Event Summary
     National Weather Service, Raleigh NC

December 26, 2004 Winter Storm
Updated 2005/01/12

Event Headlines

...”The Day after Christmas Ice and Snow Storm” in the Carolinas and Virginia, closed portions of Interstate 95, stranded motorists and airline passengers, and left 20,000 households without electrical power...
...The storm was characterized by significant differences in precipitation amounts and wintry precipitation types over short distances...
...Portions of Central and Eastern North Carolina received significant accumulations of snow, sleet, and freezing rain...

Event Overview

A winter storm brought snow, sleet, and freezing rain to central and eastern North Carolina during the morning hours of December 26th, 2004. The event was short lived as much of the precipitation had moved offshore by late afternoon. The heaviest precipitation, which consisted of a sleet and snow mixture, fell in close proximity to the Interstate 95 corridor where amounts reaching 3 inches occurred across sections of Hoke and Scotland counties in southern North Carolina. A little further to the northeast, the precipitation predominantly fell as snow, with a little sleet mixed in at times. The snow and sleet accumulated up to 6 or 8 inches across much of Wilson, Edgecombe, and Nash counties. Snow amounts in excess of 8 inches, ranging up to 11 inches, were reported across the state’s northeastern counties, including Gates, Hertford, and Northampton. Meanwhile significant amounts of ice from freezing rain fell in a narrow band just southeast of the snow and sleet regions across portions of southeastern North Carolina. A coating of a quarter inch of ice occurred in southern Cumberland, central Sampson and northern Duplin counties, while Robeson and Columbus counties experienced an ice accrual up to a half inch.

Synoptic Overview

This snow, sleet, and freezing rain event was associated with a so-called Miller “A” pattern of cyclogenesis. This pattern is characterized by a relatively simple mean sea level pressure pattern where one relatively well organized surface low tracks northeastward with a surface high pressure center located to the north. With cold air in place over the Carolina’s and Virginia, the surface low moved northeastward across the Gulf of Mexico late on December 25th. The storm system then moved across Florida, paralleling the coast of the Southeastern U.S. on the 26th.

Precipitation reached southern North Carolina shortly after dark on Christmas evening. The precipitation shield then spread northward and northeastward, covering much of central and eastern North Carolina by daybreak on the 26th. The resulting precipitation types and amounts were a function of many meteorological factors. The surface low tracked well offshore limiting the inland extent of the precipitation shield, resulting in no snow or sleet in the more climatologically favored areas of the Piedmont such as Roxboro and Triad areas. The intensification of the coastal low once it passed Cape Hatteras, contributed to the larger snowfall amounts in far northeastern North Carolina.

The surface features were again consistent with the so-called so-called Miller “A” pattern of cyclogenesis. Relative to this storm’s individual characteristics, it is important to note that the surface low’s isobars showed a rather elongated shape and relatively high central pressures as it quickly passed by North Carolina’s southern coast. Meanwhile the surface high to the north lacked a well defined ridge of high pressure extending into the Carolinas. This configuration is a signal to forecasters that there is a limited inflow of warm air into the low and cold air from the high.

The meteorological features seen at 700 AM on December 26th at the 850 MB level (typically located around 5,000 ft level above ground) showed little advection of cold and warm air into central and eastern North Carolina. When there is little or no temperature advection in the atmosphere from ground level through the 10,000 ft level or so, wintry precipitation types can be subject to frequent changes (see the meteogram from Fayetteville and note the changes to the precipitation type in the "WX" row). These changes are often associated initially with evaporative cooling as precipitation falls into the dry sub cloud layer and then later by the precipitation rates. Under these conditions the change of temperature in the vertical from the ground through about 10,000 ft is very gradual and often hovering near freezing through a deep layer. Indeed with the “Day After Christmas Storm”, there were frequent observations of changes between snow and sleet and a snow/sleet mixture. Moreover, as the precipitation rates increased, the barely above freezing melting layer extending through a relatively deep layer that existed at Fayetteville at 600 AM on the 26th, was erased by additional cooling from melting. The additional cooling allowed the sleet or sleet/snow mixture to change to snow. Once the precipitation rates diminished and cooling from melting was lost, the snow would mix with or change again to sleet.

