The study area is the mainland part of Shanghai in the flat alluvial plain of the Yangtze River Delta at the mouth of the Changjiang River with an area of 5,300 km² within 120°52′–122°12′E and 30°40′–31°53′N (Figure 1). Quaternary deposits are widely distributed in the area with a thickness of 250–350 m increasing W–E. The aquifers are comprised mainly sand and clay in loose Quaternary sediments and include aphreatic aquifers and five confined aquifers (Figure 2). This study focused on the latter.

Aquifers I–V were formed during the early to middle–late Pleistocene and their characteristics are shown in Table 1.

The groundwater system in the study area is not a stand-alone system and is part of the Yangtze River Delta system, where confined aquifers are the predominant type. In its natural state, groundwater flows mainly NW–SE. In general, the groundwater is characterized by weak hydrodynamic conditions and a low flow rate. Recharge is mainly through lateral inflow, although artificial recharge has become an important source in recent decades, and discharge is mainly from anthropogenic extraction.

For sampling, 87 wells were selected to provide a balanced regional distribution. Groundwater samples were collected in September 2018, and 19, 19, 19, 17, and 13 samples were collected from aquifers I–V, respectively (Figure 1).

Prior to filling, all sample bottles were rinsed three times with sample water. Samples for the cation analysis were collected in 500-mL HDPE bottles and then five drops of concentrated nitric acid were added; those for anion analysis were collected in 1000-mL glass bottles without preservatives. All samples were stored on ice in the field and transferred to 4°C storage until ready for analysis.

The concentrations of Na⁺, K⁺, Ca²⁺, Mg²⁺, Cl⁻, and SO₄ ²⁻ were determined by ion chromatography, HCO₃ ⁻ was determined by acid–base titration, and total dissolved solid (TDS) concentrations were determined using gravimetric analysis (drying at 105 °C). The analysis accuracy was assessed through the ion-balance error of ±5%.

The Durov diagram, which is used to study hydrochemical characteristics and facies (An et al., 2014; Gu et al., 2018), comprises two triangles, a central square, and two rectangles (Figure 5). The left and top triangles indicate the concentrations of cations and anions, respectively, the centrals square shows hydrochemical facies, and the right and bottom rectangles indicate TDS concentration and pH, respectively.

The Gibbs model is effective in elucidating hydrochemical processes (Gibbs, 1970; Gibbs, 1972). This model is based on groundwater hydrochemical processes being mainly controlled by water–rock interaction, evaporation, and precipitation (Farid et al., 2015; Wang et al., 2022). For this study area, the hydrochemical processes were adjusted to include rock–water interaction, marine-origin, and recharge (including lateral inflow and artificial recharge).

The saturation index (SI) is used to describe the solubility equilibrium of minerals in water (Rezaei et al., 2005). SI = log(IPA/Ksp), where IPA is the ionic activity product of dissolved mineral constituents and Ksp is the solubility product of the mineral. SI values of 0, >0, and <0 indicate saturated, supersaturated (precipitation may occur), and undersaturated (dissolution may occur) groundwater, respectively.

The ionic contents of the studied samples are shown in Table 2. The hydrochemical characteristics of aquifers I and II were generally similar, as were those of aquifers IV and V. The ion contents of aquifer III were the most variable with high standard deviation (SD) values, which indicated that the spatial distribution of groundwater is somewhat variable in terms of hydrochemical composition.

The pH values of groundwater were in the range of 6.99–8.36, 7.06–8.36, 6.92–9.42, 7.30–9.38, and 7.50–9.09 for aquifers I–V, respectively. Aquifers I–III were neutral to slightly alkaline, and aquifers IV–V were slightly alkaline with the pH increasing from aquifers I to V.

TDS concentrations had a range of 246–16522 mg L⁻¹ and decreased from aquifers I to V (Figure 3). All groundwater samples in aquifer I were saline water with TDS >5000 mg L⁻¹ in the west. In aquifer II, 68.4% of the samples were saline, with most in the southwest having TDS >5000 mg L⁻¹. There was a wide range of TDS concentrations in aquifer III, from 250 to 16522 mg L⁻¹, with fresh water predominating in the west and saline water in the east. In aquifers IV and V, most samples were fresh water. There were a few samples with TDS >2000 mg L⁻¹ in the area connecting aquifers III and IV.

