Hydrochemistry and environmental isotopes of spring water and their relation to structure and lithology identified with remote sensing methods in Wadi Araba, Egypt

The climate in the Eastern Desert of Egypt is arid and generally characterized by hot summers and moderate winters. Sporadic and short precipitation events take place in winter months (between October and April) and their spatial extension and intensity are extremely variable, e.g. the event on 27 October 2016 produced more than 100 mm of precipitation and caused a destructive flash flood in the town of Ras Ghareb (Saber et al. 2020), while north (Suez) and central (El Gouna) (Fig. 1), less than 20 mm of precipitation occurred.

The orographic influence of the Galala Plateaus on cloud distribution in Wadi Araba may lead to rain-bearing cloud concentration above the Northern and Southern Galala Plateaus. Observations in Suez, which has the closest climate station to Wadi Araba, show erratic precipitation events with usually low amounts, less than 6 mm/day, while only a few events produced more than 10 mm/day (Fig. 3).

The climate data presented in Fig. 4 were obtained from El Gouna climate station, situated at the Red Sea coastal plain (27.43 E, 33.66 N). The monthly averages and standard deviations are calculated using the data measured in the period between March 2014 and January 2019 with a time interval of 10 min. The range between average ± standard deviation represents 68% of the data.

The highest temperatures are recorded between June and August (>30 °C) and the lowest in December/January (<20 °C). The relative humidity with an annual average of 47% shows a weak seasonal variation with a minimum between May and August (average 45%) and a maximum between October and December (average of 50%). Evaporation data from the region were reported by Awad et al. (1996) with a maximum annual mean value in June with 10.11 mm/day and a minimum in December with 6.11 mm/day.

The geomorphological units of the landscape that are important for recharge and precipitation runoff can be differentiated. The Northern Galala Plateau shows typical conical landforms related to weathering of limestones and dolomites during more humid conditions in the past. The uncovered, flat, exposed fractured carbonate rocks show good properties for recharge.

The Southern Galala Plateau is block-wise intersected by distinct visible escarpments related to fault zones (Fig. 1). This area shows traces of an intense erosion such as deeply incised valleys. Sand and gravel sediments are accumulated in small depressions. While the northern escarpment is forming an almost continuous line, the southern border of the Wadi Araba basin is characterized by dissected front lines, intersected by larger N–S-oriented valleys. Along the northern escarpments, block gliding and rockfall towards the southern direction are a common feature. This morphology enables recharge in the hollows and fissures left behind and the steep slopes promote more surface runoff. The drainage network indicates a relatively flat bedding of the rocks and is probably the result of parallel tectonic structures.

The interaction between fluvial and tectonic processes is evident as the development of channels is influenced by subsurface structures and geotectonic movements related to plate tectonic rifting activity in the Gulf of Suez area and northward movements of the African Plate.

Wadi Araba is a structurally controlled tectonic basin (Fig. 5). The major force of the structural development in this area was the opening processes of the Red Sea, which started in the Miocene and continues today (Bosworth and Durocher 2017). The stratigraphy is divided into three major units, which are related to the evolution of the Red Sea/Gulf of Suez, i.e. the pre-rift (Pre-Oligocene), syn-rift units (Late Oligocene–Middle Miocene) and post-rift units (Late Miocene/Pliocene to Quaternary; Klitzsch and Linke 1983; Alsharhan 2003; Fig. 6).

The Nubian Sandstone Aquifer System (NSAS) in the study area overlies the metamorphic basement (Fig. 7) and contains formations from Paleozoic to Mesozoic age: the Lower Cretaceous Malha Fm. (Nubia A), the Lower Carboniferous Abu Durba Fm. (Nubia B) and the even older Nubia Fm. (C + D).

