Votre recherche

Abstract Compound flood (CF) represents a complex hazard that often leads to severe impacts. CF propagates across interconnected systems, generating systemic societal and environmental risks, particularly in coastal cities. Despite progress in data science and remote sensing, a comprehensive review of coupled hydrodynamics with the data-driven GeoAI—an integration of geospatial analysis and artificial intelligence (AI)—for systemic CF risk remains scarce. This review summarizes foundational data-driven and numerical approaches in CF modeling. It then synthesizes emergence, utilization modes, and advancements of coupled hydrodynamic-GeoAI frameworks for CF prediction and systemic impact quantification. A systematic review follows the PRISMA protocol, examining 403 articles from the Web of Science and Scopus databases. The concept of the coupled hydrodynamics-GeoAI model synergizes physics-based simulations with data-driven computational learning, enhancing predictive accuracy and spatially detailed flood risk while explicitly embedding geographic features into the framework. The model offers three utilization modes: (i) direct coupling, (ii) surrogate modeling, and (iii) stochastic statistical-hydrodynamic-ML framework. To enhance comprehensive and robust risk assessment, the review proposes four key model advancements: (1) implementing an active learning framework, (2) integration with physics-guided data-driven, (3) dynamically coupling CF drivers with external factors, and (4) incorporating spatiotemporal analysis under changing climate and socioeconomic conditions. We further advocate for integrating the quantification of both tangible and intangible cascading impacts into systemic CF risk assessments. This review synthesizes computational strategies integrating physics-based hydrodynamics with GeoAI, providing a foundation for systemic CF risk evaluation and guiding future advances in computational hydrology and resilient urban flood management. Graphical Abstract This graphical abstract visually encapsulates the core concept of leveraging coupled hydrodynamic with data-driven GeoAI models for systemic compound flood (CF) risk evaluation in coastal urban areas. The abstract integrates the three main flood drivers—coastal, pluvial, and fluvial—interacting to form the CF. The diagram flows from left to right, where the first section depicts the urban coastal zone, followed by a combination of flood drivers that illustrate the complexity and interconnectedness of factors leading to CF. On the right panel, the graphical focus shifts to the “Coupled Hydrodynamic with Data-Driven GeoAI Model” as the proposed approach for predicting the CF event, encompassing probabilistic analysis, flood propagation, risk assessment, real-time forecasting, and emergency response. The approach integrates physics-based simulations with data-driven geographic data analysis, known as GeoAI. On the bottom left panel, the illustration emphasizes the utilization of the approach through three key strategies: (i) direct coupling through ML-derived boundary conditions, (ii) surrogate modeling to emulate hydrodynamic outputs, and (iii) stochastic statistical-hydrodynamic-ML framework. These strategies demonstrate how GeoAI enhances hydrodynamic simulations to produce more accurate, timely, and spatially detailed flood predictions in various ways. The model’s advancements are also highlighted, addressing the critical need for the utilization of an active learning framework, integration with physics-guided data-driven, dynamically coupling CF drivers with external factors, and the importance of considering spatiotemporal analysis under climate and socioeconomic change. The graphic in the bottom right panel displays the CF output results from the model, which can be further utilized to assess the cascading impact on various aspects. Together, the elements of this graphical abstract convey a sophisticated, interdisciplinary approach to CF risk evaluation, focusing on integrating hydrodynamic with data-driven GeoAI models to better manage the complex challenges of systemic CF risks in urban coastal zones.

