We focus on the Afar and Ethiopian portions of the rift system (Figure 3, Figure 5 that comprise the northern 1000 km of the EARS.  This rift transects the broad Ethiopian plateau, which developed above a Paleogene mantle plume (Schilling et al. 1992; Ebinger & Sleep 1998).  The evolution of this area has been well summarized by WoldeGabriel et al. (1990) and Mohr (1978).  Recent studies in the Red Sea and Gulf of Aden show that flood basalts were erupted across a ~1000 km diameter region at ~30 Ma (Baker et al. 1996) prior to, or concurrent with, the initiation of rifting (Menzies et al. 1997).  There is little evidence for extension in the Afar/Ethiopian rift, the third arm of this triple junction, until considerably later (at ~18 Ma; WoldeGabriel et al. 1990).  Since then the Main Ethiopian Rift has linked northwards into southern Afar and into oceanic spreading in the southern Red Sea (e.g. Makris et al., 1991).  We focus on this rift specifically because it is both young and tectonically relatively uncomplicated, as the underlying continental lithosphere was last deformed in the Pan-African orogeny (c. 600 Ma, Teklay et al. 1998).

The Ethiopian rift is a region of sub-E-W extension where extensional velocities (4 mm/yr) are much less than in Afar (20-30 mm/yr) (Bilham et al. 1999).  Rift basins are asymmetric, and bounded by steep border faults showing >3 km throw (Fig. 2B).  The en echelon arrangement of the ‘new’ magmatic segmentation shows little correlation with the older border fault pattern (Figure 4).  Geodetic data show that ~80% of the strain across the rift is accommodated over a <30 km-wide zone of magmatic construction, although seismicity attests to some deformation outside this zone (Bilham et al. 1999; Ayele, 2000).  Effective-elastic-plate thickness varies along the Ethiopian rift from 17±2 km in southern Ethiopia, to 8±2 km in the central segment of the Main Ethiopian Rift, to 6±2 km in southern Afar (Hayward & Ebinger 1996), indicating that in our study area we are approaching the elastic lithosphere thickness of 6 km that is characteristic of slow-spreading ridges (Blackman & Forsyth, 1991).  Some lavas from the neovolcanic zone show high degrees of crustal contamination, implying localized heating, a hot lower crust, and long crustal-residence times (Hart et al. 1989; Schilling et al. 1992); geothermal drilling shows temperatures >350oC at depths of 2 km, with a rapid decrease away from the neovolcanic zone (EIGS, internal report 860-451).

The contrast with Afar is extreme, because the crust there is probably composed almost entirely of new magmatic material (Prodehl & Mechie 1991; Mohr 1992).  Primarily Miocene volcanism occurred on the Ethiopian plateaus flanking Afar, while in Afar volcanism, and possibly faulting, began at about 25 Ma. Beginning about 4 Ma, extensive basaltic volcanism covered most of Afar and obscured the geologic record of events.  Since that time, the history of northern and eastern Afar is more a story of sea-floor spreading than continental rifting.  Thus the northern portion of the main Ethiopian rift provides an opportunity to study the processes at work in continental rifts, and also documents the critical transition from continental to oceanic rifting.

4. Previous studies; the state of knowledge of the Main Ethiopian Rift
4.1 Seismological Studies
Previous teleseismic, seismic refraction, and gravity studies show significant features, but these data are too sparse to properly constrain even the 2D lithospheric structure.  The seismic studies in 1972 (Fig. 6) of Berckhemer et al. (1975) and Ruegg (1975) provide some basic information on crustal and, to some extent, uppermost mantle structure. However even with modern ray-tracing techniques, these refraction profiles in Afar were interpreted as indicating nearly 1D structures (Fig. 6) due to the very small number of shots and receivers (Makris and Ginzberg, 1987).  These models suggested thinned 25-km-thick crust underlain by a 10-km-thick layer with anomalously low upper-mantle P-wave velocities above apparently normal mantle (Figure 2B, Figure 6).  The 1972 study provides low-resolution information not because of any difficulty of working in Ethiopia, but simply because 30 years ago there were available exactly 15 seismic recorders, and shots were fired at separations of 100 to 300 km.  [In 2003 we will have available c. 600 seismometers and fire shots at c. 50 km spacing.]  Perhaps the most valuable aspect of these studies from our perspective is that they demonstrate that excellent signal-to-noise ratio is available out to 300 km (maximum observed in 1972) (Fig. 6) with 300 to 1600 kg dynamite shots (we will use 1000 and 2000 kg) and that the cross-over distance for PmP reflections is 150-200 km in the rift (Berckhemer et al. 1975).  We will even use some of the identical shot-point locations for which we already know the source character (Burkhardt and Vees 1975a).  Velocities near 6.0 km/s appear only in about the upper 5 km (beneath the sedimentary and volcanic layer) of the crust within Afar, and velocities of about 6.9 km/s are reached at depths of only c. 10 km (Searle, 1975).  The thinned crust beneath the Ethiopian rift and Afar is underlain by an anomalous upper mantle with low compressional seismic velocities (7.4 to 7.8 km/s), which Makris and Ginzberg (1987) infer to be caused by high temperature in the mantle (Fig. 6).  On the adjacent Ethiopian plateau, a typical continental crustal velocity structure was observed by Berckhemer et al. (1975), Bonjer et al. (1970), Hebert and Langston (1985), and Makris and Ginzberg (1987).  Analysis of inter-station Rayleigh wave phase velocities between Addis Ababa and Djibouti indicates anomalously low S-wave velocities beneath Afar between ~60 and 130 km depth, implying the presence of partial melt (Knox et al. 1999).

