Posters virtual: Fri, 8 May, 14:00–18:00 | vPoster spot 3
Agricultural production remains central to food security and rural livelihoods, yet it is increasingly exposed to compound and cascading natural hazards under a changing climate. Drought, flooding, and extreme rainfall, heat stress, storms, and soil degradation do not operate in isolation. Their impacts often accumulate across the seasonal calendar and propagate beyond the field through labor, processing, storage, and distribution constraints. This contribution synthesizes evidence on how multi-hazard pressures disrupt agricultural productivity and stability, with attention to major staple and cash crops (for example, rice, wheat, maize, sugarcane, and soybean) and to vulnerability patterns that shape disproportionate impacts on resource-constrained and smallholder systems. We review and organize recent findings around three linked questions: (1) how hazard timing and co-occurrence influence crop sensitivity across key growth stages; (2) which biophysical and socioeconomic conditions amplify losses and slow recovery; and (3) which adaptation pathways show consistent promise under multi-hazard risk. A central focus is Climate-Smart Agriculture (CSA) as an integrated response, including practices that aim to improve productivity while strengthening resilience and reducing environmental tradeoffs. However, the review also highlights barriers that frequently limit CSA uptake in high-vulnerability settings, including institutional constraints, knowledge gaps, and financing limitations. By connecting hazard mechanisms to stage-specific crop impacts and to constraints along agricultural value chains, the synthesis supports more targeted adaptation planning and more realistic resilience strategies. The paper argues for context-specific, multi-stakeholder approaches that combine policy, technology, and farmer-centered implementation to address increasing climate and hazard uncertainty.
How to cite: Hossain, T. and Mostafiz, R.: Compound hazards, crop sensitivity, and climate-smart adaptation, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-22036, https://doi.org/10.5194/egusphere-egu26-22036, 2026.
The increasing complexity of cascading and compounding effects, necessitates innovative tools for wide stakeholder engagement and decision-making, especially in uncertain situations. Risk communicators often struggle to successfully convey these complexities to diverse groups of actors. In the PARATUS project, we implemented a series of four serious games: High Water Pantano, Bucur Simulation, Saltum Montem, Paratus Systemic Risk Game; to address this gap through experiential process.
The structured, stakeholder-driven process used in the PARATUS project was grounded in the CompleCSUs framework and the design thinking methodologies. The development process included four phases, as follows: 1) Research and conceptualization, focused on the literature review and Miro app based mapping of stakeholder needs and PARATUs four case study areas (Caribbean, Bucharest, Istanbul, and the Alps); 2) Scenario and role design, focused on translating real-world impact chains co-developed with stakeholders into interactive storylines; 3) Prototyping and iterative testing, focused on stakeholders interacting with the prototypes and providing direct feedback to the tools; and 4) Implementation and evaluation, focused on the deployment of serious games in workshops and assessing their effectiveness.
Some benefits identified include increased transdisciplinary collaboration and the opportunity for stakeholder exploration of the results of inaction or certain decisions linked with the risk reduction, in a safe, simulated environment. However, the four-phase serious games approach in the PARATUS also resulted in certain critical lessons for the future implementation of co-design processes. These included the need for more flexibility in formats of the tools (analog vs. digital) to accommodate technical and context-based limitations; the importance of understanding the institutional hierarchies and factoring them into the process activities, and the need for multilingual support, especially in the transboundary context, for increase of the accessibility of the tools and trust levels of the participants. Following such four-step process, scientific risk assessment can be transformed into a scalable, user-centered and engaging tool for fostering long-term resilience.
How to cite: Kulakowska, M., Atum, F., Koelle, B., and Magnueszewski, P.: A Four-Phase Serious Games Approach in the PARATUS Project, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-20736, https://doi.org/10.5194/egusphere-egu26-20736, 2026.
Despite urgent needs for adaptive coastal risk management, operational systems still rely heavily on static triggers and fragmented information that overlook interactions between evolving hazards and response actions. Building on a completed game-like deep reinforcement learning (DRL) testbed, we present a pathway toward operational coastal decision support, progressing toward real-world case studies such as Venice in Italy and South East Queensland in Australia.
In the first phase, we developed a controllable game-like scenario that captures the essential components of coastal emergency management: a simplified representation of coastal geography and built assets, dynamic multi-hazard drivers evolving over time, and an action space reflecting plausible operational interventions under constraints. Using this environment, we demonstrated that a PPO-based DRL agent can learn adaptive policies through repeated interactions, as we gained practical lessons on state representation, constraint handling, and reward design for safety-critical objectives.
