An Ensemble-Based Coastal Flooding Index For Early Warning Applications

Abstract. Coastal flooding poses a major threat to low-lying coastal areas, particularly under increasing storm intensity and sea-level rise. Early warning systems traditionally rely on deterministic forecasts, which provide a single possible scenario and fail to represent forecast uncertainty. In this study, we evaluate a probabilistic early warning framework for coastal flooding based on short-range ensemble forecasts and assess the feasibility of ensemble reduction to support operational applications. A Coastal Flooding Index (CFI) is introduced, linking the total sea levels at the beach, computed by an atmosphere-ocean-wave modelling chain including nearshore wave processes, to the geometry of local coastal defences. The framework is applied as a pilot case to a low-lying, urbanized coastal stretch in the northern Adriatic Sea (Italy) and tested during two severe storm events (Vaia, 2018, and Detlef, 2019). Ensemble forecasts derived from a full atmospheric ensemble (50 members) are compared with a reduced ensemble (15 members) obtained through a clustering-based selection of representative members. Results show that the reduced ensemble consistently preserves the key probabilistic properties of the full ensemble, including the spatial patterns, timing, and magnitude of ensemble mean and spread of wind speed, significant wave height, nearshore sea level, and derived flooding indicators, while considerably reducing the computational cost (30 % of numerical simulations required). Although limited to two events and a single site, this study demonstrates the potential of combining CFI with ensemble reduction to retain the benefits of probabilistic forecasting for coastal flooding early warning within operational constraints.