MONITORING THE EFFECT OF CHEMICAL WEAPONS ON LAND COVER OVER HALABJA

1
MONITORING THE EFFECT OF CHEMICAL WEAPONS ON
LAND COVER OVER HALABJA CITY, IRAQbyJWAN
.M.AL-DOSKI

• Shattri B.Mansor and Helmi Zulhaidi Mohd Shafri
• Geospatial Information Science Research Centre
  (GIS RC),
• Faculty of Engineering
• Universiti Putra Malaysia 

2
Introduction

• The development and use of
  satellite imagery and aerial photographs have
  long been used for improving the effectiveness of
  military operations from aerial photographs using
  balloons to aircraft platforms and recently
  satellite remote sensing( Corson, M.W. 2004).
  Military planners have used remote sensing
  satellite imageries as a tool of warfare, while
  during the past decades, the majority of academic
  researchers have made their efforts in using this
  technology in non-military use such as Forestry
  Hydrology , Geology , Agriculture, Environment,,
  Landscape changes and finally war impacts (De
  Sherbinin, A. 2002).
• As had been shown in mankind's history,
  there are many countries in which during their
  advancement and evolution entered into wars which
  left them in devastation with huge losses in life
  and economic resources. For example, since the
  past three decades, Iraq had engaged in series of
  wars began with the bloody Iraq-Iran war which
  started from September 1980 to August 1988 after
  that Iraq- Kuwait war in 1990 which known as Gulf
  War and finally the occupation by United State in
  2003, made them the longest conventional wars in
  the 20th century (Abrahamian, E. 2008). These
  wars devastated Iraqi economy and left it saddled
  with huge debts as well as a great loss in lives
  (Dugdale-Pointon 2002) Up till this moment there
  is no conclusive figures for the number
  sacrifices and losses.

3
Problem statement

• During the course of Iraq-Iran war, many parts of
  Iraq were bombarded with hazardous chemical
  weapons such as mustard gas, nerve agents, sarin,
  Tabun, VX and cyanide. Halabja city is one of
  those places that suffered heavy bombardment with
  attendant destructive impact on human lives,
  socio-economic and the environment. Since 1994,
  many studies have been conducted on the impact of
  war on the social, cultural and economy to better
  understanding the interaction and relationship of
  the war on natural phenomena as well as for
  better managing and using available resources. No
  study has been carried out on the impact of the
  war on environment/land cover changes because of
  lack of field survey data , apart from this most
  of the areas were closed after bombing coupled
  with political undertone and the harsh conditions
  of war. In recent decades, remote sensing
  satellite imageries are extensively used for
  detecting war impacts however the ability of
  these imageries are limited and difficult to
  detect some war impacts for example,
  bullet-pocked walls whereas the mass displacement
  of local residents, re-vegetation in agricultural
  areas occurs or new service roads are constructed
  can lead to changes in land cover it would be
  detectible easily by satellite imageries.

4
Research Objectives 

• The main goal of this research is to examine
  change detection techniques to investigate
  short-term changes before and after shelling with
  chemical weapons as well as 10 years long-term
  changes in land cover that had occurred in
  Halabja. To obtain this, the following objectives
  will be achieved
• Applying three change detection methods (Image
  differencing(NDVI) , hybrid classification and
  post-classification techniques) to identify land
  cover changes.
• Create Land use/ land cover classification maps
  as accurately as possible on a regional scale of
  Halabja city.
• Create the LULC changes maps for the study area
  within the period of 1986 to 1990 as well as 1990
  to 2000 using post-classification method.
• Examine both qualitative and quantitative changes
  to show the impact of the war on Halabja city.

5
Research Questions

• In order to fulfill the above mentioned
  objectives, the following research questions can
  be made
• Can the land use/land cover Changes in study area
  be assessed using the applied satellite images?
• How effective is the hybrid change detection
  technique for image classification in the context
  of the study area?
• What are the land use/land cover changes in the
  study area in two time periods?
• How can post classification be used to strengthen
  the justification of accuracy for the Hybrid
  classification?

6
Study area

• Regionally, Halabja city with a population of
  about 70,000 is situated 80 kilometers from
  southeastern city of Sulaimanya, about 150 miles
  (241km) north-east of the Iraqi capital city
  Baghdad and 16 Km away from Iraq-Iran boarder. It
  lies in southeastern Sharazur plain. And
  surrounding by ( Hawraman ) mountain and (
  Balambo) mountain to the north and south borders
  of Halabja. Geographically, Halabja is located
  between 3510'59.22"N latitudes and 4558'59.05"E
  longitudes with land area of about 592.7 square
  miles.

