Thesis
Thesis
Thesis
===========================================
BY
ZEWDU ALEBACHEW
JUNE 2011
========================================================================
Urban sprawl mapping and land-use/land-cover change statistics 2011
==========================================================================================
June 2011
ii
------------------------------------------------------ ---- ---------------------------------------------------------------------------
Urban sprawl mapping and land-use/land-cover change statistics 2011
==========================================================================================
ADDIS ABABA UNIVERSITY
URBAN SPRAWL MAPPING AND LAND USE CHANGE ANALYSIS USING REMOTE
SENSING AND GEOGRAPHIC INFORMATION SYSTEM
Examiners
Advisor
June 2011
iii
------------------------------------------------------ ---- ---------------------------------------------------------------------------
Urban sprawl mapping and land-use/land-cover change statistics 2011
==========================================================================================
DECLARATION OF ORIGINALITY
The work has not previously been submitted in any form to the Addis Ababa University or to
any other institutions. This thesis is a presentation of my original research work. Wherever
contributions of others are involved, every effort is made to indicate this clearly, with due
--------------------------------------
Addis Ababa
Ethiopia
------------------------------------------
K.V. Suryabhavagavan(Ph.D)
iv
------------------------------------------------------ ---- ---------------------------------------------------------------------------
Urban sprawl mapping and land-use/land-cover change statistics 2011
==========================================================================================
Abstract
This study focused on the use of Remote Sensing and Geographic Information System
in mapping of urban growth and sprawl a case study of Debre Birhan town. A high
spatial resolution of Ikonos image and topographic map were used to study urban
sprawl at housing level between1995-2008. Landsat (TM) satellite images for year 1986
and Landsat (ETM+) for the years 2000 and 2005 have been utilized to quantify the
land-use /land-cover changes and the trend of urban growth. From the Land-use/land-
cover maps, the built-up structures growth was 131% between 1986 and 2000, and
89.7% between 2000 and 2005. Annually the growth rate of built-up area for these
study years was 9.38% and 17.94% respectively. The land-use/land-cover change
statistics showed that the annual conversion rate of agricultural land and bare soil to
built-up area between 2000 and 2005 accounts more than 50%. The Getis-Ord Gi*
hotspot analysis toolsets of ArcGIS 10 was used for the analysis of spatial clusters and
expansion map of the area. The validation based on of Ikonos image 2008 revealed
that the result is in good conformity with the model of suitability map. Generally, the
visual comparison of land-cover 2005 and Ikonos image 2008 indicated that all
agricultural land and grass land are under the influence of urban expansion.
v
------------------------------------------------------ ---- ---------------------------------------------------------------------------
Urban sprawl mapping and land-use/land-cover change statistics 2011
==========================================================================================
ACKNOWLEDGMENTS
First of all, I would like to thank the ‘Almighty God’ for giving me the life, patience,
wisdom and who made it possible, to begin and finish this work successfully.
for his constructive suggestions and guidance throughout the period of research work.
My sincere thanks goes to Mr. Bernard Lortic, (IRD- L’Institut de recherche pour le
Geography and Environmental Studies Senior lecturer for his encouragement and
comments on thesis.
Special thanks to Mr. Denis Gerard and Jerome Salvat, Ronan Balac for their
I am highly grateful to GeoEye Foundation for providing me with the Ikonos Satellite
And last but not least, I would like to extend my deepest gratitude to my friends,
Mesfin Tesema, Misrak Alemu, Yayeh Desalgn, Abrham Tekle, and Jemal for their
vi
------------------------------------------------------ ---- ---------------------------------------------------------------------------
Urban sprawl mapping and land-use/land-cover change statistics 2011
==========================================================================================
TBALE OF CONETENSTS
Abstract………………………………………………………………………………………………………………………………………………IV
Acknowledgments……………………………………………………………………………………………………………………………….V
INTRODUCTION
1.3. Objectives…………………………………………………………………………………………………. 3
LITERATURE REVIEW 5
7.1 Conclusions………………………………………………………………………………………………….…99
7.2 Recommendations……………………………………………………………………………….………...100
References ………………………………………………………………………………………………..….………..102
Appendices…………………………………………………………………………………………………….….…...107
ix
------------------------------------------------------ ---- ---------------------------------------------------------------------------
Urban sprawl mapping and land-use/land-cover change statistics 2011
==========================================================================================
LIST OF FIGURES
Figure 4. 1: Procedural Flowchart for urban growth and change map ....................................... 31
Figure 4. 2: Field photograph (right) and the same area with Ikonos-2 image in 3d view
(left)......................................................................................................................................................... 33
Figure 4. 3: Panchromatic 1 meter (A) and False color Multispectral pan-sharpened Ikonos
image (B) ................................................................................................................................................ 38
Figure 4. 4: Roads center lines, blocks and building footprints layers (2008)........................ 39
Figure 4. 5: GPS points, road centerline and building footprints overlaid on Ikonos image
(2008) ....................................................................................................................................................... 40
Figure 4. 7: GPS points, road centerline and building footprints draped over DEM........... 41
Figure 4. 9: removing spots, splash and blemish with WinTopo software. .............................. 43
x
------------------------------------------------------ ---- ---------------------------------------------------------------------------
Urban sprawl mapping and land-use/land-cover change statistics 2011
==========================================================================================
Figure 5. 2: Land-use/land-cover map of the study area (2000) .................................................... 52
Figure 5. 5: Land-use/ land-cover class contribute to net change to built-up area (1986-2005)
................................................................................................................................................................. 58
Figure 5. 9: Annual changes in Housing Units against annual changes in Population .......... 62
Figure 5. 10: population distribution by street blocks (1994 and 2007) ................................... 63
Figure 5. 15: Area and Growth rate of built-up area (1986-2005). ................................................ 68
Figure 5. 19: Building footprint (1995 and 2008), North east part of the town .......................... 74
Figure 5. 20: Building footprint (2008) overlaid on Ikonos image and draped on DEM
(Central part of the town). .................................................................................................................. 75
Figure 5. 21: built-up area expansion to agricultural land (field photograph of 2011) ........... 76
Figure 5. 22: Scattering of houses beyond the Kebele boundary (1995 & 2008) ....................... 76
Figure 5. 23: Building density per km2 by Kebele By block and Kebele(1995- 2008) .............. 78
Figure 5. 24: Building footprint density 1995 by Kebele (in km2) ............................................... 79
Figure 5. 25: Building footprint density 2008 by Kebele (in km2) ............................................... 80
xi
------------------------------------------------------ ---- ---------------------------------------------------------------------------
Urban sprawl mapping and land-use/land-cover change statistics 2011
==========================================================================================
Figure 5. 26: Building footprint density of 1995 by block (in km2) .......................................... 81
Figure 5. 28: Map shows Hot spot and Cold spots of housing units .......................................... 84
LIST OF TABLES
Table 2. 1: Base area type in the same location and density measurement difference. .......... 14
Table 3. 2: Areas of (in million ha) of the major soil classes in Ethiopia. ................................. 24
Table 4. 5 : Overall accuracy and Kappa (K^) statistics for the classifications ......................... 37
Table 4. 6: Producer's and User's accuracy for individual land-use/land-cover classes ......... 37
Table 5. 1: Area statistics and percentage of land-use/land-cover for each year ...................... 50
xii
------------------------------------------------------ ---- ---------------------------------------------------------------------------
Urban sprawl mapping and land-use/land-cover change statistics 2011
==========================================================================================
Table 5. 2: Change in area and percentage between 1986-2000 and 2000-2005 Error! Bookmark
not defined.
Table 5. 6(a): Land-use/ Land Cover changes (km2) from 1986 to 2000 .................................... 55
Table 5. 7 (a): Land-use/ Land Cover changes (km2) from 2000-2005 .......................................... 56
Table 5. 8 (a): Land-use/ Land Cover changes (km2) from 1986 to 2005...................................... 56
Table 5. 11: Population of Debre Birhan town by Kebele (1994 and 2007) ............................. 60
Table 5. 12 : Number of Housing units and annual growth rate by Kebele(1994-2007) ......... 61
Table 5. 13: Built-up area absolute growth and annual rate of growth rate%. ......................... 69
Table 6. 1: Pair wise comparison matrix of five factors to urban growth suitability ........... 96
xiii
------------------------------------------------------ ---- ---------------------------------------------------------------------------
Urban sprawl mapping and land-use/land-cover change statistics 2011
==========================================================================================
LISTS OF ABBREVIATIONS
TM Thematic Mapper
xiv
------------------------------------------------------ ---- ---------------------------------------------------------------------------
Urban sprawl mapping and land-use/land-cover change statistics 2011
==========================================================================================
INTRODUCTION
1.1. Background of the Study
Several studies show that urbanization is increasing in both the developed and developing
high population growth. Studies show that the expansion of cities and towns, and the
countries.
Although both developing countries and developed countries have common urbanization
processes, formerly the problem of urban sprawl was restricted to developed countries. The
reason for urban sprawl for developed countries was mainly a result of higher incomes that
make people preferring to live in the outskirts of the city, with open spaces at reasonable
distances from cities. For developing countries sprawl is largely a result of necessity- people
move to the city in search of better employment and opportunity which leads to an increase
Due to the dynamic urban growth trends, town administrations and planners are faced with
difficulties to supply basic amenities like sanitation, treated water supply, primary health
center, etc., as planners were unable to visualize such growth during planning, policy and
informal houses in and beyond the administrative boundaries of towns and cities. Moreover,
the implications of these profound changes cause a change in landuse patterns. Such highly
Maintaining up-to-date urban land cover information is both costly and time-consuming
using traditional field and air photo methods. Remote sensing technology provides an
efficient and less-expensive way for urban growth mapping (Hu, 2010).
1
------------------------------------------------------ ---- ---------------------------------------------------------------------------
Urban sprawl mapping and land-use/land-cover change statistics 2011
==========================================================================================
Therefore, the focus of this study is to prepare historical Land–use /land-cover maps from
Landsat archive images, analyzing the urban land cover change statistics, extracting GIS
vector data from high resolution image and topographic map, analyzing the residential
house pattern and sprawl, and finally mapping the physical landscape suitability for future
urban expansion.
expansion and residential houses sprawl in many of its towns and cities. Therefore, there is a
need for, mapping, analyzing and continuously monitoring of the phenomena of urban
growth patterns because it is the responsibility of urban administrators, and planners to give
the entire infrastructure and service facilities in a complex urban area. The conventional
method of mapping urban growth is the use of ground survey and aerial photographs
which is costly and slow process to get the necessary information such as the direction
where urban growth occurs and the level of services needed on the new developmental
sites.
Debre Birhan is one of the towns of Ethiopia in which various changes have been occurring
for more than 5 centuries. The town has no regularly updated maps to indicate those urban
growth changes. Although the town doesn't have a detailed updated map, a regularly
updated map with an acceptable resolution can at least give an impression about the
physical changes and growth rate of the urban area of the town. Therefore, Satellite and GIS
technologies can help the urban planners and engineers as an alternative means of rapidly
In this research, it was intended to study the urban growth and the degree of sprawl that
occurs in Debre Birhan town comparing the land use changes occurred for the study years
1986, 2000, 2005 and distribution of residential houses between 1995 and 2008. For this
purpose, first multi-temporal Remote Sensing and GIS data sets were used to map the urban
land use changes, because one of the prerequisite for understanding urban sprawl is
successful land use change detection. To work out the change detection Landsat satellite
imageries of three time period were used. Secondly high resolution satellite image and
2
------------------------------------------------------ ---- ---------------------------------------------------------------------------
Urban sprawl mapping and land-use/land-cover change statistics 2011
==========================================================================================
topographic map were used to quantify and analyze the urban sprawl that might occur in
the town. Finally, the future urban lands that would be used for urban expansion will be
predicted by combining suitability factors and constraints. The suitability model will
predicts the possibility of land use conversion for future urban sprawl.
