A
Title: Flood hazard mapping in data scarce basins: A Rain-on-Grid approach in Kumasi, GhanaVida Akyeampong 1,2, Godfred Abbey Torsah 4 Mohammad Najim Nasimi 3, Kwaku Amaning Adjei 1,2, Charles Gyamfi 1,2, Markus Disse 3, and Leonard Kofitse Amekudzi 4
1 Department of Civil Engineering, Kwame Nkrumah University of Science and Technology,
Kumasi
2 Regional Water and Environmental Sanitation Centre Kumasi (RWESCK), Kwame Nkrumah
University of Science and Technology, Kumasi
3 School of Engineering and Design, Technical University of Munich, Munich, Germany
4 Department of Meteorology and Climate Science, Kwame Nkrumah University of Science and
Technology, Kumasi
* Corresponding author: vidaakye@gmail.com, + 233 502558900
Acknowledgement
The authors acknowledge the partners of the project Current and future risks of urban and rural flooding in West Africa (FURIFLOOD project) and its funder, the BMBF.
Abstract
Flood inundation mapping is essential for assessing and managing urban flood risks. However, in many low- and middle-income countries, hydrological data scarcity limits the ability to generate reliable hazard assessments. This study developed a transferable workflow for flood hazard mapping in ungauged urban basins, demonstrated in the Aboabo Basin of Kumasi, Ghana. Hydraulic simulations were performed using the Rain-on-Grid method within HEC-RAS 2D, which was calibrated using post-flood field surveys. Synthetic design storms corresponding to 5-, 50-, and 100-year return periods were used to generate the hazard maps. The results indicated maximum flood depths of 3.5, 3.9, and 5.3 m, with inundation extents of 12.7%, 16.9%, and 23.1% of the basin area, respectively. These findings show how rapid urbanization amplifies inundation risk and demonstrate the scalability of flood impacts with return periods. Beyond the case study, the workflow highlights the potential of participatory calibration and rainfall-driven hydrodynamic modeling to overcome data scarcity in ungauged basins with limited data. This approach offers practical guidance for strengthening urban flood risk management in rapidly developing regions worldwide.
Keywords:
Urban flooding
HEC RAS 2D
Flood hazard mapping
Data scarce basin
1. Introduction
Flooding remains one of the most destructive natural hazards worldwide, with urban areas in low- and middle-income countries (LMICs) facing particularly high risks owing to rapid urbanization, unplanned development, and limited adaptive capacity (McDermott, 2022; Echendu, 2022; Snikitha et al. 2025). In 2024, a total of 7.5 million people were affected by flooding across West and Central African countries with at least 1527 people losing their lives (UNOCHA, 2025).
Effective flood risk management depends heavily on accurate flood hazard mapping, which informs planning, infrastructure design, and emergency preparedness (Mudashiru et al., 2021; Almoradie et al., 2020; Pathak et al. 2024). However, in many LMIC settings, the scarcity of hydrological data, particularly streamflow and long-term rainfall records, poses significant challenges for conventional hydrological modeling approaches (van Emmerik et al., 2015; Trinh & Molkenthin, 2021). This limitation is especially acute in rapidly growing cities in sub-Saharan Africa, where urban expansion often outpaces the establishment of monitoring networks (Amoateng et al., 2018; Campion and Venzke, 2013).
Accurate flood hazard mapping is central to disaster risk reduction (El baida et al., 2024); however, many urban basins in sub-Saharan Africa remain ungauged, constraining the application of conventional hydrological models that require long-term discharge or rainfall records (Hrachowitz et al., 2013; Akpoti et al., 2024). These knowledge gaps limit preparedness and urban planning in regions where exposure to flood hazards is increasing most rapidly (Amoateng et al., 2018; Campion & Venzke, 2013).
Recent advances in hydrodynamic modeling have enabled the use of rainfall-driven simulations and high-resolution digital elevation models (DEMs) to generate flood hazard maps in ungauged basins. Among these, the Rain-on-Grid approach within HEC-RAS 2D has emerged as a promising method, allowing the direct simulation of rainfall-runoff and flow routing across a catchment without the need for gauged discharge records (David & Schmalz, 2020; Ongdas et al., 2020; Sarchani et al., 2020). When complemented with field-based data, such models offer a feasible pathway for producing reliable hazard maps under severe data constraints (Wüthrich et al., 2024; Desalegn & Mulu 2021). However, their application in ungauged, data-scarce environments remains limited, and questions regarding model calibration, reliability, and practical transferability persist (Trinh & Molkenthin, 2021; Kreibich et al., 2025).