As a result of the regional compressional and extensional forces, along the Gulf of Suez (El-Naby et al. 2009), a complex structural system of numerous horsts and grabens with variable relief and dimensions developed (El-Naby et al. 2010). Over the entire region, the three major tectono-sedimentary features, i.e. Wadi Araba and the flanking Northern and Southern Galala Plateaus, are dissected by a complex pattern of NW–SE, N–S, and E–W directed faults. Both Plateaus consist of clastic sedimentary rocks of Late Paleozoic to Early Cretaceous age, and carbonatic strata of Cretaceous to Paleogene age (Kuss et al. 2000; El-Sadek 2009). The trend of the deep fault pattern, which affects the basement, the overlying Mesozoic, and recent sediments, oscillates around the E–W direction (El-Sadek 1998), and these fault patterns are assumed to be linked at depth with the NW–SE faults in northern Egypt (Abdel-Fattah et al. 2013).

At depth, in the study area, two almost parallel basic to ultrabasic dykes exist, which divide it into two parts: an eastern shallow part and a western deep one (El-Sadek 2009). Moreover, the study area represents a series of horsts (domes or anticlines) and grabens (troughs, basins or synclines). A meridional cutting geological profile is based on a cross section provided by Höntzsch et al. (2011) and preceding works (Moustafa and El-Rey 1993; Schütz 1994; Moustafa and Khalil 1995; Scheibner et al. 2001; Fig. 7).

Along the Gulf of Suez, groundwater emerges in the form of springs and artesian wells from the Nubian Sandstone aquifer; this was recharged in the late Pleistocene (Gat and Isaar 1974; Sturchio et al. 1996). Abouzied et al. (2020) reported that flash flood events contribute to the recharge of groundwater in the Eastern Desert along the Gulf of Suez; however, these assumptions are yet to be proven by calibrated infiltration models or age determination.

In the study area, four major aquifers exist: (1) Quaternary aquifer, (2) Upper Cretaceous aquifer, (3) Lower Cretaceous aquifer and (4) the Carboniferous aquifer (Nassim 1990; Fig. 6). The Quaternary aquifer occurs along the major dry Wadi and gets recharged during the rare occasional rainfall events. Three wells with yield between 5 and 23 m^3/day and one spring in the east of the study area derive water from that aquifer (Nassim 1990; Fig. 2).

The Upper Cretaceous aquifer (the formations of Wata, Raha, Matulla) hosts groundwater with total dissolved solids (TDS) of 1,200–8,000 ppm (Nassim 1990). The most productive aquifer is the Lower Cretaceous, which is known as the Nubian Sandstone, mainly recharged during pluvial times (Abdel Moneim 2005). This aquifer is composed of coarse sandstone and sands and has confined conditions in the central area of the case study. The groundwater in this aquifer is brackish with salinities of 1,000–2,600 ppm (Nassim 1990). The Carboniferous aquifer is mainly composed of sand and sandstone interlaced with clay layers that create confining conditions in the Zafarana area in the east. One disappeared spring, Abu Darag (Fig. 2), can be associated to that aquifer and discharged water with TDS of 2300 ppm.

Local nomads living in the area reported 30 groundwater emergences, of which 17 are springs and 13 locations are shallow hand dug wells or drilled wells that reach 86–1,200 m depth (Nassim 1990). Out of the 17 springs, 13 discharge from the Upper Cretaceous aquifer with average rates of 2.5 m³/day during summer and 5 m³/day during winter time, respectively (Nassim 1990).

The springs emerge at different elevations and from different stratigraphic units indicating that groundwater occurrence is not only influenced by lithologic properties, but also by the position and hydraulic properties of fractures and fault zones. Due to the still active neotectonic activities, faults and fractures have developed recently and play an important role in conduit development. In addition, karstification of the fractured limestones of the Galala led to the formation of cavities and a stochastic drainage system in the plateaus, hampering the exact determination of spring catchments.

Two rainwater samples were collected in the area of Hurghada in 2016 (R4) and 2018 (R5) to analyse major ion composition in rainfall. Both were supplemented by data from Askar (2014) and Hadidi (2016) for precipitation (R1, R2, R3). In the period between 2016 and 2019 water samples from 9 springs in the area of Wadi Araba were collected for chemical composition, stable isotopes and ³H analysis (Table 1; Fig. 2). In addition to these samples, a floodwater sample was collected on 18 February 2019, 3 days after a rainfall event (≈3 mm event at El Gouna weather station), directly from the area of SAM, to analyse the ³H content and stable isotopes in the provoking rainfall. The main spring at SAM was sampled for ¹⁴C analysis to determine the groundwater residence time.

