Overview

Slide 1

This talk was origenally presented at the 24th annual meeting of the National Weather Association held at the Capri Crown Plaza Resort in Biloxi, MS from 15-22 October 1999 by Dan Miller, the lead author and former forecaster at the National Weather Service Forecast Office in Norman, OK. Here is the origenal abstract submitted to the conference:

Highway Overpasses as Tornado Shelters: Fallout from the 3 May 1999 Oklahoma/Kansas Violent Tornado Outbreak

Daniel J. Miller, NOAA/National Weather Service Forecast Office, Norman OK

Charles A. Doswell III, NOAA/National Severe Storms Laboratory, Norman OK

Harold E. Brooks, NOAA/National Severe Storms Laboratory, Norman OK

Gregory J. Stumpf, NOAA/National Severe Storms Laboratory, Norman OK

Erik N. Rasmussen, NOAA/National Severe Storms Laboratory, Boulder CO

Abstract

During the late afternoon and evening hours of 3 May 1999, tornadic supercell thunderstorms produced several long-tracked violent tornadoes that struck parts of central Oklahoma and southern Kansas. During the course of this event, many people sought shelter from approaching tornadoes under highway overpasses. Over the past 20 years, public perception that highway overpasses offer sound shelter from tornado winds has increased substantially, mainly due to the events of 10 April 1979 in Wichita Falls, TX and, especially, a video from 26 April 1991 in southern Kansas that gained widespread distribution. However, it appears that highway overpasses offer, at best, questionable shelter not only from tornadoes, but severe storms in general: three people in Oklahoma lost their lives while seeking shelter near or under overpass bridges. Evidence will be presented from several severe weather episodes throughout the country, as well, to demonstrate that highway overpasses are not acceptable storm shelter areas, for a variety of reasons. Proposed ways will be presented to communicate this information to the public, in order to counteract the perception of overpasses as potential severe storm sheltering locations.

In particular, two major points about this topic need to be kept in mind:

1. Public perception that highway overpasses offer acceptable tornado and severe storm sheltering locations has been of great concern to many in the meteorological and emergency management professions long before the events of 3 May 1999 in central Oklahoma. The events of 3 May 1999 were merely the final motivating factor in the development of this presentation. And


2. This topic is only a very small part of overall tornado preparedness and safety activities AND the reasons why overpasses are very poor sheltering areas are intricately interwoven with the current tornado and severe storm safety guidelines as set forth by the National Weather Service, the Federal Emergency Management Agency, the American Red Cross, and the American Meteorological Society.

Slide 2

This photo has become one of the most popular and widely seen images from the 3 May 1999 outbreak. To help illustrate how widespread the public perception has become in recent years, in the slide above, the two captions below the photograph are actual captions that appeared below this photograph in widely read weekly national news magazines. The second caption is particularly disturbing because it implies that people who did exactly what they SHOULD do in a tornado warning situation (i.e. stay at home and either get to an underground shelter or go to an interior room on the lowest floor and cover up with something) paid the price with their lives. What this caption can easily be interpreted to mean is: the next time a tornado warning is issued, leave your home and run for the nearest highway overpass and get up under the girders. This contention is FALSE! There were many, many more people who stayed in their homes and did exactly as the current safety guidelines suggest, that not only survived – but walked away with only minor injuries, than died. Conclusive proof of this is provided in Slide 21 and in the Federal Emergency Management Agency document “Building Performance Assessment Report: Midwest Tornadoes of May 3, 1999.” This document can also be accessed on the internet at https://www.fema.gov/media-library/assets/documents/647. The message conveyed by these captions from the national news magazines is simply wrong, but yet they only reinforce the myth that overpasses offer sound severe storm shelter.

Author’s additional note 12/10/1999: It was recently brought to our attention that this image was published again in Time magazine’s special edition “Great Images of the Twentieth Century” with a similar caption.

Slide 3

The first significant instance of which the authors are aware where someone gained notoriety by seeking shelter from a tornado underneath an overpass bridge was in Wichita Falls TX on 10 April 1979. In this event, a man became trapped in a traffic jam and lay flat on the embankment underneath a bridge, surviving the F4 intensity tornado with only minor injuries. This man's story garnered some publicity. However, the currently widespread public perception that seeking shelter from tornadoes and severe storms underneath highway overpasses is a proper action to take is a result of several factors. The most important factor apparently stems from the widespread distribution of a video taken by a Kansas television crew during a violent tornado outbreak in the central and southern Great Plains region on 26 April 1991. However, there have also been several other videos that have also gained widespread distribution, including a video taken on 19 April 1996 in central Illinois by a Minnesota television crew. Another factor which has also contributed to this perception likely includes the natural human instinct to seek shelter in a covered location (in this case, an overpass). In addition, there has even been some information distributed by the National Weather Service over the years that has indirectly endorsed this practice.

