Violent Tornadoes

Johns et al. (1993 and 2000) and Edwards and Thompson (2000) have explored numerous characteristics associated with violent United States tornadoes and tornadic supercells. Their research has yielded many new findings concerning synoptic scale and mesoscale parameters. Johns et al. (2000) determined the violent tornado’s location compared to the surface low and boundaries. Edwards and Thompson (2000) and Johns et al (1993) looked at various instability, shear and moisture parameters for violent tornadoes and tornadic supercells. The purpose of this study is to build on prior research to increase the knowledge of the characteristics of violent tornado episodes and to determine the similarities and differences among violent tornado episodes across separate geographic regions of the United States. For this study, thirty-eight violent United States tornado episodes including 70 violent tornado tracks were considered from 1993 to 1999. Synoptic weather maps from the initialized ETA model and observational maps have been examined to recognize patterns at the surface and upper-levels. Surface lows, boundaries, atmospheric jets, vorticity maxima and centers of maximum CAPE (Convective Available Potential Energy) have been analyzed on synoptic maps. In addition, wind speeds, instability and moisture parameters have been analyzed. As a result, synoptic and mesoscale patterns favorable for violent tornadoes are presented for different geographic regions of the United States.

For this paper, all 38 United States violent tornado episodes including 70 violent tornado tracks from 1993 to 1999 were examined. The "Storm Data" publication was used to determine the time and number of violent tornadoes that occurred in each event. The program, "Severe Plot", version 2.0 was obtained from the Storm Prediction Center and was used to determine the location of each violent tornado. The resulting United States map is shown below. Each tornado track is in red with each event circled.

To identify similarities and differences among tornado events across separate regions of the United States, five sections of the United States were identified. The map below shows each region that was identified. Some differences of opinion as to which states belong in which regions may arise. However, for consistency, the regions for this paper have been identified on the map below. The Southern Plains are in red, the Northern Plains are in blue, the Upper Midwest is in green, the Southeast is in purple, and the East is in brown. The Southern Plains had 8 violent tornado events including 18 violent tornadoes and the Northern Plains had six violent tornado events including 14 violent tornadoes. The Upper Midwest had 7 violent tornado events including 8 violent tornadoes, the Southeast had 12 violent tornado events including 24 violent tornadoes and the East had six violent tornado events including seven violent tornadoes.

We hand analyzed the surface lows, boundaries, thermal ridges and moisture ridges using observational maps from the National Climatic Data Center (NCDC). To analyze upper-level features, we used the initialized ETA model. We obtained the 1996 to 1999 ETA model data from the Cooperative Program for Operational Meteorology, Education and Training (COMET) sponsored by the National Oceanic and Atmospheric Administration (NOAA). To interpret the 1996 to 1999 data which was GEMPAK data, we used the Scientific Applications Computer (SAC). In addition, for the 1997 to 1999 atmospheric jet study, we used archived ETA maps from the Air Resource Laboratory's (ARL) Realtime Environmental Application and Display System (READY). We obtained the 1993 to 1995 ETA model data from the Scientific Services Division (SSD) of the National Weather Service Southern Region Headquarters. To interpret the 1993 to 1995 data we used PC Gridds. The ETA model was available for all but one event. For the event where the ETA was missing, we substituted the initialized NGM model. For violent tornado events that were within a few hours of 00Z or 12Z, we used the corresponding model initialization. Where the violent tornado events occurred several hours from the initialized model time, we gathered both the model before and after the violent tornado event time. Then we interpolated to get an accurate representation of where the feature was at the time of the violent tornadoes. We then plotted the location of each feature along with the violent tornado on a map. Then, we overlaid a circle with the feature located at the center and plotted the location of the violent tornado within the circle compared to the meteorological feature near the time of occurrence.