This study addresses these challenges by combining Rain-on-Grid hydraulic simulations with post-flood field survey data in Kumasi, Ghana, to develop flood hazard maps for multiple return periods. The objectives of this study are (i) to demonstrate how participatory field data can enhance calibration in the absence of gauges, and (ii) to draw generic methodological lessons on hazard mapping that extend beyond Kumasi to other rapidly urbanizing basins worldwide.2. Methodology
2.1 Study Area
The Aboabo Basin, approximately 154 km², is located in the city of Kumasi. The population of Kumasi is 3,353,850 (GSS, 2021), and the average monthly temperature fluctuates between 24.6°C and 27.8°C, while the mean annual rainfall measures 1450 mm (Osei et al., 2019). The climate in this region experiences two distinct rainy seasons: a major rainy season spanning from March to July and a minor rainy season occurring from early September to early November (Domfeh et al., 2016). The basin’s elevation varies between 245 and 392 m mean sea level. The highest elevation was found in the northern part of the basin, whereas the lowest was found in the southern part (Fig. 1). The Aboabo River originates from Tafo Pankrono in the northern part of the Basin. It flows through Buokrom, Anloga Junction, and Moshie Zongo and is joined by the Sisa River at a confluence at Asokwa (Danquah et al., 2011). The basin (Fig. 1) is also drained by several tributaries, namely Subin, Daban, and Wiwi, and discharges into the Oda River at Sokoban.
Fig. 1
Study area showing some communities located the basin
2.2 Data sources
2.2.1 Precipitation and hydrological Data
Daily precipitation data from the Kumasi Airport Station were obtained from the Ghana Meteorological Agency (GMet). There was no discharge recording station located in or close to the basin, therefore, no observed discharge was used in the flood analysis.
2.2.1.1 Extreme rainfall events for different return periods
To generate extreme rainfall events for simulation, intensity–duration frequency curves for Kumasi were obtained from the Ghana Meteorological Agency to analyze the rainfall intensities of the various return periods in the study. A calculated synthetic design storm hyetograph was incorporated into the basin modelling using the rain-on-grid modelling approach (Borst, 2023). The design storm for the analysis in the study area was derived from a rainfall hyetograph. The alternating block method is recommended as a tool for developing synthetic hyetographs in block format that indicate incremental precipitation values during storms. The maximum precipitation value is placed at the center, and other values or blocks are arranged around the highest value in descending order (Martins et al., 2023).
2.2.2 Land use and Manning’s Data
Satellite images from Landsat 9 (Table 1) were classified using QGIS software to generate a land use/land cover map of the basin. Four classes of land use/ cover; forest, built-up, low vegetation and bare land were identified. The roughness of the various land use types in the basin was represented using Manning’s coefficients from the HEC RAS reference manual. Training samples were manually selected for analysis using the Semi-Automatic Classification plugin (SCP) of QGIS. Accuracy assessment was performed on the images using randomly generated points, and ground truthing points were chosen from Google Earth.
Satellite | Sensor ID | Path/Row | Ground Resolution | Acquisition Date |
|---|---|---|---|---|
Landsat 9 | OLI/TIRS | 194/055 | 30m | 21–12 – 2021 |
2.2.3 Digital Model Elevation Data
A high-resolution digital elevation model is essential for accurately representing the hydrological features of a basin. The performance of the 2D hydraulic model is dependent on the resolution of the DEM used in the analysis. Finer resolutions indicate detailed flood dynamics and are good for estimation of flood inundation depths (The DEM used in this study was 12.5m resolution data from the Alaska Satellite Facility (ASF) website (https://vertex.daac.asf.alaska.edu). The DEM was processed in a GIS environment to identify the hydrological features of the basin.
2.2.4 Terrain modification
One of the main issues with hydraulic modelling is that the actual topography beneath the water surface and in the channel region is frequently absent from topographical data (Brunner, 2016). During the preparation of the terrain data, it was observed that the river channels were not represented in the DEM. This could be because the width of the river is smaller than the resolution of the DEM (Fatdillah et al., 2022). Cross-sections of the river network were obtained from the Parades project, which was conducted in the basin the previous year. Terrain modification of the river channel was performed using the surveyed cross-section to establish the flow routes for the simulation. The modified DEM was then used for the simulation.
2.2.5 Post-flood survey
Gathering information on flood events in developing countries is challenging, especially in ungauged urban centers (Martins et al., 2023). There are no available measurements for the study area; therefore, other sources of information must be identified.
