Monitoring urban land use/cover changes provide spatio-temporal information on the pattern and the amount of the changes that has taken place across the land use/cover classes, and the information obtained from the urban expansion are valuable for sustainable planning and management of urban resources. Remote sensing techniques provide fast and effective means for classifying and mapping urban land cove/use change through examining spectra characteristics of multi-dates satellite images. In recent years, Kurdistan region cities have witnessed a rapid growth of the urban rates due to previous socio-economic and political variations in the area. The purpose of the study is to analyse and visualize spatial pattern of urban land use changes in Erbil city-Kurdistan and to quantify the amount of variations in the land use classes by applying remote sensing approach.
The research examines multi-dates Landsat 5 TM imageries for 1987, 2000 and 2011 by using supervised classification of maximum Likelihood classifier to display and measure the changes in the land use/classes in ERDAS 9.1 imagine processing software. The accuracy of the overall classification was measured by using confusion metrics and Kappa coefficient to test overall accuracy classification. The study achieved an overall accuracy which rate from 95, 96.43 and 94.29 in 1987, 2000 and 2011 respectively, which indicates that the research has achieved a significant level of classes.
The outcome of the study revealed that the study area has remarkably experienced changes in its land use/cover during the period of the study as built up area was increased by (3975.66 to 6123.7 hectares) over in 1987 to 2000, and (6123.7 to 12755.1 hectares) in 2000 to 2011. On the other hand, the amount of other classes has greatly declined during the period of the study from. Evidence from the post classification analysis has shown that open land and vegetation classes have experienced the most significant changes of rates in the urban land during the period of the research. For example, the rate of changes from open land to urban land is (5084.46), whereas the rate of the vegetation that converted to urban land is (2130.69).
At the very beginning, I would like to sincerely thank my almighty Allah for rewarding me patience and health that enable me to conduct this study. Then, I would like to express my gratefulness to my supervisor Andrew Jones for his remarkable supervision, constant help, objective comments and constructive critiques.
I would also like to thank all the teaching and administrative staffs at Sheffield Hallam University which assist me throughout being here for two years. Also my grateful thanks are due to all my colleagues in Sheffield Hallam University for their constant support and encouragement for conducting this work.
Also I would like to thank the companion of my life and all my family members especially my partner and dear wife that have shared me with all the difficulties that result from being in diaspora.
I want to express my gratitude to the entire staff of Geography Department at the University of Soran especially Dr. Kamaran Walie. Also I have a great appreciation for Heidi to the aid that he provided to me. I am also grateful for Mr. Bikhtiyar Omar for his constant help.
Finally, I would like to express my gratitude to the Ministry of High Education and Scientific Research/ Kurdistan Regional Government for giving me a chance to do my Msc in The United Kingdom and for their sponsorship.
List of Figure:
Figure 1: data transmission processing and analyzing in remote sensing procedure source
Figure 2: rate of population growth in Erbil governorate from1957 to 2009
Figure 3: workflow of clarify steps of study methodology
Figure 4: Landsat 5 TM Images after layer stacking process
Figure 5: Subset Study site
Figure 6: Combination of band RGB 4, 3 and 2
Figure 7: image Pre-processing Tasselled cap (TC) in 1987-2000 and 2011
Figure 8: Land cover classes for the year 1987, 2000 and 2011
Figure 9 illustrate the percentages of changes in land cover classes between 1987, 2000 and 2011
Figure 10 Map of Land use re-classification in Erbil city 1987. 2000 and 2011
Figure 11 Graph of illustrate the result of land use re- classification in the period of the study
Figure 12: Overlaid maps illustrate the urban expanded in 1987, 2000 and 2011
Figure 13: post-classification to detect Land cover change detection between 1987 and 2011
Figure 14: Summary of error matrixes for the supervised image classification of 1987, 2000 and 2011
List of Table:
Table 1: Detailed characteristics of the Landsat satellite images with a plan used for this study
Table 2: Secondary data was used for this study
Table 3: Summary results of land cover changes in 1987, 2000 and 2011
Table 4: Result of land use re- classification in the period of the study
Table 5: Land covers classes between 1987 and 2011
Table 6 accuracy assessment results for years 1987, 2000 and 2011 which shows a remarkable increased in the built-up class in the year 2011.
