The Repository Hook Echo

Bridge Creek-Moore 1999: How Radar Clocked 301 MPH

October 5, 2026 · Runtime 15:07 · Watch on YouTube

On 3 May 1999 a truck-mounted Doppler radar measured winds of about 301 mph, give or take 20, inside the tornado that crossed Bridge Creek and Moore, Oklahoma. In 2021 the same scan was reprocessed to 321 mph. We rendered the federal radar's own scans of that evening, straight from NOAA's raw archive, to show what each instrument could measure, and what it could not.

Transcript

7:27 pm, Moore 0:00

Seven twenty-seven in the evening, May 3rd, 1999. This is the radar that watches Oklahoma City, looking at a storm as it crosses the town of Moore. The bright curl at the bottom of it has a name. It is called a hook echo. At the tip of that hook, a tornado a mile wide is moving through houses at about the speed of a car on a city street. Half an hour earlier, a truck parked beside Interstate 44 had pointed a second radar at this same tornado, and written down a number nobody had written down before.

301, then 321 0:32

The number is usually quoted as 301 miles per hour. It is often called the fastest wind ever measured on Earth. In 2021 the scientists who took it revised it upward, to 321. Both numbers are real measurements. Neither one is the wind speed in the way most people picture it. This episode reads that evening the way the instruments recorded it. The federal radar that watched the whole storm. The mobile radar that caught the peak. And the damage survey that had to rate the tornado afterwards. They do not agree. The disagreement is the story.

A high-risk day 1:10

The day did not start as a famous one. The first outlook from the national Storm Prediction Center had central Oklahoma under a slight risk. At 11:15 in the morning it raised that to moderate. At 3:49 in the afternoon it raised it again, to high, the top of its scale. Fifteen minutes after four, the Norman office of the National Weather Service issued its first severe thunderstorm warning of the day.

The storm that matters here formed near Lawton, in the southwest of the state, around four o'clock. It became severe in a short time. At 4:47 the Norman office put the first tornado warning of the outbreak on it. Then it started moving northeast, toward Chickasha, and beyond it, toward Oklahoma City.

KTLX, the federal radar 1:55

The instrument you are looking at is a WSR-88D, a Doppler weather radar, one of about a hundred and sixty in the national network. This one is called KTLX. It sits at Twin Lakes, east of Moore, and in 1999 it swept the sky every five minutes. Every one of those sweeps was saved. NOAA keeps the raw files in a public archive, and the images in this episode were rendered straight from them.

The first thing the radar measures is reflectivity: how much of its pulse comes back. Rain sends some of it back. Big drops send more. Hail sends a lot. The colours run from green, for light rain, through yellow and red, to magenta at the top of the scale. Watch the southern end of this storm. The rain is starting to wrap around something.

The hook, and touchdown 2:47

That something is the updraft, the column of air feeding the storm, and it is rotating. Rain curls around the rotation and draws a hook on the screen. A hook does not prove there is a tornado on the ground. It shows where one is likely to form. At 6:23 in the evening, near Chickasha and Amber, one did.

Velocity and the couplet 3:08

The second thing the radar measures is velocity. Doppler radar can tell whether what it hits is moving toward it or away from it. Green is toward. Red is away. On the right, look at the tip of the hook. Bright green and bright red, touching. Air rushing toward the radar on one side and away on the other, less than a kilometer apart. That pair is rotation, and forecasters call it a couplet.

The ceiling: 58 mph 3:36

Here is what that picture cannot tell you. In every scan KTLX saved that evening, the fastest velocity in the file is the same: 26 meters per second, about 58 miles per hour, in either direction. That is not the wind. It is the edge of the scale. In the mode it was running, anything faster folded over and came back with the wrong colour. The federal radar could see the rotation. It could not measure the tornado.

And even if it could, it was looking from a distance. The radar beam leaves the dish almost flat, but the Earth curves away beneath it. When the tornado was in Bridge Creek, about 38 kilometers out, the lowest beam crossed it about 400 meters above the ground, and was over 600 meters wide. Whatever KTLX saw there, it saw as an average, from above the rooftops.

