A tornado moves along Interstate 85 at exit 85 near Linwood, North Carolina, on Palm Sunday, 28 March 2010. The tornado initially touched down in Spencer, North Carolina. Image taken by Jonna Bingham, courtesy of WCNC-TV Charlotte, and used by permission.

Author's Note: The following report has not been subjected to the scientific peer review process.

1.  Introduction

In the late afternoon and evening of Palm Sunday, 28 March 2010, numerous
severe thunderstorms developed over the Foothills and Piedmont of the
Carolinas.  The most significant of the storms strengthened over Gaston 
County, North Carolina, after 5 pm Eastern Daylight Time (EDT), and 
produced a tornado near Belmont at 537 pm EDT (2137 UTC).  The tornado 
struck the Park Dale Mill, damaging the metal roof of the mill building 
and overturning a tractor-trailer, and then moved over an adjacent mobile
home park before lifting seconds later.  The storm went on to produce
straight-line wind damage over western Mecklenburg County, North Carolina, 
while it quickly acquired supercell thunderstorm characteristics.  The
supercell continued to move off to the northeast and spawned a second
tornado which began in Spencer, North Carolina, at 645 pm EDT (2245 UTC).  
The tornado damaged a shopping center on North Salisbury Avenue and moved 
over the east side of town, producing tree and roof damage to several 
homes.  The tornado remained on the ground and crossed Interstate 85 
before crossing the Yadkin River into Davidson County at 649 pm EDT.  The
damage from both tornadoes was rated at EF1 intensity on the Enhanced
Fujita Scale.  No other tornadoes were reported from subsequent storms 
across the county warning area (CWA) of the Greenville-Spartanburg (GSP) 
National Weather Service (NWS) office, although several produced hail 
larger than quarters.

[Note:  All times in this report from this point forward are expressed 
in Universal Time Coordinated (UTC), which is EDT plus four hours.  To
convert from UTC to EDT, subtract four hours from the UTC time.]

Click here to view a list of wind damage and hail reports on 28 March 2010

Figure 1.  Preliminary reports of tornadoes, wind damage, and hail 
received during the 24 hours ending at 1200 UTC on 29 March 2010.  Click 
on image to enlarge.

The events of 28 March were not unprecedented as several notable severe 
thunderstorm and tornado outbreaks have occurred in late March.  Weather
forecasters have come to expect a big severe weather event across the
Carolinas in early Spring.  Nor were the events of 28 March unanticipated.  
As early as the early morning hours of 27 March, the NWS expected strong 
deep layer wind shear to be present across the Piedmont of the Carolinas 
on 28 March, as noted by the Storm Prediction Center (SPC) in the Day 2 
Severe Weather Outlook.  Subsequent outlooks issued in the afternoon of 
27 March and in the early morning hours of 28 March continued to highlight 
the threat for supercell thunderstorms and tornadoes in the afternoon and 
evening of 28 March, if enough instability could be realized.

2.  Synoptic Overview

On the morning of 28 March, the upper air analysis at 1200 UTC showed a 
closed low at 500 mb centered near St. Louis, Missouri, and a trough axis
extending back across the southern Plains (Fig. 2).  Of note was a mid-level
jet streak of 90 to 95 kt winds observed across central Alabama and central 
Mississippi.  A belt of strong winds, on the order of 50 to 55 kt, also 
appeared on the 700 mb analysis (Fig. 3) from central Alabama west across 
southern Mississippi and Louisiana.  The 850 mb analysis (Fig. 4) showed a 
southwesterly low-level jet of 40 to 50 kt extending from Alabama and 
Georgia to the Ohio River Valley.  The mid-level jet streaks at 500 mb and
700 mb were expected to move over the top of the low-level jet at 850 mb
as it translated east, contributing to an increase in deep-layer wind shear 
across the western Carolinas later in the day.

Figure 2.  Objective analysis of 500 mb geopotential height (black 
contours), temperature (red dashed contours), and wind barbs at 1200 UTC 
on 28 March 2010.  Click on image to enlarge.

Figure 3.  Objective analysis of 700 mb geopotential height (black 
contours), temperature (red and blue dashed contours), dewpoint (green 
contours), and wind barbs at 1200 UTC on 28 March 2010.  Click on image 
to enlarge.

