Twenty Years On: A Look Back at the Jarrell Tornado Catastrophe

On May 27, 1997, one of the most violent tornadoes in modern U.S. history produced close-to-unfathomable damage on the outskirts of Jarrell, TX, located about 40 miles north-northeast of Austin. There are only a few photos and videos of this monster, but it is a disaster well worth remembering. The Jarrell tornado broke many assumptions about twister behavior. It also left a grim toll, including 27 lives lost. Other than one serious injury, everyone affected by this tornado either came away relatively unscathed from an encounter with its outer edge, or was killed in its destructive core.           

Figure 1. Path of the southwestward-moving Jarrell, TX, tornado. Image credit: Adapted from the NWS Service Assessment “The Central Texas Tornadoes of May 27, 1997."

To get a sense of what a strange calamity this was, consider that the tornado moved toward the southwest along most of its 7.6-mile-long, 0.75-mile-wide path, roughly paralleling Interstate 35 and eventually angling to the west-southwest. Because of the prevailing west-to-east flow at upper levels, almost all intense U.S. tornadoes have at least some eastward component to their motion. Folks in tornado country are accustomed to keeping an eye toward the west or southwest—not the northeast—for potential threats.

What made this tornado so destructive was the blend of its raw strength and size with a fairly slow forward speed of only around 15 mph. This combination meant that some spots could have experienced tornadic winds for as long as three minutes.

In 1997, mobile Doppler radars were in their infancy, and none were deployed on the Jarrell storm. The lack of high-resolution Doppler data and the sheer level of damage makes it hard to infer just how strong the winds were. Based on its destruction, the tornado easily earned an F5 rating on the original Fujita Tornado Damage Scale, which corrresponded to estimated top 3-second gusts of 261 - 318 mph. The Enhanced Fujita Scale now tops out at EF5, corresponding to estimated peak gusts of more than 200 mph.

In an online summary, the NOAA/NWS Storm Prediction Center (SPC) tersely captured the level of destruction in Jarrell:

“This tornado blew some houses completely off the foundations and swept away the disintegrated remains. It also scoured asphalt from roads, killed and dismembered hundreds of cattle, stripped bark from trees and uprooted them, and bounced vehicles for up to half a mile from their parking places.”
 

Figure 2. A photo of the Jarrell, TX, tornado on May 27, 1997. Image credit: Lon Curtis, courtesy Adam Houston, University of Nebraska–Lincoln.

A weird, tragic day of severe weather

The environment that produced the Jarrell tornado—and 19 others that day across central Texas—was far from textbook-classic. A slow-moving cold front was draped almost parallel to a dryline, with just a few miles between them. Meanwhile, an upper-atmospheric impulse called a gravity wave, clearly visible on satellite, was pushed southwestward from a storm complex in Oklahoma and Arkansas early on May 27. This gravity wave appears to have helped kick off an early-afternoon storm complex in the vicinity of Waco, near the cold front-gust front intersection. From this point onward, new cells in the complex continued to “back-build” toward the southwest (against the upper-level flow) as the gust front continued plowing southwestward and as the cold front gradually caught up with the dryline from north to south. (To see the power of colliding boundaries, check out this impressive radar loop from just a few days ago of a cold front-dryline collision and explosive storm development near Lubbock, TX, in an environment with much stronger vertical wind shear.)

Ahead of the boundaries, the air was extremely muggy, and the instability was about as high as it gets. A research radiosonde launched at midday east of the dryline by the National Center for Atmospheric Research found a CAPE value of more than 7500 joules per kilogram. CAPE is the amount of energy an air parcel forced upward would have. Any value over 4000 J/kg is considered “extreme” by SPC; a storm going up in 7500 J/kg would have tremendously strong updrafts.

