|Above: This image made from video provided by KWTV-KOTV shows two funnel clouds associated with tornadic circulations near Crescent, Okla., on Monday, May 20, 2019. An intense storm system that weather forecasters labeled "particularly dangerous" swept through the Southern Plains Monday, spawning a few tornadoes that caused some damage and a deluge of rain but no reports of injuries. Image credit: KWTV-KOTV via AP.|
Sometimes the atmosphere humbles even the best forecasters. Monday’s meteorological setup in and around Oklahoma for severe weather, including the potential for violent tornadoes, prompted a burst of high-end outlooks that were startling even for Tornado Alley. Many school systems closed for the day, and shops and restaurants shuttered their doors in advance of the anticipated onslaught.
What emerged was a bona fide severe weather outbreak, but less fierce and extensive than the one many computer models and official outlooks had indicated was a strong possibility. Meteorologists who had stressed the extreme nature of the threat on Monday found themselves with a whole different communication challenge on Tuesday: how to explain a catastrophe that almost but didn’t quite happen.
The forecast wasn’t a total bust by any means. For the period from 7 am CDT Monday to 7 am Tuesday, the NOAA/NWS Storm Prediction Center (SPC) had logged at least 26 tornado reports by Tuesday afternoon, along with 80 reports of severe wind (4 of those higher-end “significant” reports), and 87 reports of severe hail (14 of those “significant”). The corridor of activity closely matched the moderate- and high-risk areas outlooked by SPC the night before (see below).
Oklahoma saw three noteworthy tornado events, apparently none of which produced serious injuries:
—One twister passed very near the town of Mangum in southwest Oklahoma, destroying several outbuildings and homes.
—A sequence of tornadoes, including two at one time, emerged from a compact supercell that passed just to the west of Oklahoma City.
—A strong tornado late Monday night, clearly evident on radar, carved out at least a mile-wide path and produced at least EF2 damage, according to a survey under way Tuesday afternoon by the NWS/Tulsa office.
Other tornadoes struck northwest Texas, and a mammoth 5.5”-diameter hailstone—one of the state’s largest on record—fell near the town of Wellington in the eastern Texas Panhandle.
As noteworthy as the day’s severe weather was, it wasn’t as extensive as what many short-range models had predicted, and it probably didn’t match up with many residents’ expectations of more widespread calamity. In the core high-risk area, SPC called for up to 45% odds that any of these significant events could occur within a 25-mile radius of any point: an EF2 tornado, hail of at least 2”, or wind gusts to at least 74 mph. We’ll have to wait for formal verification, but at first glance it appears the outbreak fell well short of reflecting these odds. Most of the tornadic storms were widely dispersed across the high-risk area, and there were no preliminary reports of any 2” hail in Oklahoma.
Most surprisingly, there were almost no supercell storms in the prime part of the warm sector across southwest Oklahoma, south of a warm front and well east of the cold front/dryline, where the air was most unstable and wind shear was amply supportive of high-end twisters.
There was another type of high risk on Monday–one issued by the NOAA/NWS Weather Prediction Center for excessive rain leading to flash flood risk. These heavy rains certainly materialized, albeit focused a bit further north than expected. Widespread flash flooding began Monday afternoon and was still under way on Tuesday. As of Tuesday afternoon, 345 river gauges across the contiguous U.S. were in flood stage, with several at record to near-record levels across northern Oklahoma and southern Kansas.
See the weather.com article for the latest on Tuesday’s ongoing severe weather and impacts from the outbreak. An enhanced risk was in place Tuesday afternoon for eastern Missouri and northeast Arkansas on Tuesday afternoon, with a slight risk bending back toward eastern Kansas.
A quick recap of rainfall and flood reports over the past 24 hours, compared to the WPC Excessive Rainfall Outlook at 15Z yesterday (after High Risk was expanded). pic.twitter.com/WTUt7nqhjz— Alex Lamers (@AlexJLamers) May 21, 2019
So what happened on Monday—or rather, what didn’t happen, and why? The most honest short answer is “we don’t know yet,” but there are already a few clues.
—The atmosphere was “capped” more than expected. A thin layer of warm air about two to three miles aloft flowed across the high-risk area from the deserts of northern Mexico and southwest Texas. Although just slightly warmer than expected, this layer—which moved into place just hours before the worst severe weather was expected—inhibited air parcels from rising to form thunderstorms, especially where there were no surface boundaries to force the issue.
