Assessing disaster vulnerability is crucial for understanding how natural hazards affect cities and how cities can be supported in disaster risk management. Yet, global-scale urban-risk assessments that explicitly incorporate vulnerability patterns remain limited. Composite indicators are a common tool to assess and compare vulnerability across places. However, data (un)availability is often identified as an obstacle to these analyses. In this study, we therefore aim to: (1) provide a structured overview of large-scale data for urban vulnerability assessments; and (2) assess the quality and relevance of these data for vulnerability indicators. We use an existing overview of urban vulnerability drivers (VulneraCity) and translate these drivers into measurable indicators. Concurrently, we review open-source, large-scale urban datasets of vulnerability-relevant data. Corresponding indicators and data are then matched via a semi-automated semantic similarity process, involving a large language model and extensive manual checks. The result is VC-InDat, a comprehensive overview of 1,458 vulnerability drivers, 1,959 indicators, and 1,288 data variables. VC-InDat can be used to scope urban vulnerability drivers and data, and to find which drivers of vulnerability are represented by certain data or indicators. We find 4,291 combinations of vulnerability indicators and available data. The data cover 62% of the drivers, albeit with varying quality. Also, direct quantifiable indicators are more common for certain vulnerability sub-dimensions (Economic, Demographic, and General Urban Assets), compared to others (Awareness & Information, Institution, and Governance). Based on our analysis, we identify spatiotemporal coverage and data harmonization among the most salient challenges in developing global urban vulnerability data.

We evaluate the performance of the Super-Fast INundation of CoastS (SFINCS) hydrodynamic model for simulating riverine floods, combined with a fully automated open-source data preprocessing pipeline. To do this, we assessed the simulated extent of 499 historic flood events against the satellite derived flood extents using the Critical Success Index (CSI) as a performance metric. We utilised simulated discharges from the Global Flood Awareness System (GloFAS) hydrological model and found that SFINCS performance improved with upstream basin size, with a global mean CSI of 0.42 for basins with large upstream area (>1000 km2) and a CSI of 0.29 for basins with small upstream area (<50 km2). Our results illustrate the importance of accurate discharge data input to flood hazard simulations. When the (globally simulated) GloFAS data is replaced with observed discharge data for ten events in the US, the CSI improved from 0.39 to 0.67. These results suggest that global hydrological model performance limits the accuracy of the flood hazard simulations. Our findings also showed a significant improvement in the CSI (from 0.37 to 0.57) when changing to a higher-resolution elevation input by contrasting a ∼1 m digital elevation model (DEM; 3DEP) with our default ∼30 m global DEM (FABDEM) in six US events. Sensitivity analysis of bathymetric calculations revealed a systematic underestimation of the default 2-year return period estimated by GloFAS discharge, likely driven by underrepresentation of annual block maxima, which resulted in underestimated channel dimensions. All of these factors resulted in a loss of detail, which impacted model performance, especially in smaller headwater rivers. We recommend to improve the estimation of bathymetry, for instance by employing the “gradually varying solver” method or using data from the SWOT mission. Furthermore, incorporating additional validation data which ideally includes flood depth measurements can largely enhance our understanding of the model performance.

Agricultural insurance is promoted as a drought-risk tool, yet its net long-term socio-hydrological impacts and interactions with other adaptations remain disputed. We expand the Geographical, Environmental and Behavioural model (GEB), a fully distributed hydrological model coupled with an agent-based model (ABM), a process-based crop model and a dynamic farmer adaptation behavior model. We add two adaptation options (wells, crop switching) and two insurance designs (traditional, index), calibrated to an Indian basin. Traditional insurance increases well adoption and profits but creates a lock-in to wells and higher-water-use crops, leading to 20-50% higher annual water use and 30-60% lower groundwater levels. Index insurance avoids this lock-in, shifts production toward lower-water options and delivers higher profits with lower basin-wide water use. Despite this, traditional insurance sustains greater crop diversity and a more diffuse irrigation mix via groundwater, reducing drought risk: profit variability and losses during consecutive droughts are smaller than under index insurance (similar to 0.039 vs similar to 0.085 USD m(-2); similar to 20% vs similar to 28%). Spatial patterns further show that insurance interacts with reservoir effects: uptake is lower in surface-water command areas, whereas index insurance has relatively high uptake in these zones, suggesting potential to counteract reservoir effects. Finally, we find that the level of available irrigation, rather than simple access, determines whether reservoir effects emerge. Our findings highlight design trade-offs: while hydrological and economic metrics favor index insurance, a risk-oriented perspective may prefer traditional insurance, underscoring the utility of ABMs to make these trade-offs explicit.

