MASTER THESIS Planning for climate change: an assessment of vulnerability in the city of São Paulo Submitted by: Nayara Caroline Batista Resende Date: September 14th, 2022 First examiner Prof. Dr.-Ing. habil. Jörn Birkmann Director Institute of Spatial and Regional Planning (IREUS) University of Stuttgart, Stuttgart, Germany Email: joern.birkmann@ireus.uni-stuttgart.de Second examiner Dr.-Ing. Richard Junesch Deputy Director Institute of Spatial and Regional Planning (IREUS) University of Stuttgart, Stuttgart, Germany Email: richard.junesch@ireus.uni-stuttgart.de Supervisor M.Eng. Joanna McMillan Academic Staff Member Institute of Spatial and Regional Planning (IREUS) University of Stuttgart, Stuttgart, Germany Email: joanna.mcmillan@ireus.uni-stuttgart.de Submitted to the University of Stuttgart at: Institut für Raumordnung und Entwicklungsplanung Pfaffenwaldring 7 70569 Stuttgart mailto:joern.birkmann@ireus.uni-stuttgart.de mailto:richard.junesch@ireus.uni-stuttgart.de mailto:joanna.mcmillan@ireus.uni-stuttgart.de Planning for climate change: an assessment of vulnerability in the city of São Paulo 2 Declaration of autonomy “I hereby declare that I have written the present work independently, that I have not used any sources other than those specified and that all statements taken verbatim or meaningfully from other works have been marked as such, that the submitted work has not been completely or partially the subject of another examination procedure, that I have not published the work either completely or in part, and that the electronic copy coincides exactly with the other copies.” Stuttgart, 14.09.2022. Nayara Caroline Batista Resende Planning for climate change: an assessment of vulnerability in the city of São Paulo 3 Abstract The evidence on climate change and its implications to the future generations have motivated an increasing discussion on adaptation agendas and the necessity to plan more resilient urban spaces. However, implementing effective approaches for climate change adaptation are particularly challenging in countries like Brazil, where the planning of an adequate infrastructure system was not able to accompany the accelerated urbanisation process. Particularly in the city of São Paulo, such rapid urban population growth contributed to an urbanisation that is territorially extensive and characterised by socio-spatial inequalities and severe environmental issues. As the existent inequalities are predicted to aggravate the effects of climate change in the livelihoods of the population, it becomes of utmost importance to identify and address current vulnerabilities in order to strengthen the resilience of cities. Therefore, the aim of the thesis is to identify areas within the municipality of São Paulo that are subjected to a greater risk when faced with extreme events and climate change, so that vulnerable regions, sectors or population groups can be prioritised when planning adaptation strategies. Thus, a vulnerability assessment was elaborated through the construction of a social vulnerability index and the geospatial analysis of data using GIS tools. The results of the assessment were then combined to data regarding the susceptibility to flood events in order to produce a map of risk posed by flooding, so that “territories of risk” – where social and environmental vulnerabilities overlap – could be identified for intervention. Keywords Social vulnerability index; climate change; disaster risk; floods; São Paulo. Planning for climate change: an assessment of vulnerability in the city of São Paulo 4 Table of Contents List of figures ......................................................................................................................... 6 List of tables .......................................................................................................................... 8 List of abbreviations ............................................................................................................. 10 1 Introduction ....................................................................................................................... 11 1.1 Background ................................................................................................................ 11 1.2 Problem description ................................................................................................... 11 1.3 Research question and objectives .............................................................................. 12 1.4 Research methodology .............................................................................................. 12 1.5 Outline of thesis ......................................................................................................... 13 2 Theoretical background .................................................................................................... 14 2.1 Concepts and definitions ............................................................................................ 14 2.2 Social vulnerability assessments ................................................................................ 16 3 Characterisation of the study area .................................................................................... 20 3.1 Brazilian context ......................................................................................................... 20 3.2 Urbanisation process in São Paulo ............................................................................ 21 3.3 Climate projections and risk scenarios ....................................................................... 25 3.4 Vulnerability aspects in the local context .................................................................... 28 4 Methodology ..................................................................................................................... 32 4.1 Unit of analysis ........................................................................................................... 32 4.2 Data collection ........................................................................................................... 33 4.3 Data analysis ............................................................................................................. 34 4.4 Construction of a vulnerability index ........................................................................... 34 4.5 Factor analysis ........................................................................................................... 39 4.5.1 Validation of the model ........................................................................................ 39 4.5.2 Extraction of principal components ...................................................................... 40 4.5.3 Rotation of factors ............................................................................................... 42 4.5.4 Evaluation of consistency .................................................................................... 44 Planning for climate change: an assessment of vulnerability in the city of São Paulo 5 4.6 GIS-based assessment .............................................................................................. 45 5 Results .............................................................................................................................. 46 5.1 Social vulnerability assessment.................................................................................. 50 5.2 Brief consistency check .............................................................................................. 53 5.3 Risk posed by flooding in the municipality of São Paulo ............................................. 56 5.4 Limitations .................................................................................................................. 60 6 Conclusions ...................................................................................................................... 61 References .......................................................................................................................... 63 List of appendices ................................................................................................................ 70 Appendix A .......................................................................................................................... 71 Appendix B .......................................................................................................................... 73 Appendix C .......................................................................................................................... 94 Appendix D ........................................................................................................................ 101 Planning for climate change: an assessment of vulnerability in the city of São Paulo 6 List of figures Figure 1. Interaction among weather and climate events, exposure and vulnerability producing risk. ...................................................................................................................................... 15 Figure 2: Distribution in percentages of the Brazilian population in urban and rural areas in the period between 1950 and 2010. ........................................................................................... 20 Figure 3: Location of the state of São Paulo within Brazil and the municipality of São Paulo within the homonymous state. .............................................................................................. 21 Figure 4: Expansion of the urbanised area of the municipality of São Paulo between 1881 and 2002. ................................................................................................................................... 22 Figure 5: Map of avenues, floodplains and green areas in the municipality of São Paulo. .... 24 Figure 6: Annual average temperatures in the city of São Paulo between 1933 and 2019 (meteorological station of IAG-USP). ................................................................................... 25 Figure 7: Number of days with rainfall above 80 mm and 100 mm at (A) IAG-USP and (B) Mirante de Santana stations. ............................................................................................... 26 Figure 8: Section of the Pinheiros River in the city of São Paulo showing an overlayed image of an aero photographic mapping from 1930 and a satellite image of the same area in 2017. ............................................................................................................................................ 28 Figure 9: Precarious settlement classified as landslide risk area. ......................................... 29 Figure 10: Superimposition of environmental and social vulnerabilities in Americanópolis, located in the southern zone of the municipality of São Paulo. ............................................. 30 Figure 11: Population per gender and race in Brazil. ........................................................... 31 Figure 12: Disparities in living conditions among races in Brazil. ......................................... 31 Figure 13: Illustration of inequalities within small areas of the municipality of São Paulo: Paraisópolis vs. Morumbi. .................................................................................................... 33 Figure 14: Mapping of the values estimated for factor 1 of the index (education, health and socioeconomic status) distributed into 5 categories. ............................................................ 47 Figure 15: Mapping of the values estimated for factor 2 of the index (family structure, gender and race) distributed into 5 categories. ................................................................................ 48 Figure 16: Mapping of the values estimated for factor 3 of the index (housing and infrastructure) distributed into 5 categories. ................................................................................................ 