U34A-01 INVITED
Models of Plate Motion, Mantle Flow, Mantle Plumes and Reference Frames
Models of mantle flow have been developed over the last several years that allow a unified treatment of plate motions and mantle flow. The models use the history of plate motions and density anomalies inferred fron seismic tomography as inputs and require specification of the viscosity structure. The models are adequate for investigating the last 100 My or so of Earth history; extension to earlier times is limited by the inherent indeterminancy of the energy equation, as well as by uncertainties in plate configurations and motions in the past. Mantle plumes imbedded in such models indicate the motion of hotspots and allow the testing of hotspot based reference frames. The models allow dynamical ( e.g. no net rotation) and kinematic ( e.g. hot spot) reference frames to be related. Such models have quantified the expected motions of hot spots, placed constraints of mantle rheology and density anomalies, and reconciled some apparent hot spot motions. They also allow testing the relationship between hotspot/plume locations and seismic anomalies in the deep mantle. Mantle flow is the result of 1) the return flow from subduction to ridges, 2) flow driven by viscous drag from plate motions, and 3) flow driven by internal density anomalies. The relative nagnitudes of these vary spatially, and each depends on viscosity structure. (1) may be largely accomodated in a sufficiently low viscosity upper mantle; (2) will be attenuated with depth by an increase with viscosity with depth; (3) depends on the rheological structure and magnitude and locations of density anomalies. These lead to uncertainties in the models; comparisons with hotspot motion, plume locations, slab trajectories and seismic tomography further refine our knowledge of mantle properties and structure.
U34A-02
Longitude through time
Earth scientists today have had no objective method of calculating what the palaeolongitudes of tectonic plates and other geological units were in the long eons prior to the oldest known hotspot trails, which are only of Cretaceous age (ca. 130 Ma). Before this time, palaeomagnetism is the only method by which to position plates quantitatively on the globe. Palaeomagnetic studies only directly yield latitudes and plate rotations, but the longitude uncertainty can be minimized by selecting an appropriate reference plate: if one can determine which plate has moved least, then it should be used as the reference plate. Africa has been nearly surrounded by mid- ocean ridges since the break-up of Pangea, and thus the ridge push forces should have roughly cancelled each other out. Moving hotspot-based plate motion models show minimal longitudinal motion for Africa (<10 degrees) for the past 130 million years, confirming the lack of significant longitudinal motion inferred from consideration of the plate driving forces. It is uncertain whether the 'zero-longitude' assumption about Africa holds before Pangea's break-up, but in the absence of better reference points, we have regarded zero longitudinal average motion for Africa as the best assumption. With this approach we have been able to demonstrate that virtually all Large Igneous Province (LIPs) for the last 300 million years project radially onto the edges of the African and Pacific Large Low Shear Velocity Province (LLSVPs) near the core-mantle-boundary (CMB). The LIPs must for this reason be derived from mantle plumes, and CMB heterogeneities must have remained quite stationary since the formation of Pangea. LIPs have erupted since Archean times and there is no reason to preclude that they were all derived from LLSVPs in the deep mantle. That inspired us to consider whether older LIP events would yield similar results. We attempt to reconstruct Gondwana in longitude in Cambrian times based on the substantial Antrim plateau volcanics (Australia), a LIP of ca. 510 Ma age along the Gondwanan margin. If the LIP was formed at the margin of the Africa or Pacific LLSVPs and they have remained the same throughout Earth's history there are six possible marginal sites on the CMB from which to choose, but three sites that do not position the long- lived subduction margin of Gondwana (e.g. South America, East Antarctica and East Australia) above regions of high seismic velocity (the subduction graveyards) can be eliminated. If, as recently postulated, there have only been one LLSVP (or upwelling zone) in Pre-Pangean time (Pacific LLSVP) that reduces longitude choices to two possible marginal sites on the CMB.
U34A-03
No-Net-Rotation and Indo-Atlantic Hotspot Reference Frames: Towards a New View of Tectonic Plate Motions and Earth Dynamics
A new view of plate tectonics coupled to mantle dynamics is emerging from recent paleomagnetic reconstructions of tectonic plate histories obtained in the hotspot and no-net-rotation reference frames. A number of fundamental differences relative to past plate reconstructions have been discerned. Firstly, in previous models the difference between present-day plate motions in the global hotspot and no-net-rotation reference frames consisted of a westward drift of the lithosphere due to the dominant motion of the Pacific plate in the hotspot frame. In contrast, the new plate motion reconstructions based on the Indo-Atlantic hotspot reference frame now show that the present-day global rotation of the lithosphere is mainly in the South-North direction. Second, we find a more than 100% speed-up of the Nazca plate motion at 35 Ma which we have interpreted in terms of a slab avalanche event below the Nazca-South America plate boundary. This may be the first direct geological evidence for a mantle avalanche event occurring at a time which precedes a significant plate reorganisation. Third, the speed-up of the Nazca plate does not appear to be associated with a jump of the East-Pacific rise, therefore this feature may not be completely passive as previously thought. Fourth, the Hawaiian-Emperor bend which was a key element in previous plate reconstruction based on the assumption of a fixed Hawaiian hotspot, can no longer be explained by a change of direction of the Pacific plate and this corroborates recent studies showing a southward motion of the Hawaiian hotspot. Finally, the new Indo-Atlantic hotspot reconstruction of present-day plate motions is significantly different from the one previously established by Gripp and Gordon (1990) and the model appears to be in greater accord with plate motions predicted by seismic tomography-based mantle convection models.
