Deep Earth DIALOG
This is the twelfth issue of the newsletter of SEDI, an IUGG Union Committee
to Study the Earth's Deep Interior. Requests for copies of the earlier
issues should be addressed to David Loper, Geophysical Fluid Dynamics Institute,
Florida State University, Tallahassee, Florida 32306-3017, U.S.A, faxed
to (904) 644-8972 or emailed to email@example.com. Items for the next
issue or notifications of change of address should be sent to firstname.lastname@example.org.
- 8th Sedi Symposium
- Session 1: Formation and Composition of the Core
- Session 2: Mineral Physics: Properties of Iron and Alloys, and of the Lowermost Mantle
- Session 3: Thermal History of the Geodynamo
- Session 4: Inner Core: Structure and Dynamics
- Session 5: Outer Core: Structure and Dynamics
- Session 6: Dynamos and the Deep Earth
- Session 7: Stealth Layers, D" and Core-Mantle Interactions
8th Sedi Symposium, Lake Tahoe, CA, USA, 22nd - 26th July 2002.
The eighth SEDI symposium, "Geophysical and Geochemical Evolution of the Deep Earth", was held at Granlibakken, Tahoe City, CA, USA, from July 22nd to July 26th, 2002. The local organizing committee comprised Cathy Constable and Guy Masters. A special issue of Physics of the Earth and Planetary Interiors, with guest editors Michael Bergman and Peter Shearer, will be based on talks and posters presented at the meeting. Brief summaries of the sessions are found below.
At the meeting, the Doornboos prizes for outstanding contributions from young scientists were awarded, commemorating the life and work of Durk Doornboos. The recipients were Dario Alfe, Richard Holme, and Stephan Labrosse.
At the business meeting, there was vigorous discussion of the next meeting site, with possible locations being Mexico and Germany. It was decided that the 2004 meeting will be held in Germany, with Mexico a possibility for 2006. There was also some discussion about how SEDI might broaden and increase the number of subscribers.
Session 1: Formation and Composition of the Core
Session chair: Bernie Wood
Invited speakers: Dave Stevenson and Bill McDonough
Stevenson discussed the processes responsible for core formation, including settling of immiscible iron droplets in a molten magma ocean, percolation through a solid silicate matrix, Rayleigh-Taylor instability, and magma fracturing, analogous to volcanism in the lithosphere. In actuality, the core may have formed by a mixture of all four of these processes. The size, timing, and composition of the precursor bodies are likely to have played a pivotal role in determining the role of each. The dominant process may have dictated whether the core formed in equilibrium, and hence helped shape the composition of the core. He also suggested that the core could have formed in a superheated (greater than the adiabatic) state. Although the inner core would thus not exist for awhile after the time of core formation, the age of the inner core remains uncertain. Progress can be made by a better understanding of the relevant phase diagrams at mantle and core pressures and temperatures, the fluid mechanics of impacts, and the transport of metallic/silicate mixtures.
McDonough covered the geochemical evidence that bears on the composition, formation, and evolution of the core. Although meteorites give us a guide as to the composition of the bulk Earth, and hence the core, the diversity of meteorites does not allow for a unique determination of the composition. However, 1.5-2 weight percent S is a likely candidate for part of the density deficit of the core. The remaining deficit remains an uncertain mixure of O and perhaps H, C, or P. On the other hand, the volatile element depletion suggests that there is little K in the core. Moreover, there is no evidence that U or Th are soluble in Fe alloy melts at high pressure, so that there would appear to be little radioactivity in the core. U-Pb and W-Hf isotope systems suggest that core formation continued past 25 Ma after the Earth formed, but not past about 3.8 Ga. Small variations in various isotope ratios imply less than a 1 weight percent secular interchange between mantle and core. Recent Os isotope signatures interpreted as being due to an upwelling plume from the CMB involve a mass interchange that is smaller by two orders of magnitude, and in any case there may be other interpretations that do not invoke the core.
Discussion concerned whether the concentration of K in the core must
necessarily follow the volatitlity curve (due to condensation in the Earth's
precursor bodies), and how the lack of K might imply an inner core younger
than about 1 Ga.
