Event Summary

Storm Summary

Summary of Warning and Forecast Services


At approximately 8:40 p.m. (all times Central Daylight Time [CDT]) on Saturday, May 30, 1998, a violent tornado struck the small town of Spencer, South Dakota. Spencer is in extreme western McCook County, about 45 miles west-northwest of Sioux Falls. The tornado killed six people, injured more than one-third of the town's 320 residents, and destroyed most of the town's 190 buildings. Damage is estimated at $18 million. The Spencer tornado (rated F4 on the Fujita Tornado Intensity Scale, as described in Appendix A) was one of five tornadoes, along a nearly continuous damage track approximately 30 miles long. All five tornadoes were produced by one supercell thunderstorm during a 1 hour and 5-minute period. The most devastating damage was limited to less than 1 mile along the tornado track, but significant damage (F1 to F3 intensity) was inflicted on several farmsteads in McCook County and neighboring Hanson County, including damaged or destroyed buildings, loss of crops, and loss of approximately 75 head of livestock.

The six tornado fatalities in Spencer were the first in the NWSFO Sioux Falls County Warning Area (CWA) since the Chandler, Minnesota, tornado in June 1992 and the first deaths in South Dakota in nearly 28 years. On July 14, 1970, one person was killed by a tornado in Lincoln County, just south of Sioux Falls. The last violent (F4) tornado in the Sioux Falls CWA occurred on June 7, 1993, just north of Sioux Falls.

The NWSFO in Sioux Falls issued a Tornado Warning for northern McCook County, including the town of Spencer, at 8:32 p.m. based on data from its Weather Surveillance Radar-1988 Doppler (WSR-88D) and reports from law enforcement officials. The warning was promptly disseminated over the National Oceanic and Atmospheric Administration (NOAA) Weather Wire Service (NWWS), commercial radio and television, NOAA Weather Radio (NWR), the Emergency Alert System (EAS), and the National Warning System (NAWAS). The outdoor warning siren at the Spencer Volunteer Fire Department could not be activated apparently due to loss of electricity from the storm.


Storm Summary

By late afternoon on May 30, the atmospheric environment over the north-central United States had become conducive to a significant severe weather outbreak. The air mass over southeastern South Dakota was extremely unstable (note the shaded area in Figure 1). A dry line/cold front combination was pushing slowly to the east and southeast into this unstable air, while at jet stream level in the atmosphere, a well-defined wind maximum (a jet streak) was approaching the dry line/cold front intersection. One favorable ingredient that was lacking ahead of the dry line/cold front was a low-level wind maximum (low-level jet). It was displaced farther east, over eastern Iowa and Minnesota (arrow in Figure 1). However, even with the lack of strong low-level winds, instability along the advancing dry line/cold front boundaries was sufficient for thunderstorm development, and the approaching jet streak provided more than adequate vertical wind shear for supercell evolution once storms developed.

Figure 1. A composite analysis, indicating the synoptic conditions relevant to the Spencer tornado as the parent thunderstorm, developed at 6 p.m. CDT. The (*) indicates Spencer, and FSD indicates Sioux Falls. Thick lines with arrows indicate the upper and lower level jet streams, with relevant isotachs shown in thin, labeled contours. The shaded regions indicate areas of high Convective Available Potential Energy (CAPE), implying strongest instability. The label of strong tropospheric shear indicates regions with more that 50 knots shear in the 700 to 300 mb layer. Surface low pressure centers are depicted by an "L."

Figure 2. A Geostationary Operational Environmental Satellite (GOES)-8 1 km visible image at 4:32 p.m. CDT (2132 Coordinated Universal Time [UTC]). The letter "C" corresponds to towering cumulus; "B" indicates a surface boundary; "*" indicates the location of Spencer; and "L" indicates a surface low pressure center.

