This tornado moved through the south side of the Clemson University campus on 26 August 2008. It was one of three confirmed tornadoes that affected the Upstate of South Carolina. The photo was taken from the south side of Lake Hartwell by Rob Harrison, South Carolina Department of Natural Resources.

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

1.  Introduction

Tropical Storm Fay and its remnants affected the southeastern United
States for over one week in late August of 2008.  After making landfall 
over southwest Florida early in the morning of 19 August (Fig. 1), 
the storm dumped significant rainfall as it moved across the Florida 
Peninsula.  The storm moved off the east coast of Florida and then 
slowly northward for about 36 hours beginning on 20 August.  A surface 
high pressure ridge strengthened to the north of the storm on 22 August 
which steered Fay westward for about three days, producing extreme 
rainfall amounts over parts of north Florida and south Georgia.  Fay 
weakened to a Tropical Depression on 24 August over the western Florida 
Panhandle and south Alabama and then stalled over southern Mississippi 
on 25 August.  Eventually, the remnants of Fay moved northeast and 
merged with a frontal boundary over eastern Tennessee, but not before 
it spawned tornadoes across parts of Alabama and parts of Georgia.  

Click here to view an 81 frame Java loop of HPC surface fronts and 
pressure analysis from 0000 UTC on 18 August to 0000 UTC on 28 August.

Figure 1.  Best track positions for Fay, 15-28 August 2008, as determined 
by the National Hurricane Center.  Click on image to enlarge.

The passage of the remnants of Fay brought much-needed rain to an area 
suffering from the effects of a significant drought.  Unfortunately, in 
spite of the drought, the heavy rain caused significant flooding across
parts of the western Carolinas and northeast Georgia, especially the
southern Piedmont of North Carolina and the Charlotte metropolitan area.
The remnants of Fay also provided a favorable environment for the 
development of numerous thunderstorms, including several supercells,   
on the afternoon of 26 August.  These supercells spawned a series of 
tornadoes across portions of Upstate South Carolina and northeast Georgia.  
Two tornadoes affected the Clemson, South Carolina area, with one tornado 
causing damage on the main campus of Clemson University.  A third tornado
touched down to the east of U.S. Highway 25 in the southwest part of
Greenville County.  Several reports of wind damage were received from
the western part of Upstate South Carolina as well (Fig. 2).  The 
remnants of Fay also produced significant rain across the Midlands
of South Carolina and tornadoes across central North Carolina.

(Click here to view a list of event total rainfall reports from across
the region)

Figure 2.  Large hail, damaging wind, and tornado reports compiled
by the Storm Prediction Center for the 24 hour period ending 
1200 UTC 27 August 2008.  Click on image to enlarge.

The events of 25-27 August 2008 were well anticipated by the 
National Weather Service (NWS) Weather Forecast Office (WFO) 
at the Greenville - Spartanburg Airport (GSP) in Greer, South
Carolina.  The forecast staff at WFO GSP are well versed in the 
effects of tropical rainfall across the southern Appalachians 
were able to recognize the threat when tropical cyclone remnants 
encroached upon the area.  As early as 17 August, as Tropical 
Storm Fay was still off the southern coast of Cuba, the pre-dawn 
issuance of the Hazardous Weather Outlook (HWO) highlighted the 
threat of "heavy rains and flooding" across GSP's county warning 
and forecast area (CWFA).  Although unanticipated slowing of the 
movement of Fay over Florida during the following days would 
eventually delay the onset of excessive rainfall across the CWFA 
from initial expectations,  confidence increased that Fay's 
remnants would affect the CWFA.  Again, the pre-dawn issuance of 
the HWO on 21 August stated "a chance for a significant rain 
event...next Tuesday and Wednesday."  This outlook information 
would turn out to be very perceptive.

A detailed HWO was issued during the evening of 24 August which 
stated "total rainfall amounts could reach 8 inches" over western
North and South Carolina and extreme northeast Georgia.  The most 
interesting part of the outlook information was a reference to the 
ongoing "drought and record low stream levels."  The long-standing 
practice of never issuing a Flash Flood Watch during a drought 
would be tested over the coming days, although it often has been 
said that droughts end with a flood.

