Magnesium content and groundwater flow in limestone aquifers: a relationship with potential developments for exploitation management of karst springs

Results and Discussion Mg 2+ content in Pertuso and Capodacqua springs' water is inversely proportional to the total baseflow (BF) discharges Q. For Pertuso Spring, sixteen samples showed a data cluster around a linear trend with a good correlation (R 2 = 0.86). Maximum concentrations are detected in the same month (November/December), during the minimum seasonal spring flow, except for 2017, when a strong drought event occurs, forcing the water agency to increase the exploitation rate of the spring (Figure 2). Even better is the correlation found for Capodacqua di Spigno Spring (R 2 > 0.9), where the reconstruction of seasonal spring discharge depletion for 2018-2019 highlighted a related increasing path of Mg 2+ content along the linear baseline. After reaching the minimum flow, the seasonal recharge occurring during late autumn/early winter brings back Mg 2+ values at the top of the BF line for a new recession period (Figure 3). Small deviations from the BF trendline are comparable with errors related to chemical analyses (±2%) and discharge measurements (±5%), whereas large deviations, initially supposed to be outliers, show a different hydrodynamics-geochemical mechanism of the aquifer, tracing the fast-spring response to storm events. These latter activate the discharge overflow (OF) component and the consequent mixing between long residence times groundwater and runoff water, quickly flowing through karst conduits. In fact, the marked drop in magnesium content at equal discharges is the result of the mixing. Using the mass balance equation to separate BF and OF, it was possible to set up a model which outlines overflow drifts from the baseflow line. Sensitivity analysis starts using typical rainfall Mg 2+ contents as C OF and a couple of values coming from the linear relationship C BF-Q BF. Based on available literature data, rainfall average Mg 2+ content ranges from 0.1-0.3 mg/L in mountain areas (Sapek B., 2014). The only free parameter (variable) is Q OF , related to the rainfall event intensity, giving different results of C MIX (Figure 2). For Pertuso Spring, characterized by a total substitution mechanism within the aquifer, an intense rainfall event occurring during the final phase of aquifer depletion immediately implies a drop in Mg 2+ content, whose value was enriching due to the longer residence time of groundwater in the carbonate rock. This also highlights the role of previous saturation conditions of the hydrogeological basin on the karst spring response to different precipitation events.