Detection of Interplate Earthquakes in the Source Area of the 2011 Tohoku Earthquake Using Extensive Seafloor Aftershock Observation Data

Previous studies on the source process of the 2011 Mw 9.0 Tohoku earthquake have revealed its large coseismic slip along a shallow plate interface to the Japan Trench axis. In order to further understand the complex rupture propagation along the plate interface, it is essential to elucidate recovery process of interplate coupling in the source area after the Tohoku earthquake. Estimating changes in b-values for interplate earthquakes before and after the Tohoku event is one of the available approaches to answer the above issue. To start with, we attempt to automatically detect and determine the location of interplate earthquakes using extensive seafloor aftershock observation data. We used mainly short-period pop-up type ocean bottom seismometers (OBSs) [Shinohara et al., 2011, 2012]. We applied a semblance-based method [Nakatani et al., 2015] to 23 OBSs deployed off Fukushima. A seismic tomography result [Matsubara and Obara, 2011] is used for calculation of P wave traveltimes between OBS stations and given grids along the plate interface. To confirm the validity of our method, we conducted synthetic tests by using a Ricker wavelet with several different sets of signal-to-noise (S/N) ratio and focal depths. As the results, semblance values of earthquakes with focal depths relative to the plate interface of 5 km are comparable to noise level, regardless of S/N ratio. On the other hand, earthquakes along the plate interface have significant peak semblance values. Therefore, our method is effective for detection of interplate earthquakes. We, then, applied the method to several waveforms of interplate events listed in the JMA (Japan Meteorological Agency) catalog and identified epicenters by backprojecting semblance values. We compared our resulted epicenters to those of Shinohara et al. (2011, 2012) which precisely relocated the JMA ones using P- and S-wave arrival times and maximum-likelihood estimate technique. The results show good coincidence between them. In addition, we introduced a neighborhood algorithm [Sambridge, 1999] into the semblance-based method in order to reduce computational time. We develop our method so that we determine interplate earthquakes including small ones using continuous waveform data.