ADDIS ABABA UNIVERSITY

COLLEGE OF NATURAL AND COMPUTATIONAL SCIENCES

SCHOOL OF EARTH SCIENCES

MAPPING EFFECTS OF URBAN BLUE-GREEN LANDSCAPES ON LAND

SURFACE TEMPERATURE USING GEO-SPATIAL TECHNIQUES: THE CASE OF
ADDIS ABABA, ETHIOPIA

By:
Neway Kifle ID: GSR/3066/2012
Advisor:
Binyam Tesfaw Hailu (Ph.D.)
Addis Ababa, Ethiopia
June, 2021
 ADDIS ABABA UNIVERSITY
COLLEGE OF NATURAL AND COMPUTATIONAL SCIENCES
SCHOOL OF EARTH SCIENCES

MAPPING EFFECTS OF URBAN BLUE-GREEN LANDSCAPES ON

LAND SURFACE TEMPERATURE USING GEO-SPATIAL
TECHNIQUES: THE CASE OF ADDIS ABABA, ETHIOPIA

A THESIS SUBMITTED TO THE SCHOOL OF GRADUATE STUDIES OF

ADDIS ABABA UNIVERSITY IN PARTIAL FULFILLMENT OF THE
REQUIREMENTS FOR THE DEGREE OF MASTERS OF SCIENCE IN
REMOTE SENSING AND GEO-INFORMATICS

By:
Neway Kifle
ID: GSR/3066/2012
Advisor:
Binyam Tesfaw Hailu (Ph.D.)

Addis Ababa, Ethiopia

June, 2021
 Addis Ababa University

School of Graduate Studies

Thesis Approval Form

This is to certify the thesis prepared by Neway Kifle Bekele entitled “mapping effects of urban
blue-green landscapes on land surface temperature using geo-spatial techniques. The case of
Addis Ababa, Ethiopia” is submitted in partial fulfillment of the requirements for the Degree of
Master of Science in Remote Sensing and Geo-informatics compiles with the regulations of the
university and meets the accepted standards to originality and quality.

Neway Kifle ____________ _______/______/_______

Candidate Signature Date
Signed by the Examining Committee:
Dr. Binyam Tesfaw ____________ ______/______/_______

Adviser Signature Date

Dr. Dagnachew Legesse ____________ ______/______/________

Examiner Signature Date
Dr. Tibebu Kassawmar ____________ ______/______/________
Examiner Signature Date
Dr. K.V. Suryabhagavan ____________ ______/______/________
Chairman Signature Date
Dr. Balemwal Atnafu ____________ ______/______/________
Head, Department of Earth Sciences Signature Date
 DECLARATION
I hereby declare that the dissertation entitled “mapping effects of urban blue-green landscapes
on land surface temperature using geo-spatial techniques. The case of Addis Ababa, Ethiopia”
has been carried out by me under the supervision of Dr. Binyam Tesfaw Hailu, Assistance
professor of Remote Sensing and Geo-Informatics, School of Earth Sciences, Addis Ababa
University during the year 2019-2021 as a part of Master of Science programme in Remote Sensing
and Geo-informatics. I further declare that this work has not been submitted to any other university
or institution for the award of any degree or diploma.

Addis Ababa, Ethiopia

Date: June, 2021

Neway Kifle
 Acknowledgments
First and for most, praised be to “Almighty God” for his will and help in the whole part of my life.

I am extremely indebted to my advisor Binyam Tesfaw Hailu (Ph.D.), Assistance Professor of

Remote Sensing and Geo-Informatics, School of Earth Sciences, Addis Ababa University for his
patient, valuable, constructive, and progressive comments and advice, and criticism throughout
this work.

I am also grateful to the Central Statistical Agency (CSA) of Ethiopia for sponsoring me to pursue
my post-graduate study in remote sensing and geo-informatics Stream, under the college of natural
and computational sciences at Addis Ababa University.

I would like to extend my gratitude to Balemwal Atnafu (Ph.D.), Head, School of Earth Sciences,
Addis Ababa University for frequently writing letters of support for various data providers; and
the School of Earth Sciences, Addis Ababa University, for letting us access Remote Sensing and
GIS Lab facilities.

My heartfelt gratitude further goes to Regional Centre for Mapping of Resources for Development/
Global Monitoring for Environment and Security (RCMRD/GMES), and Africa project for
covering my tuition fees, and research costs which helped me for the successful completion of this
thesis.

I am also grateful to my friend Gizachew Kabite Wedajo (Ph.D.), Associate Researcher II at

Department of Remote Sensing Research and Development, Ethiopian Space Science and
Technology Institute for reading and giving feedbacks on earlier drafts of this document; and for
all my Remote Sensing and Geo-Informatics stream classmates for their contribution of ideas and
encouragement during the study and this research work.

i
 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Table of Contents
Acknowledgments .......................................................................................................................... i

List of figures ................................................................................................................................ vi

List of Tables................................................................................................................................ vii

List of Appendices ...................................................................................................................... viii

Acronyms ...................................................................................................................................... ix

Abstract ......................................................................................................................................... xi

CHAPTER ONE ........................................................................................................................... 1

1. INTRODUCTION................................................................................................................. 1

1.1 Background ........................................................................................................................... 1

1.2 Problem Statement ................................................................................................................ 3

1.3 Objectives of the study.......................................................................................................... 5

1.3.1 Main objective ............................................................................................................... 5

1.3.2 Specific objectives ......................................................................................................... 5

1.4 Research questions ................................................................................................................ 5

1.5 Significance of the study....................................................................................................... 5

1.6 Delimitation of the study ...................................................................................................... 6

1.7 Organization of the thesis ..................................................................................................... 6

CHAPTER TWO .......................................................................................................................... 7

2. LITERATURE REVIEW ......................................................................................................... 7

2.1 Urban green landscapes ........................................................................................................ 7

2.1.1 Benefits of urban green landscapes................................................................................ 7

2.1.2 Urban green landscapes as a foe .................................................................................... 9

2.1.3 Classification of urban green landscapes ....................................................................... 9

2.1.4 An Overview of urban green landscapes in Addis Ababa ............................................ 10

ii
 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

2.2 Urban blue landscapes ........................................................................................................ 13

2.2.1 Benefits of urban blue landscapes................................................................................ 13

2.3 Land use land cover change ................................................................................................ 14

2.4 Remote sensing and GIS for urban land use land cover change ......................................... 14

2.5 Image classification ............................................................................................................ 15

2.5.1 Supervised image classification ................................................................................... 15

2.5.2 Unsupervised image classification ............................................................................... 16

2.5.3 Hybrid method ............................................................................................................. 16

2.5.4 Object-based image analysis (OBIA) .......................................................................... 17

2.5.4.1 Image segmentation .................................................................................................. 17

2.6 Land Surface temperature ................................................................................................... 18

2.7 Land surface temperature response to land use and land cover dynamics ......................... 18

2.8 Thermal remote sensing for land surface temperature ........................................................ 19

2.9 Methods of land surface temperature retrieval ................................................................... 20

2.9.1 Mono-window algorithm (MWA) ................................................................................ 20

2.9.2 Single-channel (SC) method ........................................................................................ 21

2.9.3 Split window (SW) algorithm ...................................................................................... 22

CHAPTER THREE .................................................................................................................... 23

3. STUDY AREA, MATERIALS, AND METHODS ............................................................... 23

3.1 Study area description ......................................................................................................... 23

3.1.1 Location ....................................................................................................................... 23

3.1.2 Basic population characteristics ................................................................................... 24

3.1.3 Physiography and drainage .......................................................................................... 24

3.1.4 Climate ......................................................................................................................... 26

3.1.5 Soil ............................................................................................................................... 28

iii
 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

3.2 Materials and software ........................................................................................................ 29

3.2.1 Materials ...................................................................................................................... 29

3.2.1.1 Primary data .............................................................................................................. 29

3.2.1.2 Secondary data .......................................................................................................... 31

3.2.2 Software ....................................................................................................................... 31

3.3 Methods............................................................................................................................... 31

3.3.1 Image pre-processing ................................................................................................... 31

3.3.2 Image enhancement ..................................................................................................... 32

3.3.3 Image classification ..................................................................................................... 32

3.3.3.1 Nomenclature of land cover classes .......................................................................... 32

3.3.3.2 Object-based image analysis (OBIA) ....................................................................... 34

3.3.3.3 Classification accuracy assessment........................................................................... 37

3.3.4 LULC change detection ............................................................................................... 38

3.3.5 Land surface temperature retrieval .............................................................................. 39

3.3.5.1 Conversion of DNs to at sensor spectral radiance .................................................... 39

3.3.5.2 Top of atmosphere brightness temperature ............................................................... 40

3.3.5.3 Land surface emissivity estimation ........................................................................... 40

3.3.5.4 LST estimation .......................................................................................................... 40

3.3.5.5 LST result validation................................................................................................. 41

3.3.6 Land use land cover indices ......................................................................................... 41

3.3.7 Statistical analysis ........................................................................................................ 42

CHAPTER FOUR ....................................................................................................................... 44

4. RESULTS................................................................................................................................. 44

4.1. Object-based image classification accuracy assessment.................................................... 44

4.2. Spatiotemporal dynamics in land use/and cover ................................................................ 45

iv
 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

4.2. 1 Land use/cover conversion matrix .............................................................................. 47

4.3. Land surface temperature validation result ........................................................................ 48

4.4. Spatiotemporal pattern of land surface temperature .......................................................... 48

4.5. Land surface temperature variation among LULC classes ................................................ 49

4.6 Statistical analysis ............................................................................................................... 50

4.6.1 NDVI-LST relationship ............................................................................................... 51

4.6.2 NDBI-LST relationship ............................................................................................... 52

4.6.3 MNDWI-LST relationship ........................................................................................... 54

4.6.4 Altitude-LST relationship…………………………………………………………….55

CHAPTER FIVE ........................................................................................................................ 56

5. DISCUSSIONS ........................................................................................................................ 56

5.1 Land use and land cover dynamics ..................................................................................... 56

5.2 Land use/Land cover conversion matrix ............................................................................. 57

5.3 Comparison between model LST and MODIS LST product .............................................. 57

5.4 LST trend and spatial variation ........................................................................................... 58

5.5 LST variation among land use/land cover classes .............................................................. 58

5.6 The relationship of LST with NDVI, NDBI, MNDWI and altitude ................................... 59

CHAPTER SIX ........................................................................................................................... 60

6. CONCLUSION AND RECOMMENDATIONS .................................................................. 60

6.1 Conclusion .......................................................................................................................... 60

6.2 Recommendations ............................................................................................................... 62

References .................................................................................................................................... 63

Appendices ................................................................................................................................... 77

v
 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

List of figures
Figure 2. 1: Picture of plantation forest on Mt. Entoto ................................................................ 10
Figure 2. 2: Picture of institutional forest in the old national palace ............................................11
Figure 2. 3: Picture Unity Park found in the National palace .......................................................11
Figure 2. 4: Field crop land in Addis Ababa ................................................................................ 12
Figure 3. 1: Location map of the study area ................................................................................ 23
Figure 3. 2: A) Elevation map and B) Slope map of the study area ............................................ 25
Figure 3. 3: Average monthly rainfall in Addis Ababa during the years 1998 to 2020 ............... 26
Figure 3. 4: Annual rain fall distribution in Addis Ababa during the years1998 to 2020 ............ 27
Figure 3. 5: Minimum, maximum, and mean monthly Temperature in AA from 1998 to 2020.. 28
Figure 3. 6: Soil map of the study area based on MoWR digital data ......................................... 29
Figure 3. 7: A picture of a built-up area in Addis Ababa ............................................................. 33
Figure 3. 8: An example green landscape within Friendship Square/Park .................................. 33
Figure 3. 9: A picture of onion farm in Akaki area, Southern, Addis Ababa ............................... 33
Figure 3. 10: A picture of the red ash excavation site in Southern Addis Ababa ......................... 34
Figure 3. 11: A picture of an artificial lake inside Friendship Square/Park ................................. 34
Figure 3. 12: An example of segmentation during 2006 with scale=30, shape=0.3 &
compactness=0.1 ........................................................................................................................... 35
Figure 3. 13: An example of the process tree with rule sets for 2021 classification ................... 36
Figure 3. 14: General workflow of the study ............................................................................... 43
Figure 4. 1: LULC map A) during 2006, B) during 2016 and C) during 2021 ........................... 46
Figure 4. 2: LST map A) during 2006, B) during 2016 and C) during 2021 ............................... 50
Figure 4. 3: NDVI map A) during 2006, B) during 2016 and C) during 2021 ........................... 51
Figure 4. 4: Scatter plots of NDVI and LST A) in 2006, B) in 2016 and C) in 2021 .................. 52
Figure 4. 5: NDBI map A) during 2006, B) during 2016 and C) during 2021 ........................... 53
Figure 4. 6: Scatter plots of LST and NDBI A) in 2006, B) in 2016 and C) in 2021 .................. 53
Figure 4. 7: MNDWI map A) during 2006, B) during 2016 and C) during 2021....................... 54
Figure 4. 8: Scatter plots of LST and MNDWI A) in 2006, B) in 2016 and C) in 2021 ............. 55
Figure 4. 9: Scatter plots of altitude and LST in 2006, 2016 and 2021…………………………55

vi
 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

List of Tables
Table 3. 1: Addis Ababa population by sub-city, sex composition, sex ratio, and population
density ........................................................................................................................................... 24
Table 3. 2: The classified slope of the study area ........................................................................ 25
Table 3. 3: Maximum, minimum and mean monthly temperature in AA from 1998 to 2020 ..... 27
Table 3. 4: Soil types of the study area ........................................................................................ 28
Table 3. 5: Satellite images and details of their respective bands used ........................................ 30
Table 3. 6: Segmentation parameters combinations for 3 images ................................................ 36
Table 3. 7: Meta data of thermal constants for Landsat 7 ETM+ (Band 6) & Landsat 8 TIRS
(Band 10) ...................................................................................................................................... 40
Table 4. 1: Accuracy assessment of LULC classification during 2006……………………........44
Table 4. 2: Accuracy assessment of LULC classification during 2016………………………….44
Table 4. 3: Accuracy assessment of LULC classification during 2021………………………….44
Table 4. 4: Land use/cover dynamics during 2006, 2016 and 2021 ............................................. 45
Table 4. 5: Land use/cover conversion matrix during 2006, 2016 and 2021 ............................... 48
Table 4. 6: Descriptive statistics of LST during 2006, 2016 and 2021 ........................................ 49
Table 4. 7: Descriptive statistics of LST among LULC Classes during 2006, 2016 and 2021 .... 49
Table 4. 8: R2 and r values of LST with NDVI, NDBI and MNDWI in 2006, 2016 and 2021 ... 52

vii
 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

List of Appendices
Appendix 1: Regression analysis between NDVI and LST during 2006, 2016, and 2021 ......... 77
Appendix 2: Regression analysis between NDBI and LST during 2006, 2016, and 2021 .......... 78
Appendix 3: Regression analysis between MNDWI and LST during 2006, 2016, and 2021 ..... 79
Appendix 4: Regression analysis between altitude and LST during 2006, 2016, and 2021…….80
Appendix 5: True color composite images of the study area A) SPOT 5 image during 2006, B)
Sentinel- 2A during 2016, and Sentinel-2A during 2021. ............................................................. 81
Appendix 6: A map showing conversion between LULC classes from 2006 to 2021 ................ 82
Appendix 7: Map of ground control points A) in 2006, B) in 2016 and C) in 2021……………83

viii
 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

List of acronyms
AA Addis Ababa

ASF Alaska Satellite Facility

ASTER Advanced Space borne Thermal Emission and Reflection Radiometer

AU African Union
AVHRR Advanced Very High-Resolution Radiometer
AZP Addis Zoo Park
CSA Central Statistical Agency
DEM Digital Elevation Model
DNs Digital Numbers
DOS Dark Object Subtraction
ECA Economic Commission for Africa
ENVI Environment for Visualizing Images
EPRDF Ethiopian People’s Revolutionary Democratic Front
ERDAS Earth Resource Data Analysis System
ESA European Space Agency
ESRI Environmental Systems Research Institute
ETM Enhanced Thematic Mapper
GBG Gulelle Botanic Garden
GLC Green Legacy Campaign
GPS Global Positioning System
IMW Improved Mono-Window
ISODATA Iterative Self Organizing Data Analysis
LRT Light Rail Transit
LSE Land Surface Emissivity
LST Land Surface Temperature
LULC Land Use/Land Cover
MNDWI Modified Normalized Difference Water Index

ix
 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

MODIS Moderate Resolution Imaging Spectro-radiometer

MoWR Ministry of Water Resource
MRS Multi-Resolution Segmentation
MTS Multi-Threshold Segmentation
MWA Mono-Window Algorithm
NDBI Normalized Difference Built-up Index
NDVI Normalized Difference Vegetation Index
NHS National Health System
NIR Near Infrared
NMA National Meteorological Agency
NOAA National Oceanic and Atmospheric Administration
OBIA Object-Based Image Analysis
OLI /TIRS/ Operational Land Imager/ Thermal Infrared Sensor
RS and GIS Remote Sensing and Geographic Information System
SC Single-Channel
SCP Semi-automatic Classification Plugin
SDS Spectral Difference Segmentation
SLC Scan-Line error Correction
SPOT Satellite Pour l’Observation de la Terre
SST Sea Surface Temperature
SWA Split Window Algorithm
SWIR Short Wave Infrared
TIR Thermal Infrared
TM Thematic Mapper
UHI Urban Heat Island
UN United Nations
USGS United States Geological Survey
WHO World Health Organization

x
 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Abstract
Uncontrolled, unplanned, and unprecedented urbanization characterizes most African cities.
Drastic changes in the urban landscape can lead to irreversible changes to the urban thermal
environment, including changes in the spatiotemporal pattern of the land surface temperature
(LST). Studying these variations will help us take urban climate change mitigation and adaptation
measures. This study is intended to map effects of urban blue-green landscapes on LST using geo-
spatial techniques in Addis Ababa, Ethiopia from 2006 to 2021. Object-based image analysis
(OBIA) method was applied for land use/land cover (LULC) classification using high-resolution
imagery from SPOT 5 and Sentinel 2A satellites. Moreover, LST was retrieved from the thermal
imageries of Landsat 7 ETM + (band 6) and Landsat 8 TIRS (band 10) using the Mono-Window
Algorithm (MWA). Furthermore, linear regression analysis was used to determine the relationship
of LST with normalized difference vegetation index (NDVI), normalized difference built-up index
(NDBI), and modified normalized difference water index (MNWI). Five major LULC classes were
identified namely, built-up, vegetation, urban farmland, bare land, and water. The result shows
that the built-up area was the most dominant LULC in the city and has shown a drastic expanding
trend with an annual growth rate of 4.4% at the expense of urban farmland, vegetation, and bare
land in the last 15 years. The findings demonstrated 53.7% of urban farmland, 48.1% of vegetation,
and 59.4% of bare land, was transformed into a built-up class from 2006 to 2021. The mean LST
showed an increasing trend, from 25.8oC in 2006 to 27.2oC and 28.2oC during 2016 and 2021
respectively. It was found that LST varied among LULC classes. The highest mean LST was
observed at bare land having an average LST value of 26.9oC, 28.7oC, and 30.1oC in 2006, 2016,
and 2021 respectively. While the lowest mean LST was recorded at vegetation with average LST
values of 24.3oC in 2006 and 26.0oC in 2021; and at water 25.5oC in 2016. The regression analysis
showed a strong negative correlation between NDVI and LST, a strong positive correlation
between NDBI and LST, and a weak negative correlation between MNDWI and LST. The findings
of this study have indicated that LULC alteration had contributed to the modification of LST in
Addis Ababa during the period. The regression analysis results further revealed that built-up area
and vegetation cover plays a decisive role in the variation of LST in the city compared to urban
surface water and altitude. The findings of this study will be helpful for urban planners and
decision-makers while planning and designing future urban blue-green innervations in the city.

Keywords: LULC, OBIA, LST, NDVI, NDBI, MNDWI

xi
 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

CHAPTER ONE

1. INTRODUCTION
1.1 Background
Urban blue-green landscapes, referring to all green and blue areas in the city (e.g., parks, private
gardens, rivers, ponds, and riverbanks) and the ecological services they provide, such as clean
water and air, food, fuel, hazards mitigation, and offers a favorable condition for human life (Lise
et al, 2017; UN-HABITAT, 2017). These landscapes include either water (blue), vegetation
(green), or both.

