assessment of land-use/land-cover change in muharraq island using multi-temporal and multi-source...

18
This article was downloaded by: [Karolinska Institutet, University Library] On: 04 June 2014, At: 10:13 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK Click for updates International Journal of Image and Data Fusion Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/tidf20 Assessment of land-use/land-cover change in Muharraq Island using multi- temporal and multi-source geospatial data Marjan Modara a , Mohamed Ait Belaid a & Sabah AlJenaid a a Desert and Arid Zones Sciences Program, Arabian Gulf University, College of Graduate Studies, Salmanya, Manama 26671, Bahrain Published online: 04 Jun 2014. To cite this article: Marjan Modara, Mohamed Ait Belaid & Sabah AlJenaid (2014) Assessment of land-use/land-cover change in Muharraq Island using multi-temporal and multi-source geospatial data, International Journal of Image and Data Fusion, 5:3, 210-225, DOI: 10.1080/19479832.2014.904446 To link to this article: http://dx.doi.org/10.1080/19479832.2014.904446 PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &

Upload: sabah

Post on 04-Jan-2017

214 views

Category:

Documents


2 download

TRANSCRIPT

This article was downloaded by: [Karolinska Institutet, University Library]On: 04 June 2014, At: 10:13Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Click for updates

International Journal of Image and DataFusionPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/tidf20

Assessment of land-use/land-coverchange in Muharraq Island using multi-temporal and multi-source geospatialdataMarjan Modaraa, Mohamed Ait Belaida & Sabah AlJenaida

a Desert and Arid Zones Sciences Program, Arabian Gulf University,College of Graduate Studies, Salmanya, Manama 26671, BahrainPublished online: 04 Jun 2014.

To cite this article: Marjan Modara, Mohamed Ait Belaid & Sabah AlJenaid (2014) Assessmentof land-use/land-cover change in Muharraq Island using multi-temporal and multi-sourcegeospatial data, International Journal of Image and Data Fusion, 5:3, 210-225, DOI:10.1080/19479832.2014.904446

To link to this article: http://dx.doi.org/10.1080/19479832.2014.904446

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the“Content”) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoever as tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views of the authors,and are not the views of or endorsed by Taylor & Francis. The accuracy of the Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoever orhowsoever caused arising directly or indirectly in connection with, in relation to or arisingout of the use of the Content.

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &

Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Dow

nloa

ded

by [

Kar

olin

ska

Inst

itute

t, U

nive

rsity

Lib

rary

] at

10:

13 0

4 Ju

ne 2

014

RESEARCH ARTICLE

Assessment of land-use/land-cover change in Muharraq Island usingmulti-temporal and multi-source geospatial data

Marjan Modara, Mohamed Ait Belaid* and Sabah AlJenaid

Desert and Arid Zones Sciences Program, Arabian Gulf University, College of Graduate Studies,Salmanya, Manama 26671, Bahrain

(Received 18 December 2013; accepted 2 March 2014)

Employing geographic information systems (GIS) and remote sensing (RS) techniquesis found very important these days to aid city planners and decision-makers to makeeffective decisions and designs. The updated remotely sensed imagery and theirintegration into GIS allow the planners to continuously monitor changes and makeeffective decisions for future city plans. In this multi-temporal study, RS and GIS datawere integrated to map out, assess and quantify the land-use/land-cover changes(LULCC) in Muharraq Governorate (Island), in the Kingdom of Bahrain, from 1951to 2008. The data sets used are analogue topographic maps for the years of 1951, 1970,1977, 1985 and 1994, aerial photographs taken in the years of 1961, 2001, 2007, andfinally IKONOS Satellite imagery, taken in 2008 with ground spatial resolution of 1 m.The adopted methodology was based on the collection of data sets mentioned above:map scanning; geo-referencing and/or ortho-rectification. Visual interpretation wasperformed, and seven land-use/land-cover maps (classifications) were produced. Theresults showed that the total area of Muharraq has increased from 13 km2 in 1951 to56 km2 in 2008. Accordingly, built-up area from 2.3 to 9.7 km2, the industrial area wasinexistent in the past and now increased by 14.0 km2, coastal line from 35.0 to166.0 km2, the airport area increased from 2.3 to 5.9 km2, the vegetation area droppedfrom 9.5% to 5.1%, and finally, the length of road network increased from 11.5 to450.5 km. Based on the above findings, this study recommends to develop a databasefor the city of Muharraq by the integration of RS and GIS data sets, thereby the resultswill be very useful for decision-makers and planners. Furthermore, it is recommendedto allocate at least one beach for public uses.

Keywords: mapping; land use; land cover; GIS; remote sensing; Muharraq; Bahrain

Introduction

The Kingdom of Bahrain, an archipelago that consists of 36 islands totalling more than740 km2 is located in the Arabian Gulf. Its largest island, Bahrain, measures approxi-mately 600 km2 and represents the home of the majority of its population. Causewaysconnect the main island of Bahrain to Muharraq, the second largest as well as with Sitra.The kingdom is also connected to Saudi Arabia via the King Fahad causeway.

