Flood risk and vulnerability of populations’ analysis in Southern, Burkina Faso

Abstract

This Thesis seeks to understand the space-time variations of extreme rainfall events, and the assess flood risk and community vulnerability to floods in the southern Burkina Faso. To achieve this, the specific objectives include (i) analyzing the frequency and intensity of extreme rainfall for better insight of risks associated with flood disaster occurrences in communities, (ii) examining flood-related impacts as much as possible for better awareness and preparedness of flood disaster occurrences, (iii) analyzing flood risk, including perceptions of flood risk and their influencing factors, to increase the resilience of communities that may face flooding situation, and (iv) assessing flood vulnerability using various weighting methods under the definitions of IPCC 2007 and 2014 vulnerability frameworks. As far as the analytical process is concerned, various methods such as extreme value theory modelling, econometric modelling, indicator-based analysis, and statistical tests are performed, integrating both secondary data and household survey data. The results show substantial spatial heterogeneity in rainfall trends across the study area, with more downward than upward trends, particularly for indices representing extreme rainfall. This contrasts with national and supranational studies that primarily identified increasing trends in extreme rainfall. Furthermore, the findings demonstrate a strong spatial dependency in extreme rainfall, reaching up to 300 km away and decreasing as the distance between paired sites increases, potentially impacting thousands of people simultaneously. Regarding flood risk, our findings indicate that 12 out of the 14 villages in our sample group have experienced high levels of flood risk. Households pointed out the management of runoff from the nearest urban areas and poorly designed civil engineering infrastructure, such as roads, as significant factors contributing to their increased vulnerability. Furthermore, the results suggest that flood risk perception is largely influenced by informational and behavioural factors among households. The sensitivity analysis of vulnerability to different weighting methods shows that the vulnerability index is less sensitive to whichever equal and Principal Component Analysis weighting methods are considered for assigning weights to indicators. However, the equal weighting method is preferred when dealing with numerous indicators. This thesis could offer valuable insights for disaster risk management authorities, enabling them to allocate resources effectively, plan budgets, and coordinate their actions based on the priorities of areas.