The Repository Hook Echo

Hook Echo: What the Tornado Shape on Radar Really Means

October 11, 2026 · Runtime 15:09 · Watch on YouTube

A security camera at Chickamauga Lock, Tennessee, filmed an EF2 tornado crossing the yard on 26 October 2010 -- and the radar 84 km away drew no hook at all. The hook echo, the shape this channel is named after, was first photographed by accident in Champaign, Illinois, in 1953. We render hooks from five radars, including one over Oklahoma City in 1998 that never produced a tornado, and go through the study that drove instrumented cars into thirty of them.

Chapters

  1. 0:00 A tornado on a security camera
  2. 0:24 The radar: no hook
  3. 1:21 Champaign, 1953: the first hook
  4. 2:37 A hook on a modern radar
  5. 3:17 What counts as a hook
  6. 3:36 The hook is rain
  7. 4:18 Moore 2013
  8. 4:58 Too shallow to see far away
  9. 5:55 A hook with no tornado
  10. 6:55 Thirty hooks, driven into
  11. 8:28 What did differ: the air
  12. 9:44 A tornado with no hook
  13. 11:39 What actually warns
  14. 12:22 When the hook is real
  15. 14:01 What the record supports
  16. 14:42 Next: Katrina, 2005

Transcript

A tornado on a security camera 0:00

This is a security camera at Chickamauga Lock, on the Tennessee River in Chattanooga, on the evening of October 26th, 2010. What crosses the yard is a tornado. The flashes are electrical equipment failing. The National Weather Service rated it EF2, winds of a hundred and twenty-five miles per hour. It tore the roof off an apartment complex and demolished a cement plant.

The radar: no hook 0:24

And this is the radar in northern Alabama, eighty-four kilometres away, in the scan that began at 6:41 that evening, with the storm over the dam. If you know one thing about tornadoes on radar, you are looking for a hook. There is no hook here. Just a broken line of rain.

This channel is named after that missing shape. So this episode is about it. Where the hook comes from, who first saw one, and what it tells a forecaster. And the two ways it misleads: a tornado with no hook, like this one, and a perfect hook with no tornado under it.

The Army Corps of Engineers released three angles from the lock's security system. One looks upstream across the lake. One looks over the yard where the tornado crossed. One looks downstream, toward the bridge. No sound, and no one behind the cameras. Just the record of what passed in front of them.

Champaign, 1953: the first hook 1:21

The first one was photographed by accident. On April 9th, 1953, the Illinois State Water Survey ran a research radar at the airport in Champaign, Illinois. It was built from a wartime aircraft set, and a 35 millimetre camera was mounted on its screen. It was normally switched off in the evening.

It was not a storm radar. The Water Survey used it with a network of rain gauges, to learn how strong an echo means how much rain. That evening the storm was north of the airport, with no rain between it and the radar to soak up the signal. The echo arrived clean.

That evening the technician, Don Staggs, stayed late to finish a repair. While he was testing it he noticed an unusual echo, curled at one end, and he started photographing the screen. North of Champaign-Urbana, a tornado was on the ground at that moment.

The survey's scientists matched the photographs to the damage along the tornado's path. Their report, the next year, made a careful prediction. Radar warning systems in tornado country, they wrote, might reduce the loss of life. In 1958, Tetsuya Fujita published a full analysis of those same films.

A hook on a modern radar 2:37

Here is the shape on a modern radar. Oklahoma, May 3rd, 1999, the storm from our first episode, as it crossed Bridge Creek. The main body of the storm, and hanging from its southwest corner, a curl of rain bending back on itself. That is the hook. And under it, at that moment, an F5 tornado.

The curl has a direction, and it is not random. In the Northern Hemisphere most supercell updrafts turn counterclockwise, seen from above. So the rain is wrapped around the back of the updraft in the same sense every time, and the hook bends the same way on almost every one of these screens.

What counts as a hook 3:17

Researchers needed a rule to decide what counts. The one in common use comes from a 1981 paper by Gregory Forbes. An appendage of echo sticking out at least sixty degrees to the right of the way the storm is moving. In the United States that usually means the south or southwest side.

The hook is rain 3:36

And it has a cause. A supercell is a thunderstorm whose updraft rotates. That rotating column, a few kilometres wide, is called a mesocyclone. Rain falling around it gets drawn into the spin and carried around the back of the updraft. The radar sees that rain as a curl. The hook is rain.

The dark notch inside the curl matters just as much. That is warm, moist air flowing into the storm, carrying no rain, so the radar sees almost nothing there. And on the back side, a downdraft wraps around the rotation and pushes the rain with it. It is called the rear-flank downdraft. Keep that name.

Moore 2013 4:18

Moore, Oklahoma, May 20th, 2013, from our second episode. A textbook hook, and at its tip a small round core of very strong echo. In that storm the radar could prove what the core was, and it was debris. A hook shows where the rotation is. It does not show the tornado itself.

