Jeffrey A. Chapman and Ronald T. Holmes
National Weather Service Forecast Office
Sioux Falls, South Dakota



On June 8, 1997, area balloonists participated in the annual Great Plains Balloon Race at the Tea Airport located just south of Sioux Falls, South Dakota (FSD). This annual race features balloonists from across the country who participate in various races. During one of the races, a very strong, long-lived outflow boundary swept through the area producing surface winds in excess of 25 mph. This caused considerable consternation among the balloonists who were caught up in the strong winds and forced to land at speeds approaching 30 mph. Fortunately, no one was hurt but race participants were spilled onto the ground as the balloons hit hard. Some baskets were damaged as they were dragged for considerable distances.

One of the most surprising facts about the outflow boundary was its intensity and longevity, considering the great distance the boundary moved away from its source region. It was originally produced from thunderstorms from as far east as south central Minnesota and propagated west arriving at the race site some four-to-five hours later. It then continued west into south-central South Dakota for another two-to-three hours. This paper will highlight the synoptic and mesoscale conditions that led up to this event and detail the strength and long-lived nature of the boundary through Doppler radar images and mesoscale analyses of automated surface observing stations.



The large-scale pattern exhibited an irregular and broad 500 mb closed low across the upper Ohio Valley with a narrow, high-amplitude ridge sprawled across the western High Plains. A lobe of lower heights extended back across the western Great Lakes with an associated 55 knot wind maximum upstream over northern Minnesota. The synoptic scale surface pattern featured high pressure across the northern Great Lakes with lower pressure over the Rocky Mountains. This pressure pattern was supported by an ascent/descent forcing couplet shown by the Q-vector divergence field with large-scale subsidence over southwest Minnesota and ascent over southeast South Dakota (Figure 1). The result was a maintenance of the surface high pressure system to the east and a persistent east to northeasterly flow at low levels up through 850 mb, particularly over southern Minnesota (Figure 2).

The large scale forcing for descent over Minnesota also helped to maintain a subsidence inversion around 700 mb with a considerable mass of dry, unstable air below this level seen on the 1200 UTC 7 June 1997 sounding from Chanhassen, MN (MPX) (Figure 3). Nearly dry adiabatic lapse rates were present in the 1000-700 mb layer and the 12-hour forecast for 0000 UTC 8 June indicated these unstable lapse rates would continue over south central Minnesota (Figure 4). A shallow layer of moisture was present in the sounding just below the subsidence inversion but cloud bases were high due to the deep layer of dry air below. The high-based nature of the moisture and the subsidence inversion aloft resulted in minimal convective instability despite unstable lapse rates below 700 mb. In fact, a forecast of 3-5 degree C cooling within the 700-500 mb layer during the afternoon only produced a convective available potential energy (CAPE) of 250 J kg-1. This meant that if any convection did develop it would be high-based with a considerable amount of negative energy on the sounding. Any precipitation falling into the deep, dry layer below would result in strong, evaporatively-cooled downdrafts.

Figure 1. Analysis of 850-700 mb Q-vectors (arrows) and divergence of Q (solid divergent, 10-15 K/m2s) valid 0000 UTC 8 June 1997.

Figure 2. Analysis of 850 mb geopotential height (solid, dkm), wind (barbs, knots) and isotachs (dashed, knots) valid 0000 UTC 8 June 1997.

Figure 3. 1200 UTC 8 June 1997 upper air sounding from Chanhassen, MN (MPX). Winds are plotted in knots and temperatures in Centigrade.

Figure 4. 12h Nested Grid Model (NGM) forecast of 1000-700 mb lapse rate (C/km) valid 0000 UTC 8 June 1997.



Surface winds around the FSD County Warning Area (CWA) during the afternoon preceding the race was very light (4-7 knots) and at times were calm. Clear skies resulted in abundant heating and generation of unstable lapse rates near the surface. The unstable conditions and light wind flow resulted in horizontal convective rolls that were observed on the 2238 UTC 0.5 degree base reflectivity product from the FSD WSR-88D radar during the afternoon (Figure 5). These nearly equally spaced bands of enhanced reflectivity were oriented from north-northwest too south-southeast across the radar coverage area.

Figure 5. Sioux Falls (KFSD) WSR-88D 0.5 degrees .54nm resolution base reflectivity image valid 2238 UTC 7 June 1997. The radar was operating in clear-air detection mode.

