Thunderstorms occur all summer long in North Dakota and Minnesota, and although all thunderstorms are worthy of notice, some situations pose a much greater threat and need to be watched more carefully. One such instance occurred on 22 August 1995 in northwestern Minnesota where thunderstorms developed late that afternoon, dropping hail and possibly spawning a tornado.
Morning sounding information and NGM model forecast products did indicate that the potential did exist for severe weather and the situation was monitored for development.
UND's C band Doppler radar was used to detect storm development and
also used at RWIC for nowcasting assistance to Polk County, MN's Emergency
Operations Center(EOC). RHI and PPI data provided indicators of possible
hail and tornadic development. Fortunately, the storms occurred around
00 UTC 23 August, so some incite could be obtained concerning model performance
and atmospheric conditions at the time of the storm's onset.
Morning atmospheric conditions in the northern plains indicated a good amount of instability and warranted consideration of severe weather potential. Upper-air charts, stability indices, and NGM model output all pointed toward the possibility of severe weather.
At 1200 UTC 22 August, the center of a surface low pressure system was
located just north of the North Dakota/Montana border in far southern Saskachewan
(Figure 2.1). This low pressure system and trough were progged by the NGM
to move off to the east during the day, serving as a possible trigger mechanism
for convection. Also at the surface, dewpoint temperatures were in the
60s and 70s, evidence of necessary low-level moisture. The 850 mb upper-air
chart indicated that it was extremely warm in the northern plains. Bismarck,
ND (BIS) reported an 850 mb temp of 21oC with a dewpoint depression
of 2 oC, International Falls, MN(INL) reported 17 oC
and a depression of 16 oC. Finally, Shilo Manitoba(WLO) reported
a temp of 21 oC and a depression of 9 oC. With temperatures
as warm as they were, the dewpoints were still quite high, despite the
greater depression values, indicative of a fair amount of moisture. The
700 mb level (Figure 2.2) was dry everywhere in the northern plains, with
BIS, WLO, and INL reporting dewpoint depressions of 20, 14, and 10 oC
respectively.
Figure 2.1: 1200 UTC 22 August surface map of
the north-central United States.
Figure 2.2: 1200 UTC 700 mb upper-air analysis for 22 Aug
1995. The shaded area represents the area under the 10o C capping
isotherm.
This dry layer at 700 mb served as a "cap" for potential instability below. Regarding the 700 mb capping mechanism, the Bismarck National Weather Service, in numerous forecast discussions, had mentioned the 10 oC isotherm at the 700 mb level as the edge of a convection-limiting cap. Areas possessing temps greater than 10 oC were in an area where convection was capped, or likely inhibited.
Also, at 500 mb the NGM indicated a vorticity minimum in far northwestern North Dakota with negative vorticity advection taking place in eastern and central parts of the state, which would serve to help prevent thunderstorm development in central ND.
A quick look at stability indices at 1200 UTC also shows conditions favorable for severe weather development. Total Totals (TT) and SWEAT (SWT) indices were calculated using the upper-air data taken from the charts while the lifted indices (LI) were taken from the NGM products. At 1200 UTC, BIS had a TT of 60, a SWT of about 670, and an LI of -8 oC. The conditions at INL were not nearly as conducive to severe weather. INL's TT was only 40 while its SWT and LI were only about 62 and +4 respectively.
The NGM forecasted the surface low to progress eastward throughout the day, along with the instability. The 12 hr forecast of the 1200 UTC NGM run showed a broad area of negative LIs over northern MN and eastern ND, with the axis running directly up the Red River Valley. Also, 1200 UTC upper-air maps indicated slight warm-air advection at low levels and cold-air advection at upper levels which would serve to destabilize the atmosphere ahead of the approaching trough. The surface trough was forecasted by the NGM to be located from the Ontario/Manitoba border by 0000 UTC, extending southward into northwestern MN and eastern ND.
So, taking into consideration the surface heating expected during the
day to build instability ahead of the surface trough, the 700 mb cap over
most of ND, and the availability of moisture, the area which seemed most
apt to experience thunderstorm activity on 22 Aug 1995 was northeastern
ND and northwestern MN. Those areas possessed the necessary ingredients
for severe weather. Low-level moisture was present, as evident by the dewpoints
in the 70 oF range. The instability was forecasted to be present
via negative LIs and a 700 mb capping mechanism. Finally, the trigger mechanism
was forecasted to be present also, in the form of a surface trough.
By 0000 UTC the surface trough did move into northeastern ND(Figure
2.3) as forecasted and temperatures rose into the 90s with dewpoints commonly
in the 70s.
Figure 2.3: 0000 UTC AGWINDS surface analysis for the north-central U.S. Red lines are isotherms, green lines are isodrosotherms, and vectors are of winds
Since the storm in question occurred near 0000 UTC, the 0000 UTC analyses give a reasonable representation of the atmospheric conditions near the storm time.
The 850 mb chart(Figure 2.4) still indicated a slight presence of moisture.
At 700 mb (Figure 2.5), the dry cap was still in place throughout the region
to help maintain the instability, with northeastern ND and northwestern
MN at the edge of the 10 oC isotherm. Finally, the 500 mb level
(Figure 2.6) showed some moisture(INL dewpoint depression of 1 oC),
along with anti-cyclonic shear and very slight negative vorticity advection
which may have aided in lift.
Figure 2.4: 0000 UTC 850 mb upper-air analysis
for 23 Aug 1995. Contour is of 5 degree dewpoint depression.