The main goal during our research was to determine the location of each violent tornado compared to meteorological features. Some of the features that we considered included the surface low, surface boundaries, surface dewpoints, CAPE centers, 500 mb positive vorticity center, 500 mb negative vorticity center, 850 mb jet, 700 mb jet, 500 mb jet, and 300 mb jet. The following plots show the results of our research. The tornado tracks are shown with the color from the respective region. On all the graphics, the range rings are for every 200 miles. From time to time, we may differentiate between the violent tornado tracks and the violent tornado events when giving statistical figures. When the violent tornado is referred to, it means each individual violent tornado damage track. The violent tornado event includes all violent tornadoes that occurred during the event. In addition, differentiation may be made between the Northern States and Southern States which are divided at the 40th parallel. The Northern States are considered anywhere north of Kansas and vice versa. The east and west will be divided by the Mississippi River and is referred to as east or west of the Mississippi River.

The graphic below shows each violent tornado track's relative position compared to the surface low. After detailed hand analysis, the position of the surface low to each violent tornado was determined. Each violent tornado event's surface low was plotted in the center of the circle with the vertical line being north and south. The resulting pattern shows 91.4 % of the violent tornadoes in the northeast and southeast quadrant of the low. When the events were considered separately, 35.5 % of the violent tornado events occurred in the northeast quadrant while 48.7 % occurred in the southeastern quadrant. When tracks were considered separately, 46.4 % of the violent tornado tracks fell in the northeast quadrant while 45.0 % of the violent tornado tracks occurred in the southeast quadrant. As a result, the northeastern quadrant violent tornado events were more efficient violent tornado producers than the southeast quadrant events. As shown in the table below, the northeast quadrant events produced 2.41 violent tornadoes per event while the southeast quadrant events produced 1.70 violent tornadoes per event. For the southwest quadrant, only 15.8 % of the violent tornado events and only 8.6 % of the violent tornadoes occurred. Notice that most of the Northern Plains violent tornadoes occurred in the northeast quadrant. In fact, most of the Great Plains violent tornadoes occurred in the northeast quadrant. Note, the seven violent tornado tracks in red including the one near the 400 mile range ring were associated with only one event, May 3rd, 1999.

Of the northeastern quadrant violent tornado events, 85.2 % occurred west of the Mississippi River while only 14.8 % occurred east of the Mississippi River. The graphic above shows that most of the northeastern quadrant violent tornadoes were in the Great Plains states. Conversely, of the southeastern quadrant violent tornadoes, only 29.7 % of the violent tornado events occurred west of the Mississippi River while 70.3 % occurred east of the Mississippi River. In the southeastern quadrant, the events occurring west of the Mississippi produced on average 2.21 violent tornadoes while the events east of the Mississippi River produced on average 1.47 violent tornadoes per event. As a result, in the southeastern quadrant, the violent tornado events west of the Mississippi River were more efficient at producing violent tornadoes than events east of the Mississippi River. In the southwest quadrant, 4 of the 6 violent tornado tracks occurred in the Northern States. A table below shows the violent tornado producing efficiency for each quadrant of the low and for areas east and west of the Mississippi River.

The circle plot below shows the surface low center relative to the violent tornado paths. In addition, each violent tornado event is colored to indicate the type of boundary associated with the event. Notice, most of the warm front and cold front cases occurred to the northeast or to the east of the surface low center. All of the prefrontal trough and pressure trough cases occurred south to southeast of the surface low center with the exception of the inverted pressure trough case to the northeast. Two of the dryline cases were northeast of the surface low center and one was to the south southeast. The warm sector cases did not favor a quadrant.

