You are on page 1of 21

Journal of Hydrology 626 (2023) 130139

Contents lists available at ScienceDirect

Journal of Hydrology
journal homepage: www.elsevier.com/locate/jhydrol

Research papers

Mitigating seawater intrusion in coastal aquifers: Novel approach with


treated wastewater injection and groundwater circulation
Samia Saad a, b, *, Akbar A. Javadi b, Hany F. Abd-Elhamid c, d, Raziyeh Farmani b
a
Department of Irrigation & Hydraulics, Ain Shams University, Cairo, Egypt
b
Department of Engineering, University of Exeter, Exeter, United Kingdom
c
Department of Water and Water Structures Engineering, Faculty of Engineering, Zagazig University, 44519 Zagazig, Egypt
d
Department of Environmental Engineering, Technical University of Košice (TUKE), Košice, Slovakia

A R T I C L E I N F O A B S T R A C T

This manuscript was handled by Corrado Cor­ Seawater intrusion (SWI) is a natural phenomenon that negatively impacts the potability of groundwater and is
radini, Editor-in-Chief, with the assistance of expected to worsen with rising sea levels due to climate change. Artificial recharge of freshwater is a commonly-
Clement Roques, Associate Editor used remediation method to mitigate SWI and improve freshwater supply security in affected coastal aquifers.
However, limited freshwater availability can limit the effectiveness of this approach. This study proposes a novel
Keywords:
mitigation measure, called Inj-GCW, which combines the injection of reclaimed water with the use of ground­
Seawater intrusion
water circulation wells (GCW) to enhance the effectiveness of artificial recharge in controlling SWI. GCW is a
Injection
Groundwater circulation well, sea level rise dual-screened well with isolated screens that extract and inject groundwater into the aquifer. The performance of
Nile Delta aquifer the proposed measure was quantitatively evaluated using an illustrative simplified unconfined coastal aquifer.
Mitigation Based on the findings, the design parameters were estimated for a field-scale case study of the Nile Delta aquifer
(NDA), a large Mediterranean coastal aquifer in Egypt. The study adopts a future scenario that considers Sea-
level rise due to climate change and projected population growth by 2100. Results demonstrate that intro­
ducing of GCWs into the saltwater wedge, along with injection, effectively retreated the saltwater wedge, due to
enhanced velocity, seaward fluxes, and dilution of contamination. Inj-GCW measure led to an 8.9% reduction in
SWI and a 5.2% decrease in aquifer salinity compared to the expected intrusion in 2100. Furthermore, the Inj-
GCW measure resulted in a 2.2% higher repulsion rate and 0.3% reduction in total salt mass compared to in­
jection alone. The Inj-GCW measure presents a promising solution to SWI challenges in the NDA and other
coastal aquifers facing similar issues. The formation of a brackish water bubble at the injection well screen of the
GCW and the generated vertical groundwater circulation cells acted as a hydraulic barrier and contributed to the
proposed method effectiveness.

1. Introduction to 50–60 km, affecting 30,000 ha of agricultural land (Vu et al., 2018).
Mitigating SWI requires effective measures to control or reduce it.
Seawater intrusion (SWI) occurs naturally due to salinity differences Various remediation/mitigation measures have been proposed, each
between seawater and freshwater (Jiao and Post, 2019). However, with their advantages and disadvantages, as summarized in Table 1.
coastal aquifers are experiencing interruptions in this natural equilib­ However, finding a single solution that universally applies to all specific
rium, leading to a decline in groundwater quality and ecological impacts site conditions remains elusive and unattainable. Among the measures
(Idowu and Lasisi, 2020; Saad et al., 2023). With global climate change, proposed, recharging the aquifer through positive hydraulic barriers is
especially sea-level rise (SLR) and overexploitation of groundwater, SWI the most commonly used method (Hussain et al., 2019; Shi et al., 2020).
is expected to worsen. For instance, in the Nile Delta, SLR (1.5 m) and This method increases hydraulic head, relieves over-exploitation by
groundwater abstractions (4.9 billion m3/year) could lead to no fresh­ supplying part of the demand, maintains a seaward hydraulic gradient,
water access by 2100 (Mabrouk et al., 2018). Similarly, in the Mekong and flushes out contaminated water. Different water sources can be
Delta, a 0.3 m SLR in 2050 could cause salinity intrusion into the river up used, including desalinated seawater (e.g., Ganot et al., 2018), rivers

* Corresponding author at: Department of Engineering, University of Exeter, Exeter, United Kingdom.
E-mail addresses: samia.saad@eng.asu.edu.eg, ss936@exeter.ac.uk (S. Saad), A.A.Javadi@exeter.ac.uk (A.A. Javadi), hany.abdelhamid@tuke.sk, hany_
farhat2003@yahoo.com (H.F. Abd-Elhamid), R.Farmani@exeter.ac.uk (R. Farmani).

https://doi.org/10.1016/j.jhydrol.2023.130139
Received 14 April 2023; Received in revised form 11 August 2023; Accepted 17 August 2023
Available online 11 September 2023
0022-1694/© 2023 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
S. Saad et al. Journal of Hydrology 626 (2023) 130139

Table 1 Table 1 (continued )


Remediation/mitigation measures against SWI, with their advantages and Method Advantages Disadvantages Ref.
disadvantages.
by a packer and water level cannot the regional fresh
Method Advantages Disadvantages Ref. grouting. One be raised. groundwater flux.
New water - - An easy direct - Changes in Sherif et al. screen extracts - No net loss or
resources mitigation established (2014); Song and the other gain of
legislation measure. irrigation and et al. (2018). injects groundwater.
includes - Redesigning domestic practices groundwater
reducing, the pumping due to alterations in into the aquifer.
optimizing rates is a cost- pumping rates that Combining - Increase water - Lack of suitable Abd-Elaty and
regional free measure.- have significant positive and storage.- Raise the water for recharge. Zelenakova
abstraction, Can be com­ socioeconomic and negative piezometric head. - Clogging (2022); Saad
and/or bined with other political hydraulic - Reduce saltwater - May cause a et al. (2022).
relocating the control consequences. barriers (mixed volume. reduction in
wells to zones measures. - Wells relocation is hydraulic - More efficient freshwater
less susceptible costly. barriers) than single barrier resources.
to SWI or - Temporary systems to control - Strongly reliant on
shallowest management SWI good hydraulic
portions of the approach. characterization for
aquifer barrier locations
Using physical - Physically - High initial costs, Abdoulhalik and flow rates
subsurface prevent the SWI. particularly for deep et al. (2017); relative to other
barriers to - Don’t need aquifers. Chang et al. single barrier
reduce the maintenance and (2021) systems.
aquifer repair over their Combining - Increase water - Lack of suitable Present study
permeability lifetime. injection wells storage.- Raise the water for recharge.
such as and groundwater piezometric head. - The effectiveness
subsurface dam, circulation wells - Reduce saltwater of the measure is
cutoff wall, and (Inj-GCW volume. highly influenced by
sheet pile measure). - Enhance GCW the regional fresh
Extending the - Land - It may be infeasible Abd-Elhamid efficiency by groundwater flux.
coastline reclamation can if newly reclaimed et al. (2020); coupling with - Clogging
towards the sea significantly soil has low Shen et al. injection wells, an - Strongly reliant on
using coastal increase the land permeability and (2022) external source of good hydraulic
earth fill availability there is a large tidal freshwater influx. characterization for
amplitude. - Achieve no net barrier locations
- It can negatively loss or gain of and flow rates
impact the coastal groundwater to relative to other
groundwater address potential single barrier
hydrodynamics and issues arising from systems.
marine abstraction
environment. barriers,
- Costly to be particularly the
implemented at a decrease in
large scale. available
Using horizontal - Avoid saltwater - Decrease the Yin and Tsai freshwater
wells (scavenger upconing. chance of upconing (2019). volume.
wells), skimming at local scale.
wells or
subsurface and lakes (e.g., Dibaj et al., 2021), stormwater (e.g., Mohammad-Hos­
drains seinpour and Molina, 2022), and treated wastewater (e.g., Saad et al.,
Using positive - Increase water - Lack of suitable Dibaj et al. 2022). Using unconventional water resources to store in the aquifer has
hydraulic storage. water for recharge. (2021); Zang
barriers typically - Raise the - Clogging and Li (2021)
environmental benefits, including passive treatment of wastewater and
through artificial piezometric head. - Need space for stormwater (Page et al., 2018), reduced evaporation loss (Alam et al.,
recharge using infiltration basins 2021), and decreased carbon emissions due to the raised water table and
infiltration reduced energy requirements for pumping groundwater (Dillon et al.,
basins, canals,
2020).
recharge ponds,
or injection wells Positive hydraulic barriers have limitations in the amount of water
or through available for recharge, which may hinder their effectiveness as a miti­
gaseous barriers gation measure against SWI. To overcome this limitation, previous
by injecting studies have suggested combining them with other physical or hydraulic
compressed air
instead of
barriers. For instance, Armanuos et al. (2019) experimentally demon­
freshwater strated that freshwater injection and cutoff walls can improve the
Using negative - Reduce saltwater - May cause a Kacimov et al. repulsion ratio of positive hydraulic barriers from 22.39% by up to
hydraulic volume. reduction in (2009); Pool 45.4%. Chang et al. (2021) recommended recharge facilities with cutoff
barriers through - Suitable in the freshwater and Carrera
walls to accelerate and improve desalination of coastal aquifers. Semi-
intercepting aquifer where the resources. (2010).
inflowing water level cannot - Highly sensitive to permeable subsurface dams alone may not be also sufficient to pre­
brackish/ be raised. changes in pumping vent SWI in mild seaward hydraulic gradient situations, regardless of the
saltwater. rates. extent to which the subsurface dam covers the aquifer thickness, as
Using groundwater - Reduce saltwater - The effectiveness Vats et al. demonstrated by Abdoulhalik et al. (2017). Similarly, Abd-Elaty et al.
circulation wells, volume. of the measure is (2020)
dual-screened - Suitable in the highly influenced by
(2021); Abd-Elaty and Zelenakova (2022); and Mabrouk et al. (2019a)
wells separated aquifer where the observed increased repulsion of intrusion in the combined system
compared to injection alone. Combining positive hydraulic barriers with

