V44B-01 16:05h
Where Plumes Live
From the perspective of fluid dynamics, `Plumes or not?' might be the wrong question. Let me begin by defining a few terms. Plume with a `P' is the well-known thermal structure with thin (order 100 km) tail and large, bulbous head that originates at the core-mantle boundary. The thin tail/large, bulbous-head morphology has been generated in a number of laboratory and numerical experiments. It can be seen, for example, on the cover of the famous fluid dynamics text by Batchelor. There is a clearly-defined range of parameters for which this structure is the preferred solution for instabilities arising from a bottom boundary layer in a convecting fluid. For example, a strong temperature-dependent rheology is needed. By contrast, plume with a `p' is any cylindrical or quasi-cylindrical instability originating from a thermal (or thermo-chemical) boundary layer. In fluid dynamics plume is sometimes used interchangeable with jet. Unless there is a very small temperature drop across the core-mantle boundary or a rather remarkable balance between temperature and composition at the base of the mantle, there are almost certainly plumes. (Note the little p.) Are these plumes the thermal structures with thin (order 100 km) tails and large bulbous heads or could they be broad, hot regions such as the degree 2 pattern seen in global seismic tomography images of the lower mantle, or the disconnected droplets seen in chaotic convection? To study this question, I will present a sequence of numerical `experiments' that illustrate the morphology of instabilities from a basal thermal boundary layer, i.e., plumes. Some of the aspects I will present include: spherical geometry, temperature-and pressure-dependence of rheology, internal heating, pressure-dependent coefficient of thermal expansion, variable coefficient of thermal diffusivity, phase transformations, and compositional layering at the base of the mantle. The goal is to map out the parameters and conditions where Plumes live (note the big P) and to provide insight into the structures that boundary layer instabilities at the base of the mantle may take.
V44B-02 16:20h
Implications of Heat Flow in the Triaxial Earth on Layered Convection and Plume Formation
Perception of Earth as vigorous arises from the discrepancy between model-dependent estimates of global heat flux (Q) and bulk radiogenic content, which necessitate additional sources and large secular delay. Weak, layered mantle convection is instead indicated by downward revision of these parameters, and by new theoretical models and measurements on the variation of thermal conductivity (k) with temperature. Hydrothermal circulation has been used to justify Q=44 TW derived from the half-space cooling model, rather than 31 TW obtained directly from measurements, yet MOR magmatism provides at most 4 TW. The half-space cooling model assumes inappropriate 1-D boundary conditions, resulting in infinite flux along the ridge centers over all time. Geological observations, inferred mantle overturn rates, estimated mantle cooling rates, and recent geodynamic models independently suggest that neither delayed secular cooling nor primordal heat are currently significant sources, necessitating that current heat production predominately originates in radioactive decay and is quasi-steady-state. Models of Earth's bulk composition based on enstatite chondrites are sufficiently radioactive to supply Q=31 TW, contain enough iron metal to account for Earth's huge core, and have the same oxygen isotopic ratios as the bulk Earth. That Earth is now quasi-steady state is further supported by nearly uniform release of heat over the entire surface. Weak mantle convection, suggested by quasi-steady state, is compatible with circulation within a chemically distinct mantle layers, as thinner layers mean lower Rayleigh numbers. Different dynamical styles above and below 670 km are required by k(T) variations and a change from vibrational to radiative transport mechanisms. Finally, the surface expression of mantle convection is compatible with layering: Geodesic and tomographic studies indicate that lower mantle flow is dominated by a double torus. We propose that the upper mantle system is organized in response to the non-hydrostatic triaxial stress field arising from convective motions of the lower mantle. Simple conjugate shears in the lithosphere that result from triaxial deformation are occupied by oceanic ridges and make a striking "X" pattern in polar projection. Their orientation creates alternating thermal and mechanical couplings between the upper and lower mantle systems, leading to largely E-W continental drift, and to longitudinal concentration of continents and subducting slabs. Hot-spot and ridge magmatism is attributed to thermal runaway and near-solidus temperatures, rather than to material exchange with lower mantle, which is strongly impeded.
