DI51A-01 INVITED
Early planetary evolution: The crust and mantle before convection
Terrestrial planets are expected to melt to some extent during their accretion. The melting extent depends upon planetary mass and composition and accretionary history, and the planet is likely to experience serial magma oceans of partial or even whole mantle depth. Magma ocean solidification, even in the presence of a thick atmosphere, is likely to be on the order of or shorter than the timescale of giant accretionary impacts. When the majority of accretion is complete the planet's mantle is therefore likely to consist of regions that have been melted and solidified more than once, possibly in combination with some remaining undifferentiated material. Fractional solidification of magma bodies proceeds from the bottom up and produces compositional and density stratification among the resulting silicate cumulates. The densest material is likely to be the last solidifying, as iron has been progressively enriched during solidification, and therefore it will also contain an excess of incompatible and radiogenic elements. Initially, this latest-solidifying, most iron-rich materials will form nearest the surface. This unstable stratigraphy will overturn. This densest material will likely sink deep in the mantle, displacing upward less dense differentiated and even undifferentiated materials. Material that rises up in the mantle may melt adiabatically to produce the earliest crust, or that crust may have been produced by mineral flotation in a magma ocean, as on the Moon. The mantle before convective onset may then consist of density-stratified silicate differentiates, with or without intermixed undifferentiated material. Many models produce deep dense reservoirs unlikely to be remixed by later convection, particularly in larger planets, and all models produce mantles resistant to convective onset through density stratification. Initial mantle compositions in the range of chondritic meteorite compositions can produce magma oceans that float quartz in addition to plagioclase, and mantles that during overturn produce earliest andesitic rather than basaltic crusts. The more silica-rich early crusts are likely to be more stable because of their high buoyancy.
DI51A-02 INVITED
Supercontinents, Plate Tectonics, Large Igneous Provinces and Deep Mantle Heterogeneities
The formation and break-up of supercontinents is a spectacular demonstration of the Earth's dynamic
nature. Pangea, the best-documented supercontinent, formed at the end of the Palaeozoic era (320 Ma) and
its dispersal, starting in the Early Jurassic (190 Ma), was preceded by and associated with widespread
volcanic activity, much of which produced Large Igneous Provinces (LIPs), but whether any of the heat or
material involved in the generation of LIP rocks comes from greater depths has remained controversial. Two
antipodal Large Low Shear wave Velocity Provinces with centre of mass somewhat south of the equator
(African and Pacific LLSVPs), isolated within the faster parts of the deep mantle dominate all global shear-
wave tomography models. We have tested eight global models and two D" models: They all show that deep-
plume sourced hotspots and most reconstructed LIPs for the last 300 million years project radially downwards
to the core-mantle-boundary near the edges of the LLSVPs showing that the plumes that made those
hotspots and LIPS came only from those plume generation zones. This is a robust result because it is
observed in multiple reference frames, i.e. fixed/moving hotspot and palaeomagnetic frames, and in the latter
case whether the effect of True Polar Wander (TPW) is considered or not. Our observations show that the
LLSVPs must have remained essentially stable in their present position for the last 300 million years. LIPs
have erupted since the Archean and may all have been derived from the margins of LLSVPs but whether the
African and Pacific LLSVPs have remained the same throughout Earth's history is less certain although
analogous structures on Mars do indicate long-term stability on that planet. Deep mantle heterogeneities and
the geoid have remained very stable for the last 300 million years, and the possibility is therefore open for
speculating on links to Pangea assembly. In a numerical model, Zhong et al. (2007, EPSL) argued that
Pangea assembled above a major down-welling, and calculated that, following the assembly, a sub-Pangea
upwelling developed relatively fast (within~50 Myr) as mantle return flow in response to circum-Pangea
subduction. Collision of Gondwana and Laurussia took place during the destruction of the Rheic Ocean and
parts of the Palaeotethys and bulk Pangea assembled at ca. 320 Ma. However, most of the Rheic Ocean had
gone much earlier (ca. 370 Ma) and it may therefore be more appropriate in terms of mantle modelling to
place "supercontinent formation" in Devonian times. In that case a large-scale thermal upwelling under Africa,
and the presumed chemically distinct African LLSVP beneath it, would have existed as early as 320 Ma, so
that a plume head from its edge could have impinged upon the lithosphere at the time determined for the
oldest LIP we have reconstructed. In such a model, the African LLSVP should not have existed before
Devonian times, because convection would have been dominated by a degree-1 mode with only one
upwelling, presumably above the Pacific LLSVP. The situation becomes more obscure further back in time;
TPW may have been larger during degree-1 convection, so that reconstructions in the palaeomagnetic frame
are not necessarily in relation to the deep mantle. There may again have been two LLSVPs during dispersal
of the previous supercontinent (Rodinia), in which case a second LLSVP may again have been approximately
antipodal to the one beneath the Pacific.
