T41G-01 08:00h
Upper Mantle Seismic Structure of the Northern Ethiopian Rift - a Region of Incipient Continental Breakup
The northern Ethiopian rift forms the third arm of the Red Sea, Gulf of Aden triple junction and captures the transition from continental rifting in East Africa to oceanic sea-floor spreading in Afar. As rifting proceeds to sea-floor spreading, it is expected that fault controlled (mechanical) segmentation associated with the continental rift setting will be superceded by asthenospheric and magmatic processes, which are more prevalent in oceanic settings. The EAGLE passive networks consisted of 79 broadband seismometers over an area 250x250km centred on the Boset magmatic segment 70km SE of Addis Ababa, in the centre of the rift. We image upper-mantle structure between depths of 40-300km depth using teleseismic tomographic inversion of P and S wave travel-time data. Crustal structure, unresolvable in our inversions, is accounted for using an a priori crustal model which we derive from receiver function analyses of Moho P-to-S converted phases and from the results of recent EAGLE refraction experiments in Ethiopia. Analysis of relative arrival-time residuals shows that the rift flanks are asymmetric with arrivals associated with the south-eastern Somalian Plate faster ($\sim$0.65s for the P waves; $\sim$2s for the S waves) than the north-western Nubian Plate. Our tomographic inversions image a 75km-wide tabular low velocity zone (Vp=-1.5%, Vs=-4%) between 40-300km depth, interpreted as upwelling in the upper mantle beneath the less evolved southern part of the rift. At depths of $\sim$100km, north of 9 degrees N this anomaly broadens towards the oceanic spreading centre of Afar where it appears to be connected to deeper low-velocity structure. Along axis, the low velocity upwelling beneath the rift is segmented, with low velocity material in the uppermost 100\,km always offset to whichever side of the rift has the highest rift flank topography. Lateral velocity contrasts and comparison of $P$ and $S$ travel-times suggests high temperatures and partial melt in the upper-mantle beneath the rift. We propose, that magma emplacement has occured within the lithosphere, thereby allowing rifting in an area where large scale tectonic forces may be too small for amagmatic rifting to occur.
T41G-02 08:15h
Dynamics of Slow Seafloor Spreading Constrained by Seismic Anisotropy in Atlantic Lithosphere
Seismic anisotropy within the oceanic lithosphere provides one of the most direct means to study mantle deformation associated with mid-ocean ridge and hotspot volcanism. Advection beneath a mid-ocean ridge spreading center deforms the mantle rocks, and as the rocks cool to produce the oceanic lithosphere, they retain a record of this deformation in the form of lattice-preferred orientation of olivine. In the fast-spreading Pacific, observations of seismic anisotropy suggest that spreading-center deformation is quite simple, essentially 2-D corner-flow oriented in the spreading direction. While lithospheric anisotropy is less well characterized in slow-spreading regions such as the Atlantic, the segmented nature of slow-spreading ridges and the abundance of near-ridge hotspots suggest that shallow mantle deformation in these regions may be more complex than that found beneath fast-spreading ridges. This notion is supported by two analyses of lithospheric anisotropy in the Atlantic. First, radial anisotropy imaged near the Reykjanes Ridge implies a quasi-vertical (rather than horizontal) orientation of the lithospheric fabric, which suggests a buoyant (rather than passive) mode of spreading and melt extraction in this hotspot-influenced region. Second, azimuthal anisotropy within a swatch of western Atlantic lithosphere that was formed via ultra-slow spreading has a magnitude of 3%, nearly a factor of two weaker than that found in the Pacific. This observation suggests that shallow mantle deformation at slow-spreading ridges is accommodated in part by localized (brittle) mechanisms. Here we extend these results using regional surface-wave analyses of the Atlantic basin. Earthquakes from Atlantic source regions recorded at broad-band seismic instruments located on Atlantic islands and the surrounding margins provide excellent sensitivity to oceanic lithosphere structure, without contamination by continental heterogeneity. By characterizing anisotropy in both hotspot-influenced (e.g. Iceland, Azores) and normal slow-spreading lithosphere, and comparing these structures to the Pacific, we evaluate the degree to which spreading rate and/or mantle source temperature control fabric and mantle flow beneath mid-ocean ridges.
