What funding decisions will reduce tornado fatalities?


Chuck Doswell

Posted: 01 March 2003 Updated: whenever. As usual, this is just my opinion. Those wishing to comment or discuss any aspect of this should contact me at cdoswell@earthlink.net.

1. Introduction

Although this essay concerns a very specific topic, it actually applies to a much wider range of problems. It is nominally concerned with the processes that I see ongoing in the continuing decisions our nation has to make regarding the tornado problem. I've written elsewhere about the societal impacts of tornadoes on the United States (and elsewhere).

In addition to something in the range of 50-100 fatalities per year, tornadoes cause about $1 billion in damage annually in the United States. For the individual affected by tornadoes, they are devastating, but from another perspective, we kill 40, 000 people annually in traffic accidents, and something like 10,000 people die annually in the U.S. from food poisoning! From a pure "numbers" standpoint, the risk to human life from tornadoes is trivial compared to many other hazards we face. It isn't at all unreasonable to set some limit on what we as a nation are willing to spend to reduce tornado casualties, Of course, the so-called "sanctity of human life" advocates would say something like "There's no way to set a value on human life!" My answer to that is simple: Nonsense!! We put economic value on human life all the time. Would it be worth it to us to bankrupt the national economy to save one human life? Absolutely not! If there is a limit to the amount we might want to spend to save lives, the insurance industry probably is not a bad place to look for guidance. They have settled on a dollar equivalent to a human life ... it's on the order of $8 million per life. I have no clue how they arrived at such a figure, so don't ask me. I'm simply repeating what I have heard. Nevertheless, it's a rough estimate from which I can begin. So if we spent up to $400-800 million annually on saving lives from tornadoes, such a figure would be about as much as we could ever justify, at least according to the standard set by the insurance industry.

The budget for the entire National Weather Service (NWS) is presently (as of 2002) about $740 million (of which only $670 is for operations, research, and facilities [ORF] - presumably, the rest is administrative cost), so the NWS budget seems to be close to this hypothetical spending limit of $800 million. But of course, most of that allocation is for other things besides tornado forecasts and warnings. I'll be trying to show a rough estimate of what we are actually spending on tornadoes (forecasts, warnings, and research) alone in the next section. I'll show that I believe it's reasonable to ask for an increase in the national budget for tornado-related research and development in support of forecasting and warning operations, as well as some other things. As I'll try to show below, a total of $300 million annually, which would represent a fantastic increase in our budget (a veritable bonanza, in fact), is within the crude guidelines I've given for what might be considered reasonable.

So I can argue that a potential bonanza remains to be exploited, that would still be within some bounds for the U.S. economy to absorb in seeking to mitigate tornado hazards. Is it realistic to expect such a windfall? Given the present situation with NOAA's research and forecasting managers, probably not. Too bad, though. Nevertheless, I want to forge ahead with this ...


2. So what is being spent now (figures for FY 2002)?

Quite separate from the $670 million or so being spent on NWS ORF, there is the budget for all the NOAA research laboratories. The base Federal appropriation for the NOAA Laboratories is slightly less than $120 million for ORF. Interestingly, in 2002 $16 million was allocated for the "Norman Consolidation Project" -- see here for some not yet well-known but very disturbing aspects of that project. Detailed information about NOAA's budget can be found here.


The base Federal appropriation for the National Severe Storms Laboratory (NSSL) is about $7.5 million. [NOAA laboratories can seek additional resources from "soft" money and NSSL presently has a budget that is more than twice their base Federal appropriation - in other words, more than half their budget comes from non-recurring "base" funds.] I'm guessing that about 1/3 of this total ($2.5 million) is associated, directly or indirectly, with tornadoes. [$2.5 million]

b. SPC

The Storm Prediction Center (SPC) has an annual budget within that of the National Weather Service of about $3 million -- only part of that is spent on tornadoes. Let me assign half of the SPC budget to tornadoes alone. [$1.5 million]

