Construction of an Automatic Hypocenter Location System Using Offshore Seismic Network and Distributed Acoustic Sensing Data Using Submarine Cables

Monitoring seismic activity in subduction zones with huge interplate earthquakes is important. In the Nankai Trough, huge earthquakes with magnitudes of --8 have occurred repeatedly, and seismic observation using an offshore seismic network DONET and distributed acoustic sensing (DAS) measurement have been conducted continuously using submarine cables. DAS measures the temporal change in strain rate along a fiber-optic cable with a meter-scale interval; therefore, high-density observation that cannot be achieved by seismometers can be realized. To monitor seismic activity, we are constructing an automatic hypocenter location system using the DAS and DONET data. In this system, P- and S-wave arrivals of earthquakes are picked automatically using a deep neural network model. After grouping picks, the hypocenter is located by a hypocenter location program with a one-dimensional seismic velocity structure model and that using physics-informed neural networks with a three-dimensional seismic velocity structure model. For located events, the difference in epicenters between this study and the catalog by the Japan Meteorological Agency was ~0.1° (~ 10 km) in many events. The difference is reduced by the program using the three-dimensional model. In this paper, we present the latest results of this automatic hypocenter location system.