| Popis: |
We explore the potential of the adjoint-state full waveform tsunami inversion method for tsunami warning and source imaging using S-net, an array of ocean bottom pressure gauges. Compared to finite-fault tsunami source inversions, the method we use does not require as densely gridded Green’s functions to obtain a high resolution result, thus reducing computation time. What is required is a dense instrument network with good azimuthal coverage. The S-net pressure gauges fulfill this requirement and reduce the data collection time, thus making it possible to invert the recordings for the tsunami source and issue a timely tsunami warning. We apply our method to synthetic waveforms of the 2011 Mw 9.0 Tohoku earthquake and tsunami as well as data from the 2016 Mw 6.9 Fukushima earthquake. The results of the synthetic tests show that using the first 5 minutes of the waveforms, the adjoint-state inversion method achieves good performance with an average accuracy score of 93%, with the error of predicted wave amplitudes ranging between -5.6 to 1.9 m. Our application to the 2016 Fukushima earthquake shows the required waveform duration to achieve accurate inversions for smaller events is longer than that of a larger event. However, using the first 25 minutes of the waveforms, the inversion yields a tsunami source that is sufficient for making accurate predictions of arrival times and amplitudes. Assuming a uniformly distributed fault slip, we estimated a stress drop for the latter event to be 4.6 MPa, which is in line with estimations from recent studies. |
