Event Summary
     National Weather Service, Raleigh NC

Tropical Storm Fay, August 2008
Updated 2009/06/01


Satellite Image of Tropical Storm Fay at 1615Z on 2008/08/19 - Click to enlarge


Event Headlines -

...Tropical Storm Fay made four Florida landfalls with the first occurring on August 18th at Key West and the last occurring during the early morning hours of August 23rd near Carrabelle in the Florida Panhandle...
...Very heavy rain was observed over a three day period across central North Carolina with a large area of 3 to 6 inches of rain observed with localized amounts exceeding 8 inches. The heavy rain produced significant flooding in the Charlotte and Triad areas...
...Despite indications of rotation in the lower levels of numerous thunderstorms, only three tornadoes were confirmed in central North Carolina...
...This severe weather event was one of the first significant convective events at NWS Raleigh using super resolution radar data. The event was re-examined using the Weather Event Simulator and a variety of imagery was saved and documented below...
...The local gauge corrected radar QPE product or Q2RAD_HSR_GC product available from the National Mosaic and QPE (NMQ) web page was examined and it appeared to provide a very accurate and detailed analysis of precipitation amounts...


Event Overview -

The origin of Fay can be traced back to a tropical wave that moved off the coast of Africa on August 7th. The system passed just south of the Cape Verde Islands, and tracked generally westward during the next several days. The system showed signs of development a couple of times during its westward track across the Atlantic, but the convection was not sustained and the system remained a weak area of low pressure. Reconnaissance aircraft that investigated the system on August 12th and 14th only found a broad area of low pressure. The low pressure system continued to move westward on the periphery of a subtropical ridge of high pressure located to its north during the second week of August.

The system became better organized after passing the northern Lesser Antilles, and it developed an area of deep convection as it approached the Virgin Islands and Puerto Rico. On August 15th, reconnaissance aircraft investigated the system and found flight level wind gusts of 56 MPH northeast of the center. The system was named Tropical Storm Fay at 500 PM EDT on Friday, August 15th. Fay then moved westward across the island of Hispaniola and weakened. On August 16th, the storm emerged over water and re-organized as it skirted the southern coast of Cuba. Fay turned northwestward toward the Florida Keys as the high pressure ridge to its north weakened.

At 300 PM EDT on August 18th, Fay made landfall over Key West Florida, its first of four Florida landfalls. Fay made its second landfall at 500 AM EDT on August 19 at Cape Romano south of Naples. The storm then maintained its strength or possibly strengthened slightly for a time as it crossed central Florida. Eventually, the storm weakened as it interacted with the Florida land mass and then moved offshore near Melbourne on the Florida east coast. Fay moved slowly northward, just off the Florida east coast on August 20th. Eventually, Fay was pushed westward and made another landfall at Flagler Beach during the afternoon of the 21st. Fay then moved west across northern Florida for more than a day before emerging into the extreme northeastern Gulf of Mexico during the evening of August 22nd. Fay made its fourth landfall during the early morning hours of August 23rd near Carrabelle in the Florida Panhandle.

The remnants of Fay were located about 20 miles south-southeast of Meridian Mississippi near the Mississippi-Alabama state line at 800 AM EDT on Sunday, August 24th. The remnants drifted west during the next 24 hours, nearly reaching Louisiana early Monday morning, August 25th, before drifting northeast later Monday and Tuesday. The first large areas of precipitation associated with the remnants of Fay moved into western and southern portions of North Carolina on Monday afternoon and evening. By Tuesday morning, August 26th, the remnants were centered over far northern Alabama.

On Tuesday afternoon, a cold front pushed south through much of North Carolina as an area of high pressure moved into the eastern Great Lakes region. The weak area of low pressure associated with the remnants of Fay began to interact with the western portion of the cold front. As the remnants slowly lifted northeast on Tuesday night and Wednesday, the front started lifting north as a warm front. Several rounds of precipitation including some convection moved across western North Carolina from Tuesday morning through Wednesday morning, producing a large area of 3 to 6 inches of rain with localized amounts exceeding 8 inches. The heavy rain produced significant flooding in the Charlotte and Triad areas.

By Wednesday morning, August 27th, the front extended southeast across western and southern North Carolina. The front then edged slightly northward on Wednesday as the weak area of low pressure drifted northeast. By midday Wednesday, a large area of heavy precipitation over North Carolina was moving north into southern Virginia. At around the same time, some breaks in the cloud cover were developing upstream in northern South Carolina and southwestern North Carolina. Unstable cloud streets were noted over the clearing area. Cloud streets are rows of cumulus clouds aligned parallel to the low-level wind. They typically form when the lower portions of the atmosphere are unstable, with a capping inversion aloft (RUC Analysis sounding at KFAY at 17 UTC on Wednesday, August 27.)

The region south of the stationary front destabilized as surface heating continued into the afternoon. With ample moisture, scattered convection redeveloped during the late morning hours and continued into the afternoon. Despite the fact that the low level jet was easing away from central North Carolina and that the bulk shear values were marginal for tornadoes (around 30 kts), the east and southeast surface wind flow interacting with the lingering boundaries provided enough shear for updrafts to rotate. As the convection moved across central North Carolina, three tornadoes were confirmed and dozens of spotter reports of funnel clouds or rotating wall clouds were received.


