Some thoughts, some responses, and a challenge re: The future of forecasting


Chuck Doswell

Last update: 23 May 1997

No disclaimer is needed here ... this is my personal account. In what follows, some of the basis for this work includes thoughts I have co-authored with others. However, the following thoughts are my own and should not be construed as representing the opinions of my colleagues ... who may or may not agree with any of these writings. This exposition is copyrighted [(c)1997 C. Doswell] and if you have any interest in reproducing it, please ask me first. For that purpose, or if you wish to offer feedback, my e-mail address is

I. Introduction

In a previous discussion, I went on at some length about what the future of forecasting might look like and what that future would mean for a person wishing to become (or remain in the position of) a forecaster as that future unfolds. I summarize the relevant points of that essay, for the purpose of setting the stage for what is to follow:

1. As the accuracy of numerical model and statistically-derived guidance products improves, there is a diminishing need for forecasters of the "old school," whose substantive knowledge of meteorology is limited by their education and training. As the objective forecasting systems become more accurate, forecasters need to base their forecast process firmly within the science of meteorology, a task most forecasters are neither educated nor trained to do.

2. Forecasters need to be able to demonstrate their ability to make consistent, useful improvement over the guidance products. This gets more challenging every year. If forecasters can't demonstrate their ability to add quality to the guidance products, their role is almost certain to diminish. Successful forcasters have to work hard to keep their edge over the guidance.

3. Forecast verification is a critical component of assessing forecast quality. Just making operational forecast deadlines and turning out a product every day for 25 years is no measure of the quality of the products. Rigorous verification, however necessary for assessing forecast quality and improving upon it, is not sufficient evidence of good forecasting. The real test of forecast quality is the protests (or the absence thereof) by the users when those forecasts are made by some forecaster (or automated forecasting system) other than the forecaster in question.

Further, Harold Brooks, Mike Fritsch, and I wrote a conference paper (Brooks et al. 1996) describing one vision of weather forecasting's future. Again, I summarize the salient points:

1. Technological change is inevitably going to result in the automation of many traditional forecaster duties, especially the routine products.

2. As a result of this costly technological change, and as a direct consequence of budget pressures as well, either many offices will be closed down or the number of duty forecasters is going to continue to decline radically.

3. Our vision is one with fewer offices, but each office should have a radically altered mission compared to the current situation, focusing on hazardous weather events rather than routine forecasts. With fewer offices, the emphasis can be on forecaster skills, including a rigorous forecaster certification, plus a substantive continuing education/training program for every forecaster. Also, forecast quality and value should be given much more emphasis than at present. These new offices should include a substantial investment of resources into user interaction, such that information can flow from NWS to the users, and vice-versa.

II. War and Rumors of War

Since the conference paper was presented, there have been various rumors of discontent over its content. With few exceptions, it appears that those who were offended or otherwise exercised by what Harold, Mike and I said have been loath to discuss the issues with us. It is not clear to me why these negative reactions have not been shared with us, although I have rumors to the effect that I (or we) intimidate them, or that they feel it would be a waste of time trying to have the debate. Let me say that:

a) I welcome your comments, negative or no. If you have some fear of me or what I might say, please try to overcome that fear so that I might gain insight into what you feel and know the reasons why. What can I do to hurt you? What harm can I do? If I make an intemperate statement (which has been known to happen!), why should that cause you to crumble or go into a shell?

b) If you feel it is a waste of time, please understand that I can be convinced. I will change my opinion if you can show me in some way that I am wrong, or misinformed, or suffering under some misconception. I will not roll over and submit just because you have a different opinion, but I am not so prideful that I will not admit I was wrong. I am a scientist and I have to be skeptical, even of my own viewpoints.

Please do not insult me with the inference that because I have not sat in your operational forecast seat, I cannot possibly understand why you believe what you do. I know as well as anyone that in order to understand the nature of any job deeply, one actually has to do that job. No argument there. But I have sat in an operational forecast seat (if not in yours) where lives were at stake and I lived with hard deadlines, I have issued forecasts in field experiments with many resources hanging in the balance, I have interacted with users (of weather information) of all sorts and with many forecasters who agree with me (and those who don't, also!), I have learned a thing or two about the weather, and I have spent most of my career thinking about weather forecasting in one way or another. I will not accept the idea that there is some mystical aspect of a forecaster's job that cannot be appreciated unless I sit in a particular seat. If you can't articulate that aspect of your forecasting job which makes your viewpoint different from mine in terms I can see and understand, then I am inclined to think it is not so firmly based as you might think. Your confidence in your beliefs is no guarantee of their correctness.

