Rating Tornado Intensity Based on Moblie Radar

[Brian McKibben]

The Rozel, KS tornado on 5/18/2013 was rated an EF-4 based on measured winds from a mobile radar.

http://www.crh.noaa.gov/news/display_cmsstory.php?wfo=ddc&storyid=94702&source=0

"A total of five farms were damaged by this tornado, including one farm house that lost most of its roof and some of its exterior walls. The tornado intensity has been rated EF4 based on measured winds of 165 to 185 mph by a Doppler On Wheels. The maximum damage path width was 1100 yards."

I know the El Reno/Piedmont, OK EF-5 on May 24, 2011 was also rated based on a mobile radar.

I thought the EF scale was supposed to rate tornadoes based on damage? I almost feel like these tornadoes need to have an asterisk next to them (sorry Roger Maris). But, if we are going to suddenly change a tornadoes classified intensity based on a mobile radar, doesn't that skew the statistics?

Don't get me wrong, I honestly don't mind the use of radars to classify a tornadoes strength. In ways this shows the true strength of a tornado better than the spatial variability of hitting populated areas. But, what are the guidelines for doing so? How can we differentiate between damaged assessed and radar assessed? How far away does the mobile radar have to be, proximity to tornado of ~2miles, 4miles?

Finally, if we are going to classify some tornadoes based on a mobile radar, why can't we use a stationary radar to do the same. I know the 5/20/2012 Moore tornado passed within 3 miles of this radar because I was parked right next to it while it went through Moore.

https://maps.google.com/maps?q=moor...e=+&hnear=Moore,+Cleveland,+Oklahoma&t=h&z=19

Does anyone know who's radar this is?

There have to be several other instances of a tornado passing very close to a radar, so can we use that radar to classify them?

 

[Shawn Schuman]

There was a lengthy discussion of this in the thread for the El Reno tornado, and apparently a lot of folks aren't big fans of it. I think it's a very good thing that we're incorporating mobile radar data since in some cases, like in Rozel, a damage survey simply isn't able to give an accurate picture of the tornado. But like you, I think there needs to be clear-cut guidelines for using such data if there isn't already.

Another thing I'm a bit puzzled about is, why wasn't the 5/22/11 Canton Lake tornado rated EF4? DOW7 recorded winds of 80 m/s and NOXP recorded 85 m/s. I believe this was at 45m. I'm curious why this data apparently wasn't used but the El Reno data was. I'm not sure, but I think the El Reno measurement was at a higher altitude, something like 65m. I'm not really complaining because the folks making these decisions know a lot more than I do, but I've always been puzzled by that.

 

[Caleb Chevalier]

Brian, you took the exact words out of my mouth. I was about to start a similar discussion. I also think that incorporating in situ data would be a good step, but as of now, does the EF scale allow for this? As we all know, it is a damage scale, and I don't know if using a wind speed measurement is "proper procedure," if you will.

 

[Jeff Snyder]

Another thing I'm a bit puzzled about is, why wasn't the 5/22/11 Canton Lake tornado rated EF4? DOW7 recorded winds of 80 m/s and NOXP recorded 85 m/s. I believe this was at 45m. I'm curious why this data apparently wasn't used but the El Reno data was. I'm not sure, but I think the El Reno measurement was at a higher altitude, something like 65m. I'm not really complaining because the folks making these decisions know a lot more than I do, but I've always been puzzled by that.

We measured ~275 mph winds in the El Reno tornado (5/24/11) at a height of ~65-76 m above radar level, which is *way* above the EF4/EF5 threshold. Granted, there are many caveats one must use when discussing radial velocity data measured by mobile radars in light of the EF-scale. For example, the radial velocities almost never are made <10 m above the ground. On 5/24/11, for example, the beam height was 65-75 m above radar level at the range and elevation angle of the 125 m/s winds. It's just extremely difficult to get radar observations within 10 m of the ground unless the radar is essentially in the tornado (extremely short range) and ground clutter and beam blockage for a 0 degree elevation angle are absent. In addition, radar measures the reflectivity-weighted mean radial velocity of scatterers within the radar resolution volume. As such, in areas of very high centripetal accelerations, the radial velocity that's being measured (which is typically the velocity of the debris and not the air) can be quite different from the true radial velocity of the air. In the 5/24/11 case, though, there were non-traditional DIs that indicated that the tornado had extremely high wind speeds.

