IMPROVEMENT OF REAL-TIME TSUNAMI PREDICTION BASED ON INVERSION METHOD BY USING EARTHQUAKE INFORMATION

To predict the tsunami profiles along coasts precisely and quickly based on the tsunami profiles observed at offshore locations, the present study improves the inversion method by using the location of the epicenter as a priori information. Numerical experiments are conducted around the Nankai Trough. Based on the tsunami profiles observed at 5 offshore locations for 15 minutes, the initial water surface elevation and the tsunami profiles along coasts can be predicted with good accuracy. Parametric studies are also conducted to investigate how the number of offshore observation points and the length of observation time affect the accuracy of prediction. Condition number and Dirichlet spread function are confirmed to be effective for estimating the accuracy of prediction even if the true initial water surface elevation is unknown.