We then focus on the transition from simulation to real-world settings by outlining a set of alternative state-representation options, spanning classical dimensionality reduction and feature engineering through to learned latent-state methods. We report results for selected approaches, using autoencoders as the primary entry point to compress high-dimensional spatio-temporal hazard and exposure information into compact variables that retain decision-relevant structure while improving training efficiency and robustness. This provides a practical interface to real-world, digital-twin style environments built from geospatial and socio-economic data and forecast inputs.
Finally, we propose an orchestration layer to reduce the risk of AI-driven decision making and improve usability. A large language model (LLM) ingests DRL outputs and contextualises recommendations via retrieval-augmented generation over plans, studies, and standard operating procedures, together with API calls to dynamic data feeds. The proposed orchestration layer is intended to translate DRL outputs into human-readable and auditable decision support for a human-in-the-loop operator, grounding recommendations in retrieved local documentation and live data feeds to strengthen transparency, uncertainty communication, and operational trust.
How to cite: Sano, M., Ferrario, D., Casagrande, S., Vascon, S., Torresan, S., and Critto, A.: Deep Reinforcement Learning for Operational Coastal Emergency Response With AI Agent Orchestration and Human Oversight, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-17369, https://doi.org/10.5194/egusphere-egu26-17369, 2026.
This paper presents the development and pilot evaluation of Map@Me, an RPG-based serious game designed to improve understanding of hydrogeological risks and evacuation planning. Developed within the Motivation and Engagement in Disaster Mapping in Europe (MEDiME) Horizon Project, Map@Me targets a diverse audience and was tested in formal education settings, specifically middle and high schools.
The game integrates real local hazard maps, allowing players to explore their own environments and engage in realistic evacuation scenarios. With Map@Me, the player traces a realistic evacuation route that takes into account diverse mobility conditions, including disabilities, as well as advantageous and challenging factors, such as access to local knowledge and unfamiliarity with the area. Using a randomised system to determine the fictional character’s features in a real hazard map scenario, Map@Me represents a good example of how traditional disaster education can be supported by participatory methods of learning, whereby the agents can, in a controlled environment, experiment creatively with their behavioural choices and address their intrinsic biases.
During the presentation of the preliminary results from pilot sessions conducted with students, we will highlight both traditional learning outcomes—such as knowledge of evacuation sites and emergency preparedness measures—and “soft” learning outcomes, including cooperation, empathy, and collective responsibility.
The findings suggest that serious games such as Map@Me can enhance inclusive, place-based disaster preparedness, hazard map literacy and risk awareness, and overall contribute to a more socially aware approach to risk communication among younger audiences.
How to cite: Petraroli, I., Flacke, J., and Atun, F.: Motivation and Engagement in Disaster Mapping in Europe (MEDiME): Understanding hydrogeological risks and vulnerability through serious gaming, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-17956, https://doi.org/10.5194/egusphere-egu26-17956, 2026.
Wildfires are increasingly destructive events, threatening ecosystems and human infrastructure while contributing significantly to carbon emissions. Accurate and timely burned area mapping is therefore essential for effective mitigation and recovery. Optical satellite sensors are often hindered by clouds and smoke, making Synthetic Aperture Radar (SAR) sensors like Sentinel-1, with their all-weather capability, a crucial tool for monitoring. However, SAR backscatter is significantly influenced by topography, which can distort signals and hinder accurate detection.
This study evaluates the impact of angular-based radiometric terrain normalization (RTN) on burned area mapping using Sentinel-1 SAR data and the Normalized Radar Burn Ratio (NRBR) index. We compare the performance of NRBR calculated with standard sigma nought (σ⁰) and with gamma nought (γ⁰) corrected via an angular-based RTN model implemented in Google Earth Engine. A U-Net deep learning model was used to delineate burned areas in Portugal and California. Results show that NRBR without RTN achieved better accuracy in Portugal, suggesting potential overcorrection effects in moderate terrain. In California, RTN slightly improved overall accuracy and reduced commission errors, although omission errors remained high. These findings indicate that while RTN enhances radiometric consistency, its impact on burned area detection with NRBR is limited, likely because the NRBR formulation itself already mitigates topographic effects through pre/post-fire ratios.
How to cite: Tarazona, Y. and Mantas, V.: The Impact of Radiometric Terrain Normalization (γ⁰) on Burned Area Mapping Accuracy Using Sentinel-1 data, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-15290, https://doi.org/10.5194/egusphere-egu26-15290, 2026.
