Estimating displacement risk using data-driven approaches
Flood and drought displacement risk in the Horn of Africa
Climate change is making weather-related events that trigger human displacement more frequent and more intense. It is also making communities more vulnerable and addressing the underlying drivers of displacement more urgent. In the Horn of Africa, countries are facing a future where extreme weather events, especially floods and droughts, are likely to occur more often. Without policy changes and investments today, these events could trigger millions of displacements in the region, from floods and droughts alone. Knowing where displacement might take place, and how many people are likely to be affected, can help policymakers minimise its negative impacts.
Too often, governments only assess and address displacement after it happens. This leads to responses that are largely limited to humanitarian, relief and protection interventions. If governments can combine these retrospective analyses with forward-looking models to anticipate displacement, they can identify new opportunities to prevent displacement and minimise its worst impacts. As part of the HABITABLE project, we developed new data-driven probabilistic models that forecast displacement risks under various climate scenarios. The findings from these models provide critical insights for policymakers as they work towards preventive measures that reduce displacement risks, rather than only reacting to crises after they occur.
Our new research indicates that, on average, 1.9 million people could be displaced in any given year by floods and drought under current climate conditions across Ethiopia, Somalia, and Sudan. Without measures to address climate change, the number of people at risk of displacement could double under a pessimistic scenario.
Read the report
Learn more about our data-driven probabilistic models for estimating flood and drought displacement risks under various climate scenarios.
Disaster displacement
Since 2008, floods have caused over 32 million displacements in Africa, making it the second most affected region globally. In Ethiopia, Somalia, and Sudan, floods alone have accounted for a quarter of these displacements, approximately 8.6 million. Additionally, droughts have caused nearly five million internal displacements worldwide since 2017, with Ethiopia and Somalia contributing to almost all of these displacements over the past six years.
Residential population distribution at the reference administrative level for the three countries.
Residential population distribution at the reference administrative level for the three countries.
Leaving one’s home can be the first of many further disruptions to people’s lives: they may be forced to move multiple times once they become displaced, and it can be months or even years before they can return home. Those who do return often face unsafe conditions and the prospect of becoming displaced again by the next disaster.
Internal displacements by disasters in Africa (2009-2023)
Internal displacements by disasters in Africa (2009-2023)
In some settings, people already displaced by disasters are then affected by conflict and violence, forcing them to move again, which undermines their resilience and puts their prospect to find solutions further out of reach. Conflict also pushes people towards areas exposed and vulnerable to disasters, including informal urban settlements and displacement camps which are often flooded, forcing IDPs to move again and prolonging their plight.
Background image: A man and boy wade through floodwaters in Beledweyne, Hirshabelle State, Somalia, after the Shabelle River broke its banks in May 2023. © IOM 2023/Claudia Rosel
The human impact
Droughts reduce agricultural yields, exacerbating food insecurity and economic hardship, while floods destroy or render homes uninhabitable and impact infrastructure in both urban and rural settings. Populations most at risk from drought include rural farmers, pastoralists, and those without the means to recover from repeated climate shocks. Floods, meanwhile, can pose a risk to anyone living near rivers, including people who have fled conflict and violence and are residing in IDP settlements.
Displacement compounds existing vulnerabilities, with displaced populations often facing long-term challenges such as the loss of livelihoods, food shortages, and limited access to healthcare. Social networks, which traditionally provide support, are often fractured, leaving displaced individuals and families isolated. Women, children, and the elderly are particularly affected, with fewer resources and greater barriers to accessing aid. Addressing the human impacts of displacement requires not only immediate relief but also long-term solutions that enable these populations to rebuild their lives and adapt to evolving climate conditions. Introducing novel displacement risk approaches that include data on population, livelihoods and other underlying factors that contribute to a community's vulnerability, or resilience, to potential displacement are critical for both immediate and long-term efforts to address displacement.
