Data assimilation with dispersive tsunami model: a test for the Nankai Trough

We present a method of tsunami data assimilation using a linear dispersive model in order to provide an accurate tsunami early warning. To speed up the assimilation process, we use the Green's function-based tsunami data assimilation, in which the Green's functions are calculated in advance with linear dispersive tsunami propagation models. We demonstrate a test case in the Nankai Trough off southwest Japan, with a source model similar to the main shock of the 2004 off the Kii Peninsula earthquake (M7.4) which generated tsunamis with dispersive characteristics. We show that assimilation of existing ocean bottom pressure gauge data can rapidly forecast the tsunami arrival time and the maximum height of the first tsunami peak along the coast of Shikoku and Kyushu Islands. Both the linear long-wave model and the linear dispersive model can accurately forecast the tsunami height, but the linear dispersive model can predict the tsunami arrival time more accurately for the tested earthquake. .