Urban blue-green landscapes are important for urban climate change mitigation and adaptation
measures and have a host of wider benefits to people and wildlife. As one of the urban landscape
elements, the green landscape is close to the natural environment in cities. It provides sunshade
with various forms of efficiency such as reducing the temperature, humidifying the air, improving
the partial microclimate, and hence positively affects the life quality of urban residents. Several
studies had indicated the eco-environmental advantages of urban green landscapes mainly in
releasing oxygen, consuming carbon dioxide, reducing the air and land surface temperature,
lowering noise level, rising air moisture, and lowering dust concentration. Moreover, urban green
landscapes protect biodiversity and lower air pollution through the process of photosynthesis,
transpiration and evapotranspiration, absorption, and buffer function. Urban green landscapes are
one of the best ways to improve the urban thermal environment (Su et. al, 2011). The presence of
green landscapes in cities can reduce depression, enhance physical fitness, and improve the health
and wellbeing of urban dwellers (University of Leeds, 2015). Likewise, urban blue landscapes
have also a great potential to mitigate the urban surface heat island effect (Wu et al., 2019).

The pressure on urban blue-green landscapes will most likely increase with the growing
urbanization. Fifty-four percent of the world’s population were residing in urban areas in 2014 and
it is estimated to reach 70 percent by 2050. Forecasts indicated that expansion of urbanization
together with the world’s overall population growth will possibly add additional 2.5 billion people
to urban settings by 2050. Much of the world’s urban population growth is expected to take place
in countries of developing regions, particularly Africa (UN, 2014).

Neway Kifle [2021] 1 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

LULC dynamics intensified by natural and anthropogenic processes drive biogeochemistry of the
ecology affecting the climate at regional as well as global levels. Radical changes in the LULC
specifically, the shrinkage in water bodies and vegetation cover due to human activities increases
the heat emission from atmospheric and land surface temperatures. Higher LST is predominantly
because of an increase in concentrated anthropogenic activities, paved surfaces, or barren lands
(Setturu et al., 2014).

Remote sensing particularly, thermal remote sensing is very significant in monitoring LST and
studying the thermal properties of LULC types along with their relationships (Abel Balew and
Tesfaye Korme, 2020). The basic idea is to retrieve LST from the thermal bands of satellite images
and determine its relationship with LULC classes. This method helps to overcome the weaknesses
associated with conventional methods by providing better spatial coverage and synchronicity
(Naeem et al., 2018).

Plenty of related earlier studies had demonstrated a few approaches to determine the relationship
between LST and urban green landscapes in the city. A few of them include: (i) examine the rate
and extent of urban expansion in Addis Ababa and determine the urban heat island effect from
1994 to 2011 (Ermias Teferi and Hiwot Abreha,2017); (ii) assess the cooling effect of parks in
mitigating urban heat island effect (UHI) in Addis Ababa from 1985 to 2015 (Samson Warkaye. et
al., 2018); (iii) examine the pace, extent, and pattern of urbanization and its impact on green spaces
in Bole sub-city of Addis Ababa from 1986 to 2000 (Mathias Tesfaye and Tebarek Lika, 2017) and
(iv) a comparative analysis to determine the relationship between LST and the composition,
configuration and spatial patterns of impervious surfaces and green spaces in Addis Ababa
(Ethiopia), Lagos (Nigeria), Lusaka (Zambia) and Nairobi (Kenya), (Simwanda et al., 2019).

Given the small number of earlier studies conducted in the city, their limited spatial and/or
temporal scale of analysis and type of methodology employed for LULC analysis (traditional
pixel-based method) and materials used (coarse resolution satellite imagery), further studies are
needed to develop adequate generalizations that are useful to map effects of urban blue-green
landscapes on LST in the city. Moreover, previous researches focused only on vegetation indices
to analyze the relationship between LST and land use land cover classes in the city. Furthermore,
the potential effect of urban blue landscapes, i.e., urban surface water, on LST is not well
understood in the city, although it lowers LST via evaporation (Wu et al., 2019).

Neway Kifle [2021] 2 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

1.2 Problem statement

Most African cities are under critical environmental and ecological threat due to uncontrolled
unprecedented and unplanned urban expansion. Thus, many of the African cities have emerged as
unplanned. As a result, environmental and ecological conditions in African cities are under
constant threat (Simwanda and Murayama, 2018). One of the critical environmental threats of
urbanization on urban residents is the UHI effect (Simwanda et al., 2019).

This unprecedented city growth is also evident in Addis Ababa and has reduced green landscapes
to the level that there is an indication of increasing surface and air temperatures in the city. The
problem is expected to get worse in the upcoming years due to the extensive construction of
buildings, houses, and infrastructures and the little attention given to urban green landscapes
(Mathias Tesfaye and Tebarek Lika, 2017).

In Addis Ababa, the built-up area continues to grow at a rate of 4.5 km2 per year, at the expense of
green landscapes while the remaining ecosystems get ever more polluted. The minimum threshold
value of green landscapes per capita as defined by the World Health Organization (WHO) is 9 m2.
However, Addis Ababa scores between 0.4 and 0.9 m2 per capita which is very poor. (UN-
HABITAT, 2017).

A typical characteristic of Addis Ababa and its development is that the successive Ethiopian
governments have used new urban projects, new priority areas, and shifting ideologies as material
embodiments of political rupture. For instance, Emperor Menelik is known for building the first
gravel road, the national palace, the national railway, and the marketplace which are all symbols
of modernization. Emperor Menelik was also known for introducing the Eucalyptus tree, one of
the most exotic species in Ethiopia. Emperor Haile Selassie’s regime constructed an increased
number of monuments that honor the Battle of Yekatit 12 and Adwa. The Holy Trinity Cathedral
Church and the statue of Lion of Judah were also constructed during Emperor Haile Selassie’s
regime. The Derg restored Meskel Square for its troop parades and other ceremonial occasions.
More recently, the Ethiopian people’s revolutionary democratic front (EPRDF) introduced the first
light rail transit (LRT) system in Sub-Saharan Africa. The EPRDF also introduced a housing
practice called a condominium in the city. Most recently, Prime Minister Abiy Ahmed’s
administration has announced the Beautifying Sheger (56 km river buffer restoration mega project)
and the construction of the 36-hectare luxury real estate complex at Lagahar (Biruk Terrefe, 2020).

Neway Kifle [2021] 3 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

The recent administration has also launched a new intervention called the Green Legacy Campaign
(GLC) that focuses on green space development in the country in general in the city in particular.
Hence, there exists a political shift in ideology and priority areas in the urban development of
Addis Ababa.

Launched in 2019 by Ethiopia Prime Minister Abiy Ahmed, the Green Legacy Campaign (GLC)
is part of the Government’s plan to plant 20 billion seedlings by 2024. According to the World
Agroforestry, (2020), Ethiopia planted 4 billion seedlings during the first GLC in 2019 and over 5
billion seedlings during the second GLC in 2020 organized across the country during the Ethiopian
rainy season. Addis Ababa City Government River Basins and Green Areas report on July 2020
shows that, in Addis Ababa alone, 7.8 million seedlings are planted during the second GLC
covering an area of 8,400 hectares.

The master plan of Addis Ababa city has developed strategies that integrate action plans for the
implementation of its various components, more specifically blue-green infrastructure which
includes formal parks, gardens, woodlands, rivers, river buffers, street plantations, and walkways.
Beautifying Sheger mega project clearly shows the government’s focus on blue-green
infrastructure and its urgent need. This mega project is expected to raise the green coverage of the
city from 0.3 sq.m/per capita to 7 sqm/per capita (Biruk Terrefe, 2020).

Immediately before the recent green interventions, there have been some attempts to develop green
spaces in the city. For example, 30.5 hectares of green space is developed at the Nations and
Nationalities square in the Nifas Silk-Lafto sub-city. Behind St. Estifanos church in Kirkos sub-
city, 5.2 ha of the recreational park is developed. In the Lideta urban renewal site, 4.3 ha of
recreational parks and 0.7 ha of recreational green space close to the Immigration office were
developed in the Lideta sub-city (Herslund et al., 2017).

Therefore, this study is intended to map effects of urban blue-green landscapes on land surface
temperature using geo-spatial techniques in Addis Ababa city of Ethiopia. This city is selected
based on the following criteria: (i) because of the government’s current urban blue-green space
interventions: Green Legacy and Beautifying Sheger Project; (ii) being the capital and largest city
in the country; (iii) being the main economic and commercial center of the country; and (iv)
experiencing rapid urbanization with the highest urban population than other cities in the country.

Neway Kifle [2021] 4 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

1.3 Objectives of the study

1.3.1 Main objective
The main objective of this study is to map effects of urban blue-green landscapes on land surface
temperature using geo-spatial techniques in the city of Addis Ababa, Ethiopia.

1.3.2 Specific objectives

 To estimate LST for Addis Ababa during the years 2006, 2016, and 2021;
 To map LULC using Object-Based Image Analysis (OBIA) and high-resolution images
for the city of Addis Ababa;
 To analyze the spatiotemporal pattern of LST and its variation among land cover classes;
 To study the relationship of LST with vegetation, built-up, and water indices and altitude.

1.4 Research questions

In line with the research objectives, the study has the following research questions:

 What is the LST of the study area during the years 2006, 2016, and 2021?
 Were there any major changes in the LULC classes within the study period?
 What are the spatiotemporal patterns of LST and its variation among LULC classes of the
study area?
 What is the correlation of LST with the land-use class indices (vegetation, built-up, and
water) and altitude?

1.5 Significance of the study

The study results include maps showing the spatiotemporal variations of LST, LULC, NDVI,
NDBI, and MNDWI in the city for the years 2006, 2016, and 2021. The result of this study also
provides quantitative information on LULC dynamics and LST in the study period. Therefore, it
is hoped that the study will serves as a benchmark for policymakers and urban planners to ensure
accessible and adequate urban-blue green landscapes and implement sustainable LST mitigation
strategies and create a conducive living environment in the rapidly urbanizing city of Addis Ababa
and beyond. Moreover, the results of this study will be used as initial input for interested groups
who will conduct future related studies in the area. It will also offer a better foundation on how
remote sensing and GIS technologies are applied in evaluating the urban-blue green landscape
effect on LST.

Neway Kifle [2021] 5 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

1.6 Delimitation of the study

The scope of this study is spatially delimited only to the administrative boundary of Addis Ababa
city, Ethiopia. In its temporal extent, the study is limited in investigating the urban blue-green
landscape effect on land surface temperature change using geospatial data for the last 15 years in
the city. The years 2006, 2016, and 2021 are selected to analyze the spatiotemporal pattern of
urban-blue green landscapes and their effect on LST. Moreover, theoretically, this study is
restricted in studying the spatiotemporal pattern of LULC dynamics and the resulting change in
LST, even though other factors are also affecting LST in the city.

1.7 Organization of the thesis

This thesis paper is organized into six chapters. Chapter one is devoted to the introductory part of
the paper where the background of the study, statement of the problem, objectives of the study,
research questions, significance, and delimitation of the study were described. A review of related
literature is treated under chapter two. The third chapter deals with the study area description, the
materials used and methods utilized in this study. In chapter four, results are presented and
analyzed. The results are discussed in chapter five. Finally, chapter six provides conclusions drawn
and the forwarded recommendations.

Neway Kifle [2021] 6 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

CHAPTER TWO
2. LITERATURE REVIEW
2.1 Urban green landscapes
Urban green landscape in the context of urban studies refers to several terms and definitions that
include green space, open space, urban greenery, public gardens, green infrastructure, public space,
and parks. Urban designers and planners, architects, landscape architects, and other professionals
have used these terms interchangeably. Moreover, it can be discussed from different points of view
such as design, policy, planning, management, and environmental psychology, and so on
(Rakhshandehroo et al., 2017).

Urban Green landscapes refer to those LULC that are covered with man-made or natural vegetation
in urban and other planning areas (Bilgili and Gökyer, 2012). These range from green on buildings
such as green roofs, balconies, and green walls; to green near built structures, such as street median
and side greening, railway banks, public square, playgrounds, house gardens, different types of
parks, cemeteries, institutional green spaces, sports facilities, and community gardens and
allotments; to more open kinds of green spaces like horticulture and farmland, shrublands and
woodlands (Roy et al., 2012; Pauleit et al., 2017).

2.1.1 Benefits of urban green landscapes

Bilgili and Gökyer, (2012) reported that urban green landscapes have many health, ecological,
economic, social, and aesthetic related benefits and functions which are vital to improving the
quality of life in the cities. These areas can serve as a bridge between nature and people (who live
in urban areas). Therefore, these areas are very important for the urban people and maintaining the
ecological functions of the environment.

Ecological benefits

One of the crucial benefits of urban green landscapes is the ecological and environmental functions
they provide. They can reduce environmental-related health risks associated with urban living by
improving air and water quality, mitigating impacts from extreme events, and buffering noise
pollution (WHO, 2017). Likewise, numerous studies have shown that urban green landscapes help
in mitigating the urban heat island effect in urban areas, thereby decreasing the energy costs of
cooling buildings. Urban green landscapes lower noise, air, and water pollution, and may balance
greenhouse gas emissions through CO2 absorption. The urban green landscape also provides
Neway Kifle [2021] 7 Remote Sensing and Geo-Informatics, AAU
 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

stormwater attenuation, thereby acting as a measure for flood mitigation (Pataki et al., 2010; Young
and Robert, 2010; Roy et al., 2012; Lee et al., 2015). Additional ecological benefits include a vital
role they play in providing shelter for the natural life corridors and urban forests they create. Thus,
urban green areas ensure the continuity of the species and the city's biodiversity by creating a
habitat for the existence of natural plants and animals (Dunnett et al., 2002; De La Barrera et al.,
2016).

Health benefits

From the viewpoint of health, urban green landscapes have health, physical, and psychological
benefits through physical activities and green exercises. Walk-able green areas influence the
longevity of urban senior citizens. Higher levels of greenness in urban areas have been positively
related to reduced stroke mortality. Besides, it improves recovery times after periods of ill health.
Moreover, it is also associated with a lower likelihood of being obese or overweight (Jose et al.,
2018).

Economic and aesthetic benefits

From the viewpoint of the economy, numerous studies have reported that urban green landscapes
have economic and aesthetic benefits. For instance, Arvanitidis et al., (2009) and Saraev and
Vadim, (2012) reported good quality green landscapes can increase value to the nearby property,
both residential and commercial. Consequently, helps to create a favorable image for a place,
increase tax returns to local authorities, attract tourism and inward investment in the area
encouraging employment, boost retail sales, and even exert pull-on skilled labor. Using green
landscapes has been increasingly recognized as a cost-effective way of reducing the energy costs
of cooling buildings in cities. Parts of the urban areas with adequate green landscapes are attractive
and aesthetically pleasing to the urban dwellers, tourists, and investors (Haq, 2011).

Social benefits

From the viewpoint of social benefit as a whole, green landscapes reduce the burden on the
National Health System (NHS) and absenteeism from work due to ill health by improving the
health of the population (Jennings et al., 2019). People visit urban green landscapes mostly for
enjoying various environmental elements such as fresh air, watching cascades, nature, flowers,
trees, wildlife, passive walking, taking children to play, educational opportunities, social activities,

Neway Kifle [2021] 8 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

picnics, meet friends, social interaction, meeting people, getting away from it all, and sitting. Other
reasons people visit urban green landscapes include walking a dog, hanging out, sitting either on
grass or seats, watching sport, walking socially as part of a group, smoking, passive enjoyment,
watching life go, photography, messing about on swings, readings, sunbathing, an informal pursuit
such as flying kites, etc. (Jose et al., 2018).

2.1.2 Urban green landscapes as a foe

It is helpful to understand that few studies have reported that urban green landscapes can also
negatively affect human populations. Concerning this, three major kinds of harm have been
documented.

First, allergic reactions and pollen can damage human health, infectious disease spread, and
accidental injury during outdoor green exercise or recreation. Second, green landscapes have been
used to add the surrounding property values, but this can lead to a process called environmental
gentrification where unlivable circumstances are created for communities with lower
socioeconomic status (Jennings et al., 2019). Third, other likely harmful effects of urban green
landscapes are related to fear of crime and animals (Snakes, insects, bats, rodents), disgust from
plant litter or blocked views, and psychological effects caused by trees and animals (behavior,
sound, smell) (Calaza et al., 2017).

2.1.3 Classification of urban green landscapes

There are several ways of classifying urban green landscapes such as based on their size, their
intended purpose, their location, and usage, etc. (Byrne and Sipe, 2010). According to Forest
Research, (2021) urban green landscapes include; urban parks, green areas around buildings,
gardens, wetlands, green roofs, street trees, and community woodlands. Likewise, Swanwick et
al., (2003) classified green landscapes based on the intended function of each green space in terms
of the kind of services provided, its size and the extent of its catchment area, and the nature of the
resource. These types of urban green landscapes include productive green (urban farmlands),
incidental green space (housing green space), recreation green (e.g., parks), burial grounds
(cemetery and churchyards), institutional green (e.g. school grounds), private garden (domestic
garden), wetlands and woodlands and linear green areas like transport corridors (cycleways, rail,
road, and walking route). Similarly, Panduro and Veie, (2013) identified the green landscapes in
the city of Aalborg, Northern Denmark based on the quality and quantity of facilities offered by

Neway Kifle [2021] 9 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

lakes, parks, churchyards, common areas (shared green space), agriculture fields sports fields,
nature, and green buffers.

2.1.4 An Overview of urban green landscapes in Addis Ababa

Identification of green landscape categories could vary based on the purpose of classification.
Accordingly, different cities have used different green landscape classifications. Kumelachew
Yeshitela et al, (2015) categorized the green landscapes in Addis Ababa into plantation forest,
institutional forest, public recreational parks, special function parks, private gardens, street
plantation, field crop, vegetable farm, grassland, and riverside vegetation.

Plantation forest: this type of

vegetation is mainly found in the
Northern part of Addis Ababa,
particularly, on Entoto mountain.
The eucalyptus tree is the
dominant species in plantation
forests. However, in some parts of
the plantation forest pure stands of
Cupressus lusitanica and Pinus
patula are also found.

Figure 2. 1: Picture of plantation forest on Mt. Entoto

Source: (Walta TV, 2020)

Special function parks: These are green landscapes mainly established for the conservation of
plant species, education, and research. However, they also provide recreational services to the
public. One good example of special function parks is the Gulelle Botanic Garden (GBG) which
lands on 705 hectares of land. The Addis Zoo Park (AZP), formerly called central (peacock) park
also falls into this category. AZP was previously used for providing recreational and social services
but is currently developed into a zoological park for the conservation of wild animals.

Neway Kifle [2021] 10 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Institutional forest: are forests that

are found inside the governmental
and non-governmental institution
compounds. Institutional forests in
the city include a mixture of exotic
and indigenous trees and shrubs.
These forests are largely found
within the compounds of USA,
French, British, Italy, and German
embassies, the various campus of
Addis Ababa University, within the
Ethiopian Orthodox Tewahedo
church compound, and the grand
and national palaces of Ethiopia.
Figure 2. 2: Picture of institutional forest in the old national palace

Source: (Walta TV, 2020)

Public parks: these are designed,

built, and enclosed green landscapes
mainly used for recreational and
social services such as wedding and
birthday ceremonies and other public
gatherings. These categories of green
landscapes include both shrubs and
trees. Currently, there are nearly 20
recreational parks in Addis Ababa,
including the recently built Friendship
Square/Park and Unity Park.

Figure 2. 3: Picture Unity Park found in the National palace

Source: (Walta TV, 2020)

Neway Kifle [2021] 11 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Private gardens: These are small-sized green landscapes found within the compounds of privately
owned houses and offices kept for private uses. This category constitutes vegetables, ornamental,
and fruit trees. They are mainly established for providing vegetables, fruits, and aesthetic service.
They also provide shading to houses and serve as stormwater infiltration by returning water to the
atmosphere via evapotranspiration.

Street trees: These are trees planted along road divides, street sides/pedestrian roads, and public
squares mainly for aesthetic values/beautification. Street trees play a crucial role by releasing
oxygen, consuming carbon dioxide, providing shade to vehicles and pedestrians, reducing air
pollution, and increasing the beauty of the streetscape. Moreover, they lower erosion and
stormwater management costs by reducing the intensity and amount and of precipitation striking
the ground.

