Selection of Gmpes for Probabilistic Seismic Hazard Assessment in Areas of Moderate Seismicity and Resulting Epistemic Uncertainties

Abstract eng:
In 2015, the reference probabilistic seismic hazard maps for Switzerland were updated. In order to cover epistemic uncertainties in ground-motion prediction two major sets of models were implemented: (1) stochastic and (2) empirical-based models. The stochastic models describe the synthetically generated ground-motion predictions specifically derived for Switzerland. Four models were implemented, in the Alpine and Foreland regions, segregating deep (hypocentral depth > 6 km) and shallow seismicity. All models are referenced to a well-defined rock shear-wave velocity profile and kappa0 value. This reference was derived from shear-wave velocity measurements and the assessment of kappa0 at a large number of seismic stations of the Swiss national seismic networks. The stochastic models are consistent with the magnitudes in the earthquake catalogue of Switzerland (ECOS-09), which was also used to derive the earthquake activity rates in the source model for the probabilistic hazard assessment. Geometrical spreading and attenuation were modeled from the observations in the Foreland and the Alpine regions, derived based on earthquake recordings of small events of the seismic networks in Switzerland and neighboring countries. The scaling to larger magnitudes was achieved using macroseismic data of historical events together with their calibrated moment magnitudes, resulting in an additional branch of the logic tree for the epistemic uncertainty on the stress drop. For verification, the stochastic model was successfully tested against European and Japanese strong-motion data. The empirical set of ground-motion models contains four empirical global ground-motion prediction equations (hereinafter GMPEs) extensively tested within other projects (i.e. Pegasos, SHARE). These models, however, had to be adjusted to the Vs-kappa0 conditions at the reference rock in Switzerland. The adjustment functions were obtained based on the ratio of the expected Fourier amplitude spectra at the host region and the target region. The host Vs profile for each empirical GMPE was estimated based on published information, personal communication with the GMPEs’ authors and expert evaluation. The host kappa0 values were obtained from a mixture of direct measurements and Vs30-based estimates. Additionally, the empirical GMPEs were corrected for small magnitude extrapolation using instrumental data. The adjustments and corrections were applied only to the empirical GMPEs and additional epistemic uncertainties due to these corrections were taken into account in the logic tree. The aleatory uncertainty is represented by single-station variability based on two alternative models. Finally, the two sets of ground motion models were combined together in the logic tree structure, by weighting the stochastic set and the empirical models. Here we present results that show the influence of the two sets of models on the resulting seismic hazard, and discuss the possible reason in the resulting epistemic uncertainties.

• Export as: BibTeX | MARC | MARCXML | DC | EndNote | NLM | RefWorks
• Add to your basket