The field parameters temperature (T), pH, and electrical conductivity (EC) were measured at each location using WTW portable devices. Alkalinity was determined using gran titration. To analyse anions and cations in flood water and groundwater, samples were filtered in the field using a 0.45-μm filter and filled into 50-ml bottles. The cation samples were acidified by adding 0.2 ml of HNO₃ and all samples were treated with 0.2 ml of C₄H₈N₂S, respectively. Major anions (NO₃, SO₄, Cl, Br) were analysed using a Metrohm 881 Compact Ion Chromatograph pro Anion–MCS. Major cations (Ca, Mg, Na, K) were analysed using an Agilent 715 ICP-OES. All analyses were performed in a laboratory of the Technical University of Berlin at the El Gouna Campus, Egypt.

Samples for stable isotopes of water were filtered into 2-ml bottles, finally sealed to prevent evaporation according to IAEA (2009) and later measured with a PICARRO L1102-i isotope analyser at Museum of Natural History in Berlin (Germany). The L1102-i is based on the WS-CRDS (wavelength-scanned cavity ring down spectroscopy) technique (Gupta et al. 2009).

The stable isotope ratios are expressed in the conventional delta notation (δ¹⁸O, δD) in permil (‰) versus Vienna Standard Mean Ocean Water (VSMOW). For each sample, six replicate injections were performed and arithmetic average and standard deviations (1 sigma) calculated. The reproducibility of replicate measurements is generally better than 0.1‰ for oxygen and 0.5‰ for hydrogen.

To determine the ³H content in the water, 1,000 ml could be sampled in the southern escarpment, from springs S5–S9 only (Fig. 2). Tritium analyses were carried out by applying the electrolytic enrichment using Liquid Scintillation Counting (LSC) in the Hydroisotop Laboratory, Schweitenkirchen (Germany) with a detection limit of 0.5 TU. Values are reported in tritium units (TU) where 1 TU equals an activity of 0.119 Bq/L (IAEA 1992). The tritium half-life time is (4,500 ± 8) days (≈12.3 years; Lucas and Unterweger 2000).

To estimate the groundwater age of the recent water, ³H input function of the sampled water must be reconstructed for several periods in the past using the tritium values for the rainfall in the country (if available) or neighbouring regions (IAEA 1990; Geyh et al. 1995).

Following Małoszewski and Zuber (1982), the mathematical form to describe the input-output concentration of the tracer under steady state flow condition is given by Eq. (1):where C_out(t) is the tritium output, C_in(t–τ) is the tritium input, h(τ) is the system response function, λ is the radioactive decay constant, t is the time scale and τ is the transit time.

To calculate residence time distributions, the hydrological system has to be conceptualized, which might either follow the piston flow, the completely mixed reservoir or dispersion approach (Małoszewski and Zuber 1982). Assuming that groundwater is being recharged in the catchment area and gets mixed with groundwater in the aquifer, the exponential model was applied to calculate the mean residence time (Eq. 2)where τ₀ is the turnover time of the system. τ₀ is identical to the mean transit time in the uniformly mixed reservoir (Yurtsever 1983). The system response function described by Yurtsever (1983) represents the distribution of the transient time of the tracer input–output function.

¹⁴C analysis allows the estimation of groundwater age between 1,000 and 40,000 years (Fritz and Fontes 1980, 1983; Wassenaar et al. 1991) depending on its activity in the total dissolved inorganic carbon (DIC). One sample (1 L) was collected from the main spring S5 at St. Anthony to determine the groundwater age using the ¹⁴C technique. Emphasis was placed on the minimization of gas exchange between emerging the groundwater and atmosphere. The almost neutral pH (7.82) (Table 1) was later enhanced to a value of 12 by adding NaOH. Later, CO₂ from the sample is separated by adding phosphoric acid. The ¹⁴C activity was subsequently measured using an accelerator mass spectrometry (AMS) at the Hydroisotop Laboratory, Schweitenkirchen (Germany).

