Characterization of Recharge Mechanisms and Sources of Groundwater Salinization in Ras Jbel Coastal Aquifer (Northeast Tunisia) Using Hydrogeochemical Tools, Environmental Isotopes, GIS, and Statistics

3.1. Water Sampling and Chemical Analysis

Ninety-four samples (including pumping wells and piezometers) were collected for geochemical analysis (major elements) and isotopes (δ²H, δ¹⁸O) during the wet and the dry season of March and July 2007 (Figure 5). Sampling locations were recorded using a potable GPS device. Prior to sampling, all wells were pumped for several minutes to eliminate the influence from stagnant water. Samples were collected in cleaned polyethylene bottles, tightly capped and stored at 4°C until analysis. Electrical conductivity (EC), salinity, and pH were measured in the field using a portable conductivity, salinity, and pH meter.

Chemical analyses were done using a Varian 730-ES ICP Optical Emission Spectrometer for cations and using ion chromatography (DX-120, Dionex, USA) for anions. $$ was measured by titration (Hach, USA). Samples for stable isotope analysis were collected according to the procedures described by Clark and Fritz [33]. Isotopic analyses were conducted in the isotopic laboratory of the department of Hydrology and Geo-environmental Sciences of the Faculty of Earth and Live Sciences of the Free University of Amsterdam. Isotopic ratios are expressed in per mil (δ) and oxygen and hydrogen isotope analyses were reported to d notation relative to Vienna Standard Mean Oceanic Water (V-SMOW), where d = [(Rs/RVSMOW) − 1] × 1000, Rs represents either the ¹⁸O/¹⁶O or the ²H/¹H ratio of the sample, and RSMOW is ¹⁸O/¹⁶O or the ²H/¹H ratio of the SMOW. Typical precisions are ±0.1 and ±1% for oxygen-18 and deuterium, respectively.

3.2. Statistical Analysis

The physicochemical parameters and chemical composition of the groundwater samples are presented in Table 1. All the acquired data were integrated into hydrogeochemical database in order to study the groundwater quality and to identify the groundwater salinization processes that contributed to the acquisition of the actual chemical composition.

Piper plots [34], considered as the common method for a multiple analyses on the same graph, are used to represent the different water samples and to distinguish graphically between different water types defined by the Stuyfzand classification (1993). Sample points with similar hydrochemistry tend to cluster together in the diagram [35].

Principal component analysis method (PCA) and correlations are a popular method to assess groundwater quality. One of the principle advantages of multivariate techniques such as principal component analysis (PCA) is that they are able to rapidly reveal relationships between a large number of variables. In this study, PCA and correlations are used to identify the possible sources of major ions in groundwater, hydrogeological reactions that may occur in the study area, and dominant factors that control groundwater quality.

Gibbs diagrams which are a simple plot of the TDS versus the weight ratio of Na⁺/(Na⁺ + Ca²⁺) or $$ are widely used to establish the relationships between the water composition and the lithological characteristics of the aquifer [36]. Three distinct fields, including precipitation dominance, evaporation dominance, and rock weathering dominance, constitute the segments in the Gibbs diagram.

4.1. Hydrogeochemical Characterization

Groundwater quality depends on various chemical constituents and their concentrations, which are mostly derived from the geological stratum of the particular region [6]. The pH was one of the primary indicators of the water chemistry evolution. The aquifer groundwater was neutral to slightly alkaline water, with a mean pH value of 7,23 and 7,15 in the wet and dry season, respectively. Electrical conductivity (EC) of the water samples was medium to high, suggestive of very highly mineralized waters. EC values ranged between 1240 and 6300 μS/cm in the wet season and between 1131 and 5430 μS/cm in the dry season. A problem to the water supply development in the area is the increasing electrical conductivity (EC) values of near-coastal shallow groundwater. The high EC of groundwater from wells located along the coast and in Bahirat Beni Ata can be explained by the seawater intrusion effect caused by the groundwater level drawdown due to overexploitation (Figures 6(a) and 6(b)).

