Applying the Degree of Urbanisation A METHODOLOGICAL MANUAL TO DEFINE CITIES, TOWNS AND RURAL AREAS FOR INTERNATIONAL COMPARISONS 2021 edition MANUALS AND GUIDELINES Applying the Degree of Urbanisation A METHODOLOGICAL MANUAL TO DEFINE CITIES, TOWNS AND RURAL AREAS FOR INTERNATIONAL COMPARISONS 2021 edition Manuscript completed in December 2020 The designations employed and the presentation of material in this information product do not imply the expression of any opinion whatsoever on the part of the European Commission, the Food and Agriculture Organization of the United Nations (FAO), the United Nations Human Settlements Programme (UN-Habitat), the Organisation for Economic Co-operation and Development (OECD) or The World Bank concerning the legal or development status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. The mention of specific companies or products of manufacturers, whether or not these have been patented, does not imply that these have been endorsed or recommended by the European Commission, the FAO, UN-Habitat, the OECD or The World Bank in preference to others of a similar nature that are not mentioned. The views expressed in this information product are those of the author(s) and do not necessarily reflect the official position, views or policies of the European Commission, the FAO, UN-Habitat, the OECD or of its member countries or The World Bank, its Board of Executive Directors or the governments they represent. European Union identifiers PDF: ISBN 978-92-76-20306-3 doi: 10.2785/706535 KS-02-20-499-EN-N FAO identifier ISBN 978-92-5-134073-8 UN-Habitat identifiers ISBN 978-92-1-132866-0 HS number: HS/019/20E © European Union/FAO/UN-Habitat/OECD/The World Bank, 2021. Some rights reserved. This work is made available under the Creative Commons Attribution-NonCommercial-ShareAlike 3.0 IGO licence (CC BY-NC-SA 3.0 IGO; https://creativecommons.org/licenses/by-nc-sa/3.0/igo/legalcode) (the “Licence”). Under the terms of this Licence, this work may be copied, redistributed and adapted for non-commercial purposes, provided that the work is appropriately cited. In any use of this work, there should be no suggestion that the European Commission, the FAO, UN-Habitat, the OECD or The World Bank endorse any specific organisation, products or services. The use of the European Commission, the FAO, UN-Habitat, the OECD or The World Bank logos is not permitted. If the work is adapted, including translation, then it must be licensed under the same or equivalent Creative Commons Licence. If a translation of this work is created, it must include the following disclaimer along with the required citation: “This translation was not created by the European Commission, the Food and Agriculture Organization of the United Nations (FAO), the United Nations Human Settlements Programme (UN-Habitat), the Organisation for Economic Co-operation and Development (OECD) or The World Bank. The European Commission, the FAO, UN-Habitat, the OECD or The World Bank are not responsible for the content or accuracy of this translation. The original English edition shall be the authoritative edition”. Disputes arising under the Licence that cannot be settled amicably will be resolved by mediation and arbitration as described in Article 8 of the Licence except as otherwise provided herein. The applicable mediation rules will be the mediation rules of the World Intellectual Property Organisation http://www.wipo.int/amc/en/mediation/rules and any arbitration will be in accordance with the Arbitration Rules of the United Nations Commission on International Trade Law (UNCITRAL). For any clarifications on the re-use of this work, you may contact the Publications Office copyright service at op-copyright@publications.europa.eu. The European Union does not own the copyright in relation to the following elements: cover photo, Alessandro Pinto/ Shutterstock.com; maps, administrative boundaries © EuroGeographics © UN-FAO © Turkstat. Luxembourg: Publications Office of the European Union, 2021 Theme: General and regional statistics Collection: Manuals and guidelines Foreword Foreword In March 2020, the UN Statistical Commission endorsed a new methodology to define cities, towns and rural areas and urged that a technical report on how to apply this methodology be released as early as possible. This manual responds to that request. It presents in full detail how to classify an entire territory along the urban-rural continuum into one of three distinct classes: cities; towns and semi-dense areas; and rural areas. The United Nation’s Sustainable Development Goals, and several other global agendas, call for the production of harmonised statistics for urban and rural areas. These indicators were harmonised, but the definition of the territories was left open. This meant that national statistical authorities used their own definitions of urban and rural areas. These national definitions use a variety of approaches, indicators and thresholds, which limits the international comparability of these areas. To resolve this issue, six international organisations or agencies - the European Commission, the Food and Agriculture Organization of the United Nations, the United Nations Human Settlements Programme, the International Labour Organization, the Organisation for Economic Co-operation and Development and The World Bank – worked together to develop this new definition and to produce this manual. I hope that many countries will use this manual to produce more comparable indicators by type of area, as this will enable countries to better identify the areas that are close to reaching the Sustainable Development Goals and the policies that are contributing to this success. Mariana Kotzeva Director-General, Eurostat Applying the Degree of Urbanisation — 2021 edition  3 Abstract Abstract Applying the Degree of Urbanisation — A methodological manual to define cities, towns and rural areas for international comparisons has been produced in close collaboration by six organisations — the European Commission, the Food and Agriculture Organization of the United Nations (FAO), the United Nations Human Settlements Programme (UN- Habitat), the International Labour Organization (ILO), the Organisation for Economic Co-operation and Development (OECD) and The World Bank. This manual develops a harmonised methodology to facilitate international statistical comparisons and to classify the entire territory of a country along an urban-rural continuum. The degree of urbanisation classification defines cities, towns and semi-dense areas, and rural areas. This first level of the classification may be complemented by a range of more detailed concepts, such as: metropolitan areas, commuting zones, dense towns, semi-dense towns, suburban or peri-urban areas, villages, dispersed rural areas and mostly uninhabited areas. The manual is intended to complement and not replace the definitions used by national statistical offices (NSOs) and ministries. It has been designed principally as a guide for data producers, suppliers and statisticians so that they have the necessary information to implement the methodology and ensure coherency within their data collections. It may also be of interest to users of subnational statistics so they may better understand, interpret and use official subnational statistics for taking informed decisions and policymaking. Editorial team Lewis Dijkstra (European Commission, Directorate General for Regional and Urban Policy), Teodora Brandmüller (European Commission, Eurostat), Thomas Kemper (European Commission, Joint Research Centre), Arbab Asfandiyar Khan (FAO), INFORMA s.à. r.l. and Paolo Veneri (OECD). Production and layout This publication was produced by INFORMA s.à r.l. Contact details Eurostat Bâtiment Joseph Bech 5, rue Alphonse Weicker 2721 Luxembourg E-mail: estat-user-support@ec.europa.eu For more information please consult Eurostat’s website: https://ec.europa.eu/eurostat Acknowledgements The editors of this publication would like to thank the following colleagues — Olivier Draily (European Commission, Directorate-General for Regional and Urban Policy), Bianka Fohgrub, Oliver Müller, Ruxandra Roman Enescu and Jane Schofield (European Commission, Eurostat), Sergio Freire, Luca Maffenini, Michele Melchiorri and Marcello Schiavina (European Commission, Joint Research Centre), Eva Panuska Jandova and Antoine Malherme (European Commission, Publications Office of the European Union), Pietro Gennari (FAO), Robert Ndugwa (UN-Habitat), Monica Castillo (ILO), Ellen Hamilton (The World Bank), Simon Allen and Andrew Redpath (INFORMA s.à r.l.) — who were involved in its preparation. 4   Applying the Degree of Urbanisation — 2021 edition Contents Contents Foreword..........................................................................................................................................................................3 1. Introduction................................................................................................................................................................7 References....................................................................................................................................................................................................................9 2. The legal and policy framework............................................................................................................................11 References................................................................................................................................................................................................................. 14 3. Rationale and advantages......................................................................................................................................15 3.1 Captures the urban-rural continuum in harmonised manner.......................................................................................... 15 3.2 Uses the same population size and density thresholds across the globe................................................................. 17 3.3 Starts from a population grid to reduce the bias generated by the different shapes and sizes of spatial units................................................................................................................................................................................................. 18 3.4 Measures population clusters directly............................................................................................................................................. 19 3.5 Defines areas to monitor access to services, not areas defined by access to services...................................... 20 3.6 Proposes a cost-effective approach................................................................................................................................................... 21 References................................................................................................................................................................................................................. 21 4. How the principles of official statistics and classifications are fulfilled.........................................................23 References and further information.......................................................................................................................................................... 24 5. Constructing a population grid.............................................................................................................................25 5.1 A grid based on the aggregation of point data......................................................................................................................... 26 5.2 A grid based on the disaggregation of population data...................................................................................................... 27 5.3 Extrapolating a population grid based on a partial micro-census................................................................................. 29 5.4 Alternative and emerging data sources for creating population grids....................................................................... 30 References................................................................................................................................................................................................................. 31 6. Methodology for applying level 1 of the degree of urbanisation classification..........................................33 6.1 Terminology...................................................................................................................................................................................................... 33 6.2 Short description........................................................................................................................................................................................... 33 6.3 Grid cell classification.................................................................................................................................................................................. 34 6.4 Classifying small spatial units................................................................................................................................................................. 40 6.5 Changes over time that impact on the classification given to each small spatial unit..................................... 44 References................................................................................................................................................................................................................. 46 7. Extensions to level 1 of the classification............................................................................................................47 7.1 Level 2 of the degree of urbanisation............................................................................................................................................... 47 7.2 Defining functional urban areas.......................................................................................................................................................... 51 7.3 Other possible extensions to the methodology: remoteness and land cover....................................................... 59 References................................................................................................................................................................................................................. 62 Applying the Degree of Urbanisation — 2021 edition  5 Contents 8. Which spatial units to use and adjustments to address geographic issues..................................................63 8.1 Which small spatial units to use?......................................................................................................................................................... 63 8.2 Adjustments to address specific geographic issues for the degree of urbanisation and functional urban area classifications................................................................................................................................................. 67 References................................................................................................................................................................................................................. 70 9. Selected indicators for sustainable development goals by degree of urbanisation and functional urban area..............................................................................................................................................71 References................................................................................................................................................................................................................. 88 10. Tools and training..................................................................................................................................................89 10.1 Tools.................................................................................................................................................................................................................... 89 10.2 Training............................................................................................................................................................................................................. 92 10.3 Online resources for the degree of urbanisation classification...................................................................................... 95 References................................................................................................................................................................................................................. 96 11. Conclusions.............................................................................................................................................................97 6   Applying the Degree of Urbanisation — 2021 edition 1 Introduction A United Nations Resolution adopted in September 2015, Transforming our World: the 2030 Agenda for Sustainable Development (UN (2015)) includes several indicators for sustainable development goals (SDGs) that should be collected for cities or for urban and rural areas. So far, however, no global methodology or international standard has been proposed to delineate these areas. The broad array of different criteria applied in national definitions of urban and rural areas poses serious challenges to cross-country comparisons (ILO (2018)). The Action Framework of the Implementation of the New Urban Agenda (UN-Habitat (2017)) and the Global Strategy to improve Agricultural and Rural Statistics (IBRD-WB (2011)) both highlight the need for a harmonised methodology to facilitate international comparisons and to improve the quality of urban and rural statistics in support of national policies and investment decisions. This is why six organisations — the European Commission, the Food and Agriculture Organization of the United Nations (FAO), the United Nations Human Settlements Programme (UN-Habitat), the International Labour Organization (ILO), the Organisation for Economic Co-operation and Development (OECD) and The World Bank — have been working closely together over the past four years to develop a harmonised, simple and cost-effective methodology. This new methodology allows statistics to be compiled by degree of urbanisation, identifying cities, towns and semi-dense areas, and rural areas at level 1 of the classification. By using three classes instead of only two (urban and rural), it captures the urban-rural continuum. To improve the international comparability of urban and rural indicators for SDGs, it is recommended to produce these by degree of urbanisation. The first level of the degree of urbanisation classification may be extended in two ways. The first extension, called level 2 of the degree of urbanisation classification, is a more detailed territorial typology: it identifies, cities, towns, suburban or peri-urban areas, villages, dispersed rural areas and mostly uninhabited areas. The second extension, defines functional urban areas (otherwise referred to as metropolitan areas), covering cities and the commuting zones around them. In order to produce SDG indicators by level 2 of the degree of urbanisation classification or by functional urban area, it is necessary to use surveys with large samples. As a result, it will not always be feasible to produce SDG indicators for these two extensions. To highlight the interest and the feasibility of producing SDG indicators by degree of urbanisation, this manual includes examples of indicators from 12 of the 17 goals for a range of countries across the globe. The indicators tend to have a clear urban gradient with cities at one end, rural areas at the other and with towns and semi-dense areas in between. In some cases, cities tend to fare better, for example in terms of access to education, in others, rural areas tend to do better, for example in terms of personal safety. This methodological manual is meant to complement and not replace the already existing definitions used by NSOs and ministries. Indeed, these national definitions typically rely on a much wider set of criteria which may have been refined to take into account specific characteristics, context and policy objectives. The manual has been designed principally as a practical guide for data producers, suppliers and statisticians so that they have the necessary information to implement the methodology and ensure coherency and consistency within their data collections and analyses. It may also be of interest to users of subnational statistics — such as policymakers, the private sector, research institutions, academia — so that they may better understand and interpret official subnational statistics. Applying the Degree of Urbanisation — 2021 edition  7 1 Introduction The manual was produced at the request of the 51st session of the UN Statistical Commission (UNSC), which ’endorsed the methodology for delineation of cities and urban and rural areas for international and regional statistical comparison purposes, and [the UNSC] urged the release of a technical report on the implementation of the methodology for delineation of cities and urban and rural areas as early as possible’ (1). A draft of this report was submitted for global consultation. This took place from 5 October 2020 to 5 November 2020. Input/comments were received from 22 individual countries and these were incorporated into the manuscript in November 2020. The authors would like to thank very much all the countries and experts who provided their opinions and comments. These were very enriching and certainly increased the quality of the final manual. Some of the comments received raised questions that went beyond the scope of this manual, in particular, detailed comments and questions on the production of a population grid. These issues should be addressed by a separate manual with global guidelines on how to produce an official population grid. Table 1.1: Milestones on the way to the endorsement by the UN Statistical Commission October 2016 UN-Habitat III conference, Quito The European Commission’s Commissioner for Regional and Urban development announced a joint voluntary commitment with the OECD and The World Bank to develop a global, people-based definition of cities and settlements. March 2017 UN Statistical Commission (UNSC), New York Presentation of the work plan, first results and discussion on next steps in two dedicated side events. April 2017 UN-Habitat Expert Group meeting, Brussels The Expert Group Meeting on Geospatial Definitions for Human Settlements Indicators of the SDGs concluded that a standard definition of a city is needed for global reporting and monitoring of the SDGs. November 2017 UN Statistical Division (UNSD) survey The UNSD sent a questionnaire to 20 countries to gather feedback on the proposed methodology. At least three quarters of the respondents stated that the methodology was useful for international comparisons and to compile indicators for the UN’s SDGs. January 2018 Food and Agriculture Organization of the United Nations (FAO) Expert Group meeting, Rome The Expert Group meeting on Improving Rural Statistics: Rural Definition and Indicators reviewed and made recommendations on the methodology. March 2018 UN Statistical Commission (UNSC), New York The interim results were presented at a side event of the UNSC, which highlighted the interest and support for this global development. Further consultations and communication to raise the awareness and understanding of this new methodology were planned. December 2018 FAO and the Global Strategy to improve Agricultural and Rural Statistics (GSARS) published its findings on pilot tests FAO and the GSARS tested the definition (at level 1 and level 2) for seven countries in their national contexts. The report also assessed the countries’ capacity and capability to report on a subset of core SDG indicators, applying the methodology and using existing data collection mechanisms. October 2018 – UN-Habitat regional workshops October 2019 UN-Habitat organised seven regional workshops to present the methodology and discuss how it could be improved and applied nationally. A total of 85 countries participated in these workshops (see Figure 10.5 for a complete list). January 2019 UN Expert Group meeting, New York An Expert Group meeting on the Statistical Methodology for Delineating Cities and Rural Areas (UN (2019)) concluded that both the degree of urbanisation and functional urban area classifications were useful to monitor the SDGs and should be used in parallel with national definitions of urban and rural areas. March, 2019 UN Statistical Commission (UNSC), New York The UNSC welcomed the work on developing the methodology for the delineation of urban and rural areas and the definition of cities based on the degree of urbanisation classification, and requested the submission of the final assessment, to be prepared in consultation with Member States, on the applicability of this methodology for international and regional comparison purposes to the Commission at its fifty-first session (see E/2019/24-E/ CN.3/2019/34, Decision 50/118, paragraph (d)). March 2020 UN Statistical Commission, New York The UNSC ’endorsed the methodology for delineation of cities and urban and rural areas for international and regional statistical comparison purposes’. (1) UN Statistical Commission (UNSC), Report on the fifty-first session (3-6 March 2020), Economic and Social Council, Official Records, 2020, Supplement No. 4, E/2020/24-E/CN.3/2020/37, 51/112 paragraph (i-j). 8   Applying the Degree of Urbanisation — 2021 edition Introduction 1 References FAO and GSARS (2018), Pilot tests of an international definition of urban–rural territories, Technical Report no. 37, Rome. IBRD-WB (2011), Global Strategy to improve Agricultural and Rural Statistics, International Bank for Reconstruction and Development/The World Bank, Economic and Sector Work, Report No. 56719-GLB, Washington D.C. UN (2015), Transforming our World: the 2030 Agenda for Sustainable Development, United Nations, General Assembly, A/ RES/70/1, New York. UN (2019), Expert Group Meeting on Statistical Methodology for Delineating Cities and Rural Areas, United Nations Statistics Division, New York. UN-Habitat (2017), New Urban Agenda, United Nations Conference on Housing and Sustainable Urban Development (Habitat III), United Nations, General Assembly, A/RES/71/256, New York. UN-Habitat (2017), Expert Group Meeting on Geospatial Definitions for Human Settlements Indicators of the SDGs, Brussels. Applying the Degree of Urbanisation — 2021 edition  9 2 The legal and policy framework Designing effective policies requires a good understanding of the socioeconomic conditions that exist in cities and in urban and rural areas, which in turn depends on building a solid base of knowledge about people, their activities, communities, well-being and their interaction with the environment. Reliable, timely and internationally comparable datasets for different urban and rural areas can only be produced on the basis of coherent and harmonised methodology that delineates cities, urban and rural areas in a consistent manner. 2030 AGENDA FOR SUSTAINABLE DEVELOPMENT In 2015, the United Nations General Assembly adopted the 2030 Agenda for Sustainable Development (UN (2015)). At the core of the agenda, there is a set of 17 sustainable development goals (SDGs), which provides a global policy framework for stimulating action until the year 2030 in areas of critical importance related to people, the planet, prosperity, peace and partnership. A global list of 232 indicators was developed to measure progress towards 169 targets across these 17 goals from the 2030 agenda. Cities, urban and rural areas play a crucial role for many policy areas underlying the SDGs such as eradicating poverty and hunger, housing, transport, infrastructure, land use or climate change. Beyond SDG 11 — make cities and human settlements inclusive, safe, resilient and sustainable — which focuses explicitly on cities and communities, an estimated two thirds of the 169 targets can be measured and analysed for cities and urban and rural areas which can help shape sustainable development policies from the ground up and provide support to help reach the targets set in the 2030 agenda. NEW URBAN AGENDA Urbanisation is a phenomenon that impacts economies, societies, cultures and the environment. It is projected that 55 % of the world’s population will be living in cities by 2050 (OECD and European Commission (2020)). Not only is there a growing level of interest in the rapid growth and shape of urban developments, but also in the linkages that exist between individual cities and between urban and rural areas. One particular area of policy interest is that of mega cities and large metropolitan areas that benefit from economies of agglomeration, industrial clustering and innovation, while at the same time facing significant challenges with respect to sustainable urban development (for example, congestion or environmental impacts). A United Nations Conference on Housing and Sustainable Urban Development (Habitat III) in Quito, Ecuador, on 20 October 2016 adopted the New Urban Agenda; it was subsequently endorsed by the United Nations General Assembly on 23 December 2016 (UN-Habitat (2017)). The New Urban Agenda seeks to provide a vision for a more sustainable future by promoting a new model of urban development, based on the premise that cities can be the source of solutions to, rather than the cause of, many global challenges. It provides standards and principles for the planning, construction, development, management, and improvement of urban areas following five main pillars: national urban policies, urban legislation and regulations, urban planning and design, local economy and municipal finance, and local implementation. Applying the Degree of Urbanisation — 2021 edition  11 2 The legal and policy framework RURAL DEVELOPMENT POLICIES Rural areas are intrinsically important and fundamentally different from urban areas and thus (often) require a different set of interventions and policies that aim to improve the livelihood of their populations. Research and empirical evidence show that rural areas are characterised by: slow dynamics of farm productivity, widespread income inequality and volatility of agricultural income; considerable outward migration flows to urban areas that result in depopulation of rural areas; a lack of efficient physical, technological and information technology (IT) infrastructures; public and private services that are more costly to provide and more difficult to access than in urban areas (OECD (2020)). Despite their importance, rural statistics on income and livelihoods are sparse and uncommon, mainly due to the fact that there is no consistent international definition of rural areas. Rural areas are usually defined based on national policy objectives; sometimes, as a residual, once urban areas are defined, or sometimes based on a combination of multiple criteria, for example, population size and density, the presence of agriculture, remoteness from urban areas and a lack of infrastructure and/or basic social services. It is important to highlight that rural statistics are territorial in nature, in contrast to sectoral statistics that focus on a single activity. People in rural areas are typically engaged in several different economic activities beyond agriculture, fisheries and forestry, for example mining and quarrying, as well as in craft production. Some of the main challenges facing rural areas include: malnutrition, food insecurity, poverty, limited adequate health and education services, a lack of access to other basic infrastructure and the under-utilisation of labour. In formulating a rural development policy, the FAO draws on issues identified in the 2030 Agenda for Sustainable Development, while acknowledging that rural areas have particular characteristics that present unique challenges. These include, among others: the dispersion of rural populations; topographical features (terrain and landscapes) that may act as a barrier for the efficient provision of infrastructure; an (over) reliance on the agricultural sector; ensuring that natural resources and environmental quality are protected. 