The Repository Hook Echo

EF Scale: Why No Tornado Was Rated EF5 for 12 Years

October 29, 2026 · Runtime 15:14 · Watch on YouTube

From Moore, Oklahoma, on 20 May 2013, to Enderlin, North Dakota, on 20 June 2025, no tornado in the United States was rated EF5: 12.1 years, the longest gap since records began in 1950. The Enhanced Fujita scale is read from damage, not measured. On its table for houses, the worst damage a home can show, the slab swept clean, has an expected wind of 200 mph: the top of EF4. On the old Fujita scale, the same slab was F5. When a mobile radar measured at least 295 mph near the ground at El Reno in 2013, the rating stayed EF3, because the strongest winds hit nothing a surveyor could rate.

Chapters

  1. 0:00 Moore, the last EF5 before the gap
  2. 0:58 Sixty tornadoes at the top
  3. 1:41 Fujita's scale of wind, 1971
  4. 2:52 Read from wreckage
  5. 3:08 Why it was replaced
  6. 4:05 28 damage indicators
  7. 4:37 The EF bands
  8. 5:21 The ladder for a house
  9. 6:12 The slab swept clean
  10. 6:58 The diagnosis
  11. 7:46 Our count of the record
  12. 9:13 El Reno, 2011
  13. 10:18 El Reno, 2013: measured, then EF3
  14. 11:49 Tornadoes with no rating
  15. 12:46 Enderlin, 2025: the rail cars
  16. 13:53 What the record supports
  17. 14:49 Next: Helene, 2024

Transcript

Moore, the last EF5 before the gap 0:00

On the EF scale, no tornado was rated EF5 for twelve years. The last one before the gap hit Moore, Oklahoma, in May 2013. No instrument on the ground measured its wind. The rating was read from what was left of houses like these.

We counted every rating in the national tornado record, seventy-three thousand four hundred and fifty-eight tornadoes since 1950, to see what the top of the scale actually records. This is how the radar saw Moore that afternoon: debris lofted into the storm. Radar sees the debris. It does not measure the wind at the ground.

Twelve years without an EF5 looks like a fact about tornadoes. The record says it is at least as much a fact about a table. This episode follows that table: how a destroyed house becomes a wind speed, why the worst damage a house can show stopped being enough, and what happened when an instrument did measure the wind.

Sixty tornadoes at the top 0:58

Since 1950, the National Weather Service has rated sixty tornadoes at the top of the scale. Fifty were F5, from 1953 to 1999. Ten were EF5, from 2007 on. That list is everything the top of the scale has ever held.

Our count of the gaps between them: from Moore, on May 20th, 2013, to the next one, on June 20th, 2025, is twelve point one years. The longest gap before that was eight years, from 1999 to 2007. Researchers at the National Severe Storms Laboratory had already called it the longest drought in the record, months before it ended.

Fujita's scale of wind, 1971 1:41

The scale is forty years older than the gap. In February 1971, Ted Fujita, at the University of Chicago, proposed a way to rank tornadoes and hurricanes by area and intensity. He built it as a scale of wind.

It was designed to connect smoothly the top of the Beaufort scale, the sailors' wind scale, with the speed of sound. F1 began at Beaufort twelve, seventy-three miles an hour, the start of hurricane wind. F12 was Mach one.

In between, each step followed one formula. F2 began at one hundred and thirteen miles an hour, F3 at one hundred and fifty-eight, F4 at two hundred and seven. F5 ran from two hundred and sixty-one to three hundred and eighteen: whole frame houses tossed off their foundations. The speed was the fastest quarter mile of wind; at F4, about four seconds of it.

Above F5, Fujita left six more steps on paper, from three hundred and nineteen miles an hour up to the speed of sound. He called that damage inconceivable. No tornado in the national record has ever been rated above five.

Read from wreckage 2:52

But nobody was there to measure those winds. In practice, the scale was always applied to damage. A surveyor looked at what the tornado did and picked the step whose description fit. A scale built for wind became a scale read from wreckage.

Why it was replaced 3:08

By 2001, the people using it knew its weak points. A report from Texas Tech University listed them: too few kinds of damage to read, no account of how well a building was built, and no definitive link between damage and wind. And a well-built house does not need two hundred and sixty miles an hour to be destroyed and blown away.

In March 2001, twenty users of the scale met in Grapevine, Texas. Engineers and meteorologists spent the next years building a replacement. The National Weather Service put it into use on February 1st, 2007: the Enhanced Fujita scale.

Three months later it had its first EF5. Greensburg, Kansas, May 4th, 2007. The Dodge City office put the strongest surface wind in parts of town at about two hundred and five miles an hour. Nine hundred and sixty-one homes and businesses were destroyed, and eleven people died.

28 damage indicators 4:05

The new scale starts with twenty-eight damage indicators: kinds of buildings, structures and trees. Each indicator has its own ladder of degrees of damage. Each degree has a wind speed attached, with a lower and an upper bound.

