Another Fatal, Poorly Warned Tornado

[Mike Smith]

Unfortunately, we've had another poorly warned, fatal tornado. It was in Perryton with 3 dead and 50+ in hospitals.
Here is my report: www.mikesmithenterprisesblog.com/2023/06/again-fatal-tornado-strikes-perryton.html

The question is, "What can we do to fix this?"

I've been working with some high level people and two congresspeople on trying to fix it. However, since the NWS is in denial, it is going to take much more than my efforts. Please contact your congresspeople or others with influence. Otherwise, they will just continue.

Thank you.

 

[Andy Wehrle]

I think the Matador warning wasn't great either, coming out only about 10 minutes before the town was hit (still much better than Perryton). Add in heavy rain-wrapping and a radar signature not really conclusive of such a high-end tornado, and you have a recipe for tragedy.

 

[Mike Smith]

I think the Matador warning wasn't great either, coming out only about 10 minutes before the town was hit. Add in heavy rain-wrapping and a radar signature not really conclusive of such a high-end tornado, and you have a recipe for tragedy.

I have been contacted a meteorologist in Texas who says there were a number of issues in the run-up to that tornado warning being issued. Again, for fear of upsetting the NWS, he asked for confidentiality and has not contacted anyone else.

This is one of the problems: mets and chasers across the USA are realizing that major issues exist with the NWS tornado warning program and that -- as one wrote yesterday -- "I had no idea things had deteriorated this much." But, no one (other than yours truly) is willing to go on the record. I'm not sure I understand this extreme reluctance, do they fear 'the NWS police?'

I would argue that, as professionals, we have a greater obligation to the public at large to "call it as we see it." At bare minimum, we should be contacting our congressional delegations. Any of you are welcome, in correspondence to congresspeople, to use the numerous links on my blog.

For those unfamiliar: I'm not worried particularly about unwarned EF-0's. I'm highlighting tornadoes that we would reliably warned of 15 years ago -- many of which have been EF-2/3 intensity and/or have caused fatalities.

 

[Andy Wehrle]

One thing your observations have taught me is, don't count out a storm while chasing just because it doesn't have a tornado warning on it!

 

[John Farley]

Mike knows a lot more about this than I do, but one thing that strikes me is that a number of the storms recently have been quite far from the radar sites. This has probably led to errors in both directions. I was out chasing on the 16th and 17th, going after storms that had persistent, strong rotation signatures on radar and got multiple tornado warnings, but ended up not producing tornadoes. The Springfield, CO storm on the 16th had a very persistent strong couplet on radar, and got 3 tornado warnings. But there was no tornado, although the storm did drop 4-inch hail. No doubt it was a potent supercell, but no way to tell from radar what was going on near the ground, given the distance from any NEXRAD site. It was a similar case with the Beaver, OK storm the next day, which again got 3 tornado warnings but apparently did not produce, at least until much later when it was absorbed into the surging squall line/derecho. Chasers who did better than I did got great video of rapid rotation of the wall cloud with this storm near Beaver, but even from my less desirable position 15 miles west of the storm there was obvious rotation in the wall cloud and at times, brief funnel clouds. So again, a very intense supercell, but no tornado near Beaver. Problem is, both of these storms were far from any radar site. So while rotation was obvious in the middle part of the storm via radar, and visually also at the cloud base/wall cloud, there was no way to measure by radar what was going on near the ground.

The Perryton storm (which happened on a day when I could not chase due to competing commitments), was the opposite case - no warning issued but a devastating tornado. Again, far from any radar site, so no way to tell what was going on near the ground via radar. That said, chasers on that storm said it went from nothing to rotating wall cloud to strong tornado in a very short time, so it might have been tough to accurately warn for even if it was closer to the radar site. Of course, as Mike points out in his blog post, having the radar on 80 second refresh might have helped, though near-ground rotation would have been hard to verify given the distance from the radar site.

It seems to me that the only real solution is to have a lot more radar sites. Perhaps something smaller and less expensive than a full NEXRAD site but with Doppler capabilities, located in places that are currently distant between the NEXRAD sites. I don't know what the cost of this would be, but it might be better than some of the ways money is being spent now. Perhaps another way would be to use mobile Doppler radars to fill these gaps, positioning them in areas in the gaps between NEXRAD sites in the areas with the greatest tornado risk on any given day. I don't know if the current mobile Dopplers would be practical for this or not, but the concept might be worth pursuing.

 

[JamesCaruso]

One thing your observations have taught me is, don't count out a storm while chasing just because it doesn't have a tornado warning on it!

There is also a chasing safety lesson - over-reliance on radar that may be delayed if in 240-second scan mode…

 

[JamesCaruso]

@Mike Smith in your blog post you link to here, under the “Radar Operation” section title, I think you intended to refer to an “inexperienced” meteorologist, not “experienced” (I say this from the context of the rest of the sentence and paragraph, not from any knowledge of the individual or the situation!)

 

[JamesCaruso]

Not being in the profession, it’s hard to imagine inadequate radar training being an issue, when even an amateur like me can interpret it. In addition, don’t the radars used by NWS actually visibly highlight TVS’s etc. with markers on-screen? I remember hearing once that there is even an audio alarm, is that not true?

I would think the bigger issue is over-correcting in an attempt to decrease the FAR. I guess that’s part of what you mean by “inadequate storm warning strategy training”?

