. Without any in situ data 
about the land cover change in this area, this study aims to get 
reasonable change information by using post-classifying for 
multi-temporal Landsat Thematic Mapper (TM) data. 
2.0 STUDY AREA AND AVAILABLE MATERIALS 
2.1 Study Area 
The study area, Ansan City, is located in the west coastal 
region of Korea (see Figure 1). The selected area is 324.0 km? 
(18 km * 18 km), in which the land cover types mainly 
include forest, grassland, urban and built-up land, lakes and 
reservoirs, sea and tidal zone, saltpan, agricultural land, 
wetland, etc. As the industrial development and urbanization, 
the land cover in this region changes very serious in last 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998 399