The SIR-C Paleodrainage Experiment was designed as a regional follow-up to the limited coverage obtained by SIR-A and B. These earlier imaging systems revealed an extensive network of small to large paleochannels, a few meters or less below the mobile, eolian sandsheets that cover much of western Egypt and parts of northern Sudan and eastern Libya. SIR-C mission planning was focused on getting as much contiguous coverage of the Eastern Sahara as possible. We were able to use our own coverage allocation as well as that of other experimenters where their ground tracks passed over areas of interest to us. The result was that three quarters of Egypt and large parts of adjacent countries were imaged.

These images confirmed that a complex drainage network lay beneath the relatively thin but almost ubiquitous sands of the Western Desert of Egypt and adjacent areas. These fluvial features represent a record of continental sedimentation that followed the retreat of the Tethys Sea from northeast Africa. They span the time period from about 40 Ma, through the arrival of the Nile (about 6 Ma), and on to the last Holocene Pluvial (only about 6 ka). The large valleys are remnants of regional drainage systems that were active before the Nile came into being. One, the Qena System, was the master drainage in Egypt for at least 16 to 18 years and was at its peak during the late Neogene. The southward flowing Qena System played the principal role in shaping the main geomorphic elements of Egypt--the Limestone Plateau, and the great southward opening Nubia pediplane with its apex at Aswan. Recognition of the Nile as a subsequent, pirate stream that grew southward by headward erosion, induced by the Late Miocene drop in sea level of the Mediterranean (the Messinian event) is a key element in our model. Conclusions drawn from the SIR Experiment are supported by five field studies. In addition, they are consistent with recent work by others in the Fayum Depression of northern Egypt, where the old concept of the existence of a Protonile of Tertiary age was put to rest by overwhelming stratigraphic evidence. A Protonile would have prevented drainages from the Red Sea Range from crossing into the Western Desert, where their buried valleys were detected by the SIR imaging systems.

At the beginning of Quaternary time aridity set into the Eastern Sahara, but progressive desiccation was punctuated by numerous pluvial episodes. These events coincided with the establishment of new regional gradients, so that the most recent patterns seen are complex and locally of high drainage density. The latter pattern, first seen on the specially processed SIR-C images, suggests that monsoonal conditions prevailed episodically. Fluviolacustrine conditions were widespread and early man utilized the edges of the flowing and standing water bodies whose traces were delineated in our field investigations. A satisfying new result is the definition of the full extent of a major paleodrainage system in eastern Libya, Wadi Kufra.

Study of the uncontrolled preliminary mosaic of the non-redundant SIR-C coverage of part of southwest Egypt and northern Sudan reveals six distinctive buried stream patterns, which are indicative of their climatic history and depositional environment. They are: 1) Dendritic patterns of the type commonly seen in humid areas; 2) Broad, flat-floored valleys, some of which may have very old exhumed drainages on their floors; 3) Arrays of small, straight to sinuous channels that may be the result of seasonal cloudbursts; 4) Wide, almost completely aggraded valleys that are bounded by brighter radar terrain, very bright patches of caliche, and on the ground are usually marked by Stone Age implements and hearths; 5) Braided to anabranching channels that are inset into Type 4 valleys and are the products of deposition under short-lived, high bedload conditions; 6) Anomalous channels, linear, disconnected features that are not strictly the products of surface runoff; they are probably the result of piping, that is, subsurface flowage over an impermeable horizon.

The SIR-C Mission Plan. Unlike many of the other SIR-C flight experiments, which were specific to small test sites, the Paleodrainage Experiment was directed toward obtaining as much contiguous coverage of the sand plains of the Sahara as possible. Our chief objective was to delineate thinly buried paleodrainages in order to begin to fill in about a forty-million-year gap in the geological record of Northeast Africa. We also wished to test the utility of the system for geological and archeological studies. Our focus was on the paleodrainages first seen by SIR-A (McCauley et al., 1982), and a further evaluation of the Trans-Africa Drainage System (TADS) hypothesis, (McCauley et al., 1986 a,b). Of necessity our emphasis was on Egypt and northern Sudan where we had most of our field experience (fig. 1).

Fig. 1 near here

Data Processing. Initially, we did not plan to use the multi-frequency data because of the limited penetration capability of the X- and C-band radars, nor multiple polarization because of diminution of swath widths. The C band proved, however, to have greater imaging depth than anticipated and false color composites of the three bands have been useful for other applications. The distribution of the Survey Image products to us was a problem because of its intermittent and piecemeal nature. We received our first batch of CD-ROMS in July, 1994 and the final ones from the second flight kept coming throughout 1995. Until then we were uncertain about the scope of our useful coverage and could not produce a regional mosaic for study of the distribution of the paleodrainages.

Our basic plan was to obtain L-band Survey Mode data, preferably with HH polarization and with look angles that would vary to obtain the width necessary to sidelap the images. This was the ideal, but because of operational or technical constraints imposed (because other investigators were using some of the same orbital passes to obtain different types of data) our ideal was not achieved. Several otherwise useful passes fell almost on top of one another and in other cases the swaths were too narrow to cover the area intended, leaving gores within the area of possible coverage. Nevertheless we managed to obtain from both missions 22 non-redundant mapping swaths, most of which are side-lapped ,covering about 775,000 km2, or about three quarters of the land area of Egypt (fig. 2).

The sorting out of relevant segments from the dozens of data takes, each consisting of about twenty or more 100 km long segments on the 122 CD-ROMS received, proved to be a time consuming task. The final assembly product, which required substantial manual effort, is seen in Fig. 2 and Table 1. Processing and scaling of each segment was done on a PC, using the Adobe Photoshop program to read the individual segments, and to enhance and scale them for printing on a 600 dpi laser printer. For preliminary analysis and without geometric corrections, these segments were mosaicked together using the 1:500, 000 scale grid of the Tactical Pilot Chart Series published by the Defense Mapping Agency Aerospace Center, St. Louis, Missouri. Comparisons with Landsat images then determined which features seen on the radar images were not visible at the surface.

