Original article

Simulation of impact of present and future groundwater extraction from the non-replenished Nubian Sandstone Aquifer in southwest Egypt
A.M. Ebraheem S. Riad P. Wycisk A.M. Seif El-Nasr Abstract A numerical time-dependent groundwater model of the Nubian Sandstone Aquifer in Egypt is presented. A complete database of the hydrogeological and drilling information of 850 water wells drilled in the period 19602000 was evaluated and used for the model parameter input as well as for its calibration. A steady-state condition is rejected even for the pre-development time as the subsurface inow across the Egyptian/Sudanese borders is lower than discharge, and the observed hydraulic gradient is residual gradient and not due to steady-state ow. The results of the long-period simulation (10,000 years) indicated that the Nubian Sandstone Aquifer is still under the inuence of the past humid period and has been in an unsteady deplenishing process. Therefore, groundwater development plans should be based on this concept. The calibrated model was used to investigate the hydrodynamic impacts of the present and planned groundwater extraction on the potentiometry of the aquifer. The simulation results indicate that there is a real danger of groundwater depletion, particularly in the shallow aquifer in some areas. In fact, if the planned extraction of 1,200 million m3/year in East Oweinat area is imposed, drawdown after 100 years could be as much as 200 m relative to1960s levels. By this time the cone of depression will extend to Dakhla and Kharga oases. Keywords Nubian Sandstone Aquifer Groundwater extraction Aquifer depletion Regional groundwater ow model Groundwater simulation

Introduction
Southwest Egypt is within the southwestern part of Dakhla basin, which, together with the Kufra basin in Libya, forms a major part of the well-known Nubian Aquifer System in Eastern Sahara (Fig. 1). The rocks are predominantly continental sediments of Cambrian to Late Cretaceous age. Estimates of the groundwater resources underneath the Western Desert of Egypt have been published by Ambroggi (1966: 3,000 km3), Gischler (1976: 20,000 km3), and Heinl and Thorweihe (1993: 50,000 km3). To avoid under and over estimation, an average value of 28,000 km3 is considered here. However, only a very small part of this total can be exploited, because it is distributed over an area of 630,000 km2 and extends to a depth of 3.5 km below the surface. Since the 1970s, it has become clear that successful and sustainable exploration of water resources in southwest Egypt is threatened with many problems. The most serious and obvious problem is the rapid depletion of groundwater. Despite these problems, there is increasing demand for building new settlements and reclaiming land for industrial and agricultural growth, and people from the already over-populated Nile Valley and Delta are attracted to these areas. The sustainable management of the available natural resources, particularly water, is crucial for the success of the old and new economic projects in the Western Desert of Egypt. Over the last 20 years, several groundwater ow models have been used as a groundwater management tool. However, these models were either very large-scale ow models concentrated on the determination of the degree of non-equilibrium and time response of the whole Nubian Aquifer System (e.g., Brinkmann and Heinl 1986; Sonntag 1986; Heinl and Thorweihe 1993) or local-scale ow models designed for a very restricted small area where the boundary conditions cannot be dened clearly (e.g., Nour 1996). The availability of a relatively complete database of drilling information for the last 40 years as well as the exibility and modular structure of MODFLOW software enabled us to develop the present model for both local and regional groundwater ow modeling in the Nubian Sandstone Aquifer in southwest Egypt. In the regional model, the ow system was calibrated by comparing the computed and observed potentiometric heads at the available observation wells. The calibrated model was used to predict the impact

Received: 9 March 2002 / Accepted: 9 June 2002 Published online: 10 August 2002 Springer-Verlag 2002 A.M. Ebraheem (&) S. Riad A.M. Seif El-Nasr Geology Department, Assiut University, Assiut, Egypt E-mail: ebraheem@acc.aun.eun.eg P. Wycisk Geology Department, Martin-Luther University, Halle (Saale), Germany

188

Environmental Geology (2002) 43:188196

DOI 10.1007/s00254-002-0643-7

Original article

Fig. 1 Boundary conditions and existing (2000) and planned (2012 numbers in parentheses) extraction rates in million m3/year for the Nubian Aquifer System in Egypt. (Modied after Abu Zeid and Rady 1991; Heinl and Thorweihe 1993; MPWWR 1998)

of the present and planned groundwater extraction on the aquifers potentiometry for the coming 100 years.

