NWS Central Region Impact Based Warning Experimental Product

[samuel stone]

anyone have any thought on TWX not issueing a tornado warning for that harveyville storm. I hadnt seen the velocity image until just when I looked at Adam's post. certainly appears as though it warranted a Tor Warning

 

[Mike Smith]

Cautionary Tale?

anyone have any thought on TWX not issueing a tornado warning for that harveyville storm. I hadnt seen the velocity image until just when I looked at Adam's post. certainly appears as though it warranted a Tor Warning

Meteorology is not perfect and, at least in my lifetime, never will be. That said, had this identical storm occurred on April 1 at TWX, this should have been a PARTICULARLY DANGEROUS SITUATION tornado warning. Yet, there was no warning at all.

The lack of a warning on this F-2 (by definition, "significant") tornado demonstrates the futility of tornado intensity warnings at the current state-of-the-art. This was a cycling supercell with good rotation. Obvious, right?

The problem is that things that are obvious in review are not always obvious at the time and with the pressure of dealing with multiple, fast-moving warning candidates. Things that work well in the lab don't always work as well in the real world.

From where I sit, the NWS should put its efforts into improving the accuracy of existing tornado warnings, especially as pertains to false alarm rates. Given severely limited resources, tornado intensity warnings, trying to push average lead times past 15 minutes, and warn-on-forecast should be put on the back burner for now. This is going to take new science (perhaps from Vortex II) but we need to get on it -- now.

 

[rdale]

anyone have any thought on TWX not issueing a tornado warning for that harveyville storm. I hadnt seen the velocity image until just when I looked at Adam's post. certainly appears as though it warranted a Tor Warning

It was only there for one scan, after the tornado had touched down (and possibly lifted.) There was nothing preceding this that would have indicated tornado. It was a notch on the leading edge of a line of embedded storms, and it was gone 5 minutes later. The cell had no prior reports of a tornado (the line itself apparently had one an hour previous) and no future reports of a tornado. Sometimes things just happen.

 

[Adam Simone]

It was only there for one scan, after the tornado had touched down (and possibly lifted.) There was nothing preceding this that would have indicated tornado. It was a notch on the leading edge of a line of embedded storms, and it was gone 5 minutes later. The cell had no prior reports of a tornado (the line itself apparently had one an hour previous) and no future reports of a tornado. Sometimes things just happen.

Agree with everything you said. It only showed up in 1 scan and was only on the ground for 4 minutes. I know Chad Cowan tweeted earlier today that the NWS was on the phone with him prior to the tornado trying to get information from him on the storm but he couldn't see anything. By the time the rotation was scene on radar and a warning would have been issued, the tornado would have been gone. Just a freak accident that shows no matter how good of technology we have things still happen.

 

[Mike Smith]

There was nothing preceding this that would have indicated tornado. It was a notch on the leading edge of a line of embedded storms, and it was gone 5 minutes later. The cell had no prior reports of a tornado (the line itself apparently had one an hour previous) and no future reports of a tornado. Sometimes things just happen.

That is not correct. This system had been cycling tornadoes since about 6:40pm. Here is the hook (there was also a couplet, not shown) at 6:40 in Reno Co.
View attachment 6698

And, this is 58 minutes before the tornado at Harveyville.
View attachment 6699


While you are correct that the cell was lining-out, that specific cell had a history of producing tornadoes and funnels. At the time of the Harveyville tornado, it had strong gate-to-gate shear. If we had 30 second data (2 rpm, very doable), or data from a TDWR, that tornado might have been warnable.

Instead of fixing these very fixable issues, we are going to ask meteorologists in these offices to expend their "mental bandwidth" (a finite resource) and their time (a very finite resource with storms moving at 55 mph) trying to figure out what intensity tornado warning to issue.

 

[Mike Smith]

It only showed up in 1 scan and was only on the ground for 4 minutes.

There is something to be said for age (recently celebrated my 60th).

The WSR-57's and WSR-74's operated at 3 rpm. Put another way, there would have been 12 scans showing the Harveyville tornado while it was on the ground and some (unknown number) before it touched down. With the right technology, it would have possible to have provided some warning on the storm, at least for the latter part of its track.

 

[rdale]

That is not correct. This system had been cycling tornadoes since about 6:40pm.

I see no tornado reports in the hour before Harveyville, where was one reported in that timeframe? If a storm has nothing for an hour, and shows no signs of rotation, I'm not sure they can be blamed for not going with a TOR.

At the time of the Harveyville tornado, it had strong gate-to-gate shear. If we had 30 second data (2 rpm, very doable), or data from a TDWR, that tornado might have been warnable.

Agreed, but we don't have 30 second data and/or TDWR in that area, so I'm not sure how the NWS office can be blamed for not issuing a warning.

Instead of fixing these very fixable issues

TDWR's aren't cheap (and there is no money in the NWS budget anyways, note the ending of the Wind Profiler Network this summer) and 88D's can't go high RPM because of partner requirements. So I would not consider those "very fixable". I would go so far as to call those ideas "not realistic at all."

we are going to ask meteorologists in these offices to expend their "mental bandwidth" (a finite resource) and their time (a very finite resource with storms moving at 55 mph) trying to figure out what intensity tornado warning to issue.

