Investigation of water quality origins (hydrochemical, mineralogical and isotopic) in the Jeffara of Medenine aquifer system (South-Eastern Tunisia)

Facing agriculture and urban water demand increases, managers make efforts to supply water from different sources. In arid regions, aquifers are the water sources mainly used for irrigation as well for drinking needs.

The geology complexity coupled with spatial-temporal hydrological variability and hydrodynamic flow uncertainties make it difficult to understand aquifers' water balances and the interactions within a hydrological system.

Survey studies were conducted in various arid regions to characterize aquifer systems for modeling, management and planning purposes.

Aquifer systems in arid regions are under the pressure of natural hydric deficit and seawater intrusion. Both processes, coupled with management options, are responsible for the system's sustainability.

A hydrochemical and isotopic groundwater study in the North Jeffara Aquifer, Gulf of Gabes (Southern Tunisia), was undertaken by Ben Hamouda et al. (2013). In this study, a combined hydrogeologic and isotopic investigation using several chemical and isotopic tracers, major ions, δ¹⁸O, δ²H and tritium of 29 samples was carried out in order to determine the sources of aquifer water recharge and the salinity origin. Results show that groundwater in the south of the study area represents a mixture of the Jeffara aquifer groundwater and locally infiltrated rain water. In the northern part, groundwater, which resembles that of the Sekhira aquifer, originates from locally infiltrated rain and runoff. The groundwater salinity is caused by dissolution of evaporite rocks (gypsum and halite minerals) in the aquifer system. The stable isotope data do not support the hypothesis of mixing with seawater.

In Tunisia, the north and center located towns are supplied with treated surface water transferred from the humid watershed located in the north. In the south, surface water is rare and groundwater constitutes the unique conventional water sources to supply irrigation and urban needs. The average annual hydric deficit between rain and evapotranspiration in the east part of South-Eastern Tunisia was estimated to be approximately 1700 mm in the period 1968–1995 (Yahyaoui et al. 2002).

Because of its location on the eastern foothills of the Matmata and Dahar ranges and its lithological nature, the Triassic sandstone at Sahel El Ababsa (TSE) represents the ‘water tower’ with the largest recharge from water runoff. The Jurassic limestones housing the Zeuss-Koutine (ZK) aquifer constitute a relay between the Triassic sandstones in the west and the Plio-Quaternary (PQ) and Miocene (M) formations in the east.

The PQ deposits represent the collapsed part of the Jeffara localized in the east of Dahar Mountain. It is an unconfined aquifer lodged in the Tertiary and Quaternary deposits and its thickness varies between 10 and 100 m, whereas the Miocene aquifer is confined to the entire coastal plain of Medenine. Its thickness is materialized by a hundred meters of Mio-Pliocene marls and clays. Known since the beginning of the century, this aquifer is captured by wells from 250 to 320 m in depth.

Two major fault families would be involved in this subdivision:

• A network of faults with a NW–SE direction composed of the Beni Ghzaiel, Metameur, Tejra, Grouz, Medenine, Koutine, Jorf, Sidi Maklouf and El Fje faults (Figure 1).

• A network of orthogonal faults to the last cited network oriented NE–SW composed of the Permian North faults, Kef El Anba, El Guattar and El Lehjer-Harboub faults.

Only two aquifers (TSE and ZK), which represent part of our study area, are largely solicited by the National Company of Drinking Water Supply (SONEDE) as well as industrialists and farmers (Yahyaoui et al. 2002).

Since the Tunisian revolution in 2011, the Jeffara aquifer system has experienced significant pumping increase due to the drilling of illegal wells, given the low depths of the water table, for the TSE, ZK and PQ, and the increase of the creation of wells for drinking water uses by the SONEDE company for the M aquifer.

The latest water balances have shown that SONEDE consumes 86% of the available resources of these aquifers. This situation has been aggravated by the excessive water pumping from illicit wells concentrated in areas known as irrigated perimeters, by water supply for the neighboring governorate of Tataouine and by the transfers to Zarzis and Jerba for desalination needs.

All these combined factors have led to decreases in the average annual piezometric levels (from 1965 to 2016) which were 0.40, 1.40, 0.76 and 0.58 m for the TSE, ZK, PQ and M aquifers, respectively (Sahal 2016). Previous studies on this aquifer system (Yahyaoui et al. 2002; Hamzaoui et al. 2012; Hadded et al. 2013) assume that there is a communication between the ZK and TSE aquifers and a direct infiltration of the rainwater at the level of the PQ aquifer and Triassic sandstone outcropping.