Urban flooding has emerged as a recurring challenge in Indian cities, with even minimal rainfall events triggering waterlogging and severe disruptions. This study investigates the phenomenon of frequent urban flooding in five tier-2 Indian cities: Puri, Raipur, Ranchi, Vizag and Bhopal through a qualitative lens. Using focus group discussions (FGDs) with a cross-section of stakeholders including residents, municipal officials, urban planners and environmental activists, the study aims to generate in-depth insights into the patterns of flooding, its multifaceted impacts, perceived causes and proposed mitigation strategies. The data, collected through semi-structured FGDs and analysed thematically, reveals that unplanned urban expansion, poor maintenance of drainage infrastructure, encroachment on natural waterways and inadequate storm-water management are recurrent factors contributing to urban flooding. Participants reported substantial health risks, property damage, traffic paralysis and income loss, especially among marginalized communities. This research adds to the limited qualitative evidence based on urban flooding in non-metro Indian contexts. By foregrounding lived experiences and stakeholder perspectives, the study offers valuable insights for localized disaster risk reduction strategies and urban policy reforms. It underscores the urgency of integrated urban planning, community-based resilience-building and proactive governance to mitigate the growing threat of urban flooding in emerging Indian cities.

Abstract This paper presents a systematic review of the roles and nature of public participation in climate change adaptation in coastal South Asia. In this region, coastal communities face significant climate-related risks due to their inherent vulnerability from population density, rapid changes in landuse, and extensive shoreline. They are particularly vulnerable to hazards such as sea-level rise, tropical cyclones, and storm surges. Governments in the region have initiated adaptation strategies, emphasizing the inclusion of local stakeholders. However, the extent and impact of public participation in these strategies vary across case studies. Drawing on 42 published peer review articles, this review synthesizes participatory approaches in coastal adaptation strategies. We found that public participation strengthens community cohesion, improves adaptation outcomes, and is a mandatory tool for identifying local vulnerabilities and capacities. Participation through grassroots organizations may result in wider sustainability for the vulnerable coastal communities. Additionally, the review identifies several barriers, including the prevalence of elite capture and token participation, externally driven participation processes, and unclear task definitions. We also found that to enhance the effectiveness of public participation in climate change adaptation, it is crucial to adopt inclusive, bottom-up approaches, empower marginalized groups, and ensure that participation is contextually grounded and representative. This study’s insights are applicable not only to South Asia but also to other vulnerable coastal regions in the Global South.

Abstract Compound flood (CF) represents a complex hazard that often leads to severe impacts. CF propagates across interconnected systems, generating systemic societal and environmental risks, particularly in coastal cities. Despite progress in data science and remote sensing, a comprehensive review of coupled hydrodynamics with the data-driven GeoAI—an integration of geospatial analysis and artificial intelligence (AI)—for systemic CF risk remains scarce. This review summarizes foundational data-driven and numerical approaches in CF modeling. It then synthesizes emergence, utilization modes, and advancements of coupled hydrodynamic-GeoAI frameworks for CF prediction and systemic impact quantification. A systematic review follows the PRISMA protocol, examining 403 articles from the Web of Science and Scopus databases. The concept of the coupled hydrodynamics-GeoAI model synergizes physics-based simulations with data-driven computational learning, enhancing predictive accuracy and spatially detailed flood risk while explicitly embedding geographic features into the framework. The model offers three utilization modes: (i) direct coupling, (ii) surrogate modeling, and (iii) stochastic statistical-hydrodynamic-ML framework. To enhance comprehensive and robust risk assessment, the review proposes four key model advancements: (1) implementing an active learning framework, (2) integration with physics-guided data-driven, (3) dynamically coupling CF drivers with external factors, and (4) incorporating spatiotemporal analysis under changing climate and socioeconomic conditions. We further advocate for integrating the quantification of both tangible and intangible cascading impacts into systemic CF risk assessments. This review synthesizes computational strategies integrating physics-based hydrodynamics with GeoAI, providing a foundation for systemic CF risk evaluation and guiding future advances in computational hydrology and resilient urban flood management. Graphical Abstract This graphical abstract visually encapsulates the core concept of leveraging coupled hydrodynamic with data-driven GeoAI models for systemic compound flood (CF) risk evaluation in coastal urban areas. The abstract integrates the three main flood drivers—coastal, pluvial, and fluvial—interacting to form the CF. The diagram flows from left to right, where the first section depicts the urban coastal zone, followed by a combination of flood drivers that illustrate the complexity and interconnectedness of factors leading to CF. On the right panel, the graphical focus shifts to the “Coupled Hydrodynamic with Data-Driven GeoAI Model” as the proposed approach for predicting the CF event, encompassing probabilistic analysis, flood propagation, risk assessment, real-time forecasting, and emergency response. The approach integrates physics-based simulations with data-driven geographic data analysis, known as GeoAI. On the bottom left panel, the illustration emphasizes the utilization of the approach through three key strategies: (i) direct coupling through ML-derived boundary conditions, (ii) surrogate modeling to emulate hydrodynamic outputs, and (iii) stochastic statistical-hydrodynamic-ML framework. These strategies demonstrate how GeoAI enhances hydrodynamic simulations to produce more accurate, timely, and spatially detailed flood predictions in various ways. The model’s advancements are also highlighted, addressing the critical need for the utilization of an active learning framework, integration with physics-guided data-driven, dynamically coupling CF drivers with external factors, and the importance of considering spatiotemporal analysis under climate and socioeconomic change. The graphic in the bottom right panel displays the CF output results from the model, which can be further utilized to assess the cascading impact on various aspects. Together, the elements of this graphical abstract convey a sophisticated, interdisciplinary approach to CF risk evaluation, focusing on integrating hydrodynamic with data-driven GeoAI models to better manage the complex challenges of systemic CF risks in urban coastal zones.

