V43G-01 INVITED 13:45h
Constraints on Mantle Thermal Variations from the Sedimentary Record of Large Igneous Provinces
One of the characteristics of all models for mantle plumes, whether they are deep or shallow rooted is the presence of anomalously hot asthenosphere underlying the lithospheric plate, the melting of which is believed to produce large igneous provinces (LIP). Surface uplift is driven by the upward flow of mantle, as well as a lower density caused by excess heat and melt extraction. Shallowing of the seafloor around Iceland, Hawaii and regions of French Polynesia are characteristic of modern plume activity and should be preserved in the sediment cover of old LIPs. Removal of the LIP from over the plume would result in more subsidence than observed in regular oceanic crust. However, studies of the sedimentary cover from a range of seamounts, plateaus and ridges of various ages from all major ocean basins do not always show this greater than expected subsidence. Drill sites from the North Atlantic margins show the clearest consistent evidence for subsidence anomalies that could be caused by mantle temperature anomalies of around 100?C for a 100 km thick plume layer. However, several examples (e.g., Walvis Ridge, Rio Grande Ridge, Iceland-Faeroe Ridge, MIT Guyot, Hess Rise, 90East Ridge) show no resolvable differences with the normal oceanic crust and preclude major thermal anomalies under them at or shortly after the time of their emplacement. Most unusually in some LIPs (e.g., Ontong Java, Manihiki, Magellan and Shatsky Rises) subsidence is slower than for normal oceanic lithosphere, suggesting either colder than normal mantle temperatures, or more likely the emplacement of a buoyant lithospheric root under the magmatic province at the time of its formation caused by melt extraction. Gradual emplacement of the LIP crust may also slow the net basement subsidence. In these slow subsiding examples the sediments do not preclude or require hotter than normal mantle involved with LIP generation, but instead indicate that the lithosphere in these provinces was formed by processes that are quite different from those operating at mid ocean ridges. It is not clear why the buoyant depleted root under Hawaii disperses with time after magmatism, yet those under the Pacific plateaus do not, although this implies differences in viscosity that could be temperature related.
http://www.mantleplumes.org/SedTemp.html
V43G-02 INVITED 14:00h
Hot, warm, cold; wet, damp, dry; peridotite, pyroxenite, eclogite; do petrologists know anything about mid-ocean ridge and ocean island basalt sources?
Igneous petrology can offer essential constraints on models of spreading center and intraplate volcanism to complement information drawn from seismology, geophysics, and geochemistry. However, as with all these other disciplines, inferences from petrological data are often non-unique and model-dependent. Petrology will be most useful to the general plume debate when and if it can uniquely invert for the temperature, volatile content, and major element composition of the mantle sources of erupted basaltic lavas. When instead there are ambiguities it is important to acknowledge these lest preferred models be taken as fact. Or, rather than attempting to invert for source information, petrologists might content themselves with running forward models to test hypotheses proposed by others. This is a well-defined task, free from ambiguity, and consistent with a conservative falsification approach to science. It is also of the first importance for all parties to avoid over-generalization of their arguments and false grouping of different localities into one category; proposed mantle plumes must be evaluated one at a time, rather than collectively. Furthermore, in order to be generally accepted, models must be able to explain all the observable features of a volcanic chain: the longevity and fixity (or not), the magma and buoyancy fluxes, the trace element and isotopic (lithophile, noble gases, and stable) character, and the distribution (in time and space) of major-element lava types. Such models must also be consistent with mass and energy conservation and known phase equilibria. Two tasks that are quite straightforward at present are (1) inference of the MgO content of the most primitive demonstrable parental lava in a suite from observed liquid and olivine phenocryst compositions and (2) the estimation of source parameters assuming dry peridotite melting. Although in principle an arbitrary amount of $H_{2}O$ in the primary magma might depress the liquidus temperature at any particular MgO as much as desired, in practice there are generally limits on parental $H_{2}O$ contents from melt inclusions or submarine eruptions. Among the remaining model-dependent uncertainties are the estimation of potential temperature from liquidus temperatures and the estimation of source parameters when wet melting, eclogite sources, or mixed lithology sources are considered. It is important to emphasize that a parental liquidus temperature can never be anything but a lower bound on source potential temperature. Any number of processes, most notably adiabatic melting and near-surface cooling, lower the temperature and may leave no record in the phenocryst population. To actually determine the potential temperature it is necessary to find a unique and self-consistent forward model that generates the appropriate parental melt compositions and at the same time sufficient melt volume. In this talk, I will focus on wet melting and the maximum effect that water might have on increased magma production and on over-estimation of liquidus temperatures. I will use the mid-Atlantic ridge near the Azores, the Reykjanes ridge and Iceland, the Galapagos Spreading Center, and Hawaii as examples.
