Investigating the Contribution of Stress Drop to Ground-Motion Variability by Simulations Using the Stochastic Empirical Green’s Function Method

Abstract

The stress drop (Δσ) is a fundamental parameter used to quantify source physics, and its uncertainty is closely related to seismic hazard. To reveal the relationship between Δσ uncertainty and resultant ground-motion variability, ground motions produced by the 2013 Mw6.6 Lushan earthquake, characterized by various Δσ values, are simulated using the stochastic empirical Green’s function method. First, the variability in spectral amplitudes of simulated ground motions arising from the stochastic rupture process is investigated. Generally, it increases from ~ 0.05 to ~ 0.14 as the period increases up to 2.0 s, irrespective of the Δσ value used. The ground-motion variability due to Δσ uncertainty is then explored. The synthetic spectral amplitude is found to be linearly proportional to Δσb, thus the standard deviation of Δσ (log10 unit) is equal to the standard deviation of the spectral amplitude (log10 unit) multiplied by a factor b. The regressed b values are strongly dependent on the period and generally in the range of 0.7 to 0.6 up to the period of 2.0 s. These results explain how much of the ground-motion variability is caused purely by Δσ uncertainty. Moreover, the standard deviation of the spectral amplitudes is calculated directly from simulations based on random Δσ values following a lognormal distribution. The findings further verify the reliability of the relationship between Δσ uncertainty and ground-motion variability. Assuming that the interevent standard deviation in a ground-motion prediction model is dominated entirely by Δσ uncertainty and the stochastic rupture process, we estimate the standard deviation of log10Δσ (~ 0.2–0.3) for broad regions using various models.