This session aims to bring together experts, researchers, and practitioners to present and discuss recent developments in the field of flood risk mapping, assessment and management. Topics such as 1D, 2D and 3D modelling for flood risk assessment, emergency action planning and the analysis of dam and levees breaching, as well as the design of structural, non-structural and nature-based measures, are welcome. Research on the associated uncertainties, sensitivity analysis, and flood impact modelling is also relevant to the session.
Orals: Thu, 7 May, 14:00–17:38 | Room 1.15/16
Understanding multi-hazard interactions in flood risk contexts has emerged as a critical area of environmental research. This study presents a comprehensive bibliometric analysis covering scientific publications indexed in Web of Science from 2010 to 2024, focusing specifically on compound and cascading flood events within multi-hazard frameworks. The dataset comprises 1,096 scientific documents published across 290 academic sources, authored by 4,166 scholars affiliated with 1,326 institutions from 94 countries. Annual scientific production increased significantly, rising from fewer than 10 publications per year prior to 2018 to a peak of 78 documents in 2023. Geographically, the literature is highly concentrated; China (27%) and the United States (24%) dominate publications, whereas contributions from African institutions remain minimal, reflecting critical geographic disparities in research engagement. Using paper-level counting, 505 papers (46%) include at least one European affiliation, based on a European country set of Austria, Belgium, France, Germany, Greece, Italy, the Netherlands, Norway, Poland, Portugal, Romania, Serbia, Spain, Sweden, Switzerland, and the United Kingdom. European output is led by the United Kingdom (145 papers) and Italy (134), followed by the Netherlands (94) and Germany (83). A detailed keyword evolution analysis identified a pronounced thematic shift from vulnerability and traditional hazard management toward compound-flood risk, climate-driven extremes, and resilience-oriented approaches in recent years. Topic-cluster analysis further demonstrates fragmented research efforts across methodological domains, with limited interdisciplinary integration. Citation trajectory analysis reveals that key studies published between 2022 and 2023 received more than 45 citations within two years, indicating a rapid recognition and scholarly impact within the research community. However, uncertainty quantification remains notably underrepresented (only 17% of studies), and emerging approaches such as nature-based solutions constitute less than 5% of the total literature, underscoring key research gaps. These findings offer a robust, empirical foundation highlighting underexplored themes, geographic disparities, and methodological challenges, guiding future research priorities in multi-hazard flood risk management under changing climatic conditions. International collaboration is a defining feature of the European contribution. Almost half of the European papers also include non-European affiliations (245 of 505). The strongest links are with the United States (89 papers) and China (56 papers), while 16 papers include Europe, the United States, and China together. In full counting, European countries account for 39.88% of all countries’ occurrences in the dataset (737 of 1,848). These results position Europe as both a major producer and a collaboration hub in multi-hazard flood risk research.
How to cite: Gul, E., Geykli, A. N., and Hassanzade, E.: Bibliometric analysis of global multi-hazard flood risk research: trends, knowledge gaps, and emerging topics (2010–2024), EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21974, https://doi.org/10.5194/egusphere-egu26-21974, 2026.
Azerbaijan faces significant flood risks due to its diverse terrain, which includes low-lying areas near the Caspian Sea, situated 28 meters below sea level, and mountainous regions exceeding 4,000 meters. Increasing rainfall exacerbates this threat, while current monitoring systems are inadequate for timely flood warnings. This study introduces HydroAlert Azerbaijan, an artificial intelligence-driven system utilizing satellite imagery to detect floods and assess risk levels, capable of operating in adverse weather conditions. It employs a U-Net neural network to analyze Sentinel-1 SAR data, specifically leveraging VV and VH channels for efficient flood assessment.
The system processes 512×512 pixel tiles from the SAR data, overlapping by 64 pixels to ensure comprehensive coverage. Trained on the SEN12FLOOD dataset, consisting of 209 global flood examples, HydroAlert is designed to function effectively even in areas with limited flood event data. Initial evaluations of the Sentinel-1 SAR scenes and Azersky optical images confirm its efficacy, achieving an accuracy of approximately 85% in flood identification.
The Azersky optical data, characterized by a resolution of 1.5 meters, provides detailed insights into infrastructure vulnerability and validates the extent of floods derived from SAR data. The model generates precise vector shapes on maps, improving emergency response planning by visualizing flood extents.
This study shows that a platform facilitates user interaction with flood data, incorporating historical insights from the Dartmouth Flood Observatory and high-risk area alerts. The system supports data export in user-friendly formats to assist decision-making. The Hydroalert Project, which integrates SAR and optical data sources for comprehensive flood assessment, ensures reliable monitoring capabilities through its multi-sensor integration framework.
Additionally, the system incorporates a forecasting module using ConvLSTM architecture to predict flood risks over the following week, aiding proactive decision-making in disaster preparedness. Participation in the Azercosmos Earth Observation Competition 2025 has fostered collaboration with the Azerbaijani Space Agency, leading to systematic enhancements of the HydroAlert prototype using data from the competition.
Current efforts focus on refining the model for local conditions, utilizing satellite imagery to improve operational accuracy. This project demonstrates the potential of deep learning models for flood detection in developing regions lacking robust ground-level monitoring systems. By integrating global satellite images with advanced AI techniques, HydroAlert Azerbaijan offers a viable flood monitoring and management strategy for areas with limited existing resources and information.
Keywords
Flood mapping, SAR remote sensing, Optical imagery, Deep learning, Azerbaijan, Disaster management
How to cite: Elik, F., Rustamov, P. R., and Alaskarov, E.: AI-Powered Flood Monitoring for Azerbaijan Using Multi-Source Satellite Data: Operational Prototype Development and Initial Validation, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21567, https://doi.org/10.5194/egusphere-egu26-21567, 2026.
Floods account for more than 40% of recorded natural disasters in the last two decades (EM-DAT). Given the current climate dynamics, flooding is likely to increase in intensity with shorter return periods. Flood risk is highest in the Global South countries due to lack of adequate prevention, prediction and monitoring systems.
However, the evolution of earth observation technology over the last decade carries a great potential for accurate, near real-time flood extent mapping and prediction. While this favors synthetic aperture radar (SAR) based approaches due to its ‘all weather’, day or night and canopy penetrating capabilities, the flooded scene presents a complex environment characterized by mixed signal to surface interaction mechanisms that complicate SAR imagery analysis. This necessitates fusion with other remote sensing flood detection techniques.
This study aims to develop an earth observation data fusion model for near-real time flood detection, early warning and risk as well as damage assessment in data-limited areas that fall within the exclusion mask of the Copernicus Global Flood Awareness System (GloFAS). The area of study is Tana Delta Ramsar Site in Kenya, a productive ecosystem comprising of a unique mix of fresh water, floodplain, estuarine areas and beaches supporting several ecosystem services. This analysis focuses on the April 2024 flooding event.
The initial step involved comparative analysis of three Sentinel-1 (S1) flood detection techniques namely image segmentation (Otsu threshold), multi-temporal change detection (CD) and a hybrid (Otsu + CD) technique, post-processed for removal of permanent water as well as slope and spatial-context conditions. With the limitation of missing validation data, Sentinel-2 (S2) image classification was used albeit with a 3-day acquisition date misalignment between the post-flood S1 and S2 images thus assuming no significant land cover changes took place.
Preliminary results show higher flood areas detected by the VH vis-à-vis the VV channel. In particular, the Otsu detected 240.85 km2 for VH and 196.98 km2 for the VV while the CD returned 40.64 km2 and 27.56 km2 for the VH and VV respectively with a change threshold of 1.5. Lastly, the hybrid approach detected 143.66 km2 for the VH and 47.33 km2 for the VV against S2’s 194.29 km2. This difference could be due to the depolarization of the VV backscatter in the vegetated areas.
The next steps will involve testing of operational workflows such as the Copernicus Global Flood Monitoring (GFM), UN-SPIDER recommended flood detection approach and the AUTOWADE 1.0 (AUTOmatic Water Areas DEtector) in a data limited context (Tana Delta, April 2024 flood event). Further, the Otsu threshold required initialization of a bimodal histogram for best performance while the CD results rely on the chosen change threshold hence are not suitable for automated inundation mapping. Consequently, the study will explore the use of AlphaEarth and Clay geo-foundation models for rapid flood mapping in complex land use/ land cover and data-limited areas.
Key words: SAR, Data fusion, Flood extent mapping, geo-foundational models
How to cite: Muriithi, D. M. and Vitti, P. A.: Earth Observation Data Fusion for rapid flood extent mapping in data-limited areas: A Case Study of Tana River Delta, Kenya, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-110, https://doi.org/10.5194/egusphere-egu26-110, 2026.
Floods in large transboundary basins such as the Brahmaputra pose persistent threats to lives, livelihoods, and infrastructure. Flood event database is initially created utilizing gridded routed streamflow simulations. For each event occurrence, we extract a set of flood attributes including peak discharge, flood volume, duration metrices. These hydrologic characteristics are integrated with an inundation component, enabling the Flood Severity Index (FSI) to represent not only the intensity of flooding within the channel but also the amount and duration of inundation across the adjacent floodplain. Utilizing this index, we provide a data-driven machine learning framework to predict RFSI throughout the basin. Predictor variables include key hydro-climatic inputs such as temperature, precipitation, which collectively influence the generation and evolution of flood events. Multiple machine learning models were evaluated using performance metrics including R², RMSE, MAE, and cross-validation, all of which demonstrated strong predictive skill across diverse hydrologic regimes, establishing the proposed data-driven framework as a scalable and computationally efficient tool for forecasting flood severity. This approach offers a strong basis for evaluating future flood scenarios and understanding how climate change may alter flood risk, especially in large transboundary regions with limited observational data.
