squiggly-line-land view of the earth

Squiggly-line-land view of the Earth

• What’s going on in the upper mantle?– Receiver function, powerful seismic tool

• What in the world does the structure of the inner core mean?

• Is it still rotating, like it was in 1996?

Outline of mantle discussion• USArray

• Receiver function analysis

• MOMA

• Africa

• RISTRA

• The upper mantle discontinuities– Water at 410-km-depth– A double “520”

(I just got a digital camera)

EarthScope Components

• EarthScope's facilities include the following four coupled components:

– USArray (United States Seismic Array) – SAFOD (San Andreas Fault Observatory at

Depth) – PBO (Plate Boundary Observatory) – InSAR (Interferometric Synthetic Aperture

Radar)

USArray Permanent Array

Big Foot Array

Flexible Arraysexample from recent

experiments

Why look at the upper mantle?

• Mapping seismic structure– P & S velocity, density, anisotropy

• To deduce physical characteristics– Chemical and thermal heterogeneity

• To deduce what’s going on– Stagnant or moving continental keels– Dynamics of upper thermal

boundary layer of the mantle– Mantle circulation

Seismic-style study

• Reflection for crustal structure• S-wave splitting for anisotropy

– Flow direction - aesthenosphere– Relic fabric - lithosphere

• Surface and body wave tomography– Absolute velocities in upper few 100 km

• Body wave tomography (deeper)• Receiver functions

– Best resolution of radial velocity gradients

The receiver function

• Pioneered by seismologists including Bob Phinney and Chuck Langston

• Examines echoes of the P wave to determine zones of high radial gradient in seismic velocity

• It is proving to be a very useful companion to seismic tomography, providing detailed pictures of near-receiver structure

Chuck Langston, after igniting 50 pounds of explosives in sand

Chuck Ammon’s notes

40-80° distance range best

Ray paths contributingto receiver functions

Chuck Ammon

Radial componentof receiver functionJust useful for finding the Moho

Lat

eral

var

iati

ons

Adam

Alan

Mechanics of a receiver function

• Extract the P wave from the vertical component

• Deconvolve it from the horizontal component

• This should leave a spike at the P arrival time and a string of P-S conversions

• Convert the conversions (as a fcn of time and ground motion) to structure (impedance as a function of depth)

• Average together the records from many distances and azimuths

Some limitations

• Assumes no lateral variations in structure– Migration can overcome this limitation

• Only works in a frequency pass band– Cannot recover baseline, trends, or really much

beyond about 100-200 km wavelength velocity structure

– Generally falls apart shorter than 5-10 km wavelengths

Mike Wysession Keith Koper

• Missouri to Massachusetts transect

• 19 stations placed every 100 km

• Chosen for nice graphics

MOMA

MOMAdiscontinuity imaging

KarenFischer

MikeWysession

Receiver functionsfrom events to thenorth

Eventsto the south

Stereo vision

East!? West

Tomography plus receiver functions

Farallon depression?

T < 150° C

Disagreement with individual profiles

Steve Gao

Shows trend of smaller time separation with more vertical incidence

Gao, GRL, 2002

Again, well-resolved reflections from near 410 and 660

Note the presence of clear 410 conversions at short-period

Thicker transition zone to NETransition thickness near global average of 245 km, so not cold under region, 10 km of relief may correspond to ~60° temperature difference

cooler

warmer

Receiver function migration

• Just like migrating seismic reflection data

• Benefits from adequate spatial sampling

• Ability to image structure depends on– Depth of structure– Frequency of waves recorded

• Of course, more events with more back-azimuths, and more distances are helpful

Resolution with70 km spacing

T= 15s

Resolution with10 km spacing

T= 2s

Alan

A test model

Recovery of the test model

MOMA Array: Depth Migration LP10sMOMA migration

Cheyenne Belt Receiver Functions

CB

Moho Moho

SLAB

GFSS XD

ModifiedProterozoicMantle

ArcheanMantle

From Ken Dueker

Imbricated Moho

Fast from tomography

Mantle layered

RISTRARio Grande RiftRan from Texas into Utah

Receiver functions across the 1000-km line give a good picture of the shallow structure, and show little topography on the 410 and 660.

moho

Rick Aster

Flat discontinuities

Hot off the JGR “press”

