We present a strategy for the region‐specific assessment, adjustment, and weighting of ground‐motion prediction models, with application to the 2015 Swiss national seismic‐hazard maps. The models are provided within a logic‐tree framework adopted for the probabilistic seismic‐hazard analysis (PSHA). Through this framework, we consider both aleatory and epistemic uncertainties in ground‐motion prediction in Switzerland, a region of low‐to‐moderate seismicity and consequently data poor in terms of strong‐motion records. We use both empirical models developed using global strong‐motion data and stochastic simulation models calibrated to local seismicity, characteristic wave‐propagation effects, and site conditions. The empirical models were adjusted to account for (a) the selected hazard reference rock velocity model (using VS‐κ0 adjustments) and (b) the median instrumental ground‐motion data at low magnitudes. The use of a carefully calibrated simulation model and VS‐κ0 adjusted empirical ground‐motion prediction equations allowed us to precisely define the reference‐rock profile, upon which subsequent analyses, such as microzonation and site‐specific hazard, can be applied without uncertainty related to the reference condition. We implemented partially nonergodic aleatory uncertainties in ground‐motion prediction through the single‐station sigma approach. This strategy, complemented with the known reference rock condition, leads to significant reductions in the contribution of uncertainty in ground‐motion characterization to PSHA in Switzerland. However, the application of the methodological framework outlined herein extends to any region, particularly those of low‐to‐moderate seismicity.
Online Material: Tables of adjustment factors to convert selected ground‐motion prediction equations (GMPEs) to the Swiss rock reference and figures showing trellis plots of all adjusted GMPEs with uncertainty estimates.