The threat of severe thunderstorms and tornadoes loomed larger
through the day on 25 August as it appeared that some instability
might accompany the strong wind shear as the remnant circulation
of Fay passed by to the west on 26 August.  The Storm Prediction
Center placed most of the western Carolinas in a Slight Risk of
severe thunderstorms in the initial Day 1 Severe Weather Outlook
for 26 August.  The SPC expanded the Slight Risk farther east 
across the Charlotte metro area on the morning update issued at 
1254 UTC on 26 August.

Note:  All times in this report are referenced to Universal Time 
Coordinated (UTC), which is Eastern Daylight Time plus four hours.

2.  Synoptic Overview and Antecedent Conditions

Prior to the arrival of the remnants of Fay, most of the western 
Carolinas and northeast Georgia struggled with long term drought.  
The U.S. Drought Monitor issued on 19 August (Fig. 3) highlighted 
most of the region as "Exceptional," the highest of the five drought 
categories.  Many streams ran at record low levels.

Figure 3.  U.S. Drought Monitor for the Southeast Region, issued 19 August.  
Intensity of drought is given by the color scale on the left.  Click on 
image to enlarge.

Even with the center of then downgraded Tropical Depression (TD) Fay over 
southern Mississippi during the night of 24-25 August, the air mass over 
the southern Appalachians was assuming tropical characteristics.  As 
evidenced by the 25 August 1200 UTC upper air sounding at Peachtree City, 
Georgia (FFC), precipitable water (PWAT) had increased to 1.87 inches, 
which is 160% of normal (Fig. 4).  The mean flow from the surface to the 
mid-layers of the atmosphere had a deep south southeast orientation.  
Aided by terrain lift, widespread heavy showers developed around 0800 UTC 
on 25 August across the mountains of northeast Georgia, southwest North 
Carolina and western Upstate South Carolina (Fig. 5).  With the favorable 
flow orientation and increasingly tropical airmass, a swath of 3-5 inches 
of rain fell during the pre-dawn hours, prompting the issuance of the first 
Flash Flood Warnings for the event.  Due to underestimation, the radar 
estimated rainfall (Fig. 6) should be multiplied by 1.5 to achieve 
representative totals.

Figure 4.  Skew T - log P diagram (upper left) and hodograph (upper right) 
of upper air sounding taken at FFC at 1200 UTC on 25 August 2008.  A table 
of convective parameters and indices is shown at the bottom.  Click on 
image to enlarge.

Figure 5.  Regional radar reflectivity mosaic at 0758 UTC 25 August 2008.  
The intensity of precipitation is given by the color scale at the bottom.  
Click on image to enlarge.

Figure 6.  KGSP Storm Total Precipitation estimate ending at 1358 UTC on 
25 August.  Rainfall estimate is given by the color table at the lower right.  
Click on image to enlarge.

3.  Preliminary Flooding

Several creeks were reported out of their banks in Habersham County, Georgia,
on the morning of Monday, 25 August.  To the WFO forecasters on duty that 
morning, this was an ominous sign that even with record low streamflow 
conditions, the high rainfall rate from tropical precipitation would result
in flooding.  Given the expectation that the CWFA would be under the threat 
of additional rounds of tropical rainfall for the next 48 hours, a Flood 
Watch was issued later that morning for northeast Georgia, the northwest 
Upstate, and the central and southern North Carolina Mountains through 
Wednesday morning.

Numerous if not widespread showers continued to develop across the balance 
of the CWFA throughout the day and into the evening hours on 25 August.  A 
localized rainfall maximum (Fig. 7) along the urbanized Interstate 85 
corridor, from northeast Charlotte to Concord, prompted more Flash Flood 
Warnings and caused limited flooding.  More importantly, heavy rain moistened 
antecedent conditions and raised stream and river level flows, setting the 
stage for more serious flooding across this area over the next 36 hours.

Figure 7.  As in Fig. 6, except ending at 0400 UTC on 26 August.  Click 
on image to enlarge.