Session 2: Mineral Physics: Properties of Iron and Alloys, and of the Lowermost Mantle
Session chair: Guy Masters
Invited speakers: J. Michael Brown and Don Weidner
Brown reviewed the phase diagram of iron. Although hcp iron appears to have a wide range of stability at high pressures and temperatures, some evidence also suggests the existence of a related dhcp phase. There is also evidence for the stability of bcc iron at high temperatures (and pressures), but perhaps above IC temperatures. At 330 GPa the melting temperature of pure iron is in the range 5000-5800 K. First principles calculations and laboratory experiments are giving new results on the elastic properties of iron, but there is not yet consensus. Diamond-anvil experiments incorporating non-hydrostatic pressure show the c-axis is fast, but interpretation of the experiments is not easy. On the other hand, first principles calculations at finite temperatures show the c-axis is slow, and with a greater anisotropy than at 0 K (at which the c-axis is fast).
Weidner spoke about the properties of perovskites, the minerals that dominate the lower mantle. Mg-perovskite comprises about 76 percent of the lower mantle. The elastic properties of Mg-perovskite at low pressures are strongly affected by the presence of Al in the two cation sites, but it is not yet clear whether this is true at higher pressures. Mg-perovskite is orthorhombic, deriving from a cubic phase via a simple distortion that may dominate the strength of Mg-perovskite, at least at low temperatures. Ca-perovskite comprises about 8 percent of the lower mantle. It is not clear whether this mineral is cubic or orthorhombic, but, because it is ferroelastic, the phase transition plays a large role in its physical properties, such as a low shear modulus near the transition. This could yield an increase in the shear modulus with temperature, though this has not yet been observed in the Earth.
Two posters (Alfe, et al. and Vocadlo, et al.) concerned quantum mechanical
calculations of iron under inner core conditions: the melting temperature
and the possible stability of the bcc phase, which cannot be ruled out.
Another poster (Gilder, et al.) investigated experimentally the magnetic
properties of high pressure iron, in particular, the possibility that iron
under inner core conditions might be sufficiently paramagnetic or even
ferromagnetic so as to affect the geodynamo. A fourth poster (Calderwood)
examined the thermal conductivity of the lower mantle and its consequences
on the heat budget of the Earth.
Session 3: Thermal History of the Geodynamo
Session chair: Bruce Buffett
Invited speakers: David Gubbins and Jeff Gee
Gubbins discussed the thermodynamics of the core. To maintain the adiabat, a heat flux of 4.5 TW is likely flowing out of the core. With an inner core heat capacity of 7.3 x 10^28 J, this suggest an IC age of 500 Ma. Given the existence of a geomagnetic field further back in time, this estimate for the age of the inner core seems small by nearly a factor ten. Resolutions to this include subadiabatic regions of the outer core, radioactive heating in the core, and compositional convection. The latter is likely to give a factor of two in the age of the inner core, perhaps still not enough. Gubbins' preferred resolution is radiactive heating in the core.
One problem with additional radioactivity is the additional heat this would have produced in the core in the past. The discussion that followed examined other possibilities: perhaps the density jump at the ICB is greater, perhaps the current heat flow is anomolous, perhaps the IC is not as old as the geomagnetic field, which may then have been different in the past.
Gee's talk concentrated on what precambrian paleomagnetic data can tell us about the dynamo, and what paleomagnetists would like to learn from dynamo theorists. The former includes the existence of a geomagnetic field, and reversals. The intensity and structure of the field are more difficult to ascertain. The latter includes the necessity of an IC for a dynamo, the appearance of a thermally driven dynamo, and what observations are helpful for evaluating numerical dynamo simulations. Gee hightlighted the difficulty of dating and assessing the stability of the magnetization, particularly in rocks older than 1 Ga.
