Geographically induced and the spatially differentiated dimension of flood vulnerability in Greater Kumasi Metropolis, Ghana

ABSTRACT

Over the past several years, the constant flooding events and their lethal consequences have reignited the debate on the need for vulnerability assessment of flood-prone communities in urban areas as a flood risk mitigation and adaptation measure. This article focuses on Ghana and uses remotely sensed high-resolution data, and community mapping to assess the flood vulnerability of 442 urban communities of the Greater Kumasi Metropolitan Area (GKMA). The Compound Topographic Index and the Shuttle Radar Topographic Mission were used systematically. The results showed that 43% of communities were found in low-risk areas, 30% were in medium-risk areas, and 27% were in high-risk areas. Flood intensity and frequencies were found to be dependent on location relative to slopes and undulating terrain. We argue that the identified flood hazard communities should guide policymakers in proffering specific and targeted interventions toward flood risk reduction and community resilience strategies

KEYWORDS:

• Urbanization
• flooding
• geospatial technologies
• spatial planning
• flood risk reduction

Introduction

Persistent flooding events in Ghana have captured the attention of urban scholars and policymakers with diverse theoretical perspectives (Okyere et al. 2013; Darkwah et al. 2018). The focus on flood-related research stems from its detrimental impact on infrastructure, social functionality, and economic development (Ramlal and Baban 2008; Essel 2017b). Discussions surrounding flooding often revolve around identifying causes and implementing mitigation measures (Afriyie et al. 2017). Additionally, studies highlight a direct correlation between increasing urbanisation and flooding events, underscoring the vulnerability of the urban population in flood-prone urban catchments (Feng et al. 2020).

In the context of Ghanaian cities, flooding events result from the inadequate response of city authorities to unpredictable rainfall patterns, waste generation, and sea-level rise, leading to heightened storm surges, lagoon inundation, and coastal wetland submersion. These events, compounded by existing challenges like poverty and livelihood fragility, pose significant obstacles to achieving resilient, safe, and sustainable urban areas. This situation is thwarting efforts aimed at addressing urban resilience challenges (Cobbinah et al. 2017; Darkwah et al. 2018).

Studies on the devastating effects of flooding are generally skewed towards the identification of how meteorological underpinnings have resulted in increased rainfall and consequently, flooding (Okyere et al. 2013; Amoako and Boamah 2015; Attakora-Amaniampong et al. 2016; Ansah et al. 2020) and mostly the number of lives lost, and infrastructural damages (Tabe- Ojong et al. 2020). Intervention policies also tend to be reactive while risk reduction measures are minimally discussed (De Bruijn et al. 2010; Gama et al. 2010; Ahadzie and Proverbs 2011). Meanwhile, since the ratification of the 2005 Hyogo Framework and the 2015 Sendai Frameworks on disaster governance, disaster management has shifted from a reactive and post-management intervention to disaster risk reduction which emphasises disaster preparedness, response, and recovery which forms part of the sustainable development agenda. There is therefore the need for flood researchers and policymakers to develop proactive models (reflect the current paradigm) rather than reactive ones. Consensus has been built in the literature on the indispensability of flood vulnerability and trends assessment of urban areas towards flood prevention and mitigation practices (Scheuer, Haase, and Meyer, 2011; Ouma and Tateishi, 2014; Rimba et al. 2017; Moghadas et al. 2019; Sarkar and Mondal, 2020; Asiedu, 2020).

In July 2015, the European Union (EU) and the African, Caribbean, and Pacific Group of States (ACP) launched a programme called ‘Building Disaster Resilience to Natural Hazards in sub-Saharan African Regions, Countries, and Communities’. The aim was for ACP countries to develop comprehensive frameworks for flood vulnerability assessment and disaster risk reduction (DRR). Since then many researchers have used different conventional geographic information systems (GIS) and remote sensing (RS) techniques to study flood vulnerability assessment (Ramlal and Baban 2008; Mason et al. 2014; Pourali et al. 2014; Essel 2017b; Korah et al. 2017; Higginbottom et al. 2018; Mattivi et al. 2019; Feng et al. 2020). Even with commitment from flood disaster experts to achieving reflectiveness for and support disaster risk reduction (DRR) using GIS and RS, gaps have been identified in the literature on conventional vulnerability assessment. These include the isolated application of digital elevation models without recourse to adequate community mapping and the local-level nature of assessment (see (Membele et al. 2022). Recent events in Ghana indicate that as a result of increased anthropogenic factors such as rapid urbanisation and climate impact, many urban communities are becoming more susceptible to flooding events, but such communities seem excluded from the flood disaster map of the country (Kwoyiga and Owusu-Sekyere 2022). This paper, therefore, seeks to fill this gap by assessing the flood vulnerability of communities in the Greater Kumasi Metropolitan Area, a flood-prone space in Ghana. As our contribution to the literature, the article advances the discourse on flood risk and vulnerability mapping from the local level to a contiguous metropolitan scale (2247.37 sq/km). The rest of the article is divided into sections: Section One captures the introduction followed by Section Two which captures the literature review and the historical perspective of flooding in Ghana. Section three focuses on the materials and methods used in data collection and analysis while section four presents the results and discussion. Finally, the concluding section captures policy implications and summarises the findings of the study.

