Some offices not following the NWS directives for TOR's!
DTX issued one for nearly an hour for a storm moving into the western part of a county, headed east at 40, and the county is only 20 miles wide. Cancelled early.
- Rob
Why?
Considering that nobody who has responded to this thread is currently assigned to GRB nor were any of us in the office during the events in question, why not e-mail the WCM these questions? It's hard to piece together the whole story from just a few snippets of information. Why not get it straight from the horse's mouth?
The WCM's name is Jeff Last. Here's his page at GRB:
http://www.crh.noaa.gov/grb/wcm.html
Here's his e-mail address:
jeff.last@noaa.gov
Jason
Good point - I never pictured that as the issue but I think you are on to something... Maybe if the local NWS office wasn't in a bubble then things could change - should I transfer the irate callers direct to them as a start ;> ?
- Rob
I'd give the benefit of the doubt to the forecasters making the call while in the hot seat. I'm sure they have good reasons for erring on the "long" side of things...
So if you think a storm is going to intensify, or even has the potential to do so in your county, you might issue a warning with some wide temporal boundaries, much like SPC issues watches that comfortably cover areas from initiation through the temporal and spatial lifespan of any storms they expect, and with some room to spare. I don't see this as a mistake--it's good sense.
Heck, even Six Sigma allows for an occasional mistake! :-)
Regards,
Mike
Has there been any recent independent publications on FAR/POD/CSI statistics recently? That would shed a lot of light on what may be happening in these instances.
Mike or Evan, if you're lurking, give us some OTR insight and opinions on this.
Regards,
Mike
I don't have any hard stats but sadly a significant number of people don't know their county and/or have difficulty finding their location on a county map. Many do, but many don't.
Scott
For that matter, if the NWS print warnings could go out with a small graphic that had the county in question with surrounding counties labled. In this day and age, it can't be that hard to incorporate a small pic.
:dontknow:
Mike
http://mgweather.com
Are verification numbers available anywhere?
- Rob
WFUS53 KGRB 240214
TORGRB
WIC009-087-240315-
BULLETIN - EAS ACTIVATION REQUESTED
TORNADO WARNING
NATIONAL WEATHER SERVICE GREEN BAY WI
908 PM CDT WED JUN 23 2004
Issued at 908, across the wire at 914 (strange delay) and valid to 1015p so 67 minutes in my math. And the storm left the county by 925p
Is this the warning you're referring to:
http://kamala.cod.edu/svr/0406240214.KGRB.wfus53.html
008
WFUS53 KGRB 240214
TORGRB
WIC009-087-240315-
BULLETIN - EAS ACTIVATION REQUESTED
TORNADO WARNING
NATIONAL WEATHER SERVICE GREEN BAY WI
908 PM CDT WED JUN 23 2004
THE NATIONAL WEATHER SERVICE IN GREEN BAY HAS ISSUED A
* TORNADO WARNING FOR...
BROWN COUNTY IN WISCONSIN
OUTAGAMIE COUNTY IN WISCONSIN
* UNTIL 1000 PM CDT
* AT 908 PM CDT...LOCAL LAW ENFORCEMENT REPORTED TWO POSSIBLE
TORNADOES IN OUTAGAMIE COUNTY. ONE STORM WAS PRODUCING A TORNADO
NEAR KAUKAUNA NEAR THE INTERSECTION OF I47 AND HIGHWAY 55 OVER
SOUTHERN OUTGAMIE COUNTY. A SECOND TORNADO WAS REPORTED TO BE
DEVELOPING NEAR NICHOLS OVER NORTHERN OUTAGAMIE COUNTY. THE STORMS
WERE MOVIN EAST AT 40 MPH.
* ONE TORNADO WILL BE NEAR...
SEYMOUR AROUND 915 PM...
ONEIDA AROUND 930 PM...
AND THE SECOND TORNADO WILL BE NEAR...
WRIGHTSTOWN NEAR 920 PM CDT...
MOVE TO A BASEMENT OR SMALL INTERIOR ROOM ON THE LOWEST FLOOR OF YOUR
BUILDING.