In September and October 2022, field work was carried out by the study to verify the link between the ground data and the simulated event in the model using flooding events that had occurred in the basin, focusing on the current event on the 28th September 28, 2022. On-site data, such as water marks, structural damage, and flood depth, were assessed. The residents were asked to indicate the depth of the flood during the event, which was then measured using a tape measure. The residents of the affected communities were interviewed in a semi-structured format.
2.3 Hydraulic simulation model
The open-source software, River Analysis System from the Hydrologic Engineering Center (HEC-RAS version 6.4) developed by the United States Army Corps of Engineers, was used to set up the 2D model.
HEC-RAS creates a mesh, and each cell in the in-depth hydraulic property table contains information such as the relationship between the elevation and volume/area (Ongdas et al., 2020). In the computation of flow over the cells, two hydraulic equations are applied: the shallow water equation and the diffusion wave equation (Eq. 1 and Eq. 2), which are all derived from the Saint-Venant equations (Sarchani et al., 2020). However, the Shallow Water Equation was employed to generate flood hazard maps. The Shallow Water Equation for unsteady flow is given as (Brunner, 2016)
where H is the surface elevation of the water, u and v are the velocities in the x and y directions, f is the Coriolis parameter, g is the gravitational acceleration, vt is the horizontal eddy viscosity coefficient and cf is the bottom friction coefficient. The 2D Shallow Equation used in this study disregards the eddy viscosity and Coriolis effect.
2.3.1 2D computational mesh development, boundary conditions and infiltration layer
A 2D flow area polygon was created to represent the computational mesh that fell within the boundary of the underlying elevation model. The size of the mesh cells influences the time step applied in the simulation. The runtime of the simulation increased with the increase in the number of computational cells. Therefore, to avoid runtime errors, the HEC RAS 2D manual advises modelers to utilize fewer than a million cells in the simulation analysis. A 2D computational mesh with 12.5m wide cells was generated for the study, with a total of 754000 cells for analysis. The choice of cell size was based on its capacity to accurately capture all the features of the 12.5m resolution DEM (Rangari et al., 2019). The river network in the basin was digitized along with the bank lines, with the faces of the cells aligned with the direction of flow. Break lines were added to the model to enhance the output of the model as it has been observed to play a key role in flood propagation (Ongdas et al., 2020). Two boundary conditions were selected for the model: input precipitation (P) simulated over the entire 2D flow area (David and Schmalz, 2020) and normal depth. The infiltration layer for the model was created at the intersection of the land cover and soil layers. The SCS Curve Number method was selected for analysis, in which different curve numbers of the land cover classes were used. The curve number for the different land cover classes were forest (73), built-up (83), low vegetation (79), and bare land (91).
A
Fig. 2
Flow chart of methodology
2.4 Calibration of Model
Calibration of a model involves the definition of optimal standard parameters, which will lead to the smallest change in among the observed and simulated values (Desalegn and Mulu, 2021). A crucial step in ensuring the validity and applicability of research findings is the validation of the generated flood inundation areas.
Comparing the flood depths from the simulation with the observed flood depths from the survey allowed for the validation of the model (Martins et al., 2023). In this study, calibration was performed by varying the Manning’s roughness coefficients of the model (Brunner, 2016), whereas validation was carried out using global positioning system (GPS) locations of flood-prone areas and information from elderly inhabitants during a field interview. This approach was adopted because the flow in the basin was not measured. This investigation of the ground truthing survey provided a reliable way to determine the accuracy of the model in forecasting flood dynamics. Any discrepancies found throughout the validation process were carefully evaluated to fine-tune the model parameters and lower the likelihood of prediction errors.
3. Results
3.1 Land Use Land Cover
The study revealed that the built-up area constituted the largest land cover class (Figs. 3 and 4), accounting for 88% of the total area. This was followed by low vegetation (9%), forest (2%), and bareland (1%). The built-up areas in the basin mostly comprised residential facilities, industries, and corporate buildings. Therefore, indicated the imperviousness of the basin with a lower percentage of pervious areas. An overall accuracy of 90% was achieved. The participatory assessment of flood-related disaster prevention and development of an adapted coping system in Ghana (PARADeS) project is a collaboration between the University of Bonn (U-BN), University of Freiburg (U-FR), University of Applied Sciences in Magdeburg (HS-M), and Flood Competence Center (HKC) of Germany, and the West African Science Service Centre on Climate Change and Adapted Land Use (WASCAL), Water Resources Commission (WRC), and National Disaster Management Organization (NADMO) of Ghana. The land use of the basin, as studied by the Parades project (Fig. 5), presents seven categories: high income, low income, informal settlements, industry, park, commercial, and middle income. It is observed that a high percentage of residents in the upstream basin are middle-income, whereas most low-income accommodations are located in the middle of the basin. This is due to the concentration of business activities in this part of the basin.