Personally, I think we should use whatever in-situ or remotely-sensed data we can that help fill in the gaps provided by the EF scale. Typically, one of the benefits of rating tornadoes, from a climatological standpoint at least, is to let one assess tornado frequency (spatially and temporally/seasonally) and associated tornado intensity with near-storm environmental parameters. It's well known that tornadoes can be underrated substantially in rural areas and in cases in which there are no appropriate damage indicators that can be used to assess an upper-bound to the estimate. Using additional observations (e.g., in-situ observations from anemometers like the El Reno mesonet site, high-resolution radar data near the ground, etc.) seems appropriate their use can more accurately allow one to estimate the intensity of the tornado. Heck, many of the DIs and DoDs are just estimates anyway, so even using only damage to infer tornado "intensity" can be error-prone even when there are high-quality DIs impacted. This isn't including additional complexities such as duration of tornadic winds (e.g., a tornado that churns over an area while moving very slowly can be reasonably expected to produce more damage than a tornado with the same winds that moves quickly over an area), the amount of debris loading, and the variability of the flow within a tornado.

One issue, though, when it comes to using radar data to estimate surface winds is that the vertical distribution of the horizontal wind (we're ignoring the vertical wind, which some have posited may have equally as strong as the horizontal wind near the corner flow region of tornadoes) is not well known (and probably varies as a function of time and tornado). As such, extrapolating winds from 70 m AGL to the 2-10 m AGL isn't straightforward and may be a significant source of error.Typically, it's easier to get mobile radar data near the ground (since the radars can be driven to a close distance from the tornado), and such data tend to be higher resolution (i.e., we aren't talking about gate-to-gate shear anymore, since the tornado is actually being resolved across 5-10+ radials).

 

[Andy Wehrle]

This discussion is particularly relevant given the events of the last two days. There are numerous very fortunate small towns in Kansas (Marysville, Lebanon, Ebson, Corning, Goff, Bennington and Culver to name a few) tonight, and several tornadoes with inferred violent potential that likely did not strike violent-type DIs at their peak intensity, but were well-sampled by either radar or in situ instruments such as those on the TIV and Dominator.

 

[Caleb Chevalier]

This discussion is particularly relevant given the events of the last two days. There are numerous very fortunate small towns in Kansas (Marysville, Lebanon, Ebson, Corning, Goff, Bennington and Culver to name a few) tonight, and several tornadoes with inferred violent potential that likely did not strike violent-type DIs at their peak intensity, but were well-sampled by either radar or in situ instruments such as those on the TIV and Dominator.

That was on my mind as well. NWS Hastings has, for now, surveyed the Lebanon, KS tornado as an EF3. The TIV measured wind speeds of 150-175 mph in that storm before the instruments failed. Because of some ambiguity in the EF scale, I think GID can choose to use that reading and call it an EF4, or ignore it and finish up their damage survey. There's really nothing as of now that states direct measurements must be taken into account; it's more of an option. NWS Dodge City took DOW readings into account, as mentioned in the original post. But GID might not.

 

[William Monfredo]

It's another example where bias enters the tornado database. Why? Because there's numerous locations and storms that do not have a radar standing by to provide speeds to upgrade tornadoes. Such radar or other information can be in the fine print or an extended comments section, IMO. Fujita was always interested wind speeds, true. But once the systematic processes of damage rating are compromised, the database becomes even more unreliable than it is already. I respect the people that need to file the reports at NWS, & I appreciate the research efforts of radar operators, but I'm on record from the beginning of this issue, do not casually interfere with the damage-rating processes by randomly tossing radar-derived winds into the mix.

 

[rdale]

I'd place radar estimates as MUCH more reliable than judging them based on a meteorologist's interpretation of building standards. I'm not sure I agree that using them will destroy the database. It implies that the 1967 tornado damage survey done by reading newspaper articles in 1975 was spot-on...