This is what was under that one. The tornado was on the ground for thirty-nine minutes and destroyed more than eleven hundred buildings. The hook on the screen was a few kilometres across. The tornado inside it was a little over a mile wide at most, and most of the curl around it was simply rain.

Too shallow to see far away 4:58

There is a catch in seeing it at all. The hook is a shallow feature. The meteorological glossary describes it as usually reaching only three to four kilometres up. And a radar beam climbs as it travels, because the Earth curves away underneath it.

We computed the height of the lowest beam. Fifty kilometres from the radar it is about six hundred metres up. At a hundred, one and a half kilometres. At a hundred and fifty, more than two and a half. At two hundred, four. Far enough away, the beam passes over the top of the hook.

This is what that looks like. A storm in Kansas, May 12th, 1995, about a hundred and seventy kilometres from the Dodge City radar. The lowest beam is crossing it about three kilometres up, and the beam there is almost three kilometres wide. Whatever shape the rain had near the ground, this radar could not draw it.

A hook with no tornado 5:55

Now the other way it misleads. Oklahoma City, June 8th, 1998, just after nine at night. A storm crossing the west side of the city, with a curl hanging off its southern edge, close to the radar at Twin Lakes. A team of instrumented cars was driving inside that hook. It never produced a tornado.

It did rotate. The study that documented this storm found its strongest rotation on radar at 9:11, nineteen metres per second, only about three hundred metres above the radar. For an operational radar, that is a very low and very clear view. Still no tornado.

This is that storm in Doppler velocity, the same scan. Green is air moving toward the radar, red is air moving away, and where they sit side by side, something is turning. Here it is turning about three hundred metres above the radar, near enough to the city that the radar was looking at it almost level.

Thirty hooks, driven into 6:55

That study is the one to know. In 2002, Paul Markowski, Jerry Straka and Erik Rasmussen published thirty hook echoes they had driven into. Eighteen came with tornadoes, from the weakest to the most violent. Twelve did not. And they set the radar images side by side.

Their finding about the shape was blunt. In reflectivity, they could see no systematic difference between the hooks that produced tornadoes and the hooks that did not. And the tornadic ones were shown as they looked in the five minutes before the tornado formed.

And the study had a sharper result hidden in it. In eleven of the twelve storms that never produced a tornado, the cars found rotation at the ground itself. Wind turning around a centre, under the hook, measured by an instrument on a car roof. The one exception was Oklahoma City in 1998.

What they had that a radar does not was cars. A fleet of vehicles with instruments on their roofs, driven straight into the hook, sampling the air every ten to sixty seconds. They called it a mobile mesonet, and it was built for a field project in 1994 and 1995 named VORTEX.

Each car carried a fast thermometer, a humidity sensor, a wind sensor, a barometer and a GPS. Driven through the rear of the storm, a line of them draws a map of the air under the hook, a hundred metres to a kilometre apart. No radar can do that. Radar sees rain and motion in the air above, not the temperature at the ground.

What did differ: the air 8:28

The cars measured the air coming down in the rear-flank downdraft, the one that wraps the rain into the hook. And there the two groups did differ. In the storms that produced tornadoes, that descending air reached the ground about three to four degrees warmer than in the storms that did not.

Warm air at the ground is buoyant, so it can be drawn back up into the storm more easily. Air that arrives cold sits heavy and resists. The study found tornadoes were more likely, and stronger and longer lived, as that air got warmer and more buoyant. None of that shows up in the shape.

The study also pointed to where warm downdrafts come from. Days with very humid air near the ground, and so a low cloud base, were more likely to produce the warm, buoyant kind. That is something a forecaster can read from a weather balloon before the storm forms. The hook cannot show it.

The authors were careful with all of this. Thirty storms is a small sample, and direct measurements inside hooks were, and still are, rare. What they offered is a hypothesis: in the storms that made tornadoes, less of the descending air had been chilled by evaporating rain on its way down. Not a law. A direction to look.

A tornado with no hook 9:44

Back at Chickamauga, this is the upstream camera, looking across the water. The wall moving across the lake is wind and rain coming down out of the storm, flattening the surface as it goes. The radar drew this storm as a line, not as a hook.

This is the radar from 6:20 to 6:46 that evening. A long band of storms moving east across Tennessee, with breaks and bulges along it. The radar is eighty-four kilometres away, and its lowest beam crossed the dam about one point two kilometres up. Tornadoes can form inside lines like this, often briefly and hidden in rain. And they often have no hook to look for, because the storm is not built that way.

Here is the camera again, as it reaches the lock. In a few seconds the yard disappears into rain and debris. The Army Corps of Engineers reported two maintenance trailers destroyed, a government vehicle damaged, and the roof and electrical systems of its office hit.