2
S. Saad et al. Journal of Hydrology 626 (2023) 130139

other measures is crucial to enhance freshwater safety and sustainabil­ than conventional negative barriers, groundwater was abstracted to
ity, especially in regions with water scarcity and limited access to create a hydraulic gradient that acted as the barrier. This technique is
treated wastewater and desalination, such as arid and semi-arid areas also recommended by Sharan et al. (2021) in their review study, to
and low-income countries. The Nile Delta aquifer (NDA) exemplifies prevent SWI in countries that suffer from over-abstraction.
these challenges, with excessive abstraction causing a 0.1 billion m3/ To the best of our knowledge, Vats et al. (2020) is the only study that
year decrease in groundwater potentiality (Negm et al., 2018) and SWI used GCW in controlling the SWI, they assessed its effectiveness after
reaching up to 100 km inland (Abd-Elaty et al., 2021). The Mediterra­ ceasing the groundwater abstraction on a laboratory-scale flow tank
nean coastal aquifer is severely threatened by SWI (Abd-Elaty et al., model. However, the performance of GCWs with continuous regional
2021; Abd-Elaty and Zelenakova, 2022; Mabrouk et al., 2018; Negm groundwater abstractions, and their dynamics when combined with
et al., 2018) and potentially exacerbated by the Grand Ethiopian Re­ injection wells remain unclear. Additionally, the effects of GCW design
naissance Dam’s impact on Nile River flow (Abd-Elhamid et al., 2019). parameters, including screen placement, location, and circulation flow,
Furthermore, NDA is highly vulnerable to SLR (Pörtner et al., 2022), on controlling SWI have not been studied. Furthermore, the applicability
emphasizing the urgency for combined mitigation measures. of this new methodology to reduce SWI on a large field scale, such as the
Previous studies have examined the occurrence of fresh and saline NDA (≈25,000 km2), has not been investigated. In this study, these
groundwater in the region, but fewer have focused on controlling SWI research gaps are addressed and insights in sustainable groundwater
(Abd-Elaty et al., 2021). Sherif (2003) identified the best sites for an management are provided.
additional pumping of 2.3 billion m3/yr with controlling further intru­ Therefore, this study aims to 1) construct and calibrate a 3D density-
sion of 35000 mg/l isoline, representing the seawater concentration. He dependent flow and mass transport finite element model simulating the
found that abstraction from the middle zone would reduce the intrusion SWI in the NDA, and predict its future status considering future climate
throughout the Delta, albeit with a slight increase in the eastern zone. change scenarios. 2) introduce the new mitigation measure of Inj-GCW
However, pumping from the eastern and western zones would only repel and perform a sensitivity analysis describing its behavior in reducing the
the intrusion at the middle zone. Mabrouk et al. (2019a) proposed three SWI and the aquifer salinization using an illustrative case study
adaptation measures for Sharkeya governorate (third in the adminis­ mimicking the NDA, then propose some design guidelines. 3) test the
trative districts in terms of crop production) considering the impacts of effectiveness of the measure on the NDA and evaluate its performance
SLR and the potential groundwater abstraction in 2100. The measures against other mitigation measures adopted in previous studies.
included changing crop patterns and improving irrigation practices, The proposed approach can be a promising solution to the challenges
injecting the available amount of treated wastewater, and abstracting of SWI in the NDA adding to the existing literature on mitigation mea­
brackish groundwater. The measure of changing crop patterns and sures for the protection of freshwater resources in coastal aquifers. Here,
improving irrigation practices, which resulted in a decrease in total we explore the Inj-GCW approach, which not only presents a novel
abstractions, showed the most promise compared to the other measures. avenue for further research in this domain but also has the potential to
Abd-Elaty et al. (2021) examined decreasing the abstractions, increasing be applied to other coastal aquifers facing similar SWI challenges.
the recharge, abstracting brackish water, and a combination of these
measures on the Middle Nile Delta under an SLR of 0.5 m and a potential 2. Methodology
increase of 50% in the regional abstractions. They reported 6.2%,
19.5%, 5.9%, and 21.3% retardation in the SWI for the above 4 measures 2.1. Nile Delta aquifer study area
respectively. Abd-Elaty and Zelenakova (2022) argued that subsurface
barriers and coastal earth fill measures are not effective for controlling The Nile Delta is a large Mediterranean coastal aquifer, situated on
SWI in deep aquifers such as NDA, and mixed hydraulic barriers of the northern plateau of Egypt. The Delta has 11 administrative districts
injecting treated wastewater and abstracting brackish water would have (Fig. 1) with the highest population density in the country at 1724
a great positive effect on the Nile Delta aquifer. The mixed hydraulic people per km2. The region heavily extracts water for irrigation and
barriers further repelled intrusion by approximately 1.75 km seaward. domestic purposes (Mabrouk et al., 2019b). Table 2 displays an over­
From previous studies, artificial recharge and abstraction of brackish view of the study area’s physical, geological, and hydrogeological
water are vital for the success of the mitigation measures. However, it attributes.
was found that the abstraction barriers can extract freshwater more than
brackish or saline water, eventually leading to a decrease in the avail­ 2.2. Numerical modeling of variable-density groundwater flow and solute
able freshwater volume (Hussain et al., 2019; Pool and Carrera, 2010). transport in NDA
To alleviate this problem a new mitigation measure (Inj-GCW) is pro­
posed based on injection wells and groundwater circulating wells (GCW) FEFLOW, a finite element groundwater flow software (Diersch,
to control the SWI in the NDA. This approach addresses the limitations of 2013), was used to simulate the advective–dispersive SWI process in the
the traditional method of combining negative and positive hydraulic coastal aquifer. The governing flow and mass transport equations in a
barriers by overcoming potential losses in freshwater volume. fully saturated heterogeneous porous medium were solved considering
GCW is an effective in-situ technique for remediating contaminated variable fluid density. In the present study, the viscosity was considered
groundwater. It is a dual-screened well that circulates groundwater by to be independent of the concentration, and thermal effects were
extracting from a screen, treating the water within the well, and neglected. The convective form of the transport equation was considered
injecting the treated water through the second screen (Elmore and because it is a practical choice regarding the required computational
Hellman, 2001). The induced circulation flow eventually flushes the time (Diersch, 2013). For detailed information on the governing equa­
contaminant within the well proximity (Vats et al., 2020). Unlike tions of coupled groundwater flow and salt mass transport can be found
traditional pump-and-treat systems, GCWs have no adverse effects on in Diersch (2013).
aquifer drawdown, surface discharge of treated groundwater or property The spatial and temporal discretizations utilized in the numerical
acquisition. (Elmore and Hellman, 2001). Furthermore, GCW system can model, as well as the considered hydraulic and transport parameters, are
be considered a green technique due to its inherent resource- presented in Table 3.
conservation nature of no net loss or gain of the groundwater through
the circulation (Elmore and Hellman, 2001). Although primarily used 2.2.1. Initial and boundary conditions
for removing organic compounds, (Ainscough et al., 2021), Vats et al. The hydraulic head values were initialized as zero, while the initial
(2020) took advantage of circulating the water, without any treatment, concentrations were determined based on previous study contour maps,
to create a barrier and push back the saline wedge toward the sea. Rather which accounted for the salinity of the hypersaline groundwater flowing

3
S. Saad et al. Journal of Hydrology 626 (2023) 130139

Fig. 1. Map of Nile Delta aquifer, showing the locations of different administrative districts, and the observation points.

from the bottom (Nofal et al., 2015). The simulation period lasted for Elaty and Zelenakova, 2022). The two main branches started at 15.14 m
365000 days considering the boundary conditions static. A dynamic in the north and ended at approximately 0.5 m near the sea. The water
equilibrium was achieved asymptotically after 325000 days. The simu­ heads of the surface water networks varied spatially, but they were
lation results were considered as the steady-state condition in 2013, and considered temporally constant. This is due to the irrigation structures,
the initial conditions for subsequent analyses. Fig. 2a and Fig. 3a,b, and c which were constructed along the Nile river controlling the water levels
illustrate the horizontal and vertical salinity distributions in 2013, in the canals throughout the year (Sherif et al., 2012). The salinity
respectively, with the vertical cross sections taken from the middle concentration in the main branches showed an increase from 250 mg/l
(section 1), east (section 2), and west (section 3) of the study area. (freshwater density is considered 1 mg/l) in Cairo to 650 mg/l towards
The study area is located in a semi-arid region, which is character­ the sea in the north, while most of the canals showed an average salinity
ized by limited rainfall. The aquifer is mainly replenished with the re­ of 300 mg/l (Mabrouk et al., 2019b). A constant Dirichlet BC of zero (the
turn flow of irrigation water and the leakage from the surface water mean sea level) was imposed on the north boundary along the Medi­
networks (Abd-Elaty and Zelenakova, 2022). The irrigation return flow terranean sea with a salinity of 35000 mg/l for a saltwater density of
was assigned as a specified flux (Neumann BC), and its value varied 1.025 g/cm3. On the bottom boundary, the seepage flux of the hyper­
spatially from 0.25 to 1.1 mm/day (Negm et al., 2018). Its salinity saline groundwater, expulsed from the Pliocene formation due to its
ranges from 100 mg/l south to 700 mg/l north (Mabrouk et al., 2019b), compaction, was assigned by a Neumann-type BC with a constant value
and was assigned as a Neumann-type mass flux boundary condition (BC) of 3 × 10− 6 m/day (Van Engelen et al., 2019), and its salinity concen­
on the aquifer’s top slice. The surface water networks (two main tration varied from around 40000 mg/l to 110000 mg/l (maximum
branches, main irrigation canals, and open drains) were represented by density of 1.071 g/cm3) (Nofal et al., 2015). Hence, different boundary
Cauchy BC, representing the exchange between those water bodies and conditions were imposed internally and on all the outer boundaries
the aquifer system. Cauchy BC was assigned in the first layer requiring a except the eastern boundary, which corresponds to the Suez Canal and
specified head, specified salinity, and the transfer rates of flux and mass. was considered impervious to both the flow and solute transport
The specified head for the Ismailia canal, which runs from the south to (Mabrouk et al., 2019b).
the east, varied from 16.17 to 7.01 m (Abd-Elaty and Zelenakova, 2022). The overall groundwater extraction from the NDA was estimated to
For the western boundary, El-Rayah El-Behery and Nubaria canals, the be about 3.56 billion cubic meters per year in 2013 (Nofal et al., 2015).
water head ranged from 16.96 m in the south to 0.5 m in the north (Abd- This amount was distributed among 3228 extraction wells. A single well