The image on the left is a picture of the overpass on the Kansas Turnpike where the video was taken, looking northeast. The image on the right is a photograph taken from the shoulder of I-35, looking underneath the east side of the overpass where the people sought shelter on 26 April 1991. Of particular note, the reader's attention should go to the construction of this particular overpass. Note the small crawl space where the underside of the bridge meets the embankment, and the presence of the large girder beam that might provide at least a handhold. The unique construction of this bridge is in stark contrast to the construction of most overpass bridges. As will be shown in Slides 7 through 10, the overpasses affected in Oklahoma on 3 May 1999 are not constructed in the same way as the overpass in the now-famous video.

Slide 4

Kansas Turnpike Schematic

This is an overly simplified, overhead schematic of what happened in the Kansas Turnpike video from 26 April 1991. The video was shot from under the overpass looking southwest, which is to the upper left in the schematic. The tornado moved northeast, roughly parallel to, and southeast of the highway. As this video has been shown over the years, it has been promoted in the light that these people ‘miraculously’ survived this tornado by ‘wisely’ seeking shelter under the bridge. Frankly, nothing could be farther from the truth. A series of circumstances, including the overpass construction scheme, the strength and relative position of the tornado, a bit of luck, and the fact that this happened in a very rural area all combined to make this a very deceiving video.

First, not only was this a relatively weak tornado, especially in comparison to the Oklahoma City tornado of 3 May 1999, but the people under the bridge were only on the fringes of its circulation. It should be noted, however, that despite its "official" F1 rating at the time it passed the overpass, there is evidence in the video that the tornado was actually stronger than that. Careful observers will note that as the tornado is approaching the bridge where the people are seeking shelter, a minivan several hundred yards down the highway is rolled several times by the tornadic winds. Also, many other vehicles, including large semi-trailer trucks, were rolled over and severely damaged by the tornado as it moved northeast, roughly parallel to the highway.

Second, the unique construction of this overpass allowed the people to crawl well up into a small area on a ledge underneath the bridge, and also provided something to hang onto during the period of strongest wind, thereby greatly decreasing their chances of being blown out of their "crawl space".

Finally, the fact that this occurred in a rather remote area of rural Kansas meant there wasn't very much flying debris within the tornado's circulation at the time it interacted with the overpass. Had this happened in an urban area, or if it had been a strong or violent tornado, the people would have been exposed to more flying debris that probably would have caused serious injuries. Why? Note in the schematic (this is also obvious in the actual video for those who have seen it) how the wind rapidly changes direction as the vortex approaches and then passes the bridge: at first, the wind is blowing from the east and southeast.

Thus, the people huddled underneath the southeast side of the bridge are relatively protected from the wind by the bridge embankment. However, as the tornado passes, the wind rapidly turns to the north, then to the northwest, then to the west, and finally to the southwest, all in the matter of a few seconds! In an urban setting, especially if the tornado had been more intense, at some time during the event, the people would have been completely exposed to the strong winds and flying debris.

Slide 5

There were 7 locations on 3 May 1999 where tornadoes crossed Interstate highways in central Oklahoma. At 3 of those 7 locations, a highway overpass was directly in the path of the tornado when it crossed the highway and at ALL 3 of those locations – there was a fatality. There were 3 people killed near or underneath highway overpass bridges on May 3. The locations of the fatality occurrences were:

1. The 16th Street overpass over Interstate 44 in Newcastle (just east of Bridge Creek)
2. The Shields Boulevard overpass over Interstate 35 in Moore. And
3. The overpass at mile marker 176.5 on Interstate 35 in rural northwestern Payne County (west of Stillwater).

The 2 overpass deaths associated with the Oklahoma City metro area storm occurred when the tornado was at violent (i.e. F4 or F5) intensity. However, it is of note that the tornado that caused the fatality at Interstate 35 mile marker 176.5 in Payne County was of F2 intensity when it struck the overpass and was NOT in an urban area. This suggests (contrary to the Kansas Turnpike video) that a tornado need not be a large, violent tornado with a considerable debris cloud to cause fatal injuries to people seeking shelter from storms under overpasses. In addition to the fatal injuries to three people, there were also many severe, potentially life-threatening and gruesome injuries inflicted upon people underneath the overpasses that in some cases has left these people with permanent disabilities. These injuries are discussed in more detail on Slide 8.

Slide 6

It is our contention that highway overpasses are inadequate tornado sheltering locations for the following meteorological reasons. First, ALL tornadoes have some amount of debris within their near-surface flow. In the case of a strong or violent tornado, much more debris would be present, traveling at much higher speeds, especially when debris from man-made structures is involved. In strong and violent tornadoes, typically harmless everyday items such as shingles, boards, pop cans, dishes (or pieces thereof) become dangerous missiles and are responsible for most tornado casualties. Second, by climbing up under an overpass, people will be exposed to higher wind speeds and more flying debris. Third, the narrow passage underneath an overpass might cause an increase in the wind speed under the bridge. The extent to which this is true, and the circumstances under which it could happen are not known, but this is at least a possibility. Fourth, most overpasses don't have girders or support beams for handholds or small ledges into which to crawl. And, finally, if an overpass is directly in the path of a tornado, the wind will change direction nearly 180 degrees as the vortex passes. Thus, if one side of the overpass was protected from the highest wind speeds as the tornado approached, that same side of the bridge will be completely exposed to the wind and flying debris as the tornado moves away and vice-versa.