Figure 4.  Objective analysis of 850 mb geopotential height (black 
contours), temperature (red and blue dashed contours), dewpoint (green 
contours), and wind barbs at 1200 UTC on 28 March 2010.  Click on image 
to enlarge.

The surface analysis from the Hydrometeorological Prediction Center (HPC)
showed an occluded low pressure center near the confluence of the Ohio
and Mississippi Rivers (Fig. 5).  The occluded front extended southward 
across western Kentucky and western Tennessee, with a cold front from 
northeast Mississippi to southeast Louisiana.  High pressure was 
centered over Newfoundland.  Although it was not analyzed, there was
evidence of a weak surface boundary stretching across the northern 
part of Upstate South Carolina and running southwest to northeast 
across the Foothills of North Carolina.  Surface observations taken
around 1200 UTC (Fig. 6) showed a wind shift between a northerly 
component at Rutherfordton (FQD) and North Wilkesboro (UKF), and an 
easterly to southeasterly wind across most of South Carolina and the 
eastern half of North Carolina.  The air mass on the north side of the
boundary resembled a cold air damming wedge, albeit smaller in coverage
than the usual wedge.  This boundary would play an important role in the 
development of severe weather later in the day.

Figure 5.  HPC Surface fronts and pressure analysis at 1200 UTC on 
28 March.  Click on image to enlarge.

Figure 6.  Surface observations plot for 1200 UTC 28 March, using the 
traditional station model.  A quasi-stationary boundary has been analyzed 
and displayed using traditional front symbols.  Click on image to enlarge.

The upper air sounding taken at Birmingham, Alabama (BMX, Fig. 7), was 
perhaps the most indicative of the conditions expected to occur over the 
western Carolinas and northeast Georgia late in the day.  The sounding
showed the strong shear (65 kt) and storm relative helicity (SRH, 
468 m2s-2) in the layer from the surface to 3 km that was supportive
of rotating updrafts.  However, the sounding only yielded a convective
available potential energy (CAPE) of around 600 J kg-1 when modified for 
forecast surface temperature.  The combination of strong shear and modest 
buoyancy warranted a slight risk of severe thunderstorms in the Day 1 
Convective Outlook update issued by the SPC at 1300 UTC.

Figure 7.  Skew T, log P diagram for upper air sounding taken at BMX 
at 1200 UTC on 28 March.  A hodograph is shown in the upper right corner 
and a table of convective parameters is provided at the bottom.  Click 
on image to enlarge.

An initial band of showers across northeast Georgia, the western tip of 
South Carolina, and western North Carolina at 1200 UTC (Fig. 8) moved 
slowly northeast through 1800 UTC, as seen on a loop of a regional radar 
mosaic.  The light rain showers may have strengthened the pre-existing 
weak cold air damming wedge across western North Carolina and northern 
South Carolina.  Surface observations at 1800 UTC showed light rain, a 
north wind, and temperatures in the mid-40s to low-50s at Hickory (HKY) 
and GSP, while temperatures were in the low-60s with a south wind at 
Anderson (AND), Charlotte (CLT), and Salisbury (RUQ).  The observations 
supported the analysis of a weak surface boundary running across the 
northern part of Upstate South Carolina and the western Piedmont of 
North Carolina (Fig. 9).

Figure 8.  Composite reflectivity mosaic centered on GSP at 1200 UTC on  
28 March.  Click on image to enlarge.

Figure 9.  HPC regional surface analysis at 1800 UTC on 28 March.  
Click on image to enlarge.

Destabilization was slow to occur due to extensive cloud cover during 
the morning hours, seen on a loop of visible satellite imagery, but 
enough breaks occurred for some surface-based CAPE to develop by 
midday.  The Day 1 Convective Outlook update, issued by the SPC at 
1626 UTC, highlighted the developing buoyancy and wind shear.  Water 
vapor imagery showed evidence of the mid-level jet streak moving to 
the south of the upper low seen over western Kentucky, as indicated 
by the stripe of warmer brightness temperatures as the result of 
subsidence across the middle of Mississippi and Alabama.  As expected, 
the jet streak dramatically increased the deep layer shear across the 
region, thus by the middle part of the afternoon it appeared likely 
that severe thunderstorms would occur as discussed by the SPC at 
1921 UTC.  An increase in convective development to the lee of the 
Appalachians prompted the issuance of a Tornado Watch (#0037) at 
1951 UTC.

