The Teboulba region is situated in the Monastir conurbation, some tens of kilometres to the southeast of the city (Fig. 1). This region has a semiarid to arid climate with mild, wet winters and dry, hot summers, a mean annual rainfall of the order of 375 mm and a mean annual temperature of about 20 °C. The studied sector of the region is part of the Sahelian watershed basin, built up on a low-lying coastal plain partly recovered from the sea. The monotonous topography and the geological cover, dominated by Quaternary formations, make it impossible to identify any hydrogeological basin structures. However, the depressions in the interior of the region and the wadis along the coastal zone help to identify the boundaries of more or less individual hydrological basins.

The study site is located on an alluvial plain whose geology is dominated by Quaternary deposits. The Teboulba area has relatively stable tectonics apparent in the tabular sedimentary structure. However, normal faults have clearly affected the structure of the deep layers. The Teboulba plateau, 10 km long and 4 km wide, plunges quite sharply towards the sea while a fairly gentle slope links it to the Moknine sebkha (a salt lake at the elevation of about 13 m below sea-level). It contains an alluvial water-table aquifer with lenticular geometry, short horizontal extent and irregular vertical continuity. There are also a few perched aquifers lodged inside the formation in silt-sand layers and lenses intercalated between clay and clay-sand strata. The substratum of the aquifer, consisting of an impervious Mio-Pliocene clay-marl layer, lies at a depth of over 90 m (Kraiem, 2002).

The Teboulba coastal aquifer is located in the sand-silt Plio-Quaternary formations, characterised by an effective porosity estimated at 7%, an average transmissivity of around 2 × 10⁻⁴ m²/s and a storage coefficient of the order of 3.5 × 10⁻². It is a vivid example of overexploitation. The flow in this aquifer takes several directions but converges generally on the exploitation zone (Fig. 2). The flow toward the sea and toward the Moknine sebkha, two natural outlets of the aquifer, is relatively weak due to the strong depression created by the heavy exploitation in the western part of the region. The hydraulic gradients are very variable depending on the lateral facies changes in the aquifer formation and the drawdown linked to the overexploitation. The aquifer is naturally recharged mainly by rainwater and, to a lesser degree, by domestic wastewater from infiltration pits and re-infiltration of some irrigation water. Discharge occurs mainly by pumping for agricultural purposes and into the sea and sebkha. Saltwater intrusion from the sea and/or from the sebkha may disturb the hydrodynamics of the aquifer. All these factors make the water balance of the aquifer extremely complex and its hydrodynamic equilibrium very fragile and, above all, highly variable.

The mean rainfall recharge to the aquifer was estimated with an infiltration coefficient of rainfall of 5% determined by Kamenski's method applied to triple wells aligned along a flow line and monitored every month (aquifer recharge calculated from the rise in its level after a rain event).

The Teboulba aquifer covers a surface area of around 35 km² and represents a hydrological recharge potential estimated, on average, at 0.65 Mm³/y with an exploited volume of 1 Mm³/y. This heavily negative budget (deficit of 0.35 Mm³/y) is compensated by reserves but, above all, by an inflow of saltwater from the sea and/or the sebkha and by return flow of irrigation water from the neighbouring regions, estimated at 0.53 Mm³/y (Beni-Akhy, 1998), only a part of which flows towards Teboulba. The overexploitation of the resources, begun hundreds of years ago and particularly intensified in the last decades, has caused a drawdown of the aquifer to an elevation below sea level (over −30 m at some points) in a zone that covers almost 12 km². This depression has provoked an intrusion into the aquifer layers of saltwater from the sea in the north and from the sebkha in the south and, as a consequence, the water quality has deteriorated to the extent that in some areas, it can no longer be used for irrigation (Amri, 1992). Luckily, because of the weak permeability of the water-bearing layers, the saltwater intrusion has not progressed very far. The aquifer is exploited mainly by 806 surface wells, 465 of which are equipped with motor pumps enabling them to irrigate, by sprinklers, a surface area of about 200 ha. Note that this irrigation method reduces the water losses to a minimum.

Furthermore, this aquifer is characterised by water with a fairly heavy load of salt. Its salinity may reach 4.5 g/l. The highest values have been recorded near the sea and the sebkha whereas the lowest (1.5 g/l) are found toward the centre of the aquifer (Fig. 3). The planned remedial action is to bring surface water from the overspill of the Nebhana dam and inject it into a few of the wells among the thousand existing ones.

Aside from all recharge operations, the piezometric level of the Teboulba aquifer shows a cycle of five phases over one year linked to the mode of cultivation at Teboulba (Amri, 1992; Arrondissement des Ressources en Eau de Monastir, 1973; Besbes, 1971). This piezometric cycle shows that the best times for measuring the water level is the beginning of March, for the wet period, and the end of July for the dry one. Moreover, as the irrigation network is used to recharge the Teboulba aquifer, the best time for the recharge is from November–December to March, corresponding to the first phase of the stationary piezometric cycle, since this period corresponds to an interruption in the irrigation of the end-of-season crops.

