URBAN GROWTH MODEL for Greater Abidjan TOWARDS COMPACT CITIES Deliverable 1.4 - Final comprehensive report April 2017 2 CAPSUS collaborators César Castillo José Díaz Daniela Evia Alejandra Fernández Judá García Ricardo García Tania Guerrero Karla Hernando Miguel Luis Ricardo Ochoa Carmen Valdez Guillermo Velasco World Bank team Alexandra LeCourtois National Office of Technical Studies and Development Vincent BADIE Joseph Pascal RAKOTOMALALA Jean KOUADIO Amani Didier KOMENAN Eloi Casimir BROU This work was made for the World Bank and funded by the Energy Sector Management Assistance Program Prado Sur 274 Lomas de Chapultepec Miguel Hidalgo Ciudad de México, 11000 soluciones@capitalsustentable.com www.capitalsustentable.com T: (52 55) 4744 48 32 ii Contents 1 Introduction 1 1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.1.1 Urban planning in Abidjan . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.1.2 Greater Abidjan – SDUGA 2030 . . . . . . . . . . . . . . . . . . . . . 4 1.2 Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 2 Methodology 7 2.1 Stage 1 - Identifying urban concerns and possible solutions . . . . . . . . 10 2.1.1 Urban concerns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.1.2 Possible solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.2 Stage 2 - Key indicators selection and data gathering . . . . . . . . . . . . 11 2.3 Stage 3 - Methods development and parameters definition . . . . . . . . 14 2.3.1 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 2.3.2 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 2.4 Stage 4 - Policy levers definition . . . . . . . . . . . . . . . . . . . . . . . . . . 15 2.4.1 Settlement of new population (urban growth) . . . . . . . . . . . 16 2.4.2 Land Use (density) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.4.3 Public transport expansion . . . . . . . . . . . . . . . . . . . . . . . . 18 2.4.4 Solid waste management improvements . . . . . . . . . . . . . . . 19 2.5 Stage 5 - Scenario development . . . . . . . . . . . . . . . . . . . . . . . . . . 20 2.5.1 Business As Usual (BAU) scenario . . . . . . . . . . . . . . . . . . . . 21 2.5.2 Planning scenario (SDUGA 2030) . . . . . . . . . . . . . . . . . . . . 22 2.5.3 Ideal scenario . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 2.6 Stage 6 - Knowledge exchange and results dissemination . . . . . . . . . 23 2.6.1 Technology transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 3 Results 27 3.1 Population settlement and density results . . . . . . . . . . . . . . . . . . . 28 3.2 Urban growth scenarios results . . . . . . . . . . . . . . . . . . . . . . . . . . 31 4 Conclusions 35 4.1 Main findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 4.2 Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 A Indicators methodology 39 CONTENTS A.0.1 Land consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 A.0.2 Population density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 A.0.3 GHG emissions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 A.0.4 Energy consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 A.0.5 Energy consumption for commuting . . . . . . . . . . . . . . . . . . 44 A.0.6 Energy consumption for water distribution . . . . . . . . . . . . . . 46 A.0.7 Energy consumption for public lighting . . . . . . . . . . . . . . . . 47 A.0.8 Energy consumption for solid waste collection . . . . . . . . . . . 48 A.0.9 Energy consumption for dwellings . . . . . . . . . . . . . . . . . . . 51 A.0.10 Infrastructure costs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 A.0.11 Infrastructure costs for urban expansion . . . . . . . . . . . . . . . 53 A.0.12 Infrastructure costs for upgrading existing capacity . . . . . . . . 54 A.0.13 Water consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 A.0.14 Job proximity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 A.0.15 Public transport proximity . . . . . . . . . . . . . . . . . . . . . . . . . 58 A.0.16 Services proximity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 A.0.17 Municipal service costs . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 B Data and assumptions 61 C Visualization maps of policy levers 65 C.1 Public transport levers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 C.2 Population settlement policy levers . . . . . . . . . . . . . . . . . . . . . . . 67 C.3 Land use policy levers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 C.4 Solid waste levers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 C.5 Amenities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 D Expansion model methodology 77 D.1 Expansion models methodology . . . . . . . . . . . . . . . . . . . . . . . . . . 77 D.1.1 Data sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 D.1.2 Definition of urban . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 D.2 Expansion model results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 Bibliography 80 iv 1. Introduction 1.1 Background Over the last decades, urban areas around the globe have experienced an accelerated growth. More than half of the world’s total population –4 billion people in 2016– now live in urban settlements [1]. As centers of both production and consumption, cities are playing an increasingly influential role in the global economy. However, rapid urban growth in developing countries exceeds their capacity to provide adequate services for urban dwellers. Over the next three decades, most of the world’s population growth is expected to happen in urban areas in the developing world [2]. This represents an important challenge in quality, capacity and access to urban services. As in other West African countries, in Côte d’Ivoire, cities are growing fast [3]. Between 2000 and 2010, Côte d’Ivoire’s urban population increased at an average annual rate of 3.6%. In 2017, cities accounted for 55.5% of Ivoirian total population and by 2050, an estimated 70% of Ivoirians will live in urban areas [1]. This represents an increase of more than 22 million inhabitants, which is almost twice the country’s current population (see Figure 1.1). Figure 1.1: Historic and estimates of urban and rural population in Côte d’Ivoire 50 40 Population in millions 30 Population Rural Urban 20 10 0 1975 2000 2025 2050 Year Source: World Bank Data, Population estimates Time Series for Côte d’Ivoire (http://data.worldbank.org). An increase in urbanization level is positively associated with health, as well as economic growth, and has contributed to rising incomes and poverty reduction in 1.1. BACKGROUND both urban and rural areas. Urbanization has been positive for development in Côte d’Ivoire, as it has been in other countries. For example, Côte d’Ivoire has made a steady progress in infrastructure investments, increasing access to sanitation and improving water sources. However, contrary to other countries of the region, the high migration rates and the lack of a clear housing policy caused a rise in the percentage of population living in slums since the 1990, and only recently it has started to stabilize (Figure 1.2). Figure 1.2: Slums, sanitation and water in Côte d’Ivoire and other countries A) Population living in slums B) Improved sanitation facilities C) Improved water source 90 80 80 Country % of population with access % of population with access 30 % of urban population 70 70 Benin Burkina Faso Cote d'Ivoire 60 20 60 Ghana Niger Nigeria 50 Togo 50 10 40 40 1990 1995 2000 2005 2010 2015 1990 1995 2000 2005 2010 2015 1990 1995 2000 2005 2010 2015 Source: Developed with data from WHO/UNICEF Joint Monitoring Programme (JMP) for Water Supply and Sanitation (http://www.wssinfo.org/) and from UN HABITAT, the United Nation’s Millennium Development Goals database (http://mdgs.un.org). Despite Côte d’Ivoire’s political crisis from the 1990s to 2010, there has been significant economic growth associated with urbanization. For every 1 percent increase in urbanization, Côte d’Ivoire achieved over 3 percent increase GDP per capita growth between 2000 and 2010 (see Figure 1.3). This is high in contrast to other countries of the region, like Ghana and Nigeria with 1.8 and 0.5 percent, respectively [1]. Figure 1.3: GDP per capita growth for every 1 percent of urbanization in Côte d’Ivoire and selected ECOWAS (Zone B) countries 4 q 2010's* GDP per capita growth / Urban population growth q 2010's* q 2000's q 2000's 2 q 2010's* q 2010's* 1990's 2010's* q 1960's q 1990's q q 1960's q 1970's 2000's q 2000's q q q 1990's q 2000's q 1970's q q 1960's q 1960's 1960's 2010's* q 2010's* q 0 q q q 1980's q q 1970's q 1960's 1960's 1980's 1970's q q 1990's 1970's q 2000's q 1980's 1970's q 1990's q q 2000's 1990's q 1980's q 1980's −2 q 1980's q 1970's q 1990's −4 q 1980's Benin Burkina Faso Cote d'Ivoire Ghana Niger Nigeria Togo Source: GDP per capita (constant 2010 USD). GDP per capita is gross domestic product divided by midyear population. World Bank national accounts data, and OECD National Accounts data files (http://data.worldbank.org). Despite this economic growth associated with urbanization, Côte d’Ivoire will need 2 CHAPTER 1. INTRODUCTION to overcome important challenges in terms of economic and social development as uneven wealth distribution and inequality have increased in the last decades. Thus, Côte d’Ivoire is not receiving the benefits of urbanization, in fact, the country is affected by its drawbacks. Cities are experiencing ’diseconomies of scale’, such as severe traffic congestion, air pollution, and environmental pollution risks, amongst others. In response to the above-mentioned challenges, the World Bank is providing support to the Government of Côte d’Ivoire through a technical assistance towards sustainable urbanization. Using Greater Abidjan as a case study, the aim of this project is to building local capacity to carry out coordinated, evidence-driven spatial planning, particularly with the use of scenario planning tools. This could be later replicated to other cities in Côte d’Ivoire or in other countries of the region. 1.1.1 Urban planning in Abidjan Abidjan is the economic capital of Côte d’Ivoire, it houses it administrative representatives and it has been the city with the fastest growth of the country. The city developed around the port of Abidjan, which became the engine of economic growth in the country. This attracted high national and international population migration that translated into rapidly expanding urban growth. The authorities have since then, attempted to create master plans to regulate the development of the city in an orderly manner, for example the Badani plan in 1952, the SETAP Plan in 1960 and Abidjan’s Agglomeration Plan in 1980 [4], [5]. These master plans focused on the definition of development axes, aligned with the cardinal points, zoning areas not suitable for development and determining the layout of major social infrastructure projects. These plans were quickly rendered obsolete by the extent of the development of the city, especially with the development of irregular settlements in the north of Abidjan. The urban area of Abidjan expanded beyond the communes of its core, to become the Autonomous District of Abidjan (DDA) in 2002, including thirteen communes. However, the communes lacked specific master plans, and their development depended on the overall development strategy of the city. The Ministry of Construction, Housing, Sanitation and Urban Development (MCLAU) was responsible of the creation of the Master Plan 2000-2015. However, the implementation of this plan was hindered by the socio-military crisis and lack of funding. A notable absence in the Master Plan 2000 were land use policies to implement the desired actions. There was also a need to clearly define the organizations and stakeholders responsible for implementation of the policies and the role of communes to ensure a timely and coordinated realization of the projects [6]. 3 1.1. BACKGROUND 1.1.2 Greater Abidjan – SDUGA 2030 At the end of the post-election crisis, between 2010-2011, the Japanese International Cooperation Agency (JICA), offered a donation to the Ivorian Government in order to finance the project for the development of the Urban Master Plan of Great Abidjan (SDUGA), with a target year set in 2030. The SDUGA provides an urban planning framework to guide MCLAU in fulfilling their statutory responsibility. In contrast to the Master Plan 2000, the SDUGA 2030 is an integrated urban and transport planning document, which includes two items considered essential to achieve sustainable urban development for Greater Abidjan over the master plan period. One is a comprehensive set of land use sector policies to unify the actions of the many stakeholders and thus ensure a fully integrated master plan. The other is an implementation framework that takes account of the statutory decentralization responsibilities of all stakeholders, enacted since the Master Plan 2000, to enable full coordination in plan making, funding, implementation of projects and development control [6]. Since the law on Decentralizing Local Governments 2003, Abidjan Autonomous District (DAA) has taken initiative on coordination among related organizations for urban planning and development. The coordination with local municipalities (communes) is limited and financing is a big issue, beyond the capacities of the communes. The SDUGA 2030 includes the following components: • Spatial Strategy 2030 is based on a scenario of compact city in combination with and satellite economic clusters. It sets the development framework based on a primary land use classification. • Implementation Strategy 2030 illustrates the existing and future urban expansion or urban renewal areas, as well as the current and future infrastructure projects. It is a conceptual roadmap to achieve the spatial strategy. • Land Use Framework 2030 provides more detail in terms of land uses. This is the main driver for the spatial strategy of compact growth, as it establishes desired densities and a land use distribution that aligns with transport strategies (Figure 1.4). • Master Plan of Extended Areas, includes the Detailed Urban Plans for two suburban growth areas, Bonoua and Attingué, including land use zoning guidelines. • Urban Transport Master Plan is a key component of the strategy. It includes: – Road development plan, to ensure there is infrastructure for the public transport strategy. – Traffic Control strategy, to ameliorate the current saturation of roads. 4 CHAPTER 1. INTRODUCTION – Public transport development plan, including massive transit systems, such as metro, BRT and waterbus. Figure 1.4: SDUGA Land Use Framework for 2030 SDUGA FUTURE LAND USE FRAMEWORK IN URBAN UNITS LEGEND PROTECTED LAND Institutional (Government Offices / Security, etc.) Forests and Environmental Utilities / Infrastructure / Roads Protection Area Plantation / Agriculture Public open space ! ! HT Line ! DEVELOPED LAND !! !! Tourism Centre !! !! ! Low Density Residential !! ! !! ! Medium Density Residential ! NATURAL FEATURES ! ! High Density Residential ! !! ! Water ! " Sensitive Land Industrial ! Port PUBLIC TRANSIT ! " Passenger Station ! Mixed Use (Office / retail / ! residential / hotels) ! Ferry Station ! # ! Education (Secondary Multimodal Interchange ! Schools / Universities) ! Passenger and Freight Rail Health (Hospitals) ! ! BOUNDARIES Boundaries of Urban Unit ! Cultural / Tourism / Sports ! SDUGA Planning Area ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !! ! ! ! ! ! ! ! ! ! ! ! # ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !! ! ! ! ! ! ! !! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! # ! ! !! ! ! ! ! ! ! ! ! ! ! ! ! " ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !! ! ! ! " ! ! ! !! ! ! ! ! ! !! ! ! ! ! ! ! ! ! ! ! ! ! # ! # # ! ! ! ! ! ! " ! ! ! ! " ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! # " ! ! ! ! ! ! ! ! ! ! ! # ! ! ! ! ! !! ! ! # ! !! ! " " ! ! ! ! ! " " ! ! ! ! # ! ! ! ! ! ! !! # ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! #! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! " ! ! ! ! ! ! ! ! " ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! " ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! # ! ! ! ! ! ! ! ! ! ! # ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! " ! ! ! ! ! " ! ! " ! " ! ! ! ! ! " ! ! ! ! ! " ! ! " ! ! ! ! ! ! ! ! ! ! " ! ! ! ! ! ! ! ! ! ! ! ! " ! ! ! !! ! " ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !! ! ! ! ! ! " ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! Source: JICA and MCLAU (2015). µ 1:270 000 MINISTRY OF CONSTRUCTION, SANITATION AND URBAN DEVELOPMENT JAPAN INTERNATIONAL COOPERATION AGENCY (JICA) 0 2 500 5 000 10 000 Meters It is clear that SDUGA was a major advance in urban development for Abidjan of the Map Projection: - Projected Coordinate System: WGS 1984 UTM Zone 30N ORIENTAL CONSULTANTS Co., Ltd. JAPAN DEVELOPMENT INSTITUTE (JDI) INTERNATIONAL DEVELOPMENT CENTER OF JAPAN - Projection: Transverse Mercator last 20 years, and it includes accurate and adequate strategies towards sustainable Source: - FIELD SURVEY MADE BY JICA STUDY TEAM, 2013 (IDCJ) development. However, its slow implementation caused that some of its projects and strategies became outdated of there are areas of opportunity for updating the strategy. Some of the barriers of SDUGA’s implementation have to do with coordination between different stakeholders, whose areas often have conflicting or contradictory objectives. Other barrier is the lack of financial means to carry the projects through, but it is difficult to access international financing without the lack of clear benefits associated with the projects. Urban growth scenarios are planning tools that can helps us make a rapid assessment of the cost-benefit implications of future projects, which could help us create consensus among stake holders and apply for international funding by having the numerical evidence to prove the associated benefits. Having all this in mind, the World Bank supports the Ivorian government, with funds from the Energy Sector Management Assistance Program (ESMAP), providing technical assistance to include urban growth scenarios in urban planning 5 1.2. OBJECTIVE for Abidjan. In the following sections, we present the objectives, methodology, work plan and a description of the expected outputs. 