Based on the conductivity values, the groundwater system could be classified into four groups: fresh water (<500 μS/cm), marginal water (500–1,500 μS/cm), brackish water (1,500–5,000 μS/cm), and saline water (>5,000 μS/cm). Based on our conductivity values it is evident that groundwater in Ras Jbel aquifer is in marginal and brackish waters. Few samples are of saline water type.

The study area is characterized by a wide range of salinities. Salinity values ranged between 500 and 3600 mg/l in the wet season and between 500 and 3500 mg/l in the dry season. The higher values of EC and salinity are indicators of higher ionic concentrations, probably due to the high anthropogenic activities in the region and geological weathering conditions but also due to the intrusion of sea water into the groundwater system.

The relative content of a cation or an anion is defined as the percentage of the relative amount of that ion to the total cations or anions, respectively [39]. In the study area, the strong acid anions (Cl⁻ and $$) exceed weak acid anions ($$ and $$). On the other hand, sodium and calcium concentrations exceed magnesium and potassium contents. The triangular diagram (Figure 7) shows that the groundwater chemistry was mainly characterized by two groups. The first group was Cl⁻–Na⁺ type, in which Na⁺ accounted for more than 52–65% of the total cations; the second group was Cl⁻–Na⁺/Ca²⁺ type, in which Na⁺ and Ca²⁺ accounted for 37–48% and 36–53% of the total cations, respectively. Chemical facies in the phreatic aquifer of Ras Jbel seem to be directly related to the configuration of the ante-quaternary substratum and the proximity of several sampled wells from the sea.

The Na–Cl-type indicated the presence of high chloride concentrations in the aquifer which may originate from the dissolution of halite, influx of sewage or waste water, and mainly intrusion of sea water [40]. In the downstream part of the plain near the sea and in Bahirat Beni Ata, where a marine intrusion was detected since 1980, the groundwater is mainly of Cl⁻–Na⁺ type (Figure 8). The distribution maps of Cl⁻ in both wet and dry seasons (Figures 9(a) and 9(b)) are well correlated with those of electrical conductivity and facies.

4.2. Processes Controlling Groundwater Salinization

Understanding the water salinization mechanism is the basis for regional salt management. Groundwater salinization is largely a function of the mineral composition of the aquifer through which it flows and hydrogeochemical processes such as mineral dissolution, precipitation, evaporation and transpiration, ion exchange, and the residence time along the flow path. It is also linked to various anthropogenic activities such as agriculture, overexploitation of groundwater resources, and sewage disposal.

4.2.1. Water-Rock Interaction and Origin of Groundwater Mineralization

Reactions between groundwater and aquifer minerals have a significant role on water quality. The p × p correlation matrix, revealing the existence of bivariate linear correlations between variables, allows a better understanding of the dominant water-rock interactions or source of the ions over the study area. Additionally, the use of multivariate statistics in hydr(geo)logical studies is a very common practice and numerous applications can be found in the literature ([10, 19, 41]) though most hydrologists consider that values larger than 0.5 indicate significant correlation. In the present study, PCA was carried out for 10 parameters (Ca, Mg, Na, K, HCO₃, SO₄, Cl, pH, TDS, and EC) and more than 90 observations (Tables 2 and 3). The first two factors F1 and F2 were always retained, explaining about 73% of the total variance (Table 2). Factors of a higher order generally explained the variance of a single parameter or established poorer and less significant correlations with two parameters [42].

Table 2. Principal component matrix.

	F1	F2	F3
CE	0,97	−0,03	0,05
TDS	0,98	−0,04	0,08
pH	−0,41	−0,42	0,75
Ca2+	0,82	0,17	−0,10
Mg2+	0,78	0,13	0,02
Na+	0,89	−0,20	0,17
K+	−0,11	−0,86	−0,16
Cl−	0,96	0,14	0,07
$$	0,77	−0,20	0,28
$$	0,43	−0,57	−0,49
Eigenvalue	5,86	1,40	0,95
% Explication	58,64	14	9,53
% Cumulative	58,64	72,64	82,17

• Bold values: loadings ≥ 0.5.

