Tornadoes Hit Philadelphia & New Jersey 7/29/21

Jul 5, 2009
1,247
1,193
21
Newtown, Pennsylvania
I live in Newtown PA and there were two bonafide tornadic supercells that tracked relatively close to home, one through New Hope / Washington Crossing just 5-7 miles north, and another in Bensalem less than 10 miles to the south. Meanwhile, I just about missed the entire thing because my family and I are vacationing down in Long Beach Island (LBI) NJ. You can’t imagine how pissed off I was about the irony of missing a once-in-a-lifetime local weather event! Was distracted through an entire family dinner. Typical chaser luck. I can’t remember ever seeing radar signatures like this in the area, or seeing PDS wording and “confirmed large and dangerous tornado” around here (although, even with that wording, it said “radar confirmed,” don’t remember ever seeing anything like that, isn’t it usually radar *indicated*, or confirmed only through actual sighting??)

The more southern supercell did finally arrive at the NJ coast and passed through LBI near Barnegat to my north. It was dark by then so I didn’t attempt to chase it, but enjoyed watching the lightning as the storm came across the bay, then walked to the beach to watch it out over the ocean. Enjoyed being able to get the family into it as well, in fact my wife got the lightning shot below. I am also posting a radar image from the northern storm’s close passage to my hometown (I live immediately south of where it says “Newtown Grant”) and the southern storm as it approached LBI.

There was significant damage in Bensalem, speculation that it could have been EF-3, which may tie or exceed the record for Bucks County PA. Looking forward to seeing the surveys.

Interesting that there was morning precip and little sunshine in the area during the day. Many folks on the Philly weather forum commenting, imagine if there had been sun... and another noting this should be “put in the memory banks - its all about #Dynamics”

@Saul Trabal thanks for starting this thread, don’t recall seeing you post before and always glad to meet another local chaser!
 

Attachments

Feb 19, 2021
73
92
6
Wichita
During the past ten years, essentially since Joplin, National Weather Service tornado warnings have become less accurate. That resulted in little warning for the Bensalem Tornado yesterday evening. Bensalem Tornado: Another Dangerous National Weather Service Warning Miss

I don't know why the lead time has been cut nearly in half nor why the PoD has dropped by 18%. But, this trend is extremely dangerous and a solution needs to be found as quickly as possible. My recommendation is that Congress creates a National Disaster Review Board modeled after the National Transportation Safety Board. The latter has been hugely successful at making aviation and other forms of transportation safer.
 

Dean Baron

Supporter
Sep 25, 2006
565
210
11
31
Minneapolis, MN
I can’t remember ever seeing radar signatures like this in the area, or seeing PDS wording and “confirmed large and dangerous tornado” around here (although, even with that wording, it said “radar confirmed,” don’t remember ever seeing anything like that, isn’t it usually radar *indicated*, or confirmed only through actual sighting??)
I can't confirm because I wasn't watching radar yesterday but I am willing to bet a large TDS showed up via correlation coefficient which allowed the NWS to confirm a "large and dangerous tornado" without chaser/spotter reports confirming it visually.
 
  • Like
Reactions: JamesCaruso
Jul 5, 2009
1,247
1,193
21
Newtown, Pennsylvania
During the past ten years, essentially since Joplin, National Weather Service tornado warnings have become less accurate. That resulted in little warning for the Bensalem Tornado yesterday evening. Bensalem Tornado: Another Dangerous National Weather Service Warning Miss

I don't know why the lead time has been cut nearly in half nor why the PoD has dropped by 18%. But, this trend is extremely dangerous and a solution needs to be found as quickly as possible. My recommendation is that Congress creates a National Disaster Review Board modeled after the National Transportation Safety Board. The latter has been hugely successful at making aviation and other forms of transportation safer.
Mike, I read your blog post. I can’t explain the delay from 7:03 to 7:07, except to assume that the velocity simply didn’t trip the necessary algorithmic criteria to issue a warning? As for the 6:49 image, you only showed the reflectivity, not the velocity, so it may be unfair to suggest they should have issued a warning that soon, based solely on a reflectivity appendage which is fairly common with non-tornadic storms.