List of Abbreviations
illustration not visible in this excerpt
Chapter one: Introduction
Rapid urban growth has led to shifts in land use/cover in many major cities across the world, particularly in the emergent countries (Amin Abdu-Allah, 2007). Erbil City is one of the cities that have shown a dramatic change in this respect. It is a fact that urban change results from the urbanization processes. Urbanization refers to the process of changes in the number of urban dwellers (Op cit). This indicates that a high expansion of resident’s areas is lying inside the metropolitan covering the immigration of people from rural areas to the city (Zhou, 1999; Hall, and Pfeiffer, 2000; Khaleel and Ngah, 2010).
Population growth and urbanization are posing a significant risk on the ecosystem and other socioeconomic growth across the globe (Rawashdeh and Saleh, 2006). In the last five decades, the world has experienced a significant increase in the rates of urbanization. Considering the evolution and alteration resulting from urbanization is crucial in order to examine the dynamics of urban areas and manage the resources in a better way as well as providing infrastructures to need up with the growing demands that are associated with the fast growing of the urban centres (Rawashdeh and Saleh, 2006).
Urban population grows faster than the rural area, as urban areas experience high rates of migration. Darnell (2008) states that the population of the urban areas was estimated to about three billion people, and it is expected to increase by 60 % in 2030 when nearly two-thirds of the world's population will be residing in the urban centres. Last century was recognized by the urbanization, whereas this century is witnessing urban conversion (Hall and Pfeiffer, 2000). Whitehouse (2005) states that, in the mid-20th century there were 83 mega cities in the world which had been increased to 411 cities by 2000. In Iraq there were a great demographic change of urban population as people migrate from countryside to urban areas which resulted to 63% of population lived in the cities and 37% lived in the countryside in 1987, while in 2004 urban population significantly increased to 80% , whereas rural population decreased to 20% (Erbil province, 2012).
Urban growth is associated with the changes in the land use/cover pattern. Urbanization reflects positive socio-economic and political growth which leads to developmental process in the innovation systems that alters the socioeconomic actions and transforms the practice of using land in accordance with the land use policies (Rimal, 2012). Urbanization has numerous environmental impacts, such as deforestation which could be resulted from constructional works in the process of urban expansion which turn leads to desertification and drought. Batisani and Yarnal (2009) have analysed the environmental effects of urbanization such as the changes in land use that are being converted to urban land. All over the world, agriculture land is continually experiencing transformation to urban habits in the process of urbanization (Li and Yeh, 2000). Globally, significant portion of agricultural land is being converted to urban areas and by the year 2050, 5.7% of damage agricultural land will be recorded but some regions tend to loose larger share of their agricultural land like Asia is predicted to lose over 10 per cent, southeast Asia might lose about 10 per cent, 8 per cent in the South and Central Asia, Western Asia and North Africa by 10 per cent while East Asia might lose nearly 7 per cent of its agricultural land (Rimal, 2012). The loss of agricultural land could be largely resulted from the urban expansion. Namely, the developing countries with higher population growth and higher rates of rural urban migration tend to witness significant change in the land use/cover. More importantly, the rapidity and scale of this evolution is typically determined in the developing countries which are characterized by larger urban areas and a considerable amount of extra-large cities (Cohen, 2006).
Erbil city has witnessed a significant urban growth since the year 2000 and it has passed an essential conversion phase. Earlier investigation has shown that total cultivated land decreases annually, because arable land transformed to relevant build-up land and open land due to urban growth Erbil city (Erbil province, 2012). Assessing the size and pattern of urban development in Erbil city is in a critical need.
Erbil city lack a suitable land use policies as the city planners are still adopting conventional approach of land use policies in order to support planning, management and decision makings. Rapid changes in the rates of urbanization require advance technological equipment to detect urban growth in order to assist plan better and make suitable decision. Naseem et al. (2003) suggest that urban growth can be estimated by using census data through measuring of population change, but it is not enough to analysis all patterns of urban growth. Moreover, integrating census data with remote sensing and GIS technology leads to best explanation of the direction of urban growth. Remote sensing is an effective way to monitor the land use/cover change, particularly when the land use data is inadequate in the area.
