Comment on "Why Do Modern Probabilistic Seismic-Hazard Analyses Often Lead to Increased Hazard Estimates?" by Julian J. Bommer and Norman A. Abrahamson

Abstract

This comment discusses some heuristic biases with respect to the correct treatment of uncertainty in probabilistic seismic-hazard analysis (PSHA) observed in the recent article titled “Why Do Modern Probabilistic Seismic-Hazard Analyses Often Lead to Increased Hazard Estimates” by Julian J. Bommer and Norman Abrahamson. I show that the distinction between aleatory variability and epistemic uncertainty in seismic-hazard analysis represents a think model rather than an objective property of earthquake occurrence. It is demonstrated that the separation between epistemic uncertainty and aleatory variability is model dependent. I show that a correct application of this think model does not lead to an increase of hazard estimates, because the total uncertainty to be incorporated into a PSHA model is bounded by the uncertainty observed in the real world. Ground-motion variability cannot be statistically treated as an inherent property of earthquake occurrence. Its statistical measures are model dependent. A refined definition of the terms epistemic uncertainty and aleatory variability is suggested. Furthermore, the paper addresses the observation that the methodology of traditional PSHA may lead to a violation of the energy conservation principle. Finally, a summary of some of the most problematic areas of current PSHA methodology is given.