Objective 1: Sharing the Forecast Process
On Sunday 27 February 2005,
a “Heavy Snow Warning” was posted by the National Weather
Service Forecast Office in Raleigh (NWS RAH) for six counties in the northwest Piedmont
of North Carolina, including the Greensboro, Winston-Salem, High Point Triad area.
None of the six counties, including the Triad area received heavy snow. Late Sunday
on 27 February into Monday morning on 28 February, a mixture of snow, sleet, and
freezing rain fell in the warning area. Most locations received little or no frozen
precipitation. The more notable accumulations of frozen precipitation in the warning
area included one half inch of sleet at the Greensboro Airport, and 1.5 inches of snow
and sleet at the Winston Salem Airport.
Location of Heavy Snow
Though there was no heavy snow in the warning area,
heavy snow did fall in nearby locations.
Four inches of snow accumulated about 20 miles north and west of the Winston
Salem Airport in nearby central Stokes and Yadkin counties. The snow amounts increased
through 6-8 inches in the northern foothills. Snowfall amounts then increased to a foot
of snow in Boone (northern mountains) which is located about 80 miles west of the
The Forecast Process
Area Forecast Discussions (AFD) issued on 27 February 2005 by NWS RAH revealed that
the key components of the station’s winter storm forecast process had been well
evaluated. For example, forecasters correctly anticipated that there would be a
narrow zone of mixed precipitation oriented from southwest to northeast in association
with a Miller Type “A” pattern of east coast cyclogenesis (see footnote below). The
forecasters were also correct in recognizing that heavy snow would occur just north
and west of the snow/rain line. The forecast discussions cited concern about a
suitable cold air source, noting that the colder and drier air in northern Virginia
may not be available to the warning area; however it was noted that the NWP models
were indicating a mid level closed circulation, implying strong dynamic cooling and
a high potential for banded precipitation. It was then correctly reasoned that the
combination of dynamic cooling and cooling from melting could well compensate for
the lack of significant cold air support from the surface high to the north.
NWP Model Errors
Using the Weather Event Simulator (WES ) to compare archived
NWP data from the 12Z 2/27/05 run to the real time data valid
at 12Z 2/28/05, some key model errors were found. Rather than
a closed circulation at 700 mb, as
forecast by the GFS and NAM, there was an open and
unimpressive short wave. Hence the models had overstated the potential for
banded precipitation and the associated cooling from melting that
would have occurred. The 500 mb short
wave trough lagged well west of
the open 700 mb short wave and the
closed circulation at 850 mb. The
model of choice for this event, the GFS, over forecast the 500 mb height
field by ~ 50 M, and hence overestimated the contribution from dynamic
cooling. Though these model errors were not especially large, the errors
were enough to limit dynamic cooling and cooling from melting that might
have compensated for the limited cold air support from the surface high.
Hence the significant snowfall developed and remained just north and west
of the Winston-Salem, Greensboro, and High Point Triad area.
Relative to the forecast soundings and partial thickness values, the GFS
was too cold in its 24 hour forecast of 850-700 mb thickness while its
forecast soundings had portrayed a near isothermal near freezing layer.
The NAM’s higher 850-700 mb thickness values and its forecast of a small melting
layer aloft was more accurate. While the NAM's thickness forecast had been
correct, it was for the wrong reason since it had incorrectly called for
the surface low to track inland. In time the NAM's surface low track came
more in line with the GFS which correctly called for the surface low to
move along the Carolina coast line.
Finally the impressive 992 mb surface low, located along the southeastern
coast of North Carolina at 12Z on 28 Feb 2005, appears to be largely
driven by the kinematics associated with the left forward quadrant of a
300 mb jet.
Forecast Process Summary
Given the NWP model guidance and the pattern of cyclogenesis, a forecast calling
for 2-4 inches of wet snow in the climatologically favored areas of central North
Carolina was reasonable. The limited cold air support from the surface high had
been well noted in the Area Forecast Discussion. NWS RAH forecasters showed an
extensive understanding of winter storm physical processes and their associated
meteorological patterns to foresee the potential for significant snowfall in the
northwest portion of Central North Carolina. For several model runs preceding the
event, the NWP guidance consistently pointed to more dynamic cooling and the
potential for additional cooling via melting from moderate to heavy banded precipitation.
Pinpointing the location of a snow/rain line to within ten's of miles
requires both an extensive understanding of winter weather science and precise
guidance from the NWP models. The area forecast discussions indicated the forecasters
level of understanding was substantial, while the NWP
guidance, though reasonable, was less than precise.
Product Head Lines
NWS’s policy regarding the use of product headlines has changed. The “Heavy Snow Warning”
has again returned as an option. Given the variability of winter weather, the frequent
occurrence of mixed and changeable precipitation in central North Carolina, and the users’
tendency to have their own definition for heavy snow, the “Heavy Snow Warning” banner
should be seldom used. High confidence in both p-type and snowfall total forecasts,
extensive collaborative agreement with other nearby NWS offices, and a prediction of a
snowfall total that well exceeds the minimum warning criteria are factors that should be
met before using the “Heavy Snow Warning” banner.
Footnote on the Pattern of Cyclogenesis & the Distribution of Precipitation Types
Empirical studies at NWS RAH have shown that a narrow transition zone of mixed precipitation is
often associated with a so-called Miller Type “A” pattern of east coast cyclogenesis. In North
Carolina, this pattern is characterized by a well organized single surface low developing along
a cold front and tracking northeastward while surface high pressure is located north of the low.
Differential (i.e. cold and dry vs. warm and moist) horizontal thermal advection between the cold
air surface high and the developing surface low plays a large role in establishing the location of
the snow/rain line. With this pattern of cyclogenesis, the transition from rain to snow occurs over
relatively short distances with the transition zone of mixed precipitation confined to a narrow
corridor, typically ranging from 10 to 70 miles wide.