Expanding warning times for tornadoes....

[Mike Smith]

Rob,

While I agree that this is what the thread states multiple other people associated with the program contend (and advocate) that WoF will result in "probabilistic warnings." I have the emails from two program participants to document that PW's are being touted as "replacements" for yes/no warnings. That was also what was stated in the 2007 meeting and the subsequent meeting at the AMS annual meeting in 2008.

This is why I have spent the time on this thread. As I have previously stated, this may be a valuable program that will enhance 1-2 hour forecasts but that is not what is described on the web site and the very title of the program, "warn on forecast," suggests otherwise.

Mike

 

[rdale]

I understand your concern based on those tidbits - but at some point reality would set in if they ever approached NWS offices and private sector partners and said "We aren't providing yes/no warnings ever again."

 

[Mike Johnston]

I don't meant to be rude - but have you been reading this thread? There is no chance that your wife would ever get that sort of information unless she specifically dug into the data to get it. She'd still get her warning information from hearing the sirens or getting a text message from the local TV station or seeing a crawl on the screen. There is 0% chance of her seeing percentages.

Well, if there would be absolutely no change for Joe or Jane under this WoF/probability-based system, then why are some advocates highlighting an important role for sociological studies in the development of such a system? If it's meant for sophisticated commercial users only (intermediate decision makers responsible for the safety of others and protection of their assets), then I think Mike Smith is absolutely correct - the private sector, understanding the unique needs of their clients, is best suited to handle it.

 

[Mike Gibson]

I *really* don't understand the opposition to this new gridded dataset from the NWS (if it ever sees the light of day given the comments here). If people want to use it, they'll use it. If not, they can continue using the existing warning system derived, in part, from the grids.

The private sector can derive user-specific information for clients from the grids, just as they do now from their own products, so I'm not sure where that opposition comes from.

I just wish the NWS/NSSL/whoever would start with a much less demanding and easier to produce hourly product so that those of us in the vaunted private sector can start experimenting with ways to convey the information to end users.

Mike

 

[Greg Stumpf]

Well, if there would be absolutely no change for Joe or Jane under this WoF/probability-based system, then why are some advocates highlighting an important role for sociological studies in the development of such a system? If it's meant for sophisticated commercial users only (intermediate decision makers responsible for the safety of others and protection of their assets), then I think Mike Smith is absolutely correct - the private sector, understanding the unique needs of their clients, is best suited to handle it.

Are you suggesting that the NWS shouldn't understand the unique needs of their clients (public, EMs, community gatekeepers, etc.)? Then what is the role of NOAA? Should NOAA no longer advance the science? That's exactly what WoF is designed to do - improve the accuracy of short-term hazard predictions. Should NOAA not be doing this? If so, at what point does NOAA become irrelevant?

Consider this - if a NWS forecaster is watching a storm for severe weather potential, but decides to not yet issue a warning because their confidence level hasn't reached their internal "threshold" for hitting "Send". Then, what if your present situation is more-sensitive to the hazard (perhaps out hiking and 60 minutes away from a substantial shelter), would you not want hazard information earlier? If the answer to that is "yes", then at what "threshold" would you want the forecaster to tell you about the chance of the threat. And how would the forecaster frame that message for you?

 

[rdale]

Well, if there would be absolutely no change for Joe or Jane under this WoF/probability-based system

Maybe I'm reading too much into what you are saying - but I think you are oversimplifying.

In our "watch/warning" system, I don't know of any EMA that sends out their fire trucks / police / etc for additional spotting in a watch. They wait for a warning and rely on Skywarn until that happens. Well WoF could allow them to say "If % > xx then dispatch the public safety community" which would benefit Joe & Jane.

And it's okay if you tell me I'm stretching - that's fine. But it's not okay if you tell me our current watch/warning system is PERFECT. It's great, I'll give you that. But I find it hard to declare that it cannot be improved upon in any way. WoF is one possible way of doing that, and the social science research helps tell that story. The current process was NOT developed with social science in mind.

 

[Greg Stumpf]

I *really* don't understand the opposition to this new gridded dataset from the NWS (if it ever sees the light of day given the comments here). If people want to use it, they'll use it. If not, they can continue using the existing warning system derived, in part, from the grids.

The private sector can derive user-specific information for clients from the grids, just as they do now from their own products, so I'm not sure where that opposition comes from.

I know several private sector meteorologists that are chomping at the bit to have data like these (digital hazard grids) in the future to use in their content delivery systems. In fact, one was from an unnamed private company in Pennsylvania!

 

[Greg Stumpf]

While I agree that this is what the thread states multiple other people associated with the program contend (and advocate) that WoF will result in "probabilistic warnings." I have the emails from two program participants to document that PW's are being touted as "replacements" for yes/no warnings.

I can tell you right now that this is not the vision of us folks who touting digital probabilistic hazard grids. WoF may result in probabilistic threat guidance grids, but warnings and products (including legacy) can be derived from the grids. I've regurgitated this message too many times here, so I won't say more.

This is why I have spent the time on this thread. As I have previously stated, this may be a valuable program that will enhance 1-2 hour forecasts but that is not what is described on the web site and the very title of the program, "warn on forecast," suggests otherwise.

If you could frame your notion of WoF the way I've attempted on this thread, then you might feel more comfortable with the idea.

 

[Mike Smith]

Maybe I'm reading too much into what you are saying - but I think you are oversimplifying.

In our "watch/warning" system, I don't know of any EMA that sends out their fire trucks / police / etc for additional spotting in a watch.

Then you don't live in Kansas! They do exactly that.

