June 01, 2013

Run or shelter? What to do if a tornado threatens?

A fascinating and important discussion is occurring in the media the last few days:

What should one do if a tornado is approaching your location?

A few years ago there would be no such discussion.   Tornado warnings were either non-existent or provided with insufficient lead time for doing anything other than finding the safest location close at hand.   As recently as 1987, the average tornado warning time for the National Weather Service was around 4 minutes, while today it is about 15 minutes.  Moore, OK was warned about 35 minutes before the tornado hit the city.

The trend towards increasing lead times should continue for a number of reasons: 
  • The NWS is still learning how to use its recently upgraded (to dual-polarization) Doppler weather radars, radars that can now sense debris lofted by tornadoes.
  • Next-generation radars currently being tested (phase-array radars) will produce updated images every minute rather than the 5-6 minutes of the current NWS radars.
  • Improved modeling technology, and particularly high-resolution rapid-refresh forecast systems using ensemble-based data assimilation, should substantially improved forecasts up to 3 hr out.  
  • New observing technologies, including pressure  measurements from smart phones, better satellite information, and instrumented commuter planes will be of substantial benefit.   In a few years, most cars will have internet and will be able to provide pressure and other weather information.
  • Far better delivery systems.  Smartphones are extraordinary good weather warning delivery systems.  They even know where you are!
  • Extraordinary assets being put together by local TV stations, including chase helicopters and spotters providing live feeds.
  • Improved understanding of severe convection.   My colleagues at University of Oklahoma, the National Center for Atmospheric Research, the Storm Prediction Center, the National Severe Storms Lab, and other research entities are making important strides.
So warning times will get better and better.  Our ability to distribute timely information is getting better and better.  Prediction of severe convection is hugely improved compared to a few decades ago.

But significant number of people are still getting killed and injured...what are we going to do about it?

The traditional recommendation given to the public is to find shelter.  Find a tornado shelter if possible, and if not to go to the lowest floor of a building, preferably an interior room.  In most cases, this is still good advice.

But many folks don't have access to a tornado shelter, a below-ground tornado cellar, or a specially engineered safe room in their home.  Greater warning time makes it very tempting to flee, and quite frankly many in Moore, OK did just that--which perhaps explains why only about a half-dozen folks were killed when thousands of buildings were damaged or destroyed.    But Moore had a huge warning time, is a relatively low-density community, and the tornado did not hit at rush hour.  Looking at the damage to Moore (and most tornadoes) you can see why fleeing is so attractive:  the damage path is really quite narrow (1/2 -1 mile is wide for tornadoes).  You could WALK out of the damage path with 15 minute warning IF it were known.

Furthermore,  the tracks of most tornadoes are relatively straight  (see image), so there are not a lot of surprises.  But, of course, there are exceptions to such relatively straight paths--like the El Reno tornado in OK City on Friday that took an unexpected turn, resulting in the deaths and injuries of some experienced storm chasers.

The Oklahoma City tornadoes on Friday showed what happens when folks try fleeing in a more populated region:   the roads go into grid lock and people are trapped in their cars--like sitting ducksNearly all the people who got killed in Oklahoma City were in vehicles.  Most clearly would have survived if they had sheltered somewhere.

So what is the answer:  flee or shelter?  My take (and I am not an official voice for anyone) is that we need a little bit of both.

For folks in rural locations where traffic will not be a problem, flight from danger can make a lot of sense, particularly if one does not have a tornado-safe shelter..  And they should have two flight options:  (1)  move to a location of greatly reduced tornado risk (which we can't be 100% sure of) based on the latest forecast/radar guidance or (2) go to a nearby public tornado shelter.  Clearly, Oklahoma and neighboring states need to make the public investment in large safe locations...every public school should have one, for example.  And, one needs to bite the bullet and require all new construction to have a safe room or cellar.  We require special roof tie-downs in hurricane zones, so this is not a new concept.

