Drivers and coping mechanisms for floods: experiences of residents in urban Kumasi, Ghana

Over the last few years, the occurrence and severity of climate change-induced floods in urban Kumasi, Ghana, have increased considerably, leading to devastating effects on both lives and properties. Yet, policy developments have often ignored the perspectives of residents who have experienced these floods. This study used a mixed-method approach including household surveys with 150 households and 5 key informant interviews to determine the drivers and coping mechanisms for floods in three selected communities in urban Kumasi, Ghana. Rainfall data (2001–2021) were analyzed using the Mann–Kendall trend test. A binary logistic regression was used to determine the enablers and barriers to coping mechanisms for floods. Results showed that the study respondents perceived changes in rainfall and floods through increasing amounts of rainfall, increasing duration of the rainy season, and increased incidences of floods. The perceptions of the respondents regarding rainfall changes were consistent with the analyzed rainfall data (2001–2021) which showed an increasing annual rainfall. A multiplicity of causes including poor design of drains, choked drains, and inadequate drainage infrastructure were reported by the study respondents. Destruction of properties, decreased economic productivity and soil erosion were reported as the major effects of floods. The respondents implemented various coping mechanisms including temporary migration and relying on family and friends to manage the adverse effects of floods. Findings indicated that access to climate information, access to household communication gadgets, age of respondents, and period of staying in the community significantly affected the coping mechanisms employed by the respondents (p < 0.05). Barriers impeding the implementation of these coping mechanisms reported by the study respondents included financial constraints, inadequate support from government and non-governmental institutions, and a lack of understanding of early warning systems. Accessibility to timely climate information should be prioritized to help people improve their information-sharing and decision-making processes in managing floods in urban Kumasi.

household objects, clothes, roads, and buildings, were damaged as a result of the heavy downpour (Ghana Business News 2021).
Given the adverse impacts of floods on the lives of people, it is crucial to build their adaptive capacity to manage flood risks by identifying coping mechanisms. Adger et al. (2003) define coping mechanisms as the short-term practices or interventions employed to minimize the adverse effects of climate change and extreme weather events. Coping strategies are acknowledged as critical strategies that can be employed to reduce the threats posed by floods and thereby enhance household resilience (Sultana and Rayhan 2012;Mavhura et al. 2013;Hooli 2016). However, peoples' coping mechanisms are directly related to their perception of rainfall changes and floods (Adelekan and Asiyanbi 2016;Diakakis et al. 2018).
Various studies have explored the vulnerability of people and cities to floods both globally (e.g., Dixit 2003;Dewan 2015;Hellman 2015) and in Ghana (e.g., Amoako and Inkoom 2018;Owusu-Ansah et al. 2019). For instance, Dixit (2003) highlighted the vulnerability of people in risk-prone areas of the Himalayan-Ganga region. Hellman (2015) revealed that floods are a recurring problem for people in Jakarta and as a result have intensified the people's vulnerability to floods particularly poor families living on river banks in the city center.
In Ghana, Amoako and Inkoom (2018) revealed that flood vulnerability in informal settlements in Accra has co-evolved with the dynamics of informal urbanization and dwelling processes. Owusu-Ansah et al. (2019) argued that flood vulnerability around the Weija Dam, near Accra, can be explained by the city's complex peri-urbanization trajectories. Further, Campion and Venzke (2013) reported that the extent to which urban areas in Ghana are susceptible to flooding is affected by a range of factors including housing, topography, infrastructure, and preparedness.
Nonetheless, this susceptibility could be moderated through economic and social interventions by enhancing the ability of stakeholders and institutions to address flood risks. Several frameworks for flood risk management have been developed in the past (e.g., Winsemius et al. 2013;Li et al. 2016; Barendrecht et al. 2017). For instance, Li et al. (2016) put forward an empirical framework for the evaluation of flood risk and benefit evaluation of flood management strategies in urban areas. The authors concluded that it is hard to meet the rising pressures for flood management simply depending on systemic strategies. Furthermore, using a case study of Vienna in Austria, Barendrecht et al. (2017) concluded that models may contribute significantly to the management of floods by exploring a wider range of possible futures, including unanticipated phenomena.
However, the capacity required by urban dwellers to address the challenges associated with flood risks has been relatively underexplored not only in Ghana but in sub-Saharan Africa. Moreover, the perspectives of residents who have experienced these floods have largely been ignored in policy developments and implementation. This study addresses this important gap by investigating the drivers and coping mechanisms for floods using the experiences of residents in three selected communities in urban Kumasi, Ghana. The study answers the following research questions: (i) what is the extent of rainfall changes and perception of the study communities to these changes and floods in urban Kumasi? (ii) what are the perceived causes and effects of floods in urban Kumasi? (iii) what are the coping mechanisms used to manage floods in urban Kumasi? and (iv) what are the enablers and barriers that influence the implementation of the coping mechanisms by residents in urban Kumasi? 1 3