Towering cumulus formed on the dry line by 4:30 p.m. near the Missouri River, about 90 miles west-northwest of Spencer (indicated by a "C" in Figure 2). Convective development in that area over the next hour could have been related to the intersection of a weak east-west boundary with the dry line. The enhanced band of cumulus in the area marked "B" in Figure 2 provided a hint of this boundary. Initial convection produced a brief thunderstorm by 5 p.m., however, stronger storms developed by 6:35 p.m. southwest of Wessington Springs, about 60 miles west-northwest of Spencer. The thunderstorm that would eventually produce the Spencer tornado evolved from this new development.

The Spencer storm almost immediately split into left and right moving cells with the right-mover becoming a supercell as a mid-level mesocyclone developed by 7:26 p.m. The mesocyclone grew downward toward cloud base as observed by the Sioux Falls WSR-88D between 7:45 and 7:50 p.m. During the same time, a rear flank downdraft (RFD) was developing south of the mesocyclone. RFD features are common prior to tornadogenesis in many supercells.

The Spencer supercell began to show stronger tornadic potential at 7:55 p.m., as a spotter observed a funnel cloud in northeast Davison County and the Sioux Falls WSR-88D indicated a significant mesocyclone in the storm. The first spotter report of a tornado associated with this storm came at 8:08 p.m., about 8 miles northeast of Mitchell. Tornado #1 (shown in Figure 3) actually began a few minutes earlier in that area, based on observations by members of a tornado field research project who were located on Highway 38, south of the storm. The ROTATE-98 project (Radar Observation of Tornadoes and Thunderstorms Experiment), coordinated at the University of Oklahoma, employs two truck-mounted Doppler radars, referred to as Doppler on Wheels (DOW), which are designed to gather high-resolution reflectivity and velocity data from locations within a few miles of tornadic thunderstorms.

Even though the first tornado ended northwest of Fulton at 8:15 p.m., the mesocyclone was still situated at ground-level. By 8:17 p.m., the low-level circulation produced Tornado #2 (Figure 3). This tornado passed about 2 miles north of Fulton, producing F1 to F2 damage before weakening about 4 miles west-northwest of Farmer, or 7 miles west-northwest of Spencer, at approximately 8:22 p.m. The parent mesocyclone, still appearing very strong on the NWSFO Sioux Falls WSR-88D, continued moving east-southeast.

A new damage track (Tornado #3 in Figure 3) began almost immediately, approximately 1 mile northeast of the end of Tornado #2. Reports of Tornado #3 were passed from law enforcement officers through NAWAS warning points to NWSFO Sioux Falls between 8:23 and 8:29 p.m. At 8:28 p.m., the Sioux Falls WSR-88D showed a well-defined hook echo in reflectivity data and a well-defined circulation (strong mesocyclone) in velocity data (Figure 4).

From 8:23 to 8:37 p.m., the Spencer tornado tracked through farmland, within 1 mile of the town of Farmer, prior to crossing the Hanson/McCook County line and striking the town of Spencer. Photographs from storm chasers show that it had grown to a large tornado west of Spencer. Apparently, it became so large that some chasers and storm spotters within 2 miles of the tornado became confused because of the large dust cloud surrounding the tornado, especially along the west and south sides where dust was raised by strong RFD winds. The DOW used by the ROTATE-98 project measured velocities of at least 90 meters per second (m/s) (200 mph) during this period.

Figure 3. A composite of tornado damage tracks based on aerial surveys. Eight tracks are indicated, however, only the first five are addressed in the text. Relative path length, with and F-scale damage are shown. (Brian E. Smith)

 Figure 4. A four-panel radar image from the Sioux Falls WSR-88D showing the Spencer supercell at 8:28 p.m. CDT (0128 UTC). The gray shading for the reflectivity panels progress from light to dark from 0 to 40 dBZ and again from 45 to 70 dBZ. For the velocity panels, the very light shaded region just south (north) of the tornado indicates inbound (outbound) winds more than 64 kt (more than 15 kt) respectively.

The impact time of the tornado on Spencer varies somewhat according to the source of information. Data from the ROTATE-98 project place the tornado near the western edge of town as early as 8:37 p.m., while the local electric company logged power failure in town at 8:42 p.m. and survivor accounts run as late as 8:45 p.m. Based on DOW and WSR-88D data, however, the town of Spencer experienced violent tornadic conditions from 8:38 to 8:39 p.m. Wind speeds observed by the DOW, as the tornado passed through Spencer, reached 98 m/s (nearly 220 mph) just south of the tornado center. While tracking through town, dirt and rain wrapping around the tornado might have obscured it somewhat as the parent thunderstorm took on high-precipitation supercell characteristics.