3.  Pre-Storm Environment

In the 2000 UTC to 2100 UTC time frame, convection loosely organized 
into bands across Georgia and the Carolinas (Fig. 10).  One band 
stretched from the west side of the Charlotte metro area to the west 
side of Augusta, Georgia, at 2100 UTC.  The environment was highly 
supportive of rotating updrafts and supercell thunderstorms.  At 
2100 UTC, by most measures, wind shear and SRH were large.  Effective 
bulk shear (Fig. 11a) of greater than 50 kt stretched generally along 
the surface boundary at 2100 UTC, with a maximum of 65 kt to the 
southwest of Charlotte.  The SRH in the surface to 3 km layer (Fig. 11b) 
was greater than 450 m2s-2 across all of Upstate South Carolina and the 
Foothills and Piedmont of North Carolina, with a contour of 550 m2s-2
across the Charlotte area.  Shear in the 0-1 km layer (Fig. 11c) was on 
the order of 40 kt along the boundary.  The SRH in the layer from the 
surface to 1 km (Fig. 11d) was in excess of 400 m2s-2.  This suggested
a relatively high probability for tornado-producing supercells.  The
mean height of the lifting condensation level (LCL) in the lowest 100 mb 
(Fig. 11d) was between 500 m and 750 m above ground level.  The surface
boundary was particularly apparent in the analysis of effective inflow
base (Fig. 12), which suggested the elevated nature of convective cells
to the north of the boundary and how cells to the south of the boundary
were rooted in the boundary layer.  A sharp gradient of effective SRH 
was located on the warm side of the boundary with a maximum of 500 m2s-2
on the east side of Charlotte.  Surface-based CAPE reached approximately 
500 J kg-1 on the west side of Charlotte by 2100 UTC, with no convective 
inhibition (CIN; Fig. 13).  Although modest, the amount of buoyancy was 
enough to maintain deep convection as it moved over the Charlotte metro 
area after 2100 UTC.

Figure 10.  As in Fig. 8, except for 2059 UTC on 28 March.  Click on 
image to enlarge.

Figure 11.  SPC objective mesoscale analysis at 2100 UTC on 28 March 
showing (A) effective bulk shear (kt; blue contours) and storm motion
(barbs), (B) SRH in the 0-3 km layer (m2s-2; blue contours) and storm
motion (barbs), (C) shear in the 0-1 km layer (kt; blue contours), and
(D) SRH in the 0-1 km layer (kt; blue contours), 100 mb mean lifting 
condensation level (m; dashed green contours) and storm motion (barbs).
Mecklenburg County (Charlotte), North Carolina, is shaded yellow for
reference.  Click on images to enlarge.

Figure 12.  SPC objective mesoscale analysis at 2100 UTC on 28 March 
showing effective inflow base (m; color fill) and effective SRH (m2s-2; 
blue contours).  Mecklenburg County (Charlotte), North Carolina, is 
shaded yellow for reference.  Click on image to enlarge.

Figure 13.  SPC objective mesoscale analysis at 2100 UTC on 28 March 
showing surface based CAPE (J kg-1; brown contours) and surface based 
CIN (J kg-1; color fill).  Mecklenburg County (Charlotte), North Carolina,
is shaded yellow for reference.  Click on image to enlarge.

After 2100 UTC, the convection was poised to move northeast in close 
proximity to the wedge boundary, through an environment with a low LCL 
and rich with low level shear and helicity.  Most objective criteria
supported a threat for supercell thunderstorms with the possibility
of tornadoes (Fig. 14).

Figure 14.  SPC objective mesoscale analysis at 2100 UTC on 28 March 
showing (a) Craven Significant Severe Parameter (index; blue and magenta 
contours), (b) 3 km Energy-Helicity Index (index; green and brown contours), 
(c) Supercell Composite Parameter (index; light blue contours) and Bunkers 
storm motion (kt; barbs), and (d) Significant Tornado Parameter (index; 
orange and red contours) and mixed layer CIN (J kg-1; color fill).
Mecklenburg County (Charlotte), North Carolina, is shaded yellow for
reference.  Click on images to enlarge.

