ORLANDO – Forecasters at the National Hurricane Center are not just focused on this season’s storms. They are also rethinking how past hurricanes were measured — and working with new tools that could change how future storms are predicted.
At the National Hurricane Conference in Orlando this spring, Warning Coordination Meteorologist Robbie Berg and Senior Hurricane Specialist John Cangialosi discussed rapid intensification, artificial intelligence, and what a deep dive into the historical record is revealing about storms we thought we already understood.
MORE: Download & Save The Weather Authority’s Hurricane Survival Guide
Are hurricanes getting stronger?
It is a question many people ask — and the honest answer, according to Berg, is that the science is still evolving.
“I don’t think we know for sure,” Berg said. “What we do know is that some storms are strengthening more quickly. We’re having more episodes of rapid intensification, where storms go from a depression or a tropical storm to even a major hurricane in a very short period of time. And that’s dangerous because it gives people less time to get ready.”
A second look at historic storms
The National Hurricane Center has a reanalysis team that uses modern technology to take a fresh look at hurricanes from decades past. The findings have offered a mixed picture.
Cangialosi said some storms have been upgraded after further review, while others have been downgraded.
“Andrew, for example, they upgraded to a Category 5 after the fact,” Cangialosi said. “And they’ve downgraded some, too — some in the Northeast.
He explained that some storms that had been thought to be stronger were not as strong in the new analysis.
“There’s no obvious trend in the reanalysis yet. Some are going up and some are coming down,” Cangialosi said.
AI is improving hurricane forecasts — but it is not infallible
Artificial intelligence is playing a growing role in hurricane forecasting, particularly in predicting rapid intensification. Berg pointed to Hurricane Melissa as an early success story.
“Hurricane Melissa last year down in Jamaica was a perfect example,” Berg said. “We saw that possibility days in advance, partly because of some of the new technologies, artificial intelligence that we’re seeing come online, that helped us make that forecast. Again, we’re not perfect all the time, but we’re getting better and better year after year.”
But Berg also noted that AI has not been reliable in every case.
“There were some storms last year that gave us a little bit of pause in AI. Imelda was an example. It was just off the east coast of Florida,” Berg said. “AI modeling at that time actually didn’t do very well. And so what we have to do is evaluate a lot of storms over a couple of years to see how the model’s going to behave and how much we can actually trust it.”
Human expertise remains essential
Despite AI’s promise, Cangialosi said seasoned forecasters are not being replaced.
“In my opinion, I think AI is going to take us to the next level in advancing our predictions,” Cangialosi said. “I don’t think it’s going to take away from people. You know, it still feels like a little bit of a black box to me in terms of how we use it, because it really helps us figure out the track, and it helps in the intensity, but we can’t assess the environment. We can’t really figure out ahead of time if it’s going to be right or wrong. So the traditional approaches are still there too.”