It has long been established that the convergence zone between two horizontal convective circulations results in upward vertical velocity and enhanced reflectivity along this zone, with subsidence and weaker reflectivity to either side. More recently, Weckwerth et al. (1996) showed that these convergence zones were preferred locations for convective development when the sea breeze front circulation was superimposed upon the rolls (Figure 6). In this case however thunderstorms did not form along the convergence lines associated with the horizontal rolls as the gust front moved through. A major limiting factor for deep convective development during this intersection was the lack of sufficient moisture in the low levels for parcels to reach the level of free convection, thus only an enhanced line of reflectivity occurred as the gust front intersected the horizontal convective rolls.



Early in the afternoon at 1951 UTC, the 1.1 nm 0.5 degree KFSD base reflectivity product showed isolated thunderstorms forming across south central and southeast Minnesota (Figure 7). Analysis of surface observations and radar data suggest this was the genesis region of the gust front. As noted above, this region was also favorable for elevated convection above a dry, unstable layer where evaporational cooling of falling precipitation would enhance downdrafts. Together, the downdrafts from these and other isolated cells coalesced into a large, long-lived gust front that propagated a considerable distance away from the source region. This gust front maintained its strength as it swept through the balloon race some four to five hours later and eventually continued westward past Mitchell (MHE) and Huron (HON).

Figure 6. Vertical cross-section of ground-relative wind and radar reflectivity through horizontal convective rolls and a sea-breeze front (taken from Weckwerth et al. 1996).

Figure 7. Same as Figure 5, except 1.1nm resolution valid 1951 UTC 7 June 1997.

One plausible explanation for the longevity and strength of this boundary may be that the westward moving outflow, analogous to a density current, encountered little resistance, and in fact was most likely propelled by the momentum from east to northeast synoptic scale flow in the low levels. In addition, interaction between the movement of the convectively generated cold pool in the same direction as the ambient shear vector in the low levels (pointed from east to west) likely favored an accelerating boundary to the west. The enhancement of the westward driving connection was also aided by forcing for large-scale subsidence in the low static stability environment (Bluestein 1992).

A combination of radar data and Automated Surface Observing System (ASOS) reports were used to detail the structure and movement of the gust front. This high resolution data revealed that there were actually two boundaries propagating across southern Minnesota. Mesoscale analysis at 2100 UTC (Figure 8) depicted one serpentine boundary from near Marshall, MN (MML) to Spencer, IA (SPW), and another trailing approximately 90km to the east. An interesting feature with this event was that the deformation of the boundaries had a strong correlation to topography, and in particular the Buffalo Ridge, which also follows a serpentine course across southwest Minnesota (Figure 9).

Figure 8. Mesoscale surface analysis valid 2100 UTC 7 June 1997 with temperature dashed lines (interval 2 degree C) and altimeter solid lines (interval .02 inches). Location of boundaries indicated by bold sawtooth contour. Wind barbs are in knots. Cloud cover is coded in the center of station model with none being clear, a single line few too scattered, and two lines broken.

Figure 9. Shaded topographic relief map of a region. Higher terrain indicated by shades of red and lower terrain in yellow. Stations indicated on the map include Sioux Falls, SD (FSD); Brookings, SD (BKX); Pipestone, MN (PIP); Marshall, MN (MML); Worthington, MN (OTG); Windom, MN (MWM); and Spencer, IA (SPW).

The first outflow boundary became discernable on the eastern edge of the radar umbrella just after 2238 UTC (not shown), however, due to the shallow nature of the boundary it was not fully resolved until 2308 UTC as it moved closer to the radar (Figure 10). Although the first boundary is hard to distinguish from surrounding convective rolls on the reflectivity image, the velocity data marks the leading edge of the boundary more clearly with stronger inbound winds over southwest Minnesota. This can be seen as a curved line of inbound velocity data from just north of Pipestone (PIP) to Worthington (OTG) with stronger (20-25 kt) velocity between these two sites (Figure 11). Westward propagation of the boundaries was consistent with trailing easterly wind velocities resulting in weak system-relative convergence. The deformations of the boundaries along the topography and the system-relative convergence both support density current characteristics associated with this long-lived gust front.