But it's not okay if you tell me our current watch/warning system is PERFECT. It's great, I'll give you that. But I find it hard to declare that it cannot be improved upon in any way. WoF is one possible way of doing that, and the social science research helps tell that story. The current process was NOT developed with social science in mind.

Agree with all of the above except that we have cut the tornado death rate by better than 95% since the 1930's. That is pretty darn good. Because it has taken us 50 years to educate the public about the watch/warning system radical change is likely to do more harm than good. That is why I advocate improvements (higher PoD, lower FAR, more frequent updates) within the current system.

Mike

 

[rdale]

They do exactly that.

They have all fire trucks and police officers drop their activities to spot storms for 6 hours? How do volunteer agencies handle that cost burden?

Because it has taken us 50 years to educate the public about the watch/warning system radical change is likely to do more harm than good.

Absolutely. Nobody is advocating radical change for the sake of change.

That is why I advocate improvements (higher PoD, lower FAR, more frequent updates) within the current system.

I'm with you there. But do you think calling up your local office and saying "Please lower your FAR" will be the prompting? Or might new research into storms and forecast tools and radar integration (i.e. the basis of WoF) be a better avenue?

 

[Mike Johnston]

They have all fire trucks and police officers drop their activities to spot storms for 6 hours? How do volunteer agencies handle that cost burden?



Absolutely. Nobody is advocating radical change for the sake of change.



I'm with you there. But do you think calling up your local office and saying "Please lower your FAR" will be the prompting? Or might new research into storms and forecast tools and radar integration (i.e. the basis of WoF) be a better avenue?

I had a family member who was an officer of the Wichita police thirty years ago, and during a tornado watch he explained to me that officers were assigned to certain quadrants to "keep an eye out." It didn't involve every officer, every fire truck - but they did take heed during watches.

Now, I'm all for extending the basic research, trying to understand tornadogenesis better. And if some day, that results in a numerical prediction model that equals or exceeds the capabilities of our current detection (actual observation or remote sensing ie. doppler radar), then that would call for a revision to the warning system. But, as of now, if we really don't understand tornadogenesis - then we aren't anywhere close to developing a numerical model that would serve as a warning platform. How do we even know the inputs required of such a model? Would it require observations on the spatial order of the Oklahoma mesonet - or 10x the density, or 100x the density?

I think the better approach is to work it from the top down - ie. some effort to refine the watch areas and durations, putting a little bit more pressure on the forecasters.

No, you won't get very far just by requesting "lower the FAR, pretty please" but maybe a bit more transparency (ie. prominent publication of both the SPC and WFO POD and FAR rates) would inject more urgency and motivation where it belongs - the accuracy of the forecast.

 

[rdale]

I had a family member who was an officer of the Wichita police thirty years ago, and during a tornado watch he explained to me that officers were assigned to certain quadrants to "keep an eye out." It didn't involve every officer, every fire truck - but they did take heed during watches.

I understand that, but many counties I'm involved with send every unit available out to spot during a warning. This product would let them start that process prior to the storm hitting the western edge of the county.

But, as of now, if we really don't understand tornadogenesis - then we aren't anywhere close to developing a numerical model that would serve as a warning platform. How do we even know the inputs required of such a model? Would it require observations on the spatial order of the Oklahoma mesonet - or 10x the density, or 100x the density?

I don't want to steal any of Greg's potential thunder here - but my guess is we start down that road by working with what we have and uncovering those answers as we get into the details. Throwing our hands up and saying "We can't do WoF now, so let's wait 10 years and look at it again" won't help the research process move forward.

 

[Mike Johnston]

I understand that, but many counties I'm involved with send every unit available out to spot during a warning. This product would let them start that process prior to the storm hitting the western edge of the county.



I don't want to steal any of Greg's potential thunder here - but my guess is we start down that road by working with what we have and uncovering those answers as we get into the details. Throwing our hands up and saying "We can't do WoF now, so let's wait 10 years and look at it again" won't help the research process move forward.

I'm not suggesting "throwing our hands up" on anything, but advancement should be based on actual scientific understanding of the problem (tornadogenesis) - not just hedging our bets via probability forecasts. At best, that's a diversion, at worst, it's a cop out.

As for the comments about the hypothetical person out on a hiking journey, I guess that comes back to the basic mission of the NWS. I don't think it's mission implies being all things to all people. As an individual, if you undertake an endeavor out on the edge of normal domestic life it's up to you to take that extra step to research and keep abreast of the hazards involved. The NWS is supposed to serve the general public, not necessarily every outlier.

There are alot of mixed messages here from the advocates of a WoF/probability-based warning system. From the need to be supported by sociological studies vs. Joe Public input isn't needed. From the a priori assumption of a superior short-term numerical weather model vs. no citations at all that such a thing is even on the horizon.

 

[rdale]

advancement should be based on actual scientific understanding of the problem (tornadogenesis) - not just hedging our bets via probability forecasts

Again I'm not trying to be rude - but have you looked at their work? It's not a group of forecasters sitting around and pulling up radar imagery while yelling "20%" or "40%" or "I give that one a 60!" They really are advancing our scientific understanding of tornadogenesis with that research.

 

[Mike Johnston]

Again I'm not trying to be rude - but have you looked at their work? It's not a group of forecasters sitting around and pulling up radar imagery while yelling "20%" or "40%" or "I give that one a 60!" They really are advancing our scientific understanding of tornadogenesis with that research.

OK, well since this is a storm chasing forum, the next time we see a posting on a forecast or NOW thread I'll try to take note of any mention of any comments based upon a very short-term, storm-scale numerical weather prediction model. And if I miss it, please be sure to point it out to the others. If there is such a thing, I doubt it would be overlooked by this community.