Content uploaded by Richard Gloaguen
Author content
All content in this area was uploaded by Richard Gloaguen
Content may be subject to copyright.
ORIGINAL PAPER
Environmental change detection in the central part of Iraq
using remote sensing data and GIS
Arsalan A. Othman & Younus I. Al-Saady &
Ahmed K. Al-Khafaji & Richard Gloaguen
Received: 13 August 2012 / Accepted: 25 January 2013
#
Saudi Society for Geosciences 2013
Abstract This study aims to assess the potential of sev-
eral ancillary input data for the improvement of unsuper-
vised land cover change detection in arid environments.
The study area is located in Central Iraq where deserti-
fication has been observed. We develop a new scheme
based on known robust indices. We employ Landsat
(multispectral sca nner, thematic mapper, and enhanced
thematic mapper) satellite data acquired in 1976, 1990,
and 2002. We use the Normalized Deferential Vegetation
Index, Normalized Differential Water Index (NDWI),
Salinity Index (SI), and Eolian Mapping Index. Two
new equations wer e applied for the SI and the NDWI
indices. Validation was performed using ground truth
data collected in 16 days. We show that such an ap-
proach allows a robust and low-cost alternative for pre-
liminary and large-scale assessments. This study shows
that desertification has increased in the study area since
1990.
Keywords Iraq
.
Remote sensing
.
Changes detection
.
NDVI
.
NDWI
.
SI
.
EMI
.
MSS
.
TM
.
ETM
.
Land cover
Introduction
Change detection is the process of identifying differ-
ences in t he state of an obje ct or pheno menon by
observing it at diff erent periods; it involves the ability
to quantify temporal effects using multi temporal data-
sets. Desertification means land degradation in arid,
semi-arid, and dry subhumid areas resulting from vari-
ous factors including climatic variations and human
activities. Land degradation can be considered in terms
of the loss of actual or potential productivity or utility
as a result of natural or anthropogenic factors. It is the
decline in land quality or reduction in its productivity
(Khiry 2007).
The process of desertification in Iraq has rapidly in-
creased due to the reduction of surface water by upstream
countries as well as the decreasing in precipitation especial-
ly in central and southern part of Iraq. A lot of efforts have
been devoted to define and study its causes and impacts.
(Al-Jaf and Al-Saady 2009) reported the land use land cover
and hydrochemistry study of Razzaza Lake and Bahr Al-
Najaf area. Shabanas and Zakari (1979) used remote sensing
to create a land use map and studied the de sertification
A. A. Othman
Remote Sensing Group, Institute of Geology, TU Freiberg,
Bernhard-von-Cotta-Strasse 2,
09596 Freiberg, Germany
A. A. Othman (*)
:
Y. I. Al-Saady
Iraq Geological Survey, Al-Andalus Square,
Baghdad, Iraq
e-mail: arsalan.aljaf@gmail.com
A. A. Othman
e-mail: Arsalan-ahmed.othman@student.tu-freiberg.com
Y. I. Al-Saady
e-mail: younusalsaady@yahoo.com
Y. I. Al-Saady
Dept. of Geology, College of Science, University of Baghdad,
Al-Andalus Square,
Baghdad, Iraq
A. K. Al-Khafaji
Water Science Group, Dept. of Geology, College of Science,
University of Baghdad, Baghdad, Iraq
e-mail: arams900@yahoo.com
A. K. Al-Khafaji
e-mail: ahamedobaid@gmail.com
R. Gloaguen
Remote Sensing Group, Institute of Geology, TU Freiberg,
Bernhard-von-Cotta-Strasse 2,
09596 Freiberg, Germany
e-mail: gloaguen@geo.tu-freiberg.de
Arab J Geosci
DOI 10.1007/s12517-013-0870-0
Author's personal copy
problems in Greater Al-Misayab project and followed the
salting present occurrence.
