Methodology for using precarious rocks in Nevada to test seismic hazard models

Abstract Fields of precariously balanced rocks indicate that strong earthquake motions have not occurred at that site since the precarious rocks developed. These fields can be characterized with an estimate of the peak acceleration that would be sufficient to topple the rocks and an estimate of how long the rocks have been precarious. This article uses this information to test the input to probabilistic seismic hazard analysis. The fundamental assumption is that the probability of exceeding a ground motion capable of toppling a precarious rock during a time period equal to the age of the rock is equal to the confidence level at which the inputs to the probabilistic seismic hazard analysis can be rejected. We performed a probabilistic seismic hazard analysis for 26 sites of observed precarious rocks in Nevada, using preliminary estimates of the toppling acceleration and the age of the features. Two standard models are rejected with over 95% confidence by most of the precarious rock observations. We consider several possible explanations for the inconsistency. A probabilistic seismic hazard analysis that eliminates the area sources and only includes faults is consistent with the precarious rock observations at most of the sites. Spatial clustering of small-magnitude events could maintain this effect without contradicting the precarious rock observations. However, it may not be necessary to completely reject the area sources from our probabilistic seismic hazard analysis. The physics of rock stability may allow increasing the minimum magnitude to 6.0 in the area sources, because the short duration of high-frequency accelerations in smaller events may not topple all precarious rocks. The precarious rocks could generally have small site effects. Attenuation models may overestimate ground motions on the foot wall of normal faults, and a different attenuation model in general might decrease the ground-motion predictions. The ergodic assumption that is made in estimating the uncertainty in attenuation models may be contributing to overestimate the hazard. Individually, each of these effects might allow more of the precarious rock sites to be consistent with the area source zones.