Accuweather Warnings?

[Stan Rose]

I find it all pretty amusing. AW has no clue what it's getting into. Issuing a 'warning' is much different than a tv forecast, and you'd better believe there are gonna be liability issues. No question in my mind that they will be sued to death for even the slightest messup.

As for the NWS overwarning, well...duh...of course it is better to overwarn than underwarn. Everyone here knows that tornadoes are not detectable by 88D radar, and obviously it is better to err on the side of caution. So, if AW issues a warning 10 min before the NWS, and it doesn't verify, what then? As a previous post pointed out, there's no real public accountability for FAR or POD, since they are not the "official" source of warnings. They'll trumpet their successes, and bury their mistakes.

I have no problem with the theoretical idea of issuing their own warnings in the spirit of resource sharing and seeing if they can really add anything of value to the current system. But, i wonder who it is they consider their clientelle? How are the warnings going to be put out, and to whom? Is every newspaper that has an AW forecast considered a "client", and by extension, the town that the newspaper serves? How directed or narrow in scope are the warnings gonna be? There's a lot of unanswered questions.

 

[Kiel Ortega]

Considering warning science/technique/scoring is my research, I guess I'll chime in here. At NSSL, we've had a LOT of discussion about warnings and how to make them better. The problem is, ignoring the over-complicated social/logistic issues, technology is kinda stalled on our ability to do WAY better as far as warnings. The exception to this is the polygon (actually, the NWS is calling it storm-based warnings), which Sutter and Erickson (I think MWR in press) found that by reducing warning sizes to 1/4 of a county would be like reducing the FAR to 0! Further, scoring by reports is crap IMO--it gives no sense of value or skill IMO; if you read the NWS directive on how to score warnings, it makes no sense at all. I am going to investigate another way of scoring warnings which scores by affected areas rather than how many reports were received.

- Deliver warnings with more accuracy. The NWS often over-warns the
extent of severe weather, issuing warnings that can cover hundreds of
square miles, when only tens of square miles might be affected.
(my reply): that's nice and all...that's why storm-based warnings are coming from the NWS. Yes, right now the polygons are pretty crappy, but that's because the scoring system is still county-based. As much as I want to criticize the polygons right now, I really can't (but I probably will continue to do so anyways )

- Deliver warnings that provide more detail. This detail would include
precise times various communities would be affected as well as what the
affects will be, such as 1" hail, or wind gusts of 65 mph, etc.
(my reply): this is one thing I am looking at. The first question I am asking myself is can it be done? To truly provide more detail, the warning would need to be dynamic; how do you know if one echo will grow into a storm producing quarter-sized while another will produce golf balls (A: you don't and you can't right now)? Dynamic warnings, IMO, would increase workload without automation. And warnings auto made from cell-based (i.e., SCIT) statistics or identifications would be worthless and using grids (which is what I am doing) is a bit tricky. On top of all of that, more detailed warnings means that some sort of growth/decay considerations are going to be made and there is NO way to do that with any accuracy/skill.


- More lead time of warnings. In severe weather situations, every minute
counts, and by providing a tornado warning 5-10 minutes earlier than the
NWS, lives can be saved.
(my reply): lead time is tricky because an increase in lead time might not be helpful at all. What good is a 60 min lead time if it just gets people out on the roads, rather than in shelter? And everyone has different lead times that are needed to accurately prepare. A family might need a few minutes while a hospital would need upwards of an hour. Also, lead time isn't necessarily the best thing to look at as far as WHEN people got the heads up; as seen in NC, an area that has no warning sirens and it's early in the morning, the amount of lead time may not matter or maybe there needs to be a huge amount of lead time so whatever warning processes take place there can happen. Further, by making tornado predictions 5-10 minutes earlier, the FAR must increase...I see no way that it can't.

 

[rdale]

I'm not convinced on the liability issues, espically if they are going to call these products 'warnings' and have the media distribute them (as opposed to email/cell-phone subscription service).

This is the first I've heard indications it would be openly delivered by TV / radio outlets - and not what their email indicates... What sort of arrangement do they have?

"Yes, right now the polygons are pretty crappy, but that's because the scoring system is still county-based."

That excuse always confuses me... Why would the met draw a horrible polygon? You can still hit the same counties while using an accurate box, yet many polygons still look like county outlines?!?

