Full-Lifecycle Nowcast of Thunderstorm in eIAWS®

As the civil aviation traffic volume keeps increasing, air traffic delays and diversions caused by convective weather has become a big challenge for the operational air traffic management (ATM). Nowcast of thunderstorm, which provides a couple of hours prediction, is very important to mitigate the impact of thunderstorms on ATM. In the past decades, many nowcast methods based on extrapolation techniques have been applied to predict the location of thunderstorms by estimating the advection of observed storm images. However, prediction of storm evolution at different stages like initialization, developing, maturing, and decaying is still a challenge for aviation weather nowcast. This study explores the full-lifecycle prediction of thunderstorms utilizing new-generation storm observation instruments as well as machine learning techniques. First, thunderstorm observation data from new generation geo-stationery satellite instruments like GOES-R ABI and Himawari AHI is fused with radar observation data to get a comprehensive observation of storm. Secondly, key storm lifecycle parameters are estimated by learning historical data with a machine-learning algorithm. Finally, full-lifecycle evolution of thunderstorms are predicted based on the learning results and simulations. The method is implemented in the Integrated Aviation Weather System (eIAWS^®) and is applied to the thunderstorm cases in continental United States (CONUS) and Asia using GOES-R ABI, Himawari AHI, and dual Doppler radar data, showing the method effectively predicts the full-lifecycle evolution of thunderstorm and has the potential to be used in ATM operations.

Key Words: Nowcast, Convective Weather, Artificial Intelligence