The End is In Sight for a Stunning Spell of Severe Weather

A much-needed break is finally on the horizon after a month-long siege of severe weather that ranks among the worst in modern U.S. records. A 12-day streak of daily rounds of severe weather culminated on Tuesday with a major long-track tornado that moved from the south outskirts of Lawrence, Kansas, then across the far western and northern parts of the Kansas City metro area. Dozens of homes were heavily damaged just outside the town of Linwood.

Massive amounts of debris—much of it apparently from a large greenhouse/nursery demolished near Linwood—covered the airfield at the Kansas City International Airport, forcing it to remain closed until cleanup operations were completed around midnight.

Debris picked up on the airfield that caused our Operations staff to close the airport because Foreign Object Damage to aircraft can cause catastrophe. Pots, foam, wall panels, plant ID tags over millions of square feet. Presumed from tornado damage 47 miles away in Linwood, KS. pic.twitter.com/oOhYTs7F6H

— Kansas City International Airport (@KCIAirport) May 29, 2019

The havoc this month has also included widespread flash floods across the Plains and Midwest and severe river flooding in Oklahoma and Arkansas (see our Tuesday post). Tuesday’s storms dumped 1.56” of rain at the Kansas City airport, pushing the city to an all-time May precipitation record of 12.81” (beating 12.75” from May 1995). Only two other months have been wetter in KC records going back to 1888.

The prolonged stretch of severe weather can be chalked up to an unusually cold and persistent upper low in the western U.S., a building heat dome in the southeast U.S., and a strong jet stream running south of the upper low and north of the heat dome. The same pattern has produced a raft of other noteworthy U.S. weather, from unusually late snowfalls in parts of the West to record May heat in the Southeast.

The recent rash of severe weather by the numbers

For the 12 days from 7 am CDT Friday, May 17, to 7 am Wednesday, May 29, the NOAA/NWS Storm Prediction Center (SPC) has catalogued a total of 352 preliminary tornado reports (filtered to remove duplicates) as of Wednesday morning, as shown below. 

Each calendar day since May 17 (based on midnight to midnight CST) has seen at least 16 twister reports, and at least six calendar days (shown in asterisks) have seen at least one EF3 tornado. Note that the preliminary SPC numbers below may change as additional reports come in and the paths of individual tornadoes are surveyed and clarified.

                  12Z-12Z          Calendar day  
May 17:           39                35                               
May 18:           28                22
May 19:             8                18
May 20:           35                24*
May 21:           41                50*
May 22:           43                42*
May 23:           18                21*
May 24:           16                17
May 25:           17                16
May 26:           33                33
May 27:           55                55*
May 28:           19                19*
Total:             352              352

Figure 1. The highest Day 1 risk category for severe weather issued by the NOAA/NWS Storm Prediction Center on each day this month through Wednesday, May 29. There are unlikely to be any enhanced-or-higher risk areas from Thursday into at least the coming weekend.

One outbreak, or more than that?

While this episode has seen significant severe weather each day, it’s an open question whether the entire span could be called a single tornado outbreak. SPC has no formal definition of what constitutes a tornado outbreak, warning coordination meteorologist Patrick Marsh told me. The AMS Glossary defines a tornado outbreak as “multiple tornado occurrences associated with a particular synoptic-scale system,” so the entire 12-day episode could arguably qualify based on that definition, given the persistence of the western upper low.

The event might also be seen as three separate outbreaks, according to a 2014 study by Robert (Jeff) Trapp (University of Illinois at Urbana-Champaign) titled “On the Significance of Multiple Consecutive Days of Tornado Activity.” Trapp defines an outbreak day as a calendar day with at least 20 tornado reports, again using midnight to midnight CST. By this definition, the outbreak days this month were May 17-18, May 20-23, and May 26-27.

When Trapp analyzed tornadoes during the period 1983-2012 in the SPC database, he found that F3/EF3+ tornadoes and outbreak days were more likely to occur in strings of at least three days than in one- or two-day spans. Interestingly, F3/EF3 twisters were more likely to occur near the end of a multiday outbreak period than near the beginning.