V44B-03 16:35h
Dynamics of Thermochemical Plumes
We investigate the dynamics of thermo-chemical plumes to enlighten the fundamental differences with purely thermal plumes. The key features of our 3D numerical model include: (1) a compressible mantle with an endothermic phase transition at 670km depth, (2) a mantle 'wind' induced by the imposed surface plate motion, (3) twenty million active tracers simulate denser material initially in the lowermost mantle, (4) plumes form naturally i.e., without imposing any temperature perturbation. First, we investigate the widely accepted head-tail structure of plumes. Our results show that thermo-chemical plumes reaching the surface may or may not have a head since, in some cases, only a narrow 'tail' of hot material is able to ascend in the upper mantle. Therefore, we suggest that the existence of a large igneous province at the onset of hotspot volcanism is not a valid prerequisite for a deep plume origin. Second, we investigate the entrainment of deep heterogeneities. Our results show the generation of narrow, long lasting, distinct filaments in the plume's tail. Therefore, the plume conduit is laterally heterogeneous, rather than concentrically zoned. Third, we calculate the shear wave velocity anomalies in the lower mantle, using the temperature field and the distribution of chemical heterogeneities provided by the convection model. The great variety of plume's shapes and sizes differs strikingly from the expected 'mushroom' shape of purely thermal plumes, bearing important implications for the interpretation of seismologically detected plumes. Finally, our model predictions will be compared with a variety of observations in the Central Pacific.
V44B-04 16:50h
Do unradiogenic noble gases in oceanic basalts indicate undegassed deep mantle?
Unradiogenic helium and neon isotopic compositions, found in some oceanic island volcanoes, have been interpreted by geochemists as evidence for undegassed reservoirs deep in the earth. This presentation will provide a brief review of some of the evidence for and against this standard model. The deep undegassed mantle hypothesis has been challenged by geochemists and geophysists who would prefer to explain geochemical variations by recycling and non-plume processes. One alternative explanation is that helium is more compatible than Th and U during silicate melting, which could result in unradiogenic helium isotopes (high 3He/4He ratios) in ancient depleted sources. Crystal/liquid noble gas partition coefficients are not well known, but recent laboratory studies (e.g., Brooker et al., 2003; see also Parman et al., this meeting) yield values significantly lower than earlier studies which had suggested that noble gases might behave as compatible elements during melting. Estimates based on natural basaltic phenocrysts, compared to co-existing submarine glasses, strongly suggest that helium is more incompatible than Th and U (D for olivine/melt less than 0.001). Therefore, existing data do not support the hypothesis that helium is more compatible than Th and U. Recent studies suggest that helium and neon isotopes are well correlated in submarine oceanic basalt glasses, which suggests coherent evolution of mantle (Th+U)/Ne and (Th+U)/He ratios. The correlations of helium with neon, and helium with solid radiogenic isotopes (Sr,Nd,Pb), provide important arguments against models involving complete decoupling of helium from other elements by storage in ancient lithosperic reservoirs or melting processes. The most unradiogenic helium and neon isotopic signatures are routinely found in the most active volcanic regions, also suggesting a relationship between noble gases and excess heat and melting. The global isotopic data show that unradiogenic helium and neon are most often associated with Sr, Nd and Pb isotopic compositions that are intermediate between depleted mantle and hypothetical bulk earth mantle; this demonstrates that the undegassed reservoirs are not totally primitive in geochemical composition. The early earth most likely had a huge inventory of light noble gases and a partially degassed/depleted terrestrial mantle could yield the observed isotopic characteristics. The existence of relatively undegassed mantle reservoirs in the deep earth still provides the simplest explanation for all the observations. Undegassed reservoirs could exist within the low viscosity lower mantle or the core/mantle boundary.
V44B-05 17:05h
The Standard Model for Noble Gases in Mantle Geochemistry: Some Observations and Alternatives
We evaluate the Standard Model of noble gases against a number of observational constraints of relevance to the distribution of noble gases in the Earth's mantle. These constraints include: $1$) the lack of evidence for high $^3$He/$^4$He ratios correlating with high (initial) He concentrations, $2$) that MORB and OIB $^3$He/$^4$He data do not represent two different distributions [$1$], $3$) that systematic global correlations between $^3$He/$^4$He ratios and lithophile isotopic systems are lacking, $4$) that the correlations we do observe are broadly linear, $5$) that large, local geographical $^3$He/$^4$He variations are observed, which are inconsistent with a strongly localized (i.e. plum-stem) flux of high-$^3$He/$^4$He material, and $6$) that dramatic temporal $^3$He/$^4$He variations are observed on very short time scales ($10-100$ years). Non-layered noble gas mantle models, in which the carrier of unradiogenic He is a relatively noble gas-poor phase scattered in the mantle, are more consistent with this set of constraints. We propose that the carrier of unradiogenic noble gases is primarily olivine [$2$]. Olivine-rich lithologies, produced in previous partial melting events, are a natural part of the Statistical Upper Mantle Assemblage (SUMA); a highly heterogeneous assemblage of small-to-moderate scale ($1-100$ km) enriched and depleted lithologies with a wide range in chemical composition, fertility, age and isotopic signatures. The isotopic signatures of oceanic basalts, including noble gases, are obtained by partial melting of the SUMA under slightly different P-T conditions; i.e. different degrees of partial melting and different degrees of homogenization prior to eruption [$3-5$]. Noble gas isotopic systematics do not trace deep mantle components in the source materials of oceanic basalts. They may, however, indirectly indicate potential temperature, as the order in which different mantle lithologies melt depends on pressure. References: [$1$] Anderson, EPSL $193$, $77-82$ ($2001$). [$2$] Brooker et al., Lithos, $73$, S$15$ ($2004$). [$3$] Morgan and Morgan, EPSL $170$, $215-239$ ($1999$). [$4$] Meibom and Anderson, EPSL $217$, $123-139$ ($2003$). [$5$] Ito and Mahoney, EPSL submitted ($2004$).
