Hydrogeology Journal

, Volume 15, Issue 1, pp 33–45

The Nubian Aquifer in Southwest Egypt

Authors

    • Center for Remote SensingBoston University
    • School of Professional and Continuing Studies, Department of Earth and Environmental Sciences14 Holmes Hall, Northeastern University
  • A. Werwer
    • General Company for Research and Groundwater (REGWA)
  • F. El-Baz
    • Center for Remote SensingBoston University
  • M. El-Shazly
    • General Company for Research and Groundwater (REGWA)
  • T. Fritch
    • Department of Earth and Environmental Sciences14 Holmes Hall, Northeastern University
  • T. Kusky
    • Department of Earth and Atmospheric SciencesSt. Louis University
Paper

DOI: 10.1007/s10040-006-0091-7

Cite this article as:
Robinson, C.A., Werwer, A., El-Baz, F. et al. Hydrogeol J (2007) 15: 33. doi:10.1007/s10040-006-0091-7

Abstract

Synthetic Aperture Radar (SAR) images, and topographic and groundwater data are used to understand heterogeneities of the Nubian Aquifer between 20–24.5°N and 25–32°E in southwest Egypt. New fluvial and structural interpretations emphasize that the desert landscape was produced by fluvial action, including newly mapped alluvial fans. In central locations, braided channels are spatially aligned with a NE structural trend, suggesting preferential water flow paths that are consistent with the local direction of groundwater flow. The alluvial fans and structurally enclosed channels coincide with gentle slopes and optimal recharge conditions (1–5%) derived from the new Shuttle Radar Topographic Mission (SRTM) slope map, indicating that these areas have high groundwater potential. The SAR interpretations are correlated with anomalies observed in groundwater data from 383 wells. Results suggest a relationship between the spatial organization of fluvial and structural features and the occurrence of low-salinity groundwater. Low-salinity water exists adjacent to the alluvial fans and in SW reaches of the structurally enclosed channels. Wells in the vicinity of structures contain low-salinity water, emphasizing that knowledge of structural features is essential to understand groundwater flow paths. The new approach is cost effective and noninvasive and can be applied throughout the eastern Sahara to assist in resource management decisions and support the much needed agricultural expansion.

Keywords

Arid landsHeterogeneityDrillingRemote sensingGroundwater management

Résumé

Des images Radar à Synthèse d’Ouverture (RSO) ainsi que des données topographiques et sur les eaux souterraines ont été utilisées afin d’appréhender les hétérogénéités de l’aquifère nubien, dans un secteur situé au sud-ouest de l’Egypte, entre 20 et 24,5°N et entre 25 et 32°E. De nouvelles interprétations fluviales et structurales soulignent l’importance de l’action fluviale dans le modelage des paysages désertiques, y compris les cônes alluviaux nouvellement cartographiés. Dans les secteurs centraux, les chenaux anastomosés sont alignés selon une direction structurale NE, suggérant ainsi que les cheminements préférentiels des écoulements concordent avec les directions locales d’écoulement souterrain. Les cônes alluviaux et les chenaux encaissés coïncident avec les pentes douces et les conditions de recharge optimales (1–5%) tirées des nouvelles cartes de pentes de la mission SRTM (Shuttle Radar Topographic Mission), indiquant un fort potentiel aquifère de ces secteurs. Les interprétations RSO ont été corrélées avec les anomalies observées dans les données sur les eaux souterraines issues de 383 puits. Les résultats suggèrent une relation entre l’agencement des entités fluviales et structurales d’une part, et l’occurrence d’eaux souterraines à faible salinité d’autre part. Les eaux à faible salinité se retrouvent au contact des cônes alluviaux et des chenaux encaissés. Les puits situés à proximité des structures contiennent des eaux à faible salinité, soulignant que la connaissance des éléments structuraux est essentielle à la compréhension des cheminements de l’eau souterraine. Cette nouvelle approche est peu onéreuse et non-destructive ; elle est applicable sur toute la partie Est du Sahara, en tant qu’outil d’aide à la décision pour la gestion des ressources et que support pour l’extension nécessaire de l’agriculture.

Resumen

En el sudoeste de Egipto, para comprender las heterogeneidades del acuífero Nubio, entre 20° a 24.5° N y 25° a 32° E, se han utilizado imágenes Synthetic Aperture Radar (SAR), datos de aguas subterráneas y topografía. Nuevas interpretaciones fluviales y estructurales enfatizan que el paisaje del desierto fue producido por acción fluvial, incluyendo abanicos aluviales recientemente mapeados. En ubicaciones centrales, los canales anastomosados están alineados espacialmente con un lineamiento estructural de dirección NE, sugiriendo trayectorias de flujo preferenciales coincidentes con la dirección local del flujo de aguas subterráneas. Los abanicos aluviales y sus canales estructuralmente controlados coinciden con suaves pendientes y condiciones óptimas de recarga (1–5%) de acuerdo con el mapa de pendientes derivado de la nueva Shuttle Radar Topographic Misión (SRTM), indicando que estas áreas tienen un alto potencial de aguas subterráneas. La interpretación de SAR está correlacionada con las anomalías observadas en datos de aguas subterráneas de 383 pozos. Los resultados sugieren una relación entre la organización espacial de los rasgos fluviales y estructurales y la ocurrencia de aguas subterráneas de baja salinidad. En las adyacencias de los abanicos aluviales se encuentra agua de baja salinidad y alcanza los canales estructuralmente controlados en el SW. Los pozos en la vecindad de estructuras contienen agua de baja salinidad, resaltando que el conocimiento de los rasgos estructurales es esencial para comprender la trayectoria del flujo subterráneo. El nuevo enfoque es económicamente accesible, no invasivo y puede ser aplicado a través del este del Sahara para contribuir en las decisiones de gestión de los recursos y apoyo de las muchas necesidades de la expansión agrícola.

