Mothers Day Severe Weather Outbreak of 2006
Across the Piedmont of the Carolinas
Patrick D. Moore
NOAA/National Weather Service
A tornado caused significant damage to this barn near the Rimer community of northern Cabarrus County, North Carolina, at 248 PM on May 14, 2006. Photo courtesy of Greg Kahn, Independent Tribune of Concord & Kannapolis, used by permission.
Author's Note: The following report has not been subjected to the scientific peer review process.
A tornado touched down briefly over northern Cabarrus County, North Carolina, near the community of Rimer, at approximately 248 pm EDT (1848 UTC) on Mothers Day of 2006, snapping several trees and blowing the roof off a barn. [All times in this document are referred to in Universal Time Coordinated (UTC), which is Eastern Daylight Time plus four hours.] A damage survey rated the tornado at F1 intensity on the Fujita Scale. The severe thunderstorm that spawned the tornado was one part of a larger outbreak of severe thunderstorms that raked across the Carolinas from the Piedmont to the Coastal Plain during the afternoon and evening of Sunday, 14 May 2006 (Fig. 1). (Click here to view a storm summary from the National Weather Service (NWS) Office in Raleigh, NC.) In all, the NWS Weather Forecast Office (WFO) in Greer, South Carolina, issued 18 warnings for severe thunderstorms (13) and tornadoes (5) for the event. The Rimer tornado was preceded by a Tornado Warning for northern Cabarrus County issued by WFO Greenville - Spartanburg (GSP) at 211 PM, providing a lead time of 37 minutes.
(Click here to view a summary of severe weather reports for 14 May 2006.
Figure 1. Wind damage, large hail, and tornado reports for 14 May 2006. Click on image to enlarge.
The Rimer Tornado is interesting due to its relatively close proximity to the Terminal Doppler Weather Radar (TDWR) located north of the Charlotte - Douglas International Airport. The data from the Charlotte TDWR was recently made available to WFO GSP through an agreement between the NWS and the Federal Aviation Administration. In general, the location of the Charlotte TDWR affords weather forecasters at the NWS a better view of the low level structure of thunderstorms over the Piedmont of North Carolina to the north and east of the Charlotte metropolitan area, as compared to the Weather Surveillance Radar-88 Doppler (WSR-88D) located in Greer, South Carolina. As we will see in later sections, the TDWR played an important role in the success of warning operations during the event on 14 May 2006.
2. Synoptic Features and Pre-Storm Environment
A large upper level vortex spun over the Great Lakes and Midwest for several days leading up to the event. By the morning of 14 May, cold air aloft associated with the upper low contributed to increased instability and steeper temperature lapse rates across the Carolinas (Fig. 2). Meanwhile, a mid-level jet streak rotated around the upper low and advanced over the mid-Mississippi Valley and Tennessee Valley regions. The surface analysis from the Hydrometeorological Prediction Center (HPC) at 1200 UTC showed a weak warm front lifting north across North Carolina with a cold front approaching from east Tennessee (Fig. 3). The upper air observation taken at Greensboro, North Carolina (GSO), at 1200 UTC indicated the potential for moderate instability and shear in the afternoon (Fig. 4). As a result, the Storm Prediction Center (SPC) included the Piedmont in a slight risk for severe thunderstorms in the morning convective outlook.
Figure 2. SPC objective analysis of 500 mb geopotential height, temperature, and wind at 1200 UTC 14 May. Click on image to enlarge.
Figure 3. HPC surface pressure and fronts analysis at 1200 UTC 14 May. Click on image to enlarge.
Figure 4. Skew-T log P diagram (upper left) and hodograph (upper right) for upper air sounding at GSO at 1200 UTC 14 May. The tables at the bottom summarize several objective parameters used by the SPC to determine severe weather potential. Click on image to enlarge.
