Chance Metz
EF2
Or if it was a big busniess with thousands of employees and it would take some time for everyone to get to their shelter.
Expanding warning times for tornadoes....
- Thread starter Rob Gardner
- Start date
Or if it was a big busniess with thousands of employees and it would take some time for everyone to get to their shelter.
Bart_Comstock
EF5
Not really, that is what scheduled tornado drills are for. Furthermore most large business have numerous safe areas that are spread out across the facilities and are easy to get to. In most places the code demands that this is the case. Rdale's statement still stands.
Greg Stumpf
EF5
My best suggestion would be to go here:
http://www.nssl.noaa.gov/projects/wof/
There are several links to publications here, including their 2020 vision document.
Greg Stumpf
EF5
Actually, storm-based warnings opened up a huge can of worms too. Defining warning by geographic polygons, versus geopolitical boundaries (counties) becomes very complex using legacy warning communications systems such as NOAA Weather Radio. It turned out that a lot of users were unprepared for the change.
How would it completely change the current system? As I've stated before, legacy warning information can be extracted and the output looks identical to current warnings. Then, if there are users that require higher detail and more sophisticated output, it is available to them, unlike today.
Explain to us what is the problem with increasing warning lead times (i.e., improving warning accuracy at longer forecast periods)?
I'm also told that on occasion. The unfortunate truth is that scientists simply do not understand the atmosphere, nor the complexity of our users, to tell each and every one of them exactly what to do and when! We can pander and get it wrong and lose credibility, or we can do a better job of communicating the message (which includes uncertainty).
Thanks for the offer, but I'm not qualified to lead focus groups or any other social science data gathering. And neither are you! Let's leave that to the social scientist professionals.
BTW - I am currently reading "Warnings". I may have more to say later.
Mike Smith
Greg,
First point: I disagree. While I'm not a "social scientist" one cannot possibly succeed in private sector meteorology (where the government is giving away a free version) for four decades unless you can carefully listen to and understand your clients. Then, once you listen and understand, you have to be able to respond to their concerns. No one is going to pay a private sector meteorologist unless they generate considerable value as measured by the client.
I have no axe to grind -- I want to provide my clients with the best service and I want to see lives saved amongst the public. On several occasions in the past, we have tried to pitch probabilistic forecasts and warnings to our clients and they are not interested. I have also brought the topic up with groups I have spoken with. They are not interested.
Second point: Thank you! I look forward to your comments.
Mike
rdale
EF5
I just went on a "five city tour of the great state of Indiana" and did get positive feedback from the city departments I talked with and mentioned WoF.
If they involve citizens, they don't need to be interested. They'll still get the "get into the basement" alerts and tornado sirens just like before.
Danny Neal
EF5
Thanks for the link to the WoF page. Something I haven't heard of before. Makes me wish that I continued my met. education and got involved with programs like these! I will take advantage of the documents and publications provided, but before I do I wanted to know is this aimed toward the public (if they wanted it) or more toward business/private owners. I saw most of the examples on the home page geared toward nursing homes, stadiums, airlines, etc. I'd love to see this in action and how accurate it can be. Now on to reading....
Robert Dewey
EF5
So if I'm understanding this correctly, the WoF is basically a probabilistic approach to warnings... based on NWP? As long as this is hidden from the end-user (public), I don't see a problem with it. However, what happens if someone abuses this data? Telling the public there's only a 60-70% chance a tornado will affect them would do more harm than good, IMO.
I also agree with Mike Smith that you don't need to be a social scientist to understand human desire. Invention and innovation are constantly occurring without the input of social scientists.
I also agree with Mike Smith that you don't need to be a social scientist to understand human desire. Invention and innovation are constantly occurring without the input of social scientists.
Mike Smith
I wanted to address this separately.
I do not believe that "longer is better" is correct. Beyond a certain point, there is a "diminishing returns" problem. Based on anecdotal evidence and my experience talking with tornado survivors the following seem to be true:
1. There is only so long that people will stay in the basement or a shelter area
2. Things can "go wrong" the longer the warning is in effect.
While I have several examples of #2, let me give you just one: During the Andover tornado, two latchkey girls ordered a pizza. They saw me go on the air and tell them to go to the basement. They did so. The doorbell rang and they debated what to do. They figured that if the guy was delivering pizza things could not be too bad. When they opened the front door, the family cat ran out. One of the girls chased the cat around the side of the house and she saw the tornado bearing down on them. She ran back to the house, shouted to her sister, they went back into the basement as the house blew away around them. Again, I have more examples of this (including some on video) but I think this suffices.
So, no, I don't think two hour WoF tornado warnings are a good idea.
As to what the time intervals of tornado warnings should be, my sense is that 15 to 30 minutes is about ideal. Yes, I know that WoF is proposing "floating" time intervals (i.e., that they move with the storm) and that is a good idea but I have no idea how that is going to get effectively communicated in a complex situation (i.e., more than a single supercell).
Finally, with regard to the WoF web page, I note this with interest:
Potential Benefits
Longer warning lead times are needed for tornadoes, severe thunderstorms, and flash floods:
Many hospitals and nursing homes require 30 minutes or more to move patients to safe locations.
Large venue operators such as sports stadiums require at least 30 minutes to move thousands of people from exposed locations to safety.
Towns may need more than 30 minutes to evacuate residents from low-lying areas threatened by flash flooding.
Aviation officials need time to reposition aircraft to protect it from an approaching storm.
These are all the purview of the private sector. I can assure you, WeatherData has clients in each of these areas and has already designed effective solutions for each organization's unique requirements.
To close, the NWS really needs to work on the false alarm problem but keep the current warning structure in place.
Hope this information is helpful.
