A multiscale modelling framework of coastal flooding events for global to local flood hazard assessments

Tropical and extratropical cyclones, which can cause coastal flooding, are among the most devastating natural hazards. Understanding coastal flood risk better can help to reduce their potential impacts. Global flood models play a key role in this process. In recent years, global models and methods for flood hazard simulation have improved, but they are still limited in the actionable information that they can provide at local scales. One notable limitation is the insufficient resolution of global models, which cannot accurately capture the complexities of storms and the topography of specific regions. Additionally, most large-scale hazard assessments tend to focus solely on either offshore water level simulations or overland flooding, often relying on static flood modelling approaches. In this study, we introduce the MOSAIC (MOdelling Sea level And Inundation for Cyclones) framework, a flexible Python-based framework designed to dynamically simulate both offshore water levels and coastal flooding. MOSAIC provides a multiscale modelling approach to automatically generate and nest high-resolution local models within a coarser global model. This approach seeks to simulate more accurate water levels, thereby enhancing coastal boundary conditions for dynamic flood modelling. We showcase the potential of MOSAIC using three historical storm events, with the aim of assessing the effects of temporal- and spatial-resolution refinements and bathymetry data. Our findings indicate that the importance of model refinements is linked to the topography of the study area and the storm characteristics. For instance, refining the temporal output resolution has a significant impact on small and rapidly intensifying tropical cyclones but is less critical for extratropical cyclones. Additionally, the refinement of spatial output locations is particularly relevant in regions where water levels exhibit high spatial heterogeneity along the coast. In regions with complex topography, grid refinement and higher-resolution bathymetry play a more significant role. MOSAIC provides an automated approach to provide flood maps at a local scale. Our results confirm the proof of concept that the automated approach of MOSAIC can be used to provide high-resolution flood maps without the need for calibration or other manual steps. As such, MOSAIC provides a bridge between fully global and fully local modelling approaches. In future work, further validation could be carried out to explore the optimal settings for different regions in more detail.