Flash Flooding Along the Blue Ridge
26 June 2006
Patrick D. Moore
The Pickens Jockey Lot was under five feet of water when Twelvemile Creek came out of its banks after heavy rain in the morning of Monday, 26 June 2006. Image provided by the Pickens Sentinel and used by permission.
Author's Note: The following report has not been subjected to the scientific peer review process.
Heavy rain during the early morning of Monday, 26 June 2006, resulted in flash flooding across parts of Pickens County and northern Greenville County in South Carolina, and in parts of Henderson, Polk, and Rutherford counties in North Carolina. The most significant rainfall occurred roughly in a band from the Lake Keowee area, northeast along the Blue Ridge, to Lake Lure, the Hickory Nut Gorge, and the upper reaches of the Rocky Broad River basin (Fig. 1). The most significant flooding was observed in the basin of Twelvemile Creek in Pickens County, South Carolina, and in the basin of the Rocky Broad River in the southern Foothills of North Carolina. During the course of the event, the National Weather Service office at Greer, South Carolina (GSP), issued 14 Flash Flood Warnings, along with numerous flood statements.
(Click here to view a summary of significant weather and flood reports for 26 June 2006.
(Click here to view a summary of rainfall reports for 25-27 June 2006.
Figure 1. Storm total rainfall for the period 25-27 June 2006. Click on image to enlarge.
This event is notable in that the stream gage along Twelvemile Creek near Liberty in Pickens County, South Carolina, reached its highest level ever recorded (14.78 feet). Farther north, a debris flow occurred in steep terrain inside Jones Gap State Park, which closed a popular hiking trail for at least six months afterward. Additionally, the Rocky Broad River tied its record stage at Bat Cave, in extreme northeast Henderson County, North Carolina (11.0 feet), although the Rocky Broad River is known to have reached higher levels prior to the installation of the river gage, such as 4 September 1996 (Johnstone and Burrus 1998).
2. Synoptic Features and Pre-Storm Environment
Water vapor satellite imagery on the morning of 25 June (Fig. 2, left) showed a well-defined mid-level circulation over the upper Mississippi Valley and another off the east coast of Florida. Both circulations are readily apparent in the 700 mb analysis at 0000 UTC on 26 June (Fig. 3). Over the next 24 hours, convergence at the edges of both circulations resulted in a deep plume of tropical moisture extending from the west coast of the Florida Peninsula, northward across the southern Appalachians, to the upper Ohio Valley (Fig. 2, right).
Figure 2. GOES-12 Water Vapor imagery at 1145 UTC 25 June (left) and 1145 UTC 26 June (right). Click on images to enlarge.
Figure 3. Storm Prediction Center (SPC) 700 mb objective analysis of geopotential height, temperature, and dew point at 0000 UTC on 26 June. Note the area of very high dew point stretching from the coast of Georgia and South Carolina to the northern edge of the Bahamas. Click on image to enlarge.
By 1200 UTC on 26 June, the environment aloft was favorable for deep convection across the western Carolinas. The 500 mb analysis (Fig. 4) showed a short wave over north Georgia lifting northeast out of the base of the upper trough located over the Mississippi Valley. On the 250 mb analysis (Fig. 5), the western Carolinas were under the right entrance region of a jet streak over the Ohio Valley and Great Lakes. The approaching short wave and jet entrance region were both conducive to upward vertical motion on the morning of 26 June.
Figure 4. SPC 500 mb objective analysis of geopotential height, temperature, and wind at 1200 UTC on 26 June. Click on image to enlarge.
Figure 5. SPC 250 mb objective analysis of isotachs and streamlines at 1200 UTC on 26 June. Click on image to enlarge.
The upper air sounding taken at Peachtree City, Georgia (FFC), at 1200 UTC showed a relatively modest amount of Convective Available Potential Energy (~ 1000 J/kg) for early Summer owing to a nearly moist adiabatic temperature profile (Fig. 6). However, the precipitable water was very high (1.97 inch, which is approximately 150% of normal).