In this paper, we analyze the effect of natural hazards on migration in the United States (US) and the importance of spatial dependence in such assessments. We use two measures of migration: migration rates and flows. The model for migration flows is estimated using the gravity model, whereas out- and in-migration rates are analyzed using the spatial Durbin model. Our results indicate there is a major and significant impact of economic damage caused by natural hazards on out-migration rates and outward migration flows. In the spatial Durbin model and in the gravity model, a $1,000 dollar damage per capita is associated with an increase in out-migration of 16.0% and 9.1%, respectively. However, when spatial dependence is not accounted for, the effect of natural hazards on migration is substantially overestimated: the coefficients are 1.5-2 times larger when spatial dependence is not considered.

In 2019, scientists from African countries called for more research on drought and better drought forecasting and management (Padma, 2019). Between 2020 and 2023, the Horn of Africa had experienced the worst drought in 40 years, with severe consequences related to reduced agricultural productivity and high food prices (Okoth, 2024). In this presentation, we will showcase the drought research done within the DOWN2EARTH project with a case study in Kenya.Agro-pastoral livelihoods in the Horn of Africa (HoA) are acutely exposed to climate variability due to the predominance of rain-fed systems. Yet drought risk emerges from more than rainfall deficits—it reflects interacting biophysical processes, socio-economic vulnerability, and institutional response capacity. We advance an integrated, impact-based and adaptation-informed framework by combining statistical risk modelling across Kenya’s arid and semi-arid lands (ASALs) with a coupled socio-hydrological and agent-based simulation of human–water interactions.First, using Spearman correlations and Random Forest regression, we link drought hazards to observed societal impacts and identify distinct timescale sensitivities: short (2–6 months) precipitation deficits align with increased household water trekking distances, while medium-to-long drought indices (5–24 months) better explain declines in milk production and increases in malnutrition. Clustering counties by vulnerability profiles improves predictive skill. Socio-economic clustering best captures water access outcomes, whereas environmental clustering better explains agricultural and nutrition impacts. Extending to probabilistic risk via Random Forest hindcasts (1984–2014) yields Average Annual Loss (AAL) and Probable Maximum Loss (PML) estimates, highlighting spatial heterogeneity: high water-access risk in northwestern Kenya and elevated livestock, milk, and malnutrition risk in eastern and southeastern counties. Priority adaptation pathways include sanitation and safe water access, poverty reduction, and small-scale water infrastructure.Second, the ADOPT‑AP framework couples the DRYP hydrological model with a behavioural agent model to simulate bounded-rational adaptation and policy scenarios. Sensitivity analysis identifies irrigation abstraction as the dominant driver of both drought hazard and adaptation uptake. Replacing upstream commercial farms with communities or forests increases downstream streamflow and groundwater, modestly improving water access and production in drought years. During the 2020–2023 drought, doubling extension access marginally boosts low-cost measure adoption but not capital-intensive options, underscoring finance constraints; scaling water harvesting improves milk and reduces water trekking but has mixed crop effects and downstream hydrological trade-offs.Together, these results demonstrate how vulnerability-informed, spatially targeted interventions and dynamic adaptation modelling can be used to strengthen early warning, guide equitable water governance, and build long-term resilience. However, improving drought management requires more than research. Early warnings are for example often not acted upon because of cultural values or limited resources. We therefore advocate for more transdisciplinary research, co-creation of drought adaptation solutions, and strengthening connections between communities and formal governance actors.ReferencesPadma, T. V. (2019). African nations push UN to improve drought research. Nature, 573(7774).Okoth, D. (2024). The cost of African drought. Nature Africa, doi.org/10.1038/d44148-024-00075-0.

