Keywords

1 Introduction

1.1 Water Problems in the Torrevieja Aquifer

The Torrevieja aquifer is located in the driest area of Europe where water resources are scarce and have a remarkable impact on society and economy and even political decisions are strongly linked to water management. As an example, the highest capacity desalination plant in Europe, with a yearly desalination capacity of 80 Mm3, has been constructed in the proximity of this aquifer in response to the water deficit in the entire region. The growth of the economic activities of the area has produced an increasing requirement for water and studies for the quantification of groundwater resources (García-Arostegui et al. 2003), the geometry of the aquifer (Mediavilla et al. 2007; Tabares-Ródenas et al. 2009) and the hydrogeological properties of the aquifer (Rodes and Rodes 2003). In previous hydrogeological studies and reports (IGME 1990; Ramos-González et al. 2002; Rodríguez Estrella 2003; CHS 2004), the Torrevieja aquifer has been usually defined as a confined aquifer of 147 km2 that outcrops in the northern border (16 km2), whereas the rest of the aquifer extension is covered with a confining unit composed of marls and silt (Mediavilla et al. 2007) (Fig. 1).

Fig. 1
figure 1

a Torrevieja aquifer location, water table contour map and geological units in surface. b Geological cross section with idealized conceptual model from NW to SE

Full size image

The annual pumping rates in the region are quantified as 5.50 Mm3 (Ramos González et al. 2002), whereas the recharge induced by the rain and the irrigation return flow only adds 1.16 Mm3 (Gómez-Fontalva and Calvache 2009) to the system. This indicates a clear deficit in the water budget that would not be sustainable for a long period given the size and the properties of the aquifer. The aquifer has been in this condition during the last 30 years, which shows that this imbalance should be reviewed. Based on this, two hypotheses are considered in this work: the impermeable properties of the top layer of the aquifer and the connection of the aquifer with a lower aquifer unit not very well-known in the study area. The hydrogeological information needed is scarce and the use of numerical models simulating groundwater flow is proposed as the tool for study the system complemented with hydrochemical data.

1.2 Methods

A set of 3D flow models was constructed using MODFLOW (McDonald and Harbaugh 1988), which solves the groundwater flow equation using modular finite differences. Only the Pliocene aquifer geometry and the upper unit on top were known and represented in the model. Initial steady state models were completed to decide which conceptual model would be more feasible. The calibration of the hydraulic conductivity and the transient simulations were accomplished after that. The model was designed on a regular grid of 304 rows by 370 columns at a spacing of 50 × 50 m, and two variable-thickness layers. In total, there were 22,4960 cells with their Y axis coinciding with the N-S direction. The top of layer 1 coincided with the topographic surface, and layer 2 represented the Pliocene aquifer. The morphology of both was irregular with changes in thickness and depth.

The hypothesis considered for the alternative conceptual models were:

  • Hypothesis 1: The upper Pliocene-Quaternary layer is not totally impermeable but permits certain inflows from irrigation return and precipitation. This option is based on the uncertainties about the hydraulic properties and the heterogeneity of these sediments.

  • Hypothesis 2: The extraction of water by the supply wells is partially assigned to a lower aquifer (Miocene) that decreases the outputs of the Pliocene aquifer. This possibility is proposed based on recent studies in stratigraphy (Soria et al. 2008; García-García et al. 2011) and the geometrical analysis of the lithological columns for assigning the layer from where the water is pumped to each well.

The hydrochemical dataset used was collected in several surveys from 2001 to 2003 (Rodes and Rodes 2003) by the Geological Survey of Spain (IGME). The sampling method consisted of the collection of 50 ml samples from purged wells or wells under exploitation. In many cases, these wells have long screens because their utility is to pump water for supply or irrigation, especially on the stretches with higher hydraulic conductivity, that result in samples being a mix of extensive sections of the aquifer. The information about the screen size and location was not always available. The water samples were analyzed for major ions (Ca2+, Mg2+, K+, Cl−, CO3 2−, SO42−, NO3 −) and for some minor ions (Br−, Fe). Additionally, some parameters were measured in situ, such as pH, electrical conductivity and bicarbonates content by titration with HCl. The hydrochemical composition was compared with the assumed origins (based on stratigraphical criteria) to determine if the hydrochemical differences corresponded to different hydrogeological units. A hierarchical analysis based on the average cluster considering the Euclidean distance as a similarity measurement was accomplished using the values of Cl−, SO4 2−, HCO3 −, NO3 − and electrical conductivity. These are the parameters with higher variability and could thus better differentiate the samples. Each sample was assigned one of the three categories considered geometrically: Pliocene aquifer, Miocene aquifer or surficial aquifer for wells less than 10 m deep.

