During the last century, many large irrigation projects
were carried out in arid lands worldwide. Despite a tremendous increase in
food production, a common problem when characterizing these zones is land
degradation in the form of waterlogging. A clear example of this phenomenon
is in the Nubariya depression in the Western Desert of Egypt. Following the
reclamation of desert lands for agricultural production, an artificial
brackish and contaminated pond started to develop in the late 1990s, which
at present extends for about 2.5 km
The rise of groundwater due to farmland over-irrigation, i.e., waterlogging,
is a widespread occurrence with a large global impact on the
available water resources. This comes with often contrasting effects on the water
quantity, which increases, and the water quality, which deteriorates
(Scanlon et al., 2007).
Waterlogging occurs due to a number of causes, including large percolation losses
from the irrigation water applied to the fields and seepage losses from
channels providing significant localized sources of recharge water. One
example of the groundwater level rising in semiarid regions due to man-made
activities can be found in parts of the Xinjiang Province in western China.
The diversion of river water for irrigation raised the groundwater
levels from more than 7 m below the ground surface in the 1950s to about 1 m below the ground surface in the late 1980s (Sheng and Xiuling,
2001). Another example is in the Thar Desert of India, running along the
western border with Pakistan. A mean water table rise of 1.1 m year
When over-irrigation is combined with low-permeable soils relatively close to the ground surface, low relief or depressions, and the absence of natural drainage either through surface water or groundwater systems, waterlogging can create anthropogenic perennial inland wetlands even in arid or semiarid regions. These “arid wetlands”, i.e., humid zones in an arid or semiarid climate, which may seem to be a contradiction (Lemly et al., 1993), have been the focus of hydrological research over the last few years. This is because they represent regions of high conservation value, are crucial refuges for native wildlife (fish, amphibians, snails, and plants), and provide a habitat for migratory birds. Worldwide arid wetlands are threatened by increasing anthropogenic pressure, leading to water contamination through the development of agriculture in the surrounding area and/or water shortages due to surface water and groundwater use caused by population growth (Ashley et al., 2004; Li et al., 2015; Minckley et al., 2013).
An unequivocal contradiction arises when wetlands are generated and grow in
desert areas due to the overuse of water for irrigation. In such cases, the
wetlands reflect the consequences of extremely inefficient water
management with the potential waste of a precious resource. Egypt is one of
the countries where this occurrence is becoming widespread. Since the 1952
revolution, Egypt has tried to increase its agricultural area through the
reclamation of desert land. Land reclamation in the Egyptian context means
converting desert areas into agricultural land and rural settlements primarily
by “adding water”. The canals fed by the Nile River are extended through
existing agricultural areas to supplement the new reclamation zones
(Adriansen, 2009). Hundreds of deep wells have been
drilled to tap the Nubian Sandstone Aquifer System in order to support agricultural
megaprojects developed within the depressions in the middle of the Western
Desert. Excessive pumping, accompanied by a lack of catchment
hydrogeological and geomorphological studies, has caused the formation of
large surficial ponds. An example of this occurs in the Baharia depression
where El Bastawesy et al. (2013)
have shown by remote-sensing analyses that an increase in cultivated land
from 40.9 km
In this paper, attention is focused on the Nubariya depression
(30
In the framework of the EU SWIM IMPROWARE project (Innovative means to protect water
resources in Mediterranean coastal areas through re-injection of treated
water;
Here, based on our previous experience in the use of hydrogeophysical methods
for groundwater characterization
(Behroozmand
et al., 2013; Vilhelmsen et al., 2014), a joint magnetic resonance sounding
(MRS) and ground-based transient EM (TEM) survey was performed on a
With the main goal of reconstructing the past multiyear hydrological
evolution of the area driven by the need for reclamation and its possible
future behavior in relation to artificial aquifer recharge using TWW, two
numerical models were developed: a 3-D groundwater flow model for the entire
depression area (25
In developing countries and many other parts of the world, these reclaimed desert environments are generally quite challenging to access and investigate. Moreover, the development of anthropogenic wetlands as a byproduct of over-irrigation is a process that is generally not well investigated in the hydrological community. Here, we present a unique case study where the evidence of the occurrence is profound. Finally, from a more technical point of view, the advantages of integrating MRS and TEM methods to discriminate between different deposits and groundwater quality are clearly highlighted for the first time, to our knowledge, in wetland environments.