After leaving a path of destruction nearly 1-mile wide in Spencer, the tornado curved southeast through more farmland, most likely hidden in wrapping rain curtains. This damage track finally ended just north of I-90. Tornado #4 (see Figure 3) developed just south of I-90 and tracked southeast, followed by the weaker Tornado #5, which dissipated just northeast of Stanley Corner, about 30 miles west of Sioux Falls, at approximately 9:10 p.m.


Summary of Warning and Forecast Services

NWSFO Sioux Falls is a full-service forecast office in the midst of a transition to a modernized WFO. The staff provides warning and forecast services for 45 counties in parts of four states (South Dakota, Minnesota, Nebraska, Iowa), while maintaining many of the traditional forecast office responsibilities for the entire state of South Dakota (e.g., public/zone forecasts). The office uses a combination of modern technology, such as the WSR-88D radar, and older systems, such as the Automation of Field Operations and Services (AFOS) workstations and NWR consoles employing play/record decks for magnetic tapes. (The office operates three NWR stations.)

The potential for severe weather on May 30 in parts of South Dakota, Minnesota, Nebraska, and Iowa was recognized by forecasters the day before. The NWS SPC included the Sioux Falls CWA in its Day 2 Severe Weather Outlook (SWO) first issued at 2:55 a.m. Friday, May 29. In an update issued at 12:02 p.m. Friday, the SPC forecaster specifically mentioned "EXTREME INSTABILITY," "SUPERCELL THUNDERSTORMS," and the "THREAT OF ISOLATED TORNADOES." The NWSFO Sioux Falls forecasters were also monitoring the severe weather threat for Saturday very carefully. A forecast discussion issued at 9:10 p.m. Friday mentioned increasingly favorable conditions for severe weather "...IN THE FORECAST AREA ANYTIME FROM LATE TONIGHT THROUGH SATURDAY EVENING." The scenario anticipated by forecasters on the midnight shift early Saturday morning called for two rounds of severe weather--one early in the morning and the other late in the day.

The Day 1 SWO issued by SPC at 12:54 a.m. Saturday placed a moderate risk of severe storms from eastern portions of the Sioux Falls CWA into Iowa, Minnesota, and Wisconsin. However, a slight risk area covered most of the rest of the Sioux Falls CWA, including the area around Spencer. The 5:20 a.m. update from SPC included no significant changes in the north-central United States. However, based on their local analysis, the NWSFO Sioux Falls forecasters opted to upgrade the outlook for their CWA, and in a Special Weather Statement (SPS) issued at 4:35 a.m. headlined a "...MODERATE RISK OF SEVERE WEATHER TODAY...SOME THIS MORNING AND SOME THIS EVENING...."

Strong thunderstorms developed in south-central and southeastern portions of South Dakota early Saturday morning, prompting a severe thunderstorm watch from SPC at 3:22 a.m. and a number of severe thunderstorm warnings and statements from NWSFO Sioux Falls. These storms subsided around midday. One forecaster who remained on duty from the midnight shift (midnight to 8 a.m.) in an augmented staffing configuration for the morning severe weather told the day-shift staff as he departed that he would be available to come in early if necessary that night to help again with severe weather operations.

Midday and early afternoon provided a short break. However, a Significant Weather Outlook issued by NWSFO Sioux Falls at 11:47 a.m. was already focused on the likely redevelopment of thunderstorms "...LATE THIS AFTERNOON OR EARLY THIS EVENING...." While formulating the public forecasts to be issued by 4:30 p.m., the NWSFO Sioux Falls forecasters planned staffing for the anticipated redevelopment. An updated Day 1 Outlook from SPC issued at 2:41 p.m. maintained roughly the same configuration of slight and moderate risk areas as the earlier issuance, except for an expansion of the moderate risk area to the west to include extreme northeastern South Dakota. A Mesoscale Convective Discussion (MCD) issued by SPC at 12:54 p.m. highlighted the increasing threat of severe weather in western South Dakota for mid and late afternoon. Another MCD issued at 2:19 p.m. focused on the increasing threat from east-central South Dakota into parts of Minnesota and Wisconsin. Both mentioned portions of South Dakota as candidates for severe weather watches later in the afternoon.