The range of concentrations of the major ions are shown in Figure 4. Na⁺ was the dominant cation with an order of Na⁺>Ca²⁺>Mg²⁺>K⁺. Of the anions, Cl⁻ had the highest concentrations with clear dominance among all major ions in aquifers I–III. The concentrations of K⁺, Na⁺, Ca²⁺, Mg²⁺, and Cl⁻ generally decreased from aquifers I to III but had similar concentrations in aquifers IV and V. Most samples had low SO₄ ²⁻and HCO₃ ⁻concentrations from all five aquifers (Table 2).

The hydrochemical compositions and characteristics of the groundwater samples are indicated by the Durov diagram (Figure 5), where samples from aquifers I and II are concentrated on the Na⁺ and Cl⁻ sides and those from aquifers III–V are scattered in the upper half of the central square. This indicates that the predominant hydrochemical facies of aquifers I and II are Cl–Na, whereas those of aquifers III–V are variable.

The hydrochemical facies of groundwater can also be described in terms of TDS as follows: for TDS of <1000 mg L⁻¹, hydrochemical facies were variable and included HCO₃·Cl–Na, Cl·HCO₃–Na, HCO₃·Cl–Na·Mg, and Cl·HCO₃–Na·Mg, which were mainly distributed in aquifers II–V. For a TDS of 1,000–2000 mg L⁻¹, the hydrochemical facies were Cl–Na, Cl–Na·Ca, and Cl–Na·Mg, and these are distributed mainly in aquifers I, IV, and V. For TDS of 2000–5000 mg L⁻¹, the hydrochemical facies were Cl–Na·Ca and Cl–Na, which were mainly distributed in aquifers I and II. For TDS >5000 mg L⁻¹, the hydrochemical facies were Cl–Na and distributed mainly in aquifers I–III.

The Gibbs diagrams for all groundwater samples are shown in Figure 6. Most samples from aquifers I and II are plotted in the marine-origin domain, whereas those from aquifer III are plotted in both the marine-origin and rock–water-interaction domains, and those from aquifers IV and V are plotted in the rock–water-interaction domain. However, some samples from aquifers IV and V deviate from the Gibbs model (Figure 6A). Therefore, we speculate that other factors may affect the composition of groundwater, which makes Na⁺ rich and Ca²⁺ poor. A similar pattern of evolution applies to aquifers I–III as discussed in the following section.

Quaternary strata studies (Qiu and Li., 2007) indicate that there were seven marine transgressions in Shanghai during the Quaternary period with aquifers I–III formed during transgression periods. The presence of groundwater of marine–continental- and continental-origin in the study area was verified. The distribution of the former may account for most of aquifers I and II, and the eastern portion of aquifer III. The latter applies to the remaining aquifers, including small portions of aquifers II and III and most of aquifers IV and V. Further evidence is provided in the following section based on an ion-ratio analysis.

The Quaternary sediments in the study area are comprised sand and clay. X-ray diffraction and electron microscopy data for these sediments (Qiu and Li., 2007) indicate that the diagenetic minerals in the aquifers are predominantly silicate (quartz, feldspar, kaolinite, and illite) and carbonate (calcite and dolomite) rocks. Rock weathering and dissolution are naturally controlled by the hydrochemical composition of groundwater (Bau et al., 2004; Kim, 2010) through typical reactions as follows:

Silicate dissolution: 2 N a x K 1 − x A l S i 3 O 8 + 2 C O 2 + 11 H 2 O → 2 x N a + + 2 1 − x   K + + A l 2 S i 2 O 5 O H 4 + 4 H 4 S i O 4 + 2 H C O 3 − (1)

Carbonate dissolution: C a C O 3 + H 2 O + C O 2 → C a 2 + + 2 H C O 3 − (2) C a M g C O 3 2 + 2 H 2 O + 2 C O 2 → C a 2 + + M g 2 + + 4 H C O 3 − (3)

