The Southern Minnesota Tornadoes of
March 29th, 1998
Stories from the...
The National Weather Service Perspective
While many stories of the March 29, 1998, tornadoes have been often told and are relatively easy to find in newspapers or by simply talking to residents of Comfrey, St. Peter, Le Center, and other devastated communities, one perspective that is not as readily available is that of the many National Weather Service meteorologists that provided the warnings and forecasts in the days preceding, the day of the event, and the days that followed. Many extra hours and several long days were devoted by staff members at the National Weather Service offices in the Twin Cities and Sioux Falls, as well as the Storm Prediction Center (SPC) in Norman, Oklahoma, to provide the best service possible for the residents of southern Minnesota. Here is their story.
Before the Tornadoes
The fact that severe weather occurred in southern Minnesota on the afternoon and evening of March 29 th was not a total surprise. The initial hints that a significant severe weather event might unfold that evening actually became apparent in the computer model data on the evening of the 27 th. “I was on the evening shift when the 00Z [6:00 pm CST] model runs came in,” Warning Coordination Meteorologist at the NWS Twin Cities, Todd Krause, said. “That’s when we started becoming concerned that it could be a big event.” Krause's concern also led to his decision to contact the Twin Cities Skywarn Coordinator, John Kelley, on Saturday to alert him that he might be needed the next afternoon and evening.
The very same model runs, as well as the morning data from the 28 th, also triggered action at the SPC, the agency that is responsible for creating severe weather outlooks and initiating the issuance of severe weather watches. On the 28 th, the SPC’s Day 2 Severe Weather Outlook outlined a Moderate Risk of severe weather over much of southern and central Minnesota, as well as portions of Wisconsin, Iowa, and Illinois. Brynn Kerr, a forecaster at the SPC, recalls having a conversation with his coworker, John Hart, who was responsible for issuing the midday Day 2 Outlook on the 28 th. “…the strong signal in the model data for a potentially significant tornado event is, perhaps what stands out most in my memory for the St. Peter/Comfrey event of the 29 th,” Kerr said. “Similar feelings were shared by John Hart …, in conversations we had [on the 28 th].” Kerr noted that a warm front was to be the focus for thunderstorm development, as the atmosphere south of the front was quite capped, suppressing development.
Later model runs on the evening of the 28 th only increased the awareness that a significant severe weather outbreak was likely to occur across parts of southern and central Minnesota the next afternoon and evening. Senior forecaster, Bill Togstad, and forecaster, Dan Effertz, worked the midnight shift at the Twin Cities NWS on the early morning of the 29 th, and their expert diagnosis showed that the tornado risk in parts of the area was quite high, even though it was early in the season for strong tornadoes. Togstad and Effertz noted that the wind shear and instability forecast by the models was unseasonably high, and the presence of a warm front in the area only increased the concern of tornado development. To give an estimate of the expected tornado severity, based on the current model forecasts, they used an equation devised by John R. Colquhoun and Philip A. Riley, both Australian meteorologists. The equation uses surface based lifted index, which is a measure of instability, and the deep layer wind shear to predict the Fujita scale rating. “The index estimated F3 intensity tornadoes and this information was put in our Area Forecast Discussion around 4 am CST that morning,” Togstad said. The tornado threat was also mentioned in local morning Severe Weather Outlooks from the Twin Cities, Sioux Falls, and LaCrosse NWS offices.
Steve Corfidi, lead forecaster at the SPC, was also monitoring the weather situation in the Upper Midwest during the midnight shift on the morning of the 29 th. From the SPC perspective, the primary forecast question was whether sufficient moisture would invade southern Minnesota by the afternoon, as the atmosphere that morning was relatively dry, as far as the development of surface based thunderstorms, necessary to produce tornadoes, is concerned. This latter notation is in contrast to other regional tornado events, where ample low level moisture was present over the threatened area at the start of the day. Throughout the shift, Corfidi monitored the moisture transport through the Plains using hourly surface observations, as well as measurements of the low level jet, most often responsible for rapid northward moisture transport. “Given the strength of the overnight low level jet, and the fact that the jet likely would remain in place through the day (given the continued northeastward advance of an upper level jet streak from the High Plains), by the end of the shift we were fairly confident that ample moisture would indeed be present in Minnesota – despite the fact that surface dewpoints at that time (12-13Z [6:00-7:00 am CST]) were only in the 40s, and the surface warm front was still well to the south in southern Nebraska and Iowa,” Corfidi said. As a result, the SPC’s morning Day 1 Severe Weather Outlook continued the Moderate Risk of severe thunderstorms over southern Minnesota and northern Iowa.
