Impact of high resolution convection permitting climate projections on the description of the hydrological cycle over Italian rivers and the added value of improved land surface and hydrology in the two-way regional model coupled system.

This thesis investigates the evolution of the hydrological cycle under climate change through high-resolution hydro-climate modelling. By improving model resolution and coupling strategies, it aims to enhance the representation of river discharge in time and space. Understanding present and future hydrological behaviour is essential for anticipating climate impacts and guiding water-resource adaptation strategies. Two complementary approaches were followed. In the first one, the hydrological model CHyM was driven by an ensemble of kilometre-scale convection-permitting (CP) simulations (∼3 km) that has recently become available, enabling explicit representation of convective processes in complex topography. Two CHyM configurations were tested, using as forcing either temperature and precipitation or runoff, and validated against observed river discharge. High-resolution simulations reproduced discharge more accurately compared to lower-resolution ones, driven by Euro-CORDEX and CMIP5 ensembles. Projections for the mid and end of the century under the RCP8.5 scenario revealed an intensification of extremes, with flood proxies ($Q_{100}$) increasing by up to 60 % and drought proxies ($Q_{7,10}$) decreasing by up to 50 %. Mean discharge declined over most of Italy, mainly driven by average precipitation change but modulated by extreme spring and summer precipitation. The Alpine region shows a shift in discharge seasonality, with decreased snow accumulation and earlier snow melt under climate change, with the 25th percentile of total annual discharge being reached up to two months earlier by the end of the century. Model uncertainty in the high resolution ensemble of hydrological simulations, decreased up to a factor of five, confirming the importance of high-resolution modelling for reliable discharge projections. The second approach focused on model development, updating the coupling scheme of the regional climate model RegCM5.0 to include the land-surface model CLM5.0. RegCM5.0 uses CLM4.5 as the standard land-surface scheme, and several structural adjustments were made to ensure compatibility between RegCM5.0 and CLM5.0. The update land-surface scheme includes among its advances the physically based river-routing module MOSART, allowing direct computation of discharge within coupled climate simulations. Offline and preliminary coupled experiments demonstrated realistic discharge behaviour comparable to RegCM driven CHyM results, though longer coupled simulations and further evaluation is still needed. These developments establish the foundation for future fully coupled hydro-climate simulations. To illustrate the real-world impact of reliable discharge projections, hydrological simulations were used for a climate change attribution study looking at the record-breaking 2024 floods in southern Brazil. Using local station data, ERA5 reanalysis, three climate model ensembles and CHyM simulations, both statistical and analogue-based attribution protocols confirm the role of climate change in strengthening the intensity and frequency of the event. According to the statistical framework, events of comparable magnitude have become significantly more likely in present day climate by a factor of eight for precipitation and 1.7 for discharge. Model based analyses further indicated consistent intensification of precipitation and river discharge at global warming levels above 1.5 °C. Overall, this work advances hydro-climate modelling by demonstrating the added value of kilometre-scale resolution and improved land-surface representation for simulating river discharge and extreme events. It provides methodological and technical progress toward enhanced hydrological and climate modelling and highlights how improved hydrological projections can inform climate-change attribution studies.