Seismic evidence for present-Seismic evidence for present-day plume upwelling at the day plume upwelling at the

core-mantle boundarycore-mantle boundary

Sebastian Rost Edward J. GarneroSebastian Rost Edward J. Garnero

Quentin WilliamsQuentin Williams

Michael MangaMichael Manga

University of CaliforniaUniversity of CaliforniaSanta CruzSanta Cruz

University of CaliforniaUniversity of CaliforniaBerkeleyBerkeley

ULVZ structure and detection

0.5 to 10’s km thick0.5 to 10’s km thick 10 to 30 % velocity decrease10 to 30 % velocity decrease density ?density ?

Thorne and Garnero, 2004

CMB are probed < 50 %CMB are probed < 50 % ULVZ evidence < 10%ULVZ evidence < 10%

(of CMB area)(of CMB area)

ScP waveform variationsScP waveform variationsScP waveform variations

Topography from NOAA 2’ datasetTopography from NOAA 2’ dataset

Using two small-scale arrays

- Tonga-Fiji seismicity- Tonga-Fiji seismicity

- deep earthquakesdeep earthquakes

- 97 earthquakes97 earthquakes

- Seismicity from:Seismicity from:10/1990 – 01/199810/1990 – 01/1998

WRA dataset

- 51 earthquakes- 51 earthquakes

- deep seismicitydeep seismicity

- Seismicity from:Seismicity from:11/1996 – 12/200011/1996 – 12/2000

ASAR dataset

WRA beam-trace profile

WRA beam-trace profile

All precursor eventsAll precursor events+ summation trace+ summation trace

Precursor summation tracePrecursor summation trace

Non-precursor summationNon-precursor summation

WRA double-beam

All precursor eventsAll precursor events+ summation trace+ summation trace

Precursor summation tracePrecursor summation trace

Non-precursor summationNon-precursor summation

WRA double-beam

ASAR beam-trace profile

WRA : 0.5Hz – 1.4HzWRA : 0.5Hz – 1.4Hz

ASAR: 1Hz – 3 HzASAR: 1Hz – 3 Hz

Higher ASAR resolutionHigher ASAR resolutiongives evidence for SdP gives evidence for SdP and perhaps SPcPand perhaps SPcP

ScP/P waveform comparison

ScP CMB sampling

Tomo from Ritsema and van Heijst, 2002

CRZ evidence from CRZ evidence from Rost & Revenaugh,Rost & Revenaugh,Science, 2001Science, 2001

ScP ULVZ evidence

- ~50 by 50 km- ~50 by 50 km

- northern boundary –24.5northern boundary –24.5

- southern boundary –25.5southern boundary –25.5

- some boundaries not well some boundaries not well

resolvedresolved

Forward modeling parameter space

1D Gaussian Beam Synthetics1D Gaussian Beam Synthetics

constant layer velocityconstant layer velocity

ScP, ScsP, SdP, SPcPScP, ScsP, SdP, SPcP

PREM backgroundPREM background

sharp upper boundarysharp upper boundary

4 parameter grid-search4 parameter grid-search

Forward modeling waveforms

Partial MeltPartial MeltChemical Chemical

HeterogeneityHeterogeneity

Best fit grid-search

Best-fit model properties: Best-fit model properties:

ThicknessThickness : 8.5 (: 8.5 (1) km1) km VVPP : -10 (: -10 (2.5) %2.5) % VVSS : -25 (: -25 (4) %4) % : +10 (: +10 (5) %5) %

VVPP// VVS S indicates partially molten materialindicates partially molten material

~50 by 50 km lateral extension~50 by 50 km lateral extension

small lateral extent raises stability questionssmall lateral extent raises stability questions

High-frequency data indicate very sharp upper boundaryHigh-frequency data indicate very sharp upper boundary

sharpness < 400 msharpness < 400 m

Data and modeling results

1D modeling restrictions

Experiment probes very slow mantleExperiment probes very slow mantle

(Ritsema and van Heijst, 2002)(Ritsema and van Heijst, 2002)

Region of strong lateral gradient Region of strong lateral gradient chemical heterogeneity chemical heterogeneity

(Thorne et al., 2004)(Thorne et al., 2004)

Probably dense material at CMB Probably dense material at CMB (McNamara and Zhong, 2004)(McNamara and Zhong, 2004)

Thorne et al., 2004Thorne et al., 2004

red: lowest velocities for S20RTSred: lowest velocities for S20RTS

green: strongest Vgreen: strongest VSS gradients gradients

Data and modeling results

5 to 30 vol.% melt5 to 30 vol.% melt

no spreading along CMBno spreading along CMB

trapped intercumulus trapped intercumulus

liquidliquid

incompatible-element incompatible-element

enriched liquidenriched liquid

crystals are initially over-crystals are initially over-

grown and trap residualgrown and trap residual

requires large overlying thermal anomalyrequires large overlying thermal anomaly

downward percolation of meltdownward percolation of melt

correlation to dynamic instabilities/upwellingscorrelation to dynamic instabilities/upwellings

probably a fixed base for mantle upwellingsprobably a fixed base for mantle upwellings

Preferred physical model

(from Jellinek and Manga, RoG, 2004)(from Jellinek and Manga, RoG, 2004)

Similar Tank experiment

D” aspect ratio of tank experiment !!

5 to 30 vol.% melt5 to 30 vol.% melt

no spreading along CMBno spreading along CMB

trapped intercumulus trapped intercumulus

liquidliquid

requires large overlying thermal anomalyrequires large overlying thermal anomaly

downward percolation of meltdownward percolation of melt

incompatible-element enriched liquidincompatible-element enriched liquid

correlation to dynamic instabilities/upwellingscorrelation to dynamic instabilities/upwellings

probably a fixed base for mantle upwellingsprobably a fixed base for mantle upwellings

Preferred physical model