Figure 2, Table 1 near here

Because of the time required to manually assemble the mosaic, we had to be cautious in requesting special processing of the limited number (about 100 allotted to each team) of individual 100-km-long segments to be selected from our total coverage. Originally, we had anticipated the necessary cartographic processing of the Exabyte tapes to be provided by the USGS Astrogeology facility in Flagstaff. With reorganization and downsizing of the USGS, technical support has been very limited. As a consequence, to date almost all of our work has been done with the CDs, except for the few examples of the enhanced data that we now have in areas of special interest.

For future geologic studies of the Sahara, the coregistration process of Landsat and SIR data described by Davis et al. (1993), would be useful. It has the advantage of treating the black and white radar data like a shaded relief underlay so that the Landsat surface data and the radar subsurface data can be seen simultaneously. Another useful approach for regional geologic mapping might be the use of the C-band ERS-1 data to fill in the gaps in the current L-band coverage. This task, however, is beyond the scope of this project.

The SIR-C Mosaic. The most useful data for paleodrainage study in the Eastern Sahara lies near the center of the cross point of the ascending and descending orbits, about 80 km west of Bir Safsaf at approximately the center of the Western Desert (fig. 1). North of the Gilf Kebir towards Siwa, the Great Sand Sea with its linear dunes dominates the terrain. The linear dunes are essentially opaque to the radar, and appear as long dark streaks on the images. The interdune flats, however, exhibit considerable detail, and this area is now being studied by our Co-Investigator Stefan Kroepelin, Cologne University, Germany. The Limestone Plateau region to the northeast of our prime coverage has received only a cursory examination because of the presence of a paleokarst topography and a severe wind erosion overprint that has carved the bedrock into grooved terrain and fields of yardangs, which are not relevant to the paleodrainage studies. Although very good coverage of the Eastern Desert and the Sinai was also obtained, given the focus of the project these data have not been analyzed.

Our focus has been on the types and the distribution of paleodrainages in and around the southern part of Egypt where the ascending and descending data takes overlap (Fig. 3). This region is characterized by a gently dipping sequence of Paleozoic to Mesozoic clastic rocks, mostly of continental origin, that once were covered by Tertiary marine rocks of the type still preserved in the Limestone Plateau. Underlying these rocks are the igneous and metamorphic rocks of the African Shield of Precambrian age; these basement rocks locally crop out near Bir Safsaf, to the west around the Oweinat uplift, and also at several localities along the southern edge of the Limestone Plateau. A north-south trending structural basin stretches from near Dakhla to (fig. 1) south of Bir Misaha. The bedrock within this structural and topographic low is mostly masked by an almost ubiquitous sand blanket, which varies from less than a centimeter in thickness to more than the imaging depth of the L band (estimated to be as much as two or three meters under ideal conditions; Schaber et al., this volume). Myriads of small curvilinear to dendritic channels and broad alluvial valleys, less than 1 km to tens of kilometers wide, which we have termed the "radar rivers," are visible on the radar images as areas that are dark relative to their surroundings.

In addition, the mosaic of SIR-C images also reveals many large, interconnected, irregular to roughly linear dark areas that extend from north to south about 300 km, with projections to the east and west as much as 200 km wide. These dark patches, which are opaque to the radar, represent areas of deep sand with little backscatter, and show no new geologic data. In contrast, at least 50% of the area covered by the mosaic shows ground newly visible to the radar and thus amenable to geologic analysis.

Fig. 3 near here

For example, a roughly 100-km-wide, circular dark area that lies just to the north of Selima Oasis is noteworthy because of the presence of distinctive, north trending, previously unknown buried channels (Fig 4). The channels lie a few meters or less below the flat, featureless surface sand sheet described by Haynes (1982), and Breed et al. (1987) (fig. 5). The westernmost channel follows closely the route of the Darb el Arbain, the famous camel caravan trading route that runs from the Darfur region in the Sudan to the Nile River at Asyut (fig. 1). The camel route coincides with the buried river for about 150 km to an area south of the scarp at Bir Kiseiba (fig. 5B). Selima Oasis was an important watering and rest area, and it lies within a depression that contains late Quaternary to Holocene lake beds (Haynes et al, 1979). The coincidence of the north flowing river with the trade route across this now featureless waste (fig. 5) has archeological significance. Was the knowledge of the subsurface river, which locally might be tapped by shallow wells, passed down through the ages to the desert people? At the very least, moisture retained in the buried river channel would have supported vegetation for forage, which in turn would have indicated the location of shallow groundwater to the ancient caravans.

Fig. 4, 5A, B near here

Improved geologic maps of Egypt and surrounding areas will come from the use of the CD ROM data. Its utility in the Eastern Sahara can be seen in Figure 6. This sketch map shows the extent of the radar-dark areas (opaque, and containing no new information), and the bright terrain that either is exposed at the surface or thinly buried so that it backscatters data from the shallow subsurface. When compared with the Geological Map of NW Sudan and SW Egypt (Jas et al., 1988), one sees how much more terrain is revealed by the radar as opposed to the Landsat images that were used in the preparation of the map.

Small patches of brighter terrain within the dark areas typically represent either wind-roughened bedrock outcrops or patches of rough duricrust. Our 1984 field experience showed that trenching through the crusts commonly revealed clean river sands below. Faintly brighter and locally diffuse areas commonly represent near-surface features that probably are worth a shallow pit or two for stratigraphic or structural data.

Fig. 6 near here

As one becomes familiar with the use of these images in the field, together with conventional imagery and GPS equipment for accurate field location, better regional mapping techniques will emerge. The radar will be useful not only for new regional geologic studies, but also will be invaluable as an archeological prospecting tool because it appears that early Man in Egypt was as much a riverine creature as he is today (McHugh et al., 1988 a,b; 1989).

Significance of the "Radar Rivers." Of critical importance to understanding the significance of the sand-buried rivers, first discovered by means of SIR-A, to the Cenozoic history of Egypt was firm stratigraphic evidence that the so called "Protonile" of Tertiary age did not exist (Bown et al. 1982). Unfortunately, the concept of a north flowing, ancestral, pre-late Miocene Nile has been invoked time and again to explain puzzling aspects of the geomorphology of Egypt. This cloud in the minds of early workers has unfortunately been revived several times during the last two decades, first by Kortlandt (1980). In a literature review unhindered by field data, he called upon the Protonile to explain the Tertiary ecology of the deltaic and fluvial deposits of the Fayum Depression on the west side of the present Nile. The second revival, (Burke and Welles, 1989) was a critique of our SIR-A/B interpretations, which they based primarily on the supposed existence of a north flowing, late Paleogene and early Neogene Nile River in Egypt, which would have acted as a barrier to drainages from the Red Sea Range finding their way to the west across such a Protonile valley (Also referred to as the Paleonile by some workers).

