Introduction

The salinity of groundwater has been studied intensively in recent decades, especially in coastal aquifers (Custodio and Bruggeman 1987; Richter and Kreitler 1993; Vengosh and Rosenthal 1994; Calvache and Pulido-Bosh 1997; Petalas and Diamantis 1999; Fedrigoni et al. 2001; Vengosh et al. 2005; Boschetti et al. 2011; Mastrocicco et al. 2012; Ghiglieri et al. 2012; Parisi et al. 2022; Lovrinović et al. 2023).

The sustainable management of the available reserves of groundwater calls for sufficient knowledge of the spatial distribution of fresh and saline groundwater and, in particular, the processes that determine variations in salinity over time and space. As the current distribution of fresh and saline groundwater in aquifers often reflects the previous hydrological conditions, it is not always clear to what extent the current groundwater state is due to long-term hydrological processes or to recent changes which are, in part, anthropogenic (Risacher et al. 2003; Boschetti et al. 2007; Yechieli and Sivan 2011). Thus, understanding the processes and factors that control the evolution of water with high salinity levels in aquifers, even in continental aquifers, is an indispensable scientific challenge.

Identification of the geochemical and hydrological processes that control water salinity, and the origin of the salinity, is therefore essential to predict long-term variations in the quality of groundwater. However, such problems involve more than one process and, frequently, only one or two are investigated contemporaneously. Some researchers have developed an approach that studies these issues from a purely geochemical and/or geophysical perspective (Di Sipio et al. 2006; Goes et al. 2009; Stanton et al. 2017; Torrese 2023; Torrese and Pilla 2021; Pilla and Torrese 2022; Rainone et al. 2015). Others have adopted an approach to groundwater salinity that focuses on upconing (Ma et al. 1997; Nettasana et al. 2012; De Louw et al. 2013; García-Menéndez et al. 2016). Others have linked an increase in the salinity of near-surface groundwater to mixing with ascending deep brines (Nanni and Zuppi 1986; Italiano et al. 2012).

The main aim of this work was the development of a conceptual model of the flow of high-salinity groundwater in the continental alluvial aquifer of the Oltrepò Pavese in the Po plain, northern Italy. This aquifer, one that is representative of many other continental alluvial aquifers across the globe, is distinguished by the upwelling of high-salinity paleo-water from underlying Mio-Pliocene deposits and the adjacent Apennine mountains. The study, carried out on a regional scale and which considered both the complex hydrostratigraphic architecture and the hydrogeological and hydrochemical conditions over a long period of time (6 years), highlighted that the groundwater flow is affected by a mass transfer, through a piston flow mechanism of the saline waters that are forced to rise along deep discontinuities and reach the shallow aquifer.

This article is the third in a series of articles written by the present authors that is focused on the hydrochemical and geophysical characterization of the continental alluvial aquifer of the Oltrepò Pavese Po plain. The first article, Torrese and Pilla 2021, revealed that the spatial distribution of saline-water contamination and the degree of salinisation in the alluvial aquifer are affected by the aperture of the discontinuities within the bedrock, the hydraulic conductivity of the aquifer, and the seasonal variations of freshwater recharge; the authors, based on early results, hypothesized a pressure transfer, through a piston flow mechanism, which can produce a mass transfer where saline waters are forced to rise along discontinuities within the bedrock and reach the shallow aquifer. The second article, Pilla and Torrese 2022, hypothesized a hydraulic connection between the aquifer in the nearby Apennine Mountain range and the continental alluvial aquifer and provided further clues around the piston-flow-mechanism-based hypothesis.

This article conclusively reveals evidence of the piston-flow mechanism, based on the results of the two previous studies, new, focused hydrochemical and geophysical investigations and especially continuous monitoring of electrical conductivity, temperature and piezometric level of the groundwater carried out using multiparameter probes.

In this paper, the term “contamination” was used to describe the ascent of deep saline waters into the alluvial aquifer rather than the intrusion of saline waters. Furthermore, according to Pilla and Torrese (2022), water salinity (Table 1 in Stanton et al. 2017) was classified according to these classes: freshwater with electrical conductivity (EC) < 500 μS/cm, slightly brackish water with EC ranging between 500 and 4,000 μS/cm, moderately brackish water with EC ranging between 4,000 and 8,000 μS/cm, highly brackish water with EC ranging between 8,000 and 12,000 μS/cm, saltwater with EC ranging between 12,000 and 70,000 μS/cm, and brine with EC > 70,000 μS/cm. The term ‘saline water’ includes both brackish water and saltwater.

Hydrogeological setting

The Oltrepò Pavese plain sector features Miocene – Pliocene marine rocks with very low hydraulic conductivity, formed by high-clay-content deposits (clays, sandy clays, sandy-marls, sandstones, conglomerates, gypsum-rich marls and calcareous marls) covered by alluvial Quaternary sediments (Pellegrini and Vercesi 1995).

The upper Quaternary deposits represent the area’s main water-bearing units and are deposited principally by the action of Po River and the Apennine streams. The Quaternary deposits include three distinct hydrogeological units: Pre-Würmian (>115 thousand years Before Present, kyr BP) alluvium; middle-ancient (Würmian, 115 kyr–11.7 kyr BP) alluvium; and modern (Post-Würmian, <11.7 kyr BP) alluvial deposits (Cavanna et al. 1998; Pilla et al. 2007; Fig. 1). Only the latter two units are impacted by the presence of sodium-chloride-rich waters.

Fig. 1
figure 1

a Geographical and geological setting of the Oltrepò Pavese plain sector, and b with hydrogeological sections (adapted from Pilla and Torrese 2022)

Full size image

Most of the Oltrepò Pavese plain sector is occupied by middle-ancient alluvial deposits. These are formed by sand and gravel with interbedded clays or clayey silts. The permeability of these deposits is 10–3–10–4 m/s (Pilla et al. 2007). The modern alluvial deposits are distributed mainly along the Po River and originated with the river’s Post-Würmian (<11.7 kyr BP) depositional activity. The deposition of these alluvial deposits is also contributed to by the largest Apennine streams. Further, the constant presence of a clayey silty covering with a permeability of 10–7–10–8 m/s—locally 10–9 m/s (Pilla et al. 2007)—has a varying maximum thickness of 10–15 m in the sectors close to the Apennine margin and a minimum of 2 m in the meandering area of the Po River, thereby limiting the infiltration and influencing the aquifer recharge. This limitation occurs mainly in close proximity to the coalescent fans formed from the deposition of Apennine streams (Pilla et al. 2007). An aquifer recharge contribution from the Po River is unlikely; in fact, groundwater flow direction is towards the Po River, except during occasional flooding events (Pilla et al. 2007). Therefore, the source of shallower, fresher groundwater (end member for mixing) is the Apennine streams in the sectors close to the Apennine margin. These are calcium bicarbonate waters. The aquifer is single-bedded unconfined, although it becomes locally and temporarily confined when the piezometric level of the groundwater reaches the impermeable clayey silty covering during certain periods of the year.