Digital change detection is the process of determining
and/or describing changes in land cover and land use
properties based on coregistered multitemporal remote
sensing data. The basic premise in using remote sensing
data for change detection is that the process can identify
chang e between two or more dates that is unchara cter-
istic of t he normal environmental evolution. Numerous
researcher s have addressed the problem of accurate ly
monitoring land cover and land use change in a wide
variety of environments ( Chan et al. 2001). Accurate
and up-to-date land cover changes information is neces-
sary to understanding and assessing the environmental
consequences of such changes (Giri et al. 2005). While
remote sensing has the capability of capturing such
changes, extracting the change information from satellite
data req uires effective and autom ated change det ection
techniques (Roy et al. 2002). Metternicht and Zinck
(2003)usedaerialphotographs,satelliteandairborne
multispectral sensors, microwave sensors, video imag-
ery, airborne geophysic s, hyperspec tral sensors, and
electromagnetic induction meters for detecting, mapping,
and monitoring salt-affected surface features. Shalaby
Fig. 1 Location and soil maps of the study area (modified after Buringh (1957))
Table 1 Mean annual of four
decades of climate data from Al-
Najaf and Karbala Stations
(1970–2005)
Mean annual 1970–1979 1980–1989 1990–1999 2000–2005
Evaporation (mm/year) 301.03 306.06 315.15 330.8
Rainfall (mm/year) 103.3 113.3 99 61.6
Min. temp. (°C) 16.98 17.12 17.64 18.24
Max. temp. (°C) 30.08 30.72 31.02 32.03
Arab J Geosci
Author's personal copy
and Tatieshi (2007)usedtwosetsofLandsatimages
with different time to detect land use land cover
changes in the northeastern coastal zone of Egypt.
Ayad (2009) applied five vegetative, soil, and water indices
on two Landsat thematic mapper (TM) and enhanced thematic
mapper (ETM) imageries for the period 1990–2000 in order to
asses land degradation in the central and southern part of
upper Mesopotamian Plain in Iraq. This study reveals that
most of counties in the studied area suffered from serious risks
of land degradation and drought water bodies. Hadeel et al.
(2009) used remote sensing and geographic information sys-
tem to: monitor, map, and quantify the environmental changes
from 1990 to 2003. The study demonstrated the effectiveness
of remote sensing and GIS technologies in detecting, assess-
ing, mapping, and monitoring the environmental changes.
As a desertification treatment, the Ministry of Agriculture in
Iraq has begun well and detail work in the study area, for
example Karbala green zone project. This project aims to com-
bat desertification and thus decrease sand storm. In addition,
they did sand dunes stabilization. They also started a project
encompassing the creation of earth mounds in order to grow
thousands of trees, which are resistant to salinity and drought
(Ministry of Agricultural 2012).
The main objective of our study is to apply and examine
new equations of Normalize Differential water index (NDWI)
and salinity index (SI) i n addition to detect changes in
vegetation, sand dunes, soil salinity, and water bodies that have
taken place in the last three decades in the area of interest.
Study area
The study area is located in the central part of Iraq (Fig. 1).
It covers an area of about 16,960 km
2
between longitudes
43°18′–44°33′ E and latitudes 31°39′–33°10′ N.
Fig. 2 Land use and land cover
(Al-Jaf and Al-Saady 2009 )
Table 2 Images characteristics for Landsat scenes
Instrument Acquisition date Path / Row no.
Landsat 2 (MSS) 1976/10/03 181/37
1976/10/03 181/38
1976/07/24 182/38
Landsat 5 (TM) 1990/03/04 169/37
1990/08/28 168/37
1990/03/04 169/38
1990/08/28 168/38
Landsat 7 (ETM) 2002/9/13 169/37
2002/09/06 168/37
2002/08/28 169/38
2002/09/22 168/38
Arab J Geosci
Author's personal copy
Razzaza Lake and Bahr Al-Najaf became drier in the last
decades. Drought is considered a main problem in the area
and it has far-reaching consequences on the agricultural and
pastoral systems. Most of the population depends on agri-
cultural activities in this area.
