Above: Joey Roush walks through debris at his mother's home in Beauregard, Alabama, on March 4, 2019, one day after the home was destroyed in a tornado. (Tami Chappell/AFP via Getty Images)
Nearly a year has passed since the historic tornado outbreak over the Southeast on March 3, 2019. At the time, the background weather pattern seemed, at least to me, to favor a “typical” springtime outbreak of severe weather over the region. As it turned out, however, there was a rather subtle surface-pressure feature that tipped the scales toward the deadly tornado outbreak.
My goal in writing this blog post is to identify this surface-pressure feature and to share how its impact, although initially unanticipated by forecasters, ultimately played a nuanced but pivotal role in the severity of the tornado outbreak on March 3, 2019. To do this, I will also introduce a forecasting term seldom used in forecast discussions that lay the groundwork for outbreaks of severe weather … the “geometry of severe storms.”
The story begins on the morning of March 3, 2019, when I took stock of the Storm Prediction Center’s convective outlook issued at 8 am EST (valid for the 24-hour period from 8 am March 3 to 7 am March 4). Of course, the area designated for an enhanced risk of severe thunderstorms over the Southeast wasn’t easy to miss. To gain further insight into forecasting terms such as “enhanced risk,” check out the colorful chart titled, “Understanding Severe Thunderstorm Risk Categories,” on this SPC Products page. The chart indicates that all five risk categories are based on the predicted duration, coverage, and intensity of severe thunderstorms.
With my attention now focused on the Southeast, I decided to get a better sense for the overall “big picture.” As I hinted earlier, the predicted weather pattern that I observed on the 12 UTC GFS model run seemed, at first glance, to be rather typical of spring outbreaks of severe weather over the Southeast; it just did not set off any alarms signaling that a historic tornado outbreak was just hours away. To get a sense for what I saw that morning, check out the 6-hour GFS surface and 500-mb forecasts (valid at 1 pm EST).
Specifically, a strong cold front (right panel) and its supporting 500-mb short-wave trough (left panel) were slated to move through the Southeast during midday and afternoon. For me, the screaming forecast message was a squall line that would likely develop along the strong cold front. As a result, I expected damaging straight-line winds to pose the primary risk of severe weather.
Shifting my focus away from the cold front toward the warm sector, I recognized that there would be ample vertical wind shear and modestly high convective available potential energy (CAPE), the latter being driven by invading warm and moist air from the Gulf of Mexico. By way of background, CAPE is a measure of the overall instability in the troposphere, and, thus, a measure of the potential strength of updrafts in thunderstorms. Strong vertical wind shear between the ground and six kilometers favors thunderstorms with rotating updrafts. Moreover, strong vertical shear, for sake of storm longevity, keeps the rotating updraft separated from storm downdrafts.
The next step in the forecasting process was to get a qualitative sense for any lifting of surface air in the warm sector. At this point, there weren’t any obvious and conclusive signs of scattered pockets of relatively weak lift that would favor discrete severe storms (compared to the strong lift everywhere along the cold front that would generate a solid line of severe storms). As a result, there was some uncertainty about predicting discrete, severe thunderstorms in the warm sector. All I could say for certain was that discrete supercells, which are thunderstorms with rotating updrafts, were possible in the warm sector during the afternoon. Thus, I could not dismiss the possibility of supercellular tornadoes in the warm sector.
The bottom line was that the weather pattern just didn’t seem like a classic set-up for a widespread and highly dangerous tornado outbreak. Yes, I believed that damaging winds posed the greater threat, and so I decided to get a more quantitative sense from the 13 UTC (8 A.M. EST) convective outlook in order to confirm my forecast.
As it turned out, my initial impressions seemed to be on target, because the probabilities for damaging straight-line winds expected to within 25 miles of any given point in the enhanced-risk area were roughly three times greater than the risk of tornadoes. Moreover, the 10% probability for tornadoes within 25 miles of any given point was consistent with a relatively small coverage of tornadoes.