In addition to the configuration and evolution of the surface and low level meteorological features; prominent features in the upper levels of the atmosphere were key players in the evolution of this winter event and its impact on central North Carolina. A deep trough of low pressure, a strong vorticity impulse moving around the base of the upper level trough, and a dual upper level jet structure were all contributing energy and moisture to the evolving low level low pressure system.

Prior to reaching North Carolina, the upper level trough and its closely associated vorticity impulse had combined with plentiful moisture and very cold air to produce unprecedented snowfall in South Texas. It was the energy associated with the upper level vorticity impulse that lead to the development of the surface low in the Gulf of Mexico late on the Christmas Eve that eventually produced the wintry precipitation across central and eastern North Carolina on the 26th. The strong south to southwest flow in the 500 – 300 MB layer provided abundant moisture from the Pacific. The position of the dual upper level jets produced strong divergence aloft that in turn produced low level convergence and the development of clouds and precipitation.

Snow / Sleet Accumulation Map

The corridor of heaviest snow and sleet accumulation fell near and just east of the I-95 corridor with snow and sleet amounts ranging from 3 inches in southern North Carolina amounts in excess of 8 inches across northeastern North Carolina.

Freezing Rain / Glaze Accumulation Map

Significant freezing rain accrual was confined to southern and central portions of the costal plain and eastern portions of the Sandhills. A coating to a quarter of an inch of glaze was reported over sections of Sampson and Wayne counties. Immediately south of this region, ice accrual of a half inch was reported in Robeson and Columbus counties.

High Resolution Visible Satellite Imagery Depicting Snow Cover

This image captured by the Terra MODIS instrument on Monday, December 27, 2004, shows the remaining snow cover across portions of southern North Carolina, through Virginia, and across the Chesapeake Bay into Maryland.

Note how narrow the snow/sleet band is across North Carolina. Accurately forecasting the location of this band was one of the biggest forecasting challenges associated with this event.

High Resolution Visible Satellite Imagery Depicting Snow Cover - Click to enlarge
(Click on the image to enlarge.)

Surface Analysis

Surface analysis from 15Z Sunday December 26, 2004 depicts the surface low centered east of Wilmington, North Carolina.

A Java Loop of Surface Analysis imagery from 12Z Saturday, December 25, 2004 through 12Z Monday, December 27, 2004 shows the evolution of the Miller Type "A" surface low.

Surface analysis from 12Z Sunday December 26, 2004

Satellite Imagery

Satellite was critical in monitoring the track and intensity of the strong vorticity impulse that initiated cyclogenesis, tracking the flow of moisture into the system, and the intensity of the surface low. The satellite image below shows the strong vorticity impulse just south of the Texas/Louisiana coast and a large area of higher cloud tops across the eastern Gulf of Mexico.

A Java Loop IR Satellite Imagery from 1215Z on Friday, December 24 through 2315Z on Sunday, December 26, 2004 is available.

Satellite Imagery from 1705Z on Saturday, December 25, 2004

Partial Thickness, Surface Wet Bulb Temperature and Surface Weather

The thickness of a layer of air is proportional to the layer's mean temperature. The warmer the layer, the greater it's thickness. The layer of air bounded by a pressure of 1000 - 850 MB is used by forecasters to monitor the average temperature in the lower level of the atmosphere (roughly from the surface to 5,000 ft). The 850 - 700 MB layer is closely followed by forecasters to monitor the average temperature in the layer of air (roughly from 5,000 to 10,000 ft) where elevated above freezing temperatures may exist. NWS forecasters at Raleigh have for decades now, correlated the thickness of these layers as observed from weather balloon launches to the observed wintry precipitation types. The technique has proven to be helpful in anticipating the frequent precipitation type changes often associated with winter storm events in central North Carolina. The Java loop shows the changes in the thickness levels and their associated wintry precipitation types.