Study RegionThe 21 arid and semi-arid land (ASAL) counties of Kenya.Study FocusWe quantify county-level drought risk for impacts on household and livestock trekking distance to water, milk production, and child malnutrition. Building on previously developed vulnerability-clustered Random Forest impact models, we select models with mean absolute error (MAE) ≤ 0.34, hindcast monthly impacts over 1984–2022, form county ensembles from models with observed-period Spearman rₛ ≥ 0.5, and derive average annual loss (AAL) and empirical probable maximum loss (PML) curves for 2–30-year return periods.New Hydrological Insights for the RegionDrought risk is strongly spatially differentiated. Household water-access risk is highest in northwestern counties, whereas livestock water-access and malnutrition risks are higher in eastern and southeastern counties; milk-production losses are larger in central and southern counties. Aridity and access to improved sanitation distinguish several county risk profiles, although these associations do not imply direct causation. PML curves reveal response patterns that AAL alone does not: some counties experience persistently high baseline impacts with limited additional deterioration during rarer droughts, whereas others show sharp increases with return period. These patterns distinguish persistent long-lasting water and food-system stress from sensitivity to hydrological extremes and support impact-specific drought-risk management.

To reduce current and future coastal flood risk, it is critical to better understand how adaptation measures, including nature-based solutions, can reduce that risk. Globally, hybrid coastal defenses, including a combination of coastal vegetation, such as salt marshes and mangroves, with a dike or sea wall, have been highlighted as a promising adaptation measure. Here, we present a global-scale assessment of the potential risk reduction from mangrove restoration in combination with foreshore dike systems under scenarios of climate and socioeconomic change. We provide a quantitative assessment of the benefits in terms of reduced economic damage, exposed population, and poverty exposure. We evaluate mangrove restoration fronting dikes by accounting for wave-vegetation interaction. If mangrove foreshore dike systems were established along coastlines susceptible to flooding, restoration could potentially reduce expected annual damage by US$800 million and reduce expected affected population by 140,000 annually. These values increase under future projections. Our benefit-cost analysis finds mangrove restoration economically viable for about half of the subnational regions assessed (85 to 105 out of 208). At the global scale, the benefit-cost ratio under future conditions ranges from 3 to 6, with a net present value between US$44 billion and US$125 billion. Because absolute risk values and benefit-cost analysis do not differentiate between relative wealth impacts, we also estimated restoration impacts across different wealth levels. We show that restoring mangroves disproportionately benefits people with lower incomes, as they are often more exposed to coastal flooding and located in areas suitable for mangrove restoration. As such, mangrove restoration in low- and middle-income countries could contribute to the resilience of people in poverty.

