Messages From the Deep: Reviewing Seismic Evidence for Deep Mantle Slabs Thermochemical Piles Post-Perovskite Phase Transition Ultra-Low Velocity Zones and Uncertainties Deep = below lithosphere to Earth’s center Ed Garnero Arizona State Univ. Dept. of Geological Sciences June 21, 2006 5th Annual COMPRES Meeting

Multidisciplinary research conducted in collaboration with: Avants, Megan (UCSC) Ford, Sean (UCB) Hernlund, John (IPGP) Hutko, Alex (UCSC) Igel, Heiner (U Munich) Lay, Thorne (UCSC) Manga, Michael (UCB) McNamara, Allen (ASU) Rokosky, Juliana (UCSC) Rost, Sebastian (U Leeds) Schmerr, Nick (ASU) Thomas, Christine (U Liverpool) Thorne, Mike (U Alaska) Williams, Quentin (UCSC) Deep = below lithosphere to Earth’s center

-- Today -- Some Seismo Truths:  Important modeling uncertainties/trade-offs  Outlook: possible things to come Recent results and interpretations  Focus on deep mantle ‘high resolution’ work  Draw connections to global scales/processes inferred from long wavelength studies Deep = below lithosphere to Earth’s center

Why care? …We’d like to better understand: Isaacs, Oliver, Sykes [1969] Why care? …We’d like to better understand: Mode/style of mantle convection Depth extent, nature of subduction Source of hot spot magma Mantle H2O budget/cycle Transition zone structure, dynamics Nature, structure of fluid and solid cores Thermal evolution/budget of deep interior Today I’ll predominantly focus on the lower mantle, some TZ

Seismology Report Card Structural feature Evidence Constrained? Transition zone layering/topography Deep mantle heterogeneity Deep mantle “piles” D” Vs discontinuity/layering D” Vp discontinuity/layering D” anisotropy Ultra-low velocity zone CMB topography B+ C A+ B- C C- B A+ D- What does USArray allow us to do that was not previously possible? How do we best combine USArray body wave data from different time periods / geographies? Image short wavelength regional structure Draw connections to bigger picture B C- A C F

to Seismology Report Card ao Amplitude Time Five reasons for bad “Constraints” grades 1) Poor constraints regarding where on seismic raypath observables occur (travel time delays, extra arrivals, shear wave splitting, etc) ao to Amplitude Time inner What does USArray allow us to do that was not previously possible? How do we best combine USArray body wave data from different time periods / geographies? Image short wavelength regional structure Draw connections to bigger picture

to t’ Seismology Report Card a’ ao Amplitude Time Five reasons for bad “Constraints” grades 1) Poor constraints regarding where on seismic raypath observables occur (travel time delays, extra arrivals, shear wave splitting, etc) a’ ao Amplitude to t’ Time What does USArray allow us to do that was not previously possible? How do we best combine USArray body wave data from different time periods / geographies? Image short wavelength regional structure Draw connections to bigger picture inner

to t’ Seismology Report Card a’ ao Amplitude Time Five reasons for bad “Constraints” grades 1) Poor constraints regarding where on seismic raypath observables occur (travel time delays, extra arrivals, shear wave splitting, etc) a’ ao Amplitude to t’ Time What does USArray allow us to do that was not previously possible? How do we best combine USArray body wave data from different time periods / geographies? Image short wavelength regional structure Draw connections to bigger picture inner

Seismology Report Card Five reasons for bad “Constraints” grades 1) Poor constraints regarding where on seismic raypath observables occur (travel time delays, extra arrivals, shear wave splitting, etc) 2) Modeling trade-off between discontinuity location & isotropic heterogeneity Amplitude Time New arrival! inner What does USArray allow us to do that was not previously possible? How do we best combine USArray body wave data from different time periods / geographies? Image short wavelength regional structure Draw connections to bigger picture

Seismology Report Card Five reasons for bad “Constraints” grades 1) Poor constraints regarding where on seismic raypath observables occur (travel time delays, extra arrivals, shear wave splitting, etc) 2) Modeling trade-off between discontinuity location & isotropic heterogeneity 3) Globally, only very long wavelength structure is retrievable, which involves significant smearing, and may not accurately depict physics of interior What does USArray allow us to do that was not previously possible? How do we best combine USArray body wave data from different time periods / geographies? Image short wavelength regional structure Draw connections to bigger picture