V44B-06 17:20h
Mantle Plume Upwelling Rates: Evidence from U-Series in Young Ocean Island Basalts
U-series disequilibria measured in recent lavas at intraplate volcanoes provide a powerful probe to examine the validity of the plume model. U-Th and U-Pa fractionation produced during melting is a function of the melting rate. In turn, this parameter should scale with mantle upwelling velocities. Simply stated, a larger melting rate (larger mantle upwelling velocity) yields smaller Th and Pa excess relative to their parent nuclides. A number of observations supports this approach: (1) there is a negative correlation between $^{230}$Th excess and buoyancy fluxes (2) based on new measurements of $^{231}$Pa in the Azores, Iceland and the Galapagos and literature data, we show here that there is also a well defined correlation between $^{231}$Pa excess and buoyancy flux (3) For Hawaii, Iceland and the Azores, $^{230}$Th excess (or $^{231}$Pa excess) increases as a function of the distance from the centre of the `hotspot'. These observations suggests that `hotspot' buoyancy fluxes are associated with a greater melt production per unit of time and that the centre of `hotspot' corresponds to a faster mantle upwelling velocity than its periphery. This is therefore in strong support of a model where ocean islands are associated with faster upwelling at depth. However, there is in fact not a simple relationship between melt productivity and upwelling velocities. Notably, the presence of volatiles, of mafic lithologies or of variably enriched peridotitic source could all affect melting rate and hence U-Th-Pa fractionation. We have considered these issues in great detail using a large data base for the Azores islands. While there are clear variations in mantle source composition, they cannot explain the observations of increasing $^{231}$Pa/$^{235}$U ratio with distance from the centre of the Azores hotspot . If we take into account the effect of water in the source of the Azores, it clearly affects the scaling between U-series fractionation and upwelling velocity but not the overall conclusions.
V44B-07 INVITED 17:35h
Re, Os, and Pt Fractionation by Melt Segregation
If zero-age basalts are enriched with respect to 187Os/188Os relative to present-day primitive mantle, one may assume that they tap reservoirs in the mantle that are superchondritic with respect to Re/Os and/or 187Os/188Os. Most interesting are coupled enrichments in 186Os/188Os and 187Os/188Os; if these signatures could only be derived from the outer core, they would testify that some mantle plumes indeed originate at the core-mantle boundary. We report experiments with (Fe,Ni,Cu)1-xS monosulfide in silicate mantle matrix that quantify noble metal fractionation during partial silicate melting. Our model permits the derivation of isotopically enriched melts from primitive mantle sources with time-integrated chondritic Os-isotope ratios. Melting experiments of (Fe,Ni,Cu)1-xS monosulfide in fertile mantle matrix to 1400°C and 3.5 GPa show that two sulfide phases are stable at the dry silicate solidus, a crystalline FeS-rich monosulfide and a Cu2S-enriched sulfide melt. The noble metals fractionate between the sulfide phases: Os, Ir and Ru into crystalline monosulfide, and Re, Rh, Pt, and Pt into the sulfide melt. During silicate melt segregation, crystalline monosulfide remains with silicate minerals in the mantle, concentrating Os, Ir, and Ru in the residue. The molten sulfide fraction is entrained in the silicate melt as immiscible droplets and is drained from the mantle along with the silicate melt, defining the noble metal inventory of the basaltic component. Physical processes are more important in fractionating the noble metals than chemical partitioning laws. The noble metal contribution to a basaltic melt by sulfide-silicate partitioning is small. Principally, it is possible to produce basalts with superchondritic Os-isotope ratios from chondritic mantle sources, as long as there is compositional heterogeneity among the mantle sulfides. Partial melting preferentially mobilizes and selectively entrains in the silicate melt sulfide compositions with low melting points, i.e., compositions rich in Cu, Re, Pt, and Pd. If sulfide heterogeneity is an ancient signature of the mantle, old enough to allow sufficient ingrowth of radiogenic 186Os and 187Os, the entrained sulfide component will on average be more radiogenic than bulk mantle, and so will be the basalt. There is indeed evidence for grain-scale heterogeneity among mantle sulfides both with respect to major elements, noble metals, and Os-isotopes. The key question is whether such heterogeneities may survive partial melting and may be passed on to a segregating silicate melt.