RISCO DE INUNDAÇÃO NA CIDADE DE BELÉM (PA): A PERCEPÇÃO TÉCNICA E SOCIAL DO RISCO FLOODING RISK IN BELÉM (PA) CITY: THE TECHNICAL AND SOCIAL PERCEPTION OF RISK

1 Mestre em Planejamento e Gestão do Território pela Universidade Federal do ABC, Santo André – SP, Brasil. E-mail: joaovonp@gmail.com 2 Doutor em Engenharia Civil (UFPA). Professor da Universidade Federal do Pará, Belém – PA, Brasil. E-mail: maues@ufpa.br 3 Doutora em Engenharia Civil. Professora da Universidade do Porto, Porto, Portugal. E-mail: carocha@fe.up.pt Revista Brasileira de Gestão e Desenvolvimento Regional


Introduction
Fast urbanization has brought prosperity and opportunity to people, cities have become the economic engines as well as centers of technology and innovation in their countries, but this is the case for well-planned and managed cities. On the other hand, cities have become major riskgenerating centers when growth is combined with the impacts of extreme weather events and the increase in poverty (UNISDR, 2012).
In 2016, worldwide, 569.4 million people were affected by natural disasters, the largest number ever measured since 2006. Also, approximately one billion people are living in urban slums (CRED, 2016;UNISDR, 2010). Thus, it is clear that there are two worrying situations: numerous people are vulnerable and occurrences of natural disasters are growing.
In Belém, flooding is understood as a hydrological process of overflowing water from drainage channels to marginal areas, submerging coastal areas due to the temporary rise in water level. The causes are frequent climatic events in the Amazonian winter period, usually beginning in December and ending in May, such as increased rainfall and high tides, enhanced by topographic conditions, soil waterproofing, and inefficient drainage infrastructure. (CAMPOS et al, 2014;CPRM, 2015;PINHEIRO, 2015;PONTES et al, 2017).
According to the last demographic census (IBGE, 2010), Belém has approximately 1.4 million inhabitants, with 758,524 thousand people living in subnormal agglomerations (see IPEA, 2014, p. 11), that is, 54% of the population lives in vulnerable conditions. This data becomes even more serious when compared to other important capitals, as Belém is ahead in this regard, for example: São Paulo 11%, Rio de Janeiro 22% and Salvador 33%.
Therefore, this research aims to identify flood risk in the city of Belém, so that greater understanding is obtained by integrating two paths of analysis: the technical and the social. This work gains more relevance due to this combination, not being restricted to only a perception of risk, and hence contributing to further research and discussions.

Risks e vulnerabilities
Studies on risks are individualized and fragmented for each area of knowledge according to their perspectives of understanding (MARANDOLA JR.;HOGAN, 2005apud SANTOS et al, 2015, meaning they are produced from the perspectives of geography, geology, sociology, among other areas, bringing together a set of meanings. For the French geographer, Veyret, the causes and consequences of risks must be known according to the diversity of classifications (VEYRET, 2007 apud SOUZA;LOURENÇO, 2015). Initially, damage and population exposure were considered in the approach to natural risks. Later, social and technological risks were incorporated, changing the nomenclature to environmental risks. Thus, risks are understood by the manifestation of nature together with how society has taken over the environment where it lives (EGLER, 1996 apud SOUZA;LOURENÇO, 2015).
In Geology, studies consider that the natural dynamics of the planet is comprised of processes that occur regardless of human activities, however, the action of man through to changes in land use and occupation can stimulate, streamline and increase many of those processes (CERRI; AMARAL, 1998;BARBOSA;FURRIER, 2017).