In this study, Muharraq city (island), the second-most important island in Bahrain hasbeen selected as a case study to map the land-use/land-cover changes (LULCC) within thecity for the past five decades (1951–2008) by the use of remote sensing (RS) data andgeographic information system (GIS), and to build a geospatial database for the city in aGIS environment to assess and analyse the driving forces of the urbanisation. This will

*Corresponding author. Email: [email protected]

International Journal of Image and Data Fusion, 2014Vol. 5, No. 3, 210–225, http://dx.doi.org/10.1080/19479832.2014.904446

© 2014 Taylor & Francis

Dow

nloa

ded

by [

Kar

olin

ska

Inst

itute

t, U

nive

rsity

Lib

rary

] at

10:

13 0

4 Ju

ne 2

014

help decision-makers in maintaining the city’s existing urban development and betterplanning for the future urban growth and projects.

The primary aim of this study is to map and assess the physical changes in Muharraqurban growth in the period from 1951 to 2008. Also this study is going to examine andanalyse the reclamation along the Coast of Muharraq Island and the evolution of itsshoreline during the same periods.

The specific objectives are as follows:

(1) Creating land-use/land-cover (LULC) maps of Muharraq for the years of 1951,1961, 1990, 1977, 1985, 2001 and 2008, using the collected data sets;

(2) Land-use/land-cover change (LULCC) detection will be carried out to assess thechanges occurring during the same periods of time;

(3) Mapping Muharraq Island shoreline evolution through the period 1951–2008;(4) Building and analysing Muharraq geospatial database based on ArcGIS software,

using data sets collected from different sources and Ministries.

Previous studies and study area

Previous studies

To better understand the aim of this study and its contribution to the field of LULCCdetection, it is important to consider what has been already done as well as what is knowncurrently and, thereafter, is a review of previous, but recent, literature on the topic,subdivided into four categories.

Change detection based on RS

The visual appearance of human-made constructs in IKONOS (1 m panchromatic) andRadarsat-1 (9 m resolution) satellite images were compared by Weydahl et al. (2005) whoconcluded that, although the two images had different resolutions, each contributed with aunique information useful to many mapping applications. The study also found that multi-temporal Radarsat-1 acquisitions could be used to detect human-made changes whenweather conditions hamper optical imaging techniques.

Phalke (2005) stated that the modern world needs accurate information about changesin human-made objects (especially buildings) for urban planning and geospatial informa-tion systems updating. Using positional information and feature matching techniques, thechanges between the 2001 and 2002 IKONOS images were detected by extracting thesame human-made objects (i.e. large buildings) from both images. To validate theobtained changes, the results were compared with the change obtained from both principlecomponent analysis (PCA) and supervised classification methods. The comparisondemonstrated that the new developed method provided much better results.

A visual interpretation was performed on pan-sharpen IKONOS imagery with 1 mspatial resolution (October 2000), covering representative areas situated on the coast andat the mountains in Lebanon. Afterwards, spatial and statistical comparisons were con-ducted between the results obtained from the interpretation of IKONOS imageries andexisting land-use inventories. The later were produced from Indian satellite images (IRS-1C, 5.8 m) merged with multispectral Landsat-TM (Thematic Mapper) images (30 m),both acquired in October 2000, using visual interpretation techniques. These comparisonsindicate the enhancement identification capabilities offered by IKONOS images in most

International Journal of Image and Data Fusion 211

Dow

nloa

ded

by [

Kar

olin

ska

Inst

itute

t, U

nive

rsity

Lib

rary

] at

10:

13 0

4 Ju

ne 2

014

cases and the limitations that can be offset in some other cases. IKONOS images show animprovement of land-use maps spatial accuracy in mountainous areas more than coastalareas. The visual interpretation of IKONOS imagery performed at 88.5% level of accu-racy, whereas the processing of Landsat-TM and IRS yielded 82% of accuracy. Thedegree of spatial coincidence between land-cover/land-use maps was equal to 87%(Faour and Bou Kheir 2006).

As multi-source data can be acquired more and more easily, change detection withmulti-source data has become a crucial issue. Because change detection with multi-sourcedata can eliminate the effects of the atmosphere and topography and improve the ability toidentify and extract objects, more accurate results can be obtained from change-detectionprocedures. This study aims to integrate GIS data with images into applications of changedetection. Change detection of linear, area and terrain features based on multi-source datais investigated, and change detection based on artificial neural networks and GIS data isalso analysed. As the powerful GIS functions provide efficient tools for multi-source dataprocessing and change-detection analysis, we can expect more research taking thisapproach as a generic trend in change detection (Li 2010).

Geographic information systems

Coastal land-use change for Murrells Inlet in South Carolina was studied as an integralpart of a five-year multidisciplinary and multi-institutional coastal research project fundedby the NASA/SC-EPSCoR, the design to develop and apply GIS-based methodologies foranalysis as well as modelling and prediction. The conclusion was that GIS has advantagesover conventional methods in integrating various data sources, performing spatial analy-sis, modelling spatial process and mapping the results in land-use change. A spatialmultivariate logistic regression model was successfully applied which developed 20variables that were selected for predicting the possibilities of land-use change forMurrells Inlet (Allen et al. 1999).