Description of the study area
The study was conducted within urban Kumasi (Fig. 1). The city forms part of the Kumasi Metropolis which is one of the thirty administrative districts in the Ashanti Region, Ghana. The metropolis is approximately 270 km north of the national capital, Accra ( Table 1). The Kumasi metropolis is located within the wet sub-equatorial climatic region that experiences the average minimum and maximum temperatures of 21.5 • C and 30.7 • C , respectively. The mean humidity of the area ranges from 84.16% at sunrise and 60% at sunset (Ghana Statistical Service [GSS] 2013). Kumasi metropolis is predominantly a commerce or trade service economy with the sector employing an estimated 71% of the population (GSS 2014). Industry and agriculture employ an estimated 24% and 5% of the population, respectively, in the Kumasi metropolis (GSS 2014).
Within the metropolis, three communities (Sepe-Buokrom, Atonsu, and Ahinsan) were selected to understand the experiences of residents regarding the drivers and coping mechanisms for floods. These communities were purposively selected because of their vulnerability to floods (Owusu-Ansah 2016).

Data collection
A cross-sectional research design was employed for the primary data collection. Data were collected using a mixed-method approach including household surveys and key informant interviews. A mixed-method is a research approach that allows the collection and analysis of both quantitative and qualitative data within the same study (Creswell and Clark 2017). Understanding the drivers and coping mechanisms for floods is a complex adventure; hence, the study adopted a mixed-method approach that allowed a better exploration of the varied experiences of study respondents in urban Kumasi, Ghana. Concerning the household surveys, 150 households were randomly chosen using the lottery technique. We administered a well-organized set of questions with a predetermined number of outcomes and a few undetermined ones on a face-to-face basis at the respondents' homes or any convenient place. The questionnaire was administered in the local language (Asante Twi) from April to May 2021. The questions focused on the socio-demographic characteristics of the respondents, their perception of rainfall changes and floods, their perceived causes, and effects of floods, and the coping mechanisms employed by the respondents to manage floods in the communities (Appendix 1). Monthly rainfall data for 20 years  for Kumasi were obtained from Ghana Meteorological Agency (GMet) in Accra. Data quality control was ensured using the normal ratio method to efficiently fill data gaps.
Key informant interviews were organized with five individuals purposively selected because of their in-depth knowledge of flood issues in the study communities. The key informant interviews provided the opportunity to validate the key themes that emerged from the respondents in the household surveys. Key informants included community leaders, policymakers at the municipality, and some Assembly members. With their permission, the interviews were tape-recorded and later transcribed. Each interview lasted for about 30 mins.
Ethical considerations focused on the consent of the respondents and confidentiality. Verbal consent was sought from the respondents before the administration of the questionnaire. Respondents were made to know that their participation in the study was voluntary and that they could pull out of the study at any point they felt uncomfortable with some