http://www.geodynamics.no
DI51A-03
Time Evolution of the Mantle Thermal Structure in the African Hemisphere Before and After the Formation of Pangea
The present-day mantle structure is characterized by the African and Pacific superplumes surrounded by
subduction slabs. This structure has been demonstrated to result from dynamic interaction between mantle
convection and surface plate motion history in the last 120 Ma. With similar techniques, mantle structure has
been constructed back to about 100 Ma ago. However, due to the lack in global plate motion reconstructions
further back in time, mantle structure for earlier times is poorly understood, despite of their importance in
understanding the continental tectonics and volcanisms. Zhong et al. (2007) suggested that the mantle
structures alternate between spherical harmonic degrees-1 and -2 structures, modulated by supercontinent
processes. In their model, a supercontinent forms in the hemisphere with cold downwellings, and after
supercontinent formation, the cold downwellings are replaced with hot upwellings due to return flows
associated with circum-supercontinent subduction. This model implies that the African superplume is younger
than 330 Ma when Pangea was formed, which is supported by volcanic activities recorded on continents
around Pangea time. By using paleomagnetic-geologically reconstructed continental motions between 500
and 200 Ma in a three-dimensional spherical models of mantle convection, this study, for the first time,
investigates the time evolution of mantle structures in the African hemisphere associated with Pangea
formation. We show that cold downwellings first develop in the mantle between the colliding Laurentia and
Gondwana, and that the downwellings are then replaced by upwellings after the formation of Pangea and as
circum-Pangea subduction is initiated, consistent with Zhong et al. (2007) and Li et al. (2008). We find that
the return flows in response to the circum-Pangea subduction are responsible for the upwellings below
Pangea. We also find that even if the mantle in the African hemisphere is initially occupied by hot upwellings,
the cold downwellings associated with convergence between Laurentia and Gondwana would destroy the hot
upwellings and cause the hemisphere to be cold. These results are insensitive to model parameters such as
convective vigor, internal heating rate, and the plate motions in the oceanic hemisphere. We therefore
suggest that the African superplume is younger than 330 Ma when Pangea was formed.
http://anquetil.colorado.edu/~nzhang/NanZhangCU.html
DI51A-04
Geodynamic Model of the New Zealand-Antarctica Conjugate Margin Since the Late Cretaceous
Present-day plate boundaries in the Southwest Pacific were created through continental rifting since Late Cretaceous. Spreading on the Pacific-Antarctic Ridge and the southeast Indian Ridge started around 95-90 Ma and continues to the present. The New Zealand and Antarctica conjugate margins are characterized by a number of anomalous observations. The Antarctica margin and adjacent sea floor is approximately 1 km shallower than the conjugate Campbell plateau. Backstripping of the sediments from the boreholes in Campbell plateau indicates anomalously high tectonic subsidence, with a rapid subsidence phase in the period 70-40 Ma, coincident with northward drift of the Campbell plateau. Finally, the Antarctic margin is associated with a large negative geoid anomaly and low S-wave seismic velocities. We developed a three-dimensional geodynamic model of the New Zealand-Antarctica conjugate margins for last 100 million years. We used global finite-element models of mantle flow (with CitcomS) while imposing surface velocities based on the plate reconstructions from GPlates. Models are iteratively updated based on the discrepancy between predicted dynamic topography and observed borehole subsidence on the Campbell plateau and the Antarctic margin topography. Additional constraints on the model are imposed based on the discrepancies between predicted and observed present-day geoid. Long-wavelength features of anomalous observations on the New Zealand-Antarctica conjugate margin can be attributed to time-dependent evolution of a large-scale mantle upwelling. The mantle upwelling at the present day is located on the Antarctic margin, causing a relatively long-lived topography high. Campbell plateau experiences tectonic subsidence as it drifts away from this mantle upwelling. Geoid low observed on the Antarctic margin is associated with both mantle upwelling and lower mantle "slab graveyards".
