Forecasters at the National Hurricane Center made a major bet when Hurricane Isaias first formed. Rather than relying mainly on classic physics-based models such as the European and GFS, they shaped their predictions to closely match an AI hurricane model developed by Google DeepMind. So far, the bet has paid off, reflecting the improving reliability of AI weather models and a rapid shift toward their routine use by human forecasters.
Relying heavily on DeepMind allowed the NHC to forecast Isaias as a Category 2 hurricane from its first advisory on Tuesday, before it even had a name. At that point, many physics-based models, which use mathematical equations describing atmospheric behavior, were wavering between projections and generally forecasting a weaker storm on a different track. If the forecast holds, it will be a significant win for Google and give people in the storm's path more time to prepare for a much stronger blow.
Isaias is the first hurricane of the tropical Atlantic season, largely due to a super El Niño in the Pacific, and it set a record for the latest first hurricane of the season in that ocean basin. NHC confidence in DeepMind, whose newest version is Weather Next 3, comes from experience. Last season, the Hurricane Center correctly anticipated the rapid intensification of Hurricane Melissa almost three days before it devastated Jamaica, based largely on DeepMind projections. The agency also partnered with Google to help train the model.
NHC director Michael Brennan said DeepMind performed extremely well last season and has been accurate in anticipating the intensity of many Pacific storms this season. He noted that the AI models have shown more consistency from one run to the next, and that their reliability has proven strong so far. Meteorologists still use AI alongside traditional physics models, which retain their own advantages.
Source: New York Times Top Stories · Summarized by HeadlinesBriefing