1. Probing the North American Continent Using Seismic Anisotropy
Huaiyu Yuan &
Barbara Romanowicz
Berkeley Seismological Laboratory
Jan. 26th , Shell Colloquium Series, Spring 2012, University of Oklahoma

2. Need Berkeley Logo to show different “wavelength” of the study so that the story here might not be sold

3. Mantle Nomenclature Shallow region B Transition region C
Don L. Anderson, J. Petrology 2011; mantleplumes.org
Shallow region B
Moho to 220 km (L-discontinuity)
Transition region C
L-discontinuity to 1000 km
Lower region D
1000km to core-mantle boundary
Classic mantle subdivision
Anderson, J. of Petrology 2011

4. Mantle Nomenclature Shallow region B Transition region C
Don L. Anderson, J. Petrology 2011; mantleplumes.org
Shallow region B
Moho to 220 km (L-discontinuity)
Transition region C
L-discontinuity to 1000 km
Lower region D
1000km to core-mantle boundary
Regional Inversions work on B’ to C’’
Classic mantle subdivision
Anderson, J. of Petrology 2011

5. Laminated Mantle Gutenberg’s Region B (Anderson 2011): Seismic Lid B’
G-discontinuity (velocity drop)
Low Velocity Layer (LVL) B’’
L-discontinuity (velocity jump)
“the most heterogeneous & anisotropic
region of the mantle” (Anderson 2011)
Cutting edge geodynamic model for the Earth upper mantle
Anderson, J. of Petrology 2011

6. Region B proposed in the Pacific
Promoted by the Receiver Function study in the NW pacific
(Kawakatsu et al. Science, 2009)
Predict: large Vs drop at G (100 km);
large Vs jump at L (200 km);
strong radial anisotropy ξ = (Vsh/Vsv)2 >1 in the low velocity zone
(Ekstrom and Dziewonski 1998; Nettles and Dziewonski 2008).
Gutenberg
Region B
Anderson uses the model to argue against the deep seated Hawaii plume
Lid moves due to the slab pull from subduction zones;
Shears the top of LLAMA to for aligned melts lenses (or melt rich lamellae)
Melt lamellae arrays can move horizontally thus when perturbed by fracture zones or lid thickness differences thus releases melts to the surface
Melt lamellae are low in velocity; and can add to the base of lid when it freezes (thermal addition of the lithosphere thickness)
Melt lamellae form strong radial anisotropy
LLAMA, “laminated lithology with aligned melt accumulations”
Anderson, J. of Petrology 2011;

7. Region B observed in the Pacific
Observation from the Berkeley global model:
Age progressive Seismic Lid
defined by G
Large radial anisotropy ξ
in LVZ (blue region)
Perturbed near Hawaii
Possible L (220-km)
Solves the long lived strong anisotropy at km depth in the ocean basins, illustrated by the figure
Based on Berkeley SEMum2.2 model, French et al AGU;
Strong ξ also observed by many others: Ekstrom and Dziewonski, 1998; Montagner 2002; Gung et al. 2003; Shapiro and Ritzwoller 2002; Smith et al 2004; Lekic et al., 2010; French et al. 2011

8. Region B proposed in the continents
G at 100 km depth: Anderson, J. Petrology, 2011
“The top of the LVL occurs between about 50 and 110 km depth beneath oceans and islands, and at depths of 100±20 km under continents (Thybo, 2006; Rychert & Shearer, 2009).”
L at 220 km depth: Anderson,
  “ …A thinner, deeper, and less anisotropic LVL terminates near 220 km under cratons. The 220 ± 20 km depth level (L) is a universal and fundamental tectonic boundary in the mantle.”
In the ocean basins, illustrated by the figure
Predicted strong anisotropy (radial anisotropy or

9. Region B not observed in the continents
Berkeley global model shows
High Vs extend
to > 200 km
Radial anisotropy ξ
peaks at shallow
depth
No Vs jump at
200km
Region B doesn’t work
In continents!
Obvious difference between the ocean basins and continents!
Modified from French et al., 2011 AGU

10. What happens at B’’ depth?
Where is G? Is there an LVL?
Where is the Lithosphere-asthenosphere-boundary (LAB)?
These questions fit nicely into our regional tomographic inversion.
Harvard Model
Berkeley Model
Romanowicz, Science 2009