Seismology Report Card Five reasons for bad “Constraints” grades 1) Poor constraints regarding where on seismic raypath observables occur (travel time delays, extra arrivals, shear wave splitting, etc) 2) Modeling trade-off between discontinuity location & isotropic heterogeneity 3) Globally, only very long wavelength structure is retrievable, which involves significant smearing, and may not accurately depict physics of interior 4) Only fraction of a percent of the globe has been probed at “high resolution”, and hence projection of results to global scales is conjecture What does USArray allow us to do that was not previously possible? How do we best combine USArray body wave data from different time periods / geographies? Image short wavelength regional structure Draw connections to bigger picture ~ 2502 km ~ 200 x 700 km ~ 1002 km

Seismology Report Card Five reasons for bad “Constraints” grades 1) Poor constraints regarding where on seismic raypath observables occur (travel time delays, extra arrivals, shear wave splitting, etc) 2) Modeling trade-off between discontinuity location & isotropic heterogeneity 3) Globally, only very long wavelength structure is retrievable, which involves significant smearing, and may not accurately depict physics of interior 4) Only fraction of a percent of the globe has been probed at “high resolution”, and hence projection of results to global scales is conjecture 5) Still using 1-D techniques to get 3-D answers…. What does USArray allow us to do that was not previously possible? How do we best combine USArray body wave data from different time periods / geographies? Image short wavelength regional structure Draw connections to bigger picture

A Mineral Physicist’s Guide to Disbelieving Seismologists Messages From the Deep: Reviewing Seismic Evidence for Deep Mantle Slabs, Thermochemical Piles, Post-Perovskite Phase Transition, Ultra-Low Velocity Zones A Mineral Physicist’s Guide to Disbelieving Seismologists Ignoring Embracing Ostracizing Canonizing Enslaving : Deep = below lithosphere to Earth’s center

Seismology Report Card: Grades Rapidly Improving ! Unparalleled seismic network seismometer populations (e.g., NSF-funded EarthScope’s USArray) Enables technique refinement/development Permits structural retrieval at smaller scale lengths Better computational capabilities 2- and 3-D wave propagation computations doable We are approaching capabilities of benchmarking solution structures What does USArray allow us to do that was not previously possible? How do we best combine USArray body wave data from different time periods / geographies? Image short wavelength regional structure Draw connections to bigger picture Advances in mineral physics, geodynamics, & geochemistry Provides significant guidance of our research targets and goals

Recent Seismic Results Some short seismic modeling vignettes relating to: Slabs in the lower mantle Post-perovskite phase transition Ultra-low velocity zone Deep mantle piles? What does USArray allow us to do that was not previously possible? How do we best combine USArray body wave data from different time periods / geographies? Image short wavelength regional structure Draw connections to bigger picture

Recent Seismic Results 200-800 km depth Upper mantle, transition zone structure in the central Pacific 2400-2800 km depth D” discontinuity topography 400-1000 km depth Slab detection from seismic reflections between What does USArray allow us to do that was not previously possible? How do we best combine USArray body wave data from different time periods / geographies? Image short wavelength regional structure Draw connections to bigger picture 2400-2800 km depth D” fine-scale layering 2800-2900 km depth Ultra-low velocity zone structure

Recent Seismic Results 200-800 km depth Upper mantle, transition zone structure in the central Pacific 2400-2800 km depth D” discontinuity topography 400-1000 km depth Slab detection from seismic reflections between What does USArray allow us to do that was not previously possible? How do we best combine USArray body wave data from different time periods / geographies? Image short wavelength regional structure Draw connections to bigger picture 2400-2800 km depth D” fine-scale layering 2800-2900 km depth Ultra-low velocity zone structure

Long wavelength suggestion: Some slabs continue to CMB. The ‘best’ evidence for old oceanic lithosphere sinking through the lower mantle is from seismic tomography for the region beneath the Caribbean. Here cross-sections are shown through the 2002 mantle shear wave model of Steve Grand. One can infer a contorted, thickened slab structure extending to D” then deflecting and piling up there, maybe….. dVs: Grand [2002] Hutko, Lay, Garnero, Revenaugh [Nature, 2006]

Fine-Scale D” Structure Beneath the Cocos Plate Thomas, Garnero, Lay [JGR, 2004]