Figure 6. Skew-T log P diagram (upper left) and hodograph (upper right) for upper air sounding at FFC at 1200 UTC 26 June. The tables at the bottom summarize several objective parameters used by the SPC to determine severe weather potential. Click on image to enlarge.
At the surface, a quasi-stationary frontal boundary (Fig. 7) provided the necessary low level convergence to focus the development of showers and thunderstorms over the western Carolinas during the morning of the 26 June. The air mass to the east of the front was rich with low level moisture as evidenced by dewpoints in the lower 70s. The low level southeast wind continued to supply the developing storms with Atlantic moisture. The transport of moisture can be inferred from the visible satellite imagery by the bands of low level clouds oriented from southeast to northwest across the Carolinas (Fig. 8)
Figure 7. Hydrometeorological Prediction Center (HPC) surface pressure and fronts analysis at 1200 UTC 26 June. Click on image to enlarge.
Figure 8. Visible satellite image from GOES-12 at 1345 UTC 26 June.
3. Radar observations
Between the early afternoon of 25 June and the early morning hours of 26 June, at least three distinct waves of showers moved across the area of the Blue Ridge Escarpment from Lake Keowee to Lake Lure. This early rainfall, especially between 0900 UTC and 1200 UTC on 26 June, served to saturate the soil across the area from Oconee County, South Carolina, to Rutherford County, North Carolina. New convective development over the western part of Upstate South Carolina around 1200 UTC (Fig. 9), moving northward toward the Blue Ridge, suggested an increased threat of runoff.
Figure 9. Radar reflectivity mosaic centered on KGSP radar at 1200 UTC 26 June. Click on image to enlarge.
The band of showers and thunderstorms that developed between Interstate 85 and the Blue Ridge between 0800 and 0840 UTC was ultimately responsible for producing the flooding (Fig. 10). This band of showers developed at least partly in response to low level southeast upslope flow and low level convergence on the moist side of the stationary front seen in Figure 7.
Figure 10. Radar reflectivity on 1.3 degree scan from the KGSP WSR88-D at 0840 UTC 26 June. The radar is located just to the right of the 'e' in Greenville. The point labelled 'Liberty' is the site of the stream gage on Twelvemile Creek. The point labelled 'Bear' is the location of the North Carolina ECOnet station on Bearwallow Mountain. Jones Gap State Park and Bat Cave, North Carolina, are also shown for reference. Click on image to enlarge.
By 1225 UTC, the convergence in the south-southeast upslope flow had organized the deep convection into distinct north-south oriented bands, running up the Blue Ridge Escarpment (Fig. 11). The heaviest rain was occurring in the band that stretched from Bat Cave, North Carolina, southward into Greenville County, South Carolina, and in the band that stretched from near Jones Gap State Park southward across eastern Pickens County. By this time, the initial band of precipitation seen in Figure 10 had moved over the North Carolina mountains and formed a large area of stratiform rain across Jackson, Haywood, and Buncombe counties. Flash flooding was imminent at both Jones Gap State Park and along the Rocky Broad River at Bat Cave.
Figure 11. As in Figure 10, except for 0.8 degree scan at 1225 UTC. Click on image to enlarge.
The heavy rain persisted across the Blue Ridge near the border of North Carolina and South Carolina through about 1600 UTC. The primary rain bands finally moved north and east of Lake Lure, and shifted east of Pickens County, after 1630 UTC (Fig. 12).
Figure 12. As in Figure 11, except for 1631 UTC. Click on image to enlarge.
