Quantification of source-to-site distance uncertainty in ground motion models

摘要

Among different input data, source-to-site distance plays a major role in the results of ground motion models (GMMs). In order to determine the source-to-site distance, geometric characteristics of the seismic source need to be specified. This can be challenging when the seismic source is not known thoroughly. Empirical relationships themselves which are used to determine the geometric characteristics of seismic sources contain large degree of uncertainty. In this paper, a simple algorithm based on Monte Carlo (MC) simulation method which quantifies the uncertainties in distance metric and geometric characteristics of seismic sources is proposed. The jointly effects of magnitude and distance are considered in the proposed algorithm for uncertainties modeling. Also, this algorithm has been used to quantified errors resulted from inputting inaccurate source-to-site distance metrics in the GMMs. NGA-West2 global GMMs and event-specific isotropic and non-isotropic GMMs are used in the analysis. The results demonstrate that the uncertainty in the measurement of different source-to-site distance definitions depends on the magnitude and on the location of site with respect to the seismic source. It is also observed that the distance measurement uncertainty has a direct effect on the outcomes of GMMs. GMMs' coefficient of variation maps demonstrate that the amount of uncertainty is higher around the fau for large magnitudes, outcome variation of GMMs reaches as high as 40% of the average predictions. The coefficient of variation in GMM results decreases with the increase in distance metrics considered where at a distance beyond 30 km, the coefficient of variation of GMM estimates drops from 40 to 10%.