Improving irrigation efficiency (IE) is widely promoted as a strategy to reduce agricultural water use while sustaining farm production and freeing water for other sectors. However, higher field-scale IE does not necessarily increase basin-scale water availability, a mechanism known as the IE Paradox. The paradox occurs because (1) higher IE reduces runoff and recharge that would otherwise support downstream users and environmental flows, while (2) initial withdrawal reductions may rebound if farmers use apparent 'savings' for economic gain. Here, we develop a conceptual framework for this paradox and quantify its components with the Geographical, Environmental and Behavioral model (GEB), which couples a 30 arcsec hourly distributed hydrological model, a 1.5 arcsec 'one-to-one' agent-based model, a process-based crop model and a farmer adaptation model. We apply GEB to the southern Murray-Darling Basin, Australia (similar to 75 000 km2) and compare scenarios with historical IE increase, static IE, high IE and reduced irrigation entitlements. Historical IE increased basin-wide consumed irrigation water by 17% (1.1% per percentage-point IE increase), reduced recoverable irrigation water by 25% (-1.6% per percentage point) and reduced outlet discharge by 1% (-0.03% per percentage point). Although the IE-only effect reduced withdrawals by 13%, rebound largely offset this effect, leaving a net withdrawal reduction of only 2.5% (-0.08% per percentage point). Higher IE also increased farm income, especially in dry years, because additional water availability produced nonlinear yield gains under scarcity. In contrast, reduced entitlements lowered withdrawals and consumed water and increased discharge, but caused income losses, particularly in dry years. These results show that water conservation technologies (WCTs) and entitlement reductions serve different policy goals. WCTs can support farm income and drought resilience, whereas entitlement reductions are more effective for basin-scale water recovery. Field-scale IE gains should therefore not be presented as basin-scale water savings or streamflow-recovery policy without accounting for rebound, reduced return flows, groundwater impacts and economic trade-offs.

Abstract Floods are expected to increase in frequency and severity due to climate change. Recent floods have shown that many catchments worldwide are vulnerable to floods, highlighting the need for additional adaptation measures. This study extends the Geographical, Environmental, and Behavioral (GEB) model by coupling it to a hydrodynamic and a flood risk model to assess the effects of dry‐proofing, wet‐proofing, retention ponds, reforestation, and the creation of natural grassland. A key innovation is the integration of all local‐scale models, thereby allowing for a catchment‐wide assessment of the impacts of various measures on interlinked hydrological conditions, flood extents and depths, damages, and risk. We apply our method to the Geul catchment (shared between the Netherlands, Belgium and Germany), which was heavily flooded in July 2021. Our results show that reforestation and creation of natural grassland (both 10 km2) reduce flood extent by 12% and average water depth by 10%. Damage is decreased up to 38%. Larger retention ponds (1 m deeper) have a much smaller reduction in flood extent (3%), depth (0.5%) and damage (1.6%), due to limited storage capacity compared to excess rainfall. The building‐level adaptation scenarios outperform all nature‐based solutions, with dry‐proofing reducing more damage (up to 95%) than wet‐proofing (around 55%). A cost‐benefit analysis shows that several adaptation measures are economically attractive. Overall, our findings show a coupled model is essential for comparing the relative effectiveness of different flood adaptation measures and supporting informed risk management decisions. The open‐source model is transferable to other catchments worldwide to guide decision‐making.

This paper assesses the effectiveness of Impact-based Flood forecasts (IbF) in reducing flood impacts. The new system is developed in the Geographical, Environmental, and Behavioural (GEB-IbF) model framework, where it is applied to the April–May 2024 flood in the Sinos River catchment, southern Brazil. The system correctly issued evacuation warnings to 88,883 households (76.1% of all flooded households) and recommended protective measures that reduced estimated flood damages by $4–5 million. Comparing conventional meteorological warnings (MW) and general flood warnings (FW) with impact-based warnings (IbW) demonstrates that IbW provide timely, useful, and actionable information. IbW enhances a flood early warning system in four main ways: (I) IbW reach a larger number of administrative units (593-708 for IbW vs. 580-682 for FW), especially upstream and rural areas; (II) IbW reduce falsely warned households by 19.9% (in medium flood scenarios) and 33.5% (in severe flood scenarios) compared to FW; (III) IbW prompt more affected households to act, with 4–11% more households correctly evacuating and implementing protective actions; and (IV) IbW achieve greater damage reduction (25.2-30.6% of total damage) than FW (21.9-21.6%), under the medium and severe flood scenarios, respectively. These results highlight the need for investment in operational flood forecasting and continuous gathering of impact-related data. Besides, this study shows that improved human response to warnings can double avoided damage and evacuations, emphasizing the urgency for continuous training and awareness raising among the local population. Therefore, local and community authorities are key actors in further increasing the warning effectiveness.