The planned experiment was conceived not only as a means of combating the saltwater intrusion, but especially as a potential storage in the aquifer depression of part of the excess water from the Nebhana dam. This dam, situated 47 km north-west of Kairouan, was built in 1966–1968 and started to operate in 1972. It was expected that there would be some mixing of the freshwater in the dam (0.95 g/l of salt) with the saltier groundwater at Teboulba, thus increasing the total volume of freshwater available for agriculture (Arrondissement des Ressources en Eau de Monastir, 1973).

The history of this operation goes back to the 1970s. It concerns the recharge of the aquifer via a connection to the main pipe of the irrigation network (Besbes, 1971; Maalel, 1998; Rekaya and Amri, 1992). As there was no excess water in the Nebhana dam because of the rainfall deficit during the period from 1978 to 1991, the recharge operation was interrupted at the end of 1978 and restarted in July 1992, lasting until 2000 (Fig. 4).

Preliminary experiments were carried out in 1971 for 128 days during which dam water was injected into four surface wells (diameter 1.2 to 1.7 m). The water used was initially not treated and cascaded freely into the wells from the soil surface, becoming aerated and thereby clogging the flow paths. For each well, the flow rate was measured by a meter and controlled by a valve; the injected volume was 186,000 m³ (i.e. close to 4 l/s per well). The aim of this operation was first to establish an order of magnitude for the infiltration capacity of each well. On the basis of these first results, and taking into account the hydrodynamic characteristics specific to the aquifer, given above, a simple mathematical finite-difference model was constructed to interpret the first 128-day experiment, in order to estimate by extrapolation the volume that would be necessary to replenish the aquifer. During this first period, the exploitation amounted to 78,000 m³, only 11,000 m³ of which were taken into account in the modelled zone because of the distance between the injection zone and the withdrawal points. In order to bring the piezometric level back up to sea-level, it was found necessary to inject a volume of around 7 Mm³. In case such a volume was not available, another scenario proposed, injecting the water into some 15 wells for five years, in the first year at a rate of 0.5 Mm³ during four months, then 0.25 Mm³ every following year to reach a maximum piezometric recovery of 25 m. The depression would not be entirely filled, but this method would prolong the life of the aquifer by around 40 years by decreasing the hydraulic gradients that caused the saltwater intrusion and diluting the aquifer water by less salty water, if the withdrawal rate remained at 1.1 Mm³/y. This simulation also showed that, for such a withdrawal rate, the aquifer would be permanently protected if it were artificially recharged by, on average, 0.5 Mm³/y and if 20 to 25 wells were equipped, assuming an annual rainfall of 375 mm with an infiltration coefficient of 5% (Arrondissement des Ressources en Eau de Monastir, 1973).

During the 1972–1978 period, a total volume of around 1.52 Mm³ was injected (i.e. on average 220,000 m³/y). Three-inch injection pipes penetrating the wells below the water level were used. At the same time, filters were fixed to the injection devices in order to reduce to a minimum the clogging by suspended matter contained in the dam water. Despite these precautionary measures, a clear decrease was noted in the absorbed flow rates and it seemed doubtful that it would be possible to inject more than 3 or 4 l/s into each well over the long term unless the water went through a complete treatment process. Moreover, the farmers were always reluctant to allow access to their wells for the recharge. In order to solve this problem, plots of land were acquired where injections could be made into deep trenches, but this technique was never implemented and the injection continued in wells.

Despite the considerable difference between the initial model predictions and the water volume really injected into the Teboulba aquifer, the results were quite noticeable. The piezometric depression cone disappeared to a great extent (Rekaya and Amri, 1992). Whereas the model predicted only a piezometric recovery to an elevation between −10 and −15 m, it actually reached −5 m and even, at some points, zero, despite the fact that the injected volume was less than a quarter of that used in the model. This made it possible to bring back into production a number of wells that had been dry for several years (Amri, 1992).

During the 1992–2000 phase of artificial recharge, a total volume of around 1.76 Mm³ of water was injected (i.e. on average 196,000 m³/y), always through wells.

In order to follow the variations in time and space of the aquifer piezometric level and salinity and to define the effect of the artificial recharge, series of measurements were made of the piezometric level of the aquifer as well as samplings of the water for physical and chemical analyses which made it possible to make rough draughts of piezometric and salinity maps.