V43G-03 14:15h
Hot Hawaii, Cold Ridges, Mantle Heterogeneity, and Plumes
We use model-system phase relations in the CaO-MgO-Al$_{2}$O$_{3}$-SiO$_{2}$ (CMAS) and CaO-MgO-Al$_{2}$O$_{3}$-SiO$_{2}$-Na$_{2}$O-FeO (CMASNF) systems at 1 atm to 6 GPa to compare melt generation and crystallization processes of tholeiitic basalts at Hawaii and oceanic ridges. At both localities, erupted melt compositions are strongly controlled by low-pressure fractional crystallization of magmas generated at greater depths. Also, the Mg numbers of the most primitive melts from each locality are nearly the same (MORB, 72.1; Hawaii, 72.4 - when Fe$^{2+}$/(Fe$^{2+}$ + Fe$^3+}$) = 0.91). However, in other respects, the compositions of these most primitive basalts are quite different, and the phase relations indicate that in both cases they are only slightly less primitive than their respective parental primary melts. At Hawaii, the phase relations support generation of picritic tholeiitic melts at $\sim$5 GPa and $1565\deg$C (Gudfinnsson and Presnall, 2004), whereas at ridges, the conditions are $\sim$0.9-1.5 GPa and 1260-$1280\deg$C (Presnall {\it et al}., 2002). In Hawaii, the trend of picritic melt compositions indicates olivine-controlled fractionation, not a polybaric melting column like that suggested by Klein and Langmuir (1987) for MORBs. For the MORB modeling of Klein and Langmuir (1967) and Langmuir {\it et al}. (1992), which employs polybaric melting columns extending to 4 GPa, the phase relations show that aggregate melts would be produced that require significant low-pressure olivine-controlled fractionation in order to reach the field of observed MORB glasses. No trace of this fractionation has ever been observed in MORBs, even at Iceland. Furthermore, because the phase relations show that an inverse correlation of Na8 with Fe8 can be produced by melting of a heterogeneous mantle in the 0.9-1.5 GPa pressure interval (Presnall{\it et al}., 2002), this correlation cannot be used as an indicator of widely varying temperature. Mantle heterogeneity produced by recycling of oceanic crust and underlying depleted peridotite back into the source region for ridge volcanism would produce little change in the temperatures required for MORB generation in the plag/sp lherzolite transition. However, strong variations in melt productivity would be expected and the compositional range of basalts erupted would be expanded. No petrological evidence for ascending plumes driven by high temperatures appears to exist anywhere along the oceanic ridge system. However, some volcanic centers ({\it e. g}. Galapagos) may be caused by diapirism of low-density, major-element depleted peridotite recycled into the mantle at subduction zones (Presnall and Helsley, 1982). Low-velocity regions extending to depths $>$200 km beneath Iceland, Afar, and Easter (Ritsema and Allen, 2003) could be caused by carbonate-induced melting at low melt-fractions in an eclogite-enriched source rather than by elevated temperature. If temperatures in the central Pacific are generally high due to lithospheric blanketing, the high temperature indicated at Hawaii may not indicate a plume.
V43G-04 14:30h
How Many Hotspots are on Present-day Earth, and are all Plumes hot?
The petrological characteristics of primary magmas that exit the melting regime are sensitive indicators of mantle potential temperature. However, most primary magmas partially crystallize some olivine during transit to the surface, and erupted lavas are typically hybrid mixtures of olivine and solidified liquid. Primitive glass on the surface can have an MgO content that is lower than a parental magma from which it was derived, and a parental magma can differ from its primary magma by partial crystallization of olivine in a crustal magma chamber. However, the parental magma composition can be restored using a simple petrological procedure when olivine is the sole phenocryst phase. On Kilauea the most primitive magnesian glass has been reported to contain 15% MgO, and the most magnesian olivines contain Fo 90-91. The exchange coefficients (Kd) for FeO and MgO between these olivine and glass compositions are 0.25-0.28, much lower than 0.33-0.34 for olivine equilibrated with liquid in melting experiments. The only way to obtain the correct Kd is by computing the effects of dissolving olivine into a 15% MgO liquid composition. This procedure results in a crustal parental magma with 17-19% MgO and a mantle primary magma with 18-20% MgO. The potential crystallization temperature for Kilauea is 1400C, an estimate that includes the effects of 0.34% H2O. Hawaii is therefore a hotspot. This is the most fundamental geological constraint that all models are required to satisfy. It is independent of ongoing questions concerning the role of subducted crust and pyroxenite in the melting regime. A primary magma with 18-20% MgO is successfully reproduced by decompression melting in a hot plume with potential temperatures in excess of 1550C. Hawaii is the only hotspot Earth at the present time. The mantle below Iceland is comparatively cooler, warmer than oceanic ridges, but it was hotter during the early Tertiary. A preliminary analysis of volcanics in and around the African and South Pacific superplumes indicates low extents of wet melting and potential temperatures that may be comparable to oceanic ridges (~1300-1400C). More work is needed for a quantitative petrological evaluation, but it is clear that these volcanoes cannot be hotspots even though they are associated with broad and narrow regions of slow seismic velocities that extend deep into the mantle. The implication is that most plumes or superplumes are buoyant for compositional reasons at or close to ambient mantle temperatures, and they are distinct from the Hawaiian hotspot.
