1 Tom Parsons, U.S. Geological Survey, 345 Middlefield Rd. Menlo Park, CA, 94025 Earthquake probability calculated from paleoseismic observations on the south Hayward fault by Tom Parsons 1 Abstract. A recent statewide earthquake probability report issued by the Working Group on California Earthquake Probabilities calculated a mean 19.4% probability of a M≥6.7 earthquake rupturing the south Hayward fault in the next 30 years. This value was based on earthquake rate calculations inferred from observations of fault slip rate. Here, south Hayward fault probability values are presented that are based solely on the paleoseismic record. A recent consistency test between time dependent and time independent recurrence distributions was made using a Monte Carlo method to replicate the paleoseismic series on the south Hayward fault, which concluded that large Hayward fault earthquakes are quasi-periodic, and are most consistent with a stress renewal process. A by-product of the analysis yielded the full range of recurrence parameters that are consistent with paleoseismic observations. In this paper, these values are used to calculate rupture probability on the south Hayward fault. The resulting mean 10-yr probability (2010-2020) is 6.2%, while the mean 30-year probability (2010-2040) is 17.9%. Taking account of coseismic and post-seismic stress reduction from the 1906 earthquake on the south Hayward fault reduces probabilities to 5.2% in 10 years, and 15.6% in 30 years. Two independent approaches to earthquake probability calculations have now yielded similar, and relatively high mean 30-year results (17.9% and 19.4%). Thus expedient retrofitting of vulnerable structures along the south Hayward fault would be a prudent public investment. INTRODUCTION The remarkable paleoseismic sequence (Figure 1) developed by Lienkaemper and Williams (2007) for the southern Hayward fault in the San Francisco Bay area of California (Fig. 2) was examined quantitatively for consistency with time dependent and time independent recurrence models by Parsons (2008a). Individual time dependent distributions produced >5 times more matches to the observed record than the most common time independent distributions (Fig. 2). Within the framework of the test, the most likely south Hayward fault recurrence distribution is time dependent, with mean recurrence interval of μ=210 yrs, and coefficient of variation of α=0.6. This result is somewhat different than the mean of 170 yrs reported by Lienkaemper and Williams (2007) for the reasons described by Parsons (2008a, 2008b). Briefly, the difference arises because the arithmetic mean of a small sample of intervals is likely to be shifted towards the mode (most frequent value) rather than the true mean of a skewed, or asymmetric underlying recurrence distribution. Current consensus is that earthquake intervals distribute with strongly asymmetric shapes (e.g., Nishenko and Buland, 1987; Hagiwara, 1974; Kagan and Knopoff, 1987; Matthews et al., 2002). In this paper, implications on south Hayward fault rupture probability calculations are explored as caused by: (1) a time-dependent earthquake renewal process, (2) the distribution of allowable recurrence interval and coefficient of variation pairings as found from Monte Carlo analysis, and (3) potential delays caused by static stress transfer and post-seismic viscoelastic relaxation from the 1906 great San Francisco earthquake. 179 Proceedings of the Third Conference on Earthquake Hazards in the Eastern San Francisco Bay Area October 22-24, 2008