The Atlantic hurricane season is coming to an end, with Hurricane Melissa becoming the most destructive storm of the year. A Category 5 hurricane hit Jamaica in late October. It was the strongest storm to ever hit Jamaica, killing dozens of people and destroying many neighborhoods. In the days leading up to landfall, forecasts were uncertain. But one particular model got it exactly right.
Ricardo Makin/AFP via Getty Images
hide signature
switch signature
Ricardo Makin/AFP via Getty Images
MIAMI – The Atlantic hurricane season, which comes to an official end on Sunday, predictions that came true it will be an active year.
There were 13 named storms and three Category 5 hurricanes recorded. But for the first time in a decade, a hurricane did not make landfall in the United States.
Season most destructive hurricaneMelissa was one of the strongest Atlantic storms ever recorded. It hit Jamaica with winds of 185 mph. destructive communities and killing dozens of people.
However, in the week before the hurricane made landfall, forecast models disagreed on where it would go. One model that got it right—accurately predicting Melissa's path and its Category 5 intensity—was new: Google's DeepMind Hurricane model based on artificial intelligence.
James Franklin, a former branch manager at the National Hurricane Center, analyzed how the forecasts are modeled. performed this yearand says Google's DeepMind has eclipsed them all. “The model performed very, very well, which was very impressive,” he says. “This was the best leadership we have seen this year.”
Artificial intelligence has been used in weather forecasting models for some time. However, Google's DeepMind marks a significant step forward, suggesting that AI could soon overtake the physics-based models that meteorologists have long relied on.
Models such as Global Forecast System — NOAA's GFS is based on equations that calculate how wind, moisture and heat move through the atmosphere. Models use these equations to predict what might happen in the atmosphere, including the direction and intensity of hurricanes.
Track errors for the 2025 Atlantic hurricane season based on the most common forecast models and the NHC's official forecast (OFCL – black line). Google DeepMind (GDMI – red line) showed the lowest overall forecast error.
James Franklin
hide signature
switch signature
James Franklin
On the other hand, artificial intelligence models such as Google's DeepMind know nothing about physics and instead focus on history. “They were designed to go back and look at historical records and identify patterns and relationships in what happened in the past in very subtle ways that a person couldn't figure out on their own,” Franklin says.
To develop its hurricane model, Google engineers worked closely with scientists from the National Hurricane Center and Colorado State University. Cooperative Institute for Atmospheric Research. Keith Musgrave, a research fellow at CIRA, analyzes the performance of artificial intelligence models, including one developed by Google.
In the past, she said, AI models have done a good job at one part of hurricane forecasting—tracking the storm's path: “Because that's determined by large-scale impacts in the atmosphere. However, the intensity and how severe the hurricane will be is not well reflected by AI models.” But she said Google's model did a very good job of predicting intensity because it added historical data detailing how hurricanes have progressed in the past.
Musgrave believes AI modeling could be the future not only for hurricanes, but also for forecasting other weather events, from tornadoes to cold snaps.
As for hurricanes, as artificial intelligence models advance, she believes meteorologists will be able to predict the trajectory and intensity of hurricanes earlier than ever before, a vital improvement. She says: “As coastlines become more populated, it takes us longer and longer to get people out of the way. Thus, forecasts for the future become increasingly important.”
People walk along a flooded street after Hurricane Melissa in Petit Goave, Haiti, October 30, 2025. The storm, a destructive Category 5 with sustained winds of 185 mph, killed dozens of people across the Caribbean.
Clarence Siffroy/AFP via Getty Images
hide signature
switch signature
Clarence Siffroy/AFP via Getty Images
The National Hurricane Center has adopted Google's new DeepMind model, citing it in many of its discussion of forecastsespecially since he was tracking Hurricane Melissa.
Wallace Hogsett, a research scientist at the National Hurricane Center, says, “I think it's clear at this point that AI will be a component of the hurricane forecasting process in the future.” Additional artificial intelligence models are being developed by NOAA and the European Center for Medium-Range Weather Forecasts. He says, “I expect this pace of innovation to continue.”
But for forecasters accustomed to seeing wind, pressure, humidity and sea surface temperature data and how they are interpreted by physical models, says former NHC forecaster Franklin, relying on AI can be unsettling.
“AI models are something of a black box for a forecaster,” he says. “There’s a lot of data coming in. You get the forecast that comes out. But you don't really know how he got there.”
While artificial intelligence models will become increasingly important, Franklin and Musgrave don't expect them to replace long-standing physics-based models or the judgment of expert forecasters.