Research synthesis

Shifting the emphasis on disaster management

Disaster hazards are commonly classified based on factors such as size, likelihood, and frequency. Flood disasters, as characterised by an extreme temporary inundation of land by water, often exceed the capacity of affected communities to protect lives and livelihoods. This necessitates external assistance in managing the resulting losses (Ayariga 2014; Duy et al. 2019). Globally, flood-related disasters are acknowledged as the most significant natural events, having claimed more lives than any other disasters (Oteng-Ababio 2013). Since the 1990s, there has been a shift in disaster management strategies from a focus on post-disaster activities to pre-disaster activities emphasising disaster risk reduction (UNISDR 2011).

This strategic shift requires all countries to formulate appropriate policies and legislation to minimise the lethal impact of flood disasters on lives and livelihoods. According to the Food and Agriculture Organization (FAO 2008), Disaster Risk Reduction (DRR) is a framework with the potential to minimise vulnerabilities and disaster risks throughout society within the context of sustainable development. Disaster management should thus incorporate DRR measures, including prevention, mitigation, and preparedness.

The importance of disaster risk reduction was a prominent topic during the World Conference on Disaster Reduction in Japan, leading to the establishment of the Hyogo Framework for Action (HFA) (Holgren and Moe 2012). The framework recognised the significance of proactive incorporation of DRR in disaster management strategies, alongside reactive emergency relief efforts. In 2010, the United Nations International Strategy for Disaster Reduction (UNISDR) introduced the ‘Resilient Cities’ campaign, emphasising the necessity for countries to adopt a proactive stance towards disasters (UNISDR 2011). Addressing the risk from climatic extremes, the Intergovernmental Panel on Climate Change (IPCC) has called for a combination of incremental and transformational adjustments (IPPC 2011). Given the increasing frequency, unpredictability, and severity of disasters, these actions have become imperative, elevating risk management to a top priority. Despite the ongoing challenges in systematically integrating DRR into development planning and operations, these interventions increasingly recognise DRR as a key component in achieving sustainable development.

Following the launch of the Sustainable Development Goals (SDGs) framework, the UN launched the Sendai Framework for Disaster Risk Reduction (SFDRR) in 2015 to guide the targets of the SDGs. The Sendai framework is built on the pillars of the Hyogo Framework for Action and articulates the need for governments to add DRR principles to all dimensions of the disaster governance structure. The framework also encourages governments globally, to intensify efforts at promoting sustainable development and create safer and more resilient communities. This is because the SFDRR recognises that the preparedness of state actors through its regionalised institutions and local communities is of great importance for the effective management of ecological disasters such as floods. Involving local communities in disaster management efforts to identify and assess their susceptibilities and capacities as well as to create and carry out action plans is an efficient strategy for reducing flood disasters (Gu et al. 2021). This is important as it demonstrates the significance of community participation in the disaster management continuum.

Historical perspectives of flooding in Ghana

Since historical times, urban Ghana has been impacted by numerous flood disasters of different scales and magnitudes (see Kwoyiga and Owusu-Sekyere 2022). On many occasions, the affected communities struggle to deal with their impact due to the inability of government agencies and their development partners to address the people’s extensive vulnerabilities (Karley 2009). Table 1 presents some of the significant flood disasters in Ghana over the years.

Table 1. Incidence of flooding in Ghana.

Flooding Year	No. of People affected
1968	25, 000
1983	12, 500,000
1991	2, 000,000
1995	700,000
1999	324,602
2001	144,025
2007	332,600
2008	58,000
2009	139,790
2010	33,602
2011	81,473
2015	46,370

Source: Authors construct from multiple sources (Ahadzie and Proverbs 2011; Asumadu-Sarkodie et al. 2015; Tengan and Aigbavboa 2016; Ansah et al. 2020).

The flood disasters have largely been attributed to natural conditions such as rainfall and topography, the effect of climate change (Youssef et al. 2015; Erena and Worku 2018), and anthropogenic factors such as modified hydrological characteristics to land use change as well as improper flood control systems (Ali et al. 2011; Alexakis et al. 2014; Shanableh et al. 2018; Sholihah et al. 2021). The anthropogenic is the offshoot of increased urbanisation and changes in land cover patterns. These factors have triggered the loss of urban green spaces to the built environment and have in turn increased run-offs exacerbated by increased impermeable surfaces from pavements and compacted soils. Again, issues of poor planning and development control that have led to the erection of unauthorised structures and blocked waterways have also been cited as responsible for flood disasters in urban Ghana. Songsore (2017) also indicated that flood disasters in Ghana are due to inadequate waste collection and disposal systems in urban areas where solid waste is disposed of in drainage systems and bushes. These flood disasters have had a devastating impact on infrastructure and livelihood support systems (Amoako and Inkoom 2018).