The image below depicts the partial thickness values for 1000 - 850 MB and 850 - 700 MB along with surface wet bulb temperatures and the observed weather at 14Z on Sunday December 26, 2004.

A Java Loop of Partial Thickness, Surface Wet bulb and Weather imagery from 02Z through 22Z Sunday December 26, 2004 is available.

Partial Thickness, Surface Wet bulb and Weather Imagery from 14Z on Sunday December 26, 2004

Raleigh - KRAX Radar Imagery

Raleigh, KRAX WSR-88D base reflectivity imagery from 1331Z on Sunday December 26, 2004 is shown below. Note the areas of enhanced reflectivity northwest of Fayetteville and southwest of Roanoke Rapids.

Areas of enhanced reflectivity were critical in the distribution of the precipitation types. The numerous changes between snow and sleet were associated with the changes in precipitation intensities. As the precipitation rates increased, the slightly above freezing layer was erased by additional cooling from melting, allowing sleet to change to snow.

A Java Loop of KRAX Radar imagery from 02Z Sunday December 26, 2004 through 22Z Sunday December 26, 2004 is available.

Raleigh, KRAX Radar imagery from 1331Z Sunday December 26, 2004

Regional Radar Imagery

Regional composite reflectivity radar imagery from 13Z on Sunday December 26, 2004 is shown below. Note the sharp gradient on the back edge of the precipitation shield and the bands of enhanced reflectivity oriented from southwest to northeast.

A Java Loop of Regional Radar imagery from 02Z Sunday December 26, 2004 through 22Z Sunday December 26, 2004 is available.

Regional Radar from 13Z Sunday December 26, 2004

Using PV Thinking to Diagnose Model Differences in Cyclone Evolution in the 12 UTC operational model run from 25 December 2004

Significant differences were seen in the solution of the NCEP Eta and GFS model runs in the track of the surface cyclone and the lower-tropospheric wind field, which directly impacted the transport of moisture into the region and the total QPF in the Carolinas and Virginia. These differences can be at least somewhat diagnosed by examining the differences in convective precipitation in the models and their impact on the lower-tropospheric PV distribution, as latent heating produced by the model convective schemes can lead to the development of PV maxima in the lower-troposphere and impact the track of the surface cyclone by generating sea-level pressure falls below the level of maximum heating.

These differences are already seen in the 12-h forecast valid at 00 UTC 26 December. The GFS model shows a band of convective precipitation (red contours) occurring northeast of the surface cyclone across central Florida and into the offshore Atlantic waters well south of the Carolinas. A lower-tropospheric PV maximum (color shading) is already seen offshore of the east coast of Florida beneath this precipitation region due to the strong diabatic heating underway in this region. Strong troughing in the sea-level isobars is also seen in this region, caused by pressure falls at least partially due to latent heating. At this time the Eta model solution at this time is quite different, with a band of stronger convective precipitation occurring farther west offshore of Georgia through the Carolinas.

By 12 UTC 26 December, the surface cyclone in the GFS forecast is located east-southeast of Hatteras with a lower-tropospheric PV maximum associated with the band of convective precipitation both ahead of and behind the surface low. The 850-mb winds offshore of the Carolinas are northeasterly, consistent with a cyclonic flow associated with the PV maximum. The orientation of these winds would tend to limit moisture flux and warm air advection into the Carolinas and Virginia in this situation.

The Eta forecast at this same time is quite different, again showing a large area of convective precipitation closer to the coast and a PV maximum just offshore of the Carolinas. At this time the surface cyclone is in the process of “jumping” from its position over Florida to the latent heating maximum offshore of the Carolinas, much closer to the coast in the Eta model forecast relative to the GFS. The 850-mb wind field is consistent with the influence of the PV maximum closer to the coast, as the cyclonic flow associated with the PV max results in a more easterly component to the flow to its north, implying stronger warm advection and moisture transport into the Carolinas and Virginia.

