Modelling and assessment of sustainable urban drainage systems in dense precarious settlements subject to flash floods

ABSTRACT In developing countries, dense, precarious and informal settlements are common. In the tropics, the occurrence of intense rainfall and the increased impervious surfaces have led to disastrous flash floods. Sustainable urban drainage systems (SUDS) are an alternative to enhance stormwater management and runoff control, providing benefits related to social and environmental domains. This study aims to ground the development of procedures for SUDS implementation to mitigate flooding, and to maximize urban benefits. We present a modelling framework of SUDS implementation, using the Storm Water Management Model (SWMM) to simulate various SUDS scenarios. An application to a small catchment in Brazil considered permeable pavements, bioretention systems, infiltration trenches and rainwater harvesting systems, introduced in both public and private areas. The results demonstrate that SUDS implemented in public spaces, although they may increase access to public facilities, do not completely avoid flooding in the denser parts of the catchment. In such cases, measures must be implemented in private lots due to the reduced amount of available public spaces. Such evidence shows that one of the key elements for flash flood management in consolidated dense and precarious urban settlements is the integration of SUDS measures with the engagement of local inhabitants in decision-making.


Introduction
In the developing world, rapid, increasing and frequently unplanned urbanisation leads to the formation of large impervious surface areas, which, in conjunction with flash floods during extreme precipitation events, can aggravate local flooding and create significant problems downstream due to faster runoff, overwhelming traditional and highly centralised drainage systems (Silveira, 2002). During heavy rainfall events, entire communities are harassed by flooded roads and pathways, making walking and vehicle access difficult for some, while run-off from roads and paths pours into the houses of others, causing problematic circumstances (Jiusto & Kenney, 2015). The conventional grey infrastructures may be inefficient depending on the rainfall intensity, as demonstrated by hydraulic simulations performed for Ho Chi Minh City, Vietnam (Ho et al., 2014).
CONTACT Luma Gabriela Fonseca Alves lumagabriela2010@hotmail.com In informal and precarious settlements, which can be seen as another outcome of unequal and unplanned urbanisation in developing countries, flooding problems are even more remarkable due to socioeconomic vulnerability, extremely dense occupation patterns and a lack of basic infrastructure services, as well as the presence of houses in vulnerable locations, such as floodplains (UN-Habitat, 2003). However, precarious settlements are often neglected and not considered an integral part of a city, excluding their residents from the right to the city and its services (Cities Alliance, 2014). Furthermore, in most cases there are attempts to implement, in the informal and highly dense settlements, successful social and technical approaches for developed areas without considering their particularities, which can lead to disastrous failures. Any intervention in such areas must be adapted thoughtfully to their particular social, environmental and occupation characteristics (Jiusto & Kenney, 2015).
In contrast to the traditional and highly centralised grey infrastructure, sustainable urban drainage systems (SUDS) have emerged as a preferred alternative to enhance stormwater management and runoff control by capturing it at its source, through the processes of evapotranspiration, infiltration and detention, approximating the hydrological cycle in its natural conditions (Fletcher et al., 2015;Sletto et al., 2019). As a result, SUDS measures and techniques can help to enhance the urban landscape, reduce flooding vulnerability throughout the catchment, and aggregate benefits related to social and environmental aspects (Fletcher et al., 2015).
A range of SUDS techniques can be applied in both public and private spaces, such as permeable pavements, bioretention systems, infiltration trenches and rainwater harvesting systems. However, their effectiveness will strongly depend on the characteristics of their locations. According to Silveira (2002), these types of modern urban drainage solutions are applicable in developing countries and consolidated areas only if they explore spaces still available or under-utilised. It is important to note that SUDS will not behave the same in all types of spaces. This is why it is necessary to develop urban planning involving SUDS measures based on the community's and catchment's needs and stakeholders' context (Ariza et al., 2019). Urban planning integrating sustainable drainage systems in the catchment can avoid irremediable and vulnerable development (Ho et al., 2014).
The trend observed in SUDS allocation is implementing them in places with an already adequate urban infrastructure, or in areas not yet consolidated, which facilitates the process due to the availability space and the possibility of integration with urban planning since its conception. Some research has been done to identify potential areas for SUDS implementation in already consolidated urban catchments via an analysis of the physical, urban and socioeconomic variables (Ariza et al., 2019). However, few studies have focused their attention on SUDS implementation in precarious settlements considering their urban heterogeneity and highlighting the difficulties of modelling the impacts of extreme events in these areas (Garcia-Cuerva et al., 2018).
SUDS can motivate residents to engage in environmental protection (Perales-Momparler, 2015). since they are decentralised solutions applied on a local scale and involving several stakeholders. Such characteristics enable the local community to take a leading role in the processes of planning and decision-making, in contrast to the lack of community participation observed in traditional implementation of drainage systems. Jiusto and Kenney (2015) highlight that social participation is much more important to urban drainage planning in informal settlements than in developed areas, because their communities' members are the main active agents of interventions in their urban space.
Modelling flooding in the urban space requires high accuracy and it is necessary to estimate the range of uncertainty of these models (Paquier & Bazin, 2014). However, modelling rainfall-runoff events to identify areas susceptible to flooding and estimate the effectiveness of SUDS interventions in highly dense informal and precarious settlements may be even more challenging. It requires a significant level of detail and careful spatial observation, due to their unique patterns of occupation and land use, markedly different from those of the surrounding areas, with varying sizes of lots. Divergence from the normative legislation and lack of data present a considerable challenge to the process of planning and decision-making. Moreover, the absence of some drainage elements throughout these areas, such as underground pipeline networks and storm sewer conduits, impairs the parameterisation of the model used to simulate urban flooding, as well as other factors that are very difficult to quantify, such as variations of streets' surface roughness and topography, and the presence of moving obstacles (Paquier & Bazin, 2014).
For an integral approach to select and model flooding mitigation measures, this article presents a SUDS modelling framework suited to precarious informal settlements and low-income communities. The approach is applied to a suburban catchment in a Brazilian city. Flood-susceptible areas as well as suitable locations for SUDS measures implementation were identified, based on spatial patterns of urban configuration. The effectiveness of the alternative measures was analysed and evaluated through hydrological and hydraulic modelling simulations.