T41G-03 08:30h
Stratification of Anisotropy in the Pacific Upper Mantle
Based on the use of broad-band (25 sec - 150 sec) Rayleigh wave group speeds to estimate the 2-psi component of azimuthal anisotropy, we present evidence for a stratification of anisotropy in the uppermost mantle at large-scales across the Pacific basin. We confirm previous surface wave studies that established that the fast-axis directions of azimuthal anisotropy for intermediate and long period Rayleigh waves approximately align with present-day plate motions. At shorter periods (25 sec - 50 sec), however, fast-axes align nearer to the paleo- or fossil spreading direction which differs from present-day plate motion in the old Pacific. These observations, as well as observations of the age dependence of the amplitude of azimuthal anisotropy, imply that azimuthal anisotropy in the Pacific upper lithosphere is fixed or ``fossilized'', on average, reflecting the strain conditions extant during the early evolution of the lithosphere rather than the current ambient flow direction. In the deeper lithosphere and asthenosphere, anisotropic fast axis directions align nearer to present-day plate motions, apparently having re-oriented to conform to the current conditions of mantle flow. The mechanism of anisotropy stratification remains unclear, but observations are consistent with the anisotropy of the shallow lithosphere being fixed because the shear flows that can produce dislocation creep and a change in anisotropy will occur at increasing depths as the plate ages.
T41G-04 INVITED 08:45h
Oceanic Intraplate Volcanism: a consequence of mantle dynamics at scales small compared to deep mantle plumes and plates.
Intraplate volcanism occurs at spatial and temporal scales not explained by relatively stationary, long lived hotspots that generate age progressive volcanic chains like Hawaii, which may be due to upwelling in plumes from the deep mantle. This small scale intraplate volcanism is most abundant on the Pacific plate and occurs over wide areas in the form of a) short volcanic chains aligned with plate motion along which volcanism persists for only 1-10 Myr and b) volcanic ridges, some of which are known to propagate at rates fast compared to plate motions. Recent intraplate volcanic activity is particularly high within the South Pacific Superswell, an area of relatively shallow seafloor for its age beneath which global seismic tomography indicates a large scale region of low seismic velocity. If low seismic velocity corresponds to low density, a large scale mantle upwelling is predicted. Radial seismic anisotropy is particular strong beneath the Pacific plate, and regions of faster horizontal than vertical shear wave velocity appear to correlate spatially with the abundance of seamounts. Several mechanisms to explain small scale volcanism have been suggested. One possibility may be thermal plumes that develop at the top of a hotter, but chemically more dense lower mantle upwelling (e.g. Davaille, et al., EPSL, 2002). Alternatively in regions where the mantle is initially at its melting temperature, decompression melting in upwellings that result from buoyancy associated with melting itself may occur (e.g. Raddick, et al., JGR, 2002). This spontaneous buoyant decompression melting, which might be termed "magmatic convective storms", should be limited by accumulation of chemically buoyant and creep resistant mantle depleted in Fe/Al and water, respectively, as melt is extracted. Modelling studies indicate that this mechanism can plausibly explain volcanism with the volumes and time scales observed. Water is much more soluable in the transition zone than in lower pressure upper mantle phases. Thus, an alternative mechanism currently being considered is the release of water due to the upwelling of water-rich transition zone mineral phases. Rapidly propagating volcanic ridges may be due to the triggering of buoyant melting by initial upwelling caused by small amounts of plate extension, an amount too small to explain melting due to the upwelling associated with extension alone. Alternatively, rapidly propagating volcanic ridges may be explained by fingering instabilities as horizontally moving low viscosity mantle displaces higher viscosity mantle (Weeraratne, et al., Fall AGU, 2003). Seismic observations must continue to play a major role in examining these possible mechanisms of intraplate volcanism.
T41G-05 09:05h
Shear Wave Splitting Forward Model for the 3-D Hawaiian Plume : the Sensitiveness of Interpretation
The Hawaiian volcanism is thought to be generated by a thermal mantle plume. A 3-D convection code using the hybrid spectral/finite difference technique of Christensen & Harder (1991) describes the interaction of the Hawaiian plume with the moving Pacific lithosphere (Ribe & Christensen, 1994). A steady state flow field solution allows to calculate the velocity gradient field and the path of rocks. This can be used to integrate along streamlines either the finite deformation or the Lattice Preferred Orientation (LPO) of upper mantle minerals (olivine, orthopyroxene). The code D-Rex describing the deformation of a mineral aggregate by dislocation creep and dynamic recrystallization provides the necessary tool to obtain the LPO (Kaminski, Ribe & Browaeys, 2004). The equivalent elastic tensor at each position of the model is calculated using both the Orientation Distribution Function related to the LPO and tabulated elastic tensor of minerals. The tensor is then reduced to its transverse isotropic part by decomposition into different symmetry class components (Browaeys & Chevrot, 2004). The convolution of the sensitivity kernels for shear wave splitting in a transverse isotropic medium (Favier & Chevrot, 2004) with the 3-D distribution model of transverse isotropy gives variation of the splitting intensity $I(\alpha)$ with the backazimuth $\alpha$ of the teleseismic wave. The splitting intensity can be related to the parameters of classical ray theory (time delay $\delta t$ and fast axis direction $\phi_0$) by the relationship $I(\alpha)=\delta t\sin2(\alpha-\phi_0)$. A comparison between the fast axis orientations in the 3-D model and the fast direction deduced from the classical expression of $I(\alpha)$ is done. It shows that in regions where the wave propagates through inhomogeneous anisotropic properties compare to the typical wavelength of a teleseismic wave (i.e. $\lambda\sim$ 50 km), there is no simple relation between the seismic anisotropy observations and the underlying anisotropic structure in the upper mantle.