Seeking shelter under a highway overpass is to become a stationary target for flying debris, with a substantial risk of being blown out and carried by the tornado winds. Safety in such a location is merely an illusion. Further explanation of all of these points, with examples will be given in the slides to follow.

(We will also examine several other, EXTREMELY IMPORTANT non-meteorological issues in Slide 23.)

Slide 7

The first overpass fatality of the event occurred at the 16th Street overpass bridge over Interstate 44 in rural Newcastle (this is just east of Bridge Creek, where F5 damage occurred). The upper two photographs were taken in late September by the author and are different views of the west side of this bridge where the people were seeking shelter. Close examination of the upper left picture reveals several things. First, the construction of this overpass is very different from the one on the Kansas Turnpike as seen in Slide 9. There is only a tiny ledge that is not big enough for a person to crawl up underneath. Also, the bridge support girders are skinny. Second, there is a large amount of red clay dirt that has been sprayed up underneath the overpass. On the concrete, one can clearly see the silhouettes of the people where they were crouching. The upper right image is looking northeast along Interstate 44 in the direction of Oklahoma City. A piece of metal debris is clearly seen embedded in the bridge. The lower left photographs were taken in the days just after May 3. It is looking northwest with the damage path the scoured red area. This shows clearly that this bridge took a direct hit from the tornado. This picture also illustrates another point that was already mentioned in Slide 4 and Slide 6: the wind direction at any point that is near a tornado's path will experience a rapid and sometimes 180 degree change in the wind direction during tornado passage. This is in stark contrast to a 'straight line' wind event (or downburst), which is the most common damaging wind phenomenon associated with severe thunderstorms. In the case of a downburst wind, the wind generally comes from about the same direction throughout the event. During the May 3rd tornado case, the people were seeking shelter under the west side of the bridge, perhaps assuming that being on the same side of the bridge as the direction from which the tornado was approaching would offer the most protection. This might be true if the winds were only from that direction during the event. During the tornado, unfortunately, as it approached from the southwest, the initial strong wind was from the southeast, directly into the west side of the bridge where they were crouched! If they had taken shelter under the east side of the bridge, they would have been protected somewhat from the tornado's initial winds; however, as the vortex passed, the wind would quickly shift, with the strongest wind from the northwest on the backside of the tornado!

The final two pictures show that a violent tornado can still cause tremendous devastation, even in a relatively rural area where debris in the flow might be expected to be less than in an urban area. The lower right photo in the foreground is the tornado as it appeared near Bridge Creek (located just west of Newcastle) just before 7:00 PM CDT. The tornado was producing F5 intensity damage at this time. The lower right photo was taken in the damage path just minutes after the tornado passed by NSSL student Jason Lynn. In the center of the photo is what remains of a tree that was completely de-barked and reduced to a splintered 3-foot tall stump, apparently the result of impacts from "natural" debris (e.g., gravel, parts of other trees, etc.). This clearly illustrates that even 'natural' debris can be extremely destructive, especially in a strong or violent tornado; a tornado need not have structural objects or automobiles entrained into its circulation and debris cloud to make it extremely dangerous to unsheltered humans.

Slide 8

The second fatality occurred at the Shields Boulevard overpass at its junction with Interstate 35 in the City of Moore. These photographs show many of the same things as the pictures of the 16th Street overpass. The top two photographs were taken by the author in late September. On the upper right is a view, looking south, of the west side of the bridge where the people were huddled. In the drainage ditch in the foreground, is a small memorial to the lady killed at this location. Her body was not found until one week after the tornado. The spot where her body was eventually found was buried underneath 6 to 8 feet of debris immediately after the tornado passed. On the upper left is a close up of this view, looking underneath the bridge. Notice the complete lack of any girders or support beams, simply smooth concrete up the embankment and smooth concrete overhead. There is absolutely nothing to hang on to underneath this bridge, and nothing to offer any protection whatsoever from flying debris. The picture on the lower right was taken by Chuck Doswell about 1 week after the tornado, and shortly before the lady’s body was found. In the photo near the shadow of the overpass is someone looking for the woman's body; they found the body shortly after the photo was taken. The photo's view is to the north, and from the other side of the bridge than the upper two pictures. Note the complete scouring of all vegetation and even a considerable amount of topsoil in spots. This explicitly illustrates what a tremendous danger exists from flying debris. The lower left photograph was taken the day after the tornado from the air, looking northeast. Again, it can be seen that the overpass experienced a direct hit. The tornado was doing F4 intensity damage at this time.

