Speaker: Sarah Minson
The objective of earthquake early warning (EEW) is to provide population centers with a useful warning of impending strong ground shaking: the warning must be sufficiently rapid and accurate for users to take action to mitigate damage. By considering the time scale of evolving moment release and observed ground motion variability, we find that EEW systems may perform quite differently than previously understood. Many current-generation EEW systems estimate the location and magnitude of an earthquake in real-time and input this information into a Ground Motion Prediction Equation (GMPE) to identify and alert people and infrastructure expected to experience strong shaking before that shaking arrives at their location. Much of our theoretical knowledge of how these systems behave relies on the assumption that the final earthquake magnitude can be known at the beginning of the rupture and that the predicted ground motion from this source is accurate. Removing these assumptions, we find that the timeliness of alerts depends strongly on the level of ground motion for which users want to be warned. To estimate that a user will experience ground motion exceeding a higher threshold, the EEW system must wait until the rupture grows larger, which cuts into potential warning time. Thus, we cannot expect long warning times for destructive levels of ground motion. Further, even if alerts are timely, they often will be inaccurate. Observed ground motion typically has a factor of two variability (one standard deviation) about the median expected ground motion predicted by a GMPE. As a result, ground motion predictions – including those based on accurate source parameters – will occasionally greatly under-estimate or over-estimate the ground motion the user will experience, resulting in missed alerts and false alarms, respectively. For false-alarm-tolerant users, EEW performance is optimized by taking action at low ground motion levels, improving alert timeliness and decreasing the number of missed events at the expense of more false alarms.