Field crop: These are crops grown based

on rain-fed urban agriculture. The land
covered with this crop appears green
during the rainy season that runs from
June to September. The crops are usually
harvested during November and
December, therefore, the land remains
bare till the next crop season. During the
dry season, the land is mainly used for
keeping livestock. The common field
crops grown in the city include Wheat
(Triticum aestivum), Tef (Eragrostis tef),
and Barley (Hordeum vulgare).
Figure 2. 4: Field cropland in Addis Ababa

Source: (Kumelachew Yeshitela et al, 2015, pp. 11)

Grassland: This category includes green areas mainly covered with grasses which are located
within the compounds of the church, airport, military, and golf field. Grasses appear to be bare
land during the dry season changing the morphology of the landscape.

Neway Kifle [2021] 12 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Riparian vegetation. Addis Ababa is endowed with several streams and rivers. These rivers
originate from the Northwest, North, and Northeastern parts of the city draining the city southward,
and finally joins the Awash River. The major rivers that drain the city include Kebena, Great Akaki,
Kechene, Little Akakai, Ginfile, Jemo, Bantiketu, etc. There is no documented information on the
plant species composition of the riparian vegetation. However, various types of vegetation grow
along the river banks of the city.

Vegetable farms: Are found in parts of the city where vegetables are grown using irrigation from
the nearby rivers. The main vegetables planted in Addis Ababa include onion (Allium cepa), garlic
(Allium sativum), tomato (Lycopersicon esculentum), cabbage (Brassica oleravea), lettuce
(Lactuca sativa), and carrot (Daucus carota).

2.2 Urban blue landscapes

Blue spaces consider all the hydrographic features including lakes, ice masses, ponds, estuaries,
playas, reservoirs, and swamps and marshes or a linear surface water drainage network such as
streams, rivers coastlines, canals, and ditches) (Haeffner et al., 2017). The term “urban blue
landscapes” refers to the flowing open surfaces water courses and standing water bodies such as
lakes, reservoirs, streams, rivers, and ponds, etc. located within urban areas (Gunawardena et al.,
2017; Radek and Renata, 2019). Several scholars have claimed that blue landscapes provide very
different sensory experiences and are used in different ways with different outcomes and benefits
that are often overlooked and remain poorly understood. However, it is evident that there exists an
overlap between the green and blue landscapes (Haeffner et al., 2017; Easkey et al., 2020)

2.2.1 Benefits of urban blue landscapes

The widely acknowledged ecological benefits provided by urban blue landscapes include oxygen
production, noise reduction, microclimate regulation, and carbon sequestration. Moreover, the blue
spaces are used for recreation and social interaction purposes like physical exercise, sport, etc.
People also visit urban blue landscapes for recreational reasons like watching various types of wild
animals (e.g., fish, ducks) (Haeffner et al., 2017). Furthermore, a recent study has reported that
walking along or in blue spaces gives additional benefits for the mental health and well-being of
people by decreasing anger, tension, depression, and improving restoration time (Vert et al., 2020).

However, water bodies in urban areas are not always considered with a positive attitude only.
Ponds, rivers streams, and canals in urban areas can attract rubbish resulting in an unattractive

Neway Kifle [2021] 13 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

looking. There are also dangers from either the water itself (e.g., flooding) or associated potential
waterborne disease (e.g. malaria) at least during the life cycle of mosquitoes (Gledhill and James,
2008).

2.3 Land use land cover change

Land cover is defined as the physical and biological cover over the land surface, including
vegetation, water, bare soil, and/or man-made structures. On the other hand, land use refers to
activities on the land or classification of land according to how it is being used, such as residential,
industrial, commercial, agricultural, recreational, etc. Hence, both land use land cover is directly
related to the activities of humans in their environment. Unlike land-use, land-cover information
can be obtained from remote sensing. The land cover properties detected using remote sensing
techniques concerning land use can be inferred, particularly with ancillary data or prior knowledge
(Bhatta, 2011). Land-use change refers to a change in the way a certain area of land is being
managed or used by humans, whereas, land cover change means an alteration in a given continuous
property of the land like soil properties, vegetation type, etc. (Patel et al., 2019).

2.4 Remote sensing and GIS for urban land use land cover change
The method of detecting and monitoring the physical features of an area by measuring it's reflected
and emitted radiation from a distance, such as from a satellite, drone, or airplane, is known as
remote sensing (RS) (USGS, 2021). Whereas, GIS is an integrated system of computer software
and hardware for capturing, storing, retrieving, manipulating, analyzing, and displaying
geographically referenced or spatial information to aid development-oriented management and
decision-making processes (Aboyade, 2001).

Improvements in remote-sensing data have enhanced spatial accuracy and availability of free to
less-expensive satellite imageries. Geographic information system (GIS), combined with RS
allows quantitative analyses of the pattern and rate of urban LULC dynamics with better accuracy
and reasonable cost (Epstein et al., 2002). One of the advantages of RS data for earth observation
applications is its wide-area coverage (synoptic view) and studying of areas hard to reach.
Moreover, RS data is usually actual, timely, and temporal, or repetitive. High spatial resolution is
also essential for LULC studies as the sensor may detect data on more diversified LULC types.
Another advantage of remote sensing is its long-lasting records of data for more than 40 years of
observations by now (Bobrinskaya, 2012).

Neway Kifle [2021] 14 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

GIS creates a flexible environment for storing, displaying, and analyzing digital spatial data needed
for detecting urban LULC dynamics (Wu et al., 2006). RS imagery is the most important data input
of GIS (Reis, 2008). Accordingly, RS imageries can serve as a source of data for monitoring and
extraction of LULC types and their dynamics over different periods that occur in urban settings.
Integrating GIS with the digital image classification has proved its capability to analyze a wide
range of information about the rate, nature, direction, and location of urban land expansion and its
consequential LST rise. Thus, the integration of RS and GIS was found to be effective in analyzing
and monitoring the rate, extent, and patterns of urban growth and its impact on LST (Matiwos
Belayhun, 2018).

2.5 Image classification

Image classification is the process of categorizing and labeling individual/groups of pixels or
image objects within an image based on specific rules (Shinozuka and Mansouri, 2009). The
principal aim of image classification is to categorize and assign all the pixels of a particular image
into a proper land-cover class. Nowadays, there are numerous techniques for classifying satellite
images. These classification techniques can be grouped into pixel-based (Supervised,
unsupervised, and hybrid classification) and object-based image classification (Gašparović, 2020).

2.5.1 Supervised image classification

In supervised classification, the process of pixel categorization is governed by the image analyst
who sets numerical descriptors of the different land cover types found in the image scene to the
computer algorithm. To perform this, training areas that represent sample sites of known land cover
types are used to compile a numerical “interpretation key” that defines the spectral characteristics
for each land cover category. Every single pixel in the image scene is then compared numerically
against each land cover class in the interpretation key and finally assigned to the name of the land
cover class that it “looks most like” (Lillesand et al., 2015). Currently, there are several supervised
image classification techniques. It should be noted that traditional methods used in nowadays
research include Maximum Likelihood, Minimum Distance, as well as, contemporary machine
learning-based methods like Random Forest Artificial, Neural Network, and Support Vector
Machines (Gašparović, 2020).

Previous knowledge of the dataset is essentially needed for testing and training of classifiers in
supervised classification. The analyst collects training samples based on prior knowledge of the

Neway Kifle [2021] 15 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

dataset. One important advantage of these classification techniques is detecting classification

errors and correcting them. Another advantage of such techniques is that, after completing the
classification process, they represent the actual land-cover features, i.e., the training samples based
on which the classification is carried out. The main disadvantage of these techniques is that it is
costly and time-taking. Moreover, the analysts or scientists may not consider all situations and
land-cover classes during the selection of training samples. Thus, these techniques may lead to
human errors during classification (Borra et al., 2019).

2.5.2 Unsupervised image classification

The unsupervised classification does not need prior knowledge of the area and training samples
but only requires the number of land cover categories. In this type of classification, the entire
satellite image is automatically classified into land cover classes by grouping pixels of similar
radiometric characteristics. After classification, the class must be assigned automatically or
manually to the proper land-cover category in the real world. The most commonly used researchers
are Iterative Self Organizing Data Analysis (ISODATA) and k-means (Gašparović, 2020).

There are three main advantages of applying the unsupervised classification method. First, during
the initial separation of image pixels, extensive prior knowledge of the area being classified is not
needed. Second, the analyst who makes decisions while classification is not needed, hence human
errors are less likely to occur, Third, some unique land cover categories ignored in a supervised
classification may be recognized by unsupervised classification. However, there are limitations to
the use of such classification methods. For instance, the land cover classes of interest may not
necessarily correspond with the natural groupings identified by the classification process which is
based on spectrally homogeneous pixels. Moreover, the analyst has limited control over the classes
chosen by the classification process (Enderle and Weih, 2005).

2.5.3 Hybrid method

The hybrid method is intended to retain efficiency and/or accuracy and involve aspects of both
unsupervised and supervised classification (Lillesand et al., 2015). This technique is primarily
applied when one is unable to define well the differences between types of subclasses such as type
of material used as roof cover, vegetation type, crop or soil type. Rule-based classification methods
fall into hybrid methods. Such methods are based on some rules or conditions, e.g., water is an
area where the MNDWI>0 or built-up represents the area where the NDBI>0 (Gašparović, 2020).

Neway Kifle [2021] 16 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

2.5.4 Object-based image analysis (OBIA)

In contrast to pixel-based image classification, the object-based or object-oriented classification
uses both spectral and spatial information for classification. This method uses texture, color, shape,
pattern, size, and context information to categorize pixels into meaningful objects. This makes it
similar to the human visual interpretation of digital images, which works at multiple scales
simultaneously. OBIA is a two-step process involving (1) segmentation of the imagery into
separate image objects, followed by (2) classification of image objects to meaningful classes. The
basic idea behind this is that the image being classified consists of relatively similar “patches” that
are larger than individual pixels (Lillesand et al., 2015).

The object-based method has advantages that overcome the traditional pixel-based technique in
two aspects. First, the so-called salt-and-pepper effects that are common in pixel-based
classification are removed as the changes in classification units from pixels to image objects reduce
within-class spectral variation. Second, complementary information on the direct spectral
observations that improve classification accuracy can be derived from a large set of features
characterizing objects such as textural, spatial, and contextual properties (Guoet et al., 2007; Liu
and Xia, 2010). On the other hand, Liu and Xia, (2010), reported the potential limitation in using
object-based classification is the dependence on segmentation parameters like scale. This may
result in under or over-segmentation of image objects. Therefore, segmentation could be carried
out with an appropriate scale for better classification accuracy while using object-based methods.

2.5.4.1 Image segmentation

Segmentation is a process that involves the grouping of pixels in an image scene into non-
overlapping object segments, or features with homogeneous spatial and spectral properties
(Witharana and Civco, 2014). The scale parameter is the main variable that determines the size of
image objects during segmentation. To illustrate, in the case of the urban landscape, individual
vehicles parked on streets may be better segmented (changed into image objects) at a fine-scale,
whereas, buildings might be well segmented with a courser scale. Likewise, in the forest landscape,
segmentation at a medium scale can generate image objects that represent trees of the same sizes
and species, while segmentation at a coarser scale, could aggregate large areas of the forest into a
single image object. Determination of the scale parameter used in OBIA depends on some factors
such as the interest of the analyst on the general scale of the objects on the landscape to be
identified and the resolution of the sensor (Lillesand et al., 2015).
Neway Kifle [2021] 17 Remote Sensing and Geo-Informatics, AAU
 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

After image segmentation, there are various image object spectral and spatial characteristics that
help to classify or categorize the objects into proper land cover classes. These characteristics are
broadly grouped into two types. The first is related to the intrinsic characteristics of each object
itself (e.g. its texture, its spectral properties, its shape, its size, etc.). The second characteristics
describe the relationships among image objects such as boundary shared with the same or other
objects, their proximity to objects of the same or other types, their connectivity, and so forth. For
instance, an object having a smooth texture, a linear shape, a similar spectral signature with asphalt,
and topologically connected to other road objects would most likely be the road itself (Lillesand
et al., 2015).

2.6 Land Surface temperature

Land surface temperature (LST) is how cold or hot the Earth’s surface is when touched on a
specific location. From the viewpoint of satellites, the Earth’s surface is whatever it sees while it
looks to the ground, through the atmosphere. It could be the roof of a building, sea surface, snow
and ice, the leaves in the canopy of a forest, or the grass on a lawn. Thus, the land surface
temperature is not the same as the air temperature that is included in the daily weather report
(NASA, 2000).

LST, including also sea surface temperature (SST), is the principal geo–biophysical variable to be
estimated from thermal bands of satellite images. Most of the thermal energy directly emitted by
the land surface features is remotely detected by the thermal sensors that operate within this
spectral region. It is a vital parameter in different disciplines that focuses on environmental studies
including meteorology, ecology, climatology, geology, hydrology, oceanography, and so forth
(Jimenez-Munoz et al., 2008).

2.7 Land surface temperature response to land use and land cover dynamics
Land use/land cover (LULC) dynamics caused by natural processes or anthropogenic activities
drive the biogeochemistry of the Earth affecting the climate both at regional as well as global
levels. Radical changes in the land cover combined with the shrinkage in water bodies and
vegetation due to human activities increase the heat emission from the land surface and
atmospheric temperatures. The higher concentration of human activities, paved land cover, or
barren lands mainly leads to increased LST (Setturu et al., 2014). Landscape changes involving
LULC dynamics have contributed to the increase in LST. The various land cover classes show the

Neway Kifle [2021] 18 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

variability with different land surface temperatures (Weng et al., 2007; Wichansky et al., 2008).
Moreover, LULC dynamics influence evaporation rates, the absorption of solar radiation, storage
of heat, wind turbulence, land surface temperature, and can considerably change the conditions of
the near-surface atmospheric temperature. Furthermore, the key indicators considered to explain
the global climate change including land cover/use, clouds, land surface temperature, exchanges
of energy, and moisture can vary drastically both temporally and spatially. Thus, it is essential to
monitor LULC dynamics at proper scales to mitigate its impacts on changing climate particularly
land surface temperature (Setturu et al., 2014).

In Ethiopia, several studies have investigated the temporal and spatial pattern of LST in
relation to the changes in LULC classes. For instance, Abel Balew and Tesfaye Korme, (2020)
used NDVI and NDBI to analyze the spatial pattern of LST and its variation with different LULC
types in Bahir Dar City and its surrounding. Likewise, Matiwos Belayhun, (2018) used NDVI to
study its correlation with LST in the city of Dire Dawa, Ethiopia.

It is also evident from the literature that in Ethiopia, particularly Addis Ababa, the urban heat island
(UHI) phenomenon and its relation to urban green landscapes has been studied by a few studies
(e.g. Mathias Tesfaye and Tebarek Lika, 2016; Samson Workaye et al., 2018; Ermias Teferi and
Hiwot Abreha, 2017). However, recently, (Simwanda et al., 2019), focused on the relationship
between LST and the composition, spatial patterns, and configuration of impervious surfaces and
green landscapes in four African cities, Addis Ababa (Ethiopia), Lagos (Nigeria), Nairobi (Kenya),
and Lusaka (Zambia).

2.8 Thermal remote sensing for land surface temperature

Remote sensing, particularly, thermal remote sensing is very important for monitoring LST and
analyzing the thermal characteristics of the earth’s surface features as well as their relationships.
In this regard, LST has been retrieved from low-resolution thermal satellite sensors, for example,
Moderate Resolution Imaging Spectroradiometer (MODIS) and NOAA Advanced Very High-
Resolution Radiometer (AVHRR). Moreover, LST can also be estimated from medium resolution
thermal sensors such as Advanced Spaceborne Thermal Emission and Reflection Radiometer
(ASTER), Landsat Thematic Mapper (TM), Enhanced Thematic Mapper Plus (ETM+), and
Operational Land Imager (OLI)/ Thermal Infrared Sensor (TIRS) (Abel Balew and Tesfaye Korme,
2020). Thermal data is detected by satellite sensors that operate within the thermal infrared spectral

Neway Kifle [2021] 19 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

region. The radiance at the top of the atmosphere is captured by Thermal Infrared (TIR) sensors.
However, radiation recorded by the sensor is influenced by atmospheric effects. Therefore, to
obtain meaningful values, this row data should be corrected for atmospheric effects and emissivity.
This radiometrically corrected temperature can be used to retrieve LST in Celsius degrees or
Kelvin (Bobrinskaya, 2012).

2.9 Methods of land surface temperature retrieval

Many applications such as agro climate, meteorology, and environmental studies strongly require
knowledge of land surface temperature. Satellite remote sensing in the infrared provides an
interesting alternative for the continuous and global records of this parameter. Two main problems
arise in using thermal remote sensing data for retrieving LST. These are the land emissivity effect
and atmospheric effect. Hence, different algorithms have been developed to correct these two
effects (Pozo et al., 1997). There are three major methods of retrieving LST (Wang et al., 2019)
that include the Mono-Window Algorithm (MWA), developed by Qin et al., (2010), the Single-
Channel (SC) Method, developed by Jimenez-Munoz and Sobrino, (2003), and the Split Window
Algorithm (SWA), developed by McMillin, (1975).

2.9.1 Mono-window algorithm (MWA)

Qin’s mono-window algorithm is shown in equation (1). MWA requires three parameters. These
are land surface emissivity (Ɛ), the effective mean atmospheric temperature (Ta), and atmospheric
transmittance (τ). However, equation (1) is proposed for Thematic Mapper (TM) thermal images
(Wang et al., 2019).

Ts = [a6 (1 − C6 − D6) + (b6 (1 − C6 − D6) + C6 + D6) T6 − D6Ta] / C6 (1)

Where;

Ts-LST in K; T6 is the brightness temperature of Landsat 5 TM Band 6; Ta is the effective mean

atmospheric temperature; C6 and D6 are the internal parameters for the MWA based on the
atmospheric parameters and ground emissivity, and a6 and b6 are the coefficients used to
approximate the derivative of the Planck radiance function for the thermal band.

Later, Wang et al., (2015) improved the algorithm for retrieving LST from Band 10 of Landsat 8.
This improved algorithm is called Improved Mono-Window (IMW) algorithm as is shown in
equation (2).

Neway Kifle [2021] 20 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Ts = [a10 (1 − C10 − D10) + (b10 (1 − C10 − D10) + C10 + D10) T10 − D10Ta] / C10 (2)
where Ts is the LST estimated from Band 10 of Landsat 8 TIRS image; Ta is the effective mean
atmospheric temperature; a10 and b10 are the constants used to approximate the derivative of the
Planck radiance function for the TIRS Band 10; T10 is the brightness temperature of Landsat 8
TIRS Band 10; C10 and D10 are the internal parameters for the algorithm.

2.9.2 Single-channel (SC) method

The single-channel (SC) method was proposed in 2003 by Jiménez-Muñoz and Sobrino, (2003)
and improved in 2009 by Jiménez-Muñoz et al., (2009). This algorithm requires only two input
parameters, the land surface emissivity, LSE (Ɛ), and the atmospheric water vapor content (ω).
Unlike MWA, the SC method does not require the effective mean atmospheric temperature (Ta)
parameter, and the atmospheric water vapor content (Wang et al., 2019). The advantage of the SC
algorithm is its simplicity. However, the SC algorithm applies to sensors with only one TIR band.
Another disadvantage of the SC algorithm is the substantial effects of the uncertainty of
atmospheric patterns on the accuracy of LST retrieval (Rongali et al., 2018).

Jiménez-Muoz et al. (2014), updated the SC method for Landsat 8 and calculated the
corresponding parameters; the formula for the SC method is provided in equation (3).