Response to flood disaster management

In order to manage disasters and reintegrate disaster victims into normal life, the Ghanaian government formed the National Disaster Management Organization (Act 517) in 1996 and this was revised in 2016 and replaced with the National Disaster Management Organization Act 927 (Adelekan 2010; Abu and Codjoe 2018). According to the Act, NADMO’s primary responsibilities include creating disaster management plans at all levels, from national to the district, for the prevention and/or mitigation of disaster effects, providing adequate services to ensure technical training institutions for educational activities to raise public awareness, warning systems, and preparation for its staff and the public, as well as organising local and international assistance for relief services and reconstruction (Oteng-Ababio 2013). Despite the challenges faced by the NADMO such as limited funding and lack of institutional collaboration, the organisation has established early warning and forecasting systems, one of its primary responsibilities is to educate communities to take the proper steps to mitigate flood losses (Owusu-Sekyere et al. 2017). The organisations have also been carrying out public sensitisation programmes on the structural and non-structural measures that local authorities can adopt to reduce the impact of flood disasters. Figure 1 provides examples of the structural and non-structural measures that local authorities are using to reduce the effects of flood disasters (Yeo 2000).

Figure 1. Strategies to reduce flood damage.

Source: Yeo (2000)

The non-structural approaches are described in the top right quadrant, while the structural measures are highlighted in the top left quadrant. While non-structural methods manage the floodplains to lessen the effects of floods, structural measures are used to prevent extreme events like floods. In recent studies, the economic vulnerabilities with recourse to the capacities of households to avoid and mitigate floods as well as the financial inadequacies of flood management organisations are known (Adedeji et al. 2012; Møller-Jensen et al. 2022). Also, many studies including (Duy Can et al. 2013; Mwale et al. 2015) have looked at the environmental vulnerabilities and the use of the livelihood vulnerability index to estimate flood vulnerability (Salazar-Briones et al. 2020), water-related vulnerability index (Balica et al. 2015). Santo et al. 2020 and others to estimate economic and social vulnerabilities on smaller scales. This paper focuses on the spatially differentiated dimension of vulnerability.

Methods

Study setting and data

The GKMA covers the Kumasi metropolitan area and its adjoining seven municipal and district assemblies including Bosomtwe, Ejisu, Atwima Kwanwoma, Kwabre, Afigya Kwabre, and Atwima Nwabiagya which collectively covers approximately 2247.37 square kilometres [See Figure 2] (Oduro et al. 2014). The study area is found in relatively the southern part of the Ashanti region and experiences an annual rainfall range of 1300–1500 mm per annum with a mean temperature of between 24°C with the highest temperature being 30°C (Ministry of Environment, Science, Technology and Innovation 2013). The study area has a population of about 3,345,623 which represents about two-thirds of the Ashanti region with an annual growth rate stood at 4.62% which is about double the regional figure (GSS, 2010). The GKMA has seen unprecedented and unplanned spatial expansion over the past decades both outwardly and internally.

Figure 2. Map of greater Kumasi metropolitan area. Source Author’s construct based on data extracted from the land use planning and management system (2020).

Correspondingly, the built-up area has also expanded from about 98 square kilometres in 1972 to about 478 square kilometres in 2012 (Oduro et al. 2014). The area is also characterised by urban sprawl, waste disposal challenges, and housing facilities that mimic informal settlements (Benneh et al. 1993; Owusu-Sekyere et al. 2021). The area is also drained by some of the seven main rivers in the Ashanti Region such as Offin, Oda, Aboabo, and Wiwi among others which flow over impervious rocks. The high population growth without corresponding increase in development; the high rainfall without proper drainage channels and the uncontrolled urban growth with poor spatial planning have all worked together to increase human vulnerability to incessant flooding events. The story of GKMA is not complete without acknowledging the numerous flood events that characterise the area every year. Without digging deep into history, the flood events in 2012 killed seven people and rendered over flood 391 homeless. Similarly, in 2017, another flood event killed over 38 and rendered over 3236 people homeless (NADMO 2017). In all these cases, the after-flooding discourse has been that there should be a proper scientific model that could help institute risk-reduction intervention spatially in vulnerable areas (Šoltésová 2017; Lebre 2021; Moulds et al. 2021)

Data types and sources

Based on the objectives of the study, several data sources were employed with the Digital Elevation Model (DEM) being paramount. Data types and sources are indicated in Table 2. The use of a courser resolution DEM is influenced by the high cost of higher resolution datasets and less availability of public and open-source data with topographical characteristics (Essel 2017a), especially in the global south. The progression in use of DEM in topographic assessment has evolved over the years. For instance, in more recent works, the use of Synthetic Aperture Radar and Moderate Resolution Imaging Spectrometer (MODIS) to indicate time series of floods and detection of flood inundations have gained much prominence (Vichet et al. 2019).

Table 2. Data sources and types.

Analyses	Data	Resolution	Format	File type	Data source
Topographic analyses	NASA Shuttle Radar Topographic Mission (SRTM), Digital Elevation Model (DEM)	30 M	Raster	GeoTiff	USGS Earth Resources Observation and Science (EROS) Center Archives
Area of interest (AOI)	Greater Kumasi Metropolitan Area (GKMA) District Boundaries		Vector	Shapefiles (Polygons)	Land Use Planning and Management Information System
Communities	Greater Kumasi Metropolitan Area (GKMA) District Settlements		Vector	Shapefiles (Points)	Land Use Planning and Management Information System. Community Mapping

However, work on the comparison of courser resolutions DEMs (such as from SRTM) and finer resolution DEMs (such as from LiDARs) by (Saksena and Merwade 2015), revealed that the use of courser resolutions for flood mapping with a better spatial resolution of 30m (Essel 2017a) produces similar results since it produces significant improvement of surface elevations. On account of the aforementioned, the use of STRM DEM was considered for this study.