4
S. Saad et al. Journal of Hydrology 626 (2023) 130139

Table 2 Table 3
Characteristics of the study area. Model Discretization and Hydraulic/Transport Parameters for the NDA.
Characteristic Description Model discretization of NDA
Feature Description
Physical description
Location Nile Delta, Egypt, latitudes 30˚00‘ and 31˚45‘N, and Horizontal discretization Using 68,266 linear triangular prism elements per
longitudes 29˚30‘and 32˚30‘E layer, with refinements near the coast, model
Surface area ≈ 25,000 Km2 boundaries, and high-velocity areas (e.g., pumping
Thickness The NDA is remarkably thick, reaching up to 1000 m depth wells, Nile branches, canals, and drains). Mesh
at the coast. Across the globe, only 0.3% of the coastal resolution ranged from 0.3 to 2.3 km.
aquifers are deeper (Van Engelen et al., 2019). Vertical discretization Using 17 layers, from top: 1 layer (7.5 m)
Domain boundary NDA has a triangular shape with a 300 km coastal base on considering surface water depths, 8 layers (max 50
the Mediterranean sea and an apex near Cairo, 200 km m) considering extraction well screens, 7 layers
away from the sea. The Delta is centered by the Nile River, (max 100 m), and 1 layer (5 m) accounting for the
which divides into two branches at the apex - Damietta and flux from Pliocene deposits.
Rosetta. The delta is bordered by the Suez Canal to the east, Temporal discretization Using the Automatic time-stepping method with a
Ismailia Canal to the southeast, El-Rayah El-Behery to the predictor–corrector scheme, adjusting time step
southwest, and Nubaria Canal to the west (Fig. 1). based on convergence behavior with an error
Geological tolerance of 1 × 10-3. Initial time step: 0.001 day.
information The maximum growth rate between subsequent
Geological formation Quaternary deposits consist of the Holocene and Pleistocene time steps was 2.
layers (Sestini, 1989). Below these two layers, is an Hydraulic and transport parameters of NDA
impervious Pliocene formation. Feature Description
Holocene layer Consists of medium to fine-grained silt with clay and peat in Horizontal hydraulic 50–500 mm/day (Nofal et al., 2015).
some locations (Pennington et al., 2017). Its thickness conductivity of Holocene
decreases towards the delta fringes in the south and reaches layer
up to 50 m near the sea. Vertical hydraulic 25 mm/day (Nofal et al., 2015).
Pleistocene layer Consists of coarse sands with a few clay intercalations. It is conductivity of Holocene
capped by the Holocene layer (RIGW, 1992). Its thickness layer
ranges from 200 m in the south up to 1000 m near the sea in Hydraulic conductivity of 23–177 m/day, average of 75 m/day (Elshinnawy
the north. It is the main highly productive aquifer. Pleistocene layer et al., 2015).
Pliocene layer Consists of marine clays. It is generally considered the Specific yield of top Holocene Gradual decrease from 12% to 0% towards the
aquifer hydrogeological base (Abd-Elaty and Zelenakova, layer north (Eltarabily and Negm, 2017; Eltarabily and
2022; Mabrouk et al., 2019b). Negm, 2018).
Hydrogeological Specific yield of Pleistocene Constant average value of 35%.
System layer
Aquifer classification The aquifer is unconfined in the east and west with a thin Specific storage Average value of 10–5 (1/m) (Van Engelen et al.,
top Holocene layer and confined in the northern middle 2018).
Delta with a thick top Holocene layer. It is also leaky in the Effective porosity of Varies spatially from 0.12 near the coast to 0.2 in
southern and middle parts. (Sherif et al., 2012). Pleistocene layer the south (Sherif et al., 2012).
Depth to water table Groundwater exists at shallow depths ranging from 1.0 near Effective porosity of clay cap Constant average value of 0.5 (effective porosity =
(m) the coast to 5 m in Cairo below the ground surface. 0.067) (Mabrouk et al., 2019b).
Transfer coefficient of surface 0.01 (1/day) (Van Engelen et al., 2019). It is used
water beds to simulate the rate of flux exchange between
can represent several nearby wells due to the relatively coarse grid of the surface water and groundwater.
model. The extraction wells were assigned vertically in different layers Longitudinal dispersivity 10 m (Nofal et al., 2018).
Transverse dispersivity 1 m (Nofal et al., 2018).
based on the location and length of the well screen. Their depth varied Molecular diffusion 8.64 × 10–5 m2/sec (Van Engelen et al., 2018).
between 45 and 180 m below the ground surface. Transfer rate of mass transport 1.5 m/day, adapted from a similar system for the
coefficient watershed of the Colombia River, USA (Fang et al.,
2.2.2. Model calibration 2020).
The calibration process was undertaken using manual trial-and-error
due to the long computation time. The calibrated parameters included 2.2.3. Future scenario
the irrigation return flow, seepage flux of the hypersaline groundwater, The current conditions for the flow and salinity in 2020 were ob­
horizontal hydraulic conductivity for the Pleistocene layer, vertical tained by running the simulation model using the calibrated parameters,
hydraulic conductivity for the Holocene layer, and longitudinal and the above-mentioned BCs but with the consideration of the
dispersivity. groundwater extraction increase and the sea level rise (SLR). The SLR
A total of 984 observation points (246 points at depths of 60 m, 200 boundary condition was updated to comply with the high-level climate
m, 400 m, and 600 m) were located on a pseudo-regular grid covering change scenario RCP 8.5 (Representative Concentration Pathways 8.5),
the entire model domain (Fig. 1). They were separated by approximately which corresponds to the highest CO2 emission. According to the esti­
10 km. The observed data of salinity concentration at depths 200, 400, mate provided by the IPCC, a gradual SLR was considered 4–5 mm/year
and 600 m and the shallow piezometric head at depth 60 m were (Torresan et al., 2020). Assessing the impacts of gradual SLR is more
extracted from the contour maps found in the literature, which were realistic than that of instantaneous SLR (Ketabchi et al., 2016), however,
developed based on collected field measurements in the same year 2013 the impacts of the landward migration of the coastline in response to SLR
(Nofal et al., 2015; Nofal, 2016). were not considered. The increase in groundwater extraction was
The model calibration performance was assessed based on the assumed to correlate with the increase in the population. The total
following statistical indices: 1) Root mean square error (RMSE), which population was expected to reach 200 million in 2100 with an average
measures the difference between the observed and simulated datasets, growth rate of 8.11% per year (Mabrouk et al., 2018). The future sce­
2) Nash-Sutcliffe model efficiency (NSE), which estimates the ratio of nario until 2100 was applied to assess the vulnerability of the NDA to the
the model error to the variability of the data, 3) Coefficient of deter­ combined effect of gradual SLR and increase in groundwater extraction
mination (R2) (the squared value of Pearson correlation coefficient) to to reach 45 cm and 10.95 BCM/year, respectively.
quantify the similarity between the two datasets.

5
S. Saad et al. Journal of Hydrology 626 (2023) 130139

Fig. 2. Plan view of the salinity distributions at the average depth of − 100 m, where most extraction wells are located, (a) in 2013 (initial salinity), (b) in 2020 (just
before implementing a mitigation measure), (c) & (d) in 2100 at the end of the management period, with no management strategy adopted and after implementing
the mitigation measure, respectively.

2.3. SWI management strategy injection rate per well of 40000 m3/day at depths ranging from 50 to
275 m. Due to the grid size used, a single well could represent multiple
The goal of the management was to mitigate the SWI. The 1000, nearby wells. The next section will discuss the GCW technique.
2000, 10000, 17500, and 35000 mg/l isolines (salinity concentrations
contour lines) were used to fully represent the intrusive saltwater 2.3.1. Groundwater circulation well technology
wedge. The average intrusion length of each isoline (Lx ) was calculated Groundwater circulation wells (GCW) allow quasi-in-situ ground­
as in Eq.1. water remediation, where the aquifer is not subjected to a net loss or
gain of the groundwater. They include the following components:

nc
Lx = avg lx (1)
i=1 - A well with two screened intervals separated by vertical distance and
where, lx is the isoline intrusion length, measured per each node that isolated from each other by an impermeable plate,
lies on the coastline (m), nc is the number of nodes that lie on the - A pumping system that induces vertical groundwater circulation by
coastline, and × corresponds to the isoline of interest. The objectives of extracting groundwater from the extraction well interval and dis­
the management were to minimize L17500 (50% isochlor) by the end of charging it to the recharge well interval.
2100 and the total salt mass within the aquifer. These objectives were
chosen as evaluation criteria in previous studies (Abd-Elaty and Zele­ The GCW could operate in two flow modes; standard mode and
nakova, 2022; Saad et al., 2022). The mitigation management started in reverse circulation mode. In standard mode, water is extracted from the
2020 and continued until 2100 considering the future scenario settings. lower screen and injected through the upper screen, while in the reverse
The management utilized the proposed approach of Inj-GCW, which mode, the process is reversed. Vats et al. (2020) tested the effect of both
combines injection wells for recharging the aquifer with available modes on SWI and found that the standard GCW was ineffective.
treated wastewater and groundwater circulation wells to reduce further Therefore, the reverse circulation mode was adopted in the present
toe advance. The treated wastewater used was tertiary treated with a study (Fig. 4).
salinity of 500 mg/l. GCWs are commonly used alongside other remediation techniques
The national development plan (Tawfik et al., 2021) expected a total (such as air stripping, ozone addition, and surfactant usage) to remove
treatment capacity of 0.019 BCM/day for the Nile Delta by 2030, which volatile and nonvolatile organic compounds. However, these techniques
was distributed among the 11 administrative districts based on extracted cannot treat salinity contamination. In this study, GCWs were used
groundwater. 386 injection wells were assigned with an average without passing through an external treatment. They were used as

6
S. Saad et al. Journal of Hydrology 626 (2023) 130139

Fig. 3. Vertical salinity distributions at three cross sections, their locations are shown in Fig. 2. (a,b,c) in 2013 (initial salinity), (d,e,f) in 2020 (just before
implementing a mitigation measure), (g,h,i) & (j,k,l) in 2100 at the end of the management period, with no management strategy adopted and after implementing the
mitigation measure, respectively.

injected.
To simulate the GCW in the numerical model, its upper and lower
parts were designated as the extraction and injection sections, respec­
tively, with each section treated as a single well. Injection and extraction
rates were kept equal. The wellbore was represented as a one-
dimensional discrete feature with infinite hydraulic conductivity. To
formulate the mass flux boundary condition, Neumann-type BC should
be used. This requires applying the divergence form of the mass trans­
port equation to achieve mass conservation on the boundary (Diersch,
2013). However, the divergence form poses difficulties at the outflow
boundary, while the convective form is more stable (Panday et al.,
2018). To address this issue, the terms in Eq. (2) of the Neumann-type
BC for the divergence form were modified to become approximately
equivalent to Cauchy–type BC for the convective form (Eq. (3)).
− (D.∇C).n + C qn = qk (2)
where, D: hydrodynamic dispersion tensor, C is the unknown mass
concentration on the outflow boundary, n corresponds to the normal
unit vector, qn the normal fluxes across the boundary, qk are prescribed
Neumann mass fluxes on the boundary.
− (D.∇C).n − qn (Ck − C) = 0 (3)
where Cis the unknown mass concentration on the outflow bound­
Fig. 4. Groundwater circulation well (reverse circulation mode) sche­ ary, Ck is the prescribed value of mass concentration on Cauchy BC,
matic diagram. which equals the mass concentration of the water extracted by the GCW,
and qn represents the prescribed Neumann fluxes (≈ Cqkk ). A custom script
hydraulic barriers to mitigate SWI, benefiting from the developed ve­ was developed in Python to simulate the GCW in FEFLOW.
locity, which can potentially reduce the inland lateral flow from the sea. GCW design parameters of spacing, screen placement, and injection/
The velocity is enhanced due to the combination of the induced negative extraction flow rate (circulation flow rate) vary according to the aquifer
pressure at the extraction well interval with the overpressure at the in­ conditions such as permeability, anisotropy, saturated thickness, etc.
jection well interval that generates a vertical hydraulic gradient. GCWs (Tatti et al., 2019). However, to implement the new measure of Inj-
can also dilute the initial salt concentration, as the salinity of extracted GWC, it was first examined in an illustrative case study. The hydro­
water from the upper layer is less than that in the lower layer where it is geological settings were set to mimic the NDA.

7
S. Saad et al. Journal of Hydrology 626 (2023) 130139

2.3.2. Illustrative coastal aquifer case study were set apart by 2 km. However, the abstraction wells in this study were
The aquifer is an unconfined, anisotropic, homogenous, deep coastal replaced by GCWs.
aquifer. The model boundary was considered a rectangular shape. The Table 4 lists the conducted simulation runs, to explore various con­
horizontal dimensions were × = 7000 m, y = 3000 m, and the vertical figurations of GCW, including the horizontal location (in the x-axis di­
dimension was z = -1000 m below the sea level, where the aquifer base rection), penetrating depth, injection and extraction screen locations of
was located. The regional abstractions were supplied by 8 partially wells, and the circulation rate. Fig. 5a and Fig. 5b provide a 3D view of
penetrating pumping wells placed between 1000 and 3000 m from the the coastal aquifer model for example run 2, showing the imposed flow
shoreline at depth of − 200 m below the ground surface with a constant and transport boundary conditions, initial hydraulic head, and salinity
pumping rate of 6250 m3/day per well. The abstractions were estimated distributions just before the implementation of the management plan.
based on the average ratio of abstractions to volume in the NDA of 2.14 In light of the performed sensitivity analysis which described the
× 10-6 1/day. More detailed description of the simulation model, and the dynamics of Inj-GCW mitigation measure under different settings (dis­
considered hydraulic parameters are found in the supplementary data. cussed later in Section 3.3.1), the design parameters were determined.
Constant head Dirichlet boundary condition was imposed on the They were then applied to the NDA. The GCWs were placed spatially to
right boundary with 0 m above the mean sea level and 35000 mg/l align with the location of 17500 mg/l isoline at the bottom of the
constant salinity concentration, while for the left side, a constant inflow Pleistocene layer, just before implementing the management strategy.
of 0.015 m/day with a 250 mg/l salinity concentration was considered, 93 GCWs with a spacing of 2 km were used. Vertically, the extraction
estimated using the average inflow to outflow ratio in the NDA of well screens were located near the top of the Pleistocene layer at depths
approximately 0.85. The initial toe position (Xtoe), just before imple­ varying from − 13 m to − 60 m, while the injection well screens were
menting the measure, was located at 6081 m from the coast. The man­ placed at an average depth of − 350 m (ranging from − 118 m to − 517
agement period was 100 years. m) (near the bottom of the Pleistocene layer). The total circulating flow
A single injection well was located at the interface toe, penetrating was considered as 5% of the injected water (950000 m3/day) after
the entire aquifer and positioned at the midpoint of the model domain testing different flows (Section 3.3.3).
along the y-axis (Fig. 5). The impact of the recharge rate was examined
using two rates: 75000 m3/day, equivalent to 1.5 times the total ab­ 3. Results and discussion
stractions, and 32000 m3/day, which maintained a similar ratio to the
injection and regional abstractions reported in the NDA (approximately 3.1. Calibration results of NDA model
63%). For GCWs, two wells were centered on the y-axis, separated by 2
km (Fig. 5). The initial and calibrated parameters are shown in Table 5. The
In previous studies, the spacing between adjacent GCWs was deter­ initial values and their ranges were based on the existing literature. In
mined to guarantee that no fluid particles can escape from the interval general, the hydraulic conductivities in the Quaternary and Holocene
between them (e.g., Elmore and Hellman, 2001). Cleaning up all the salt layers underestimated the salinities considering the initial values. The
particles is technically and economically infeasible. Hence the selected seepage flux increased to mimic the hypersaline groundwater salinity
spacing was adopted from Abd-Elaty and Zelenakova (2022) study. They distribution within the aquifer. A remarkable change was made to the
used mixed hydraulic barriers of abstraction and injection wells to longitudinal dispersivity to enhance the representation of the extent and
control SWI in NDA which included abstracting brackish water and thickness of the freshwater-seawater transition zone, and to avoid the
recharging the aquifer using reclaimed water. The abstraction wells non-convergence of the solute transport equation. It can also be

Fig. 5. 3D illustrative coastal aquifer model with applied flow boundary conditions (a) hydraulic head distribution, (b) salinity distribution before the application of
management strategy (e.g., run2).