On the other hand, the weak surface winds beneath moderate westerly upper-level flow meant that vertical wind shear was quite low, as was helicity (the rotational part of the wind profile). There was no sign of the strong vertical wind shear and helicity that are almost always found in major tornado outbreaks. In fact, a 2007 Monthly Weather Review analysis of this outbreak in led by Adam Houston (University of Nebraska–Lincoln) concluded: “Comparison of this environment to other cases of back-building storms revealed that the observed wind shear was more in line with the vertical shear typically associated with nonsevere back-building storms.” 
The paper estimated that storm-relative helicity from the surface to 3 kilometers was only about 60 m2 s-2, compared to the 200-300 m2 s-2 typically found with significant tornadoes.

There are just a handful of comparable events where violent, long-lived tornadoes have developed in environments of modest wind shear and extreme instability. Among those cited in Houston’s 2007 paper:

—an F5 that leveled the tiny town of Jordan, Iowa, on June 13, 1976

—the seven Grand Island, Nebraska, tornadoes of June 2-3, 1980 (later made famous in the “Night of the Twisters” children’s book and made-for-TV film).

—an F5 that moved through Plainfield, Illinois, and neighboring areas on August 28, 1990, killing 29 people

Because there is so little large-scale vertical wind shear in these environments, it appears that tornado production is more dependent on very strong updrafts becoming juxtaposed with the more localized shear found along gust fronts, drylines, and/or cold fronts. The study by Houston and colleagues found that interactions between short segments of all three boundary types may have helped the storm near Jarrell to generate multiple tornadoes, with the tornadoes themselves forming along the southward-moving gust front.

Figure 3. A view of the Jarrell, TX, tornado early in its lifespan. Image credit: Scott Guest.

Despite the less-than-classic nature of the day, the extreme instability was enough to prompt SPC to issue a moderate risk of severe weather early on May 27. A tornado watch was issued at 12:54 pm CDT and a tornado warning for the Jarrell area at 3:30 pm CDT. Sadly, the act of taking shelter would not be enough to save more than two dozen people in Jarrell.

At its outset, between around 3:15 and 3:20 pm CDT, the Jarrell tornado was a pencil-thin rope-like twister, as evident in this mesmerizing video by Scott Guest recently posted to YouTube. Around 3:30 pm, the tornado rapidly transitioned into a much larger, multiple-vortex beast as it was approaching Jarrell. The tornado passed less than a mile northwest of the heart of town around 3:40 pm.

It’s difficult to overstate how completely the Double Creek Estates subdivision of Jarrell was demolished. Of the 38 homes, only slab foundations remained, with the homes reduced to bits. There were no basements, and all 27 deaths in Jarrell occurred here. "One of the most striking signs in approaching this area was the distinct lack of debris of any size," noted the National Weather Service assessment of the event. "Closer inspection showed lots of little debris, but no sign of large items. At least half a dozen cars were identified from the air lying in the open areas, most of them flattened and encrusted with mud and grass. Later, a ground survey revealed that most of the debris that was left in the area was extremely small, indicating the power of the tornadic winds."

The Jarrell tornado was preceded and followed by two F3 tornadoes and an F4, all produced by the afternoon’s southwestward-backbuilding series of storms. The F4 just west of Lake Travis killed one man, but none of these tornadoes left the kind of intense destruction found at Jarrell.

"Since the Jarrell event, I have yet to see anything comparable in terms of both storm evolution and tornado production," Houston told me in an email. He added: "There is evidence to suggest that, although most assuredly a supercell, the Jarrell storm was not maintained via conventional supercell dynamics.  In other words, this might be a (very unique) example of another way to 'get a supercell'."

Figure 4. Radar reflectivity image collected by the Aggie Doppler Radar (ADRAD) at Texas A&M University at 2024Z (3:24 pm CDT) on May 27, 1997, shortly after the Jarrell tornado developed. The tornado was located near the south end of the storm, close to Jarrell, but had not yet reached the Double Creek Estates area. Image credit: Michael Biggerstaff and Gordon Carrie (University of Oklahoma).

What if it happened today?

