Performance of the earthquake early warning system for the 2024 Noto Peninsula earthquake

Abstract The Noto Peninsula earthquake (Mj7.6), which occurred on New Year’s Day of 2024, had two characteristic features: multiple tremors at the initiation of the rupture and a long fault rupture exceeding 100 km. The source process included three significant tremors for 15 s: Mj ~ 3 event, Mj 5.9 event, and Mj 7.6 event. The rupture started at the tip of the Noto Peninsula and propagated bilaterally in northeast and southwest directions. We evaluated the performance of the Japanese earthquake early warning (EEW) issued to the public. The source determination process of the EEW was triggered by the preceding Mj ~ 3 event and the warning threshold was exceeded by the Mj 5.9 event, so there was at least a 13-s lead time before the S-arrival of the Mj 7.6 event, allowing many residents to take protective measures. The first warning was issued to only the Northern part of the Ishikawa prefecture. However, the second warning that was distributed to as far as a few hundred kilometers was issued 27.1 s after the first warning, which was longer than expected. This is because the magnitude was underestimated during the rupture process and the warning was issued based on the shaking observation of the Mj7.6 event. We recomputed the shaking estimation from the Integrated Particle Filter (IPF) method and the Propagation of Local Undamped Motion (PLUM) method used in the Japanese EEW, and additionally, the XYtracker method to evaluate the effect of fault finiteness. At the initial part of the rupture, the fault finiteness is difficult to capture, and the finite-source approach produced a similar shaking estimation to the point-source approach. As the rupture propagates, shakings in the western area near the fault were significantly underestimated by the point-source approach. For large earthquakes, considering fault finiteness may be able to capture the rupture directivity and improve the accuracy of shaking estimation. Graphical Abstract