The first revival (Kortlandt, 1980) was thoroughly discredited by the careful field work of Bown et al. (1982) on the stratigraphy, sedimentology, and paleontology of the Jebel Qatrani Formation of Oligocene age that crops out on the west and north flanks of the Fayum. They state that there is no geologic evidence for the existence of a "Protonile" to explain the Tertiary ecology of deltaic and fluvial deposits of the Fayum Depression on the west side of the Nile Valley. They likened the concept to a romantic holdover from the nineteenth century! An elegant and detailed account by Bown and Kraus (1988) provides additional stratigraphic and provenance data. They show that the Jebel Qatrani and its underlying and overlying units were derived, first by deltaic sedimentation followed by marine regression and then deposition by meandering streams emanating from the Red Sea Range, where Mesozoic, Paleozoic and igneous basement rocks were already exposed in Oligocene and Miocene time. These provenance conclusions are supported by dominantly westward paleocurrent directions and the presence of diagnostic clasts such as felsic feldspars and chert grains from units that lie below the Fayum stratigraphically, but which were eroded from the highland to the East.

The second revival (Burke and Welles, 1989) was addressed in the paper of Issawi and McCauley (1992), which asserted that Egypt had been drained by a succession of at least three different river systems since the Tethys regression in the late Eocene and that a single master stream (the "Protonile") did not exist during the Tertiary. These newly described drainage systems competed with one another for survival and the rivers with the overall gradient advantages replaced, over time, the earlier less efficient systems. Competition took place in response to tectonic uplifts, sea level changes and climate variations. The present landscape of Egypt was produced by the combined geomorphic effects of these old drainage systems. The modern Nile in Egypt, that is, a north-flowing river connected to the Ethiopian highlands, dates only to the very late Pleistocene, ~.25 Ma (the age of the earliest Nile deposits--the Dandara Formation--at Qena: Paulissen and Vermeersch, 1987). It is made up locally of parts of the prior rivers and it is neither an antecedent nor a superimposed stream, and it is not the descendant of the imaginary "Protonile."

We attribute the "radar rivers" of the western Desert to downstream parts of the Qena System, which consisted of powerful subsequent streams that flowed southward across eastern Egypt along the west flank of the Red Sea Range and then south-westward along the southern edge of the Limestone Plateau into northwest Sudan (Issawi and McCauley, 1992). Beginning at about 24 Ma, in Neogene time, volcanism and accelerated uplift occurred all along the Red Sea area, with more uplift at the northern margin of the Red Sea rift than farther to the south (Garfunkel, 1988). This uplift, which gave Wadi Qena a gradient advantage to the south, is evidenced by the bedrock stratigraphy exposed in the wadi, where as one proceeds northward (upslope within the wadi) into higher terrain, progressively older rocks are exposed.

The Qena System was chiefly responsible for producing the cliffs along the southeastern edge of the Limestone Plateau and the broad, southward opening pediplain that extends from near Aswan into the Sudan (fig. 1). We believe (McCauley, 1986a) that elements of the Qena system were able to flow to the west into the Chad Basin and from there into the Niger River System. Discussions with our French Coinvestigator, Hugues Faure (Faculty of Sciences at Luminy, Marseilles, France) indicate that reddened stream gravels of probable Tertiary age (the "Continental Terminal") have been reported in the Marhdogoum Gap (fig. 1), one of the probable late-stage streamcourses to Lake Chad. There is cartographic evidence as well as Spot and Landsat data indicating that a major river once flowed through this breach in the Erdi Plateau. Unfortunately, this area was out of range of SIR-C because of mission constraints. Another probable outlet for the later stages of the Qena System is the E-W trending Mourdi Depression, a structural trough that as late as Quaternary time allowed the passage of waters back and forth between the Chad and Nile basins (Pachur and Rottinger, in press).

As a topographic divide between Egypt and Chad began to develop, by uplift and later volcanism in the Darfur region, the Qena System was forced progressively more westward around the outlying extremities of the Eocene marine deposits left by the former Tethys Sea. In so doing, the Qena System formed, by northward lateral erosion, the south facing edge of the Limestone Plateau (the Sinn al-Kaddab scarp). (The sandy, beach deposits of the Tethys Sea are well preserved about 200km farther to the south of the Sinn al-Kaddab scarp across the Red Sea in Saudi Arabia. We examined exposures of these rocks near the town of Turabah (Lat 21_ 30'N, Long 41_ 45'E), where they have escaped erosion by Tertiary rivers). The Qena System must have flowed on successively lower erosion surfaces before it was beheaded by the headward (southward) eroding Nile System. The dying remnants of the Qena System are marked by playa and mud pan deposits near the base of the scarp, and by buried channels such as those we examined south of Dungul Oasis (described later in this report).

The SIR missions added little or our knowledge of the origin of the Nile. It had already been established that this system came into being as a result of a spectacular drop in the level of the Mediterranean of about 1000 m or more (Hsu, 1972) during late Messinian time (about 6 Ma). A recent comparison between the Nile and the Colorado (McCauley and Breed, 1993), which applies the results of much new work by others in the United States, indicates that the old ideas of superposition and antecedence are as untenable for the Nile as they are for the Colorado. The sequence of events that led to the modern Nile, i.e., a river system that flows through Egypt from south to north, has been described by Said (1981, 1993). He defines a series of late Tertiary and Quaternary Niles that varied in regime as a result of tectonism and climate variations. The oldest of these was the Eonile, a very energetic river (pirate stream) that cut by headward erosion the steep walled, youthful appearing valley, through the Limestone Plateau between Cairo and the vicinity of Luxor. Deep entrenchment also occurred as far as Aswan, thereby beheading the old Qena System of its Red Sea Range headwaters. With the onset of aridity in Egypt at about the beginning of the Pleistocene, reversals of many of the former Qena channels took place during episodic pluvial events, as these aggraded channels attempted to adjust to the new regional gradients. The modern Nile with its headwaters deep in equatorial Africa is an exotic river that continues to be the topic of much research (Paulissen and Vermeersch, 1987) although it is a very recent arrival on the scene, having developed its present set of tributaries only about 200 to 250 ka.