Figure 5. Aerial view of Albertsons grocery store showing damage from the Cedar Park tornado. Image credit: Long Phan, NIST, from the NWS Service Assessment “The Central Texas Tornadoes of May 27, 1997.”

Figure 6. Ground view of damage to Albertsons grocery store. Image credit: Michael Trainor, NOAA, from the NWS Service Assessment “The Central Texas Tornadoes of May 27, 1997.”

An event similar to the outbreak of May 27, 1997, in central Texas could inflict even more harm if it were to occur now, as the population of the exurbs north of Austin has mushroomed. The city of Round Rock, located 10 miles south-southwest of Jarrell, has grown from 61,000 residents in 2000 to an estimated 122,000 in 2016.  Even Jarrell itself now has roughly twice the population it had in 1990.

Tornadoes as intense as the ones in Jarrell, Jordan, Grand Island, and Plainfield--together with their potential for unusual evolution and motion--could easily spell trouble for storm chasers, whose numbers have ballooned since 1997, as well as for any other motorists. Seeking shelter above ground turned out to be a fruitless exercise for those caught in the core of the Jarrell tornado, yet driving away from a tornado is fraught with its own risks (many deaths occur that way, including one earlier this month near Elk City, Oklahoma). In-home storm shelters are more common now than in the 1990s; we can only hope that basements and/or in-home shelters might spare at least a few lives in any reprise of a Jarrell-type tornado.

One bright spot in the 1997 outbreak was the life-saving response at a supermarket heavily damaged by one of the F3 tornadoes noted above:

“The manager of the Albertsons grocery store in Cedar Park went outside into the parking lot to watch the approaching thunderstorm. While approximately 60 customers were shopping inside, he watched for a few minutes as the tornado developed. Going back inside the store, he instructed that an announcement be made asking everyone to gather in the center of the store so he could lead them into the store’s cooler in the rear of the building. He also asked customers trying to leave the store to stay inside.

"Although the store’s wide-span roof collapsed into the center of the store, the customers survived in the cooler with only a few minor injuries. One store employee who did not make it into the cooler suffered serious injury. The actions of the store manager saved many from certain serious injury and possibly death.”

A hundred years ago: One of the worst U.S. tornado strings on record

Harold Brooks (National Severe Storms Laboratory) called my attention to another red-letter event in American tornado history. Starting on May 25, 1917--100 years ago Thursday--an eight-day outbreak sequence delivered a total of at least 63 tornadoes rated at least F2. An estimated 383 people were killed in that weeklong span. “There were 9 separate double-digit fatality tornadoes in 7 states on 5 different days,” noted Brooks. The worst of the lot was a tornado family that rampaged across a 155-mile track in east central Illinois on May 26. It killed 101 people and inflicted major damage on Charleston and Mattoon as well as numerous other towns.

Figure 7. Tornadic destruction produced in Mattoon, Illinois, on May 26, 1917. Image credit: National Weather Service, Central Illinois, via Wikimedia Commons.

Tornadoes could threaten on Saturday across Southern Plains

The day with the most intense thunderstorms this Memorial Day weekend is likely to be Saturday, when an upper-level impulse racing across very unstable air will produce potentially severe storms from the Southern Plains into Virginia and North Carolina. The greatest tornado threat is expected over and near eastern Oklahoma, where a cold front and dryline will interact with outflow from early-day storms that are likely to be moving across Kansas and Missouri. The latter complex could end up producing severe wind gusts over a long swath as it progresses across the mid-Mississippi Valley. Update: In its revised Day 2 outlook for Saturday, SPC has placed southern Missouri and northern Arkansas under a moderate risk, citing the potential for a "severe convective windstorm" across the area on Saturday afternoon and evening. (Setups like these sometimes produce derechoes, such as the "super derecho" that plowed across Saturday's region of concern on May 8, 2009, with winds that topped 100 mph.) SPC also noted the potential for "significant tornadoes" in parts of Oklahoma, Kansas, and Missouri.