3-6 hours difference is what saved Oklahoma today. That is beyond our current abilities to forecast in any longer-term forecast. It wasn’t even apparent until *maybe* 21z special OUN RAOB at the earliest.— Elizabeth Leitman (@WxLiz) May 21, 2019
—Overcast skies limited surface heating. Dense low clouds prevailed across most of the high risk area, which cut down on surface heating that might have helped more storms overcome the weak cap.
—Updrafts tended to be “skinny”. Some of the storms were quite narrow, which made them more vulnerable to disruption from the very strong wind shear.
—There was considerable haze and smoke in the air, some of it apparently related to smoke from wildfires in southern Mexico. Such aerosols, which are not incorporated in traditional weather models, may affect the amount of heat absorbed or reflected at various heights and/or the microscale cloud physics driving the storms. Modeling studies published in 2008 and 2015 found that smoke intrusions can actually intensify tornado-producing environments. However, smoke's effects on supercells might actually depend on the particulars of a given weather day, as suggested in a 2014 modeling study.
“The Mangum tornado is an example of what could have been,” said Rick Smith, warning coordination meteorologist at NWS/Norman, on Twitter. “Mobile radars had winds well within the violent category (I know, I know). What if we had had 5 of those yesterday in the warm sector? That was [what] I was imagining when I walked in the door yesterday. We weren't that far off.”
SPC warning coordination meteorologist Patrick Marsh also pondered the day's perplexities on Twitter. Marsh showed atmospheric profiles collected on Monday from Norman, OK, and in the catastrophic 2011 Super Outbreak from Birmingham, AL, and reflected on how similar they were. One saving grace Marsh mentioned to me was the southward-surging pool of rain-cooled air from the day's early storms, which spread out ahead of the dryline-generated storms that eventually moved through late Monday night. "I have a lot of hypotheses, but no answers," said Marsh.
One of the most reflective of the meteorologists weighing in on Tuesday was Roger Edwards, a longtime SPC forecaster who was on duty early Monday. In a thread on his personal Twitter account, Edwards offered wide-ranging thoughts on what happened and what we can learn from it.
1/9 Those rooting for long-track tornadoes & destruction yesterday probably were wondering what went "wrong" with the forecast. Those not as sadistic wondered what we missed w/overforecasting in a results-based sense. I have ideas from deep professional/scientific experience.— Roger Edwards (@SkyPixWeather) May 21, 2019
According to Edwards, “Dedicated scientific forecasters are like best NFL cornerbacks: shake off previous blown coverage, be better next play, but learn from it in film room. Data is our ‘film room’.”
Monday will offer plenty of material for review. For example, the wealth of computer model runs issued before the event can be compared with data from special efforts like the TORUS field campaign, to figure out what the models did and didn’t capture ahead of time.
Ultimately, Monday was a day with truly sobering high-end potential, so we can be grateful that supercells didn’t end up developing in the right places and at the right times to take full advantage of the situation. As Edwards put it, “Amazing parameter spaces don't produce [without] storms in them.”
In a new weather.com clip, Ari Salsalari and I discuss Monday's forecast and why it didn't quite pan out as expected.
Monday's #HighRisk #severe forecast wasn't as dire as expected. Meteorologists @AriWeather and @bhensonweather explain what happened and why it was or wasn't a "bust": https://t.co/durkL9acaS pic.twitter.com/fZudyh2klN— The Weather Channel (@weatherchannel) May 21, 2019
|Figure 1. Trees groan under the weight of snowfall in Boulder, Colorado, on Tuesday morning, May 21, 2019. Snow totals of 3" to 7" were widespread from Denver to Boulder. Some 10,000 Colorado customers lost power during the storm, according to Xcel Energy. Image credit: Richard Rotunno.|
Monday's ingredients may realign on Thursday
The cold upper low in the western U.S. that’s been helping to trigger this week’s severe weather—as well as a few surprisingly intense late-May snows over parts of the West and Upper Midwest—will twirl in place while slowly weakening this week. Another strong wave will emerge from the low on Thursday, and a moist tropical air mass is in place to resurge northward, so our next major ramp-up in severe weather is likely to take shape from Texas to Kansas. An enhanced-risk area is already in place for Thursday.