The Repository Hook Echo

Moore 2013: The Day Radar Saw the Debris

October 6, 2026 · Runtime 15:13 · Watch on YouTube

On 20 May 2013 a mile-wide EF5 tornado crossed Newcastle and Moore, Oklahoma. National Guard footage shows what it left. The Oklahoma City radar, upgraded to dual polarization 225 days earlier, saw something no radar there could see in 1999: the debris itself, lifted almost four miles into the storm. We measured it scan by scan from NOAA's raw archive.

Chapters

  1. 0:00 Moore, the next day
  2. 0:27 3:16 pm on the radar
  3. 1:14 A day forecast in advance
  4. 1:45 The 2:40 warning
  5. 2:11 Touchdown, 2:56
  6. 2:36 225 days of dual-pol
  7. 3:09 What correlation measures
  8. 4:02 Counting the debris
  9. 4:49 Tornado Emergency, 3:01
  10. 5:57 Briarwood and the first EF5
  11. 6:17 Debris six kilometres up
  12. 6:57 Plaza Towers
  13. 7:43 Moore Medical Center
  14. 8:31 The loop
  15. 8:53 Six radars
  16. 9:45 174 to 201 mph
  17. 10:31 The ground survey
  18. 11:13 Nine houses
  19. 12:12 The debris outlives the tornado
  20. 13:01 What it cannot see
  21. 13:44 1999 and 2013
  22. 14:17 What the record supports
  23. 14:46 Next: Kilauea, 2018

Transcript

Moore, the next day 0:00

This is Moore, Oklahoma, filmed by the National Guard two days after a tornado. On the afternoon of May 20th, 2013, it crossed this city in about twenty minutes. It was more than a mile wide. It destroyed more than eleven hundred buildings, two elementary schools, and the only hospital in a city of nearly fifty-eight thousand people. The National Weather Service calls it the deadliest tornado of that year in the United States.

3:16 pm on the radar 0:27

And this is how a radar about fourteen miles away saw it, at 3:16 in the afternoon. On the left, the storm. On the right, a measurement that radar could not make fourteen years earlier. That pink blot is not rain, and it is not hail. It is pieces of houses, carried up into the storm, and the radar was reporting them every four minutes.

The next morning a Black Hawk flew the whole damage path, from the first farm to the last street, and the Guard released the footage uncut. This episode follows that path, and holds it up against what the instruments recorded while it was happening. The radar that saw the debris. The radar that came closest to the wind. And the survey that had to rate it afterwards, house by house.

A day forecast in advance 1:14

Nobody was surprised that there were storms. At one in the morning the national Storm Prediction Center already gave central Oklahoma a ten percent chance of a strong tornado within twenty-five miles of any point. At eight, the warning coordinator in Norman emailed emergency managers and school administrators. By nine, Moore's schools had cancelled every after-school activity. At 1:30 in the afternoon the district decided not to send the buses out at three. The children would shelter in the buildings.

The 2:40 warning 1:45

The storm that mattered formed southwest of Oklahoma City in the early afternoon. At 2:40 the National Weather Service in Norman issued a tornado warning for Newcastle, Moore and part of Oklahoma City. Moore sounded its sirens moments later. That warning came from Doppler velocity, the same way it had in 1999: wind moving toward the radar beside wind moving away, close together, in the hook of the storm.

Touchdown, 2:56 2:11

At 2:56 the survey puts the tornado on the ground, in the countryside on the northwest side of Newcastle. In the radar scan that started at 2:55, fourteen small cells sit right on the rotation, where the instrument says the echo has stopped looking like weather. The scan four minutes earlier had seven. Within four minutes of touchdown, the damage was already rated EF3 and EF4.

225 days of dual-pol 2:36

The radar in 1999 sent its pulse in one orientation, horizontal. In October 2012, KTLX was taken off the air for four and a half days. When it came back, it sent two pulses at once, one horizontal and one vertical. We checked this in the archive itself. The last scan in the old format is from October 2nd. The first scan in the new one, with the new measurement in it, is from October 7th. Two hundred and twenty-five days before Moore.

What correlation measures 3:09

The new measurement is called the correlation coefficient. It asks one question about every small volume of the storm: do the horizontal and vertical echoes change together, pulse after pulse? Raindrops are all roughly the same shape, so they do. In rain the number sits near one. Hail drops it a little. Shingles, insulation, branches and pieces of siding are every shape at once, tumbling. The two echoes stop agreeing, and the number falls.

So here is the rule we used, the one the research literature has used since 2005. A cell counts as debris when its correlation drops below 0.8, the echo is strong, and it sits within three kilometres of the rotation. Below about 0.2, the radar does not report a lower value at all. Over Moore, much of the debris sat right on that floor.

Counting the debris 4:02

The tornado grew fast after it crossed the Canadian River and moved into the rural edge of southwest Oklahoma City. Watch the count. Fourteen debris cells at 2:55, twenty-six at 2:59, seventy-eight at 3:03. The rotation signal on this radar stayed about the same from scan to scan. What changed is what the tornado was reaching, and what it had to pick up.

