T23E-01 INVITED
Seismic Evidence for a Vertical Tear in the South American Lithosphere Offshore Venezuela
In the southeast corner of the Caribbean plate, westward subduction of oceanic (Atlantic) South America transitions to east-west transform with continental South America, forcing the South American lithosphere to tear along its former passive margin. The most widely accepted model for lithospheric detachment, the tensile tear model, envisages oblique convergence causing northwest-dipping subduction. In this model, tensile stress on the subducting slab results in detachment orthogonal to the motion vectors. Conversely, the less commonly accepted shear tear model predicts detachment parallel to the motion vectors along a vertical plane, with shear stress focused at the edge of the propagating transform boundary. The tensile tear model requires significant crustal shortening to produce the 7 km of uplift in the coastal Serrania del Interior and 10 km of subsidence in the Maturin Basin onshore Venezuela. These features have been widely interpreted as the foreland fold and thrust belt and foreland basin of a collisional orogeny. Recent numerical modeling of the lithospheric response to shear tearing has shown, however, that much of the topography of eastern Venezuela can be generated from the propagating detachment, requiring relatively little shortening. We present five types of initially independent seismic data analyses to resolve the lithospheric tear: local seismicity locations, fault plane solutions, receiver functions, wide-angle seismic data inverted for velocity structure, and a regional, balanced cross-section constrained by petroleum industry data. These five datasets image a shear tear extending near-vertically through the entire lithosphere. Confirmation of this type of detachment has implications for diverse locations with similar plate boundary geometries, such as Tonga or the Carpathians, and improves our general understanding of vertical tectonics and associated mantle driving forces.
T23E-02
Shear Wave Splitting Observations on the Caribbean-South American Plate Boundary: Evidence for Vertically Coherent Deformation Induced by Trench Rollback
We utilized data from the BOLIVAR project and the permanent national seismic network of Venezuela to investigate the boundary between Venezuela and southeastern Caribbean. We made shear wave splitting measurements on SKS and SKKS phases using techniques developed by Silver and Chan (1991). We examined 3000+ station event pairs and found approximately 300 with visible SKS and/or SKKS phases. From these data we obtained robust measurements for 43 of the 82 stations. Two areas show unusual split times larger then 2.0 seconds: (1) two observations on the northeastern edge of the Maracaibo block and (2) a 200 km2 area between Trinidad and Margarita Island. We interpret the large split times over the Atlantic subduction as further evidence of a shear tear subduction geometry. We hypothesize that the slab edge at depth is forcing the creation of a vertical shear zone. Split times larger then 2.0 seconds are difficult to explain using typical horizontal shear deformation with normal lithosphere thicknesses. We use mineral physics models to predict maximum split times for different geometries and anisotropic layer thicknesses to support the vertical shear zone hypothesis. In addition, we hypothesize that a similar geometry could be created by the northern edge of a paleo- Caribbean subduction and may explain the northeastern Maracaibo split time anomaly.
T23E-03 INVITED
Terminal Stage Subduction and Slab Detachment: From Structure to Dynamics
In this contribution we focus on terminal stage subduction, often triggered by continent-continent or arc-continent collision. Rupturing of subducted lithosphere, including slab detachment, is a key process in this stage. Seismic tomography information on the 3D structure of subduction zones has been key towards developing a better understanding of subduction processes. We show how structural information on lithosphere and mantle structure, obtained from seismic tomography, can be used to formulate hypotheses concerning the evolution and dynamics of subduction zones which (1) can be subjected to numerical modeling to investigate the physical soundness of the hypotheses, and (2) can subsequently be tested against geological, geophysical, and geodetical data to assess their general or regional validity. In combination, this approach increases our understanding of the relation between deep processes and the geological processes at the surface. The landlocked basin setting of the Mediterranean region provides unique opportunities to study terminal stage subduction and its consequences. Subducted lithosphere is subject to various types of rupturing: Slab detachment, vertical slab tearing and the formation of STEP (tear) faults have been identified and analysed. The different Mediterranean arc systems being in slightly different stages of evolution provides additional clues towards unraveling the dynamics of the process. Using examples from the western-central and eastern Mediterranean, we show how temporal and spatial variations in slab structure affect the dynamics and stress field of convergent plate boundaries, with arc migration and back-arc extension as a prominent expression. In addition, terminal stage subduction appears to be have played a crucial role in Italian magmatism. Terminal stage subduction involving slab detachment is a strongly transient stage with great impact on geological processes, on a wide variety of scales. The information obtained from Mediterranean subduction zone studies sheds light on the evolution of other and/or older convergent plate margins, as well.