I know you listed some possible theories for the reasons behind the alleged delays, but I always thought there was an algorithm that determines when the velocity couplet exceeds a threshold, so I’m not sure how there could be a deterioration in warning lead time, wouldn't this imply that someone was literally ignoring an alarm?? My understanding is that human intervention can issue a warning earlier than the threshold is exceeded, but not later ... But admittedly I don’t know too much about it, this is just my amateurish anecdotal understanding...

EDIT: I went back and watched the video in the Tweet you linked, where you say the tornado is visible and then the warning is issued after. I couldn’t actually see the tornado, couldn’t that have just been hard-wrapping RFD, forming the couplet that prompted the warning?
 
  • Like
Reactions: Saul Trabal
Feb 19, 2021
73
92
6
Wichita
Hi James,

Happy to elaborate.

I have attached the 6:49p velocity and reflectivity for reference. In the preceding two minutes, the inbounds greatly increased. So, we had a supercell's pendant echo that was starting to grow a hook between Horsham and Abington and a rapidly increasing couplet (likely from an increasing RFD). Given it was the "tail-end Charlie" with a SIGTOR of 2 and the northern supercell was producing tornadoes, it was a pretty easy warning. That a warning was obviously needed became even evident in the next few minutes as the Bensalem supercell was clearly a right mover.

Had I been working the WeatherData/AccuWeather warning desk and had I had a client in Bensalem, I would have tornado warned at 6:49. However, I don't necessarily fault the NWS for not warning at 6:49. Reasonable people can disagree there. But, by 6:55-7p, it was absolutely unmistakable. Addition at 2:40pm. Mt Holly NWS has rated the storm EF-3.

Some NWS offices rely more heavily on various techniques and algorithms while others don't pay much attention to them. I don't know the specifics of how Mt Holly does it.

If this were an isolated incident, I would not have written today's blog post about Bensalem nor my May 21 piece for the Washington Post (while there is a link in my blog post, for convenience here is my WaPo article: https://www.washingtonpost.com/weather/2021/05/21/joplin-tornado-warning-improvement-nws/ ).

Unfortunately, there have been multiple obvious, unwarned tornadoes between May 21 and yesterday. For example, here is one from northern Indiana: Friday: Another Dangerous Missed Tornado Warning This was rated by the NWS was "upper EF-2."

Right now, very straightforward tornadoes up to perhaps F-3 intensity, are improperly warned. If this continues, people are going to lose their lives.

That is why I urge Congress to quickly create a National Disaster Review Board modeled on the hugely successful NTSB.

Thank you for asking.

Mike

Screen Shot 2021-07-30 at 1.31.26 PM.png

P.S. I forgot to include the PHL TDWR data. This is from 6:59pm.
Screen Shot 2021-07-30 at 1.57.06 PM.png


Mike, I read your blog post. I can’t explain the delay from 7:03 to 7:07, except to assume that the velocity simply didn’t trip the necessary algorithmic criteria to issue a warning? As for the 6:49 image, you only showed the reflectivity, not the velocity, so it may be unfair to suggest they should have issued a warning that soon, based solely on a reflectivity appendage which is fairly common with non-tornadic storms.

I know you listed some possible theories for the reasons behind the alleged delays, but I always thought there was an algorithm that determines when the velocity couplet exceeds a threshold, so I’m not sure how there could be a deterioration in warning lead time, wouldn't this imply that someone was literally ignoring an alarm?? My understanding is that human intervention can issue a warning earlier than the threshold is exceeded, but not later ... But admittedly I don’t know too much about it, this is just my amateurish anecdotal understanding...

EDIT: I went back and watched the video in the Tweet you linked, where you say the tornado is visible and then the warning is issued after. I couldn’t actually see the tornado, couldn’t that have just been hard-wrapping RFD, forming the couplet that prompted the warning?
 
