Real‐Time Detection of Rupture Development: Earthquake Early Warning Using P Waves From Growing Ruptures

Abstract

Large earthquakes with long rupture durations emit P wave energy throughout the rupture period. Incorporating late‐onset P waves into earthquake early warning (EEW) algorithms could contribute to robust predictions of strong ground motion. Here I describe a technique to detect in real time P waves from growing ruptures to improve the timeliness of an EEW algorithm based on seismic wavefield estimation. The proposed P wave detector, which employs a simple polarization analysis, successfully detected P waves from strong motion generation areas of the 2011 Mw 9.0 Tohoku‐oki earthquake rupture. An analysis using 23 large (M ≥ 7) events from Japan confirmed that seismic intensity predictions based on the P wave detector significantly increased lead times without appreciably decreasing the prediction accuracy. P waves from growing ruptures, being one of the fastest carriers of information on ongoing rupture development, have the potential to improve the performance of EEW systems.

A Framework for Evaluating Earthquake Early Warning for an Infrastructure Network: An Idealized Case Study of a Northern California Rail System

“…Second, the lack of models for S to P ratios and for how shaking amplitudes evolve with time during a rupture prevents us from considering how lowering the alerting threshold for on-site ground-motionbased EEW systems could increase warning time. If the seismological community were to develop models for shaking covariance and evolution, it could potentially have significant impacts on the EEW community including improved performance assessments of ground-motion-based EEW methods such as PLUM, quantifying the potential utility of EEW methods that aim to predict peak shaking from P-wave amplitudes (Kodera, 2018), and improving real-time shaking forecasts.…”

Generalized Seismic Phase Detection with Deep Learning

“…By looking for signals that are reminiscent of P-waves, rather than arbitrarily polarized large amplitude signals, GPD can provide a far more stable base to EEW systems. Furthermore, the additional information of what phase-type a detection corresponds to (P vs. S) is valuable information in an EEW context, since it can be used for excluding S-wave observations from magnitude-estimation schemes that have been optimized only for P-waves [4] and to assess if an ongoing rupture is still growing [13].…”

An Earthquake Early Warning Method Based on Huygens Principle: Robust Ground Motion Prediction Using Various Localized Distance‐Attenuation Models

“…The algorithm's prediction timeliness could be improved if P wave intensities were used for attenuation estimations before the S wave arrivals. Kodera (2018) proposed an on-site algorithm that detects P wave arrivals based on a polarity analysis and predicts the corresponding S wave intensities from the P wave ground motions. He showed that the on-site method enhanced the Ground motion overpredictions at distant sites were caused by the model's use of the peak intensity at the station located closest to the M7.3 epicenter.…”

“…I also applied ALPHA to offshore and inland earthquakes with M ≥ 7 and calculated prediction errors and lead times (Figures c, d). The event data set was the same as that used by Kodera (), that is, 18 offshore and three inland events (excluding the Tohoku‐Oki and Kumamoto events) with M ≥ 7 that occurred from 2000 to 2016 for which JMA reported observed I JMA ≥ 5L.…”