 

[Scott Olson]

This is the first I've heard indications it would be openly delivered by TV / radio outlets - and not what their email indicates... What sort of arrangement do they have?

Their email doesn't specificy who they are planning on disseminating it to, which is why I've raised that as one of my primary concerns.

 

[rdale]

Gotcha - everything I've seen, including discussions on the TV met forum, indicates this is a subscriber-based service. I find it hard to believe any TV station would go on-air using the word "warning" unless it comes from NWS.

We go on-air and tell people to take cover from a potential tornado even when there's nothing from NWS, but I can't imagine any on-air met announcing an AW warning.

 

[Kiel Ortega]

This is the first I've heard indications it would be openly delivered by TV / radio outlets - and not what their email indicates... What sort of arrangement do they have?

"Yes, right now the polygons are pretty crappy, but that's because the scoring system is still county-based."

That excuse always confuses me... Why would the met draw a horrible polygon? You can still hit the same counties while using an accurate box, yet many polygons still look like county outlines?!?

It's not an excuse. It's how WarnGen works in AWIPS. Basically, the forecaster can make up a warning polygon--basically a warning polygon pops up that is a skewed, smaller on one end and grows larger as it goes downstream. The counties overlaid by the polygon are then put in the warning. The forecaster then has the option to click off counties that they don't want to warn...unfortunately, they don't see the resulting polygon...so the new, crappy polygon is sent out. To clarify, by clicking off certain counties the polygon gets all messed up and looks stupid to users who can see them. The other problem is that even the polygon clipping is even county based (obviously), which makes it look like certain areas are not under warning. The new system will allow just for POLYGONS and the counties have no effect (Greg Stumpf and I have talked about experimenting to see what would happen by turning off county boundaries in the current system and letting forecasters loose in such an environment)...and old polygons can be displayed in the new system (the current system doesn't show older polygons because warnings are county-based). Hopefully I explained it right or Greg or any NWS forecaster can yell at me later

 

[Danny Neal]

i think this all has to do with "whats your cup of tea" It is kind of like the evening news.....which one do you prefer to watch? in my area you have ch 2 5 7 9 and 32 news which one is the most accurate? this is just another option for people to use, you have the NWS, the TV mets, the SPC(not for warnings obviously but for watches and outlooks) its all about who you trust the most when it comes down to it. i think its kind of useless for someone to pay or suscribe to get there own weather service in essence when hypothetically any body with the technology to look at storms can do it and issue warnings.......I dont think its a great idea because at 4 AM if you use the ACCUWX system then you can draw conclusions that you dont have much faith in the NWS but if ACCUWX fails you too where do you turn then? I think instead of everyone whose tied up in the same issue pushing in different directions to try to be the best that they share info with one another and try to make what we have already better, this is human life this isnt about whose better at doing what.....The Nws can issue public warnings like they have been doing and doing a great job like Dan R. pointed out its not the Adminstrations fault people dont take light of the situation, but if the NWS does allthe public warnings then i dont see why ACCUWX cant do what was mentioned in the posts and articles where they can warn different businesses and schools who need more than 5 mins to evacuate.......

to make it clearer........a storm develops outside of dallas tx at 1 pm, lunch hour for most schools........the NWS and ACCUWX confer and talk about the possibility of something happening, AccuWx contacts their suscriber about a THREAT in the near future, the atmosphere is dynamic or w/e the conditions may be that day and they make a call to whoever suscribes to them and say "ok at 105 PM NWS radar detects a growing thunderstorm 45 miles to the southwest of your area we are under a SVR or TOR watch, please take extra caution, and listen for an official warning.....so that school can at least know someone has their back other than getting a warning with the storm bearing down on them and them having to worry about getting 500-1000 children into shelter with only minutes to spare, the same with hospitals and businesses....now this is all hypothetically speaking and probably will never see the light of day but if ACCUwX is serious about this then i dont think they should try to compete and say they are b etter than the NWS i think they should work hand in hand,

just my thoughts on an issue im not 100% familiar with :/

 

[rdale]

It's not an excuse. It's how WarnGen works in AWIPS.

I wasn't directing blame at the line forecasters - I'm saying that the end result right now sucks. Obviously someone developed the software and intended it to do that, and it was tested, and released for that reason...

Regarding NWS conferring with AccuWeather - I'm not sure that's what they want. Even if they did - it wouldn't happen that way.