Fig. 2 Simplied cross sections of the Nubian Aquifer System in Egypt (after Thorweihe 1990)

Geological and hydrogeological settings

Thorweihe (1990) subdivided the Nubian Sandstone Aquifer in Egypt into three structural units, which are not completely independent of one another (Fig. 2). 1. The Northwestern Basin, which is north of the Cairo Bahariya uplift. Only the southernmost strip of this unit stretches south of the salinefreshwater interface. Therefore, it is of minor importance for the Nubian Aquifer System. 2. The Dakhla Basin, which is the largest and most important unit in Egypt. Except for the southwesternmost corner, which is related to Kufra Basin in Libya, the southern part of Dakhla Basin is lled with mainly neto coarse-grained continental sandstone of early Cretaceous and Cenomanian age. This sediment pile is about 1,000 m thick and has a good permeability. It is intercalated by two regional formations of low permeability. In the southwesternmost corner, it consists of ne- to coarse-grained Paleozoic sandstone, which crops out in the area of Abu Ras and reaches more than 1,500 m in thickness. In the northern part of the Dakhla Basin, the Cretaceous sediments overlie Paleozoic sediments and are covered by some hundred meters of shale and clay, which form the conning bed of the Nubian Aquifer System. 3. The Upper Nile Platform is east of the Dakhla Basin and is separated from it by the Kharga uplift. Until the end of Cenomanian time, the area of the Upper Nile

Platform was a high; thus no sediments of early Cretaceous to Cenomanian age were deposited. In Turonian time, this area probably became a shallow basin and was lled with approximately 500 m of sediments with much higher proportions of shale than those of the Dakhla Basin. Therefore, the upper Cretaceous sediments of the upper Nile Platform have a much lower permeability than their equivalent in Dakhla Basin (Amer and others 1981; Hesse and others 1987; Thorweihe 1990; Meissner and Wycisk 1993). Regional groundwater flow model Recent groundwater ow modeling studies of the whole Nubian Sandstone Aquifer indicate that a steady state does not exist because subsurface inow at the southwest boundary is lower than total outow, and the whole system had been in the non-equilibrium state for a long time (Brinkmann and Heinl 1986; Heinl and Thorweihe 1993). To determine the long-term groundwater extraction that can sustain the foreseen large-scale agricultural development in southwest Egypt, a numerical groundwater ow model was developed using MODFLOW software. The model boundaries were chosen as no ow boundaries in the east and southeast (basement outcrops). The no ow boundary conditions can also prescribed in the north (salinefreshwater interface). Since the groundwater inow across the Sudanese borders is approximately known through Darcian ow calculations (Sonntag and Christmann 1987), the grid cells between Nasser Lake and Gabel Oweinat were considered as known constant ux cells. The grid cells of Nasser Lake were considered as constant head cells (Fig. 1). 189

Environmental Geology (2002) 43:188196

Original article

Fig. 3 Sections of hydraulic conductivities of Dakhla Basin (after Hesse and others 1987)

Based on bore logs, pumping test results, and hydrogeological cross sections, the Nubian Sandstone Facies were approximated into a two-layer model (Fig. 3). Transmissivity was calculated by integration of the permeability over the whole sediment thickness along their cross section and assigned to grid cells. A rst estimate of storativity in the order of 0.1 for the unconned and 0.001 for the conned part of the aquifer was used (Diab 1972; Ezzat 1976; Heinl and Thorweihe 1993; Ebraheem and others 1998). Groundwater ow by extraction was assumed to affect the whole sediment thickness. Once the geological parameters

are considered as given values, the groundwater ow is only dependent on extraction, the boundary conditions, and the natural seepage at surface in depressions. Evaporation rate in southwest Egypt ranges between 10 mm/day during January and 31 mm/day during July, with an average of 22 mm/day (Nour 1996). Steady-state and long-period transient simulations As the initial point for calculations, a steady-state ow with maximum storage at the end of the pluvial period (Fig. 4) was simulated. The resultant groundwater