Precipitation Totals from Tropical Storm Fay

The map below contains a local precipitation analysis across North Carolina for the period in which Fay generated precipitation across North Carolina from Monday, August 25th through Thursday, August 28th, 2008.

Precipitation totals from the remnants of Tropical Storm Fay from Monday, August 25 through Thursday, August 28, 2008


The map below contains a precipitation analysis across the southeastern U.S. from the Hydrologic Prediction Center for the period of August 17th through August 29th, 2008. Click on the map to open a larger image.

Precipitation analysis from Tropical Storm Fay and its remnants from August 17 through August 29, 2008 - click to enlarge



Severe Weather reports from Tropical Storm Fay

Text of severe weather reports across central North Carolina

The map below contains severe weather reports received by the National Weather Service during the period in which Fay impacted North Carolina from Monday, August 25th through Thursday, August 28th, 2008.





Regional Radar Loop

A Java loop of hourly regional radar imagery from 0558 UTC on August 25th through 0258 UTC on August 29th, 2008 is available here. Note - this loop includes 90 frames.

The regional reflectivity image below is from 1158 UTC (758 AM EDT on August 27, 2008). A multi-cell cluster of convection was moving across the Sandhills and Southern Coastal Plain while a larger area of rainfall was moving across the Northern Piedmont into southern Virginia. This area of precipitation produced rainfall amounts approaching 1.5 inches per hour near the Triad area based on one hour precipitation estimates ending at 1125 UTC.


Regional reflectivity image - click to load loop



KRAX Radar Loops

A Java loop overview of the entire event with images from every hour between 0559 UTC August 27 through 0559 UTC August 28, 2008 is available here. Note - this loop includes 25 frames

A Java loop overview of the entire event with images from every 15 minutes between 0559 UTC August 27 through 0559 UTC August 28, 2008 is available here. Note - this loop includes 97 frames

The KRAX reflectivity image below is from 1827 UTC or 227 PM EDT on August 27, 2008 when three Tornado Warnings were in effect along with three Flash Flood Warnings.





Surface Analysis

The surface analysis from 21 UTC on Wednesday, August 27th, 2008 shown below depicts the larger scale surface features during the last significant day of the event. A stationary front extends southeast from a weak surface low located over northeast Tennessee which was the surface remnants of Fay. The stationary front extends across western and southern North Carolina providing a focus for moisture and enhancing the severe weather event.

A Java loop of surface analysis imagery from 12 UTC Sunday, August 17th, through 00 UTC Friday, August 29th, 2008 shows the path that Tropical Storm Fay took across Florida, the Deep South and the southern Appalachians over a period of 12 days.

Surface analysis from 21 UTC Wednesday, August 27, 2008



The Randleman Tornado

A National Weather Service storm survey confirmed an EF-0 tornado touched down in Randleman in Randolph County. The tornado had an estimated path length of 500 yards and a width estimated at 50 yards. The tornado occurred just after 1230 PM on August 27. Winds were estimated at 60 mph.

An account of the event, which occurred on Applewood Road, was of a dark funnel cloud that appeared from the south behind an area of trees and then hit several residences, particularly affecting three residences along the east side of Applewood Road. Most of the debris from damage to two outbuildings and trees was blown in an easterly direction. Moderately sized, cement yard statues were also tossed to the east under the skirting of a manufactured home. However, a light metal carport was thrown well across Applewood Road to the southwest, and a dog was apparently picked up by the tornado and carried some distance to the northwest. Fortunately, the dog survived, was found a few hours later and returned to its owners. The last observed damage was to a tree and a snapped large limb, just southeast of the intersection of Applewood Road and Creekridge Country Road. There was some siding from residences found in those trees. Spotters down wind of the tornado touchdown area near the intersection of New Salem Road and Old Greensboro Road reported thousands of leaves falling from the sky. There were no reported injuries.

Tornado: EF-0
Peak Wind: 60 mph
Path Length: 500 yards
Path Width: 50 yards
Time/Date: 1230 PM EDT Wednesday, August 27, 2008
Injuries: None
Fatalities: None

Special thanks to the Randolph County emergency services group and other local emergency services for assisting with the survey and providing access to damaged areas.


Randleman Tornado Track -





Photos and video of the Randleman Tornado -

Hal Pugh, a former weather observer for the Agricultural Weather Station in Randolph County that was associated with NC State University and NOAA in the 1970's and 1980's passed along the report below as well as the photographs and video.

At around 1245 PM EDT, on August 27th, after a storm had passed, Hal and his wife observed several thousand green leaves from a variety of trees floating from the sky for as high as one could see. A neighbor located about 1000 feet to the east of the Pugh residence, also witnessed the event at the same time. Hal later heard that a tornado had occurred and realized the storm path would have been toward him and that the leaf fall was likely associated with the tornado that occurred 10 minutes or so earlier.

In addition, Hal obtained some photographs and video of the same tornado from Ashley Lemons. Ashley Lemons took the photographs and video via a cell phone camera on New Salem Road looking southwest.

A map of the tornado touchdown location and the location in which the photos and video were taken is shown below.

still image from Randleman Tornado video taken by Ashley Lemons
Note - the video is 30 seconds in length and approximately 0.7 MB in size.