Data analysis
The Mann-Kendall trend test was used to analyze the rainfall trend in the study area while Sen's slope estimator was used to calculate the magnitude of the slope. The procedure to compute the Mann-Kendall trend test has been documented in prior research including Baffour-Ata et al. (2021). The variability of rainfall for the study period was computed using the coefficient of variation (CV). With the aid of Statistical Package for Social Sciences (SPSS) version 21, household survey data were analyzed using descriptive statistics including frequencies and percentages. Qualitative data were analyzed with thematic analysis which allowed the categorization of key themes that emerged from such analysis (Braun and Clarke 2019). Relevant quotes were used to enrich the discussions, where necessary. A binary logistic regression was employed to determine the enablers and barriers to coping mechanisms for floods. This model has been used in previous studies to assess the factors influencing adaptation or coping strategies (see, e.g., Alemayehu and Bewket 2017;Ali et al. 2021). The model is represented as: where Y = response variable; 0 = intercept; 1 , 2 , z = slopes; X 1, X 2, X z = explanatory variables; e = residual (error). The model's response variable is coping mechanisms used by the respondents in managing floods, while the independent variables are the socio-demographic characteristics including gender, age, household size, educational level, period of staying in the community, access to climate information, access to household communication gadgets and the main occupation of the study respondents. The explanatory variables incorporated in the model and their hypotheses are indicated in Table 3. Using a variance inflation factor (VIF), the explanatory variables were evaluated for collinearity in the regression model (Zainodin and Yap 2013) (Table 2).

Socio-demographic characteristics of respondents
Results showed that the majority of the respondents were females (n = 78; 52%) as compared to the males (n = 72; 48%) ( Table 4). This is following the recent census report for Kumasi which revealed that the female population outnumbers that of their male counterparts (Kumasi Metropolitan Assembly 2021). About 56% of the respondents (n = 84) were in the age category of 21-40 years. In terms of education, 68 of the respondents (representing 45%) had attained basic education. Only 7.3% of the respondents reported having completed tertiary education. Education is an important socioeconomic factor because it tends to improve life circumstances and reduce environmental risk factors (Aikens and Barbarin 2008;Morgan et al. 2009). Most of the respondents (n = 98; 65%) had stayed in the study communities for more than 10 years. This implies that most of them had in-depth knowledge and understanding of flood issues in the study communities. About 63% of the respondents reported having access to climate information. Climate information is defined as "the generation, provision, and contextualization of information and knowledge derived from weather and climate research for decision-making at all levels of society" (Vaughan and Dessai 2014, p. 587). A significant coping step that can improve the ability of vulnerable communities to address climate-related disasters is the timely access and utilization of accurate climate information (Antwi-Agyei et al. 2021a, b;Baffour-Ata et al. 2022). Hence, most of the respondents accessing climate information implies that they could manage floods in the study communities.

Extent of rainfall changes and respondents' perception of these changes and floods
The results showed an erratic rainfall pattern from 2001 to 2021 (Fig. 2). The positive values of Kendall's tau and Sen's slope suggest increasing annual rainfall over the study period, although the trend is not statistically significant (p > 0.05). The highest annual rainfall (1956.8 mm) was recorded in the year 2021, while the lowest (1185.8 mm) was recorded in the year 2001. The rainfall variability under the period of consideration is about 15%. Table 5 presents the respondents' perception of changes in rainfall and floods. Perception is relevant in environmental and climate change studies because peoples' coping mechanisms to weather extremes are related to their perception of changing weather conditions (Afriyie et al. 2018;Ge et al. 2021). For instance, Ge et al. (2021) reported that comprehension and enhancing public risk perception have become key components in flood risk management. About 54% of the respondents (n = 81) reported that the amount of rainfall had increased in the study communities. For instance, one of the respondents stated: Recently, there has been an increase in the amount of rainfall, particularly in Kumasi. The rains can start around 4 pm today and end tomorrow at dawn. The intensity at which it rains these days is also very high so you can imagine the total amount that could be measured when the rain stops -(Male respondent, Sepe-Buokrom, April 2021) About 48% of the respondents (n = 72) reported that the duration of the rainy season had also increased. Regarding floods, 81 of the respondents (representing 54%) reported that flood incidence had increased. Slightly more than half (55%) of the study respondents (n = 82; 55%) indicated that storms had also increased in the study communities. These findings agree with earlier studies suggesting that rainfall amounts have increased recently in Ghana contributing to increased incidences of floods (for example Amoateng et al. 2018;Tazen et al. 2019;Lai et al. 2020;Jamal et al. 2021). The trend analysis of rainfall data was consistent with the perception of the respondents indicating that the respondents correctly perceived the rainfall changes and floods. The increase in storms and strong winds in the study communities have the potential to increase flood incidence. This is because studies have shown that increased storms and strong winds are the main drivers of floods (Slingo et al. 2014;Muis et al. 2016).