Four posters (Stepanov & Starchenko; Buffett; Khristoforova; Labrosse,
et al.) discussed the thermal evolution of the core and the power requirements
of the geodynamo. Labrosse et al. used paleointensity measurements to estimate
the CMB heat flux before IC crystallization. This and Buffett's study suggested
some unreasonable temperatures before the IC began to solidify, unless
additional heat sources are present in the core. Four other posters (Tauxe,
et al.; Scott, et al.; Tarduno, et al.; Hulot & Gallet) made paleomagnetic
measurements or analyzed paleomagnetic data to study paleointensity and
the nature of the geomagnetic field during superchrons. Conclusions include
that paleosecular variation may be weak (Tauxe, et al.), and that the field
is primarily dipolar during superchrons (Tarduno, et al.). Moreover, superchrons
may appear with little warning, suggesting they occur as a result of the
non-linearity of the geodynamo rather than as a result in the change in
the core boundary conditions (Hulot & Gallet).
Session 4: Inner Core: Structure and Dynamics
Session chair: John Vidale
Invited speakers: Ken Creager and Shun-ichiro Karato
Creager reviewed the current state of seismic inferences on IC structure and super-rotation. The upper IC (250 +/- 100 km in the west, thicker in the east) is nearly isotropic. The west has an average P wavespeed that is .8 % less than the east. Q is 250 in the east, 600 in the west. The lower IC exhibits as great as 5 % anisotropy, with less east-west variation in the average wavespeed. There is also evidence in the form of scattering for heterogeneity at the km scale. The large velocity gradients in the IC require care in ray tracing. Some have suggested that apparent IC anisotropy could result from D", but Creager argued that this would require correlated 15 % velocity variations, which seems unlikely. Tromp questioned why normal modes don't see degree 1 variations as thick as 250 km, and wondered whether east-west variations could be an artifact of assuming spin axis symmetry. Masters pointed out that the mantle does not have a strong degree 1 signal, arguing against mantle control of the east-west variations.
Three types of data have been used to study IC super-rotation: PKP(DF)-PKP(BC) or PKP(AB), phase shifts of scattered waves recorded at the LASA array, and normal modes. All are consistent with prograde IC rotation of perhaps .1 - .3 deg/yr, though the normal modes do not require any rotation.
Karato discussed the origin of the IC anisotropy. Possible mechanisms inlcude a preferred orientation of the solid state and aligned partial melt. Karato argued against the latter due to the low porosity and because partial melting may not be necessary to explain IC attenuation. A preferred solid state orientation may arise during growth of the IC, though it is not clear how an isotropic layer at the top of the IC would then form. Moreover, deformation may modify the texture. Deformation has also been suggested to give the IC preferred orientation. Deformation texture requires dislocation rather than diffusion creep. This requires sufficiently large grain size and stress. Moreover, the solid state flow must have the right geometry to give the observed symmetry of the anisotropy. One such model involving zonal flow resulting from Lorentz forces also has difficulty explaining the isotropic layer. Thus, no single mechanism is currently entirely satisfactory for understanding IC anisotropy, in particular, the depth dependence and lateral variations in the anisotropy. Stevenson argued that anisotropy in thermal conductivity may play a role in inner core thermal structure.
Several posters addressed various aspects of inner core seismology. Koper, et al. used the IMS array to investigate CMB topography, the density jump at the ICB, and the sharpness of the ICB. Ishii & Dziewonski used a joint inversion of normal mode splitting and absolute and differential travel times to find a distinct innermost inner core about 300 km thick, which suggests a change in the core environment. Cormier & Li inverted PKIKP waveforms to find that scattering attenuation may be the predominant IC attenuation mechanism, with scatterers of size 9.8 +/- 2.4 km and velocity perturbations 8.4 +/- 1.8 %. Wen & Niu, Sun & Song, and Isse continue to find different IC seismic properties between the two hemispheres, and between the upper and lower inner core. The latter also found no evidence for super-rotation though cannot rule out a relative rotation of less than .2 deg/yr. Laske & Masters revisited super-rotation as detected by normal modes and found the rate to be mode dependent with a mean super-rotation of .13 +/- .11 deg/yr. Xu and Song estimated a .41 +/- .12 deg/yr super-rotation using the South Sandwich Islands-Beijing Sesimic Network path.
Bergman, et al. found that the preferred orientation of hcp Zn alloys
is affected by fluid flow during solidification, which could be responsible
for the complexity of IC anisotropy. Two posters (Rosat, et al. and Rogister)
studied the Slichter mode, from an observational and a theoretical standpoint.