c. Local WFOs

For a local Weather Forecast Office (WFO), in round numbers, the annual budget is of order $2.75 million (most of which is salaries). For 120 offices, that amounts to $330 million - or about half of the operations budget of the the NWS. Some fraction of the budget for operating the local NWS offices is tied directly and indirectly to tornadoes, but there is no simple way to know just what that fraction is. Permit me to guess that of the local office operating budgets, about 5% is spent in some way specifically on tornadoes (this is just a wild guess, of course, but I think it might be in the right "ballpark" as an estimate. This would vary from office to office. The NWS office in Norman, OK would spend a lot more on tornado-relatied operations than, say, Fairbanks, AK! If I am anywhere near correct, this is the largest single expenditure annually on tornadoes, and most of it is being spent on salaries for the people who issue tornado warnings, do spotter training, collect tornado data, etc. [$16.5 million]

d. Modernization

But wait! This isn't everything. Recently, the NWS completed a "modernization" exercise, that included buying and installing the new WSR-88D weather radars. The modernization cost several $billion, spread over a period of several years. A major factor used in "selling" this program was the argument that improved tornado warnings would result from the implementation of the new WSR-88D radars. To show my readers exactly what happened, here are the annual fatality figures from tornadoes in the U.S. (Fig. 1). I will let you draw your own conclusions about the implementation of the WSR-88Ds on fatalities.

Figure 1. A figure showing the annual number of tornado fatalities from 1950-2001. The WSR88D network was essentially in place by, say, 1995, but began to be implemented after 1988.

Modernization is not an annual cost, of course; it was a "one-time" series of annual appropriations designed to replace outmoded equipment. In this case, since so much of this cost was associated with the new radars, some fraction of that total (say, 20 percent) could be tied directly or indirectly to tornadoes. It's inevitable that a renewed need for "modernization" will arise in the future. I'll simply assign a generous annual figure for this "modernization" activity tied specifically to tornadoes, under the assumption that some fraction of this will be used for expensive new radars, to come. If this annual figure were "banked" every year into a "Modernization fund" (which is probably illegal under some stupid, shortsighted Federal law!), there would be enough after 20 years to buy some pretty neat radars, etc. [$100 million]


But wait! There's still more! The legions of weather satellites put up on behalf of weather research and forecasting have also been justified, at least in part, by their purported value for weather forecasting (including tornadoes, of course). The part of NOAA that is associated with weather satellites is the National Environmental Satellite, Data and Information Service (NESDIS). They have an annual "systems acquisition" budget of about $560 million. I assume that this includes the cost of their geostationary and polar orbiting satellites. The NESDIS ORF base funding appropriation is about $140 million. Of this total of $700 million, perhaps as much as 5 % is associated with tornadoes in some way. [$3.5 million]

Estimated total annual NOAA expenditures for tornadoes without including any annual contributions to a "fund" for modernization: [$24 million]

There might be some modest amounts spent in other Federal agencies, like NASA or the Department of Defense, but they probably do not represent more than, say, an additional $1 million annually.

This total is quite far short of the $400-800 million (let's call it $600 million, for short) we could reasonably ask to spend and still "break even" economically, at least by my crude reckoning in terms of the insurance industry's value assigned to human life.


3. How should we use any new allocations for tornado research and operations?

A question naturally arises if any plausible improvement in tornado forecasts and warnings could reduce that roughly $1 billion figure in tornado damage. I think not. For tornado forecasts and warnings to offer much potential for reducing the damage, they would need to be pretty accurate at ranges of 24 h or longer, if people were to take substantial measures to limit property damage (as they now do with hurricanes, for instance). We could do things to limit property damage in response to the climatological threat of tornadoes (building more tornado-resistant homes), but this would not have any relevance to the value of tornado forecasts and warnings unless those forecast products could attain an unrealistically high level of accuracy at ranges of 24 h or more. Without accurate advance warnings on the order of a day or more, it's hard to imagine what we could do to reduce damage. With hurricanes, it's possible to board up windows and so on, because of the long lead time for hurricane warnings. This is not going to be possible for tornadoes for many, many years, if ever.