The structural setting of the Oltrepò Pavese plain (Tibaldi et al. 2023) is highly conditioned by a significant tectonic discontinuity called Vogherese Fault (Boni 1967; Fig. 1). This fault is situated below several tens of meters of alluvial deposits. On a regional scale, it has a NE–SW direction, stretching from west of Casteggio to the Colle of S. Colombano (in the Pavia plain area), crossing the confluence of the Ticino and Po rivers (Boni 1967). Boni (1967) believed it to be a normal fault with a hanging wall to the NE, up to the Barbianello area. Here, it becomes a reverse fault, south of Pinarolo, and gradually deviates westwards, north of Casteggio and Voghera. As no specific recent studies have been carried out, and given that the definition of the fault type is outside the scope of this paper, Vogherese Fault has always been traced as a vertical fault on the cross-sections of this study. Vogherese Fault is the cause of the sudden deepening of the hydrogeological bedrock in the northern sector of the Oltrepò Pavese plain and is also responsible for the aquifer’s highly variable thickness (Fig. 1). The aquifer displays a thickness of a few meters in the southern sector, which represents the up-thrown block, and >100 m in the northern sector, in the down-thrown block (Braga and Cerro 1988; Cavanna et al. 1998; Regione Lombardia and ENI Divisione AGIP 2002; AGIP 1972; Fig. 1).

Two deep boreholes (500 m) resulting from drilling for oil exploration are also in the area (AGIP 1972; Regione Lombardia and ENI Divisione AGIP 2002). At a depth of 275 m, the Casanova Lonati 1 oil borehole (Fig. 1), located on the down-thrown block of the Vogherese Fault, intersects the underlying marine deposits. At 136 m depth, this borehole intercepts brackish groundwater that rises along the fault and flows into the continental deposits. Self-potential and resistivity logs, as well as formation testing, have revealed that groundwater intercepts a body of saltwater at 400 m depth. Further, formation tests identified sodium chloride up to a concentration of 34.4 g/L in the bedrock at depths of several hundred meters. The Stradella 1 oil borehole (Fig. 1)—the second of the oil exploration boreholes—is located on the up-thrown block of the fault, intercepting the underlying marine deposits at a depth of only 36 m.

This particular setting facilitates the ascent of saline waters. It is most evident in the southern sector of the Po plain where the aquifer is thinner and where the sodium-chloride-rich waters cannot be diluted by the more abundant calcium bicarbonate groundwaters. This setting affects the chemistry of the groundwater greatly. The brines (very high-density fluids) that are remnants of evaporated marine waters in the late Messinian, trapped at the bottom of the Po plain aquifer, are thought to be the source of the sodium-chloride-rich waters (end member for mixing; Conti et al. 2000; Regione Lombardia and ENI Divisione AGIP 2002). Structural discontinuities localized in the bedrock of marine origin also facilitate the ascent of saline waters. These discontinuities represent the saline waters’ preferential flow paths and facilitate the flow towards the alluvial aquifer.

However, this contamination can prevent the exploitation of groundwater in much of the Oltrepò Pavese alluvial plain. This not only concerns drinking water supply but also agricultural and industrial use. The realization of new wells that can be fully exploited is, therefore, very problematic (Cameron et al. 2018, 2019).

Hydrochemical framework

The Oltrepò Pavese alluvial aquifer is characterized by the presence of waters from two types of chemical hydrofacies: waters with a calcium bicarbonate hydrofacies, which are the most common and distinguished by an average salinity level (EC < 1,500 μS/cm), and waters with a sodium chloride hydrofacies, which presented a higher degree of salinity (EC >> 3,000 μS/cm). Due to the ascent of deep saltwater into the alluvial aquifer, waters that derive from a more or less pronounced mixture of these two hydrochemical poles are also present. They have been localized in specific sectors of the investigated area straddling a tectonic discontinuity such as Vogherese Fault (Pilla et al. 2007, 2010, 2015; Torrese and Pilla 2021; Pilla and Torrese 2022) and, probably, in some sectors affected by another tectonic discontinuity known as Marginale Fault, as well as lesser tectonic discontinuities associated with them.

The waters with a sodium chloride hydrofacies are mostly localized in three areas along Vogherese Fault, as hydrochemical investigations have shown: to the north of Casteggio where the chloride values can exceed 10,000 mg/L; west of Barbianello where the highest concentrations (above 4,000 mg/L) of chloride were recorded, and the sectors that include Mezzanino and Albaredo Arnaboldi, where chloride concentrations can reach 3,000 mg/L (Pilla and Torrese 2022; Fig. 2).

Fig. 2
figure 2

Zones in the Oltrepò Pavese aquifer with the greatest chloride contamination, with reference to the Vogherese Fault and to the main NE–SW trends of high-conductivity anomalies revealed by VLF-EM surveys and which can be correlated with the ascent of saline waters along secondary tectonic discontinuities (adapted from Pilla and Torrese 2022)

Full size image

The studied area also presents, locally, a third water type: a calcium-sulphate hydrofacies and a relatively high salinity (SO42–: 800–2,700 mg/L; EC: 1,500–5,000 μS/cm). Its origin is associated with evaporite layers within the Gessoso-Solfifera formation that are present locally at the base of the Oltrepò Pavese aquifer (Bersan et al. 2010). These waters were sampled at a single well (well 15) sited in Barbianello. In other sectors of the Oltrepò Pavese plain, waters with lower concentrations of sulphate were identified (Broni: 210–220 mg/L; Redavalle: 240 mg/L; Casanova Lonati: 230–270 mg/L; Mezzanino: 250 mg/L; sector north of Casteggio: 200–210 mg/L).

