Improving predictability of deep convective storms with icon hectometric-scale ensembles

Extreme precipitation events represent an increasing global challenge, particularly affecting the Mediterranean basin, where their frequency and intensity have been rising in recent decades. Accurately forecasting these events remains a key objective for numerical weather prediction, yet operational models still struggle to reproduce the magnitude, spatial organization, and timing of intense convective systems. These limitations mainly stem from the relatively coarse grid spacing and the continued reliance on convection parameterizations, which hinder the realistic simulation of localized and short-lived storms. This first part of this work explores the potential of hectometric-scale modelling, employing the ICON model with nested simulations characterized by a grid spacing as fine as 500 m approximately at the innermost domain to reproduce the devastating floods that impacted Italy’s Marche region in September 2022. The study combines sensitivity experiments and ensemble simulations to assess the impact of enhanced resolution and improved turbulence representation. The results reveal substantial improvements in the accuracy of the simulated precipitation fields when convection is treated explicitly and the three-dimensional turbulence scheme is adopted. The ensemble analysis shows that without changes in microphysical parameterizations, high-resolution simulations already provide more realistic precipitation distributions and skilful forecasts. Moreover, the results highlight the crucial role of initial and boundary conditions. In the second part of the thesis, the focus shifts to the role of cloud microphysics and the representation of hydrometeor species. Perturbations are introduced in the two-moment microphysics scheme to assess their influence on ensemble spread, convective intensity, and precipitation type. This analysis is complemented by a second case study characterized by intense hail formation. Overall, this study underlines the added value of hectometric-scale ensemble modelling in capturing the complexity of convective storms. Such developments are essential to strengthen early warning capabilities and inform risk management strategies in a changing climate.