1. The Existence, Longevity and Composition of Mantle Plumes and Hotspot Volcanoes Mark Jellinek Dept. Earth and Ocean Sciences U. British Columbia Michael Manga Dept. Earth and Planetary Science U. California, Berkeley

2. Mars Venus Earth

3. Hotspots related to (deep mantle) plumes from CMB (e.g. Wilson, 1963; Morgan, 1971; 1981; Richards et al., 1989; Campbell and Griffiths, 1992; Clouard and Bonneville, 2001; Courtillot et al., 2003) HS island chain w/monotonic age progression Flood basalt at start (unless subducted) High melt production rates Large axisymmetric swell (strong B-flux) Significant -  V s in underlying mantle Large  T ; O(100+) viscosity variations Long term spatial stability High 3 He / 4 He Duncan and Richards, 1989

4. 1.Earth-like mantle plumes require large temperature and viscosity variations in TBL at CMB. 2.Large temperature and viscosity variations may require strong mantle cooling due to plate tectonics. 3.Sources for Pacific and African hotspots involve dense, low viscosity material that is composed of solid or partially-molten silicate and outer core material. 4.Interaction between convection due to core cooling and dense layer is required for long- lived spatially stable mantle plumes in the Earth, consistent with long-lived hotspots. 5.Earth is …. improbable? The story…

5. “Earth-like” Plumes vs. Thermals Plumes Large O(100) viscosity variations Head / Tail structure Tails persist >> 1 rise time. Thermals Small O(1) viscosity variations Discrete “heads” +/- transient tails Tails persist ≤ 1 rise time.

6. The simplest model of planetary mantle convection: Convection in a fluid with T-dependent viscosity under conditions of thermal equilibrium Heat In Heat Out Can Earth-like plumes occur?

7. Stagnant lid convection weak cooling = small viscosity variations in hot TBL Concepts: Flow at high-Ra has 3 layers: 2 Thin thermal boundary layers of unequal thickness; well mixed interior Cold TBL is a “rheological boundary layer” Stagnant lid part Active part Internal T > T average, close to T hot Small  T to hot boundary = small (order 1) viscosity variations in hot TBL. Earth-like plumes not possible. h  = 4 Simulations by A. Lenardic Ra = 10 6 = 10 6

8.  i ≈ constant in Stag-Lid limit h ≈ constant in Stag-Lid limit Isoviscous convection 1 / (1+ -1/6 ) Isoviscous convection Cold Boundary Hot Boundary

9. Role of subduction: stir in stagnant lid Strong cooling = large viscosity variations Ra = 10 6  = 10 6 h  = 4 h  = 10 3 2D Numerical Simulation: Stir in lithosphere, obtain large viscosity ratio required for plume formation. Subduction and Recycling of the lid

10. h  = 10 3 Role of subduction: stir in stagnant lid Strong cooling = large viscosity variations Ra = 10 6  = 10 6 Imposed stirring of stagnant lid into interior: Low viscosity upwellings with large heads and narrow tails Ra = 10 6,  = 10 6 Ra = 1.2 x 10 6,  ≈ 10 4 h  ≈ 10 2

11. Interactions between low viscosity plumes not consistent with long-term stability at high Ra. Large viscosity variations necessary but insufficient condition for longevity. Do large viscosity variations guarantee plume stability and hotspot longevity?

12. Seismic velocity at the base of the mantle along with (mostly) Pacific hotspots The base of the mantle is laterally heterogeneous. Hotspot positions correlate with low velocity material. (e.g. Williams et al., 1998) Low velocity regions shown are buoyant and likely deep mantle return flows (e.g. Forte and Mitrovica, 2001) V s model from Ritsema, 2004

13. The base of the mantle is laterally (chemically) heterogeneous Chemical heterogeneity in lower mantle: V s and V b anticorrelated Acute (i.e. non-diffusive) lateral and vertical seismic velocity gradients ULVZ (5 - 40 km thick): V s and V p reduced 5-10%, 10-30% -  V s /-  V p ≈ 3 to 3.5 / 1 Monotonic increase in Poisson ratio with depth African / Pacific hotspots. Not Iceland. ULVZ composed of dense material Joint analysis: normal mode and free air gravity constraints (Ishii and Tromp, 1999).

