No Advance Warning for Fatal Clarksville Tornado

Investigative Report  
© 2023 Mike Smith Enterprises, LLC

Update: December 20....
Fox Weather, Nashville
The tornado's length turned out to be nearly 48 miles. Tragically, four people were killed by this storm. The NWS warning of this tornado left a lot to be desired. That story is told below. 

--- original posting ---

If I was a resident of Tennessee, I would be writing my congressional delegation demanding a congressional investigation into the National Weather Service's increasingly serious tornado warning issues -- and, that Congress create a Natural Disaster Review Board modeled after the National Transportation Safety Board.

Yesterday, I wrote about the tornado near Sharon, TN Saturday that was on the ground 13 minutes before a tornado warning was issued – even though it was an obvious warning situation. 


Today, I want to inform you that the fatal tornado that struck Clarksville, TN was not properly warned. Three people were killed and 62 injured. There was no tornado warning for Clarksville until after the storm was on the ground doing damage in spite of it being an obvious warning situation.  As tornado situations go, this was an easy one. 


I invite you to watch this heartrending interview conducted by Charles Peek of The Weather Channel. The man lost his brother-in-law in the storm. He stated there were just “seconds” between the tornado alarm activation and the beginning of the shaking of his building. 

As you'll recall from yesterday's investigative report, there was a tornado watch in effect for the area. There had also been a tornado warning to the west and southwest of Clarksville that did not include the metro area. Clarksville is an unusual metro area in that there is more or less continuous populated area from the word just south of the word "Clarksville" on the radar maps then on northwest to Ft. Campbell, Kentucky. 

The tornado was EF-3. EF-3 intensity is defined as a "strong" tornado by the National Weather Service (NWS). The NWS's published goal is to have a tornado warning out 13 minutes before a tornado develops. In this case, the "lead-time" (interval of time between when a warning is issued to when the tornado touches down) was, at best, negative two minutes as the tornado was on the ground before the warning was issued. A tornado warning for Clarksville should have been out no later than 1:25pm! I explain below. 

Here ist the radar data from 1:19pm. Please click to enlarge. 

The left panel shows the reflectivity data -- the type of weather radar you see on television. There had been a hook echo minutes before, but the circulation of the "mesocyclone" (a small area of lower pressure and circular winds) had caused it to wrap around 360°. Note the lighter color in the middle, which was the likely location of the tornado. On the right is the tell-tale "tornado vortex signature" -- a sure indication a tornado is present. Had I been at the warning desk of my company (WeatherData, Inc.) I would have instantly issued a tornado warning for that area and points northeast to Clarksville and, perhaps, beyond.

The Nashville radar, given the tornado watch, should have been operating in the 80-second "tornado SAILS" mode. It was unfortunately operating in 3-minute mode, so we next data we have 1:22pm. 
At this point, a tornado warning is easy and straightforward. At upper left, we have a "hook echo" which has been a known tornado indicator since the 1950's. At upper right, you see the tornado's rotation on the Doppler wind display. This type of tornado signature has been known since it was discovered by NOAA's National Severe Storms Laboratory in the 1970's. At lower right, lofted debris from the tornado being detected by the dual-polarization feature of the radar. We've known about this signature for a decade. Given 2-3 minutes to prepare it, there should have been a tornado warning for Clarksville by 1:25pm at the latest.

That no tornado warning was issued until 1:45pm is utterly shocking. Words do not readily come to mind as to how bad a meteorological performance this was.

Note: The intermediate radar data between 1:22 and 1:41pm has been placed in a supplement below.

At 1:41pm, when the NWS said the tornado touched down just north of Sabre Airfield, there is a hook echo present (I enhanced it with a thin, dark line) which looks like an upside-down candy cane. On the right is the rotation from the tornado. At this point, the tornado is on the ground in the Clarksville area. Again, there is no tornado warning!

I saw this and did what I could to warn the people of Clarksville. But, it isn't my job to issue tornado warnings. It is the job of the NWS.
The NWS  tornado warning appeared on Twitter moments after I posted the above. While the 
warning says it was issued at 1:43, the times I can find on a couple of the communication channels say it didn't become available until 1:45pm. 

By 1:43, the radar again showed a hook echo (upper left), rotation (upper right) and considerable lofted debris (lower right). 

The Director of the NWS in October flippantly disregarded these concerns in a television interview last month -- calling them, "Monday morning quarterbacking." There are three more people dead and others in hospitals because his agency often cannot issue tornado warnings in a timely and accurate manner even when the evidence is overwhelming a tornado is present. 

From 2000 to 2011, the people of the United States could count on the National Weather Service for high-quality tornado warnings. It met its self-imposed goals of 13 minutes of lead-time (average) and accurately warned of 73% of tornadoes. This is no longer true. Since 2011,  average lead time has dropped to just over 8 minutes and the percent of tornadoes warned in advance has dropped to 59% -- in other words, the NWS has regressed to pre-Doppler quality warnings (statistically).

