INNOVATIVE APPROACH FOR CHARACTERIZING CARBONATE MOUNTAIN AQUIFERS UNDER CHANGING CLIMATE AND LAND USE CONDITIONS

The progressive decline of groundwater resources in the carbonate aquifers of the central Apennines has become increasingly evident over recent decades, particularly in the Sibillini Mountains, where several springs that traditionally sustained local water supplies now show persistent reductions in discharge. Identifying the drivers of this trend is essential for both scientific understanding and water management, given the strategic value of these springs. Although climate change is often cited as the primary cause, the interplay among climatic variability, land use transformations, snow dynamics, and the structural complexity of Apennine aquifers suggests that a broader, integrated analysis is required. This thesis addresses this need by disentangling the individual and combined effects of these factors, with particular attention to the profound land use changes that have reshaped the Apennine landscape since the mid twentieth century. The study area covers 1,225 km2 in the northern Apennines, largely corresponding to the Sibillini Mountains. This region is dominated by a thick and tectonically complex carbonate succession forming fractured aquifers of high permeability. Numerous high discharge perennial springs originate here, supplying water of exceptional quality. Recharge processes, however, are influenced not only by geology but also by steep topography, east–west climatic gradients, and an Apennine snow regime that strongly modulates seasonal infiltration. Against this complex backdrop, extensive forest expansion—at the expense of grasslands and pastures abandoned during the second half of the 20th century—introduces additional hydrological implications through its effects on evapotranspiration, canopy interception, and soil moisture dynamics. To evaluate these interactions, historical climatic, hydrological, and land use datasets from the mid 1900s to the present were reconstructed from heterogeneous institutional sources. While precipitation series revealed gaps and inconsistencies, they nevertheless showed no significant long term change in cumulative rainfall, in line with regional assessments. Temperature records, by contrast, display a clear and widespread increase since the 1950s, with marked implications for evapotranspiration and the snow to rain ratio, and thus for groundwater recharge. Land use evolution was reconstructed using 1954 and 1978 aerial photographs and CORINE datasets for 2000 and 2018, harmonized into forests, shrublands, and grasslands. The resulting classification shows that forest cover has expanded by roughly 200% since 1950s, at a rate significantly higher than the Italian average. This reforestation, driven by the abandonment of traditional mountain livelihoods, represents a major shift in the hydrological functioning of the landscape. Spring discharge analyses reveal widespread declines before the 2016–2017 earthquakes, which produced abrupt and sometimes irreversible hydrogeological changes. Because discharge regimes before and after the seismic sequence are not comparable—as confirmed by split sample modelling—quantitative analyses were restricted to the pre earthquake period. Given data limitations and the spatial heterogeneity of the area, two representative spring catchments were selected for quantitative assessment: Pescara d’Arquata and Tennacola. These basins differ in elevation, size, forest expansion trajectories, and record length, providing a robust basis for comparison. Two methodological approaches were employed: a water budget calculation and a conceptual rainfall–runoff model, which includes a snow module essential for capturing recharge dynamics in mountainous settings. In the water budget approach, effective infiltration was estimated as the difference between precipitation and losses due to evapotranspiration, runoff, and other components. To isolate the effect of land use change, two scenarios were compared: one assuming constant 1954 land cover and one incorporating observed changes up to 2018. Crop coefficients (Kc) for forest, shrubland, and grassland were expressed as ranges derived from multiple literature sources and NDVI based estimates to explicitly account for uncertainty. The rainfall-runoff model was calibrated using the actual land use scenario and then run without recalibration under the constant land use scenario to quantify the effect of vegetation changes on simulated discharge. Results for the Pescara d’Arquata basin are particularly significant. Forest cover increased from 13% to 38% over the study period, and the water budget approach indicates that recharge decreased by about 17% during 1995–2015 using the lower Kc bound and by about 14% using the upper bound. For the entire 1960–2015 period, reductions remain substantial at approximately 10–12%. The rainfall-runoff model, though more conservative, confirms this trend, showing discharge decreases of about 7% (lower Kc) and 5.5% (upper Kc) attributable solely to land use change. According to the simulation snowfall accounts for roughly 38% of total annual precipitation in this basin, and snowmelt delays the recharge peak by more than one month, highlighting the crucial role of snow processes. The agreement between the two approaches, despite structural differences, indicates that the hydrological effect of forest expansion is robust. In the Tennacola basin, forest expansion was earlier and less pronounced. Most land use change occurred before 2000, so the 1998–2015 period analysed corresponds to relative stability in vegetation cover. Here, the water budget method suggests a recharge reduction of 3–4%, depending on the Kc range, while rainfall-runoff model yields comparable decreases of 2–3%. Although lower than in Pescara d’Arquata, these values confirm that the magnitude of hydrological impact is proportional to the extent of forest expansion. Snowfall contributes only 3% of annual precipitation in this basin, underscoring the marked influence of elevation and morphoclimatic context on recharge processes. Overall, the two case studies demonstrate that forest expansion has exerted a strong and previously underestimated influence on groundwater recharge in the Sibillini Mountains. While rising temperatures contribute to increased evapotranspiration and reduced effective infiltration, land use change emerges as an equally important driver, particularly in basins experiencing substantial reforestation. The methodological convergence between two distinct modelling approaches reinforces the robustness of the results. Furthermore, the study highlights the necessity of explicitly representing snow dynamics when investigating recharge in high altitude carbonate aquifers. In conclusion, the decline in groundwater resources in the Sibillini Mountains results from the combined effects of climatic warming, extensive land use transformations, and the inherent sensitivity of fractured carbonate aquifers to shifts in recharge timing and magnitude. The integrated framework developed in this thesis—combining historical reconstruction, dual modelling approaches, and explicit treatment of uncertainty—offers a transferable methodology for similar mountainous regions facing environmental change. By showing that land use change alone can reduce recharge by up to 17% in some basins, this work underscores the need to consider vegetation dynamics in long term groundwater assessments and water resource planning under future climate scenarios.