Eastern United States Winter Storm of 14-15 December 2013
Value of Anomalies for Situational Awareness
Richard H. Grumm
National Weather Service State College, PA
A winter storm brought heavy snow, freezing rain and rain from the Midwest to New England on 14-15 December 2013. The storm was over the southern Plains (Fig. 1a) at 0000 UTC 14 December 2013 and moved into the Ohio Valley by 1200 UTC 14 December (Fig. 1b). The system followed a Miller-B type (Miller 1946, Kocin and Uccellini 1990, Mote et al. 1997) of evolution with a secondary cyclone developing along the coast around 0000 UTC 15 December. The coastal cyclone became the primary cyclone bringing snow and rain to New England.
A key feature in many snow storms over the eastern United States is the presence of a strong anticyclone north and east of the cyclone (Kocin and Uccellini 2004). This pattern, associated with standardized anomalies, was presented as a potential predictor of snow storms (Root et. al 2008). The anticyclone is often critical in maintain low-level cold air. Similar to many studied snow storms, this storm possessed a strong anticyclone which tracked across southern Canada and northern New England as the cyclone moved across the eastern United States (Fig. 1a-e).
The flow around the anticyclone often produces strong eastern flow north of the surface cyclone. Stuart and Grumm (2006) showed that many historic and high impact snowstorms in the eastern United States were associated with strong 925 and 850 hPa easterly flow and u-wind anomalies in the -2 to -5 range. Root et al (2008) using re-analysis (R-Climate) data showed that heavy snow events in the eastern United States had a clear signal in the low-level u-wind anomaly fields. As this storm tracked across the southern Plains, easterly flow developed between the cyclone and the anticyclone to the north. The 850 hPa winds (Fig. 2) show the negative u-wind anomalies which developed north of the cyclone center. The 850 hPa winds peaked in northern New England (Fig. 2d) as the secondary cyclone deepened.
Other features associated with East Coast winter storms include a strong jet circulation (Kocin and Uccellini 2004) with a jet entrance circulation north and east of the surface cyclone and a strong southern stream jet and jet exit region. Early in this event, a strong 250 hPa jet was present over the eastern United States with an implied jet entrance region over the Great Lakes (Fig. 3a). A southern stream jet was present over New Mexico and western Texas. The standardized anomalies implied a modest coupled jet over the eastern United States by 0000 UTC 15 December 2013 (Fig. 3c).
The synoptic setting for this event showed a fairly typical scenario for a winter storm over the eastern United States. The predictability of the event from numerical guidance and leveraging these data to produce useful forecasts was the key issue in this and all winter storms. Standardized anomalies (Hart and Grumm 2001) can aid in identifying patterns associated with extremes. When used with forecast guidance (Grumm and Hart 2001; Stuart and Grumm 2006; Junker et al. 2008) they can aid in predicting extreme events and identify when forecast systems are predicting potentially extreme events, relative to a base climatology. It is immaterial whether the climatology is re-analysis based (R-Climate) or model or ensemble prediction system based (M-Climate).
The final component of predicting winter storms is the precipitation amount and type. Models historically have a difficult time predicting QPF, and in tight baroclinic zones, the precipitation type can produce uncertainty issues. Ensemble prediction systems are extremely useful when dealing with uncertainty in the QPF and precipitation type.
This paper will document the winter storm of 14-15 December 2013. The focus is on how to use standardized anomalies to characterize and identify the potential event and how ensemble forecast data may have contributed to a successful forecast of the storm. In this case, the QPF was relatively well predicted which likely contributed to the success of the forecasts.
Data and Methods
The large scale pattern was reconstructed using the 00-hour forecast of the NCEP Global Forecast System (GFS) as a first guess at the verifying pattern. The standardized anomalies were computed in Hart and Grumm (2001). All data were displayed using GrADS (Doty and Kinter 1995). The standardized anomalies and the probability distribution functions are based on the re-analysis climate (R-Climate). Though not shown here, they could be produce from internal model system climatologies (M-Climate).