O.K. -- to "justify" any additional expense to the taxpayers, it must be in terms of human lives, not property damage.

But wait. In order to have that $600 million expenditure be justified, strictly speaking, we would have to reduce the casualty level to ... zero! And it would have to stay there indefinitely. That might not be a realistic goal, so perhaps we could hope realistically to cut the average annual fatality rate in half, and so could only "justify" (in the terms I've been using) an annual expenditure of half the $600 million, or $300 million. This would double the annual expenditures we are now making, even if we account for new radars every 20 years!! If we consider only the annual expenditures and ignore the cost of annual contributions to a "modernization fund," the $300 million figure is waaaaaaayyyy more than what we are spending now. A veritable bonanza, indeed!

If by some miracle (and that's probably what it would take!) we were to obtain this funding windfall, how should we choose to spend those resources?

Granting the unrealistic expectation that the annual budget for tornado research and forecasting would increase to this new limit of $300 million, what might be our best strategy for cutting the annual U.S. tornado fatality rate in half?


4. The situation now regarding tornado fatalities

What is now known about tornado fatalities? Although significant tornadoes (F2 and higher on the Fujita scale) account for around 1/4 of the annual total of reported tornadoes, they account for 95 % of the fatalities. If we consider only violent tornadoes (F4 and higher), these are about 2 % of the total each year, but they account for 67 % of the fatalities.

Under the existing "state of the science" of tornado forecasting and warning operations, it is precisely these significant events that are most likely to be forecast and warned for most effectively. In cases like 03 May 1999, or any of a large number of major tornado outbreaks in recent history, the existing system of forecasting and warning procedures has generally done quite well. In parts of the metropolitan area of Oklahoma City, OK on 03 May 1999, some warnings were issued with a "lead time" (the time elapsed between the warning issuance and the arrival of the tornado) of 45 min or more. Certainly most of the warnings had 30 min lead times. Not every significant tornado event can achieve such outstanding lead times for warnings, but the majority of significant tornado events produce lead times that double the average for all tornadoes. A few years back, a significant tornado (I believe it wound up being rated F3) struck with virtually no warning in Salt Lake City, UT, resulting in only a single fatality. Even that one death is regrettable, and especially so in light of the lack of warning, but the low toll underscores the dominance of violent, long-track events in the fatality counts. When these violent, long-track events strike populated areas, even with excellent warnings, fatalities are apparently inevitable.

It's the relatively brief, and mostly weak tornadoes that are so challenging to warn for with accuracy and long lead times. No tornado that hits your house seems "insignificant" to you, but in most cases, the tornadoes most likely to cause fatalities are generally those most likely to have been warned for by the NWS. We already have the lowest false alarm ratios, the highest probabilities of detection, and the longest lead times for violent storms! If we are to improve our tornado warning statistics, the majority of the improvement would necessarily have to be in association with the weaker tornadoes and these just don't kill very many people year in and year out. From the perspective of establishing national priorities, it seems we're not going to save many lives by spending resources to improve upon something we're already doing rather well with, on the whole.

Within the past several decades, there has been some attempt to keep track of the circumstances under which people are killed in tornadoes. In particular, their location: in motor vehicles, in mobile homes, in site-built homes, in multifamily dwellings, at work, church, or school, at some place of business, at recreation, and so on. What is becoming clear from this information is the following: although they account for less than 10 percent of homes nationally, 50 percent or more of the U.S. annual tornado fatalities now occur in mobile homes. In some areas of the Southeastern U.S., mobile homes constitute on the order of 20 percent of the homes, and so such populations are seriously at risk from tornadoes.

Although much safer than mobile homes, about 20 percent of the annual U.S. tornado fatalities occur in site-built homes. Generally, these fatalities are in the strongest tornadoes in homes without basements or nearby tornado shelters. Whereas in the past, in the tornado-prone areas of the U.S., many people chose to have "tornado shelters" nearby, most people currently do not have such shelters available. In some parts of the U.S. - again, notably in the Southern Plains and the Southeast - homes do not always have basements. In a violent tornado, where even interior walls do not remain standing, the residents of homes without shelters or basements are vulnerable.

