Urban anthropization: community vulnerability and resilience to flood hazards in eastern Democratic Republic of Congo

The effects of the 2020 floods in Uvira were exacerbated by urban anthropization and climate change. Floods are causing severe human, material, economic, and environmental losses as well as affecting socioeconomic and ecological systems beyond the affected community’s ability. The purpose of this study was to analyze land use changes in Uvira from 1995 to 2021 and to assess the vulnerability and effectiveness of the community’s resilience strategies in response to both the impact of flooding and the risk of future flooding. Data were collected from victims in the municipalities of Kanvinvira and Mulongwe in the city of Uvira, which were severely impacted by the 2020 floods. This study used a mixed approach that combined the spatial analysis of land use and land change with the anthropization of community perceptions of vulnerability and flood mitigation measures. Spatial analysis revealed that anthropogenic land use increased by 4.73% per year between 1995 and 2021. In comparison to the initial surface covered in 1995, this has increased the surface occupied by human activities by 127.7%. Gender, education, and pre-flood professional activity of the household head, combined with pre-flood house construction quality, property ownership, degree of house damage, and family member death, explained households’ resettlement capacity. The Mann–Whitney U test results revealed that households’ vulnerability and resilience to 2020 floods were influenced by their residence municipality (Kanvinvira and Mulongwe) and lifestyle (living at the host site or reintegrating into the community). During and after floods, disaster victims received assistance from the government and stakeholders (churches, civil society, and non-governmental organizations) to mitigate flood effects, but they were unable to restore the lives of hundreds of flood victims. The study’s findings suggest that provincial and local disaster management authorities, as well as stakeholders, should invest in preventive and sustainable flood risk management. Public awareness of flood prevention and mitigation should be increased through capacity-building training, preparedness, and sensitization. Finally, effective adaptation practices are required to reduce future potential damage.


Introduction
Flood risk is an old hazard that has been exacerbated in today's society due to the effects of global warming and land-use changes.Aside from being a 'natural' disaster, the main threats are anthropogenic activities, such as rapid urbanization and agricultural and non-agricultural practices (Bronfort 2017, Waghwala andAgnihotri 2019).Human activities (such as the conversion of grassland to cultivated land, raw material extraction, urban expansion, and industrialization) are transforming the world's landscapes and altering the conditions, both in collective sites and in host families (OCHA. 2020, PACIF 2020).Flooding is exacerbated by climate variability and unpredictability, irregular rainfall, and reduced agricultural harvests, exacerbating poverty in conflict-affected areas (Sulser et al 2010, Kassegn andEndris 2021).Flood damage in Uvira is linked to a lack of systematic urban planning, adequate drainage infrastructure, and regular maintenance of existing drains, technical capacity, and financial resources.The poorest neighborhoods are the most vulnerable to threats (Lavigne et al 2010).In response, local communities developed strategies to assess and protect themselves from future threats (Colbeau-Justin andWeiss 2002, Dutozia andVoiron-Canicio 2019).
The purpose of this study was to (1) analyze land use changes in the city of Uvira from 1995 to 2021 and (2) assess the vulnerability and effectiveness of the community's resilience strategies in response to both the impact of flooding and the risk of future flooding.This study, which was conducted directly with flood victims and flood-risk communities in the city of Uvira, established the relationship between self-assessment measures of flood vulnerability and resilience in Uvira.Being severely impacted by the 2020 floods, data were collected from floods' victims in the municipalities of Kanvinvira and Mulongwe in the city of Uvira.This paper discussed the impact of sociodemographic factors on the community's vulnerability and resilience to the risks of pluvial and lacustrine flooding.The resilience model used in this article reflected the characteristics of the flood resilience elements, namely exposure, sensitivity, adaptive capacity, and recovery capacity.Based on chronological data  on land use and land change (Giampieri and Vialle 2020) and cross-sectional field data, this study assumed that Uvira's inhabitants were experiencing anthropogenic disasters.Anthropogenic activities have had a significant impact on land use-change, resulting in natural hazards such as flooding, which is the most recurrent in the region (Ekaka Azanga et al 2016).The effects of the April 2020 flood are still being felt three years later.Many individuals have become vulnerable as a result of the disruption of river dynamics and the catastrophic erosion of riverbanks caused by excessive sedimentation (Mugisho Bachinyaga et al 2022).Preventing or limiting anthropization is driven by the community and supported by environmental protection policy.The mixed approach used in this study allowed for the triangulation of results from the spatial analysis on land use and land change and from anthropization on community perceptions of vulnerability and flood mitigation measures.Spatial analysis was thus utilized to monitor the long-term effects of urban development on land-use change.Long-term analysis was required to evaluate changes and support adaptation strategies to climate change in order to minimize human impact on the environment (Plisnier et al 2018).The findings of this investigation revealed a link between urban anthropization and its catastrophic impacts on community wellbeing.Data from interviews, surveys, and focus group discussions were used to fill information gaps on community well-being and flood mitigation policies in Uvira (Nsabimana et al 2023).The research participants' experience was extremely important in reconciling findings from spatial analysis and field data analysis.The findings were compared to the literature, which drew on other flooding experiences from across the country and around the world.The outcomes were used to develop recommendations for improving the environmental, social, and economic dimensions of flood risk management programs in Uvira.