To aid decision making, the NWS should develop a special tornado app for smartphones.   It would know your location and  direct you to the nearest shelter (perhaps even within walking distance) or tell you which way you should drive if you wished to take that option.   The app would have traffic information (via a feed from the google maps capability perhaps) that would be used to tell you the best routing to your choice.  And it would tell you whether you would be hopelessly trapped in traffic.   If this app was really sophisticated, it would route people to minimize overcrowding of shelters and traffic. And with knowledge of real-time rainfall, it could even keep folks away from low-lying areas that might flood.  There are already a number of smartphone apps dealing with severe weather...we need the next generation approach that provide guidance on how to find safety. 

In urban areas, the potential for grid lock increases rapidly and taking shelter should be the main approach.   Again, we need to insure there are a large numbers of safe locations.  Some will need to be built, but urban areas have more strong structures (steel frame, reinforced concrete buildings, parking garages) that are better able to take EF-4 or 5 winds.

In short, with improved tornado forecasts, an investment in shelter infrastructure, development of appropriate smartphone apps, and a more organized and rational approach to the tornado threat, I suspect we could reduce tornado injury and deaths by a huge proportion. But it will take a will do to do so and some public investment.


  1. Have you heard the rumors about Tim Samaras his son and Carl Young? Absolutely devastating.

  2. This comment has been removed by the author.

  3. GPS coupled with a modicum of storage can be a huge safety boost. Many years ago I saw a small box in a private airplane. It had only one button, which when pressed gave the pilot a list of the closest airports, sorted by distance, with the heading and distance to fly to get there. Very simple and obviously a real life saver in an emergency. If forecast tornado paths were broadcast in a defined data format, you could have that emergency box in every smart phone, as Cliff proposes. No rocket science required!

    Also, a background app which places a call to a data center automatically when atmospheric pressure drops very suddenly would improve public safety. The phone owner would not have to do anything for this to work. Again, no rocket science required.

  4. What do proper weather authorities (which especially includes, of course, Prof Mass) think of "storm chasers"? That is, do they really provide useful information (otherwise unavailable), or are they just testosterone fueled thrill seekers that are useful to the media circus? (Veteran Storm Chaser Among Those Killed In Oklahoma)

  5. I've been testing pressure sensors for paraglider use (altitude, climb rate) and observed that in a car on the freeway, vehicles in other lanes will momentarily but violently affect the pressure readings. Some pretty fancy filtering or outlier removal would be necessary for a vehicle pressure network. All the parts required are now mainstream mainly due to mobile phones : Cheap easy to use pressure sensors, cheap low power processors, cheap self contained GPRS modems. In the paraglider world there are now tic-tac box sized tracking units that send location and altitude (pressure) data via the cellular network every 10 seconds.

  6. I was surprised to learn how many buildings are constructed in Tronado Alley that lack either basements or dedicated tornado shelters. I grew up in south eastern Wisconsin, which gets more tornadoes than most people realize, and the farms there had tornado cellars made of concrete with a sturdy door on top.

    Until the coverage of the twister that took out the school, I had assumed that significant buildings in places like Oklahoma were built with tornado shelters. This should be changed. This should be considered part of the building code like, say, seismic codes on the West Coast.

  7. There are a lot of reasons why basements aren't practical in Tornado Alley.

    In some areas, the clay soils combined with high annual temperature variability make basements impractical. In others, there is a high water table and thus nowhere to put a basement.

    A more practical solution would be to put hurricane ties on roofs and permanent storm shutters on windows. This would raise the price of the typical house by about $5,000 for a typical single family dwelling. This would allow houses to survive a tornado of up to one category higher and also help reduce the damage to other houses from flying debris.

    Many municipalities have considered requiring such measures in the building code, but they're consistently defeated, because the political climate strongly opposes building codes (moreso now than any time I could remember when I lived there as a child).