Those wind speeds were not measured either. They came from expert elicitation: six recognized experts, chosen under a formal protocol, estimated the wind each kind of damage takes. Judgment, written down carefully, in place of an instrument.

The EF bands 4:37

The wind also changed meaning: a three-second gust, not Fujita's quarter mile. EF0 is sixty-five to eighty-five miles an hour. EF1, eighty-six to one hundred and ten. EF2, to one hundred and thirty-five. EF3, to one hundred and sixty-five. EF4, one hundred and sixty-six to two hundred. EF5 is anything over two hundred. And the Storm Prediction Center warns that this three-second gust is not the same wind that a weather station reports.

The Storm Prediction Center says it plainly on its own page. The Enhanced F-scale still is a set of wind estimates, not measurements, based on damage.

The ladder for a house 5:21

The indicator a surveyor meets most often is the house. One- and two-family homes are indicator number two, and they have ten degrees of damage, from the first visible damage to nothing left at all.

Here is the ladder, at its expected values. The threshold of visible damage: sixty-five miles an hour. Exterior walls collapsed: one hundred and thirty-two. All walls down: one hundred and seventy.

This is degree six: large sections of the roof structure removed, most walls still standing. Expected wind, one hundred and twenty-two miles an hour, anywhere from one hundred and four to one hundred and forty-two. Nashville, March 3rd, 2020. The tornado that did this was rated EF3, one hundred and sixty-five miles an hour at its worst.

The slab swept clean 6:12

Degree ten is the last one. Destruction of a well-built house, the slab swept clean. Expected wind: two hundred miles an hour. Lower bound, one hundred and sixty-five. Upper bound, two hundred and twenty.

Now put that number next to the scale. EF4 ends at two hundred. EF5 begins above it. The most complete destruction a house can show reads, at its expected value, as the very top of EF4.

To go higher, a surveyor has to argue for the upper bound: that this house was well built and well anchored, and took more than the expected wind to sweep away. On the F scale, the same empty slab was simply F5.

The diagnosis 6:58

In March 2025, a team led by Anthony Lyza of the National Severe Storms Laboratory published the diagnosis. The typical rating for a home swept off its foundation is EF4 on the EF scale, and was F5 on the F scale. This, they wrote, is likely the fundamental cause of the lack of EF5 tornadoes. Their fix is not a new scale. They suggest a few simple adjustments to the wind speeds attached to the damage.

The warning had been in print for years. In 2009, Charles Doswell, Harold Brooks and Nikolai Dotzek wrote that the minimum criteria for EF5 damage had effectively been raised: the complete destruction of a typical frame home would no longer be enough for EF5, and perhaps not even for EF4.

Our count of the record 7:46

Our count shows the same step. From 1950 to January 2007, the record holds fifty F5 tornadoes in fifty-eight years: zero point eight six a year. From February 2007 on, ten EF5 in nineteen years: zero point five three a year. Rated F4, eight point six a year. Rated EF4, five point two.

The fall did not start in 2007. The share of tornadoes rated four or five was two point five percent in the 1950s, zero point eight in the 1980s, and zero point three so far in the 2020s. Part of that is many more weak tornadoes being reported. Part is how the damage is read. The record alone cannot separate the two.

The EF scale did produce EF5 ratings. Six in 2011 alone, among them Joplin, Missouri, on May 22nd. The national record gives it one hundred and fifty-eight deaths.

And Moore, two years later. The survey team documented four thousand two hundred and fifty-three damaged objects along the path. The Norman office counted more than three hundred homes with EF4 or EF5 damage. For the strongest level of damage, EF5, the survey found only nine homes. Nine houses carried the top of the scale.

El Reno, 2011 9:13

Twice, near the same Oklahoma town, an instrument was in the way. On May 24th, 2011, a tornado passed by the Oklahoma Mesonet station northwest of El Reno. The station logged a gust of one hundred and fifty-one miles an hour, and kept working. That is rare. A 2013 review found only thirty-one direct measurements of a tornado by a station on the ground, from 1894 to 2011.

That tornado was rated EF5. A federal review of the EF scale notes that its final rating was influenced upward by mobile radar data, in a stretch with few objects to damage. In 2011, a measured wind helped set the number.

It had happened before, the other way round. On May 3rd, 1999, at Bridge Creek, a mobile Doppler radar measured three hundred and one miles an hour, give or take twenty, about thirty meters above the level of the radar. That tornado was rated F5. The damage had agreed with the instrument.

El Reno, 2013: measured, then EF3 10:18

Two years later, May 31st, 2013, a few miles south of El Reno. This is the KTLX radar that evening. The tornado that formed at 6:03 PM grew to two point six miles wide, by the National Weather Service's survey.

A mobile research radar from the University of Oklahoma, called RaXPol, was scanning it from a few miles away. The weather service's own summary says it: RaXPol data shows winds of at least two hundred and ninety-five miles an hour very close to the surface. A second research radar was there too. Both measured winds of more than two hundred.

On June 4th, the Norman office upgraded the tornado to EF5, based on that velocity data. On August 30th, it changed the rating to EF3.

