The Crywolf Issue in Earthquake Early Warning Applications for the Campania Region

Abstract

Earthquake early warning systems (EEWS), based on real-time prediction of ground motion or structural response measures, may play a role in reducing vulnerability and/or exposure of buildings and lifelines. Indeed, seismologists have recently developed efficient methods for real-time estimation of an event’s magnitude and location based on limited information of the P-waves. Therefore, when an event occurs, estimates of magnitude and source-to-site distance are available, and the prediction of the structural demand at the site may be performed by Probabilistic Seismic Hazard Analysis (PSHA) and then by Probabilistic Seismic Demand Analysis (PSDA) depending upon EEWS measures. Such an approach contains a higher level of information with respect to traditional seismic risk analysis and may be used for real-time risk management. However, this kind of prediction is performed in very uncertain conditions which may affect the effectiveness of the system and therefore have to be taken into due account. In the present study the performance of the EWWS under development in the Campania region (southern Italy) is assessed by simulation. The earthquake localization is formulated in a Voronoi cells approach, while a Bayesian method is used for magnitude estimation. Simulation has an empirical basis but requires no recorded signals. Our results, in terms of hazard analysis and false/missed alarm probabilities, lead us to conclude that the PSHA depending upon the EEWS significantly improves seismic risk prediction at the site and is close to what could be produced if magnitude and distance were deterministically known.