When doing the research, all events appeared to be associated with moisture advection tongues of varying strength. Although somewhat subjective, a technique was developed to plot the circle below to the right. It shows point X determined by the dewpoint advection tongue and moisture gradient in relationship to the violent tornado paths. To execute the technique, the surface dewpoint moisture advection tongue was identified using 2 degree Fahrenheit intervals. The highest dewpoint pool in the tongue was noted and the average surface wind direction at the tip of the area was determined. The wind vector was projected from the tip of the area within the tongue to the first tight moisture gradient with a moisture density of greater than 2 degrees for a 30 mile distance. The point of intersection of the vector to the edge of the tight gradient was designated as point X. An example of this method is shown below to the left for May 3, 1999. Point X is shown in purple with the approximate violent tornado tracks. The analysis below is only valid for the violent tornado that produced F4 or greater damage in and southwest of Oklahoma City because the violent tornadoes farther north occurred much later than this dewpoint analysis. Another dewpoint analysis was done along with interpolation for the violent tornadoes farther north. All the events were done like the May 3, 1999 example. After Point X was determined for all events, Point X was plotted at the center of the circle with each violent tornado family plotted with respect to Point X. In the graphic below to the right, the violent tornado families (not individual tornado paths) are shown relative to Point X. North and south run along the vertical line. Notice the tight relationship found for this technique. Most of the tornadoes were within 100 miles of this point. In fact, 57.1 % of the violent tornadoes were found within the two small areas just southwest and just north of point X. This indicates that a large number of the violent tornadoes formed along tight moisture gradients oriented in a southwest to north or northeast direction downstream from a tongue of surface moisture advection.

A few other characteristics were determined, including the location of the moisture ridge and thermal ridge compared to the violent tornado event's location. In the Great Plains, it was found that 78.6 % of the violent tornadoes occurred between the moisture ridge and thermal ridge. In 85.7 % of the cases, the thermal ridge was west of the moisture ridge. The violent tornado events in the Upper Midwest showed similarities to the Great Plains violent tornado events. Conversely, the Southeast and East violent tornado events were different. Of nearly half or 47.1 % of the violent tornado events, the violent tornadoes occurred west of both the moisture and thermal ridge. Opposite to the Great Plains cases, in 52.9 % of the cases, the thermal ridge was east of the moisture ridge.

The table below shows the average temperature, dewpoint and lifting condensation level at the location of the violent tornado for each region. Notice the Southern Plains average temperature at the point of the violent tornado was much warmer than any other region. This contributed to much higher lifting condensation levels in the Southern Plains. The highest dewpoints with the violent tornadoes were found in the Upper Midwest. In contrast, the lowest temperatures and dewpoints were found in the Northern Plains. Notice the lowest lifting condensation levels occurred east of the Mississippi River.

The circle plot below to the left was generated by locating the CAPE center of maximum instability associated with each event and plotting the position of each violent tornado in relationship to the CAPE center. Although, most of the violent tornado events had a clear CAPE center nearby, a few of the cases had only higher CAPE areas that were not completely closed off. For those cases, we looked for a large area of higher CAPE closed off on at least three sides. These areas were largely evident, and only a minimum of subjective reasoning was needed. Another problem was that the 1993 PC Gridds data did not show CAPE. We studied a sample of events and determined that the CAPE center and Lifted Index center were well related with a minimum of variability. As a result, we substituted Lifted Index for CAPE for the six 1993 events to determine the instability center and orientation. In the graphic to the left, the vertical and horizontal lines represent north-south and east-west directions. It was found that almost two-thirds or 65.0 percent of the violent tornado events in the Southern Plains, Northern Plains, and Upper Midwest were clustered in the northern quadrant within 225 miles from the CAPE center. The violent tornado events in the Southeast and East tended to be further from the CAPE centers. There was a small percentage of events that were more than 400 miles from the CAPE center.

Below to the right, the circle plot was generated by noting where the violent tornado event occurred in relation to the orientation of the instability axis extending from the CAPE center. The low-level wind was determined to verify the orientation of the instability axis. In the graphic to the right, the vertical line represents the direction of the axis of instability associated with the CAPE center and violent tornadoes. The violent tornadoes tended to cluster tightly around the instability axis and were less spread out when compared to the circle plot to the left. Note that the Northern States violent tornadoes tend to occur along or just to the west of the instability axis, while the Southern States violent tornadoes tend to occur along or just to the east of the instability axis. Of all the regions, the Upper Midwest violent tornadoes occurred closest to the CAPE center with the Southeast violent tornadoes spread out the most. In addition, a table below the two graphics shows the average CAPE, precipitable water and 850 mb through 500 mb mixing ratio values at the location and time of the violent tornadoes. For several events, CAPE and precipitable water were not available. Out of 38 violent tornado events, five were left out for CAPE and six were left out for precipitable water. Note the much higher CAPE values and higher 850 mb mixing ratios associated with the Southern Plains and Upper Midwest violent tornadoes. In contrast, note the low CAPE values for violent tornadoes across the Southeast and East. In addition, the precipitable water values, 700 mb mixing ratio and 500 mb mixing ratio were lower for the Great Plains.