How to cite: Parashar, T., Chakma, S., and Saharia, M.: A Data-Driven Approach for Predicting Riverine Flood Severity Index in the Transboundary Brahmaputra River Sub- Basin, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-754, https://doi.org/10.5194/egusphere-egu26-754, 2026.
Accurate global flood forecasting, particularly in ungauged basins, remains a primary challenge for operational hydrology and disaster risk reduction. While data-driven approaches using Long Short-Term Memory (LSTM) networks have set new benchmarks in global streamflow prediction, limitations regarding forecast lead time, temporal consistency, and operational robustness to missing data persist.
In this work, we present the next-generation Google global hydrologic model, which introduces three major advancements over previous state-of-the-art systems. First, we integrated AI-based medium-range weather forecasts as additional meteorological forcing, alongside traditional deterministic products. Second, leveraging recent contributions to the Caravan community dataset, we expanded the training dataset three-fold to include nearly 16,000 streamflow gauges globally. Third, we implemented a novel masked mean embedding LSTM architecture. This design eliminates the traditional encoder-decoder state hand-off issue (which introduces temporal inconsistencies or forecast hairs) and enables the model to remain operational during weather data outages by dynamically averaging embeddings from available input sources.
Our results demonstrate a significant extension of the reliable forecast horizon: the new model achieves accuracy at a 7-day lead time comparable to the 5-day lead time performance of its predecessor. Furthermore, the model continues to outperform other global operational systems, such as GloFAS and GeoGlows, across both gauged and ungauged basins. These advancements represent a significant step toward providing more timely and reliable flood warnings in regions where traditional monitoring infrastructure is scarce.
In conjunction with this update, we released two new community resources. The Google Runoff Reanalysis & Reforecast (GRRR) dataset provides a comprehensive, multi-decade reforecast archive generated by the current operational global model. Additionally, we have launched the GoogleHydrology GitHub repository, which provides an open-source research implementation that closely approximates our operational environment. This release is intended to facilitate the reproduction of our findings and provide the scientific community with a robust baseline for future global hydrologic modeling research.
How to cite: Loike, G., Nearing, G., and Cohen, D. and the Google Research - Floods Forecasting: The Next-Generation Google Flood Forecasting Model & Community Resources, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6973, https://doi.org/10.5194/egusphere-egu26-6973, 2026.
The MOVIDA (Modello per la Valutazione Integrata del Danno Alluvionale) project, launched in 2020, aims to develop GIS-based procedures for the quantitative and qualitative assessment of exposure and damage to assets (e.g., residential buildings, transportation networks, agricultural areas) located within the Po River Basin District Authority (AdBPo) and potentially affected by flood hazard. These procedures were implemented in the QGIS Graphical Modeler as automated geoprocessing workflows integrating multiparametric models for damage and exposure evaluation.
This development phase was supported by the construction of a district-scale geodatabase of exposed elements, obtained through the harmonization and integration of national and institutional datasets, as well as additional open-source datasets (e.g., OpenStreetMap).
In the second phase of the project, a modular and service-oriented WebGIS platform was designed and implemented by leveraging open-source geospatial software technologies to enable the dissemination and operational use of the developed methodology. The platform provides end-to-end support for the ingestion, processing, and analysis of both institutional flood hazard maps derived from the Flood Risk Management Plans (FRMPs) and user-defined flood scenarios. All input data are persisted within a spatial database, i.e., PostGIS and programmatically coupled with the processing and computation modules. The execution of the processing workflows is asynchronous, and users are automatically notified upon completion via email-based alerts.
The system provides both interactive visualization of the outputs and access to the generated datasets.
The MOVIDA platform is based on a containerized architecture composed of three Docker services: (i) a Redis service for caching, (ii) a PostgreSQL/PostGIS service for data storage, and (iii) an application service integrating Django and QGIS. Django manages the entire web application layer, including the web interface, user interactions, and notification services, while QGIS performs the geoprocessing tasks. The results are then returned to the web application and made available through the user interface.
How to cite: Zazzeri, M. and the MOVIDA Team: The MOVIDA platform: a WebGIS tool for the assessment of flood risk–related impacts, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-17715, https://doi.org/10.5194/egusphere-egu26-17715, 2026.
Two-dimensional hydrodynamic models are computationally expensive. This drawback can limit their application to solving problems requiring real-time predictions or several simulation runs. To resolve fine-scale physical processes, allowing for local impact assessments, downscaling techniques are essential. Super-resolution is an innovative technique that upscales the resolution of an image and thus enables to reconstruct high-fidelity images from low-resolution data. This study performs super-resolution analysis for spatial downscaling of hydrodynamic data using various deep learning techniques to reconstruct high-resolution flow fields from low-resolution flow field data. It increases the spatial resolution of coarsened water depth and unit discharge norm from 4 m to 80 cm. The training data for these models was generated using a physically based hydrodynamic model. To evaluate their performance and accuracy, multiple tests were conducted using synthetic events. Our experiments indicate that these models successfully predicted water depths in the testing flood scenario for the dynamic case but could not preserve the steady states during the reconstruction. Furthermore, these models cannot satisfactorily generalize to flood scenario outside the training datasets with different boundary conditions. The results demonstrate that the proposed models are up to 30 times faster than the hydrodynamic model and promising in terms of accuracy. Therefore, it bridges the gap between detailed flood modelling and real-time applications.
How to cite: Ait-Ameur, K., Guinot, V., Marti, L., Rousseau, A., and Toulemonde, G.: Exploring super-resolution for the downscaling of urban floodsimulations, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21384, https://doi.org/10.5194/egusphere-egu26-21384, 2026.
Flooding is one of the most frequent and destructive natural disasters in Japan, particularly in river basins with high population density and urbanization. This study aimed to use a GIS-based probabilistic Certainty Factor (CF) model to evaluate flood susceptibility in the Arakawa River basin, Japan. Nineteen flood conditioning factors were incorporated: soil, land use/land cover (LULC), normalized difference built-up index (NDBI), normalized difference water index (NDWI), normalized difference vegetation index (NDVI), curvature, elevation, precipitation, slope, topographic position index (TPI), sediment transport index (STI), stream power index (SPI), topographic wetness index (TWI), drainage density (Dd), distance to streams, distance to roads, flow accumulation, population, and aspect were included to assess their impact on flood frequency. A flood susceptibility map (FSM) was generated by applying the Certainty Factor model. A total of 230 flood locations within the study area were examined and geostatistically processed in ArcGIS for model validation. The resulting FSM was categorized into five susceptibility classes: very low, low, moderate, high, and very high. The spatial distribution of these classes showed that 22.5% of the area falls under moderate susceptibility, 37.1% under high, and 22.6% under very high susceptibility. The model's performance was evaluated using the area under the receiver operating characteristic curve (AUC), yielding an accuracy of approximately 71%. The results indicate that the most influential factors affecting flood susceptibility in the basin are elevation, stream power index (SPI), sediment transport index (STI), flow accumulation, and distance to roads. The suggested framework offers useful spatial insights that can assist in supporting decision-makers to reduce both economic losses and risks to human life.
How to cite: M. Elsadek, W., Ahmed, H. S., and Kanae, S.: Probabilistic flood susceptibility assessment using a GIS-based Certainty Factor approach in the Arakawa River basin, Japan, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-2015, https://doi.org/10.5194/egusphere-egu26-2015, 2026.
Accurate flood inundation modelling in monsoon-dominated regions remains fundamentally limited by the scarcity of discharge observations, particularly along small tributaries that contribute substantially to flood peaks. Operational hydraulic models such as LISFLOOD-FP typically use boundary conditions derived from a small number of main-stem gauging stations. Consequently, floodplain dynamics in headwater and fringe zones are systematically underestimated, especially in peninsular India where ungauged tributaries and side valleys can supply 20–40% of peak flow during extreme rainfall events. This omission introduces major errors in hazard assessment and reduces the usefulness of model outputs for early warning and risk preparedness. This study presents a data-efficient geomorphic–recession-based method for reconstructing discharge hydrographs for ungauged tributaries without requiring additional gauge infrastructure. The approach integrates three components: (1) Recession constant estimation from CAMELS-IND catchments using baseflow separation and multi-event recession analysis; (2) Geomorphic prediction of tributary-specific recession behaviour based on drainage area, basin slope, and land-cover characteristics, enabling flow recession prediction where gauge data are unavailable; (3) Event-based hydrograph disaggregation and scaling, where synthetic hydrograph shapes are generated and apportioned across tributaries according to drainage area ratios and their predicted recession behaviour. Reconstructed tributary hydrographs are automatically introduced into LISFLOOD-FP as distributed lateral boundary conditions at tributary–floodplain junction nodes, enabling both main-stem and tributary-driven flood dynamics to be simulated simultaneously. The framework is tested on multiple monsoon flood events across peninsular India. LISFLOOD-FP simulations are conducted under two forcing scenarios: (i) conventional configuration using only main-stem discharge data, and (ii) the proposed distributed tributary inflow scheme. Model outputs are evaluated against Sentinel-1 SAR flood extent maps, which provide an independent, satellite-based benchmark of observed inundation patterns. Key performance metrics include spatial correspondence (F1-score, precision, recall), headwater and fringe-zone inundation extent, and the model's ability to capture compound flooding behaviours (e.g., tributary–main-stem surge interactions). This work provides the first operational framework for generating tributary-scale discharge inputs for flood inundation models in data-scarce monsoon basins using only regional hydrological signatures and topographic data. The method is computationally efficient, scalable across complex tributary networks, and requires no additional hydrometric infrastructure. By explicitly representing ungauged tributary forcing, the approach aims to substantially improve flood hazard mapping and forecasting in regions that are often highly exposed to tributary driven flooding yet poorly resolved in existing operational models. The framework offers a practical and transferable pathway for enhancing flood early warning systems in peninsular India and comparable monsoon-affected regions globally.