4.  Convective Evolution

An isolated severe thunderstorm (the first of the day) occurred between 
2000 UTC and 2100 UTC, producing hail up to the size of golf balls in the 
Mooresville, North Carolina area.  Between 2100 and 2300 UTC, convection 
gradually increased in coverage and intensity over the GSP CWA, with 
severe thunderstorms becoming more numerous.  Although severe weather 
during this time was mainly limited to large hail ranging from quarter 
to golf ball size, two tornadoes occurred across the North Carolina 
Piedmont: one in the Belmont area of Gaston County, and another that began 
in Spencer in Rowan County.  We will focus closely on these two tornadoes 
through the remainder of this section.

Click here to view a 26 frame Java loop of base reflectivity on the 
0.5 degree elevation scan from the KGSP radar from 2103 UTC to 2259 UTC.

Click here to view a 26 frame Java loop of storm relative motion on the 
0.5 degree elevation scan from the KGSP radar from 2103 UTC to 2259 UTC.

a. The Belmont Tornado

The convective cell that spawned the Belmont tornado moved into southwest 
Gaston County from South Carolina around 2108 UTC (Fig. 15).  At this 
point, the Belmont storm had a slight cyclonic curvature at the 0.5 degree 
elevation slice from the NWS Doppler radar located at the GSP airport (the
KGSP radar).  There was also echo overhang on the storm’s right flank, 
indicative of a strong updraft.  However, the high reflectivity core 
(greater than 50 dBz) extended only to about 16 kft.  Higher elevation 
slices from KGSP revealed an echo top of around 30 kft.  In addition, the 
width of the storm core in the mid-levels was only two to three miles.  The 
cyclonic curvature and echo overhang, along with the shallow and small 
nature of this cell, was suggestive of a mini-supercell.  However, 
examination of the storm relative velocity (SRV) at this time (Fig. 16) 
revealed only a broad, divergent area of cyclonic shear on the 1.3 degree 
elevation slice, which was sampling the storm at about 10 kft MSL.  
Rotational shear associated with this feature was only 3.2 x 10-3 s-1, which 
does not qualify it as a "minimal mesocyclone" according to the Rotational 
Shear Nomogram.  Therefore, while the storm exhibited some supercell-like 
characteristics, the absence of a mesocyclone eliminated it as a mini-
supercell candidate at this point in time.  It was also important to note 
that based on the thermodynamic character of the environment, the storm did 
not meet objective Severe Thunderstorm Warning criteria at this time due to 
the fact that it was unlikely to produce large hail given the limited 
vertical extent of the high reflectivity core.  Indeed, no severe weather 
was reported with this storm as it moved over southwest and central Gaston 
County. 

Figure 15.  Four-panel display of base reflectivity from the KGSP radar at 
(a) 0.5 degrees, (b) 1.3 degrees, (c) 2.4 degrees, and (d) 3.4 degrees 
elevation at 2108 UTC on 28 March.  Rainfall intensity is given by the 
color table at the upper left of each figure.  Click on image to enlarge.

Figure 16.  Four-panel display of storm relative motion from the KGSP radar 
at (a) 0.5 degrees, (b) 1.3 degrees, (c) 2.4 degrees, and (d) 3.4 degrees 
elevation at 2108 UTC on 28 March.  Motion toward or away from the radar 
is given by the color table at the upper left of each figure, where warm
colors are motion away from the radar and cool colors show motion toward 
the radar.  Click on image to enlarge.

The lowest four elevation slices of base reflectivity from KGSP at 2126 UTC 
showed persistent echo overhang as well as cyclonic curvature (Fig. 17). 
The right-flank appendage became elongated in a north/south orientation. 
The images also revealed a line of scattered convective cells intersecting 
the appendage and extending to the southwest. The surface observations 
overlaying the 0.5 degree elevation scan suggested that these cells were 
oriented roughly parallel and just behind (i.e., in the cool sector of) 
the synoptic surface boundary. 

Figure 17.  As in Fig. 15, except at 2126 UTC.  Click on image to enlarge.

The narrower beam width of the Terminal Doppler Weather Radar (TDWR) near 
Charlotte, North Carolina (TCLT), and its closer proximity to the cell 
offered an opportunity for a more detailed analysis of the storm (Fig. 18).   
The reflectivity image showed the elongated appendage extending from the 
"main" cell over east-central Gaston County.  From the TDWR’s perspective, 
the storm system appeared to have a quasi-linear, as opposed to single-
cellular, appearance, especially at the 1.0 degree and 2.4 degree elevation 
angles.

