As the Atlantic hurricane season comes to an end, the most destructive storm of the year was Hurricane Melissa. A Category 5 hurricane devastated Jamaica in late October. It was the strongest hurricane ever to hit Jamaica, killing dozens of people and destroying many neighborhoods. There was a forecast of uncertainty in the days before the landslide. But one particular model got it exactly right.
Ricardo Makin/AFP via Getty Images
hide caption
toggle caption
Ricardo Makin/AFP via Getty Images
MIAMI – The Atlantic hurricane season, which officially ends Sunday, meets predictions that it will be an active year.
There were 13 named storms and three Category 5 hurricanes. But, for the first time in a decade, no hurricane hit America
The most destructive hurricane of the season, Melissa, was one of the most powerful Atlantic hurricanes ever recorded. It struck Jamaica with winds of up to 185 mph, devastating communities and killing dozens.
However, a week before the storm’s arrival, forecast models disagreed on where it would move. The one model that got it right — accurately predicting Melissa’s path and its Category 5 intensity — was a new one: Google’s DeepMind AI-based hurricane model.
James Franklin, former branch chief of the National Hurricane Center, analyzed how forecasting models performed this year, and says Google’s DeepMind outperformed them all. “The model performed very, very well, which was very impressive,” he says. “This was the best guidance we’ve seen this year.”
Artificial intelligence has been used in weather forecasting models for some time. However, Google’s DeepMind is an important step forward, suggesting that AI may soon overtake the physics-based models that meteorologists have long relied on.
Models like the Global Forecast System – GFS – developed by NOAA are based on equations that calculate how air, moisture, and heat move around in the atmosphere. Models use these equations to predict what might happen in the atmosphere, including a storm’s track and intensity.
Track errors for the 2025 Atlantic hurricane season from the most common forecast models and the NHC official forecast (OFCL – black line). Google DeepMind (GDMI – red line) had the lowest overall prediction error.
james franklin
hide caption
toggle caption
james franklin
On the other hand, AI models like Google’s DeepMind know nothing about physics, but instead focus on history. “These were developed to go back and look at historical records and tease out patterns and relationships in very subtle ways about what’s happened in the past that a person can’t figure out on their own,” says Franklin.
To develop their hurricane model, Google engineers worked closely with scientists at the National Hurricane Center and the Cooperative Institute for Research in the Atmosphere at Colorado State University. Kate Musgrave, a research scientist at CIRA, analyzes the performance of AI-based models, including one developed by Google.
She says that in the past, AI models have done well with one part of hurricane forecasting – tracking the storm’s path, “because it is controlled by large-scale effects in the atmosphere. However, the intensity, how strong a storm will be, is not captured well in AI models.” But the Google model performed very well at predicting intensity, she says, because it combined historical data about how past storms had developed.
Musgrave believes AI modeling could be the future in predicting not only hurricanes, but other weather events, everything from tornadoes to cold snaps.
As far as hurricanes are concerned, as AI models develop, he believes meteorologists will be able to predict the track and intensity of hurricanes much earlier than ever before, which is a significant improvement. “As the population on the beach increases, we need more and more time to move people out of the way. Therefore, predictions about the future become more important,” she says.
People walk through a flooded road after Hurricane Melissa in Petit-Goave, Haiti on October 30, 2025. The devastating Category 5 hurricane, with sustained winds of 185 mph, killed dozens of people across the Caribbean.
Clarence Sifroy/AFP via Getty Images
hide caption
toggle caption
Clarence Sifroy/AFP via Getty Images
The National Hurricane Center has adopted the new Google DeepMind model, and references it in many of its forecast discussions, particularly when tracking Hurricane Melissa.
“I think it’s clear at this point that AI will be a component of the hurricane forecasting process going forward,” says Wallace Hogsett, science operations officer for the National Hurricane Center. Additional AI models are being developed by NOAA and the European Center for Medium-Range Weather Forecasts. “I hope this pace of innovation continues,” he says.
But former NHC forecaster Franklin says relying on AI can be unsettling for forecasters accustomed to looking at data on wind, pressure, humidity and sea surface temperature and seeing how it is interpreted by physics-based models.
“AI models are like black boxes for a forecaster,” he says. “A lot of data goes in. You get a forecast that comes out. But you don’t really know how it came out.”
Although AI models will become increasingly important, Franklin and Musgrave do not expect them to replace long-standing physics-based models or the judgment of experienced forecasters.
<a href