Introduction

A new approach is presented that utilizes space-borne radar imagery, Digital Elevation Models (DEMs) and groundwater data to characterize the Nubian Aquifer in the East Oweinat area of southwest Egypt, between 20–26°N, 25–32°E (Fig. 1). The project is jointly conducted by the Center for Remote Sensing at Boston University in the USA, and the General Company of Research and Ground Water (REGWA) in Egypt. It is the first study of this kind. REGWA was initially assigned by the Government of Egypt to drill wells in the East Oweinat region as part of economic development plans that include the expansion of agriculture into desert regions. Mathematical groundwater models that have been applied by several organizations (including REGWA) show that it is economically feasible to extract 1,200×106 m3 of water per year for 100 years in this area (CEDARE 2001). Hence, there are plans for continued agricultural development.
https://static-content.springer.com/image/art%3A10.1007%2Fs10040-006-0091-7/MediaObjects/10040_2006_91_Fig1_HTML.gif
Fig. 1

Coverage of the Radarsat-1 images overlain on a Landsat ETM+ color composite (7, 4, 2). The ETM+ image shows the nature of the surface sand fields in the study area. Principal place names mentioned in the text are labelled. Darb El-Arbeain is a road leading from Sudan to Egypt and passing between Bir El-Shab and Bir Tarfawi

Nubian aquifers underlie southwest Egypt and also exist in Libya, Chad and Sudan. Detailed analysis of pumping test data by REGWA reveals that the local aquifer is unconfined in the East Oweinat area but not in the entire Nubian region. The groundwater resides in continental sandstone that unconformably overlies basement rocks and is covered by Quaternary deposits. The sandstone is highly porous with an average bulk porosity of 20%, excluding any secondary porosity introduced by fracturing. Drilling records have not yet revealed any losses of circulation during drilling operations, indicating fracture zones have not been penetrated in these instances.

The local aquifer is not homogeneous. Heterogeneity is observed in groundwater data obtained from 383 productive wells. For example, average Total Dissolved Solid (TDS) values may vary by 500 mg/L over a distance of 120 km. They have so far been unaccounted for and do not show any correlation with well depth. The latter is mainly determined by the relief of the basement rocks and their structural configuration, which affect the thickness of the Nubian Aquifer. In general, the total thickness of the aquifer is between 300–600 m, although in some places it does not exceed 200 m. The structurally controlled depth to basement rocks forms the major structural control affecting the geometry of the aquifer system.

Heterogeneity of the aquifer is also implied by the variation of drainage styles, drainage densities and the variable structural trends and densities observed in Synthetic Aperture Radar (SAR) images. From this perspective, it is possible that contributions to the heterogeneities arise because the groundwater present depends on: (1) fluvial recharge in the geologic past; and (2) fractures that introduce structural complexities, which influence groundwater distributions. In this situation, components of the original hydrological signatures are retained over time. Fractures introduce secondary porosity into the rocks and can serve as storage zones, collectors and transmitters of groundwater (e.g., Bisson and El-Baz 1991; El-Baz 1995; National Research Council 1996; Babiker and Gudmundsson 2004; Kusky et al. 2005). For example, open tensional faults are ideal conduits for groundwater flow in hard rocks and have added hydrogeological significance in that they constitute the shortest path for groundwater to travel from highlands to low ground. A well-known example of this is from the Rocky Mountains to the Central Plains in the U.S.A. (e.g., Garven 1995). The presence of fault breccias and unconsolidated material also enables groundwater storage and transmission. Finally, longer structures affect deeper, water saturated strata.

The goal of this research is to understand the fluvial and structural feature distributions as interpreted from SAR images and establish if there is any correlation of their spatial distribution with anomalies observed in the groundwater data. In the area of interest (Fig. 1), most drainage and structures exist in the near surface and are sand-covered, thus they require SAR data for their detection. This is because radar waves are uniquely able to penetrate the desert sand cover to reveal the courses of ancient rivers and streams (called palaeochannels) in the near surface that reflect the rainfall during previous humid phases of climate (e.g., McCauley et al. 1982, 1986; Robinson et al. 2000; Schaber et al. 1997). Near-surface imaging occurs when the sand is fine-grained relative to the radar wavelength, physically homogeneous and dry, with a moisture content of less than 1% (Hoekstra and Delaney 1974; Roth and Elachi 1975). Radar data are also useful in identifying structural features because radar waves can clearly depict scarp faces and regional structural trends (e.g., Roth and Elachi 1975; Sabins et al. 1980; Robinson et al. 1999a). For C-band radar (the principal radar wavelength used for the research in this paper), the depth of imaging near-surface geology is up to 0.5 m (Schaber et al. 1997; Robinson et al. 1999b).