Another part of our research focused on the position of the violent tornado in relationship to the orientation of the 500 mb positive and negative vorticity center. North-south direction was not taken into account, only the orientation of the vorticity center. The orientation of the vorticity center was considered perpendicular to the direction of the average geostrophic flow running through the vorticity center or perpendicular to the average 500 mb height contour running through the vorticity center. Most of the time, the orientation was perpendicular to the movement of the vorticity center; however, there was not a perfect relationship because vorticity centers do not always move exactly with the average geostrophic flow. The circle plots below show the vorticity center's orientation relative to the violent tornado's position. We used positive vorticity centers and lobes with a minimum value of 10 or above. In most cases, the associated positive vorticity center was clearly evident. However, a few of the PC Gridds events were harder to interpret because some of the PC Gridds model data had a greater minimum contour interval than the GEMPAK data on the SAC. We used our best judgment to identify each vorticity center. In one case, no positive vorticity center was evident. The circle plot below to the left resulted after we plotted each violent tornado's position with respect to the positive vorticity center's orientation associated with it. A striking pattern emerged showing that over three-quarters or 76.3 % of the violent tornadoes occurred in the lower right quadrant with respect to the orientation of the positive vorticty center, regardless of geographic location in the United States. However, geographic location did play an important role in distance from the positive vorticity center. Most of the Northern Plains violent tornadoes occurred about 200 miles from the positive vorticity center in the lower right quadrant. All but two of the Upper Midwest violent tornadoes occurred about 250 miles in the lower right quadrant while, most of the Southern Plains violent tornadoes occurred about 325 miles away from the positive vorticity center in the lower right quadrant. A general notation was made that as one goes north or northwest across the regional map, the violent tornadoes occurred closer and closer to the positive vorticity center and the value of the vorticity center increased. This is evident in the graphic to the lower left and in the table below that. One possibility in the Great Plains is that the higher buoyancy usually prevalent in the Southern Plains compensated for the weaker dynamics there. In the Northern Plains, the stronger positive vorticity centers and greater dynamics may have compensated for the less overall CAPE and buoyancy there. As we will see in the next section, in the East and Southeast, higher low-level wind speeds may have compensated for the less overall CAPE and buoyancy there. The circle below to the right shows the negative vorticity center in relationship to each violent tornado path. Notice that most of the violent tornadoes occurred in the upper left quadrant with the highest concentration located from just behind the negative vorticity center out to about 300 miles in the upper left quadrant. However, the relationship was not nearly as significant as that found for the positive vorticity center.

A section of research was done on the atmospheric jets associated with the violent tornadoes. Not only was GEMPAK and PCGridds data used, the Air Resource Laboratory's (ARL) Realtime Environmental Application and Display System (READY) was used to display archived atmospheric jet data using the ETA model. One advantage from using READY was that three hour increments using interpolation of the ETA model were available instead of just 12 hour initializations. First, the position of the closest atmospheric jet core to the violent tornado event was identified. Then, the position of the highest wind speed in the jet at or nearest to the time of the violent tornado was plotted for 850 mb, 700 mb, 500 mb, 300 mb. Interpolation was used if the violent tornado event fell in between model time increments. Then, the orientation of the jet was noted by determining the average wind direction running through the core of the jet at that level. As a result, the orientation is parallel to the average wind direction running through the jet. After plotting the jet position and orientation, the circle plots below were generated. In each graphic, the center of the atmospheric jet for each violent tornado event is in the center of the circle with each violent tornado plotted with respect to the jet's position. The orientation of each jet is along the vertical line in the center of the circle. Note that the atmospheric jet's movement is not always parallel to the jet's orientation. North-south direction was not taken into account. The nose of each jet would be in the middle to top portion of the circle plot.

