Thursday, May 7, 2026

View in browser

More than a dozen people were injured in Mississippi last night when several supercell thunderstorms spawned strong tornadoes — something that is happening more in the Southeast as the planet warms. Combined with the higher population density in the Southeast (compared to, say, Tornado Alley) it's a recipe for major disasters.

Fortunately, our tornado warnings have gotten a lot better in the past century. Maggie is here to tell us about the earliest attempts to keep people safe.
—Angela Fritz, Meteorologist

How the US military learned to stop worrying ... and love tornado warnings

From Maggie Koerth, with illustrations by Ashley Burke

Captain Robert Miller left work on March 25, 1948, feeling like a failure.

A meteorologist with the US Air Force, Miller was stationed at Tinker Air Force Base, near Oklahoma City. Just five days before, a tornado barreled through the unprepared base, destroying the control tower and 34 airplanes that were sitting on the tarmac.

Part of Miller’s job was protecting all that expensive equipment from severe weather — a responsibility easier said than done. In 1948, there was still no reliable way to accurately predict tornadoes. Even radar was still a new technology, and no one yet knew what to look for on the shuddery black and white screen that could signify an oncoming twister. In fact, no one in the US had issued a public tornado forecast in more than 60 years.

But on March 25th, Miller and his colleague, Lt. Colonel Earnest Fawbush, gave it a shot. Carefully comparing atmospheric data, they saw the conditions were almost identical to the ones that had spawned the March 20th tornado. The two men wrote up a forecast — a tornado was likely late that afternoon — and sent it off to base command.

Then they watched as the afternoon came, and went, with little more than some spitting hail and 26 mph winds.

Turns out, there was a reason two generations had passed since the last time someone issued a tornado forecast — they just weren’t very good. At least, not in the military's estimation.

The issue was specificity. Researchers collected data on the atmospheric conditions that seemed to presage tornadoes, using a network of human storm spotters scattered across the country. By 1884, there were nearly 1,000 of these reporters and a US Army Signal Corps Seargent named John Finley was issuing twice-daily tornado predictions with what he claimed was a 96.6% accuracy rate.

But Finley’s apparent success relied on the fact that he issued two different kinds of predictions: “tornadoes likely” and “tornadoes unlikely." He’d also sliced the US up into 18 large districts that were biased by historic tornado activity. So it was one thing to predict a no-tornado day for District II, which stretched from southern New Jersey to Lake Erie, and another thing entirely to determine the likelihood of a tornado in District XIII, which encompassed a large swath of what we now consider Tornado Alley.

While Finley’s predictions of “unfavorable conditions for tornadoes” had a failure rate of just 1%, his positive tornado predictions were wrong 72% of the time.

The first tornado forecasts were issued in 1884 — but the military quickly decided the public couldn't be trusted with such sensitive information. The ban on predicting tornadoes would last for more than 60 years.

To the Signal Corps, this looked like a recipe for chaos and mass hysteria. In 1887, the Chief Officer issued a report stating, “It is believed that the harm done by such a prediction would eventually be greater than that which results from the tornado itself.”

Now, to Finley’s credit, there’s no record of his predictions sending Americans screaming into the streets. But that’s probably at least partly because a “tornado likely” forecast in 1884 rarely reached the public.

The telegraph had made fast communication over long distances possible, but there wasn’t mass media for that proverbial last mile between the telegraph operator and the public. The forecast districts were huge and the ability to tell people what was happening was small.

Even if a funnel touched down, there was no way to warn the people in its path. Although, God knows, folks did try. One researcher, inspired by Finley’s data, proposed a plan where every city would have a telegraph wire strung along the southwest side of town that was designed to break in high winds, triggering alarms or — more delightfully — cannon fire.

Five decades later, when Miller and Fawbush issued their tornado watch for Tinker AFB, meteorologists’ prowess had improved — but so had the ability to use such a forecast in a practical way, provided it was correct.

Miller left the base in the afternoon thinking he’d gotten the prediction wrong. But later that night, around 6 p.m., he heard something odd on the radio: the announcers were talking about a destructive tornado at Tinker. Thinking they were still talking about the storm from five days ago, he tried calling the base to see if there was some new damage they’d discovered. But the phone lines were dead.

So Miller drove to Tinker — and discovered that the base had been hit, again, by a totally different tornado. But this time, the storm was expected, and the most valuable equipment had been secured. Miller and Fawbush had gotten their forecast right, after all.

For the next two years, Miller and Fawbush’s techniques remained the sole purview of military meteorologists. Tornado watches were issued for military bases, where the storms could threaten planes and trucks, but not for the public, who were still assumed to be too skittish to deal with them.

Word got out anyway.

Soldiers told their families, who told their neighbors, of course. But the “leaks” were bigger, even, than that. At one point, the lead meteorologist of the US Weather Bureau’s Kansas City office was taking tornado forecasts from the Air Force and passing them directly to the press, who read them on the radio.

By July of 1950, the public pressure had built enough that the Weather Bureau officially rescinded the rule that its predecessor organization had set down in 1887. Tornadoes could be forecast now with better accuracy. And the public, it seemed, could handle it.

In fact, media at the time described the effect as calming. One Saturday Evening Post article in 1951 wrote about a farmer now having the time to “walk — not run — to his ‘scarehole.’”

Photo of the day

A double rainbow frames Chicago’s Willis Tower on May 4. (Credit: Barry Butler)

Meet the team

Maggie Koerth, Weather & Climate Editor

Raised in Kansas, Maggie still believes that taking a beer to the front porch is an important part of thunderstorm season.

Ashley Burke, Multimedia Designer

Ashley is an artist and animator who is delighted to illustrate weather history.

Angela Fritz, Meteorologist, Senior Director, CNN Weather

Angela loves stories about meteorology pioneers, and is thankful for modern science and communications.

Unlock deeper analysis and exclusive videos on the stories you care about. Subscribe here.

Did you enjoy this newsletter?

Let us know. We love to hear from you!

Drop us a line

You are receiving this newsletter because you signed up for CNN Weather.
To stop receiving this newsletter, unsubscribe or sign up to manage your CNN account