[S-CG21] Recent advances and future directions in slow earthquake science

[SCG21-04] A detection method for P and S waves of deep low-frequency earthquakes using a 3D array in the Tokai area and its application to hypocenter determination

*鈴木 貞臣¹、大久保 慎人²、今西 和俊³、武田 直人³ (1.公益財団法人 地震予知総合研究振興会 東濃地震科学研究所、2.高知大学、3.独立行政法人　産業技術総合研究所)

キーワード：deep low-frequency earthquake, 3D array, P- and S-waves, semblance analysis, plate boundary

We have developed a novel method that uses a 3D array to detect the P and S waves of deep low-frequency earthquakes (LFEs) that occur along the subduction zone of the Philippine Sea plate in southwest Japan. Obtaining accurate hypocenters of LFEs is very difficult because their seismic waves are characterized by low amplitude and the absence of sharp pulses. In particular, identifying P phase arrivals is not readily possible using conventional methods and seismic networks. To determine their hypocenters accurately-not only their epicenters but also their depths-we tried to find their P and S wave pairs and obtain S-P times. We constructed a 3D array (6 km x 4 km area, see Fig.1) using 14 seismic stations in the Tokai area with three component seismographs, including ones with deep (600 m at the deepest) borehole seismographs. We observed remarkable LFE activity occurring in the Tokai area over November 10-30, 2010. We successfully detected not only S waves but also very weak P waves of LFEs using the 3D array data and the semblance method. Assuming a homogeneous half-space model with P wave velocity=4.5 km/s and S wave velocity=2.2 km/s in the 3D array, we calculated the semblance distributions for more than 20 LFEs to obtain their propagation parameters (back azimuth and the incident angle of seismic waves) and to identify P and S-waves. Using the time of the maximum value of the semblance in each component, we detected the direct P wave in the vertical component and the S wave in the horizontal component, providing S-P time. Fig.2 shows an example of hypocenter determination (red star) using estimated S-P time and propagation parameters, where we found 8.2 km difference in depth between the hypocenter in this study and that (green star) listed in Japan Meteorological Agency (JMA) catalogue. This example suggests that the inclusion of the S-P time strongly reduces the uncertainty on source depth, because the LFEs in the JMA catalogue were generally located using only S-arrival times. Choosing 4 LFEs with reliable results obtained from the semblance analysis, we located their hypocenters and found they distribute in the depth range from 28 km to 35 km approximately along the plate interface inclining in depth from 30 km to 32 km.Because a single array inherently has a limitation in the precise location estimate, especially for epicenter, we also tried to locate hypocenters (for example, a blue star in Fig.2) of LFEs using 3D array data together with arrival times (in many cases, S-arrivals) of surrounding stations that listed in JMA catalogue. For the LFE in Fig.2, we found 4.5 km difference (between a blue star and a red star) in the epicenter by combining the arrival times of surrounding stations, which is not always negligible for better understanding the spatial and temporal distribution of LFEs.