The results were divided into four parts. The first section describes the results obtained from the assessment of sectors under risk of flooding and unplanned settlements, paying particular attention to the population at risk. The second section presents the assessment of individual indicators, focusing on household income, population age groups, sanitation services and housing conditions. The third section presents the vulnerability index assessment for the urban sectors of the ADE.
Assessment of flood risk areas and unplanned settlements
The distribution of the urban sectors along the ADE is shown in Fig. 3. The sectors are classified as:
under risk of flooding;
considered unplanned settlements;
both situations (under risk of flooding and considered unplanned settlement), and;
More than 1.2 million people representing 41 % of the total urban population are exposed to risk of flooding in extreme water level events. Of these people at risk, 307,392 live in the SUS, 686,347 live in the BMR and 239,660 live in the MMR (Fig. 4). About 41 % of the total urban population lives in unplanned settlements. Over 1.1 million people in the BMR are living in unplanned settlements, while MMR and SUS together reaches almost hundred thousand inhabitants. Results indicate that about half million people or 19 % of the population under risk of flooding are also living in unplanned settlements, mostly in the BMR (see Fig. 4).
Floods are a natural process in the Amazon but also the main natural hazard posing the greatest threat to the urban population in the ADE (CEPED UFSC 2012). Severe flood episodes are seasonal (Benatti 2011), characterized by a gradual rise of water that influences social life and organization in the urban ADE (Costa and Brondizio 2011; Vogt et al. 2016). There is no significant loss of life because of this gradual rise in water level and only eight deaths due to flood episodes were reported between 1991 and 2012 in the urban ADE (CEPED UFSC 2012).
Nevertheless, 30 events of severe floods and 20 events of flash floods in the urban ADE between 1991 and 2012 were declared as public disasters. These resulted in the displacement of thousands of people as well as affecting infrastructure (CEPED UFSC 2012). The magnitude of extreme flood impacts can greatly affect other aspects of these urban areas. Floods can affect people directly (e.g., through injuries) as well as indirectly (e.g., through displacement, the destruction of homes, spread of waterborne diseases, water shortages, disruption of essential services, impact on resources and financial loss), (EEA 2015; CEPED UFSC 2012).
Poorer households tend to live in riskier areas of the urban settlements (Adger 2006; Newton et al. 2012). Urban spaces in the ADE have little or poor planning due to:
high rates of urbanization (Costa and Brondizio 2011);
increased migration from rural to urban centers (Costa and Brondizio 2011; IBGE 2011);
occupation of marginal and lowlands areas, and (Benatti 2011);
land invasion and unclear property titles (Benatti 2011; IBGE 2011).
These factors contribute to increase the exposure of people to risk from flooding, disease and other chronic stresses (Adger 2006). Living in unplanned, often overcrowded settlements, increases the exposure of inhabitants to other risks, including pathogens, insect vectors, indoor pollution, violence, drug abuse and other social problems (WHO 2003).
Most of the unplanned settlements in the ADE are densely packed spatially and the housing units are crowded (IBGE 2010). They are often built of flimsy materials or on stilts, above the rivers. Stilt houses compose 83 % of housing units in unplanned settlements of the MMR (IBGE 2011). Residents in the unplanned settlements of BMR represent about 10 % of the total for Brazil, ranked the third by size after Rio de Janeiro and São Paulo. Moreover, our results indicate that 56.6 % of the residents of BMR live in unplanned settlements. This proportion is the largest among all metropolitan areas in Brazil (IBGE 2011).
Assessment of individual indicators
The descriptive statistics and overview of the results are shown in Table 2. The results indicate a huge variability in the indicators for the three region of the urban ADE elucidated by the large standard deviation for these indicators.
Figure 5 shows the percentage of households with a total monthly income of: (1) no income and income less than one minimum wage, (2) between one and five minimum wage, (3) more than five minimum wage. Results indicated a very low percentage of households (a mean of less than 5 % in all regions) receives more than the minimum income that is considered necessary to supply the basic needs of the individual and family (DIEESE 2015). A mean of about 65 % of the households receives less than one minimum wage in urban sectors of MMR and BMR, while in urban sectors of SUS this mean reaches 85 % of the households, placing the ADE region among the poorest urban areas in Brazil (IBGE 2010). This situation results in a high dependency of families on federal subsidy programs, such as the Bolsa Familia (Costa and Brondizio 2011; Brondizio et al. 2013).
The cities in the ADE lack an adequate fiscal framework and administrative capacity to collect taxes (Brondizio 2011). This results in a high dependency on federal subsidies for investment in urban infrastructure (Brondizio et al. 2013). This insecurity and the lack of economic well-being, represented by the high number of poor households, contribute to the vulnerability of people and assets along the urban spaces of the ADE.
Figure 5 also shows that urban sectors of the ADE have similar proportions of children and the elderly population, indicating a minor proportion of population with limited physical capacity to deal with the impacts of hazards (Nobre et al. 2010).