2 Objectives

Develop alternative conceptual models of aquifer functioning based on the study of the lithological columns of boreholes that can explain the imbalance of the aquifer.

Test the new hypotheses by applying simplified groundwater numerical models to decipher the feasibility of each case.

Verify the results with the assessment of hydrochemical characteristics of groundwater based on the data collected in previous surveys.

Analyze the hydrogeological status of the Torrevieja aquifer using the perspective of the new conceptual model.

3 Results

Based on the numerical model of Hypothesis 1, the recharge proceeding from the semipermeable top layer (Plio-quaternary) should be considered minor because of the large impact that would have the recharge over the water table in the areas where there was no pumping (the water table would exceed the topographic surface). Even if this layer could store a certain amount of water it would not have a relevant impact on the global water balance. In contrast, the possibility that a percentage of the water extraction came from a lower aquifer unit (Miocene aquifer) seems to be more plausible as a conceptual model. The implementation of the numerical model for Hypothesis 2 showed that the depth of the water abstraction was a very sensitive parameter and that the match between water table observations and model results present much lower errors compared with Hypothesis 1 (from 20–30 m to less than 5 m). This conceptual model greatly depended on the classification of the wells (from which layer is the water pumped), which was also connected with the stratigraphy model. There are uncertainties in the location of the well screens and in the contribution of each layer in cases when the wells had a long screen that penetrated both aquifer layers.

Hypothesis 2 was considered more adequate and a transient model of the year 2002 with monthly time steps was constructed using a percentage of the pumping abstracted from the upper Pliocene aquifer (25% of total) and without leakage from the top Quaternary layer. The pumping rates and recharge were assigned monthly based on the information collected in previous surveys. The model was manually calibrated by trial and error modifying the hydraulic conductivity and the specific storage in the range of the hydrogeological materials that were observed. The hydraulic conductivity of the Pliocene aquifer varied from 0.1 to 2.5 m d−1. The homogeneity of the values was related to the spatial location of the observation wells, most of which were near the central area, which made the hydraulic conductivity closer to the borders more uncertain. The mean absolute error of the model for the 14 wells with monthly values was 1.56 m (Fig. 2).

Fig. 2
figure 2

Observations VS model results for the transient simulation of year 2002 including the modifications proposed based on the steady state models

Full size image

The differentiation between the Pliocene aquifer and the Miocene aquifer established for the pumping could also be tested by analysis of their hydrochemical properties. Following the same geometrical criteria established for the flow model, the samples were plotted based on the SO4 2− and Cl− values that could distinguish between saltwater from the sea or from sediments because of the presence of gypsum or other sulfates. There was a chemical differentiation between the two groups designated geometrically in a high proportion, but there were still samples that were not possible to differentiate. In general, the Miocene aquifer groundwater had a higher concentration of Cl− relative to SO4 2−, possibly related to older saltwater intrusion events, the presence of connate water during the sedimentation process or ongoing processes that have not yet been deeply studied. The hierarchical analysis showed that 20 of 25 samples for the Miocene aquifer had similar hydrochemical properties, and 33 of 42 for the Pliocene aquifer (Fig. 3). A third group was detected where the properties of the groundwater were mixed and it was not possible to differentiate them based on the chemical properties of water. These were interpreted as wells with mixed water because of the location of the screen at an intermediate position or bad isolation during the drilling process. These results highlight the differences between the aquifers and support the geometrical classification initially considered.