Background image: Gedaref, East Sudan, 2023. A woman walks through a flooded IDP shelter after rainy season in September. © Ahmed Omer/NRC
Results
The findings reveal a complex interplay of risks across Ethiopia, Somalia, and Sudan, with drought and flood displacement hazards showing contrasting trends under optimistic and pessimistic scenarios. In most cases, the results highlight an anticipated increase in displacement risk, emphasising the pressing need for adaptive measures to mitigate future impacts.
|
Displacement risk metrics |
|
|---|---|
|
Average Annual Displacement (AAD) is a measure of the magnitude of future displacement by hazard type that a country is likely to experience. It does not reflect the number of displacements it will face each year, but the number it can expect per year considering all the events that could occur over a long timeframe. AAD is a compact metric with low sensitivity to uncertainty. |
Probable Maximum Displacement (PMD) is the maximum displacement expected within a given time period. It answers the question: What is the maximum expected displacement within a range of X years? It represents the outlier event that could occur during a specific time frame. PMD can be used to determine the size of shelters and other assets that a government needs to provide to cope with the potential magnitude of displacement. |
|
Climate scenarios |
|
To explore the potential impact of climate change on the forthcoming frequency and intensity of extreme events, we compared 15 models from the Inter-Sectoral Impact Model Intercomparison Project (ISIMIP3b) from which we have selected two: “Optimistic”-the scenario closest to the 20th percentile, corresponding to an average temperature rise of about +1°C by 2100 “Pessimistic”-the scenario closest to the 80th percentile, corresponding to an average temperature rise of over +5°C by 2100 We did not consider changes in exposure and vulnerability between current and future climate conditions, but it is worth noting that factors such as population growth and distribution – for example, the rapid urban sprawl that shrinks natural areas available to absorb floodwater, as deforestation and agriculture change - may greatly alter the future “riskscape”. |
Flood displacement risk
The findings reveal a complex interplay of risks across Ethiopia, Somalia, and Sudan, with drought and flood displacement hazards showing contrasting trends under optimistic and pessimistic scenarios. In most cases, the results highlight an anticipated increase in displacement risk, emphasising the pressing need for adaptive measures to mitigate future impacts.
The results show that flood displacement risk may increase under these optimistic scenarios, with the Average Annual Displacement (AAD) two to four times higher than current conditions, depending on the country. Under pessimistic scenarios, the risk might increase by a factor of three for Somalia, four for Ethiopia and even nine for Sudan.
Flood displacement risk in the Horn of Africa: In any given year in the future, on average, the number of er of people that could be displaced by floods under different climate scenarios
Flood displacement risk in the Horn of Africa: In any given year in the future, on average, the number of er of people that could be displaced by floods under different climate scenarios
Through this research, we have examined displacement risk from a new perspective, analysing how water depth values impact not only residential buildings and populations but also industrial and service buildings, affecting workers in these sectors. Additionally, we considered the effects on crop and grazing areas, which are critical for those employed in agriculture and can act as a push factor for internal displacement.
While the lack of services may not be the primary driver of displacement after rapid-onset events, the combined effects of housing loss, job loss, or both—alongside a general lack of services in the affected area—can exacerbate the situation. Regions with high displacement rates due to uninhabitable homes or disrupted livelihoods often also experience significant disruptions in access to education, particularly in flood-affected areas. This lack of access can increase vulnerability, ultimately heightening the likelihood of displacement.
Various factors that exacerbate displacement risk: Probable Maximum Displacement (PMD) for a 100-year return period in Somalia disaggregated by cause of displacement under different climate scenarios.
Various factors that exacerbate displacement risk: Probable Maximum Displacement (PMD) for a 100-year return period in Somalia disaggregated by cause of displacement under different climate scenarios.
Lastly for riverine floods, for Sudan we have explored how an 100 years return period flood could impact both formal and informal IDP settlements. Given Sudan's overlapping crises of conflict and disasters, the analysis sheds light on how future floods may exacerbate displacement, especially for IDPs in vulnerable shelters or overcrowded conditions.
The findings, based on IOM-DTM’s Weekly Displacement Snapshot (8 September 2023), show potential impacts, such as in the River Nile Region where 43% of IDP settlements face severe flood damage under an optimistic scenario, escalating to 80% under a pessimistic one.
These figures might underestimate risks, as they rely on point-based data and do not account for the spatial area needed for adequate shelter. Nevertheless, the insights emphasise the urgency of preparedness to mitigate future climate impacts on IDPs.
