Epistemic Uncertainties in Local Earthquake Locations and Implications for Managing Induced Seismicity

Abstract

Earthquake hypocentral location is perhaps the most classical problem in seismology, the solution of which is often affected by significant uncertainty. In monitoring the effects of underground anthropogenic activities, the earthquake hypocentral location, magnitude, and ground motions are important parameters for managing induced seismicity (as e.g., for operating traffic‐light systems). Such decisional systems define the operative reactions to be enacted once an earthquake, exceeding some magnitude or ground‐motion threshold, occurs within a monitoring volume defined in the neighborhood of a certain anthropogenic underground activity. In this case, a reliable evaluation of the hypocentral location, along with its uncertainty, becomes crucial for rational decision making. In this article, we analyze different sources of uncertainty that can be relevant for the determination of earthquake source locations, and introduce a logic‐tree‐based ensemble modeling approach for framing the problem in a decision‐making context. To demonstrate the performance of the proposed approach, we analyze uncertainties in the location of a seismic event that occurred on 22 July 2019 within the perimeter of the monitoring domain defined in the Val d’Agri oil field (southern Italy). We cast the result as a model ensemble that allows us to obtain samples from a parent distribution that better represents both aleatory and epistemic uncertainties of the earthquake location problem. We find that often‐neglected epistemic uncertainties (i.e., those that arise when considering alternative plausible modeling approaches or data) can be considerably larger and more representative of the state of knowledge about the source location, than the standard errors usually reported by the most common algorithms. Given the consequential repercussions of decision making under uncertainty, we stress that an objective evaluation of epistemic uncertainties associated with any parameter used to support decisional processes must be a priority for the scientific community.