A study of the urban planning of Isa town in the Kingdom of Bahrain was conducted,and its stages of development since the implementation of the town plan in 1963 until theyear 2005 were assessed by using GIS techniques. Different multi-temporal data setsconsisting of statistical reports, aerial photography of the study area dated 1951 and threetopographic maps (scale 1:10,000) for the years 1977, 1991 and 2005 were used. Themethodology of acquisition and review of the data sets were adopted for preprocessing,including digitising of the maps and converting the data using Micro-station, ESRI andArcGIS, in order to produce an urban database including land-use maps of the study areafor three decades. Finally, the produced land-use maps were used for analysis of thedevelopment of the study area, and recommendations were proposed for future planningof Isa town (Haider et al. 2008).

GIS and RS integration

The Pudukkottai Taluk in Tamil Nadu, India, was characterised using RS data to deter-mine LULC classes and a combination of RS and GIS data to identify changes in LULCpatterns. Both basic data, which included information from IRS satellites, topographicmaps, and other ancillary reports, and ground truth data were utilised. The image analysiswas performed using ERDAS Imaging Processing software and a supervised classificationchange-detection method. The LULC classifications were obtained by applying the max-imum likelihood algorithm to various defined training areas. Changes in LULC patterns

212 M. Modara et al.

Dow

nloa

ded

by [

Kar

olin

ska

Inst

itute

t, U

nive

rsity

Lib

rary

] at

10:

13 0

4 Ju

ne 2

014

were determined using the ArcView (geo-processing extensions) package, which inte-grated the existing RS and GIS data sets. The authors concluded that satellite RS and GIStechnologies are valuable techniques for mapping and monitoring LULC and providingconsistent and accurate baseline data (Balakrishnan and Alomary 2004).

Ait Belaid (2003) analysed urban–rural land-use change using RS and GIS techniquesby conducting two multi-temporal studies for two different sites. The main objective wasto investigate the effects of the urbanisation on agricultural lands in three cities in theKingdom of Morocco (the first study), and in Al-Ahsa Oasis in Saudi Arabia. The firststudy was based on six panchromatic satellite imagery (two for each city) acquired bySPOT-HRV satellite during two decades, namely, 1980 and 1990. The second studyconcerning Al-Ahsa Oasis was based on two multi-spectral satellite imagery of Landsat-TM during two periods of time, namely, 1987 and 2001. In the two case studies,topographical maps and current ground observations were also used. Laboratory facilitiesused in terms of GIS and RS software were Arc Info and ERDAS Imagine.

Once again, Ait Belaid (2006) stated the potential of satellite imagery and GIS inmapping urban areas and monitoring urban changes over time. In this study, the mainobjective was to assess the extent of urbanisation and investigate its effects on the otherland-use categories in Al-Ahsa Oasis in Saudi Arabia and the city of Doha in Qatar. Thestudy was respectively on two (1987, 2001) and (1986, 2002) satellite images taken of thestudy sites from TM sensor of Landsat satellite, topographical maps and ground observa-tions. It was concluded that the cartographical and statistical results obtained after map-ping the land-use change in the studied areas are valuable tools for the protection ofagricultural lands and planning of urban and sub-urban areas located in the oases andcoastal fragile cities.

Light detection and ranging (lidar) technology offers precise information about thegeometrical properties of the surfaces and can reflect the different shapes and formationsin the complex urban environment. Generating a monitoring system that is based onintegrative fusion of hyper-spectral and lidar data may enlarge the applications of eachindividual technology and contribute valuable information on urban-built environmentsand planning. A fusion process defined by a data-registration algorithm and includingspectral/spatial and three dimension (3-D) information is developed and presented. Theproposed practical 3-D urban environment application for photogrammetric and urbanplanning purposes integrates the hyper-spectral (spectrometer, ground camera and air-borne sensor) and lidar data. This application may provide urban planners, civil engineersand decision-makers with tools to consider quantitative spectral information in the 3-Durban space (Brook et al. 2013).

LULCC detection

The multispectral scanner triplicates data in the early 1970s, mid-1980s and early 1990scollected from every national forest in the United States by the North AmericanLandscape Characterization (NALC) project, a part of the US National Aeronautics andSpace Administration’s (NASA) Landsat Pathfinder Program, were used to develop studyfor a change-detection training course. The study was designed to cover the basic methodsand options for developing change-detection maps, and included visual, post-classificationcomparison, image algebra, PCA, and unsupervised techniques. The study concluded thatall of these techniques were useful and important under different circumstances and,therefore, the advantages and disadvantages of each should be considered in the contextof one’s research before selecting a specific method (Grey et al. 1998).

