JMA has introduced the tsunami forecast system with a numerical simulation technique to issue quantitative tsunami warnings. Tsunami propagations originating from various types and locations of faults were simulated in advance and the calculated tsunami arrival time and amplitude are stored on the database along with magnitudes and hypocenter locations (presumed epicenters are shown in Figure below). Once an earthquake occurs and its hypocenter and magnitude are determined, the best approximation of tsunami propagation is retrieved from the database according to the earthquake data.

  Tsunami propagation is calculated using the nonlinear long wave theory including effects of Coriolis force [e.g. Satake (2002)]. The nonlinear long wave theory is applied under the assumptions that “the wavelength of tsunami is much longer than the sea depth”.

  Tsunami amplitude at the calculation points is converted into that at a designated point on the coast (“forecast point”) based on the Green’s law.

Green's law

Hh^1/4 = const.  (H:wave amplitude of tsunami, h:sea depth)

Meanwhile, tsunami arrival time at the calculation points, which is obtained by the numerical simulation, is directly applied as that at the forecast points without any conversion. The arrival time is the time when estimated amplitude first exceeds 5 cm.

  It should be noted that actual tsunami arrival time and amplitude may change depending on the coastal topography and the sea bed topography especially near the coasts where fine-mesh bathymetric data are not used in the numerical simulation of tsunamis. Further, the nearest simulation point, which is not always close to the actual epicenter, is chosen to estimate tsunami arrival time and amplitude. Therefore, even though the arrival time of a tsunami at each forecast point is given on a minute time scale, the time does not mean that they are estimated with the accuracy of such a time scale. Tsunamis may arrive at coasts earlier or later than the estimated arrival time in NWPTA messages.