New Tsunami Forecast Tools for the French Polynesia Tsunami Warning System Part II: Numerical Modelling and Tsunami Height Estimation

Tsunami warning is classically based on two fundamental tools: the first one concerns the source parameters estimations, and the second one is the tsunami amplitude forecast. We presented in the first companion paper how the seismic source parameters are evaluated, and this second article describes the operational aspect and accuracy of the estimation of tsunami height using tsunami numerical modelling on a dedicated supercomputer (2.5 T-flops). The French Polynesian tsunami warning centre developed two new tsunami forecast tools for a tsunami warning context, based on our tsunami propagation numerical model named Taitoko. The first tool, named MERIT, that is very rapid, provides a preliminary forecast distribution of the tsunami amplitude for 30 sites located in French Polynesia in less than 5 min. In this case, the coastal tsunami height distribution is calculated from the numerical simulation of the tsunami amplitude in deep ocean using an empirical transfer function inspired by the Green Law. This method, which does not take into account resonance effects of bays and harbour, is suitable for rapid and first estimation of the tsunami danger. The second method, named COASTER, which uses 21 nested grids of increasing resolutions, gives more information about the coastal tsunami effects about the flow velocities, the arrival time of the maximal amplitude, and the maximal run-up height for five representative sites in 45 min. The historical tsunamis recorded over the last 22 years in French Polynesia have been simulated with these new tools to evaluate the accuracy of these methods. The results of the 23 historical tsunami simulations have been compared to the tide-gauge records of three sites in French Polynesia. The results, which are quite encouraging, shows standard errors of generally less than a 2 factor : the maximal standard error is 0.38 m for the Tahauku Bay of Hiva-Oa (Marquesas islands).