T23E-04 INVITED
Numerical Models of Subduction and Slab Detachment: Placing a Lower Bound on the Strength of Slabs
Subduction provides the main driving force for the motion of tectonic plates at the Earth's surface through slab- pull and sinking-induced flow in the surrounding mantle. The ability of the slab to directly transmit slab-pull forces to the tectonic plate at the surface depends on the minimum strength and rheology (e.g., viscous, plastic) of the slab. Previous models have shown that observations including the state of stress in slabs, dynamic topography and the geoid above slabs, the evolution of slab and the kinematic history of subduction can be well-matched by a variety of models with either low viscosity (i.e., 100-1,000 times more viscous than the surrounding mantle) or high viscosity slabs (i.e., more than 10,000 times more viscous than the surrounding mantle). However, in many of the models in which a good match to observations is found for low viscosity slabs, the maximum slab viscosity is imposed as a cut-off value, which forces the entire slab to have a more or less uniform viscosity independent of strain-rate or stress magnitude, rather than a plastic yielding-type rheology. We present numerical models demonstrating that when the non-Newtonian viscosity of the upper mantle and plastic yielding behavior of slabs are taken into account, the minimum yield strength that allows for continuous subduction is approximately 300- 500 MPa, which leads to high viscosity slabs with some localized lower viscosity regions. A yield stress of 10-100 MPa is required to form uniformly low viscosity slabs, but these slabs detach from the subducting plate, due to localized weakening, when the slab reaches a length of 200-300 km, even when subduction is facilitated by a low viscosity shear zone and kinematically-imposed surface velocities. In contrast, detachment of higher strength slabs in fully-dynamic models only occurs when the shear zone is removed and prevents further subduction.
T23E-05
Seismic Anisotropy Across a Boundary Between Compression and Extension Above a Subducting Plate, Western North Island, New Zealand.
The North Island of New Zealand lies on the Australian Plate, above the subducted Pacific Plate. The highest point on the North Island is near its centre on Mt. Ruapehu Volcano, which is the southernmost in a line of active volcanoes associated with extension in the Taupo Volcanic Zone, the southern limit of the Lau-Havre Trough. South of Mt. Ruapehu, compression occurs and volcanism stops, but subduction continues. An east-west trending boundary between Mt. Ruapehu and Mt. Taranaki volcanoes is delineated by a strong gravity gradient, change in crustal thickness, and abrupt changes in seismic attenuation, and has been termed the "Taranaki- Ruapehu Line" (Salmon, 2007). Shear wave splitting determined from SKS phases on broadband stations exhibit strong, trench-parallel (extension-perpendicular) anisotropy both south of the line and north east of the line near the center of the extending region. However, SKS waves recorded north of the line and west of the extending region do not split, suggesting isotropy. (Greve and Savage, 2007). Anisotropy measurements from local S phases recorded on sparsely-spaced stations also suggest changes from well-aligned shear wave polarisations south of the line to more scattered measurements north of the line (Audoine et al., 2004). Several models have been proposed to explain the north-south changes. These include: 1) in the north, extension carries fluids away from the plate, allowing the fluids to spread westward while to the south, compression and plate- boundary parallel flow confine the fluids to the slab itself. (Audoine et al. 2004) 2) Thickened crust to the south shut off the vertical flow of fluids, driving northeastward flow of the mantle, increasing the fluids available in the north (Reyners et al. 2006). 3) In the west and central NI Miocene shortening of ~ 100 km led to a series of a Rayleigh-Taylor instabilities that have since detached to be replaced by an asthenospheric upwelling (Stern et al. 2006). We use seismic data gathered by seven temporary three-component stations set up across the Taranaki- Ruapehu line (TRL) to determine anisotropy in local S phases in an attempt to delineate more accurately the location of the change in anisotropy, and to explain the mechanism for the change. Preliminary results of shear- wave splitting delay times (dt) and fast polarisations are made from 409 phases. No overall correlation between dt and depth is found, which is attributed to a complex and rapidly changing anisotropic structure. Yet the anisotropy is consistent for phases arriving within close incidence angles and back azimuths; we attribute this to lateral variations in anisotropy. North of the line, within the extending region, extension-perpendicular fast orientations dominate with delay times from 0.5 to 0.8s, consistent with large splitting observed on SKS phases. Closer to the TRL, E-W trending fast orientations are more frequent and small delay times (0.1 to 0.3s) are common. We propose that the smaller delay times are caused by re-splitting in the crust due to anisotropic crustal structures parallel to the TRL. We plan to incorporate more measurements by using a newly developed automatic technique to test if the preliminary results hold.