The activity ratio of radiocarbon is given in percent modern carbon (pMC) as described in Eq. (3):where A_SN represents the normalized specific activity of the sample; A_ON represents the normalized specific activity of oxalic acid (Stuvier and Polach 1977; Mook and van der Pflicht 1999). Groundwater ages (t) are calculated from the general law of radioactive decay and are reported as conventional radiocarbon age in years before present, i.e. before the year 1950 (Geyh 2000; Eq. 4).

A₀ is the initial ¹⁴C activity in the recharge, A is the measured activity in the sample, and λ_C is the Cambridge decay constant (λ_C = 1/8267 year⁻¹) which is based on the physical half-life of (5730 ± 30) years. For uncorrected water ages, A/A₀ is equal to A_SN/A_ON. To correct the ¹⁴C value, ¹³C (δ¹³C-DIC) was measured using an isotope ratio mass spectrometer (IRMS) and expressed as permil (‰) referenced to the Vienna Pee Dee Belemnite (VPDB) standard: 1 sigma = ± 0.3‰. The sample was analysed in Hydroisotop Laboratory, Schweitenkirchen (Germany).

All determined analyses are presented in Table 1. Since only one floodwater sample could be collected, additional flood water analyses from the El Quseir area (F1) and from El Gouna and Hurghada area (F2–4) were added from literature (Awad et al. 1996).

Between Nov 2018 and Feb 2019, the discharge and temperature of the major spring S5 in St. Andrews was permanently (30 min interval) recorded applying a temperature and water-table recorder at the artificial tunnel feeding S5.

For the present study, optical, radar and digital elevation data were integrated into the GIS data bank. The satellite data were digitally processed using image processing (ENVI, SNAP) and ArcGIS software. The chosen image processing and RGB-combinations were focused on the enhancement of geologic, tectonic and surface water information. Information on the soil/sediment properties is very important in the scope of these studies as they have an effect on the infiltration capacity of surface water and water storage.

With its 14 spectral bands from the visible to the thermal infrared wavelength region, and its spatial resolution of 15–90 m, ASTER images were included in these investigations, especially the thermal bands. The RGB images created with the bands 10, 12 and 13 were converted from raster into polygon-shapefiles. The lowest values were extracted to get information on potential soil humidity lowering the reflectance in the thermal spectrum. The thermal bands of Landsat 8 were included as well.

Special attention was directed at precise mapping of traces of the tectonic pattern visible on satellite imageries, predominantly on areas with distinct expressed linear features (tonal linear anomalies, geomorphologic linear features, etc.). Lineament analysis helps to reveal traces of the tectonic structure. The term lineament is a neutral term for all linear, rectilinear or curvi-linear elements. Lineaments are often expressed as scarps, linear valleys, narrow depressions, linear zones of abundant watering, drainage network, and geologic anomalies. Tonal linear anomalies such as the linear arrangement of pixels depicting the same colour/grey tone were visually mapped as linear features as well as lineaments.

Structural features and lithologic units are better visible when merging RGB image products with filter-tool enhanced images. Low-pass and high-pass filters and directional variations were used for the detection of subtle surface structures. Merging the “morphologic” image products derived from “morphologic convolution” image processing in ENVI software with RGB imageries, the structural/tectonic evaluation feasibilities were improved. Surface-water input and resulting infiltration might be influenced by the fault and fracture pattern in the subsurface leading to a relatively higher permeability and hydraulic conductivity.

The floodwater sample F1 lies on the local meteoric water line (LMWL) and the global meteoric water line (GWML; Craig 1961; Fig. 7), indicating no fractionation of the rainfall’s signature due to evaporation. Comparing this sample with rainfall samples from the Sinai Peninsula (El Sayed 2006; Eissa et al. 2013) shows that rainfall in Wadi Araba is isotopically heavier than the Sinai rainfall. That might be related to the air mass movement from S to N (NOAA HYSPLIT model using GDAS meteorological data), indicating a rainout of heavier isotopes along the path and depleted signatures in Sinai.

After Dansgaard (1964), the deuterium excess value (d) can be determined with d = δ²H – 8δ¹⁸O. This parameter is strongly dependent on the relative humidity and kinetic effect during evaporation (Merlivat and Jouzel 1979). A d-excess of 8.2‰, as calculated for F1, indicates a relative humidity of about 80–85% (Gonfiantini 1986; Pfahl and Sodemann 2013), preventing significant evaporation prior to or during runoff.