11. North American Regional Inversion
Rich tectonic history
Whitmeyer & Karlstrom 2007

12. North American Regional Inversion
Rich tectonic history
USArray coverage

13. North American Regional Inversion
Rich tectonic history
USArray coverage
Highest station density
for a continent

14. Target: the NA upper mantle
What is continent lithosphere?
Is there a continent asthenosphere?
Where is the Lithosphere-Asthenosphere Boundary (LAB)?
Is there a G-discontinuity in the continent lithosphere?
What is the nature of the intra-lithosphere discontinuity?
Fundamental questions tie into the formation and evolution of the stable continent

15. Berkeley Regional Waveform Inversion
Global technique applied to regional case (since 2004)
Fundamental and overtone surface wave 3-component waveforms
Inverting for Isotropic Vs, radial (ξ) and azimuthal (Gc, Gs) anisotropy
Shear wave splitting station averages
Regional phases
Here is some technical background for our tomographic inversion.
The model is developed under the frame work of a normal model asymptotic formalism, NACT. NACT provides broadband 2D sensitive kernels for the body waves and surface waves, which represent better ray propagation in the mantle than the standard surface wave theory PAVA, path average approximation.
e.g. nact shows the finite width of the body wave phase SS sensitivity kernel around the ray and the spatial variation along the ray, while PAVA assumes averaged 1D sensitivity kernels.
French et al AGU

16. 3D Isotropic and radially & azimuthally anisotropic inversion
Isotropic Vs=[ (2VSh2+VSV2)/3]1/2
Radial anisotropy: ξ = (VSh/VSv)2
Azimuthal anisotropy (Gc, Gs):
VSV = VSV0[1+ Gc cos2ψG+ Gssin2ψG]
Strength (G) and fast symmetry axis direction (ψG ):
Montagner et al. 2000; Panning et al. 2006; Marone et al. 2006; Romanowicz & Yuan 2011

17. Non-linear Waveform Sensitivity Kernels
Normal mode Asymptotic Coupling Theory (NACT; Li & Romanowicz, 1995)
Non-linear waveform sensitivity kernels:
Finite-width of Fresnel zones; non-uniform sensitivity along the rays
Sensitivity down to the transition zone
Here is some technical background for our tomographic inversion.
The model is developed under the frame work of a normal model asymptotic formalism, NACT. NACT provides broadband 2D sensitive kernels for the body waves and surface waves, which represent better ray propagation in the mantle than the standard surface wave theory PAVA, path average approximation.
e.g. nact shows the finite width of the body wave phase SS sensitivity kernel around the ray and the spatial variation along the ray, while PAVA assumes averaged 1D sensitivity kernels.
Marone et al. 2007
French et al AGU

18. Berkeley Regional Models
Model 2007: Vs, ξ, and G, Marone & Romanowicz 2007
Pre-TA waveforms and SKS dataset, global events (to 2004) & stations
Model 2010: Vs, ξ, and G; Yuan and Romanowicz 2010
TA (to 2008) waveforms and SKS dataset, global events & stations
Model 2011/12: Vs, ξ; Yuan et al 2011 AGU
TA (to 2011) waveforms, regional/local events, regional stations
Published two generations of models; working on the third one;

19. NA Regional Model Setup
Model 2010: Including TA data; a lot better vertical and lateral resolution

20. Shear wave splitting data
Model 2007
Model 2010
SKS: sensitive to the same azimuthal anisotropy parameters (Gc and Gs)
Complementary sensitivity to surface waveforms due to “smearing” along the ray path
Shallow anisotropy constrained by surface waveforms.
And a lot more SKS measurements and the combined dataset improves deep model sensitivity .

21. Results in the cratonic upper mantle
Seismic observations for anisotropic layering in the craton
Rapid change of anisotropy with depth
Craton scale LAB
Revealed by surface wave anisotropy model
Layering in Craton Lithosphere: MLD
Associated with the chemical layer

22. 1D Profiles
Vs ξ G

23. Continental Isotropic Vs
Shallow depth: Fast Craton; slow WUS
300 km depth: asthenospheric instabilities; subducting JdF slab

24. Radial Anisotropy (ξ)
ξ = (VSh/VSV)2: sense of shear between horizontal and vertical
Horizontal shear (blue): oceans and frozen in the lithosphere
Vertical shear (red): subduction zone, sutures and upwellings
Craton and WUS separated by the Rocky Mountain Front (RMF)
Positive in ocean and continent down to > 250 km