DI51A-05
Transitions in Tectonic Mode Based on Calculations of Self-Consistent Plate Tectonics in a 3D Spherical Shell.
In the past decade, several studies have documented the effectiveness of plastic yielding in causing a basic approximation of plate tectonic behavior in mantle convection models with strongly temperature dependent viscosity, strong enough to form a rigid lid in the absence of yielding. The vast majority of such research to date has been in either two-dimensional, or three-dimensional cartesian geometry. Also, scalings for mixed internally and bottom heated convection are not well established. In our previous study (van Heck and Tackley, 2008), mantle convection calculations were performed to investigate the planforms of self consistent tectonic plates in three-dimensional spherical geometry. We found, for internally heated convection and fixed Rayleigh number, that when yield stress of the lithosphere is low a "great circle"-subduction zone forms. At low-intermediate yield stresses plates, spreading centers and subduction zones formed and were destroyed over time. At high-intermediate yield stresses two plates form, separated by a great circle boundary that is a spreading centre on one side and a subduction zone on the other side. At high yield stresses a rigid lid was observed. Here, the planforms found by van Heck and Tackley (2008) are investigated further, leading to a more general understanding of how different parameters determine which planform prevails. New calculations are performed to investigate the effect of varying Rayleigh number and different internal/bottom heating ratios. Several diagnostics are used to analyze how successful each model is in producing tectonic plates. Cases with zero internal heating are compared to cases which have both internal heating and bottom heating. The results are compared to analytical scalings for boundary regimes as well as scalings for heat flux. This allows us to scale to different planets of different sizes and can be applied to the evolution of Earth, Mars and Venus as well as terrestrial extra-solar planets. Also, we can study the tectonic evolution of a cooling planet. As radioactive heat production decreases over time the tectonic mode (e.g. changes in plate size, rigid lid convection to tectonic plates, smoothly evolving plates to more episodic, time dependent, tectonics) is likely to change. Grigné et al. (2005) showed that surface heat flux depends on the wavelength of convection so the scalings for heat flux will not only depend on the internal/bottom heating ratio but also on the planform since the wavelength of convection (i.e. plate size) changes for different planforms (van Heck and Tackley, 2008).
DI51A-06 INVITED
Past and present seafloor age distributions and the temporal evolution of plate tectonic heat transport
Variations in the rates of seafloor generation and recycling have potentially far-reaching consequences for sea level, ocean chemistry and climate. A parameterized framework to describe such variations could guide the study of non-uniformitarian plate tectonic activity, but there is little agreement on the appropriate mechanical description of the surface boundary layer. A strong constraint on the statistics of oceanic convection systems comes from the preserved seafloor age distribution, and additional inferences are possible when paleo-seafloor is modeled based on plate motion reconstructions. Based on previously reconstructed seafloor ages, we recently inferred that oceanic heat flow was larger by ~15% at 60~Ma than today. This signal is mainly caused by the smaller plates that existed previously in the Pacific basin with relatively larger ridge-proximal area of young seafloor. The associated decrease in heat flow is larger than any plausible decrease due to cooling, and therefore hint at cyclic behavior in plate tectonics. We also consider area-per-age statistics for the present-day and back to 140~Ma from new paleo-age reconstructions. Using a simplified seafloor age evolution model we explore which physical parameterizations for the average behavior of the oceanic lithosphere are compatible with broad trends in the data. In particular, we show that a subduction probability based on lithospheric buoyancy ("sqrt(age)") leads to results that are comparable to, or better than, that of the probability distribution that is required to obtain the "triangular" age distribution with age-independent destruction of ocean floor. The current, near triangular distribution of ages and the relative lull in heat flow are likely only snapshots of a transient state during the Wilson cycle. Current seafloor ages still contain hints of a ≤sssim 60~Myr period, cyclic variation of seafloor production, and using paleo-ages for 140~Ma, we find a ~ 400~Myr best-fitting variation that is broadly consistent with geologically based reconstructions of production rate variations. From the new reconstructions, a consistent decrease of total oceanic heat flow by ~ -0.25%/Myr over the last 140~Ma is inferred. Our study provides some of the required input for an improved understanding of the non-uniformitarian evolution of plate tectonics and the interplay between continental cycles and the self-organization of the oceanic plates.
