Analysis of atlantic tropical cyclone landfall forecasts in coupled GCMs on seasonal and decadal timescales

Abstract

In this chapter we present advances in forecasting Atlantic tropical cyclone (TC) landfall statistics at both seasonal and multi-annual timescales using coupled global climate models. First, we demonstrate potential for forecasting TC landfall frequency on seasonal timescales using the Met Office seasonal forecast system, GloSea5, in some regions: statistically significant skill is found in the Caribbean and moderate skill is found for Florida. In contrast, low skill is found along the US Coast as a whole. We show that the skill over the Caribbean is likely due to a good model response to El Niño-Southern Oscillation (ENSO) forcing. Lack of skill along the US Coast may be due to a weaker influence from ENSO compounded by a low bias in model storm tracks crossing the US coastline. Secondly, we demonstrate that it is possible to construct reliable 4-year mean forecasts of landfalling hurricane numbers in the Atlantic using initialised global climate models to predict an index that relies on subpolar gyre temperature and subtropical sea level pressure, two quantities with links to hurricane activity. Furthermore, we give evidence that the forecast system anticipates large changes in at least one of the two components of this index, which suggests that the technique could be used to forecast shifts between active and inactive regimes of hurricane activity in the Atlantic.