Toward operational coastal flood forecasting: a reduced-ensemble early-warning index for low-lying coastal areas

Coastal flooding represents a major hazard for low-lying coastal environments, arising from the combined effects of atmospheric forcing, sea-level fluctuations, and wave-induced processes, which can jointly produce rapid and severe coastal impacts. In a complex context of climate change, including sea-level rise and potential shifts in storm characteristics, both the frequency and intensity of such events are expected to increase. This context calls for the development of robust operational early warning systems capable of explicitly resolving atmosphere-ocean interactions and their role in driving coastal extremes. In this work, we introduce and evaluate a Coastal Flooding Index (CFI), conceived as an impact-oriented Early Warning Index that links meteo-marine conditions with local coastal characteristics. The index combines total water level at the shoreline – accounting for tides, storm surge, and wave contributions – with the geometry of coastal defences, allowing a classification of flooding conditions from no impact to inland inundation.
The proposed framework is based on a fully coupled modelling chain, including atmospheric, hydrodynamic, wave, and nearshore components. Atmospheric forcing is provided by WRF, while sea level and offshore wave conditions are simulated through SHYFEM and WAVEWATCH III, and nearshore wave processes are resolved with XBeach. This setup enables a consistent representation of air–sea interactions across scales, from synoptic forcing to local coastal impacts.
To support operational applications, a reduced-ensemble approach is adopted. Starting from a larger ensemble, a subset of representative members is selected through clustering techniques applied to key atmospheric variables. This underline to retain the essential spread and extreme scenarios while reducing computational cost, making the system suitable for real-time forecasting. The methodology is tested over a vulnerable sector of the Northern Adriatic coast (Jesolo, Italy) during two severe storm events (Vaia 2018 and Detlef 2019), both characterized by strong surge-wave interactions. Results suggest that the reduced ensemble is able to preserve the probabilistic structure of the full system and provides meaningful estimates of both exceedance probability and timing of coastal flooding. Overall, the proposed approach suggests a practical way to integrated, probabilistic early warning systems, bridging weather forecasting and coastal impact assessment, and supporting emerging applications in climate services and coastal risk management.