Application of compound topographic index (CTI) in flood mapping

The CTI, also known as the Topographic Wetness Index (TWI) is a complex and widespread index in the field of flood mapping, modelling, assessment, and prediction of flood due to its use for the estimation of where there will be cell accumulation of water taking to consideration, the elevation differences (Mattivi et al. 2019). As demonstrated in Figure 3, mass balance is the primary concept of CTI. Thus, the description of the tendency of the draining contour length and local slope implicit in the SCA (the tendency of water reception – parameter) to evacuate water and produces an index scaled by the natural flow routing logarithm (Gruber and Peckham 2009) which establishes the direction of the flow for each cell. Explanatory, low slope angle (large contributing drainage areas) areas that are deemed as water accumulating prone areas are likened to high CTI values. Likewise, steep slopes (well-drained areas) are associated with low CTI values.

Figure 3. CTI scheme. Source: adapted from (Gruber and Peckham 2009).

Many writers have employed tenets of CTI and have been deemed as an effective and cost-efficient approach to the identification of flood-prone areas compared to other conventional hydrodynamic models (Gruber and Peckham 2009; Pourali et al. 2014; Higginbottom et al. 2018). However, limitations to the use of CTI are with its scale-dependent computation and it does not apply to areas with flat terrains. For this work, the systematic procedure for CTI employed is shown in Figure 4. The SRTM with entity IDs SRTM1N06W002V3 and SRTM1N07W002V3 were used. The two raster files covering different extents were first mosaicked using the mosaic raster algorithm and subsetted to the boundaries (in.shp) of the GKMA also using the raster processing (clip) tool in ArcGIS. Two spatial analytical methods were employed, thus, hydrological and surface (Refer to Figure 4), as inputs to the realisation of CTI. Under the hydrology, fill was calculated to avert sink levels of the SRTM which was followed by flow direction. The result was an input for the generation of flow accumulation. Finally, the flow accumulation output was scaled to the resolution of the SRTM. On the surface side, the slope of GKMA was generated using the SRTM. An application of a constant (PI) aided in the generation of the radiant slope of the area which was later enhanced by the removal of undefined values. The CTI was generated using the formula:

Figure 4. Compound topographic index workflow. Source: authors construct.

$\mathit{CTI}=\mathit{IN}\frac{\mathit{SFA}}{\mathit{ERS}}$

Where SFA represents Scaled Flow Accumulation.

ERS represents Enhanced Radiant Slope

Enhanced framework for assessing geographically induced vulnerabilities of urban settlements

The conventional CTI was augmented with literature documentation and enhanced community mapping as well as a spatial query in the GIS environment. Through academic database searches, various studies carried out on a variety of issues on flooding in GKMA were identified as part of the criteria for the identification of flood-intense areas and the purpose of categorisation in the study area. Key concepts/words such as ‘Flood’, ‘flood inundation in Kumasi’, and ‘Flood mapping in GKMA’ were used to gather research works and publications in repositories (Google Scholar, JSTOR, Scopus, Web of Science), local journals news portals (both electronic and print media). As a result, 13 study areas were identified as communities studied concerning floods in GKMA. Spatial distribution and patterns identified, and the intensity of the study (multiple mentions) were instrumental and contributed to the categorisation of flood zones for this study. Observation, as a primary data collection method, was used as an additional effort to understand the spatial differences in the intensity of topographic-induced flooding. As Abass et al. (2020) corroborated, the use of these methods aids in the triangulation of data collected to ensure more reliable and credible results. Visits, as a ground-truthing technique, were made to the communities identified from the spatial analyses and those derived from the literature search. Observation and field mapping, therefore, afforded researchers to ascertain the spatial variances of flood in the study areas and the categorisation into zones. Zones were further determined using the geostatistical analyst with natural breaks (Jenks) algorithm. Summarily, this study proposes a framework for the mapping of flood zones as a great input in understanding flooding in contemporary times. The author suggests the integration of aspatial data into conventional flood mapping techniques such as CTI. Resultantly, this enhances flood mapping and identification of topographic-induced flood risk areas in urban and Peri-urban settings (Figure 5).

Figure 5. A proposed enhanced framework for flood mapping.

Results and discussions

Flood risk assessment of GKMA

The determination of various flood zones was based on multiple factor considerations with the CTI being the primary basis. The major components included slope, flow accumulation, and flow direction. Each output was generated using the hydrology toolset, classified (into descriptive levels), and symbolised using natural breaks (Jenks) in ArcGIS. The results from the spatial analyses (see Figure 6a,b) especially the DEM (A) present an undulating terrain for GKMA with about two-sevenths of the entire landmass highly susceptible to flooding and concentrated at the southern part. The slope model shows the areas with low and high gradients.