49 Planning for climate change: an assessment of vulnerability in the city of São Paulo 7 Figure 17: Categorisation of the value range obtained with the social vulnerability index into five categories utilising the method of natural breaks. .......................................................... 50 Figure 18: Social vulnerability assessment in the municipality of São Paulo. ....................... 52 Figure 19: Location of the human development units (UDHs) selected as exemplars for a brief consistency check. ............................................................................................................... 54 Figure 20: Assessment of the exposure to flood events in the municipality of São Paulo. .... 57 Figure 21: Estimation of the risk posed by flooding in the municipality of São Paulo. ........... 58 Figure 22: Example 1 of very high risk to flood events - Favela do Sapo in São Paulo. ....... 59 Figure 23: Example 2 of very high risk to flood events - Complexo Cantinho do Céu in São Paulo. .................................................................................................................................. 59 Figure 24: Example 3 of very high risk to flood events - Jardim Piratininga in São Paulo. .... 59 Planning for climate change: an assessment of vulnerability in the city of São Paulo 8 List of tables Table 1: Compilation of social vulnerability concepts identified by Cutter et al. in relevant research. .............................................................................................................................. 17 Table 2: Adoption of concepts compiled by Cutter et al. (2003) by four vulnerability assessments in Brazil (adopted concepts marked in green). ................................................ 19 Table 3: Projections for the Southeast region of South America until 2100. ......................... 27 Table 4: Main climate projections indicated for the municipality of São Paulo and associated risk scenarios. ...................................................................................................................... 28 Table 5: Summary of data collected and utilised in the research along with their respective format, source and online availability. .................................................................................. 34 Table 6: Name and description of variables selected for the construction of an index for the assessment of social vulnerability in the municipality of São Paulo. ..................................... 35 Table 7: Chosen variables and respective relation to vulnerability, abbreviation and data source. ................................................................................................................................. 37 Table 8: Results for the Kaiser-Meyer-Olkin (KMO) method and Bartlett's test of sphericity. 39 Table 9: Appropriacy of a factor analysis on a given data set according to the Kaiser-Meyer- Olkin (KMO) test. ................................................................................................................. 39 Table 10: Evaluation of the communalities of each variable in relation to the extracted factors (variables which presented values below 0,5 were excluded from the analysis). .................. 40 Table 11: Definition of the number of components to be extracted based on Eigenvalues above 1. ......................................................................................................................................... 41 Table 12: Rotated component matrix. .................................................................................. 42 Table 13: Total variance explained by the three extracted components (factors). ................ 43 Table 14: Variables (and respective loadings) that best explain each of the components extracted by the factor analysis. ........................................................................................... 43 Table 15: Verification of the internal consistency of the factors through Cronbach’s alpha. .. 44 Table 16: Reliability statistics for component 1 - education, health and socioeconomic status. ............................................................................................................................................ 44 Table 17: Reliability statistics for component 2 - family structure, gender and race. ............. 44 Table 18: Reliability statistics for component 3 - housing and infrastructure. ........................ 44 Planning for climate change: an assessment of vulnerability in the city of São Paulo 9 Table 19: Statistics for each category of social vulnerability and corresponding mean values of factor 1, 2 and 3. .................................................................................................................. 51 Table 20: Association between the classifications of the social vulnerability index constructed in this research and the correspondent UDHs, affected population and area (absolute and relative values). .................................................................................................................... 51 Table 21: Characterisation of the proposed exemplars (UDHs) according to their respective population, social vulnerability index (SVI) and factor scores. .............................................. 53 Table 22: Exemplar areas. ................................................................................................... 55 Table 23: Classification of the risk posed by flooding. .......................................................... 56 Planning for climate change: an assessment of vulnerability in the city of São Paulo 10 List of abbreviations CO2 - carbon dioxide CRED - Centre for Research on the Epidemiology of Disasters C40 - C40 Cities Climate Leadership Group EMPLASA - Metropolitan Planning Company of São Paulo GIS - Geographic Information System HAND - Height Above the Nearest Drainage IBGE - Brazilian Institute of Geography and Statistics INPE – Brazilian National Institute for Space Research IPCC - Intergovernmental Panel on Climate Change IPEA - Brazilian Institute of Applied Economic Research i.e. - in other words km - kilometres mm - millimetres PCA - principal component analysis SIGRC - Integrated Citizen Relationship Management System of the Municipality of São Paulo SMDU - Department of Urban Development of the Municipality of São Paulo SMUL - Department of Urbanism and Licensing of the Municipality of São Paulo SoVI - Social Vulnerability Index developed by Cutter et al. (2003) SVMA - Department of Green and Environment Affairs of the Municipality of São Paulo UDH - human development unit °C - degrees Celsius Planning for climate change: an assessment of vulnerability in the city of São Paulo 11 1 Introduction 1.1 Background In its last report, the IPCC (2021) acknowledged that the global surface temperature will continue to increase at least until 2050 under all considered emissions scenarios and that, unless deep reductions in greenhouse gas emissions occur in the next decades, a global warming of about 2°C will be exceeded during this century. Such scenario could threaten the survival of animal and plant species, melt glaciers, and significantly affect agriculture and the water supply of hundreds of millions of people (Nobre et al., 2010). In the context of a changing climate, the urban areas constitute one of the biggest challenges at present, as the urban population has surpassed 4.2 billion people and it is predicted that this number will reach 5 billion in 2030 (United Nations, 2018). In Central and South America, climate variability has been affecting social and natural systems at various time scales at the same time as extreme events have been distressing large regions of the continent (Magrin et al., 2014). Just in Brazil, in the period between 2000 and 2020, 105 climatological and hydro-meteorological extreme events have occurred, resulting in more than 2900 fatalities, approximately 50 million people affected and estimated economic losses that exceed 20 billion US dollars (CRED, 2022). 1.2 Problem description The evidence on climate change and the apprehension about its future implications have motivated an increasing discussion on how to rethink cities and plan more resilient and adaptable urban spaces. However, when it comes to implementing effective approaches for climate change adaptation, countless challenges are faced. These challenges are particularly severe for developing countries like Brazil, where an accelerated and chaotic urbanisation took place. In Brazil, more specifically in the city of São Paulo, as a consequence of this accelerated urbanisation process, the investments in planning and supply of an adequate urban infrastructure system were unable to accompany the fast pace of the urban population growth. In addition to that, a lack of governmental regulations greatly contributed to an urban development that prioritised political and economic interests. As a consequence, a territorially extensive urbanisation contributed to an unequal distribution of population and economic opportunities, as much as to severe environmental problems. With the changing climate and increased frequency of extreme events, it has become ever more evident that the existent inequalities within the municipality of São Paulo aggravate the impacts that these events pose to the livelihoods of the population. With that in mind, identifying and spatialising these inequalities have an essential role in the planning of more resilient cities. Planning for climate change: an assessment of vulnerability in the city of São Paulo 12 1.3 Research question and objectives The long-lasting inequalities present in São Paulo have only aggravated environmental and social crises in the city. As exemplified by the recent COVID-19 pandemic, several discrepancies within the municipality have determined how deeply different communities and neighbourhoods were affected and responded to the crisis. Therefore, as these discrepancies tend to become even more prominent due to the effects of climate change, it is necessary to identify different patterns of environmental and social vulnerability in the city of São Paulo in order to prioritise public policies and identify suitable solutions for different cases. As climate change will not have a uniform effect on all communities, an understanding of vulnerability in different parts of the city becomes critical as a planning instrument for climate change adaptation. Therefore, this research aims to identify the areas within the municipality of São Paulo that are subjected to a greater risk when faced with extreme events and climate change, considering their social vulnerability and capacity to recover from disasters. In order to elaborate a vulnerability assessment of the municipality, the research aims to achieve the following objectives: identify major environmental risks to which the municipality is subjected, choose relevant variables to evaluate social vulnerability according to the local context and available data, map the measured social vulnerability and superimpose it to an exposure (to hazards) map using GIS tools. 