Most of the wells with the highest electrical conductivity values are sited within a distance of 1.5 km from Vogherese Fault in the southern sector and within a distance of 1.0 km from the fault in the northern sector. A few contaminated wells were also noted some distance from the fault (~4 km), both in the northern and southern sectors. These are most likely connected to secondary tectonic discontinuities that convey deep saltwater into the alluvial aquifer. Further, in the most contaminated zone (<1 km), while some wells are not contaminated by chloride, some are characterized by low electrical conductivity values (<1,500 μS/cm up-well; <2,000 μS/cm down-well; Pilla and Torrese 2022).

The sodium chloride groundwater has notably higher salinity levels (EC values can be above 12,000 μS/cm and are, on average, between 2,000 and 5,000 μS/cm). As the concentrations of other major ions are somewhat similar to those of the calcium bicarbonate water of the Oltrepò plain, the EC values are related to the dissolved chloride and sodium (Pilla and Torrese 2022). The correlation between the chemistry of the aquifer and the trend of the Vogherese Fault is not always seen: some wells, though close to the fault, show low groundwater salinity (wells 32, 34 and 35).

Materials and methods

Hydrochemical investigations

A total of 146 wells, which are mainly located close to Vogherese Fault (Fig. 3), were periodically sampled and analyzed as part of the hydrogeochemical study. Sampling was conducted between July 2007 and July 2013 and involved monitoring hydrogeochemical characteristics, groundwater depth, in situ temperature (T), electrical conductivity (EC) and redox potential. In addition, vertical logs of EC and T were carried out on 63 wells from July 2008 and were used to monitor the freshwater–saltwater interface.

Fig. 3
figure 3

Map that shows the distribution of wells used in this study along with the investigations carried out in the area

Full size image

Piezometric levels were measured with a PASI BFK 100 hydrostatic probe; electrical conductivity and temperature data were acquired using a WTW LF597 conductivity meter; redox potential was measured using a WTW pH 340/ION. Groundwater sampling from the wells was conducted using submersible electric pumps supplied with the well, a Cellai 504S peristaltic pump or a bailer.

On three wells in the study area (Fig. 3), those characterized by water with a high salinity, continuous monitoring over a long period (2008–2014) of the parameters including electrical conductivity, temperature and groundwater level was conducted using Diver CTD (conductivity, temperature, depth) multiparametric probes. The data acquired were corrected with barometric compensation of the atmospheric pressure with the Baro CTD probe.

Geophysical investigations

A four-dimensional (4D) time-lapse, electrical resistivity tomography (ERT) survey was used in the geophysical study to track the time-varying groundwater contamination. The time-lapse monitoring involved the repetition of a three-dimensional (3D) ERT survey, labelled as 3D_ERT in Fig. 3. This survey was carried out in June 2013 and again in May 2015. A surface snake grid was used, whereby one comprised 12 × 4 electrodes spaced 10 m apart along both the X and Y axes. A sequence of 540 quadrupoles, employing all feasible X, Y and diagonal dipole–dipole array set-ups, was used. ERTLab Solver (Release 1.3.1, by Geostudi Astier s.r.l., Multi-Phase Technologies LLC) based on tetrahedral finite element modelling (FEM) was used to perform data inversion.

Both forward and inverse modelling used tetrahedral discretization. The foreground region was discretized using a 5-m cell size, i.e., half the electrode spacing, in order to produce a highly accurate model. The background region was discretized by gradually increasing element size towards the outside of the domain, following the subsequent sequence 1×, 1×, 2×, 4× and 8× the foreground element size. Forward modelling was carried out using mixed boundary conditions (Dirichlet-Neumann) and a tolerance (stop criterion) of 1.0E-7 for a symmetric successive over-relaxation conjugate gradient (SSORCG) iterative solver.

A least-squares smoothness constrained approach was used for data inversion (LaBrecque et al. 1996). A data-weighting algorithm that allows the variance matrix to be adaptively changed after each data point iteration poorly fitted by the model was used to manage noise (Morelli and LaBrecque 1996). Inverse modelling was conducted with a maximum number of internal inverse preconditioned conjugate gradient (PCG) iterations of five and a tolerance (stop criterion) for inverse PCG iterations of 0.001. The level of roughness from one iteration to the other was controlled to evaluate maximum layering: a low value of reweight constant (0.1) was established to obtain maximum heterogeneity. Inversion included the application of homogeneous starting models that set the average measured apparent resistivity value at each node. Based on the minimum data residual (or misfit error), the final inverse resistivity models were chosen.

Cross-validation of geophysical results with well logs

Piezometric level measures, water sampling and analysis, electrical conductivity, temperature and redox potential logs carried out at well 30 provided the opportunity for 3D ERTs to be cross-validated with extensive ground truth. Well 30, sited at the center of 3D ERT, is 50 m west of a saltwater contamination area, which is why the well intersects brackish groundwater and not saline groundwater. Well 30 logs found a drop in resistivity (EC log) at a depth of 11 m as well as an increase in temperature and a decrease in the redox potential of the groundwater. This well transition zone occurs in a section of the aquifer in which, based on the bulk resistivity suggested by the ERT survey, the aquifer should feature coarser deposits. Precisely, maximum resistivity values can be identified at a depth of 11–12 m, which would indicate the probable presence of a clean sand layer contaminated by brackish water. This would suggest that the distribution of saline water contamination is a hydraulic conductivity-controlled process at the plume scale within alluvial deposits.

Well 30 provided resistivity and temperature logs conducted on 5 June 2013 and 4 July 2013, as well as the redox potential log carried out on 4 July 2013. Detailed analysis of the resistivity log undertaken on 5 June 2013 and used for the cross-validation of 3D and time-lapse ERTs, found a transition between slightly brackish groundwater (1,430 μS/cm on average, ≈1,000 mg/L TDS, 1–10 m of depth, June 2013) to underlying moderately brackish groundwater (5,262 μS/cm on average, ≈3,700 mg/L TDS, 11–15 m of depth, June 2013) at 11 m depth. Similarly, the temperature log (5 June 2013) shows a transition between shallower relatively colder groundwater (with an average temperature of 12.8 °C, 3–10 m of depth, June 2013) and deeper relatively warmer groundwater (with an average temperature of 13.5 °C, 11–15 m of depth, June 2013) at 11 m depth.