With the increased forecast confidence, Corfidi made the decision to share his thoughts with the local forecasters via telephone. “I remember calling the forecaster on duty at the MSP [Twin Cities] NWS office around 13Z (7 am local time) to discuss the situation,” he said. “Some elevated thunderstorms were beginning to form on the northern edge … of the low level jet across South Dakota and Iowa. We were afraid that some of these might pose a threat for severe hail a bit later in the morning as the jet brought increasingly moist air at elevated levels northward. … We both agreed that a watch wasn’t needed at the moment. But I wanted to note that daytime heating in the upstream warm sector across Nebraska and Iowa likely would allow the elevated moisture to suddenly mix down to the surface, resulting in a sudden northward jump of the warm front. This, in turn, would allow for a rapid northward advance of the surface-based severe storm (i.e., tornado) threat into southern Minnesota by afternoon.”
As the day shifts at the SPC and the Twin Cities and Sioux Falls NWS offices arrived, the same mantra of significant severe weather potential was broadcast in various forecast products. The 11:30 am CST State Forecast Discussion from the Twin Cities NWS pinpointed southern Minnesota as having the best chance for severe thunderstorm development during the afternoon. Perhaps more striking was a Special Weather Statement, issued from the Twin Cities NWS at 12:26 pm CST. The statement proclaimed that “[t]hunderstorms located south of a Redwood Falls to Minneapolis to Rice Lake line could be particularly strong with the potential of tornadic thunderstorms.”
One of the main late morning and early afternoon concerns at the Twin Cities and Sioux Falls NWS offices was verifying that adequate staffing would be available during the afternoon and evening to provide the best service possible to the public. Quite often, forecasters and other support staff at the forecast offices either stay late on severe weather days or even sacrifice days off to provide the best warning and forecast services. March 29 th was no exception. Both Todd Krause and then-intern, Dan Miller were not scheduled to work that Sunday at the Twin Cities NWS. Both ended up working that Sunday. “I was called in to help,” Krause recalls, “and played coordinator.”
As many meteorologists plan to do on scheduled off days in which severe weather is a possibility, Dan Miller had plans to storm chase “to get [his] first fix of thunderstorms for the spring”, and pinpointed Mankato as a good beginning point for his chase. Before departing for the Mankato area, however, Miller stopped into the Twin Cities NWS to inspect the latest data and also inquire as to whether he would be needed that afternoon in support of the expected severe weather operations. His inquiry to the senior forecaster was met with the response “…if you’re offering, why don’t you hang around for a while and we’ll see what happens. It can’t hurt to have an extra body or two on a moderate risk day.” With elevated thunderstorms ongoing across east central and south central Minnesota and parts of western Wisconsin, most of which were severe, Krause, Miller, and others at the Twin Cities NWS were thrust into action before the first hint of actual thunderstorm development over southeastern South Dakota or southwest Minnesota became apparent.
At the SPC, Brynn Kerr, who was again on the day shift, and then-lead forecaster, Joe Rogash were closely monitoring the conditions across Minnesota, especially toward the noon hour, within a few hours of maximum daytime heating. All available observational and model data strongly indicated that a significant severe weather event would unfold, and strong action needed to be taken. “[I]t was felt that the best course of action was to issue a mesoscale discussion highlighting the anticipated threat, location, timing and likelihood of a watch, with perhaps a bit extra lead time than normal,” Kerr said. “This was done just before noon CST…” The mesoscale discussion included the following:
Given strength of vertical shear profiles, CAPE on the order of 2000 J/kg will support increasing potential for tornadic supercells during the afternoon hours, especially along an axis roughly from Yankton through Sioux Falls into Redwood Falls and Minneapolis/Rochester areas. We will continue to monitor. Present indications are WW (Weather Watch) will be necessary within the next 2 to 3 hours.
With continued high confidence that a significant severe weather event would transpire and thunderstorms beginning to initiate and strengthen near Sioux Falls, Kerr, Rogash, and John Hart concluded the time had come to issue a watch. Just shy of two hours after the mesoscale discussion was issued, Tornado Watch 132 was issued by the SPC at 1:35 pm CST. This watch included the Potentially Dangerous Situation designation, reserved for the most high impact tornadic situations.