International Journal of Image and Data Fusion 213

Dow

nloa

ded

by [

Kar

olin

ska

Inst

itute

t, U

nive

rsity

Lib

rary

] at

10:

13 0

4 Ju

ne 2

014

A report, conducted in 2005, provides an overview of the statistical and empiricaltechniques used in the spatial analysis of LULCC. The analysis of land-cover patterns andchanges is discussed, as well as changes in land-use and farming systems. The main focusof the report is on techniques for data reduction, structure detection and regressionanalysis. All methods are illustrated with examples. The report concludes that there isno single perfect method of approach for the analysis of LULCC. The selected methodshould fit the research questions, available data and resources. For a full understanding ofthe land-use change processes at work, a combination of methods and approaches mightprovide most insight, applied to a cohesive theoretical framework. Comparisons andcontradictory results should be used to signal weaknesses and demand a re-evaluationof the data and methods of analysis, and perhaps the underlying theory (Lesschen et al.2005).

Study area characteristics

The Kingdom of Bahrain

The Kingdom of Bahrain, an archipelago that consists of 36 islands totalling more than740 km2 is located in the Arabian Gulf. The largest and most populated is Bahrain Islanditself, where the capital Manama is located to the North. It represents 85% of the total landarea. The topography is flat, and the mean height do not exceed 60 m above the sea level.The highest point is located at Jabal Aldoukhan with 120 m height. In less than 20 years,the coastal zone of Bahrain has increased by about 40 km2 due mainly to rapid urbanisa-tion, land reclamation from the sea and dredging. In 2007, the ‘National StrategyDevelopment’ was submitted to the government of Bahrain, under the title of Bahrain2030 National Development Strategies. This plan is the first project to make inventory andpropose every aspect of new infrastructures. Furthermore, it addresses the role thatsustainable land-use development can play in guaranteeing stable, predictable and long-term economic growth.

The study area: Muharraq Governorate

The Kingdom of Bahrain consists of five Governorates of which Muharraq is the secondlargest in terms of population density (3813.46 people/km2) and the second in importanceafter the Capital governorate, according to Land Survey and Registration Bureau(Figure 1). Muharraq Governorate is the study area of the present research.

Located in the north east of the main island, it is connected to the mainland and to thecapital Manama by three bridges. Muharraq’s importance lies in the fact that it containsthe Kingdom’s International Airport, an industrial area which is to be 640 ha uponcompletion of the reclamation planned by the Ministry Of Industry (MOI 2004), andone of the biggest ship repair yards in the world (ASRY).

Rapid industrialisation combined with the urban sprawl is causing major pressures onthe land area of Bahrain. In recent years, there has been a considerable reclamation of landfrom the sea which has caused blocking of natural drainage channels killing date palmtrees and affecting mangrove swamps (Choudhury 2004).

Currently, there are few existing efforts to map and analyse the dramatic changes thathave taken place in the Muharraq Island. Ben-Hamouche (2004) presented a paper foranalysing the historical development in Muharraq using RS data and GIS techniques. The

214 M. Modara et al.

Dow

nloa

ded

by [

Kar

olin

ska

Inst

itute

t, U

nive

rsity

Lib

rary

] at

10:

13 0

4 Ju

ne 2

014

paper was more historically oriented rather than mapping the change in the urbanstructure.

In the past five decades, the island has been under dramatic and constant changes. Inthe 1971 census, the Island had an area of 13.9 km2 and a population of 49,540inhabitants with 7256 housing units. By 1981, its area had increased to 17.9 km2, witha population of 68,964 inhabitants and 10,260 housing units. Ten years later, the censusindicated an area of 31.45 km2 and a population of 82,855 inhabitants with 13,834housing units. And in 2001, the area was 33.98 km2, the population was 103,576inhabitants, and there were 17,388 housing units. The most current area estimation ofMuharraq Governorate is 51 km2 (Table 1) according to Central Informatics Organization(CIO 2007).

Figure 1. Location of study area (Muharraq Governorate).

International Journal of Image and Data Fusion 215

Dow

nloa

ded

by [

Kar

olin

ska

Inst

itute

t, U

nive

rsity

Lib

rary

] at

10:

13 0

4 Ju

ne 2

014

By studying the existing documentation particularly topographic maps and aerialphotography and satellite imagery over the period 1951–2008, the development andgrowth of Muharraq for eight periods were recognised: 1951, 1961, 1970, 1977, 1985,1994, 2001 and finally 2008.