T23E-06 INVITED
Modeling Three-dimensional Variations in Mantle Flow due to Rollback Subduction and Back- arc Extension: Implications for Arc Magmagenesis
Circulation and melting of the mantle are directly related to large-scale plate motions. We use laboratory experiments to characterize how subduction of a discrete plate segment and extension in an overriding plate influence 3D return flow trajectories to the sub-arc mantle wedge and spatial-temporal heterogeneity in slab and wedge temperature fields. Our goal is to identify which combinations of subduction parameters favor either decompression melting, slab melting or melting of the hydrated slab boundary layer due to vertical advection. Models utilize a glucose working fluid with a temperature dependent viscosity to represent the upper 2000km of the mantle. The subducting plate is modeled with a Phenolic sheet and extension in the overriding plate is produced using Mylar sheets. We recreate basic subduction styles observed in previous dynamic subduction models using simplified, kinematic forcing. Slab plate segments, driven by hydraulic pistons, move with various combinations of downdip, rollback and steepening motion. Two modes of producing back-arc extension (1. trench retreat vs. 2. overriding plate retreat) are investigated. Models reveal strong vertical shear within the wedge produced by lateral return flow around the retreating slabs and coupling between the wedge fluid and the overriding plate. Results show that vertical velocities in the wedge (e.g., favorable to decompression melting) and both slab and wedge temperatures are strongly related to the mode of back-arc extension, the ratio of rollback to downdip motion and the degree of plate steepening. Mode 1(2) back-arc spreading produces laterally heterogeneous (homogeneous) temperature distributions in the slab and beneath the spreading center. Experiments predict rapid onset, short-lived periods of slab-influenced melting after the initiation of slab rollback and back-arc extension due to direct vertical advection of material from the hydrous boundary layer above the downgoing slab.
T23E-07 INVITED
Subduction dynamics and its expression in plate motions
The dynamics of plate tectonics are strongly related to those of subduction. We gain new insights in the thus far elusive dominant forces in subduction, by comparing relations between subduction motions and dips as predicted by a fully dynamic model for free subduction (i.e., driven solely by downgoing plate buoyancy while resisted passively by the mantle and upper plate), with data for the major subduction zones from the recent compilation by Sdrolias and Muller (2006). We find that at the present day, subduction and sinking velocities are as for slabs driven by their own upper-mantle buoyancy, with Stokes velocities that require effective average upper mantle viscosity variations of less than 20% and effective slab widths of 2000-3000 km. Dips and subduction velocities imply that effective plate resistance to bending is low, as for viscosities two orders of magnitude above that of the upper mantle. Steep dips and high plate advance velocities require a low-drag asthenosphere (as when its viscosity is 2 to 3 orders of magnitude lower than the upper mantle average). Back through the Cenozoic, old plates in the Pacific subducted at rates expected for upper-mantle slab pull, while young plates often subducted at velocities up to two or three times those expected from their upper mantle buoyancy, accompanied by very low trench retreat rates. Such high rates require an additional driving force, which is most likely supplied by lower mantle slab penetration. Young lithosphere may be more prone to penetrate the lower mantle, as its lower density and strength limit trench retreat and facilitate slab thickening in the transition zone which increases Stokes' sinking velocity.
T23E-08 INVITED
The influence of lower mantle viscosity on plate velocity time-dependence in numerical mantle convection models
Plate boundary locations, and therefore plate dimensions,can change significantly over time periods that are relatively short compared to the mantle overturn time-scale. However, the influence of plate geometry evolution on mantle convection has not been widely studied in 3D convection models. We examine the effect of mobile plate boundaries on the time-dependence of the mean surface velocity, individual plate velocities and heat flow in models featuring different mantle viscosity profiles.. The three-dimensional Cartesian geometry models feature evolving plate geometries, high Rayleigh numbers, periodic boundary conditions, and multiple plates with dynamically determined motion. Plate motion is determined by specifying that each plate move rigidly with a velocity that results in a net shear stress of zero at the base of the thick, viscously defined, lithosphere. This condition ensures that the specified plate motion neither drives nor resists the buoyancy driven flow. The time-dependent plate velocities determine the evolution of the plate geometry. Plate boundaries evolve as triple junctions and are moved with a velocity that is equal to the average of the velocity of the three surrounding plates. In a 3×3×1 solution domain geometry, we compare the time-dependence of the convection obtained in cases where plate geometry is able to evolve, with cases where the geometry remains fixed. We investigate time-dependence in three distinct viscosity stratification models. We compare the time-dependence of the plate velocities and mantle heat flux in these calculations and examine the consistency of the results in a pair of larger geometry calculations (obtained with 6×6×1 Cartesian geometry solution domains featuring a minimum of 9 plates).We find that an evolving plate geometry results in more rapid and more dramatic variations in all global measures of the vigor of the convection. However, the presence of a high viscosity lower mantle dampens the time-dependence of the convection, even over periods in which the plate geometry evolves considerably.