An Empirical Rainfall Threshold Approach for the Civil Protection Flood Warning System on the Milan Urban Area

Non-structural risk mitigation tools such as civil protection alerts for citizens proves highly beneficial in mini-mizing the impacts linked to floods. Flood forecasting represents a challenge due to complex and non-linear hydrological processes involved, especially in highly urbanized areas. In this study, a Flood Warning System (FWS) based on the development of catchment-specific empirical Rainfall Thresholds (RTs) is proposed. Seven river catchments in the "Hydraulic node of Milan," northern Italy, were analyzed using a dataset of 25 years (1998-2022) of hourly rainfall and discharge data. An empirical methodology, based only on historical rainfall-runoff data and applicable to any river catchment, is proposed with the aim to validate and improve the existing Rainfall Threshold (RT) defined on the same area by the Lombardy Region civil protection. The RTs obtained using the proposed method showed improvements with respect to the existing civil pro-tection RTs, because it allows to derive time-continuous and catchment-specific RTs. Additionally, accounting for the Antecedent Moisture Conditions (AMC) with the proposed "equivalent rainfall" approach results in more accurate RTs, suggesting its consideration for issuing civil protection alerts. The accuracy and uncertainty of the RTs were analyzed by means of binary classification measures coupled with bootstrap resampling. The proposed procedure for constructing RTs, which is applicable to any river catchment having sufficiently long time series of rainfall and runoff data, and not necessarily on urban areas only, indicates potential for being an additional and simple FWS to mitigate flood risks for civil protection purposes.