1.2 Objective The general objectives are to contribute to local urban development planning using data analysis tools and to model urban growth scenarios that help urban planning and the decision-making processes, which would lead Great Abidjan to become more sustainable, resilient and inclusive. There are two specific objectives for this project: • Develop urban growth scenarios that visualize impacts of different public policies on the environmental, social and economic dimensions for Abidjan. • Assess the benefits and drawbacks of different combinations of public policies, projects and conditions to reach consensus about the best development path for the city. 6 2. Methodology This study used the methodology for urban growth scenario modeling. This methodology was designed to provide key information for city managers and simplify their decision-making process. This is done by forecasting how stakeholders’ present decisions might impact the city’s future conditions. The following section provide key definitions for concepts use thorough this chapter, as well as insights on the methodology of: urban growth scenarios, expansion and population distribution models. Definitions Urban concerns are the most significant problems that a city is facing. These “challenges” can be derived from different sources of information, such as literature review, benchmark comparison and interviews with local experts. In the present study, urban concerns are strongly based on the perspective of local stakeholders. Possible solutions are projects, instruments or policies which local stakeholders envision as potential alternatives to deal with urban concerns. Possible solutions can be a detailed plan or a conceptual idea. Indicators are numeric values which describe the conditions and issues of a city. Indicators simplify the evaluation, monitoring and communication of the status of a city and are key for integrated urban planning. Indicators can be used to assess (or model) how a city is (or will be) dealing with a specific urban concern. Scenarios are “possible future conditions” which can be projected using statistical models and spatial data. Developing scenarios helps forecasting how a city will be like in the future. To do so, practitioners analyze historical data and identify the key factors which led the city to its present conditions. Base year is the present year, or the year with the latest available information. Defining a base year is highly relevant in scenario modeling because it represents a startup-point for the forecasting process. Horizon year is a selected year in the future in which scenarios take place. Defining a horizon year is key to avoid bias when comparing scenarios. The definition of a horizon year strongly depends on the availability of historical data. Normally, the range between the base year and the horizon year is equal or less than the range of the historical data. Urban growth scenarios are planning tools that facilitate the understanding of various possible outcomes related to specific urban policies. These policies might include transport or infrastructure investment plans, land use changes and housing policies, among others. Scenarios contribute to an efficient communication of urban initiatives. They rely on indicators, which provide a “common language” based on numerical data and represent a consistent, transparent and systematic approach to urban concerns. Scenarios can serve as a dialog platform between stakeholders. They can be used to assess synergies between initiatives, to develop integrated solutions, to understand interdependency of possible solutions or to create a multilevel and multi-sectorial consensus, among others. Policy levers are abstractions of the real world. Levers allow decision makers to test the application of a project, instrument or policy in a computer model. By activating (or deactivating) a lever, stakeholders can visualize the potential impact of implementing a “possible solution” in a simulation platform. Policy levers are commonly designed to model the effects of a decision that is made in the present on a range of indicators which describe the city in the future —i.e., in the horizon year. Overview The scenarios provide a platform for understanding how integrated solutions could work and help different actors understand their interdependency, creating a multilevel and multisector consensus. To perform growth scenario modeling, the first step is to analyze the city characteristics of the past that led to the present conditions, in order to forecast how the city will be like in the future if no measures are taken. The current year, or the year with the latest available information, is called base year, and the forecasted year is called horizon year. For this project, the base year is 2014 (based on the latest available census) and the horizon year is 2030 (based on the SDUGA). The performance of the 2030 forecasted scenario is measured in a variety of indicators, which are calculated from the characteristics of the city (input variables). For example, indicators as public transport proximity and job proximity depend on the population density, the public transport system, and the employment density across the city. Figure 2.1 depicts the scenario modeling conceptual framework. Urban growth scenario modeling for Abidjan was developed following six main stages. The urban concerns were discussed and analyzed in close coordination with local stakeholders. In order to do so, a series of meetings and workshops with government officials and specialists were scheduled. Figure 2.2 shows the distribution of these meetings along the seven stages defined for urban growth scenario modeling. The following sections describe each stage. 8 CHAPTER 2. METHODOLOGY Figure 2.1: Conceptual framework for the urban growth scenarios Indicators Input variables Land consumption Public transportation network Job proximity Job density Public transportation proximity Population density Health centers proximity Street network 2030 School proximity Solid waste management Public space proximity Wastewater treatment Exposure to hazards Urban development plans Water and energy consumption Infrastructure costs Green House Gas emissions Existing urban amenities Infrastructure costs Water and energy demand Financial balance of local govt. Figure 2.2: Project Stages STAGES OUTPUTS 1 Urban concerns and 1.0 First mission relevant policies ollow up ollow up report 1st visit meetings meetings 2 Indicators and data 1.5 Data sets, materials gathering and sources 3 Methods development 1.1 Draft Methodology Report 4 Model baseline and 1.2 Preliminary Growth policy levers Model report 1.3 Validation workshop 5 Scenario development report 2nd visit ollow up meetings 1.4 Final comprehensive 6 Capacity building report 3rd visit 9 2.1. STAGE 1 - IDENTIFYING URBAN CONCERNS AND POSSIBLE SOLUTIONS 2.1 Stage 1 - Identifying urban concerns and possible solutions 2.1.1 Urban concerns We identified a set of preliminary urban concerns for Abidjan after a thorough literature review, including academic literature and policy documents. Urban concerns are the most urgent problems that the city is currently facing. These included services or infrastructure deficiencies, environmental issues and social conflicts, among others. After identifying the main urban concerns, we focused on relevant policies or projects that could solve those concerns, thus, improving the quality of life in the city. Those concerns and possible solutions were corroborated with local stakeholders after the first visit to Abidjan, when held meetings with different ministries involved with urban planning. Those meetings included: Ministry of Construction and Urban Planning, Ministry of Transportation, National water and sewage offices and electricity companies. Each stakeholder shared the most urgent and relevant concerns, from their own perspective, and identified possible solutions to address those concerns. The status of those solutions ranged between: projects or policies recently implemented, planned or vague proposals. A final list of urban concerns was defined considering the following criteria: those who have a feasible solution, those that could be abstracted into one or more indicators, those that could be modified in relation to a policy lever, those with either direct or indirect data to assess them. For Abidjan, the main urban concerns where the uncontrolled pace of urban growth caused by irregular settlements, the environmental degradation compromising natural protected areas, and the poor distribution of infrastructure along the city. See figure 2.3 for the complete list of identified urban concerns, including the number of stakeholders who identified it. 2.1.2 Possible solutions In a similar way to the urban concerns, we evaluated the most feasible solutions, in terms of projects, policies or programs, for those concerns. Usually, the local stakeholders will have already identified possible solutions for their concerns. These may be existing projects, projects in process of implementation or ideas of a project for which preliminary or complementary studies are needed. After the first visit to Abidjan, it was made clear by different stakeholders that the Urban Master Plan of Greater Abidjan 20301 (SDUGA) was addressing most of those urban concerns. SDUGA includes a new land use strategy that promotes a compact city through the establishment of urban growth limits. It also has a strong transport strategy, with the implementation of metro, BRT and water bus networks. 1 Schema Directeur d’Urbanisme du Grand Abidjan 10 CHAPTER 2. METHODOLOGY Figure 2.3: Urban concerns identifies by stakeholders Urban concerns Count Uncontrolled growth of informal settlements obstruct implementation of new projects 6 Environmental problems associated with uncontrolled growth 5 Lack of sanitation and natural preservation strategies in SDUGA 4 Inefficient and limited infrastructure distribution 4 High transport demand is limited by the infrastructure and covered by informal sector 3 Urban expansion caused by population migration from the center to the peripheries 3 Lack of solid waste management strategy, currently only one open dumpsite 2 Lack of social infrastructure (mainly schools, hospitals and public spaces) 2 Urban planning unlinked form socio-economic and demographic information 1 Lack of affordable housing policy because of uncontrollable market 1 SDUGA is a strong planning framework and has the potential of solving many urban concerns of Abidjan. However, as many stakeholders pointed out, its fully adoption will be a long-term process, with different priority projects. In addition, in some cases, the SDUGA is already outdated or is missing key areas of coverage. This is why the strategies of SDUGA’s compose a great part of the policy levers that will define the scenarios, which are also complemented with new policy levers that could generate an idea of the impacts of and “improved” SDUGA. 2.2 Stage 2 - Key indicators selection and data gathering The second stage of the project focused on identifying a list relevant indicators that reflected the urban concerns identified in stage 1. Simultaneously, we collected, validated, organized and integrated data into a single platform. To ensure high data validity, the National Office of Technical and Development Studies of Côte d’Ivoire (BNETD) acted as our local counterpart and performed the task of data collection and pre-processing. In this way, both the list of urban concerns and the integration of available data into a single platform provided a realistic perspective for the definition of urban indicators. Urban indicators are numeric values which describe the conditions and issues of a city. Indicators simplify the evaluation, monitoring and communication of the status of a city and are key for integrated urban planning. The list of indicators is crucial to evaluate the sustainability of each urban growth scenario and, therefore, they had to be tailored to the specific urban concerns of Abidjan. A first list of indicators was drafted after the first visit, and after assessing the data availability, it was verified with stakeholders during the second set of workshops and meetings. Indicators like water consumption, energy consumption and GHG emissions 11 2.2. STAGE 2 - KEY INDICATORS SELECTION AND DATA GATHERING measure the environmental aspect of sustainability, and the social sphere is covered by indicators like proximity to schools, jobs, health facilities and other urban services. Some relevant indicators, like air quality, environmental degradation and affordable housing provision could not be included due to lack of data. However, we consider those indicators are crucial to address some of the urban concerns like the expansion of irregular settlements, environmental impacts on natural protected areas and severe traffic congestion affecting the quality of life. Figure 2.4 shows the final list of indicators, which are described in detail below. Figure 2.4: Indicators for Abidjan Land Population GHG Energy Infrastructure Municipal consumption density emissions consumption costs service costs [km2] [population/km2] [kgCO2eq/capita*yr] [kWh/capita*yr] [FCA] [FCA] Water Job Public transport School Public space Sport facility consumption proximity proximity proximity proximity proximity [m3/capita*yr] [%] [%] [%] [%] [%] Worship place University Health facility Public building Cultural space Exposure to proximity proximity proximity proximity proximity hazards [%] [%] [%] [%] [%] [%] Land consumption Amount of land predicted to change from natural habitats or agricultural uses into urban human settlements. This indicator reflects how much will the city expand at the horizon year, therefore, it has a direct impact on other indicators, such as population density, energy consumption, infrastructure costs and all the proximity indicators. Population density. Number of inhabitants per built-up area, expressed as inhabitants per square kilometer. Therefore, this indicator refers to net population density. GHG emissions. Average greenhouse gases emissions released per capita annually. These GHG emissions are related to energy consumed in: public lighting, municipal water supply, solid waste collection, residential use and commuting (public transportation and private vehicles). 12 CHAPTER 2. METHODOLOGY Energy consumption. Average annual energy consumed per capita in public lighting, municipal water supply, solid waste management, residential use and commuting. The energy consumption indicator is composed out of 5 sub-indicators, all expressed in kWh per capita per year: • Energy consumption in commuting: Average annual energy consumed per capita in commuting within the city, by public transportation or private vehicle. • Energy consumption in water distribution: Average annual energy (per capita) required to supply the total volume of water demanded by the city’s dwellings. A water loss factor due to municipal network leakages is included in this calculation. • Energy consumption in public lighting: Average annual energy consumed per capita for public lighting. • Energy consumption in solid waste collection: Average annual energy consumed per capita in the solid waste management system of the city, including collection, transportation and energy consumed in landfill and transfer stations. • Residential consumption: Average annual household electricity use per capita. Infrastructure costs. Total capital investment required to build roads, water and sewerage network, public lighting and electricity grids for each square kilometer that the city expands. It also includes the investment required to increase the carrying capacity of existing networks within the current built up area of the city. The infrastructure costs indicator is composed out of two sub-indicators: • Infrastructure costs for urban growth outside the existing built up area (expansion): Total costs of building roads and water, sewerage, public lighting and electricity networks of each km2 of estimated city expansion. • Infrastructure costs for upgrading existing capacity of the built up area (infill): Total costs of upgrading roads and water, sewerage, public lighting and electricity networks for each km2 of the existing city where the population is estimated to double. Municipal service costs. Average annual municipal expenditure needed to provide public lighting, potable water, solid waste collection services and road maintenance per resident. Water consumption. Average annual household water use per capita. Job proximity. Percentage of the population that lives within 1,000 meters from areas of the city with a high job density. 