The objectives of the study are to analyze the distribution of palaeodrainage features beneath the desert sand and the influence of structures on drainage; identify those structures that were potentially recharged by groundwater in the past (that is, in which groundwater resides or in which it was transmitted); and determine locations with enhanced groundwater accumulation. These operations are performed within a Geographical Information System (GIS) database, supplemented with a new slope map derived from the Shuttle Radar Topographic Mission (SRTM) elevation data set, and correlated with REGWA groundwater data in the same database. The scale transformations are simply resolvable within a GIS.

A question that could be raised is: How can there be groundwater in an arid and sandy desert? El-Baz (1982) explained this by recognizing that two main dynamics are superposed to form the features of the eastern Sahara. The first is the fluvial system that worked from the south to the north during humid phases of climate and is responsible for laying down the original material of the desert landscape at the mouths of the channels. The water from these channels ponded and percolated into the substrate through the porous Nubian sandstone rocks to be stored as groundwater (e.g., El-Baz 1988, 1998, 2000; El-Baz et al. 2000). The second dynamic is an aeolian system that continues to work from the north to the south and became operative once the climate changed and dry conditions prevailed. With the wind as the principal modification regime, the fluvially deposited sand began to be shaped into the dunes and the sand sheets that cover today’s desert surface (El-Baz 1982, 1998, 2000). Other major sand accumulations worldwide also illustrate the same spatial correlation of features, e.g., the Rajasthan of NW India, the Simpson of Australia, the Taklimakan of China and the Wahiba Sands of Oman (e.g., El-Baz 1982, 1998).

Present understanding of the Nubian Aquifer in southwest Egypt

East Oweinat exists in a huge sandy plain, called the Selima Sand Sheet which covers an extensive area of southwest Egypt and northwest Sudan. It overlies a sandstone formation first called “the Nubian Sandstone” by Russegger (1837). The Nubian Sandstone overlies basement rocks that are crosscut by extensive EW faults in southern Egypt that caused their vertical uplift (Issawi 1968, 1971, 1973, and 1978a,b,c). In these papers, Issawi also reported that the Nubian Sandstone of southern Egypt is the southern facies of Upper Cretaceous-Lower Tertiary marine beds that are also found in the north. On the other hand, Klitzsch et al. (1979) classified the Nubian strata in southwest Egypt into six distinct units, ranging in age from Jurassic to Late Cretaceous. Abdallatif et al. (1997) viewed the sandstone differently. They divided the rocks in the East Oweinat area into three units: a surface cover together with a dry part of the Nubian Sandstone, a water saturated Nubian Sandstone and the basement rocks. These authors concluded that the study area was affected by faults that caused the depth to the basement rocks to increase to the west and southwest, corresponding with an increase in aquifer thickness in these directions.

With respect to recharge processes, the groundwater in the East Oweinat area has been considered as part of a regional study of a “Nubian Aquifer System” that straddles Sudan, Chad, Libya and Egypt (e.g., Thorweihe 1990), and extends as far north as the Sinai and into southern Israel (Issar et al. 1972). In his early perspective, Beadnell (1901) suggested that the groundwater is recharged from the Darfur region of Sudan. Ball (1927) then constructed the first piezometric map that covers parts of Sudan, Chad, Libya and Egypt, and shows that the groundwater is recharged by regional groundwater movement from the south and southwest. Hellström (1940) modified Ball’s map and concluded that the recharge area is located in the Chad highlands as well as the northern part of Wadi Howar in Sudan. He also calculated that the groundwater velocity through the Nubian Sandstone is in the order of 15 m/year. More recently, Thorweihe and Heinl (1993) suggested that the groundwater of the Nubian Aquifer System is fossil water that infiltrated locally in the past and that the groundwater surface slopes to the NE, indicating inflow across the Libyan and Sudan borders to the Egyptian depressions. Sabed and Zeid (2003) also concluded that the groundwater formed from in situ precipitation during the Late Pleistocene to Early Holocene times. Using environmental isotopes and hydrochemical techniques, they determined that the main bulk of water is “fossil” and that contributions from recent recharge are insignificant.

Youssef and El-Saady (1964) were the first to research the aquifer’s hydrochemical properties. Based on their analysis, they suggested that the quality of the Nubian aquifer water is determined by its structural setting, where water in anticlines and the upthrown side of faults is of better quality than those in synclines or the downthrown side of faults. Atta et al. (1989) also investigated hydrochemical properties of the aquifer and showed that groundwater salinity in the East Oweinat area increases from the southwest towards the northeast. They concluded that the groundwater resulted from local precipitation and it is unlikely that it migrated many hundreds of kilometers from the south and southwest. Himida (1965, 1966) used geological, hydrological and hydrochemical data in his detailed studies to interpret the regional hydrology and evolutionary history of the waters of the Nubian aquifer, which he considered as the most extensive of a series of major artesian basins in North Africa. Based on hydrochemical evidence, he subdivided the basin into three main regions that contain different genetic types of groundwater, within which many more hydrochemical zones can be recognized (such as the very fresh groundwater in the Nubian beds beneath the Siwa Oasis). He interpreted the occurrence of a saline water-bearing formation in northern Egypt to be a consequence of deposition in marine environments (primarily during the Late Cretaceous to the Early Palaeogene marine transgressions), whereas the extensive fresh water-bearing formations in the central and southern regions are a consequence of extended infiltration of fresh meteoric water in continental conditions. There is an irregular boundary between these zones which contains evidence that sea water was displaced by meteoric water, with a general decrease in salinity to the northeast.