12   Applying the Degree of Urbanisation — 2021 edition The legal and policy framework 2 INTERNATIONAL STATISTICS DIFFERENTIATING BETWEEN URBAN AND RURAL AREAS The idea of differentiating between urban and rural areas for international statistics dates back several decades. In 1991, the European Union labour force survey introduced a variable to indicate the characteristics of the areas where respondents lived. However, its results had limited comparability internationally. In 2012, the OECD together with the European Commission developed a new way to measure metropolitan areas (OECD (2012), later extended in Dijkstra et al. (2019)). It seeks to ensure that statistics on urban development are made more robust through the provision of an internationally recognised definition of cities and their commuting zones as functional economic units that may guide policymakers better in areas such as planning, infrastructure, transport, housing, education, culture and recreation. The European Commission’s Directorate-General for Regional and Urban Policy (DG REGIO) published A harmonised definition of cities and rural areas: the new degree of urbanisation (Dijkstra and Poelman (2014)). It describes the degree of urbanisation classification and distinguishes three different classes: cities, towns and suburbs, and rural areas (or densely, intermediate and thinly populated areas) that are based on information for population grids to provide more robust data (greater comparability and availability). Prior to 2017, territorial typologies and their related methodologies within the European Statistical System (ESS) did not have any legal basis. On 12 December 2017, an amending Regulation (EU) 2017/2391 of the European Parliament and of the Council was adopted as regards territorial typologies (Tercet), followed on 18 January 2018 by a consolidated and amended version of Regulation (EC) No 1059/2003 of the European Parliament and of the Council on the establishment of a common classification of territorial units for statistics (NUTS). The main objectives of Tercet include: establishing a legal recognition of territorial typologies for the purpose of European statistics by laying down core definitions and statistical criteria; integrating territorial typologies into the NUTS Regulation so that specific types of territory may be referred to in thematic statistical regulations or policy initiatives, without the need to (re-)define terminology such as cities and urban or rural areas; ensuring methodological transparency and stability, by clearly promoting how to update the typologies. As part of the Global Strategy to improve Agricultural and Rural Statistics (GSARS), the FAO published Guidelines on defining rural areas and compiling indicators for development policy (FAO (2018)). These guidelines provide a definition of which territories should be considered as rural and a more detailed breakdown of different types of rural places to promote like-for-like comparison internationally. The guidelines seek to provide information on concepts and methods to improve the quality, availability and use of rural statistics. The United Nations Statistics Division (UNSD) plays a pivotal role in the coordination of the world population and housing census programme and, in 2017, the United Nations published Principles and Recommendations for Population and Housing Censuses (UN (2017)). In a similar vein, the Conference of European Statisticians Recommendations for the 2020 Censuses of Population and Housing was published by the United Nations Economic Commission for Europe (UNECE (2015)), providing a set of recommendations tailored specifically to the needs of European statisticians. Both documents provide guidance and assistance in the planning and execution of censuses and, among others, aim to facilitate improvements in the comparability of subnational data. Two different approaches are identified for the coding of housing or population units: the first is based on coding units to their lowest-level enumeration area, while the second is based on a coordinate or grid-based system. European countries were urged to adopt the use of grid data and identifiers for coordinate references so that the results of their next censuses could potentially provide a wide spectrum of spatial analyses. Applying the Degree of Urbanisation — 2021 edition  13 2 The legal and policy framework References Dijkstra, L. and H. Poelman (2014), ‘A harmonised definition of cities and rural areas: the new degree of urbanisation’, Regional Working Paper 2014, WP 01/2014, European Commission Directorate-General for Regional and Urban Policy. Dijkstra, L., H. Poelman and P. Veneri (2019), ’The EU-OECD definition of a functional urban area’, OECD Regional Development Working Papers, No. 2019/11, OECD Publishing, Paris. Eurostat (2019), Methodological manual on territorial typologies — 2018 edition, Publications Office of the European Union, Luxembourg. FAO (2018), Guidelines on defining rural areas and compiling indicators for development policy, Food and Agriculture Organization of the United Nations (FAO), Rome. ILO (2018), Rural-urban labour statistics, 20th International Conference of Labour Statisticians, International Labour Office, ICLS/20/2018/Room document 3/Rev. 3, Geneva. OECD (2012), Redefining “Urban”: A New Way to Measure Metropolitan Areas, Organisation for Economic Co-operation and Development, OECD Publishing, Paris. OECD (2020), Rural Well-Being: Geography of Opportunities, OECD Publishing, Paris. OECD and European Commission (2020), ‘Cities in the World: A New Perspective on Urbanisation’, OECD Urban Studies, OECD Publishing, Paris. UN (2015), Transforming our World: the 2030 Agenda for Sustainable Development, United Nations, General Assembly, A/ RES/70/1, New York. UN (2017), Principles and Recommendations for Population and Housing Censuses — Revision 3, ST/ESA/STAT/SER.M/67/ Rev.3, Department of Economic and Social Affairs, Statistics Division, United Nations, New York. UNECE (2015), Conference of European Statisticians Recommendations for the 2020 Censuses of Population and Housing, United Nations Economic Commission for Europe, United Nations, New York and Geneva. UN-Habitat (2017), New Urban Agenda, United Nations Conference on Housing and Sustainable Urban Development (Habitat III), United Nations, General Assembly, A/RES/71/256, New York. 14   Applying the Degree of Urbanisation — 2021 edition 3 Rationale and advantages Different countries use different criteria to define urban and rural areas which reflect their various perspectives as to what constitute urban and rural areas. It is clear that individual countries need to have their own national definitions that can be implemented in their statistical systems and used to disaggregate indicators by urban and rural areas for their own national policy purposes. Nonetheless, in order to have meaningful international comparisons of statistical indicators by urban and rural areas there is also an undisputed need for a definition that is nationally relevant and internationally comparable at the same time. Such a definition was lacking for international official statistics and international statistical standards. Without a harmonised global methodology, comparisons of the level of urbanisation and indicators for urban and rural areas were difficult to interpret as the differences in definitions could affect the results. The proposed solution was to develop a global definition of cities, urban and rural areas that could be used generally across the world based on the same delineation criteria for all regions/countries. This proposal should result in a harmonised and universal mapping of cities, towns and semi-dense areas and rural areas. Having internationally comparable statistical information is fundamental for solid evidence-based policymaking and measuring progress towards the sustainable development goals in both urban and rural areas. This new methodology has been designed not to replace national definitions, but to complement them with a definition that is both nationally relevant and internationally comparable. There are six clear advantages of the new methodology, namely that it: • captures the urban-rural continuum through three different classes at level 1 of the degree of urbanisation classification and through seven different classes at level 2 (see Chapter 6 and Chapter 7); • uses the same population size and density thresholds across the globe (see Chapter 6 and Chapter 7); • starts from a population grid to reduce the bias of using spatial units with different shapes and sizes (see Chapter 5); • measures population clusters directly instead of indirectly by using building clusters as an approximation of population clusters (see Chapter 6 and Chapter 7); • defines areas independently from their access to services to ensure that this access can be monitored reliably, in other words, without interference from the definition; • proposes a relatively cost-effective approach that can be applied to existing data collections (see Chapter 5, Chapter 9 and Chapter 10). 3.1 Captures the urban-rural continuum in harmonised manner The UN’s World Urbanization Prospects (1) presents data for urban areas and for rural areas. Many countries, however, use an approach with multiple classes to better capture the urban-rural continuum. For example, the 2011 census in India defined three types of urban areas: statutory towns, census towns and outgrowths. The United States census used urbanized areas, urban places outside of urbanized areas, and rural places and territory. The census in Portugal used predominantly urban areas, medium urban areas, and predominantly rural areas, while South Africa used three geography types: urban areas, rural areas and traditional areas. (1) United Nations Department of Economic and Social Affairs: Population Dynamics (https://population.un.org/wup/). Applying the Degree of Urbanisation — 2021 edition  15 3 Rationale and advantages The degree of urbanisation classifies the entire territory of a country along an urban-rural continuum. It combines population size and population density thresholds to capture three mutually exclusive classes: cities, towns and semi-dense areas, and rural areas (level 1 of the degree of urbanisation classification). Comparing level 1 of the degree of urbanisation classification to the traditional urban-rural dichotomy, and depending on the country under consideration, national definitions may include towns and semi-dense areas in an urban class or a rural class (see Figure 3.1). For example, the population of towns and semi-dense areas is almost entirely classified as urban according to the national definitions employed in Portugal, Brazil, France and the United States, while in Uganda and India the population of towns and semi-dense areas is generally classified as rural. By creating a separate class for areas where there is often no general agreement within national definitions, the degree of urbanisation classification proposes a compromise which acknowledges both approaches and enhances international comparability. There are two principal extensions to the methodology. The first (level 2 of the degree of urbanisation classification) provides a further breakdown for towns and semi-dense areas and for rural areas, with each divided into three separate subclasses (see Chapter 7). The second extension, defines functional urban areas (also referred to as metropolitan areas). These complement the degree of urbanisation classification by extending the concept of a city to include its surrounding commuting zone. This provides a more economic perspective of the urban-rural continuum. It can also be combined with level 1 of the degree of urbanisation classification to distinguish rural areas inside and outside a metropolitan area. Figure 3.1: Share of total population according to the degree of urbanisation classification and national urban-rural definitions, selected countries, mixed reference years (%) 100 90 80 70 60 50 40 30 20 10 0 Cities Rural areas Cities Rural areas Cities Rural areas Cities Rural areas Cities Rural areas Rural areas Rural areas Rural areas Rural areas Rural areas Rural areas Rural areas Towns and semi-dense areas Towns and semi-dense areas Towns and semi-dense areas Towns and semi-dense areas Towns and semi-dense areas Cities Towns and semi-dense areas Cities Towns and semi-dense areas Cities Towns and semi-dense areas Cities Towns and semi-dense areas Cities Towns and semi-dense areas Cities Towns and semi-dense areas Cities Towns and semi-dense areas Portugal Brazil France United Germay Mexico Spain Costa United South Uganda India States Rica Kingdom Africa National urban de nition National rural de nition Note: this figure shows for each degree of urbanisation the proportion of the population that was classified as urban and as rural according to national definitions. Countries are ranked by the share of their population living in towns and semi-dense areas that are classified as urban according to national definitions. Reference years vary from 2010 to 2018 depending on the selected country. 16   Applying the Degree of Urbanisation — 2021 edition Rationale and advantages 3 3.2 Uses the same population size and density thresholds across the globe National definitions often use very different population size and density thresholds (see Figure 3.2), which can potentially reduce the international comparability of the resulting data. The degree of urbanisation classification uses the same thresholds across the globe. These harmonised population size and density thresholds drew inspiration from national definitions: • out of the 103 countries that use a minimum population size threshold to define urban areas, 84 use a threshold of 5 000 inhabitants or less — this minimum threshold of 5 000 inhabitants was employed to define urban clusters; • Japan uses a minimum population size threshold of 50 000 inhabitants — this criterion was employed to define urban centres; • China and the Seychelles use a minimum population density threshold of 1 500 inhabitants per km² — this criterion was employed to define urban centres. Figure 3.2: Distribution of minimum population size thresholds used to define urban areas (count of countries) 35 30 25 20 15 10 5 China Japan Mali 0 0 5 000 10 000 15 000 20 000 25 000 30 000 35 000 40 000 45 000 50 000 55 000 60 000 65 000 70 000 75 000 80 000 85 000 90 000 95 000 100 000 Minimum population size threshold Source: UN World Urbanization Prospects Applying the Degree of Urbanisation — 2021 edition  17 3 Rationale and advantages An extensive sensitivity analysis was performed both on official grids in the EU and two global grids (GHS-POP (2) and WorldPop (3)). Using GHS-POP, the combination of a density threshold of 1 500 inhabitants per km² and a minimum population size threshold of 50 000 inhabitants identified at least one city in every country of the world that had at least 250 000 inhabitants (4), with Vanuatu as the only exception. Using GHS-POP, all small island developing states (SIDS) either have a city or a town. Those SIDS estimated to have a town with at least 5 000 inhabitants were Antigua and Barbuda, Dominica, Grenada, Kiribati, the Marshall Islands, Micronesia, Nauru, Palau, Saint Kitts and Nevis, Saint Lucia, Saint Vincent and the Grenadines, Samoa, Seychelles, Tonga, Tuvalu and Vanuatu. 3.3 Starts from a population grid to reduce the bias generated by the different shapes and sizes of spatial units More than half of the countries in the world have a national definition with a minimum population size threshold to classify urban areas. However, applying these thresholds to spatial units that differ in shape and size will influence the results and reduce international comparability. Furthermore, the application of national thresholds can lead to some small rural areas being classified as urban only because they are part of a large(r) administrative unit. For example, Plockton in Scotland has just 387 inhabitants, but it is part of the Highland Council which has more than 230 000 inhabitants. Using a population size threshold would classify Plockton as a rural area, but Highland Council as urban, while both are perceived as rural. To avoid classifying rural areas as urban, some national definitions add a population density requirement. However, a large city can have a very low population density if it is part of an administrative unit with a very large area. For example, Ulaanbaatar in Mongolia has 1.4 million inhabitants, but it has a relatively low population density of only 270 inhabitants per km2. This bias created by the shape and size of spatial units is called the modifiable areal unit problem. It can be addressed by using spatial units of the same shape and size, such as the population grid. Figure 3.3 shows how there is only one settlement identified when using the population density of the administrative units, whereas the population grid reveals that there are actually two settlements (circled in red) when analysed across identical spatial units (a grid). The method proposed here starts with a classification of a 1 km² population grid. This creates a classification which is independent from the administrative units of a country and is typically far more detailed. For example, the European Union has around 120 000 local administrative units, but more than 4 million grid cells of 1 km². Some national definitions are applied census enumeration areas, which are typically much smaller than local administrative units. As they are designed to capture roughly the same number of households, they tend to be (very) small in urban areas and (very) large in rural areas. As a result, the population density of enumeration areas will be higher in urban areas and lower in rural areas as compared with units of the same shape and size. For example, the Australian mesh block vary in size by a factor of one billion from 0.0001 km2 to more than 100 000 km2. Such large differences in size are bound to have a significant impact on population density figures and thus also on a definition that relies on population density. The benefit of using the grid is that all the cells have the same shape and size and their borders are stable over time. This produces a classification which is more comparable across space and more stable over time. The second step of this method classifies administrative or statistical spatial units, which reintroduces the problem of working with units of varying shapes and sizes. Therefore, it is recommended to use small administrative or statistical spatial units; this should ensure a good match with the grid classification. Applying this method to very large units, such as regions, may significantly alter population shares when compared with the grid classification. (2) A spatial raster dataset with the distribution and density of population, expressed in terms of the number of inhabitants per cell (https://ghsl.jrc.ec.europa.eu/data.php). (3) Spatially detailed information on the number of inhabitants mapped to administrative boundaries (https://www.worldpop.org/focus_areas). (4) Testing the degree of urbanisation at the global level (https://ghsl.jrc.ec.europa.eu/CFS.php). 18   Applying the Degree of Urbanisation — 2021 edition Rationale and advantages 3 Figure 3.3: Population density of administrative units and grid cells in Veenendaal, the Netherlands, 2011 (inhabitants per km²) ! ≤ 150 301 - 500 1 001 - 1 500 2 501 - 5 000 0 5 10 km 151 - 300 501 - 1 000 1 501 - 2 500 > 5 000 Source: Eurostat (GEOSTAT 2011) 3.4 Measures population clusters directly The United Nations Principles and Recommendations for Population and Housing Censuses (UN (2017)) defines a locality or settlement as a distinct population cluster (Section 1.8, p. 187). In the past, however, it was not possible to measure where people were clustered, although buildings were often mapped at a much higher spatial resolution than the population. For example, a cadastral map with the outline of each building has a very high spatial resolution and can be used to identify which buildings are within 200 m of each other. Population data, however, was only available at a much coarser spatial resolution. Therefore, some national and academic definitions used clusters of buildings to identify settlements. Today, however, far more precise information is available on the distribution and location of populations. With the advent of geo-coded censuses, geo-referenced population registers and high resolution population grids, the spatial resolution of population data has increased dramatically and allows the direct identification of population clusters. As a result, it is no longer necessary to approximate a population cluster by using a cluster of buildings. Measuring population concentrations directly makes them more comparable across different levels of (economic) development. Cities in high-income countries tend to have far more built-up area per inhabitant than cities in low- income countries (for example, because cities in high-income countries tend to have bigger houses, as well as more spacious offices and shops). Using only the built-up area to define cities would mean that a high-income country would have more cities and each city would be bigger (in terms of area) than for a low-income country, even if they had exactly the same urban structure in terms of population clusters. Measuring population concentrations directly also makes them more comparable over time. In many countries, the amount of built-up land grows faster than the size of the population. This means that over time, less and less people would be needed for a certain size built-up area to be reached. As a result, definitions based on built-up areas are likely to inflate the share of the urban population over time, whereas people-based definitions are not affected by this problem. Applying the Degree of Urbanisation — 2021 edition  19 3 Rationale and advantages 3.5 Defines areas to monitor access to services, not areas defined by access to services The sustainable development goals include multiple indicators that monitor access to services or infrastructure. Examples include indicators measuring access to electricity, safely managed drinking water, a mobile phone network and all-weather roads. To properly monitor access to these services in urban and rural areas, they should not be part of the definition of such areas. For example, if the definition of an urban area includes a criterion that everyone should have access to electricity, this would mean, by definition, that the entire urban population would have to have access. This would make it impossible to monitor access to electricity in urban areas, as some large and dense settlements lacking electricity would not be classified as urban areas. To avoid this problem, the degree of urbanisation does not use access to services or infrastructure as criteria. This means that it can be used to identify cities, towns and semi-dense areas, and rural areas that lack or have successfully acquired such a service. This can facilitate international policy exchanges on how to provide, for example, electricity to different types of areas. Furthermore, the degree of urbanisation does not use the share of agricultural employment for both conceptual and empirical reasons. Rather, the methodology is people-based and this means that settlements of the same size are consistently classified in the same way. If a maximum threshold for agricultural employment was employed as part of the methodology to identify different areas, then settlements with the same population size could be classified either as urban or rural, undermining the central principle of the methodology. Empirically, the share of employment in agriculture varies from more than 50 % to less than 1 % between different countries of the world. Using a fixed threshold for the share of agricultural employment would result in some countries being classified as entirely rural or entirely urban. This, in turn, would undermine the goal of facilitating international comparisons and measuring the sustainable development goals in a harmonised manner. Because, agricultural employment is not part of the methodology, it may be distributed across all three classes. For example, in the EU-27, some 6 % of the people working in agriculture live in cities, 24 % live in towns and semi- dense areas and the remaining 69 % in rural areas. The presence of agricultural employment outside rural areas should not be seen as a problem, but rather as a benefit of this method. For example, farmers living in cities, towns and semi-dense areas will have better access to markets, allowing them to focus on more perishable and higher value added produce. They may also have more opportunities to combine farming with working in a different economic sector. The United Nations Principles and Recommendations for Population and Housing Censuses (UN (2017)) mentions the lack of a single definition of urban and rural areas. It suggests that some countries may wish to use additional criteria including ’the percentage of the population engaged in agriculture, the general availability of electricity or piped water in living quarters and the ease of access to medical care, schools, recreation facilities and transportation’. The method presented here aims to fill the lack of a harmonised method to delineate cities, urban and rural areas. This method deliberately avoids the suggested additional indicators to ensure that a) settlements of the same size are classified in the same way and b) access to services can be monitored over time and space. 