A few % discontinuous dVs increase is consistent with the post-perovskite phase transition Observations: 200-300 km above CMB 0.5-3% velocity increase consistent with onset of post- perovskite phase e.g., Murakami et al. [Science, 2004] Lay et al. [EOS, 2005] Lay et al [PEPI, 2004]

Results: Vertical Step in D” Discontinuity: Fine-Scale D” Structure Beneath the Cocos Plate Results: Vertical Step in D” Discontinuity: Height of D” increases by 100 km over >200 km horizontally Hutko, Lay, Garnero, Revenaugh [Nature, 2006]

Fine-Scale D” Structure Beneath the Cocos Plate Hutko, Lay, Garnero, Revenaugh [Nature, 2006]

D” anisotropy Seismic anisotropy and shear wave splitting After Crampin [1981]

D” anisotropy After Crampin [1981]

D” anisotropy The apparent step in the D” layer coincides with a change in D” anisotropy parameters Hutko et al. [Nature, 2006, in press] S, Sdiff ScS Garnero, Maupin, Lay, Fouch [Science, 2004] Maupin,Garnero,Lay,Fouch [JGR, 2005] Rokosky, Lay, Garnero [EPSL, 2006, in press]

Recent Seismic Results 200-800 km depth Upper mantle, transition zone structure in the central Pacific 2400-2800 km depth D” discontinuity topography 400-1000 km depth Slab detection from seismic reflections between What does USArray allow us to do that was not previously possible? How do we best combine USArray body wave data from different time periods / geographies? Image short wavelength regional structure Draw connections to bigger picture 2400-2800 km depth D” fine-scale layering 2800-2900 km depth Ultra-low velocity zone structure

Raw array data: Fiji EQ recorded on the Canadian Yellowknife Array Rost,Garnero,Williams [in prep., 2006]

Each trace = stack at a different Incoming angle to the YKA array Array processed data: Each trace = stack at a different Incoming angle to the YKA array Rost,Garnero,Williams [in prep., 2006]

Back projecting along determined back azimuth and slowness of each precursor permits estimation of reflection location that matches differential time between precursor and the direct PP wave Rost,Garnero,Williams [in prep., 2006]

Rost,Garnero,Williams [in prep., 2006]

Recent Seismic Results 200-800 km depth Upper mantle, transition zone Structure in the central Pacific 2400-2800 km depth D” discontinuity topography 400-1000 km depth Slab detection from seismic reflections between What does USArray allow us to do that was not previously possible? How do we best combine USArray body wave data from different time periods / geographies? Image short wavelength regional structure Draw connections to bigger picture 2400-2800 km depth D” fine-scale layering 2800-2900 km depth Ultra-low velocity zone structure

Ultra-low velocity zones at Earth’s core mantle boundary This model type is appropriate for partial melt of the deepest mantle: dVs=3*dVp Williams and Garnero [Science, 1996]

Best attempt at global coverage ( ~40 % ) Ultra-low velocity zones at Earth’s core mantle boundary Not detected globally Isolated anomalous zones Best attempt at global coverage ( ~40 % ) Thickness depends of several assumptions Uncertainties in global ULVZ details quite large This model type is appropriate for partial melt of the deepest mantle: dVs=3*dVp ULVZ Thickness (km) Thorne and Garnero [JGR, 2004]

Ultra-low velocity zones at Earth’s core mantle boundary Best-fit model properties: Thickness : 8.5 (1) km DVP : -8 (2.5) % DVS : -25 (4) % Dr : +10 (5) % Issues: this is the only spot on Earth, at present, sampled with this degree of sampling density Rost,,Garnero,Williams,Manga [Nature, 2005]

Conceptual model possibility Ultra-low velocity zones at Earth’s core mantle boundary Conceptual model possibility 5 to 30 vol.% melt no spreading along CMB trapped intercumulus liquid incompatible-element enriched liquid crystals are initially overgrown and trap residual requires large overlying thermal anomaly downward percolation of melt correlation to dynamic instabilities/upwellings probably a fixed base for mantle upwellings Rost, Garnero, Williams, Manga [Nature, 2005]

