Diagnostic of urban flooding and livability in Ghana

Diagnostic Of Urban Flooding and Livability in Ghana 1 SEPTEMBER 22nd, 2025 Organization: ITC-BMS, University of Twente Delivered by: SPACE4ALL Edited by: Florencio Campomanes V & Lorraine Trento Oliveira Funded by: Dutch Research Council (NWO) DIAGNOSTIC OF URBAN FLOODING AND LIVABILITY IN GHANA Report 2025 Diagnostic Of Urban Flooding and Livability in Ghana 2 Executive Summary This report, produced by the Space4All project, explores the intersection on urban flooding and livability in Accra and Tema, Ghana – two rapidly expanding cities facing effects from climate change, unplanned urbanization and inadequate infrastructure. Despite reports and media coverage about the living conditions of the residents, especially those living in deprived settlements, spatial evidence at the city scale is absent. Flood exposure analysis shows that while Accra has a larger total flood-prone area, Tema is more severely affected in relative terms, especially at hazardous depths above 50 cm. Overall, about one-third of the flooded areas in both cities are builtup, with deprived communities disproportionately exposed—59% of exposed deprived areas in Accra and 26% in Tema. At the same time, livability patterns reveal that most at-risk communities are also those with the lowest quality of living. The overlap between flood hotspots and low-livability clusters shows how vulnerabilities compound: poor infrastructure, exposure to flooding, and low livability reinforce each other. The core strength of the project lies in its participatory approach. Residents of deprived urban areas (DUAs) are actively engaged throughout the process with local workshops, transect walk activities and mapping surveys. City actors – government representatives, NGOs and academia were also engaged in multi-stakeholder workshops. Participatory activities confirmed that flood models often underestimate risk in deprived neighborhoods, particularly in Tema, where tidal surges, clogged drains, and artificial flooding exacerbate impacts. Ultimately, the results show that flood exposure is likely higher on the ground, underscoring an urgent need for targeted infrastructure and risk reduction strategies. With citizen science at the heart, our approach advocates for more livable and safer cities for all. Diagnostic Of Urban Flooding and Livability in Ghana 3 Contents Executive Summary ....................................................................................................2 Contents .....................................................................................................................3 1 Introduction ........................................................................................................4 Project Context .......................................................................................................4 Project goals and contributions ...............................................................................6 Understanding risk ..................................................................................................6 Understanding livability ..........................................................................................7 2 Methodology ......................................................................................................8 2.1. Stakeholders Engagement ..........................................................................8 2.2. Local Data Collection ................................................................................ 10 2.3. Flood modelling ........................................................................................ 13 2.4. Livability modelling .................................................................................. 15 2.5. Limitations ............................................................................................... 17 3 Key Findings ...................................................................................................... 18 3.1. Flood Exposure ......................................................................................... 18 3.2. Perceived urban livability ......................................................................... 23 4 Key Learnings .................................................................................................... 25 4.1. Flood Exposure ......................................................................................... 25 4.2. Perceived Urban Livability ........................................................................ 26 5 Recommendations for Planning Agencies .......................................................... 27 6 Outlook ............................................................................................................. 28 7 Acknowledgements ........................................................................................... 