Halite dissolution: N a C l → N a + + C l − (4) K C l → K + + C l − (5)

Ancient seawater is an important source of Na⁺ and Cl⁻ in the groundwater of marine–continental-origin. The Na⁺/Cl⁻ equivalence ratio in seawater is 0.87 (Kunwar and Kawamura, 2014). A ratio of <1 indicates that the groundwater is affected mainly by seawater, whereas ratios of >1 indicate that silicate dissolution is more important (Eq. 1) and a ratio of 1 indicates that halite dissolution is the primary source of these ions (Eqs 4, 5; Gianguzza et al., 2004; Panno et al., 2006).

Bivariate plots for Na⁺ and Cl⁻ are shown in Figures 7A–B, which shows two groups of samples separated by the Cl⁻ = 7.5 meq L⁻¹ line (Figure 7B). Samples from aquifers I and II are distributed mainly below the 0.87:1 line with Cl⁻ >7.5 meq L⁻¹, and samples from aquifers IV and V are distributed mainly above the 1:1 line with Cl⁻ <7.5 meq L⁻¹. Therefore, the Cl⁻ = 7.5 meq L⁻¹ line acts as a demarcation index to identify groundwater origins in the study area. Some samples from the aquifer V plot near the 1:1 line with Cl⁻ >7.5 meq L⁻¹ is due to halite dissolution rather than seawater. Using the demarcation index of Cl⁻ = 7.5 meq L⁻¹, it was determined that the proportions of continental-origin groundwater samples in aquifers I–V were 0%, 26.3%, 52.6%, 70.6%, and 84.6%, respectively.

Carbonate dissolution and silicate weathering are the primary sources of Ca²⁺, Mg²⁺, and HCO₃ ⁻ as indicated by the (Mg²⁺+Ca²⁺)/HCO₃ ⁻ ratio (Kenoyer and Bowser, 1992; Kim, 2010): a ratio of <1 indicates that Ca²⁺ and Mg²⁺ were derived primarily from silicate weathering, and ratios of ≤1 indicate they were derived primarily from carbonate dissolution.

Most samples from aquifers I and II and the east part of aquifer III have (Mg²⁺+Ca²⁺)/HCO₃ ⁻ ratios of >1 (Figure 7C), which suggests carbonate dissolution as a source of Ca²⁺ and Mg²⁺. Most samples from aquifers IV and V and the west part of aquifer III have ratios of ≤1, which reflects the predominant contributions of silicate weathering and carbonate dissolution. Calcite and dolomite reach saturation or oversaturation in most samples (SI >−0.5; Figures 10C, D), which indicates that carbonate dissolution occurred throughout geological history. In summary, carbonate dissolution made the greatest contribution to marine–continental-origin groundwater, whereas both silicate weathering and carbonate dissolution contributed to continental-origin groundwater.

In the absence of anthropogenic sources, SO₄ ²⁻ is derived mainly from evaporite deposits, such as gypsum, through hydrochemical reactions as follows: C a S O 4 → C a 2 + + S O 4 2 − (6) M g S O 4 → M g 2 + + S O 4 2 − (7)

The SO₄ ²⁻ content of most samples was low (Figure 7F), and the correlations between (Mg²⁺+Ca²⁺−HCO₃ ⁻) and SO₄ ²⁻ were poor (Figure 7E). Furthermore, most samples were undersaturated in gypsum (SI <−2; Figure 10A). Confined aquifers in the study area have excellent sealing properties, low mobility, and long retention times, so evaporite deposits such as gypsum are at low levels and SO₄ ²⁻ is mainly derived from insoluble sulfides (Lang et al., 2011), although the high SO₄ ²⁻ contents of some samples are attributable to anthropogenic activity.

