Tropical cyclones are highly variable and, in many areas of the world, are the main cause of extreme wind and associated waves, surge and current conditions. At a given location, cyclones that cause a significant impact are relatively rare but severe events, which means that the number of historical events for which data are available is often quite small. In addition, the effects, particularly surge, can be relatively localized and affected by the local bathymetry and topography. This causes considerable difficulty in making quantitative predictions of extreme events for design of offshore or coastal structures in areas affected by tropical cyclones.

A new probabilistic method has been developed to increase the sample of tropical cyclones by producing 10,000 years of synthetic cyclone tracks with a range of paths, intensities and sizes based on Hall and Jewson [1] and Casson and Coles [2]. From this set of synthetic tracks, those tropical cyclones most likely to affect the site of interest are modelled using time-varying wind fields based on the Holland model [3] with surge, current and waves then modelled using the hydrodynamic model TELEMAC-2D coupled to the SWAN wave model. As it is impractical to model 10,000 years of tropical cyclones, a Gaussian process emulator is employed to relate the resultant conditions to parameters defining the cyclones, such as track position, heading, intensity and radius to maximum wind.

The result is a synthesized 10,000 years of cyclone events from which design conditions for a range of return periods can be predicted with a greater degree of certainty than by extrapolating from historical events.

This content is only available via PDF.
You do not currently have access to this content.