The Turkey Flat Blind Prediction Experiment for the September 28, 2004 Parkfield Earthquake: Comparison with Other Turkey Flat Recordings

A blind prediction experiment was conducted for the Turkey Flat test area data after the September 28, 2004 Parkfield earthquake. The results of the test were reviewed at the Turkey Flat Blind Prediction workshop in 2006. The predicted ground motions at the center of the valley significantly exceeded the observed ground motions. In this study, records of Parkfield aftershocks and two other events, including the 2003 San Simeon earthquake, are analyzed and compared with the 2004 Parkfield mainshock records. Comparison of the observed ground motions at the surface and at depth at the rock outcrop station shows little amplification from a depth of 24 m to the surface. This observation may be expected as the instruments at both the surface and at depth are located in the same rock formation. Average rock motions beneath the valley center site are considerably weaker than those recorded 800 m away at a similar depth below the rock outcrop. This observation implies that the recorded motion at depth under the rock outcrop should not be applied directly as the input motion at depth under the sedimentary valley for the purpose of predicting ground motion at the surface. The peak ground accelerations at the surface at the center of the valley are largest for nearly all events. The ratio of the peak acceleration from the surface to depth varies from about 1.5 to more than 6. The smallest amplification is for the distant M6.5 San Simeon earthquake of 2003, which has a lower dominant frequency content than the frequencies amplified at the valley center site.