Joplin 2011: The Tornado Warning 17 Minutes Early

Chapters

1. 0:00 Joplin from the air
2. 0:23 The radar at 5:38
3. 1:08 The forecast
4. 1:55 The first storms, in Kansas
5. 2:39 Two warnings and a siren
6. 3:23 A radar 100 km away
7. 4:19 Folded velocity
8. 5:02 Touchdown, 5:34
9. 6:04 The debris ball
10. 7:17 The path through the city
11. 7:50 St. John's and the stores
12. 9:20 How strong was the wind?
13. 10:01 What people did
14. 11:02 The protective bubble
15. 11:49 False alarms, and the sirens
16. 12:28 What made people move
17. 13:30 The whole clock
18. 14:30 What the record supports
19. 14:52 Next: the hook echo

Transcript

Joplin from the air 0:00

This is Joplin, Missouri, filmed from a National Guard Black Hawk in the days after May 22nd, 2011. A tornado crossed six miles of this city in thirteen minutes. It damaged nearly eight thousand buildings. The federal investigation counted one hundred and sixty-one people killed, the most for any single tornado since official records began in 1950.

The radar at 5:38 0:23

And this is how the radar near Springfield saw it, in the scan that began at 5:38 that afternoon. The hook of the storm is wrapped around Joplin, and at its tip sits a small, intense blot. Twenty-one minutes earlier, the National Weather Service had already issued a tornado warning covering the whole city. Seventeen minutes before the tornado touched down.

Seventeen minutes is more than the national average. The federal assessment called Joplin a warned event. So this episode lays the record of that afternoon against the clock. What the radar could see, and from how high. What the warnings and the sirens said, and when. And what a hundred interviews found about what people did with those minutes.

The forecast 1:08

The storms were not a surprise. On Friday, two days ahead, the Storm Prediction Center already gave southwest Missouri a thirty percent chance of severe weather. On Sunday morning at 7:55 it raised the area to a moderate risk, with a ten percent chance of a strong tornado nearby. At 1:30 in the afternoon came a tornado watch, about four hours before Joplin was hit.

Through the morning, the forecasters in Springfield expected very large hail to be the main danger, with only a small chance of tornadoes. At 1:06 the Storm Prediction Center wrote that the wind shear near the ground was a bit marginal, but more than sufficient for tornadoes, given extreme instability. By mid-afternoon it was warning of tornadoes that could form one after another.

The first storms, in Kansas 1:55

The first storms formed in southeast Kansas between two and three o'clock and quickly turned severe. At 4:25 and again at 4:51, Springfield issued the first tornado warnings of the day, for counties west of Joplin. At 4:33 a forecaster called the emergency manager of Jasper County, which contains Joplin, to brief him on what was coming.

In Springfield, the office had six people on shift, two of them working the radar. The lead radar operator noticed that the storms were moving further to the right of the wind than forecast, a sign of more spin near the ground than the analyses showed. That, and two cells merging over southeast Kansas, led directly to the early warning, the assessment says.

Two warnings and a siren 2:39

Then the timeline gets crowded. At 5:09, a tornado warning for a different storm, covering the northeast corner of Joplin. At 5:11, Joplin's sirens sound for three minutes, set off by reports of funnel clouds in Kansas. At 5:17, a second warning, this time for the storm that mattered, covering all of the city.

Joplin had twenty-five outdoor sirens. Its policy was to sound them once, in a single three-minute blast, and then stop. Several people told the assessment team they took the silence afterwards to mean the danger had passed. Others heard the siren, then heard on television about a warning seven miles north, and decided it was for someone else.

A radar 100 km away 3:23

The radar watching all of this stood near Springfield, about a hundred kilometres east of Joplin. Here is the whole storm complex at 5:24, ten minutes before touchdown. Two other radars, in Tulsa and near Kansas City, were also used for the warning decision. None of them was close.

The Earth curves away beneath a radar beam. We computed where Springfield's lowest beam crossed this tornado. Its centre passed about one point six kilometres above the ground. More than five thousand feet. And at that range the beam itself was about as wide as the tornado was.

It also took its time. That afternoon the radar ran a scanning pattern that finished one complete volume every four minutes and fifty seconds. The assessment noted that a faster pattern, at about four point two minutes, would have given roughly four more scans in the ninety minutes that mattered.

Folded velocity 4:19

And the Doppler velocity, the measurement that shows rotation, was compromised near Joplin, the assessment says. We checked. In every scan we decoded, the strongest motion around the tornado sits pinned at the radar's own limit, thirty-three and a half metres per second. That is folded data, not a measurement.

The second radar used for the warning, in Tulsa, had the mirror-image problem. Its velocity data was obscured over Joplin itself, at the height of the tornado. The assessment traced both failures to the same choice: scanning modes that hold the pulse rate fixed at the lowest angles, which blinds the velocity at certain distances from the radar. It recommended a different mode.