The circle plot below shows each violent tornado path relative to the 850 mb jet. Again, the movement of the jet is up along the vertical line with the jet core at the center. Notice the strong relationship of violent tornadoes to the position of the 850 mb jet. Most of the violent tornadoes occurred to the left of the 850 mb jet. Most of the Great Plains and Upper Midwest violent tornadoes occurred in the left front quadrant of the jet. Two of the Upper Midwest violent tornadoes occurred well to the right of the 850 mb jet. The violent tornadoes in the Southeast occurred both in the left front and left rear quadrants of the jet, while the Eastern States occurred mainly in the left rear quadrant.

The circle plot below shows the relationship of violent tornadoes to the position of the 700 mb jet. It shows many of the violent tornadoes occurred in the jet nose, just ahead of the jet core or in the left front quadrant of the 700 mb jet. The majority of the Southern Plains violent tornado events occurred just ahead of the jet core. In contrast, most of the Northern Plains and Eastern States violent tornado events occurred to the left or in the left front quadrant of the 700 mb jet. The Southern Plains, Upper Midwest and Southeast violent tornadoes were mostly concentrated along and just to either side of the axis of the 700 mb jet. Also, note the many Southeast violent tornadoes that occurred in the right rear quadrant.

The circle plot below shows the relationship of violent tornadoes to the 500 mb jet. In this graphic, many of the violent tornadoes occurred in the jet nose or in the left front quadrant of the 500 mb jet. A few violent tornado events occurred to the right of or in the right rear quadrant of the 500 mb jet. Similar to the 700 mb jet, many of the Southern Plains violent tornado events occurred just ahead of the jet core in or near the jet nose. Again, like 700 mb, most of the violent tornadoes in the Northern Plains and East occurred to the left of the jet axis or in the left front quadrant. Also, note that most of the Upper Midwest violent tornadoes occurred from about 125 miles to the right of the jet core to about 125 miles just ahead of the jet core.

The circle plot below shows each violent tornado path relative to the 300 mb jet. This graphic shows no definite pattern, unlike the circle plots at lower levels. However, a weak pattern can be seen with many of the violent tornadoes occurring in the left front quadrant or to the right of the 300 mb jet. Similar to the 850, 700 and 500 mb, most of the Northern Plains violent tornadoes occurred in the left front quadrant. Also, note that most of the Southern Plains violent tornado events occurred in the left front quadrant or just to the right of the 300 mb jet. Most of the Southeast violent tornadoes occurred in the rear quadrants.

In the table below, average wind speeds at 850 mb through 300 mb at the location and time of the violent tornadoes are shown. In the second table below, the average wind speed difference between different levels is shown. One thing to note is that the Great Plains violent tornadoes were associated with lower wind speeds at low to mid-levels than those found further east. The higher CAPE and buoyancy in the Southern Plains and the greater dynamics in the Northern Plains may have compensated for the weaker low to mid-level wind speeds found in the Great Plains, whereas the higher wind speeds at low to mid-levels in the Southeast and East may have helped compensate for the weaker CAPE and buoyancy there. In the second table below, note that the highest low to mid-level speed shear was located in the East. It is a possibility that this was a factor coupled with low Lifting Condensation Levels that helped violent tornadoes in the East where dynamics were relatively weak and CAPE was relatively low. Also, note in the second table that the average speed difference from 850 to 500 mb had a fairly tight range of around 16 knots in the East to around 23 knots in the Southeast with an overall average of about 20 knots.

Geographic Region
Southern Plains
Northern Plains
Upper Midwest
Southeast
East

850 to 700 mb
Speed Shear
        10.5
          8.1
          7.2
          6.8
        14.4

700 to 500 mb
Speed Shear
          8.5
        13.4
        13.0
        16.3
          1.4

500 to 300 mb
Speed Shear
          8.4
        19.6
          9.7
          8.1
        11.2

850 to 500 mb
Speed Shear
        19.0
        21.5
        20.2
        23.1
        15.8

700 to 300 mb
Speed Shear
         16.9
         33.0
         22.7
         24.4
         12.6