The paper begins with a section that describes the environmental setting and hydrological evolution of the study area. Second, the MRS and TEM methods are briefly introduced, the results from the surveys carried out in Nubariya are summarized, and a hydrogeological model obtained by the joint inversion method is shown. The setup of the depression-scale and local-scale numerical models is presented and the simulation outcomes are shown. The final sections discuss the value of the proposed approach and draw the conclusions.
The study area is located in a complex sedimentary setting along the transition between the Quaternary deposits of the Nile River Delta to the west and the Pliocene and Miocene deposits of the Western Desert to the east (Fig. 1). The area is characterized by low relief with an elevation from 0 to 100 m above the mean sea level (a.m.s.l.). The main landforms developed through the interaction of the geological structures, the processes of wind and surface water erosion, and the climatic conditions, such as temperature and humidity (Dawoud et al., 2005; Hassan et al., 2012; RIGW and IWACO, 1991). Hydrogeological investigations at the regional scale (Sharaky et al., 2008) highlight the presence of loose coarse Miocene sands with clay lenses overlain by Pliocene deposits in the upper 200 m of subsoil. The Pilocene units consist mainly of estuarine clayey facies at the base and fluviomarine and shallow marine limestones at the top. The uppermost units are exposed in the lowest parts of the landscape and their vicinities. Quaternary deltaic sediments from sands to silt dominate in the eastern part of the study area. The Quaternary layer gradually decreases in thickness from the Rosetta branch of the Nile River to almost zero along the linear depression from Nubariya to the north and Wadi El-Natrun southward, where they interdigitate more or less sharply with the Pliocene deposits.
The Nubariya depression represents the northernmost tip of the Wadi El-Natrun elongated depression situated approximately 50 km to the south. Unlike the latter, which has been extensively investigated from the hydrological, hydrogeological, and environmental perspectives over the last few decades due to the presence of alkaline lakes (Atwia et al., 2012; Khalil and Santos, 2013), very little hydrogeological information is available on Nubariya, even though it has been subject to more recent reclamation and development.
With a minimum elevation of about 7 m a.m.s.l., the Nubariya depression is
surrounded by sandy hills with a mean elevation ranging between 30 and 100 m a.m.s.l. (Fig. 2). A digital elevation model (DEM)
of the area has been obtained from the ASTER GDEM, which was calibrated by a
kinematic DGPS in situ survey carried out in 2014. Over the last three
decades, large portions of this desert area have been converted into
productive farmland. A remote-sensing investigation with a series of Landsat
images acquired between 1984 and 2014 clearly shows that a pond started to
develop in 1999 and expanded significantly until the 2014 extent reached
about 2.5 km
A digital elevation model of the Nubariya depression and its surroundings derived from the ASTER GDEM (Advanced Spaceborne Thermal Emission and Reflection Radiometer; Global Digital Elevation Model) (Tachikawa et al., 2011) and calibrated by a kinematic DGPS in situ survey (red alignment). The boundaries of the regional and local models developed in the framework of the IMPROWARE project are shown by the black and white boxes, respectively. The blue lines and associated dates, which separate the desert (to the southwest) from the irrigated areas (to the northeast), indicate how the latter has encroached southwestward over time. This information was obtained form an analysis of Landsat images. The results from the 3-D FE pond-scale groundwater flow model are shown along the black alignment E–E.
Landsat images showing the time evolution of the farmland (blue lines) and the pond in Nubariya between 1984 and 2014. The extent of desert reclamation is illustrated with reference to the red line, which marks the contemporary (2014) extent of the irrigated area. The Landsat images are available from the US Geological Survey.
A photo of the Nubariya pond dated November 2013 showing houses and a road abandoned due to the pond expansion. The photo location and direction are shown in the inset.