8
S. Saad et al. Journal of Hydrology 626 (2023) 130139

Fig. 6. Spatial distribution of the calibrated parameters (a) Holocene layer horizontal hydraulic conductivity, (b) Pleistocene layer horizontal hydraulic conductivity,
and (c) Irrigation return flow rate.

attributed to that longitudinal dispersivity in laboratory experiments are from 2020 to 2100) were investigated. The initial intrusions in 2020 of
usually much smaller than those in the field (Gelhar et al., 1992). As the 1000, 2000, 10000, 17500, and 35000 mg/l isolines were 110 km, 86
irrigation return flows were the main source of recharge, their values km, 44.7 km, 36.4 km, and 18.1 km. The results showed further in­
significantly influenced the groundwater levels. trusions of 1.9 km, 5.8 km, 15.8 km, 12.4 km, and 6.6 km of the isolines
The statistical errors between the simulated and published observed respectively, due to the combined effects of the SLR and over-
hydraulic heads and salinities were computed. The RMSE, scaled-RMSE, abstraction. Salinity distributions in 2020 and 2100 at three vertical
NSE, and R2 of the hydraulic heads were 1.2 m, 9.8%, 0.76, and 0.87, cross-sections are presented in Fig. 3d, e, f, g, h, and i. The total salt mass
and of the salinities at different depths were 4668 mg/l, 6.23%, 0.92, within the aquifer increased from 6.04 × 1010 tons to 7.39 × 1010 tons.
and 0.9 respectively. The scaled-RMSE (RMSE scaled by the range of
measurements) values indicated an acceptable level of model-
measurement mismatch (Barnett et al., 2012). The R2 results showed a 3.3. Implementation of the SWI management strategy
high correlation between the simulated and observed measurements,
where values were > 0.7, indicating a significant relationship. NSE co­ 3.3.1. Performance of Inj-GCW measure in controlling SWI under different
efficients were close to 1, hence the groundwater levels and salinities settings (based on the illustrative case study)
were closely matched by the calibrated model. The statistical errors of To gain a clear understanding of the mechanism of GCW and its ef­
the model calibration showed overall satisfactory results, and are shown ficacy when combined with the injection to control SWI, an illustrative
in Fig. 7. case study was conducted to allow fast evaluations of different config­
urations. The performance of the Inj-GCW measure, injection-only
measure, and no-management scenario were compared according to
3.2. Impacts of SLR and groundwater extractions on NDA the comparison points: repulsion of the intrusions of 1000, 2000, 10000,
17500, and 35000 mg/l isolines and the reduction in the total salt mass
The extent of SWI in NDA was found to be highly affected by the within the aquifer.
combined impacts of SLR induced by climate change and over­ The results showed that the injection-only scheme reduced the
exploitation of groundwater, as highlighted in previous studies (Abd- aquifer’s salinity by 37.5% and repulsed the intrusions of 1000, 2000,
Elaty et al., 2021; Mabrouk et al., 2018; Sefelnasr and Sherif, 2014). 10000, 17500, and 35000 mg/l isolines by 37%, 38%, 47.8%, 49.7%,
Therefore, quantifying their impacts was crucial for sustainable and 12.8%, respectively, compared to the no-management scenario
groundwater management in the region. The effects of the SLR (0.45 m) (Fig. 8). The velocity vectors indicated that the injection scheme was
and population growth (leading to a 2.45 times increase in abstractions able to shift back the hydraulic gradient toward the sea (Fig. 9).

9
S. Saad et al. Journal of Hydrology 626 (2023) 130139

Table 4
Different physical and hydraulic settings of Inj-GCW simulation runs
No. of runs GCW Injection
Distance from the coast (m) Extraction depth (m) Injection depth (m) Total flow (m3/day) Total flow (m3/day)
1 0.25Xtoe ≈ 1500 − 200 − 1000 100,000 75,000

2 0.50Xtoe ≈ 3000 − 200 − 1000 100,000


3 0.75Xtoe ≈ 4500 − 200 − 1000 100,000
4 Xtoe ≈ 6000 − 200 − 1000 100,000
5 0.25Xtoe ≈ 1500 − 500 − 1000 100,000
6 0.50Xtoe ≈ 3000 − 500 − 1000 100,000
7 0.75Xtoe ≈ 4500 − 500 − 1000 100,000
8 Xtoe ≈ 6000 − 500 − 1000 100,000
9 0.25Xtoe ≈ 1500 − 900 − 1000 100,000
10 0.50Xtoe ≈ 3000 − 900 − 1000 100,000
11 0.75Xtoe ≈ 4500 − 900 − 1000 100,000
12 Xtoe ≈ 6000 − 900 − 1000 100,000
13 0.25Xtoe ≈ 1500 − 200 − 900 100,000
14 0.50Xtoe ≈ 3000 − 200 − 900 100,000
15 0.75Xtoe ≈ 4500 − 200 − 900 100,000
16 Xtoe ≈ 6000 − 200 − 900 100,000
17 0.25Xtoe ≈ 1500 − 500 − 900 100,000
18 0.50Xtoe ≈ 3000 − 500 − 900 100,000
19 0.75Xtoe ≈ 4500 − 500 − 900 100,000
20 Xtoe ≈ 6000 − 500 − 900 100,000
21 0.25Xtoe ≈ 1500 − 200 − 500 100,000
22 0.50Xtoe ≈ 3000 − 200 − 500 100,000
23 0.75Xtoe ≈ 4500 − 200 − 500 100,000
24 Xtoe ≈ 6000 − 200 − 500 100,000
From 25 to 48 Similar settings to their corresponding runs from 1 to 24 75,000
From 49 to 72 Similar settings to their corresponding runs from 1 to 24 50,000
From 73 to 96 Similar settings to their corresponding runs from 1 to 24 25,000
97 Similar settings to its corresponding run 2 100,000 32,000

Table 5
Values of initial and calibrated model parameters.
Parameter Initial values Units Lower Upper limit Calibrated Sources for the limits
limit parameters

Longitudinal dispersivity 10 m 0.1 500 100 (Abd-Elaty et al., 2021; Mabrouk et al., 2019b;
Negm et al., 2018; Nofal et al., 2015; Sherif et al.,
2012)
Holocene vertical hydraulic Varied spatially
conductivity
0.005–0.025 m/day 0.0025 16 0.007–0.016
(Fig. 6a)
Pleistocene layer horizontal Varied m/day 5 240 44––150
hydraulic conductivity spatially23.22–––177.49
(Fig. 6b)
Irrigation return flow Northern part 2.5 10-4 m/ 2.5 11 2––11
day
(Fig. 6c)
Middle part 10 2.5 10
Southern part 11 8 11
Eastern part 2.9 2 50
Western part 5 1 15
Hypersaline groundwater 3 × 10-6 3 × 10-6 6 × 10-6 3.75 × 10-6 Crude estimate
seepage flux

Fig. 8 shows the performance of the Inj-GCW measure under could impact the workability of Inj-GCW as a mitigation measure against
different simulation runs (1 to 24), showing that all the runs, relative to SWI.
using injection only, showed further improvement at least in some At a circulating flow rate of 50,000 m3/day per each GCW, the
comparison points. The overall decrease in the total salt mass ranged performance of the Inj-GCW measure was compared to the injection-
from 0% to 6%, while the repulsion of the isolines ranged from − 4.1% to only scheme at various horizontal locations (in the x-axis direction).
16.6%. Fig. 9 displays the vertical salinity distributions along the aquifer Across the six runs (runs 1,5,9,13,17, and 21) where the GCW was
length and the velocity vectors for the initial status, at the end of the placed near the sea, the Inj-GCW measure could reduce the salinity of
management period for the no-management scenario; after imple­ the aquifer, particularly in terms of pushing back the 35000 mg/l isoline
menting the injection scheme; and the Inj-GCW measure. The results by 2.6% to 5.8% relative to using injection only. The reduction in overall
indicated that both the injection and Inj-GCW measures were able to aquifer salinity varied between 0 and 5.2% (Fig. 8). It was found that the
shift back the hydraulic gradient toward the sea, while combining in­ inland migration of the 10,000 mg/l and 17500 mg/l isolines ranged
jection with GCW was a more effective measure for controlling SWI. from 0.59% to 0.71% and 0.16% to 1.26%, respectively (e.g. Fig. 9d for
However, it was found that the horizontal and vertical position of the run1). This migration was caused by the upward and inland lateral flows
GCW significantly affected the amount of improvement achieved, which induced by the injection head buildup, and the change in fluid density

10
S. Saad et al. Journal of Hydrology 626 (2023) 130139

Fig. 7. Goodness of fit of model outputs and their measured counterparts. Calibration results of (a) salinity at different depths, (b) hydraulic head.