20   Applying the Degree of Urbanisation — 2021 edition Rationale and advantages 3 3.6 Proposes a cost-effective approach This method is highly cost-effective for two reasons. First, a population grid can be created for a relatively low cost using existing data. Second, compiling statistics by degree of urbanisation can be done through aggregating existing data. A population grid can be created using a geo-coded census or a geo-coded population register for little extra cost. These sources provide the exact location of the residents of a country. All that is further required is to add up the population per 1 km² grid cell and, if needed, treat the results to protect confidentiality. If the exact location of the population is not available, a population disaggregation grid can be created by combining the population of census enumeration areas with high resolution land use or land cover data; these data can be produced using remote sensing. Several organisations offer a free global layer, including the Global Human Settlement Layer (5). Compiling data according to the degree of urbanisation can be relatively simple. If, for example, in a household survey, the location of where respondents live or the small spatial unit in which they live is available, then responses can be aggregated accordingly to compile statistics according to the degree of urbanisation. As the degree of urbanisation classification often has a quite balanced population distribution across its three classes, surveys will generally have a sufficiently large sample in each of the classes to produce reliable results. Other types of data, such as administrative data, can also be aggregated and compiled according to the degree of urbanisation as long as they are collected for small spatial units. (5) Joint Research Centre, Global Human Settlement Layer (https://ghsl.jrc.ec.europa.eu). References UN (2017), Principles and Recommendations for Population and Housing Censuses — Revision 3 , ST/ESA/STAT/SER.M/67/ Rev.3, Department of Economic and Social Affairs, Statistics Division, United Nations, New York. Applying the Degree of Urbanisation — 2021 edition  21 4 How the principles of official statistics and classifications are fulfilled This chapter reviews the methodology that is used to compile statistics by degree of urbanisation according to the 10 principles specified in Best Practice Guidelines for Developing International Statistical Classifications (UN (2013)). • Conceptual basis: the degree of urbanisation classification relies on population density and size. Population size is also used in most national definitions of urban and rural areas. The functional urban area classification additionally uses commuting data, which is often used for national definitions of metropolitan areas. Each of these elements is clearly defined. Tests have shown that the methodology captures settlements of different sizes and economic relations between cities and their surrounding commuting zones. • Classification structures: the degree of urbanisation classification is hierarchical with two levels, the functional urban area classification has a single level. • Classification types: the methodology proposes two international reference classifications. As a result, the classifications may require some adaption to meet country specific conditions. There may be categories defined for international use which do not apply in country specific circumstances, or there may be country specific circumstances which are not catered for in the international reference classifications. In such cases, producers of statistics are advised to provide details of the correspondence linking country specific circumstances to the international classifications. • Mutual exclusivity: the classes at each level (levels 1 and 2) of the degree of urbanisation classification for both the grid cell and the small spatial unit classification and the functional urban area classification are mutually exclusive. • Exhaustiveness: levels 1 and 2 of the degree of urbanisation classification are exhaustive, in other words, they classify the entire territory of a country. The functional urban area classification is also exhaustive, insofar as it covers metropolitan and non-metropolitan areas that together make up the entire territory of a country. • Statistical balance: estimates based on the Global Human Settlement Layer (GHSL) population grid show that the classifications produce classes where the populations are not too disparate in size. As a result, they will allow for effective cross-tabulation of data. • Statistical feasibility: the classifications were kept simple so as to make them feasible to apply across all countries of the world. The degree of urbanisation classification requires a population grid, which has already been estimated globally. A growing number of countries have produced or are planning to produce such a grid. The functional urban area classification also requires commuting data, which are not widely available across countries. However, auxiliary data sources such as from mobile telephones or employment registers can help to fill this gap. • Classification units/statistical units: the classifications propose simple classes (such as cities, towns and semi- dense area, rural areas or metropolitan areas) which can be used with a wide variety of statistical units such as people, jobs, enterprises, buildings, farms, land use, and so on. • Time-series comparability: estimates based on the GHSL population grid show that data using the degree of urbanisation classification capture changes over time, but are not too volatile. Applying the Degree of Urbanisation — 2021 edition  23 4 How the principles of official statistics and classifications are fulfilled References and further information UN (2013), Best Practice Guidelines for Developing International Statistical Classifications, Expert Group on International Statistical Classifications, Department of Economic and Social Affairs, Statistics Division, United Nations, New York. UN (2014), Fundamental Principles of Official Statistics, United Nations, General Assembly, A/RES/68/261, New York. UNECE, ‘Part B Metadata Concepts, Standards, Models and Registries‘, Common Metadata Framework, online publication, United Nations, Geneva. 24   Applying the Degree of Urbanisation — 2021 edition 5 Constructing a population grid A population grid is a powerful tool: its main advantage is that it standardises reporting units. Population grids may be used to analyse issues that require a consistently high spatial resolution, such as access to public transport, exposure to flooding or patterns of urbanisation. Census enumeration areas provide a high level of spatial resolution in urban areas, but usually a much coarser resolution in rural areas, which makes them less suitable for this type of analysis. Because a population grid is so useful, a number of organisations are promoting their production and use, including the United Nations Global Geospatial Information Management (UN GGIM), the United Nations Population Fund (UNFPA) and the POPGRID Data Collaborative initiative (1). Population grids have a number of important advantages: • grid cells all have the same size allowing for easy comparison; • grids are stable over time (2); • grids integrate easily with other data (for example, meteorological or air quality data); • grid cells can be assembled to form areas reflecting a specific purpose and study area (mountain regions, water catchment areas, metropolitan areas). The first modern population grids were produced in Scandinavia based on geo-coded population registers in the 1970s. Today, over 30 countries have an official population grid, including Brazil and all the countries in the European Statistical System (ESS). In addition, a substantial number of countries have recently conducted a geo-coded census or are preparing one. Such a census can produce a high quality official population grid (see Subchapter 5.1). In the absence of a geo-coded census or population register, a disaggregation grid can be created by combining the population of census units (enumeration areas) with high-resolution land use data from national or global sources (see Subchapter 5.2). If census population data for an entire country are not available, models can estimate grid cell population data for areas not covered by the census (see Subchapter 5.3). Finally, a number of emerging sources of big data from mobile phones or social media can also be used to estimate a population grid, although these sources pose a number of issues of reliability and stability over time (see Subchapter 5.4). To apply the degree of urbanisation, the population grid needs to be turned into a population density grid. For cells that are entirely covered by land, the calculation for population density is simple in an equal area projection: for example, if the number of inhabitants living in a 1 km² grid cell is 100, the population density is simply 100 inhabitants per km². However, for grid cells that are partially covered by water, the share of land in the total (surface) area needs to be calculated to adjust the population density. This can be done by combining the grid with a GIS layer identifying rivers, lakes and seas. (1) POPGRID Data Collaborative initiative (https://www.popgrid.org/). (2) Grids can be kept stable for future data collections, but it is difficult to construct reliable population grids for the past. Applying the Degree of Urbanisation — 2021 edition  25 5 Constructing a population grid 5.1 A grid based on the aggregation of point data Ideally, a population grid is based on a geo-referenced point dataset with a high spatial accuracy (see Figure 5.1). This guarantees a high quality grid and avoids any need for estimations or disaggregations. These points can be derived from a variety of sources. A growing number of countries have or will conduct a digital census where the exact geographical location of each household is recorded (3). Countries with a geo-coded cadastre, a building register or an address register can use these to generate a set of points with population data. Once the point data have been created, they can simply be aggregated to square grid cells. Figure 5.1: Example of point-based data overlaid on a statistical geo-coded grid of 1 km² (left) and population counts in shades of orange according to population density per 1 km² cell (unpopulated grid cells in white) for aggregated point-based information (right) A B C D A B C D 1 1 1kmN4627E5033 1kmN4627E5034 1kmN4627E5035 1kmN4627E5036 1kmN4627E5033 1kmN4627E5034 1kmN4627E5035 1kmN4627E5036 2 2 1kmN4626E5033 1kmN4626E5034 1kmN4626E5035 1kmN4626E5036 1kmN4626E5033 1kmN4626E5034 1kmN4626E5035 1kmN4626E5036 3 3 1kmN4625E5033 1kmN4625E5034 1kmN4625E5035 1kmN4625E5036 1kmN4625E5033 1kmN4625E5034 1kmN4625E5035 1kmN4625E5036 4 4 1kmN4624E5033 1kmN4624E5034 1kmN4624E5035 1kmN4624E5036 1kmN4624E5033 1kmN4624E5034 1kmN4624E5035 1kmN4624E5036 The exact location of each household is considered confidential. However, aggregating these data to grid cells of 1 km2 is often sufficient to address confidentiality concerns. Some countries also apply a limited amount of record swapping to provide an even higher guarantee of confidentiality (Eurostat (2019) and GEOSTAT 1B (4)). (3) United Nations Statistics Division, Guidelines on the use of electronic data collection technologies in population and housing censuses (https://unstats.un.org/unsd/demographic/standmeth/handbooks/data-collection-census-201901.pdf). (4) European Forum for Geography and Statistics (EFGS), GEOSTAT 1B (https://www.efgs.info/geostat/1B/). 26   Applying the Degree of Urbanisation — 2021 edition Constructing a population grid 5 5.2 A grid based on the disaggregation of population data In the absence of point data, a population grid can be produced by disaggregating population data from census enumeration areas or administrative units (such as municipalities, districts or provinces) using auxiliary data with a higher spatial resolution, such as land cover or built-up area data, that are linked to the presence of people (see Figure 5.2). In a disaggregation grid, the total population of a census unit or Figure 5.2: Simplified workflow for population grid administrative unit is distributed across the grid cells covering that creation by disaggregation of existing counts unit based on other data that are linked to the presence of people. This disaggregation can be done in a variety of ways. The simplest method relies on a single covariate and allocates the population Population count Spatial proportionally to that covariate. GHS-POP R2019A (Freire et al. (2016); by census unit covariate(s) Schiavina et al. (2019)) is a good example of such an approach (5). A slightly more complex method uses multiple covariates. For example, the population may be allocated proportionally to all built-up areas with the exception of non-residential areas and roads Combine and railways. The European Settlement Map (Corbane and Sabo (2019); Corbane et al. (2020)) is an example that distinguishes between Population grid residential and non-residential buildings (6). grid A more complex method uses multiple co-variates combined with a ‘random forest’ estimation technique to determine the weights to distribute the population. WorldPop (Tatem (2017)) is a good example of such an approach (7). Regardless of the disaggregation method selected, two key issues will determine the quality of the resulting population grid. First, the size (area) of the units for which population data are available: the smaller the spatial unit, the higher the quality of the grid. Second, the quality of the covariate: a covariate that is closely linked to the presence of people and that avoids errors of omission and commission will produce a higher quality grid. For example, a geospatial layer of built-up areas or building footprints with high spatial resolution is considered to be highly suitable for such a purpose. Such sources are often based on remote sensing, which may not detect all built-up areas or buildings (omission) or may mistakenly identify some areas as built-up or as covered by a building (commission). Several organisations offer open access global layers based on remote sensing data, including the Global Human Settlement Layer (GHSL) produced by the European Commission’s Joint Research Centre (JRC). To allocate proportionally the population within a census unit based on a single covariate involves a number of steps that are presented in Figure 5.3. The first map shows a census unit and its population (p). The second map shows the boundary of this census unit rasterised using a 250 m grid. Through this process each 250 m cell is assigned to one and only one census unit (8). This process can also be done at a finer resolution (100 m or smaller) to ensure a closer match between the original census unit and the assigned cells, although this requires a more powerful computer. The third map shows the built-up areas (b), which are mapped at 30 m resolution in binary fashion, in other words, built-up or not. The fourth map shows, for each 250 m cell, the built-up area within that cell as a share of the total built-up area within the census unit (b % = b in cell / b in census unit). The fifth map shows the population that has been allocated proportionally based on the share of the built-up area (POPcell = p * b %). Because the sum of the shares of built-up areas in all the cells in a census unit is 100 %, the sum of the population in these cells will exactly match the population of the census unit. The sixth map shows the population for a set of 1 km grid cells (in yellow). Note that the sum of the three 1 km² grid cells (113 people) is higher than the population of the census unit (104 people) because these three grid cells include the population of a few 250 m cells that belong to neighbouring census units. (5) Joint Research Centre, Global Human Settlement Layer (https://ghsl.jrc.ec.europa.eu). (6) Copernicus, European Settlement Map (https://land.copernicus.eu/pan-european/GHSL/european-settlement-map). (7) WorldPop (https://www.worldpop.org/). (8) With the exception of census units that do not have a raster equivalent; the population of these units can be distributed across the cells with which it intersects. Applying the Degree of Urbanisation — 2021 edition  27 5 Constructing a population grid Figure 5.3: Example of the process used to generate the GHS-POP layer (extract from a location in France) 104 People BU: 3.5% BU: BU: 12.3% 8.8% BU: BU: 31.6%24.5% BU: 3.5% BU: 15.8% POP: 3.7 POP: POP: 12 80 11.8 9.7 POP: POP: 33.4 24.5 POP: 4.5 21 POP: 16.2 Note: Esri, HERE, Garmin, Intermap, increment P Corp., NPS, GeoBase, IGN, METI, © OpenStreetMap contributors and the GIS User Community. Processed by JRC. 28   Applying the Degree of Urbanisation — 2021 edition Constructing a population grid 5 The GHS-POP (Freire et al. (2016); Schiavina et al. (2019)) is produced in this way. It disaggregates residential population estimates for four target years using the best available census units, adjusted to UN WPP estimates (the population input is the Gridded Population of the World v4.10 (CIESIN (2018)). The disaggregation is done using the built-up areas as detected by the GHSL. 5.3 Extrapolating a population grid based on a partial micro-census Comprehensive and accurate population data for small areas can be costly and logistically challenging to collect, but they represent a fundamental basis for government decision and policymaking. In resource-constrained settings, national population and housing census data can be outdated, inaccurate, or missing specific groups, while registry data can be lacking or incomplete. In addition, certain areas of a country may not be included in national data collections due to conflict, inaccessibility or cost limitations. In such cases, a different approach is needed to produce a complete population grid. When a geo-referenced census is not available or it is considered Figure 5.4: Simplified workflow for unsuitable due to a lack of completeness, freshness, or reliability, a different population grid creation in the absence of approach can be employed to create a population grid. This technique census counts is more challenging as it does not start from pre-existing population counts for the entire country; instead, the total is estimated using a Micro-census Spatial population distribution model. Such an approach requires the availability pop. counts covariate(s) of detailed and reliable data from a micro-census or survey which does not cover the entire country to develop a model. This technique estimates a count — at the level of grid cells — through combining sampling with ancillary data, typically remotely-sensed (for example, the density of Sampling buildings, urban areas). Given such a spatial covariate covering the whole country and surveys (micro-census) for a subset of the country, these data are combined to derive parameters or weights in a statistical model Weights Combine characterising the population’s distribution. This model is then used to predict the population’s distribution in non-surveyed areas (see Figure 5.4) under the assumption that the surveyed area is representative of the whole area. Population grid Recent advances in the availability of detailed satellite imagery, geo- positioning tools for field surveys, statistical methods and computational power are providing opportunities to complement traditional collection methods for data on population by modelling and estimation into areas that were missed from enumeration (Wardrop et al. (2018)). Bayesian geostatistical modelling approaches to predict population numbers and age/sex structures from small area micro- census surveys, or incomplete census enumeration, have been developed and applied for multiple countries where instability, funding or other obstacles have limited recent national data collection exercises. Using a set of spatially complete datasets as covariates, including satellite-derived building footprints, along with a spatial covariance structure makes it possible for models to predict population by age and sex in unobserved areas across a country, together with associated uncertainty metrics (Wardrop et al. (2018)). Cross-validation typically shows high model accuracies at subnational levels (9). This technique has the potential to fill gaps where enumeration could not be undertaken and to provide contemporary, regularly-updated and accurate population information to support decision-making and development in challenging contexts (10). Datasets built using these approaches for Nigeria, Zambia and the Democratic Republic of the Congo are available from WorldPop (11). (9) For example, the United Nations Population Fund (https://www.unfpa.org/resources/new-methodology-hybrid-census- generate-spatially-disaggregated-population-estimates). (10) For example, the United Nations Population Fund or GRID3 (https://grid3.org/solution/high-resolution-population- estimates). (11) WorldPop Open Population Repository (https://wopr.worldpop.org/). Applying the Degree of Urbanisation — 2021 edition  29 5 Constructing a population grid 5.4 Alternative and emerging data sources for creating population grids In recent years, a number of emerging data sources and technologies have been explored for direct mapping of the population or as alternative proxies for its disaggregation; at present, this work has mainly been carried out as a proof-of-concept. Examples include data from mobile phones (Deville et al. (2014)), crowdsourcing/volunteered geographic information (Bakillah et al. (2014)) and location-based social media (Aubrecht et al. (2011) and (2017)). For example, in countries with a high mobile phone penetration rate and many mobile phone towers, the night- time location of mobile phones could be used to generate a high-resolution population grid. Some promising approaches involve the integration of conventional with unconventional data sources, for example, combining official statistics with big data from remote sensing, volunteered geographic information, social media and mobile phones (Aubrecht et al. (2018)). However promising, there are a number of issues concerning these types of data and technologies, for example, the sustainability of such approaches, data access and ownership, privacy and anonymity of social media users, or representation bias (Zhang and Zhu (2018)). The main challenge for developers is how to scale-up highly localised approaches to wide geographical areas (continents, the world) to provide datasets that are open and free (in a sustainable way). Given these as yet unsolved challenges, such data cannot currently be used as a reliable substitute for an official population and housing census that — in addition to complying with strict technical and statistical specifications — collects a wealth of additional information on population characteristics and living conditions. 30   Applying the Degree of Urbanisation — 2021 edition Constructing a population grid 5 References Aubrecht, C., D. O. Aubrecht, J. Ungar, S. Freire and K. Steinnocher (2017), ‘VGDI — advancing the concept: volunteered geo-dynamic information and its benefits for population dynamics modeling’, Transactions in GIS, Volume 21, Issue 2, pp. 253-276. Aubrecht, C., J. Ungar and S. Freire (2011), ‘Exploring the potential of volunteered geographic information for modeling spatio-temporal characteristics of urban population: a case study for Lisbon metro using foursquare check-in data’, Proceedings of the 7th International Conference on Virtual Cities and Territories, pp. 57-60. Aubrecht, C., J. Ungar, D. O. Aubrecht, S. Freire and K. Steinnocher (2018), ‘Mapping land use dynamics using the collective power of the crowd’, Earth Observation Open Science and Innovation, ISSI Scientific Report Series, Volume 15, pp. 247-253. Bakillah, A., S. Liang, A. Mobasheri, J. J. Arsanjani and A. Zipf (2014), ‘Fine-resolution population mapping using OpenStreetMap points-of interest’, International Journal of Geographical Information Science, Volume 28, Issue 9, pp. 1 940-1 963. Center for International Earth Science Information Network (CIESIN), Columbia University (2018), Documentation for the Gridded Population of the World, Version 4 (GPWv4), Revision 11 Data Sets, NASA Socioeconomic Data and Applications Center (SEDAC), Palisades, NY. Corbane, C. and F. Sabo (2019), European Settlement Map from Copernicus Very High Resolution data for reference year 2015, Public Release 2019, European Commission, Joint Research Centre (JRC). Corbane, C., F. Sabo, V. Syrris, T. Kemper, P. Politis, M. Pesaresi, P. Soille and K. Osé (2020), ‘Application of the Symbolic Machine Learning to Copernicus VHR Imagery: the European Settlement Map’, IEEE Geoscience and Remote Sensing Letters, Volume 17, Issue 7, pp.1 153-1 157. Deville, P., C. Linard, S. Martin, M. Gilbert, F. R. Stevens, A. E. Gaughan, V. D. Blondel and A. J. Tatem (2014), ‘Dynamic population mapping using mobile phone data’, Proceedings of the National Academy of Sciences of the United States of America, Volume 111, No. 45, pp. 15 888-15 893. Eurostat (2019), Methodological manual on territorial typologies — 2018 edition, Publications Office of the European Union, Luxembourg. Freire, S., K. MacManus, M. Pesaresi, E. Doxsey-Whitfield and J. Mills (2016), Development of new open and free multi- temporal global population grids at 250 m resolution, Conference paper for AGILE 2016 — Helsinki, June 14-17, 2016, Association of Geographic Information Laboratories in Europe (AGILE). Schiavina, M., S. Freire and K. MacManus (2019), GHS-POP R2019A — GHS population grid multitemporal (1975, 1990, 2000, 2015), European Commission, Joint Research Centre (JRC). Tatem, A. (2017), ‘WorldPop, open data for spatial demography’, Scientific Data 4, Article No. 170004. Wardrop, N. A., W. C. Jochem, T. J. Bird, H. R. Chamberlain, D. Clarke, D. Kerr, L. Bengtsson, S. Juran, V. Seaman and A. J. Tatem (2018), ‘Spatially disaggregated population estimates in the absence of national population and housing census data’, Proceedings of the National Academy of Sciences, Volume 115, No. 14, pp. 3 529-3 537. Zhang, G. and A-X. Zhu (2018), ‘The representativeness and spatial bias of volunteered geographic information: a review’, Annals of GIS, Volume 24, Issue 3, pp. 151-162. Applying the Degree of Urbanisation — 2021 edition  31 6 Methodology for applying level 1 of the degree of urbanisation classification This chapter presents the key methodological recommendations on how to apply level 1 of the degree of urbanisation classification, which is the recommended level for a territorial classification of indicators on sustainable development goals. 6.1 Terminology Two sets of terms have been developed to describe level 1 of the degree of urbanisation classification. The first set uses short and simple terms such as cities and rural areas. The second set uses more technical and neutral language. The second set can be helpful to avoid overlaps with the terms used in national definitions. Table 6.1: Short and technical terms for classifying grid cells by degree of urbanisation Short terms Technical terms Urban centres High-density clusters Urban clusters Moderate density clusters Rural grid cells Mostly low-density cells Small spatial units can be administrative units — such as municipalities — or statistical areas — such as census units (enumeration areas). Table 6.2: Short and technical terms for classifying small spatial units by degree of urbanisation Short terms Technical terms Cities Densely populated areas Towns and semi-dense areas Intermediate density areas Rural areas Thinly populated areas 6.2 Short description Level 1 of the degree of urbanisation classifies small spatial units as (i) cities or densely populated areas, (ii) towns and semi-dense areas or intermediate density areas and (iii) rural areas or thinly populated areas. This is done using 1 km² grid cells, classified according to their population density, population size and contiguity (neighbouring cells). Each small spatial unit belongs exclusively to one of these three classes. Urban areas consist of cities plus towns and semi-dense areas. Because level 1 of the degree of urbanisation classification was developed to capture the urban-rural continuum, it is recommended to report indicators for all three classes instead of only for the urban-rural dichotomy. This is important because towns and semi-dense areas may differ significantly both from cities and from rural areas. Semi-dense areas in low- and middle-income countries are often described as peri-urban areas. In high-income countries, they are usually described as suburbs. In both cases, these areas have a moderate density and are at the transition between a rural area and a city or town. Within national statistical systems, there is generally a high level of agreement concerning the two outermost classes: cities are typically classified as being urban, while villages and sparsely-populated areas are typically classified as being rural. By contrast, the classification of intermediate areas is less clear-cut: some countries prefer to classify them as urban, Applying the Degree of Urbanisation — 2021 edition  33 6 Methodology for applying level 1 of the degree of urbanisation classification others as rural, with a third group of countries choosing to create an intermediate class between these two extremes. The degree of urbanisation classification tries to accommodate these intermediate areas and different points of view to emphasise that towns and semi-dense areas are partway between a city and a rural area. This is important because policymaking that is uniformly applied across the three classes may not be suitable and could benefit from being tailored to the specific requirements of cities, towns and semi-dense areas or rural areas. 6.3 Grid cell classification The basis for the degree of urbanisation classification is a 1 km² population grid (for more details on how to construct a population grid, see Chapter 5). Each grid cell has the same shape and surface area, thereby avoiding distortions caused by using units varying in shape and size. This is a considerable advantage when compared with alternative approaches such as those based on the use of population data for local administrative units (for example municipalities). The use of relatively small (1 km²) and uniform grid cells means that the basic concept underlying the methodology is to look inside larger local administrative units to detect the presence of individual cities, towns and semi-dense areas as well as rural areas. This makes it possible to create a more accurate classification. Grid cells of 1 km² were selected instead of smaller cells for two reasons. They strike a balance between spatial detail, availability of official data, concerns about confidentiality and computational complexity. For example, grid cells of 1 km² have been used by many national statistical authorities with few or no confidentiality concerns and can be processed by a regular desktop computer. Although a grid composed of cells that are 100 m by 100 m would provide more spatial detail, this would also increase the number of cells one hundred-fold. In addition, the method would have to be modified for two reasons. First, smaller cells would follow a different, more skewed population density distribution. Second, using a higher resolution grid could lead to the fragmentation of single settlements. A small linear park could be enough to split a settlement into two parts, which could also result in it falling below the population size threshold. Understanding contiguous cells Before looking at the identification of the three cluster types, it is necessary to understand the concept of contiguous cells. Figure 6.1 shows an array of nine grid cells, with the focus on the central cell which is surrounded by eight others, numbered 1 to 8. Figure 6.1: contiguous grid cells 1 2 3 4 5 6 7 8 Two types of contiguous grid cells can be identified: (i) four-point contiguity, which is a narrower definition excluding diagonals — all cells that touch each other excluding those cells that only touch each other on a diagonal; only cells numbered 2, 4, 5 and 7 are contiguous to the central cell in Figure 6.1 according to this narrower definition. (ii) eight-point contiguity, which is a broad definition including diagonals — all cells that touch each other in any way, including cells that are linked only on a diagonal; all cells numbered 1 to 8 are contiguous to the central cell in Figure 6.1 according to this broader definition. 34   Applying the Degree of Urbanisation — 2021 edition Methodology for applying level 1 of the degree of urbanisation classification 6 Stage 1: classifying grid cells Each cluster type is identified by classifying 1 km² population grid cells according to characteristics that are based on their total population and population density. Groups of 1 km² population grid cells are plotted in relation to their neighbouring cells to identify: • An urban centre (high-density cluster) — a cluster of contiguous grid cells of 1 km² (using four-point contiguity, in other words, excluding diagonals) with a population density of at least 1 500 inhabitants per km² and collectively a minimum population of 50 000 inhabitants before gap-filling; if needed, cells that are at least 50 % built-up may be added (see Subchapter 8.2.1). • An urban cluster (moderate-density cluster) — a cluster of contiguous grid cells of 1 km² (using eight-point contiguity, in other words, including diagonals) with a population density of at least 300 inhabitants per km² and a minimum population of 5 000 inhabitants. In a final step, the grid cells identified as an urban centre are removed from the urban cluster. • Rural grid cells (or mostly low-density cells) — grid cells that are not identified as urban centres or as urban clusters. Note that in Eurostat’s Methodological manual on territorial typologies — 2018 edition (Eurostat (2019)), a grid cell can belong to an urban centre and an urban cluster. Applying the Degree of Urbanisation proposes a different approach, whereby every cell is allocated to one and only one class, by excluding those cells that belong to urban centres from urban clusters. This difference has no impact on the classification of small spatial units. Rather, the benefit of this mutually exclusive grid layer is that it will closely match the classification of spatial units and also adheres to the guidelines for an international reference classification. 