1. 1 Ivan Wong Principal Seismologist/Vice President Seismic Hazards Group, URS Corporation Oakland, CA Uncertainties in Characterizing the Cascadia Subduction Zone and Their Seismic Hazard Implications Cascadia Subduction Zone Earthquakes and Critical Infrastructure Oregon State University, Corvallis, OR 18 July 2012

2. 2  The four most significant seismic source characteristics of the Cascadia subduction zone (CSZ) relevant to hazard in the Pacific Northwest are: 1) segmentation, 2) maximum magnitudes, 3) time-dependent and time-independent recurrence intervals, and 4) the eastern edge of the megathrust rupture zone.  Studies in the past 25 years since the CSZ was recognized as being capable of producing large earthquakes (M ≥ 8) have focused on these issues.  We have evaluated the impacts of these parametric uncertainties on seismic hazard in the Pacific Northwest using a seismic source model developed in a 4-year SSHAC Level 3 PSHA. Introduction

3. 3  Our seismic source model of the CSZ megathrust was developed as part of a regional seismic hazard analysis of 42 dams in British Columbia sponsored by BCHydro.  The CSZ model was developed by a TI team of 8 with input from 12 subject matter experts.  We have investigated the sensitivities to probabilistic hazard at three cities in the Pacific Northwest by calculating the hazard along branches of our logic tree.  The choice of recurrence models (e.g., characteristic, maximum magnitude and truncated exponential) is typically critical to PSHA but we assume that the CSZ can produce the full range of earthquakes and so we model it as such using the paleoseismic/historical seismicity record. Introduction (cont.)

4. 4 BCHydro Redwood (Logic) Tree

5. 5 Cascadia Subduction Zone GSC-McCrory Model

6. 6 Wang et al. (2003) Model of CSZ

7. 7  We have adopted the eastern rupture extent from the model of Wang and Hyndman (2011). We have assumed that M 9 rupture extends down to (1) the base of the coseismic transition zone (CTZ), (2) 30 km downdip, and (3) 10 km updip of the base of the CTZ.  We consider three possible modes of rupture for the megathrust: (1) full rupture events around M 9; (2) intermediate-sized earthquakes of M 8 to 8.8, which rupture one of two possible segments in the southern half of the subduction zone; and (3) smaller earthquakes (M < 8).  Segmentation of the Cascadia subduction zone is based on the model of Goldfinger et al. (2012). Paleoseismic evidence supports full rupture and intermediate magnitude earthquakes. Uncertainties

8. 8 Cascadia Subduction Zone GSC-McCrory Model

9. 9 Space Time Diagram for the Cascadia Margin from Goldfinger et al. (2012) Source: Goldfinger et al. (2012)

10. 10  Time-independent and time-dependent recurrence intervals have been estimated for full rupture events based on the Holocene turbidite history of Goldfinger et al. (2012).  Statistical analysis of the turbidite record indicates that full- rupture events are likely clustered in time. In the Holocene there have been four clusters consisting of 5 earthquakes.  Temporal clustering of the M 9 earthquakes was addressed by including time-dependent intercluster and intracluster recurrence intervals. Uncertainties (cont.)

11. 11 Turbidite Event Chronology for Full Rupture Events from Goldfinger et al. (2012) Source: Goldfinger et al. (2012)

12. 12 BCHydro Logic Tree for Recurrence

13. 13 Seismic Source Contributions to Mean PGA Hazard Eastern Edge SeattlePortlandEureka

14. 14 Seismic Source Contributions to Mean PGA Hazard Time Dependence SeattlePortlandEureka

15. 15 Seismic Source Contributions to Mean PGA Hazard Clustered/Non-Clustered SeattlePortlandEureka

16. 16 Seismic Source Contributions to Mean PGA Hazard Clustered/Gap SeattlePortlandEureka

17. 17 PGA Sensitivities at 2% in 50-Year Exceedance Probabilities UncertaintySeattlePortlandEureka Eastern Edge Shallow Preferred Deep 0.184 0.201 0.223 0.223 0.238 0.262 0.847 0.837 0.800 Time Dependence Independent Dependent 0.214 0.242 0.224 0.252 0.468 0.552 Clustered/Non-Clustered Independent Clustered Non-clustered 0.214 0.272 0.166 0.224 0.280 0.182 0.468 0.634 0.348 Clustered/Gap Independent Dependent – In a gap Dependent – In a cluster Dependent – Non-clustered 0.214 0.128 0.309 0.166 0.224 0.142 0.315 0.182 0.468 0.244 0.792 0.348

18. 18 Time-Dependent 100-Year Probabilities Probability Clustered – in a cluster38% Clustered – in a gap< 1% Not clustered14% Overall (T2 excluded)25% Overall (T2 included)< 5%

19. 19  The probabilistic hazard along the Pacific Coast is most sensitive to the time-dependent aspects of the CSZ recurrence.  Whether the full rupture earthquakes (M 9) are clustered or not has a significant impact on the time-dependent hazard.  Similarly if at present, we are in a gap or within a cluster also has a significant impact on the hazard in the Pacific Northwest.  The hazard is not very sensitive to the location of the eastern edge of the megathrust.  Our modeling of the CSZ recurrence is heavily dependent on the Goldfinger et al. (2012) interpretation of the turbidite record. Continued critical review of that interpretation is needed to improve our ability to estimate the probabilistic hazard from the CSZ. Summary

20. 20  “IT AIN’T WHAT YOU DON’T KNOW THAT GETS YOU INTO TROUBLE ---IT’S WHAT YOU KNOW FOR SURE THAT JUST AIN’T SO”. Final Comment Final Comment