Last edited:
  • Like
Reactions: Saul Trabal
Jul 5, 2004
450
15
6
New Jersey
During the past ten years, essentially since Joplin, National Weather Service tornado warnings have become less accurate. That resulted in little warning for the Bensalem Tornado yesterday evening. Bensalem Tornado: Another Dangerous National Weather Service Warning Miss

I don't know why the lead time has been cut nearly in half nor why the PoD has dropped by 18%. But, this trend is extremely dangerous and a solution needs to be found as quickly as possible. My recommendation is that Congress creates a National Disaster Review Board modeled after the National Transportation Safety Board. The latter has been hugely successful at making aviation and other forms of transportation safer.
I wonder if part of the problem, in this area at least (I live in northeastern New Jersey) is that tornadoes are fairly rare around here. Yesterday was practically unprecedented. I recall one TV report that said there were 14 tornado-warned storms (radar indicated?) and two confirmed tornadoes. People around here seem pretty blasé about tornadoes because they are so infrequent. It also seems fairly rare that we get anything above an EF1 or EF2 in the Tri-State area-and I should probably include Pennsylvania in that.
 
Feb 19, 2021
73
92
6
Wichita
Saul,

I don't know but it is hard for me to imagine that is the reason. I have attached an actual textbook illustration of a toranadic supercell and it was more or less identical to yesterday evening's PA-NJ tornadoes. It is difficult for me to understand why this wouldn't have been immediately recognized for the threat it was, especially since SPC had issued a tornado watch and there was a SIGTOR of 2.

There was an F-5 tornado in Pennsylvania in 1985 and several F-4's. It is necessary for meteorologists of the region to be well-versed in warning techniques.

Mike
Screen Shot 2021-07-30 at 3.04.19 PM.png

I wonder if part of the problem, in this area at least (I live in northeastern New Jersey) is that tornadoes are fairly rare around here. Yesterday was practically unprecedented. I recall one TV report that said there were 14 tornado-warned storms (radar indicated?) and two confirmed tornadoes. People around here seem pretty blasé about tornadoes because they are so infrequent. It also seems fairly rare that we get anything above an EF1 or EF2 in the Tri-State area.
 
  • Like
Reactions: Saul Trabal

John Farley

Supporter
Apr 1, 2004
1,627
880
21
Pagosa Springs, CO
www.johnefarley.com
One thing I wonder is whether this trend is related to some effort to reduce the false alarm rate. Certainly seems to me like these storms should have been tornado warned, as you say. I think that reducing the false alarm rate is a worthy goal, but it should not come at the expense of delayed warnings for storms that do produce tornadoes, especially strong ones as in these examples. One could, I suppose, argue that there is an inherent tradeoff, but from the images posted by Mike above, the PA storm seems pretty obvious with regard to tornado potential.
 
  • Like
Reactions: Saul Trabal
Feb 19, 2021
73
92
6
Wichita
Hi John,

Thank you for your comments.

Harold Brooks and Victor Gensini (B&G) say exactly that: the big drop in PoD is related to an attempt to cut false alarms. Please see: https://www.washingtonpost.com/weather/2021/06/11/improving-tornado-warnings-preparation-nws/

Their hypothesis is that it is worthwhile to have "one unwarned F-3 tornado every three years" in order to decrease the NWS' tornado FAR from 74.2 to 70.4 percent. I rather strongly disagree. My full reply is here: Friday: Another Dangerous Missed Tornado Warning

The problem is not confined to the reduced PoD. Lead-time has regressed to 1994 levels. The lead time on the Bensalem Tornado was minus three minutes. That tornado would have been an easy warning, with about ten minutes of lead time, in, say, 2008.

As I note in my reply to B&G, if they are correct, one person will die each year in order to achieve the imperceptibly small (to the public) decrease in FAR. This seems like a terrible trade to me. But, it is actually worse. I'm not particularly concerned about EF-0 tornadoes. But, the NWS is missing EF-2's and EF-3's (like Thursday's Bensalem EF-3) at a far greater rate than Brooks and Gensini predict. That means, inevitably, more will be killed. Given what occurred in Joplin, another catastrophic fatality tornado is in the realm of possibility. This is why I am hitting this topic so hard.