13 2.3. STAGE 3 - METHODS DEVELOPMENT AND PARAMETERS DEFINITION Public transport proximity. Percentage of the population that lives within walking distance from a public transport station. Walking distance is considered to be 800 meters for structured transport systems –like a BRT or subway, and 300 meters for buses. School proximity. Percentage of the population that lives within a radius of 700 meters from an elementary school. Public space proximity. Percentage of the population that lives within a radius of 700 meters from a public space or park. Worship place proximity. Percentage of the population that lives within a radius of 1,000 meters from a Mosque, Church, Synagogue or other worship place. Health facility proximity. Percentage of the population that lives within a radius of 1,500 meters from a hospital, clinic, or doctor. Public building proximity. Percentage of the population that lives within a radius of 2,000 meters from the city’s town hall or a public service office. Cultural space proximity. Percentage of the population that lives within a radius of 1,000 meters from a cultural facility, community center, library, social facility or theater. 2.3 Stage 3 - Methods development and parameters definition Stage three focused on developing accurate methods to calculate each indicator. In this stage, we also defined a series parameters which could be adjusted in the future, as better or updated data is available. 2.3.1 Methods The calculation methods were also defined in line with the policy levers identified and validated by the stakeholders (see section 2.4). For this stage, it was important to reflect all the possible drawbacks of policy levers, both positive and negative, and its impact on other indicators. For example, creating a transfer station within the solid waste collection system can reduce the volume of diesel consumed by the collection vehicles, but the new transfer station will also consume energy to operate. Therefore, is important to have both aspects in the calculation method of the energy consumption indicator. To facilitate methods replication, index cards for each indicator can be found in Appendix A. They contain a description of each indicator, the calculation methods and equations, units and the information sources used in the present study. Appendix B lists the data sources used to calculate the indicators for Abidjan. 14 CHAPTER 2. METHODOLOGY 2.3.2 Parameters In addition to the methods for calculating indicators, we established a set of parameters or criteria that could help us establish a baseline to evaluate each indicator. Some parameters are based on international standards, like recommended proximity to certain services, while others are city-specific, like per capita energy or water consumption. Appendix B list of all the parameters used for this project. 2.4 Stage 4 - Policy levers definition Possible solutions to the urban concerns identified in stage 1, like projects and public police suggested by the local actors, were analyzed and structured into policy levers that can change the urban development of the city. These policy levers were complemented with international experience, creating new levers or enriching the ones identified with local counterparts. The outcome of this stage was a set of policy levers that can impact the selected indicators. For example, the construction of a new BRT line will expand the public transportation network, enhancing the public transport proximity indicator and decreasing the greenhouse gases emissions indicator, as shown in Figure 2.5. Figure 2.5: Policy lever example and impact on two indicators A preliminary list of policy levers was created based on the identified possible solutions to the urban concerns during the first mission to Abidjan. As mentioned in section , SDUGA Master Plan offers some of the possible solutions for the urban concerns identified by the local stakeholders. This is why three of policy levers are based on variations on land use, public transportation and density, which are strategies included in the SDUGA. In addition, complementary policy levers were defined based on specific projects, policies or programs that provided a possible solution for the remaining urban concerns. This list of complementary policy levers was presented to the stakeholders during the second mission, who assigned different scores according to their relevance. The list below shows the main outcome of the validation exercise. • Improve wastewater management - 31 points 15 2.4. STAGE 4 - POLICY LEVERS DEFINITION • Improve solid waste management - 28 points • Include intra urban affordable housing projects - 16 points • Increase school infrastructure according to demographic needs - 12 points • Increase generation of distributed clean energy - 9 points • Increase green and public areas - 6 points • Use vacant intra-urban land - 2 points The validation of additional policy levers helped us identify relevant improvements to the SDUGA that would define our Ideal scenario. However, due to the lack of available data regarding some of these policy levers, not all of them were incorporated in the model. Until now four policy levers have been incorporated in the urban growth model; their calculation methods are explained below. 2.4.1 Settlement of new population (urban growth) Different urban growth scenarios can be created according to where the forecasted population for 2030 will be distributed geographically. If no specific measures are taken to tackle urban expansion, the population is more likely to occupy new expansion areas, which are predicted using the urban expansion models explained in Appendix D. However, urban growth can be distributed according to land uses and densities, specified in the SDUGA. In addition, if there is a particular urban contention policy being enforced by the government, the new population settlement could be concentrated within the specified boundaries or polygons. The settlement lever directs the location where future population will distribute using different priority polygons. The model allocates population to each priority polygon until it reaches a pre-defined saturation level. Next, it allocates the remaining population in to a second polygon and continues as such until it reaches the forecasted population for 2030. It has four options: • 0 where the future population occupies the predicted urban expansion polygon; • 1 where the future population settles in areas close to jobs and public transportation, first within the current boundaries of the city, and then in the urban expansion polygon; • 2 where the future population is allocated according to the expansion areas of the Master Plan. Once it saturates its cap, remaining population is settled in the residential land use areas of the Master Plan within the built up area. Finally, remaining population settles outside Master Plan Zoned areas; and • 3 where the future population is allocated according to residential land use areas of the Master Plan within the built up area. Second, remaining 16 CHAPTER 2. METHODOLOGY population allocates according to expansion areas of the Master Plan. Finally, remaining population settles outside Master Plan Zoned areas. Figure 2.6: Settlement policy levers 2.4.2 Land Use (density) The lever for settlement of new population works in combination with the Land Use lever, which specifies the way that population is distributed according to different densities. The quantity of new population that the model allocates in each priority polygon is based on the maximum population density (max_hu) allowed per residential land use. These land uses are established in the Master Plans. The model saturates this maximum allowed density according to different percentages, from 30%, 50%, 70% and 95%. For example, if in a particular location the authorized population density cap (max_hu) is 100 inhabitants per hectare, and we set the saturation level on 95%, then the model will only allocate 95 inhabitants in that block. If there is not a defined maximum allowed density, because there is no Master Plan or it is an un-zoned area, a maximum number is assumed (assumed_hu). The Land Use lever can also reflect changes to the land uses and building norms, i.e., a new or modified Master Plan. This is modeled by introducing a different set of values for max_hu. In the case of Abidjan, we modeled three different Land Uses with a saturation of 95 %: • 0 Maximum population density caps max_hu according to the current Master Plan 2010; • 1 maximum population density caps max_hu according to the SDUGA 2030; and • 2 maximum population density caps max_hu according to new residential land uses near transport corridors and employment hubs, where the assumed density (assumed_hu) equals the high density areas of the SDUGA. 17 2.4. STAGE 4 - POLICY LEVERS DEFINITION Figure 2.7: Land Use policy levers 2.4.3 Public transport expansion In Abidjan, mobility problems are some of the most challenging to address. Among the main problems are road congestion and traffic jams, an increase of car ownership ratio, insufficient road capacity, lack of reliable public transportation system and absence of BRT and metro systems. These problems cause an increase of travel time, travel cost, air pollution and increase accident risk rates. The aim of the Public Transportation lever is to increase the total population that lives within walking distance from a structured public transport station, aiming to reach at least 80% of the population in proximity to structured public transport. SDUGA transport strategy includes the implementation of two metro, five BRT and two water bus lines for Abidjan. Some of these lines are already in process of implementation, but the the full strategy will take over 15 years to complete. This is why local stakeholders requested results by type of project. By differentiating the impacts and benefits of one line against the other they could prioritize investment decisions. The specific public transportation policy levers used for Abidjan were: Figure 2.8: Public transportation policy levers 18 CHAPTER 2. METHODOLOGY 2.4.4 Solid waste management improvements In some cities of the country, the waste problem has escalated, caused mainly by urban expansion. Especially in cities like Abidjan, this issue is highly relevant. The city of Abidjan currently generates between 0.7 to 0.9 tons of annual solid waste per person. The lack of transfer stations and of appropriate solid waste disposal centers worsens the problem of limited waste collection coverage, as the collection trucks must travel larger distances to the dumpsites, increasing collection intervals, consuming more energy and affecting total management costs. The construction of transfer stations could be part of the solution to the solid waste management problems of Abidjan. According to the National Agency of Waste Management (ANAGED), the Autonomous District of Abidjan has a new plan to build seven new transfer stations: two in Yoponugon community, one in Bingerville, one in Adjamé, one in Anyama and two in Port Bouet. These stations will have a higher storage capacity which will allow them to receive than 3,400 tons of solid waste every day. Another relevant improvement will be the creation of new landfills, which cuts down environmental and public health risks posed by the current open dumpsites. There are three landfill projects planned for Abidjan: one in Kossihouen, one in Attiéko and, one in Ayewahi. Based on this information provided by the technical experts in the field, a policy level was created to test the impacts of improvement int he solid waste management system of Abidjan, as shown in figure 2.9. Figure 2.9: Solid waste management policy levers 19 2.5. STAGE 5 - SCENARIO DEVELOPMENT Table 2.10 summarizes all the policy levers for Abidjan. Figure 2.10: Policy levers for Greater Abidjan 2.5 Stage 5 - Scenario development After the definition of policy levers, a set of scenarios is created to visualize different urban growth trends. The first step of this stage was the creation of a base model which reflects the current conditions of Abidjan, for the base year (2014). Using this base, we created a preliminary set of results using de methods defined in stage 3 that would help us compare future improvements. This base model was the be forecasted for the 20 CHAPTER 2. METHODOLOGY horizon year (2030) and, in combination with the activation of deactivation of different policy levers, created other scenarios. Three alternative scenarios were analyzed for Greater Abidjan: 2.5.1 Business As Usual (BAU) scenario The business as usual (BAU) scenario reflected the base model growth trends forecasted to 2030. An expansion model methodology was developed to forecast the urban footprint growth to the horizon year. The resulting scenario is called Business As Usual (BAU Scenario) and it does consider any policy lever to improve its growth trend. Figure 2.11: Policy levers for BAU scenario Expansion models The expansion models work with machine learning algorithms and are based on trends of change in historical density and land use data. They analyze three different moments in time, therefore, based on the existing data, the forecasted year is set at 2030. This data is then complemented with input variables that reflect the physical characteristics of the urban environment to find a causality for change, also known as explanatory variables. For these models, the explanatory variables include: built-up area, population, elevations (slopes), gross domestic product distribution areas, highways, landmarks (schools, airports, universities, worship places and hospitals), and water bodies. In order to make results comparable among cities, global data sources with similar temporal information were used. This facilitates future replication of the methods, once the databases are updated. Three different expansion modeling methods were used for Abidjan: random forests, extratrees and logistic regression with regularization. Despite having the same input variables, results from each method vary due to the use of different statistical methods. Appendix D.1 provides a detailed description of the expansion models used for Abidjan. 21 2.5. STAGE 5 - SCENARIO DEVELOPMENT The method with the highest degree of confidence was selected to reflect the future footprint area of the BAU scenario. The resulting urban expansion predictions are shown in Section D.2. 2.5.2 Planning scenario (SDUGA 2030) The second scenario includes the strategies of the Greater Abidjan Urban Plan for 2030 (SDUGA), incorporating new land uses, new residential cap densities, and an important structured transport strategy. The SDUGA scenario includes the following configuration of policy levers: Figure 2.12: Policy levers for SDUGA scenario 2.5.3 Ideal scenario Finally, a third scenario –ideal– was based on the SDUGA scenario, but enhanced with complementary strategies identified by the stakeholders. This was the opportunity to prioritize among the different projects. In addition, this scenario aligns to sustainable development goals, making an efficient use of the available land –maximizing resources and infrastructure–, ensuring a balanced mixed of land uses, and incorporating adequate public transport solutions. The Ideal scenario has assigns the priority of allocating future population along transport lines, which are defined with a higher residential land use. By doing this, the scenario seeks to align to a Transport Oriented Development strategy, where the investment in infrastructure is capitalized by developing higher densities along. The Ideal scenario includes the following configuration of policy levers: 22 CHAPTER 2. METHODOLOGY Figure 2.13: Policy levers for IDEAL scenario (Transport Oriented Development) An alternative Ideal scenario was created to reflect minor improvements to the SDUGA. For example, it includes a different distribution for the future population settlement, which prioritize allocating population within infill areas instead of expansion areas. This scenario assumes the implementation of complementary instruments to the SDUGA, which will promote infill development. These instruments could be inner city re-densification programs, rise in property taxes of well-served sub-utilized plots, among others. Without these instruments there are no incentives for inner city development, so it is assumed that it would happen in the peripheries of cities, where land values tend to be lower. Figure 2.14: Policy levers for IDEAL scenario (SDUGA infill) 2.6 Stage 6 - Knowledge exchange and results dissemination Findings were communicated to decision-makers and other stakeholders during the workshops and follow-up meetings. In addition to the final report and process 23 2.6. STAGE 6 - KNOWLEDGE EXCHANGE AND RESULTS DISSEMINATION databases, which are the final deliverables of this project, a series of disseminations materials were produced, including a project brief summarizing the project (delivered in the first visit) and the scenario visualization online tool. The online visualization tool was created to display the scenario’s results and policy levers. Through the use of an interactive map, the user can turn on or off different layers with geo referenced information. For example, the location of particular amenities or transport network lines considered for each scenario. The tool can be accessed through http://up.technology/up_ci 2.6.1 Technology transfer The project involved a series of capacity building activities developed through the duration of the project. The purpose of these activities was to ensure an adequate technology transfer to the local counterpart and to create a sense of ownership of the methodology. Each of these capacity building activities involved a visit to Abidjan, which are explained below. First visit The first visit aimed at creating awareness of the importance of scenario modeling tools for decision making processes in urban planning. The capacity building activities will included: 1. Explaining the relevance of urban planning tools. 