Statistical behavior of temporally compounding multi-peak flood events

Abstract

Floods are the most frequent natural hazard globally (2003–2023) and rank second in economic losses, according to EM-DAT, the International Disaster Database. When flood events occur in close succession, they pose significant challenges for emergency management due to limited recovery time between events. This study focuses on multi-peak (MP) floods, where peaks occur within hours to days. Using discharge data from 77 hydrometric stations in Northern Italy’s Po district, we examined statistical differences between MP and single-peak (SP) flood events and analyzed their seasonal patterns and generating mechanisms across diverse river regimes. We demonstrated that SP and MP events exhibit distinct statistical behaviors. The first type of events has more skewed distributions with heavy tails, while the second displays flatter distributions with lighter tails and higher mean values. Seasonal analyses suggest that MP floods in glacial and nival-pluvial regimes are influenced by glacier and snow melting, whereas those in the Padanian regime are driven by tributary routing effects. These triggering mechanisms seem to be responsible of the lighter tails of the distribution of MP floods. By highlighting the distinct statistical behaviors and generating mechanisms of MP and SP floods, we identified recommendations for designing MP flood hydrograph, supporting flood-risk management.