Abstract
In order to solve the problems of the inversion methods commonly used in the field of real-time tsunami prediction using the observation data of tsunami gauges, the ensemble Kalman filter method which is one of the data assimilation methods is applied to this filed. We tested its effectiveness by twin experiment in the area where DONET2 of JAMSTEC is set up. As a case study, using the case (3) model which the Cabinet Office supposes as earthquake tsunami model in Nankai trough, numerical calculation is executed in this study. As a result, it is clear that using the ensemble Kalman filter method, tsunami heights in the coast can be predicted in real time with good accuracy if the data are observed at tsunami gauges for 20 minutes. However, as long as DONET2 tsunami gauges are used, it is impossible to predict tsunami heights with good accuracy before arrival of tsunamis in the south of Wakayama and Tokushima Prefecture.