Randleman Tornado photo - photo courtesy of Ashley Lemons - Click to enlarge           Randleman Tornado photo - photo courtesy of Ashley Lemons - Click to enlarge           Randleman Tornado photo - photo courtesy of Ashley Lemons - Click to enlarge           Randleman Tornado photo - photo courtesy of Ashley Lemons - Click to enlarge          

Map of the Randleman Tornado and location where the photos and video were taken - Click to enlarge


Randleman Tornado Damage Photos -

Photos courtesy of the National Weather Service.
(Click on the image to enlarge)


Photo courtesy of the National Weather Service - Click to enlarge           Photo courtesy of the National Weather Service - Click to enlarge           Photo courtesy of the National Weather Service - Click to enlarge           Photo courtesy of the National Weather Service - Click to enlarge          


Randleman Tornado Radar Imagery -

The severe thunderstorm that produced the Randleman Tornado did not possess an impressive radar signature. The images below are from the NWS Raleigh KRAX WSR-88D Doppler radar located near Clayton North Carolina or about 10 miles southeast of Raleigh. The images are from 1633 UTC or just a few minutes after the tornado touchdown and a few minutes before the pictures and video of the tornado (above) were taken. In all of the four-panel images, the image in the upper left is from the 0.5 degree elevation angle, the upper right is the 0.9 degree elevation angle, the lower right is from the 1.3 degree elevation angle and the lower left is from the 1.8 degree elevation angle. The images at 0.5, 0.9, and 1.3 degrees are super resolution data with the 1.8 degree elevation angle image a legacy 8-bit product.

The thunderstorm was located approximately 65 miles from the RDA at the time of the tornado touchdown and the radar signature was not very impressive. The Storm Relative Velocity signature was rather poor and not indicative of a tornadic thunderstorm. The rotational pattern was ill-defined and a velocity couplet was not identified. It is believed that the circulation in the storm was likely very shallow and confined to the lowest portions of the thunderstorm. The Reflectivity pattern was more impressive with an implied hook or kidney bean pattern but it should be noted that many of the thunderstorms on August 27th showed a similar reflectivity structure and were rotating but only a few produced confirmed tornadoes.

Four-panel reflectivity imagery from the 1633 UTC volume scan - click to enlarge           Four-panel storm relative velocity imagery from the 1633 UTC volume scan - click to enlarge          



The Silk Hope Tornado

A National Weather Service storm survey confirmed an EF-0 tornado just west of Silk Hope in Chatham County on August 27th. The tornado had an estimated path length of 1.5 miles and a maximum width estimated at 100 yards. The tornado occurred around 415 PM EDT, with the time based on eyewitness accounts. Winds were estimated at 70 mph.

Accounts of the event, which began along Jesse Bridges Road between Smith Hudson Road and Rufus Brewer Road, was that the tornado was in the midst of very heavy rain with considerable thunder and lightning. Persons had heard warnings for the area, either from telephone calls from family or through media reports. Most of the damage occurred along Jesse Bridges Road with very large trees blown down in different directions. In addition, considerable damage was done to an outbuilding, and in one case a window screen was placed on top of a weathervane on the roof of a home. Isolated cases of peeled aluminum siding, flattened grass, and downed trees were noted near the intersection of Rufus Brewer and Silk Hope Liberty Roads, and on Will Brown Road. There were no reported injuries.

Tornado: EF-0
Peak Wind: 75 mph
Path Length: one and a half miles
Path Width: 100 yards
Time/Date: 415 PM EDT Wednesday, August 27, 2008
Injuries: None
Fatalities: None

Special thanks to the Chatham County emergency services for assisting with the survey and providing access to damaged areas.


Silk Hope Tornado Track -




Silk Hope Tornado Damage Photos -

Photos courtesy of the National Weather Service.
(Click on the image to enlarge)


Photo courtesy of the National Weather Service - Click to enlarge           Photo courtesy of the National Weather Service - Click to enlarge           Photo courtesy of the National Weather Service - Click to enlarge           Photo courtesy of the National Weather Service - Click to enlarge           Photo courtesy of the National Weather Service - Click to enlarge          


Silk Hope Tornado Radar Imagery -

At the time of the Silk Hope Tornado, the amount of instability was modest with MLCAPES of 500 to 1000 J/Kg. The low level shear values were moderate with 0 to 1 km storm relative helicity (SRH) values of 100-150 m²/s² and 0 to 3 km SRH values of 150-250 m²/s². The tornado appears to have developed when the thunderstorm crossed a persistent surface boundary located in the Western Piedmont.

The four-panel images below are from the NWS Raleigh KRAX WSR-88D Doppler radar located near Clayton North Carolina or about 10 miles southeast of Raleigh. The images are from 2008 UTC or 2012 UTC, just a few minutes before the tornado touchdown. In all of the four-panel images, the image in the upper left is from the 0.5 degree elevation angle, the upper right is the 0.9 degree elevation angle, the lower right is from the 1.3 degree elevation angle and the lower left is from the 1.8 degree elevation angle. The images at 0.5, 0.9, and 1.3 degrees are super resolution data with the 1.8 degree elevation angle image a legacy 8-bit product.

