DATA NEEDS FOR IMPROVED SEISMIC HAZARD ANALYSIS John G. Anderson, James N. Brune, Abdolrasool Anooshehpoor and Matthew D. Purvance Seismological Laboratory and Department of Geological Sciences University of Nevada Reno, Nevada 89557 Figure 1. Map of part of southern California. The heavy line shows the approximate extent of rupture of the 1857 Fort Tejon, California, earthquake, Mw=7.9. The large 4-pointed star shows the location of Lovejoy Buttes (~34.60N, 117.85W). The Lovejoy Buttes are significant for the presence of 10,000 year old, semi-precarious rocks that have not been toppled by earthquakes. The inset photo (upper right) shows some of these rocks, with the San Andreas fault in the background. The inset perspective view (lower left) views the region along the direction of rupture propagation in the 1857 earthquake. A B Figure 6. A) Illustration of a geometry where several stations all record the ground motion at the same distance from a vertical strike-slip fault. B) Peak acceleration results of a numerical experiment by Yuehua Zeng (personal communication) in which the experiment in part A is implemented using a numerical scheme to calculate synthetic seismograms (Zeng et al, 1994). The experiment was performed at several distances, as given by the horizontal axis. C) Estimated standard deviation at each distance. Peak Acceleration (g) Figure 7. Illustration of the difference between epistemic and aleatory uncertainty, based on the numerical experiment in Figure 6. The green line shows the distribution of mean values of peak acceleration over the set of equidistant stations from the fault. The blue line shows the probability distribution for repeated recordings of the ground motion at a single site. Precarious Rock at Lovejoy Buttes, a 4 T =0.38 , mu=0.38g Purvance probability of topplin g GM prob density, d F /d x 2 0 0 0.5 1 1.5 2 2.5 3 1.5 b Probability density of toppling, per event Cumulative probability of toppling, per event 1 0.5 0 0 1 0.5 1 1.5 Peak Acceleration (g ) c 2 2.5 3 Probability of surviving (250 year repeat time) 0.5 Age of this rock 0 0 1000 2000 3000 4000 5000 Age of Rock 6000 7000 8000 9000 10000 Figure 8. A) Fragility curve for toppling a balanced rock with parameters typical of Lovejoy Butte. The monotonic increasing curve gives the probability that the rock will topple as a function of increasing peak acceleration in a suite of records with frequency content appropriate to a magnitude 8 earthquake. This is a highly nonlinear problem, where the outcome for any individual record is either toppling or not, and switches between the two domains as peak acceleration increases. This curve is smoothed and approximately monotonic due to summing the curves for a large suite of records. The peaked curve on this plot is the probability density distribution for the peak acceleration based on the Sadigh et al (1997) ground motion prediction equation. conditional on the earthquake happening. B) The peaked line gives the probability density for the rock to topple, conditional on the earthquake happening, obtained as the product of the two curves in part A. The monotonic increasing line is the cumulative probability of toppling per event. Under the assumptions used here, the overall probability of toppling is about 0.3 in each characteristic earthquake. C) Probability of the rock surviving as a function of it’s age. With a probability of toppling of about 50% per earthquake, the half life of an individual rock of this geometry is about 400 years, or two earthquake cycles. The bold vertical line shows the age of the rock. Lovejoy Buttes Constraint 0.5 Boore et al., 1997 0.45 Abrahamson and Silva, 1997 0.4 Sadigh et al., 1997 0.35 0.3 Not Allowed 0.25 0.2 Allowed 0.15 t 1/2 0.1 0.25 t 1/2 =5000 year s 0.3 0.35 =2000 year s 0.4 0.45 0.5 Mean Peak Acceleration (g) Figure 9. One way of evaluating ground motion prediction equations for the specific case of Lovejoy Buttes. The contours show two half-live lines in the peak acceleration – aleatory uncertainty plane. Since the rocks at Lovejoy Buttes are over 10,000 years old, ground motion prediction equations substantially to the right of the 2000 year half-life line are inconsistent. Lovejoy Buttes Constraint 0.5 Boore et al., 1997 0.45 Abrahamson and Silva, 1997 0.4 Sadigh et al., 1997 0.35 0.3 Not Allowed 0.25 0.2 0.15 90% 0.1 0.25 50% 0.3 10% 0.35 Prob. to survive 10 0.4 4 years 0.45 0.5 Mean Peak Acceleration (g) Figure 10. An alternative way of evaluating ground motion prediction equations. The contours on the peak acceleration – aleatory uncertainty plane give the probability for a rock with a certain geometry to survive 104 years.