V43G-05 14:55h
Plumes or Not? Yes, and Plenty!
We present confirmation of the detection of deep mantle plumes, earlier imaged using P waves (Montelli et al., Science 2004) using a finite-frequency inversion of long period S waves. Our data set comprises 69079 S traveltimes, 26337 SS-S and 13856 ScS-S differential traveltimes. We invert for both velocity anomalies, origin times, and the relocation of the 6834 hypocenters, using the banana-doughnut kernels derived by Dahlen et al. (GJI, 2000). The S-wave images confirm the presence of well resolved deep mantle plume beneath Ascension, Azores, Canary, Easter, Samoa and Tahiti. Among the deep plumes that were not very well resolved in the earlier P-wave study, the S wave inversion shows a robust extension all the way to the CMB of the plumes beneath Cape Verde, Cook Island and Kerguelen. The presence of plumes rising from the base of the mantle but not reaching yet the surface in the Coral Sea, East of Solomon and South of Java is validated. Plumes such as Bowie, Eifel, Etna and Seychelles remain mostly confined to the upper mantle. However, the new S-wave images reopen the question on the depth extent of Iceland and Galapagos plumes. The weakening of the plume in the mid-mantle beneath Iceland is confirmed, but the S inversion clearly shows the presence of a low velocity zone at greater depth that was not visible in the P-wave images. For Galapagos, the new S-wave images show more clearly a possible connection of the plume with a broad low velocity anomaly in the lowermost mantle that feeds Easter as well. We will present the final S-wave plume images and will provide a synthesis of our findings in the light of existing ideas about plume characteristics and their superficial signature.
V43G-06 15:10h
The Boundary Between the Upper and Lower Mantle
The debate on the style of mantle convection continues partly because information from seismic tomography appears to be contradictory. There are strong arguments in favor of a major obstacle to the flow across the 650 km discontinuity. These include a change in the spectrum of lateral heterogeneity above and below the boundary, with the "red" spectrum of the velocity anomalies in the transition zone changing, either abruptly or very rapidly, to the white spectrum in the middle mantle. There are large wavelength (2,000-3,000 km) perturbations in the topography of the 650 km discontinuity with an amplitude of up to 20 km; the pattern of these perturbations matches well that of the velocity anomalies in the transition zone, with fast velocities being underlaid by depressions. The topographies of the 410 km and 650 km discontinuities are uncorrelated, indicating that they are not caused by a large scale mantle flow. Ponding of the subducted material in the transition zone seems to be consistent with all these facts. In addition, there is information on the deep earthquakes that supports the hypothesis of slab accumulation. The most outstanding example is a 500-600 km long zone of earthquakes under the Fiji Plateau with focal depths greater than 600 km. Isolated events, distant by 200-300 km from the main Wadati-Benioff zone, are also observed in South America, Tonga, Kuriles, Izu-Bonin and Banda Sea. These events are rare on the decadal time scale, but may be widespread over thousands or millions of years. Also, there is a frequent change in the orientation of the principal stress axes as the slab approaches the 650 km boundary. On the other hand, the tomographic images of the lowermost mantle bear resemblance to the near-surface tectonics. The circum-Pacific ring of fast velocities correlates well with the past locations of the subduction zones. There is also a significant concentration of the hot-spots in the regions of slow velocities in the lowermost mantle: in particular, above the Equatorial Pacific Plume Group and the Great African Plume (Dziewonski {\it et al.}, 1993). It appears that some connection between the upper and lower mantle must exist; whether it is direct or indirect cannot be determined from seismic tomography alone.