Significance of seismic tomography within the wider geoscience

communityComputational seismology works on improving imaging methods. Other geoscientists care mainly about the images themselves.

Hopefully, tomography results make sense in light of surface studies (geology, tectonics) and mantle convection simulations (geodynamics).

Solutions are non-unique -- but how bad is it? And how do I convey that to a non-specialist?

Example study: the mantle under North America -- window into a distant pastSubduction on the west coast: an oceanic plate gets submerged beneath the continent

A simple tectonic history(?)

Ren et al 2007, after Engebretson 1985

A single large plate has been subducting beneath the west coast for 180 million years. No significant interference from other plates.

Geologists puzzle about episodes of rapid change in the past

5-10 million years later: the sea is gone; 4-km-high mountains have risen.

75 million years ago: A shallow inland sea covers the Rocky Mountain area

Western U.S., 75 Myr ago 65 Myr ago

Geometry of the body-wave tomography experiment

Signal sources are P-waves generated by large but distant earthquakes

637 earthquake sources

1125 broadband receivers (seismometers)

Sensitivity kernels (shaded red) map the areas sampled by the body waves used.

Nissen-Meyer, 2007

How body waves sample the earth

time in sec

synthetic seismogram

North America

To solve the inverse problem, expand the sensitivity kernels on a global grid

Inverse problem (linear)

KT

reg

dT

reg

dln(vp )*=

data physical model (rows = kernels)

3-D earth structure (unknowns)

Result: a 3-D model of seismic P-wave velocities under North America

The subducted slab (blue=fast=cold) in the mantle down to 1800 km depth

B49N

42N

A

CD

PA

F

42NS2S2 F2F2

F1F1S1S1CC

dVp /Vp in %

Image of the subducted Farallon slab in the mantle

Seismically fast material is contoured (fast means cold).

Color signifies depth. We can confidently image ~1500 km deep.

Crust and lithosphere not rendered.d

e

p

t

h

/

k

m

Sigloch, McQuarrie, Nolet 2008, Nature Geo

The current subduction system

d

e

p

t

h

i

n

k

m

S1

S1N1

N2S2

S2

W

N1

SGMe

Birds-eye view from eastern Canada on the currently active western subduction system.

Notice the large vertical offsets of some adjacent fragments.

Resolution tests

Assume some hypothetical earth structure

Compute synthetic data from it

Invert the data. Is the input structure recovered?

Questions

New model is surprising but plausible (because it explains more geological observations that earlier models).

How different can other plausible models be that also fit the data? Possible to generate such models a priori?

Alternatively, at what level of confidence can we say that certain interesting features are real? (Example: tears in the plate, which are geodynamically important.)

Inverse problem

KT 0

KA KQ

regreg

dT

dA

reg

dln(vp )

dln(Q-1)*

=

data physical model (rows = kernels)

3-D earth structure (unknowns)

Significance of seismic tomography within the wider geoscience communityExample study: the mantle under North America -- window into a distant pastA simple tectonic history(?)Geologists puzzle about episodes of rapid change in the pastGeometry of the body-wave tomography experimentHow body waves sample the earthTo solve the inverse problem, expand the sensitivity kernels on a global gridInverse problem (linear)Result: a 3-D model of seismic P-wave velocities under North AmericaThe subducted slab (blue=fast=cold) in the mantle down to 1800 km depthImage of the subducted Farallon slab in the mantle The current subduction system Resolution testsQuestionsInverse problem