2. Describing the following concepts: indicators, scenarios and policy levers. 3. Identifying international examples of solutions to specific urban concerns. 4. Visualizing demo tools and understand the basic principles of the methodology. 5. Propose locally appropriate policy levers and scenarios. Second visit The second visit involved a workshop to validate the proposed scenarios and policy levers with local stakeholders. During this visit, we used a beta visualization tool to navigate through the scenarios and policy levers. Inputs from local and federal governments helped us calibrate the scenarios to suit better the local needs. Capacity building activities included: 1. Describing the basic concepts and calculations used in the tool: input variables, indicators and assumptions. 2. Explaining data requirements for local input variables and indicators. 3. Identifying combination of projects or policy levers to alleviate specific urban concerns. 24 CHAPTER 2. METHODOLOGY Third visit Finally, the third visit focus on disseminating the final outcomes of the project and training on how to use the visualization tool to analyze public policies and urban projects. The activities include: 1. Show the process of policy or selected project integration in the tool. 2. Explain how to create new calculation methods based on specific local needs. 3. Assess how assumptions affect the tool’s outputs. 4. Propose appropriate maintenance and dissemination programs for the tools. The workshops were focused on two groups of actors: decision makers and operational staff. The sessions for decision makers included dissemination material with the highlights of the project. The workshops for operational staff were custom-made for a more technical audience. These sessions included a description of methods, platforms, programming languages and any additional information relevant to understand the project. In addition to the above-mentioned activities, the present scenario methodology report, includes accurate policy recommendations and and implementation roadmap. 25 2.6. STAGE 6 - KNOWLEDGE EXCHANGE AND RESULTS DISSEMINATION 26 3. Results Urban scenarios are “possible future conditions” which can be projected using statistical models and spatial data. For this project, four scenarios were forecasted for the year 2030 using the combination of policy levers described in sections 2.4 and 2.5: a Business as usual (BAU) scenario, a SDUGA or planning scenario, and two versions of the Ideal Scenario. 3.1. POPULATION SETTLEMENT AND DENSITY RESULTS Figure 3.1 shows a graphic representation of the scenarios and policy levers for Greater Abidjan. Figure 3.1: Urban Growth Scenarios for Abidjan 3.1 Population settlement and density results The combination of population settlement and land use dictates how population is distributed on each scenario. They work together to define, where, how much and in which order shall the “future” population be distributed spatially. The base scenario reflects the current conditions of the city in the year 2014. In figure 3.2 we see the distribution of the population reaching different density levels. We observe that the denser areas of the city are in the commune of Koumassi, 28 CHAPTER 3. RESULTS reaching between 2,000 and 2,500 inhabitants per square km. The commune of Adjamé follows, with 1,500 to 2,000 inhabitants per square km. Medium density areas include Yopougon and Abobo, while the lowest density is observed in Plateau, Cocody and Port Bouët, with maximum of 500 inhabitants per square km. Figure 3.2: Population density scenario result - Base After the base condition was defined, the future scenarios create new configurations for the population settlement according to the specified policy levers. Figure 3.3 shows the BAU, SDUGA, IDEAL (infill) and IDEAL (TOD) scenarios. In BAU scenario, upper left from 3.3, we see that new population is distributed in the expansion areas defined in section 2.5.1. This is an hypothetical scenario, and is evidently not realistic – especially considering that densities in this area reach more than 6,000 inhabitants per square km. However, it is an important exercise to see the degree of urban expansion that could follow if now policies were applied. SDUGA scenario, in the upper right, creates new patches of high density areas appear along the city. Particularly, north of Abobo, in central areas of Yopougon and the area of Petit-Bassam, reaching between 3,000 and 3,500 inhabitants per square km. This responds to land use specifications of the SDUGA plan. However, in this case the model simulates that the first areas to populate would be those further form the city centers. It starts distributing population in the expansion areas, at low density 29 3.1. POPULATION SETTLEMENT AND DENSITY RESULTS rates, and continues gradually to more central locations, where higher densities are allowed. Figure 3.3: Population density scenarios results In contrast to SDUGA, IDEAL (infill) scenario alters the order in which population is distributed. This means that it respects the same land uses as SDUGA scenario, but it starts distributing population in the denser areas, which tend to be located in more central locations. In the lower left corner of figure 3.3 we see larger zones with high density spread along Abobo, Adjamé and Koumassi. Finally, IDEAL (TOD) scenario offers contrasting results (lower right corner). In this case the population is distributed according to the areas with proximity to employment and transport, disregarding the land uses of SDUGA. It is evident that the central areas of Abidjan, meaning Plateau, Adjamé, Treichville and Cocody are the best served areas, and therefore they receive population under this scenario. 30 CHAPTER 3. RESULTS 3.2 Urban growth scenarios results Based on the population distribution and land use maps presented before, a set of results from the selected scenarios is presented in figures 3.4, 3.5, 3.6 and 3.7. It shows 25 different indicators, which include urban growth and density, proximity to services, energy, GHG emissions and costs indicators. Land consumption and density indicators Figure 3.4: Main results by scenario and city IDEAL IDEAL BASE BAU SDUGA (infill) (TOD) land consumption - 25.25 49.62 9.5 - [km ] 2 land consumption 6% 12% 2% - - (incr.) [%] infill area - - 39.64 59.85 56.17 [km2] total footprint 360.56 385.81 410.18 370.06 360.56 [km2] population density 12,246.94 18,355.64 17,264.85 19,136.87 19,641.1 [pop/km2] The results show additional land consumption for BAU, SDUGA and IDEAL (infill) scenarios. Interestingly, SDUGA shows larger land consumption than BAU scenario, 49.5 against 25.2 square km from the latter. This means that the plan has authorized urban growth areas that go beyond the trend of expansion. In contrast, IDEAL (infill) shows an increment of 9.5 square km, just 2 percent more than the current urban footprint. As expected, because all the scenarios assume the the same future population for 2030, average population density mirrors land consumption results, as it directly depends on the expansion areas. In this case, TOD scenario presents the higher density, while SDUGA shows the lowest. 31 3.2. URBAN GROWTH SCENARIOS RESULTS Proximity indicators Figure 3.5: Main results by scenario and city IDEAL IDEAL BASE BAU SDUGA (infill) (TOD) proximity to + 65% 40% 58% 67% 71% health facilities [%] proximity to 11% 7% 8% 10% 16% cultural facilities [%] proximity to 35% 22% 30% 35% 49% public offices [%] proximity to 51% 32% 48% 56% 58% schools [%] proximity to 61% 38% 55% 64% 64% workship facilities [%] proximity to 9% 6% 42% 49% 59% public transport [%] proximity to 37% 23% 29% 35% 61% employment hubs [%] Proximity to social amenities (including schools, health, cultural, worship and public facilities) increases in all scenarios, as compared to BAU scenario. In all cases SDUGA scenario presents the smallest increment, and Ideal (TOD) performs slightly better than Ideal (Infill) scenario. The most dramatic increase is regarding transport proximity, which varies from a coverage of only 6 percent of the population in the BAU scenario, to a coverage between 42 and 59 percent in the SDUGA and Ideal(TOD) scenarios, respectively. A less dramatic but still representative, increase results from the employment proximity, which shifts from from 23 percent coverage, to 35 and 61 percent coverage in the Ideal (Infill) and Ideal (TOD) scenarios. It is important to note that Base scenario shows better performance than BAU, SDUGA and, in some cases IDEAL(Infill) scenarios. This is because of the definition of future scenarios and the rules to distribute future population, i.e., mainly in future expansion areas. Besides, by having different horizon years and, therefore, 32 CHAPTER 3. RESULTS total population, is not appropriate to compare Base scenario against any other. Energy and GHG emissions indicators Figure 3.6: Main results by scenario and city IDEAL IDEAL BASE BAU SDUGA (infill) (TOD) GHG emissions 2,496.2 2,927.97 2,168.54 2,084.46 2,022.18 [Kwh/pp/yr] total energy 7,756.1 9,473.59 6,494.98 6,172.56 5,929.97 consumption [Kwh/pp/yr] energy use 51.48 34.35 36.52 32.95 32.1 (public lighting) [Kwh/m3] energy use 343.69 325.86 328.16 324.37 323.48 (water) [Kwh/pp/yr] energy use 5,167.8 6,919.3 3,936.06 3,644.89 3,404.11 (transport) [Kwh/pp/yr] energy use 3,430.22 4,592.81 2,612.63 2,419.36 2,259.54 (gasoline) [Kwh/pp/yr] energy use 1,737.58 2,326.49 1,323.43 1,225.53 1,144.57 (diesel) [Kwh/pp/yr] energy use 44.04 42.75 42.91 19.01 18.95 (solid waste) [Kwh/pp/yr] energy use 2,149.08 2,151.33 2,151.33 2,151.33 2,151.33 (building) [Kwh/pp/yr] Total energy consumption reduces considerably in SDUGA and both Ideal scenarios, in comparison to BAU scenario. This is because energy consumption from transport (public and private) comprises 70 percent from the total energy consumption. Since SDUGA includes an important public transport strategy, total energy consumption reduces 12%, 14% and 16% in SDUGA, IDEAL (Infill) and IDEAL (TOD) scenarios, respectively. 33 3.2. URBAN GROWTH SCENARIOS RESULTS Thanks to the solid waste policy levers, which are active in both Ideal scenarios, energy consumption reduces in half, as compared to the BAU and SDUGA scenario. This means that transport collection is more efficient because of the additional transfer stations. Infrastructure investment and municipal service costs indicators Figure 3.7: Main results by scenario and city IDEAL IDEAL BASE BAU SDUGA (infill) (TOD) total infrastructure 338,713.58 684,021.02 155,214.74 26,068.24 costs - [Mill FCFA] new infrastructure 338,713.58 665,622.49 127,436.79 - - costs [Mill FCFA] infill rehab. costs - - 18,398.53 27,777.94 26,068.24 [Mill FCFA] municipal service 0.3295 0.3328 0.3278 0.3266 costs - [Mill FCFA/pp/yr] Infrastructure costs include new infrastructure in expansion areas, infill growth rehabilitation costs – to increase urban carrying capacity, or both, depending on the scenario. Results are therefore directly related to land consumption. This is why infrastructure cost from SDUGA scenario increase a twofold in comparison to BAU scenario, because their growth is allocated using twice as much expansion area than the BAU scenario. In contrast, Ideal(Infill) costs reduce by half those of BAU scenario and Ideal(TOD) represent less than 10%. Municipal service costs include the operation and maintenance of public lighting, potable water provision, solid waste collection and management services, as well as road maintenance for the city per resident. In general, these results show minimal variation because all the scenarios assume the same building energy and water consumption. Only road surface and solid waste collection services showed variations. Results show a slight increase in SDUGA scenario, as compared to BAU scenario, and a minimal decrease in Ideal scenarios. 34 4. Conclusions 4.1 Main findings The urban growth scenarios for Abidjan assessed the possible outcomes of implementing different projects, instruments or policies that aim to deal with the main urban concerns of the city for 2030. These possible solutions relied largely in the implementation of a new Master Plan (SDUGA) for 2030, which included an important public transport strategy, a reconfiguration and intensification of land uses. Through the use of different policy levers, variations in future conditions are reflected in different scenarios. Three main urban growth scenarios were analyzed: A Business As Usual scenario (BAU scenario) following historical growth, a Planning scenario according to the city’s Master Plan (SDUGA), and an Ideal scenario with two variations: one that couples SDUGA Master Plan with compact growth policies, and another one that focalizes growth into well-served polygons – with transport and job proximity. The overall aim of comparing these scenarios was to identify the most sustainable growth paths for the future of Abidjan. Results show that most indicators are closely related to urban expansion. Therefore, scenarios that have greater land consumption, like BAU and SDUGA scenarios, tend to perform worst than Ideal scenarios.The greatest impact seems to be related to amenity proximity indicators and infrastructure costs. Interestingly, even though SDUGA scenario has twice as much land consumption than BAU, it performed much better, by having a strong transport strategy. This is true, except for the infrastructure costs, which are higher in SDUGA scenario because of the large expansion areas. However, even when SDUGA’s land use strategy opens the possibility for higher densities in central locations, without complementary mechanisms that motivate infill development – like those shown in Ideal(infill) scenario, it is unlikely that SDUGA by itself will achieve its goals of creating a compact city. Finally, Ideal(TOD) scenario proves that is possible to house all the projected population increase within high density priority polygons, which are closer to public transit and employment areas. This was by far the scenario with the best performance. Its strategy of intensification along transport corridors that are 4.2. RECOMMENDATIONS currently planned for Abidjan, proved to increase substantially the proximity to amenities using only 7% of the cost from the BAU scenario. The generation of the above mentioned scenarios is the first step towards assessing tangible proposals that could support informed decision making. In this case, compact growth strategies yielded transversal benefits on the environmental, social and economic aspects. Annual GHG emissions could be reduced one third, proximity to employment doubled, and infrastructure costs reduced a tenfold. In the future, new or complementary projects and policies can be modeled in order to aid urban planning processes in Abidjan, that could bring further benefits for the Ivorian population. 