On 24 March 1996, an intense negatively tilted upper-level trough moved across the Central Plains region. This trough was accompanied in the lower levels by strong cold air advection that led to the rapid development of locally heavy snow and blizzard conditions across portions of northwest and north-central Kansas into a large part of Nebraska. Intense convection also developed in the warm sector over eastern Kansas, western Missouri and Oklahoma where some thunderstorms became severe and produced strong, straight-line winds and hail. This study will examine the evolution of the low pressure system with observed surface, upper air and satellite data, then compare how well the ETA model from 1200 UTC 23 March performed from a forecaster's point of view.
What made this storm system rather difficult to forecast some 24 hours ahead of time was the fact that the areas of the Central Plains that eventually experienced blizzard conditions on March 24 were reporting high temperatures on March 23 in the 60s, 70s, and even some 80s over northwest Kansas. The 0000 UTC 24 March 1996 high temperature map is shown in Figure 1. It can be seen on this map that the Arctic air mass was already surging southward across Montana, Wyoming and western portions of North and South Dakota where high temperatures were generally in the teens and 20s. At 0000 UTC 24 March, the 500 mb trough was deepening near the four corners area with the geopotential height fall center located near Flagstaff, AZ (not shown). The 300 mb analysis from this time also indicated a 100 kt jet max on the backside of the trough over western Nevada (not shown). The flow over the Central Plains at 500 and 300 mb was southwesterly but somewhat diffluent. At the surface, a low pressure area was located over east central Colorado near Limon with strong southerly flow from Kansas southward into Texas where dew points were increasing into the 50s. Light snow was occurring behind the Arctic cold front across the northern High Plains region.
Figure 1. High temperature plot valid 0000 UTC 24 March 1996, contours every 10 degrees F.
By 1200 UTC 24 March, the Arctic cold front had rapidly moved southward into eastern Colorado, northwest Kansas and western Nebraska, dropping temperatures into the teens and 20s (Figure 2). Six hour pressure tendencies show rises of 2-3 mb were observed in this region. Light and some blowing snows were being reported from western Nebraska into northeast Colorado while light freezing rain was observed over northwest Kansas where the leading edge of the Arctic air was still relatively shallow. New snow amounts ending at 1200 UTC 24 March were generally in the one-to-two inch range from eastern Wyoming into the Nebraska Panhandle and far northeast Colorado. At 500 mb, the trough had moved east-southeastward into northern New Mexico with the 12 hour geopotential height change fall center of 140 meters located near Albuquerque (Figure 3). The 300 mb jet maximum of 110 kts had also rotated into the base of the trough as the flow aloft was strongly backing over the Central Plains as the wave was becoming increasingly negative tilted at this time. The 1215 UTC 24 March MB curve satellite image (Figure 4) shows colder cloud tops extending north of the 500 mb height fall center in the diffluent upper level flow from Colorado northward into Wyoming and western Nebraska.
Figure 2. Surface plot and fronts with mean sea-level pressure analysis (every 4 mb) valid 1200 UTC 24 March 1996.
Figure 3. 500 mb plot and analyses of geopotential heights (every 60 decameters), temperatures (every 5 degrees C), and 12 hour geopotential height falls (every 4 decameters) valid 1200 UTC 24 March 1996.
Figure 4. A satellite image (MB curve) valid 1215 UTC 24 March 1996.
Blizzard conditions quickly developed shortly after 1500 UTC 24 March across northeast Colorado, northwest Kansas and southwest Nebraska as the strong dynamics associated with the upper level wave interacted with the strong low level cold air advection and northwesterly winds on the backside of the surface low pressure now located over central Kansas. Winds by this time were gusting between 35 and 40 kts over this area, producing wind chills of 20 to 30 degrees below zero, where less than 24 hours ago temperatures were in the 60s, 70s and 80s.
By 1800 UTC, the surface low pressure system had moved eastward into northeastern Kansas while the Arctic cold front continued to surge southeastward across Nebraska, central Kansas, northwest Oklahoma and the Texas Panhandle (Figure 5). Six hourly pressures' rises of 2-4 mb were observed behind the front while blizzard conditions continued over portions of northwest Kansas, central and southwest Nebraska and northeast Colorado. Goodland, KS reported visibilities at or below one-quarter mile from 1700 UTC through 0000 UTC 25 March with northwesterly winds of 40 mph with gusts to 55 mph. Temperatures remained in the middle teens which produced wind chill readings of 30 to 40 degrees F below zero. Interstate 70 from Hill City, KS to Limon, CO was closed as was Interstate 80 from Grand Island, NE to east of Cheyenne, WY during the storm.