7
Study area
8
Data Landsat scenes
9
(No Transcript)
10
(No Transcript)
11
(No Transcript)
12
Climate data
Iraq crop calendar
13
Methodology

•  

14
Pre- processing1. Radiometric correction

• Absolute radiometric correction method was used
  which is generally a two step process. The
  first step is to convert the digital number (DN)
  o f the sensor measurements to spectral radiance
  (L). Secondly, conversion of measured DN to top
  of atmosphere (P) reflectance units by using
  following equations

15
SATELLITE SUNSOLAR ELEVATION() SUNSOLAR ZENITH() SCENE_CENTER_SCAN_TIME
1986 imagery TM 61.1776237 105.0757542 065700379
1990 imagery TM 60.5546184 105.1066741 06532424
2000 imagery ETM 66.0654401 111.8446727 07245393
16
2. Geometric correction

• Geometric correction procedures address
  errors in the relative position of pixels . The
  images in this study were level-IT products which
  mean geometrically corrected for making sure
  further rectify and standardize were done on all
  images by using automatic Image-to-image
  registration to 2000 image (base image) with root
  mean square error 0.5 consider acceptable
  followed by projecting all images to a common
  coordinate system (UTM) with a World Geodetic
  System (WGS) 84 datum Zone 38 North. In order to
  easy to compute, speed up and reduce time of
  processing all images subsetted to (1477 samples,
  929 lines) including only the area of interest
  followed by resampling to a 30 m pixel size using
  the polynominal warping method with nearest
  neighbor algorithm to keep the original
  brightness values of the pixels

Base Image Warp Images GCPs RMS
2000 Image 1986 Image 131 0.47
2000 Image 1990 Image 116 0.43
17
Atmospheric correction

• Dark Object Subtraction (DOS) is a simple
  image-based atmospheric correction . It was
  performed on each image to remove absorption and
  scattering of electromagnetic radiation or reduce
  the influence of atmospheric scattering within
  each scenes

18
NDVI 1986

• NDVI is related to photosynthetic
  activity of vegetation and it can calculate using
  the following equation
• NDVI NIR- R / NIR R

NDVI values range from -1.0 to 1.0. Typical
values are from 0.2 to 1 or above 0,6
represents a large amount of high
photosynthesizing and healthy vegetation.
According to (Wang, J. 2004) NDVI values between
-1.0 to 0.1 represents non- vegetative and
correspond to barren areas of rock, sand ,snow
,built- up area and water body. Conversely
sparse values 0.2 indicate shrub and grassland,
Moderate values(0.5) indicate vegetation field
while forests are represented by high NDVI values
(0.6)
19
(No Transcript)
20
NDVI 1990
21
NDVI 1990 Classification
22
NDVI 2000
23
(No Transcript)
24
NDVI statistic analysis
NDVI VALUE NDVI VALUE NDVI VALUE NDVI VALUE
IMAGES Min Max Mean Stdev
1986 -0.521281 0.770415 0.178504 0.137779
1990 -0.485578 0.771176 0.158516 0.13129
2000 -0.173588 0.562458 0.136942 0.059818
25
Band differencing
Red band differencing
Green band differencing
Color composite image
Blue band differencing
26
NDVI differencing methodIt can be calculate by
using NDVID NDVI time 1-NDVI t2
27
(No Transcript)
28
NDVI 2000-1990
29
NDVI 1990-1986
30
Prepare General Base map
31
Spectral Angle Mapper 1986
32
Spectral Angle Mapper 1990
33

• The three satellite images were classified into
  land use/ land cover classes base on( Anderson
  et al (1976) land use/ land cover
  Classification Systems) were modified to
  classify the images into five classes as shown
  in this table

Classes Definition
Water bodies Rivers, lakes, ponds, lagoons, dams, marsh wetlands
built-up area Residential and commercial services, office blocks, roads, rails
Forest land protective forests, timber forest , economic forest, firewood forest and forests of special use
Agriculture land All cultivated areas such as farmlands, crop fields including vegetable gardens, plantations, fallow plots
Rangeland Herbaceous Rangeland, Shrub and Brush Rangeland ,Mixed Rangeland
34

• Thank you