1.3. Objectives
1.1.1. Main objectives
To investigate the basic urban land-use/ land-cover changes and patterns of urban
sprawl over the study area based on the analysis of high and medium resolution
To study the land requirement for urban expansion using GIS based Multi-Criteria
Evaluation.
Did urban sprawl takes place in the study area based on characteristics of urban
growth and population growth? The population of Debre Birhan town is increasing
roughly by 2 times by the year 2007(65214) from the 1994(38717) census. This
indicated that the study area has undergone a tremendous change in urban growth
Is GIS and Remote Sensing used to investigate the urban land-use/land-cover change
and urban sprawl on the study area? There was no study made on the urban sprawl
and land use changes on the study area, therefore, this study uses GIS and Remote
3
------------------------------------------------------ ---- ---------------------------------------------------------------------------
Urban sprawl mapping and land-use/land-cover change statistics 2011
==========================================================================================
sprawl and land-use/land-cover change studies are important tools for urban or regional
planners and decision makers to consider the impacts that can occur on sustainable urban
development of the study area. The results of this study would provide information relevant
Generally, the study after the completion of the research will have the following
significances:
It will help the concerned bodies understand the sprawl/scatter development of residential
houses and the rate of continuous land-use/ land-cover change in the study area. This may
provide some idea to find appropriate solutions for scattering or non regulated urban
It will provide local engineers, urban planners and policy makers with the necessary
information to depend on
expansion and dynamic population growth which create a burden on basic supply of public
services and infrastructural developments. Given the fact of current and future
unmanageable urban growth problems, there is a need of decision making and planning to
maintain at least some control of urban expansion. This is not possible without up-to-date
information about various aspects in the urban areas. Remote sensing and GIS techniques
can help to get up-to-date information of the urban expansion with frequent coverage and
low cost. Therefore, this research will investigate the rate of land use/ land-cover change that
occur in the study area over those 19 years (1986 to 2005). Similarly, it was intended to
investigate the level of urban sprawl in the study area between 1995 and 2005. The study
also includes urban expansion suitability map to identify the most probably growth area of
LITERATURE REVIEW
currently, the rapid urban growth and the associated urban land cover changes caused by
rapid population growth are a characteristic of cities and towns of developing countries.
According to the United Nations, it is predicted that by 2030, the number of city dwellers
will reach 60% of the world’s population. This urban population growth will be most
significant in low income countries, notably in Africa and Asia. Among African countries,
East Africa will experience urban population growth rates significantly higher than the
African average. In descending order, Addis Ababa, Nairobi, Dares Salaam, Antananarivo,
Kampala and Mogadishu will remain the region’s largest cities in the foreseeable future.
Comparing to the world cities, Dare Salaam, Kampala, Nairobi, Antananarivo and Addis
Ababa already rank among the 31 fastest growing cities. (UN-HABITAT, 2008)
UN-HABITAT (2010) stated that the uncontrolled physical expansion of cities and towns in
developing country create an impact on urban environment and economy. This unplanned
and uncontrolled development makes provision of housing, roads, water supply, sewers
and other public services too expensive. The other impact mentioned on the report was the
loss of agricultural land as most cities and towns built on productive agricultural land.
demographic changes such as, rapid natural population growth and rural-urban migration.
This type of rapid urban growth make specially poor African countries incapable of
This rapid urban growth rate is attributed to population growth in urban residents and rural
urban migration. The high population growth which does not match with the socio-
economic growth and infrastructure service requirement creates high demand for housing
which leads to unplanned and uncontrolled horizontal expansion and contributes to urban
land use changes. The extent of urbanization or the sprawl is one such phenomenon that
----------------------------------------------------------- 5 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
drives the change in urban land use patterns. These changes caused by anthropogenic
Remote sensing and Geographic Information Science (GIS) technologies have been utilized
to detect and quantify changes in the landscape and the consequential environmental
impacts. In modern times, several studies have utilized remote sensing data to examine
urban land changes of different patterns of urban expansion and development (Longley et al
2005).
urban sprawl is an expansion of urban area to agricultural land caused by high population
growth and rural urban migration. This spread of development across the landscape is
Urban sprawl is an extent of urbanization mainly caused by population growth and large
scale migration (Sudhira, et al 2004). But urban sprawl is not to be considered as increase of
urban lands in a given area. Urban sprawl is a scatter development which is mainly
unplanned and uncontrolled. Glaster, et al (2007), describes urban sprawl as one of the
dimensions of land use patterns such as low density, continuity, concentration, nuclearity,
diversity, proximity and centrality, decentralization from of urban core to the periphery.
Urban growth takes place either in radial direction around the core of the city/town or
linearly along the highways. The form of development along highways, or surrounding the
city and in rural countryside is often referred as urban sprawl (Theobald, 2001).
The spatial development of urban sprawl can be divided in to three basic forms (Barnes et
al., 2001)
Low density sprawl: This type of sprawl is characteristic by exploitative use of land
for housing purpose along the margins of existing urban areas. This type of
Ribbon sprawl: The development of this type of sprawl is following major road
corridors outward from the urban centers. Lands adjacent to the major roads and
----------------------------------------------------------- 6 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
highway are developed but lands with some distant from the roads and highway
remain undeveloped.
expensive to give services such as water, power and roads for this form of
development.
From an economist point of view, as population raises in an urban area it becomes more
difficult to locate the same percentage of residential houses and business activity if the urban
boundaries remain fixed. In addition, higher income residents generally demand larger
quantities of housing and the inexpensive land to build it on is more likely on the fringe of
Another cause of urban sprawl to occur as stated by UN-Habitat (2010) report was because
authorities pay little attention to slums, land, services and transport. Moreover, they lack the
ability to predict urban growth and, as a result, fail to provide land for the urbanizing poor.
In addition, the urban poor are denied land rights which are one of the main factors driving
people to the periphery of towns, associated with urban sprawl in developing countries.
----------------------------------------------------------- 7 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
In developing country the causes of urbanization are twofold: people migrate from rural
areas to cities for employment opportunities, and fertility rates of the people already living
There is a critical key difference between urban sprawl development in developed and
developing countries. In developed countries, people do not want to live in the cities and
they choose to move out. However, in developing countries, people move out because there
is not enough space for them to live in the city (Haregewoin, 2005). The movement of the
people from the city center to the sub urban part due to lack of space results spread of
problems such as the wastage of agricultural land become a burden for service provision.
Currently, the high population growth in cities is seen to create pressure and competition for
land related resources, which has also caused upward pressure on land values, particularly
in sub-urban areas. Clearly, not enough attention is paid to developing solutions to the
expansion of cities in sub-urban areas and the challenge this is causing in regard to the
In many African cities the urban land market is skewed against the poor. The rising costs of
urban land and other factors prevent poor people and other vulnerable and disadvantaged
groups from gaining access to suitable land, which does not pose economic, environmental
Another cause for urban sprawl is lack of clear regulation mechanism of the new
development in sub urban area. The majority of local land use regulation is accomplished
with zoning. The goal of zoning land is to regulate the use and intensity of development
such as housing type and density. Although technical in their implementation, zoning
regulations are often a community’s expression of the unwritten rules of social order
(Harris, 2002).
regulations can restrict the supply of land available for development operates to increase
land prices. This can is useful to maintain environmental quality open space or agricultural
However, absence of strict zoning regulations leads to rich men to buy a land on sub-urban
area and to build houses for rental purposes. This provides cheap accommodation for other
----------------------------------------------------------- 8 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
low-income resident or migrants who lack the means of constructing their own house.
areas
Diminished capacity to attract and retain business in the state due to deterioration in
quality of life
The cost of building and maintaining new suburban infrastructure become high
The UN-Habitat report says, urban sprawl has a negative impact on infrastructure and the
sustainability of cities. In most cases, sprawl translates to an increase in the cost of transport,
metropolitan areas require more energy, metal, concrete and asphalt than do compact cities
because homes, offices and utilities are set farther apart (UN-Habitat, 2010).
In recent years the fiscal problems of many municipal governments have rapidly increased.
The large investments in road, water and sewerage and other public services made in the
last decade in many countries often comes from encouragement and financial assistance of
the major international development donor agencies. The external aid helped to provide the
initial capitalization for these extensive projects, but left the problem of their maintenance
and repair to local governments. As time has passed, the costs of maintenance have
increased, forcing the government units responsible to search for new sources of revenue
(Dobele, 1984).
----------------------------------------------------------- 9 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
about an object without coming into physical contact. Schowngerdit (2007) defined Remote
sensing as the measurement of object properties on the earth’s surface using data acquired
Although the earth is the most frequent target in remote sensing study, the science of remote
sensing in its broadest sense includes aerial, satellite, and spacecraft observations of the
surfaces and atmospheres of the planets in our solar system (Smith, 2006).
The modern era of earth remote sensing from satellites began when the Landsat
Multispectral Scanner System (MSS) provided, for the first time, in 1972. Currently, the
available remotely sensed images, ranges from high resolution panchromatic images
(10m) and the 15-m low-resolution panchromatic images by Landsat 7. These satellites have
become one of the most important sources of spatial data used in the GIS community.
cover map. The basic ideas are that the spectral characteristic in a multi-band image can
ways, depending on the needs. The most widely used satellite image for land-cover/land
economic impacts of greater damage than benefit to the area (Mohsen, 1999). Therefore data
planners in monitoring the consequences of changes that could occur on a given place.
Although urban monitoring and modeling applications usually rely on extensive historic
tends to focus on high resolution satellite imagery. High resolution images such as Ikonos
include a large variety of regions occurring in complex configurations that are challenging to
----------------------------------------------------------- 10 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
definitions recognize the use of spatial data. Many alternative definitions of GIS have been
suggested, but a simple definition is that a GIS is a computer-based system for the capture,
"A geographic information system (GIS) is a computer-based tool for mapping and
analyzing things that exist and events that happen on earth. GIS technology integrates
common database operations such as query and statistical analysis with the unique
What differentiated GIS from other spatially related systems is by its analytical capacity and
geology, geography, etc. GIS works with many types of data. The spatial data in GIS
databases are predominately generated from remote sensing through the direct import of
images and classified images, but also through the generation of conventional maps (e.g.
topographic maps) using photogrammetry. Today, remote sensing is an integral part of GIS.
Remote sensing data, such as satellite images and aerial photos allow us to map the
variation in terrain properties, such as vegetation, water, and geology, both in space and
time. Satellite images give a synoptic overview and provide very useful environmental
information for a wide range of scales, from entire continents to details of a meter (Longley,
et al., 2005).
urban sprawl research. Progress in modern Remote Sensing and Geographic Information
Systems allow us to collect a lot of physical data with relatively low cost and high speed and
on repetitive basis than the traditional ground surveying method. The remote sensing data
together with GIS helps us to analyze the data spatially and very important to make
----------------------------------------------------------- 11 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
Therefore, it is ideal to use a tool such as a GIS as part of research on urban sprawl because
of its capacity to handle many different types of spatial data. Several spatial and non spatial
techniques of estimating the urban sprawl level exist. However, all techniques used are
Galster and Cutsinger (2007) examined six different measures of urban sprawl development:
4 Centrality:- the degree to which development is located close to the central business
district;
of development;
6 Proximity of land uses: - the degree to which different land uses is close to one
another.
The above mentioned measure of urban sprawl is summarized by Ewing, et al., 2002 in to
four factor of sprawl Index which each of these factors is in turn composed of several
1 Residential density;
sprawl. However, there is a debate which variable is best to use in representing density, the
density level at which a city can be regarded as sprawling, the scale at which density should
be measured and the extent of the space over which density should be characterized (Alberti
----------------------------------------------------------- 12 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
Concerning the scale to be used to calculate density, different form of urban density
measurement is available. The some type of density that is commonly used includes:
1 The gross density: The calculation is based on the total area of to the city/town
understudy.
2 Net density: a density figure for a given area of land that excludes land not directly
3 Floor-to-area ratio: The Floor Area Ratio or Floor Space Index is the ratio of the total
floor area of buildings on a certain location to the size of the land of that location.