Fig. 3
Land cover map (2021) indicating the different land cover classes
Fig. 4
Percentage of land cover (2021)
Fig. 5
Landuse map (2021) revealing the different uses of the land
Source: Parades project
3.2 Case Study
On September 28, 2022, a significant flood event occurred in the basin owing to heavy rainfall during the late hours of the day. The following day, a field visit (Fig. 6) was conducted to assess the impact of the flood as reported by the media. During this visit, watermarks from the flood were identified, along with remnants of properties that had been destroyed (Fig. 7).
Gaining insight into the effects of flooding through direct observation and community input is crucial for developing effective flood management and mitigation strategies. By integrating scientific measurements with local knowledge, understanding flood dynamics and strengthening resilience in vulnerable areas can be deepened.
Interviews were conducted with the affected residents to gather first-hand information on the flood event. One crucial aspect explored during these interviews was the level of water that inundated their properties. To convey this information effectively, the interviewees used relatable references—the human body. Specifically, they described floodwater depth in terms of body parts: feet, ankles, knees, waist, and shoulders. This approach allowed for a qualitative assessment of the flood depths observed during the event.
Fig. 6
a. Measurement of flood depth b. Flood mark showing on the wall
Fig. 7
a. Flooding on Duncanson Road – KNUST b. Level of flood water marked on the structure during the event
3.3 Hydrodynamic model simulation (Case study)
A past flood event that occurred on the 28th September 28, 2022, was simulated in the model. The HEC RAS 2D model was initially simulated using the diffusive wave equation for the stability of the model and finally simulated with the shallow water equation for maximum results. The total inundated area in the September 28, 2022, event was 34.47 km2, and the peak discharge was 67.2 m3/s, with the maximum depth simulated being 5.5 m (Fig. 8).
Fig. 8
Flood map for case study (28th September, 2022)
3.4 Calibration of model
The water levels observed during the field survey were compared with the levels of the simulated events in HEC-RAS. The GPS coordinates of the observed depth locations were overlaid on the simulated flood map to record the depth of the simulated flood. A visualization of the flood depths at the different checkpoints indicated a correlation between the observed and simulated events (Fig. 9). A Pearson correlation analysis (Table 2) conducted on the values showed a moderate correlation between the observed and simulated events. The Root Mean Square Error (RMSE), which indicates the measure of accuracy of the predictions, was 0.44m, which showed that the model accuracy was moderate.
Fig. 9a
Field Survey locations
Fig. 9b
Field survey check points against simulated event (28th September, 2022) in HEC RAS
No. | Statistic | Value |
|---|---|---|
1. | Pearson correlation | 0.62 |
2. | RMSE | 0.44 |
3.5 Generation of synthetic hyetographs for various return periods
Synthetic rainfall hyetographs were generated for return periods of 5, 10, 50, and 100 years using the intensity–duration–frequency (IDF) curves for Kumasi obtained from the Ghana Meteorological Agency (Fig. 10). The rainfall intensity at a certain return period for each duration was obtained from the IDF curve. The alternate block method is used when increments, or blocks, are recorded into a time sequence with the maximum intensity occurring at the center of the required duration and the blocks are arranged in descending order alternately to the right and left of the central block to form the storm plots (Duka et al., 2018). The alternate block method was employed in this study because it creates a realistic distribution of rainfall intensities over a given period of time. The simulation was conducted for a duration of 3 h, which is the time frame typically associated with maximum storm events in Ghana. The maximum depths recorded for the respective return periods were as follows: 19.61 mm (5-year), 22.16 mm (10-year), 27.51 mm (50-year), and 29.78 mm (100-year).
Fig. 10
Temporal distribution of rainfall using the alternating block method
3.6 Flood hazard mapping for different return periods
The model employed rainfall data extracted from the synthetic rainfall hyetograph that was developed using the rain-on-grid approach within HEC RAS. The results obtained for flood depths and extents were obtained by simulating various return periods (Fig. 12). The 5-year return period produced a maximum flood depth of 3.5 m, the 50-year period showed a maximum flood depth of 3.9 m, and the 100-year period exhibited the highest maximum flood depth of 5.3 m. In the basin area was inundated with water, and 5-year return period, 12.7%, 16.9%, and 23.1% of the area by 50-year return period and 100-year return period precipitation (Table 3).
