Recently, Hurricane Melissa rapidly intensified into a Category 5 hurricane. During this process, scientists used a powerful new tool—an artificial intelligence (AI) model developed by Google's DeepMind—to successfully predict its path and intensity changes, issuing warnings for Jamaica and other countries severely affected by the storm.

This model was used for the first time this year by the National Hurricane Center (NHC), a primary source of hurricane information. It's a machine learning tool specifically designed for hurricanes, more advanced than earlier models developed by Google DeepMind that accurately predicted weather conditions.

"I've never seen a new model achieve such good results so quickly," said James Franklin, a retired atmospheric scientist who previously headed the NHC's hurricane specialists division.

Brian McNoldy, a hurricane researcher at the University of Miami, stated that compared to other hurricane models used by the NHC, the DeepMind model is "certainly in the leading tier, if not the absolute best."

The developers trained the model on two datasets: a large global database of weather observations and a smaller database containing nearly 5,000 cyclones from the past 45 years. Franklin added that the addition of the second cyclone-focused database is likely the reason why the DeepMind model outperforms other AI models in hurricane forecasting, particularly in the long-standing and challenging area of ​​storm intensity prediction.

The DeepMind team stated that as early as October 21st—before signs of Melissa becoming a major storm emerged—the model predicted a 50% to 60% chance that Melissa would develop into a Category 5 hurricane. By October 23rd, this probability had risen to over 80%. The following day, NHC forecasters cited the model's prediction, stating that "Melissa is very likely to become a Category 5 hurricane."

Nevertheless, DeepMind researcher Ferran Alet said, "We are pleased to provide useful guidance to the NHC, but we caution against relying on a single case or indicator to represent the model's capabilities."

The NHC used the DeepMind model at the start of the Atlantic hurricane season in June. McNoldy said the DeepMind model has generally performed well in predicting the tracks and intensity of the 13 named storms that have formed so far.

It's worth noting that hurricane forecasting is taking place during the U.S. government shutdown. Most federal employees—including those who fly into storms on Hurricane Hunter missions to collect measurement data—have not received pay. Franklin said the shutdown means less scientific research is being conducted on Hurricane Hunter missions. "We missed an opportunity to collect more detailed data."

Melissa is going down in meteorological records. On the morning of October 28th, its central pressure reached 892 millibars, with maximum sustained winds of 298 kilometers per hour, comparable to the strongest Atlantic storm ever recorded. "I've been studying this for a long time, and I can't recall ever seeing a more textbook-perfect hurricane structure," wrote Marshall Shepherd, an atmospheric scientist at the University of Georgia, on social media.

Rising ocean temperatures caused by human-induced climate change are fueling stronger storms in the Atlantic. Preliminary analysis indicates that global warming has increased Melissa's maximum sustained winds by 13 to 16 kilometers per hour compared to its previous potential.

Currently, Hurricane Melissa has moved away from Jamaica, Cuba, Haiti, and other areas, heading towards the Bahamas, where disaster assessment and reconstruction efforts are slowly underway. As of October 29, the hurricane has caused at least 28 deaths.