In the past decades, and notably the last few years, droughts have severely impacted various interconnected socio-economic sectors and ecosystems across the EU. These impacts encompass, among others, extensive losses in both rain-fed and irrigated agriculture, challenges and constraints in public water supply, disruptions in inland shipping, diminished production of hydropower and thermoelectric energy, impaired functioning of terrestrial and freshwater ecosystems, and implications for the tourism industry. In order to better prepare for future drought events in Europe, knowledge on the drivers, spatial patterns and dynamics of drought risks is urgently needed. The European Drought Risk Atlas responds to that need by mapping hotspots and risk drivers across diverse systems and regions within the EU. Combining conceptual risk models (impact chains) and a data-driven quantitative drought risk assessment based on machine learning, this Atlas represents a significant stride toward impact-driven drought risk analysis in present and projected global warming levels (+1.5°C, +2.0°C, +3.0°C). It provides a detailed and disaggregated perspective on the risks posed by droughts to societies and ecosystems, with a particular focus on agriculture, public water supply, energy, river transportation, freshwater, and terrestrial ecosystems.The data-driven analysis reveals that current levels of drought risk in the EU are already notable, with average annual losses presenting economic and environmental threats in nearly all regions. As expected, the Mediterranean region, particularly the Iberian Peninsula, faces high drought risk under both current and projected climate conditions, driven by the escalating dry conditions associated with global warming. However, while drought risk of certain sectors in Europe follows a north-south gradient of overall mean drying (south) and wetting (north) under climate change, the analysis underscores that each sector reacts distinctly to current and projected hazard conditions, exhibiting sector-specific sensitivity. Eastern and Western Europe may experience complex dynamics due to the interplay between drying and wetting patterns and precipitation variability, resulting in different risk conditions depending on the considered sector. While the analysis may still be refined as new data (observations and future climate simulations) become available, this Atlas represents a unique tool of unparalleled value that can shape future EU preparedness and adaptation policies.

Over the past decade, the Horn of Africa (HoA) has been plagued by recurrent drought events that have had devastating impacts on the population. The frequency, duration and severity of these droughts are expected to increase in the wake of global warming, leading to higher losses and damages if the vulnerability of the population is not reduced. Monitoring and early warning systems for droughts are based on various drought hazard indicators. However, assessments of how these indicators are linked to impacts are rare. For adequate drought management, it is essential to understand and characterise the drivers of drought impacts, especially in the HoA, where most studies focus either on meteorological droughts, agricultural droughts or the propagation of droughts through the hydrological cycle, without considering the relationship between hazard and impact. Drought hazard indices alone cannot capture the vulnerability of the system. In this study, we identify meaningful indices for the occurrence of region- and sector-specific impacts. We assess the effectiveness of socio-economic clustering in categorising counties based on common characteristics and their correlation with historical drought impacts (malnutrition, milk production and trekking distances to water sources). Using Random Forest (RF) and Spearman correlation analyses, we examine the link between drought indices (Standardised Precipitation Index, Standardised Precipitation Evapotranspiration Index, Standardised Soil Moisture Index, Standardised Streamflow Index and Vegetation Condition Index) with different accumulation periods and the impact data. We find that clustering regions based on vulnerability proxies significantly improves the hazard-impact relationship, emphasising the importance of considering vulnerability factors in drought risk assessment. Our results indicate an impact-specific relationship that is strongly influenced by the vulnerability of the region. In particular, household and livestock distance to water is most strongly associated with medium- to long-term precipitation-based indices (2-10 months), while milk production can be associated with a variety of indices with different accumulation periods (5-24 months), and malnutrition is correlated with precipitation- and streamflow-based indices (5-24 months). Household and livestock distance to water is well modelled by clusters reflecting low access to improved sanitation and safe water sources, high poverty, aridity and gender disparities. Malnutrition was well modelled by clusters related to aridity, average precipitation, food consumption score, access to water sources, improved sanitation and poverty levels. The type of clustering used in modelling the impact of drought on milk production does not have a major impact on the performance of the models. We then apply this relationship to hindcast drought indices to obtain impact data on individual counties for periods when no impact monitoring was done yet. With that information we estimate the associated risk under specific climatic conditions. By recognising the drivers and vulnerability factors that influence the sensitivity of counties to drought, communities can better prepare and mitigate the impacts of drought.