Figure 2.5: 0000 UTC 700 mb upper-air analysis
for 23 Aug 1995. 10 oC capping isotherm is bold. Hatched area
is capped.
Figure 2.6: 0000 UTC NGM Initial Panel for 500
mb Heights and Vorticity.
0000 UTC stability indices indicated that the atmosphere did indeed
destabilize in the storm area. At 0000 UTC, BIS had its TT drop to 52 and
SWT drop to 273; however, INL's TT jumped to 58 (from 40) and its SWT rose
to 642 (from 62). Also, lifted indices at 0000 UTC were near -8 oC,
very conducive to severe weather development.
The first radar echo of storms began to show up on the UND Doppler radar around 2100 UTC, during the time of maximum surface heating, much as expected. By 2300 UTC, three significant thunderstorms were well underway in Minnesota's Kittson County and the studied storm was just beginning to develop.
UND Radar also detected a fine line in the reflectivity field. Convergence could clearly be seen in the velocity field which served as the trigger for these storms. The storm of interest grew quickly and tracked to the east at approximately 24 - 27 mph for a while before veering to the southeast. Explosive grow was exhibited by the storm such that by 2337 UTC, and RHI through the storm indicated echo tops to 53,400 ft MSL with 54 dBz echo extending to 37,800 ft. Reflectivity returns of this magnitude, at such a great height above the freezing level(approximately 13-14,000 ft.), indicated that there was a definite possibility of hail within the storm. Dennis et al (1970) concluded that reflectivities near or above 50 dBz at altitude of 2 - 3 km provides for a reasonable chance that hail is present within the storm and even possibly reaching the ground. Also, Cheng and Rogers (1988) concluded that any reflectivity greater than or equal to 60 dBz reasonably assures that hail would be reaching the ground.
By 0000 UTC (Figure 3.1), a hook echo was clearly present at approximately 54o and 87 km from the UND radar. At 0013 a new radar scan was created so as to greatly increase the maximum unambiguous velocity of the radar. At this time a meso-low was clearly detected in the wind field showing rotation. At the 0.5o elevation the maximum toward velocities of the meso-low are on the order of 15 m s-1 while away velocities are approximately 20 m s -1. It is this low-level rotation that may have spawned a tornado.
Figure 3.1: UND Doppler Radar scan at 0000 UTC 23 August 1995. A hook echo is visible at the northwestern edge of the storm. An 'x' indicates the approximate location of the hail report and a '+' indicates the possible tornado report.
At approximately 0035 UTC, a funnel cloud was reported by a sheriff
over the Lake Agassiz Refuge in eastern Marshall County in MN. Also, at
0049, a possible tornado was reported by a citizen 19 miles east of Thief
River Falls, MN. These reports will be discussed further in the following
section.
IV. Discussion
Monitoring the storms carefully with the radar allowed forecasters at RWIC to communicate with the Polk County EOC in Minnesota. Although storm observed here occurred primarily in Marshall County to the north, phone communications helped forecasters keep the EOC aware of situations involving possible hail and meso-cyclone activity which may have move into their area.
Was the effort successful? The Polk County EOC was very appreciative of RWIC's assistance for decision making and notification of current and impending weather. Severe storm reports from that day indicated that the storms indeed were strong to severe. A copy of the National Severe Storms Forecast Center's Tornado and Severe Thunderstorm Report was obtained along with a preliminary storm report from the Fargo, ND National Weather Service office. There was a report of a possible tornado located 19 miles east of Thief River Falls, MN (TVF) at 7:49 PM CDT (0049 UTC) and a report of a funnel cloud at 7:35 PM CDT (0035 UTC) over the Lake Agassiz Refuge in northwestern MN. At 0040 UTC quarter-sized hail was reported by a coop observer near the town of Goodridge in Marshall County. Figure 3.1 from 0000 UTC, shows the approximate locations of the hail and possible tornado reports. These locations match up favorably with the storm considering its trajectory to the southeast.
Radar observations indicated a strong likelihood of hail within the storms and hail reported on the ground verified the nowcasting of the forecasters and radar operator.
Regarding the possibility of tornadic activity, the radar did detect a meso-low within the storm; however, at 90 km ( 56 mi), the radar beam (assuming 1o beamwidth) would be about 1 mile across, making it nearly impossible to actually detect a tornado in the velocity pattern through a tornado vortex signature(TVS). The amount of averaging done due to the wide beam, would cause the loss of detail necessary to actually pinpoint the tornado. The reports of a funnel cloud and possible tornado may not definitively verify the existence of a tornado on the ground, but they do indicate that the meso-low tracked by the radar was likely capable of producing a tornado. All that can really be said is that conditions were favorable for tornadic development and the storm studied did indeed show signs of rotation in the form of a meso-low.
Of primary importance is that the severe weather occurred where it was
expected to occur. As mentioned earlier, the most likely place for severe
weather to occur on 22 Aug 1995 was northeastern ND and northwestern MN.
Morning analyses and forecasts indicated the possibility of severe weather.
Radar operations detected and monitored the development and evolution of
the storm(s), enabling forecasters to assist people in decision making.
Cheng Lawerance, and D.C. Rogers, 1988: Hailfalls and hailstorm feeder
clouds-an Alberta case study. J. of Atm. Sci., 45, 3533-3545.
Dennis, A.A., C.A. Schock, and A. Kocielski, 1970: Characteristics of
hailstorms in western South Dakota. J. of Appl. Meteor., 9,
127-135.
Last Modified on 06/97.