The Wadi Watir is one of the main wadis in the region of the Aqaba rift province of the Sini Peninsula in Egypt, and it is located between Taba in the north and Sharm El-Sheikh city in the south. The Wadi Watir basin is an important water resource area for Nuweiba city, which is located on its delta. The Wadi Watir basin is bounded by the range of basement mountainous to the west and the shoreline of the Gulf of Aqaba to the east [34]. The catchment area of Wadi Watir (3860 km²) has a steep mountainous topography that consists of 12 sub-basins and about 32 main wadis, which drain toward the Gulf of Aqaba. High flows occur here mainly during two main storm periods, one in spring and the other in autumn. There are 5 to 10 rainy days per year with a mean annual rainfall amount ranging from 30 to 40 mm, and the bulk of the volume of rainwater falls within a period of a few hours [35]. Annually, the amount of rainfall affecting the Wadi Watir catchment areas is about 200 million m³/year. The major proportion of this rainfall is lost to the Gulf of Aqaba as surface water runoff because of the steep nature of the surface topography and the dominant presence of hard rocks; Table 1 shows historical information for flash flood occurrences and volumes in this region [35] [36].

Quaternary deposits constitute the main water-bearing formation (Quaternary aquifer) in the Wadi Watir delta. This aquifer is unconfined, heterogeneous, and composed mainly of fine to coarse sands, gravels, sandstone, boulders of carbonate, and granitic rock embedded in a clay and silt matrix [37] [38]. The Quaternary aquifer in the study area can be categorized into five layers, and the

thickness and composition of each layer are shown in Figure 2, which illustrates hydrogeological cross sections in the study area [39] [40] [41].

The Quaternary aquifer in the Wadi Watir delta is an unconfined aquifer in which the water depths range from 2.4 to 40.8 m [41]. The fresh water layer with a salinity of less than 2500 ppm attains a thickness ranging between 12 m near the outlet of the wadi to approximately 7 m in the middle part of the Wadi Watir delta, while values reach to about 4 m near the shoreline. Groundwater over pumping from the Wadi Watir delta aquifer has caused seawater intrusion that is controlled to some extent by factors such as the distance from the shoreline, hydraulic conductivity, and aquifer recharge conditions. The first water table map was constructed based on measurements of water levels from October to November 1994 by Himida (1994) [4], while the recent water level contour map was constructed based on measurements of water levels from October to November 2012 by Eissa et al. (2013) [40]. Generally, groundwater flows from the northwestern to southeastern parts of the Wadi Watir delta, and the absolute groundwater level ranges between 0 to 2.5 m above mean sea level (amsl) (see Figure 3(a) and Figure 3(b)). The groundwater hydraulic gradient in the study area varies between 2.3 × 10⁻⁴ in the south and 2.7 × 10⁻³ in the north with an average value of about l.5 × 10⁻³.

The hydraulic conductivity map was created based on the results of the pumping tests that had been carried out by the Desert Research Center [4]. As shown in Figure 4, the values of hydraulic conductivity ranged between 1 and

12 m/day. Low values were found in the south and high values were observed at the central part of the Wadi Watir delta, whereas moderate values were observed in northern localities. The storativity of the water-bearing horizon in the Wadi Watir delta was estimated to vary between 3.16 × 10⁻⁴ and 0.01. The groundwater salinity map was created based on measurements taken in 75 shallow dug wells and deep wells distributed throughout the Wadi Watir delta. As shown in Figure 5, the values of groundwater salinity ranged between 2000 ppm and 8000 ppm. Low values of groundwater salinity were detected in the west at the wadi outlet and high values were observed at the eastern and southern parts of the Wadi Watir delta [34] [40].

Groundwater natural recharge values have been estimated by many researchers such as [4] [5] [35] [36] [42] [43] [44]. The pumping rates in the study area varied through the period 1982 to 2012 depending on the groundwater availability and flash flood intensity and/or frequency. The average groundwater recharge in the Wadi Watir delta ranged between 3900 m³/day (1982 to 1987) to 7770 m³/day (1987 to 2002) with an annual daily average of 6000 m³/day through the period from 1982 to 2009 (see Table 2). However, for each hand dug wells (40 hand dug wells), the pumping rates ranged between 0.5 and 7 m³/day [40].

To assess the hydrogeological suitability of installing ASR systems in the Wadi Watir delta, a numerical groundwater flow and transport model was developed to simulate the horizontal and vertical groundwater movement and upward migration of deep saline groundwater due to groundwater pumping and seawater intrusion; then, simulations of the mixing process between injected water and ambient groundwater were conducted because this process is one of the main factors governing the recovery efficiency in ASR systems. Finite-difference code MODFLOW linked to the particle tracking code MODPATH and solute transport code MT3DMS (Codes link https://www.usgs.gov/mission-areas/water-resources/science/modflow-and-related-programs?qt-science_center_objects=0#qt-science_center_objects) were used; all three codes were run by using the graphical user interface. Visual MODFLOW flex 6, (https://www.waterloohydrogeologic.com/introducing-visual-modflow-flex-6-0) which is capable of simulating the groundwater flow in cases of injection and recovery water, was also used. MODPATH uses the average linear velocity calculated from heads that are generated by MODFLOW to track the movement of imaginary particles that are added to the injected water by advection. These particles were tracked throughout the duration of the storage and recovery cycles [45]. In order to quantify the volume of injected water recovered from the aquifer, the transport code MT3DMS was used to calculate the concentrations of water in the aquifer while considering both advection and dispersion and assuming that the aquifer is an equivalent porous medium with connected pore space; the code also assumes miscible flow and no density effects.