Looking more broadly, the 12-day total of 352 tornadoes is roughly on par with the May 2003 outbreak analyzed by Thomas Hamill and colleagues. They found that more than a dozen tornadoes struck each day from May 3 to 11, leading to a 9-day total of 361 tornadoes. Only three other comparable outbreaks occurred in the prior 88 years, they found. “An analysis of tornado statistics and environmental conditions indicates that extended outbreaks of this character occur roughly every 10 to 100 years,” they concluded.

As reported by weather.com’s Jon Erdman, the past 30 days are the first 30-day period to generate more than 500 tornado reports since April 2011, which included the catastrophic Super Outbreak in the Southeast.

Just how "active" have the last 30 days been for tornadoes? We're currently sitting at 500 filtered *eyewitness tornado reports* during this time period.

Only four periods in the official database ever exceed 500 *observed* tornadoes in 30 days: 2003, 2004, 2008, and 2011. pic.twitter.com/rV2KTC3Gmz

— Patrick Marsh (@pmarshwx) May 28, 2019

At the same time, Erdman noted, no single day in the past month has generated more than 400 reports of tornadoes, severe wind, or severe hail, a total that typically occurs one to five times each year. “Instead, it's been just a steady stream of severe weather, often over the same, relatively confined area, with an elevated number of tornado reports each day,” Erdman said.

Signs of progress: The human impact of this month’s severe weather

One big difference between this year’s outbreak and that in May 2003 is the human toll. In May 2003, the nine-day extended outbreak led to 41 deaths and 642 injuries, with some 2300 homes and businesses destroyed. The tornadoes in April 2011—the last month-long period that saw as many tornadoes as the past 30 days—caused 363 deaths, according to SPC.

Amazingly, only seven deaths have been directly attributed to this month’s tornado swarm as of Wednesday morning, with a few other deaths related to flooding and other storm hazards. A man in Celina, Ohio, was killed late Monday when an EF3 tornado flung a tree onto his home as he slept, and two people in manufactured homes were killed late Saturday when an unusual EF3 tornado struck El Reno, Oklahoma (see below). After midnight on May 22, a woman was killed at a farmstead struck by a tornado near Adair, Iowa, and three people died later that day when a tornado hit Golden City, Missouri. Several dozen tornado-related injuries have been reported over the last few days, though it appears only a small fraction of those have been critical or life-threatening.

The lack of violent tornadoes (EF4s and EF5s) in the latest siege is one reason why the death toll has been so low. The 2003 multiday outbreak included seven F4 and F5 tornadoes, which are comparable to EF4s and EF5s in the current rating system. Update: Not a single tornado since May 17 had been classified as EF4 or stronger until Wednesday, when a preliminary NWS survey showed that Tuesday's 31-mile-long track passing near Lawrence and Linwood included EF4 damage.

Tornado climatology expert Harold Brooks (National Severe Storms Laboratory) points out that the most violent tornadoes tend to be far more deadly than their weaker counterparts. In the SPC database going back to 1950, all but six of the 59 tornadoes rated F5 or EF5 produced fatalities, including every one of the F5s/EF5s since 1977 (a total of 13 tornadoes). In contrast, only about 60% of EF4s led to fatalities, and the percentage drops even more as you get to EF3 and weaker twisters.

As a rule, the worst tornadoes this month were well warned by local NWS offices, and the severe weather in general was amply signaled by convective outlooks and severe weather watches issued by SPC. These alerts were widely disseminated via broadcast and social media, factors that have surely helped reduce the death toll as well.

Figure 2. A semi-truck sits among the rubble after a tornado struck the American Budget Value Inn May 26, 2019 in El Reno, Oklahoma. At least two people were killed in this Oklahoma City suburb after a tornado barreled through, destroying much of the motel, a trailer park and a car dealership on the night of May 25. (Photo by J Pat Carter/Getty Images)

The El Reno tornado of 2019: A forecaster’s nightmare

The EF3 tornado that killed two people in El Reno, Oklahoma, on Saturday night, May 25, was an especially potent example of a type of twister that bedevils meteorologists. Tornadoes that form along squall lines—or what are technically known as quasi-linear convective systems (QLCSs)—are notoriously difficult to warn for.

