Sub-basin forecasts of seasonal Atlantic hurricane activity: skill and challenges

Abstract Seasonal North Atlantic tropical cyclone (TC) activity is largely regulated by the El Niño-Southern Oscillation (ENSO), Atlantic Meridional Mode (AMM) and Multidecadal Variability (AMV), and associated difference in sea surface temperatures (SSTs) between the Atlantic main development region (10°-20°N, 60°-20°W) and the equatorial Pacific. However, the fluctuating amplitudes and phases of ENSO, AMM, and other climate modulators contribute to a spatial inhomogeneity of seasonal hurricane activity throughout the North Atlantic basin, over which TC impacts may vary significantly. For instance, ENSO is known to modulate Atlantic vertical wind shear more acutely in the western tropical North Atlantic, thus influencing TC development in close proximity to the US. Additionally, Atlantic Niño has recently been shown to enhance the development of powerful TCs in the eastern tropical North Atlantic. Therefore, analysis of sub-basin seasonal TC activity would enable the development of “user-oriented” outlooks (e.g., Gulf of Mexico energy sector) to better inform regional stakeholder interests. While seasonal outlooks of North Atlantic hurricane activity have been available for decades, skilled predictions of sub-basin TC activity have been elusive and are not yet incorporated in official seasonal forecasts. We refine the Pacific-Atlantic interbasin SST relationship to develop seasonal probabilistic outlooks targeting the operational forecast criteria of NOAA’s Seasonal Atlantic Hurricane Outlook (i.e., terciles of below-, near-, and above-normal TC activity) for five specific North Atlantic sub-basins (e.g., Caribbean, Gulf of Mexico, US East Coast, Subtropical North Atlantic and Tropical North Atlantic). We find skill in forecasting seasonal sub-basin accumulated cyclone energy for above- and below-normal seasons using SST predictors from the North American Multi-Model Ensemble, though strong correlations between the sub-basin forecasts limit the predictive usefulness. Overall, there are marginal improvements in forecasting sub-basin seasonal TC activity when compared with a “full-basin” forecast, with the Caribbean achieving the most substantial increase in probabilistic forecast skill.