The images below contain four-panel reflectivity imagery from the 2008 UTC volume scan and the four-panel storm relative velocity imagery from the 2008 UTC volume scan. Note that the rotational couplet is embedded in the precipitation core. In the 0.5 degree storm relative velocity panel, the rotational velocity is 34 knots at a distance of approximately 45 miles from the RDA which corresponds to a borderline weak to moderate mesocyclone.

Four-panel reflectivity imagery from the 2008 UTC volume scan - click to enlarge           Four-panel storm relative velocity imagery from the 2008 UTC volume scan - click to enlarge          

The images below contain four-panel reflectivity imagery from the 2012 UTC volume scan and the four-panel storm relative velocity imagery from the 2012 UTC volume scan. The signature in the storm relative velocity imagery is not as apparent in this volume scan.

Four-panel reflectivity imagery from the 2012 UTC volume scan - click to enlarge           Four-panel storm relative velocity imagery from the 2012 UTC volume scan - click to enlarge          



The Fremont Tornado

The National Weather Service, in conjunction with Wayne County Emergency Management, found that an EF-0 tornado touched down to the northeast of Fremont in northern Wayne County during the early morning hours of Wednesday, August 28th, 2008. The tornado tracked north after touching down in the middle of a corn field, just southeast of the intersection of Antioch Church Road and Old Black Creek Road. At 188 Aycock Church Road, the EF-0 tornado with winds estimated around 75 mph destroyed a barn and damaged a pick up truck. Minor roof and siding damage was also noted to the house at this same location. Debris from the barn and house were scattered about 100 yards. The tornado crossed the Aycock Church Road before moving across a field of soy beans and damaging several trees. After striking the tree line, the tornado then caused a roof to collapse on a second barn and damaged two fields of tobacco. The overall path length of this first touchdown was about a half a mile with a path width of about 50 yards.

The EF-0 tornado then lifted off the ground before touching down again briefly near Beaver Dam road causing minor damage. Several mobile homes along Beaver Dam Road had shingles removed and several trees were blown down. A few small out buildings were also damaged. This second tornado touchdown was also EF-0, but weaker than the first with winds of 65 mph. The path length of the second touchdown was about 150 yards with a path width of 25-50 yards.

First Touchdown at Aycock Church Road Tornado: EF-0
Peak Wind: 75 mph
Path Length: one half mile
Path Width: 50 yards
Time/Date: 1258 AM EDT – 100 AM EDT Wednesday, August 28, 2008
Injuries: None
Fatalities: None

Second Touchdown at Beaver Dam Road
Tornado: EF-0
Peak Wind: 65 mph
Path Length: 150 yards
Path Width: 25-50
Time/Date: 102 AM EDT – 103 AM EDT Wednesday, August 28, 2008
Injuries: None
Fatalities: None


Special thanks to Wayne County emergency services and Wilson County emergency services for assisting with the survey.


Fremont Tornado Track -




Fremont Tornado Damage Photos -

Photos courtesy of the National Weather Service.
(Click on the image to enlarge)


Photo courtesy of the National Weather Service - Click to enlarge           Photo courtesy of the National Weather Service - Click to enlarge           Photo courtesy of the National Weather Service - Click to enlarge          


Fremont Tornado Radar Imagery -

The thunderstorm that rapidly developed and produced the Fremont tornado was likely enhanced as it moved across and into the cooler side of a persistent surface boundary. The mesocyclone associated with the Fremont tornado developed between 0441 UTC and 0445 UTC. The circulation continued to tighten at 0450 UTC and by 0454 UTC the outbound velocities had increased significantly. A hook signature clearly developed in the reflectivity imagery by 0454 UTC. The mesocyclone intensified even more by 0459 UTC as indicated in the rotational shear value of 34 knots and a VR Shear value of 0.104/s at 28 nm from the RDA. The 34 knots of rotational shear at 28 nm from the radar corresponds to a marginal moderate mesocyclone with the VR Shear value of 0.104/s corresponding to a "Tornado Possible" based on guidance from the Rotational Shear Nomogram for Tornadoes.

A Java loop in which the user can stop, start, and animate the imagery of KRAX base reflectivity data from 0432 UTC through 0517 UTC August 28, 2008 is available here. Note - this loop includes 11 frames

A loop of the KRAX base reflectivity imagery from 0432 UTC August 28 through 0517 UTC is shown below.




A Java loop in which the user can stop, start, and animate the imagery of the KRAX storm relative velocity data from 0432 UTC through 0517 UTC August 28, 2008 is available here. Note - this loop includes 11 frames

A loop of the KRAX storm relative velocity imagery from 0432 UTC August 28 through 0517 UTC is shown below.





Comparison of Super Resolution Data and Legacy 8 Bit Data

During the spring and summer of 2008, a long term project to upgrade the resolution of several WSR-88D radar products was fielded. The improved data resolution is called super resolution. The WSR-88D data resolution before super resolution was installed was 1 degree in azimuth by 1000 m in range for reflectivity products and 1 degree in azimuth by 250 m in range for velocity products. Super resolution data improves reflectivity and velocity resolution to 0.5-degree azimuth by 250 m range. Initially super resolution data will only be produced for “split cut” elevation angles or those elevation angles at or below around 1.5 degrees when the radar is in storm mode. Additionally, the range of the radar will be extended from 230 km to 300 km. The improved data resolution will not be used in the various radar algorithms at this point. Simulations using super resolution data show that mesocyclone and tornado signatures can be detected at greater ranges than with legacy resolution data. In addition, other smaller scale features should be detectable in base products sooner or with greater reliability.