Case study
The area of study is located in Campina Grande, a city in Northeastern Brazil that has faced rapid urbanisation and a remarkable increase of impervious surface within the last few years. Furthermore, informal and precarious settlements of low-income communities in the study area are affected by flooding problems aggravated by high socioeconomic vulnerability, highly dense occupation patterns and basic infrastructure deficits. The Ramadinha catchment is one such area ( Figure 1). It has about 2.4 km of streams over an urban area of 1.29 km 2 . It is a highly dense area characterised by predominantly residential occupation, with a lack of infrastructure and public facilities. In recent years, infrastructure works were planned in the delimited precarious settlements of the municipality, aiming to eliminate the risks caused by floods and inundations to which the communities were exposed. One of these interventions took place at the Ramadinha catchment, where a concrete canal was constructed along the main catchment stream.
Although the canal construction played an important role in mitigating flood vulnerability at the Ramadinha catchment, flooding issues still persist in some areas. The community still faces significant challenges regarding the lack of proper infrastructure, such as paved roads, appropriate sidewalks, pedestrian spaces, public facilities, and green and open spaces for social interaction. Furthermore, many houses do not present setbacks from the roads (Figure 2), which increases the chances of runoff from roads and paths pouring into the houses. In some areas of the catchment, this problem has led residents to build barriers to protect their houses against the entrance of water, as shown in Figure 3.
Such provocative challenges, along with the fact that urban patterns and configuration are not homogeneous throughout the catchment, were the main factors that made the Ramadinha catchment appropriate for this study. It faces problems that may be overcome by the multiple benefits that decentralised solutions, such as SUDS implementation, bring to the catchment and the community, considering the variety of urban space characteristics and the possibilities for a more participatory social engagement.

