A Twist to Tuesday’s Severe Weather: Tornadoes Missing

Tornado chasers scanned the skies fruitlessly on Tuesday, while residents of the Southern Plains breathed huge sighs of relief, as a bumper crop of severe thunderstorms produced buckets of large hail and high wind while spinning up only a handful of twisters. By mid-morning Wednesday, NOAA’s Storm Prediction Center (SPC) had received only five tornado reports: two in Texas, and one each in Kansas, Missouri, and Indiana. All of these tornadoes were relatively minor, with no serious damage or injury reported. High wind and large hail were far more plentiful, with more than 200 reports of each. Most of the hail was 1” to 2” in diameter, but several reports of 3” hail (larger than baseball size) came in from Marshall County in northeast Kansas. Winds gusted to 87 mph around 10:30 pm as a fierce squall line tore through Copan, Oklahoma, northeast of Tulsa.


Figure 1. Preliminary reports of severe weather received by the NOAA/NWS Storm Prediction Center for the period from 12Z (8:00 am EDT) Thursday, April 26, 2016, through 12Z Wednesday. Image credit: NOAA/SPC.


Figure 2. Lightning strikes along Interstate 70 near Junction City, Kansas, late Tuesday, April 26, 2016. Image credit: AP Photo/Orlin Wagner.


Figure 3. Susan Goodwyn holds large hail deposited by thunderstorms that swept through Wichita, KS, on Tuesday afternoon, April 26, 2016. Image credit: Susan Goodwyn/SMGPhotos via AP.


What happened to the tornadoes?
Tuesday’s outcome was far better than many residents had feared that morning, when forecasters were calling for widespread severe storms and the potential for strong tornadoes. This shortfall in tornado production wasn’t a total surprise. As we noted in Tuesday’s post, upper-level winds were not ideal for a major tornado outbreak, but it seemed plausible that adequate wind shear would accompany the extreme instability in at least a few locations that would be difficult to pinpoint well in advance. Along these lines, the large tornado watch issued for parts of Texas and Oklahoma was a PDS watch--meaning a “particularly dangerous situation,” with “several tornadoes and a few intense tornadoes likely.”

As it turned out, the upper-level winds were far from optimal during the afternoon, and by the time wind shear had improved, a squall line had formed, making it hard for any isolated supercells to develop and rotate. Thus, it was a day with high-end potential for strong tornadoes, but relatively low confidence as to how many there might be. NOAA’s midday outlook on Tuesday bore this out, with relatively low probabilities for tornadoes (maximum 10%) but much higher odds for severe hail (maximum 45%).

SPC had noted Tuesday’s potential for severe weather, including tornadoes, as far back as last Thursday, April 20. So was the forecast a bust? That depends on your perspective, and how you interpret forecaster guidance. For all time periods beyond 24 hours, the SPC convective outlooks (explained here) indicate only two things:

--the categorical risk of severe weather (from marginal to high)
--the numerical odds of getting any type of severe weather within 25 miles of any spot on the map

“Severe weather” means any or all of these:
--a tornado
--winds of at least 50 knots (58 mph), or lesser winds that inflict damage
--hail at least 1” in diameter

For Days 1, 2, and 3, the outlooks also include a crosshatched area where severe weather might be “significant”, meaning:

--a tornado inflicting at least EF2 damage
--winds of at least 65 knots (75 mph)
--hail at least 2” in diameter

With these definitions in hand, we can see that Tuesday’s outlook was far from a total bust. Figure 3 (below) shows how the outlooks issued at various lead times stacked up against the final outcome (top row, against the midday outlook issued at 12:30 pm EST Tuesday). The tornado count was undeniably low--and two of the day’s five reports were entirely outside the probabilistic risk area, associated with a bow echo across the Midwest--yet the tornado odds were relatively low to begin with, maxing out at just 10%. The hail outlook appears to verify quite well over the Southern Plains, given that about half of the 45% max-threat area (purple) experienced large hail within 25 miles. The Ohio Valley bow echo did produce more large hail than expected.


Figure 3. The progression of guidance issued by the NOAA/NWS Storm Prediction Center leading up to Tuesday, April 26, 2016. At bottom left of each panel is the lead time: Day 7 corresponds to a forecast issued six days prior (in this case, Wednesday, April 20). Day 4-8 outlooks include only the probability of any type of severe weather within 25 square miles of any point. Days 3, 2, and 1 include subjective categorization of the total risk (marginal, slight, enhanced, moderate, or high). Along the top are the Day 1 outlooks issued at 1630Z (12:30 pm EDT) Tuesday for the probability of any severe weather (left), tornadoes (center), and severe hail (right). Crosshatching indicates the forecasted possibility of significant tornadoes (EF2 or stronger) or very large hail (at least 2” in diameter). Symbols on the Day 1 outlooks along the top show actual reports from Tuesday for each category. Image credit: NOAA/NWS/SPC, courtesy James Correia (OU/CIMMS).

The question of lead time
The week-long build-up to Tuesday’s severe weather led to some soul-searching in the meteorological community: what is the value, and potential downside, of giving people five or six days’ notice that an outbreak may occur? An AP analysis by Seth Borenstein and Kelly Kissel, published on Saturday, covers the issues well. WXGeeks host and WU contributing blogger Marshall Shepherd (University of Georgia) also explored the topic thoughtfully in a Forbes essay last Thursday. (Shepherd also has some great perspective on yesterday's presumed forecast bust.]

There is some evidence that specific tornado warnings could be counterproductive if issued well in advance (such as two hours), as people may put themselves at increased risk rather than taking immediate shelter. More general outlooks, issued days ahead of time, may be a different matter. Research is still scant in this area, but long-lead outlooks could give emergency managers and response agencies extra time to raise awareness and assemble resources. Veteran atmospheric researchers such as Lance Bosart (University at Albany, State University of New York) point to the scientific accomplishments that have made such extended outlooks possible. In an email to colleagues, Bosart said:

“The achievement of being able to recognize the potential that a high-impact severe weather event will occur and where it will likely occur a week in advance is testimony to how far numerical weather prediction has advanced, how well new convection-allowing ensemble models have been integrated into the forecast process, how much probabilistic thinking has been incorporated into forecasts of high-impact severe weather events, and how well the dedicated professionals at the SPC have been able to take full advantage of these advances to produce operationally informative and useful guidance well in advance of expected severe weather outbreaks.”


Figure 4. As of mid-morning Wednesday, SPC’s convective outlooks show a slight risk of severe weather on each of Days 1, 2, and 3 (Wednesday through Friday, April 27-29, 2016).

It’s not over yet: more severe weather in the cards
With upper-level troughiness persisting in the West, and plenty of low-level moisture at hand, severe storms are a continuing possibility into this weekend. Wednesday’s threat straddles the Mississippi Valley from eastern Missouri to southern Louisiana. The day should yield mostly garden-variety severe weather, with the tornado risk quite low. More storms are possible over Texas and Oklahoma late Thursday night, with activity building on Friday as the upper-level low recharges and a warm front pushes north. Moderately strong wind shear and very high instability suggest the potential for a tornado threat worth monitoring on Friday across parts of Oklahoma and north Texas. The active pattern shows signs of continuing off and on as we roll into the first week of May and the peak of U.S. tornado season.

Bob Henson