After a severe weather watch was issued, at that time, the watch was reinforced with a status report from the SPC once every two hours. With this likely being a high impact event, the decision was this was not sufficient. “I decided to issue status reports earlier and more frequently than ‘normal’, with the hope of contributing to a ‘higher’ state of awareness among those forecasters at the warning desks, and all other users of the product,” Kerr said. Almost an hour after the watch issuance, at 2:34 pm CST, the first watch status report was issued, alerting that “indications continue to become increasingly favorable for tornadic supercells within the next 1 to 2 hours near northwest Iowa/Minnesota border areas bounded by Sioux Falls, Redwood Falls, and Spencer [Iowa].”
Now, at 3:00 pm CST, with Severe Thunderstorm Warnings already issued and expired in southeastern South Dakota and ongoing in Rock County in southwestern Minnesota, as well as a Tornado Watch in effect, the stage was set. The hours and days of analysis and preparation carried out by the meteorologists in Sioux Falls, the Twin Cities, and at the SPC in Norman, Oklahoma, were all about to be put to use. It was only 23 minutes until the first tornado would touchdown.
The Tornado Event
The first Tornado Warning of the event came out of the Sioux Falls NWS office at 3:09 pm CST, including parts of Murray, Nobles, and Rock counties in southwest Minnesota. The strongest portion of the storm was located about 17 miles southeast of Pipestone near the Rock/Nobles county border. At 3:23 pm CST, the supercell produced its first tornado over northern Nobles County near Lismore. The initial touchdown was followed other brief twisters near Leota, Wilmont, St. Killian, and Fulda over the next 25 to 30 minutes. At 3:50 pm, a not-so-brief tornado touched down just west of the Murray/Cottonwood county line – 7 miles east of Avoca; at the same time, a Tornado Warning for eastern Murray and Cottonwood counties was issued by the Sioux Falls NWS.
At the SPC, a second watch status report was composed and issued at 3:42 pm CST by Brynn Kerr, signaling that “Strong moisture flux convergence and vertical shear profile near surface low now east of Sioux Falls will likely maintain intense tornadic supercell across southern Minnesota counties between Sioux Falls and Minneapolis.”
The staff at the Twin Cities NWS was closely monitoring the action in southwest Minnesota, as the storm was rapidly approaching their own area of warning and forecast responsibility. Warning operations were already ongoing at the Twin Cities NWS, mainly in the Redwood Falls, Mankato, and Albert Lea vicinities. “The development and intensification was explosive with storms reaching severe levels and becoming supercells within about 30 minutes after initiation,” Dan Miller recalled. At that time, the technology only allowed for one warning forecaster monitoring the radar, although multiple assistant warning forecasters could be employed. Typing statements was the main duty of the assistants. On that day, then-forecaster, Greg Tipton was the primary warning forecaster, with Miller and, eventually, then-forecaster, Marc Kavinsky being his assistants. Miller primarily typed warnings and statements for the thunderstorms in east central and south central Minnesota. With this initial round of elevated thunderstorms, Miller said, “It got very, very busy very, very quickly.”
Even back then, information flowed between surrounding NWS offices, and news of possible touchdowns in southwest Minnesota was noted by meteorologists at the Twin Cities NWS. “There were fleeting reports of a tornado or funnel in Murray County, then two or three more solid reports came in from western/central Cottonwood County,” Todd Krause said.
With confirmed tornadoes in southwest Minnesota and very impressive reflectivity and velocity signatures from the relatively new Doppler Radar (WSR-88D), the decision to issue a Tornado Warning for Brown and Watonwan counties was made by Tipton when the storm was still in central Cottonwood County. This early of an issuance was rare. “There might still have been some old-style NWS concern about not issuing warnings for storms outside the CWA [County Warning Area],” Krause said. At 4:21 pm CST, Kavinsky, the main assistant for the tornadic supercell, hit the Enter button on SRWarn to send the warning for Brown and Watonwan counties. Nine minutes later, the tornado hit Comfrey.
Just prior to the tornado moving into Comfrey, John Kelley recalled that an eerie spotter report came across the amateur radios in the office. A rotating wall cloud had just collapsed into a "fog bank".