Methodology and used equipment

The adopted methodology

Bahrain topographic maps are the only systematically collected data resource covering theentire kingdom since the 1950s. These maps are important tools to study and analyse theLULCC for geographical and environmental studies. Analysing land-cover data in aspatially clear way over longer time periods requires a combination of historical topo-graphic maps, aerial photographs and recent RS imagery. This study will utilise thetemporal capabilities of RS data to monitor the LULCC between 1951 and 2008. Thedata obtained from satellite images will be integrated with various geo-coded physio-graphic and socioeconomic data within the framework of a GIS spatial database to analysethe dynamics of urbanisation in Muharraq Governorate. The adopted methodology willfocus mainly on the following steps (Figure 2):

(1) Collection of available data sets composed of topographic maps, aerial photo-graphs, satellite imagery and statistical data;

(2) Satellite image and aerial photographs registration and/or ortho-rectification usingtopographic (e.g. basic maps) as reference maps;

(3) Classification based on visual interpretation (on screen digitising) of rectifiedor registered data sets (topographic map, aerial photographs and satelliteimage);

(4) LULCC detection was performed by differencing of earlier produced maps;(5) Evolution of shorelines of Muharraq was also studied for the same period of time

on the coastline of the island.

The used data sets

RS and GIS data were collected from different agencies, e.g. Ministry of Housing,Municipalities and Environment (MOHME 2001) for national statistics, and LandSurvey and Registration Bureau (LSRB 2007) for topographic maps and aerial photo-graphy, and finally the Arabian Gulf University (AGU) for IKONOS satellite imagery

Table 1. Muharraq socio-economic development (from 1965 to 2007).

YearArea(km2)

Population(Inhabitant)

Housing(Unit)

1965 13 46,373 67681971 13.9 49,540 72561981 17.9 68,964 10,2601991 31.45 82,855 13,8342001 33.98 103,576 17,3882007 50.23 15,9937 No Data

216 M. Modara et al.

Dow

nloa

ded

by [

Kar

olin

ska

Inst

itute

t, U

nive

rsity

Lib

rary

] at

10:

13 0

4 Ju

ne 2

014

Change-Detection Analysis

Output map production

Create a Vector data Model Buffering

Data entry intoArcGis9

Yes

Georeferenced No

Aerial Photos

Building a Geodatabase for Muharraq andMapping LULCC

Yes

No

Screening the dtacollected

Data collected issatisfactory to

use in the study

Analogue Maps

Scan

Clipping

Geometric correction toassign coordinate system

using ERDAS IMAGINE

Union

Editing

Rasterisation

Reclassification

Overlay

Satellite Images

Data Collection

Figure 2. Methodological framework including data input, processing and output.

International Journal of Image and Data Fusion 217

Dow

nloa

ded

by [

Kar

olin

ska

Inst

itute

t, U

nive

rsity

Lib

rary

] at

10:

13 0

4 Ju

ne 2

014

(Figure 3). It was then scrutinised and screened for most possibly to be used in thestudy so far as presented in Table 2.

Geo-referencing or geo-coding means geometric correction, while ortho-rectificationmeans advanced geometric correction using digital elevation model. The geo-referencingparameters used for this study are the following: World Geodetic System (WGS84) asGeodetic Datum, and Universal Transverse Mercator (UTM, Zone 39) as Map ProjectionSystem. Ground spatial resolution means the smallest area that can be detected in RS (e.g.pixel), which can also be called pixel size.

Figure 3. Ortho-rectified IKONOS image covering Muharraq City © IKONOS 2008 – ArabianGulf University.

218 M. Modara et al.

Dow

nloa

ded

by [

Kar

olin

ska

Inst

itute

t, U

nive

rsity

Lib

rary

] at

10:

13 0

4 Ju

ne 2

014

The used softwares

Topographic maps, aerial photographs and satellite images must undergo several prepro-cessing steps prior to use in LULCC detection mapping and analysing. These data setswill be utilised and processed using the following software provided by the AGU:

● Earth Resources Data Analysis System (ERDAS Imagine Ver. 9) software (ERDAS1999) is used to import generic binary data, display images, and perform geometriccorrection, as well as image analysis and conversion.

● ArcView and ArcGis (version 9.3) software (ArcGIS 2008) with the spatial analyst,imagine image support and geo-processing extensions were used to view raster andvector data, polygon overlay, preparing, editing and geo-coding descriptive infor-mation, topology building, overlaying and evaluating change detection.

● Microsoft Office Access 2007, Microsoft Word Office 2007 and Power PointOffice 2007, owned by the researcher, will serve as the main database, wordprocessing and presentation facilities.

LULC processing (output results)

LULC mapping

After studying and assessing the multi-source data sets of Muharraq during theperiod 1951–2008, and based on visual interpretation of data sets including topo-graphic maps, aerial photographs and satellite imagery, LULC categories weredivided into seven feature classes. LULC categories that were noted to be prepon-derant in Muharraq Island, or city or Governorate were as follows: roads, airport,cemetery, industrial, vacant, urban and vegetation. A spatial geo-database was cre-ated in arc-catalogue for each year of the study, containing all types of layers: shapefile and text file related to that year (Table 3). The average time period for theanalysed data sets is about 8 years. Road network has been increasing since the year1951 until 2008, from 0.07 to 5.41 km2. The area of the airport has been increasedfrom 2.34 to 5.95 km2. The area of the cemetery was changed slightly, from 0.28 to0.38 km2. The industrial zones did not exist until the year 1977, thus the areaincreased from 0 to 14.14 km2. The urban area has increased continuously from2.34 to 9.74 km2. Vegetation area increased from 1.22 to 2.71 km2, whereas thevacant area decreased from 6.07 to 4.3 km2 in the year 1994 and after that increaseduntil 18.46 km2 in 2008. This is due to the land reclamation to compensate the lackof land for other land-use categories.