4.2 Recommendations As part of this project, we propose a series of recommendations that could multiply the benefit of the latent projects of Abidjan’s urban transformation. One of the biggest limitations of the project development was data accessibility. In some cases data was available but not shared among different ministries or particular stakeholders. During the validation workshop, public officials showed interest in visualizing the current urban conditions of Abidjan, in terms of social infrastructure coverage, transport and employment hubs and having a common platform to discuss them with other sectors. The creation of a Geoportal would facilitate comprehensive urban planning in Abidjan. An open access web-based platform integrating urban spatial information into one single database can foster proper urban assessment and adequate planning. Its public access to Abidjan communes and urban planning experts in the region is essential. This Geoportal could gradually collect information that would allow the future development of new policy levers. For example, one regarding the provision of public space or expansion of social amenities, like schools or hospitals. Unfortunately, because of the lack of specific future policies regarding this issue, these policy levers could not be modeled. Second, as we have seen from the results, most of the future growth could be allocated within the current urban footprint of Abidjan. Many well-served areas have not reached their densification potential and, therefore, causing unnecessary urban expansion. For example, an inventory of vacant or “fraiche” land could be combined with a strategic infill policy. Similarly with the abandoned or unoccupied buildings, that could be refurbished and maximize its potential. This is not an easy task, particularly in contexts like Abidjan, where land is not acknowledged as a finite resource and private lots tend to be of larger size. However, once the amount of vacant land is accounted for, the impacts of other complementary instruments could be measured, for example a tax on undeveloped property when located in prime areas. Finally, even with adequate planning instruments like SDUGA, unless there is a strong housing strategy, the majority of the low-income population will continue to 36 CHAPTER 4. CONCLUSIONS settle in informal arrangements. Thus, exacerbating urban expansion. Therefore, we recommend the use of priority polygons to target specific types of development, in this case low-income. These polygons could have special regulations, in which higher density caps allow to accommodate more development. In combination to policies such as inclusionary housing, where a percentage of the development is reserved for lower-income housing, a high density policy could also alleviate housing alternatives for the most needed. In general, urban development policies and programs work better when combined with different approaches. Land use, transport, housing and infrastructure provision must interact in a coherent way and align their objectives towards a more sustainable future for Abidjan. This project is an attempt to bring those aspects closer together and generate dialogue among inter and intra governmental stakeholders. 37 4.2. RECOMMENDATIONS 38 A. Indicators methodology This Appendix contains a detailed description of the calculation methods and data sources used to build each of the indicators explained in section 2.3. Appendix B summarizes the values used to calculate the indicators in each city and policy lever. A.0.1 Land consumption Description Amount of land predicted to change from natural habitats or agricultural uses into urban human settlements. Measurement units Square kilometers [km2 ] Methodology Land consumption (land_consumption_km) is calculated as the difference between the city footprint in the horizon year (fp_horizon) and the footprint in the base year (fp_base). The city footprint refers to the total built-up area of a city, including streets, open space and inner vacant land. Urban footprint for the horizon year is estimated using artificial neural networks based on orography, roads, built-up area, population and employment historical data, using at least two points in time (e.g. years 2000 and 2015). The time gap between these two points in the past determine how far into the future can the forecast go; for example, 2030 would be the horizon year forecasted with 2000 and 2015 information. Calculation land_consumption_km = fp_horizon - fp_base Sources • Built-up area: Developed by CAPSUS as explained in Appendix D. A.0.2 Population density Description Number of inhabitants per built-up area, expressed as inhabitants per square kilometer. Measurement units Inhabitants per square kilometers [pop/km2 ] Methodology The population density (pop_dens) is calculated by dividing the total number of inhabitants (tot_pop) by the built-up area of the city (footprint_km2 ). Calculation pop_dens = tot_pop / footprint_km2 Sources • Population: Population and housing census 2014 and projections from Institut National de la Statistique (INS) [6, 7]. • Built-up area: Developed by CAPSUS as explained in Appendix D. Desirable range Recommended urban densities range between 80 to 110 housing units per hectare (15,000 to 35,000 inhabitants per km2 ) [8–10]. 40 APPENDIX A. INDICATORS METHODOLOGY A.0.3 GHG emissions Description Average per capita greenhouse gases emissions released annually related to the energy consumed for public lighting, municipal water supply, solid waste collection, electricity in dwellings and commuting (public transportation and private vehicles). Measurement units Kilograms of CO2 eq per person per year [kgCO2 eq/capita/annum] Methodology Annual per capita GHG emissions (ghg_tot) are the result of multiplying annual energy consumption per person by the carbon factor of each type of energy. Carbon factors refer to the amount of CO2 eq released by unit of energy consumed. For electricity consumption, the carbon factor (carbon_factor_elect) is specific to the national energy mix. Gasoline (carbon_factor_gasoline) and diesel (carbon_factor_diesel) carbon factors are used to estimate emissions from the consumption of fuels in transportation. The types of energy considered are: electricity for public lighting (energy_lighting), electricity for water supply (energy_water) and energy for transportation (energy_gasoline) and (energy_diesel). The carbon electric factor (carbon_factor_elect) is calculated by dividing the electric emissions (elec_emi) between the total energy generated in the country (gen_tot) plus the total energy imported in the country (imp_tot). The electric emissions (elec_emi) value is calculated by adding the emissions of the energy generated in the country (gen_emi) plus the emissions from the energy imported to the country (imp_emi). To calculate the emissions of the energy generated in the country (gen_emi) is necessary to make a calculus adding the generation of each type of energy multiplied by the emissions factor of each type of energy, The steam energy generation (ste_gen) is multiplied by the steam energy emissions (ste_emi), plus the diesel powered gas turbine generation (gdie_gen) multiplied by the diesel powered gas turbine emissions (gdie_emi), plus the natural gas powered gas turbine electric generation (gnat_gen) multiplied by its emissions (gnat_emi), plus the heavy fuel oil powered diesel engine electric generation (hfo_gen) multiplied by its emissions (hfo_emi), plus the natural gas powered diesel engine electric generation (dnat_gen) multiplied by its emissions (dnat_emi), plus the diesel engine electric generation (die_gen) multiplied by its emissions (die_emi), plus the electric generation of hydropower units (hyd_gen) multiplied by its emission (hyd_emi), plus the eolic electric generation (win_gen) multiplied by its emission (win_emi), plus biogas plants electric generation (bio_gen) multiplied by its emissions (bio_emi), plus combined cycle electric generation (com_gen) multiplied 41 by its emissions (com_emi), plus the existing generation of solar energy (sol_gen) plus the calculation of new solar power plants (capacity of the solar plant (sol_cap) multiplied by the efficiency of the plant (sol_eff) and by the hours of sun that the plant receives every year (sol_hours)) multiplied by 365 and by the emissions of solar electric generation (sol_emi). The emissions of the imported energy (imp_emi) are calculated by substracting the generation of the new solar power plants (capacity of the solar plant (sol_cap)) multiplied by the efficiency of the plant (sol_eff) and by the hours of sun that the plant receives every year (sol_hours) from the total energy imported in the country (imp_tot), all multiplied by the imported energy emissions factor (imp_fact). Carbon_factor_diesel and carbon_factor_gasoline are constant values assumed from the national average. Calculation ghg_tot=(energy_water+energy_lighting+energy_buildings)*emission_fact+ energy_gasoline* carbon_factor_gasoline+(energy_diesel+energy_swaste)*carbon_factor_diesel carbon_factor_elect=(elec_emi/(gen_tot+imp_tot))/1000000 elec_emi=gen_emi+imp_emi gen_emi=(ste_gen*ste_emi)+(gdie_gen*gdie_emi)+(gnat_gen*gnat_emi)+(hfo_gen *hfo_emi)+(dnat_gen*dnat_emi)+(die_gen*die_emi)+(hyd_gen*hyd_emi)+(win_gen* win_emi)+(bio_gen*bio_emi)+(com_gen*com_emi)+((sol_gen+((sol_cap/1000)* (sol_eff/100)* (sol_hours*365))*sol_emi) imp_emi=(imp_tot-((sol_cap/1000)*(sol_eff/100)* (sol_hours*365)))*imp_fact Sources • Electricity generation in Abidjan obtained from the Ivorian Electricity Company (CIE). • Emissions factors per type of generation obtained from the IPCC [11]. 42 APPENDIX A. INDICATORS METHODOLOGY A.0.4 Energy consumption Description Total average energy consumed per person during a year for public lighting, municipal water supply, solid waste management, electricity in dwellings and commuting by public transportation and private vehicles. The solid waste management energy consumption involves collection, transportation to transfer stations and final disposal sites, transference, and final disposal in landfill (if applicable). Measurement units Kilowatts hour per person per year [kWh/capita/annum] Methodology The indicator energy_consumption embraces the energy consumed by the city’s population for commuting (energy_transport), for the electricity they consume in their homes (energy_buildings), to supply the water they consume (energy_water), for public lighting (energy_lighting), and to manage the solid waste produced by the city (energy_swaste). Each of these consumptions is explained as one indicator in the next charts. Calculation energy_consumption=energy_water+energy_lighting+energy_swaste+ energy_buildings+energy_transport 43 A.0.5 Energy consumption for commuting Description Total average energy consumed per person during a year for commuting within the city, by public transportation or private vehicle. Measurement units Kilowatts hour per person per year [kWh/capita/annum] Methodology Energy consumption associated with transportation (energy_transport) is calculated by adding the energy consumed by type of fuel available in transportation vehicles in the city (energy_diesel and energy_gasoline) divided by the total population(tot_pop). The overall consumption in the city, by type of fuel, is the sum of the energy consumed in all the analysis points in the city. The energy associated to transportation by fuel type in each analysis point was calculated by multiplying the costs incurred in each type of fuel (transport_cost_diesel and gasoline) per person, by the population in the analysis point, by a factor that converts them to energy. This factor combines the diesel and gasoline calorific value (diesel_cv and gasoline_cv) and the diesel and gasoline density (gasoline_density and diesel_density). The transport cost associated to each type of fuel was calculated by multiplying the transport (transport_cost) by the fraction each type of transport represents of the whole array (gasoline and diesel_transp_frac). Transport costs were calculated per each analysis point by applying a linear multivariable regression model developed with expenditure and income data from Mexico. The model’s resulting units are costs incurred per household per trimester in mexican pesos. As the units desired are the local currency per person per year, the result is multiplied by 4 trimesters in a year, the exchange rate to the desired currency (FCFAMXN_exrate), average inflation of the mexican Peso from the date the model was developed until the current date (avge_inflation), and divided by average household size in the city (hu_size). Calculation energy_transport=(sum(energy_dieseli )+sum(energy_gasolinei ))/tot_pop energy_diesel=sum(energy_dieseli )/tot_pop energy_gasoline=sumenergy_gasolinei )/tot_pop energy_dieseli =popi *transport_cost_dieseli *(diesel_cv/diesel_cost* diesel_density/1000) energy_gasolinei =popi *transport_cost_gasolinei *(gasoline_cv/gasoline_cost* gasoline_density/1000) transport_cost_dieseli =transport_costi *diesel_transp_frac/100 44 APPENDIX A. INDICATORS METHODOLOGY transport_cost_gasolinei =transport_costi *gasoline_transp_frac/100 transport_costi =max(0,[4/FCFAMXN_exrate*(1+avge_inflation/100)*(- 620.06+3.06*transit_distance+(-19.10)*job_density_avge+(- 0.69)*pop_density_avge+213.09*avge_area+661.16*socioeco_level)]/hu_size) Sources • Percentages of diesel and gasoline vehicles borrowed from data in Jordan [12]. • Population: Population and housing census 2014 and projections from Institut National de la Statistique (INS) [7, 13]. • Diesel and gasoline calorific values and densities [14]. 45 A.0.6 Energy consumption for water distribution Description Per capita annual amount of energy required to supply the volume of water demanded by the city’s dwellings. Water losses due to the municipal network leakages are included in the calculation. Measurement units Kilowatts hour per person per year [kWh/capita/annum] Methodology The total annual energy consumption for water distribution (energy_water) is calculated by multiplying the energy needed to supply and distribute one cubic meter of water (water_factor) by the sum of the total volume of water consumed by the city (tot_water * tot_pop) and the water lost through leakages; which is estimated as the multiplication of the kilometers of roads in one square kilometer of the city (prim_road_km2 + sec_road_km2 + ter_road_km2 ), the square kilometers of the city (footprint_km2 ) and the volume of water lost by kilometer (loss). This total is then divided by the total population of the city (tot_pop). Calculation energy_water=water_factor*(tot_water*tot_pop+footprint_km2 *(prim_road_km2 + sec_road_km2 +ter_road_km2 )*loss)/tot_pop Sources • Total water consumed by the city provided by by the National Water Office (ONEP). • Loss of water per net length provided by the National Water Office (ONEP). • Energy consumption of the municipal water grid to supply 1 m3 of water [15]. • Population: Population and housing census 2014 and projections from Institut National de la Statistique (INS) [7, 13]. • Roads: Open Street Maps. • Built-up area: Developed by CAPSUS as explained in Appendix D. 46 APPENDIX A. INDICATORS METHODOLOGY A.0.7 Energy consumption for public lighting Description Annual average energy consumption for public lighting per person. Measurement units Kilowatts hour per person per year [kWh/capita/annum] Methodology The energy consumption for public lighting (energy_lighting) considers the total number of bulbs in the city (tot_bulb), how many of these are LED bulbs (num_led), the voltage of conventional bulbs (volt_bulb) and LED bulbs (volt_led), the dalily number of hours day that the bulbs are on (h) and 365 days to calculate the annual energy required to illuminate the streets. This number is divided by the total kilometers of primary, secondary and tertiary roads in the city (prim_road_km + sec_road_km + ter_road_km) to obtain the energy required per kilometer of street. As the precise number of kilometers of street that the city will have is uncertain, this number is estimated by multiplying the total built-up area of the city (footprint_km2 ) by the kilometers of primary, secondary and tertiary roads per square kilometer that the city has in the base year (prim_road_km2 + sec_road_km2 + ter_road_km2 ). At last, the energy required per kilometer of street is multiplied by the estimated kilometers of street and divided by the total population (tot_pop) to obtain the annual per capita energy consumption for public lighting. Calculation energy_lighting=((tot_bulb- num_led)*volt_bulb+num_led*volt_led)*hours_day*365/(prim_road_km+ sec_road_km+ter_road_km+ped_road_km)*(prim_road_km2 +sec_road_km2 + ter_road_km2 )*footprint_km2 /tot_pop Sources • LED and common bulbs voltage [16] • Population: Population and housing census 2014 and projections from Institut National de la Statistique (INS) [7, 13]. • Roads: Open Street Maps. • Built-up area: Developed by CAPSUS as explained in Appendix D. 47 A.0.8 Energy consumption for solid waste collection Description Average per capita energy consumed annually by the solid waste management system of the city, including collection, transportation and energy consumed in the landfill and transfer stations. Measurement units Kilowatts hour per person per year [kWh/capita/annum] Methodology The energy consumption associated with solid waste management (energy_swaste) embraces the energy consumed in every step the management system: solid waste collection (collection_energy), its transportation to the transfer stations and/or landfills (transport_energy), the energy consumed in the transfer station (TS_energy) and the energy used in the landfill (landf_energy). The result of this sum is divided by the total population (tot_pop). The first step is to calculate the energy used in the collection stage (collection_energy). This includes the efficiency of the collection truck (truck1_ef) in liters of diesel consumed per km, multiplied by the diesel density (diesel_den), the diesel calorific value (diesel_cv) and the total of kilometers traveled in a year, which is estimated by multiplying the kilometers of primary roads per km2 (prim_road_km2 ) by the percentage of the primary roads that the truck uses (prim_road_fact) plus the kilometers of secondary roads per km2 (sec_road_km2 ) multiplied by the percentage of the secondary roads that the truck uses (sec_road_fact) plus the kilometers of tertiary roads per km2 (ter_road_km2 ) multiplied by the percentage of the tertiary roads that the truck uses (ter_road_fact). All the last is multiplied by the number of times the truck collects garbage per week (collections) and multiplied by the weeks of the year. Then, this total is multiplied by the total built-up area (footprint_m3 ) to obtain the energy