29 Diagnostic Of Urban Flooding and Livability in Ghana 4 1 Introduction In August 2024, the Space4All team conducted an intensive data collection and engagement in Ghana, working in close collaboration with local stakeholders - including residents from six deprived communities, local authorities, NGOs and academia - to examine challenges of flooding and livability in Accra and Tema. Both cities face the effects of unplanned urban growth, social disparities and recurring flood events. Additionally, mass eviction has contributed to increasing inequalities, underscoring the need for locally relevant, spatially grounded knowledge to inform inclusive and resilient urban planning. Together with communities and institutional partners, the project produced two key outputs: a high-resolution flood model and a cutting-edge AI tool for livability mapping. The fieldwork created a space for dialogue about climate impacts and underlying vulnerability aspects of the deprived communities, shedding light into possible strategies to respond to their local realities. Project Context Accra and Tema, two coastal Ghanaian cities of the Greater Accra Metropolitan Area (5 million people), face many of the same challenges. Accra, the capital, has approximately 300,000 inhabitants, while Tema hosts almost 200,000 residents (Ghana Statistical Service, 2022). The cities face the wet season from April to October, during which floodwaters regularly overwhelm already strained housing, drainage and waste management systems. Despite improvements from the work of the Greater Accra Resilient and Integrated Development (GARID) Project, these impacts still fall the heaviest on the most deprived settlements, amplifying underlying vulnerabilities. Despite their proximity and shared challenges, important differences remain. The population density of Accra is nearly 3 times higher than Tema, intensifying the pressures in the capital city. Another difference is that Accra benefits from comparatively more accessible urban documentation and government engagement, including clearer communication with the city authorities. In the project, six deprived communities, three in each city, were selected for indepth engagement (Opetekwei, Old Fadama, Accra New Town, Ashaiman-Zenu, Diagnostic Of Urban Flooding and Livability in Ghana 5 Ashaiman-Tulaku and Tema New Town). These areas were identified based on their deprivation patterns and exposure to yearly flood events. Unlike previous work, this project seeks to directly examine links between flood exposure and livability patterns. Although it is widely recognized that deprived communities often suffer the worst impacts, there remains a gap in spatially explicit analysis at both city and community scales. The struggle for social and economic space in the cities is deeply rooted in complex structural inequalities, shaped by political and socio-cultural factors derived from the colonial urban development. On the one hand, residents of deprived settlements defy the planning regulations, settling in hazardous zones, pushed by the lack of affordable housing. On the other hand, local administrative systems lack institutional capacity and tools to ensure equitable infrastructure provision and just land governance, contributing to the displacement of low-income population. This double bind renders many residents as passive victims of flooding, while systematic neglect fosters spatial inequality, in terms of provision, hazard exposure and attention from the state. In this report, we address these gaps by generating new spatial data and emphasizing local knowledge and participation. Our goal is to inform inclusive and equitable planning, particularly of underserved communities most affected by these challenges. Figure 2 - Opetekwei junction, August 27th 2024. Figure 1 – Old Fadama Community, August 27th 2024. Diagnostic Of Urban Flooding and Livability in Ghana 6 Project goals and contributions Based on the current challenges, this project aims: I. To engage local stakeholders in the research process. II. To map the flood hazard extent and exposure in Accra and Tema. III. To map local perceptions of urban livability in Accra and Tema. Understanding risk The concept of risk plays a central role in flood assessments, particularly in the context of climate changes impacts. The Intergovernmental Panel on Climate Change (IPCC) Sixth Assessment Report defines disaster risk as the potential for adverse consequences on lives, livelihoods, ecosystems, and infrastructure due to climate-related hazards. Disaster risk arises from the interplay between three components: hazard, exposure, and vulnerability. Figure 3 - IPCC Disaster risk framework (Sixth Assessment Report). A hazard is defined as the physical event or phenomenon that could potentially cause harm. In the case of urban flooding in Accra and Tema, there are three recognized flood types: riverine flooding from overflow of rivers into settlements, coastal flooding from storm surges and artificial localized floods from poor drainage and waste systems (WRI, 2011). Exposure refers to the presence of people, assets, or ecosystems in areas that are subject to hazards. In cities, formal and unplanned settlements located in flood-prone areas are especially exposed. Vulnerability is the predisposition of exposed elements to be adversely affected when impacted by hazards. Vulnerability