Density varies greatly depending on the base land area used in the density calculation. The
parcel or site density is almost always higher than the neighborhood density, because at a
neighborhood scale much land is included in the base land area calculation that does not
The gross density is calculated by considering the total area of the city or a town including
open spaces and sometimes non habitable area. The argument behind this calculation is the
different use of the base area calculation where more and more nonresidential uses added
into the calculation considered. The difference between these numbers is that as the base
land area being considered increases there are more and more nonresidential uses added
----------------------------------------------------------- 13 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
Table 2. 1: Base area type in the same location and density measurement difference.
Dus=Dwelling Units
Density type Measurement
Site density 10 DUs per acre
Block density 8 DUs per acre
Net residential density 10 DUs per acre
Net neighborhood density 6 DUs per acre
Gross neighborhood density 5 DUs per acre
City density 4 DUs per acre
Metropolitan density 3 DUs per acre
(Adopted from Forsyth, 2003)
Density, measures overall activity intensity in urban area and is the most commonly used
although density by itself, does not address the pattern of activity distribution (Tsai, 2005).
that advances these technologies have a strong influence on spatial statistics. Classic spatial
The classic spatial autocorrelation is related to Tobler's first law of geography‛. Waldo
1970. Tobler's first law of geography states: ‚everything is related to everything else, but
near things are more related than distant things‛. In recent years it has become even popular
and important in geography and GIS science which is the basis of spatial auto-corelaltion
between entities, and "auto" refers to the fact that a single variable is being related to itself
(Griffith, 1984).
The importance of housing to both the ecological and social sciences has focused increasing
attention on the spatial patterns of housing growth. Furthermore, because housing growth is
not a spatially random process, understanding its patterns requires a spatially explicit view
----------------------------------------------------------- 14 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
In spatial data analysis, especially in order to get additional hints for the existence of urban
Currently, new measures of local and global spatial statistics have been developed.
The Global Moran I Spatial autocorrelation is used to identify the overall patterns or trends
dispersion. Global Moran’s I tool in desktop ArcGIS software can be used to calculate the
values cluster together, and/or if low values cluster together. Similarly, Global Moran I
statistics the High/Low Clustering (Getis-Ord General G) tool assess the overall pattern and
trend of spatial data. Global spatial statistics, including the High/Low Clustering tool, are
most effective when the spatial processes being measured are consistent across the study
The Local Moran I pattern analysis and Local statistics tools (like Hot Spot Analysis) assess
each feature within the context of neighboring feature. Local analysis is based on the Local
Moran statistic visualized in the form of significance and cluster maps (Anselin 1995).
Spatial autocorrelation index such as Moran’s I and Geary’s C requires measuring a spatial
weights matrix that defines a local neighborhood around each geographic unit. The value at
each unit is compared with the weighted average of the values of its neighbors. A weights
file identifies the neighbors. Weights can be constructed based on spatial relationship of
objects such as contiguity to the polygon boundary (shape) files, or calculated from the
Inverse distance: The impact of one feature on another decrease with distance
Fixed band distance: Everything with specified critical distance is included in the
Polygon contiguity: The neighbor of each feature is only those neighbors which share
a boundary.
----------------------------------------------------------- 15 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
Hot Spot Analysis (Getis-Ord Gi*) (Spatial Statistics) identifies statistically significant spatial
clusters of high values (hot spots) and low values (cold spots). It creates a new output
feature class with a z-score and p-value for each feature in the input feature class. At the
same time, it returns the z-score and p-value field names as derived output values for
potential use in custom models and scripts. The z-scores and p-values are measures of
statistical significance which indicates whether or not to reject the null hypothesis, feature by
feature. In effect, they indicate whether the observed spatial clustering of high or low values
is more pronounced than one would expect in a random distribution of those same values. A
high z-score and small p-value for a feature indicates a spatial clustering of high values. A
low negative z-score and small p-value indicates a spatial clustering of low values. The
higher (or lower) the z-score, the more intense the clustering and a z-score near zero
has significant success in monitoring and managing urban land-use/land cover changes and
to measure the degree of urban sprawl. A number of urban sprawl- related studies were
conducted using GIS and Remote Sensing. For example, Yeh and Li (2001) used remotely
sensed data to measure and monitor the degree of urban sprawl for cities and towns in
China. Sudhira et al. (2003) integrated IRS 1C and LISS satellite imagery product with
Mangalore and Udupi region in India. Gu(2006) used Spatial Analysis and Spatial Statistics
There are many examples of related woks that can be mentioned. However, it is important
to understand that the accuracy of the techniques used to measure urban sprawl largely
depend on the input data used and the scale. In this thesis, housing density and GIS based
----------------------------------------------------------- 16 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
3.1. Location
Debre Birhan town is the capital of North Shoa Zone and is located 130 km North East of
Addis Ababa city. The town was established during the regime of Atse Zara Yakob in 1454
The study area covers 181 km2 (Figure 3.1) and lies between UTM Zone 37, 552083, 563954
Easting and 1064175, 1075864 Northing and has an average elevation of 2750m meters above
sea level.
----------------------------------------------------------- 17 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
3.2. Topography
Based on the local topographic characteristic (the difference between highest and lowest
elevation) of the study area, the major landform elements were mapped using SRTM 30m
DEM. The major land form components of the study area include plateau, valley, and some
3.2.1. Plateau
The study area lies on an extensive upland region at high elevation ranging from 2050m and
2850m a.s.l. The plateau area (Figure 3.2) is surrounded by escarpments, undulating ridge
chains and deeply dissected valleys. Studies show that the flat lying topography (plateau)
is mostly represented by the Tertiary basalts and Quaternary sediments and some irregular
small hills formed by trachyte and erosion remnant of ignimbrites (Daniel,et al., 2010).
----------------------------------------------------------- 18 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
3.2.2. Valleys/gorges
Most of the northern and north western part of the study area is characterized by valley
depression (Figure 3.3) which is not far from the town. The part of the Jama canyon is one of
----------------------------------------------------------- 19 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
3.2.3. Drainage
There are rivers and streams that cross Debre Birhan town. These streams and rivers start
from the eastern part of the plateau and drained from southeast to northwest -towards Jema
River. Among the rivers that cross Debre Birhan town, Bersinaa is a perennial river and
Dalecha is an intermittent river. The drainage pattern of the study area has a form of
----------------------------------------------------------- 20 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
3.3. Climate
September, and the short rainy season which usually falls between January and April. The
short rainy season however, is highly unreliable being characterized by variability, and
absence of rains. The monthly rainfall for 10 years is presented in Figure 3.5. The maximum
and minimum monthly average rainfall varies from year to year. The highest monthly
rainfall was registered in 2006. In 2008, the mean annual rainfall was 946mm. The analysis of
rainfall data for 10 years (Appendix 2) revealed that there was high annual average rainfall
----------------------------------------------------------- 21 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
3.3.2. Temperature
Debre Birhan's 10 year average temperatures graph shows that there was a decrease in 2000,
2005, and 2007. The lowest average temperature was registered during November 1999 and
1993. Speaking at Durban's International Conference Center, Jam Gibson said the
relationship of the town with Blanc Mesnil has resulted in significant investments in the
town from the Blanc Mesnil municipality and other French agencies. One of the investments
that have been realized was the upgrade of the town’s water and sewerage systems. This
makes residents of Debre Birhan town to get a reliable water service 24 hours per day
(Gibson, 2010).
----------------------------------------------------------- 22 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
Water Pump
Well Field Capacity Aquifer Capacity
3.5. Soil
Currently, there is no large scale digital soil map for the study area. For general information,
the soil map description was taken from FAO Digitized Soil and Terrain Database of East
Africa (Burundi, Djibouti, Egypt, Eritrea, Ethiopia, Kenya, Somalia, Sudan, Rwanda, and
Uganda) at as scale 1:1 Million. The ARC/INFO version (7.3) of the SEA (Soil and Terrain
Database of East Africa) files has a summary info of dominant soil type, dominant texture
----------------------------------------------------------- 23 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
The two dominant soil classes in the study area are Leptosols(Lp) and Vertisols(CM)(
Appendix 2:). Studies show that vertisols develop mainly on the gentle slopes-usually less
than 3%-of terraces, plains and valley floors in association with vertic inceptisols. Vertisols
are important to Ethiopian agriculture. They are dark-colored clays and account for 24% of
all highland soils. Vertisols develop cracks when expanding and contracting with changes in
moisture content. The physical properties of Vertisols and their moisture regime create
production constraints. Workability of these soils is also hampered by their stickiness when
wet and hardness when dry, and water-logging and erosion greatly affect crop production
(Bull 1987). Vertisols occupy more than 10% of the soils in all administrative zones of the
country. The Central Zone of the country (Shawa), comprises 35% vertisols among other soil
Table 3. 2: Areas of (in million ha) of the major soil classes in Ethiopia.
NW North West(Gojam)
E East (Harerge
S South (Sidamo,Gamogofa
----------------------------------------------------------- 24 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
3.6. Geology
The geology of the study area can be generalized from geological study of Debere Birhan
area which is associated to the Mesozoic and Cenozoic Era. The two lithographic units
The Mesozoic Sediments exposed in highly dissected plateau area, within the gorge
and canyon of Jama drainage basin which comprises stratified mud-stone and cross
bedded sandstone,
The Cenozoic volcanic rocks of Tertiary and quaternary age which covers the largest
part and exposed on the plateau and the NE-SW rift margin. These volcanic rocks
and Tarmaber-Megezez basalt. According to the report, the most widely expanding
geological rock in the study area is the Tertiary Debre Birhan Ignimbrite (Tdig)
Debre Birhan town is completely, within the ignimbrite lithology which is locally called Sela
Dingay Debre Birhan. The eastern part of the town is within quaternary eluvial lithology.
Kessem Basalt is found on the low land part of the study area. The Tarmaber Megeze Basalt
is mostly located on the Eastern, Western and North western part of the study area (Figure
3.8).
----------------------------------------------------------- 25 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
Table 3. 3: Classification of lithologies and stratigraphy of the map sheet (Debre Birhan Map sheet).
units
Deposits
Eluvium (Qel)
Trachyte (Tt)
----------------------------------------------------------- 26 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
----------------------------------------------------------- 27 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
----------------------------------------------------------- 28 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
acquired from one of the best source of free satellite imagery on the web: Global Land Cover
historical land-use/land-cover maps of the study area. Source and acquisition date of this
Shawa District, Central Ethiopia was donated by the GeoEye Foundation as part of an
----------------------------------------------------------- 29 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
Table 4. 2: Ikonos satellite data description
information was calculated using the old town boundary and the population data of each
Kebele.
Data
Source
Topographic map(1995) Debre Birhan town Municipality
A 30m SRTM Digital elevation model (DEM) was used for the derivation of slope and
stream features across the study area. Digital Elevation Model was also used for the
----------------------------------------------------------- 30 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
change detection. The amount of land-use/land-cover change over the study years and the
land-use/land-cover category loss and gain were investigated by mapping the land-
use/land-cover of the study area for 1986, 2000 and 2005 with Landsat image archives.
From Landsat images, seven land use categories were identified including built-up,
agricultural land, grass and pasture, vegetation, shrub land, bare land and water (lake). The
class name agricultural land includes, crop land, fallow land, etc. After post classification
processing has been done for each land-cover/land-use maps, accuracy assessment has been
performed. Then, the land-cover/ land-use change statistics was performed and the rate of
----------------------------------------------------------- 31 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
4.2.1.1. Pre Processing of Landsat images
The digital image processing is largely concerned with four basic operations: image
1999). The image restoration is concerned with the correction and calibration of images in
The challenges of urban mapping using Landsat imagery include spectral mixing of diverse
land cover components within pixels and spectral confusion with other land cover features
(Lu and Weng 2006). To get a better enhanced image, different image enhancement
techniques are available with different digital image processing software. One of the image
enhancement works done in landsat image in this work was the spectral enhancement
with the help of different sources such as high resolution images (Ikonos), Google Earth and
topographic map.
A high resolution Ikonos image was used to access the low resolution Landsat image.
