Forecasting High-Impact Weather in Landfalling Tropical Cyclones Using a Warn-on-Forecast System

During the 2017 Atlantic hurricane season, several strong tropical cyclones impacted the United States causing significant loss of life and billions of dollars in property damage. While medium to long range track and intensity forecasts have become increasingly skillful over the years, much less emphasis has been placed short-term (0-6 hour) forecasts of high-impact weather events such as tornadoes, localized severe winds, and flash flooding generated within landfalling tropical cyclones. Each of these phenomena occurred, often with multiple hazards impacting the same location, during the landfall of Hurricane Harvey in south TX on 25-26 August and Hurricane Irma in Florida on 10 September 2017. Currently, no operational tool exists to generate rapidly-updating, probabilistic forecasts of these multi-hazard events.

The Warn-on-Forecast (WoF) project aims to generate short-term forecasts of high-impact weather using a rapidly-cycling, ensemble-based data assimilation and forecasting system. A prototype real-time system has been successfully demonstrated for high-impact weather generated by continental convective storms in the Hazardous Weather Testbed during the springs of 2016–18. This study extends the application of this system to tropical cyclones by generating short-term, probabilistic forecasts of precipitation, rotation, and wind speed during and after landfall of Hurricanes Harvey and Irma. Comparison with radar observations and storm reports indicates that the WoF system often identified the areas corresponding to locations specific high impact weather events 1–3 hours prior to their occurrence. Given the success of these initial tests, it is hoped that the system can be applied to future tropical systems in real time and provide coastal forecast offices with a powerful new tool for short-term guidance of the multi-hazard environment of landfalling tropical cyclones.