13:45 〜 14:00
[HDS11-06] Inverse modeling of tsunami source from the 2018 Anak Krakatau eruption and its sensitivity analysis
キーワード：The 2018 Anak Krakatau tsunami, Volcanic tsunami, Inverse modeling, Source characteristics
The volcanic eruption and collapse of Anak Krakatau on 22 December 2018 caused a tsunami that severely impacted and damaged coastal areas in the Sunda Strait. Tsunami waveforms during the event were recorded by four tidal gauges located along coasts of the Sunda Strait. Previous studies utilized inverse modeling with the observed tsunami waveforms to develop source models, reproducing the tsunami at the tide gauge stations. Thoroughly investigating the tsunami source model, which accurately reproduced the tsunami, is essential to establish a correlation between the tsunami and collapse characteristics. However, the number of high-quality observational data for the 2018 event has been limited, and the developed source model has large uncertainties. The objective of this study was to develop a source model for the 2018 tsunami based on an inverse analysis and investigate how the source characteristics varied with different assumptions and numerical conditions.
Green’s functions from unit sources with a Gaussian shape were employed in this study. By linearly superimposing Green’s functions at the tide gauge stations for the observed tsunami waveforms, we developed a static initial sea-surface deformation as a source model for the 2018 tsunami, where the initial velocity components are all zero. Effects of three critical factors were investigated in this study: the origin time of the tsunami generation, the horizontal scale of the Gaussian source, and the time window for the inversion. While the 2018 eruption time has been reported in previous studies, it may not necessarily represent the origin time of the tsunami generation.
Based on results of a checkerboard test, we concluded that the Gaussian scale was reasonable at 0.75 km for our inverse modeling. Subsequently, we developed the source models with different tsunami origin times and time window for the inversion. Our results revealed that reasonable deformations, capable of reproducing observed tsunami waveforms, could be obtained under several scenarios. The developed deformation peaked at approximately 90 m when the origin time and Gaussian scale were assumed at 13:55:00 UTC and 0.75 km, respectively, which was a representative result. We introduced a water volume, calculated as the product of cumulative values of water surface levels at grid points in a zone and the unit area, to evaluate the source model. The absolute values of the volumes, respectively calculated as 0.4 and 0.2 km3 for the uplifted and subsided water surface zones produced by the model, respectively, were consistent with the total collapse volume of Anak Krakatau (0.22 – 0.30 km3 ) at the order of magnitude. However, the developed source model and its characteristics, including the peak value and water volume, varied greatly when different origin time and time window for the inversion were applied. We will present variations in the developed deformation models and discuss their characteristics during our presentation.
Green’s functions from unit sources with a Gaussian shape were employed in this study. By linearly superimposing Green’s functions at the tide gauge stations for the observed tsunami waveforms, we developed a static initial sea-surface deformation as a source model for the 2018 tsunami, where the initial velocity components are all zero. Effects of three critical factors were investigated in this study: the origin time of the tsunami generation, the horizontal scale of the Gaussian source, and the time window for the inversion. While the 2018 eruption time has been reported in previous studies, it may not necessarily represent the origin time of the tsunami generation.
Based on results of a checkerboard test, we concluded that the Gaussian scale was reasonable at 0.75 km for our inverse modeling. Subsequently, we developed the source models with different tsunami origin times and time window for the inversion. Our results revealed that reasonable deformations, capable of reproducing observed tsunami waveforms, could be obtained under several scenarios. The developed deformation peaked at approximately 90 m when the origin time and Gaussian scale were assumed at 13:55:00 UTC and 0.75 km, respectively, which was a representative result. We introduced a water volume, calculated as the product of cumulative values of water surface levels at grid points in a zone and the unit area, to evaluate the source model. The absolute values of the volumes, respectively calculated as 0.4 and 0.2 km3 for the uplifted and subsided water surface zones produced by the model, respectively, were consistent with the total collapse volume of Anak Krakatau (0.22 – 0.30 km3 ) at the order of magnitude. However, the developed source model and its characteristics, including the peak value and water volume, varied greatly when different origin time and time window for the inversion were applied. We will present variations in the developed deformation models and discuss their characteristics during our presentation.