Mike
Greg Stumpf
EF5
I've already addressed this issue a few posts back. Point-specific time of arrival and time of departure information is key.
By that logic, then we should do away with tornado watches and convective outlooks, all which present tornado forecast data at longer forecast intervals.
Perhaps the misunderstanding is based on how WoF is being advertised. It's not really an extension of warning lead times. WoF should provide more-accurate hazard forecasts at longer forecast periods. That will allow users to feel more comfortable in taking specific actions in which their cost-benefit ratio is now improved by the better forecasts/warnings.
Using NWP to improve hazardous weather short-term forecasts means better accuracy, i.e., higher POD and lower FAR, and can be done through the legacy hazard information delivery structure.
Greg Stumpf
EF5
You are talking about market research, which is very different from social science research.
Mike Smith
1. Huh? I'm talking about warnings where people should take cover in this discussion. I've previously written that there is potential for these new techniques to improve watches and similar products.
2. Greg, if you go to the web site you linked to, there are numerous statements about increasing warning lead time. The very first sentence on the web site says,
Warn-on-Forecast is a National Oceanic and Atmospheric Administration (NOAA) research program tasked to increase tornado, severe thunderstorm, and flash flood warning lead times.
I don't know how anyone would reasonably expect that this program is not about increasing warning lead times. I think that -- as stated above -- is bad idea. If the program is about something else, then I urge you and NSSL to reword the descriptions of the program on the web site.
3. Great! Then, we are in agreement in that regard.
Mike
Last edited by a moderator:
Mike Smith
Other than the target audience, how do they differ?
Patrick Marsh
EF5
- Joined
- Apr 18, 2006
- Messages
- 783
What's the different between a Warning and a Watch? Typically (and simplistically), we use the term warning (in the convective sense) when a hazard is imminent (and use a <1 hour time scale) and a watch means a hazard is possible (and use a <8 hour time scale). What happens when we starting developing products in between the watch and warning time scale? One that begins to blend the two? What do we call it?
Let's say that I value a missed detection considerably more than a false detection. In other words, I'd rather a high FAR than a low POD. In this case, I might take the same action I would for warning (based on current paradigm) whenever my threat increased to or exceeded 25%. What if I value false detections more than missed detections? In this case I might not take action until my threat increased to 75%.
This gives rise to the question, "What is the threat threshold at which the warning forecaster will issue a warning?" Is it 25%, 50%, 75%? We don't know because it varies for each forecaster. WoF aims to remove this ambiguity by making the underlying probabilities available.
Consider this example:
An end user wants to be notified anytime his threat of a tornado within some long time period (say 6 hours) is greater than 10% and he wants to be notified again if the probability of a tornado is greater than 50% within a shorter time period (say 1 hour).
With the probabilities generated by WoF, the end user can be notified the moment his 6 hour probability of a tornado reached 10%. This would be his short term tornado outlook. As long as the 10% threshold is exceeded, he knows he's at a risk of tornadoes -- the user begins to maintain weather awareness. He also gets notified whenever the 3 hour tornado probabilities exceed 25%. He can consider this his tornado watch. As long as his probability of a tornado is greater than 25%, he is in a tornado watch. When the probability of a tornado within an hour exceeds his 50% threshold, he can consider this his personal tornado warning. As long as his probability remains above 50%, he's in a tornado warning.
They key is that each user can use his or her thresholds (both in time and threat) to determine what constitutes a watch or warning, instead of relying on someone else's thresholds that may not suit the end user's need. For users who don't have any knowledge of what their thresholds need to be, private weather companies can help develop the appropriate thresholds. For the general public, they can just rely on the default thresholds set by the local NWS office for that event. In this example, the current warning structure remains in place, but end users can alter the default settings to best reflect their needs.
WoF should be considered as a continuous flow of information from short-term outlook to warning scales.
WoF is being developed by OAR, not the NWS. NSSL (which is a part of OAR) and OAR are supposed to push the edge of capabilities and develop the next generation products and WoF is this next generation of convective forecasting. NWS already assigns probabilities to convective watches. Shouldn't NSSL proceed to work on making these probabilities as scientifically robust as possible? In any event, even if NSSL/OAR develop these technologies, it's up to the NWS to implement them. But regardless of implementation, it doesn't mean that NSSL/OAR shouldn't work toward this goal. A lot of important things look to come out of this research that will benefit a lot more areas than just convective warnings.
Mike Smith
Patrick,
Thank you for jumping in. Here are my comments...
It isn't the job of the NWS to worry about "an end user." That is the purview of the private sector. I want to emphasize this again: WeatherData (and many other commercial weather companies) has hospitals, aviation companies, cities worried about flash flood evacuations, and numerous sports venues as clients. During the recent NYC tornado, the Mets moved their early arriving fans into shelter based on our warning long before the NWS issued its first warning.
Because we work one-on-one with clients we can fashion solutions that meet each client's unique needs and we have been extremely successful in these endeavors. Given today's budgetary climate, I'm not sure why the Norman branch of NWS/NOAA/NSSL/OAR doesn't seem to realize or acknowledge that the private sector has complementary capabilities and the NWS does not need to duplicate them.
Greg said of the program, "It's not really an extension of warning lead times." If you go to the home page of http://www.nssl.noaa.gov/projects/wof/ you will find that extending lead times is the only goal mentioned. You say it is not NWS doing the work. The NWS is talked about on the home page and not NSSL/OAR. What the two of you are saying in this discussion is inconsistent with the official documentation.
The result is a confusing set of mixed messages that makes me wonder what this program is really about. The description that Greg gives is quite different than the one John Snow gives. I suggest that some thought be given to reworking the descriptions and messaging regarding WoF. That way, we all might be able to generate more light than heat on this topic.
I hope none of this comes across as harsh because I do not intend it that way.
Mike