Similar to the floodwater sample F1, groundwater samples (S1–S9) from the springs (with an average value of δ¹⁸O = −6.55‰, δ²H = −40.9‰) lie on the GMWL too. That fact confirms a fast infiltration process where evaporation does not play a role (Fig. 7a), usually provided by fast infiltration into highly permeable sediments or open (karst) conduits, which swallow precipitation immediately.

Groundwater from S6 and S7 show the most depleted δ¹⁸O and δ²H ratios, revealing either cooler recharge conditions or a higher elevated catchment area. Contrastingly, S4 which is located on the northern flank of Wadi Araba, might be recharged at a lower morphologic position compared to the other springs. Isotope values from S1, S2, and S3 resemble those of the famous Ain Sokhna, which discharges in Suez from the NSAS and is assumed to have been recharged during the late Pleistocene (Sturchio et al. 1996). That similarity may point to the same feeding aquifer and will be discussed in section ‘Concentrations of selected major ions and trace elements’.

In contrast to the spring waters, all groundwater from the deep wells in the area of Hurghada, having an average signature of δ¹⁸O = –5.8‰ and δ²H = −48‰, plot well below both the LMWL and GMWL, indicating a systematic evaporation effect. Since the trend line formed by the deep groundwater samples (Fig. 8a) starts from the GMWL close to the isotope ratio of S7, the source of recharge might be similar to S7, but the recharge process is much slower, providing room for evaporation. In that case, the trend line is an evaporation line. Alternatively, the recharge occurs similarly fast as for S7 but mixes later in the subsurface with a second, highly evaporated fluid. In that case, the trend line represents a mixing line. In any case, these deep groundwater samples are more enriched than the samples originating from the Nubian Sandstone in the Western Desert, proven to have been recharged in the Pleistocene era (Sturchio et al. 2004).

Plotting the d-excess of the water against δ²H (Fig. 8b), two groups of groundwater appear: (1) the deep groundwater with d-excess of <3 and (2) the springs from the Northern and Southern Galala Plateaus with higher d-excess values of >5.

Precipitation along the Gulf of Suez is dominated by Cl, SO₄, Na and Ca ions, indicating either the uptake of sea spray or the wash out of halite and sulphate from the atmosphere (Fig. 9a). According to the isotope results, the enhanced mineralization of sampled floodwater (F1) from Wadi Araba, which is 5–15 times higher in salinity than floodwaters from Hurghada and El Quseir, respectively (Fig. 9b), results from interaction of precipitation with the bedrock formations, which collect and transport the runoff. The variability of the Na/Cl ratios in the floodwater in the area of Hurghada (0.82–2.34) indicates different origins, while a value of 0.83 resembles the ratio in seawater, the Na/Cl of 2.34 refers to a different source of Na, probably from weathered plagioclase-rich rock formations (Siebert et al. 2014, 2019). As these formations do not exist in Wadi Araba, the Na/Cl ratio of the floodwater (F1) 0.97 refers to dissolution of halite, which is abundant in the region and has Na/Cl ratio of 1. The enhanced Ca, Mg and SO₄ concentrations in floodwater F1 compared to rainwater are the result of dissolution of gypsum, limestones and dolostones, which are exposed in the Southern Galala Plateau, and the release of these ions during the weathering process. Detectable fluorine (F) concentrations in the Hurghada area indicate release of that element from apatite and fluorite, which occur in the igneous metamorphic rocks of the Red Sea Hills. Since these rocks are missing in Wadi Araba, F is missing in floodwater from Wadi Araba.

To bring spring waters into context to earlier studies, analyses from Ain Sokhna spring (Sturchio et al. 1996) and Ain El-Boweirat (Nassim 1990; Fig. 2, currently dry) were included in Fig. 9c, where spring water composition is given. Groundwater in the southern Galala seem to have lower fluorine concentrations than groundwater from the northern rim.

While Ca and SO₄ content in spring waters is lower than in floodwater, Cl, Na and Mg concentrations in spring waters are larger than those of the floodwater F1. That issue refers to the dissolution of aerially imported gypsum by floodwater, while groundwaters dissolve halite and Mg-minerals, which may exist in the local aquifer.