6.3.1 URBAN CENTRES (HIGH-DENSITY CLUSTERS) The identification of urban centres (high-density clusters) is done in three steps. The first step involves identifying groups of contiguous cells: • all cells with a population density of at least 1 500 inhabitants per km² are selected (light blue shading in Figure 6.2); • groups of contiguous grid cells are identified (groups G1 and G2 in Figure 6.2). If available, cells that are at least 50 % built-up can be added (see Subchapter 8.2.1). Contiguous cells are grouped together, however, when identifying urban centres diagonal contiguity is excluded. As such, in the example of Figure 6.2, cells C2 and D3 are not considered as contiguous; rather, they are each part of different groups (G1 and G2). Figure 6.2: Contiguous groups for urban centres A B C D E F A B C D E F 1 15 000 16 500 5 000 1 G1 G1 G1 2 15 000 6 000 2 G1 G1 3 15 000 18 500 2 500 3 500 3 G1 G1 G2 G2 4 15 500 7 000 4 G1 G2 Population ≥ 1 500 inhabitants/km² G1 Group 1 of contiguous cells Population < 1 500 inhabitants/km² G2 Group 2 of contiguous cells Applying the Degree of Urbanisation — 2021 edition  35 6 Methodology for applying level 1 of the degree of urbanisation classification In a second step, each group of contiguous grid cells is analysed in relation to its total number of inhabitants and only those groups of contiguous cells with collectively at least 50 000 inhabitants are selected (see Figure 6.3). Continuing with the same example, Group G1 is considered an urban centre as it has a population of 106 500 inhabitants, as shown in Figure 6.3, while G2 is not an urban centre as its population is only 13 000 inhabitants. Figure 6.3: Identifying urban centres A B C D E F A B C D E F Population 1 106 500 1 Urban centre 2 2 Population 3 13 000 3 4 4 The third step for identifying urban centres is taken to fill gaps and smooth borders. This is done by applying an iterative majority rule. This rule is applied to individual urban centres (1): in other words, only cells for a particular urban centre and not the cells of other nearby urban centres are taken into account. In some cases, urban centres can become contiguous due to the majority rule, but they should not be combined and should remain as two separate entities. The majority rule was introduced to address several issues. It adds areas that have a lower population density (but are surrounded by densely populated neighbourhoods) and are likely to be heavily used during the day-time by city residents. These areas include industrial and commercial areas, transport hubs, parks and urban forests. The majority rule generates areas that are more suitable for monitoring the sustainable development goal indicators. For example, to measure the share of urban green areas, these areas should (ideally) be included within the urban centre or to measure the total area that needs to be served (or crossed) by public transport lines, industrial and commercial areas, parks and urban forests should also be included. The majority rule fills such gaps in urban centres (2) and produces a shape that is more rounded / lacks sharp angles. As a result, urban centres that have been modified to fill gaps and smooth borders are more likely to include transport lines that connect different parts of the urban centre. The iterative ‘majority rule’ If five or more of the (eight) cells surrounding a particular cell belong to the same unique urban centre, then that cell is also considered to belong to the same urban centre; this process is repeated (iteratively) until no more cells may be added. Note that the criterion for gap-filling following the majority rule includes cells that are linked only on a diagonal. For example, cell B2 on the left-hand side of Figure 6.3 has seven of its eight surrounding cells that belong to the same urban centre. This cell should therefore subsequently be added to the urban centre to smooth borders (as shown on the right-hand side of Figure 6.3). (1) When two or more urban centres are located close together, the outcome of the majority rule may lead to different results depending on which urban centre is treated first. The DUG tool (see Chapter 10) identifies all cells that could be allocated to more than one urban centre. It attributes the cells to one urban centre if the majority rule considering all urban centres leads to a single allocation. The remaining cells are not attributed to any urban centre. This ensures consistency in terms of how cells are allocated. (2) In some cases, a large rectangular gap will not be filled by the majority rule. Note that the DUG tool (Chapter 10) fills all gaps that are smaller than 15 km². 36   Applying the Degree of Urbanisation — 2021 edition Methodology for applying level 1 of the degree of urbanisation classification 6 6.3.2 URBAN CLUSTERS (OR MODERATE-DENSITY CLUSTERS) The technique used to identify urban clusters (moderate-density clusters) is similar to that used for urban centres (high-density clusters). Rather than using a threshold of at least 1 500 inhabitants per km², the identification of urban clusters is based on grid cells with a population density of at least 300 inhabitants per km² (see Figure 6.4). The initial identification of urban clusters is done in two steps: • all cells with a population density of at least 300 inhabitants per km² are plotted (light blue shading in Figure 6.4); • groups of contiguous grid cells are identified (groups G1 and G2 in Figure 6.4); note that contiguous grid cells may include cells that are linked only on a diagonal (eight-point contiguity) — as shown, for example, by cell C2. Figure 6.4: Contiguous groups for urban clusters A B C D E F A B C D E F 1 400 550 2 100 1 G1 G2 G2 2 500 1 000 400 2 G1 G1 G2 3 1 500 350 3 G1 G1 4 2 000 1 250 4 G1 G1 Population ≥ 300 inhabitants/km² G1 Group 1 of contiguous cells Population < 300 inhabitants/km² G2 Group 2 of contiguous cells Thereafter, each group of contiguous grid cells is analysed in relation to its number of inhabitants and those groups of contiguous cells with collectively at least 5 000 inhabitants are selected; these are urban clusters. Note that if there are cells that are also part of an urban centre they are removed. Continuing with the same example, Group G1 is considered an urban cluster as it has a population of 7 000 inhabitants, as shown in Figure 6.5, while G2 is not an urban cluster as its population is only 3 050 inhabitants. Figure 6.5: Identifying urban clusters A B C D E F A B C D E F Population 1 3 050 1 2 2 3 3 Population Urban 7 000 cluster 4 4 Applying the Degree of Urbanisation — 2021 edition  37 6 Methodology for applying level 1 of the degree of urbanisation classification Figure 6.6 shows a schematic overview from grid cell classification through to the identification of urban centres. In the first image, grid cells with a population density of at least 300 inhabitants per km² are identified. The second image overlays these grid cells showing urban clusters (moderate-density clusters) that are composed of contiguous grid cells linked by eight-point contiguity and at least 5 000 inhabitants before removing any cells that are also part of an urban centre. The final image shows the urban cluster cells after the removal of the urban centre cells and the urban centre — a set of contiguous grid cells that have a population density of at least 1 500 inhabitants per km² and at least 50 000 inhabitants (before applying the iterative ‘majority rule’). Figure 6.6: Schematic overview of identifying urban clusters and urban centres Grid cells with ≥ 300 inhabitants/km² Urban clusters (moderate-density clusters) Urban centres (high-density clusters): before excluding the urban centre cells: a a cluster of contiguous grid cells of 1 km² with a Local administrative unit (LAU) cluster of contiguous grid cells of 1 km² with a density of at least 1 500 inhabitants per km² and boundaries density of at least 300 inhabitants per km² and a minimum population of 50 000 inhabitants a minimum population of 5 000 inhabitants after gap-filling Urban clusters (moderate-density clusters) after excluding the urban centre cells: a cluster of contiguous grid cells of 1 km² with a density of at least 300 inhabitants per km² and a minimum population of 5 000 inhabitants Source: Eurostat, JRC and European Commission, Directorate-General Regional and Urban Policy and Directorate-General Agriculture and Regional Development 38   Applying the Degree of Urbanisation — 2021 edition Methodology for applying level 1 of the degree of urbanisation classification 6 6.3.3 RURAL GRID CELLS Rural grid cells are those cells that are not identified as urban centres or as urban clusters. The majority of rural grid cells have a population density that is less than 300 inhabitants per km², although this is not necessarily the case. Some rural grid cells may have a higher number of inhabitants if they do not form part of a cluster that meets the criteria for an urban centre or an urban cluster. In Figure 6.7, cells A3, B4 and F1 each meet the population criterion for an urban centre (at least 1 500 inhabitants per km²), while cells B3, C2 and E1 each meet the population criterion for an urban cluster (at least 300 inhabitants per km²). Figure 6.7: Detecting rural grid cells A B C D E F A B C D E F 1 550 2 100 1 G1 G1 2 450 2 G2 3 1 500 350 3 G2 G2 4 1 600 4 G2 Population ≥ 300 inhabitants/km² G1 Group 1 of contiguous cells Population < 300 inhabitants/km² G2 Group 2 of contiguous cells Each group of contiguous grid cells (groups G1 and G2 in the right-hand side of Figure 6.7) may be analysed in relation to their total number of inhabitants and those groups of contiguous cells with collectively at least 5 000 inhabitants are selected. In Figure 6.8, it can be seen that neither group G1 with a total population of 3 900 inhabitants nor group G2 with a total population of 2 650 inhabitants reaches the population threshold for an urban cluster. As such, each cell in these two groups is classified as a rural grid cell, as shown on the right-hand side of Figure 6.8. Figure 6.8: Identifying rural grid cells A B C D E F A B C D E F Population 1 2 650 1 2 2 Population 3 3 900 3 4 4 Rural grid cells Note also, as mentioned above, that it is possible for grid cells with a population density of less than 300 inhabitants per km² to be classified as part of an urban centre, due to gap-filling or as a result of adding cells that are at least 50 % built-up (see Subchapter 8.2.1). Applying the Degree of Urbanisation — 2021 edition  39 6 Methodology for applying level 1 of the degree of urbanisation classification 6.4 Classifying small spatial units Stage 2: classifying small spatial units by degree of urbanisation Once all grid cells have been classified and urban centres, urban clusters and rural grid cells identified, the next step concerns overlaying these results onto small spatial units, as follows: • cities (or densely populated areas) — small spatial units that have at least 50 % of their population in urban centres; • towns and semi-dense areas (or intermediate density areas) — small spatial units that have less than 50 % of their population in urban centres and no more than 50 % of their population in rural grid cells; • rural areas (or thinly populated areas) — small spatial units that have more than 50 % of their population in rural grid cells. Cities (or densely populated areas) consist of one or more small spatial units with at least 50 % of their population in an urban centre. A small spatial unit can be either an administrative unit or a statistical area. Examples of administrative units include a municipality, a district, a neighbourhood or a metropolitan area. Some of these administrative units also have a political role as electoral districts or in terms of local government. Statistical areas can be census units/enumeration areas, census blocks, census tracts, wards, super output areas, named places or small areas. In some countries, not all the small spatial units contain inhabitants. To classify the spatial units without any population, the same rules should be applied to their area instead of to their population. For example, a small spatial unit without any population that has more than 50 % of its area in rural grid cells should be classified as a rural area. Map 6.1 shows the grid cell classification for Durban in South Africa and Map 6.2 shows the classification of small spatial units. Map 6.1: Grid cell classification around Map 6.2: Classification of small spatial units around Durban, South Africa Durban, South Africa Umzinyathi Umzinyathi Umzinyathi Umzinyathi Umzinyathi Umzinyathi Umzinyathi Umzinyathi iLembe iLembe UMgungundlovu UMgungundlovu Legend Legend eThekwini Municipalities eThekwini Municipalities district 2011 district 2011 Durban GHS-SMOD grid GHS-SMOD grid Urban centre Urban centre Urban cluster Urban cluster Rural grid cells Rural grid cells Sisonke Water Sisonke Water Ugu km Ugu km 0 10 20 40 0 12.5 25 50 Source: Florczyk et al. (2019) Note that each small spatial unit should be classified to one and only one of the three classes within level 1 of the degree of urbanisation classification. However, in order to classify small spatial units based on the population grid, these units have to be transformed into a raster as well, which can lead to some situations which require case-by- case solutions (see Subchapter 7.2.4 and Chapter 8 for more information on different types of adjustments that may be made). 40   Applying the Degree of Urbanisation — 2021 edition Figure 6.9: Schematic overview of the degree of urbanisation classification nals) nals and with gap filling) Contiguous cells (including diago Contiguous cells (without diago inhabitants/km² inhabitants/km² with a density of at least 300 with a density of at least 1 500 Grid cells itants; urban itants and a minimum of 5 000 inhab and a minimum of 50 000 inhab urban clusters centres are removed from the Raster cells of 1 km² are classified ers Grid cells outside urban clust using criteria of population density and contiguity. The population per grid cell is ideally derived from a geo-coded census or population register. Alternatively, it can be estimated by downscaling population data from larger spatial units using covariates with a high spatial resolution Applying the Degree of Urbanisation — 2021 edition  At least 50 % of population cells and living in urban centres Small spatial units < 50 % of population in rural grid centres At least 50 % of population < 50 % of population in urban living in rural grid cells The degree of urbanisation is a classification of small spatial units based on the share of population living in urban centres, urban clusters and rural grid cells. Thinly populated areas Intermediate density areas Densely populated areas Note: for more information, see http://ec.europa.eu/regional_policy/sources/docgener/work/2014_01_new_urban.pdf. Source: Directorate-General Regional and Urban Policy, based on data from Eurostat, JRC, national statistical authorities Methodology for applying level 1 of the degree of urbanisation classification 41 6 6 Methodology for applying level 1 of the degree of urbanisation classification Map 6.3 shows that, when classifying small spatial units as cities, it may be necessary to consider more than one urban centre. In this example, there were 65 593 people living in the urban centre of Haarlemmermeer in the Netherlands, which equated to just 46 % of the total population of the small spatial unit for Haarlemmermeer (below the threshold of 50 % that is required to identify a city). Nevertheless, as shown in the example, there were two adjacent small spatial units — Amsterdam and Haarlem — and their urban centres spill over into Haarlemmermeer. Aggregating the total population of the three urban centres that are located within the boundaries of Haarlemmermeer results in the share of those living in urban centres rising to some 54 % of the total population; as such, Haarlemmermeer is classified as a city. Map 6.3: More than one urban centre needed to define a city — an example for Haarlemmermeer, the Netherlands Haarlem Amsterdam Haarlemmermeer 0 1 2 3 4 km Administrative boundaries: © Eurogeographics © OpenStreetMapContributors Local administrative unit (LAU) boundaries LAU boundary of Haarlemmermeer Urban centre (cluster of high-density cells with population of ≥ 50 000 inhabitants) Note: GEOSTAT population grid from 2011 and small spatial units for 2016. Note: Source: Eurostat, Development GEOSTAT population JRC and European grid from 2011 Commission, Directorate-General and Regional LAU and Urban 2016. Policy and Directorate-General Agriculture and Regional SMALL SPATIAL UNITS WITH NO POPULATION IN THE RASTER EQUIVALENT Some small spatial units will be too small to have a 1 km² grid cell equivalent. When determining their class within level 1 of the degree of urbanisation, these small spatial units are not assigned any population as they are physically too small (smaller than one grid cell); as such, they are given no initial classification. 42   Applying the Degree of Urbanisation — 2021 edition Methodology for applying level 1 of the degree of urbanisation classification 6 After the initial classification, these remaining small spatial units can be selected. For each small spatial unit a centroid falling within its boundaries should be determined. These centroids can be used to classify the remaining small spatial units. They should be spatially joined to the grid-based typology, whereby the small spatial unit gets the classification of the grid type in which its centroid falls. In the EU, such small spatial units were found to be exclusively in urban centres. An example is provided for Dublin in Ireland (see Map 6.4). Map 6.4: Small spatial units with no population in the raster equivalent — an example for Dublin, Ireland Dublin 0 1 2 3 4 km Administrative boundaries: © Eurogeographics © OpenStreetMapContributors Local administrative unit (LAU) boundaries LAU without raster equivalent Urban centre (cluster of high-density cells with population of ≥ 50 000 inhabitants) Note: GEOSTAT population grid from 2011 and small spatial units for 2016. Source: Eurostat, JRC and European Commission, Directorate-General Regional and Urban Policy and Directorate-General Agriculture and Regional Development Note: GEOSTAT population grid from 2011 and LAU 2016. This issue can also be resolved by using a raster with a higher spatial resolution, for example using cells that are 50 by 50 m. At this scale, virtually all small spatial units should have a raster equivalent. If a population grid is available or can be estimated at this scale, these small spatial units without raster equivalent in a 1 km2 grid can still be classified based on their population distribution between the three types of grid cells. The type of grid cells would still be defined at 1 km2, but the population distribution would be determined using the 50 by 50 m cells. Subchapter 10.1.3 describes the degree of urbanisation territorial unit classifier tool (GHS-DU-TUC) which facilitates the process of using smaller grid cells to classify spatial units. Applying the Degree of Urbanisation — 2021 edition  43 6 Methodology for applying level 1 of the degree of urbanisation classification 6.5 Changes over time that impact on the classification given to each small spatial unit The classification given to each small spatial unit according to level 1 of the degree of urbanisation classification should be updated to reflect any changes to the underlying sources of information that are used to determine their class. As such, the classes may be updated to reflect: changes to small spatial unit boundaries or changes to population distributions for 1 km² grid cells. The frequency of such updates varies according to the source of information. Changes to the classification given to each small spatial unit resulting from a revision of population distributions for 1 km² grid cells are less common and these may be expected every 5 or 10 years, when new census data become available. Annual updates of the degree of urbanisation classes assigned to small spatial units should be made to reflect changes to small spatial unit boundaries. These modifications can be implemented in two ways: applying the methodology for the degree of urbanisation classification as described above for the new layer of small spatial units; or estimating the degree of urbanisation based on changes to small spatial unit boundaries. The first approach is more labour intensive, while the second is particularly suitable if boundary changes for small spatial units are relatively minor or consist principally of merging small spatial units, especially if these have the same class at level 1 of the degree of urbanisation classification. UPDATING TO REFLECT CHANGES IN SMALL SPATIAL UNIT BOUNDARIES Small spatial unit boundaries may change over time in three different ways: small spatial units may merge, they may undergo a boundary shift, or they may be split. The most common change for small spatial units within the EU in recent years has been for two or more small spatial units to be merged; boundary shifts have been less common, while splitting small spatial units has been rare. Case 1: small spatial unit mergers Merging two small spatial units with different degrees of urbanisation may be resolved by giving precedence to the more densely populated spatial unit: • when merging small spatial units composed of a city and a town or semi-dense area, reclassify the new small spatial unit as a city; • when merging small spatial units composed of a town or semi-dense area and a rural area, reclassify the new small spatial unit as a town or semi-dense area. Such a process may be further refined by taking into account the relative population sizes of the two small spatial units. Case 1a: small spatial unit mergers involving the same degree of urbanisation The degree of urbanisation classification is additive, meaning that if two small spatial units classified as rural areas are subsequently merged into a single small spatial unit then they will remain a rural area; this is also true for the other classes in the classification. Case 1b: small spatial unit mergers involving rural areas and towns and semi-dense areas These mergers can be addressed in two simple ways: using the population of the urban cluster or using the population of the small spatial units. In the first case, if the population of the relevant urban cluster(s) is available then add the population inhabiting the urban cluster for each of the small spatial units and divide this by the total population of the new small spatial unit to determine the new degree of urbanisation class. If more than 50 % of the population of the new small spatial unit lives in an urban cluster, the new small spatial unit should be classified under towns and semi-dense areas. If the population share is less than 50 %, then the new small spatial unit should be classified under rural areas. 44   Applying the Degree of Urbanisation — 2021 edition Methodology for applying level 1 of the degree of urbanisation classification 6 In the second case, if the population living in the urban cluster cannot be identified, then the degree of urbanisation class may be determined based on the population distribution between the small spatial units. If more than 50 % of the population of the new small spatial unit comes from rural areas, the new small spatial unit should be classified under rural areas. If more than 50 % of the population of the new small spatial unit comes from towns and semi-dense areas, the new small spatial unit should be classified under towns and semi-dense areas. Case 2: small spatial unit boundary shifts Whereas mergers can be dealt with using simple techniques, boundary shifts cannot always be as reliably addressed. Indeed, in some rare cases, boundary shifts between small spatial units that have the same degree of urbanisation class can lead to a change in the classification given to the small spatial units. Such complexity means that a simple rule of thumb is often the preferred and most efficient approach. A simple rule may be established whereby if a small spatial unit loses less than 25 % of its previous population or gains less than 50 % of its population due to boundary shifts, then the degree of urbanisation class does not change. This rule of thumb is likely to cover 90 % of all boundary shifts and ensures continuity. If this is not the case, then further investigation is required, as described below. Case 2a: changes in the degree of urbanisation classification due to boundary shifts are excluded For each small spatial unit, the share of population in the three different types of population grids cells is known. For example, if as the result of a boundary shift the population of a small spatial unit that has 100 % of its population in rural grid cells shrinks, then it will remain classified under rural areas. Equally, if a boundary shift for a small spatial unit that has 100 % of its population in rural grid cells rises, then the new small spatial unit would need to more than double its population before it could (potentially) be classified under towns and semi-dense areas. As a result, if a boundary shift leads to a change in population that is too small to tip the population share of the revised small spatial unit below 50 % of the relevant grid cells, it remains in the same degree of urbanisation class. Case 2b: changes in the degree of urbanisation classification due to boundary shifts are unlikely (but cannot be excluded) If the boundary shift leads to a change in population that is theoretically sufficient to tip the population share of the revised small spatial unit below or above 50 %, but the shift is between small spatial units with the same classification by degree of urbanisation, then the same class should be maintained. Case 2c: changes in the degree of urbanisation classification due to boundary shifts are likely In some cases, changes in the degree of urbanisation class are likely. As an example, if a city were to gain part of a suburb (classified under towns and semi-dense areas) as a result of a boundary shift. The city gains a small number of additional inhabitants (which does not have an impact on its classification by degree of urbanisation). The suburb loses some of its population (that is reclassified to the city). As a result, the population in the revised small spatial unit covered by the suburb may have less than 50 % of its population living in an urban cluster in which case it should subsequently be reclassified under rural areas. Case 3: splitting small spatial units This type of change is relatively rare. Therefore, the main recommendation is one of continuity; in other words, maintain the same degree of urbanisation class. If a small spatial unit is split, the new small spatial units should have the same degree of urbanisation class as the old small spatial unit. If there are concerns that the new small spatial units may have a different degree of urbanisation class, the same approaches as described for boundary shifts may be used. Applying the Degree of Urbanisation — 2021 edition  45 6 Methodology for applying level 1 of the degree of urbanisation classification References Eurostat (2019), Methodological manual on territorial typologies — 2018 edition, Publications Office of the European Union, Luxembourg. Florczyk, A., C. Corbane, D. Ehrlich, S. Freire, T. Kemper, L. Maffenini, M. Melchiorri, M. Pesaresi, P. Politis, M. Schiavina, F. Sabo, and L. Zanchetta (2019), GHSL Data Package 2019, JRC 117104, EUR 29788 EN, Publications Office of the European Union, Luxembourg. 46   Applying the Degree of Urbanisation — 2021 edition 7 Extensions to level 1 of the classification The first two sections of this chapter describe possible extensions to level 1 of the degree of urbanisation classification: how to compile statistics for level 2 of the degree of urbanisation classification and how to compile statistics for functional urban areas (otherwise referred to as metropolitan areas). Both of these extensions have the potential to provide additional useful insight into the spatial structure of a territory/country. The final section details how specific geographic issues should be addressed from a methodological standpoint and provides information on further possible extensions. 7.1 Level 2 of the degree of urbanisation The three classes assigned under level 1 of the degree of urbanisation provide an important first step to assess the urban-rural continuum. Cities are clearly defined settlements which can be organised by population size. The other two classes, however, are quite heterogeneous and do not identify specific types of settlement. The level 1 class of towns and semi-dense areas includes towns, but it does not separate them from semi-dense areas. Equally, rural areas may contain villages, but the degree of urbanisation level 1 does not separate them from other thinly populated areas. Therefore, a second level or sub-classification has been introduced to capture the full settlement hierarchy of large, medium and small settlements or, in simpler terms, cities, towns and villages. 