5:15 PM - 6:45 PM
[SCG53-P04] Performance of the Earthquake Early Warning System for the 2024 Noto Earthquake
Keywords:Earthquake Early Warning, Noto earthquake, IPF method, PLUM method
On 1 January at 16:10 (Japan local time), a large earthquake (Mj 7.6) occurred on the Noto peninsula, in central Japan. It was the evening of New Year’s Day, and many people went back to their hometown for the family gathering. More than 200 people were killed mainly due to the collapse of the housing. The strong shaking was observed in the wide area, the PGV exceeded 100 cm/s along the fault about 100km from the epicenter.
The first earthquake early warning was issued at 6s after the first detection of the P-wave arrivals. The earthquake rupture was quite complicated; the very small foreshock (Magnitude undetermined) occurred at 16:10:09, and most of the near-source seismometers picked this P-wave. The M5.9 earthquake occurred at 16:10:10, with a very sharp onset. The largest M7.6 earthquake occurred at 16:10:23. These three earthquakes occurred nearby area (<3km), so the EEW has 14 s extra warning time before the S-wave arrival of the largest earthquake in this sequence. It was fortunate for the warning purpose, and most of the people received the warning before or right after the strong shaking started. There were multiple movies of dashboard cameras showing that drivers stopped their cars after receiving the warning message.
The first warning was issued to the area of Noto region, but the warning was not updated till 16:10:43, 27 s after the first warning. This seems to be too long to capture the size of the M7.6 event. The time history of the magnitude estimate increased for 6s after the first warning, but it stayed at M6.6 for about 30s. The second warning, extended from the first warning region, was issued with no change of location and magnitude estimations. Therefore, the second warning was issued by the PLUM method, not the IPF method based on the source location. The third warning was issued by the PLUM method for the same reason.
The magnitude was estimated from the closest 5 stations. Since September 2023, Hi-net stations have also been processed by the IPF method, along with the JMA acceleration data. If the instrumental correction were properly applied, the underestimation of the magnitude due to the sensor characteristics would be minimized. We compare the results of the IPFx method and JMA EEW system and discuss the possible improvement.
The first earthquake early warning was issued at 6s after the first detection of the P-wave arrivals. The earthquake rupture was quite complicated; the very small foreshock (Magnitude undetermined) occurred at 16:10:09, and most of the near-source seismometers picked this P-wave. The M5.9 earthquake occurred at 16:10:10, with a very sharp onset. The largest M7.6 earthquake occurred at 16:10:23. These three earthquakes occurred nearby area (<3km), so the EEW has 14 s extra warning time before the S-wave arrival of the largest earthquake in this sequence. It was fortunate for the warning purpose, and most of the people received the warning before or right after the strong shaking started. There were multiple movies of dashboard cameras showing that drivers stopped their cars after receiving the warning message.
The first warning was issued to the area of Noto region, but the warning was not updated till 16:10:43, 27 s after the first warning. This seems to be too long to capture the size of the M7.6 event. The time history of the magnitude estimate increased for 6s after the first warning, but it stayed at M6.6 for about 30s. The second warning, extended from the first warning region, was issued with no change of location and magnitude estimations. Therefore, the second warning was issued by the PLUM method, not the IPF method based on the source location. The third warning was issued by the PLUM method for the same reason.
The magnitude was estimated from the closest 5 stations. Since September 2023, Hi-net stations have also been processed by the IPF method, along with the JMA acceleration data. If the instrumental correction were properly applied, the underestimation of the magnitude due to the sensor characteristics would be minimized. We compare the results of the IPFx method and JMA EEW system and discuss the possible improvement.