The non-political National Transportation Safety Board (NTSB) revolutionized safety in aviation. I believe a Natural Disaster Review Board (NDRB), if by law, was:
  1. Non-political
  2. Had to focus on natural disasters only and could not delve into side issues like climate change (the U.S. already has the National Climate Assessment and others for that),
  3. Was located outside of Washington to minimize political interference, 
  4. Was staffed with appropriate subject matter scientists and social scientists,
  5. Had the power to subpoena all involved parties (including local and state emergency managers and private sector entities)
  6. Was not an enforcement agency just as the NTSB is not. It would recommend best practices in the field of disaster forecasting and recovery. 
... could revolutionize the way we handle disasters in the United States, save billions, and, most importantly, save lives! Please write your congressional Otherwise, major unwarned tornadoes will keep recurring. I fear another Joplin (May, 2011, when the NWS and local emergency management made a number of mistakes) -- with 161 tornado deaths--  is inevitable. 

Supplemental Radar Data 
The software I use is called "RadarScope" and it displays professional-quality radar data. The data below are all of the frames between 1:22 and 1:41pm Saturday.


  1. This has become nothing short of gross negligence. Negligence on the part of any government entity (local up to federal) can beget legal action, as well all know. It may be impossible, but since nothing else seems to affect change with the NWS perhaps a lawsuit from affected families could force change. I am not a huge fan of lawsuits, but they do have their place, and are a necessary evil when entities fail to act reasonably. I wish someone in the media would press the Secretary of Commerce on this, but it would probably be the same bucket of denial NWS/NOAA directors have offered up. /s/Bill Eckrich, Meteorologist

  2. I am on Clarksville Tennessee and no sirens sounded, it all happened so fast with like you said no warning other than something about possible rotation above the dover area of Clarksville but from what I understand the tornado was already on the ground ripping through the fort Campbell area when this warning was issued. I really hope something is done to help the warning systems because this was truly a fail

  3. I live in Clarksville and on my phone I got 3 alerts to a tornado watch and 2 warnings about 5 minutes before it touched down and 1 notice as it touched down.

  4. Mike Smith here: Thanks for the comment. Are there tornado sirens in Clarksville? From where did the two warnings come from? Thank you.

  5. Mike, I am a bit confused. According to the IEM archived warning database, the tornado warning that was issued by NWS Nashville at 1:22, centered on the shear and obvious tornado signatures near Dover, does not appear on your RadarScope maps. That warning (number 24 for Nashville) went into NW Montgomery county, or NW of Clarksville. However, I am having a tough time cross referencing between databases and zoom levels on the maps, as I am not familiar with this county warning area. If I am seeing things correctly, this warning did not include Clarksville, but at least showed that the warning forecaster was aware of the signature. As a retired NWS employee, I would like to believe that such an obvious signature would not have been missed. I do share your concern with the NWS’s recent poor warning performance and back your support for the need for an independent review of NWS warnings.

  6. Mike Smith here: I doubled checked my archive against the warning database at Iowa State University and the 48-hr warning retention feature at College of DuPage. I'm virtually certain my sequence is correct. If you look at my Tweet (above, time-stamped 1:45pm) you see a warning west of Clarksville but didn't include Clarksville.
    Interesting, there was a TOR warning west of Clarksville and across the border in KY for this storm -- but not for Clarksville in the middle.

    1. I think the source of yhe confusion may be

    2. The source of the confusion may be that the four panels are not showing Nashville tornado warning # 24 in effect at that time, covering the TDS.

    3. That said, obviously the warning forecaster lost situational awareness and let the feature track well out of the warning box. This is inexcusable and needs to be investigated.

    4. Mike here: I figured out the confusion. For some reason RadarScope only displays radar when in archive mode, not warnings. If you look at my urgent message (above) you see the warning to the west. I does not show up on the archived images. I will, right now, make a suggestion to DTN (which owns RadarScope) to add archived warnings.

  7. Nothing of substance will happen, unfortunately. No one has the stones to implement rigorous training using realistic scenarios. These things cost a lot of money to produce and implement and the NWS is beyond broke. Training is always the first thing to be cut in austere budget situations.

    If you miss an EF-3 practically in your back yard, that’s okay. Monday-morning quarterbacking. But if an employee misses a DEIA critical element on their performance plan, however, they’re PIP’ed.

  8. AccuWeather warned its clients in Clarksville about the same time as I identified in the (above) blog post:


Post a Comment

Popular posts from this blog

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

Dangerous Travel Conditions - People Reportedly Stranded

Dangerous Tornado Situation Developing Tuesday and Tuesday Night