Upgrade to Pro — share decks privately, control downloads, hide ads and more …

New approaches in the kinematic imaging of global large earthquakes

0f828d1e52998fda294448484de2e409?s=47 @Résif
October 14, 2015

New approaches in the kinematic imaging of global large earthquakes

Présentation d'Ana MG Ferreira (University College London) aux 2èmes Rencontres Scientifiques et Techniques Résif | 12-14 octobre 2015, La Grande Motte

0f828d1e52998fda294448484de2e409?s=128

@Résif

October 14, 2015
Tweet

Transcript

  1. New approaches in the kinematic imaging of global large earthquakes

    Ana M G Ferreira1, Jennie Weston2, Jose-Angel Comino3, Gareth Funning4, Daniel Stich3, John Woodhouse5 1 University College London, UK 2 Arizona State University, USA 3 University of Granada, Spain 4 University of California Riverside, USA 5 University of Oxford, UK
  2. Dramatic progress in the past 30 yrs … massive new

    data sets … new theoretical developments … new computational developments Earthquake models help addressing fundamental and practical questions … how do faults slip? … what are the geometrical properties of faults? … how quickly can the size of an earthquake be estimated and reliable tsunami warnings issued? There are new ideas on earthquake source processes … new hypotheses to be tested Earthquake source models 1861 1930s Today www.globalcmt.org USGS geoscope.ipgp.fr/index.php/en
  3. Global earthquake source models 16 November 2000 New Ireland earthquake

    Mw: 7.6 Mo: 3.30x1020 Nm Depth: 13 km Strike: 306o Dip: 82o Rake: 79o USGS GCMT Mw: 8.0 Mo: 1.24x1021 Nm Depth: 24 km Strike: 328o Dip: 43o Rake: 3o
  4. Global earthquake source models Mw 7.1 Hector Mine earthquake 16

    October 1999 Weston et al., 2012
  5. Global earthquake source models !  Why such discrepancies? !  How

    can we robustly constrain source models? Mw: 7.6 Mo: 3.30x1020 Nm Depth: 13 km Strike: 306o Dip: 82o Rake: 79o Mw: 8.0 Mo: 1.24x1021 Nm Depth: 24 km Strike: 328o Dip: 43o Rake: 3o USGS GCMT
  6. Multidisciplinary data e.g., GPS, InSAR, strain, tilt, rotations, … Forward

    modelling e.g., spectral elements, ray theory, … Inverse modelling e.g., least squares, Monte Carlo, … Earthquake source imaging QUEST www.quest-itn.org
  7. Multidisciplinary data e.g., GPS, InSAR, strain, tilt, rotations, … Forward

    modelling e.g., spectral elements, ray theory, … Inverse modelling e.g., least squares, Monte Carlo, … Earthquake source imaging QUEST www.quest-itn.org
  8. Earthquake source models Data and approach Data: !  Long-period (T=135-150s)

    surface waveforms Modelling: !  Linearised (least-squares) inversion for point source parameters: lat, lon, depth, origin time, moment tensor fault strike, dip, rake and moment magnitude !  Use two different wave propagation formulations and four different 3-D Earth models 8 different solutions for each earthquake
  9. Ferreira & Woodhouse, GJI, 2006 M, nm Earthquake source models

    The 1998 Mw 6.6 Fandoqa, SE Iran earthquake Lat/Lon M0 (x1025dyne-cm) Strike/Dip/Rake m2 BB 1D - PREM 30.4/57.4 25.50 229/86/285 0.28 3D - S20RTS 30.1/57.5 8.60 156/68/191 0.17 3D - TW95 30.1/57.5 11.90 151/40/189 0.23 Berberian et al. 2001 30.1/57.6 9.09 156/54/195 ----
  10. Ferreira & Woodhouse, GJI, 2006 M, nm Earthquake source models

    The 1998 Mw 6.6 Fandoqa, SE Iran earthquake Lat/Lon M0 (x1025dyne-cm) Strike/Dip/Rake m2 BB 1D - PREM 30.4/57.4 25.50 229/86/285 0.28 3D - S20RTS 30.1/57.5 8.60 156/68/191 0.17 3D - TW95 30.1/57.5 11.90 151/40/189 0.23 Berberian et al. 2001 30.1/57.6 9.09 156/54/195 ---- 3D Earth structure helps constraining the seismic source
  11. Ferreira & Woodhouse, GJI, 2006 M, nm Earthquake source models