Figure 18.  Four-panel display of base reflectivity from the TCLT radar at 
(a) 0.2 degrees, (b) 1.0 degrees, (c) 2.4 degrees, and (d) 5.0 degrees 
elevation at 2126 UTC on 28 March.  Rainfall intensity is given by the 
color table at the upper left of each figure.  Click on image to enlarge.

The SRV data from TCLT at 2126 UTC (Fig. 19) did not reveal anything too 
enlightening.  However, there was a broad, weak zone of cyclonic shear 
associated with the elongated appendage.  This was especially evident at 
2.4 degrees.  Meanwhile, surface observations indicated the storm system 
was located just on the cool side of the frontal boundary (denoted by the 
red line) at this time.  Since analyses indicated that no CAPE existed on 
the cool side of the boundary (Fig. 9), it was unlikely that this cell 
was surface-based. 

Figure 19.  Same as in Fig. 18, except for Storm Relative Velocity.  Click 
on image to enlarge.

By 2132 UTC, the storm system was exhibiting interesting structure (Fig. 20). 
A slight bulge and rear-inflow notch was evident in the middle of the 
appendage at 0.2 degrees.  There was also a subtle cyclonic curvature of 
the 40-45 dBz echo at the southern end of the appendage at the 5.0 degree 
elevation slice.  The storm relative velocity image at this time (Fig. 21) 
indicated this curvature was associated with an area of cyclonic rotation. 
Rotational shear was calculated at 1.33 x 10-2 s-1, placing this couplet 
in the "tornado possible" category of the rotational shear nomogram. 
However, the velocity couplet was not evident below 5.0 degrees.  Meanwhile, 
there was a weak area of inbound velocity along the back edge of the high 
reflectivity at 5.0 degrees.  This was above the bulging segment seen in 
the reflectivity field in lower levels, and may be indicative of a 
descending air current (i.e., a rear-flank downdraft [RFD]).  Velocities 
were weak due to the fact that the RFD would have a strong cross-radial 
component.

Figure 20.  Same as in Fig. 18, except at 2132 UTC.  Click on image to 
enlarge.

Figure 21.  Same as in Fig. 19, except at 2132 UTC.  Click on image to 
enlarge.

This circulation intensified and descended after 2132 UTC, as revealed in 
the time-series of rotational shear shown in Fig. 22.  By 2134, the 
circulation became evident at the 0.2 degree elevation scan.  At 2135 UTC, 
values of rotational shear were marginally within the "Tornado Probable" 
category of the nomogram.  This was only a minute or two before a tornado 
touched down on the southeast side of Belmont.  After 2135 UTC, the 
circulation intensified rapidly.  Unfortunately, due to the 3 to 6 minute 
resolution of the higher elevation scans of the TDWR, data were only 
available at 0.2 degrees during the brief time the tornado was on the 
ground.  The reflectivity and SRV at the time of the tornado (2137 UTC) 
showed a hook echo and associated small (~0.5 km diameter), but intense 
circulation in the vicinity of the tornado (Fig. 23). Although the actual 
velocities were not that high (39 kt outbound and 14 kt inbound), the very 
small diameter of the couplet is resulting in rotational shear of 4.8 x 10-2 s-1. 
This value was actually "off the chart" with respect to the rotational shear 
nomogram, suggesting a tornado was likely imminent or already occurring.

Figure 22.  Time series of rotaional shear of the Belmont tornadic storm 
as calculated from TCLT between 2129 UTC and 2141 UTC.  The time of the 
tornado (red line) and downburst occurrence (light blue line) are marked 
along the X-axis.  Click on image to enlarge.

Figure 23.  TCLT base reflectivity (a) and storm relative velocity (b) at 
2137 UTC on 28 March.  The location of the damage produced by the Belmont 
tornado is indicated by the white triangle.  Click on image to enlarge.