Figure 6a. Hydrological characteristics of GKMA. Source: authors construct based on data extracted from USGS earth resources observation and science (EROS) center archives (2020).

As shown in model F and the enhanced slope radiant (undefined values removed) in G, approximately 24% of the area (representing 539 sq km) was found to have steep slopes from the northern part to areas around the lake. This notwithstanding, such terrain was found to be not concentrated in a particular area but across the entire terrain. Indicatively, such areas will have increased runoff due to flow intensity resulting from high slopes area-wide. Flow accumulation (D in a scaled form, E) was estimated using the flow direction (C) with the accumulated weight of all cells flowing into each downslope cell in the output raster as the proxy (Zhou et al. 2019). As indicated, 19% of the landmass of GKMA representing 382 sqKm was found to have high flow accumulation values which makes it easier to form run-offs. These corroborated findings by the Ministry of Environment, Science, Technology and Innovation (2013) on the overall terrain of the region. Among other geographical and topographical characteristics, the study identified the terrain to be made up of flat-topped inter-fluvial ridges running north-southwards direction with widths ranging from 1500m to 2500m. The terrain also influences the flow of major rivers in the study area such that they flow north-southwards including the flow of surface runoffs.

Additionally, the irregular terrain noticed is attributable to the location of the study area. Amoateng et al. (2018) and the Ministry of Environment, Science, Technology and Innovation (2013) indicate that it is dominated by dissected landforms since it is located within a forest-dissected plateau physiographic region and hence disposed to advanced tertiary erosion. Lastly, the result of the CTI is presented in model H (Figure 6b) and symbolised with a red-to-blue colour spectrum. In the extreme ends, red indicates low CTI values (less wetness) and blue is indicative of higher CTI values (high wetness). It was revealed that GKMA is generally characterised by a moderate level of wetness making the entire area susceptible to flash flooding. This is attributable to the nature of the terrain, drainage pattern, and the interspersed nature of low-lying areas influencing high tendencies of flow accumulation and inundation (Saksena and Merwade 2015).

Figure 6b. Hydrological characteristics of GKMA. Source: authors construct based on data extracted from USGS earth resources observation and science (EROS) center archives (2020).

Terrain variability of GKMA

Figure 7 reveals the terrain variability of the GKMA. It was shown that the entire area has an undulating terrain. Altitudes range from as low as 73 metres to 460 metres.

Figure 7. Terrain variability of the GKMA. Source: authors construct based on data extracted from USGS earth resources observation and science (EROS) center archives (2020).

A careful look at the map disclosed that areas with higher altitudes were found to be primarily on the Northern and the North-Western sides. Districts like Afigya Kwabre, Kwabre, and some portions of Kumasi Metropolis (KMA) and Ejisu Municipality fall within high-altitude areas. Medium-range areas were found to be in the southwestern part (second quadrant) of the region. Districts in these ranges were found to primarily be Ejisu, Bosomtwe, and a portion of KMA and Atwima Kwanwoma. The third quadrant (South-Eastern) was found to be in the low range. The quadrant covers portions of districts such as KMA, Atwima Nwabiagya, and Atwima Kwanwoma. There is an interplay of low and high ranges at the North-Eastern part of GKMA. The fourth quadrant consists of some parts of the Afigya Kwabre and Atwima Nwabiagya districts.

Upon careful examination of the above, combined with the analysis of profile graphs from Google Earth Pro and spatial analysis parameters such as clustering and outlier analysis (Pourali et al. 2014), the terrain variability of the Greater Kumasi Metropolitan Area (GKMA) can be categorised into four zones. These zones serve as the basis for delineating flood risk assessment zones, namely flood zones, high-risk zones, low-risk zones, and elevated areas with a comparatively lower risk of flooding. As illustrated in Table 3, approximately 17% of the GKMA land falls within the flood zone, while areas at a higher risk of flooding constitute around 33%. Low-risk areas make up about 40%, and elevated areas account for approximately 10%. The deduction is that around 90% of the GKMA is susceptible to risk; however, the spatial distribution of these districts (communities) influences the frequency and intensity of flooding events.

Table 3. Flood risk assessment (GKMA).

Area of Interest (AOI)	Area (SqKm)	Categories	Land area (sq KM)	Percentage (%)
Greater Kumasi Metropolitan Area (GKMA)	2247.37	Flood Zone	379.41	16.88
		High risk	742.89	33.06
		Low risk	897.38	39.93
		Higher areas	227.69	10.13

Explanatorily, Flood zones were deduced on 9.75 km distance denoted areas with elevations gain/loss of 102m, −108m. The maximum slope was found to be 8.7% with an average slope of 1.9%. The next zone identified was the high-risk zone. The profile graph revealed that such zones had elevations up to 254m. The graph distance of about 7.42 km indicated an elevation gain/loss of 128m, −147m. The zone had a maximum slope of 10.4% and an average slope of 3.6%. The low-risk areas were identified to have elevations ranging from 254m to 303m based on a distance of 15.7km. The zone had an average elevation of 274m with elevation gain/loss being 245m, −233m. The maximum slope was found to be 8.6% with an average slope of 3.1%. High areas signified areas with 304m, 329m, and 358m being the minimum, average and maximum elevations. An 8.18km distance on the profile graph indicated an elevation gain/loss of 111m, −128m. The average slope was found to be 3.1% with a maximum slope of 10.3%. The varying nature of the zones influenced by topography agrees with conventional literature that establishes a connection between low areas with DEMs to the high susceptibility of flooding and flood hazards (Essel 2017b).

