Orographic control on bias and uncertainty in extreme sub-daily precipitation simulations from a convection-permitting ensemble

• Elevation strongly influences extreme precipitation biases across complex terrains. • Errors in extreme precipitation simulation increase with decreasing rainfall duration. • Normalised sampling uncertainty ranges between 12% and 19%. • Inter-model uncertainty dominates at short durations in complex terrains. Climate change adaptation strategies and flood risk management in complex terrain rely on accurate projections of sub-daily precipitation extremes and on the realistic evaluation of the associated uncertainty. However, these aspects are still insufficiently analysed, especially concerning the most updated convection-permitting climate models (CPM). This study provides a comprehensive approach to assess the performance of CPMs in complex terrain by analysing bias and uncertainty of sub-daily extreme precipitation. We evaluate extreme sub-daily precipitation in the north-eastern Italian Alps using an ensemble of CPM simulations. We assess return levels for 2, 5, 10, 20, 50, and 100-year return periods across five durations (1, 3, 6, 12, and 24 h) and associated uncertainty. We compare estimates based on CPM simulations with the ones based on rain gauge observations. Our results show that CPM performance improves with longer durations, with underestimations in the lowlands and overestimations in the highlands, particularly for short durations. The CPM ensemble captures the orographic effect, although with a weaker magnitude with respect to what is observed. We quantify intra-model, inter-model, and total uncertainties, finding that inter-model uncertainty dominates, particularly at higher elevations for short durations. The normalised intra-model uncertainty ranges between 12% and 18% and matches well the observed uncertainty for the different elevation ranges, suggesting that the sampling uncertainties are, on average, roughly proportional to the return level values (and the relevant biases). This work provides critical insights into the performance of CPMs in complex terrain and has important implications for improving the accuracy of extreme precipitation estimations, which are crucial for climate change adaptation and flood risk management in mountainous regions.