Regional probabilistic flood displacement risk assessment: the Horn of Africa case study

The risk of displacement caused by natural hazards has been increasingly impactful and emerges as a topical issue point in the field of disaster risk management. Given the potential escalation of this phenomenon due to climate change, population growth and urbanization, enhancing displacement risk assessment through reliable models and data has become increasingly crucial. Different applications require approaches that can be adapted at different spatial scales, from local to global scale. In pursuit of this goal, we have devised a probabilistic procedure for estimating the potential displacement of individuals due to riverine floods. The methodology is based on a novel approach to vulnerability assessment which considers that people’s vulnerability depends on several physical and social factors such as direct impacts on houses, livelihoods and critical facilities (such as schools and hospitals). These concepts are seamlessly woven into a comprehensive probabilistic risk assessment. A modelling chain that incorporates climatic, hydrological, and hydraulic and exposure/vulnerability models can be run different resolution to predict impacts at different special scales, from local to global scale.

This approach already demonstrated its validity for in Fiji and Vanuatu, where the small size of the countries allows for the definition of a building scale exposure model. In the present study, our focus turns towards adjusting the methodology for large countries, where using a high-resolution exposure model becomes impractical.

For our case study, we selected three countries in the Horn of Africa—Ethiopia, Somalia, and Sudan—acknowledging their particular vulnerability to the challenges posed by recurrent floods and the resulting internal displacement.

To properly match the 90m resolution of riverine flood hazard maps and avoid distortions in the final risk computations, a specific procedure for downscaling global exposure dataset, such as the 1-km resolution Global Exposure Socio-Economic and Building Layer (GESEBL), was implemented using high-resolution population distribution products. The resulting exposure layers are a set of population distributions associated to different sectorial assets (residential, industrial and agricultural production, services), characterized in terms of physical vulnerability to floods.

Impacts of current and future flood scenarios on those assets may render them unable to provide their function, thus causing people to forcedly move. In this procedure we took special care to avoid double counting, i.e. those cases where people lose both habitual place of residence and livelihoods.

Displacement risk expressed in annual average displacement and probable maximum displacement was evaluated under current and future climate conditions with optimistic and pessimistic scenarios. The results indicate a potential 2 to 4 times increase in average annual displacement for optimistic scenarios compared to current conditions, with even higher risk for pessimistic scenarios.

The application of this methodology in larger countries paves the way for its implementation on a global scale.