5.1. SEASONAL VARIATION OF PHYSICOCHEMICAL PARAMETERS
As can be observed in Fig. 10, the temperature exhibits a "V" pattern, generally ranging from 23 to 27°C in the summer (represented by campaigns S1, S5, and S8) and between 18 and 16°C in the winter (S3 and S7), with intermediate values (18 to 21°C) in the autumn (S2) and spring (S4 and S6). Additionally, the data show corresponding trends between shallower and deeper wells, as described earlier, with a greater propensity for seasonal variation in the former. Regarding pH, it is noted that it tends to fluctuate in the range corresponding to a neutral to slightly acidic characteristic, regardless of the climatic season. For shallow wells (M1A, M1B, M2A, M2B, M4A, and M4B), the pH varied relatively consistently between 5.7 and 6.5, remaining nearly in that range from S1 to S6, with a slight increase in the last campaign (S8). Alternatively, the pH of wells M3B and M4C tended to stay within the range of 5.2 to 5.6 during most campaigns, except for the last one, when it showed an increase above 6.1.
EC exhibited a more homogeneous behavior throughout all campaigns, with a slightly higher level during the summer of different years. Among wells M3B, M3C, and M4C, this variation remained in the lower range, between 100 and 200 µS/cm, while for wells M1A to M2B, it generally ranged between 200 and 300 µS/cm. For wells M4A and M4B, this level was mostly between 400 and 500 µS/cm. The ORP data showed a variation in an inverted "V" pattern, with predominantly oxidizing conditions - highlighted by values above 0 (zero) mV - between campaigns S2 and S7, and reducing situations (with negative values) in campaigns S1 and S8. Overall, the values remained relatively constant between + 85 and + 180 mV for almost all wells, except for M4A and M4B. In these wells, the values frequently fluctuated between − 50 and + 100 mV during S2 and S7, while they had values below − 70 mV in campaigns S1 and S8.
Lastly, DO presents slightly higher values during the winter months (S3 and S7) and, to a lesser extent, in the spring (S4). During the summer months (S1, S5, and S8) and autumn (S2), the values are slightly lower. However, the values fluctuate in the range between 2 and 3 mg/L of O2. This behavior is more evident in shallow wells (M1A to M2B) but imperceptible in deeper wells (especially M3C and M4C), where the values show a decreasing pattern between S2 and S8, transitioning from values between 6 and 7 mg/L to values below 4 and 2 mg/L, respectively.
5.2. SEASONAL INFLUENCE ON GROUNDWATER GEOCHEMISTRY AND MINERAL SATURATION INDICES
From a seasonal perspective, the concentrations of Ca, Mg, K, Na and NO₃⁻, and trace elements (Fe, Ba, Sr, Mn, Al, B, Ni, and Zn) are proportionally higher in the summer, spring, and autumn periods (S2, S4, S5, S6, and S8) compared to winter (S3 and S7). This pattern can be seen in Fig. 11, where the upper right quadrant - representing the area with a groundwater temperature above 20.4 ºC (average of autumn and spring) and a concentration higher than the median for the parameter - appears more populated by data from these climatic seasons. This behavior suggests that higher temperatures favor the increase of dissolved ions in groundwater, which is linked to a higher rate of chemical weathering (Riedel 2019), but also possibly linked to higher PCO₂, which enhances the release of ions present in the aquifer matrix (Keating et al. 2010). Solute concentration increase due to a higher evaporation rate during summer cannot be discarded as another mechanism influencing the results (Zhu and Schwartz 2011).
Higher NO₃⁻ concentrations during this period are primarily related to the Aa unit wells (M4C and M3C) connected to the deeper fractured crystalline aquifer. These align with values reported by (Roisenberg et al. 2003), indicating an inherited hydrogeochemical condition from the granite. Elevated NO₃⁻ levels can also result from organic fertilizer degradation and nitrogen residues, especially at higher temperatures (Pastén-Zapata et al. 2014). Abnormal concentrations of cations and trace elements (especially Ca, Mg, Na, K, Fe, Mn, Sr, and Ba) in wells M4A and M4B - compared to other wells in St unit - can be explained by the accumulation of dissolved ions in topographically low areas, where there is convergence of surface runoff and groundwater flow (Adams et al. 2001; Delin and Landon 2002; Yan et al. 2017), causing a local hydrochemical anomaly. Conversely, concentrations of the anions SO₄²⁻, Cl⁻, and PO₄³⁻ remain elevated across all seasons, exhibiting weak correlation with temperature. This pattern suggests that these solutes originate independently from the aquifer's mineralogy, indicating external sources. The most plausible origins for these elements are atmospheric deposition, residual NPK fertilizer dissolution, and/or decomposition of organic matter from animal sources (Roisenberg et al. 2003; Domagalski et al. 2012; Porowski et al. 2019; Torres-Martínez et al. 2020).
As the increase in dissolved CO₂ will contribute to the acidity of groundwater, this factor is expected to alter the hydrogeochemical stability of minerals present in the aquifer. Therefore, understanding SI in background water establishes a baseline to assess climate-driven changes. The data show that although seasonal influence affects SI, shifts from undersaturation to supersaturation are rare, except for wells M3B and M4C (goethite, lepidocrocite, and illite). During the CO₂ injection phase (which will be carried out in a future phase), the alteration of saturation conditions and the interactive process with the aquifer mineralogy will be comprehensively examined.