IDP settlements at risk of riverine floods in Sudan
IDP settlements at risk of riverine floods in Sudan
Drought displacement risk
The results indicate that in Somalia, drought displacement risk may increase slightly under both optimistic and pessimistic scenarios, by 22 per cent and 54 per cent respectively, in comparison with current conditions. Conversely, in Sudan, there might be a decrease in risk by 5 per cent to 9 per cent under both scenarios. Ethiopia exemplifies the East African Climate Paradox even more starkly, with the risk of drought displacement potentially doubling under the optimistic scenario, while dropping to one-sixth of current levels under the pessimistic scenario.
Although Sudan lacks empirical data on drought-induced internal displacement, millions of Sudanese are regularly affected by drought episodes. The results highlight that between 800,000 and 900,000 people could be displaced in any given year when drought severely impacts the region.
Drought displacement risk in the Horn of Africa: In any given year in the future, on average, the number of people that could be displaced: by droughts under different climate scenarios.
Drought displacement risk in the Horn of Africa: In any given year in the future, on average, the number of people that could be displaced: by droughts under different climate scenarios.
While the AAD provides an average estimate of expected displacements from drought events, it may obscure potential outliers. To address this, the publication also presents PMD Probable Maximum Displacement (PMD) curves for current and projected conditions, illustrating the number of people potentially displaced during frequent (low return period) and rare (high return period) events.
At the country level, the PMD results show that under optimistic conditions, approximately 4.5% of Somalia’s population could be displaced during a 5-year event—around 800,000 people.
PMD curves for Somalia. Displacements, as a function of return period of the event, are expressed as a percentage of total population. The continuous black line represents the current climate scenario, while the continuous red line and the red dotted line depict optimistic and pessimistic climate scenarios respectively
PMD curves for Somalia. Displacements, as a function of return period of the event, are expressed as a percentage of total population. The continuous black line represents the current climate scenario, while the continuous red line and the red dotted line depict optimistic and pessimistic climate scenarios respectively
Estimating displacement
Probabilistic risk assessments (PRA): a new approach to forecasting displacement
Our probabilistic risk assessments (PRA) models estimate the likelihood and potential impacts of different displacement scenarios caused by floods and droughts. This approach integrates data from climate projections, historical data, and machine learning models to provide a nuanced understanding of displacement risks under various climate conditions. They also incorporate exposure data such as population, and country-specific building typology, usage, and value in both urban and rural settings. This approach allows policymakers and humanitarian actors to better anticipate where and when displacement is likely to occur and allocate resources accordingly.
The displacement risk equation
Maximum and annual displacement metrics
Our models provide two key metrics, the Probable Maximum Displacement (PMD) and Average Annual Displaecment (AAD). PMD estimates the maximum number of people who may be displaced by extreme events over a specified period of time, while AAD offers an annual average based on potential events. Our models demonstrate that, under both optimistic and pessimistic climate scenarios, displacement in the Horn of Africa could increase significantly in the coming years. Governments can use the information from this data-driven approach to inform national disaster risk reduction strategies that scale to meet future challenges.
Two hazards: floods and droughts
We chose to focus on floods and droughts because these are the hazards that trigger the most disaster displacement in the Horn of Africa.
Flood modeling: this involves analysing hydrological patterns, topography, and human exposure to assess the potential impact of floods on displacement.
Drought modeling: this is done using machine learning models (e.g., decision trees) to identify which drought factors lead to displacement and how they may evolve in the future.
Estimating exposure
While hazards are one part of the equation, the people affected are another. Population data constitutes the exposure layer, a fundamental component in risk modelling. Using existing exposure database of the Global Infrastructure Risk Model and Resilience Index (GIRI) project on exposure that combines population data with country-specific building typology, usage, and value in both urban and rural settings.
Additionally, for drought, we considered a wide range of hydrometeorological indicators to capture the various ways drought impacts can develop in a region. These indices serve as proxies to indicate abnormalities in water availability within the hydrological cycle, including the atmosphere (meteorological drought indices), the hydrological system (hydrological drought indices), and the soil (agricultural drought indices).