7.1.1 TERMINOLOGY Two sets of terms have been developed to describe level 2 of the degree of urbanisation classification. The first set uses simple and short terms such as city, town and village. The second set uses more neutral and technical language. The second set can be helpful to avoid overlaps with the terms used in national definitions. Table 7.1: Short and technical terms for classifying grid cells for levels 1 and 2 of the degree of urbanisation classification Level Short terms Technical terms 1 Urban centres High-density clusters 2 Urban centres Dense, large clusters 1 Urban clusters Moderate-density clusters 2 Dense urban clusters Dense, medium clusters 2 Semi-dense urban clusters Semi-dense, medium clusters 2 Suburban or peri-urban grid cells Semi-dense grid cells 1 Rural grid cells Mostly low-density cells 2 Rural clusters Semi-dense, small clusters 2 Low-density rural grid cells Low-density grid cells 2 Very low-density rural grid cells Very low-density grid cells Applying the Degree of Urbanisation — 2021 edition  47 7 Extensions to level 1 of the classification Small spatial units can be administrative units — such as municipalities — or statistical areas — such as census units (enumeration areas). Table 7.2: Short and technical terms for classifying small spatial units to levels 1 and 2 of the degree of urbanisation classification Level Short terms Technical terms 1 Cities Densely populated areas 2 Cities Large settlements 1 Towns and semi-dense areas Intermediate-density areas 2 Dense towns Dense, medium settlements 2 Semi-dense towns Semi-dense, medium settlements 2 Suburban or peri-urban areas Semi-dense areas 1 Rural areas Thinly populated areas 2 Villages Small settlements 2 Dispersed rural areas Low-density areas 2 Mostly unhabitated areas Very low-density areas Semi-dense areas in low- and middle-income countries are often described as peri-urban areas. In high-income countries, they are usually described as suburbs. In both cases, these areas have a moderate density and are at the transition between a rural area and a city or town. The technical terms that are used for level 2 of the degree of urbanisation classification follow a specific logic. The population density thresholds each have a specific term: dense means at least 1 500 inhabitants per km2, semi-dense means at least 300 inhabitants per km2, low-density means at least 50 inhabitants per km2 and very low-density means a density of less than 50 inhabitants per km2. The terms large, medium and small each refer to a specific population size threshold: large means a population of at least 50 000 inhabitants, medium means a population of at least 5 000 inhabitants and small means a population between 500 and 4 999 inhabitants. The technical terms for small spatial units that refer to a city, town or village include the word ‘settlement’, while the others use the word ‘area’. The technical terms for the grid cells follow the same approach: the word ‘cluster’ is used if linked to a settlement, while the word ‘cell’ is used for those cells that are not linked to a settlement. 7.1.2 SHORT DESCRIPTION Level 2 of the degree of urbanisation classification is a hierarchical sub-classification of level 1. It was created to identify medium and small settlements, in other words, towns and villages. Practically, it splits two classes into six sub-classes. • Towns and semi-dense areas are split into three subclasses: • dense towns; • semi-dense towns; • suburban or peri-urban cells; and • Rural areas are split into three subclasses: • villages; • dispersed rural areas; • mostly uninhabited areas. Level 2 of the degree of urbanisation classification is implemented with the same two-stage approach as level 1 of the classification. Firstly, grid cells are classified based on population density, population size and contiguity. Subsequently, small spatial units are classified according to the type of grid cells in which their population resides. 48   Applying the Degree of Urbanisation — 2021 edition Extensions to level 1 of the classification 7 7.1.3 GRID CELL CLASSIFICATION Stage 1: classifying grid cells An urban centre is identified in the same manner as for the degree of urbanisation level 1. • An urban centre consists of contiguous (using four-point contiguity) grid cells with a density of at least 1 500 inhabitants per km2. An urban centre has a collective population of at least 50 000. Gaps in this cluster are filled and edges are smoothed. If needed, cells that are at least 50 % built-up may be added (see Subchapter 8.2.1). The urban cluster cells that are not part of an urban centre can be subdivided into three types. • A dense urban cluster consists of contiguous (using four-point contiguity) grid cells with a density of at least 1 500 inhabitants per km2, with a collective population of at least 5 000 and less than 50 000 in the cluster. • A semi-dense urban cluster consists of contiguous (using eight-point contiguity) grid cells with a density of at least 300 inhabitants per km2 and has a collective population of at least 5 000 (in other words, an urban cluster) and this cluster is neither contiguous with nor within 2 km of a dense urban cluster or an urban centre (1). • Suburban or peri-urban cells are the remaining urban cluster cells, in other words those not part of a dense or semi-dense urban cluster. These grid cells are part of an urban cluster that is contiguous (using eight-point contiguity) or within 2 km of a dense urban cluster or an urban centre. Rural grid cells can be categorised into three types. • A rural cluster consists of contiguous (using eight-point contiguity) grid cells with a density of at least 300 inhabitants per km2 and a collective population between 500 and 4 999 in the cluster. • Low-density rural grid cells are rural grid cells with a density of at least 50 inhabitants per km2 and are not part of a rural cluster. • Very low-density rural grid cells are rural grid cells with a density of less than 50 inhabitants per km2. 7.1.4 CLASSIFYING SMALL SPATIAL UNITS Stage 2: classifying small spatial units For level 2 of the degree of urbanisation classification, small spatial units are classified as cities in the same manner as in level 1. • A city consists of one or more small spatial units that have at least 50 % of their population in an urban centre. Within level 2 of the classification, small spatial units classified as towns and semi-dense areas can be divided into three subclasses. • Dense towns have a larger share of their population in dense urban clusters than in semi-dense urban clusters (in other words, they are dense) and have a larger share of their population in dense plus semi-dense urban clusters than in suburban or peri-urban cells (in other words, they are towns). • Semi-dense towns have a larger share of their population in semi-dense urban clusters than in dense urban clusters (in other words, they are semi-dense) and have a larger share of their population in dense plus semi- dense urban clusters than in suburban or peri-urban cells (in other words, they are towns). • Suburban or peri-urban areas have a larger share of their population in suburban or peri-urban cells than in dense plus semi-dense urban clusters. Dense and semi-dense towns can be combined into towns. This reduces the number of classes that are identified for level 2 of the classification and may be useful especially if the population share in semi-dense towns is low. (1) Measured as outside a buffer of three grid cells of 1 km2 around dense urban clusters and urban centres; this ensures that adjacent, but not contiguous suburbs are taken into account. Applying the Degree of Urbanisation — 2021 edition  49 7 Extensions to level 1 of the classification In a similar vein to towns and semi-dense areas, within level 2 of the classification small spatial units classified as rural areas can be divided into three subclasses. • Villages have the largest share of their rural grid cell population living in a rural cluster. • Dispersed rural areas have the largest share of their rural grid cell population living in low-density rural grid cells. • Mostly uninhabited areas have the largest share of their rural grid cell population living in very low-density rural grid cells. In some countries, not all the small spatial units contain inhabitants. To classify the spatial units without any population, the same rules should be applied to their area instead of to their population. Estimates indicate that, for most countries, a considerable share of the population is classified to each of the three classes of the degree of urbanisation level 1. For level 2 of the degree of urbanisation classification, in one or more of the seven classes some countries may only have a relatively small share of their population. Map 7.1 and Map 7.2 show the application of the methodology to Toulouse and its surroundings. Map 7.1: Grid cell classification around Toulouse, France Map 7.2: Small spatial unit classification around for level 2 of the degree of urbanisation classification Toulouse, France for level 2 of the degree of urbanisation Grid classification Local Units REGIOgis Urban centre Suburban cells Very low-density City Suburb Mostly uninhabited area Dense urban cluster Rural cluster rural grid cells Dense town Village Semi-dense urban cluster Low-density rural grid cells Semi-dense town Dispersed rural area 0 20 km 0 20 km 50   Applying the Degree of Urbanisation — 2021 edition Extensions to level 1 of the classification 7 Figure 7.1 provides a simplified and schematic overview of level 2 of the degree of urbanisation classification. Figure 7.1: Schema for the grid cell classification for level 2 of the degree of urbanisation classification Population size thresholds of the cluster of cells No population (settlement size) size criterion ≥ 50 000 5 000 - 49 999 500 - 4 999 (not a settlement) Dense urban Population density of cells, ≥ 1 500 Urban centres clusters inhabitants per km² Semi-dense Suburban or ≥ 300 Rural clusters urban clusters (1) peri-urban grid cells Low-density ≥ 50 rural grid cells Very low-density < 50 rural grid cells (1) Semi-dense urban clusters can have a population of more than 49 999. 7.2 Defining functional urban areas The degree of urbanisation classification may be complemented by a classification of functional urban areas (FUAs) (2). A functional urban area (or metropolitan area) is composed of a city plus its surrounding, less densely populated spatial units that make up the city’s labour market, its commuting zone. This commuting zone generates a daily flow of people into a city and back (home to their dwelling). Such areas are often referred to as ‘functional’ because they capture the full economic function of a city. A functional urban area classification is particularly useful to inform policymaking in a number of domains, including transport, economic development and planning. Several national statistical authorities, including those of Brazil, Italy, Japan and the United States, complement their urban and rural area classifications with a classification of metropolitan areas. The functional urban area classification and the degree of urbanisation classification are linked because they use exactly the same concept of a city. The functional urban area classification is exhaustive, in other words it covers all of the small spatial units in a territory, as those areas that are not classified as functional urban areas (metropolitan areas) are classified as areas outside a functional urban area (non-metropolitan areas). It should be noted that not all of the areas within a functional urban area need to be classified as urban areas (in other words, cities plus towns and semi-dense areas) and that, as such, a functional urban area may contain rural areas if these belong to the commuting zone of a city. In a similar vein, it is possible for an urban area (in other words, cities plus towns and semi-dense areas) to be located outside a functional urban area, but only if the particular urban area is only composed of towns and semi-dense areas and therefore does not have a city. In other words, because cities are systematically included as part of a functional urban area, only towns and semi-dense areas (as well as rural areas of course) can be located outside a functional urban area. (2) This subchapter is adapted from Dijkstra et al. (2019). Applying the Degree of Urbanisation — 2021 edition  51 7 Extensions to level 1 of the classification 7.2.1 TERMINOLOGY This section summarises the terms that are necessary to distinguish the different concepts that are used to define functional urban areas. Table 7.3: Terminology related to functional urban areas Preferred term Synonym Geographic level Urban centres High-density clusters (HDCs) Grid Cities Densely populated areas Small spatial unit Commuting zones Small spatial unit Functional urban areas (FUAs) Metropolitan areas Small spatial unit Areas outside functional urban areas (non-FUAs) Non-metropolitan areas Small spatial unit 7.2.2 SHORT DESCRIPTION A functional urban area (metropolitan area) can be defined in four steps: • Identify an urban centre — a set of contiguous grid cells with a density of at least 1 500 inhabitants per km² and with a collective population of at least 50 000. • Identify a city — one or more small spatial units that have at least 50 % of their population in an urban centre. • Identify a commuting zone — a set of contiguous small spatial units that have at least 15 % of their employed residents working in a city. • A functional urban area (metropolitan area) is a city plus its commuting zone. Consequently, within the functional urban area classification, all the areas of a territory outside of cities and their commuting zones may be considered as areas outside a functional urban area (non-metropolitan areas). Figure 7.2 shows visually the different concepts that are used in the classification of functional urban areas, notably the urban centre, the city, and the commuting zone. The following data sources are required to compile statistics for functional urban areas: • a residential population grid with the number of inhabitants per km² of land area (in other words, excluding water bodies); • digital boundaries for small spatial units; • commuting flows between the small spatial units and the number of employed residents per small spatial unit. Figure 7.2: Urban centre, city, commuting zone and functional urban area of San Luis Potosí, Mexico 0 10 km 0 20 km 0 50 km 0 50 km Urban centre Urban centre City Functional urban area Small spatial unit City Commuting zone Small spatial unit 7.2.2.1 Definition of an urban centre The first step focuses on the concentration of population in space, which is the simplest and most uncontroverted feature of a city — the starting point for this definition. The idea of a city as a place with a relatively high concentration of population in space is common to many disciplines that describe a city including economic, social, cultural and geographical ones. 52   Applying the Degree of Urbanisation — 2021 edition Extensions to level 1 of the classification 7 How to estimate commuting flows? Several countries do not collect commuting data as part of their census. Other sources such as linked population and employment registers or mobile phone data could be used to estimate such flows. Estonia offers an illustrative example where — as reported in two studies commissioned by the Ministry of the Interior and conducted by the Mobility Lab of the University of Tartu (Ahas et al. (2010); Ahas and Silm (2013)) — mobile positioning data made it possible to delineate functional urban areas (metropolitan areas). The movements between individuals’ anchor points (in other words, residence, work, and so on) are aggregated at the level of small spatial units (in other words, municipalities) in order to produce a matrix of flows. Such a matrix has the benefit of providing an estimation of mobility patterns for the entire population, rather than for employees only, at a highly disaggregated spatial scale. The Netherlands has also produced a flow matrix between all small spatial units within its territory using mobile phone data (Van der Valk et al. (2019)). Many national definitions of a city rely on the population size and density of a small spatial unit. This causes two types of problems. A big city in a relatively large spatial unit may have a very low or rural population density. For example, Ulaanbaatar, the capital city of Mongolia, has a population of 1.4 million but a density of only 270 inhabitants per km². The population of a city is difficult to determine when it is spread out over multiple small spatial units. For example, how many people live in Paris? An urban centre, as defined in this methodological manual, relies on a population grid which can identify spatial concentrations of population independently from political or administrative boundaries, using spatial units of the same shape and size. An urban centre or high-density cluster is a spatial concept based on grid cells of 1 km². It is defined in three steps, as indicated below and represented in Figure 7.3. • Step 1: all grid cells with a density of at least 1 500 inhabitants per km² of land are selected. If needed, cells that are at least 50 % built-up may be added (see Subchapter 8.2.1). • Step 2: contiguous high-density cells are then clustered. Only those clusters with at least 50 000 inhabitants are kept. To avoid over-aggregation, four-point contiguity is used (in other words, cells with only the corners touching are not considered). • Step 3: gaps in each cluster are filled separately and its edges smoothed. Figure 7.3: High-density cells, high-density clusters, urban centre of Toulouse, France Cells Clusters Urban centre 1 924 2 708 1 512 7 345 2 301 32 269 2 636 26 348 1 961 9 726 1 611 969 795 5 191 4 454 1 655 4 037 2 814 35 269 6 009 4 124 1 970 1 676 7 872 5 467 1 837 1 861 2 699 1 700 16 991 1 684 High-density cell < 50 000 inhabitants Remaining cluster (≥ 1 500 inhabitants per km²) ≥ 50 000 inhabitants Gap lling Small spatial unit Applying the Degree of Urbanisation — 2021 edition  53 7 Extensions to level 1 of the classification 7.2.2.2 Definition of a city A city consists of one or more small spatial units with at least 50 % of their population in an urban centre. A small spatial unit can be either an administrative unit or a statistical area. Examples of administrative units include a municipality, a district, a neighbourhood or a metropolitan area. Some of these administrative units also have a political role as electoral districts or in terms of local government. Statistical areas can be census units/enumeration areas, census blocks, census tracts, wards, super output areas, named places or small areas. Examples of small spatial units used in OECD countries include communes in France, municipalities in Italy, sigungu in South Korea and census subdivisions in Canada. The best small spatial unit for this definition is the smallest unit for which commuting data are available (3). Figure 7.4 shows the process through which a city is identified by intersecting the grid-based urban centre with small spatial units. Figure 7.4: Urban centre and city for Toulouse, France Urban centre Urban centre Urban centre Small spatial unit Small spatial unit < 50 % of its population City in an urban centre Small spatial unit ≥ 50 % of its population in an urban centre 7.2.2.3 Definition of a commuting zone Once all cities have been defined, commuting zones can be identified using the following steps: • if 15 % of employed persons living in one city work in another city, these cities are treated as a single city — this step is referred to as a ‘polycentricity check’; • all small spatial units with at least 15 % of their employed residents working in a particular city are identified as part of the commuting zone for that city (see Figure 7.5, second panel); • enclaves, in other words, small spatial units entirely surrounded by other small spatial units that belong to a commuting zone or a city are included and exclaves or non-contiguous small spatial units are excluded (see Figure 7.5, third panel). (3) In principle, commuting data at grid level would be another usable option, if available. 54   Applying the Degree of Urbanisation — 2021 edition Extensions to level 1 of the classification 7 It can happen that, due to a low intensity of commuting flows, there is no commuting zone for a specific city. In this case, there is a perfect correspondence between the functional urban area and the city. The delineation of functional urban areas is summarised in Figure 7.5. Figure 7.5: City, commuting zone and functional urban area for Genova, Italy Commuting area after including City Commuting area enclaves and dropping exclaves City Small spatial unit with ≥ 15 % Functional urban area of its employed population Small spatial unit Added enclave commuting to the city Removed exclave Figure 7.6: Defining a functional urban area Methodology to de ne functional urban areas Apply the threshold to identify densely populated grid ≥1 500 inhabitants per km cells STEP 1: create urban centres Select a cluster of contiguous high-density grid cells with ≥50 000 inhabitants a total population above the population threshold Select the small spatial units that have the majority of if at least 50 % of the population of the local unit lives their residents living in an urban centre to create a ‘city’ within the urban centre STEP 2: create cities if at least 15 % of the population of one city commutes to work Check whether two or more cities will belong to the same in another city; if this applies, those cities will be treated as a functional urban area (’polycentricity check’) single functional urban area in the next step STEP 3: Select small spatial units that surround the city if at least 15 % of the population of a local unit create commuting and that are part of its labour market commutes to work in the city zones STEP 4: create a functional Combine the city and its commuting zone urban area Applying the Degree of Urbanisation — 2021 edition  55 7 Extensions to level 1 of the classification 7.2.3 DEFINING AN URBAN CENTRE The approach to identify an urban centre as part of the functional urban area classification is identical to that described for level 1 of the degree of urbanisation classification (see Subchapter 6.3.1). To identify an urban centre (high-density cluster): • Select all 1 km² grid cells with a density of at least 1 500 inhabitants per km² of land area (in other words, for each cell the density should be calculated by excluding bodies of water); if needed, cells that are at least 50 % built-up may be added (see Subchapter 8.2.1). • Cluster all contiguous cells above this density threshold using only four points of contiguity and keep those clusters with at least 50 000 inhabitants (high-density clusters); remove any clusters that have less than 50 000 inhabitants. • Fill any gaps and smooth borders using the ‘majority rule’ iteratively until no more cells may be added. The identification of an urban centre is based on a population grid. Several statistical authorities already produce their own population grids. For example, the 2011 GEOSTAT grid covers all EU Member States (4). Australia, Brazil, Colombia and Egypt either have their own grid or are developing one. Other national statistical authorities plan to produce an official population grid by geo-coding their next census. Because these grids are based on points, they are called ‘bottom-up’ grids. In other words, the grid is created from the bottom-up using data with a higher spatial resolution. Various institutions provide modelled global population grids that are publicly available (see Chapter 5). In countries with relatively low-density urban development, a very accurate population grid and a strong separation of land uses, this approach may lead to an excessive fragmentation of urban centres. In such places, grid cells with shopping centres, transport infrastructure or business parks will not reach the residential density threshold to be included in the urban centre and this has the potential to create breaks between adjacent areas. The quality of the population grid also plays a role. In a disaggregation grid, some population would still be attributed to commercial or industrial areas, whereas in a bottom-up grid this would not be the case. Therefore, fragmentation is less likely to occur when using a disaggregation grid. To resolve this issue, grid cells that are at least 50 % built-up may be added to the urban centre. This resolves the problem in this specific type of city and has little to no impact on higher- density cities, as virtually all the cells that are at least 50 % built-up have a high enough population density or are added as part of the gap-filling process. (4) GEOSTAT 2011 (http://ec.europa.eu/eurostat/web/gisco/geodata/reference-data/population-distribution-demography/ geostat). 56   Applying the Degree of Urbanisation — 2021 edition Extensions to level 1 of the classification 7 7.2.4 DEFINING A CITY In most cases, defining a city is simple. There is a single urban centre located in a single small spatial unit. This means that all of the urban centre population is located in that small spatial unit and the share of its population in that urban centre is very high (see Figure 7.7). Figure 7.7: High-density cells, urban centre and city for Graz, Austria High-density cells Urban centre (cluster of Urban centre Urban centre (≥ 1 500 inhabitants per km²) high-density cells with Small spatial unit with City Small spatial unit population ≥ 50 000 inhabitants) ≥ 50 % of its population in an urban centre However, in some cases the relationship is more complex. Two cases are discussed below: (i) if a city contains more than one urban centre; and (ii) if an urban centre covers two distinct cities. 7.2.4.1 A city contains more than one urban centre It may be that a wide river, a steep slope or an industrial area has led to a split in the urban centre. In this case, the small spatial unit simply represents both urban centres. For example, Budapest has two separate urban centres Figure 7.8: Example of two urban centres within the same small spatial unit — Budapest, Hungary High-density cells Urban centre (cluster of Urban centre City (≥1 500 inhabitants per km²) high-density cells with population ≥ 50 000 Small spatial unit ≥ 50 % of its Small spatial unit population in an urban centre inhabitants) Applying the Degree of Urbanisation — 2021 edition  57 7 Extensions to level 1 of the classification (Buda on the west bank of the Danube and Pest on the east bank). They both fall within the same small spatial unit (see Figure 7.8). 7.2.4.2 An urban centre covers two distinct cities Some urban centres cover two (or more) distinct cities, in the sense of two distinct urban settlements with their own centre and their own name. This can happen because these cities have grown towards each other but remain functionally distinct. If the population grid is estimated, this situation might occur because the estimated population is often more evenly distributed than the actual population. In some cases, an urban centre can become too big to be plausible as the centre of a daily urban system, meaning that it is too large to be considered as a space encompassed by the daily movements of people between residence and place of work. When a single urban centre covers two or more distinct cities, a national statistical authority can choose to create multiple cities. For example, Poole and Bournemouth in the United Kingdom share a single urban centre (see Figure 7.9) but are two separate cities. However, each of these cities should have a population of at least 50 000 inhabitants. If there is at least a one-way commuting flow of more than 15 % between these two cities, they should Figure 7.9: Example of two cities with a single urban centre — Poole and Bournemouth, the United Kingdom High-density cells Urban centre (cluster of Urban centre City (≥1 500 inhabitants per km²) high-density cells with population ≥ 50 000 Small spatial unit ≥ 50 % of its Small spatial unit population in an urban centre inhabitants) have a joint commuting zone and therefore be part of the same functional urban area. If, instead, the flow of commuting between the two cities is less than 15 %, then each city should have its own commuting zone and its own functional urban area. In addition, the urban centre can also be split into two parts along the border between the two cities. 7.2.4.3 What is a greater city? In some situations, an urban centre may stretch far beyond the boundaries of the small central spatial unit that gives it its name. This is often the case for (large) capital cities that have outgrown their central spatial unit, such as Athens, Copenhagen, Paris or Valletta. To avoid confusion, the pre-fix ‘greater’ is often added to their name. This is already common practice in several countries, for example Greater London, Greater Dublin, Grand Paris and so on. The functional urban area classification ensures that the most comparable boundaries are selected. It does this by first defining an urban centre independently from administrative boundaries and only in a second step identifying the administrative boundaries that correspond best to this urban centre. In this way, it is possible to ensure that comparisons are not made between, for example central Paris (within the confines of the périphérique) and the full urban sprawl of Berlin or London. Countries with relatively small spatial units, such as France and Switzerland, are more prone to this problem of ‘underbounding’ (see also Chapter 8). 58   Applying the Degree of Urbanisation — 2021 edition Extensions to level 1 of the classification 7 In short, a greater city is a city. The addition of the term ‘greater’ functions only as a warning to the data users that this definition of the city contains more small spatial units than the central spatial unit which gives this city its name. 7.2.5 DEFINING A COMMUTING ZONE 7.2.5.1 Checking for connected cities: the polycentricity check The delineation and definition of a commuting zone starts with the polycentricity check, in other words, a check to see if two or more cities are linked by strong commuting flows. If city A has 15 % of its employed residents commuting to city B, then these two cities will share a single commuting zone. Note: it is sufficient that the flow of commuters reaches 15 % in a single direction. For example, if city B has a commuting flow of less 15 % to city A, it will still share the same commuting zone. The polycentricity check is applied only once; it is not an iterative rule. For example: City C has a commuting flow of 20 % to city D. City E has a commuting flow of 10 % to city C and 10 % to city D. Then cities C and D will have a shared commuting zone, but city E will have its own commuting zone because the commuting flow to each individual city is too small. If city H and city I both have a commuting flow over 15 % to city J, then all three cities will share a single commuting zone. 7.2.5.2 Creating the commuting zone The next step is to identify all small spatial units with at least 15 % of their employed residents working in a single city (or both cities in the case of cities linked by commuting flows). If a small spatial unit has a commuting flow of more than 15 % to two different cities, it will become part of the commuting zone of the city to which the flow is biggest. If a small spatial unit has a commuting flow of 20 % to city K and 17 % to city L, it will be classified as part of the commuting zone of city K. Enclaves, in other words, small spatial units surrounded by a single functional urban area, are included and exclaves (or non-contiguous small spatial unit) are excluded. An enclave is defined as a small spatial unit that shares 100 % of its land border with the functional urban area (city or commuting zone); water borders are not considered. An exclave is defined as a small spatial unit that does not share any border with the functional urban area (city or commuting zone); in other words, it is a non-contiguous spatial unit. The city destination for commuting flows should be the best approximation of the urban centre, in other words, all the units with at least 50 % of their population in the urban centre. If the city boundary is adjusted by adding or dropping a few small spatial units or shifted to a higher administrative level (see next section), this adjusted city should not be used for the commuting analysis; the only exception is where a single urban centre covers multiple cities. 