Photo illustration by Slate. Photo by Hector Reamal/AFP/Getty Images.

The Meteorologist’s Lament

Weather forecasting is better than ever—but the public doesn’t seem to realize that.

Imagine a situation where Jordan Spieth stands at a tee 87 yards from the hole. The crowd takes for granted that he will hit a hole-in-one. Any other outcome from that swing is unacceptable. Spieth tees off, and the ball lands 6 inches to the left of the cup. While some cheer, others in the crowd curse as Spieth taps the ball in for a birdie and gains a stroke on the rest of the field.

That is a reasonable analogy to the forecasts for Hurricane Irma and the subsequent grumbling response from much of the public and some of the media. Irma was a Category 4 storm when it struck the Florida Keys and destroyed 1 out of 4 homes. Fortunately, the storm weakened while passing over the almost unpopulated Everglades before it got to the heavily populated Marco Island-Naples area. AccuWeather, where I work, estimates the total cost of Hurricanes Harvey and Irma to be $290 billion.*


The forecast was not perfect. When the evacuation decisions had to be made, the anticipated path of the eye of the storm (which contains the highest winds and the most destructive storm surge) was about 60 miles too far east. That caused more people to be evacuated than absolutely needed to be. Many seemed to think that this “birdie”-quality forecast wasn’t good enough.

But consider this. On Sept. 13, 1928, an unanticipated Category 5 hurricane struck Puerto Rico. Estimates of deaths range from 300 to more than 1,000. Four days later, again without warning, it struck Palm Beach County (then lightly populated), brushed the northeast side of Lake Okeechobee, and passed over Orlando and Amelia Island before making its way up the U.S. East Coast. The death toll for Florida alone was 2,500 to 3,000. The names of most of the victims are unknown, as they are buried in mass graves.

Let’s say, then, that 2,700 people died. Adjusted for the increase in Florida’s population since, the death toll in an identical hurricane would be 38,000. Weather science made America’s lifestyle migration to Florida possible. According to the Miami Herald, 47 died in Florida in Hurricane Irma.

But weather science and Florida may both be victims of their own success.

It now takes about 72 hours to do an orderly evacuation of southeast Florida (Miami-Dade, Broward, and Palm Beach counties) for a major hurricane. Even over three full days, there are major problems. If it takes officials 12 hours to coordinate the evacuation, and if everyone is to be out 12 hours before the storm begins, meteorologists have to give Florida officials 96 hours’ notice. But optimal evacuation decisions require a level of precision that is greater than weather science can consistently provide that many hours before landfall. As the below graphic demonstrates, the Florida forecast was nearly perfect 48 hours out and saved countless lives. But we were not able to make a “hole-in-one forecast” when Irma was four days away.

The white line is the actual path of the eye of Hurricane Irma. Red dots are the forecast locations of the eye 12, 24, 72, and 96 hours into the future. The purple tint is the width of hurricane strength winds (above 75 mph). The brown tint is the width of tropical storm force (39-74 mph).

For meteorologists like me, this is frustrating. Our increasingly accurate forecasts save untold numbers of lives. For instance, the death rate from tornadoes has been cut by more than 95 percent since the 1930s, the last decade when there was no tornado warning system of any kind. In the 1970s and ’80s, thunderstorm-related downbursts—violent downrushing air that spreads out when it reaches the ground, creating wind shear—were the No. 1 cause of airliner crashes. Thanks to the work of weather scientists, the last wind shear-related airline crash was in 1994.

It’s easy to overlook these major advances because when meteorology is successful, nothing happens. Planes don’t crash. People don’t die, or at least they die in far smaller numbers than before the storm warning system existed. It’s a bit like national security—stopping a terrorist attack before it happens doesn’t make for as good a news story.

In most other countries, people get their forecasts from the government. That isn’t the case in the United States. Meteorologists on television and companies like AccuWeather provide the forecasts, while the government maintains the common meteorological infrastructure and provides storm warnings to the public at large. Businesses that require tailored forecasts and storm warnings receive them from private-sector companies. That way, the public is not subsidizing specialized services for business. We call this amalgamation “America’s Weather Enterprise.” And it is all done at amazingly low cost to the public. The budget of the National Weather Service is $1,124,149,000 per year, or $3.45 per person. It is one of the best investments the federal government makes.


Although forecasts have improved markedly, they can get even better. We are still learning to fully exploit a new technology known as dual-polarization radar that has been installed on National Weather Service radars in the past two years. It is already resulting in more accurate flash-flood warnings and better water resource management.

Predictions are—as I know so well!—tricky, but by the early 2020s, I expect we will be able to forecast the approximate path of supercell thunderstorms—the kind that produce the most violent tornadoes, the largest hail, and often the highest thunderstorm straightline winds—with a reasonable degree of reliability. This will allow hospitals, local governments, and others that require additional preparation time to take the needed precautions in an orderly manner.

Beyond that, who knows? I have been a meteorologist for 46 years, and the pace of innovation is surprising even me. In 20 years, we may be able to consistently hit those holes-in-one.

And, by the way, weather science hit a hole-in-one with the forecasts of Hurricane Maria. Just as predicted, it was a Category 5 when it struck Dominica and St. Croix and Category 4 when it unleashed savage winds and flash flooding on Puerto Rico. As of this writing, with power not expected to be restored for months, we cannot know what the ultimate death toll might be. However, it is almost certain to be less than the 1928 storm, even though the population of the island has tripled.


Correction, Sept. 21, 2017: This piece originally misstated AccuWeather's estimate for the total cost of Hurricanes Harvey and Irma. It is $290 billion, not $390 billion. (Return.)

This article is part of Future Tense, a collaboration among Arizona State University, New America, and Slate. Future Tense explores the ways emerging technologies affect society, policy, and culture. To read more, follow us on Twitter and sign up for our weekly newsletter.

More from Technology

Uranus Is About to Get More Visible. Here’s How to View It.

Google and Facebook Helped Campaigns Confuse Voters, Stoke Bigotry Before the Election

Snapchat Was Built on Intuition, Not Data. Is That Why It’s Struggling?

Trump’s FTC Pick Doesn’t Appear to Share Trump’s Desire to Crack Down on Amazon Out of Spite

Alphabet, Google’s Parent, Leads $1 Billion Investment Round in Lyft

Congress Should Pass the Bill That’d Help Stop Russia From Buying Online Election Ads