It is possible that the transition from relatively thick crust (-40 km) on the plateau to thinned, rifted crust is as abrupt in Ethiopia as it is in Kenya (Braile et al., 1994).  However the velocity structure in the Kenya rift does not show evidence of the large extension and thinning of the continental crust and pervasive modification by magmatism implied by the Ethiopian rift and Afar crustal seismic models.  Lower-crustal seismic velocities in the Ethiopian rift and Afar are about 6.9 km/s and the lower crustal layer is present at a relatively shallow depth (about 10 km) as compared to the lower crust of Kenya.  Significant crustal thinning has apparently occurred in the Lake Turkana area of the Kenya rift but lower-crustal seismic velocities (about 6.4 km/s) are not unusually high.  Where the rifted crust is thick in Kenya, a 6.9 km/s layer is present at the base of the crust and represents underplated material (Hay et al., 1995).  Mohr (1989) suggests that new igneous material has replaced the bulk of the original Afar crust and the geophysical data appear to support this view.

In the Kenya rift, where there is the greatest amount of deep structural information, largely from seismic studies (Keller et al., 1994a, Birt et al., 1997), the rift is relatively narrow (about 70 km wide) at the surface, and crustal thinning and anomalously low upper-mantle velocities are confined to a relatively narrow zone directly beneath the rift.  Crustal models derived from recent seismic studies in the Kenya rift show no evidence for major intrusions into and densification of the upper crust beneath the rift as had been inferred primarily on the basis of gravity models.  The Kenya rift also displays considerable east-west symmetry, at least at the latitude of the KRISP90 seismic survey (Braile et al., 1994).  Along-rift (north-south) variations in volcanism, structural style and crustal and upper mantle seismic velocities in the Kenya rift are substantial (Mechie et al. 1994; Keller et al., 1994b) and attest to the need for the proposed axial seismic profile in Ethiopia. In fact, the crust thins by 15 km in <300 km in a fashion that correlates with topography and independent measures of extension (Morley et al., 1992).

4.2 Gravity field of Ethiopia
The gravity field of Ethiopia has been studied since the early 1960’s (Gouin and Mohr, 1964; Mohr and Rogers, 1966) in order to determine the region’s general crustal structure.  The few existing gravity studies have concentrated on determining the crustal structure of the main Ethiopian rift (Makris et al., 1970; Searle and Gouin, 1972; Mahatsente et al., 1999, 2000; Jentzsch et al., 2000) and the Afar region of northeastern Ethiopia (Makris et al., 1972).  The earlier studies (Makris et al., 1970; Makris et al., 1972; Searle and Gouin, 1972) determined that the main Ethiopian rift and the northern Afar region coincided with a regional Bouguer gravity minima that is related to crustal attenuation, while the southern Afar region is less attenuated and may contain continental crust.  The most detailed study to date has been by Mahatsente et al. (1999, 2000) and Jentzsch et al. (2000) who collected additional gravity data and merged them with all available gravity data into one dataset to produce the most complete Bouguer gravity anomaly map to date.  The Mahatsente et al. study [the three papers by this group essentially come to the same conclusions, the difference in the papers are the three-dimensional (3D) inversion routines that they developed] uses 3D inverse theory to determine crustal thickness and density distribution beneath the main Ethiopian rift.  A zone of crustal thinning (31 km) exists under the main rift coinciding with inferred intrusions within the upper crust.  Southeast and northwest of the main rift, the crust thickens to approximately 50 km.  The lateral and vertical size of the intrusions led Mahatsente et al. (2000) to conclude that the main Ethiopian rift is underlain by continental crust.  Note that C. Ebinger (pers. comm., 2001) re-measured elevations at 3 IGSN gravity base stations in the rift valley in 10/01, and found USCGS heights in error by 93-140 m!  This means that rift valley anomalies will be more positive (implying more crustal thinning in the northern Main Ethiopian Rift).  Corrections to gravity values are now being made, and one outcome of US and UK EAGLE will be a significantly improved gravity field over the Ethiopian Rift.