Study area
Uvira is located between 3°20' and 4°20' latitude south and 29°30' longitude east.It is bounded on the north by Kawizi, on the south by the Gengeza (Katongo) River, on the east by Lake Tanganyika, and on the west by the Munanira mountain range.The city of Uvira is characterized by a humid tropical climate with two distinct seasons.From October to May, the rainy season lasts about eight months, followed by the dry season from July to September.The temperature outside ranges from 24 to 28 °C.Uvira is located on the shores of Lake Tanganyika and is traversed by several rivers that cause flooding.The main rivers, which originate in the Minembwe highlands and flow directly into Lake Tanganyika, are the Kavimvira, Mulongwe, and Kalimabenge (figure 1).
Uvira, a city of less than 50 km 2 is experiencing rapid population growth.It has approximately 280,000 inhabitants (a very high population density of 5,600 inhabitants/km 2 ).People are engaged in agriculture, fishing, extensive livestock farming of cattle, pigs, goats, and poultry, and small and large-scale trade of agricultural products and necessities.Uvira is divided into three municipalities (Mulongwe, Kagando, and Kalundu), and each municipality is further divided into quarters.Mulongwe is the commercial and administrative center of Uvira.Kalundu Port is in Kalundu's municipality.Because of its strategic importance, Kalundu Port is the DRC's second-largest port (Kirangwa 2012).

Research design
This study used a mixed approach.The analysis of the spatiotemporal distribution of land use and land change, as well as the impact of anthropization on community well-being and the environment, was done using chronological data collection.Data from cross-sectional interviews, surveys, and focus group discussions were used to assess the impact of sociodemographic characteristics on community vulnerability and resilience, as well as to examine perceptions of vulnerability and resilience among people who are typically affected by these hazards.
The adaptive logical framework was used to combine the aforementioned parts as well as to demonstrate the logical sequences of the effects of urban anthropization and to provide means of coping with their uncertainties.This dilemma was exacerbated by ambiguity about the effects of flood mitigation efforts on land, water, and other natural resources, climatic conditions, demographic evolution, and people's overall well-being.The proposed approach involved an in-depth analysis of the community's past experience and lessons learned.Land use changes, flood control methods, and flood management planning processes and policies were all measured in this study (Poon et al 2009, Mah et al 2011).

Data sources and data collection
Observation, surveys, and focus group discussions (FGD) were used to collect primary data from residents and key informants.A spatial analysis of selected flood events was performed, as well as the impact of changing the Universal Time Coordinated (UTC) on flood risk.The data collected from the community was specifically about the early 2020 floods, but participants who had been involved in previous floods attempted to relate them to the current floods.Data was collected between 2020 and 2022.The data collection process was divided into three stages.The first stage involved observing the various flood-affected areas (Kavimvira, Mulongwe, and Kalundu) in 2020.During this phase, three local guides (one for each site) were used to identify the flood-affected areas.In each area, we collected geographical data about the flood-affected area.This enabled the selection of the two most flood-affected sites, Mulongwe (with more than 57,000 victims) and Kanvimvira (with at least 10,000 victims).Until our last fieldwork in December 2022, 336 flood victims' households were hosted at the Kanvimvira site and 212 households at the Kasenga site.We conducted semi-structured interviews with 142 affected households.We interviewed 64 survivors from the Kasenga and Kanvimvira sites, as well as 78 survivors who had been reintegrated into the community following the floods.Based on their geographical location, we surveyed 80 households in Mulongwe and 62 in Kanvimvira.Key informants from government agencies and civil society organizations (NGOs, churches) participated in in-depth interviews.The in-depth interviews were conducted with 12 key informants.During the interviews, we sought to comprehend the resilience strategies developed by participants to deal with flood risks.The participants identified the anthropogenic causes of the flooding's severity.The study participants were selected by convenience sampling and had to be either the household head (husband or wife) or another active adult in the household.The interview guide was distributed in advance to key informants so they to prepare for the discussion.The final stage included focus group discussions with community members who had hosted flood victims or helped with disaster relief (physical and material).The organized FGDs were used to collect information on the perceived severity, consequences, and community involvement in future flood risk prevention.Two focus group discussions (FGDs) were held in Mulongwe and Kanvimvira.Each FGD had 12 participants.The research authorization issued by the Uvira town council was shown to the chief of the quarters, who indicated the avenues where the victims could be found.The enumerators explained clearly the objectives and purpose of the study.The respondents were asked to give their verbal consent for the interviews and photography.Representatives from each population group were invited to a feedback meeting.During this meeting, participants validated the study's findings and made recommendations for the final field report.
The flood interview guide included open-ended questions (description of flood conditions, immediate actions taken, interventions received from community members, government, and other stakeholders, prospects, and so on), closed questions on socio-demographic characteristics and effects of flooding, and scaled questions on vulnerability and resilience indicators.Data collected on the vulnerability and resilience of flood victims was thematized into different components (Shah et al 2018).Vulnerability components include exposure (previous flood experience and the number of affected houses or houses constructed near the river or streams), susceptibility (building materials, physical and psychological disability, illiteracy, coping mechanisms, deaths, and livestock losses), and adaptive capacity (access information and credit, social networks, education, households working age group between 15-60 years, income sources and employment).We considered sociodemographic (households' social capacities, age, education, gender of household head, household size, religious and social beliefs, social network), economics (house ownership, employment, and multiple sources of income), physical (building materials, local infrastructure, houses near government-built structural measures and flood protection), and institutional components for resilience (quality of disaster management services, flood warning information, hazard mitigation training, first aid training, recovery assistance, livelihood restoration, and water sanitation and hygiene) (Qasim et al 2016, Shah et al 2018).We did not systematically collect data from all affected households because we were unable to collect data on vulnerability and resilience immediately after the floods and due to limited resources.We collected data on vulnerability and resilience using 5-point Likert scale questions (1 = Not important 5 = Important (de Brito et al 2017).