Meanwhile, the center of TD Fay assumed a northeastward motion, bringing 
her into west central Alabama during the night of 25-26 August.  As 
evidenced by the upper air sounding taken at FFC at 0000 UTC on 26 August
(Fig. 8), the airmass across the CWFA became increasingly favorable to 
support heavy tropical showers.  The sounding was saturated up to near 
400 mb, the warm air advection flow had strengthened, and PWATs had 
risen to 2.11 inches, or about 170 percent of normal.  Even though 
rainfall was relatively light north of Interstate 40 during 25 August, 
the perceptive midnight shift forecast team expanded the Flood Watch 
northward to encompass the North Carolina Foothills and northern North 
Carolina Mountains.

Figure 8.  As in Fig. 4, except at 0000 UTC on 25 August 2008.  Click on 
image to enlarge.

After a period of relatively limited, but isolated heavy shower activity 
during the pre-dawn hours of 26 August, widespread heavy showers increased 
rapidly across the Charlotte metro area shortly after daybreak (Fig. 9). 
This prompted another round of Flash Flood Warnings for Mecklenburg and
Cabarrus counties in North Carolina and York County in South Carolina, 
which were issued at 1318 UTC and then extended until 1730 UTC.  During 
this same timeframe, widespread heavy showers developed northward across 
the Mountains and Foothills, as shown by the increase in the coverage and
magnitude of total rainfall across the western CWFA and from York, South
Carolina, to Salisbury, North Carolina (Fig. 10).

Click here to view a 19 frame Java loop of regional radar mosaic
imagery from 2356 UTC on 25 August to 1756 UTC on 26 August.

Figure 9.  As in Fig. 5, except for 1159 UTC 26 August 2008.  Click on image to enlarge.

Figure 10.  As in Fig. 6, except ending at 1558 UTC on 26 August.  Click 
on image to enlarge.

4.  Synoptic Characteristics on 26 August

The surface analysis at 1200 UTC on 26 August (Fig. 11) indicated the
center of TD Fay located over north central Alabama.  A quasi-stationary 
boundary extended from the cyclone across middle Tennessee into extreme 
southern Kentucky and Virginia.  The upper air sounding from FFC at 
1200 UTC on 26 August (Fig. 12) revealed an atmospheric profile typical 
of the northeast quadrant of a remnant tropical cyclone (McCaul 1991).  
The sounding possessed high moisture content (precipitable water of 
5.76 cm), with lapse rates tending toward moist adiabatic through a deep 
layer.  However, due to the very warm and moist boundary layer, there was 
potential instability in the sounding, with Convective Available Potential 
Energy (CAPE) calculated at 675 J kg-1.  A strongly veering wind profile 
was noted, with a deep layer of 40 to 50 kt winds in the lowest 6 km.  
This resulted in strong shear parameters, with Storm Relative Helicity 
(SRH) of 411 m2 s-2 and 365 m2 s-2 in the 0-3 km and 0-1 km layers, 
respectively.

Figure 11.  National surface analysis of fronts and pressure at 1200 UTC 
on 26 August 2008.  Click on image to enlarge.

Figure 12.  Observed upper air sounding from FFC at 1200 UTC on 
26 August 2008.  Click on image to enlarge.

Although widespread cloud cover north of the frontal boundary 
inhibited surface heating for much of the day, the visible 
satellite image in Fig. 13 indicated some clearing had occurred 
south of the boundary over the South Carolina Midlands by early 
afternoon.  A regional surface analysis at 1800 UTC (Fig. 14) 
on 26 August revealed the evolution of the lower atmosphere 
during the late morning and early afternoon.  As the strong 
high pressure system over the northern part of the United 
States (Fig. 11) continued to move to the southeast, the 
frontal boundary pushed south across the western Carolinas.  
The location of the boundary can be inferred from Fig. 14 by 
the shift from southerly to east and southeast winds.  There 
was also a change in air mass across the front, with lower-to-
middle 80s air temperatures south of the boundary, and lower-
to-middle 70s temperatures to the north.  This temperature 
gradient was enhanced by the clearing skies in the warm sector.

