Using Coastal Arrays to Improve Earthquake Early Warning and Offshore Detection in Subduction Zones

Abstract:

Earthquake Early Warning (EEW) in subduction zones can be challenging due to the difficulty in locating offshore earthquakes using only a small number of P-wave observations. Many EEW methods are subject to a tradeoff between their location and magnitude estimates determined from the initial P-wave data. For instance, the ShakeAlert EEW system routinely detects moderate M4-6 earthquakes offshore of the Mendocino Triple Junction (MTJ) in Northern California, but it sometimes shows 50-100km location errors that could bias a magnitude estimate by as much as 0.5 units, which in turn could generate unnecessary alerts as far away as San Francisco for M6 earthquakes.

We deployed an array of short-period nodal seismometers around the Northern California Seismic Network station KCT located just north of the MTJ. The array consists of 20 instruments with a spatial footprint of less than 1 km. We find that standard beamforming measurements on the first 1.0 seconds of P-wave recording are sufficient to achieve a back-azimuth resolution of the incoming P-waves from offshore earthquakes within 10 degrees. These measurements are more robust than single station polarization measurements that use longer P-wave time windows and often have errors of tens of degrees. From the observed back-azimuth error distribution, we find that adding array measurements at the first few stations can significantly improve location errors for earthquakes up to 100km offshore. Adding coastal arrays could improve EEW alerting performance in subduction zones.

Motivated by the 2011 Tohoku-Oki and 2014 Iquique earthquakes, one key goal of the SZ4D Initiative is to monitor offshore megathrusts for earthquake swarms and determine if they show anomalous characteristics before large ruptures. Coastal arrays could be a cost-effective approach to achieve this goal. In particular, the array constraints on offshore earthquake locations could be optimized by calibrating the arrays with temporary ocean bottom seismometer experiments or airgun shots, allowing improved performance even when seafloor instrumentation is unavailable. Moreover, the array approach can reduce detection thresholds by approximately 1 magnitude unit at 50-100km offshore, allowing more comprehensive searches for anomalous precursory seismicity before large megathrust ruptures.