April 16th, 2011 Tornado Outbreak

Thirty confirmed tornadoes occurred in North Carolina on 16 April 2011, the greatest one-day total for North Carolina on record.

A total of 24 individuals lost their lives in North Carolina.

Thirteen tornadoes were classified as strong ( EF-2 or greater ), with some hitting highly-populated areas.

Nine tornadoes occurred in the National Weather Service (NWS) Raleigh (RAH) County Warning Area (CWA). Among the nine, there were two EF-3 tornadoes, four EF-2 tornadoes and three EF-1 tornadoes.

There were 8 fatalities in the Raleigh Weather Forecast Office (WFO Raleigh) forecast area, the most in April on record in central North Carolina (official tornado database begins in 1950) and second only to the  28 March 1984 outbreak.  There were a total of 304 injuries reported in central North Carolina, although the actual number is likely higher. Two tornadoes, the Sanford-Raleigh Tornado and the Fayetteville-Smithfield Tornado, were responsible for more than 100 injuries each.

The NC Department of Crime Control & Public Safety reported that over 900 homes and businesses were destroyed and more than 6,400 were damaged across the state. Total structural damage in central North Carolina was estimated at $328,610,000. The Sanford-Raleigh Tornado produced $172,075,000 in damage alone and the Fayetteville-Smithfield Tornado produced $116,100,000 in damage.

One tornado came within 1.75 miles (2.8 km) of the WFO Raleigh office. The staff executed a phased evacuation from the operations area, which is on the third floor of a three story building. WFO Blacksburg, VA, backed up the Raleigh office for around 7 minutes. As the storm drew closer, the power went out, but there was no damage to the building or in the immediate area. The staff noted a strong odor of pine in the air after returning to the operations area.

The threat of severe weather on Saturday, 16 April 2011, was first mentioned during the morning of the Tuesday, 12 April Area Forecast Discussion (AFD) and the Hazardous Weather Outlook (HWO). On the morning of 13 April, forecasters became even more concerned as noted in the AFD with confidence increasing on 14 April. By the afternoon of 15 April, forecasters expressed a growing confidence of a significant tornado outbreak in the AFD. On Saturday morning, forecasters at WFO Raleigh and the Storm Prediction Center (SPC) issued statements and alerts that emphasized the unique and serious dangers presented by this particular weather pattern.

Dr. Victor Gensini, professor and researcher at Northern Illinois University, created this map showing the evolution of this event over a three day period ending on 16 April 2011. Shown are snapshots of the surface pattern on the 14th (left-most set of fronts), 15th (center), and 16th (right) of April 2011, along with the official storm reports as logged by the Storm Prediction Center (SPC).

Additional details of this event and reports can be found on  SPC's event page .

Here's a closer look at the four supercell thunderstorms that produced nine separate tornadoes across central North Carolina on 16 April 2011: Alamance-Person, Sanford-Raleigh, Fayetteville-Wilson, and Cumberland-Wayne.

Of these nine tornadoes,  two were EF-3, four were EF-2, and three were EF-1 intensity . These nine tornadoes were produced by four supercells, with each supercell producing at least two tornadoes. These four supercells were actually responsible for as many as 17 of the 30 total tornadoes across North Carolina on 16 April. The tornado tracks are shown in red in the map to the left, and the path of each of the four tornadic supercells is shown with the purple highlighting and identified by a number, in blue.

The thunderstorm that would eventually produce the Alamance County Tornado and the Person County Tornado initially developed at around 1015 AM about 20 miles northwest of Spartanburg, South Carolina. During the next hour, this thunderstorm along with other  storms across the western Carolinas would grow into an organized line of convection . As the low level instability increased, the  thunderstorms became more intense by around 1200 PM . The broken line of thunderstorms continued to intensify and move east with embedded thunderstorms developing into supercells with persistent rotating updrafts. Tornadoes were reported in Davie and Rowan Counties at around 1245 PM. 

The same complex cluster of thunderstorms that produced the tornadoes in Davie and Rowan Counties weakened slightly as they moved across Davidson and Guilford Counties between around 100 PM and 145 PM. The weak low-level rotation across eastern Guilford County at 150 PM quickly  strengthened as the storm moved into northwestern Alamance County at 204 PM  when a tornado touched down around 6 miles northwest of Graham. The tornado moved across northwestern Alamance County with the  storm relative velocity imagery becoming very impressive at 213 PM . The tornado remained on the ground for another few minutes before exiting Alamance County at around 216 PM and entering Caswell County. 