Fig. 8. Intrusion length at the aquifer bottom of low and high salinity isolines and the total salt mass for runs 1 to 24 considering the same circulating flow of
100,000 m3/d, 75,000 m3/day for injection, and different locations for the GCW.

reflecting the replacement of higher salinity groundwater with injected across this interface can be encountered (e.g. Fig. 9e for run2). This
water. The study found that the effect of hydraulic gradient increase circulating flow assisted in diluting the salinized groundwater and then
vanished for the 1000 mg/l and 2000 mg/l isolines, resulting in no it was flushed out due to the seaward fluxes. Regarding the 35,000 mg/l
significant impact. isoline, it did not exhibit significant repulsion. Injection rises the water
Locating GCW, midway of the saltwater wedge, near 17500 mg/l table in its vicinity, and hence inland migration is inevitable (Al-Yaqoubi
isoline (runs 2,6,10,14,18, and 22) caused more retreat for the 10000 et al., 2021). As compared to using injection only across runs
mg/l and 17500 mg/l isolines than the other configurations, where the 2,6,10,14,18, and 22, the inland migration by the rest of the isolines
repulsion of 10000 mg/l and 17500 mg/l ranged from 0.8% to 11% and ranged from 0.4% to 1.9%. The maximum reduction in the total salt
5.2% to 16.3% respectively relative to using injection alone. This can be mass (by 1.4% to 6.1%) was experienced when GCW was placed near the
attributed to the curved shape of the freshwater-saltwater interface. The midway of the interface (Fig. 8).
maximum difference in salinity between the extracted water (from Placing the GCW at around 75% of the distance of the interface toe
fresh/brackish water) and where it would be injected (at saline water) measured from the coast (runs 3,7,11,15,19, and 23), resulting in the

11
S. Saad et al. Journal of Hydrology 626 (2023) 130139

Fig. 9. Salinity distributions at a vertical cross-section along the aquifer passing through the GCW location and the velocity vectors (bullet shape tracks the di­
rection), (a) just before implementing the mitigation measure, (b) at the end of the management period for the no-management scenario, (c) injection only, (d),(e),(f),
and (g) Inj-GCW at 0.25, 0.5, 0.75 and 1 Xtoe. (h) Reduced injection rate with GCW at 0.5 Xtoe. Rectangles indicate the GCW location.

maximum retreat for the 1000 mg/l and 2000 mg/l isolines relative to 100 m (elevation − 900 m) respectively. As can be seen in Fig. 8, run 2
all other configurations (ranged from 2% to 5.77%). This emphasizes was the most effective in controlling and mitigating SWI, which was
that the closer a site is to the GCW, the greater is the reduction in salinity because of the large contrast in salinity between the extracted water and
(e.g., Fig. 9f for run3). The reduction in total salt ranged from 0.5% to where it would be injected. Freshwater floats on the top of the aquifer,
2%. while the salinity increases getting deeper. Thus, the extracted water in
In contrast, placing the GCW at the interface toe (runs 4, 8, 12, 16, run 2 had lower salinity than runs 6 and 10, resulting in the most
18, and 24) resulted in an insignificant reduction in aquifer salinity, dilution effect and flushing out of the saline groundwater. In run 2, the
ranging from 0 to 0.72%. The repulsion for all the isolines ranged from reduction in aquifer salinity and 50% isoline repulsion were 6.1% and
0 to 2.1%, with higher values in low salinity isolines. No inland 16.3%, respectively; while they were 4.4% and 15.8% for run 6, and
migration was observed for all the isolines (e.g., Fig. 9g for run4). 1.4% and 12.3% for run 10. In runs 14 and 22, where the extraction well
These simulation runs illustrated that the horizontal location of GCW screen was at − 200 m below the ground surface (near the top of the
has a significant impact on SWI in terms of the total salt mass within the aquifer) and the spacing to the injection well screen was at 600 m
aquifer and the intrusion length. The significant decrease in both com­ (elevation − 900 m), and 300 m (elevation − 500 m) respectively, the
parison points, in any subregion of the model domain, was strongly achieved SWI reduction in run 14 (4.6%) was less than that of run 2,
controlled by its distance from the GCW. Therefore, the chosen evalu­ whereas, in run 22 it was much less (1.9%) (Fig. 8). Although the
ation criteria strongly influence the mitigation strategy. The optimal spacing between the screens was the same in runs 18 and 22, and in runs
location for pushing back one of the isolines might not be optimal for 6 and 14, the SWI retreated more when the salinity difference between
repelling the others. This is similar to the finding reported by Ebeling the saline groundwater and the injected water across the circulating
et al. (2019). The present study adopted the criterion of pushing back flow was higher. The initial average salinity differences between both
the 17500 mg/l isoline, and the optimum location was found to be extraction well and injection well screen locations for runs 18 and 22
midway through the interface toe and the coast (within the saline were 5768 mg/l and 904 mg/l respectively, while for runs 6 and 14 they
wedge). were 8430 mg/l and 6672 mg/l (Fig. 8). The salinity difference between
On the other hand, the vertical spacing between the extraction and extracted and injected water and their locations play a crucial role in
injection well screens and their locations varied to investigate their ef­ reducing SWI.
fects on the Inj-GCW measure with the same flow rates as previous runs Additional runs were conducted with varying circulating flows to
and by placing GCW at 0.5 Xtoe. Runs 2, 6, and 10 considered placing the further investigate the effect on intrusion lengths. Runs 25 to 48, 49 to
injection well screen at − 1000 m below the ground surface (at the 72, and 73 to 96 used the same configurations as runs 1 to 24, but with
bottom of the aquifer), and the extraction well screen at separating 75%, 50%, and 25% of the circulating flow, respectively. Fig. 10 shows
distances of 800 m (elevation − 200 m), 500 m (elevation − 500 m), and the intrusion lengths of 1000, 2000, 10000, 17500, and 35000 mg/l

12
S. Saad et al. Journal of Hydrology 626 (2023) 130139

Fig. 10. Intrusion length of low and high salinity isolines for runs 2,6,10,14,18 &22 and their corresponding runs considering different circulating flows. The GCW
location was fixed within the saltwater wedge at a distance of 0.5 Xtoe.

isolines with GCW placed at 0.5 Xtoe and varying circulating flows of repulsion of 1000 mg/l, 2000 mg/l, 10000 mg/l, 17500 mg/l, and
12500, 25000, 37500, and 50000 m3/day per well. For runs 2, 6, and 14, 35000 mg/l isolines decreased by 30%, 26.5%, 21%, 33%, and 6.6%,
and their corresponding runs increasing the discharge resulted in more respectively. Additionally, the total salt mass increased by 22.2%. This
retreats for the 10000 mg/l and 17500 mg/l isolines. However, the in­ deterioration may be due to the reduced ability of the measure to reverse
crease in repulsion was disproportionately impacted by higher circu­ back the overall flow direction towards the sea and relieve regional
lating flow rates (Fig. 10). For instance, the repulsion of the 10000 mg/l water deficiency.
isoline increased from 4.7% to 11.1% for runs 2 and their corresponding Based on our results the optimal location for GCWs is where there is a
runs, from 3.3% to 10.3% for runs 6 and their corresponding runs, and significant difference in salt concentration between the injection and
from 1.3% to 8.7% for runs 14 and their corresponding runs, when the extraction well screens. This concentration difference leads to improved
flow rate increased from 25% to 100%. Similarly, for the 17500 mg/l mixing efficiency, resulting in a seaward retreat of the saltwater-
isoline, the repulsion increased from 11.4% to 16.25%, from 8.4% to freshwater interface and a reduction in the overall salt mass. We have
16.6%, and from 7.2% to 14.8%, respectively. The rest of the isolines found that a higher salinity difference yields a greater reduction in
showed insignificant improvement with varying flows. Regarding runs saltwater intrusion (SWI). However, injecting water into a saline wedge
10, 18, 22, and their corresponding runs in the other sets, they did not to achieve a high salinity difference may result in inland migration of
show a substantial difference in the intrusion length with different flows saltwater and increased salinity in some areas. Nonetheless, we have
(Fig. 10). This can be attributed to the short spacing between the in­ observed a significant decrease in salinity near the location of the GCW.
jection well and extraction well screens (100 m, 300 m, and 400 m), Our study has also shown that increasing the circulation flow may
which resulted in short-circuiting. Short-circuiting occurs when there is enhance SWI reduction, but the relationship is not strictly proportional.
a direct flow from the injection interval of the higher hydraulic head to Similarly, a higher injection rate results in a greater repulsive force on
the extraction interval of the lower hydraulic head without flowing the saltwater wedge, which pushes it back toward the sea. Therefore, it
radially away from the well, thus limiting the area of effectiveness. is essential to consider the injection rate, circulating flow, GCW location,
To investigate the influence of the injection rate on the effectiveness and the location of injection and extraction well screens, as well as the
of the Inj-GCW measure, the injection rate was decreased to 32000 m3/ salinity difference when designing and implementing the Inj-GCW
day in run 97 using the same configuration as run 2, which was the most measure for controlling SWI.
effective mitigation measure. The measure’s ability to control SWI Overall, our findings provide valuable insights into the potential
significantly deteriorated compared to run 2, as shown in Fig. 9h. The effectiveness of GCW as a mitigation measure for SWI and underscore

13
S. Saad et al. Journal of Hydrology 626 (2023) 130139

the importance of carefully evaluating and selecting the optimal GCW It was found that a significant deterioration of the desalinization
location and operating parameters to achieve effective SWI reduction efficiency occurred for deep penetrating wells (injected at the bottom of
while minimizing any potential adverse effects. the aquifer). This can be attributed to two factors. First, the hydraulic
gradient produced by the high rate of injection caused the migration of
3.3.2. Controlling SWI in NDA using injection wells as hydraulic barriers brine and high-salinity water from the bottom of the aquifer, leading to
The potential impact of climate change and population growth on the an increase in salinity. The salinity at the bottom of the aquifer, prior to
NDA system and freshwater availability has been studied, with a focus the implementation of the injection scheme, ranged from about 500 mg/
on mitigating the expected increase in salinization. The study proposed l to 108,500 mg/l, which was mainly caused by seawater intrusion,
injecting high-quality treated wastewater and using groundwater con­ downward seepage mass flux, and the upward transport of hypersaline
trol wells (GCWs) to control SWI. groundwater that was assumed to be originated from the compaction
To assess the efficacy of the injection scheme alone, the study (Van Engelen et al., 2018; Van Engelen et al., 2019). Second, the in­
examined its impact on the salinity distribution in three vertical cross- jection wells were located far from the position of the toe which caused
sections of the aquifer system (Figs. 11-13). As a baseline, the result­ lateral saline water inland migration. Al-Yaqoubi et al. (2021) also
ing aquifer salinization based on the future scenario, without imple­ observed in their physical experiments that shallow injection was more
menting the management strategy, was considered (Section 3.2). The effective in reducing salinity and cleaning salts from the upper zone of
injection scheme successfully pushed back the average intrusion of the the saline wedge than deep injection, especially when the injection was
17500 mg/l isoline towards the sea, reducing it from 48.9 km (baseline placed far from the original interface toe, near the freshwater and
scenario) to 45.6 km. Additionally, other isolines were also repulsed by seawater boundaries of the sand tank.
distances of 9.3 km, 15.1 km, 7.1 km, and 2.2 km for 1000 mg/l, 2000 The efficiency of the injection scheme decreased as the injection rate
mg/l, 10000 mg/l, and 35000 mg/l isolines, respectively. The total salt increased, despite the reduction in SWI into the aquifer. This decrease in
mass decreased from 7.39 × 1010 to 7.03 × 1010 tons, resulting in an efficiency can be attributed to the loss of groundwater flowing to either
overall decrease of 6.89% in salinity concentration relative to the the sea (Russo et al., 2015) or inland (Al-Yaqoubi et al., 2021). We used
baseline concentrations of all model nodes. However, a small percentage Eq (4) to quantify the efficiency and found that doubling and halving the
increase in salinity (0.15%) was observed, possibly due to the pressure injection rates led to a decrease and increase in efficiency by 0.29 and
increase induced by the injected water, resulting in the inland flow of 0.92, respectively. Our injection scheme was compared with previous
relatively high salinity water. This issue can be resolved through studies that applied the same scheme but with different configurations
continuous groundwater extraction from the wellfields. The overall and injection rates, focusing on parts of the aquifer. Our results showed
hydraulic gradient towards the sea can also gradually remove residual good agreement with previous studies, with an estimated efficiency of
saltwater, although this process may take longer than the intrusion time 1.97 in our study, 2.31 in Abd-Elaty et al. (2021), and 1.89 in Mabrouk
(Gao et al., 2021). et al. (2019a). This indicates that the injection scheme alone is efficient
Previous studies using physical or hydraulic barriers have also compared to other previous studies, providing a fair comparison when
observed similar behavior of the high-salinity water inland migration combined with groundwater circulation wells (GCWs) as a subsequent
(Ebeling et al., 2019; Gao et al., 2021; Saad et al., 2022). While injecting measure (Inj-GCW measure).
water directly at the interface toe or inland between the interface toe
L17500red × Ared
and the extraction wells would have been more effective in controlling E= (4)
T yrs × rinj
SWI (Al-Yaqoubi et al., 2021; Ebeling et al., 2019), it was not feasible in
the present study due to the high cost and technical difficulties in where, E: Injection scheme efficiency, L17500red is the average reduc­
conveying treated wastewater over long distances from its source to the tion of the 50% isochlor intrusion length, Ared is the average area
injection wells. reduced by the repulsion,Tyrs is the management period in years, and rinj
To determine the locations of injection wells, several factors were is the annual injection rate.
considered (Fig. 14): 1) the boundary of each administrative district was
considered to reduce the inconvenience and high cost of water 3.3.3. Controlling SWI in NDA using Inj-GCW measure
conveyance from its wastewater treatment plants to the injection wells In the previous section, it was shown that injection of treated
(El-Rawy et al., 2019), 2) injecting closer to the toe of the SWI interface wastewater has the potential to displace brackish and saline water,
to effectively act as hydraulic barriers controlling the SWI, 3) injecting thereby increasing the availability of freshwater. However, the feasi­
the aquifer at the moderate hydraulic conductivity zone (≥8.64 m/day) bility of this approach is limited by water availability. Alternatively,
to store the water, thus the water can be readily available (Alam et al., using GCW only (aligned spatially with the 17500 mg/l isoline) was able
2021) (Fig. 6b), 4) injecting in areas with a relatively thicker vadose to push back the 2000 mg/l and 17500 mg/l isolines by 9.1 km and 0.33
zone to provide significant volume for the groundwater storage km towards the sea, respectively. However, approximately 0.2 km
(Mohammad-Hosseinpour and Molina, 2022), and 5) injecting near the further intrusion was observed for the 1000 mg/l and 10000 mg/l iso­
irrigated lands to relieve the stress caused by over-exploitation of the lines compared to the no management scenario. The 35000 mg/l isoline
irrigation wells (Bachtouli and Comte, 2019). and the total salt mass within the aquifer remained unaffected.
To evaluate the impact of well screen location on the efficiency of the Comparing the GCW scheme to injection scheme, no further improve­
injection scheme, different injection depths were assessed. The locations ment in terms of the aquifer salinity and the repulsion of the isolines was
of the well screens were varied between an average depth of 60 m observed (Figs. 11- 13). To address the constraints of the injection
ranging from − 10 m to − 75 m below the ground surface and 500 m scheme and enhance GCW performance, GCWs are proposed as a com­
ranging from − 200 m to − 900 m. Simulations were performed with the plementary measure to be employed in conjunction with the injection
same horizontal location and injection rates, and the intrusion lengths of scheme (Inj-GCW measure).
1000, 2000, 10000, 17500, and 35000 mg/l isolines were determined. The effectiveness of the Inj-GCW in controlling SWI is highly
The results showed that the efficiency of the injection scheme was dependent on the salinity difference between the extracted and injected
significantly enhanced for shallow injection depths, as depicted in water in the circulating flow. Fig. 14 illustrates the initial salinity dif­
Figs. 11- 13. The intrusion lengths of all the considered isolines ference between the extracted water at the top of the Pleistocene layer
increased with increasing injection depth. Specifically, for an increase in and the water that would be injected at the bottom of the layer. At the
injection depth from 60 m to 500 m, the intrusion length of 1000, 2000, locations of the extraction well screens, the salinity ranged from 547
10000, 17500, and 35000 mg/l isolines increased by 14.1, 10.4, 7.4, 5.1, mg/l to 14,700 mg/l, with an average of 4,280 mg/l. While, at the
and 1.8 km, respectively.