 

[Mike Smith]

Fixed, Jim. Thank you.

 

[Mike Smith]

Thanks for everyone chiming in on this vital issue. I have some comments that I hope will be helpful.

With regard to distance from radar: If anything the radar data available to the NWS is better today than it was 15 years ago. There are radars at MLU and COU. The -88D's are now dual-polarized so lofted debris can be seen. There is a network of C-band radars in western North Dakota available to them.

Given better-quality data, there is no reason for the lead-time to be nearly halved and the PoD to drop 14 points (NWS's own statistics) since 2011. Here are the latest figures (from my WaPo article) that I could obtain.

That is a terrible drop considering the improved radar data mentioned above. I'd love to post more current figures but the NWS, which used to post these figures out in the open, has now put them behind a login. That is yet another indicator that NWS/NOAA management is in denial.

Remember: the NWS used to -- routinely -- provide better quality tornado warnings than they have in recent years. They have regressed.

As to distance from the radars: As demonstrated in my interpretation of the AMA radar as it pertained to Perryton [here: Again! Fatal Tornado Strikes Perryton Without Sufficient Warning ] there is no reason the NWS couldn't have warned in time. There was a meteorologist chasing in Perryton (who has requested confidentiality) who called and said my interpretation was "right on." He said as meso #1 passed, the wind shifted to the NW. Then, it rapidly shifted back to the NE then ESE as meso #2 approached (the one that produced the tornado).

AccuWeather's business meteorology group provided 18-minutes of warning to its client in Perryton. The NWS? About one minute. And that was based on a call from a chaser!

Even the radar-obvious Matador Tornado had just 10 minutes of warning. The NWS's published standard? Thirteen minutes that they used to routinely obtain (see table).

Two NWS mets have called me (they, too, have requested confidentiality) in the past two months and said there is no quality radar training anymore. Meteorologists who are well-trained (AccuWeather) are able to issue these warnings. So, the "poor quality radar training in the NWS" seems like a reasonable hypothesis.

Further, along those lines, when the 1:20am June 8, 2022, tornado went unwarned across the south KC area, SPC had a severe thunderstorm watch that said, "a tornado or two possible" in effect. So, how did EAX operate their radar? On seven-minute mode! A Udall-type tornado (since it was the middle of the night) could have developed and they would never have known! For more info: www.mikesmithenterprisesblog.com/2022/06/regarding-overnight-tornado-situation.html

John, I agree with you that we need more radars. I worked very hard with Congress to cue them up to give them to the NWS. Then, the NWS said they didn't want them! See: www.washingtonpost.com/weather/2020/11/21/radar-gaps-weather-service/

While more training will help -- greatly -- the management of the NWS is in denial about the depths of the problem in the agency. That is why we need a National Disaster Review Board to hold NWS, FEMA, Red Cross, emergency management and everyone involved in disaster planning and response accountable in the way the NTSB does for transportation.

I'd be happy to answer any questions or comments anyone else may have. Thanks again, everyone, and call your congresspeople.

 

[Warren Faidley]

Just throwing this out, but I believe these odd, 2023 events of storms moving south, mergers, splitting, etc., are giving Doppler algorithms fits. This was the case when I chased in May, from personal observations. Just a thought.

 

[Mike Smith]

Just throwing this out, but I believe these odd, 2023 events of storms moving south, mergers, splitting, etc., are giving Doppler algorithms fits. This was the case when I chased in May, from personal observations. Just a thought.

Hi Warren, while that is correct, the warnings are supposed to be issued by well-trained human meteorologists. As mentioned above, these odd storm movements, etc., are not deterring AccuWeather's extreme weather meteorologists (they provided superior warnings in Perryton and Matador).

The stats I quote above show the decline in the NWS tornado warning quality over a period of years ending in 2020.

 

[Mike Smith]

I meant to include this in my longer post above:

I mentioned that I was told by a meteorologist/chaser, who heard the telephone exchange, that a chaser calling AMA was why they issued their belated tornado warring. That is why AMA and LBB were tweeting these the afternoon of the Matador storm.

Do we really want to go back to the 70's and 80's when spotters/chasers were the primary input for tornado warnings?

 

[Mike Smith]

12:52pm Sunday:

A tornado watch is in effect for most of Indiana and southern Michigan. It is 10% on the 1730Z convective outlook. They are running their radar on five minute intervals even though storms have fired.

I sent them a note via Twitter. We'll see what happens. But, it should not be our job to coach NWS offices in these situations.

 

[Andy Wehrle]

Nowhere near as important as shorting the warning on a significant tornado, but perhaps indicative of the overall trend in decreasing quality of NWS warnings.

I was chasing in Iowa yesterday (tornado potential turned out to be a huge dud), and caught hail from two separate storms, one between Indianola and Pleasantville and again from a second storm that fired up back over Indianola. Left one noticeable dent in the roof of my car.

1930	100	Indianola	Warren	IA	4136	9356	Relayed from NWSchat. (DMX)
1931	175	Indianola	Warren	IA	4136	9357	Delayed report with time estimated. Via KCCI viewer.

Indianola was not under a severe thunderstorm warning at the time indicated on these reports. A warning went out for the first storm (for Pleasantville and points east, after I'd already encountered hail with it) at 1926Z.