More recently, Fathy et al. (2001) analyzed hydrochemical properties of groundwater sampled from 14 newly drilled water production wells around 22°19’N, 29°46’E at Bir El-Shab (Fig. 1). Regarding salinity, they determined the principal dissolved chemical type to be sodium chloride and calculated the TDS values to range from 770 ppm to 1,202 ppm, which increase southward in this area. They concluded that the water quality is permissible for domestic purposes and suitable for agriculture uses. In a later study in Darb El-Arbeain (Fig. 1), Fathy et al. (2002) concluded that the hydrochemical composition of the groundwater indicates that the northern part is of marine origin, while it is of meteoric origin in the other parts of the area. Again, they determined the chemical water type to be sodium chloride. Finally, Sabed and Zeid (2003) performed hydrochemical analysis on the groundwater in East Oweinat and determined the chemical water type to be chloride, sulfate and calcium bicarbonate, with salinity ranges from 336 ppm and 688.8 ppm. Thus, they concluded that the East Oweinat water belongs to a freshwater class of meteoric origin that is suitable for irrigation purposes for most soil types.

The application of a wide array of age dating techniques shows that the water in the Nubian aquifers comes from several past pluvial episodes. For example, tangible evidence for earlier pluvial periods is provided by artifacts in numerous archaeological sites associated with remnants of playa or lake deposits in the eastern Sahara (Haynes 1982; Szabo et al. 1995). An Early Holocene pluvial cycle is well documented by geoarchaeological investigations at Neolithic playa sites in Egypt (Brookes 1983). Late Pleistocene lake deposits associated with Early and Middle Palaeolithic archaeological sites are best known from work in the Bir Tarfawi area of southwest Egypt at 22°55’N, 28°53’E (Wendorf et al. 1987). Similar associations occur in northwest Sudan (Haynes 1982). Further, radiocarbon dating also proves that the eastern Sahara experienced a period of greater effective moisture, with five palaeo-lake forming episodes occurring at about 320–250, 240–190, 155–120, 90–65 and 10–5 ka (Szabo et al. 1995). Four of these five pluvial episodes may be correlated with major interglacial stages (Szabo et al. 1995). Patterson et al. (2005) support the idea of variably aged water in the aquifers. They determined that shallow horizons contain younger water from native recharge events, compared with lateral flow from a southern recharge area in the deep horizons.

Variations in the age of the groundwater are also observed spatially. For example, Sadek et al. (2001) estimated the age of groundwater in Darb El-Arbeain to be from 21,187 and 23,790 years and that of the East Oweinat and Tushka areas (Fig. 1) to be from 14,780 to 17,070 years. Using 14C, Thorweihe (1990) estimated the age of the water in East Oweinat to be between 1,000 and 14,000 years and the water in the new valley further north to be more than 20,000 years old (the upper limit could not be established because it is outside the dating range of 14C). More recently, the application of 81Kr and 36Cl dating methods has allowed the upper limit of groundwater age to be estimated (Sturchio et al. 2004; Patterson et al. 2005). For example, by using long-life krypton isotopes, Sturchio et al. (2004) determined the upper ages of groundwater at six sites, ranging from 200,000 years in East Oweinat to 1,000,000 years in the Bahariya Oasis at 28°N.

What is clear then, is that much heterogeneity exists in the Nubian aquifer water, both vertically and horizontally, with respect to hydrochemical properties and age. There is a general consensus that there is negligible recharge in the present arid conditions; thus, the water is described as “fossil” water. Since the water is considered to be fossil, it is commonly understood that mixing processes in a Nubian Aquifer System led to homogenization over time, and that hydrochemical variations that exist arise because of chemical interaction of the groundwater with the aquifer matrix along its flow path. This paper challenges this perception by examining the local nature of aquifer in the East Oweinat area (using new groundwater data) and understanding whether common controls on groundwater recharge (that is, fluvial and structural controls) have left an imprint on the quality of the fossil groundwater and contributed to its heterogeneities. If so, this would favor the models that call for local infiltration in the past.

Materials and methods

The research makes use of standard mode Radarsat-1 and SIR-C SAR data (Parashar et al. 1993; Jordan et al. 1995). Both image types have 25-m spatial resolution. Radarsat-1 data are collected at one wavelength (C band) and one polarization (HH) and are used in this project to produce the base map of fluvial and structural features. These data are favored over SIR-C data since a continuous mosaic covering the study area can be produced (compared with discontinuous orbital strips for SIR-C data). SIR-C data were collected at two wavelengths (L band and C band) and multiple polarizations (HH, VV, HV, and VH). Depending on scene availability, they are used to supplement the Radarsat-1 analysis if drainage directions are difficult to discern. In these cases, SIR-C color composites are produced as they provide additional detail and information concerning geometric discontinuities. A combination of bands and polarizations found to produce the best high-contrast images in southwest Egypt is C band horizontally polarized and received, L band horizontally polarized and received, and L band horizontally polarized and vertically received images (Robinson et al. 2000). L-band radar can image up to 2–5 m in the subsurface (Elachi and Granger 1982; Schaber et al. 1997).

