Improvements and applications of state_of_the_art numerical models for simulating tsunami hazard

The large recent catastrophic events of Indian Ocean with over 300,000 fatalities in 9 different countries and the 2011 Tohoku tsunami in Japan with about 20,000 fatalities and over $100B damage to the Japanese economy, have shown the limitations of some of the modeling approaches used in the past and have stimulated the development of both new models and novel modeling methodologies. In this thesis, some improvements in tsunami modeling is contributed, in part to better simulate tsunami generation by the co-seismic seafloor displacement caused by megathrust earthquakes like the one in Manuscript 1 (Pure and Applied Geophysics, 170, 1333-1359, 2013), and also model tsunamis generated by Submarine Mass Failures (SMFs) on or near the continental shelf break Manuscript 2 (published online on Natural Hazards, 42pps. Nov. 15th,2014). Besides, when assessing coastal tsunami hazard along simple coastlines, one usually sets the static reference level in tsunami models to the largest astronomic tide, typically with 10% exceedence; one also accounts for a potential sea level rise. However, in complex estuaries such as Chesapeake Bay or New York/Hudson River harbor, the dynamic effects of tidal currents on the incoming tsunami waves could, in some situations, enhance tsunami impact. In the Manuscript 3 (to be submitted) of this thesis, a new approach is implemented to simulate dynamic tide-tsunami interactions and is applied to the full case study of tsunami hazard assessment for the Chesapeake Bay mouth and the James River, which are very vulnerable, low lying, coastal environments equipped with major ports (e.g., Norfolk, Virginia Port authority) and resort areas (e.g., Virginia Beach). In the Manuscript 4 (To be submitted), by applying the new modeling tools together with the most recent bathymetric and geophysical data, we revisit the simulation of perhaps the most significant and damaging SMF tsunami case study in modern history: the 1998 Papua New Guinea (PNG) tsunami, to make a valid benchmark for landslide generated tsunami waves. Accurate tsunami hazard assessment for a specific coastal area requires modeling tsunami impact and inundation from all the possible extreme near- and far-field tsunami sources in a given ocean basin. Tsunami generation from each of these sources must first be performed, which usually requires applying a separate model representing the considered geophysical phenomenon (e.g., subaerial landslide, earthquake, volcanic eruption, submarine mass failure,...). Tsunami propagation to the study site must be conducted. Because of the many spatial and temporal scales involved, this requires using a variety of numerical grids. In this thesis, to perform tsunami propagation and coastal impact simulations with a one-way coupling modeling approach, in a series of nested grids of increasingly fine resolution (with commensurately accurate bathymetric data) is implemented. The tsunami propagation model used here, FUNWAVE-TVD (Shi eta l., 2012; Kirby et al., 2013), is a fully nonlinear and dispersive Boussinesq long wave model that features accurate dissipation by breaking and bottom friction processes. The model has Cartesian and spherical implementations, which are used for simulating nearshore and deep water ocean nested grids, respectively. The model is used to simulate both Submarine Mass Failure tsunami sources and complex co-seismic sources such as for is NHWAVE (Ma et al., 2012), which is a three-dimensional non-hydrostatic sigma-layer model. Coupling between NHWAVE and FUNWAVE will also be applied in the work