Climate
The stu dy area has two main water bodies (Razzaza
Lake and Bahr Al-Najaf). Climate data were obtained
from two meteorological stations Karbala and Najaf for
three successive decades (Table 1). It is clear that the
average of annual evaporation has considerably in-
creased, the average of annual rainfall has considerably
decreased, and the average of annual maximum and
minimum temperature has considerably increased since
1970. Using the Köppen-Geiger climate classification
(Markus et a l. 2006)thetypeofclimateisBWh,an
arid climate with prevailing hot summers and limited
seasonal rains. The rainfall season extends from
November to April.
Fig. 3 The NDVI in 1976, 1990, and 2002
Fig. 4 Vector of the NDVI in 1976, 1990, and 2002
Arab J Geosci
Author's personal copy
Soil type and land cover
The study area consists of several types of soils as a
result of a varied geologic environment and fluvial
processes. These types are: basin depression, gypsi fer-
ous gravel, marsh, mixed gypsiferous desert land, poorly
drained phase, river basin, river levee, and sand dunes
land soils (Fig. 1).
The study area has a high intensive urban land. The
agricultural lands represent the main class in the eastern
part of the study area, which is covered by the
Mesopotamian Plain. The water class includes two sub-
classes: shallow and deep water. The barren land, which
is divided to: (1) d ry salt flats, (2) Razzaza beaches, (3)
sand dunes transported by wind, which reach to 6 m in
height. These accumulations have NW–SE tending, (4)
Fig. 5 The NDWI in 1973, 1990, and 2002
Fig. 6 Vector of the NDWI in 1976, 1990, and 2002
Arab J Geosci
Author's personal copy
sand sheet, the sand dunes, and sand sheet consist of
very fine sand and rich carbonate or gypsum grain and
(5) bare exposed rock includes areas of bedrock expo-
sures (Al-Jaf and Al-Saady 200 9;Fig.2).
Topography and geological setting
The study area is generally characterized by horizontal plain
of low relief with an average altitude ranging between 3 and
254 m. The area consists of many formations with approx-
imately thickness of 80 m, covered by Quaternary deposits.
Quaternary deposits are represented by river terraces, gyp-
crete, slope sediments, residual soil, shallow depression
sediments, sand dunes and sand sheets, flood plain sedi-
ments, sabkha sediments, marsh sediments, valley fill sedi-
ments (Jassim and Goff 2006).
Methodology
In this study, three sets of data were used (Table 2).
Unfortunately, there is no recent available coverage TM data
for the whole study area in a same year after 2002. After 31
May 2003, there is a problem with the sensor (the data has
Scan Line Corrector).
Three scenes of multispectral scanner (MSS) images,
four TM images, and four ETM images were mosaiqued
and subsetted using an area of interest file. Nearest neighbor
polynomial correction has b een applied with the aid of
ERDAS 9.1 software. The images were geometrically cor-
rected and projected (WGS84 datum and UTM 38N projec-
tion) using nearest neighbor resampling. The results were
layout with the aid of ArcGIS 9.1 software. Postprocessing
technique results allowed the production of thematic maps
Fig. 7 Sand dune in the study
area
Fig. 8 EMI image of MSS 1976, TM 1990, and ETM 2002 of the study area
Arab J Geosci
Author's personal copy
and thus the quantification of changes for each phenomenon
in the study area.
Normalized Differential Vegetation Index
Lyon et al. (1998) concluded that the Normalized Differential
Vegetation Index (NDVI) is one of the most suitable vegetation
indices for change detection. The NDVI is expressed as the
difference between the near infrared (NIR) and red (R) bands
normalized by the sum of those bands (Eq. 1 in Main (2007)).
NDVI ¼
NIR " R
NIR þ R
ð1Þ
Three periods of Landsat MSS, TM, and ETM data were
used to assess changes in vegetation, using NDVI in 1976,
1990, and 200 2, respectively (Fig. 3). The value range
between −1 to 1 with eight bits and vegetation pixels have
values greater than 0.3. The raster map was divided into two
classes using a threshold of 0.3. All raster data of the NDVI
were then converted to vector data (Fig. 4).
Normalized Differential Water Index
The NDWI was used to investigate the situation of water in
the study area. Equation 2 created by the authors makes use
of the ETM and TM bands 4 and 5. The use of MSS sensor
depends on bands 6 and 7 that represent NIR and SWIR
bands.