Chemical analyses of groundwater samples collected in 2003 (after
Sharaky et al., 2008) and 2011 (after Masoud, 2014) and
groundwater and pond water samples collected in 2014 within the IMPROWARE
project
The artificial pond has grown despite the climatic conditions. The typical
climate that prevails in the area is characterized by high temperatures that
span from a high of 30
Hydrogeological and chemical investigations were carried out to collect information about the link between the subsurface lithostratigraphy and the water quality in the pond, thus providing specific ancillary data for the calibration of geophysical measurements.
With this aim, five boreholes (Fig. 5) were drilled with continuous core sampling and pumping and recovery tests. The depth of the drilling ranges between 25 and 55 m from the ground surface. Moreover, a number of slug tests were carried out to characterize the infiltration rate, and hence the soil type, of the uppermost deposits.
The location map of the boreholes (red triangles) and the TEM (blue dots) and MRS (green squares) measurements carried out in the surrounding area of the Nubariya pond. The black line represents the trace of the cross section through the resistivity model shown in Fig. 6. The location of MRS03 is highlighted. The domain of the local-scale model is shown by the yellow box. The background is a Landsat image acquired in 2014 available from the US Geological Survey.
The lithological analyses of the recovered core revealed the presence of a
top 10 to 15 m thick sandy unit underlain by a 15 to 20 m thick clay layer
overlying the limestone (fine sand) at the base borehole. The interpretation of
the well tests provides estimates of the limestone hydraulic conductivity
The results of the chemical analyses of the groundwater and pond water samples are summarized in Table 1. The measurements highlight a deterioration of the groundwater quality, mainly in terms of NaCl concentration. Moreover, the similar proportional concentrations of various anions and cations in the water samples taken in the pond and in the boreholes suggest significant mixing between the two water bodies.
The transient electromagnetic (TEM) method has been widely used in
groundwater studies
(Auken
et al., 2003; Danielsen et al., 2003; Siemon et al., 2009) and offers a
relatively fast and cost-effective way to obtain information about the
ground electrical resistivity down to depths of a few hundred meters. The
ground-based TEM method employs a transmitter loop deployed on the ground
surface, which generates a primary magnetic field. After a short duration
(
The TEM survey was carried out in December 2013 using the WalkTEM system
(Nyboe et al., 2010). The full survey was comprised of 127 soundings, 110 of which were located on a 200 m spacing grid within the 2.7 km
The WalkTEM system setup.
A west-to-east section of the resistivity model obtained from the TEM data along the black line in Fig. 5. The color is faded white below the depth of investigation (gray line). The bars on the plots represent three MRS results in the vicinity of the profile in terms of the free water content (top) and relaxation times (bottom). Figure 5 provides the location of the MRS soundings.
The quality of the collected data is high and appropriate for the planned
analyses. For the majority of the measurements, there are usable data points
from 5
Magnetic resonance sounding (MRS; also called surface NMR) is an electromagnetic geophysical method that noninvasively measures water content and pore structure. The method works based on the physical principle of nuclear magnetic resonance (NMR) and is directly sensitive to water molecules and their interaction with the pore space. In short, using Earth's magnetic field, resonance excitation is induced by passing a tuned AC current into a large wire loop deployed on the ground surface. The corresponding energizing magnetic fields propagate in the subsurface; at any position in the subsurface, a component of the energizing field excites the water molecules. The measurement continues by turning off the energizing pulse and recording the responses from all subsurface excited protons as they gradually return to their initial (equilibrium) state. This experiment is called a free induction decay (FID) and is mostly used for MRS data acquisition. The experiment is repeated for a number of energizing pulses at different pulse moments (defined as the product of the current amplitude and the pulse duration) through which different volumes of the subsurface are sampled, and therefore depth information is provided. For more information about the principles and application of the MRS method, see Behroozmand et al. (2015).
