2014 NWS trends – Commentary

by Chuck Doswell

Page created: 27 October 2014 last updated: whenever.

This page contains my opinions and since it is on my home Website, no disclaimer is needed. If you want to discuss the material contained herein, e-mail me at cdoswell@earthlink.net.

As my friends and colleagues in the operational weather forecasting world move toward an inevitably uncertain future, my professional career received its highest possible validation:  the National Weather Association’s Special Lifetime Achievement Award.  The award was made in 2013, but I was unable to attend the awards ceremony that year, so it was recently given to me at the 2014 NWA Annual Meeting in Salt Lake City.  During my time at the meeting, I saw that some trends evidently are underway in the National Weather Service (NWS) that will have major impacts on the services provided by the NWS.  And if these trends continue, it will have major impacts on the staff of the NWS - the very people I've dedicated my career to support. 

Although not elaborated upon in this essay, there is also the ongoing juggernaut leading toward full automation of the existing weather forecasts and services.  I have numerous essays on that and related topics here, and I have no wish to repeat all of that. Nevertheless, that ongoing process of automating the forecasts has a bearing on the present discussion.  Here, I’m simply noting that this is another issue of concern, but one I (and others) have written about extensively in the past.

Impact-based Warnings (IBW)

In 2012, the NWS Central Region initiated an "experiment" that provided graded threat levels to their warnings based on the potential impacts of the warned-for events.  At the time, I was critical of this "experiment" - describing it as ill-timed, ill-conceived, and poorly executed:


The timing of this "experiment" is poor, in that the whole notion of impact-based warnings (IBW) had not (and still has not) been explored in any systematic, comprehensive way prior to the execution of the initial experiment.  In no way was the notion of IBW based on large amounts of concrete information as to its efficacy, and virtually no consideration was given to possible unintended consequences and negative side effects.   One should not provide revised critical services to the public without first having done the work to test the ideas in a purely experimental setting (i.e., not going directly to the public users of these products).  Prior to the introduction of new ways of providing those services, moreover, there should have been an extensive public information campaign to inform the public about what changes were coming and how to use the new products.  This was simply someone's idea of how to fix a problem that had not been examined thoroughly, and so was not clearly enough understood even to begin proposed new products.  This experiment was egregriously premature.  Hence, ill-timed.


Given the absence of a reasonably thorough understanding of the problem(s) associated with the existing warning products, this plan likely was created by someone in NWS management, somewhere.  Its potential impacts on the forecasters tasked with executing this experiment were not given any sort of thorough considerations.  Surely, the proper way to assess that would have been to experiment with these new products in an operational setting but without transmitting them to the users.  Training in their use should have been provided, as well, but I doubt if anyone in the system was really qualified to train forecasters in how to do much of anything that forecasters are asked to do.  It was a half-baked idea to start with.

Consider the forecast experiment design (taken from their own website):

1.  Optimize the convective warning system within the existing structure

Optimize with respect to what?  What does "optimize" mean in this context - just do the best you can?  Is it realistic to expect truly to optimize something done within the context of an experiment?  It seems to me that if anything "optimal" could be designed, that would come after the experiment, rather than as it was ongoing.

2.  Motivate proper response to warnings by distinguishing  situational urgency

Just what's considered "proper response to warnings"?  The responses by people to being informed of potentially hazardous weather can vary considerable within the non-monolithic population known as "the public".  How can warnings manage to cover that vast spectrum of possible responses?  Do we really want people to respond like automatons to our warnings, and do immediately what we tell them to do?  Are forecasters aware of and knowledgeable about enough aspects of a situation to assess the urgency?  It's reasonable to expect forecasters to know about the meteorological situation, but that"s not what seems to be the goal here.

3.  Realign the warning message in terms of societal impacts

Just what does "realign" mean?  Reword?  Revise?  How are forecasters supposed to "realign" the warning messages?  On what social science finds can such a realignment take place?  It seems to me that a lot of research needs to be done before we start playing around with the warning messages.  A primary directive to consider in any proposed changes:  First, do no harm!

4.  Communicate recommended actions/precautions more concisely

As I'll elaborate upon below, the time for calls-to-action is notduring the few minutes prior to the arrival of hazardous weather.  This should take place months before, and if so, there's no particular need to be concise. The message of informing people about proper preparation for hazardous weather shouldn't be conveyed at the last minute by means of some simple catch phrase or two.  And the "proper" actions for different people will be different.  There's no set of "one size fits all" recommendations we can make!  Many segments of the public necessary will be left out of any CTAs we cram onto the tail end of a warning.  Whose needs take priority?

The mandated graded warnings in this experiment were simply a rehash of some things done unofficially by forecasters in the past (e.g., the "tornado emergency").  Obviously there was no serious research put into this experimental design.

5.  Evaluate ability to distinguish between low impact and high impact events

This is so absurd, it boggles the mind to consider it!  Surely the evaluation of forecasters' ability to do this should be done long before we expect them to perform this task at a high level in the heat of a hazardous weather situation!

Poorly executed

All of the preceding makes is pretty evident that the original experiment was flawed in several ways.  I’m sure that forecasters did (and are doing) the best they could under the circumstances they’d been given.  This cancerous experiment now (2014) has metastasized to other offices and other NWS regions, so the juggernaut of IBW rolls forward, without regard to the issues I've raised.

