Probabilistic Tsunami Forecast

While PTHA calculates long term probabilities of tsunami inundation at a given location, Probabilistic Tsunami Hazard Forecasts (or PTF) take us into the realm of Urgent Supercomputing.

We are in a situation where a large earthquake has just occurred somewhere on earth and we wish to compute the likely consequences for far-off near-coastal populations faster than the water waves travel. In the open ocean, tsunami waves travel at about the speed of a jet plane – so it will take about 24 hours for a tsunami generated by an earthquake at one coast of the Pacific Ocean to reach a coastal region on the opposite extreme of the Pacific rim.

However, this is an extreme case of the time constraint – for almost all other combinations of earthquake source and coast, you will have far less time to make a forecast. When a large earthquake first occurs, there is much uncertainty about the actual source. Seismic waves travel faster than water waves, but it can still take many tens of minutes before very accurate information about the source is available. By the time this information is available, it is probably too late to start the tsunami calculation and so, like in PTHA, we run many ensembles – or different scenarios of likely sources.

Unlike in PTHA – where we need to consider every possible tsunamigenic earthquake – in PTF our potential sources are narrowed down at the start by the initial seismic data. However, very large earthquakes are associated with very long ruptures (sometimes of the order of 1000 km for magnitude 9) and the form of the tsunami will vary a lot depending upon the actual pattern of slip.