Meteo-climatic data

The meteo-climatic data used in this study were taken from seven pluviometric stations (owned by Gallini State Agricultural Technical Institute of Voghera, ARPA Lombardia and Regione Liguria) and five nivo-pluviometric stations (owned by ARPAE Emilia Romagna and ARPA Piemonte). The sites of the pluviometric and nivo-pluviometric stations used appear in Fig. 4.

Fig. 4
figure 4

Locations of pluviometric and nivo-pluviometric stations used in the study

Full size image

In the Oltrepò plain, the wet season normally lasts from October to May and the dry season from June to September. The average annual rainfall, calculated at the Voghera station in the period 2008–2014 was as follows—for 2008, 908 mm; for 2009, 896 mm; for 2010, 1,010 mm; for 2011, 659 mm; for 2012, 764 mm; for 2013, 1,033 mm and for 2014, 1,370 mm.

Results

Spatial and temporal variability of groundwater salinity inferred from conductivity logs

The distribution of saltwater determined through the electrical conductivity logs is particularly variable (Fig. 5). The contributing factor for the spatial variability of saltwater ascent into the Oltrepò Pavese alluvial aquifer is primarily governed by the location of the tectonic discontinuities—Vogherese Fault and associated tectonic discontinuities, see also Fig. 8 in Pilla and Torrese (2022). Also significant is the thickness of the alluvial aquifer, which is conditioned by the plano-altimetric development of the underlying Mio-Pliocene deposits of marine origin and, ultimately, well location.

Fig. 5
figure 5

Electrical conductivity (EC) logs taken in the wells in September 2010

Full size image

In general, groundwater salinity begins to rise at depths of 5–8 m but can increase to depths of 10–20 m. In other instances, the high-salinity depth is just below ground level (bgl), such as in certain subareas of the zone investigated in which saline waters were found in the surface drainage network situated in the first few meters of depth of the drift deposits.

Furthermore, another decidedly interesting aspect emerges from the creation of electrical conductivity logs, i.e., the temporal variability of water salinity detected in certain wells along the entire water column. As seen in Fig. 6, the minimum and maximum salinity levels in some wells are detected at different time periods—for example, in wells 27 and 36, the highest salinity levels were observed in October 2008 (well 27) and in May 2011 (well 36) while the lowest levels were recorded, respectively, in May 2011 (well 27) and in October 2008 (well 36). The same situations were also well evidenced from logs conducted at wells 123 and 132. Maximum and minimum salinity levels at the bottom of the well were registered, respectively, in March, November and December 2011 and in October 2011. At well 123, the maximum and minimum salinity levels at the bottom of the well were observed in October 2011 and November 2011, respectively. Such temporal variation and the apparent anomalies in the salinity of the Oltrepò Pavese alluvial aquifers will be discussed further in subsequent sections.

Fig. 6
figure 6

Variation of water electrical conductivity recorded at different times in wells a 27, b 36, c 132, and d 123. The data shown in this figure were partially shown in Figure 7 in Pilla and Torrese (2022)

Full size image

Temporal variability of groundwater salinity inferred from the use of multiparameter probes

Continuous monitoring of electrical conductivity, temperature and piezometric level was carried out using multiparameter probes (Fig. 2) at three wells in the study area (wells 27, 36 and 136). These wells all featured highly saline water. The study made it possible to check variations in water salinity over time due to ascending saline waters in alluvial groundwaters.

Evidence of mass transfer induced by piston flow from hydrochemical investigations

Well 27

Well 27, situated in Albaredo Arnaboldi (eastern sector of the study area; Fig. 2), was monitored from December 2008 to August 2009. The well is used for agricultural purposes and, owing to the high salinity of its waters, has not been used for a long period. Therefore, throughout the monitoring period, groundwater was not pumped from the well. The well is 8 m deep and a multiparametric Diver CTD probe was positioned at a depth of 7.50 m. Data were acquired every 12 h (at 12:00 and at 24:00).

Electrical conductivity of the waters in this well showed a constant decrease in the period from December 2008 to August 2009, from a value over 9,000 μS/cm to little more than 5,000 μS/cm (Fig. 7). In the period from January to April 2009, electrical conductivity of the groundwaters experienced sudden variations, from a minimum of 140 μS/cm to a maximum of ~10,000 μS/cm.

Fig. 7
figure 7

Variation of the piezometric level and the electrical conductivity of the waters of well 27 as a function of the rainfall (Voghera pluviometric station) in the period December 2008–August 2009

Full size image

Such variations occurred during a period of intense and prolonged precipitation that impacted the groundwater level of the groundwater. The groundwater level varied from an approximate depth of 1.5 m in December 2008 to an average depth of 0.1–0.5 m in January–April 2009. On two occasions, it exceeded ground level: 6–12 February 2009 and 27 April–7 May 2009. The last two episodes coincided with two intense rainy periods that caused the flooding of agricultural fields in areas adjacent to the well.

In particular, during 6–12 February 2009, rainfall was measured at 135 mm in the high Apennines (Passo del Giovà rainfall station), at ~75 mm in the low-mid Apennines (Fortunago and Canevino rainfall stations) and 67 mm in the Oltrepò Pavese plain (Voghera rainfall station).

Ongoing monitoring of well 27 allowed the causes of the continual oscillation in electrical conductivity to be assessed, a phenomenon that had already been evidenced by chemical analysis. This revealed that, for these waters, extremely variable concentrations in chloride were found, from a maximum of 3,000 mg/L (October 2008) to a minimum of 174 mg/L (June 2010). Electrical conductivity logs also suggested clear variations over time in the depth of the transition band between freshwater and brackish/saltwater (Pilla and Torrese 2022), which varied between 6 and 8 m depth. The well is, in fact, sited in correspondence to a slight depression where, in conjunction with heavy rains, rainwater tends to stagnate on the surface and infiltrate the interspace between the well and the surrounding terrain. Such infiltrations led, during periods of more intense rain, to the immediate dilution of groundwater and the lowering of electrical conductivity from inside the well (Fig. 7). Over time, electrical conductivity of the waters gradually increased, returning to the values before these rainfall events. With each new intense precipitation event, rainwater again entered the aquifer, leading to a further and significant dilution of the groundwater with a consequent and instantaneous decrease in the salinity of the water in the well being monitored.

An interesting fact emerged from the data acquired: there was a sharp increase in electrical conductivity of the waters that was registered after a delay of a few days (respectively, 3 and 7 days) from the two events with the most rainfall (6 February 2009 and 27 April 2009). On these occasions, the electrical conductivity of the waters reached, respectively, 9,470 and 9,980 μS/cm, markedly higher than that measured in the periods before January 2009.