1.4 Research methodology With the aim of assessing vulnerability to climate change in the municipality of São Paulo, an index of social vulnerability based on the methodology proposed by Cutter et al. (2003) was constructed whilst taking into consideration the local context and the data available for the chosen unit of analysis. The decision to utilise an index to measure social vulnerability was taken on account of it being a useful tool that helps to identify social and spatial inequalities in order to prioritise vulnerable regions, sectors or population groups. After the selection of appropriate variables, the computation of the index was achieved through statistical analysis. Following the calculation of the vulnerability index for each unit of analysis, GIS tools can be used to support the geospatial analysis of data as well as assist the interpretation of abstract concepts through spatial mapping and visualisation. Thus, the georeferencing of the different levels of vulnerability identified in the municipality can be used to emphasise areas of the territory where impacts on society are expected to be most severe and therefore facilitate the identification of priority areas for intervention. Planning for climate change: an assessment of vulnerability in the city of São Paulo 13 1.5 Outline of thesis The thesis is structured by an introduction and five other chapters organised in the following order: Chapter 2. Theoretical background: this chapter outlines concepts associated with vulnerability to climate change – such as vulnerability, exposure and risk – and briefly reviews approaches used to measure vulnerability in order to set a theoretical basis for the elaboration of the vulnerability assessment proposed by the thesis. Chapter 3. Characterisation of the study area: this chapter depicts the environmental and socioeconomic contexts of the selected study area (municipality of São Paulo) and identifies aspects of its urbanisation process that could characterise the existent social inequalities and projected risk scenarios for the area. Chapter 4. Methodology: this chapter presents and describes the methods used for data collection and analysis as well as the approach employed for the construction of the social vulnerability index and spatial assessment of the results. Chapter 5. Results: this chapter presents the results of the thesis, including the computation of the constructed vulnerability index, the spatialisation of the results using GIS tools and their overlay with different levels of exposure to floods in the municipality of São Paulo. Chapter 6. Conclusions: this chapter summarises the main findings of the research, acknowledges its limitations and suggests approaches for future research on the topic. This thesis also contains four appendices: • Appendix A contains a map of the delimitation of the human development units (UDHs) in the municipality of São Paulo. • Appendix B contains the georeferenced data for each variable used in the construction of the social vulnerability index. • Appendix C contains additional tables resulted from the factor analysis in the software IBM SPSS statistics 28.0. • Appendix D contains the data per UDH (human development unit) of each variable used in the construction of the social vulnerability index. Planning for climate change: an assessment of vulnerability in the city of São Paulo 14 2 Theoretical background This chapter briefly presents the concepts which are relevant to the research and describes a few methodological approaches for measuring vulnerability, setting a theoretical basis for the development of the proposed vulnerability assessment. First and foremost, climate change is one of the most complex challenges faced by society today. The projected intensification and raise in frequency of climatic events – such as storms, heat waves and droughts – threaten particularly the urbanised areas in the world. In Brazil, about 85% of the country’s population lives in urbanised areas (IBGE, 2010a). And these urban agglomerations increase the propensity of economic, environmental and social damage arising from climate change (GERICS and KfW, 2015; Kim et al., 2021). Therefore, cities pose themselves as fundamental elements to the success of adaptation agendas and should be subject of research that aims to assist authorities and policymakers in planning for a changing climate. 2.1 Concepts and definitions Considering that some of the concepts addressed in the thesis, such as disaster risk, vulnerability, and exposure, can be interpreted in various ways by different research fields, this section seeks to present the definitions of these terms in the perspective of this research. First of all, it is important to specify the meaning of risk in the context of climate change. The IPCC (2014a, p. 1048) defines risk as the potential for adverse consequences where something of value is at stake and where the occurrence and degree of an outcome is uncertain. In the context of the assessment of climate impacts, the term risk is often used to refer to the potential for adverse consequences of a climate-related hazard, or of adaptation or mitigation responses to such a hazard, on lives, livelihoods, health and well-being, ecosystems and species, economic, social and cultural assets, services (including ecosystem services), and infrastructure. The IPCC (2014a) also describes risk as a combination of certain components: the probability of climate-related hazards occurring, the exposure of a system to them and the vulnerability of the affected system to the impacts or consequences of these hazards (see Figure 1). Exposure refers to a situation in which an element is subjected to the probability of harm. As stated by the IPCC (2014a, p. 1048), exposure can be defined as “the presence of people, livelihoods, species or ecosystems, environmental functions, services, and resources, infrastructure, or economic, social, or cultural assets in places and settings that could be adversely affected”. However, it is critical to highlight that, while exposure to a hazard could be the same among a population, the impacts it could pose on each family or individual will be largely determined by their capability to cope with and recover from such events. Planning for climate change: an assessment of vulnerability in the city of São Paulo 15 Figure 1: Interaction among weather and climate events, exposure and vulnerability producing risk. Source: IPCC, 2012. In order to account for this capability to cope and recover, it is necessary to consider the vulnerability of the population, which refers to socioeconomic characteristics or institutional dimensions which affect their susceptibility to climate change impacts (Sherbinin et al., 2019). As the thesis attempts to assess social vulnerability in the context of climate change planning, it is important to establish an understanding of what vulnerability represents in the context of this research. As stated by Birkmann (2013), the concept of vulnerability has been continuously developed over time, however, its exact meaning, framing and assessment remain contested among scientific communities. Adopting the definition given by the IPCC in its fifth assessment report, vulnerability can be described as “the propensity or predisposition to be adversely affected. Vulnerability encompasses a variety of concepts and elements including sensitivity or susceptibility to harm and lack of capacity to cope and adapt” (IPCC, 2014a, p. 1048). According to this perspective, vulnerability is a multidimensional concept that can change and evolve over time according to the characteristics of the system concerned. Cutter and Finch (2019, p. 130) also pointed out that social vulnerability measures “both the sensitivity of a population to natural hazards and its ability to respond to and recover from the impacts of hazards”. Thus, it could be considered as a combination of social inequalities – such as age, race and income – and place inequalities, for instance access to basic services and infrastructure (Cutter et al., 2003). According to Infield et al. (2019), the impacts of climate change are not evenly distributed across regions or populations, and the disadvantaged are often concentrated in areas subjected to natural hazards. Planning for climate change: an assessment of vulnerability in the city of São Paulo 16 Therefore, for the population which is already affected by poverty, food insecurity, inadequate water supply or other issues, the combination of these existing concerns with the impacts of climate change will amplify inherent risks and threaten the livelihoods of the population. According to Bolay (2020), the impoverished communities are the first to suffer from the lack of infrastructure and services and they are often informally settled, particularly in areas classified as unbuildable due to risks such as steep hillsides and floodplains. Thereby, as Hardoy and Satterthwaite (1991) argue that much of the urban population in the Global South lives in poverty, it becomes vital to identify and address vulnerability in these regions. 2.2 Social vulnerability assessments There is a wide variety of approaches utilised to assess vulnerability in the context of disaster risk and climate change. Vulnerability is, nevertheless, a multifaceted abstract concept which cannot be easily measured or quantified. “The analysis of different conceptual frameworks and assessment methodologies has shown that no single conceptual approach can capture and explain vulnerability comprehensively” (Birkmann, 2013, p. 552). In spite of the limitations and obstacles, vulnerability assessments are yet an important tool that helps to bridge the gaps between the theory and the actual decision-making processes. Considered as a latent variable, i.e., inherent to a person or a place but not directly observable, social vulnerability can only be measured indirectly by utilising measurable parameters, such as the demographic attributes of an area (Spielman et al., 2020). In 2003, Cutter et al. identified in relevant research the most mentioned characteristics which influence social vulnerability (see Table 1) and developed one of the most widespread social vulnerability indicators, the Social Vulnerability Index (SoVI), based on demographic and socioeconomic data from the United States. The index attempted to synthesise the characteristics which were capable of influencing social vulnerability to natural hazards into a single model. Throughout the years, the SoVI has achieved a modicum of success as an empirically based approach for measuring social vulnerability and has been widely replicated and adapted around the world (Cutter and Morath, 2013). In Brazil, the SoVI was replicated and adapted by Loyola Hummell et al. (2016) in an assessment of vulnerability that included all 5565 Brazilian cities as study units. And although the analysis and measurement of vulnerability proposed by some other Brazilian researchers do not directly replicate the SoVI, they still corroborate the perspective of Cutter et al. (2003). In Table 2, the author has summarised which of the concepts compiled by Cutter et al. were utilised in four different vulnerability assessments proposed for the context of Brazil. The most frequently adopted concepts are socioeconomic status (mostly measured by income), gender, age, infrastructure (focus on adequate water supply and sanitation conditions), family structure and education. Planning for climate change: an assessment of vulnerability in the city of São Paulo 17 Table 1: Compilation of social vulnerability concepts identified by Cutter et al. in relevant research. Source: Cutter et al., 2003. Concept Description Socioeconomic status (income, political power, prestige) The ability to absorb losses and enhance resilience to hazard impacts. Wealth enables communities to absorb and recover from losses more quickly due to insurance, social safety nets and entitlement programs. Sources: Cutter, Mitchell and Scott (2000); Burton, Kates and White (1993); Blaikie et al. (1994); Peacock, Morrow and Gladwin (1997, 2000); Hewitt (1997); Puente (1999); and Platt (1999). Gender Women can have a more difficult time during recovery than men, often due to sector-specific employment, lower wages and family care responsibilities. Sources: Blaikie et al. (1994); Enarson and Morrow (1998); Enarson and Scanlon (1999); Morrow and Phillips (1999); Fothergill (1996); Peacock, Morrow and Gladwin (1997, 2000); Hewitt (1997); and Cutter (1996). Race and ethnicity Imposes language and cultural barriers that affect access to post-disaster funding and residential locations in high hazard areas. Sources: Pulido (2000); Peacock, Morrow