NDWI ¼
NIR þ SWIR
2
ð2Þ
The idea of the NDWI is based on the nature of the very
high contrast between water and land. The low reflections of
SWIR and NIR bands of the water allow for their detection.
Fig. 9 Vector of EMI in 1976, 1990, and 2002
Fig. 10 Saline soil in the study
area
Arab J Geosci
Author's personal copy
The interval between 0 and 255 was then thresholded for
values less than 50 to clearly enhance water bodies (Fig. 5).
All raster data of the NDWI were then converted to vector
data (Fig. 6).
Eolian Mapping Index
In order to analyze and evaluate wind erosion in the study
area the Eolian Mapping Index (EMI) was generated. The
EMI is a simple model which has been developed to
generate images that emphasize areas with low vegeta-
tion density and high soil reflectance. MSS, TM, and
ETM data were used to generate this model. The index
uses the R/NIR spectral bands. An RGB color compos-
ite of n ear-infrared and red spe ctral bands, with the ratio
of the red/near-infrared bands (NIR, R and R/NIR) is
generated (Khiry 2007). The produced image shows
various shades of yellow color indicating levels of low
vegetation density and high soils reflectance that serves
as a guide to estimate the relative level of erosion
Fig. 11 The SI image in 1976, 1990, and 2002
Fig. 12 Vector of the SI in 1976, 1990, and 2002
Arab J Geosci
Author's personal copy
potential/vulnerability by wi nd. Sand dunes and sand
sheet areas were observed in the western part of the
study area (Fig. 7).
Pixels that have duller shade of yellow are generally
areas with medium to high vulnerability to wind than
brighter pixels, while the non-yellow areas represent a
little or no wind eros ion potential (Fig. 8). All raster
data of the EMI were then converted to vector data
(Fig. 9).
Salinity Index
Soil salinity can be determined by measuring the TDS of
solution extracted from water-saturated soil paste. The effect
of soil salinity is shown in Fig. 10. The study showed the
possibilities to detect the salinity by using ETM , TM, and
MSS data. The suggested index SI makes use of green and
red bands.
SIðÞ¼
Green þ Red
2
ð3Þ
The idea of the SI is based on the nature of the very high
contrast between salt and their background. The high reflec-
tions of green and red bands of the salts and saline soil allow
for their detection. Saline soil (Eq. 3)wasappliedon
Landsat MSS, TM, and ETM data (Fig. 11). The higher
reflection represents high saline soil.
All raster data of the SI were converted to vector data, the
SI as a vector mode were monitored in Fig. 12 for three
periods (1976, 1990, and 2002).
Results
The vegetation cover area was about 1,587.8 km
2
in 1976; it
has increased to 2,079.8 km
2
in 1990 and then to
Fig. 13 Change in vegetation detection in Shithatha area 1976–1990, 1990–2002, and 1976–2002
Fig. 14 a Left vegetation density, b right positive and negative vegetation in study area for the three periods
Arab J Geosci
Author's personal copy
2,166.9 km
2
in 2002 (Fig. 14a ). The change in vegetation in
the eastern part of the plateau at the three periods is dis-
played in Fig. 13b. These results were obtained after con-
verted all raster data of the NDVI to vector data (Fig. 14).
The maximum density of vegetation is located in the eastern
part of the study area; the typical changes in vegetation of
three periods are shown in Shithatha area (Fig. 13). The
stable vegetation has a dark green color, the vegeta tion loss
has a yellow color, and regrowth vegetation has a light green
color.
Figure 14b show s the h ighest di fference b etween re-
growth and degradation of vegetation was in the period from
1976 to 1990, the regrowth was more important than deg-
radation; in other words, the high vegetation incre ase hap-
pened during the period 1976–1990. The period for highest
vegetation regrowth was in 1990. Results showed a general
increase in water bodies in 1990 and a general decrease in
2002.
