TOWARD REAL-TIME EARTHQUAKE MONITORING AND EARLY WARNING WITH DAS: INSIGHTS FROM THE SHINKANSEN RAILWAY
The Shinkansen, Japan’s high-speed rail system, stands as a symbol of the Japanese technological prowess. Nevertheless, historical data indicates the potential for significant earthquakes to occur in proximity to the Shinkansen railway system, raising concerns among seismologists and JR leaders. This has prompted the development of research aimed at establishing an independent early warning system for the Shinkansen, requiring the establishment of suitable infrastructure for this purpose. Recently, seismologists have explored the Distributed Acoustic Sensing (DAS) technology for earthquake characterization. DAS transforms telecommunications fiber optic cables into seismometers, demonstrating its reliability as a tool for retrieving earthquake information. Recent studies have demonstrated the potential of DAS for early warning purposes, prompting ongoing research in this area. Consequently, this work presents preliminary results suggesting that this knowledge can be effectively applied for earthquake monitoring along the Shinkansen railway.
To accomplish this objective, we conducted an analysis of a specific segment of the JR Tokaido Shinkansen line, spanning from Atami to Tokyo stations. The first step for our research was calibrating body-wave scaling relations. Then, two distinct approaches can be conducted to address the early warning challenge, utilizing the initial 3-second P-wave DAS records: 1) we employ a method to determine the earthquake’s location and magnitude, thereby enabling the prediction of the S-wave peak strain-rate amplitude (PSRA), and 2) we establish empirical relationships for the S-to-P-wave amplitude ratio, which essentially facilitates the prediction of PSRA even without the precise estimation of the earthquake’s location.
Our findings demonstrate that the first approach exhibits remarkable efficacy in accurately recovering the earthquake’s location and magnitude for near-cable earthquakes recorded by DAS, thereby enabling reliable prediction of the PSRA. Conversely, for far-cable events, approach one appears to successfully adjust the P-wave PSRA, although the capability of accurately retrieving the earthquake’s location and magnitude is reduced. This discrepancy between the actual and estimated locations may be attributed to the limited sensitivity of the logarithmic scaling relation in the presence of significant distance-magnitude misfits. Consequently, approach two emerges as a potentially more effective solution for this type of events.
Finally, taking advantage of the relationship between the displacement and strain fields, we propose an empirical relation to convert S-wave PSRA to peak-ground acceleration (PGA). Our findings demonstrate that the correlation between these two quantities is predominantly governed by the near-cable material’s S-wave velocity. Subsequently, once the S-wave PGA is predicted, the potential JMA intensity along the cable can be determined. All these procedures can be completed within the initial 3 seconds following the first P-wave arrival. Consequently, our method can be positioned as a comprehensive early warning procedure.