The traditional standardized anomalies were produced from the GFS 00-hour forecast using Climate Forecast System based means and standard deviations. The climatology spans a 30 year period. This 30-year period was used in the situational awareness (SA) tables to show the return period of the anomalies. Standardized anomalies were applied to analysis and forecasts of the event.
The forecasts in the SA tables were derived from the 42-member North American Ensemble Forecast System (NAEFS). These data are displayed using the NAEFS mean fields relative to the climate. In addition to standardized anomalies, the return period and interval of the anomalies are produced relative to the 21-day centered values. The return period is in years such that a 30-year return period represents the extreme in the climatological data set. The percentile data represent where along the climatological probability distribution the current NAEFS forecast lies. Extremes in the tails would be in the 97th to 100th percentile (MAX) or in the 3rd to 0th percentile (MIN). In addition to finding the MAX and MIN relative to the time, the MAX and MIN are computed using the 4 climatological times of 00, 06, 12, and 18 UTC. This provides the ability to estimate the MAX (MIN) at the specified time or for all 4 times in the same 21-day centered climatological window.
The estimated quantitative precipitation estimates were from Stage-IV data. Snow amounts were retrieved locally and from reports collected by the National Weather Service.
Pattern of storm in Mid-Atlantic and eastern United States
The larger scale 500 hPa pattern (Fig. 4) showed a deep trough over southeastern Canada at 1200 UTC 13 December and a ridge building over the eastern Pacific and western North America (Fig. 4a-f). A short-wave in the flow was evident over New Mexico at 1200 UTC 13 December and was associated with the surface cyclone (Fig. 1a) over northwestern Texas at 0000 UTC 14 December. A second 500 hPa wave in the northern stream deepened the trough over the Great Lakes (Figs. 4b-c) and there was some apparent phasing of these two waves.
Over the northeast, the GFS analyzed a deep cold air mass at 1200 UTC 14 December with the -8C contour at 850 hPa (Fig. 5) over Pennsylvania and Ohio. As the surface cyclone moved north (Fig. 6) into the Ohio Valley, the 850 hPa temperatures rose rapidly. By 0000 UTC 15 December, (Fig. 5c) the 850 hPa temperatures rose above 0C in southwestern and southern Pennsylvania, while at 700 hPa the 0C isotherm covered much of the eastern two thirds of the state (Fig. 7). Not surprisingly, the snow over most of Pennsylvania changed to ice pellets and freezing rain over all but the northwestern part of the state1. The cold air with sub-0C temperatures moved back into the region as the primary low deepened and moved to the north (Fig. 6c-f).
The 6-hour increment surface pattern over the eastern United States (Fig. 6) showed that the secondary cyclone remained intact as it moved into northeastern Ohio (Fig. 6c). The flow ahead of this system was important as it related to the surge of warm air aloft to the east (Figs. 5 & 7) . The secondary low was clearly identified in the GFS by 0600 UTC 15 December. As the secondary low circulation deepened, cold air moved in from the west. The precipitation ended as light snow over much of western and central Pennsylvania as the cold air moved back in (Figs. 5 & 7).
The resulting estimated precipitation in 6-hour increments (Fig. 8) showed the precipitation arrived in the Mid-Atlantic region mainly after 1200 UTC 14 December. The period of highest QPF amounts in central Pennsylvania was the 6-hour period ending at 0000 UTC 15 December. The total 24-hour QPE over Pennsylvania (Fig. 9) showed an area of 16 mm (0.63 inches). Most of the region covered by the 16 mm contour had over 6 inches of snow and some mixed precipitation. The pattern of the 16 mm contour implied the evolution of snow bands which were evident and analyzed on radar (not shown). The inset shows the higher QPE amounts in northern New Hampshire and Quebec where estimated QPE ranged from 24 to 48 mm. This was in close proximity to the strongest 850 hPa jet and u-wind anomalies (Fig. 2d).