Topography plays a key role in flood outcomes in the study region. Insights gleaned from JICA (2013), Owusu-Ansah (2015), and especially the outcomes of interviews conducted by Amoateng et al. (2018) suggest that the region’s undulating topography generally shields it from flooding hazards in a topographical context and underscoring the beneficial role of the undulating terrain in mitigating flood disasters within the city. According to feedback from institutional representatives and certain property owners in the same study, the elevated nature of the land prevents runoff from flowing upward to higher grounds, mitigating the risk of floods. Instead, runoff generated on higher grounds and other lowland areas is slowed down and accumulated in valleys, thereby diminishing the likelihood of flooding. In line with this, Owusu-Ansah (2015), employing cross-sectional analysis, observed a near-complete absence of flood incidents in areas characterised by the steepest slopes in the city. This is therefore attributable to less incidence witnessed in Afigya Kwabre and Kwabre districts over the years.

Spatial distribution of settlements per flood risk zones

Figure 8 shows a spatial distribution of settlement in the GKMA. However, their number, name, and location are referenced to data available in the Land Use Planning and Management Information System of Ghana. As indicated, the selection of settlement per zone was done using the select by location Standard Query Language (SQL)) in ArcMap. Additionally, names of settlements were coordinated using urban town flooding in GKMA as captured in literature. Although this may not capture all settlements, it captures major towns since there is less spatial data on small towns and rural areas in Ghana (Korah et al. 2017).

Figure 8. Spatial distribution of zones and settlements in GKMA. Source: authors construct based on data extracted from USGS earth resources observation and science (EROS) center archives (2020).

As identified and grouped in Table 4 (District and community-based), there is less variability in terms of flooding for the various districts. However, its variability is influenced by the location and geography of the area. All districts but Kwabre and Afigya Kwabre had communities in the flood zone. About 23 communities (34%) of the Bosomtwe district were found to be in the flood zone. This was followed by 28% (25) of the communities of Atwima Nwabiagya District then Atwima Kwanwoma which had 13 (20%) of its communities in the flood zone. Other districts: Mampong, KMA, and Ejisu had fewer communities in the flood zone with 18%, 6%, and 2% respectively. Apart from the Kwabre district, all other districts in the GKMA had communities in the high-risk zone. However, a greater proportion (45%) of communities in the Atwima Kwanwoma district were found in this flood zone. Likewise, 42%, representing 38 communities of the Atwima Nwabiagya were found in the high-risk zone. The Kumasi Metropolis had 37% of its communities in this zone while Bosomtwe had 27. Fewer percentages: 18, 14, and 6 were recorded for Asokore Mampong Municipality, Ejisu Municipality, and Afigya Kwabre District respectively.

Table 4. Communities and categories.