14
S. Saad et al. Journal of Hydrology 626 (2023) 130139

Fig. 11. Salinity distribution at the east vertical section presented in Fig. 14 (section 2) and the velocity vectors (arrows show the overall direction and bullet shape
tracks the direction). (a) No management applied, (b) Shallow injection, (c) Deep injection, (d) GCW, (e) Inj-GCW measure. The GCW location is indicated by dashed-
line rectangle.

15
S. Saad et al. Journal of Hydrology 626 (2023) 130139

nodes in the salinity was 0.21%, while the increase was by 0.05%.
The Inj-GCW measure significantly improved the injection scheme,
leading to an 8.9% increase in the repulsion of the 17500 mg/l isoline
and a 5.2% reduction in aquifer salinity compared to the no manage­
ment scenario. This surpassed the improvements achieved using the
injection scheme alone, which resulted in a 6.7% increase in repulsion
and a 4.9% reduction in salinity. In addition, Inj-GCW significantly
reduced the salinity concentration distribution compared to no man­
agement (Fig. 2c and 2d), demonstrating its potential as a promising
strategy for improving the performance of the injection scheme and
controlling SWI. Notably, Inj-GCW outperformed other measures
implemented by Abd-Elaty et al. (2021), including a combined measure
of Treatment and Recharge of wastewater, Abstraction, and Desalination
of brackish water (TRAD) along with pumping reduction, which resulted
in only a 1.8% higher repulsion rate compared to injection alone in
controlling SWI in the NDA. While Inj-GWC resulted in a 2.2% increase.
These findings suggest that Inj-GWC is a highly effective method for
controlling SWI in the NDA, outperforming other measures.
A quantitative evaluation was conducted to determine the impact of
the circulating flow on SWI repulsion in NDA using the Inj-GCW measure
while fixing the injection amount. Our findings indicate that doubling
the circulating flow (to around 20000 m3/day per each well) exacer­
bated the inland migration of higher salinity water (≥10000 mg/l),
resulting in a further intrusion of all isolines except 10000 mg/l. The
10000 mg/l isoline retreated slightly by 50 m. On the other hand,
halving the circulating flow decreased the negative impact of the inland
migration, resulting in a maximum intrusion of 223 m, while the
repulsion of the 17500 mg/l isoline significantly decreased to 643 m.
Therefore, further research should focus on optimizing the circulating
flow to minimize intrusion while also reducing the adverse impact of
inland migration of brackish water. Various approaches for groundwater
management optimization, as described in the literature (e.g., Abarca
et al., 2006; Saad et al., 2022), can be adopted and implemented.

3.4. Investigating the mechanism of Inj-GCW measure as a hydraulic SWI


barrier

Fig. 12. Salinity distribution at the middle vertical section presented in Fig. 14 The formation of a hydraulic barrier to repel SWI, including the use
(section 1) and the velocity vectors (arrows show the overall direction and of GCWs, relies on two key factors: the significant difference in salinity
bullet shape tracks the direction). (a) No management applied, (b) Shallow between the extracted water and the ambient groundwater at the in­
injection, (c) Deep injection, (d) GCW only, (e) Inj-GCW measure. The GCW jection well screen of the GCW, and the enhanced velocity induced by
location is indicated by dashed-line rectangle. the circulating flow. These factors give rise to three observed phenom­
ena: (1) the creation of a bubble of fresh/brackish water within the
injection well screen locations, the salinity varied from 14,372 mg/l to higher saline water, (2) the dilution of the salinized groundwater due to
22,713 mg/l, with an average value of 17,928 mg/l. The amount of the difference in salinity across the circulating flow, and (3) the change
improvement achieved due to the dilution effect and increase in the of the groundwater flow path near the GCW due to the enhanced
velocity in the proximity of the GCW was higher using the Inj-GCW velocity.
measure than using the GCW measure only (Figs. 11 - 13). The higher The bubble of fresh/brackish water gradually expanded upward and
effectiveness of the Inj-GCW measure can be attributed to the induced connected to the upper zones of lower salinity (Figs. 11 - 13), as brackish
seaward fluxes. The seaward fluxes induced by the injection scheme play water was injected into the saline water through each GCW. This process
a crucial role in the hydrodynamics of SWI. This agrees with the findings displaced the higher salinity water in all directions, effectively building
of previous studies (Abdoulhalik et al., 2017; Armanuos et al., 2020; a hydraulic barrier against intrusion. Our findings align with those of Al-
Chang et al., 2021) on the influence of seaward fluxes. Yaqoubi et al. (2021), who experimentally observed similar dynamics of
The salinity distributions at three vertical sections are presented in the SWI interface when freshwater was injected near the saline water
Figs. 11 -13, to demonstrate the efficacy of the Inj-GCW measure in boundary. A bubble was created, which was topologically isolated from
repelling the 17500 mg/l isoline by 1.05 km beyond what the injection the fresh groundwater, retarded the intrusion of the saline water when
scheme alone achieves. Additionally, the 2000 mg/l isoline was slightly the injection stopped by a time of around 500% longer than the other
pushed back (by 52 m), while no significant improvement was observed scenarios where no bubble was formed. They stopped the injection to
for the 35000 mg/l isoline. However, the implementation of the Inj- mimic the cyclic managed aquifer recharge schemes experimentally.
GCW measure resulted in the inland migration of the 1000 mg/l and Additionally, our study supports the findings of Zang and Li (2021), who
10000 mg/l isolines by 152 m and 324 m, respectively. The maximum suggested that injecting compressed air as an alternative to freshwater
repulsion was observed for the 17500 mg/l isoline because the GCW was also depends on forming a bubble due to the air’s lower density that can
located just upstream of this isoline before implementing the manage­ create a desaturated zone within the aquifer, impeding the SWI.
ment strategy. The total salt mass was reduced from 7.03 × 1010 to 7.0 The dilution of the groundwater in the lower part of the GCW is
× 1010 tons. The percent of the overall further decrease across all model facilitated by the circulating flow. The abstracted water from the upper
part of the GCW is injected and mixed with the ambient groundwater

16
S. Saad et al. Journal of Hydrology 626 (2023) 130139

Fig. 13. Salinity distribution in the west vertical section presented in Fig. 14 (section3) and the velocity vectors (arrows show the overall direction and bullet shape
tracks the direction). (a) No management applied, (b) Shallow injection, (c) Deep injection, (d) GCW only, (e) Inj-GCw measure. The GCW location is indicated by
dashed-line rectangle.

17
S. Saad et al. Journal of Hydrology 626 (2023) 130139

Fig. 14. Spatial locations of extraction well screen and injection well screen of GCW, and injection wells. (a) Salinity distribution at an average depth of − 60 m
where the injection screens and the extraction screen of GCW were located. (b) Salinity distribution at an average depth of − 350 m where the injection screen of GCW
was located.

near the injection well screen. The mixing ratio between the regional to dilute the salinized groundwater and reduce its salinity, further
flow, the injected water by the GCW, and the ambient groundwater at impeding the progress of SWI. This mechanism is similar to that
the injection well screen, defined the salinity concentrations near and explained by Li et al. (2021) in their study of assessing the performance
seaward the GCWs. This mixing is mainly dominated by forced con­ of aquifer storage and recovery scheme by injecting freshwater into a
vection due to the injection and free convection resulting from the un­ layered saline aquifer.
stable density stratification when less saline water is injected into the Additionally, the installation of the GCWs changed the flow path of
higher salinity water (the formed bubble). The mixing mechanism helps the groundwater near the wells to a circulating pattern, increasing the