190

Environmental Geology (2002) 43:188196

Original article

Fig. 4 Groundwater contour lines for initial lled-up conditions (digitized from Brinkmann and others 1987)

Fig. 6 Contour map of the initial hydraulic heads (digitized from Ball 1927)

thirds of the estimated annual natural discharge in the pre-development time before 1960. Tremendous annual wadi oods from all sections of the mountainous terrain are needed to balance the decit between annual groundwater recharge and natural discharge (which obviously does not exist). Therefore, it is concluded that the aquifer was under non-equilibrium conditions far before the pre-development stage in 1960 (Sonntag and Christmann 1987). To support this conclusion, groundwater ow in the Nubian Sandstone Aquifer was simulated for 10,000 years, starting with initial lled-up conditions. In this simulation, an arid condition with no articial stresses (i.e., no groundwater abstraction) is considered. Groundwater drains through depressions in the vicinity or is affected by local evaporation or evapotranspiration. Groundwater inow is limited to 120 million m3/year at the southwestern borders. The resultant groundwater contours at the end of the long-term simulation in 1960 (Fig. 7) indicate the following: The drawdown due to the natural discharge always kept the regional groundwater ow from southwest to northeast. contours (Fig. 5) show a similar ow pattern to that ob- The computed groundwater contour lines at the end of the long simulation period are very similar to the observed in the pre-development time (Fig. 6). Despite this served groundwater contours of the pre-development similarity, the steady-state equilibrium is impossible due period before 1960 (Fig. 6). to the following: Water balance calculation at the end of this period in1. Even under the humid conditions (obviously not the dicated that the total discharge is 200 million m3/year; case for the last 9,000 years), the groundwater might two thirds of it is inow and one third is local change in not reach the ground surface in hilly areas. The hystorage. draulic gradient times the transmissivity is insufcient Despite the similarity in ow pattern in Balls map to maintain the lled-up conditions. (Fig. 6) and the contour map of the simulated ground2. The main ow is not a regional ow from southwest to water contours (Fig. 7), groundwater heads in Balls northeast but a ow between several subareas. map seem to be lower than the simulated contours in 3. Taking into account that the regional hydraulic gradiFig. 7. This obviously indicates that initial lled-up ent in the map of Ball (1927) is 0.5103, and average conditions never existed. hydraulic conductivity, k = 0.86 m/day, the eld velocity of groundwater will be about 0.16 m/year. Taking These ndings conrm the conclusion that the system was the average cross sectional area as 750 km2, the total already in an unsteady deplenishing process before artiinow rate will be 120106 m3/year, which is only two cial irrigation started in 1960. The present groundwater
Fig. 5 Simulated groundwater contour lines assuming a steady-state condition