This severe weather event was one of the first significant convective events at RAH using the super resolution data. The event was re-examined using the Weather Event Simulator and a variety of imagery was saved and documented.


Examples

The image below is a four-panel radar image from the KRAX WSR-88D valid 2003 UTC on August 27, 2008 just before the Silk Hope Tornado touchdown. The images on the top row are KRAX base reflectivity at 0.5 degrees while the images in the bottom row are storm relative velocity at 0.5 degrees. The reflectivity images on the left hand side are super resolution radar data at 0.5 degrees and 250 m resolution while the images on the right side are legacy 8-bit reflectivity imagery at 1 degree and 1000m resolution. The storm relative velocity images on the left hand side are super resolution radar data at 0.5 degrees and 250 m resolution while the images on the right side are legacy 8-bit storm relative velocity imagery at 1 degree and 250 m resolution.

The improvement in the super resolution data is very apparent. The super resolution reflectivity imagery is very detailed when compared to the legacy reflectivity imagery while the improvement in storm relative velocity imagery is not nearly as great since the storm relative velocity resolution was only improved from 1 degree and 250 m resolution to 0.5 degrees and 250 m resolution. In the larger image (click on the image below to open a larger version), you can see a few pixels of 65 dBZ in the super resolution reflectivity panel (as noted by the white pixels). In the legacy 8 bit reflectivity product, there are no pixels of 65 dBZ as these larger values likely were reduced or averaged out in the larger bin.

click to enlarge          



The image below is similar to the image above with a four-panel radar image from the KRAX WSR-88D valid 2008 UTC on August 27, 2008 just before the Silk Hope Tornado touchdown. The images on the top row are KRAX base reflectivity at 0.5 degrees while the images in the bottom row are storm relative velocity at 0.5 degrees. The reflectivity images on the left hand side are super resolution radar data at 0.5 degrees and 250 m resolution while the images on the right side are legacy 8-bit reflectivity imagery at 1 degree and 1 km resolution. The storm relative velocity images on the left hand side are super resolution radar data at 0.5 degrees and 250 m resolution while the images on the right side are legacy 8-bit storm relative velocity imagery at 1 degree and 250 m resolution.

As in the previous image, the improvement in the super resolution data is very apparent. In the larger image (click on the image below to open a larger version) you can see an inbound storm relative velocity value of 35 knots (as noted by the bright green pixel) and an outbound velocity value of 33 knots white pixels). In the legacy 8 bit reflectivity product there are no pixels of 65 dBZ as these larger values likely were reduced or averaged out in the larger bin.

click to enlarge          

The image below is also valid from 2008 UTC on August 27, 2008 and it compares the storm relative velocity super resolution data at 0.5 degrees and 250 m with the storm relative velocity legacy 8-bit resolution at 1 degree and 250 m. The added detail can be seen and the small differences in the max inbound and outbound velocities between the data sets are noted in the image.

click to enlarge

The image below is similar to the image above with a four-panel radar image from the KRAX WSR-88D valid 0459 UTC on August 28, 2008 as the Fremont Tornado was touching down. The images on the top row are KRAX base reflectivity at 0.5 degrees while the images in the bottom row are storm relative velocity at 0.5 degrees. The reflectivity images on the left hand side are super resolution radar data at 0.5 degrees and 250 m resolution while the images on the right side are legacy 8-bit reflectivity imagery at 1 degree and 1 km resolution. The storm relative velocity images on the left hand side are super resolution radar data at 0.5 degrees and 250 m resolution while the images on the right side are legacy 8-bit storm relative velocity imagery at 1 degrees and 250 m resolution.

click to enlarge          

The image below is also valid from 0459 UTC on August 28, 2008 and it compares the storm relative velocity super resolution data at 0.5 degrees and 250 m with the storm relative velocity legacy 8-bit resolution at 1 degree and 250 m. The added detail can be seen and the small differences in the max inbound and outbound velocities between the data sets are noted in the image.

click to enlarge


Why Did So Few Tornado Warnings Verify

Nearly 20 Tornado Warnings were issued during the event and despite indications of rotation in the lower levels of many of these thunderstorms, only three tornadoes were confirmed in central North Carolina. There are likely many factors responsible for the limited number of confirmed tornadoes.

  • While the event was produced by the remnants of a tropical cyclone, the atmosphere contained less deep layer shear than recent significant tropical cyclone tornado events such as Hurricane Jeanne, Hurricane Ivan, Hurricane Frances, or Tropical Storm Cindy. The low and mid level flow was also weaker with Fay although the low level storm relative helicity values were more similar. Unlike some of these previous storms, Fay began to impact central North Carolina 5 to 6 days after its landfall on the Gulf Coast of Florida.



    In the 1834 UTC Mesoscale Discussion SPC noted that the 30 kts of effective bulk shear was modest for supercells but thunderstorms that interact with the boundaries across North Carolina possess a greater potential for low level rotation and possible tornadoes.

