Climate change and anthropogenic activities are progressively inducing dramatic changes in hydrologic processes and large-scale changes in water availability (Alcamo and Henrichs, 2002; Becker et al., 2010; Ceola et al., 2016; Wang et al., 2020). These changes affect key aspects of the water column (including surface water, vadose zone, and groundwater), surface water-groundwater interactions, and water quality (Hamdi et al., 2018; Jerbi, 2018; Tang et al., 2017; Zhu et al., 2020). In mid-latitude arid and semi-arid areas, where surface water is only available for short periods, groundwater generally represents a long-term water supply; even during severe droughts when river flows are significantly reduced (Guezgouz et al., 2017; Loaiciga and Leipnik, 1996; Lin et al., 2015; Zambrano et al., 2018). Groundwater demand is increasing profoundly with population growth, while global changes are putting additional stress on water resources and increasing the likelihood of groundwater scarcity. Declining groundwater levels are by far the most alarming indicator of aquifer overexploitation and depletion of groundwater storage. This decline can reach tens of meters per year in the most anthropized Mediterranean areas such as the Crevillente aquifer in Spain (e.g. Garrido et al., 2005) and a few meters per year in other regions such as the Kairouan plain in central Tunisia (e.g. Jerbi et al., 2018) or the Saïss plain in Morocco (e.g. Ameur, 2017). In the Nefza aquifer considered in this paper, the groundwater decline is about 1 m/year according to the official exploitation report (DGRE, 2018). In this case, examining how groundwater levels are declining in response to climatic and anthropogenic effects is essential for sustainable water management.
Groundwater availability is a function of storativity, the volume of water released (or brought into) storage per unit area of the aquifer per unit decrease (or increase) in the hydraulic head (Freeze and Cherry, 1979). While easily defined in words, groundwater storage is much more difficult to quantify in practice. Understanding the discontinuous and heterogeneous occurrence and pattern of hydraulic parameters is essential (Rane and Jayaraj, 2020). Spatial variation in hydraulic parameters is controlled by lateral and vertical variation in hydrostratigraphic units. For example, evaluation of aquifer geometry is mandatory for estimating groundwater storage. However, developing a realistic aquifer geometry has never been an easy task for groundwater systems of detrital origin due to their inherent structural heterogeneity (Galloway, 1977; Pham and Tsai, 2017). In these types of aquifers, the most important factor influence groundwater flow patterns is the heterogeneity of lithological composition and complexity of geodynamics. A realistic hydrostratigraphy can be developed using a large amount of well log data to capture the complexity of an aquifer system. Logging data can provide excellent information on vertical variability of the sediments but only limited information on lateral variability. Surface geophysics, such as electrical resistivity surveys or electromagnetic methods, can be a quick and easy way to complete the data set when no logging data are available. Vertical electrical sounding (VES), a one-dimensional array method, is particularly useful when correlated with existing hydro-stratigraphic information. However, VES can be subject to ambiguity problems because low resistivity can be due to various factors such as groundwater chemistry and formation materials (Song et al., 2007). Flathe (1955) critically reviewed the applicability and limitations of the vertical electrical sounding method for solving hydrogeological problems. To construct a reliable and realistic 3D hydrostratigraphic model, all types and sources of geologic and geophysical data must be used.
To address the gaps between local information and model hydrostratigraphy for different scales of heterogeneity using various input datasets (e.g., well logs, geophysical data, seismic data, and pumping tests), a range of specialized geospatial algorithms have been used worldwide. These include methods widely used in hydrogeology such as two-point variogram statistics, such as indicator geostatistics (Proce et al., 2004; Tartakovsky, 2013); simple kriging and co-kriging methods (Aboufirassi and Mariño, 1984; Allard et al., 2012; Vo Thanh et al., 2019), indicator geostatistics based on transition probabilities (Carle and Fogg, 1996; Lee and Kitanidis, 2014; Zhao and Illman, 2018), Markov chain method (Fogg et al., 1998; Xiao et al., 2016), Monte Carlo simulations (Delhomme and Delfiner, 1973; Pakyuz-Charrier et al., 2018; RamaRao et al., 1995) and multi-point simulation (MPS) (Bai and Tahmasebi, 2020; dell’Arciprete et al., 2011; Journel, 2005; Strebelle, 2002). Reviews of these methods can be found in de Marsilly et al. (2005). In his research, de Marsilly et al. (2005) recommend that future geologic modeling work focuses on improving facies models, comparing them, and designing new in-situ testing procedures that would help identify facies geometry and properties.
The aquifer of the Nefza dunes, in the North of Tunisia, is one of those places where the storage capacity of groundwater must be studied taking into account the complex spatial distribution of hydraulic properties. Furthermore, the aquifer is only exploited by the national company of drinking water supply (SONEDE) in its southern part due to the difficulty of access to other parts covered by forests. The monitoring of the groundwater levels in this exploited part of the aquifer shows that the piezometric levels have decreased by about 20 m over the last 30 years (DGRE, 2018). In this case, the exploration of all the extensions of the aquifer has become an obligation concerning the drinking water supply of the residents. Previous studies in the region (e.g. de Marsilly, 1972; Djebbi and Gabtni, 2018; Manaa, 1987) were mainly oriented on the identification of the flows and barely on the storage. In addition, the geometry of the aquifer has never been updated since de Marsilly (1972). This study is a component of a larger investigation to assess the groundwater potential of the massive dunes of the Nefza basin. Thus, the objective of this paper is to evaluate the aquifer geometry and spatial distribution of groundwater storage. This work provides valuable information on the spatial distribution of groundwater storage that can be useful in predicting groundwater availability in the future. It also represents an entry point for future groundwater flow modeling. Overall, the methodology described in this paper can be applied to any region of the world because of its simplicity and best estimate of groundwater storage distribution.