Forecast goes astray
Predicting Hurricane Irma creates challenges
The forecast for Hurricane Irma, which correctly predicted a direct hit on Florida about five days in advance, was excellent.
The devil, as always, is in the details — and that was because of the state’s narrow shape.
From Wednesday to Friday of last week, it seemed like Florida’s east coast might see the worst impacts, according to the National Hurricane Center. It wasn’t until very late Friday that the forecast track shifted west, indicating that the eye of the storm would rake up the state’s Gulf Coast.
The forecast track for the core of Irma “had been placed on the west coast of Florida since Friday night at 11 p.m.,” according to Dennis Feltgen, a spokesman at the hurricane center.
The storm made its first landfall in the Florida Keys at 9:10 a.m. ET Sunday, then a second landfall on Marco Island along Florida’s west coast at
3:35 p.m. ET Sunday.
Irma’s path across the Atlantic was primarily steered by a large ridge of high pressure over the northwest Atlantic, according to Weather Underground meteorologist Robert Henson. Irma rode west along the south side of that ridge all the way to Cuba, he said.
“The forecast challenge throughout that time was exactly where the west tip of that ridge would be located, since we knew Irma was going to arc around it and head north,” Henson said.
Exactly how far west that ridge extended was the tough part of the forecast. “On the whole, the track forecasts were excellent: The distance between the west and east coast of Florida isn’t that large, and the entire state was in the hurricane center’s ‘cone of uncertainty’ as Irma approached.”
The main problem is that the biggest track uncertainty — how far west or east the northward track would occur — happened to coincide with the width of the peninsula of Florida.
That’s simply a case of geo- graphic and demographic bad luck, Henson said.
The storm finally made that long-awaited and long-predicted right turn Sunday morning.
Irma continued the comparison over whether the top American or top European weather model is better: For most of the storm, the European model was better, forecaster Ian Livingston said on the Capital Weather Gang, noting it accurately predicted the storm’s interaction with Cuba that caused it to weaken from a Category 5 to a Category 3.
As Irma made its final approach to Florida, U.S. models performed better, Livingston said.
Overall, the hurricane center did an “exceptional job forecasting,” said meteorologist Jeff Masters, also of the Weather Underground. “When you consider that Florida is 90 miles wide, and the average error in a two-day hurricane center track forecast is 90 miles, that makes it tough for evacuation decisions,” he said.
Still, as Henson noted, “Irma makes it clear that even though track forecasts have improved more steadily than intensity forecasts, a small track error can still have big implications.”