hide
Free keywords:
-
Abstract:
Ground motion variability is often left aside in the validation framework of physics-based ground motion simulation. In this study, we evaluate how the distributions of model parameters control ground-motion variability, and whether the simulations are consistent with the empirical ground-motion model in terms of both median and standard deviation.
Our analyses consist of two main steps. First, we establish a database that contains a series of simulated ground motions from a stochastic catalog (i.e., a set of earthquake scenarios consistent with the source model developed recently for the German Probabilistic Seismic Hazard map) that could occur in the Rhine Graben area. Ground-motion simulations are performed adopting the simulation method of Graves and Pitarka (2010, 2015) implemented in the Southern California Earthquake Center (SCEC) broadband platform (BBP), which we have tailored for use in the Rhine Graben. In a second step, we ablaze ground motion variability by deconstructing it into its between-event and within-event components
The analyses are performed through six different modeling assumptions reflecting various choices of stress parameters, velocity structure, site effects, and source parameters. The residual analysis through random-effect splitting captures the main components of the variability of resulting simulations. The between-event variability, for instance, is mainly controlled by stress parameter distribution at short-period and by the source parameters and slip distributions at long-period. If these distributions are well-chosen the simulations reproduce the variability of empirical models. Our results however highlight that simulations based on one-dimensional velocity models cannot reproduce the observed within-event (path-to-path) variability.