In the horizontal direction, an orientation grid (100 rows × 50 columns) along the north-south axes and east-west axes of the modeled area was created (see Figure 6), while in the vertical direction, the model grid consisted of two layers. The first layer with a hydraulic conductivity of 4 m/day represents the alluvial deposits, and the second layer was comprised of Pliocene to Pleistocene aged sand with a hydraulic conductivity from 4 to 12 m/day [34]. For each layer, the vertical hydraulic conductivity values were assumed to be one-tenth of the values of the horizontal ones. Aquifer materials were assumed to be homogeneous and mainly composed of poorly sorted sand, silt, gravel, and clay; hence, the total

porosity was set to 0.1 [46] [47]. The storage coefficient value used for the model ranged from 3.16 × 10⁻⁴ to 0.01; also, the longitudinal dispersivity was assumed to be homogeneous and set to 10 m, the ratio of horizontal transverse dispersivity to longitudinal dispersivity was assumed to be 0.1, and the ratio of vertical transverse dispersivity to longitudinal dispersivity was assumed to be 0.01 [48].. The diffusion coefficient was assumed to be 9 - 10 cm/s [40]. The 10-m longitudinal dispersivity according to Gelhar (1992) was located within the category of high confidence in the small-scale field suite (1 - 1000 m) and in the category of low confidence for the entire scale of the Wadi Watir delta (for large scales of 10³ - 10⁶ m) according to contaminant events tracer tests [48].

The model base was assumed to be the NO flow boundary, and for the flow and transport model, the constant head boundary was set to sea level at the shoreline, where the constant concentration was assumed to be 41,000; meanwhile, the western boundary was set to the source points boundary with a concentration value of 1000 ppm [40]. An initial water level map for the steady state was created from the values of the water levels that were measured in February 1986 [37]. Steady state conditions of groundwater flow and solute transport models were identified from simulations under initial conditions until reaching system stabilization and equilibrium for salt and physical flow within the models. The groundwater salinity values, which were obtained from the steady state, were modified with the salinity data for deep saline groundwater from an earlier study [34], where salinity reached 18,000 ppm at a depth of 34 m below sea level. In accordance with the data availability, simulations were carried out for the period from 1982 to 2012. Figure 7 shows the boundary condition observation points and abstraction wells for both the flow and transport models.

The natural recharge in the model area was assumed to be zero because the annual average rainfall for the Wadi Watir area only ranges from 10 mm/year in the Wadi Watir delta to 50 mm/year in the highland areas [35]. On the other hand, the main recharge to the aquifer comes mainly from subsurface groundwater flow from the mountain blocks [5] [49], and the main source of subsurface recharge is flash floods that occur on average about every six years [4].

The comprehensive groundwater flow and solute transport model that was developed was calibrated by obtaining general agreement between the observed and modeled salinity and water level values; this model was found to be more sensitive to porosity, hydraulic conductivity, and dispersivity values than the other factors. The recharge was the main factor controlling the head, while the recharge rates, pumping rates, dispersivity, and effective porosity were the main factors controlling the salinity. A total of 37 head observations [5] [34] [40] were used to calibrate the model in addition to the salinity observations (see Figure 8). The pumping rate through the drilled wells changed with time (8% to 15%) for each stress period through the modeled time (1982 to 2012) according to the original data that were collected. The model calibration was complete when a reasonable value of the relative error of less than 10.5% was achieved between the modeled and observed water levels. The sensitivity of the model was tested to the full range of hydrogeological and operational parameters. The model was found to be valid for most regions within the model area except for the area close to the coast and the area located in the Sabkha deposits.

The ASR systems were simulated in five specific locations in the Wadi Watir

delta by using different injection rates. The relative effects of selected hydrogeological parameters on the recovery efficiency were similar to those discussed by Lowry (2004) [50]. Notably, the dispersivity that describes the mixing process was one of the hardest parameters to quantify in terms of the recovery efficiency impacts in ASR systems. Thus, the effects of recovered water salinity on the recovery efficiency of ASR systems were evaluated at the five locations with different dispersivity values of 3, 10, 20, 50, 100, and 200 m. Figure 10 shows the effects of dispersivity on the recovery efficiency of the aquifer model with a total volume of injected water of 12,000 m³ and a storage period of 180 days.

The main operational parameters that affected the recovery efficiency were the storage period and the volume of injected water, as shown in Figure 11, where the recovery efficiency decreased as the storage period increased; Figure 12 shows that the recovery efficiency initially increased as the injected water volume increased, but then it decreased with large volumes of injected water because of the high hydraulic gradients created by water moving away rapidly from the ASR areas.