Some Issues Arising from the 3 May 1999 Central Oklahoma Tornado Outbreak

Roger Edwards

DISCLAIMER: These opinions are mine only, and do not represent my unnamed employer or anyone else in particular. If you have problems with what I am saying, address them directly to me, not my co-workers, supervisors, friends, relatives or neighbors...none of whom have any control over what I write on my own time and equipment (re: First Amendment).

Last updated: 2-7-2

There are many good lessons and reminders from the 3 May 1999 central Oklahoma/southern Kansas tornado disaster. Atmospheric scientists, media members, and the broader public alike need to stand back and face these issues head-on....

When a large, violent tornado plows across many square miles of a metro area, people will die. If this statement is ever proven wrong, I will happily remove it from this website. My fear, however, is that the Bridge Creek/Moore tornado probably had within a few people of the lowest possible death toll for such immense devastation. Consider... The central Oklahoma populace is (in general) more tornado-attentive and weatherwise than that of other major metro areas. This is not a provincialistic statement; it is based on exposure and experience with tornadoes unmatched elsewhere. The Greater OKC SMSA has experienced more significant (F2 damage, Grazulis 1993) tornadoes in its history than any other metro area. TV stations have conducted intense "weather wars" of storm coverage for at least 15 years. The big tornado itself was in a tornado watch. It was in a tornado warning. It was in daytime and quite visible. Spotters by the hundreds were positioned at every conceivable distance and direction, observing and reporting on its progress. The tornado was audible from a half mile or more away -- and moving "only" 25-30 mph. Sirens sounded in advance. NOAA Weather Radio functioned throughout, beaming warnings and statements without interruption. Live video of this tornado and its predecessors was broadcast almost constantly on Oklahoma City TV for well over an hour before it hit Bridge Creek and the OKC metro. Radio stations of all formats broadcast warnings; and some of the most popular stations switched to continuous audio feeds of the full-time TV coverage. No tornado in history has ever had such widespread, intensive, advance warning in so many ways! Yet 38 people were directly killed -- and not just outdoors or in mobile homes, either. Yes, some people made fatal mistakes in the panic of the moment; and others (such as those who died hiding under bridges) fell prey to bad safety myths. But many who perished were doing the "right" thing: taking shelter in windowless interiors of their homes. This is because...

There is no guaranteed safe place inside a violent tornado. NONE! Tornado safety is a relative term. For the great majority of tornadoes, the well-known safety advice, for people to move to the windowless interior of their home (if no basement) is correct. Even within the path of violent tornadoes, this is still true; because most well-built houses will not be utterly flattened (F4) or scoured off the slab (F5). But in a violent tornado, some will; and there is always the very tiny risk it could happen to your house, with you inside. The odds are like a lottery ticket -- one you never want to cash.

Interior safe rooms, reinforced with concrete and/or steel, are an excellent idea and should be standard practice for all new house construction east of the Rockies. [Don't forget they offer good protection against winds from the most intense hurricanes, downbursts and derechoes also.] But a car, truck or tractor hurtling through the air at 100 mph or more into a "safe room" presents an uncertain but potentially lethal risk. Taking cover under a sturdy basement workbench is a good thing to do, but not fail-proof. Massive, crushing pieces of debris (e.g., cars, refrigerators, farm implements, and big chunks of the houses themelves) sometimes fall into basements, as I have seen during damage surveys at Spencer, SD, and other places. The underground storm cellar is, by far, the best option; but there is still the risk of flooding, door removal followed by flying debris, or heavy debris landing on the door, trapping occupants within. In short, the safety rules exist because they are proven to work most of the time; they minimize, but do not eliminate, your risk of being seriously hurt or killed.

Overpasses are death traps in severe weather. Avoid them! The old NOAA film, Terrible Tuesday, documented someone who "survived" the Wichita Falls tornado of 10 April 1979 under a bridge. This film has been seen by millions, and may be the source of some bad lessons about bridges and safety being taught in schools. But another even more hideous video has led to a much more vast spread of this safety myth. During the 26 April 1991 outbreak, a TV news crew (who could have easily evaded the tornado at interstate driving speeds) instead decided to become the news and make glory for themselves. They drove along with, and just ahead of the tornado, then stopped to film the tornado from under some bridge girders, keeping the camera rolling as the tornado passed nearby, not overhead (little-publicized fact: They were in the surface inflow jet!)...and filming the tearful and lucky survivors. Even in the inflow jet, any of those people under that bridge could have been impaled by flying debris and killed. Had the tornado vortex itself moved overhead, it may have blown them out from under that bridge.