The main objectives of this research are to study the relations between the various aquifer levels mentioned above by integrating major and minor ion geochemistry with isotopic data.

Isotopic tools are used to identify the hydrochemical processes occurring between water and reservoir. They are also used to provide precise information about the interconnection between different aquifers that constitute the entire groundwater system of the Medenine region.

The study area is located in South-Eastern Tunisia, where the Jurassic Zeuss Koutine (ZK), the Triassic sandstone at Sahel El Ababsa (TSE), the Plio-Quaternary (PQ) and the Miocene (M) aquifers are housed. It perfectly characterizes the geographical area known as the Jeffara of Medenine.

The Jurassic ZK is limited in the north and north-east, respectively, by Zeuss Wadi and the Oum Zessar sebkhas, in the south by Ellaba Wadi and in the west by the Tejra, Tebaga, and Saikha reliefs.

The TSE is bordered on the east by Koutine plains and Koutine mountains, in the north by the Tebaga Mountains, in the west by the Dahar reliefs and in the south by El Khil Wadi.

The Plio-Quaternary and Miocene aquifers are limited in the west by the eastern limits of the above mentioned (TSE and ZK) aquifers. In the south, they are limited by Fessi Wadi and in the north and the east by Bou Ghrara Gulf (Figure 2).

The study area is characterized by an arid climate. The annual rainfall recorded at the Medenine station between 1994 and 2016 is estimated at 163.27 mm. At the seasonal scale, the winter is the wettest season with 36.96% (60.36 mm) of the annual rainfall. The autumn provides an average of 35.39% (57.79 mm), and the spring 25.26% (41.8 mm). Summer is characterized by an almost total drought with an average rainfall of 3.32 mm representing 2% of the average rainfall recorded at this station (Direction Générale des Ressources en Eau (DGRE) 2016).

The prevailing winds come from the NW. In winter, these winds are cold and dry while during summer winds are very hot (Sirocco). Winds from the east and NE are usually accompanied by irregular rainfalls, which may be very violent.

The combined effect of high temperature, irregular and torrential precipitation and dry winds promotes very high evapotranspiration, which can reach 20 times the values of precipitation in very dry years (Hamzaoui et al. 2012).

The Sahel El Ababsa groundwater was first recognized in the 1960s and delimited laterally by:

• The eastern foothills of the Dahar in the south-west and west where the sandy Triassic is the outcrop (Figure 3).

• The periclinal closure of the Jeffara dome to the north-west, which corresponds to outcrops formed by Permian reef limestones.

• Outcrops of Upper Jurassic and Cretaceous in the north-east (Hamzaoui et al. 2012).

This aquifer covers an area of 600 km² and encloses the watersheds of the Ennakim, Metameur, Eddouamiss and Et Taam wadis (Figure 1). It is lodged in the sandstones and dolomites of the Lower Triassic and recharged along the main wadis crossing the Sahel El Ababsa plain. This aquifer is unconfined at the foot of the Dahar and El Ouara (Raulin et al. 2011).

The Sahel El Ababsa is a multi-layered aquifer consisting of two levels:

• The upper sandy-dominated level is captured in the Medenine-Bir Lahmar, El Mzar region, Guelib Lemsene and in the El Ababsa North plain.

• The lower level, which is the most clayey and constitutes a thick cover, does not allow direct recharge of the water table.

The TSE is unconfined in the west part of the study area and it changes configuration to a confined aquifer in the east (Figure 4).

This aquifer covers an area of 537 km², in the west of the Tejra Mountains. The Jurassic ZK is considered as an unconfined aquifer because it is covered by a low PQ cover not exceeding 10 m (not shown in Figure 2). In the east of the Tejra relief, the Jurassic ZK aquifer becomes more and more confined from west to east, semi-confined near the Tejra relief and confined close to the coast under the effect of Miocene and Plio-Quaternary deposits.

It is housed mainly in the fissured limestones and dolomites of the Callovo-Oxfordian and Kimmeridgian age (Raulin et al. 2011) in the study area (Figure 4). In the northern and southern parts of the study area, it is lodged in the limestones of the Albo-Aptian to the Turonian and in the lower Senonian limestone, respectively (Raulin et al. 2011).