With the much higher rate of missed significant tornadoes (≥EF-2) than predicted by Brooks and Gensini, it is safe to say the misses are not due to the attempt to decrease false alarms. Something else is at work, but I don't know what it is. I can speculate that some of the factors might be:
  • The retirement of my generation of meteorologists and our expertise. This is especially true since we learned to warn tornadoes based on hooks, right-movers and other non-Doppler wind techniques. Given the funnel cloud report at 6:49pm along with the hook (see image) west of Bensalem at 6:47, it would have been an easy warning even without Doppler.
  • The hiring of meteorologists fresh out of college. I am aware of multiple instances in three months where the NWS hired meteorologists who graduated in May, even when a more experienced candidate had applied.
  • Degradated radar training in the NWS. A person currently in the NWS, when discussing the tornado warning problem said to me, "Twenty years ago, radar training was a month in Kansas City. A decade ago, it was a week in Kansas City. Now, it is a couple of hours from the SOO." I don't know whether this is true everywhere or in all cases. But, if it is largely true, it is a serious issue.
  • Lack of attention. A quarter-century ago, the NWS worried about the weather wire and NOAA Weather Radio. Now, there is a tremendous amount of social media that has to be accomplished. Is this drawing attention away from mesoscale analysis and monitoring?
Note: I do not know to the extent any of the above are true. I have heard from people who believe there is no problem at all.

This is why we need a National Disaster Review Board. I explain in detail, here: Editorial: Renewing My Call For a National Disaster Review Board

The Board would have two responsibilities:
  • Like the NTSB, National Chemical Incident Board, and the others, it will comprehensively investigate major disasters to learn what went wrong and what went well. It would not be limited to reviewing the NWS. It would review the NWS, FEMA, Red Cross and other agencies, public and private, just like the other federal boards do. It is a terrible policy for the NWS to investigate itself.
  • It will remove warning verification from the NWS. Lots of people, under the promise of confidentiality, send me things. Correctly or not, there are widespread allegations the NWS "fudges" times and other information when they did not have a warning out. On one occasion, a NWS office in Kansas elected not to add a tornado report of mine (even though it was documented with photos) to their database in a situation where they did not have a tornado warning. That gives me a sense those allegations have credibility.

The huge death toll in Joplin -- by far, the worst in the history in the tornado warning era -- was due to mistakes made by the NWS and local emergency management. See: Amazon.com: "When the Sirens Were Silent" How the Warning System Failed a Community eBook: Smith, Mike: Kindle Store

The NWS's JLN "Service Assessment" was, at best, very poor. Others have called it a "cover-up." But, most meteorologists in and out of the NWS don't know that. As a result, a number of sub-optimal decisions have stemmed from the information in the JLN report.

Before another Joplin, Sandy, or a June 2016 Greenbrier Flash Flood (23 dead) can occur, we must fix this.

I am a Reagan Conservative and I believe we would be better off if the federal government was smaller. That stipulated, we must expand the government to create the National Disaster Review Board.

John's comments were appreciated and I would be happy to read any comments or answer any questions you might have. Please, fire away. Thanks, everyone, for reading.

Mike


Screen Shot 2021-07-31 at 5.21.17 PM.png


One thing I wonder is whether this trend is related to some effort to reduce the false alarm rate. Certainly seems to me like these storms should have been tornado warned, as you say. I think that reducing the false alarm rate is a worthy goal, but it should not come at the expense of delayed warnings for storms that do produce tornadoes, especially strong ones as in these examples. One could, I suppose, argue that there is an inherent tradeoff, but from the images posted by Mike above, the PA storm seems pretty obvious with regard to tornado potential.
 
Last edited:
Feb 19, 2021
73
92
6
Wichita
Thank you, James.

The Bensalem Tornado touched down at 7:04 and reached the town at 7:11. It lifted one minute later. That made the 'lead time' a minus three minutes. Bensalem, ABC6, 710p.png
Above is a screen shot from ABC6, the most-watched news operation in Philadelphia. At that moment, she is speaking the words that a tornado warning had been issued for Bucks Co., north Philly and Bensalem. Note the time (lower right). Even though the warning was officially issued at 7:07, it takes some time to get this on the air as there were multiple other tornado warnings they were covering. Essentially the tornado was in Bensalem as the warning was going out. The tornado would lift just two minutes later. It is fortunate no one was killed.