1
LST = γ [𝜀 (ψ1Lsen+ψ2) +ψ3] +δ (3)

Where; ε is LSE, Lsen is at-sensor registered radiance (W/ (m2·sr·µm)), γ and δ are variables
based on the Planck’s function given by:

γ = C2LsenTsen2λ4LsenC1+1 (4)
δ = -γLsen+Tsen (5)
Where Tsen is the at-sensor brightness temperature; for band 10 of Landsat 8 and
C1 and C2 are physical constants (C1=1.191×108 Wμm4 sr−1 m−2, C2=1.439×104 μm·K) (Käfer
et al., 2020).
The three atmospheric function parameters; ψ1 ψ2 and ψ3 are calculated as follows;

ψ1=-7.2122w2+0.00005Ta2-2.452321w-0.026275Ta 0.00005Ta2w+ 0.02317Taw+

0.04663Taw2-0.00007Ta2w2+4.47297, (6)

Neway Kifle [2021] 21 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

ψ2=89.61569w2-0.00038Ta2+106.55093w+0.21578Ta+0.00141Ta2w-0.78444Taw-
0.5732Taw2+0.00091Ta2w2-30.37028, (7)
ψ3=-14.65955w2-0.0001Ta2-79.95838w+0.04181Ta-0.00091Ta2w +0.54535Taw+
0.09114Taw2-0.00014Ta2w2-3.76184. (8)

2.9.3 Split window (SW) algorithm

The split window algorithm (SWA) was originally proposed by McMillin, (1975). It is an
algorithm for observing the ocean surface temperature based on Advanced Very High-Resolution
Radiometer (AVHRR) thermal infrared data. The basic principle behind this algorithm is that two
adjacent TIR bands have different absorption properties; the difference between the brightness
temperatures of the two TIR bands is used to determine the attenuation information of the
atmosphere on thermal radiation. The main advantage of SWA that it is simple to operate and is
less dependent on atmospheric parameters (Rongali et al., 2018; Wang et al., 2019). According to
Jiménez-Muñoz et al., (2014), the equation to estimate LST using SW algorithm from Landsat 8
TIRS data is:

LST = Tisen+1.378(Tisen−Tjsen)+0.183(Tisen−Tjsen)2−0.268+(54.3−2.238w)(1−ε)+

(−129.2+16.4w)Δε (9)

Where;

Ti sen is at-sensor brightness temperatures for bands i (band 10) in Kelvins,

Tj sen is at-sensor brightness temperatures for band j (band 11) in Kelvins,

ε is the mean emissivity, ε = 0.5(εi + εj),

Δε is the emissivity difference, Δε = (εi −εj), and

w is the total atmospheric vapor content (in g·cm−2).

Neway Kifle [2021] 22 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

CHAPTER THREE
3. STUDY AREA, MATERIALS, AND METHODS
3.1 Study area description
3.1.1 Location
Addis Ababa is Ethiopia's capital and largest city. Addis Abba means "New Flower" in Amharic.
It is the headquarters of the African Union (AU) and the United Nations Economic Commission
for Africa (ECA). Hence, Addis Ababa is frequently referred to as Africa's "political capital." The
city is sub-divided into 10 sub-cities: Akaki-Kalty, Arada, Bole, Lideta, Gulele, Kirkos, Kolfe,
Nifas silk lafto, and yeka. The ten sub-cities are further divided into 116 woredas. The city is
located in the central highlands of Ethiopia and is in-circled by mountain ridges of Entoto in the
North, Wochecha in the West, Furi in the South West, and Yerer in the North East. In absolute
terms, Addis Ababa lies between 38° 39´2.76´´ to 38° 54´19.4´´ East Longitude and 8° 50´13.06´´
to 9° 6´4.62´´ North Latitude covering a total area of 527 square kilometers (See Figure 3.1).

Figure 3. 1: Location map

Neway Kifle [2021] 23 Remote Sensing and Geo-Informatics, AAU
 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

3.1.2 Basic population characteristics

According to CSA, (July 2017) population estimation, Addis Ababa has a total human population
of 3,433,999 with a population density of 6518 persons per square kilometers. The sex composition
of the city reveals that 1,809,999 (47.3%) are males and the rest 1,809,000 (52.7%) are females.
Whereas, the sex ratio indicates that there are 90 males for every 100 females in the city (Table
3.1). The 2007 Ethiopian population and housing census show that the city has an average
household size of 4.1 and an annual population growth rate of 2.1. The major ethnic groups
residing in the city include Amhara (47.1%), Oromo (19.5%), and Guragae (16.3%). The religious
composition of the city shows that 74.7 % Orthodox Christians, 16.2 % Muslims and 7.8%
Protestants (CSA, 2007).

Table 3. 1: Addis Ababa population by sub-city, sex composition, sex ratio, and population density

Male Female Total Population Sex

Sub-city Population Population Population Density Ratio
Addis Ketema 155,478 164,575 320,053 43,250 94
Akaki-Kaliti 110,435 116,747 227,182 1,924 95
Arada 123,445 141,696 265,141 26,755 87
Bole 180,782 206,573 387,355 3,212 88
Gulele 161,078 174,356 335,434 11,144 92
Kirkos 128,841 148,505 277,346 18,970 87
Kolfe-Keraniyo 258,480 279,081 537,561 8,777 93
Lideta 119,843 132,999 252,842 27,573 90
Nefas-Silk-Lafto 185,461 211,025 396,486 5,805 88
Yeka 201,156 233,443 434,599 4,970 86
Total 1,624,999 1,809,000 3,433,999 6518 90
Source: CSA, 2017

3.1.3 Physiography and drainage

The topography of Addis Ababa is characterized by contrasting plain areas and very steep slopes.
In general, flatness dominates the southern part of the city and the slope is steeper in the northern
part of the study area. USGS, (2001) slope classification is modified and adapted to reclassify the
slope gradient in the study area (Table 3.2). The result shows that 34.4 % is slopping, 28.4 % is
moderately steep and 16.7% gently sloping. Whereas 7.2% of the area in the city is characterized
by flat to very gently sloping topography and only 0.5% is very steep. The topography forms plateau
in the northern, western, South Western, and North-Eastern parts of the city. As a result, the
streams drain towards the south from the Entoto mountain range, southeast from Mt. Wechecha,
towards southwest from Mt. Yerer and Mt. Furi resembling a dendritic drainage pattern (Teklit
Neway Kifle [2021] 24 Remote Sensing and Geo-Informatics, AAU
 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Berhe, 2015). The major rivers that drain the city include Little Akaki, Great Akaki, Kebena,
Kechene, Kutume, and Banchi yiketu. The altitude in the city ranges from 2044 to 3141 meters
above mean sea level (MSL) with a mean elevation of 2347 meters above MSL (Figure 3.2).

Figure 3. 2: A) Elevation map and B) Slope map

Table 3. 2: Reclassified slope

Area Coverage
Slope gradient in (%) Description
Area in sq. km. Percentage
0-2 Flat to very gently sloping 37.9 7.2
2-5 Gently sloping 88.1 16.7
5-10 Slopping 181.2 34.4
10-20 Moderately steep 149.5 28.4
20-50 Steep 67.3 12.8
Above 50 Very Steep 2.6 0.5
Source: Computed from DEM (12.5m by 12.5m)

Neway Kifle [2021] 25 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

3.1.4 Climate
A. Precipitation
Precipitation data was obtained from the NMA recorded at 12 stations (Abyssinia school, Bole,
Addis Ababa observatory, Akaki, Ayertena, Intoto, Kolfe keranyo, Kotebe Luke, Kotebe TTC,
Medihanealem School, Megabit 28 School, and Yekatit 23 School) for the last 23 years (1998 to
2020). The result shows that the average precipitation in the city is 1126.48mm. Most of the
precipitation in the city was recorded during the rainy season months of June, July, August, and
September. The month with the most precipitation on average is August with 286.0mm of
precipitation. The month with the least precipitation on average is December with an average of
6.6mm (Figure 3.3). The precipitation ranges from 694.9mm to 1452.0mm during the years 2012
and 2004 respectively (Figure 3.4).

350.0

300.0
Rainfall (mm)

250.0

200.0
150.0

100.0

50.0

0.0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Rainfall (mm) 10.8 15.8 59.6 69.4 85.8 137.7 277.4 286.0 154.1 29.8 8.5 6.6
Months

Figure 3. 3: Average monthly rainfall in Addis Ababa during the years 1998 to 2020

Neway Kifle [2021] 26 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

1600.0

1400.0

1200.0
Rainfall (mm)

1000.0

800.0

600.0

400.0

200.0

0.0
2003
1998
1999
2000
2001
2002

2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
Years

Figure 3. 4: Annual rainfall distribution in Addis Ababa during the years1998 to 2020

B. Temperature
The average recorded temperature in Addis Ababa from Abyssinia school, Bole, Addis Ababa
observatory, Akaki, Ayertena, Intoto, Kolfe keranyo, Kotebe Luke, Kotebe TTC, Medihanealem
School, Megabit 28 School, and Yekatit 23 school stations during the years 1998 to 2020 was
17.20C. The warmest month, on average, is March with an average maximum temperature of
25.90C. While the coolest month on average, is December, with an average minimum temperature
of 8.30C (Table 3.3 and Figure 3.5).

Table 3. 3: Maximum, minimum and mean monthly temperature in Addis Ababa from 1998 to 2020

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Temp Max 24.1 25.4 25.9 25.5 25.2 23.5 21.2 20.7 21.9 23.3 23.6 23.3
Mean 16.5 17.8 18.6 18.8 18.7 17.6 16.3 16.0 16.5 16.7 16.3 15.8
Temp Min 8.9 10.1 11.4 12.1 12.2 11.6 11.4 11.3 11.2 10.1 9.1 8.3
Source: NMA, 2021

Neway Kifle [2021] 27 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

30.00

25.00
Temperature (oC)

20.00
Temp Max
15.00 Mean
Temp Min
10.00

5.00

0.00
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Months

Figure 3. 5: Minimum, maximum, and mean monthly Temperature in Addis Ababa from 1998 to 2020

3.1.5 Soil
The information obtained from the Ministry of Water Resource (Ethiopia) indicates that the
dominant soil types in the study area are eutric nitisols, calcic xerosols, and chromic luvisols
constituting 21.2%, 6.9%, and 2.2% respectively (Table 3.4 and Figure 3.6). Other soil types in the
study area include pellic vertisols, orthic solonchaks, and leptosols.

Table 3. 4: Soil types

Soil type Area (sq. km.) Percentage

Eutric nitisols 111.6 21.2
Calcic xerosols 36.3 6.9
Chromic luvisols 11.4 2.2
Orthic solonchaks 4.9 0.9
Leptosols 2.8 0.5
Pellic vertisols 2.8 0.5
Source: Based on MoWR digital data, 2004

Neway Kifle [2021] 28 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Figure 3. 6: Soil map, based on MoWR digital data

3.2 Materials and software

3.2.1 Materials
3.2.1.1 Primary data
Remotely-sensed imageries data have been increasingly used as a primary source of information
for remote sensing and geographical information systems (GIS) applications (Craig, 1991). With
the obtainability of high-resolution remote sensing images and multi-source geospatial data, there
is a boundless requirement to transform Earth observation data into useful information essential
for urban planning and decision making (Vatseva et al., 2016). Accordingly, in the present study,
high-resolution imagery from commercially available SPOT 5 and freely available Sentinel-2A
satellite images were obtained from European Space Agency (ESA) and the USGS website
(http://earthexplorer.usgs.gov) respectively. The SPOT 5 and sentinel-2 images were used as input
Neway Kifle [2021] 29 Remote Sensing and Geo-Informatics, AAU
 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

for the OBIA. Landsat 7 ETM+ and Landsat 8 OLI/TIRS thermal and multispectral images were
also freely downloaded from the USGS website to retrieve LST and calculate LULC indices
respectively. Additionally, LST data products of MODIS (MOD11A1) Version 6 daily 1km spatial
resolution imagery were downloaded freely from the website of Earth Data and used for the
retrieved LST validation. Moreover, a 12.5m (resampled) spatial resolution digital elevation model
(DEM) was acquired from the Alaska Satellite Facility (ASF) (www.asf.alaska.edu) to generate
data on drainage, elevation, and slope gradient layer for the study area (Table 3.5). Furthermore,
for ground-truthing and image classification accuracy validation reference data was collected using
Google Earth image and handheld GPS devices with an accuracy of ± 3 m.

Table 3. 5: Satellite images and details of their respective bands used

Central
Spatial Acquisition
Data set Band name band/band Purpose Source
resolution date
width (µm)
B1 (Green) 0.49-0.61 10mX10m
OBIA and
B2 (Red) 0.61-0.68 10mX10m
SPOT 5 Jan, 2006 LULC ESA
B3 (NIR) 0.78-0.89 10mX10m
Classification
SWIR 1.58-1.70 20mX20m
Band 2 (Blue) 0.49 10mX10m
Band 3 (Green) 0.56 10mX10m Jan 25, 2016 OBIA and
Sentinel-
Band 4 (Red) 0.665 10mX10m and LULC USGS
2A
Band 8 (NIR) 0.842 10mX10m Jan 18, 2021 Classification
Band 11 (SWIR) 1.61 20mX20m
Band 1 (Blue) 0.45 - 0.52 30mX30m
Band 2 (Green) 0.52 - 0.60 30mX30m LULC
indices
Landsat 7 Band 3 (Red) 0.63 - 0.69 30mX30m
Jan 04,2006 calculation USGS
ETM+ Band 4 (NIR) 0.77 - 0.90 30mX30m
and
Band 5 (SWIR 1) 1.55 - 1.75 30mX30m LST retrieval
Band 6 (Thermal) 10.40 - 12.50 30mX30m
Band 2 (Blue) 0.452 - 0.512 30mX30M
Band 3 (Green) 0.533 - 0.590 30mX30M LULC
Jan 24, 2016 indices
Landsat 8 Band 4 (Red) 0.636 - 0.673 30mX30M
and calculation USGS
OLI/TIRS Band 5 (NIR) 0.851 - 0.879 30mX30M Jan 05, 2021 and
Band 6 (SWIR 1) 1.566 - 1.651 30mX30M LST retrieval
Band 10(Thermal 1) 10.60 - 11.19 30mX30m
Band 31 (TIR) 10.78 -11.27 1kmX1km Jan 04,2006,
MODIS- Jan 24, 2016 For LST Earth
LST Band 32 (TIR) 1kmX1km & Jan 05, validation Data
11.77 - 12.27 2021
slope,
12.5mX12.5m
DEM Not Applicable Not Applicable April 2008 drainage, and ASF
(resampled)
elevation

Neway Kifle [2021] 30 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

The study period covers 15 years and the years 2006, 2016, and 2021 were selected to analyze the
spatiotemporal pattern of urban-blue green landscapes and their effect on LST. These years were
selected due to the availability of high-resolution images at hand. All the images were acquired
during the dry month of January to minimize atmospheric problems such as cloud and haze effects.

3.2.1.2 Secondary data

Shapefiles with attributes of administrative boundaries, roads, rivers as well as population data
from the Central Statistical Agency (CSA) of Ethiopia were collected and used as secondary data
sources in this study. Besides, Meteorological station data on temperature and precipitation, and
digital soil data were collected from the National Meteorological Agency (NMA) of Ethiopia, and
the Ministry of Water Resource-Ethiopia (MoWR) respectively.

3.2.2 Software
The main software packages that were used in this study include ESRI product ArcGIS 10.6.1,
ERDAS Imagine 2015, QGIS 3.10.5, eCognition Developer 9.1, ENVI 5.3, Stata 15, Google Earth
Pro, and Microsoft Office products (Microsoft Office and Excel).

3.3 Methods
3.3.1 Image pre-processing
For better feature identification and other related applications, raw satellite images cannot be used
due to mainly radiometric, atmospheric, and geometric effects. Hence, pre-processing was carried
out before data processing and information extraction. Preprocessing carried out for SPOT 5,
Sentinel-2A, Landsat 7 ETM+, and Landsat 8 OLI/TIRS imageries include radiometric correction
(haze and noise reduction), geometric correction (by providers), layer stacking, and sub setting.
Image mosaicking was also implemented on SPOT 5 and Sentinel-2A image scenes on ERDAS
Imagine 2015. Another task of pre-processing includes scan-line error correction (SLC) which is
applied for the Landsat 7 ETM+ images (Bands 1 to 6). The SLC was performed using the gap-fill
tool in QGIS 3.10.5. For the thermal bands of Landsat 7 ETM+ (Band 6) and Landsat 8 TIRS
(Band 10), atmospheric correction, particularly the Dark Object Subtraction-1 (DOS-1) method as
proposed by (Chavez Jr, 1988) was applied using the Semi-automatic classification plugin (SCP)
in QGIS (Filgueira et al., 2019).

Neway Kifle [2021] 31 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

3.3.2 Image enhancement

Image enhancement refers to improving the image quality to a better and more understandable
level for information extraction and interpretation. It is carried out for better interpretation of
images by human viewers (Bhatta, 2011). Numerous ways help to enhance an image. In this study,
band rationing or indices (arithmetic operations performed on images) were used. Accordingly,
during image classification, NDVI, NDBI, and MNDWI were computed within eCognition
Developer 9.1 to boost the interpretability of vegetation, built-up, and water land use/cover classes
respectively.

3.3.3 Image classification

Image classification is the process of assigning and labeling groups of image objects or pixels
based on spectral and spatial information for classification. There are various techniques for
classifying remotely sensed images. The common ways of classification include pixel-based such
as unsupervised, Supervised, and hybrid classification, and image object-based. In this study, the
object-based image analysis was applied (see section 3.3.3.2 for details). The true color
combination of the land cover classes in the study area is presented in Appendix 5.

3.3.3.1 Nomenclature of land cover classes

To standardize the classification United States Geological Survey (USGS) land cover
classification system was contextualized and adopted (USGS, 1975 as cited in Hillel and Hatfield,
2005). Thus, five land cover classes were identified in the study area. These are the built-up,
vegetation, urban farmland, bare land, and water. The description of each land use classes is
presented as follows:

Neway Kifle [2021] 32 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Built-up: include areas

allotted to residential,
commercial and services,
industrial, transportation,
communication and
utilities, mixed urban and
other paved surfaces.

Figure 3. 7: A picture of a built-up area in Addis Ababa, taken from Mt. Yeka

Vegetation: include
forests, grasslands,
shrubs, and other green
spaces (except cropland).

Figure 3. 8: An example green landscape within Friendship Square/Park

Urban farmland:
include urban cultivated
land (farmland with and
without crop).

Figure 3. 9: A picture of onion farm in Akaki area, Southern Addis Ababa

Neway Kifle [2021] 33 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Bare land: bare exposed

rock, excavation sites,
quarries, and gravel pits.

Figure 3. 10: A picture of the red ash excavation site in Southern Addis Ababa

Source: (Google Earth, 2021)

Water: include both the

flowing and standing
urban surface waters
such as rivers, ponds, and
lake and wastewater
treatment sites.

Figure 3. 11: A picture of an artificial lake inside Friendship Square/Park

3.3.3.2 Object-based image analysis (OBIA)

For this particular study, Object-based image analysis was applied using high-resolution imageries
of SPOT 5 and Sentinel-2A satellites as an input. OBIA is more robust than the pixel-based
paradigm as its potential for image classification in heterogeneous areas like urban areas has been
proved (Bisquert et al., 2014).

Unlike pixel-based classification, OBIA always analyzes images as objects instead of pixels that
enables to reduce the ‘‘salt and pepper” effect which emanates from neighboring pixels and results
in better classification accuracy (Ziwei et al., 2019). OBIA is a rule set-based technique and
involves a two-step process: (i) image segmentation followed by (ii) classification and merging.
eCognition Developer 9.1 software was used to implement the OBIA.

Neway Kifle [2021] 34 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Segmentation

Segmentation is a process where an image scene is decomposed into image objects that have
similar textural, spectral, and spatial attributes. Segmentation is very important because the
resulting image objects are inputs for the subsequent step (classification) (Witharana and Civco,
2014).

In eCognition Developer software, there are numerous algorithms for segmentation. A few of them
include Multi-Resolution Segmentation (MRS), Multi-Threshold Segmentation (MTS), and
Spectral Difference Segmentation (SDS). MRS was used in this particular study, as it has been
proven to be one of the most effective image segmentation algorithms in the OBIA framework
(Witharana and Civco, 2014). MRS is a region growing method that starts with a pixel forming an
object and merging the neighboring homogeneous pixels until the parameters criterion (set by the
user) is attained (Bisquert et al., 2014). The main parameters that govern MRS are the scale
(Controls the size of resulting objects), Shape (controls the weight of color during segmentation
as shape + color =1), and Compactness (controls the compactness of the resulting object as
compactness + smoothness =1). Different combination of these parameters was experimented
using the Trail-and-Error approach to get the best combination (Gülçin and Akpinar, 2018; Labib
and Harris, 2018). An example of segmentation result and segmentation parameters combination
is presented in Figure 3.12 and Table 3.6 respectively.