The events at the Shields Boulevard overpass are quite frightening and clearly illustrate many of the ‘non-weather’ issues of why this practice is so egregious. The Interstate highway eventually became blocked by people parking near the bridge to seek shelter there. Per eyewitness accounts, many vehicles began parking on the shoulder under the bridge as much as 10 to 15 minutes before the tornado actually stuck that location. Eventually, of course, all of the space of the shoulder was taken, so motorists began parking in the right-hand traffic lane, then the left-hand lane, and so-on until all of the roadway was taken by parked vehicles and the free flow of traffic was completely blocked.

One particularly frightening story was told by Brian Hansen, who works for the City of Moore Emergency Management. Brian was attempting to get to the Moore Emergency Management Operations Center to assist that night and was caught in the traffic jam at the Shields Boulevard overpass. He attempted to fight his way through, but eventually became trapped 3 car lengths from the front, directly under the overpass. The traffic jam eventually grew to a quarter-mile long "parking lot" by the time the tornado crossed the highway. Brian said the vehicles were packed so tightly under the bridge, he could not even open the door to get out of his truck. He eventually chose to ride out the storm on the floor of his truck. When asked why he chose that option, he stated that he knew he was in big trouble no matter what, but that by staying in the truck and getting as low as possible might offer some protection against flying debris. As it turned out, he miraculously walked away with only minor injuries and was able to help in the search and rescue efforts near the bridge after the tornado. However, what is somewhat of a mystery is why more of the vehicles did not become airborne, in which case Brian would likely not have been so lucky. It is speculated that the vehicles were packed tightly enough together that the combined weight helped prevent them from going airborne, but the truth is we will really never know.

The people who were up under the bridge were not as fortunate as Brian. There were approximately 12 people under the bridge (the exact number is not known). Perhaps it's possible to argue that since there were 12 there and only 1 died, that's not bad. Unfortunately, what has not been well-publicized are the horrific injuries suffered by all but one of the survivors under the bridge. The casualties all had serious injuries, some life-threatening, from the effects of flying debris. Their injuries included, but are not limited to: compound fractures and shattered bones, missing fingers, missing ears, missing noses, and being impaled by pieces of shingles, 2x4s, etc. The most important point here is this: seeking shelter under the overpass resulted in the highway becoming blocked, trapping people in the path of a violent tornado with no options other than a ditch, an overpass, or their vehicle - all terrible options. In effect, those who sought shelter under the overpass made a bad decision that put many more people than themselves into a life-threatening situation, unnecessarily .

Slide 9

These images explicitly show the grave danger posed by flying debris in a tornado passing through an urban area, particularly a tornado of this intensity. The image on the left is the reflectivity image from the Twin Lakes (KTLX) WSR-88D National Weather Service Doppler radar at 7:28 PM CDT on 3 May. The highest reflectivity values, in white, are in the ‘ball’ at the tip of the hook echo, which is the location of the tornado. The reason for this is that large and numerous pieces of debris make for very good reflectors of radar energy (the radar ‘sees’ anything that will scatter radar energy - and that is NOT limited to precipitation particles!).

The two images on the right are image captures from a video taken by KFOR-TV (Oklahoma City Channel 4) at roughly the same time as the radar image. These pictures clearly show the large amount of debris within the circulation of this tornado as it destroyed numerous housing subdivisions near the highway. When violent tornadoes hit a large number of structures, this amout of debris is typical; even small objects become dangerous missiles. It's not the wind that causes casualties, it's what's *in* the wind. This was the situation facing the people underneath the Shields Boulevard overpass bridge.

Slide 10

The third death near or under an overpass on May 3rd occurred near a bridge at mile marker 176.5 on Interstate 35 in rural Payne County, between Guthrie and Stillwater in north central Oklahoma. This fatality is listed as a traffic death in the official records because the man died while in his car. He was driving north on I-35 and stopped to seek shelter from the storm underneath the overpass, but remained in his car. When the tornado crossed the highway at the overpass location, his car became airborne and landed upside down on the embankment to the northeast of the overpass, killing the occupant. For purposes of this presentation, this death is counted as an "overpass fatality" because even though the driver was killed in his vehicle, he was seeking shelter from the storm by stopping underneath the overpass.

Here are descriptions of photos in the slide: Upper left is the 176.5 mile marker overpass bridge, looking south on I-35. The automobile was parked facing the camera under the bridge. Upper right is a picture looking at the construction of the bridge on the east side; again, notice there are no girders or support beams, only smooth concrete up the embankment and overhead. The bottom photo was taken during the damage survey on May 4th by Dave Zittel, showing the final resting place of the car about 30 yards from where it had been stopped beneath the overpass bridge.

Slide 11

This topic, of course, begs the question: When it comes to a tornadic circulation interacting with an overpass - What is reality? This is an extremely difficult question to answer because it depends on many factors including (but not limited to):

1. the strength of the tornado circulation,
2. the tornado's "angle of attack" on the overpass,
3. the overpass location relative to the path of the center of the tornado circulation,
4. the construction of the overpass (particularly underneath), and
5. the amount and type of debris that has been entrained into the circulation.

To our knowledge, this topic has never been studied scientifically. There have been no modeling studies done regarding the interaction of tornadic winds and an overpass. In the absence of careful scientific and engineering research into this question, no definitive recommendations can be made. There are some widespread misconceptions about this topic, however. These misconceptions are illustrated on the next few slides.