Data analysis
Based on the type of information collected, multiple data sets were analyzed.For quantitative data, descriptive and comparative analyses were used.The Chi-square test and the non-parametric Mann-Whitney U test were used to relate socio-demographic data on household vulnerability and resilience at significance levels of 1%, 5%, and 10% (de Brito et al 2017, Jamshed et al 2020).Thematic Framework Analysis was used to analyze the qualitative data (Kablan et al 2017).Geographical analyses were used to estimate the number of households and people affected by flooding and land use dynamics (Membele et al 2022).To allow data grouping, field data were transcribed and codified in a standard language.The analysis included merging the many approaches used and identifying important themes based on similarities to understand how they integrate into anthropogenic processes affecting land use change and community vulnerability and flood mitigation.

Estimation of households and population affected
The areas impacted by the 2020 floods were quantified using AIRBUS high-resolution imagery from 2019 (the year before the floods) and 2021 (the year after the floods).The images were captured with Google Earth 7.3 and georeferenced with ArcGIS 10.5.Flood-affected areas around the Mulongwe and Kanvimvira rivers were imaged using very high-resolution images in January 2021.The Kavimvira and Mulongwe rivers, as well as floodaffected areas, have been digitized using sediments left by previous flooding rivers as an indicator.
The affected population was estimated using 0.5 m spatial resolution AIRBUS images from 2019.This study focused on areas near the Kanvimvira and Mulongwe rivers.ArcGIS sampling tools were used to divide images into 50 m * 50 m grids.Households were digitized in each grid, and each house corresponded to a point.The number of households in each grid cell was assigned.The total population (figure 2) was calculated by multiplying the grid cell value by the household size (an average of 8 people).The same method was used to calculate the population density per pixel (grid cell).
Where P = total population, N = Household size, F = number of households, n = Observation The April 2020 floods severely impacted people living along the rivers, as shown in table 1.The Kavimvira River flooded 0.35 km 2 and the Mulongwe River flooded 0.56 km 2 .The riverside populations impacted by the Kavimvira and Mulongwe rivers were 5,446 and 13,125, respectively.The average affected population density was 31 people per 250m 2 , compared to 48 people per 250m 2 in Kavimvira and Mulongwe, respectively.

Land use change in Uvira between 1995 and 2021
The population of Uvira has grown over time, and it now occupies land that was previously forbidden for construction.One of the primary spatial indicators for quantifying the severity of anthropogenic environmental impact is land use assessment (Asmat et al 2016).The temporal and spatial dynamics of land cover were investigated in this study for the years 1995, 2000, 2010, and 2021.The study years were chosen using a quasidecadal approach.
To assess the dynamics of land occupation and use, Landsat level 2 images with a spatial resolution of 30 meters were downloaded from the USGS website (https://earthexplorer.usgs.gov/).Table 2 contains descriptions of the images used.Three types of Landsat satellite images were analyzed to determine LULCC (land use and land cover change) for Uvira in the current and previous years.These images are the 1995 Landsat Thematic Mapper (TM) image, the 2000 and 2010 Landsat Enhanced Thematic Mapper (ETM) images, and the Landsat OLI-TRS image.The images were chosen based on their cloud cover and acquisition date.Empires were extracted, and the four images considered were returned to the extensions and boundaries of the city of Uvira.The images were classified using supervised classification with the RANDOM FOREST algorithm.The classification was evaluated using the overall accuracy and the kappa index.
Post-classification approaches were used to detect and evaluate changes in land cover and determine the expansion rates of each type of occupation between the first year (1995) and the last year (2021).The LULC maps have been interpreted, and urbanization changes have been identified.The different land-use classes observed in the study area were divided into four major groups: urban area (anthropogenic occupation), vegetated area, open area (bare soil), and water bodies.The overall and annual expansion rates of each of these classes were used to evaluate their growth between 1995 and 2021.
The rate of change (Tv) used by Ahononga et al (2020) was used to assess the evolution of land cover units.
= Land area of occupation unit in year 1 and S2 = Land area of occupation unit in year 2. When Tv (%) is negative, the land cover unit regresses from year 1 to year 2, whereas a positive rate indicates an increase in the land cover unit.If the rate is zero, the land cover unit remains stable.