Fig. 11
Aboabo basin showing major transportation routes and some key facilities
Fig. 12a
5-year return period flood event
Fig. 12b
50-year return period flood event
Fig. 12c
100-year return period flood event
Fig. 13
5-year return period simulation with some flood prone areas
Fig. 14
100-year return period simulation with some flood prone areas
The maps (Fig. 12) show areas prone to flooding during 5-year, 50-year, and 100-year return period events. Several flood-prone areas have been identified along the Accra-Kumasi Highway, including the eastern and southern bypasses, with particular emphasis on hotspot communities such as Anloga Junction, Susanso, KNUST, Airport Roundabout, and Atonsu. Residents along these routes experience varying water depths, depending on the rainfall return period. The Anloga Junction was identified as a critical hotspot because of the overtopping of floodwaters during events, given its role as a major transportation route in the basin. This route is vital for accessing the Komfo Anokye Teaching Hospital, Kumasi Airport, and other key facilities (Fig. 11).
Additionally, communities such as Aboabo, Asawase, Asokore Mampong, Moshie Zongo, and Sawaba are identified as hotspots affected by these events as these areas are inundated even with the lowest return period (Fig. 13). In the 100-year return period event (Fig. 14), severe flooding of tributaries and rivers was observed. The identified locations were recognized as high-risk zones during a rainfall of a 100-year return period (Fig. 13).
The direct impact of flooding on key services in the basin is very limited; however, routes to access these services during an event are greatly compromised. This is because the primary assets exposed during a flood are culverts, bridges, and footbridges located in the communities.
Return Period (years) | Flooded Area (km2) | Flooded area in percent | Peak Discharge (m3/s) |
|---|---|---|---|
5 | 19.7 | 12.7% | 22.0 |
50 | 26.1 | 16.9% | 43.9 |
100 | 35.5 | 23.1% | 60.2 |
4. Discussions
4.1 Land use distribution and flood response
The land-use map of the study area indicates that built-up areas dominate the catchment, interspersed with smaller patches of vegetation and riparian zones. This spatial pattern directly influences hydrological responses. Built-up surfaces, characterized by high imperviousness, are associated with elevated Curve Numbers (CN), leading to rapid runoff generation and shallow infiltration. Vegetated and riparian zones, though limited in extent, provide localized areas of higher infiltration and temporary storage, moderating flows at the micro-scale.
Flood simulations demonstrated that the inundation depth and extent were concentrated in areas where built-up land cover coincided with low-lying terrain and insufficient drainage. Conversely, zones with vegetative cover, even when located in flood-prone areas, exhibit slightly reduced runoff potential due to higher infiltration capacity. This highlights the importance of land-use configuration, not just the overall proportion of urban areas, in shaping the spatial variability of flood hazards.
Even in ungauged basins where only a single land-use snapshot is available, flood hazard assessment can reveal how different land cover classes modulate hydrological responses. The spatial interplay between impervious cover and topography is a transferable diagnostic for other rapidly urbanizing catchments worldwide, underscoring the need to integrate land-use zoning and green infrastructure into flood management strategies.
4.2 Rain-on-Grid simulation under data scarcity
Using Rain-on-Grid in HEC-RAS 2D allowed us to directly simulate runoff and inundation without gauged discharge data. The calibration step, based on field surveys of past flood extents, demonstrated that community-based participatory data could effectively substitute for traditional gauge records. Although this introduces uncertainties, it provides a viable low-cost pathway for modeling ungauged basins where formal data networks are absent.
Participatory calibration enhances the credibility of hazard models in ungauged catchments and offers a replicable approach for other LMIC cities facing similar data gaps.
4.3 Flood hazard dynamics across return periods
Flood maps for 5-, 50-, and 100-year return periods showed that hazard depth and extent scale rises with increase in rainfall intensity. Depths reached up to 5.3 m in the 100-year scenario, with inundation expanding from 12.7% to 23.1% of basin area. The spatial pattern of flooding revealed hotspots along riparian zones and poorly drained urbanized areas.
The methodology highlights how flood hazard increases with return period, a feature that is transferable to other rapidly urbanizing basins. This indicates the importance of integrating return-period-based hazard maps into zoning and disaster preparedness, particularly where rapid land cover change accelerates risk.
4.4 Limitations and uncertainties
Some limitations encountered in this study are; First, the 30 m DEM resolution may not capture fine-scale drainage features critical in dense urban environments. Second, land-cover classification errors, particularly in peri-urban mosaics, introduce uncertainties into CN estimation. Third, the CN method’s assumptions regarding initial abstraction and uniform rainfall distribution simplify complex hydrological processes. Finally, participatory calibration relies on recall and localized observations, which may miss broader spatial patterns.