Google Earth provides an access of high resolution image on the internet. The study
area has a high resolution Google Earth image, dated 2007. Clearly recognized
objects on Google earth image were digitized and taken as AOI (Area of Interest) and
GPS points and digital camera were used to locate some sample ground cover
situation at the time of filed visit. Some land-use/land-covers in urban area, like
water body, woodlots, and built-up area stay for long time without change to other
----------------------------------------------------------- 32 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
Figure 4. 2: Field photograph (right) and the same area with Ikonos-2 image in 3d view (left)
intended to be mapped for the study area. These classes include: Built-up, agricultural,
vegetation, grass and pasture, shrub land, bare land and water body.
Vegetation Economic forests such as timber and fuel woodland other natural forests.
----------------------------------------------------------- 33 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
Built-up
Urban or Built-up land is comprised of areas of intensive use with much of the land covered
by structures. Included in this category are all manmade features such as houses, roads,
Agricultural land
Agricultural Land for the purpose of this study defined broadly as land used primarily for
production of food. On high spatial resolution imagery, agricultural activity can be seen
with distinctive geometric field and road patterns on the landscape. However, on Landsat
From Landsat imagery alone, it generally is not possible to make a distinction between
Cropland and Pasture with a high degree of accuracy. However, grass land and pasture had
different pixel reflectance property on Landsat imagery used for this work. High resolution
image was referred for the spatial location of grass and pasture. Therefore, grass and pasture
in this work includes open field with grass and land used for grazing purpose.
Shrub land
Shrub land in this study defined as land where the potential natural vegetation is
Bare Land
Vegetation
In this study the vegetation includes planting of trees around compounds, eucalyptus wood
Waterbody
The water body in this thesis was the water dam which is clearly seen on landsat images.
----------------------------------------------------------- 34 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
4.2.1.4. Post Classification
Classified data often manifest a salt-and-pepper appearance due to the inherent spectral
and Kiefer, 1994). It is often desirable to ‚smooth‛ the classified output to show only the
dominant classification.
One means of classification smoothing involves the application of a majority filter. In such
operations a moving windows is pass through the classified pixel in the window is not the
majority class, its identity is changed to the majority class. If there is no majority class in the
window, the identity of the center pixel is not changed. As the windows progresses through
the data set, the original class code are continually used, not the labels as modified from the
Majority filters can also incorporate some form of class and/or spatial weighting function.
Data may also be smoothed more than once. Certain algorithms can preserve the boundaries
between land cover regions and also involve a user-specified minimum area for any given
land cover type that will be maintained in the smooth output (Lillesand and Kiefer, 1994).
The Spatial Analyst extension in ArcGIS software was used for the post-classification
processing task. Three tasks have been employed to smooth the classified output from
Landsat images.
Filtering the classified output. By using Majority filter tool, the isolated pixels were
Smoothing class boundaries and clumping classified output. These steps smooth the
ragged class boundaries and clump the classes. The Boundary Clean tool is used.
reclassifies small isolated regions of pixels to the nearest classes. The Region group,
----------------------------------------------------------- 35 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
4.2.1.5. Accuracy Assessment
In order to determine classification accuracy, it is necessary to determine if the output map
meets, exceeds, or does not meet certain predetermined classification accuracy criteria. One
of the most common and typical method used to assess classification accuracy is the use of
an error matrix (sometimes called a confusion matrix or contingency table (Lillesand and
Kiefer, 1994)). Currently, accuracy assessment is considered as an integral part of any image
may classify pixels or group of pixels to wrong classes. The most obvious types of error that
has been done to make sure the classification is accurate enough for use. To assess
classification accuracy the classified map derived from remotely sensed data and the
topographic map, and ground truth data were used as a means of time and cost-efficient
error checking.
To do this a set of randomly selected pixels whose true values well known were used and
ERDAS Accuracy Assessment tool was used to conduct an accuracy assessment. ERDAS
IMAGINE uses a square window to select the reference pixels. Among the three different
types of distribution parameters namely random (no rules will be used), stratified (the
number of points will be stratified based on the extent and area of each thematic class), and
equalized random (each class will have an equal number of random points regardless of its
extent), simple random was used to evaluate the accuracy assessment of the land-
cover/land-cover classes.
The study assessed accuracy of the classification results with 100 randomly sampled
reference points for each classified image. The error matrix for each classification was
created and then overall accuracy and Kappa coefficient were calculated and evaluated
(Table 4.5).
----------------------------------------------------------- 36 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
Table 4. 5 : Overall accuracy and Kappa (K^) statistics for the classifications
Waterbody - - - - - -
The results show that the achieved overall classification accuracies were 85.00 %, 88.75% and
87.65% and overall Kappa (κ^) statistics were 0.7903, 0.8297 and 0.7968 respectively for the
classification of 1986, 2000 and 2005 Landsat images(Table 4.4). The producer's and User's
For individual classes, the producer's accuracy ranges from 66.7% -94.6 %( 1986), 77.8% -
91.8 %( 2000) and 60% - 97.8% (2005) whereas user's accuracy were 71.4% - 92 %( 1986), 80-91
%( 2000) and 71%-92 %( 2005). Higher producer's and user's accuracies were achieved
----------------------------------------------------------- 37 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
4.2.2. Automatic feature extraction and digitizing
Most of GIS applications to urban planning and modeling have relied upon vector GIS and
polygon data. Vector GIS includes points, lines and areas with attached various attributes.
Features in a vector GIS have a very high graphic resolution and they are often used to make
informative and visually appealing maps. Therefore, vector GIS is the main part of this
work. To find the existence of residential and commercial housing sprawl, buildings
footprints and roads were digitized from toposheet of Debre Birhan town (1995) and Ikonos
image 2008. These data were used to show the growth of urban area and the pattern of
development relatively at a better scale map than data from Landsat image. The urban
sprawl metrics were computed based on the digitized vector files on Ikonos image and
topographic map.
A. Pre-processing
The availability of a 1m panchromatic band, in conjunction with 4m multispectral bands,
images. The spatial enhancement tools of ERDAS Imagine software used to merge these
Figure 4. 3: Panchromatic 1 meter (A) and False color Multispectral pan-sharpened Ikonos image (B)
----------------------------------------------------------- 38 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
B. Manual digitize of building footprints and road features.
Geospatial features and buildings in particular are required for varieties of applications such
as urban planning, creation and updating of Geographic Information Systems (GIS)
databases and creation of urban city models. To get this data, numerous efforts have been
made in the past few years to automate building extraction from digital images (Gulch,
2000).
For mapping from high resolution space imagery such as Ikonos, automatic object-based
image analysis using GIS has been generally used for remote sensing applications in recent
years. However, in practice, the extraction of buildings from digital images is complex
because buildings can have various complex forms and roofs of various compositional
materials.
Therefore, a high resolution Ikonos image was used to digitize building footprints and road
centerlines with the Savedit module of SAVGIS software. The GIS data layers include road
layer, street blocks and building footprints (Figure 4.4. Street Blocks were generated from
streets using the data analysis and mapping module the Savene 9.05 of SavGIS software.
2008
Figure 4. 4: Roads center lines, blocks and building footprints layers (2008)
----------------------------------------------------------- 39 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
To evaluate the quality of spatial information digitized from the image, it was necessary to
measure the accuracy of the data with information from an independent source. Therefore,
the reference data that was used include GPS points (Figure 4.5), topographic map and
visual assessment of digitized data layers on Google Earth image (Figure 4.6).
Figure 4. 5: GPS points, road centerline and building footprints overlaid on Ikonos image (2008)
----------------------------------------------------------- 40 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
Figure 4. 7: GPS points, road centerline and building footprints draped over DEM
----------------------------------------------------------- 41 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
4.2.2.2. Automatic feature extraction from Topographic Maps
Deriving GIS data from historical map will provide the ability to compare historical
geographic pattern to more recent layers (Junk, 2000). One of the input data sources for
Geographic Information System is the paper-based maps. The information from paper based
maps has been digitized to be used in GIS. But manual tracing of paper maps is costly and
tedious process, therefore, an effective way for conversion of paper maps to vector data was
needed.
Currently, raster to vector conversion software are available. ESRI offers a wide range of
GIS software with a high emphasis on vector applications. ArcScan is an ArcGIS extension
used to extract vectors from scanned images. Another tool which was used for vectorization
In this work, both WinTopo pro and ArcScan tools were used to extract vectors from the
----------------------------------------------------------- 42 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
A. Pre- Processing of raster image
The first process for raster vectorization is preparing the raster for vectorization. After the
scanned image has been digitized (converted to digital image), it was converted to gray level
scale to work on it. All the processes filtering, thinning, pruning and Gap filling were
Pruning removes shorter lines that prude out of the longer line while thinning helps to
smoothing the lines. The gap filling process is one of the most important task that used to
reconnect or recover the missing part of lines that are disconnected by unwanted cells that
were removed by pruning information layers like symbols, grids and other non outline
Thinning is another most important process that can be used to reduce a shape to a simpler
B. Raster Vectorization
The raster painting and cleaning toolbar of ArcScan were used to remove unwanted cells
from the raster cells. Finally, the vectorization of the raster accomplished with user defined
sitings. These settings affect the geometry of the feature that can be produced unless the
vectorization settings are carefully done. The polygon outline features of the topographic
image produced by semi-automated process of ArcScan were presented with figure 4.10.
----------------------------------------------------------- 43 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
direction of land-use/ land-cover change on a given urban environment. Spatial data used in
GIS analysis come from a wide variety of sources, in a number of different formats,
resolutions, scales, and projections. One of the data source for GIS is Digital Elevation.
Digital elevation models are essential for a number of topographic and hydrological analysis
including slope computation, stream generation and watershed delineation etc. This study
will attempt to combine various forms of Digital elevation data to study urban spatial
pattern and growth. The most important physiographic data that were extracted from DEM
----------------------------------------------------------- 44 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
4.3. METHOD OF DATA ANALYSIS
use/land-cover change on the study area. Absolute change was calculated by subtracting the
area of land-use/land-cover of the first from the area of later land-use/land-cover image
(Area at Time 2)-(Area at Time 1) to get an aggregate of area change of each class.
percentage ((% area at Time 2) – (% area at Time 1)). Thematic change analysis of the
individual land-use/land-cover was performed using ENVI 4.7 software. The relative change
detection matrix result shows results from one class change to another.
growth rate of population. A simple linear growth rate formula was used to calculate the
annual rate of population growth for each Kebele which was used to compare the growth
The density of the population at Kebele level and street block level was calculated in order
to understand the population distribution within the administrative units. I assumed this
data would help me to identify the location of population concentration and the
concentration of housing units using both housing units derived from satellite data and
topographic map.
The most common and useful method to calculate population growth rate is percentage
growth rate. Percentage growth rate or growth rate is useful as an indicator to look at how
much a population is growing or declining in a particular area. The rate of change (growth
rate) of population of Debre Birhan town for the census years of 1994 and 2007 will be
----------------------------------------------------------- 45 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
<<<<<<<<<<<<<<<<<<. (1)
beginning of period).
and 2005. The rate of urban land cover change has been calculated from the Landsat images.
Data extracted from topographic map of 1995 and on screen digitized of housing units on
Ikonos image of 2008 were used to study the urban growth at housing level. By comparing
dwelling units from these two sources, the level of housing scatter (sprawl) has been
identified.
level over the study years (1995 and 2008). The pattern and distribution of housing units in
and towns. Density, measures overall activity intensity in urban area and is the most
compactness/sprawl.
In this thesis housing density using Kebele and street blocks as base land area was
calculated. The gross density (considering all area of the urban administration boundary)
was compared to the net density (considering all area bounded by street bock and excluding
----------------------------------------------------------- 46 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
roads, agricultural land and all non inhabitable area) of housing density. The result would
help to understand the location of low density housing development and sprawl.
local neighborhood around each geographic unit. The value at each unit is compared with
the weighted average of the values of its neighbors. A weights file identifies the neighbors.
