Process-oriented surface-wave tomography

1
Process-orientedsurface-wave tomography
Nikolai Shapiro
University of Colorado at Boulder
Collaborators
Michael Ritzwoller Peter Molnar Shijie
Zhong Jeroen van Junen

• Plan of presentation
• How surface-wave tomography works
• Deformation of Tibet
• Old continental lithosphere
• Cooling of the oceanic lithosphere
• High-resolution imaging from seismic noise

Jean-Claude Mareschal Claude Jaupart Anatoli
Levshin Mikhail Barmin Michel Campillo Laurent
Stehly
2
common belief
main point of this talk
Surface-wave tomography can be used to study
processes in the Earths crust and upper mantle

1. Improvement of resolution (short-period
   measurements)
2. Conversion into physical parameters
3. Combining with other types of geophysical
   information
4. Physically motivated parameterization of seismic
   models

3
Seismic data
How surface-wave tomography works
Body waves sample deep parts of the Earth Surface
waves sample the crust and upper mantle
4
Seismic surface-waves
How surface-wave tomography works

1. Two types Rayleigh and Love
2. Dispersion travel times depend on period of wave
3. Two types of travel time measurements phase and
   group

5
1. Dispersion measurements
How surface-wave tomography works

• More than 200,000 paths across the Globe
• Rayleigh and Love wave phase velocities (40-150
  s)
• (Harvard, Utrecht)
• Rayleigh and Love wave group velocities (16-200
  s)
• (CU-Boulder)

6
2. Dispersion maps
How surface-wave tomography works
100 s Rayleigh wave group velocity
7
3. Local dispersion curves
How surface-wave tomography works
All dispersion maps Rayleigh and Love wave group
and phase velocities at all periods
8
4. Inversion of dispersion curves
How surface-wave tomography works
All dispersion maps Rayleigh and Love wave group
and phase velocities at all periods
Monte-Carlo sampling of model space to find an
ensemble of acceptable models
9
3D seismic model
How surface-wave tomography works
combination of 1D seismic profiles from all
locations
horizontal slices at 150 km depth
vertical slices
10
regional studies
JGR, 2002
JGR, 2003
Geology, 2005
Nature, 2002
11
Andaman-Sumatra region
12
Quantitative interpretation of seismic
tomographic models

• S-wave speed - temperature (composition)
• anisotropy - strain

• Problems
• inversions are over-parameterized and non-unique
• (combining with other geophysical data
• physically motivated parameterization)
• limited resolution
• (measurements at short periods and along
  short paths)

13
Seismic anisotropy and deformation of Tibetan
crust
Discrepancy between inversely dispersed Rayleigh
waves and normally dispersed Love waves at
periods lt 30 s
14
Discrepancy between inversely dispersed Rayleigh
waves and normally dispersed Love waves at
periods lt 30 s
15
Explanation of short-period Rayleigh-Love
discrepancy radial anisotropy within mid-lower
crust
34N 84E

1. Anisotropy maximizes In the Western Tibet
2. lttSV-tSHgt 0.50.18 s

16
Radial anisotropy vertical slow axes of
symmetry- Likely mechanism of the radial
anisotropy near-horizontal alignment of mica
crystals

• Tectonic process that can cause the radial
  anisotropy
• Shearing during the underthrusting
• Crustal thinning and flow

17
Relation between crustal thinning and radial
anisotropy

• Mid-crustal deformation (30) is stronger than
  the surface deformation (10)
• Tibetan middle crust is mechanically weak

30 km layer with 30 of mica
18
Thermal models of the old continental lithosphere
from Jaupart and Mareschal (1999)
from Poupinet et al. (2003)

1. Constrained by thermal data heat flow, xenoliths
2. Derived from simple thermal equations
3. Lithosphere is defined as an outer conductive
   layer
4. Estimates of thermal lithospheric thickness are
   highly variable