Results and discussion
Socio-demographic factors were among the most frequently used indicators of post-flood vulnerability and resilience.They are used to determine which factors expose households to risk and which promote post-flood resilience.Table 3 summarizes the socio-demographic characteristics of the flood survivor households studied.Data were collected from 50% of male and female household heads.The Pearson Chi-square test demonstrated, at the 5% threshold, a significant gender effect on household living conditions in Uvira after the 2020 floods.Female-managed households (64%) continued to live in host sites, while male-managed households (62%) reintegrated.In host sites, approximately 89% of household heads did not complete secondary education, whereas 50% of households reintegrated into the community.After the floods, 41% of household heads with a high school certificate or higher were back in the community.Professional activity before floods explained households' ability to respond positively to the flood effects.About 60% of household heads who worked in a lucrative activity other than farming were reintegrated into the community, compared to 42% of other victims living in host sites.Approximately 73% of the houses in Uvira were built with semi-durable materials.77% of reintegrated households were still living in their own homes.The floods destroyed the residences of 93% of the victims surveyed at the host sites.The owners repaired their partially destroyed homes before returning.Half of those who had been reintegrated into the community had built a new home.Household heads (96%) surveyed in host sites stated didn't have enough money to rebuild a new house, and some (4%) couldn't rebuild in the same place because they weren't authorized to rebuild on their plot.13% of households had lost an average of 1.3 people (SD = 0.744).Household heads who had lost relatives were shocked and stated they would never return to their plots.Eighty-nine percent of households had lost at least half of their assets.46% of households had about 4 sick people (SD = 3.1), 37% of households had about 2 physically injured people (SD = 0.92), and 13% of households had about 1 mentally injured person (SD = 0.3).Several illnesses, including cholera, malaria, typhoid fever, urinary infections, headaches, and high blood pressure, had emerged, in addition to increased malnutrition.Lowe et al (2013) and Rufat et al (2015) discovered multiple similarities with this study relating to socio-demographic characteristics in their systematic case reviews.Lowe et al (2013) found similar findings, despite being more interested in factors associated with health vulnerability.They discovered that men, persons over 65, and those with a low education level or socioeconomic class appeared to be facing a greater risk of physical health impacts, whereas women appeared to be at a higher psychological risk.While age, marital status, and race/tribe of household head were not shown to have statistically significant results, multiple studies have demonstrated that they are, particularly age, among the risk factors for post-flood vulnerability (Rufat et al 2015).Lim et al (2016)'s results discovered that gender, education level, presence of children, house ownership, number of house floor levels, and type of house material had been strongly associated with households' ability to reduce flood risk.
The socio-demographic information gathered demonstrates the quality and relevance of the data collected from Uvira flood-affected households.These findings were utilized to forecast households' ability to cope with the effects of land use change and control and manage human activities.Sections 3.1 and 3.2 provide supporting evidence.