Bektas and Goksel (2005) argued that the geographic information systems (GIS) aggregation with satellite remote sensing have been broadly used, and they offer excellent means of detecting urban changes in the cities all over the world. Satellite remote sensing entails assembling of multi- data (multi- spectral, multi-resolution and multi-temporal data), and converts them into appreciated data to consider and monitor the procedure of urban land for structure urban land cover data arrangements. Likewise, GIS is an excellent tool that has been broadly used for monitoring urban growth as well as visualizing the level of change (Rimal, 2011). The study examines urban growth in the study area by using remote sensing techniques.
1.2: Significance of the study
The cities in Iraq have responded to the changes in economic and political changes that took place at the level of local, national and global. It is observed that the rapid growth of urban areas was challenging for city planners and authorities. Remote sensing and GIS is efficient tools that have the ability to display the rate and the pattern of changes in urban areas, like Erbil city. This research measures and analyses the changes of urban land between 1987 and 2011, thus it is expected to:
- Offer basic data about the situation and changes of Land cover in the urban areas by using multi-date satellite imageries; also it is expected to clarify the level of Land Cover changes and to measure the rate of urban growth in different years.
- Provide information to help regional and urban planners and policy makers to get benefit from Geo-spatial tools for planning and monitoring the urban expansion.
- Assist in monitoring and observation for a long term planning.
- Offer some recommendations for further studies.
1.3: Statement of the problem
Recently, rates of urbanization are on the increase across the globe particularly in developing countries (Matuschke, 2009). Cities have developed in population and size; the issue of urbanization attract urban planners, ecologists, sociologists, engineers and policy makers (Riggs and Simmons, 2002). In Erbil city, during the last 24 years large areas of open land and agriculture are transformed to urban areas deprived of considering its effect on social, economic, physical and environment. In the developing countries, cities facing many problems due to rapid urbanization, as the process of urban growth characterised by lack of plan and information to predict the direction of growth. this limitation in this filed should be addressed by carrying out such studies and applying remote sensing and GIS techniques in order to filed the gap in this area.
Erbil city expands haphazardly as a result of population growth, Political changes and economic growth. In Erbil city, between 1987 and 2011, a large area of open land and agricultural land has been transformed to urban areas (Khaleel and Ngah, 2010). Transformation in agricultural lands in the urban areas greatly affects the land components and the environment. In Erbil city, urban expansion rose dramatically in the last two decades and it is continued to grow. This poses a need for efficient planning and information, which can be obtained by using remote sensing techniques to examine accurately urban growth in the study area within the period of the study. The outcome of this study will assist policy makers to inform future urbanization policies and this research is valuable worth of studying. The research adds a value to the existing body of literature through adopting the unique techniques as few studies were conducted by using different methods in the area. The outcome of the study will be helpful in acknowledging town planners with the significance of using advance technological equipment in monitoring urban growth.
1.4: Aim of the study
The key purpose of this study is to detect urban growth in the study by making use of utilising remote sensing techniques that will be achieved through the following objectives:
- To detect urban changes by using Landsat satellite images for examining the land use classification and for determining the land use alteration of Erbil City for the period of the study.
- To identify the quantity of changes in urban transformation (built-up) land on different dates (1987, 2000 and 2011).
- To identify and monitor urban expansion by using change detection methods.
- To make suggestion for further research.
1.5: Research question
- What are the nature urban expansion and the associated trends and patterns?
- What is the rate of land use change that has occurred in the Erbil city between 1987, 2000 and 2011?
- Which classes of the land use were mostly converted to urban land?
- How the rate of the urban growth between 1987 to 2000 differs from the rate of the urban growth from 2000 to 2011?
1.6: Organization of the Thesis
This thesis consists of five main chapters. Chapter one is an introductory one that involves the aims of the study, the research question, the problems as well as the significance of the research. Chapter two discusses the literature review related to urban growth. Chapter three is devoted to the research method and the methodology that is used in the research. Chapter four interprets the results and discuss the study. Finally, chapter five covers conclusion, limitation and recommendations for future study.