In terms of hydrochemical processes and ion sources, Ca²⁺, Mg², and Na⁺ deviated slightly from the standard model, which implied that the groundwater was affected by other factors. Cation exchange commonly influences the evolution of groundwater composition (Tournassat et al., 2009; Zheng et al., 2021) through typical reactions as follows: 2 N a + r o c k + C a 2 + w a t e r ⇌ 2 N a + w a t e r + C a 2 + r o c k (8) 2 N a + r o c k + M g 2 + w a t e r ⇌ 2 N a + w a t e r + M g 2 + r o c k (9)

The reactions on the right are positive cation exchange, and those to the left are reverse cation exchange. The relationship between (Mg²⁺+Ca²⁺−HCO₃ ⁻−SO₄ ²⁻) and (Na⁺+K⁺−Cl⁻) indicates the mechanism of cation exchange (Figure 8).

Changes in (Mg²⁺+Ca²⁺−HCO₃ ⁻−SO₄ ²⁻) and (Na⁺+K⁺−Cl⁻) for the samples were opposite with a correlation coefficient of 0.98, which indicated a strong negative correlation. In general, the anion changes (Cl⁻, HCO₃ ⁻, SO₄ ²⁻) in the groundwater were not significant, so this negative correlation mainly reflects (Ca²⁺, Mg²⁺) and (Na⁺, K⁺). A clear relationship between (Ca²⁺, Mg²⁺) and (Na⁺, K⁺) is thus noted with cation exchange playing a major role in controlling the hydrochemical compositions of groundwater.

The direction of cation exchange depends on the ionic adsorption energy and concentration. The energy decreases in the order of Ca²⁺>Mg²⁺>K⁺>Na⁺, so positive cation exchange is more common and leads to an increase in Na⁺ and K⁺ and a decrease in Ca²⁺ and Mg²⁺ concentrations in groundwater. However, 61% of the samples in the study area exhibited reverse cation exchange with negative (Na⁺+K⁺−Cl⁻) and positive (Mg²⁺+Ca²⁺−HCO₃ ⁻−SO₄ ²⁻) values.

Reverse cation exchange is more common in water with Cl⁻ content of >7.5 meq L⁻¹ (Figures 8C, D), whereas positive cation exchange is predominant with Cl⁻ <7.5 meq L⁻¹. This is consistent with the demarcation index for groundwater origins (Section 3.4). Reverse cation exchange is more significant in marine–continental-origin groundwater, and positive cation exchange is more significant in continental-origin groundwater with cation exchange explaining the deviation of some samples from the Gibbs model and why the Na⁺/Cl⁻ ratio is below the 0.87:1 line.

As a drainage area of the Yangtze River Delta, aquifers in the study area have received freshwater recharge throughout geological history through lateral inflow, which compresses saline water within a certain range and limits its expansion and forms stable boundaries. However, boundaries and hydrochemical characteristics have changed over the past century due to groundwater exploitation.

Annual groundwater use exceeded 200 million m³ in 1963, but it is currently decreasing each year. During the heavy-use periods, groundwater funnels develop due to excessive use, and saline water flows into the fresh water. Thus, continental-origin groundwater becomes marine–continental-origin groundwater (Figure 9). For example, saline water from the east flowed into a groundwater funnel to the northwest of Qingpu in aquifer III.

The expansion of saline water due to human activities over the past century is widespread and has serious implications for freshwater resources. Fortunately, Shanghai has adopted strict controls on groundwater exploitation with annual use limited to 2 million m³. Artificial recharge has also been implemented. During the 13th Five Year Plan period, the annual artificial recharge was ∼20 million m³ and through this the groundwater level rises each year. However, artificial recharge has caused other problems, notably, desalination and mineral re-dissolution.

Saturation indices were calculated with respect to gypsum, aragonite, calcite, and dolomite in groundwater and are plotted in Figure 10. It is generally considered that groundwater is saturated at SI values of −0.5 to +0.5 (Liu, 2019). Most samples were supersaturated or saturated in aragonite, calcite, and dolomite (SI >−0.5; Figures 10B–D), which is consistent with weak hydrodynamic conditions in the study area. However, a few samples had SI values of <−0.5, which is possibly because of the influx of fresh groundwater and artificial recharge and resulted in the breakdown of hydrochemical equilibrium and renewed mineral dissolution (Kanagaraj and Elango, 2019).