Radar data are unique in that they image the near-surface environment of southwest Egypt. Their use for the detection of palaeochannels in this manner is well documented (e.g., McCauley et al. 1982, 1986; Robinson et al. 1999a,b, 2000; Robinson 2002; Blumberg et al. 2004). The depth of near-surface imaging varies according to the moisture content of the sand at the time of imaging and the radar wavelength used for imaging. Experiments show that dry desert sand has a skin depth of 5 m or more (Elachi and Granger 1982). Calculations by Schaber et al. (1997) favor shallower penetration depths of 0.5 m for C band and 2 m for L band. In reality, L band may be imaging anywhere between 2 to 5 m in the surface, as indicated by field studies (McCauley et al. 1982).

Channels observed so far in L-band (23.5 cm) show less detail than those in C-band (5.6 cm), which is likely to be an effect of roughness. This is because near-surface imaging is only possible if the cover material is at least one fifth of the imaging wavelength, that is, it is radar smooth (Roth and Elachi 1975). Thus, longer wavelengths may encourage misinterpretation of these features in spite of the greater penetration depths by not imaging morphological details of the channels.

The Radarsat-1 scenes used in this research are n2355e2723; n2354e2825; n2357e2928; n2238e2706; n2240e2809; n2240e2912; n2122e2650; and n2123e2752. All images are selected with high look angles of 45°, as this allows for more refraction at the air-sand interface, thus enhancing near-surface imaging capabilities (Elachi and Granger 1982). The SIR-C scenes used are pr45032 and pr45033; pr45034 and pr45035; pr44044 and pr44045; pr45036; and pr11490 and pr11491. All have incidence angles greater than 50°, except for pr44044 and pr44045, which have an incidence angle of 26.5°.

The images are georeferenced using PCI’s Geomatica OrthoEngine and GCPworks software to the UTM coordinate system, zone 35, and WGS84 datum. Radarsat-1 georeferencing proceeds by transferring Ground Control Point (GCP) and orbital layers contained in the original 16-bit image header file and using the information contained therein to apply the transformation. A second-order model is applied for the transformation, which is preferred for orbital data, with the cubic resampling mode selected to avoid autocorrelation. SIR-C images are georeferenced using the satellite header information. A first-order polynomial transform model is applied for correction to minimize geometric distortion in the central parts of the scene, which lack control points.

Digital enhancement methods are applied to these data in combinations that maximize drainage and structural details (e.g., Robinson et al. 1999b). The application of histogram equalization and gaussian stretches, and edge filters (either simple 3×3 Laplacian filters or modified subtraction filters) are found to be particularly effective.

Digitizing of the structural and fluvial features is carried out in a GIS spatial database created using ESRI’s ArcGIS9 software. The satellite images provide the cartographic basis for feature analysis. Digital maps of fluvial and structural features are generated using individual scenes rather than a mosaic, in order to preserve the dynamic range of the individual images. Drainage is digitized by tracing the center of the channels, except for alluvial fans which are digitized as polygons. Fault traces that highlight the structural control on drainage, as well as independent geological structures, are digitized in a separate vector layer.

The DEM produced from the SRTM data set is also employed in the research. The SRTM mission was flown in February 2000 and was jointly undertaken by NASA, the National Geospatial-Intelligence Agency (NGA), the German Aerospace Center (DLR), and the Italian Space Agency (ASI). It produced two DEMs, one at C-band wavelength (5.6 cm) and one at X-band (3.1 cm). The DEM used in this paper is at the C-band wavelength — the same as the Radarsat-1 data used. Robinson et al. (2006) and Ghoneim et al. (2006) have demonstrated that SRTM data map the palaeotopography of desert regions. Thus, penetration depths for the C-band DEM used here would be expected to be no greater than 50 cm.

SRTM data were acquired in 225-km swath widths for 80% of the Earth’s landmass. They are freely available internationally at a 3-arc second (90 m) horizontal resolution and a 16-m vertical accuracy with a 90% confidence level (USGS-mission 2003). It is considered to be the best-known DEM ever generated at the global scale (Suna et al. 2003).

In most regions of the world, DEMs are used in hydrological modeling to understand surface drainage where precipitation rates are known. Most drainage in southwest Egypt is near surface and precipitation amounts at the time of their formation are not known. Thus, the DEMs are used alternatively. The principal function of the DEM is to establish the slope and confirm the flow direction of the drainage analyzed from the SAR images. With respect to flow direction, the SAR images and vector layers are overlain onto the DEM during analysis so that the topography that influenced the independent drainage systems can be checked. This approach is preferred to automatic drainage extraction in low gradient or flat terrain where the latter method is less reliable. A slope map is also derived from the DEM using the spatial analyst in ArcGIS9. Slopes allow infiltration rates to be appraised where gentle slopes (1–5%) have the best recharge potential for runoff, while building a certain hydraulic gradient for near surface water to continue flowing downwards. Areas with optimal slope conditions and high drainage-structure intersections and densities would have been most favorable to groundwater recharge in the past, thus they may depict areas with enhanced groundwater accumulation.

Subsurface groundwater salinity data, hydrogeological cross sections, and depth to basement rocks (e.g., Fig. 2a,b), which are available as a result of drilling, are also combined in the GIS database to determine if variations observed therein can be understood in the framework of fluvial and structural controls. The sections in Fig. 2a,b illustrate that the aquifer is mainly composed of sandstone intercalated with siltstone in places. The siltstone layers and lenses do not act as a confining layer in the aquifer as they do not separate the sandstone rocks.
https://static-content.springer.com/image/art%3A10.1007%2Fs10040-006-0091-7/MediaObjects/10040_2006_91_Fig2_HTML.gif
Fig. 2

Examples of two hydrogeological cross sections, CC’ (a) and DD’ (b), showing the depth to the basement and groundwater Total Dissolved Solid (TDS) levels. These profiles were compiled from subsurface data obtained as a result of drilling wells in the region. The information that they contain is included in the GIS database for correlation with results of the remote sensing analyses

Results

The accessibility of new groundwater data, the process-flow routine established for analyzing SAR data, the SRTM slope map, and the correlation of these datasets, have allowed a new and improved understanding of the local aquifer in the East Oweinat area of southwest Egypt.