To make understanding easier, risk analysis categories were established, followed by an adjective that qualifies them and with a clear determination of the threats that constitute them (VEYRET, 2007;CASTRO et al, 2005apud ALMEIDA, 2012CERRI;AMARAL, 1998apud ROCHA, 2005apud SANTOS et al, 2015: • Natural hazards: Hurricanes, droughts, storms, hail, lightning, flooding, inundations, earthquakes, volcanic activities, tsunamis, landslides, erosion, diseases caused by viruses, bacteria, poisonous animal bites, etc. • Technological risks: Leaks of toxic, flammable, radioactive products, accidental fires, vehicle collisions, aircraft crashes, etc. • Economic, geopolitical, and social risks: Economic crises, insecurity and violence due to urban socio-spatial segregation, political-ideological conflicts, etc. Environmental risks are considered the largest category of analysis, covering natural, technological, economic, geopolitical, and social risks. It is noteworthy that this classification is carried out to aid the reasoning of the types of existing risks, given that a phenomenon may be present in more than one of the groups. For example, urban flooding may have a purely natural cause, but it can also be influenced by human activities of soil impermeabilization, insufficient urban drainage, and housing in flooded areas. When expanding the concept of risk in the light of Sociology, an important issue is the social perception of risk, which is understood as a social product that only exists because the society or individual can perceive it (constructivist perception). However, some defend the existence of an independent risk of being perceived, without interfering in the impact that can happen (objectivist or realistic perception). Thus, it is understood that the social perception is undeniable although it cannot be restricted in itself, so the analysis of risks in the urban space must embrace the combination of social judgments and scientific parameters (GUIVANT, 1998;VEYRET, 2007;TAVARES et al, 2017).
Another concept intrinsic to risk is vulnerability, that It can be defined by the characteristics of a society that enhance the susceptibility of negative consequences when a threat is manifested (UNISDR, 2009apud KELMAN, 2018. For authoress Cutter, the different approaches to existing vulnerability result in three main attitudes (CUTTER, 1996 apud MARANDOLA JR;HOGAN, 2005 apud SOUZA;LOURENÇO, 2015). The first vulnerability is premised on the spatial or geographical dimension, understood as a result of the physical aspects of a given region, designated by the extent to which an environment is susceptible to the occurrence of natural threats, in other words, it concerns the areas where the manifestation of a natural phenomenon occurs (TAGLIANI, 2003;ALVES, 2006apud ESTEVES, 2011PINHEIRO, 2015).
The second vulnerability is related to the characteristics of the community, contrasting the physical sense of the risks, taking into account economic issues, the offer and access to public services, the ways of life of the population living in risk areas, among others. In this thought, vulnerability concerns not only the populations that are exposed but also as a result of the social needs that affect them (CUTTER, 2006;2010apud MENDES et al, 2011PINHEIRO, 2015).
The third one is the combination of social and environmental dimensions in the identification and analysis of the vulnerability. Thus, the use of the terminology "socio-environmental vulnerability" becomes relevant because environmental risks depend on social, economic, technological, cultural, environmental factors, etc. (ESTEVES, 2011;PINHEIRO, 2015).
Lastly, another relevant issue regarding threats and vulnerabilities is the possibility of all these elements being viewed through Geographic Information System (GIS) for the construction of physical and social indicators. GIS can integrate diversified data sources and assist in the understanding of risks as well as vulnerabilities, in addition to being able to assist in decision making to intervene in the territory (CUTTER, 2003 apud SOUZA;LOURENÇO, 2015).
According to Marcelino (2008), an indicator includes a diversity of data (maps, measurements in the field, satellite images, questionnaires, etc.) that enable the identification of the characteristics and the context of the environment. For the author, "whenever possible, quantitative data should be used from reliable sources, as well as long historical series and methods of analysis involving mathematical and physical models". The goal is that the risk assessment is as close to the reality of the place and not the perception of the technician responsible for the analysis. Also, this assessment has to be made possible for replication and comparison with other areas.
For KOBIYAMA (2004( apud MARCELINO et al, 2006, mapping risk areas is the objective of an analysis instrument since, based on a map, it is possible to elaborate a series of preventive, emergency, and joint actions between population and power public to provide a permanent defense of society against a natural disaster.

Method
This exploratory research is constituted by the method of mathematical modeling, survey, and case study according to theoretical reference, consisting of two stages of analysis. The first stage is related to the technical point of view with the use of a mathematical model that allows viewing and analyzing the different levels of exposure, vulnerability, and flood risk through the collection of social, economic and environmental indicators, resulting in the creation of cartographic maps using GIS.
The second stage approach is the social point of view through the application of a questionnaire to a portion of the population to build an understanding of how the population perceives the flood events, resulting in percentage data on the frequency of floods and the losses experienced.