4. Hydrologic Observations
Flooding developed first across eastern Henderson County from Saluda to the upper part of the Rocky Broad River above Bat Cave between 1200 UTC and 1330 UTC. The heaviest rain fell across the Blue Ridge from Caesar's Head up to Bat Cave, where most locations measured in excess of 9 inches for the event. Rain fell at the highest rate at the North Carolina ECOnet site on Bearwallow Mountain between 1200 UTC and 1300 UTC, when over one inch was recorded (Fig. 13). It is likely that rain fell at an even greater rate closer to Bat Cave, as the band of highest reflectivity was located just east of Bearwallow Mountain during this time frame (see Figure 11 and the reflectivity loop from 0840 UTC to 1220 UTC). In fact, the automated recording station at Bat Cave received 1.46 inches of rain in the 30 minute period from 1130 UTC to 1200 UTC (Fig. 14). This period of intense rainfall quickly sent the Rocky Broad River out of its banks at Bat Cave. The river gage showed a very rapid rise from less than 4 feet at 1000 UTC to nearly 9.5 feet by 1400 UTC (Fig. 15). Flood stage was exceeded probably around 1215 UTC, with a crest of 11 feet around 1500 UTC. The river fell below flood stage around 2000 UTC.
Figure 13. Accumulated rainfall at the North Carolina ECOnet site at Bearwallow Mountain in northeast Henderson County. The time axis is local time. Click on image to enlarge.
Figure 14. Accumulated rainfall at the Army Corps of Engineers site at Bat Cave in extreme northeast Henderson County. The time axis is in UTC. Click on image to enlarge.
Figure 15. River stage on Rocky Broad River at Bat Cave, North Carolina. The red line indicates the flood stage at 7.5 feet. The purple line indicates the level above which major flooding can be expected, which is 9.5 feet. The time axis is in UTC. Click on image to enlarge.
Heavy rain fell across most of Pickens County in two main batches. The first period of heavy rain fell from about 0900 UTC to 1100 UTC. The second, more significant period of heavy rain fell from about 1300 UTC to 1600 UTC, as seen by the radar imagery in the second reflectivity loop. The level of Twelvemile Creek came up at the rate of 3 feet per hour after 1400 UTC, eventually pushing it out of its banks near Pickens between 1430 UTC and 1500 UTC. Although flooding was reported in low lying areas, the creek did not rise above flood stage at the gage site near Liberty (13.5 feet) until much later, around 0200 UTC on 27 June (Fig. 16), as runoff moved down from the upper reaches of the basin north and east of Pickens. A crest of 14.78 feet was reached around 0500 UTC and the creek dropped below flood stage around dawn on June 27, although flooding persisted at the Pickens Jockey Lot until late in the morning.
Figure 16. River stage on Twelvemile Creek near Liberty, South Carolina. Flood stage is 13.5 feet. The time axis is in UTC. Click on image to enlarge.
The axis of heaviest rainfall extended across the northern part of Greenville County, South Carolina, including Jones Gap State Park. Rainfall of greater than 9 inches was sufficient to cause a debris flow within Jones Gap State Park att some time prior to 1400 UTC, when the aftermath was discovered. Numerous trees, boulders, and other debris were carried several hundred yards down a steep slope, closing a main hiking trail (Fig. 17).
Figure 17. The aftermath of a debris flow in Jones Gap State Park during the morning of Monday, 26 June. The large boulder in the middle of the image is several feet in diameter.
Studies of previous devastating flash flood events in the area of the Hickory Nut Gorge area of western North Carolina have focused on a substantial low level easterly moist upslope flow relative to the unique terrain features of the gorge as a significant contributor to the production of heavy rain (Lee and Goodge 1984, Johnstone and Burrus 1998). In contrast, the atmosphere on the morning of 26 June 2006 had a deep unidirectional southerly flow, as seen in the upper air soundings at Peachtree City, Georgia (Fig. 7), and Greensboro, North Carolina. However, when viewed in hindsight from the perspective of an ingredients- based methodology for forecasting flash floods (Doswell et al. 1996), the events of 26 June 2006 could have been anticipated just as easily.
