Far-field tsunami potential and a real-time forecast system for the Pacific using the inversion method

Abstract

Estimating tsunami potential is an essential part of mitigating tsunami disasters. We proposed a new method to estimate the far-field tsunami potential by assuming fault models on the Pacific Rim. We find that a tsunami that generates in the areas where there is no tsunami in the history can damage the Japanese coast. This shows that it is important to estimate tsunami potential by assuming fault models other than the past earthquake data. Another important activity to mitigate tsunami disasters is to provide appropriate warnings to coastal communities when danger from a tsunami is imminent. We applied a new inversion method using wavelet transform to a part of the real-time tsunami forecast system for the Pacific. Because this inversion method does not require fault location, it is possible to analyze a tsunami in real time without all seismic information. In order to check the usability of the system, a numerical simulation was executed assuming an earthquake at sea off Taiwan. The correlation coefficient for the estimated initial waveform to the assumed one was calculated to be 0.78. It takes 90 min to capture time-series waveform data from tsunami gauges and 5 sec to estimate the 2-D initial waveform using the inversion method. After that, it takes 2 minutes to forecast the tsunami heights at the Japanese coast. Since the sum of these times is less than the 105 minutes transit time of the tsunami from Taiwan to Japan, it is possible to give a warning to the residents before the tsunami attacks the Japanese coast. Comparing the tsunami heights forecasted by this system with those calculated by the fault model, the average error was 0.39 m. The average error of the arrival time was 0.007 min.