Initially, an attempt is made to develop a flood risk index on a intra-municipal scale, adapted from the Disaster Risk Indicators in Brazil (DRIB) index, which is developed on a municipal scale. The DRIB index is based on the World Risk Index, whose theoretical concepts, in the context of natural disasters, state that the risk derives from a combination of physical factors and the vulnerability of exposed elements, having equation 1 as calculation premise (ALMEIDA et al, 2016; WRR, 2016): = × Eq. 1 Where the risk index ( ) is the product between the exposure index ( ) and the vulnerability index ( ), with the numerical values of these components between zero and one: 0 (zero) indicates that there is no exposure, vulnerability or risk and 1 (one) indicates that these are maximum. To display the indexes, cartographic maps are elaborated on Quantum Geographic Information System (QGIS) and the numerical data are classified qualitatively into five classes: very low (0.00 -0.20); low (0.21 -0.40); medium (0.41 -0.60); high (0.61 -0.80); and very high (0.81 -1.00).
That said, the first step is to know the exposure through the following georeferenced files: areas susceptible to flooding (CPRM, 2015; GEOFABRIK, 2018); territorial map and population data (IBGE, 2010). On QGIS, population data are linked to the territorial map to identify the different demographic densities. Thus, one can overlay the map of areas susceptible to flooding on the territorial map and apply the intersection between these two layers.
Subsequently, the different levels of exposure are identified at the district scale -a unitary element of the Municipal System of Urban Planning and Management (Sistema Municipal de Planejamento e Gestão Urbana) (BELÉM, 2012) -and then, the exposed population of a district is divided by the total population, resulting in cartography that represents the spatial dynamics of risk, see equation 1.1 shown below (adapted from ALMEIDA et al, 2016): The second step is to know the vulnerability through thirty-two indicators that embrace the social, economic, and environmental conditions of the Brazilian territory, being divided into susceptibility ( ), coping capacity ( ), and adaptive capacity ( ). Socioeconomic and cultural conditions as well as the performance of public institutions in dealing with risks are strongly associated with vulnerability, taking into account that the series of inequalities and the inefficiency of the State result in barriers to risk reduction. The following are the vulnerability indicators divided into categories: The selection of indicators is based on the World Risk Index, being related to the eight Millennium Development Goals and the United Nations Hyogo Board for Action on Disaster Risk Reduction (ALMEIDA et al, 2016). It is noteworthy that the indicators treated here refer to technical data collected from several official and publicly available sources, meaning that the exposure, vulnerability, and the risk itself are identified from numerical and georeferenced data. Table 2 shows the categories, weights (weighting), databases, years of publication, scales, and units of the indicators that make up , , and . Concerning Table 2, the susceptibility refers to the population's predisposition to suffer damage, comprising nine indicators (a, b, c, d, e, f, g, h, i) divided into four categories (public infrastructure; housing conditions; poverty and dependence; economic capacity and income).
Coping capacity refers to the ability of the municipality to prepare before, endure during, and recover after flood impacts. Eight indicators are used (a, b, c, d, e, f, g, h) divided into four categories (government and authorities; disaster preparedness and early warning; medical services; material coverage).
Adaptive capacity concerns the capacity of the municipality and its population to selftransform as a society. For example, gender equity and the preservation of the environment are considered contemporary themes that may indicate the extent to which a society can adapt. Fifteen indicators are used (a, b, c, d, e, f, g, h, i, j, k, l, m, n, o) divided into five categories (education and research; gender equity; environmental conditions/ ecosystem protection; adaptation strategies; investments).

1.2.3
Regarding the calculation adjustments, most indicators are in percentage, so has been divided by 100 (one hundred) to be in a range from 0 (zero) to 1 (one). Also, there is the "ratio" unit, which is the division of the indicator's value by its total. Thus, based on the census information, it is known how many measures have already been and should be implemented.
Lastly, for comparative analysis, indicators are sought on the smallest scale possible, with three scales being identified: district; municipality; and state. Despite that, this limitation does not prevent the analysis from being carried out; it only reduces a greater distinction between areas.
The social perception of flood risk is developed through the application of a questionnaire to survey how the population lives with flood events. Therefore, in the elaboration of the questionnaire, simple questions, which were easy to answer, were the focus to achieve faithful interpretations. It should also be noted that a representative sample of the population is used, free from any tendency or pre-judgment. The aim is really to collect the population's perception, without providing any prior information or selecting any group prone to risk and socially vulnerable. Hence, the questionnaire is applied by different means and locations to seek the largest and most diverse participation of the population.