7.3 Other possible extensions to the methodology: remoteness and land cover Global Strategy’s, Guidelines on defining rural areas and compiling indicators for development policy, published in 2018 (5) identifies three dimensions of ’rurality’: (i) sparse settlement; (ii) remoteness (from urban areas); and (iii) land cover. While consideration of all three dimensions is potentially useful for policy design and analysis, it is the dimension of sparse settlement (population size and density) that is captured by the degree of urbanisation classification. Sparse settlement reflects the idea that at one end of a continuum (as measured by population size or density) there are rural areas that are more sparsely populated and settled, while at the other there are urban areas that are the most populous and densely populated parts of a country. Remoteness affects the opportunities people have to gain access to markets (for goods, services and labour) and to public services. It is most often represented by (5) Global Strategy (http://gsars.org/wp-content/uploads/2018/12/GS-GUIDELINES-RURAL-AREAS-EN-FINAL-2018.pdf). Applying the Degree of Urbanisation — 2021 edition  59 7 Extensions to level 1 of the classification the difficulty of physical travel to places where markets and services are more (widely) available. Land cover is the physical cover on the land including vegetation (either planted or naturally occurring) and any buildings, other structures or features constructed by humans. Land cover reflects and determines land use, which is related to the human activities that take place there. Remoteness In general, remoteness (or distance from urban areas) is considered an important dimension of rurality. In combination with low population density, remoteness characterises rural areas that face particular challenges concerning their development. Remote areas are generally those where population densities are low, markets of all kinds are thin, and the unit costs of delivering most social services and many types of infrastructure are high. Additionally, in these areas that are distant from urban centres, farm-gate (or factory-gate) prices of outputs are often low and prices of inputs are often high, while it is usually difficult to recruit skilled personnel to work in public services or private enterprises. In contrast to remote areas, urban areas are characterised by agglomeration economies, in other words benefits that come when firms and people locate near to one another in cities and industrial clusters, effectively lowering the costs of transporting goods and sharing knowledge. More specifically, remoteness signifies the extent of opportunity people have to gain access to markets. The dimension of remoteness can be included in the methodology for analytical purposes, though the identification of an empirical measure of remoteness to be used depends on the context in each country. Instinctively, a remote area is far from a city in terms of distance or the time it takes to travel physically from one place to another. The mode and speed of transportation, however, would be expected to vary depending on terrain and on the presence or absence of infrastructure. Travel by road or train might be the most common means of transport in one place, but travel by water or foot may be more common in another. While the variables chosen might be different across countries, or even within countries, the underlying supposition is that physical access to a city is key, however it is achieved. Physical distance is not a perfect proxy, however, as distance may not restrict access to one service (for example, access to online education) but it may be a significant barrier to another (for example, access to a surgery at a regional hospital). Some disadvantages of remoteness may be overcome by telecommunication or internet services, as for example with the provision of health care services through satellite video. However, remoteness in terms of travel time is likely to be the most expedient approach when selecting a variable. Concept of remoteness: an example for small regions The OECD has used the concept of remoteness in a classification of small regions based on their access to functional urban areas (Fadic et al. (2019)). Based on this, a small subnational region (or territorial level 3 region, TL3) is classified as either a ‘metropolitan region’ — if at least half of its population lives in a functional urban area of at least 250 000 inhabitants — or as a ‘non-metropolitan region’. The concept of remoteness is thereafter used to further characterise non-metropolitan regions, more specifically: • if at least half of the population in a non-metropolitan region cannot reach a functional urban area within a one hour drive, then that region is sub-classified as ‘remote’; • if at least half of the population in a non-metropolitan region can reach a functional urban area within a one hour drive, then that region is sub-classified depending on the size of the functional urban area, as a non-metropolitan region: • ‘with access to a metropolitan region’ (for functional urban areas of at least 250 000 inhabitants); or • ‘with access to a small functional urban area’ (for functional urban areas with less than 250 000 inhabitants). 60   Applying the Degree of Urbanisation — 2021 edition Extensions to level 1 of the classification 7 In short, though the concept of remoteness seems straightforward, it is not always clear how to represent it with data. For example, is remoteness always a function of physical distance? Or might this barrier be reduced/removed, for example by access to telecommunications that allow commercial transactions to take place virtually or social services like health care to be delivered remotely? Furthermore, data on road networks and their use are hard to come by on a global scale, although there have been recent attempts at improvement. The use of mass/public transportation might also be complicated to measure. In any case, remoteness might be considered less a permanent aspect of rurality and more a condition to be addressed by taking steps to improve access to markets and services in rural areas themselves. If that is the case, then a definition of remoteness should not include any elements that themselves are policy targets. Land cover Land cover consists of vegetation (occurring naturally or cultivated), buildings, roads and other man-made features and describes cover by forest, grassland, impervious surfaces, cropland and other land and water types (such as wetlands and open water). This is in contrast to land use that defines what people do on the landscape (for example, work in factories, live in houses, use parks and gardens for recreation, graze cattle on agricultural land) with the intention of getting benefit from its use. A given type of land cover, say tree cover, may support multiple land uses: for example, recreation, logging and/or conservation. For rural development policies and analytical purposes, countries may use land cover as an additional dimension to further enrich their understanding of rural areas and augment rural development policies (FAO (2018)). Applying the Degree of Urbanisation — 2021 edition  61 7 Extensions to level 1 of the classification References Ahas, R. and S. Silm (2013), ‘Regionaalse Pendelrände Kordusuuring’ (Re-study of regional commuting), Regionaalministri Valitsemisala. Ahas, R., S. Silm, O. Järv, E. Saluveer, and M. Tiru (2010), ‘Using Mobile Positioning Data to Model Locations Meaningful to Users of Mobile Phones’, Journal of Urban Technology, Volume 17, Issue 1, pp. 3-27. Fadic, M., J. E. Garcilazo, A. I. Moreno-Monroy and P. Veneri (2019), ’Classifying small (TL3) regions based on metropolitan population, low density and remoteness’, OECD Regional Development Working Papers, No. 2019/06, OECD Publishing, Paris. FAO (2018), Guidelines on defining rural areas and compiling indicators for development policy, Food and Agriculture Organization of the United Nations (FAO), Rome. Van der Valk, J., M. Souren, M. Tennekes, S. Shah, M. Offermans, E. De Jonge, J. Van der Laan, Y. Gootzen, S. Scholtus, A. Mitriaieva, B. Sakarovitch, S. Hadam, M. Zwick, M. Rengers, A. Kowarik, M. Weinauer, J. Gussenbauer, M. Debusschere and A. Termote (2019), City data from LFS and Big Data, Publications Office of the European Union, Luxembourg. 62   Applying the Degree of Urbanisation — 2021 edition 8 Which spatial units to use and adjustments to address geographic issues 8.1 Which small spatial units to use? The population grid helps to address what is referred to as the modifiable areal unit problem (1). However, when these grid concepts are used to classify small spatial units, the problem that different shapes and sizes of spatial units will lead to different results reappears. The general recommendation is to use the smallest spatial unit for which regular data can be produced to compile statistics by degree of urbanisation. It is not necessary to be able to produce reliable data for each individual small spatial unit, rather the goal is to compile statistics for the aggregation of these spatial units by degree of urbanisation. Household sample surveys, for example, cannot produce data for all small administrative or statistical spatial units, but if respondents are coded by these spatial units, the results of the survey can subsequently be aggregated to compile statistics by degree of urbanisation. Many countries have more than one local administrative level and more than one potential type of statistical area that might be chosen as the small spatial units to delineate cities and functional urban areas. Smaller spatial units will normally lead to a closer match between an urban centre and a city. However, national statistical authorities may not be able to provide annual data for many indicators at such a detailed level. Furthermore, smaller spatial units, such as wards or districts, may not have as strong a political role as larger spatial units (such as municipalities). This section describes some of the issues a national statistical authority may encounter when classifying spatial units by degree of urbanisation and proposes a range of options for how they may be addressed. 8.1.1 LARGE SPATIAL UNITS MAY LEAD TO THE OVER-, UNDER- OR NON- REPRESENTATION OF AN URBAN CENTRE BY A CITY The population of an urban centre and that of a city can differ by a considerable amount if a country has relatively large spatial units. Below are three types of issues that may potentially arise when using relatively large spatial units to define a city. Overrepresentation A city can have almost double the population of an urban centre. For example, an urban centre of 50 001 inhabitants in a spatial unit of 100 000 would mean that this spatial unit will be defined as a city (see Subchapter 7.1.4). This is a tricky problem to solve as the only alternative to the overrepresentation is non- representation, in other words, by not defining this spatial unit as a city. Underrepresentation A city can also have a much smaller population than the urban centre it represents. Take for example, an urban centre of 200 000 inhabitants that is split across four spatial units. One spatial unit (A) has a population of 50 000 and all of its inhabitants live in the urban centre. The other three spatial units (B, C, D) each have a population of 150 000 inhabitants of which respectively 60 000, 50 000 and 40 000 live in that urban centre. As a result, the city will consist of just the one spatial unit (A) with a population of 50 000 inhabitants and not the other three spatial units (B, C or D). (1) The modifiable areal unit problem (or MAUP) highlights that using different boundaries can produce different results. For example, altering the boundaries of electoral districts can change the outcome in first-past-the-post systems. When using larger spatial units, the degree of urbanisation classification tends to categorise fewer people as living in rural areas and cities and more people as living in towns and semi-dense areas. The MAUP was originally identified by Gehlke and Biehl (1934) and further developed by Openshaw (1984). Applying the Degree of Urbanisation — 2021 edition  63 8 Which spatial units to use and adjustments to address geographic issues This underrepresentation can be reduced by adding the spatial unit with the highest share of its population in that urban centre to the city (spatial unit B with 60 000 of its 150 000 inhabitants in the urban centre). This would bring the population of the city up to 200 000 inhabitants, of which 110 000 would be living in the urban centre. Non-representation The most extreme form of under-representation is non-representation. For example, a spatial unit with a population of 200 000 inhabitants with a single urban centre of 75 000 inhabitants will not be classified as city. As a result, this urban centre will not be represented by a city, in other words, non-representation, something which is more likely to happen for small urban centres. In a country where all the spatial units are relatively large, it is likely that not all of the small urban centres will be represented by cities. This would create a quite skewed representation of urban centres as all small urban centres would be missing. One option to address this problem is that for half of the small urban centres without a city, their spatial unit is classified as a city even though their share of population in an urban centre is less than 50 %. 8.1.2 SMALL SPATIAL UNITS MAY LEAD TO A LOSS OF THE LINK TO LOCAL GOVERNMENT OR TO LESS STATISTICAL DATA In a country with relatively large spatial units, most cities will consist of a single spatial unit. As a result, each city will have a single local government. This makes it easier to communicate indicators to local politicians/representative groups and helps to ensure good inputs for policymaking. In countries with relatively small spatial units, most cities will consist of multiple spatial units. These small spatial units will ensure that there is a close match between the population in the urban centre and the population in the city. The trade-off is that the city will not match a single local government, which makes it more complicated to communicate data to local politicians/representative groups. This effect can be shown in Portugal, which has both municipalities (municipio or concelho) and parishes (freguesia). If the urban centre of Braga in Figure 8.1 is used to define the municipal level (left panel), there is a simple one-to-one relationship; the local government of Braga is organised at the municipal level. If the urban centre is used to define a city at the parish level (right panel), the relationship becomes a more complicated one-to-many relationship; the simple link with the local government of Braga is also lost. When statistical areas are used as building blocks to define a city and/or a functional urban area, the latter can be adapted ex post to the closest local administrative units. For example, cities and their commuting zones in the United States have been delineated using census tracts as building block units, but subsequently adapted to the closest county boundaries, by including the counties where the share of population living in cities and functional urban areas was higher than 50 %. The imperfect match between the cities and functional urban areas and their respective urban centres can be informative for policymakers. Administrative boundaries of cities often remain unchanged for decades, while cities can expand or shrink. Many OECD countries, following the urban expansion that occurred in the last few decades, have created new levels of government for large cities encompassing multiple spatial units. For example, France has created métropoles to help govern its 21 biggest cities. 64   Applying the Degree of Urbanisation — 2021 edition Which spatial units to use and adjustments to address geographic issues 8 Figure 8.1: Example of the influence of the choice of type of spatial unit — municipal and parish levels, Braga, Portugal Municipal level Parish level Urban centre Urban centre Medium spatial unit Small spatial unit Medium spatial unit ≥ 50 % of its population in an urban centre Small spatial unit < 50 % of its population in an urban centre Small spatial unit ≥ 50 % of its population in an urban centre 8.1.3 ADJUSTING THE CITY TO ENSURE A BETTER REPRESENTATION OF THE URBAN CENTRE OR A BETTER LINK TO LOCAL GOVERNMENT If a national statistical authority wishes to adjust the delineation of its cities to get a better link between a city and its urban centre or a city and its local government, it can add or drop a spatial unit as long as the two following rules are respected: • Rule 1: a spatial unit with less than 50 % of its population in an urban centre can be added to a city if at least 50 % of the population of this expanded city lives in an urban centre. • Rule 2: a spatial unit with at least 50 % of its population in an urban centre can be excluded from a city as long as at least 75 % of the population of that urban centre lives in a city after excluding the spatial unit. These two rules were designed to provide statistical limits to these optional changes that can be made. Furthermore, national statistical authorities are encouraged to limit the number of adjustments that they make, as these may weaken the international comparability of results compiled according to the degree of urbanisation classification. City adds a few spatial units Returning to the example of Braga in Portugal: if the urban centre is used to define the city at the parish level, this city would only contain some of the parishes in the municipality of Braga. Defining Braga at the municipal level amounts to adding these surrounding parishes to the city. As still more than 50 % of the population of the municipality of Braga lives in the urban centre, this complies with rule 1; it also ensures a direct link to Braga’s local government. Applying the Degree of Urbanisation — 2021 edition  65 8 Which spatial units to use and adjustments to address geographic issues City drops a few spatial units An example of the application of rule 2 is presented for Vienna in Austria. A number of small spatial units just south of the city of Vienna have 50 % or more of their population in the urban centre of Vienna. As more than 75 % of the population of the urban centre lives in the city of Vienna, these smaller spatial units can be excluded without significantly compromising the comparability of the results (see Figure 8.2). Figure 8.2: Dropping a few spatial units from a city, Vienna, Austria High-density cells Urban centre (cluster of Urban centre City (≥1 500 inhabitants per km²) high-density cells with population ≥ 50 000 Small spatial unit ≥ 50 % of its Small spatial unit population in an urban centre inhabitants) Cities without an urban centre The definition that has been developed provides an estimate of the population of an urban centre. Two elements may reduce the accuracy of this estimate: (i) geographic features and (ii) the source of the population grid data. The definition does not take into account the specific geography of a city. Some geographic features, such as steep slopes, cliffs or bodies of water may lead to an underestimation of the population of an urban centre. This affects in particular cities with a small centre. The definition works best when a bottom-up grid (based on point data) or a high-resolution, hybrid grid (based on a mixture of points and smaller statistical areas) is available, which ensures that the density of the population (per km²) is very accurate. In countries where such a grid is not yet available, the population of a small spatial unit has to be disaggregated based on a given criterion, such as land use data in the case of the GHS-POP grid produced by the European Commission’s Joint Research Centre (JRC). This is called a top-down approach, which is generally less accurate. It tends to underestimate the population cells with a moderate to high-density and overestimate population in those grid cells with a low population density. Due to this imprecision, there remains a margin of error, especially for smaller centres. Therefore, a national statistical authority may opt to classify a small spatial unit as a city when it lacks an urban centre of at least 50 000 inhabitants, but fulfils the following two conditions: • the presence of an urban centre of at least 50 000 inhabitants, which the definition does not capture due to geographic features or population grid estimation techniques; • the small spatial unit has a population of at least 50 000 inhabitants. For example, a small spatial unit which has two clusters of high-density cells separated by a river or a bay which together have a collective population of at least 50 000 inhabitants can be argued to have an undetected urban centre. A small spatial unit with a high-density cluster of 49 000 inhabitants based on a top-down population grid can be argued to have an undetected urban centre (see Subchapter 8.2.1 for more details). 66   Applying the Degree of Urbanisation — 2021 edition Which spatial units to use and adjustments to address geographic issues 8 8.2 Adjustments to address specific geographic issues for the degree of urbanisation and functional urban area classifications This section describes how the degree of urbanisation classification can be adjusted in the presence of certain geographic issues that may skew the results. These adjustments are optional. In most countries, the original classification without these adjustments will produce robust results. 8.2.1 RAILWAYS, HIGHWAYS, MALLS, OFFICE PARKS AND FACTORIES In countries with a strong separation of land use functions and relatively low-density urban developments, the methodology may generate multiple urban centres for a single city. For example, Houston in the United States has nine urban centres if the methodology is applied without considering cells that have at least 50 % of their land classified as built-up areas (see Map 8.1). This is often because highways, railways, shopping centres, office parks and factories typically have little or no residential population and can occupy enough of a single grid cell that it does not reach the population density threshold of at least 1 500 inhabitants per km². Although many people may use these areas during the daytime, the methodology is designed to be applied to the residential population, broadly speaking the night-time population. As a consequence, areas which are intensively used by city residents during the day but which have few, if any, residents might not be considered to be part of a city. Creating urban centres using both criteria — cells with a density of at least 1 500 inhabitants per km² and cells that are at least 50 % built-up — resolves this issue. For example, in Houston the nine separate urban centres are all connected by cells that are at least 50 % built-up (see Map 8.2). Map 8.1: Grid cell classification without Map 8.2: Built-up cells, urban centres and dense considering built-up cells, Houston, United States urban clusters without considering built-up cells, Houston, United States Urban centre Dense urban cluster Semi-dense urban cluster Urban centre Suburban cells Dense urban cluster Rural cluster Built-up area coverage share Low-density rural grid cells 0 - < 50 % 0 20 km 0 20 km REGIOgis Very low-density rural grid cells REGIOgis ≥ 50 % Applying the Degree of Urbanisation — 2021 edition  67 8 Which spatial units to use and adjustments to address geographic issues When the urban centre is defined using both of these criteria, the nine separate urban centres become one (see Map 8.3). In addition, a few separate dense urban clusters are also combined such that they reach the 50 000 population threshold and become an urban centre (see Map 8.4). As official, up-to-date, high-resolution data on built-up areas are generally not available for many countries, this adjustment is optional. If high-quality data on built-up areas are available, however, adding the cells that are at least 50 % built-up to the urban centres is encouraged. Map 8.3: Built-up cells, urban centres and dense Map 8.4: Grid cell classification considering built- urban clusters considering built-up cells, Houston, up cells, Houston, United States United States Urban centre Dense urban cluster Urban centre Semi-dense urban cluster Dense urban cluster Suburban cells Built-up area coverage share Rural cluster 0 - < 50 % Low-density rural grid cells 0 20 km 0 20 km REGIOgis ≥ 50 % REGIOgis Very low-density rural grid cells 8.2.2 WATER BODIES, STEEP SLOPES AND PARKS IN A CITY The presence of water bodies, steep slopes and parks may have an impact on the capacity of the methodology to identify a city. These elements can lead to gaps or separations which result in a single urban centre being fragmented into multiple centres or — when these fail to reach the minimum population threshold of 50 000 inhabitants — multiple dense urban clusters. To overcome these problems, the methodology can be adapted to address gaps or separations that are due to the presence of waterways, parks and/or areas with steep slopes. This optional process should be applied to clusters of high-density grid cells before evaluating the minimum population of urban centres. Hence, the initial input of the workflow are clusters of contiguous grid cells characterised by a population density threshold of at least 1 500 inhabitants per km², without any criterion for the total population of the cluster. For the purpose of this process description, they are called sHDCs (small high-density clusters), as no minimum population threshold was applied. Each of these sHDCs is stored as a polygon and receives its unique number, which is required in further steps of the workflow. Additional spatial data are needed to represent the areas that will be taken into account in a special exercise to fill gaps in or separations between sHDCs: • Waterways should ideally be portrayed as polygon features. If these are not available, waterway line features should be buffered to model the actual width of the waterway. Furthermore, waterway polygons can 68   Applying the Degree of Urbanisation — 2021 edition Which spatial units to use and adjustments to address geographic issues 8 (optionally) be buffered by a limited width (for instance, a maximum of 50 m) to portray adjacent zones which are assumed not to be suitable for the construction of buildings. • Zones with steep slopes should be retrieved from a layer with appropriate spatial detail. Usually this will be a selection of raster cells, with resolution equal to or higher than 1 km². The selection of steep areas should be converted to polygons. • Parks will also be represented by polygons; these should be retrieved from dedicated thematic layers. The polygons representing waterways, steep slopes and parks are merged into a common polygon layer. Next, only the areas in the close neighbourhood of sHDCs should be taken into account for this special potential gap or separation filling. To assess this spatial relationship, each of the sHDCs is expanded by applying a buffer. The size of this buffer should be between 500 m and 2 000 m depending on the local circumstances (in other words, depending on the size of the water bodies, areas with steep slopes and parks). Then the common polygon layer for waterways, steep slopes and parks is intersected with the expanded sHDCs. Hence, the aim is to keep only those parts of waterways, steep slopes and parks that are located close to a sHDC. The selected waterways, steep slopes and parks are converted to 1 km² grid cells by selecting those cells that are at least 50 % covered by the common polygon layer for waterways, steep slopes and parks. In the next step, the grid cells of selected waterways, steep slopes and parks are merged with the sHDC grid cells. If this results in changes to the boundaries of the sHDCs, the result can be twofold: • two or more sHDCs are linked by the grid cells added for waterways, steep slopes and parks; • the coverage of a single sHDC has been expanded by adding adjacent grid cells for waterways, steep slopes and parks. The goal of this adapted methodology is to capture only the first case when overlaying the adjusted sHDCs with the original ones. If an adjusted sHDC contains more than one original sHDC then the adjustment should be kept; a new sHDC has been created, covering two or more original sHDCs. If the adjusted sHDC only contains a single original sHDC then the adjustment should be discarded, reverting to the original classification of grid cells (as there is no need to expand the sHDC by adding nearby waterways, steep slopes or parks). Map 8.5: Grid cell classification, Canberra, Map 8.6: Water and parks, Canberra, Australia Australia Dense urban cluster Suburban cells Low-density rural grid cells Water 0 5 km 0 5 km REGIOgis Very low-density rural grid cells REGIOgis Parks Applying the Degree of Urbanisation — 2021 edition  69 8 Which spatial units to use and adjustments to address geographic issues Only those new sHDC which reach a minimum population threshold of 50 000 inhabitants are kept. Thereafter, the normal smoothing and gap-filling process is applied to turn them into an urban centre. Adjusting the results for cities As the degree of urbanisation classification and the functional urban area classification share a common definition of cities, any changes that are made to the delineation of cities should be adopted for both of these classifications (using the same rules). More information on adjustments that might be made when delineating cities is provided in Subchapter 7.2.4. Map 8.7: Dense urban clusters and cells covered Map 8.8: Grid cell classification taking into by water and/or parks, Canberra, Australia account water and parks, Canberra, Australia Urban centre Urban centre Dense urban cluster Dense urban cluster Suburban cells Cell covered at least 50 % by water and parks Low-density rural grid cells 0 5 km 0 5 km REGIOgis within the 2 km bu er REGIOgis Very low-density rural grid cells References Gehlke, C. E. and K. Biehl (1934), ‘Certain Effects of Grouping upon the Size of the Correlation Coefficient in Census Tract Material’, Journal of the American Statistical Association Supplement, Volume 29, Issue 185A, pp. 169-170. Openshaw, S. (1984), The Modifiable Areal Unit Problem, CATMOG 38, Geo Books, Norwich. 