A relatively clear sky in the early morning across the Foothills and Piedmont of North Carolina allowed for solar heating to destabilize the lower part of the atmosphere. Low level moisture moving northward in the warm sector across the Carolinas, as evidenced by the dewpoints in the mid to upper 50s on the surface analysis at 1500 UTC, contributed further to the increasing instability over the Piedmont. By 1500 UTC, visible satellite imagery showed the development of cumulus clouds across western North Carolina (Fig. 5). Over the next hour, showers and thunderstorms fired in the warm sector over the Foothills.
Click here to view a 11 frame java loop of GOES-12 visible satellite imagery.
Figure 5. Visible satellite image from GOES-12 at 1445 UTC 14 May.
The environment became increasingly favorable for severe thunderstorms across the Piedmont by midday. The Storm Prediction Center updated the Day 1 Convective Outlook at 1610 UTC, followed shortly thereafter by a Mesoscale Discussion at 1620 UTC, highlighting the increasing atmospheric shear with the approach of the mid-level jet streak from the west. By this time, WFO GSP had already issued the first Severe Thunderstorm Warning for the event, for southern Chester County, South Carolina. By 1700 UTC, surface based Convective Available Potential Energy (CAPE, a measure of the potential energy of rising air parcels related to buoyancy) reached the expected level of 1000 J/kg with very little Convective INHibition (CINH, a measure of resistance to convection at low levels) (Fig. 6). Meanwhile, wind shear in the layer from the surface to 6 km above ground level (AGL) grew in excess of 60 kt (Fig. 7) in response to the mid-level jet streak rotating around the upper low. As a result, Storm Relative Helicity (SRH, a measure of the potential for rotation in thunderstorm updrafts) in the layer from the surface to 3 km AGL topped 200 m2/s2 (Fig. 8). Shear and SRH of this magnitude are generally indicative of an environment favorable for the development of supercell thunderstorms.
Figure 6. SPC objective analysis of surface based CAPE and CINH for 1700 UTC 14 May. Click on image to enlarge.
Figure 7. SPC objective analysis of 0-6 km shear and shear vector at 1700 UTC 14 May. Click on image to enlarge.
Figure 8. SPC objective analysis of 0-3 km SRH and projected storm motion at 1700 UTC 14 May. Click on image to enlarge.
By the early part of the afternoon, the threat for an outbreak of severe thunderstorms, including supercells, was apparent to the forecasters at the SPC. At 1720 UTC, a Severe Thunderstorm Watch was issued by the SPC for most of the Piedmont of North Carolina to the north and east of the Charlotte metropolitan area. Over the next hour, the surface boundary stretched west to east across the Piedmont of North Carolina, further enhancing moisture convergence and storm relative helicity across and north of the Charlotte metro area.
3. Radar observations of the Western Piedmont Supercell
At 1657 UTC, a thunderstorm developing near the South Mountains very close to the surface boundary prompted the issuance of a Severe Thunderstorm Warning for southeast Burke, northern Cleveland, Catawba, and Lincoln counties in North Carolina (Fig. 9). Large hail was produced along the Burke - Cleveland county line shortly after 1700 UTC.
Click here to view a 27 frame java loop of 0.5 degree reflectivity from the KGSP radar.
Figure 9. Radar reflectivity on 0.5 degree scan from the KGSP WSR88-D at 1657 UTC. The radar is located off the bottom left corner of the image. Click on image to enlarge.
The severe thunderstorm gained supercell characteristics as it moved nearly due east along the Catawba - Lincoln county border from 1706 UTC to 1731 UTC, including the development of a deep mesocyclone, weak echo region, and a well-defined hook echo. Additional reports of large hail were received from both sides of the county border through this time. The path of the supercell would take it farther away from the WSR-88D at the GSP airport (KGSP), and much closer to the TDWR north of Charlotte (TCLT). By 1731 UTC (Fig. 10), the hook echo was located 71 nautical miles distant from the KGSP radar, but only 23 nautical miles to the northwest of the TCLT radar. The movement of the supercell brought the center of the severe thunderstorm closer and closer to TCLT and farther and farther away from KGSP after 1731 UTC.
