Christian Sonntag
Institute of Environmental Physics, University of Heidelberg, Germany

The environmental isotope and noble gas data of groundwaters from all over the Sahara provide strong arguments that these palaeowatcrs have mainly been formed by infiltration of local precipitation during moist climatic periods in the past. They show almost late-pleistocene 14C-ages of >20,000 y B.P., which compare with ages derived from radiogenic He accumulated in groundwater. Because of their stable isotope pattern across the Sahara with pronounced West to East decrease of dD and d18O the late-pleistocene palacowaters have been formed by infiltration of local rainfall from the Western Drift, i.e. Mediterranean winter rain climate has prevailed in Northern Africa in last glacial time. Palacowaters from holoccne humid phases appear to be relatively rare. This. however, is an artefact of non-representative groundwater sampling. The last pluvial when full groundwater reservoirs prevailed, has ended at some 5,000 to 6,000 y B.P. Since that time, unbalanced groundwater discharge has caused decreasing groundwater levels (hydraulic heads) in the catchment areas around the various morphological depressions, where groundwater evaporates from barren soils, wild vegetation, and even from groundwater lakes. This hydraulic head decay is discussed for the groundwater reserves in the Nubian Aquifer System (NAS) in context with the hydraulic interpretation of the isotope and noble gas ages of East-Saharan palaeowaters.

The Nubian Aquifer System extends over a region of 2 Mi. km2 which comprises the North Sudan-, Kufra-, Dakhla- and the North Egyptian-Basin, where sediment thicknesses up to several thousand meters arc reached in the basin centres. Natural groundwater discharge from East-Saharan depressions occurs at a total rate of approx.1x109m3/y (Ahmad 1983), either by capillary rise from the groundwater table or by ascend of confined or artesian groundwater through leaky confining beds followed by water vapour diffusion through top layers of dry soils and by transpiration of wild vegetation. In comparison to this, groundwater replenishment under the present arid climatic conditions by infiltration of episodic local rainfall and by subsurface inflow from rainy mountaineous terrain at the southern margin of the Eastern Sahara is very small (Sonntag 1986). If this unbalanced groundwater discharge is related to the areal extension of NAS mentioned above, a mean local discharge rate of D=0.5 mm water/year is obtained. Thus very small replenishment means a local recharge rate which is by at least one order of magnitude smaller than D, i.e. R<0.05 mm/y. Hydrological estimates of extremely low groundwater recharge like this are very difficult and need many convincing arguments.

Exponential Decay of the Hydraulic Head

Under the assumption of negligible recharge since the beginning of the present arid period at 5,000 y to 6,000 y B.P., the hydraulic head h (with the depression surface as reference level) has exponentially been decteasing, h(t)=hoxexp(-t/t), from its initial level of h0 =IOOm as an estimate for the mean elevation of the catchment areas around East-S aharan depressions (full groundwater reservoirs assumed!) to the present one of approx. 25-30 meters. The time constant t=gx(nxL2)/(K1xH) depends on the one hand on the hydraulically dischargeable groundwater mass, which is represented by the effective porosity n, by the hydraulic head h, and by the length scale L of the catchment area. On the other band, the time constant depends on the areal mean effective transmissivity KtxH (Kt=hydraulic conductivity of the sediments, H=penetration depth of the groundwater flow towards the depression). The dimensionless geometrical factor g depends on the morphological shape of the depression and of its catchment area around. In case of cylindrical depressions (radius r) with ring shaped catch ment areas (outer radius R), L has to be replaced by the depression radius r, and in g the ratio Rir of the radii comes into play; for example: R/r=4.3 gives g=8.2. Using R=130 km, r=30 km, n=0.1, Kt=l.2x105 m/s, and H=500m as typical numbers, a time constant of t=3,900 y is calculated for the hydraulic head decay in the catchment areas around (Sonntag 1986).

If steady state groundwater circulation at full reservoirs (pluvial period) is considered, this hydraulic time constant is equivalent to the mean groundwater age in the hydraulically dischargeable sediment layers above the depression floor, which are ho meters thick. The groundwater, however, which has reached the depression area, has passed through a much larger reservoir of ho+H meters vertical extension, where H denotes the mean penetration depth of the groundwater circulation below the depression floor. With ho=I00 m and H=500 m in use, this total reservoir is six times the dischargeable one. Since the ratios of the reservoirs and of the mean groundwater ages therein are identical, mean groundwater ages in the order of 6x4,000=24,000 years are to be expected for deep groundwater ascending to the depression surface. However, this is a hydrodynamic estimate of the deep groundwater age in plu vial time. In the arid period like now, this mean age is higher by the actual duration of the arid period, i.e by 5,000 y to 6,000 y

It should be mentioned here that the number M=500 m used for the penetration depth of the deep groundwater circulation stems from Burdon (Burdon 1977) Multiplied with the Krvalue above, an areal mean transmissivity of T=1.2xl0-5x500=6xl0-3 m2/s=520 m2/d is obtained. Transmissivity numbers like this are very important for long-term prognoses for economical palaeowater exploitation, i.e. to find out whether a wide and flat the depression cone will develop at fixed pumping rate or a narrow and steep one with intolerable groundwater draw-down in the well field.

Changing Groundwater Circulation Pattern in the Past

Autochthoneous palaeowaters in widely extended deep aquifers like the Nubian Aquifer System do not contradict the existence of large-scale groundwater flow from (rainy) montain ranges at the southern margin of the Sahara, like from Tibesti across Southern Libya into the Western Desert of Egypt, under the present arid climate. This paradox can be explained by changing groundwater circulation pattern after the transition from humid climate with local groundwater recharge, enough to maintain full aquifers, to arid climate with unbalanced groundwater. High ground-water levels everywhere in pluvial time imply a hydraulic head surface which has followed even small-scale topographic ups and downs of the (palaeo-) ground surface. Thus strong small-scale hydraulic gradients have driven small-scale groundwater circulations with discharge in maiiy topographic lows, which have fallen dry one by one in the course of the present arid period; i.e., most of them have already disappeared. At the beginning, small-scale cells of intensely circulating shallow groundwater have penetrated even through thick leaky confining beds. Thus large-scale groundwater flow driven by the regional hydraulic gradient like now was suppressed. However, in the present arid period, which lasts since 5,000 y or 6,000 y, hydraulic head decay has smoothed out small-scale topograhic variations of the hydraulic head. This implies that the small-scale, but intense circulation of the shallow and of the deep groundwater as well has ceased, so that larger or even regional scale flow has or will come up. This large-scale flow is less intense than the former circulation cells of the deep groundwater, i.e. there is no increase of the groundwater age along the regional flow lines. Due to the local groundwater formation everywhere in Palaeo-Sahara and particularly in last pluvial time, there is an increase of the groundwater age with depth below ground rather than a lateral increase in the direction of the present regional flow. The regional flow shifts this age profile only through the sediment package with a low distance velocity.

The vertical groundwater flow through the confining beds downward and upward, so that last glacial ages appear in the confined aquifer, need leakage coefficients for the confining beds in the order of 10-9 to 10-8 m/S and, of course, hydraulic gradients for driving the deep ground-water by small scale circulations, which are about one order of magnitude larger than the regional hydraulic gradient. If 0.3 permille is taken for the hydraulic gradient forcing the regional groundwater flow, then palaeo-gradients of 3 permille should have existed to drive the small-scale circulations of shallow and deep groundwater in the past.

This picture of changing groundwater circulation patterns over the paleoclimatic history of the Sahara follows the idea of Toth (Toth 1963).