Despite these limitations, explicit acknowledgment and sensitivity testing (varying hydrologic soil group assumptions, CN abstraction ratios) help increase the transferability of findings and transparency for application in other basins.
4.5 Policy and practical implications
The results have direct implications for urban flood management. Hazard maps identifying high-risk zones can support:
Land-use planning thus restricting settlement expansion into high-risk floodplains.
Infrastructure design by guiding stormwater drainage upgrades and prioritizing detention basins in rapidly urbanizing sub-catchments.
Disaster risk reduction through informing contingency planning, early-warning dissemination, and climate resilience initiatives.
These insights align with the Sendai Framework for Disaster Risk Reduction and broader international agendas to strengthen urban resilience under climate change.
Flood hazard mapping in ungauged basins, even under severe data constraints, can produce actionable products for policy and planning, contributing to global hazard and risk management agendas
5. Conclusion
This study applied a rainfall-driven hydrodynamic modeling approach using HEC-RAS 2D Rain-on-Grid, calibrated with participatory post-flood surveys, to map flood hazards in the ungauged Aboabo Basin of Kumasi, Ghana. Results showed substantial increases in inundation extent and depth across 5-, 50-, and 100-year return periods, indicating the escalating flood risk associated with urbanization and climate extremes.
Beyond the local findings, three broader lessons emerge. First, participatory field surveys provide a practical, low-cost alternative for model calibration in data-scarce basins. Second, Rain-on-Grid simulations are effective for hazard assessment in ungauged urban catchments, with sensitivity analyses highlighting the need to account for soil and abstraction assumptions. Third, flood hazard maps generated through this workflow offer direct value for policy by informing zoning, infrastructure siting, and disaster preparedness in rapidly urbanizing regions.
By articulating a replicable methodology under data constraints, this study contributes to advancing flood hazard science and risk management practice in LMICs. The approach is transferable to other ungauged basins globally, aligning with international efforts to improve prediction in ungauged basins and enhance urban resilience to natural hazards.
A
References
A
Abass K (2023) Battling with urban floods: Household experience, coping and adaptation strategies in Ghana. Cities, 140. https://doi.org/10.1016/j.cities.2023.104430A
Ahadzie DK, Mensah H, Simpeh E (2022) Impact of floods, recovery, and repairs of residential structures in Ghana: insights from homeowners. GeoJournal 87(4):3133–3148. https://doi.org/10.1007/s10708-021-10425-2Akpoti K, Mekonnen K, Leh M, Owusu A, Dembélé M, Tinonetsana P, Velpuri NM (2024) State of continental discharge estimation and modelling: challenges and opportunities for Africa. Hydrol Sci J 69(15):2124–2152
Almoradie A, de Brito MM, Evers M, Bossa A, Lumor M, Norman C, Yacouba Y, Hounkpe J (2020) Current flood risk management practices in Ghana: Gaps and opportunities for improving resilience. J Flood Risk Manag 13(4). https://doi.org/10.1111/jfr3.12664
Amoateng P, Finlayson CM, Howard J, Wilson B (2018) A multi-faceted analysis of annual flood incidences in Kumasi, Ghana. Int J Disaster Risk Reduct 27:105–117. https://doi.org/10.1016/j.ijdrr.2017.09.044
A
Arseni M, Rosu A, Calmuc M, Calmuc VA, Iticescu C, Georgescu LP (2020) Development of flood risk and hazard maps for the lower course of the Siret River, Romania. Sustain (Switzerland) 12(16). https://doi.org/10.3390/su12166588Borst R (2023) Evaluating the Effect of Nature-Based Solutions on Urban Runoff using Hydraulic Modelling: A Case Study in Sekondi-Takoradi. Dissertation, University of Twente
Brunner GW (2016) HEC-RAS River Analysis System Hydraulic Reference Manual. www.hec.usace.army.mil
Campion BB, Venzke JF (2013) Rainfall variability, floods and adaptations of the urban poor to flooding in Kumasi, Ghana. Nat Hazards 65(3):1895–1911. https://doi.org/10.1007/s11069-012-0452-6
