The Afar Volcanic Province within the East African Rift System

Edited by G. Yirgu, C. J. Ebinger and P. K. H. Maguire


The seismically and volcanically active East African Rift System is an ideal laboratory for continental break-up processes: it encompasses all stages of rift development. Its northernmost sectors within the Afar volcanic province include failed rifts, nascent seafloor spreading, and youthful passive continental margins associated with one or more mantle plumes. A number of models have been proposed to explain the success and failure of continental rift zones, but there remains no consensus on how strain localizes to achieve rupture of 125–250 km thick plates, or on the interaction between the plates and asthenospheric processes. This collection of papers provides new structural, stratigraphic, geochemical and geophysical data and numerical models needed to resolve fundamental questions concerning continental break-up and mantle plume processes. It focuses on how mantle melt intrudes and is distributed through the plate, and how this magma intrusion process controls along-axis segmentation and facilitates break-up.

  1. Page 1
    1. Page 9

      Although the East African Rift (EAR) System is often cited as the archetype for models of continental rifting and break-up, its present-day kinematics remains poorly constrained. We show that the currently available GPS and earthquake slip vector data are consistent with (1) a present-day Nubia–Somalia Euler pole located between the southern tip of Africa and the Southwest Indian ridge and (2) the existence of a distinct microplate (Victoria) between the Eastern and Western rifts, rotating counter-clockwise with respect to Nubia. Geodetic and geological data also suggest the existence of a (Rovuma) microplate between the Malawi rift and the Davie ridge, possibly rotating clockwise with respect to Nubia. The data indicate that the EAR comprises at least two rigid lithospheric blocks bounded by narrow belts of seismicity (<50 km wide) marking localized deformation rather than a wide zone of quasi-continuous, pervasive deformation. On the basis of this new kinematic model and mantle flow directions interpreted from seismic anisotropy measurements, we propose that regional asthenospheric upwelling and locally focused mantle flow may influence continental deformation in East Africa.

    2. Page 23

      The development of the Afar rift–rift–rift triple junction is analysed from the viewpoint of the Nubia, Arabia and Somali plate kinematics. A variety of constraints allow definition of a range of kinematic models that approximate well to the plate motions with a resolution of a couple to tens of kilometres. Rigid plate kinematics probably cannot resolve smaller motions. The size and location of the new area that opened between the major plates is inferred from plate kinematics and this provides a framework in which to assess the structural development that accommodates plate separation.

      The development of the Afar region was complicated by the presence of microplates – the Danakil and Aisha blocks – which results in a complex plate boundary geometry. This led to local deformation that does not directly reflect the divergence of the major plates, e.g. rotations of microplates and of minor blocks about vertical axes and strike–slip faulting. The opening of new area was accommodated by various crustal growth and accretion mechanisms, e.g. building of thick new igneous crust, normal seafloor spreading, and/or crustal stretching. Thus, plate motions by themselves do not determine the development of the plate boundaries, as this is strongly influenced by other factors such as lateral variations of the rates of magma supply (e.g. away from, and over, the Afar plume).

      The plate boundaries changed – e.g. the Gulf of Aden spreading centre propagated westward c. 2 Ma ago and normal seafloor spreading began along portions of the Red Sea axis since c. 5 Ma ago – while there were no resolvable changes in the plate motions. Such changes therefore signify a reorganization of the way in which plate divergence and addition of new area is accommodated: diffuse extension may give way to separation along a narrow spreading centre, or new plate boundaries may form at the expense of other boundaries that became inactive.

