Matador: Another Fatal Tornado Without Adequate Warning

After initially believing the warning for last week's Matador, Texas, Tornado was adequate, my opinion has -- strongly -- changed.

Investigative Report
© 2023 Mike Smith Enterprises, LLC

After receiving suggestions from chasers in the field and others, I've reviewed the National Weather Service (NWS) "chat" from the Lubbock and Amarillo field offices. Those concerned individuals are correct. This incident is just inexplicable.  

It was Lubbock that had the responsibility to provide advance warning for Motley County on June 21, 2023. Keep in mind the National Weather Service's stated goal is 13 minutes' advance warning for a tornado. From 2000 to 2010, the NWS routinely achieved that goal. It has regressed considerably since. 

The first issue was -- yet again -- the radar was being improperly operated during a tornado watch. Instead of operating the radar at 80 second data collection intervals, the radar was set to collect data at 360 second intervals -- six minutes between scans! When the radar is set at the more frequent interval, it is possible to provide emergency storm warnings more quickly. 

Here is a timeline of the meteorological events that evening in west Texas:

7:40pm CDT Radar
At left is the reflectivity data (the type of radar you see on TV). The storm is moving nearly due south (arrow). The developing "hook echo" (circle) is a sign of a tornado but there is a weak area in the center. That is evidence of a rotating updraft spinning precipitation (raindrops and hail stones) out of its center -- which is another precursor to a tornado. The velocity data (right) is contaminated by what meteorologists call "side lobes." However, using the radar's next step up vertically (not shown) clearly shows strong rotation. 

A tornado warning should have been issued by this time for northern, or all of, Motley County. 

At 7:42, the Lubbock NWS meteorologist received a report of a utility pole leaning at a 45° angle due to high winds (caused a tornado or otherwise) from this storm. 

7:45pm Radar
At this time, the radar is screaming, "issue a tornado warning!"
At left, here is now a definite hook echo (red). The hail forms the shape of an arc which means the supercell's updraft is strong and able to be concentrated into a strong tornado. At right, the velocity data shows strong rotation in the forward part of the hook, which is the classic tornado signature. 

But, instead of issuing a tornado warning, the Lubbock office issued a severe thunderstorm warning for hail and high winds. It has the option, in that type of warning, to add the phrase, "tornado possible." It did not. The tornado is just nine miles from the town and closing. After the fact, the Lubbock NWS chooses not to survey this area. I won't speculate on the reason but a survey should have been conducted. I am confident it would have demonstrated a tornado was on the ground with no warning,  

Meteorologist Kevin Selle was live-broadcasting the tornado on his video stream. We know the NWS was aware that the tornado was being broadcast because it is on their Amarillo chat! Amarillo states they are passing that info on to Lubbock.
12:49 UCT = 7:49pm CDT.
This was Amarillo NWS's office's chat. 
Adding together: the radar was on the wrong scan strategy + what the Lubbock office knew from the 7:45pm radar data + plus the knowledge a tornado was already on the ground, to have not issued a tornado warning for Motley County = a level of ineptitude even beyond the previous cases I've documented. 

Instead of issuing a tornado warning, the Lubbock NWS meteorologist notes in the chat (but not for the public, only on the internal chat with TV and other meteorologists), "Very, very strong updraft coupled with solid mid-level rotation." While he adds "tornado possible" to the severe thunderstorm warning already in effect, the public does not see that addition.

It is simply incredible that a tornado warning had not been issued. 

7:51pm Radar
This is the type of radar image of a tornado they put in college meteorology texts. 
I have outlined the hook (left) in the reflectivity data. Classic. At right, the velocity data shows the tight rotation (known as a "gate-to-gate couplet") is centered on the leading edge of the hook. This is what they teach students is a tornado as sensed by radar. The tornado is just six miles from the town and moving in. 

Keeping in mind they are aware that Kevin Selle has been broadcasting the tornado -- live -- the following statement on the chat is inexplicable:

Lubbock NWS chat
12:51 UTC = 7:51pm CDT
Rear flank downdraft appears to be quite strong with a sizable downburst wind threat. Circulation 'not sufficiently tight' to suggest a tornado but isn't far from that. Storm bears close monitoring.

This was another inexplicable statement. Radar meteorology 101 tells us that gate-to-gate is as tight as it gets. And, how is this data not "suggest[ing]" a tornado? One is being broadcast.

Lubbock's chat has the following reports of a tornado in progress:
  • Chaser Daniel Shaw is 3 mi. NNW of Matador, a tornado is 2 miles to his northeast. 
  • Corbin Voges, a storm chaser, has a brief video (note it is the video that is brief) of a tornado north of Matador.
  • Media meteorologist Matt Massey points out to the NWS, which has responsibility for issuing a tornado warning, writes that the velocity data (right panel of above radar image) is "interesting." I believe that was a diplomatic attempt to prod Lubbock to issue a tornado warning. 
  • Sammy Brence, a chaser meteorologist, was reporting a "large tornado" 3 mile NNW of Matador and that is rain-wrapped, meaning it cannot easily be seen. 
The NWS -- at long last -- issues a tornado warning for southern Motley County even though there is a tornado in progress in the center of Motley Co., 6 mi. north of Matador.
Note they do not disclose there is a tornado in progress. They note a "wall cloud" which implies a tornado has not formed yet. At the bottom where it says "TORNADO" they can put "OBSERVED," "RADAR CONFIRMED" or "RADAR INDICATED." Inexplicably, they chose "radar indicated."