Weights can be constructed based on contiguity to the polygon boundary (shape) files, or
calculated from the distance between points such as points in a point shape file or centroids
of polygons(Anselin, 2003)
Before the analysis of spatial clustering the polygon centroid of each the building footprint
units was converted to polygon centroid to get the mean center of each polygon that
represent housing units. That was because point data were straightforward to measure
spatial distribution of data. All individual areal features of polygons were aggregated to its
----------------------------------------------------------- 47 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
==========================================================================================
own centroid to get the point data. The centroid of each building units were generated from
the digitized buildings using ArcGIS software. Therefore, the polygon centroid represents
point feature (shapefiles) that contains the number of dwelling units per point (Figure 4.11).
The Analyzing Patterns tool set in ArcGIS desktop contains methods that are most
appropriate for understanding broad spatial patterns and trends of residential housing
units. Spatial statistics tools of ArcGIS such as Mapping Clusters help to identify where
spatial clustering occurs. Some standard global and new local spatial statistics, including the
Moran I, Getis-Ord G were used to detect the sprawl pattern on the study area.
To implement this analysis, a spatial weights matrix, should be computed. Spatial weights matrix
analysis to evaluate the suitability of the study area for urban expansion by examining and
weighting different types of factors and constraints. The factors that influence the suitability
proximity to the main road and housing density. Constraints that restrict further expansion
of urban area include: topography, water bodies (open water-body and rivers), and
woodlots.
----------------------------------------------------------- 48 ---------------------------------------------------------------------------
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
land -use/ land-cover use change detection approaches are grouped in to two (Singh 1989).
These are:
In the post-classification approach, images belonging to different dates are classified and
After post classification, the absolute change and relative change were calculated using
Envi 4.7 software. Absolute change is calculated by subtracting the area of land-use/land-
cover of the first from the area of later land-use/land-cover image (Area at Time 2- Area at
Time 1). This method was used to get an aggregate of area change of each class.
percentage (% area at Time 2 - % area at Time 1). The relative change method allows
supervised image classification techniques on each image of 1986, 2000 and 2005. After
performing an image classification, scattered individual pixels were filtered and cleaned
up for better visual display. Then the subtractive change detection based on their area
----------------------------------------------------------- 49 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
extent was performed on the classified images. The summery of area extent of individual
1986 2000
2005
Area(%)
Class Name Area(km2)) Area (%) Area(km2) Area(km2) Area (%)
Built-up 1.05 0.60 2.43 1.40 4.61 2.55
Agricultural land 93.40 51.60 96.82 54.00 110.97 61.31
Grass and pasture 20.93 11.56 15.96 8.82 14.78 8.17
Vegetation 31.06 17.16 14.66 8.10 14.53 8.03
Waterbody 0.00 0.00 0.04 0.02 0.11 0.06
Shrub land 16.60 9.17 25.00 13.81 23.49 12.98
Bare land 17.90 9.89 26.12 14.43 12.55 6.93
Class Total 181 99.99 181 100.00 181 100
The built-up area (Table 5.1) in 1986 occupied the least area with 0.6% of the total class.
Agricultural land occupied more than 90 % of the total area in each of the study year. This
may be due to the conversion of grass land and bare soil to agricultural land. But the most
probable reason might be the time of the year in which the area was imaged which could
be a major contributing factor of the high percentage of the observed classification. The
water body class (a dam) exists only for Landsat image scene of 2000 and 2005.
----------------------------------------------------------- 50 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
1986
Land-use/Land-cover Category Area(km2) Area (%)
Built-up 1.05 0.60
Agricultural 93.40 51.60
Grass and Pasture 20.93 11.56
Vegetation 31.06 17.16
Water body 0.00 0.00
Shrub land 16.60 9.17
Bare land 17.90 9.89
Total 181 100
----------------------------------------------------------- 51 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
2000
Land-use/Land-cover Category Area(km2) Area (%)
Built-up 2.43 1.34
Agricultural 96.82 53.49
Grass and Pasture 15.96 9.00
Vegetation 14.66 13.90
Water body 0.04 0.022
Shrub land 25.01 13.82
Bare land 26.12 14.48
Total 181 100
----------------------------------------------------------- 52 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
----------------------------------------------------------- 53 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
Change %
1986 2000 2005
Class Name Area(km2) Area (%) Area(km2) Area (%) Area(km2) Area (%) (1986-2000) (2000-2005)
Built-up 1.05 0.600 2.430 1.400 4.610 2.55 131.429 89.712
Agricultural
land 93.40 51.602 96.820 54.000 110.970 61.31 3.662 14.615
Grass and
pasture 20.93 11.564 15.960 9.000 14.780 8.17 -23.746 -7.393
Vegetation 31.06 17.160 14.660 0.040 14.530 8.03 -52.801 -0.887
Waterbody 0.00 0.000 0.040 13.600 0.110 0.06 No data 175.000
Shrub land 16.60 9.171 25.000 8.000 23.490 12.98 50.602 -6.040
Bare land 17.90 9.890 26.120 0.030 12.550 6.93 45.922 -51.953
Class Total 180.94 99.987 181.030 100.000 181.040 100.02 0.050 0.006
120
Change % (1986-2000)
Change % (2000-2005)
70
20
-30
-80
Built-up Agricultural Grass and Vegetation Shrub land Bare land
land pasture
decrease of a given land-use/land cover. But the change of one land-use/land-cover class
to another land-cover class may not be known without further analysis. Therefore, the
----------------------------------------------------------- 54 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
ENVI 4.7 software. The matrix of change detection result shows results from one class
change to another. This method is useful to understand how classified land-covers have
Table 5. 5(a): Land-use/ Land Cover changes (km2) from 1986 to 2000
Class Name Built-up Agricultural Grass & pasture Vegetation Waterbody Shrub land Bare land Raw Total
Built-up 0.88 0.38 0.00 0.51 0.00 0.02 0.63 2.42
Agricultural land 0.02 66.44 8.36 10.45 0.00 2.56 8.90 96.72
Grass and pasture 0.00 3.57 10.22 1.04 0.00 0.00 1.06 15.88
Vegetation 0.05 3.82 0.80 7.07 0.00 0.47 2.74 14.96
Waterbody 0.00 0.02 0.02 0.00 0.00 0.00 0.01 0.04
Shrub land 0.01 2.72 0.00 8.47 0.00 12.17 1.46 24.84
Bare land 0.08 16.44 1.54 3.52 0.00 1.38 3.11 26.07
Class Total 1.05 93.40 20.93 31.06 0.00 16.60 17.90 0.00
Class Changes 0.16 26.96 10.72 23.98 0.00 4.43 14.79 0.00
Image Difference 1.38 3.32 -5.05 -16.09 0.04 8.24 8.17 0.00
The diagonal values from cross tabulation matrix (table 5.5 (a), 5.6 (a) and 5.7 a)) show
land-use/land-covers that were unchanged in the given years. This thematic change
detection was calculated by subtracting the area of each land cover classification of 1986
from 2000 and 2000 from 2005. The resulting thematic change detection statistics matrix
shows how classes have changed from one class at time 1 to another class at time 2.
Class total shows the total area of each land-use/land-cover for the base year and row total
shows the total area of the later year. Image difference shows the difference of the total
area in each class between the study years. Class change is the total change for each land-
----------------------------------------------------------- 55 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
use/ land-cover between the later and the base year under study. The diagonal values
shaded with orange color shows the unchanged total area of the land-use/land-cover
categories.
Table 5. 7 (a): Land-use/ Land Cover changes (km2) from 1986 to 2005
Grass &
Class Name Built-up Agricultural pasture Vegetation Waterbody Shrub land Bare land Raw Total
Built-up 0.96 1.02 0.14 1.26 0.00 0.09 1.10 4.58
Agricultural land 0.01 70.05 12.79 13.40 0.00 4.68 9.81 110.73
Grass and pasture 0.00 5.04 6.87 1.58 0.00 0.03 1.17 14.68
Vegetation 0.02 4.41 0.09 6.79 0.00 1.42 1.68 14.41
Waterbody 0.00 0.03 0.05 0.01 0.00 0.00 0.02 0.11
Shrub land 0.03 5.53 0.11 5.92 0.00 9.66 2.41 23.66
Bare land 0.02 7.32 0.89 2.10 0.00 0.72 1.72 12.77
Class Total 1.05 93.40 20.93 31.06 0.00 16.60 17.90 0.00
Class Changes 0.08 23.35 14.06 24.26 0.00 6.94 16.19 0.00
Image Difference 3.53 17.33 -6.25 -16.64 0.11 7.06 -5.13 0.00
----------------------------------------------------------- 56 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
Table 5. 7(b): Land-use/ Land Cover changes (%) from 1986 to 2005
Grass &
Class Name Built-up Agricultural pasture Vegetation Waterbody Shrub land Bare land Row total
Built-up 92.016 1.096 0.648 4.057 0.000 0.548 6.170 100.000
Agricultural land 1.008 74.997 61.092 43.134 0.000 28.208 54.768 100.000
Grass and pasture 0.000 5.394 32.816 5.092 0.000 0.161 6.533 100.000
Vegetation 2.016 4.726 0.427 21.878 0.000 8.542 9.364 100.000
Waterbody 0.000 0.029 0.244 0.031 0.000 0.000 0.100 100.000
Shrub land 2.946 5.920 0.528 19.048 0.000 58.177 13.479 100.000
Bare land 2.016 7.839 4.245 6.758 0.000 4.364 9.586 100.000
Class Total 100.000 100.000 100.000 100.000 0.000 100.000 100.000 0.000
Class Changes 7.984 25.003 67.184 78.122 0.000 41.823 90.414 0.000
Image Difference 336.977 18.555 -29.848 -53.586 0.000 42.493 -28.668 0.000
The primary focus of this study was to understand the level of change in built-up area for
the given years. The major land-use/land-cover that contributed to the increase of built-up
area is presented in table 5.8. Based on the land-use/land-cover change statistics between
1986 and 2000, 0.38 km2 of agricultural land, 0.51 km2, vegetation, and 0.02 km2 of shrub
land and 0.62 km2 bare lands were converted to built-up area. Between 2000 -2005, 0.29
km2 of agricultural land, 0.08 km2, of grass land, 0.83 of km2 of vegetation, 0.09 km2 of
shrub land and 1.05 km2 of bare land were converted to built-up area (Figure 5.5).
land 25% of the total area change converted to built-up land (Table 5.9. The vegetation
(tree canopy along roads) cover converted to built-up was 33%. These changes may seem
to be classification errors. But in some cases there is a probability that vegetation area may
----------------------------------------------------------- 57 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
be converted to roads. Roads are mostly included to built-up area when classified from
low resolution image. But when tree canopies along the roads grow, it is common to see
mixed pixels on low spatial resolution satellite image. Therefore, pixels near the roads
changes have always some drawbacks. One of the problems occurs is the existence of
mixed pixels on different objects which become spectrally similar one another. For
example, from land-use/land cover conversion statistics, there were errors indicating the
area to other land cover classes is unlikely to happen especially within 5 years (2000 and
2005). One solution that has been done to correct this problem was to compare the result
Figure 5. 5: Land-use/ land-cover class contribute to net change to built-up area (1986-2005)
and may help to identify the current problems, trends, and issues facing the sustainable
----------------------------------------------------------- 58 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
Knowledge of past and current population trends and projecting population growth into
the future helps to determine the level of municipal services that will be needed for future
The population census data from Central Statistics Agency of Ethiopia were used to
calculate total population growth between 1994 and 2007. The most recent census data
(2007) show that Debre Birhan town grew by 65231 people, and had a growth rate of 5.3%
annually. Table 5.10 presents population growth of the census years of October 1994 to
May 2007.