Well 36

Well 36, located north of Santa Giuletta, close to Vogherese Fault (Fig. 2), is a well for domestic use that is a little less than 6 m in depth. It is used especially during summer months for the irrigation of a small plot of land dedicated to the cultivation of vegetables and is practically unused during winter. Well 36 was monitored for the period June 2009–May 2014 with data acquired every 12 h (at 12:00 and 24:00), with the Diver CTD probe positioned at a depth of ~5 m.

In summer 2011, the period of the year in which the well is used most, monitoring and scans were also taken every 5 min for short periods, from 11 to 25 July (Fig. 8), in order to verify how the use of the well would alter the groundwater level and salinity. In this period, the static level of the groundwater presented a groundwater level of a 2.5 m depth from ground level and electrical conductivity that settled at ~4,200 μS/cm. During the pumping of the well (water flow of a few liters per second) a sudden decrease in both the groundwater level and electrical conductivity was observed. This may be interpreted by an initial drawing of less saline water (800 μS/cm) from lateral and shallower sections of the aquifer. When the well was turned off, a rapid restoration of the well’s groundwater level and an increase in electrical conductivity of the waters up to 6,000–7,000 μS/cm was observed, induced by the recall towards the well of highly saline waters from deeper sections of the aquifer due to upconing. The preperturbation balance was re-established after a few days.

Fig. 8
figure 8

Depth to groundwater level and the electrical conductivity of well 36 in the period 11–25 July 2011 with a 5-min data acquisition scan

Full size image

Figure 9 instead reports the data acquired for the entire monitoring period, with scans taken every 12 h. The groundwater level tended towards the minimum in later summer (–4 m summer 2009; –3.5 m summer 2010, –3.8 m summer 2011 and –3.5 summer 2012), due to the high levels of groundwater use for irrigation and low rainfall. The groundwater volumes generated by the abundant precipitation in subsequent months (autumn–winter) brought a significant increase in groundwater levels, reaching maximum values (1 m from ground level). During autumn–winter 2009/2010, the maximum groundwater levels (ca. –1.5 m) were reached gradually from around February 2010. In contrast, in autumn–winter 2010/2011, the maximum groundwater levels (ca. –1 m) were reached more quickly, from the beginning of December 2010, due to summer values that were ~50 cm higher than the previous year’s and by more rainfall. On the other hand, spring 2010 was decidedly rainier than 2011; the groundwater level stayed at maximum levels for almost all of spring up to mid-May, when it began to fall. Instead, during mid-March 2011, the groundwater level started to increase: intense rainfall had an immediate effect on the groundwater level, causing it to rise. The year 2012 saw low precipitation and the groundwater level in well 36 was markedly lower (ca. 2.5 m). In 2013 and 2014, the groundwater levels reached values similar to those in 2010 and 2011 but more gradually, especially in 2013.

Fig. 9
figure 9

Electrical conductivity and depth to groundwater level in well 36, and rainfall (Voghera pluviometric station) recorded in the period June 2009 to May 2014

Full size image

The salinity of the waters monitored did not present seasonal variation as the groundwater levels did. In Fig. 9, short-term pulses of increased salinity can be observed; these periods are due mainly to groundwater being drawn from the well, especially in summer. Those variations that develop over longer periods do not appear to be due to anthropic reasons. The main groundwater ‘salinity crises’ were identified as May–June 2010, December 2010–January 2011, February –April 2011, April –June 2011, April 2013–February 2014 and February 2014 to beyond April 2014. A more modest salinity crisis, though a significant one, as will be discussed later, occurred on 6–13 November 2011.

In 2013 and 2014, electrical conductivity values of the Oltrepò Pavese aquifer waters, in winter–spring, reached ~5,400 and 8,000 µS/cm, respectively. In May 2010, electrical conductivity rose by ~2,100–5,000 µS/cm, in December 2010 by ~2,300–5,800 µS/cm, in February 2011 by ~4,300–9,200 µS/cm, and in May 2011 by ~5,800–9,000 µS/cm. In 2012, it remained at markedly lower values ​​(1,700–1,900 µS/cm) which, generally, represent the basic level of groundwater salinity.

The increases in the salinity described seem anomalous as they occurred at the same time as the groundwater’s presented maximum groundwater levels (late autumn–winter–spring), bringing certain episodes of groundwater dilution from the arrival of low salinity waters. Further, from the diagram in Fig. 9, it can be observed that these crises of salinity had ever-increasing magnitudes and reached a peak in spring 2011 and spring 2014. These periods were distinguished by, as mentioned previously, particularly elevated groundwater levels that lasted for a long period due to abundant autumnal, winter and spring rains.

As suggested, this anomalous behavior cannot be exclusively the result of well pumping; the periods of greater perturbation of the groundwater induced by anthropic factors, that is the pumping of well 36, can be seen clearly during summer, while in the rest of the year fluctuations in groundwater level, even sudden variations, are due to natural phenomena, as will be illustrated, namely the increase in the groundwater owing to precipitation.

From what has been stated, one plausible explanation for the increase in salinity of the groundwater is determined by a larger quantity of saltwater coming straight from the substrate of marine origin towards the aquifer and not from the mobilization owing to anthropic phenomena of saltwater from the lower sections of the aquifer, the presence of which was underlined by the electrical conductivity logs made. This does not exclude the fact that the two mechanisms of well salinization may, in certain periods, coexist.

In 2010 and 2011, two events occurred that clarified which mechanisms could be the cause of the increase in salinity of the groundwater. As shown in Fig. 10, during November 2010 in the mid-high Apennines, abundant snowfall events were recorded in the entire northern Apennines. At the Lagdei snow-rain station, snow cover reached 70 cm depth. In other snow-rain stations, snow cover varied from ~30 cm to 50–60 cm.

Fig. 10
figure 10

Snow depth in the Northern Apennines in winter 2010. The locations of the snow stations (nivo-pluviometric stations) are shown in Fig. 4

Full size image

From 8 December, thanks to a significant rise in atmospheric temperatures and to several precipitation events, though not excessive, the depth of snow cover was reduced by a few centimeters or completely melted. This certainly induced a rapid recharge of the mountain aquifers, triggering a piston flow along the main discontinuities in the Tertiary terrains and a mass transfer towards the Oltrepò Pavese aquifer that produced a notable increase in the salinity of the groundwater, whereby salinity values of the groundwater recorded in well 36 increased by 2,300 to ~6,000 μS/cm. This rise occurred after ~3 days.