General Types of Paleodrainages. Our earlier work, based on SIR-A data, led to the recognition of three types of paleochannels, which we referred to as RR-1, RR-2 and RR-3, i.e. Radar River 1 etc. (McCauley et al., 1986a). The RR-1 type consists of broad aggraded valleys, filled with alluvium, which are 10 to 20 km wide and can be traced for hundreds of kilometers. The RR-2 type are set into the flood plains of the RR-1 type, and they are braided, as at Safsaf, or consist of bunches of short stream-like features that are darker than their surroundings. In contrast to the RR-1 type they are shallow, relatively recent features. The RR-3 type are long, narrow (< 1 km wide), slightly sinuous channels incised into the local bedrock; these often can be seen on Landsat or Spot images, or on air photos.

The three types of Radar Rivers bear some similarity to the three basic types of channels recognized by Schumm (1985): 1) The bedrock channel that is well fixed in position and stable over time; 2) The semi-controlled channel that is fixed only locally by indurated alluvial deposits or resistant bedrock. Changes in these channels are determined by the strength of bank materials; 3) The alluvial channel that is inset into a floodplain and subject to pattern changes as the sediment load and the discharge vary. Of the alluvial channels, there at least three subtypes, which are defined by their degree of sinuosity (meandering being an end member that occurs under very limited conditions), their braiding or anastamosing, and their anabranching. The latter is a type of pattern in which the river is divided by islands whose width is three times or more greater than the average channel width (this pattern is seen south of Safsaf, fig. 12). Anastamosing channels are distinct from anabranched by multiple channels, which have major secondary channels that separate and then rejoin to form a complex network within the main channel. Stream patterns reveal clues as to their internal shape. As an example, Schumm (1985) points out that in the Great Plains region of the United States, narrow and deep channels are generally sinuous whereas relatively straight channels tend to be wide and shallow. These insights may be applicable to the paleodrainages of the Western Desert, which at times during the Cenozoic shared some climatic and general geologic similarities with the southern Great Plains.

Analysis of the region around the orbital crossovers reveals two general paleodrainage characteristics: 1) low sinuosity and no evidence of real meandering; 2) most channels bear no relation to the Nile, which was earlier thought to be the Cenozoic master stream of northeast Africa. The only clear examples of drainages on the West Bank that are graded to the Nile are Wadis Tushka and Kalabsha and the dendritic complex of small channels described below.

Types of Paleodrainages on SIR-C Images. A survey of the paleodrainage patterns that can be seen on the SIR-C images of the Western Desert shows that the multitude of partly to completely invisible fluvial features fall into six general classes. These six distinctive patterns are about half the number of those generally described in the literature of fluvial geomorphology. As pointed out by Schumm (1985), an understanding of these patterns is the key to understanding their environment of deposition and climatic history. All of these newly revealed riverine features deserve more study with the enhanced SIR-C images and in the field by systematic backhoe trenching. The types of paleochannels we recognize are:

Type 1: Dendritic patterns, as in Figure 7. A nexis of interfingered channels (fig. 7A), on the west bank of the Nile, lies about 75 km northwest of the third cataract near the Temple of Sulb (Lat 20_ 15'N, long 30_ 15'E). The channels extend over an area about 30 km long and 20 km wide, and trend toward the east. Individual channels are about 0.5 km wide. Another pattern far to the west of the Nile (fig. 7B) lies at Lat 22_ 25'N, Long 26_ 30'E, within an area that on Landsat images shows no visible channels. We believe these channels were carved in either late Pleistocene or early Holocene time. The presence of dendritic patterns, in general, suggests relatively homogeneous, easily erodible materials.

Fig. 7 A,B near here

Type 2: These consist of broad, flat-floored valleys, some of which are 30 to 40 km wide and hundreds of kilometers long. An example is shown in Figure 8, located near Lat 20_ 30'N, Long 26_ 10'E. The banks of these valleys have subtle topographic expression ( tens of meters; parts of these features are recognizable on some Landsat images). Their major trends appear to be to the south but some that lie to the west of the Misaha structural trough trend eastward and widen as they disappear beneath the thick, radar-opaque sand sheets of this large depression. These features may be controlled in part by regional structures which have been exploited by streams that follow the present regional hydrologic gradient in Egypt (to the northeast). The east trending arm of the valley seen in this illustration shows disconnected dendritic paleochannels on its floor. Relations with the surrounding lighter-toned, flat lying strata, mapped by Jas et al. (1988) as the Lakia Formation of Permian Age, suggest that these particular dendritic channels could be exhumed features as old as late Paleozoic. The existence of such very old channels, some of which are still active, has been well documented in Australia (van de Graaff et al., 1977; Stewart et al., 1986), but is yet to be established in the Eastern Sahara.

Fig. 8 near here

Type 3: This pattern is made up of either isolated examples or scattered arrays of small, straight-to-weakly-sinuous channels that are common throughout the northern and eastern part of the area near Lat 22_ 15'N, Long 27_ 45'E, (fig 9). Individual channels are typically a few hundred meters wide, and as long as 20 to 30 km. Usually only two stream orders can be seen on the CD data, that is, a main channel with several tributaries. These arrays occur mostly in terrain of intermediate radar tone. According to Schumm (1985), straight channels tend to be wide, shallow features that are unstable when on older alluvial surfaces. When incised into bedrock they are stable and can be narrow and deep. Many of these channel clusters could be interpreted as the result of local cloudbursts or heavy seasonal rains. Those seen in the bedrock must be of a more permanent nature.

