Distortion of the seismic radiation pattern in a long period band: Absence of clear node observations

Abstract:

Sakai et al. (2014, 10th ASC) estimated the source amplitudes in a long-period band of 50 – 100 s using broadband seismic records of earthquakes that occurred in the Philippine and Indonesia, and showed that the source amplitudes fell within a constant band against the moment magnitudes (M_w). Long-period waveforms are thought to be less influenced by structural heterogeneities, and the radiation pattern in a long-period band may not be distorted. However, the results of Sakai et al. (2014) suggest that the nodes are not clearly observed in the long-period band. In this study, we analyzed seismic data from the Japan broadband seismograph network (F-net), which is denser than the networks in the Philippine and Indonesia, and carried out numerical tests using synthetic waveforms to investigate the characteristics of long-period wavefields. We used earthquakes that occurred in Japan with M_w between 4 and 8 since 2003, and estimated their source amplitudes in a long-period band of 50 – 100 s using waveform data from F-net. In each event, we calculated the ratios of the minimum source amplitude to other source amplitudes. We found that the ratios for most of events had values ranging up to roughly 10 regardless of their moment magnitudes and the nodes were also not clearly observed in the F-net data. Using the discrete wavenumber method, we calculated synthetic seismograms assuming uniform station distributions with horizontal intervals of 50 and 100 km as well as the Philippine and F-net station distributions. We assumed an earthquake located in the center of the distributed stations, and systematically changed the dip, strike, and rake angels. We estimated the source amplitudes in the long-period band from the seismograms synthesized with these different fault angles. Our numerical tests indicated that the ratios of the source amplitudes become larger as the density of stations is higher and the ratios depend on the mechanisms. These estimated ratios were larger than the observed ratios. Our numerical results did not reproduce the observed features of the radiation patterns in the long-period band. Our study implies that the radiation pattern may be distorted by structural inhomogeneity and/or source complexity in the long-period band.