  • There were 3 to 4 dozen reports of rotating wall clouds, funnel clouds, or tornado touch downs across central North Carolina. Several reports were from reliable sources that cited specific tornado touch downs or real time reports of tornadoes moving across roads or farms. Many of these reports were from storms that moved across Johnston, Wilson, Nash, and Edgecombe Counties. A large number of reports such as these in an environment potentially conducive for tornadoes. These factors made it very difficult for the radar operator to be conservative and hold of issuing tornado warnings. The storm survey team could not locate any damage or indications of tornadoes despite the numerous reports of tornadoes on the ground, moving across farm land or crossing roads.

  • While nearly every storm that had a tornado warning issued on it possessed rotation or a mesocyclone most of these features can be best described as "weak shear'" or "minimal mesocyclones" on the mesocyclone recognition chart.

  • The new super resolution data may have played a very minor role by making the reflectivity structure, including features such as hooks, much more apparent. The super resolution reflectivity imagery is much more detailed when compared to the legacy reflectivity imagery. The improvement in storm relative velocity imagery is not nearly as great since the velocity resolution was only improved from 1 degree and 250 m resolution to 0.5 degrees and 250 m resolution. Comparisons of the storm relative velocity super resolution data at 0.5 degrees and 250 m with the storm relative velocity legacy 8-bit resolution at 1 degree and 250 m showed that the change in velocity values was rather small, mainly on the order of a few kts.

  • The three confirmed tornadoes all occurred in the immediate vicinity of a pronounced surface boundary.



Precipitation Estimates and the National Mosaic and QPE (NMQ)

An accurate analysis of the amount of precipitation that has fallen from a tropical cyclone is extremely important to forecasters. The amount of precipitation is a critical factor in determining the likelihood of flash flooding and eventual river flooding. Because flooding is typically a result of significant precipitation over an area such as a river or stream basin, relying on point observations of precipitation (rain gauges) only provides a portion of the needed information. The estimation of the amount of precipitation that has fallen is often referred to as Quantitative Precipitation Estimation or QPE. Studies have shown that algorithms which combine sensor inputs such as radar, gauge, and satellite yield more accurate precipitation estimates than those which rely on a single sensor. Recent advances in QPE integrates radar data with other data sources such as rain gauges and satellite information in a process called Muti-sensor Precipitation Estimation or MPE.

In February 2000, the National Severe Storms Laboratory (NSSL), National Sea Grant (NSG) College Program, University of Oklahoma, North Carolina State University (NCSU), and the North and South Carolina Sea Grant programs established a joint project, centered in North Carolina areas affected by Hurricane Floyd. The original collaborators were later joined by the National Weather Service Office of Hydrologic Development and the National Environmental Satellite, Data and Information Service (NESDIS). The primary demonstration area was the Tar-Pamlico River basin. This project, called CI-FLOW, has established a research and demonstration program for the evaluation and testing of new technologies and techniques to produce accurate and timely identification of inland and coastal floods and flash floods. CI-FLOW leverages scientific expertise from three ongoing NOAA research activities namely the NSSL NMQ/Q2, the NWS/OHD MPE/EMPE program, and the NESDIS HydroEstimator program.

The National Mosaic and QPE (NMQ) Web Page provides real time evaluation and display of experimental techniques and applications used for high resolution 3D mosaics of radar reflectivity data and QPE. The NMQ serves as a test bed for research, development, and evaluation of data and methods for the monitoring and warnings of floods and flash floods and in support of comprehensive hydrology and ecosystem modeling.

There are 4 major components of the NMQ system:
  • Data ingest
  • Products
  • Analysis tools
  • Verification

    The NMQ system ingests data from a number of sensors and products from various sources:
  • 128 WSR-88D radars (5 min)
  • Gauge data set that includes 5500 gauges from many different networks (hourly)
  • Satellite IR images (15 min)
  • Rapid Update Cycle 20 km resolution (RUC) model analysis variables (hourly)
  • NWS Hydro Estimator (satellite-based) precipitation (hourly)
  • NWS Stage IV precipitation (1, 6, and 24 hour)

    Data and processing flow:
  • Ingest all data (they arrive at different rates)
  • Quality control radar data
  • Derive vertical profile of reflectivity from each radar
  • Analyze radar data to 8 tiles and stitch the tiles together to form the CONUS 3D grid (1 km x 1 km x 31 levels
  • Derive hybrid scan reflectivity and other products
  • Produce experimental Q2 products
  • Use gauges to validate Q2 and other products



    Comparison of QPE Products

    QPE products for the same 24 hour period ending at 12 UTC on 08/28/2008 are shown below. The various QPE products provide a slightly different analysis but the most detailed and most representative analysis is produced by the Local gauge corrected radar QPE product otherwise known as the Q2RAD_HSR_GC.



    Q2GAUGE

    The Q2GAUGE is gauge-only precipitation analysis that uses a local gauge correction scheme based on an inverse distance weighted (IDW) mean scheme. The two parameters in the IDW scheme, exponent and radius of influence, are determined through a cross-validation procedure. The interpolated gauge bias field is used to create the Q2GAUGE product.

    Precipitation estimate



    HYDRO_EST

    The HYDRO_EST is a NWS operational satellite precipitation hydroestimator product. Precipitation rates are primarily based on the cloud top temperature obtained from GOES 12 and GOES 10 (10.7 micron). Instantaneous, 1 hour, 3 hour, 6 hour, and 24 hour precipitation estimates are available. Numerous other factors, including the cloud-top geometry, the available atmospheric moisture, stability parameters, radar, and local topography, are used to further adjust the rain rate.