The strongest U.S. tornadoes tend to develop on the south sides of supercell thunderstorms. In these cases, strong circulations may be evident on radar a half hour or more before a tornado forms. Precipitation wrapping around the circulation often forms a distinct hook-like echo. A single supercell may produce tornadoes for hours in cyclic fashion, one after the other. All these provide ample clues to assist with early warning. A tornado warning was issued for most of the Dayton area on Monday night roughly 30 minutes before an intense tornadic supercell reached the north side of the metro area.

In contrast, the front edge of a QLCS may appear very linear on radar, until a kink develops in the line, in which case a tornado may form within minutes. The saving grace with QLCS tornadoes is that they tend to be relatively small and brief, and their damage rarely exceeds the EF2 range. The El Reno QLCS tornado developed right at the south end of El Reno and lasted only about four minutes. Its compact damage path—about 225 feet wide and 2.2 miles long—began less than a mile away from the Skyview Mobile Home Park and the adjacent American Budget Value Inn. Estimated peak winds of 140-145 mph were solidly in the EF3 range. The first photo in the embedded tweet below shows how the tornado hit the southeast corner of the mobile home park and the south and east wings of the hotel.

My goodness. First images from the El Reno tornado damage via SkyNews 9 and @jimintheair. The damage path is extremely narrow and not more than 1/4 to 1/2 long but boy is it devastating. Search and rescue still ongoing. #okwx @NEWS9 pic.twitter.com/y3kxdQdGT3

— Matt Mahler (@themahler) May 26, 2019

The tornado’s rapid spin-up and its timing (10:28-10:32 pm CDT on the Saturday night of a holiday weekend) made it even more challenging to warn residents. A tornado warning was issued for the El Reno area at 10:26 pm, just two minutes before the twister developed. Central Oklahoma had been under a severe thunderstorm watch rather than a tornado watch.

In a typical year, about 5-10% of all U.S. tornadoes (excluding EF0) are rated EF3 or stronger. In most years, that equates to about 20 to 40 EF3 tornadoes, as shown in NOAA graphics. In turn, only a small percentage of those EF3+ tornadoes—perhaps as little as 3%, based on a multi-year climatology put together in 2012 by SPC’s Bryan Smith and colleagues—are associated with QLCSs. Given the typical small size of QLCS tornadoes, it’s especially striking and unfortunate that the El Reno twister happened to pass right over over a particularly vulnerable spot.

Figure 3. The severe weather outlook for Wednesday, May 29, updated at 12:30 pm EDT by NOAA’s Storm Prediction Center (SPC). An enhanced risk of severe storms capable of producing large hail, damaging winds and tornadoes are predicted for this afternoon into tonight along a wide swath of both the Midwest and Mid-Atlantic. This risk level is one notch down from Tuesday’s “moderate risk” scenario. Image credit: SPC.

One more day of significant severe weather

Tornadoes are possible in northeast Texas and far southeast Oklahoma on Wednesday, as one more piece of the relentless upper low in the western U.S. moves through the Plains. Wind shear is relatively weak, though, and "the tornado threat does not appear to be particularly high," said SPC in its midday update (see Figure 3). Other intense thunderstorms are expected to trek from Pennsylvania toward New Jersey, perhaps affecting the I-95 corridor from Washington to New York. Large hail, torrential rains, and high winds are the main threats, but a tornado can't be ruled out.

Beyond Wednesday, we can expect increasing upper ridging and a distinct and welcome shift toward a much quieter mode of severe weather. As is often the case in early June, there may be one or more days with scattered severe storms on the High Plains, but the extended-range models show no sign of any return to the strong western trough and the resulting cascade of severe weather we've seen in late May.

Dr. Jeff Masters contributed to this post. Thanks also to Jon Erdman for additional analysis.