----------------------------------------------------------- 59 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
The population grew 51% between 1984 and 1994, and 68.88% percent between 1994 and
2007. The annual growth rate between 1984 and 1994 and between 1994 and 2007 was 5.1%
and 5.3% respectively (Table 5.10). Generally, the analysis of population statistics shows
high population size and the average annual population growth rate of the town generally
shows very fast. From this analysis it is understood that population pressures in the town
----------------------------------------------------------- 60 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
According to the 1994 and 2007 housing and census report of Central Statistics Agency of
Ethiopia, there were 9317 housing units in October 1994 and 19725 housing units in May
2007(Table 5.12) in Debre Birhan town. Between 1994 and 2007, the town's housing stock
increased by 10408 units, or 111.60%. This percentage increase is greater than the
According to the statistics (Figure 5.6), the number of housing development and provision
has not been proportionate with the rate of population growth in Debre Birhan town.
Factors such as the lack of available finances, limited access to land and overall poor
planning for the provision and distribution of housing has resulted in this disproportional
----------------------------------------------------------- 61 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
Figure 5. 9: Annual growth changes in Housing Units against annual growth changes in Population
----------------------------------------------------------- 62 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
Figure 5.9 shows the annual growth rate in housing units compared to the population
growth rate. The rate of housing growth shows faster than town’s population growth rate.
The population data of 1994 and 2007 were used to calculate the population density for
each block using ArcGIS 10 software. Figure 5.10 shows the population size the town by
block. In 1994 blocks in Kebele 3 had the highest population size and blocks in Kebele 7
had the lowest population size. In 2007, blocks in Kebele 3 and Kebele 1 had the highest
The population density of any region is found by dividing its population by its area:
Due to the absence of census block map, I used Kebele boundary and street block map to
examine population density of the study area. Density per administrative unit area is
----------------------------------------------------------- 63 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
normally calculated by taking all land area such as open spaces, water body, etc to density
5000.0
4000.0
Density/km2
3000.0
2000.0
1000.0
0.0
1 2 3 4 5 6 7 8 9
Density 1994 1381.2 2017.2 3117.4 2318.3 2379.9 2581.9 390.8 1698.3 1155.4
Density 2007 3466.9 2734.8 5373.8 3487.8 4037.7 2225.0 1047.9 3645.0 1753.1
Kebele 1 2 3 4 5 6 7 8 9
----------------------------------------------------------- 64 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
Figure 5.11 and Figure 5.12 shows that the highest population density was exhibited in
Kebele 3 for both 1994 and 2007 and relatively the lowest population density has been
There are a wide variety of density measures. All density measures fall in to: gross density
and net density. Gross density indicates that the entire land area is used as the basis for
calculation and net densities exclude certain land uses or features such as regional open
space (Forsyth, 2003). Therefore, the density of population by Kebele can be considered as
gross density but by excluding some open space from the Kebele, the block density has
----------------------------------------------------------- 65 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
The map in Figure 5.13 shows the population density per block area (km2) in each Kebele.
This calculation was performed by dividing the total population of the Kebele by total
The difference of this population density from the Kebele population density of this map
shows the population density based on the land areas that contain housing units and
The block density map of 1994 and 2007 (Figure 5.14) indicated that the highest population
density occurred in Kebele 3 which was the same as population density by Kebele in
Figure 5.14. The most striking difference of population density by Kebele and street block
was that the block density had a higher density. The reason for this was that the total base
----------------------------------------------------------- 66 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
area used for block density calculation was lower than the base area used in each Kebele.
Kebele 7 and Kebele 9 had a large potential land for residential house development
between the study year of 1995 and 2007. These Kebeles had the lowest population density
than other Kebeles. In general, the north east and central part of the town had a higher
Debre Birhan town shows a fast growth within the last seven years of time (2000- 2008).
The growth of the Debre Birhan town area in the last study years (2000-2008) has occurred
through the linking of a number of smaller urban settlements spread along the main roads
to northeast and west direction from its center. The central part of the town is
characterized by relatively high rise buildings and the peripheral part of the town is
Plate 5.1: Horizontal expansion (left) and vertical growth (right) of built-up area
----------------------------------------------------------- 67 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
Wards). Geographic Information System and Remote Sensing data and technique was
used study the physical extent of urban growth and the spatial pattern of dwelling units’
growth.
Data from Landsat image helps us to map urban growth change results a change in
5
4.5
4
3.5
3
2.5
2
Area in km2
1.5
1
0.5
0
1986 2000 2005
Built-up area 1.05 2.43 4.61
Figure 5. 15: Area (km2) and Growth rate of built-up area (1986-2005).
Between 1986 and 2000 the urban development in Debere Birhan town was very small
compared to the development between 2000 and 2005. From the classified images, the
urbanization growth was mainly on bare soil and agricultural land. The graph of the town
population obviously indicated that there was a massive increase of population between
----------------------------------------------------------- 68 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
The percentage change per year of the urban land uses/ land cover has been calculated
and presented in Table 5.13. The rate of change of land use/land cover per year in the
Table 5. 5: Built-up area absolute growth and annual rate of growth rate%.
To calculate the urban growth speed, one approach is to use observed population growth
rate as the basis for assuming the rate of increase in the urban land uses. The effect of the
Where: ‚A‛ is area; ‚f‛ is the future year, ‚b‛ base year. And % is rate of change per year.
Applying this formula for 2000 and 2005 urban growth on the study area, using the annual
A 2015 =4.61*(1.17)(10)
A 2015 =4.61*4.81
A2015=53.93 km2
----------------------------------------------------------- 69 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
At current growth rate, the urban area of Debre Birhan town will be 53.93 km2 after 10
year (2015). The estimation is using a linear growth rate annually disregarding rules and
regulations that may be practical by the town administration. This prediction may occur if
there is no any regulation mechanism on the growth rate of population and urban sprawl
----------------------------------------------------------- 70 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
Population increases, so does the need for new housing, schools, and transportation
networks. Especially in developing country where population growth is high, the urban
urban area associated with population growth creates a significant change on urban
landscape. Therefore, the urban area is characterized by very low density, leapfrog and
Due to the heterogeneous nature of most of the urban environment, data from high
resolution imagery and topographic map helps to understand the pattern and sprawl
are some indicators which are commonly used by researchers in urban sprawl assessment.
Some of the sprawl metrics as has been mentioned on the literature of this thesis includes
Cutsinger, 2007).
In this work, high resolution Ikonos satellite image (2008) and data from topographic map
(1995) was used to study the urban growth and sprawl of the town. To identify housing
sprawl, a GIS based map was prepared and tried to show where sprawl occurred on the
study area. In order to analysis the pattern of residential houses on the study area:
A GIS map was used to asses visually where scatter (sprawl) development
occurred on the study area in two time frames (1995 and 2005).
Housing density from polygon centroid of each housing units was used to identify
One of the core functions of ArcGIS the spatial statistics was used to analyze the
----------------------------------------------------------- 71 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
Visual inspection of these points reveals some clusters of high housing units, but
By visually displaying land based resource including potential environmental and cultural
resource impacts and economic implications, GIS can help planners, public officials, and
----------------------------------------------------------- 72 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
A map provides the visual aspect from which studies on urban sprawl can begin in
relation to urban growth. Mapping urban phenomena helps to identify where urban
sprawl occurs and where compact form of development can be identified. A Geographic
Information System is useful for mapping the spatial pattern, direction and distribution of
urban areas.
The distribution and pattern of housing in Debre Birhan town from 1995 through 2008 can
be directly examined through direct visual observation from a GIS map presented in
Figure 5.17. From the map of housing spatial distribution, it has been shown that the
development of houses in 2008 were mostly towards the north east and west peripheral
part of town. The remaining houses development distribution shows development inside
the central part of the town and along the major roads.
----------------------------------------------------------- 73 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
Figure 5. 19: Building footprint (1995 and 2008), North east part of the town
----------------------------------------------------------- 74 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
Figure 5. 20: Building footprint (2008) overlaid on Ikonos image and draped on DEM
(Central part of the town).
The expansion of the town needs an in-fill process which can reduce the sprawl of housing
development on agricultural land. The sprawling pattern is the results of a fast- growth
and, absence of tight control of developers according to a defined plan. If there is a town
belt that restricts further informal housing development, sprawling growth begins to slow
down and the growth of the urbanized area expands into filling up the vacant lands left
A recent field photograph with digital camera (January 2011) of Figure 5.21 shows an
expansion of the built-up are to agricultural land (2008, Ikonos image draped on DEM).
----------------------------------------------------------- 75 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
Figure 5. 21: built-up area expansion to agricultural land (field photograph of 2011)
The sprawling of houses was not limited to the town’s administrative boundary. Figure
5.22 shows the development of housing beyond the Kebele boundary(1995& 2008).
Figure 5. 22: Scattering of houses beyond the Kebele boundary (1995 & 2008)
----------------------------------------------------------- 76 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
Spatial analysis based on building population is very rare or absent in GIS. Spatial
provide humanitarian assistance, for customer’s physical distribution is one of the key
However, there is always a problem to get building population data to use in GIS analyses
since building population data is not available at polygon level for public use. Therefore,
building footprints digitized from high resolution Ikonos image and automatically
extracted from scanned topographic map were used to investigate spatial distribution of
Visual assessment of the residential house which contains the results from the digitized
polygon of residential houses from Ikonos image and the automatically extracted polygon
A. Housing Density
Density has been one of the most employed variables to look at the spatial structure of
cities. Probably because of that, it has also been one of the phenomena to which more
attention has been paid in order to measure sprawl, and several ways have been proposed
Block Density can be calculated as dwelling unit divided by block area measured to the
curb if the boundaries of the street block are clear. This is relatively easy to measure from
We have data such as blocks and buildings, for the areas under study, based on which we
can compute the density of buildings per square kilometer. As shown in Table 5.14, there
is a strong relation between the density of buildings per square kilometer at block level
and at Kebele level for the 1995 and 2008 years of data.
----------------------------------------------------------- 77 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
Kebele Block
Building Building Building
density density Building density
2 2 2 2
Area Block Buildings Buildings (km ) (km ) density (km ) (km )
2 2
Kebele km Area(km ) 1995 2007 1995 2008 1995 2008
1 2.39 0.396777 481 1366 201.26 571.55 1212.27 3442.74
2 2.04 0.699911 1243 3065 609.31 1502.45 1775.94 4379.13
3 1.49 0.329168 969 1993 650.34 1337.58 2943.79 6054.66
4 2.62 1.102837 1120 3337 427.48 1273.66 1015.56 3025.83
5 1.54 0.596104 975 1757 633.12 1140.91 1635.62 2947.47
6 1.88 0.642195 1138 2760 605.32 1468.09 1772.05 4297.76
7 6.54 1.100492 875 2012 133.79 307.65 795.10 1828.27
8 2.38 0.755699 939 2880 394.54 1210.08 1242.56 3811.04
9 4.73 0.777518 961 2845 2031.92 601.48 1235.98 3659.08
7000.00
6000.00
0.00
1 2 3 4 5 6 7 8 9
Figure 5. 23: Building density per km2 by Kebele By block and Kebele(1995- 2008)
The comparison of density of buildings per km2 at block level and Kebele shows that there
higher density of buildings in 2008 at block level. In 1995, the density magnitude was
nearly similar to the density of housing units by Kebele at 2008. This means that in 1995,
----------------------------------------------------------- 78 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
there was large open space within the block that was not occupied by buildings because
Figure 5.24, 25, 26 and 27 shows the spatial distribution of building footprint at block and
Kebele
----------------------------------------------------------- 79 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
----------------------------------------------------------- 80 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
The density of housing units was calculated by dividing the total housing units by total
area of Kebele boundary. The highest density of houses were exhibited in Kebele 5 and
Kebele 3 in 1995 (Figure 5.24) and Kebele 6 and Kebele 2 in 2008 (Figure 5.25). The density
of housing units more accurately was calculated using street blocks. The comparison of
housing density by Kebele area in km2 and block area showed remarkable difference. The
highest density of ousing units were calculated in Kebele 5 and 3 for the 1995 and Kebele 6
and Kebele 2 for 2008. The highest housing density at block level exhibited in Kebele 3
with maximum density of 2900 housing units per km2 (1995) and with maximum of 6100
housing unit density per km2 in 2008( Figure 5.26 and Figure 5.27).