is shaped by a range of factors, reflecting socio-economic, infrastructure and governance conditions that may exacerbate risks. Diagnostic Of Urban Flooding and Livability in Ghana 7 Measuring these concepts and analyzing their interaction with the human and environmental factors is not an easy task. It requires a lot of (spatial) information whose acquisition can also be a challenge. In this report, we focus on mapping the hazard (flood prone areas) and exposure (urban built-up structures in flood-prone areas), assessing the spatial patterns of floods and quantifying the exposed areas. Understanding livability Urban livability is a multifaceted concept that describes how well a city supports the well-being and quality of life of its residents. It goes beyond providing basic infrastructure to include factors like safety, environmental quality, social cohesion, cultural amenities, economic opportunities, and participatory governance (Leach et al., 2017). Livability is dynamic and context-specific, meaning its interpretation varies across different cities, communities, and individuals. This inherent subjectivity challenges the effectiveness of typical top-down urban planning models. Therefore, urban planning decisions must consider the lived experiences of diverse communities, including those in deprived urban areas (DUAs). Accurately understanding these perspectives requires robust data and inclusive spatial mapping practices. Without these, the already unseen communities in DUAs continue to be marginalized. This report focuses on mapping livability (the least and the most livable places) across Accra and Tema based on the perspectives of city planners and DUA residents. Diagnostic Of Urban Flooding and Livability in Ghana 8 2 Methodology The Space4All project follows a participatory data-driven approach (Figure 4). In the sub-sections below we will detail each of the dimensions of the project, these include: engagement of local stakeholders; state-of-the-art modelling processes to map urban flood exposure and livability; workshops for data collection and critical dialogue; participatory mapping training and surveys as local evidences to support advocacy of marginalized residents; transect walks to complement the quantitative methods ethnologically; and the dissemination of outputs and documentation back to local stakeholders. 2.1. Stakeholders Engagement Citizen science is the foundation of this project - the involvement of residents in the scientific process - including different types and groups of stakeholders to incorporate ensuring their information needs into the project. This brings a wider range of people into the research, with benefits diversity and inclusion, enhancing the understanding of the lived realities and their existing urban challenges. Figure 4 - The Space4All research methodology diagram. Diagnostic Of Urban Flooding and Livability in Ghana 9 The first task, therefore, was to connect and select a local co-coordinator to support the fieldwork activities, connecting the team to key stakeholders that can benefit from the project and assisting in logistics and cultural aspects. Table 1 summarizes the stakeholders involved in the engagement activities. At the city level, institutional and governmental actors were contacted to participate in local workshops. At the community level, residents from six communities were invited to participate in workshops, training for a mapping survey and the survey itself. Figure 5 illustrates the location of each community involved in the research process. Table 1 - Stakeholders involved in the project. Stakeholder Category Stakeholders Involved Activities Community members Invited residents from six neighborhoods, including Assembly and opinion leaders Capacity training, Communityscale workshops, Participatory mapping Civil Society Institutions People’s Dialogue (SDInet), Smart Nature Freak Youth Foundation City-scale workshop Government Agencies Accra Metropolitan Assembly (AMA), Greater Accra Resilient and Integrated Development (GARID) City-scale workshop Research Institutions University of Ghana City-scale workshop Private organizations Big Data Ghana City-scale workshop Figure 5 - Study areas and selected communities. Diagnostic Of Urban Flooding and Livability in Ghana 16 livability preferences of the DUA residents and another of the city planners. These two models would then amplify the limited data that was collected to be able to conduct city-wide livability mapping. Our AI model, also called AI-voter, had a structure that would take in two radar satellite images, specifically Sentinel-1, of two different areas and the corresponding collected vote as to which area is a better place to live (Figure 10). The AI-voter is trained to find patterns in the two images that make one of them a better place to live. With enough data, the models are then able to simulate selecting which of two areas is a better place to live as if they were either a DUA resident or a city planner. After training the DUA residents’ AI-voter and the city planners’ AI voter, thousands of images from randomly paired areas across the entire area