The tornado remained on the ground for another few miles in Caswell County before dissipating. The parent supercell weakened slightly but still maintained a broad area of rotation as it moved across eastern Caswell County. As the supercell moved into western Person County, the  rotation strengthened  while the  reflectivity structure improved . At around 240 PM, the Person County Tornado touched down with an impressive radar signature at 245 PM (see interactive swipe panel below). The tornado remained on the ground for 10 minutes and nearly 10 miles before the storm weakened and moved into Virginia.

The thunderstorm that would eventually produce the Sanford-Raleigh Tornado and the Roanoke Rapids Tornado initially developed to the southeast of Lancaster, South Carolina just after 115 PM. The thunderstorm intensified as it moved into North Carolina and approached Wadesboro. The storm then became a supercell as it developed a persistent rotating updraft as it moved into Moore County just after 230 PM. At 253 PM the storm produced a tornado in northeastern Moore County. The tornado strengthened as it moved across southeastern Sanford. The circulation in the storm remained intense as the storm moved across Chatham County into Wake County ( reflectivity image  &  storm relative velocity image ). The rotation within the storm weakened as it moved through Holly Springs with EF-0 to EF-1 damage observed from Holly Springs to southern Raleigh ( reflectivity image  &  storm relative velocity image ). The storm weakened a bit as it moved through downtown Raleigh before intensifying as it moved across northeastern parts of the city. A persistent path of EF-2 damage was noted in Raleigh from Stony Brook Drive ( reflectivity image  &  storm relative velocity image ) to Buffalo Road and Forestville Drive. The tornado began to weaken as it moved across far northeastern Wake County with damage becoming sporadic and isolated as it moved into Franklin County. 

The thunderstorm associated with the Sanford-Raleigh Tornado weakened somewhat as it moved across Franklin County but it remained strong as it moved through Warren County into southern Halifax County with no confirmed tornado damage. The supercell possessed an impressive reflectivity structure and broad rotation at 508 PM. The storm intensified markedly during the next few minutes and by the next volume scan the storm relative velocity imagery was indicating a new and intensifying area of rotation with an impressive reflectivity pattern. At 517 PM, both the  storm relative velocity imagery  and the  reflectivity imagery  had become even more alarming with a strong velocity couplet approaching Roanoke Rapids. Just a few minutes later, an EF-2 tornado touched down in the city. The tornado moved across the Roanoke River and dissipated in a few minutes after moving into Northampton County. 

The thunderstorm that would eventually grow into the Fayetteville-Smithfield, Micro, and Wilson tornadic supercell can be traced back to convection that developed around 25 miles west of Columbia, South Carolina at around 1230 PM. This thunderstorm was rather strong as it moved northeast across northern and northeastern South Carolina exhibiting periods of strong radar signatures with reports of damaging winds and hail. As the thunderstorm moved into far southern North Carolina near Cheraw at around 245 PM, it began to strengthen. The thunderstorm exhibited broad rotation across central Hoke County at 331 PM. The first tornado that this supercell would produce touched down near the Wayside and Johnson Mills communities at around 340 PM. The  reflectivity image  and  storm relative velocity image  were very impressive and well defined at 345 PM. The tornado remained on the ground and tracked northeast across northern Cumberland County maintaining a strong radar signature at 404 PM. As the tornado approached and crossed Interstate 95, the tornadic circulation became separated from the main updraft at 417 PM. The storm maintained an impressive structure as it moved south of interstate 95 in the 427 PM  reflectivity image  and  storm relative velocity image . As the tornado crossed interstate 95 for the second time and approached Smithfield, a secondary area of rotation developed near U.S. 70, around 3 to 4 miles southeast of the main circulation center. After the tornado went through Smithfield, it abruptly dissipated as a new tornado was forming to the southeast. 

The Fayetteville-Smithfield, Micro, and Wilson tornadic supercell developed a new updraft just as the first tornado was approaching Smithfield. The 450 PM radar imagery not only shows a new updraft developing to the northeast of Smithfield but also a new low level circulation developing to the east of the weakening circulation associated with the first tornado. This is a remarkable image that shows both the weakening circulation from the first tornado and an intensifying circulation associated with the second tornado (see interactive swipe panel below). The Micro Tornado was relatively short lived and the most impressive radar signature associated with the storm occurs at around 454 PM when the tornado touches down. The low level circulation weakened a few minutes later and the tornado dissipated after being on the ground for around 3 miles. 

