il 2004
)ported
ints.
| status
etry &
nmetry
etry &
ki, M.,
ing and
itoring.
1996.
phasis
APRS,
Digital
is and
93.
irborne
vation
imetrie
anning
iology.
irborne
edings,
dels in
lournal
ter, H.
. 2001.
ciers —
nt and
Spatial
nt, pp.
g - à
99, PP-
ctueller
eines
de und
MONITORING SPATIAL AND STRUCTURAL CHANGES OF FOREST COVER IN
YENICIFTLIK WATERSHED WITH MULTITEMPORAL SATELLITE DATA
M.INAN', K. ERDIN?
Istanbul University, Forestry Faculty, 34473 Bahçekôy, Istanbul, Turkey — (inan, erdink)@istanbul.edu.tr
KEY WORDS: Remote Sensing, Monitoring Forest Resources, Natural Resources Planning, Change Detection, Landsat
ABSTRACT:
Recent surveys indicate that changes in forest cover and land use have a direct and enormous effect on wildlife, water quality,
climate and carbon cycling. The forest ecosystem is threatened constantly by both human impacts like forest fires, air pollution,
clearing for agricultural uses, illegal cutting and also natural phenomena like storms and droughts. The monitoring and control of this
dynamically structured forest ecosystem give the best usage possibilities for the sustainable operation and protection of forest
resources. In this study, the land use of Yeniciftlik watershed located in Beykoz-Istanbul- Turkey, and spatial and structural changes
of forest cover are investigated by using the monitoring system with Lands
at MSS, TM and ETM" images evaluated with the data of
ten year time periods belonging to the years of 1974, 1984, 1994 and 2001. In addition, the combined uses of GIS and remote
sensing tools have been highlighted with respect to their advantages in forestry applications.
1. INTRODUCTION
At present, change detection and land use mapping studies are
important because false land use decisions and deforestation
have caused negative impacts on soil erosion, run-off and
flooding, CO2 concentration and climate changes. Change
detection is the method used to highlight or extract differences
through imagery acquired from different epochs time (ERDAS,
1997). Satellite data has become the major data source in the
change detection of diverse applications including forest cover
changes, because of the its advantages in fast, cost effective,
synoptic, accurate, flexible and up to date properties and digital
data acquisition characteristics.
There are several methods for the determination of change
detection, but basically, there are two main approaches:
Comparative analysis of independently-produced classifications
and simultaneous analysis of multi-temporal data (Sing 1989).
The method of post-classification comparison provides
complete change information reduces the impact of atmospheric
and environmental differences to a minimum and make it
possible to the classify images recorded at different time
periods.
This study covers the analysis of the changes in land-use
preferences over a 30 year time period in Yeniciftlik Watershed
and determination of structural and area changes in the forest
covers by using the Landsat satellite data belong to the years
1974, 1984, 1994 and 2001.
2. STUDY AREA AND DATA SET
The study area is located in Turkey, roughly between latitudes
41° 09’ - 41° 03’ N and longitudes 29° 10° — 29° 09’ E. The
total area is 22.86 km” and average height of the area is 152.04
759
m. The main data used in this study is Landsat MSS/TM/ETM*
images recorded on 31* June 1975 (MSS), 24" June 1984
(TM), 02" July 1984 (TM) and 16" July 2001 (ETM”).
Additionally, tabular data derived from Forest Management
Plans made in the relevant years and the map of these plans
were used to compile the terrestrial data to be used as
supporting data for the study.
3 METHODS
Figure 1 illustrates the flow chart of land use change detection
associated with forest cover change detection. The multi-
temporal Landsat images were geometrically rectified and
registrated into UTM projections. All Landsat images were
atmospherically and topographically corrected with ERDAS
Imagine ATCOR 2.1.Modules.
| Multi Temporal Landsat Images (1975, 1984, 1994, 2001)
[ Geometric Rectification and registrations { Atmospheric & Topographic correction |
| j | (ATCOR 2.1)
M
M t
Ancillary data | | Linear regression between 19754:1994, 198441994 and 1994 2001
(Tabular data & forest maps) |
1. Image Enhancement (Veg. Indices, Tasseled Cap, adco)
2. Determining Training Sample Area
3. Supervised Classifications (1975. 1984, 1994, 2001) |
) Y ! f E
1974Landuse | | 1984 Land Use i | 1994LandUse | | 2001 Land Use |
image (raster) | | Image (raster) | | Image (raster) Image (raster) |
À MÀ tt pee. Ne nr te ee ee i & — —
rte I bbe M T = 3 P
*
( GIS Software f )
l. Data Conversion (raster to vector) | Change detection maps |
2. Overlay Process ” and statistics |
3. Query Process C
J
Figure 1. Flow chart for classification and change detection
The atmospheric and topographic correction models eliminate
or reduce the effect caused by sun zenith angle, solar radiance,
atmospheric scattering and absorption, but they cannot
eliminate the reflectance difference among the multi-temporal
images caused by different environmental conditions such as
soil moisture (Lu D. at all. 2002).