U34A-04 INVITED
Global mantle circulation models with thermodynamically self consistent mineralogy: bridging the geodynamic/seismic gap
Seismic tomography is a powerful tool to aide plate reconstructions. Still its use has been hampered mainly because we don't understand very well how to interpret seismic images in terms of temperature and composition. Notable examples are observations of anti-correlation of bulk sound and shear velocity near the bottom of the mantle and a general lack of strong compressional heterogeneity in the lower mantle, both of which have given rise to a variety of speculations on lower mantle dynamics and its relation to past plate motion. Here we address this fundamental problem directly by employing a newly published, thermodynamically self consistent mantle mineralogy model, derived from considerations of Gibbs free energy minimisation of the mantle phase assemblage, which provides us with estimates of density and elastic constants for a wide range of lower mantle P,T conditions. We combine the model with simulations of global mantle circulation, where the very high numerical grid point resolution of less than 20 km throughout the mantle, amounting to more than 100 million grid points totally, is sufficient to achieve a vigorous regime of high Rayleigh number thermal convection that lies within the parameter range for which the mineralogic model was validated. We find a number of important results. For example, the hot spot flux is likely to exceed 10 TW, giving hot spots a more prominent role than is commonly believed. We also find that the anti-correlation of bulk sound and shear, and the low level of compressional wave speed arise naturally in our mineralogy/geodynamics simulations under the assumption of an isochemical mantle, greatly facilitating the interpretation of seismic heterogeneity in terms of past subduction. We will discuss these findings and explore their consequences for the development of next generation geodynamic earth models.
U34A-05
Motion of Pacific mantle plumes
The Hawaiian--Emperor hotspot chain, and its distinctive bend at 47 Ma, have figured prominently in the development of ideas on the nature of mantle plumes, plate motion, and frames of reference. However, paleomagnetic data from Ocean Drilling Program (ODP) Leg 197, together with results from plate circuit and geodynamic modeling studies, indicate southward motion of the Hawaiian hotspot during formation of the Emperor Seamounts. These analyses confirm the idea that mantle plumes should be influenced by mantle flow, and that such motion must be considered when constructing frames of reference for plate motion. An important corollary of this finding on hotspot motion is that long-term polar wander of Earth, which has been inaccurately assessed by viewing paleomagnetic data in a fixed hotspot reference frame, has been far less than previously thought. Here we extend the ODP Leg 197 analysis in three ways. We examine i. paleomagnetic data and their uncertainties relative to volcanic propagation rates; ii. consistency tests of plate circuit models and global paleomagnetic data and iii. intra-basin motion of plumes through new analyses of Late Cretaceous lavas from New Zealand. The first analysis suggests that while Late Cretaceous--Paleogene hotspot motion was the dominant factor in forming the Emperor track (and thus the famous bend morphology), smaller- scale plate motion changes might still be preserved in the track. The second set of analyses help point to deficiencies in the global paleomagnetic data set, whereas the third highlights that motion between groups of hotspots is a dominant feature during mid-Cretaceous to Paleogene times.
U34A-06
Izanagi-Pacific Ridge Subduction and its Geodynamic Consequences
As part of a global plate tectonic model for 140 Ma to the present we present a revised plate reconstruction for the western Pacific and investigate its geodynamic consequences. In our plate model, mid-ocean ridge subduction beneath southern Japan occurs at 60-55 Ma, 20 million years later than proposed for Kula-Pacific or Farallon- Izanagi ridge subduction. The difference arises because Izanagi-Pacific (I-P) spreading ceases in previous models after 110 Ma while our model incorporates continued spreading until the I-P ridge subducts beneath eastern Asia at 60-55 Ma. We regard cessation of spreading at the I-P ridge between 110 and 80 Ma as unlikely as the Izanagi plate was undergoing rapid motion, driven by net slab-pull force, from the north-northwest, immediately prior to the proposed spreading cessation. Metamorphism of the Ryoke Belt in southern Japan has previously been attributed to Kula-Pacific ridge subduction at 85 Ma, but the high-T/low-P Ryoke Belt cannot be uniquely linked to a ridge subduction event. We propose that sub-parallel subduction of the I-P mid-ocean ridge beneath Japan at 60-55 Ma resulted in nearly simultaneous slab break-off along the length of the Japanese trench (approximately 2700 km). Geological evidence for this model includes cessation of a major accretion phase in the late Cretaceous, emplacement of the Okitsu Melange due to subduction of hot, buoyant material at 55 Ma, and cross-cutting fault fabrics that indicate a counter-clockwise rotation in relative plate motions between Eurasia and the I-P plate, consistent with palaeothermal and palaeopressure data, some time between 55 and 34 Ma. Rapid subduction of the I-P ridge, over a vast distance, may have triggered a chain reaction of tectonic plate reorganizations. With complete subduction of the I-P ridge at 55 Ma, forces acting on the western edge of the Pacific Plate would have changed from ridge-push to slab pull, changing Pacific absolute plate motions from northwest to west. A combination of Australian and Pacific plate motion changes between 53 Ma and 50 Ma then initiated both the Tonga-Kermadec subduction system and the Izu-Bonin-Marianas subduction systems around 50Ma, likely due to convergence across a fracture zone caused by the Pacific plate motion change. We suggest that the observed slowdown of sub-Pacific mantle flow at 47 Ma was due to progressive impediment of lateral sub-Pacific mantle flow by the descending slabs of the Izu-Bonin-Marianas and Tonga-Kermadec subduction zones.
