Enhancing coastal flood hazard assessments through spatio-temporal event modelling at large scales
Abstract
The definition of spatio-temporal extreme coastal water level events along the coast is critical for flood hazard assessments. Traditional site-by-site analyses of extreme events at large scales (thousands of kilometers) often oversimplify spatial dependencies and neglect the temporal evolution of storms across space, leading to unreliable hazard estimates. In this study, we implement a spatio-temporal statistical model that integrates wave and storm surge contributions to estimate water level footprints and their associated flooding over large spatial scales. To translate these water levels into flood areas, we employ a hybrid calculation approach that combines physics-based hydrodynamic modelling with machine learning. Demonstrated in the Bay of Biscay (1 230 km of shoreline), we reproduced historical events in both time and space and generated large synthetic databases of extreme events. The synthetic database enables the analysis of extreme event progression along the coast, and allows for a more spatially-coherent determination of flooded areas return levels by incorporating spatial dependencies. The results revealed notable differences between the event-based approach (coherent spatiotemporal phenomena) and the standard site-by-site analysis, particularly at larger spatial scales. In the Bay of Biscay, our comparison shows that spatially uniform 100-year return levels are substantially less frequent when spatial consistency is enforced. For atmospheric water levels, comprising storm surge and wave setup but excluding astronomical tide, the spatially uniform 100-year return-level footprint is expected to occur approximately once every 494 years when spatial dependence is considered. Separately, for flooded areas computed under a high-tide scenario, the spatially uniform 100-year flooded-area estimate corresponds to an occurrence of approximately 175 years. These findings highlight the importance of considering the spatio-temporal evolution of extremes to improve large-scale coastal flood hazard assessments and provide more spatially coherent hazard information to support coastal adaptation and resilience
Related
- Uncertainty in broad-scale coastal flood damage cost assessments under sea-level rise: critical needs for more accurate protection data
- ShyBFM v1.0: unstructured grid advection-diffusion-reaction modelling for coastal biogeochemical processes
- Phase-resolved coastal wave modelling using a non-hydrostatic multilayer formulation
- Assessing Spatial and Temporal Variability of Vertical Land Motion Along Coastal Massachusetts Using Sentinel-1 InSAR and GNSS Time Series Analysis
- Enhancing urban water resilience through green infrastructure in Kathmandu, Nepal
BibTeX
@article{alvarezcuesta2026enhancing,
title = {{Enhancing coastal flood hazard assessments through spatio-temporal event modelling at large scales}},
author = {Moisés Alvarez-Cuesta and Jérémy Rohmer and Alexandra Toimil and Íñigo J. Losada and Gonéri Le Cozannet},
journal = {Weather and Climate Extremes},
year = {2026},
month = sep,
doi = {10.1016/j.wace.2026.100952},
url = {https://hal.science/hal-05743530v1},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - JOUR
TI - Enhancing coastal flood hazard assessments through spatio-temporal event modelling at large scales
AU - Moisés Alvarez-Cuesta
AU - Jérémy Rohmer
AU - Alexandra Toimil
AU - Íñigo J. Losada
AU - Gonéri Le Cozannet
PY - 2026
DA - 2026/09/09
JO - Weather and Climate Extremes
DO - 10.1016/j.wace.2026.100952
UR - https://hal.science/hal-05743530v1
AB - The definition of spatio-temporal extreme coastal water level events along the coast is critical for flood hazard assessments. Traditional site-by-site analyses of extreme events at large scales (thousands of kilometers) often oversimplify spatial dependencies and neglect the temporal evolution of storms across space, leading to unreliable hazard estimates. In this study, we implement a spatio-temporal statistical model that integrates wave and storm surge contributions to estimate water level footprints and their associated flooding over large spatial scales. To translate these water levels into flood areas, we employ a hybrid calculation approach that combines physics-based hydrodynamic modelling with machine learning. Demonstrated in the Bay of Biscay (1 230 km of shoreline), we reproduced historical events in both time and space and generated large synthetic databases of extreme events. The synthetic database enables the analysis of extreme event progression along the coast, and allows for a more spatially-coherent determination of flooded areas return levels by incorporating spatial dependencies. The results revealed notable differences between the event-based approach (coherent spatiotemporal phenomena) and the standard site-by-site analysis, particularly at larger spatial scales. In the Bay of Biscay, our comparison shows that spatially uniform 100-year return levels are substantially less frequent when spatial consistency is enforced. For atmospheric water levels, comprising storm surge and wave setup but excluding astronomical tide, the spatially uniform 100-year return-level footprint is expected to occur approximately once every 494 years when spatial dependence is considered. Separately, for flooded areas computed under a high-tide scenario, the spatially uniform 100-year flooded-area estimate corresponds to an occurrence of approximately 175 years. These findings highlight the importance of considering the spatio-temporal evolution of extremes to improve large-scale coastal flood hazard assessments and provide more spatially coherent hazard information to support coastal adaptation and resilience
SN - 2212-0947
ER -