Variation analysis of multiple tsunami inundation models

Researchers have developed tsunami inundation models based on nonlinear shallow water equations to estimate tsunami propagation and inundation. However, their empirical results are not in perfect agreement with those of other research institutes, even though the same governing equations are used. Therefore, we quantitatively evaluated the variability of tsunami simulations in this study. Several research institutes have conducted tsunami simulations under the same input conditions using tsunami inundation models adopted for tsunami hazard assessment, resulting in a certain degree of variability among them. By examining the spatial and temporal differences in various physical quantities, we identified the characteristic topography where the variability between tsunami simulations increases. A novel method for calculating statistics from the area integrals of physical quantities was proposed to demonstrate the variability in the overall simulation results. In addition, the effects of different setting parameters and computational environments on the simulation results of a single model were evaluated. The findings of this study are expected to not only serve as a basis to verify the reliability of source codes employed by users of the tsunami inundation model, but also contribute useful technical information to advance probabilistic tsunami hazard assessment in the future.