Warnings, 13 to 65 minutes 4:29

But seeing was enough to warn. The forecasters in Norman watched that hook and that couplet scan after scan. Grady County, where the tornado touched down, had its tornado warning 65 minutes before the tornado arrived. That night the Norman office issued 70 tornado warnings and 46 severe thunderstorm warnings across 32 counties.

Here are the four counties this one tornado crossed, and how much warning each had before it arrived. Grady, 65 minutes. McClain, 18. Cleveland, where Moore is, 31. Oklahoma County, 13. The government's own review of the event, published that August, called the average across the metro area remarkable.

The review listed the reasons. Forecasters who knew how to read the radar. A trained spotter network on the ground. And a change in how the office worked that night: several workstations with the same data, and the warning area split by geography, so each forecaster owned one piece of the map instead of everyone watching everything.

Bridge Creek 5:36

Now the tornado reaches Bridge Creek, a spread-out rural town along Interstate 44. This is where the damage survey would later find some of the most intense damage of its entire path. And this is where the second radar was waiting.

The Doppler on Wheels 5:51

It was called a Doppler on Wheels: a radar dish on the back of a truck, built by research meteorologists to drive close to tornadoes. Its beam is much narrower than KTLX's, and from up close it could scan the tornado a few dozen meters above the ground, not hundreds. Just before seven o'clock, near the interstate, it scanned the tornado and recorded the strongest winds it would measure that day.

The paper that published the result came out in 2007. It reported winds of 135 meters per second, about 32 meters above the level of the radar. In miles per hour, about 301, give or take 20. The plus or minus matters. It is part of the measurement, not a disclaimer. The number has been quoted ever since, usually without it.

What a radar number measures 6:40

So what does a radar number like that actually measure? The team that took it lists four limits. First, radar sees what the wind carries: rain, and in a tornado over a town, debris, which is heavier and slower than the air. Second, it measures above the ground, not at it. Third, each reading is an average over a volume of air, not a point. Fourth, it only sees motion straight toward it or away from it.

Revised to 321 7:08

All four tend to make the number too low, not too high. So in 2021 the same group reprocessed the 1999 data with improved methods, correcting for the averaging and for the motion the beam could not see, and published a new peak: 321 miles per hour. The scan did not change. The reading of it did.

Is it a record? 7:29

Is it the fastest wind ever measured? It depends on what you allow. Mobile radar has determined winds above 300 miles per hour in three tornadoes. Bridge Creek is one. El Reno, in 2013, was reported as a range up to 336, in small vortices that passed any one spot in a fraction of a second. Greenfield, Iowa, in 2024, came out at 309 to 318.

The first Tornado Emergency 7:59

Back to the radar that was warning people. At 6:57, with the tornado crossing Interstate 44 west of Newcastle, the Norman office sent out a statement that began with two words no National Weather Service product had used before as a headline. Tornado emergency.

This is the text, as the government's review printed it. A large tornado moving along Interstate 44. It will enter the southwest of the Oklahoma City metro area between 7:15 and 7:30. Persons in Moore and south Oklahoma City should take immediate tornado precautions. It went out on NOAA Weather Radio with the alert tone, 25 minutes before the tornado entered those areas.

Two minutes later the Storm Prediction Center replaced the tornado watch with one labelled a particularly dangerous situation. Oklahoma City television was already live. Two news helicopters flew alongside the tornado, and one station kept a box on screen listing the streets it would cross next, and roughly when.

The radar draws the path 9:02

At 7:11 the tornado crossed the South Canadian River into Cleveland County. The red line now drawing itself on the map is not the official damage path. It is the radar's own record. In each scan, we found the point where KTLX's velocity changed most sharply between neighbouring beams, and joined them up. Five minutes apart, they pass through each town in the order, and at the times, the Weather Service logged.

Moore, 7:25 pm 9:29

At 7:25 the tornado entered Moore. It crossed a part of the town that another tornado had damaged in October 1998, less than seven months earlier. On the radar, something changes here. At the tip of the hook the reflectivity jumps to the top of the scale, into the magenta.

