What Can We Learn From Images of Seismic-Wave Attenuation? Colleen Dalton and Göran Ekström

1. What Can We Learn From Images of Seismic-Wave Attenuation? Colleen Dalton and Göran Ekström Harvard University

2. Global S-Velocity Tomography • Good agreement on • large-scale features • magnitude of velocity variations • The challenge • better resolution of smaller-scale features

4. Attenuation: low attenuation high attenuation Global Shear-Attenuation Tomography • The challenges • consistent resolution of large- and intermediate-scale features • agreement on magnitude of Q-1 variations

5. Resovsky et al. 2005 Bulk Attenuation is very difficult to constrain Depth (km) Bulk Q

6. frequency, Hz Liu et al. 1976 Faul & Jackson 2005 Frequency Dependence • Seismology • Absorption band model • Constant Q-1 in absorption band • Experimental results • Q-1 ~ -, =0.2-0.4

7. Why Study Attenuation? 1. Velocity and Q-1 have different sensitivities

8. Why Study Attenuation? 1. Velocity and Q-1 have different sensitivities • 2. Physical dispersion of velocities

9. Why Study Attenuation? 1. Velocity and Q-1 have different sensitivities • 2. Physical dispersion of velocities • 3. Sensitivity of wave amplitude to velocity • valuable data set once attenuation effects removed

10. 75 seconds Data • 1. Surface-wave amplitudes • algorithm: Ekström et al. (1997) • 16,000 - 32,000 data at each period • 50-125 sec: R1 only • 150-200 sec: R1 & R2 • 225-250 sec: R1, R2, R3, R4 • 1993-2002; 347 earthquakes & 179 stations 2. Rayleigh wave phase anomalies • 22,000 - 90,000 at each period R1 R2 R3 R4

11. Data Amplitudes are sensitive to more than just attenuation 1. Source amplitude 2. Propagation effects: attenuation, focusing, scattering 3. Receiver amplitude We invert amplitudes simultaneously for: Coefficients of attenuation model Coefficients of velocity models (focusing effect) Amplitude correction factors for each earthquake Amplitude correction factors for each receiver

12. Rayleigh Wave Sensitivity Kernels Radial Basis Functions

13. low attenuation high attenuation 3-D Q Model

14. Vertical Average Profiles GTR1 oceans GTR1 continents

15. Source, receiver, & focusing corrections • Importance of corrections • 150-second Rayleigh waves • North America Source & receiver corrections No corrections low attenuation high attenuation Synthetic Q Map -if amplitudes were contaminated by focusing effects

16. Phase Velocity Maps from Amplitudes Only Ekström et al. 1997

17. Comparison with Previous Results

18. Comparison with Velocity Models

19. (Kustowski 2006) (Kustowski 2006) 100 km 150 km

20. 100 km GTR1 oceans GTR1 continents

21. 150 km GTR1 oceans GTR1 continents

22. Comparison with Laboratory Results • Laboratory measurements of Q and  • (Jackson et al., 2002; Gribb and Cooper, 1998) • Melt-free olivine • Periods: 1-100 sec • Temperatures:1000-1300oC • Grain size: 3-23 m • Faul & Jackson (2005) developed model to explain these results • Attenuation & modulus depend on • Frequency • Temperature • Pressure • Grain size • Fits the data; allows extrapolation Faul & Jackson 2005

23. Comparison with Laboratory Results • Comparison of their model to 100-Myr oceanic lithosphere • Appropriate frequencies for surface waves • Grain size: increases with depth Four profiles: • Tpot=1300oC, V*=1.2x10-5 m3/mol, grain-size increase at 140 km • Tpot=1300oC, V*=1.2x10-5 m3/mol, grain-size increase at 160 km • Tpot=1350oC, V*=1.6x10-5 m3/mol, grain-size increase at 140 km • Tpot=1400oC, V*=1.6x10-5 m3/mol, grain-size increase at 140 km • Solid-state mechanism can match low Q values observed

24. Lateral Variations in Velocity and Q-1 • From laboratory-based model • Depth = 150 km

25. Lateral Variations in Velocity and Q 150 km

26. Lateral Variations in Velocity and Q GTR1 oceans GTR1 continents

27. Calculated Q model (using Faul & Jackson 2005) Temperature model Q tomography Geodynamic model from Conder, Wiens & Morris 2002 Data from Roth et al 1999;reinverted by J. Conder Modeling Attenuation Structure - Tonga/Fiji(from Doug Wiens)

28. Effect on Velocities is Important Input: Q Models Output: Velocities

29. Conclusions • We inverted Rayleigh wave amplitudes for four quantities • Maps of attenuation • Maps of phase velocity • Source factors • Receiver factors • Accounting for extraneous effects on amplitude is essential • Focusing effects; source & receiver uncertainty • Results show: • Strong correlation between attenuation and velocity • Change in pattern of heterogeneity below 200 km • Results broadly consistent with experimental results • Joint interpretation of attenuation and velocity shows great potential

30. 150 seconds Global Correlation of Q-1 maps with Synthetic Q Map

31. Comparison with Other 2-D Studies 150 seconds

32. Receiver Factors • 75 seconds • 136 stations • 150 seconds • 157 stations BOSA-GT LBTB-GT BGCA-GT VNDA-GT LVZ-II KEG-MN BOSA-GT LBTB-GT LVZ-II BJT-IC QIZ-CD BJT-IC QIZ-CD BORG-II

33. scaling factor Blue - observed seismograms Red - synthetic seismograms Ekström, Dalton & Nettles 2005

34. Scaling Factors at LVZ-II, 1993-2004 scaling factor

35. (Ekström et al., 2005) (This Study) Comparison of Receiver Terms

36. Phase Velocity Maps

37. Power per Degree

38. 75 seconds Ekström et al. 1997 This Study Correlation = -0.77

39. Ekström et al. 1997 150 seconds Correlation = -0.75

40. Source Factors