Local Climatological Data (LCD) indicates that the average thunderstorm season in North Dakota begins in early May and lasts until early September, with a fairly even distribution of thunderstorm days in June, July, and August. Hirt (1985) found that June is the primary month for tornado occurrence in the state. The typical thunderstorm in North Dakota does not produce a tornado, but when a tornado does form it is generally weak, short lived, and does little, if any, damage.
In the late afternoon of June 24, 1998, thunderstorms began developing in south central North Dakota. These storms exhibited both typical and atypical characteristics. Radar data showed cells forming in the same environment, very close to one another, yet exhibiting very different movements. The northern cell moved northeast and produced large hail at Gackle, in Logan County, then dissipated. The southern cell remained nearly stationary with new development on its southwest flank. This southern cell produced a tornado near Ashley, in McIntosh County, and later moved southeast and produced large hail and flooding rains.
The fact that the second storm remained nearly stationary and with new development for such a long time, and produced a tornado, was the motivation for this paper. This paper will examine the atmospheric conditions that led to storm development and examine subsequent radar data from both the Bismarck, North Dakota Doppler Radar (BIS 88D) and the Aberdeen, South Dakota Doppler Radar (ABR 88D).
Atmospheric data was examined using GEMPAK Analysis and Rendering Program (GARP) and the Skew-T/Hodograph Analysis and Research Program (SHARP). Radar data was examined using Level IV archive data from the BIS 88D and Level II archive data from the ABR 88D, using the WSR-88D Algorithm Testing and Display System (WATADS) version 10.0. WSR-88D and National Severe Storms Laboratory (NSSL) algorithms were used to evaluate radar data. NSSL algorithms used in this study were not operationally available the day of the event.
On the morning of June 24, a 1003-mb surface low was near Goodland, Kansas with a warm front extending to the northeast. This system was bringing high dew point air into the northern plains. At the same time a 1004-mb mesoscale surface low was between Pierre and Huron, South Dakota with a warm front into southwest Minnesota (Figure 1). During the day these systems remained nearly stationary and the higher dew point air continued to advect northward with the leading edge into North Dakota and Manitoba, Canada. Dew points in the south central and southeast parts of North Dakota rose from around 60°F in the morning to around 70°F by early evening.
During the day, a ridge of high equivalent potential temperature air at 850-mb nosed up through the eastern Dakotas (Figure 2). At 850-mb was a trough from Alberta, Canada to the central plains. During the late afternoon a circulation began forming along this trough, in Nebraska, with a 40 knot low level jet from Texas to Nebraska. At 500-mb, weak impulses were imbedded in southwest flow ahead of a long wave trough through the Rockies. Analysis of the 200-mb level revealed the main jet from the Desert Southwest across the central plains into the Great Lakes. There was a jet streak of 70 knots extending across South Dakota with the left front quadrant over southeast North Dakota.
McNulty (1995) used the term "extreme convective instability" to describe that instability with Convective Available Potential Energy (CAPE) values above 3,000 Jkg-1 or Lifted Index (LI) values more negative than -6°C. In this case study both parameter values exceeded the lower boundaries stated by McNulty, implying that extreme convective instability was present.
At 2100 UTC 24 June, showers were moving through southern South Dakota. The only convection in North Dakota was an isolated shower that developed 40 miles north of Fargo and was moving northeast. This shower developed into a thunderstorm and by 2130 UTC crossed into Minnesota. The storm continued to intensify and just before 2200 UTC radar showed two distinct reflectivity cores. The northern storm moved northeast and dissipated while the southern one moved east-northeast and continued to intensify. The National Weather Service Office in Grand Forks, North Dakota received numerous severe weather reports from 2230 UTC to 0022 UTC with this southern storm. Hail as large as golf-ball-size fell in Polk and Norman Counties in Minnesota. There was also an unconfirmed report of a funnel cloud. This severe weather occurred about 180 miles northeast of the convection that is the focus of this paper.
