An early warning system for urban fluvial floods based on a predefined library of hazard maps and rainfall depth-duration thresholds: the case study of Palermo (Italy)

The growing expansion of urban areas has profoundly modified the processes of generation and transferring of surface runoff; the increase in the density of infrastructures, human activities and population in urban settlements also implies a considerable rise of the elements exposed to flood risk. In the last years, climate change has induced an intensification of the hydro-climatic extreme events, and it has been globally observed an increase in the frequency of water related catastrophic events in urban areas, such as fluvial and pluvial floods.

Among the possible different approaches to mitigate hydraulic risk, non structural mitigation measures probably offer the highest potentialities in the short term. This study proposes a prototypal Early Warning System (EWS) for fluvial flood risk developed over a densely populated subarea of Palermo (Italy) threatened by the presence of the Oreto River. The proposed EWS is capable of determining a relationship between possible event scenarios and some potential precursors. Each event scenario identifies potential first points of flooding, flooding areas and water levels, providing hazard maps specific for people, vehicles and infrastructures. Forecast rainfall, antecedent soil moisture and initial river stage conditions are considered as event precursors. Rainfall characteristics considered are the 24/48 h forecast of the expected cumulative depth and duration, retrieved from the National Surveillance Bulletins based on a Sicily Local Area Model. Water levels monitored in real-time by the Oreto a Ponte Parco hydrometric station are used to derive a proxy measure for the antecedent soil moisture and the initial river stage conditions.

The system is based on numerical rainfall depth-duration thresholds previously defined for the basin under analysis that, for fixed antecedent moisture condition and rainfall hyetotype, provide a family of isocritical discharge curves. From such curves, once the characteristics of the expected rainfall event are known, it is possible to estimate the expected hydrograph peak. Coupling this information with the initial river stage, the EWS is able to retrieve a corresponding event scenario, exploiting a pre-built library of event scenarios previously developed offline through a coupled hydrological-hydraulic model, considering several project hydrographs under different initial river stage conditions. The computation domain was previously accurately defined, starting from a 2m DEM, opportunely corrected to account for the exact geometry of bridges and other infrastructures. The EWS was tested with respect to an historical flood event occurred in 2018, demonstrating satisfactory performances in terms of predicted flooding area and water levels.