The El Reno tornado of 2013, on radar

The El Reno tornado of 31 May 2013 was an EF3 tornado in Canadian County, Oklahoma, that grew to 2.6 miles wide — the widest on record — while mobile Doppler radars measured winds near 300 mph inside it. It killed eight people, all in vehicles, including three storm researchers.

The tornadoes on radar guide describes the signatures a forecaster reads on an operational radar. El Reno shows the distance between those signatures and the wind field beneath them, because research radars scanning every few seconds sampled it at close range.

The record

The figures below come from the National Weather Service office in Norman, which surveyed the tornado, and the NWS service assessment on the May 2013 Oklahoma tornadoes.

UTC times are converted from the Norman event page; the office’s database lists the end two minutes later. The same supercell produced other tornadoes; Bluestein et al. (2018, Monthly Weather Review) count nine.

A path that would not hold still

NWS Norman’s survey describes a tornado that changed direction and speed repeatedly. It formed northeast of the intersection of Heaston and Reuter roads and moved southeast at 20 to 25 mph, turned east at 6:09 while expanding and speeding up to 30 to 40 mph, and crossed U.S. Highway 81 at 6:19. It then turned abruptly north while accelerating to more than 50 mph, reached its greatest size and intensity near Radio and Reuter roads, and slowed as it approached Interstate 40. Between 6:27 and 6:35 it made a loop just north of the interstate at less than 10 mph, then moved east and dissipated.

Vortices inside the vortex

El Reno was a multiple-vortex tornado: smaller secondary vortices orbiting inside a very large parent circulation. The research radars showed how extreme those were.

• Wurman, Kosiba, Robinson and Marshall (2014), in the Bulletin of the American Meteorological Society, used Doppler on Wheels data to track subvortices moving in trochoidal-like patterns at ground-relative speeds from 0 to 79 m s⁻¹, which they describe as the fastest ever documented. That upper figure is about 177 mph by conversion.
• Bluestein, Thiem, Snyder and Houser (2018) tracked 24 secondary vortices with the University of Oklahoma’s RaXPol radar, which updated a single elevation every 2 seconds. Most long-lived vortices formed near the radius of maximum wind in the left-rear quadrant of the parent tornado and dissipated closer to its center; some lasted a minute or more. One translated about 700 m in 9 seconds.
• Wakimoto et al. (2016), combining an aerial damage survey with two mobile radars, matched a suction vortex seen on video to both a radar couplet and a swath mark in a field.

What the mobile radars measured

The service assessment reports that the Doppler on Wheels sampled winds slightly above 250 mph about 375 feet above ground, and RaXPol sampled winds in excess of 290 mph about 500 feet above ground. Wurman et al. report more than 115 m s⁻¹ at 114 m above ground in one subvortex, with peak speeds of 130–150 m s⁻¹ implied. Wakimoto et al. resolved ground-relative winds above 90 m s⁻¹ near several chaser fatality locations, which they note exceeds the minimum speed for EF5 damage.

A Doppler radar measures only the component of motion along its beam, averaged over a sample volume, above the ground. Snyder and Bluestein (2014, Weather and Forecasting) used this tornado, with radar-relative velocities exceeding 135 m s⁻¹ “in rural areas essentially devoid of structures,” to argue that comparing radar velocities with the EF scale requires understanding what the radar data represent. Bluestein et al. (2018) add that their very highest estimates, 136.5 and 140 m s⁻¹, carry limited confidence because of ambiguity in manual dealiasing, while a 135 m s⁻¹ measurement had only one reasonable solution. They attribute much of the ground-relative extreme to the vortices’ fast translation.

Why the rating is EF3

The EF scale is a damage scale. The Storm Prediction Center’s table places EF3 at 136–165 mph and EF5 at over 200 mph, but those speeds are estimates inferred from damage indicators, not measurements.

According to the service assessment, the survey found only EF3 damage. NWS Norman, after consulting NSSL and the University of Oklahoma, assigned a preliminary EF5 on the strength of the mobile radar winds, citing at least six earlier tornadoes whose ratings had been raised the same way. NWS policy requires a damage-based rating. In a memo dated 6 June 2013, Louis Uccellini wrote that “EF ratings are determined by observed damage rather than measured wind because we have no consistent way to measure wind speed for every tornado that occurs.” The Norman event page now states that surveyors “could not find any damage that would support a rating higher than EF-3.”

Wurman et al. give a physical reason for the gap: few well-built structures in the path, and the most intense winds confined to small, fast subvortices, “resulting in only subsecond gusts.”

What the operational radar showed

The nearest WSR-88D, KTLX at Twin Lakes near Oklahoma City, was roughly 65 km from the tornado (our calculation from approximate positions). An NWS training page gives the average WSR-88D half-power beamwidth as 0.95°, about 1 nautical mile across at 62 nautical miles. At 65 km that is a beam roughly 1 km wide, with the 0.5° beam center about 800 m above the radar under standard refraction (both calculations). RaXPol, at 5 to 7.5 km range for most of its measurements, had an effective azimuthal resolution of about 125–200 m (Bluestein et al. 2018).

So KTLX could depict the large parent circulation, which the service assessment says “was evident on radar,” but not individual subvortices. Time resolution was the other limit. Kuster, Heinselman and Austin (2015, Weather and Forecasting), working with the forecaster who warned on the storm, compared KTLX with the experimental phased-array radar in Norman, whose volumes arrived at 70-second intervals. The rapid scans helped most with sudden mesocyclone intensification and “abrupt, short-term changes in tornado motion.” For the rotating updraft behind those changes, see the mesocyclone.

The deaths

All eight people killed by the tornado were in vehicles. Per the service assessment, four were storm chasing: three experienced members of the TWISTEX research group, and an amateur chaser who had left his nearby home. None was operating under a NOAA-sponsored project. The other four were motorists, two of them parked along I-40.

The assessment records chasers’ reports of the storm’s “visually-deceiving nature” and states that “many chasers did not realize they were in, or as close to, the large circulation.” During rapid intensification the forward speed rose from about 35 to more than 50 mph, combined with a sharp left turn and intense inflow winds. A chaser analysis reproduced in the assessment compares the visible width of the condensation funnels with the much larger circulation width derived from RaXPol. What the event showed about observation is how far the visible tornado could understate the circulation, and how quickly its motion could change between one look and the next.

WeatherStory Radar reads full-resolution Level II from all 159 WSR-88D radars, including velocity and correlation coefficient, straight from NOAA’s public archive.