2.2 Estimation of focal mechanisms
We estimated the focal mechanisms based on the amplitude ratios of the waveforms using the method of Yoshida et al. (2019b), which is similar to that of Dahm (1996). We used six focal mechanisms determined by the JMA (Fig. 1b) to represent effects of the path and site on the waveform. We determined the focal mechanisms of 161 earthquakes with ML ≥ 2. We used displacement waveforms obtained by integrating the velocity waveforms recorded at the 20 stations (green triangles in Fig. 1a) surrounding the hypocenters. The vertical component was used for the analysis of the P-wave, whereas radial and transverse components were used for that of the S-wave. We applied a 2–5 Hz bandpass filter to the waveforms, cutting them out with time windows of 2.8 s for P-waves and 4.3 s for S-waves starting 0.3 s before their arrival.
We used waveform cross-correlations to measure the amplitude ratios between target and reference events. The amplitude ratios were obtained for pairs with absolute correlation coefficients above 0.75. We used principal component analysis (PCA) to measure the amplitude ratios.
We only estimated the mechanism solutions when amplitude ratios were obtained for more than 20 channels. We eliminated the results when the Variance Reduction (VR) was below 80:
\(\text{VR}=\ \left(1-\frac{\sum_{k=1}^{n}\left(d_{k}-s_{k}\right)^{2}}{\sum_{k=1}^{n}d_{k}^{2}}\right)\ \bullet 100,\)(1)
where \(d_{k}\) and \(s_{k}\) are the observed and calculated displacement amplitude ratios, respectively, at channel \(k\).