In the meantime, at 2234 UTC a shower began developing on the Logan-LaMoure county line in south central North Dakota. At 2240 UTC the ABR 88D identified this as a storm moving northeast into LaMoure County. This storm would move northeast and dissipate with no warnings issued and no severe weather reports received. In addition, at 2240 UTC, other convection was developing just to the south of this in McIntosh County and this is the focus of the paper. By 2252 UTC the ABR 88D identified this as a storm. Together the convection from Minnesota to south central North Dakota appeared oriented in a broken line from northeast to southwest (Figure 3). The broken line of convection was forming along a line that separated air with dew points in the 50s from dew points in the 60s. This line may have marked the position of the surface warm front mentioned earlier.
From 2315 UTC to 2332 UTC a cluster of three radar identified storms, all in McIntosh County, formed (Figure 4). The northeast most storm moved northeast and dissipated while the northwest most storm moved northeast and began decreasing in intensity. A new storm formed immediately north of it by 2358 UTC. This new storm would move northeast and at 0045 UTC produce 3-inch diameter hail at Gackle in Logan County. Figure 5a and 5b show base reflectivity (R) and storm relative mean-radial velocity (SRM) respectively, at 0044 UTC.
Newspaper reports from Gackle residents indicated that there was no rain and no wind before large hail started falling. The large hail lasted from five to ten minutes. The storm then decreased rapidly in intensity with no further severe weather reports. The southern most storm had regeneration on its southwest flank over the same area where it originally formed. This regeneration continued for more than an hour and resulted in a quasi-stationary storm motion. Through 0003 UTC only weak circulation was indicated.
At 0008 UTC, ABR 88D detected a mesocyclone with this regenerative storm, just northwest of Ashley in McIntosh County. At 0010 UTC law enforcement reported a tornado 2-½ miles west of Ashley. At 0013 UTC the NSSL algorithm identified a tornadic signature with the storm (Figure 6). The previous and subsequent volume scans did not show the signature. At this time neither the ABR 88D nor the BIS 88D detected a tornado vortex signature (TVS). However, the ABR 88D did detect a mesocyclone from 0008 UTC until 0024 UTC while the BIS 88D did not detect any mesocyclone with this storm until an hour later. Reflectivity on the ABR 88D showed a tight reflectivity gradient on the southwest flank as well as a Weak Echo Region (WER), but no noticeable hook echo. In addition to the initial tornado report at 0010 UTC there were several other tornado reports with the last one at 0022 UTC. All of these reports indicated the tornado west or northwest of Ashley.
From 0023 UTC until 0039 UTC there were no tornadic signatures and no reports of tornadoes. During this time regeneration on the southwest flank continued. At 0039 UTC and 0044 UTC the NSSL algorithm detected a tornadic signature west of Ashley. At 0040 UTC law enforcement reported a tornado near Ashley, the last tornado report in McIntosh County that night. There were no reports of damage from any of the tornadoes that occurred in McIntosh County the evening of June 24.
Up to this point there were no reports of large hail. During the next hour the storm would remain nearly stationary but with no additional severe weather reports. Also during this time, in addition to the regeneration on the southwest flank, new development began occurring on the southeast flank. By 0140 UTC this new development on the southeast flank became the more dominant feature and the line of convection began moving southeast (Figure 7). Figures 8a and 8b show base reflectivity (R) and a reflectivity cross-section (RCS) respectively, at 0151 UTC. The figures indicate high reflectivity aloft and a Bounded Weak Echo Region (BWER) just one minute before 1.75 inch diameter hail was reported 5 miles southeast of Ashley. The SRM from the next volume scan showed nearly 100 knots gate-to-gate shear (Figure 9). At 0305 UTC 1.75 inch diameter hail fell 16 miles southeast of Ashley near the South Dakota border. During the period of large hail reports the 88D hail detection algorithm indicated hail size varying from 1.00 to 3.00 inches in diameter. North Dakota Atmospheric Resource Board and NWS Cooperative observer reports indicated rainfall amounts ranging from 1.50 to 2.50 inches in southern McIntosh County.