Mound and Buffett developed a theoretical model for free osciallations
of the gravitationally and electromagnetically coupled core-mantle system,
predicting a surface gravity effect due to inner core rotation that may
be observable by the GRACE satellite.
Session 5: Outer Core: Structure and Dynamics
Session chair: Gauthier Hulot
Invited speakers: Richard Holme and Steve Lund
Holme reviewed the progress that has been made in using surface observations of the geomagnetic field to infer core dynamics. Magsat, in addition to providing high quality data, gave confidence in the methods used to downward continue the field to the CMB. These methods have since been used for historical, paleomagnetic, and archeomagnetic data. Holme then discussed the assumptions (frozen flux and some additional constraint to reduce non-uniqueness) that go into inverting for the core flows that cause secular variation. He stressed the implicit assumption that the flow is large scale. However, if one associates the surface flows with torsional oscillations, one can predict changes in the length of day that agree with what is observed, providing confidence in the inferred flows. Holme then discussed the various mechanisms-viscous, electromagnetic, topographic, gravitational- that could be responsible for decadal angular momentum exchange. The problem is difficult because determination of the torques often requires cancellation of large numbers. Current and future data will allow for resolution of core fields and flows at smaller length and time scales.
Lund discussed the issues concerning extracting paleomagnetic secular variation (PSV) (direction and intensity) from thermoremanence (TRM) and depositional remanence (DRM). Especially with DRM it is important to have data replication on a small spatial scale to evaluate systematic errors. PSV studies document that the field has been, on average, dipolar, but with deviations from a dipole field that existed for millions of years. The statistics of the field can be different between time scales of 10^6 versus 10^4 years, the latter perhaps more representative of the dynamo. Whereas the high-latitude flux lobes appear to be persistent features, there is no evidence that westward drift has been a common feature. PSV time series indicate some cyclic variability on a time scale of 500 - 3000 years. Excursions are associated with times of low magnetic field intensity.
Six posters concerened paleomagnetism. Korte and Constable produced a continuous global geomagnetic field model for the past 3000 years. McMillan, et al. used 'data' generated by dynamo calculations and stacking procedures employed in paleomagnetism to understand the errors associated with paleomagnetic studies. Herrero-Bervera and Valet studied PSV from lava sequences in Hawaii. Using archeomagnetic data Genevey, et al. studied secular variation in western Europe over the past 3000 years. Baker and Aldridge attempted to relate paleointensity time series (over 10,000 years) with changes in the core flow regime. Lund and Constable explored a model for PSV over the past 3000 years that considered whether the high latitude flux lobes might result from dynamo waves superimposed on the dipole field.
Two papers (Jackson and Finlay and Jackson) examined the geomagnetic field more recently. The former used satellite data and a maximum entropy method to produce maps of the field at the CMB. The latter performed spectral analysis on the historical geomagnetic data. A third poster on more recent geomagnetic fields, by Simonyan, modelled geomagnetic jerks. Botvinovsky and Starchenko studied electrical current sources in the core. Two more papers studied core surface flow as inverted from geomagnetic data. Whaler, et al. studied inversions undertaken with both one-norm and two-norm measures of misfit. Eymin and Hulot used the most recent satellite data to invert for surface flow, eximining the effects of truncation errors.
Vanyo and Asari, et al. studied coupling between the core and mantle, the former concluding that none of the proposed mechanisms can be discounted, but that magnetic coupling between the core and conducting lower mantle is likely to be important. The latter attempts to invert for core surface flow assuming topographic coupling and using changes in LOD as an additional constraint. Zatman and Bloxham presented work on torsional oscillations in the core, finding evidence in the time dependent core flow for short period (order 10 years), growing waves. Dumberry and Bloxham studied torsional oscialltions in numerical geodynamos.
Four more posters in this session also studied core dynamics. Noir, et al. performed an experimental study of non-linear interaction between retrograde precession and the tilt-over mode, while London did work on resistive instabilities. Takehiro and Lister considered zonal flows in a stably stratified layer beneath the CMB driven by underlying thermal convection, finding that the flows are affected by the convection even when the stratification is strong. Ivers studied analytically thermal instability in a rotating oblate spheroid.
