Abstract Climate change is significantly modifying the characteristics of extreme precipitation events, with relevant implications for society and water-related risks. While most studies focus on changes in the magnitude or frequency of this hazard, this work investigates shifts in the seasonality of extreme precipitation across the Sicily Island over a period spanning almost 100 years. We use circular statistics to characterize the seasonality of annual maximum precipitation at canonical durations. By applying two different circular-linear regression tests (i.e., a Generalized Linear Model developed for circular outcomes within the Bayesian framework and a circular-adapted version of the Theil-Sen slope estimator), we also analyze changes in the extreme events timing over time. Our findings reveal a general anticipation in the timing of extreme events, particularly for short durations, reaching up to ~ 20 days over the study period (1928–2018) and aligning with previous studies reporting increasing severity of short-lived extremes in the region. By integrating circular statistics with spatial clustering methods, we also identify seasonal clusters that reflect the spatial coherence of extreme precipitation timing. In this case as well, especially at shorter durations, a particularly distinct cluster of stations emerges in the northeastern part of the island, an area known to be highly affected by intense convective events and prone to flash floods. This study has a twofold significance. First, it challenges the conventional paradigm in the analysis of seasonality of extreme precipitation by introducing a novel, robust methodology that fully respects the circular nature of the data and can be extended to other geographical contexts or to different variables. Second, it highlights how changes in the timing of extremes are more pronounced for short-duration events, an aspect that could have substantial consequences for flash flooding, urban drainage systems, and agricultural planning, particularly in regions already vulnerable to climate change impacts. Graphical abstract Climate change is altering not only the intensity of extreme precipitation but also its seasonal timing. Using almost one century of rain-gauge observations over Sicily, this study investigates long-term changes in the seasonality of annual maximum precipitation at canonical durations (1, 3, 6, 12 and 24 h) through a fully circular statistical framework. Event occurrence dates are treated as angular variables and analyzed using circular descriptors and two complementary circular–linear regression tests to detect temporal shifts in the timing of extremes. To account for spatial coherence, circular information is further integrated within a spatially constrained clustering approach, allowing the identification of regions sharing similar seasonal behavior of extreme precipitation. The results reveal a widespread anticipation of extreme events, with the most pronounced shifts emerging for short-duration rainfall (1 and 3 h), in agreement with recent evidence of increasing short-lived precipitation severity in the region. Spatial clustering highlights coherent seasonal patterns across Sicily, particularly at sub-daily timescales, suggesting that changes in timing are not spatially random but organized at regional scales. Overall, the study demonstrates that changes in the seasonality of extremes represent an additional and relevant dimension of climate change impacts. By introducing a robust and transferable methodology that explicitly respects the circular nature of timing data and integrates spatial structure, this work provides new insights into how and where extreme precipitation is occurring earlier in the year, with important implications for flash-flood hazard, urban drainage design and climate-adaptation planning in Mediterranean environments.
Treppiedi et al. (Mon,) studied this question.