Since then, I and other storm observers have noticed dangerous concentrations of people and vehicles under bridges during storms, doubtless inspired by that video which has been played to worldwide audiences for years. Sometimes, traffic lanes are illegally blocked, impeding emergency vehicles and risking catastrophic wrecks in rain-slicked and low-visibility conditions. People are crowding under bridges who could easily drive out of the storms' paths, or go to true shelter. Recognizing the great danger of this popular practice, several prominent atmospheric scientists, including Erik Rasmussen and Chuck Doswell of NSSL (personal communications), have tried to get the media and the American Meteorological Society to strongly discourage using bridges for storm shelters.

Fast forward to 3 May 1999. One person each was blasted out from under bridges on I-35 and I-44 in Moore and Newcastle, Oklahoma. The former wasn't found for days...the latter, dismembered across adjacent fields. Several others were injured. Yet a photo in Time magazine shows people hiding under a bridge with a violent tornado and a satellite vortex in the background, as if it is the correct and safe thing to do. [The photographer claimed to be a 25-year veteran "storm chaser" despite admitting to never seeing a tornado before, despite being utterly unknown to any similarly experienced chasers, and despite doing something no conscientious storm observer would dare try. He's full of B.S.] When will mass-media outlets (besides the Daily Oklahoman, which prominently publicized the overpass problems) make serious efforts to discourage using overpasses for shelter? [NOTE: Since I originally wrote this, Dan Miller of NWS Norman has prepared an excellent 25-slide online presentation about this problem, available online. There is much education left to do, though; I still see concentrations of vehicles parked under bridges during thunderstorms.]

The spirit of neighbors helping neighbors rebuild is happening in wonderful ways. But local-media hyperbole to the effect that this is a uniquely Oklahoman characteristic, as if Oklahomans are inherently superior people, is insulting to disaster victims and their helpers everywhere else in the nation. Volunteers from across the state, and across the U.S., have dedicated incalculable time and money into helping tornado victims recover. There can be no better testimonial to the goodness in people than this. Such was the case as well during the Missouri Floods in 1993, and Hurricane Andrew in South Florida in 1992. I know. I was living in those places during those disasters also. Like Oklahomans now, Missourians and Floridians rushed to the aid of their stricken neighbors -- as would residents living near any disaster area. [Unfortunately, a few slimeballs from Missouri, Florida, and Oklahoma looted and pillaged their neighbors after these disasters. May they suffer great pain for their evil, which also knows no state boundaries.] Aid rolled in from many other states in each of these events. The enormous outpouring of help in all these cases reminds me, even in my most turbid moments of cynicism, that people are capable of great good. Let's not, then, turn the beautiful spirit of giving into an exercise in provincialistic chest-thumping. How shameful. After all, goodness is goodness, wherever it is performed, and whoever performs it; and no state or locality has a monopoly on it.

Forecasters earned their pay, and everyone's respect, on this day. The 3 May 1999 outbreak was not a textbook, completely "synoptically evident" outbreak, characterized by all the classic signals well in advance (e.g., progressive, deepening southwestern cyclone, intense precursory theta-e advection in the boundary layer, strong and convergent dryline and fronts progged for days in advance, and ideal storm-relative flow profiles already in place). This is evident in how the storms formed. The initial cumulonimbus (Cb) formed and died fast S of the Red River. The first supercell developed in a cirrus hole SW of Lawton, well away from any apparent surface boundaries; and the second appeared to developed from a cluster of altocumulus towers near Altus, close to an apparently non-convergent dryline. [Radar analysis in the months since the outbreak indicates the eventual OKC supercell formed on a large boundary layer roll; see this page with a slow-loading javascript applet for more details. Also, for a more complete review of the meteorology of this outbreak, see the formal paper by Rich Thompson and I from the Dec. 2000 issue of Weather and Forecasting entitled An Overview of Environmental Conditions and Forecast Implications of the 3 May 1999 Tornado Outbreak.]

Despite the non-obvious nature of this event in advance, SPC outlook risks were upgraded during the day as the situation became more ominous. Watches and warnings were timely and accurate. The Norman forecast office, not following any regulation but instead compelled by the urgency of the situation, issued a "TORNADO EMERGENCY" statement after the warning, highlighting in certain terms the great danger to the populace of the southern OKC metro:

657 PM CDT MON MAY 3 1999





To think of and act on that initiative in such frantic circumstances was an unprecedented act of courage, skill and public service by a meteorologist. The entire afternoon and evening crews of WFO OUN and SPC deserve the highest possible Federal honors for their superior performance under conditions never before endured by their surrounding communities.

