Facing water scarcity, the Berlin-Brandenburg region explored underground water storage to capture excess surface water during wet periods and store it in deep aquifers for use during droughts. We identified extensive areas suitable for storage and found substantial downstream water availability under ecological flow constraints. The approach could help offset local water demand and provide a cost-effective, transferable strategy for strengthening water security in other water-stressed regions.

 

European windstorm damages varied by a factor three over recent multidecadal periods, and a better understanding of these changes could improve how this risk is managed. Here, we explored the impacts of anthropogenic aerosols (AA) using results from climate model experiments, and found AA boosted European wind losses by an average of 45 % in the late 20th century, though varying from zero to 100 % between the six models. Validation data suggest the signal may be at the higher end of this range.

We use several machine learning models to explore which factors best explain landslide release during an extreme rainfall event in eastern Norway. As landslides often occur in clusters, methods must be chosen carefully to account for any spatial effects. When considering this, we find that south-facing slopes, thicker soils and more water made landslides most likely. On forested slopes, landslides are most likely in deciduous rather than spruce or pine stands.

Post-fire debris flows or mudflows, can occur during intense rainfall following wildfires. Yet, it remains unclear how seasonal variation affects debris-flow initiation years after the fire. Using simulations of the 2020 Lake Fire in California, we show that debris-flow likelihood varies drastically between drier and wetter months, even four years after the fire. This study highlights the need to consider climatic and seasonal patterns in post-fire debris-flow assessment.

Extreme weather events are becoming more frequent and severe, creating a strong need for rapid and reliable information. We developed an open tool that automatically detects and tracks heatwaves, cold spells, heavy rain, and strong winds across Europe, both in real time and in past decades. By comparing current events with long historical records, the tool shows how unusual an event is and reveals clear increases in heatwaves, while other hazards show more mixed changes.

We investigated the Central Texas flash flood on July 2025 to understand why forecasts underestimated its severity. We analyzed operational streamflow forecasts and evaluate flood inundation maps against benchmark flood extent from High Water Mark. Results show that peak flooding was missed due to spatial error in rainfall and failures in monitoring stations, leading to an underestimation of impacts. We highlight the need for uncertainty aware flood predictions to support emergency response.

This study presents a quantitative method to capture how damage caused by concurrent and consecutive hazard events evolves, including how earlier damage and ongoing recovery affect later events. Applied to a sequence of hurricanes and earthquakes in Puerto Rico, it shows that ignoring residual damage significantly underestimates the impacts caused by consecutive events.  By accounting for recovery and interaction effects, the method supports more realistic damage assessment and risk planning.

Our research lays the groundwork for the next generation of disaster risk modelling by improving how building-level value and use are estimated across Germany. By testing multiple data sources and methods, we identify a transparent, adaptable approach that enhances forecasts of damage and recovery – helping to protect lives, property, and communities.

Earthquakes cause major losses worldwide, but their impacts differ across regions. Using global disaster records from 1980 to 2024, this study shows that while recorded disasters and economic losses increased, deaths declined relative to population size. Social and economic conditions are linked to disaster impacts, highlighting the importance of resilience planning.

The health burden of extreme heat necessitates the development of effective early warning systems. These systems depend critically on the underlying weather forecasts. We compare forecasts of temperature-related mortality based on data-driven weather forecasts with those based on physics-based weather forecasts for summer 2024 in Europe. We find that forecasts based on data-driven weather forecasts could represent a promising avenue for the development of heat-related health impact forecasts.

Deadwood supports biodiversity and contributes to the protective function of mountain forests, but under certain conditions it can itself become a hazard when sliding downslope. This study examines when and why sliding deadwood occurs and how far it can travel. Using a physics-based simulation model and forest data, we show that sliding is favoured by steep slopes, wet surfaces, and specific decay stages, helping identify where hazards can be reduced while preserving protective forest functions.