Recent disasters have highlighted the severe local impacts of extreme precipitation, including flash floods, landslides, and urban inundation. Despite significant investments in early warning systems, these events often catch many people off guard, emphasizing the need for a better translation of warnings into early actions. In this study, we directly address this gap by translating precipitation forecasts from the ECMWF, specifically the extreme forecast index (EFI) and shift of tails (SOT), to concrete triggers for early action. Our analysis reveals that such actions, triggered with a 2-3-day lead time, have a high potential economic value (PEV) across large parts of Europe. The SOT forecasts generally have higher economic value than EFI, especially at longer lead times. However, the effectiveness of both disappears across most of Europe with a 5-day lead. These results are based on comparing forecasts to the E-OBS dataset for extreme rainfall (>5-yr return period) over 8years. We apply the optimal warning thresholds found in this analysis to the rainfall event that triggered the July 2021 western Europe flood disaster. Our results indicate that the EFI and SOT forecasts provided accurate and timely warnings at least 2-3 days in advance, aligning with flood impacts recorded in the Emergency Events Database (EM-DAT) dataset. Notably, on a 1-day lead time, the SOT forecasts accurately pointed to the Ahr catchment in Germany with highly exceptional values, providing strong indications of the disaster that unfolded there. This study underscores the value of these rainfall indicators and calls for further testing on more high-impact events. SIGNIFICANCE STATEMENT: Recent floods triggered by extreme rainfall, such as the 2021 western Europe floods, the 2023 Emilia-Romagna floods, and the 2022-23 California floods, have had an enormous impact. These events highlight the urgent need to scale up early actions. In this study, we use two rainfall indicators from ECMWF: the extreme forecast index (EFI) and the shift of tails (SOT). We find evidence that these indicators have high economic value when used to trigger early actions 2-3 days ahead of extreme rainfall events in Europe. Furthermore, these indicators provided accurate warnings at least 2 days before the 2021 western Europe floods. This underscores the vital role of early warning and preparedness in reducing future disaster risk.

Global flood and salinity risk is increasing due to sea-level rise (SLR). Farmers in coastal areas will experience this risk through rising salinity levels that reduce crop yields and direct flood damage to their property. In response, farmers can take various actions such as protecting their homes, changing cropping patterns, or migrating to safer areas if adaptation is not feasible. We developed the agent-based model (ABM) called DYNAMO-M to assess these responses and their influence on climate risk for farmers. This is a global ABM that simulates the actions of 13 million farming households based on economic decision theory across 674 coastal zones from 2015 to 2080. Results show that globally, 23 billion USD of private farming assets will be at risk annually, and coastal areas of Florida, New York, and Oregon in the United States, as well as the coasts of Spain, France in Europe, Japan, China, Indonesia, and the Philippines in Asia, Brazil and Australia will experience significant farmer outmigration to inland regions. Under the RCP4.5-SSP5 scenario, approximately 518,000 farming households are projected to migrate, with the highest numbers in East Asia and the Pacific (about 376,000 households), followed by South Asia (about 86,000 households). The highest household adaptation rates (over 40%) are observed in Indonesia, parts of Africa (e.g., Nigeria, Mozambique), Argentina, and some Mediterranean regions. However, many farmers in these areas cannot adapt due to limited income. Such budget constraints are not limited to poorer countries. The United States also shows a high percentage of households unable to afford adaptation. Our findings suggest that small government subsidies can significantly enhance the adaptive capacity of poorer households, partially offsetting migration flows and increases in climate risk. Results indicate that globally, about 15,000 fewer households migrate when provided with 30% insurance coverage. Government subsidies that reduce farmers’ adaptation costs by 30% decrease global farmer migration by up to 10% in some floodplains. These findings are useful for informing migration policies, options for managed retreat, and efforts to reduce the burden on receiving areas while supporting vulnerable, often impoverished, communities in coastal zones.