In near-surface geophysics, MRS is commonly used to estimate free water content (or porosity in the case of saturated porous media) and hydrogeological properties such as pore size and hydraulic conductivity. The initial amplitude of the MRS decaying signal is proportional to the water content, while its relaxation rate provides information about the pore structure. A small relaxation time typically indicates fine-grained material, whereas a high relaxation time indicates coarse materials. The method is well established for near-surface characterization (Chalikakis et al., 2008; Günther and Müller-Petke, 2012; Knight et al., 2012). Different studies have also dealt with estimating hydrogeological parameters from MRS and have found good correlation between MRS-derived parameters and those from borehole aquifer tests (Boucher et al., 2009; Herckenrath et al., 2012; Plata and Rubio, 2008; Vilhelmsen et al., 2014, 2016; Vouillamoz et al., 2012, 2015).
The field campaign consisting of six MRS soundings was performed in January 2014 under the same hydrogeological conditions as the TEM survey. The locations of the MRS soundings are shown in Fig. 5. Prior to the survey, a noise scouting study was carried out over the entire area to investigate the ambient electromagnetic noise condition and to propose potential locations for acquiring MRS data. The results of our noise scouting presented low noise levels in the area, which suggests that it is a good site for measuring MRS data using a relatively low number of repeated measurements (free induction decays; FIDs). We used the NUMISPoly system (IRIS Instruments, Orleans, France) for the data acquisition, and the system was configured in coincident loop geometry with a 100 m side square loop as both a transmitter and a receiver. In addition, up to two reference loops were deployed for recording noise. Table 3 summarizes the measurement configuration used in the study. The measurement sequence consisted of a noise record before excitation, an energizing pulse, and a measurement dead time (to switch from the transmit to receive phase) followed by a recorded FID (see Table 3 for details). We measured FIDs for 16 pulse moment values, and a stack size of 30 was more than enough to obtain high-quality data.
The MRS measurement configuration used in this study.
The inversion results for sounding MRS03 (see Fig. 5 for the location of the sounding). The MRS and TEM data were inverted jointly. Row one shows the MRS observed (column one) and simulated (column two) data and the weighted data residuals (column three). Column four shows the TEM data fit. Row two shows the inversion results (black lines) in terms of resistivity, free water content, and relaxation time (columns one to three, respectively) together with the model parameter uncertainties (gray error bars). Column four shows the lithological information obtained from a nearby borehole.
The data were processed following the approach described by Dalgaard et al. (2012) and Larsen et
al. (2013) and inverted jointly with the TEM data
following Behroozmand et al. (2012a, b). We used the AarhusInv code
(Auken et al., 2015) for the inversion of
the data. As an example, the bars on the top of the TEM plots in Fig. 6 show the results from three MRS soundings in
the vicinity of the profile. In the top figure, the MRS bars show the free
water content, whereas in the bottom plot they display the relaxation times.
By comparing the TEM results with the MRS results at profile distances 180,
850, and 1900 m, one can translate the geophysical results into geology. The
MRS measurements suggest no free water in the first layer. This, together
with a relatively high resistivity of 10–200
Figure 7 displays the detailed inversion results for sounding MRS03. The MRS and TEM data were inverted jointly. The top row shows the MRS observed data (column one), the simulated data (column two), and the weighted data residuals (column three) as well as the TEM data fit (column four). The estimated model fits both datasets very well, as shown in columns three and four. Row two shows the inversion results (black lines) in terms of resistivity (column one), free water content (column two), and relaxation time (column three). The gray error bars denote the model parameter uncertainties, shown as the 68 % confidence intervals. The model is well determined and is in good agreement with the lithological information obtained from a nearby borehole (column four). It is noteworthy that the MRS results vary along the profile in Fig. 6, suggesting that the clay content of the second layer may vary.
Overall, the outcomes of the hydrogeophysical surveys integrate satisfactorily with the lithological and hydrological characterization available from the wellbore information.
The derived hydrogeophysical model was used as an input in the hydrological model of the area, as will be discussed in the following section.
Numerical modeling was developed to gain some fundamental insight into the processes that explain the observed pond development and provide some preliminary information on the possibility of using TWW for managed aquifer recharge (MAR) in the area. Based on the hydrogeophysical characterization of the area, we set up a simplified groundwater flow depression-scale model to verify the possible connection between the desert reclamation and the wetland growth and a density-dependent flow and transport local-scale model to test the potential and the effect of recharging the saline aquifer by fresh TWW.