    3. Page 43

      The initiation of the Afro-Arabian Rift System on three nearly straight segments occurred shortly after massive amounts of basalt poured out of the triple junction of those segments in Afar. The synchroneity of magmatism and rifting may reflect the fact that normal continental lithosphere is too strong to rift without magmatic dyke intrusions and the straightness of rifts reflects a localized source for the magma feeding those dykes. Simple relations are derived for the minimum extensional force needed for lithosphere cutting dyke intrusions as functions of the density structure and thickness of the lithosphere. As long as the density contrast between continental crust and magma is small compared to the density contrast between the mantle and magma, then the force needed to rift scales with the square of the thickness of the mantle lithosphere. Thus, continental regions with normal-thickness lithosphere may rift when reasonable levels of extensional force and sufficient magma are available. Very thick mantle lithosphere may not rift at levels of force that are likely to arise on Earth. Two main sections of the Afro-Arabian Rift System, the Red Sea and the Ethiopian Rift, appear to have developed as magma-assisted rifts in normal continental lithosphere. The northern and southern ends of the system are bounded by regions of very thick mantle lithosphere where dykes could not open. In the south, the Tanzanian Craton, with normal-thickness crust and a very deep lithospheric root, was also not split by the rift. In the north, the rift opened along a nearly straight line from the centre of the flood basalt province 2000 km to the edge of the Mediterranean Sea. The old oceanic lithosphere of this margin may be no thicker than the adjacent continental lithosphere of Egypt, but the thinner oceanic crust means that Mediterranean lithosphere may be too thick and dense to rift magmatically. The role of magma in the third branch, the Gulf of Aden, is not so clear given the lack of syn-rift dykes.

    4. Page 55

      The rifting of continents and eventual formation of ocean basins is a fundamental component of plate tectonics, yet the mechanism for break-up is poorly understood. The East African Rift System (EARS) is an ideal place to study this process as it captures the initiation of a rift in the south through to incipient oceanic spreading in north-eastern Ethiopia. Measurements of seismic anisotropy can be used to test models of rifting. Here we summarize observations of anisotropy beneath the EARS from local and teleseismic body-waves and azimuthal variations in surface-wave velocities. Special attention is given to the Ethiopian part of the rift where the recent EAGLE project has provided a detailed image of anisotropy in the portion of the Ethiopian Rift that spans the transition from continental rifting to incipient oceanic spreading. Analyses of regional surface-waves show sub-lithospheric fast shear-waves coherently oriented in a northeastward direction from southern Kenya to the Red Sea. This parallels the trend of the deeper African superplume, which originates at the core–mantle boundary beneath southern Africa and rises towards the base of the lithosphere beneath Afar. The pattern of shear-wave anisotropy is more variable above depths of 150 km. Analyses of splitting in teleseismic phases (SKS) and local shear-waves within the rift valley consistently show rift-parallel orientations. The magnitude of the splitting correlates with the degree of magmatism and the polarizations of the shear-waves align with magmatic segmentation along the rift valley. Analysis of surf ace-wave propagation across the rift valley confirms that anisotropy in the uppermost 75 km is primarily due to melt alignment. Away from the rift valley, the anisotropy agrees reasonably well with the pre-existing Pan-African lithospheric fabric. An exception is the region beneath the Ethiopian plateau, where the anisotropy is variable and may correspond to pre-existing fabric and ongoing melt-migration processes. These observations support models of magma-assisted rifting, rather than those of simple mechanical stretching. Upwellings, which most probably originate from the larger super-plume, thermally erode the lithosphere along sites of pre-existing weaknesses or topographic highs. Decompression leads to magmatism and dyke injection that weakens the lithosphere enough for rifting and the strain appears to be localized to plate boundaries, rather than wider zones of deformation.

    1. Page 77

      The major and trace element and radiogenic isotope compositions of basalts from throughout the East African rift system are reviewed in the context of constraints from previous geophysical studies. The data indicate the presence of two mantle plumes, the East African and Afar plumes, which dynamically support the East African and Ethiopian plateaus. Rifting across the plateaus is accompanied by the generation of large volumes of basaltic magma and associated evolved derivatives. Relatively few mafic magmas have an unambiguous Afar mantle plume signature, notably the MgO-rich picrites and ankaramites from the 29–31 Ma Ethiopian traps, and the most recent basalts (<5 Ma) from Afar. The Eocene Amaro basalts from southern Ethiopia also have a plume source but their lower source temperatures and isotopic characteristics are distinct from those of Afar. The remaining basalts from the Ethiopian rift, and throughout the Kenya and Western rifts, have a lithospheric source region as reflected in both radiogenic isotope and trace element characteristics. The Amaro basalts are suggested as the first manifestations of magmatism from the East African plume; subsequent magmatic activity being represented by progressively younger episodes further south through Turkana, Kenya and into Northern. Tanzania, as the African plate migrated north. Despite their clear lithospheric characteristics, U-series data on geologically recent basalts from the axis of the Kenya rift show that they were generated in a dynamic melting regime. Melting is effected when lithospheric mantle heats up and becomes incorporated into the convecting mantle, hence leading to greater degrees of lithospheric thinning than are indicated by extension across individual rift basins.