Table 3. Correlation matrix of dissolved species and the TDS for the study area in wet season.

	CE	TDS	pH	Ca2+	Mg2+	Na+	K+	Cl−	$$	$$
CE	1
TDS	0,98	1
pH	−0,34	−0,32	1
Ca2+	0,84	0,83	−0,44	1
Mg2+	0,73	0,73	−0,30	0,43	1
Na+	0,84	0,87	−0,18	0,60	0,65	1
K+	−0,04	−0,05	0,23	−0,11	−0,28	0,01	1
Cl−	0,95	0,96	−0,37	0,84	0,76	0,84	−0,19	1
$$	0,73	0,76	−0,12	0,50	0,52	0,75	−0,02	0,65	1
$$	0,36	0,37	−0,20	0,20	0,37	0,42	0,30	0,27	0,32	1

The correlations established between the TDS and concentrations of major elements (Table 3) show that the TDS is well correlated with the concentrations of chloride (r² = 0,96), sodium (r² = 0,87), calcium (r² = 0,83), magnesium (r² = 0,73), and sulphates (r² = 0,76). The high correlation of TDS with chloride, sodium, magnesium, sulphate, and calcium indicated that these elements are mostly contributed by mineralization. These ions have been dissolved into groundwater continuously and resulted in the rise of TDS. The contribution of carbonates and potassium is negligible (r² = 0,37 and r² = −0,05, resp.). The low correlation between TDS and pH suggests that the dissolution of the salts is not related to acidic conditions of groundwater but it is related to their degrees of solubility. $$ and pH apparently have little association with the other variables.

Bicarbonates are not correlated to calcium r(HCO₃Ca) = 0,20 indicating another source other than the calcite dissolution. However, considerable correlation coefficients between sodium and chlorides r(NaCl) = 0,84 and between calcium and sulphates r(CaSO₄) = 0,50 suggest halite and gypsum dissolution, respectively.

The first factor F1 accounts for 58% of the total variance, and it is contributed by the following variables: EC, TDS, Mg, Ca, Na, Cl, and SO₄. This factor is associated with the salinity component (NaCl salt source with Ca and SO₄ enrichment) and the cation exchange. The second factor F2 accounts for 14% of the total variance, and it is negatively determined by K and HCO₃ (Figure 10). It suggests carbonates weathering and pollution by fertilizer application.

In order to understand the origin of groundwater mineralization in Ras Jbel plain, the saturation index (SI) was calculated. The mineral facies are chosen based on the analysis result of groundwater quality, the main components of groundwater, and the occurrences conditions ([6, 43, 44]). In the study area, the main cations are Na⁺, Ca²⁺, and Mg²⁺ and the main anions are $$, $$, and Cl⁻; thus gypsum, anhydrite, calcite, dolomite, aragonite, and halite are chosen to be the mineral facies.

The positive values of the calculated SI with respect to calcite and dolomite for all groundwater samples (Figure 11) suggest their oversaturation in respect to these minerals (0,05 < SI_calcite < 1,32 and 0,08 < SI_calcite < 1,18 in the wet and dry seasons, resp., and −0,10 < SI_dolomite < 2,29 and 0,06 < SI_dolomite < 1,98 in the wet and dry seasons, resp.). As described by Appelo and Postma [45], the dissolution of calcite and dolomite is as follows:

•  

  Calcite:

  $$ (2)

•  

  Dolomite:

  $$ (3)

However, focusing on the scatter plots of bicarbonate versus calcium and calcium + magnesium versus bicarbonate we notice that groundwater samples are not plotted on the 1 : 1 straight lines of calcite and dolomite dissolution (Figures 12(a) and 12(b)). Groundwater samples show an excess of Ca²⁺ that can be explained by the gypsum dissolution.