People also are vulnerable in motor vehicles. They may not even be aware of the developing weather situation. In fact, people have been known to drive directly into ongoing tornadoes. People killed in motor vehicles account for about 10 percent of the annual total.

Therefore, about 80 % of the fatalities are associated with: mobile homes, homes without shelters, and people in motor vehicles. These seem to be the most likely situtations to put our resources toward reducing tornado fatalities.

Given the preceding information, it seems to me that although I definitely want to see the warning accuracy and lead time for tornado warnings increase on the average, doing so does not seem likely to have a significant impact on the annual U.S. tornado fatality rate!


5. Technology and the fatality rate

Some people in the tornado "business" (some forecasters, researchers, and their management) believe that warning lead time is the end-all and be-all goal for both research and operations. I've already said elsewhere that this does not seem reasonable to me. I don't think that increasing the lead time of tornado warnings is going to have much of an impact on the fatality rate - some, perhaps, but it is just not likely to make an important difference. I certainly don't believe that increasing lead times significantly will reduce the rate to half the current rate.

Furthermore, the accuracy of tornado warnings, in terms of increasing the probability of detection (POD) and reducing the False Alarm Ratio (FAR) is strongly dependent on the state of the science. Unless the state of the science improves, increasing the POD is inevitably associated with increasing the FAR, and decreasing the FAR is inevitably going to result in a decrease of the POD. Improving both at the same time is probably possible, but there are only a few avenues for actually accomplishing this:

(a) "cook" the numbers in the verification game

Sad to say, this option has been used in the past and probably will continue to be used. Although such an approach strikes hard at the integrity of the system as a whole, managers and some forecasters are ready to do anything to improve their "numbers," even to the point of cheating. Suffice it to say that I detest even the thought of such an approach, but ...

(b) improve the technology for detection of tornadic potential

Our nation has a love affair with technology and neither politicians nor NOAA managers are very excited about programs that involve people (i.e., training and education). I'll have more to say about training and education shortly, but it's clear that radars and satellites have powerful advocates in the system. Radars and satellites are indeed very useful tools for a host of weather forecasting applications. Radar is the single tool that is arguably the most responsible for the existing state of the science of tornadoes, and its role will always be large in the tornado warning process. Moreover, both radar and satellites have many applications besides their use in tornado forecasting, warning, and research. There probably are many good reasons to explore and implement new and better radars and satellites that have nothing to do with tornadoes.

Satellites, in particular, do not give good value in relation to their enormous cost when we look at their relevance to the tornado problem. Satellite images are nice for tornado forecasting, but if we didn't have them, we could still function reasonably well. Satellite information derived from expensive and complex processing of multispectral observing capability has been disappointing in the past and I frankly don't see any reason to spend more on such methods. Why spend huge amounts of new money trying to make something of a system that has failed repeatedly to live up to its promises?

While I don't dispute the critical role for radars in the tornado warning problem, their huge expense raises questions about how much "bang for the buck" they represent if we want to improve the fatality rate. The existing WSR-88D is a pretty decent radar, and will continue to be the backbone of the warning system for a long time to come. The problem with justifying huge new investments in radar technologies is that warnings for the tornadoes that cause most fatalities are already pretty darned good. A new radar might jump the statistics up a bit, but most of the improvement will be for events that tend not to kill anyone.

Radar and satellites are not cost-effective ways to reduce tornado fatalities. Although I don't believe we should stand still in the area of developing new and better radars, such technological advances are just not a very good investment in terms of the fatality rate. Conceivably, some new system (such as the shiny new toy at NSSL, the Phased Array Radar) could eventually lead to improved warning accuracy and/or lead time for tornadoes (although that remains to be shown!). As noted, I'm not convinced that warning issues are the key to reducing tornado fatalities, so the cost of a new radar system would need to be justified in other ways than improved tornado warnings. Those other reasons for investing in radar and satellites might well be important, but they are not pertinent to this issues I'm considering here.