A
Ceppi A, Gambini E, Ravazzani G, Lombardi G, Cerri L, Meucci S, Mancini M (2025) Attribution of Flood Forecasting Errors From a Multi-Model Perspective in Milan Urbanized River Basins. J Flood Risk Manag 18(1). https://doi.org/10.1111/jfr3.70023Danquah L, Abass K, Nikoi AA (2011) Anthropogenic Pollution of Inland Waters: the Case of the Aboabo River in Kumasi, Ghana. J Sustainable Dev 4(6). https://doi.org/10.5539/jsd.v4n6p103
David A, Schmalz B (2020) Flood hazard analysis in small catchments: Comparison of hydrological and hydrodynamic approaches by the use of direct rainfall. J Flood Risk Manag 13(4). https://doi.org/10.1111/jfr3.12639
Desalegn H, Mulu A (2021) Mapping flood inundation areas using GIS and HEC-RAS model at Fetam River, Upper Abbay Basin, Ethiopia. Scientific African, 12. https://doi.org/10.1016/j.sciaf.2021.e00834
Domfeh MK, Anyemedu FOK, Anornu GK, Adjei KA, Odai SN (2016) Assessment of the water balance of the Barekese reservoir in Kumasi, Ghana. J Sci Technol (Ghana) 35(3):34. https://doi.org/10.4314/just.v35i3.4
Duka MA, Lasco JDD, Veyra CD, Aralar AB (2018) Comparative assessment of different methods in generating design storm hyetographs for the Philippines. J Environ Sci Manage 21(1):82–89. https://doi.org/10.47125/jesam/2018_1/08
Echendu AJ (2022) Flooding in Nigeria and Ghana: opportunities for partnerships in disaster-risk reduction. Sustainability: Sci Pract Policy 18(1):1–15. https://doi.org/10.1080/15487733.2021.2004742
El baida M, Boushaba F, Chourak M Classification machine learning models for urban flood hazard mapping: case study of Zaio, NE Morocco. et al (2024) https://doi.org/10.1007/s11069-024-06596-zFarooq, M., Shafique, M., & Khattak, M. S. (2019). Flood hazard assessment and mapping of River Swat using HEC-RAS 2D model and high-resolution 12-m TanDEM-X DEM (WorldDEM). Natural Hazards, 97(2), 477–492. https://doi.org/10.1007/s11069-019-03638-9
Fatdillah E, Rehan BM, Rameshwaran P, Bell VA, Zulkafli Z, Yusuf B, Sayers P (2022) Spatial Estimates of Flood Damage and Risk Are Influenced by the Underpinning DEM Resolution: A Case Study in Kuala Lumpur, Malaysia. Water (Switzerland) 14(14). https://doi.org/10.3390/w14142208
A
Frimpong BF, Koranteng A, Atta-Darkwa T, Junior OF, Zawiła-Niedźwiecki T (2023) Land Cover Changes Utilising Landsat Satellite Imageries for the Kumasi Metropolis and Its Adjoining Municipalities in Ghana (1986–2022). Sensors 23(5). https://doi.org/10.3390/s23052644GSS (2021) Ghana 2021 Population and Housing Census. General Report Volume 3A - Population of Regions and Districts
Hrachowitz M, Savenije HHG, Blöschl G et al (2013) A decade of Predictions in Ungauged Basins (PUB)—a review. Hydrol Sci J 58(6):1198–1255
A
Hu S, Fan Y, Zhang T (2020) Assessing the effect of land use change on surface runoff in a rapidly Urbanized City: A case study of the central area of Beijing. Land 9(1). https://doi.org/10.3390/land9010017A
Ibrahim S, Brasi B, Yu Q, Siddig M (2022) Curve number estimation using rainfall and runoff data from five catchments in Sudan. Open Geosci 14(1):294–303. https://doi.org/10.1515/geo-2022-0356Kreibich H et al (2025) Panta Rhei: Towards a new hydrological synthesis. Hydrological Sciences Journal
A
Lumbroso D (2020) Flood risk management in Africa. In Journal of Flood Risk Management (Vol. 13, Issue 3). Blackwell Publishing Inc. https://doi.org/10.1111/jfr3.12612Martins LF, Studart TM, de Filho CP, Porto JD, de Filho VCS, Costa FR (2023) da S. Flash Flood Reconstruction and Analysis—A Case Study Using Social Data. Climate, 11(1). https://doi.org/10.3390/cli11010020
McDermott TKJ (2022) Global exposure to flood risk and poverty. In Nature Communications. Issue 1) Nat Res 13. https://doi.org/10.1038/s41467-022-30725-6
A
Mensah H, Ahadzie DK (2020) Causes, impacts and coping strategies of floods in Ghana: a systematic review. In SN Applied Sciences. Issue 5) Springer Nat 2. https://doi.org/10.1007/s42452-020-2548-zMudashiru RB, Sabtu N, Abustan I, Balogun W (2021) Flood hazard mapping methods: A review. J Hydrol 603. https://doi.org/10.1016/j.jhydrol.2021.126846. Elsevier B.V.