 

[Mike Johnston]

The tornado database will always be imperfect. Changes in population and technology over the years assures that. It's a work in progress, more like a history book written by many authors that grew up in different times and places with different tools at their disposal. If new technologies can bring better measurements, so be it.

 

[Lou Ruh]

Here's another case ... http://www.crh.noaa.gov/news/display_cmsstory.php?wfo=top&storyid=94990&source=0 ... the May 28th Bennington, KS stationary beast. Looks like it is becoming a common practice.

 

[rdale]

Read the May '13 BAMS for more information on the future of the EF-Scale. (Well, that was probably written pre-sequester )

 

[William Monfredo]

Looks like it is becoming a common practice.[/QUOTE]

At this point, we can count on one hand the number of times this has been done, out of thousands of tornadoes over the years. But... if we are going to go down this radar road, then institutionalize the process in an official manner, place an appropriately-spaced network around the country, but that's probably too expensive. The question becomes will you use the database for you research in graduate school, for a university or private company? If so, consistency is needed, not hodge-podge efforts.

 

[MClarkson]

I think all technology that increases the accuracy of the measurement should be used. One should simply keep in mind that the older tornado records have less reliable/fewer observations.

 

[rdale]

if we are going to go down this radar road, then institutionalize the process in an official manner, place an appropriately-spaced network around the country, but that's probably too expensive

Why? What harm is there in adding value when value exists, and not adding value when it doesn't?

The question becomes will you use the database for you research in graduate school, for a university or private company? If so, consistency is needed, not hodge-podge efforts.

I think you are unaware of the inconsistent hodge-podge efforts that are already part of the scale

http://www.bama.ua.edu/~jcsenkbeil/gy4570/doswell%20et%20al.pdf

Most everything in there pre-70's was rated simply by college students reading newspaper reports. If you think that information is of the highest standards, far more reliable than DOW readings, you might be mistaken.

In the current era ratings come from professional meteorologists, building engineers, TV meteorologists, Skywarn spotters, emergency managers, and probably a few more. There is no way that every person doing an EF-scale survey knows all he or she needs to know to file a 100% accurate report. The F-scale database is not completely reliable. Pretty good for most efforts, but not 100%.

I guess my real question is - what sort of research or private company needs completely accurate EF-scale reports?

 

[William Monfredo]

You have brought up several good points here. But, in a general sense, I don't want to see a climatology of violent tornadoes become a climatology of DOWs. Where will violent tornadoes show up? Where advanced radars roll out, and that's a very limited region, with a lot of researcher and spatial bias.

 

[BrandonHeadley]

I see a few potential problems with mixing damage reports and DOW readings. IMHO we should be using any and all methods to better understand the damage caused by tornadoes.

The first would be standardizing on when to use data from DOWs. Say one office decides to consider radar data only when it impacts rural areas and another decides to use it every time there is a tornado report. This could cause a skew in the data from a certain office. Also, if a DOW indicated a EF3, but the damage survey indicated EF4 damage then which do you go with? Let's say the damage survey found EF2 damage but a DOW indicated EF4 wind speeds who decides what it should be rated? I see the potential for the higher rated to be given every time which causes the data to be skewed toward more violent tornadoes.

Second, DOWs can't be everywhere. As someone already brought up Tornado Alley has a large concentration of them during peak time. This obviously makes sense monetarily and for the research they are doing. The more tornadoes they get in a shorter amount of time the more data they have. However, it may bias reports to more violent tornadoes in tornado alley vs other parts of the country.

Finally, I hear a lot of Media reports that there are more violent tornadoes now then __ years ago. I believe this is at least somewhat because of the change from F-scale to EF-scale. Additionally this can be attributed to the increase in research and data gathered during the storms. The data gathered isn't a bad thing at all and the more data we gather the better we understand what causes a tornado or doesn't cause them. I would urge people/media to be careful when comparing the violence of tornadoes from year to year.

Just my thoughts on the matter. I do think the NWS should look into some standards. Who knows maybe they already have them in place.

