Automated Multimode Inversion

1
Automated Multimode Inversion
of global tomography
and SEM benchmarking
Sergei Lebedev Rob van der Hilst
Dublin Institute for Advanced Studies Utrecht
University MIT
2
Seismic tomography
3
Seismic tomography
4
Seismic tomography
5
Waveform tomography

• Constrain Earth structure with full seismic
  waveforms,
• not one-per-seismogram observables
• Extract information from both the fundamental
  mode and S and multiple-S waves
• (interference ? individual arrivals
  often unidentifyable).
• Resolution in the upper mantle from near the
  surface down to 660-km depth.
• Nolet (1990) Partitioned Waveform Inversion
  (PWI)

6
Automated Multimode Inversion (AMI)
Lebedev, Nolet, Meier, van der Hilst (2005)
7
Automated Multimode Inversion (AMI)
Lebedev, Nolet, Meier, van der Hilst (2005)
8
Approximate sensitivity areas (kernels)on the
integration and model grids
Lebedev van der Hilst (2008)
9
Dense global coverage
10
3-D reference model
- for waveform inversions - for the 3-D
tomographic inversion
11
3-D shear-speed model

• Lateral resolution a few hundred km, varying
  with data sampling.
• Known seismic-velocity contrasts across
  tectonic boundaries resolved and located
  correctly (ARROWS), confirming accuracy of the
  imaging.

Lebedev van der Hilst (2008)
12
3-D shear-speed model

• 110 km depth. Most prominent high-velocity
  anomalies lithosphere of
• ancient cratons (stable continents).
• Largest low-velocity anomalies mid-ocean
  ridges and backarc basins,
• mantle regions with partially molten
  rock.

13
3-D shear-speed model

• Below 200 km amplitude of heterogeneity
  decreases.

14
Benchmark-of-the-benchmark SEM resolution test
Qin et al (2006), Lebedev van der Hilst
(2008)
15
SPICE-2 spectral-element test

• Below 200 km amplitude of heterogeneity
  decreases.

Qin, Capdeville, Montagner et al. (2008)
16
SPICE-2 the crust
17
SPICE-2 spectral-element test
Qin et al (2008)
18
SPICE-2 spectral-element test
19
SPICE-2 spectral-element test
20
Conclusions

• Automated Multimode Inversion (AMI)
• - uses the JWKB approximation largely
  within its limits of applicability
• - validated, confirmed to be implemented
  accurately
• JWKB approximation
• - neglects seismic-wave scattering and
  diffraction
• - global upper-mantle imaging at scales
  down to a few hundred km,
• depending on path coverage
• Benchmarking with SEM'' datasets
• - effective means of testing and validating
  tomographic methods