----------------------------------------------------------- 81 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
----------------------------------------------------------- 82 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
outliers are located. The importance of housing to both the ecological and social sciences
has focused increasing attention on the spatial patterns of housing growth. Furthermore,
because housing growth is not a spatially random process, understanding its patterns
The following question can be used to guide to understand the spatial pattern of housing
units in this work.
What is the pattern of points in terms of their nearest distances from each other?
Or how spread is the housing units across the total urban area? The point data
from polygon centroid was used because point data are the straight forward for
this analysis.
To identify housing growth and spatial pattern, local Getis-Ord G* statistic was used to
determine where housing growth hotspots exist in the study area. The Spatial Analyzing
Patterns tool set in ArcGIS was used to identify the spatial patterns of residential housing
units. The patterns with more clustering of housing units that contain hot spots and cold
Figure 5. 28: Region of Hot spot and Cold spots in a normal distribution graph
(Zscore below -1.96 is the locations of cold spots and Zscore above is the location of 1.96 hotspots)
----------------------------------------------------------- 83 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
From the analysis, local Getis- Ord G* hotspot and cold spot map indicated that there was
a clear spatial pattern of clustering of housing units on the north east part of the town. The
hotspot of housing unit location indicated by cyan color clustered polygon centroids and
the cold spots of the housing unit analysis shown with blue color polygon centroids
(Figure 5.29). According to this statistics, we can say that point patterns between G zscore
>-2 and < 2 are randomly distributed, point values on the normal distribution with
Gzscore < -2 and >2 can be clustered which shows a compact development.
Figure 5. 29: Map shows Hot spot and Cold spots of housing units
----------------------------------------------------------- 84 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
From Fig 5.30, considering the Getis- OrdG* map, high Z-score value >2, shows high
values are clustered together in the Northeastern part and low z-scores value < -2 indicate
low values are generally clustered in the west parts of the town. The Z-score value
between -2 and 2 are generally a random pattern of housing units. That mean the
----------------------------------------------------------- 85 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
Generally, the higher (or lower) the z-score, the more intense the clustering is . A z-score
near zero indicates no apparent spatial clustering. The spatial statics of housing unit
distribution analysis indicated that the majority of the housing distribution didn’t show a
cluster pattern. The spatial relationship of the point features therefore indicates a random
----------------------------------------------------------- 86 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
urbanization is undergoing with the mixture of activities and fragmented land- use that
Land suitability evaluation defines suitability of certain kinds of land use based on natural
overall land use planning, which requires a scientific approach to guide development and
analysis may be used to attain a specific goal from multiple alternatives where there is a
need of land use suitability modeling for urban expansion. In this thesis, an evaluation
model of multi-criteria urban expansion suitability using GIS platform was developed for
the study area to point out where the future sprawl and urban expansion will occur. A
number of factors, important in reality, will not be considered in this thesis. Some of the
factors that are not included are: reserved land, some of the environmental constraints,
This approach will define a suitability map for urban land cover based on the conditions
that are associated with current urban land and areas of recent urban land cover change.
valleys, drainages, land-use/land-cover and other factors such as distance from the center
of the town, distance from the main road, will be used to define appropriate conditions for
urban growth. Socio-economic variables, including lands that are protected such as
Because the suitability model is focuses on urban landscape characteristics, and depend
on only available data, it will have shortcomings to include, geological factors, soli
----------------------------------------------------------- 87 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
suitability, and socio economic conditions to map and to have a better representation of
The basic assumption of urban expansion suitability analysis is guided by the following
ideas. All lands based on land use/land cover classification (2005) are available for urban
expansion. The expansion model was, based on physical factors and constraints which are
their influence land use suitability. Identifying the excluded area that restrict
development, and mapping the suitable area for urban expansion by ranking those factors
The first procedure as it is indicated in figure 6.1, was designing a suitability model to
assess the amount of suitable area for urban expansion. The suitability factors were rated
by combining them according to their rank. The final suitable map was produced as
----------------------------------------------------------- 88 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
highly suitable, moderately suitable, and unsuitable. There were also areas that cannot be
used for any construction activity. These areas called constraints and were excluded from
Areas where development cannot occur because of either strict policy enforcement or
physical barriers for development are called constraints. Thus constraints are physical
factors or areas that require high costs and technology for development, areas that cannot
Constraints are the Boolean conditions which limit the alternatives under consideration
(Eastman, 2006). Some of the constraint that restricts urban expansion on the study area
includes open water body (rivers and lake) topographic conditions and woodlots. A 200
----------------------------------------------------------- 89 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
meter buffer on the two rivers that pass through the town, woodlots, and the valley and
gorges found in the vicinity of the town, were excluded as constraints. The selection of
200m buffer for the rivers was based on the agreements from local experts in the study
area.
6.1.1.1.Topography
Both topography and slope play an important part in determining urban development in a
given site. While topography is the general configuration of the land surface which is
measured by elevation and slope is the percentage change in the elevation over a certain
distance.
Debre Birhan town lies on a plateau and surrounded by ridges oriented north-east, south-
west situated between the two broad valleys found in the vicinity of the town (Figure 6.3).
The general topography of the town is therefore includes plateau, undulating hills and
valleys. These undulating hills and valleys considered as constituents for future expansion
of the town.
6.1.1.2. Drainage
To avoid risks from river flood, runoff, and water pollution, generally, construction at
some distance from open water body and rivers is chosen. Debre Birhan town has two
rivers starts from the eastern part of the plateau and drained from southeast to northwest
towards Jema River: River Beresa is a perennial river and Dalecha is an intermittent river.
These rivers have therefore limitations for development because they must employ costly
building and landscape measures to insure that destructive damage does not occur to
residential and commercial housing expansion. Although there were not densely
----------------------------------------------------------- 90 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
vegetated urban forests in the study area, there are eucalyptus woodlots around the town.
For the purpose of this study these areas were excluded from urban expansion suitability
site selection.
Figure 6. 3: Topography.
Factor is a general category of information concerning the site being evaluated. The factors
factors (e.g. slope, geologic hazards, land cover, etc), built environment factors (e.g.
existing land use, planned land use, accessibility to roads, availability of utilities, etc.),
economic factors (e.g. land value, tax rate, etc.), and social factors (e.g. population present,
----------------------------------------------------------- 91 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
In this study, economic and social factors, municipal land use restrictions, and factors such
as soil suitability, water availability were not used due to lack of information.
The central part of the town was considered as very suitable for urban development.
Hence, the suitability will decrease from the center gradually. In general, areas closer to
currently developed land are more suitable than areas farther from developed land
Distance from the main road not affect suitability very strongly although a large distance
involves some extra cost and increase in travel time. But a new road might be constructed
to change the accessibility. This factor, therefore, will have low weighting factor compared
to others. The major road from Addis Ababa to Dessie crosses Debere Birhan town. Other
main roads are road to to Ankober and Jiru. The residential roads are not included for
suitability analysis. Like distance from the center, the suitability of urban growth also
decreases gradually from the main road. Buffer distance from the main road 500m and
C. Slope factor
Low slopes make housing and road construction less expensive. A slope also will cause
considerably higher construction costs, which strongly affects its suitability. Therefore, in
most cases the slope < 15% are mostly taken as suitable for construction and slope > 15%
are unsuitable.
would be the future development than the sub-urban area. That means, replacing or
renewing the densely built-up area requires, a larger cost for developers. Therefore,
suitability from highly densely populated area increases gradually until certain distance
limit. Sometimes densely populated area may attract developers due to market reasoning.
This may facilitate urban growth on densely populated area. However, the assumption in
this work is to know what factors should be considered for horizontal urban expansion.
The point density of individual housing units was calculated and classified to standard
----------------------------------------------------------- 93 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
E. Land-use/ land-cover
Agricultural and grass lands are being extensively used for urban expansion compared to
other land cover land uses such as bush land and shrub.
the comparison of each factor possible. The common measurement scale helps to
determines how suitable a particular location for urban expansion. Numeric scores to a
total of ten (one as least and ten as most suitable) were assigned to each factor attribute
class in spatial analysis of ArcGIS 10.0 platform. ArcGIS 10 (ESRI) with the Spatial Analyst
Extension that includes the Model Builder tool is well suited for land suitability modeling
in this study. The datasets were raster (grid) format with cells size of 30 meters.
----------------------------------------------------------- 94 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
All the criteria were aggregated with the most commonly used decision rule called
formula:
S=ΣWiXix Πcj……………………………………………………..4
Where:
The weighting of a suitability factor also depends on the intended land use:
In this study slope has been given higher weighting factor because it causes considerably
higher construction costs, which strongly affects its suitability. The distance to a main
road, on the other hand, may not affect suitability very strongly: a large distance involves
some extra cost and increase in travel time. Moreover, a new road might be constructed to
change
----------------------------------------------------------- 95 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
Table 6. 1: Pair wise comparison matrix of five factors to urban growth suitability
The eigenvector weight of each factor is calculated using IDRISI 15.0 software. Based on
the calculated weights, slope gradient is the first factor that influences urban expansion
and therefore has been given the highest weight, town proximity has been given 33%
weighting value. The next influencing factors include the densification of buildings on the
existing built-up area and proximity to the main road which have weights 17 % and 8%
ID Factor Weight %
----------------------------------------------------------- 96 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
By multiplying all layers of factors and excluding those area which restrict future
expansion the urban growth suitability was produced. According to the suitability
analysis of urban expansion map (Figure 6.7), the most suitable location of land for urban
expansion was agricultural land, grass and bare soil. The existing built-up area ranked as
the next suitability area. There are also areas that were not suitable for urban growth
inside the town because of the high density of the current built-up area.
The assessment of urban expansion suitability map indicated that the most suitable
available land for urban expansion are agricultural and grass land. This indicated that for
the next coming years, the rapid increase of urban sprawl will create a pressure on
intensively cultivated agricultural land that found near and around the existing urban
based Multi- Criteria Evaluation (MCE) techniques used to identify the urban growth,
for this study, Ikonos image of 2008 was used. According to this suitability analysis result,
all agricultural land and grass land were most suitable for urban growth. It is necessary to
keep in mind here that this approach does not intend to model land suitability map for
urban sprawl but aims at generating land suitability map for sustainable urban growth
process.
but urban sprawl is an informal scatter development which results loss of environmental
resources and the expansion doesn’t consider the level of economic development.
Moreover, all parameters and their values used in this suitability analysis are subjective in
reality because urban growth suitability map needs experts from different departments
such as geology, hydrology, urban planning etc. But we assume that the contribution of
----------------------------------------------------------- 97 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
this work may be appreciated from GIS and remote sensing perspective since it provides
the most valuable data sources and information for urban planners.
----------------------------------------------------------- 98 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
----------------------------------------------------------- 99 ------------------------------------------------------------------------
---
2011 Urban sprawl mapping and land-use/land-cover change statistics
========================================================================================
This study demonstrates the use of Remote sensing and Geographic Information System
(GIS) to analyze the urban growth and sprawl mapping in the study area. Attempt was
made to capture as accurate as possible seven land use-land cover-classes as they change
through time. Satellite data are found to be useful in mapping and quantifying the
historical land-use/land cover and the rate of urban growth in different time periods.
In order to give answers of the research questions framed in the present study, different
experiments were carried out using multi-resolution satellite images and data from
topographic map. From the classified land-use/land-cover classes, more emphasis has
been given on built-up land as it affects all other class categories. The land-cover/land-use
change statistics indicated that the growth rate of built up area occupied the largest share
among other land-cover/land-use categories. The result of this work shows a rapid
growth in built-up land between 2000 and 2005 while the periods between 1986 and 2000
witnessed relatively slow growth. The urban area development occurred mainly on bare
GIS based analysis of the pattern of urban expansion over the demographic change and
landuse modifications has also indicated that urban growth has mainly taken place
The vector data extracted from Ikonos imagery and topographic map of my findings were
used to map whether housing development was sprawling or compact. Density of houses
by Kebele and blocks was computed and mapped to identify the location of densely
populated area with population and housing units. Residential house scattering and
compactness has been measured with spatial statistics of hotspot analysis. The GIS map
shows that residential housing development has already sprawled during the study
period.