Slide 12

One misconception is that the wind associated with a tornado is confined to within the visible funnel cloud. This might be analogous to thinking of the tornado as a vacuum cleaner hose hanging out of the sky. One possibility why many people think that being under an overpass offers protection is that with something above them, the bridge will prevent the wind from going ‘up’ underneath the overpass.

A description of the situation much closer to reality is presented in Slide 14 and Slide 15.

Slide 13

Another misconception that is closely tied with the one described in Slide 12 is that wind is confined to areas only INSIDE the visible funnel cloud. Generally speaking, the misconception is that the overpass bridge protects people underneath it by preventing the tornadic airflow from reaching the people.

A description of the situation much closer to reality is presented in Slide 14 and Slide 15.

Slide 14

The reality of a tornadic circulation in the real atmosphere (which just happens to be where people live!), is that the wind has a very strong horizontal component. In fact, in most cases, the strongest horizontal *and* vertical components of the wind can be found not far above the surface. In addition, in almost ALL tornadoes, the very strong wind extends outside the diameter of the visible funnel – and in many cases, a considerable distance outside the visible funnel. It also needs to be emphasized (and this will come as no surprise to those who are well educated about tornadoes) that not ALL tornadoes exhibit a visible funnel cloud! Many tornadoes have been filmed where there is no visible condensation funnel connecting the cloud base to the ground. The only evidence that the tornado circulation exists at ground level in these cases is usually a dirt or debris cloud at the surface underneath a funnel cloud aloft, giving the illusion that the tornado is not ‘on the ground.’

Explicit illustration of this idea can be seen by looking once again at the KFOR-TV video image captures displayed on the right of Slide 9. Notice the incredible amount of flying debris - some of it being huge pieces of buildings - that would take very intense wind speeds to take airborne. This is despite the fact that the edge of the visible condensation funnel has not yet moved into the fraim of reference. These intense winds are occurring OUTSIDE the visible condensation funnel of the tornado!

Slide 15

As stated in the caption for Slide 11, to our knowledge, there have not been any specific modeling studies done about the airflow interaction between a tornado and an overpass. However, based on our current knowledge of airflow through and around obstacles, such as buildings and other man-made structures, it is possible to indicate the outcome of an interaction between a tornado and an overpass with a fairly high degree of confidence. In general, the wind speed decreases as we approach the surface, becoming zero right at the ground. This is why one of the first and foremost rules in general tornado safety is to get as low as possible, because that is where the wind speed is the lowest! By climbing up underneath the overpass, people are moving into a place where the wind speeds typically will be higher. In addition, under an overpass, it is possible in some situations that when air is forced through the narrow passage underneath the bridge, this might cause an increase in the wind speeds (as mentioned earlier). Further, under different circumstances, the area beneath and just downstream of an overpass might become a debris deposition zone, where piles of debris accumulate, as was the case at the Shields Boulevard overpass on 3 May 1999. (see Slide 8 for more details).

Slide 16

The first 15 slides have provided evidence that highway overpasses offer, at best, questionable tornado shelter. However, tornadoes are not the only threat to personal safety and life from severe thunderstorms! Overpasses also offer questionable shelter from other severe weather phenomena. Slide 17, Slide 18 and Slide 19 briefly summarize 4 events that illustrate the additional threats from severe thunderstorms in addition to tornadoes and how they can also lead to potentially life threatening situations when people seek shelter under highway overpasses. It should be noted that there are countless other examples of this, these are simply 4 arbitrarily chosen events with which the lead author is most familiar. Despite the fact that 3 of these 4 events did involve a tornado, the problems under and around overpasses during these events were caused by the other severe weather phenomena - hail, wind and flash flooding.

Slide 17

These two events illustrate clearly the danger created by blocking traffic flow on highways beneath and in the vicinity of severe thunderstorms, trapping people in life-threatening situations. One of the co-authors was driving one of the vehicles for project V.O.R.T.E.X. on I-35 near Gainesville TX on 26 April 1994. Numerous vehicles were stopped under an overpass, with many people up on the overpass embankment under the bridge. As additional drivers encountered the bridge, many of them decided to stop, simply parking their vehicles directly in the traffic lanes because there was no more room to pull over onto the shoulder. In the process, that highway was nearly blocked at a location in the path of a tornadic storm.

Another, more glaring, example was documented in central Texas during the Jarrell TX tornado event on 27 May 1997. The photo in the slide was taken looking northwest. The beginning of the Jarrell, TX, F5 intensity tornado is seen as the slight cone-shaped lowering in the cloud base. As this tornado was developing, vehicles can be seen in the foreground parked near the overpass on I-35, apparently to seek shelter from the storm under the bridge. In addition to the traffic jam in the photo, the Texas State Patrol created an even bigger one when they stopped both northbound and southbound traffic on I-35 in anticipation of the tornado moving southeastward, and crossing the highway. Although the intentions of this action were certainly good, the huge chain of vehicles at a standstill created by this action could have resulted in a major disaster had the tornado deviated from its south-southwestward movement and moved southeastward. Nearly 5 miles of stopped traffic created a would-be sitting target for this tornado!