Figure 3. 12: An example of segmentation during 2006 with scale=30, shape=0.3 & compactness=0.1

Neway Kifle [2021] 35 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Table 3. 6: Segmentation parameters combinations for 3 images

SPOT 5 Sentinel-2A Sentinel-2A
Algorithm Segmentation parameters
(2006) (2016) (2021)
Layer weights (Blue, Green, red, NIR) 1,1,1 1,1,1,2 1,1,1,2
Multi-Resolution Scale parameter 30 30 30
Segmentation
(MRS) Shape 0.3 0.2 0.3
Compactness 0.1 0.2 0.3
No of resulting image objects 62,605 128,269 167,454
Source: Computed from eCognition Developer 9.1 software
Classification and merge
This is the second stage in OBIA where the image objects generated were categorized into built-
up, vegetation, urban farmland, bare land, and water land use/cover classes. Various classification
algorithms are integrated with eCognition Developer 9.1 software. Quite a few of them include
assign class, classification, and hierarchal classification.

Figure 3. 13: An example of the process tree with rule sets for 2021 classification in eCognition

Neway Kifle [2021] 36 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

The assign class algorithm is the simplest algorithm among them and has been applied in this study
(Akmalov et al., 2017). The Assign class algorithm requires the user to set at least four parameters:
(i) level, (ii) class filter (the class from which the image objects are assigned to the user class), (iii)
threshold condition (determines whether an image object matches a condition or not), (iv) use class
(the new or existing class to which the image objects are supposed to be assigned).

In this study, five major image object features are used for classifying the objects into proper land
use classes. These are (i) layer values such as the brightness and standard deviation, (ii) geometry
of objects (e.g., areal extent in pixels), (iii) shape (e.g., asymmetry and rectangular fit), (iv) class-
related features like relative border to other classes and (v) customized object features, for instance,
NDVI, NDBI, and MNDWI which were used to extract vegetation, built-up and water land cover
classes respectively. An example of a process tree with rule sets is presented in Figure 3.13.

During classification, there was confusion between vegetation and urban farmland (with crop).
However, an attempt has been made to differentiate the classes using rectangular fit (as croplands
have higher rectangular fit value) and a manual editing tool. Bare land was also mixed with urban
farmland (without crop). Brightness value was used to discriminate between bare land and
farmland (without crop) given the brightness values for bare land are relatively higher in general.

Finally, each land cover class is merged using the merge region algorithm in eCognition and the
final result was exported in ERDAS Imagine images (*.img) format for layout and further analysis.

3.3.3.3 Classification accuracy assessment

The process of estimating accuracy is the next step that follows classification. Inaccurate
classification results might lead to misleading analyses, and consequently to misleading
conclusions. The confusion matrix (sometimes called contingency table or error matrix) was used
to estimate accuracy measures such as overall accuracy, Producer’s accuracy, user’s accuracy, and
Kappa statistics. Overall accuracy is calculated by dividing the total number of properly classified
pixels (sum of the values on the confusion matrix's main diagonal) by the total number of reference
pixels, which is a fairly coarse assessment. The producer’s and user’s accuracy are used to
measuring the accuracy of individual classes. In the confusion matrix, the producer's accuracy is
computed by dividing cell (i i) by the sum of column i while the user's accuracy is computed by
dividing (i i) by the sum of row (Gašparović, 2020).

Neway Kifle [2021] 37 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

The Kappa coefficient, K is a more robust measure of classification accuracy than overall accuracy
as it takes into account the off-diagonal elements and the diagonal terms. (Ahmad and Quegan,
2012). It is computed as (Gašparović, 2020):

𝑟 𝑟
N∑ X𝑖𝑖−∑𝑖=1(X𝑖+∗X+𝑖)
k= 𝑖=1
𝑟 (3.1)
𝑁2 −∑𝑖=1(𝑋𝑖+∗𝑋+𝑖)

Where:

r - number of rows in the confusion matrix,

Xii – the values in column i and row i (on the main diagonal),

xi+ - the total number of measurements in row i,

x+i - the total number of measurements in column i, and
N -is the number of measurements in the matrix as a whole.
Map of the ground control points is presented in Appendix 7.
3.3.4 LULC change detection
Change in LULC pattern was computed in square kilometers for each land cover class during the
periods. LULC conversion matrix (image differencing) technique was applied in a GIS
environment to detect the change directions i.e. which land use/cover class has converted into what
(Binyam Alemu et al., 2015). The percentage share (Ebrahim Esa and Mohammed Assen, 2017)
of each land-use class and the annual rate of change (Belaybeh Bufebo and Eyasu Elias, 2021) in
square kilometers per year was calculated to obtain the magnitude of the changes undergone
between the study periods using equations 3.2 and 3.3 respectively as follows:

T2 − T1
Percentage share = ( ) * 100, (3.2)
T1

T2 − T1
Rate of change (Sq.km/year) = ( ) (3.3)
N

Where: T1 and T2 area of land use/land cover type at the initial and final year respectively and

N is the time interval/number of years between the initial and final years.

Neway Kifle [2021] 38 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

3.3.5 Land surface temperature retrieval

Land Surface Temperature (LST) is defined as the temperature of the earth’s surface which we feel
when we directly touch or contact it. In other words, it is the skin temperature of the surface
(NASA, 2000). In this study, the thermal bands of Landsat 7 ETM+ (Band 6) and Landsat 8 TIRS
Band 10 are used to calculate LST. Landsat 8 TIRS Band 10 is preferred due to the higher stary
light error and larger calibration uncertainty associated with band 11 (Guha, 2018; USGS, 2019b;
Abel Balew and Tesfaye Korme, 2020). To estimate LST, several steps have been followed:

3.3.5.1 Conversion of DNs to at sensor spectral radiance

The first step in estimating LST is converting row digital numbers (DNs) values to at sensor
radiance. Equations 3.3 (USGS, 2019a) and 3.4 (USGS, 2019b) are used to convert DNs into
radiance units for Landsat ETM+ thermal band (band 6) and Landsat 8 TIRS band 10 respectively.

𝐿𝑀𝐴𝑋𝜆−𝐿𝑀𝐼𝑁𝜆
Lλ = ( ) * (QCAL- QCALMIN) + LMINλ (3.3)
𝑄𝐶𝐴𝐿𝑀𝐴𝑋−𝑄𝐶𝐴𝐿𝑀𝐼𝑁

Where:

Lλ is at sensor spectral radiance in (Watts/ (m² * sr * µm)),

QCAL stands for quantized calibrated pixel value (DNs),

LMINλ is spectral radiance scaled to QCALMIN in (Watts/ (m² * sr * µm)) = 3.2

LMAXλ is spectral radiance scaled to QCALMAX in (Watts/ (m² * sr * µm)) = 12.65

QCALMIN in DN = 1 is the minimum quantized calibrated pixel value (equivalent to LMIN),

QCALMAX in DN =255 is the maximum quantized calibrated pixel value (equivalent to LMAX).

Lλ = ML * Qcal + AL (3.4)

Where: Lλ is at sensor spectral radiance (W/ (m2 * sr * μm)), MLis RADIANCE_MULT_BAND_n

from the metadata (radiance multiplicative scaling factor for the band) = 0.0003342, AL is

RADIANCE_ADD_BAND_n from the metadata (Radiance additive scaling factor for the band)
= 0.1 and Qcal = Standard product pixel values that have been quantized and calibrated (DN).

Neway Kifle [2021] 39 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

3.3.5.2 Top of atmosphere brightness temperature

This refers to the effective at-satellite temperature and is computed as follows (USGS, 2019a;
USGS, 2019b):

K2
T= 𝐾1 (3.5)
ln( L𝜆 +1)
Where:
𝑇 - Effective at-satellite temperature in Kelvin,

Lλ - Spectral radiance at the sensor’s aperture in (Watts/ (m² * sr * µm)),

K1- Band-specific thermal conversion constant one from the metadata and K2 - Band-specific
thermal conversion constant two from the metadata as presented in Table 3.7.
Table 3. 7: Meta data of thermal constants for Landsat 7 ETM+ (Band 6) & Landsat 8 TIRS (Band 10)

Landsat 7 ETM+ (Band 6) Landsat 8 TIRS (Band 10)

Thermal constant (K1) Thermal constant (K2) Thermal constant (K1) Thermal constant (K2)
666.09 1282.71 774.8853 1321.0789
Source: Metadata file of Landsat 7 ETM+ (2006) and Landsat 8 OLI/TIRS (2016 and 2021)

3.3.5.3 Land surface emissivity estimation

Before LST retrieval, land surface emissivity was computed after calculating NDVI. Equations
3.8 and 3.6 (Van de Griend and Owe, 1993) are utilized to obtain NDVI and LSE respectively.

LSE (Ɛ) = 1.009 + 0.047 * ln (NDVI) (3.6)

3.3.5.4 LST estimation

Finally, Mono-Window Algorithm (Kumari et al., 2018; Rongali et al., 2018; Simwanda et al.,
2019) is implemented to retrieve LST on ERDAS Imagine 15 model maker using equation 3.7.

𝑇
LST (oC) = 𝑇 – 273.15 (3.7)
1+(𝜆 𝑋 𝜌) 𝑙𝑛Ɛ

Where 𝜆 = average wave- length of emitted radiance (λ = 11.45 µm and 10.89 µm for Landsat 7

ETM+ Band 6 and Landsat-8 band 10 respectively; ρ = h × c/σ (1.438 × 10-2 mK),
σ = Boltzmann constant (1.38 × 10-23 J/K), h = Planck’s constant (6.626 × 10-34 Js) and c =
velocity of light (2.998 × 10-8 m/s); and Ɛ is the land surface emissivity. Therefore, ρ = 14380
(Rongali et al, 2018).
Neway Kifle [2021] 40 Remote Sensing and Geo-Informatics, AAU
 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

3.3.5.5 LST result validation

The common methods used for validating LST derived from TIRS sensors include near-surface air
temperature, ground-based measurements (Avdan and Jovanovska, 2016), and LST retrieved from
other TIR sensors (Yosef Mengistu et al., 2017). In this study, LST retrieved from another sensor
i.e., Moderate Resolution Imaging Spectrometer (MODIS) product MOD11A1 of 1 km spatial
resolution was used for comparison and validation purposes.

The validation was carried out in several steps. First, the MODIS LST product obtained from the
Earth Data website was downscaled multiplying by 0.02. Second, a total of 500 random points
were generated and LST values were extracted to the generated points using ArcGIS 10.6.1 for
each study year. Then, for the sake of comparison and validation the original Landsat 7 ETM+ and
Landsat 8 LST result (30m) is resampled to 1km spatial resolution to match with MODIS LST
product using bilinear interpolation on ERDAS Imagine (Yosef Mengistu et al., 2017). Eventually,
the difference between the MODIS LST product and retrieved LST was computed for each of the
generated points.

3.3.6 Land use land cover indices

There are several kinds of LULC indices used in geospatial studies. In this study, LULC indices
such as Normalized Difference Vegetation Index (NDVI), Normalized Difference Built-up Index
(NDBI), and Modified Normalized Difference Water Index (MNDWI) has been used to analyze
the relationship of LST against Vegetation, the built-up environment, and the blue landscape land
use classes respectively. The indices were computed from Landsat ETM+ (for the year 2006) and
Landsat 8 OLI/TIRS (for the years 2016 and 2021) imageries. NDVI was developed by (Rouse et
al., 1973) and computed as follows:

(ρNIR− ρred)
NDVI = (3.8)
(ρNIR+ ρred)

Where; ρred − reflectance in the red band (Band 3 for the Landsat 7 ETM+ sensor and Band 4
for the Landsat 8 OLI/TIRS sensor) and ρNIR- reflectance in the Near-Infrared band (Band 4 and
Band 5 for the Landsat 7 ETM+ and Landsat 8 OLI/TIRS sensors respectively). NDVI value
ranges between -1 and +1 and it enhances all vegetation and resulting in a positive value. Soil may
have a nearby zero value, while waterbody features have negative values (John et al., 2020).

Neway Kifle [2021] 41 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

The Normalized Difference Built-up Index has been developed by (Zha et al., 2003) and expressed
as follows:

(𝜌SWIR1 − 𝜌𝑁𝐼𝑅)
NDBI = (3.9)
(𝜌SWIR1 + 𝜌𝑁𝐼𝑅)

Where; 𝜌SWIR- reflectance in the Short-Wave Infrared band (Band 5 in Landsat 7 ETM+ and
Band 6 in Landsat 8 OLI/TIRS). The values of NDBI ranges from -1 to +1 where built-up areas
tend to have higher positive values. Water bodies have negative NDBI values and values closer to
zero NDBI values indicate areas with vegetation cover (Abel Balew and Tesfaye Korme, 2020).

The MNDWI is calculated using equation 3.9 as suggested by (Hanqiu Xu, 2006).

(𝜌Green − 𝜌𝑆𝑊𝐼𝑅1)
MNDWI = (3.10)
(𝜌Green + 𝜌𝑆𝑊𝐼𝑅1)

Where; 𝜌Green- reflectance in the green band (Band 2 for Landsat 7 ETM+ and Band 3 for Landsat
8 OLI/TIRS). The result of the MNDWI produces three results: (i) water with greater positive
values. (ii) Built-up areas having negative values; and (3) soil and vegetation with negative values.

3.3.7 Statistical analysis

The relationship between LST with NDVI, NDBI, MNDWI and altitude was analyzed using linear
regression. The regression was carried out for all the study years with LST as the dependent
variable and NDVI, NDBI, MNDWI and altitude as the independent variables. To perform the
regression analysis, 3 steps have been followed.

First, a total of 1000 random points were generated from NDVI, NDBI, MNDWI and (DEM)
altitude raster for the study years 2006, 2016, and 2021.

Afterward, values were extracted from the raster files for each generated point. Create random
points tool and extract multi values to points tools were utilized from ArcGIS 10.6.1 to generate
the random points and extract the respective values respectively.

Finally, spatial scatter plots with Pearson’s correlation coefficient (r) and coefficient of
determination (R2) values were calculated to determine the degree of relationship and the amount
of information shared between LST and LULC indices using Stata 15 software (Naeem et al., 218).
The linear regression coefficient (r) is interpreted based on the strength of ±0.50 to ±1 represents
a strong correlation, ±0.30 to ±0.49 represents a medium correlation, and below ±0.30 represents
Neway Kifle [2021] 42 Remote Sensing and Geo-Informatics, AAU
 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

a weak correlation (Statistics Solutions, 2021). The general workflow of the study is shown in
Figure 3.14

Figure 3. 14: General workflow of the study

Neway Kifle [2021] 43 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

CHAPTER FOUR

4. RESULTS
4.1. Object-based image classification accuracy assessment
Tables 4.1, 4.2, and 4.3 shows several LULC classification accuracy measures such as overall
accuracy, producer’s accuracy, user’s accuracy, and Cohen’s kappa coefficient.

Table 4. 1: Accuracy assessment of LULC classification during 2006 using a confusion matrix
Urban Bare Row User's
Classified Water farmland Vegetation Built-up land total accuracy
Water 27 1 1 0 0 30 90%
Urban farmland 0 47 0 1 2 50 94%
Vegetation 1 1 38 0 0 40 95%
Built-up 0 2 2 56 0 60 93%
Bare land 0 1 2 0 37 40 92%
Column total 28 52 43 57 39 220
Producer's accuracy 96% 90% 88% 98% 95%
Overall Classification Accuracy = 93% Overall Kappa Statistics = 91%

Table 4. 2: Accuracy assessment of LULC classification during 2016 using a confusion matrix
Urban Bare Row User's
Classified Water farmland Vegetation Built-up land Total accuracy
Water 29 0 1 0 0 30 96%
Urban farmland 1 46 1 1 1 50 92%
Vegetation 0 2 38 0 0 40 95%
Built-up 0 0 1 57 2 60 95%
Bare land 0 0 1 2 37 40 90%
Column total 30 48 42 60 40 220
Producer's accuracy 96% 95% 90% 95% 92%
Overall Classification Accuracy = 94% Overall Kappa Statistics = 92%

Table 4. 3: Accuracy assessment of LULC classification during 2021 using a confusion matrix
Urban Bare Row User's
Classified Water farmland Vegetation Built-up land total accuracy
Water 27 1 2 0 0 30 90%
Urban farmland 1 37 1 0 1 40 92%
Vegetation 0 0 49 0 1 50 98%
Built-up 0 2 0 58 0 60 96%
Bare land 0 0 2 0 38 40 95%
Column total 28 40 54 58 40 220
Producer's accuracy 96% 92% 90% 100% 95%
Overall Classification Accuracy = 95% Overall Kappa Statistics = 93%

Neway Kifle [2021] 44 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

The final object-based LULC classification result shows an overall accuracy of 93%, 94%, and
95% during 2006, 2016, and 2021 respectively. Whereas, Cohen’s kappa coefficient became 91%,
92%, and 93% for 2006, 2016, and 2021 classification results respectively. (Tables 4.1 - 4.3).

4.2. Spatiotemporal dynamics in land use/and cover

Spatiotemporal analysis of LULC was carried by applying an object-based image analysis
approach using high-resolution satellite imagery as an input. The land cover of the study area was
categorized into the built-up, vegetation, urban farmland, bare land, and water land cover classes.
Table 4.4 portrays the land cover classification and analysis result during January 2006, 2016, and
2021.

In January 2006 the largest share of the LULC goes to built-up with a total of 211. 7 sq.km (40.2
%) followed by urban farmland, vegetation and bare land accounting 141.3 sq.km (26.8%), 105.0
sq.km (19.9%) and 67.7 sq.km (12.9%) respectively.

In January 2016, the built-up had grown more in aerial coverage having a total area of 301.1 sq.
km which is 57.1% of the total land cover in the city. During this year the aerial extent of
vegetation, urban farmland, and bare land had noticeably diminished to 95.3 sq. km (18.1%), 93.2
sq. km (17.7%), and 36.1 sq. km (6.8%) respectively.

Table 4. 4: Land use/cover dynamics during 2006, 2016 and 2021

2006 2016 2021 Annual rate of change
Land cover
Area Area Area Area
class Percent Percent Percent Percent
(sq.km) (sq.km) (sq.km) (sq.km)
Built-up 211.7 40.2 301.1 57.1 350.6 66.5 9.3 4.4
Vegetation 105.0 19.9 95.3 18.1 95.9 18.2 -0.6 -0.6
Urban farmland 141.3 26.8 93.2 17.7 57.7 10.9 -5.6 -3.9
Bare land 67.7 12.8 36.1 6.8 21.4 4.1 -3.1 -4.6
Water 1.6 0.3 1.5 0.3 1.6 0.3 0.0 0.0
Total 527.2 100.0 527.2 100.0 527.2 100.0
Source: Computed using ArcGIS 10.6.1
Likewise, in January 2021 the built-up land remains the dominant land cover class considerably
expanding its coverage to 350.6 sq. km representing 66.5% of the total land coverage. Vegetation
accounts for 95.9 sq. km (18.2%) resembling a very slight increase i.e., 0.6 sq. km increase between
the period January 2016 and 2021. In contrast, urban farmland and bare land had revealed a drastic
decline in aerial coverage bringing the figures down to 57.7 sq. km (10.9%) and 21.4sq.km (4.1%)
respectively. The water land cover class accounts 1.6 sq.km (0.3%), 1.5 sq.km (0.3%) and 1.6
Neway Kifle [2021] 45 Remote Sensing and Geo-Informatics, AAU
 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

sq.km (0.3%) during January 2006, 2016 and 2021 respectively. The result also depicts that, during
the period between January 2006 and 2021 the built-up had substantially grown by 138.9 sq. km
which is 65.7% of the built-up in 2006. Whereas, urban farmland, bare land, and vegetation have
reduced by 83.7 sq. km, 46.2 sq. km, and 9.2 sq. km respectively. The annual increasing rate of
change in built-up from 2006 to 2021 was 4.4%, but, bare land, urban farmland, and vegetation
declined annually by 4.6%, 3.9%, and 0.6% respectively.

Figure 4. 1: LULC map A) during 2006, B) during 2016 and C) during 2021

Neway Kifle [2021] 46 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

The spatial LULC distribution presented in figure 4.1 shows that the vegetation cover is mainly observed
in the Northern part of the city and along the river banks throughout the study periods. Bare land was
observed in the North-Eastern, Eastern, Central, Western, Southern, South-Western, and South-Eastern
parts of the city. While urban farmland was distributed in the North-Eastern, Eastern (only during 2006),
Western (only during 2006), Southern, South-Western, and South-Eastern parts of the city. Figure 4.1 also
reveals that the city expansion is heading towards the Eastern, Western, and Southern parts of the city.