Figure 6 shows the infrastructure patterns for the urban sectors along the ADE. Solid waste collection is provided to 100 % of households in most sectors of the metropolitan regions and around 97 % at SUS. However, a significant number of sectors lack a waste collection service and most municipalities lack appropriate landfills and treatment facilities. Solid waste accumulates in streets and is often dumped into watercourses, which clogs natural water flow and contributes to the flooding in the urban spaces of the ADE (Hardoy and Pandiella 2009; Costa and Brondizio 2011). Waste dumping on streets creates spaces for disease sprawl, insects, rodents and drug consumption. This situation also contributes to increased pollution of the land and water systems.
Domestic waste water collection is still very limited with less than 20 % of households connected to this service in the metropolitan regions and almost totally absent in sectors located in the SUS (see Fig. 6). Due to this deficit, domestic effluents are disposed of in urban streams and rivers, affecting water quality.
There is a variation in the public water supply service within all sectors. Although 80 % of households are connected to public water supply, there are still households with no, or minimal, access to this service (see Fig. 6). These households mainly depend on water collection from the river, rain-harvesting or collective tap water, where the water is provided for the neighborhood as a whole. Residents that extract drinking water from unsafe water sources, such as directly pumping from rivers, are constantly exposing themselves to potential health risks (Hardoy and Pandiella 2009).
In terms of housing conditions, the percentage of households without a drainage system is very large. In the SUS, almost 100 % of the houses have no drainage system (see Fig. 6). The drainage system is limited in households of the two metropolitan regions in the ADE. Lack of drainage systems increases the risk of floods while the accumulation of uncollected wastes blocks drains and surface runoff (Hardoy and Pandiella 2009).
Despite the comprehensiveness of solid waste collection service within the urban ADE, there is still a large variability of households with accumulation of solid waste. The mean percentage of households with incidence of open-air waste water reaches near 50 % in BMR and MMR and 34 % at SUS, reflecting the undesirable consequence of the poor housing conditions along the ADE (see Fig. 6).
Results indicate the deficiency of urban infrastructure along the ADE, especially in the small urban spaces. Lack of public services and infrastructure such as drinking water, sewage, proper waste collection and disposal, increases the health risks of population, posing additional challenges to advancing social and economic progress (IPS Amazônia 2014). Flood impacts can exacerbate these risks, since rising water levels can expose people by direct contact with contaminated water after a disaster (CEPED UFSC 2012). Health risks are related to waterborne diseases, including: leptospirosis, cholera, hepatitis, amoebic dysentery, typhoid fever, diarrhea caused by Escherichia coli, and vector-borne diseases that are related to water, such as yellow fever, malaria, dengue and zika virus.
The findings of this study corroborate those of Costa and Brondizio (2011), who analyzed data from IBGE 2000 and found no significant difference between cities built at different historical periods and available public services and infrastructure. Our study found no recent relevant improvements with respect to the services and infrastructure of urban spaces of the ADE. Exceptions exist to some urban infrastructure not assessed in our index, such as electricity service, which has improved when compared to census data 2000 presented in Costa and Brondizio (2009, 2011). Today, nearly 100 % of the households in almost all urban sectors of the ADE are served by electricity (IBGE 2010), possibly due to investments made during the Federal Government Program called Luz Para Todos (“Light for All”).
Some other investments in sanitation and infrastructure on unplanned settlements along the ADE were made since 2007 when the federal government initiated a Program of Accelerated Growth (PAC) dedicated to urban infrastructure (BRASIL 2015). Yet, according to Avelar et al. (2013), only 10 % of the population of the Belem Metropolitan Region (BMR) living in unplanned settlements have benefited from the incentives of the PAC. The integrated system for waste water treatment for BMR was planned but not implemented (Avelar et al. 2013).
Relationship between indicators
A correlation matrix was generated (see supplementary material) to explore the relationships between household income, sensitive age groups, sanitation services and housing conditions. Some of the major findings of the correlation analysis yielded the following
Income is positively correlated with access to drainage, and negatively with effluent collection and solid waste collection. Households with income below less than minimum wage or with no income had the lowest level of access to these services. Households with income between one and five minimum wage have significant more access to domestic effluent collection and solid waste collection;
Households income below one minimum wage and no income tend to have a higher proportion of sensitive population (children and elders). This may be attributed to limited sources of income for these two major age groups as children are generally economically inactive population and elderly may represent a source of income through retirement pension.
There is no significant correlation between sanitation services (water supply, effluent collection and solid waste collection) and housing conditions (accumulation of solid waste and open-air sewage. Although solid waste collection is very representative in across urban sectors, solid waste is also dumped in street corners, river borders or drainage channels and river ways. Very often sewer pipes are clogged, so waste water is drained in the open air even in sectors served with sewage collection.