Geotechnical centrifuge modeling provides an effective approach to reproduce prototype-relevant stress states for high-speed dry granular flows. Yet, in a rotating reference frame, the Coriolis acceleration induced by rapid granular motion can become comparable to the centrifugal acceleration, thereby markedly modifying run-out behavior and impact responses and complicating the interpretation of physical modeling results. This study integrates a suite of centrifuge model tests with discrete element method (DEM) simulations to systematically elucidate how Coriolis effects govern both the mobility of dry granular flows and their impact on rigid barriers. For run-out processes, a DEM framework incorporating both centrifugal and Coriolis accelerations is employed to compare granular mobility under three Coriolis configurations: dilative, compressive, and deflective conditions. The results indicate that the dilative Coriolis condition substantially enhances flow mobility and kinetic energy, whereas the compressive condition suppresses run-out and promotes flow densification. In contrast, under the deflective Coriolis condition, the sensitivity of the final run-out distance and overall flow scale to Coriolis effects is significantly reduced. This reduced sensitivity is attributed to two opposing deflection stages during propagation and deposition, suggesting a practical advantage for mitigating Coriolis-induced bias in centrifuge modeling. For impact processes, centrifuge experiments combined with DEM simulations are used to characterize granular impact behaviors on rigid barriers under different Coriolis conditions. The Coriolis effect has a limited influence on the peak magnitude of the total impact force, but it significantly alters the force time history and spatial distribution by modifying the velocity structure, flow thickness, and particle-scale momentum transfer. Notably, impact responses obtained under the dilative Coriolis condition are closer in force level to Coriolis-free reference cases, whereas the resultant force application point is comparatively insensitive to the Coriolis configuration. Overall, the results demonstrate that Coriolis effects should not be treated as a uniform experimental disturbance. Instead, they represent a key control factor whose influence depends on the specific quantities of interest. The findings provide methodological guidance for configuring centrifuge experiments and interpreting results in the modeling of high-speed dry granular flows, with explicit implications for both run-out and impact simulations.
How to cite: Zhang, B.: Understanding Coriolis effects in centrifuge modeling of high-speed dry granular flows, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7212, https://doi.org/10.5194/egusphere-egu26-7212, 2026.
Landslides are among the natural hazards that significantly impact human life and infrastructure, making accurate landslide mapping essential for hazard assessment, risk reduction, and land use planning. However, mapping landslides, particularly in vegetated areas, remains challenging, as traditional field-based and manual mapping approaches are time-consuming and require substantial expert knowledge. Semi-automatic mapping methods based on high-resolution Digital Terrain Models (DTMs) have improved landslide inventory preparation; however, their transferability to larger and diverse environmental settings remains limited and require further assessment. Therefore, this study aims to assess the transferability of a Geographic Object-Based Image Analysis (GEOBIA) landslide mapping approach using optimal moving window sizes, and to examine whether model performance varies across specific land use classes and improves with higher-quality DTM data.
A GEOBIA-based model, originally developed for forest covered landslides in the cuesta landscape of Jena region (Zangana et al., 2025), was transferred and applied to landslides at the Swabian Alb escarpment in south-western Germany, which are located not only in forests, but also in grasslands and settlements. The study area is characterized by Jurassic limestones overlying marls and clays. It is affected mainly by rotational slides, slump-earthflows, and translational landslides, some of which show repeated reactivation. The manually mapped landslide inventory was used for result validation and accuracy assessment. DTM derivatives (from the 2003 and 2023 data) were prepared using optimal moving window sizes following Zangana et al. (2025). The semi-automatic landslide detection workflow involved multi-resolution segmentation (MRS) and support vector machine (SVM) classification, followed by expert-based refinement and accuracy assessment against the reference map. Finally, transferability was further examined through land use class-based performance analysis and by evaluating the effect of higher-quality 2023 DTM data on model results.
The results indicate that the model developed for the Jena region is transferable to the Swabian Alb. When applied to the 2003 dataset, without differentiating between land use types, the model achieved a 75% detection rate for landslide body areas. Using the 2023 dataset increased detection accuracy to 86% compared to the 2003 data. The area-based detection accuracy in this study is approximately 30% higher than reported for the Jena region by Zangana et al. (2025). When considering only forested areas—for which the model was originally developed—the true positive rate increased by about 15%, while false positives decreased by a similar margin. Although the approach effectively identifies landslides, particularly in vegetated areas, it currently performs best for cuesta-related rotational slides. Further assessment and refinement are needed to extend its applicability to other landslide types. Nevertheless, the method shows strong potential for detecting landslides with distinct geomorphological signatures in high-resolution DTMs worldwide.