68.5 dBZ 9:48

We measured it. This is the strongest reflectivity KTLX recorded within three kilometers of the rotation, scan by scan. Over open country it stays mostly between 45 and 60. When the tornado reaches the streets of Moore it climbs to 68.5. Rain alone rarely returns that much. Large hail can. So can a cloud of lofted debris.

Which one was it? In 1999 KTLX had no way to tell. Reflectivity says how much came back, not what it was. Telling debris from hail would need a different kind of radar, one that sends its pulse in two orientations and compares the echoes. That upgrade reached this radar years later. We will come back to it.

The damage survey 10:37

The third instrument is not an instrument at all. It is the ground. After the storm, survey teams walked the path and rated each stretch by what was left standing. In 1999 they used the Fujita scale, built in 1971, which ties each kind of damage to a range of wind speed.

This is the scale, as printed in the same government review. F5, incredible tornado: strong frame houses lifted off their foundations and carried away. Wind speed, 261 to 318 miles per hour. That is the top row. There is nothing above it on this page.

Five numbers, one tornado 11:17

Put the numbers side by side. The ceiling of the federal radar's velocity scale, 58. The bottom of F5, 261. The 2007 reading, 301. The top of F5, 318. The 2021 reading, 321. The reprocessed radar peak sits just above the highest wind the damage scale was written to describe. They were measured at different heights, so this is not a contradiction. It is a warning about what each number can say.

The survey found F5 damage in rural Grady County and in the suburbs. In Moore, whole neighborhoods were reduced to foundations. In all, 36 people were killed by this tornado, in the National Weather Service's count. Damage across Oklahoma that day reached about a billion dollars. It was the first F5 tornado to strike Oklahoma City.

Overpasses 12:15

The review also wrote down a lesson that is still repeated every spring. During the outbreak, people stopped their cars and climbed under highway overpasses. Two deaths were attributed to it. An overpass is no shelter from a violent tornado's wind and debris, and the report asked for every tornado safety leaflet to say so.

And it wrote down what worked. In Grady County, a man living among mobile homes heard the warning early and gathered his neighbors into his storm cellar. Thirty-five people crowded in. The tornado pulled the cellar door open. Every one of them survived. The mobile homes did not.

At West Moore High School an award ceremony was under way when the tornado came. The assistant principal moved everyone into interior hallways and bathrooms. The tornado tore the roof off the gym. Nobody at the school was killed.

72 tornadoes 13:10

Bridge Creek-Moore was one tornado out of many. The National Weather Service in Norman counted 72 over the two days of that outbreak. Thirty-three F0s, twenty-one F1s, eight F2s, seven F3s, two F4s, and one F5, the one this episode is about.

And it was not over. Between 8:30 and 9 that night, about 150 miles north, a violent F4 tornado struck Haysville and the south side of Wichita, Kansas. Its warning came from a radar signature too, and it set off the sirens. Six people were killed there.

Stronger than the ratings 13:50

There is one more thing the Doppler on Wheels changed. In 2021 its team published low-level winds from 120 tornadoes they had scanned. In their study area, the National Weather Service had rated about 1 percent of tornadoes EF4 or EF5. By the radar, about 1 in 5 had winds capable of that kind of damage. A damage rating can only be as high as what the tornado happened to hit.

What the record supports 14:16

So, how fast was the wind in Bridge Creek? Here is what the record supports. Faster than any wind the federal radar could measure that night. Fast enough for the top of the damage scale. And, a few dozen meters above the ground, at one moment, 301 miles per hour, give or take 20, or 321 as the data are read today. Nobody measured it at the height of a house.

Next: Moore, 2013 14:43

Fourteen years and seventeen days later, on May 20th, 2013, the same radar at Twin Lakes watched another violent tornado cross Moore. By then KTLX could send its pulse in two orientations, and this time it could tell debris from hail. That is the next episode. Moore, 2013: the day the radar saw the debris.