used by the truck per year. Additionally, is important to add the energy used by the truck’s compactor system, to calculate this part is necessary to multiply the compactor efficiency in m3 of diesel by m3 of garbage (comp_ef) by the diesel density (diesel_den) and the diesel calorific value (diesel_cv). Then multiply the result by the total population (tot_pop) multiplied by the waste generation per person per day (waste_per) multiplied by 365 days per year, all between the waste density (waste_density). The second stage is the energy used by the waste transport (transport_energy) from the center of the city to the transfer stations and from the transfer stations to the landfill, if there are not transfer stations it is assumed that the transport is 48 APPENDIX A. INDICATORS METHODOLOGY from the center of the city to the landfill area. The first part assumes that there are not transfer stations, it includes the efficiency of the collection truck (truck1_ef) converted to m3 per km multiplied by the diesel density (diesel_den) and the diesel calorific value (diesel_cv). Then is multiplied by the total waste volume collected annually (tot_wvol) divided by the capacity of the collection truck (truck1_cap), multiplied by the average of the distances from the center of the city to the landfill or landfills (dist_land). The second part of the calculation assumes one or more transfer stations exist, therefore it is the same than the first part but multiplied by the average of the distances from the center of the city to the transfer stations (dist_ts) and plus the efficiency of the transfer station truck (truck2_ef) converted to m3 per km multiplied by the diesel density (diesel_den) and the diesel calorific value (diesel_cv) and multiplied by the total population (tot_pop), the waste generation per person per day (waste_per), 365 days per year, and divided by the capacity of the transfer station truck (truck2_cap) multiplied by the average of the distances from transfer stations to the landfill (dist_tsland). The third stage is the energy consumed in the transfer stations (TS_energy) which includes the multiplication of the total population (tot_pop) by the waste generated per person per day (waste_per) multiplied by 365 days per year, multiplied by the energy consumed by the waste segregation machinery (energy_tonTS). The fourth and last stage is the calculation of the energy consumed in the landfill. This is obtained by multiplying the total population (tot_pop) by the waste generation per person per day (waste_per) by 365 days per year, divided by the efficiency of the landfill in ton per year (land_ef), all multiplied by the efficiency of the landfill trucks (truck3_ef). Calculation energy_swaste=(collection_energy+transport_energy+ TS_energy+landf_energy)/tot_pop collection_energy=truck1_ef/1000*diesel_den*diesel_cv*(prim_road_km2 * prim_road_fact/100+sec_road_km2 *sec_road_fact/100+ter_road_km2 * ter_road_fact/100)*collections*52*footprint_km2 +comp_ef/1000*diesel_den* diesel_cv*tot_pop*waste_per*365/1000/waste_density transport_energy=(((truck1_ef/1000)*diesel_den*diesel_cv)*(((tot_pop*waste_per *365)/1000))/truck1_cap)*dist_land)+(((truck1_ef/1000)*diesel_den*diesel_cv)* (((tot_pop*waste_per*365)/1000))/truck1_cap)*dist_ts)+(((truck2_ef/1000)* diesel_den*diesel_cv)*(((tot_pop*waste_per*365)/1000))/truck2_cap)*dist_tsland) TS_energy=((tot_pop*waste_per*365)/1000))*energy_tonTS landf_energy=(((tot_pop*waste_per*365)/1000))/land_ef)*truck3_ef 49 Sources • Solid waste generation and efficiencies and capacities of the trucks obtained from the National Agency for Waste Management (ANAGED). • Population: Population and housing census 2014 and projections from Institut National de la Statistique (INS) [7, 13]. • Roads: Open Street Maps. • Built-up area: Developed by CAPSUS as explained in Appendix D. 50 APPENDIX A. INDICATORS METHODOLOGY A.0.9 Energy consumption for dwellings Description Average annual housing electricity consumption per capita. Measurement units Kilowatts hour per person per year [kWh/capita/annum] Methodology The annually energy consumed per person in housing units (energy_buildings) reflects the energy savings expected from implementing a Green Building Code in the houses to be built between the base year and the horizon year (HU_new). It is estimated by multiplying the number of housing units existing in the base year (HU_existing) by the average energy consumption per household established as baseline (ener_baseline), plus the multiplication of the new houses (HU_new) by the penetration percentage of the green building code (GBC_pen /100) by the reduced energy consumption per household (GBC_ener), plus the multiplication of the new houses (HU_new) by the percentage that do not implement the green building code (1-GBC_pen/100) by the baseline housing unit energy consumption (ener_baseline). This total volume is then divided by the total population (tot_pop) to obtain the annual housing energy consumption per capita (energy_buildings). The number of new housing units (HU_new) is calculated as the difference between the total number of housing units in the horizon year (HU_tot_h) and the total number of housing units in the base year (HU_tot_b). Calculation energy_buildings=(HU_existing*ener_baseline+(HU_new*(1- GBC_pen/100)*ener_baseline+HU_new*GBC_pen/100*GBC_ener))/tot_pop HU_existing=HU_tot_b HU_new=HU_tot_h-HU_tot_b Sources • Energy of equipment: Energy of air conditioner, heater and lighting [17]. • Saving of equipment: Obtained from the USA Environmental Protection Agency (EPA). • Population: Population and housing census 2014 and projections from Institut National de la Statistique (INS) [7, 13]. • Roads: Open Street Maps. • Built-up area: Developed by CAPSUS as explained in Appendix D. 51 A.0.10 Infrastructure costs Description Total investment required to build the roads, water network, sewer network, public lighting and electricity grids for the square kilometers that the city will expand, and the investment required to increase the capacity of the existing networks were the urban population will have a two-fold increase. Measurement units Millon FCFA [Mill FCFA] Methodology The infrastructure costs indicator is the sum of the investment required to build the infrastructure of the city’s expansion (infrastructure_new_costs) and to upgrade the areas of the city that will increase their population (infrastructure_infill_costs). Calculation Infrastructure_costs=Infrastructure_new_costs + Infrastructure_infill_costs 52 APPENDIX A. INDICATORS METHODOLOGY A.0.11 Infrastructure costs for urban expansion Description Total costs to build the roads and water, sewage, public lighting and electricity networks of the km2 that the city is estimated to grow. Measurement units Millon FCFA [Mill FCFA] Methodology Combines the parametric costs to build the infrastructure of one new km2 of city by the total km2 of urban expansion. Infrastructure costs embrace construction costs per kilometer of primary road (cost_prim_road), secondary road (cost_sec_road), tertiary road (cost_ter_road), municipal water network (cost_water), sewage (cost_swge), electric grid (cost_elec) and public lighting (cost_light). Construction of roads considers walkways and road pavement. As the precise number of kilometers of street that the city will have is uncertain, this number is estimated by multiplying the total built-up area of the city (footprint_km2 ) by the kilometers of primary, secondary and tertiary roads per square kilometer that the city has in the base year (prim_road_km2 + sec_road_km2 + ter_road_km2 ). This estimate is multiplied by each of the parametric costs to obtain the total infrastructure costs of expanding the city. Calculation infrastructure_new_costs = ((cost_prim_road * prim_road_km2 ) + (cost_sec_road * sec_roads_km2 ) + (cost_ter_road * ter_roads_km2 ) + ((cost_light + cost_elec+cost_watr + cost_swge) * (prim_road_km2 + sec_roads_km2 + ter_roads_km2 ))) * land_consumption_km 53 A.0.12 Infrastructure costs for upgrading existing capacity Description Total costs to upgrade the water, sewage and electricity networks of the areas of the existing city that are estimated to have a two-fold increase in their population. Measurement units Millon FCFA [Mill FCFA] Methodology Considers costs associated with retrofitting the infrastructure needed to promote infill development. Includes parametric costs of water pipes retrofit (retro_watr), electricity lines retrofit (retro_elec), and sewer infrastructure retrofit (retro_swge) per km2 that needs to be retrofitted to cope with the demand of the new population density. The parametric costs are then multiplied by the number of kilometers of primary, secondary and tertiary roads per square kilometer of city (prim_road_km2 + sec_roads_km2 + ter_roads_km2 ), and by the amount of total land designated for infill development (infill_area_km2 ). If the infill area is not indicated by the user, it is calculated as the sum of the areas of all points of analysis that had population since the base scenario (pop0 >0) and their population had a two-fold increase or more, i.e. that the population in the scenario of analysis (popi ) is twice or more than the population in the base scenario (pop0 ), which was higher than 0. Calculation infrastructure_infill_costs=(retro_elec+retro_watr+retro_swge)*(prim_road_km2 +sec_roads_km2 +ter_roads_km2 )*infill_area_km2 infill_area_km2 =0.01*sum of area where pop0 >0 and popi /pop0 >=2 54 APPENDIX A. INDICATORS METHODOLOGY A.0.13 Water consumption Description Total average volume of water consumed per capita in the households of the city in one year. Measurement units Cubic meters per person per year [m3 /capita/annum] Methodology The total volume of water consumed annually per capita (tot_water) reflects the water savings expected from implementing a Green Building Code in part of the new houses to be built between the base year and the horizon year (HU_new). It is estimated by multiplying the number of housing units existing in the base year (HU_existing) by the average consumption of water per household established as baseline (HU_water0), plus the multiplication of the new houses (HU_new) by the penetration percentage of the green building code (GBC_pen /100) by the reduced water consumption per household (HU_water1), plus the multiplication of the new houses (HU_new) by the percentage that do not implement the green building code (1-GBC_pen/100) by the baseline housing unit water consumption (HU_water0). This total volume of water is then divided by the total population (tot_pop) to obtain the annual water consumption per capita. The number of new housing units (HU_new) is calculated as the difference between the total number of housing units in the horizon year (HU_tot_h) and the total number of housing units in the base year (HU_tot_b). Calculation tot_water=((HU_existing*HU_water0)+(HU_new*(1- (GBC_pen/100))*HU_water0+(HU_new*(GBC_pen/100)*HU_water1))/tot_pop HU_existing=HU_tot_b HU_new=HU_tot_h-HU_tot_b Sources • Average water savings per equipment (toilets, showerheads, sinks and washing machine) [18]. • Total Housing from: Population and housing census 2014 and projections from Institut National de la Statistique (INS) [7, 13]. 55 A.0.14 Job proximity Description Percentage of the population that lives within a distance of 1,000 meters from the areas of high job density of the city. Measurement units Percentage [%] Methodology This indicator identifies the areas of the city that concentrate employment and then quantifies the population that lives close to these areas as a percentage of the total city population. The process is divided in 4 stages: First, job density is quantified for each analysis point within a 250x250 meters grid. Job density (job_density) is measured as the number of jobs in a radius of 1,000 meters divided by the area of that circle, i.e. by 314.16 hectares. Second, a buffer of the maximum distance recommended (max_dist_job) is created from the center of each analysis point with a job density equal or higher than the minimum job density recommended (min_job_density). Third, the population (pop) of all the analysis points contained in the buffer is added up. This is the population that lives close to high job density areas (pop_prox_job). Fourth, this population is divided by the total population of the city (tot_pop) to obtain the percentage of the population that lives close to employment (job_prox). Calculation job_prox=pop_prox_job/tot_pop pop_prox_job=sumpopif(distance<=max_dist_job) from (job_density>min_job_density) max_dist_job=800m min_job_density=10 jobs per hectare If jobs_lever = 0 job_prox = ( subset(pop_prox_job, footprint_base) + pop_expansion * d_job_prox) / tot_pop Sources • Number of jobs: Obtained of an estimate of gross domestic product (GDP) derived from satellite data made by the National Oceanic and Atmospheric Administration (NOAA) [19] [20]. 56 APPENDIX A. INDICATORS METHODOLOGY • Population: Population and housing census 2014 and projections from Institut National de la Statistique (INS) [7, 13]. 57 A.0.15 Public transport proximity Description Percentage of the population that lives within walking distance from a public transport station. Walking distance is considered to be 800 m for structured transport systems like a BRT or subway, and 300 m for buses and similar. Measurement units Percentage [%] Methodology Public transport proximity (transit_prox) is calculated by dividing the population (pop_prox_transit) that lives within the maximum distance recommended to a public transport station (max_dist_transit), by the total population (tot_pop). First, a buffer of the maximum distance recommended (max_dist_transit) is created from the center of each public transport station. Second, the population (pop) of all the neighborhoods or blocks contained in the buffer is added up to obtain the population that lives close to public transport (pop_prox_transit). Third, this population is divided by the total population of the city (tot_pop) to obtain the percentage of the population that lives close to public transport (transit_prox). Calculation transit_prox=pop_prox_transit/tot_pop pop_prox_transit=sumpopif(distance<=max_dist_transit) If transit_lever=0 transit_prox=(subset(pop_prox_transit,footprint_base)+pop_expansion* d_transit_prox)/tot_pop Sources • Public transport stations: Obtain from SOTRA and the Ministry of Transport. • Population: Population and housing census 2014 and projections from Institut National de la Statistique (INS) [7, 13]. Desirable range 80% to 100% 58 APPENDIX A. INDICATORS METHODOLOGY A.0.16 Services proximity Description Percentage of the population with access to each of the following urban public services and amenities: schools (elementary, secondary or high school), universities, health facilities (clinics and hospitals), nurseries, public buildings, cultural spaces (community centers, libraries, theaters), worship places, markets, sport facilities, and public spaces. One indicator is calculated per each urban service or amenity. Measurement units Percentage [%] Methodology One indicator is calculated for each of the following classes of amenity: school, university, health facility, nursery, public building, cultural facility, place of worship, market, sports and public space. The proximity is calculated for each amenity class by dividing the population (pop_prox_ami ) that lives within the maximum distance recommended for that type of amenity (max_disti ), by the city’s total population (tot_pop). Table A.1 shows the maximum distance considered for each indicator. Table A.1: Classes and maximum distance recommended Class Landmarks Maximum distance School Elementary school, secondary school and high school 700m Health Clinic, doctors, hospital 1500m Public building Public building, town hall 2000m Cultural facility Community center, library, social facility, theater 1000m Place of worship Mosque, Church 1000m Public space Park, garden, public space 700m Calculation amen_proxi =(pop_prox_ami )/tot_pop pop_prox_ami =sumpopif(distance<=max_disti) If amenity_lever=0 amen_prox=(subset(pop_prox_am,footprint_base)+pop_expansion* d_amen_prox)/tot_pop 59 A.0.17 Municipal service costs Description Average per capita annual municipal expenditure needed to provide public lighting, potable water, and solid waste collection services and maintenance to the city roads. Measurement units Millon FCFA per capita [Mill FCFA/person] Methodology The annual municipal expenditure in public services per person (municipal_service_costs) is estimated from the energy needed to provide public lighting (energy_lighting) in all the streets of the city multiplied by the cost the municipality pays per each kWh of electricity consumed for public lighting (elighting_cost), plus the energy needed to provide potable water (energy_water) multiplied by the cost the municipality pays per each kWh of electricity consumed for provide potable water (ewater_cost), plus the diesel consumed by the trucks used for solid waste collection and management (energy_swaste) multiplied by the cost of one liter of diesel (diesel_cost), plus the average amount of money that the municipality spends to maintain one kilometer of road (road_maintenance) multiplied by the road kilometers per square kilometer of the city (prim_road_km2 +sec_roads_km2 +ter_roads_km2 ), by the built-up area of the city (footprint_km2 ) and divided by the total population (tot_pop) to obtain the municipal expense per capita. Calculation municipal_service_costs = energy_lighting*elighting_cost + energy_water*ewater_cost + ( diesel_cost*1000 * (collection_energy+transport_energy) / (diesel_den*diesel_cv) + road_maintenance * (prim_road_km2 + sec_roads_km2 + ter_roads_km2 ) * footprint_km2 ) / tot_pop Sources • Population: Population and housing census 2014 and projections from Institut National de la Statistique (INS) [7, 13]. • Roads: Open Street Maps. • Built-up area: Developed by CAPSUS as explained in Appendix D. • Diesel calorific value and density [14]. 