Figure 5. As in Figure 2 except valid 1800 UTC 24 March 1996.
The MB curve satellite imagery at 1815 UTC (Figure 6) showed cloud tops had cooled rapidly over western Nebraska, northwest Kansas and northeast Colorado where moderate to heavy snow and blizzard conditions were occurring at this time. A line of strong to severe thunderstorms extended along and ahead of the Arctic cold front from eastern Kansas into central Oklahoma as the low pressure circulation took on a comma shape appearance, indicating rapid deepening and intensification aloft.
Figure 6. As in Figure 4 except valid 1815 UTC 24 March 1996.
A dry slot was also beginning to move into western Oklahoma and west Texas while a large area of blowing dust can be seen in the image over the southern Texas Panhandle. At 0000 UTC 25 March, the surface low had accelerated into central Iowa while the Arctic cold front moved rapidly southeastward into western Missouri and north-central Texas (Figure 7). The MB curve satellite imagery at 0015 UTC March 25 (Figure 8) showed a deformation zone of colder cloud tops over northwest Kansas and central Nebraska where blizzard conditions in moderate to heavy snow were occurring. The cold cloud tops were associated with strong upper level divergence near the left exit region of the jet. The line of convection has moved rapidly eastward into Missouri and Arkansas while the leading edge of the Arctic air mass can be seen over south central Oklahoma and northwest Texas. The 500 mb chart for 0000 UTC (Figure 9) also showed the height fall center had moved from central New Mexico at 1200 UTC into eastern Kansas while the trough became negatively tilted. The 0000 UTC 25 March 300 mb analysis (not shown) also indicated the 110 kt jet maximum had rotated around the base of the trough into eastern portions of Oklahoma and Kansas.
Figure 7. As in Figure 2 except valid 0000 UTC 25 March 1996.
Figure 8. As in Figure 4 except valid 0015 UTC 25 March 1996.
Figure 9. As in Figure 3 except valid 0000 UTC 25 March 1996.
The upper level trough continued to deepen through 1200 UTC 25 March over the middle Mississippi Valley region. The heaviest snow also shifted northeastward into Iowa and Minnesota as the surface low pressure system moved into southern Wisconsin. Figure 10 shows the observed snowfall with this storm system with the axis of a heaviest snow band (4-6") extending from northwest Kansas across the southern half of Nebraska.
Figure 10. Plot and analysis (every 2 inches) of observed snowfall for the 24-hour period from 1200 UTC 24 March through 1200 UTC 25 March 1996.
Gridded model output from the 1200 UTC 23 March 1996 ETA run was examined to see how well this significant winter storm was forecast 24 hours in advance of the observed blizzard conditions over portions of the Central Plains. From a forecaster's point of view, some of the weather elements critical to the accuracy of a public-forecast package were related to the movement of the 500 mb vorticity maximum and height fall center for the location of the snow band, the location of the strongest vertical motion fields and the speed and magnitude of the low-level cold advection which would impact the location of critical rain/snow thickness fields. The ETA model forecast of temperatures, heights and winds at 850 mb will also be compared to what was actually observed at both 1200 UTC 24 March 1996 and 0000 UTC 25 March 1996.
The strong upper level dynamics associated with this storm system are evident in the 700-300 mb layer Q-vector divergence field shown in Figure 11. This four-panel diagram shows both the 24 and 36 hour forecasts and the corresponding initial analysis at 1200 UTC 24 March and 0000 UTC 25 March respectively. It can be seen that this upper level trough had impressive dynamics associated with it with a deep layer of quasi-geostrophic forcing for upward vertical motion in the 700-300 mb layer Q-vector divergence field. The 24-hour forecast indicated the strongest forcing in the 700-300 mb layer across northeastern Colorado into the Nebraska panhandle. Far northwest Kansas was on the southern fringe of this strong forcing. The initial analysis from the 1200 UTC March 24 ETA model showed this forcing to be further to the south of the 24-hour forecast.