Click here to view a 43 frame Java loop of GOES-12 water vapor 
satellite imagery from 1145 UTC to 2345 UTC on 26 August 2008.

Click here to view a 12 frame Java loop of GOES-12 visible satellite
imagery from 2345 UTC on 24 August to 2245 UTC on 26 August 2008.

Figure 13.  GOES-12 visible satellite image from 1745 UTC on 
26 August 2008.

Figure 14.  Regional surface observations plot at 1800 UTC on 
26 August 2008.  The plots follow the traditional station model.  
Click on image to enlarge.

It is well established that baroclinic boundaries can play a key 
role in enhancing the potential for tornadogenesis in severe 
weather environments, especially on the "cool" side of boundaries 
(Markowski et al. 1998).  This occurs due to baroclinic generation 
of horizontal vorticity that results from gradients in buoyancy.  
If there is positive surface-based CAPE on the "cool" side of the 
boundary, convective updrafts can reorient this vorticity vertically, 
resulting in cyclonically rotating updrafts and greater potential 
for tornadogenesis.  Moreover, it is thought that weak east-west 
oriented boundaries may have played an important role in previous 
tornado events associated with the remnants of tropical cyclones 
across the Carolinas and northeast Georgia (Schneider and Sharp 
2007; Lane 2005; McCaul et al. 2004).

5.  Convective evolution

A sequence of regional radar images (Fig. 15) showed the effect that 
destabilization had on the character of precipitation.  While the 
morning hours mainly saw widespread stratiform rain, convective 
activity, including discrete thunderstorms, became more common by 
mid-afternoon in the vicinity of the front.  Of particular interest 
was the band of showers and embedded thunderstorms moving across 
northeast Georgia at 1800 UTC on 26 August.

Figure 15.  Regional radar mosaic from 1459 UTC (left) and 1756 UTC 
(right) on 26 August 2008.  Click on images to enlarge.

a.  The Reed Creek - Clemson Tornado

Figure 16a is a reflectivity image at 0.5 degree elevation from the 
Greer, South Carolina (KGSP) Weather Service Radar (WSR-88D) at 
1803 UTC.  Although the cell appeared rather innocuous, there was 
a cyclonic curvature in the reflectivity pattern on the southern 
flank of the cell.  The storm relative velocity (SRV) images at 
1803 UTC in Fig. 16b-e were more revealing.  The images show a 
shallow, but significant circulation associated with the convective 
updraft near Hartwell, Georgia.  The rotational shear associated 
with this vortex was calculated at 17.1 x 10-3 s-1.  This value of 
shear was within the "Tornado Possible" range of the rotational 
shear nomogram (Falk and Parker 1998).  Contrast this with 
the rotational velocity (Vr) of the vortex, calculated at 25.3 knots 
at a distance of 47 km from KGSP.  According to the mesocyclone 
detection nomogram (Andra 1997) this value suggests only a 
minimal mesocyclone.  It is important for forecasters to remember 
that the mesocyclone detection nomogram is based upon studies of 
classic supercells.  Therefore, it has little, if any utility in 
assisting with warning decisions in environments that are 
unsupportive of classic supercells (i.e., weakly buoyant and highly 
sheared).  This is because the diameter of the circulation is critical 
in assessing the strength of the vortex.  Weakly unstable environments 
will result in small updrafts, and therefore compact vortices.  A 
rotational velocity over a distance of 1 km is more significant than 
the same value of rotational velocity over a distance of 5 km.

Figure 16.  Base reflectivity at the 0.5 degree scan (a) and storm 
relative velocity at the 0.5 degree (b), 1.5 degree (c), 2.4 degree 
(d), and 3.4 degree (e) elevation angles from KGSP at 1803 UTC on 
26 August 2008.  Click on images to enlarge.

According to a study by Schneider and Sharp (2007), most of the 
Carolinas tornadoes that occurred in association with tropical 
cyclones during the active 2004 season occurred in association with 
vortices possessing WSR-88D derived rotational velocities of 20 kt 
or more over an average distance of less than 3 km.  Based largely 
upon this study, a tornado warning was issued for portions of Hart, 
Anderson, and Oconee counties at 1801 UTC.