Around 10 minutes after the Micro Tornado dissipates, the same supercell featured broad low level rotation and a modest reflectivity pattern. As the supercell approached Wilson, a  small but tightening low level circulation developed at 513 PM  with a  reflectivity appendage . The tornado touched down southwest of Wilson and then moved through western Wilson before dissipating just north of town. 

The thunderstorm that would eventually produce the Cumberland-Sampson and Wayne County Tornadic Supercell can be traced back to convection that developed around 25 miles southwest of Columbia, South Carolina at around 1230 PM. This thunderstorm intensified after it moved east of Columbia and approached Darlington and Florence. The thunderstorm intensified further into a supercell as it moved into northeastern South Carolina and produced an EF-1 tornado near Little Rock in Dillion County, South Carolina. After crossing into North Carolina, the same supercell produced an EF-1 tornado near Rowland in Robeson County and another EF-1 tornado near Barker Ten Mile. 

This supercell moved into southern Cumberland County and a tornado touched down just north of the Bladen-Cumberland County line at 433 PM ( reflectivity image  &  storm relative velocity image ). The tornado moved northeast crossing into southwestern Sampson County and weakened slightly. As the storm moved to the north of Clinton, the circulation strengthened while the reflectivity pattern remained impressive. The circulation subsequently weakened and the tornado dissipated at around 500 PM. 

The same supercell continued to move northeast across Sampson County and then produced a relatively brief tornado as it moved across northern Duplin County near Faison. As the supercell moved into and across southern Wayne County, the  storm relative velocity signature was generally convergent  while the  reflectivity pattern showed a developing appendage . Just before the storm was about to exit Wayne County, the  storm relative velocity  and especially the  reflectivity  pattern became more impressive on radar. An EF-0 tornado touched down near Parkstown and was on the ground for a mile before exiting Wayne County and moving into Greene County. 

As one might expect, this severe weather outbreak received a lot of media coverage, including before, during, and after the event.

This screenshot from WTVD-TV (Raleigh-Durham, NC) shows the kind of coverage most seen during the event -- real-time sky "cams" along with annotated radar imagery showing the circulation associated with the tornado, this one in eastern Wake county.

This video above shows coverage of the tornado that passed over Raleigh from another local station, WRAL-TV (Raleigh-Durham, NC).

Below is audio from the Central Carolina SKYWARN amateur radio transmissions on 16 April 2011. Two Central Carolina SKYWARN amateur radio operators were staffed at the NWS Raleigh during event and they collected 96 reports from 57 different spotters during the event. Many of the reports were first hand observations by trained spotters with additional reports relays of Public Safety transmissions heard on a scanner. Many of the initial reports of severe damage near Sanford were relayed by SKYWARN spotters listening to 911 traffic via scanners. (Duration: 3:34:28)

There have been many advancements made in the 10 years since this severe outbreak. These include improvements and innovations in radar systems and applications, satellite imagery, and even social media. Let's look at how these new tools could have helped us in April 2011.

Any weather radar scans multiple vertical levels to get a cross-section of an entire storm. The combination of different vertical levels in a single volume scan is a Volume Coverage Pattern (VCP). The default setting for WSR-88Ds is for the lowest scan to be 0.5 degrees above the horizon. This is designed to create a balance between keeping radar radiation from entering homes and businesses, but scan as close to the ground as possible. No matter how low the lowest scan is, the earth’s curvature means that the radar beam increases in height above ground level the farther a location is from the radar. A general rule of thumb is that for every mile a location is away from a radar, the lowest radar scan is about 100 feet off the ground. However, after additional study, some radars have been configured to add a scan below 0.5 degrees. In August of 2019, a 0.2 degree scan was added to KRAX, increasing radar coverage area at the 2,000 foot above site level by 74.5%. If this scan had been available in 2011, the NWS would have had even higher confidence that tornadic circulations were reaching the ground, instead of having to assume that the circulations noted at 0.5 degrees were reaching the ground.

Here's an example comparing the 0.2 degree radar scan (storm relative velocity, top L, and reflectivity, top R) with the 0.5 degree radar scan (storm relative velocity, bottom L, and reflectivity, bottom R) from the KRAX radar for an EF-0 tornado in Wayne county, NC, in 2019. The 0.2 degree scan shows the strong inbound and outbound velocity signatures right next to each other ("gate-to-gate"), strongly indicative of a tornado, while the 0.5 degree scan's velocity extremes are not gate-to-gate, making the presence of a tornado less certain.