The outstanding watches and warnings for this event show the skill and dedication of forecasters -- but belie the pitiful, bare-bones funding of our nation's severe weather forecast services. Only two years ago, during a NOAA/NWS budget shortfall, meteorologist jobs were almost eliminated from both the Storm Prediction Center and the National Hurricane Center! How could this asinine bureaucratic decision have almost come to pass? Only through the coordinated bipartisan efforts of the Florida delegation was NHC spared the axe; and Rep. JC Watts (R-OK) almost singlehandedly saved SPC. The rest of the Oklahoma congressional delegation (including both Senators) was alerted to the problem, and did nothing. All members of Congress outside Oklahoma and Florida did nothing.

Severe weather only matters in Florida and parts of Oklahoma? Tell that to any hurricane survivor in the Carolinas or Louisiana, or tornado victims in Alabama, Kansas or Iowa. Voters take note: Talk is cheap; results speak for themselves. Unless you live in JC Watts' central Oklahoma district or Florida, your Senator and Representative showed he/she was not interested in your severe weather safety!]

It is my hope that severe weather research and forecasting will not only be made immune by law to the chronic budget hatchets, but that they will finally be funded at a level much greater than the present "barely adequate" state. [This is up to you and your congressman!] Standing on the foundation of a shattered home, the President spoke of renewed federal emphasis in funding the government's share of the long-proposed Weather Center facility at OU; we'll see if these words translate to action. [This is also up to you and your congressman!]

Because of the outbreak, the Norman WFO MIC (Meteorologist in Charge) and several big names in severe weather research testified before the House Basic Research Subcomittee on 16 June 1999. Is this as far as it will go? The money which went into making the movie Twister could fund the real-life project V.O.R.T.EX. for over 60 years! How important is the weather in your daily life?

Although the forecasts and warnings worked fabulously in this event, the public must accept that our science is imperfect, and that some deadly weather events will go unwarned. Grand pronouncements of "no surprise" forecasting make cool sound bites; but they ignore scientific reality. "No surprise" promises made by members of high NWS management (most of whom have never issued an NWS watch or warning or public forecast of any kind!) are setups for failure...and who will be blamed? Not the managers making these asinine pledges, of course, but the field troops -- the forecasters who have now been publicly obligated to do the impossible. The Salt Lake City tornado of 11 August 1999 was the first of many inevitable "surprises" on which the media will pounce. And the resulting ugly fallout from their management's unkeepable promise will certainly be "no surprise" to most forecasters.

Is this a cop-out for inadequate performance? Absolutely not! There are valid reasons why "no surprise" meteorology will never be possible. You see, weather forecasters deal with an infinitely complex and interdependent global physical system: our atmosphere, oceans and land surface. Within it, oceanic and atmospheric data is gathered at far too-coarse resolution in time and space to detect the small subtleties which make the difference between thunderstorm and tornado-spawning thunderstorm.

Yes, the science should always strive to improve -- and is doing so. But "no surprise" implies the unattainable: absolute perfection. Pure perfection in forecasting won't be seen at least until we have

  1. Surface sensor sets every mile or more dense across the whole nation and adjacent oceans, reporting highly precise wind, thermal, drosothermal, radiative and pressure observations by the minute,
  2. Direct-measurement thermodynamic and kinematic soundings every few miles worldwide, several times an hour,
  3. Full physical understanding of fluid dynamics and chaos theory as applied to every possible contingency of air/water and air/ground interaction (ask any atmospheric dynamicist how close we are to that!),
  4. Numerical models which represent every cubic meter of our atmosphere, incorporating that perfect theoretical understanding and extreme observational input density,
  5. Machines of currently unfathomed processing power to run such models,
  6. Technicians who can keep those machines running without fail, regardless of power outages and software and hardware failures, and most importantly,
  7. Human forecasters, intensely motivated, trained with unprecedented rigor, undistracted by irrelevant tasks, who can interpret and account for the biases and undetected features which will still be endemic to the models.
Oh, and there must of course be no human error anywhere along the way. Consider all that, and the net lesson on forecast perfection is: Get real. It ain't gonna happen! The best we can hope for is to forever improve, never level off in our understanding and skill, and minimize surprises. That is the noble goal of all atmospheric science.

I have heard talk that forecast discussions as a whole should be less detailed from now on, so that less can appear publicly to go wrong. Rubbish! Forecasters are not in the business of withholding information from the users of their forecasts. This cover-your-*ss (CYA) sentiment arises from failures of very specific and/or precise forecasts of storm evolution in both national outlooks and local discussions -- for this and perhaps a few other events. On 3 May, national and local outlooks did a great job upgrading the threat during the day as more ominous clues became apparent. The forecast system should work that way when outbreaks (like this one) are not "synoptically evident."

