Early estimation of tsunami heights with a Kalman filter from the collocated ocean-bottom pressure gauges and accelerometers

Abstract:

Dense and wide-span ocean-bottom observation networks such as Dense Oceanfloor Network system for Earthquake and Tsunamis (DONET) and Seafloor observation network for earthquakes and tsunamis along the Japan Trench (S-net) enable us to observe tsunamis at near-fault sites. At these networks, tsunamis are designed to be recorded in terms of water pressure changes. During an earthquake, however, the water pressure is changed not only by tsunamis but also by ocean-bottom displacements, vertical acceleration changes, and radiated seismic waves (e.g., Saito and Tsushima, 2016). To remove these non-tsunami components, we developed a method to extract tsunami and displacement components from coseismic ocean-bottom pressure (OBP) records (Mizutani et al., submitted).

In this study, we developed an alternative scheme to extract tsunami components from near-fault OBP records by taking advantage of the seismometers installed at the same locations as OBP gauges. We combined the records of OBP gauges and accelerometers with a Kalman filter, and estimated coseismic displacement waveforms on the real-time basis. This is analogous to studies on on-land sites by the combination of GNSS and accelerometer records (e.g., Bock et al., 2011). From the comparison of OBP records with the displacement seismograms, derived from the accelerometer records to remove non-tsunami signals, we can estimate tsunami heights by applying a Kalman filter with the selected parameters.

We applied our method to the records of the 2016 Off-Mie earthquake (Mw 6.0) and the 2016 Off-Fukushima earthquake (Mw 7.4). We successfully suppressed displacement components in OBP records (Figure 1). To yield good results, however, we need (1) to select an optimal value as the variance of measurement noise (i.e., parameters of the Kalman filter), and (2) to remove the small offset of accelerometer seismograms (i.e., control inputs of the Kalman filter). For practical tsunami early detection, we may use such optimal parameters before or during an earthquake, based on the performance with the records in the past.