• Hersh, Dueker, Sheehan, and Molnar, JGR• 410 and 660 topography under western US• 20-30 km topography, with 500 km scale length• No relation to surface tectonics• Sharpness not easily related to depth• Conclusions:

– Either transition zone has smaller scale convection than deep mantle

– Or there is a lot of compositional variation down there

Ken Dueker

Field areaAnne Sheehan

Average receiver function structure

Seymour Hersh

410

660

410 topography

660 topography

+/- 10 km

+/- 15 km

Science 6 June 2003

Seismic evidence for waterdeep in the Earth’s upper mantle

Mark van der Meijde

Suzanne van der Lee

Federica Marone

Domenico

Science 6 June 2003 - van der Meijde et al.

1000 ppm water broadening the 410-km-discontinuity?

Main points of van der Meijde • Conversion from “410” stronger at low

frequency than high, but conversion from “660” is steady

• So “410” must be broader, in fact very broad, 20-40 km wide

• Subduction has been pervasive, so water might be common near 410-km-depth

• Entire story is consistent if about 1000 ppm water is present.

~1 s period 6 s period

9 stations

The general trend is consistent, and statistics can be constructed to support the significance of the trend.

X

Earthquake

Station P'P'df P'P'ab

Mantle

OuterCore

InnerCore

Figure 1

The phase P’P’

Jim Whitomb

DLA

JGR, Fei, Vidale and Earle

• Rounded 3 good datasets of P’P’– California networks

– LASA recordings

– Highly selected GSN seismograms

• We’ll see– Sharp 660-km-depth discontinuity

– Somewhat less sharp 410, sometimes

– (but MUCH sharper 410 than claimed for Europe)

– No 520

0

0.2

0.4

0.6

0.8

1

-200 -150 -100 -50 0 50 100

Envelope stack:1/19/69 earthquake at LASA

Time relative to P'P'(ab) (sec)

P'P' onset

P'660P' P'410P'

Several minute envelope stack

0.00

0.05

0.10

0.15

-200 -150 -100 -50

P'P' precursory interval

Amplitude relative to P'P'

Time relative to P'P' (sec)

P'660P' P'410P'Raw stack

Noise-corrected

Figure 4b

The 660 and 410 corrected for steady noise

0

0.2

0.4

0.6

0.8

1

-200 -150 -100 -50 0 50 100

Stack of best 91 GSN traces

Raw stack

Noise-corrected

Time relative to P'P' (sec)

P'P' onset

P'660P'onset P'410P'

onset?

Figure 8

A global average

-200 -150 -100 -50 0 50

Summary of envelope stacks

Time relative to P'P' (sec)

P'P'

P'660P' P'410P'CSN

LASA

GSN

LASA + CSN

Figure 9

More 660than 410 energy,Nothing else

Fei Xu

0

0.01

0.02

0.03

0.04

0.05

-200 -150 -100 -50

Precursors to P'P'

Time relative to P'P' (sec)

P'660P'P'410P'

XXlong-period

reflectionamplitudes

Comparison to long-period reflections

Corrected for attenuation

0.00

0.02

0.04

0.06

0.08

0.10

2200 2240 2280 2320

LASA stacks at two frequencies

0.7 Hz stack1.0 Hz stack1.3 Hz stack

Time Figure 11

"660"

"410"

No visible 410 at higher frequencies

This means• 660 sharp enough to efficiently

reflect 1 Hz waves - less than 2 km thick transition

• 410 not so sharp - our data is fit by half a sharp jump, half spread over 7 km

SS precursors as a probe of layering near their bounce point

Peter Shearer

(Also has claims to see PKJKPand a “250”)

ArwenDeuss

Science, 2001. Sees 520 sometime simple, sometimes split.

Interprets this as the 520 having phase changes in two components, olivine and garnet, whose depths don’t always coincide.

Transects that indicate lateral continuity of structure

Transectsof the 520

Lateral continuity of

structure

A global map, where there is

coverage

JohnWoodhouse

Some high points• “410”

– Why is it’s brightness variable?– Can we map the pattern globally to learn more?– Is topography real?

• “520”– Why does it flicker?

• “660”– Is topography a thermometer?