(c) improve the level of education and training about tornadoes in the forecast offices

Forecaster training has long been a major issue with me. You can read more about my concerns here and here and here. For purposes of this discussion, I believe that some improvement in tornado forecast accuracy would be possible if we spent the resources to do a credible job of training NWS forecasters. My guess is that this increase in accuracy through training is comparable to what has been proposed recently by some NOAA management for the next 25 years in the NWS mostly through technology (see above) but also perhaps with some better science (see below). That is, training alone could increase tornado warning PODs to roughly 0.8 and reduce FARs to about 0.5, with perhaps an increase of the average lead time for all tornadoes to about 15 min. Of course, the warning verification "numbers" for all tornadoes are not very important for the fatality rate, as I've tried to suggest. A credible training (and education) program might produce some increases in warning accuracy for significant tornadoes, as well. But I'm not holding my breath that such a boost in training resources for a credible training program will ever happen. I note that a training program of the sort I'm advocating would be a lot cheaper (on an annual basis) than buying new radars! NOAA and the NWS are simply not interested in investing in people!

(d) advance our scientific understanding

NOAA has spent $billions, literally, in recent years implementing new technologies in the forecast offices. Unfortunately, the NWS has never had a real commitment to research ... that sad lack of interest in research goes back to a time early in the history of the Weather Bureau (what the NWS used to call itself) ... in the 19th century.* The only way to realize real operational value from the huge investments the NWS has made in technology is to put some fraction (say 10 %) of that investment into long-term, sustained research. New capabilities can have a substantial impact on operations, but only if the new capabilities are explored in a systematic way - that is, research.

For tornadoes, we need to learn what makes most supercells nontornadic. Forecasters have to be able to distinguish tornadic from nontornadic supercells before the action begins. If we learn what makes the difference, our tornado warning accuracy could improve by leaps and bounds. There still will be the problem of nonmesocyclonic tornadoes, but they are typically not the ones that do the most damage, nor are they responsible for most fatalities.


6. Where I believe our priorities should be

Some might find it surprising that I don't believe that improved warning accuracy is the most important issue. I believe that improving the accuracy of our warnings should always be a priority. I would prefer that the NWS put out the best product it can, given the technology and science it has to apply to the task.

But if I were given the job of setting priorities for spending the "bonanza" of resources represented by my fantasy: a "tornado" budget of $300 million annually, my choices would not be to spend resources on satellites (my last choice), radars, or even big scientific programs. The following is the program that presents my proposal for what would be most effective way to lower the fatality rate. These items are not in any particular order, and their proposed funding is in 2002 dollars. The proposed numbers constitute a crude priority system. Note that all of these items should be considered as new monies -- additions to the existing expenditures of about $24 million (give or take a few $million), annually.

(a) Make NSSL a true Federal scientific research lab once again.

Increase their base funding to equal their current annual budget. If "soft money" resources are found, fine ... use them to support temporary "job shop" projects as opportunity permits, [These soft money-supported programs should be charged some reasonable "overhead" to help underwrite the NSSL science infrastructure. The existing use of soft money to pay recurring bills is spiralling out of control, such that the core scientific process in NSSL is being threatened.] NSSL no longer has the base funding (thanks to decades of level base funding during a time of double-digit inflation) even to pay the salaries of their existing staff and keep the buildings open. Increase their federally-funded FTEs to include order 5 new Ph.D.-level scientists, with most of them committed to tornado-related research. Use some of the funds to create continuing support for field observations. Become a funding agency once again, as they were in the past - use some of the additional funds to support tornado-related external research programs via a proposal-based process. [$4 million]

(b) Develop an interdisciplinary program to explore the psychological, economic, and sociological aspects of tornadoes.

We have precious little hard information about "the public" use of tornado warning information. If we're to develop plans and procedures to provide information to the public about tornadoes, we need to get away from the ad hoc approach to creating infrastructure - the history of the development of the NWS "system" for dealing with tornadoes is not one of systematic exploration of alternatives. Rather, hasty decisions made in response to some unfocused management directive have "hardened" into the existing system. It's virtually certain that if we knew enough about how best to create this infrastructure, we would choose to do things very differently than what is currently being done.

