DocumentCode :
1994259
Title :
Hybrid Classification of Landsat Data for Land Cover Changes Analysis of the Halabja City, Iraq
Author :
Al-Doski, Jwan ; Mansor, S.B. ; Shafri, Helmi Zulhaidi Mohd
Author_Institution :
Fac. of Eng., Univ. Putra Malaysia, Serdang, Malaysia
fYear :
2013
fDate :
9-11 Oct. 2013
Firstpage :
84
Lastpage :
98
Abstract :
We all witness that the earth´s surface has changed attributable to two factors, natural and human activities. Human activities include land usage, deforestation and wars. To study the impacts of wars we have to identify changes to land cover on the earth´s surface. For this purpose, remote sensing technology has played a significant role to monitor land cover changes. This study conducted to examine the impacts of wars on land cover in the Halabja city in the northern part of Iraq (study area) using remote sensing technology. During the last thirty years, Iraq has been immensely influenced by several wars, including the Iran-Iraq War (1980 to 1988), the Gulf War and Early Sanctions (1990 to 1993), and Operation Iraqi freedom OIF (2008 to 2011). These wars have caused huge losses in life and economic and left an aftermath of physical devastation to the land. To study these changes along the study area, the Landsat 5 TM and Landsat 7 ETM+ data acquired in 1990 and 2000 were used. A post-classification technique base on hybrid classification was utilized. After pre-processing an unsupervised K-means classification was carried out first on six reflective bands with the assistance of ancillary data for two images individually followed by maximum likelihood supervised classification to classify all images into five land cover classes, water bodies, vegetation fields, forests, built-up areas and bare lands. The ground truth information collected from Google earth images, visual interpretation and expert knowledge of the area were used to assess the accuracy of the classification images. The overall accuracy was measured to be 91 and 95% with Kappa Coefficient 0.8 and 0.9 for 1990 and 2000 respectively. Post-classification change detection technique was used to produce change image through cross-tabulation to assess changes between different land cover classes. During 1990-2000, cross-tabulation highlight that 54% of water bodies, 25% of forests, 26% of built-up areas a- d 5% of bare lands decreased whilst, the vegetation fields increased about 90%.
Keywords :
geophysical image processing; learning (artificial intelligence); maximum likelihood estimation; pattern classification; remote sensing; Gulf War; Halabja City Iraq; Iran-Iraq War; earth surface; hybrid classification; land cover changes analysis; landsat data analysis; maximum likelihood supervised classification; operation Iraqi freedom OIF; physical devastation; post classification change detection technique; post classification technique; remote sensing technology; unsupervised K-means classification; Accuracy; Cities and towns; Earth; Land surface; Maximum likelihood detection; Remote sensing; Satellites; Hybrid Classification; Land Use/Land Cover; Post-Classification Technique;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Geo-Information Technologies for Natural Disaster Management (GiT4NDM), 2013 Fifth International Conference on
Conference_Location :
Mississauga, ON
Print_ISBN :
978-1-4799-2268-0
Type :
conf
DOI :
10.1109/GIT4NDM.2013.12
Filename :
6937485
Link To Document :
بازگشت