International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
(Evrendilek and Doygun, 2000; Wali et al., 2002). For example,
increasing the amount of land, water and nitrogen fertilizers
used to meet the ever-increasing food consumption which in
turn increases environmental degradation leads to an
unsustainable cycle, for environmental degradation makes it
more difficult to meet our basic human needs of food and clean
water in the long-term (Kilic et al., 2003; Kilic et al., 2004). On
the other hand, multi-temporal high-resolution, remotely sensed
data and geographic information systems (GIS) have facilitated
the derivation of ecological inventories and the monitoring of
LULC changes on the local, regional and global scales
(Berberoglu, 2003; Berberoglu et al., 2004). In particular, the
visible to shortwave infrared (VSWIR) bands of Landsat
Multispectral Scanner (MSS) and Thematic Mapper (TM) data
have been extensively used for surveys and analyses of forest
and agricultural resources since the initiation of Landsat
program in 1972 (Berberoglu et al., 2000; Townshend and
Justice, 2002).
This study region was selected because it is located in one of
the most biologically diverse regions as well as the most water-
short regions of the Mediterranean Basin. There is a lack of
quantification and identification of LULC changes in
ecologically productive and hotspots of the study region.
Sustainability of vital ecosystems in the region are threatened
by increased rates of population growth, consumption and waste
disposals, and the keen competition among LULCs for both
rate-limited ecological services and stock-limited natural
resources. For example, draining of the Amik lake and its
surrounding wetlands has led to continued problems with water
quality and quantity and escalated flood damages. Preventive
and mitigative measures are urgently needed in the supply- and
demand-side managements of natural resources in order to
assure long-term productivity and health of ecosystems before
the point of irreversible losses is reached. The main objective of
this study was to reconstruct past LULC changes over the last
28 years at the scale of the province of Hatay, Turkey, based on
a time series of Landsat imagery acquired from 1972, 1987, and
2000.
2. MATERIALS AND METHODS
2.1 Study Region
The study region is located in the province of Hatay (Turkey)
(35? 47'-36? 24' E; 35? 48'-36? 37' N) and has a total area of ca.
3930 knY? with an elevation range from sea level up to 2100 m
(Figure 1). The study region is delimited by the Mediterranean
Eel TUE El
Figure 1. Location of the study region in Turkey.
Sea in the west, by Syria in the south and east, and by the
province of Gaziantep in the east. The prevalent climate regime
is Mediterranean climate characterized by a mild winter during
which about 67% of the annual precipitation of 1124 mm falls,
and a hot dry summer. Average annual temperature reaches a
maximum of 44°C in the summer and a minimum of —15°C in
the winter, with an average annual temperature of 18°C. Parent
materials in the study region consist mostly of sedimentary
rocks of highly calcareous clays, limestone, dolomites, and
sandstones. The major soil orders include entisols, inceptisols,
vertisols, mollisol, and alfisol (Kilic, 1999). The alluvial soils
formed in the Amik plain by the Orontes river are the most
productive soils covering the northeast of the study area. The
dominant Mediterranean vegetation consists of evergreen
forests of Pinus brutia, and shrublands composed of maquis,
and garrigues. Regressive succession stages of evergreen forests
have taken place under drastic human-induced disturbances
such as overgrazing, overcutting, and clearance for cultivation.
2.2 Data Processing
The following remotely sensed cloud-free data were used in the
classification of LULCs in the province of Hatay through
ERDAS Imagine and TNTmips software: Landsat MSS of
December 15, 1972, Landsat-5 TM of October 1, 1987, and
Landsat-7 ETM+ of June 22, 2000. The images were
geometrically corrected and geo-coded to the Universal
Transverse Mercator (UTM) co-ordinate system, using 1:25 000
scale topographic maps and aerial photographs taken in 1992.
Approximately 45 evenly distributed ground control points
were selected from each image. These were used to spatially
resample the images using a nearest neighbor algorithm, which
takes the value of the pixel in the input image that is closest to
the computed co-ordinate. This method is fast and does not alter
the original pixel values. The transformation had a root mean
square (RMS) error of between 0.4 and 0.7 pixel, indicating that
the image rectification was accurate to within one pixel. The
acquired Landsat images were classified using a maximum
likelihood classifier method of ERDAS Imagine and TNTmips
software. A supervised maximum likelihood classification was
performed with 2682, 11689 and 8107 pixel training data sets,
and the images were classified into 6 (1987 and 2000) and 7
(1972) LULC classes.
3. RESULTS AND DISCUSSION
On the basis of a time series of the Landsat images, LULCS of
the study region was classified into seven categories: evergreen
forest, shrublands-orchards, the Amik lake and its related
wetlands, croplands, water bodies, settlements, and bareground
(Table 1). The study area occupied 392,970 ha. The annual
rates and cumulative amounts of LULC changes were estimated
over the two periods: 1972 to 1987 (period 1), and 1987 to 2000
(period 2). Land use/cover changes were of highest amount in
croplands, evergreen forests, shrublands-orchards, and the
Amik lake-related wetlands in the first period and in settlements
and bareground in the second period. In both periods,
bareground decreased, while croplands and settlements
increased. In the first period, the rate of change of the Amik
lake-its related wetlands, and shrublands-orchards was negative,
whereas the rate of change of evergreen forests was positive. In
the second period, a decrease in evergreen forest took place,
whereas an increase in shrublands-orchards occurred.
Comparisons of LULCs in 1972, 1987 and 2000 derived from
the Landsat images indicated that the Amik lake and its
314
Internat
Table 1.
surroundi
annihilate
of a larg
used for
Amik lak
croplands
the lake
lake and
ecosyster
regulation
region. V
its relate
ecologica
altered tl
some spe
2 oc = DR = = =
Changes in LULC (x10' ha)
Figure 2.
In both p:
to the de
Were neg;
settlemen
There w
evergreen
Between