Water bodies covered 1,544 km
2
in 1976 and 1,677 km
2
in 1990 and 1,079 km
2
in 2002 (Fig. 15). The decreases in
the surface water bodies in the study area can be attributed
to many reasons, for example, decrease in the flow of the
Euphrates River from the upstream countries as well as the
use of river water for irrigation in the study area. Figure 15b
shows the high difference between positive and negative
changes of water bodies during 1990–2002. The highest
rate of increase in water bodies happened during 1976–
1990; the highest loss was 598 km
2
when the water bodies
covered an area about 1,079 km
2
during 1990–2002. The
period for the highest water surface increase was in 1990.
From visual interpretation of the EMI imagery for the
years 1976, 1990, and 2002, it is clear that in 1976, the
potential of wind erosion is medium while the highest rate of
erosion were observed in 2002 in the western part of the
study area. The dull tone of yellow color indicates high
density of vegetation cover and low reflectance of soil.
The brighter tone of yellow color indicates low density of
vegetation cover and high reflectance of soil. The EMI of
ETM 2002 explained that very brighter tone of yellow
observed in western part especially in sand dunes and sand
Fig. 15 a Water body in the study area for the three periods, b variation in water body in the study area for the three periods
Fig. 16 a Eolian density in study area for the three periods, b variation in eolian density in the study area for the three periods
Arab J Geosci
Author's personal copy
sheet areas. The most increase of the EMI was in the year
2002. We were calculated the areas of the EMI (Fig. 16a). It
was 1,267 km
2
in 1976, 576 km
2
in 1990, and 3,102 km
2
in
2002. Figure 16b shows the high difference between eolian
and non-eolian, it was 2,165 km
2
in 1990–2002. The higher
increase in eolian erosion happened in 1990–2002. Generally,
the best period for non-eolian erosion was in 1990. The
visualization and interpretation of the EMI imagery of MSS
1976, TM 1990, and ETM 2002 gives a guide to the relative
level of erosion potential and vulnerability to wind
Figure 17a shows a general increase in soil salinity in the
study area. The saline area covered 51 km
2
in 1976 and
83 km
2
in 1990. The highest increase of saline soil
(124 km
2
) was in 2002. Figure 17b shows the high differ-
ence between saline and nonsaline soil in 1990–2002. The
saline soil is more than nonsaline; the highest saline increase
happened in 1990–2002. The period for highest non saline
soil cover was 1990.
Discussion of the desertification
In order to quantify desertification changes, we combined the
previously mentioned indices. We produce a ratio between the
areas of threshold indices at two dates. When the NDVI and the
NDWI ratios are less than (1) then the area is considered
desertified. The same when the EMI, and the SI ratios have a
value more than (1) (3, 4, and 5). We then consider negative
changes (desertification) and positive changes (regrowth). In
1990 when the NDVI, NDWI, and EMI ratios indicate positive
changes and the SI negative change, the area considered
repaired and reformed (Table 3). The NDWI, EMI, and SI
changes are negative in 2002 while the NDVI change is posi-
tive. Therefore, the area is considered desertified (T able 4). We
attribute this decline to the human activities when the area that
appeared from the regression of Razzaza Lake water was used
for agricultural activities (Table 5). Generally, the process of
desertification in the area has increased.