The focus here will be on the general pattern in the NCEP GEFS and SREF and the quantitative precipitation forecasts (QPFs). The QPF fields shown the 16mm contour and the probability of exceeding it as this relates closely, using a quick first guess 10:1 ratio to about 6 inches of snow. All systems predicted a Miller-B evolution though the track of the secondary low showed considerable variability at longer ranges which at times suggested the storm would stay farther south. The GEFS mean sea-level forecasts from 6 GEFS cycles (Fig. 10) showed a weaker storm farther south in forecasts issued prior to 0000 UTC 13 December (Figs. 10d-f).
The SREF (Fig. 11) showed a similar cyclone evolution as the GEFS and shorter range forecasts appeared to have better positions with the cyclone and secondary cyclone evolution. The uncertainty, not shown, indicated a convergence of solutions as forecast length decreased as the dispersion about the mean was relatively small from forecasts issued after 1200 UTC 12 December.
The GEFS and SREF mean QPF fields (Figs. 12 & 13) show that the QPF shield orientation and area of maximum QPF shifted westward from forecasts in the GEFS issued after 1200 UTC 11 December. This westward shift was when the uncertainty in the mean sea-level field got smaller and the secondary cyclone was predicted to stay closer to the coast. The SREF showed a similar shift between the 0300 and 2100 UTC 12 December runs (Fig. 13) though the shift was actually in the 1500 UTC SREF run (not shown). The shift put the 16mm contour over much of Pennsylvania with conditions supporting mainly snow and freezing rain (not shown).
The GEFS (Fig. 14) and SREF (Fig.15) probability of 16mm or more QPF. The shorter range forecasts all indicate a high probability (over 90%) of 16mm or more QPF over most of Pennsylvania and southern New York in forecasts issued on or after 1200 UTC 12 December. The convergence of solutions in the two ensemble forecast systems combined with the conditions supporting mostly snow indicated a potential winter storm with over 6 inches of snow. Though not shown probabilities of QPF over 25 mm (1 inch) were generally 0 to 5 percent except along the coast where most of the QPF was forecast to fall as rain. Along the cast the probability of 25mm or more QPF was over 90%.
It would be exhaustive to show the temperature profiles and pattern to determine snowfall forecasts from each GEFS and SREF run. Thus one 24 hour summary of snowfall and ice from the SREF is presented here. The GEFS was very similar and not shown. The SREF snow precipitation type forecasts (Fig. 16) show the high probability of snow mainly north of the track of the surface cyclone. All SREF forecast had 6 inches of snow over most of Pennsylvania with the exception of the SREF initialized at 0300 UTC 12 December (Fig. 16f) which had a 4 inch contour. The SREF initialized at 2100 UTC 13 December showed an 8 inch contour (Fig. 16b). Overall the SREF consistently predicted a 4-8 inch snowfall over much of Pennsylvania and southern New York. Though not shown earlier accumulation intervals showed more snow in the Ohio Valley and later times showed more snow over New England. One limiting factor in the snow in southeastern Pennsylvania was the precipitation type issue as the SREF had freezing rain (Fig. 17) and ice pellets (not shown). The SREF probabilities of 10-30% chance of freezing rain were low, instantaneous values in 3-hour and 6-hour time increments showed periods where the probabilities of freezing rain into central Pennsylvania exceed 60%.
A surface cyclone moved out of the southern Plains on 13-14 December 2013 producing rain and snow as it tracked into the Ohio Valley. As the surface cyclone move into the Ohio Valley, a secondary cyclone developed along the coastal plain. This Miller-B cyclone evolution allowed warm air aloft to change snow to ice pellets, freezing rain and rain between the developing secondary and primary cyclone. The warm air long the coast produced locally heavy rainfall.
The standardized anomalies as shown and as predicted indicated a potential winter storm. Due to the lack of large u-wind anomalies at 850 hPa the snow and rain was not expected to be extreme and snowfall was not excessive with this storm. Observations showed the heaviest snow fell over eastern New York, New Hampshire and Maine was 12 to 17 (16.5) inches were observed. Over Pennsylvania snowfall ranged from an inch to 12 inches (11.8 in Avis). In north-central Pennsylvania 6-10 inch snowfall amounts were common, the higher amounts likely the result of higher than 10:1 snow ratios early in the event as liquid equivalents were generally in the 16 (0.63) to 20mm (0.78 in) range.