As the tornado tracked to the northeast, damage reports were very scarce, despite the considerable havoc the tornado was wreaking. The continued impressive Doppler Radar data, however, provided all the motivation to continue Tornado Warnings as the thunderstorm approached the Minnesota River and Blue Earth and Nicollet counties. At 4:56 pm, Blue Earth and Nicollet counties were placed under a Tornado Warning. As the tornado approached the city of Nicollet, a helpful spotter report was received by Amateur Radio operators inside the office. “A HAM radio operator reported one tornado on the ground two miles west of Nicollet, and another two miles east of Nicollet,” Krause recalled. It later became apparent that this spotter witnessed the thunderstorm undergoing what is called cyclic tornadogenesis, in which the original tornado is unable to sustain itself and another begins to the east of the dissipating original tornado. The weakening tornado west of Nicollet would be the end of the 67-mile long tornado track that began in far eastern Murray County.
Miller’s recollections of spotter reports mirrored Krause’s. “We didn’t get too many reports in real-time, but I seem to recall one particular report from west of St. Peter that made me think something ominous was going on,” Miller said. At 5:23 pm CST, Le Sueur County was placed under a Tornado Warning, and the city of St. Peter was hit at 5:35 pm CST.
This tornadic supercell was not the only show in the area, which provided additional complications. Another impressive, potentially tornadic thunderstorm developed southwest of the one carving a path through parts of the Minnesota River Valley. “[A] seemingly similar storm developed west-southwest of Mankato and headed straight for Mankato. A [Tornado Warning] was issued for that storm but fortunately nothing happened,” Krause said. “But I do remember being terribly worried for the Mankato area.”
Tornadoes continued to touchdown as the storm moved to the east-northeast, damaging parts of Le Center. Additional Tornado Warnings and supporting Severe Weather Statements poured out of the Twin Cities NWS office, as the busy warning team worked to keep the public, emergency services, and the news media fully aware of the danger moving ever closer to southern sides of the Twin Cities metro area. Tornado Warnings were issued for Le Sueur and Rice counties at 6:08 pm CST and Dakota County at 6:25 pm CST. Reports of damage near Lonsdale trickled into the office.
After the last tornado touched down at 6:43 pm CST approximately 6 miles southwest of Hastings, the storm finally decreased in intensity, losing its tornadic capabilities. It was only then that more information about what had actually transpired over southern Minnesota that afternoon began to pour into the Twin Cities NWS.
After the Tornadoes
“Later that evening, after the long tracked supercell had finally dissipated while crossing the Mississippi River near Hastings, more details on the damage began to trickle in, and it was really only at that point that it started to hit me what had just happened – Comfrey, Nicollet, St. Peter, Le Center – all had damage, and some of it sounded quite bad,” Dan Miller said. “There were reports of severe damage to Gustavus Adolphus College in St. Peter and that the downtown area of Comfrey was ‘destroyed’.”
“[A]lthough we knew there was some damage, I don’t think we had any idea how bad it was, at least not until a long time later when we started getting the reports of severe damage,” Todd Krause recalled.
The full extent of the damage would come to light the next day, March 30 th, when NWS personnel ventured to the affected areas to perform an official damage survey. With it being such a large event, many staff members participated in the surveys, led by Krause. Marc Kavinsky, Dan Miller, and then-intern, Jen Hacker contributed to the ground and aerial surveys.
Since the tornadoes wreaked such havoc over a large area, the ground survey took three days – Monday the 30 th through Wednesday the 1 st. The survey team from the Twin Cities NWS ventured from the office in Chanhassen on Monday, bound for St. Peter and Comfrey. Separate vehicles were taken, as, ironically, it was the middle of Skywarn training season and Kavinsky was scheduled for a few talks in western Minnesota, beginning that Monday night. Upon reaching the north side of St. Peter, the severe damage was immediately visible. “I was struck by how wide the tornado must have been,” Krause recalled. “Damage along Highway 169 started on the north side of town and continued well into the southern section.”
The team first stopped at the Emergency Operations Center (EOC) to get information on the hardest hit areas, as inspecting the worst damage is necessary to determine the official Fujita scale rating of the tornado.
As the team ventured further into the damage areas, members of the team were particularly affected by the damage done to the campus of Gustavus Adolphus College. Kavinsky remembered, “…one of the dorms at Gustavus Adolphus in St. Peter had almost all of its windows broken out.”
Miller was personally taken by the sight of the damage at Gustavus Adolphus. “…I had a close friend who was attending college at Gustavus at the time and had visited St. Peter and the Gustavus campus several times in the past couple of years. Most of the town as I knew it was heavily damaged.”