Much of the reason for this situation has been the "top-down" management so characteristic of NOAA and the NWS. Some high-level bureaucrat makes a decision based on precious little actual working knowledge, and this becomes the basis for implementation, come what may. For example, the NWS seems committed at the highest levels to converting to probabilistic forecasts and warnings. However, this remarkable commitment (which I wholeheartedly support) needs to be backed up with a serious investigation of how best to implement probability-based forecasts and warnings. Especially for warnings, I have little faith in the current NWS management to do anything substantial to plan for the implementation of probabilistic warnings operationally. Much needs to be done to learn how to do this most effectively for the heterogeneous "public" and to help that public understand precisely what they will be getting from probability-based warnings, as well as how best to use it for their benefit.

Further, there are many economic issues associated with the tornado problem. Installing shelters is an expensive proposition and I'm personally opposed to unfunded Federal mandates. We need to explore a number of issues related to the cost of saving lives and what we might be able to do to reduce property damage by changes to construction in tornado-prone parts of the U.S. What are the economic trade-offs that need to be considered? What economically viable recommendations should we make? Is the reason for the current lack of tornado shelters purely an economic decision or is it caused by something else? The same goes for construction practices - most homes in the tornado-prone parts of the United States (everything east of the Continental Divide) are not very tornado-resistant. How do we address these issues in a way that something could actually be done to increase public safety without bankrupting individual home owners, home builders, and municipalities? Clearly, the mobile home safety problem is reaching critical proportions, so how do we deal with the issue in a way that is equitable and practical and yet actually improves the likelihood of reducing tornado-caused mobile home fatalities?

Human psychology also plays an important role. How do people perceive tornadoes? What variability in this perception exists within the public? What actions are they likely to take when threatened? To what extent do people have action plans in case of a tornado? What differences in perception exist between those who have and those who have not experienced a tornado?

Recently, the existing safety recommendations regarding motor vehicles have been questioned. We need to do more research to find out what is the best recommendation to make that will actually be beneficial in reducing fatalities in vehicles.

We need to know more about the injuries, as well as the fatalities. Fatality figures for tornadoes can be somewhat controversial, owing to a lack of clear-cut definitions and to an absence of follow-up work after events - some people die of injuries but are never counted. Some fatalities are indirect (not directly caused by the tornado) such as heart attacks and motor vehicle accidents while fleeing a tornado. Including the medical profession systematically in morbidity studies would be of considerable assistance, and we would be able to say more definitively what are the medical causes of tornado injuries and fatalities.

Since this is necessarily a multidisciplinary effort, it would require substantial funding to create an appropriate center for multidisciplinary interaction, and to provide "seed" money funding for the research. [$50 million]

(c) Support the development of a quality tornado database

At the moment, the collection, interpretation, and archival of imformation about the occurrence of tornadoes and their associated impacts is mostly being done by forecasters within the NWS as an extra duty, and then administered through NOAA's National Climate Data Center (NCDC) -- their publication Storm Data. The Storm Prediction Center also maintains a log of severe weather events for their own purposes, mainly because NCDC is incompetent, and the formal procedure is glacially slow. Thanks mostly to some volunteers, a "quick response" survey team (QRT) is presently being set up by the NWS to do systematic scientific surveys for at most a few tornado events per year. This QRT is essentially unfunded, except for travel support. Increasing the QRT budget would permit more comprehensive surveys of more events. [$1 million]

If we're to improve forecasting, we need as accurate as possible a record of what did (and did not) happen regarding tornadoes every year. If we're to reduce fatalities, we need to learn in detail from the events that produce injuries and fatalities just what is going on that results in those casualties. The existing plan is grotesquely inadequate to the task. Gathering of information about tornado events cannot be limited to imposing an "extra duty" on NWS forecasters. We should fund storm climatology "experts" in all the regions who would be responsible for supporting the NWS field personnel in this duty. And there should be a small group at NSSL - the logical place for such an activity - to coordinate the surveys and archive the data. It can be argued that NCDC would be the logical place for this activity. That would be so if NCDC was even remotely competent, but they are not and I simply don't trust them to do this properly. [$10 million]