Keywords: CAMELS-IND; LISFLOOD-FP; recession analysis; Sentinel-1; ungauged tributaries.
How to cite: Tripathi, V., Singh, H., and Mohanty, M.: Ungauged Tributary Discharge Reconstruction for Monsoon Flood Inundation Modelling: A Geomorphic-Recession Approach Applied to Peninsular India, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-815, https://doi.org/10.5194/egusphere-egu26-815, 2026.
Historical flood events provide critical insights into extreme flood dynamics that are often underrepresented in instrumental records. This contribution presents the application of hydrodynamic modelling to analyse major historical flood events in the Iberian Peninsula and to assess their relevance for flood risk assessment and mitigation. Several catastrophic floods affecting Portugal and Spain between the late 19th and the 20th centuries are investigated, including riverine and flash-flood events with severe societal impacts. A modelling framework is implemented using the physically based Iber+ hydrodynamic model. It integrates precipitation reconstructed through multiple methodologies, including interpolation of gridded and measured data, and topography compiled using different sources, ranging from historical maps to field-based measurements. This framework enables the estimation of peak river flows, one of the main unknowns in historical events, and reproduces flood propagation, inundation extent, water depths, and flow velocities. Model performance is evaluated against historical watermarks, documentary evidence, and witness testimonies, showing good agreement despite the scarcity of direct measurements and the associated uncertainties. The simulations enable a detailed analysis of the key drivers controlling flood severity, including the exploration of plausible scenarios, providing insights into the main causes of these events where uncertainties in flood development persist. The results highlight the role of hydraulic bottlenecks, infrastructure blockage, and local topographic constraints in amplifying flood impacts. Scenario-based simulations further demonstrate the potential of hydrodynamic modelling to explore mitigation strategies, such as optimized dam operation and improved infrastructure maintenance, and to assess extreme but plausible flood scenarios under current terrain and infrastructure conditions, supporting flood mitigation. Overall, the results emphasize the value of integrating historical flood information with modern hydrodynamic modelling. This approach contributes to improving flood hazard mapping, reduce uncertainties in flood risk estimation, and support more robust flood risk management strategies under current and future climatic conditions, in which flood events of comparable or even greater intensity than historical floods are expected.
How to cite: Fernandez-Novoa, D., Gonzalez-Cao, J., Garcia-Feal, O., Barreiro-Fonta, H., Trigo, R. M., and Gomez-Gesteira, M.: Hydrodynamic modelling of historical flood events in the Iberian Peninsula: implications for flood risk assessment and mitigation, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-20999, https://doi.org/10.5194/egusphere-egu26-20999, 2026.
Dam breach flooding is low probability yet high consequence hazard particularly in arid and semi-arid regions where urban and rural population is deployed along the ephemeral streams. The study analyses the flood hazard and flood risk assessment of the Wadi Baysh Dam in southwestern part of Saudi Arabia, based on a dam breach analysis as a result of an observed extreme rainfall event happened in July 2011. The target is to evaluate the flood impacts on the downstream under various breach scenarios and to support risk informed mitigation and emergency planning.
Three dam breach scenarios were studied using HEC-RAS 2D hydrodynamic model simulations: (i) full breach (S1), (ii) half breach (S2), and (iii) one-third breach (S3). The 2D simulations produced spatially distributed inundation depth, maximum flow velocity and inundation duration, which were later used to derive composite dam-breach flood hazard maps. By integrating composite flood hazard and landuse map as a vulnerability measure (including urban areas, agricultural land, roads, rangeland, bare ground and tree cover classes), risk maps of dam-breach floods were created and finally exposure analysis was conducted.
Results show strong non-linear relationship between breach severity and downstream impacts. Under full-breach conditions, the total inundated area reaches approximately 371 km², with large portions classified as moderate to high flood hazard. Agricultural land and rangeland exhibit the greatest exposure, while urban areas, although spatially limited, experience locally elevated hazard levels due to high flow depths and velocities. Partial-breach scenarios substantially reduce inundation extents (~278 km² for half breach and ~32 km² for one-third breach); however, hazardous conditions persist along the main wadi channel and low-lying floodplain zones, indicating that partial structural failure still poses significant downstream risk.
Composite flood risk assessment shows that low-to-moderate risk dominates across all scenarios, yet localized high-risk zones emerge near critical infrastructure and densely cultivated areas, particularly under full- and half-breach conditions. The results further demonstrate that reductions in breach size do not translate linearly into risk reduction, underscoring the importance of explicitly considering multiple breach scenarios in dam-safety assessments.
The study highlights the value of scenario-based 2D dam-breach modelling for flood-risk assessment in arid environments, where observed extreme storms can act as credible compound triggers. The proposed framework supports advances in flood-risk modelling by integrating hazard characterization, land-use exposure, and risk classification, providing actionable insights for emergency action plans, evacuation zoning, and long-term risk mitigation strategies.
How to cite: Zaidi, S., Alomary, M., and Al-Gafari, Y. H.: Scenario-Based 2D Hydrodynamic Modelling of Dam-Breach Flood Hazard and Risk under an Extreme Storm, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4751, https://doi.org/10.5194/egusphere-egu26-4751, 2026.
This study analyzes embankment breach outflows resulting from overtopping using probabilistic modeling. The effects of uncertainties in the inflow hydrograph and embankment breach parameters on the breach outflow were evaluated using a numerical model that considers (1) breach parameters, including final bottom elevation, width, side slopes, formation time, weir coefficient, and water surface elevation triggering the breach; and (2) inflow hydrograph parameters, such as peak flow rate and time to peak, as probabilistic variables. The Monte Carlo method was employed to conduct 10,000 simulations for each scenario. Histograms and exceedance probability curves of the resulting peak outflows were generated, and probability density functions were fitted and evaluated using the Chi-square goodness of fit test. It was found that both the range and type of the final bottom elevation distribution significantly influence the breach outflow, with observed values ranging from 353 to 2170 m3/s depending on the parameter combinations. Modeling the inflow as either deterministic or probabilistic did not significantly impact the discharge; however, a normal distribution is recommended for representation. The deterministic breach model yielded a peak outflow that was approximately 40% lower than the maximum value produced by the probabilistic simulations, underscoring the importance of incorporating uncertainty into breach analyses.
How to cite: Gödek Hayal, N. P., Calamak, M., and Yanmaz, A. M.: Assessing the Uncertainty of Embankment Breach Outflow due to Overtopping, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4042, https://doi.org/10.5194/egusphere-egu26-4042, 2026.
Accounting for nearly 30% of all the losses due to natural disasters, flood emerges as one of the most havoc-creating extreme events in the world. Recently, as an adverse effect of climate change, the tropical river basins have witnessed recurring extreme flood events leading to significant devastation of agricultural production. We have studied the climate change-induced flood risk associated with crop damage in a paddy crop-dominated tropical river basin in India. We have considered a 50-year return period design flood and three (2010-2039, 2040-2069 and 2070-2099) future scenarios in the most extreme representative concentration pathway 8.5 conditions in a hydrodynamic modelling framework as the test case. Nine different Global Climate Models (GCMs) are used here. Analysis of the three best performing (HadGEM2-AO, IPSL-CM5A-MR and MIROC-ESM-CHEM) GCM data-driven flood inundation depth and extent, and the associated net loss/benefit from the cultivation of normal rice variety indicates increased flood risk in the projected scenarios as compared to the historical period. In contrast to high water level, occurrence of comparatively low inundation depth but for a longer period of time is found to increase the flood vulnerability of the paddy crops in the future projected time frames. As a significant alteration in the cultivation pattern is highly subjective on the adoption/willingness of the local farmers, we suggest an alternate rice planning, considering cultivation of an alternate rice variety as a probable adaptation strategy to minimize climate change induced flood risk. Considering the near-future period (2010s) and the MIROC-ESM-CHEM model, our study shows that cultivation of shallow, medium deep or deep water rice varieties in high flood inundation areas can reduce the very high flood risk from about 35% to 17%. The methodology adopted herein encourages the application of hydrodynamic modelling in analyzing projected flood-agriculture risk and paves avenues for more novel scientific research.