Other sights within St. Peter that were noted by the team members were the stripped, damaged, and destroyed trees in the downtown area, a car that was thrown about an eighth of a mile off the road, and a house on the southwest edge of town that was completely destroyed, swept off its foundation. “As respectfully as I could, I walked onto that property and asked the owner if it would be OK for us to look at the damage and take pictures,” Krause said. “A few days later, at the Iowa NWA [National Weather Association] Severe Storms Conference, I showed the photos to Tim Marshall, who thought that the house sustained F3 damage.” Seeing no areas with more significant damage than that one house, the maximum damage in St. Peter was deemed to be of the F3 variety.
The team now ventured to the southwest to Brown County and Comfrey, where the National Guard was staffing the checkpoints. “I had called ahead to ensure that county officials knew we were coming, so that we would be able to get into town,” Krause recalled. Similar to what they did in St. Peter, the team visited the mobile command post to get information from the Incident Commander. With an escort, the team set out to inspect the worst damage.
Both Miller and Krause were hit by the desolation within the town, as Comfrey was evacuated soon after the tornado moved through town. “It was very sobering to walk through streets surrounded by storm debris and few people,” Miller recalled.
The worst damage within Comfrey was deemed to be another house that was completely off its foundation. The home was found to not be attached to the foundation prior to the tornado, and the team once again rated the damage within Comfrey to be F3. Darkness finally fell, and Miller and Krause returned to Chanhassen. Kavinsky had already departed for his Skywarn presentation in Redwood Falls.
One question that the Monday survey had not answered was exactly how long the tornado or tornadoes remained on the ground continuously. On Tuesday the 31 st, the team departed once again into Nicollet County to determine if the tornado that destroyed Comfrey was the same one that also raked through St. Peter. “There was a three or four mile long area that appeared to have no damage,” Krause said, “and we were aware of the Ham radio spotter report from the city of Nicollet.” This report indicated that two tornadoes were on the ground at once – one west of Nicollet and the other east. The ground survey showed that the Comfrey tornado ended two miles west of Nicollet, while the St. Peter tornado began two miles east of Nicollet. Similar investigations were performed in Le Sueur County to determine that a different tornado moved through Le Center than the one that devastated St. Peter and also that the Lonsdale tornado was a separate tornado, as well. Krause also recalled that Tuesday was when “we saw the van that little six year old Dustin [the first fatality from the storms] had been in. It was about 1/8 mile off the road. Exceedingly sad.”
The team returned to Rice County on Wednesday, April Fool’s Day, to survey the remainder of the damage near Lonsdale, eastward to near Hastings in Dakota County, where the last reports of tornadic damage occurred.
Krause, Miller, and Hacker took to the air on Thursday to perform the aerial survey. Aerial surveys are often done during large tornado events, in which there are multiple touchdowns and/or long tracked tornadoes. They are best at determining whether a path is continuous or whether multiple tornadoes were involved in the destruction. The State Patrol provided the airplane, flying past Comfrey and across much of Cottonwood County. Krause took still photographs, Hacker operated the video camera, and Miller plotted the damage path on maps. Ironically, the weather did not fully cooperate, with low ceilings providing less than optimal flying conditions, and the team was unable to perform the exhaustive aerial survey that such an event deserved. “Would we have seen a continuous path near Nicollet? Between St. Peter and Le Center? Near Montgomery and beyond?” Krause posed these questions, but concluded, “I’m satisfied we did the best we could at the time.” Besides the weather, the time of year was a complicating factor for ascertaining accurate track information, as the ground remained quite hard, wet, and without ample vegetation, making typical aerial signals such as damage paths through agricultural fields and scoured earth very difficult or impossible to see. A fully exhaustive aerial survey may still not have answered these questions.
As Friday arrived, the surveying was almost complete – but not quite. Kavinsky ventured once again into eastern Brown County to inspect additional damage in the rural areas around Lake Hanska. Area residents, seeing the complete devastation to the community of Comfrey, were initially quite unhappy with the F3 rating given to the tornado that caused so much destruction. News of potentially worse damage in the Hanska vicinity led to the deeper investigation. Upon surveying one flattened, well built home, the tornado was deemed to have produced F4 damage. The final, official Fujita scale rating for the Comfrey tornado was F4.