5:15 PM - 7:15 PM
[SSS11-P23] Aiming to automate the analysis of earthquake source processes using near-field data
Keywords:source process analysis, automation
Results by source process analyses (or finite fault models) are more useful than an epicenter or a centroid moment tensor (CMT) solution because they provide more detailed information about the slip areas etc. However, it depends on the accuracy of the input data and the degree of freedom of parameters, and it is often difficult to obtain a reasonable optimal solution early (e.g., Yagi (2009 Zisin)). On the other hand, from the viewpoint of contributing to a more rapid interpretation of earthquakes that have occurred, it would be useful if only limited elements such as the horizontal position of the main slip area could be obtained redundantly and automatically. Research aimed at obtaining analysis results automatically includes, for example, Fujita et al. (2017 JpGU), which uses teleseismic wave data. In this study, we aim to automatically obtain analysis results using near-field seismic wave data from strong motion observation networks, etc., which allow for faster analysis because seismic waves arrive earlier, and we consider the analysis results and issues extracted when aiming for automation.
The method is almost the same as the analysis of near-field data by Iwakiri et al. (2014). The hypocenter and CMT solution analyzed after the event were used to determine the rupture starting point and the fault plane, which was placed on a single plane. Errors that may occur in immediate analysis were ignored here. In addition, parameters from previous analyses (Japan Meteorological Agency website: https://www.data.jma.go.jp/eqev/data/sourceprocess/index.html) were referenced, and the size of small faults, etc. were determined by creating an approximation equation for Mw. For parameters with poor coefficients of determination for the approximate equations, only the range of values was referenced. Koketsu et al. (2012) was referenced for creating the velocity structure.
As an example, for the Hyuga-nada (M6.6) earthquake on January 13, 2025, the analysis by the Japan Meteorological Agency (2025 Earthquake Research Committee https://www.static.jishin.go.jp/resource/monthly/2025/2025_01.pdf p.44) was considered to be the true value, and the results of this manual analysis were compared with the automated analysis results. As a result, it was found that the automated analysis was able to obtain the same characteristic as the manual analysis, that the main slip area was located to the northeast of the epicenter. We plan to evaluate the performance of this analysis in other cases in the future.
Acknowledgements: We used K-NET and KiK-net data from the National Research Institute for Earth Science and Disaster Resilience (2019 doi:10.17598/NIED.0004).
The method is almost the same as the analysis of near-field data by Iwakiri et al. (2014). The hypocenter and CMT solution analyzed after the event were used to determine the rupture starting point and the fault plane, which was placed on a single plane. Errors that may occur in immediate analysis were ignored here. In addition, parameters from previous analyses (Japan Meteorological Agency website: https://www.data.jma.go.jp/eqev/data/sourceprocess/index.html) were referenced, and the size of small faults, etc. were determined by creating an approximation equation for Mw. For parameters with poor coefficients of determination for the approximate equations, only the range of values was referenced. Koketsu et al. (2012) was referenced for creating the velocity structure.
As an example, for the Hyuga-nada (M6.6) earthquake on January 13, 2025, the analysis by the Japan Meteorological Agency (2025 Earthquake Research Committee https://www.static.jishin.go.jp/resource/monthly/2025/2025_01.pdf p.44) was considered to be the true value, and the results of this manual analysis were compared with the automated analysis results. As a result, it was found that the automated analysis was able to obtain the same characteristic as the manual analysis, that the main slip area was located to the northeast of the epicenter. We plan to evaluate the performance of this analysis in other cases in the future.
Acknowledgements: We used K-NET and KiK-net data from the National Research Institute for Earth Science and Disaster Resilience (2019 doi:10.17598/NIED.0004).