A
Nogherotto R, Fantini A, Raffaele F, Di Sante F, Dottori F, Coppola E, Giorgi F (2022) A combined hydrological and hydraulic modelling approach for the flood hazard mapping of the Po river basin. J Flood Risk Manag 15(1). https://doi.org/10.1111/jfr3.12755Ongdas N, Akiyanova F, Karakulov Y, Muratbayeva A, Zinabdin N (2020) Application of hec-ras (2d) for flood hazard maps generation for yesil (ishim) river in kazakhstan. Water (Switzerland) 12(10):1–20. https://doi.org/10.3390/w12102672
Osei MA, Amekudzi LK, Wemegah DD, Preko K, Gyawu ES, Obiri-Danso K (2019) The impact of climate and land-use changes on the hydrological processes of Owabi catchment from SWAT analysis. J Hydrology: Reg Stud 25. https://doi.org/10.1016/j.ejrh.2019.100620
A
Pathak AA, Mathad AR, Gagnon AS (2025) Integrating socio-economic variables in urban flood damage assessments: a case study of Bengaluru, India. Nat Hazards 121:5387–5412. https://doi.org/10.1007/s11069-024-07019-Rangari VA, Umamahesh NV, Bhatt CM (2019) Assessment of inundation risk in urban floods using HEC RAS 2D. Model Earth Syst Environ 5(4):1839–1851. https://doi.org/10.1007/s40808-019-00641-8
A
Sahoo SN, Sreeja P (2017) Development of Flood Inundation Maps and Quantification of Flood Risk in an Urban Catchment of Brahmaputra River. ASCE-ASME J Risk Uncertain Eng Syst Part A: Civil Eng 3(1). https://doi.org/10.1061/ajrua6.0000822Sarchani S, Seiradakis K, Coulibaly P, Tsanis I (2020) Flood inundation mapping in an ungauged basin. Water (Switzerland) 12(6). https://doi.org/10.3390/W12061532
Snikitha S, Kumar GP, Dwarakish GS (2025) A Comprehensive Review of Cutting-Edge Flood Modelling Approaches for Urban Flood Resilience Enhancement. Water Conserv Sci Eng 10:2. https://doi.org/10.1007/s41101-024-00327-y
Trinh MX, Molkenthin F (2021) Flood hazard mapping for data-scarce and ungauged coastal river basins using advanced hydrodynamic models, high temporal-spatial resolution remote sensing precipitation data, and satellite imageries. Nat Hazards 109(1):441–469. https://doi.org/10.1007/s11069-021-04843-1
van Emmerik T, Mulder G, Eilander D, Piet M, Savenije H (2015) Predicting the ungauged basin: Model validation and realism assessment. Front Earth Sci 3. https://doi.org/10.3389/feart.2015.00062
A
Vojtek M, Petroselli A, Vojteková J, Asgharinia S (2019) Flood inundation mapping in small and ungauged basins: Sensitivity analysis using the EBA4SUB and HEC-RAS modeling approach. Hydrol Res 50(4):1002–1019. https://doi.org/10.2166/nh.2019.163Wüthrich D, Korswagen PA, Selvam H, Oetjen J, Bricker J, Schüttrumpf H (2024), March 1 Field survey assessment of flood loads and related building damage from the July 2021 event in the Ahr Valley (Germany). Journal of Flood Risk Management. https://doi.org/10.1111/jfr3.13024
Statements and Declarations
A
Funding
This study was made in the context of a PhD funded by the BMBF (Bundesministerium für Bildung und Forschung) through the project Current and future risks of urban and rural flooding in West Africa (FURIFLOOD project), Grant number: 01LG1001A.
A
Competing InterestsThe authors declare no conflicts of interest.
A
Author Contribution
Conceptualization: Vida Akyeampong, Kwaku Amaning Adjei, Leonard Kofitse Amekudzi, Data curation: Vida Akyeampong, Godfred Abbey Torsah, Formal Analysis; Vida Akyeampong, Godfred Abbey Torsah, Mohammed Najim Nasimi, Methodology: Vida Akyeampong, Kwaku Amaning Adjei, Software: Vida Akyeampong, Mohammed Najim Nasimi, Supervison: Kwaku Amaning Adjei, Charles Gyamfi, Markus Disse, Leonard Kofitse Amekudzi, Visualization: Vida Akyeampong, Kwaku Amaning Adjei, Markus Disse, Leonard Kofitse Amekudzi, Writing: Vida Akyeampong, Godfred Abbey Torsah, Mohammed Najim Nasimi, Kwaku Amaning Adjei, Charles Gyamfi, Markus Disse, Leonard Kofitse Amekudzi. All authors read and approved the final manuscript.