(d) NWS infrastructure improvements

Although I think that radar "modernizations" and, to an even greater extent, weather satellite upgrades need to be justified mostly in terms of other benefits, apart from tornado issues, I believe that some fraction of those programs (mostly radar) could be included in my proposed enhancements to the "tornado-related" budget. Since this is not currently included in our annual expenditures, its inclusion is actually a proposal of mine. This would eventually help to buy radars and other needed upgrades as technology improves. It should not be used exclusively for radar, since the justification for radars should include more factors than just tornadoes. Let someone else who will benefit from radars ante up their fair share of the cost of radar (and satellites) [$100 million]

(e) Create an NWS Forecaster's "Postgraduate School"

Since this is only a fantasy exercise, I want to include my fantasy about a substantive forecaster's education and training institution. It will need a substantial start-up budget for facilities, equipment, moving expenses for the permanent staff, etc., and then will need to include enough resources to support in-class participation for forecasters (not the so-called "distance learning" which has a place but cannot serve to replace having a real instructor in a real classroom), equipment upgrades, extended visits by scientists, etc. See here for a discussion. [$32 million]

If the original total relatied to tornadoes (see above) is $24 million, then that means we could spend up to $276 million to improve things to the point where we would cut the annual fatality total in half and still "break even" by my dubious accounting. What I've just proposed totals $197 million ... $79 million less than the potential "bonanza"! In priority order:

  1. Multidisciplinary program
  2. Forecaster's Postgraduate School
  3. Tornado database
  4. NSSL base budget
  5. NWS infrastructure

That is, if we were able to spend more than what I've proposed, this list is the priority order in which those additional resources should be allocated, in the interest of saving lives.


7. Some things to ponder

Note that the production budget for the movie "Twister" was about $70 million. The last major effort to study tornadoes was the so-called VORTEX Project of 1994-1995. It cost about $2 million during its two-year existence. Thus, for the price of producing one Hollywood movie of dubious quality and virtually no scientific reality, we could have run VORTEX for 70 years, - or, during its two year existence, the research budget for the project could have been increased by a factor of 35! The movie itself grossed about $300 million, which paid for its production and offered the producers a pretty substantial profit. How much of that profit was plowed into tornado research to save lives ... the ostensible goal pursued by the protagonists in the so-called "plot" of the movie? Exactly ... nothing!

The entire budget for the NWS is, as I have said, on the order of $700 million per year. If we make the assumption that there are roughly 100 million taxpayers in the United States, this means that the cost to each taxpayer to fund the entire range of things the NWS does is about $7 per year. Such a figure is roughly equal to the price of one "meal" in a fast-food restaurant - or about the price of seeing one movie (without the parking fee or popcorn, candy, and drinks from the lobby). It strikes me this represents quite a bargain, even given the problems I like to gripe about within the NWS. If everyone was willing to give up one fast-food meal or movie (say, the sequel to "Twister" if it ever comes out), we could afford to double the budget of the National Weather Service!

However, let me put this another way. Remember ... my hypothetical "bonanza" that we could spend on tornado forecasting and research is about half the current budget of the NWS. So if each taxpayer were to give up that one meal in a fast-food restaurant or choose to miss one movie per year, this is about 3.5 times the cost of what I've proposed to really do something serious to improve our society's ability to learn about and deal with tornadoes. If this investment were to reduce the annual fatality rate by one-half, it would essentially pay for itself, with money left over.

So where are your priorities? How do you think we should allocate our nation's resources? Let's talk about your proposal.

* See: Cox, J.D., 2002: Storm Watchers: The Turbulent History of Weather Prediction From Franklin's Kite to El Nino. John Wiley, 252 pp. ISBN 0-471-38108-X

Return to Chuck Doswell's Essays Page

Return to Chuck Doswell's Home Page