    The 1998 Mw 6.6 Fandoqa, SE Iran earthquake Lat/Lon M0 (x1025dyne-cm) Strike/Dip/Rake m2 BB 1D - PREM 30.4/57.4 25.50 229/86/285 0.28 3D - S20RTS 30.1/57.5 8.60 156/68/191 0.17 3D - TW95 30.1/57.5 11.90 151/40/189 0.23 Berberian et al. 2001 30.1/57.6 9.09 156/54/195 ---- Which 3-D Earth model(s) should we use? Significance of misfit differences? Need independent benchmark
  12. Earthquake source models Interferometric Synthetic Aperture Radar (InSAR) Mw 6.1,

    1993 Eureka Valley earthquake !  High spatial resolution and accuracy !  Alternative, independent data and technique to model earthquakes !  Over 100 earthquakes studied using InSAR since the 1992, Landers earthquake Weston, J., Ferreira, A.M.G. & Funning, G., JGR, 2011; Tectonophys., 2012 ICMT archive First earthquake catalogue independent of seismic data
  13. Earthquake source models Multi-data earthquake source analysis in a 3D

    Earth Weston, Ferreira, & Funning, GJI, 2014 Data !   SAR images are processed to produce unwrapped interferograms !   Three-component long period surface (T~150 s) and body (T~30 s) waveforms
  14. Earthquake source models Multi-data earthquake source analysis in a 3D

    Earth Data !   SAR images are processed to produce unwrapped interferograms !   Three-component long period surface (T~150 s) and body (T~30 s) waveforms Inversion scheme ! Fix the earthquake location to the InSAR estimate ! Model the InSAR data using a forward elastic dislocation model ! Model seismic data using the SEM for the 3D mantle model S20RTS + CRUST2.0 ! Use a Monte Carlo method to find the global minimum misfit Weston, Ferreira, & Funning, GJI, 2014
  15. Earthquake source models Multi-data earthquake source analysis in a 3D

    Earth Inversion scheme ! Fix the earthquake location to the InSAR estimate ! Model the InSAR data using a forward elastic dislocation model ! Model seismic data using the SEM for the 3D mantle model S20RTS + CRUST2.0 ! Use a Monte Carlo method to find the global minimum misfit Estimate moment, strike, dip, rake, average slip, fault length, width Data !   SAR images are processed to produce unwrapped interferograms !   Three-component long period surface (T~150 s) and body (T~30 s) waveforms
  16. Earthquake source models Multi-data earthquake source analysis in a 3D

    Earth 22 May 1998, Mw 6.6, Aiquile, Bolivia Weston, Ferreira, & Funning, GJI, 2014
  17. Joint earthquake source inversions 22 May 1998, Mw 6.6, Aiquile,

    Bolivia Data !  InSAR: 1 descending Interferogram (ERS-2); spans over 800 days. !  Seismic: Surface waves (34 stations) + body waves (19 stations)
  18. Joint earthquake source inversions 22 May 1998, Mw 6.6, Aiquile,

    Bolivia
  19. Joint earthquake source inversions 22 May 1998, Mw 6.6, Aiquile,

    Bolivia
  20. Joint earthquake source inversions 22 May 1998, Mw 6.6, Aiquile,

    Bolivia
  21. Joint earthquake source inversions 22 May 1998, Mw 6.6, Aiquile,

    Bolivia
  22. Joint earthquake source inversions 22 May 1998, Mw 6.6, Aiquile,

    Bolivia
  23. Weston, Ferreira, & Funning, GJI, 2014 Joint earthquake source inversions

    22 May 1998, Mw 6.6, Aiquile, Bolivia 358.0 179.0 15.0
  24. Discrepancies in seismic and InSAR solutions are reconciled in the

    joint inversions Joint earthquake source inversions 22 May 1998, Mw 6.6, Aiquile, Bolivia 358.0 179.0 15.0
  25. Extended source inversion through falsification •  Generate a representative set

    of slip distributions: 10,000 pseudo-random heterogeneous slipmaps assuming a von Karman autocorrelation function (Mai & Beroza, 2002) - rake angle, rupture velocity and slip velocity function fixed •  Compare the corresponding forward modelling predictions to real data (teleseismic P and S waves) •  Assemble all those trial models that have not been falsified on account of unacceptable fits. Assess uncertainties and non-uniqueness of the problem
  26. None
  27. Highly ambiguous problem! Comino, Stich, Ferreira & Morales, GJI, 2015

  28. Conclusions •  In order to build robust earthquake source models

    it is important to use the full spectra of seismic and geodetic (e.g., InSAR) data along with sophisticated forward and inverse modelling •  3-D Earth structure effects need to be properly taken into account, particularly when using surface wave and normal mode data •  Combining InSAR and seismic data helps resolving previous inconsistencies in source models (e.g., location) and better constraining source parameters (e.g., strike, dip and rake) •  Teleseismic variable slip inversions are highly ambiguous - inversions through falsification are useful to assess uncertainties and non-uniqueness of the problem (apply to other data types)