The reason is in the office's summary. The strongest winds were in small vortices inside the larger tornado, and none of them hit a structure in rural Canadian County. Despite the measured wind speeds, surveyors could not find damage that would support a rating higher than EF3.

Eight people were killed, all of them in vehicles, three of them storm researchers. On the scale, El Reno is an EF3.

The weather service said at the time that its policy for ratings is based on surveys of ground damage, and that it was exploring whether that policy should change. A measured wind does not set the number. Damage does.

Tornadoes with no rating 11:49

That rule produces the strangest line in the record. A tornado that crosses only open fields, with nothing a surveyor can rate, gets no rating at all. EF unknown.

Our count of those. In 2016, three percent of the year's tornadoes were unrated. In 2019, twelve percent. In 2023, twenty. In 2025, two hundred and thirty-four of one thousand three hundred and eighty-three tornadoes, about one in six, have no number at all. The database keeps the label, not the reason.

The gap was known. In 2013, six federal and engineering authors wrote of a long-known tendency to underestimate tornado intensity where there is little to damage. They listed what might fill the gaps: center-pivot irrigators, farm machinery, grain bins, silos, rail cars.

Enderlin, 2025: the rail cars 12:46

June 20th, 2025, 11:02 PM, south of Enderlin, North Dakota. This is the KMVX radar. The tornado ran twelve point one miles in nineteen minutes, up to a mile wide, and killed three people.

It crossed a railroad. In October, the Grand Forks office upgraded it to EF5, greater than two hundred and ten miles an hour. The estimate came from the wind it takes to tip fully loaded grain hopper cars, and to loft tanker cars: one empty tanker was thrown about four hundred and seventy-six feet.

The office added that the maximum wind also matches the strongest rotation the KMVX radar measured. In the river valley, the trees were left as stubs of trunks, the bark sanded off. The rating that ended the gap came from rail cars, one of the indicators listed in 2013, and from a radar that agreed.

A house on the same path was not enough. At a farmstead, the foundation was swept clean. The office wrote that a lack of proper anchoring limits the rating there.

What the record supports 13:53

Here is what the record supports. An EF rating says how strong the wind had to be to do the worst damage the surveyors found, on something they could rate. It says nothing about the wind where there was nothing to break.

And twelve years without an EF5 came, at least in large part, from a table: the one that put a swept-clean house at two hundred miles an hour, one step below the top.

Here is what it cannot say. Whether tornadoes got weaker. With a stricter reading of houses, more weak tornadoes reported every decade, and one in six now rated by nothing at all, the ratings by themselves cannot answer it.

One question we could not close. A mobile radar measured at least two hundred and ninety-five miles an hour near the ground at El Reno in 2013, and the rating stayed EF3. Should a measured wind be allowed to set the rating? Tell us in the comments.

Next: Helene, 2024 14:49

Next, a hurricane whose worst flooding came in the mountains, hundreds of miles from where it came ashore. On the French Broad River at Asheville, North Carolina, a stream gauge with records back to 1895 read twenty-four point six seven feet, a foot and a half above the flood of 1916. That is the next episode. Helene, 2024.

Description and sources

Ted Fujita proposed his scale in February 1971 as a wind scale, from Beaufort 12 (F1, 73 mph) to Mach 1 (F12); in practice it was always applied to damage. The Enhanced Fujita scale, used by the National Weather Service since 1 February 2007, reads 28 damage indicators, each with degrees of damage and a wind range from expert elicitation: 'a set of wind estimates (not measurements) based on damage' (SPC). For homes, degree 10, slab swept clean, is 200 mph expected (165 to 220). Lyza, Brooks and Krocak (BAMS, 2025) call that likely the fundamental cause of the EF5 drought. Our count of the SPC database (73,458 tornadoes, 1950-2025): 50 F5 in 58 years against 10 EF5 in 19; tornadoes rated 4 or 5 fall from 2.5% in the 1950s to 0.3% in the 2020s; unrated tornadoes (EF-U) rise from 3% in 2016 to 17% in 2025. El Reno 2013 was rated EF5, then EF3; Enderlin 2025 was rated EF5 from rail cars and the KMVX radar.

PRIMARY SOURCES

Hook Echo reads a disaster through the instruments that recorded it: radar, sensor, survey and report, with the primary sources above.

About the material. Video: the Oklahoma National Guard's uncut Black Hawk flight over Moore on 21 May 2013, the Tennessee Air National Guard's aerial survey and FEMA's ground footage after the Nashville tornado of 3 March 2020, and federal footage from Joplin (2011), Tuscaloosa (2011), Mayfield (2021) and Rolling Fork (2023), all published on DVIDS. Radar: NOAA NEXRAD Level II (KTLX, KMVX), rendered by us with the National Weather Service colour scales. Our measurement uses the Storm Prediction Center's tornado database, 1950-2025. Every diagram is drawn in code from the sources listed. Nothing in this video is AI-generated imagery.

#HookEcho #EFScale #Tornado