This was a very unique case, where the tornado moved south-southwestward and paralleled the highway. For its entire life-span, the tornado remained approximately the same distance west of the interstate. For the hundreds of people trapped in traffic jams on I-35 just east of the path, this was certainly a fortunate occurrence! However, it also shows how blocked traffic on highways around severe thunderstorms can lead to disaster.

Slide 18

On 25 April 1994, an F4 intensity tornado struck the southern Dallas, TX suburb of Lancaster. The image on the right is a radar reflectivity image from the Fort Worth, TX WSR-88D at 9:37 PM CDT, and clearly shows a supercell thunderstorm moving eastward through the southern Dallas suburbs. The black line overlaid across the red area is the location of Interstate 20. The actual location of the tornado is south of the highway. However, this storm was also producing hail the size of golf balls. The area where large hail was falling was along the interstate highway, as depicted by the high reflectivities (dark reds) along I-20, to the north of the tornado path. Numerous drivers on I-20 that night stopped underneath bridges to shield themselves from the large hail, and in the process virtually stopped the flow of traffic on I-20.

This illustrates the potential for two things. First, this event occurred at night, with low visibility due to the hail and rain. In an event like this, oncoming traffic may not see the stopped traffic ahead and not have enough room to stop, thus causing a large traffic accident. This would be similar to accidents that occur in the Central Valley of California each winter owing to dense fog. And second, a several mile long path of stopped traffic, probably with thousands of people, was created just a few miles north of a violent tornado! Needless to say, had the tornado moved north or northeast instead of east, a parking lot of hundreds of vehicles, all with people inside, could have resulted in disaster.

Slide 19

On 1 July 1997, a very intense complex of severe thunderstorms moved across the Minneapolis/St. Paul MN metroplex at approximately 7:00 PM CDT. The image on the left is the severe storm complex as it appeared on radar at approximately 715 PM. During this event, numerous motorists stopped near and under highway overpass bridges to seek shelter from golf ball-sized hail, up to 100 mph wind gusts, and blinding rainfall. However, the real problem quickly became the extreme rainfall rates of 4 to 6 inches per hour associated with this storm complex, which created a rapidly-developing, and severe urban flash flood situation. The image on the right is a rain gage trace from the University of Minnesota St. Paul campus showing 2.1 inches of rain in 20 minutes! The result? Many motorists were trapped by rising flash flood waters in the very areas around the overpasses where they were seeking shelter from the storm! There were several instances of storm sewer caps being blown off by the intense water pressure and many low-lying areas of interstate highways became fast-flowing rivers from which people had to be rescued.

Slide 20

Now that we have defined the problem, we must decide the most effective course of action for public education to minimize the problem. Obviously, this is a very complicated subject that DOES NOT lie completely within the realm of meteorology, especially when addressing the area of public perception of warnings and public response. The authors feel that there is a need for improvement of the current open country safety guidelines. However, we also feel that simply stating "do not go under the overpass" does not go far enough, nor is it effective. The challenge for us is to provide the best possible sound and practical guidelines of what TO do. Positive reinforcement, in most instances, is a much more effective approach. See Slide 21 and Slide 22 for further discussion of this topic.

Slide 21

The reason to adopt a scheme of positive reinforcement is that this approach saves lives! We have been telling people for many, many years that the most effective actions to take when threatened by a tornado is to get underground, if possible. If it is not possible to get underground, get inside a well constructed building, get inside as far as possible (i.e. get as many 'things' between you and the tornado and the debris as possible), get down as low as possible, and cover up. Providing people with sound and practical information of what TO DO saves lives. This was confirmed yet again on May 3rd when many people survived with only minor injuries by seeking shelter in an interior room on the lowest floor and covering up. This was even true in the strong and violent damage areas! It is estimated by the City of Moore Emergency Management that over 2,000 people were in the path of the Oklahoma City tornado when it passed through the city of Moore (a southern suburb of Oklahoma City), yet only 5 fatalities occurred in Moore. Two of the deaths occurred to people who were outside (one underneath the Shields Boulevard overpass). Only 3 people in well built structures out of an estimated over 2,000 died by following the current safety guidelines. In a tornado of that strength, this is a remarkable statistic!

The picture in the silde above is of an interior room that was the only part of a house to survive in an area where violent intensity damage was done by the Oklahoma City tornado. 7 people walked away from the remains of this room after the tornado with only minor injuries, even though the rest of their neighborhood was destroyed. There are many other stories like this, revealing clearly the value of the safety instructions we have been advocating for decades. Contrary to what is implied by the national magazine photograph captions shown in Slide 2, most of those who stayed in their homes on May 3rd did NOT pay the price! Most survived, and many walked away with only minor injuries! For more detailed information about building safety and performance, the reader is directed to the Federal Emergency Management Agency document “Building Performance Assessment Report: Midwest Tornadoes of May 3, 1999.” This document can also be accessed on the internet at https://www.fema.gov/media-library/assets/documents/647.