Click here to view a 16 frame java loop of 1.0 degree reflectivity from the TCLT radar.
Figure 10. Radar reflectivity on 0.5 deg scan from KGSP WSR-88D at 1731 UTC. The KGSP radar is located at the bottom left corner on the image. Note the location of the TCLT radar in northern Mecklenburg County. Distance to the hook echo from both radars is shown by light brown lines on the figure. Click on image to enlarge.
When viewing Doppler radar data, there are certain aspects that must be considered related to how the radar beam propagates through the atmosphere. One of the physical limitations is that the radar beam spreads out with increasing distance from the antenna (Fig. 11). As a result, at great distance from the radar antenna, the radar sampling volume is larger than it would be at a distance closer to the antenna. The larger sampling volume can obscure some of the smaller scale details of a thunderstorm.
Figure 11. The effect of the spread of the radar beam with increasing distance from the antenna. Graphic courtesy of NWS Southern Region Headquarters.
For example, compare the view of the low level reflectivity from the KGSP radar in Figure 10 with the low level reflectivity from the TCLT radar taken at 1730 UTC (Fig. 12). The smaller sample volume of the TCLT radar allows for more detail to be seen in the hook-like appendage on the storm's southwest flank.
Figure 12. Radar reflectivity at 1.0 degree scan from the TCLT TDWR at 1730 UTC. Click on image to enlarge.
Under normal atmospheric conditions, the radar beam refracts slightly as it propagates away from the antenna, but still climbs in altitude. The increase in altitude of the beam with distance from the antenna is made even greater by the curvature of the Earth (Fig. 13). Thus, the lowest scan of the radar (0.5 deg for most WSR-88Ds) may only cut through the mid-levels of a distant thunderstorm target (beyond 90 nm or so), or shoot over the top of shallow convection at great distances, missing it entirely. In general, the closer the radar is to the thunderstorm, the better it will be able to scan the lowest levels of the storm. As a result, the TCLT radar had a much better "view" of the low level structure of the supercell past 1731 UTC.
Figure 13. The effect of the curvature of the Earth on radar beam altitude. Graphic courtesy of the Australian Bureau of Meteorology.
Although the KGSP radar indicated a persistent and strong mesocyclone at mid-levels of the storm (rotational velocity of 35 kts at 60 nm from the radar), the lack of strong rotation at low levels on the TCLT radar allowed the forecaster to conclude correctly that a tornado warning was not needed at that time. Hail the size of tangerines (2 inches diameter) was reported east of Maiden in Catawba County at 1745 UTC, but there was no report of wind damage along the storm's path. As the supercell approached the southern end of Lake Norman, a Severe Thunderstorm Warning was issued for southern Iredell and northern Mecklenburg counties at 1751 UTC. A classic hook echo redeveloped as the storm crossed Lake Norman, prompting an upgrade to a tornado warning for southern Iredell and northern Mecklenburg counties at 1803 UTC. Although no tornadoes touched down in those counties, additional reports of large hail were received. A tornado warning was issued for southern Rowan and northern Cabarrus counties at 1811 UTC as the storm moved across the area between Mooresville and Davidson.
A comparison between the KGSP radar and the TCLT radar at the lowest elevation scans as the supercell moved along the Rowan - Cabarrus county line illustrates the value of the TDWR for investigating the low level structure of thunderstorms across the Piedmont of North Carolina. The 0.5 degree reflectivity scan from the KGSP radar at 1844 UTC (Fig. 14) showed an inflow notch of lower reflectivity over northern Cabarrus County north of Mount Pleasant. However, the center point of the radar beam cut through the storm at approximately 11,000 feet AGL, below which the KGSP radar was not able to observe. At 1842 UTC, the 1.0 deg scan from the TCLT radar (20 nm distant) showed a well-defined, classic hook echo to the east of Kannapolis and northwest of Mount Pleasant (Fig. 15). The center point of the radar beam from TCLT cut through the storm at only 2500 feet AGL. The hook echo also has more detail on the image from the TCLT radar because of the smaller sampling volume, due in part to the configuration of the TDWR and in part to its closer proximity to the storm.
