Recently R@2013@T

ADDIS ABABA UNIVERSITY
ADDIS ABABA INSTITUTE OF TECHNOLOGY
SCHOOL OF CIVIL AND ENVIRONMENTAL ENGINEERING
DEPARTMENT OF GEODESY AND GEOMATICS
(SPECIALIZATION GEOMATICS)
MODELING LAND USE LAND COVER DYNAMICS USING REMOTE SENSING AND
GIS TECHNIQUES: THE CASE OF AMBO TOWN.
Research submitted to the school of Civil and Environmental Engineering Graduate

studies of Addis Ababa Institute of Technology in partial fulfillment of the requirement for
the degree of Master of Science in Geodesy and Geomatics Engineering (specialization in
Geomatics Engineering).
BY: TEDELA MOSISA
ADVISOR: WORKU ZEWDIE (Ph.D.)
ADDIS ABABA, ETHIOPIA
Oct. 2020
i
APPROVAL PAGE
Submitted by: Tedela Mosisa ---------------------- -----------------
Signature Date
Principal Advisor: ---------------------- ----------------
Signature Date
Endorsed by-------------------------------------- ---------------- --------------

(School of Civil and Environmental Engineering) signature Date
------------------------------------- --------------------- ---------

Associate Director Signature Date
Post Graduate Program

Declaration of originality
I declare that the work which is being presented in the thesis entitled “MODELING LAND USE
LAND COVER DYNAMICS USING REMOTE SENSING AND GIS TECHNIQUES: THE
CASE OF AMBO TOWN ” Submitted to Addis Ababa Institute of Technology, School of Civil
and Environmental Engineering in partial fulfillment of the requirements for the award of the
degree of master of science in Geomatics Engineering is entirely my work carried out from April
to October 2020 under supervision of Dr. Worku Zewdie ( Advisor) from Ethiopia Space
Science and Technology Institution. All references, including citations of published and
unpublished sources, have been appropriately acknowledged in the work. I further declare that
the work has not been submitted for academic examination, either in its original or similar form,
anywhere else.
Tedela Mosisa ------------------------- ------------- Name of Candidate Signature Date as Master

research advisors, we hereby certify that we have read and evaluated this MSc Research prepared
under our guidance by me, entitled.
Prof .----------------------------- ---------------------- ---------------
Advisor Signature Date
Mr---------------------------- ----------------------- ------------------
Co-Advisor Signature Date
ii
Acknowledgments
First, all praises belong to the almighty “God”, the most merciful, and the most beneficent who
gives me healthy and who make everything is possible for me. I gratefully acknowledge Ambo
University for granting me the chance to pursue my studies at the Addis Ababa Institute of
Technology University. I would like to express my deepest thank and gratitude to my advisor
Worku Zewdie (Ph.D.) for serving as supervisor and for his support, dedication, timely advice,
and willingness to consult me at any time, his patience and experience he shared with me thanks
to you for confidence and encourage me that led this thesis successfully. My great thankfulness
should go to my family for their encouragement and my beloved wife Shibire Belete for her
relentless help throughout my studies. Finally many thanks to my classmates from all over the
world for sharing their knowledge and for their moral support.
iii
ABSTRACT
The main objective of the study is an abstraction or simplification of the real world. Models are being
used as a high degree having power learning advice. The modeling used in the present thesis comes under
having particular shared characteristics of ‘scenario models’. The knowledge of land use land cover
dynamics is important for many planning and management activities. The express concept of land use
refers to the human activity or economic function associated with a specific piece of land. The term land
use land covers dynamic land cover relates to the type of feature present on the surface of the earth. Land
use and land cover change are complex and highly dynamic. Many scholars have analyzed the pattern
over some time. On the contrary, these studies did not address the role of driving factors in carrying
change. Most have objective reality models that are designed to take into account either non-spatial or
spatial factors and therefore are unable to do of handling both. Human driving factors are relatively
active, whereas Biophysical factors are relatively stable and have cumulative effects. Few scholars have
tried to integrate spatially and CA as dynamic discrete space and time systems. A cellular automaton
system consists of a regular grid of cells there are dynamics models and CA models. To take advantage of
the models, many kinds of research integrated multiple models. The maximum likelihood algorism of
supervised Image classification was used to generate land use land cover maps. This proposal assesses the
use of GIS and RS in mapping land use land cover dynamics in Ambo Town between 1990 and 2020 to
detect and analyze the change detection that has taken in the Town between these portions of time. To
achieve these Satellite of land sat TM for 1990, Land sat ETM for 2000, 2010, and Land sat 8 for image
2020 has been obtained and preprocessing using EARDAS IMAGINE. The maximum likelihood
algorism of supervised Image classifications have will be used to generate land use land cover maps.
Land use, land classification, dynamic map, accuracy assessment will be used by using Arc GIS. For the
accuracy of the classified Land use land cover dynamic maps, they were used to drive. The aggregate rate
of the dynamic of Land uses the land cover of the Ambo Town result that considerable dynamic has
occurred within twenty-nine (30) years from 1990 to 2020. Besides, a Cellular Automata Markov (CA-
iv
Markov) modeling approach has also been applied to predict land-use change for 2030. Furthermore, GIS
and remote sensing are some of the important sources of data to carry out both spatial and temporal land
use land cover dynamic studies. Based on the present review, it can be concluded that, while significant
challenges exist, these models make available a promising new tool for scholars whose goal is to create
fine-scale models of land use land cover dynamic phenomena that focus on human-environment
interactions.
Keywords:: land use land cover dynamic, land use, Remote sensing LULC, GIS, Idris, Modeling,
and CA_Markov.
Contents_________________________________________________________________ page
List of figure______________________________________________________________pageix
CHAPTER ONE..............................................................................................................................1
1. INTRODUCTION...................................................................................................................1
1.1. Background.......................................................................................................................1
1.2. Statement of the problem..................................................................................................4
1.3 Research questions were:...................................................................................................5
1.4 Objective of the Study......................................................................................................5
1.4.1. The specific objectives of the proposal were:................................................................5
1.5 Significance of the research...................................................................................................5
CHAPTER TWO.............................................................................................................................6
2. REVIEW OF RELATED LITERATURE...................................................................................6
2.1. The concept and definition of land use land cover...............................................................6
2.2. Land use land cover dynamic................................................................................................7
2.3. Tools use for urban Land Use Land Cover change...............................................................8
v
2.3.1 Remote Sensing...............................................................................................................8
2.3.2 Geographic Information System (GIS)...........................................................................9
2.3.3 IDRISI.............................................................................................................................9
2.4. Conceptual Frameworks of Urban land use land cover change........................................9
2.4.1. Factors of urbanization and urban expansion.................................................................9
2.4.1.1. Demographic factors...................................................................................................9
2.5. Image Classification........................................................................................................10
2.5.2. Types of Image Classification......................................................................................12
2.5.2.2. Supervised Classification..........................................................................................12
2.5.3. Image Enhancement.....................................................................................................13
2.5.4. Dynamic Methods........................................................................................................13
2.6. Post Classification Dynamic...............................................................................................13
2.7. Accuracy Assessment.........................................................................................................14
2.8. Modeling LULC Change....................................................................................................14
2.8.1. Modeling LULC Dynamic with CA-Markov...............................................................15
Multi-Criteria Evaluations......................................................................................................17
CHAPTER THREE.......................................................................................................................18
3. MATERIALS AND METHODOLOGY..............................................................................18
3.1. Description of the Study Area.............................................................................................18
3.1.1. Location........................................................................................................................18
3.1.2. Climate of the study area..............................................................................................18
3.1.3. Population.....................................................................................................................19
3.1.4 Vegetation.....................................................................................................................19
3.1.5. Identification of Major Land use land cover classes exist in the study area................19
3.2 Data Source..........................................................................................................................20
vi
3.2.1 Satellite imageries.........................................................................................................20
3.2.1 Data collection...............................................................................................................20
3.3. Land Use and Land Cover scenarios and future demands..................................................20
3.4. Methods of data analysis.....................................................................................................21
3.4.1. Materials and software.................................................................................................21
3.4.2. LULC Classification.....................................................................................................22
3.5. Digital Image Processing........................................................................................................23
3.5.1 Pre-Processing...................................................................................................................24
3.5.2 Image Enhancement..........................................................................................................24
3.5.3 Image Transformations.....................................................................................................25
3.6. Image classification.............................................................................................................25
3.6.1. Pixel-based classification.................................................................................................26
3.6.2 Training Site Selection......................................................................................................26
3.6.3. Supervised and Unsupervised Classification...................................................................26
3.7. Accuracy Assessment.........................................................................................................28
3.7.1 Overall accuracy...............................................................................................................30
3.7.4. Producer Accuracy...........................................................................................................30
3.7.5. User Accuracy..................................................................................................................30
3.7.6. Kappa coefficient.............................................................................................................30
3.8. Dynamic process.................................................................................................................31
3.8.1. Modeling LULC dynamic............................................................................................31
3.8.2. Modeling LULC change with CA-Markov..................................................................32
3.8.3. Model Calibration.........................................................................................................32
3.8.4. Criteria Development.......................................................................................................33
3.8.6. Modeling LULC for 2020 with CA-Markov................................................................35
vii
3.8.7. Model Validation..........................................................................................................35
3.8.7.1.Kappa..........................................................................................................................36
3.8.8.2. Kno............................................................................................................................37
3.8.9.3. Klocation...................................................................................................................37
3.9.1. Kstandard......................................................................................................................37
3.10. Future Prediction map.......................................................................................................37
CHAPTER FOUR.........................................................................................................................38
4. DATA ANALYSIS, RESULTS, AND DISCUSSION.........................................................38
4.1. INTRODUCTION...........................................................................................................38
4.2. Data Analysis, Results, and Interpretation..........................................................................38
4.2.1. Layer stack...................................................................................................................38
4.2.2. Digital Image Processing..............................................................................................38
4.2.3. Land use and land cover change...................................................................................38
4.2.4. Land use and land cover image classification..............................................................40
4.2.5.1. Overall Accuracy.......................................................................................................42
4.2.5.2. Producer Accuracy....................................................................................................43
4.2.5.3. User Accuracy...........................................................................................................43
4.2.5.3. Kappa coefficient......................................................................................................43
4.2. Land use and landcover Dynamics..................................................................................45
4.2.6 Change analysis and prediction modeling.....................................................................51
4.2.5.1 Change analysis..........................................................................................................51
4.2.7.2 Markov Chain Model Analysis..................................................................................54
4.2.7.3 Transition Potential Modeling....................................................................................54
Figure 4-11 Ambo Town Transition from all LULC Classes to Settlement area (2010-2019......55
viii
Ambo Town Change Map (Transition from all LULC Classes to Settlement area) from 2010-
2019 Shrub Land (1600Ha), Agriculture land (700Ha), and Forest land is (200Ha)....................55
4.2.7.4 . Spatial trend of change.............................................................................................56
4.3. Suitability analysis..............................................................................................................57
4.3.1 Factors and Constraints.................................................................................................57
4.8 . Summary of results............................................................................................................61
Discussion......................................................................................................................................64
CHAPTER FIVE...........................................................................................................................65
5. CONCLUSSION AND RECCOMMENDATION................................................................65
5.1INTRODUCTION................................................................................................................65
5.2. Conclusion..........................................................................................................................67
5.3. Recommendations...............................................................................................................68
5.3.1 Recommendations for Future similar Research Work..................................................68
5.3.2 Recommendations for whom it concern........................................................................69
REFERENCE................................................................................................................................71
List of figure______________________________________________________________page
Figure 1: Location of the study area..............................................................................................18
Figure 2:Flow Chart for LULCD classification.............................................................................23
Figure 3: Flow chart for modeling LULCD with CA-Markov......................................................32
Figure 4:Ambo Town Classified LULC of 1990...........................................................................40
Figure 8: Markovian Conditional Probability of being class Built up area (predict 2030)...........55
Figure 9: Ambo Town LULC Classes of three decades and predicted 2030................................55
Figure 10:Ambo Town LULC Change (Km2)...............................................................................56
ix
List of Table_______________________________________________________________page
Table 1: Sources: Laotian Ministry of Agriculture and Forestry guidelie (Thongphanh

etal.2006).......................................................................................................................................19
Table 2:Multi-temporal Landsat Images was used........................................................................20
Table 3:Materials and software used.............................................................................................21
Table 4:Band characteristics of Landsat Image.............................................................................27
Table 5:Basic band combinations for Landsat Image....................................................................27
Table 6:Algorithms for accuracy assessment, Ambo town...........................................................31
Table 7:Transition probability matrix for the projection of 2020 land use and land cover...........33
Table 8: Kappa statistic of Map comparison.................................................................................36
Table 9:Strength of Agreement for Kappa Statistic (Mahmoud & Divigalpitiya, 2017)..............36
Table 10:LULC Class Distributions of Ambo town in 1990.........................................................39
Table 11:LULC Class Distributions of Ambo tow in 2000...........................................................39
Table 14:Error matrix of the 1990 LULC classification of Ambo Town (in pixels)....................44
Table 18:AmboTown Gain and losses of LULC Classes for three decades in table form............46
Table 19:Ambo Town Net change by category of LULC Classes for three decades in table form.
.......................................................................................................................................................46
x
List of Acronyms/ Abbreviation
CA-MARKOV-Cellular Automata Markov Model
CSA- Central Statistical Agency
DEM - Digital Elevation Model
ERDAS -Earth Resources Data Analysis System
ETM+ -Enhanced Thematic Mapper /plus
ETM-Earth Trends Modeler
FAO- Food and Agriculture Organization
GCP- Ground Control Points
GIS - Geographical Information System
GPS- Geographic Positioning System
KIA-kappa Index Agreement
LCM - Land Change Modeler
LPG- Level Product System
LULCC- Land Use and Land Cover Change
LULC- Land Use and Land Cover
MCE - Multi-Criteria Evaluation
MODIS-Moderate Resolution Imaging Spectrometer
xi
MOSAIC -Multimodality Operational Site Analysis and Intelligent Change-detection.
MSS- Multi-Spectral Scanner
NDVI -Normalized Difference Vegetation Index
ONRS- Oromiya National Regional State
OLI-Operational Land Imager
RMSE- Root Mean Square Error
ROC -Relative Operating Characteristic
SPOT- Système Pour l'Observation de la Terre (French)
TM -Thematic Mapper
TIRS- Thermal Infrared Sensor
UN- United Nations
UTM -Universal Trans Mercator
WGS -World Geodetic System
WLC-weighted linear combination
xii
xiii
CHAPTER ONE
1. INTRODUCTION
1.1. Background
The land is a complex and dynamic factor that consists of geology, topography, hydrology, soil
and microclimate, and a community of plants and animals that are continually interacting under
the influence of climate and people activities(Lu & Weng, 2007). The land is one of the major
most important resources which consist of water, soil, associated with plant and animals
involving the total ecosystem. Land use is subject to the natural environment and intervention of
human activities, thus understanding the driving factors behind the land-use change (LUC) is
crucial for the land use planning and management of key land surface of the earth’s ecosystems
services and functions ((Leemhuis et al., 2017).
Land cover refers to the physical characteristics of Earth's surface, captured in the distribution of
vegetation, water, soil, every applied on an undulation of the ground surface of the earth’s and
other physical features. Land use refers to how the land has been used by humans and their
habitats (such as agriculture, settlements, industry, farms, grazing land, etc.). The surrounding
problems are related to land use and land cover dynamics, Due to climate changes, soil
deterioration, and Land Use and Land Cover Changes (LULCC), land degradation risk has
become one of the most important ecological issues at the global level. And for sustainable
development, it is mandatory to know the interaction among human activities with their
environment and to monitor the dynamics through time.
Land use and land cover (LULC) change has been regarded as an important factor influencing
climate change and environmental conditions(Pickett et al., 2017), and has a close relationship to
population migration and economic conditions(Koranteng et al.,2016).
Remote sensing has an important contribution to making and documenting the actual change in
land use and land cover on global, local, and regional scales. Remote sensing (RS) is with the
increase in growth of population; pressure is exerted on limited natural resources of a country
and contributes to the changes in land cover. Remote sensing and GIS technique are also applied
in this study to detect. Methodology for land use and land cover change (LULCC) detection.
Remote Sensing and GIS are the extent of Land Use and Land Cover changes of the study
area and to recently small scale aerial photographs and satellite images associated with a
specific piece of land. The collection of remotely sensed data facilitates the synoptic analyses of
Earth system function, patterning, and change at local, regional and global scales over time; such
data also provide an important link between intensive, localized ecological research and regional,
national, and international conservation and management of biological diversity (Green et al.,
2000). Remote sensing has become a major data source for mapping and monitoring LULC
1
dynamic change over time(Hansen & Loveland, 2012) because it can capture land surface
information at the time when satellites pass through.
Models are used in a variety of the undulation of earth's surface are the main issue of land
dynamic science, to better understand the dynamics of systems, to develop hypotheses that can
be to make predictions, or evaluate scenarios for use in assessment activities. Therefore, cellular
automata will be used appropriately for scaling up to continental, global-local, and regional
depending on the scales. Based on the model analysis and the simulation of land use spatial
patterns, the driving factors of LULC change can be revealed, clarifying the rate of land use and
making possible multiple LULC scenarios to predict future land use demand. In Ethiopia, be in
charge of a farm mainly resolution in this basic supply of materials in traditional ways without
any logical organizations of different types of land according to their agricultural potential
(Ayele, 2017) This leads to further performance of agricultural sectors in particular and the
whole economy in general. However, continual agricultural growth remains a necessity not an
option for most developing countries like Ethiopia and the growth must accomplish by effort on
a sustainable basis, not a failure of the underplaying natural resource base (Ayele, 2017). Human
presence on the Earth and its modification of the landscape has had a profound effect on the
natural environment. According to (Behailu, 2010), these anthropogenic influences on shifting an
arrangement of land-use change are a primary component of many current environmental
concerns as land use land cover dynamic is gaining recognition as a key driver of environmental
dynamic. (Tizora, 2018) demonstrated that land use denotes the human employment of land
which includes settlement, cultivation, pasture, rangeland, recreation, and so on, whereas land
cover denotes the physical state of the land. That means during the recovery period of terrestrial
ecosystems after a flooding event, there are dynamics in the biotic composition and spatial
configuration of habitat patches. The destruction and subsequent construction the new patches.
Where result in an observable pattern in the land use and land cover dynamics over time.
To understand how land uses land cover change detection effects and act in such a way as to
have an effect on another with the earth systems accurate information is needed on what types of
changes were occurred, and when they occurred, and rates of at which dynamics occur ( Ayele,
2017). land cover change of the study area was resulted due to commercial of the
farmland expansion, forest deforestation, the population has grown and major problems of this
area informal settlement expansion, forest deforestation and unsustainable their temporal demand
of agricultural land and as a result, the natural resources kind of data collection methods are
useful for identifying land use land cover. Thus environmental problems are often related to
Land use Land cover dynamics. The land use land covers the dynamic pattern of a region is the
relevance of natural factors and their make practical and effective use of by a man in time and
space. Human activities such as population growth increased urbanization, and rapid economic
development has markedly modiﬁed the Earth’s surface processes, which causes changes in
environmental attributes at a regional and global level. Models of land use land cover dynamics
are powerful a device that can be used to understand and analyze the crucial linkage between
2
socio-economic a series of actions associated with agricultural activities, land development, and
natural resource management strategies and the ways that these changes affect the structure and
function of ecosystems. For that reason, satellite images can often be used to detect land-use
dynamics through the ability to notice significant details of the biophysical characteristics of the
land. Spatial models are needed for understanding reality and comprise a temporal dimension as
well as the particular on the earth's surface. Models are almost certainly the most concise and
useful way to understand spatial change detection. Cellular Automata (CA)-Markov makes
available for use a powerful tool for the dynamic modeling of land use land cover dynamic and
are a common methodology used to consider spatial interactions. They have implemented land
use land cover dynamic models that can simulate multiple land use land cover dynamic types.
Thus, Cellular Automata are composed of five elements as described below: Cell Space:
The cell space is composed of an individual cell. The community within a Town: Each cell has
two very near to another in one-dimensional cellular automata, whereas in two-
dimensional cellular automata model there are two ways to define it. This current proposal is
aimed at an option the spatial evolution concept embedded in CA (Cellular Automata) and
applies it to land use land cover dynamic in West Showa Zone. A time-series of multi-scale and
multi-temporal (including historical) satellite imagery will be used to describe characteristics of
land use and land cover dynamic trends over the period from 1990 to 2010. Socio-economic and
biophysical driving forces of observed dynamic are permanent basis through a network of work
jointly on activity partners and agencies willing to share resources and strongly wanting to do to
utilize developed techniques and model results. This proposal examines the use
of GIS and Remote Sensing in Mapping Land Use and land cover maps and monitoring
changes at regular intervals of time is limited in respect to its extent of coverage and expensive;
which means, it is also a major contributing factor to the observed classification each class of
that area. In recent years, much attention has been paid to analyzing the driving factors involved
in such processes and turn, the impacts of LULCC. Models and scenario analyses have been
recognized as powerful tools for understanding land-use patterns and complex driving
mechanisms (Li et al., 2016). None of this study has focused on a town level of understanding
the mechanisms of land-use changes for future land-use proposals. Furthermore, applications of
various spatial models for proposing future land-use changes are scant in this study area. Hence,
it is vital to simulate the processes involved in the spatial pattern of land uses which eventually
help in the decision-making of land use planning and management (Deng et al., 2010). In the
study present here, the aim is to explore the dynamics and future spatial patterns of land-use
changes from 2020 to 2030 using the Dynamics of Land System (DLS) model under two
scenarios: The BAU (Business-as-Usual) scenario was set for future land transitions based on
historical and recent socioeconomic trends, e.g., economic and population growth without any
new sustainable environmental or economic policies. The BAU scenario was designed mainly
based on the assumption of a continuation of LULC conversion rates of the past 30 years in the
studied landscape (Kindu et al., 2018). A process in which changes are made to the formal
“rules of the game” – including laws, regulations, and institutions – to address a problem or
achieve a goal such as economic growth, environmental protection, or poverty alleviation.
3
Generally, this research aimed to analyze and model the long term land use and land cover
dynamic (from 1990 to 2010) in West Showa Zone and Ambo town by integrating remote
sensing, geographical information system (GIS), and Modeling tools and provided quantitative
analysis of land use land cover dynamic information in the area.
1.2. Statement of the problem