Abstract This case study concerns a major flood event occurred in July 1996 in the Saguenay region (Québec, Canada) induced by heavy and persistent rainfall over this river basin. Various configurations of the CRCM6/GEM5 regional climate model (RCM) using 12-km (0.11° x 0.11°) and convection-permitting (CP) 2.5-km (0.0225° x 0.0225°) resolutions are used to evaluate added value from CP simulations on the simulated extreme precipitation characteristics. The effects of spectral nudging (SN) and initial soil moisture conditions (ISMCs) on surface are also tested on the simulated rainfall. The evaluation of all simulations shows a significant improvement in reproducing precipitation extremes with the convection-permitting model (CPM) at 2.5-km, and substantial influences from SN, and ISMCs. The SN in the CP simulation improved the spatial and temporal patterns of precipitation extremes. Additionally, forced ISMC from a long-term simulations at a 12-km resolution significantly enhanced the model’s ability to capture rainfall intensity, using rainfall observed stations as a reference dataset. This research contributes to the understanding of extreme precipitation events and its reliability as simulated by various configurations of our RCM, and the need to apply higher resolution and accurate surface conditions in the CRCM6/GEM5 for future projections, and its use in design infrastructures, and flood risk management strategies.

Abstract. Coastal cities are becoming more vulnerable to flood risks due to climate change, rising sea levels, intense storm surges, population growth and land subsidence. Developing emergency preparedness and response strategies can reduce the impact of coastal flooding and improve a city's resilience. This article presents a flood relief logistics planning approach aimed at providing decision-makers with a feasible framework. The framework integrates geographic information system (GIS) network analysis and resource allocation optimisation models. Considering the equity of resource allocation, a bi-objective allocation model that minimises the total transportation cost and maximum unsatisfied rate is developed. This flood relief logistics planning approach is applied to Shanghai, China, to present feasible distribution strategies. The case study indicates that the current spatial distribution of emergency reserve warehouses (ERWs) and emergency flood shelters (EFSs) in Shanghai may be vulnerable to extreme flood events. Under a 1000-year coastal flood scenario, the existing emergency resources are insufficient to meet the needs of the affected elderly population. In situations of resource scarcity, reducing the maximum unsatisfied rate can help to improve the equity of resource allocation. Furthermore, incorporating private warehouse clubs (WHCs) into government emergency logistics through public–private collaboration could reduce the governmental burden and improve system efficiency and resilience. This study provides a scientific reference for developing flood relief logistics plans in Shanghai, and it presents a transferable framework that is applicable to other coastal cities.