The analysis of the SAR images has provided the first complete fluvial map of the southwest desert of Egypt and emphasizes that the desert landscape is one produced by fluvial action (Fig. 3). This is evidenced by the gullied escarpments around the Gilf Kebir plateau; the chaotic drainage in the limestone plateau to the north; the large alluvial fans mapped for the first time at the foothills of Gilf Kebir (and smaller ones upstream and to the east); structurally-controlled fluvial passageways; braided channels; and dendritic drainage in the highlands (Fig. 3). In central locations, some drainage lines terminate in kankar deposits, the latter having been mapped by Wendorf et al. (1993; Fig. 3). Regional structures are also pronounced throughout the region (Fig. 4). In the north, the trend is eastwest, compared with the newly mapped east northeast and northeast-trending structures in southeast locations. The eastwest structures have been discussed in earlier work and are considered to represent a structural high of the Neoproterozoic crystalline basement and Palaeozoic sediments, sometimes described as the Nubian Swell (Issawi 1978a; Thurmond et al. 2004).
https://static-content.springer.com/image/art%3A10.1007%2Fs10040-006-0091-7/MediaObjects/10040_2006_91_Fig3_HTML.gif
Fig. 3

A new fluvial map of southwest Egypt interpreted from SAR images. The fluvial interpretations overlain on the Radarsat mosaic show dendritic drainage in the highlands (the outer limits mark a drainage divide); gullied escarpments at the edges of Gilf Kebir; alluvial fans; chaotic drainage leading to the northernmost alluvial fan; and braided channels in the low-lying plains. In central locations some drainage lines end in kankar deposits (the latter are mapped from Wendorf et al. 1993)

https://static-content.springer.com/image/art%3A10.1007%2Fs10040-006-0091-7/MediaObjects/10040_2006_91_Fig4_HTML.gif
Fig. 4

A new structural map of southwest Egypt interpreted from SAR images. The structural interpretations overlain on the Radarsat mosaic show that regional faults are pronounced throughout the region. In the north, the trend is eastwest (already known), compared with east northeast and northeast in southeast locations (newly mapped)

Four principal alluvial fans have been mapped and are located at the foothills of Gilf Kebir (Fig. 3). They measure up to 85 km in length and 46 km in width. The three southern large alluvial fans all have well developed and highly integrated channel networks leading to them. Their major tributaries extend for over 200 km. The northern most alluvial fan is fed by less clearly defined drainage described as “chaotic”, a term originally introduced by Bagnold (1931).

The new fluvial analysis shows that braided stream patterns emerge toward the east. They form when surface slope and surface material favor stream deposition rather than erosion and constitute low-flow, high-volume discharge. They represent a common morphological transition observed in drainage passing from highland to lowland terrain and are visible in SAR images owing to coarser material in their beds; thus, they represent permeable water courses. It is possible that these channels abound throughout the low-lying areas in East Oweinat but are only visible where the sand cover is <0.5 m. Their presence, therefore, could potentially be confirmed by acquiring continuous, longer-wavelength radar scenes in the future such as those offered by ALOS-PALSAR. In central locations, the braided channels are spatially aligned with the northeast structural trend, suggesting preferential water flow paths (Fig. 5). They drain through Wadi Arid and Wadi Safsaf in the direction of the Tushka lakes. Interestingly, the direction of groundwater flow here is also to the northeast, thus emphasizing the influence of structural control on subsurface geology. Further, they are intersected by several east northeast-trending structures extending from the highlands that likely brought water from Gilf Kebir to the lowlands at this location in periods of wetter climate (Fig. 5).
https://static-content.springer.com/image/art%3A10.1007%2Fs10040-006-0091-7/MediaObjects/10040_2006_91_Fig5_HTML.gif
Fig. 5

Fluvial and structural interpretations overlain on the Radarsat mosaic showing that the newly-mapped braided channels centered on 21°N, 28°E are spatially aligned with the NE structural trend. This suggests preferential water flow paths in this direction. Also, note that braided channels are intersected by ENE-trending structures around 22°N, 29°E. The latter may have constituted potential recharge areas to the local aquifer in periods of wetter climate

The new slope map derived from the SRTM dataset (Fig. 6) shows that both the alluvial fan complexes and structurally enclosed channels coincide with gentle slopes (1–5%). Gentle slopes have the best recharge potential for runoff, while presenting the ideal hydraulic gradient for near surface water to continue flowing downwards. If these conditions prevailed during earlier pluvials, then both the alluvial fans and structurally controlled fluvial passageways would represent areas with high groundwater potential.
https://static-content.springer.com/image/art%3A10.1007%2Fs10040-006-0091-7/MediaObjects/10040_2006_91_Fig6_HTML.gif
Fig. 6