    Precipitation estimate



    STAGE4

    STAGE4 data is based on operational 24 hour NWS Stage IV hourly and 24-hr precipitation analyses from NCEP. Stage IV is a final stage term used to describe nationwide mosaicing of manually-edited, regional MPE products produced by each of River Forecast Center (RFC) on an hourly basis. Stage IV is a readily available operational product (real time Stage IV data for North Carolina).

    Precipitation estimate



    Q2RAD_HSR

    The Q2RAD_HSR is a radar-based precipitation rate and accumulation product with data available over various time periods. The rate is derived from the hybrid scan reflectivity (HSR) using convective, stratiform and tropical Z-R relations.

    Precipitation estimate



    Q2RAD_HSR_GC

    The Q2RAD_HSR_GC is a local gauge corrected radar QPE field. The local gauge correction is applied onto the one hour Q2RAD_HSR precipitation field. It runs hourly and uses hourly rain gauge observations from the HADS (Hydrometeorological Automated Data System) data sets at NCEP. In the local gauge correction scheme, radar-gauge biases are calculated at each gauge site and then interpolated onto the NMQ grid using an inverse distance weighted (IDW) mean scheme. The two parameters in the IDW scheme, exponent and radius of influence, are determined through a cross-validation procedure. The interpolated radar-gauge bias field is applied back to the Q2RAD_HSR one hour precipitation field and a local gauge bias corrected one hour precipitation field is obtained. Longer-term accumulations are computed by aggregating the one hour local gauge corrected precipitation fields.

    Precipitation estimate




    KRAX Precipitation Estimate

    The krax precipitation estimate is a radar-based precipitation accumulation produced at the RPG. The precipitation estimate is derived from the Hybrid Scan Reflectivity (HSR) product from the Enhanced Precipitation Preprocessing (EPRE) algorithm using a specific Z-R relationship. The image below is the product from the KRAX RDA displayed in AWIPS (click on the image to enlarge.)

    Precipitation estimate from KRAX


    Comparison of the local gauge corrected radar QPE product with the Stage IV QPE product

    The local gauge corrected radar QPE product or Q2RAD_HSR_GC generally provided the most accurate and detailed analysis of the precipitation associated with the remnants of tropical storm Fay. The comparison below shows the local gauge corrected radar QPE product from the NMQ web site with the Stage IV QPE product from the NWS AHPS web site for the 24 hour period ending at 12 UTC 08/28/08 (click on the image to enlarge).

    Comparison of the Local gauge corrected radar QPE product with the Stage IV QPE product - click to enlarge


    • Mesoscale Data

    Analyzed surface pressure and wind barbs from SPC at 18 UTC on Wednesday, August 27, 2008
    The circulation center associated with the remnants of Fay can be seen across far eastern Tennessee. A surface trough extends eastward across the central Appalachians, the central Foothills, into the Piedmont and Coastal Plain of North Carolina with southeast winds south of the trough and east to northeast winds north of the trough.

    SPC Analysis at 18 UTC on Wednesday, August 27, 2008



    Analyzed surface theta-e (green) and theta-e advection (purple) from SPC at 18 UTC on Wednesday, August 27, 2008
    The surface trough can clearly be seen extending west to east across central North Carolina. The cooler and more stable air mass can be seen in the theta-e gradient along and just north of the trough. Max theta-e values around 360K can be seen in the far southern Sandhills.

    SPC Analysis at 18 UTC on Wednesday, August 27, 2008



    850 MB heights, temperatures (red/blue), dew points (green), and wind barbs (black) from SPC at 18 UTC on Wednesday, August 27, 2008
    The broad circulation with modest velocities associated with the remnants of Fay can be seen across the southern Appalachians. A southeast to southwesterly flow at around 20 to 25 kts was present across central North Carolina. Much stronger flow was noted across northwestern Virginia and West Virginia with a southeasterly flow at 35 to 40 kts. Dew points were in the 15 to 16 degree range.

    SPC Analysis at 18 UTC on Wednesday, August 27, 2008



    Analyzed mixed layer convective available potential energy (MLCAPE) (red) and mixed layer based convective inhibition (MLCIN) (blue lines - shaded) from SPC at 18 UTC on Wednesday, August 27, 2008
    The surface trough can clearly be seen extending west to east across central North Carolina with MLAPE values greater then 500 J/kg noted south of the trough along with indications of some modest convective inhibition (MLCIN) north of the boundary.

    SPC Analysis at 18 UTC on Wednesday, August 27, 2008



    0-1 km Storm Relative Helicity (blue) and storm motion (brown) from SPC at 18 UTC on Wednesday, August 27, 2008 Note that the 0-1 km SRH values range between 150 and 250 m²/s² across the northern portions of the Piedmont. The SRH is a measure of the potential for cyclonic updraft rotation in right-moving supercells. Studies have shown that larger values of 0-1 km SRH, greater than 100 m²/s², suggests an increased threat of tornadoes and that very large values of 0-1 km SRH (perhaps greater than 200 to 300 m²/s²) are indicative of significant tornado potential.