The models, particularly the one developed at the local scale to investigate possible MAR scenarios, rely strongly on the outcome of the hydrogeophysical investigations. Both the geometry of the hydrogeological layers (maps of the top and bottom of the various units) and the distribution versus depth of the groundwater quality have been derived through the integration of the TEM and MRS results.
The objective was to investigate the possibility of a surplus of irrigation water being responsible for the formation and growth of the pond at the boundary of the desert area by modeling the hydrological response of the zone to the reclamation carried out from 1984–2014. By simulating the decadal rise of the water table, the model also aimed to show the inefficient use of the available water resources. Furthermore, the results from the regional-scale hydrological model can be used to provide boundary conditions for the local-scale MAR model.
For this purpose, we used a 3-D Richards' equation solver that is part of the
more general code CATHY (Camporese et al., 2010).
The model domain is showed in Fig. 2 by the black
box. The area size amounts to 800 km a seepage face condition on the depression zone, i.e., for the surface nodes
around the pond with an elevation of less than 15 m a.m.s.l.; a net specific recharge of 400 mm year a null infiltration/exfiltration in the desert zone, i.e., the portion of the
domain complementary to the reclamation area. Due to the relatively large
depth to the water table and the lack of specific data, a null balance
between rainfall and evaporation has been assumed for simplicity.
Fixed and time-dependent Dirichlet conditions were prescribed along the
northwest–southeast bounds toward the desert (A–D,
Fig. 2) and the Nile Delta (B–C,
Fig. 2), respectively. The water table was set to
0 m a.m.s.l. along A–D with a sensitivity analysis that has demonstrated a
negligible influence of the selected value on the piezometric evolution in
the area of interest. The information at the regional scale reveals that the water
table was continuously rising in the Nile deltaic region
(El-Sayed et al., 2012) . However, due to the lack of
specific data, we have used the rise of the water table along B–C as a
calibration parameter in order to match the date of the pond formation and
the pond enlargement versus time. This represents the best information
available to calibrate the model at the scale of interest. The approach
suggested a linear rise from
The DEM was used to derive the grid discretization in the horizontal direction, with a characteristic element dimension that reduces from 1000 m on the external boundaries to 50 m around the pond to allow for a more accurate representation of the infiltration/exfiltration processes and changes in the time and space of the groundwater table and saturation degree. The 3-D finite element (FE) grids consists of 146 275 nodes and 830 808 elements with 24 layers of tetrahedra that are used to represent the three main lithostratigraphic units defined from the hydrogeological and hydrogeophysical investigations (Fig. 8).
The 3-D mesh of the regional hydrogeological model. A 2-D triangular grid was projected vertically to generate the 3-D FE tetrahedral mesh used in the CATHY simulations.
The hydrological properties (
A steady-state simulation was initially performed to fix the initial
condition in the whole domain for the transient run. The solution was
obtained by prescribing a water table elevation of 10 and 0 m a.m.s.l. on
the B–C and D–A (Fig. 2) boundaries,
respectively. The initial distribution of pressure and saturation degree
The evolution of the saturation degree along the vertical cross section E–E traced in Fig. 2, i.e., along the main groundwater flow direction as obtained by the 3-D depression-scale model.
Then, the model was applied over the period between 1984 and 2014.