Let me put it to you another way. During my career as a weather forecaster, I observed multiple springtime weather patterns that, at first glance, appeared to be much more threatening than the weather pattern I saw on the morning of March 3, 2019. Based on my experience, enough of these seemingly more-threatening weather patterns subsequently underperformed in regard to the severity of expected tornado outbreaks. Such forecasting experiences made a lasting impression on me, and so, after I turned off my laptop on the morning of March 3, I began the day’s routines with little expectation that a devastating tornado outbreak was about to happen in the Southeast.
Later in the day, while watching the evening news, I was stunned to learn just how prolific the tornado outbreak turned out to be. And my jaw just about dropped to the floor when I learned about the long-tracked EF4 tornado that ripped through Lee County in eastern Alabama, killing 23 people and injuring 90. The death toll was more than twice the number of tornado fatalities in the entire United States for all of 2018. This killer tornado was the first violent tornado (EF4 or EF5) in the United States in almost two years, and the nation’s deadliest tornado since the EF5 twister that devastated Moore, OK, on May 20, 2013.
On radar, the supercell that spawned the Lee County EF4 tornado was quite impressive. The 2016 UTC (2:16 pm CST) image of base reflectivity (below) from the NEXRAD Doppler radar at Maxwell Air Force Base near Montgomery, Alabama, revealed a classic hook echo, indicating a well-developed rotating updraft (mesocyclone).
The high reflectivity (above) visible within the hook echo (the rather small, red, circular area) is the signature of airborne debris, which is confirmed by the 2016 UTC image of correlation coefficients (CC) below.
The small area of dark blue on the CC image above translates to a relatively low correlation between sizes and shapes of the radar targets, which is consistent with the chaos of airborne debris caused by a tornado. Finally, the corresponding 2016 UTC image of storm-relative velocities indicated a tornado vortex signature (TVS), meaning that the radar actually sampled violent winds in the EF-4 tornado.
Impressive storm, don’t you agree? Recall that, on the morning of March 3, I conceded that supercells spawning a few tornadoes in the warm sector were possible during the afternoon. But this? A long-lived supercell producing a long-track EF4 tornado? And dozens of other tornadoes? All from a weather pattern that seemed, at least to me, to be a set-up for a “typical” outbreak of severe weather?
Something was amiss, and I was determined to turn over every stone, delve deeper, and embark on what was for me, a compelling quest to learn more. Yes, there had to be a theory out there that would help me to explain the difference between what initially seemed like a “typical” springtime outbreak of severe weather in the Southeast and, ultimately, a historic tornado outbreak.
I began my quest by reading Mesoscale Discussion #144 issued by the Storm Prediction Center at 12:28 p.m. on March 3. As of late morning, eastern Alabama remained in the enhanced-risk area, and the tornado probabilities were virtually unchanged. However, in this discussion, SPC forecaster Bryan Smith connected the already impressive supercell in Lee County to a persistent, well-defined isallobaric fall center, which is a relative maximum in negative pressure tendencies—in other words, the location where surface pressure were decreasing most rapidly.
To get a better sense of negative pressure tendencies and their role in the Lee County tornado outbreak, check out, below, SPC’s 18 UTC (noon CST) mesoanalysis of surface pressure, surface wind barbs, and surface pressure tendencies. The dashed blue contours correspond to pressure falls, labeled in millibars per two hours (a common unit of pressure tendency). Solid red contours indicate pressure rises.
The figure above revealed a rather broad isallobaric fall center (the closed, dashed-blue contour of negative pressure tendencies … surface pressure decreasing with time) over eastern Alabama and western Georgia. I’ll have more to say about isallobaric fall centers in just a moment.
This pressure fall center first appeared over east-central Alabama during the late morning, and then tracked eastward into Georgia by early afternoon in concert with an east-northeastward moving, sub-synoptic-scale low. The pressure tendencies within this fall center at this time were negative 3 to negative 4 millibars per two hours. That’s not chicken feed, folks!