Districts	Categories/Communities
	Flood Zone	High Risk	Low Risk	Higher Lands
Atwima Kwanwoma	Anensogya, Winiso Eduyaokrom, Akrampa, Aboabo Sekyekurom, Esipan, Bedase, Kurofoforo, Ampabame, Nkroransa, Konkori	Domakwae, Afrantwo, Bebu, Brofoyeduru, Ahenema Kokoben, Kotei, Trede Kwanwoma, Kanfinchiri, Ampabame, Odumase No. 1, Nkwanta No. 1, Nyamebekyere, Taabuom, Yaokwachikrom, New Ahenema, Kokoben, Heman, Afrentia, Adumase, Deikurom, Duakoben, Bekwame, Kyekyewere, Duopam, Dentekurom, Paa	Camp, Twedee, Yabi (Yabea), Krumuase, Aburaso, Akosomo, Apemanim, Takyiman, Mmerewagya, Aduwomasi, Mpatasee, Daku, Gyakye, Kyekyebon, Foase, Amoafo, Winsa, Sebin Akrofrom, Kokoben, Dida, Ampetua, Akyeremade, Aponsakwa
Afigya Kwabre		Pampatia, Nsutam, Denasi	Tetrem, Ntirbuoho, Afrantwo Krabobuoho, Buonkabin, Nkukua Buoho, Bomso, Sasa, Besease, Agyaakobooho, Brofoyeduru, Ampabame, Atrama, Esen, Ahodwo, Patase, Heman, Duaponko, Brukon, Edwenase, Atimatim, Nkwanta Penten, Akuroforam, Amoako, Maase, Adwumakasekese, Soko, Adukro, Abroma, Akom, Kwaman, Esaase, Aduman, Ahenkro, Nkwanta-Kese No. 2, Swedru No. 2, Kodie, Mmowere	Kyekyewere, Heman, Abrade, Aboabogya, Adwiratia, Mpobi, Dabaa, Ankaase
Atwima Nwabiagya	Aboabo, Akrofuom, Apuayem, Bedabuokrom Besease, Brosanko, Duasi, Kobi, Nkwantaya, Ntabanu, Ohiasumawu, Sampruso, Tananse, Anyamesemkurom, Bankyease, Ntabeemukura, Akwaboa, Sokwae, Ntensere, Kodwobiowuokurom, Sepaase, Dakowaakurom, Akuapem, Adankwame, Bari, Mfensi, Owabi	Asakraka, Gyankoba, Kontomrekrom, Nkuran, Seidi, Toase, Yawkrakurom, Boafo, Oforikurom No. 2, Amanokurom, Tweapiase No. 1, Boaso, Kumi, Mabai, Kyekyewere, Afigya, Asakraka, Adagya, Aninkroma, Pasoro, Esaase, Asenemaso, Abuakwa, Manhyia, Mmakuro, Kokoben, Dabaa, Asaman, Kapro, Asumenya, Fufuo, Akropon, Asuofua, Koforidua, Hwidiem No. 2, Agogo	Ntensere, Tailorkurom, Rubykurom, Woradon Bungalow, Sapongkurom, Ohiayeanimguase, Atuntuma, Adubofuokurom, Afrekurom, Kwabenankrumakurom, Asikamu, Nkasasu No. 2, Nkasasu No. 1, Kokoben, Nkatia, Amoaman, Worapon, Bokwankye, Esaso, Akyease, Barekese, Anwoma Ofin, Amisare, Tonto Kokoben, Abira, Adankwame
Asokore Mampong	Aboabo No1, Aboabo No2	Owusu Ansa, Asabi	Buabai, Mesuom, Pakuso, Aperede, Sepetimpom, Nyame Yehene, Mampong
Kwabre			Brofoyeduro, Abirem, Gyenyasi, Boyiasi, Adense, Akyease, Asennua, Asumenya, Bamparasi, Manhyia, Adwuman, Abira, Kenyasi No.2	Mmedoma, Saaman, Sisirase, Aketebommo, Safo, mape, Adanwomase, Kasaem, Ahwaa, Fawoade, Bosore, Nwomasi, Dumanafo, Adeesena, Krobo, Gyamfi Wono, Sakorawono, New Nsoromaso, Asonomaso, Antoa, Adwumakasekese, Aboaso, Ntonso, Mamponteng
Bosomtwe	Nyinatiase, Adunku, Behenase, Adwuampon Abaase, Anyinatiase, Aborodwam, Adwafo, Nkawi, Obo, Saalem, Atafram, Brodekwano 1, Dedesua, Adwaden, Adwumam, Apinkra, Konkoma, Abonu, Pipie 1, Amakom, Jwaman, Brodekwano 2	Kokobiriko, Nkwainkwanta, Onwe 2, Esereso, Adagya, Oyoko, Nnuaso, Pramso, Soaduro, Worakese, Akokofe, Adwinase, Mmem, Pipie 2, Onwe 1, Amanase, Beposo, Nkwanta	Ayuom, Aputuogya, Adwafo Naw Kurom, Patiensa, Toamfom New Kurom, Youth Camp, Patiensa, Perabon, Sawia (Sawua), Aboaso, Feyiase, Abuontem, Atobiase, Tetrefu, Akwaduo, Amankwaa Adei, Adwafo Town, Yaase, Deduako Kokodee, Asisiriwa, Hampanasi, Nyameani, Pease, Kuntanase
Kumasi Metropolis	Atonsu 2, Sakoban, Atonsu 1, Kaase	Atafua, Apere, Konkromase, Dabanpanin, Sakoben. Anyinam, Wamase, Ahensan, Dakodwo, Apirabo,Emena, Ohwim, Apatropa, Tanoso, Takyiman, Behinase, Adeemmra, Atasomanso, Aprabon Deduako, Asago, Asukwa, Amanforom	Kronom, Mpatasee, Abrepokese, Old Soldier Settlement, Ehwemase, Edwenase, Denkyemmoso, Kwadaso, Adeemmra, Kyirapatere, Gyenyasi, Boadi, Dompoase, Duom, Bebiri, Daban, Nyankye Remasi, Kuronum, Anomannye, Abrepo, Ampabame, Amanfrom, Nsema, Ahodwe Fankyenetra, Boko, Buokurom No.1, Duase, Sipe, Kotei, Ayeduasi, Anwomaso, Domeabra, Kokoben, Bohen, Agogo No.2, Hwediem, Kwadaso, Breman, Nhiaso, Buokurom. Asuoyeboa, Santase, Tafo, Kumasi	Pankrono
Ejisu Municipality	Hwereso	Domase, Okorase, Boankra, New Ampabame, Kwaso, Boankra No.1	Ekyem, Kokobra, Asaman, Asopon, Donuaso, Akyease, Tikurom, Baworo, Akokoamon, Kerekurom, Gyakye, Doduako, Sape, Odaho, Asawasi, Ampabame, Adadeentem, Krapa, Abankurom, Manhyia, Krapa No.2, Krapa No.1, Edwenase, Abanase, Apromase, Akyenakurom, Baman, Fumisua, Kwamo, Akyawkurom, Onwe, Asotwe, Donyina, Esienimpon, Besease, Ejisu	Bonwire

