Friday, August 30, 2013

Extremely Odd Decision From the National Weather Service

The National Weather Service has downgraded the EF-5 El Reno Tornado of May 31 to EF-3. The ostensible reason is because (along a very sparsely populated path) the worst damage was EF-3 (138 to 167 mph). It was upgraded to EF-5 because of a close-up mobile radar reading of 290 mph winds. The NWS wants to "honor the intent" of the EF scale (according to one NWS meteorologist), thus the downgrade. I believe this is a poor, and dangerous, decision.

Some background, the 'Enhanced' Fujita Scale replaced the original Fujita tornado intensity scale for NWS purposes in 2007. It differs from the original by lowering the wind speeds for strong tornadoes. F-5 was 261 to 319 mph. EF-5 is 200 mph or greater.

We have two extreme wind readings from mobile radars in the State of Oklahoma: 318 mph in 1999 and 290 mph in the El Reno storm. Those readings are much more compatible with Ted's original than the EF. Plus, Ted's original explicitly allowed instrumental measurements (see image below) where I

have circled the wind speed, as opposed to damage, scale.

Unfortunately, this is not academic. If a nuclear power plant, for example, is built in the tornado belt, it needs to be about to withstand 300+mph gusts (worst case scenario). By intentionally putting the El Reno tornado into the database weaker than it really was, it could mislead non-meteorologists (i.e., architects and building engineers) into under-designing buildings. For example, this FEMA design document references only the EF-scale.

Given its better grasp of tornado's upper-end wind speeds and that using the best data is an impediment to the EF-Scale, I believe we need to go back to the original F-Scale.


  1. The "NWS wants to "honor the intent" of the EF scale?" That sounds good. Wind 'damage estimates' were later created to have a secondary method to confirm estimated wind velocities, based on later discovered ground damage data, in lieu of not having an upfront direct means to directly acquire local wind velocity measurements, as we do have today via DOW measurements. Here is the irony, not only was there one DOW XPOL with EF5 wind velocity data, confirmed by Dr. Howie bluestein, but also EF5 confirmed dual Doppler data from several other confirmed DOW sources, one source including Dr. Josh Wurman, CSWR, and no one recently thought that just maybe they should have waited ..until peer reviewed science data was published, before down grading the tornado based on building debris findings not located within the core flow region of the El Reno tornado?

    Oops, someone at NWS may now be be popping Xanex for making a controversial rush to judgement in 2013.

    On the plus side, since it was originally rated EF5, downgraded to EF3, and later confirmed in peer reviewed papers as EFX, then it will be easy in the future for NWS to say, after such highly regarded data formally hits the street, that after all due consideration of the facts studied and presented over time, it is of the esteemed opinion of weather service to reissue a conclusive finding that the El Reno tornado will in fact finally be rated EFX after all.

    The irony of the recent short sighted decision-making is the following, government agencies are looking for places to cut wasteful spending. Since the entire WSR-88D network is costly, and now deemed not useful by WS in determining tornado velocity, local governments can go back to the process of letting tornadoes hit, doing damage estimates, and determining EF scale. Really?

  2. Hi Mike,

    I understand your perspective, but if you'll bear with me, please allow me to explain why I disagree as briefly as possible.

    While tornado intensity rating has allowed for direct (and here I will include radar sampling for the sake of argument) measurement of tornadic winds, particularly as you note the original F-scale, the harsh reality has been that 99.999...% of all intensity determinations have been made using damage (~~1870's -present). Damage, while quite imperfectly, nearly continuously samples a tornado's surface winds, something no observational method can rival at the moment. As such, regardless of the availability, surface obs (particularly those necessitating unknown reduction factors to bring them to the same level as the historical obs) would do more harm to the historical record than good.

    The above said however, I am fully for including (and seeking to increase our ability to include) direct measurements (as defined above) into the official record. This would best be accomplished as a new field / comment in the archive specifically for direct intensity measurements. In time, my hope would then be that we could accomplish a sufficient bridge between damage and direct intensity measures that (and that observing technology would allow for) a smooth and physically consistent transition to a directly measured intensity scale that would not render the previous century of data useless.