    2. Page 95

      Primitive recent mafic lavas from the Main Ethiopian Rift provide insight into the structure, composition and long-term history of the Afar plume. Modern rift basalts are mildly alkalic in composition, and were derived by moderate degrees of melting of fertile peridotite at depths corresponding to the base of the modern lithosphere (c. 100 km). They are typically more silica-undersaturated than Oligocene lavas from the Ethiopia–Yemen continental flood basalt province, indicating derivation by generally smaller degrees of melting than were prevalent during the onset of plume head activity in this region. Major and trace element differences between the Oligocene and modern suites can be interpreted in terms of melting processes, including melt-induced binary mixing of melts from the Afar plume and those from three mantle end-member compositions (the convecting upper mantle and two enriched mantle sources). The Afar plume composition itself has remained essentially constant over the past 30 million years, indicating that the plume is a long-lived feature of the mantle. The geochemical and isotopic compositions of mafic lavas derived from the Afar plume support a modified single plume model in which multiple plume stems rise from a common large plume originating at great depth in the mantle (i.e. the South African superplume).

    3. Page 121

      Structural and geochronological relations indicate that the felsic rocks at the top of the Oligocene flood basalt sequences in the Afar volcanic province were erupted coevally with the initial rifting in the Red Sea and Gulf of Aden. In this study, we use the newly established volcanic— tectonic history to examine the geochemical evolution with time of felsic volcanics as rifting has progressed to seafloor spreading in the southern Red Sea and northern Main Ethiopian Rift. Geochemical analyses (major and trace elements; Sr, Nd and O isotopic compositions) of syn-rift rhyolites ranging in age from 28 to 2.5 Ma indicate that the rhyolites can be derived from mantle-sourced basaltic magma through fractional crystallization accompanied by variable amounts of crustal contamination (e.g. 87Sr/86Sr = 0.70489−0.70651; 143Nd/144Nd = 0.51254−0.51283; δ18O = +4.5 to +6.4‰). The input of crust tends to increase with time, which suggests the weakening and heating of the crust in response to lithospheric thinning and magma injection in the past c. 30 Ma. These results support earlier structural and thermomechanical models for rift formation in the southern Red Sea rift and the younger, less-evolved northern Main Ethiopian Rift system.

    1. Page 133

      The May 2000 earthquake cluster, around 10° N and 41° E in southern Afar, has been studied using high quality data from 12 temporary and permanent broadband seismic stations deployed in the area. 140 earthquakes have been located using P- and S-wave arrival times, a well-constrained velocity model, and a double-difference location algorithm. Source mechanisms and moment magnitudes for the four largest events (M > 4) have been obtained from moment tensor inversion. There is no clear alignment of the epicentres along a fault zone; however, the events are clustered slightly southeast of Mount Amoissa along WNW—ESE extension of the Ayelu—Amoissa (Abida/Dabita) lineament. Focal mechanisms show fault motion along WNW—ESE to east—west striking normal faults, with extension oblique to the orientation of the Main Ethiopian Rift. The non-double-couple components of the source mechanisms range from 18–25%, suggesting that the seismic activity is of tectonic origin and not volcanic. Source depths are ≤7 km, in good agreement with estimates of the elastic thickness of the Afar lithosphere. We suggest that the Gewane earthquake swarm represents remnant strain accommodation along a previous line of weakness in southern Afar related to the separation of Arabia from Africa because the focal mechanisms show north—south extension similar to many of the events in central Afar at the triple junction where Arabia is presently rifting away from Africa.