LAT...LON 4457 8856 4429 8834 4428 8801 4460 8805
$$
TH
I, like Kevin, never found any warnings issued at 9:08 that lasted until 10:15.
Jason
Why?
The DDC CWA does not have very many towns compared to most other parts of the country. There is a lot of open rural area, and if there's no town impacted, we would get a lot of "X miles south of Y town at Z time". If a thunderstorm is quite large, the arbitrary location with reference to a town can be confusing and sometimes misleading. A location may be getting strong winds and heavy rain in advance of a pathcast time as well. Yes, customers can be confused when we tell them that a severe thunderstorm will be there at Z time and they may be getting non-severe thunderstorm conditions well before the predicted severe thunderstorm time. There can also be inconsistencies in the times from initial warning and subsequent SVSs, and this can be due to only subtle change in the warning operator's track of the storm. Again, more unnecessary confusing to the public, in our eyes.
This is something fairly new our office is doing, and we'll just have to wait and see how the public/media responds. In our eyes, there's just too much uncertainty in nailing down an exact minute for a town in the warning. We prefer to keep it simple by saying "hey, this location will be impacted at some point during the valid time of this warning."
Mike U
In theory, the text product should not be touched unless additional information, ie. the inclusion of a spotter report, etc.. are inserted. Warngen, however, is far from perfect, and there are many times when the SVR, TOR, whatever needs to be edited. I find it strange that the expiration time was changed... because if you change the expiration time, you MUST change the UGC coding on the top (and soon to be VTEC coding)... it is a NO-NO (in our office at least) to edit UGC coding!! If there's an error in expiration or counties, we re-gen it with warngen.
Anyhoo... warning duration philosophy is never easy, and the warning forecaster will usually error on the side of "longer warning" vs. "shorter", at least that's why I do. It is very simple in warngen to gauge the duration of the warning if one has the storm motion correct... simple linear extrapolation, which warngen shows graphically. Of course, storm forward motion is never always constant.
It takes experience, obviously, as a radar meteorologist when issuing sound warnings. I can't imagine they've had a ton of svrwx in wisconsin so far this season... and with new builds always occuring with AWIPS, etc.. sometimes Warngen templates get messed up, etc.
On another SVR & TOR note... the DDC office no longer gives pathcast TIMES... only the pathcast towns or the proverbial "rural portion of X county". ;-) I love the desolate high plains!
Mike U in DDC
Severe Weather Statements should be issued at least once during a warning and may be issued at the conclusion.
- Rob
I have also notice a few times this year, seeing a Tornado Warning was issued for such and such county and never seeing a SVS sayiing the Tornado Warning was cancelled for that county.
Mike
Mike, of course if the tornado warning goes until expiration and never needs to get cancelled, there's not going to be a cancellation weather statement. If the metr has timed it correctly, there really shouldn't be too many cancellations. :)
- Rob
I agree personally.... However, in the name of verification statistics, a 2 30-minutes tornado warnings that don't verify is worse than one 60-minute tornado warning that doesn't verify. I have a hunch that this is largely driven in the name of verification stats... PoD seems to be more valued than the FAR, and I assume much of that is because of political pressure. You know as much as anyone, that it's much better from a political viewpoint to warn many and hit all (high FAR, PoD=1) than to warn few and miss one (low FAR, lower PoD). The county EM will get many calls from local reisdents and media complaining that they had no warnings, they'll call their local legistlators, and next thing ya know you have a bad performance evaluation becuase you missed a tornado... Yes, I think it's all crap, but I don't think that'll change anytime too soon unless people really realize that meteorology, while more advanced than ever, is still fuzzy in many areas...
Again this is NOT confined to any one office. GRB happened to have the warning I posted primarily because they screwed up the UGC code so it appeared worse than it was. But it's happening more and more often and others in this thread have covered the fact that it's a CYA vs scientific.