Radar shows the tornado is now just 3 miles from the city but, per reports, the tornado may have been invisible from Matador because it was wrapped in rain. This seems plausible because the hook is rather thick at this point. 
The right panel, the wind velocity data, shows the strong rotation still gate-to-gate. 

Storm chaser reports at 7:57pm:
  • Ben Holcomb is 3 mi. NNE of Matador reporting the tornado is approximately 2 miles north of his position. 
The Lubbock NWS meteorologist writes on chat -- the public does not see this  -- "multiple confirmations of a tornado on the ground in Motley County."

7:58pm CDT
The Lubbock meteorologist updates the tornado warning to an "observed" tornado and says it is 4 miles north of the town. 

Reports of the tornado continue to be received, including from the Motley Co. sheriff's office. 

The radar mode, now that the tornado is on Matador's doorstep, is switched to 180 second mode. You'll recall it had been set to 360 second mode. This is an improvement but, unquestionably, should have been set to 80 second "tornado mode."

Skipping Ahead to the 8:03pm radar 
Based on the radar, the tornado has reached Matador. 

We can zoom in and add an additional type of data for confirmation. 
If you look closely, the streets of Matador show up as thin blue lines. The velocity data (upper panel) shows the center of the couplet which corresponds to the location of the tornado is at the northernmost street of the city (give or take a small distance).

The lofted debris data (known as a lowered "correlation coefficient") shows lofted debris over the city. So, within a minute or so, 8:03pm is the time of arrival of the tornado in Matador. 

Later Reports
  • 8:14pm, law enforcement reported that, as of 8:11, there was damage in the City of Matador.
  • 8:14pm, Val Castor's (well-respected storm chaser from Ch 9 in Oklahoma City) video feed showed severe damage on the north side of Matador. 
From the News Media

Barbara, who recently moved to Matador, recalled little warning when the tornado hit



Nicky Dempsey was heading home with her daughter when she noticed how dark the sky was to the north of her house. She thought it was hail or rain, since she didn’t see anything on the radar. Then she heard the faint sound of sirens and told her family to get in the basement as quickly as they could. 

                                                 --- Via Texas Tribune

Four people were killed and 9 were injured by the tornado. The tornado was indeed "strong" -- EF-3 intensity. 

The city depended on the NWS for a warning because, per storm chaser reports, the tornado was invisible due to it being surrounded by rain. Had the tornado been visible, the emergency manager may have had the option to operate the sirens independently.

Tornado Warning Lead Time
The tornado warning was issued at 7:56pm. Given Selle's of the video of the tornado in northern Motley county, at best, the lead time would be -7 minutes. However, that is really giving the NWS the benefit of the doubt as they never issued a tornado warning for that area.

For the City of Matador, the National Weather Service's tornado warning was issued at 7:56pm. With the time of arrival in the city of 8:03pm, that would mean the lead time was +7 minutes. Given the NWS's published goal of 13 minutes, this was far shorter than it could have been. 

Had the tornado warning been issued at 7:40pm -- which is when it should have been issued -- the lead time would have been 23 minutes. 


On May 22, 2011, the NWS botched the warnings for the Joplin Tornado. One hundred sixty-one people lost their lives. While a storm of that intensity crossing a densely populated area would have caused dozens of deaths, the extreme loss of life was confirmation of the inadequate warnings as it, too, was rain-wrapped and invisible as it crossed the city. 

Unfortunately, the National Weather Service seemed to learn the wrong lessons from that storm. And, since, the tornado warning program has fallen apart due to those program issues combined with the retirement of its well-trained meteorologists. 

The service’s own figures demonstrate that their warning accuracy has deteriorated by 24%. In addition to the decrease in "probability of detection" (whether a tornado warning is out before the tornado touches down), the lead time (the interval between and warning and the arrival of the storm) has collapsed by a whopping 58% (below). 

Since 2020, we don't know what has occurred, although I suspect continued deterioration. The accuracy figures, when they were good, were out in the open. Now, they are behind a password and login. 


The bottom line: people are dying because of these issues. 

And, I am terrified, as are other meteorologists, that -- absent strong, immediate action -- another Joplin is inevitable. 

Now that I have documented the Matador storm and the attendant warning issues, I am going to write up a summary of these issues and my recommendations for repairing them. I hope to have it by the end of the week.


Popular posts from this blog

[1:10am Update] Tornado Forecast for Rest of the Night

First Tornado Watch of the Day Issued

Hilary's Forecast Path Shifts West; Updated 9:20am PDT