Olga Petrucci
Rainfall-triggered shallow landslides (SLs) are rapid slope failures affecting soil layers typically thinner than 3 m. Despite their limited extent, cumulative impact can be significant, particularly in vulnerable hilly environments. This study presents an empirical approach based on a 104-year historical inventory (1921–2025) covering an area of approximately 539 km2, where 467 records were aggregated into 41 shallow landslide events. Because historical datasets primarily document damage, the most complete records are associated with densely populated municipalities. This exposure-driven bias is explicitly exploited to derive rainfall triggering conditions and assess their transferability to neighboring data-scarce areas. A reference municipality is used to define rainfall thresholds based on 1-, 3-, and 5-day cumulative precipitation. Results highlight the importance of rainfall accumulation over short to intermediate durations (1–5 days), particularly during intense rainfall episodes. The threshold derived from second-lowest values reduces the influence of individual minima and provides a more representative description of the observed triggering conditions. Event-based analysis shows that most rainfall events exhibit high spatial consistency, while lower consistency is limited to small number of cases. Station-based analysis confirms that several rain gauges are highly consistent with the reference station, supporting the transferability of rainfall-triggering conditions associated with the threshold, whereas a few stations show lower agreement due to local topographic and climatic factors. These findings suggest that the proposed approach provides a practical framework for landslide hazard assessment in regions characterized by fragmented historical datasets.