Fig. 9 near here

Type 4: These features consist of dark, broad, 10 to 20 km wide southwesterly trending valleys that have little or no surface expression, but that can be traced for hundreds of kilometers (fig. 9). They were first seen on the SIR-A data, and are marked by flat, firm, highly trafficable surfaces. These valleys are bordered by discontinuous, subtle highs that are bright on the radar but lie only a few meters above the present, aggraded valley floor. These bright areas mark the edges of former river banks or terraces as described for Wadi Arid (McCauley et al., 1986a). They also contain a series of living surfaces that range from Paleolithic to Neolithic as described by McHugh et al. (1988, a,b; 1989). Of note is that on their north sides, these banks locally exhibit thick layers of groundwater calcrete that show as very bright patches on the radar and probably indicate a reversed hydrologic gradient at the time these old rivers were active. Type 4 valleys differ from Type 2 because they are less clearly influenced by structure, and appear almost fully aggraded.

Similar broad, diffuse valleys occur along the base of the Limestone Plateau south of the Dungul Oasis (fig 10). The base of this part of the Sinn al-Kaddab scarp is marked by numerous west trending channels that debouch from the scarp and which are plotted on 1:250,000 scale maps of Egypt. A few kilometers farther to the south, at about Lat 23_ 16'N, Long 31_ 40'E, concealed beneath the sand sheets of the Atmur el Kibeish, similar broad, diffuse channels are recognized on the SIR-C enhanced pictures. The presence of fluvial sediments there was confirmed by means of shallow hand-dug pits during a brief reconnaissance after the November, 1996 GSE Centennial Conference. We believe these channels to be the resequent remnants of west flowing elements of the Qena System of Middle to early Late Miocene age, as originally described by Issawi and McCauley (1992).

Figs. 10, 11 near here

Type 5: Braided to anabranching channels (fig. 12) are singularly prominent south of the Safsaf oasis (Lat 22_ 35'N, Long 29_ 20'E), and have been described previously by McCauley et al. (1986a), and Schaber et al. (1986). SIR-C shows these in greater detail than did SIR-A. The islands within the braided channels consist of calcified patches of reeds (bullrushes) that produce a strong radar return at both the L- and C-band wavelengths. Some of these reeds have been dated at 18 ka (Szabo and others, 1990). These are cut through by a multiplicity of anastomosing dark channels indicative of high bedload conditions. Several anabranches are present on the south side of the complex. The channels are typically 200 to 500 m wide and the braided complex extends over an area some 40 km long, and 20 km wide. Individual channels are cut into the uppermost layers of the alluvial fill that has aggraded the diffuse, deep valley (Type 4). A model showing the relation of these channels to the deeper valley fill is given in McCauley et al. 1986b, Fig 20).

Fig. 12 near here

Type 6: Anomalous patterns are prevalent in the northern part of the Darfur Province in the Sudan, north of Merga (fig. 1) at about (Lat 20_ 30', Long 27_ 30') in strata mapped as the Lakia Formation of Permian age (Jas et al, 1988), Many short (tens of km) linear, disconnected channel-like features are present in arrays that appear to feed into more traditional appearing fluvial valleys, which trend northward toward the Misaha Trough (Figure 13). The features appear to be similar to those ascribed to the process of "piping," which is widespread in the semi-arid Southwestern United States. The piping process is the result of percolating waters above a soluble underground surface, producing caving and the formation of narrow conduits or pipes through which soluble or suspended material is moved. It is distinct from sapping, which occurs in rocks exposed to the atmosphere, and is more akin to but distinct from the development of karst. The piping features were not originally exposed to the air, and are not the result of surface runoff. The implication is that the surface of the Lakia Formation has been lowered without any trace of a surface drainage net remaining except for the crenulate channels that have taken over the former paths of the underground pipes.

Fig. 13 near here

We now have in hand about three dozen specially processed images derived from the SIR-C exabyte tapes by the U.S. Geological Survey Data Facility, Flagstaff. About half of these have been distributed for detailed analyses to foreign Coinvestigators in Germany and France.

Moonsoonal ? paleodrainage in Selima Basin. Here we will discuss another example from the Selima area, which provides a new insight to the climatic history of the Eastern Sahara. Figure 14 shows the east side of the semicircular Selima basin (the west side is shown in Figure 4) which is floored by lower Cretaceous units called the Abu Ballas Formation and the Gilf Kebir Formation (Jas et al., 1988). The enhanced image shows a completely different and complex fluvial terrain beneath the featureless, surficial sand sheet originally described by Haynes (1979), and depicted here as Figure 5. The existence of an unconformity beneath the thin, flat sand sheet was noted by Breed et al. (1987), who recognized the presence of fluvial sediments below the eolian cover, and who argued against the inclusion of these distinctive subjacent materials in the same stratigraphic unit with the overlying eolian blanket as done by Haynes (1979).

Fig. 14 near here

The C-band shows an extremely dense drainage net as well, if not better, than the L-band, indicating that this complex pattern lies only about a meter or less below the surface. The drainage net is especially apparent on large-format prints of the European Radar Satellite (ERS-1) C-band images. This pattern shows a great deal of disorder and complex, curving tributaries that do not fit the traditional drainage patterns of the literature. Although they approach a badlands topography, the valleys and divides do not have enough relief to qualify as such. The closest analog we have found to this pattern is described by Baker (p 6, fig 1-3, in Short and Blair, 1986). The much larger, disordered patterns seen in the Flinders and Diamantina rivers in southwestern Queensland, Australia, were formed by monsoonal flooding in early 1974. The disordered pattern observed in the Selima Basin may be evidence for episodic monsoonal events sometime prior to the onset of the present aridity in this part of Egypt.