This low-end winter storm was generally well predicted by ensemble and deterministic forecast models with 2-3 days of lead-time. The relatively successful QPFs and pattern forecasts combined with the convergence of solutions likely contributed to the successful forecasts of the event. In this event the QPF in the SREF and GEFS supported a low-end winter storm as the QPF in areas to be affected by mostly snow averaged 16 to 20 mm of QPF with few if any locations expected to see 25mm or more QPF. With both systems predicting sufficient QPF for 6 inches or more snow, this case was likely a human success due to the model success. The verifying QPE (Fig. 4) showed that 16mm was the area average QPF over most of the affected area2.
Two earlier winter storms in the Mid-Atlantic region were poorly predicted. The event of 6 December had large positive QPF errors and the event of 8 December most forecast systems missed the heavy snow from northern Maryland into southern New Jersey. Despite some successes and some failures each event has to be considered with the highest confidence forecast available. But the variable nature of QPF in these forecast systems suggest some type of Bayesian corrections would be of long term value.
The Pennsylvania State University for gridded data access and Elyse Colbert for editing.
Associated Press, 2013a: LeSean McCoy runs for 217yards, Eagles beat Lions 34-20 in snow storm. And similar stories. 8 December 2013
Associated Press, 2013b: California citrus growers fight off cold spell. And similar stories 6-8 December 2013.
Doty, B.E. and J.L. Kinter III, 1995: Geophysical Data Analysis and Visualization using GrADS. Visualization Techniques in Space and Atmospheric Sciences, eds. E.P. Szuszczewicz and J.H. Bredekamp, NASA, Washington, D.C., 209-219.
Graham, R.A, T. Alcott, N. Hosenfeld, and R. Grumm, 2013: Anticipating a rare event utilizing forecast anomalies and a situational awareness display: The Western Region US Storms of 18-23 January 2010, BAMS, 2013.
Graham, R A. and R. H. Grumm, 2010: Utilizing Normalized Anomalies to Assess Synoptic-Scale Weather Events in the Western United States. Wea. Forecasting, 25, 428-445.
Grumm, R.H. and R. Hart. 2001:Standardized Anomalies Applied to Significant Cold Season Weather Events: Preliminary Findings. Wea. and Fore., 16,736–754.
Grumm, Richard H., 2011a: The Central European and Russian Heat Event of July–August 2010. Bull. Amer. Meteor. Soc., 92, 1285–1296.
Grumm, R.H. (2011b) New England Record Maker rain event of 29-30 March 2010. NWA. Electronic Journal of Operational Meteorology, EJ4 12: 1-31.
Hart, R. E., and R. H. Grumm, 2001: Using normalized climatological anomalies to rank synoptic scale events objectively. Mon. Wea. Rev., 129, 2426–2442.
Junker, N.W, M.J.Brennan, F. Pereira,M.J.Bodner,and R.H. Grumm, 2009:Assessing the Potential for Rare Precipitation Events with Standardized Anomalies and Ensemble Guidance at the Hydrometeorological Prediction Center. Bulletin of the American Meteorological Society, 4 Article: pp. 445–453.
Junker, N. W., R. H. Grumm, R. Hart, L. F. Bosart, K. M. Bell, and F. J. Pereira, 2008: Use of standardized anomaly fields to anticipate extreme rainfall in the mountains of northern California. Wea. Forecasting, 23, 336–356.
Kocin, P. J., and L. W. Uccellini, 2004: Northeast Snowstorms, Volume I: Overview. Meteor. Monogr., Vol. 32, No. 54, Amer. Meteor. Soc., 1-296.
Kocin, P.J, and L.W. Uccellini. 1990. Snowstorms along the Northeastern Coast of the United States: 1955 – 1985. Meteorological Monograph No. 44, American Meteorological Society
Miller, J.E. 1946. Cyclogenesis in the Atlantic Coastal region of the United States. Journal of Meteorology 3: 31–44.