A Bouguer gravity anomaly map (Figure 7) has been constructed from 3467 data points obtained from the National Imaging and Mapping Agency and additional data from C. Ebinger (pers. comm.).  The Mahatsente et al. (2000) study had other data in the rift valley, but the total number of stations was not stated.  Clearly obvious is the gravity minimum (~-250 mGal) that corresponds to the main Ethiopian rift with gravity maxima to the northwest and east.  The present dataset is concentrated along the main roads.  This lack of data inhibits the interpretation of smaller scale features and thus the overall structure of the Ethiopian rift.  Nonetheless, intriguing features are already visible.  Superimposed on the northeastwards-increase of 150 mGal along our proposed 400-km axial refraction profile (Fig. 7), there are along-axis jumps of 20-25 mGal associated with the tips of several of the magmatic segments that will be studied by the 3D fan profiling.  These rapid jumps must represent either steps in the Moho, perhaps indicating punctuated along-axis extension, or mantle segmentation driving the emplacement of the magmatic centers.  The gravity jumps could possibly represent successive increases in the mafic component of the crust, but we find this less plausible, in part because we note that the young, highly mafic, off-axis volcanism lacks a gravity signature at this scale.  We expect our combination of 2D and 3D seismic inversions to more tightly constrain the sources of the gravity anomalies and hence to allow a clearer interpretation of their significance

4.3 Petrologic and Geochemical Studies
The variety and differences in the volume of magmatic products extruded along the branches of the EARS are striking.  There have been many discussions and estimates of the actual volumes (e.g., Baker et al. 1972; Williams 1982; Karson and Curtis 1989; Mohr 1989; MacDonald 1994).  Recently, Mohr (1992) compared the initial flood volcanism in the Ethiopian, Kenya, and Western rifts, and estimated eruptive volumes of 400,000 km3 (including Afar and Yemen), 220,000 km3, and 100,000 km3 respectively.  Though volume estimates for the Western rift vary widely (from 2000 km3 [Barberi et al. 1982]) to 100,000 km3 [Mohr 1992]), the volume is clearly subordinate to that in other rift sectors.  The large volume of flood basalts in Ethiopia (primarily in Afar) is consistent with the suggestion that the original crust has been largely replaced by magmatic products (Mohr 1989).  In Kenya, there are significant volumes of phonolites in addition to the flood basalts.  Through an integrated geophysical and petrological analysis (Hay et al. 1995), these phonolites have been linked to the velocity models derived from the KRISP experiments.  The high-velocity lower crust observed has been interpreted as an underplate that served as the source of the phonolites.

The Ethiopian rift is an impressive magmatic province on a global scale.  It is characterized by largely bimodal magmatism, with transitional basalts that straddle the tholeiitic-alkaline boundary and associated trachyte-pantellerite ignimbrites and lavas.  The magmatic evolution of the Ethiopian rift is a key issue to understanding rift processes.  Peak eruption volumes associated with the rift occurred during the period 32-21 Ma (Mohr & Zanettin 1988), even though volcanism started earlier.  In the main Ethiopian rift, WoldeGabriel et al. (1990) recognize six episodes of volcanic activity starting in the mid-Oligocene or earlier, and they all involve the eruption of transitional basalts.  In Afar, volcanism started during late Oligocene-early Miocene (ca. 25 Ma) with a second episode in late Miocene (7 Ma) and oceanic crust development from Pliocene to present.  Magmatism is ongoing and includes historic mafic and felsic lavas - with minor intermediate volcanics - concentrated along faults within the rift axes.  Late Quaternary magmatic segments in the northern Main Ethiopian Rift are separated by narrow fault zones (the Wonji fault belt; Mohr 1962, 1967) and defined by volcanic centers located within the central rift valleys, including the caldera complexes of Dofan and Fantale (Sabure segment), and Kone and Bosetti (Nazret segment) (Figure 4).  Lavas from these areas comprise the Wonji Group (e.g., Boccaletti et al. 1999; Chernet & Hart 1999).  Pre-Pleistocene volcanism occurred within the same structural segments but outside the modern axial magmatic zones as manifest, for example, in the Bishoftu basalts and the silicic center of Bede Gebabe (Figure 4).  Field observations suggest that the locus of extension and magmatic activity relocated to the narrow axial segments in Pleistocene time, with fissures and small cinder cones becoming active ~1.6 Ma (WoldeGabriel et al. 1990; Boccaletti et al. 1999).

Most of the geochemical research conducted in this region has surveyed the older Wonji Group lavas outside the modern axial rift (Hart et al. 1989; Gasparon et al. 1993; Chernet et al. 1998; Barbiero et al. 1999; Boccaletti et al. 1999; Chernet & Hart 1999).  Investigation of the abundant felsic lavas (which make up >75% of the eruptives) indicate they may be derived by fractionation in shallow crustal reservoirs (Gasparon et al. 1993; Chernet & Hart 1999), which is consistent with a high degree of underplating beneath the rift axis.  Investigations of the mantle source regions of young Ethiopian basalts have documented the presence of at least three components: depleted mantle, Ethiopian lithospheric mantle, and an asthenospheric plume (e.g., Betton and Civetta 1984; Hart et al. 1989; Barrat et al. 1990, 1993; Chernet & Hart 1999; Pik et al. 1999; Furman et al. 2000, 2001).
 
 

RETURN to EAGLE proposal

RETURN to Simon Klemperer's Home page

Last updated 5/02