Urban anthropization: land use change in uvira between 1995 and 2021
The presented findings (table 4) highlight the dynamics of land use between 1995 and 2021.Figure 2 depicts the rate of growth of each land use and land cover unit over these two years.The cartographic analysis clearly shows the areas affected by the April 2020 floods.A comparison of these images revealed that population growth, combined with a lack of urban planning measures, deforestation, and ignorance of river boundaries, are the precursors of flooding in Uvira.Waghwala and Agnihotri (2019) discovered consistent results that support these findings using geophysics in Surat City.They demonstrated that anthropogenic causes and a lack of floodwater management in urban areas are far more responsible for the flood risk in Surat City.The transition from a lowto-high urbanization model is the primary cause of increased flood risk.
For each year studied, table 4 shows the proportions assigned by land cover type.Following classification theory, the ArcGIS classifications were excellent because the Kappa coefficients obtained for the four years were well above 0.7 (Arvor et al 2011).Table 4 shows that 2021 had the highest classification accuracy, with a kappa coefficient of 0.94.In 1995, vegetation dominated the land cover in Uvira, accounting for 49.4% (or 11.318 km2) of the total area.In that year, bare soil occupied the smallest area, accounting for 17.1% (3,907 km2).The vegetation dominance in the 2000 image analyses was 54.7%.Anthropogenic land cover was the second most prevalent (20.8%).In 2010, vegetation and human occupation co-dominated land cover, with 37.7% and 38.5%.Human occupation outnumbered all other land use classes in the 2021 image analyses.In 2021, human occupation occupied 8,899 km 2 or 38.8% of the city's total surface area.
Table 4 and figure 3 show a gradual increase in human occupation and a steady decline in vegetation.Annual expansion rates ranged from −3.04 to 4.73%.Anthropogenic land use increased by 4.73% per year between So far, the discussion of land-use change has assumed that anthropogenic use endangers natural landscapes and must have consequences for local societies and landowners.Anthropization demand, which has been exacerbated by human pressure to improve productivity, has resulted in new occupation of risk and urbanization-prohibited zones.In a context where the urbanization process had not been thoroughly appraised, the people of Uvira arbitrarily occupied land to meet their needs.The land use change in Uvira was characterized by an overall decrease in soil quality.Land degradation in high-risk locations has been discovered to be a dramatic phenomenon, increasing the vulnerability of the local community.The field findings (section 3.2) allowed for an assessment of the consequences of this transformation on vulnerability as well as the identification of the factors related to vulnerability and resilience.

Vulnerability assessment
The average vulnerability scores of Uvira households to flooding in 2020 are shown in table 5.The five items most affected by the floods were loss of household assets (4.1), sources of income (3.77),employment (3.77), coping mechanisms (3.55), and livestock loss.The Mulongwe and Kanvimvira rivers had overflowed, and rising Lake Tanganyika waters had submerged riverside homes.The distressing cries of these households had alerted other households far away to evacuate the hazard site.Flood victims living far from the riverbeds had received help from relatives and donations from friends to rebuild their homes, which had been destroyed by the floods.Other households were relocated to areas unaffected by the floods.According to the victims, rising water from Lake Tanganyika caused a massive displacement of the population to host families, host sites, and Burundi.Faced with this situation, victims were housed in tarpaulin huts constructed by the Red Cross organization.Other families were housed in school classrooms after the disaster.Affected households were relocated to public host sites (Kasenga in the municipality of Mulongwe and Kilomoni in the municipality of Kanvimvira).Government initiatives to rebuild houses on safer sites for the victims have gone unfulfilled, leaving the victims in a state of stress and despair.Similarly to the findings of this study, Dube et al (2018) discovered that floods in Zimbabwe's Tsholotsho district damaged human shelters, roads, bridges, and dams.Farm locations near rivers and dams, poor quality building materials, and a lack of flood warnings contributed to the region's vulnerability to flooding.These floods increased the risk of contamination in host sites and in families during the COVID-19  2018) discovered, similarly to this study, that those who constructed near the lake and even pushed back the water's bounds were more exposed to Lake Tanganyika's rising floodwaters.The outcomes of this study show that Uvira's people have been victims of its egocentric acts against the environment.The Mann-Whitney U test results presented in table 5 were used to compare household vulnerability to flooding.This analysis was linked to municipality and household status.
In terms of household vulnerability, the difference was highly significant (p < 1%) between households in Mulongwe and Kanvimvira, as well as those in host sites and those reintegrated into the community.Households in Kanvimvira (41%) were more exposed than those in Mulongwe (24%).Households in Kasenga and Kanvimvira were more vulnerable than those in the community.The poor quality of the inhabited house (p < 5%), the physical handicap of the head of household (p < 10%), the loss of livestock (p < 5%), and the loss of household assets (p < 1%) were likely to exacerbate the vulnerability of households differently depending on the municipality.Poor households with low incomes and living in unsustainable housing were more vulnerable to the floods' effects (Dube et al 2018).Previous flood victims (such as those in 2000) reported controlling the flow of rainwater and evacuating their homes.Our results support Salami et al (2017) findings of a significant relationship between residents' flood awareness and previous flooding experience, but not between their flood risk awareness and preparedness level.Mulongwe households lost 36% of their livestock and assets.Flooding destroyed approximately 80% of food reserves.Floods, as discovered in Uvira, have had a significant impact on the food insecurity status, livelihoods, and development gains of households in African countries (Kassegn andEndris 2021, Reed et al 2022).Households living in the host sites and where the household heads were physically disabled (p < 10%), mentally disabled (p < 1%), had not studied (p < 5%), or had experienced the death of at least one household member (p < 5%) were vulnerable.The evacuation of disabled people exposed households to asset loss.Each household was assigned an active person to take care of people with mental health issues.The education of the household head explained the effectiveness of flood prevention and management.Households in Mulongwe had higher adaptive capacities in the social network, diversity of income sources, and employment, with a mean rank of 35% for each item, compared to households in Kanvimvira, which had less than 27%.Families with weak social networks were left to fight the floods alone.The priority in these households was to save children and other family members.Households surveyed on host sites had significantly higher means ranks than those interviewed in communities in terms of access to climate information (37%), access to credit (38%), fewer than 10 years' education for the head of household (36%), and working-age people (37%), whereas 29% of reintegrated households had a lower means rank.During the first night of rain, men, women, and young people banded together to rescue flood victims.Many of the women and girls at the host sites were sexually abused.Because girls and boys were packed together and mixed in the same room and/or location, young girls were victims of unwanted pregnancies and abortions.Girls were found to be victims of sexual violence in temporary shelters in Bangladesh.Early marriage was a coping strategy for dealing with the negative consequences of extreme weather (Ahmed et al 2019).During the floods, children were also victims of unintentional physical abuse and injury by their parents (Biswas et al 2010).As a result of living in critical sanitary conditions characterized by promiscuity in the reception sites, the women developed urinary tract infections and other transmissible infections (Lowe et al 2013).Furthermore, studies conducted in Sub-Saharan Africa have explained the link between floods and the spread of epidemic infections such as cholera, scabies, taeniasis, Rhodesian sleeping sickness, malaria, alphaviruses, and flaviviruses (Suhr and Steinert 2022).HH: Household, Mul : Mulongwe, Kanv : Kanvimvira, *** Significant at 1% threshold, ** at 5% and * at 10% Lowe et al (2013) found that women, children, and the elderly are most vulnerable to health and mental health risks during and after flooding, which is supported by this study.Young males and adults are at risk of physical harm and death.Floods impacted households' social and economic living conditions, affecting employment and income sources (Khandlhela and May 2006).Victims met in the community scored higher in terms of employment and income sources, with a mean rank of 35% each, compared to households surveyed at the host sites, which scored 28% and 27%, respectively.People who worked in the informal sector or engaged in certain income-generating activities lost their only source of income.Several economic units (kiosks, depots, products, plots of land, money, and so on) were washed away by the rivers.