The SPC issued a Severe Thunderstorm Watch at 3:33 p.m. for much of South Dakota, north of I-90, and a Tornado Watch at 3:50 p.m. for an adjacent area from eastern South Dakota through southern Minnesota. McCook County was included in the tornado watch. Both watches were in effect until 10 p.m. The watches were disseminated before 4 p.m. and the public forecasts were issued by 4:30 p.m. As often occurs at NWS forecast offices during late afternoon severe weather events, the two forecasters who had worked the day shift remained on duty to augment staffing for the evening shift. One forecaster stationed himself at the WSR-88D Principal User Processor (PUP) workstation while the other went to the personal computer (PC) workstation used to generate warnings and statements.

The staffing configuration at the start of the evening severe weather event consisted of four forecasters (two in the standard public and aviation forecaster roles and two handling warnings and statements) and one Hydrometeorological Technician (HMT). One of the staff interns was called at home and asked to report for duty by 7 p.m. to take over NWR broadcasts. This brought total staffing to six just as severe storms entered the Sioux Falls CWA. The staffing level remained at six until 9 p.m. At that time, the forecaster who had already worked an extra

4 hours from 8 a.m. until noon reported 3 hours early for his upcoming midnight shift. This brought the total staff to seven for about 1 hour as the severe storms moved through the Sioux Falls area.

During the severe weather episode on the evening of May 30, WSR-88D data analyzed at the PUP workstation and spotter reports became the primary input for warning decisions. Most of the key spotter reports were relayed from law enforcement officials through NAWAS warning points to the NWSFO. A few came from storm chasers who were in the area.

The first warning issued by the Sioux Falls NWSFO for this event was a severe thunderstorm warning at 6:45 p.m. for Beadle County, the farthest northwest county in the CWA. Over the next 4 hours, the Sioux Falls staff issued a total of 11 Severe Thunderstorm Warnings, covering 15 counties, and eight Tornado Warnings, covering nine counties. In addition, forecasters issued five Short Term Forecasts and 26 Severe Weather Statements during the evening.

The first severe weather report of the evening (three-quarter-inch diameter hail) came from 3 miles west of Huron in Beadle County at 7:19 p.m., more than 30 minutes after the first warning was issued. The first significant report on the storm that would eventually produce the Spencer tornado was the 7:55 p.m. funnel cloud report in northeast Davison County from the County Sheriff. A Severe Thunderstorm Warning had been in effect since 7:44 p.m. for northeast Davison and northern Hanson Counties, but the funnel cloud report combined with radar data led to the first tornado warning, issued at 8:01 p.m. for northeast Davison and northern Hanson Counties. The first tornado report on the Spencer storm was the 8:08 p.m. report from approximately 8 miles northeast of Mitchell.

The 8:01 p.m. tornado warning received limited dissemination, mainly through NAWAS and secondary law enforcement channels. It did not go from the PC used to prepare warning products to AFOS and, consequently, did not reach NWWS and other systems feeding the media and EAS. The apparent reason for this dissemination problem was a communications traffic "bottleneck" at a peripheral sharing device (PSD) serving as the interface between a cluster of four PCs and AFOS. One of the NAWAS warning points, having been notified of the tornado warning at 8:02 p.m. by NAWAS phone, but not seeing a hard copy of the message on the state law enforcement telecommunications system, notified the forecast office at approximately 8:10 p.m. that the message had not been received. By this time, the tornadic thunderstorm had crossed the Hanson County line.

