The Virtual Seismologist (VS) Method: a Bayesian Approach to Earthquake Early Warning

Abstract

The goal of earthquake early warning is to provide timely information to guide damage-mitigating actions that can be taken in the few seconds between the detection of an earthquake and the onset of large ground motions at a given site. From a subscriber’s perspective, effective early warning consists of both real-time information about the expected ground motions, as well as a methodology of how to use this information, and the inherent uncertainties, to guide decision-making. The Virtual Seismologist (VS) method is a Bayesian approach to early warning that provides a unified framework for the real-time earthquake source estimation, as well as the subscriber’s decision-making problem. The introduction of prior information into the source estimation problem via Bayes’ Theorem distinguishes the VS method from other paradigms for earthquake early warning. Station locations, previously observed seismicity, and known fault traces are among the type of information that can be used to resolve trade-offs in magnitude and location that are unresolved by the ground motion observations alone at the initial stages of earthquake rupture. The benefits of prior information are most evident in regions of low station density, where large inter-station distances result in source estimates based on a relatively sparse set of observations. The drawback of prior information is the increased complexity of information that must be communicated to the user, as the resultant earthquake source estimates can no longer be adequately described by Gaussian distributions. We illustrate the performance of the VS method in regions of high and low stations density, and discuss how subscriber requirements ultimately dictate how the real-time source estimation problem must be addressed.