Flood impacts are intensifying due to the increasing frequency and severity of factors such as severe weather events, climate change, and unplanned urbanization. This study focuses on Briar Creek in Charlotte, North Carolina, an area historically affected by flooding. Three machine learning algorithms —bagging (random forest), extreme gradient boosting (XGBoost), and logistic regression—were used to develop a flood susceptibility model that incorporates topographical, hydrological, and meteorological variables. Key predictors included slope, aspect, curvature, flow velocity, flow concentration, discharge, and 8 years of rainfall data. A flood inventory of 750 data points was compiled from historic flood records. The dataset was divided into training (70%) and testing (30%) subsets, and model performance was evaluated using accuracy metrics, confusion matrices, and classification reports. The results indicate that logistic regression outperformed both XGBoost and bagging in terms of predictive accuracy. According to the logistic regression model, the study area was classified into five flood risk zones: 5.55% as very high risk, 8.66% as high risk, 12.04% as moderate risk, 21.56% as low risk, and 52.20% as very low risk. The resulting flood susceptibility map constitutes a valuable tool for emergency preparedness and infrastructure planning in high-risk zones.

The growing frequency of extreme rainstorms has increasingly exposed tunnels to flooding risks, underscoring the urgent need for effective flood prevention and drainage measures. In this context, an evaluation framework for tunnel flood hazards was developed based on three criteria—hazard-inducing factors, hazard-formative environment, and disaster-bearing body—encompassing nine specific indicators. This study employs the Decision Making Trial and Evaluation Laboratory (DEMATEL) method to construct a causal analysis model and assess the interrelationships and influence levels of risk factors associated with tunnel flooding disasters. Rainfall intensity (C1), rainfall duration (C2), ground elevation (C4), road slope (C5), and impervious surface area (C6) exhibit high causal values, acting as external input factors that drive the occurrence of tunnel flooding incidents. Conversely, water depth (C3), tunnel drainage capacity (C7), emergency flood control measures (C8), and infrastructure aging (C9) display high centrality values, serving as internal factors that reflect the tunnel’s flood prevention capability and determine the extent of disaster losses. Simply enhancing tunnel drainage capacity from the perspective of internal factors alone is insufficient; optimizing the tunnel’s flood resilience requires a combined consideration of both internal and external factors.

This study aims to analyse the effect of partially clogged inlets on the behaviour of urban drainage systems at the city scale, particularly regarding intercepted volumes and flood depths. The main challenges were to represent the inlet network in detail at a rather large scale and to avoid the effect of sewer network surcharging on the draining capacity of inlets. This goal has been achieved through a 1D/2D coupled hydraulic model of the whole urban drainage system in La Almunia de Doña Godina (Zaragoza, Spain). The model focuses on the interaction between grated drain inlets and the sewer network under partial clogging conditions. The model is fed with data obtained on field surveys. These surveys identified 948 inlets, classified into 43 types based on geometry and grouped into 7 categories for modelling purposes. Clogging patterns were derived from field observations or estimated using progressive clogging trends. The hydrological model combines a semi-distributed approach for micro-catchments (buildings and courtyards) and a distributed “rain-on-grid” approach for public spaces (streets, squares). The model assesses the impact of inlet clogging on network performance and surface flooding during four rainfall scenarios. Results include inlet interception volumes, flooded surface areas, and flow hydrographs intercepted by single inlets. Specifically, the reduction in intercepted volume ranged from approximately 7% under a mild inlet clogging condition to nearly 50% under severe clogging conditions. Also, the model results show the significant influence of the 2D mesh detail on flood depths. For instance, a mesh with high resolution and break lines representing streets curbs showed a 38% increase in urban areas with flood depths above 1 cm compared to a scenario with a lower-resolution 2D mesh and no curbs. The findings highlight how inlet clogging significantly affects the efficiency of urban drainage systems and increases the surface flood hazard. Further novelties of this work are the extent of the analysis (city scale) and the approach to improve the 2D mesh to assess flood depth.

