El Reno Tornado Downgraded to EF-3

[Tim Cermak]

Here's a great read on Chuck Doswell's thoughts regarding the recent downgrade from EF-5 to EF-3 of the El Reno tornado from May 31, 2013. What are your thoughts on the downgrade? Just curious to see what everyone else thinks about this.

Link to Doswell's post: cadiiitalk.blogspot.com/2013/08/operations-versus-science-case-of-storm.html

[Derek Mallia]

I saw this earlier this weak, rather disappointing that the NWS is choosing to ignore radar data. From what I've read these radar measurements seem relatively accurate so I am not entirely sure why the NWS would choose to ignore good data? It seems like a tornado's classification is a function of what it hits and the damage it causes to the object it hits. If a strong tornado happens to hit nothing (i.e middle of nowhere) odds are it would get a lower rating than a weaker tornado that happened to hit a lot of "stuff". Seems like flawed logic to me if you are trying to accurately measure the "intensity" of a tornado.

[Keith Sherburn]

There has been a pretty lengthy debate on Facebook regarding the downgrade. After thinking about it quite a bit, I feel like the best way to handle tornado ratings moving forward is to have two scales: 1) a destruction scale, based on damage and 2) a wind speed scale, based solely on *measured* wind speeds. The F- and EF-scales have essentially represented the "destruction scale", as the vast majority of tornadoes have been rated based on damage, rather than measured wind speeds. Going forward, we will have more measured wind speeds within tornadoes due to the increased prevalence of mobile radars around tornadoes.

Ultimately, no matter the scale, we will struggle to ever consistently estimate wind speeds based purely on damage. The relationship is not linear, though we try to treat it as such with the EF scale.

[kgriffin]

I've always been up in the air about this, and can remember getting into some debate (even at home!) over how this should be ranked. My opinion was to ignore the radar data for the integrity of the scale, and it looks like that's what has eventually been done here. Although I wouldn't fault the NWS for using the radar wind speeds, it is clearly subject to sampling bias. Not to say that Doppler winds haven't been used before, but that doesn't mean they necessarily belong either.

But overall, Keith's last point really summarizes things nicely in that we'll never get a true estimate from damage. Maybe we can talk in 10-20 years when phased array radars exist every couple of miles...

[Tony Lyza]

There are numerous reasons why I can't support the downgrade of the El Reno tornado. Doswell sums some of them up pretty well, and I sum them up as follows:

1) The consistency argument is already off the shelf. It's widely accepted that the standards for rating tornadoes in the '70s-'90s were different from present-day standards. And from the '50s-early '70s, the ratings had to be applied retroactively with only photographic evidence. Even when considering the present-day rating system, ratings standards may vary not only from office to office, but from surveyor to surveyor. For instance, the snapped power poles that one office may decide to rate as EF2 damage to go along with the expected wind value on the EF-scale may only garner an EF1 rating from a slightly more conservative surveyor/office (speaking from personal experience here), creating a dichotomy between what the two offices would consider as a "significant" tornado, a fairly important distinction in tornado climatology.

2) Sampling bias has already occurred numerous times in the past with regards to applications of EF-scale ratings to tornadoes. I'll share a few examples:

a) Philadelphia/Preston, MS (EF5; 27 April 2011): This tornado was rated EF5 purely on the basis of digging a couple enormous, 2-ft. deep trenches in the ground. The argument has been made, and it's a good point, that nothing like this was observed with the El Reno tornado, but in this instance, the ground that was scoured only had short grass covering it. I'm not a soil expert, but it does leave me to wonder if there was some property of that soil that made it fairly "easy" to scour.

b) Plainfield, IL (F5; 28 August 1990): This tornado was rated F5 by Fujita, who claimed in his paper summarizing the event that the damage "was among the worst [he had] ever seen." The basis for the F5 rating was not structural, however. It was corn damage. He compared corn damage in fields directly adjacent to homes that experienced F4 damage in the tornado path to the worst of the corn damage and concluded that the tornado must have been F5 intensity where the worst corn damage occurred. I strongly doubt any NWS office today would go out and rate a tornado EF5 in the same manner.

c) Valley Mills, TX (F5; 6 May 1973): Grazulis cites this tornado as having received an F5 rating from wind engineers for throwing a pickup truck 1/2 mile. I certainly don't believe we'll see a present-day tornado rated EF5 on the same premise.

I place these events in the same category as utilizing Doppler wind retrievals for rating tornadoes. Will the standards in either set of cases be able to be applied evenly in every case? Or, even more importantly, would they be, given the opportunity?

3) What little published research there is (admittedly not much) on the wind-height relationship in tornadoes seems to suggest that tornado winds peak quite low within the vortex. The results indicate that they may peak anywhere from 30-50 m (Wurman et al. 2013) to perhaps as low as even 3-5 m AGL (Kosiba and Wurman 2013). Though I'd be very hesitant to use these values at face value as the main chip in this argument, they certainly suggest that the wind-height relationship in a tornado is (as should be expected) not similar to that within a hurricane (an argument I've heard too many times).

I think Doswell covers a couple other points very well, and I'll leave those to his blog. I'll simply finish by saying that, as a scientist, it just seems wrong to ignore this data.