Mote, T.L., D.W. Gamble, S. J. Underwood, M.L. Bentley. 1997. Synoptic-scale features common to heavy snowstorms in the Southeast United States. Weather and Forecasting 12: 5–23.
Stuart, N.A and R.H. Grumm 2006: Using Wind Anomalies to Forecast East Coast Winter Storms. Wea. and Forecasting,21,952-968.
Figure . GFS 00-hour forecasts of mean sea-level pressure and pressure anomalies in 12 hour periods from a) 0000 UTC 14 December 2013 through f) 1200 UTC 16 December 2013. Isobars every 4 hPa and standardized anomalies in standard deviations from normal as in color bar. Return to text.
Figure . As in Figure 1 except for 850 hPa winds (kts) and u-wind anomalies. Return to text.
Figure . As in Figure 2 except for 250 hPa winds and total wind anomalies. Return to text.
Figure . As in Figure 1 except for 500 hPa heights (m) and height anomalies for 24 hour periods from a) 1200 UTC 13 December through f) 1200 UTC 18 December 2013. Return to text.
Figure . As in Figure 1 except for 850 hPa temperatures and temperature anomalies in 6-hour increments from a) 1200 UTC 14 December through f) 1800 UTC 15 December 2013. Return to text..
Figure . As in Figure 5 except for mean sea-level pressure and pressure anomalies. Return to text.
Figure . As in Figure 6 except for 700 hPa temperatures. The enhanced red isotherm is the 0C isotherm to illustrate the warm air loft which caused a precipitation type issue. Return to text.
Figure . Stage-IV estimated QPE (mm) for the 6-hour periods ending at a) 1800 UTC 14 December, b) 0000 UTC 15 December, c) 0600 UTC 15 December and d) 1200 UTC 15 December 2013. Return to text.
Figure . As in Figure 8 except total QPE for the 24 hour period ending at 1200 UTC 15 December 2013. The inset in the upper right shows the total QPE for the 48 h period ending at 1200 UTC 16 December and the higher QPE amounts in New England. Return to text.
Figure . NCEP GEFS ensemble mean forecast of mean sea-level pressure and anomalies from forecasts initialized at a) 0000 UTC 14 December , b) 1200 UTC 13 December, c) 0000 UTC 13 December, d) 1200 UTC 12 December, e) 1200 UTC 11 December and f) 1200 UTC 10 December 2013. Return to text.
Figure . As in Figure 10 except NCEP SREF initialized at a) 0900 UTC 14 December, b) 2100 UTC 13 December, c) 1500 UTC 13 December, d) 0900 UTC 13 December, e) 2100UTC 12 December and f) 0300 UTC 12 December 2013. Return to text.
Figure . As in Figure 10 except GEFS ensemble mean QPF (shaded) and each members 16mm contour and a thick black line for the ensemble mean 16mm contour for the 24 hour period ending 1200 UTC 15 December 2013 Return to text.
Figure . As in Figure 11 except SREF ensemble mean QPF (shaded) and each member 16mm contour and the ensemble mean 16mm contour in thick black for the 24 hour period ending 1200 UTC 15 December 2013. Return to text.
Figure . As in Figure 12 except GEFS probability of 16mm or more QPF in the 24 hour period ending 1200 UTC 15 December 2013. Return to text.
Figure . As in Figure 13 except SREF probability of 16mm or more QPF in the 24 hour period ending 1200 UTC 15 December 2013. Return to text.
Figure . As in Figure 15 except SREF 24 hour snow precipitation (inches) and the probability of snow in the 24 hour period. A simple 10:1 snow ratio is employed as a first guess. Return to text.
Figure . As in Figure 16 except 24 hour accumulated freezing rain and probability of freezing rain over the 24 hour period. Return to text.
1 In State College ice pellets and snow grains mixed in around 0030 UTC 15 December after 7 inches of snow fell. By 0300 UTC freezing rain was observed though little accumulating precipitation was observed after 0400 UTC.
2 Personal observation 0.69 melted with 7 inches snow and 1 hour freezing rain.