Sensitivity study for stochastic finite-fault model in eastern North America

Abstract

The reliability of structures depends on the uncertainties of random characteristics of strong ground motions. This study investigates the sensitivity level of the stochastic finite-fault ground-motion simulation model in response to major uncertain input variables, with special focus on eastern North America. The finite-fault model is one of the most useful methods to simulate ground motion for a large earthquake. The variance-based sensitivity analysis using the Monte Carlo-based procedure is considered to compute the set of sensitivity indices of the input variables to estimate the contribution of each uncertain input parameter. The variables that have a dominant influence on the uncertainty of simulated ground motions are identified. The analytical results show that the stress parameter, fault model variables (including faulting geometry, heterogeneous slip distribution and location of hypocentre within fault plane) and site-amplification factor are the main sources of uncertainty, while path variables, such as distance, anelastic attenuation and upper crust attenuation, have relatively little effect.