1. INTRODUCTION Winter Storm of 11-13 February 2008 By Richard H. Grumm National Weather Service Office State College PA 16803 A potent winter storm brought snow, sleet, freezing rain, and rain to the Midwest and eastern United States from 11-13 February 2008. Snow was observed from Illinois to New York. In the Midwest, 3-5 inches of snow were common before the precipitation transitioned from snow to rain. A secondary cyclone developed early on the 13 th heavy rains brought flooding to portions of the Mid Atlantic region. To the south, the storm system produced severe weather in the Gulf States (Fig. 1). There were 158 reports of severe weather including 17 tornadoes on the 12 th. This was the second in a series of winter storms to produce a significant severe weather event. The event of 5-6 February produced over 453 severe reports to include 91 tornadoes. The association with winter storms with severe weather and tornado events was shown by Galway and Pearson (1981). This storm is another event fitting this winter storm archetype. The super storm of March 1993 also produced significant severe weather across Florida (Kocin et al. 1995). There were some unique aspects of this storm in the Mid Atlantic region. First, the primary cyclone moved northward through the western Appalachians into southern Ontario. This is normally not a preferred storm track for a significant snow storm. Though there were widespread reports of 4-8 inches of snow in Pennsylvania. Another point was that there was no anchoring anticyclone. Most winter storms Figure 1Storm Prediction Center (SPC) storm report by type for the 24 hour ending 1200 UTC 13 February 2008. Reports are color coded by severe weather event type. http://www.spc.noaa.gov/climo/reports/ and ice storms typically have strong easterly winds on the north side as the storm approaches. This storm lacked this feature. Despite the strong easterly flow in most snow and ice storms, this storm produce a snow to sleet to freezing rain event over much of the Mid Atlantic region. Freezing rain was observed in Pennsylvania, Maryland, and Virginia. A contributing factor may have been the extremely cold air mass that moved over the region on the 10 th. Temperatures fell into the single digits and below zero in Pennsylvania on the morning of the 11 th. This cold air mass affected most of the eastern United States. This low-level cold air mass was slow to retreat. The secondary development on the 13 th was not the typical Miller-B cyclonic event. These storms normally form on a
quasi east-west frontal zone. This storm appeared as a wave on a more north-south frontal zone. The development of this storm cut-off the warm air and much of central Pennsylvania never reached freezing. This paper will document the winter storm of 11-13 February 2008. The focus will be on the impact in the Mid Atlantic region with some evaluation of the NCEP ensembles forecast of this system. 2. METHOD and DATA Data for this study include re-analysis climatological data from the NCEP/NCAR global re-analysis project (GR: Kalnay et al 1996). The means and standard deviations were used to compute standardized anomalies, displayed in standard deviations from normal (SDs). The 00-hour forecasts from the NCEP NMA are used to provide an overview of the large scale pattern and the evolution of the storm system. Ensemble data shown here were primarily limited to the NCEP GEFS and SREF. Displays focus on the storm system and the precipitation type issues. The climatological data used to compute anomalies was restricted to those produced by the NCEP/NCAR GR data set (Hart and Grumm 2001). They will be presented in relation to both NAM and GEFS output. All data was displayed using GrADS. Anomalies were computed as described Hart and Grumm (2001) and Grumm and Hart (2001). Shaded values show the standardized anomalies computed as: SD = (F M)/σ ( ) Where F is the value from the reanalysis data at each grid point, M is the mean for the specified date and time at each grid Figure 2 NAM 00-hour forecasts of 850 hpa temperatures (C ) and heights (m) and anomalies of these fields valid at (left) 1200 UTC 11 February 2008, (center) 0000 UTC 12 February 2008, and (right) 1200 UTC 12 February 2008.
