• Welcome to TalkWeather!
    We see you lurking around TalkWeather! Take the extra step and join us today to view attachments, see less ads and maybe even join the discussion.
    CLICK TO JOIN TALKWEATHER

Severe WX May 2019 Plains Severe Event

In hindsight was it wise for SPC to say yesterday was 4/27 all over again? If I recall that day there were ZERO possible limiting factors. I think in the future SPC should wait to pull the trigger o. higher risks It's much easier to ramp up a threat than to ramp down.
 
warneagle said:
I don’t think they were wrong to go with the high risk given the model output and observed conditions yesterday, but I do think they may have communicated more confidence in the forecast than was actually realistic (particularly in the extreme probabilities in the two PDS watches). You’re never going to get it 100% right and you’ll inevitably be criticized whether your forecast underperforms or overperforms, but being transparent about uncertainty and communicating the level of confidence accurately takes some of the sting out of those criticisms. Hopefully this will be a learning experience both scientifically (still mind-blowing that a day with those parameters Atmospheric Anti-Climax) and in terms of communication/public relations.
Click to expand...
I agree with you to some extent, but we have to be realistic; was there very much uncertainty at all to most of us yesterday? If there was notable uncertainty in the forecast, wouldn't somebody have mentioned something? The strong cap came virtually out of nowhere, leaving every single one of us scratching our heads. I think most if not all of us expected a massive outbreak, something we'd see only a couple times in a decade. I absolutely stand by the SPC using the strong wording it did for that watch as at the time it was absolutely the right thing to do, and we have to acknowledge the privilege we have of being able to talk about this in retrospect, hoping to learn something noteworthy.
 
 
Staying w/ the post-mortem theme on yesterday's event:

The NAM may have performed the best.
While the HRRR was in Armageddon mode all day the NAM did 2 things well:

1) The NAM kept SW OK relatively quiet until around 6-7pm (00z). Then it had the area being hit by a more linear, less organized (ie, weaker) storm mode. This is pretty much what happened though I believe it was a little later than 00z.

2) The NAM consistently showed warming in the mid levels through 00z, with 700mb temps in the 10-12C range over SW OK. I believe this also verified and resulted in the CAP issues.

The problem for the forecaster is which model do you go with after you have looked at all of them. I, personally am an HRRR fan but this event has taught me a good lesson about not drinking the Koolaid of the model data that tastes (looks) the most pleasing.
 
bwalk said:
Staying w/ the post-mortem theme on yesterday's event:

The NAM may have performed the best.
While the HRRR was in Armageddon mode all day the NAM did 2 things well:

1) The NAM kept SW OK relatively quiet until around 6-7pm (00z). Then it had the area being hit by a more linear, less organized (ie, weaker) storm mode. This is pretty much what happened though I believe it was a little later than 00z.

2) The NAM consistently showed warming in the mid levels through 00z, with 700mb temps in the 10-12C range over SW OK. I believe this also verified and resulted in the CAP issues.

The problem for the forecaster is which model do you go with after you have looked at all of them. I, personally am an HRRR fan but this event has taught me a good lesson about not drinking the Koolaid of the model data that tastes (looks) the most pleasing.
Click to expand...
Very well said. I too get stuck in a single model sometimes and tend to forget about the others. Especially when one model performs well time after time and then goes nuts. We should still look at the other models and look at the mean results instead of getting Tunnel Vision.
 
bwalk said:
Staying w/ the post-mortem theme on yesterday's event:

The NAM may have performed the best.
While the HRRR was in Armageddon mode all day the NAM did 2 things well:

1) The NAM kept SW OK relatively quiet until around 6-7pm (00z). Then it had the area being hit by a more linear, less organized (ie, weaker) storm mode. This is pretty much what happened though I believe it was a little later than 00z.

2) The NAM consistently showed warming in the mid levels through 00z, with 700mb temps in the 10-12C range over SW OK. I believe this also verified and resulted in the CAP issues.

The problem for the forecaster is which model do you go with after you have looked at all of them. I, personally am an HRRR fan but this event has taught me a good lesson about not drinking the Koolaid of the model data that tastes (looks) the most pleasing.
Click to expand...

Thank you as that answers my question from earlier, credit to the NAM it is.
 
warneagle said:
I think the analysis of what the cap meant for the forecast may have been the result of confirmation bias—the assumption was that the main failure mode would be too many storms forming, not too few, so people took the stronger cap to mean that the main failure mode had been eliminated rather than seeing it as a potential failure mode in its own right. It didn’t help that none of the models really forecast a cap of that strength and that it wasn’t obvious in the early observations, but I think there was that preconceived idea of what a stronger-than-expected cap would mean.