My challenge to you will be to back up those beliefs with their basis in facts, figures, and unambiguous observations, as well as good, logical arguments. Much of what we said in our conference paper was associated with our studies in forecast verification. If our conclusions do not match yours, then you should consider that our results still might well be representative and yours may not be so typical.

III. A Gauntlet at Your Feet!

I challenge anyone who is bothered by our vision of what the future might look like to articulate and present their vision. I have not seen a deluge of contrary position papers. If you have a point of view and fear to present it because of what NOAA and the NWS managers might do to you, I consider that to be a cop-out. You challenge our ideas but choose not to risk presenting your ideas to our scrutiny (much less anyone else's scrutiny) and, instead, hide behind the defense of a fear of bureaucratic retribution. This is, of course, a rather safe defensive strategy but it does not advance anyone's understanding ... not mine, and not yours, not anyone's.

Are you willing to put your ideas on paper and show them to me, if not to anyone else? Do you have a vision or are you just able to criticize? Do you want to challenge us and have a useful discussion of the issues, or is it enough for you to talk negatively about our ideas behind our backs ? How many offices do you think there should be? What tasks do you believe the staff should be doing? Are you arguing for thestatus quo or do you believe operational forecasting should change in some way that is significantly different from what we presented? How do you see the impact of the virtually inevitable budget decline being optimally mitigated on behalf of the users of public weather forecasts? Do you believe that a miracle will happen and vast resources will be pouring into the budget of the NWS soon? Do you see us primarily as a threat to your job security (or to your personal vision of the future), or do you have a substantive reason to disagree with us? Come on now, let's get this discussion going!

IV. Some Negative Feedback ... Mostly Second-Hand

1. A Contrary View

Some, it seems, have a vastly different view of what state the National Weather Service is in at the moment. Their arguments look something like this:

What Brooks, Fritsch and Doswell have said is based on old information. Things are changing in the NWS. There is lots of operational research going on, through COMET partnerships, university interactions, etc. The training programs of the NWS are changing in response to the needs of the forecasters, with COMET, the OSF, and NWSTC providing all sorts of classroom and distance-learning, computer-based modules. Furthermore, the mix of tasks in operational offices is changing ... much of the routine, synoptic-scale forecast may done by centralized objective forecasting systems, with local forecasts focusing on non-routine, important weather "events."

I see no choice but to disagree. In the current plan for the modernized NWS, there are something like 120 WFOs. The key jobs are the Science and Operations Officer (SOO) and the Warning Coordination Meteorologist (WCM). Thus, the NWS has to find something like 240 folks to fill these key positions. Unfortunately, the necessary skills for these two jobs are such that if anyone actually has all of the needed skills, they will have to be supermen/women! The SOOs have to be forecasters (they will be working shifts), researchers (they will be expected not only to carry out research, but to guide the research of others on-station), trainers (they will be the primary leaders of the huge amounts of on-station training that will be substituted for formal classroom-type training), managers (they will be considered part of the management team), effective at interpersonal relations (they will have a lot of interaction with the whole staff on a regular basis), ... and on and on. The WCMs also will be forecasters, trainers (of spotters, etc.), researchers, public-relations experts, effective at interpersonal relations, good public speakers, managers ... and on and on.

I submit that there is no more than a handful of people who even come close to fulfilling all these requirements. What we actually have in these positions is a group of people who are seriously deficient in one or more of these skills. Please do not take this as an implied criticism of these people!! Many (perhaps most) of them are well-motivated, hard-working, and have many valuable skills ... it's just that the system is asking way too much from these positions for most of us ordinary humans to be able to carry out all these diverse tasks with the skill needed. For every absent but necessary skill, you have that component of the SOO/WCM responsibilities being done poorly, if at all. And every part of the SOO/WCM job is a critical component in the successful implementation of the modernization.

An unfortunate aspect is that these are well-paid positions that can be seen simply as springboards to higher positions. Some SOOs and WCMs are no more than careerists, whose ambition is to rise to power positions using the SOO/WCM slot as a temporary waystation on their way to some fat-cat bureaucrat slot, safely out of shiftwork and away from the challenges of forecasting. Furthermore, some of these positions are going to be filled by persons who lack more than one or two of the needed skills ... in 240 positions, it is inevitable that some incompetent people will be selected. For every SOO/WCM position filled with a careerist (an incompetent by definition ) or an incompetent, you have an office that is severely handicapped.