Chapter two: Literature review
Human activities such as urbanization and mining among others have greatly influenced the patterns of land use/cover across the earth surface. Over the centuries, the earth surface has experienced significant changes in land use/cover types as transformations have occurred from one class to another (Agriculture, forest, open land, barren land, and cropland). For example, agricultural land converted to built-up area. The rapid changes in the Land use/cover can be better monitored by using remotely sensed devices. In the previous chapter, an overview of the research was covered while this chapter explores the theoretical issues in relation to monitoring urban land cover/use detections. The chapter two is structured as follows; section two covers a brief history of remote sensing and application of remote sensing in identifying urban land use change by using satellite imageries, section three examines the urban change detection and the fourth and last part outlines image classification and accuracy assessment.
2.2: History of application Remote Sensing
The concept of remote sensing focuses on collecting data from a distance without having physical contact through a platform (sensor). Lillesand et al., (2008) describe remote sensing as the science of providing information about objects, places and phenomenon by analysing the data obtained from devices which has no direct contact with the phenomenon, area or object that is under study.
Campbell and Wynne, (2011, p.6) pointed out that “remote sensing is practice of deriving information about the Earth's land and water surface using images acquired from an overhead perspective, using electromagnetic radiation in one or more regions of the electromagnetic spectrum, reflected or emitted from the earth's surface".
Campbell and Wynne, (2011) and Lillesand et al. (2008) explained that the remote sensing concept offers a valuable insight to the research, because the research adopts the concept of remote sensing in collecting data (remotely sensed data), processes, analyses the data to obtain information on urban land use/cover change through an integrated remote sensing and Geographic Information System (GIS) approach.
Sensors are classified into passive and active. Passive remote sensing system obtains energy from the exterior source (solar energy) (Aggarwa, 2003; Navalgund, 2002), while active remote sensing system generates illumination from itself, such as a laser fluourosensors and a synthetic (Canada Centre of Remote sensing, 2007; Levin, 1999). The key benefit of active is the ability to collect data at any time of the day with irrespective of the terrain (op cit). The energy source such as electromagnetic spectrum (EMS) is the vital medium for transferring information from the object to the satellite sensor, which depends on the quantity of energy radiated, reflected, and emitted by objects into the atmosphere (Aggarwa, 2003) (see figure 1). The radiation from the visible part of the electromagnetic spectrum (ultraviolet and infrared, 0.4 to 0.7μm) can be seen by human eyes as colours (rainbow colours) (Walsh, 2003). Some parts of electromagnetic spectrum are not visible to the human eyes (X-ray gamma ray, and radio waves), but visible to remotely sensed devices (Op cit). The invisible part of the spectrum contains vital information about the earth surfaces. Every object imitates radiation on electromagnetic spectrum based on its physical properties (temperature and emissivity) and objects are identified or differentiate based on their spectra signatures for instance, vegetation reflects light, while rocks and buildings do not (NASA’s Landsat, 2006).
illustration not visible in this excerpt
Figure 1: data transmission processing and analyzing in remote sensing procedure source (Ali, 2010).
Systems of remote sensing provide 4 direct elements of recording and quantifying data around the earth surface remotely; According to Adam, (2010, p. 14), “these components include the energy source, the transmission path, the target and the satellite sensor”. Remote sensing is an interesting tool that provides data on natural resources in different temporal and spatial resolutions (Wulder et al., 2004). Technological advancements have been made in modern remote sensing, for instance, advancements in satellite platforms, sensors, optics (creation of camera), display place, computer data processing and transmission (Adam, 2010). In the past, different types of air platform were applied to monitor the earth surface (DE Sherbinin and Giri, 2001). In the 1859 in Paris, aerial photograph from a balloon was used to monitor the earth surface; it represents an unrivalled source for observing urban area, particularly by using the existed tools that supplied by “photogrammetry” (Baudot, 1992; Patino and Duque, 2013).