U34A-07 INVITED
Mantle convection and reference frames: Inverse models with plate motions, tomography and sea level
Paleo plate motions, the vertical motion of plates, and seismic tomographic images place fundamental constraint on models of mantle convection. We have developed forward and inverse models of mantle convection using these constraints self consistently. Using new paleogeographic concepts and reconstructions, implementations of adjoint and inverse models of convection, and stratigraphic forward models in the plate frame, we apply our method to North America over the last 100 Myr. In order to make a closer link between plate motions and mantle convection, we have developed a new paleogeographic concept (‘dynamically closing plate polygons') and developed plate reconstructions 140 Ma to the present. Seismic tomography is used explicitly by using an adjoint of the equations in CitcomS. We have implemented a forward-adjoint looping that solves for the initial conditions while minimizing the difference between predicted and observed present day structure. Since the scaling from seismic anomalies to mantle temperature and mantle viscosity are both uncertain, we apply additional constraints from regional sea level observations on an inverse problem. Finally, prediction of vertical motions (dynamic topography) from the forward models is translated into the plate frame and tested with paleo-shorelines, sediment isopachs, and tectonic subsidence curves. Applied to North America since 100 Ma, we find that as North America moved westward, a long wavelength dynamic topography depression swept eastward over the continent. The Farallon slab, as imaged by seismic topography, is currently below the central Atlantic coast of the US. Sea level fall since the Cretaceous, inferred from boreholes on the Atlantic coast are estimated to be less than 100 m, 100 to 200 meters lower than inferred from either ridge volume or the average flooding of continents. Using this discrepancy between regional and eustatic sea level, as well as the well known anomalous Cretaceous subsidence (and subsequent uplift) of the western interior seaway, we are able to constrain the inverse model of mantle convection. The self-consistent model suggests that the marine flooding of the western interior sea way was not an unusual ‘event' in which North America regionally subsided. Rather, North America moved over a more or less fixed downwelling and the "event" was only recorded when eustatic sea level was elevated. The use of the US Atlantic margin as a stable reference frame for sea level is called into question and the unusually small sea level fall may have been caused by a gradual subsidence of the east coast as it moved over the Farallon slab downwelling.
U34A-08 INVITED
Seismic Anisotropy as a Constraint on Global Mantle Flow and Plate Motions
Shear flow in the asthenosphere tends to align olivine crystals in the direction of shear, producing a seismically anisotropic fabric that can be detected by shear-wave splitting or surface wave observations. These observations thus provide a strong constraint on the pattern of asthenospheric shear that accommodates the relative motion between the surface plates and the flowing mantle. To utilize this constraint, we developed models of global mantle flow driven by either surface plate motions above a passive viscous mantle (plate-driven flow) or tomographically-inferred density heterogeneity in the viscous mantle beneath a rigid lithosphere (density-driven flow). In the no-net-rotation reference frame (NNR), the relative motions of the plates control the pattern and amplitude of plate-driven flow. For reference frames that incur net lithosphere rotation (e.g., the Pacific hotspot (HS3) frame induces strong westward motion), the NNR shear pattern is superimposed upon a global net shearing of the asthenosphere because some of the net shear between the plates and the deep mantle occurs within the asthenosphere. For density-driven flow, the amplitude of asthenospheric shear flow scales inversely with the absolute mantle viscosity. Thus, the combination of plate- and density-driven flow fields, which should represent the net mantle flow, is sensitive to both the absolute mantle viscosity and the plate motion reference frame. To constrain our models, we used anisotropy inferred from global surface wave tomography studies (e.g., Debayle et al. [2005]) and SKS splitting observations from mid-plate ocean island stations where lithospheric anisotropy appears to be small. We find that the orientation of anisotropy at these stations is best fit by a flow model with a viscosity structure consistent with constraints from postglacial rebound (asthenospheric viscosity of ~ 3-7 × 1020 Pa s). We also use the data to evaluate the most appropriate deep mantle reference frame for plate motions by comparing the misfit among several commonly-used reference frames that include varying degrees of net rotation (e.g., NNR, HS3, HS2).