- Rob
TORGRB
WIC009-087-240315-
BULLETIN - EAS ACTIVATION REQUESTED
TORNADO WARNING
NATIONAL WEATHER SERVICE GREEN BAY WI
908 PM CDT WED JUN 23 2004
Issued at 908, across the wire at 914 (strange delay) and valid to 1015p so 67 minutes in my math. And the storm left the county by 925p
DTX wasn't cancelled til :40 into the warning but the storm had cleared the county at :31 into the warning...
I understand the need to pad things - but again if a storm is moving at 40 and the county is 20 wide it'd take a LOT to still be in that county an hour later. ESPECIALLY for storms that only briefly showed TOR characteristics on radar. The DTX one was halfway into the next county when the warning was originally set to expire.
- Rob
On a similar note, I notice some NWS offices do a excellent job, when it comes to issuing timely Severe Weather Statements, telling you the severe thunderstorm warning or tornado warning continues for such and such county and having statements telling you when the warning is cancelled. Some offices I have notice do a fair job in issuing severe weather statements.
I have also notice a few times this year, seeing a Tornado Warning was issued for such and such county and never seeing a SVS sayiing the Tornado Warning was cancelled for that county.
Mike
The NWS directives quite clearly state that Tornado Warnings should not exceed 45 minutes when issued. It does not mean that you can issue an hour+ warning and cancel early - it says that when tornado warnings are issued they are to be for a duration of 15-45 minutes.
A lot of this mess could be solved if severe weather warnings could be issued by polygon alone, and throw out the notion of warnings tied to geopolitical boundaries - afterall storms aren't guided by these boundaries. The technology has been here for years. Even though NWS folks use the polygon tools in AWIPS, the polygons are more often than not still "tweaked" to the county borders, and the final warning product is mainly tied to the counties. The polygon coordindates are an "afterthought" placed at the very bottom of the warning.
Then, for each severe weather event, 15 or 30 minute updates could be issued with another polygon. Instead of Severe Weather Statements, this is a new warning (and polygon) replacing the old warning (and polygon). Whatever counties, parts of counties, or communities affected (inside the polygon) would receive the warning. Re-issuances of polygons takes care of the situations where storms move out of a region. If a storm weakens, then the cancellation is tied to the final polygon warning.
Did I make sense?
greg
I've been notified of that difference since then... I can understand on occasion the need would exist for a longer warning, but in talking to Jeff @ GRB he sent some average TOR times from neighboring offices and he was the smallest of the lot -- but several averaged 40-44 minutes. If the average is on the high end of the "should" limit then the "should" is apparently being interpreted too loosely.
- Rob
Compared to a TV station - yep! Ask anyone on the street where the NWS office responsible for warnings is and I guarantte NOBODY will know it's 60 miles away.
"Once again, have you actually contacted the forecaster or office who issued this 67 minute warning to ask them what was going through their heads?"
I have talked with GRR and they are trying to get keep their times down as well as issuing partial-county warnings if possible. I talked to DTX about a tornado warning that was issued for a non-existing thunderstorm (the tornado had occurred 30+ minutes prior, radar was EMPTY, and the report was third-hand) and it was explained to me by the WCM that meteorologists can interpret radar in different ways as the science isn't perfect. No sense in continuing THAT conversation. Here's what they posted to a wx list:
"Several cases have suggested we re-issued too many warnings for the same county if the storm is slower or redeveloping."
- Rob
where it reads 240214, 0214 which is 914 pm CDT
to 240315, 0315 is 1015 pm CDT,
compare to the 908 pm CDT to 1000 pm CDT in the text.
008
WFUS53 KGRB 240214
TORGRB
WIC009-087-240315-
908 PM CDT WED JUN 23 2004
THE NATIONAL WEATHER SERVICE IN GREEN BAY HAS ISSUED A
* TORNADO WARNING FOR...
BROWN COUNTY IN WISCONSIN
OUTAGAMIE COUNTY IN WISCONSIN
* UNTIL 1000 PM CDT
Mike
A Tornado Warning was issued for Peoria Co at 10:04pm until 11pm, then a Tornado Warning was issued for Peoria Co at 10:23pm til 11:30pm. Not quite sure why that happened but in any case 10:04 til 11:30 is a 86 minute warning. 10:04p - 11p is 56 minutes, 10:23p til 11:30p is 67 minutes. The NWS directives quite clearly state that Tornado Warnings should not exceed 45 minutes when issued. It does not mean that you can issue an hour+ warning and cancel early - it says that when tornado warnings are issued they are to be for a duration of 15-45 minutes.