Table 2. Data sets used in the research study.

Data type Source Scale FormatGeo

codingOrtho

rectification

Topography (1951) SLRB 1:12,000 Scanning Yes NoAerial Photo (1961) SLRB 1:20,000 Aerial Photo No NoTopography (1970) SLRB 1:5000 Scanning Yes NoTopography (1977) SLRB 1:10,000 Scanning Yes NoTopography (1985) SLRB 1:10,000 Scanning Yes NoTopography (1994) SLRB 1:10,000 Scanning Yes NoAerial Photo (2001) SLRB 1:25,000 Ortho-photo Yes YesIKONOS (2008) AGU GSR (1 m) Ortho-image Yes Yes

International Journal of Image and Data Fusion 219

Dow

nloa

ded

by [

Kar

olin

ska

Inst

itute

t, U

nive

rsity

Lib

rary

] at

10:

13 0

4 Ju

ne 2

014

The area is given in km2, for each class category and each period of time. In total,there are seven classes and eight periods of time (Figure 4). Figure 4 represents the area ofthe classes versus the periods of time. Figure 5 gives the area versus land-use classes.

LULCC detection

LULCC were determined by differencing between subsequent periods of time, separated byeight years intervals in average. It was realised that the changes for the periods of 1951, 1961and 1970 were minimal. Therefore, the starting point for change detection was from 1951 and1970. Furthermore, the changes were calculated for the following periods: 1951–1977, 1977–1985, 1985–1994, 1994–2001, 2001–2008 and finally the last column gives the total changesfrom 1951 to 2008 in km2 (Table 4). Table 4 allows illustrating the LULCC occurred duringthe entire period of time from 1951 until 2008. The area of Roads increased continuouslybetween the periods of time, but increased more between 2001 and 2008. In contrast, airportarea increased more between 1951 until 1977 followed by slow increase afterwards. Nonotable change has been noticed for the cemetery area. Industrial area increased slowly andmostly between 1994 until 2008. Urban area increased continuously, but mostly between2001 and 2008. Vacant area decreased between 1951 until 1985 and after that increased for therest of the period, but mostly between 2001 and 2008. Vegetation area increased during theentire period, but deceased between 1994 and 2001.

Table 3. Land-use/land-cover statistics (from 1951 to 2008).

Classcategories

1951(km2)

1961(km2)

1970(km2)

1977(km2)

1985(km2)

1994(km2)

2001(km2)

2008(km2)

Roads 0.07 0.25 0.37 0.65 1.61 2.42 3.28 5.41Airport 2.34 2.58 2.94 5.2 5.36 5.51 5.92 5.95Cemetery 0.28 0.28 0.32 0.38 0.38 0.38 0.38 0.38Industrial 0 0 0 0.95 2.59 2.64 7.13 14.14Urban 2.34 2.83 2.96 3.43 5.38 6.84 7.36 9.74Vacant 6.07 5.21 6.01 4.75 2.91 4.3 8.1 18.46Vegetation 1.22 1.73 1.65 2.22 2.3 2.46 2.18 2.71

0

2

4

6

8

10

12

14

16

18

20

1951 1961 1970 19851977 1994 2001 2008

Land-Use/Land-Cover Statistics (in km2)

from 1951 to 2008

Roads

Airport

Cemetery

Industrial

Urban

Vacant

Vegetation

Figure 4. Land-use/land-cover statistics per periods of time (from 1951 to 2008).

220 M. Modara et al.

Dow

nloa

ded

by [

Kar

olin

ska

Inst

itute

t, U

nive

rsity

Lib

rary

] at

10:

13 0

4 Ju

ne 2

014

Shoreline change and evolution

The shallow waters around the Muharraq Island made it possible to economicallyviable to reclaim land for new development and particularly for industrial areas, thuschanging the shoreline all around the island. The shoreline of the Kingdom of Bahrainwas drawn on the topographic map as the average high water mark by the Survey andLand Registration Bureau (SLRB) and it was identified on the existing topographicmaps of the years 1951, 1977, 1985 and 1994. It was determined by interpretation ofaerial photos for 1961 and 2001 and finally for satellite image for 2008. The length ofthe shoreline has increased successively from 34.85 km in 1951, 50.40 km in 1977,and 165.57 km in 2008. Table 5 gives the length and the corresponding area of theshoreline from 1951 to 2008.

Road network assessment

The only existing road in 1951 was the highway connecting the swing bridge fromManama to Hidd, with total length of 11.5 km. In 1961, construction of the airport ringroad, totalising a length of 21 km. In 1971, construction of old Muharraq Road, totalising

0

2

4

6

8

10

12

14

16

18

20

1951

1961

1970

1977

1985

1994

2001

2008

Figure 5. Land-use/land-cover statistics per classes categories (from 1951 to 2008).