Mike, I posted a link to your blog post to the Philadelphia weather forum (PhillyWx.com) thread for the 7/29/21 event to spark some discussion. Here is an interesting post from someone who agrees with you that the Bensalem warning was late:

 

John Farley

Supporter
Apr 1, 2004
1,627
880
21
Pagosa Springs, CO
www.johnefarley.com
Mike, are you a member of the WAS-IS group on Facebook? It is a group that looks at the integration of weather and social sciences, and I think the points you are making would be of interest to that group. For others here with an interest in the relation of social science and societal issues to weather, I would also suggest that group.
 
  • Like
Reactions: JamesCaruso
Feb 19, 2021
73
92
6
Wichita
John,

I used to be. However, I got off Facebook due to the sleazy antics of Mark Zuckerberg. The last straw was when he sent his employees into a hospital to attempt to break into a computer to get dirt and data on people.

I don't think this is a social science issue unless you mean "human factors" as social science, in this case "Why didn't they pull the trigger?"
The problem is that many in NOAA are in denial and I doubt they will be cooperative with a social science study. However, I'd love to be proved wrong.

Mike




Mike, are you a member of the WAS-IS group on Facebook? It is a group that looks at the integration of weather and social sciences, and I think the points you are making would be of interest to that group. For others here with an interest in the relation of social science and societal issues to weather, I would also suggest that group.
 

John Farley

Supporter
Apr 1, 2004
1,627
880
21
Pagosa Springs, CO
www.johnefarley.com
Social science would include analysis of organizational issues that could lead to less effective/timely warnings. Everything from organizational culture and structure to funding trends. For example, understanding WHY they reduced training and moved toward hiring people with less experience, if that is the case. So very much the domain of WAS-IS. Regarding Facebook, yes I have a lot of issues with them, too, but if you want to be on social media there are not a lot of choices. If I avoided every organization I have some disagreement with, I would have nowhere to go. ;-) Not trying to get too off-topic here, but I do think a discussion of why lead time for tornado warnings has been decreasing would very much be of interest to WAS-IS.
 
  • Like
Reactions: JamesCaruso
Jul 5, 2009
1,247
1,193
21
Newtown, Pennsylvania
Here is a link to the damage surveys documented so far. I do not think this is completed yet, for example I don’t see the tornado that hit Long Beach Island (LBI) in the Barnegat area, causing damage both on the mainland and on the barrier island.


The image is the Bensalem tornado path (source - NWS Mt. Holly tweet via PhillyWX.com)

I still have that weird crappy feeling of missing an event as a chaser. Two tornados within a 10 mile radius of home!!! A once in a lifetime event even in Tornado Alley, let alone in southeastern PA!!! What a cruel irony to have been out of town. I could have theoretically seen the first tornado to my north and then still gotten down to Bensalem. Yes a tornadic storm did make it to LBI where I am vacationing, but it was after dark. Oh well, I guess even if I were home it would have been difficult to plot an intercept given the roads, traffic, buildings, trees, hills, etc. around here. I suspect there were a lot of low clouds, limited visibility, and little or no structure to be seen. I suspect most of the tornados were rain-wrapped, although I don’t know for sure. I haven’t seen too much in the way of video actually, but I haven’t looked too hard either. It was great to see the lightning over the bay and ocean; I would not have had that kind of visibility at home. I guess I have to just be grateful I’m able to chase the Plains and don’t have to depend on local events. I’m quite sure nothing on Thursday came close to what I have seen, or hopefully will see, on the Plains, so I shouldn’t be too worried about missing this, but still there’s got to be something incredibly surreal about seeing and experiencing something like this in your own backyard...

2600BD6A-FEE9-4B7E-B471-2B86B91C5EF0.jpeg
 
Sep 5, 2019
29
124
1
Newtown, PA
I still have that weird crappy feeling of missing an event as a chaser. Two tornados within a 10 mile radius of home!!! A once in a lifetime event even in Tornado Alley, let alone in southeastern PA!!! What a cruel irony to have been out of town. I could have theoretically seen the first tornado to my north and then still gotten down to Bensalem.
If it's any consolation, visibility here was terrible - there was a lot of heavy rain with those storms (in addition to the usual trees, buildings and traffic in the area) - and it's highly unlikely that you would've been able to get into position to see either tornado. The only video I've seen that appears to show the Bensalem tornado was taken with a Ring camera - 1:39 in this video:
 
  • Like
Reactions: JamesCaruso
Jul 5, 2009
1,247
1,193
21
Newtown, Pennsylvania
Someone posted this link on PhillyWx.com. I’m sure many of you are familiar with this tool, but it’s the first time I have seen it.


