This study provides an overview of the impacts to critical infrastructure and how recovery has progressed after the July 2021 flood event in Germany, Belgium and the Netherlands. The results show that Germany and Belgium were particularly affected, with many infrastructure assets severely damaged or completely destroyed. This study helps to better understand how infrastructure can be affected by flooding and can be used for validation purposes for future studies.

The L-ML-HySEA (Landslide Multilayer Hyperbolic Systems and Efficient Algorithms) model of the tsunami triggered by the Storfjorden LS-1 landslide provides new insights into the sliding mechanism and bathymetry controlling the propagation, amplitude values and shoaling effects as well as coastal impact times. This case study provides new perspectives on tsunami hazard assessment in polar margins, where global climatic change and its related ocean warming may contribute to landslide trigger.

Extreme events have increased in the last few decades; having a good estimation of the outflow of a reservoir can be an advantage for water management or early warning systems. This study analyzes the efficiency of different machine learning techniques to predict reservoir outflow. The results obtained showed that the proposed models provided a good estimation of the outflow of the reservoirs, improving the results obtained with classical approaches.

The influence of model resolution and settings on drought reproduction in Germany between 1980–2009 is investigated here. Outputs from a high-resolution model with settings tailored to the target region are compared to those from coarser-resolution models with more general settings. Gridded observational data sets serve as reference. Regarding the reproduction of drought characteristics, all models perform on a similar level, while for trends, only the modified model produces reliable outputs.

Sandy dunes protect the hinterland from coastal flooding during storms. Thus, models that can efficiently predict dune erosion are critical for coastal zone management and early warning systems. Here we develop such a model for the Dutch coast based on machine learning techniques, allowing for dune erosion estimations in a matter of seconds relative to available computationally expensive models. Validation of the model against benchmark data and observations shows good agreement.

Assessing landscape wildfire connectivity supported by wildfire spread simulations can improve fire hazard assessment and fuel management plans. Weather severity determines the degree of fuel patch connectivity and thus the potential to spread large and intense wildfires. Mapping highly connected patches in the landscape highlights patch candidates for prior fuel treatments, which ultimately will contribute to creating fire-resilient Mediterranean landscapes.

Seafloor instruments at the bottom of the Sea of Marmara recorded disturbances caused by earthquakes, addressing the minimum magnitude that may be recorded in the sediment. A magnitude 4.7 earthquake caused turbidity but little current. A magnitude 5.8 earthquake caused a mudflow and strong currents that spread sediment on the seafloor over several kilometers. However, most known earthquake deposits in the Sea of Marmara spread over larger zones and should correspond to larger earthquakes.

A multiscale analysis, where the historical and future precipitation data from the CORDEX project were used as input in a hydrological model (HEC-HMS) that, in turn, feeds a 2D hydraulic model (Iber+), was applied to the case of the Miño-Sil basin (NW Spain), specifically to Ourense city, in order to analyze future changes in flood hazard. Detailed flood maps indicate an increase in the frequency and intensity of future floods, implying an increase in flood hazard in important areas of the city.

Volcanic edifices result from successive construction and dismantling phases. Thus, the geological units forming volcanoes display complex geometries. We show that such geometries can be reconstructed thanks to aerial views, topographic surveys and photogrammetric models. In our case study of the Samperre cliff (Martinique, Lesser Antilles), it allows us to link destabilizations from a rocky cliff to the existence of a filled paleo-valley and estimate a potentially unstable volume.

Accurate predictions of storm activity are desirable for coastal management. We investigate how well a climate model can predict storm activity in the German Bight 1–10 years in advance. We let the model predict the past, compare these predictions to observations, and analyze whether the model is doing better than simple statistical predictions. We find that the model generally shows good skill for extreme periods, but the prediction timeframes with good skill depend on the type of prediction.

Debris flows are dangerous sediment–water mixtures in steep terrain. Their formation takes place in poorly accessible terrain where instrumentation cannot be installed. Here we propose to monitor such source terrain with an autonomous drone for mapping sediments which were left behind by debris flows or may contribute to future events. Short flight intervals elucidate changes of such sediments, providing important information for landscape evolution and the likelihood of future debris flows.

Fire weather indices are used to assess the effect of weather on wildfires. Fire weather risk was computed and combined with large wildfires in Portugal. Results revealed the influence of vegetation cover: municipalities with a prevalence of shrublands, located in eastern parts, burnt under less extreme conditions than those with higher forested areas, situated in coastal regions. These findings are a novelty for fire science in Portugal and should be considered for fire management.

Models are needed to understand and predict pollutant output from forest fires so fire agencies can reduce smoke-related risks to human health. We modelled air quality (PM_2.5) based on fire area and weather variables. We found fire area and boundary layer height were influential on predictions, with distance, temperature, wind speed and relative humidity also important. The models predicted reasonably accurately in comparison to other existing methods but would benefit from further development.

Barrier islands in tropical regions are prone to coastal flooding and erosion during hurricane events. The Yucatán coast was impacted by hurricanes Gamma and Delta. Inner shelf, coastal, and inland observations were acquired. Beach morphology changes show alongshore gradients. Flooding occurred on the back barrier due to heavy inland rain and the coastal aquifer's confinement. Modeling systems failed to reproduce the coastal hydrodynamic response due to uncertainties in the boundary conditions.

Humans build levees to protect themselves against floods. We need to know where they are to correctly predict flooding, for example from sea level rise. Here we have looked through documents to find levees, and checked that they exist using satellite imagery. We developed a global levee map, available at www.opendelve.eu, and we found that 24 % of people in deltas are protected by levees.

In this article, we propose an equivalent hazard magnitude scale and a method to evaluate and compare the strengths of natural hazard events across different hazard types, including earthquakes, tsunamis, floods, droughts, forest fires, tornadoes, cold waves, heat waves, and tropical cyclones. With our method, we determine that both the February 2021 North American cold wave event and Hurricane Harvey in 2017 were equivalent to a magnitude 7.5 earthquake in hazard strength.

The Tsunami Alert Centre of the National Institute of Geophysics and Volcanology (CAT-INGV) has been promoting the study of tsunami risk perception in Italy since 2018. A total of 7342 questionnaires were collected in three survey phases (2018, 2020, 2021). In this work we present the main results of the three survey phases, with a comparison among the eight surveyed regions and between the coastal regions and some coastal metropolitan cities involved in the survey.

An innovative multi-strategy-mode waterlogging-prediction framework for predicting waterlogging depth is proposed in the paper. The framework selects eight regression algorithms for comparison and tests the prediction accuracy and robustness of the model under different prediction strategies. Ultimately, the accuracy of predicting water depth after 30 min can exceed 86.1 %. This can aid decision-making in terms of issuing early warning information and determining emergency responses in advance.