How to cite: Chatterjee, C., Khatun, A., and Sahoo, B.: Exploring possibilities to reduce climate change induced flood risk and crop production losses using hydrodynamic modelling, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7289, https://doi.org/10.5194/egusphere-egu26-7289, 2026.
Assam, a flood-prone state, comprises multiple national parks and wildlife sanctuaries (Kaziranga, Orang, Manas National Parks, and Pabitro, Laokhawa, and Burachapori Wildlife Sanctuaries) located in the southern part of the Brahmaputra River. Floods significantly influence wildlife in these conserved areas, which are less discussed. The greater one-horned Rhinoceros (Rhinoceros unicornis) is listed as vulnerable on the International Union for Conservation of Nature (IUCN) Red List. While floods are ecologically integral to the Brahmaputra floodplain, extreme and frequent flood events increasingly threaten rhino habitat suitability, mobility, and survival, underscoring the need for spatially explicit risk assessment to support conservation planning.
The present study focuses on mapping the spatial risk index of the one-horned Rhino using Earth observation (EO) based Analytical Hierarchy Process (AHP) in the Google Earth Engine (GEE) platform. The Shuttle Radar Topographic Mission (SRTM) Digital Elevation Model (DEM) is used in conjunction with GEE to generate flood-influencing parameters, including elevation, slope, aspect, flow accumulation, distance to drainage, drainage network, and topographic wetness index. Flood depth, distance from roads, and distance from built-up areas have been used to develop layers for the Flood Risk Index (FRI) in rhino conservation. Rhino locations were collected from the Assam Biodiversity portal.
The primary rhino habitat occupies approximately 23% of the study area, yet a substantial proportion of these habitats overlaps with zones of elevated flood risk. The Flood Risk Index (FRI) indicates that nearly 78% of the region falls within moderate to high flood-risk categories, with several rhino habitat areas consistently exposed to high inundation susceptibility. Spatial overlay analysis highlights critical habitat patches where flood risk, anthropogenic proximity and low-lying terrain converge, and need priority zones for intervention during extreme flood events. The findings provide actionable insights for flood-responsive rhino management, including targeted evacuation planning, habitat restoration, and infrastructure placement, contributing to more resilient conservation strategies under intensifying hydro-climatic extremes.
How to cite: Ghosh, S., Sarkar, U., Kour, S., Bhattacharyya, S., and Nandy, S.: Greater One-Horned Rhino Habitat Risk Mapping Due to Flood Using Google Earth Engine for Informed Conservation Management in Assam, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-19513, https://doi.org/10.5194/egusphere-egu26-19513, 2026.
The "Prediction in Ungauged Basins" (PUB) problem remains a central challenge in global hydrology, as the accuracy of rainfall-runoff models is fundamentally constrained by the availability of local streamflow observations for training and calibration. While recent advancements in data-driven modeling have improved our ability to generalize across catchments, the global distribution of streamflow gauges is characterized by severe geographical and socio-economic biases. Most available data are concentrated in the Global North, leaving vast regions, particularly in the Global South, functionally "ungauged" or under-represented in the training sets of global models.
In this study, we shift the focus from simply counting the fraction of ungauged watersheds to estimating the quantitative effect of this geographical bias on global flood forecasting skill. Using a large-sample machine learning framework based on Google’s flood forecasting model, we quantify the relationship between gauge network density (specifically upstream and downstream coverage fractions) and predictive performance. We utilize cross-validation experiments to isolate the information loss associated with geographical distance and hydrological connectivity from gauged locations.
Our analysis indicates that hydrological factors are the main driver of predictive performance, with basin aridity being a larger factor in model skill than whether a basin is gauged or ungauged. However, if streamflow gauges were hypothetically installed in all of the world’s watersheds, we could expect a 20% increase in the Nash-Sutcliffe Efficiency (NSE) skill score for state-of-the-art global models, including causing almost half of the basins globally currently scoring below NSE = 0.50 to rise above that threshold, with an average skill improvement of about ΔNSE = 0.1. Critically, this potential for improvement is not uniform, with Africa being the continent where our model predicts that largest overall skill improvement with higher density gauging networks.
These findings emphasize that the path toward equitable global flood safety requires not just better algorithms, but a concerted effort to address the structural biases in the global hydrological data ecosystem.
How to cite: Nearing, G., Gauch, M., and Rothenberg, J.: The Effect of Geographical Bias in Streamflow Gauge Distribution for Global Flood Forecasting, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13530, https://doi.org/10.5194/egusphere-egu26-13530, 2026.
Flood susceptibility mapping in the steep and geomorphically complex Himalayan terrain remains inherently challenging due to sparse observational data, strong process nonlinearity, and uncertainty embedded in judgment-based decision frameworks. Conventional Multi-Criteria Decision Making (MCDM) approaches, most notably the crisp Analytic Hierarchy Process (AHP), rely on deterministic pairwise judgments that inadequately represent the vagueness, subjectivity, and cognitive bias associated with hazard assessment in high-relief mountain environments.
This study addresses these limitations by systematically comparing classical AHP with a Fuzzy AHP (FAHP) framework for flood susceptibility mapping in the data-scarce Rudraprayag District of Uttarakhand, India (1984 km²), a region frequently impacted by extreme hydro-meteorological events. Thirteen geo-environmental conditioning factors were integrated within a GIS environment at 30 m spatial resolution, encompassing topographic attributes (elevation, slope, curvature, aspect), hydrological indices (HAND, TWI, drainage density, distance to river), and environmental controls (rainfall, geology, LULC, NDVI, distance to roads). To robustly handle the full 13×13 comparison matrix and avoid zero-weight artifacts commonly associated with fuzzy extent analysis, FAHP was implemented using Buckley’s geometric mean method with triangular fuzzy numbers, explicitly capturing uncertainty bounds in pairwise comparison judgments.
Results demonstrate that FAHP yields a smoother and more balanced weight distribution compared to crisp AHP. While slope remains the dominant control, its rigid dominance is reduced, allowing geomorphically subtle yet physically meaningful factors such as curvature and aspect to exert greater influence. Validation against an independent flood inventory derived from Google Earth Engine, evaluated using ROC–AUC analysis, confirms the superior predictive performance of FAHP (AUC = 0.837) relative to classical AHP (AUC = 0.806).
Overall, the findings highlight that incorporating fuzzy uncertainty into MCDM frameworks significantly enhances the robustness and defensibility of flood susceptibility assessments. FAHP thus provides a more uncertainty-aware and process-sensitive hazard baseline, particularly suited for data-scarce Himalayan regions where judgment-based weighting remains unavoidable in disaster risk reduction and spatial planning.
How to cite: Yadav, S. and Kumar, P.: Quantifying Uncertainty in Himalayan Flood Susceptibility Mapping: A Comparative Analysis of AHP and Fuzzy AHP in Rudraprayag District, Uttarakhand, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-20654, https://doi.org/10.5194/egusphere-egu26-20654, 2026.
Due to climate change, previously safe buildings will face flood risk, while for others risk will grow. The trend can be inverted by retrofitting buildings to achieve low losses (e.g., economic loss, downtime). Retrofit should focus on property flood resilience (PFR) measures to restricting water entry (e.g., flood skirts) and/or reducing its impact (e.g., water-resistant plasters). Currently, PFRs are selected with prescriptive checklists, rather than based on an explicit model of building water ingress (only possible with research-oriented computational fluid dynamics). Consequently, there is no guarantee on the effectiveness of the selected PFRs, nor the trade-off between their cost and reduced future losses. This work shows the preliminary implementation of a simplified water ingress model overcoming this gap.
For a selected hydrograph (i.e., time-variant depth and velocity of the exterior water), water ingress through a building envelope is modelled with a 1D dynamic flow model, using a quasi-steady, fixed-step, explicit Euler scheme. Each ingress pathway is treated as an orifice-like opening, with flow regulated by both hydrostatic water head difference and a velocity-dependent correction to account for drag effects. Calibration of the opening areas and discharge coefficients is based on available experimental data. PFR measures are explicitly considered: for example, a waterproofing membrane renders inactive all orifices below a certain height, while causing a sudden influx if a calibrated pressure strength is exceeded. After aggregating the flows, the interior water height is calculated separately for the building and the basement using mass conservation.
The model is illustrated for an archetype consistent with a ~1980s terraced masonry building in the United Kingdom. Its materials, plan dimensions, height, and water entry points are characterised according to relevant statistics. Inventories of finishes and contents are derived using public commercial listings (e.g., Zoopla). Apart from the as-built configuration, three retrofit solutions are defined considering combinations of PFR measures (e.g., waterproofing membranes, self-closing airbricks, flood doors/windows, non-return valves, raising power sockets and contents). The flood hazard profile is characterised consistently with an ideal site exceeding 1% annual flood probability. Using several hydrographs, the probability distribution of peak interior water depth is computed. The results are used as inputs of an analytical, component-based flood vulnerability assessment. Expected annual economic losses are finally calculated and compared.
The preliminary results show that this approach is simple enough for the early design phase yet accurate enough to allow identifying the marginal benefit/cost of PFR measures. However, benchmarking against refined computational fluid dynamics models is identified as a required step to fully validate the approach and generalise its results.
How to cite: Gentile, R.: Modelling flood water ingress in buildings: towards a simplified, orifice-based hydraulic model, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-2984, https://doi.org/10.5194/egusphere-egu26-2984, 2026.