Urban configuration and components
The urban space presents different arrangements of urban structure, with varying patterns of vegetated and impervious surface cover, a mixed configuration of building forms, green space configurations, street network patterns, and distribution of public and private open spaces. Such arrangements affect stormwater runoff  generation and propagation, the quality of life of the inhabitants and the appropriate approach to mitigate issues, including the allocation of SUDS measures and techniques. The diversity of the urban configuration and land cover can be analysed at several scales, "from the city scale to a single land unit" (Palme et al., 2020, P. 4).
In the case of the Ramadinha catchment scale, some urban components are of more importance: street network and pavements, sidewalk configurations, public and private parking areas, side roads, public green and open spaces (parks, squares and open spaces inside government institutions, such as public schools and hospitals), empty spaces and vacant lots with potential to be converted into public social interaction spaces, and pervious and open spaces inside private lots (residential and business lots). These components can play an essential role in the approach needed for the implementation of SUDS.
Taking this into account, we identified four urban configuration types throughout the Ramadinha catchment ( Figure 4). They present different urban densities, layouts of open spaces and urban components available to implement SUDS measures. Such heterogeneity makes the analysis of these different spatial configurations even more important to evaluate SUDS suitability and effectiveness in the urban space.
Urban Configuration 1 is an area characterised mainly by regular building configuration, density attenuation due to the presence of public open spaces, a similar configuration to the catchment surroundings and the presence of paved roads and regular sidewalks in almost all its extension. Urban Configuration 2 is the area delimited by the precarious settlement at the catchment. It presents a high building density, absence of open public spaces, and very few areas with the potential to be converted into public green spaces. It is composed mainly of unpaved roads along with the absence of sidewalks, which may increase the vulnerability of households to flooding, as most of the houses are not set back from the streets. Urban Configuration 3 is composed of a regular building form. It presents a high building density in some areas but large unoccupied spaces in others. It includes paved roads and regular sidewalks throughout all its extension. Urban Configuration 4 is characterised by an irregular building form with low density and a significant quantity of empty spaces with the potential to be converted into green areas, along with the presence of unpaved roads and absent or irregular sidewalks, also increasing the vulnerability of households to flooding, as most of the houses are not set back from the streets.

Storm event
Extreme rainfall events in the region are mostly generated by convective storms, with short duration. For this study, we selected one of the heaviest rainfall events recorded in the city in the year 2020, which resulted in 60 mm of precipitation over approximately 1 hour ( Figure 5).

Selection of SUDS measures
Four types of SUDS measures were taken into account to determine the potential areas for implementation: bioretention systems, permeable pavements, infiltration trenches and rainwater harvesting systems. These techniques were chosen considering that they can be more easily implemented in already established areas and housings, and considering the socio-economic profile of the inhabitants of the catchment, since they present a low level of complexity in their maintenance and installation (Woods-Ballard et al., 2015). Furthermore, these SUDS measures can be integrated into areas that could potentially be transformed into social interaction spaces and, thus, appropriated by the community (Woods-Ballard et al., 2015).

Suitable places for SUDS implementation
The implementation of SUDS throughout the Ramadinha catchment was analysed according to the specific characteristics of the urban space (configuration and urban components identified for the catchment) and the physical condition of the area (slope): (i) permeable pavements (PP): at locations with absent and irregular sidewalks, parking lots (including private areas with large paved surfaces for parking areas), and unpaved low traffic roads and areas where the slopes are less than 10% ( (residential, commercial and institutional) with available open space within the catchment. A rainwater harvesting system with tanks with a total capacity of 200 L was assumed for simulation. Rainwater harvesting has the potential to reduce stormwater runoff and increase water supply for indoor usage (Aladenola & Adeboye, 2010) and has been widely used in the city, which faces water supply shortages during the frequent droughts in the region.
Parking areas, unpaved roads, absent and irregular sidewalks, side roads, parks, squares, and empty spaces were identified as georeferenced polygons and lines with the aid of a high-quality orthophoto and registry information system (SEPLAN/PMCG, 2014). For open space analysis inside lots (both public and private), the building area was subtracted from the total lot area, obtained by the buildings' vectorisation (Souza, 2015). The slopes were calculated from a digital elevation model with 2 m resolution accuracy. The urban components used to identify the suitable areas for each SUDS technique, in both public and private spaces, are shown in Table 1.

SUDS management scenarios
SUDS scenarios were composed based on iterative simulation by SWMM to evaluate the reduction of flooding vulnerability throughout the catchment, according to the different levels of suitable urban spaces (public and private spaces) for SUDS implementation. First, permeable pavements were located in public spaces. In cases where this did not lead to the complete mitigation of flooding vulnerability at the catchment, other scenarios comprising infiltration trenches, followed by bioretention and rainwater harvesting systems, were introduced, located as in Figure 6.

