3:45 PM - 4:00 PM
[SCG55-02] A posteriori estimation of seismic intensity distribution using backpropagation of seismic energy
Keywords:Seismic intensity distribution, seismic wave propagation, real-time analysis
The distribution of seismic intensities should be used to the early response of an earthquake disaster, hence early estimation of it is important. Seismic networks have become denser recently, so simple interpolation of seismic intensity may work well. However, a simple interpolation technique cannot retrieve seismic intensity distribution where there are no observations due to the failure of powers and/or communication lines. The method to estimate seismic intensity distribution is required to be robust for the lack of observations.
In this study, we propose a method based on the backpropagation of seismic energy. Hoshiba and Aoki (2015) proposed the Numerical Shake Prediction scheme. In the scheme, similar to the numerical weather prediction, observations and predictions are combined with the data assimilation technique to estimate the current wavefield, then future ground motion is predicted with the current wavefield and physics of wave propagation. In the proposed method, we first estimate the current wavefield based on the Numerical Shake Prediction scheme till a certain time, then we backpropagate the wavefield till the origin time of the earthquake. In the proposed method, even if some observations near the epicenter could not retrieved, the distribution of seismic intensities might be retrieved from the observations that are far from the epicenter. However, a simple backpropagation unfortunately did not reproduce the past wavefield that was estimated by the Numerical Shake Prediction scheme. It seems that there is a room for the improvement of the proposed method because we are able to adopt more time-consuming method than the Numerical Shake Prediction scheme, which should be calculated as fast as possible.
Acknowledgment
We used seismograms recorded by K-NET and KiK-net operated by the National Research Institute of Earth science and Disaster resilience.
In this study, we propose a method based on the backpropagation of seismic energy. Hoshiba and Aoki (2015) proposed the Numerical Shake Prediction scheme. In the scheme, similar to the numerical weather prediction, observations and predictions are combined with the data assimilation technique to estimate the current wavefield, then future ground motion is predicted with the current wavefield and physics of wave propagation. In the proposed method, we first estimate the current wavefield based on the Numerical Shake Prediction scheme till a certain time, then we backpropagate the wavefield till the origin time of the earthquake. In the proposed method, even if some observations near the epicenter could not retrieved, the distribution of seismic intensities might be retrieved from the observations that are far from the epicenter. However, a simple backpropagation unfortunately did not reproduce the past wavefield that was estimated by the Numerical Shake Prediction scheme. It seems that there is a room for the improvement of the proposed method because we are able to adopt more time-consuming method than the Numerical Shake Prediction scheme, which should be calculated as fast as possible.
Acknowledgment
We used seismograms recorded by K-NET and KiK-net operated by the National Research Institute of Earth science and Disaster resilience.