Resilience assessment
The resilience of disaster victims was determined by the severity of the losses suffered by households during the floods.Each component's average score ranged from 2.06 (41.2%) to 4.34 (86.8%).Households were more resilient in ensuring their subjective well-being (social capacities, religious beliefs, and social network) as well as objective well-being (house ownership, employment, multiple sources of income, and proximity to government-built structural measures).Poverty and infrastructure reconstruction costs, as well as a lack of cooperation between communities and officials, are all barriers to managing flood disasters (Dube et al 2018).Table 6 displays the comparative mean rank results for the various items based on respondents' addresses and current household living conditions.HH social capacities (p < 10%), religious beliefs (p < 5%), and social beliefs (p < 5%) differed significantly by the municipality of disaster victims' residence.Households in Mulongwe developed greater resilience in their HH social capacities and religious beliefs, whereas households in Kanvimvira developed greater resilience in their social beliefs.Households were able to reintegrate into the community due to HH social capacities (p < 1%) and male household heads (p < 5%).Illnesses, deaths, and flood-damaged house repair costs have exacerbated poor households' vulnerabilities by affecting their subjective well-being through a lack of peace of mind (Kikwasi and Mbuya 2019).Female-headed single-parent families were more vulnerable to the floods' effects.Due to the inability of people to evacuate their belongings from their homes, these women saw their household assets dispersed (loss of means of subsistence) and were vulnerable to mental and physical insecurity.High and middle-income women in the 2011 Lagos floods appeared to be less constrained in their recovery due to access to economic resources, strong social networks, and the social choice to relocate temporarily or permanently from flood-prone areas.Women in low-income areas, on the other hand, were unable to relocate due to a lack of funding (Ajibade et al 2013).Household head age (>60 years) was more prevalent in host site households, with a mean rank of 34% compared to 30% for reintegrated community households.As a way of dealing with the economic consequences of the floods, 38% of spouses in Kanvimvira municipality were involved in income-generating activities.To mitigate the effects of the floods, household heads and working adults began to engage in occasional activities such as farm work, labor on construction sites, shoe repair and other craft work, market handling, and domestic work (Ajibade et al 2013).Households with 4 to 8 people earned between USD 10 and USD 50 per month.To compensate for unmet needs in their households, women began cutting and sewing, selling cakes, doughnuts, vegetables, and a variety of other agricultural products.Government and civil society organizations around the world were providing social assistance to vulnerable groups (Ajibade et al 2013).Household ownership was significantly higher in reintegrated households (35%) than in host site households (p < 10%).In terms of water distribution and access infrastructure, the physical resilience of households in Kanvimvira (p < 10%) was significantly higher than in Mulongwe.Building materials and housing near government-built structural measures helped reintegrated disaster victims be more resilient, with 36% and 37%, respectively, compared to 26% and 24% of households in host sites.Housing quality is an indicator of a household's standard of living (Dube et al 2018).Households living in houses constructed of durable materials were easily reintegrated into the community.Although houses built with durable materials were affected by the floods, they were not destroyed.Similarly to Fatemi et al (2020) in Bangladesh and Kikwasi and Mbuya (2019) in Dar es Salam, we discovered that older buildings constructed with natural materials are more likely to sustain high flood damage than those constructed recently with durable materials.According to Kikwasi and Mbuya (2019), vulnerability is exacerbated by income levels, house location, and inadequate stormwater drainage.Kanvimvira (36%) outperformed Mulongwe (28%), in terms of government assistance (p < 10%) and access to roads and other communications (p < 5%).Households in the host sites reported benefiting more from institutional interventions than households that developed their resilience strategies.Access to hazard mitigation training, first aid training, government recovery assistance, and livelihood restoration, for example, were ranked higher among disaster victims at reception sites than among those in the community.The government and civil society organizations were reported to have provided various forms of assistance for those most vulnerable to flooding.Those affected stated they had received help from community members, churches, non-governmental organizations, and the state.According to the testimonies of the victims, the 8th Community of Pentecostal Churches in Central Africa (CEPAC) intervened quickly by making the land In addition to providing emergency assistance, NGOs have been working to raise awareness among disaster victims and the entire community about how to avoid illness and infection during a flood crisis.Psychosocial support for those affected and those with special needs, as well as care for separated and unaccompanied minors.Around ten health centers provided free health care to disaster victims, including children aged 0 to 59 months, pregnant or breastfeeding women, and the elderly.To ensure adequate water supply, water treatment units and nearly 50 chlorination sites have been installed along the Mulongwe and Kamvivira rivers.In Uvira, public awareness campaigns have been organized to promote community relations, flood prevention, and disaster management education.NGOs and local governments organized sanitation, hygiene promotion, and cholera prevention assistance.Children were given school supplies to assist them in returning to school.Approximately forty schools that were destroyed by the floods have been rebuilt and equipped with school supplies.Local government concentrated on draining the rivers.Reforestation efforts in the Kitundu district, the flood epicenter, have been ineffective; very few trees have been planted tree years after the floods.
Similarly to other African countries, government initiatives were mixed in with community strategies.The actions were based on strengthening crisis management resources, relocating vulnerable populations living in flood-prone areas, and running awareness campaigns (Lavigne et al 2010).In the absence of effective government interventions, Amoako (2018) discovered emerging and sustainable responses to flooding and coping practices shaped by residents' social networks, political alliances, and sense of place in three informal communities in Accra.Residents are active agents in managing their vulnerability, working with socially and politically organized institutions, and their responses to flood events have included ongoing housing unit restructuring, the construction of communal drains, and the establishment of local evacuation teams and shelters (Fatemi et al 2020).Although these actions are critical, they are not long-term because they do not eliminate the risk of flooding.In its national urbanization plan, the Congolese government should include structural strategies based on hard and light engineering (Carrick et al 2019, He et al 2021).This study's findings have provided extensive empirical evidence about the impact of urban anthropization combined with climate change on the vulnerability and resilience status of households in flood-prone municipalities in Uvira.Anthropization, together with climate change, enhanced floods in the study area.These floods have had a significant impact on the livelihoods, assets, and coping strategies of affected households.Flood mitigation responses from governments and other stakeholders were inadequate and ineffective.This study's findings have offered policymakers and stakeholders with insights into how to increase flood preparedness and resilience for Uvira's communities.The study's methodological limitations prevented the collection of quantitative and detailed data from many respondents (direct and indirect victims) regarding disaster impacts, as well as the effectiveness of government and stakeholder policies and measures for flood mitigation and future flood prevention mechanisms.The study's weaknesses were the small but representative sample size, the lack of baseline data, and potential bias in self-reported data, and local leaders and stakeholders involved in natural disaster management's reticence to contribute information.The outcomes of this study should be interpreted with caution, and future research should focus on unfilled gaps.Accordingly, this study serves as a foundation for future flood research and government efforts to reduce flood-related vulnerability and resilience in flood-affected communities.Additional study is required to evaluate and generalize these findings, as well as to investigate the long-term effects of floods on the social, economic, and environmental aspects of Uvira households.Indeed, this study provides numerous potential measures, including upgrading the early warning system, increasing flood protection infrastructure, encouraging community engagement and awareness, and assisting with livelihood diversification and resettlement.