The lunched of Landsat imagery (medium 30 meter spatial and 16 days temporal resolution) in the early 1970s and it continuous series up to seven has led to the broad and increasing adoption of remote sensing for resource and environmental management (Bekalo, 2009). The viability of satellite imageries, rapid advancement in computer and GIS technology further resulted in the wide applications of remotely sensed data in diverse disciplines, such as ecology, land use/ cover change, forestry and resource management. Currently, other high resolution satellite images like Ikonos1-2 with (1 m) resolution in panchromatic and 4 m in (MSS) and Quick bird satellite which has a resolution (2 m) in (MSS) and (0.60 cm) in Panchromatic offer consistent, current and precise geospatial data were developed based on the Landsat criticism by United State (US) government and the high resolutions satellite images have being widely accepted into diverse discipline; it is valuable for the study to understand the concept of remote sensing and Landsat imageries (Bekalo, 2009).
Landsat imageries are excellent in analysing urban land use land change through classification. The high resolution satellite imageries like QUICKBIRD and IKONOS (less than 1meter spatial) are important in transport modelling, because small items like buildings and cars can be seen (Rashed and Jürgens, 2010). Several studies analyses urban growth through analysing medium resolution imageries due to its free access, in monitoring and measuring the patterns of urban expansion particularly in large areas that witnesses rapid changes in the land use (Yeh, and Xia, 2001). In addition, 30 meter spatial resolution of Landsat TM imagery is suitable for identifying/visualizing changes and quantifying the amount of variations in the urban Land use/cover classes and this rationalize the choice of the medium resolutions imageries for the study. Earlier studies ((Kärdi, 2007; Donnay et al, 2003; Verma et al, 2009) have suggested that urban monitoring model requires high resolution satellite imageries due to the intricacy and mixed nature of urban land use classes and the enhance of the spatial resolution that reduces difficulties in mixing pixels. In addition, detail landscape such as individual buildings and details of transport networks can be easily identified with high resolution satellite imagery in monitoring urban sprawl. Although high resolution satellite imageries (IKONOS,1 meter pixel resolution) is excellent for classifying land use/cover, it is expensive, whereas medium resolution imagery (Landsat-5 TM) is free, but still is ideal for detecting spatio-temporal land use/cover changes in the study.
2.3: Identifying urban land use change
‘Land cover describes the physical material at the surface of the earth. Land cover types such as trees, bare soil, grass, water etc. Whereas land use is a description of how people utilize the land and socio-economic activity such as urban land, agricultural land, water land, grazing land etc’ (Bekalo, 2009, p.17).
Urban sprawl is an inevitable phenomenon, but the spatial pattern of urban expansion can be analysed, planned and managed effectively in order to sustainably develop urban resources. Conventional techniques (aerial photos, ground survey etc.) of collecting data about the earth surface are not effective in delivering the essential data in a timely and cost effective ways and urban planners require current urban information for better planning and decision making processes (Ayele, 2011; Herold et al., 2002).
Remote sensing provides detailed spatial coverage, however it is limited in terms of distinguishing characteristics of the urban environment based on their spectral reflectance characteristics, classification, analysis and modelling of urban sites which depends on the support of spatial information and the analytical ability of GIS (Donnay et al., 2003). As a result of remote sensing restrictions, it has been integrated with a global positioning system (GPS) and GIS to analyse changes in the land use/cover more efficiently than it was with only remotely sensed data (Xiao et al., 2006). Remote Sensing and GIS techniques have been widely used to monitor urban land use/cover changes through collecting, storing, analysing, and presenting of geospatial data for different purposes (Bahrain, 2003). Remote sensing and GIS based approaches have proved to be excellent in providing valuable insights on identifying the type and quantity of land use/cover change that prevails in an area (Sundarakumar et al, 2012; Gupta, 2011).