As it turned out, the center of large negative pressure tendencies over eastern Alabama and western Georgia made the severe weather set-up much more threatening by producing a strengthening isallobaric wind—a component of the surface wind that helped to accelerate the flow of warm, moist and unstable air from the Gulf of Mexico toward the isallobaric fall center. Moreover, this strengthening component of the isallobaric wind also increased the low-level vertical wind shear, an unfortunate double whammy that tipped the scales toward the historic tornado outbreak. Let’s take a closer look at the isallobaric wind and the pivotal role it played on March 3, 2019.
The isallobaric wind is a component of the surface wind that occurs in response to centers of maximum pressure falls with time (negative pressure tendencies) and maximum pressure rises with time (positive pressure tendencies). For the record, the isallobaric wind blows directly toward centers of maximum pressure falls. Why would this happen? The even-keeled atmosphere always tries to compensate for irregularities. For example, air converging around a surface low-pressure system (an “irregularity”) is forced to rise in order to avoid major (and unrealistic) congestions of air near the ground. In much the same way, an isallobaric fall center qualifies as an “irregularity,” to which the orderly atmosphere responds by generating the isallobaric wind. In turn, the isallobaric wind helps to transport more air toward the fall center in order to keep the pressure from decreasing unrealistically quickly. In this way, the isallobaric wind is a tool that the atmosphere uses to compensate for the “irregularity” of an isallobaric fall center.
In the case of March 3, 2019, surface winds, aided by their isallobaric component, blew generally from the south, transporting warm and moist air (modestly high CAPE) inland from the Gulf of Mexico. Without the presence of the isallobaric fall center over eastern Alabama and western Georgia, surface winds would have blown pretty much exclusively from the southwest rather than from the south. In other words, surface winds over eastern Alabama and western Georgia “backed” in response to the isallobaric fall center.
Now I’m getting close to unveiling the geometry of this outbreak of severe weather. The key factor that came into play was the position of the isallobaric fall center relative to the area of modestly high CAPE (backed southerly winds transporting warm, very moist air inland from the Gulf of Mexico) and the relative position and movement of the discrete supercells in the warm sector.
That’s a mouthful, I realize. I’ll put it another way. The geometry of the tornado outbreak was such that discrete supercells in the warm sector were able to remain in the zone of backed southerly flow for a considerable length of time. Stated another way, the favorably moist and warm low-level environment moved in tandem with the axis of the isallobaric fall center as well as with the supercells themselves. This set-up allowed supercells to maximize their potential to spawn tornadoes because the storms spent a comparatively long time in the tornado-favorable environment.
Quantifying such warm-sector "geometry" is extremely difficult, and often easier in hindsight than in real time. Had I anticipated the geometry of these severe storms on the morning of March 3, I might have had a completely different attitude about the pending outbreak of severe weather. Yes, I learned this lesson the hard way.
A caveat: I can't say for certain, of course, that the "backing" of the low-level flow that occurred in response to the isallobaric fall center was directly tied to the very focused nature of the tornado outbreak. Nonetheless, this case illustrates how analyzing fields like pressure tendency can help to identify smaller, specific areas (within broader volatile environments) that are at a greater risk for severe weather. Such careful analysis can yield increased watch and warning lead times, and, ideally, save lives.
A closing thought to my professional colleagues and fellow forecasters: The backed southerly winds associated with the isallobaric fall center also enlarged low-level hodographs over eastern Alabama and western Georgia. In other words, wind profiles traced out longer and more curved paths (when plotted on a hodograph) than they otherwise would have. These enhanced hodographs likely increased the storm-relative helicity (SRH) that supercells in the warm sector ingested, thereby increasing the threat of tornadoes.
The bottom line here is that there is indeed more meteorology to be learned from this historic tornado outbreak. Perhaps I can write another blog in the future which discusses these concepts in greater detail. For now, however, your takeaway from this blog should be that the "geometry" component of any severe-weather event likely plays a prominent role in determining the event's intensity as well as its extent.
Here endeth the lesson.
Lee