4.2.1 Land use/cover conversion matrix

The land use/land cover matrices show the extent and direction of what is changing to what i.e.
which land use/cover class is transforming into another and which remains unchanged. The results
from change matrices revealed 20 types of conversion between the land use/cover categories
(Table 4.5 and Appendix 6). However, the conversion between water and other land use classes
was not taken into account as there was no considerable conversion from and to water. The result
illustrates that during the whole study period (2006 to 2021), most of the urban farmland was
converted into built-up by 75.9 sq. km (53.7%) and followed by the conversion into vegetation by
13.3 sq. km (9.4%) and bare land by 7.4sq.km (5.3%). In this manner, 40.2 sq. km (59.4%), 9.9
sq. km (14.6%), and 8.1 sq. km (11.9%) of bare land were converted into built-up, urban farmland,
and vegetation respectively. Similarly, 50.5 sq.km (48.1%), 2.5 sq.km (2.3%) and 1.5 sq.km (1.5%)
of vegetation was converted in to built-up, urban farmland and bare land cover categories
respectively. The result further reveals that the built-up land use category was converted into
vegetation, bare land, and urban farmland by 22.2 sq. km (10.5%), 3.7 sq. km (1.8%), and 2.9 sq.
km (1.8%) respectively.

Whereas, 182.2 sq.km (86.0%) of built-up, 49.5 sq.km (47.2%) of vegetation, 41.0 sq.km (29.1%)
of urban farmland and 8.5 sq.km (11.9%) of bare land remained unaltered during the last 15 years.

Neway Kifle [2021] 47 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Table 4. 5: Land use/cover conversion matrix during 2006, 2016 and 2021
Observed change
Between 2006 and Between 2016 and Between 2006 and
2016 2021 2021
Change type
Area Area Area Area Area Area
(Sq.km) (%) (Sq.km) (%) (Sq.km) (%)
Built-up to Bare land 6.8 3.2 4.7 1.5 3.7 1.8
Built-up (unchanged) 170.8 80.8 271.9 90.3 182.1 86.0
Built-up to Urban farmland 6.9 3.3 5.7 1.9 2.9 1.4
Built-up to Vegetation 26.7 12.6 18.9 6.3 22.2 10.5
Vegetation to Bare land 3.5 3.3 0.8 0.8 1.5 1.5
Vegetation to Built-up 43.8 41.7 26.5 27.8 50.5 48.1
Vegetation to Urban farmland 4.9 4.7 1.3 1.4 2.5 2.3
Vegetation (unchanged) 52.4 49.9 66.5 69.7 49.5 47.2
Urban farmland to Bare land 12.5 8.8 3.4 3.7 7.4 5.3
Urban farmland to Built-up 55.1 38.9 31.4 33.6 75.9 53.7
Urban farmland (unchanged) 63.9 45.3 46.8 50.2 41.0 29.1
Urban farmland to Vegetation 9.2 6.5 5.38 5.8 13.3 9.4
Bare land (unchanged) 12.9 19.1 12.2 33.9 8.5 12.5
Bare land to Built-up 31.1 46.1 18.9 52.3 40.2 59.4
Bare land to Urban farmland 16.3 24.1 2.5 6.9 9.9 14.6
Bare land to Vegetation 7.1 10.5 2.1 5.9 8.1 11.9
Source: Computed using ArcGIS 10.6.1
4.3. Land surface temperature validation result
The estimated LST from Landsat 7 ETM+ band 6 and Landsat 8 TIRS band 10 were compared
and validated against LST data products of MODIS (MOD11A1) Version 6 daily 1km spatial
resolution. A total of 500 random points were generated for each study year and the LST values
were extracted from both MODIS LST and estimated LST raster. The validation result
demonstrated that 86.6% (433-pixel LST values), 81.4% (407-pixel LST values) and 80.2% (401-
pixel LST values) extracted shows a maximum difference of ±2oC between MODIS LST product
and estimated LST during January 2006, 2016, and 2021 respectively.

4.4. Spatiotemporal pattern of land surface temperature

In this study, the Landsat 7 ETM+ thermal band (band 6) was used to retrieve LST during the year
2006. Whereas, Landsat 8 TIRS band 10 was utilized to estimate LST during 2016 and 2021 for
the city of Addis Ababa. Table 4.6 shows the minimum, maximum, mean, and standard deviation
of LST during the years 2006, 2016, and 2021. The result depicts that the highest minimum
temperature was estimated during January 05, 2021, having a minimum LST value of 15.9oC
followed by January 24, 2016, and January 04, 2006, with a minimum LST value of 15.1oC and

Neway Kifle [2021] 48 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

9.6oC respectively. Similarly, the highest maximum temperature was recorded during January 05,
2021, with a maximum LST value of 37.3oC. In contrast, the maximum temperature was lower
during January 24, 2016, and January 04, 2006, representing a maximum LST value of 34.9oC and
32.0oC respectively.

Overall, the mean LST result showed an increasing trend from 2006 to 2016 and 2021 in Addis
Ababa. The mean LST value during January 05, 2021, was 28.2oC which is comparatively higher
than 27.2oC and 25.8oC on January 24, 2016, and January 04, 2006, respectively. In other words,
the average LST value increased by 1.4oC (between 2006 and 2016), by 1.0oC (between 2016 and
2021), and by 2.4oC (between 2006 and 2021).

Table 4. 6: Descriptive statistics of LST during 2006, 2016 and 2021

LST (oC) Mean LST difference
MIN MUX Mean STD 2006-2016 2016-2021 2006-2021
04 January, 2006 9.6 32.0 25.8 2.6 1.4 1.0 2.4
24 January, 2016 15.1 34.9 27.2 2.5
05 January, 2021 15.9 37.3 28.2 2.8
Source: Computed
4.5. Land surface temperature variation among LULC classes
The estimated value of LST was extracted using the zonal statistics tool in ArcGIS 10.6.1 software
for each land use/cover category. The result presented in Figure 4.2 and Table 4.7 confirmed that
there was LST variation among the land use/cover classes in the study area.

Table 4. 7: Descriptive statistics of LST among LULC Classes during 2006, 2016 and 2021
LST (oC), 2006 LST (oC), 2016 LST (oC), 2021
Class name
Min MUX Mean STD Min MUX Mean STD Min MUX Mean STD
Built-up 16.4 32.0 25.8 2.6 17.2 34.9 27.6 1.6 17.7 36.7 28.6 2.1
Vegetation 9.6 29.9 24.3 3.0 15.1 33.5 25.9 2.9 15.9 35.8 26.0 3.4
Urban farmland 14.0 31.3 26.9 1.8 15.1 34.6 27.9 2.7 16.9 37.0 29.5 2.1
Bare land 13.1 32.0 26.9 1.9 18.9 34.9 28.7 1.8 18.8 37.2 30.1 2.1
Water 11.6 29.9 24.8 2.1 17.1 31.9 25.5 2.2 16.7 34.5 26.0 2.9
Source: Computed

Table 4.7 portrays that bare land experienced the highest mean LST during the whole range of the
study period having an average LST value of 26.9oC, 28.7oC, and 30.1oC in January 2006, 2016,
and 2019 respectively. Urban farmland tends to have the second-highest mean LST of 26.9oC,
27.9oC and 29.5oC in January 2006, 2016, and 2021 respectively. Built-up exhibited the third-
highest average LST having mean LST values of 25.8oC, 27.6oC, and 28.6oC during 2006, 2016,

Neway Kifle [2021] 49 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

and 2021 respectively. On the other hand, the lowest mean LST was shown by vegetation with a
mean LST of 24.3oC in 2006 and 26.0oC in 2021. While the mean LST of water slightly lowers
that of vegetation (25.5oC vs. 25.9oC) during 2016.

Figure 4. 2: LST map A) during 2006, B) during 2016 and C) during 2021

4.6 Statistical analysis

Linear regression analysis was performed for all the study years, with LST as the dependent
variable and NDBI, NDVI, MNDWI and altitude as predictor variables using Stata 15 software

Neway Kifle [2021] 50 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

(Appendices 1, 2, 3 and 4). Table 4.7 gives the results for Pearson’s correlation coefficient (r) and
the coefficient of determination (R2) while the scatter plots are shown in Figures 4.4, 4.6, 4.8 and
4.9. The results of R2 values during all the study years reveal that NDBI is a better predictor of
LST compared to NDVI and MNDWI.

4.6.1 NDVI-LST relationship

Figure 4.3 shows that the maximum NDVI value decreased from 0.60 in 2006 to 0.58 between
2016 and 2021. While minimum NDVI value increased from -0.37 in 2006 to -0.15 and -0.13 in
2016 and 2021 respectively.

Figure 4. 3: NDVI map A) during 2006, B) during 2016 and C) during 2021
Neway Kifle [2021] 51 Remote Sensing and Geo-Informatics, AAU
 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

The statistical results in Table 4.8 and the scatter plot in Figure 4.4 show a strong negative
correlation between NDVI and LST with r values of -0.56, -0.55, and -0.50 during the years 2006,
2016, and 2021 respectively. This means that the lower the NDVI, the higher the LST and vice
versa. The R2 values were found to be 0.31, 0.30, and 0.23 indicating that 31%, 30%, and 23% of
the variations in LST were explained by NDVI (vegetation cover) in 2006, 2016, and 2021
respectively.

Table 4. 8: R2 and r values of LST with NDVI, NDBI, MNDWI and altitude in 2006, 2016 and 2021

NDVI NDBI MNDWI Altitude

Year
r R2 r R2 r R2 r R2
2006 -0.56 0.31 0.57 0.33 -0.27 0.07 -0.39 0.16
2016 -0.55 0.3 0.63 0.4 -0.29 0.08 -0.47 0.22
2021 -0.5 0.23 0.53 0.29 -0.24 0.06 -0.31 0.08
Source: Computed

A B C
Figure 4. 4: Scatter plots of NDVI and LST A) in 2006, B) in 2016 and C) in 2021

4.6.2 NDBI-LST relationship

It is evident from Figure 4.5 that the maximum NDBI values were found to be 0.48, 0.34, and
0.30 during 2006, 2016, and 2021 respectively. Whereas the minimum NDBI values were -0.48, -
0.43, and -0.45 for the years 2006, 2016, and 2021 respectively. The linear regression carried out
indicated that there was a strong positive relationship between NDBI and LST with a correlation
coefficient of r =0.57, 0.63, and 0.53 in 2006, 2016, and 2021 respectively (Table 4.8). The result
also portrays that the coefficient determination (R2) value was the highest during 2016 (0.40) and
reduced to 0.33 and 0.29 in 2006 and 2021 respectively. In other words, NDBI contributed by 40%,
33%, and 29% in LST variation during the respective years 2006, 2016, and 2021.

Neway Kifle [2021] 52 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Figure 4. 5: NDBI map A) during 2006, B) during 2016 and C) during 2021

A B C

Figure 4. 6: Scatter plots of LST and NDBI A) in 2006, B) in 2016 and C) in 2021

Neway Kifle [2021] 53 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

4.6.3 MNDWI-LST relationship

Figure 4.7 illustrates that the MNDWI value ranges from -0.54 to 0.59 in 2006; -0.45 to 0.27 in
2016; and -0.47 to 0.20 in 2021. The Pearson’s correlation results presented in table 4.8 and Figure
4.8 shows a weak negative association between MNDWI and LST with r = -0.27, -0.29, and -0.24
during 2006, 2016, and 2021 respectively.

Figure 4. 7: MNDWI map A) during 2006, B) during 2016 and C) during 2021

Neway Kifle [2021] 54 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

The R2 values were found to be 0.07, 0.08, and 0.06, meaning that LST was influenced by urban
blue landscapes (urban surface water) with an estimate of 7%, 8%, and 6% variations during 2006,
2016, and 2021 respectively (Table 4.8).

A B C
Figure 4. 8: Scatter plots of LST and MNDWI A) in 2006, B) in 2016 and C) in 2021

4.6.4 Altitude-LST relationship

The model values for Pearson correlation coefficient (r) were found to be -0.39, -0.47 and -0.31 in
2006, 2016 and 2021 indicating a moderate negative relationship between altitude and LST. On
the other hand, R2 values became 0.16, 0.22 and 0.08 in 2006, 2016 and 2021 respectively. This
means that 16%, 22% and 8% of the variations in LST were explained by altitude during the years
2006, 2016 and 2021 respectively.

A B C
Figure 4.9: Scatter plots of LST and Altitude A) in 2006, B) in 2016 and C) in 2021

Neway Kifle [2021] 55 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

CHAPTER FIVE
5. DISCUSSIONS
The image classification, which was carried out by implementing Object-based Image Analysis
(OBIA) identified 5 major land use/land cover classes in the study area. These are built-up,
vegetation, urban farmland, bare land, and water. Acceptable accuracy results were attained for
overall accuracy (93%, 94%, and 95 %) and Cohen’s kappa coefficient (91%, 92%, and 93%)
during 2006, 2016, and 2021 respectively (Tables 4.1-4.3). Therefore, this indicates that there is a
high level of agreement between the classified image and the ground truth, suggesting that it is
possible to use the result for further analysis. These accuracy results are in good agreement with
similar studies conducted in the country and elsewhere (Binyam Tesfaw et al., 2014; Gülçin and
Akpinar, 2018).

5.1 Land use and land cover dynamics

The LULC change analysis indicated that the study area has experienced a noticeable land
use/cover change over the past 15 years both spatially and temporally. Overall, in all the study
periods the built-up land cover class experienced a rapid expansion at the expense of vegetation,
urban farmland, and bare land. In contrast, in general, urban farmland, urban green landscapes,
and bare land have shrunken in aerial extent during the last 15 years. This has resulted due to the
unplanned and uncontrolled city urbanization along with population pressure. These results
confirm the research results and conclusions of several previous studies conducted in the city (Rosa
Assaye et al., 2015; Mathias Tesfaye and Tebarek Lika, 2017; Ermiayas Teferi and Hiwot Abreha,
2017; Simwanda et al., 2019).

However, the urban green landscapes showed a very slight increase between 2016 and 2021. This
is due to the recent green interventions like the green legacy campaign and beautifying
sheger/Addis Ababa project as reported in an earlier study by Biruk Terrefe, (2020). On the other
hand, water remains relatively unaltered in aerial coverage and holds the least share of the LULC
in the city during the years 2006 to 2021. Figure 4.1 reveals the city is mainly expanding towards
the South, East, and West directions instead of North which is steep and densely vegetated (Rosa
Assaye et al., 2015).

Neway Kifle [2021] 56 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

5.2 Land use/Land cover conversion matrix

The land use and land cover transformation detected by the conversion matrix depicts that the city
experienced a multidimensional change between different land-use classes. As seen in Table 4.5,
most of the bare land (59.4%), urban farmland (53.7%), and urban green (48.1%) were transformed
into built-up between the study period 2006 and 2021. Mathias Tesfaye and Tebarek Lika, (2017)
conducted research to assess the pace, extent, and pattern of urbanization and its impact on urban-
green areas in the Bole sub-city of Addis Ababa. Their result showed a similar pattern of LULC
transformation among the land use/cover classes. On the other hand, the largest lost share of the
built-up goes to the urban green landscape (10.5%) compared to the conversion of built-up to other
land use classes between the years 2006 and 2021. This implies that the urban redevelopment and
river buffer projects like beautifying Sheger and associated relocation programs which demolished
settlements have contributed to the conversion of built-up into vegetation. This is evident in areas
like Arat killo, Lideta, Kirkos (behind St, Estfanos church), Dejach Wube, etc., where, many
upgrading or renewals, and slum improvement activities have been undergoing (Herslund et al.,
2017; Mathias Tesfaye and Tebarek Lika, 2017). Another possible reason for this is that, according
to the Addis Ababa drainage basin and green areas development office, 2020 during the recent
years, large hectares of trees have been planted along the roadsides, street medians, and public
squares which shades/ hides the paved surface below the street tree canopies.

5.3 Comparison between model LST and MODIS LST product

To assess the reliability of the LST result derived from thermal bands of Landsat 7 ETM+ (Band
6) and Landsat 8 TIRS (Band 10), LST data products of MODIS (MOD11A1) Version 6 daily 1km
spatial resolution was used. The comparison result yielded that 86.6%, 81.4% and 80.2% pixel
LST values extracted lied within ±2oC surface temperature between MODIS LST product and
retrieved LST during January 2006, 2016, and 2021 respectively. This validation result is in line
with some previous studies (Yosef Mengistu et al., 2017; Srivastava et al., 2009) that found a
maximum difference of ±2oC while comparing MODIS LST product and estimated LST from
Landsat 8 TIRS sensor. It should be taken into consideration that there exists a spatial resolution
difference between the sensors i.e. 30m for Landsat 7 ETM+ and Landsat 8 TIRS versus 1km for
MODIS LST product. Besides, MODIS LST is retrieved using the split-window algorithm on
bands 31 and 32 (Wan and Dozier, 1996), while the mono window algorism is utilized in this study.
Therefore, the LST differences are normal and anticipated.

Neway Kifle [2021] 57 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

5.4 LST trend and spatial variation

The results from this study revealed (Table 4.6) that the average LST has increased progressively
during the last fifteen years in the study area. Likewise, the maximum and minimum LST values
also showed an increasing trend during the years under study. Similar LST trends have been
identified in the study area by the results of previous studies (Ermiyas Teferi and Hiwot Abreha,
2017; Dissanayake et al., 2019).

As it is evident from the maps in Figure 4.2, in its spatial pattern, lower LST values were observed
in the Northern part of the city (at foot of Mt. Entoto), along with the river courses and vegetation-
covered areas. While most of the southern, eastern, and central parts of the city are relatively
characterized by higher LST values. These results are supported by the research findings of Rosa
Assaye et al., (2015) and Samson Warkaye et al., (2018) who conducted similar studies in the city.
The standard deviation of the 2021 LST exceeds that of 2006 and 2016 implying that the surface
temperature exhibited a substantial variation in that specific period.

5.5 LST variation among land use/land cover classes

Table 4.7 portrays the statistic values for the five major land cover classes identified. The result
shows that the highest mean LST was observed in bare land with a mean LST value of 26.9oC in
2006, 28.7oC in 2016, and 30.1oC in 2021. Urban farmland experienced the second-highest average
LST compared to vegetation and water having a mean LST of 26.9oC, 27.9oC, and 29.5oC in 2006,
2016, and 2021 respectively. These results confirm the findings of earlier studies (Simwanda et
al., 2019; Abel Balew and Tesfaye Korme, 2020) who stated that radiation-exposed land or bare
soil (farmland without crop) had higher thermal property and radiate energy that is responsible for
raising LST.

The built-up mean LST was found to be higher as compared to the mean LST of vegetation and
water. This implies that non-evaporating and non-transpiring built-up areas (impervious surfaces)
such as residential sites, parking lots, industrial parks and shades, street networks, and other paved
surfaces contribute a lot in increasing LST over the area. These results are coincident with the
findings of previous studies conducted elsewhere (Abu Bakar et al., 2016; Priyankara et al., 2019).
However, in contrast to a previous study (Matiwos Belayhun, 2018) the average LST of built-up
tends to be lower as compared to the average LST of urban farmland and bare land. Conversely,
the mean LST for vegetation and water classes was lesser compared to other land use/cover classes

Neway Kifle [2021] 58 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

throughout the study period. Other scholars (Su et. al, 2011; Ogunjobi et al., 2018; Jovish John et
al., 2020) also witnessed that vegetation cover plays a vital role in lowering LST by providing
shade and cooling the surrounding via the process of evapotranspiration. Similarly, the urban blue
(urban surface water) also has the potential to contribute a cooling effect through the process of
evaporation as reported by Wu et al., (2019).

5.6 The relationship of LST with NDVI, NDBI, MNDWI and altitude
In this study, NDVI-LST, NDBI-LST, MNDWI-LST and altitude-LST relationships were
investigated using linear regression analysis and the statistical results are presented in Table 4.8.
The result showed that NDVI tends to have a strong negative linear relationship with LST in all
the study years. This suggests that the higher the vegetation cover the lower the surface temperature
and vice versa. These results are supported by the findings of previous studies conducted elsewhere
(Sun et al., 2012; Adebowale and Kayode, 2015; Bakar et al., 2016; Hua and Ping, 2018; Mwangi
et al., 2018; Malik et al., 2019).

