High Resolution Modeling of Tropical Cyclones Using Rare Event Simulation

Abstract:

Tropical cyclones (TCs) present a challenge to modeling using general circulation models (GCMs) because they involve processes and structures that are too fine for GCMs to resolve. TCs have fine structures – e.g. the eye, eyewall, and rain bands – with length scales on the order of 10 km, while GCMs have typical resolutions on the order of 50-100 km. High resolution GCM runs that are sufficiently long to exhibit multiple TCs can be prohibitively computationally expensive. Thus, while GCMs exhibit TC-like vortices with similar spatial and temporal frequencies to observed TCs, the ability of GCMs to reproduce fine TC structures remains largely untested. In this study, we use recently developed rare event analysis and simulation methods to selectively simulate TCs under GCMs at very high resolution. These rare event simulation methods have been developed mostly in the context of computational chemistry, but are broadly applicable. They allow (either by careful manipulation of the model or by selection of trajectories) direct and detailed interrogation of the event of interest without introducing error and without the need to simulated for long periods of time to see the event. By creating targeted, high resolution GCM simulations with many TCs, we hope to determine whether or not GCMs can capture fine TC structures such as eyewalls and individual rain bands.