An EF1 tornado touched down near the community of Reed Creek, 
Georgia, at around 1818 UTC (Fig. 17).  Once again, the reflectivity 
from the KGSP radar at 1820 UTC was rather non-descript (Fig. 18). 
However, the series of SRV images from this time showed the 
circulation evident in Fig. 16 had intensified significantly and 
deepened slightly.  Rotational velocity at 0.5 degrees was 38.4 kt. 
Meanwhile, rotational shear was 58.0 x 10-3 s-1, an increase of 70% 
from 1803 UTC.  This is well within the "Tornado Likely" category 
of the rotational shear nomogram.

Figure 17.  Map indicating the track of the Reed Creek - Clemson 
tornado.  Movement of the tornado was from the lower part of the image 
to the upper part (southwest to northeast).  Click on image to enlarge.

Figure 18.  As in Fig. 16, except for 1820 UTC.  Click on images to enlarge.

The Reed Creek tornado proceeded to cross Lake Hartwell into South 
Carolina, skipping along an intermittent path across western Anderson 
County and extreme southeast Oconee County before affecting the 
Clemson area.  The 0.5 degree SRV image at 1850 UTC (Fig. 19b) 
indicated a gate-to-gate rotational shear couplet calculated at 
65.3 x 10-3 s-1. This value is "off the chart" according to the 
rotational shear nomogram.

Figure 19.  As in Fig. 16, except for 1850 UTC.  Click on images to enlarge.

A time series of the rotational velocity and rotational shear values 
associated with the convective cell that produced the Reed Creek/Clemson 
tornado showed that both parameters were highly variable during this 
two-hour window (Fig. 20).  However, rotational shear during the 
lifetime of the tornado was generally contained within the range of 
15 x 10-3 s-1 and 30 x 10-3 s-1.  The two "spikes" in shear values 
coincided with the most significant damage near Reed Creek and Clemson.

Figure 20.  Time series of rotational shear (red line) and rotational 
velocity (blue line) from the KGSP radar for the Reed Creek/Clemson storm.  
Times of significant damage are indicated as are times of Tornado Warning 
issuances (TOR 042 and TOR 043).  Click on images to enlarge.

b.  The Pendleton Tornado

Another tornadic cell developed across northwest Anderson County, 
South Carolina, shortly after 1930 UTC.  At 1920 UTC, the rotational 
shear observed from KGSP was quite weak (Fig. 21).  However, the 
reflectivity image revealed a rather pronounced cyclonic curvature 
on the southern flank of the storm.  A Tornado Warning was issued 
at 1927 UTC for northwest Anderson and southern Pickens counties, 
based mainly on the Vr of 25.5 kt and the cyclonic curvature in the 
reflectivity field.  By 1937 UTC (Fig. 22), shear increased to 
16.1 x 10-3 s-1.  Shear and Vr increased dramatically after 1937 UTC, 
peaking at 75.3 x 10-3 s-1 at 1941 UTC.  This rapid increase coincided 
with the development of a tornado between Pendleton and Townville, 
which tracked into extreme southwest Pickens County before dissipating 
(Fig. 23).

Figure 21.  As in Fig. 16, except for 1920 UTC.  Click on images to enlarge.

Figure 22.  As in Fig. 16, except for 1937 UTC.  Click on images to enlarge.

Figure 23.  Map indicating the track of the Pendleton tornado.  
Movement of the tornado was from the lower part of the image to the 
upper part (south-southeast to north-northwest).  Click on image 
to enlarge.

The time series of rotational velocity and rotational shear for the 
Pendleton tornadic storm (Fig. 24) illustrated the rapidity at which 
tornadogenesis occurred.  In just one volume scan, the strength of 
the vortex as sampled by KGSP increased from the lower bound of the 
"tornado possible" category (10.4 x 10-3 s-1) at 1933 UTC to 
"tornado likely" (70.0 x 10-3 s-1) at 1937 UTC.  This was an 
intensification of 70% in four minutes.  This illustrated the 
importance to warning forecasters of thoroughly evaluating each radar 
volume scan and taking quick, decisive action in making warning 
decisions during tropical cyclone events.