The 2138 UTC volume scan (Fig. 24) revealed the presence of the hook echo 
through the lowest four elevation slices.  In fact, high reflectivity had 
almost wrapped completely around the circulation at the 2.4 degree and 
5.0 degree slices.  Although the circulation weakened in the lower levels 
after the tornado dissipated (Fig. 22), it and the hook echo persisted for 
approximately 30 minutes after the tornado dissipated (not shown).  
Nevertheless, an NWS storm survey indicated that the tornado dissipated 
before moving into Mecklenburg County, with the parent storm evolving into 
a downburst-producer.

Figure 24.  Same as in Fig. 18, except at 2138 UTC.  Click on image to 
enlarge.

The circulation developed rapidly as the shear zone moved across the 
surface boundary.  As the storm crossed the boundary, it would have 
ingested and tilted air that was rich in horizontal vorticity.  In 
addition, it would have experienced more unstable air and enhanced 
low-level convergence.  Indeed, there was a notable increase in updraft 
strength over eastern Gaston County at the time of the tornado, suggested
by an increase in Vertically Integrated Liquid on the 2140 UTC scan. 
This suggested that in addition to tilting of the horizontal vorticity 
that existed along the boundary, vortex tube stretching within the updraft 
likely played a key role in tornadogenesis as well.  Despite the elongated 
appearance of this storm system in TDWR reflectivity data, the presence 
of a cyclonic circulation in the mid-levels of the storm (i.e., around 
10 kft) prior to tornadogenesis suggested supercellular processes were 
at play.

b. The Spencer Tornado

As the storm that produced the Belmont tornado moved across Mecklenburg 
County, it generally maintained its intensity in terms of the depth of the 
storm.  However, other than the downburst that occurred across western 
Mecklenburg County subsequent to the dissipation of the Belmont tornado, 
no additional severe weather was reported with the storm as it moved across 
northern Mecklenburg County and northwest Cabarrus County into southwest 
Rowan County.  However, the storm underwent a significant intensification 
as it moved into central Rowan County.  During this time, the high 
reflectivity core developed to over 20 kft (Fig. 25), while the storm top 
extended to over 40 kft.  This was accompanied by a substantial increase 
in mid-level rotation (Fig. 26).  Images from TCLT showed the supercellular 
character of the storm during this time.  The reflectivity images revealed 
an appendage extending southeast from the main core of the cell (Fig. 25).  
A large weak echo region (WER) was observed at the lower elevation slices, 
topped by a high reflectivity core in the mid-levels (around 14 kft).

Figure 25.  Four-panel display of base reflectivity from the TCLT radar at 
(a) 1.0 degrees, (b) 2.4 degrees, (c) 5.0 degrees, and (d) 7.7 degrees 
elevation at 2232 UTC on 28 March.  Rainfall intensity is given by the 
color table at the upper left of each figure.  Click on image to enlarge.

Figure 26.  Same as in Fig. 25, except for Storm Relative Motion at 
2232 UTC.  Click on image to enlarge.

TCLT images from 2244 UTC (Fig. 27 and Fig. 28) revealed the impressive 
structure of the mini-supercell minutes prior to tornadogenesis.  The 
appendage observed at 2232 UTC evolved into a pronounced hook echo.  The 
"hole" in reflectivity at the 2.4 degree elevation scan (above the weak 
reflectivity region on the 1.0 degree scan) was topped by an area of high 
reflectivity (greater than 50 dBz) at 5.0 degrees, indicating the presence 
of a bounded weak echo region (BWER).  The SRV data at this time (Fig. 28) 
did not reveal a rotational couplet, owing to range folding in the region 
of the BWER.  However, there was an area of intense (60 kt or more) 
outbound velocities associated with the hook echo.

Figure 27.  Same as in Fig. 25, except at 2244 UTC.  Click on image to enlarge.

Figure 28.  Same as in Fig. 26, except at 2244 UTC.  Click on image to enlarge.

At 2250 UTC, the tornado was in progress.  By this time, it had likely 
exited the GSP CWA and was just inside the Davidson County border.  TCLT 
reflectivity imagery revealed a persistent hook echo and BWER (Fig. 29). 
However, the SRV data was most impressive at this time, especially at the 
2.4 degree slice, where a gate-to-gate circulation of around 90 kt was 
evident (Fig. 30).

Figure 29.  Same as in Fig. 25, except at 2250 UTC.  Click on image to enlarge.

Figure 30.  Same as in Fig. 26, except at 2250 UTC.  Click on image to enlarge.