Flash floods are among the most difficult natural hazards to deal with because they are usually associated with short, intense, and localised rainfall events in small, ungauged catchments. Their frequency and associated impacts have increased over the last few decades due to climate change impacts on extreme precipitation and land-use changes. The Northern Apennines River Basin District (Autorità di bacino distrettuale dell’Appennino Settentrionale - ADAS in Italian), which covers much of the river basins in Tuscany and Liguria, is particularly prone to such hazards due to its steep topography and proximity to the Tyrrhenian coast. Indeed, many flash flood events have occurred recently in these regions.
Because systematic observations of flash floods are scarce, especially in small river basins, regional-scale approaches are essential to support decision-making and to identify susceptible areas. In this context, the ADAS developed a rapid assessment procedure, known as the “Metodo Arno”, to map flash flood-prone areas at the management-district scale, which has been recognised as a best practice and adopted at the national level. Recently, a joint agreement between ADAS and the Department of Civil and Environmental Engineering of the University of Florence has been signed to further improve and harmonise the method.
In its recent version, the “Metodo Arno” is a multi-criteria flash flood susceptibility index derived from three indicators: the time of concentration and the average curve number of the river basin, and the return period associated with 50 mm in an hour. These three indicators are selected because they represent, respectively, the subbasin's response time, its propensity to generate runoff, and the frequency of extreme events. The indicators are normalised to a common scale and then combined using Simple Additive Weighting (SAW) to map the Flash Flood Potential index. All the river basins in the ADAS were delineated from a 10-meter-resolution Digital Elevation Model, with a minimum watershed size of 5 km2. The time of concentration for each river basin was extracted using the Lekan software, which applies the more commonly used formulas in the literature. The return period associated with 50mm/h was estimated using local Intensity-Duration-Frequency curves. The curve number was estimated by combining information on the hydrological soil group and the Corine Land Cover.
A flash flood database has been generated using an automated online news search with keywords to identify locations where past flash flood events occurred. The database is used to validate the procedure. The results show that the approach can map river basins prone to flash flooding. Future steps will focus on assessing the impact of climate and land-use change on the Flash Flood Potential Index.
How to cite: Lompi, M., Sadun, S., Checcucci, G., Spicchi, R., Franceschini, S., and Caporali, E.: Regional-scale mapping of flash flood susceptibility using a simple and operational approach, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-20146, https://doi.org/10.5194/egusphere-egu26-20146, 2026.
India's low-lying floodplains and heavy monsoon rains cause frequent flooding in the Ghaghara River basin. Each year, these floods catastrophically damage infrastructure, agriculture, and human life. To effectively mitigate these impacts, understanding flood dynamics through precise and timely assessment techniques is crucial. Therefore, the study combines Machine Learning (ML) with hydrological and hydraulic models to create a strong modeling chain. Using historical hydrological and meteorological data, the LSTM model is trained to reconstruct continuous streamflow in an ungaged basin. The Soil and Water Assessment Tool (SWAT) then utilizes the ML-derived outflows to support model validation. The predicted runoff in SWAT is used in the HEC-RAS model to assess urban flood inundation and depth in the basin. The ML model achieved a good result for both training and testing. Similarly, the SWAT model demonstrated reliable performance, with a validation accuracy of 0.71 and a calibration accuracy of 0.75, making the model's results suitable for further analysis and interpretation. The available water level and flood depth were used to validate the HEC RAS flood result, which demonstrated satisfactory results. The hybrid ML, hydrological, and hydraulic approach effectively identifies vulnerable flood zones in the Ghaghara basin, thereby improving the accuracy of streamflow, runoff, and flood inundation predictions. The framework supports more efficient planning and mitigation efforts, offering a dependable method for flood assessment in areas with limited data.
How to cite: Bashir, M., Berama, S. M. A., and Ahmad, R.: Hybrid Machine Learning with Hydrological and Hydraulic Models for Runoff Prediction and Flood Risk Assessment in Ghaghara Basin, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-646, https://doi.org/10.5194/egusphere-egu26-646, 2026.
Fluvial floods are one of the most common types of flooding worldwide. Therefore, accurate flood prediction is essential for effective flood preparedness and risk management. This study investigates the prediction of fluvial flood extent under three flood scenarios (Q10, Q100, and Q1000) using deep learning (DL), in particular, the U-Net model. The U-Net model was trained on official flood maps, created as part of the second cycle of the EU Flood Directive (2007), along with seven high-resolution predictors derived from the LiDAR DEM (1 m resolution), orthophotos (20 cm resolution), and ZBGIS spatial database: slope, stream power index (SPI), topographic wetness index (TWI), height above the nearest drainage (HAND), distance from river, roughness, and normalized difference vegetation index (NDVI). Multicollinearity among predictors was tested using the Pearson correlation and Variance Inflation Factor (VIF) with thresholds for Pearson correlation ≤0.7 and VIF ≤5. The model performance was evaluated using three quantitative metrics (Recall, Precision, and F1-score) and training time. The study focused on four river sections in Slovakia (Kysuca, Gidra, Torysa, and Topľa). In each U-Net application, three sections were used for training and one for testing and performance evaluation. The results indicate that the highest model performance was achieved when predicting flood extents on river sections that were similar in width and length. This was particularly evident in the cases where the training/testing river sections included combinations Torysa/Kysuca, Topľa/Kysuca or Kysuca/Torysa. When DL models were trained on narrow/short river sections and then tested on wider/longer sections, the number of false negative (FN) pixels tends to be high. Conversely, when these models are trained on wider/longer river sections and tested on narrow/short ones, the number of false positive (FP) pixels increases. Based on these findings, we recommend avoiding both of these training/testing strategies when transferring the prediction of fluvial flood extent across distinct river sections. Furthermore, the optimized U-Net model showed relatively fast training times with the maximum equal to 15 minutes. The findings of this study suggest strong potential for near-real-time or even real-time flood mapping and operational use. Acknowledgment: Funded by the EU NextGenerationEU through the Recovery and Resilience Plan for Slovakia under the project No. 09I03-03-V03-00085.
How to cite: Vojtek, M., Držík, D., Kapusta, J., and Vojteková, J.: Evaluating the river-to-river transferability of deep learning-based fluvial flood extent predictions, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-3447, https://doi.org/10.5194/egusphere-egu26-3447, 2026.
As urbanization and climate change accelerate, extreme flood events in urban areas have significantly increased, posing a major threat to human and property safety. As an efficient research tool, numerical simulation technology has shown significant application value in mitigating the impacts of urban flood disasters. This study integrates multiple data sources, including Digital Elevation Models (DEM), topographical features, underground sewer systems, rainfall intensity, water level dynamics, and pump station operations, to construct an urban flood inundation simulation. Additionally, using the incipient velocity formula, the dynamic flood risk levels for humans and vehicles were quantitatively analyzed. The main results include: 1. The numerical model accurately simulated the hydraulic characteristics of flood events. The results indicating high model reliability and providing a solid foundation for subsequent risk assessments. 2. During peak rainfall periods, the risk level for humans and vehicles escalates significantly. After the peak, the slight risk for humans decreases, while the magnitude of extreme risks in later stages becomes more severe with larger rainfall return periods. Conversely, the flood risk for vehicles steadily increases, surpassing that of humans overall. 3. In the later stages of rainfall events, both humans and vehicles encounter extensive areas where water depths exceed danger thresholds, transforming them into extreme risk areas. The results obtained in this research contribute to enhancing public awareness of urban flood risks and revealing the spatiotemporal evolution of these risks. They also provide important theoretical support and practical guidance for enhancing urban resilience and promoting sustainable development.
How to cite: Zhu, W. and Xu, Z.: Risk Assessment of Human and Vehicle Stability in Extreme Weather Events in Coastal Cities, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4726, https://doi.org/10.5194/egusphere-egu26-4726, 2026.
This study develops a high-resolution, grid-based explainable machine learning (XAI) framework to systematically identify dominant flood-influencing factors across Jeju Island, Korea, by integrating historical flood trace maps with multi-source spatial datasets. Flood occurrence was classified at a 100 m grid resolution using four state-of-the-art tree-based ensemble algorithms, enabling robust modeling of nonlinear interactions between hydro-meteorological, geomorphological, and infrastructural variables. Model performance was rigorously evaluated across multiple subregions to quantify spatial heterogeneity in predictive skill and controlling mechanisms. The models achieved moderate to high classification performance, with maximum recall and F1-scores reaching 0.81 and 0.75, respectively, demonstrating strong capability in detecting flood-prone conditions.
Explainability analyses based on feature-importance metrics consistently identified short- and long-duration extreme rainfall (3-hour and 12-hour maxima), 5-day antecedent precipitation, maximum wind speed, groundwater level, and proximity to detention facilities and river networks as the most influential predictors of flood occurrence. Notably, their relative contributions exhibited pronounced spatial variability. In inland and high-elevation basins, flood dynamics were primarily governed by rainfall persistence and subsurface hydrological responses, whereas in coastal and highly urbanized zones, flood occurrence was more strongly modulated by drainage connectivity and proximity to hydraulic infrastructure.
