Statistical Analyses of Springs Discharges Fed by Karstified and Fractured Limestone Aquifers (Central Italy)

The investigation of spring discharge response to meteorological variables is a preliminary but fundamental step towards the understanding of aquifer dynamics and improving their predictability over time. We designed a comprehensive statistical analytical workflow to analyse and predict spring discharges and their dependence on changing meteorological and climatic conditions at different temporal scales. This was applied to four springs monitored at a daily scale for at least 30 years (Bagnara, Capo d’Acqua di Nocera Umbra, Alzabove, and Lupa), which have been affected by prolonged drought periods in the last decades. The studied springs are fed by limestone aquifers characterised by different degrees of karstification and fracturing, mainly exploited for domestic water use. They are located along the Umbria-Marche Apennine (Central Italy) chain in areas with low anthropic pressure (i.e., the spring discharge only depends on the recharge). The proposed workflow first aims to characterise the completeness and inconsistencies of discharge and meteo-climatic time series through univariate statistical tests (e.g., Pettitt, Re-Cusum) at the annual scale. As a second step, the correlation of discharge with reference climate change indices is investigated and quantified at annual time scales, using statistical univariate (i.e., autocorrelation functions) and bivariate metrics (i.e., correlograms, cross-correlation functions). In addition, spring discharge dependence on meteorological variables (precipitation and temperature) is analysed at a daily scale, using bivariate metrics (i.e., correlograms and cross-correlation functions). The response and sensitivity of spring discharges to aggregated meteorological variables over sub-yearly periods are also quantified using cross-correlation function analyses. Finally, springs' recession curves are analysed to check the different behaviour during no-recharge periods.