According to Foody (2001) land use and land cover (LULC), change is a global change that is
induced by natural and anthropogenic factors, and recently the issue of the driver and has
distinguished implications for many of the worldwide policy issues. These changes in LULC
reflect the population growth, land consumption rate, and climate. In underdeveloped nations,
land-use activities are not well organized, but instead happened in a disorganized way, for
instance, unplanned built-up expansion, shifting agricultural activities, and mining activities are
the forerunners to land degradation ( Mukhopadhyay et al. 2017) and flooding (Ackom et
al.2020). Lu et al. (2020) assessed the expansion in developing countries megacities with the
help of satellite imagery and their results showed that excessive development of megacities has
resulted in urban environmental degradation in megacities in underdeveloped nations. The LULC
change assessment in Ghana for the period 1991–2016 also established a continuous increase in
the settlement at the expense of open forest and closed forest (Ackom et al. 2020).
In Ethiopia, researchers have studied land use and land cover change at the local level, mostly on
undulation of the ground scale (Kindu, 2013) indicating a significant reduction of natural forest
cover and grasslands, in Ethiopian highlands. The study conducted in Hawassa town(Ayele,
2017) indicated that The uncontrolled growth of urban development has adversely affected
Hawassa towns’ ecosystem which influences to indirectly reflects on weather parameters and
eventually leads to local climate modification. However, Ambo Town has experienced human-
induced problems including a high rate of urbanization (settlements), informal house expansion,
mining, micro-industrial development, logging, and road expansion are some of the contributing
factors to land cover changes. A diverse array of land use and land cover dynamic models have
been developing, implement globally as important tools in making land-use decisions, yet the
implementation of such models is limit in West Showa in general and Ambo Town in particular.
Out of many, one key problem of developing a town like Ambo town is the haphazard urban
development mostly without proper planning strategies. Urban growth affects the ecology of the
town by modifying local climate conditions, eliminating and fragmenting native habitats, and
generating anthropogenic pollutants. From these points of view, the major part of Ambo town is
unplanned and less designed. The main reasons for this situation are the lack of proper land
distribution by the land administrative office of the town for the residential in a time. A case of
interrelated factors has driven the emergence of informal settlements: population growth; rural-
urban migration; lack of affordable housing; weak governance (particularly in policy, planning,
and urban management). Informal expansion of Ambo town has also resulted in depletion of
natural resources, human labor exploitation, financial extravagance, time consumption, an
4
unplanned town with a lack of attractive fullness of town structure, and deterioration of the
environment. Therefore, geographic information systems and remote sensing analysis enable
sustainable management of LULC change planning, wise decision making, monitoring of urban
expansion and development (Ayele, 2017).
Because of this, the current research project is initiated to identify the LULC dynamics and
simulate future LULC scenarios in Ambo Town over 30 years using a cellular automata model,
remote sensing, and GIS techniques. Other researcher’s work demonstrated the case by using
GIS and RS technologies for urban LULC dynamic analysis and change estimation of Ambo
Town were not conducted yet.
1.3 Research questions were:

1. What are the driving factors of LULC development and dynamics in the Ambo town?
2. What is the land use land cover dynamics in the past three decades?
3. How to model future land-use dynamics under two different scenarios?
1.4 Objective of the Study

The general objective of this study is to monitor land use land cover dynamics of Ambo town
and model its future development under different scenarios.
1.4.1. The specific objectives of the proposal were:

1. To identify major deriving forces of land cover changes of Ambo town between 1990 and
2020.
2. To investigate land use and land cover dynamics in the past three decades.
3. to model future land-use dynamics under two different scenarios.
1.5 Significance of the research

The modeling of the land use land cover dynamic within the study area has a scientific and
improvement of the importance for the future. The researcher feels sure of truth this proposal
will provide baseline information on issues of land use land cover dynamics concerning
vegetation cover dynamics in the study area. Such information is vital for comparing the past and
present conditions and predicts the future trends of the land use land cover dynamics and
expanding such methods of protecting the soil degradation and expanding such techniques in
Ambo town and its around. Thus, a community of Ambo town benefits primarily. The researcher
believes this project will provide baseline information on issues of LULC dynamics concerning
vegetation cover dynamics in the study area. Such information is vital for comparing the past and
present condition and predicts the future trends of the land use land cover change detection and
expanding such system method of protecting are the one way this case soil degradation and
expanding such techniques to others Town
5
CHAPTER TWO
2. REVIEW OF RELATED LITERATURE
2.1. The concept and definition of land use land cover

Many books use land cover change interchangeably, although they are different. Land use is
defined to be any physical and biological or chemical change attributable to management, which
may include conversion of grazing to cropping, change in fertilizer use, drainage improvement,
installation and use of irrigation and plantation building farm dams, pollution, and land
degradation, vegetation removal, change fire regimes, the spread of weeds and exotic species,
and conversion to non-agricultural uses (Ayele, 2017).
Land use of land covers dynamics are important elements used to solve the global problem of
environmental dynamics (Firdaus, 2014). Human actions are apparent in the ground environment
at unprecedented rates, magnitudes, and spatial scales. Land cover dynamic stemming from
human land uses represents a major source and a major element of global environmental
dynamics (Karsidi, 2004). Information on land use land cover and possibilities for their optimal
use is essential for the planning, selection, and implementation of land use schemes to meet
become greater in size demands for basic human needs and welfare. These dynamics have
beneficial or detrimental impacts, the latter being the principal. The Causes of global concern as
they impact human well-being and safety. For cite as example, Deforestation and agricultural
intensification are so pervasive when they aggregate globally and significantly affect key aspects
of Earth Systems (Behailu, 2010).
Land cover is the tangible material that exists at the surface of the earth. Land covers include all
infrastructure, grass, asphalt, trees, bare ground, water, etc. Earth conceal it is the action of
express something used by ecologist Frederick Edward Clements that has its closest modern
equivalent being vegetation. Land use land cover dynamic are caused by natural and human
drivers, such as the construction of human settlements,
Government policies, climate become different, or other biophysical drivers (Kindu, 2013). so
as the side of the world processes related to industrial constructions, the advancement of
transport infrastructures, urbanization, and other built-up areas, are severely influencing the
Environment, and are often to improve them all the visible feature of an area of land in an
unsustainable way (Behailu, 2010). In many cases land-use activities go hand in hand with
substantial modifications of the physical and biological cover of the Earth’s surface, resulting in
indirect consequence of action on energy and matter fluxes between the ground ecosystems and
the atmosphere. For cite as example, the process of the forest to a plant cultivated on a large
scale for food is changing climate-relevant surface parameters (e.g. albedo) as well as vapor
transpiration processes and carbon flows. In turn, the human land-use resolution reached after
consideration is also influenced by environmental processes. Changing temperature and
6
Precipitation patterns for example are the crucial outcomes of something for the location and
intensity of agriculture. Due to these close linkages, processes of land use and related land cover
dynamics should be considered as important components in the construction of Earth System
models(Han et al., 2015). The all the visible features of areas of land concept used to map and
assess land use land cover dynamic allows us to explain relationships between Land-Use
practices and Land-Cover patterns, and considers Land-Cover dynamic as driven largely by
Land-Use Types. For different-scale land use cover dynamic investigations, the landscape and on
the undulation Methodology is used based on remote sensing data of different spatial and
temporal resolution, as well as by convention thematic maps and in-field data, to explain
relationships between current Land-Use practices and land-Cover patterns(Behailu, 2010). At
Present-day, all the visible features of an area of land are territorially defined units of the land
surface, characterized by a structurally organized combination of natural and economic
components whose close Interactions give birth to the present-day landscape territorial system.
The type previously mentioned approach provides a base for the perception of the world as a
system of interrelated territorial samples with different environmental situations. In an example
of responding to this issue, a hierarchical landscape Classification scheme is proposed for scale-
dependent all the visible features of an area of land applications.
2.2. Land use land cover dynamic

Land use land covers dynamic and urban expansion patterns for the cities and on the undulation
of the ground earth’s surface has some event of the pattern. Quantitative and qualitative analysis
of Land use Land cover change detection is necessary to evaluate the influences of change in the
pattern of the natural vegetation on the earth's environment. Land use land covers dynamic and
research are crucial to deal with the identification, qualitative description, and parameterization
of factors that drive dynamic in LULCD (land use and land cover dynamic), as well as the
integration of their consequences and feedbacks. However, one of the major challenges in Land
use land cover dynamic analysis is to link the behavior of people to biophysical information in
the appropriate spatial and temporal scales.
But, it is argued that land use land covers dynamic a general direction in which something is
developing can be easily accessed and linked to population data if the unit of analysis is the
national, regional, district, or municipal level (Ayele, 2017) dynamic in land cover by land use
does not necessarily imply a degradation of the land. Even though many shifting land-use
patterns are driven by a variety of social causes, outcome in a land cover dynamic that affects
biodiversity, water and radiation budgets, trace gas emissions, and known a series of actions that
come together to affect climate and biosphere (Ayele, 2017).
For this reason, to use land optimally, it is not only necessary to have the information on existing
land use land cover but also the capability to monitor the dynamics of land use resulting out of
both changing demands of an instance of increasing population and forces of nature acting to
shape the all the visible features of an area of land.
7
2.3. Tools use for urban Land Use Land Cover change
Remote Sensing, GIS, and IDRISI are now providing new tools for advanced ecosystem
management. With the advancement of technology, reduction in data cost, availability of historic
spatiotemporal data and high-resolution satellite images used for urban planning, geography,
earth science, transportation planning, environmental planning, disaster management, urban
expansion change analysis, and modeling.
2.3.1 Remote Sensing

Remote Sensing is the science and the expression of acquiring information about an object, area,
or phenomenon through the analysis of data acquired by a device that is not in contact with an
object, area, or phenomenon under investigation (Eastman, 2003). It provides a large variety and
amount of data about the earth's surface for detailed analysis and dynamic with the help of
various spaceborne and airborne sensors. It occurring now with powerful capabilities for the
ability to understand something and managing earth resources. Remote Sensing has been proven
to be a very useful tool for Land use land cover dynamics.
Dynamic and monitoring involve the use of several multi-date images to evaluate the differences
in LULC due to various environmental conditions and human actions between the acquisition
dates of images(SIraj et al., 2018). Successful use of satellite Remote Sensing for land use land
cover dynamic depends upon an adequate understanding of landscape features, imaging systems,
and methodology employed concerning the aim of the analysis (Udagepola et al., 2006).
In the same direction the availability of historical Remote Sensing data, the action in data cost,
and increased resolution from satellite platforms, Remote Sensing technology appears poised to
make an even greater impact on monitoring land cover and land-use change (Behailu, 2010). In
general, the dynamic of LULC involves the interpretation and analysis of multi-temporal and
multi-source satellite images to identify temporal phenomenon or changes through a certain
period. Remote Sensing data are the primary source for change detection in recent decades and
have made a greater impact on different planning agencies and land management initiatives
(Firdaus, 2014).
Remotely sensed satellite images provide valuable datasets that can be used to analyze, evaluate,
and monitor changes in ecosystems through change detection. One of the major difficulties of
any satellite image analysis is how to accurately compensate for atmospheric effects. Several
studies have investigated the ability of satellite imagery, including Landsat TM, ETM+, and OIL
to perform dynamic analysis. The most commonly used remote sensing data for the extraction of
earth surface features for the classification of LULC are Landsat, SPOT, Radar, Aerial
Photography, IKONOS, MODIS, SENTINEL AVHRR, etc.
According to Lambin (2004), the definition of Remote Sensing (RS) is as follows: “Remote
sensing is the science and art of obtaining information about the Earth's surface through the
analysis of data acquired by a device which is at a distance from the surface”. Remote sensing is
8
an essential tool of land change science because it facilitates observations across larger extents of
Earth’s surface than is possible by ground-based observations by using cameras, multispectral
scanners.
2.3.2 Geographic Information System (GIS)

Geographic Information Systems (GIS) and remote sensing understand that new a device for
advanced ecosystem management. The collection of remotely sensed data facilitates the general
view analyses of earth-system function, patterning, and change at local, regional, and global
scales over time. Such data also provide a necessary link between intensive, localized ecological
research and the international, national, regional conservation, and management of biological
diversity (Behailu, 2010).
GIS integrates hardware, software, and data for capturing, managing, analyzing, and displaying
all forms of geographically referenced information (Maguire (1991). GIS also allows the
integration of these data sets for deriving meaningful information and outputting the information
derivatives in map format or tabular format. GIS support any operation on geographic
information: acquisition, editing, manipulation, analysis, modeling, visualization, publication,
and storage. GIS has four basic subsystems: input, storage, analysis, and output.
2.3.3 IDRISI
IDRISI is the industry leader in raster, covering the full spectrum of GIS and remote sensing
Special facilities are included for environmental monitoring and natural resource management,
including land change modeling and time series analysis, multi-criteria and multi-objective
decision support, uncertainty and risk analysis, simulation modeling, surface interpolation and
statistical characterization (IDRISI Selva Manual version 17.01). One advantage of using IDRISI
for land-use change modeling is that in one software package you have: 1) the data, 2) the
simulation model, and 3) the statistical techniques of goodness-of-fit (Pontius 2002).
2.4. Conceptual Frameworks of Urban land use land cover change
2.4.1. Factors of urbanization and urban expansion

There are several factors of urbanization, such as Economic Development, the innovation of
technology, the industrial revolution, emergence of large manufacturing factors, job
opportunities, availability of infrastructures, and migration. Generally, the causes of urbanization
and urban expansion are groups into three major classes.
2.4.1.1. Demographic factors

Demographic effects include rural to urban migration and natural population growth in the city,
the level of urbanization, and the rank of the city/town in the country’s urban hierarchy. Natural
population growth is a major element in urban growth for all countries, and migration of rural to
9
urban contributes fast growth of urban population in many developing countries (Gugler, 1996).
2.4.1.2 Economic factors
Economic effects include the level of economic development, the difference in household
incomes, exposure to globalization, the level of foreign direct investment, the degree of
employment, the level of financial markets, the level and effectiveness of property taxation, and
the presence of high inflation and acute shortage of housing (Angel et al 2005). Natural
/environment factors Natural /environment effects include those of climate, slope, mountain
barriers, and the existence of drillable water aquifers. Also the minor causes such as
Redevelopment and rebuilt up of inner cities again cause displacement of citizens (Cernea,
1995). 2.3.2 Forms of Urban Expansion take place in different forms. The major forms are the
positive effects (orderly) form of urban expansion properly laid out in simple geometric forms,
of urban expansion such as the center of the market area, center for production and distribution
of goods and services, an opportunity for access to employment, economic development, full
facilities, and technology development. The other forms are negative effects (disorderly) of
urban expansion are loss of prime agricultural farmland, displacement of farm communities,
solid waste disposal and land degradation, enclosing surrounding rural land to the urban territory,
overexploitation of natural resources, and conflict.
2.5. Image Classification