Expanding Wadi Kufra--Desert Hydrology The ground water oasis of Kufra, of historical and strategic importance, has undergone extraordinary ground water exploitation during the past several decades (fig. 15). Expensive and time consuming geophysical techniques have been used in this development program (Ahmed, 1978). It has long been known, from the presence of numerous watering holes, that the oasis lies along the course of a major north trending wadi that has been traced in a general way hundreds of kilometers to the south of the Kufra Oasis by Pachur (1993). SIR-C reveals the broad, previously unmapped East Branch, a tributary that emanates from the Gilf Kebir-Oweinat area (about Lat 22_ N, Long 24_E) and trends northwest at least 300 km to join the trunk stream of the Kufra System just to the south of the circular irrigation fields near the oasis (figs. 15, 16). Pachur (1993) began his tracing of a large paleodrainage system (shown as Wadi Yangara on a map by Pachur and Altmann, in press) near Lat 22_. This headwaters area is just below the Libya-Chad border, and consists of an array of channels that debouch from the east flank of the great Tibesti volcanic edifice. A mosaic of side-lapped SIR-C data takes 60.1, 108.4, and 156.1 allows us to trace the Kufra channel much farther south, via circuitous bends to an area about 30 km west of the village of Ounianga-Kebir, near Lat 19_N, Long 20_E in Chad and about 260 km south of the border with Libya (fig. 16). Of note is the reversal of the junction angles of the tributaries from northward to southward along newly recognized parts of the Kufra System south of Maatan as Sarra, which suggest that one or more episodes of stream capture may have diverted north-flowing Kufra drainage southward into the Bodele Depression, which lies now at least 150 m lower than the presumed elbow of capture. One of these probable capture points is illustrated in Figure 17. We believe the Wadi Kufra coverage to be a dramatic demonstration of the SIR systems' capability in the field of desert regional hydrology. The SIR-C data provides the basis for extended studies of this major wadi system, which formerly flowed northward, parallel to the Nile but about 1000 km away, across the Western Desert, but which is now entirely dry along all but its headwaters in the Tibesti Mts.

Figs. 15, 16, 17 near here

The fifteen year period of SIR investigations in Northeast Africa has led to a much improved understanding of the geomorphic evolution of the Western Desert of Egypt and the relatively minor role of the Nile in this history. The nonredundant SIR-C coverage of Egypt, as seen in the CD-ROM data, shows convincingly that collected runoff in rivers, streams, and locally lakes was the principal agent of erosion that produced the major elements of the Egyptian landscape. The present-day eolian sand sheets and dunes of various types are only a thin blanket dating mostly to the Late Pleistocene and Holocene Epochs. Although southwestern Egypt lies within the most arid part of the Earth, and has a powerful and persistent wind regime, the radar shows the eolian landscape to be superficial. We estimate that 50 percent more of the total surface of Egypt can now be seen through the "eyes" of SIR-C than was previously accessible. The ability of the SIR system to see through the Sahara sands depends mostly on the wavelength of the signal (the longer the wavelength, the deeper the penetration), the dryness of the sand, and the roughness of the surface (Schaber et al., this volume). Rough surfaces produce strong backscatter of the signal and are bright, whereas smooth surfaces either reflect the signal away from the spacecraft or absorb it, producing dark tones.

The Paleodrainage Experiment has exceeded our expectations, for not only are different types of paleodrainages seen but much new shallow subsurface stratigraphic and structural information has emerged. Both L-band and C-band imaging have proven to be effective tools in those parts of the Earth's deserts that are masked by thin sand deposits. Important new field information is also emerging from the Western Desert and surrounding areas by our German Co-Investigators and their colleagues dealing with the Quaternary. These works include: Gabriel and Kroepelin, 1984, Pachur et al, 1987, Pachur and Kroepelin, 1987, Pachur et al, 1990, Thorweihe and Schandelmeier, 1993, and Pachur and Rottinger, 1996. Their work demonstrates that a variety of lakes and swamps of small to large size, existed during the various wet periods, in a wide variety of low places. These data are totally consistent with the SIR-C data. F. Rottinger (this volume), using SIR-C data to identify small, residual, wind-eroded knobs of lake sediments, shows that these knobs are the last remaining physical evidence of the existence of a previously unknown lake of at least 7000 km². Rottinger further suggests that the use of all three bands may be a tool for detection of subsurface moisture, as also demonstrated by Schaber et al. (1997).

The data base achieved by SIR-C is now available to all workers interested in North Africa. We are pleased to have participated in this project and to have had the opportunity to work in long collaboration with the Geological Survey of Egypt. We have received much cooperation during the ten visits or field expeditions to Egypt that were part of the SIR effort. We now leave a legacy of these radar data to those who will follow, knowing that they will be useful for many years in the geological and archeological exploration of this grand and important place where we have done just a little more than peel off parts of its surface layers, thereby getting a better peek at its history.

REFERENCES

Baker, V.R., 1986, Introduction: Regional landforms analysis, in Short, N.M., and Blair, R.W., Jr. (eds.), Geomorphology from Space: National Aeronautics and Space Administration (NASA) SP-486, p. 1-26.

Bown, T.M., Kraus, M.J., Wing, S.L., Fleagle, J.G., Tiffney, B.H., Simons, E.L., and Vondra, C.F., 1982, The Fayum Primate Forest revisited: Journal of Human Evolution, v. 11, p. 503-560.

Bown, T.M., and Kraus, M.J., 1988, Geology and paleoenvironment of the Oligocene Jabel Qatrani Formation and adjacent rocks, Fayum Depression, Egypt: U.S. Geological Survey Professional Paper 1452, 60 p.

Breed, C.S., McCauley, J.F., and Davis, P.A., 1987, Sand sheets of the eastern Sahara and ripple blankets on Mars: in Frostick, L., and Reid, I. (eds.), Desert Sediments: Ancient and Modern: Geological Society (London) Special Publication 35, p. 337-359.

Burke, Kevin, and Welles, G.L., 1989, Trans-African drainage system of the Sahara: Was it the Nile?: Geology, v. 17, p. 743-747.

Davis, P.A., Breed, C.S., McCauley, J.F., and Schaber, G.G., 1993, Surficial geology of the Safsaf region, south-central Egypt, derived from remote-sensing and field data: Remote Sensing of Environment, v. 46, p. 183-203.

Gabriel, B., and Kroepelin, Stefen, 1984, Holocene lake deposits in northwest Sudan: in Zinderen Bakker, E.M.V., and Coetzee, J.A. (eds.), Paleoecology of Africa and surrounding islands, v. 16, p. 295-299.

Haynes, C.V., Mehringer, P.J., Jr., and Zaghloul, E.S.A., 1979, Pluvial lakes of north-western Sudan: Geographical Journal, V. 145, part 3, p. 437-445.

Haynes, C.V.,Jr., 1982, Great Sand Sea and Selima Sand Sheet, Eastern Sahara: Geochronolgy of desertification: Science, v. 217, p. 629-633.

Issawi, Bahay, and McCauley, J.F., 1992, The Cenozoic rivers of Egypt: The Nile problem: in Friedman, R., and Adams, B. (eds.), The followers of Horus: Studies in memory of M.A. Hoffman: Egyptian Studies Association Publication no. 2, Oxbow Monograph 20, Oxbow Books, Oxford, England, p. 121-138.

