MODELLING CURRENT AND FUTURE WATER RESOURCES AVAILABILITY IN AN ALPINE CATCHMENT EXPLOITED FOR HYDROPOWER PRODUCTION

A Precise and dependable assessment of present water resource availability and their future projections are crucial for addressing the challenging crisis resulting from climate change and ongoing population growth. It is strategically important to analyze regions traditionally known for their abundant water resources but highly susceptible to climatic variations and human pressures, such as the Alpine environment. This study focused on the hydrologic modelling of the Valgrosina valley (Italian Alps). It covers an area of about 130 km2 ranging from 1200 to 3400 m a.s.l. and is divided into two main branches: the Sacco valley (western side) and Eita valley (eastern side), where a dam for hydroelectric exploitation is located. The main objective was to accurately simulate the hydrologic cycle through the calibration of the partially distributed hydrologic model (GEOframe), focusing on both hourly and daily scales. Specific objectives include: (i) refining precipitation input for optimal model representation, (ii) evaluating model performance forced by weather forecasts input, and (iii) quantifying the impacts of climate change on water resource availability within the study area. The thesis is composed of seven Chapters. The first three introductive Chapters illustrate the motivation behind the research, a thorough description of the study area, and the modelling choices. In the fourth Chapter, the hourly weather series of 22 rain-gauges and hourly precipitation from a radar dataset (1-km × 1-km resolution, from MeteoSWISS) from 2005 to 2020 were used. Given a general underestimation of radar compared with ground-based values (about 20%), two approaches were tested to adjust the radar input: radar precipitation is corrected in every centroid of the hydrologic model sub-basins (point-based correction); the radar precipitation is adjusted by spatializing the radar-station error (interpolation-based correction). The precipitation corrections and the streamflow simulations were evaluated by comparing them with the observed data by performance indices. Among the corrections tested, the spatial method obtained by summing the radar trend with the ordinary kriging interpolation of residuals proved to be the best with a KGE of 0.56 resulting from hourly comparison between the corrected precipitation series and those of the stations. The integration of this corrected series with the hourly hydrologic model led to excellent results with an average KGE for the calibration and the validation period of 0.79 and 0.77, respectively. The coupling of the forecast data with the hourly hydrologic model is described in Chapter five. The database of hourly weather forecasts derived from the MOLOCH model (ISAC-CNR) over the period 2016-2020 was exploited to test the predictivity of the hourly hydrologic model. The model was forced with weather forecasts (temperature and precipitation) for two months, namely the rainiest month (Oct 2018) and an average precipitation month (May 2018) in the period 2010-2020. Model outputs were analyzed hourly and over three- and six-hour aggregation periods, which are considered useful for taking actions by the hydropower management company. The aggregated six-hour hydrologic model outputs showed good predictions of the general hydrologic response (streamflow trends, hydrograph shape, time of peak), but some quantitative bias. The error calculated on an hourly basis (-88% ÷ +54%) decreases (-37% ÷ +10%) considering the 6h series. Also, simulations forced by weather forecasts reproduced well the timing of the streamflow peak related to the extreme event of October 2018, with a maximum lag of one hour. In Chapter six, Climate Change effects on the hydrologic cycle of the study area were analyzed. First, the hydrologic model was calibrated on a daily timestep (KGE>0.70). Then, temperature and precipitation timeseries from 21 RCMs (EURO-CORDEX domain) were downscaled to investigate possible variations of the hydrologic regime for four different scenarios characterized by global temperature increases of 1.5, 2.0, 3.0, and 4.0°C above the pre-industrial levels. In the study area, the downscaling results revealed an average temperature increase up to 5°C in July for the most severe scenario, respect to the 2005-2020 reference period. No clear trend was observed for precipitation, with differences ranging from +4% to +7% compared to the observed data. The corresponding discharge simulations revealed significant alterations in the hydrologic seasonal response. In detail, reductions of the summer peak around 9-10% were observed, considering the most severe scenarios (+4.0°C) in comparison to the reference period (2005-2020). Finally, Chapter seven summarizes all the analyses performed and provides perspectives on future research. By investigating the complexities of local hydrologic processes, this study makes a significant contribution to enhancing our comprehension of the water systems within the region. The insights gained have the potential to serve as a foundation for informed decision-making in water resource management, facilitating the formulation of adaptive policies tailored to the specific challenges and opportunities inherent in this geographical area. Consequently, the study not only advances scientific knowledge but also holds practical implications for sustainable water resources utilization and resilience-building to face climate changes.