It's really not that complex. For example, look at the Buffalo MO tornado. The warning was initially issued as a "radar indicated." After the debris ball, all they did was add text to a SVS that nobody in the public or private industry would ever see. In the new tier structure, they could have issued a new TOR with "confirmed" tornado. Makes sense to me, and doesn't involve any extreme levels of "mental bandwidth" more than they already exerted. It just allows that information to be better relayed to the end users, and to me that's a good thing.

 

[Mike Smith]

Rob,

I wish you would reply to what I wrote rather than creating your own interpretation. I don't "blame" TWX. That word does not appear anywhere in my posts.

I've said all along that no one has the scientific skill forecast the ultimate intensity of a tornado and put it into a warning. This, because it affected one of the offices that will be engaged in the intensity warning experiment on April 1, it tends to prove my point. I believe the NWS is shooting itself in the foot because it will set itself up for criticism for not getting the intensity correct:

• Congressman Gasbag (after being briefed by his staff): "Was this a 'significant' tornado?"
• NWS manager: "Yes, sir."
• Congressman Gasbag: "But, you only issued an 'ordinary' tornado warning, didn't you?!"

Truncating volume scans is an idea whose time has come based on numerous conversations (including with aviation interests) the past year. No one -- and I mean no one -- is using layer composite reflectivity and some of these other products. Truncating the volume scans in these situations is the smart thing to do. It just requires some will and leadership.

Finally, w/r/t the hour before, look at this link: www.stormtrack.org/forum/attachment.php?attachmentid=6699&d=1330576836 You'll see TWX's TOR (purple polygon) where I have added an arrow to denote the hook. As I understand it, there was a funnel cloud report (it was darkness). This was less than one hour before Harveyville.

I've always said we should tell people what we know and your Buffalo, MO example is a good one. But, there is a huge difference between noting what is occurring and trying to forecast what is going to occur in the form of tiered intensity warnings.

Mike

 

[rdale]

I wish you would reply to what I wrote rather than creating your own interpretation. I don't "blame" TWX. That word does not appear anywhere in my posts.

Agreed - and my apologies if it came across that way. The first post to bring up this topic clearly looked to pass blame, so that's what I was clearing out.

Truncating volume scans is an idea whose time has come based on numerous conversations (including with aviation interests) the past year.

I agree - and from my understanding the FAA has said they will not allow truncated scans with precip occurring. As I understand, AVSET took a lot more work to get through than needed because of it.

As I understand it, there was a funnel cloud report (it was darkness). This was less than one hour before Harveyville.

Does a funnel cloud in the dark necessitate the need for an hour long blanket TOR? I don't think so, but now we're getting too close to Monday morning QB. I checked the chatroom logs and didn't see anyone suggesting anything out of the ordinary with that cell, so if all the TV, NWS, private sector mets and EMs had no issues during the event, it is hard for me to go back and point out the "obvious" at this stage.

 

[Mike Smith]

but now we're getting too close to Monday morning QB.

I fear something about my position is getting lost in the translation: By representing themselves as able to do tornado warnings tiered by intensity they set themselves up for more "Monday morning quarterbacking," not less. I'm concerned that the April 1 experiment will serve neither the public nor the NWS well.

...it is hard for me to go back and point out the "obvious" at this stage.

Precisely.

 

[Jason Haller]

another issues not being addressed with the Harveyville Tornado was its proximity to the Topeka radar. Harveyville is within 30 miles or so of the physical radar. I also believe that Harveyville is directly south of the actual radar as well with which would have made any velocity scans very hard to read, correct? Could this have been a factor in not seeing rotation in the storm since it was so close to the radar and a bit far for the Wichita radar for good velocity scans?

Edit: Harveyville is not directly south of the radar but is south and east of it. My mistake.

 

[rdale]

No, that's a sweet spot for seeing rotation. Tough for straightlines though.

 

[Jason Haller]

No, that's a sweet spot for seeing rotation. Tough for straightlines though.

Thanks for the info. I had heard differently.

 

[Rodney Price]

I'm in a unique position with respect to the Harveyville tornado, as a former resident of the town. My folks' house was damaged, the garage destroyed and worst of all, they lost their next door neighbor.

View attachment 6706View attachment 6707

Above are the radar images from 0257 and 0302Z 29 Feb. (thanks to George Phillips, SOO at NWS TOP for providing these).
Mike suggests a faster scanning strategy for the 88-D. Tuesday night, KTWX 88-D was running in VCP-212 mode. Let's say the radar could be set up to complete the volume scan in half the time. Would we have been able to spot the vortex and provide enough warning, given it was moving at 75 mph? Hard for me to say.

As is the case with any event (even life in general) there's always room for improvement.

 

[Mike Smith]

Here is more on the topic in case anyone is interested: www.kansas.com/2012/03/02/2239323/harveyville-tornado-struck-without.html

For me, this situation further demonstrates the futility of trying to warn for tornado intensity. We just do not have the scientific skill or the data (4+ min. intervals with the WSR-88D's) to do it.