|Above: Among the structures damaged by a tornado that moved through Albany, GA, on Sunday, Jan. 22, 2017, was this gas station. Image credit: AP Photo/Branden Camp.|
A new study finds that the months of November to February are seeing an increase in average tornado activity, with a shift away from the Southern Plains and a ramp-up over the favored terrain of “Dixie Alley,” including Arkansas, Louisiana, Mississippi, Alabama, and Tennessee.
Published this month in Weather and Forecasting, the study was produced by Samuel Childs (Colorado State University), along with Russ Schumacher (CSU) and John Allen (Central Michigan University). The authors examined 4293 tornadoes reported between 1953 and 2015 during the cool-season months from November to February (NDJF).
The latest findings add to an expanding body of work suggesting that tornadoes are becoming a bit less likely in the classic Tornado Alley region and more common across the South and mid-South, especially during the cooler months. “Given that this region has one of the highest societal vulnerabilities in the country, an increase in cold-season tornado activity poses many risks and warrants investigation into potential meteorological influences,” the authors note.
|Figure 1. Regression coefficient for the number of tornadoes observed in each box from 1953 through 2015 during the months from November through February (NDJF). Areas of blue indicate a decreasing trend; red areas denote an increasing trend. The trends were statistically significant in eastern Oklahoma (decrease) and in western and central Tennessee (increase). Similar results were found when the box size was increased or decreased. Image credit: Figure 3 in Childs et al. (2018), “Cold-season Tornadoes: Climatological and Meteorological Insights,” Weather and Forecasting, courtesy American Meteorological Society.|
U.S. tornado reports have more than doubled since the 1950s. Most of that growth is attributed to increased awareness and to more people filing reports and collecting video. Tornadoes rated F1/EF1 or stronger have increased at the insignificant rate of just 1 twister per decade nationally, according to Childs and colleagues.
However, when looking only at the cool-season months of NDJF, the increase in reported tornadoes is five times greater, or about 5 twisters per decade. That’s enough to qualify as a statistically significant increase—albeit with great year-to-year variability. The total number of cool-season tornadoes F1/EF1 or stronger in the study period was as high as 154 (1973-74) and as small as 14 (1993-94).
An eastward shift in tornado prevalence?
The increase in cool-season tornadoes appears to be concentrated east of traditional Tornado Alley, as shown above in Figure 1. Similar results were found for the year as a whole in a study published in the Journal of Applied Meteorology and Climatology in 2016 and led by Ernest Agee (Purdue University), with Childs as a coauthor. That analysis compared two 30-year periods, 1954–83 and 1984–2013; it found a sharp decrease in summer tornadoes, centered on Oklahoma, and a jump in autumn tornadoes, centered on western Tennessee and Mississippi, with winter and especially spring showing a drop in tornado counts over the Southern Plains and a rise across the South.
“It is proposed that the new ‘heart of Tornado Alley’ as based on annual totals (and not on any particular season) is now located in central Tennessee/northern Alabama and not in eastern Oklahoma,” wrote Agee and colleagues.
|Figure 2. Change in the number of tornadoes of at least F1/EF1 strength reported during the period 1984-2013 versus 1954-83. Legend at right shows the change for each box of 2.5° longitude by 2.5° latitude. Areas of blue indicate a decreasing trend; red areas denote an increasing trend. Image credit: Agee et al. (2016), “Spatial Redistribution of U.S. Tornado Activity between 1954 and 2013,” Journal of Applied Meteorology and Climatology, courtesy American Meteorological Society.|
It’s not yet clear whether the apparent shift is simply the result of multi-decade variability or perhaps linked to long-term climate change. The second period of the Agee study (1984-2013) was considerably warmer than the first period, which might signal a tendency for tornado production to shift east in a warming climate.
Harold Brooks (National Severe Storms Laboratory), a world expert on tornado climatology, said the study’s methods appear to be solid. “We know from basic principles that warm winters give you more tornadoes, and warm summers give you fewer tornadoes. If we warm uniformly across the year, you’d expect the annual cycle to flatten out a bit,” Brooks told me.
As for the eastward shift, Brooks said, “we don’t have a good physical mechanism right now to explain this. Have we changed the total number of tornadoes, or are we rearranging them? I’d like to have a few thousand years’ worth of data to address those questions.”
Examining the meteorology behind cool-season (NDJF) tornadoes, Childs and colleagues found a familiar set of culprits at work during the busiest years. These include a tendency toward upper-level troughing in the West, ridging in the East, and an upper-level jet streak in between, across the Great Plains. Tornado-prone winters also featured up to 8% more moisture in lower levels of the atmosphere, with the western Gulf of Mexico tending to run warmer than average and the eastern Gulf cooler than average.
Vertical wind shear is notorious for enhancing the odds of tornado-producing supercell storms, even when the air is only moderately unstable. Interestingly, the authors didn’t find that wind shear and storm-relative helicity (SRH) were useful for distinguishing between the most- and least-tornadic winters. This could be because there is usually plenty of wind shear available during the winter months, so it’s the arrival of moisture that makes the difference (as opposed to summer, when moisture is usually plentiful but wind shear tends to be weak).
The La Niña effect
The number of tornadoes per cool season tends to peak every 3 to 7 years, which jibes well with the known relationship between the El Niño/Southern Oscillation (ENSO) and wintertime U.S. tornadoes. As we discussed in a post on March 9, tornadoes are most common during La Niña winters, when the polar jet stream is typically located farther to the north and warm, moist, unstable surface air has a better chance to sweep in from the Gulf of Mexico.
Consistent with earlier studies, the new paper found that more NDJF tornadoes occurred during La Niña winters: about 90, versus about 55 during neutral conditions and 62 during El Niño. A similar bump-up was found for winters when the Arctic Oscillation (AO) was mainly positive. In those years, the stratospheric polar vortex tends to be stronger and tighter, and the storm track typically stays further to the north. Neither ENSO nor the AO had an overall impact on the average tornado strength across a four-month cool season.
Can we use ENSO or the AO to say something about a winter's tornado risk months ahead of time? The AO's high variability makes this a challenge (though researchers are making inroads on long-range AO prediction). There's more immediate potential for ENSO, especially since we often have a good sense by summer or autumn of whether El Niño or La Niña will be in place by wintertime. Based on anticipated ENSO conditions, the authors assert, “One could envision a seasonal probabilistic forecast for NDJF tornadoes issued in summer or fall, which would serve to heighten public awareness and help city officials and emergency managers budget for resource allocation.”
Multiple risks to consider
In a companion paper now in early release at Weather, Climate and Society, Childs and Schumacher dive into the warning and communications challenges of cold-season tornadoes. They note that more than 900 people were killed by NDJF tornadoes between the months of November and February in the 1953-2015 period. Forested terrain, an increasing elderly population, and a high prevalence of mobile homes make the Southeast especially vulnerable. The long nights of winter only add to the threat that an approaching tornado may go unseen or unrecognized.