Assessment of the vulnerability index
Figure 7 shows the overall vulnerability for the urban sectors of the ADE. Results indicated that about 3.37 % of the urban ADE area presented a low degree of vulnerability, against 34.72 % with moderate degree, 60.30 % with high degree and 2.60 % with very high degree of vulnerability. While populations are more concentrated in areas with moderate degree of vulnerability in all regions, about 34 % of these populations are located in areas with high to very high degree of vulnerability, reaching over 1 million inhabitants.
Our results indicate that vulnerability of ADE cities increases from the city center to the recent peri-urban expansion (see Fig. 7), corroborating with results found by Perz (2000) which pointed out that the urban population growth during the 1980s have led to an intra-regional difference within the city in terms of environmental quality. He indicated that old established urban areas showed higher environmental quality, suggesting a deterioration which occurred in part due to the establishment of newly formed occupations. His study detected that households in new urban areas have considerably less resources and services, such as waste collection and water supply and are more exposed to environmental hazards than the older urban areas, a trend still observed in this study.
A correlation analysis was performed between an independent variable which in our case was the average household income within each sector and the overall vulnerability. While looking at percentage households in different income categories allows us to assess vulnerabilities within sectors, the average household income looks at income and vulnerability relationships between different sectors. The Spearman’s correlation value of r = −0.533 with a p value <0.01 suggests an inverse correlation between the average household income and the vulnerability of urban sectors. Thus, the higher the average income of the households, the lower the vulnerability within the urban sectors of the ADE. This suggests that poorer urban spaces present higher degrees of vulnerability.
The degree of vulnerability of the urban ADE will define the impacts of extreme events on society and urban systems. Poor planning and rapid urbanization result in high density, high concentration of poverty, and higher levels of exposure to environmental hazards. Coupled with inadequate investments in infrastructure and public services, the increasing exposure of urban settlements and inhabitants to flood hazards will escalate the severity of potential impacts due to future climate change (IPCC 2014).
Hence, to cope with the accelerated urbanization of the ADE (Brondizio 2013), Costa and Brondizio (2009, 2011) argue that municipalities of the ADE should seek ways to diversify their source of income and decrease their dependency on federal government subsidies. While most municipalities in the ADE have strong and active resource economies (agroforestry, fishing, mining, agriculture, and ranching), industries aggregating value and taxation to these resources are virtually inexistent.
The vulnerability assessment presented in this study shows the distinct roles for socio-economic sensitivity, infrastructural vulnerability, and flood exposure. These are dependent not just on the available resources and services in this region but also on the adaptive capacity of the population living in these areas to make use of existing services. This reiterates the importance of considering different dimensions of vulnerability as outlined in our conceptual model and how their relationships affect and are affected by cross-scale processes. In other words, the analysis presented here shows the importance of understanding relationships between vulnerability processes and patterns at different levels of aggregation. Analyzing interactions among social, physical, and ecological conditions at finer units of analysis allow one to gage the role of social and economic inequalities on levels of vulnerability otherwise not observable at the scale of whole urban areas or municipalities.
Methodological discussions and implications
Using a larger scale such as a city-level analysis and aggregating the data from the sector level reduces the visibility of heterogeneity present within these datasets. On the other hand, though the higher resolution of the sector level data does provide for greater accuracy in terms of assessment by separating the variability present in terms of socio-economic and environmental characteristics, these data sets often have limited base of information (Richards and VanWey 2015). This explains the limitations in terms of choice of scale and variables used for the vulnerability analysis and also explains the subjectivity involved in making this choice. Future work will involve perceived vulnerability using in situ semi-structured interviews to better assess socio-economic sensitivity and adaptive capacity of individual households within the sectors.
Lack of higher resolution topographic data at the sector level limited our flood risk analysis in the urban sectors of the ADE. Other studies used similar methodologies to characterize inundation areas in this region, these include Valeriano and Rosseti (2008) and Sadeck et al. (2012), and in other regions, including Kebede and Nicholls (2011) and Fluet-Chouinard et al. (2015). However, the use of the SRTM dataset to characterize elevation raises the question of data accuracy, as SRTM measurements can be influenced by vegetation in areas that have been substantially altered by human actions, such as built-up urban areas. Nevertheless, despite these limitations, the analysis provides important results as a first attempt to better understand urban flooding along the ADE. For a more robust assessment of flood exposure and risk, future analysis will require local specific environmental monitoring, such as long-term measurements of tides and finer resolution of topographic data.
Finally, this study offers a holistic view of the interaction of multiple factors affecting the vulnerability of the ADE. The results can provide a knowledge base for decision makers and managers to prioritize and develop sustainable management efforts to enhance resilience in the urban ADE (e.g., improvement of sanitation conditions, adaptive capacity). The authors focus on the need for the transition of these management goals to move from a theoretical discourse to practice and implementation for this approach to be truly holistic (Szabo et al. 2015).