Four other posters investigated various aspects of the core. Kotelnikova
and Starchenko looked at core cooling and inner core formation, and inner
core rotation and magnetic field generation. Van Hoolst, et al. investigated
the accuracy that free core nutation and free inner core nutation frequencies
and damping time scales can be determined from VLBI. Yu, et al. found east-west
variations in PKiKP-PKP residuals that could be explained by structure
at the base of the outer core, but they wouldn't commit to such structure.
Cardin, et al. studied turbulent viscosity experimentally. By studying
spin-up times they concluded that the concept of turbulent viscosity is
Session 6: Dynamos and the Deep Earth
Session chair: Phillipe Cardin
Invited speakers: Peter Olson and Daniel Lathrop
Olson described the inputs and outputs of numerical dynamos. The dimensional inputs and model parameters include geometry, rotation rate, energy sources, CMB and ICB boundary conditions, viscosity, conductivities, resolution, treatment of turbulence, and duration. Outputs inlcude the magnetic field spectrum, the secular variation and the time averaged behavior, and the velocity field. Current numerical models produce an Earth-like spectrum and secular variation, but with a time-average that is too dipolar. The velocity tends to the small scale when the CMB boundary conditions are homogeneous, but becomes larger scale when heterogeneous boundary conditions are imposed. Models can produce reversals and inner core rotation, though not always prograde. Other features of numerical dynamos include heterogeneous inner core growth, reversal frequency increasing with increasing Rayleigh number, and westward drift inside the tangent cylinder, but not outside. Magnetic flux patches are concentrated near downwellings, and are a feature of the secular variation, but do not appear to be primarily associated with dynamo action. To compare the output from numerical dynamos with data, one needs to view the output through the crustal filter (about l=12). Issues that need to be settled include how to deal with the cascade of energy to the small scale (hoping that the closure scheme will not affect what is observable), and the effects of using magnetic Prandtl numbers that are several orders of magnitude higher than that of the core. It is also time for systematic parameter searches.
Lathrop gave an overview of the current state of experimental dynamo research. Self-generation has been achieved by two different groups (Karlsruhe and Riga), but current dynamos are kinematic in the sense that the magnetic field has little effect on the carefully chosen flow. Experiments are in the range of magnetic Reynolds number order one (much lower than in the core or numerical simulations), magnetic Prandtl number 10^-5 (as in the core), and Reynolds numbers exceeding 10^5, large enough so that turbulence is prevalent. Much of the focus of current work is understanding the saturation mechanism and the role of turbulence. The Riga group finds that the saturated field is oscillatory. Perhaps related to this Lathrop's group finds that the magnetic field tends to suppress turbulence. Other groups are studying other mechanically forced flows and convectively driven flows.
In the discussion Cardin drew a triangle, one vertex representing magnetostrophy, one turbulence, and one viscosity. Theorists appraoch the dynamo problem from the first, experimentalists from the second, and numerical analysts from the third, with the actual core being somewhere in the middle.
Two posters presented their experiments concerning the core. Sumita and Olson performed experiments at high Rayleigh number, so that the flow was dominated by geostrophic turbulence. Cardin, et al. showed the design for their small-scale liquid sodium experiment, driven by differential inner core rotation. To go along with this experiment, Schaeffer and Cardin carried out a numerical study of the destabilization of Stewartson layers, finding agreement with the results of aympototic scalings.
Six posters concentrated on turbulence and small scale motion. Reshetnyak, et al. approached hydrodyanmic turbulence with a shell model, calculating radial dependent turbulent coefficients that can be used for the large scale solution. Buffett and Glatzameier approached the problem from a similarity standpoint, using the structure of the resolved large scale flow to estimate that of the unresolved small scale. Matsushima examined numerically the effects of magnetic fields and rotation on the anisotropy of turbulence. They found that intermittency can be important. Phillips and Ivers sought a better way to truncate numerical simulations, in particular, seeking not to artifically introduce anisotropy via hyperdiffusivity. Siso-Nidal and Davidson studied how inertial waves influence the evolution of small scale motion, and Loper, et al. looked to model small scale flow in an effort to be able to parameterize it for large scale simulations.
