'A Summary of Recent Achievements and Future Strategic Planning for Hurricane Forecast Improvement Program (HFIP)'

The Hurricane Forecast Improvement Program (HFIP), first established in 2007 and revisioned under the Weather Research Forecasting Innovation Act of 2017 (Wx Act) in 2017, has evolved significantly over the years. The HFIP team is currently in the midst of formulating a new strategic plan to set new objectives in 5-year and 10-year timeframes. The purpose of this presentation is to highlight continuity with previous strategic plans, while also expanding upon a slight shift in focus and emphasis on some newer objectives. The original 10-year goals of HFIP to reduce tropical cyclone (TC) model track and intensity errors by 50% and to significantly improve our ability to predict rapid intensification (RI) have been met. Multi-year reforecasts also demonstrate that the new modeling system, Hurricane Analysis and Forecast System version 1 (HAFSv1), meets the refined 5-year goals established in 2017, and is on-track to meet the 10-year goals for reducing track, intensity, and rapid intensification errors.

In the next strategic plan, we still expect HFIP to pursue ambitious advancements in terms of reduction in error for track, intensity, and rapid intensification criteria. However, given the recent paradigm shift associated with the availability of artificial intelligence (AI), machine learning (ML), cloud computing, and graphics processing units (GPUs), we foresee a disproportionately greater potential for advancements in terms of efficiency and speed of NWP forecasting versus during the earlier years of HFIP. This opens up the possibility for much larger ensemble systems, and more advanced bias correction and post processing techniques. We also foresee a greater emphasis on the analysis and prediction of TC structure, since structure has just as significant (and in some cases greater) an impact on TC hazards as does intensity. Inherent in the problem of improving the analysis is the forthcoming shift to JEDI-based data assimilation, and the continued pursuit of assimilating an ever-increasing dataset of available observations. Lastly, the critical importance of how to properly communicate the threat of individual hazards to the public will continue to be explored.