Reference: Zangana, I., Bell, R., Drăguţ, L., Sîrbu, F., and Schrott, L.: Mapping forest-covered landslides using Geographic Object-Based Image Analysis ( GEOBIA ), Jena region , Germany, Nat. Hazards Earth Syst. Sci., 25, 4787–4806, https://doi.org/10.5194/nhess-25-4787-2025, 2025.
How to cite: Zangana, I., Bell, R., Drăguţ, L., and Schrott, L.: Transferability of Semi-Automatic Landslide Mapping Approach Using High-Resolution DTMs: a Case Study from the Swabian Alb, Germany, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11366, https://doi.org/10.5194/egusphere-egu26-11366, 2026.
Translating coarse-resolution climate projections into actionable, city-scale hazard information remains a critical challenge for coastal infrastructure planning worldwide. We present a transferable framework that combines adaptive-mesh numerical modeling with a physically consistent pseudo-global warming (PGW) methodology to generate high-resolution, climate-adjusted tropical cyclone (TC) scenarios. Here, we employ the CPAS (ClusterTech Platform for Atmospheric Simulation) model at variable resolutions (96-48-24-12-3 km), coupled with bias-corrected CMIP6 data under SSP5-8.5 forcing. Climate perturbations are applied using a physically consistent approach that also helps reduce model spin-up. The methodology incorporates a scale-aware physics scheme specifically validated for TCs. It bridges the scale gap between global climate models (~100 km) and decision-relevant hazard assessment (~1 km), offering a pathway applicable to coastal megacities globally.
We demonstrate the framework using five representative TCs impacting the South China coast during 2008-2021, spanning a range of intensities, sizes, and approach characteristics. Historical control simulations accurately reproduce observed storm tracks and structures, establishing confidence in the climate-perturbed scenarios. Systematic climate change signals emerge across the event portfolio: (1) variable intensity amplification (3.1-8% °C⁻¹ climate sensitivity), dependent on storm structure, with the strongest storms exhibiting the largest response; (2) nonlinear precipitation enhancement, with median increases of 30-35% and amplification up to 50% at extreme percentiles; and (3) diverse structural responses, with some storms contracting while others expand their damaging wind field.
Event-to-event differences (e.g., initial intensity, storm size, track angle, and rapid intensification) drive diverse climate responses, making uniform adjustment factors potentially misleading. The framework provides physics-based, scenario-specific hazard simulations at 3 km resolution (extendable to < 1 km), directly linkable to exposure databases for “what-if” stress-testing of historical events under future climate conditions. Although demonstrated for TCs, the framework is transferable to other storm types and regions, with adaptive meshing enabling efficient, decision-relevant hazard modeling over complex coastal terrain.
How to cite: Zheng, Y., Tam, C., Cheung, C.-C., and Sze, W.-P.: Integrating Climate Models and Coastal Risk Assessment in relation to Tropical Cyclones using an Adaptive Mesh Framework , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4650, https://doi.org/10.5194/egusphere-egu26-4650, 2026.
Probabilistic precipitation, such as the 100-year rainfall, is widely used as the design storm for planning flood control structures. However, due to climate change, the return periods estimated 50 years ago are no longer valid. This shift necessitates a fundamental reconsideration of how we determine design levels for construction. In other words, there is an urgent need for more sophisticated methodologies capable of handling non-stationary precipitation data.
To address these challenges, we present two key topics:
-
In Japan, the national and local governments have issued guidelines suggesting that future extreme rainfall intensities can be estimated by multiplying present-day values by a change factor of 1.1 to 1.15, assuming a 2.0°C increase in global temperature. While these guidelines tend to treat the change factor as largely uniform across regions for practical simplicity, we contend that it should be estimated with greater geographical precision. Consequently, we estimated the change factors specifically for Mie Prefecture in central Japan. Our results demonstrate that even within a single prefecture, the factor varies significantly depending on the specific location.
-
We have been developing a novel approach to estimate future 100-year precipitation levels through multivariate analysis. The details of this methodology and our findings will be presented in our poster session.
