17:15 〜 19:15
[SSS10-P08] Deconvolution Approach to Estimate Radiation Process of Large (Mw7.0-9.0) Earthquakes
キーワード：破壊過程、大地震、2024年能登半島地震、モーメントレート関数
When large earthquakes (Mw>7.0) occur, researchers immediately begin estimating their rupture processes and slip distributions. While slip distributions can be relatively easily inferred from the spatial pattern of raw geodetic data, interpreting raw seismic data is not straightforward, particularly when multiple reflected phases overlap and Green's functions become complex. Different researchers and analysis methods sometimes yield different results.
This study presents an approach using Green's function deconvolution to readily identify rupture patterns of large earthquakes (Mw6.9-9.0). This approach enables straightforward interpretation of data constraints through differences in apparent moment rate functions (AMRFs) at each station, and has been systematically applied to Mw3.0-7.0 earthquakes by Yoshida & Kanamori (2023) by using empirical Green's functions (eGF). The obtained AMRFs have also been used as waveform inversion data in several studies (Ross et al., 2017; 2018; Yoshida et al., 2020 JGR, 2022 JGR, 2023 JGR, 2023 GRL; Yoshida, 2023 JGR). While this method has been previously applied to regional earthquakes, we now utilize teleseismic data to analyze larger events.
We analyzed 27 earthquakes (Mw6.9-9.0), including 19 Japanese and 8 foreign events. BHZ channel waveforms from the GNS network were obtained through IRIS. We first determined WCMT solutions for each event using WCMT inversion by Duputel et al. (2012), then calculated synthetic waveforms for stations with available data based on these solutions. Apparent Moment Rate Functions (AMRFs) were estimated by deconvolving observed waveforms with synthetic ones. The results showed coherent azimuth-dependent variations for many events. The results showed coherent azimuth-dependent variations for many events, enabling interpretation of their physical meaning.
For example, for the 2024 Noto Peninsula earthquake, the AMRFs reveal that the rupture remained relatively quiet for the first 15 s, followed by a moderate rupture lasting about 10 s, and then two large-amplitude rupture episodes. The azimuthal dependence suggests that the second of the two episodes propagated eastward. These characteristics were consistently observed in both synthetic and empirical Green's function analyses.
In our talk, we will discuss individual results, their characteristics, and features among different earthquakes.
This study presents an approach using Green's function deconvolution to readily identify rupture patterns of large earthquakes (Mw6.9-9.0). This approach enables straightforward interpretation of data constraints through differences in apparent moment rate functions (AMRFs) at each station, and has been systematically applied to Mw3.0-7.0 earthquakes by Yoshida & Kanamori (2023) by using empirical Green's functions (eGF). The obtained AMRFs have also been used as waveform inversion data in several studies (Ross et al., 2017; 2018; Yoshida et al., 2020 JGR, 2022 JGR, 2023 JGR, 2023 GRL; Yoshida, 2023 JGR). While this method has been previously applied to regional earthquakes, we now utilize teleseismic data to analyze larger events.
We analyzed 27 earthquakes (Mw6.9-9.0), including 19 Japanese and 8 foreign events. BHZ channel waveforms from the GNS network were obtained through IRIS. We first determined WCMT solutions for each event using WCMT inversion by Duputel et al. (2012), then calculated synthetic waveforms for stations with available data based on these solutions. Apparent Moment Rate Functions (AMRFs) were estimated by deconvolving observed waveforms with synthetic ones. The results showed coherent azimuth-dependent variations for many events. The results showed coherent azimuth-dependent variations for many events, enabling interpretation of their physical meaning.
For example, for the 2024 Noto Peninsula earthquake, the AMRFs reveal that the rupture remained relatively quiet for the first 15 s, followed by a moderate rupture lasting about 10 s, and then two large-amplitude rupture episodes. The azimuthal dependence suggests that the second of the two episodes propagated eastward. These characteristics were consistently observed in both synthetic and empirical Green's function analyses.
In our talk, we will discuss individual results, their characteristics, and features among different earthquakes.