On the manifestation of ground motion model differences on seismic Hazard sensitivity in North Iceland

Abstract

In this study, probabilistic seismic hazard assessment (PSHA) for North Iceland is explored in terms of its sensitivity to one of its key elements, the selected ground-motion models (GMMs). The GMMs in previous PSHA studies for Iceland are reviewed and in some cases recalibrated to the Icelandic dataset using a Markov Chain Monte Carlo (MCMC) algorithm which is useful in regions where the earthquake records are scarce. To show the ground motion model variability as it is manifested in PSHA uncertainties, the hazard maps of standard deviation and coefficient of variation (CV) of PGA at two hazard levels for GMMs before and after recalibrating are shown. The results indicate that the recalibrated models are promising candidates to be applied for future hazard studies in Iceland, but more importantly they show how to what extent and how the epistemic uncertainty of the GMMs contribute to patches of heightened hazard uncertainties, especially at near- and far-fault distances where there is a particular lack of data.