were generated for each AI-voter to “vote” on. These “votes” were then used in a rating tool called TrueSkill which is typically used in online gaming. The idea is that each vote is a “game” in a tournament and as one image is chosen as the better place to live, it signifies a “win” in the tournament and its rating increases while the loser’s rating goes down. After several thousands of pairs (or games played), the final rating of each image is equivalent to its perceived livability. Figure 10. Configuration of the AI-voter. Diagnostic Of Urban Flooding and Livability in Ghana 17 The perceived livability maps from both perspectives were aggregated into hotspots, both positive and negative clusters of livability. These areas would be considered the most livable and least livable from each perspective. Finally, the areas of agreement and disagreement across perspectives of what was considered most livable and least livable were calculated and mapped. 2.5. Limitations The flood exposure mapping methodology has a few limitations: (1) the data collectors using Google My Maps which can only map flood points along riverbanks and streets. It introduces a bias for validation; (2) many areas could not be reached, and therefore could not measured during transect walks, mostly for physical and social barriers; (3) the number of flood observation points collected was not significant for a quantitative analysis, thus we did not include these points in this report; (4) the methodology applied in this project works with an absence of discharge measurements. It is possible that the results would be different if the local discharge behaviors could be analyzed. This would require local data collection campaigns to determine the base discharge conditions. Planning agencies should push integration of the slums into the formal urban fabric, including the residents into participatory mapping approaches and, consequently, in disaster risk policies. The methods to develop the perceived livability results were limited by the following restrictions: (1) our data collection was limited due to time and budgetary constraints. We were only able to collect data from Accra city planners and not Tema as they did not respond to our invitation. We also were only able to go to three DUAs per city. Although our participants were relatively representative of age, gender, and location, there were still other DUAs that could have been included if more time and money had been available; (2) the subjectivity of the “a better place to live” makes it difficult to validate if a participant is making random choices or if that is actually their choice; (3) some participants had technical difficulties or were more indecisive than others leading to a low number of votes which could affect the performance of the trained AI-model leading to potentially inaccurate livability scores. Diagnostic Of Urban Flooding and Livability in Ghana 18 3 Key Findings This section is split into two parts. As the flood and the livability models were conducted with different spatial resolutions, the findings are not aggregated in a single unit of analysis and present separate findings. 3.1. Flood Exposure The results provide a comprehensive overview of flood patterns in Accra and Tema, including estimated flood depth and extent for the event of May 22nd, 2022. Figure 11 shows the spatial distribution of flood exposure of Accra and Tema, overlaid with the deprived (slum) areas. In Accra, higher flood exposure is concentrated along the Odaw River in the western part of the city, where several informal settlements are located. These riverbank areas were originally designated as buffer zones, but the population migrating from within and outside Accra has increasingly settled there in the past decades. Some highly exposed pockets can be seen in the eastern of the city in more deprived areas like Teshie but also in more formal areas like Osu. In Tema, flood exposure hotpots are scattered across the city, but highest exposure levels can be found in the Ashaiman district, right off the Ashaiman Dam. This district has dense population living in areas initially planned for agriculture, lacking infrastructure provision suitable for residential use. In Tema West, the areas around Odum Road and Nungua Road appear especially susceptible to floods, while in Tema Metropolitan, the area between the Sakimono Lagoon and the industrial hubs shows higher flood exposure. Table 4 presents the statistical assessment of flooded areas. Considering both built-up and non-built-up areas, Accra, for its larger size, has a considerably larger flood prone territory when compared to Tema. However, when looking at relative proportions and more life-threatening flood depths (above 50cm), Tema is significantly more exposed. In relative numbers, Accra and Tema face similar relative exposure to floods, respectively 14% and 15% of the territory is prone to floods above 10cm. However, in absolute numbers, this accounts for 32km² of the Accra’s territory and 15km² of the Tema’s territory. Diagnostic Of Urban Flooding and Livability in Ghana 19 Figure 11 - Flood Exposure Maps of Accra and