19
Seismic models of the old continental lithosphere

1. Based on ad-hoc choice of reference 1D models and
   parameterization
2. Complex vertical profiles that do not agree with
   simple thermal models
3. Seismic lithospheric thickness is not uniquely
   defined

Additional physical constraints are required to
eliminate non-physical vertical oscillations in
seismic profiles and to improve estimates of
seismic velocities at each particular depth
20
Using heat-flow data to constrain seismic
speedsat the top of the mantle

• Computation of end-member crustal geotherms
• Conversion into seismic velocity bounds
• Extrapolation of temperature and seismic speed
  bounds over a large areas

21
Using heat-flow data to constrain seismic
speedsat the top of the mantle
seismically acceptable models
22
Using heat-flow data to constrain seismic
speedsat the top of the mantle
seismically acceptable models
23
Using heat-flow data to constrain seismic
speedsat the top of the mantle
seismically acceptable models
24
Thermal parameterization of the old continental
uppermost mantle
25
Lithospheric thickness and mantle heat flow

1. Seismic inversions can be reformulated in terms
   of an underlying thermal model.
2. Lithospheric thickness beneath cratonic cores
   exceeds 250km.
3. Mantle component of the heat flow beneath cratons
   is low ( lt 15 mW/m2).
4. Power-law relation between lithospheric thickness
   and mantle heat flow is consistent with the model
   of Jaupart et al. (1998) who postulated that the
   steady heat flux at the base of the lithosphere
   is supplied by small-scale convection.

26
Cooling of the Pacific lithosphere
Half-space cooling (HSC) model
T(z, A) Tm erfc(0.5z(kA)-0.5) A - lithospheric
age
A0.5
A-0.5
predictions of the standard model agree with
observations only at young ages
27
Possible explanations of the arrested cooling of
the Pacific lithosphere

• mantle plumes
• elevated mantle temperature in the past
• small-scale convection (SSC)

Numerical simulations of SSC by S. Zong, J. van
Hunen et J. Huang
details of the predicted structure depend on
mantle rheology
28
Dispersion maps
29
Dispersion maps
30
Inversion of dispersion curves

• non-physical oscillations in resulting thermal
  models
• physical interpretation problematic

Traditional approach ad-hoc parameterization of
seismic models
32N 160W

• thermally consistent modes
• inversion directly for physical parameters

32N 160W
31
3D seismic model
32
Apparent thermal age of the Pacific lithosphere
33
Small-scale convection can explain the observed
arrested cooling
34
Constraints on the upper mantle rheology
Ea activation energy
Laboratory measurements (Karato and Wu,
1993) Linear rheology (n1) Ea 240-300
KJ/mol Non-linear rheology (n3) Ea 430-540
KJ/mol
Best agreement between observations and
laboratory measurements is obtained with
non-linear rheology (dislocation creep)
35
Conclusions

• Seismic inversions can be reformulated in terms
  of an underlying thermal model.
• This allows us to infer simple thermal parameters
  that can be directly compared with predictions of
  geodynamical models.
• Comparison with numerical simulations show that
• - the structure of the old Pacific
  lithosphere is controlled by small-scale
  convection
• - deformation of the uppermost mantle is
  controlled by the dislocation creep (non-linear
  rheology)

36
Resolution of seismic tomographic models of the
crust and the uppermost mantle

• Seismic tomography is traditionally based on
  direct waves from earthquakes
• Distribution of earthquakes and seismic stations
  is inhomogeneous
• In many regions, surface-wave tomography is based
  on long pathes

• Problems with long paths
• Extended sensitivity kernels
• Difficult to make short-period measurements
• Consequence limited resolution