Conclusions and recommendations
We used new equations of the NDWI and SI that give good
results to detect the water and salinity from multispectral satellite
data. The application of multitemporal (MSS, TM, and ETM)
remote sensing data offers an effective opportunity for mapping
desertification processes in the study area as well as in arid lands
at relatively low cost. The NDVI results show an increase in
vegetation in 1976, 1990, and 2002 for the study area. The
NDWI results show an increase in water bodies in 1990 and a
Fig. 17 a Saline density of soil in the study area for the three periods, b variation in saline soil density in the study area for the three periods
Table 3 The relationship between the NDVI, NDWI, EMI, and SI
during the periods 1990 and 1976
Area in km
2
Rate of change
Year 1976 Year 1990
NDVI 1,587 2,079 0.763
NDWI 1,544 1,677 0.92
EMI 1,267 576 2.2
SI 51 83 0.61
Table 4 The relationship between the NDVI, NDWI, EMI, and SI
during the periods 2002 and 1990
Area in km
2
Rate of change
Year 1990 Year 2002
NDVI 2,166 2,079 0.96
NDWI 1,079 1,677 1.55
EMI 3,102 576 0.186
SI 124 83 0.669
Arab J Geosci
Author's personal copy
decrease in 2002. The EMI hybrid color composite results show
adecreaseineoliandepositsin1990andanincreasein2002.SI
results show an increase in saline area in 1990 and 2002. In
general, the desertification processes in the study area have
increased. The combination of NDVI, NDWI, EMI, and SI is
apowerfultechniqueincharacterizationandmappingofdesert-
ification process in the study area by providing direct measure-
ments. Salt layer presence on Razzaza Lake bottom and islands
attributed to high evaporation. The authors recommend a de-
creasing of intensive culture, areductionofovergrazingade-
crease of tree falling for lumber or firewood for heating and
cooking. In addition, using newly irrigated methods allowed
decreasing soil salinity.
References
Al-Jaf AA, Al-Saady I, (2009) Integration of remote sensing data and
GIS application for land cover land use and environmental
changes detection in Razaza Lake Bahr Alnajaf area GEOSURV
int.rep.no. 3150
Ayad FM (2009) Land degradation detection using geo-information
technology of some sites in Iraq. J Al-Nahrain Univ 12(3):94–108
Buringh P (1957) Directorate General of Agricultural Research and
Projects–Ministry of Agricultural, Baghdad, Iraq
Chan JC, Chan KP, Yeh AGO (2001) Detecting the nature of change in
an urban environment: a comparison of machine learning algo-
rithms. Photogramm Eng Remote Sens 67:213–225
Giri C, Zhu Z, Reed B (2005) A comparative analysis of the Global
Land Cover 2000 and MODIS land cover data sets. Remote Sens
Environ 94:123–132
Hadeel AS, Jabbar MT, Chin X (2009) Environmental changes mon-
itoring in the arid and semi arid regions: a case study Al-Basra
Province, Iraq
Jassim SZ, Goff JC (2006) Geology of Iraq. Dolin, Czech Republic
Khiry MA (2007) Spectral mixture analysis for monitoring and map-
ping desertification processes in semi-arid area in north Kordofan
state, Sudan, PhD thesis, Univ. of Dresden, p 126
Kottek M, Grieser J, Beck C, Rudolf B, Rubel F (2006) World Map of
the Köppen–Geiger climate classification updated. Meteorol Z 15
(3):259–263
Lyon JG, Yuan D, Lunetta RS, Elvidge CD (1998) A change detection
experiment using vegetation indices. IEEE Photogramm Eng Re-
mote Sens 64:143–150
Main RS (2007) A remote sensing change detection study in the arid
Richtersveld Region of South Africa, M.Sc. thesis, Univ. of the
Western Cape
Metternicht GI, Zinck JA (2003) Remote sensing of soi l salinity:
potentials and constraints. Elsevier Remote Sens Environ 85:1–20
Ministry of Agricultural (2012) http://www.zeraa.gov.iq/ar/index.php?
name=alscosearch
Roy DP, Lewis PE, Justice CO (2002) Burned area mapping using
multi-temporal moderate spatial resolution data—a bi-directional
reflectance model-based expectation approach. Remote Sens En-
viron 83:263–286
Shabanas and Zakari (1979) Remote sensing - land use mapping study
of desertion problems in Greater Al-Misayab project and follows
the salting, present occurrence, GEOSURV int.rep.no. 984
Shalaby A, Tatieshi R (2007) Remote sensing and GIS for mapping
and monitoring land cover and land-use changes in the North-
western coastal zone of Egypt. Appl Geogr 27:28–41
Table 5 The relationship between the NDVI, NDWI, EMI and SI
during the periods 2002 and 1976
Area in km
2
Rate of change
Year 1976 Year 2002
NDVI 1,587 2,166 0.73
NDWI 1,544 1,079 1.43
EMI 1,267 3,102 0.408
SI 51 124 0.41
Arab J Geosci
Author's personal copy