Convective precipitation in the late evening hours of 25 June and the early morning hours of 26 June provided the antecedent conditions for increased runoff. The atmosphere on the morning of 26 June possessed copious amounts of moisture with little dry air aloft and enough available potential energy to support strong convective updrafts, which suggested that high precipitation rates were possible. Several low level mechanisms, in the presence of a favorable synoptic environment for upward vertical motion, forced the development of numerous showers and thunderstorms.
A large stratiform rain region formed over the North Carolina mountains by 1300 UTC as the initial wave of deep convection moved up the eastern slopes of the Appalachians and over the quasi-stationary front (Fig. 18). It is hypothesized that the resulting rain-cooled air mass across the North Carolina mountains effectively relocated the surface front to a position very near the Blue Ridge by 1300 UTC. Surface observations at 1300 UTC (Fig. 19) show a light north wind at Asheville (AVL) and a south-southeast wind at reporting stations across upstate South Carolina (GSP, AND) and the southern Foothills of North Carolina (FQD).
Figure 18. As in Figure 11, except at 1254 UTC. The deep convection is shown by the higher reflectivity values extending in an arc across Oconee, Pickens, and northern Greenville counties to the area around Bat Cave. The stratiform rain region is denoted by the relatively low values of reflectivity across Jackson, Haywood, and Buncombe counties. Click on image to enlarge.
Figure 19. Regional surface plot at 1343 UTC 26 June. In spite of the time stamp, most observations are valid at 1255 UTC. Observations are plotted according to the standard station model. Click on image to enlarge.
The south-southeasterly flow of unstable and very moist low level air up against the newly-relocated surface boundary, nearly coincident with the sharp terrain rise of the Blue Ridge, enhanced precipitation production. New cells developing in the convergent low level flow over upstate South Carolina moved in a direction nearly parallel to the new boundary, which significantly increased the duration of heavy rain along the boundary.
The environment across the western Carolinas on the morning of 26 June had enough moisture, instability, and forcing for deep convection to develop. Flash flooding across the basins of Twelvemile Creek and the Rocky Broad River resulted when the necessary ingredients were present in an area with favorable hydrologic conditions. The repeated movement of convective cells with a high rainfall rate directed nearly parallel to a low level boundary resulted in heavy rain of sufficient duration to produce flooding.
More images from the flooding along Twelvemile Creek and Town Creek near Pickens, South Carolina, on 26 June 2006. Images courtesy of the Pickens Sentinel. Used by permission.
More images from the debris flow in Jones Gap State Park on 26 June 2006. The large boulders in the uppermost images are at least five or six feet in diameter. The debris flow left a long and deep scar on the slope and deposited a large amount of mud and tree trunks below.
Sandy Foster, editor of the Pickens Sentinel, provided the images of the flooding along Twelve Mile Creek and Town Creek near Pickens, South Carolina. Greg Schoor (NWS) prepared the map of the storm total rainfall for the event. The satellite imagery, radar mosaic imagery, and the surface observations plot were obtained from the University Corporation for Atmospheric Research. Upper air analyses and sounding plots were obtained from the Storm Prediction Center. The surface analyses were obtained from the Hydrometeorological Prediction Center. The local radar graphics were prepared with the Java NEXRAD Viewer, obtained from the National Climatic Data Center. The rainfall data for Bearwallow Mountain was obtained from the State Climate Office for North Carolina.
Doswell, C. A., III, H. E. Brooks, and R. A. Maddox, 1996: Flash flood forecasting: An ingredients-based methodology. Wea. Forecasting., 11, 560-581. Johnstone, T. P., and S. A. Burrus, 1998: An analysis of the 4 September 1996 Hickory Nut Gorge Flash Flood in western North Carolina. Preprints, 16th Conf. on Weather Analysis and Forecasting, Phoenix, AZ, Amer. Meteor. Soc., 275-277. Lee, L. G., and G. W. Goodge, 1984: Meteorological analysis of an intense "east-slope" rainstorm in the southern Appalachians. Preprints, 10th Conf. on Weather Analysis and Forecasting, Clearwater Beach, FL, Amer. Meteor. Soc., 30-37.