• Other discontinuities?• Better images on the way from USArray

The enigmatic inner core• Layering

• Anisotropy

• Rotation

• Possible origins of structure

• Combined my slides with those of Ken Creager and Shun Karato

Some slides lent byKen Creager and Shun Karato

Seismic characteristics of the inner core

• A large Poisson’s ratio, close to that of a liquid

• High attenuation (Qs~100-200)

• Strong anisotropy

Anisotropic Lower inner Core

Transition R

egion

Isotropic Upper Inner C

ore

A current working model

IMIC

Upper Inner Core:

Isotropic, finely heterogeneous

West: 0.8% slower

250 km thickQ = 600

East: thickerQ = 250 in east

Middle Inner Core:

Strong anisotropy

Isotropic Voigt average is homogeneous

Innermost Core:

Different anisotropy?

Niu and Wen, 2001Red - western hemisphereBlack - eastern hemisphere

Comparing polar and equatorial data

Ouzounis and Creager, GRL, 2001

Beghein and TrampertScience, 2003 Adam and

Miaki Ishii

Summed slant stackSlowness (s/km)

predicted for PKiKP

1000 1050 1100 1150 1200 1250

-0.10

-0.05

0.00

0.05

0.10

Time after event (s)

Slowness (s/km)

0.00 0.25 0.50 0.75 1.00

Amplitude

Stack of envelopes of slant stacks13 earthquakes and 4 nuclear tests

X

direct P coda slowness

(Vidale & Earle)

Proposed mechanisms of inner core anisotropy

Jeanloz & Wenk, GRL, 1988

Convective flow due to high Rayleigh number aligns crystals (most effective near surface)

Yoshida et al., JGR, 1996

Inhomogeneous growth of inner core drives convective flow that restores isostatic equilibrium

Michael Bergman, Science, 1997 (modified by Michael Wysession)

Dendritic growth of crystals aligns a-axes radially with heat flow direction (assumes c-axis is fast)

Modified from Annie Souriau, Science, 1998

Strong heterogeneities, various crystal alignment orientations

Bruce Buffett, Nature, 2001

Rotationally wrapped magnetic field around inner core causes Maxwell stresses that align crystals (c-axes cylindrically radially out)

Modified from Shun-Ichiro Karato, Nature, 1999

Lorentz forces produce axisymmetric, sustained flow that aligns crystals

Hemispherical asymmetrySumita and Olson (1999)

Hemispherical asymmetry might be due to heterogeneous thermal boundary conditions at the inner-core boundary caused by core-mantle interaction.

[Time-scale for anisotropic structure formation must be comparable to or shorter than the time scale for changes in mantle structure.]

Does the inner core rotate with respect to the mantle?

Song and Richards, 1996 yes 1.1 deg/yr

Creager, 1997, yes 0.2-0.3 deg/yr

Vidale et al., 2000, yes 0.15 deg/yr

Souriau, 2001, no, at least not very fast, <0.1 - 0.2 deg/yr

Laske and Masters , 2002, maybe 0.13±0.11 deg/yr

Song, 2002, yes 0.5-1.0 deg/yr

Why do we care?

• I think it’s interesting

• Would mean the core has either– Quite low viscosity

• Can deform fast enough to keep moving

– Quite low viscosity• Deforms so little that there is little viscous drag

• Would prevent association of IC structure with mantle structure

25 years of data

Xiao-Dong Song and Paul Richards

Li and Richards, submittedSouth Sandwich Islands Doublet

Song, AGU Monograph, 2002

More Sandwich doublets

Laske and MastersNormal mode analysis

AGU Monograph, 2002

Geometry

Explosions LASA array

ICS

View from Equator

PKKP comparison

-6

-4

-2

0

2

4

6

1870 1875 1880 1885 1890 1895 1900

Stacked PKKP waveforms

9/27/718/29/74

Time after blast (s)

PKiKP waveform correlation

-6

-4

-2

0

2

4

6

1070 1075 1080 1085 1090 1095 1100

Stacked PKiKP coda waveforms

9/27/718/29/74

Time after blast (s)

P’660P’ correlation

-3

-2

-1

0

1

2

3

2220 2225 2230 2235 2240 2245 2250

Stacked P'660P' waveforms

9/27/718/29/74

Time after blast (s)

Bottom line:Inner core maylap Earth every

2000 years

Wild card - Does the outer corechange over time?

Quick Review• Mantle discontinuities still

remain interesting after 40 years

• Inner core is being mapped but not yet understood

• Inner core is likely turning slowly

• Seismology and mineral physics must progress together