Measuring floods by their scars: toward a data-driven flood severity scale

Abstract To classify and compare extreme events, it is useful to rely on metrics that capture their impacts. In the natural hazard field, widely accepted impact-based classification systems already exist, such as the Fujita scale for tornadoes, the TORRO scale for hail, and the Mercalli scale for earthquakes. While magnitude-based measures like the Richter scale have become standard for seismic events, phenomena such as tornadoes and hailstorms are still most effectively characterized through the impact rather than through direct physical measurements. For flood events, assessing and comparing their “magnitude” is often problematic, as peak discharge data are frequently unavailable, incomplete, or difficult to relate to a meaningful river section. Moreover, even with available precipitation data, it is known that similar rainfall amounts can produce very different effects in distinct environments. For these reasons, this paper suggests the introduction of a flood severity (FLOSEV) scale. The FLOSEV scale is devised using the historical information from the AVI (Aree Vulnerate Italiane) project, an extensive collection of flood and landslide events in Italy spanning centuries but with a unique systematic inventory from the early years of 1900. Each AVI record contains detailed information on affected municipalities, number of fatalities, and damage to properties, here used to suggest an approximate log-scale classification of flood events based on five severity levels, ranging from minor disruptions to widespread destruction. The proposed criteria have been applied to recent major events and appear to attribute them to reasonable classes, which supports the robustness and practical relevance of the scale.