Based on more than twenty years of NOAA observations, the model reconstructs continuous wind profiles from sparse probe measurements, and can extend surface winds observed by satellite into three dimensions.
Near the ocean surface, the wind of a tropical cyclone changes very quickly with altitude. However, direct measurements remain rare: probes dropped by plane (dropson) only give scattered and irregular points. Researchers, including Robert Rogers and Jun A. Zhang, propose an AI model that combines machine learning and physical constraints to reconstruct continuous wind profiles from these sparse observations.
True-to-surface, realistic at altitude
The "implicit" network respects the winds observed at the surface while generating realistic variations in speed and direction with height. Tested on more than two decades of NOAA hurricane observations, it replicates the vertical structure of winds for a wide range of storm intensities.
The authors see it as a tool for storm surge forecasting, offshore engineering and coastal risk assessment. The model also makes it possible to extend satellite measurements of surface wind into three-dimensional near-surface wind fields.
Read the full article on the original source
Go to the source →