Recent Seismic Results 200-800 km depth Upper mantle, transition zone structure in the central Pacific 2400-2800 km depth D” discontinuity topography 400-1000 km depth Slab detection from seismic reflections between What does USArray allow us to do that was not previously possible? How do we best combine USArray body wave data from different time periods / geographies? Image short wavelength regional structure Draw connections to bigger picture 2400-2800 km depth D” fine-scale layering 2800-2900 km depth Ultra-low velocity zone structure

Central Pacific D” Layering dVs [Grand, 2002] Lay, Hernlund, Garnero, Thorne [Science, 2006, in review] Avants, M., T. Lay, S. Russell, and E.J. Garnero [JGR, 2006] Avants, M., T. Lay, and E.J. Garnero [GRL, 2006]

Central Pacific D” Layering Lay, Hernlund, Garnero, Thorne [Science, 2006, in review]

Central Pacific D” Layering ~ 500 km Bin3 Bin2 Bin1 ~ 1000 km Lay, Hernlund, Garnero, Thorne [Science, 2006, in review]

Recent Seismic Results 200-800 km depth Upper mantle, transition zone structure in the central Pacific 2400-2800 km depth D” discontinuity topography 400-1000 km depth Slab detection from seismic reflections between What does USArray allow us to do that was not previously possible? How do we best combine USArray body wave data from different time periods / geographies? Image short wavelength regional structure Draw connections to bigger picture 2400-2800 km depth D” fine-scale layering 2800-2900 km depth Ultra-low velocity zone structure

Probing the Transition Zone Regionally: Central Pacific SS waves and precursors Schmerr and Garnero[2006, JGR, in press]

Probing the Transition Zone Regionally: Central Pacific Thinning of the TZ ( ~ 15 km average) 410 disc slightly depressed - 660 disc upwarped Schmerr and Garnero[2006, JGR, in press]

What about chemically distinct piles in the deep mantle? Step in D” discontinuity: consistent w/ slab piling/folding Slabs: at least to 1000 km Thinned transition zone What about chemically distinct piles in the deep mantle? This supports the geodynamic calculation: hot spots appear to overly edges of low velocities. Very localized ULVZ: dense, partial melt, base of upwelling D” stratification: LLSVP, pPv, ULVZ Thorne, Garnero, Grand [2004, PEPI]

Sharp “edges” to low velocities inferred from seismic waveforms mantle Caribbean anomaly core Sharp top African anomaly Pacific anomaly Ni and Helmberger [2003,2005, Science, EPSL] Ford, Garnero,McNamara [2006, JGR] Sharp “edges” to low velocities inferred from seismic waveforms South pole

Indirect evidence from global tomography red: lowest velocities for S20RTS green: strongest lateral VS gradients Thorne, Garnero, Grand, PEPI., 2004

Deep mantle shear velocity and hot spots This supports the geodynamic calculation: hot spots appear to overly edges of low velocities. Model: Grand 2002 Iso-velocity contour: - 0.7% Hotspots: Thorne, Garnero, Grand [2004, PEPI]

Summing up Topic More constrained Less constrained Deep mantle heterogeneity Long wavelength dVs patterns Short scale structure, Vp Deep mantle anisotropy It exists, it laterally varies Strength, depth distribution, geometry D” discontinuity dVs reflector strength, location Sharpness, height above CMB, dVp disc, deeper disc assoc. w/ pPv phase ULVZ Except for one spot: internal properties, geographical distribution, sharpness Deep mantle slabs Existence in upper half of lower mantle Structure/behavior in lower half of the mantle Deep mantle piles Abrupt transition between low velocities and mantle Density, internal structure What does USArray allow us to do that was not previously possible? How do we best combine USArray body wave data from different time periods / geographies? Image short wavelength regional structure Draw connections to bigger picture

“there are known unknowns Today: lots of seismically imaged short scale details in just a few spots of the volume of the interior Point: Where we’re afforded the ability to image in great detail, richness in complexity is apparent “there are known unknowns and unknown unknowns” What does USArray allow us to do that was not previously possible? How do we best combine USArray body wave data from different time periods / geographies? Image short wavelength regional structure Draw connections to bigger picture

garnero@asu.edu http://garnero.asu.edu Thus, do slabs make it all the way to the CMB? At least in this location, the evidence is pretty good. Note however, in other locations the evidence is not so compelling. This may be a unique location. garnero@asu.edu http://garnero.asu.edu Garnero [Ann. Rev. , 2000] Garnero, Maupin, Lay, Fouch [Science, 2004]