Figure 14. Radar reflectivity at 0.5 degree scan from the KGSP WSR-88D at 1844 UTC. Click on image to enlarge.
Figure 15. Radar reflectivity at 1.0 degree scan from the TCLT TDWR at 1842 UTC. Click on image to enlarge.
A comparison of velocity images has even greater ramifications in this case. The corresponding storm relative velocity image from KGSP at 1844 UTC (Fig. 16) shows the strongest gate-to-gate shear on the Rowan side of the county line. However, the corresponding velocity image from TCLT (Fig. 17, storm relative velocity calculations were unavailable) shows the best velocity couplet on the Cabarrus side of the county line. The information from the TCLT radar allowed the warning forecaster to issue a more precise pathcast for the potentially tornadic part of the storm. In fact, a tornado touched down briefly near the Rimer community in northern Cabarrus County north of Mount Pleasant at approximately 1848 UTC (Fig. 18).
Figure 16. Storm relative velocity at 0.5 degree scan from the KGSP WSR-88D at 1844 UTC. Targets moving toward the radar are indicated by green colors while targets moving away from the radar are indicated by red shades. Note the contrast between green and red along the border between Rowan and Cabarrus counties. Click on image to enlarge.
Figure 17. Velocity at 1.0 degree scan from TCLT radar at 1842 UTC. Click on image to enlarge.
Figure 18. Approximate location of tornado damage along Cline School Road near the community of Rimer, in northern Cabarrus County, North Carolina. The map on the right shows the location of the small scale map on the left. Graphics made with Delorme Street Atlas USA. Click on images to enlarge.
The supercell continued to move east along the county line through 1857 UTC, exhibiting a low level pendant at 0.2 degrees, a hook echo at 2.4 degrees, and even an echo-free bounded weak echo region on the 5.0 degree scan. The supercell eventually moved along the Rowan - Stanly county line and out of the WFO GSP county warning area after 1910 UTC.
A tornado of F1 intensity briefly touched down in northern Cabarrus County, North Carolina, near the Rimer community at 1848 UTC (248 pm EDT). A Tornado Warning was issued by the National Weather Service at 1811 UTC (211 pm EDT) fornorthern Cabarrus County, providing a lead time of 37 minutes. The availability of the TDWR data from the TCLT radar had an important and positive impact on warning operations during the event. The location of the TCLT radar in close proximity to the supercell as it tracked across the western Piedmont allowed for better interrogation of low level storm features. A tornado warning false alarm was avoided for Lincoln and Catawba counties. Once low level rotation increased as the storm moved north of Charlotte, the TCLT radar allowed the warning forecaster to issue a more precise projection for the path of the tornado across northern Cabarrus County.
Hail stones at least one inch in diameter fell near Pumpkin Center, North Carolina, northeast of Lincolnton in Lincoln County, at 1742 UTC. Photo courtesy of Justin Reid.
The storm report, 500 mb analysis, upper air sounding, and mesoscale analysis graphics (Figs. 1, 2, 4, 6-8) were obtained from the Storm Prediction Center. Jonathan Blaes (NWS Raleigh) assisted with obtaining the mesoscale graphics. The surface analysis (Fig. 3) was obtained from the Hydrometeorological Prediction Center. The satellite imagery (Fig. 5) was obtained from the University Corporation for Atmospheric Research. The author wishes to thank the Australian Bureau of Meteorology for the use of the radar beam propagation graphic (Fig. 12). As usual, Neil Dixon debugged the html code.