"so 2 SVSs for the Peoria county warning"
I was responding to Mike & Jeff above.
"Did you ever contact Jeff last at GRB? "
You may have missed my previous post - it appears GRB was more of a coding error because the met tried to tweak things he wasn't supposed to tweak and the UGC was messed up. That's not clueless, just a mistake.
"Why don't you contact them about your concerns, and we could all benefit from a better understanding of the basis for their office's decisions on warning times."
I'm just forwarding these to the HQ contact for the directives, I figure since he is the one responsible for making sure that TOR's never exceed 45 minutes he'd get a little more respect from the offices. It amazes me to see so many offices going against the regulations that were written to make sure all offices do things the same way instead of individual forecasters making things up on their own...
- Rob
- Rob
- Rob
- Rob
If you read the directives carefully, you'll see that the rule is "Valid times should be 15 to 45 minutes..."
The way the directive-ese jargon works, it means that 15-45 minutes is a recommended guideline, not a hard rule. A hard rule would have used the term "SHALL."
Offices are given the ability to extend the valid times if they think the situation warrants it. You may disagree with the meteorological validity of the warning times, but that's a whole other arguement, one that would best be addressed by contacting the forecaster or WCM from the office that actually issued the warning, not someone in a glass bubble up at HQ.
You'd be amazed at the frustrating flaws in WarnGen, which is luckily being sent out to pasture soon. Hopefully whatever replaces it isn't even worse. I'd be willing to bet many forecasters would rather issue a warning for longer than the recommended 45 minutes, rather than having to deal with re-issuing the warning, especially if software issues may result in something not being transmitted correctly, causing even more confusion amongst the public.
Oh yeah... clearly,
*These opinions are my own, and not my employer's, and are being submitted from the comfort of my own home on my own time...*
-Mike
HA! In a bubble!?!? OK...
Once again, have you actually contacted the forecaster or office who issued this 67 minute warning to ask them what was going through their heads?
Hate to rehash an old topic but was checking on ILX (they had Peoria County under a TOR for 86 minutes) and found that the NWS directives clearly state that Tornado Warnings should not exceed 45 minutes. SVR's can be up to an hour.
Severe Weather Statements should be issued at least once during a warning and may be issued at the conclusion.
- Rob
The Peoria county warning was cancelled at 11:15 :
390
WWUS53 KILX 060414
SVSILX
ILC143-179-203-060430-
SEVERE WEATHER STATEMENT
NATIONAL WEATHER SERVICE LINCOLN IL
1115 PM CDT MON JUL 5 2004
...A TORNADO WARNING CONTINUES FOR WOODFORD AND TAZEWELL
COUNTIES UNTIL 1130 PM CDT...
...THE TORNADO WARNING FOR PEORIA COUNTY HAS BEEN CANCELLED...
So the warning was in effect for 52 minutes, not 86. A tornado was reported with this cell at 10:40 - so a lead time of 17 minutes, which is actually pretty good. An SVS was also issued at the time of the tornado report - so I don't really see anything wrong with how the situation was handled there (so 2 SVSs for the Peoria county warning). Regulations may advise that the warning times are limited, but there could be reasons behind the choices to sometimes have extended warnings times that are not outwardly apparent, so it seems unfair to lambaste the weather service on every occasion.
Brown county in in NE was under a TOR for 57 minutes last night - but in two separate warnings. To me, this is no better or worse than if they had issued a single hour long warning and put out an SVS at at the mid-point (they actually issued several SVSs).
I currently live in the ILX CWA, and I do have my fair share of gripes with their performance - but these are more based on whether a warning should be issued for a particular storm and for what area (county-wide vs. partial county - the latter is NOT done here). Warning duration just doesn't seem like that big of a problem so long as there is still active convection within the warned area.