In contrast, the linear regression analysis revealed that the NDBI-LST relationship was found to
be strong and positive in 2006, 2016, and 2021. This means that as the coverage and density of
built-up areas (impervious surfaces) such as residential sites, parking lots, industrial parks and
shades, street networks, and other paved surfaces increases, LST increases (Tyubee and Anyadike,
2015; Bakar et al., 2016; Malik et al., 2019).

The regression analysis also demonstrated that MNDWI has a weak negative correlation with
surface temperature in the study area. The weak and indirect relationship between MNDWI and
LST indicates that the potential of the urban blue landscapes in affecting LST in the study area is
limited as compared to NDVI, MNDWI and altitude. However, this result is different from Sun et
al., (2012) and Hua and Ping, (2018) who found a strong negative correlation between MNDWI
and LST. The result further shows that moderate negative relationship between altitude and LST
suggesting that the higher the altitude the lower the LST and vice versa (Peng et al., 2020).

Furthermore, based on the R2 values presented in Table 4.8, it is predicted that NDBI and NDVI
are the two most important parameters responsible for the variation of LST in the study area
compared to altitude and MNDWI.

Neway Kifle [2021] 59 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

CHAPTER SIX
6. CONCLUSION AND RECOMMENDATIONS
6.1 Conclusion
This study was mainly focused on mapping effects of urban blue-green landscapes on land surface
temperature using a geospatial techniques in the city of Addis Ababa. The study applied the OBIA
method for LULC classification and identified five major LULC classes i.e. built-up, vegetation,
urban farmland, bare land, and water. The findings of the study revealed that there has been a
substantial transformation between the LULC classes in the city since 2006 in Addis Ababa. The
built-up area was found to be the most dominant land use/cover class and has shown a considerably
expanding trend with an annual growth rate of 4.4% (9.3 sq. km) per year during the last 15 years.
On other hand, the area coverage under urban farmland, vegetation (except between 2016 and
2021), and bare land have shrunken continuously and extensively between the study periods 2006-
2016, 2016-2021, and 2006 -2021. Between the study periods, 2016 to 2021 vegetation showed a
slight increase in area coverage. This is due to the contribution of recent green interventions like
the green legacy campaign, river buffer project/ beautifying Sheger project in raising the green
landscapes coverage in the city. Whereas, the area covered under the urban surface water remained
relatively unaltered for the last 15 years.

While considering the whole range of study periods, the findings from the LULC conversion
matrix demonstrated that 75.9 sq. km (53.7%) of urban farmland, 50.5 sq. km (48.1%) of
vegetation, and 40.2 sq. km (59.4%) of bare land, was transformed into built-up class. Therefore,
it can be concluded that the expansion of built areas such as residential sites, parking lots, industrial
parks and shades, street networks, etc. is the most prominent phenomenon and is responsible for
the shrinking of urban farmland, bare land, and vegetation coverage in the city since 2006. The
spatial analysis result of this study reveals that the city expansion is heading towards the Eastern,
Western, and southern parts of the city instead of the North which is characterized by dense
vegetation and steep slopes.

Moreover, LST was retrieved from the thermal imageries of Landsat 7 ETM + (band 6) and
Landsat 8 TIRS (band 10) using the Mono-Window Algorithm (MWA). The findings indicated
that LST varied both spatially and temporally throughout the study periods. The spatial variation
was mainly due to the variation of different LULC types in the city. Whereas the changes in the

Neway Kifle [2021] 60 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

LULC have contributed to the temporal variation of LST in the city. It was observed that the
minimum, maximum and mean LST values showed an increasing trend since 2006. The descriptive
statistics for LST were extracted for each land use/cover class to examine how LST varied among
the LUL classes. The result showed that the highest mean LST was observed on bare land followed
by urban farmland and built-up land use/cover classes throughout the periods of study. It was also
observed that the average LST showed an increasing trend among all the LULC classes in the last
15 years.

Therefore, the present study had demonstrated that the state of LULC dynamics during the study
period and its potential for modifying LST in the city. The result also revealed the effect of
anthropogenic activities, particularly the built-up areas which experienced a drastic change over
time as a result of the increase in population and the associated infrastructure development in the
city. Hence, the findings of this study have indicated that LULC alteration had contributed to the
modification of LST in Addis Ababa during the period.

Furthermore, the relationship of LST against NDVI, NDBI, MNDWI and altitude was examined
quantitatively using linear regression analysis at a pixel level. The results for Pearson’s coefficient
(r) values during all the study years indicate that a strong negative correlation was observed
between NDVI and LST meaning that the higher the healthy vegetation cover the lower will be
the LST. Unlike NDVI, NDBI showed a strong positive correlation with LST, indicating that the
higher coverage of built-up areas such as residential sites, parking lots, industrial parks and shades,
street networks, etc., the higher will be the surface heating in the city. MNDWI tends to have a
weak negative correlation with LST. Altitude showed a moderate negative relationship with LST.
The regression analysis further indicated that the results of Coefficient of determination (R2) values
were found to be the highest for NDBI followed by NDVI, altitude and MNDWI throughout the
study period. This implies that NDBI, NDVI and altitude are the three most important parameters
to predict LST in the study area compared to MNDWI. Therefore, it can be concluded that built-
up area, vegetation cover and altitude plays the most decisive role in the variation of LST in the
city compared to urban surface water during the last 15 years.

Lastly, the study witnessed the potential of integrating remote sensing, particularly thermal remote
sensing and GIS for investigating the urban thermal environment with respect to different LULC
classes.

Neway Kifle [2021] 61 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

6.2 Recommendations
Based on the findings and conclusions drawn, the following recommendations are suggested:

 This study demonstrated that the drastic growth in urbanization and considerable loss in
vegetation cover have contributed a lot to the alteration of the urban thermal environment.
Therefore, it is highly recommended that when planning and designing, urban policy-
makers and planners should consider diversified green landscapes such as green buildings
(urban buildings with vegetation cover on exterior walls and roofs), street greening, small
and large parks, green trails for walking/cycling, playing grounds, etc.
 Urban blue-green landscapes should better be planned and designed in such a way that, it
could be possible to make adjustments to adapt to changing and forthcoming demands.
 Blue-green development opportunities should be promoted and extended into other sectors
and projects like the greening of business areas, schools, shopping areas, residential
housing estates, industrial areas, and private green areas and gardens, etc.
 While planning and designing for future blue-green interventions in the city, it would be
beneficial to consider the construction of man-made lakes within the green landscapes to
strengthen the cooling effect, as was the case in Friendship Square/Park.
 The results of this study were discussed with the focus on the spatiotemporal pattern of
LULC dynamics and the consequential LST modification in the city. The author realizes
that some other driving forces of LST other than LULC dynamics such as slope, latitude,
albedo, etc., were not considered in this study. Therefore, it is recommended that future
studies should focus to fill this gap. Moreover, the temporal extent should also be widened
to better understand the LULC dynamics and its effect on the LST trend in the city.

Neway Kifle [2021] 62 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

References
A. A. Van De Griend & M. Owe. (1993). On the relationship between thermal emissivity and the
normalized difference vegetation index for natural surfaces, International Journal of Remote
Sensing, 14:6, 1119-1131, DOI: 10.1080/01431169308904400

Abel Balew &, Tesfaye Korme. (2020). Monitoring land surface temperature in Bahir Dar city and
its surrounding using Landsat images. The Egyptian Journal of Remote Sensing and Space
Science. 23. 10.1016/j.ejrs.2020.02.001.

Aboyade, O. (2001). Geographic information systems: application in planning and decision-making

processes in Nigeria. Unpublished paper presented at the Environmental and Technological
unit in the Development Policy Centre, Ibadan.

Abu Bakar, Suzana Binti & Pradhan, Biswajeet & Lay, Usman & Abdullahi, Saleh. (2016). Spatial
assessment of land surface temperature and land use/land cover in Langkawi Island. IOP
Conference Series: Earth and Environmental Science. 37. 012064. 10.1088/1755-
1315/37/1/012064.

Addis Abba city drainage basin and green areas development office. (2020). Annual report on Green
spaces development.

Addis Ababa City Structure Plan. (2017-2027). Draft Final Summary Report [Online] Available at:
https://c40productionimages.s3.amazonaws.com/other_uploads/images/2036_Addis_Ababa_
Structural_Plan_2017_to_2027.original.pdf?1544193458 Retrieved on November 13, 2020

Adebowale, B. I., & Kayode, S. E. (2015, July). Geospatial assessment of urban expansion and land
surface temperature in Akure, Nigeria. In Proceedings of the ICUC9—9th International
Conference on Urban Climate Jointly with 12th Symposium on the Urban Environment,
Toulouse, France (pp. 20-24).

Ahmad, Asmala & Quegan, S. (2012). Analysis of maximum likelihood classification on

multispectral data. Applied Mathematical Sciences. 6. 6425-6436.

Akmalov, S., Blanpain, O., & Masson, E. (2017). Study of ecological changes in syrdarya province
by using the remote sensing and geobia analysis method. Irrigation and Melioration, 2017(2),
15-19.

Neway Kifle [2021] 63 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Ang Kean Hua & Owi Wei Ping. (2018). The influence of land-use/land-cover changes on land
surface temperature: a case study of Kuala Lumpur metropolitan city, European Journal of
Remote Sensing, 51:1, 1049-1069, DOI: 10.1080/22797254.2018.1542976

Arvanitidis, Paschalis & Lalenis, Konstantinos & Petrakos, George & Psycharis, Yannis. (2009).
Economic aspects of urban green space: a survey of perceptions and attitudes. International
Journal of Environmental Technology and Management. 11. 143-168.

Basudeb Bhatta. (2011). Remote Sensing and GIS. Textbook. – Oxford University Press. Textbook.
- Second Edition – 716p.

Bayram Cemil Bilgili and Ercan Gökyer. (2012). Urban Green Space System Planning, Landscape
Planning, Dr. Murat Ozyavuz (Ed.), ISBN: 978-953-51-0654-8, In Tech, [Online] Available at:
http://www.intechopen.com/books/landscape-planning/urban-green-space-system-planning
Retrieved on December 31, 2020

Belaybeh Bufebo & Eyasu Elias. (2021). Land Use/Land Cover Change and Its Driving Forces in
Shenkolla Watershed, South Central Ethiopia. The Scientific World Journal, 2021.

Binyam Alemu, Efrem Garedew, Zewudu Eshetu, & Habtemariam Kassa. (2015). Land use and land
cover changes and associated driving forces in northwestern lowlands of
Ethiopia. International research journal of agricultural science and soil science, 5(1), 28-44.

Binyam Tesfaw & Maeda, Eduardo & Hurskainen, Pekka & Pellikka, Petri. (2014). Object-based
image analysis for distinguishing indigenous and exotic forests in coffee production areas of
Ethiopia. Applied Geomatics. 10.1007/s12518-014-0136-x.

Biruk Terrefe. (2020). Urban layers of political rupture: the ‘new’ politics of Addis Ababa’s
megaprojects, Journal of Eastern African Studies, 14:3, 375-395, DOI:
10.1080/17531055.2020.1774705

Bisquert, Mar & Bégué, Agnès & Deshayes, Michel. (2014). Object-based delineation of
homogeneous landscape units at regional scale based on MODIS time series. International
Journal of Applied Earth Observation and Geoinformation. 47. 72-82.

Bobrinskaya. M., (2012). Remote Sensing for Analysis of Relationships between the Land cover and
Land Surface Temperature in Ten Megacities.

Neway Kifle [2021] 64 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Byrne, Jason & Sipe, Neil. (2010). Green and open space planning for urban consolidation – A
review of the literature and best practice. Urban Research Program.

Calaza, Pedro & Carinanos, Paloma & O'Brien, Liz & Calfapietra, Carlo. (2017). The Cost of
Greening: Disservices of Urban Trees. 10.1007/978-3-319-50280-9_9.

Chavez Jr PS. (1988). An improved dark-object subtraction technique for atmospheric scattering
correction of multispectral data. Remote sensing of environment 24(3):459-479. DOI:
https://doi.org/10.1016/0034- 4257(88)90019-3

Craig m. Trotter. (1991). Remotely-sensed data as an information source for geographical

information systems in natural resource management a review, International Journal of
Geographical Information Systems, 5:2, 225-239, DOI: 10.1080/02693799108927845

CSA. (2007). Summary and Statistical Report of the 2007 population and Housing Census of
Ethiopia. Addis Ababa, Ethiopia

CSA. (2017). Population Projection of Ethiopia for All Regions at Wereda Level from 2014 – 2017.
Addis Ababa, Ethiopia

D. Pozo Vázquez, F.J. Olmo Reyes, L. Alados Arboledas. (1997). A comparative study of algorithms
for estimating land surface temperature from AVHRR Data, Remote Sensing of Environment,
Volume 62, Issue 3, Pages 215-222, ISSN 0034-4257, https://doi.org/10.1016/S0034-
4257(97)00091-6.

De La Barrera, F., Reyes-Paecke, S., Banzhaf, E. (2016). Indicators for green spaces in
contrasting urban settings. Ecological Indicators, 62, 212–219

Dissanayake, D. M. S. L. B., Morimoto, T., Murayama, Y., & Ranagalage, M. (2019). Impact of
landscape structure on the variation of land surface temperature in the sub-Saharan region: A
case study of Addis Ababa using Landsat data (1986–2016). Sustainability, 11(8), 2257.

Dunnet, N., Swanwick, C., Wooley, H. (2002). Improving Urban Parks, Play Areas and
Open Spaces. 217 pp., University of Sheffield. Queen’s Printer. London.

Easkey Britton, Gesche Kindermann, Christine Domegan, Caitriona Carlin. (2020). Blue care: a
systematic review of blue space interventions for health and wellbeing, Health Promotion
International, Volume 35, Issue 1, Pages 50–69, https://doi.org/10.1093/heapro/day103
Neway Kifle [2021] 65 Remote Sensing and Geo-Informatics, AAU
 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Ebrahim Esa Hassen & Mohammed Assen. (2017). Land use/cover dynamics and its drivers in Gelda
catchment, Lake Tana watershed, Ethiopia. Environmental Systems Research. 6.
10.1186/s40068-017-0081-x.

Elif Sertel & Ugur Alganci. (2016). Comparison of pixel and object-based classification for burned
area mapping using SPOT-6 images, Geomatics, Natural Hazards and Risk, 7:4, 1198-1206,
DOI: 10.1080/19475705.2015.1050608

Enderle, D.I., & Weih, R.C. (2005). Integrating Supervised and Unsupervised Classification
Methods to Develop a More Accurate Land Cover Classification. Journal of the Arkansas
Academy of Science, 59, 65-73.

Epstein, J., Payne, K., and Kramer, E. (2002). Techniques for mapping suburban sprawl. Photogr.
Eng. Remote Sens. 63, 913–918.

Ermias Teferi & Hiwot Abreha. (2017). Urban Heat Island Effect of Addis Ababa City: Implications
of Urban Green Spaces for Climate Change Adaptation. 10.1007/978-3-319-49520-0_33.

Filgueiras, R., Mantovani, E. C., Althoff, D., Dias, S. H., & Cunha, F. F. D. (2019). Sensitivity of
evapotranspiration estimated by orbital images under influence of surface
temperature. Engenharia Agrícola, 39(SPE), 23-32.

Forest Research. (2021). What is greenspace/green infrastructure (GI)? Types of greenspace [Online]
Available at: https://www.forestresearch.gov.uk/tools-and-resources/urban-regeneration-and-
greenspace-partnership/greenspace-in-practice/types-of-greenspace/ Retrieved on January 9,
2021

Gašparović, M. (2020). Urban growth pattern detection and analysis. In Urban Ecology (pp. 35-48).
Elsevier.

Gledhill, David & James, Philip. (2008). Rethinking Urban Blue Spaces from a Landscape
Perspective: Species, scale and the human element. Band. 42. 151-164.

Guha, Subhanil & Govil, Himanshu & Dey, Anindita & Gill, Neetu. (2018). Analytical study of land
surface temperature with NDVI and NDBI using Landsat 8 OLI and TIRS data in Florence and
Naples city, Italy. European Journal of Remote Sensing. 51. 667-678.
10.1080/22797254.2018.1474494.

Neway Kifle [2021] 66 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Gülçin, Derya & Akpinar, Abdullah. (2018). Mapping Urban Green Spaces Based on an Object-
Oriented Approach. Bilge International Journal of Science and Technology Research. 2. 71-
81. 10.30516/bilgesci.486893.

Gunawardena, Kan-Chane & Wells, Mike & Kershaw, Tristan. (2017). Utilizing green and bluespace
to mitigate urban heat island intensity. Science of The Total Environment. 584-585. 1040-1055.
10.1016/j.scitotenv.2017.01.158.

GUO, Q., KELLY, M., GONG, P. and LIU, D. (2007). An object-based classification approach in
mapping tree mortality using high spatial resolution imagery. GIS science & remote sensing,
44, pp. 24–47.

Hanqiu Xu. (2006): Modification of normalized difference water index (NDWI) to enhance open
water features in remotely sensed imagery, International Journal of Remote Sensing, 27:14,
3025-3033

Haq, S. M. A. (2011). Urban green spaces and an integrative approach to sustainable

environment. Journal of Environmental Protection, 2(5): 601-608.

Herslund, Lise & Backhaus, Antje & Fryd, Ole & Jørgensen, Gertrud & Jensen, Marina &
Limbumba, Tatu & Liu, Li & Mguni, Patience & Mkupasi, Martha & Habtemariam, Liku &
Yeshitela, Kumelachew. (2017). Conditions and opportunities for green infrastructure –
Aiming for green, water-resilient cities in Addis Ababa and Dar es Salaam. Landscape and
Urban Planning. 180. 10.1016/j.landurbplan.2016.10.008.

Hillel, D., & Hatfield, J. L. (Eds.). (2005). Encyclopedia of Soils in the Environment (Vol. 3).
Amsterdam: Elsevier.

Jennings V., Browning M.H.E.M., Rigolon A. (2019) Friend or Foe? An Overview of the Services
and Disservices from Urban Green Spaces. In: Urban Green Spaces. Springer Briefs in
Geography. Springer, Cham. https://doi.org/10.1007/978-3-030-10469-6_2

Jiménez-Muñoz, J.C.; Cristóbal, J.; Sobrino, J.A.; Sòria, G.; Ninyerola, M.; Pons, X. (2009).
Revision of the Single-Channel Algorithm for Land Surface Temperature Retrieval from
Landsat Thermal-Infrared Data. IEEE Trans. Geosci. Remote Sens. 47, 339–349.

Neway Kifle [2021] 67 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Jimenez-Munoz, Juan-Carlos & Sobrino, Jose & Skokovic, Drazen & Mattar, Cristian & Cristóbal
Rosselló, Jordi. (2014). Land Surface Temperature Retrieval Methods from Landsat-8 Thermal
Infrared Sensor Data. Geoscience and Remote Sensing Letters, IEEE. 11. 1840-1843.
10.1109/LGRS.2014.2312032.

Jimenez-Munoz, Juan-Carlos & Sobrino, Jose. (2003). A generalized single-channel method for
retrieving land surface temperature from remote sensing data (vol 109, art no D08112, 2004).
Journal of Geophysical Research. 108. 10.1029/2003JD003480.

Jimenez-Munoz, Juan-Carlos & Sobrino, Jose. (2008). Split-Window Coefficients for Land Surface
Temperature Retrieval from Low-Resolution Thermal Infrared Sensors. Geoscience and
Remote Sensing Letters, IEEE. 5. 806 - 809. 10.1109/LGRS.2008.200163

Jose G V H, Karina P, Patricia H. (2018). Urban Green Spaces as a Component of an Ecosystem

Functions, Services, Users, Community 027 Involvement, initiatives, and Actions. Int J
Environ Sci Nat Res. 8(1): 555730. DOI: 10.19080/IJESNR.2018.08.555730.