These spatially differentiated controls reflect the complex volcanic hydro-geomorphological setting of Jeju Island and highlight the limitations of uniform flood warning criteria. The findings underscore the necessity of region-specific, dynamically adaptive warning thresholds that explicitly account for local hydrological processes and infrastructure configurations.
Overall, this study demonstrates the methodological advantages of grid-based explainable machine learning for physically interpretable and spatially adaptive flood risk assessment. The proposed framework provides a transferable blueprint for data-driven disaster risk management in volcanic island environments and other hydrogeomorphologically complex regions under intensifying climate extremes.
Acknowledgements
The research for this paper was carried out under the KICT Research Program (Project no. 20260161-001, Development of Digital Urban Flood Control Technology for the Realization of Flood Safety City) funded by the Ministry of Science and ICT.
How to cite: Moon, H., Kim, K.-T., and Kim, G.: Identification of Dominant Flood Influencing Factors with Grid-Based Explainable Machine Learning, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4695, https://doi.org/10.5194/egusphere-egu26-4695, 2026.
Earthen levees represent one of the primary structural measures for flood protection in floodplain areas. However, they can paradoxically increase hydraulic risk by fostering a false sense of security among the exposed population and urban planners (Castellarin et al., 2011). Moreover, these structures are susceptible to failure through various mechanisms triggered by physical processes during flood events, including overtopping and seepage or piping (Palladino et al., 2019).
Monitoring levee conditions using non-invasive geophysical techniques, such as Electrical Resistivity Tomography (ERT), is a highly effective tool for assessing internal structural integrity and detecting potential weaknesses. Such approaches can provide early warning of seepage or piping processes, thereby helping to prevent breach formation and enhance flood risk management (Dezert et al., 2019).
This work describes the results of an experimental monitoring system developed on an earthen levee along the Tatarena stream, central Italy. In particular, the outcomes of an ERT monitoring system is used to collect geophysical parameters (i.e. electrical conductivity and permittivity) correlated to the main hydraulic characteristics of the investigated soil (levee body and foundation), such as porosity, water content, permeability. The observations are correlated with rainfall and groundwater measurements and used as a reference data to address numerical modelling of water infiltration process for seepage vulnerability assessment. The developed methodology, based on coupling experimental ERT monitoring and numerical modelling, can provide valuable insights into water infiltration processes, enabling the assessment of the hydraulic condition of levees, which is essential for identifying potential critical areas.
Castellarin, A., Di Baldassare, G., Brath. (2011). A floodplain management strategy for flood attenuation in the River Po. River Res. Appl. 27 (8), 1037–1047.
Dezert, T., Palma Lopes, S., Fargier, Y., Côte, P. (2019). Combination of geophysical and geotechnical data using belief functions: Assessment with numerical and laboratory data. Journal of Applied Geophysics, Vol. 170. https://doi.org/10.1016/j.jappgeo.2019.103824.
Palladino M.R., Barbetta S., Camici S., Claps P., Moramarco T. (2019). Impact of animal burrows on earthen levee body vulnerability to seepage, J Flood Risk Management.2020;13 (Suppl. 1): e12559, https://doi.org/10.1111/jfr3.12559.
How to cite: Bonaccorsi, B., Rizzo, E., Boldrin, P., Giampaolo, V., De Martino, G., Aronica, G. T., Dionigi, M., Ciabatta, L., Moramarco, T., and Barbetta, S.: Experimental ERT Monitoring for Evaluating Seepage Vulnerability in Earthen Levees: A Case Study of the Tatarena Stream., EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7500, https://doi.org/10.5194/egusphere-egu26-7500, 2026.
Recent increases in intense rainfall have exacerbated urban flooding, driven by impervious surfaces, drainage limitations, and topography. For predicting urban flood susceptibility, models have to consider the spatial configuration of urban hydrological infrastructure, such as urban water detention facilities (UWDF). Conventional physics-based hydrological and hydraulic models are constrained by extensive data requirements and long setup times. In contrast, machine learning (ML) models have been increasingly applied to large-scale flood prediction due to their ability to capture complex relationships among multiple factors. This study aims to assess flood susceptibility in Busan, Republic of Korea using ML models, categorize the characteristics of high-susceptibility areas, and propose optimal locations for UWDF. In this study, the dependent variable for binary classification was constructed by extracting flooded (1) and non-flooded (0) points at a 1:1 ratio, based on flood inventory maps from 2019 to 2023. The explanatory variables consisted of topographic, meteorological, land-use, and drainage infrastructure factors related to flooding (a total of 16 variables). All input datasets were prepared in raster format and resampled to a spatial resolution of 50 m, consistent with the Digital Elevation Model. The constructed dataset was randomly divided into training and testing sets at an 8:2 ratio and applied to Random Forest, Extreme Gradient Boosting, and Support Vector Machine models. Hyperparameter optimization was conducted for each model via Random Search. Then, model performance was evaluated using Accuracy and ROC-AUC metrics. For the best-performing model, Variable Importance and Partial Dependence Plot analyses were performed to interpret the relationships between key explanatory variables and flood susceptibility. Subsequently, the calculated flood susceptibilities were classified into five levels. K-means clustering was applied to high-susceptibility areas to categorize flooding event types based on shared topographic and environmental characteristics. Based on these results, area types with high potential effectiveness for UWDF were identified, and optimal installation sites were derived. The ML–based flood susceptibility analysis would quantitatively reveal and visualize the complex drivers of flooding in high-susceptibility areas.
How to cite: Koo, J., Kim, G., Yim, J., Lee, S., Kim, T., Lee, Y., and Lee, S.: A Machine Learning-based Assessment of Urban Flood Susceptibility and Priority Location of Urban Water Detention Facilities: A Case Study of Busan, South Korea, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-8967, https://doi.org/10.5194/egusphere-egu26-8967, 2026.
Global flood hazard assessments increasingly rely on large-scale modelling frameworks, yet practical use is often constrained by trade-offs between spatial resolution, computational cost, and interactivity. Due to higher computational costs, often global flood products are commonly available only as static maps, which are often limited to predetermined return periods. Such high-resolution global flood maps are not interactive and often restrict event-specific analysis and rapid response mapping. We introduce a new dynamic global flood model, FastFlood Global, which combines higher spatial resolution, interactivity, and computational efficiency to provide flood information for any point in the world with no additional input.
FastFlood Global integrates a computationally efficient flood simulation core with automated global parameterization derived from openly available Earth observation datasets, including topography, land cover, and soil parameters. It supports simulations of fluvial, flash flood, and dam break events and enables the generation of interactive flood maps with custom mitigation options integrated. It also supports an efficient global early warning system for pluvial and flash floods, which is based on the forecasts by the European Centre for Medium-Range Weather Forecasts (ECMWF).
We will present this new model with a series of showcase applications demonstrating the model’s performance across diverse geographic and hydrological contexts, highlighting its potential for global early warning and rapid impact assessment.
How to cite: Kolaparambil, F. J., van den Bout, B., van Roon, K., and Meijvogel, D.: FastFlood Global: Enabling Rapid High-Resolution Flood Modelling at Global Scale , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11931, https://doi.org/10.5194/egusphere-egu26-11931, 2026.
Floods are one of the costliest types of natural disasters. Flood simulation models play a critical role in flood risk prediction and damage prevention by delineating areas at risk of flooding and aiding the design of flood protection infrastructure. 2D hydrodynamic models can be used to simulate floods with high accuracy in complex topography or when detailed hydraulic outputs are required. These models are typically composed of terrain data, a computational mesh, arcs and polygons that form the mesh structure and affect cell size and orientation, boundary conditions, and a set of numerical equations representing flow dynamics. 2D hydrodynamic models can be time-consuming to configure, specific model generation steps can be subjective and based on modeller judgement, and they can be challenging to reproduce. This research focuses on increasing the accessibility of critical flood risk information by creating an automated workflow to generate hydrodynamic models in 2D HEC-RAS.
Models will be generated in the Assiniboine River in Manitoba, Canada. This river is bordered by urban areas and flat prairie topography, which present unique hydraulic modelling challenges. Models will be developed in test sites with varying topography to ensure the generalizability of this work. The 2D hydrodynamic software 2D HEC-RAS will be used, as it is publicly available and widely used across North America.
The workflow includes managing GIS data, generating the computational mesh, and adjusting computational parameters based on desired runtimes, model purpose, desired accuracy, and characteristics of the computational mesh. Automatic mesh generation is already an active area of research; however, the selection of mesh cell size is seldom well-justified, and only semi-automated approaches have been implemented in 2D HEC-RAS. The mesh is designed based on terrain characteristics, flow characteristics, numerical stability, and model accuracy and efficiency. Models will be developed at each test site using the automated workflow to generate an unstructured mesh, a manually generated unstructured mesh, and a manually generated structured orthogonal mesh with a consistent cell size. Each model will be compared in terms of accuracy and computational effort, and qualitatively in terms of mesh configuration. Given that the comparison between the manually and automatically generated unstructured meshes will depend on the modeller, a workflow will be created for the manually generated unstructured grid to increase transparency and reproducibility and to highlight the subjective steps involved in manually developing a model mesh. Both automated and manual workflows will be designed to ensure models and data are findable, accessible, interoperable, and reusable (FAIR principles).
