In 1992 when Hurricane Andrew swirled offshore, a tight ball of wind just becoming a hurricane and still three days away, forecasters plotted its track and drew a landfall-possibility circle in and near Florida that stretched from north of the Georgia border to Cuba’s north coast. At the time, that was the best their models could do.
Twenty-five hurricane seasons later, that map looks a lot different.
With hurricane season starting Thursday, and expected to be busier than usual, National Hurricane Center forecasters have a vastly better chance of predicting a storm’s course. Using Andrew as an example, they took data collected in 1992, compared it to 2016 forecasts and came back with dramatic results. The circle shrank by about 200 miles. If the lethal storm struck today, forecasters and emergency managers could be much more precise in their warnings, and residents much more confident in what they were hearing.
“It really is amazing just how far we’ve come,” said senior forecaster Dan Brown, who modeled the storm to better demonstrate the changes. “The three-day error today is smaller than the one-day back then.”
Sign Up and Save
Get six months of free digital access to the Miami Herald
Andrew leveled more than 100,000 homes, killed 15 people and, for a state with a long dramatic history of lethal storms, became the modern marker of just how bad a hurricane can be. In the years since, forecasting hurricanes has evolved into a high-tech, multi-platform undertaking. Forecasts extend five days, rather than three, and come with a suite of graphics, detailed discussions and a margin of error in track accuracy 65 percent lower.
From the GOES 16 satellite to models that incorporate vast amounts of information streamed on Twitter, Facebook, Instagram, YouTube or Periscope, forecasters have a slew of ways to reach the public with a whole new arsenal of forecast tools. This year, they will begin issuing watches and warnings for storm surge and tropical disturbances that could potentially become a cyclone within the 48 hours of hitting land. In 1992, Andrew’s hurricane warning came just one day before the storm hit.
But some experts worry that so much progress presents another risk: that the public will overlook the uncertainty that remains a part of forecasts.
“There’s a lot more work to be done,” said outgoing hurricane center director Rick Knabb. “It’s going to be years and years, maybe not even in my lifetime, where hurricane forecasts get perfect.”
The biggest gap remains predicting intensity. Five years into a 10-year plan to improve intensity predictions by 50 percent, forecasters are far from their goal. Last year, Matthew caught them off guard when it unexpectedly swirled into a fierce Cat 5 storm in the Caribbean before ripping across Haiti’s southern peninsula. Forecasters also struggle with weak, sloppy storms that lack definition.
“No doubt about it, of all the forecast improvement needs, intensity forecasting is number one and it has been for a long time, with the greatest concern being rapid intensification, especially near land,” Knabb said. “If people go to bed with a Category 1 and wake up with a Category 5 like Matthew did in the Caribbean, then that’s putting preparations at risk.”
With President Donald Trump’s proposed 6 percent cut to the National Weather Service and other cuts for research, including a proposed $318 million for the GOES satellite program, there is also some concern that progress will slow. But Knabb and other atmospheric scientists don’t believe Congress will approve the cuts. And on Thursday, NOAA officials announced that the latest in the series of stationary satellites, the GOES 16, would be posted off the U.S. east coast, with three additional satellites set to launch between 2018 and 2024.
“We’re going to have to continue to advocate for what we need to not only maintain the mission that the hurricane center and the weather service carry out, but what is needed to advance that mission,” he said. “You do that on a year to year basis.”
Emergency managers say this year could also present another unforeseen problem.
“Because we have new products coming out this year, there may be more confusion,” Monroe County Emergency Management Director Martin Senterfitt said this week.
And then there’s the question of expectation. Technology has allowed forecasts to begin much sooner with more accuracy. But forecasts still come with a margin of error, conveyed as a percentage of probability, that usually gets overlooked with the focus on the cone of uncertainty. A quarter century ago, 24-hour cable was still new and the Weather Channel was mostly for science geeks.
“We can’t meet the expectations of people today. The science can’t meet it,” said Bryan Norcross, who became Andrew’s ‘voice in the storm’ after manning his on-air TV post through the hurricane, and whose book recounting the storm was published earlier this month. “In 1992 we could because nobody had any expectation.”
Two days after Andrew formed off the coast of Africa, former huuricane center director Max Mayfield, now a hurricane specialist for WPLG, issued the first advisory for a tropical depression at 11 p.m. seven days before it was poised to hit south Dade. The advisory was an all-caps bulletin marked “FOR INTERGOVERNMENTAL USE ONLY,” warning of more strengthening. Twelve hours later, Andrew was born when forecasters upgraded the cyclone to a tropical storm. For the next three days, it seemed to bobble between weakening and intensifying.
Norcross was keeping a close watch, worried that South Florida was ill-prepared after going so long without a hurricane.
“I was a little squeamish about it,” he said. Conditions weren’t great for Andrew to strengthen, but the location was prime for a Florida hit, he said. “For Miami, that’s where they come from.”
So Norcross kept broadcasting the same message: Get ready for a brutal storm. Even when other TV stations backed off, Norcross kept insisting South Florida prepare.
When a hurricane warning was finally issued, Andrew was only a day away, moving far faster than expected.
“People responded. Sunday the roads were empty. People were in place,” Norcross said. “The bottom line was the forecast was actually excellent because people didn’t expect to know anything until about 24 hours out.”
Follow Jenny Staletovich on Twitter @jenstaletovich