Hurricane Forecasting Upgraded, but Public Doesn’t Quite Get It

Hurricane forecasting models have developed steadily in the last decade, and increased data from aircraft over the last three years has greatly improved forecasting. But getting the message to the public is a challenge.

by Jim McKay / September 27, 2019

Hurricane forecasting has improved markedly over the last 10 years, and especially the last three as improved models have benefited from much more data.

That allows forecasters to identify potential storms and to begin developing policy and warnings even before the naming of a storm.

“That’s new,” said meteorologist Frank Marks, director of the federal Hurricane Research Division for the National Oceanic and Atmospheric Administration (NOAA). “Having a model do that before there’s a storm with enough confidence that the National Hurricane Center would start alerting people with watches and warnings, that’s a big leap.”

But the increased ability to see what might be out there has yet to translate into a better-prepared public. And the reason is probably a lack of understanding by the public about where a storm is and what that means in relation to its future danger, and the inability of the forecasters to educate the public.

Marks explains that when a storm is predicted to be 75 to 100 miles off the coast, forecasters express concern, putting out warnings. But the public really doesn’t understand what that 75 to 100 miles means. The forecast means the center of the storm is 75 to 100 miles out, but the storm’s width may mean that it’s a lot closer.

“While our guidance has gotten better, our ability to communicate our uncertainty with the tools we currently have is very challenging,” Marks said.

“As the average error continues to get smaller, we almost always see impacts outside the cone,” Michael Brennan, forecaster for the National Hurricane Center, told the NOLA Media Group recently. “And while the average track forecast errors are improving, we still have outliers, and it’s really hard to determine when that will happen, when we will have really big forecast errors.”

The models have improved by about 2 to 3 percent per year for about the last decade, but forecasting has vastly improved during the last three years because of an increase in data being fed to the models.

The increased data, particularly the data coming from the United States Air Force and the NOAA aircraft that fly into the storms, have coincided with technical improvements to the models that have greatly improved the models’ ability to forecast potential storms.

The aircraft that fly into a storm take measurements like temperature, pressure, humidity and wind, and that data is added to the model through a process called data simulation. The model takes into consideration the data and develops analysis for that moment in time.  When new data is added hours later, the model still remembers the previous data in its analysis.

“As you do that, each time you get memory of the past data, so it improves the way the model sees the storm. So each cycle that we put data in improves it a little bit more because of the memory part of the model,” said Marks.

In the case of Hurricane Barry, the model predicted that a storm would develop just south of the Florida Panhandle and that it had a likelihood of affecting New Orleans and even Texas at the time it was forming off the coast two or three days out.

That’s good information for emergency managers but doesn’t help the public as much, Marks said.

“Emergency managers want as much lead time as they can because they have to activate EOCs and do evacuations. So for many activities, this idea of having a potential tropical cyclone model was … a game-changer,” Marks said.

Platforms & Programs