Respondents' perceived causes and effects of floods in urban Kumasi
Study respondents had varying opinions on the causes of floods with the majority of them reporting that choked drains (n = 145; 97%) and poor design of drainage infrastructure (n = 145; 97%) were the major causes of floods in the communities (Fig. 3; Table 6). For example, one respondent remarked:

3
The poor design of drainage infrastructure is the primary cause of floods in this community. The drains are not large enough to accommodate large volume of water and ensure easy flow of water. The result is that when the rains fall heavily, there is an overflow of water, even in areas considered to be dry -(Female respondent, Atonsu, May 2021).
Lack of adequate drains was also reported by a majority of the respondents (n = 140; 93%) as a major cause of floods in the study communities. However, few of the respondents

I believe the main cause of floods in this community is an act of God. Signs like these have already been predicted in the Bible to happen during the end times so I am not surprised -(Female respondent, Ahinsan, May 2021).
The results are consistent with earlier studies conducted in other parts of Ghana (e.g., Amoako and Frimpong Boamah 2015; Bempah and Øyhus 2017) and sub-Saharan African countries including Kenya (e.g., Okaka and Odhiambo 2019). The authors revealed that poor environmental attitudes (e.g., bad refuse disposal), poor urban planning and development, inadequate drainage facilities, and Acts of God were perceived as the main causes of floods.
The arrangement of structures and layout plan of the study communities were haphazardly scattered without adequate spaces created in between these buildings. Buildings for human settlements, offices, and businesses were not properly and orderly arranged. These unplanned settlements associated with a lack of adequate drainage infrastructure resulted in increased flood occurrences in the study communities. Results further revealed that improper waste disposal practices were a key cause of increased incidence of floods in the study communities. The fewer drains situated in these areas were choked with solid waste due to improper waste disposal practices of inhabitants. This prevents the easy flow of water when it rains heavily. Low-lying nature of reliefs in these communities contributes to the incidence of floods. Whenever there are heavy rains, such areas experience floods. The uncontrolled generation of waste has been a major constraint to many city authorities in Africa. Poor waste management practices may elevate flood hazards in several ways. Improper disposal of waste along roads could physically block the drainage system, influencing the flow of runoff in the canal system. This directly causes flash floods in urban areas (Mensah and Ahadzie 2020). This driving factor is worsened by the increased rate of ignorance and poor enforcement of environmental laws in the country (Ahadzie et al. 2016).
Regarding the perceived effects of floods, a greater number of the respondents (n = 148; 99%) disclosed that their properties have been damaged as a result of the floods (Table 7; Fig. 3). Floods pose serious risks to domesticated animals including dogs, pigs, and goats. This was closely followed by decreased economic productivity (n = 147; 98%). According to the respondents, floods adversely affect the flow of trade and finance that make up both individual and national economies. This, they attributed to the fact that when floods occur, people are unable to go about their various economic activities and this tends to affect their incomes and eventually their standard of living. Most of the respondents (n = 128; 85%) also pointed out soil erosion as one of the major effects. The fast movement of water can wash the top soil which accommodates the majority of the nutrients needed by plants, eventually depleting the soil (Abayomi 2014). Soil erosion weakens buildings and creates channels and gullies on streets and roads, making some of them impassable. Nonetheless, few of the respondents (n = 36; 24%) revealed that lives were lost during these floods in the communities. Such findings compare favorably with previous studies (Memon 2015;Mensah and Ahadzie 2020) indicating that floods exert significant adverse impacts on the lives of people including the destruction of properties, crops, and lives, decreased economic productivity, erosion of soil and waterlogged lands that eventually leads to an outbreak of water-related diseases such as cholera. For example, Memon (2015) reported that heavy rains and floods have been responsible for unparalleled losses to the lives of humans and the environment in Pakistan.