and Gladwin (1997, 2000); Bolin with Stanford (1998); and Bolin (1993). Age Extremes of the age spectrum affect the movement out of harm’s way. Parents lose time and money caring for children when day-care facilities are affected; elderly may have mobility constraints or mobility concerns increasing the burden of care and lack of resilience. Sources: Cutter, Mitchell and Scott (2000); O’Brien and Mileti (1992); Hewitt (1997); and Ngo (2001). Commercial and industrial development The value, quality, and density of commercial and industrial buildings provides an indicator of the state of economic health of a community, potential losses in the business community and longer-term issues with recovery after an event. Sources: Heinz Centre for Science, Economics and the Environment (2000); Webb, Tierney and Dahlhamer (2000). Employment loss The potential loss of employment following a disaster exacerbates the number of unemployed workers in a community, contributing to a slower recovery from the disaster. Source: Mileti (1999). Rural / urban Rural residents may be more vulnerable due to lower incomes and more dependent on locally based resource extraction economies (e.g., farming, fishing). High-density areas (urban) complicate evacuation out of harm’s way. Sources: Cutter, Mitchell and Scott (2000); Cova and Church (1997); and Mitchell (1999). Residential property The value, quality and density of residential construction affect potential losses and recovery. Expensive homes on the coast are costly to replace; mobile homes are easily destroyed and less resilient to hazards. Sources: Heinz Centre for Science, Economics and the Environment (2000); Cutter, Mitchell and Scott (2000); and Bolin and Stanford (1991). Infrastructure and lifelines Loss of sewers, bridges, water, communications and transportation infrastructure compounds potential disaster losses. The loss of infrastructure may place an insurmountable financial burden on smaller communities that lack the financial resources to rebuild. Sources: Heinz Centre for Science, Economics and the Environment (2000); and Platt (1995). Planning for climate change: an assessment of vulnerability in the city of São Paulo 18 Renters People that rent do so because they are either transient or do not have the financial resources for home ownership. They often lack access to information about financial aid during recovery. In the most extreme cases, renters lack sufficient shelter options when lodging becomes uninhabitable or too costly to afford. Sources: Heinz Centre for Science, Economics and the Environment (2000); and Morrow (1999). Occupation Some occupations, especially those involving resource extraction, may be severely impacted by a hazard event. Self-employed fishermen suffer when their means of production is lost and may not have the requisite capital to resume work in a timely fashion and thus will seek alternative employment. Those migrant workers engaged in agriculture and low- skilled service jobs (housekeeping, childcare, and gardening) may similarly suffer, as disposable income fades and the need for services declines. Immigration status also affects occupational recovery. Sources: Heinz Centre for Science, Economics and the Environment (2000); Hewitt (1997); and Puente (1999). Family structure Families with large numbers of dependents or single parent households often have limited finances to outsource care for dependents and thus must juggle work responsibilities and care for family members. All affect the resilience to and recovery from hazards. Sources: Blaikie et al. (1994); Morrow (1999); Heinz Centre for Science, Economics and the Environment (2000); and Puente (1999). Education Education is linked to socioeconomic status, with higher educational attainment resulting in greater lifetime earnings. Lower education constrains the ability to understand warning information and access to recovery information. Source: Heinz Centre for Science, Economics and the Environment (2000). Population growth Counties experiencing rapid growth lack available quality housing and the social services network may not have had time to adjust to increased populations. New migrants may not speak the language and not be familiar with bureaucracies for obtaining relief or recovery information, all of which increase vulnerability. Sources: Heinz Centre for Science, Economics and the Environment (2000); Cutter, Mitchell and Scott (2000); Morrow (1999); and Puente (1999). Medical services Health care providers, including physicians, nursing homes and hospitals, are important post-event sources of relief. The lack of proximate medical services will lengthen immediate relief and longer-term recovery from disasters. Sources: Heinz Centre for Science, Economics and the Environment (2000); Morrow (1999); and Hewitt (1997). Social dependence Those people who are totally dependent on social services for survival are already economically and socially marginalised and require additional support in the post-disaster period. Sources: Morrow (1999); Heinz Centre for Science, Economics and the Environment (2000); Drabek (1996); and Hewitt (2000). Special needs populations Special needs populations (infirm, institutionalised, transient, homeless), while difficult to identify and measure, are disproportionately affected during disasters and, because of their invisibility in communities, mostly ignored during recovery. Sources: Morrow (1999); and Tobin and Ollenburger (1993). Planning for climate change: an assessment of vulnerability in the city of São Paulo 19 Table 2: Adoption of concepts compiled by Cutter et al. (2003) by four vulnerability assessments in Brazil (adopted concepts marked in green). Source: table elaborated by the author. Vulnerability assessments in Brazil Author Alves (2013) Almeida (2010) Deschamps (2004) Cunha et al. (2004) Unit of analysis Census tract Census tract Dissemination area Dissemination area Location of the analysis 16 municipalities on the coast of the State of São Paulo Basin of Maranguapinho River in the State of Ceará Metropolitan region of Curitiba in the State of Paraná Municipality of Campinas in the State of São Paulo C o n c e p ts i d e n ti fi e d b y C u tt e r e t a l. ( 2 0 0 3 ) Socioeconomic status Gender Race and ethnicity Age Commercial and industrial development Employment loss Rural / urban Residential property Infrastructure and lifelines Renters Occupation Family structure Education Population growth Medical services Social dependence Special needs populations The concept of employment is adopted in a different perspective by some of the assessments, which take into consideration the number of unemployed and/or the formality rate of the employed. Moreover, the disregard of the concepts of medical services and social dependence on these assessments could be particularly attributed to the availability of such data for smaller territorial units, such as census tracts and dissemination areas. Lastly, the social vulnerability index (SoVI) proposed by Cutter et al. (2003) seems to provide a reliable theoretical basis for the elaboration of similar assessments, nonetheless it is typically adapted and reshaped to reflect the context under consideration as well as adjusted to the available data. Planning for climate change: an assessment of vulnerability in the city of São Paulo 20 3 Characterisation of the study area 3.1 Brazilian context The extensive and accelerated urbanisation that took place in the last centuries and is still ongoing in different parts of the world has led to grave consequences to the environment and the quality of life in urban areas. The situation is particularly severe in developing countries, where predominantly the urbanisation process tends to unfold at an accelerated pace. Brazil is no exception: the tremendous speed of the urbanisation that occurred in the country, especially in the second half of the twentieth century, has profoundly affected the patterns in which Brazilian cities were shaped and it is one of the structural determinants of the constitution of modern Brazilian society (Brito and Pinho, 2012). The shift from a rural to a more predominantly urban society took place in a very short interval of time in Brazil (see Figure 2). The urban population of the country grew just in the period from 1950 to 2010 from around 18 million to more than 160 million inhabitants (IBGE, 2010a). As a consequence of this accelerated process, the investments in planning and supply of an adequate urban infrastructure system were unable to accompany the fast pace of the urban population growth. Figure 2: Distribution in percentages of the Brazilian population in urban and rural areas in the period between 1950 and 2010. Source: elaborated by the author based on data from the Brazilian Institute of Geography and Statistics (IBGE, 2010a). 36,2 45,1 56 67,7 75,5 81,2 84,4 63,8 54,9 44 32,3 24,5 18,8 15,6 0 10 20 30 40 50 60 70 80 90 100 1940 1950 1960 1970 1980 1990 2000 2010 2020 P e rc e n ta g e o f th e t o ta l p o p u la ti o n [ % ] Year Distribution of the Brazilian population in urban and rural areas in the period between 1950 and 2010 Urban Rural Planning for climate change: an assessment of vulnerability in the city of São Paulo 21 3.2 Urbanisation process in São Paulo São Paulo is a municipality located in the homonymous state in south-eastern Brazil (see Figure 3). In addition to other 38 municipalities, it shapes the largest metropolitan region in the country. The development process of the city of São Paulo reflected the urbanisation process in Brazil, characterised by one of the most accelerated rural-urban transitions in the world (SMDU, 2012). In 1872, São Paulo had a population of slightly over 31 thousand inhabitants. At present, São Paulo is the most populated municipality in Brazil and South America with an estimated population of around 12 million inhabitants (IBGE, 2022). Figure 3: Location of the state of São Paulo within Brazil and the municipality of São Paulo within the homonymous state. Figures by TUBS / CC-BY-SA-3.0 and Shadowxfox / CC-BY-2.5. Source: Wikimedia Commons contributors, 2022a and 2022b. São Paulo was founded by Jesuit missionaries in 1554 and established on top of a small hill between two rivers, the Tamanduateí and the Anhangabaú, in an area characterised by a phenomenon of regular river floods (Gouveia, 2016). As stated by Travassos (2004), on one hand, the urban occupation of the region depended on the existence of the nearby rivers and streams, but on the other hand, as the city spread out across the territory, various conflicts with its natural support were surfaced. Although the numerous water courses that used to meander the region played a crucial role in the development of the city, they ended up over time acquiring the mere function of transporting away the garbage and sewage produced by its inhabitants. Planning for climate change: an assessment of vulnerability in the city of São Paulo 22 Figure 4: Expansion of the urbanised area of the municipality of São Paulo between 1881 and 2002. Source: adapted from EMPLASA, 2002/2003. Districts Parks Urban landmarks (public university, airports etc.) Water bodies Boundary of springs’ protection area Time periods Until 1881 From 1882 until 1914 From 1915 until 1929 From 1930 until 1949 From 1950 until 1962 From 1963 until 1974 From 1975 until 1985 From 1986 until 1992 From 1993 until 2002 Planning for climate change: an assessment of vulnerability in the city of São Paulo 23 In association with the fast and unorganised urban growth of the city that happened predominantly in the twentieth century, the provision of enough infrastructure to satisfy the needs of the millions of people arriving in São Paulo led to a series of problems. That includes a drastic reduction of the green areas within the city, the channelling and canalisation of many rivers and streams and a spread urbanisation (Marques, 2017). This territorially extensive