5.3. CARBON ISOTOPES (δ¹³C-DIC) AND CO₂ SOURCES
Due to the complexity of the isotope signature in background groundwaters, the Miller-Tans plot (Miller and Tans 2003, Fig. 12), which plots δ¹³C-DIC x DIC against DIC concentration, were used to differentiate possible CO₂ sources to the DIC of the groundwater. The line depicting the overall linear trend of data (R2 = 0.76) effectively segregates two groups: i) those above the line, mostly comprising wells with a δ¹³C-DIC median around − 6‰ or higher, and ii) those below the line, primarily including wells with a δ¹³C-DIC median below − 6‰. Applying the analytical data of δ¹³C-DIC and referring to the equations by Clark and Fritz (1997) and (Clark 2015) (Equations (1) to (7), Supplementary Information), the isotopic ratio of the DIC source (δ¹³C-CO₂) could be deduced, considering an isotopic equilibrium between them.
As can be noted, δ¹³C-CO₂ values for wells M1A, M1B, M2A, M2B, and M3B generally fall between − 6 and − 8‰. This range corresponds to the isotopic ratio of atmospheric CO₂ (Sharp 2017; Hoefs 2021), implying these shallow wells maintain isotopic equilibrium with the atmosphere. However, M4A, M4B, M4C, and occasionally M3C exhibit slightly depleted δ¹³C-CO₂ values, spanning from − 8.1 to 11.3‰, aligning with CO₂ produced in soil through C4 plant decay (Vieth and Wilkes 2010). Though these samples show a somewhat enriched ratio compared to C4 plants, it can be attributed to the diffusive effect of 12C towards the atmosphere, causing a slight ¹³C enrichment in the soil. This results in a δ¹³C value less negative than − 10‰. These findings corroborate the pattern observed in the Miller-Tans plot, revealing two main carbon sources influencing the two hydrostratigraphic units differently, although not exclusively to the St or the Aa unit.
5.4. THE IMPORTANCE OF SEASONAL BACKGROUND MONITORING
The usefulness of groundwater monitoring in CCS projects relies on the potential of a leakage signal detection using conventional hydrogeochemistry, which offer a well-established, usually low-cost, and reliable set of approaches. Nonetheless, as pointed out by Jones et al. (2015), tracking CO₂-impacted groundwater plume is complex due to its spatially restricted nature (as they tend to be long and narrow due to a constant groundwater flow direction of the overlying aquifer), which makes the monitoring and the discovery of its presence much more difficult. Zielinski et al. (2023) also showed in a thorough review that numerous studies on controlled-release CO₂ in shallow groundwater exhibited the detection of elements freed from the aquifer mineral framework in a high frequency sampling setting, which is not always the case. Besides, they also found that seasonal background monitoring is usually not considered in this type of studies, making its reproducibility in real cases not straightforward.
One possible approach to overcome these issues is to monitor the natural range of seasonally influenced overlying shallow aquifers and establish the upper and lower limits within which groundwater parameters usually fall. A similar idea was proposed by Berger et al. (2019), although they argued in favor of creating statistically determined limits that depend on each parameter's sensitivity to CO₂. Here, the data regarding the seasonal hydrochemical variation of the analyzed parameters in the two aquifer units were assessed using individual control charts, also known as Shewhart Charts (Swamidass 2000; Iglesias et al. 2016). The aim was to observe the data points collected at specific time intervals to comprehend variations and trends in the background and establish the upper and lower bounds (Upper Control Limit – UCL and Lower Control Limit - LCL, respectively) of the hydrogeochemical processes involving water-mineral matrix interaction.
Hence, the study conducted by Do et al. (2022), which carried out CO₂ injection tests in a lithological and hydrogeological context that is very similar to the study area of this paper (granite-derived setting) was used as a reference. In their study, the main parameters seen to vary were Ca, Mg, Na, K, HCO3-, Mn, Sr, Ba, pH and EC (apart from pCO₂ and δ¹³C-DIC). Assuming that the variation (increase or decrease) from the average background values were the same as in their case, these parameters would display the following behavior (seen in yellow dashed lines in Fig. 13): Ca (+ 56.10 mg/L), Mg (+ 9.72 mg/L), Na (+ 28.74 mg/L), K (+ 2.15 mg/L), HCO3- (+ 3,355 mg/L), Mn (+ 148.18 µg/L), Sr (+ 9.19 µg/L), Ba (+ 65.99 µg/L), pH (-1.28 unit) and EC (+ 488 µS/cm).
Tacking this example, depending in which season the CO₂ controlled release would occur (or, in a real case, when the CO₂ leakage might be manifested), K, Sr and Ba could not be used as hydrochemical tracers as they fall within the natural variation range and could not be reliably used as indirect indicators of CO₂ leakage. On the other hand, Ca, Mg, Na, HCO3-, Mn, pH and EC could potentially be used since their natural seasonal variation are restricted to their respective UCL and LCL, and theoretically expected increase and decrease are at least some units out of this range (Fig. 13).