Download
slide1 n.
Skip this Video
Loading SlideShow in 5 Seconds..
Use of Precarious Rocks to Test Earthquake Ground-Motion Simulations PowerPoint Presentation
Download Presentation
Use of Precarious Rocks to Test Earthquake Ground-Motion Simulations

Use of Precarious Rocks to Test Earthquake Ground-Motion Simulations

90 Views Download Presentation
Download Presentation

Use of Precarious Rocks to Test Earthquake Ground-Motion Simulations

- - - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript

  1. Use of Precarious Rocks to Test Earthquake Ground-Motion Simulations SCEC CyberShake Collaboration: Robert Graves, Ned Field (USGS); Scott Callaghan, Thomas H. Jordan, Philip Maechling, Kevin Milner, Gideon Juve, David Okaya, Patrick Small (USC/SCEC); EwaDeelman, Gaurang Mehta, Karan Vahi (USC/ISI) • CyberShake: Simulation Based Seismic Hazard Modeling • Possibly inconsistent with PBRs • Rupture Characterization Issues • Magnitude-Area scaling relations (depth of rupture) • Aleatory magnitude variability • Spatio-temporal rupture complexity

  2. CyberShake Seismic Hazard Platform • Replace ground motion prediction equations (GMPEs) with physics-based simulations • Account properly for rupture directivity and basin effects • Predict full time-series of ground motion • “Scenario-based” seismic hazard calculation, incorporating many thousands of scenarios • Simulates ground motions for potential fault ruptures within 200 km of each site • 40,000 sources (Mw > 6) in Southern California from UCERF2.0 (2008) • Extends UCERF2.0 to multiple hypocenters and slip models for each source • 440,000 ground motion simulations for each site

  3. CyberShake 1.0 computation (225 sites, f < 0.5 Hz) • 440,000 simulations (i.e. EQ scenarios) per site • 50-day run on Ranger (5.3 million hrs, 4,400 cores) • 189 million jobs • 46 petabytes of total I/O • 176 terabytes of total output data • 2.1 terabytes of archived data • Site-based approach: • Naturally includes source and path effects, and their variability • Alleviates need for ergodic assumption LA region

  4. 3 sec SA Hazard Curve at Aliso PBR Site • CyberShake PSHA is lower than, but still close to GMPE predictions PBR might not survive (PGA?) • CyberShake motions continue to grow at low probabilities

  5. Magnitude-Area Scaling Relations • UCERF2.0: • Ellsworth-B 50% • Hanks and Bakun 50% • Wells and Coppersmith 0% • Somerville 0% • For Mw > 7.5 Ellsworth-B/Hanks and Bakun predict about 0.2 magnitude units higher than Wells and Coppersmith/Somerville (about twice Mo) • Or for same magnitude, this represent a difference of about 50% - 60% in rupture area (and 25% in slip) • Initial CyberShake calculations were noticeably larger than expected and this was attributed to Mag-Area scaling

  6. Down-dip widths in UCERF2.0 had to be increased to be consistent with waveform-based estimates of rupture areas. UCERF2 Weighting Ellsworth-B 50% Hanks-Bakun 50% Wells-Coppersmith 0% Somerville 0% CyberShakeERF Weighting Ellsworth-B0% Hanks-Bakun 0% Wells-Coppersmith 50% Somerville 50%

  7. Magnitude-Area Verification Many recordings for Mw 6 – 7 within 70 km Magnitude-Area correction brought CyberShakeinto better agreement with GMPE predictions where empirical constraints are strong.

  8. Aleatory Magnitude Variability • For a given rupture area, aleatory magnitude variability in UCERF2.0 can be > 0.7 magnitude units • GMPEs depend only on magnitude (not stress drop), and thus implicitly adjust stress drop as the magnitude changes • Numerical simulations are much more sensitive to this variability than GMPEs because it translates directly into slip • Consequence is large magnitude, high stress drop events produce very large simulated ground motions

  9. San Andreas Example • 470 km rupture length • Median Mw 8.0

  10. CyberShake Kinematic Ruptures

  11. Average Horizontal SA

  12. Spatio-Temporal Rupture Complexity Effects on Rupture Directivity strong weak

  13. Recent enhancements to rupture generator provide increased heterogeneity in rupture propagation speed and slip rise time => Decrease in coherence of radiated motions

  14. strong weak

  15. Summary • Need to resolve Magnitude-Area scaling discrepancies:GMPE’s are insensitive to changes in rupture area for same magnitude; however, long period simulations scale almost directly with slip (1/area). • Magnitude variability: For a given rupture area,aleatory magnitude variability in UCERF2.0 can be >0.7 units. This produces a range in simulated ground motions of nearly a factor of 10; however, median GMPE’s ground motions vary by only factor of 2 or less (double counting?). • More sophisticated rupture characterization: Current set of ruptures are relatively coherent at long spatial and temporal length scales, which may over estimate directivity effects. New class of rupture generators contain greater intra-event variability of rupture speed and rise time. Must quantify expected variability of these parameters and develop appropriate joint probability density functions (e.g., via rupture dynamics).