Figure 9 shows the evolution of
The contour lines of the saturation degree at the land surface around
the depression in
MAR using treated wastewater is one of the main challenges in Egypt
with the reduction of freshwater due to climate change, the growth in
water demand due to population increase, and the enlargement of desert
areas reclaimed for agricultural purposes. The reuse of TWW (municipal and to
some extent industrial wastewater) is considered an effective water saving
measure in areas where this water would otherwise be lost from the Nile
system. The primary use of TWW is for the irrigation of green areas (landscape
development) and non-food agriculture or the improvement of aquifer quality
by mitigating or countering saltwater intrusion and contamination. The
planned extent based on treated wastewater is some 250 000 feddan
(approximately 1000 km
Within this general context, one of the aims of the IMPROWARE project was to
perform a preliminary evaluation of the effectiveness of MAR in the Nubariya
zone using TWW provided by the Nubariya wastewater treatment plant (WWTP). An
existing WWTP located in the area surrounding the pond has been updated with a
constructed wetland (CW) tertiary system in order to improve the quality of
the plant effluent to a level compatible with MAR. The CW treatment
capability in the present condition amounts to 160 m
A number of preliminary simulations were performed at a local scale to
understand the effect of recharging the aquifer system in the Nubariya. This
local scale coincides with the area where hydrogeophysical surveys have led
to an in-depth understanding of the geologic setting. The US Geological
Survey SUTRA code (Voss and Provost, 2002), which can handle
density-dependent flows under saturated–unsaturated conditions, was used to
simulate a number of scenarios of artificial aquifer recharge. Based on the
hydrogeological setting of the study area, the limestone unit identified in
the depth range between 0 and
The model domain extends approximately 2.7 km in the southwest–northeast
direction and 2.85 km in the northwest–southeast direction
(Fig. 2), roughly coinciding with the area
characterized by the MRS and TEM surveys at the southwestern tip of the
pond. The model spans the depth range from the ground surface to the bottom
of the fine-sand confined aquifer, approximately
Elevation maps (m a.m.s.l.) of
The domain was discretized into prismatic elements. A 2-D grid was initially developed composed of 7077 nodes and 6965 quad elements with the dimension ranging between 50 m on the external boundaries and 20 m in the central area where injection is planned. Then, the 2-D FE grid was “projected” vertically to generate a 3-D FE mesh for a total of 219 387 nodes and 208 950 elements. A vertical discretization into 30 layers 0.5 to 15 m thick allows for an accurate reconstruction of the geological formations detected in the area (Fig. 12).
The 3-D FE grid used in SUTRA to simulate possible scenarios of
artificial aquifer recharge at Nubariya:
The distribution of the salt concentration Null groundwater and concentration flux along the southwest–northeast
boundaries (side A Dirichlet constant conditions on the northwest–southeast bounds (side
A A Neumann no-flux condition on the land surface and the domain bottom.
A steady-state simulation was initially performed to acquire an equilibrated
condition in terms of pressure and concentration in the whole domain to be
used as the initial state for the transient runs. Due to the lack of
specific tracer tests, a sensitivity analysis on the hydrodynamic
dispersivity (longitudinal dispersivity
Three major scenarios were used to address the variability on the injection
layout:
The salt concentration (g L
The results in terms of
Scenario
The fate of the injected wastewater can play an important role in water
reuse because it may result in a further quality improvement due to the
well-known phenomenon of soil aquifer treatment (SAT)
(Idelovitch et al., 2003). Therefore, we have performed a
last simulation in order to investigate the possibility of withdrawing the
water previously injected into the limestone aquifer. With reference to the
well layout implemented in scenario
Generally, the numerical models are simplifications of real aquifer systems. The model results are affected by (1) the numerical approximations used to solve the groundwater flow and transport equations, (2) the discretization of the modeled area, and (3) the availability and accuracy of the hydrogeological data used to define the spatial distribution of the physical parameters, the boundary conditions, and the factors forcing the system (Idelovitch et al., 2003; Tsang, 2005). Another important limitation of calibrated numerical models is the non-uniqueness of their solutions.
The models presented here rely on a number of assumptions, starting from the
hydrogeological structure of the system and the distribution of the
hydrological properties (
Regarding the issue related to lithological and hydraulic heterogeneity, the
two models did not address the spatial variation of the hydrological parameters,
namely the hydraulic conductivity, elastic storage, and longitudinal and
transversal dispersivity. The construction of the models in this manner would
have required detailed hydrogeological information from aquifer tests and
decadal piezometric records, which are not available. However, an accurate
calibration of the depression-scale model is beyond the scope of the study.