    If one is of some means, it's possible to get prefab survival shelters that you can put in a garage or something. They're similar to the earthquake-proof desks you can buy on the West Coast--generally impervious to collapsing buildings, external debris, and even being thrown some distance. Those run $10k for deluxe models.

  8. I like the custom alerts for your location idea, but as someone who doesn't have a smartphone, I'd like a backwards-compatible option, as well.

    I just took a flight and recieved a text alert that my connection had changed gates. This type system could be employed to a greater degree (there are already alerts of this type).

    I realize the gps location options are very useful, but just want to keep in mind that the poor are often hit hard by tornadoes & hurricanes.

  9. Although I have lived in the Puget Sound area since 1996, I was born and raised in Oklahoma City. Almost all of my family still lives in south Oklahoma City, Moore, and Norman. Although the last two weeks have been very stressful, nobody in my family is hurt outside of some minor property damage. This of course is not true to other people in their neighborhoods.

    People ask why there were so many people on the roads especially when there was so much advance warning? I forwarded a Twitter post from the National Weather Service in Norman that was posted that morning stating about the probability of very severe weather between 4 - 6 pm. We were all discussing the forecast throughout the day. It was not a surprise to anyone. Many people said that their employers let them go home early because the forecast was so dire.


    You want to see how accurate these guys were? Look at the Twitter posts BEFORE the first tornado warning. Look at the times of the May 20 and May 31 posts.


    I digress.

    Everyone in my family knew the warnings. We were discussing them throughout the day. As the first tornado dropped outside of Union City, my mother and middle brother both got in their cars and drove south to get out of the way. This was about an hour before things got really bad in Oklahoma City.

    As for my youngest brother, aunts, uncles, cousins, nieces, and nephews, they stayed behind because they all had underground storm shelters available to them. I talked to many of them and they felt relatively safe. They were keeping an eye on the local news as they continuously updating the progress of storm.

    As the tornadoes entered the Metro Oklahoma City area, everyone that stayed behind took shelter. This is when the unforeseen problem occurred.

    Not only where there multiple tornado touchdowns, large hail, straight line winds of 60 to 95 mph, and severe lightning, there was torential rain. Many rainfall totals were broken that day. It was raining so hard that the ground was starting to flood. I knew of family members in three different storm shelters and the story was all the same. They were all taking in water at a very fast rate. Once all was said and done, the storm shelters were completely submerged in water. Some people actually had water damage in their homes.

    Everyone was stuck with a choice, stay in the shelter and possibly drown or leave the shelter and possibly get blown away by the tornado. As the storms got closer, almost all of the south side of Oklahoma City lost power. Other than small battery operated radios, the visual confirmation of where the storms were was gone.

    Many people started to panic. They heard on the radio where the storms were and they were close. Too close. Some got in their cars to try to out run them. The storms actually when through their neighborhoods.

    I am not saying that what they did was ok. The problem was that the May 20th tornado, a rare F5, was fresh in all of their minds. The pictures of houses being completely wiped off of their foudations gave them little desire to stay at home when the storm shelters flooded.

    The question we should all ask is how can we improve our preparation? As a whole, forecasting is improving but it still has a long way to go as it only gets you so far however. Ok we now know a major windstorm is going to hit the Puget Sound region. What then? Think about the Hanuakkah Eve storm. That was predicted and people STILL lost their minds because they were without power for over a week. People were still unprepared.

  10. The arguments against basements involving water tables and soil types are basically not meaningful. The primary factor regarding basements is cost. Having a basement in order to take shelter from a tornado is impractical - a tornado shelter (above or below ground) is much more cost-effective for this purpose.

    The decision to shelter in place or to flee is correctly understood as one of situation awareness. Under conditions of low traffic volume, fleeing is a practical alternative if you have no adequate shelter.


Please make sure your comments are civil. Name calling and personal attacks are not appropriate.

Winter-Like System in May

Tomorrow will not feel like late May, with a winter-like Pacific weather system making landfall, bringing lots of rain, wind, and cool tempe...