This is that stretch from the air the next morning. Ponds, pasture, scattered houses with long driveways. Even here the path is easy to read: a band of bare ground and broken trees running across fields that are green on both sides of it. Out here there was less to lift. That is a limit of the method, and we will come back to it.

Tornado Emergency, 3:01 4:49

At 3:01 the National Weather Service upgraded the warning to a tornado emergency, its most urgent wording. According to the federal review, the Norman office had used that phrase only twice before. The first time was May 3rd, 1999, for the tornado in our last episode. The second was 2003. Both of those also hit Moore.

Moore's outdoor sirens went off six times that afternoon. The first was moments after the 2:40 warning. The last was at about 3:20, when the tornado was already in the city. At 3:11 the sirens played a recorded voice message, the only time that day they were used for that. The federal review is plain about it: the sirens are meant for people outdoors. Inside, the warning has to arrive some other way.

Here is the whole series, one point for every scan of the radar. The count climbs through southwest Oklahoma City, jumps when the tornado enters the dense neighborhoods of Moore, and peaks at about four hundred cells. And it does not stop when the tornado does. We will get to that.

Briarwood and the first EF5 5:57

At about 3:15 the leading edge reached Briarwood Elementary School. The researchers who ran a fast university radar in Norman put the first EF5 damage near that school at 3:16 and about forty seconds. In the KTLX scan that began at 3:16, the debris signature reaches its tallest column of the day.

Debris six kilometres up 6:17

The radar tilts its antenna up through fourteen angles in every scan. At 3:16, every one of them up to nearly sixteen degrees found debris above that spot, without a gap. The highest of those beams crossed it about six kilometres above the ground. Close to four miles. Pieces of houses, carried that high inside the storm, and the radar counted them there.

This is what was under that column. From the air, whole blocks are reduced to the outlines of their streets. The concrete slabs are still there. The houses that stood on them are spread downwind, in a fan that the camera has to pan across to follow.

Plaza Towers 6:57

Plaza Towers Elementary was hit a few minutes after Briarwood. The children there were where the school's plan told them to be, in the area it had designated as safe. Part of that building collapsed, and seven children died. The federal review found that none of the three buildings it studied, the two schools and the hospital, had a room built as a storm shelter, and none had a basement.

Sixteen minutes passed between the warning and touchdown. The national average lead time for a tornado warning at the time was about fourteen. The federal review puts the deaths in exactly those terms. They happened even though the storm was forecast days ahead, even though the warning came early, and even though the children at Plaza Towers were where the plan said they should be.

Moore Medical Center 7:43

At about 3:20 the tornado reached the Moore Medical Center. Thirteen patients were inside. Staff moved people into the safest rooms they had, one of them a walk-in freezer. Nobody died there. Wind, rain and debris went through the outer walls and gutted the inside. The concrete and steel frame held, but the building never reopened. It was demolished, and Moore was left without a hospital.

FEMA counted two thousand three hundred and ninety-three structures in the path. One thousand one hundred and twenty-eight destroyed. Three hundred and thirty-five with major damage. Six hundred and twenty-nine with minor damage, and three hundred and one more affected. Only eighty-five of them were not homes.

The loop 8:31

Near that hospital the tornado did something unusual. It slowed, made a small loop, and then went on east. On the university radar, which scanned every twenty seconds, the loop came with bursts of debris thrown out of the circulation. Every radar that was watching shows the wind estimate falling sharply at this moment, around 3:22.

Six radars 8:53

That is the other half of this story. Six radars scanned this tornado at close range. Two were test units of the same model as KTLX, in Norman. One was the FAA's radar at the city airport. One was a university trailer. One was an experimental phased-array radar. And KTLX, the one that issued the warnings, was the farthest away of all of them, between twelve and thirty-three kilometres.

A radar does not measure the wind at the height of a roof. It measures how fast the things in its beam are moving toward or away from it. The closest radar was the Terminal Doppler radar that guards Will Rogers airport. At 3:17 and thirty seconds its estimate of the tornado's wind peaked at eighty-six metres per second. That is about one hundred and ninety-two miles per hour.

174 to 201 mph 9:45

The researchers converted that peak into the three-second gust that the damage scale is built on. Their answer was a range, from about one hundred and seventy-four to two hundred and one miles per hour. EF5 starts above two hundred. The best wind measurement anyone made that day ends right on the line.

And the team listed why even that number is not the whole answer. The beam was forty to eighty metres up, not at ten, where the scale is defined. The debris itself was in the beam, and debris thrown outward by the spin does not move exactly with the air. And after the loop, the damage path narrowed to just twenty to forty metres wide. Still EF4. Too narrow, they wrote, for any of the radars to sample.

The ground survey 10:31

So the rating came from the ground. A survey team from the National Weather Center and two private companies started walking the path on the morning of May 21st and kept going for weeks. They did not sample the damage. They tried to log every damaged object. Four thousand two hundred and fifty-three of them.

