11:45 AM - 12:00 PM
[STT36-10] Development of the Earthquake Early Warning Utilizing DAS with Optical Fiber Cables along the Shinkansen Railway
Keywords:DAS, EEW, Shinkansen, fiber cable
The early earthquake warning (EEW) used in the Shinkansen estimates parameters such as epicentral distance and magnitude using data from individual seismometer. Processing at individual seismometer has the disadvantage of poor estimation accuracy, but the advantage of being able to issue warnings immediately. The Shinkansen EEW algorithm estimates seismic parameters using short data (usually 1-2 seconds) immediately after the P-wave arrival, obtained from seismometers installed along the railway line (approximately every 20 km) or on the coast. However, there are estimation errors of about half the epicentral distance, ±30° in epicentral direction, and ±0.6 in magnitude.
Distributed Acoustic Sensing (DAS) uses the phase change of scattered wave in optical fiber cables to measure strain changes along the cable (Katakami et al., 2024). This technology can measure data every few meters, making it possible to detect seismic waves at multiple points almost simultaneously (within, for example, 0.5 seconds). Therefore, it is possible to apply a method for determining the epicenter using data from multiple observation points and to statistically evaluate the results of the estimated magnitude for each point, which may lead to higher accuracy of the warnings. In this study, we introduce a method to estimate seismic parameters quickly after detecting seismic waves using DAS acquired from existing optical fiber cables along the Shinkansen.
We conducted earthquake observations using DAS applied to existing optical fiber cables along the Kyushu Shinkansen. The observation periods were from January to February 2022, February to March 2023, and December 2023 to April 2024. In 2022, we used a 75 km section from Shin-yatsushiro Station to Shin-oomuta Station, and in 2023 and 2024, we used a 100 km section from Shin-yatsushiro Station to Kurume Station. We used DAS from AP Sensing and installed the interrogator at Shin-yatsushiro Station.
In this presentation, we developed a method to estimate the epicenter and magnitude in real time as needed. Additionally, we paid attention to minimizing the computational cost for use in EEW. Yin et al. (2023) showed a magnitude scaling law for strain rate, magnitude, and epicentral distance measured by DAS. We set the parameters of the above the law using the maximum strain rate for 1 second after the P-wave arrival time in the DAS data along the Kyushu Shinkansen, and calculated the magnitude scaling equation in the DAS data. To determine the epicenter in real time, we developed a method to robustly determine the arrival times of the P-wave and S-wave based on STA/LTA. When the number of channels that detected the P-wave exceeded 10, we determined the epicenter using hypomh (Hirata and Matsu'ura, 1987) and used the estimated result as an initial value. Next, if the number of channels that detected the P-wave exceeded 20, we determined the epicenter again using hypomh and updated the initial value. We repeated this process until the number of channels that detected the P-wave exceeded 500.
We applied the developed method to earthquake data with a magnitude of Mj2.0 or higher and an epicentral distance of 150 km or less. Using the data of the P-wave arrival time obtained within 1 second after detecting the P-wave for the first time, we were able to determine the epicenter with an error of several kilometers for earthquakes with an epicentral distance of less than 100 km. However, for earthquakes with an epicentral distance exceeding 100 km, the error in the epicenter determination was large. This is thought to be due to the difficulty in obtaining significant arrival time differences with respect to the observation point configuration (approximately 100 km) as the epicentral distance increases. For earthquakes with an epicentral distance of less than 100 km, we estimated Mdas. We calculated the maximum strain rate for 1 second after the P-wave detection in each channel and substituted it into the magnitude scaling law. As a result, it was found that Mdas could be estimated with an error of about ±0.5.
Distributed Acoustic Sensing (DAS) uses the phase change of scattered wave in optical fiber cables to measure strain changes along the cable (Katakami et al., 2024). This technology can measure data every few meters, making it possible to detect seismic waves at multiple points almost simultaneously (within, for example, 0.5 seconds). Therefore, it is possible to apply a method for determining the epicenter using data from multiple observation points and to statistically evaluate the results of the estimated magnitude for each point, which may lead to higher accuracy of the warnings. In this study, we introduce a method to estimate seismic parameters quickly after detecting seismic waves using DAS acquired from existing optical fiber cables along the Shinkansen.
We conducted earthquake observations using DAS applied to existing optical fiber cables along the Kyushu Shinkansen. The observation periods were from January to February 2022, February to March 2023, and December 2023 to April 2024. In 2022, we used a 75 km section from Shin-yatsushiro Station to Shin-oomuta Station, and in 2023 and 2024, we used a 100 km section from Shin-yatsushiro Station to Kurume Station. We used DAS from AP Sensing and installed the interrogator at Shin-yatsushiro Station.
In this presentation, we developed a method to estimate the epicenter and magnitude in real time as needed. Additionally, we paid attention to minimizing the computational cost for use in EEW. Yin et al. (2023) showed a magnitude scaling law for strain rate, magnitude, and epicentral distance measured by DAS. We set the parameters of the above the law using the maximum strain rate for 1 second after the P-wave arrival time in the DAS data along the Kyushu Shinkansen, and calculated the magnitude scaling equation in the DAS data. To determine the epicenter in real time, we developed a method to robustly determine the arrival times of the P-wave and S-wave based on STA/LTA. When the number of channels that detected the P-wave exceeded 10, we determined the epicenter using hypomh (Hirata and Matsu'ura, 1987) and used the estimated result as an initial value. Next, if the number of channels that detected the P-wave exceeded 20, we determined the epicenter again using hypomh and updated the initial value. We repeated this process until the number of channels that detected the P-wave exceeded 500.
We applied the developed method to earthquake data with a magnitude of Mj2.0 or higher and an epicentral distance of 150 km or less. Using the data of the P-wave arrival time obtained within 1 second after detecting the P-wave for the first time, we were able to determine the epicenter with an error of several kilometers for earthquakes with an epicentral distance of less than 100 km. However, for earthquakes with an epicentral distance exceeding 100 km, the error in the epicenter determination was large. This is thought to be due to the difficulty in obtaining significant arrival time differences with respect to the observation point configuration (approximately 100 km) as the epicentral distance increases. For earthquakes with an epicentral distance of less than 100 km, we estimated Mdas. We calculated the maximum strain rate for 1 second after the P-wave detection in each channel and substituted it into the magnitude scaling law. As a result, it was found that Mdas could be estimated with an error of about ±0.5.