A second, more minor, event was registered during a catastrophic flood event in the city of Genoa. Precipitation values at northern Apennine rain stations along a route towards Genoa recorded decidedly elevated precipitation, especially in the sectors at higher altitudes. In the mid and high Apennines, recorded precipitation was, respectively, over 200 and 400 mm. Along the Oltrepò Pavese flood plain, precipitation remained at nonexcessive values (75 mm at the Voghera station).

In well 36, this event was recorded with a raising of the groundwater level from a depth of 325–250 cm and by an increase in the salinity of the waters of 3,000 to ~3,250 μS/cm (Fig. 11) some 3 days after the flood event. Also in this case, the variations in water salinity could be due to piston flow phenomena with a successive mass transfer. In the post-flood period, water salinity returned to levels slightly lower than the pre-flood period.

Fig. 11
figure 11

Electrical conductivity and depth to groundwater level in well 36, and rainfall (Voghera pluviometric station) recorded in the period 21 Oct 2011–20 Nov 2011

Full size image

Well 136

In the period from 20 January 2012 to 14 March 2012, the waters in well 136 were continuously monitored thanks to the installation, at a depth of 5.3 m, of a Diver CTD multiparametric probe. Electrical conductivity, temperature and groundwater level data were acquired every 12 h (at 12.00 and at 24.00).

The well, located to the west of Santa Giuletta, is in a sector between the Vogherese and Marginale faults (Fig. 2), intersected by a confined aquifer with a spontaneous flow of around 1 L/s with a piezometric level of around 2.25 m above ground level (agl). It was therefore necessary to install at the mouth of the well a pipe measuring around 2.5 m to prevent the recall of saltwater and, thus, prevent natural upconing in order to lower and stabilize the well water’s salinity which, initially, showed electrical conductivity values of ~5,500 μS/cm. These values subsequently dropped by an average of 1,100–1,250 μS/cm.

In order for comparisons to be made, Fig. 12 reports precipitation data, groundwater levels and the electrical conductivity values recorded for well 36. It is worth recalling that the lowest salinity values at well 36 were recorded during the investigation period of well 136, due to the modest rainfall at the end of 2011 and beginning of 2012.

Fig. 12
figure 12

Electrical conductivity and piezometric level in well 136 (confined aquifer) and well 36 (unconfined aquifer), and rainfall (Voghera pluviometric station) recorded in the period 20 January 2012 to 11 May 2012

Full size image

Examining Fig. 12, it can clearly be observed that, overall, the water from the two wells, which are 2.5 km apart, had similar trends both in terms of piezometric levels and salinity. Between 21 February 2011 and 14 March 2012, each significant increase in piezometric level, due to precipitation or the melting of snow in the Apennines, led to a corresponding rise in water salinity.

Evidence of mass transfer induced by piston flow from geophysical investigations

The location of 3D ERT (Figs. 13 and 14), centered on well 30, is in a region of the aquifer where deep saline waters that migrate upward reach the alluvial aquifer, diffuse, and mix with the fresh groundwater of the shallow aquifer. This results in the formation of groundwater that is fresh, slightly brackish, moderately brackish, and highly brackish. Different degrees of groundwater salinity can be related to different resistivity range extractions from the 3D resistivity block (Fig. 13a–c).

Fig. 13
figure 13

3D ERT carried out on 3 June 2013 found that deep saltwaters reach the alluvial aquifer during upward migration. They diffuse and mix with the fresh groundwater from the shallow aquifer, resulting in different degrees of groundwater salinity within the aquifer. The figure shows: a a saltwater to highly brackish water plume that contaminates a sandy body in the alluvial aquifer and shallow low-resistivity anomalies in the upper clayey deposits; b slightly to highly brackish water contaminating clayey sandy deposits; c slightly brackish to fresh water contaminating clayey sandy deposits. The resistivity model shows good correlation compared to the logs for well 30, which found a drop in resistivity at a depth of 11 m, together with an increase in temperature and a decrease of the groundwater’s redox potential. This suggests a well transition zone between slightly and moderately brackish water. To be compared with the subsequent figure

Full size image
Fig. 14
figure 14

3D ERT carried out on 20 May 2015 shows that water bodies (a–c) may be directly compared to bodies in shown in Fig. 13a–c; a reduction in size of the saltwater to highly brackish water plume (panel a) and a rise in the alluvial aquifer’s resistivity values during late spring of 2015 are due to 52% lower rainfall during spring 2015 compared to spring 2013; the reduction in size of the saltwater to highly brackish water plume owing to lower recharge suggests a pressure transfer, via a piston flow mechanism, and, consequently, a mass transfer that forced a rise of saline waters towards the shallow aquifer

Full size image

Here, 3D geophysical imaging cross-validated by well logs allowed the mapping of the irregular-shaped, shallow, highly brackish water plume that contaminates the alluvial aquifer. The water plume, which contaminates a sandy body within the alluvial deposits, corresponds to the plume extracted from the 3D ERT (4–9.3 ohm·m, Fig. 13a). The 2013 3D resistivity block (3 June 2013) provides good correlation with well 30 logs (5 June 201; Fig. 13). Moderately brackish water presents lower redox potential and a higher temperature than the slightly brackish water that overlies it.

Inverting the two data sets collected in late spring 2013 and 2015 allowed the time-lapse monitoring of the highly brackish water plume contaminating the alluvial aquifer. This analysis revealed the regions of the model in which a change in electrical resistivity has been verified and provided suggestions about the temporal variability of groundwater salinity. A reduction in size of the highly brackish water plume (Figs. 13a and 14a) and a rise in resistivity values of the alluvial aquifer (Figs. 13b,c and 14b,c) owing to groundwater salinity dilution in late spring 2015 were observed.

Groundwater salinity dilution revealed by the 2015 block model (Fig. 14) compared to the 2013 block model (Fig. 13) is probably a result of the 52% lower rainfall during spring 2015 than in spring 2013. It is likely that the ascent of saltwater is induced by an increase in hydraulic head within the main Apennine groundwater following rainfall or snow melting events as the temporal variability of groundwater salinity is associated with the variability in the discharge of ascending saltwater. This system seems to be in hydraulic connection with a deeper aquifer that hosts saline waters, which suggests a pressure transfer, through a piston flow mechanism, which can produce a mass transfer in which saline waters are forced to rise through discontinuities to the shallow aquifer. This was seen on one hand to coincide with aquifer recharge (maximum values of water table elevation), while on the other, the 2015 block model displays a fall in the resistivity of the upper clayey deposits (Fig. 14). This is due to rainfall the previous day affecting only very shallow resistivity values.