Fig. 3
figure 3

Dendrogram for hierarchical cluster analysis for the 126 hydrochemical samples collected in the study area. M: Miocene aquifer, P: Pliocene aquifer, S: Surficial aquifer based on geometric criteria. The differentiation between the different aquifers was established based on geometrical data

Full size image

4 Discussion

The Torrevieja aquifer presents an imbalance that can be explained based on numerical models and hydrochemistry analysis to the exploitation of different aquifers in the region. The increases of the outputs would lead to a compensation by seawater input assuming the connection of the Pliocene aquifer with the sea. The negative gradients farther inland indicate an ongoing saltwater encroachment process. Considering that the pumping regime in the area has been approximately constant during the last 30 years, the aquifer must be in a slow process of saltwater movement inland because of the tendency of decrease of the levels and the pumping cones in the water table surface with values beneath the sea level. The morphology of the aquifer as well as the hydraulic connection would determine if this process would reach the most sensitive areas for the population where most of the wells are located and how long it would take.

Climatic conditions play an essential role for the water resources in this region; the precipitation is very variable with annual oscillations from a maximum of 633–88 mm in a period of 39 years. This hinders the development of forecasts for the groundwater perspectives. Additionally, there are recharge inputs connected with anthropogenic sources, such as the irrigation return flows and the derivations from other catchments that can also be a major element of the water budget, especially during dry periods, although they are totally unpredictable. Other actions that can modify the water budget are, for example, the use of desalination plants for urban supply, the artificial recharge of treated water that can reduce pumping or the potential decrease in the pumping rate if the quality of the groundwater is reduced. This creates a very dynamic environment where it is challenging to establish strong guidelines for groundwater use and forecasting connected for example with climate change scenarios. Nonetheless, a trend of decreasing water resources can be detected in the current aquifer conditions both in the water balance of the model and in the water table elevation of the wells with longer series, even if the climatic conditions did not show a dryer tendency during the period 2002–2007.

The potential exploitation of the lower (Miocene) aquifer in the region can have a great impact for the aquifer management. In this study, it is assumed that most of the pumping wells were located in the lower aquifer, especially the ones with higher rates, whereas that ones with lower volumetric abstraction were pumping in the Pliocene aquifer. This highlights the relevance of this aquifer and opens new questions for further studies of recharge sources, reserves, the degree of exploitation and the connection with other nearby aquifers. Another possibility would be the merging between the Pliocene and Miocene aquifers into one bigger hydrogeological unit in the North and West sectors. In this case, the transfer between aquifers could act as a buffer when any of the aquifers has an excessive degree of exploitation or lack of recharge, although, at the same time, the contamination of one of them could directly affect both.

The differentiation between the Pliocene aquifer and the Miocene aquifer established for the pumping could also be tested by analysis of their hydrochemical properties. Following the same geometrical criteria established for the flow model, the samples were plotted based on the SO4 2− and Cl− values that could distinguish between saltwater from the sea or from sediments because of the presence of gypsum or other sulfates. There was a chemical differentiation between the two groups designated geometrically in a high proportion, but there were still samples that were not possible to differentiate. In general, the Miocene aquifer groundwater had a higher concentration of Cl− relative to SO4 2−, possibly related to older saltwater intrusion events, the presence of connate water during the sedimentation process or ongoing processes that have not yet been deeply studied. The hierarchical analysis showed that 20 of 25 samples for the Miocene aquifer had similar hydrochemical properties, and 33 of 42 for the Pliocene aquifer (Fig. 3). A third group was detected where the properties of the groundwater were mixed and it was not possible to differentiate them based on the chemical properties of water. These were interpreted as wells with mixed water because of the location of the screen at an intermediate position or bad isolation during the drilling process. These results highlight the differences between the aquifers and support the geometrical classification initially considered.

5 Conclusions

Two conceptual models were hypothesized to explain the imbalance in the water budget in the Torrevieja aquifer (South East Spain), a region with high demand for water and low resources. The assessment of the feasibility of each conceptual model was tested by the construction of groundwater numerical models and comparison with hydrochemical analyses of water samples. Based on this, it is proposed a conceptual model differentiating the pumping into two aquifer units: the one studied in this work and another underlying aquifer composed of Miocene sediments that would provide up to 75% of the water abstracted for human supply. The differentiation between the two aquifer units, Pliocene and Miocene, was supported also by the distinct hydrochemical signal of groundwater.