The questionnaire was designed to be applied using an electronic form (Google Forms) and printed on paper, having the data processed on Microsoft Office Excel 2016. The ideal sample size (n) was defined according to AYRES et al (2015), through equation 2: = + Eq. 2 In which is the size of the population and the size of the provisional sample that considers the margin of error (Er) in probabilistic terms of 5%. Thus, firstly, the calculation of was done based on the margin of error to know the provisional sample size: = = ( . ) = 400 Eq. 2.1 Therefore, for a population size equal to 1,392,332 inhabitants (IBGE, 2010), the quantity for the questionnaire to be applied is given by the following calculation: = . .
. . + = 400 questionnaires Eq. 2 Initially, the questionnaire was validated through an experimental application to determine the quality of the questions, in other words, what they were proposed to measure.
With the necessary adjustments, the final application was done through a request for collaboration sent to the e-mails of students, employees, and teachers at the Faculty of Civil Engineering (FCE) of the Federal University of Pará (FUPA) including the link to the questionnaire in Google Forms and then, the questionnaire was randomly applied in-person to the students at FCE. The questionnaire (electronic one) was applied to people who attend FUPA due to the fact the authors had access to their e-mails. Nevertheless, they cover a large and representative sample of the city, as the interviewees live in different districts of the municipality targeted by the survey.
Finally, to diversify the public, the authors had assistance from the Civil Defense agents, who allowed that the interview was applied to the population in the social action promoted by the City of Belém in the neighborhood of Pedreira on April 4, 2018(BELÉM, 2018. In the first section of the questionnaire, through five questions such as gender, age, education, occupation, and district of residence, it is intended to characterize the interviewees. In the second section, the five questions aim to assess the awareness of risk, memory, frequency, and if they have already been affected as well as the damage experienced (Table 3).  It is noteworthy that question 5 allowed a comparative analysis among each district of the municipality, in addition to question 7, which intended to qualify interviewees to answer the following questions in the questionnaire, emphasizing that if the answer was negative to this question, the interviewee should discontinue the application of the questionnaire.

Results e discussions
It is known that cities are increasingly threatened by natural events, which have worried public managers to fight risks. In this regard, this research contributes to regional development in both planning and management. For example, the promotion of spatial planning that reduces exposure and population vulnerability or the adoption of management measures focused on building urban resilience.
The matter is that urban risks need to be taken seriously through a holistic view, for a comprehensive understanding of the phenomena be sought. It is in this perspective that the object of this study is explored, by combining technical and social perception, contributing to future research and discussions on regional development.
At this moment, the results of the technical perception will be presented according to the official district division of the municipality: Belém, Benguí, Guamá, Entroncamento, Icoaraci, Mosqueiro, Outeiro, and Sacramenta (BELÉM, 2012).
The exposure index found is between 0.61 to 1.00, representing a worrying situation, as the numbers are high: Benguí (0.637), Belém (0.659), Entroncamento (0.665), and Icoaraci (0.721) presented high indexes whereas Outeiro (0.829), Mosqueiro (0.935), Sacramenta (0.946), and Guamá (0.994) presented very high indexes. Bellow, it is shown the official division map of the municipality, as well as the exposure indexes found for each district. According to the figure above, it can be seen that the population exposure occurs along the adjacent drainage channels that cut through the territory, in addition to large coastal areas prone to flooding. A strong feature is the occupation of areas that are susceptible to flooding as well as unsuitable for housing by irregular housing.
Regarding the vulnerability indexes, all districts are between 0.41 and 0.60, so they have a medium vulnerability index. The results show that the population is more exposed than vulnerable, however, it is worth mentioning that most of the vulnerability indicators used are on a municipal or state scale, which influences less distinction among districts.
It cannot be overlooked that Belém has large areas of subnormal agglomerations and that these occupations influence the increase in vulnerability due to its characteristics. In Figure 2, the vulnerability index map is presented together with the map of the subnormal agglomerations.  When analyzing the answers about the district of residence, it is observed that a large portion lives in Belém (29.06%) and Sacramenta (23.40%). The percentages of Benguí (15.27%), Guamá (12.56 %), and Entroncamento (11.58%) are close, while the percentages of Icoaraci (3.45%), Mosqueiro (2.22%), and Outeiro (2.46%) are relatively low, considering that these districts are located at a much greater distance from the city center, making it more difficult to apply questionnaires in those areas. The results of the questions that involve the population living with floods are presented below, classified by district: Regarding awareness, the risk is best translated by the word robbery, chosen by the majority of interviewees in all districts (above 73% in each district). This data reveals that the population does not perceive flooding as the main risk and, in some districts, it is not even perceived as a risk. Instead, the risk of being robbed is what most concerns.