These differences related to convective precipitation differences in the model forecasts are consistent with the different scenarios of cyclone evolution, precipitation coverage and thermal advection, with the Eta model forecasting more inland precipitation, but with a warmer thermal structure. Using the RUC analysis as verification at 12 UTC 26 December, neither model was entirely correct with its scenario. The RUC shows the surface low center closer to the Eta forecasted position, but with an extended PV maximum more similar to the GFS solution in character, if not location. However, the Eta forecast of 850-mb winds seems to verify better with the RUC analysis showing a more onshore component to the flow offshore of the Carolinas.

The image below contains the RUC analysis as verification at 12Z on December 26 displaying 900-700 MB PV, Mean Sea Level Pressure, Convective Precipitation, Total Precipitation and 850 MB wind.

RUC analysis imagery used as verification at 12Z on December 26

Final Thoughts

Satellite imagery early on Saturday afternoon was used to determine that the system was stronger than the models were indicating, giving credence to the higher QPF forecast by the eta.

Early recognition of a single low in a “Miller A” fashion with cold air in place and high pressure to the north was used to correctly anticipate a relatively narrow band of freezing rain.

Forecasters recognized that the eta’s above freezing warm nose aloft was overdone given the evolution of the surface low and the 850 low. This resulted in limited WAA, which allowed more frozen precipitation to fall.

Rather than using a 10:1 ratio, considerable emphasis was correctly placed upon using lower ratios associated with wet snow and sleet/snow combinations. A 3:1 ratio was used for those areas when mostly sleet was expected. Ratios of 6-8:1 were used for areas where a snow/sleet mixture was expected.

Rather than using the p-type tools as a black box, physical processes were inferred from the partials correctly leading to the conclusion that this event would be characterized by snow/sleet combination. The 06Z run of the eta on 12/26 showed the snow/sleet combination for RWI and also indicated at least the potential for the same at FAY with a narrow band of freezing rain adjacent to and just southeast of the snow/sleet area.

Areas of enhanced reflectivity were critical in the distribution of the precipitation types. The numerous changes between snow and sleet were associated with the changes in precipitation intensities. As the precipitation rates increased the small above freezing melting layer in the 925 – 800 MB layer was eliminated, allowing sleet to change to snow.

Relative to a sound forecast process, it is essential that forecasters closely observe and monitor real time data well upstream. The recognition of the potential development of a rare snow event in coastal Texas provided much insight and an early awareness for a potential winter event in North Carolina. Our forecasters were tracking the early stages of the winter weather system well in advance of its subsequent movement into the region.

NWS forecasters do not relay upon any single tool to predict precipitation types, precipitation amounts, and geographical distributions of precipitation types. Rather there is an established winter storm forecast process (subject to modifications as the science advances) consisting of key components. All of the key components have been found to be well interconnected. By evaluating these key components, forecasters are able to connect the dots to visualize conceptual winter storm models based upon well documented previous patterns and by understanding the physical processes which account for the evolution of the various storm patterns. The forecaster’s conceptual models are not solely qualitatively. They are substantiated by a large ongoing empirical data base consisting of observations of precipitation types and amounts, state wide precipitation type distributions, and objective values of key winter storm predictors. Since winter storms in this region are so often marked by mixed precipitation types and changeovers, special emphasis has been devoted by the NWS Forecast Office in Raleigh and its collaborative partners to devise local forecast techniques to more accurately forecast wintry precipitation changes and amounts (see the National Weather Service - NCSU Collaborative Research & Training Web Site, Case Studies and Event Summaries Web Page, and the Introduction to Forecasting Predominate Precipitation Type Trends Web Page for more information).

Case Study Team

Michael Strickler
Michael Brennan
Scott Sharp
Phil Badgett
Douglas Schneider
Gail Hartfield
Kermit Keeter
Jonathan Blaes

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

  • NWS Disclaimer.