Kinematics of the East African Rift from GPS and earthquake slip vector data E. Calais, C.J. Ebinger, C. Hartnady, J.-M. Nocquet To cite this version: E. Calais, C.J. Ebinger, C. Hartnady, J.-M. Nocquet. Kinematics of the East African Rift from GPS and earthquake slip vector data. The Geological Society, London, Special Publications, Geological Society of London, 2006, 259, pp.9-22. 10.1144/GSL.SP.2006.259.01.03. hal-00407586 HAL Id: hal-00407586 https://hal.archives-ouvertes.fr/hal-00407586 Submitted on 26 Jul 2019 HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. Kinematics of the East African Rift from G PS and earthquake slip vector data 1 2 3 4 E. CALAIS , C. EBINGER , C. HARTNADY , & J.M. NOCQUET 1 Purdue University, Department of Earth and Atmospheric Sciences, West Lafayette, Indiana, USA (e-mail: [email protected]) 2 Department of Geology, Royal Holloway, University of London, Egham, UK 3 Umvoto Africa (Pty) Ltd, PO Box 61, Muizenberg, South Africa 7950 4 CNRS, UMR6526 Geosciences Azur, Valbonne, France Abstract: 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 (l) a present-day Nubia-Somalia Euler pole located between the southern tip of Africa and the South­ west 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, poss­ ibly 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 loca­ lized deformation rather than a wide zone of quasi-continuous, pervasive deformation. On the basis of this new kinematic model and mantle flowdirections interpreted from seismic anisotropy measurements, we propose that regional asthenospheric upwelling and locally focused mantle flow may influence continental deformation in East Africa. The East African Rift (EAR), a 5000 km-long plates. For instance, Sella et al. (2002) used two series of fault-bounded depressions straddling GPS sites on the Somalian plate and four on the east Africa in a roughly north-south direction, Nubian plate, while Fernandes et al. (2004) used marks the divergent boundary between two major three GPS sites on the Somalian plate and 11 on tectonic plates, Somalia and Nubia (Fig. 1). the Nubian plate and longer data time series. Although the EAR is often cited as a modern Again, these two GPS estimates of the Nubia­ archetype for rifting and continental break-up and Somalia angular velocity differ significantly from a Cenozoic continental flood basalt province, its each other, as well as from previous results cunent kinematics is still poorly understood and derived from oceanic data (Fig. 1). quantified, owing in part to its tremendous extent In addition to far-field plate motions, the and inaccessibility. kinematics of the EAR itself remains an open Jestin et al. (1994) were among the first to quan­ question. Some authors have proposed that tify the kinematics of the EAR by estimating a 3 Ma the EAR consists of a mosaic of rigid litho­ average Nubia-Somalia angular velocity. Using spheric blocks bounded by localized deformation Arabia-Nubia and Arabia-Somalia relative within nanow seismically and volcanically active motions determined from marine geophysical data rift valleys (e.g. Ebinger 1989; Hartnady 2002). in the Red Sea and Gulf of Aden, they find a Others favour a broad deformation zone (e.g. Nubia-Somalia Euler pole south of the Southwest Gordon 1998) implicit in models that assume a Indian Ridge (Fig. 1). Their result differs signifi­ weak mantle lithosphere beneath continents (e.g. cantly from that of Chu & Gordon (1999), who esti­ Jackson 2002). But the distribution of strain mate a 3 Ma average Nubia-Somalia angular across and along the EAR is cunently unknown velocity from marine geophysical data along the and no quantitative kinematic data are presently Southwest Indian ridge and the Antarctica­ available for that plate boundary. In this paper, Somalia and Antarctica-Nubia plate closure we use an updated GPS and earthquake slip circuit (Fig. 1). More recently, direct estimates of vector data set to estimate the Somalia-Nubia the Nubia-Somalia plate motion have been made angular velocity and propose a first-order kin­ possible thanks to a limited number of permanent ematic model for present-day deformation in the Global Positioning System (GPS) stations on both EAR. 1 -20 -10 0 10 20 30 40 50 60 70 RAB ... � . ! � 30 ____ :'" ... MASl 20 Indian 