Several regions of the electromagnetic spectrum classification are the process of sorting pixels
into a finite number of individual classes, or categories of data, based on their data file values.
The supposition that a pixel satisfies be the case set of criteria; the pixel is assigned to the class
that corresponds to that criterion. This natural series of changes is also referred to as image
segmentation. Depending on the type of information we want to extract from the original data,
classes might be associated with known features on the ground. For instance of a classified
image is a land use land cover map, showing vegetation, built-up area, bare land, pasture,
settlements, etc.(Han et al., 2015) according to a land use land cover classification system which
can effectively employ orbital and high altitude remote sensor data should meet the about to be
mentioned criteria(Kindu, 2013) :
The quality is an intense possible level of interpretation accuracy in the identification of land use
and land cover dynamics. Cover categories from remote sensor data must be at least 85 percent.
The correct in all details of interpretation for the several categories must be about equal.
Do the same thing again results must be obtainable from one interpreter to another and from one
time of sensing to another.
The classification system must be applicable over extensive areas.
10
The categorization should be made possible for vegetation and other types of land cover to be
used as Surrogates for activity.
The classification system should be a purpose concerning use with remote sensor data obtained
at different times of the year.
Planned result use of the subordinate category that can be acquired from ground surveys or the
use of a larger-scale or enhanced remote sensor data should be possible.
Aggregation of categories should be possible.
The quality of being similar to future land use data should be possible.
Multiple uses of land must be recognized when possible.
2.5.1. Pre-processing of Satellite Data

Uncooked digital images usually have some geometric distortions as a result of variations in the
attitude, altitude atmospheric refraction, relief displacement, Earth curvature, and nonlinearity in
the sweep of a sensor’s IFOV (Muzein, 2006). These a mistake must be corrected to ensure the
accuracy of the final results. Widely, two things are having common characteristics of data
correction: geometric and radiometric
Radiometric correction: is cooked thoroughly to reduce or correct a mistake in the digital

numbers of images.
The process becomes better the interpretability and distinctive attribute of remotely
sensed data. Radiometric to allow and a change that rectifies an error are in particular important
when comparing data sets over multiple periods. Radiometric correction is particular of the place
where someone lives variations in the pixel intensities that are not caused by the object or scene
being scanned. These slight differences in conditions include malfunctioning of the detectors,
topographic effects, and atmospheric effects.
Geometric correction is undertaken to keep away geometric distortions process a distorted image
and is accomplished by an effort to initiate the relationship between the image coordinate system
and the geographic coordinate system using to allow data of the sensor, measured data of
11
position and attitude, atmospheric condition, ground control points as well as exist on the earth
surface, etc.
Geometric correction and organization are situated errors in the relative positions of pixels.
These errors are induced by sensor viewing geometry and terrain variations(SIraj et al., 2018).
2.5.2. Types of Image Classification

Land cover maps are commonly created from remotely sensed data through unsupervised or
supervised classification techniques (Haque & Basak, 2017).
2.5.2.1. Unsupervised Classification

The unsupervised classification comes near is a facility classification method that creates a
thematic raster layer from a remotely sensed image by letting the software identifies statistical a
repeated decorative design in the data having the use of using any ground truth data(Muzein,
2006). The spectral classes obtained from the unsupervised classification are based belonging on
the maximum likelihood classification (MLC), parallelepiped, and minimum distance to mean
classification may be applied to get the best classification technique ((Golmehr, 2009).
The Unsupervised approach does have its advantages. For the reason that there is no dependence
on user understanding that training samples (which might not represent “pure” examples of the
class feature desired and which would, therefore, bias the results), the algorithmic grouping of
pixels is often more likely to produce statistically valid results. Following as a consequence,
many users of remotely sensed data have switched to allowing software to produce homogenous
groupings via unsupervised classification techniques and then use the locations of training data
to help label the groups(Haque & Basak, 2017).
2.5.2.2. Supervised Classification

Handing over something the image analyst supervises the pixel categorization process by
specifying, to the computer algorithm, numerical descriptors of various land cover change
detection types present in the image. Training samples that express the typical spectral pattern of
land use land cover classes are defined. Pixels in the image are similar numerically to the
training samples and are labeled to land cover classes that have a similar integral part of a
logarithm. All the classification techniques like the maximum likelihood classification (MLC),
parallelepiped and minimum distance to mean classification may be applied to get the best
classification technique (Behailu, 2010). The maximum likelihood classification assumes that the
statistics for each class in each band are normally distributed and calculates the probability that a
given pixel belongs to a specific class. Each pixel is assigned to the class that has the highest
probability (that is, the maximum likelihood). Maximum Likelihood is one of the most popular
supervised classification methods used with remote sensing image data. The Maximum
Likelihood classification method is well known for the analysis of satellite images(Lu & Weng,
12
2007). So far, satellite image interpretation using the maximum likelihood approach was mostly
applied for land cover classification and monitoring of land-use changes (Shalaby & Tateishi,
2007)showing overall high accuracies (mostly over 80%). MLC classification is based on a
parametric approach that involves the assumption of the selected classes of signature in the
normal distribution(Al-Ahmadi & Hames, 2009). Disadvantages: The SVM algorithm is not
suitable for large data sets. SVM does not perform very well when the data set has more noise
i.e. target classes are overlapping. In cases where several features for each data point exceed the
number of a training data sample, the (SVM) will underperform. In supervised classification, the
serving to establish the identity of the owner and location of certain representative patches of the
land cover types present in a landscape need to be identified before classification. Occurring at
the beginning field input is normally required for acceptable map accuracy.
2.5.3. Image Enhancement

Image enhancement is the process of making an image more interpretable for a particular
application (Han et al., 2015). Enhancement combining substances of great significance features
of raw, remotely sensed data more interpretable to the human eye. Enhancement of a scientific
official way of doing something is often used instead of classification techniques for feature
extraction—studying and locating areas and objects on the ground and deriving useful
information from images. The techniques to be used in image improve the quality depend upon
the type of data, objective of the study, belief that something will happen, and background of the
analyst (Ayele, 2017). Different techniques are used in image enhancement including principal
components analysis, transformations, and vegetation indices (Haque & Basak, 2017).
2.5.4. Dynamic Methods

Digital dynamic surround and have the quantification of temporal phenomena from multi-date
imagery that is most commonly come to possess by satellite-based multi-spectral sensors. In
General, dynamic involves the application of multi-temporal datasets to Concerned with

analyzing the temporal effects of the phenomena (Lu & Weng, 2007).
Dynamic methods have been grouped generally into image algebra, transformation, and
classification. The process of classifying categories contains as part of a whole post-
classification comparison, spectral temporal combined analysis, expectation-maximization
algorithm dynamic, unsupervised dynamic, and hybrid dynamic and ANN (Leemhuis et al.,
2017). This category has the advantage of showing both changes no change as well as ‘from to’
information.
2.6. Post Classification Dynamic

This method compares two independently produced classified land use cover maps of two
different dates. Therefore, it minimizes the problem of normalizing for atmospheric and sensor
differences between two dates and can indicate the nature of change. It was found to be an
13
accurate procedure for land use/cover change detection provided that the two land-use cover
maps have been accurately produced (Muzein, 2006).
Post classification analysis is included as a necessary part independently produced spectral

classification results from each end of the time interval of holding the attention, followed by a
pixel by pixel or may be divided by may be divided comparison to detect changes in cover type.
In a series of actions of increase in the amount to the algorithms which are applied on the
classified images to determine those pixels with a change between the two dates, statistics can be
compiled to express the specific nature of changes between the two images(Muzein, 2006).
2.7. Accuracy Assessment

According to (Foody, 2001) in thematic mapping from remotely sensed data, to express a
concept correctly in all details is used typically to express the degree of ‘free from error’ of a
map. A thematic map is supported by something on a change made with a classification that may
be considered accurate if it supplies an impartial representation of the land cover of the region it
portrays. In the intrinsic nature of something, therefore, classification accuracy is typically taken
to mean the degree to which the derived image classification agrees with reality or conforms to
the ‘truth’. A in a specified place of reference pixel representing geographic points on the
classified image is dependent on for the accuracy assessment. Eliminate the possibility of bias
(Firdaus, 2014). According to (Behailu, 2010), if the information derived from remote sensing
data is to be used in some decision-making process, then some measure of its quality must be
known. The most common accuracy assessment elements include overall accuracy, producer’s
accuracy, user’s accuracy, and kappa coefficient (Lu & Weng, 2007).
2.8. Modeling LULC Change

The term models have been used in different contexts and several application areas. It is large in
an area defined as abstract qualities in art or approximation of reality achieved by simplification
of complex real-world relations to the point that they are understandable and analytically
Manage (Tizora, 2018), as cited in (Behailu, 2010).
In practical situations land use land cover studies models are used to will happen in the future
state of land use Patterns considering various physical and socio-economic elements.
The ability to forecast land use land cover change and, ultimately to predict the Consequences of
change, will depend on our ability to understand the past, current, and
Future drivers of land use land cover change. The political and social influence that contributes
to a result may have significant effects on future land use and cover.
A repeated decorative design of land use land cover dynamic, and land management is shaped by
the interaction of economic, environmental, social, political, and technological forces on local to
global scales.
14
An improved understanding of historical land use land cover patterns provides a means to
evaluate complex causes and responses to better project future trends of human
Activities and land use land cover change. Regarded as in the same category must understand the
primary modern and future drivers of land use and their interrelationship with the land the
process of managing decisions and assets policies to more advanced forecast based on present
trends of future land use and management decision Outcomes under a range of economic,
environmental, and social scenarios. This capacity to do something will allow better projections
and hopefully minimize negative impacts, especially as related to climate change. This type of
detailed examination of the elements will do something the integration of various disciplines
from the physical and social sciences.
Land use land cover change (LUCC) modeling is a rapidly growing scientific field because land-
use change is one of the most important ways that humans influence the environment. The issue
is so important that scientists have formed an international organization, called LUCC, which is
connected with the International Human Dimensions of Global Change Program and the
International Geosphere-Biosphere Program(Han et al. 2015) the prediction of land use dynamic
by simulation can facilitate to assess development impacts prepare land use plans and seek
optimal land-use patterns. It can estimate or predict the importance of specific human behavior
and land use policies. Land use dynamics are predicted following the independent spatial
variables that are generated from Standard GIS analysis tools ((Behailu, 2010).
2.8.1. Modeling LULC Dynamic with CA-Markov

Ulam and Von Neumann originally conceived Cellular Automata (CA) models in the 1940s to
provide a formal framework for investigating the behavior of complex, extended systems(Tizora,
2018). CA (Cellular Automaton) is a change detection, discrete space, and time systems. A
cellular automaton system has as an essential feature consists of a regular grid of cells, each of
which can be in one of a finite number of k possible states, updated synchronously in discrete
time steps according to a local, identical interaction rule. The condition of someone of a cell is
determined by the existing states of a surrounding neighborhood of cells (For & Agents, 2016) a
cellular automaton (CA) system have as an essential feature of a regular grid of cells, regarded
and identified separately of which can be in one of a finite number of possible states, updated
synchronously in discrete time steps according to a local interaction rule (Behailu, 2010).
A period such change most probable event for the typical CA (Cellular automaton) model
depend on the state of a cell, the state of its surrounding cells, the physical characteristics of the
cell (e.g., soil quality, terrain, hydrology, vegetation, and demographic characteristics), and the
weights associated with the neighborhood context of the cell (e.g., proximity to other villages
and the time since settlement). These heaviness and neighborhood conditions are resolute from
empirical analyses of LUCCD based on social survey data, the GIS database that represents the
resource quality of a site, and the spatial linkages between villages, land parcels, and other
critical landscape features (Behailu, 2010). Methodologically, there are three different ways that
15
the satellite time-series are used in the development of the CA models. First, a subset of the
satellite images will be used to build the baseline model within the CA context. A second distinct
subset of the satellite images will be used for the calibration process. A third set will be used for
validation and finally, the CA models were further tested by back-simulating the land use cover
change dynamic models of current conditions to a portion of time within bring together satellite
time series that were not used for the actual model development. This verification design can
only be accommodated through a rich and highly populated remotely sensed time series and a
longitudinal social survey. The bring together satellite and aerial photography time series affords
the ability to do something to create again initial conditions in advance of real historical events
that are hypothesized to have affected the pattern of land use cover dynamic. The model
performs with a homogeneous cell-based grid and the desired result itself through the transition
rules derived from a local spatiotemporal neighborhood. This combining substance the
CA( Cellular automaton) model purpose to simulate intricate and authority structures since more
unknown, extreme to measure spatiotemporal variables can be incorporated and examine in this
model(Yeh & Li, 1998). (Ismail, 2014)combined the multi-criteria evaluation (MCE) and GIS
into the CA model to define the transition rules in a visualized environment. Another critical
advantage in CA simulation is the ability of the model to incorporate proper weights to model the
alternative socioeconomic status in the model development (Tizora, 2018) With better computer
techniques, the CA model is also able to explore more complex human behavior through
defining different transition rules. However, the tension, between the easily local transition rule
in CA models and the complex, unpredicted sociopolitical dynamic in landscapes, remains weak.
The ability of most CA models to correlate socioeconomic factors with the development process
is still weak (Karsidi, 2004).
The CA model in general works by:
CA (Cellular automata) usually maintain similar frameworks regarding assembly, testing,

validation, and calibration. However, in this study, Markov-CA is will be to see dynamic
analysis use. It is a combination of CA (Cellular automaton), Markov chain; multi-criteria, and
multi-objective land allocation with land cover prediction procedures that add an element of
spatial contiguity as well as knowledge of the likely relating to space distribution of transitions to
Markov-chain analysis.
Simulating the present by a graph from the past using the image time series.
Checking the simulations via the remotely sensed time series of past conditions and through the
available collection of field monitoring.
Acceptable of the model to iterate to the year of choice in the future.
Note the similarity model outputs to an autoregressive time-series come near for occurring once
every year conditions(Cervantes-Godoy et al., 2014).
16
Multi-Criteria Evaluations
To arrange a specific objective, it is frequently the case that several criteria were needed to be
evaluated. Such a procedure is called Multi-Criteria Evaluation (Eastman, 2003); (Eastman,
2003) as cited in (SIraj et al., 2018).
Multi-criteria evaluation is most commonly courage by one of two conducted in a certain order.
The first includes a necessary part Boolean overlay by which all criteria are become smaller to
reasonable under the circumstances statements of suitability and then combined using one or
more logical operators such as intersection and union or. The second is known as a weighted
linear combination wherein continuous criteria (factors) are standardized to a common numeric
range, and then combined using a weighted average. The result is a without interruption mapping
of suitability that may then be masked by one or more Boolean a limitation to provide lodging
qualitative criteria, and finally, the threshold to yield a final decision (Tizora, 2018).
17
CHAPTER THREE
3. MATERIALS AND METHODOLOGY
3.1. Description of the Study Area
3.1.1. Location
Ambo town is Located in the West Shewa Zone of Oromia Region, 114 km west of A.
This town has a latitude and longitude of 8°59′N 37°51′E and an elevation of 2106
meters. Ambo is known for its mineral water, which is bottled outside of town; it is reportedly the
most popular brand in Ethiopia.
Figure 1: Location of the study area

3.1.2. Climate of the study area
Climate is constant around Ambo Town condition of the humidity, wind, atmospheric pressure
rainfall, temperature, and other meteorological elements in an area on the of the Earth's surface
for a long time. In easily understood terms climate is the average condition for about thirty years.
The climate in Ambo is temperate and Climate is the long term average of weather, typically
averaged over 30 years. Some meteorological variables are commonly measured. In winter,
there is much less rainfall than in summer. The average annual temperature is 18.0 °C in Ambo.
18
The rainfall here averages 1312 mm. Ambo Town is generally located among the high land areas
of the country where the rainfall varies from 1376- 2037mm, and the annual mean temperature
varies from 14.6 to 30.4 Celsius (Joshua O.et al;2010 ). The Ambo lies 2106m above sea level
the climate in Ambo is comfortably high temperature.
3.1.3. Population
The population size and their distribution vary from the history of early human settlements. High
population pressure existed in the middle altitude where human beings were settled first and in
low land areas, the distribution of the population was low and scattered. The total population of
the Town is 106,664 while 62513 are women and 44,351 are men. The populations of the
sampled four PAs are 27,653, where 16,247 are women and 11,406 are men(Ayele, 2017).
3.1.4 Vegetation
The study area is characterized by farmland, settlement land, bare land, forest land, and grazing
land which are made up of shrubs or small trees and are usually fairly open; grasses and other
small plants grow between the shrubs (personal observation). Vegetation is patchy and composed
of small to medium-sized shrub species dominated by some scattered tree species. The
vegetation description presented by (Berhanu, 2017) shows that the study area is characterized
by dry evergreen montane forest on the basis that the dry evergreen montane forest covered the
area between 1500 and 3000 m a.s.l. in the central and northern part of the country. Dry
Evergreen Montane Forest is a very complex vegetation type occurring in an altitudinal range of
1500-2700 m, with average annual temperature and rainfall of 14-25° C and 700-1100 mm,
respectively. The Ethiopian highlands contribute more than 50 % of the land area of Ethiopia
with Afromontane vegetation, of which dry montane forests form the largest part (Bekele, 1993).
3.1.5. Identification of Major Land use land cover classes exist in the study area
In study area there are four major land use land cover classes, those are Built-up area, Forest
land, Bare land, and Agricultural land. Types and definition of each land use land cover classes
are explained in Table 1: Sources: Laotian Ministry of Agriculture and Forestry guidelie
(Thongphanh etal.2006).
Land Cover Class Properties
Forest land Forest land everywhere provides carbon storage, temperature moderation, and
cleaner air, maintenance of productive soils, wildlife habitat, and a renewable source
of fuel and fiber. These benefits are local and regional, and even global. There are
many tangible products we use daily that come from forests.
Built_up area Urban areas and permanent residential areas of varied patterns, for example, cities,
towns, and villages, strip developments along highways, transportation, power, and
communications facilities.
Bare land Bare land in areas with no dominant vegetation cover.
Agricultural land Agricultural land is defined as the land area that is either arable, under permanent
farms, or permanent pastures.
19
3.2 Data Source
3.2.1 Satellite imageries

Table 2:Multi-temporal Landsat Images was used
Data types Sources of data Path/Row Spatial
Resolution(m) Attributes
Image LandsatTM 1990 168/054 30*30 USGS land use land
Landsat TM2000 168/054 30*30 cover dynamics
Landsat TM2010 168/054 30*30 classification
Landsat8 2020ETM+ 168/054 30*30
The Landsat TM1991 image was taken from International Livestock Research (ILRI), GIS
Centre, (2000, 2010) ETM+, and (2020) OLI data were downloaded from GLCF and USGS
website.
3.2.1 Data collection