Solution measurements independent of
earthquake locations
37
(No Transcript)
38
(No Transcript)
39
(No Transcript)
40
(No Transcript)
41
By computing cross-correlation of seismic noise
records at two stationswe can extract
surface-waves propagating between these stations
42
Cross-correlation of seismic noise in California
43
(No Transcript)
44
Sierra Nevada
Sacramento basin
Franciscan formation
Peninsular Ranges
Salinean block
San Joaquin basin
45
Central Valley
Vantura basin
Imperial Valley
LA basin
46
Comparison between noise-based and
earthquake-based tomographies
47
Extraction of surface waves from seismic noise
Measurements without earthquakesImproved
resolutionPossible applications -
imaging of the crust and the uppermost mantle
- structure of sedimentary basins for seismic
hazard - seismic calibration for nuclear
monitoring Remaining questions - optimal
duration of noise sequences - spectral
range - optimal inter-station distances
- Other than Rayleigh waves (Love, body waves)
48
the end
49
(No Transcript)
50
Tibet cartes de dispersion longues-périodes
At short periods, group velocities are slow
because of the thick, slow crust At long periods,
group velocities are neutral to fast because the
crust is compensated by fast material in the
upper mantle
51
Tibetan Mantle Structure depth slices
52
(No Transcript)
53
(No Transcript)
54
(No Transcript)
55
(No Transcript)
56
Anisotropie azimutale de vitesse de groupe
Amplitudes larger in the young than the old
Pacific. Large amplitudes trend NW from the
EPR. Fast axes align perpendicular to the EPR
at all periods. 100 s 150 s maps similar
across the Pacific. 25 s 50 s maps mostly
similar. Contrast 50 s 150 s maps in the
old Pacific.
57
Comparaison de lanisotropie azimutale avec la
direction de mouvement des plaques
Present-Day Plate Motion
HS3-NUVEL-1A (Gripp Gordon, 2002)
58
Comparaison de lanisotropie azimutale avec la
direction de mouvement des plaques
Present-Day Plate Motion
HS3-NUVEL-1A (Gripp Gordon, 2002)
59
Comparaison de lanisotropie azimutale avec la
direction de mouvement des plaques
50 sec Avg difference 34 deg lt30 Ma -- 20
deg gt70 Ma -- trend from 25 - 60 deg
150 sec Avg difference 17 deg lt30 Ma -- 10
deg gt70 Ma -- 20 deg
60
Différence entre lanisotropie azimutale et le
mouvement des plaques
Near the EPR fast axes align nearly perpendicular
to PPM at all periods. At 70 My, agreement
begins to break down at short and
intermediate periods. Beyond 100 My,
disagreement is severe at periods below 80
s. Thus, in the shallow lithosphere (lt75 km) in
the old Pacific fast axes are not aligned with
present plate motions, but they are in the
underlying asthenosphere (gt125 km).
61
Comparaison de lanisotropie azimutale avec la
direction de paleo-extension aux dorsales
océaniques
Paleo-Spreading Directions
Determined from the gradient of lithospheric age.
Mueller et al., 1997
62
Comparaison de lanisotropie azimutale avec la
direction de paleo-extension aux dorsales
océaniques
Paleo-Spreading Directions
63
Comparaison de lanisotropie azimutale avec la
direction de paleo-extension aux dorsales
océaniques
50 sec Avg difference 17 deg lt30 Ma -- 15
deg gt70 Ma -- no trend
150 sec Avg difference 25 deg lt30 Ma --
20 deg gt70 Ma -- trend from 30 - 45 deg
64
Stratification de lanisotropie sous le Pacifique

• At large scales the story that emerges is a
  pretty simple one of
• anisotropy stratification
• In the deep lithosphere
• asthenosphere, anisotropic
• fast axes conform with current
• plate motions.
• In the shallow lithosphere
• (lt75 km), fast axes appear to
• be fossilized, aligning closer
• to the paleo-spreading directions.

65
Nouvelle méthode tomographie sismique de haute
résolution à partir de bruit sismique
66
(No Transcript)
67
Mechanisms to Produce Stratification of
Anisotropy Beneath the Pacific
68
(No Transcript)
69
Azimuthal Anisotropy Fast Axis Directions vs
Paleo-Spreading Directions
Paleo-Spreading Directions
70
Amplitude of Azimuthal Anisotropy as a Function
of Age