 

[Danny Neal]

Regarding NWS conferring with AccuWeather - I'm not sure that's what they want. Even if they did - it wouldn't happen that way.

i know i was just thinking outside the box to make both sides seemingly happy, but what do i know maybe accuweather is on to something ill believe it when i see the results

 

[Matthew Havin]

After one botched warning on a large deadly tornado.....

LAWSUIT TIME!!!!!

I know that AccuJack has not been successfully sued before, but the tide will turn when it comes to tornado deaths and the associated emotional toll.

 

[Jonathan.Brazzell]

It's not an excuse. It's how WarnGen works in AWIPS. Basically, the forecaster can make up a warning polygon--basically a warning polygon pops up that is a skewed, smaller on one end and grows larger as it goes downstream. The counties overlaid by the polygon are then put in the warning. The forecaster then has the option to click off counties that they don't want to warn...unfortunately, they don't see the resulting polygon...so the new, crappy polygon is sent out. To clarify, by clicking off certain counties the polygon gets all messed up and looks stupid to users who can see them. The other problem is that even the polygon clipping is even county based (obviously), which makes it look like certain areas are not under warning. The new system will allow just for POLYGONS and the counties have no effect (Greg Stumpf and I have talked about experimenting to see what would happen by turning off county boundaries in the current system and letting forecasters loose in such an environment)...and old polygons can be displayed in the new system (the current system doesn't show older polygons because warnings are county-based). Hopefully I explained it right or Greg or any NWS forecaster can yell at me later

We do get to see the polygons on Awips. Ever since the polygon test two years ago that our office participated in, we have used the polygon method to issued warning based solely on the storm and it's movement. It is unfortunate that most offices have not went to this method yet. It is slated to become official in 2007 or 2008, and hopefully this will take care of that problem.

 

[Joe Nield]

We do get to see the polygons on Awips. Ever since the polygon test two years ago that our office participated in, we have used the polygon method to issued warning based solely on the storm and it's movement. It is unfortunate that most offices have not went to this method yet. It is slated to become official in 2007 or 2008, and hopefully this will take care of that problem.

Exactly. We have what essentially amounts to total freedom to shape the polygons as we see fit using our expertise. Of course, with competing verification systems, there is a tight rope that is being walked right now, and I really believe we'll have to stop serving two masters (county-based and polygon-based verification) in order to succeed fully.

This will also involve our media partners and their vendors catching up to the technology, which it fortunately sounds like this is beginning to occur.

In the end, the main problem remains where it has always been: not with the warnings themselves or their quality, although great strides have been made in that over the last 20 years and room likely remains for improvement, but with the quality and quantity of public response and preparedness. All the new warning sources and longer lead times in the world will do nothing to improve this.

The perfect warning means nothing if it does not reach its intended target. This has been true throughout the history of our young science.

***DISCLAIMER: ALL OPINIONS EXPRESSED HEREIN ARE MINE AND MINE ALONE, AND DO NOT REFLECT IN ANY WAY, SHAPE, OR FORM UPON THE OPINIONS OR POLICIES OF MY EMPLOYER.***

 

[rdale]

This will also involve our media partners and their vendors catching up to the technology, which it fortunately sounds like this is beginning to occur.

Not sure I follow - in what way would TV stations use polygon warnings? I see polygons (when they work) as being a nice way to directly notify people via cellphone / pager / etc but I wouldn't imagine a TV station would plot them...

 

[Joe Nield]

Not sure I follow - in what way would TV stations use polygon warnings? I see polygons (when they work) as being a nice way to directly notify people via cellphone / pager / etc but I wouldn't imagine a TV station would plot them...

This is exactly what I mean. It is mainly a matter of waiting for the vendors, as our partners around Indianapolis have told us, but many of them are apparently close to being able to display the polygons over their radar data of choice on-air. This has already seen some measure of success in the Kansas City market from what I've heard. Our local guys are pretty excited about the prospect of it.

 

[rdale]

Sorry - forgot you work in that market ;> Ones with real meteorologists obviously wouldn't be using the NWS polygons, Barons has had it for years and I can't imagine why a met would show that vs something they draw themselves.

My concern is with the private sector distribution - so far there's been no response but I think that's because the NWS has not pushed this in any way. Hopefully in a few years when NWS is fully polygoned something will change, but I'm afraid AccuWeather may beat them to the punch.