I feel like there’s probably a debate to be had about over-reliance on CAMs and the tendency of social media to create feedback loops and promote groupthink as well, but that struck me as the most obvious issue with the nowcasting yesterday evening.

I don’t think they were wrong to go with the high risk given the model output and observed conditions yesterday, but I do think they may have communicated more confidence in the forecast than was actually realistic (particularly in the extreme probabilities in the two PDS watches). You’re never going to get it 100% right and you’ll inevitably be criticized whether your forecast underperforms or overperforms, but being transparent about uncertainty and communicating the level of confidence accurately takes some of the sting out of those criticisms. Hopefully this will be a learning experience both scientifically (still mind-blowing that a day with those parameters Atmospheric Anti-Climax) and in terms of communication/public relations.
Click to expand...

Great post. Confirmation bias via group-think and/or over-reliance on one model is what I was hinting at yesterday, but I lacked the words to put it this clearly.

The post-mortem here on Talkweather has been amazing to watch. Respectful discussion, a variety of viewpoints, and numerous posters bringing new information or ideas to the table - - including input from Mets and other experts.

Patrick Marsh's Twitter thread that someone recently posted also reminds me that we have to careful in avoiding additional group-think as it pertains to post-mortem analysis. A lot of us have touched on the stouter cap, warm mid-levels, and weak lapse rates, but as Patrick Marsh pointed out there's not a ton of difference between the 4-27-11 BMX sounding and yesterday's OUN sounding. He's asked about 0-3km lapse rates by someone on Twitter and he responds with a sounding from 04-15-2012 that shows 0-3km lapse rates a tad weaker than yesterday.

So, one of those (or all) previously mentioned factors might explain some things or they might not. Other variables are also going to need further scrutiny. As an example, perhaps weaker lapse rates 0-3km can be made up for with stronger lapse rates higher up UNLESS X or Y also occurs. We know the stouter cap couldn't be overcome -- but that's just one issue from yesterday. Why was storm isolation so poor? Even in areas that had better discrete spacing the tendency for storms to undercut storms further to the north was a constant issue. It's all quite fascinating and I'm glad for the excellent discussion here on Talkweather.
 
South AL Wx said:
Click to expand...


Wow. It's kind of disturbing, from a logic perspective when you view the 2 soundings side-by-side (OUN/BMX). Both appear to have equal CAPS around 700mb, yet produced astoundingly different outcomes. With all the other parameters being equal between the 2 soundings it just does not make any sense that the results were polar opposites (at least in the warm sector/highest risk area). That the 2 systems did produce such differing results causes a degree of cognitive dissonance in me, as I like things to make sense from an intellectual/evidence-based perspective. IOWs: both systems should have behaved the same given all the variables appearing to be equal. 2+2 ALWAYS = 4.

What is the 1 or 2 almost invisible variables that are being missed?
 
bwalk said:
Wow. It's kind of disturbing, from a logic perspective when you view the 2 soundings side-by-side (OUN/BMX). Both appear to have equal CAPS around 700mb, yet produced astoundingly different outcomes. With all the other parameters being equal between the 2 soundings it just does not make any sense that the results were polar opposites (at least in the warm sector/highest risk area). That the 2 systems did produce such differing results causes a degree of cognitive dissonance in me, as I like things to make sense from an intellectual/evidence-based perspective. IOWs: both systems should have behaved the same given all the variables appearing to be equal. 2+2 ALWAYS = 4.

What is the 1 or 2 almost invisible variables that are being missed?
Click to expand...


Very interesting perspective. I can see several differences, but not one that jumps off the screen and says that one is going to produce more damaging tornadoes than the other. It looks bad for both soundings.
 
This is when special soundings from upstream offices at 18z/onward would've proved super helpful in picking up the warming aloft that stunted the surface based storms being able to take advantage of the incredible surface-based instability.

In the end, you get a bunch of elevated convection, which then stabilizes the higher up layers of the atmosphere (basically using up all of the elevated CAPE) combined with poor low-level lapse rates resulted in storms cannot tap the powder keg that is the surface.
 
Storms in Kansas on the back end of the system are beginning to rotate... a confirmed tornado just came very close to Chapman.
 
Of course the longer tracked tornado family is in Kansas on the slight risk back side of the system on QLCS day instead of in Oklahoma in one of the most volatile setups in modern years
 
You must log in or register to reply here.
Back
Top