18
S. Saad et al. Journal of Hydrology 626 (2023) 130139

regional flow velocity and consequently creating a higher hydraulic fold increase in groundwater abstractions, the intrusion is expected
gradient. The high velocity at the injection well screens repulsed the SWI to extend another 12.4 km by year 2100. Therefore, implementing a
by affecting the lower part of the mixing zone close to these wells, while sustainable management strategy is imperative.
the upper part remained largely unchanged (compared to that using the 2) The location of GCWs relative to the interface toe of the saltwater
injection scheme only) (Figs. 11 -13). wedge had a significant impact on intrusion reduction and aquifer
The formation of the bubble, the dilution effect induced by the salinity. Placing GCWs near the toe pushed back low salinity isolines
circulating flow, and the enhanced velocity can delay the progress of the due to increased velocity induced by the circulating flow. However,
SWI. However, the lateral retreat of the saltwater wedge is commonly circulating flow within the saline wedge repelled high salinity iso­
dominated by the advective forces driven by the seaward fluxes (Motz lines and resulted in a greater aquifer salinity decrease. This was due
and Sedighi, 2013). The regional flow in NDA was insufficient to turn to the dilution effect from the large salinity contrast between the
the overall flow direction towards the sea, as can be seen from the ve­ extracted water and the ambient groundwater at the GCW’s injection
locity vectors in Figs. 11 - 13. Even though the Inj-GCW measure suc­ well screen, as well as the velocity increase due to flow circulation.
ceeded to achieve repulsion of the SWI and reduction to the total salt Efficiency was enhanced when GCWs were placed within the saline
mass, the system of NDA was still dominated by the density-driven flux wedge, with extraction at the top and injection at the bottom of the
under the combined effect of the worst-case climate change scenario aquifer.
(RCP 8.5) and projected population growth. By the end of the man­ 3) Placing the injection well screen of the GCW within the saline wedge
agement period, the overall groundwater flow was still flowing inland resulted in the formation of a fresh/brackish water bubble. This
(from north to south) (Figs. 11-13). The study observed that a decrease bubble displaced the saline water as it grew and acted as a hydraulic
in the injection rate of the Inj-GCW measure led to a change in the barrier.
overall groundwater flow direction from seaward to inland, as seen in 4) The retreat of isolines near the GCW is positively correlated with the
the illustrative case study (run 2 and run 97). The decrease of 42%, circulating flow rate, but it can lead to inland migration of saline
negatively affected the effectiveness of the Inj-GCW measure, reducing water upstream of the GCW. However, short-circuiting may occur
the repulsion of 17500 mg/l isoline by 33% and the aquifer salinity by due to the short spacing between injection well and extraction well
22.2% (Fig. 9). The difference in improvement achieved in controlling screens, resulting in insignificant differences in the retreat of isolines
the SWI between the two runs can be attributed to the magnitude and with varying flows. To determine the optimal design variables, such
direction of the regional flow, highlighting the importance of the as circulating flow, GCW locations, and screen placement, a coupled
regional freshwater fluxes in determining the extent of the SWI and the simulation–optimization model with objective functions of minimum
salinization of the aquifer, and in enhancing the workability of the total cost and maximum reduction in saltwater intrusion (SWI) is
mitigation measure. necessary.
The regional freshwater flux is affected by the over-exploitation of 5) The efficiency of shallow injection in repulsing saline water intrusion
groundwater and sea level rise. Mitigation strategies such as altering and reducing salt mass was higher than deep injection. The injection
crop patterns and using modern irrigation techniques, particularly in wells were distributed across the entire model domain, not just near
water-scarce regions, can complement the Inj-GWC measure in reducing the interface toe for sewage treatment plant proximity. Conse­
their adverse impacts (e.g., Mabrouk et al., 2019a). Additionally, con­ quently, injecting in less saline upper layers mitigated adverse im­
trolling deforestation and adopting clean and renewable sources of en­ pacts of inland migration of saline water.
ergy can mitigate greenhouse gas emissions and reduce sea level rise (e. 6) The hydrodynamic mechanism of the Inj-GCW measure relies on
g., Yoro and Daramola, 2020). Coastal protection structures can also be several key factors. Firstly, the seaward freshwater flux induced by
constructed to mitigate SWI (e.g., Le Xuan et al., 2022). Further studies artificial recharge creates a repulsive force that acts on the entire
are required to assess the effectiveness of combining these measures saltwater wedge. Secondly, the relatively high velocity at the injec­
with Inj-GCW in mitigating SWI in the NDA. tion well screen results from the circulating flow, which increases the
Additional research on biogeochemical processes in NDA is essential hydraulic head at the bottom of the GCW and alters the salinity
as clogging is inevitable in any artificial recharge scheme during its distribution in its proximity. Lastly, the salinity difference across the
operational life including inj-GCW measure. However, effective design circulating flow also plays a significant role.
addressing the biogeochemical processes (Dillon et al., 2020) and reg­ 7) The effectiveness of the Inj-GCW measure for SWI control increases
ular maintenance can help alleviate this concern. as the seaward hydraulic gradient becomes steeper.
8) The Inj-GCW measure demonstrated superior effectiveness in
4. Conclusions retreating the SWI in the NDA compared to other measures,
including injection scheme only, GCW measure only, and a combi­
In this study, we examined the potential impacts of sea level rise and nation of injecting treated wastewater and abstracting brackish
population growth on SWI in the NDA, which could lead to intensified water (TRAD measure), as well as reducing regional abstractions (in
salinization without appropriate intervention. To address this issue, we previous studies identified as the most effective measure). This em­
proposed a new mitigation measure called Inj-GCW, which involves phasizes that the circulating flow with no net loss or gain of the
injecting the available treated wastewater and circulating groundwater groundwater can alleviate the potential problem of abstracting
near the initial 50% isochlor (17500 mg/l isoline) to promote sustain­ freshwater, and enhance the feasibility of the injection scheme
able management of the coastal aquifer. Our analysis relied on a cali­ considering water availability constraints.
brated 3D transient model of density-dependent groundwater flow and
salt transport in the NDA. The main conclusions from the results of the In summary, the study has demonstrated the hydraulic effectiveness
simulation model and the implementation of the Inj-GCW mitigation of the Inj-GCW measure in mitigating SWI in the NDA and has high­
measure are as follows: lighted the need for further research to optimize its design and assess its
economic feasibility in comparison to combining injection with other
1) The NDA is severely deteriorating due to salinization from SWI, hydraulic or physical barriers. These findings contribute to the under­
hypersaline groundwater flow, and mass fluxes from various sources standing of SWI mitigation measures and support the implementation of
such as river branches, water networks, and irrigation return flows. sustainable management strategies for coastal aquifers facing saliniza­
This has resulted in the intrusion of the 50% isochlor up to an tion challenges.
average distance of 36 km from the shoreline. Furthermore, under
the future scenario of a gradual sea level rise of 0.45 m and a three-