The plot of SIGypsum and SIAnhydrite versus TDS exhibits a proportional and parabolic shape evolution with negative values of the saturation indexes (Figure 11) (−1,86 < SI_gypsum < −0,34 and −1,72 < SI_gypsum < −0,35 in the wet and dry seasons, resp., and −2,08 < SI_anhydrite < −0,56 and −1,94 < SI_anhydrite < −0,56 in the wet and dry seasons, resp.). Thus, both calcium and sulphate are derived from the same origin, which is the dissolution of gypsum and anhydrite.

$$ (4)

However, the plot of sulphate versus calcium (Figure 12(c)) shows an excess of Ca²⁺ ions for the majority of the groundwater samples. For these samples, the (Ca/(Ca + SO₄)) ionic ratio greater than 0,5 ratio (from 0,53 and 0,79) confirms the ionic exchange process [46]. The (Ca/(Ca + SO₄)) ionic ratio close to 0,5 confirms that the main source of Ca²⁺ is the gypsum dissolution [46].

A bivariate diagram of sodium versus chloride (Figure 12(d)) reveals two main groups: for the first group halite dissolution was maintained for a slope equal to unity where majority of the samples are situated on the 1 : 1 straight of halite dissolution given by the following reactions [45]:

$$ (5)

The second group includes the high-salinity samples (Na–Cl type), which do not follow the halite dissolution line and show enrichment in chloride compared to sodium. Thus, another phenomenon other than geological effect is controlling their salinization, and this may be the salt water intrusion.

Water samples were plotted in the Gibbs diagrams, which takes into account the major role of natural mechanisms (rock weathering, evaporation, and precipitation). Figure 13 clearly shows that the mechanism controlling water chemistry seems to be a combination of the weathering of carbonates minerals as well as the evaporation-precipitation processes. However, low rates of the groundwater samples were obtained in areas that were dominated by rock-water interactions.

Samples with Na⁺/(Na⁺ + Ca²⁺) or Cl⁻/(Cl⁻ + $$) ratios greater than 0,5 and TDS levels between 783 and 4323 mg/l showed that the groundwater chemistry was controlled mainly by the saline water mixing or evaporation. Evaporation results in increased TDS in relation to high ratios of dominant cations and anions and CaCO₃ precipitates by losing Ca²⁺ and $$.

4.2.2. Ionic Exchange Processes and Freshwater-Saline Water Mixing

Ion exchange is one of the important natural processes responsible for the concentration of ions in groundwater and has significant impact on the evolution of groundwater chemistry [18]. The dominance of salty groundwater dominated by sodium and chloride ions in Ras Jbel shallow aquifer provides evidence of mixing with an external salinity source, which could be the seawater from the coastal part of the aquifer. Cation exchange, responsible for the salinity signature, is described by two mixing mechanisms (freshening and saline water intrusion). Equations (6) and (7) show the gain or loss related to Na⁺ and (Ca²⁺ + Mg²⁺) within the exchanger X.

The freshening process or direct ion exchange: where Ca²⁺ from freshwater displaced the marine cations Na⁺ and Mg²⁺ from the exchanger complex. The resulting loss of Ca²⁺ from solution decreases the saturation state for calcite and possibly causes calcite dissolution.

$$ (6)

The intrusion of seawater or reverse ion exchange also triggered cation exchange reactions where Ca²⁺ was expelled from the exchanger by seawater Na⁺ and Mg²⁺. The released Ca²⁺ is being flushed from the aquifer by groundwater flow [45, 47].

$$ (7)

The plot of $$ versus (Na⁺–Cl⁻) determines the exchange of Na⁺ against (Ca²⁺ or Mg²⁺) through the clay matrix. The relationship [(Ca²⁺ + Mg²⁺)–($$)] is the gain or loss of (Ca²⁺ + Mg²⁺) due to the carbonates and gypsum dissolution. The relationship of (Na⁺–Cl⁻) determines the gain or loss of Na⁺ relative to the halite dissolution. If there is no ion exchange, all water samples will be placed in the origin of diagram [46].