 

[rdale]

Brandon, if DOW shows EF4 winds and the damage did not match the DI listed in the EF Scale, it more likely means that the DIs are incorrect. I'll take an actual measurement over a hypothetical damage determination any day of the week,

 

[Brian McKibben]

I think if this trend continues, we might find that there are a lot more violent tornadoes than what the databases say. I wouldn't doubt both the Rozell and Bennington were violent. They sure did look like beast. I still wish we could see some radar data from the KFOR south Moore doppler from the May 20th tornado. It was 3 miles away from EF-5 damage, so no telling how strong the winds were as measured by radar.

 

[Jeff Duda]

I think if this trend continues, we might find that there are a lot more violent tornadoes than what the databases say. I wouldn't doubt both the Rozell and Bennington were violent. They sure did look like beast. I still wish we could see some radar data from the KFOR south Moore doppler from the May 20th tornado. It was 3 miles away from EF-5 damage, so no telling how strong the winds were as measured by radar.

Can't thinko f the reference right now (check SLS conference proceedings from 06-08 or EJSS) but I believe Wurman et al. already did such a study using the database of DOW-sampled tornadoes and found that EF1 and EF2 tornadoes were much more common than had been previously measured by just damage surveys. It had been thought that the tornado intensity distribution was more of an inverse exponential (i.e., of the form y = exp(-x)) so that the most common tornado rating was EF0. This new evidence suggests there is more of a gamma distribution (i.e., a bell curve skewed right - with a heavy right tail) to fit the intensity ratings to be consistent with EF1/EF2 tornadoes being more common.

 

[Shawn Schuman]

Can't thinko f the reference right now (check SLS conference proceedings from 06-08 or EJSS) but I believe Wurman et al. already did such a study using the database of DOW-sampled tornadoes and found that EF1 and EF2 tornadoes were much more common than had been previously measured by just damage surveys. It had been thought that the tornado intensity distribution was more of an inverse exponential (i.e., of the form y = exp(-x)) so that the most common tornado rating was EF0. This new evidence suggests there is more of a gamma distribution (i.e., a bell curve skewed right - with a heavy right tail) to fit the intensity ratings to be consistent with EF1/EF2 tornadoes being more common.

I think you're referring to:

Curtis R. Alexander and J. Wurman, 2008: Updated mobile radar climatology of supercell tornado structures and dynamics, Proceedings 24th Conference on Severe Local Storms, Savannah, GA, American Meteorological Society

 

[Jake Orosi]

I think all technology that increases the accuracy of the measurement should be used. One should simply keep in mind that the older tornado records have less reliable/fewer observations.

I tend to agree.

Tornadoes are typically EF-rated based on damage, true; but remember the whole point of that: the extent of the damage is used as a gauge to estimate wind speeds which are usually extremely local and therefore not measured by actual instruments that were not in the path of the tornado. Certain wind speeds are expected to cause a certain severity of damage. It's exactly like using the Beaufort scale when you don't have an anemometer.

I think that, wherever actual scientific instruments have a confirmed reading, we should absolutely defer to them when assigning the EF-rating of a tornado. And of course, where such data aren't available, we should continue to measure as best we can using observed structural damage. That's not bias; it's merely being as accurate as possible on a case-by-case basis. There's no such thing as tornado being "unfairly" rated "only" an EF-4 based on actually-measured wind speed for instance when in somebody's opinion it "should have been an EF-5", or vice-versa. It is a wind-speed classification; no more or less.

 

[rdale]

El Reno went from EF3 to EF5 based on mobile Doppler radar...

 

[James Gustina]

They also updated the width, 2.6 miles wide. Beats out Hallam by .1 for widest tornado ever.

 

[Brian McKibben]

Saw that too Rob. Also, it looks like it has overtaken the Hallam NE as the widest.

Gary England ‏@garyeOK3m
El Reno, OK tornado now rated EF5, Doppler measured 296 mph. Largest ever documented: 2.6 miles wide. gary

 

[Brian McKibben]

Moving at 30 mph at 2.6 miles wide means tornado winds for 5.2 minutes.