Discussion

The main aspects that emerged from this study are:

  • The high variability in salinity of the alluvial aquifer waters in many periods of the year in which continuous monitoring was active in some wells. Of these, the well that provided the best information was well 36.

  • The high variation in space and time of the salinity of the waters, which emerged from the execution of electrical conductivity logs carried out along the water column of the monitored wells and by the 3D ERTs created.

Concerning the first aspect, the salinity crisis recorded in wells monitored continuously (wells 27, 36 and 136) did not originate from seismic activity, even at low magnitudes, which could have induced ‘forced’ lifting and squeezing of highly saline waters along tectonic discontinuities in quantities that might be modest but are, in any case, able to significantly modify the salinity of the groundwater (Nanni and Zuppi 1986; Italiano et al. 2012).

In order to test this hypothesis, data related to earthquakes that occurred between 2008 and 2014—the period the wells were monitored—with a magnitude equal to or greater than 3.0 were gathered in a large zone around the actual study area (Moschillo and Pignone 2020). An examination of this data—Fig. S1 in the electronic supplementary material (ESM)—did not show any temporal correlation between salinity crisis and seismic activity.

As suggested by Nanni and Zuppi (1986), an influence linked to the migration of CO2, CH4 and H2S gases, as has been hypothesized in similar contexts, can be ruled out, as the gas dissolved in the waters with greater salinity at the areas with thermal-mineral centres (Salice T.: Cl = 35,250 mg/L; San Colombano al Lambro: Cl = 23,865 mg/L; Rivanazzano: Cl = 14,854 mg/L) show low concentrations that are only slightly higher than reference ASW (air-saturated water) values (Bocca 2006).

The apparently anomalous behavior must, therefore, be due to other factors. One of the most accredited theories is that at the origin of the ascent of saline waters is a pressure transfer followed by a mass transfer induced on the deeper water circuits where the highly saline waters are found, originating from the subterranean circuits of the nearby Apennines. Fantoni and Franciosi (2010) highlight how several extremely deep tectonic discontinuities, among them the Vogherese Fault, which cross the Tertiary terrains and come into contact with tectonic discontinuities of the nearby mid-high Apennines, correspond to the study area’s strip at the foot of the Apennines. This implicates the possibility of a hydraulic link between Apennine water and deeper Mio-Pliocenic and Tertiary circuits present below the alluvial aquifer in the plain sector and in which sodium chloride waters originate from the brines at the base of the Po plain aquifers (Conti et al. 2000; Regione Lombardia and ENI Divisione AGIP 2002).

In conjunction with significant recharge episodes that affect the Apennine water circuits—that is, extensive periods of rainfall, to which is added (in spring) the melting of snow at mid-high altitudes—there is a transfer of pressure able to be transmitted along the water circuits in which saline waters reside. This pressure provides the water with the necessary push to rise along discontinuities and reach the alluvial aquifer (Fig. 15). In the case of well 36, this theory would explain the sharp increase in electrical conductivity of the waters recorded during the ‘salinity crises’ that affected the area in which the well is located and would be confirmed in the aquifer’s piezometric levels which, as previously suggested, reach maximum levels in autumn–winter. In light of what has been stated, the sudden rise in water salinity registered in the other monitored wells around the time of the rainfall events would also find an explanation.

Fig. 15
figure 15

Conceptual diagram illustrating the mechanism at the origin of the saline water ascent phenomenon in the Oltrepò Pavese alluvial aquifer

Full size image

The possibility of a hydraulic connection between two systems, fractured in mountainous areas and porous in foothills and in alluvial plains, would allow piston flows to propagate with a subsequent mass transfer able to reach the porous medium. Pilli et al. (2012) verified a hydraulic connection between the carbonate units of the pre-Alpine plateau whose formations are buried under the Veneto alluvial plain (northern Italy). This is facilitated by the main discontinuity systems that intersect the entire depth of the carbonate unit between the two deep fractured and superficially porous systems.

The mix of deeper, highly saline waters with fresher more superficial waters was also illustrated by Re and Zuppi (2011) in an analysis of the factors able to influence coastal aquifers in the Mediterranean basin. Ultimately, in the Oltrepò Pavese, the different variations in salinity recorded at the monitored wells, as seen clearly at well 36, suggest that water flow—which takes place through the fractured environment in the Mio-Pliocene terrain and in the older Apennine terrain towards more porous deposits of Mio-Pliocene soils and the more recent overlying terrain—is mainly due to a problem of flow through a dual-porosity (fractured and porous) system. The piston flow phenomenon—induced by the principal recharge events of the Apennine aquifers on the saltwaters present in the deep sectors of the Oltrepò Pavese alluvial plain—provokes a mass transfer on the same highly saline waters that use two distinct ‘paths’: one that is faster and probably turbulent that flows along more developed discontinuity systems (e.g., Vogherese Fault and Marginale Fault), and a slower flow in secondary discontinuities and in more porous sectors (Sanz Escudé 2007). In Oltrepò Pavese, this also affects the response of the system to these phenomena, one that translates into an immediate response or a delay that varies from several days to 10 days in the salinity of the alluvial aquifer waters. It is worth underlining that mass and pressure transfers have very different propagation velocities—mass transfer is a function of hydraulic conductivity (usually ranging from 10–11 to 102 cm/s), while pressure transfer is defined by compressional seismic wave velocity (typically 1,450 m/s in water).

In the international literature, it has been possible to find a similar situation to that described in this study in a Mexican coastal aquifer. Escolero et al. (2007), in research conducted on a karst aquifer close to Merida, in the Yucatan, observed that, concurrent with exceptional recharge events, a 20-m ascent in the freshwater–saltwater interface in correspondence to a continuously monitored well at a time in which the high hydraulic charge of freshwater should, instead, have countered the ascent in deep saltwater. The authors also observed how such waters presented an electrical conductivity of 80,000 µS/cm, much higher than seawater (~63,000 µS/cm, Appelo and Postma 2005), without, however, focusing on the origin of this phenomena. The authors stated that a piston flow effect induced by the extraordinary recharge that had occurred accounted for this behavior.