19
S. Saad et al. Journal of Hydrology 626 (2023) 130139

CRediT authorship contribution statement Chang, Q., Zheng, T., Chen, Y., Zheng, X., Walther, M., 2021. Influence of inland
freshwater influx on the natural desalination of coastal aquifers with a cutoff wall.
Desalination 499, 114863. https://doi.org/10.1016/j.desal.2020.114863.
Samia Saad: Conceptualization, Methodology, Software, Formal Dibaj, M., Javadi, A.A., Akrami, M., Ke, K.-Y., Farmani, R., Tan, Y.-C., Chen, A.S., 2021.
analysis, Investigation, Validation, Resources, Visualization, Writing – Coupled three-dimensional modelling of groundwater-surface water interactions for
original draft, Writing – review & editing, Funding acquisition. Akbar A. management of seawater intrusion in Pingtung Plain, Taiwan. J. Hydrol. Regl. Stud.
36, 100850.
Javadi: Supervision, Conceptualization, Writing – review & editing. Diersch, H.-J.-G., 2013. FEFLOW: finite element modeling of flow, mass and heat
Hany F. Abd-Elhamid: Visualization, Writing – review & editing. transport in porous and fractured media. Springer Science & Business Media.
Raziyeh Farmani: Supervision, Writing – review & editing. Dillon, P., Fernández Escalante, E., Megdal, S.B., Massmann, G., 2020. Managed aquifer
recharge for water resilience. MDPI 12 (7), 1846.
Ebeling, P., Händel, F., Walther, M., 2019. Potential of mixed hydraulic barriers to
remediate seawater intrusion. Sci. Total Environ. 693, 133478 https://doi.org/
Declaration of Competing Interest 10.1016/j.scitotenv.2019.07.284.
Elmore, A.C., Hellman, J.B., 2001. Model-predicted groundwater circulation well
performance. Practice Periodical of Hazardous, Toxic, and Radioactive Waste
The authors declare that they have no known competing financial
Management, 5(4): 203-210.
interests or personal relationships that could have appeared to influence El-Rawy, M., Al-Maktoumi, A., Zekri, S., Abdalla, O., Al-Abri, R., 2019. Hydrological and
the work reported in this paper. economic feasibility of mitigating a stressed coastal aquifer using managed aquifer
recharge: a case study of Jamma aquifer, Oman. J. Arid Land 11 (1), 148–159.
Elshinnawy, H., Zeidan, B., Ghoraba, S., 2015. Impact of hydraulic conductivity
Data availability uniformity on seawater intrusion in the Nile Delta aquifer, Egypt.
Eltarabily, M.G.A., Negm, A.M., 2017. Groundwater management for sustainable
Data will be made available on request. development east of the nile delta aquifer. Groundwater in the Nile Delta 687–708.
Eltarabily, M.G.A., Negm, A.M., 2018. Groundwater management for sustainable
development plans for the Western Nile Delta. Groundwater in the Nile Delta
Acknowledgments 709–727.
Fang, Y., Chen, X., Gomez Velez, J., Zhang, X., Duan, Z., Hammond, G.E., Goldman, A.E.,
Garayburu-Caruso, V.A., Graham, E.B., 2020. A multirate mass transfer model to
The first author is funded by the Ministry of Higher Education of the represent the interaction of multicomponent biogeochemical processes between
Arab Republic of Egypt, Netwon Mosharafa scholarship [ID: NMM26/ surface water and hyporheic zones (SWAT-MRMT-R 1.0). Geosci. Model Dev. 13 (8),
17]. The authors would like to express their gratitude to the DHI group 3553–3569.
Ganot, Y., Holtzman, R., Weisbrod, N., Russak, A., Katz, Y., Kurtzman, D., 2018.
for providing the free license of FEFLOW. The authors would also like to Geochemical processes during managed aquifer recharge with desalinated seawater.
thank Mohammed Marzouk, research assistant at Nile research institute, Water Resour. Res. 54 (2), 978–994.
National water research center, Egypt for discussions on seawater Gao, M., Zheng, T., Chang, Q., Zheng, X., Walther, M., 2021. Effects of mixed physical
barrier on residual saltwater removal and groundwater discharge in coastal aquifers.
intrusion modeling of the Nile Delta aquifer. Hydrol. Process. 35 (7), e14263.
Gelhar, L.W., Welty, C., Rehfeldt, K.R., 1992. A critical review of data on field-scale
dispersion in aquifers. Water Resour. Res. 28 (7), 1955–1974.
Appendix A. Supplementary data
Hussain, M.S., Abd-Elhamid, H.F., Javadi, A.A., Sherif, M.M., 2019. Management of
seawater intrusion in coastal aquifers: a review. Water 11 (12), 2467.
Supplementary data to this article can be found online at https://doi. Idowu, T.E., Lasisi, K.H., 2020. Seawater intrusion in the coastal aquifers of East and
org/10.1016/j.jhydrol.2023.130139. Horn of Africa: a review from a regional perspective. Scientific African 8, e00402.
Jiao, J., Post, V., 2019. Coastal hydrogeology. Cambridge University Press.
Kacimov, A., Sherif, M., Perret, J., Al-Mushikhi, A., 2009. Control of sea-water intrusion
References by salt-water pumping: Coast of Oman. Hydrogeol. J., 17(3): 541-558.
Ketabchi, H., Mahmoodzadeh, D., Ataie-Ashtiani, B., Simmons, C.T., 2016. Sea-level rise
impacts on seawater intrusion in coastal aquifers: Review and integration. J. Hydrol.
Abarca, E., Vázquez-Suñé, E., Carrera, J., Capino, B., Gámez, D., Batlle, F., 2006. Optimal
535, 235–255.
design of measures to correct seawater intrusion. Water Resour. Res. 42 (9).
Le Xuan, T.u., Ba, H.T., Thanh, V.Q., Wright, D.P., Hasan Tanim, A., Tran Anh, D., 2022.
Abd-Elaty, I., Javadi, A.A., Abd-Elhamid, H., 2021. Management of saltwater intrusion in
Evaluation of coastal protection strategies and proposing multiple lines of defense
coastal aquifers using different wells systems: A case study of the Nile Delta aquifer
under climate change in the Mekong Delta for sustainable shoreline protection.
in Egypt. Hydrogeol. J., 29(5): 1767-1783.
Ocean Coast. Manag. 228, 106301.
Abd-Elaty, I., Zelenakova, M., 2022. Saltwater intrusion management in shallow and
Li, H., Ye, Y., Lu, C., 2021. Aquifer Storage and Recovery in Layered Saline Aquifers:
deep coastal aquifers for high aridity regions. J. Hydrol.: Reg. Stud. 40, 101026.
Importance of Layer-Arrangements. Water 13 (18), 2595.
Abd-Elhamid, H., Abdelaty, I., Sherif, M., 2019. Evaluation of potential impact of Grand
Mabrouk, M., Jonoski, A., Oude Essink, G.H., 2019a. Regional Groundwater Modelling
Ethiopian Renaissance Dam on seawater intrusion in the Nile Delta aquifer. Int. J.
For Determining Adaptation Strategies In The Nile Delta Aquifer. E-Proceedings of
Environ. Sci. Technol. 16 (5), 2321–2332.
the 38th IAHR World Congress, Panama City, Panama. 10.3850/38WC092019-1397.
Abd-Elhamid, H.F., Abd-Elaty, I., Hussain, M.S., 2020. Mitigation of seawater intrusion
Mabrouk, M., Jonoski, A., Oude Essink, H.P., G., Uhlenbrook, S., 2018. Impacts of sea
in coastal aquifers using coastal earth fill considering future sea level rise. Environ.
level rise and groundwater extraction scenarios on fresh groundwater resources in
Sci. Pollut. Res. 27 (18), 23234–23245.
the Nile Delta Governorates. Egypt. Water 10 (11), 1690.
Abdoulhalik, A., Ahmed, A., Hamill, G., 2017. A new physical barrier system for seawater
Mabrouk, M., Jonoski, A., Oude Essink, G.H., Uhlenbrook, S., 2019b. Assessing the
intrusion control. J. Hydrol. 549, 416–427.
fresh–saline groundwater distribution in the Nile delta aquifer using a 3D variable-
Ainscough, T.J., Oatley-Radcliffe, D.L., Barron, A.R., 2021. Groundwater remediation of
density groundwater flow model. Water 11 (9), 1946.
volatile organic compounds using nanofiltration and reverse osmosis membranes—A
Mohammad-Hosseinpour, A., Molina, J.-L., 2022. Improving the Sustainability of Urban
field study. Membranes 11 (1), 61.
Water Management through Innovative Groundwater Recharge System (GRS).
Alam, S., Borthakur, A., Ravi, S., Gebremichael, M., Mohanty, S.K., 2021. Managed
Sustainability 14 (10), 5990.
aquifer recharge implementation criteria to achieve water sustainability. Sci. Total
Motz, L.H., Sedighi, A., 2013. Saltwater Intrusion and Recirculation of Seawater at a
Environ. 768, 144992.
Coastal Boundary. J. Hydrol. Eng. 18 (1), 10–18. https://doi.org/10.1061/(ASCE)
Al-Yaqoubi, S., Al-Maktoumi, A., Kacimov, A., Abdalla, O., Al-Belushi, M., 2021. Fresh-
HE.1943-5584.0000594.
saline water dynamics in coastal aquifers: Sand tank experiments with MAR-wells
Negm, A.M., Sakr, S., Abd-Elaty, I., Abd-Elhamid, H.F., 2018. An overview of
injecting at intermittent regimes. J. Hydrol. 601, 126826.
groundwater resources in Nile Delta aquifer. Groundwater in the Nile Delta 3–44.
Armanuos, A.M., Ibrahim, M.G., Mahmod, W.E., Takemura, J., Yoshimura, C., 2019.
Nofal, E.R., 2016. Seawater Intrusion in Nile Delta in perspective of climate changes and
Analysing the combined effect of barrier wall and freshwater injection
aquifer heterogeneity. Faculty of Engineering, Cairo University, Egypt. Ph.D. Thesis.
countermeasures on controlling saltwater intrusion in unconfined coastal aquifer
Nofal, E., Amer, M., El-Didy, S., Fekry, A., 2015. Sea water intrusion in Nile Delta in
systems. Water Resour. Manag. 33 (4), 1265–1280.
perspective of new configuration of the aquifer heterogeneity using the recent
Armanuos, A.M., Al-Ansari, N., Yaseen, Z.M., 2020. Assessing the Effectiveness of Using
stratigraphy data. J. Am. Sci. 11 (6), 281–292.
Recharge Wells for Controlling the Saltwater Intrusion in Unconfined Coastal
Nofal, E.R., Fekry, A.M., Ahmed, M.H., El-Kharakany, M.M., 2018. Groundwater:
Aquifers with Sloping Beds: Numerical Study. Sustainability 12 (7). https://doi.org/
Extraction versus recharge; vulnerability assessment. Water Sci. 32 (2), 287–300.
10.3390/su12072685.
Page, D., Bekele, E., Vanderzalm, J., Sidhu, J., 2018. Managed aquifer recharge (MAR) in
Bachtouli, S., Comte, J.-C., 2019. Regional-Scale Analysis of the Effect of Managed
sustainable urban water management. Water 10 (3), 239.
Aquifer Recharge on Saltwater Intrusion in Irrigated Coastal Aquifers: Long-Term
Panday, S., Bedekar, V., Langevin, C.D., 2018. Impact of Local Groundwater Flow Model
Groundwater Observations and Model Simulations in NE Tunisia. J. Coast. Res. 35
Errors on Transport and a Practical Solution for the Issue. Groundwater 56 (4),
(1), 91–109.
667–672. https://doi.org/10.1111/gwat.12627.
Barnett, B. et al., 2012. Australian groundwater modelling guidelines. Waterlines report
series, National Water Commission, Canberra: 203.

20
S. Saad et al. Journal of Hydrology 626 (2023) 130139

Pennington, B.T., Sturt, F., Wilson, P., Rowland, J., Brown, A.G., 2017. The fluvial Adv. Water Resour. 144, 103704 https://doi.org/10.1016/j.
evolution of the Holocene Nile Delta. Quaternary Science Reviews 170, 212–231. advwatres.2020.103704.
Pool, M., Carrera, J., 2010. Dynamics of negative hydraulic barriers to prevent seawater Song, J., Yang, Y., Wu, J., Wu, J., Sun, X., Lin, J., 2018. Adaptive surrogate model based
intrusion. Hydrgeol. J. 18 (1), 95–105. multiobjective optimization for coastal aquifer management. J. Hydrol. 561,
Pörtner, H.-O., et al., 2022. Climate change 2022: Impacts, adaptation and vulnerability. 98–111.
IPCC Sixth Assessment Report. Tatti, F., Petrangeli Papini, M., Torretta, V., Mancini, G., Boni, M.R., Viotti, P., 2019.
RIGW, 1992. Hydrogeological map of Egypt: map sheet of Nile Delta. Research Institute Experimental and numerical evaluation of Groundwater Circulation Wells as a
for Groundwater, National Water Research Center, Kanater El Khairia, Egypt. remediation technology for persistent, low permeability contaminant source zones.
Saad, S., Javadi, A.A., Chugh, T., Farmani, R., 2022. Optimal Management of Mixed J. Contam. Hydrol. 222, 89–100.
Hydraulic Barriers in Coastal Aquifers using Multi-objective Bayesian Optimization. Tawfik, M.H., Hoogesteger, J., Elmahdi, A., Hellegers, P., 2021. Unpacking wastewater
J. Hydrol. 612, 128021. reuse arrangements through a new framework: insights from the analysis of Egypt.
Saad, S., Javadi, A.A., Farmani, R., Sherif, M., 2023. Efficient uncertainty quantification Water Int. 46 (4), 605–625.
for seawater intrusion prediction using Optimized sampling and Null Space Monte Torresan, S., Furlan, E., Critto, A., Michetti, M., Marcomini, A., 2020. Egypt’s coastal
Carlo method. J. Hydrol. 620, 129496. vulnerability to sea-level rise and storm surge: Present and future conditions. Integr.
Sefelnasr, A., Sherif, M., 2014. Impacts of seawater rise on seawater intrusion in the Nile Environ. Assess. Manag. 16 (5), 761–772.
Delta aquifer, Egypt. Groundwater 52 (2), 264–276. Van Engelen, J., Essink, G.H.O., Kooi, H., Bierkens, M.F., 2018. On the origins of
Sestini, G., 1989. Nile Delta: a review of depositional environments and geological hypersaline groundwater in the Nile Delta aquifer. J. Hydrol. 560, 301–317.
history. Geological Society, London, Special Publications 41 (1), 99–127. van Engelen, J., Verkaik, J., King, J., Nofal, E.R., Bierkens, M.F.P., Oude Essink, G.H.P.,
Sharan, A., Lal, A., Datta, B., 2021. A review of groundwater sustainability crisis in the 2019. A three-dimensional palaeohydrogeological reconstruction of the groundwater
Pacific Island countries: Challenges and solutions. J. Hydrol. 603, 127165. salinity distribution in the Nile Delta Aquifer. Hydrol. Earth Syst. Sci. 23 (12),
Shen, C., Fan, Y.u., Wang, X., Song, W., Li, L., Lu, C., 2022. Effects of Land Reclamation 5175–5198.
on a Subterranean Estuary. Water Resour. Res. 58 (6). Vats, O.P., Sharma, B., Stamm, J., Bhattacharjya, R.K., 2020. Groundwater circulation
Sherif, M., Sefelnasr, A., Javadi, A., 2012. Incorporating the concept of equivalent well for controlling saltwater intrusion in coastal aquifers: numerical study with
freshwater head in successive horizontal simulations of seawater intrusion in the experimental validation. Water Resour. Manag. 34 (11), 3551–3563.
Nile Delta aquifer, Egypt. J. Hydrol. 464–465, 186–198. https://doi.org/10.1016/j. Vu, D., Yamada, T., Ishidaira, H., 2018. Assessing the impact of sea level rise due to
jhydrol.2012.07.007. climate change on seawater intrusion in Mekong Delta, Vietnam. Water Sci. Technol.
Sherif, M., Sefelnasr, A., Ebraheem, A.A., Javadi, A., 2014. Quantitative and qualitative 77 (6), 1632–1639.
assessment of seawater intrusion in Wadi Ham under different pumping scenarios. Yin, J., Tsai, F.-T.-C., 2019. Steady-state approximate freshwater–saltwater interface in a
J. Hydrol. Eng. 19 (5), 855–866. two-horizontal-well scavenging system. J. Hydrol. Eng. 24 (10), 06019008.
Sherif, M., 2003. Assessment, modeling and management of seawater intrusion in the Yoro, K.O., Daramola, M.O., 2020. Chapter 1 - CO2 emission sources, greenhouse gases,
Nile delta aquifer. Tecnología De La Intrusión De Agua De Mar En Acuíferos and the global warming effect. In: Rahimpour, M.R., Farsi, M., Makarem, M.A.
Costeros: PAÍSES MEDITERRÁNEOS© IGME, Madrid. (Eds.), Advances in Carbon Capture. Woodhead Publishing, pp. 3–28. https://doi.
Shi, L., Lu, C., Ye, Y., Xie, Y., Wu, J., 2020. Evaluation of the performance of multiple- org/10.1016/B978-0-12-819657-1.00001-3.
well hydraulic barriers on enhancing groundwater extraction in a coastal aquifer. Zang, Y., Li, M., 2021. Numerical assessment of compressed air injection for mitigating
seawater intrusion in a coastal unconfined aquifer. J. Hydrol. 595, 125964.

21

You might also like