Coping mechanisms used by households to manage floods
Results showed that the respondents employ a host of coping mechanisms in response to floods in the study communities ( Table 8). The majority of the study respondents (n = 149; 99%) reported using temporary migration as a key coping mechanism for floods in the study communities. Others (n = 146; 97%) also relied on family or friends during flood events. About 48% of the respondents (n = 72) relied on support from government or nongovernmental organizations during floods. For instance, one of the respondents indicated that:

Government officials like National Disaster Management Organization (NADMO) come to our aid two weeks after the flood incident and provide us with blankets and mattresses which will not be beneficial to our situation at that moment. They do not show up again until there is another flood situation -(Female respondent, Ahinsan, May 2021).
Social capital is an important aspect of a community's ability to recover quickly and 'bounce back after a disaster. Coordination and collaboration for mutual benefit are made easier by these social networks, norms, and social trust (Okayo et al. 2015). The more networks a household head is a member of, the easier it is for such a family to seek assistance in dealing with flood disasters (Nji and Balgah 2019).
The majority of the respondents also relied on family and friends for assistance which has often proven to be very helpful (Armah et al. 2010;Casagrande et al. 2015). This emphasizes the crucial role played by social networks as critical tools for the management of floods. These findings agree with earlier research works (for example Bola et al. 2014;Musyoki et al. 2016;Owusu-Ansah et al. 2019) conducted in other parts of Ghana and Africa suggesting that households have managed floods through the implementation of coping strategies such as reliance on family and friends and temporary migration. A study in Mississippi in the USA showed that flood-affected people relied mostly on immediate family for support (Casagrande et al. 2015).
Temporary migration in this study is defined as migration to another community that is not intended to be permanent, for a specified and limited period, and is usually undertaken because of the floods (Call et al. 2017). The respondents reported that after the floods, they return to their communities and clean up their houses hoping to recover some of the properties that were not destroyed by floods. Few of the respondents relied on governmental and non-governmental organizations (Table 8). This could be attributed to the fact that most of these organizations do not have specific policies targeted at flood victims (Armah et al. 2010). This situation subverts the sustainability of the relief programs initiated by governmental and non-governmental organizations (Armah et al. 2010).

Determinants of coping mechanisms used by the respondents
Results showed that the coping mechanisms employed by the respondents to manage floods were significantly influenced by socio-demographic factors including access to climate information (p = 0.000, β = 1.983), access to household communication gadgets (p = 0.001, β = 2.371), period of staying in the community (p = 0.012, β = 0.670) and age of the respondent (p = 0.040, β = 5.350) ( Table 9). These factors have been reported in prior research to significantly affect coping mechanisms for floods Nji et al. 2019;Navarro et al. 2020). For instance, climate information services provide sciencebased and user-specific information that can be used to manage flood risks and exploit opportunities created by floods (Machingura et al. 2018). The predominant source of dissemination of climate information to respondents is the media. This implies that acquiring household communication gadgets such as mobile phones, radio, and television is very crucial to accessing climate information. Most of the respondents in this study had access to household communication gadgets (Table 4) thereby explaining the reason for it being a significant factor.

3
The period of staying in the communities also facilitated the implementation of coping mechanisms in the study communities. Results showed that the majority of the respondents had lived in the selected communities for more than 10 years (Table 4). There is a high probability that people who have lived in a flood-prone community for a long time will cope better with flood risks as compared to those who have lived there for a shorter period. Such people become familiar with the environment and can cope with local indigenous practices. This is consistent with prior research indicating that factors including the period of stay in a community significantly influenced coping mechanisms to floods in Colombia (Navarro et al. 2019). Age was also a significant determinant, and this is in line with a previous study conducted in northern Ghana indicating that age plays a crucial role in the implementation of coping strategies for floods (Lolig et al. 2014). This is because older residents tend to depend on social support as compared to younger ones (Berman et al. 2015).
Factors including household size, educational level, and occupation were not significant determinants. This compares favorably with a study conducted in the past where the authors found some of those factors to be non-significant (Wang et al. 2018). Household size can significantly or insignificantly influence urban residents' coping mechanisms for floods. A very large household size could contribute to the conversation, confirmation, and sharing of flood-related issues thereby enhancing the management of floods significantly. However, in this study, household sizes were not too huge (Refer to Table 4).