The Jurassic limestones and the Sahel El Ababsa sandstones are linked via the Tajera fault (Raulin et al. 2011). Towards the east, the aquifer deepens; it is located at depths that can reach 1,000 m (Figure 4).

The fissured Jurassic limestones of ZK play the role of a relay between the sandstones of the Triassic in the west and the Plio-Quaternary and Miocene formations in the east (Figure 4).

The reservoir architecture presented in the hydrogeological section in Figure 4 highlights the role of the faults. A series of NW–SE faults generate architecture in blocks tilted towards the east, and another series NE–SW favor architecture in horsts and grabens of the Triassic and the Jurassic series housing the TSE and ZK aquifers.

This is represented only by the marine Vindobonian composed of a marno-clay unit at the summit, an intermediate sandy unit whose thickness varies from 50 to 150 m, and a basal marl-clay unit (Vindobonian to Lower Miocene) with gypsum and sandy-gypsum intercalations with diffuse gypsum.

The vertical communication between the M and the PQ aquifers is ensured by the permeable sediment deposits of the Plio-Quaternary aquifer.

This aquifer forms a thick conglomerate deposit arranged in terraces on the banks of the wadis. These deposits result from the dismantling of the existing structures during the long phases of erosion and relief shaping. The groundwater recharge is manifested through direct infiltration of rainwater and runoff in the wadis.

To reach the objective of this research, the proposed methodology is based on hydrogeological, chemical and isotopic field survey, organized as follows.

A first piezometric campaign was carried out in September 2016 to record the piezometric levels of 27 water points tapping the Triassic Sahel El Ababsa, Jurassic Zeuss-Koutine, Miocene and Plio-Quaternary aquifers.

A complementary campaign was deemed necessary to allow the extension of the isopiestic curves on the potentiometric map and to have a clearer idea about communication between the different aquifers. This last campaign took place in October 2016 and concerned 16 additional points.

The two surveys were carried out very closely to make the potentiometric map as plausible as possible while recording the levels of the water table almost simultaneously throughout the study area.

Chemical analyses of the major elements (Table 1) concerning 57 samples distributed on the TSE, ZK, Miocene and Plio-Quaternary aquifers (24, 13, 5 and 15 samples, respectively) were carried out at the Laboratory of Applied Hydro-Sciences of the Higher Institute of Water Sciences and Techniques of Gabes (Tunisia) in December 2016 (Figure 5).

The temperature, pH and electrical conductivity (EC) were measured in situ using a sympHony SP80PC portable instrument. Accuracy was ±0.1 °C for temperature, ±0.05 for pH and 1% for EC.

Samples for major ion analyses were filtered (using 0.45 μm filters) and stored in polyethylene containers. Major elements were analyzed by professional chromatography, Metrohm IC 850 (r² = 0.999) in the Laboratory of the Higher Institute of Water Sciences and Techniques of Gabes (Tunisia).

The sum of the anions expressed in meq/l displayed values greater than 10 meq/l for all the samples representing the different aquifers, except for sample number 4 taken from the TSE with 9.77 meq/l.

The obtained error for the totality of samples varies between ±0.34 and 5.00%. The error obtained on the ionic balance of sample number 4 corresponds to 0.34%, which is in perfect concordance with Standard Methods for the Examination of Water and Wastewater 20th edition, published by American Public Health Association, American Water Works Association, Water Environment Federation (1999).

The statistical analyses of the physico-chemical data of the sampled waters were carried out by XLSTAT2018. In order to compare the importance of the correlations between the ions taken two by two, the Pearson coefficient of determination was calculated by this application. This coefficient was used to find relationships between two or more variables. The results are shown in Table 3.

Stable isotopes of ¹⁸O and ²H (Table 1) were determined by using the equilibration technique for oxygen and water reduction. This survey concerns 57 water points distributed on the ZK, TSE, Miocene and Plio-Quaternary aquifers (Figure 5). The analyses of these samples were performed at the Laboratory of the Higher Institute of Water Sciences and Techniques of Gabes (Tunisia) in January 2017.

Measurements were performed using a Finnigan Delta Plus mass spectrometer and AP 2003 automatic preparation system coupled with Isotope Ratio Mass Spectrometry (IRMS). This equipment is specially designed for measurement of the light environmental stable isotopes ²H, ¹³C, ¹⁵N, ¹⁸O, ³⁴S. This sensitive and selective instrument has been applied in hydrology, geology and environmental protection.