Another similar situation was found by Torrese et al. (2006) in the continental alluvial aquifer of San Pedro Sula (Honduras). Here, the plain aquifer is bordered by the Merendon mountain range to the north-west, west and south. Results of cross-correlations between the water levels in the piezometers in the plain and the precipitation values of the pluviometers located in the mountains has shown that there is a rapid aquifer reaction to precipitation, which appears to suggest that sections of the aquifer on the alluvial plain react to rainfall inflows from the mountain areas within 15 days (the maximum sampling rate of piezometry measurements). On the other hand, the extremely limited variations of piezometric levels between dry and rainy periods, linked to the fact that some water tables show minimum elevations that are much delayed compared to those of precipitation, led the authors to state that the system has slow reaction and high inertia. To summarise, the authors suggested both a deeper supply from the bottom and shallow water influx, via fractured bedrock (see Figure 4 in Torrese et al. 2006).

The second aspect to emerge from the study is, as mentioned previously, the variability in the salinity of the waters along the verticals of the wells being monitored over different locations and at different times. In fact, analyzing the results of electrical conductivity well logs conducted at different times, it emerges that the salinization of groundwater is extremely variable, as is evident in Fig. 16.

Fig. 16
figure 16

Spatial and seasonal variations of the salinity of the Oltrepò Pavese groundwater identified through the electrical conductivity logs of the contaminated wells

Full size image

In some cases, an increase in groundwater salinity is observed in conjunction with low piezometric levels, and the contrary is seen in other cases—an increase in groundwater salinity during the main recharge events of the aquifers of the nearby Apennines; and recharge events which, in any case, are also recorded with a rise in the piezometric levels from the Oltrepò Pavese alluvial aquifer. Instead, in other wells, no salinity variations of the groundwater were observed over time. These behaviors can be associated with the spatial relationships that exist between tectonic discontinuities able to convey deep, saline fluids and the location of monitored wells.

Wells constructed in correspondence with or close to hydraulically active tectonic discontinuities are affected by the arrival of highly saline groundwaters with variable delay, while the waters in wells sited in sectors not influenced by tectonic discontinuities are affected by recharge events of the alluvial aquifer. In contrast, in episodes of aquifer depletion, there is no arrival of deep saltwater from tectonic discontinuities or the arrival of lower volumes of saltwaters. Therefore, in many sectors, phenomena of dilution by uncontaminated aquifer water prevails, with a contextual reduction in the overall salinity of the waters or a slight increase in salinity, owing to the remobilization of more saline water present at the base of the alluvial aquifer.

In coastal aquifers affected by the intrusion of saltwater of marine origin, the maximum salinity expected by the water occurs in periods of little or no precipitation, at times of practically no recharge and when pumping from the well is high. From the data obtained by continuous monitoring, it was possible to observe that the most significant increases in the salinity of groundwater were reached in conjunction with the maximum piezometric levels of the aquifer recorded at the end of the highest aquifer recharge periods. Surprisingly, the least groundwater contamination was recorded in conjunction with the minimum piezometric levels.

It should also be noted that, even modest volumes of highly saline sodium-chlorinated water can generate notable increases in surface water salinity. In these conditions, the role of groundwater extractions is secondary in increasing the saline contamination of the waters. Significant extraction during summer, the period in which the aquifer is most exploited, appears not to lead, in the period investigated, to a worsening of saline contamination of the waters; increased salinization seems, instead, to be conditioned by purely natural factors.

The data collected so far seem to confirm that groundwater extractions have only a temporary effect on the ascent in saltwater–freshwater interface. However, the observation period was too brief to allow the long-term effects of anthropic interventions on the amplification of this particular contamination phenomenon of the Oltrepò alluvial aquifer. The saline water ascent that characterizes the Oltrepò Pavese area is, then, very different from that of coastal aquifers, making it, therefore, more complicated to manage.

Conclusion

This study used hydrochemical and geophysical methods to evaluate saline paleo-water mass transfer induced by piston flow mechanisms in the continental alluvial aquifer of the Oltrepò Pavese plain (Po Plain, northern Italy). The study’s findings indicate that the variation in space and time of groundwater salinity in the plain is due to a transfer of pressure and, subsequently, of mass induced on the deeper water circuits in which these saline waters reside, from the groundwater circuits of the nearby Apennine mountains. This implies a hydraulic connection between the fractured water circuits of the Apennines and the deeper, more porous Mio-Pliocene and Tertiary circuits present in the alluvial aquifer of the section of the plain and in which saline waters emerge. In conjunction with significant recharge episodes that affect the Apennine water circuits, such as prolonged periods of rainfall to which the melting of snow at mid-high altitudes is added in spring, there is a pressure transfer capable of being transmitted along the circuits in which the salty waters reside, providing them with the necessary impulse to flow along the discontinuities and reach the alluvial aquifer.

The mass transfer on the deep saline waters occurs along two distinct ‘paths’: one that is faster and probably turbulent that flows along more developed discontinuity systems (e.g., Vogherese Fault and Marginale Fault), and a slower flow in secondary discontinuities and in more porous sectors. These affect the system’s response to water salinization phenomena: the response is either immediate or with a delay of the salinization of groundwater that ranges from several to ~10 days.

Further, wells constructed in correspondence or close to hydraulically active tectonic discontinuities are affected by the arrival in the aquifer of saline waters with variable delays, while water in wells sited in sectors not affected by tectonic discontinuities are diluted by ‘fresh’ waters connected to recharge events of the alluvial aquifer.

The phenomenon of saline water ascent that characterizes the Oltrepò Pavese area is, thus, very different from that found in coastal aquifers, making them more complicated to manage. The data gathered so far appear to confirm that, while extractions from the aquifer have only a temporary influence on the upwelling of the freshwater–saltwater interface, the observation period is too short to allow predictions of the long-term effect of anthropic interventions on the amplification of the contamination of the Oltrepò alluvial aquifer.

Findings from this study may be extended to other continental alluvial aquifers contaminated by deep saline water and sited in similar geological-geomorphological contexts and in which an alluvial plain is located near to or is bordered by mountain ranges. The results of this study have highlighted many aspects of continental saline water ascent that were not previously completely understood. The findings contribute to the current debate and have significant repercussions on the management of groundwaters in similar hydrological settings.