70   Applying the Degree of Urbanisation — 2021 edition Selected indicators for 9 sustainable development goals by degree of urbanisation and functional urban area The methodology described in this manual has been developed to facilitate the international comparison of cities and urban and rural areas. The UN’s sustainable development goals (SDGs) include numerous indicators that should be compiled for individual cities or for urban and rural areas. This chapter shows that many of these indicators can already be calculated by degree of urbanisation using a wide variety of sources. These examples not only show the feasibility of this approach, but also underscore its interest. In particular, they show the benefit of compiling data separately for cities, towns and semi-dense areas, and rural areas. In most countries, these indicators follow a clear urban gradient with an increasing or decreasing performance as one moves from one end of the continuum, through towns and semi-dense areas, to the other end of the continuum. The degree of urbanisation classification can be used with a wide variety of data sources. It can be integrated into household surveys: for example, the European Union labour force survey (EU-LFS) codes its respondents according to level 1 of the degree of urbanisation classification using the municipality in which the respondent lives. Face-to- face interviews are increasingly geo-coded, which makes the application of the degree of urbanisation even easier. For example, recent Demographic and Health Surveys (USAID/WHO) and the face-to-face World Poll (Gallup) are all geo-coded. To ensure robust results, these surveys should have a large enough sample in each of the degree of urbanisation classes. As a result, it is easier to produce data by level 1 of the degree of urbanisation classification using surveys than by level 2 or by individual functional urban area. Therefore, producing SDG indicators by degree of urbanisation level 1 is considered the most suitable approach for international comparisons. The degree of urbanisation classification can also be used with geospatial data, such as remote sensing and point locations. For example, air pollution, changes in the built-up area and the distance to the nearest health facility can all be calculated by degree of urbanisation. The examples below are organised by SDG and include one or more examples for most, but not all, goals. One of the many benefits of geospatial data is that they typically cover the entire territory. As a result, indicators can be reliably provided not only for level 1 of the degree of urbanisation classification, but also for level 2 and even for individual cities and functional urban areas. Applying the Degree of Urbanisation — 2021 edition  71 9 Selected indicators for sustainable development goals by degree of urbanisation and functional urban area SDG 1 — END POVERTY IN ALL ITS FORMS EVERYWHERE Securing tenure rights may help ensure sustainable social and economic opportunities that contribute to eradicating poverty and hunger. Such rights are considered key to responsible land governance, enhancing the productive use of land through efficient and effective appropriation. Prindex (1) collects data, by degree of urbanisation, about how secure people feel their property rights are. It shows that perceived tenure insecurity for the main property was generally higher among adults living in cities than it was for adults living in rural areas. Across all 76 countries for which data were collected, perceived tenure insecurity was, on average, 5 percentage points higher for adults living in cities compared with adults living in rural areas. Towns and semi-dense areas occupied an intermediate position: as perceived tenure insecurity was 2 percentage points higher than in rural areas, but 3 percentage points lower than in cities. The data presented by Prindex are collected through interviews with a nationally representative sample of adults aged 18 years or older. The data presented refer to the main property that a respondent has rights to access or use. The indicator assesses perceived tenure security using the question: ‘in the next five years, how likely or unlikely is it that you could lose the right to use this property, or part of this property, against your will?’ People who consider it ‘somewhat likely’ or ‘very likely’ are classified as insecure. This indicator may be used to analyse progress towards SDG 1.4.2 — the proportion of the total adult population with secure tenure rights to land, with legally recognised documentation and who perceive their rights to land as secure; the only difference being that it refers to each individual’s main property instead of tenured land. The analysis that is presented may be extended to other land or to property when referring to the publicly available raw dataset and its methodology (Prindex (2020)). (1) Prindex: measuring global perceptions of land and property right (https://www.prindex.net/data/). Figure 9.1: Share of adult population aged 18 years or older with tenure insecurity, by degree of urbanisation, selected countries, 2019 (%) 50 45 40 35 30 25 20 15 10 5 0 Philippines Gabon Republic of Congo Botswana Ethiopia Zimbabwe Malaysia South Africa Bangladesh Sierra Leone Mali Afghanistan Guatemala Sri Lanka Guinea Dominican Republic Togo Mauritania Laos Mongolia Chad Lebanon Brazil Nicaragua Georgia Egypt Gambia Algeria Pakistan El Salvador Nepal Moldova Argentina Paraguay Uruguay Belarus Uzbekistan Turkmenistan Iraq India Chile Cities Towns and semi-dense areas Rural areas Source: Prindex 72   Applying the Degree of Urbanisation — 2021 edition Selected indicators for sustainable development goals by degree of urbanisation and functional urban area 9 Figure 9.2 shows the share of the population at risk of poverty for a number of European countries. A household is classified as being at risk of poverty if its income is below 60 % of the national equivalised median income after taxes and transfers. This is an example for SDG indicator 1.2.1: it reveals significant disparities in the situation along the urban-rural continuum. In around 40 % of European countries, the poverty rate was (considerably) higher in rural areas than in cities. This was most notably the case in countries with relatively low ratios of GDP per inhabitant, for example Bulgaria and Romania. In several western and northern European countries with higher levels of GDP per inhabitant, the risk of poverty was higher in cities than it was in towns and semi-dense areas, or rural areas. This was the case in Austria, Belgium, Denmark, the United Kingdom, the Netherlands, Germany, Norway and Switzerland. Figure 9.2: Share of the population at risk of poverty, by degree of urbanisation, selected European countries, 2017 (%) 40 30 20 10 0 Bulgaria Romania Serbia Croatia Latvia Spain Greece Estonia Portugal Italy Malta Poland Cyprus Ireland Belgium Hungary Luxembourg Sweden Switzerland Germany Slovakia Slovenia United Kingdom Finland Norway Netherlands France Austria Denmark Lithuania Czechia Cities Towns and semi-dense areas Rural areas Source: Eurostat (online data code: ilc_li43) Applying the Degree of Urbanisation — 2021 edition  73 9 Selected indicators for sustainable development goals by degree of urbanisation and functional urban area SDG 2 — END HUNGER, ACHIEVE FOOD SECURITY AND IMPROVED NUTRITION AND PROMOTE SUSTAINABLE AGRICULTURE Statistics on moderate or severe food insecurity are based on the food insecurity experience scale (FIES), as developed by the Food and Agriculture Organization of the United Nations (FAO). An FIES survey module forms part of the World Poll (Gallup), from which national estimates of the prevalence of moderate and severe food insecurity may be produced. For each country, this indicator was computed on combined sub-samples for each year in which geo-referenced data were available. Therefore, the statistics presented are not intended to be representative of the population by degree of urbanisation. Food insecurity is principally, but not exclusively, a rural problem: rural areas are often found to be significantly more food insecure than cities. Across the seven most food insecure countries shown in Figure 9.3, the prevalence of food insecurity at a moderate or severe level for the adult population living in rural areas was, on average, 11 percentage points higher than for the corresponding share recorded for people living in cities. For example, 73 % of the adult population living in rural areas of Botswana experienced this type of food insecurity during the period 2016-2018, compared with 60 % of adults who were living in cities. Rural areas were not systematically more food insecure than urban areas. For example, in Armenia, Mongolia, Bulgaria and Moldova there was little or no difference in the prevalence of food insecurity between adults living in cities and those living in rural areas. By contrast, food insecurity was significantly higher across the adult population living in the cities of Greece (22 %) than it was for the rural population (16 %). Figure 9.3: Share of the adult population aged 15 years or over facing moderate or severe food insecurity, by degree of urbanisation, 2016-2018 (%) 80 70 60 50 40 30 20 10 0 Botswana South Africa Philippines Cambodia Albania El Salvador Georgia Armenia Nepal Mongolia Moldova Ukraine Bulgaria Serbia Greece Cities Towns and semi-dense areas Rural areas Note: each data point is shown with error bars that indicate the 95 % confidence interval; in those cases where error bars by degree of urbanisation overlap, the differences between point estimates are not statistically significant. Source: FAO 74   Applying the Degree of Urbanisation — 2021 edition Selected indicators for sustainable development goals by degree of urbanisation and functional urban area 9 Among countries with a high overall prevalence of food insecurity, the share of adults living in towns and semi- dense areas facing food insecurity was generally situated between the extremes observed for people living in cities and those living in rural areas. Food insecurity for adults living in towns and semi-dense areas was lower than the share recorded for people living in rural areas for seven of the countries shown in Figure 9.3, while there were nine where the prevalence of food insecurity among adults living in towns and semi-dense areas was higher than the share recorded for people living in cities. Across the three classes of the degree of urbanisation, the prevalence of food insecurity was lowest for adults living in towns and semi-dense areas of six of the countries shown. By contrast, adults living in towns and semi-dense areas of Serbia were considerably more likely to face food insecurity (than those living in cities or in rural areas); this pattern was repeated (although it was far less pronounced) in Nepal. SDG 3 — ENSURE HEALTHY LIVES AND PROMOTE WELL-BEING FOR ALL AT ALL AGES In most countries covered by the Demographic and Health Survey (USAID), infant mortality is notably higher in rural areas than in cities (see Figure 9.4). In six countries (Mali, Nigeria, Lesotho, Guinea, Cambodia and Angola) the infant mortality rate was at least 20 deaths per 1 000 live births higher in rural areas than it was in cities. In a few countries, cities had a higher infant mortality rate, but the difference tended to be smaller. In five countries (Mozambique, Haiti, Kenya, Zambia and Tanzania), the infant mortality was between 5 and 10 deaths per 1 000 live births higher in cities than in rural areas. Note: this is not an SDG indicator, but it is closely linked to the under-5 mortality rate and the neo-natal mortality rate (respectively SDG 3.2.1 and SDG 3.2.2). Figure 9.4: Infant mortality rate, by degree of urbanisation, selected countries, 2012-2016 (per 1 000 live births) 100 90 80 70 60 50 40 30 20 10 0 Sierra Leone Chad Haiti Ivory Coast Mozambique DR Congo Cameroon Zambia Burkina Faso Zimbabwe Angola Mali Bangladesh Nambia Dominican Republic Rwanda Nepal Burundi Tanzania Nigeria Guinea Ethiopia Liberia Gabon Senegal Malawi Togo Kenya Ghana Lesotho Benin Uganda Comoros India Myanmar Timor-Leste Honduras Guatemala Colombia Cambodia Cities Towns and semi-dense areas Rural areas Note: the infant mortality rate is defined as the probability of a child dying before their first birthday and is expressed per 1 000 live births; the sample is limited to births that took place between one and five years prior to the interview. Source: Demographic and Health Survey as calculated by Henderson et al. (2020 Applying the Degree of Urbanisation — 2021 edition  75 9 Selected indicators for sustainable development goals by degree of urbanisation and functional urban area Infant mortality may be influenced by the distance to the nearest health facility, which tends to be larger in rural areas than in cities; Figure 9.5 shows this distance for a selection of sub-Saharan countries. As these data are very comprehensive, data can be calculated for level 2 of the degree of urbanisation classification. This reveals a very clear urban-rural gradient with distances increasing from cities to suburbs, to towns, to villages and so on. In cities, the nearest health facility was, on average, only 1.7 km away, less than a 30-minute walk. People living in suburbs were generally closer to a health facility (on average 2.5 km) than people living in dense and semi- dense towns (3.2 km and 3.8 km respectively). Within rural areas, those living in villages tended to live closest to the nearest health facility (4.7 km) followed by people living in dispersed rural areas (5.6 km), while people living in mostly uninhabited areas had the furthest distance to travel (12 km), equivalent to a three-hour walk. Note: this is not an SDG indicator but it is closely linked to health worker density and distribution (SDG 3.c.1) and the proportion of health facilities that have a core set of relevant essential medicines available and affordable on a sustainable basis (SDG 3.b.3). Figure 9.5: Average distance to the nearest health care facility, by degree of urbanisation, sub-Saharan countries, 2012-2016 (km) 26 24 22 20 18 16 14 12 10 8 6 4 2 0 Cape Verde Kenya Mauritius Sao Tome and Principe Tanzania Burkina Faso DR Congo Cameroon Liberia Botswana Mali Comoros Somalia Ethiopia Angola Eritrea Zambia Seychelles Burundi Republic of Congo Senegal Benin Uganda Namibia South Africa Guinea Nigeria Cote d'Ivoire Sierra Leone Rwanda Gambia Equatorial Guinea Chad Niger Ghana Mauritania Mozambique Gabon Madagascar Malawi Djibouti Central African Republic Zimbabwe South Sudan Lesotho Swaziland Togo All sub-Saharan countries Dispersed rural areas Villages Suburban/peri-urban areas Semi-dense towns Dense towns Cities Source: JRC calculation using GHS-POP and data from Maina et al. (2019) 76   Applying the Degree of Urbanisation — 2021 edition Selected indicators for sustainable development goals by degree of urbanisation and functional urban area 9 SDG 4 — ENSURE INCLUSIVE AND EQUITABLE QUALITY EDUCATION AND PROMOTE LIFELONG LEARNING OPPORTUNITIES FOR ALL In virtually all of the countries shown in Figure 9.6, 16-year-olds living in cities are far more likely to have completed eight years of schooling compared with those living in rural areas. Across the selected countries that are shown, 55 % of 16-year-olds living in cities had completed eight years of schooling compared with only 31 % in rural areas. The share of 16-year-olds living in towns and semi-dense areas that had completed eight years of schooling was in between, at 41 %. The only exceptions (among those countries shown) to the pattern described above were: India and Bangladesh where the differences by degree of urbanisation were very small; Kenya where 16-year-olds living in towns and semi-dense areas were most likely to have completed eight years of schooling, followed by those living in rural areas with a slightly lower share recorded for those living in cities. Note: this is not an SDG indicator, but it is closely linked to the proportion of children and young people (a) in grades 2/3; (b) at the end of primary education; and (c) at the end of lower secondary education achieving at least a minimum proficiency level in (i) reading and (ii) mathematics, by sex (SDG 4.1.1). Figure 9.6: Share of 16-year-olds having completed eight years of schooling, by degree of urbanisation, selected countries, 2012-2016 (%) 90 80 70 60 50 40 30 20 10 0 Zimbabwe Nambia Malawi Ghana Guatemala DR Congo Angola Haiti Burkina Faso Liberia Nepal India Timor-Leste Colombia Myanmar Dominican Republic Nigeria Comoros Honduras Zambia Sierra Leone Benin Cambodia Cameroon Bangladesh Lesotho Kenya Gabon Togo Uganda Ivory Coast Tanzania Chad Ethiopia Senegal Mali Mozambique Guinea Rwanda Burundi Cities Towns and semi-dense areas Rural areas Source: Demographic and Health Survey as calculated by Henderson et al. (2020) Applying the Degree of Urbanisation — 2021 edition  77 9 Selected indicators for sustainable development goals by degree of urbanisation and functional urban area SDG 5 — ACHIEVE GENDER EQUALITY AND EMPOWER ALL WOMEN AND GIRLS Among the countries shown in Figure 9.7, on average 29 % of married women living in rural areas had experienced domestic violence, compared with 28 % for married women living in cities and 27 % for towns and semi-dense areas. In some countries, the share of married women having experienced domestic violence was considerably higher for those living in rural areas compared with those living in cities, for example in Uganda the difference was 19 percentage points and in Timor-Leste it was 17 points. In Mozambique, however, this pattern was reversed as married women living in cities were more likely to have experienced domestic violence than those living in rural areas (with a gap of 10 percentage points). This indicator captures SDG 5.2.1 with the only difference being that it does not ask if the domestic violence experienced by married women occurred during the 12 months prior to the Demographic and Health Survey. Figure 9.7: Share of married women who have been the victim of domestic violence, by degree of urbanisation, selected countries, 2012-2016 (%) 60 50 40 30 20 10 0 DR Congo Sierra Leone Gabon Cameroon Zambia Kenya Angola Zimbabwe Colombia Tanzania Ivory Coast Mozambique Mali Rwanda Uganda Malawi India Chad Timor-Leste Haiti Nepal Honduras Namibia Ethiopia Togo Dominican Republic Guatemala Nigeria Cambodia Comoros Burkina Faso Myanmar Cities Towns and semi-dense areas Rural areas Source: Demographic and Health Survey as calculated by Henderson et al. (2020) 78   Applying the Degree of Urbanisation — 2021 edition Selected indicators for sustainable development goals by degree of urbanisation and functional urban area 9 SDG 6 — ENSURE AVAILABILITY AND SUSTAINABLE MANAGEMENT OF WATER AND SANITATION FOR ALL Figure 9.8 shows that in most countries included in the Demographic and Health Survey a higher share of households in cities had access to safely managed drinking water than the share recorded for households in towns and semi-dense areas, which in turn had a higher share than for households in rural areas. On average, across all of the countries shown, 56 % of households in cities had access to safely managed drinking water compared with 26 % of households in rural areas, while households in towns and semi-dense areas had an intermediate share (37 %). This indicator corresponds to SDG 6.1.1. Figure 9.8: Share of households having access to safely managed drinking water, by degree of urbanisation, selected countries, 2010-2016 (%) 100 90 80 70 60 50 40 30 20 10 0 Bangladesh India Lesotho Ivory Coast Angola Burkina Faso Colombia Honduras Timor-Leste Guatemala Dominican Republic Senegal Gabon Nepal Nambia Zimbabwe Uganda Ghana Comoros Malawi Mali Ethiopia Guinea Zambia Rwanda Mozambique Kenya Chad DR Congo Togo Liberia Myanmar Tanzania Benin Cameroon Sierra Leone Nigeria Burundi Haiti Cities Towns and semi-dense areas Rural areas Note: safely managed drinking water is defined by the DHS-WHO Joint Monitoring Programme as all improved water sources that take zero minutes to collect or are on the premises; improved water sources encompass all piped water and packaged water, as well as protected wells or springs, boreholes, and rainwater. Source: Demographic and Health Survey as calculated by Henderson et al. (2020) Applying the Degree of Urbanisation — 2021 edition  79 9 Selected indicators for sustainable development goals by degree of urbanisation and functional urban area SDG 7 — ENSURE ACCESS TO AFFORDABLE, RELIABLE, SUSTAINABLE AND MODERN ENERGY FOR ALL The share of households in cities with access to electricity was generally much higher than that recorded for households in rural areas. On average, across all of the countries shown in Figure 9.9, 73 % of households in cities had access to electricity compared with 31 % in rural areas. Households in towns and semi-dense areas had an intermediate share (45 % had access to electricity). In 11 out of the 39 countries shown in Figure 9.9, the share of households in rural areas with access to electricity was within the range of 0-10 %. This indicator corresponds to SDG 7.1.1. Figure 9.9: Share of households having access to electricity, by degree of urbanisation, selected countries, 2016 (%) 100 90 80 70 60 50 40 30 20 10 0 Dominican Republic Gabon Ivory Coast Comoros Zimbabwe Bangladesh Togo Uganda Haiti Angola Lesotho Zambia Mozambique Ethiopia Sierra Leone Malawi Burkina Faso DR Congo Colombia Nepal Timor-Leste Guatemala Cambodia India Ghana Cameroon Myanmar Nigeria Senegal Benin Nambia Guinea Rwanda Mali Tanzania Kenya Chad Liberia Burundi Cities Towns and semi-dense areas Rural areas Source: Demographic and Health Survey as calculated by Henderson et al. (2020) SDG 8 — PROMOTE SUSTAINED, INCLUSIVE AND SUSTAINABLE ECONOMIC GROWTH, FULL AND PRODUCTIVE EMPLOYMENT AND DECENT WORK FOR ALL Financial services can help people to escape poverty: for example, they can make it possible for people to invest in education, to finance health care or to start a business. Having a bank account is a first important step to accessing such services or taking such initiatives. A bank account also makes it easier to manage payments safely. However, most people in low-income countries do not have a bank account. The share of the adult population (persons aged 15 years or over) living in low-income countries with a bank account was highest in cities (30 % of adult city-dwellers had a bank account; see Figure 9.10). A much lower share (18 %) of the adult population in rural areas of low-income countries had a bank account. By contrast, the share of the population with a bank account in high-income countries was above 80 % for all three classes by degree of urbanisation. In the two groups of middle income countries, adults living in rural areas were also less likely to have a bank account than people living in towns and semi-dense areas or in cities. 80   Applying the Degree of Urbanisation — 2021 edition Selected indicators for sustainable development goals by degree of urbanisation and functional urban area 9 Figure 9.11 shows that in most Figure 9.10: Share of the population aged 15 years or over with a bank European countries the share of young account, by degree of urbanisation and income group, 2017 people (aged 15-24 years) neither in (%) employment nor in education or training (the NEET rate) was often considerably 90 higher for young people living in rural 80 areas than it was for those living in cities; this was most notably the case in 70 Bulgaria, Greece, Romania and Hungary. However, in six of the countries shown, 60 the NEET rate was higher for young people living in cities than it was for 50 young people living in towns and semi-dense areas or in rural areas; this 40 was most notably the case in Belgium 30 and Austria, and was also observed in Slovenia, Malta, the United Kingdom 20 and the Netherlands. This indicator corresponds to SDG 8.6.1. 10 0 Low-income Lower-middle Upper-middle High-income countries income countries income countries countries Rural areas Towns and semi-dense areas Cities Source: Global Findex (2017) Figure 9.11: Share of young people (aged 15-24 years) neither in employment nor in education or training, by degree of urbanisation, selected European countries, 2018 (%) 40 30 20 10 0 Italy Bulgaria Greece Romania Cyprus Croatia Hungary France Slovakia Spain Lithuania Ireland Poland Finland Latvia Portugal United Kingdom Estonia Belgium Sweden Norway Slovenia Czechia Malta Luxembourg Germany Austria Switzerland Netherlands Denmark Cities Towns and semi-dense areas Rural areas Source: Eurostat (online data code: edat_lfse_29) Applying the Degree of Urbanisation — 2021 edition  81 9 Selected indicators for sustainable development goals by degree of urbanisation and functional urban area SDG 9 — PROPORTION OF POPULATION COVERED BY A MOBILE NETWORK, BY TECHNOLOGY Mobile phone ownership has increased over the last few decades. Nevertheless, only half the rural population living in low-income countries owned a mobile phone, compared with almost three quarters of city-dwellers living in low-income countries (see Figure 9.12). The gap in mobile phone access between rural areas and cities narrows as average income levels increased. Nevertheless, in high-income countries there remained a 5 percentage point gap in mobile phone ownership in favour of city-dwellers. Note that this indicator differs from the core SDG indicator 9.c.1 in that it measures mobile phone ownership and not the population covered by a mobile network. Figure 9.12: Share of the population aged 15 years or over with a mobile phone, by degree of urbanisation and income level, 2016-2018 (%) 100 90 80 70 60 50 40 30 20 10 0 Low-income Lower-middle Upper-middle High-income countries income countries income countries countries Rural areas Towns and semi-dense areas Cities Source: Gallup World Poll 82   Applying the Degree of Urbanisation — 2021 edition Selected indicators for sustainable development goals by degree of urbanisation and functional urban area 9 SDG 11 — MAKE CITIES AND HUMAN SETTLEMENTS INCLUSIVE, SAFE, RESILIENT AND SUSTAINABLE Access to public transport in cities is considered critical to encourage low-carbon mobility and ensure that people can get where they need or want to go. This is especially the case for those people who cannot drive, do not want to drive or cannot afford to drive. The core SDG indicator 11.2.1 measures the share of city-dwellers living within 500 m walking distance of a transport stop. A secondary indicator takes into account the frequency of departures and expands the distance under consideration so that transport stops within a 1 km radius by foot are taken into account if they provide access to a faster mode of transport (such as bus rapid transit, metro or rail). Figure 9.13 shows this secondary indicator. The selected South American cities and most of the selected European cities had a relatively high level of access to public transport with a high frequency of departures. In the selected cities of North America and Oceania, access to public transport was somewhat lower (in particular in Houston and Atlanta), while the frequency of departures was generally lower than in European or South American cities. In the selected cities of Africa and Asia, the situation was more mixed. Some cities, including Cape Town, Taichung or Tel Aviv, offered a relatively high level of access combined with a relatively high frequency of departures. In most other cities selected for Africa and Asia, less than half the population had access to public transport. Figure 9.13: Share of city-dwellers with access to public transport by frequency of departure, selected cities, 2015- 2019 (%) 100 90 80 70 60 50 40 30 20 10 0 Accra Cairo Douala Kigali Nairobi Cape Town St Petersburg Oslo Chiang Mai Manila Enzan Taichung Tel Aviv Buenos Aires Houston Atlanta Washington Boston Vancouver Toronto Auckland Sydney Vilnius Reykjavík Rīga Dublin Stockholm Ljubljana Warszawa København Amsterdam Berlin Tallinn Praha Helsinki Roma London Athina Budapest Bern Luxembourg Brussels Wien Paris Madrid São Paulo South North Africa Asia America America Oceania Europe Low frequency (< 4 departures per hour) Medium frequency (4-10 departures per hour) High frequency (> 10 departures per hour) Source: European Commission and International Transport Forum calculated using General Transit Feed Specification (GTFS) data from various sources and population data from GHS-POP Applying the Degree of Urbanisation — 2021 edition  83 9 Selected indicators for sustainable development goals by degree of urbanisation and functional urban area To measure sustainable urbanisation, SDG indicator 11.3.1 is based on the ratio between land use change and population change. The methodology proposed for this indicator is rather complex (a unitless ratio of two logarithmic changes derived from boundaries that change over time). The indicator presented in Map 9.1 is simpler. It compares the amount of built-up land per person for two points in time using the most recent metropolitan boundary. This means that the indicator has a more understandable unit (built-up land in m² per person) and the changes can be compared with the amount of built-up land per person for the first reference period. The amount of built-up land is a secondary indicator for SDG 11.3.1. Map 9.1 shows that most metropolitan areas in the world reduced their ratio of built-up land per inhabitant between 2000 and 2015 (those metropolitan areas shaded in green). Some metropolitan areas increased their amount of built-up land per inhabitant because their built-up land grew at a faster rate than their total number of inhabitants or because their total number of inhabitants declined, as was the case for many metropolitan areas of China, central Asia and eastern Europe. The data for metropolitan areas reducing their amount of built-up land per inhabitant should be interpreted cautiously and with regard to the initial level of built-up land. Those with very low amounts of built-up land per inhabitant may be characterised by low levels of infrastructure and high numbers of inhabitants living in crowded conditions. Map 9.1: Change in the ratio of built-up land per inhabitant, selected metropolitan areas, 2000-2015 ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 0 5 000 km ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! Change in built-up area per inhabitant in metropolitan areas 2000-2015 m² per inhabitant Population < −20 250 000 - 500 000 −20 - −10 500 000 - 1 000 000 ! ! ! ! −10 - −5 1 000 000 - 5 000 000 ! ! −5 - 0 5 000 000 - 10 000 000 0-5 10 000 000 - 20 000 000 5 - 10 > 20 000 000 10 - 20 Only functional urban areas with at least 0 1 000 km 0 1 000 km > 20 250 000 inhabitants in 2015 are shown. Source: GHS-BUILT using boundaries from Moreno-Monroy et al. (2020) 84   Applying the Degree of Urbanisation — 2021 edition Selected indicators for sustainable development goals by degree of urbanisation and functional urban area 9 The spatial concentration of people and economic activities in cities can lead to high levels of air pollution, which may potentially harm people’s health and reduce their life expectancy, as well as having other consequences. Many cities in China and India had high concentrations of fine particulate matter (PM2.5 — particles with a diameter of 2.5 micrometres (µm) or less) of at least 60 micrograms per cubic metre (µg/m³), which was six times higher than the World Health Organisation’s limit for protecting human health (10 µg/m³). Map 9.2: Annual mean concentration of fine particulate matter (PM2.5), selected cities, 2014 ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 0 5 000 km ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! Mean concentration of PM2.5 in cities, 2014 μg/m³ Population < 10 250 000 - 500 000 10 - 25 500 000 - 1 000 000 ! ! ! ! ! ! 