Touchdown, 5:34 5:02

Between 5:29 and 5:34 the rotation near the ground tightened very fast. At about 5:34 the tornado touched down just southwest of the city limits. In the radar scan that began at 5:34, the strongest echo near it was fifty-four dBZ, heavy rain. Nothing on the screen yet said how large this was.

And from the ground there was little to see. The tornado was wrapped in rain. The assessment collected account after account of people who could not recognize it until it was very close, and who ran for shelter in their homes just as it struck, despite a warning that had gone out a quarter of an hour before.

At 5:36 it reached the western suburbs, and at 5:38 it crossed Schifferdecker Avenue, where the damage became EF4. At that moment Joplin's sirens sounded a second time. Because reports were coming in of a tornado on the ground, the emergency manager broke the one-time policy. The tornado was already inside the city.

The debris ball 6:04

This is the scan that began at 5:38, the one the assessment refers to as 5:39. At the tip of the hook, over western Joplin, the echo jumps past sixty dBZ. The forecasters on duty later said this was their first indication of the size and strength of the tornado. The assessment calls it a very large and distinctive debris ball.

Here is that echo, one point per scan. Fifty-four at 5:34. Sixty-four and a half at 5:38. Sixty-seven at 5:43. And seventy-one and a half at 5:48, as the tornado crossed the middle of the city. For comparison, rain alone rarely goes much past fifty-five. Sixty usually means hail, or something that is not weather at all.

That is also the limit of what this radar could say. In 2011 it measured how strong the echo was and how it moved, not what shape the targets were. Hail and pieces of houses can return the same number. Two years later, over Moore, a radar with dual polarization could tell them apart. Over Joplin, the radar could only show that something enormous was there.

The path through the city 7:17

Under that echo were neighborhoods. From the helicopter, the path is a band of houses stripped to their frames, running east across the city, with the streets still marking where the blocks were. Inside Joplin the tornado moved at about twenty miles an hour.

The investigators mapped who lived inside the damage. About twenty thousand eight hundred people, forty-one percent of the city. Six miles of Joplin in thirteen minutes means that for any one street, the worst of it lasted well under a minute, and the warning had been on the air for twenty minutes or more.

St. John's and the stores 7:50

This is St. John's Regional Medical Center, hit at about 5:40. Its two towers, seven and nine stories, kept their frames standing. The tornado tore off the outer walls and windows, and wind, rain and debris went through the inside. The federal investigators found that the building lost all of its function anyway. It never reopened as a hospital.

Further east, the tornado reached a large retail area. A Walmart and a Home Depot collapsed in large parts. In both, the light roof lifted first, and the tall walls it had been bracing fell. Both stores had designated refuge areas for tornadoes. The investigators found no evidence that those areas made a difference.

The investigation counted the buildings. Seven thousand four hundred and eleven homes damaged, three thousand and sixty-nine of them heavily or totally. Five hundred and fifty-three other buildings, and every one of those took heavy damage. Among them a hospital, ten public schools, twenty-eight churches and two fire stations.

The tornado crossed Interstate 44 on the east side of the city and kept going, weakening, for another twelve miles. It lifted at about 6:12. On the ground for twenty-two miles, and up to a mile wide. The worst of it, the EF4 and EF5 damage, filled a band about six miles long, through Joplin and Duquesne.

How strong was the wind? 9:20

So how strong was the wind? The rating is EF5, winds over two hundred miles per hour. But no instrument measured it. The nearest radars were more than sixty miles away. The federal investigators worked from fallen trees and damaged buildings instead, and arrived at one hundred and seventy-five miles per hour, give or take a quarter.

The method is to map which way thousands of trees fell, and then find the wind that would knock them down in that pattern. It also showed something the rating hides. In the Joplin tornado, as much as ninety percent of the area hit, and about forty percent of the deaths, were under winds no stronger than an EF3.

What people did 10:01

Two weeks after the tornado, the assessment team went back to Joplin and conducted nearly a hundred interviews. Survivors, emergency managers, broadcasters, dispatchers, people who worked in the stores and the hospital. They were trying to answer one question: what did people do with the warning?

The answer was consistent. Most people did not take shelter at the first sign of danger, which for most of them was the siren. They went looking for more information first. They turned on the television, stepped outside, called someone. They needed a second, stronger signal before they believed the first.

One man described nine separate signals before he took cover. He heard the first siren while driving to a restaurant. That restaurant had locked its doors. He drove to a second one, where business was carrying on as usual, sat down, and ordered dinner. Only when a patron said there was a tornado in Joplin did the staff move everyone to shelter.

The protective bubble 11:02

Behind the waiting was a belief. People told the team that storms always missed Joplin, that the city sat inside a protective bubble. The record shows why that felt true. Since 1950 the city itself had been hit by one strong tornado, in 1971. Within eighty miles of it, one hundred and eighty-two.