    2. Page 143

      Active deformation within the northern part of the Main Ethiopian Rift (MER) occurs within approximately 60 km-long, 20 km-wide ‘magmatic segments’ that lie within the 80 km-wide rift valley. Geophysical data reveal that the crust beneath the <1.9 Ma magmatic segments has been heavily intruded; magmatic segments accommodate strain via both magma intrusion and faulting. We undertake field and remote sensing analyses of faults and eruptive centres in the magmatic segments to estimate the relative proportion of strain accommodated by faulting and magma intrusion and the kinematics of Quaternary faults. Up to half the ≤10 km-long normal faults within the Boset—Kone and Fantale—Dofen magmatic segments have eruptive centres or extrusive lavas along their length. Comparison of the deformation field of the largest Quaternary fault and an elastic half-space dislocation model indicates a down-dip length of 10 km, coincident with the seismogenic layer thickness and the top of the seismically imaged mafic intrusions. These relations suggest that Quaternary faults are primarily driven by magma intrusion into the mid- to upper crust, which triggers faulting and dyke intrusion into the brittle upper crust. The active volcanoes of Boset, Fantale and Dofen all have elliptical shapes with their long axes in the direction N105, consistent with extension direction derived from earthquake focal mechanisms. Calderas show natural strains ranging from around 0.30 for Boset, 0.55 for Fantale, and 0.94 for Dofen. These values give extension strain rates of the order of 0.3 microstrain per year, comparable to geodetic models. Structural analyses reveal no evidence for transcurrent faults linking right-stepping magmatic segments. Instead, the tips of magmatic segments overlap, thereby accommodating strain transfer. The intimate relationship between faulting and magmatism in the northern MER is strikingly similar to that of slow-spreading mid-ocean ridges, but without the hard linkage zones of transform faults.

    3. Page 165

      We report the first palaeomagnetic results from the Main Ethiopian Rift (MER), the northernmost sector of the East African rift system. This part of the MER shows an along-axis tectono-magmatic segmentation pattern similar to that of slow-spreading mid-ocean ridges, which developed during the past 1.9 Ma. The aims of our palaeomagnetic, structural and geochronological studies are to test plate kinematic models for the right-stepping, en echelon 60–80 km-long magmatic segments. Twenty palaeomagnetic sites were sampled on either basalt or ignimbrite outcropping in the region adjacent to, and within, the < 1.9 Ma-old tectono-magmatic segments of Gademsa—Koka, Boset and Fentale—Dofan. Five K—Ar age determinations were made to bracket the age of units studied in the palaeomagnetic analyses. The natural remanent magnetization intensity possibly exhibited a unimodal distribution with a value of 6.6 A/m (σ = 5.6 A/m) for the basalts and a bimodal distribution with magnetization intensity of 0.69 A/m (σ = 0.55 A/m) and 0.03 A/m (σ = 0.02 A/m), statistically similar to values from previous studies in the Afar triple junction zone (e.g. Kidane et al. 1999, 2002). Progressive heating, alternating field analysis, and susceptibility vs. temperature measurements indicated unblocking temperature ranging between 300 °C—600°C for basalts and between 500 °C—660 °C for ignimbrites, suggesting the magnetic mineralogy to be titanomagnetite and magnetite for the former and magnetite and titanohematite for the latter. Palaeomagnetic measurements using both TH and AF technique revealed quasi-single component of magnetization with viscous remanent magnetization (VRM) on a few samples. Principal component analysis and statistical averaging resulted in an overall mean palaeomagnetic direction of (Ds = 2.3°, Is = 7.8°, α95 = 7, K = 26.9, N = 17) which is statistically identical to the expected direction (D = 1.9°, I = 13.5°, α95 = 2.5, K = 105.6, N = 32) from the Apparent Polar Wander Path reference curve for Africa at 1.5 Ma (Besse & Courtillot 2003). The angular difference between the observed and expected directions above with their uncertainty is calculated to be 0.4° ± 7.5°. These results indicate that the Late Pliocene—Pleistocene rocks of the MER in the studied region do not suffer vertical axis rotation, arguing against transtensional and seafloor-spreading—transform kinematic models. We suggest that magma intrusion, rather than large offset faults, produce the right-stepping, en echelon magmatic segments of the MER, which is at the transition from continental to oceanic extension.