60 B. Data and assumptions Table B.1: Data used for Abidjan Description Value Units Category ID Lever Source Number of houses per point that will be added 187.5 hu beyond_udp beyond_udp 0 in the expansion of the city in unzoned areas Parking area per parking spot 32.5 m2 building_norms parking_sqm 0 Building construction costs 0.24046055 Mill FCFA/m2 building_norms constr_cost_sqm 0 Underground parking construction costs 0.360690971 Mill FCFA/m2 building_norms parking_cost_sqm 0 Primary road (6 lanes) construction 1800 Mill FCFA/km costs cost_prim_road 0 [21] Secondary (4 lanes) construction 500 Mill FCFA/km costs cost_sec_road 0 [21] Tertiary road (2 lanes) construction 400 Mill FCFA/km costs cost_ter_road 0 [21] Electricity network construction 14 Mill FCFA/km costs cost_elec 0 Public lighting construction 20 Mill FCFA/km costs cost_light 0 [22] Water network construction 128 Mill FCFA/km costs cost_watr 0 Sewer network construction 11.27159103 Mill FCFA/km costs cost_swge 0 Electric network improvement 21.058 Mill FCFA/km costs retro_elec 0 [22] Water network improvement 1.3134195 Mill FCFA/km costs retro_watr 0 [23] Sewage network improvement 0.65670975 Mill FCFA/km costs retro_swge 0 Average road maintenance costs for the 0.65962846 Mill FCFA/km costs road_maintenance 0 municipality Increment cost factor for consolidation 1.5 factor costs incr_consolidation 0 Cost that the municipality pays per kWh of 0.00006 Mill FCFA/km costs ewater_cost 0 [24] electricity consumed to provide potable water Cost that the municipality pays per kWh of 0.00008 Mill FCFA/km costs elighting_cost 0 [24] electricity consumed for public lighting Commercial cost per liter of diesel 575 FCFA/lt costs diesel_cost 0 [25] Median of municipal expenses 2311044.341 Mill FCFA/km costs median_expense 0 [26] Median of municipal revenues 3053161.663 Mill FCFA/km costs median_revenue 0 [27] Primary Road Width 21 m costs prim_road_wid 0 [28] Secondary Road Width 15 m costs sec_road_wid 0 [28] Tax collection rate percentage 1 % costs tax_collection 0 Tertiary Road Width 10 m costs ter_road_wid 0 [28] Additional annual energy fees per consolidated 15.02878804 Mill FCFA/km costs energy_fee 0 km2 Additional annual water fees per consolidated 15.68958398 Mill FCFA/km costs water_fee 0 km2 Average street area % of km2 of development 36 % general %_street_area 0 [28] Average inflation from (dic 2010/2012) to Agu 27.84 % general avge_inflation 0 [29] 2017 in Mexico Housing unit size: number of habitants per 3.7 Inhabitants/hu general hu_size 0 [30] housing unit Avge. Gasoline market price (90 and 95 octane, 584 FCFA/L general gasoline_cost 0 [31] 50/50 split) Diesel density 837.52 kg/m3 general diesel_den 0 [14] Gasoline density 676.13 kg/m3 general gasoline_den 0 [14] Diesel net calorific value (NET CV) 11.93 kWh/kg general diesel_cv 0 [14] Gasoline net calorific value (NET CV) 12.62 kWh/kg general gasoline_cv 0 [14] Exchange rate for 1 Franco CFA (FCFA) to 0.034 MXN/FCFA general FCFAMXN_exrate 0 [32] Mexican Peso (MXN) Area of each analysis cell is different because 0 general census_data 0 it corresponds to the urban blocks provided official local sources (1=yes, 0=no) Primary roads total lenght in the city 454.88 km general prim_road_km 0 [28] Secondary roads total lenght in the city 340.491 km general sec_road_km 0 [28] Tertiary roads total lenght in the city 6471.251 km general ter_road_km 0 [28] Pedestrian roads total lenght in the city 190.307 km general ped_road_km 0 [28] Biogas units emissions factor 0 kgCO2/GWh ghg_emissions bio_emi 0 [11] Biogas units electricity generation per year 4.1 gCO2/kWh ghg_emissions bio_gen 0 [7] Combined cycle units emissions factor 371600 kgCO2/GWh ghg_emissions com_emi 0 [11] Combined cycle units electricity generation per 15108.2 GWh ghg_emissions com_gen 0 [24] year Diesel engines units emissions factor 0 kgCO2/GWh ghg_emissions die_emi 0 [11] Diesel engines units electricity generation per 28.6 GWh ghg_emissions die_gen 0 [24] year Continues in the next page Description Value Units Category ID Lever Source Natural gas diesel engines emissions factor 469000 kgCO2/GWh ghg_emissions dnat_emi 0 [11] Natural gas diesel engines electricity generation 648.7 GWh ghg_emissions dnat_gen 0 [24] per year Diesel gas turbines emissions factor 0 kgCO2/GWh ghg_emissions gdie_emi 0 [11] Diesel gas turbines electricity generation per 1 GWh ghg_emissions gdie_gen 0 [24] year Quantity of energy that the country consumes 19171.5 GWh ghg_emissions gen_tot 0 [24] generated in the country Natural gas turbines emissions factor 469000 kgCO2/GWh ghg_emissions gnat_emi 0 [11] Natural gas turbines electricity generation per 329.5 GWh ghg_emissions gnat_gen 0 [24] year HFO diesel engines emissions factor 0 kgCO2/GWh ghg_emissions hfo_emi 0 [11] HFO diesel engines electricity generation per 94.5 GWh ghg_emissions hfo_gen 0 [24] year Hydro units emissions factor 19000 kgCO2/GWh ghg_emissions hyd_emi 0 [11] Hydro units electricity generation per year 41.6 GWh ghg_emissions hyd_gen 0 [24] Emissions factor of imported energy 96000 kgCO2/GWh ghg_emissions imp_fact 0 [33] Quantity of imported energy that the country 435 GWh ghg_emissions imp_tot 0 [24] consumes Solar energy plants emissions factor 29000 kgCO2/GWh ghg_emissions sol_emi 0 [11] Solar energy units electricity generation per 491 GWh ghg_emissions sol_gen 0 [24] year Steam units emissions factor 769 kgCO2/GWh ghg_emissions ste_emi 0 [11] Steam units electricity generation per year 2033.6 GWh ghg_emissions ste_gen 0 [24] Wind units emissions factor 15000 kgCO2/GWh ghg_emissions win_emi 0 [11] Wind units electricity generation per year 390.7 GWh ghg_emissions win_gen 0 [24] Capacity of the solar plant 0 MW ghg_emissions sol_cap 0 Efficiency of the solar plant 86.12 % ghg_emissions sol_eff 0 [34] Number of sun hours that the country receives 8.5 h/year ghg_emissions sol_hours 0 [17] in average Emission factor of the national electricity grid 0.458 kgCO2/kwh ghg_emissions emission_fact 0 Greenhouse gases emissions per kWh of diesel 0.26751 kgCO2eq/kWh ghg_emissions carbon_factor_diesel 0 [14] Greenhouse gases emissions per kWh of 0.24906 kgCO2eq/kWh ghg_emissions carbon_factor_gasoline 0 [14] gasoline Penetration percentage of Green Building Code 0 % green_b_code GBC_pen 0 measures in housing units Average energy consumption per household 7973.76 kWh/yr per hu green_b_code ener_baseline 0 [35] established as baseline Reduced energy consumption per household by 10466.77 kWh/yr per hu green_b_code GBC_ener 0 [17] implementing green building code measures Other sources of water consumption 0 m3/yr green_b_code others_water 0 [23] Baseline water demand per housing unit 518.35 m3/yr per hu green_b_code HU_water0 0 [18] Efficient water demand per housing unit 235.7 m3/yr per hu green_b_code HU_water1 0 [36] Infill percentage 0 % infill infill_pct 0 Infill percentage 30 % infill infill_pct 1 Infill percentage 50 % infill infill_pct 2 Infill percentage 70 % infill infill_pct 3 Infill percentage 95 % infill infill_pct 4 Infill percentage 150 % infill infill_pct 5 Restricts infill to only vacant land 0 only_vacant only_vacant 0 Population increase between base year and 2666000 inhabitants population pop_increase 0 [37] horizon year Base year 2015 year population base_year 0 Horizon year 2030 year population horizon_year 0 Indicates that new population will settle 2 prioritize_tod prioritize_tod 2 according to the Master Plan or zoning Indicates that new population will settle in new 0 prioritize_tod prioritize_tod 0 urbanized areas, expanding the city Indicates that new population will settle near 1 prioritize_tod prioritize_tod 1 employment and public transportation Indicates that new population will settle 3 prioritize_tod prioritize_tod 3 according to the Master Plan, infill first (high to low density) Indicates that new population will settle 4 prioritize_tod prioritize_tod 4 according to the Master Plan, infill first (low to high density) Led penetration 0 % public_lighting led_pen 0 [38] Number of hours a day that the street lighting 12 h public_lighting hours_day 0 [22] works Voltage of the common bulbs used for street 0.25 kW public_lighting volt_bulb 0 [22] lighting Voltage of the led bulbs used for street lighting 0.139 kW public_lighting volt_led 0 [16] Distance between public lighting posts 35 m public_lighting interpost 0 [22] Percent of diesel vehicles in the country 29.87 % transport_energy diesel_transp_frac 0 [12] Percent of gasoline vehicles in the country 70.13 % transport_energy gasoline_transp_frac 0 [12] Minimum vacancy rate considered as natural 0 % vacant_hu min_vhu_rate 0 Activates the reduction of the vacancy rate to a 0 % vacant_hu vhu_reduction 0 natural minimum Percentage of primary roads used by the 90 % waste prim_road_fact 0 [39] collection truck Percentage of primary roads used by the 90 % waste prim_road_fact 1 [39] collection truck Percentage of secondary roads used by the 90 % waste sec_road_fact 0 [39] collection truck Continues in the next page 62 APPENDIX B. DATA AND ASSUMPTIONS Description Value Units Category ID Lever Source Percentage of secondary roads used by the 90 % waste sec_road_fact 1 [39] collection truck Percentage of tertiary roads used by the 90 % waste ter_road_fact 0 [39] collection truck Percentage of tertiary roads used by the 90 % waste ter_road_fact 1 [39] collection truck Total solid waste generated per person per day 0.9 kg/day waste waste_per 0 [40] Total solid waste generated per person per day 0.9 kg/day waste waste_per 1 [40] Waste Density (compacted volume) 350 kg/m3 waste waste_density 0 [41] Waste Density (compacted volume) 350 kg/m3 waste waste_density 1 [41] Average distance from the city center to the 45 km waste dist_land 0 [39] landfill or dumpsite Average distance from the city center to the 0 km waste dist_land 1 [39] landfill or dumpsite Average distance from the city center to the 30 km waste dist_ts 0 [39] transfer station Average distance from the city center to the 15 km waste dist_ts 1 [39] transfer station Average distance from the transfer station to 15 km waste dist_tsland 0 [39] the landfill or dumpsite Average distance from the transfer station to 15 km waste dist_tsland 1 [39] the landfill or dumpsite Efficiency of the landfill truck (compactor) 2264 kWh/day waste truck3_ef 0 [39] Efficiency of the landfill truck (compactor) 2264 kWh/day waste truck3_ef 1 [39] Energy used by the transfer station by each ton 12 kWh/ton waste energy_tonTS 0 [42] of segregated waste Energy used by the transfer station by each ton 12 kWh/ton waste energy_tonTS 0 [42] of segregated waste Efficiency of the collection truck 0.84 L/km waste truck1_ef 0 [39] Efficiency of the collection truck 0.84 L/km waste truck1_ef 1 [39] Efficiency of the transfer truck 1.2 L/km waste truck2_ef 0 [39] Efficiency of the transfer truck 1.2 L/km waste truck2_ef 1 [39] Efficiency of the collector truck compaction 0 L/m3 waste comp_ef 0 [39] system Efficiency of the collector truck compaction 0 L/m3 waste comp_ef 1 [39] system Times the solid waste is collected per week 6 times per week waste collections 0 [41] Times the solid waste is collected per week 6 times per week waste collections 1 [41] Capacity of the collection truck 7 ton waste truck1_cap 0 [39] Capacity of the collection truck 7 ton waste truck1_cap 1 [39] Capacity of the transfer truck 40 ton waste truck2_cap 0 [39] Capacity of the transfer truck 40 ton waste truck2_cap 1 [39] Efficiency of the landfill expressed in quantity 143 ton/h waste land_ef 0 [39] of waste that can handle by hour Efficiency of the landfill expressed in quantity 143 ton/h waste land_ef 1 [39] of waste that can handle by hour Total annual solid waste collected in the base 1450556 ton/yr waste tot_wvol_base 0 [39] year Total annual solid waste collected in the base 1450556 ton/yr waste tot_wvol_base 1 [39] year Energy needed to supply one m3 of water 2.07 kWh/m3 water water_factor 0 [43] Water distribution loss 16000 m3/km per yr water_loss loss 0 [44] End of the table 63 64 C. Visualization maps of policy levers As we have seen from the methodology, different combinations of policy levers define each scenario. The following maps were used to produce each policy lever included in the model and they help understanding the rules by which, scenarios are calculated. C.1 Public transport levers The specific public transportation policy levers used for Abidjan were: Figure C.1: Public transportation policy levers Figure C.2 shows Policy lever 0, which only includes the existing bus and ferry stations included in BAU scenario. In contrast, figure C.3 shows the transportation strategy included in SDUGA, considering new BRT, Metro and Water bus lines in addition to the existing bus stations. C.1. PUBLIC TRANSPORT LEVERS Figure C.2: Public transport- Current bus system Source: CAPSUS with data from SOTRA. Figure C.3: Public transport - New transport system Source: CAPSUS with data from JICA and MCLAU (2015). 66 APPENDIX C. VISUALIZATION MAPS OF POLICY LEVERS C.2 Population settlement policy levers The specific population settlement policy levers used for Abidjan were: Figure C.4: Settlement policy levers Figure C.5 shows Policy lever 0, in which future population occupies the predicted urban expansion areas. Figure C.5: Population settlement - Predicted urban expansion areas Source: CAPSUS. 67 C.2. POPULATION SETTLEMENT POLICY LEVERS In contrast, figure C.6 shows the allocated population in policy level 1, where the population settles first in areas near employment and public transport, second, in areas only with public transport, and third in areas with employment. Figure C.6: Population settlement - Employment and public transportation Source: CAPSUS. Figure C.7 shows level 2, where population is allocated in expansion areas first. After this area is filled to a minimum density, population settles in the low, medium and high zoned areas, progressively. 68 APPENDIX C. VISUALIZATION MAPS OF POLICY LEVERS Figure C.7: Population settlement - According to the Master Plan (priority in expansion areas) Source: CAPSUS with data from JICA and MCLAU (2015). In contrast in lever 3, figure C.8, population settlement is reversed. First, it is allocated in high density areas of the SDUGA. Second, in mid density areas, followed by low density. And lastly, it allocates it in expansion areas, when it is needed. 69 C.3. LAND USE POLICY LEVERS Figure C.8: Population settlement - According to the Master Plan (priority in infill areas) Source: CAPSUS with data from JICA and MCLAU (2015). C.3 Land use policy levers Land use policy levers determine how much population settles in the areas defined before. The specific land use policy levers used for Abidjan were: Figure C.9: Land Use policy levers 70 APPENDIX C. VISUALIZATION MAPS OF POLICY LEVERS Figure C.10 shows level 0, which respects the land uses of the Masterplan 2010, but it assumes a maximum density cap of 7,000 inhabitants per square km. Figure C.10: Land use - Current Master Plan (2010) Source: CAPSUS with data from JICA and MCLAU (2015). In contrast, figure C.11 shows the maximum population density caps according to the SDUGA 2030. It includes low density areas with a maximum density cap of 7,000 inhabitants per square km. The mid density caps are assumed at 14,500 inhabitants per square km, while the high density reaches 22,000 inhabitants per sq km. 71 C.3. LAND USE POLICY LEVERS Figure C.11: Land use - SDUGA Master Plan (2030) Source: CAPSUS with data from JICA and MCLAU (2015). Figure C.12 shows level 2, where polygons in proximity to employment and transport get a maximum population density caps of 22,000 inhabitants per square km, disregarding if they have authorized residential land uses or not in SDUGA. 72 APPENDIX C. VISUALIZATION MAPS OF POLICY LEVERS Figure C.12: Land use - New high density residential areas near transport corridors and employment hubs Source: CAPSUS. 73 C.4. SOLID WASTE LEVERS C.4 Solid waste levers The specific solid waste levers used for Abidjan were: Figure C.13: Solid waste management policy levers Figure C.14 shows the different levels of the solid waste management levers. Policy lever 0, which includes the existing open dumpsites and transfer stations, making longer tracts. Policy lever 1 shows the new transfer stations in addition to the existing ones. Figure C.14: Solid waste management Levers Flow Source: CAPSUS with data from ANAGED. 