14. Constraints on ULVZ / Dense Layer properties: Plausibly a mixture of partial melt and OC material Seismolgy 6-30% partial melt within TBL (and / or) Metals from the outer core Geodetic studies Gravitational and electromagnetic coupling at CMB Length of Day (e.g. Holme; Zatman; Domberie) Gravitationally-forced nutations (e.g. Buffet) Metallic conductance in thin layer at CMB Geomagnetic / Paleomagnetic studies Observations of time-averaged radial field in Pacific: Link to thermal (electrical?) BC at CMB Behavior of non-dipole component of radial field during reversals Metallic conductance in thin ULVZ-like patches

15. Geodynamic studies: Mantle convection models (theoretical, exp., numerical): Subduction and mantle stirring, entrainment and longevity of layer, spatial stability of plumes etc. Dense (2-5%) low viscosity layer beneath deep-mantle upwellings : “Piles” beneath Africa and central Pacific Distribution governed by subduction zones Geochemical studies Silicate component of deep mantle plume source 3He / 4He in high-Mg OIB lavas? others …. ? Core component (e.g. Walker; Brandon; Humayun) Coupled Os isotopic excesses in high-Mg OIB Os systematics over large spatial scales Fe/Mn in MORB vs high-Mg OIB lavas Entrainment of ≤ 1% core material (implies a density increase of a few %)

16. Structure of time-averaged (non-dipole) radial field and core-mantle coupling Indicative of physical properties of ULVZ/dense layer? (1840-1980) Bloxham and Jackson, 1992 (0 - 5 Myr) Johnson et al., 2004 (0 - 3 kyr) Constable et al., 2000 3 Observations in Pacific matter: Complicated structure. Radial field varies with latitude and longitude. Structure persists over times >> core overturn Low radial field and low secular variation centered on HI. Hypothesis derived from simulation and theory: Spatial variations in thermal and/or electrical coupling at CMB…

17. Conductive patches and VGP paths during reversals (Costin and Buffett, 2003) Indicative of physical properties of ULVZ/dense layer?

18. Data from Sediment Cores VGP paths from observations VGP paths from Costin and Buffett Model* *Using same spatial sampling

19. What is ULVZ? Geochemical characteristics of plume source: A mix of LM and core material? I.Tracer for Silicate component: High 3 He / 4 He (“primitive, undegassed” ?) mantle Plume Buoyancy Flux MORB Most hotspots related to deep mantle plumes have elevated 3 He / 4 He relative to MORB.

20. Geochemical characteristics of plume source: ULVZ a mix of LM and core material? II. Core component: Siderophile elements Hawaii, Siberia, Galapagos, S. Africa, NOT Iceland Two Observations related to Re-Os systematics (Walker, Brandon and coauthors) Coupled 187 Os / 188 Os, 186 Os / 188 Os excesses in lavas associated with Hawaiian, Siberian and Galapagos plumes consistent with presence/ entrainment of 0.8-1.2% outer core material.

21. Geochemical characteristics of plume source: ULVZ a mix of LM and core material? II. Core component: Siderophile elements Hawaii, Siberia, Galapagos, S. Africa, NOT Iceland Two Observations related to Re-Os systematics (Walker, Brandon and coauthors) Coupled 187 Os / 188 Os, 186 Os / 188 Os excesses in lavas associated with Hawaiian, Siberian and Galapagos plumes consistent with presence/ entrainment of 0.8-1.2% outer core material. Intersection/convergence: One interpretation is that each linear array reflects mixing between two distinct Os isotopic components where a common radiogenic isotopic component is present in the sources of all of these materials. Identical systematics in Siberia, Hawaii and Gorgona (Galapagos origin?) require a spatially extensive reservoir consistent with a large, well-mixed outer core. Brandon et al., 2003 H,S G

22. Tracers for silicate/outer core mixture in source for hotspots overlying ULVZ? Modified from Brandon et al., 1999 MORB Plume DePaolo et al., 2002 MORB source Plume Source Linear mixing of outer core and LM silicate consistent with data from Hawaii. N.B.: No obvious correlation exists for Icelandic lavas. No evidence of core material identified (also no ULVZ sightings).

23. How does a dense, low viscosity layer influence convection from the hot boundary? Heat In Heat Out

24. Experimental Apparatus Dense layer experiments Two additional parameters: “Viscosity Ratio” Sabilizing compositional density difference Note: free-slip and no-slip bottom boundaries studied

25. Control Experiment:No Dense Layer Stagnant Lid Convection in the form of thermals Cold Boundary Hot Boundary Shadowgraph Image

27. Entrainment from a dense layer: Topography on the layer. Lateral variations in temperature and viscosity. “Free Slip”, Constant-T Lower BC “No Slip”, Constant-T Lower BC

28. Entrainment of dense, low viscosity fluid leads to formation of long-lived conduits “Free Slip”, Constant-T Lower BC “No Slip”, Constant-T Lower BC

29. Thermal Coupling: Initial decline in  following input of dense fluid  : Fewer new plumes form for the same heat flux.  governed by convection in dense layer Steady flow into conduits ultimately established (  = 0).