Vulnerability
For riverine floods, our results are based on a physical vulnerability analysis. Physical damage (as a percentage) is directly derived from vulnerability curves, which relate the hazard intensity to the expected damage. In the displacement risk evaluation procedure, concerning the home and workplace for services and industries, we adopted a specific damage threshold, or loss, that could cause displacement
For drought displacement, vulnerability is statistically derived by analyzing the relationship between specific drought drivers (combinations of hydrometeorological indicators and observed impacts) and displacement. This relationship is calibrated using observational data from Somalia, where comprehensive data has been collected over time under varying weather patterns.
Read the report
Learn more about our data-driven probabilistic models for estimating flood and drought displacement risks under various climate scenarios.
Policy implications
Preventing and managing future displacement
The insights from probabilistic risk models underscore the urgent need for proactive policy interventions to prevent and manage future displacement driven by natural hazards in a changing climate. They also provide insights into how and where that displacement may occur. To minimse the risk of displacement and its most adverse impacts, governments in Ethiopia, Somalia, and Sudan can use this information to move beyond reactive responses to displacement events and adopt forward-thinking strategies that anticipate risks and build resilience in vulnerable communities.
Key policy actions include improving preparation and response to floods, such as through investments in resilient infrastructure, drought-resistant agricultural systems and early warning systems. Governments must also ensure that emergency shelters and evacuation centers are equipped to handle the scale of displacement expected under future climate scenarios. Investment in social safety nets, reforestation efforts, and diversified livelihoods will be crucial in addressing the impacts of weather related displacement in a changing climage. International collaboration and financial support from donors will be vital to implementing these strategies and ensuring long-term stability.
Background image: Emergency drought response project in 2022 at NRC's water distribution site in Baidoa, Somalia. © Abdulkadir Mohamed/NRC
Global relevance
A model for other regions
While our analysis focuses on Ethiopia, Somalia, and Sudan, its findings have global relevance. Disaster displacement is a growing issue worldwide, affecting communities from coastal cities threatened by sea-level rise to arid regions facing severe droughts. The probabilistic risk models used in this study provide a scalable approach that can be adapted to estimate displacement risks in other regions.
By adopting and building on these models, governments can shift from reactive crisis management to proactive disaster planning, reducing the long-term human and economic costs of displacement. The lessons learned from the Horn of Africa can inform global discussions on climate adaptation and displacement policies, highlighting the need for international cooperation to mitigate displacement risks and protect vulnerable populations worldwide.
Background image: Flooding in Beledweyne, Hirshabelle State, Somalia. Heavy downpours and floods triggered around 97,000 displacements in the south of the country in April and May 2023. © IOM 2023/Claudia Rosel
Conclusion
The growing risk of disaster displacement in the Horn of Africa demands urgent, proactive action. As extreme weather events become more frequent and intense, Ethiopia, Somalia, and Sudan will face increasingly severe challenges, particularly in regions reliant on agriculture and fragile infrastructure. These impacts go beyond immediate displacement, leading to long-term disruptions to livelihoods, food security, and social stability.
Probabilistic risk models offer a critical tool for policymakers to anticipate future displacement events and inform targeted interventions that build resilience before disasters strike. Investments in infrastructure, early warning systems, and social safety nets will be essential in preventing mass displacement and mitigating its impacts on vulnerable populations.
These models in the Horn of Africa highlight the potential for use in other regions. Governments, researchers, and international organizations can adopt this approach to improve disaster risk assessments and develop comprehensive, data-driven climate adaptation strategies. Global collaboration is key to addressing the challenges posed by disaster displacement, ensuring that the world’s most vulnerable populations are protected from the increasing threat of displacement in the years to come.
Contact us at info@idmc.ch to learn more about applying this model in your country or region.
Background image: A young boy trudges through deep mud and river water on his way to school after his village in Ethiopia was flooded by heavy rains. © Muse Mohammed / IOM
Acknowledgments
The HABITABLE project focuses on exploring the interconnections between climate change and migration/displacement patterns. This particular report emphasizes the significant risks posed by floods and droughts in Ethiopia, Somalia, and Sudan, which are critical for both current and future displacement scenarios.
We would like to thank our partners and contributors:
© Internal Displacement Monitoring Centre, Geneva, Switzerland.
Download the full report here.
Subscribe to our newsletter to get the latest data and analysis delivered to your inbox.
For any questions about how to use the model or the sources used for this piece, reach out to us at info@idmc.ch
Built with Shorthand