1. Faults, Earthquakes, and Simulations
David D. Jackson, UCLA
and
Anna E. Holt, MIT

2. Standard recipe for hazard estimation
Determine maximum magnitude from fault geometry.
Estimate relative magnitude distribution (characteristic, Gutenberg-Richter, etc.).
Set earthquake rate to match fault slip rate.
Adjust for cascades, time dependence, etc.
Estimate probable shaking for each hypothetical quake.

3. Procedures Establish selection criteria California Mw>5.5
Prior fault map (implies after 1975)
In Wells & Coppersmith, +Hector
Estimate fault length from prior fault map
Compare prior fault length with rupture length (from Wells and Coppersmith).
Correlate Mw with prior fault length

4. The Problem
Regressions of magnitude on fault dimensions (as in Wells and Coppersmith) use rupture length, unknown before quake.
Future magnitude limits are inferred from prior mapped fault length.
Equivalence assumes that
rupture length is limited by prior fault length, and
magnitudes for fault-limited earthquakes behave the same as those for ordinary earthquakes (as used by Wells and Coppersmith)
Few or none of Wells and Coppersmith quakes were limited by fault dimension.

5. Possible relationships between fault and rupture

6. New #4

7. Old #4

9. Mw vs. Surface Rupture Length

10. Mw Vs. Prior Mapped Fault Length

11. Conclusions
California earthquakes commonly (5/14) rupture past mapped fault ends.
Mapped fault length is not a good predictor of quake size limit.
Wells and Coppersmith relation ok for posterior scaling, but not for predicting magnitude limits.
Earthquakes may be larger than generally assumed in past hazard estimates, but limiting factors unknown.
Simulations of earthquakes limited to mapped faults clearly inadequate.
Simulations of earthquakes based on existing faults, even unmapped, are probably also inadequate.

12. New# 1

13. Old #1

14. New #2

15. Old #2

16. New #3

17. Old #3