SAILS was based on the concept that during certain weather regimes, while it is still important for multiple vertical levels to be scanned to get cross sections of storms, it may also be important to get more frequent low-level scans. SAILS allows the radar operator to add an extra base level scan in the middle of a volume scan from bottom to top. The radar starts at the base elevation, scans the bottom half of elevations, rescans the base elevation, then returns to the “middle” elevation and finishes the rest of the volume scan. MESO-SAILS allows the radar operator to add as many as 3 extra base level scans throughout a volume scan. While a typical volume scan from bottom to top may take 4.5 minutes, meaning that the base scan would only occur every 4.5 minutes, the use of MESO-SAILS 3 could allow for base scans to be scanned as quickly as every 75-80 seconds. While the use of SAILS increases the amount of time for an entire volume scan to be completed, this is balanced by the additional data gained at the surface. MESO-SAILS was added to radar software in 2014. MESO-SAILS works with the lowest level of the radar, so for the Raleigh radar, this is the 0.2 degree scan, not the 0.5 degree scan. If MESO-SAILS had been available in 2011, the NWS would have received low-level scans with a frequency 3 times greater than what was observed.

By default, the radar will scan all angles in a Volume Coverage Pattern (VCP), even if there is no precipitation in the highest levels of the atmosphere. When using the AVSET algorithm, the radar can sense that there are no precipitation returns above a particular threshold and height, terminate the volume scan, and return to the base elevation to begin a new scan. However, if there are storms close to the radar, all elevations will have precipitation returns, and AVSET will not activate. AVSET was added to radar software in the fall of 2011. If AVSET was available earlier in 2011, some of the volume scans could have been completed more quickly, allowing for additional low-level scans. Looking back at radar data, it is likely that AVSET would have allowed for some shorter volume scans at the beginning of the event, but by the storms reached Greensboro and Interstate 74, the storms were tall enough that AVSET would become inactive. AVSET might have been active during the Alamance and Person County tornadic supercell, but likely would have been inactive during the Sanford-Raleigh, Fayetteville-Smithfield-Wilson, and Cumberland-Wayne County tornadic supercells.

Legacy “single-polarization” radar only measured the horizontal properties of weather targets such as rain, snow, hail, and debris from a tornado. Dual-polarization radar allows both the horizontal AND vertical properties of weather targets to be measured. While there are several applications of dual-pol radar, the application most related to this event would be the use of correlation coefficient (CC) (values ranging from 0 to 1). If a single type of precipitation is occurring, the correlation in size and shape between all targets will be very similar, usually 0.99 and higher. If multiple types of precipitation are occurring, the correlation value will be lower, possibly between 0.9 and 0.95. With some (but not all) tornadoes, the CC will drop even lower if the radar is detecting debris that is lifted into the air, possibly as low as 0.5, providing near-real time confirmation of a tornado on the ground. The CC values are significantly lower with tornadoes. In November of 2012, KRAX was upgraded to dual-polarization. If dual-polarization data had been available in 2011, instead of simply noting high values of reflectivity paired with strong rotation couplets near tornadoes, reduced values of CC would have provided near-real time confirmation of tornadoes on the ground.

At left is a image showing 88D radar products for an EF-2 tornado in Alabama on 2 March 2012, with the tornado location circled. Focusing on the bottom right, you can see the area of lower CC values that indicates lofted debris (also called a debris ball) and confirms a destructive tornado on the ground. (Image courtesy of WFO Jackson, MS.)

The MRMS system integrates data streams from multiple radars to create composite products. The system creates products at a spatial resolution of 1 km and a temporal resolution of 2 minutes. While there are several applications of MRMS, the application most related to this event would be the use of MRMS Rotation Tracks. Even though a strength of this system is to integrate multiple radars, which may not have been applicable to all tornadoes from this event, the rotation tracks show both a location history and intensity trend of rotation. MRMS became operational in 2014, although Rotation Tracks existed during the 4/16/11 event and can be seen in the image below (more imagery can be found on the  original NWS Raleigh 4/16/11 case study page ). If MRMS had been available in 2011, the NWS would have access to the rotation tracks in near-real time during the event, highlighting the individual tornado tracks and enhancing their longevity.