This is the third camera, looking downstream toward the bridge. The water is calm. Then the far bank goes grey, the bridge fades, and spray runs across the lock wall. Inside that grey, somewhere close, is the tornado. This is what rain-wrapped means, seen from the ground.

The survey afterwards put it on paper. EF2. A hundred and twenty-five miles per hour. One point two miles long and a hundred yards wide. A real tornado, measured on the ground, from a storm whose radar picture never matched the textbook.

It was not the only one that day. The National Weather Service office in Morristown surveyed three more tornadoes in its area on October 26th, all of them weaker, rated EF0 and EF1. Chickamauga was the strongest, and the only one with a camera pointed at it.

What actually warns 11:39

That is the part of radar that actually warns. Doppler velocity, air moving toward the radar beside air moving away, close together. This is the pair from 1999, the same storm whose hook we just saw. Green coming in, red going out. The tightness of that pair is what forecasters look at.

Velocity has limits too. In our first episode we measured that this radar's velocity, that evening in 1999, could not go past twenty-six metres per second. The tornado's winds, measured by a research truck, were about five times faster. The pair says where to look. It does not give the wind speed at the ground.

When the hook is real 12:22

And sometimes the hook is real, and the proof sits at its tip. Joplin, Missouri, 2011, our last episode. A clear hook over the city, and inside it an echo stronger than rain usually gets. The forecasters on duty said this was their first sign of how big the tornado was. We measured that echo peaking at seventy-one and a half dBZ, in the scan that began at 5:48.

Without dual polarization, that core could have been hail. Two years later, in Moore, the radar could measure whether the echo looked like itself from pulse to pulse. Rain does. Debris does not. Here the correlation collapses exactly at the tip of the hook.

So the order of evidence is clear. The hook says rotation is organized and rain is wrapping around it. The velocity pair says how tight and how low that rotation is. And only the debris signature, where it exists, says that something on the ground is being torn apart.

And the debris does not stay inside the hook. In our second episode we followed that signature up through the radar's tilts over Moore. In the scan at 3:16 it reached about six kilometres above the ground. The hook is shallow. What a violent tornado lifts goes far higher than the curl of rain around it.

The prediction from Champaign came true in pieces. A national weather radar network followed in the late 1950s, and the Doppler network that recorded every storm in this episode was installed across the country in the 1990s. Every one of those radars is still drawing hooks.

What the record supports 14:01

Here is what the record supports. The hook is rain wrapped around a rotating updraft. It is a real sign, and the first one radar ever showed. But tornadoes form without it, and it forms without tornadoes. What decides is invisible to the shape: how fast the rotation is near the ground, and how warm the air is that comes down beside it.

This channel kept the name anyway. Because the hook is exactly what an instrument record is: a shape that means something, read by someone who knows what it cannot say. The camera at Chickamauga saw the tornado. The radar never drew one. Both were telling the truth.

Next: Katrina, 2005 14:42

Next, a hurricane that broke the instruments meant to measure it. In 2005, Katrina's storm surge knocked out tide gauge after tide gauge along the Gulf Coast. The highest surge in the federal record, twenty-seven point eight feet at Pass Christian, Mississippi, was read from marks the water left behind. That is the next episode. Katrina, 2005: the surge they measured afterward.

Description and sources

A hook echo is rain drawn into the spin of a rotating updraft (a mesocyclone) and wrapped around it by the rear-flank downdraft; it shows organized rotation aloft, not the tornado itself. It is shallow, usually below 3-4 km, and the lowest radar beam climbs with distance: about 0.6 km above the radar at 50 km, 1.5 at 100, 2.6 at 150, 4.1 at 200. Markowski, Straka and Rasmussen (2002) sampled 30 hooks with a mobile mesonet: 18 with tornadoes, 12 without, and found no systematic difference in the shape; in 11 of the 12 without tornadoes there was rotation at the ground. What differed was the descending air: about 3-4 K warmer in the tornadic storms. Tornadoes also form with no hook, like the EF2 at Chickamauga Dam (125 mph, 1.2 miles) inside a line of storms. Warnings come from Doppler velocity, and proof of damage from the debris signature.

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: U.S. Army Corps of Engineers security-camera video from Chickamauga Lock, Chattanooga (26 October 2010), and Oklahoma National Guard aerial footage of Moore (2013), released through DVIDS, public domain. Every radar image is rendered by Hook Echo from NOAA's public NEXRAD Level II archive (KHTX 2010, KTLX 1998, 1999 and 2013, KDDC 1995, KSGF 2011; unidata-nexrad-level2 on AWS), in the National Weather Service colour scale. Beam heights are our own computation. Every diagram is drawn in code from the sources listed. Nothing in this video is AI-generated imagery.

#HookEcho #Tornado #Radar