Figure 24.  As in Fig. 20, except for the Pendleton tornadic storm.  
Click on image to enlarge.

c.  The Piedmont Tornado

The third tornadic storm on 26 August 2008 developed over southern 
Greenville County shortly after 2000 UTC.  The radar reflectivity 
and SRV images from KGSP at 1954 UTC (approximately 10 minutes 
prior to tornado occurrence) revealed weak rotational shear 
(6.8 x 10-3 s-1) and a cyclonic curvature in the reflectivity field 
(Fig. 25).  However, Vr at this time was 25.0 kt, within the range 
suggested by Schneider and Sharp (2007) as a tornado warning threshold. 
Images from KGSP at 2003 UTC (Fig. 26) revealed a rapid intensification 
of this circulation from 1954 UTC.  An EF1 tornado touched down just 
east of the Pelzer community at 2005 UTC (Fig. 27).  Rotational shear 
increased to 24.5 x 10-3 s-1, which was an increase of over 70% from 
the previous volume scan.

Figure 25.  As in Fig. 16, except for 1954 UTC.  Click on images to enlarge.

Figure 26.  As in Fig. 16, except for 2003 UTC.  Click on images to enlarge.

Figure 27.  Map indicating the track of the Piedmont tornado.  
Movement of the tornado was from the lower part of the image to the 
upper part (south to north).  Click on image to enlarge.

Figure 28.  As in Fig. 20, except for the Piedmont tornadic storm.  
Click on image to enlarge.

Due to the unexpected, rapid increase in the strength of the vortex, 
a warning was issued with negative lead time for southern Greenville 
County (at 2007 UTC).  Using the Vr guidelines suggested by Schneider 
and Sharp (2007) would have provided much greater opportunity for a 
warning with positive lead time than rotational shear analysis, as
suggested by Figure 28.  While Vr was at or above 25 kt (with 
vortex diameter of less than 3 km) for 32 minutes prior to tornado 
occurrence, rotational shear did not "spike" to the "Tornado Likely" 
category until two minutes before the tornado.

6.  Heavy Rain and Flooding 

Rain totals within the region continued to increase throughout the 
afternoon (Fig. 29).  It was at this juncture that accumulated rainfall 
and excessive runoff contributed to several streams and creeks 
approaching and exceeding flood stage across York, Mecklenburg and 
Cabarrus counties.  For example, Figs. 30 and 31 show these effects on 
river levels at two locations, Mallard Creek in northeast Mecklenburg 
County, which is a tributary of the Rocky River, and the Rocky River 
gage in southern Cabarrus county, just above Irish Buffalo Creek.  Both 
of these gages exceeded flood stage during the afternoon of 26 August, 
and river levels continued to rise into the 27th, cresting at record 
levels.

Figure 29.  As in Fig. 6, except ending at 2203 UTC on 26 August.  Click 
on image to enlarge.

Figure 30.  Time series of stream gage height for Mallard Creek below 
Stony Creek near Harrisburg, North Carolina, for the period 24-31 August.  
Flood stage is shown by the yellow line.  Click on image to enlarge.

Figure 31.  Time series of stream gage height for Rocky River above 
Irish Buffalo Creek near Rocky River, North Carolina, for the period 
24-31 August.  Flood stage is shown by the yellow line.  Click on image 
to enlarge.

Excessive rainfall also increasingly became a concern across the western 
CWFA during the afternoon of 26 August as Flash Flood Warnings were 
issued for the North Carolina counties of McDowell, Rutherford, Mitchell 
and Yancey, where various roads were closed due to flooding and mudslides 
into the evening hours.  With the expanding bullseye of rainfall across 
northeast Georgia into the southern escarpment of the North Carolina 
Mountains, additional Flash Flood Warnings were issued for the Georgia 
counties of Rabun and Habersham, the North Carolina counties of Macon, 
Jackson, Haywood and Transylvania, and Oconee County, South Carolina, 
into the evening.

