stanbul 2004
proaches to
n, European
>
s.
vironnement
iltiples, Télé
ntribution de
nples choisis
scientifiques
an.
's during the
iguished by
geology. 170
Application
Nouakchott,
)0. Coastal
Ote sensing
wironmental
fred ehlers ,
ural remnant
nts from the
d Planetary
GIS BASED ANALYSIS OF LANDCOVER CHANGES ARISING FROM COAL
PRODUCTION WASTES IN ZONGULDAK METROPOLITAN AREA-TURKEY
H. Akein* *, S. Karakis*, G. Büyüksalih*, M. Oruc*
? ZKÜ Engineering Faculty, Geodesy & Photogrammetry Eng. Dept. , 67100 Zonguldak Turkey, (akcin,karakis,
buyuksal, oruc) @jeodezi.karaelmas.edu.tr
Commission PS ICWG I/IV
KEY WORDS: GIS, change detection, object oriented image analysis, Landsat-5 TM, Ikonos, Coal production waste, Zonguldak
ABSTRACT:
Zonguldak Metropolitan Area (ZMA) is located in North-western part of Turkey and a mining and industry region which occuppies
the 80% of the Zonguldak Hardcoal Basin (ZHB). In this region, hardcoal mining has been made since 1848 and, based on the
records kept since 1865, 328 million tons pit run coal have been produced. Resulted amount of waste from the processing of coal in
the coal washing factories is about 114 million tons. Some parts of these coal wastes have been stocked along the coastal lines in
ZMA by governmental and private companies while the others have been stored in the valleys and hilly side of inner regions.
Additionally, waste materials have been used as filling materials in road and plant constructions.
In this study, coal waste areas and use of wastes in the settlement areas have been determined by means of the analysis carried out
on the Landsat TM images of the region using object-oriented image analysis program eCognition 3.0. After segmentation and
required classification procedures, results are integrated to GIS environment. At this phase, classified images are superimposed with
the high-resolution Ikonos images and then necessary analyses have been realized. Based on the findings, in ZMA, in1992, there
was about 2.1 sq.km coal and coal waste areas, their dimensions increased to 4.2sq.km in 2000. In the urban areas, while the size of
coal waste cover was about 9.7sq.km in 1992, it was measured as 16.5sq.km in 2000.
1. INTRODUCTION
Based on the statistics from 2000, ZMA is a metropolitan area
with 70 sq.km size and consist of 5 sub-settlement units in
which 176511 population lives (Erkin 1999). 20%, 20%, 40%,
7% and 15% of ZMA are covered by forested areas, settlements
neighboured to forested areas, industrial site, agriculture area
and coal waste and idle part respectively.
ZTH contains ZMA is the only hardcoal production basin in
Turkey. This basin covers an area of 10000sq.km which extends
in 20 km width from east to west. However, 8096 of the
production takes place in the ZMA. While the hardcoal reserve
is about 0.25% of world reserve, amount of total production has
been only 10% of the world production (DPT report 1996 and
www. taskomuru.gov.tr 2004). In the basin, until the 2004, total
amount of production is 412 million tons and decomposed
waste is around 143 million tons. Corresponding amounts for
ZMA is 328 million tons production and 114 million tons waste
respectively. Amount of annual production and waste are given
in Table 1.
In this study, first, Landsat- TM images of the interest area was
segmented, and then classified. Afterwards, they superimposed
to high-resolution Ikonos image with the required queries which
results necessary statistics for temporal analysis.
Corresponding author.
Year Annual Total Annual Total
Pit Run Pit Run Waste Waste
Production Production
1865-92 365271539 129152730
1992 4790719 370062258 1961553 131114283
1995 4608668 374670926 1824880 132939163
1994 4210622 378881548 1384559 134320722
1995 3251570 382133118 1003400 135324122
1996 3320074 385453192 897090 136221212
1997 3118688 388571880 813205 137034417
1998 2865824 391437704 736672 137771089
1999 2601175 394038879 654502 138425591
2000 3196564 397235443 940337 139365928
509
Table 1. Amount of total and annual production and waste
from tons unit in ZTH.
2. METHODOLOGY
In this study, two types of images have been used. First one
includes the Landsat images (only bands 1,4 and 7) dated as
1992 and 2000 with the ground resolution of 30m. These
spectral channels were analysed by PCI Geomatica V9.1.1
software package. In these analyses maximum likelihood
classification technique was used (Lillesand and Kiefer, 2000).
Second one is the pan-sharpened lkonos images with 1m pixel
size.