The IRMS is a universal tool for ¹³C/¹²C, ¹⁵N/¹⁴N, ¹⁸O/¹⁶O, ³⁴S/³²S, and ²H/¹H determination in dual inlet and continuous flow modes.

Prior to stable isotope ratio measurements, different samples must be brought into gaseous form. A host of preparation methods have been developed and improved to convert different sample compounds to an appropriate gas including H₂, CO, CO₂, N₂, and SO₂.

The average precision was ±0.1 ‰ for δ¹⁸O and ±1.0‰ for δ²H.

The compartments shown in the potentiometric map (October 2016) (Figure 6(a)–6(c)) coincide perfectly with the compartments delimited by the fault networks shown in Figures 1 and 2. In the Triassic aquifer, a main flow direction of west–east was identified going from the outcrop of the Dahar to the Jurassic ZK aquifer (Figure 6(a)).

However, there is a convergence towards the potentiometric contour of 100 m, which corresponds to an agricultural zone which strongly operates the TSE aquifer and which corresponds to the Megarine area. Moreover, the piezometric level of the ZK aquifer shows a general flow trend going from east to west with two zones of over-exploitation between potentiometric contours of 30 and 40 m.

There is a hydraulic continuity between the TSE and the ZK aquifers by the transit of the potentiometric contour of 120 m (downstream of the TSE aquifer) to the 110 m potentiometric contour (upstream of the ZK aquifer) (Figure 6(b)).

The number of boreholes collecting the Miocene aquifer is very limited in the north-east part of the study area and does not allow the establishment of a potentiometric map covering the entire study area. On the other hand, the Miocene is covered by the Plio-Quaternary deposits formed of permeable sediments allowing a vertical communication recharging the underlying Miocene aquifer. Consequently, a potentiometric map covering this aquifer system formed by the Miocene and Plio-Quaternary aquifer was established.

Several surface wells in the south-eastern part of the study area served as piezometers for the establishment of the Plio-Quaternary and Miocene potentiometric map.

The cited potentiometric map (Figure 6(c)) shows a general flow direction towards the sea. However, in the northern part, there is a hydraulic continuity of the ZK aquifer towards Boughrara Gulf. The flow direction in the southern part originates from Daher Mountains towards the Mediterranean Sea (Bhirett el Bibene).

The EC of the groundwater was very heterogeneous and varied from 817 μS/cm to 9,165 μS/cm. The lowest conductivity values were recorded in wells and boreholes tapping the Triassic sandstone in the foothills of the Dahar Mountain, which corresponds to the main recharge zone of the aquifer system in the study area. The highest values were measured in wells attributed to the Plio-Quaternary aquifer, suggesting a contamination by the brines of the depressions dotting South-Eastern Tunisia and the proximity of the coastal outlet.

Temperatures of water samples in the surface aquifer of the PQ range between 21 and 30 °C, in the same array as the ambient temperature. Their pH oscillated from 6 to 8, reflecting the dissolution of bicarbonates (Adams et al. 2001).

To characterize the groundwater facies of the Jeffara of Medenine aquifer system, the ‘Schoeler-Berkaloff diagram’ was used. It uses the major elements as cations (Na⁺, K⁺, Ca²⁺ and Mg²⁺) and as anions (Cl⁻, SO₄²⁻ and HCO₃⁻) measured in milliequivalents per liter. The chemical analysis results have been plotted on a Schoeler-Berkaloff diagram using the ‘Diagram’ software. The obtained results show the dominance of two water facies: the first in which (the Ca + Mg) takes precedence over the alkaline metals (Na + K), and the strong acids (Cl + SO₄) take precedence over the weak acids (HCO₃). The Ca-Mg-SO₄/Cl type group corresponds to waters collected from the most superficial layer of the Plio-Quaternary and from the upstream part of the Triassic and Zeuss Koutine sandstones (Figure 7).

The second group is the Na-Cl type and corresponds to samples collected in the Miocene sands (evaporite deposits of the Upper Miocene in the south of Tunisia, consequence of the Messinian crisis) and near the outlets of the sandstones of the Triassic and ZK. This evolution from sulphated facies to chlorinated facies is demonstrated by the Schoeler-Berkaloff (Figure 7).

The continuing decline in the piezometric head of the aquifer system, due to overexploitation in the Jeffara of Medenine to supply mainly potable water and greenhouse crops, has largely contributed to water quality deterioration through the formation of a crust, in most cases gypsum, mainly in old dry springs and the majority of irrigated perimeters.