Barriers to coping mechanisms by respondents in urban Kumasi
Results suggested that study respondents are confronted with several barriers in the implementation of coping mechanisms for floods (Table 10). The key barrier reported by the respondents was financial constraint (n = 145; 97%). This was followed by a lack of inadequate support from governmental and non-governmental institutions (n = 142; 95%) and inadequate logistics to respond to floods (n = 142; 95%). For example, one key informant stated: Government support flood victims in the municipality with emergency relief items including mattresses, clothes, sugar, rice, boxes of cooking oil, soap, blankets, plastic buckets, bowls, cups and mats, bags of sachet water, as well as boxes of mosquito coils and treated mosquito nets. However, the flood victims tend to want cash donations and I think that is why they report that we do not support them-(Key informant, May 2021) Lack of climate information was mentioned as the least barrier (n = 55; 37%). Concerning the financial barrier, key informant interviews revealed that insufficient funds are usually disbursed to medium and small-sized organizations to spend and these are usually prioritized for urgent matters. Organizations often have competing needs with inadequate human, financial and technical resources to address issues of floods. The organizations are then left with inadequate resources to embark on comprehensive awareness programs aimed at sensitizing residents on coping mechanisms for flood risk management. These findings are consistent with previous studies (e.g., Biesbroek et al. 2013;Halkos and Skouloudis 2020) suggesting that barriers to coping mechanisms for floods are often related to financial, informational, cognitive, institutional, and social issues.
The respondents also disclosed that they do not have enough funds to relocate to communities that are less prone to floods. Furthermore, while there is a huge agreement on climate change, there is unpredictability about the extent of the changes. Although scientific knowledge will keep growing, however, for numerous forms of climate risks including flood risks, it will be difficult to have accurate predictions of the timing as well as the magnitude of floods. People with low comprehension of the changes in climate and weather extremes will also not prepare adequately for its effects or even welcome the idea that other people should be made to cope (Antwi-Agyei et al. 2015).

Conclusion and recommendations
This study examined the drivers and coping mechanisms for floods in urban Kumasi, Ghana. Study respondents reported an increase in the amounts of rainfall which had also increased incidences of floods in the study communities. The perceived drivers were building on/close to water resources, poor waste management practices, unplanned settlements, and inadequate drainage infrastructure. Results indicated that flood incidences have had a devastating impact on the lives of the people across the city. To moderate the effects of floods, study respondents reported adopting several coping mechanisms including temporary migration and relying on assistance from family and friends. These are crucial mechanisms that allow flood-affected communities to sustain their livelihoods. However, it should be stressed that such interventions may not be sufficient with the projected changes in rainfall patterns closely linked to climate change and variability. Results showed that socioeconomic factors including access to climate information, age, period of staying in a flood-prone community, and access to household communication gadgets were significant in determining the implementation of flood coping mechanisms. In their attempt to moderate the adverse effects of floods, study respondents enumerated several significant barriers including financial constraints, inadequate logistics to respond to floods, uncertainty about the impact of floods, and lack of understanding of early warning systems. It is therefore important that policymakers take these factors into consideration in the design of flood interventions.
Accessibility to timely climate information should be prioritized to help people improve their information-sharing and decision-making processes. Further, public education campaigns should be intensified and aimed at changing the behavior and attitudes of inhabitants to encourage proper waste disposal practices. Major river channels in the Kumasi metropolis should be regularly dredged to improve drainage and lessen the probability of the occurrence of floods. This should be supported with regular public education and awareness creation campaigns aimed at ensuring proper waste management practices in the metropolis. Additionally, we recommend that there should be effective enforcement of environmental laws by institutions and other regulatory bodies such as the Kumasi Metropolitan Assembly, the Environmental Protection Agency (EPA), and the National Disaster Management Organization (NADMO) to address improper waste disposal practices in the metropolis.