Still, discussions accompanying the earliest forecasts (which tend to be based largely on model guidance) did contain detailed elements regarding how storms were supposed to evolve. Some of those forecast details were underdone; storms lasted longer, and produced more intense tornadoes for longer periods, than anticipated 12-36 hours in advance. This does and should pale in importance to the recognition that the valid afternoon outlooks, watches and warnings were excellent at alerting the public and other meteorologists to the impending danger. But let's mash, like a cockroach on the floor, the CYA thoughts of deliberate vagueness which occasionally crawl from the more old-school walls of operational meteorology.

Forecasters are supposed to examine and discuss, within time constraints, all relevant clues to the weather threat. This includes environmental shear and thermodynamic parameters which influence storm character (e.g., LP/CL/HP), associations between storm-relative flows and the occurrence of tornadoes in supercells, and supercell evolution case studies, all of which have been well-documented in the literature (e.g., Davies-Jones and Brooks 1993, Brooks et al. 1994, Thompson 1997, and Rasmussen and Straka 1998). Scientifically sound severe weather forecasts -- no matter the source -- must discuss in some detail such formal scientific concepts as they apply to the daily situations, or the forecaster is shortchanging his audience and his own ability.

One argument has often been raised against forecasting storm character based on the expected environment is: "We don't have the ability." That's crap. "Ability" and "perfection" are not synonyms! If the concepts discussed in a forecast have been recently formally published -- passing the rigors of scientific peer review -- and have been used successfully in other events (despite the occasional, unavoidable exceptions) -- then YES, WE DO HAVE THE ABILITY. Nothing is guaranteed in meteorology; and to hypotheses and theories there will inevitably arise failures. The supercells of 3 May 1999 provided some important ones. Let's do research and learn from them! Instead of using these exceptions as excuses to revert to primitive, overgeneralized, CYA forecast methods and vague discussion reasoning, severe weather forecasters should continue to aggressively dig for and discuss detailed clues about the situations at hand. Forecasters also must learn to accept, and use for self-education, the occasional forecast failures.

I refuse to feel a shred of remorse for my fascination with severe storms; so don't even try to lay any kind of "How could you possibly enjoy...?" type of guilt trip on me. It will not work. Severe storm observing is, and will always be, a rewarding, educational, and yes, enjoyable hobby, wherever the storms may happen to strike. While I feel compassion for the victims, there is no reason for any of us who carefully observe severe weather to stash away our weather clothing and wallow in shame when it destroys property or hurts people. We don't control tornadoes or wish any harm on people in their path. Those of us who observe and study them would be quite happy if they somehow avoided houses, always remaining in open fields.

But I am pragmatic enough to realize that these columns of rotating air we call tornadoes sometimes lay waste to the things we build in front of them. Tornadoes are basically air -- to me, a normal and integral part of God's creation, like any other form of wind and water; and they were here long before we were. We just have to deal with them. How? Work hard to reduce the human toll when they do happen, and continue to learn and improve our ability to protect people and build safer buildings. Meanwhile, know that responsible storm observers don't celebrate human misery or cheer for death. Anyone who does is a demented sadist and subhuman garbage. And if I ever see any so-called "chaser" doing so out in the field, I can and will crash the party.


Brooks, H.E., C.A. Doswell III, and J. Cooper, 1994: On the environments of tornadic and nontornadic mesocyclones. Wea. Forecasting, 9, 606-618.

Davies-Jones, R.P., and H.E. Brooks, 1993: Mesocyclogenesis from a theoretical perspective. The Tornado: Its Structure, Dynamics, Prediction, and Hazards (C. Church et al., Eds.), Geophysical Monograph 79, Amer. Geophys. Union, 105-114.

Grazulis, T.P., 1993: Significant Tornadoes: 1680-1991. Environmental Films, St Johnsbury VT, 1326 pp.

Rasmussen, E.N., and J.M. Straka, 1998: Supercell morphology variations. Part I. Observations of upper-level storm-relative flow. Mon. Wea. Rev., 126, 2406-2421.

Thompson, R.L., 1997: Eta model storm-relative winds associated with tornadic and non-tornadic supercells. Wea. Forecasting, 13, 125-137.

_____, and R. Edwards, 2000: An overview of environmental conditions and Forecast implications of the 3 May 1999 tornado outbreak. Wea. Forecasting, 16, (pending Dec. 2000).


[This does not imply any endorsement by these folks of the opinions expressed in this essay.]

I appreciate these individuals for their constructive comments, discussions, ideas and/or and e-mails over the past year or two on the issues discussed here: Harold Brooks, Steve Corfidi, Chuck Doswell, Doris Grazulis, Michael James, Dan Miller, Vince Miller, Erik Rasmussen, Joe Schaefer, Karl Schulze, Rich Thompson.

Tornado Intercept: 3 May 1999

Roger Edwards Home Page

SkyPix weather photo gallery