17:15 〜 18:30
[SCG53-P01] 津波増幅率による津波即時予測の高精度化
キーワード：Realtime tsunami prediction、robustness
We have developed a realtime tsunami prediction system based on the tsunami amplification. The system predicts tsunami arrival times, the heights, inundation area and inundation depth for each target points in turn, in proportion to observed pressure values on observatories of the oceanfloor network like Dense Oceanfloor Network system for Earthquakes and Tsunamis (DONET). The prediction information is extracted from a database of the system for the observatories, which composed of calculated waveforms using many fault models changing the magnitude, depth, dip angle and strike. The merits are simplified system using only one PC playing a roles data transmission, making prediction information and visualization, easy revise even though coastline changes, and robustness of the prediction under the local anomalies of tsunami by landslides on the seafloor. We implemented it on local government of Wakayama and Mie prefectures and an electric power company using DONET. Recently, it was implemented on Chiba prefecture using Seafloor observation network for earthquakes and tsunamis along the Japan trench (S-net).
The error of the prediction is brought by following items, which are noises on observatories of the oceanfloor network, complicated tsunami propagation for the local topography, and contamination of the crustal displacement on the observatories. S-net data sometimes includes large offset for the strong motion due to rotation and unstable posture of the sensor. Sometimes observatories subside by liquefaction on seafloor. We resolved this noise issue statistically by comparison of the observed data among the triggered sensor and dismissal of abnormal data. Second issue of dispersion of the tsunami height due to complicated tsunami propagation are resolved by specification of fault models using detected direction of the source. For the last issue of contamination of the composition of crustal displacement, we introduce distribution of the observed crustal displacement on the oceanfloor network and omit fault models with large difference of the composition of crustal displacement. In particular, The crustal displacement around the prediction target point brings large dispersion on the inundation areas. We realized the high precision of the tsunami prediction by devise on settings and extraction of the fault models.
The error of the prediction is brought by following items, which are noises on observatories of the oceanfloor network, complicated tsunami propagation for the local topography, and contamination of the crustal displacement on the observatories. S-net data sometimes includes large offset for the strong motion due to rotation and unstable posture of the sensor. Sometimes observatories subside by liquefaction on seafloor. We resolved this noise issue statistically by comparison of the observed data among the triggered sensor and dismissal of abnormal data. Second issue of dispersion of the tsunami height due to complicated tsunami propagation are resolved by specification of fault models using detected direction of the source. For the last issue of contamination of the composition of crustal displacement, we introduce distribution of the observed crustal displacement on the oceanfloor network and omit fault models with large difference of the composition of crustal displacement. In particular, The crustal displacement around the prediction target point brings large dispersion on the inundation areas. We realized the high precision of the tsunami prediction by devise on settings and extraction of the fault models.