urbanisation (see Figure 4) contributed to an unequal distribution of population and economic opportunities, as much as to severe environmental problems. Among those environmental problems are the occupation of springs surroundings and environmentally fragile areas; the occupation of valleys, especially for the implementation of large avenues; and the high impermeabilization of the urban soil, which has resulted in a velocity increase of the stormwater runoff and intensified the floods (SMDU, 2012). According to an analysis made by Marques (2017), it was identified that about 20% of the water streams in the city of São Paulo are currently underground and approximately 5% of those still running on the surface are confined in concrete channels. The analysis also detected that almost 100% of the water streams that were confined underground or in concrete channels are located in the urban areas of the municipality, which turned large portions of the city completely arid due to the absence of green areas that originally accompanied the water network. It is also relevant to point out that the water courses in the urban area of the city are heavily polluted. Since the basic sanitation and drainage systems do not cover the entire occupied area of the municipality, the generated sewage and the stormwater, which collects all the dispersed pollution and garbage from the streets, flow into the affluents and contaminate the entire river network (Ikeda, 2016). The occupation of riverbanks and conservation areas by irregular housing, a result of an unattended demand for social housing, also contributes to the water pollution and increases the risk of the exposed population to events such as floods. Moreover, the floodplains, where since the foundation of the city a phenomenon of regular river floods has occurred, were drought and deforested to give place to rigid grey infrastructure and occupied mostly by roads and expressways. Once the main avenues, the floodplains and the green areas in the municipality of São Paulo are combined in a single map, it is noticeable how the main roads and expressways concentrated themselves in the floodplains, and drastically reduced the network of natural green areas in the regions where the urbanisation is more intense (see Figure 5). As a result of narrowing the riverbeds and sealing the soil, the storm events cause ever more frequently an overflow of the water streams, in a way that the annual floods have turned into catastrophes (Ikeda, 2016). These events already occur often in the metropolitan region of São Paulo, yet it has been predicted that, due to the consequences of climate change, these events will be even more frequent and intense (Delgado, 2020). Planning for climate change: an assessment of vulnerability in the city of São Paulo 24 Figure 5: Map of avenues, floodplains and green areas in the municipality of São Paulo. Source: adapted from Marques, 2017. Planning for climate change: an assessment of vulnerability in the city of São Paulo 25 3.3 Climate projections and risk scenarios The climate in the municipality of São Paulo is characterised by a warm and humid summer and a cold and dry winter (Nobre, 2011). As indicated by SVMA (2021), the climate of the municipality was strongly influenced by the presence of the native Atlantic Forest and its water network at the beginning of the 19th century, displaying mild temperatures and drizzle, which for many years characterised São Paulo. The urbanisation, deforestation and frequent sealing of the soil, however, have increasingly contributed to shifts in the local climate and the aggravation of weather and climate events. The extensive urbanisation of São Paulo, where vegetation was often replaced by grey infrastructure, has resulted in an increase of the surface temperature, contributing to the formation and intensification of heat islands (SVMA, 2021). For instance, temperature data of the municipality show that there has been an increase in average temperatures in the period between 1933 and 2020 (see Figure 6). Additionally, pattern changes in the region’s climate have been noticed in the last decades (SVMA, 2021) and a significant increase in heavy rainfall has been observed (see Figure 7). According to Marengo et al. (2020), the number of rainfall events above 100 mm/day in the last 20 years has exceeded the accumulated records of the previous six decades in São Paulo. Figure 6: Annual average temperatures in the city of São Paulo between 1933 and 2019 (meteorological station of IAG-USP). Source: adapted from SVMA, 2021. Annual average Summer Winter T e m p e ra tu re ( °C ) Planning for climate change: an assessment of vulnerability in the city of São Paulo 26 Figure 7: Number of days with rainfall above 80 mm and 100 mm at (A) IAG-USP and (B) Mirante de Santana stations. Source: elaborated by the author based on Marengo et al. (2020) and SVMA (2021). Currently, some of the main concerns regarding urban areas are related to climate variability at present and possible future changes, especially concerning an increase in frequency and intensity of extreme weather events (SVMA, 2021). For the Southeast region of South America, where São Paulo is located, global and regional models have indicated for the coming decades an increase on air temperature, precipitation, consecutive dry days, hot days and nights, heat waves and greater irregularity in the distribution of rainfall throughout the year, as well as a decrease in the humidity of the air and the number of cold days and nights (see Table 3). 2 2 6 3 3 3 12 13 1 1 2 1 2 3 4 7 0 2 4 6 8 10 12 14 1931-1940 1941-1950 1951-1960 1961-1970 1971-1980 1981-1990 1991-2000 2001-2010 2011-2020 N u m b e r o f d a y s p e r d e c a d e A Rainfall above 80 mm Rainfall above 100 mm 3 5 4 9 9 11 1 2 2 2 4 0 2 4 6 8 10 12 1931-1940 1941-1950 1951-1960 1961-1970 1971-1980 1981-1990 1991-2000 2001-2010 2011-2020 N u m b e r o f d a y s p e r d e c a d e B Rainfall above 80 mm Rainfall above 100 mm Planning for climate change: an assessment of vulnerability in the city of São Paulo 27 Table 3: Projections for the Southeast region of South America until 2100. Source: adapted from Morais, 2019. Climatic event Trend Reliability Source Air temperature Increase - Giorgi and Diffenbaugh, 2008 Increase - Marengo et al., 2011 Increase - Ambrizzi and Araujo, 2014 Increase High Torres, 2016 Precipitation Increase - Giorgi and Diffenbaugh, 2008 Increase - Marengo et al., 2011 Increase - Ambrizzi and Araujo, 2014 Increase Medium IPCC, 2014b Irregularity in the distribution of rainfall throughout the year Increase Medium Torres, 2016 Relative humidity of the air Decrease Low Torres, 2016 Consecutive dry days Increase Medium Torres, 2016 Warm nights Increase Medium Sillmann et al., 2013 Increase High IPCC, 2014b Increase Medium Torres, 2016 Warm days Increase Medium Sillmann et al., 2013 Increase High IPCC, 2014b Increase Medium Torres, 2016 Cold days Decrease Medium Sillmann et al., 2013 Decrease High IPCC, 2014b Decrease Medium Torres, 2016 Cold nights Decrease Medium Sillmann et al., 2013 Decrease High IPCC, 2014b Decrease Medium Torres, 2016 Heat waves Increase Medium IPCC, 2014b Increase High Torres, 2016 The projected increase in the intensity and frequency of rainfall events as a result of climate change observed in the municipality of São Paulo highlights the importance to analyse current as well as future trends in risk scenarios and vulnerability conditions (Nobre, 2011). For the three principal climate projections indicated for the municipality of São Paulo, the main associated risk scenarios were identified (see Table 4). Among those, the two most frequent and severe risk scenarios in the municipality – floods and landslides – manifest themselves in periods of intense precipitation and cause significant damage to society from both human and economic perspectives (Nobre, 2011). Although floods and landslides impact society as a whole, it is usually the low-income population living in risk areas that is affected the deepest. Planning for climate change: an assessment of vulnerability in the city of São Paulo 28 Table 4: Main climate projections indicated for the municipality of São Paulo and associated risk scenarios. Source: table elaborated by the author. Climate projections Associated risk scenarios Increase of precipitation and occurrence of extreme rainfall events Floods Landslides Spread of waterborne diseases Increase of temperature and irregularity in the distribution of rainfall throughout the year Droughts Disruption of the water supply Heat waves Mortality associated with temperature extremes 3.4 Vulnerability aspects in the local context As a consequence of being part of the largest and most populous metropolitan region in Brazil and having experienced such an accelerated urbanisation process, a large portion of São Paulo’s population is subjected to vulnerable conditions, poor infrastructure and difficult access to basic services. In addition to that, the possibility of an increase in frequency and intensity of extreme climate events predicts concerning risk scenarios (Nobre et al., 2010). Figure 8: Section of the Pinheiros River in the city of São Paulo showing an overlayed image of an aero photographic mapping from 1930 and a satellite image of the same area in 2017. Source: Goldenstein, 2017. In Brazil, due to climatic conditions and land use patterns, the most recurring natural hazards are related to alterations in the natural hydrological cycle, which are especially noticeable in urban areas (Almeida, 2010). In São Paulo, for instance, the Pinheiros River, which previously meandered the city, was over time rectified and had the direction of its course reversed (see Figure 8). As a consequence of such urbanisation pattern, which sealed the urban soil, narrowed the riverbeds and canalised the watercourses, “natural” hazards such as floods and landslides are enhanced and aggravated. These are some of the major reasons for the observance of disasters caused by floods and landslides reoccurring every year in the municipality of São Paulo, causing enormous economic losses and numerous fatalities. Planning for climate change: an assessment of vulnerability in the city of São Paulo 29 From the perspective of Rolnik and Klink (2011), due to an urban development that prioritised economic interests, the natural geography of the cities turned into an obstacle to be overcome in a design that seeks to minimise territorial losses in favour of the land market. As a consequence, in Brazil, as in several other developing countries, urban rivers and riverbanks ceased to be seen as attractive environments and became synonymous with degraded and devalued spaces, thus becoming the alternative of access to urban land for a growing contingent of urban poor (Almeida, 2010). With the projected increase of extreme weather events in the municipality of São Paulo, Nobre et al. (2010) predicts a high probability of accidents related to floods and landslides occurring due to a combination of risk areas, high concentration of people in those areas and the vulnerability of the existing occupation due to the irregularity and precariousness of the dwellings (see Figure 9). Figure 9: Precarious settlement classified as landslide risk area. Photo: IPT, 2010. Source: Nobre et al., 2010. The significant concentration of poverty in the metropolis of São Paulo is characterised by the duality between the “formal city”, which is the focus of public and private investments, and the “informal city”, which is demoted from equivalent benefits and grows rapidly in urban illegality, exacerbating social and environmental disparities within the territory (Grostein, 2001). These contexts of socio-spatial inequalities can be seen as consequences of the intersection of degraded environments, improvised occupations and unfavourable socio-economic conditions, which establish as a result territories of risk (Almeida, 2010), where environmental