Social researchers’ perceptions differ from geospatial specialists in applying remote sensing for urban study. Social scientists perceive remote sensing as just a visualizing tool that is only limited to display functionality while geospatial scientists consider remote sensing as a tool that is valuable in identifying physical characteristics of an urban area and disregarding the effect of social phenomenon and ecological consequences of different social, economic or demographic processes on urban centres (Netzband, 2010). Understanding the theories of social and geospatial scientists of remote sensing is valuable for the research because the theories provide conceptual insights that support examining urban growth through identifying the physical (spectra) characteristics of urban land use/cover on the electromagnetic spectrum (spectra signature) and the effects of socioeconomic phenomena on urbanization which is the focus of the research. Socioeconomic and demographic changes are essential parts of urban growth.
Remote sensing GIS applications in urban studies have been well documented by (Weng, 2002; Yeh and Xia, 2001; Xue, 2012; Xiao et al., 2006). Soffianian et al., (2010) examined urban sprawl in Isfahan, Iran using multi-date remotely sensed imageries. They analysed both aerial photographs and Landsat MSS, TM, ETM+ by visual interpretation and image processing approaches. The outcomes of their research indicates that built up class grew by nine times and significant portion of green space in the study area has been converted to residential areas while the population of the study area rose by about six times. The findings of their research have shown that the pattern of urban growth in Isfahan city was parallel to the areas of road network expansion, main industrial and infrastructural development (schools and hospitals). Soffianian et al., (2010) research is relevant to the current study, because it analysed Landsat satellite multidates imageries and examined spatio-temporal pattern of urban expansion in relation to other socioeconomic development in their study area. The study area has experienced a remarkable political and socio-economic development over the last two decades. The spatial pattern of urban land use/cover transformation of the study area will be explored through analysing aerial photographs with Landsat TM and IRSLISS-III satellite imagery using maximum likelihood supervised classification. The current study might explore a similar or different insight since it applies a different approach (remote sensing based); and the two study area differs in their recent political and socioeconomic characteristics from the Soffianian et al. (2010).
Weng (2002) observed urban land use/cover change using remote sensing, GIS and markov model in the Zhujiang Delta, China. Their research analysed Landsat TM imageries for different years and group them into six classes using maximum likelihood algorithm (supervised classification) with high level of accuracy. The results of their study show that a remarkable change in the land use classes have occurred within the period of their study with a total of 12.8 per cent as built up and arable rose from 47.68 per cent to 88.66 per cent while cropland declined by 48.37 per cent. Their study observed that urban sprawl was irregular in different part of the study area. Weng (2002) findings reveal that population and economic development was the key factor that contributed to the urban growth of their study area. Their study is valuable for the present research because of the similarities in the data obtained (Landsat TM) and the method of classification and the insight from their research useful for better understanding the impact of demographic and other socioeconomic factors on the urban growth of the study area.
Landsat satellite imageries are popularly applied for monitoring urban growth but recently high resolution imageries, like IKONOS and QUICKBIRD are gaining acceptance in urban studies. Chen et al., (2000) conducted a study in four Chinese cities Beijing, Shanghai, Dongguan and Chongqing to examine spatial and temporal patterns of urban growth in their study area by using both low resolution (Landsat TM image, Landsat MSS), high resolution satellite imageries (SPOT 10m) and aerial photographs the outcome of their research shows that there was a remarkable increasing in the build-up areas in all the Chinese cities. In Beijing the built-up areas increased nearly 39 per cent from 1987 to 1994, whereas in Shanghai the maps visualized show a great expanding in the built up areas due to rapid rates of industrialization, population growth and migration. Their findings have revealed that high resolution satellite imageries (IKONOS, QUICKBIRD) are more effective in identifying physical characteristics of the urban areas than the low or medium resolution satellite imageries. High resolution imageries are not appropriate for the present study, because it cannot be obtained for free and very expensive. Earlier studies discussed above have highlighted that medium resolution satellite imageries (Landsat imageries) have been successful in identify urban land use classes through identifying spectra characteristics of the urban area and their findings consistently reveal that urban expansions evolved parallel to areas of major socioeconomic development like industries, major roads and high population density centres among others. Their studies were conducted in different areas with different political and socioeconomic setting from the study area of the current study; as a result, the findings from the current research might be consistent or different from the earlier studies above. A factor responsible for urban expansion in one area might be the key factors that influence urban land use/cover in the study.