In the graphic below, the actual wind speeds are shown as a dot colored for each region. Note that at 850 mb and 700 mb the range of wind speeds is the least. The Great Plains show the weakest wind speeds at low to mid-levels. Of the 11 lowest speeds at 850 mb and 700 mb, 81.8 % are Great Plains violent tornado events. In contrast, of the top 10 highest wind speeds at 850 and 700 mb, 90.0 % are in the Southeast or East where dynamics and instability were relatively low. The trend also continued up to 500 and 300 mb. Out of the lowest 12 wind speeds at 500 mb, nine were Great Plains violent tornado events. Out of the top 11 wind speeds at 500 mb, eight were violent tornadoes in the Southeast or East. A similar but much weaker pattern shows up at 300 mb.

The graphic below shows the relationship between CAPE and 700 mb wind speeds associated with the location and time of the violent tornado. Notice the high concentration of violent tornadoes that occurred with CAPE values of less than 2000 J/kg and 700 mb wind speeds above 40 knots. Most of these violent tornado events were in the Southeast and East. In contrast, most of the Great Plains violent tornadoes were associated with relatively low 700 mb wind speeds. In the Southern Plains, CAPE was generally high and in the Northern Plains CAPE was generally low. The lower bound threshold for CAPE was around 450 J/kg. All but three violent tornado events occurred with 700 mb wind speeds of higher than 30 knots with almost two-thirds or 63.6 % of the violent tornado events occurring between 40 and 55 knots.

Many similarities and differences were found among violent tornado episodes across separate geographic regions of the United States. The following summarizes this paper's results. With respect to the surface low, most violent tornado episodes in the Great Plains occurred in the northeast quadrant. In contrast, most violent tornado episodes east of the Mississippi River occurred in the southeast quadrant or southwest quadrant of the surface low. With respect to surface boundaries, each region had boundaries that favored that particular region. For example, prefrontal troughs favored the Southeast, cold fronts favored the Southern Plains and Southeast, drylines were solely in the Southern Plains, warm fronts and pressure troughs favored the Northern States, and the inverted trough case was solely in the Northern Plains. As far as low-level moisture is concerned, most of the violent tornado episodes were associated with low-level moisture advection tongues. Typically, the violent tornado event occurred downstream of the moisture advection tongue near a tight moisture gradient often, but not necessarily associated with a boundary. The violent tornadoes in the Great Plains tend to be between the moisture and thermal ridge with the moisture ridge to the east. In contrast, the violent tornadoes east of the Mississippi River commonly occurred west of both the moisture and thermal ridge with the thermal ridge east of the moisture ridge. The Great Plains had higher lifting condensation levels, lower precipitable water values and lower 500 mb and 700 mb mixing ratios while the Upper Midwest, Southeast and East had lower lifting condensation levels, higher precipitable water values and higher 500 mb and 700 mb mixing ratios. With respect to CAPE maximum centers, most of the violent tornadoes occurred from northwest to northeast of the CAPE center from the CAPE center out to 325 miles. The violent tornado episodes were plotted tighter around the orientation of the CAPE axis than they were directionally from the CAPE center. The Southern Plains and Upper Midwest were typically higher CAPE cases and the Southeast and East were typically lower CAPE cases. Most of the Northern States violent tornadoes occurred just to the west of the instability axis whereas most of the Southern States violent tornadoes occurred just to the east of the instability axis.

Moving up to the 500 mb level, with respect to the 500 mb positive vorticity center's orientation perpendicular to the average 500 mb height, almost all of the violent tornadoes occurred in the lower right quadrant from the positive vorticity center out to 400 miles. The Northern Plains and to some extent the Upper Midwest violent tornadoes were associated with higher vorticity center values, corresponding to greater dynamics there. And considering atmospheric jets, most of the violent tornado episodes were clustered tightly to the left of the 850 mb jet from the middle of the left front quadrant down to the middle of the left rear quadrant out to 225 miles from the 850 mb jet center. With respect to the 700 mb jet, most violent tornado episodes were clustered in the nose, in the left front quadrant or along the axis of the 700 mb jet. And with respect to the 500 mb jet, most violent tornado episodes were clustered in the left front quadrant and to the right of the 500 mb jet. The 300 mb jet circle plot showed more events occurring in the left front quadrant and to the right of the jet core; however, it was nearly an indistinguishable pattern. Finally, considering wind speeds, the violent tornadoes in the Great Plains were generally associated with weaker low to mid-level wind speeds which may have been compensated in the Southern Plains by higher CAPE environments and in the Northern Plains by greater dynamics. The violent tornadoes in the Southeast and East were associated with greater low to mid-level wind speeds which may have helped compensate for the lower CAPE environments.