Tema including the delineated deprived urban areas (DUAs). Diagnostic Of Urban Flooding and Livability in Ghana 20 Table 4 – Flood statistics of Accra and Tema (absolute and relative values) at different flood depths. Accra Tema Flood-prone Area (in km²) Floods above 10cm 32km² 15km² Floods above 50cm 16km² 9km² Percentage of flood-prone areas (%) Floods above 10cm 14% 15% Floods above 50cm 8% 9% Percentage of areas exposed to floods above 10cm (%) High exposure 35% 10% Moderate exposure 45% 42% Low exposure 20% 48% When focusing solely on built-up areas, this picture shifts. In Accra, the findings show that from the 32km² of exposed areas, 35% are high density areas and 45% moderate-density areas. In Tema, out of the 15km², 10% is highly dense and 42% is moderately dense. Given Accra’s overall density is about three times higher than Tema’s, the flood risks there are proportionally more severe. At the intra-urban level, the most flood prone districts are Tema Metropolitan, Accra Metropolis, Ablekuma Central and Tema West, all of them with over 20% of their area susceptible to floods above 10cm. However, exposure patterns differ when considering density (Figure 12). For example, Ayawaso Central has only 10% of its territory prone to floods, but 90% of that is high-density, exposing a larger number of structures and people to flood impacts. The presence of dense deprived settlements such as Accra New Town further compounds vulnerability and, thus, higher flood risk. Ayawaso North, Accra Metropolis and Ablekuma North comes in sequence with 63%, 59% and 53% of their flood-susceptible areas as highly exposed. In absolute values, even though Tema districts do not seem highly exposed, in absolute values, the districts of Tema West and Tema Metropolitan have, respectively, 7km² and 6 km² exposed to floods, that is, the highest exposed areas between the two cities (Table 5). In Accra, the district of La Dade-Kotopon also presents a high absolute area exposed to floods, 5.5km², with over 60% moderately built-up in early-stage formal development near the Kpeshie Lagoon. This Diagnostic Of Urban Flooding and Livability in Ghana 21 demonstrates that both planned and unplanned areas are highly exposed to floods and require different mitigation strategies. In comparing flood exposure in deprived and non-deprived areas, significant differences emerge. In Accra, we found that 26% of the non-deprived (formal) areas are densely built-up, against 59% of the deprived areas (Figure 11). This suggests that deprived communities – already more vulnerable – face disproportion high flood exposure. In Tema, even though with smaller relative values, the comparison is dramatic. While only 4% of the exposed non-deprived areas are densely built-up, 26% of the exposed deprived areas are highly built up. This indicates over ¼ of the deprived areas are at high risk. As a secondary city, Tema receives less attention from research and governmental initiatives, which puts the exposed populations in worst conditions. Figure 12 - Percentage of built-up areas exposed to floods per built-up density class in each district. Diagnostic Of Urban Flooding and Livability in Ghana 22 Table 5 - Total exposed area (square meters) at each density class. The total sum of exposed areas (excluding non-builtup areas) per district is provided in square kilometers. District Non Built-up Low-Density Medium-Density High-Density Total (km²) Ablekuma Central 55000 377500 1175000 1480000 3.03 Ablekuma North 2500 245000 2265000 2787500 5.30 Ablekuma West 65000 505000 795000 715000 2.02 Accra Metropolis 10000 185000 325000 760000 1.27 Ayawaso Central 0 0 15000 142500 0.16 Ayawaso East 0 137500 187500 102500 0.43 Ayawaso North 0 107500 292500 670000 1.07 Ayawaso West 435000 1130000 1722500 910000 3.76 Korle Klottey 0 132500 470000 520000 1.12 Krowor 120000 307500 1327500 365000 2.00 La Dade-kotopon 542500 2037500 2695000 597500 5.33 Ledzokuku 90000 470000 1970000 597500 3.04 Okaikwei North 682500 505000 700000 1382500 2.59 Tema Metropolitan 5290000 2525000 2455000 447500 5.43 Tema West Municipal 2950000 3637500 2812500 447500 6.90 Tema-Ashaiman 880000 1000000 1075000 560000 2.64 When we look closely at the six selected deprived areas, the participatory activities conducted on site revealed important differences on the ground. During the workshops, the participants indicated underestimation trends of the model, in coastal areas – where the model cannot capture tidal surges – and spots of artificial floods. For example, residents of Zenu and Tulaku mapped many streets that flood regularly but appear as safe spots in the flood simulations. In contrast, the residents from Accra and the city planners mostly agreed with the simulation outputs. This suggests that the flood model most likely underestimates flood exposure in Tema, especially in deprived neighborhoods, as it cannot capture the complexity of localized flooded areas. During the transect walks in Accra, we witnessed some mitigation measures from both the government and residents (Figure 8). A large-scale dredging work was underway in Old Fadama, which residents reported to occur on a yearly basis to reduce floods in the Korle Lagoon. In internal alleys, residents have been selfconstructing roads and drainage lines, often raising road levels over time, forcing houses to sit progressively lower (thus more vulnerable to floods). In Mpoase, blocked gutters were widespread and stagnant waters persisted long after rainfall, contributing to water borne diseases – often caused by shallow floods. There, Diagnostic Of Urban Flooding and Livability in Ghana 23 residents have tried to protect homes and schools by laying rubber foundations and building stone barriers. These findings reinforce that localized flood drivers – especially drainage infrastructureplays a major role in determining flood impacts and should be prioritized in mitigation plans. 