This research will decrease the time required to develop 2D hydrodynamic models for applications such as flood mapping, producing training and validation data for machine learning models, water quality and sediment transport analyses, and stream crossing, control structure and flood protection infrastructure design. With reduced model development time, more time can be spent on model analysis. In addition, this work will increase model reproducibility, enabling more efficient uncertainty and sensitivity analyses, greater transparency in scientific experiments, and the repetition or expansion of experiments in other geographic locations or under different flood conditions.
How to cite: Riddell, P., Asadzadeh, M., Stadnyk, T., and Razavi, S.: Flood modelling in the Assiniboine River Basin with automated mesh generation using 2D HEC-RAS, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12797, https://doi.org/10.5194/egusphere-egu26-12797, 2026.
Natural Flood Management (NFM) interventions and nature-based solutions are increasingly advocated as sustainable flood-mitigation strategies, yet empirical evidence of their hydrological impact at the catchment scale remains limited. This study uses an ensemble approach to characterise runoff-response distributions, drawing on long-term observed data (2011–2023) from the Eddleston catchment in Scotland. Unlike event-focused approaches, this method synthesises system behaviour across diverse hydrometeorological conditions to identify “typical” responses under pre- and post-intervention states.
Results from a small headwater catchment (2.3 km²) reveal statistically significant changes in runoff dynamics, including a delay in peak timing and a reduction in peak height after the installation of a series of leaky barriers. Comparable patterns in a larger catchment (34 km²), within which this smaller headwater catchment is nested, indicate that NFM effects extend beyond headwater sub-catchments. Ensemble-based summaries further highlight the dominant role of antecedent wetness in runoff generation, while also indicating increased infiltration and reduced runoff coefficients under high-flow conditions post-NFM installation.
By integrating ensemble hydrograph separation and impulse-response analysis, this framework provides a transferable tool for assessing NFM effectiveness across multiple scales. Findings strengthen the evidence base for NFM design optimisation and policy integration, supporting long-term strategies for flood resilience.
How to cite: Knapp, J. L. A., Jones, A., Reaney, S. M., Pattison, I., and Black, A.: Catchment-Scale Changes in Runoff Dynamics Following Natural Flood Management Interventions, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-19814, https://doi.org/10.5194/egusphere-egu26-19814, 2026.
Coastal regions are projected to experience a continuous increase in storm surge heights due to climate change–induced mean sea level rise and the intensification of typhoons. These changes substantially exacerbate the risk of coastal inundation in low-lying areas, necessitating a reassessment of existing design standards and disaster mitigation frameworks. To proactively respond to the evolving coastal inundation environment, it is essential to move beyond deterministic design approaches based solely on historical maxima and instead adopt probabilistic analyses of Extreme Sea Levels (ESLs). This study was conducted as part of a project in Korea aimed at developing storm surge–induced coastal inundation prediction maps. A total of approximately 4,100 ESL datasets for each return period, derived through computation and analysis for return periods of 50, 100, 150, and 200 years over a 13-year period (2011–2024), were used to construct spatial distribution maps of ESL heights along the entire Korean coast. To minimize inconsistencies arising from temporal and regional differences in reference sea levels, all ESLs were standardized to a common datum based on the Approximate Highest High Water (AHHW) referenced to the mean sea level at Incheon. For coastal areas where inundation prediction maps were not available, ESLs were estimated using frequency analysis based on the Gumbel distribution. To evaluate the reliability of the constructed ESL distribution maps, the estimated ESLs were compared with ESLs derived from observed tide gauge records, as well as results from extreme value analyses based on the Empirical Simulation Technique (EST) and the Annual Maximum Series (AMS) approach. The comparisons showed similar magnitudes and spatial distribution patterns across regions and return periods, indicating overall consistency in the estimated ESL characteristics. The nationwide coastal ESL distribution maps developed in this study are expected to serve as fundamental baseline data for coastal municipalities in the era of climate crisis, supporting the establishment of comprehensive natural disaster mitigation plans, coastal inundation risk assessments, designation of coastal inundation hazard zones, and the design of coastal and harbor structures.
How to cite: Lee, H.-Y., Cho, W.-H., Park, J.-J., Gu, B.-H., Jeong, K.-Y., Kim, H., and Seo, G.-H.: Development of Return-Period-Based Extreme Sea Level Maps along the Korean Coast, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13005, https://doi.org/10.5194/egusphere-egu26-13005, 2026.
The Danube River Basin in the heart of Europe covers an area of just over 800,000 km², and fully or partly overlaps the territories of 19 countries and the living places of approximately 79 million inhabitants. Many of these people live and work in areas threatened by river floods. Despite 170 years of international coordination by various Danube river commissions and a long, sad record of disasters there are hardly any joint efforts for dealing with this hazard across borders. Barriers to better coordination arise from at least eleven different languages spoken in the basin, political tensions between some of the countries, and the heterogeneity of economic conditions. More basin-wide research about actual and future river flood hazards could play an important role in raising common problem awareness and joint action.
Our contribution – funded by the EU HORIZON project DIRECTED (grant no. 101073978) – is the further development and application of PIK’s Danube Model. Basically a combination of the eco-hydrological model SWIM (similar to SWAT) and the hydrodynamical model CaMa-Flood, it is capable of calculating river water stages and flood heights for ten-thousands of subcatchments from daily weather inputs. Using bias-adjusted ISIMIP CMIP6 climate scenario data we produced respective flood projections under SSP 370 and 585 scenarios.
We are going to present maps of the river system with each river reach coloured after the average return intervals expected for floods at the end of the century which are currently assigned 100-year return periods. Both scenario maps for SSP 370 and 585, respectively, show shifts towards longer and shorter return periods with many extremes: Current 100-year floods are often projected to occur over three times more or less frequently. Trends of shortening return periods however dominate both scenarios – river flood hazards are likely to increase under climate change. The maps also show rather different spatial patterns which indicate the high uncertainty in the projections and in the estimation of extreme value distributions despite having used 600 model years for the analysis.
Besides discussing possible systematic errors of the return interval estimations caused by single extreme climate realizations, the importance of considering levees in river flood models is practically exemplified. Especially in level areas like the Pannonian Basin the simulated flood events show more than 1 m water level differences between model runs with and without levees.
How to cite: Conradt, T.: Expected changes of river flood hazards across the Danube Basin, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7164, https://doi.org/10.5194/egusphere-egu26-7164, 2026.
Flash floods pose significant hazards for settlements and infrastructure in small rural catchments, especially when steep terrain, erodible sediments, and intensive land use come together. Wagenhausen, a village in Lower Franconia (Germany), is located within a 3 km² catchment, dominated by mixed forest and agricultural land. It lies along a small creek, a tributary of the Main River. In response to repeated flooding during intense rainfall events in the past, a series of cascading retention basins was constructed upstream of the settlement in the summer of 2024. In addition, a land-use change is being implemented by reforesting an agricultural field, a potential sediment source, upstream.
This study presents a work-in-progress, comparative assessment of the effects of these hydrological and sediment-related mitigation measures. Event-based simulations are conducted using the SIMWE hydrological model (r.sim.water). We analyse changes in runoff generation and flood intensity under different scenarios, including retention basins, reforestation, and their combined implementation. The modelling approach focuses on relative differences between scenarios rather than absolute discharge values, on account of the data-scarce nature of the catchment and the absence of gauge measurements.
Field investigations include soil analyses, infiltration measurements, and UAV-based surveys, conducted before and after the implementation of the retention basins. This provides information on soil properties, topography, and potential sediment source areas.
The study aims to improve process understanding of flash flood generation and sediment mobilisation in small catchments and to evaluate the role of land-use-based and structural nature-based solutions. The work is part of the EFRE-funded MainPro project, which investigates future hazards and ecosystem-based adaptation strategies for geo-ecosystems, settlements, infrastructure, and agriculture.
How to cite: Schmieg, S., Dittmaier, F., Wolf, T., Krech, M., Kraus, D., and Terhorst, B.: A comparative assessment of nature-based flood mitigation measures in a small catchment, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12270, https://doi.org/10.5194/egusphere-egu26-12270, 2026.
Dike breaches along navigation canals can lead to rapidly evolving flood dynamics, posing significant risks to populations and critical infrastructures. This contribution presents a two-level modeling framework for assessing the hydrodynamic consequences of dike failures. It builds upon previous developments in real-time flood mapping and breach modeling by extending them to both detailed scenario analysis and long-term risk assessment under changing hydroclimatic conditions.
The first component of the methodology relies on a computationally efficient chain of simplified models combining: (i) a 1D shallow-water hydraulic model of the waterways network, (ii) a lumped, semi-empirical breach-growth model accounting for the multi-scale processes governing dike failure, and (iii) a simplified floodplain representation based on pre-computed inundation maps. This framework is applied both for short-term forecasting and for long-term assessments under future climate conditions. The latter uses ensembles of climate projections (seven climate models under three emission scenarios) to evaluate the evolution of breach likelihood and flood hazard up to 2100.
The second component of the methodology consists of detailed “what-if” simulations based on a GPU-accelerated 2D hydrodynamic model (WolfGPU) solving the full shallow-water equations and coupled with the same breach-growth model as in the simplified approach. This technique provides high-resolution predictions of inundation depth, arrival time, and flow velocity fields for selected breach scenarios, enabling a refined characterization of local impacts.