The assessment of urban expansion suitability map indicated that the most suitable
available land for urban expansion are agricultural and grass land. This indicated that for
the next coming years, the rapid increase of urban sprawl will create a pressure on
intensively cultivated agricultural land that found near and around the existing urban
area. This may highlight some clues to decision makers and planners to understand the
environmental change due to anthropogenic factors and to find solutions for sustainable
urban development
7.2. Recommendations
There is a need to plan for balanced physical urban growth and population growth in our
cities and towns. This can only be possible if there is an understanding of the nature of
urban growth change, the urban demographic pressure, the level of service available to
supply, the source and amount of resources to future growth of cities and towns.
Urban planning authorities and town planners should think about the future growth of
the town and should understand the consequence of unbalanced physical urban growth
Urban planning authorities should have to depend on a GIS data base and information
system to regulate their urban development in a sustainable way such that they will
manage the supply of public services and infrastructures that will be needed as a result of
Satellite remote sensing with repetitive and synoptic viewing capabilities, as well as
multispectral capabilities, is a powerful tool for mapping and monitoring the ecological
changes in the urban core and in the peripheral land-use planning. The use of remote
sensing needs to be introduced for monitoring the activities of developers. This will help
in reducing unplanned urban sprawls and the associated loss of agricultural lands
The last section of this thesis looked towards mapping the land requirements for urban
expansion. By using GIS based multicriteria evaluation method, the future suitable urban
expansion land has been mapped which can be used as a visual guide for urban planners.
This will help them to plan urban growth with effective design, not just architecturally,
but also at the ‘urban design’ level which is crucial to understand that higher-density
1. The use of GIS and remote sensing based urban land management and urban
sprawl controlling
To reduce the impact of urban sprawl, the following four policy measures are suggested:
4. Controlling land use and residential development through land price regulations
buildings and infill development of high rise buildings to meet the demands and
References
Alberti, M. and Torrens, P. (2000). Measuring Sprawl. Working Paper Series No.
27, Center for Advanced Spatial Analysis. University College London.
Anselin, L. (2003). GeoDa 0.9 User’s Guide, Center for Spatially Integrated Social
Science, Urbana-Champagne
Barnes, K.B., Morgan III, J.M., Roberge M.C. and Lowe, S. (2001). Sprawl
development: Its patterns, consequences, and measurement. Towson
University: A white paper. Available at:
http://chesapeake.towson.edu/landscape/ ... _paper.pdf
Bull, T.A. (1988). Agroecological assessment of Ethiopian Vertisols. In: Jutzi ,S. C.,
Haque, I. Proceedings of a conference held at ILCA, Addis Ababa Ethiopia
31 August-4 September 1987. ILCA (International Livestock Centre for
Africa), Addis Ababa Ethiopia, pp: 89-105.
Campagna, M. (ed.) GIS for sustainable development. London: Taylor and Francis,
2006.
Central Statistics Agency (CSA). The 1994 and the 2007 population and Housing
Census of Ethiopia. Summary Report at Country and Regional Level.
Central Statistical Authority (CSA), 1996. The 1994 Population and Housing
Census of Ethiopia: Results for Amhara Region. Statistical Report on
Population Size and Characteristics. Volume I: Part I. Ethiopia, Addis Ababa
Central Statistical Authority (CSA), 2010. The 2007 Population and Housing
Census of Ethiopia: Results for Amhara Region. Statistical Report on
Population Size and Characteristics. Volume I: Part I. Ethiopia, Addis Ababa
De Jong S.M., F.D. van der Meer & J.G.P.W. Clevers.(2004). Basics of Remote
Sensing. In: De Jong S.M. & F.D. Van de Meer F.D. (Eds.), Remote Sensing
Eastman, J. R. (2006). IDRISI Andes Tutorial. Clark Labs Clark University, 950
Main Street Worcester, MA 01610-1477 USA. Manual Version 15. Web:
http://www.clarklabs.org
Ewing, R. (1997). ‚Is Los Angeles-Style sprawl desirable?‛ Journal of the American
Planning Association, 63(1): 107-126.
Ewing, R., Pendall, R. and Chen, D. (2002). Measuring Sprawl and Its Impact.
Smart Growth America. Washington, DC, USA.
FAO (Food and Agriculture Organization of the United Nations) (1997). The
Digital Soil and terrain Database of East Africa (SEA). Notes on the Arc/Info
files. Version 1.0 completed 3rd April 1997.
Getis A., Ord J.K. (1992). The analysis of spatial association by use of distance
statistics, Geographical Analysis, 24: 189-206.
Griffith, D.A. (1984). Theory of Spatial Statistics. In: Gaile, G.L. and Willmott, C.J.
(eds.) Spatial Statistics and Models. Dordrect: D. Reidel Publishing Company
Gou, Y.H. (2003).Using GIS to Examine Urban Growth Pattern in Alachua County,
Florida. Master thesis in Urban Planning at Virginia, Polytechnic Institute
and State University.
Irwin, E.G. and Bockstael, N., and Cho, H.J. (2006) Measuring and Modeling Urban
Sprawl: Data, Scale and Spatial Dependencies, unpublished working paper.
Lee, J. and Wong, D. (2005). Statistical Analysis with ArcView GIS. New York.
Wiley, 2nd Edition.
Lepczyk, C.A.; Lortie, C.J.; Anderson, L.J. (2008) . An ontology for landscapes.
Ecological Complexity 5(3): 272-279
Lillesand, T.M., and Kifer, R.W. (1994). Remote Sensing and Image Interpretation.
(Third Edition). United States of America: John Wiley & sons, Inc.
Lu, D., and Weng, Q. (2006). Use of Impervious Surface in Urban Land-Use
Classification. Remote Sensing of Environment. 102: 146-160.
Lwin, K., (2010). Online Micro-spatial Analysis Based on GIS Estimated Building
Population: A Case of Tsukuba City. Doctoral Program in Geoenvironmental
Sciences. University of Tsukuba
Myers, D., & Kitsuse, K. (1999). The Debate over Future Density of Development:
An Interpretive Review. Working Paper. Washington, D.C.: Lincoln Institute
of Land Policy.
National Meteorological Agency (NMA). Climate data for Debre Birhan Station
(1999-2008), Central Statistical Agency, Addis Ababa, Ethiopia.
Ryan J. H. (2002). Measuring sprawl in Rhode Island: a Gis based analysis of land
use change in the Pawcatuck river watershed. Master thesis,University of
Cornell.
Schowengerdt, R.A. (2007). Remote sensing: models and methods for image
processing. 3rd Edition. Amsterdam: Elsevier; Burlington: Academic Press.
Singh, A. (1989). Digital change detection techniques using remotely sensed data.
International Journal of Remote Sensing 10: 989–1003.
Sudhira, H. S., Ramachandra, T.V. and Jagadish, K.S. (2004). Urban sprawl:
metrics, dynamics and modeling using GIS. International Journal of Applied
Earth Observations and Geoinformation 5(1): 29-39.
Sun, H., Forsythe, W. and Waters, N. (2007). Modeling Urban Land Use Change
and Urban Sprawl: Calgary, Alberta, Canada. Networks and Spatial
Economics, 7(4): 353-376.
Theobald, D. M. (2001). ‚Quantifying urban and rural sprawl using the sprawl
index‛, Paper presented at the annual conference of the Association of
American Geographers in New York, on March 2nd, 2001
Tsai, Y.H. (2005). Quantifying Urban Form: Compactness versus Sprawl. Taylor &
Francis. Urban Studies 42(1): 141-161.
UN-Habitat (2010). State of the World Cities, Bridging the Urban Divide Report
2010/2011. London: Earthscan.
Yeh, A., and Li, X., (2001). Measurement and monitoring of urban sprawl in
rapidly growing Region using entropy. Photogrammetric Engineering &
Remote Sensing. 67(1):83-90.
Appendices
Appendix 1: Monthly temperatures data (1999-2008)
Year Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov Dec
1999 Maximum 19.4 21.5 20.5 21.8 22.6 22.4 17.1 18.1 18.5 17.9 17.7 18.5
Minimum 2.2 3.1 5.9 5.2 5.8 6.3 8.6 8.6 6.2 4.6 -1.2 1.8
2000 Maximum 19.7 20.5 21.5 20.8 21.8 22.7 18.3 17.7 18.5 18.4 18.5 19.4
Minimum 1 3.4 4.8 7 7.1 5.7 8.6 8.1 6.8 3.3 1.7 1
2001 Maximum 19.9 20.8 19.2 21.2 22.2 21.5 17.9 17.6 19.6 19.9 19.6 19.3
Minimum 1.5 4.5 8.3 6 8.2 8 9.2 9.2 6.4 3.3 2.9 4.6
2002 Maximum 19.5 21.2 20.7 21.4 23 22.8 21.5 17.5 18.6 19.6 19.8 19.3
Minimum 5 6.3 8.4 7.6 7.7 7.6 9.1 9.1 7.5 3.1 2.8 7.2
2003 Maximum 20.3 21.4 21.2 20.8 21.8 22.5 18 18.3 19 19.4 18.5 19.4
Minimum 5.3 6.9 7.9 9.5 7 8 9.4 9.6 8.1 3 2.7 1.6
2004 Maximum 20.7 20.4 21 20.4 23.1 21.2 18.7 18.7 19.1 18.6 19.5 19.6
Minimum 5 4.4 5.6 8.9 6.7 8.1 9.2 9 7 3.3 2.7 5.2
2005 Maximum 20.1 22.1 22 21.3 20.4 21.6 18.3 18.9 19.3 19 18.8 18.9
Minimum 5.3 7.3 8.3 9.2 9.2 7.8 9.2 9.2 8.3 3.9 2 1.7
2006 Maximum 20 21 20.5 20 21.9 22.6 19 18 18.6 19.6 18.3 19.1
Minimum 5.9 8 7.5 8.5 7.4 8.2 9.8 9.8 7.4 5.4 2.8 4.7
2007 Maximum 19.9 20.6 21.8 20.8 22.7 21.2 18 18.2 18.7 19 18.6 20
Minimum 6.6 7.5 7.7 8.5 8.1 9 9.6 9.2 7.5 3.4 3.9 1.5
2008 Maximum 19.4 20.2 22 20.8 21.3 21.2 18.9 19 19.4 19.2 18.4 19.1
Minimum 5.4 5 5.5 7.7 8.6 8.5 8.6 9.2 7 4.2 4 4.6
Average 12.16 13.329 13.87 14.28 14.84 14.705 13.85 13.617 13.075 11.35 10.62 11.2914
Source: Source: National Meteorological Agency of Ethiopian
Annual
Year Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov Dec Rainfall
1999 6.9 0 26.6 2.8 11 48.9 362.4 365.1 52.4 59.6 1.4 0 937.1
2000 0 0 25.9 47.3 37.1 45.8 352.4 317.5 105.2 28.5 18.8 6.8 985.3
2001 0 33.8 71.2 45 64.6 34.9 406.7 260.4 32.2 4.1 0 3.4 956.3
2002 18.1 28 60.6 46.1 18.4 29.1 214.4 294.8 109.1 3.1 0 8.4 830.1
2003 15.6 36.3 60.2 85.7 3.8 99.5 334.7 288.7 74.2 7.1 0 7.4 1013.2
2004 24.4 9.7 29.7 113.3 5.6 99.7 334.7 301.3 78.9 14.1 11.8 0 1023.2
2005 34.3 4.5 28.6 49.5 76.4 91.1 310.7 228.3 106.8 0.7 1.5 0 932.4
2006 17.3 24.4 61 38.3 19.8 35.2 432.6 224.2 59.8 8.6 4.5 26.3 952
2007 0.3 30.4 8.9 71.8 13.6 93.2 309.9 414.6 128.5 4.9 5.7 2 1083.8
2008 0.3 1.7 0 34.6 68.9 66.4 397.7 234.8 76.6 9.9 54.6 1.2 946.7
Source: National Meteorological Agency of Ethiopian