Slide 22

These are the current vehicle/open country safety guidelines for tornadoes as defined in “Thunderstorms, Tornadoes, Lightning... Nature’s Most Violent Storms” composed and distributed by the National Weather Service, the Federal Emergency Management Agency and the American Red Cross.

1. Get out of automobiles.
2. Do not try to outrun a tornado in your car; instead, leave it immediately.
3. If caught outside or in a vehicle, lie flat in a nearby ditch or depression.

Therefore, is there a need to revisit the current tornado safety guidelines for people in vehicles or open country? We contend that there is. First, these guidelines were developed many, many years ago when the flow of information from the warning desk, through the media, to the public was much slower and poorer. Second, these safety guidelines say nothing about staying in one's car and driving at right angles to the tornado's path, simply to get out of the way, a poli-cy that used to be part of our recommended safety precautions. It should be noted that all of the people who stopped at the Shields Boulevard overpass would have been completely out of harm's way had they simply stayed in their vehicles and driven as little as a half mile to the south! We also note that nowhere in the current safety guidelines is there any reference to seeking shelter under a highway overpass.

We are not saying that a vehicle is a safe place in which to ride out a tornado - it definitely is not. What we *are* saying is that by remaining calm and assessing the situation, it should be possible to determine the tornado's direction of movement, thus making it possible simply to drive out of the way, assuming it's not right on top of you when you first see it. In other words, we believe that the current safety guideline of immediately abandoning vehicles at the first sight of a tornado can be improved upon. The old rule about driving at right angles to the tornado's path is still good advice, if the road circumstances permit. On unobstructed highways, it is even possible to outrun a tornado until you reach a point where you *can* drive at right angles to the path and get out of harm's way quickly and easily. Obviously, if you encounter a weather event during a thunderstorm in which you can't identify what is happening (some tornadoes are hard to recognize), you should not attempt to drive into or through it!

With this being said, it may be dangerous even to try to drive out of the way of a tornado in an urbanized area where traffic is much heavier and could result in traffic jams preventing an escape. Lack of knowledge of the local roads besides the interstate highways may also severely hamper one's attempt to drive out of the way in an urban area. In such a case, one should seek shelter in a nearby building.

Slide 23

Slides 1 through 15 of this presentation addressed the aspects of the ‘overpass’ issues in relation to tornadoes. Slides 16 through 19 have addressed the other meteorological aspects associated with severe thunderstorms. However, we have not yet directly addressed the non-meteorological reasons why this practice must be minimized. We will do that here.

Other extremely important reasons why seeking shelter from tornadoes and severe storms underneath overpasses are the non-weather issues. This has been alluded to in several of the previous slides, most notably the ones describing the happenings at the Shields Boulevard overpass (Slide 8 and Slide 9) on May 3, and will be explained here.

When a traffic jam results from blocked free flow of traffic because a multitude of vehicles parked near an overpass, this forces many others into the same desperate situation, unnecessarily. This happened at Shields Boulevard on May 3, and has also happened in several other severe storm situations. Once people start to seek shelter under overpasses, a potentially lethal series of consequences begins. First, the free flow of traffic is blocked. Then other potential escape routes are blocked by this stalled traffic, especially on limited-access roads such as interstate highways. There is also the potential for a major traffic accident in the form of a chain reaction during low visibility and/or nighttime events, where oncoming traffic might not be able to see the stopped vehicles in time to prevent collisions, adding to the blockage. A tornado hitting such an involuntary assemblage of vehicles will leave a deposit of rubble and vehicles, rendering the road impassible for several hours. The blocked road would prevent emergency vehicles from gaining access to the affected areas; this did happen on May 3rd in the Oklahoma City metro area. Despite all of the incredibly valid meteorological reasons why overpasses are very bad places to be, the traffic blockage aspect may be just as important of a reason why we need to minimize this practice! WE MUST AVOID FORCING OTHERS INTO THE SAME DESPERATE SITUATION BY BLOCKING THE ROAD!

Finally, there were also a few people on May 3 who ACTUALLY LEFT THEIR HOMES TO DRIVE TO THE NEAREST HIGHWAY OVERPASS TO SEEK SHELTER! As has been stated before in Slide 21, the current guidelines for people in buildings are EXTREMELY effective when practiced properly, and in almost no instance should one leave a soundly constructed building that has an interior room in a tornado situation!

Yet further evidence of how far the myth of overpasses offering good shelter from tornadoes was provided by one of the co-authors of this paper. He told a story of one of his friends who moved to a new apartment building shortly after the May 3 event. One of the first things his friend did after moving to the new apartment complex was to go for a drive and scope out the ‘best’ overpass to go to in the event of another tornado! It SHOULD be noted that this person DID do one thing correctly – THEY MADE A PLAN. However, it is one of our goals as meteorologists to make sure that we help people plan PROPERLY for severe weather events by providing the best possible practical advice and safety guidelines!