Jas, C., Klitzsch, Eberhard, Schandelmeier, Heinz, and Wycisk, P., (eds.),1988, Geologic Map of NW-Sudan and SW-Egypt, Sheet NF 35, Jebel'Uweinat; Scale 1:1,000,000, Special Research Project 69 (E3-G1) of the German Research Foundation, Bonn.

McCauley, J.F., Breed, C.S., and Schaber, G.G., 1986a, Megageomorphology of the Eastern Sahara: Proceedings of the Second Spaceborne Imaging Radar Symposium, National Aeronautics and Space Administration (NASA) Jet Propulsion Laboratory Publication 86-26, p. 25-35.

McCauley, J.F., Breed, C.S., Schaber, G.G., McHugh, W.P., Issawi, B., Haynes, C.V., Jr., Grolier, M.J., and El Kilani, A., 1986b, Paleodrainages of the Eastern Sahara - The radar rivers revisited (SIR A/B implications for a mid-Tertiary trans-African drainage system): Institute of Electrical and Elctronic Engineers (IEEE) Transactions on Geoscience and Remote Sensing GE24, no. 4: 624-648.

McCauley, J.F., G.G. Schaber, C.S. Breed, M.J. Grolier, C.V. Haynes, B.Issawi, C. Elachi, and R. Blom, 1982, Subsurface valleys and geoarchaeology of the Eastern Sahara revealed by Shuttle Radar: Science, v. 218, n. 4516, p. 1004-1020.

McHugh, W.P., Breed, C.S., Schaber, G.G., McCauley, J.F., and Szabo B.J.,

1988a. Acheulian sites along the "radar rivers", southern Egyptian Sahara, Journal of Field Archaeology 15: p. 361-79.

McHugh, W.P., McCauley, J.F., Haynes, C.V., Breed, C.S., and Schaber, G.G., 1988b, Paleorivers and geoarchaeology in the southern Egyptian Sahara: Geoarchaeology, v. 3, no. 1, p. 1-40.

Pachur, H.-J., 1993, Palaodrainage systeme im Sirte-Becken und seiner Umrahmung: Wurtzburger Geograph. Arb., v. 87, p. 17-34.

Pachur, H.-J., and Altmann, Norbert, in press, the Quaternary (Holocene, ca 8000 BP), Plate 17, in Schandelmeier, H., Reynolds, P.O., and Semtner, A.-K., p. 83-137.

Pachur, H.-J., and Braun, G., 1980, The paleoclimate of the central Sahara, Libya, and the Libyan Desert; In Zinderen Bakker, E.M.V., and Coetzee, J.A. (eds.), Paleoecology of Africa and the surrounding islands, v. 12: Sahara and surrounding seas (ed: Sarnthein, M., Seibold, E., and Rognon, P.), Balkema, Rotterdam, p. 351-363.

Pachur, H.-J., and Hoelzmann, 1991, Paleoclimatic implications of late Quaternary lacustrine sediments in western Nubia, Sudan: Quaternary Research, v. 36, p. 257-276.

Pachur, H.-J., and Kroepelin, Stefan, 1987, "Wadi Howar: Paleoclimatic evidence for an extinct river system in the southeastern sahara," Science v. 237, p. 298-300.

Pachur, H.-J., Kroepelin, Stefan, Hoelzmann, Philipp, Goschin, Michael, and Altmann, Norbert, 1990,"Late Quaternary fluvio-lacustrine environments of Western Nubia," Berliner Geowissenschaftliche Abhandlunger, Reihe A,v. 120.1, p. 203-260.

Pachur, H.-J., Roper, H.P., Kroepelin, Stefan, and Goschin, Michael, 1987, Late Quaternary hydrography of the Eastern Sahara: Berliner Geowissenschaftliche Abhandlungen, Reihe A, Band 75.2, p. 331-384.

Pachur, H.-J., and Rottinger, Frank, in press, SIR-C/X-SAR detection of sediemtns from a large extended paleolake in the eastern Sahara (NW-Sudan) - preliminary results: Remote Sensing of Environments.

Paulissen, Etienne, and Vermeersch, P.M., 1987, Earth, Man, and Climate in the Egyptian Nile Valley during the Pleistocene: in A. Close (Ed.), Prehistory of North Africa: Southern Methodist Univ. Press, Dallas, Texas, p. 29-67.

Schaber, G.G., McCauley, J.F., and Breed, C.S., 1997, this volume, The application of SIR-C/X-SAR images to mapping of sand-blanketed basement rocks in southern Egypt, 15 p. and 17 figs.

Schaber, G.G., McCauley, J.F., Breed, C.S., and Olhoeft, F.R., 1986, Shuttle imaging radar: physical controls on signal penetration and subsurface scattering in the Eastern Sahara: Institute of Electrical and Electronic Engineers (IEEE) Transactions on Geoscience and Remote Sensing, GE-24, no. 4, p. 603-623.

Schaber, G.G., McCauley, J.F., and Breed, C.S., 1997, The use of multifrequency and polarimetric SIR-C/X-SAR data in geologic studies of Bir Safsaf, Egypt: Remote Sensing of Environment, v. 59, no. 2, p. 337-363.

Schumm, S.A., 1985, Patterns of alluvial rivers: Annual Review Earth Planetary Sciences, v. 13, p. 5-27.

Stewart, A.J., Blake, D.H., and Ollier, C.D., 1986, Cambrian river terraces and ridgetops in central Australia: Oldest persisting landforms? Science, v. 233, p. 758-761.

Szabo, B.J., McHugh, W.P., Schaber, G.G., Haynes, C.V., Jr., and Breed, C.S., 1989, Uranium-series dated authigenic carbonates and Acheulian sites in southern Egypt: Science, v. 243, p. 1053-1056.

Tactical Pilotage Chart, 1981, 1:5,000,000 scale, Edition 5 GSGS, London.

Thorweihe, Ulf and Schandelmeir, Heinz, (eds.),1993, Geoscientific research in Northeast Africa: Rotterdam, Balkema, 776.