One question that emerged from this case study was why there was such a 
long duration (i.e., more than an hour) between the two tornado events 
produced by this single mini-supercell.  The surface analysis at 2200 UTC 
offers a potential explanation.  We hypothesize that the Belmont tornado 
occurred as the mini-supercell crossed from the cool side of the boundary 
into the warm sector.  This may have been the result of a combination of 
vertical tilting of horizontal vorticity along the boundary, and vortex 
stretching resulting from updraft intensification as it ingested more 
unstable surface-based air.  Between 2145 UTC and 2230 UTC, the supercell 
was generally within the warm sector air mass and moving away from the 
boundary.  However, the analysis at 2200 UTC (Fig. 31) indicated that the 
surface boundary took on more of a zonal orientation across the northwest 
Piedmont of North Carolina. Therefore, it appeared that the supercell may 
have begun to interact with the surface boundary again as it moved northeast 
across the Piedmont after 2230 UTC.  This, along with the significant 
intensification of the updraft (vortex tube stretching) that occurred 
after 2230 UTC may be responsible for tornadogenesis near Spencer.

Figure 31.  SPC surface mesoanalysis at 2200 UTC on 28 March.  Red lines 
are surface temperature of 60 oF or greater.  Dashed blue lines are surface 
dewpoint temperature in oF, with color fill denoting values greater than
56 oF.  Black lines are sea level pressure in hPa.  Heavy blue line is 
the hand-analyzed position of the surface front.  Click on image to enlarge.

5.  Summary 

The Belmont and Spencer tornadoes appeared to be a classic case of a 
supercell thunderstorm producing multiple tornadoes as it interacted with 
a surface boundary.  The combination of high shear in the 0-1 km layer 
(around 40 kt) and low LCL (mean 100 mb value around 625 m) suggested a 
high probability of tornado occurrence.  It appeared that the Belmont
tornado occurred as the supercell moved from the cool side of the boundary
to the warm sector.  A bit more than one hour later, the Spencer tornado
appeared to touch down as the supercell began to interact with the boundary
once more over Rowan County.  Thus, the events of 28 March reinforced the
idea that boundaries should be identified and monitored closely when
the potential for tornadic supercells exists.

Selected Damage Survey Images

Most of the damage from the Belmont tornado occurred at the Parkdale Mill
plant 61 in Belmont, North Carolina, on 28 March 2010.  Part of the roof 
was torn off by the tornado.  Note the shreds of roof insulation strewn 
across the ground and in the trees downwind of Parkdale Mills.  Images
taken by Blair Holloway, NWS.  Click on images to enlarge.

The Spencer tornado touched down at the North Carolina Transportation
Museum and moved east paralleling N Salisbury Avenue through East Spencer, 
North Carolina, on 28 March 2010.  Most of the damage was concentrated 
near the intersection of N. Salisbury Avenue and E. Jefferson Street.  
Images taken by Rodney Hinson, NWS.  Click on images to enlarge.

References

Craven, J. P, H. E. Brooks, and J. A. Hart, 2001:  Baseline climatology
     of sounding derived parameters associated with deep, moist 
     convection.  Preprints, 21st Conf. Severe Local Storms, San Antonio,
     Amer. Meteor. Soc., 643-646.

Markowski, P. M., and Y. P. Richardson, 2009:  Tornadogenesis:  Our
     current understanding, forecasting considerations, and questions
     to guide future research.  Atmospheric Research, 93, 3-10.
	 

Acknowledgements

The author wishes to express his gratitude to Ms. Jonna Bingham for
graciously allowing the NWS to use the image of the tornado as it
moved east of Spencer.  Thanks are also given to Brad Panovich and 
Amy Lehtonen of WCNC-TV 36 Charlotte for providing the image.  Upper
air analyses and soundings were obtained from the Storm Prediction
Center.  The objective mesoscale analysis was created by the Storm
Prediction Center and obtained from the archive at the NWS Omaha
office.  Satellite imagery, surface observation plots, and radar
mosaics were obtained from the University Corporation for Atmospheric
Research.  The surface analyses were obtained from the Hydrometeorological
Prediction Center.  The survey of the Belmont tornado was conducted
by Tony Sturey and Blair Holloway, NWS.  The survey of the Spencer
tornado was conducted by Rodney Hinson and Jeffrey Taylor, NWS.