While the Bensalem tornado looks like a damage path I would expect, others such as the New Hope / Washington Crossing tornado are a completely straight line, which can’t be an actual damage path. Any help with interpreting this map would be appreciated.

Also the Barnegat / LBI tornado is not shown on this tool or in the NWS Mt. Holly survey report. Was that determined to be straight line wind damage and not a tornado? Even so, I would assume the survey outcome would still be documented in the report...

@Lou Ruh I saw your survey participation mentioned in the report, any light you can shed on this (and further on-the-ground insights) would be appreciated!
 
  • Like
Reactions: Todd Lemery
To put it lightly, there's a lot of misinformation in this thread that needs to be clarified and checked. Not all of the comments below necessarily fall into that category.

but I always thought there was an algorithm that determines when the velocity couplet exceeds a threshold, so I’m not sure how there could be a deterioration in warning lead time, wouldn't this imply that someone was literally ignoring an alarm?? My understanding is that human intervention can issue a warning earlier than the threshold is exceeded, but not later ... But admittedly I don’t know too much about it, this is just my amateurish anecdotal understanding...
There are a couple of algorithms that have been developed over the years to assess the tornado potential of a storm. The only one that I know that is widely used currently is ProbTor, part of the recently developed suite of probabilistic severe guidance called ProbSevere. "Widely used" is a generous term, as I don't want it to seem that NWS forecasters rely on this algorithm (or any algorithm) to determine whether or not to warn on a storm. If they do, they shouldn't be on the warning desk. However, it is one of many useful tools that can help forecasters determine the severe/tornado potential of a storm. Nothing beats "manual" radar analysis, though.


In the preceding two minutes, the inbounds greatly increased. So, we had a supercell's pendant echo that was starting to grow a hook between Horsham and Abington and a rapidly increasing couplet (likely from an increasing RFD).
If you're an NWS warning meteorologist and you issue a tornado warning on a supercell because it has a hook/pendant in reflectivity, you'll at least be removed from warning duties if you're lucky. To be a supercell, the updraft has to be spinning, which is typically going to lead to a hook. Not all supercells produce tornadoes. As for the velocity data at 2249z, there is no couplet there. What I see is cyclonic convergence, which can (not always) be associated with tornadogenesis in the near term.


Given it was the "tail-end Charlie"
There is research that suggests a storm being the "tail-end Charlie" doesn't mean much for tornado potential. Interestingly, when the storm in question produced the tornado, it was no longer the tail-end storm, as another storm had formed in its updraft region, completely destroying any "classic" reflectivity signatures of the first storm. I believe this second storm produced the Bustleton EF0 that crossed Highway 1 and produced a hell of a debris ball for that weak of a tornado.


with a SIGTOR of 2
There is also research that shows the unreliability of STP during the warm season.


it was a pretty easy warning. That a warning was obviously needed became even evident in the next few minutes as the Bensalem supercell was clearly a right mover.
Given all that was said above in addition to there was not really much of a well-defined, gate-to-gate couplet prior to tornadogenesis, no, it was not a "pretty easy warning". TPHL provided a little better of a look at the couplet, but both it and KDIX looked to be plagued by glitchy level II velocity data. TPHL did show what may have been a low-level current of air emanating from the forward flank toward the mesocyclone, which research has recently picked up on possibly being a pre-cursor to tornadogenesis. Also "a warning was obviously needed became even evident in the next few minutes" - hindsight is 20/20. This is a terrible bias we has humans have when reviewing past events. Of course it was evident - it happened. This train of thought can lead to poor decision making and missed opportunities to properly learn. There are many books on this topic and how to think probabilistically. Here's a good one.

Some NWS offices rely more heavily on various techniques and algorithms
Techniques...yes. Any time you issue a warning, you've applied a technique to do so. It's a matter of if an office (forecaster, really) is up on all the latest and greatest research...or if they're still several decades back issuing warnings based on hooks in reflectivity. Algorithms....just no.