25. Depth dependent azimuthal anisotropy (G)
Anisotropy strength G= sqrt(Gc2+Gs2) and directions as black sticks
Absolute plate motion (APM) direction in hotspot frame as arrows
Craton: weak G suture-zone parallel at 70km; north-south at 150 km
WUS: NA APM parallel at70 km; rotates toward Pacific APM
Positive in ocean and continent down to > 250 km

26. Shallow depth G: Suture-zone parallel direction
NE/SW structure trends for most sutures
NW/SE trends for Torngat & New Quebec sutures
West-East sub-province structural trends in Superior (Percival et al. 1994)
Trans-Hudson Orogen parallel
Positive in ocean and continent down to > 250 km

27. Deep depth G: Plate motion parallel
Craton: NA APM parallel at 250 km;
Hotspot frame (HS3 NUVEL 1A; red) better fits than no-net-rotation model (NNR-NUVEL 1A , Gripp & Gordon 2002; green)
WUS: Pacific APM parallel at 250 & 400 km;
Positive in ocean and continent down to > 250 km

28. Anisotropy layering in the craton
Rapid changes of fast axis direction with depth
Layered craton upper mantle
Deep apm parallel direction has two immediate implication to us: First application

29. LAB and MLD defined by anisotropy
Laying shown in both fast axis directions
and anisotropy strength
Lithosphere-Asthenosphere-Boundary (LAB)
APM parallel + anisotropy strength jump: plate shear!
Intra-lithospheric layer boundary: correlates with MLD (Abt et al. 2010; Fischer et al., 2010)
MLD
LAB
Anisotropy Strength
Change of velocity associated with the gradient change of anisotropy strength

30. LAB and MLD in the craton
First continent wide LAB map;
Craton LAB km; consistent with numerous local geochemistry, electric conductivity and other tomographic studies
MLD thickness km: layer 1 suture zone parallel direction (craton collision)
In the following I will focus more on the Layer 1 and 2 boundary and try to draw some conclusions from its correlation with other geodynamic observations.

31. Correlation with S Receiver Functions
Negative (red) phase matches with layer boundary
MLD
MLD

32. Correlation with Geochemical Layering
Layer 1- layer 2 boundary resides in Mg# contours from xenocryst samples (Griffin et al. 2004)
Layer 1: chemically depleted layer
Magnesium content from thermo-barometric analysis.
Indicate Layer 1 is old, chemically deplete stuff.

33. Correlation with geodynamic layering
Layer 1/layer 2 boundary matches with chemical/sub-thermal layers from geodynamic modeling (Cooper et al. 2004; King 2005; Lee 2006)
Layer 1: chemically depleted strong layer
Lithosphere thickens due to diffusive cooling

34. Correlation with results from other fields
8°-discontinuity from long range profiling (Thybo, Science, 1997)
Low velocity zone (LVZ) from tomography (Lekic et al. 2011; Chen et al. 2007)
High conductivity layer from MT (Jones et al., Geology, 2003)
Kimberlite layer (Sleep, G-cubed, 2009)

35. Correlation with results from other fields
Anisotropic receiver function layer (Bostock, JGR, 1998; Hale’s discontinuity, Levin & Park, Tectonphys. 2000)
Effective elastic layer (Audet & Burgmann Nature Geosciences, 2011)
Water solubility boundary (Green et al., Nature, 2010)
G-discontinuity (Anderson, J. Petrology, 2011)?

36. Correlation with water solubility boundary
Decrease of water solubility due to break down of Pargasite
Flat Clapeyron slope ( km)
LVZ at 100 km, both continents and oceans
Green et al. 2010

37. Continent Lithosphere
Continent lithosphere defined by azimuthal anisotropy: rapid direction change to the current APM.
Continent asthenosphere reflects the current plate shear
The LAB is around 200km, consistent with other estimates
MLD occurs at the proposed G-discontinuity depth, associated with Vs drop, but no L observed below
MLD: anisotropic, elastic, chemical, and/or water permeability
Fundamental questions tie into the formation and evolution of the stable continent

38. Layer Cake North American upper mantle
MLD
Plate Shear
To summarize
Craton LAB:
rapid change toward current APM; increased anisotropy strength underneath;
Craton MLD:
anisotropy domain boundary; bottom of chemical strong layer

39. WUS Anisotropy Fascinating region due to rich tectonic history and
abundant data from TA

40. Anisotropy and SKS in WUS
Circular pattern of the SKS splitting (Savage and Sheehan, 2000)
This 3D depth dependent anisotropy model can explain some of the hot topics in SKS measurements in the western US.
More than one-layer of the anisotropic domain needed.