Figure 3 As in Figure 2 except NAM MSLP (hpa) and precipitable water (mm) and anomalies. point, and σ is the value of 1 standard deviation at each grid point. Anomaly data were applied to GFS forecasts and GEFS output. When anomaly data are used with EPS data it should be noted that the large anomalies are often a good indication of agreement between members. Thus, the large anomalies in the pressure and temperature fields, as well as indicating a significant event also indicated high confidence in the forecast. For brevity times are presented in the format of 13/0000 UTC which signifies 13 February 2008 at 0000 UTC. Both model and EPS initial and valid times are presented in this format. 3. RESULTS i. Overview Figure 2 (left) shows the 850 hpa temperatures in the 00-hour NAM analysis at 11/1200 UTC. These data show a cold air mass, with temperatures running 1 to 3 standard deviations below normal over the region. A broad east-west baroclinic zone was present from the plains to the southeastern United States. The mean sea level pressure and precipitable water (PW) at 11/1200 UTC showed large anticyclone and dry air over the Midwest and eastern United States. The hint of some return moisture was present over eastern Okalahoma where 16mm of PW with a +1SD anomaly was present. During the afternoon hours of the 11 th snow developed over the Midwest from Illinois southeastward into Kentucky. The strong baroclinic zone was present in this region with above 0C temperatures over central and southern Kentucky at 12/0000 UTC (Fig. 2-center). In the cold air the 850 hpa temperatures were still 1 to 2SDs below normal. At the surface (Fig 3-
center) a trough was present from Kentucky westward and into a weak cyclone over northeastern Texas. The large anticyclone and its pressure anomalies appeared to dominate over the eastern United States. A surge of warm air with +1SD PW anomalies was present and moving into the frontal zone. The 850 hpa winds (Fig 4) showed a strong southwesterly jet associated with the surge of anomalous PW. Absent in the cold air was a strong easterly jet often present in major winter storms (Stuart and Grumm 2007). The combination of cold air and strong southwesterly flow produced moderate to heavy snow in the Ohio Valley. Louisville would received around 5 inches of snow (KSDF observation) during the evening hours of 11 February, a few hours either side of 12/0000 UTC before changing to sleet. Though not shown, the precipitation shield spread rapidly eastward with snow arriving in western Pennsylvania around 12/0300 UTC. By 12/1200 UTC the 850 hpa 0C line had pushed north of the Ohio Valley and was in south-westernmost Pennsylvania (Fig. 2). An 850 hpa low was present over Illinois. At the surface, Fig 3 (right) the retreating anticyclone was present over the eastern United States with +1 to +1.5 SD above normal pressure values and weak cyclone and trough were present over the lower Mississippi River Valley. A surge of high PW was present in this trough with PW values on the order of 1 to 2 SDs above normal. Snow was falling in central Pennsylvania with surface temperatures Figure 4 As in Figure 3 except NAM 850 hpa winds and u- and v-wind anomalies valid at (left) 0000 UTC 12 Feb 2008 (center) 1200 UTC 12 February 2008 and (right) 1800 UTC 12 February 2008.
The 850 hpa winds at 12/1200 UTC (Fig 4) showed strong southsouthwesterly jet with anomalies in the core of the southerly jet on the order of 3 to 4SDs above normal. Easterly wind anomalies were present to the north near and north of the 850 hpa low center. By 12/1800 UTC the 850 hpa jet had strengthened and moved northward with the maximum v- wind anomalies over Ohio (Fig. 4 right). At the surface a cyclone was now located over Kentucky and the anticyclone was over New England (Fig. 5). The PW fields indicated a frontal boundary from Texas to Indiana with above normal PW values in he warm air ahead of this feature. Figure 5 As in Figure 3 except valid at 1800 UTC 12 February 2008. The high PW air and the strong low-level jet were associated with the severe convection present at around 10F. Figure 6 NAM 00-hour analysis valid at 0000 UTC 13 February 2008 showing (left) 850 hpa winds and a) u- wind anomalies and b) v-wind anomalies, and (right) mean sea level pressure and anomalies and b) precipitiable water and anomalies.