    4. Page 185

      A levelling line consisting of 43 benchmarks was established between the towns of Wolenchiti and Metehara in the northern part of the Main Ethiopian Rift in July 1995 by the Geophysical Observatory of Addis Ababa University and the Ethiopian Mapping Authority. The measurement was repeated in November 2003 on 30 of the surviving benchmarks. In both epochs of measurement, the standard accuracy attained is ± 4 mm √K corresponding to a first order levelling, where K is the inter-benchmark distance in kilometres. The line crosses the northern and southern parts of the Nazret (Boset—Kone) and Sabure (Fantale—Dofen) magmatic segments respectively where 80% of rift deformation is believed to be localized. In eight years, interval height differences ranging from +3 mm to −22 mm are found along the line with the maximum subsidence rate of 2.8 mm a−1 corresponding to the Kone-Gariboldi volcanic complex in the northern part of the Nazret magmatic segment. On the other hand, at the eastern end of the line, despite the existence of large fissures and an expanding lake in the vicinity suggesting possible significant subsidence, on the contrary relatively small vertical deformation is found. The strong subsidence measured at the Kone-Gariboldi volcanic complex is interpreted to be due to remnant processes of subsidence at these volcanic centres following withdrawal of magma in the recent past. Regarding the eastern end of the line where Lake Beseka is located, the result is particularly important in verifying that the rapid expansion of the lake associated with elevation increase of 40 cm a−1 could not be attributed to tectonic subsidence or uplift in the region.

    5. Page 191

      The Wonji Fault Belt (WFB), Main Ethiopian Rift, forms a network of faults oriented NNE—SSW with a Quaternary direction of extension oriented c. N95° E. Faults are spaced between 0.5 and 2 km, show a fresh steep scarp, recent activity and slip rates of up to 2.0 mm a−1. This high value of deformation along the rift floor with respect to the plate separation rates suggests that most of the active strain could be accommodated by magma-induced faulting within the rift. However, the mountain front morphology associated with a displacement of 300–400 m since the Middle Pleistocene, tilted-blocks, brittle-seismic fault rock fabric and historical earthquakes with M>6 support a tectonic origin of the Asela boundary fault. Therefore, we propose a model that considers the possible coexistence of both magmatic deformation at the rift floor and brittle faulting at the rift margin. We also report the data relative to a GPS network installed in December 2004, along two transects across the WFB, between Asela and the Ziway Lake.

    6. Page 209

      The Turkana magmatic rift (Northern Kenya) initiated at 45 Ma as one of the nucleation zones of rifting in the East African Rift. It forms an anomalously broad-rifted zone (c. 200 km) striking with a north—south trend and lying within a NW—SE topographic depression, floored on both sides of the Turkana area by Cretaceous rifts in the Sudan and Anza plains. From a compilation of available data, combined with newly acquired remote sensing and DEM dataset, we propose a five-stage tectono-magmatic model for the Turkana rift evolution (45–23 Ma; 23–15 Ma; 15–6 Ma; 6–2.6 Ma and 2.6 Ma—Present). The corresponding ‘restored’ maps clearly show the changing spatial distribution of magmatism and fault/basin network with time, hence supplying some clues about dynamics of continental extension. First-order basement-rooted transverse faults zones are identified and their influence on nucleation and propagation of strain is demonstrated, whereas the role of magmatic ‘soft-spots’ as concentrating strain is minimized. Blocking of deformation as well as rift jump and lateral transfer of strain are discussed in relation to various types of fault interaction (dip direction, strikes and acute/ obtuse angle of the intersecting faults). The causal links between rift nucleation ‘cells’ and inherited transverse weakness zones in the Turkana rift might also exist elsewhere along the eastern branch of the East African Rift, hence suggesting a complex and discontinuous mode of rift propagation.

