DeepMind’s hurricane breakthrough has surprised weather scientists
Open source WeatherNext model can make accurate predictions with lower-resolution weather data.
In October 2025, a powerful storm emerged over the Caribbean Sea, prompting differing predictions from weather models regarding its path. Would it stay relatively mild and impact Haiti, or would it strengthen and threaten Jamaica? DeepMind's WeatherNext AI model opted for the latter, forecasting a Category 5 hurricane hitting Jamaica with 80% confidence five days before landfall.
This prediction proved accurate, as Hurricane Melissa resulted in devastating flooding and landslides across Jamaica. However, the AI model played a crucial role in giving weather forecasters an earlier warning, allowing communities to better prepare for the impending disaster. Researchers have now published a paper in Nature detailing how WeatherNext AI model can predict cyclones with unparalleled precision.
On average, it provides forecasters an additional day of lead time compared to existing models, meaning its predictions three days in advance are as reliable as previous models' predictions two days out. This extended lead time can make a significant difference on the ground.
Written by urgent.news from Ars Technica's reporting — not their text. Machine-written; read the original for the full account.
This story
This is one outlet's version. Read the fullest account.
