11:00 AM - 1:00 PM
[HDS10-P09] Tsunami source estimation of the 2009 Samoa earthquake using tsunami waveforms observed by ocean-bottom electro-magnetometers
Keywords:Tsunami, The ocean-bottom electro-magnetometers
Interplate earthquakes and outer rise earthquakes often tend to occur in tandem, such as the 1896 and 1933 Sanriku earthquakes, the 2006 and 2007 Kuril Islands earthquakes, and the 2009 Samoa earthquakes. In the case of Sanriku and Kuril Islands, the outer rise earthquakes occurred later than the interplate earthquakes. On the other hand, in 2009 at Samoa, they occurred almost at the same time, with slightly earlier occurrence of the outer rise earthquake. This means that the linkage pattern of the interpolate and the outer rise earthquakes is not unique.
Hossen et al. (2018) performed inversion using the tsunami waveforms of the tide gauges and the offshore observatories (DART) and proposed the tsunami source of the 2009 Samoa earthquakes. However, the reproducibility of the tsunami waveforms observed at the tide gauges were not sufficient. In this study, we added new offshore tsunami waveform data and performed advanced tsunami calculation method to estimate the tsunami source with higher accuracy. The newly added data are the tsunami waveforms converted by Lin et al. (2021) from the magnetic field fluctuation observed by the ocean-bottom electro- magnetometers. In the tsunami calculation, the Earth's elasticity, and seawater density stratification were taken into consideration in the linear frequency dispersive equations. Subfault models used in the inversion were those proposed by Hossen et al. (2018). After calculating the crustal movements caused by the subfaults, the effect of the horizontal displacement of the seafloor slope was added to the vertical displacement, and the Kajiura Filter was applied to obtain the initial water levels. The tsunami propagation was calculated by the finite difference method and Green's functions were generated. The least squares method was used in the inversion. A regularization term was adopted to stabilize the solution The weight (λ) between the data misfit and the penalty function was determined by the reproduction accuracy in the subsequent waves were not used in the analysis.
In the optimal fault model, the root-mean-square error (RMSE) of the observed and calculated waveforms was 0.0076 m. In the fault model of Hossen et al. (2018), RMSE = 0.0184 m, so the reproducibility was improved about 60% by this study. In the fault model of this study, a large slip area was located on a little east of that by Hossen et al. (2018). On the fault plane at the plate boundary, the slip area is compact and narrow in the east-west direction. This is because the slip area was restricted in the east-west direction by the newly added tsunami waveforms converted from the magnetic data on the east side of the epicenter.
Acknowledgments: This study was supported by ERI JURP 2021-S-B103 in Earthquake Research Institute, the University of Tokyo. This research was conducted using the Fujitsu PRIMERGY CX600M1/CX1640M1 (Oakforest-PACS) in the Information Technology Center, the University of Tokyo.
Hossen et al. (2018) performed inversion using the tsunami waveforms of the tide gauges and the offshore observatories (DART) and proposed the tsunami source of the 2009 Samoa earthquakes. However, the reproducibility of the tsunami waveforms observed at the tide gauges were not sufficient. In this study, we added new offshore tsunami waveform data and performed advanced tsunami calculation method to estimate the tsunami source with higher accuracy. The newly added data are the tsunami waveforms converted by Lin et al. (2021) from the magnetic field fluctuation observed by the ocean-bottom electro- magnetometers. In the tsunami calculation, the Earth's elasticity, and seawater density stratification were taken into consideration in the linear frequency dispersive equations. Subfault models used in the inversion were those proposed by Hossen et al. (2018). After calculating the crustal movements caused by the subfaults, the effect of the horizontal displacement of the seafloor slope was added to the vertical displacement, and the Kajiura Filter was applied to obtain the initial water levels. The tsunami propagation was calculated by the finite difference method and Green's functions were generated. The least squares method was used in the inversion. A regularization term was adopted to stabilize the solution The weight (λ) between the data misfit and the penalty function was determined by the reproduction accuracy in the subsequent waves were not used in the analysis.
In the optimal fault model, the root-mean-square error (RMSE) of the observed and calculated waveforms was 0.0076 m. In the fault model of Hossen et al. (2018), RMSE = 0.0184 m, so the reproducibility was improved about 60% by this study. In the fault model of this study, a large slip area was located on a little east of that by Hossen et al. (2018). On the fault plane at the plate boundary, the slip area is compact and narrow in the east-west direction. This is because the slip area was restricted in the east-west direction by the newly added tsunami waveforms converted from the magnetic data on the east side of the epicenter.
Acknowledgments: This study was supported by ERI JURP 2021-S-B103 in Earthquake Research Institute, the University of Tokyo. This research was conducted using the Fujitsu PRIMERGY CX600M1/CX1640M1 (Oakforest-PACS) in the Information Technology Center, the University of Tokyo.