4.1 Hydrodynamic impact

4.1.1 First recharge phase (1972–1978)

After injection of around 1.52 Mm³ of water (average 220,000 m³/y), the piezometric map of 1978 (Fig. 5) clearly shows the disappearance of the piezometric depression, revealed in 1971. Moreover, it shows the emergence of a hydrodynamic divergence zone to the west of the town of Teboulba but with a partial recovery of the hydrodynamic equilibrium of the aquifer in the recharge zone where the piezometric level exceeds 5 m. Note that the hydraulic gradient between wells 369 and 546 has increased. This may be explained by the high recharge rate (10 l/s for each injection well) and the weak permeability of the aquifer layer. This situation lasted until 1980 when the injection site still appeared as a high piezometric level zone in the aquifer albeit with a downstream propagation of the recharge wave (Amri, 1992).

The evolution of the piezometric level between 1971 and 1978 (Fig. 6) shows a general recovery that may exceed 30 m in the zone adjacent to the recharge site. The monitoring of the depth of the water-table shows a very significant increase since the beginning of the recharge. The response of the aquifer is expressed by an attenuation of the injection pulse and a regularisation of the effective water quantity that actually reaches the aquifer.

The surface wells situated a few hundred meters from the irrigated perimeter showed only weak recovery. In the most exploited zones and those furthest away from the recharge zone, particularly in the western part of the region, a relatively large decrease of the aquifer level was recorded. An analysis of the map of the recovery during the first recharge phase led to an estimate of the water volume stored in the aquifer porosity of around 4 Mm³. Although a slight increase in exploitation was noticed because of the quality improvement due to the artificial recharge, this volume is much greater than that injected during the same period (1.52 Mm³), meaning that a large fraction of the aquifer water comes from other sources, e.g. pre-artificial recharge water, re-infiltration of the irrigation return water and domestic wastewater discharged in infiltration pits.

The cessation of artificial recharge after 1978 has caused a significant lowering of the aquifer piezometric level. The 1988 piezometric map (Fig. 7) reveals a depression to the west of the town of Teboulba which reaches 20 m below sea level. The hydraulic gradient is on the order of 16%, which demonstrates the weak permeability of the formation. This situation lasted until the beginning of 1992 and caused a general draw-down of the aquifer, in some places exceeding 10 m (Fig. 8).

4.1.2 Second recharge phase (1992–2000)

During the second phase of artificial recharge, a total volume of around 1.76 Mm³ of water was injected (average 196,000 m³/y). Although the available volume of water in the dam varies from year to year, for every well, attempts were made to keep the injected volume at a fixed percentage of the annual natural recharge at the site. The second phase, begun with an injection of 120,000 m³, revealed some improvement in the reserves stored in the aquifer compared to 1988. The situation in 1992 (Fig. 9) showed some piezometric levels reaching −11 m (versus −22 m in 1988).

Due to the great rainfall deficit in 1994 and to the negligible replenishment of the Nebhana dam that year, the recharge in 1994 was reduced to 153,000 m³. For this reason, the exploitation wells were strongly affected and recorded a fall in the piezometric levels of up to 8 m compared to the situation in 1992, creating a new depression further west of the town of Teboulba (not shown). In fact, a concentric and convergent flow was detected in the direction of this piezometric depression. However, the injection zone had a fairly normal piezometric level with a general flow toward the sea in the north and toward the sebkha in the south. West of Teboulba, there is still a noticeable depression around wells No. 190, 196 and 456 which may be explained either by a very high withdrawal rate at this point or by a lack of influence of the artificial recharge in the zone.

In 1995 and 1996, the recharge was weak although water was available in the dam. Indeed, work being carried out on the old above-ground irrigation network to replace it by a buried one had hampered the recharge operations. The total injected volume was reduced to 67,000 and 97,000 m³, respectively. The mean absorption rate per well was low, estimated at 1.63 l/s. There was no artificial recharge in 1997 but the piezometric levels in the aquifer did not show any great decrease. There was a surplus of rainfall in the autumn of 1997 which contributed to the natural recharge of the aquifer and to fairly stable piezometric levels. The piezometric map for the year 1997 (not shown) shows the emergence of a flow toward the sebkha to the south. The depression is still present to the west of Teboulba. A few, highly localised, piezometric domes appear in the north-east of the urban area, in the far north-west (well No. 905) and in the vicinity of well No. 531 where the injection wells are concentrated.

The artificial recharge could not be continued until April 1998, when the new irrigation network had been installed. Thus, with 19 injection wells and an overall instantaneous flow rate of 23 l/s, it was possible to increase the recharge rate of the aquifer. After intense artificial recharge from 1998 to 2000, an analysis of the piezometric situation of the aquifer, in 2000 (Fig. 10), further confirmed the existence of the localised domes, revealed in 1997, but definitely enhanced. The flow direction of the aquifer remained constant but still retained the depression in the western part of the aquifer, where the piezometric level reached –17 at well No. 190. Furthermore, the map shows a narrowing of the depression cone compared to earlier situations. This may be explained by the increase in the water volume injected into the wells in the neighbouring zones which exceeded 297,000 m³ in 2000. An analysis comparing the situations in 1992 and 2000 revealed that the water level had risen, sometimes by more than 13 m underneath the recharge site (Fig. 11), and that the aquifer had received around 1 Mm³. The difference between the injected volume and that stored in the aquifer porosity (0.76 Mm³) can be explained by the increase in the withdrawal rate which may have been as high as 1.5 Mm³/y instead of 1.1 Mm³/y. Note that the Teboulba aquifer is also recharged by infiltration return flow of treated wastewater used for irrigation. These irrigation volumes were 144,000 m³ in 1992, 114,000 in 1993 and 255,000 in 1994 (Beni-Akhy, 1998).