Floods are among the most devastating natural disasters; predicting their depth and extent remains a global challenge. Machine Learning (ML) models have demonstrated improved accuracy over traditional probabilistic flood mapping approaches. While previous studies have developed ML-based models for specific local regions, this study aims to establish a methodology for estimating flood depth on a global scale using ML algorithms and freely available datasets—a challenging yet critical task. To support model generalization, 45 catchments from diverse geographic regions were selected based on elevation, land use, land cover, and soil type variations. The datasets were meticulously preprocessed, ensuring normality, eliminating outliers, and scaling. These preprocessed data were then split into subgroups: 75% for training and 25% for testing, with six additional unseen catchments from the USA reserved for validation. A sensitivity analysis was performed across several ML models (ANN, CNN, RNN, LSTM, Random Forest, XGBoost), leading to the selection of the Random Forest (RF) algorithm for both flood inundation classification and flood depth regression models. Three regression models were assessed for flood depth prediction. The pixel-based regression model achieved an R2 of 91% for training and 69% for testing. Introducing a pixel clustering regression model improved the testing R2 to 75%, with an overall validation (for unseen catchments) R2 of 64%. The catchment-based clustering regression model yielded the most robust performance, with an R2 of 83% for testing and 82% for validation. The developed ML model demonstrates breakthrough computational efficiency, generating complete flood depth predictions in just 6 min—a 225× speed improvement (90–95% time reduction) over conventional HEC-RAS 6.3 simulations. This rapid processing enables the practical implementation of flood early warning systems. Despite the dramatic speed gains, the solution maintains high predictive accuracy, evidenced by statistically robust 95% confidence intervals and strong spatial agreement with HEC-RAS benchmark maps. These findings highlight the critical role of the spatial variability of dependencies in enhancing model accuracy, representing a meaningful approach forward in scalable modeling frameworks with potential for global generalization of flood depth.

Monitoring urban pluvial floods remains a challenge, particularly in dense city environments where drainage overflows are localized, and sensor-based systems are often impractical. Physical sensors can be costly, prone to theft, and difficult to maintain in areas with high human activity. To address this, we developed an innovative flood detection framework that utilizes publicly accessible CCTV imagery and large language models (LLMs) to classify flooding conditions directly from images using natural language prompts. The system was tested in Bandung, Indonesia, across 340 CCTV locations over a one-year period. Four multimodal LLMs, ChatGPT-4.1, Gemini 2.5 Pro, Mistral Pixtral, and DeepSeek-VL Janus, were evaluated based on classification accuracy and operational cost. ChatGPT-4.1 achieved the highest overall accuracy at 85%, with higher performance during the daytime (89%) and lower accuracy at night (78%). A cost analysis showed that deploying GPT-4.1 every 15 min across all locations would require approximately USD 59,568 per year. However, using compact models like GPT-4 nano could reduce costs by up to seven times, with minimal loss of accuracy. These results highlight the trade-off between performance and affordability, especially in developing regions. This approach offers a scalable, passive flood monitoring solution that can be integrated into early warning systems. Future improvements may include multi-frame image analysis, automated confidence filtering, and multi-level flood classification for enhanced situational awareness.

