Far-Field Tsunami Potential and a Real-Time Forecast System for the Pacific Using the Inversion Method

Estimating tsunami potential is anessential part of mitigating tsunami disasters. Weproposed a new method to estimate the far-fieldtsunami potential by assuming faultmodels on the Pacific Rim. We find thata tsunami that generates in the areas wherethere is no tsunami in the history can damagethe Japanese coast. This shows that it isimportant to estimate tsunami potential byassuming fault models other than the pastearthquake data. Another important activity to mitigate tsunamidisasters is to provide appropriatewarnings to coastal communities when dangerfrom a tsunami is imminent. We applied anew inversion method using wavelet transformto a part of the real-time tsunami forecastsystem for the Pacific. Because this inversionmethod does not require fault location, it ispossible to analyze a tsunami in real timewithout all seismic information. In order tocheck the usability of the system, anumerical simulation was executed assuming anearthquake at sea off Taiwan. The correlationcoefficient for the estimated initialwaveform to the assumed one was calculatedto be 0.78. It takes 90 min to capturetime-series waveform data from tsunamigauges and 5 sec to estimate the 2-D initialwaveform using the inversion method. After that,it takes 2 minutes to forecast thetsunami heights at the Japanese coast. Since thesum of these times is less than the 105minutes transit time of the tsunami fromTaiwan to Japan, it is possible to give a warningto the residents before the tsunami attacksthe Japanese coast. Comparing the tsunamiheights forecasted by this system with thosecalculated by the fault model, the averageerror was 0.39 m. The average error ofthe arrival time was 0.007 min. Copyright Kluwer Academic Publishers 2003