Conclusion and recommendations
Spatial analysis revealed that anthropogenic land use increased by 4.73% per year between 1995 and 2021.In comparison to the initial surface covered in 1995, this has increased the surface occupied by human activities by 127.7%.Gender, education, and pre-flood professional activity of the household head, combined with pre-flood house construction quality, property ownership, degree of house damage, and family member death, explained households' resettlement capacity.The Mann-Whitney U test results revealed that households' vulnerability and resilience to 2020 floods were influenced by their residence municipality (Kanvinvira and Mulongwe) and lifestyle (living in the host site or reintegrating into the community).Accordingly, household vulnerability components (i.e., exposure, susceptibility, and adaptive) were differentially impacted by municipality and household lifestyle.The commune of residence and lifestyle of Uvira households had a similar effect on socio-demographic, economic, physical, and institutional components.During and after floods, disaster victims received assistance from the government and stakeholders (churches, civil society, and non-governmental organizations) to mitigate flood effects, but they were unable to restore the lives of hundreds of flood victims.The study's findings suggest that provincial and local disaster management authorities, as well as stakeholders, should invest in preventive and sustainable flood risk management.Further measures are needed to increase the resilience of vulnerable communities by strengthening their social, physical, economic, and institutional capacities.Public awareness of flood prevention and mitigation should be increased through capacity-building training, preparedness, and sensitization.Finally, effective adaptation practices are required to reduce future potential damage.
Did women and young people play a specific role in the process of supporting flood victims in Uvira?Theme 3: Considering the consequences of the April 2020 floods, what has been done to prevent future flooding in Uvira?What motivates you to continue living in this city?
What are the resources/capacities/assets that allow you to survive despite the losses (shock, loss of savings) that the floods have caused at the: Individual, family, Community/Institutional, for women, Children, elders and people with disability Do you think the practices (activities) you have developed to survive will be sustainable?In your opinion, will the skills/practices you've developed improve your socio-economic situation in the long term or not?Why?
Theme 4: What measures and actions could you take to sustain your lifestyle?For family?For the community?
What exactly should be the institutional role (government, religions, stakeholders, family, etc)?