During the evening hours of 26 August, as the center of Fay degenerated 
into a remnant low over north Alabama and moved into southeast Tennessee, 
less favorable air was beginning to encroach from northeast Georgia. 
This was exhibited on the upper air sounding at FFC at 0000 UTC on 
27 August (Fig. 32), as notable drying aloft occurred, and the mean 
lower and mid level flow veered to west-southwest.

Figure 32.  As in Fig. 4, except at 0000 UTC on 27 August 2008.  Click on 
image to enlarge.

Rainfall trends throughout the evening featured diminishing activity 
across the central and southern North Carolina Mountains, and then 
across northeast Georgia and western Upstate South Carolina (Fig. 33).  
The combination of lingering rainfall and antecedent conditions prompted 
additional Flash Flood Warnings for the northern North Carolina Mountains 
and Foothills during the latter half of the evening, with flooding 
occurring on both sides of the Blue Ridge (Figs. 34 and 35).

Figure 33.  Storm Total Preciptitation estimate from KGSP for the 48 hour 
period ending 0359 UTC 27 August 2008.  Click on image to enlarge.

Figure 34.  Time series of stream gage height for the South Toe River 
near Celo, North Carolina, for the period 24-31 August.  Flood stage is 
shown by the yellow line.  Click on image to enlarge.

Figure 35.  Time series of stream gage height for the Johns River at 
Arneys Store, North Carolina, for the period 24-31 August.  Flood stage 
is shown by the yellow line.  Click on image to enlarge.

Widespread and locally heavy rain gradually diminished from the southwest 
across the Foothills and Piedmont during the overnight hours, but not 
before several more Flash Flood Warnings were issued for much of the North 
Carolina Piedmont through early Wednesday morning.

Click here to view a 19 frame Java loop of regional radar mosaic
imagery from 1856 UTC on 26 August to 1156 UTC on 27 August.

Event total rainfall was expectedly impressive, and the magnitude and 
orientation of the amounts were well forecast across the western CWFA, 
although a minimum was noted in the Smokies and lower French Broad Valley. 
Radar rainfall totals (Fig. 36) underestimated reality by a rough factor 
of 1.5 to 1.75 in comparison to the observed raingage values (Fig. 37).  
The underestimation of the secondary maximum across the Interstate 85 
corridor from Rock Hill, South Carolina, to Charlotte to Salisbury, 
North Carolina, was notable, especially given that there was record 
flooding in spots.  No main stem river flooding occurred, although the 
French Broad River at Rosman, North Carolina, did just reach flood stage, 
and the downstream gage at Blantyre crested just below flood stage, 
probably due to the low flows at the onset of the event.  The main 
Piedmont precipitation axis was aligned downstream of the forecast 
point at Lowell, North Carolina, and there was no main stem flooding 
of the Catawba River downstream either.

Figure 36.  Storm Total Preciptitation estimate from KGSP for entire event
ending at 1200 UTC 27 August 2008.  Click on image to enlarge.

Figure 37.  Total observed rainfall for the period 25-27 August 2008.  
Click on image to enlarge.

7.  Summary and Conclusions 

Several tornadoes occurred on 26 August 2008 across Upstate South 
Carolina and northeast Georgia in association with the remnants of 
Tropical Storm Fay.  Examination of environmental data indicated 
the synoptic pattern was very similar to previous tropical cyclone-
associated tornado events across the region.  Low-level helicity 
was very high, while CAPE was adequate for strong convection.  In 
addition, a weak baroclinic boundary was analyzed across the region, 
with surface-based potential instability noted even on the cool 
side of the boundary.  The three tornadoes mentioned in this study 
occurred within the cool air 30 to 40 km north of the boundary.