risks, exposure to hazards and social vulnerability overlap (see Figure 10). Planning for climate change: an assessment of vulnerability in the city of São Paulo 30 Figure 10: Superimposition of environmental and social vulnerabilities in Americanópolis, located in the southern zone of the municipality of São Paulo. Photo by TV Globo. Source: Mancuso, 2017. Another aspect of vulnerability is that extreme weather events in Brazil significantly affect the health of communities, either by causing casualties due to accidents or triggering outbreaks of climate-sensitive infectious diseases (FIOCRUZ, 2007). The occurrence of heavy rainfall and floods specifically, increases the likelihood of contracting infectious waterborne diseases, notably intestinal parasitosis, viral hepatitis, leptospirosis and enteroviruses, while creating the conditions for the breeding of mosquitoes that transmit diseases such as dengue, yellow fever and malaria (Nobre et al., 2010). This situation is even more severe in areas with inadequate sanitation conditions and an absence of garbage collection services because the probability of contact with contaminated water is very high. Lastly, when analysing social vulnerability in Brazil, it is necessary to take an intersectional approach of class, gender and race. Firstly, because gender and race discrimination are issues which concern the majority of the population (see Figure 11). And, secondly, because in any social indicator considered (education, employment, access to basic infrastructure etc.), there is a systematic disadvantage for women in relation to men, and for pardo1 and black individuals of both sexes in relation to white individuals (Abramo, 2004). The greater levels of economic and social vulnerability among black, pardo and indigenous populations in Brazil can be explained by factors such as occupational segregation, fewer educational opportunities and lower incomes in similar occupations (IBGE, 2019). 1 Pardo is the term used in Brazil to describe individuals with multiracial background (IBGE, 2010b). Planning for climate change: an assessment of vulnerability in the city of São Paulo 31 Figure 11: Population per gender and race in Brazil. Source: elaborated by the author based on data from IBGE (2010a) and IBGE (2019). With regard to living standards, inequalities by race are also revealed by differences in housing conditions, concerning both the spatial distribution of households as well as the access to basic services and infrastructure. For instance, disparities in the access to adequate water supply and sanitation can be identified for the black (and pardo) population when compared to the white population (see Figure 12). In addition to that, it has been observed that excessive household crowding – situations in which there are more than three residents per room used as dormitory – occurred among the black and pardo population with a frequency two times higher than the white population, being especially common among households consisting of women without a spouse (IBGE, 2019). Figure 12: Disparities in living conditions among races in Brazil. Source: elaborated by the author based on data from IBGE (2019). 48,97%51,03% Population per gender in Brazil (2010) Men Women 47,73% 50,74% 1,52% Population per race in Brazil (2010) White Pardo and black Others 6 11,5 26,5 3,6 12,5 17,9 42,8 7 0 5 10 15 20 25 30 35 40 45 Absence of garbage collection services Absence of water supply from the general network Absense of sewage collection through the general network or a septic tank Household overcrowding % of the respective population Disparities in sanitation and housing conditions among races (2018) White population Black and pardo population Planning for climate change: an assessment of vulnerability in the city of São Paulo 32 4 Methodology This chapter presents and describes the processes of data collection and analysis as well as the methods used for the assessment and mapping of vulnerability in the municipality of São Paulo. The assessment was performed through the construction of a vulnerability index, and the choice of the most adequate set of variables depended on the socio-economic attributes of the local context and the availability of data. The values calculated for the index were georeferenced and associated with the chosen unit of analysis. Finally, as one of the greatest challenges regarding climate change identified in the municipality of São Paulo are the impacts resulting from flood events, the risk posed by flooding was mapped by multiplying the identified social vulnerability to the levels of exposure of the population to the climatic event. 4.1 Unit of analysis Most studies regarding vulnerability to climate change in Brazil utilise a national or regional level of analysis, assessing the vulnerability levels per municipality. In contrast, the studies that focus on the local or intra-urban scales utilise mostly the census tract as the unit of analysis. In Brazil, a census tract is defined by IBGE (2010b) as the territorial unit for the collection of census data, with identified physical limits and continuous area. Some of the data collected by the census is, however, not published per tract because of restrictions due to data protection. In some cases, the smallest territorial unit for which data is published is the so-called dissemination area, formed by a grouping of contiguous census tracts with the objective of statistical calibration (IBGE, 2010b). Yet, IPEA (2015) argues that the delimitation of the IBGE's dissemination areas meets the technical requirements related to the collection and sampling processes but misses to consider the socioeconomic diversity and portray inequalities that occur within intra-metropolitan spaces (as seen in Figure 13). For that reason, human development units (UDHs) were suggested by IPEA (2015) as divisions that generate more homogenous areas from a socioeconomic point of view and better describe the diversity of situations related to human development that occurs within the metropolitan regions. As the aim of this research is to analyse vulnerability within the municipality of São Paulo, an intra-municipal level of analysis should be adopted. Thus, with the objective of characterising areas that present a more homogeneous territory and socioeconomic attributes, the human development unit (UDH) was chosen as the unit of analysis of this research. It gathers data of groups of census tracts mapped in the 2010 Census in Brazil. As stated by Oliveira et al. (2020), these units capture the diversity of social contexts and characteristics that occur within the territory of the municipality, better portraying the inequalities found in reality. In the study area of this research – the municipality of São Paulo – there are a total of 1593 human development units (see Appendix A). Planning for climate change: an assessment of vulnerability in the city of São Paulo 33 Figure 13: Illustration of inequalities within small areas of the municipality of São Paulo: Paraisópolis vs. Morumbi. Photo: Tuca Vieira, 2004. Source: The Guardian, 2017. 4.2 Data collection All the data used in this research were obtained digitally via the online platforms of the responsible agencies. The data used for constructing the social vulnerability index in the research were mainly acquired from the 2010 Census in Brazil – carried out by the Brazilian Institute of Geography and Statistics (IBGE) – and the Human Development Atlas developed by the Brazilian Institute of Applied Economic Research (IPEA) with data from 2010. The utilisation of data from 2010 is due to the non-realisation of the 2020 Census, which was postponed due to the COVID-19 pandemic and is taking place at present. The spatial analysis and maps of the municipality of São Paulo, elaborated in the software ArcGIS Pro by the author, were produced with data provided by the Municipality of São Paulo through the digital platform Geosampa (tool that gathers detailed and georeferenced public information about the municipality of São Paulo). Additionally, the data used for the assessment of exposure to floods was obtained through a database made available by an initiative of several research institutions which assesses the vulnerability of Brazilian cities to climate change. All the collected data, their format and sources were summarised by the author and can be viewed in Table 5. Planning for climate change: an assessment of vulnerability in the city of São Paulo 34 Table 5: Summary of data collected and utilised in the research along with their respective format, source and online availability. Source: table elaborated by the author. Data Format Source Available at Administrative boundaries of the municipality of São Paulo shapefile SMDU, 2013 http://geosampa.prefeitura.sp.gov.br Water bodies in the municipality of São Paulo shapefile SMUL, 2021 http://geosampa.prefeitura.sp.gov.br Delimitation of the census tracts in the municipality of São Paulo shapefile IBGE, 2010 http://geosampa.prefeitura.sp.gov.br Delimitation of the UDHs in the municipality of São Paulo shapefile IPEA, 2010 http://ivs.ipea.gov.br Data per census tract of the 2010 Census in Brazil xls / csv IBGE, 2010 http://geosampa.prefeitura.sp.gov.br Human development data per UDH in São Paulo xls / csv IPEA, 2010 http://geosampa.prefeitura.sp.gov.br http://ivs.ipea.gov.br HAND model for assessment of flood exposure in the municipality of São Paulo tiff INPE & IPT, 2010 http://megacidades.ccst.inpe.br Flood occurrences in the municipality of São Paulo in the years 2018, 2019, 2020 and 2021 shapefile SIGRC, 2022 http://geosampa.prefeitura.sp.gov.br 4.3 Data analysis The data obtained from the 2010 Census and Human Development Atlas was initially organised and analysed in Microsoft Excel and subsequently processed using the statistical analysis software IBM SPSS statistics 28.0. Furthermore, the spatial data was analysed and processed using the geoprocessing software ArcGIS Pro with the association of databases and overlays of georeferenced information, which enabled the spatialisation of the calculated vulnerability index and the identification of areas where there is a coincidence between exposure and social vulnerability. 