Table 4. Land-use/land-cover changes (from 1951 to 2008).

Classcategories

1977–1951Area (km2)

1985–1977Area (km2)

1994–1985Area (km2)

2001–1994Area (km2)

2008–2001Area (km2)

2008-1951Area (km2)

Roads 0.58 0.96 0.81 0.86 2.13 5.34Airport 2.86 0.16 0.15 0.41 0.03 3.61Cemetery 0.1 0 0 0 0 0.1Industrial 0.95 1.64 0.05 4.49 7.01 14.14Urban 1.09 1.95 1.46 0.52 2.38 7.4Vacant −1.32 −1.84 1.39 3.8 10.36 12.39Vegetation 1 0.08 0.16 −0.28 0.53 1.49

International Journal of Image and Data Fusion 221

Dow

nloa

ded

by [

Kar

olin

ska

Inst

itute

t, U

nive

rsity

Lib

rary

] at

10:

13 0

4 Ju

ne 2

014

a length of 37 km. In 1977, construction of old town of Muharraq, Dair Samaheej andGalili road network, totalising 107 km. In 1985, construction of new roads for the newreclaimed areas, in addition to Arad and Hidd, totalising a length 152 km. In 1994, thelength of road network was 172 km. In 2001, north of Busaiteen got equipped by roads,totalising 318 km. In 2008, the length of road network becomes 451 km. The roadnetwork represents 0.57% of the total area of Muharraq Island in 1951 and has increasedto 9.53% in 2008 (Figures 6 and 7).

Table 5. Muharraq shoreline evolution in terms of length and area (from 1951 to 2008).

Shoreline 1951 1961 1970 1977 1985 1994 2001 2008

Length (km) 34.85 34.78 37.35 50.40 62.28 68.87 101.00 165.57Area (km2) 12.88 13.53 14.16 17.81 20.42 24.47 35.28 53.51

Figure 6. Muharraq area growth (from 1951 to 2008).

222 M. Modara et al.

Dow

nloa

ded

by [

Kar

olin

ska

Inst

itute

t, U

nive

rsity

Lib

rary

] at

10:

13 0

4 Ju

ne 2

014

Estimation of accuracy of LULC maps

The accuracy of the total LULC maps assessed and calculated within this study wascompared to the total area evaluated for each year separately, and a standard deviation wasestimated by the comparison of the two figures from existing data and produced in thepresent study, in terms of total area of the study through the entire period from 1951 to2008.

The differences in the values were due to the error in the interpretation of theboundaries between different LULC as well as to the complexity to differentiatebetween land-cover categories some time. The standard deviation ranges from −4.80%in the year 1961 to +6.13% in the year 2008, and the total mean were estimated at3.33%. The 6% standard deviation in 2008 is due primarily to estimate of roadsbuffering (Table 6).

Figure 7. Muharraq road network for the year 2008.

International Journal of Image and Data Fusion 223

Dow

nloa

ded

by [

Kar

olin

ska

Inst

itute

t, U

nive

rsity

Lib

rary

] at

10:

13 0

4 Ju

ne 2

014

Conclusions and recommendations

The objectives of this multi-temporal study were to provide consistent LULC maps andstatistical inventory, along with changes which occurred in Muharraq Island from 1951 to2008, in order to integrate photo-interpretation with statistical data using GIS, and to examinethe potential of the GIS and RS integration in studying LULC and shoreline evolution.

RS data showed to be very efficient source to produce LULC maps and also LULCCand shoreline change/evolution through the use of GIS techniques by storing, manipulat-ing and analysing RS images and analogue maps. Most importantly was to produce theresults by presenting clearly LULCC and shoreline digitised maps.

Conclusions

(1) The LULCmaps alongwith corresponding statistical inventorywere produced properlyfor eight successive periods of times and for seven class categories, from 1951 to 2008.

(2) The LULC changes were also produced by differencing documents and productsestablished earlier.

(3) Muharraq Island development has been studied in concert with the developmentof its shoreline.

(4) The LULC were governed by geographical and socio-economic conditions,including population growth, development policies and economic development.

(5) Urban development in Muharraq has been driven mostly by the increase ofpopulation and urbanisation, and also by the improvement of road network.

(6) With the government’s economic diversification policy in the 1970s, the industrialarea was added to the LULC classification in Muharraq.

(7) The shallow water around the island made it economically viable to make landreclamation from sea, for new developments particularly for industrial area, thuschanging the shoreline all around the island.

Recommendations

(1) Developing an up-to-date geospatial database (or urban GIS) for Muharraqincluding existing data sets in various Ministries and Department in Bahrain,for urban planning and monitoring of the city.

(2) This will provide a framework to the decision-makers (e.g. municipalities) todevelop an action plan for better services and management of the existinginfrastructures, facilities and services.

Table 6. Estimation of the accuracy of LULC maps.

