Source Constraints and Model Simulation of the December 26, 2004, Indian Ocean Tsunami

The December 26, 2004 tsunami was perhaps the most devastating tsunami in recorded history, causing over 200,000 fatalities and widespread destruction in countries bordering the Indian Ocean. It was generated by the third largest earthquake on record $(M_w=9.1–9.3)$ and was a truly global event, with significant wave action felt around the world. Many measurements of this event were made with seismometers, tide gauges, global positioning system stations, and a few satellite overpasses. There were numerous eyewitness observations and video digital recordings of coastal tsunami impact, as well as subsequent coastal field surveys of runup and flooding. A few ship-based expeditions also took place in the months following the event, to measure and map seafloor disturbances in the epicenter area. Based on these various data sets, recent seismic analysis estimates of rupture propagation speed, and other seismological and geological constraints, we develop a calibrated tsunami source, in terms of coseismic seafloor displacement and rupture timing along $\mathrm{1,200}\mathrm{km}$ of the Andaman–Sunda trench. This source is used to build a numerical model of tsunami generation, propagation, and coastal flooding for the December 26, 2004 event. Frequency dispersion effects having been identified in the deep water tsunami wavetrain, we simulate tsunami propagation and coastal impact with a fully nonlinear and dispersive Boussinesq model (FUNWAVE). The tsunami source is specified in this model as a series of discrete, properly parameterized, dislocation source segments [Okada, 1985, Bull. Seismol. Soc. Am., 75(4), 1135–1154], triggered in a time sequence spanning about $\mathrm{1,200}\mathrm{s}$ . ETOPO2’s bottom bathymetry and land topography are specified in the modeled ocean basin, supplemented by more accurate and denser data in selected coastal areas (e.g., Thailand). A $1\min$ grid is used for tsunami simulations over the Indian Ocean basin, which is fine enough to model tsunami generation and propagation to nearshore areas. Surface elevations simulated in the model agree well, in both amplitude and timing, with measurements at tide gauges, one satellite transect, and ranges of runup values. These results validate our tsunami source and simulations of the December 26, 2004 event and indicate these can be used to conduct more detailed case studies, for specific coastal areas. In fact, part of the development of our proposed source already benefitted from such regional simulations performed on a finer grid $\left(15\mathrm{s}\right)$ , as part of a Thailand case study, in which higher frequency waves could be modeled (Ioualalen et al. 2007, J. Geophys. Res., 122, C07024). Finally, by running a non-dispersive version of FUNWAVE, we estimate dispersive effects on maximum deep water elevations to be more than 20% in some areas. We believe that work such as this, in which we achieve a better understanding through modeling of the catastrophic December 26, 2004 event, will help the scientific community better predict and mitigate any such future disaster. This will be achieved through a combination of forecasting models with adequate warning systems, and proper education of the local populations. Such work must be urgently done in light of the certitude that large, potentially tsunamogenic, earthquakes occur along all similar megathrust faults, with a periodicity of a few centuries.