Although the depression persisted in the western part of the Teboulba aquifer, the piezometric map for 2001 (not shown) shows a slight deepening of the depression cone compared to the situation in 2000. Moreover, the map reveals that the local piezometric domes mentioned in 2000 have disappeared. The reason is that there had been no recharge to the aquifer since the summer of 2000, which had lowered the water level by 1.2 m in the Teboulba region.

4.2 Hydrochemical impact

4.2.2 Situation after the artificial recharge

During the whole period from 1972 to 1993, a clear zonal distribution of the salinity was observed in the aquifer together with a persisting zone of weakly concentrated water (TDS lower than 1.5 g/l) localised on the coastal zone. Around this zone, the salinity increased sharply and exceeded 4.5 g/l in the vicinity of the sea and the sebkha (Amri, 1992). Because of a lack of available salinity data from the period of the first recharge phase, this study only examines the evolution of the aquifer water quality between 1994 and 2001. The salinity maps for the different states, based on electric conductivity data, clearly confirm the zonal distribution and the increasing salinity in the direction of the sea and the sebkha. The situation in 1994 (Fig. 12) shows quite low salinity values (lower than 2 g/l) in the western part of the aquifer, the zone of artificial recharge, whereas the highest ones are found especially in the coastal zones and at the boundaries of the sebkha where they exceeded 3 g/l.

Although the Teboulba aquifer was not recharged in 1997, the salinity variations of its water, recorded in the control wells, were not very significant, due to the great surplus of rainfall that year which contributed to the natural recharge of the aquifer, thus maintaining both the piezometric levels and the water quality. The salinity map for the year 2000 (Fig. 13) shows almost the same general pattern as that for 1994 with a slight salinity reduction toward the west of the aquifer: the zone characterised by a salinity lower than 2 g/l and situated near the recharge site becomes more extensive. This shows the influence of the mixing with less salty water from the Nebhana dam.

In 2001 there was no artificial recharge, which resulted in a degradation of the groundwater quality in the region. The 2001 salinity map (not shown) shows a shrinking of the less-than-2 g/l salinity zone and reveals saline intrusions especially in the northern and southern parts of the aquifer where the salinity exceeds 4 g/l. It is important to note that the salinity increase at the aquifer boundaries and the existence of the piezometric depression west of the town of Teboulba are directly related to the heavy exploitation, on the one hand, and to the very low volume of recharge water available that year, on the other. The continuity of the aquifer flow toward the depression and the disappearance of the local piezometric domes clearly confirm that the aquifer needs continuous recharge every year, if its status is to be maintained.

4.3 Social and economic impacts

Artificial recharge of coastal aquifers, which are especially overexploited, can offer an efficient means of combating seawater intrusion and thus of preventing an inevitable degradation of the water quality which might prove irreversible. It has produced good results in recent decades and seems to have a promising future in arid and semiarid regions given that it is considered to be a conservation measure as well as a means of developing resources. In a country such as Tunisia, which suffers from lack of water and a capricious climate, artificial recharge by surface water would be very valuable for the aquifers because of the availability of the water and its remarkable annual regularity of occurrence.

Artificial recharge by dam water represents a valuable tool for managing water resources. The installations needed for the recharge operations are simple and relatively inexpensive. Although there are other methods of artificial recharge, the recharge by injection directly into the aquifer is the most efficient one for creating freshwater barriers against saline intrusion. However, clogging of the wells, which was quite random, and the doubts of the farmers concerning this technique might jeopardise the success of the recharge operation. Tunisia has acquired an extensive experience in the field of recharge. However, until now the recharge has remained low and discontinuous in time and space.

Now the aim is to promote an active research and experimentation programme to evaluate the additional resources and, in particular, to achieve an integrated and participative management of the water resources, including partnerships with local communities and the private sector, and to convince the users of the necessity to save water and protect its quality. Note that the storage of excess water in wet years would be extremely valuable in particularly dry years and would lead to a beneficial joint management of groundwater and surface water. Only a determined acceleration of the recharge operation could help to avoid the predicted water crises by guaranteeing supply without putting increased pressure on already degraded resources. The social, economical and environmental impacts of such an action would be considerable.