    1. Page 239

      Crustal structure beneath the GEOSCOPE station ATD in Djibouti has been investigated using H-κ stacking of receiver functions and a joint inversion of receiver functions and surface wave group velocities. We obtain consistent results from the two methods. The crust is characterized by a Moho depth of 23 ± 1.5 km, a Poisson’s ratio of 0.31 ± 0.02, and a mean Vp of c. 6.2 km s−1 but c. 6.9–7.0 km s−1 below a 2–5 km-thick low-velocity layer at the surface. Some previous studies of crustal structure for Djibouti placed the Moho at 8 to 10 km depth, and we attribute this difference to how the Moho is defined (an increase of Vp to 7.4 km s−1 in this study vs. 6.9 km s−1 in previous studies). The crustal structure we obtained for ATD is similar to crustal structure in many other parts of central and eastern Afar. The high Poisson’s ratio and Vp throughout most of the crust indicate a mafic composition and are not consistent with models invoking crustal formation by stretching of pre-existing Precambrian crust. Instead, we suggest that the crust in Afar consists predominantly of new igneous rock emplaced during the late syn-rift stage where extension is accommodated within magmatic segments by dyking. Sill formation and underplating probably accompany the dyking to produce the new and largely mafic crust.

    2. Page 253

      The northern Main Ethiopian Rift captures the crustal response to the transition from continental rifting in the East African rift to the south, to incipient seafloor spreading in the Afar depression to the north. The region has also undergone plume-related uplift and flood basalt volcanism. Receiver functions from the EAGLE broadband network have been used to determine crustal thickness and average Vp/Vs for the northern Main Ethiopian Rift and its flanking plateaus.

      On the flanks of the rift, the crust on the Somalian plate to the east is 38 to 40 km thick. On the western plateau, there is thicker crust to the NW (41–43 km) than to the SW (<40 km); the thinning taking place over an off-rift upper mantle low-velocity structure previously imaged by traveltime tomography. The crust is slightly more mafic (Vp/Vs ~ 1.85) on the western plateau on the Nubian Plate than on the Somalian Plate (Vp/Vs ~ 1.80). This could either be due to magmatic activity or different pre-rift crustal compositions. The Quaternary Butajira and Bishoftu volcanic chains, on the side of the rift, are characterized by thinned crust and a Vp/Vs > 2.0, indicative of partial melt within the crust.

      Within the rift, the Vp/Vs ratio increases to greater than 2.0 (Poisson’s ratio, σ > 0.33) northwards towards the Afar depression. Such high values are indicative of partial melt in the crust and corroborate other geophysical evidence for increased magmatic activity as continental rifting evolves to oceanic spreading in Afar. Along the axis of the rift, crustal thickness varies from around 38 km in the south to 30 km in the north, with most of the change in Moho depth occurring just south of the Boset magmatic segment where the rift changes orientation. Segmentation of crustal structure both between the continental and transitional part of the rift and on the western plateau may be controlled by previous structural inheritances. Both the amount of crustal thinning and the mafic composition of the crust as shown by the observed Vp/Vs ratio suggest that the magma-assisted rifting hypothesis is an appropriate model for this transitional rift.