this time over the Gulf States. Radar echoes (not shown) indicated organized generally north-south lines of convection with some line echo wave patterns in them. By 13/1800 UTC the secondary cyclone was analyzed off the Delmarva. The PW values and anomalies had dramatically increased (Fig. 7). The 850 hpa winds showed over 5SD anomalies in the v- winds. This configuration, with anomalous v-winds and above normal PW in a northsouth frontal zone is the dominant heavy rain producing event type in the eastern United States. ii. Rains and snowfall The unified precipitation data set (UPD) showing the total 24 hour rainfall for the periods ending at 1200 UTC 12 and 13 February 2008 are shown in Figure 8. The east-west band with the initial baroclinic zone and frontal wave show light rain with an embedded area of 32 to 48 mm of precipitation over the Mississippi and Ohio valleys. The northern edge of this was snow as shown in Figure 9. The UPD precipitation ending at 13/1200 UTC is shown in Figure 8b. The impact of the developing southerly jet along the coast and the eastward shift of the precipitation shield is evident. Note the 48-64 mm area over eastern Pennsylvania, New Jersey and New York. The broad area of 8 to 16 mm over Pennsylvania was mainly snow and sleet. Heavier amounts in southeastern Pennsylvania fell as freezing Figure 7 NAM 00-hour analysis valid at 0000 UTC 13 February 2008 showing (left) 850 hpa winds and a) u- wind anomalies and b) v-wind anomalies, and (right) mean sea level pressure and anomalies and b) precipitiable water and anomalies.
Figure 8 As in Figure 5 except valid at 1800 UTC 13 February 2008. rain and rain (Fig. 10). iii. Ensemble Forecasts The threat of a winter storm, with precipitation type issues was forecast 5-7 days in advance of the event. Showing all products related to this would be prohibitive. A few images from 1200 UTC cycles are shown to illustrate the salient points. The GEFS plume diagrams from the 08/1200 and 10/1200 UTC cycles are shown in Figure 12. Both forecast cycles show a potential precipitation event in 12-14 February time frames. The forecasts from 08/1200 show two potential events and considerable precipitation type issues. Forecasts from 10/1200 UTC still show precipitation type issues, though leaning more toward snow and ice than the rain indicated in the earlier forecasts. There are also hints of a prolonged event with a delayed onset in some members. Precipitation between 12/1200 and 13/1800 appeared to be a high probability outcome. The 2m temperatures forecasts (Fig. 12) from 10/1200 UTC showed the surge of arctic air and then the gradual warming as the storm system approached. Note that
Figure 8 Unified precipitation data set showing 24 hour accumulated precipitation for the 24 hour periods ending at a) 1200 UTC 12 and b) 1200 UTC 13 February 2008. the 2m temperatures barely get to 32F in most members and the 925 and 850 hpa temperatures barely get to 0C suggesting a potentially long period of freezing rain or ice pellets. Experientially, the GEFS precipitation type is not overly reliable and the temperature plumes can help with some of these issues.
Figure 13 shows forecasts from the Figure 10 Spotter reports of snow and ice from freezing rain for the period of the storm ending about 1800 UTC 13 February 2008. All snow values in inches and ice in tenths of inches.
Figure 11. GEFS forecasts of precipitation (gray) and accumulated precipitation (colored by type) for a point near Dubois, Pennsylvania. Forecasts are initialized at 1200 UTC 8 and 10 February 2008. 10/1200 UTC GEFS. These forecasts show the primary precipitation shield with the primary low and the evolving precipitation shield with what would be the secondary cyclonic development. With some timing and placement errors, the GEFS got the large scale features correct. Comparing these images to Figure 7, the GEFS mean MSLP field had an anomalous anticyclone retreating to the east, a trough of low pressure, too far east and not low enough pressure, and an anomalous south-southwesterly jet. The ensemble mean 850 hpa jet was about 100 km northeast of the verifying system in the NAM. The wind anomalies were slightly under done. But the pattern was remarkable.