Environmental Geology (2002) 43:188196

191

Original article

In this simulation the articial discharge for irrigation and domestic uses was imposed at the pumping centers and the resultant groundwater contour lines and drawdown for the year 2000 are shown in Figs. 9 and10. The computed drawdowns for Kharga and Dakhla oases approximately agree with the reported values during the last 40 years. Results of parameter calibration are shown in Table 1. To obtain a more general picture of the behavior of the aquifer system, a groundwater balance of the whole aquifer in southwest Egypt was calculated. The groundwater balance can be expressed as: Outflow Inflow Decrease in storage where the system outow is the sum of natural discharge into depressions and articial extraction from pumping centers. Inow to the Nubian Aquifer System is limited to the underground inow through the southern borders since recharge in southwest Egypt stopped 9,000 years ago. Figure 11 indicates that during the 19602000 period, articial extraction increased by 500%, inow increased only by 20%, and groundwater volume decreased by 41 km3. If the present extraction rate is kept the same, the predicted groundwater volume decrease is about 300 km3 by year 2100. This is only 1% of the average value of the total groundwater mass below southwest Egypt. However, as a consequence of past and future groundwater extraction, drawdown cones have been formed in the vicinity of pumping centers and have become larger and deeper with time. At present, a pumping head of more than 100 m is considered to be uneconomical for irrigation. The attained good match between the observed and calculated groundwater contours in the short period simulation (19602000) was considered as a verication of the present regional ow model. Thus, it was used to predict groundwater contours, drawdowns, and water depths for the coming hundred years. Two scenarios were simulated: the rst scenario assumed that the present extraction rate of year 2000 is constant for a hundred years starting from year 2000. In the second scenario, the planned increase in extraction rates was imposed (Fig. 1).

Fig. 7 Simulated groundwater contour lines for the Nubian Aquifer System in 1960 (10,000 years after the initial lled-up conditions)

ow model is based on the non-steady-state concept (currently more than 95% of groundwater extraction is considered as groundwater mining of unrenewable resources). The digitized piezometric contours of Ball (1927) were assigned as the initial hydraulic heads for the model grid cells in all the subsequent transient simulations (Fig. 6). Short-term transient simulation The ofcial records of groundwater extraction from shallow wells (<100 m below ground level) and deep wells (>100 down to 1,000 m b.g.l) were grouped together, the extraction was added up, and was nally given as a mean value for a year in m3/day for the pumping grid cells. For calibration purposes, the observations of hydraulic head in the period 1960 to 2000 at the available observation wells were compared to the computed hydraulic heads in this period (an example is shown in Fig. 8).

Fig. 8 The observed and computed hydraulic head (m above m.s.l.) of the Nubian Sandstone Aquifer in the middle of Kharga oasis

192

Environmental Geology (2002) 43:188196

Original article

Fig. 9 Simulated potentiometric heads by year 2000 in aquifer 2 imposing the actual extraction rates starting from year 1960

The simulated decline in the potentiometric head for the years 2030, 2060, and 2100 (assuming the rst scenario extraction rates) are shown in Figs. 12, 13, and 14. The areas surrounding Gabel Oweinat, Dakhla, and Kharga oases will be the most affected in terms of groundwater depletions, particularly in the shallow aquifer. For example, drawdown in the potentiometric surface of this aquifer in the middle of Dakhla oasis will continuously increase from the simulated value of 55 m in year 2000 (drawdown values of 4550 m were recorded in the existing observation wells in this year) to 100 m in year 2100. The situation will be aggravated more in East Oweinat area after imposing the planned increase in extraction rate (scenario 2). The predicted drawdown in this area will increase from less than 10 m in year 2000 to more than 180 m in year 2100 (Fig. 15). Finally, the predicted water depth at each grid cell for year 2100 was computed by subtracting the computed hydraulic head from the ground elevation. Since no topographic maps are available for the major part of the area, a large-scale topographic sheet published by Meissner and Wycisk (1993) was used for computing ground elevations at each grid cell. The resultant water depth maps for both scenarios are shown in Figs. 16 and 17. In both scenarios, water depth in the areas surrounding Dakhla and Kharga oases will be approaching the 100 m limit by year 2100. This means that all shallow wells in this area will dry up and groundwater exploitation for irrigation will be uneconomic. In scenario 2, water depth in East Oweinat area will be around 200 m from the ground level. However, these predictions are based on the regional model and more accurate predictions are expected after imposing the rened grids on the developed areas.