10 .Plate 0 -10 -20 -30 -40 -50 Antarctic Plate * .. Smm/a J94 · ,. .: ,. •• measured � .. �..... j• : •"'I , •• -60 :·.;��1• modeled (3-plate pole) � Fernandes et al., 2004 � CGPS sites not used o Fig. 1. GPS sites used in this study. Dots show seismicity (NEIC catalog). Stars are Euler poles for Somalia-Nubia relative motion with associated 1-sigma error ellipse. S02 = Sella et al. 2002; PB04 = Prawirodirdjo & Bock 2004; F04 = Femandes et al. 2004; CG99 = Chu & Gordon 1999; J94 = Jestin et al. 1994. This work 2-plate = two­ plate inversion using GPS velocities at MALI, REUN and SEYl +nine earthquake slip vectors along the Main Ethiopian Rift. This work 3-plate = three-plate inversion using GPS velocities at MBAR, MALI, REUN, SEY l and 47 earthquake slip vectors along the East African Rift (see Figure 2 and Table 1). Curved dashed line shows the Nubia­ Somalia plate boundary in the SW Indian Ocean proposed by Lemaux et al. 2002. 2 GPS data the vector of unknowns. The least-squares solution is then given by: We processed GPS data from 10 permanent GPS sites operating on the Nubian plate (MAS1, GOUG, NKLG, ZAMB, RBAY, SUTH, SUTM, HRAO, HARB, SIMO), the three sites available on the Somalian plate (MALI, SEY1, REUN), and We estimated the Nubia-ITRF2000 angular vel­ one site in the EAR (MBAR). The results presented ocity by inverting horizontal velocities at sites here include all the publicly available data from MAS1, NKLG, SUTH, SUTM and GOUG (Fig. 1). August 1998 to April 2005. We processed the We formed the data covariance matrix using the GPS data using the GAMIT software version 10.2 2-sigma velocity uncertainty (95% confidence) on (King & Bock 2005). We solved for station coordi­ GPS velocities and their site-by-site NS-EW corre­ nates, satellite state vectors, one tropospheric delay lation (i.e. we do not account for intersite corre­ every four hours at each site, and phase ambiguities lations). We find a reduced y (Y divided by the using double-differenced GPS phase measurements, degrees of freedom) close to unity with a weighted with International GPS Service (IGS) final orbits RMS of 0.7 mm a -I for horizontal velocities and International Earth Rotation Service (IERS) (Table 2). We then use an F-ratio statistic to test Earth orientation parameters relaxed. We then com­ whether the velocity at additional sites in Africa is bined our regional daily solutions with global Sol­ consistent with the rigid rotation defined by the pre­ ution Independent Exchange (SINEX) files from vious subset of sites. The F-ratio, defined by: the IGS daily processing routinely done at Scripps Institution of Oceanography and imposed the refer­ ence frame by minimizing the position deviations of _ [K(PI)- K(Pz)]/(P - Pz) F - 2 I ' 38 globally distributed IGS core stations with X (pz)/pz respect to the International Terrestrial Reference Frame 2000 (ITRF2000; Altamimi et al. 2002), tests the significance of the decrease in y from a while estimating an orientation and translation model with p2 versus p1 degrees of freedom, with transformation. Our primary result consists of p1 > p2• This experimental F-ratio is compared to precise positions and velocities at 14 continuous the expected value of a F(p 1 - p2, p1) distribution GPS sites on the Nubian and Somalian plates, for a given risk level a% (equivalent to a expressed in ITRF2000 (Fig. 2; Table 1). In a (lOO - a)% confidencelevel) that the null hypothesis second step, explained hereafter, we rotate these (the additional site is consistent with the rigid plate velocities in a Nubia-fixed frame. model) can be rejected. The degrees of freedom of GPS velocities, in a Cartesian geocentric frame, a rigid rotation estimation is p1 = 2 x N - 3 for N can be modelled as: site velocities, it becomes p2 = 2 x N- 3- 2 with an additional site. For site ZAMB, the F-test value is F = [(8.52- 7.53)/(9- 7)]/[7.53/7] = 0.4 (using values provided in Table 1), corresponding to where P(x,y,z) is the unit vector defining the pos­ a = 0.32. The velocity at ZAMB is consistent ition of the GPS site, V(vx, vy, vz) is the velocity with a Nubian plate model (defined by sites vector at that site, and O(wx,Wy,wz) is the rotation MAS1, NKLG, SUTH, SUTM and GOUG) at a vector defining the motion of the plate carrying 68% confidence level.