Abstract
Unprecedented tropical cyclones can result in catastrophic devastation due to their unforeseen impacts. This paper conducts an intercomparison of selected approaches to identify plausible unprecedented tropical cyclone scenarios. We review datasets from statistical tropical cyclone track models, hindcast archives from numerical weather prediction models, and a coupled approach of rainfall and flood modelling, comparing how each represents tropical cyclones that would be unprecedented in the historical record. Whilst highlighting the fundamental and incidental advantages and limitations of each dataset, our results demonstrate that these can and should be used to develop diverse scenarios of unprecedented tropical cyclones. We show that the plausible events that fall outside the observational record in these datasets provide a wealth of opportunities to build scenarios of unprecedented tropical cyclone for humanitarian disaster management in a way that would be both scientifically robust and imaginative, going beyond current practice. We recommend greater access and use of these opportunities by disaster risk managers and call for greater collaboration between scientists and practitioners on these questions.