The joint applied research projects going on in conjunction with university professors or with COMET, some papers appearing in conferences, a precious few formal publications ... these do not mean, in my opinion, that we are seeing a renaissance of research-operations interaction. I cannot accept these as true signs that a new age is dawning ... it is more a matter of research projects being a factor in performance ratings than it is a signal of a vast groundswell of solid on-station research. A few of the really good SOOs and WCMs are accounting for a big fraction of the really substantive work. I also think that a goodly fraction of these papers and projects are substandard efforts that will end up having little or no real impact.[1] If you believe that the shoe doesn't fit you, then by no means am I asking you to wear it, but I have not seen an outpouring of really solid accomplishments from these projects.[2]

The continuing lack of a meaningful entry-level forecaster training program, where many applicants will fail, and the absence of a challenging forecaster certification[3] mean that what we have for training in the NWS is simply a pitiful effort to provide a smokescreen for the very real lack of training! Present classroom training is either more of what folks got at their university (which most all forecasters agree is not very useful in operations), or is pretty simple "knobology" exercises that do not address the big deficiencies in applying the science of meteorology to forecasting.

The idea that "on-station" training in dribs and drabs with a few interactive videodisks can substitute for real training is at best simply a self-serving delusion. The NWS has no intention of providing anything of substance for its people, when it is struggling to justify the huge expenditures it has made for its technology! Billions (literally) can be spent for technology, but it seems that the NWS can't create anything but wasteful, sub-critical training programs. NWS training is a disorganized collection of competing, duplicative programs with no coherent direction, churning out little more than glitzy packaging. These certainly consume millions of dollars, so someone can say "We're spending millions on training!" but there is no evidence that any of the meteorological training done by the NWS makes any real difference in the forecasting. The lack of any useful measure (other than "How do you feel about the training?" surveys done occasionally after the classroom training) of the impact of the training is a dead giveaway that the process is one with no more than superficial effect. I assert that no one knows what influence on weather forecasts has been generated from NWS training ... and I have participated in some of that training.[4] Meanwhile, I have no reason to disbelieve that all the current NWS "training" fails utterly in addressing the major training (and educational!) gaps in the forecasting staff.

I also believe that the myth of the forecaster as a "mesoscale expert" in the future is no more than a dream, and a dangerous one at that. I do not believe that anyone is much of a mesoscale "expert" although some might pretend to be ... the complexity and dizzying nonlinearity of mesoscale meteorology make it unlikely that anyone (forecaster or researcher) understands it very well. Separating forecasters from the synoptic scale forecasting process is going to make them poor mesoscale forecasters, in my opinion. I believe that if a forecaster is not involved in synoptic-scale processes, he or she is not likely to be a very good mesoscale forecaster, either. Of course, our paper does advocate turning a good deal of the routine forecasting over to objective guidance most of the time, but human forecasters need to be monitoring those products regularly, and occasionally nudging them back toward reality when they become obviously wrong. They can do this if they are deeply involved in synoptic-mesoscale meteorological analysis and diagnosis, spending much of their time looking at data (observations) instead of fruitlessly trying to decide which part of the conflicting model guidance to believe. If forecasters focus on hazardous weather, the "quality control" process for the centralized guidance is a natural outgrowth of the process of looking at observations and guidance.

Moreover, our paper developed its ideas in the context of what is being done currently (and what has been done in the past). I hope that things are changing, and changing for the better, but we should not be faulted for not commenting on how current NWS management hopes to change the operational forecasting system. I'll be happy to reserve judgement on that until (and if) those changes are implemented.

The NWS may have changed the leopard's spots, but it still is a leopard.

2. Our Vision as a Threat

To others, the Brooks, Fritsch, and Doswell paper represents a threat. That is, they believe that NOAA and the NWS are likely to use our paper as justification to bring harm to the NWS. To that, I can say that NOAA and the NWS have shown very limited response to me and my writings over the years. In fact, their silence is deafening. I find it quite unlikely that they are going to implement some idea of mine simply because I say so (verbally, in an essay, or in a formal publication).