25- 45 1 000 000 - 5 000 000 45 - 60 5 000 000 - 10 000 000 > 60 10 000 000 - 20 000 000 > 20 000 000 0 1 000 km 0 1 000 km Only cities with at least 250 000 inhabitants in 2015 are shown. Source: JRC Urban Centre Database from Florczyk et al. (2019) Applying the Degree of Urbanisation — 2021 edition  85 9 Selected indicators for sustainable development goals by degree of urbanisation and functional urban area SDG 16 — PROMOTE PEACEFUL AND INCLUSIVE SOCIETIES FOR SUSTAINABLE DEVELOPMENT, PROVIDE ACCESS TO JUSTICE FOR ALL AND BUILD EFFECTIVE, ACCOUNTABLE AND INCLUSIVE INSTITUTIONS AT ALL LEVELS People living in rural areas are more likely to feel safe when they are walking alone at night than city-dwellers. This information is covered by SDG indicator 16.1.4. People living in rural areas felt safer walking alone at night than people living in cities for all four groups of countries based on average income levels (as shown in Figure 9.14). This urban gradient was clearly visible for low-, upper- middle and high-income countries. The gap in the proportion of people feeling safe between those living in rural areas and those living in cities was greater for high-income and upper-middle income countries than it was for low- income countries. In lower-middle income countries, people living in towns and semi-dense areas felt safer walking alone than people living in rural areas or in cities. Figure 9.14: Share of the population aged 15 years or over who considered it was safe to walk alone at night, by degree of urbanisation and income group, 2016-2018 (%) 90 80 70 60 50 40 30 20 10 0 Low-income Lower-middle Upper-middle High-income countries income countries income countries countries Rural areas Towns and semi-dense areas Cities Source: Gallup World Poll 86   Applying the Degree of Urbanisation — 2021 edition Selected indicators for sustainable development goals by degree of urbanisation and functional urban area 9 SDG 17 — STRENGTHEN THE MEANS OF IMPLEMENTATION AND REVITALISE THE GLOBAL PARTNERSHIP FOR SUSTAINABLE DEVELOPMENT SDG indicator 17.8.1 concerns use of the internet. Cities typically have a higher share of internet use than rural areas (see Figure 9.15). The gap between cities and rural areas was biggest in low-income countries where, on average, 54 % of people aged 15 years or over in rural areas used the internet in the seven days prior to the Gallup World Poll survey, compared with 75 % in cities. As the average income level in a country goes up, the gap in internet use between rural areas and cities tends to narrow. Nevertheless, a 5 percentage point gap remained for high-income countries. Figure 9.15: Share of the population aged 15 years or over having made use of the internet in the previous seven days, by degree of urbanisation and income group, 2016-2018 (%) 100 90 80 70 60 50 40 30 20 10 0 Low-income Lower-middle Upper-middle High-income countries income countries income countries countries Rural areas Towns and semi-dense areas Cities Source: Gallup World Poll Applying the Degree of Urbanisation — 2021 edition  87 9 Selected indicators for sustainable development goals by degree of urbanisation and functional urban area References Dijkstra, L. and H. Poelman (2012), ‘Cities in Europe: the new OECD-EC definition’, Regional Focus, No. 01/2012, European Commission. Eurostat (2019), Methodological manual on territorial typologies — 2018 edition, Publications Office of the European Union, Luxembourg. Florczyk, A., C. Corbane, M. Schiavina, M. Pesaresi, L. Maffenini, M. Melchiorri, P. Politis, F. Sabo, S. Freire, D. Ehrlich, T. Kemper, P. Tommasi, D. Airaghi and L. Zanchetta (2019), GHS Urban Centre Database 2015, multitemporal and multidimensional attributes, European Commission. Global Findex (2017), The Global Findex Database 2017, The World Bank. Henderson, J. V., V. Liu, C. Peng and A. Storeygard (2020), Demographic and health outcomes by degree of urbanisation — perspectives from a new classification of urban areas, Publications Office of the European Union, Luxembourg. Maina, J., P. O. Ouma, P. M. Macharia, V. A. Alegana, B. Mitto, I. S. Fall, A. M. Noor, R. W. Snow and E. A. Okiro (2019), ‘A spatial database of health facilities managed by the public health sector in sub Saharan Africa’, Scientific Data, Volume 6(1). Moreno-Monroy, A. I., M. Schiavina and P. Veneri (2020), ’Metropolitan areas in the world. Delineation and population trends’, Journal of Urban Economics. Prindex (2020) Comparative report: A global assessment of perceived tenure security from 140 countries, London. 88   Applying the Degree of Urbanisation — 2021 edition 10 Tools and training The degree of urbanisation classification is a geospatial concept that can be implemented in geographic information systems (GIS) — computer systems designed to analyse spatial data. However, this requires adequate expertise to operate the GIS in an appropriate way and the availability of the necessary population and, optionally, built-up density grids. There is a strong demand for ready-to-use tools that facilitate the application of the degree of urbanisation classification as well as for capacity building that assures a conscious implementation of the methodology on which it is based. This chapter describes tools that are currently available and training materials that have been produced by the European Commission’s Joint Research Centre (JRC) Global Human Settlement Layer (GHSL) project to support the development of a harmonised global definition of cities and settlements. 10.1 Tools The tools described in this subchapter address three production steps that are described in the previous chapters. The first step is the construction of a regular-spaced population grid from given geospatial population data in the form of points or polygons (see Chapter 5). The second step is the application of the methodology to a given population grid and additional optional layers (see Chapter 6 and Chapter 7). Finally, in the last step, the derived grid cell classification is used to classify small spatial units into cities, towns and semi dense areas, or rural areas (see also Chapters 6 and 7). Figure 10.1 displays the workflow to operationalise the tools that have been produced within the framework of the GHSL. Figure 10.1: Conceptual workflow to apply the methodology using GHSL tools Stage 1 Stage 1 grid classi cation territorial unit classi cation Tool GHS-POP2G GHS-DUG GHS-DU-TUC Population layer Population grid Population grid Input Built-up area layer Built-up area layer Land layer Settlement grid Regions layer Territorial units layer Settlement grid Units classi cation layer Output Population grid statistics statistics Applying the Degree of Urbanisation — 2021 edition  89 10 Tools and training The tools described below are open and available free-of-charge from the GHSL tools website (1). They require the installation of MATLAB Runtime (2), which is a standalone set of shared libraries that enable the execution of compiled MATLAB applications. The tools are also available as an ArcGIS toolbox, compatible with ArcMap 10.6. The tools were developed to run on standard computers. They all run on Windows 10 operating systems with any processor and require at least 16GB RAM. It is important to note that more memory is required for processing larger data sets. More details can be found in the corresponding user manuals (see below for more information on specific user guides). 10.1.1 CONSTRUCTION OF A POPULATION GRID (POPULATION TO GRID TOOL — GHS-POP2G) A population grid is the key input to produce the grid cell classification that is necessary in order to compile data by degree of urbanisation. A population grid is obtained by re-allocating population counts from points and/ or polygons to gridded surfaces of regular and standardised grid cells or pixels. The population grid is produced through geospatial and geo-statistical processing of geo-coded population data (as available). Population grids can be produced in alternative ways depending on the type of data available. One process is that of aggregation. The aggregation approach is generally used when micro-census source data have higher spatial detail (resolution) than the selected cell size of the population grid. A point-based micro-census is usually conducted at the building or census block level, and this high level of spatial detail should be the only one for which this aggregation technique should be deployed. Population grids are more generally produced through disaggregation of population counts attached to small spatial units — statistical areas or administrative units. The GHS population grid layers (GHS-POP) are produced through disaggregation (the population input is the Gridded Population of the World v4.10 (CIESIN (2018)). The disaggregation is driven by the density of built-up areas as a proxy for the location of the resident population. To support the uptake of this methodology, the GHSL project has developed a population to grid tool — GHS-POP2G (version 2). This is a flexible tool to produce geospatial population grids in GeoTIFF format from census data. It operationalises the workflow developed for the production of the GHS-POP. GHS-POP2G offers the possibility to create population grids at 50 m, 100 m, 250 m and 1 km spatial resolutions, handling census data stored as point or polygon vector data (the latter case requires an additional covariate as input for dasymetric disaggregation); it is available as standalone software or as an ArcGIS toolbox (see Figure 10.2). The principal purpose of the tool is to produce a population grid that may be used as an input for the degree of urbanisation grid tool (GHS-DUG) which has also been produced within the GHSL framework. However, potential uses of the tool and population grids extend far beyond this principal application. The GHS-POP2G user manual (Maffenini et al. (2020a)) explains all of the functionalities and requirements to run the tool. Figure 10.2: GHS-POP2G interface window — standalone tool (left); ArcGIS toolbox (right) (1) Global Human Settlement Layer (https://ghsl.jrc.ec.europa.eu/tools.php). (2) Available from MathWorks (https://mathworks.com/products/compiler/matlab-runtime.html). 90   Applying the Degree of Urbanisation — 2021 edition Tools and training 10 10.1.2 CLASSIFYING GRID CELLS (DEGREE OF URBANISATION GRID TOOL — GHS-DUG) The degree of urbanisation grid tool — GHS-DUG (version 4) is an information system to produce geospatial grids for degree of urbanisation classes and related statistics. GHS-DUG 4 is designed as a scalable tool allowing the application of methodology to available population grids or to data made available in the GHSL Data Package 2019 (Florczyk et al. (2019)). The GHS-DUG implements the workflow developed for the production of the GHS-SMOD. It produces a grid cell classification for the entire area of interest in GeoTiff format at 1 km spatial resolution according to both level 1 and level 2 of the degree of urbanisation classification. GHS-DUG requires a population grid (at 1 km resolution) and optionally a built-up surface and land fraction layers. When a shapefile delimiting territorial units is provided, the tool compiles statistics by degree of urbanisation class. The principal purpose of the tool is the production of a classification of grid cells by degree of urbanisation. The GHS-DUG grid output is used to operationalise stage 2 of the methodology (the classification of small spatial units) and is used as an input for the degree of urbanisation territorial unit classifier tool (GHS-DU-TUC) also produced within the GHSL framework (see Subchapter 10.1.3). The GHS-DUG user manual (Maffenini et al. (2020b)) explains all of the functionalities and requirements to run the tool. Figure 10.3 shows the graphical interface of the GHS-DUG tool both for the standalone tool and for the ArcGIS toolbox. Figure 10.3: GHS-DUG graphical interface — standalone tool (left); ArcGIS toolbox (right) Applying the Degree of Urbanisation — 2021 edition  91 10 Tools and training 10.1.3 CLASSIFYING SMALL SPATIAL UNITS (DEGREE OF URBANISATION TERRITORIAL UNIT CLASSIFIER TOOL — GHS-DU-TUC) The methodology classifies the entire territory of a country along the urban-rural continuum into regularly spaced grid cells. However, often it is required to classify small spatial units, for example a commune or municipality. The GHS-DU-TUC tool implements this transition from the grid cell classification to the classification of small spatial units based on the type of grid cells in which the majority of their population resides. The degree of urbanisation territorial unit classifier tool — GHS-DU-TUC (version 1.0) is designed as an operational tool that classifies small spatial units based on the grid cell classification already derived using the GHS-DUG tool. It requires the following inputs: a classification of grid cells, a population grid and a geospatial layer containing the small spatial units. The input population grid must be the one used for the production of the grid cell classification through the GHS-DUG tool. GHS-DU-TUC produces a geospatial layer in vector format (a shapefile) that contains the classification of small spatial units according to levels 1 and 2 of the degree of urbanisation classification, plus a statistical table with the classification of the small spatial (territorial) units and their population counts. The GHS-DU-TUC user manual (Maffenini et al. (2020c)) explains all of the functionalities and requirements to run the tool. Figure 10.4 shows the graphical interface of the GHS-DU-TUC tool both for the standalone tool and for the ArcGIS toolbox. Figure 10.4: GHS-DU-TUC graphical interface — standalone tool (left); ArcGIS toolbox (right) 10.2 Training The tools described in the previous subchapter are distributed with detailed manuals that encourage autonomous use (see References at the end of this subchapter for further details). Nevertheless, additional training materials or training courses are available to expedite the correct selection and application of the different options. In preparation for the 51st session of the UN Statistical Commission, partner organisations supported a range of countries in different ways to increase their capacity to understand and implement the methodology. UN-Habitat together with the European Commission organised seven regional workshops between 2018 and 2019 to present the methodology underlying the degree of urbanisation classification and discuss how this could be improved and applied to national data. A total of 85 countries participated in these workshops (see Figure 10.5). 92   Applying the Degree of Urbanisation — 2021 edition Tools and training 10 Figure 10.5: Overview of regional workshops presenting the methodology Abuja, Nigeria, Abidjan, Ivory Coast, Lusaka, Zambia, 15-19 October 2018 13-16 November 2018 22-25 January 2019 with representatives from: with representatives from: with representatives from: Nigeria, Burundi, Botswana, Ghana, Burkina Faso, Malawi, The Gambia, Central African Republic, Tanzania, Sierra Leone, Chad, Mauritius, Kenya, Republic of Congo, Angola, Ethiopia, Comoros, Zimbabwe, South Sudan, Democratic Republic of Congo, Mozambique, Liberia, Madagascar, South Africa, Uganda Djibouti, Eswatini, Mali, Lesotho, Niger, Namibia, Senegal, Zambia Guinea, Togo, Ivory Coast Cairo, Egypt, Lima, Peru, Delhi, India, 18-21 March 2019 25-28 June 2019 23-26 September 2019 with representatives from: with representatives from: with representatives from: Egypt, Argentina, Azerbaijan, Morocco, Bolivia, Armenia, Sudan, Brazil, Bangladesh, Tunisia, Chile, Bhutan, Bahrain, Costa Rica, India, Iraq, Colombia, Kyrgyzstan, Jordan, Cuba, Maldives, Kuwait, Dominican Republic, Nepal, Lebanon, Ecuador, Sri Lanka, Oman, Mexico, Uzbekistan Palestine, Peru, Saudi Arabia, Uruguay Syria, Yemen Kuala Lumpur, Malaysia, 22-25 October 2019 with representatives from: Afghanistan, Australia, China, Iran, Kazakhstan, Lao PDR, Malaysia, Mongolia, Myanmar, New Zealand, Thailand, Timor-Leste, Vietnam Applying the Degree of Urbanisation — 2021 edition  93 10 Tools and training As a follow-up to the workshops, the European Commission’s Joint Research Centre (JRC) conducted dedicated training in the United Arab Emirates at the request of the Federal Competitiveness and Statistical Authority, and at the UN-Habitat Headquarters in Kenya for UN-Habitat staff. Further events and a comprehensive training package are under preparation. The objective of training courses is to provide an overview of the data, methods and tools developed by the GHSL project, to provide examples of how data and tools can be used to apply the methodology, and which applications it can support. The course includes presentations and practical exercises. The presentations are targeted at a general audience with a background in regional and urban development and to those working for national statistical authorities; the practical exercises require some basic knowledge in GIS and spreadsheets and the installation of dedicated software prior to the exercise (see Subchapter 10.1 for more details of the specific requirements). Training courses address four broad themes through presentations: • The first module addresses the need for a global definition of urban and rural areas for international statistical comparisons. • The second module explains the GHSL datasets: the built-up area spatial grids (GHS-BUILT), the population spatial grids (GHS-POP), the settlement model spatial grid (GHS-MOD) and the urban centre database (GHS-UCDB). • The third module explains the JRC’s solution to operationalise the degree of urbanisation classification into a settlement classification grid and into the classification of small spatial units by degree of urbanisation for urban and rural areas. • The fourth module shows examples of GHSL data applications, to support policymaking with new findings on human settlements. Those taking part in the practical training exercises can expect to learn the following skills: • Construction of a population grid with the population to grid tool (GHS-POP2G). • Classification of grid cells with the degree of urbanisation grid tool (GHS-DUG). • Classification of small spatial units by degree of urbanisation with the degree of urbanisation territorial unit classifier tool (GHS-DU-TUC). • Disaggregation of statistics and indicators according to the degree of urbanisation classification. • Estimation of sustainable development goal (SDG) 11.3.1 for urban areas (LUE tool). The production of stand-alone online courses and webinars is under preparation. 94   Applying the Degree of Urbanisation — 2021 edition Tools and training 10 10.3 Online resources for the degree of urbanisation classification In order to support the discussions with interested countries and stakeholders, the JRC has also created a dedicated web presence for the degree of urbanisation classification (3). The homepage contains everything that is needed to understand and implement the degree of urbanisation classification (see Figure 10.6). Figure 10.6: The degree of urbanisation website (3) Global Human Settlement Layer (https://ghsl.jrc.ec.europa.eu/degurba.php). Applying the Degree of Urbanisation — 2021 edition  95 10 Tools and training The different sections include: • An introduction to the methodology: why there is a need for a global, people-based definition of cities and urban and rural areas. • A summary of the methodology. • Country fact sheets summarise the application of methodology based on data from the GHSL and publicly available country borders. • Interactive maps and the application of the methodology in the Urban Centres Database. • A data section that provides open data for the global grids of the GHSL dataset including built-up area grids, population grids and settlement classification grids (GHS-SMOD layers). • The tools section with links to the available set of tools for implementing the methodology. • A list of essential documents. • A section summarising materials and initiatives for capacity building. References Center for International Earth Science Information Network (CIESIN), Columbia University (2018), Documentation for the Gridded Population of the World, Version 4 (GPWv4), Revision 11 Data Sets, NASA Socioeconomic Data and Applications Center (SEDAC), Palisades, NY. Florczyk, A., C. Corbane, D. Ehrlich, S. Freire, T. Kemper, L. Maffenini, M. Melchiorri, M. Pesaresi, P. Politis, M. Schiavina, F. Sabo, and L. Zanchetta (2019), GHSL Data Package 2019, JRC 117104, EUR 29788 EN, Publications Office of the European Union, Luxembourg. Maffenini, L., M. Schiavina, S. Freire, M. Melchiorri, M. Pesaresi and T. Kemper (2020a), GHS-POP2G User Guide, JRC 121485, Publications Office of the European Union, Luxembourg. Maffenini, L., M. Schiavina, M. Melchiorri, M. Pesaresi, and T. Kemper (2020b), GHS-DUG User Guide, JRC 121484, Publications Office of the European Union, Luxembourg. Maffenini, L., M. Schiavina, M. Melchiorri, M. Pesaresi, and T. Kemper (2020c), GHS-DU-TUC User Guide, JRC 121486, Publications Office of the European Union, Luxembourg. 96   Applying the Degree of Urbanisation — 2021 edition 11 Conclusions The endorsement of the UN Statistical Commission in March 2020 of the methodology for the delineation of cities and urban and rural areas was a key milestone. However, work in this area is not over. As part of the endorsement process, the UN Statistical Commission made two additional requests. First, that a technical report on the implementation of the methodology for the delineation of cities and urban and rural areas was made available as quickly as possible; this manual responds to that request. Second, that the UN Statistics Division and the sponsoring organisations review the implementation of the methodology for the delineation of cities and urban and rural areas and report back to the UN Statistical Commission at one of its future sessions. As a result, the focus of the work will now shift to three different lines of action. First, to encourage and support countries applying the methodology for compiling statistics by degree of urbanisation (level 1). The current census round presents an opportunity to apply this methodology using data with a high spatial resolution. In particular, countries that have conducted or will conduct a digital census and collect the GPS location of all households can produce a high- quality population grid. Such a population grid will create a highly robust and accurate classification of a country’s settlements. This methodological manual presents a number of tools to make it easier to compile statistics by degree of urbanisation. Nevertheless, hands-on training and responding to specific questions will be needed to ensure that as many countries as possible apply the methodology in a consistent and coherent manner. Several of the organisations behind this work are ready to provide such training and technical support. This experience will then be summarised to report back on the implementation phase to the UN Statistical Commission. Second, to improve and update global data. To support this work, the Joint Research Centre (JRC) of the European Commission has produced a global, estimated population grid for the years 1975, 1990, 2000 and 2015. Using new imagery, the Sentinel 1 and 2 satellites and improved methods relying on artificial intelligence and cloud computing, the JRC will publish improved population grids and produce regular updates for free. This will ensure that national administrations, NGOs, the academic community and other interested parties have access to coherent, complete and up-to-date information. In addition, the JRC will explore how to project these population grids up to 2050 and even 2100 by incorporating the latest UN World Population Projections. Third, to integrate this new methodology in the documentation of the relevant sustainable development indicators. To facilitate the comparison of data for cities, towns and rural areas, the methodology should be included in the metadata of relevant SDG indicators. This will encourage more countries to produce the SDG indicators in such a way that they can be reliably compared across national borders. To this end, the organisations involved in this work will reach out to the custodian agencies of the various SDG indicators that might be analysed by degree of urbanisation (level 1). Applying the Degree of Urbanisation — 2021 edition  97 Getting in touch with the EU In person All over the European Union there are hundreds of Europe Direct Information Centres. You can find the address of the centre nearest you at: https://europa.eu/european-union/contact_en On the phone or by e-mail Europe Direct is a service that answers your questions about the European Union. You can contact this service - by freephone: 00 800 6 7 8 9 10 11 (certain operators may charge for these calls), - at the following standard number: +32 22999696 or - by electronic mail via: https://europa.eu/european-union/contact_en Finding information about the EU Online Information about the European Union in all the official languages of the EU is available on the Europa website at: https://europa.eu/european-union/index_en EU Publications You can download or order free and priced EU publications at: https://publications.europa.eu/en/ publications. Multiple copies of free publications may be obtained by contacting Europe Direct or your local information centre (see https://europa.eu/european-union/contact_en). EU law and related documents For access to legal information from the EU, including all EU law since 1952 in all the official language versions, go to EUR-Lex at: http://eur-lex.europa.eu Open data from the EU The EU Open Data Portal (https://data.europa.eu/euodp/en) provides access to datasets from the EU. Data can be downloaded and reused for free, for both commercial and non-commercial purposes. Getting in touch with the Centre for Entrepreneurship, SMEs, Regions and Cities (CFE) — OECD In person The OECD headquarters are located in Paris. More information available here: http://www.oecd.org/contact/ By e-mail You can contact: RegionStat@oecd.org or CFE.Contact@oecd.org Finding information about the CFE Online You can find more information on these topics on the following websites: http://www.oecd.org/ cfe/ or http://www.oecd.org/regional/regional-statistics/ Publications and datasets You can consult our publications and datasets at: https://www.oecd-ilibrary.org/ If you wish to subscribe to our newsletter please visit: http://oe.cd/CFEnews Social media Follow us on: LinkedIn: https://www.linkedin.com/company/oecd-local Twitter: https://twitter.com/OECD_local KS-02-20-499-EN-N Applying the Degree of Urbanisation A METHODOLOGICAL MANUAL TO DEFINE CITIES, TOWNS AND RURAL AREAS FOR INTERNATIONAL COMPARISONS 2021 EDITION Applying the Degree of Urbanisation — A methodological manual to define cities, towns and rural areas for international comparisons has been produced in close collaboration by six organisations — the European Commission, the Food and Agriculture Organization of the United Nations (FAO), the United Nations Human Settlements Programme (UN-Habitat), the International Labour Organization (ILO), the Organisation for Economic Co-operation and Development (OECD) and The World Bank. This manual develops a harmonised methodology to facilitate international statistical comparisons and to classify the entire territory of a country along an urban-rural continuum. The degree of urbanisation classification defines cities, towns and semi-dense areas, and rural areas. This first level of the classification may be complemented by a range of more detailed concepts, such as: metropolitan areas, commuting zones, dense towns, semi-dense towns, suburban or peri-urban areas, villages, dispersed rural areas and mostly uninhabited areas. The manual is intended to complement and not replace the definitions used by national statistical offices (NSOs) and ministries. It has been designed principally as a guide for data producers, suppliers and statisticians so that they have the necessary information to implement the methodology and ensure coherency within their data collections. It may also be of interest to users of subnational statistics so they may better understand, interpret and use official subnational statistics for taking informed decisions and policymaking. For more information https://ec.europa.eu/eurostat/ https://ec.europa.eu/info/departments/regional-and-urban-policy_en https://www.fao.org/home/en/ https://unhabitat.org/ https://www.oecd.org/ https://www.worldbank.org/ ISBN 978-92-76-20306-3 doi:10.2785/706535