The rotational track product for this event from 1200 to 2358 UTC on 16 April 2011 is shown on the left. The product shows numerous corridors of enhanced rotation across central and eastern North Carolina which are associated with the many supercell thunderstorms that moved across the region that afternoon and evening. The northeast storm motion is easy to see, as is the longevity of the rotating thunderstorms, with some of the tracks extending more than 100 miles (160 km). The two long track tornadoes (the Sanford-Raleigh Tornado and the Fayetteville-Smithfield Tornado) can be identified in this rotational track product (denoted by the black lines). 

ProbSevere combines several types of data to give a prediction as to whether a particular storm has reached severe limits. Some of the data that ProbSevere integrates include the Rapid Refresh (RAP) computer model, Geostationary Operational Environmental Satellite (GOES) imagery, and MRMS products. The first version of ProbSevere was created in 2014, simply giving a percentage whether a particular storm was severe or not. The second version of ProbSevere now gives individual probabilities for whether a storm may have severe wind, severe hail, or a tornado. If ProbSevere had been available in 2011, the algorithms may have given NWS forecasters additional confidence about whether an individual storm required a Severe Thunderstorm Warning or Tornado Warning.

At left is an example of ProbSevere output (colored contours) overlaid on GOES-East satellite imagery for a severe weather event on 5 May 2020. Based on several parameters, the algorithm highlights areas with a high probability of being severe. This kind of data is used by forecasters to increase confidence in the decision to issue a warning. (Image courtesy of CIMSS.)

The United States has had GOES in orbit since 1975. Compared to the previous GOES-East satellite, GOES-16 has 3 times as many spectral channels for sensing data (visible and infrared), 4 times greater spatial resolution (as fine as 0.5 km by 0.5 km), and 5 times greater temporal resolution (some mesoscale sectors can be scanned as frequently as every 60 seconds). The current GOES-East satellite, GOES-16, was launched in November 2016. If GOES-16 had been available in 2011, NWS forecasters would have had the capability to see satellite imagery update even more frequently than radar data. In addition, the additional data recorded by the satellite for upstream locations could have been fed into computer models, allowing for improved forecasts of the upcoming storm. Below is an example of such rapid-update imagery, from tornadic supercells in Iowa on 28 June 2017.

There have been tremendous advances in high resolution modeling over the last ten years, with several new datasets now in use for severe weather forecasting. One such model, one that became operational in 2014, is the High Resolution Rapid Refresh (HRRR). The HRRR runs hourly and is constantly reading in real-time observations, and even radar data, as inputs to the model. It creates forecasts of numerous fields that help forecasters identify potential severe weather threats, including forecast radar reflectivity. In 2011, the tightest grid spacing available in an hourly operational model was 13 km (from the Rapid Update Cycle [RUC]) model, while the HRRR is now run at a grid spacing of 3 km. One grid cell from the 2011 RUC model would now contain 16 HRRR grid cells. If the HRRR had been available in 2011, forecasters would have been able to view model data with much finer detail, even at an hourly scale, and would have had greater confidence in the storm types and how they would evolve during the day.

In addition, forecasters routinely examine output from the High Resolution Ensemble Forecast (HREF) system. The HREF is a set of 10 high-resolution models (including the HRRR), each set up in a slightly different manner than the next, that is available 4 times per day. By looking at a number of model solutions at the same time, forecasters can gain increased confidence in severe weather occurring in a specific area or at a specific time. While other ensemble forecast systems were in place in 2011, none of them updated as frequently, nor contained high resolution models. If the HREF had been available in 2011, forecasters would have increased confidence in how the storms would develop, not from just looking at a single high-resolution model like the HRRR, but from viewing multiple models at a similar high resolution.

Social media networks such as Facebook and Twitter can be a wealth of knowledge, especially in a rapidly developing situation. Members of the public will frequently share information about the weather that is currently impacting them. One of the biggest concerns about social media is that there are occasionally false reports or inaccurate information, something that any forecaster needs to be aware of. Any reports need to be vetted before being used. Facebook was approved for use in the NWS just a month after these tornadoes occurred, with Twitter approval coming in 2012. If Facebook and Twitter had been available during these storms, the NWS would have received confirmation about tornadoes on the ground more quickly and could have added that information into warning and follow-up statements that were issued, creating additional urgency to take shelter from the storms.

NWS Raleigh issued 31 tornado warnings during the 16 April 2011 event, with a probability of detection of 97%, false alarm rate of 48%, and an event average lead time of 19.5 minutes. 

This tornado outbreak is one that those of us at NWS Raleigh who worked it will never forget. Our hearts go out to all of those affected by these tragic storms.