Figure 12. As in Figure 11 except GEFS plumes from the 10 February cycle at 1200 UTC showing 2m temperatures (F), 925 hpa temperatures ( C) and 850 hpa temperatures for each member. The ensemble mean is in thick black. Figure 8b suggest a fairly good spatial distribution of precipitation in the forecasts. The ensembles mean under forecast the areas of heaviest precipitation and was too slow to bring the precipitation up the coast. Moving forward 2 days, the GEFS QPF forecasts improve markedly (Fig. 14). Comparing this QPF to Figure 5 reveals a good overall pattern. The amounts are too light over the Ohio Valley, but the ensemble prediction system (EPS) has begun to pickup the potential for heavy rainfall along the East Coast. The surface pressure pattern forecast and the 850 hpa wind forecasts (Fig 14) also show marked improvement verse the verifying analysis (Fig 7). The 850 hpa cyclone is nearly spot on in southwestern Michigan. The anomalous 850 hpa westerly winds are also well forecast. Cleary, shorter range forecasts provide better guidance which more realistically depicts the verifying conditions. Though not shown, the GEFS forecast a secondary cyclone to develop over the coastal plain of Virginia and North Carolina around 13/1200 UTC and be off the New Jersey coast by 13/1800 UTC. The SREF also forecast the larger scale features with considerable skill. The 12/0300 UTC SREF MSLP forecasts are shown in Figure 15. They look surprisingly similar to those presented by the GEFS. Note the secondary cyclonic development over North Carolina in the pressure field and the area of -2SD pressure anomalies in these forecasts. Additionally, the SREF forecast the anomalous southerly at 13/1200 UTC and the surge of anomalously high PW along the coast which produced the heavy rainfall observed in Figure 10b. For brevity SREF 24 hour QPF forecasts are shown in a summary format in Figure
16. The forecasts are presented for comparison with Figures 8a and 8b. The SREF captured the pattern well, to include the stripe of precipitation coming in from the northwest on the 11 th ending at 12/1200 UTC. The SREF under forecast the heavy rainfall in the northeast. As indicated by the 36-hour accumulated QPF (Fig 16 right) the QPF error was more of a timing and time of accumulation error in the SREF EPS rather than a missed forecast. 4. CONCLUSIONS A strong winter storm affected the eastern United States from 11 through 13 February 2008. This storm brought snow from Illinois to New York. The snow changed to sleet and freezing rain from Kentucky, eastward across Pennsylvania. Eventually, many of these areas saw the freezing rain turn to rain as the warmer air moved in ahead of the initial surface cyclone. In the warm air over the Mid Atlantic region, a strong low-level jet and above normal PW air brought heavy rain and minor flooding to portions of eastern Pennsylvania, New Jersey, and southern New York State. Farther south, the storm produced a line of showers and thunderstorms. For the second time in about a week, a winter storm produced a tornado outbreak in the southern United States. As shown in Figure 1, over 158 reports of severe weather and 17 tornadoes were observed. The winter of 2007-2008, a La Nina winter has been a prolific producer of winter storms with severe weather. Over Pennsylvania, the broad area of 3-8 inches of snow was a rare heavy snow event without a strong easterly jet. Heavy snow with strong south-southwesterly flow also known as warm advection snow is not a common phenomena. The snowfall totals were based on 24 hour amounts and few locations received more than 3-5 inches in any 12 hour period. Southern Pennsylvania saw the snow turn to sleet, freezing rain and rain. Central areas ended as freezing rain until the cold air transitioned the precipitation back to snow. In this case, the GEFS provided a long lead forecast on the potential for the event. It showed some uncertainty with the precipitation