The retirement of my generation of meteorologists and our expertise. This is especially true since we learned to warn tornadoes based on hooks, right-movers and other non-Doppler wind techniques. Given the funnel cloud report at 6:49pm along with the hook (see image) west of Bensalem at 6:47, it would have been an easy warning even without Doppler.
I've seen horrendous warnings produced by folks of your generation. Experience does not always equal expertise. Again, if you put a tornado warning on a storm just because it's a right mover and has a hook, you're going to be taken off the warning desk in today's world. Also, the number of false funnel reports far outweigh the real ones. Good luck going solely off that and how a storm looks in reflectivity.


The hiring of meteorologists fresh out of college. I am aware of multiple instances in three months where the NWS hired meteorologists who graduated in May, even when a more experienced candidate had applied.
As someone who graduated in 2019...oof. What the hell are fresh college grads expected to do? This sounds like the classic "Needs 10 years of experience" for an entry-level job. If you think that newly hired NWS mets hit the forecast and warning desk on their first day, you are very mistaken. It takes YEARS of training. Also, coming from someone who generally lacks confidence in many things (I don't know if I'm even good enough to have imposter syndrome ;)), I can say confidently that there are times (not all the time) when I can out-forecast and out-warn people who have been in the business longer than I've been living. Again, experience does not equal expertise.


Degradated radar training in the NWS. A person currently in the NWS, when discussing the tornado warning problem said to me, "Twenty years ago, radar training was a month in Kansas City. A decade ago, it was a week in Kansas City. Now, it is a couple of hours from the SOO." I don't know whether this is true everywhere or in all cases. But, if it is largely true, it is a serious issue.
Your source is either bitter about something and/or purposefully lying to you, or they are severely detached from how warning training is done. It takes over 100 hours of training that includes guided instruction from the SOO and facilitators in Norman BEFORE you can even get to the week long in-person training in Norman. This then is/should be complemented by at least yearly training from the SOO to ensure you are still up on your skills and the latest techniques and research.


Lack of attention. A quarter-century ago, the NWS worried about the weather wire and NOAA Weather Radio. Now, there is a tremendous amount of social media that has to be accomplished. Is this drawing attention away from mesoscale analysis and monitoring?
No. The warning meteorologist's sole duty is to warn on storms. They are not doing anything else. During severe weather, there is/should be someone dedicated to social media, comms, etc. If there is multi-tasking going on, it's not being done by the warning meteorologist.
 
Feb 19, 2021
73
92
6
Wichita
Hi Alex, Thank you for your comments. FYI, I have been issuing tornado warnings since 1971. I have published papers on this topic and am aware of the tornado research you cited. I do not care comment on every point your raised except the one pertaining to "hindsight."

Attached is my tweet from 6:49pm Thursday evening -- the time I said I would have warned. I don't issue my own "tornado warnings" because I do not wish to confuse people with the official warnings from the NWS. I was extremely concerned about that storm and couldn't believe the NWS didn't issue a tornado warning until 7:07pm.

Part of the basis for my concern at 6:49 was the 6:47pm hook echo (second image). While I agree the velocity data quality left something to be desired, the hook plus the report of a funnel cloud in the area was plenty for me given the atmospheric environment.

I don't agree with all of the research. For example, I've had very good luck with "tail-end Charlies" over the years.

Bottom line: If you feel this was an adequate performance from the PHI NWS, with the warning having a lead time of minus three minutes, that's fine. Everyone looks at things differently. However, I emphatically disagree. Bensalem, My tweet, 649 .png Bensalem Tornado, hook echo, 647p.png I


Also "a warning was obviously needed became even evident in the next few minutes" - hindsight is 20/20. This is a terrible bias we has humans have when reviewing past events.
 
  • Like
Reactions: JamesCaruso
FYI, I have been issuing tornado warnings since 1971. I have published papers on this topic and am aware of the tornado research you cited. I do not care comment on every point your raised except the one pertaining to "hindsight."
An interesting approach after I said "experience does not equal expertise". There's a correlation, but it is far from perfect and usually does not need to be self-proclaimed. If you spoke as if you were up to date with how things are done and the research involved, I would not have felt compelled to comment in the first place.


