Unravelling the interplay between event frequency and intensity in driving rainfall erosivity future changes: Insights from convection-permitting climate simulations in a mediterranean island

Rainfall erosivity is controlled by both rainfall intensity and erosive event frequency, making it highly sensitive to climate change. Convection-permitting climate models (CPMs) demonstrated good performance in representing intense sub-daily rainfall, potentially providing reliable projections of future extreme events, but they are still limitedly used in the context of soil erosion studies. This study evaluates the performance of a convection-permitting climate model at hourly resolution in simulating historical erosivity in Sicily and projects future changes under RCP4.5 and RCP8.5 scenarios. Hourly rainfall data are firstly corrected using intensity thresholds and scaling factors derived from high-resolution observations. CPM consistently underestimate maximum rainfall intensity, erosive event frequency, and mean annual erosivity, particularly in lowland and coastal areas. These biases reflect model limitations in representing short-duration, high-intensity events and sea–atmosphere feedbacks, as well as differences between point-based and gridded data. Bias-corrected future projections reveal contrasting outcomes: under RCP8.5, a 17% decrease in event frequency outweighs intensification, leading to reduced erosivity, whereas under RCP4.5, smaller frequency reductions result in a net increase. Overall these results emphasize that bias correction procedures should consider topographic dependence and different variables, and that future erosivity cannot be assessed from intensity changes alone, but event frequency plays a decisive role. High-resolution models and frequency–intensity interactions must therefore be considered in erosion risk assessments and adaptation strategies.