type, timing, and amounts, which became a smaller problem at shorter ranges. The GEFS forecast the large scale pattern remarkably well and it showed the pattern of the precipitation quite well. Clearly, the GEFS missed some of the important details, such as the maximum precipitation amounts and the location of the surface and 850 hpa cyclones. But these errors were quite smaller at shorter ranges. The SREF showed similar skill and made quite good forecasts of the pattern and amounts of the rainfall. The synoptic pattern was also quite well forecast. Similar to the GEFS, the SREF correctly forecast the secondary cyclone along the coast. Though not shown, this cyclonic development likely limited the intrusion of warm air into central and northern Pennsylvania. The 2m temperatures in the SREF were forecast to go above freezing in many locations where the warm air never made it. The SREF had a good QPF forecast but appeared to suffer from some timing issues related to when the heavy rain would fall. Thus, the models and EPS s have some limitations and shorter range
forecasts are, as in this case, more accurate than longer range forecasts. 5. Acknowledgements Data access and images provided by Ron Holmes, National Weather Service State College. 6. References Doty, B. E., and J. L. Kinter III, 1995: Geophysical data and visualization using GrADS. Visualization TechniquesSpace and Atmospheric Sciences, E. P. Szuszczewicz and Bredekamp, Eds., NASA, 209 219. Dalcher, A., E. Kalnay, R.N.Hoffman,1988: MediumRange Lagged AverageForecasts. Mon. Wea. Rev.,116,402-416. Doty, B. E., and J. L. Kinter III, 1995: Geophysical data and visualization using GrADS. Visualization TechniquesSpace and Atmospheric Sciences, E. P. Szuszczewicz and Bredekamp, Eds., NASA, 209 219. Du et al 2006 New Dimension of the NCEP Short-Range Ensemble Forecasting (SREF) System: Inclusion of WRF members. Grumm, R.H. and R. Hart. 2001:Standardized AnomaliesApplied to Significant Cold Season Weathe r Events:Preliminar y Findings. Wea. and Fore.,16,736 754. Hart, R. E., and R. H. Grumm, 2001: Using normalized climatological anomalies to rank synoptic scale events objectively. Mon. Wea. Figure 13. GEFS forecasts initialized at 1200 UTC 10 February 2008 showing (left) a) MSLP and spread and the ensemble mean MSLP fields; (center) a) 850 hpa winds and u-wind anomalies and b) 850 hpa winds and v-wind anomalies and (right) a) probability of 0.30 inches of QPF in 24 hours ending 1200 UTC 13 February 2008 and lower panels shows the ensemble mean QPF and each member 0.30 inch contour along with the ensemble mean value (thick black).
Rev., 129, 2426 2442. Sivillo, S.K,J.E. Ahlquist, and Z. Toth,1997: An ensemble forecasting primer. Wea. Forecasting.,12, 809-818. 1999: Using Ensembles for Short- Range Forecasting, Mon. Wea. Rev., 127, 433-446. Stensrud D. J., H. E. Brooks, J. Du, M. S. Tracton, and E. Rogers, Figure 24 As in Figure 13 except initialized at 1200 UTC 12 February 2008.
Figure 9 Snow fall analysis showing snow fall on the 12 and 13 th of February 2008. Data from the National Snowfall analysis website http://www.nohrsc.noaa.gov/nsa/
Figure 15 SREF forecasts initialized at 0300 UTC 12 February 2008 valid at 1200 UTC 13 February 2008 showing (left) a) PW spaghetti and spread and b) the ensemble mean PW Field and the anomalies and (right) a) 850 hpa winds and u-wind anomalies and b) 850 hpa winds and v-wind anomalies.
Figure 16 SREF precipitation forecasts showing (upper panels) the probability of the specified thresholds and (lower panels) the ensemble mean (shaded) and each members forecast of the specified contour. The left panels was initialized at 0300 UTC 11 February 2008 and shows the probability of 1 inch or more QPF for the 24 hour period ending at 1200 UTC 12 February 2008. The center and left panels are from the 0300 UTC 12 February cycle showing the 24 and 36 hour accumulated QPF ending at 1200 UTC 13 February and 0000 UTC 14 February 2008.