DI51A-07
Signatures of Downgoing Plate-Buoyancy Driven Subduction in Cenozoic 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 Cenozoic compilation by Sdrolias & Müller (2006). We find that: (a) Around 90% of Cenozoic plate convergence is achieved by advance of the downgoing plate towards the trench, requiring a low-drag asthenosphere (as for a viscosity 10-2 to 10-3 times the upper-mantle average), as well as plate widths > 2000 km. (b) Present-day sinking velocities, and downgoing-plate motions throughout the Cenozoic are as expected for slabs driven by their own upper-mantle buoyancy. In a few cases, young plates move at velocities that require a higher driving force, most likely supplied by lower-mantle-slab induced flow (c) Steep present-day slab dips imply that plate resistance to bending is low, as for effective viscosities 102 times that of the upper mantle. Dips either in- or decrease with plate age, evidence of a nonlinear response of the plate to high bending stresses. (d) Throughout the Cenozoic, 80% of the trench sections retreat. Trench-plate motion correlations range from strongly positive to strongly negative. This variability can be explained by regional factors that encourage/hamper plate motion or hamper/encourage trench motions. On average, trench motion is small, and often very oblique (mean angle of 73°) to the direction of downgoing-plate motion, most likely due to constraints imposed by the upper plate. Thus emerges a picture of (upper-mantle) slab pull driven, relatively free, subduction, where motion partitioning and slab geometry adjust to external constraints on trench motions.
DI51A-08 INVITED
Ridges and Hot Spots: Reconciling Isotopes and Major Elements
Meyzen et al. (2007) combined the radiogenic isotope data of several hundred MORB samples along a single mid-ocean ridge profile extending from the northernmost Atlantic to the Indian over to the Pacific Ocean covering >400 degrees. A remarkable finding was that the total reduced variance on Sr-Nd-Hf-Pb data, hereafter referred to as "isotopic variance", showed conspicuous maxima and that a harmonic analysis of this variance showed a periodic spacing of the maxima with a mean value of ~35° (actually a doublet, which is a consequence of modulation by a hemispheric contrast). The strong but unexpected hint was that hot spots are nearly periodically spaced along the ridge systems. To explore whether the isotopic variations in the mantle are controlled by physical properties of the mantle source, such as its thermal state and major element composition, we estimated the apparent temperatures and pressures of equilibration using newly calibrated thermometers and barometers of Lee et al. (submitted) based on the most recent compilation of experimental data. We calculated T-P couples for over 3000 MORB samples after correcting for olivine fractionation. Peaks appear for pressure and even stronger for temperature at the precise same localities along the ridges where the presence of hot spots has been inferred. Periodograms of temperature estimates were calculated which produced a spectrum similar to that of the isotopic variance with the same conspicuous doublet and the same mean spacing of ~35° with a total power >40%. Pressure estimates show similar features with a lesser signal/noise ratio but we suspect that these features may largely reflect the rather strong correlation between errors on Tand P. These results, based on two independent data sets, leave little doubt about deep mantle upwellings with high potential temperatures underpinning mid-ocean ridges. However, the regular spacing of hot spots along mid-ocean ridges remains an unsolved conundrum. Meyzen etal. (2007) Nature, 447, 1069-1074.