The objective of this paper was to give the forecaster a clearer idea of how the characteristics of violent tornado episodes are both similar to each other and different from each other across separate geographic regions of the United States. More research is needed to further develop our understanding of violent tornado events. Radar could be examined for these violent tornado episodes. In addition, sounding analysis could be examined showing wind shear parameters and mesoscale environments in greater detail. With continued research, a greater awareness of the synoptic and mesoscale characteristics associated with violent tornadoes will emerge which should help us better forecast these deadly events in the future.

We would like to thank Bernard Meisner of the Scientific Services Division (SSD) at Southern Region Headquarters of the National Weather Service for getting us PC Gridds model data. Also, we would like to thank Patrick O'Reilly and Liz Page of the Cooperative Program for Operational Meteorology, Education and Training (COMET) for getting us model data for the Scientific Applications Computer. In addition, Sam McCown of the National Climatic Data Center (NCDC) provided surface observational maps for hand analysis which we appreciate. Finally, we would like to thank Ed Andrade for editing the paper in a web format.

Bluestein, H. B., 1993: Synoptic-Dynamic Meteorology in Midlatitudes, Volume II, Observations and Theory of Weather Systems, Oxford University Press, 465-526.

Doswell, C. A., III, S. J. Weiss, R. H. Johns, 1993: Tornado Forecasting: A Review. The Tornado: Its Structure, Dynamics, Prediction, and Hazards. Geophys. Mongr., 79, Amer. Geophys. Union, 557-571.

Edwards, R., and R. L. Thompson, 2000: RUC-2 supercell proximity soundings, Part II: an independent assessment of supercell forecast parameters. Preprints, 20th Conf. Severe Local Storms, Orlando FL, Amer. Meteor. Soc., 435-438.

Grazulis, T. P., 1997: Significant Tornadoes Update, 1992-1995, The Tornado Project of Environmental Films, 1392.

Guerrero, H., J. C. Broyles, and D. Eastlack, 1998: Forecasting tornado location across the Dakotas and Minnesota, Preprints, 19th Conf. Severe Local Storms, Minneapolis, MN, Amer. Meteor. Soc., 301-304.

Johns, R. H., J. C. Broyles, D. Eastlack, H. Guerrero, K. Harding, 2000: The role of synoptic patterns and temperature and moisture distribution in determining the locations of strong and violent tornado episodes in the north central United States: a preliminary examination, Preprints, 20th Conf. Severe Local Storms, Orlando FL, Amer. Meteor. Soc., 489-492.

Johns, R. H., J. M. Davies, and P.W. Leftwich, 1993: Some wind and instability parameters associated with strong and violent tornadoes. Part II: Variations in the combinations of wind and instability parameters. The Tornado: Its Structure, Dynamics, Prediction, and Hazards. Geophys. Mongr., 79, Amer. Geophys. Union, 583-590.

Leftwich, P. W. and W. R. Sammler 1986: Forecasting violent tornadoes, Preprints, 11th AMS Conference on Forecasting and Analysis, 114-118.

National Oceanic and Atmospheric Administration (NOAA), U.S. Department of Commerce, 1993-1999: Storm Data

Whiting, R. and R. Bailey, 1957: Some Meteorological Relationships in the Prediction of Tornadoes. Monthly Weather Review, 85, 141-149.

Dewpoint Analysis of May 3, 1999 with Point X	Point X with Respect to the Violent Tornadoes

CAPE Max Center with Respect to Direction	CAPE Max Center/Instability Axis Orientation

500 mb Positive Vorticity Center Orientation	500 mb Negative Vorticity Center Orientation