Conclusions
A steady state with initial lled-up conditions and recharge in southwest marginal mountains was simulated. The ow condition was assumed as natural conditions for the time before 1960. The results indicate that the gradient in connection with the transmissivity distribution Fig. 10 at the southwest boundary is not sufcient to maintain Simulated decline potentiometric heads (drawdown) by year 2000 in discharge in the Egyptian oases. This means that the
aquifer 2 imposing the actual extraction rates starting from year 1960 Table 1 Results of parameter calibration Subregion Aquifer 1 Transmissivity T1 (m2/day) Dakhla Kharga Farafra Bahariya Siwa East Oweinat Tushka 3 10.5 7.5 12 22.5 0.03 0.03 Storativity S1 3.4104 3.4104 3.4104 3.4104 3.4104 3.4104 3.4104 Vertical conductivity Vc1 (m/day) 6.48103 6.48103 6.48103 6.48103 6.48103 6.48103 6.48103 Aquifer 2 Transmissivity T2 (m2/day) 8,000 1,100 6,270 3,795 5,775 4007,500 1,500 Storativity S2 3.3103 3.3103 3.3103 3.3103 3.3103 3.3103 3.3103 Vertical conductivity Vc2 (m/day) 3.3103 3.3103 3.3103 3.3103 3.3103 3.3103 3.3103

Environmental Geology (2002) 43:188196

193

Original article

Fig. 11 Simulated balance for the Nubian Aquifer System during the shortterm simulation (a annual changes, b total changes)

Fig. 12 Simulated decline potentiometric heads (drawdown) by year 2030 in aquifer 2 assuming that the present extraction rates of year 2000 is constant for the coming 100 years

Fig. 13 Simulated decline potentiometric heads (drawdown) by year 2060 in aquifer 2 assuming that the present extraction rates of year 2000 is constant for the coming 100 years

194

Environmental Geology (2002) 43:188196

Original article

Fig. 14 Simulated decline potentiometric heads (drawdown) by year 2100 in aquifer 2 assuming that the present extraction rates of year 2000 is constant for the coming 100 years

Fig. 16 Computed water depth from the ground surface by year 2000 in aquifer 2 imposing the actual extraction rates starting from year 1960

Fig. 15 Simulated decline potentiometric heads (drawdown) by year 2100 in aquifer 2 imposing the planned extraction rates starting from year 2001

Fig. 17 Computed water depth from the ground surface by year 2100 in aquifer 2 imposing the planned extraction rates starting from year 2001

system was already in an unsteady deplenishing process before the articial irrigation started. This conclusion is supported by the long-period transient simulation results. Starting from the initial lled-up condition, they show the reaction of the aquifer to climate change. A long time after the transition to arid climate, a large-scale groundwater ow component was still observed. This indicates that, during arid conditions, groundwater drains through seepage and springs in depressions or is affected by local evaporation from shallow groundwater tables. The results from the rst scenario run demonstrate that, at the present extraction rate, the groundwater of the Nubian Sandstone Aquifer is being mined. Recharge from lateral inow is much less than the combined natural and articial discharge (approximately 1:22). Potentiometric surfaces will continuously decline at various rates until year 2100 (the end of the simulation period) and beyond. The upward

leakage to the rst aquifer will continuously decline and consequently affect phreatic water levels and springs. With the establishment of the new well elds in East Oweinat and Tushka, the situation will be aggravated further, and by year 2100 aquifer 1 may be dewatered in several areas. The simulation results of the second scenario, which consider the planned increase in the extraction rate, indicate that the cone of depression in the potential surface due to intense ground water extraction in East Oweinat will extend into both Dakhla and Kharga oases. This will cause further lowering of water levels and water depths will exceed 100 m in these areas as well as in East Oweinat area. Within the development regions, the telescoping mesh program (Leake and Claar 1999) should be applied, which allows for a rened grid to be superimposed over the regional model. Input from the regional model 195

