1:45 PM - 2:00 PM
[SSS03-01] Hypocenter determination using S-net data of shallow earthquakes off Fukushima and Ibaraki prefectures, Japan
★Invited Papers
Keywords:Hypocenter determination, S-net data, Normal fault earthquakes, Hypocenters from before and to after the Tohoku-Oki earthquake
Accurate hypocenter distribution and focal mechanisms of small earthquakes offer essential information about the fault structures and the intraplate stress state. Since the 2011 Mw 9.0 Tohoku-Oki earthquake, normal fault earthquakes have intensely occurred in the upper plate of the northeastern Japan subduction zone. However, it was difficult to capture the details of their spatial distribution due to the absence of nearby stations. Since 2016, seismic data from the S-net stations have become available off the coast of northeast Japan, making it possible to estimate hypocenters of the offshore earthquakes more accurately. Constraining hypocenters accurately using S-net data allows us to recognize systematic bias in travel time that mainly stem from the structural differences between onshore and offshore. Correcting for the travel time residuals may improve the accuracy of hypocenters using only onshore stations for the period even before the S-net data became available. In this study, we determined hypocenters and investigated the spatial distribution for 19 years in off Fukushima and Ibaraki prefectures, where many normal fault earthquakes occurred after the Tohoku-Oki earthquake.
We first estimated source locations, origin times, station correction values for the S-net and land stations, and 1-D velocity structure simultaneously using the method of Kissling et al. (1994). The phase arrival time data were from Suzuki (2022) and Uchida et al. (2023) based on machine learning. We used 1,993 earthquakes of M≧3 from August 15, 2016, to August 31, 2020. For the initial velocity structure, we adopted the JMA2020A model, which was used by the JMA for routine processing of travel time calculations for S-net stations. As a result, hypocenters located in the upper plate were mainly concentrated on a plane, which may represent the fault of the 2016 Mw 7.0 normal fault earthquake. This result suggests that the hypocenters were accurately estimated by using the S-net data. In the land region, backarc stations tend to have positive station correction values representing the delays from the theoretical times, while forearc stations tend to have negative values. This tendency may reflect the influence of the three-dimensional structure beneath the Japanese island.
With the station correction values and the velocity structure thus obtained, it may be possible to improve the accuracy of hypocenters before the S-net data became available. With only land station data, we relocated hypocenters in the offshore area from March 8, 2003, to March 31, 2022. The targets were 8,897 events with M≧3, listed in the JMA unified catalog. The relocated hypocenters were used as the initial hypocenters and re-relocated by the Double-Difference method (Waldhauser & Ellsworth, 2001) using highly accurate arrival time difference data obtained by waveform correlation.
Our re-relocation results allow us to study the temporal changes in the hypocenter distribution in detail since before the Tohoku-Oki earthquake. Before the Tohoku-Oki earthquake, few earthquakes occurred in the upper plate, but immediately after the Mw 9.0 event, seismic activity increased in a wide range at shallower than 30 km depth. One year later, the seismicity deeper than 20 km decreased whereas the the shallow seismicity remained noticeably high. Plotting the NIED MT solutions on our re-relocated hypocenters shows that normal fault type events predominate down to a depth of about 30 km just above the plate boundary after the Tohoku-Oki earthquake. This suggests the normal-fault type stress conditions in this region are not concentrated in the shallow part, but extends to deeper parts. The factors responsible for the formation of the normal fault field are not clear, but it is possible, for example, that gravity effects due to the topography and density structure of the forearc region of the subduction zone are greater than the compressional stress caused by plate convergence.
We first estimated source locations, origin times, station correction values for the S-net and land stations, and 1-D velocity structure simultaneously using the method of Kissling et al. (1994). The phase arrival time data were from Suzuki (2022) and Uchida et al. (2023) based on machine learning. We used 1,993 earthquakes of M≧3 from August 15, 2016, to August 31, 2020. For the initial velocity structure, we adopted the JMA2020A model, which was used by the JMA for routine processing of travel time calculations for S-net stations. As a result, hypocenters located in the upper plate were mainly concentrated on a plane, which may represent the fault of the 2016 Mw 7.0 normal fault earthquake. This result suggests that the hypocenters were accurately estimated by using the S-net data. In the land region, backarc stations tend to have positive station correction values representing the delays from the theoretical times, while forearc stations tend to have negative values. This tendency may reflect the influence of the three-dimensional structure beneath the Japanese island.
With the station correction values and the velocity structure thus obtained, it may be possible to improve the accuracy of hypocenters before the S-net data became available. With only land station data, we relocated hypocenters in the offshore area from March 8, 2003, to March 31, 2022. The targets were 8,897 events with M≧3, listed in the JMA unified catalog. The relocated hypocenters were used as the initial hypocenters and re-relocated by the Double-Difference method (Waldhauser & Ellsworth, 2001) using highly accurate arrival time difference data obtained by waveform correlation.
Our re-relocation results allow us to study the temporal changes in the hypocenter distribution in detail since before the Tohoku-Oki earthquake. Before the Tohoku-Oki earthquake, few earthquakes occurred in the upper plate, but immediately after the Mw 9.0 event, seismic activity increased in a wide range at shallower than 30 km depth. One year later, the seismicity deeper than 20 km decreased whereas the the shallow seismicity remained noticeably high. Plotting the NIED MT solutions on our re-relocated hypocenters shows that normal fault type events predominate down to a depth of about 30 km just above the plate boundary after the Tohoku-Oki earthquake. This suggests the normal-fault type stress conditions in this region are not concentrated in the shallow part, but extends to deeper parts. The factors responsible for the formation of the normal fault field are not clear, but it is possible, for example, that gravity effects due to the topography and density structure of the forearc region of the subduction zone are greater than the compressional stress caused by plate convergence.