The El Reno tornado from 5/31/13 has officially been rated as an EF3; OUN officially submitted the Storm Data events for May 31st today. I know that OUN felt very comfortable with the original EF5 rating, so I am left to deduce that the "decree" came from up high. Downgrading the El Reno tornado back to EF3 despite high-quality, near-ground measurements of radial, quasi-horizontal winds easily exceeding 250 mph from a rapid-scan radar (i.e., RaXPol) and corroborating observations from a DoW seems like a step back to me. More bluntly, I find it ridiculous and scientifically disingenuous. The TTU "Recommendation for an Enhanced Fujita Scale (EF-Scale)" dated June 2004 notes on page 14 that "the technology of portable Doppler radar should also be a part of the EF scale process, either as a direct measurement, when available, or as a means of validating the wind speeds estimated by the experts."
I don’t know the final implementation of the EF-Scale by the NWS, but the verbiage above is pretty clear. If observed data from 5/31/13 are not to be used in the EF scale determination process, then I assume this means that all other tornadoes that have used radar observations (e.g., the tornado in Rozel on 5/18/13, the Bennington tornado a week after that, the 5/19/2013 Clearwater, KS, tornado, the Goshen County tornado from 6/5/09, and others) will be downgraded as well to remain consistent with the "rule" that only damage shall be used to rate tornadoes. This should also be extended to any rating that used non-traditional damage indicators since they, by definition, aren’t a part of the EF scale. If we say that El Reno was an EF3 tornado, that means it had winds of 135-165 mph when observations <50 m AGL are nearly double that. This is logically inconsistent. Essentially, it was an EF3 with 250-300+ mph winds. Adjusting the radar observations for a fair apples-to-apples comparison (e.g., 3 s wind gust) still yields winds significantly above 200 mph.
Many have noted that the EF-Scale is a damage scale. This is partly true – damage is assessed using Degree of Damage (DoD) levels for a series of Damage Indicators (DIs). However, as I'm sure we know, the reason why we rely upon damage assessment is largely because that’s the best way we have to estimate the strength of the winds within a tornado. For the vast majority of tornadoes that occur within the U.S. each year, damage at near-ground level is the only data available from which to infer tornado intensity. If wind speeds were irrelevant to the EF-scale, there wouldn’t be a need to attach wind speed ranges to EF-scale level. The entire development of the EF scale was based upon a survey from experts to determine the wind speeds that would produce the DoDs for each DI. So, wind speed and the EF-scale are very intricately tied together.
I don't think many have noticed but the winds assigned to each DoD for each DI varied, sometimes very significantly, from one expert to the next when the EF scale was developed. This is particularly true at the higher DoDs and stronger wind speeds. For example, from Appendix B (“Elicitation of Damage Versus Wind Speed”) of the EF scale document, the wind speeds estimated from the EF-scale expert panel associated with DoD 8 for the “Automobile Service Building” DI range from 120 to 220 mph – a range that spans from EF1 to EF5! Many DoDs for DIs span multiple EF categories. I point this out not to comment on the expertise of the panel (they are, indeed, experts in wind damage assessment!) but to point out the tremendous uncertainty that exists with the wind speeds associated with the EF scale. This uncertainty is compounded by uncertainties in the field when one has to assign a DoD to a DI. So, not only do the wind speeds assigned to a DoD span a sometimes-significant range, but it’s sometimes very difficult to determine which DoD should be assigned to a piece of damage. This is all to point out that one should not assume that damage assessments are far from an optimal way to arrive at an estimated tornado intensity if one intends to infer wind speeds.
To add to the above rant: Of course, scientific instruments such as mobile radars and anemometers are not without error either. There are well-known sources of error when one examines radial velocity data from a radar in an attempt to determine the winds within a tornado. A few are listed as follows: the radial velocity data represent the reflectivity-weighted mean velocity of scatterers within the radar resolution volume (which likely are not moving the same direction and speed as the air within a tornado); the radar resolution volume is often much larger and with a mean center above the 3 m standard height used in the EF scale; the volume of the atmosphere illuminated by the radar isn’t known with complete accuracy since there are uncertainties with beam propagation and partial beam blockage that differentially affects the power within the illuminated volume. It’s also very important to make sure we’re doing an apples-to-apples comparison when comparing radar observations (which are nearly instantaneous) to the 3-s wind gust specification of the EF scale. In RaXPol’s case, since we have PPIs every 2 s when we’re scanning at 180 deg. / s, we can average scans together to get an Eularian-frame average wind.
For the record, I do think it IS very important that standards be in place to guide offices on HOW and WHEN to use radar and other observational data. As I noted, a radial velocity estimate is essentially instantaneous, and it is an estimate. There are all sorts of complications involved when relating these measurements to the 3 s wind gust at 3-10 m AGL that the EF scale uses. In this particular case, however, it seems extraordinarily unlikely that, given the observed radar measurements, winds were <200 mph at 3-10 m AGL and when averaged over 3 s. The 3 s gust likely wasn't in the 290-300+ mph range we measured with RaXPol, but I can't imagine it wasn't comfortable over 200 mph. In addition, the radial velocity estimates are averaged over a resolution volume (and weighted towards the larger objects), which means that winds on a scale smaller than the radar volume may be appreciably higher (and winds that likely are moving faster than larger debris given inertia with the debris).
Indeed, there needs to be a clear method by which which such data are considered so that observations (be they remote sensing or in situ) are used consistently and reasonably (e.g., can't use data from a radar that's 85 km away with minimum beam height of 1.5 km AGL). Just this May, we saw one office determine an EF scale rating by averaging the damage-based rating and the radar-based rating, whereas others set the EF rating based on radar data on account of a lack of DIs. I just am not sure why wind speeds are attached to the EF scale if objective wind speed measurements (after considering the myriad of uncertainties and complications involved in radial velocity estimates) are not to be used. This seems particularly true to me when considering the potential error with damage estimates in the first place (in terms of winds "assigned" to each DoD and DI and in terms of accurately assigning a DoD to the damage).
.