The main sulphated salts associated with these crusts are gypsum (CaSO₄.2H₂O), anhydrite (CaSO₄), epsomite (MgSO₄7H₂O), burkeite (Na₂CO₃.2Na₂SO₄), thenardite (Na₂SO₄) and mirabilite (Na₂SO₄, 10H₂O). Halite (NaCl) and magnesium chloride (Mg Cl₂) are the main chloride salts (Kamel et al. 2006).

The dissolution of this crust by the return flow irrigation of pumped water from the sandstones of the TSE and ZK and by rainy events may be the principal cause of water salinity. Water saturation indices calculated for the aquifer system show that they are largely undersaturated with respect to halite, thenardite, mirabilite and to a lesser extent with respect to gypsum and anhydrite, indicating a possible dissolution of these salts (Table 4).

These possible dissolutions are confirmed by the strong correlations between Na/Cl, Na₂/SO₄, Mg/Cl₂ and Mg/SO₄ (Figure 8) and the saturation indices of the main salts with the minerals resulting from their dissolution (Figure 9). For gypsum and anhydrite dissolutions, the main sulphated salts release as much Ca as SO₄, whereas Figure 10(a) shows a calcium deficiency to the detriment of sulphates.

These same results are confirmed by the Pearson coefficient of determination (Table 3) obtained from the statistical analyses of the physico-chemical data of the sampled water.

The values of R² greater than 0.5 (shown in bold in Table 3) obtained by XLSTAT2018 reflect strong correlations between the ions taken two by two and are in clear agreement with those obtained by the binary correlations (Na/Cl, Na₂/SO₄, Mg/Cl₂ and Mg/SO₄).

This deficiency in the aquifer system can be interpreted by calcium precipitation according to two probable mechanisms: (i) the replacement of evaporites with carbonates generated by bacterial reduction of sulphates, a hypothesis based on ¹³C concentrations reinforcing the organic origin of carbonates (Henchiri & Slim-S'himi 2006); and (ii) de-dolomitization of calcite: since almost all the samples were saturated with calcite (Figure 9), dolomite (Table 4) and gypsum and anhydrite (Figure 9), the dissolution of the anhydrite continues and the concentrations of Ca²⁺ and SO₄²⁻ ions increase to reach super-saturation and consequently precipitation, which leads to decreasing water calcium concentrations.

Decreasing HCO₃⁻ concentrations, also resulting from calcite precipitation, can lead to a de-dolomitization process, adding Mg²⁺ to water and a drop in the molar ratio of Ca/Mg (Kamel et al. 2006).

On the other hand, cation exchanges can generate a calcium deficit to the detriment of sulphates. The evidence of these exchanges is highlighted by the correlation (Ca + Mg) – (HCO₃ + SO₄) as a function of (Na + K – Cl) (Garcia et al. 2001). In the absence of exchange, representative points should be placed close to the origin (McLean et al. 2000), which is not the case for the Medenine aquifer system (Figure 11), confirming cation exchanges probably on the surface of the clayey lenses, particularly abundant in the aquifer housed in the TSE.

A graphical representation of the groundwater of TSE, ZK, Miocene, and PQ in an ¹⁸O/²H diagram, in relation to the Global Meteoric Water Line (GMWL) and the Local Meteoric Water Lines of Sfax (Abid et al. 2011) and Nefta (Kamel 2011), shows the following:

• Stable isotope contents are very heterogeneous and spread over an approximate range of 4 δ¹⁸O (between −6.88 (sample No. 41) and −3.89 δ¹⁸O (sample No. 47)). Figure 12 shows three groups.

• The first group formed exclusively from samples collected in the PQ aquifer and aligned in a line whose slope is less than that of the GMWL and that of the local Meteoric Water Lines of Sfax and Nefta, in the zone of Water evaporation (Kamel 2011). The intersection of the line passing through the majority of the PQ samples with the local and global lines corresponds more or less to 0.5 δ¹⁸O, near the isotopic signal recorded in the Medenine region (about −6 ‰ δ¹⁸O).

• The second group is constituted exclusively of waters of the Miocene with an estimate of an isotopic signal of the original rains of their recharge evaluated to −7.3 δ¹⁸O.

• The third group consists of the different water sampled in the aquifer system of the Jeffara of Medenine with an estimate of an isotopic signal evaluated between −6.5 and −4.5 δ¹⁸O.