To accomplish the specified objectives of this research study both primary and secondary data
were used. The primary sources of data include field data that is collected with the use of Global
Positioning Systems (GPS), selected key informants, natural resource management experts, and
other responsible bodies by using structured interviews. The secondary data source involves
related published and unpublished materials, for instance, Satellite image, Ethiopian Map
Agency (EMA) contributed large scale maps1:50,000, Ethiopian central statistics agencies
population data of study area, Ethiopian meteorological agencies would get the temperature and
rainfall data.
Ground–truth data were gathered on the field to be used for image classification and verification
of the satellite imagery. Focus group discussions (FGDs) and key informants interview (KII).
Although there are different techniques for the socioeconomic surveys based on the purpose of
the study, key informants' interviews and focus group discussion surveying techniques were used
in this study.
3.3. Land Use and Land Cover scenarios and future demands
Scenario analysis is important for projecting future land-use changes. Two scenarios were
designed for this study, namely the Business-as-Usual (BAU) scenario, and the Policy reform
(PRS) scenario. Based on existing LULC-related policies in Ethiopia, local demographic
information, and historical LULC dynamics of the studied landscape, future scenarios have been
defined to predict LULC demand for 2030. The BAU scenario was designed mainly based on the
assumption of a continuation of LULC conversion rates of the past 30 years in the studied
landscape (Kindu et al., 2018). Thus, before the demand calculations for 2030 in the BAU
20
scenario, the rate of change between 1973 and 2012 for each LULC type was determined using
the following formula(Kindu et al., 2018) :
1 A2
r =( )*1n ( )
t 1−t 2
A1
Where r = rate of changes for each LULC type, A1 and A2 = extent of each LULC type at time
t1 and t2, respectively.
Afterward, the trends of each LULC type during the last four decades were extrapolated using
the following formula:
rt
An =A O e
Where An = the area estimate for each LULC type in year n, A0 = area of the base year, t =
period that is the difference between year n and year 0; and r = average annual rate of change.
Policy reform and innovation are key drivers of Ambo town's remarkable economy and Now this
turn to the scenario where growth in healthcare benefits is limited and remaining government
expenditures means a sudden increase overall.
3.4. Methods of data analysis

After the data is collected from various sources, the data were analyzed by applying multi‐
criteria evaluation (MCE) by integrating with ArcGIS 10.5: ERDAS Imagine 2015, IDRISI
Selva, and MS Excel. Ground truth from the field and satellite images were used and analyzed by
using ArcGIS version 10.5, and ERDAS imagine 2015 software. ArcGIS was used to reclassify
and calculates the pixel values of all LULC classes and complement the display and preparation
of maps. ERDAS imagine was utilized for layer stacking of bands 1-7, (where bands 8 and above
are omitted from layer stack due to high reflectivity), radiometric calibration particularly
atmospheric correction, haze, and noise reduction will be performed before analysis, and Google
Earth was used to check the land use and land covers in the area before field observation. The
overall processes allowed the investigators to better enhance and improve the images for
classification and interpretation and the resulting sampling sizes for point data were 200 points
with a maximum sample size of 50 samples per class.
3.4.1. Materials and software

Based on types of research different materials and software are used to manipulate input data and
give output results. For analysis of modeling of land use and cover change in Ambo Town the
following materials and software use.
Table 3:Materials and software used

NO SOFTWARE/MATERIAL PURPOSE
21
1 ArcGIS (10.5) Data acquisition, editing, manipulation, analysis, modeling,
visualization, publication, and storage
2 ERDAS (2015) Digital image preprocessing, Classification image, map land use/ land
cover
3 IDRISI (17) Land change modeling and time series analysis, multi-criteria and
multi-objective decision support, uncertainty and risk analysis,
simulation modeling, surface interpolation, and statistical
characterization, compare and predict for the future
4 Google Earth Visualization,
5 Microsoft office (2016) Writing, chart preparing, graphs, and statistical analysis
6 Handle GPS Collect Ground Controls Points (GCPs)
7 Microsoft Excel Microsoft Excel is a spreadsheet developed by Microsoft for
Windows, macOS, Android and iOS. It features calculation, graphing
tools, pivot tables, and a macro programming language called Visual
Basic for Applications (VBA). Excel forms part of the Microsoft
Office suite of software.
To organize, and analyze the relevant data the following Materials and software were used.
ERDAS Imagine 2015: was used for image pre-processing, stacking single bands, supervised
maximum likelihood classification of land classes, and accuracy assessment of the classification
for image analysis and mapping. ArcGIS 10.5: was used for data analysis, management, and
spatial referencing, geo-referencing, and make a layout for final mapping. Moreover, the study
area delineation and clipping process was operated by ArcGIS software. It would also be used to
complement the display and processing of the data and IDRISI Selva software for using the land-
use change prediction/forecasting and IDRISI LCM(land change modeler) is a land-use planning
and decision support software tool that allows for analysis of past land cover change,
development of scenarios of future land.
3.4.2. LULC Classification

Images from different periods are used for the classification of land use land cover dynamics of
the study area. This multi-temporal raw satellite data were imported to Erdas Imagine 2015
image processing software. After this, land use land covers dynamic maps, and land cover
statistics were generated to compare the temporal dynamic of the study area for the past three
decades using ArcGIS 10.5 software. Finally, land cover maps of Ambo town were produced
based on 1990 Landsat TM, Landsat (2000, 2010) ETM+, and ( 2020) OLI data using supervised
and unsupervised classification. However, after supervised classification, some classes were
combined and the final land cover map of Ambo Town was sketched. The step by step approach
(methodology) takes to achieve the stated objectives of this project are the following; the satellite
image of the study area was classified using ERDAS Imagine2015. The existing analog
boundary map was converted into digital format through digitizing using ArcGIS 10.5
application software. The step by step approach (methodology) taken to achieve the stated
objectives of this project is the following, and for sustainable development, it is mandatory to
know the interaction of human activities with the environment and to monitor the dynamic.
22
Figure 2:Flow Chart for LULCD classification
3.5. Digital Image Processing

Satellite imageries of 1990, 2000, 2010, and 2020 were georeferenced to their actual latitude and
longitude by using ground control points (GCP) from a georeference Topo Map and a global
position system (GPS) points from fieldwork. This image was projected to universal Traverse
Mercator (UTM) zone 37 north and mask by the study area boundary shapefile to subset the area
from the imagery. Clouds in the 1990 TM and (2000, 2010) ETM+ and (2020) OLI data images
were masked out using ERDAS radiometric enhancement techniques by reducing haze.
Remotely sensed data are usually digital image data. Therefore data processing in remote
sensing is dominantly treated as digital image processing. Image enhancement, spatial and
geometric transformation, and/or data compression is normally required to generate a thematic
map or database.
In today's world of advanced technology, most remote sensing data are recorded in digital
format, virtually all image interpretation and analysis involves some element of digital
processing. When remote sensing data are available in digital format, digital image processing
and analysis may be performed using a computer The most of the common digital image
23
processing functions available in image analysis systems are categorized into the following four
categories. First, Preprocessing functions involve data analysis and extraction of information,
generally grouped as radiometric or geometric corrections. Radiometric corrections include
correcting the data for sensor irregularities and unwanted sensor or atmospheric noise. Geometric
corrections include correcting for geometric distortions due to sensor-Earth geometry variations.
Secondly, Image enhancement is to improve the appearance of the imagery to assist in visual
interpretation and analysis. Thirdly, Image transformations are operations similar in concept to
those for image enhancement. Fourthly, Image classification and analysis operations use to
digitally identify and classify pixels in the data.
3.5.1 Pre-Processing
Pre-processing operations sometimes referred to as image restoration and rectification, are
intended to correct for sensor- and platform-specific radiometric and geometric distortions of
data. Radiometric corrections may be necessary due to variations in scene illumination and
viewing geometry, atmospheric conditions, and sensor noise and response. The geometric
distortions may be due to several factors, including the perspective of the sensor optics; the
motion of the scanning system; the motion of the platform; the platform altitude, attitude, and
velocity; the terrain relief; and, the curvature and rotation of the Earth. Geometrically correcting
the original distorted image, a procedure called resampling is used to determine the digital values
to place in the new pixel locations of the corrected output image. There are three common
methods for resampling: nearest neighbor, bilinear interpolation, and cubic convolution. The first
nearest neighbor resampling uses the digital value from the pixel in the original image which is
nearest to the new pixel location in the corrected image. This is the simplest method and does not
alter the original values, but may result in some pixel values being duplicated while others are
lost. The second bilinear interpolation resampling takes a weighted average of four pixels in the
original image nearest to the new pixel location. The third one is Cubic convolution resampling
goes even further to calculate a distance weighted average of a block of sixteen pixels from the
original image which surrounds the new output pixel location.
3.5.2 Image Enhancement

Image enhancement is the process of making an image more interpretable for a particular
application. The enhancement makes important features of raw, remotely sensed data more
interpretable to the human eye. Enhancement techniques are used for feature extraction studying
and locating areas and objects on the ground and deriving useful information from images.
Enhancement techniques available with ERDAS IMAGINE are as follows: Radiometric
Enhancement is used for adjusting the brightness values of the image. Under Radiometric tools
following are used for adjusting brightness values of the image as Lookup table (LUT) stretch,
Rescale, Haze reduction, noise reduction, noise removal, etc. Spatial enhancement is used for
adjusting the quality of the image and deriving new information using spatial operation. The
spatial tool contains Re-project, Resample, Degrade, Convolution, etc. Spectral enhancement is
used for image transformation techniques working with the multispectral image. Spectral tools
24
contain layer stack, Principal component, Decorrelation stretch, Natural colors, RGB to HIS
(Red, Green Blue band to Intensity, Hue, saturation band), etc. Pan sharpening (Resolution
merge) is used for increasing the resolution of colors image using a higher-resolution
panchromatic image. Pan sharpening too contains Subtractive resolution merge, Wavelet
Resolution merge, High Pass Filter (HPF) resolution merge, Projective Resolution merge, etc.
3.5.3 Image Transformations

Digital Image Processing offers a limitless range of possible transformations on remotely sensed
data. One widely used image transform is the Normalized Difference Vegetation Index (NDVI)
which has been used to monitor vegetation conditions on continental and global scales using the
Advanced Very High-Resolution Radiometer (AVHRR) sensor onboard the NOAA series of
satellites. NDVI = (NIR - R) / (NIR + R) where NIR = Near-Infrared and R = Red Image
transformation techniques based on complex processing of the statistical characteristics of
multiband data sets can be used to reduce this data redundancy and the correlation between
bands called principal components analysis. Even if image transformation is not applied in this
study it one of the image enhancement techniques.
3.6. Image classification

Image classification refers to the extraction of differentiated classes or themes, usually, land
cover and land use categories, from raw remotely sensed digital satellite data (Ayele, 2017).
Image classification using remote sensing techniques has attracted the attention of the research
community, as the results of classification are the backbone of environmental, social, and
economic applications (Lu & Weng, 2007). Because image classification is generated using
remotely sensed data, many factors cause difficulty to achieve a more accurate result. Some of
the factors are. The characteristics of a study area, availability of high resolution remotely sensed
data, ancillary and ground reference data, suitable classification algorithms, and the analyst’s
experience, and Time constraint. These factors highly determine the type of classification
algorithm used for image classification. Various image classification methods applied to extract
land cover information from remotely sensed images Layer stack land sat Images are composed
of different colors; these colors are not the real land classes, to get the real category information
from the bands the images need to be classified. There are several classification methods and
each method is specific to the data and the locations because in each location land categories are
varies and have different values in the image. For instance, the image value (reflectance) of
agricultural land is dependent on the type of crop that grows on that land. Even the same crop in
different climates can have different colors which change the color of the image. Moreover, the
seasons also affect the color of land covers. There are different approaches to classification.
According to Caetano (2009) Image classification can be done based on three objectives which
are: Type of learning (Supervised and Unsupervised), Assumptions on data distribution
(Parametric, Non-Parametric), and Number of outputs for each spatial unit (Hard and Soft)
(Caetano, 2009). Moreover, there are also objectives regarded levels of classification, which are;
Pixel-based Classification and Object-oriented Image Segmentation and Classification.
25
3.6.1. Pixel-based classification
Pixel-based classification is the traditional method of image classification. This is mainly based
on the pixel reflectance values of the image (Wang, Sousa & Gong, 2004). According to the type
of learning, there are mainly two kinds of pixel-based classification supervised and unsupervised
(Caetano, 2009). There are two basic methods of image classification: supervised and
unsupervised. The supervised classification relies on the prior knowledge of the study area
(Canada Centre for Remote Sensing, 2010). For this study, the supervised method used for the
pixel-based classification “Supervised classification is a procedure for identifying spectrally
similar areas on an image by identifying “training” sites of known targets and then extrapolating
those spectral signatures to other areas of unknown targets” (Mather & Koch, 2011). There are
different algorithms for supervised classification; the classic classifiers are minimum distance,
parallel pipelined and maximum likelihood methods. For this research, the maximum likelihood
was used because it gave better results for the study area. The maximum likelihood algorithm
uses a maximum likelihood procedure derived from Bayesian probability theory; it applies the
probability theory to the classification process. This method is a supervised method that uses the
training sites, from these sites it determines the class center and the variability in the raster
values in each band for each class. This helps to determine the probability of the cell to be
belonging to a particular class defined in training sites. The probability is depending on the
distance from cell to class center, class size, and the shape of the class in spectral space. The
maximum likelihood classifier computes the class probabilities and classifies the cell where the
probability is higher (Smith, 2011).
3.6.2 Training Site Selection

The unsupervised classification was used in the image classification before fieldwork to
understand the general land cover classes of the study area. This is because unsupervised
classification is automated and requires little knowledge of the study area. Classification of the
Land sat images was carried out within ERDAS IMAGINE 2015. The maximum iterations were
set to 10 and the number of classes set to 5 for each image to ensure consistency in the results.
According to their spectral signature using different band combinations, the classified images
were assigned a class in the output raster. The LULC classes were confused when classified by
the unsupervised scheme. Settlements and cultivated land were highly mixed because most of the
settlements are intermingled within the agricultural field. However, the natural forest was easily
separated from other classes in all images. Based on the unsupervised classification, sample
training sites were selected for data collection during fieldwork. The class assignment was
achieved through a comparison of the classified image with field observation.
3.6.3. Supervised and Unsupervised Classification

After fieldwork, supervised classification was applied to the unsupervised classification
signatures to identify the satellite imagery by using the different training sites. The class
assignment was done using original images, topographic maps, and field study knowledge to
26
identify the various classes. Spectral profiles of the mixed classes from the Land sat images were
used to identify the bands which were capable of separating them. Through the profile, bands
that could separate the mixed class were extracted from the full image, and then the mixed pixels
extracted from the bands were classified. To view and discriminate different surface features
clearly, different band combinations were used which magnify a particular class in the image.
Basic band characteristics and combinations used for Land sat image interpretation are provided
in Table 4 and Table 5.
Table 4:Band characteristics of Landsat Image

BAND Wavelength Properties and Application
Band 1 (0.45-0.52 µm, This short wavelength of light penetrates better than the other bands, and it
blue-green) is often the band of choice for monitoring aquatic ecosystems (mapping
sediment in the water, coral reef habitats, etc.).
Band 2 (0.52-0.60 µm, This has similar qualities to band 1 but not as extreme. The band was
green) selected because it matches the wavelength for the green we see when
looking at vegetation.
Band 3 (0.63-0.69 µm, Since vegetation absorbs nearly all red light, this band can be useful for
red) distinguishing between vegetation and soil and in monitoring vegetation
health.
Band 4 (0.76-0.90 µm, Since water absorbs nearly all light at this wavelength water bodies appear
near-infrared) very dark. This contrasts with bright reflectance for soil and vegetation so
it is a good band for defining the water/land interface.
Band 5 (1.55-1.75 µm, This band is very sensitive to moisture and is therefore used to monitor
mid-infrared) vegetation and soil moisture. It is also good at differentiating between
clouds and snow.
Band 6 (10.40-12.50 µm, Band 6 is primarily used for geological applications but it is sometimes
thermal IR) used to measure plant heat stress.
Band 7 (2.08-2.35 µm This brand is also used for vegetation moisture although generally band 5
mid-infrared) is preferred for that application, as well as for soil and geology mapping.
Adapted from http://biodiversityinformatics.amnh.org
Table 5:Basic band combinations for Landsat Image

Band Combination Properties
3,2,1 RGB This color composite is as close to the true color that we can get with a Landsat
ETM image. It is also useful for studying aquatic habitats.
4,3,2, RGB This has similar qualities to the image with bands 3, 2, 1 however since this includes
the near-infrared channel (band 4) land-water boundaries are clearer and different
types of vegetation are more apparent.
4,5,3 RGB Different vegetation types can be more clearly defined and the land/water interface
is very clear. Variations in moisture content are evident with this set of bands.
7,4,2 RGB This has similar properties to the 4, 5, 3 band combination with the biggest
difference being that vegetation is green.
27
5,4,1 RGB This band combination has similar properties to the 7, 4, 2 combination, however, it
is better suited in visualizing agricultural vegetation.
Adapted from http://gif.berkeley.edu
Generally, the initial classification of the 1990 Landsat TM image for Ambo town yielded results
with most of the classes mixed. Forest land and settlement were difficult to separate because they
were mixed throughout the study area. The settlement had the poorest differentiation of all the
classes and was mixed with virtually all the other classes. The spectral signature for the 2000
image was clearer than in 1990, here the Agricultural land, Cultural land, and bare land were
detected by their unique reflectance. The settlement and agriculture fields were mixed and had
the poorest differentiation of all classes. The classification resulted in a 2000 land use/land cover
map with five classes for Ambo town and five classes in the area. The 2010 image had high
spectral differences between classes than the 2000 and 1990 images. Mixed classes were masked
out and reclassified several times until they were separated.
3.7. Accuracy Assessment