41. Circular pattern of the SKS splitting
Schutt and Humphreys, Pure and Applied Geophysics, 1998
Savage and Sheehan, JGR, 2000

42. Circular pattern of the SKS splitting
NA-SWS-1.1, Liu, G-Cubed 2009
Eakin et al., EPSL, 2010

43. Proposed Interpretations from others
Savage and Sheehan (2000): Active upwelling
Zandt and Humphreys (2008): Passive edge/toroidal flow
West et al (2009): Lithospheric drip + associated mantle flow

44. Savage and Sheehan Models
Plume model
Model 3
Mantle Upwelling
Savage and Sheehan 2000

45. Zandt and Humphreys model
Slab Rollback + Slab window
Passive edge/toroidal Flow
Zandt and Humphreys, Geology, 2008

46. West et al. model Center of the circular pattern Low Volcanism
Low Heat flow
Large Vp variation
Small Splitting Time
They first observed a high velocity body from ~100 down to below the transition zone.
Downward Drip + mantle flow 
West et al., Nature Geo., 2009

47. Berkeley Model Circular Pattern Predicted by 3D model
Not good along the Wasatch front/western border of the CP where extremely large change of Velocity and fast axis directions over 50km distance have been reported from the LA Ristra array studies, which is beyond our current resolution.

48. Anisotropy and SKS in WUS
More than “one-layer” of anisotropic domain is needed.
Deep east-west direction beneath Oregon
More than one-layer of the anisotropic domain needed.

49. Deep anisotropy & Subducted slab
Fast Vs ( km) under Oregon:

50. Origin of the deep anisotropy
Isotropic Vs
So what causes the deep anisotropy?

51. Origin of the deep anisotropy
Anisotropy direction
Isotropic Vs
Fast velocity correlates with the east-west fast axis direction, indicating there is contribution from the slab!
There are a lot of slab segments in the WUS transition zone.

52. Origin of the deep anisotropy
Anisotropy direction
Isotropic Vs
Slab
Fast velocity correlates with the east-west fast axis direction, indicating there is contribution from the slab!
There are a lot of slab segments in the WUS transition zone.

53. Stagnant Slab Frozen-in/structural anisotropy
in the stagnant/flattened slab
Slab rollback caused stagnant segments above the 660-km and slab flattening above the transition zone.
Also the rollback may cause east-west flow around 150-km depth
Schmid et al., EPSL, 2002;
Fukao, Annu. Rev. Earth Planet. Sci. 2009

54. Depth dependent anisotropy in the WUS
Shallower than 150 km NE-SW plate shear
At 150 km circular flow due to slab rollback
At > 350 km east-west frozen-in/structural anisotropy in the stagnant/flattened slab
Slab Rollback
Plate shear
Pacific Plate
NA Plate
Plate shear
Slab rollback caused stagnant segments above the 660-km and slab flattening above the transition zone.
Also the rollback may cause east-west flow around 150-km depth
Stagnant
slab
Frozen-in fabric
660 km
Modified from Fukao, Annu. Rev. Earth Planet. Sci. 2009

55. Global Stagnant Slab in Transition Zone
SEMum2.2 Vs model (French et al 2011 AGU)
Expect to see

56. Future Directions
Higher frequencies: better vertical resolution in the lithosphere
Numerical approach: Spectral Element Method and Adjoint methods
Other regions: East Asia and Middle East
Global detection of the LAB and MLD (on going)
Plunging symmetry axis: geodynamic implications

57. Acknowledgements
David Abt, Heather Ford and Karen Fischer at Brown University
K. Liu, M. Fouch, R. Allen, D. Schutt, A. Frederiksen, A. Courtier, G. Soto, I. Bastow and V. Levin for sharing their results
S. Whitmeyer and K. Karlstrom for providing their Proterozoic Laurentia structural map
P. Cupillard, F. Marone, V Lekic, S. French, and D. Shan for modeling setup and figure preps.
EarthScope program, IRIS DMC, and the Geological Survey of Canada
NSF EAR Program