Indeed, its main objective was to evaluate whether a groundwater flow model is
able to reproduce the formation and growth of the pond as observed by
satellite data by implementing reasonable values of
As reported above, the main assumption in the modeling of MAR is related to the disregard of aquifer clogging in the surrounding area of the injection wells. The deterioration of the aquifer properties due to biological, chemical, and physical clogging strongly depends on the quality of the water used in MAR (Bouwer, 2002). A number of methodologies have been proposed in the literature to account for the decrease in porosity and thus the conductivity of the medium due to clogging (Pérez-Paricio and Carrera, 2000). However, at the present stage of this study, the lack of specific information on the quality of the effluent from the Nubariya WWTP and the chemical and physical characteristics of the limestone composing the aquifer do not allow for a specific numerical investigation on the clogging effects.
Concerning the dispersivity values, a sensitivity analysis on
Finally, we highlight that the hydrogeophysical data acquired in this study are of high quality with good spatial coverage of the layered aquifer model. In a more complex environment, e.g., with a larger lateral heterogeneity, a higher density of hydrogeophysical measurements might be required to provide accurate hydrogeological models to be used in hydrological modeling.
The salt concentration versus time at the pumping well intake for the
two selected values of
In this study, we evaluated the possible reuse of treated wastewater in the Nubariya depression in Egypt using extensive hydrogeophysical input and numerical modeling. The geophysical surveys successfully characterized the aquifer system in the study area where distinct layers of unsaturated sand, saturated sand, and sandy clay were found. In addition, the combined use of the MRS and TEM methods provided information about the spatial distribution of the layers as well as the water quality and clay content. Within the sandy layers, we found that the water quality is inappropriate for use as drinking water or for agricultural purposes. In addition, a relatively shallow confining clay layer was found as one of the reasons for the evolution of the artificial pond in the area. Below the clay layer, a 100 m thick aquifer with a high free water content was nominated as the potential aquifer for recharge.
The estimated hydrogeophysical model was used as an input for building the hydrological model of the area. The simulations carried out at the depression scale clearly pointed out the inefficient use of the freshwater resources in the study area. Although not very well known, the problem of waterlogging and pond development within desert zones caused by over-irrigation has been reported for several locations worldwide. This study investigates a phenomenon that represents a contradiction in areas where long-term programs are underway to effectively reuse treated wastewater: large volumes of good quality freshwater are wasted and large economic investments are made to improve and reuse TWW.
The same low-permeability layer responsible for the formation of the artificial pond in Nubariya confines a relatively thick limestone aquifer and bounds the upward migration of the treated water injected into the system. The continuity of the clay unit, an important requirement in view of the recharge, was clearly highlighted by the geophysical surveys. Although preliminary, a local-scale density-dependent groundwater flow and transport model allowed for the development of an optimized MAR scheme.
This research highlights the hydrological challenges related to the effective management of water resources in reclaimed desert areas and the effectiveness of using advanced geophysical and modeling methodologies to characterize the subsurface environment, investigate naturally and/or anthropogenically driven exchanges between groundwater and surface water, and plan appropriate interventions aimed at the efficient use of available water.
All data are available upon request to Ahmad Ali Behroozmand (abehrooz@stanford.edu).
The authors declare that they have no conflict of interest.
This article is an outcome of the IMPROWARE Project funded by the European Union in the framework of the SWIM program (Sustainable Water Integrated Management). Ahmad A. Behroozmand was also supported by funding from The Danish Council for Independent Research, Natural Sciences. The authors are much indebted to Ehab El-Hemady and Amr Fadlelmawla at the Egyptian Environmental Affairs Agency for their essential support of the field surveys in Egypt and to Andrea De Angelis at the Italian Ministry for the Environment, Land and Sea for his effort in managing the IMPROWARE activities. We are extremely grateful to Simon Rejkjaer, Mads S. Christensen, and Henrik Boejer, who in the challenging times of 2013 and 2014 collected the geophysical data together with a great crew from Egypt, particularly Sayed Bedair and Ibrahim Yehia Ibrahim. Edited by: Y. Fan Reviewed by: L. Tosi and one anonymous referee