Figure 14 shows that reverse ion exchange is a dominant process. To confirm the effect of reverse ion exchange, chloroalkaline index CAI-1 was calculated in milliequivalents per liter according to the relationship proposed by Schoeller [48]:

$$ (8)

If reverse ion exchange occurs in groundwater, CAI-1 values are positive. The calculated CAI-1 values are positive for more than 70% of the water samples which confirmed that reverse ion exchange is a dominant process. This shows that the interaction between the seawater and groundwater in the study area is playing a major role in the contamination of the aquifer by seawater intrusion. These results indicate that seawater/freshwater interface is in a continuous evolution despite the artificial recharge operations since 1993, and this probably because of the permanent heavy pumping.

Water samples concentrated in the origin of the plot $$ versus (Na⁺–Cl⁻) indicated the absence of the ion exchange process which can be attributed to the evaporation process followed by carbonate precipitation [49]. This result confirmed the results obtained using saturation states of minerals and Gibbs diagrams.

The seawater fraction in the groundwater is often estimated using chloride concentration [50]. Chloride ion has been considered as a conservative tracer not affected by ion exchange [51]. For conservative mass balance of the mixture, the equation used is as follows [45]:

$$ (9)

where Cl_mix is the Cl⁻ concentration of the sample, Cl_seawater is the Cl⁻ concentration of the Mediterranean Sea, and Cl_freshwater represents the Cl⁻ concentration of the fresh water. The fresh water sample will be chosen considering the lowest measured value of the electrical conductivity.

The rate of mixture varies from 0,32% (well n°23) in the north of the Raf Raf region and where the configuration of the ante-quaternary substratum prevents the marine intrusion to 13% (well n°1) near the shoreline (close to the coast). The highest value of the mixing fraction corresponds to the highest measured values of Cl⁻ and EC (1792 mg/l and 5430 μS/cm, resp.).

4.3. Isotopes and Groundwater Origin

The stable isotope ratios of oxygen and hydrogen in the groundwater are useful tools to differentiate between salinity origins [52, 53] and to help us understand various sources of recharge processes to groundwater because they are sensitive to physical processes such as atmospheric circulation, groundwater mixing, and evaporation ([33, 54]).

In arid and semiarid regions evaporation could be an important process influencing groundwater chemistry [19]. To understand the relationship between isotopic composition of groundwater of the shallow aquifer of Ras Jbel and those of precipitation measured at the station of Tunis Carthage situated at 50 km from the plain of Metline-Ras Jbel- Raf Raf, a bivariate diagram δ²H versus δ¹⁸O is plotted in Figure 15(a).

A local meteoric water line (LMWL) for Tunis Carthage was used to interpret the data in this study. The local meteoric water line (LMWL) is controlled by local hydrometeorological factors, including the origin of the vapor mass, reevaporation during rainfall, and the seasonality of precipitation [33]. The isotope composition of the precipitation was plotted along the LMWL using the following equation: δ²H (‰) = 8∗δ¹⁸O (‰) + 12,4 (which had a correlation coefficient R² = 0,99) [55, 56].

Figure 15(a) shows that the isotopic composition of most of the groundwater samples collected in the wet season (except for sampling site number 35) lies within a narrow range, confirming that these groundwater samples had the same recharge source. Furthermore, all groundwater samples are scattered around the LMWL indicating that the recharge of the Ras Jbel shallow aquifer originates from infiltration of recent precipitation from Mediterranean vapor masses. Based on their isotopic composition, two groups of groundwater samples were identified (Table 4). The first group is relatively depleted in isotopic values and includes samples with δ¹⁸O and δ²H values ranging from −6,03 to −5,37‰ versus V-SMOW and from −33,94 to −29,66‰ versus V-SMOW, respectively. This may be explained by the fact that nonevaporated water is rapidly infiltrated to the saturated zone.

As per the second group, values vary from −5,35 to −3,50‰ versus V-SMOW and from −29,37 to −22,15‰ versus V-SMOW, for δ¹⁸O and δ²H, respectively. The relatively enriched isotopic values of the group 2 samples demonstrate that this groundwater is affected by evaporated open water or soil water.