How to cite: Kurata, M., Hasegawa, M., Mizuki, C., and Kuzuha, Y.: Estimation of Future 100-year Precipitation in Mie Prefecture, Japan, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-16126, https://doi.org/10.5194/egusphere-egu26-16126, 2026.
Landslides are among the most destructive natural hazards in the Himalayan region, where steep terrain, complex lithology, heterogeneous soil cover, and intense hydro-meteorological forcing collectively govern slope instability. Despite growing recognition of ARs as major contributors to extreme rainfall, their explicit integration into physically informed slope stability assessments in the Himalayas remains limited. This research aims to investigate the impact of atmospheric-river-driven precipitation on slope stability across the Himalayan region by coupling landslide inventory data, soil characteristics, topographic controls, and slope stability theory. landslide occurrences are analyzed with respect to topographic parameters derived from digital elevation models, such as slope angle, elevation, and terrain morphology. Given the limited availability of site-specific geotechnical data over large mountainous regions, soil mechanical properties specifically cohesion and angle of internal friction are inferred from soil type and texture classes obtained from global soil databases. Representative ranges of shear strength parameters are assigned based on established values reported in the literature.
Temporal characteristics of AR events, including shape, movement, intensity, duration, and antecedent moisture conditions, are linked with observed landslide occurrences to identify critical thresholds associated with slope failure. Slope stability is evaluated using the factor of safety (FOS) concept derived from limit equilibrium principles for infinite and shallow slope conditions. The influence of atmospheric rivers is incorporated through rainfall-induced changes in pore-water pressure and effective stress, enabling assessment of strength reduction and progressive destabilization under extreme precipitation scenarios. The outcomes of this research are expected to quantify the degradation of slope stability associated with atmospheric-river-driven rainfall, identify soil and terrain combinations most susceptible to AR-induced failures, and provide a physically meaningful explanation for observed landslide spatial clustering during extreme precipitation events. By integrating atmospheric processes with geotechnical slope stability analysis, this study advances the understanding of hydro-geomorphic hazards in the Himalayas and contributes to improved landslide susceptibility assessment, risk mitigation, and climate-resilient land-use planning in mountainous regions.
How to cite: Raina, B. A. and Nayak, M. A.: Atmospheric Rivers as Triggers of Slope Instability and Landslides in the Himalayas, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9527, https://doi.org/10.5194/egusphere-egu26-9527, 2026.
Landslides are a global geological phenomenon that constitute serious threats for human lives and engineering infrastructure, making the susceptibility assessment of these landslides a critical step for risk mitigation. The Al Haouz province, which was heavily struck by the Mw 6.8 earthquake of 2023, recorded several slope instabilities caused by seismic motion. In this context, the present study aims to evaluate co-seismic landslides susceptibility using machine learning models to support effective risk mitigations.
Logistic Regression LR and Random Forest RF models were employed to generate the susceptibility maps. The landslide inventory map with 302 landslide points and 600 non-landslide points was utilized with a 70:30 split for training/testing purposes. Sixteen conditioning factors were considered in the modelling process.
The results indicate RF performed better than the LR method, with an accuracy of 97.34% compared 92.92% for LR. The area under the curve (AUC) values ranged between 0.98 for LR and 0.99 for RF. reflecting the high predictive capability of both models. Elevation, Slope, PGA and rainfall had the highest contribution scores amongst the factors identified by both models.
The outcomes indicate the effectiveness of machine learning algorithms, specifically the RF model, for susceptibility mapping related to landslides in a seismic area. Elevation and slope are the most important factors influencing landslides from a geomorphological perspective in Al Haouz province. PGA is the most significant parameter among all factors as landslides are primarily triggered by seismic acceleration associated with earthquake events. Rainfall is a significant parameter that triggers landslides as a result of steep slopes associated with heavy rainfall either continuously or with high intensity.
The co-seismic landslide susceptibility maps produced in this study provide valuable information for identifying vulnerable zones and constitute an effective tool for land-use planning and disaster risk reduction aimed at protecting human lives, infrastructure, and the environment.
Keywors: Landslide susceptibility; Al Haouz earthquake; Machine learning; Morocco
How to cite: Edoudi, A., Cherif, S., Ibouh, H., Ahmadian, N., El Wahidi, F., Chourak, M., Kurtz, R., and Dubovyk, O.: Co-seismic Landslide Susceptibility Mapping after the 2023 Al Haouz Earthquake (Morocco) Using Machine Learning, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13589, https://doi.org/10.5194/egusphere-egu26-13589, 2026.
