Probabilistic Translation of Tsunami Inundation Depths across Tides
Published by Geoscience Australia Data
About this dataset
Tides can significantly influence the extent of tsunami inundation. Tsunamis produce a series of waves beginning at some random point in the tidal cycle and the largest inundation typically occurs when the combined tide and tsunami is highest, which is not necessarily the time of the largest tsunami wave. However, tsunami hazard assessments are usually conducted assuming a static tide. Selecting an appropriate static tidal level or set of levels is therefore a key modelling decision. While it is desirable to comprehensively represent tidal uncertainties in hazard assessments, simulating each scenario at many sea levels becomes computationally burdensome. The problem is especially acute for probabilistic inundation hazard assessments, which require inundation modelling for hundreds or thousands of scenarios. A simple and efficient approach to represent different static tides is to run the model at a single high sea level and approximate the solution for other sea levels by reducing the maximum depth by the change in sea level. However, this simple approach does not capture hydrodynamic non-linearities, which may be prevalent near the inundation zone, and in some cases can cause increased tsunami amplification at lower tides. To improve upon this simple treatment of static tide variation, we developed a censored statistical model to translate inundation simulations across tidal levels. This model uses simulations at a high sea level for every scenario, combined with limited simulations at lower sea levels to statistically predict (with uncertainty) changes in tsunami inundation depths. The approach enables hazard assessments to efficiently approximate the effect of different static tidal levels, including due to sea level rise. For example, we can translate results modelled at the highest astronomical tide (HAT) down to the highest tide during the tsunami. While this is still conservative because it convolves over all possible arrival times, it is potentially much less inundation than at HAT. We explore these issues for an onshore probabilistic tsunami hazard assessment of Geraldton in Western Australia, a region prone to tsunamis originating from the Sunda arc, Indonesia.