4.4 Construction of a vulnerability index The vulnerability index constructed in this research is based on the Social Vulnerability Index (SoVI) developed by Cutter et al. (2003). However, the mentioned index was elaborated in the context of the United States and a direct use of the methodology without adaptations for the context of Brazil did not seem appropriate. The main concepts of the index of reference were taken into consideration for the selection of variables among the data available in Brazil and the particularities of the local context. Although there is some debate on its utilisation, an index is yet a useful tool when attempting to quantify an abstract concept such as vulnerability, allowing an approximate representation of the socioeconomic conditions of the population. Planning for climate change: an assessment of vulnerability in the city of São Paulo 35 It is important to tailor vulnerability indexes according to the local context, taking into consideration the different scenarios and levels of development of the analysed location (Loyola Hummell et al., 2016). Therefore, the SoVI was utilised in this research as a basis for the selection of variables that could have an effect on social vulnerability. The main driver concepts identified by Cutter et al. (2003) – for instance, socioeconomic status, gender, race and ethnicity, age, infrastructure, education and family structure – were essential for the choice of variables adopted by this research. Another important factor when selecting the variables for the construction of the index was the availability of data, especially for smaller territorial units as the UDHs (human development units), which were the chosen unit of analysis for this research. Indicators of health coverage and social security, for example, were not published in the 2010 Census and the available data from other sources was collected only at a district level and not disaggregated into smaller units. After taking all these aspects into consideration, a set of variables was selected for the construction of a vulnerability index in the context of São Paulo (see Table 6). Table 6: Name and description of variables selected for the construction of an index for the assessment of social vulnerability in the municipality of São Paulo. Source: table elaborated by the author based on data from IPEA (2010) and IBGE (2010). Variable Description S O C IO E C O N O M IC S T A T U S Population vulnerable to poverty Percentage of the population living in households earning up to half a minimum salary per capita per month (R$ 255,00 Brazilian Reais as of August 2010). Population living in extreme poverty Percentage of the population living in households earning up to R$ 70,00 per capita per month, in Brazilian Reais, as of August 2010. Occupation status of the household Number of permanent private households in another condition of occupation (not owned, rented or leased). Population living in households with more than 2 residents per dormitory Ratio between the population living in permanent private households with a density greater than 2 people per dormitory and the total population living in permanent private households, multiplied by 100. The density of the dwelling is given by the ratio between the total number of residents in the household and the total number of rooms used as dormitories. Per capita income Ratio between the sum of the income of all individuals residing in permanent private households and the total number of these individuals. Values in Brazilian Reais (from 2010). G E N D E R Female population Ratio between the female population and the total population, multiplied by 100. A G E Dependency ratio Ratio between the number of people aged 14 years or less and 65 years or older (dependent population) and the number of people aged 15 to 64 years (potentially active population), multiplied by 100. Population aged 65 years or more Number of individuals aged 65 years or more. Planning for climate change: an assessment of vulnerability in the city of São Paulo 36 IN F R A S T R U C T U R E Inadequate water supply and sanitation Ratio between the number of people living in houses where the water supply does not come from the general network and where sewage is not collected through a sewerage network or septic tank, and the total population living in permanent private households, multiplied by 100. Households without a bathroom or toilet Number of households without a bathroom for exclusive use by the residents and without a toilet. Absence of garbage collection service Ratio between the population living in urban households without garbage collection service and the total population living in permanent private households, multiplied by 100. Population living in households with low quality external walls Ratio between the people living in houses with walls that are not made of masonry or wood and the total population living in private permanent households, multiplied by 100. E M P L O Y M E N T Unemployment rate of the population aged 18 years or more Percentage of the economically active population in this age group that was unoccupied, i.e., that was not employed in the week prior to the census date but had looked for work during the month prior to the date of the survey. Formality rate of the employed - population aged 18 years or more Ratio between the people aged 18 years or more formally employed and the total number of employed people in this age group, multiplied by 100. Formally employed workers were considered to be those with a signed employment contract, military personnel in the army, navy, air force, military police or fire brigade, those employed under the civil service regime, as well as employers and self-employed workers who were contributors to an official social security institute. F A M IL Y S T R U C T U R E Women who are heads of the household Number of women who are heads of the households. Women who are heads of the household, without primary education and with a child under 15 years of age Ratio between the number of women who are heads of the household, who have not completed primary education and have at least one child under the age of 15 living in the household, and the total number of women who are heads of household (multiplied by 100). Women aged 10 to 17 years who have had children Ratio between the number of women aged 10 to 17 years who have had children, and the total number of women in this age group, multiplied by 100. Households without males Number of households without male residents. Households with 5 or more residents Number of households containing 5 or more residents. E D U C A T IO N Illiteracy rate of the population aged 15 years or more Ratio between the population aged 15 years or older who cannot read or write a simple note, and the total number of people in this age group (multiplied by 100). Population aged 18 years or more with complete secondary education Ratio between the population of 18 years of age or older that concluded secondary education and the total number of people in this age group, multiplied by 100. H E A L T H Infant mortality Number of children not expected to survive the first year of life, for every 1000 children born alive. Life expectancy at birth Average number of years that people are expected to live from birth, if the level and pattern of mortality by age prevailing in the Census year remain constant. R A C E Black population Number of individuals that declared themselves as black. Pardo2 population Number of individuals that declared themselves as pardo. Indigenous population Number of individuals that declared themselves as indigenous. 2 Pardo is the term used in Brazil to describe individuals with multiracial background (IBGE, 2010b). Planning for climate change: an assessment of vulnerability in the city of São Paulo 37 Of the total of 26 chosen variables, 22 have a direct relation with social vulnerability, i.e., the higher the value of the indicator, the more vulnerable the population residing in the analysed area tends to be. The other 4 indicators have an inverse relationship, which means the higher the value of the indicator, the less vulnerable is the population living in a given area. The relation to vulnerability that each variable presents, as well as the abbreviation used for their identification in further analyses and the source of data are described in Table 7. Table 7: Chosen variables and respective relation to vulnerability, abbreviation and data source. Source: table elaborated by the author. Abbreviation Variable Relation to vulnerability Source VULN_POV Population vulnerable to poverty direct IPEA, 2010 EXTR_POV Population living in extreme poverty direct IPEA, 2010 OCC_STA Occupation status of the household direct IBGE, 2010 DENS_DORM Population living in households with more than 2 residents per dormitory direct IPEA, 2010 PC_INC Per capita income inverse IPEA, 2010 FEM_POP Female population direct IPEA, 2010 DEP_RATIO Dependency ratio direct IPEA, 2010 ELD_POP Population aged 65 years or more direct IPEA, 2010 WAT_SAN Inadequate water supply and sanitation direct IPEA, 2010 WIT_BATH Households without a bathroom or toilet direct IBGE, 2010 GARB_COL Absence of garbage collection service direct IPEA, 2010 EXT_WALL Population living in households with low quality external walls direct IPEA, 2010 UNEMP_RATE Unemployment rate of the population aged 18 years or more direct IPEA, 2010 FORM_RATE Formality rate of the employed - population aged 18 years or more inverse IPEA, 2010 FEM_HEAD Women who are heads of the household direct IBGE, 2010 FEM_HOUS Women heads of the household, without primary education and with a child under 15 years of age direct IPEA, 2010 WOM_CHILD Women aged 10 to 17 years who have had children direct IPEA, 2010 WIT_MEN Households without males direct IBGE, 2010 RES_HOUS Households with 5 or more residents direct IBGE, 2010 ILL_RATE Illiteracy rate of the population aged 15 years or more direct IPEA, 2010 SECO_EDU Population aged 18 years or more with complete secondary education inverse IPEA, 2010 INF_MORT Infant mortality direct IPEA, 2010 LIFE_EXP Life expectancy at birth inverse IPEA, 2010 BLA_POP Black population direct IBGE, 2010 PAR_POP Pardo population direct IBGE, 2010 IND_POP Indigenous population direct IBGE, 2010 Planning for climate change: an assessment of vulnerability in the city of São Paulo 38 All the data obtained from IPEA was already published for each UDH in the municipality of São Paulo. However, the data obtained from IBGE was disaggregated into census tracts and had to be processed and grouped for each UDH using spatial analyses in the software ArcGIS Pro, which could generate some uncertainty on those variables. The values per UDH of all the variables utilised in this research were compiled and are presented in Appendix D. After the selection of variables that could have a significant role in determining social vulnerability, a standardisation procedure was carried out to ensure that the variables are comparable. The selected method was refitting the range of values of each variable to relative positions across a scale from 0 to 1. In that way, the highest value of the range equals to 1 and the lowest value, to 0. For the variables that display an inverse relationship with vulnerability, a transformation is applied so that the highest values always represent a higher vulnerability. For the elaboration of the index, a factor analysis3 was utilised. This statistical method is useful when looking for an explanation for the correlation between the observable variables. It is a method used for the reduction of a large number of variables in a smaller set of hypothetical variables (factors or components) that summarise the essential information contained in the initial variables (Almeida, 2010). A factor or component represents a set of variables that are highly correlated among themselves and weakly correlated to the remaining variables. Cutter et al. (2003) expresses that the factor analysis cannot be performed with missing values, so, as recommended, a value of zero was assigned to these cases. As recognised by the reference, this procedure might underestimate the true vulnerability for the affected territorial unit, yet it is considered by the author as a better procedure than excluding the affected units from the assessment. In the case of this research, very few cases of missing values were identified in the data per UDH collected from IPEA. As for the data collected from IBGE, which was disaggregated into census tracts, the missing values might indicate tracts without population or where data was omitted for protection due to a very low number of households. As adopted by Almeida (2010), Deschamps (2004), Nguyen (2015), Cutter et al. (2003), Barbosa et al. (2019) and Loyola Hummell et al. (2016), a factor analysis using principal component analysis4 (PCA) was conducted, utilising Kaiser normalisation and Varimax rotation to provide the most robust set of independent factors. The software used for statistical processing and analysis was IBM SPSS statistics 28.0. 3 Factor analysis attempts to identify underlying factors that explain the pattern of correlations within a set of observed variables. It is often used to identify a small number of factors that explain most of the variance that is observed in a much larger number of variables (SPSS Statistics, 2021a). 