The tornadoes that occurred during the afternoon of 26 August 2008 
were associated with vortices characterized by rotational shear 
values of 25 x 10-3 s-1 or greater, all within the "tornado 
probable" or "tornado likely" portion of the rotational shear 
nomogram.  Values of rotational velocity were generally between 25 
and 40 kt.  This is consistent with the findings of Schneider and 
Sharp (2007) for tropical cyclone tornadoes.  However, it should be 
noted that there were a number of rotating updrafts that produced 
Vr values of greater than 25 kt that were apparently not associated 
with tornadoes.  Rotational shear seemed to be a better discriminator 
between tornadic and non-tornadic vortices on 26 August 2008.  
However, from a warning decision-making perspective, sole reliance 
on the rotational shear nomogram over the warning guidelines suggested 
by Schneider and Sharp (2007) would have resulted in short, and in 
some cases, negative lead time.  It is recommended that forecasters 
use both of these tools in conjunction with reflectivity analysis 
in making tornado warning decisions in tropical cyclone environments.

The most significant flash flooding occurred across the heavily
urbanized area around Charlotte, North Carolina, late on 26 August
and early on 27 August.  Some smaller streams exceeded flood stage, 
most notably Mallard Creek and the Rocky River tributary of Irish
Buffalo Creek.  However, none of the main stem rivers across the
Foothills and Piedmont of the Carolinas reached flood stage.  The
same was true for the mountains, with the exception of the headwaters
of the French Broad River.  Most likely, this was due to very dry
antecedent conditions with most streams and rivers at all-time record
low flows prior to the arrival of the rain associated with Fay.
Although heavy rain was widespread across most of the region, not
enough fell to mitigate the ongoing drought across the western 
Carolinas.  Most of Upstate South Carolina remained in the 
exceptional category into early September after Fay's rains, but 
the North Carolina Piedmont saw some relief.

Pictures

References

Andra, Jr., D. L., 1997:  The origin and evolution of the WSR-88D 
     mesocyclone recognition nomogram.  Preprints, 28th Conference 
     on Radar Meteorology, Austin, TX, Amer. Meteor. Soc., 364-365.

Falk, K., and W. Parker, 1998:  Rotational shear nomogram for 
     tornadoes.  Preprints, 19th Conference on Severe Local Storms, 
     Minneapolis, MN, Amer. Meteor. Soc., 733-735. 

Lane, J. D., 2005:  Environmental aspects of two tornado outbreaks 
     associated with landfalling tropical cyclones.  Preprints, 4th 
     Southeast Severe Storms Symposium, Starkville, MS, Mississippi 
     State University.

Markowski, P. M., E. N. Rasmussen, and J. M. Straka, 1998:  The 
     occurrence of tornadoes in supercells interacting with boundaries 
     during VORTEX-95. Wea. Forecasting, 13, 852-859.

McCaul, E. W., Jr., 1991:  Buoyancy and shear characteristics of 
     hurricane-tornado environments. Mon. Wea. Rev., 119, 1954-1978.

McCaul E. W. Jr., D. E. Buechler, S. J. Goodman, and M. Cammarata, 
     2004:  Doppler radar and lightning network observations of a 
     severe outbreak of tropical cyclone tornadoes. Mon. Wea. Rev.,
     132, 1747–1763.

Schneider, D., and S. Sharp, 2007:  Radar signatures of tropical 
     cyclone tornadoes in central North Carolina.  Wea. Forecasting, 
     22, 278-286.

Acknowledgements

The authors wish to thank Mr. Rob Harrison of the South Carolina
Department of Natural Resources for providing the image of the 
Clemson Tornado near Lake Hartwell.  The track plot of Tropical
Cyclone Fay was obtained from the National Hurricane Center.  
Surface analysis graphics were obtained from the Hydrometeorological
Prediction Center.  Satellite imagery, radar mosaics and surface 
observation plots were obtained from the University Corporation 
for Atmospheric Research.  Time series graphics for rotational
shear and velocity were created using Microsoft Excel.  Radar 
images used in the tornado study were created using the GR2Analyst 
software.  The upper air sounding image in figure ii was created 
using the RaOB (Radiosonde Observation Program) version 5.8 for
Windows.  Tornado damage path graphics were prepared using Delorme 
Street Atlas USA 2006 Plus.  Storm Total Precipitation estimate
graphics were created using the Java NEXRAD viewer obtained from
the National Climatic Data Center.  Stream gage plots were obtained
from the United States Geologic Survey.  Blair Holloway created
the rainfall map.