Calculation of molar Ca/Mg ratio shows that groundwater in springs from northern Galala (S1–S4) receive slightly more Mg from aquifer rocks than those from the southern Galala (S5–S9). While the former show low ratios (0.83–0.98) those of the latter range from 1.25 to 1.7, indicating different aquifers of origin. Floodwater ratios are much higher (6.02–6.52), which is the result of additional gypsum, which does not originate locally. Considering SO₄/Cl equivalent ratios, differences are again observable between both spring regions. Groundwater in the northern rim shows ratios of 0.24–0.49, while those in the southern Galala springs are higher (0.56–0.87). Separation is also given comparing Mg/Cl equivalent ratios, which are high in the north (0.41–0.49) and low in the south (0.18–0.41), which indicate a larger content of dolomite in the aquifer feeding the northern springs compared to that supplying the southern group of springs. This is supported by saturation indices (calculated applying PHREEQC Interactive, version 3.3.9 (USGS 2020)). Groundwaters in the north are oversaturated in respect to dolomite, while those in the south are undersaturated.

Concerning the absolute salinity expressed as TDS, all groundwaters are brackish (1,500–2,600 mg/L) and not suitable as potable water. The modified Piper diagram (the position of Cl and HCO₃ + CO₃ is switched for a better representation, Fig. 10) shows the hydrochemical characteristics of the spring water.

Groundwater samples S5 and S8 seem to be not influenced by salinities which increase in S9 at St. Paul and continue to increase in S7 and S6. Ain Sokhna is most saline and has a different composition than all other groundwater samples in the northern group, referring to a different aquifer of origin. The dissolution of calcite and dolomite is lowest in groundwaters from S6 and S7, while the alkaline concentrations are highest. Field observations indicate the spring water seems to be partly recharged from the thick rubble overburden, which serves as storage. Though springs at S6 and S7 emerge close to S5 and S8, they emerge from different aquifers. While S5 and S8 originate from fractures and fissures in limestone, S6 and S7 obviously flow out of the shallow unconsolidated material. Hence, the hydrochemistry between both types of springs differ as stable isotope do.

The structural/geologic evaluation of optical satellite images and of radar data allows a reasonably precise mapping of larger lineaments, many of which are probably fault zones (Fig. 11). The evaluation of Landsat 8 and Sentinel 2 based filter images (Fig. 10a), which enhances surface topography, reveals prominent linear and parallel features. In parallel, equidistant linear features become visible on the satellite images processed with directional filters in ENVI in the areas covered by the youngest sediments. These parallel lineaments might reflect the stress pattern of the actual geodynamic activity. The most prominent lineaments are oriented in the NW–SE and NNW–SSE directions. The orientation and occurrence of these prominent linear features on the satellite images are obviously related to the subsurface tectonic pattern, characterized by block faulting (Nassim 1990). Larger fault and fracture zones might have an influence on groundwater flow as indicated in Fig. 11c, showing the example of the area of SAM. Fracture zones are most important for infiltration, migration, and storage of groundwater. The transmissive properties of the dominant rocks in this area are dependent on the presence of fracture zones to provide porosity and permeability. Whenever surface-water infiltrates after precipitation, it is likely that the infiltrated water flow is concentrated along the more permeable fracture zones. In the case of the St. Anthony area, several larger fault zones are intersecting, striking in SW–NE, NW–SE, N–S and NNW–SSE directions. The springs are aligned along the SW–NE-oriented fault zone bordering the Wadi Araba basin in the south.

The lowest and flattest areas are more susceptible to higher surface water flow. During extreme rainfall events in the flatter areass, the run-off concentration is slower and infiltration in sediments and structures is possible. Wadis were developed along weak zones representing fractures or faults; therefore, groundwater recharge may take place in the flat part and the wadis of the Galala Plateaus.

Over the period of observation, the hydraulic head in spring S5 oscillated between 1.14 and 1.15 cm. Because of the exchange of air in the cave where the spring occurs, the temperature differences in the measurements reflect the differences of air temperature (Fig. 12).