The second boundary became evident as a very thin line of reflectivity on the eastern edge of the radar coverage area by 2327 UTC (not shown). Figure 12 depicts the boundaries at 0000 UTC 8 June, by which time the separation between the two had closed to an average of 20 km. The reflectivity at 0006 UTC (Figure 13) showed two strong boundaries to the east of the radar and the velocity data (Figure 14) showed two areas of strong inbound velocity just behind the leading edge of each boundary. Between 0000 and 0100 UTC this second, a stronger boundary accelerated west and caught up to the first boundary. Figure 15 shows the second boundary closing on the first as it moved past the race site and the velocity data (Figure 16) shows a widespread area of strong (>26 knots) inbound velocity over the area. Note how the horizontal convective rolls dissipate due to turbulent mixing behind the boundaries and the increase in outbound velocity as the boundaries traversed the area.

Figure 10. Same as Figure 5, except valid 2307 UTC 7 June 1997.

Figure 11. Same as Figure 10, except .54nm base velocity. Green and blue shadings indicate inbound velocity while yellows and oranges indicate outbound.

Figure 12. Same as Figure 8, except valid 0000 UTC 8 June 1997.

Figure 13. Same as Figure 5, except valid 0006 UTC 8 June 1997.

Figure 14. Same as Figure 11, except valid 0006 UTC 8 June 1997.

Figure 15. Same as Figure 5, except valid 0115 UTC 8 June 1997.

Figure 16. Same as Figure 11, except valid 0115 UTC 8 June 1997.

Figures 17 and 18, respectively, present time series of 20 minute observations from Windom (MWM) and Worthington, MN (OTG) and reveal differing characteristics of each boundary. The former wind shift was associated with a minor pooling of moisture, but no appreciable disturbance in temperature and pressure fields. The trailing discontinuity had characteristics of a more classic density current in 1) a sharp pressure rise; 2) a temperature drop; and 3) a distinct change in wind velocity.

Figure 17. Time series of 20 minute resolutions automated surface observations from Windom, MN (MWM). Temperature (solid) and dew point (dashed) in degree C are indicated in the upper chart. Altimeter (inches) and wind barbs (knots) are plotted in the lower chart. Gusts are noted by "G", and passage of the first and second boundaries by B1 and B2, respectively.

Figure 18. Same as Figure 17, except for Worthington, MN (OTG).

The FSD ASOS station afforded an opportunity to observe the passage of the two boundaries with a five minute resolution. The wind perturbation from the east at 15 knots associated with the leading boundary occurred around 0040 UTC (Figure 19). After gradually easing to 10 knots through 0120 UTC, passage of the second and stronger discontinuity pushed winds to 20 to 25 knots for the next hour, past local sunset time of 905pm CDT.

Figure 19. Same as Figure 17, except for Sioux Falls, SD (FSD) and at 5 minute resolutions.



The gust front that swept through the Great Plains Balloon Race turned out to be a significant mesoscale event, especially to race participants! The synoptic environment in which this event occurred was quite tranquil and was what normally would be considered a "fair weather" situation thought to require minimal analysis and observation. Given the influence of a large, synoptic scale high pressure system over the area and surface dew point around 10 degree C (mid 40's to low 50's), there was little evidence for strong afternoon convection. The MPX morning sounding and forecast lapse rates for the afternoon suggested the possibility for weak afternoon convection over eastern Minnesota bit it was certainly surprising that a gust front of such magnitude and longevity was produced.

High resolution ASOS and WSR-88D data was crucial in detecting the "surprise" gust front that moved into the FSD CWA during the race. Forecasters should not be lulled into a false sense of security under fair weather regimes and should be vigilant in using all data to keep a good met-watch. In this case, the base reflectivity product was not as useful as the velocity data because the approaching gust front was oriented nearly parallel to and "blended" with existing horizontal convective rolls. This made it difficult to discern the structure and strength of the boundary. However, base velocity data revealed more information, especially when compared to surface observations, by showing more definition in the structure and strength of these boundaries. A loop of base velocity data enabled forecasters to see more clearly the leading edge of strong winds approaching the race site.



Bluestein, H.B., 1992: Synoptic-Dynamic Meteorology in Midlatitudes: Principles of Kinematics and Dynamics, Vol. 1, 331p.

Weckwerth, T.M., N.A. Atkins, and R.M. Wakimoto, 1993: Convection Initiation at the Intersections of Convective Rolls and the Sea Breeze Front: A Detailed Kinematic Analysis. Preprints, 26th International Conference on Radar Meteorology, Norman, AMS (Boston), 501-503. is the U.S. government's official web portal to all federal, state and local government web resources and services.