New SRTM-derived slope map with drainage features overlain. It shows that both the alluvial fan complexes and braided channels primarily coincide with gentle slopes (1–5%) that have the best recharge potential for runoff; thus groundwater potential should be high at these locations. The map also shows that kankar deposits coincide with areas having very low slope (0–1%), which indicates unfavorable infiltration conditions; thus, slope conditions support the idea of ponding and evaporation of surface water at these locations in the past

The kankar deposits, mapped by Wendorf et al. (1993) and included on Fig. 6, correspond with anomalously bright SAR signatures (SAR backscatter averages 5–6 dB higher than the surrounding terrain) as would be expected in a playa-like setting (Bergen et al. 1998). This is because they are characterized by rough surface residual deposits and possibly a shallow water table (0.5–1 m). Kankar deposits are often indicative of a closed basin environment (that is, depressions) and the slope map confirms that they coincide with areas having very low slope (0–1%). This suggests unfavorable infiltration conditions and possible ponding and evaporation of surface water in the past. Thus, near-surface water would be expected to be salty at these locations since salt is left behind upon evaporation. Other potential kankar deposits, that could be identified using these SAR principles, exist around 22°15’N, 28°13’E, but confirmation of their presence awaits field study.

These remote sensing observations were correlated against groundwater data retrieved from over 383 productive wells lying within two independent well fields (Fig. 7a). The westerly well field abuts alluvial fan deposits below Gilf Kebir (Awkaf; Fig. 7b). The easterly one (Garpad; Fig. 7c) sits in the NE-trending structurally controlled fluvial domain leading to Wadi Safsaf. Thus, they lie in two discrete fluvial domains. The data correlations within the GIS point to a relationship between fluvial features and salinity values. With respect to the Awkaf wells (Fig. 7b), TDS values of 500 mg/L exist adjacent to the alluvial fans, compared with 700 mg/L 10 km to the east (Fig. 7a and b). With respect to the Garpad Wells (Fig. 7c), TDS values of 400 mg/L occur in the SW portions of the structurally controlled Wadi Arid–Wadi Safsaf fluvial passageway, compared with 700 mg/L 10 km to the NE. Similarly, Himida (1965, 1966) and Atta et al. (1989) also determined that salinity increases to the NE. In both situations, kankar/playa deposits are not apparent in the SAR images in the immediate vicinity of high salinity values. It can be inferred, therefore, that low-salinity water exists closer to the water sources interpreted from the radar images (Fig. 7a).
https://static-content.springer.com/image/art%3A10.1007%2Fs10040-006-0091-7/MediaObjects/10040_2006_91_Fig7_HTML.gif
Fig. 7

a Radarsat mosaic with fluvial features and two separate well fields overlain (Awkaf and Garpad). The purple rectangles denote the boundaries of the salinity contours in (b) and (c). b and c Interpolated salinity contours retrieved from groundwater data in (a). Low-salinity water exists closer to the water sources interpreted from the radar images, that is, the alluvial fans to the west and the NE-trending structurally aligned braided channels. Arrows are included to emphasize these general directions of increasing salinity

The hydrogeological significance of the structures mapped from SAR data was considered in relation to five new hydrogeological cross-sectional profiles that include the depth to basement rocks and TDS levels (e.g., Figs. 2a,b and 4). These profiles are in addition to the well fields shown in Fig. 7 and cover a larger area (Fig. 8). The depth to the basement allowed the vertical component of the sense of movement along the faults to be assessed. The analysis indicates that the structures are hydraulically open faults which are able to have transmitted groundwater. TDS levels along these profiles were also correlated with the structures within the GIS. The results show that wells in the vicinity of structures contain lower-salinity water than those that are not (Fig. 8), implying that the structures acted as freshwater carrying conduits during past pluvials (Fig. 8).
https://static-content.springer.com/image/art%3A10.1007%2Fs10040-006-0091-7/MediaObjects/10040_2006_91_Fig8_HTML.gif
Fig. 8

Map showing transects of hydrologic cross-sectional profiles and the associated TDS values (e.g., Fig. 2) and their relation to structures mapped from the Radarsat-1 images. The cyan circles denote areas with decreased salinity values, which occur where wells exist in the immediate vicinity of structures, indicating that these structures transported groundwater during past pluvials. Note that TDS values otherwise decrease toward the east at Awkaf and northeast at Garpad (see Fig. 7)

Specifically, along profile EE’, TDS values are as low as 392 mg/L at 22°25’N, 28°13’E in the immediate vicinity of the mapped NE-trending structure. Salinity values around 600 mg/L may otherwise be expected around these longitudes (Figs. 7 and 8). Similarly along profile BB’, at 22°35’N, 28°37’E, salinity values dip to 584 from 704 mg/L in the immediate vicinity of the continuation of the same structure described for EE’. TDS values in the order of 700 mg/L may otherwise be expected here. Finally, salinity values dip to 608 mg/L along profile AA’ at 22°17’N, 28°50’E, adjacent to the mapped NNE-trending structure. At these longitudes, values between 700 and 850 mg/L are typically encountered where the locations are removed from interpreted fluvial and structural features. Finally elevated salinity values of 780 mg/L do occur at the northern limits of the C’C profile (Fig. 8). This would be expected since the evaporation processes that produced the kankar deposits would raise the salinity levels also.