Van de Graff, W.J.E., Crowe, R.W.A., Bunting, J.A., and Jackson, M.J., 1977, Relict Early Cainozoic drainages in arid Western Australia, Zeitschrift f¸r Geomophologie, v. 21, no. 4, p. 379-400.

1. Index map to localities in Eastern Sahara discussed in text. (Modified from Issawi and McCauley, 1992).

2. Map of SIR-C swaths that provide non-redundant coverage of three-fourths of Egypt. Numbers refer to datatakes listed in Table 1.

3. Mosaic of SIR-C images (Survey data, from CD-ROMs) that cover most of the area between 20_N, 26_E to 24_N, 26_E and between 20_N, 31_E and 24_N, 31_E.

4. Specially-processed SIR-C image (from Exabyte tape) shows previously unknown, buried river channel (arrows) extending from Selima Oasis (fig. 1) across basin. Channel is coincident with route of historic camel caravans (Darb el Arbain, fig. 5B), which may have utilized it for shallow ground water and/or forage. (SRL-1 d.t. 146.4, L-HH, segment 14)

5. Ground photographs illustrate the extraordinarily flat, bland surface of the Western Desert: A Absence of surface expression of underlying fluvial topography in the Selima Sand Sheet (photograph taken in 1982 in northwest Sudan, south of Bir Misaha); B Darb el Arbain north of Selima (photograph taken in 1978 south of Bir Kiseiba).

6. Dotted areas on sketch map of mosaicked images (fig. 3) overlain on the best available regional geologic map (Jas et al., 1988) represent areas that are radar-opaque (dark, showing little or no geologic features). Clear areas represent areas that are newly accessible to geologic analysis using the radar data.

7. Type 1 paleodrainages: A Dendritic channels on West Bank in Nile Valley in Egypt (arrow) a few km north of border with Sudan (1a on fig. 3). B Dendritic channels (arrow) in apparent playa basin (1b on fig. 3) near Gilf Kebir Plateau (arrow). (SRL-1, d.t. 76.10. segment 17 L-HH); SRL-1, d.t. 146.4, segment 15 L-HH)

8. Type 2 paleodrainages: Broad, flat-floored valleys with subtle topographic expression; some dendritic elements possibly exhumed. (SRL-2, d.t. 44.11, L-HH, segment 32)

9. Type 3 paleodrainages: Scattered arrays of relatively small, weakly sinuous channels (black arrow) (3 on fig. 3) and Type 4 paleodrainages: Large, broad valleys, fully aggraded (white arrow) (4 on fig. 3). (SRL-2, d.t. 28.10, L-HH, segment 32)

10.Type 4 paleodrainage valley aligned parallel to Sinn al-Kaddab scarp south of Dungul Oasis (fig. 1), revealed on SIR-C specially processed data. (SRL-2, d.t. 50.4, C-HH, segment 12). Field investigations (fig. 11) indicate this valley, now wholly obscured by sand, was part of the pre-Nile Qena system described by Issawi and McCauley (1992).

11.Ground photograph in Type 4 valley south of Dungul Oasis , at west arrow (fig. 9). View north across aggraded valley toward Sinn al-Kaddab scarp.

12.Type 5 paleodrainage: Anabranching channels (arrow) inset into fill of type 4 valley south of Bir Safsaf (5 on fig. 3). (SRL-1, d.t. 130.4, L-HH, segment 9)

13.Type 6 paleodrainage: Anomalous channels, probably formed by piping (6 on fig. 3). Segment is 100 km long. (SRL-2, d.t. 28.10, L-HH, segment 30)

14.Specially processed SIR-C image (from Exabyte tape) shows complex, disordered pattern of buried paleodrainage channels that may represent episodes of monsoonal flooding under earlier, wetter climatic conditions. Large paleochannel in northern part of basin lacks tributaries on west side, suggesting slope to west in basin at that time. (SRL-1, d.t. 130.4, L-HH, segment 10)

15.Wadi Kufra, fully revealed for the first time on SIR-C images (A), compared with Landsat data (B) for this area. East branch of wadi system (D) was not known to exist prior to SIR-C. It can be traced farther to the southeast (toward the Gilf Kebir) on ERS-1 images). Note irrigated fields (A) at junction of two branches in the Kufra Oasis (SRL-1, d.t. 172.21, L-band, HH, segments 12, 13; Landsat 1, path 194, rows 43, 44)

16. Map of Wadi Kufra, based on SIR-C images, shows new details of regional paleodrainage system. Arrow marks probable point of capture of the north-flowing river from the Tibesti Mts (W. Yangara of Pachur and Altmann, in press) by headward erosion of another paleodrainage system that flowed southward toward the Bodele Depression, south of Faya Largeau in the Chad Basin. Capture point is in wide radar-dark section of wadi that probably represents a former lake. Most of these drainage patterns are now concealed beneath a cover of sand sheets and dunes, but wells produce groundwater at Bishara and Maatan as Sarra, along the camel track (dashed line) that follows the old wadi.

17. SIR-C images show details of area "X" along the course of Wadi Kufra (at arrow, fig. 16) where the trunk stream, which exits form the Tibesti Mts. as Wadi Yangara, was apparently captured by a south-flowing, headward-eroding stream. (SRL-2 d.t.60.10, C-band, HH, segments 28, 29)

Table 1. Twenty-two SIR-C datatakes (d.t.'s) provide areal coverage of

three-fourths of Egypt (shown on map, figure 4).

	d.t.	(km)	CD#
SRL-1 Descending	146.4	67	FL1-47
	130.	56	FL1-21
	66.5	57	FL1-32
	50.4	44	FL1-50
	34.5	30	FL1-05
	18.31	75	FL1-19
SRL-1 Ascending	172.22	54	FL1-23
	76.1	47	FL1-23
	28.1	62	FL1-24
	12.1	30	FL1-08
	156.1	50	FL1-41
	44.11	30	FL1-26
SRL-2 Descending	66.5	58	FL2-38
	50.4	76	FL2-65
	34.51	35	FL2-36
	18.31	77	FL2-26
SRL-2 Ascending	108.1	45	FL2-57
	76.1	7	FL2-67
	44.11	50	FL2-66
	28.1	66	FL2-41
	12.1	50	FL2-27
	60.1	80	FL2-37