T41G-06 09:20h
The Transition Zone Beneath the South Pacific Superswell From Receiver Functions
The geometry and thickness of the transition zone (TZ) is a key question to discuss the upper or lower mantle origins of the mantle plumes and more generally the thermal state of the upper mantle. We investigate the TZ thickness in the south Pacific by detecting P to S converted waves at the 410 and 660 km discontinuities from receiver function techniques. The south Pacific and particularly the French Polynesia area are interesting for several reasons: i) The huge and low seismic velocity anomaly in the lower mantle beneath the south Pacific may indicate the presence of hot mantle rising toward the Earth's surface. ii) The presence of several volcanic alignments indicates the present or recent hotspots activity (e.g., the Society, Austral, Marquesas and Pitcairn alignments). iii) The presence of the large-scale bathymetry anomaly (the South Pacific Superswell) suggests a low-density anomaly at large depth. We used seismic data recorded during the last decades by the Geoscope, IRIS and LDG/CEA permanent stations running in the South Pacific but also recorded by a temporary network of 10 broad band seismic stations operating in French Polynesia for the Polynesian Lithosphere and Upper Mantle Experiment (PLUME) since the end of 2001. The P410S and P660S phases can be observed on high quality individual receiver functions but are more visible by stacking the individual receiver functions. Even if the results at the permanent stations are better constrained by a large number of individual receiver functions (between 50 and 200), the PLUME temporary stations generally provided 10 to 20 individual receiver functions and their stacking suggest coherent and homogeneous signatures throughout the studied area. The prominent result is that the P410S phases arrive generally late of 3 to 6 s after the expected arrival (44 s after the P wave in the IASPEI91 model), whereas the P660S phases arrive either on time (68 s after the P-arrival) or slightly later (1 to 3 s). The sum of the P660S and P410S travel times are anomalously high (between 2 and 10 s) suggesting an anomalously slow upper mantle, in the range 2-5 %, compatible with the available tomographic models. The P660S-P410S differential times are anomalous low of 1 to 5 s, suggesting a large-scale thinning of the TZ that could result from a temperature positive anomaly of 100 to 250 K at the base of the TZ.
http://www.isteem.univ-montp2.fr/TECTONOPHY/polynesia/Polynesia_home_page.html
T41G-07 09:35h
3-D Spherical Models of Mantle Convection: A Study of the Impact of Tectonic Plates, Nuvel 1 Plate Motions and Seismic Tomography
We explore here the influence of realistic characteristics of mantle convection such as tectonic plates, the surface velocity field from the Nuvel-1 No-net-rotation (NNR-1) model (Argus & Gordon, 1991), and the 3-D mantle structure described by global seismic tomographic reconstructions (Grand, 2002). We first established a `reference' 3-D spherical convection model which incorporates surface plates in a dynamically self-consistent way (Monnereau & Qu\'er\'e, 2001). This reference model, with a simple two-layer viscosity profile, reproduces the main features of plate tectonics: linear subduction zones and spreading along mid-ocean ridges and deep-seated hotspot plumes (numbering three in this model) which extend to the surface. The next level of modelling involved taking the average radial temperature field from the reference model and imposing the velocity field from NNR-1 in a time-dependent thermal convection simulation. This second convection model is distinctly different from the reference model because the balance of flow-induced torques acting on the plates (e.g., Ricard & Vigny, 1989) is no longer verified. We now observe as many as six hotspot plumes, close to the number of principal or major hotspots which have been identified for the Earth (Courtillot et al., 2004). In a final series of convection simulations, the 3-D mantle structure from seismic tomography is employed as a starting condition. We now observe another distinct manifestation of evolving hotspot plumes in the mantle. We will also report here on the implications of these models for global heat flow at the surface and at the core-mantle boundary.