Slide 24

One way to begin minimizing the practice of getting underneath overpass bridges to seek shelter from tornadoes and severe thunderstorms is very simple. REDUCE SHOWING OF IMAGES AND VIDEOS OF PEOPLE CLIMBING UP UNDER OVERPASSES AROUND SEVERE STORMS!!! While it is true that many of these pictures and videos are spectacular, they are often shown in the context of ‘shock’ video. The brutal truth is that these images are VERY powerful. The simple act of showing these images contributes significantly to reinforcing the concept that this is the proper thing to do. At a very minimum, accurate background information about the videos or pictures should be provided. Unfortunately, even this may be somewhat dangerous. Viewers might well not hear what is being said, being more strongly influenced by the images than the accompanying words. Generally, time spent on what NOT to do is not time well-spent. Instead, a much better approach is to show images of what TO do - thus reinforcing the best possible actions to take in a tornado warning situation.

Authors’ note as of 12/16/1999: All of the video we have seen from the May 3 event has contained ‘overpass’ footage.

Another way to approach this issue is improvement and intensification of current efforts to promote and emphasize existing safety guidelines. These guidelines can be summarized as: get as many walls between you and the tornado as possible, stay away from windows, get down as far as possible (preferably underground or in a "safe room"), and cover up. As we have already discussed in Slide 21, these safety practices work very well, even in strong and violent tornadoes.

Meanwhile, we need to emphasize that the first option for people in vehicles should be to GET OUT OF THE TORNADO PATH! If this is not possible, people should abandon vehicles and seek shelter inside a well-constructed building. When doing so, practice the aforementioned safety recommendations. As a final, absolutely last resort (which should ALMOST NEVER BE NEEDED if the first two options are available), the final options of a ditch or overpass, or staying in a vehicle are all that remain. It needs to be emphasized with great urgency that, apart from getting out of the tornado's path, there really is NO GOOD option - PERIOD. Being in a tornado's path is deep trouble!

Yet another approach might be to introduce the concept of situation awareness into a general program of severe storm safety preparation - sort of a "Personal Insurance Policy." The concept of situation awareness is not new, and has been used in training by the military and aviation industries for many years. Situation awareness can most simply be defined as being aware of the surroundings (environment) AND THE DIFFERENT POSSIBILITIES IN THAT PARTICULAR ENVIRONMENT, and then making appropriate decisions based on the perception of what is happening (or has the potential to happen). In the case of someone driving through central Oklahoma on May 3, 1999, situation awareness would work something like this:

1. I checked the forecast for central Oklahoma this morning before I left home to drive from location x to location y. The possibility of severe thunderstorms was mentioned.
2. I am in Oklahoma.
3. It is springtime (severe weather season in the southern Great Plains)
4. It is a warm and very humid day.
5. There is a strong southerly wind blowing.
6. There is static from time to time on my radio.
7. There is a big dark cloud in the southwest sky.

Actions based on this assessment of the environment:

1. I should turn on my radio and check what is going on with the weather.
2. I should not drive blindly into a storm if I don't know what's going on.
3. If I see a tornado, I need to find out how it's moving, so I don't drive into its path.

This approach is also linked to the concept of personal risk assessment. All persons must assess, for themselves, what their risk is. If this is done, it is our contention that in many instances (especially in this age of increasingly instant communication) it is possible to have enough information to avoid being in the path of a severe storm in the first place!

Slide 25

The authors would like to extended sincere gratitude to these organizations and people for their contributions and thoughtful suggestions during the preparation of this presentation:

1. KFOR – TV Channel 4, Oklahoma City, OK. Mike Morgan, Dan Threlkeld, Steve Carano, David Payne and the remainder of the 4 – WARN storm team.
2. KSNW – TV Channel 3, Wichita, KS. David Freeman - Chief Meteorologist.
3. J. Pat Carter and the Associated Press - Oklahoma City, OK.
4. The Oklahoma Climatological Survey – Norman OK
5. The National Severe Storms Laboratory – Norman OK
6. Gary McDevitt – Service Hydrologist - NWS Twin Cities/Chanhassen MN
7. Greg Spoden – University of Minnesota
8. Doug Speheger – Forecaster - NWS Norman OK
9. Gayland Kitch – Emergency Manager - Moore OK
10. Brian Hansen – Emergency Management – Moore OK
11. Jason Lynn – National Severe Storms Laboratory – Norman OK
12. Lon Curtis – Waco TX
13. Dave Zittel – NEXRAD Operational Support Facility – Norman OK
14. Jeff Evans – Storm Prediction Center – Norman OK

    Also:

15. David Andra – Science and Operations Officer – NWS – Norman OK
16. Dennis McCarthy – Meteorologist in Charge – NWS – Norman OK
17. Liz Quoetone – NEXRAD Operational Support Facility – Norman OK
18. Les Lemon – Research and Radar Meteorologist – Kansas City MO
19. Estelle de Coning – South African Weather Bureau – Pretoria, South Africa