Extreme windstorms pose significant societal and economic challenges, ranking among the costliest natural disasters in Europe. This study addresses the complex task of quantifying windstorm impacts, with a specific focus on the Netherlands. Despite their substantial economic cost, windstorm risks in the Netherlands have been underexplored in dedicated regional studies. Existing large-scale investigations often rely on hazard-loss relationships derived from data from other European countries. We aim to enhance the accuracy of windstorm risk assessment by utilizing not only higher-resolution hazard data but also higher-resolution Dutch damage data. Our methodology involves analyzing high-resolution data to identify hazard variables that best correlate with losses. This is done by leveraging post-disaster loss data from a private Dutch insurance company. In particular, we use the aggregated losses per postal code 4 area, which delivers a nuanced understanding of the spatial distribution of losses. Simultaneously, we account for hazard intensities using the wind climatology data from KNMI North Sea Wind (KNW). This data is derived from 40 years (1979-2019) of ERA-Interim re-analyzed data and downscaled to a higher resolution (2.5 x 2.5 km) tailored specifically for the Netherlands. Through statistical analysis, the study aims to determine the most suitable hazard components for a regional windstorm damage assessment model. This approach aims to move beyond the conventional use of daily maxima wind speed or gust speed by evaluating the appropriateness of hazard variables concerning observed losses. This meticulous integration of proprietary loss records and refined wind climatology enables developing new spatial windstorm hazard maps and a high-resolution windstorm risk database, which provide a solid basis for risk assessment.

The characterization of drought hazards remains a complex endeavor, primarily due to the absence of a universally accepted definition for a "drought event." Different deficits across various parts of the water cycle contribute to a spectrum of drought consequences, rendering the definition contingent upon the impacts incurred. Moreover, quantifying drought vulnerability poses challenges given the intricate interplay among socioeconomic, political, and environmental factors that influence the relationship between a drought event and its impacts on exposed production systems, people and nature. Our work addresses these challenges by introducing a novel data-driven methodology employing an array of drought indices and several datasets on observed drought impacts. Applying decision tree-based AI techniques, this method identifies combinations of hydrometeorological conditions known to generate societal consequences, and as such is able to estimate probabilistic drought disaster risk.The presented impact-based approach is generalizable and impacts evaluated include energy production losses, internal displacement, crop and livestock damage, malnutrition, ecosystem health degradation, and strains on drinking water utilities. Illustrated through a case study in the Horn of Africa, this contribution exemplifies the quantification of expected annual drought impact, whereby impact is measured as the number of drought-induced internally displaced persons (IDPs). Drawing on the latest IDMC Displacement Tracking Matrix data, we assessed drought displacement risks under current and projected climate scenarios for Somalia and Ethiopia. Both countries grapple with complex human mobility dynamics, driven by a multitude of push and pull factors. Our findings reveal average annual IDPs up to 2% in some regions in Ethiopia, rising to 3% with unmitigated climate change. In Somalia, the majority of regions are anticipated to experience on average >10,000 drought-induced IDPs annually, under all future projections. Our model demonstrates proficiency in distinguishing prolonged and flash droughts as drivers for displacement. Furthermore, it facilitates the identification of hotspot areas, thereby supporting drought disaster risk reduction decisions and proactive policies.