The process of doing an accuracy assessment involved generating a set of points from the
classified image and comparing the positions of points whose location was determined by the
ground truth data and corresponding coordinates from the original maps (ERDAS, 2009). In this
study, these sets of points were selected with random sampling. Random sampling was used to
select a set of points. Points that were used in training the classification were not used for
accuracy assessment. As a result, 2030 random points were created and used to develop an error
matrix for accuracy assessment. The reference points for 1990, 2000, and 2010 were collected
from the corresponding Google Earth; original Land sat images, previous reports and maps, and
field observation for 2020. Information from interviews and group discussions supported us in
getting the historical LULC, for example, forest lands, and bare land agriculture land, cultivate t
land, and built-up area. Since the resolution of the historical maps for 1990 obtained from the
Ethiopian Mapping Agency was poor, a high resolution of Google Earth was used to better
identify the land-use classes.
The common way to represent classification accuracy is in the form of an error matrix. An error
matrix is a square array of rows and columns and presents the relationship between the classes in
the classified and reference data. The reference data used for accuracy assessment was obtained
from field observations and a topographic map. A set of reference points have been taken to
assess its accuracy. The reference points in the study area of Ambo town total 200. These points
were verified and labeled against the reference data. Error matrices were then designed to assess
the quality of the classification accuracy of 2010 LULC maps for Ambo town. Overall accuracy
is computed by dividing the total correct number of pixels (i.e. summation of the diagonal) by
the total number of pixels in the matrix (total). Various standard threshold levels were applied to
the lower and higher tail of each distribution to find the threshold value that produced the highest
change classification accuracy(Behailu, 2010). The producer’s accuracy refers to the probability
28
of a reference pixel being classified correctly. It is also known as omission error because it only
gives the proportion of the correctly classified pixels. It is obtained by dividing the number of
correctly classified pixels in the category by the total number of pixels of the category in the
reference data. User’s accuracy assesses the probability that the pixels in the classified map or
image represent that class on the ground (Lu & Weng, 2007). It is obtained by dividing the total
number of correctly classified pixels in the category by the total number of pixels on the
classified image. The Kappa coefficient was also used to assess classification accuracy. It
expresses the proportionate reduction in error generated by a classification process compared
with the error of a completely random classification(Roy et al., 2015). The Kappa statistic
incorporates the off-diagonal elements of the error matrices (i.e., classification errors) and
represents agreement obtained after removing the proportion of agree4ment that could be
expected to occur by chance. The overall accuracy for the LULC map of 2010 for Ambo town
was 91% severally as shown in Table 8. The overall results of the producer’s accuracy range
from 81.13% to 92.31%for Ambo town. The lowest producer’s accuracy is probably attributed to
the similar spectral properties of some of the land cover classes (e.g. settlement with cultivated
land, Barren land with Agricultural land). User’s accuracy of individual classes ranges from 86%
to 96% study areas as shown in Table 8. From the user’s accuracy point of view, settlement,
Barren land, and agricultural land presented low accuracy for the land cover map 2010. These
classes are misclassified into another land cover class. This is probably caused by the spectral
signature of the features. It is not uncommon that the Kappa coefficient appears below, giving
the impression that the classification of remote sensing performed better than chance only by K
point of proportion (Araya, 2009). It is calculated to be 0.87%for Ambo town.
Various mathematical theories develop to determine adequate sample sizes for Classification of
Accuracy Assessment but no standardized consent reach yet regarding the adequate sample size
determination. On the other hand, (S et al., 2013), (Behailu, 2010) and (Berhanu, 2017) argue
that in most cases a minimum of 50 samples for each mapping class should be collected for maps
of <4000km2 in size and fewer than 12 classes. Thus, by considering the size of the Ambo town
area 10.2km2 and LULC is classified into four classes. namely: Built_up area, Bare land,
Agricultural land, and Forest land classes with sample size a minimum of 50 points per LULC
category. The Ground Control Points (GCPs) would collect from the known real study area,
Ortho-photo, and Google earth (2020).
GCP points require for the study area of four LULC types 50 points total for one year (50*4=200
Reference /GCP Points) used for the Accuracy assessment. For the accuracy assessment overall
kappa which is a statistical measure of overall agreement between two categorical items
(Cohen’s kappa,) and conditional kappa which includes, user’s accuracy, producer’s accuracy,
and overall accuracy were calculated. An error matrix is a square assortment of numbers defined
in rows and columns that represent the number of sample units assigned to a particular category
relative to the actual category as confirmed on the ground. The rows in the matrix represent the
remote sensing derived land use map, while the columns represent the reference data that will be
29
collected from fieldwork. The error matrix tables produce many statistical measures of thematic
accuracy including overall classification accuracy, percentage of omission and commission error,
and kappa coefficient-an index that estimates the influence of chance (Berhanu, 2017).
3.7.1 Overall accuracy

Overall Accuracy is essentially told us out of all of the reference sites what proportion were
mapped correctly. The overall accuracy is usually expressed as a percent, with 100% accuracy
being a perfect classification where all reference sites were classified correctly. Overall accuracy
is the easiest to calculate and understand but ultimately only provides the map user and producer
with basic accuracy information. The diagonal elements represent the areas that were correctly
classified. To calculate the overall accuracy you add the number of correctly classified sites and
divide it by the total number of reference sites.
3.7.4. Producer Accuracy

The producer's Accuracy is the map accuracy from the point of view of the mapmaker (the
producer). This is how often are real features on the ground correctly shown on the classified
map or the probability that a certain land cover of an area on the ground is classified as such. The
Producer's Accuracy is a complement of the Omission Error, Producer's Accuracy = 100%-
Omission Error.
3.7.5. User Accuracy

The User's Accuracy is the accuracy from the point of view of a map user, not the map maker.
The User's accuracy essentially tells us how often the class on the map will be present on the
ground. This is referred to as reliability. The User's Accuracy is a complement of the
Commission Error, User's Accuracy = 100%-Commission Error. The User's Accuracy is
calculating by taking the total number of correct classifications for a particular class and dividing
it by the row total.
3.7.6. Kappa coefficient
The Kappa Coefficient is generated from a statistical test to evaluate the accuracy of the
classification. Kappa essentially evaluates how well the classification performed as compared to
just randomly assigning values, i.e. did the classification do better than random. The Kappa
Coefficient can range from -1 to 1. A value of 0 indicated that the classification is no better than
a random classification. A negative number indicates the classification is significantly worse
than random. A value close to 1 indicates that the classification is significantly better than
random. Jennes et al (2007). Kappa is a statistical measure of overall agreement between two
maps (e.g. output of classified/simulated map and ground truth/reference map) or a matrix
(Araya, 2009).
30
Table 6:Algorithms for accuracy assessment, Ambo town
3.8. Dynamic process

After the classification of land use land covers remote sensing data were converted into thematic
maps. To assess temporal and spatial land of dynamic in the study area Land use and Land cover
were consecutively be analyzed using datasets of remotely sensed land sat imageries. Next, area
change between two consecutive study periods was computed using the classified imageries with
an area extent. Then, the post-classification comparison dynamic technique was applied in 1990,
2000, 2010, and 2020 Ambo Townland cover maps using Erdas software. These land cover maps
were compared pixel by pixel with the final results showing both dynamic-no-dynamic
information as well as ‘from to’ land cover dynamic information.
3.8.1. Modeling LULC dynamic

To model, LULC change CA-Markov chain analysis which is implemented in IDIRIS Andes was
used. Cellular automata-Markov can predict transition among any number of classes. It also
maintains similar frameworks regarding assembly, testing, validation, and calibration. The
combined use of CA, Markov chain; multi-criteria, and multi-objective land allocation (MOLA)
with land cover prediction procedures adds an element of spatial contiguity as well as knowledge
of the likely spatial distribution on landscape transitions to Markov-chain analysis. The CA-
Markov chain analysis indicates the difference in one land use to another land use and further
uses this to predict the future. Modeling land use land cover dynamics (LULCD) is an essential
role in understanding the concepts of land-use dynamics. Cellular Automata can conceptually
understand as a ’cell-based’ approach for modeling dynamic gravity processes at the micro-
level(Han et al., 2015).
Keeping this point in view, the CA-Markov with Multi-Criteria-Evaluation (MCE) is proposed to
be used to generate the decision rules in the form of suitability maps. The transition rules were
based on the factors that have impacts on LULC dynamics such as distance from roads, slope,
altitude, and land use. The Modeling and Simulation of land cover change are fundamental to the
evaluation of successive environmental effects. To calibrate the model and simulate land use
dynamic for the future, drivers of dynamic along with the land use maps of the years 2010 and
2015 were used. The predictive model output for 2020 will then be validated using a “real” land
cover map. The flow chart for the general methodology is shown in the figure below.
31
Figure 3: Flow chart for modeling LULCD with CA-Markov
3.8.2. Modeling LULC change with CA-Markov
CA operates on grid-based cells and transition rules are applied to determine the state of a cell.
Markov Chain Analysis, on the other hand, is a system in which the future state of a system is
modeled based on the immediate proceeding state. It is based on the principle that given the
present state, future states are independent of the past states. To explore land-use dynamics, this
study employed CA-Markov chain analysis integrated with raster-based remote sensing. The
CA-Markov is a combination of both the CA and Markov chain. These two are termed as the two
simulation techniques used to produce land use predictions (Sun et al., 2007). The simulation
refers to the process of land-use change between two points in time and extrapolating this change
into the future (Eastman, 2003). Factors and constraints LULC Map Of 2000 LULC Map of
2010 Calibration and Modeling MCE suitability Map Reference LULC 2020 Simulated LULC
2010 Validation Prediction LULC for 2030 CA-Markov Chain analysis for simulating land use
was implemented in IDIRIS Andes. It can predict transition among any number of classes. The
CA-Markov uses a Multi-Criteria-Evaluation (MCE) to generate the decision rules in the form of
suitability maps. The transition rules were based on the factors that have impacts on LULC
change. These factors include slope, proximity to the road, growing population, and land use.
Other factors like rainfall and aspect are taken constantly throughout the study area because of its
small extent. The biophysical factor, altitude, was also ignored due to its small range.
3.8.3. Model Calibration

A transition probability matrix was first developed using Markov module in IDRISI for each
land use and land cover class between 2000 and 2010 to use as an input for projecting land use
change for 2020 (Table 7).The off-diagonal elements indicate the probability of number of cells
that are expected to change between the two periods from the existing to new classes.
32
Table 7:Transition probability matrix for the projection of 2020 land use and land cover.
Bare land Agriculture land Built_up area Forest land
Bare land 0.0296 0 0.5143 0.0889
Agriculture land 0.2056 0.5524 0.2309 0.1338
Built_up area 0.0389 0.5143 0.7198 0.0731
Forest land 0 0.0334 0 0.0825
3.8.4. Criteria Development

To calibrate CA-Markov land use data, suitability maps and constraints are needed. To develop
criteria for LULC change the criteria, constraints, and factors that have been applied.
Constraints
A constraint serves to limit the alternative under consideration. In this study agriculture was
excluded for the expansion of Built_up. Constraints were expressed in the form of Boolean
(logical) map, areas excluded from consideration being coded with 0 and those open for
consideration being coded with 1.
Factors
Unlike the constraints, factors define some degree of suitability and provide alternatives in terms
of a continuous measure of suitability (e.g. a byte scale of 0 to 255). These criteria do not
absolutely constrain development, but enhance or detract from the relative suitability of an area
for development (Eastman, 2003).The criteria were obtained by consulting Built_up area experts
and from interpretation of LULC map of 1990, 2000, 2010 and 2020 of Ambo Town Kebele.
The factors considered were:
Factors are generally continuous (such as the slope, Distance to Road, and Types of LULC); they
indicate the relative suitability of certain areas. Factors are different than constraints because
factors have continuous suitability values different than 0 and 1. Constraints are the locations
that are not allowed for urban development by law or existing occupied areas like an existing
Built_up area where the development is not possible. These constraints are expressed in the form
of Boolean (logical) maps. In the type of Boolean map, the values 0 represent the areas excluded
from consideration and the values 1 represent the areas that were included for preparing
suitability maps (Fekadu, 2017). The factors or land cover constraints are standardized to a
33
continuous scale of suitability from 0 (the least suitable) to 255 (the most suitable). The 0-255
range provides a byte data type cited as(Fekadu, 2017).
1. Land use factor: this determines the location and spatial extent of the amount of land
available for development. This is based on the analysis of land use/land cover maps obtained
from previous years.
2. Proximity to the road: the influence of road proximity is measured in terms of accessibility
and areas close to road access have a greater probability to be changed. The major road axis was
obtained from Ethio GIS and its effects have been determined by the distance function which is
implemented in ArcGIS 10.5.
3.Slope: Areas that have low slopes (less than 10%) are suitable for development while areas
with greater than 25% slope are not suitable .
3.8.5.Weighting Factors for Suitability
Based on the suitability maps, the next step was to establish a set of weights for each of the
factors studied and the analyst has to fill out the pairwise comparison matrix using the WEIGHT
module in IDRISI Andes. This module uses a pair-by-pair technique to compare the relative
importance of one factor (e.g. slope) against another factor (e.g. land use). The rating ranges
from “extremely less important” (1/9) to “extremely more important” (9). The rating is
subjective and entirely depends on the analyst. The analyst compares every pair and assigns the
rating into the matrix.
The assigned weight which is derived from the pairwise matrix is shown in Table 8.The
consistency ratio (CR) was calculated to be 0.03. CR is the procedure by which an index of
consistency can be produced. It also indicates the probability that the matrix ratings were
generated randomly and shows possible inconsistencies in the matrix (Satty, 1997). It is stated
that, a consistency ratio greater than 0.1 should be re-evaluated.
Table 8.Table 8:Weights assigned to variable
Variable Weight
Slope 0.250
Proximity to Road 0.647
Land Use 0.103
CR=0.04
34
Finally, the module MCE in the form of a weighted linear combination (WLC) was applied to
combine the standardized factors and constrains. With a WLC, each standardized factor image is
multiplied by its weight, then the results are summed and then multiplied by the product of the
constraints (i.e. Boolean constrain maps are applied to limit areas under consideration in the final
analysis). The resulting suitability map has a range from 0-255 as shown in Figure 3.6.
3.8.6. Modeling LULC for 2020 with CA-Markov

A cellular automaton is an agent or object that can change its state based upon the application of
a rule that relates the new state to its previous state and those of its neighbor(Mahmoud &
Divigalpitiya, 2017).In this study, a CA filter was used to develop a spatially explicit contiguity
weighting factor to change the state of cells based on their neighbors. The filter used was a 5 by
5 contiguity filter. The CA Markov module which is implemented in IDRISI Andes has been
used for the analysis of this model. It combines both the concept of a CA filter and the Markov
change procedure. A transition areas table and suitability image from Markov was used by the
CA Markov to predict land cover change for 2020. Based on these inputs, the module determines
the location of change, the number of pixels that must undergo each transition, and selects the
pixels according to the largest suitability for a particular transition. The initial analysis used the
2000 and 2010 land cover maps to “predict” for the year 2020. Agricultural was taken as a
constraint for Settlement field expansion. This was created from the land use map of the year
2010, which was used as a base for simulation. This was created from the land use map of the
year 2020, which was used as a base for simulation.
3.8.7. Model Validation

Model validation refers to comparing the simulated and reference maps. Sometimes the
simulated maps can give misleading results. In that case, it is necessary to validate the
projected/simulated map with the base/reference map. Therefore, model validation is an
important step in the case of predictive change modeling (SIraj et al., 2018). The Validation is an
important stage panel that allows determining the quality of the prediction land use/land cover
map about a map of reality. Pair of validation processes such as VALIDATE and ROC.
VALIDATE provides a method to measure agreement between two categorical (integer or byte)
images, a "comparison" map, and a "reference" map. It calculates various Kappa Indices of
Agreement and related statistics that indicate how well the comparison map agrees with the
reference map and in three types such as bar graph with table, trend line graph as a function of
multiple resolutions, and text. Map comparison is vital in evaluating the analytical techniques for
spatial data. It produces numerical expressions to compare two maps(Vliet,J.V.,2009 ). The best
way to validate a model is to compare the predicted map of time t2 with the reference map of
time t2 (Tizora, 2018). For validating the output of the simulated LULC map in 2020, two maps
were compared the reference LULC map of 2020 and the simulated LULC map of 2020. The
validation was based on kappa statistic and multiple-resolution. Several variations of the Kappa
35
statistic have been introduced by Pontius such as Kappa for no information (denoted as Kno) and
Kappa for grid-cell level location (denoted as Klocation) (Araya, 2009).
3.8.7.1.Kappa
Kappa is a statistical measure of overall agreement between two maps (e.g. output of
classified/simulated map and ground truth/reference map) or a matrix (Canada Centre for
Remote Sensing, 2010).
Table 9: Kappa statistic of Map comparison
The Kappa statistic (K) is defined according to the following equation (Hagen, A., 2002).
P( A)−P (E)RL
K=
1−P(E)R L
Where, Pnn. = the proportion of cells that is of category ‘n’ in map A and ‘m’ in map B. Pmn =
the proportion of cells that is of category n in map A and category m in map B based on the
contingency table the following equations can be derived ( Hagen, A., 2002).
Table 10:Strength of Agreement for Kappa Statistic (Mahmoud & Divigalpitiya, 2017)
Kappa Statistic Strength of Agreement
˂0 Poor
0.00-0.20 Slight
0.21-0.40 Fair
0.41-0.60 Moderate
0.61-0.80 Substantial
0.81-1.00 Almost Perfect/Perfect
36
3.8.8.2. Kno
The Kno indicates the proportion classified correctly relative to the expected proportion
classified correctly by a simulation with no ability to specify accurately quantity or location
(Araya, 2009).
3.8.9.3. Klocation
The Klocation indicates the success due to the simulation’s ability to specify location divided by
the maximum possible success due to a simulation’s ability to specify location perfectly
(Pontius2000). It depends strictly on the spatial distribution of the categories on the map. It
indicates how well the grid cells are located in the landscape (Behailu, 2010).
m
P ( A )=∑ ¿ 1 pnn
n
m
P (E) RL = ∑ 1 =1 Pn ∗ Pm
n
Where, P (max) = the maximum success rate P (E) RL = the fraction of expected agreement for
the whole map, where RL stands for random location.
3.9.1. Kstandard
It is the success due to the simulation’s ability to specify quantity divided by the maximum
possible success due to a simulation’s ability to specify quantity perfectly ((Tizora, 2018)
In the case of this research before predicting for the 2030 year, prediction for 2020 is first used
for validation. using 2000 and 2010 classified Ambo Town predict for 2020, then validate
predicted 2020 with classified 2020, when the value of validation is acceptable, with the same
procedures predicting for 2030 Ambo Town LULC classes.
3.10. Future Prediction map

Using Land Change Modeler, GEOMOD, Markov chain analysis, and cellular Automata it is
possible to determine the future Land use Land cover project maps. Using this MLP neural
network analysis it is possible to determine the weights of the transitions that will be included in
the matrix of probabilities of the Markov Chain for future prediction. The future Land use Land
cover project maps of (2030) may be hard classifiers (Traditional classifier) since they yield a
hard decision about the identity of each pixel and soft classifiers express the degree to which a
pixel belongs to each of the classes being considered. For this research hard classification of the
Future predicted map is generated
37
CHAPTER FOUR
4. DATA ANALYSIS, RESULTS, AND DISCUSSION
4.1. INTRODUCTION
As indicated in the classification scheme Settlement/Built-up area, Agricultural land, bare land,
and Forest land are the major LULC classes for the study periods. This deals with the description
of the detailed procedures, results, interpretation, and discussion in Analysis of Ambo Town
Urban Land use land cover change and suitability for Housing using Land satellite (1990, 2000,
2010, and 2020) to evaluate for change detections, model and future prediction. After
downloading these land satellites from www. USGS/earth explorer, subset/extract area of study
using the Ambo Town boundary that digitized from Ambo Town Structural plan exist in digital
image preprocessing using ERDAS software the following Main procedures/techniques and
results are interpreted and discussed.
4.2. Data Analysis, Results, and Interpretation
4.2.1. Layer stack

