This study focused on a catastrophic post-earthquake rockslide in Wenchuan, Southwestern China. On-site investigations and aerial photography were used to ascertained the basic landslides characteristics. A 3D discrete element method with discrete fracture network was employed to assess the dynamic process of the rockslide. The effects of fractures density and friction angle on kinetic characteristics were analyzed, and the role of the check dam as a prevention work has been verified.

Indicators contain observable and measurable characteristics to understand the state of a concept or phenomenon and/or monitor it over time. There have been limited efforts to understand how indicators are being used in multi-hazard and multi-risk contexts. We find most of existing indicators do not include the interactions between hazards or risks. We propose a set of recommendations to enable the development and uptake of multi-hazard and multi-risk indicators.

Intense rainfall events cause increasingly severe damages to urban areas globally. We use unique insurance claims data to study the effect of nature-based and other adaptation measures on damage. We compare an area in Amsterdam where measures have been implemented to a similar, adjacent area without measures using an innovative method. We find a significant reduction of damage where the adaptation measures were implemented. Urban areas can reduce rain damage by implementing adaptation measures.

This study proposes an innovative method for predicting drought in the Huaihe River Basin of China using advanced machine learning and interpretable artificial intelligence techniques. By analyzing more than 50 years of data, the model successfully predicted four drought categories with an accuracy of 79.9 %. It used explanatory methods to analyze the contribution of different drought influencing factors, providing key insights for early warning systems and water resources planning.

INSYDE-content is a novel probabilistic model designed to estimate flood damage to household contents. Based on a synthetic “what-if” approach, it integrates multiple input variables describing both exposed items and damage mechanisms, while accounting for uncertainty. The model expands the range of available tools for flood damage assessment by including an often overlooked asset in current literature.

Tornado is a destructive meteorological phenomenon that can injure, kill people, and damage infrastructure. Indonesia has no tornado database yet. We collected tornado reports from numerous scattered reports between 1800 and 2024 to analyze the climatological tornado for Indonesia, and the results are published in an open repository to catalyze future tornado studies in the tropics.

Protecting urbanized alluvial fan canals and levees from flooding requires effective sediment retention measures, such as check dams, terraces, and trees on steep basins. However, their effectiveness declines over time due to sedimentation and aging, increasing the risk of catastrophic breaching floods. To enhance urban resilience, we propose preserving natural mountain basins and allocating about 35 % of the alluvial fan to channel migration and sediment deposition corridors.

Urban flooding is a growing issue in cities, often disrupting daily life, especially at night, when the extent of flooding is harder to identify. This study introduces NWseg, a new deep learning model designed to identify the extent of urban flooding at night. Using a dataset of 4000 nighttime images, we found that NWseg outperforms existing models in accuracy. This research offers a practical solution for real-time flood monitoring, helping improve urban disaster response and management.

With growing demand for decision support in recreational and professional use of avalanche terrain, we applied machine learning for automated Avalanche Terrain Exposure Scale (AutoATES) mapping in Bulgaria. A Random Forest model, trained on expert-labelled data from the Pirin Mountains, accurately classifies avalanche terrain and reduces reliance on manual expert mapping, offering an effective and scalable solution for large-scale regional AutoATES applications.