All districts had communities in the low-risk zone. More specifically, Ejisu Municipality had 36 of its communities (81%) in this zone. This is followed by 42 communities (79%) in the Afigya Kwabre district and Mampong being 64%. Kumasi Metropolis had 36 communities (57%) in the low-risk zone. This is followed by Bosomtwe, Atwima Kwanwoma, and Kwabre with 39%, 35%, and 34% respectively. Concerning areas on higher lands with extremely low risk of flooding, few communities were found in this zone. For instance, four out of the eight districts making up the GKMA had no settlement in this zone. Explicitly, Kwabre District had a greater share of its communities (65%) in the high areas. Additionally, Afigya Kwabre, Kumasi Metropolis, and Ejisu Municipality had 15%, 2%, and 2% respectively. In sum, 91.9% of (399 out of the 434) major towns in the GKMA are susceptible to flooding. More so, the intensity and frequency of flooding are dependent on the location. From Figure 9, though all districts in GKMA are at risk of flood, there exists differentiation in flood susceptibility among districts under GKMA. For instance, about 180 sq km (48%) of the total land area of Atwima Nwabiagya District is highly susceptible to flooding. Likewise, 136 sq km representing 36% of the land area of Bosomtwe District is decidedly susceptible to flooding. On the contrary, other districts that have a higher proportion of their land area are in high areas with relatively low susceptibility to flooding. An example is Afigya Kwabre District with 123 sq km (54%) in areas that have low susceptibility to flooding incidence. This notwithstanding, as observed by (Lwasa 2010), increased levels of anthropogenic factors (such as blocking drainages) increase susceptibility to flooding. On aggregate (Table 4) the majority (49%) of settlements in GKMA were found to be in the low-risk zone. Those in high-risk areas constituted 27% with high lands having a share of approximately 8%. Significantly, approximately 16% of communities in GKMA are in flood zones.

Figure 9. Settlement categories relative to flood risk in GKMA.

Source: Authors Construct

Conclusion and recommendations

The research addressed the call for innovative techniques to identify geographically vulnerable areas prone to flooding in the Greater Kumasi Metropolitan Area (GKMA) of Ghana. Utilising the Compound Topographic Index and other non-spatial data and methods, the study uncovered the primary communities in the GKMA at risk of flooding. The results indicated that the combined number of communities in high-risk and medium-risk zones exceeded those in low-risk zones. The severity and frequency of flooding varied based on location, with the study attributing persistent flooding in the communities to factors such as high rainfall, inadequate spatial planning, and topographic conditions.

While acknowledging that flood events, like most disasters, cannot be entirely prevented, our study supplied essential information for disaster managers in the area for effective disaster risk reduction. The methodology highlighted specific communities requiring focused flood management interventions and identified reasons behind the recurring flooding incidents. The findings underscored the importance of avoiding one-size-fits-all policies, recognising that what works in one location may not be applicable elsewhere.

Given indications of ongoing population growth, uncontrolled urban expansion, and consistently high annual rainfall in the study area, the research suggests a holistic approach to reducing flood-related disasters, with Geographic Information Systems (GIS) playing a key role. GIS, capable of producing hazard and vulnerability maps, emerges as a valuable tool for pre- and post-hazard (flood) intervention and mitigation strategies. This recommendation stems from the understanding that GIS can contribute significantly to comprehensive flood risk management strategies.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Notes on contributors

Anthony Kwabena Sarfo

Anthony Kwabena Sarfo is an expert in Planning and geospatial technologies. He is presently a Technical Advisor (GIS and Remote Sensing) in GIZ-Ghana on the Resilience Against Climate Change Project (EU-REACH). His interest is in Planning, Climate Impact, Climate Adaptation measures and Geospatial Technologies.

Ebenezer Owusu-Sekyere

Ebenezer Owusu-Sekyere is an Associate Professor of Geography at the University for Development Studies, Ghana. He is an accomplished academic with diverse experience and expertise in Geography, Research, Higher Education Management and Curriculum Development. His area of research includes Urban Environmental Management, Disaster Risk Management and Climate Change.

Alfred Toku

Alfred Toku holds an M.Phil. in Development Studies and a BSc. in Integrated Development Studies with a specialization in Development Studies. He is a lecturer in the Department of Planning and Land Administration, University for Development Studies, Tamale, Ghana. He is an urban development planner with comprehensive knowledge of urban and peri-urban studies. His research interests include urban planning, sustainable development, and environmental planning. He has published extensively on regional planning, sanitation, spatial modelling, agriculture, climate change, land acquisition, local livelihood challenges, and adaptation

Nelson Nyabanyi N-Yanbini

Nelson Nyabanyi N-yanbini is a lecturer at the Department of Urban Design and Infrastructure Studies, SD Dombo University of Business and Integrated Development Studies, Wa. He holds an MPhil in Planning and a Bachelor of Science in Human Settlement Planning, both from the Kwame Nkrumah University of Science and Technology (KNUST). Nelson previously worked as a Project Coordinator in Family Support Lifeline (NGO) and as a Teaching and Research Assistant at the Department of Planning, KNUST. His areas of research interests are climate change adaptation, urban/spatial planning, urban sustainability, eco-tourism and land governance.