These observations, therefore, point to a relationship between hydraulically open structures (able to have transported groundwater during past pluvials) and reduced salinity values. An association between Nubian aquifer water quality and structural setting has been made before (Youssef and El-Saady 1964). These authors suggested that water in anticlines and the upthrown side of faults is of better quality than those in synclines or the downthrown side of faults. Results both from this study and the earlier study of Youssef and El-Saady (1964) highlight the importance of the knowledge of structural features in the development of groundwater flow paths. It should be stressed that other factors that could contribute to variations in salinity — such as the transport processes that the solutes undergo in the aquifer and the eventual mineralization processes — cannot be appraised in the context of the present study.

The joint analysis of the surface (remote sensing) and subsurface (groundwater) data also allows areas to be recommended that likely have high groundwater potential (Fig. 9). First, the newly mapped alluvial fans are hydrologically significant in that they likely served as both conduits and storage space for shallow groundwater trapped from the rivers that led into them in the past. Due to the surface gradient, the water storage capacity will extend beyond the limits of the alluvial fans to percolate into the porous Nubian Sandstone beneath (Fig. 9). Groundwater data obtained in the vicinity of the middle fan show a shallower aquifer thickness with low TDS levels (within 200–350 mg/L) and well discharges averaging high yields of 200 m3/hour (Figs. 2a,b and 8). Aquifer thickness is clearly affected by the geometry of the basement rocks at these locals, although the hydraulic conductivity of the wells remains high, with groundwater production of 200–250 m3/hour irrespective of aquifer thickness.
https://static-content.springer.com/image/art%3A10.1007%2Fs10040-006-0091-7/MediaObjects/10040_2006_91_Fig9_HTML.gif
Fig. 9

Radarsat-1 mosaic showing selected areas with high groundwater potential based on the analysis of surface and subsurface data

It is also expected that the regional east northeast structures crosscutting the braided channels at 22°80’N, 28°37’E result in areas with high groundwater potential (Fig. 9), where the east northeast structures likely transported groundwater from the surrounding highlands of the Gilf Kebir to this area in past pluvials. This is in addition to groundwater supplied by the structurally controlled NE trending drainage at these locations. Currently, no wells have been drilled near this location, thus yield rates are not known.

Discussion

Prior to the work completed in this study, the practical role of SAR data utilization to impact groundwater resource management decisions in the Western desert of Egypt had not been exercised. Rather, the focus of the research studies pertained to: confirming the ability of radar waves to penetrate the veneer of desert sand to image sand-covered courses of wadis and river beds and to delineate underlying geological diversity (e.g., Elachi and Granger 1982; Elachi et al. 1984; McCauley et al. 1982, 1986; Schaber et al. 1997; Daniels et al. 2003); understanding of the palaeochannel distribution with respect to Cenozoic history (e.g., McCauley et al. 1982, 1986; Burke and Wells 1989; Issawi and McCauley 1992); improving the interpretative capabilities of SAR data with respect to palaeodrainage direction definition (Robinson et al. 2000); examining the role of SAR data analysis to facilitate dating pluvial episodes (Sultan et al. 2004). Studies prior to the year 2000 relied on the analysis of Shuttle Imaging Radar-A (SIR-A) data or low-resolution SIR-C survey images. Studies from 2000 and beyond employed high-resolution, multi-wavelength, multi-polarization Spaceborne Imaging Radar (SIR-C) or Radarsat-1 data for fluvial feature analysis, combined with DEMs. However, the coverage of SAR data used in these later studies is of limited areal extent.

The analysis presented here shows the complexity of palaeodrainage morphologies and structural feature distribution for the entire area between 20 to 24.5°N and 25 to 32°N for the first time. The fluvial and structural interpretations produced are tested against anomalies observed in groundwater data to understand if the latter can be understood in terms of fluvial and structural controls. It is the first study of this kind. Results of the work have improved the understanding of the heterogeneity of the local aquifer in this area by showing that a relationship exists between the spatial organization of fluvial and structural features and the occurrence of low-salinity groundwater. It has also emphasized the utility of employing multi-source datasets for addressing complex problems. The remote sensing and GIS methods used are cost effective, nondestructive and noninvasive and can be applied elsewhere in the eastern Sahara.

Future work should include the analysis of multispectral data for complete surface feature identification; further field documentation to confirm and supplement the SAR image interpretations, and to perform structural analysis aimed at determining the hydrological potential of the structures and faults. This may require geophysical techniques for the detection of structures buried by sand. Additional drilling of exploratory wells is required where water samples will be collected using packers to isolate different water zones in order to understand water stratification, both with respect to water quality and age. Thus, vertical variations can be comprehended as well as the horizontal. Hydrochemical and isotopic analysis is also desired to complement the salinity data, in order to better understand the direction of groundwater movement and to discern between different mineralization processes (e.g., Verhagen 1995). This in turn could provide a criterion to understand how much the water is affected by evaporation or dissolution in addition to the suggested contributions from fluvial and structural controls noted here. Further, a distributed groundwater flow and transport model, in which the role of the structures are considered as preferential flow and transport paths could be constructed by choosing appropriate hydraulic and transport parameters. In this situation, the fluvial features can be considered as recharge sources.

Acknowledgments

Funding for this project has been provided by National Science Foundation, Award Number OISE-0417704 (NSF INT 8702-5), and the Egyptian Academy of Sciences, Award number OTH8-011-001. Remote sensing software is provided by PCI Geomatics. The authors would like to thank one anonymous reviewer for the very useful insights and helpful suggestions.

Copyright information

© Springer-Verlag 2006