SWMM model parameterisation
SWMM is a well-known dynamic rainfall-runoff and hydraulic simulation model developed by the US Environmental Protection Agency, frequently used for simulating urban flooding (Rossman, 2008). The successful use of this model depends on the accuracy and efficiency of its parameterisation (Baiti et al., 2017). SWMM only considers flow within closed or open pipes. Thus, to model Ramadinha catchment, the streets were assumed to be rectangular open canals, and junctions as points scattered along them, since the runoff propagation in the study area occurs over the roads, until reaching the concrete canal, due to the absence of micro drainage infrastructure. The parameterisation of the current land use and occupation, representing the patterns of lots and streets that determine runoff propagation, took into account several physical characteristics of the catchment.
To obtain a greater level of detail of the study area, the catchment was discretised according to the boundaries of each block, which included the streets that contribute to the flow propagation, totalling 199 sub-catchments. For each sub-catchment, area, slope, flow direction, imperviousness and width were determined. The infiltration process throughout the sub-catchments was modelled using the curve number (CN) method. The average terrain slope and flow direction vectors were measured using a digital elevation model with 2 m resolution Parking areas X Open spaces inside residential and commercial lots X X Figure 6. Sustainable urban drainage systems (SUDS) management scenarios.
accuracy. Each sub-catchment slope was calculated as an area-weighted average and its values range from 2% to 24%. The percentage of the impervious area for each subcatchment was defined based on visual analysis of orthophotos, where the impermeable areas, comprising constructed areas and paved roads, were differentiated from the permeable areas, comprising unpaved roads, green and open spaces, and permeable spaces inside private lots. The imperviousness values for each sub-catchment range from 3% to 80%. Width was calculated for each sub-catchment with the expression suggested by Collodel (2009). The model was parameterised to identify flooding whenever water depth exceeds a particular value above street level which makes walking and vehicle access difficult, and also enters the houses. This maximum depth varies according to the street characteristics, and it is shown in Table 2, along with the Manning roughness coefficient for different surface areas and the depth of depression storage.
SWMM can be used to analyse the effect of SUDS techniques at the catchment, which will "capture surface runoff and provide some combination of detention, infiltration, and evapotranspiration to it" (Rossman, 2008, P. 53). Therefore, through alterations in the catchment parameters and using the appropriate SUDS data in the simulation model (Brown et al., 2009;Rossman, 2008;Woods-Ballard et al., 2015) it was possible to evaluate the effectiveness of the selected SUDS techniques. Furthermore, in this article, SUDS measures were placed in the existing sub-catchments working in parallel, each one "treating a different portion of the runoff generated from the fraction of the sub-catchment without SUDS measures" (Rossman, 2008, P. 54). Figure 7 shows the spaces suitable for each SUDS technique, and Figure 8 shows the percentage of SUDS suitability for each urban configuration type (UC) throughout the catchment. For public spaces, bioretention systems were the most suitable SUDS for all urban configuration types except UC 2, of which only 3.5% of its area is suitable for this measure. This UC is characterised by a high building density, with an absence of open public space, and very few unoccupied areas with the potential to be converted into green spaces. In contrast, UC 2 is the most suitable area for implementing permeable pavements, with 15.8% suitability, because it presents a significant quantity of unpaved roads and absent sidewalks. Thus, future interventions aiming to improve urban mobility can consider the use of permeable pavement techniques, which could improve urban infrastructure and enhance flooding mitigation. The low suitability of permeable pavements in UC 3 (only 1.1%) lies in the fact that this area is covered by paved roads and regular sidewalks throughout almost all of its extension, presenting already good infrastructure in the urban mobility domain. For the public SUDS interventions analysed, the least suitable technique, with less than 2% of suitability for all the UCs, is the infiltration trenches. Although this technique did not present relevant urban restrictions, its use was limited by the slope values in the catchment, mainly in UCs 3 and 4, which show slope ranges greater than 5%.

The suitability of the catchment for SUDS implementation
For private spaces, the most suitable SUDS technique for all of the UCs is the rainwater harvesting system, which does not require the availability of vast spaces and, as a consequence, can be integrated even in the denser areas of the catchments, such as UCs 2 and 3. Also, bioretention systems have a great potential for application in private lots throughout the whole catchment, including UC 2, with 25.1% suitability of this technique in private areas, in contrast with only 3.5% for the public spaces. This fact emphasises the importance of considering interventions inside the private lots in the denser parts of the catchment. However, the potential areas for permeable pavements in private spaces are almost zero because of the reduced availability of the urban component analysed to implement it (private parking areas) throughout the catchment.

