EVALUATING AND IMPROVING REAL-TIME STRATEGIES FOR ENGINEERING GROUND MOTION PREDICTIONS

Because, from the engineering perspective, the effectiveness of earthquake early warning systems (EEWS) depends only on the possibility of immediately detecting the earthquake and estimating the expected loss, or a proxy for it, for an engineered system of interest in order to undertake actions to manage/mitigate the risk before the strike, it is worthwhile to assess the efficiency of strategies to predict in real-time the earthquake’s destructive potential. The simplest engineering ground motion parameter is the peak ground acceleration (PGA) which may be predicted through probabilistic seismic hazard analysis in the framework of EEW conditional on some measures the seismologists use to estimate the magnitude from the early recorded signal. The effects of different sources of uncertainty on the prediction of PGA are assessed with reference to the ISNet (Irpinia Seismic Network) EEWS, although results can be considered general. The analyses show how the uncertainty of the ground motion prediction equation (GMPE) dominates those of magnitude and distance, almost independently of the information available for the event. Because the uncertainty related to GMPE is usually very large, it seems that the estimation of PGA should be where to put effort rather than improving the estimation of magnitude and/or earthquake location. An attempt to reduce the uncertainty in the estimation of PGA is made by adding more information (i.e., a second parameter measured in the early part of the signal from real-time seismology) and using the intra-event spatial correlation of peak accelerations at different sites. Based on these analyses distance-related bounds to uncertainty and information-dependent lead-time maps are defined and illustratively computed for the Campania (southern Italy) region.