Four thousand two hundred and twenty-two of those objects were damage indicators: houses, schools, trees, power poles, each one with a table that says what wind it takes to do what was seen. About half were rated EF0. Of the rest, thirty-eight percent were EF1, twenty-four EF2, twenty-one EF3, seventeen EF4.

Nine houses 11:13

And EF5, the rating the whole tornado carries, was four tenths of one percent. Nine houses. Before it could award that rating at all, the team had to write down what a well-built house was. A house is only as strong as the way it is fastened to its foundation, and only a well-built one could earn EF5.

On the ground the Guard filmed what the indicators look like before anyone rates them. Street signs bent flat. Trees stripped of their bark and most of their branches. Cars rolled into what had been living rooms. Each of these is a data point in the survey, and a street in somebody's neighborhood.

That night, Guard airmen searched the rubble with thermal cameras, house by house, in the dark. The final count was twenty-four people killed, and three hundred and eighty-seven treated for injuries at local hospitals. Federal estimates put the number of buildings affected at two thousand three hundred and ninety-three.

The debris outlives the tornado 12:12

Back to the radar. At 3:35 the tornado lifted, about five miles east of Moore, after roughly thirty-nine minutes on the ground and fourteen miles of track. In the scan that began at 3:38, the debris signature is still there, four hundred and one cells of it. At 3:42, one hundred and seventy-three. At 3:46, fifty-one. The tornado was gone. What it had lifted was still coming down.

At the end of its flight the Black Hawk climbed, and from there the path stops being streets and becomes a single brown stripe across the green, running off toward the horizon. The survey measured it at up to one point seven kilometres wide, and twenty-three long. Everything in this episode happened inside that stripe, in thirty-nine minutes.

What it cannot see 13:01

So what did the new measurement change? It lets a forecaster confirm that a tornado is on the ground and destroying things, even when rain hides it and no one can report it. But it needs something to lift. Over empty pasture the signal is weak, and the farther a tornado is from the radar, the higher the beam passes over it, and the more of the debris it misses.

Over Moore, KTLX's lowest beam crossed the tornado about one to three hundred metres above the ground. At one hundred kilometres from the radar, that same beam is nearly a kilometre and a half up, and at a hundred and fifty, more than two and a half. Moore sits about twenty kilometres from Twin Lakes. Few places are that lucky.

1999 and 2013 13:44

Set the two Moore tornadoes side by side. In 1999 the track ran thirty-eight miles and thirty-six people were killed. In 2013 the track was fourteen miles, and twenty-four were killed. The same radar watched both. In 1999 it could see the storm and the spin. In 2013 it could also see what the spin was carrying. Neither number on its own says the second warning worked better. They were different storms, on different paths, at different times of day.

What the record supports 14:17

Here is what the record supports. The radar knew, within a minute of touchdown, that this tornado was hitting things. It watched the debris climb six kilometres over the first EF5 damage. It could not measure the wind at the height of a house, and neither could any other radar that day. And none of it could make a hallway into a shelter. Moore had no public storm shelter that day. Its residents had registered more than three thousand of their own.

Next: Kilauea, 2018 14:46

Next, the ground itself moves. In 2018, the summit of Kilauea, on the Big Island of Hawaii, collapsed sixty-two times in three months. Each collapse released the energy of a magnitude 5.3 earthquake, and the instruments around the summit recorded every one. That is the next episode. Kilauea, 2018: the volcano that fell in on itself.

Description and sources

The tornado touched down at 2:56 pm CDT and lifted at about 3:35, 14 miles and roughly 39 minutes later; 24 people were killed, seven of them children at Plaza Towers Elementary, and 387 were treated at hospitals. The warning came 16 minutes before touchdown, and a Tornado Emergency at 3:01. KTLX began recording the correlation coefficient on 7 October 2012. Counting cells where that value fell below 0.8 beside the rotation, the debris signature appears in the 2:55 pm scan, grows from 14 cells to about 400 over Moore, and reaches a continuous column about 6 km high in the 3:16 pm scan, the minute of the first EF5 damage near Briarwood. It outlives the tornado: 173 cells at 3:42 and 51 at 3:46, debris still falling after the tornado lifted at 3:35. The closest radar, the airport TDWR, put the peak wind at 86 m/s, a 3-second gust of 174 to 201 mph: the EF5 line. The ground survey logged 4,222 damage indicators; nine houses rated EF5.

PRIMARY SOURCES

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

About the material. Video: U.S. National Guard footage released through DVIDS (Oklahoma National Guard: Lt. Col. Lindy White's uncut Black Hawk flight of 21 May 2013, plus ground and night search footage), public domain. Every radar image is rendered by Hook Echo from NOAA's public archive of KTLX Level II data (NEXRAD, unidata-nexrad-level2 on AWS); reflectivity uses the National Weather Service colour scale, and the correlation-coefficient scale is our own, with the 0.8 debris threshold marked on screen. The debris counts, the debris column and the date of the dual-polarization upgrade are our own measurements from those files. Every diagram is drawn in code from the sources listed. Nothing in this video is AI-generated imagery.

#HookEcho #Tornado #Moore2013