Figure 5b. 0044 UTC 25 June 1998 0.5° SRM from ABR 88D.
Figure 9. 0156 UTC 25 June 1998 0.5° and 1.4° SRM from ABR 88D.
Recent studies have attempted to correlate positive cloud-to-ground lightning flashes to severe thunderstorms. MacGorman and Burgess (1994) conducted a study of 15 severe thunderstorms to examine relationships of positive ground flashes to various storm characteristics, especially reports of large hail and tornadoes. In only four of the 15 storms they examined, ground flash activity was dominated by positive cloud-to-ground lightning during most of the life of the storm. In the remaining storms the dominant polarity of ground flashes switched from positive to negative some time during the mature stage of the storm. They also stated that no study thus far had found frequent positive strikes in heavy precipitation supercell storms. On June 24, 1998, the storms that developed over south central North Dakota were dominated by positive cloud-to-ground lightning flashes throughout the life of the storms. National Lightning Detection Network (NLDN) lightning data obtained from the North Dakota Atmospheric Resource Board (NDARB) indicated that the percentage of positive flashes never fell below 60 percent with the storms in south central North Dakota. In fact most of the time the percentage was above 70 percent.
These thunderstorms in south central North Dakota on the evening of June 24, 1998, continued to move southeast and into South Dakota. The NWS Aberdeen office issued various severe weather warnings on these storms. Only heavy rain, estimated wind gusts to 50 mph, and a few funnel cloud reports were received. The storms eventually moved through Aberdeen and the NWS office received 2.54 inches of rain in about one hour. A peak wind gust of 38 knots (44 mph) was measured at the Aberdeen office at 0438 UTC. The storms would continue to move southeast and merge with other storms moving into northeast South Dakota. Together these storms evolved into a Mesoscale Convective System (MCS).
This case study documented a tornadic supercell thunderstorm from development stage in south central North Dakota during the evening of June 24, 1998, until it moved into South Dakota and developed into an MCS. Atmospheric conditions that led to the development of this storm were examined. Conditions conducive to storm development included abundant low level moisture with dew points of 65 to 70°F, a ridge of high equivalent potential temperature air, and extreme convective instability with CAPE values above 3,000 Jkg-1 and LI more negative than -6°C. Other contributing factors were the left front quadrant of a 70 knot jet streak over southeast North Dakota and a surface boundary. Radar imagery indicated several interesting features, including splitting storms and a quasi-stationary movement of the main storm. The authors stress the importance of a complete surface and upper air analysis prior to a severe weather threat. A complete analysis can lead to a better understanding of the resulting storm motions and to more precise warnings and statements.
The authors thank Director Bruce A. Boe and Meteorologist Darin Langerud of the NDARB for supplying lightning and rainfall data. We also thank Vic Jensen, SOO at NWS Bismarck for his encouragement in conducting this case study and for his review, Daniel Noah, WCM at NWS Bismarck for his assistance in preparing several of the graphics, and Bill Abeling, Lead forecaster at NWS Bismarck for his technical advise. Thanks also go to the staff of the Aberdeen NWS and Grand Forks NWS offices for supplying various data, and to Preston Leftwich, Central Region Scientific Services Division for his review of the paper and suggestions for improvement.
Hirt, W.D., 1985: Forecasting Severe Weather in North Dakota. Preprints, 14th Conference on Severe Local Storms, Indianapolis, AMS (Boston), 328-331.
MacGorman, D.R., and D.W. Burgess, 1994: Positive Cloud-to-Ground Lightning in Tornadic Storms and Hailstorms. Mon. Wea. Rev., 122, 1671-1697.
McNulty, R.P., 1995: Severe and Convective Weather: A Central Region Forecasting Challenge. Wea. Forecasting, 10, 187-201.
NOAA, NWS, NCEP, Climate Prediction Center, 1998: Daily Weather Maps, June 22-28.