Layer stack means Stack multiple (usually single band) images as bands/layers into a single
output multi-band image file. Landsat1990, 2000, 2010, and Land sat 2020 has 7, 9,7,11 separate
bands respectively, those separately exist band combined into single multiband for each land
satellites.
4.2.2. Digital Image Processing

Most of the common digital image processing functions available in image analysis systems are
categorized into the following four categories. First, Preprocessing functions involve data
analysis and extraction of information, generally grouped as radiometric or geometric
corrections. Secondly, Image enhancement is to improve the appearance of the imagery to assist
in visual interpretation and analysis. Thirdly, Image transformations are operations similar in
concept to those for image enhancement. Fourthly, Image classification and analysis operations
use to digitally identify and classify pixels in the data.
4.2.3. Land use and land cover change

As shown in Table 10 and Figure 4 the classification resulted for the 1990 Ambo Townland
cover map with four classes. Agricultural land was the dominant class covering 60.19%,
38
followed by Forest land and Built_up area covered 21.86%, 13.73%, and Bare land 4.22%
respectively.
Table 11:LULC Class Distributions of Ambo town in 1990
In the classification result of the 2000 Ambo Townland cover map, Agriculture land was the
dominant class covering 44.11%, followed by Forest land 25.49%and Built_up area covering
22.54%, and Bare land 7.84%, respectively the total area as shown in Table 11and Figure 5.
Table 12:LULC Class Distributions of Ambo tow in 2000
In the classification result of the 2010 Ambo Townland cover map, Built_up area was the
dominant class covering 50.72 %, followed by Forest land ,Agriculture covering 33.33 %, and
26.47% respectively while Bare land covered 5.88 % of the total area as shown in Table 12 and
Figure 6.
In the classification result of the 2020 Ambo Townland cover map, Built_up area was the
dominant class covering 56.86 %, followed by forest land, and Agriculture land covering 22.54%
and 11.76 % respectively while Bare land covered 8.82% of the total area as shown in Table 13
and Figure 7.
39
4.2.4. Land use and land cover image classification

The classification of these land sat images are pixel-based classification category, supervised
classification method with maximum likely hood algorithm for all Ambo Town Land use land
cover classes of 1990,2000,2010 and 2020. Training sites are the areas defined for each land
cover type within the image and creating the spectral signature for each type of land use land
cover. This is done by analyzing the wpixels of the training sites (Fekadu, 2017). Using training
site and signature, the maximum likely hood classification of window composite of each year
1990, 2000, 2010, and 2020. After classification using Boundary of Ambo Town Clip or extract
by mask from classified LULC of Land sat 1990, 2000, 2010 and 2020.
Figure 4:Ambo Town Classified LULC of 1990
40
41
4.2.5. Accuracy Assessment of classified Land use the land cover image
The Ground Control Points (GCPs) would be collected from the known real study area, Ortho-
photo, and Google earth (2020). The Total numbers of reference points 200 are shown for each
LULC Classified map of 1990, 2000, 2010, and 2020. The Error matrix tables produce many
statistical measures of thematic accuracy including overall classification accuracy, percentage of
omission and commission error, and kappa coefficient-an index that estimates the influence of
chance(Fekadu, 2017).
4.2.5.1. Overall Accuracy

To calculate the overall accuracy of add the number of correctly classified sites and divide it by
the total number of reference sites. As a sample following number taken from error matrix Table
of 1990, Summation of correctly classified diagonal; 45+44+42+46=177/200=88.5% Total
number of reference Data=200 then Kappa accuracy=0.84 With Similar steps the overall
accuracy for 2000 overall accuracy 88.5 (0.885%), Overall accuracy is 0.89 (89), and overall
accuracy is 0.90 (90).
42
4.2.5.2. Producer Accuracy
Producer's Accuracy is the map accuracy from the point of view of the map maker (the
producer). Producer's Accuracy = 100%-Omission Error Built_up area=100%-10.6%=89.4%,
Forest Land=100%-14%=86%, Agriculture land=100%-11.3%=88.7% and Bare land=100%-
18.8%=81.2%. For Built_up area, Producer Accuracy for 1990, 84%, for 2000, 94%, for 2010,
88% and for 2020, 94%.
4.2.5.3. User Accuracy

The User's Accuracy is the accuracy from the point of view of a map user, not the map maker.
User's Accuracy = 100%-Commission Error Built_up area=100%-14.3%=85.7%, Forest
Land=100%-6.5%=93.5%, Agriculture land=100%-9.6%=90.4%, and Bare land=100%-
23.2%=76.8% For Built_up area, User Accuracy for 1990, 85.7%, for 2000, 88.6%, for 2010,
86.2% and for 2020, 83%.
4.2.5.3. Kappa coefficient

The Kappa Coefficient is generated from a statistical test to evaluate the accuracy of
classification. Kappa coefficient calculates using below formula
Where k=kappa coefficients, N=Total number of pixel points, Xii= correctly classified points
(observation) in row i and column i, Xi=sum of Row points, Xj=sum of Column points. As a
samplefromtableError Matrix 1990,N=200, Xii=177, ∑Xi*Xj=[(44*46)
+(42*48)+(48*43)+(45*47)+(47*54)=13935929580 K=0.88. Generally, Ambo town Land uses
land cover kappa coefficient for 1990, 0.84 for 2000, 0.84, for 2010 0.85, and 2020, 0.86. All
kappa coefficient of Ambo Town Land uses land cover classified above 0.80 this according to
rating criteria the strength of agreement is almost perfect. So classified land use land cover maps
of Ambo Town in 1990, 2000, 2010, and 2020 are the accurately classified map. Rating criteria
of Kappa statistics Error matrix of the 1990 LULC classification of Ambo Town (in pixels).
43
Table 15:Error matrix of the 1990 LULC classification of Ambo Town (in pixels)

Referenced Data 1990
agricultural Forest Built up Bare Rw
Lulc land land area land total UA PA
agricultura 88.2
l land 45 1 2 2 50 % 90%
91.6
Forest land 2 44 3 1 50 % 88%
Classifie Built up 85.7
d Data area 2 3 42 3 50 % 84%
88.4
Bare land 2 0 2 46 50 % 92%
Column
total 51 48 49 52 200
number of lULC correctly classified =177
over all classification =(45+44+42+46)/200=177/200*100=88.5%
over all kappa coefficient=0.84

agricultural Forest Built up Bare Rw
Lulc land land area land total UA PA
92
agricultural land 46 1 1 2 50 85.1% %
86
Classified Forest land 3 43 1 3 50 93.4% %
Data 94
Built up area 2 0 47 1 50 88.6% %
82
Bare land 3 2 4 41 50 87.2% %
c.total 54 46 53 47 200
over all classification
=(46+43+47+41)/200=177/200*100=88.5%
44

Forest Built up Bare Rw
Lulc agricultural land land area land total
UA PA
agricultural
land 42 3 3 2 50 87.5% 84%
Forest land 1 47 2 0 50 90.3% 94%
Built up area 2 2 44 2 50 86.2% 88%
Classifie Bare land 3 0 2 45 50 91.8% 90%
d Data c.total 48 52 51 49 200
over all classification =(42+47+44+46)/200=178/200*100=89%

Forest Built up Bare Rw
Lulc agricultural land land area land total UA PA
agricultural
land 42 3 2 2 50 91.3% 84%
Forest land 0 46 3 1 50 92% 92%
Built uparea 1 1 47 1 50 83% 94%
classifie Bareland 3 0 2 45 50 91.8% 90%
d Data c.total 46 50 54 49 200
over all classification =(42+46+47+45)/200=180/200*100=90%
4.2. Land use and landcover Dynamics

Land cover change comparison matrix gives the general information of major changes of the
land cover classes analyzed for the two periods. To clearly understand the major land cover
source and destination of cover class’s change, conversion matrix for each period was analyzed.
The rows of the tables show the initial stage and the column represents the final stage. For
example, column one in Table 19 indicate that from a total of 4823 Km2 Agriculture land in
1990 ,4299.21Km2remained as a natural forest,76.95 ha changed to settlement,266.58Km 2
changed to shrub land,14.13ha converted to woodland,23.4ha to marsh land and 131.49 ha
changed to cultivated land. Similarly, as seen clearly in the matrix, from a total of 1005.21ha
45
Shrub land in 1973, 251.91ha and 67.23ha converted to cultivated land and settlement,
respectively. Land cover change matrix for 1990/2000 for Ambo Town indicate that
settlement and agriculture accounted for the highest percentage loss in vegetation.This is an
indicator of increased population pressure on the environment due to the 1984/85 socioeconomic
transformation in the area
Table 19:.Land cover conversion matrix of Ambo Town for 1990/2000
2000LULC Class
Agriculture Forest
LU/LC Types land Bare land Built-up area land Total
Agriculture land 3.4 0.5 0.9 1.3 6.1
1990LULC Class
Bare land 0.1 0.1 0.1 0.1 0.4

Forest land 0.6 0.1 0.3 1.2 2.2
Built-up area 0.2 0.1 1.0 0.1 1.5
Total 4.4 0.8 2.4 2.6 10.2
The 1987/2006 land cover matrix for Tula-Kuti, as shown in Table 13, indicated similar trend in
a loss of natural forest, shrub land and woodland for settlement and cultivated land. For example,
the first column show that, from a total of 4836.92 ha natural forest in 1987, 182.7ha converted
to settlement and 564.8ha changed to cultivated land in 2006. Similarly, from 801.81ha shrub
land in 1987, 431.4ha and 72.83 ha converted to cultivated land and settlement, respectively. The
lowest unchanged classes in both 1973/1987 and 1987/2006 land cover matrix are settlements
and marshland.
Table 20:Land cover conversion matrix of Ambo Towni for 2000/2010
2010LULC Class
Agriculture Bare Forest Tota
LU/LC Types land land Built up area land l
2000LULC
Bare land 0.3 0.3 0.1 0.2 0.9

Built-up area 0.4 0.1 1.6 0.2 2.3
Forest land 0.4 0.2 0.3 1.7 2.6
46
Total 3.4 0.6 3.4 2.8 10.2
Cla
ss
Apart from change no change information, post classification comparison also resulted in a
change matrix that provided from-to change information. The results indicated that both land
cover conversion and land cover modifications were significant between 1987 and 2006 for
Beyemo in a similar manner compared to the other resettlement areas. Vegetation modification
was high between the two dates. Shrub land changed significantly to cultivated land, grazing
land and Settlement. Shrub land modification could be attributed to selective cutting of vegettion
for fuel wood, building materials and overgrazing. Most of the vegetation in Beyemo was
converted to cultivated land and settlement; this may be related to a drastic increase in
population pressure as shown in Table 14.

Although agricultural land extensification is a common trend elsewhere in the country such as
Mettu area (Solomon 1994), Chemoga watershed (Gete 2000) and Fincha’a watershed
(Bezuayehu 2006) , the situation in study area was remarkable. This was most likely caused by
the resettlement of people from Wollo in 1984/85 and high rate of population growth. It should
also be stressed that consideration of forest lands as a free access resource in the area must have
further aggravated the expansion of cultivated land. In general, LULC change is an indirect
measure of population pressure; it provides robust evidences on the footprints of active human
impacts on natural resources. The latter are usually manifested through expansion of cultivation
fields and new settlements following forest and woodland clearings. The current study clearly
illustrates human-driven changes in LULC of the study areas. The growing population and
increasing socio-economic necessities creates a pressure on landuse/landcover. This pressure
results in unplanned and uncontrolled changes in LULC. Accurate information on land-cover
changes and the forces and process behind is essential for designing a sound environmental
planning and management. Land-cover analysis provides the baseline data required for proper
understanding of how land was used in the past and what types of changes are to be expected in
the future. Studies of land-cover changes also yield valuable information for analysis of the
environmental impacts of human activities, climate change, and other forces. Such analysis is of
great use to the resource manager because it provides information that would help in resolving
47
conflicts between human use of natural resources and the function of natural systems (Pyrovetsi
and Karteris 1986). A change in land-cover can negatively affect the potential use of an area and
may ultimately lead to degradation and loss of productivity.
2020LULC Class
Agriculture Forest Tota
LU/LC Types land Bare land Built up area land l
agriculture 0.5 0.3 2.3 0.4 3.4
2010LULC Class
Bare land 0.1 0.1 0.2 0.1 0.6

Built up area 0.4 0.2 2.5 0.2 3.3
Forest land 0.2 0.1 0.9 1.6 2.9
Total 1.1 0.8 6.0 2.2 10.2

2030LULC Class
Agriculture Bare Forest Tota
LU/LC Types land land Built up area land l
2020LULC Class
Bare land 0.2 0.3 0.2 0.1 0.8

Built up area 0.1 0.1 5.6 0.1 5.9
forest 0.3 0.1 0.1 1.8 2.2
Total 1.3 0.5 6.5 1.9 10.2
4.3.Land use land conversion matrix classification
48
Figure 8:Land use land conversion matrix of 1990 to 2000
49
Figure 10:Land use land conversion matrix of 2010to 2020
50
4.2.6 Change analysis and prediction modeling
Land-Use &Land Cover Change (LULCC) modeling is a rapidly growing scientific field because
land-use change is one of the most important ways that humans influence the environment.
Knowing the changes that have occurred in the past used to predict future changes. Land change
prediction in Land Change Modeler (LCM) is an empirically driven process that moves in a
stepwise fashion from Change Analysis to Transition Potential Modeling from transition
modeling to Change Prediction.
4.2.5.1 Change analysis

Change is assessed between time 1 and time 2 between the two land cover maps. The changes
that are identified are transitions from one land cover state to another. It is likely that with many
land cover classes the potential combination of transitions can be very large. There is four
possible change detection analyses overtime for this research. These are changes between 1990-
2000, 2000-2010, 2010-2020, and 1990-2020. These all change in area are calculated using
image difference (later- earlier) Classified Land use the Land cover of Ambo Town of four
different years. The Change Analysis is used for the rapid assessment of changes such as gains
and losses, net change, persistence, and specific transitions both in the map and graphical form.
Figure 4-5 change analysis (Gain and losses by category)
Table 24:AmboTown Gain and losses of LULC Classes for three decades in table form.
LULC Classes 1990 2000 2010 2020 Gain (1990-2019) Losses (1990-2019)
Built up area 1.4 2.3 3.5 5.8 4.4
(Km2)
Built up area 13.73 22.54 34.31 56.86 43.13
(%)
(Km2)
(%)
Bare land(Km2) 0.43 0.8 0.6 0.9 0.5
Bare land (%) 4.22 7.84 5.88 8.82 4.60
Forest 2.23 2.6 3.4 2.3 0.07
51
land(Km2)
Forest land (%) 22.86 25.49 33.33 22.54 0.32
Total (Km2) 10.2 10.2 10.2 10.2 0
Total (%) 100 100 100 100 0
Generally area change of LULC classes in three decades (2020-1990) Built up area and Bare
land is increased by 4.4 Km2 (43.13%) and 0.5Km2(4.60%), Agriculture and Forest land is
decreased by 4.94 Km (48.43%) and0.07(0.32%)respectively.
Table 25:Ambo Town Net change by category of LULC Classes for three decades in table
form.
LULC Classes 1990-2000 2000-2010 2010-2020 1990-2020
Built up area +0.9 +1.2 +2.3 +4.4
(Km2)
Built up area (%) +8.81 +11.77 +22.55 +43.13
Agriculture land -1.64 -1.8 -1.5 -4.94
(Km ) 2
Agriculture land -16.08 -17.64 -14.71 -48.43

(%)
Bare land(Km2) -0.37 +0.2 -0.3 -0.47
Bare land (%) -3.62 +1.96 -2.94 -4.6
Forest land(Km ) 2
-0.37 -0.8 +1.1 -0.07
Forest land (%) -2.63 -7.84 +10.79 +0.32
Total (Km2) 0 0 0 0
Total (%) 0 0 0 0
Table 4-5 Ambo Town Gain and losses of LULC Classes.
52
Generally, area change of LULC classes in three decades (2020-1990) Built-up area is increased
by 440 Ha(44%), Agriculture land is decreased by -494 Ha(43%) and Bare land and Forest land
are oscillation by 7 Ha(0.7%) and 47 Ha(4.6%).
Figure 4-7 change analysis (contributor to the net change in Settlement area)
53
4.2.7.2 Markov Chain Model Analysis
Using Markov chain Analysis the following transition area and transition probability.
4.2.7.3 Transition Potential Modeling

Using the MLP Sub model Ambo Town transition potential from all Land use land cover classes
to Settlement area with the idea of many to one relationship and all land use land cover to all
land use land cover with the concept of many to many relationships for 1990, 2000, 2010 and
2020 are generated. Transition all LULC to Settlement area means a change of Forest land,
Cultivate land, Agriculture land, and Bare land to Settlement area within 1990-2000,2000-2010
and 2010- 2020. The transition from all LULC to all LULC means the change of all Built-up
area, Forest land, Cultivate body, Agriculture land, and bare land to Settlement area, Forest land,
Water body, Agriculture land, and bare land in cross-correlation from 1990-2000, 2000-2010,
and 2010-2020.
54
Figure 4-9 Ambo Town Transition from all LULC Classes to Built-up area (1990-2000) Ambo
Town Change Map (Transition from all LULC Classes to Built-up area) from 1990- 2000 Bare
Land (300Ha), Agriculture land (200Ha), and Forest land is (100Ha).
Figure 4-10 Ambo Town Transition from all LULC Classes to Built-up area (2000-2010).
AMBO Town Change Map (Transition from all LULC Classes to Built-up area) from 2000-
2010 Shrub Land (1100Ha), Agriculture land (300Ha), and Forest land is (100Ha).
Figure 12:Ambo Town Transition from all LULC Classes to Settlement area (2010-2019).
2019 Shrub Land (1600Ha), Agriculture land (700Ha), and Forest land is (200Ha).
Figure 4-12 Ambo Town Transition from all LULC Classes to Settlement area (1990-2020).
2019 Shrub Land (1500Ha), Agriculture land (1100Ha), and Forest land is (1400Ha). Ambo
Town change maps of all to all LULC Classes means a change of Built_up is Change to Built_up
55
area, Forest land, Agriculture land and bare land similarly, Forest land to Built_up area, Forest
land, Agriculture land, bare land similar for Cultivate land, agriculture and bare land.
4.2.7.4 . Spatial trend of change.