Total area1951(km2)

1961(km2)

1970(km2)

1977(km2)

1985(km2)

1994(km2)

2001(km2)

2008(km2)

Existingstatistics

12.88 13.53 14.16 17.81 20.42 24.47 35.28 53.51

Calculatedstatistics

12.32 12.88 14.25 17.58 20.53 24.55 34.35 56.79

Standarddeviation (%)

−4.35 −4.8 0.64 −1.12 0.54 0.33 −2.64 6.13

224 M. Modara et al.

Dow

nloa

ded

by [

Kar

olin

ska

Inst

itute

t, U

nive

rsity

Lib

rary

] at

10:

13 0

4 Ju

ne 2

014

(3) The dramatic changes and development of the shoreline during the investigated periodof time (e.g. 47 years) is due mainly to dredging and land reclamation on the coastalzones. This will have a negative impact of course on marine and coastal biodiversity.

(4) Possibility to allocate at least one beach free for the public usage equipped withrecreation services and sport facilities.

ReferencesAit Belaid, M., 2003. Urban-rural land use change detection and analysis using GIS and RS

technologies. In: 2nd international federation of surveyors (FIG) regional conference, 2–5December, Marrakech. Morocco: FIG.

Ait Belaid, M., 2006. Remote sensing of desert cities and urbanization management in West Asiaregion. In: Map Middle East 2006, 26–29 March, Dubai. Map Middle East.

Allen, J.S., et al., 1999. A GIS-based analysis and prediction of parcel land use change in a coastaltourism destination area. In: The world congress on coastal and marine tourism, Vancouver,British Columbia, Canada.

ArcGI, S., 2008. Getting to know ArcGIS desktop. 2nd ed. Updated for ArcGIS 9.3. Redland, CA:ESRI Press.

Balakrishnan, P. and Alomary, A.Y., 2004. Detection of land use/land cover changes using remotesensing and GIS techniques and its application to monitoring Pudukkottai Taluk area. JordanJournal of Applied Science and Natural Sciences, 6 (2), 8–31.

Ben-Hamouche, M., 2004. GIS, Remote sensing and urban history, analyzing Muharraq citydevelopment. In: Third international conference & exhibition (GIS 2004), 27–29 September,Kingdom of Bahrain. Bahrain Society of Engineer.

Brook, A., Ben-Dor, E., and Richter, R., 2013. Modelling and monitoring urban built environmentvia multi-source integrated and fused remote sensing data. International Journal of Image andData Fusion, 4 (1), 2–32. doi:10.1080/19479832.2011.618469.

Choudhury, P., 2004. Land use/land cover mapping in Bahrain using landsat-tm data. In: Thirdinternational conference & exhibition (GIS 2004), 27–29 September, Kingdom of Bahrain.Bahrain Society of Engineer.

CIO, 2007. Bahrain in numbers 2007. Kingdom of Bahrain: Central Informatics Organization.ERDAS, 1999. Earth resources data analysis system, field guide. 5th ed. Atlanta, GA: ERDAS.Faour, G. and Bou Kheir, R., 2006. Effectiveness of using very high resolution imagery (IKONOS)

for land use mapping [online]. Available from: http://www.gisdevelopment.net/technology/ip/techip_002.htm [Accessed 15 March 2010].

Grey, D.R., et al., 1998. Change detection using NALC MSS triplicates to set forest planningcontext. In: Integrated tools, 16–20 August, Boise, ID.

Haider, L., Ait Belaid, M., and Khalil, A., 2008. Analysis of urban planning in Isa town using GIStechniques. International Archives of Photogrammetry, Remote Sensing and Spatial InformationSciences, 37, (Part B8).

Lesschen, J.P., Verburg, P.H., and Staal, S.J., 2005. Statistical methods for analyzing the spatialdimension of changes in land use and farming systems. Nairobi: The International LivestockResearch Institute.

Li, D., 2010. Remotely sensed images and GIS data fusion for automatic change detection. InternationalJournal of Image and Data Fusion, 1 (1), 99–108. doi:10.1080/19479830903562074.

LSRB, 2007. Provider of topographic maps and aerial photography’s, land survey and registrationBureau, Bahrain.

MOHME, 2001. Bahrain Report 2001 for Istanbul +5. Bahrain: Ministry of Housing,Municipalities & Environment.

MOI, 2004. Kingdom of Bahrain industrial facts & opportunities. Bahrain: Ministry of Industry.Phalke, S., 2005. Change detection of man-made objects using very high resolution images. Thesis

(MSc). University of Calgary, CA.Weydahl, D.J., Bretar, F., and Bjerke, P., 2005. Comparison of RADARSAT-1 and IKONOS satellite

images for urban features detection. Information Fusion, 6 (3), 243–249. doi:10.1016/j.inffus.2004.07.001.

International Journal of Image and Data Fusion 225

Dow

nloa

ded

by [

Kar

olin

ska

Inst

itute

t, U

nive

rsity

Lib

rary

] at

10:

13 0

4 Ju

ne 2

014