  3. Two very thoughtful comments, thank you.

    Jacob, in the original F-scale, Ted used the following nomenclature,
    -- F-3 (capital F meant it was a measured wind speed)
    -- f-3 (lower case f meant it was estimated from damage)
    See the illustration in the posting.

    I don't understand why going back to his system would cause a discontinuity in the records since people would know which were measured and which were estimated.

    If you disagree, by all means chime in. I'd like to better understand that perspective.

  4. I greatly miss Dr. Fujita, who has probably rolled over in his grave several times by now (and I don't just mean over the El Reno tornado - but countless other issues with poorly performed "surveys" that lack thoroughness and consistency, to say the very least).

    As you'll probably recall, Dr. Fujita rated the Plainfield, IL tornado of 1990 an F-5 based largely on the damage patterns that he observed in a corn field. Yes, that's right - a corn field.

    In my opinion, we haven't seen a strong focus on attention to detail and thoroughness in storm damage surveys since his passing. The focus today seems to be on how quickly the survey can be completed, and how fast we can get the rating assigned.

    For example, aerial examinations aren't even part of the protocol anymore. Dr. Fujita would never complete a tornado survey without taking a look from up top. Today, if someone has aerial data, great, they'll look at it, but it's not part of standard operating procedure, and I think it should be, particularly in long track situations.

    I know "budget concerns" are likely to be blamed for the lack of aerial surveys now. Solution: there isn't a TV station with a helicopter in any market that wouldn't jump all over the chance to take a survey team member up following a long track tornado. There are also other resources for helicopters and airplanes (such as within the law enforcement community) that could contribute as well.

    I agree that the entire storm damage survey situation needs to be overhauled, but no one seems to want to take it on. Until we find the next Dr. Fujita (if there is or ever could be such a person), I don't see it going very far...

  5. Mike,

    I had utterly forgotten that distinction (kind of like gradually loosing old SA code). That would tend to the problem better than nothing, and with no notable discontinuity. I imagine though that you can understand that I am still partial to my solution. ;)

    Thank you for your kind reply.


  6. Mike, I certainly agree that this does set up a potentially dangerous situation in the minds of structural engineers downplaying these strong winds. It also seems that this would be a step backwards in using advancing technology to keep moving the science forward.

    As Jacob said in his comments, there are certainly merits to assignments based on damage (it is an indicator which has very little bias, perhaps unfortunately so), but the ability to sense tornadoes' winds accurately with radar has been something that has been researched and tested a lot over the past decade or so. It's been 14 years since the 318mph measurement of the May 3, 1999 tornado from a DOW and that number is now very highly cited. In the interim, we've conducted a huge research project in VORTEX2 which used many different types of radars and other remote sensing tools.

    I think it's time that remote sensing at least have some weight in the tornado-ranking system - be it the F(f) scale, the EF scale or something else.

  7. Mike,
    Doesn't pg. 18 of the EF-Scale allow for radar measurements to be incorporated into rating the intensity? I understand the implications regarding the continuity of the tornado record, but perhaps the EF-Scale just needs to be updated to address radar measurements with more than just a passing comment and an additional field be included for the radar measurements if available. The historical record already changed with the implementation of the EF-Scale, and at some point the EF-Scale will most likely be changed again, altering the record as well. Doesn't seem reasonable to resist needed updates to the EF-Scale just for the sake of the record, but consideration needs to be given with any changes as to how we can maintain some degree of continuity.

  8. @87. Thanks for the comment.

    You are correct about p. 18. The problem is that NWS policy forbids using measured winds speeds (regardless of source) in lieu of estimates from damage.

    By going back to the original Fujita Scale, we accomplish several things:

    1) The top value (319 mph) is far more useful than "more than 200 mph."

    2) The Rozell, Kansas, tornado would be rated:
    F-4 (measured winds) AND
    f-2 (estimate from damage)
    People doing a database search would be free do an "estimate-only" search and it would appear nothing has changed.

    3) Some much work was done with the original scale (Tom Grazulis "Significant Tornadoes", and others) we would not have to go back and re-rate the tornadoes prior to 2007 to accommodate changes in EF or some third index.

    What's not to like?