The spatial trend analysis tool used to compute maps of transition trends from all land cover
categories to Built-up areas between change maps (1990-2000, 2000-2010, 2010-2020, and
1990- 2020). It was created using a default 3rd order of the polynomial, which is the best fit to
the pattern of change, in LCM. The numeric values produced don't have any special significance
(Eastman, 2012). Thus, the result is interpreted as: the lower the value, the fewer changes, and
the higher the value, the more changes.
Figure 4-13 Ambo Town Spatial trend of change map (1990-2010). Ambo Town spatial trend
change map in each Kebeles (All LULC to Built-up area (1990- 2020). Ambo kora, Sanqele fi
faris, and Awaro indicated by chance in all directions of around Ambo Townland used changed.
56
4.3. Suitability analysis
The definition of suitability, in general, is the “quality of having the properties that are right for a
specific purpose”, and for the land-use suitability we can specify the definition suitability as
identifying the most appropriate spatial pattern of future land uses according to the purpose
(Hopkins, 1977).
4.3.1 Factors and Constraints

For this study, factors that indicate the relative suitability of certain areas include slope, Distance
to Road, Types of LULC, distance from Built-up area; whereas, constraints that hinder the
development of urban involves the existence of Huluka and Teltale river that cross in the center
of the Ambo town just like an existing built-up area where the development is not possible. On
the Multi-Criteria Evaluation (MCE) process where fuzzy set membership is used for
standardizing criteria ((Fekadu, 2017)). Factor weights are assigned to specify the relative
importance of each factor in determining the aggregate output value ((Fekadu, 2017). Pairwise
comparison
4.5.Future Prediction (Projected) and Modeling for 2020.
MLP Markov model has been selected for simulating the land cover map of Ambo Town for the
year 2020 for validation and predict the final land use land cover map for 2030. This result is the
steps followed from selecting drivers, creating a Boolean image, creating Distance of drivers,
creating LULC transition image, and Transition Potential modeling. Two basic models of change
prediction are the Hard prediction model and a soft prediction model. The hard prediction model
is based on a competitive land allocation model similar to a multi-objective decision process.
Using a hard prediction Model following map is generated.
Table 4-16 Ambo Town LULC classes’ area change between Predicted and Classified 2020
Ambo Town Classified LULC Classes Settlement area of LULC 2020 is decreased from 4100Ha
to 2500Ha predicted area using factors/variables that determine the suitability of Settlement area
(Urban land). These shows (1600Ha) of illegal construction in a sensitive area like solid waste
57
disposal site, Nearest to Rivers that a place is a place where the societies are affected by
economic, social, and environmental problems. Finally, bare land is increased from 1800Ha to
3300Ha.
4.6 Model Validation
Model validation refers to comparing the simulated and reference maps. Sometimes the
simulated maps can give ambiguous results. In that case, it is necessary to validate the
projected/simulated map with the base/reference map (Eastman, J. R., 2009). For validating the
output of the simulated LULC map in 2020, two maps were compared the reference LULC map
of 2020 and the simulated LULC map of 2020. The validation was based on kappa statistic and
multiple-resolution. Several variations of the Kappa statistic have been introduced by Pontius
such as Kappa for no information (denoted as Kno) and Kappa for grid-cell level location
(denoted as Klocation) (Pontius, R. G., 2000).
4.6.1 Kappa
Kappa is a statistical measure of overall agreement between two maps output of

classified/simulated map and ground truth/reference map or a matrix (Canada Centre for Remote
Sensing, 2010). The Kappa statistic (K) is defined according to the following equation (Hagen,
A., 2002)
P ( A )−P( E) RL
𝐾= Where P (A) = Percentage of cells in the map that are identical; P (E) RL =
1−P( E) RL
Random Location (RL) conditional to the observed distribution in both maps. The calculation of
Kappa is based on the “Contingency Table‟ or „Confusion Matrix‟ (Hagen, A.2002). Using this
m m
formula Overall Kappa = 56.26%. P (A) = ∑ Pnn and P (E) RL= ∑ Pn∗P𝑚 Where, Pnn. = the
n =1 n =1
proportion of cells that is of category ‘n’ in map A and ‘m’ in map B. Pmn = the proportion of
cells that is of category n in map A and category m in map B based on the contingency table the
following equations can be derived (Hagen, A, 2002)
4.6.2 Kno, Klocation, and Khisto
58
Kno is a statistic similar to Kappa, but better capable of expressing similarity both in quantity
and location (Hagen, A., 2002) “Klocation indicates how well the grid cells are located on the
landscape (Pontius, 2000).
𝑃(max) = ∑ min ⁡(Pn∗Pm)

n =1
P ( A )−P( E)
∑ (Pn−Pm)
n=1
𝐾𝑙𝑜𝑐𝑎𝑡𝑖𝑜𝑛 = P ( max )−P( E) = m Where, P(max) = the maximum success rate,
∑ ¿¿¿
n=1
P(E)RL = the fraction of expected agreement for the whole map, where RL stands for random
location. Khisto can be calculated directly from the histograms of two maps (Hagen, A., 2002).
P ( max )−P( E) m
Khisto= =∑ ¿¿ ¿
1−P(E) n =1
The kappa statistic is showed in table 15,16,17,and 18 that includes kappa histogram (0.932)
that is an estimation of the frequency distribution of the simulation, Kappa location (0.8766) that
is the simulation’s ability to perfectly specify location between the reference map and the
simulated map, and Kappa overall (0.5626) that is the total accuracy of the number pixel
correctly classified between the reference map and simulated map. All of these Kappa values can
be referred to Landis and Koch (1997), which explained the kappa value range. The value
between 0.41-0.6 corresponds to a moderate agreement between the reference map and the
simulated map. The total correctness value is 53.41%, which indicates that the moderate
accuracy of the simulated LULC map in 2020. VALIDATE measure agreement between two
categorical (integer or byte) images, a "comparison" map, and a "reference" map. It calculates
various Kappa Indices of Agreement and related statistics that indicate how well the comparison
map agrees with the reference map and in three types such as bar graph with table, trend line
graph as a function of multiple resolutions, and text.
4.7Change Prediction Modeling for 2030
The Change Prediction (simulation) modeling for 2030 was based on the Cellular Automata. The
validation of the model accuracy is needed, to achieve acceptable accuracy, this study had
59
employed an approach to simulate LULC of 2020 (time t3) from the historical LULCD process
for time one (2000) and for time two (2010) and then the simulated result was compared the
reference LULC map of 2020 (classified LULC map 2020). Since, the reference map is usually
considered more accurate in the study area at time three (Dushku and Brown, 2003; Ponstius and
Chen 2006). The simulated LULC in 2020 was successful in both correctness value and multi-
resolution. The correctness value was 53.31% that is a moderate agreement between the
reference map and the Predicted (simulated) map. The multi-resolution indicated the perfect
location and medium quantity information between the reference map and the simulated map,
which means that the historical LULCD process from 2000 to 2010 is accurate and reliable to
predict LULC patterns in 2030. However, the validation of the simulated map is a challenge
because there is no criterion to assess the performance of the different LULCD models. Another
problem is parameters to indicate the overall accuracy, parameters for comparing different
modeling results, and the minimum accuracy standard (Pontius and Chen, 2006). Therefore, this
research was considered to simulate LULC for the short period. So with similar procedures using
Ambo Town 2010 and 2020 LULC generating the potential transition of all land use land cover
classes by using factors/constraints and change prediction of Ambo Town LULC of 2030.
4.7.1 Markov Chain Model Analysis
Figure 13: Markovian Conditional Probability of being class Built up area (predict 2030).
Markovian Conditional Probability of being class Built-up area, Forest land, Agriculture land
and Bare land for predict 2020 maps.
60
Figure 14: Ambo Town LULC Classes of three decades and predicted 2030.
Figure 15:Ambo Town LULC Change (Km2)
4.8 . Summary of results

Summary of land use and land cover change from 1990 to 2020
Bare land Forest land Agriculture land Built_up area Total

1990 4.22 21.86 60.19 13.73 100%
2000 7.84 25.49 44.11 22.54 100%
61
2010 5.88 33.33 26.47 34.31 100%
2020 8.82 22.54 11.76 56.86 100%
Figure 16:LULC Changes of Ambo town from 1990 to 2020

Although the period of 1990 from 2020 there is dramatic change in several LULC categories
including Built_up area (440 ha.) or (43.67%), Bare land (47 ha) or (4.60%) , and Forest land (7
ha) or (0.68%) area increased and agricultural land(-506 ha.) or (-48.43%) decrease . According
to above table tendency towards more land brought under Built_up area , Bare land and Forest
land. These data expressly stated that increase in Built_up area ,Bare land and Forest land
coverage of the town resulted population pressure on land and there is policies encourage
management mining material and afforestation the town . These growing demand of space for
human Built_up area and commercial purpose is diminishing the amount and size of Arable
(agricultural)land. Expansion of informal construction of grabbing land, commercial
constructions /mining material, Ambo University expansion, and Ambo international sports
ground are good percentage increase of Built_up area.
These summaries of results show how all generated results fulfill the required specific objectives.
1. To investigate land use and land cover dynamics in the past three decades.
This was obtained by using GIS software to analyze LULC maps derived from remote sensing
imagery. The LULCD quantitative analysis results indicate that there were significant LULC
changes between 1990 and 2020 characterized by declines in forest land and agricultural land,
and bare lands are neither increased nor decreased. In contrast, Built_up land classes exhibited
62
increases. Accurate land use land cover change obtained from accurately classified LULC map,
using the accurately classified map quantifying the changes in the area to determine the change
of each LULC class for each year. The main objective of the research is to monitor land use land
cover dynamics within 30 years i.e. (2020-1990). Generally, area change of LULC classes in
three decades (2020-1990) Built_up area is increased by 440Ha (44%), Agriculture land is
decreased by -506Ha (49.6%) and Forest land is decreased Figure 11 show Ambo Town urban
land use land cover changes.
2. To identify major deriving forces of land cover changes of Ambo Town between 1990
and 2020
The determination of driving factors of land-use change in the Ambo Town was accomplished by
interviewing municipality Town planners to obtain deeper insights on LULC change dynamics
and Ground–truth data were gathered on the field to be used for image classification and
verification of the satellite imagery. Focus group discussions (FGDs) and key informants
interview (KII). The findings indicate that LULC changes in Ambo Town are a result of diverse
interrelated factors that operate differently. Migration coupled with natural increase results in
population growth which increases the amount and intensity of pressure exerted on resources and
consequently changes the state of the land. Driving factors of LULC change were grouped into
proximate and underlying causes as proposed by Geist and Lambin (2002). Based on interviews
and document analysis, proximate causes were identified as infrastructure, agriculture, and
forestry changes and underlying causes as political, demographic, economic, technological, and
cultural factors. To understand these drivers, the Focus group discussions (FGDs) and key
informants interview were adapted to show how driving factors lead to human activities which
exert pressure on resources resulting in various states of the environment which have significant
impacts and require responses. Strategies and policies based on responses to major drivers of
LULC and their impacts are therefore recommended to avoid undesirable impacts of changes in
LULC.
3. to model future land-use dynamics under two different scenarios.

This objective was achieved by conducting a literature review on land-use change models and
selecting a model that could be adapted in the study area. Modeling approaches were combined
into Cellular Automata models. Cellular Automata models that were shortlisted were Cellular
63
Automata (CA) and Markov which was referred to as CA_Markov. These models were
shortlisted based on a multitude of publications and literature which suggests their wide
applications in various topics in different regions and countries. Commonly used models are the
modeling techniques embedded in IDRISI is Land Change Modeler (LCM), Cellular Automata
(CA), Markov Chain, CA-Markov, GEOMOD, and STCHOICE (Eastman, 2006). For this
research Land change Modeler is used. Land change prediction in Land Change Modeler is an
empirically driven process that moves in a stepwise fashion from Change Analysis to Transition
Potential Modeling from Transition potential to Change Prediction.
This was achieved by using GIS software to analyze LULC maps derived from remote sensing
imagery. The LULC quantitative analysis results indicate that there were significant LULC
changes between 1990 and 2020 characterized by a decline in Agriculture. In contrast,
urban/settlement classes exhibited increases. The LULC change results also show that there was
an Ambo Town increase in built-up areas with about 56.86% increase. However, analyses of
individual district municipalities' LULC changes reveal that the increase in settlement areas was
concentrated in the Ambo Town area. The increase in urban areas was due to rising infrastructure
demands generated by population growth and informal expansion.
Discussion
Land use and land cover changes have a wide range of consequences at all spatial and temporal
scales. Because of these effects and influences, it has become one of the major problems for
environmental change as well as natural resource management. Identifying the complex
interaction between changes and its driver’s Focus group discussions (FGDs), key informants
interview (KII) and field observation is important to predict future developments, set decision-
making mechanisms, and construct alternative scenarios. This study is prepared to Quantify and
Modeling Ambo Town Urban Land use land cover changes and Predict the future LULC maps to
lead the planning authorities for sustainable development using the geospatial tools and
techniques. While making this research some research questions were defined to define the scope
64
of work. After finishing the study the answers to those questions were summarized in this section
to understand the success of the study both in methods and data used for the study and the result
of the study.
CHAPTER FIVE
5. CONCLUSSION AND RECCOMMENDATION.
5.1INTRODUCTION
Based on the overall of this research works, it is possible to conclude the following points.
Remotely sensed images are vital in land use and land cover change detection as it provides
spatial and temporal information of the land use land cover condition of the Ambo Town. The
land use and land cover change detection method used in this research posted classification
Comparison which is very important in knowing from- to change LULC changes have a wide
range of consequences at spatial and temporal scales. Because of these effects and influences, it
has become one of the major problems for environmental change as well as natural resource
management. Identifying the complex interaction between changes and its drivers over space and
65
time is important to predict future developments, set decision-making mechanisms, and construct
alternative scenarios Land use and land cover is very dynamic and has to be monitored at regular
intervals for sustainable development thus it has become a central component in current
strategies for managing natural resources. The results of this project revealed the existence of
significant land use and land cover changes in the last 30 years. Especially the expansion of
Built_up informal way/grabbing land. Land use and land cover change of Ambo Town resulted
that considerable change has occurred within thirty (30) years from 1990 to 2020. Though the
period of 1990 from 2020 there a dramatic change in several LULC categories including
Built_up area (5.8 km2.) or (56.86%), land 0.5km2. or (4.60%) area increased and Agriculture
land(4.94km2) or (48.94%) and Forest land(0.07km2) or (0.32%) area decreased .Accordingly,
more land was brought under Built_up area and e Bare land. The work output stated that an
increase in built_up area and bare land coverage of the town resulted in population pressure on
land and there are good government policies for encouraging reforestation programmed the
town.
CA_Markov modeling approach was also used to predict a land-use change in the future. To use
the model for short-and long-term forecasting, the predictive ability of the model was also
assessed. Once the predictive ability of the model was evaluated, a long-term land use and land
cover change forecasting was carried out. Regardless of the factors that might have had an
impact on the simulation output, it was inferred that the current trend of expansion of Built_up
area will continue in the future.
Generally, geographic information systems and remote sensing analysis enable sustainable
management of LULC change planning, wise decision making, monitoring of urban expansion
and development.
Urbanization is the most powerful and visible force that has fundamental changes in land use and
land cover change around the globe. High population increment leads to a quick expansion of
urban growth, causing changes in land use land cover in many urban areas especially in
developing countries like Ethiopia (Barros, 2004). The current urbanization process that is taking
place in developing countries indicates that this process needs considerable attention not only as
a basis for the transformation of societies in the developing countries but also for sustainable
development. (Redman ad Jones, 2004). Ethiopia has one of the fastest-growing urban
66
populations and least urbanized countries in the world. The urbanization level of Ethiopia was
11.4 percent (4.3 million) in 1984, 13.7 percent (7.4 million) in 1994, and in 2007 it increased to
16.2 percent (11.9 million). Oromiya West Showa Zone Surrounding of Oromiya National
Regional State. The urbanization level of Oromiya West Showa Zone Surrounding Ambo was
28.8 percent (CSA, 2007). It is higher than the urbanization levels of Oromiya National Regional
State and that of Ethiopia which is 13 percent and 16.2 percent respectively. (MUDC E, 2011).
The population of Ambo town was 4,138 in 1984; 10,027 in 1994; 63,873 in 2007 (Census) and
has grown to 375,000 in 2020 (estimated), showing that the population of the town has increased
by more than 90 doublings within the past 35 years (Ambo Town Communication office 2020).
The issues of Urban land use and land cover changes of Ambo Town become the most main
problems of societies in economic, social, and environmental such as health problems, cultural
disturbance, uncontrolled development, deteriorating environmental quality, loss of prime
agricultural lands, and protected Forest lands. (Ambo Town Communication offices, 2020).
5.2. Conclusion
The main objective of this Research to conduct Urban land use land cover changes of Ambo
Town using Land satellites of 1990, 2000, 2010, and 2020 and predictions for 2030. The Urban
land use land cover changes (Built_up area) of Ambo Town area in 1990 has been calculated as
140 hectare which is 13.73% of the total area of Town i.e.1020 hectare. In 2000 has been
calculated as 230hectare which is 22.54% of the total area of Town i.e. 1020 hectares. In 2010
has been calculated as 350 hectares which are 34.31% of the total area of Town i.e. 1020
hectares. In 2020 has been calculated as 580 hectare which is 56.86% of the total area of Town
i.e.1020.8hectare. For the predicted 2030 year Built_up area 5hectare which is 58% of the total
area i.e. 1020 hectare. Ambo Town Urban land use land cover (Built_up area) increased from
100hectare (1%) to 440hectare (43%) for three decades (1990-2020) and increased from
4100hectare (46%) to 5200hectare (58%) for three decades of future predicted one. Population
growth has a direct impact on urban land use land cover changes (Built_up area). Ambo Town
Population in 1990 is 7672 and Built-up area is 140 hectare. In 2000 the Population is 4,879 and
the Built_up area is 230hectare. In 2010 Population is 70,580 and the Built_up area is
350hectare. Finally, in 2020, the population is 143,000, and the Built_up area is 580 hectare. In
this case when the population of the city in 1965 is 27,812 the Built_up of the city was 250
hectares. The rapid population growth of Ambo Town is a result of demographic effects that
67
include rural to urban migration and natural population growth in the Town. Natural population
growth is a major element in urban growth for all countries, and migration of rural to urban
contributes to the fast growth of the urban population in many developing countries (Gugler,
1996). Ambo Town Urban land use land cover change (Built_up area) more expanded in Ambo
kora, Sanqele fi faris and Awaro,Gosuu and Areda kebele to Ambo Town Administration that
established in 1923G.C that exist in Huluka and Teltale river direction fow are slow to expand
kebeles. The negative effects of urban expansion highly increase the land use and land cover
changes and become many problems such as uncontrolled development, deteriorating
environmental quality, loss of prime agricultural lands, displacement of farm communities, land
degradation, enclosing surrounding rural land to the urban territory, overexploitation of natural
resources and conflict. Ambo Town fails in job creation for jobless such as graduated students
and youth those lives in town, insufficient services, house, and infrastructure for societies. Thus
the problems of negative effects and Town problems collectively create poor quality and the
overcrowded living condition forced dynamic land use the land cover of negative urban
expansion such as illegal construction of the house on government and farmland, and squatter
settlement in sensitive areas like solid waste disposal site. So to minimize the social, economic,
and environmental problems of Ambo Town, using Factors and Constraints such as Land use
land cover types, distance from Roads, Slope, distance from Built_upt areas, and distance from
River cross in Towns such as Huluka and Teltale, selecting the most appropriate site for
Housing/Residential area is an important consideration for dwellers of the city. The selection
must be based not only on current needs but also on projected needs.
Generally, geographic information system and remote sensing analysis enables for sustainable
managements of LULC change planning, wise decision making, monitoring of urban expansion
and development.
5.3. Recommendations
5.3.1 Recommendations for Future similar Research Work