4 Principal component analysis (PCA) is a factor extraction method used to form uncorrelated linear combinations of the observed variables (SPSS Statistics, 2021a). Planning for climate change: an assessment of vulnerability in the city of São Paulo 39 4.5 Factor analysis A principal components analysis was carried out for the initial set of variables selected by the author. After the analysis, the 26 initial variables were reduced to three factors (components), composed of a total of 20 variables, that explain about 78% of the total variance in the data. The procedure adopted by the author for the application of the factor analysis is described below. 4.5.1 Validation of the model The selected variables must be correlated in order to apply the factor analysis. Therefore, the chosen set of variables should be validated using the Kaiser-Meyer-Olkin method and Bartlett's test of sphericity5 (Barbosa et al., 2019; Santos, 2010). The analysis of the adequacy of the variables in this research resulted in a KMO (Kaiser-Meyer Olkin) above 0,9 (see Table 8), which indicates a very good appropriacy of the analysis for the selected data (see Table 9). Additionally, the Bartlett's test of sphericity resulted in a significance lower than 0,05 (see Table 8), which confirms the hypothesis that the correlation matrix is not an identity matrix, i.e., that there is correlation between the variables (Barbosa et al., 2019). Table 8: Results for the Kaiser-Meyer-Olkin (KMO) method and Bartlett's test of sphericity. Source: table elaborated by the author based on data resulted from this research. KMO and Bartlett's Test Kaiser-Meyer-Olkin Measure of Sampling Adequacy 0,904 Bartlett's Test of Sphericity Approx. Chi-Square 53025,224 df 190 Sig. 0,000 Table 9: Appropriacy of a factor analysis on a given data set according to the Kaiser-Meyer-Olkin (KMO) test. Source: adapted from Santos (2010). KMO Factor analysis 1,0 - 0,9 Very good 0,8 - 0,9 Good 0,7 - 0,8 Medium 0,6 - 0,7 Acceptable 0,5 - 0,6 Weak <0,5 Inappropriate 5 “The Kaiser-Meyer-Olkin measure of sampling adequacy tests whether the partial correlations among variables are small. Bartlett's test of sphericity tests whether the correlation matrix is an identity matrix, which would indicate that the factor model is inappropriate” (SPSS Statistics, 2021a). Planning for climate change: an assessment of vulnerability in the city of São Paulo 40 Furthermore, the variables should present a sample adequacy measurement6 above 0,5 in the anti-image correlation matrix (Barbosa et al., 2019). If any of the variables does not fulfil this requirement, it should be excluded before proceeding with the factor analysis. All selected variables displayed a value above 0,5 (see Appendix C). 4.5.2 Extraction of principal components Table 10: Evaluation of the communalities of each variable in relation to the extracted factors (variables which presented values below 0,5 were excluded from the analysis). Source: table elaborated by the author based on data resulted from this research. Variables Communalities Initial Extraction VULN_POV 1,000 0,938 EXTR_POV 1,000 0,688 OCC_STA 1,000 0,320 DENS_DORM 1,000 0,920 PC_INC 1,000 0,618 FEM_POP 1,000 0,367 DEP_RATIO 1,000 0,461 ELD_POP 1,000 0,567 WAT_SAN 1,000 0,543 WIT_BATH 1,000 0,471 GARB_COL 1,000 0,684 EXT_WALL 1,000 0,790 UNEMP_RATE 1,000 0,756 FORM_RATE 1,000 0,522 FEM_HEAD 1,000 0,886 FEM_HAUS 1,000 0,849 WOM_CHILD 1,000 0,347 WIT_MEN 1,000 0,709 RES_HOUS 1,000 0,920 ILL_RATE 1,000 0,639 SECO_EDU 1,000 0,921 INF_MORT 1,000 0,881 LIFE_EXP 1,000 0,913 BLA_POP 1,000 0,874 PAR_POP 1,000 0,897 IND_POP 1,000 0,230 Extraction method: Principal Component Analysis. 6 “The sample adequacy measurement for a variable is displayed on the diagonal of the anti-image correlation matrix. In a good factor model, most of the off-diagonal elements will be small” (SPSS Statistics, 2021a). Planning for climate change: an assessment of vulnerability in the city of São Paulo 41 For the extraction of principal components, it is firstly important to analyse the communalities, which represent the proportion of each variable’s variance that can be explained by the retained factors (UCLA: Statistical Consulting Group, 2021). Variables that display high values of communalities are well represented by the extracted factors. Therefore, values lower than 0,5 mean a weak relation to the extracted factors and, thus, the respective variables should be excluded before proceeding with the analysis. Among the 26 variables analysed, six displayed values below 0,5 (see Table 10) and were excluded: occupation status of the household (1); female population (2); dependency ratio (3); households without a bathroom or toilet (4); women aged 10 to 17 years who have had children (5); and indigenous population (6). The next step consists of the determination of the number of components to be extracted. For a number of variables above 30, the Kaiser criterion was used: selection of the factors whose explained variance is greater than 1, i.e., a characteristic value (Eigenvalue) superior to 1 (Santos, 2010). The results pointed 3 components which fulfil this requirement (see Table 11). Together, they are responsible for the explanation of more than 78% of the total variance. Table 11: Definition of the number of components to be extracted based on Eigenvalues above 1. Source: table elaborated by the author based on data resulted from this research. Component Initial Eigenvalues Total % of Variance Cumulative % 1 10,152 50,761 50,761 2 3,997 19,984 70,745 3 1,630 8,150 78,895 4 0,769 3,847 82,742 5 0,737 3,684 86,425 6 0,511 2,554 88,979 7 0,406 2,032 91,011 8 0,394 1,972 92,983 9 0,331 1,656 94,639 10 0,273 1,365 96,004 11 0,250 1,248 97,252 12 0,189 0,947 98,199 13 0,131 0,655 98,854 14 0,070 0,349 99,203 15 0,062 0,310 99,513 16 0,031 0,157 99,670 17 0,030 0,148 99,818 18 0,026 0,128 99,946 19 0,009 0,045 99,991 20 0,002 0,009 100,000 Extraction method: Principal Component Analysis. Planning for climate change: an assessment of vulnerability in the city of São Paulo 42 4.5.3 Rotation of factors Usually, the first solution provided by the factor analysis does not present results which allow for an adequate interpretation, so equivalent solutions can be obtained through methods of rotation of the factors (Almeida, 2010). The rotation can be used to improve the interpretation of the factors, i.e., to explain each factor with the smallest number of variables. In this research, the rotation method which allowed the best interpretation of the factors was the Varimax7 rotation. After rotation, the interpretation of the factors is done through the obtained matrix of “loadings” (rotated component matrix) by identifying loadings that equal or surpass the value of (+/-) 0,5 (Santos, 2010). In each line of the matrix, the loadings which contribute the most to the respective factors were highlighted and used to distinguish the theoretical concepts represented by each one of the factors (see Table 12). Table 12: Rotated component matrix. Source: table elaborated by the author based on data resulted from this research. Variables Components (factors) 1 2 3 VULN_POV 0,838 0,008 0,493 EXTR_POV 0,635 -0,021 0,534 DENS_DORM 0,886 0,012 0,376 PC_INC 0,796 -0,006 -0,049 ELD_POP -0,724 0,076 -0,074 WAT_SAN 0,307 -0,030 0,700 GARB_COL 0,190 -0,057 0,823 EXT_WALL 0,252 -0,103 0,858 UNEMP_RATE 0,692 0,008 0,537 FORM_RATE 0,517 -0,019 0,489 FEM_HEAD -0,299 0,917 0,017 FEM_HAUS 0,837 -0,004 0,396 WIT_MEN -0,487 0,706 0,090 RES_HOUS 0,133 0,955 -0,065 ILL_RATE 0,752 -0,001 0,263 SECO_EDU 0,911 0,019 0,317 INF_MORT 0,878 0,016 0,346 LIFE_EXP 0,906 0,025 0,321 BLA_POP 0,130 0,938 -0,111 PAR_POP 0,217 0,923 -0,090 Extraction method: Principal Component Analysis. Rotation method: Varimax with Kaiser normalisation. 7 Varimax is an orthogonal rotation method that minimises the number of variables that have high loadings on each factor, simplifying the interpretation of the factors (SPSS Statistics, 2021a). Planning for climate change: an assessment of vulnerability in the city of São Paulo 43 The first component has the highest influence on social vulnerability – accounts for about 40% of the total variance (see Table 13) – and is predominantly explained by 12 variables, which compile information on education, health and socioeconomic status of the population (see Table 14). Component two has the second highest influence (about 20%) and is explained mainly by 5 variables, which illustrate the connections between family structure, gender and race. Finally, component three explains about 18% of the total variance and is described largely by 3 variables that portray housing and infrastructure conditions of the households. Table 13: Total variance explained by the three extracted components (factors). Source: table elaborated by the author based on data resulted from this research. Total variance explained Component Rotation sums of squared loadings Total % of Variance Cumulative % 1 8,082 40,410 40,410 2 4,005 20,025 60,435 3 3,692 18,460 78,895 Table 14: Variables (and respective loadings) that best explain each of the components extracted by the factor analysis. Source: table elaborated by the author based on data resulted from this research. Components Variables Loading Component 1: Education, health and socioeconomic status Population aged 18 years or more with complete secondary education 0,911 Life expectancy at birth 0,906 Population living in households with more than 2 residents per dormitory 0,886 Infant mortality 0,878 Population vulnerable to poverty 0,838 Women heads of the household, without primary education and with a child under 15 years of age 0,837 Per capita income 0,796 Illiteracy rate of the population aged 15 years or more 0,752 Population aged 65 years or more -0,724 Unemployment rate of the population aged 18 years or more 0,692 Population living in extreme poverty 0,635 Formality rate of the employed - population aged 18 years or more 0,517 Component 2: Family structure, gender and race Households with 5 or more residents 0,955 Black population 0,938 Pardo population 0,923 Women who are heads of the household 0,917 Households without male residents 0,706 Component 3: Housing and infrastructure Population living in households with low quality external walls 0,858 Absence of garbage collection service 0,823 Inadequate water supply and sanitation 0,700 Planning for climate change: an assessment of vulnerability in the city of São Paulo 44 4.5.4 Evaluation of consistency The construction of indexes presupposes the verification of their internal consistency, which can be confirmed by Cronbach's alpha8 (Santos, 2010). Component 1 – education, health and socioeconomic status – and component 2 – family structure, gender and race – presented a Cronbach’s alpha above 0,9 (see Table 16 and Table 17), which indicates a very good internal consistency (see Table 15). For component 3 – housing and infrastructure – the Cronbach’s alpha displayed a value of approximately 0,84 (see Table 18), which suggests a good internal consistency (see Table 15). Table 15: Verification of the internal consistency of the factors through Cronbach’s alpha. Source: adapted from Santos (2010). Cronbach’s alpha Internal consistency >0,9 Very good 0,8 - 0,9 Good 0,7 - 0,8 Acceptable 0,6 - 0,7 Weak <0,6 Unacceptable Table 16: Reliability statistics for component 1 - education, health and socioeconomic status. Source: table elaborated by the author based on data resulted from this research. Reliability statistics - component 1 Cronbach's alpha Cronbach's alpha based on standardised items Number of items (variables) 0,935 0,929 12 Table 17: Reliability statistics for component 2 - family structure, gender and race. Source: table elaborated by the author based on data resulted from this research. Reliability statistics - component 2 Cronbach's alpha Cronbach's alpha based on standardised items Number of items (variables) 0,932 0,935 5 Table 18: Reliability statistics for component 3 - housing and infrastructure. Source: table elaborated by the author based on data resulted from this research. Reliability statistics - component 3 Cronbach's alpha Cronbach's alpha based on standardised items Number of items (variables) 0,837 0,841 3 8 Model of reli