The Horn of Africa drylands (HAD) encompassing Kenya, Somalia, and Ethiopia, recently endured an unprecedented multi-year drought from 2020 to 2023, causing devastating impacts. This study investigates these impacts and the dynamics of human adaptation in response to the drought, comparing it to earlier drought events (i.e., 2016-2018) to identify key lessons. First, drought impact data-covering milk production, trekking distances to water sources, and internally displaced persons (IDPs)-are analysed over time to provide a detailed overview of drought dynamics. Second, household survey data (n = 752) are used to examine community perceptions of the drought period and their adaptation strategies. Finally, agent-based modeling (ABM) simulations explore the interactions between mitigation, adaptation decisions, and drought impacts. The results reveal that, on average, the 2020-2023 drought had more severe impacts than the 2016-2018 drought, although the latter exhibited greater variability in impacts. Communities have adopted various adaptation measures to cope with drought effects; however, limited knowledge and financial resources remain significant barriers to scaling these efforts. ABM simulations indicate that enhancing extension services can boost the adoption of adaptation strategies, leading to increased crop and milk production. Additionally, the simulations suggest that water harvesting can mitigate drought impacts upstream, though it may reduce water availability downstream. These findings highlight the critical need for sustained investments in adaptation measures, timely and well-informed decision-making, and region-specific interventions while carefully considering the trade-offs associated with these strategies.

Ecosystems in Europe are increasingly faced with more frequent and more intense drought events. The impacts of droughts do not only undermine ecosystem health and the provision of ecosystem services, but can lead to the deterioration of the system’s long-term resilience to droughts. In order to effectively assess, reduce, and manage the risks posed by droughts on ecosystems, it is first necessary to gain a thorough understanding of how droughts affect a particular ecosystem, what the underlying risk drivers and root causes are, and how these interact to produce that risk. Addressing this need, we have developed conceptual models of drought risks for two highly-relevant European ecosystem types, forest and freshwater ecosystems. The conceptual models were developed and visualized using the impact chain methodology, building on extensive literature review and expert consultations, and validated in a series of expert workshops. Following this process, the risks of decreased primary production, forest die-off, and soil degradation and desertification were identified for forest ecosystems, and the risk of disruption of environmental flow for freshwater ecosystems.The resulting impact chains provide insights into how different climatic, ecological, and societal risk drivers interact to produce drought risks for ecosystems, with some drivers being specific to a certain risk (e.g. forest composition), and others shared across them (e.g. societal water demand and abstractions). While some of these drivers relate to purely ecological features (e.g. plant physiology or soil conditions), many relate to how ecosystems are managed, and to the influence of other sectors/systems upon them (e.g. hydropower, river transportation or intensive agriculture and their adverse effects on freshwater ecosystems). Moreover, the impact chains also highlight some of the root causes (e.g. increased demand for energy, farmers' lack of awareness about agriculture's impacts on freshwater ecosystems, or incentives to enhance navigability) behind these drivers, indicating potential entry points for risk reduction and adaptation. The visualization in impact chains is useful to enhance the understanding and at the same time break down the complexity of the risks for these systems, which can support data-driven risk assessment, as well as the identification of entry points for risk management and adaptation. While in this work the risks posed by droughts for forests and freshwater ecosystems were assessed on European level, the impact chain approach presented here can be used at different scales and transferred to different ecosystems at risk from droughts. Moreover, it can be used to identify common risk drivers between ecosystems that could be addressed jointly, contributing to a more systemic drought risk management.

Food insecurity is a global concern resulting from various complex processes and a diverse range of drivers. Due to its complexity, it is one of the most challenging drought impacts to predict. In this study, we introduce a novel machine learning model designed to forecast food crises in the Horn of Africa up to 12 months in advance. We trained an “XGBoost” model using more than 20 different input datasets to capture key food security drivers such as drought, economic shocks, conflicts and livelihood vulnerability. The model shows a promising ability to predict food security dynamics several months in advance (R2>0.6, three months in advance). Notably, it accurately predicted 20% of crisis onsets in pastoral regions (n = 84) and 40% of crisis onsets in agro-pastoral regions (n = 23) with a 3-month lead time. We compared these results to the established FEWS NET early warning system, and found a similar performance over these regions. However, our model is clearly less skilled in predicting food security for crop-farming regions than FEWS NET. This study underscores the importance of integrating machine learning into operational early-warning systems like FEWS NET and expanding these techniques to the continental or global-scale.