3.2. Perceived urban livability The next section shows the level of perceived urban livability in Accra and Tema based on the perspective of city planners and of DUA residents. For city planners, livability scores were generally high (>0.5) across both Accra and Tema. However, specific areas known to be deprived like Old Fadama and Accra New Town in Accra and Zenu in Tema, had low livability scores. The highest livability scores were found in the central area between the urban centers of Accra and Tema. In the hotspot analysis, Old Fadama and Accra New Town were the largest clusters of low livability in Accra while it was Zenu for Tema. Figure 13. Perceived urban livability score and hotspot maps of Accra and Tema. Diagnostic Of Urban Flooding and Livability in Ghana 24 For DUA residents, livability was perceived more evenly across the territory but still with several areas having low livability scores. For example, the coastal area near Opetekwei in Accra had low livability scores. The surrounding areas of the airport had high livability and like the city planners, Accra New Town also had low livability. The hotspot analysis revealed that the coastal area of Accra and Accra New Town had the largest clusters of low livability in Accra. Meanwhile, parts of Zenu and Tema New Town had clusters of low livability in Tema. The high livability clusters were the areas surrounding the airport. There were both convergences and divergences between the DUA residents’ and city planners’ perceived livability maps (Figure 14). In Accra, both considered livability in the dense urban area of Accra New Town as low (Figure 14-1). The livability in the coastal area of Opetekwei was found to be low only by the DUA residents (Figure 14-2) while the city planners perceived the livability in the dense urban area of Old Fadama along the Odaw river as low (Figure 14-3). In Tema, fewer areas were found to be low in general but there was some agreement between both perspectives that the livability in Community 1 and Tulaku was relatively low. Only the DUA residents considered livability in Tema New Town as low while livability in a newly developed area west of the Sakumono lagoon was also perceived as low. Figure 14. Agreement and disagreement of livability hotspots between planners and DUA residents. Diagnostic Of Urban Flooding and Livability in Ghana 25 4 Key Learnings This report provides a foundation for informed decision-making and strategic planning to mitigate flood exposure and to enhance livability in Accra and Tema. The analysis reveals important insights: 4.1. Flood Exposure 1. Flood exposure is widespread but unevenly distributed. Accra has a larger total area exposed to floods, but Tema shows proportionally higher exposure, especially at more dangerous flood depths exceeding 50cm height. At the same time, a third of the flood exposed areas are highly dense built-up areas in Accra, against only 10% in Tema. 2. Deprived settlements face disproportionate flood impacts. Flood exposure is strongly concentrated in deprived, densely built-up areas. In Accra, almost 60% of flood-exposed deprived areas are high-density, compared to 26% in formal areas. In Tema, the gap is even starker: 26% of deprived areas are highly built-up, while only 4% of non-deprived areas are. 3. Local factors strongly influence flood exposure. Many flood hotspots in Accra and Tema result from local conditions, named artificial flood factors, such as poor drainage, blocked gutters and unplanned development. As natural water flows are impeded by the unplanned urban layout, informal infrastructure and inadequate waste disposal, artificial floods increase the severity of impacts. These factors cannot be well captured by our models and suggest that the true extent of flood exposure on the ground is likely higher than reported here. 4. Participatory processes are essential for local understanding of flood vulnerability. Involving the local communities has proven essential: residents spotted model inaccuracies, critically discussed the possible flood causes, shared adaptations measures. The communities are aware of the problems of building in riparian zones and are interested in contributing to mitigation and prevention measures. However, while these actions are resourceful, they are not substituted for coordinated investment in drainage, sanitation and flood management infrastructure.