All samples of Group 1 located in the eastern part of the Medenine Fault (Figure 4) suggest a permeable roof and a wall with relatively impermeable clay predominance, which does not allow for the effect of vertical ascending drainage, which explains why the isotopic concentrations of these samples are similar to those of rainwater (Figure 12).

This is confirmed by the position of these same samples in Figure 13, highlighting the low depth of capture of these samples (less than 100 m depth) and displaying isotopic concentrations close to the local rainfall signal.

The second group corresponds to the Miocene samples, collected at a depth between 250 and 350 m (Figure 13) and corresponding to the waters that are the most depleted in stable isotopes and probably recharged by rains with a signal of −7.3 δ¹⁸O, and which would correspond to colder conditions than the current one (Figure 12). This group is the second member of the waters captured in the study area.

The third group is formed by a mixture of the waters of the different aquifer levels known in the region and closely related to the geological configuration of these aquifer levels; in fact, west of the study area ZK rests directly on the TSE, between the faults of Tejra Esghira and Medenine, and the PQ rests on ZK, itself based on the TSE (Figure 4). In this case, upward and downward vertical exchanges occur according to the pressures of the aquifer levels, communicating from the highest load to the lowest load.

Figures 12 and 13 show the clear separation of the three mentioned groups.

The ¹⁸O levels of four rainfall samples collected in Medenine in 2016 showed values from −5.5 to −6.0 δ¹⁸O‰ vs SMOW. This justifies the recharge of the TSE and ZK aquifer forming Group 1 in Figure 12. The Triassic sandstones and the intensely fissured limestones of the Jurassic of ZK would be favorable for an efficient and rapid infiltration without evaporation.

Therefore, the third group is recharged by more impoverished rains than those of the present one.

If we exclude the effect of altitude in the south of Tunisia (Kamel et al. 2006), this recharge would have occurred during the rainy and cold Quaternary periods with temperature differences of five degrees Celsius and 2.0 to 3.0 δ¹⁸O (Fedrigoni et al. 2001).

The distinction between dissolution and evaporation, which are the two main mechanisms of the mineralization of the Jeffara aquifer system, is highlighted by the TDS/¹⁸O relationship. Figure 14 shows two poles with a predominance of dissolved mineralization for Group 2, formed by Plio-Quaternary samples recharged mainly by infiltration through the unsaturated zone up to 30 meters. This allows a large evaporation, which leads to an enrichment that can exceed 2 δ¹⁸O.

The dilution leads to a position of samples quasi-parallel to the X-axis, as well as the representative samples of the Miocene (Group 3) and those of the Plio-Quaternary (Group 2).

The dissolution of gypsum (very abundant in the detrital fillings of the PQ) by sulphate release would be the main actor of Group 2 mineralization, and the dissolution of halite would be the main actor for Group 3.

The deposits of the evaporative formations during the upper Miocene, largely discussed by most sedimentologists of southern Tunisia and known as the Messenian crisis, would explain the chloride-sodium facies of Group 3 (Saghari 2012).

Group 1, formed by the TSE and ZK samples, shows a tendency towards evaporation for the former and towards dissolution for the latter.

The combined investigation in chemical and isotopic data in the Jeffara of Medenine aquifer system indicates that the dissolution of sulphated and chlorinated salts would contribute to the mineralization of the Jeffara aquifers, in perfect agreement with the good binary correlation of the major elements in the composition of these salts and the state of water's sub-saturation towards these minerals for most of the sampled points.

The isotopic study revealed that the aquifers hosted in the Triassic sandstones and in the Jurassic limestones are recharged by current precipitation without significant evaporation. The Plio-Quaternary aquifer would also be recharged by the same precipitation with an evaporation effect, evidenced by the position of its representative samples on the ¹⁸O/²H diagram.

The combination of the geological study with the new information from drilling and the contribution of isotopic geochemistry converge towards an unequivocal communication of the water bodies of the various aquifer levels via lateral hydrogeological relays in the subsurface and faults going deep in the Medenine region (communication between the two aquifers of ZK and TSE, that between the ZK and the PQ, and communication between the PQ and Miocene (M)).

Particular attention has been paid to understanding the geochemical and isotopic characteristics of the Miocene reservoir, which has different characteristics from those of other aquifers, presumably in relation to the upper Miocene evaporite deposits in southern Tunisia known as the Messenian crisis.