Overall, I have no problem with warnings that have graded threat levels - quite the opposite, in fact.  But the graded threat I believe needs to be incorporated is not that of the impact on the public but rather the forecaster’s perception of the threat of hazardous weather.  That is, the probability of hazardous weather reaching some threshold, however that might be communicated effectively.  I’ve talked about that in essays here and here

A forecaster should be able to describe his/her confidence that a particular event will occur, given that such an occurrence probability has met some threshold where the threat is now at the point where it needs to be mentioned.  Deciding on how best to express that confidence remains to be determined, and a lot of work will be needed to reach some recommendations based on that research.  But for a forecaster to know all the external factors that affect people within the path of a storm is impossible.  Forecasters should be allowed to forecast the weather and not be required to give any consideration to impacts.  Those impacts are best delineated by those decision-makers whose decisions could be influenced by the weather forecast but which are influenced by many factors besides the weather.  Given that forecasters can't possibly know the factors influencing everyone's decisions, they have no legitimate basis for making impact estimates.

If the NWS accepts some responsibility for what happens in the world outside of the NWS after a forecast/warning is issued (and I believe they should), there surely is a need for knowledgeable experts to help train users put NWS-provided weather information to use to serve their needs.  But forecasters are not educated and receive no training in how do fulfill such tasks.  Forecasting the weather is quite challenging - challenging enough to occupy a forecaster’s full attention.  If we convolve impacts with our forecasts, it can only degrade forecasting performance.  I recall some discussions in SELS (the "old days"!) about issuing tornado watches in rural Montana, because there were too few people to be affected.  Should forecasters forecast the weather or the impacts of that weather?  I believe they need to forecast the weather and nothing else!

Although I've commented extensively on "call-to-action" statements here, I’d like to bring up a topic that arose at the NWA Annual Meeting:  the notion of "nuisance" floods as opposed to major flooding events.  Someone commented "It's not a nuisance flood if someone dies!"  I beg to differ with that statement!  There's a clear distinction between a truly devastating flash flood (e.g., the Big Thompson flood event of 31 July 1976) and some case involving minor urban flooding that just happens to include a rare fatality.  Over the years, I've found that it's impossible to prevent people from making bad decisions and getting themselves killed even in what are well-forecast but meteorologically relatively minor events.  Would it truly be possible to anticipate the impact of one person's stupid mistake that resulted in their death in an otherwise marginal event?  I think not, and it's foolish even to consider that it might be possible.  Otherwise, we'd have to issue warnings for virtually every storm, no matter how meteorologically weak it might be.

I hope the NWS does indeed move toward including impacts in the process of serving public needs.  However, I don't think the forecasters should be expected to take that on as an additional aspect of their job.  People qualified to get involved with impacts would need a very different skill set than what meteorologist/forecasters are given to qualify for their position.  Therefore, if the NWS goes after impacts, it will have to change the character of their staffing.  If they're not going to hire a bunch of new people to address impacts, then they're going to have to retrain and reassign existing staff to cover that responsibility.  This likely would require employing fewer forecasters, and a probable shift of most of the the forecast/warning products to an automated system.

Weather Ready Nation (WRN)

It seems there's another juggernaut rolling in the NWS:  the drive toward a so-called "Weather/Storm Ready Nation".  I’ve already commented on that topic at some length here and here

I continue to describe the existing "certification" of storm readiness to be little more than lip service to being truly storm ready.  The basic concept of minimizing the impact of hazardous weather events is one that I certainly believe is worthwhile, but I don't believe the NWS is willing to commit to making that become actual reality.  This means a lot of public education, a lot of research into existing structures, and an assessment of the cost vs. benefits of retrofitting existing structures to become tornado resistant.  I endorse any such efforts in principle, but the practical reality is that the task is large and resources are limited.

If we consider seeking a "Weather Ready Nation" to be simply an overarching philosophy rather than some specific set of tasks, then this entails developing a vast array of new efforts for the wide variety of hazards the weather might bring our way:  storms of all sorts (including very rare events such as a tornado hitting a large entertainment venue), heat waves, cold waves, seiches, avalanches, etc.  This doesn't mean I disagree with any specific efforts to get the NWS involved - far from it.  I think most any effort is better than no effort at all (but first, do no harm!).  However, the investment of resources required to do the job of preparing everyone for all weather-related hazards of all sorts seems far in excess of what the NWS could ever commit.  Yes, one doesn't expect this to happen all at once, but the task seems gargantuan to me, and I'm not convinced the NWS recognizes how little their existing resources allow them to do compared to the need.  The goal of reaching the status of a truly weather-ready nation is likely many decades away, if it ever can be attained.

As an example of a major challenge:  it seems obvious to me that the existing home construction standards in the US are far less stringent than they need to be to become "weather ready" in more than name only.  I think the same standards that exist along the coastal zones to resist the impact of hurricanes should be applied also to the tornado-prone regions of the US (everything east of the Continental Divide).  This is a complex topic and people have to consider their choices carefully.  Unfortunately, the structural integrity of your home in a tornado (for instance) depends in part on the construction quality of your neighbors' homes!  Debris from poorly-constructed nearby structures can hit your home and cause damage well beyond what the wind would have done by itself.  The flying debris from flimsy homes magnifies the damage along a tornado track.  Much research needs to be done to deal with such issues in practical terms, and people need to understand what their real risks are in order to make decisions that are good for them as individuals and simultaneously good for their communities.  Needless to say, most builders are strongly (and effectively) opposed to any enhanced construction standards.

Another thorny issue is development in flood-prone areas, including along the coasts.  As it stands, the nation is paying a huge price for people to live in the center of a bullseye, with the inevitable result virtually every year that someone is devastated by flood or oceanic surges due to storms.  In a truly weather-ready nation, no such development would be permitted and vulnerable locations would be abandoned.  Good luck with getting that implemented!

The whole problem of making us a WRN is a lot more challenging than the NWS realizes or is ready to commit to make happen - in my opinion, of course.  I find the headlong rush to WRN goals to be na´ve and potentially problematic.  We are creating an illusion of security that isn’t commensurate with the true security of communities.  That could come back to haunt us - no, it will come back to haunt us.