Jovish John, G. Bindu, B. Srimuruganandam, Abhinav Wadhwa. (2020): Land use/land cover and
land surface temperature analysis in Wayanad district, India, using satellite imagery, Annals of
GIS, DOI: 10.1080/19475683.2020.1733662

Käfer, Pâmela Suélen, Rolim, Silvia Beatriz Alves, Diaz, Lucas Ribeiro, Rocha, Nájila Souza da,
Iglesias, María Luján, & Rex, Franciel Eduardo. (2020). Comparative Analysis of Split-
Window and Single-Channel Algorithms for Land Surface Temperature Retrieval of a Pseudo-
invariant Target. Boletim de Ciências Geodésicas, 26(2), e2020008.
.https://doi.org/10.1590/s1982-21702020000200008

Kumari, Babita & Tayyab, Mohammad & Shahfahad & Salman, & Mallick, Javed & Khan,
Mohammad Firoz & Rahman, Atiqur. (2018). Satellite-Driven Land Surface Temperature
(LST) Using Landsat 5, 7 (TM/ETM+ SLC) and Landsat 8 (OLI/TIRS) Data and Its
Association with Built-Up and Green Cover Over Urban Delhi, India. 1. 63-78.
10.1007/s41976-018-0004-2.

Kumelachew Yeshitela, Liberatus Mrema, Alazar Assefa and Simon Mpyanga. (2015). Water
Resilient Cities in Africa: Green area typologies and mapping of green structures in Addis
Ababa and Dar es Salaam. [Online] Available at:
Neway Kifle [2021] 68 Remote Sensing and Geo-Informatics, AAU
 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

https://ign.ku.dk/english/research/landscapearchitectureplanning/landscapetechnology/water-
green-africa/projectpublications/filer/Report1-WP1-GREEN-AREA-TYPOLOGIES-AND-
MAPPING-OF-GREEN-STRUCTURES.pdf Retrieved on January 06, 2021.

Lee, Andrew & Jordan, Hannah & Horsley, Jason. (2015). Value of urban green spaces in promoting
healthy living and wellbeing: Prospects for planning. Risk Management and Health care
Policy. 8. 131. 10.2147/RMHP.S61654.

Lillesand, T., Kiefer, R. W., & Chipman, J. (2015). Remote sensing and image interpretation. John
Wiley & Sons.

Liu, Desheng & Xia, Fan. (2010). Assessing object-based classification: Advantages and limitations.
Remote Sensing Letters. 1. 187-194. 10.1080/01431161003743173.

Malik, M. S., Shukla, J. P., & Mishra, S. (2019). Relationship of LST, NDBI, and NDVI using
Landsat-8 data in Kandaihimmat watershed, Hoshangabad, India.

Mathias Tesfaye Abebe & Tebarek Lika Megento. (2016). The city of Addis Ababa from ‘forest city’
to ‘urban heat island’: Assessment of urban green space dynamics. Journal of Urban and
Environmental Engineering. 10. 254-262. 10.4090/juee. 2016. V10n2.254262.

Mathias Tesfaye Abebe & Tebarek Lika Megento. (2017). Urban green space development using
GIS-based multi-criteria analysis in Addis Ababa metropolis. Applied Geomatics, 9, 247-261.

Matiwos Belayhun Haylemariyam. (2018). Detection of Land Surface Temperature in Relation to

Land Use Land Cover Change: Dire Dawa City, Ethiopia. J Remote Sens GIS 7: 245.
doi:10.4172/2469-4134.1000245

McMillin L.M. (1975). Estimation of sea surface temperatures from two infrared window
measurements with different absorption. J. Geophys. Res.; 80:80–82. Doi:
10.1029/JC080i036p05113.

Melissa Haeffner, Douglas Jackson-Smith, Martin Buchert, Jordan Risley. (2017). Accessing blue
spaces: Social and geographic factors structuring familiarity with, use of, and appreciation of
urban waterways, Landscape and Urban Planning, 167, 136-146, ISSN 0169-2046,
https://doi.org/10.1016/j.landurbplan.2017.06.008.

Neway Kifle [2021] 69 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Mwangi, P.W., Karanja, F.N. and Kamau, P.K. (2018). Analysis of the Relationship between Land
Surface Temperature and Vegetation and Built-Up Indices in Upper-Hill, Nairobi. Journal of
Geoscience and Environment Protection, 6, 1-16.

Naeem, Shahid; Cao, Chunxiang; Qazi, Waqas A.; Zamani, Mehdi; Wei, Chen; Acharya, Bipin K.;
Rehman, Asid U. (2018). "Studying the Association between Green Space Characteristics and
Land Surface Temperature for Sustainable Urban Environments: An Analysis of Beijing and
Islamabad" ISPRS Int. J. Geo-Inf. 7, no. 2: 38. https://doi.org/10.3390/ijgi7020038

NASA-National Aeronautics and Space Administration, (2000). Land Surface Temperature [Online]
Availableat:https://earthobservatory.nasa.gov/globalmaps/MOD_LSTD_M#:~:text=Land%20
surface%20temperature%20is%20how,the%20atmosphere%20to%20the%20ground.
Retrieved on October 30, 2020.

Ogunjobi KO, Adamu Y, Akinsanola AA, Orimoloye IR. (2018). Spatiotemporal analysis of land use
dynamics and its potential indications on land surface temperature in Sokoto Metropolis,
Nigeria. R. Soc. open sci. 5:180661.

Panduro, Toke & Veie, Kathrine. (2013). Classification and valuation of urban green spaces—A
hedonic house price valuation. Landscape and Urban Planning. 120. 119–128.
10.1016/j.landurbplan.2013.08.009.

Pataki, Diane & Carreiro, Margaret & Cherrier, Jennifer & Grulke, Nancy & Jennings, Viniece &
Pincetl, Stephanie & Pouyat, Richard & Whitlow, Thomas & Zipperer, Wayne. (2010).
Coupling biogeochemical cycles in urban environments: ecosystem services, green solutions,
and misconceptions. Frontiers in Ecology and the Environment. 9. 27-36. 10.2307/41149674.

Patel, S.K., Verma, P., Singh, G.S. (2019). Agricultural growth and land use land cover change in
peri-urban India. Environmental Monitoring and Assessment 191 (9), 600.

Pauleit, Stephan & Hansen, Rieke & Rall, Emily & Zölch, Teresa & Andersson, Erik & Luz, Ana &
Száraz, Luca & Tosics, Iván & Vierikko, Kati, (2017). Urban Landscapes and Green
Infrastructure. 10.1093/acrefore/9780199389414.013.23.

Neway Kifle [2021] 70 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Peng, Xiaoxue & Wu, Wenyuan & Zheng, Yaoyao & Sun, Jingyi & Hu, Tangao & Wang, Pin. (2020).
Correlation analysis of land surface temperature and topographic elements in Hangzhou,
China. Scientific Reports. 10. 10451. 10.1038/s41598-020-67423-6.

Priyankara, P., Ranagalage, M., Dissanayake, D., Morimoto, T., Murayama, Y. (2019). Spatial
Process of Surface Urban Heat Island in Rapidly Growing Seoul Metropolitan Area for
Sustainable Urban Planning Using Landsat Data (1996– 2017). Climate 7.

Qin, Zhihao & Karnieli, Arnon & Berliner, Pedro. (2010). A Mono-Window Algorithm for
Retrieving Land Surface Temperature from Landsat TM data and its Application to the Israel-
Egypt Border Region. International Journal of Remote Sensing. 22. 3719-3746.
10.1080/01431160010006971.

Radek Dušek & Renata Popelková. (2019). Water in most important towns of the Czech
Republic, Journal of Maps, 15:2, 425-435, DOI: 10.1080/17445647.2019.1601136

Rajan KS, Ramachandra TV. (2013). Land Surface Temperature Responses to Land Use Land Cover
Dynamics. Geoinfor Geostat An Overview 1:4. doi:10.4172/2327-4581.1000112.
10.4172/2327-.

Rakhshandehroo, Mehdi & Afshin, Sahrakar & Mohd Yusof, Mohd Johari. (2017). Terminology of
Urban Open and Green Spaces.

Rongali, Gopinadh & Keshari, A.K. & Gosain, A. & Khosa, Rakesh. (2018). Split-Window
Algorithm for Retrieval of Land Surface Temperature Using Landsat 8 Thermal Infrared Data.
Journal of Geovisualization and Spatial Analysis. 2. 10.1007/s41651-018-0021-y.

Rongali, Gopinadh & Keshari, A.K. & Gosain, Ashvani & Khosa, Rakesh. (2018). A Mono-Window
Algorithm for Land Surface Temperature Estimation from Landsat 8 Thermal Infrared Sensor
Data: A Case Study of the Beas River Basin, India. Pertanika Journal of Science and
Technology. 26.

Rosa Assaye, K.V. Suryabhagavan and M. Balakrishnan. (2015). Geo-Spatial Approach for Urban
Green Space and Environmental Quality: A case study in Addis Ababa. M.Sc. Thesis. Addis
Ababa University.

Neway Kifle [2021] 71 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Rouse, J.W., Haas, R.H., Schell, J.A. and Deering, D.W. (1973). Monitoring Vegetation Systems in
the Great Plains with ERTS. 3rd ERTS Symposium, NASA SP-351, Washington DC, 10-14
December 1973, 309-317.

Roy, Dr & Byrne, Jason & Pickering, Catherine. (2012). A systematic quantitative review of urban
tree benefits, costs, and assessment methods across cities in different climatic zones. Urban
Forestry & Urban Greening. 11. 351-363. 10.1016/j.ufug.2012.06.006.

S. M. Labib & Angela Harris. (2018). The potentials of Sentinel-2 and Landsat-8 data in green
infrastructure extraction, using object-based image analysis (OBIA) method, European
Journal of Remote Sensing, 51:1, 231-240, DOI: 10.1080/22797254.2017.1419441

Samson Workaye & Suryabhagavan, Karuturi & Belliethathan, Satishkumar. (2018). Urban Green
Areas to Mitigate Urban Heat Island Effect: The Case of Addis Ababa, Ethiopia. International
Journal of Ecology and Environmental Sciences. 44. 353-367.

Selçuk, Reis. (2008). Analyzing Land Use/Land Cover Changes Using Remote Sensing and
GIS in Rize, North-East Turkey. Sensors. 8. 10.3390/s8106188.

Setturu, Bharath & Rajan, K. & Ramachandra, T V. (2014). Land Surface Temperature Responses to
Land Use Land Cover Dynamics. Geoinformatics & Geostatistics: An Overview. Bharath S,
Rajan KS, Ramachandra TV (2013) LandSurface Temperature Responses to Land Use Land
Cover Dynamics. Geoinfor Geostat An Overview 1:4. doi:10.4172/2327-4581.1000112.
10.4172/2327-.

Simwanda, M.; Ranagalage, M.; Estoque, R.C.; Murayama, Y. (2019). Spatial Analysis of Surface
Urban Heat Islands in Four Rapidly Growing African Cities. Remote Sens. 2019, 11, 1645.

Simwanda, Matamyo., & Murayama, Yuji. (2018). spatiotemporal patterns of urban land-use change
in the rapidly growing city of Lusaka, Zambia: Implications for sustainable urban
development. Sustainable Cities and Society https://doi.org/10.1016/j.scs.2018.01.039

Statistics Solutions. (2021). Pearson’s Correlation Coefficient. [Online] Available at:

https://www.statisticssolutions.com/pearsons-correlation-coefficient/ Retrieved on April 06,
2021.

Neway Kifle [2021] 72 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Su,Y. & Huang, G. & Chen, X. & Chen, S. & Li, Z. (2011). Research progress in the eco-
environmental effects of urban green spaces. Shengtai Xuebao/ Acta Ecologica Sinica. 31.
7287-7300.

Sun, Q., Wu, Z., & Tan, J. (2012). The relationship between land surface temperature and land
use/land cover in Guangzhou, China. Environmental Earth Sciences, 65(6), 1687-1694.

Swanwick, Carys & Dunnett, Nigel & Woolley, Helen. (2003). Nature, Role, and Value of Green
Space in Towns and Cities: An Overview. Built Environment. 29. 94-106.
10.2148/benv.29.2.94.54467.

Teklit Berhe. (2015). Remote Sensing and GIS-Based Urban Land Suitability Modeling at Parcel
Level Using Multicriteria Decision Analysis; A Case Study of Bole Sub City, Addis Ababa,
Ethiopia. Unpublished thesis work

Tyubee, B. T., & Anyadike, R. N. C. (2015, July). Investigating the effect of land use/land cover on
urban surface temperature in Makurdi, Nigeria. In Proceedings of the ICUC9—9th
International Conference on Urban Climate Jointly with 12th Symposium on the Urban
Environment, Toulouse, France (pp. 20-24).

Uça Avcı, Zehra & Karaman, Muhittin & Ozelkan, Emre & Papila, Ibrahim. (2011). A Comparison
of Pixel-Based and Object-Based Classification Methods, A Case Study: Istanbul, Turkey.

Ugur Avdan, Gordana Jovanovska. (2016). "Algorithm for Automated Mapping of Land Surface
Temperature Using LANDSAT 8 Satellite Data", Journal of Sensors, vol. 2016, Article
ID 1480307, 8 pages, 2016. https://doi.org/10.1155/2016/480307

UN- HABITAT. (2017). The State of Addis Ababa: The Addis Ababa we Want [Online] Available at:
https://unhabitat.org/the-state-of-addis-ababa-2017-the-addis-ababa-we-want Retrieved on
October 12, 2020

UN-HABITAT. (2010-2013). The State of African Cities 2010: Governance, Inequality, and Urban
Land Markets. Cities, 31, 425–429. Doi: 10.1016/j.cities.2012.07.007

United Nations (Department of Economic and Social Affairs - Population Division. (2014). World
Urbanization Prospects: The 2014 Revision Highlights

Neway Kifle [2021] 73 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

University of Leeds. (2015). A Brief Guide to the Benefits of Urban Green Spaces [Online] Available
at:https://www.euro.who.int/__data/assets/pdf_file/0010/342289/Urban-Green-
Spaces_EN_WHO_web3.pdf%3Fua=1 Retrieved on October 23, 2020

USGS. (2001). Percent-Slope Classes to Characterize the Unsaturated Zone in North Carolina.
[Online]Available at:
https://www2.usgs.gov/water/southatlantic/publications/ofr01495/#metadat.Retrieved on
February 16, 2021.

USGS. (2019a). Landsat 7 Data Users Handbook [Online] Available at: https://www.usgs.gov/core-
science-systems/nli/landsat/landsat-7-data-users-handbook Retrieved on January 10, 2021.

USGS. (2019b). Landsat 8 Data Users Handbook [Online] Available at: https://www.usgs.gov/core-
science-systems/nli/landsat/landsat-8-data-users-handbook. Retrieved on April 1, 2021.

USGS. (2021). What is remote sensing and what is it used for? [Online] Available at:
https://www.usgs.gov/faqs/what-remote-sensing-and-what-it-used?qt-
news_science_products=0#qt-news_science_products Retrieved on January 2021.

Vatseva, R., Kopecká, M., Otahel, J., Rosina, K., Kitev, A., & Genchev, S.E. (2016). Mapping Urban
Green Spaces Based on Remote Sensing Data: Case Studies in Bulgaria and Slovakia.

Vert, C.; Gascon, M.; Ranzani, O.; Márquez, S.; Triguero-Mas, M.; Carrasco-Turigas, G.; Arjona,
L.; Koch, S.; Llopis, M.; Donaire-Gonzalez, D.; et al. (2020). Physical and mental health
effects of repeated short walks in a blue space environment: A randomized crossover
study. Environ. Res.188, 109812.

Walta TV. (2020). Shegre Park Project. [Online] Available at:

https://www.youtube.com/watch?v=tkdovTyx_uM&t=1478s Retrieved on January 10, 2021.

Wan, Z., Dozier, J. A. (1996). A generalized split-window algorithm for retrieving land surface
temperature from space. IEEE Trans. Geosc. Rem. Sens.34 (4), 892–905,

Wang, Fei & Qin, Zhihao & Caiying, Song & Tu, Lili & Karnieli, Arnon & Zhao, Shuhe. (2015).
An Improved Mono-Window Algorithm for Land Surface Temperature Retrieval from Landsat
8 Thermal Infrared Sensor Data. Remote Sensing. 7. 4268-4289. 10.3390/rs70404268.
Neway Kifle [2021] 74 Remote Sensing and Geo-Informatics, AAU
 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Wang, L., Lu, Y., & Yao, Y. (2019). Comparison of Three Algorithms for the Retrieval of Land
Surface Temperature from Landsat 8 Images. Sensors (Basel, Switzerland), 19(22), 5049.
https://doi.org/10.3390/s19225049

Weng Q, Liu H, Lu D. (2007). Assessing the effects of land use and land cover patterns on thermal
conditions using landscape metrics in the city of Indianapolis, United States. Urban Ecosyst.
10. 203-219. 10.1007/s11252-007-0020-0.

Wichansky PS, Steyaert LT, Walko RL, Weaver CP. (2008). Evaluating the effects of historical land
cover change on summertime weather and climate in New Jersey: part I: land cover and surface
energy budget changes. J Geophys Res 113: D10107.

Witharana, Chandi & Civco, Daniel. (2014). Optimizing multi-resolution segmentation scale using
empirical methods: Exploring the sensitivity of the supervised discrepancy measure Euclidean
distance 2 (ED2). ISPRS Journal of Photogrammetry and Remote Sensing.
10.1016/j.isprsjprs.2013.11.006.

World Agroforestry. (2020). Ethiopia to grow 5 billion trees in the Second Green Legacy Campaign
[Online] Available at: https://www.worldagroforestry.org/blog/2020/06/09/ethiopia-grow-5-
billion-trees-second-green-legacy-campaign Retrieved on November 13, 2020

World Health Organization (WHO). (2017). Urban Green Spaces: A Brief for Action [Online]
Available at: https://www.euro.who.int/__data/assets/pdf_file/0010/342289/Urban-Green-
Spaces_EN_WHO_web3.pdf%3Fua=1 Retrieved on January 02, 2021

Wu, C., Li, J., Wang, C., Song, C., Chen, Y., Finka, M., & Rosa, D.L. (2019). Understanding the
relationship between urban blue infrastructure and land surface temperature. The Science of
the total environment, 694, 133742.

Wu, Q.; Li, H.Q.; Wang, R.S.; Paulussen, J.; He, H.; Wang, M.; Wang, B.H.; Wang, Z. (2006).
Monitoring and predicting land-use change in Beijing using remote sensing and
GIS. Landscape and Urban Planning, 78, 322–333.

Yosef Mengistu Darge & Binyam Tesfaw Hailu & Ameha Atnafu Muluneh & Tesfaye Kidane.
(2017). Detection of geothermal anomalies using Landsat 8 TIRS data in Tulu Moye
geothermal prospect, Main Ethiopian Rift.

Neway Kifle [2021] 75 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Young, Robert. (2010). Managing municipal green space for ecosystem services. Urban Forestry &
Urban Greening - URBAN FOR URBAN GREEN. 9. 313-321. 10.1016/j.ufug.2010.06.007.

Zha, Yong & Gao, Jingqing & Ni, S. (2003). Use of normalized difference built-up index in
automatically mapping urban areas from TM imagery. International Journal of Remote
Sensing - INT J REMOTE SENS. 24. 583-594. 10.1080/01431160304987.

Ziwei Deng, Xiang Zhu, Qingyun He, Lisha Tang. (2019). Land use/land cover classification using
time-series Landsat 8 images in a heavily urbanized area, Advances in Space Research, Volume
63, Issue 7,2019, Pages 2144-2154 https://,doi.org/10.1016/j.asr.2018.12.005.

Neway Kifle [2021] 76 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Appendices
Appendix 1: Regression analysis between NDVI and LST during 2006, 2016 and 2021

Neway Kifle [2021] 77 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Appendix 2: Regression analysis between NDBI and LST during 2006, 2016 and 2021

Neway Kifle [2021] 78 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Appendix 3: Regression analysis between MNDWI and LST during 2006, 2016 and 2021

Neway Kifle [2021] 79 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Appendix 4: Regression analysis between Altitude and LST during 2006, 2016 and 2021.

Neway Kifle [2021] 80 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Appendix 5: True color composite images of the study area A) SPOT 5 image during 2006, B) Sentinel-
2A during 2016 and Sentinel-2A during 2021.

Neway Kifle [2021] 81 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Appendix 6: A map showing conversion between LULC classes from 2006 to 2021

Neway Kifle [2021] 82 Remote Sensing and Geo-Informatics, AAU

 Mapping effects of urban blue-green landscapes on land surface temperature using geo-spatial
techniques: The case of Addis Ababa, Ethiopia

Appendix 7: Map of ground control points A) in 2006, B) in 2016 and C) in 2021

Neway Kifle [2021] 83 Remote Sensing and Geo-Informatics, AAU