Bayesian Treatment of Induced Seismicity in Probabilistic Seismic-Hazard Analysis

This article proposes a Bayesian treatment of uncertainty in probabilistic seismic-hazard assessment to account for the possibility of anthropogenic changes in the properties of future earthquake occurrences. Operational factors are used to establish prior distributions for future seismicity, and observations are used to dynamically update the distributions and resulting hazard calculations. The updating process is relevant for regions of the central and eastern United States where large increases in the rate of earthquakes, apparently triggered by anthropogenic activities, are difficult to address using conventional quantitative methods for assessing seismic hazards. The approach is consistent with conventional seismic-hazard analysis initially, allows a number of types of data to be used for updating in an informative and transparent manner, and may incentivize improved monitoring of at-risk regions. Examples are provided to illustrate the approach, including the use of Gibbs sampling to characterize updated distributions in the case for which analytical solutions to the updating equations are not available.