Hurricane track forecasts improve; strength still hard to predict

In the world of tropical storm science, and the public reaction to storms, some things never change, even though year after year, forecasts of where storms are going get more accurate.

When a panel of University of Miami scientists and other officials concerned with tropical storm forecasting gave a briefing to reporters last week, many of the messages were familiar:

Despite better forecasts for storm tracks, intensity forecasts have not improved much.

Many people don’t understand the uncertainty factor in storm forecasts and the meaning of the “cone of error” in forecast graphics.

Larger coastal populations need more time to evacuate than in previous decades.

At the same time, scholars and government officials are excited by advances in equipment and research programs that could give people more time to prepare and ultimately save lives.

At the University of Miami Rosenstiel School of Marine and Atmospheric Science on Thursday, May 12, Bill Read, director of the National Hurricane Center, and a panel of academic and government researchers talked about ways that hurricane forecasts are improving and how new research could give scientists better understanding of how hurricanes form and where they go.

The panel included John Cangialosi, hurricane specialist at the NHC and Shirley Murillo, field director of the National Oceanic and Atmospheric Administration’s Atlantic Oceanographic and Meteorological Lab on Virginia Key, and other scientists and forecasters.

Better track accuracy

Read told the audience that he knew nothing about hurricanes when, as a young Navy ensign, he became an aviator on the first hurricane hunter flight into 1972’s Hurricane Agnes.

Now he leads the agency that provides hurricane forecasts for the U.S. and other countries whose coasts lie on the Atlantic hurricane basin.

Read said scientists have made big improvements in hurricane forecasts over the years. From the 1960s through 1983, the average error in hurricane forecasts was 400 miles at 72 hours. In other words, predictions of where a storm would be in three days were off by 400 miles on average — sometimes more, sometimes less.

In the years from 2003 to 2008, forecasts had improved so that the average error was down to less than 200 miles at 72 hours, and less than 100 miles at 48 hours.

“If you gave somebody an average error now of 400 miles at 72 hours out, you’d probably be fired,” Read said.

According to data Read showed, the NHC track error has gone down 60 percent since 1990. A forecast for five days from now is as accurate as forecasts for three days away were a decade ago.

Meanwhile, forecasters haven’t made much improvement in forecasting the intensities of hurricanes in the last 15 to 20 years. “We probably don’t do much better forecasting intensity today than we did then,” Read said of his early years tracking hurricanes.

In the next decade, Read said, he said, he expects track forecasts for six to seven days will be as accurate as they are today for four to five days ahead.

Although improvements in track forecast accuracy show signs of leveling off in the last few years, “I’m not that worried about the track forecast,” Read said. “I think we will gain on that as new improvements come along.”

Knowing the cone

Because of improved forecasts, the familiar “error cone” will be smaller this year. The hurricane center adjusts the cone every year. The size is set so that the cone is as wide as two-thirds of the average error on each side of the track, for each forecast period.

This year, the cone is substantially smaller, Read said, partly because the record-setting 2005 hurricane season — which was not a “stellar” year for track forecasts, he noted — is no longer in the five-year error calculation.

Because the error could be to the right or left of the track, the circles that define the cone are twice as wide as the average forecast error.

This year’s error cone, based on 2006 through 2010, will be only 72 nautical miles wide at the 12-hour forecast point. At the furthest forecast point, five days (120 hours) away, the cone will be 478 nm wide. That’s about 550 statute (land) miles.

But the biggest problem, mentioned by several speakers on the panel, is that people can completely misunderstand what the cone represents.

Even when the cone is carefully explained, it’s hard for people to not see it as a kind of picture of the storm — a forecast of how big the storm is, how far out the winds will be or where the storm surge will strike the coast. The cone does not represent any of those things.

Read emphasized that the cone has absolutely nothing to do with a storm’s size, or the size of its effects. Even if a forecast is very accurate, and the center of a storm is exactly in the center of the forecast cone, hurricane-force winds or storm surge, or both, can easily affect locations outside the cone.

“A lot of people think if they’re outside the cone they don’t need to worry,” said John Cangialosi, the NHC hurricane specialist. He said in a storm like Hurricane Ike, “the cone tells you nothing about the wind field,” and Ike’s worst storm surge hit beyond the cone.