Urban flooding caused by short and high-intensity rainfall events presents increasing challenges for cities, threatening infrastructure, public safety and economic activity. Accurately representing infiltration processes in hydrodynamic models is critical, as oversimplifying infiltration can lead to significant errors in predicted flood extents and water depths. This study systematically compares two widely used infiltration models—Green-Ampt and Curve Number—implemented within two leading 2D hydraulic models, HEC-RAS and IBER, to assess their influence on urban flood predictions. Simulations were conducted for 26 rainfall events, including both observed and synthetic hyetographs, across two urban neighbourhoods in Pamplona metropolitan area, Spain. Model performance was evaluated using root mean square error, mean absolute error and confusion matrix-derived metrics such as precision, accuracy, specificity, sensitivity and negative predictive value. Results indicate that the choice of infiltration method significantly affects both water depths and inundation extents: while Green-Ampt yields more conservative water depth estimates, Curve Number tends to underestimate flood extents. The comparison between the two hydraulic models has shown that IBER simulates broader flood extents and lower water depth errors compared to HEC-RAS. The findings highlight the importance of selecting appropriate infiltration methods and hydraulic models for reliable urban flood risk assessment, as well as providing guidance for model selection in urban inundation studies.

Abstract. Large-scale atmospheric dynamics modulate the occurrence of extreme precipitation events and provide sources of predictability of these events on timescales ranging from days to decades. In the midlatitudes, regional dynamical drivers are frequently represented as discrete, persistent and recurrent circulation regimes. However, available methods identify circulation regimes which are either predictable but not necessarily informative of the relevant local-scale impact studied, or targeted to a local-scale impact but no longer as predictable. In this paper, we introduce a generative machine learning method based on variational autoencoders for identifying probabilistic circulation regimes targeted to spatial patterns of precipitation. The method, CMM-VAE, combines targeted dimensionality reduction and probabilistic clustering in a coherent statistical model and extends a previous architecture published by the authors to allow for categorical target variables. We investigate the trade-off between regime informativeness of local precipitation extremes and predictability of the regimes at subseasonal lead times. In an application to study drivers of extreme precipitation over Morocco, we find that the targeted CMM-VAE regimes are more informative of the impact variable of interest, compared to two well-established linear approaches, while maintaining the predictability of conventional non-targeted circulation regimes in subseasonal hindcasts, hence resolving the trade-off identified in previous studies. Furthermore, the targeted regimes and their predictability are physically interpretable in terms of known subseasonal teleconnections relevant to the region, the Madden-Julian Oscillation and variability of the stratospheric polar vortex. The proposed method therefore allows to identify predictable, interpretable and locally relevant representations of regional dynamical drivers given a target variable of interest. These results highlight the potential of the method for a variety of applications, ranging from subseasonal forecasting to attribution and statistical downscaling.

Abstract Context Floodplain forests are being transformed by multiple pressures, prompting widespread management and restoration efforts. It is uncertain how disturbances, including hydrologic change, and management actions will interact to influence the ecology of these threatened forests. Objectives This study examined the effects of alternative management and hydrologic regimes on forest succession at an Upper Mississippi River floodplain site with a restoration project in planning. Methods We used the spatially explicit forest landscape model, LANDIS-II, to simulate forest succession for 100 years under four hydrogeomorphic management scenarios, three forest management scenarios, and two scenarios of future hydrologic conditions. We evaluated changes in forest biomass and composition over time and assessed the relative importance of management actions and hydrologic change on succession. Results Forest aboveground biomass decreased in all management-hydrology scenarios, especially in the wetter hydrological scenario. Intensified hydrogeomorphic and forest management scenarios reduced the magnitude and extent of biomass declines; however, they were unable to prevent overall declines in biomass or cause large shifts in tree species composition. Silver maple ( Acer saccharinum ) was projected to decrease in biomass, while increases in biomass were projected for several late-successional species including swamp white oak ( Quercus bicolor ). Among the factors influencing variation in biomass, forest management had the largest influence in the first 50 years of our simulations, but hydrological regime became the most important factor by the end of the century. Conclusions Our simulations indicate that management actions could play an important role in the conservation of floodplain forests, but their effectiveness will likely be limited if recent upward trends in flooding conditions in this system continue in the future. Thus, our results highlight both the potential benefits and limitations of management actions in the face of hydrologic change.