Description and sources

The Bridge Creek-Moore tornado was on the ground from 6:23 to 7:48 pm CDT, 38 miles long and up to a mile wide; the National Weather Service counts 36 deaths. The Oklahoma City radar, KTLX, scanned it every five minutes. In every one of its saved files the fastest velocity is 26 m/s, about 58 mph: the edge of its scale, not the wind. Over Bridge Creek its lowest beam passed about 400 m above the ground. Joining the point of sharpest velocity shear in each scan, the radar draws the tornado's route through Bridge Creek, Newcastle and Moore; at Moore the strongest echo near the rotation reached 68.5 dBZ, which 1999 radar could not tell apart from hail. Warnings came 13 to 65 minutes ahead, and at 6:57 pm the Norman office issued the first Tornado Emergency. The damage survey could only say F5: 261 to 318 mph, by the table in the government's own review.

PRIMARY SOURCES

NOAA National Weather Service, Service Assessment: Oklahoma/Southern Kansas Tornado Outbreak of May 3, 1999 (August 1999): https://www.weather.gov/media/publications/assessments/ok-ks99.pdf
NWS Norman, The Great Plains Tornado Outbreak of May 3-4, 1999: https://www.weather.gov/oun/events-19990503
NOAA NEXRAD Level II, KTLX, 3-4 May 1999 and 20 May 2013 (public archive): https://unidata-nexrad-level2.s3.amazonaws.com/?prefix=1999/05/03/KTLX/
Wurman, J., Alexander, C., Robinson, P. & Richardson, Y. (2007). Low-level winds in tornadoes and potential catastrophic tornado impacts in urban areas. Bulletin of the American Meteorological Society 88: 31-46. https://doi.org/10.1175/BAMS-88-1-31
FARM Facility (2024). Statement on preliminary peak wind speeds from DOW data, Greenfield, Iowa, 21 May 2024 (with the Bridge Creek revision to 321 mph, Wurman et al. 2021): http://www.farmfacility.org/pdf/best-greenfield-windspeed-note-2024-0623bp2.pdf
Wurman, J., Kosiba, K., White, T. & Robinson, P. (2021). Supercell tornadoes are much stronger and wider than damage-based ratings indicate. PNAS 118(14): e2021535118. https://doi.org/10.1073/pnas.2021535118

ARCHIVE PLATES — public-domain documents, with author, date, archive and licence.

Mike Eilts, National Severe Storms Laboratory — 1999-05-03 | Wikimedia Commons | Public domain
U.S. Department of Commerce, NOAA, National Weather Service — 1999-08 | NOAA National Weather Service | Public domain
Andrea Booher, FEMA — 1999-05-04 | Wikimedia Commons | Public domain
Andrea Booher, FEMA — 1999-05-04 | Wikimedia Commons | Public domain
U.S. Department of Commerce, NOAA, National Weather Service — 1999-08 | NOAA National Weather Service | Public domain
Andrea Booher, FEMA — 1999-05-04 | Wikimedia Commons | Public domain
Andrea Booher, FEMA — 1999-05-04 | Wikimedia Commons | Public domain
Andrea Booher, FEMA — 1999-05-04 | Wikimedia Commons | Public domain
U.S. Department of Commerce, NOAA, National Weather Service — 1999-08 | NOAA National Weather Service | Public domain
Andrea Booher, FEMA — 1999-05-04 | Wikimedia Commons | Public domain
Andrea Booher, FEMA — 1999-05-04 | Wikimedia Commons | Public domain
National Weather Service Norman, Oklahoma — 2009-11-24 | Wikimedia Commons | Public domain

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

About the material. Every radar image is rendered by Hook Echo from the public NOAA archive of KTLX Level II data (NEXRAD, unidata-nexrad-level2 on AWS), with the National Weather Service colour scales; county lines and interstates are U.S. Census TIGER data. The couplet track and the reflectivity series are our own measurements from those files; the 7:07 pm volume is truncated in the archive and is skipped. Photographs: NOAA/NSSL (Mike Eilts) and FEMA (Andrea Booher), public domain. Every diagram is drawn in code from the sources listed. Nothing in this video is AI-generated imagery.

#HookEcho #Tornado #NEXRAD #Oklahoma #WeatherHistory