Regarding memory, interviewees were asked if they remember the occurrence of floods in the district of residence. Mosqueiro and Outeiro had higher percentages for the answer "No". All other districts had higher percentages for the answer "Yes" and those interviewees who answered "No" or "I don't know" did not answer the following questions. Consequently, it is understood that throughout the municipality, part of the population has already witnessed some flood events.
Regarding the frequency of flood events in the district of residence, the option "Monthly" is the most chosen among Benguí, Entroncamento, Icoaraci, Mosqueiro, and Outeiro. Belém presented the highest percentage for the answer "Annually", which can be considered the lowest frequency in the municipality. Guamá and Sacramenta had the worst ranking of interviewees who indicated the answer "Weekly".
The district of Belém had the highest percentage of interviewees whose cars or motorcycles have been damaged. In Benguí and Icoaraci, the interviewees suffered from interdiction on public roads to access their homes, similar to Entroncamento, where the interviewees suffered from traffic disorder. In Mosqueiro, Outeiro, Guamá, and Sacramenta, the interviewees had both their furniture and equipment damaged. In this way, it is noticed that the damages and losses are related to the means of transport and urban mobility or furniture and equipment.
Therefore, when completing the development of the analyses through the combination of technical and social perception, four areas of risk are identified according to the similarities and differences that characterize them. Certainly, these perceptions converge, making it possible to logically structure a table with the results obtained and consolidate the possibility of integrating them, as shown in Table 5: The spatial heterogeneity of risk is perceived on a direct link to the historical process of urbanization due to the dynamics of the city's development. The formation of the central area (green area) was configured on the high and sanitized lands, occupied by families of higher income while in its surroundings, low and floodable land (red area) became alternatives for poorer families (RODRIGUES et al, 2013).
In the expansion area (yellow area), the occupation had occurred spontaneously, in most cases with precarious urban infrastructure, having large land lots with physical configurations of areas prone to flooding (RODRIGUES et al, 2013). The districts of Outeiro and Mosqueiro (blue area), also part of the expansion area but formed by groups of islands, concentrate the lowest demographic densities and urban infrastructure (PEREIRA, 2009).

Conclusion
Firstly, the flood risk index was a useful tool for producing information, resulting in thematic maps of risk and vulnerability by the use of GIS. It is worth noting that there is a substantial generalization of vulnerability indicators, considering that they are available, in the vast majority, in numerical tables incompatible with the GIS format or solely in percentages. Therefore, it is interesting that these data are made available as georeferenced files.
Another issue is the lack of a probabilistic component, given that demographic information from the 2010 Census was used and also due to the difference in years among the indicators' databases.
Then, the social perception of flood risk was developed through a participatory process of the population, resulting in collected information that contributed to the characterization of the risk. The application of the questionnaire has become a challenge of communication and consultation to citizens, successfully overcome beyond expectations, given the production of information close to the reality of the municipality of Belém.
Concerning its limitations, the questionnaire presented a relative diversity of the interviewed public since each district showed a different percentage of responses, this would allow the analysis of the most reliable social perception for each area.
Therefore, it can be concluded that the quantitative survey of the exposure and vulnerability indicators, as well as popular participation, allowed a qualitative comparative assessment among the districts that compose the study area. The flood risk index is an essential spatial analysis tool to support decision making, especially because it considers indicators from different sorts for a global analysis of the real situations that the municipality has faced and, in the same way, the questionnaire was established as a tool for social analysis.
It is also recommended for future research to work together with experts from environmental, sanitary, and water resources engineering, among others, to enrich this study through a holistic view, by adding a greater amount of information from different points of view. Such a study would bring a greater understanding of the extent of flood risk by adopting a multifaceted approach, resulting in a document rich in details for flood equalization.
Lastly, it is recommended to carry out a comparative study between cities that present flood events and have implemented projects that have been successful in solving these problems.