Much work was presented on different numerical dynamos. Matsui developed a dynamo code using finite elements, with the advantage of ease of parallelization. The difficulty comes from the treatment of the electrical boundary conditions, but results of this code are similar to those employing spectral methods. Rotvig and Jones looked at fully self-consistent, 3-D, low Ekman number dynamos, in a plane layer. No hyperdiffusivity was used, and they found a magnetostrophic balance with an Elsasser number order 10. The effects of Taylor-Proudman remain detectable, though the ageostrophic flow comprised about 80% of the flow. Kono and Roberts looked at low Rayleigh number, relatively high Ekman number, and less hyperdiffusivity, emphasizing the large scale flow and the drift (both east and westwards) of features. Kutzner and Christensen dispense with hyperdiffusivity for Rayleigh numbers up to 40 times critical, finding that as they increase the Rayleigh number the dipole becomes relatively less important than higher moments. By looking at both electrically conducting and insulating inner core boundary conditions, and running his code for up to 100 magnetic diffusion times, Wicht found that the inner core does not seem to stabilize against reversals, primarily because of few linking poloidal field lines. Cupal, et al. explored the possibility of control volume methods over grid methods, while Harder and Hansen explored finite volume methods, with the hope that they would be more efficient.
By looking at marine sediments over the past 2.25 million years, Yamazaki
and Oda found a 100,000 year periodicity in inclination and intensity,
suggesting a connection with the orbital eccentricity. Kuang and Chao argued
that mass redistribution due to core convection may be observable by modern
gravimetry. Livermore, et al. pointed out that the magnetic induction equation
can have non-orthogonal eigenfunctions, which may not be suitable for understanding
transient magnetic field bahevior of kinematic dynamos.
Session 7: Stealth Layers, D" and Core-Mantle Interactions
Session chair: Louise Kellogg
Invited speakers: Quentin Williams and Mike Kendall
Williams reviewed the observations concerning the lowermost mantle: a discontinuity 130-340 km above the CMB, negative velocity gradients in the lowermost 300 km, anisotropy, an ultralow velocity zone (ULVZ) in the lowest 40 km, 1000 km regions of mega-upwelling, and core-side low rigidity. The complexity in D" is comparable to that in the lithosphere. The origin for these phenomena may include some combination of a thermal boundary layer, partial melting, shear flow, core-mantle reactions, core sediments, subduction-related phenomena, primordial stratification, and changes in chemistry such as iron enrichment. Williams prefers a model where partial melt increases from .2 % above the ULVZ (to explain the negative velocity gradients) to perhaps 10% within the ULVZ. Within the ULVZ small changes in chemistry could then produce large changes in partial melt, and hence seismic properties. Motion of the partial melt could then enhance iron incorporation from reaction zones or core sediments. Unresolved geochemical questions include the composition of the partial melt, the presence of core sediments, which hinge on core topography, and the nature of CMB reaction products. Seismological questions include the nature of the ULVZ, in particular, whether it is global, and what is the density contrast.
Kendall discussed the evidence for D" anisotropy, which comes from shear wave splitting. The regions beneath the Caribbean, Alaska, Inidan Ocean, and Siberia are .5 - 3 % anisotropic, but the region beneath the central Pacific is more complicated, with smaller scale structure. Moreover, Kendall suggests that the anisotropy may be more complex than a simple transverse anisotropy. Lattice preferred orientation (LPO) can produce anisotropy, but we need better knowledge of the minerals present, their single crystal elastic constants, and the degree of alignment. Experimental and theoretical work is ongoing towards this. Shape preferred orientation (SPO) or layering can also produce anisotropy, which is being studied by effective medium modelling. Kendall suggests that the origin of anisotropy beneath the central Pacific, an area with an ULVZ and high partial melt, might be due to SPO, whereas other regions associated with paleo-slabs might have anisotropy due to LPO.