The overall framework is demonstrated through applications to dike-breach scenarios along the Albert Canal in Belgium. Results from both approaches are compared in terms of predicted flood extent, maximum water depth, and warning lead times. The complementarity between fast simplified modeling for real-time support and high-resolution 2D simulations for scenario exploration is highlighted. The study further demonstrates the operational relevance of the framework for waterway managers, particularly in evaluating preventive measures such as anticipatory drawdown of the waterways.
Overall, this work delivers an integrated multi-scale modeling strategy for dike-breach hazard assessment under both present and future hydrological conditions, combining efficiency, physical consistency, and operational applicability.
This research is co-funded by the European Union’s Horizon Europe Innovation Actions under grant agreement No. 101069941 (PLOTO project: https://ploto-project.eu/)
How to cite: Hardy, J., Archambeau, P., Mastricci, D., Schmitz, V., Champailler, S., Melitsiotis, A., Erpicum, S., Pirotton, M., and Dewals, B.: Two-level modeling of dike breach scenarios: from GPU-accelerated 2D hydrodynamics to a simplified real-time modeling chain, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9741, https://doi.org/10.5194/egusphere-egu26-9741, 2026.
Estimates of hydrological hazards such as floods and landslides are highly sensitive to the spatial representativeness of precipitation inputs in data-scarce mountainous basins, where precipitation heterogeneity is strong and observational coverage is limited. This study assesses how precipitation dataset choice and station-network configuration affect SWAT streamflow simulations in the Barandouz catchment, a mountainous sub-basin of Lake Urmia in northwest Iran. The 1158 km² basin is characterized by elevations ranging from 1298 to 3483 m a.s.l. and predominantly pasture land cover.
Daily gauge precipitation from four stations was combined with ERA5-Land reanalysis data to construct three forcing scenarios: (S1) observed in-catchment gauges; (S2) gauges augmented with two upstream virtual stations forced by ERA5-Land; and (S3) the same configuration with bias-adjusted ERA5-Land precipitation. Comparison of daily gauge and ERA5-Land precipitation shows moderate agreement (R² ≈ 0.33–0.44), improving after bias adjustment (R² ≈ 0.43–0.47).
Daily streamflow simulations were evaluated at three main-stem gauges. Model calibration and uncertainty analysis are being conducted using SWAT-CUP (SUFI-2) with split-sample calibration (2006–2013) and validation (2014–2022). Preliminary simulations indicate systematic underestimation of observed discharge across all forcing scenarios, pointing to remaining inconsistencies in precipitation forcing and/or runoff generation. Ongoing calibration will quantify the extent to which these biases can be reduced and identify the precipitation forcing configuration that yields the most robust daily streamflow estimates, with direct implications for hydrometeorological hazard assessment in the Lake Urmia basin.
How to cite: Javanmard Ghoshouni, S., Montaseri, M., Hessari, B., Naderi, S., Prieto, C., and Fenicia, F.: Sensitivity of hydrological hazard estimates to precipitation representativeness in a data-scarce mountainous basin in Iran, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18380, https://doi.org/10.5194/egusphere-egu26-18380, 2026.
Climate change has led to increasingly complex alterations in flood characteristics across South Korea, with changes in flood frequency, magnitude, and duration occurring in different and sometimes opposing directions. This evolution has resulted in the expansion of multidimensional flood risk, which cannot be adequately captured by conventional flood assessments focusing solely on peak discharge. In particular, the increasing occurrence of extreme rainfall events and localized torrential storms highlights the need for a new assessment framework that integrates multiple flood characteristics to better anticipate future flood risks.
In this study, an Integrated Flood Risk Index (IFRI) was developed using IPCC AR6-based future climate scenarios and nationwide runoff simulations to comprehensively assess future flood risk across South Korea. Daily runoff was simulated from 1981 to 2100 using the Soil and Water Assessment Tool (SWAT), with 774 sub-basins across the five major river basins adopted as spatial analysis units. Among 20 climate change scenarios provided by the Korea Meteorological Administration, seven representative RCM–SSP combinations were selected based on a climate variability and extremity screening method proposed by Kim et al. (2025). Analyses were conducted for a historical reference period (1991–2020) and two future periods: mid-century (2031–2060) and late-century (2061–2090).
The IFRI was constructed by integrating information on both flood frequency and intensity derived from runoff simulations. As a core component, the Standardized Flood Index (SFI) was calculated by standardizing short-term accumulated runoff using a log-normal distribution, with flood events defined when SFI values exceeded +1.0 or corresponding high-percentile thresholds. Based on the IFRI, future flood regime changes were quantitatively classified into four distinct types (Type 1–4), representing different patterns of flood risk evolution.
The results reveal pronounced spatial variability in future flood risk across South Korea, with a marked intensification of flood hazards in many regions during the mid-century period, followed by partial moderation in the late century while maintaining elevated risk levels. The shortening of flood return periods indicates an increased likelihood of more frequent and severe flood events. These findings provide a robust scientific basis for national-scale flood risk assessment and emphasize the need to strengthen climate-adaptive flood management and planning strategies.
How to cite: Kim, S., Kim, J., and Kang, H.: Assessing Multidimensional Future Flood Risk across South Korea Using an Integrated Flood Risk Index Based on IPCC AR6 Scenarios, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-8910, https://doi.org/10.5194/egusphere-egu26-8910, 2026.
Reservoirs are critical infrastructures for regulating natural flow regimes and reducing flood discharges, yet their effectiveness during extreme events strongly depends on operational strategies, particularly the initial storage level at the onset of a flood. Here we investigate the non-linear relationship between initial reservoir conditions and flood-attenuation efficiency for about 250 large dams across Italy, adopting a comprehensive, data-driven modelling framework. Flood hydrographs are generated using a simplified hydrological model and subsequently routed through each reservoir, using a “no gates management” approach and full hydraulic routing. We investigate different scenarios of input hydrograph and initial reservoir storage, derived from historical time series of stored volumes from approximately 70 reservoirs across the country and informed by regional flood seasonality.
The results indicate that the reduction in peak discharge is neither spatially homogeneous nor uniform with increasing flood return periods when initial storage levels are reduced. This relationship is strongly non-linear; for instance, as reservoirs reach their capacity limits, doubling the incoming flood peak leads to abrupt reductions in attenuation efficiency. Based on the initially available storage capacity for flood control and the actual reservoir geometries, dams were classified according to the flood severity level needed to cause a significant reduction in their attenuation capacity. This classification allows us to distinguish between dams that experience a gradual decline in performance with increasing flood return periods and those that undergo a threshold effect, which is often not accounted for in conventional regional dam-safety assessments. Notably, the commonly used assumption of a fully filled reservoir at the onset of a flood proves to be overly conservative: under this scenario, about 20% of dams reach their maximum allowable water level for events with return periods of 100 years or less.
By providing a national-scale assessment, the findings of this study can offer helpful insights for dam managers on the effectiveness of maintaining unfilled storage capacity for flood mitigation: by quantifying the impact of initial reservoir storage on flood attenuation, this research provides a data-driven basis for optimizing reservoir operations, particularly in balancing water storage needs with flood risk management. For instance, notable differences can be recognized between reservoirs in the Alpine region, primarily used for hydropower generation, and those in southern Italy, which serve mainly for irrigation and drinking water supply.
How to cite: Evangelista, G., Bertola, M., Blöschl, G., and Claps, P.: Nonlinear flood peak mitigation driven by initial reservoir conditions, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14531, https://doi.org/10.5194/egusphere-egu26-14531, 2026.
Flooding is among the most recurrent natural hazards globally, with its impacts intensifying in urban areas due to rapid urban expansion, land use transformation, and the increasing occurrence of high intensity rainfall events. Flood modeling in ungauged urban catchments remains particularly challenging because of limited hydrological observations and the dominant role of impervious surfaces on runoff generation. This study presents a coupled hydrological- hydraulic and data-driven modeling framework to simulate flood inundation in an ungauged urban region of Hyderabad, India with a specific focus on Zone 5 of the Greater Hyderabad Municipal Corporation. Two sets of modelling scenarios were employed. In the first scenario, flood inundation mapping was simulated coupling HEC-HMS and HEC-RAS. For the hydrograph generation, SCS Curve number method was used with rainfall of finer temporal resolution and ward wise land use land cover data. These hydrographs were used as boundary conditions in the HEC-RAS model. In the second scenario, an Artificial Neural Network (ANN) model was developed using rainfall intensity and other meteorological variables along with lagged simulated discharges from the HEC-HMS model. The ANN-predicted discharges were then coupled with HEC-RAS to generate inundation depths. For validation with ground truth data, both the scenarios were validated using geotagged crowd sourced flood images. The second modelling scenario integrating data driven and hydraulic-hydrologic modelling, performed better than the conventional HEC-HMS HEC-RAS approach, as it showed closer agreement with inundated flood depth. Overall, the findings demonstrate that coupling data driven techniques with hydrologic and hydraulic models significantly enhances urban flood simulation capabilities in data scarce environments.
How to cite: Bora, S., Saipriya, S., and Regonda, S. K.: A coupled hydrological- hydraulic and data-driven modeling framework for flood modelling in ungauged urban catchments, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-16513, https://doi.org/10.5194/egusphere-egu26-16513, 2026.