Just as a new Tornado Warning was being issued for northern Hanson County (at 8:12 p.m.), a telephone call came from the Kansas City NWSO relaying a report of a tornado in northern Hanson County, west of the town of Farmer. The report came from the WCM from the Kansas City NWSO, one of the many storm chasers in southeast South Dakota that evening. Hearing no active amateur radio networks or other accessible spotter networks at the time, and not having the telephone number of the Sioux Falls office handy, this NWS employee used his personal cell phone to call back to his own office. Just 1 minute later at 8:13 p.m., a report of a large tornado northeast of Mitchell was relayed by the Mitchell NAWAS warning point.

At 8:16 p.m., the SPC issued a new Tornado Watch for extreme southeast South Dakota, northeast Nebraska, and northwest Iowa. This watch emphasized the "PARTICULARLY DANGEROUS SITUATION" (a PDS watch) evolving in southeast South Dakota. The NWSFO included McCook County in this new Tornado Watch.

The next tornado reports came from several law enforcement officials between 8:23 and 8:29 p.m. The first, relayed through the Mitchell NAWAS point at 8:23 p.m., indicated a tornado 5 miles northwest of Fulton. Between 8:25 and 8:29 p.m., NAWAS points at Mitchell and Sioux Falls (Metro Communications Center) relayed reports from a South Dakota Highway Patrol officer, the McCook County Sheriff, and a deputy sheriff all indicating a tornado west of Spencer. A Tornado Warning was issued for northern McCook County at 8:32 p.m., and it was promptly disseminated through usual means to the media and law enforcement officials (i.e., NWWS, NAWAS, EAS, etc.). The NWR log indicates that the warning was broadcast live within 1 minute of issuance.

At 8:50 p.m., the tornado was reported as it reorganized on the south side of I-90, approximately 5 miles southeast of Spencer. Reports of damage in Spencer started reaching the forecast office just after 9 p.m. However, details regarding fatalities and the scope of the destruction did not reach the forecast office until after 10 p.m.

With the exception of the PSD problem, resulting in the missed transmission of the 8:01 p.m tornado warning to AFOS, all of the equipment and computer systems used by the staff at the Sioux Falls NWSFO functioned well throughout the severe weather episode on the evening of May 30. Forecasters changed power at the WSR-88D radar from commercial to generator just after issuing the 8:12 p.m. Tornado Warning for northern Hanson County rather than risk loss of radar data as severe storms continued to approach Sioux Falls. The power change resulted in some loss of archive data but no loss of real-time data critical to the warning process.

Assessment team interviews with key representatives of Sioux Falls television stations and area radio stations and service feedback questionnaires distributed to the media by the Sioux Falls WCM indicate a universal satisfaction with the service provided by the NWSFO during the May 30 severe weather events. Broadcasters who were on the air that evening all indicated they had timely and accurate information. One TV meteorologist described the service provided as "excellent" and "perfect."

Interviews with Spencer residents and second-hand information gathered through media accounts indicate a wide range in the level of awareness prior to the tornado and a wide variety of safety precautions taken during the event. Many survivors had some indication, either through media reports or by first-hand observation, of the potential severity of the event, although many residents were quite unaware. Many had time to take shelter in a basement or some other part of their homes, while others were caught with just enough time to get behind a piece of furniture or to lie on the floor. All six fatalities were adults, ranging in age from 62 to 93, and occurred in the southern half of Spencer where destruction was worst.

Apparently, most of those who were watching television were tuned to one of the local network affiliates. The three local stations did an excellent job of broadcasting warnings as they were issued and providing "cut-ins" with displays of radar images. There were also many comments that the commercial radio stations did an excellent job of relaying critical warning information and reports. Two stations located in Mitchell played an especially vital role because of their proximity to Spencer and their first-hand accounts of upstream severe weather. The cable TV provider for Spencer has approximately 95 subscribers, but there was no cable-override system for the town. NWR reception from the Sioux Falls and DeSmet transmitters is marginal at best in Spencer. No one from the Service Assessment Team or the NWSFO is aware of any regular NWR listeners in Spencer. A few residents knew the radio frequency used by local spotters and received very timely and valuable information while monitoring on a scanner.

Return to Spencer Tornado Table of Contents or to Disaster Survey page.