The groundwater samples collected in the dry season were enriched compared to those collected in the wet season. The results show isotopic content ranging from −5,74 to −0,06‰ versus V-SMOW for δ¹⁸O and from −31,46 to −10,31‰ versus V-SMOW for δ²H. This may be explained by heavy isotope enrichment in the groundwater caused by strong evaporation given that there was effectively no precipitation in the study area between the two sampling periods; meanwhile, the groundwater may have been mainly recharged by lateral inflow from outside the study area, resulting in seasonal fluctuations in the isotopic values. The enriched groundwater samples (n°15, 20, 25, 30, 32, 35, 38, 41, 45, and 48) might be an indicator for evaporation of the recharge water before its infiltration to the aquifer.

The isotopic data for the groundwater collected in the wet season were linearly fit using the regression equation δ²H (‰) = 4,01∗δ¹⁸O (‰) − 8,37 (R² = 0,82). This regression line can be interpreted as the groundwater evaporation line (GEL). The gel has a δ²H/δ¹⁸O slope <8, which reflects evaporation during or after rainfall and/or mixing with an external water source (e.g., return of irrigation water) with high δ¹⁸O and δ²H values. Furthermore, the GEL of the wet season intersected the LMWL at values of δ¹⁸O = −5,22‰ versus V-SMOW and δ²H = −29,37‰ versus V-SMOW. These values are estimated as baseline for δ²H and δ¹⁸O in recharging rainfall (Figures 15(b) and 15(c)). If samples are plotted above the lines, significant groundwater evaporation process can be confirmed.

Additionally, the isotopic data from the dry season were linearly fit using the regression equation δ²H (‰) = 3,09∗δ¹⁸O (‰) − 11,81 (with a correlation coefficient R² = 0,83). The GEL has a smaller slope than the LMWL, because evaporation tends to enrich heavy isotopes in water ([19, 57]). The increase in groundwater salinity due to evaporation can thus result in simultaneous increase in heavy isotopes ([19, 35]). The GEL of the dry season intersects the LMWL at values of δ¹⁸O = −4,93‰ versus V-SMOW and δ²H = −27,32‰ versus V-SMOW, which are chosen as baselines (Figures 15(d) and 15(e)). It is observed that 80% of the groundwater samples were plotted above the baselines and this demonstrates that evaporation has a significant contribution to groundwater salinity in the study area.

Furthermore, the deuterium excess calculated as d-excess = δ²H − 8δ¹⁸O [54] has been widely used in hydrological studies. The d-excess is used to identify secondary processes that influence the atmospheric vapor content in the evaporation–condensation cycle in nature ([54, 58]). The d-excess plotted against δ¹⁸O shows a negative correlation for the whole set of samples (Figure 16). The decrease in d-excess is an indication that evaporation has occurred during the recharge process which again confirms the previous results.

4.4. Irrigation Return Flow

Irrigation return flow is defined as the excess of irrigation water that is not evapotranspirated or evacuated by direct surface drainage and which returns to an aquifer or surface water [59, 60]. Irrigation return flows may induce salt and nitrate pollution of receiving water bodies [61]. Indeed, $$ is the most common water contaminant, and $$ pollution is increasing because the number of anthropogenic sources is increasing [26].

71% of groundwater samples are contaminated by nitrates where the concentration exceeds the permissible value of 50 mg/l set by WHO [62]. The spatial distribution of nitrates (Figure 17) shows that high nitrate contents are observed especially in the upstream of Ras Jbel aquifer. Groundwater contamination by nitrate is due to the intensive use of nitrogen fertilizers (Ca(NO₃)₂, KNO₃, and MgSO₄). In recent years, the agricultural land area in Ras Jbel plain has increased and copious amounts of nitrogenous fertilizer have been used, which have increased the groundwater $$ concentrations.

Furthermore, return flow from irrigation water also seems to contribute notably to the recharge process. Most of groundwater samples shows a correlation between NO₃ and δ¹⁸O, reflecting the significant role of evaporated and contaminated irrigation water to the groundwater salinization (Figure 18(a)). Huge quantities of irrigation return flow elevated groundwater level, hence increasing evaporation and inducing salinization [63]. As highlighted in Figure 18(b), the contamination by return of irrigation water is observed in the shallower horizons (depth ≤ 13 m).