    3. Page 269

      The Ethiopia Afar Geoscientific Lithospheric Experiment (EAGLE) was undertaken to provide a snapshot of lithospheric break-up above a mantle upwelling at the transition between continental and oceanic rifting. The focus of the project was the northern Main Ethiopian Rift (NMER) cutting across the uplifted Ethiopian plateau comprising the Eocene–Oligocene Afar flood basalt province. A major component of EAGLE was a controlled-source seismic survey involving one rift-axial and one cross-rift c. 400 km profile, and a c. 100 km diameter 2D array to provide a 3D subsurface image beneath the profiles’ intersection. The resulting seismic data are interpreted in terms of a crustal and sub-Moho P-wave seismic velocity model. We identify four main results: (1) the velocity within the mid- and upper crust varies from 6.1 km s−1 beneath the rift flanks to 6.6 km s−1 beneath overlying Quaternary axial magmatic segments, interpreted in terms of the presence of cooled gabbroic bodies arranged en echelon along the axis of the rift; (2) the existence of a high-velocity body (Vp 7.4 km s-1) in the lower crust beneath the northwestern rift flank, interpreted in terms of about 15 km-thiek, mafic under-plated/intruded layer at the base of the crust (we suggest this was emplaced during the eruption of Oligocene flood basalts and modified by more recent mafic melt during rifting); (3) the variation in crustal thickness along the NMER axis from c. 40 km in the SW to c. 26 km in the NE beneath Afar. This variation is interpreted in terms of the transition from near-continental rifting in the south to a crust in the north that could be almost entirely composed of mantle-derived mafic melt; and (4) the presence of a possibly continuous mantle reflector at a depth of about 15–25 km below the base of the crust beneath both linear profiles. We suggest this results from a compositional or structural boundary, its depth apparently correlated with the amount of extension.

    4. Page 293

      18 audio-frequency magnetotelluric (MT) sites were occupied along a profile across the northern Main Ethiopian Rift. The profile covered the central portion of the Ethiopia Afar Geoscientific Lithospheric Experiment (EAGLE) line 1 along which also a number of broadband seismic receivers were deployed, a controlled-source seismic survey was shot, and gravity data were collected. Here, a two-dimensional model of the MT data is presented and interpreted, and compared with the results of other methods. Shallow structure correlates well with geologically mapped Quaternary to Jurassic age rocks. Within it, a small, shallow conducting lens, at less than 1 km depth, beneath the Boset volcano may represent a magma body. The 100Ωm resistivity contour delineates the seismically inferred upper crust beneath the northern plateau. The Boset magmatic segment is characterized by conductive material extending to at least lower crustal depths. It has high velocity and density in the upper to mid-crust and upper mantle. Thus, all three results suggest a mafic intrusion at depth, with the MT model indicating that it contains partial melt. There is a second, slightly deeper, more conductive body in the lower crust beneath the northern plateau, tentatively interpreted as another zone containing partial melt. The crust is much more resistive beneath the southern plateau, and has no resistivity contrast between the upper and lower crust. The inferred geoelectric strike direction on the plateaus is approximately parallel to the trend of the rift border faults, but rotates northwards slightly within the rift, matching the orientation of the en echelon magmatic segments within it. This follows the change in orientation of the shear wave splitting fast direction.

    5. Page 307

      We present analysis of new gravity data to produce a 2D crustal and upper mantle density model across the northern Main Ethiopian Rift (NMER). The magmatic NMER is believed to represent the transitional stage between continental and oceanic rifting. We conclude that beneath our profile, magma emplacement into the upper crust occurs in the form of a 20 km-wide body beneath the axis of the rift, and a 12 km-wide off-axis body beneath the NW margin of the rift. These are coincident with Quaternary volcanic chains, anomalies in seismic velocity and conductivity identified by the Ethiopia Afar Geoscientific Lithospheric Experiment (EAGLE) along the same profile. We also identify a shallow, high-density body beneath the axial Boset volcano interpreted as either a dyke zone or a magma reservoir that may have fed Quaternary felsic volcanism. Our results provide supporting evidence for a c. 15 km-thick mafic underplate layer beneath the northwestern rift flank, imaged by the EAGLE controlled-and passive-source seismic data. A relatively low-density upper mantle is required beneath the underplate and the rift to produce the long wavelength features of the gravity anomaly. The resulting model suggests that the lithosphere to the SE of the rift is unaffected by rifting processes. Our results combined with those from other EAGLE studies show that magmatic processes dominate rifting in the NMER.

Purchase Chapters

Recommended Reading