Hurricane Nate: No One Knows How To Read Hurricane Forecasts—Here's Why

As Hurricane Nate approaches the Gulf Coast, tens of thousands of Americans are turning to hurricane forecasts to understand the risks they face. But despite decades of improvements in meteorology, there’s no perfect way to display that information to the public.

Meteorologists have been interested in risk communication for many years, and an ongoing project is pairing their expertise with that of cognitive scientists to understand the strengths and weaknesses of different hurricane forecast displays. The newest paper from the project, published this week in the journal Cognitive Research: Principles and Implications, offers more insight into how current displays can trip people up.

10_07_hurricane_nate_forecast The "cone of uncertainty" forecast for Hurricane Nate as of October 7. National Hurricane Center

"We were really inspired to learn how people think about uncertainty," lead author Lace Padilla, a doctoral student at the University of Utah, told Newsweek. In this study, they focused on two different types of forecast displays: the so-called ensemble model, which shows individual lines for individual predicted paths, and the summary model, which shows one swath over which the hurricane might pass.

If you’ve seen a forecast released by the National Hurricane Center, that’s an example of the summary model, specifically known as a “cone of uncertainty” visualization. It's been around for decades, but being a bit old-fashioned isn’t necessarily a bad thing, since it means people who have lived in regions that often see hurricanes are pretty familiar with the format, Padilla says.

The team had already compared the “cone of uncertainty” with a few variants, like one displaying a central line or one with fuzzy boundaries for the cone—but people tend to parse these variants in more or less the same way, and with the same errors. That can include believing that the storm and its impacts will stay within the cone, even though the cone represents only 67 percent of historical error.

10_07_hurricane_forecast_models Summary models on the left and ensemble models on the right showing forecasts for two different hurricanes, top and bottom. University of Utah

The ensemble model is newer and still somewhat less common, but depicts individual tracks from different forecast runs. Here, people tend to over-emphasize the individual lines that make up the visualization. "We were very surprised that people tended to focus on the one ensemble member that sort of hit their area of interest," Padilla says, like their home or business. If a line does happen to fall on that specific place, the authors found, people are more concerned than if individual tracks fall on either side of that particular point.

"That's really surprising because the scientists that make these pick the number of lines they want to plot, so those lines don't represent all the possible paths," Padilla says. That means if the visualization cuts off at 20 paths and path number 21 would hit your hometown, you may underestimate the risk you face.

Padilla and her colleagues aren’t certain yet what the best solution to the problem will be, since right now it seems that each visualization approach has its own strengths and weaknesses. Animation may be able to address some of the current problems, but may carry its own issues. "There's work showing that people tend to forget the information at the beginning of an animation and only remember the last bits that they saw," Padilla says, plus the animation itself can be distracting.

Even if meteorologists and data visualization experts do find an ideal way of conveying hurricane forecasts to the public, that won't necessarily mean everyone at risk makes the best decision possible, says Kenneth Broad, who studies human interactions with the environment at the University of Miami.

"While the cone, for example, may be one piece of information, it's not the only one people take into account," Broad says, noting that social cues like whether friends and neighbors are preparing for the storm also affect decisions. And while these displays focus on where a storm will go, there are other important pieces of information that can't be included, like the intensity of the storm, its size, and what the relative risks of wind and water damage.

He says local meteorologists have taken the lead in adapting forecast displays to make them more easily understandable and in trying to find better ways to convey risk to their audiences. That work also includes explaining what precisely the forecast visualizations show. "People absolutely need more information to even understand what's being represented," Padilla says. "They need to know what is being shown because it's just not inherent in the visualization."

Join the Discussion