Atmospheric and radar scientists at Colorado State University are developing a novel forecasting system that relies on artificial intelligence to better detect the approach of destructive hailstorms known for shredding crops, ruining roofing and denting vehicles — thus impacting safety, livelihoods and insurance costs.
Current forecasting models often do not provide hail warnings more than 30 minutes in advance, leaving little time to prepare. By contrast, CSU researchers aim to provide hail forecasts several hours or even days in advance in Colorado and Wyoming.
Their project, funded by the National Science Foundation, links a short-term “nowcasting” model recently developed with AI industry leader NVIDIA to an existing CSU machine learning model for longer-term storm prediction. The combination is expected to deliver accurate forecasts both further in advance and as storms rapidly evolve…