P-FLOOD: a probabilistic framework for flood hazard mapping in levee-protected floodplains

The paper presents a probabilistic framework for mapping hazard induced by river floods in levee-protected floodplains. Large-scale hazard maps of water depth (or any other hazard variable) are derived from the corresponding cumulative probability distributions, estimated throughout the domain from a set of flood simulations that combine different inflow hydrographs and multiple levee-failure scenarios. For each simulation, levee breach probabilities are estimated using fragility functions conditioned on the local hydraulic loading. By accounting for the joint probability of hydrological and breaching events, the method enables a probabilistic characterization of hazard metrics and produces hazard maps associated with any exceedance probability, without being constrained by the discrete return periods of the hydrological inputs. Flood simulations require integrated two-dimensional hydrodynamic models to include flood routing, breach formation, and ensuing floodplain inundation. To limit the computational load, a scalable parent–child strategy is proposed to reduce the number of simulations while retaining acceptable accuracy in the estimated probability field. Moreover, the available simulations can be reused to recompute hazard maps at negligible cost to consider changes in the probability of hydrological and/or breaching events (due to climate change, levee reinforcements, etc.). The framework is applied to a pilot case study in the Adige River (Northern Italy), considering fragility functions for overflow-induced failures and a flood-prone area of ∼100 km2. The results suggest that, if supplied with realistic estimations of levee fragility, P-FLOOD can provide improved hazard predictions compared with traditional approaches, especially for frequent flood events, thereby supporting more informed flood risk management and planning.