The same assessment put national numbers beside it. Across the country, from late 2007 to early 2011, three out of four tornado warnings ended without a tornado. Yet the system almost never missed the violent ones. Ninety-four percent of EF3 to EF5 tornadoes were warned, on average almost eighteen minutes ahead. A warning just did not say which kind it was.

False alarms, and the sirens 11:49

And residents said the sirens went off all the time. Here the federal investigation could check the instruments' own record. Of the eighteen tornado warnings issued for Joplin since 2005, fourteen ended without a tornado. From 2007 through 2011, the false alarm rate for Joplin was ninety-two percent.

But the sirens were a different story. Since 2007 Joplin had sounded its sirens for weather about once a year at most, not counting the one-minute test every week. The false alarms people remembered were real. They were in the warnings, not in the sirens.

What made people move 12:28

What moved people, according to both reports, was anything that broke the routine. The second siren, which the city's policy did not call for. Broadcasters dropping their regular programming to say take cover now. And seeing the tornado itself, which was hard, because it was wrapped in rain until it was very close.

And then there was where to go. Eighty-two percent of homes in Joplin had no basement. The city had no public storm shelter. The investigators found no death among people who were in the basement of a house that was destroyed. Most people found the best shelter they had, and in many buildings in the path, that was not enough.

One more channel was never used. Jasper County had a reverse 911 system that could call every phone in the area. It stayed off that afternoon, because in earlier emergencies it had taken up to three hours to get the calls out. And the investigators found that people over sixty died at about four times the rate of everyone younger.

The whole clock 13:30

Line the clock up one more time. Warning at 5:17. Touchdown at 5:34. The second siren at 5:38. The debris ball on radar at 5:39. The first sign of the tornado's true size reached the forecasters at almost the same moment the tornado reached the city.

The assessment made two recommendations that came straight from those minutes. Warnings that say how bad, not just what, so that a violent tornado does not sound like every other warning. And radar scans that come back to the lowest angle every minute, because, in its words, the average tornado lives shorter than one full scan.

This is the same ground in January 2013, twenty months later. The last debris is being cleared, and new houses stand on lots that were slabs. The final count of people killed, in the federal investigation, is one hundred and sixty-one, and more than a thousand injured.

What the record supports 14:30

Here is what the record supports. The forecast was right days ahead. The warning came seventeen minutes early. The radar was a hundred kilometres away, scanning a mile above the ground every five minutes, and it showed the size of the tornado only after it was in the city. And the people under it needed one more signal than the system was built to give.

Next: the hook echo 14:52

Next, the shape this channel is named after. On April 9th, 1953, a radar technician in Champaign, Illinois, stayed late to finish a repair and photographed an echo curled like the number six. It was the first tornado ever documented on radar. That is the next episode. The hook echo: what the shape means, and what it does not.

Description and sources

The forecast was right days ahead and Tornado Warning #31 went out at 5:17 PM CDT. The radar was about 100 km away: its lowest beam crossed the tornado about 1.6 km above the ground, one volume every 4 min 50 s, with the velocity folded at its 33.6 m/s limit. The echo near the tornado went from 54.5 dBZ at touchdown (5:34) to 64.5 in the 5:38 scan and 71.5 at 5:48 -- the 'debris ball' that first showed the forecasters its size. Federal interviews found most residents waited for a second, stronger signal. They remembered constant false alarms: the warnings for Joplin were wrong 14 times in 18 since 2005, but the sirens had sounded for weather about once a year. NIST counted 161 killed, 7,411 homes and 553 other buildings damaged, and estimated the peak wind at 175 mph, give or take 25%.

PRIMARY SOURCES

• NWS Central Region Service Assessment: Joplin, Missouri, Tornado -- May 22, 2011 (July 2011). https://www.weather.gov/media/publications/assessments/Joplin_tornado.pdf
• NIST NCSTAR 3 (2014). Final Report, NIST Technical Investigation of the May 22, 2011, Tornado in Joplin, Missouri. https://nvlpubs.nist.gov/nistpubs/NCSTAR/NIST.NCSTAR.3.pdf
• NOAA NWS NEXRAD Level II, KSGF, 22 May 2011: https://unidata-nexrad-level2.s3.amazonaws.com/?prefix=2011/05/22/KSGF/
• NOAA NCEI Storm Events Database, 2011 (the three segments of the Joplin tornado): https://www.ncdc.noaa.gov/stormevents/

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: Missouri National Guard, U.S. Army Corps of Engineers and Missouri National Guard Public Affairs footage released through DVIDS, public domain. Every radar image is rendered by Hook Echo from NOAA's public archive of KSGF Level II data (NEXRAD, unidata-nexrad-level2 on AWS), in the National Weather Service colour scale. The beam height and the echo series 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 #Joplin #Tornado