Bottom line: If you feel this was an adequate performance from the PHI NWS, with the warning having a lead time of minus three minutes, that's fine. Everyone looks at things differently. However, I emphatically disagree.
I did not come here to judge PHI's warning performance, nor did I say I do or do not agree with what they did. I came here to correct/clarify some points with definitive facts so that folks reading this thread can have a proper view of the situation.


Attached is my tweet from 6:49pm Thursday evening -- the time I said I would have warned. I don't issue my own "tornado warnings" because I do not wish to confuse people with the official warnings from the NWS. I was extremely concerned about that storm and couldn't believe the NWS didn't issue a tornado warning until 7:07pm.

Part of the basis for my concern at 6:49 was the 6:47pm hook echo (second image). While I agree the velocity data quality left something to be desired, the hook plus the report of a funnel cloud in the area was plenty for me given the atmospheric environment.
I'm also not saying you were incorrect. If you would have issued a tornado warning at that time as you said, you would have clearly been correct. My point was if you base tornado warnings solely on a supercell's reflectivity, broad rotation, and the environment it is in, your POD will be superb, but your FAR will be abysmal. Those are definitely nudgers, but far from telling the entire picture. If that's all it took, training to issue warnings would not take as long as it does.


I don't agree with all of the research. For example, I've had very good luck with "tail-end Charlies" over the years.
I can't say what you experience was/is wrong. However, your experiences (or anyone's for that matter) are not an objective measure of reality. If you disagree with the research, produce your own objective, peer-reviewed counter argument.
 
  • Like
Reactions: JamesCaruso
Feb 19, 2021
73
92
6
Wichita
Alex,

Again, thank you for commenting. I have reproduced some snippets of your just-published comment below and wish to comment on them as a whole.

If your approach is superior to the techniques I have developed and my experience, why have the NWS's own tornado warning statistics cratered the last ten years?

The 2012-2020 numbers are approximately equal to the numbers in the mid-90's. In most of science the goal is to get better with time. Instead, the NWS has regressed by about twenty years (note: NWS hurricane warnings have improved during this period). The figure below is from my May 21 piece on this topic in the Washington Post, https://www.washingtonpost.com/weather/2021/05/21/joplin-tornado-warning-improvement-nws/ .
Screen Shot 2021-08-04 at 1.19.32 PM.png

If the many examples I have produced are not sufficient, here is another example of a completely unwarned Lake City EF-3 -- no severe thunderstorm or tornado warning. This was the first tornado of the day at the beginning of the underforecasted regional outbreak in Iowa on July 14. Below is my tweet trying to let people in the path of the supercell they were in danger.
Lake City Iowa tornado tweet, July 14, 2021.png
I provided by Twitter followers with +4 minutes of lead time. The NWS lead time was -5 minutes as the tornado moved into Lake City.

While I think I understand what you are saying, I wish to confirm: When you wrote, "IF you would have issued a tornado warning at that time..." You aren't suggesting that I issue my own tornado warnings, are you? When the NWS is clearly missing a major, and obvious, tornado situation, I feel an obligation to let my followers know but I would be extremely hesitant to call them "tornado warnings" as I do not wish to cause confusion with the NWS. Do you disagree?

As to, "folks reading this thread can have a proper view of the situation," I would say this: The National Weather Service's tornado warning program has serious issues at this time. The reasons are not clear to me. That is why we desperately need a National Disaster Review Board -- modeled on the highly successful NTSB -- to, 1) review the performance of all parties when a major disaster occurs and to make recommendations for improvement, and to 2) take over warning verification from the NWS. Perhaps tornado warnings are simpler than the NWS is making them these days.

Again, thank you for the respectful dialog.

Best wishes,

Mike Smith

An interesting approach after I said "experience does not equal expertise".

folks reading this thread can have a proper view of the situation.

If you would have issued a tornado warning at that time as you said, you would have clearly been correct.

I can't say what you experience was/is wrong. However, your experiences (or anyone's for that matter) are not an objective measure of reality. If you disagree with the research, produce your own objective, peer-reviewed counter argument.