The 2023 SSJ Fall Meeting

Presentation information

Room A

Regular session » S09. Statistical seismology and underlying physical processes

[S09] PM-2

Wed. Nov 1, 2023 3:15 PM - 4:45 PM Room A (F205+206)

chairperson:Katsuhiko Shiomi, Satoru Baba

3:30 PM - 3:45 PM

[S09-04] Seismic activity around plate boundary near westernmost Nankai trough revealed by ocean bottom seismometer observation

*Chingyu Hu1,2, Masanao Shinohara2, Yusuke Yamashita3, Tomoaki Yamada2, Takeshi Akuhara2, Kimihiro Mochizuki2 (1. Graduate School of Science, The University of Tokyo, 2. Earthquake Research Institute, The University of Tokyo, 3. Disaster Prevention Research Institute, Kyoto University)

The Nankai trough region has numerous historically devastating earthquakes. However, a earthquake greater than magnitude 8 has not been recorded in the Hyuga-nada region covering the westernmost area of the Nankai trough. A lack of a large earthquake may be attributed to a coupling between the subducting Philippine Sea plate and a landward plate. It is widely known that low-frequency tremors and very low frequency earthquakes frequently occur in the region. Accurate distribution of normal earthquakes is one of important information for understanding the plate subduction in the study region. To gain insights into it, several Long Term Ocean Bottom Seismometer (LTOBS) networks were deployed to observe seismic activities. This study focuses on the analyses of three datasets among these observations in 2015-2016, 2017-2018 and 2022, respectively.
The very first step for data processing is to extract three minutes of the waveform to include the event from collected raw data. Since that raw data were continuous time series waveforms, establishing a reference for event detection was essential. The Japan Meteorological Agency (JMA) provided an earthquake catalog based on terrestrial observations, which included date and time information. Utilizing this catalog, we identified origin times from the JMA catalog, setting a reference three seconds prior to origin time for extracting the seismic waveforms. After cutting out the waveform, we manually picked up P-wave and S-wave arrivals, S-wave converted S-P wave arrivals, first arrival polarity, and maximum amplitude. P- and S-wave arrivals and maximum amplitude were employed in a location program that integrated absolute travel times and a 1-D velocity structure. During this process, station corrections were applied to compensate for structural heterogeneity beneath the LTOBSs, and station corrections were the calculation result of difference between S-wave arrival and S-P arrival, P-wave velocity from initial velocity model and assumed S-wave velocity in sediment layer, etc. Subsequently, we relocated the events by a program which determines hypocenters and 3-D velocity structure simultaneously using a double-difference technique, to enhance the location accuracy. Furthermore, data on first arrival polarity were inputted into the computer program to infer focal mechanisms.
The first dataset was collected by LTOBSs deployed from January 9, 2015, to January 6, 2016, in the westernmost part of the Nankai trough. The relocated result revealed that the earthquakes occurred within the Philippine Sea Plate. Although some earthquakes had thrust-fault type focal mechanisms, the dominant focal mechanisms were normal-fault and/or strike-slip types.
The second dataset was obtained by LTOBSs in the northernmost part of the Ryukyu trench adjacent to the first analysis, and we processed data from July 25, 2017, to September 30, 2018. According to the result of this dataset, we found that most events concentrated in small regions and occurred during limited time periods. These hypocenters were also positioned within the subducting plate. The focal mechanisms of the events were predominantly strike-slip and/or normal faulting.
The third dataset was collected by LTOBSs in the same area to the first analysis, which deployed in 2021 and 2022. We then selected the 13 events occurred within the network between January 9, 2022, to August 13, 2022. Since that only 13 events could be selected to be processed, the double-difference technique has limitations due to event pair scarcity. We obtained hypocenter distribution by applying to a location program utilizing absolute travel times. As the result, it shows similar to the results from observation from 2015 to 2016.