© by PSP Volume 20 – No 5. 2011 Fresenius Environmental Bulletin
1190
A CHANGE VECTOR ANALYSIS TECHNIQUE TO MONITOR
LAND-USE/LAND-COVER IN THE YILDIZ MOUNTAINS, TURKEY
Murat Özyavuz
1,
*, Onur Şatır
2
and Bayram Cemil Bilgili
3
1
Namık Kemal University, Faculty of Agriculture, Department of Landscape Architecture, 59030,Tekirdağ, Turkey
2
Çukurova University, Faculty of Agriculture, Department of Landscape Architecture, Adana, Turkey
3
Karatekin University, Faculty of Forestry, Department of Landscape Architecture, Çankırı, Turkey
ABSTRACT
This paper presents the method to detect the intensity
of change and the dimension of change using Change Vector
Analysis (CVA) method. The aim of this study is to con-
firm the changes in the Yıldız Mountains using proof from
natural ecosystems (natural deciduous forest, wetland and
sand dune ecosystems) and other various ecosystems
ecologically significant all over Europe and Turkey by using
the CVA technique. Two Landsat TM scenes recorded in
1990 and 2009 were used to minimize change detection
error introduced by seasonal differences. Images were geo-
metrically and atmospherically corrected. As a result of the
ratings, it is observed that agricultural areas decreased by
17.5 and Crimean pine forests by 16.4% whereas oak for-
ests increased by 4.1%. Ash trees, sand dunes and bare
ground reflected insignificant changes, too.
KEYWORDS: CVA analysis, change detection, remote sensing,
habitat monitoring.
1. INTRODUCTION
Land-use/land-cover change is an important field in
the global environmental change research. Inventory and
monitoring of land-use/land-cover changes are indispensa-
ble aspects for further understanding of change mechanism,
and modeling the impact of change on the environment and
associated ecosystems at different scales [1, 2].
Remote sensing is a valuable data source from which
land-use/land-cover change information can be extracted
efficiently [3]. Information derived from remote sensing
has played an important role in natural resources manage-
ment and planning by providing insight into land-cover and
land use patterns and multi-temporal trends [4, 5]. The
measurements of remote sensing in visible bands and ther-
mal infrared can be used to extract characteristics of land-
cover, derive biophysical parameters of vegetation, and
support many study fields including global change [6].
* Corresponding author
Landsat satellite imageries are widely used in remote
sensing. The launch of Landsat-1 in 1972 initiated a new
era of providing satellite data in digital format to users.
Efforts to develop algorithms for image classification have
continued since the 1970s. Combined with the rapid evolu-
tion of Geographic Information Systems, researchers have
quickly created computer-aided analysis tools to produce
land-use and land-cover (LULC) maps [7]. These maps have
been used to analyze the impacts of land-use change on the
environment [8], improve land-use planning and natural
resource management [5], and better understand ecological
processes on earth [9]. Change detection is the process of
identifying differences in the state of an object or phenome-
non by observing it at different times [10]. Timely and accu-
rate change detection of Earth surface features provides the
foundation for greater understanding of the relationships
and interactions between human and natural phenomena.
Different change detection algorithms have their own mer-
its and no single approach is optimal and applicable to all
cases. In practice, different techniques are often compared
to find the most useful change detection results for a spe-
cific application [11]. These algorithms have a common
characteristic in that they all involve selecting a threshold
to determine the changed areas.
Four change detection techniques, image differencing,
image rationing, image regression and change vector analy-
sis (CVA), are used widely in the remote sensing context.
Among them, CVA is a valuable technique for land-use/
land-cover change detection [3]. Image differencing is a
common change detection approach for forested and agri-
cultural areas [10, 13-15]. CVA is typically applied to multi-
spectral images acquired by passive sensors, by using all
the spectral channels that contain useful information with
respect to the considered kind of change. The CVA tech-
nique is based on three steps: (1) image comparison by
vector subtraction, (2) magnitude of the spectral change
vectors computation (sometimes also the direction of SCVs
is computed), and (3) thresholding. The first step computes
the vector difference of spectral feature vectors associated
with pairs of corresponding pixels in two images acquired
on the same geographical area at two different times, and
results in a multispectral difference image. Each pixel in
this image is associated with a multidimensional vector