Environmental Geology (2002) 43:188196

Original article

groundwater. Modeling of large-scale ow regimes. Proceedings serves as boundary conditions in the rened grid. This of a workshop on hydrogeology, 2022 May 1985. SFB 69, Berl approach allows for more precise analysis of pumping and effects of stresses on individual regions of the study Geowiss-Abh (A) 72 Brinkmann PJ, Heinl M, Hollander R, Reich G (1987) Retroarea. spective simulation of groundwater ow and transport in the To facilitate this work, available data on hydrogeology Nubian Sandstone Aquifer System. Berl Geowiss-Abh (A) should be organized and additional eld experiments 75(2):465516 should be carried out to better specify the important Diab MS (1972) Hydrogeological and hydrochemical studies of the Nubian Sandstone water-bearing complex in some localities hydrogeological parameters. in United Arab. Republic. PhD Thesis, Geol Dept, Faculty of Despite the fact that the development of groundwater resources in these areas is highly recommended, it has to be Science, Assiut University, Egypt U, Wycisk sustainable. A techno-economical feasibility study should Ebraheem A, Riad S, Doering of the New P (1998) Large scale groundwater ow modeling Valley area, SW Egypt. therefore be carried out periodically. TGA, Reihe C, No. 36, pp 118120 Ezzat M (1976) Exploration of groundwater, Kharga Oasis. Desert Acknowledgements We would like to acknowledge Professors irrigation, Ministry of Irrigation, Cairo Gunter Dorhofer, Sigrid Dorhofer, and James LaMoreaux from Gischler CE (1976) Present and future trend in water resources the editorial board of the Environmental Geology journal for their development in Arab Countries. UNESCO report valuable comments on the manuscript. This paper would not Heinl M, Thorweihe U (1993) Groundwater resources and manpossible without the database of drilling information together agement in SW Egypt. In: Meissner B, Wycisk P (eds) Geopowith geological and hydrogeological data. Many of the drilling tential and ecology. Catena Suppl 26:99121 activities, which have been going since the 1970s, were supported Hesse K, Hissne A, Kheir O, Schnacker E, Schneider M, by the USA-AID program. Thorweihe U (1987) Hydrogeological investigations in the Nubian Sandstone Aquifer System, Eastern Sahara. Berl Geowiss-Abh (A) 75(2):397464 Leake SA, Claar DV (1999) Procedures and computer programs for telescopic mesh renement using MODFLOW. US Geol Surv Open-File Rep 99-238 Meissner B, Wycisk P (1993) Geopotential and ecology. Catena Abu Zeid M, Rady MA (1991) Egypts water resources manageSuppl 26 ment and policies. Comprehensive Water Resources ManageMPWWR (1998) Southern Egypt Development Project. Internal ment Policy Workshop. External Affairs, Middle East and North Report, Ministry of Public Work and Water Resources, Cairo Africa Region, World Bank, Washington, DC Nour S (1996) Groundwater potential for irrigation in the east Ambroggi RP (1966) Water under Sahara. Sci Am 214:2129 Oweinat area, Western Desert, Egypt. Environ Geol 27:143154 Amer A, Nour S, Meshriki M (1981) A nite element model for Sonntag C (1986) A time-dependent groundwater model for the the Nubian Aquifer System in Egypt. Proceedings of the InterEastern Sahara. Berl Geowiss-Abh (A) 72:124134 national Conference on Water Resources Management in Egypt Sonntag C, Christmann D (1987) Groundwater evaporation from (Cairo), pp 327361 east-Saharian depressions by measuring deuterium and oxygenBall J (1927) Problems of the Libyan Desert. Geogr J 70:2138, 18 in soil moisture. Berl Geowiss-Abh (A) 75(2):385396 105128, 209224 Thorweihe U (1990) Nubian Aquifer System. In: Said R (ed) The Brinkmann P, Heinl M (1986) Numerical groundwater model. In: geology of Egypt, 2nd edn. Balkema, Rotterdam Thorweihe U (ed) Impact of climatic variation on East Saharan

References

196

Environmental Geology (2002) 43:188196