Did you ever contact Jeff last at GRB? Chris Miller is the WCM at ILX: chris.miller@noaa.gov
Why don't you contact them about your concerns, and we could all benefit from a better understanding of the basis for their office's decisions on warning times.
Glen
Mike U
That makes sense to me...
The by-the-minute feature relative to the general uncertainty of future storm motion, etc, reminds me of another thing quirk I have -- street level mapping... I always get a kick out of local media when they pull up a radar map and zoom into a storm, oh, 120-150 miles away. Then, they zoom in and activate street-level mapping, and proceed to say EXACTLY (i.e. the intersection) where the rotation is over... But wait, being 120 miles from the radar site, they are telling where the mid-level rotation is, not necessarily where the low-level rotation is.. By being so detailed in their analysis, I think it portrays a false sense of security for those in the area but not living on the exact street they say the rotation is... There've been cases where the actual tornado was 5-10-15 miles from the center of the rotational couplet on the lowest scan... The only time when real time street-level mapping should ever be used is when the feature-of-interest is close enough to the radar to make using street-level mapping valid. I know there are politics / ratings behind a lot of this, but it just bugs me...
here is the link to Tulsa
http://www.srh.noaa.gov/tulsa/sver.html
Speaking as a user of those products, I always found it confusing to hear "17 miles SSW of East Podunk or 12 miles E of Hicksboro" as a storm location/path forecast. In the same token, leaving out all pathcast times doesn't make a lot of sense to me, either. I like a blend of sorts...when it's going to pass over a town, use the town name, and pathcast the time as "by (time)" instead of "at (time)." When no town is in the path during the forecast period, "rural portions of (county)" works very well.
Good thought which highlights the public awareness problem... probably even worse in the Northeast where there is no county government. What good is a warning when the people who actually receive them (another issue) ask themselves, "where the heck is that!?"
Main problem here is with the systems that receive the warnings. I don't know the details, but we are limited to 69 characters per line, all caps, and no commas, quotations, or apostrophes, etc.
Given those limitations, you can pretty much discount graphics embedded in warnings. Some web sites will, like you mention, show the counties, including NWS office homepages. Some of the graphics, like you imply, could be better.
Besides, if the NWS put graphics right in the warnings, the private sector would probably find some reason to get their knickers in a knot over it.
*See previously used disclaimer!*
Unfortunately the wx infrastructure has not developed - probably due to lack of a push from NWS. Our Barons display can utilize the polygons but the WSI can't. Many counties don't have siren systems that allow for partial sounding - either all or none. For non-visual delivery methods (i.e. NWR, radio, pager) I think it would be difficult to explain but in that case you go with the "county method" and direct people to other sources for a detailed display?
- Rob
With that logic, one could take the extreme. One office could issue a continuous tornado warning for their entire CWA for the entire year. As long as there is one tornado in their CWA that year, they have a POD of 1 and an FAR of 0.
Yes, the NWS GPRA goals do emphasize higher POD more than lower FAR, but a low FAR is still very important. If not, then offices would tend toward the above.
The whole procedure for warnings and warning verification needs an overhaul. It needs to be area and temporally based, perhaps on a grid like the SPC now uses for watch verification, rather than based on individual warnings and counties warned. There are some NSSL researchers that have proposed the idea.
greg
- Rob
The technology has been here for years. Even though NWS folks use the polygon tools in AWIPS, the polygons are more often than not still "tweaked" to the county borders, and the final warning product is mainly tied to the counties. The polygon coordindates are an "afterthought" placed at the very bottom of the warning.
Did I make sense?
That makes sense, I only wonder how customer-friendly that product would be. People know what county they live in and tend to know their position relative to surrounding counties. If they heard a polygon described over NWR or even in scrolling text on their television, could they visualize the warned area to determine their own position relative to it? That is, if their community isn't mentioned or if they live in an unincorporated rural area?
Even with as much practice as chasers have, it's still a trick to visualize perfectly a tornado watch parallelogram from coordinates over the radio.
#If you have any other info about this subject , Please add it free.# |