    SPC Analysis at 18 UTC on Wednesday, August 27, 2008



    0-3 km Storm Relative Helicity (blue) and storm motion (brown) from SPC at 18 UTC on Wednesday, August 27, 2008
    The maximum 0-3 km SRH values were centered over western Virginia in the area near the greatest 850 mb flow. An axis of fairly impressive SRH values extend southeast into central North Carolina. The SRH is a measure of the potential for cyclonic updraft rotation in right-moving supercells. Larger values of 0-3 km SRH (greater than 100 m²/s²) suggest an increased threat of supercells and tornadoes. Some studies suggest that the 0-3 km SRH is a better indicator of storm rotation, which is related to tornadoes, but not directly the potential for tornadoes themselves.

    SPC Analysis at 18 UTC on Wednesday, August 27, 2008



    Analyzed Lifting Condensation Level (red, blue, and green) from SPC at 18 UTC on Wednesday, August 27, 2008
    The LCL (Lifting Condensation Level) is the level at which a parcel becomes saturated. It is a reasonable estimate of cloud base height when parcels experience forced ascent. Note that much of central NC has LCL values less than 1000 meters. Research has shown that LCL heights generally are low (< 1000 meters) for most significant tornado cases.

    SPC Analysis at 18 UTC on Wednesday, August 27, 2008



    NWS Composite Reflectivity Imagery from 2130 UTC on Tuesday, July 8, 2008 (530 PM EDT).
    The composite reflectivity imagery is from the approximate time in which the analysis imagery above is valid.

    Composite Reflectivity Imagery from 2130 UTC on Tuesday, July 8, 2008



    Archived Text Data from the Event

    Select the desired product along with the date and click "Get Archive Data."
    Date and time should be selected based on issuance time in GMT (Greenwich Mean Time which equals EDT time + 4 hours).

    Product ID information for the most frequently used products...

    RDUAFDRAH - Area Forecast Discussion
    RDUZFPRAH - Zone Forecast Products
    RDUAFMRAH - Area Forecast Matrices
    RDUPFMRAH - Point Forecast Matrices
    RDUHWORAH - Hazardous Weather Outlook
    RDUNOWRAH - Short Term Forecast
    RDUSPSRAH - Special Weather Statement
    RDULSRRAH - Local Storm Reports (reports of severe weather)
    RDUSVRRAH - Severe Thunderstorm Warning
    RDUSVSRAH - Severe Weather Statement
    RDUTORRAH - Tornado Warning


     from 



    Lessons Learned

    Every confirmed tornado was in very close proximity to the surface trough that extended eastward across the central Appalachians, the central Foothills, into the Piedmont and Coastal Plain of North Carolina.

    The local gauge corrected radar QPE product or Q2RAD_HSR_GC product (select "Gage Corrected QPE [HSR]" from the drop down menu) available at the National Mosaic and QPE (NMQ) Web Page appears to provide a very accurate and detailed analysis of precipitation amounts. It can provide additional detail to the commonly used NWS Stage IV precipitation product. This is shown with a comparison of the Local gauge corrected radar QPE product from the NMQ web site with the Stage IV QPE product from the NWS AHPS web site for the 24 hour period ending at 12 UTC 08/28/08.

    Mesoscale discussions produced by the Storm Prediction Center again provided valuable insight into the location and tendency for severe weather on the mesoscale or county scale. Forecasters need to insure that the information contained in these products reaches the forecasters who are issuing warnings.

    Adjusting schedules and providing the resources necessary to conduct two storm survey teams was invaluable.

    The improved resolution associated with super resolution radar data is very apparent. The super resolution reflectivity imagery is much more detailed when compared to the legacy reflectivity imagery. Storm structures were more easily identified. The improvement in storm relative velocity imagery is not nearly as great since the velocity resolution was only improved from 1 degree and 250 m resolution to 0.5 degrees and 250 m resolution.

    Further investigation into the utility of the NOAA GPS-MET data during this event may demonstrate the utility of this often underutilized data source. For example, GPS-MET data from locations across central North Carolina showed the rapid increase in precipitable water values on August 26th. The GPS-MET network Provides unattended, autonomous, frequent, and accurate observations at very low cost that are unaffected by weather conditions or time-of-day.



    Acknowledgements

    Many of the images and graphics used in this review were provided by parties outside of WFO RAH. The surface analysis graphic and regional Fay precipitation analysis were obtained from the Hydrometeorological Prediction Center. The upper air analysis images were obtained from the University of Wyoming. GOES satellite data was obtained from National Environmental Satellite, Data, and Information Service. SPC meso-analysis graphics provided by the Storm Prediction Center. Base maps for the tornado tracks were provided by Google Maps and Google Earth - Google Earth map imagery used under license. The various QPE products were provided by the National Mosaic and QPE (NMQ) Web Page. Photos courtesy of Ashley Lemons as well as Russ Henes, Jeff Orrock, and Darin Figurskey of the National Weather Service in Raleigh.



    Case study team -
    Phillip Badgett
    Darin Figurskey
    Gail Hartfield
    Russ Henes
    Jeff Orrock
    Barrett Smith
    Jonathan Blaes

    For questions regarding the web site, please contact Jonathan Blaes.


  • NWS Disclaimer.