74 APPENDIX C. VISUALIZATION MAPS OF POLICY LEVERS C.5 Amenities Because of the lack of a proper amenity improvement strategy, it was not included as a policy lever. However the current amenity provision and its ratio of coverage are included as indicators. Figure C.15 shows the location of amenities considered in the analysis. Figure C.15: Existing Amenities Source: CAPSUS with data OSM. 75 C.5. AMENITIES 76 D. Expansion model methodology D.1 Expansion models methodology Urban growth projection models are not new, in fact, some of these processes were conceptualized in the 1940s and, since then, they have been used in a variety of environments to estimate patterns of urban expansion. Recent examples of their use includerecent studies in Zimbabwe, Belgium, The Netherlands and Iran [45], [46], [47], [48]. The models work with machine learning algorithms and are based on trends of change in historical density and land use data, for three moments in time. This data is then complemented with input variables that reflect the physical characteristics of the urban environment to find a causality for change, these variables are also known as explanatory variables. These models “learn” about the perceived tendencies and “train” to predict a future change in land use. Three different growth modeling methods were used for Abidjan: random forests, extratrees and logistic regression with regularization. Despite having the same input variables, results from each method may vary due to the use of different statistical methods. Random forest was the method with highest degree of confidence and it was selected as the future footprint area for the BAU scenario. Random Forest The Random Forest method is made up of a collection of decision trees, which are used to control the variance. This method can be described as a set (collection) of models that use aggregated sampling bootstraps to construct different decision trees, to later combine these models in a final classification. The Random Forest method has several advantages: they can handle many variables, they are quickly trained, they do not require distribution assumptions like the rest of the methods, they are generally robust in the treatment of outlier data and noise, and they provide a way to calculate the importance that each variable has in the model. Extratrees The Extratrees method is a variant of the Random Forest classifier [49] that uses the complete sample in each step with random decision limits (variables). Some advantages of Extratrees, compared to Random Forest, are that it represents a lower computational cost, the randomization makes the limits of each decision smoother and the use of fewer variables in each tree avoids overfitting. D.1. EXPANSION MODELS METHODOLOGY Logistic Regression The Logistic Regression method allows to predict the outcome of a dependent variable (categorical) based on a series of independent variables. In general terms, it allows to model the probability of an event that occurs as a function of other factors. To reduce the possibility of overfitting, a Ridge and Lasso regularization process was integrated. The Logistic Regression method allows to identify the variables that are significant for a particular model. D.1.1 Data sources Some of the most frequently used explanatory variables include proximity to urban centers, roads and metro stations [50] or proximity to built-up areas, roads, industrial centers, schools, universities, hospitals, airports, downtown area of the city, topographic characteristics, per capita income, altitude, average slope and population density [45]. In order to make results comparable among cities, global data sources with similar temporal information were used. This facilitates future replication of the methods, once the databases are updated. The data sources used in this work are: Built-up Grid: Data contain an information layer on built-up presence as derived from Sentinel1 image collections. • Source: Global Human Settlements [51]. • Temporality: 1990, 2000 and 2014. • Format: Raster with a 250 by 250 meters pixel resolution. Population Grid: Generated using census data combined with built-up index and aerial weights to generate the spatial distribution expressed as the number of people per cell. • Source: Global Human Settlements [52]. • Temporality: 1990, 2000 and 2015. • Format: Raster with a 250 by 250 meters pixel resolution. Digital Elevation model (DEM): From the GTOPO30, which is a global digital elevation model (DEM) with a horizontal grid of 1 square kilometer. GTOPO30 was derived from several raster and vector sources of topographic information. With the elevation model, a raster file with slopes in percentage was generated, which considers the maximum difference between each cell to its adjacent cells. • Source: U.S. Geological Survey ’s Center for Earth Resources Observation and Science (EROS) [53]. • Temporality: 1996. • Format: Raster with a 1,000 by 1,000 meters pixel resolution. Gross Domestic Product distribution: Gross Domestic Product spatial distribution derived from night lights satellite data. 78 APPENDIX D. EXPANSION MODEL METHODOLOGY • Source: National Oceanic and Atmospheric Administration (NOAA) [20]. • Temporality: 1995, 2000 and 2013. • Format: Raster with a 1,000 by 1,000 meters pixel resolution. Highways • Source: Open Street Maps. • Temporality: starting from 2008. • Format: lines geometry. Geolocations: airports, schools, universities, worship places and hospitals. • Source: Open Street Maps. • Temporality: starting from 2008. • Format: points geometry. Water Bodies: Provides a base map for the lakes, seas, oceans, large rivers, and dry salt flats of the world. • Source: Esri Data and Maps [54]. • Format: polygons geometry. D.1.2 Definition of urban There are no standard criteria to define an urban area, but many definitions include elements like population size and density, economic activity, level of infrastructure, or a combination of them. For this project, an adaptation to the harmonized urban cluster defined by the European Commission [55] was used, where a cell is considered urban if: 1. The maximum value of built-up is greater or equal than 40%. 2. The mean value of the population is greater or equal than 75 people per 0.16 km2 . 3. The total number of people in adjacent cells is greater or equal than 5,000. The three models for Greater Abidjan were estimated using the data sources and definition of urban mentioned in this section. The resulting urban expansion predictions are shown in Section D.2. D.2 Expansion model results This section contains the results of the urban expansion predicted for Abidjan. Three different machine learning methods were used to predict the expansion between the years 2015 and 2030: Random Forests, Extratrees and Logistic Regression. 79 D.2. EXPANSION MODEL RESULTS Table D.1 contains accuracy parameters for the results of the three different growth models, which helps evaluating the performance of each model against the others and, therefore, to choose the best fitting one. Table D.1: Accuracy of the urban expansion models Model TN TP FP FN Number Precision Recall F1 of units Extratrees 6,159 3,802 129 126 10,216 0.96 0.96 0.96 Random Forest 6,129 3,818 159 110 10,216 0.96 0.97 0.96 Logistic Regression 6,044 3,754 244 174 10,216 0.93 0.95 0.94 Where: • TN = True Negative - number of cells correctly predicted as rural • TP = True Positive - number of cells correctly predicted as urban • FP = False Positive - number of cells where urban for a rural area was predicted • FN = False Negative - number of cells where rural for an urban area was predicted • Precision: measures the proportion of urban cells that are correctly classified from all the actual positive. precision = T P T P + F P • Recall: measures the proportion of urban cells that are correctly identified as such over all cells that are classified as urban. recall = T P T P + F N • F1: it is the harmonic mean between precision and recall. It is a single measure of performance of the test for the positive class, in this case urban cells. F 1 = 2T P 2T P + F P + F N Although both, Random Forests and Extratrees, models achieved better results than Logistic Regression in overall terms, Random Forests achieve a higher score in recall. This means that the model does not usually mispredict an area that is actually rural as urban. Table D.2 presents the expansion predicted by each model compared with the historical expansion of the city in 1990-2000 and 2000-2014. The urban expansion predicted by the Random Forest model was used for the analysis of Abidjan’s BAU scenario, which was 25.5 square kilometers for 2030, equivalent to 1.60 square kilometers per year. Maps displayed in figure D.1 plot the urban footprint of each city in the years 1990, 2000 and 2014, and the expansion predicted for 2030 by the three expansion models. 80 APPENDIX D. EXPANSION MODEL METHODOLOGY Table D.2: Urban expansion between 1990 and 2014 and expansion predicted for 2030 Expansion model 1990 - 2000 [km2 ] 2000 - 2014 [km2 ] 2015 - 2030 [km2 ] Extratrees 35.82 35.23 18.82 Random Forest 35.82 35.23 25.55 Logistic Regression 35.82 35.23 36.02 Figure D.1: Abidjan's urban footprint in 1990, 2000 and 2014 (upper left) and urban growth prediction for 2030 using Logistic Regression (upper right), Extratrees (lower left) and Random Forest model (lower right) 81 D.2. EXPANSION MODEL RESULTS 82 Bibliography [1] WBG. World Bank Data, 2017. [2] Barney Cohen. Urbanization in developing countries: Current trends, future projections, and key challenges for sustainability. Technology in Society, 28(1- 2):63–80, 2006. [3] Kouamé Appessika. The case of Abidjan. Technical report, University College London, Abidjan, 2003. [4] René Parenteau and François Charbonneau. Érudit est un consortium interuniversitaire sans but lucratif composé de l Abidjan: une politique de l’habitat au service du plan urbain. Cahiers de géographie du Québec, 36(99):415– 437, 1993. [5] Yao Gnamba. Rétrospective de l’aménagement du territoire en Côte d’Ivoire: le cas d’Abidjan. Revue Canadienne de Géographie Tropicale, 1(1):36–48, 2014. [6] JICA and MCLAU. Project For The Development Of The Urban Master Plan In Greater AbidjanProject For The Development Of The Urban Master Plan In Greater Abidjan. Technical report, Japan International Cooperation Agency and Ministry of Construction, Housing, Sanitation and Development, Abidjan, 2015. [7] JICA. Republique de Cote d’Ivoire Le projet de developpement du schema directeur d’urbanisme du Grand Abidjan (SDUGA). Technical report, Japan International Cooperation Agency (JICA), Abidjan, 2015. [8] CABE. Better Neighbourhoods: Making higher densities work. Comission for Architecture and the Built Environment, Londres, 2005. [9] Campaign to Protect Rural England. Compact Sustainable Communities. Number November. Londres, 2006. [10] ITDP. Hacia una estrategia de desarrollo orientado al transporte para el distrito federal. Instituto de Políticas para el Transporte y el Desarrollo México, Ciudad de México, 2015. [11] Volker Krey, Omar Masera, Geoffrey Blanforde, Thomas Bruckner, Roger Cooke, Karen Fish-Vanden, Helmut Haberl, Edgar Hertwich, Elmar Kriegler, Daniel Müller, Sergey Paltsev, Lynn Price, Steffen Schlömer, Diana Uerge-Vorsatz, Detlef Van Vuuren, and Timm Zwickel. Annex II: Metrics & Methodology. Climate Change 2014: Mitigation of Climate Change. Contribution of Working Group III to BIBLIOGRAPHY the Fifth Assessment Report of the Intergovernmental Panel on Climate Change, pages 1281–1328, 2014. [12] Jamal Jaber and Ahmed Al-ghandoor. Prediction of energy consumption of passenger transportation and GHG emissions in Jordan transportation and GHG emissions in Jordan. (April 2015), 2012. [13] The World Bank. Côte d’Ivoire Urbanization Review. Technical report, The World Bank, Washington DC, 2015. [14] Energy & Industrial Strategy Department for Business. Greenhouse gas reporting: conversion factors 2017, 2017. [15] Hammond, G. and Jones C. Inventory of Carbon and Energy (ICE). Number 1.6a ed. University of Bath, Bath, 2008. [16] Shane L Silsby. HPS to LED Conversion – A City of Phoenix Experience. 2013. [17] Ahmad Qasaimeh, Mohammad Qasaimeh, Zaydoun Abu-salem, and Mohammad Momani. Solar Energy Sustainability in Jordan. Computational Water, Energy, and Environmental Engineering, 3(April):41–47, 2014. [18] SENER-GIZ. Reporte técnico del Simulador de Ahorro de Agua en la Vivienda – SAAVi, Versión 2.0 (2016). 0, 2016. [19] Tilottama Ghosh, Rebecca L Powell, Christopher D Elvidge, Kimberly E Baugh, Paul C Sutton, and Sharolyn Anderson. Shedding Light on the Global Distribution of Economic Activity. The Open Geography Journal, 3:148–161, 2010. [20] National Oceanic and Atmospheric Administration. An Estimate of Gross Domestic Product (GDP) Derived From Satellite Data, 2010. [21] BNETD. Infrastructure costs data collected for the project Urban Growth Scenarios Abidjan, 2018. [22] Ci-Energies. Energy data collected for the project Urban Growth Scenarios Abidjan, 2018. [23] Office National de l’Eau Potable. Demande infos Modèle Coissance Abj 04 11 2017 01. [24] National Electric Power Company. NEPCO Annual Report 2016. [25] MANASEER Oil & Gas. Fuel Prices Up, 2017. [26] REN21. Renewables 2013 Global Status Report. Renewable Energy Policy Network for the 21st Century, 2013 edition, 2013. [27] UNESCO. The United Nations World Water Development Report 3: Water in a Changing World. World Water Assessment Programme, Paris, FR, first edit edition, 2009. [28] Open Street Map. [29] INEGI. Calculadora de inflación, 2017. 84 BIBLIOGRAPHY [30] Peter M. Ward and C. B. Smith. Housing rehab for consolidated informal settlements: A new policy agenda for 2016 UN-Habitat III. Habitat International, 50(2015):373–384, 2015. [31] GPP. Global Petrol Prices. [32] XE.com Inc. XE Currency Converter, 2017. [33] Egyptian Electricity Holding Company. Annual report 2015/1016. page 64, 2016. [34] B. Shiva Kumar and K. Sudhakar. Performance evaluation of 10 MW grid connected solar photovoltaic power plant in India. Energy Reports, 1:184–192, 2015. [35] UN Habitat. State of the World’s Cities. Technical report, UN Habitat, Nairobi, 2013. [36] INFONAVIT. Hipoteca Verde. Instituto del Fondo Nacional de la Vivienda para los Trabajadores, 2016. [37] Yao Godefroy Konan. La planification intégrée, outil de développement durable des transports urbains : cas du Grand Abidjan. In CODATU XV Changement Climatique, Qualité de l’Air et Défis Énergétiques, Istanbul, 2015. [38] SENER. Inventario Nacional de Energías Renovables, 2016. [39] BNETD. 2 dans la commune de Yopougon ( après la zone industrielle ) ; 1 à Bingerville ; 1 à Adjamé ; 1 à Anyama ; la collecte et le transport des déchets depuis les ménages jusqu ’ aux centres de transfert ; le transfert des déchets des centres de transfert au C. 2018. [40] SEMARNAT. Residuos. page 380, 2016. [41] Harinaivo A Andrianisa, Yves O.K. Brou, and Alphonse Séhi bi. Role and importance of informal collectors in the municipal waste pre-collection system in Abidjan, Cote d’Ivoire. Habitat International, 53:265–273, 2016. [42] Eunomia Research & Consulting. Kerbside Collections Options : Wales. Technical Report January, 2011. [43] EEA. Towards efficient use of water resources in Europe. European Environment Agency, Luxemburg, Lu, first edit edition, 2012. [44] PIGOO. Programa de Indicadores de Gestión de Organismos Operadores, 2016. [45] Courage Kamusoko and Jonah Gamba. Simulating Urban Growth Using a Random Forest-Cellular Automata (RF-CA) Model. ISPRS International Journal of Geo- Information, 4(2):447–470, 2015. [46] Ahmed Mustafa, Mario Cools, Ismaïl Saadi, and Jacques Teller. Coupling agent- based, cellular automata and logistic regression into a hybrid urban expansion model (HUEM). Land Use Policy, 69(October):529–540, dec 2017. [47] Tong Wang, Qi Han, and Bauke de Vries. A semi-automatic neighborhood rule discovery approach. Applied Geography, 88:73–83, nov 2017. 85 BIBLIOGRAPHY [48] Hossein Shafizadeh-Moghadam, Ali Asghari, Mohammad Taleai, Marco Helbich, and Amin Tayyebi. Sensitivity analysis and accuracy assessment of the land transformation model using cellular automata. GIScience & Remote Sensing, 54(5):639–656, sep 2017. [49] Pierre Geurts, Damien Ernst, and Louis Wehenkel. Extremely randomized trees. Machine Learning, 63(1):3–42, 2006. [50] Yan Liu, Yongjiu Feng, and Robert Pontius. Spatially-Explicit Simulation of Urban Growth through Self-Adaptive Genetic Algorithm and Cellular Automata Modelling. Land, 3(3):719–738, 2014. [51] Martino Pesaresi, Daniele Ehrilch, Aneta J. Florczyk, Sergio Freire, Andreea Julea, Thomas Kemper, Pierre Soille, and Vasileios Syrris. GHS built- up grid, derived from Landsat, multitemporal (1975, 1990, 2000, 2014), 2015. European Commission, Joint Research Centre (JRC) [Dataset] PID: http://data.europa.eu/89h/jrc-ghsl-ghs_built_ldsmt_globe_r2015b. [52] European Commision Joint Research Centre (JRC) and Columbia University Center for International Earth Science Information Network (CIESIN). GHS population grid, derived from GPW4, multitemporal (1975, 1990, 2000, 2015)., 2015. European Commission, Joint Research Centre (JRC) [Dataset] PID: http://data.europa.eu/89h/jrc-ghsl-ghs_pop_gpw4_globe_r2015a. [53] U.S. Geological Survey ’s Center for Earth Resources Observation and Science. Global 30 Arc-Second Elevation (GTOPO30), 1996. [54] Inc. Esri, Garmin International. World Water Bodies, 2017. [55] Lewis Dijkstra and Hugo Poelman. A harmonised definition of cities and rural areas: the new degree of urbanisation Working Papers. 2014. 86 BIBLIOGRAPHY 87 www.capitalsustentable.com