Flooding vulnerability
For the rainfall event simulated in all six scenarios, flooding hotspots were identified throughout the Ramadinha catchment, impacting the UCs in distinct ways (Figure 9). The results show that a centralised drainage system, represented by the Ramadinha Canal in the baseline scenario, was not capable of eliminating the flooding issues. Also, this system did not take into consideration the improvement of the urban landscape, especially in the areas farther away from the canal, which revealed the necessity to consider a decentralised and effective approach for these issues.
As a result of the great suitability of permeable pavements in public spaces at UC 2, the flooding vulnerability in this area, represented by the total flooding volumes in Figure 10, was reduced by 60.7%, with the application of Scenario 1 only, which shows that this type of intervention can provide urban infrastructure, hydrological and environmental benefits. However, its implementation had a limited effect on flooding in the other parts of the catchment. The infiltration trenches of Scenario 2 produced no additional effectiveness regarding flooding mitigation in the entire catchment, even in the spaces most suitable for it, such as at UC 2. On the other hand, with the integration of bioretention systems in public areas in Scenario 3, greater reduction in flooding vulnerability was achieved at UC 3 (24.3%) and at UC 4 (88.0%). These results show that although these areas do not benefit from high hydrological performance with the application of permeable pavements, their urban landscapes and flooding reduction can be considerably enhanced with the addition of bioretention systems within parks, unoccupied areas and other vacant lots with the potential to be converted into social interaction spaces.  SUDS measures implemented only in public spaces, although they may propitiate better urban conditions for the community, such as paved streets, regular sidewalks and access to public parks and squares, were not sufficient to avoid flooding altogether in the denser areas of the catchment. Reasons for this were the lack of sufficient vacant lots and open spaces in UC 2, and the priority given to traditional pavements in UC 3. For Scenario 4, the implementation of rainwater harvesting systems had a remarkable effect in reducing flooding in UC 2 (52.9%) and UC 3 (85.7%), which can be explained by the high density of private households in these areas. Although larger capacity tanks could be even more effective in reducing the total volume of flooding, they would require more financial resources and difficult maintenance, which might impair the implementation of these systems in all private properties throughout the catchment. Rainwater harvesting is important during extreme precipitation events, especially in precarious settlements located in arid and semiarid regions, combining stormwater attenuation with water supply (Le Jallé et al., 2013). However, the water stored inside the water tanks must be drained before the heavy rainfall events, which can cause conflicts between the needs of different stakeholders (Dinh et al., 2019). The integration of permeable pavements in private areas simulated by Scenario 5 only had an effect in UC 3 (10.0%). However, with the incorporation of bioretention systems in Scenario 6, a considerable reduction in flooding volume was achieved, with only 12.2% of the total flooding remaining in UC 3.

Conclusions
This study demonstrated the importance of selecting, allocating and simulating SUDS measures based on the singular characteristics of each urban space, to maximise their urban and hydrological benefits, especially in the heterogeneous territory of precarious settlements. The scenarios simulated for an intense rainfall event resulted in various configurations of SUDS coverage in the catchment, which led to different results for SUDS suitability and flooding mitigation across the different urban configurations. In the denser parts of the catchment, the reduced possibilities for implementing SUDS in public areas were compensated by their effectiveness when introduced in private households. However, the interventions in private spaces require an appropriate approach towards the stakeholders, which can be achieved by supporting programmes aiming to integrate community members into the process of planning and decision-making, and offering education opportunities regarding flooding issues as well as water conservation and its benefits for the community in which they live. Although this study focused on the quantitative aspects of SUDS implementation and flooding mitigation in the particular study area, further research could be done to evaluate the social, economic and environmental benefits and challenges to integrate SUDS measures in the different suitable areas of precarious settlements.

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
This work was supported by the Brazilian agencies CAPES, CNPq, FINEP, Rede Clima, and the National Institutes of Science and Technology on Climate Change, and on Metropolis Observatory. Camila Silva provided some of the maps used in the study.

Data availability statement
The data that support the findings of this study are available from the corresponding author, LGFA, upon reasonable request.