Global changes in extreme rainfall patterns are increasing the frequency and intensity of flood events. Populations in low-income countries are disproportionately affected, in part due to rapid, often uncontrolled urban expansion in flood-prone zones. Flood events can impact drinking water services directly (e.g., causing infrastructure damage) or indirectly (e.g., through power supply interruptions). Floods can also create an environment that favours infectious disease transmission and outbreaks and, in the long term, may lead to development setbacks. Especially in fragile contexts, where data collection can be challenging and infectious disease burdens high, there has been limited research on urban water infrastructure vulnerabilities during flood events. Geospatial information on flood-related risks to water services is needed to better anticipate and manage these risks.

Objectives

Our study aims to map the extent of various flood events and to identify vulnerabilities of the piped water supply service and users in the fragile, urban context of Uvira (South Kivu, Democratic Republic of the Congo). In order to achieve this goal, we combine various data sources – satellite images, infrastructure plans and utility data, and household survey data. A secondary objective of this study is to confirm the feasibility of using open-access satellite images for mapping floods in a dense, data-scarce urban setting. 

Study site

Uvira is a town of approximately 280,000 inhabitants (in 2020) in South Kivu, a region that has been experiencing protracted conflict for about three decades. Access to basic services including drinking water and sanitation is limited in Uvira, and the burden of cholera and diarrhoeal diseases remains high. The town is situated on a hilly terrain along the shore of Lake Tanganyika and crossed by five rivers that flow into the lake. Floods are frequent, especially during the rainy season. In April 2020, the town experienced an extreme flood event which affected around 80,000 people.

Methods

We use three complementary approaches to study flood events and related drinking water service vulnerabilities in Uvira:

1. We mapped the extent of three flood events (April 2020, December 2020, March 2024) using high-resolution, commercially available satellite images as well as open-access, satellite images (optical and radar images) – and compared these approaches;
2. We overlaid flood extent maps based on satellite images with plans of the water supply infrastructure to identify exposures and potential vulnerabilities, and analysed operational records of the water utility showing pumping hours per day and volumes of water supplied;  
3. We carried out a survey-based assessment of 148 households 12 weeks after the April 2020 flood, focused on drinking water-related practices.

Preliminary findings

Mapping and analysing household survey data suggest that nine out of fourteen districts in Uvira were exposed to flooding in April 2020, mostly in the vicinity of rivers.

By overlaying flood extent maps with plans of the piped water supply infrastructure (Graph 1), we identified critical points of the system exposed to flooding during the event. These included the main water intake, treatment plant and pumping station, as well as 20% of community taps. Information from the utility confirmed that the main water intake for the water supply network was destroyed, leading to a six-week disruption of services and longer-term limitations in the water production capacity.  

Almost half (45%) of household survey participants reported to have changed their main drinking water source due to the flood event in April 2020. However, at the time of the survey, two and half months after the event, the water sources used were similar to those reported to have been used before the event (Graph 2). This suggests limited long-term impacts of the extreme flood event. Despite regular interruptions of water services, storage capacities within households were modest at the time of the survey (22L per person) and household drinking water treatment practices uncommon (15% of households).

Analysis is in progress to map the extent of smaller flood events recorded in December 2020 and April 2024, and validate the use of open-access images for flood mapping in the dense urban environment of Uvira.