Figure 1 .
Figure 1.Overview of the study area.

Figure 2 .
Figure 2. Airbus image showing the affected areas and the 50m x 50m grid at Kanvimvira (A) and Mulongwe (B).

Table 1 .
Delimitation of the affected area and estimation of affected people.

Table 2 .
Description of selected images.

Table 3 .
Azanga et al (2016))aracteristics (%) of the flood survivor households.Household, SD = Standard deviation, *** Significant at 1% threshold, ** at 5% and * at 10% 1995 and 2021.Overall expansion rates ranged from −82.13 to 127.7% between the two years, and the area occupied by human activities increased by 127.7% compared to the initial area covered in 1995.Several studies conducted in DRC cities discovered that changes in land use and land cover (LULC) associated with climate change caused several disasters.For example,Mahamba et al (2022)discovered that the change in the LULC in Butembo City was overall characterized by the growth of bare land and buildings to the detriment of low vegetation (grassland).Between 1987 and 2011, the classes bare land and buildings (Overall change rate (Tv) = 61.33%)andwoodland(Tv = 34.2%)increased,whilegrassland (Tv = −39.5%)decreased.Similarly, EkakaAzanga et al (2016)discovered that anthropization is represented in land use, with farmland being the main vegetative cover in the Lake Tanganyika micro-watershed, followed by forest/tree plantations, grassland, and urban settlements.As a result of human activity, environmental consequences in Yangambi are worsening (Lateef et al 2016, Mawemba 2016, Holenu Mangenda et al 2020)ULC (the patch density of crop land, the edge density of bare soil and residential land, and the mean patch area of grassland have increased while the proportion of dense forest patches tends to decrease in the catchment areas)(Chishugi et al 2021).The uncontrolled urbanization of Bukavu(Muhaya et al 2022)and Kinshasa(Lateef et al 2016, Mawemba 2016, Holenu Mangenda et al 2020), combined with human activities, is responsible for pollution, accelerated soil erosion, flooding, and an unbalanced municipality, highlighting the geological constraints on socio-urban setting, sustainable urban planning, and human well-being.

Table 4 .
Kappa coefficient for classification.
pandemic.Overcrowding caused by jostling during rescue operations contributed to the spread of COVID-19 (Simonovic et al 2021, Reed et al 2022).Plisnier et al (

Table 5 .
Analysis of household vulnerability to flooding in Uvira in 2020.

Table 6 .
Analysis of Uvira affected households' resilience to the 2020 floods.Significant at 1% threshold, ** at 5% and * at 10% at Kasenga available to the victims.Tents had been set up to house victims, and believers had donated food and non-food items.The Red Cross, WFP, POPOLI, AVIS, CARITAS, OCHA, ACTED, UNICEF, HCR, IFRC, ADRA, ASED, NRC, CARE, Oxfam, Save the Children, AVREO, UNFPA, INTERSOS, and AFPDE provided emergency assistance to victims in the city of Uvira in the form of essential household and non-household items.