Figures 11a-d. Eta model 24- and 36-hour forecast 700-300 mb Layer Q-Vector's (arrows) and Q-Vector divergence (dashed) (a and b) and initial analyses (c and d) valid 1200 UTC 24 March and 0000 UTC 25 March 1996 respectively.
The 36-hour forecast valid at 0000 UTC 25 March showed the 500 mb vorticity maximum moving northeastward into central Kansas (not shown) while the 700-300 mb layer Q-vector convergence implied strong forcing for upward vertical motion over eastern South Dakota and north-central Nebraska. This Q-vector divergence forecast was also farther north than what verified as can be seen in the initial conditions from the 0000 UTC 25 March ETA model. The 500 mb trough became negatively tilted after 1200 UTC 24 March, as stronger than a forecast mid-level cold air advection caused the system to dig further to the south than the 1200 UTC 23 March ETA model forecast of 500 mb heights, temperature and vorticity. Satellite data also confirmed this with the colder cloud tops associated with the deformation zone and a heavy snow band located across portions of Nebraska and northwest Kansas by 0000 UTC 25 March. Both Younkin (1968) and Weber (1979) related the location of the heaviest snow band using the projected path of the 500 mb vorticity maximum and the 500 mb height fall center for snowstorms over the central United States. They both found that the heavy snow band was some 1.5 to 2.5 degrees latitude to the left of these features. Figure 12 shows the observed 500 mb height fall center track compared with the ETA model forecast. These methods worked very well for this case, but again had to be adjusted farther south than the 1200 UTC 23 March 1996 ETA model forecast.
Figure 12. Comparison of Eta model forecast and observed 500 mb geopotential height fall center track.
The 850 mb ETA model forecast of heights, temperatures and winds valid for 1200 UTC 24 March and 0000 UTC 25 March were compared with the observed conditions (not shown). The ETA model forecast the location of the 850 mb low fairly accurately, but under forecast the speed and intensity of the cold air advection and the wind speeds over the Central Plains region. The model was about 5 degrees C too warm over the northern and central High Plains region, which caused the model to miss the location of the 0 degrees C and -5 degrees C isotherms by a fairly large distance. These errors in the forecast 850 mb temperatures mirrored the errors in the critical rain/snow thicknesses that were displaced farther northwest than what actually occurred (not shown). The ETA model forecast movement of the 500 mb vorticity maximum/height fall center, the strongest forcing for vertical motion in the 700-300 mb layer Q-vector field and the 850 mb temperatures would imply that the heaviest snow and eventual blizzard conditions would not reach into northwest Kansas, but since the 500 mb vorticity maximum ejected farther south and was more intense, the significant winter weather was also observed farther south.
Overall, the 1200 UTC 23 March 1996 ETA model did a reasonably good job in forecasting the development of a significant late winter/early spring storm over the Central Plains region. This model data was preferred by forecasters in this case over both the NGM and AVN solutions (not shown) since it was more aggressive and faster in moving the Arctic air mass southward. However, model errors in both the speed and magnitude of the low level cold air advection and the eventual track of the 500 mb height fall center and vorticity maximum affected the location of the observed heavier snowfall and blizzard conditions.
The forecast that was issued to the public in northwest Kansas at 2200 UTC 23 March 1996 from the NWSFO in Topeka, KS indicated that there was a good chance for snow during the day Sunday 24 March, along with blowing and drifting conditions. Temperatures were also to turn sharply colder. A winter storm watch was not issued at this time, but a Special Weather statement detailing the expected rapid change to a much colder and windy conditions was issued. Snowfall amounts of 1 to 3 inches were also included in the statement. These forecast products gave an accurate description of the upcoming weather event and gave the public enough lead time to take precautions for travel and to prepare for the arrival of the Arctic air mass.
Weber, E.M., 1979: Meteorological Techniques, Air Weather Service Pamphlet 105-56. Headquarters Air Weather Service, Scott AFB, 2-3.4-1-2-3.5-2.
Younkin, R. J., 1968: Circulation Patterns Associated with Heavy Snowfall Over the Western United States. Mon. Wea. Rev., 96, 851-853.