There was some discussion of core sediments. Issues concerning core sediments include their electrical conductivity, which will be shaped by the porosity, their connection with VGP's, the need for large scale regions of high conductance, and whether 1 km CMB topography could have the same effects. There was also some discussion on the continued lack of evidence for stealth layers.
Seven posters dealt with the seismology of the lower mantle. Ritsema, et al. found that PP/P, but not SS/S, amplitude ratios can be accounted for by crustal thickness variations between oceans and continents. Ni and Helmberger mapped out a 1200 km ridge-like structure beneath Africa, correlating with the geoid, suggesting a dynamic origin. Using stacking to eliminate source/receiver effects, Warren and Shearer investigated Q of the lower mantle, finding small scale correlation but not large scale correlations at different depths. To study the information contained in the coda that follows the main phases, Shearer and Earle attempted a Monte Carlo approach on seismic 'photons', finding that some lower mantle scattering is likely. Lay and Garnero investigated the relationships between differential travel time anomalies, shear wave splitting, and lower mantle triplications to better understand the structure beneath the Caribbean. Wen studied the seismic structure beneath south Africa, finding what looks like a compositional anomaly that correlates with geochemical anomalies. He argued this compositional variation was produced early in Earth's history. Reif, et al. looked at the resolution of seismic methods used for lower mantle tomography, and found that indiviudal anomalies should be larger than model errors.
Five posters presented the results of experiments bearing on the mantle.
Namiki described an experiment to model heat transfer in two-layer convection,
as a model for the D" and lower mantle. In general the D" would inhibit
heat transfer from the core, but this may not be the case if that layer
is much less viscous. Jellinek and Manga also
looked at a low viscosity lower layer, studying how the lateral variations that develop influence geometry and timescale of mantle plumes. They found that the low viscosity but dense layer can cause the plumes to become fixed in space, as is observed. Gonnermann, et al. also studied a stratfied system to understand the rate at which such a system becomes homogenized, finding that after scaling, even an initial 2 % density difference can cause distinct layers to persist for billions of years. Kavner and Walker studied means to achieve geochemical mixing at the CMB, along with laterally variable chemistry fluxes. They studied electrochemical reactions on iron/silicate/sulfide systems at 1350 deg C and 20 kbar with platinum/rhenium electrodes. The reactions at the anode and cathode differed. Lee, et al. looked at the density and bulk modulus of mineral assemblages at lower mantle temperatures and pressures, finding inconsistencies with those inferred seismically. This suggests a lower mantle that is enriched in iron, with some segregation between upper and lower mantle.
Seven posters used numerical simulations to study convection and heat
flow in the mantle. Hernlund and Tackley
put partial melt in the ULVZ at the base of the mantle, finding that regions of high melt exhibit small scale convection that may play a role in heat transfer across the CMB. Nakagawa and Tackley investigated compositional heterogeneity at the base of the mantle. Tackley studied the problem of why layering in the lower mantle is not more strongly seen, even if there is cancellation between thermal and compositional effects. The numerical simulations suggest that global layering is unlikely, but isolated piles of dense material are possible. Stegman, et al. also looked at mantle layering, also concluding that it is likely that the mantle cannot avoid evolving towards a uniform layer (in contrast to Gonnermann, et al.?). Steinberger and Holme sought an improved model for CMB topography, combining geodynamic and seismic constraints. Mantle flow models are parameterized by density heterogeneities and viscosity structure, and the misfit to geophysical data is minimized. Xie and Tackley studied mixing of geochemical reservoirs, finding that depth-dependent viscosity and plate tectonics reduces lateral mixing. Labrosse and Tackley examined heat transfer with plates. With larger plates, heat transfer may have been less efficient, keeping the Earth hotter longer.
Costin and Buffett considered the possibility that there could be a connection between lateral variations in electrical conductivity, perhaps due to CMB sediments in topographic highs, and preferred VGP paths. Prominent locations in the paleomagnetic database predict reversal paths that agree with the paleomagnetic observations. Kawasaki searched for silent earthquakes in D" using superconducting gravimeters. Silent earthquakes were inferred, but the data remains inconclusive as to the location.
[Report by Michael Bergman]