30. Theoretical Scaling Analyses Goals: - Condition for long-term stability of plumes. - Topography on dense layer. - Entrainment from dense layer. Applications to Earth (and other planets): -Long-lived mantle plumes? - Bumps on ULVZ material? New way to look for plumes seismically? -Understand composition of hotspot lavas in terms of mechanics governing formation of plumes?

31. Topography can stabilize plumes: Theory and Experiment Height of topography h/  µdµd µcµc L z x µ  h U µdµd µ h/  = C

32. How high is the topography? Theory and Experiment h u’ UdUd UdUd  µdµd µ L Height of topography h/  1/B 1/2

33. Tendril Thickness Theory and Experiment Tendril thickness Ra h µcµc µ µdµd    L U x z

34. Entrainment and Plume Spacing? Spacing between conduits approximately fixed Hypothesis: Spacing set by 1st R-T instability to TBL

35. Applications to Earth: 10 6 < Ra bot < 10 8 1 < B < 2 Longevity h/  > 0.7; topography comparable to TBL thickness Plumes expected to be stable for life of dense layer Topography on dense layer order 40 - 200 km; comparable to observed 5 - 40 km. Entrainment Low viscosity material enhances structure due to large  T. Influence composition.

36. Entrainment from dense layer and composition of source for volcanics: 3 He / 4 He: A thermophysical parameter? Plume Buoyancy Flux Good Medium Poor B-Flux Constraints MORB

37. Large Temperature differnces: Subduction and stirring of lithosphere Large viscosity variations: Earth-like plumes Subduction and stirring of lithosphere Entrainment from dense, low viscosity layer (ULVZ?). Long-lived plumes and hotspots Topography on dense layer comparable to TBL. Composition of hotspot lavas Entrainment from dense layer explains average composition of melt source.

38. Moving Forward: Effect of mantle stirring on longevity and composition of mantle plumes and hotspots? Outstanding questions How will large-scale mantle flow affect the dynamics of plume formation in the presence of a dense, low viscosity layer? Low viscosity outer core: Expect negligible shear stresses at CMB -- patches expected to be a slave to subduction. How will mantle shear influence the dynamics, rise and composition of plume conduits? Azimuthal stirring within the conduit important? Thermal entrainment? How will plate motions influence the spreading and composition of plume material ponded beneath the lithosphere? Farnetani et al., 2002 Kerr and Meriaux, 2005

39. Abouchami et al., Nature, 2005 Weis, unpublished

40. Internal chemical variation in plume conduits and hotspots (Kerr and Meriaux, 2005): What matters: Shear by mantle flow (cf. Richards and Griffiths, 1988; 1989): ratio of velocity of horizontal mantle motions to centerline plume rise velocity. Viscosity variations across plume conduit. Ra Q, Aspect ratio of conduit. Density and viscosity of tracer ??? Further implications of this work: Spreading of plume material beneath lithosphere Chemical variations within spreading plume material (e.g Farnetani and Samuel, 2004) (New) Dynamics of plume rise in the presence of both shear and a lithosphere: Implications for hotspot tracks predicted from global models and internal chem. variation: Thermal entrainment important Drag on the lithosphere important

41. Side ViewTop View Ra = 2.4E6, Viscosity Ratio = 56 Increasing Shear Velocity Ratio = 0.35 Velocity Ratio = 0.85 Velocity Ratio = 2.05 Some results (K&M, 2005):

42. Steinberger et al., 2004 Hotspot tracks and the dynamics of plume conduits in a Convecting Mantle … more to do on this problem Some Implications: Azimuthal and/or radial chemical variations among hotspot volcanoes: Relate length scale of variation to buoyancy flux Diagnostic of structure and composition of plume source.

43. Dense layer at CMB: Mixture of melt and core material? Constitution and transport properties Physical properties of melt phase (ab initio Stixrude, in progress) Distribution of melt across TBL Transport properties of outer core material in silicate melt vs. solid phases? Physical and electrical properties of mixture? Connectivity of core material in matrix? “Robust” geochemical tracers for core component? Physical processes within dense layer: Compaction? Internal Convection? Thermal, mechanical and electromagnetic coupling to core and mantle? Garnero

44. (0 - 5 Myr) Johnson et al., 2004 Is there a direct relationship between patches of dense layer and the spatial and temporal structure of the radial geomagnetic field observed in the central Pacific and Africa? Can the structure and secular variation of the time-averaged field constrain the geometry and physical properties of such patches as well as their influence on core cooling and the geodynamo? Geomag observations and geodynamic models

45. Concluding Remark Long-lived mantle plumes and hotspots are likely a direct consequence of the interactions between plate tectonics, core cooling and dense low viscosity material within D”