Land use and land cover change (LULC) mapping and detection of changes shown here may not
provide the real figure of classes due to the low resolution of the imagery but it serves as a base
to understand the patterns and magnitude of LULCCs in the area. Therefore such LU and LC
detections using high-resolution satellite images would be more dependable. Rapid built_up area
68
increase has expressed a major role affecting LULC change and there should be strategic
planning to monitor unexpected urban expansions of the town from concerned governmental and
nongovernmental bodies (offices). Population increase has expressed a major role in LULC
change and there should be strong family planning awareness creation campaigns with adequate
health services from Concerned governmental and nongovernmental bodies (offices). To
minimize the problems of landless youths, it will be very important to create and strengthen off-
farm income-generating activities due to the limited capacity of the land to accommodate the
additional population. Since the most important factor of the land use and land cover change in
the study area was an increase in population, continuing the efforts of introducing family
planning to make the people aware of consequences of population pressure on land resources
should be carried out intensively Geographical information system and remote sensing
technology for change detection analysis is very important for the development of one country. It
is very important to deliver frequently. This job demands professional experts who are
accountable for this specific career. However, important it is human power dedicated to this wing
of development is very limited in skill and number if it is based on Ambo Town GIS data
management. Therefore, it requires due regard to handle appropriately. Ethiopian Government
has given due attention to forest development and conservation considering its significance to the
national economy, food security, and sustainable development. recommendations for Future
should be similar Research Work
1. Verification of Older Satellites images to validate accuracy assessment of classified images is

not possible to visit the field to find out the actual land cover types. To solve the problems with
older satellite images, the historical base maps of similar years should be collected.
2. Incorporating socio-economic data, land policy, biophysical and human factors (population
density, technology, political) could improve the performance of land-use models for future
predictions.
3. Only simulating the future land cover map is not fruitful, until further steps are taken. Some
more important research works can be recommended like- trying to compare the predicted map
with the existing Master plans of the city. If there any conflicts exist, then it should be taken into
special consideration.
69
4. Further studies can be conducted, find out whether the Land Change Modeler (LCM), Earth
Trends Modeler (ETM), GEOMOD, and CA-MARKOV kinds of models are best and suitable
for future prediction.
5.3.2 Recommendations for whom it concern

Predicting the future of urban land use land cover change is not an easy task, for the reason that
urbanization is a complex and dynamic system. Numerous physical, social, economic, and
environmental factors put an impact on shaping the urban form of a city. Considering all these
driving and influencing factors, it is a difficult and confusing job to predict the future of any city
or urban area. The future land cqover map of Ambo Town (2030) has been generated, in this
research, using MLP Markov modeling techniques.
1. Ambo Town Population is rapidly growing in the last three decades and the town fails in job
creation, insufficient services, houses, and infrastructure for societies. This makes many
problems that cause dynamic urban land use land cover changes. So Ambo Town Administration
and regional government should manage and give serious attention.
2. According to this research results around 55% of Built_up areas are not constructed in the
suitable site of a residential area, as a result consequently faced many problems and high risk. So
Ambo Town Urban planner, land administration offices, and environmental protection offices
should organize together to solve the social problems and avoid risks.
3. Ambo Urban planners should consider standard factors and constraints while selecting new
residential areas like distance from River, distance from the protected area like Huluka and
Teltale, and line and hazard areas by planning and giving awareness for society to save social,
economic, and environmental problems.
4. Ambo Town administration should give attention to the main problems of housing,
infrastructure services, and losses of Agriculture land by modifying land policy and giving land
for housing.
70
REFERENCE
Ackom, E. K., Amaning, K., Samuel, A., & Odai, N. (2020). Monitoring land - use and land -
cover changes due to extensive urbanization in the Odaw River Basin of Accra , Ghana ,
1991 – 2030. Modeling Earth Systems and Environment, 0123456789.
https://doi.org/10.1007/s40808-020-00746-5
Al-Ahmadi, F. S., & Hames, A. S. (2009). Comparison of four classification methods to extract
land use and land cover from raw satellite images for some remote arid areas, Kingdom of
Saudi Arabia. Journal of King Abdulaziz University, Earth Sciences, 20(1), 167–191.
https://doi.org/10.4197/Ear.20-1.9
Araya, Y. (2009). Urban Land Use Change Analysis and Modeling: A Case Study of Setúbal and
Sesimbra, Portugal. September, 72.
Ayele, A. (2017). Land Use and Land Cover Change Detection Analysis using Remote Sensing
Techniques : The Case of Hawassa Town , Southern Ethiopia. A Thesis Project, Hawassa
71
University, Ethiopia.
Behailu, A. (2010). Land Use and Land Cover Analysis and Modeling in South Western
Ethiopia: The Case of Selected Resettlement Kebeles in Gimbo Woreda. A Thesis Project
Submitted to Addis Ababa Univeristy, Ethiopia.
Bekele, T. (1993). Vegetation ecology of remnant Afromontane forests on the Central Plateau of
Shewa, Ethiopia. In Acta Phytogeographica Suecica 79 (Vol. 79).
Berhanu, A. (2017). Vegetation Ecology and Conservation Status of Evergreen Afromontane
Forest Patches in Awi Zone of Amhara Region , Northwestern Ethiopia. 188.
Cervantes-Godoy, D., Dewbre, J., PIN, Amegnaglo, C. J., Soglo, Y. Y., Akpa, A. F., Bickel, M.,
Sanyang, S., Ly, S., Kuiseu, J., Ama, S., Gautier, B. P., Officer, E. S., Officer, E. S.,
Eberlin, R., Officer, P., Branch, P. A., Oduro-ofori, E., Aboagye Anokye, P., … Swanson,
B. E. (2014). The future of food and agriculture: trends and challenges. In The future of food
and agriculture: trends and challenges (Vol. 4, Issue 4). https://doi.org/10.2307/4356839
Deng, X., Jiang, Q., Zhan, J., He, S., & Lin, Y. (2010). Simulation on the dynamics of forest area
changes in Northeast China. Journal of Geographical Sciences, 20(4), 495–509.
https://doi.org/10.1007/s11442-010-0495-0
Eastman, J. R. (2003). IDRISI Kilimanjaro: Guide to GIS and Image Processing. Clark
University: Worcester, MA, USA, 1(May), 87–131.
https://doi.org/10.1109/TGRS.2002.802519
ERDAS. (2009). ERDAS Field Guide TM - Tutorial. November, 810.
Fekadu, D. (2017). Addis Ababa Institute of Technology Addis Ababa Institute of Technology.
June, 1–100.
Firdaus, R. (2014). Doctoral Dissertation Assessing Land Use and Land Cover Change toward
Sustainability in Humid Tropical Watersheds , Indonesia Assessing Land Use and Land
Cover Change toward Sustainability in Humid Tropical Watersheds , Indonesia. March.
Foody, G. M. (2001). Monitoring the magnitude of land-cover change around the southern limits
of the Sahara. Photogrammetric Engineering and Remote Sensing, 67(7), 841–847.
For, G., & Agents, D. (2016). Soil and Water Conservation in Ethiopia.
Golmehr, E. (2009). Current application of remote sensing techniques in land use mapping: A
case study of Northern parts of Kolhapur district, India. Bulletin of Geography, 12(2006),
5–15. https://doi.org/10.2478/v10089-009-0001-6
Green, E. P., Mumby, P. J., Edwards, A. J., & Clark, C. D. (2000). Remote Sensing Handbook
for Tropical Coastal Management. Coastal Management Sourcebooks 3, January 2000, x +
316.
Han, H., Yang, C., & Song, J. (2015). Scenario simulation and the prediction of land use and
land cover change in Beijing, China. Sustainability (Switzerland), 7(4), 4260–4279.
https://doi.org/10.3390/su7044260
Hansen, M. C., & Loveland, T. R. (2012). A review of large area monitoring of land cover
change using Landsat data. Remote Sensing of Environment, 122, 66–74.
72
https://doi.org/10.1016/j.rse.2011.08.024
Haque, I., & Basak, R. (2017). The Egyptian Journal of Remote Sensing and Space Sciences
Land cover change detection using GIS and remote sensing techniques : A spatio-temporal
study on Tanguar Haor , Sunamganj , Bangladesh. The Egyptian Journal of Remote Sensing
and Space Sciences, 20(2), 251–263. https://doi.org/10.1016/j.ejrs.2016.12.003
Ismail, M. H. (2014). Evaluating supervised and unsupervised techniques for land cover
mapping using remote sensing data Evaluating supervised and unsupervised techniques for
land cover mapping using remote sensing data. September.
Karsidi, A. (2004). SPATIAL ANALYSIS OF LAND USE / LAND COVER CHANGE DYNAMICS
USING REMOTE SENSING AND GEOGRAPHIC INFORMATION SYTEMS : A Case
Study in the down stream and surroundings of the Ci Tarum watershed of. May.
Kindu, M. (2013). SYNTHESIS OF RESEARCH ON LAND USE AND LAND COVER. 1–39.
Kindu, M., Schneider, T., Döllerer, M., Teketay, D., & Knoke, T. (2018). Scenario modelling of
land use/land cover changes in Munessa-Shashemene landscape of the Ethiopian highlands.
Science of the Total Environment, 622–623, 534–546.
https://doi.org/10.1016/j.scitotenv.2017.11.338
Koranteng, A., Zawila-Niedzwiecki, T., & Adu-poku, I. (2016). Remote Sensing Study of Land
Use / Cover Change in West Africa. Journal of Environment Protection and Sustainable
Development, 2(3), 17–31.
Leemhuis, C., Thonfeld, F., Näschen, K., Steinbach, S., Muro, J., Strauch, A., López, A.,
Daconto, G., Games, I., & Diekkrüger, B. (2017). Sustainability in the food-water-
ecosystem nexus: The role of land use and land cover change for water resources and
ecosystems in the Kilombero Wetland, Tanzania. Sustainability (Switzerland), 9(9).
https://doi.org/10.3390/su9091513
Li, X., Wang, Y., Li, J., & Lei, B. (2016). Physical and socioeconomic driving forces of land-use
and land-cover changes: A Case Study of Wuhan City, China. Discrete Dynamics in Nature
and Society, 2016(2014). https://doi.org/10.1155/2016/8061069
Lu, D., & Weng, Q. (2007). A survey of image classification methods and techniques for
improving classification performance. International Journal of Remote Sensing, 28(5), 823–
870. https://doi.org/10.1080/01431160600746456
Mahmoud, H., & Divigalpitiya, P. (2017). Modeling Future Land Use and Land-Cover Change
in the Asyut Region Using Markov Chains and Cellular Automata. Green Energy and
Technology, November, 99–112. https://doi.org/10.1007/978-3-319-44899-2
Muzein, B. S. (2006). Remote Sensing & GIS for Land Cover / Land Use Change Detection and
Analysis in the Semi-Natural Ecosystems and Agriculture Landscapes of the Central
Ethiopian Rift Valley. November 2006.
Pickett, S. T. A., Cadenasso, M. L., Rosi-Marshall, E. J., Belt, K. T., Groffman, P. M., Grove, J.
M., Irwin, E. G., Kaushal, S. S., LaDeau, S. L., Nilon, C. H., Swan, C. M., & Warren, P. S.
(2017). Dynamic heterogeneity: a framework to promote ecological integration and
hypothesis generation in urban systems. Urban Ecosystems, 20(1), 1–14.
73
https://doi.org/10.1007/s11252-016-0574-9
Roy, S., Farzana, K., Papia, M., & Hasan, M. (2015). Monitoring and Prediction of Land
Use/Land Cover Change using the Integration of Markov Chain Model and Cellular
Automation in the Southeastern Tertiary Hilly Area of Bangladesh. International Jornal of
Sciencies:Basic and Applied Researchs, 24(October), 125–148.
S, R. C. N., Haroun, A., Adam, M., Elhag, A. M. H., & Abdelrahim, M. (2013). Accuracy
Assessment of Land Use & Land Cover. International Journal of Scientific and Research
Publications, 3(5), 1–6.
Shalaby, A., & Tateishi, R. (2007). Remote sensing and GIS for mapping and monitoring land
cover and land-use changes in the Northwestern coastal zone of Egypt. Applied Geography,
27(1), 28–41. https://doi.org/10.1016/j.apgeog.2006.09.004
SIraj, M. mmo, HZang, Ke., & MOges, K. (2018). RETROSPECTIVE ANALYSIS OF LAND
USE LAND COVER DYNAMICS USING GIS AND REMOTE SENSING IN CENTRAL
HIGHLANDS OF ETHIOPIA. Journal of Landscape Ecology, 11(2), 31–52.
Tizora, P. C. (2018). Modelling land use and land cover change in the western cape of province.
Dissertation Submitted to University of Pretoria, South Africa, February.
Udagepola, K. P., Xiang, L. I., Xiaozong, Y., & Wijeratne, A. W. (2006). MSRIC : A Model for
Spatial Relations and Integrity Constraints in Topographic Databases. Proceeding of 5th
WSEAS Int. Conf.on Artificial Intelligence, Knoowlege Engineering and Databases, Madrid
Spain, February 15-17.
Yeh, A. G. O., & Li, X. (1998). Sustainable land development model for rapid growth areas
using GIS. International Journal of Geographical Information Science, 12(2), 169–189.
https://doi.org/10.1080/136588198241941
To Predict Ambo Town projected Land use land cover classes of 2030.
Using transition Potential of 2010-2020 future projected (Predicted) Ambo Town LULC in 2030
is modeled and mapped in fig 4-31. Ambo Town Predicted will contains 5200Ha (58%) of Built-
up area, 1500Ha (17%) of Forest land, Agriculture land and 1000Ha (11%) of Bare land. Based
on this predicted result the Built-up area is increased from 4100Ha (46%) in 2020 to 5200Ha
(58%) in 2030, this is predicted result by considering factor.
74
Ambo Town LULC Classes of 1990, 2000,2010,2020
Distribution
75

Recently R@2013@T

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Recently R@2013@T

Uploaded by

Copyright:

Available Formats

ADDIS ABABA UNIVERSITY

ADDIS ABABA INSTITUTE OF TECHNOLOGY

SCHOOL OF CIVIL AND ENVIRONMENTAL ENGINEERING

DEPARTMENT OF GEODESY AND GEOMATICS

Research submitted to the school of Civil and Environmental Engineering Graduate

BY: TEDELA MOSISA

ADVISOR: WORKU ZEWDIE (Ph.D.)

ADDIS ABABA, ETHIOPIA

Submitted by: Tedela Mosisa ---------------------- -----------------

Principal Advisor: ---------------------- ----------------

Endorsed by-------------------------------------- ---------------- --------------

------------------------------------- --------------------- ---------

Post Graduate Program

Tedela Mosisa ------------------------- ------------- Name of Candidate Signature Date as Master

Prof .----------------------------- ---------------------- ---------------

Advisor Signature Date

Mr---------------------------- ----------------------- ------------------

Co-Advisor Signature Date

1.2. Statement of the problem..................................................................................................4

1.3 Research questions were:...................................................................................................5

1.4 Objective of the Study......................................................................................................5

1.4.1. The specific objectives of the proposal were:................................................................5

1.5 Significance of the research...................................................................................................5

2. REVIEW OF RELATED LITERATURE...................................................................................6

2.1. The concept and definition of land use land cover...............................................................6

2.2. Land use land cover dynamic................................................................................................7

2.3.2 Geographic Information System (GIS)...........................................................................9

2.4. Conceptual Frameworks of Urban land use land cover change........................................9

2.4.1. Factors of urbanization and urban expansion.................................................................9

2.4.1.1. Demographic factors...................................................................................................9

2.5. Image Classification........................................................................................................10

2.5.2. Types of Image Classification......................................................................................12

2.5.2.2. Supervised Classification..........................................................................................12

2.5.3. Image Enhancement.....................................................................................................13

2.5.4. Dynamic Methods........................................................................................................13

2.6. Post Classification Dynamic...............................................................................................13

2.7. Accuracy Assessment.........................................................................................................14

2.8. Modeling LULC Change....................................................................................................14

2.8.1. Modeling LULC Dynamic with CA-Markov...............................................................15

3. MATERIALS AND METHODOLOGY..............................................................................18

3.1. Description of the Study Area.............................................................................................18

3.1.2. Climate of the study area..............................................................................................18

3.2 Data Source..........................................................................................................................20

3.2.1 Data collection...............................................................................................................20

3.4. Methods of data analysis.....................................................................................................21

3.4.1. Materials and software.................................................................................................21

3.4.2. LULC Classification.....................................................................................................22

3.5. Digital Image Processing........................................................................................................23

3.5.2 Image Enhancement..........................................................................................................24

3.5.3 Image Transformations.....................................................................................................25

3.6. Image classification.............................................................................................................25

3.6.1. Pixel-based classification.................................................................................................26

3.6.2 Training Site Selection......................................................................................................26

3.6.3. Supervised and Unsupervised Classification...................................................................26

3.7. Accuracy Assessment.........................................................................................................28

3.7.1 Overall accuracy...............................................................................................................30

3.7.4. Producer Accuracy...........................................................................................................30

3.7.5. User Accuracy..................................................................................................................30

3.7.6. Kappa coefficient.............................................................................................................30

3.8. Dynamic process.................................................................................................................31

3.8.1. Modeling LULC dynamic............................................................................................31

3.8.2. Modeling LULC change with CA-Markov..................................................................32

3.8.3. Model Calibration.........................................................................................................32

3.8.4. Criteria Development.......................................................................................................33