It has been suggested that if they chose to carry out only part of the plan that Harold, Mike, and I articulated in that conference paper (e.g., reducing the number of NWS offices to something on the order of 10)[5] without doing the whole program (e.g., failing to develop a meaningful training program), you can rest assured that I will be vigorous and unrelenting in my criticism of such a development.

If the idea is that NOAA and the NWS have some nefarious agenda and they would seize on this our paper as a smokescreen for carrying it out, I submit that it is unreasonable to condemn us for that. Should I censor myself because of what the NOAA and the NWS might do? Do I keep my thoughts to myself to avoid potential misuse? Nonsense! How can I possibly anticipate all the potential misuses of our proposals? Why should I do so? It seems to me that fear (of unnamed actions) is a major concern to some of our detractors. If this is the case, do they really believe that we drive out fear by self-censorship and silence? Is not having the debate going to prevent the political and economic forces for change in the operational forecasting system from carrying out draconian budget cuts?

V. The Future?

What I believe is at the heart of many of the criticisms of our vision in the Brooks, Fritsch, and Doswell paper is a reluctance to accept change. Given the national budget situation, crying about the reductions in the NWS and calling for "holding the line" is not a relevant aspect of the debate we are trying to have. Do you really expect to see an operational weather forecasting system (including the NWS, private forecasting services, media forecasters, etc.) like that of today's go on basically unchanged indefinitely? The hammer is falling even as I write this, and I cannot imagine a sudden improvement in the NWS budget situation at any time in the near future.

Change is both a threat and an opportunity. The "modernization" is a reality, for the time being, like it or not. When things are in turmoil, we can take the turmoil as a chance to carry out important, fundamental changes instead of trying to create a program that is a modified version of the old, with perhaps enhanced automation allowing for reductions in staffing. Oh yes, those reductions are almost certain to happen, like it or not. If we choose to keep lots of offices open, do you really think we are going to be able to staff them properly? Dream on! The biggest cost driver in the NWS is almost certainly the payroll, not the hardware and expendables. The only realistic choices in the future are limited to two, as I see it:

1) Order 100 offices, with skeletal staffing, or

2) Order 10 offices, with enough high-quality, well-trained staff to do the job.

What Harold, Mike, and I have done is try to suggest one way to deal "proactively" with the virtually inevitable changes. Rather than sticking our heads in the sand and then being terribly upset when the changes hit, why not have the debate now? Why not develop a strategy that allows us to design a trimmed-down and more efficient forecasting system, rather than having a structure forced down our throats later?

If a real commitment to user outreach is to be implemented (e.g., support for hazardous weather preparedness programs in local communities, and/or a substantive program for teaching users how to use the weather information they receive), there may have to be more locally-based folks to carry out this process successfully than can be contained in "order 10" offices. I am willing to accept that possibility but I am not willing to take it on faith. We need to experiment and see what works and what doesn't work, rather than taking someone's arguments for the status quo on faith. We also need to design those experiments carefully, to prevent them from being manipulated for political ends. Any argument for the status quo worries me, owing to the inevitability of continuing budget pressure.

Note added: 23 May 1997. I have recently become aware that high-placed NWS officials have made veiled threats to the effect that if Harold and I continue to write papers like Brooks et al. (1996), then there may be reprisals against NSSL and/or the SPC! This amazing revelation suggests the extent to which some power-drunk people in the NWS headquarters hierachy are willing to go to suppress dissenting viewpoints (see my recent essay about NOAA/NWS management). In this case, the individual making the threats used to be a scientist for whom I had considerable respect. For a scientist to reach the point where he would consider using threats of personal retribution for expressing a contrary view ... a sad day, indeed. It is astonishing that our organizations might be threatened by NWS officials for things that we have said. This clearly is intended to put pressure on the local managers here in Norman to shut me (and others) up. I hope that this is a baseless rumor, but I am certainly not going to pretend I am not concerned about it. If it becomes impossible in this organization to express contrary viewpoints as part of my job, then I fear for our science and for the NWS itself. We are not trying to sabotage the NWS! We are trying to force the NWS to look long and hard at the directions it is taking and the methods it is using. We want a dialog, not top-down imposition of a pre-ordained structure. Is that reason to punish our organizations?


Brooks, H.E., J.M. Fritsch, and C.A. Doswell III, 1996: The future of weather forecasting: The eras of revolution and reconstruction. Preprints, 15th Conf. Wea. Analysis and Forecasting, Amer. Meteor. Soc., 523-526.