The Valentine’s Day Snowstorm of 2007


Meteorological Conditions:


            This storm occurred while the NAO was neutral, which had been the NAO phase since the beginning of February (Fig. 1A).  The PNA was positive, and had been in this phase since mid January (Fig. 1B).  The PNA was not very strongly positive, but positive enough so that the teleconnection with Polar Vortex centered around Hudson Bay Canada produced a fairly deep mean upper trough across the eastern U.S. from mid January into mid February.  This upper-level pattern across North America contributed to below normal temperatures across the eastern U.S. from mid January into mid February, but also resulted in a suppressed storm track and below normal precipitation, with our region being affect by mainly clipper systems. 



Figure 1.  Observed and forecasted NAO (left) and PNA (right) from 1 November 2006 through 1 March 2007.


The potential for a significant storm became evident 5 to 7 days before the storm, around 7-9 February.  As has been typical this winter, the GFS and ECMWF solutions diverged in terms of timing, track and strength of the system, but both models suggested a Miller-B type storm redeveloping off the U.S. east coast.  The GFS suggested a consistently more southerly track, off the DELMARVA, due to colder air being entrenched in the northeastern U.S., while the ECMWF suggested a more northerly track, south of Long Island, also suggesting the Polar Vortex temporarily retreating a bit north in the Hudson Bay region of Canada.  Even 5 to 7 days before the storm, the forecasted upper patterns in both the GFS, ECMWF and MREF suggested a relatively low amplitude +PNA, which supported a less amplified downstream upper trough over the eastern U.S., and more northerly storm track, suggesting the GFS was too cold and too far south with the storm.

Once the storm was expected in 3 to 5 days, the ECMWF and GFS began to converge toward a common timing track and strength of the storm.  However, the GFS was still suggesting a more southerly track and colder air in the northeastern U.S., but differences with the ECMWF were much less.  Even with the differences in the long-range forecast guidance, there was a decent consensus that a southern stream upper system would phase with a northern stream upper low dropping southeast out of Canada into the Great Lakes and northeastern U.S.  The 2 streams were forecasted to phase somewhere along the mid-Atlantic or New England coast late Wednesday 14 February which would result in rapid deepening of the surface low near New England.  About 72 to 84 hours before the storm was expected, the Short Range and Medium Range Ensemble Forecasts were suggesting the potential for a historical storm, particularly in the wind anomalies (Figs. 2A and 2B), QPF probabilities for 1.00” liquid equivalent (Figs. 3A and 3B), and plume diagrams (Figs. 4A and 4B).  The wind anomalies in figure 2 did not quite meet the -4 SD threshold 3+ days prior to the storm, but subsequent MREF and SREF anomaly displays edged just into the -4 to -5 SD range.  Upper level wind anomalies at 250 hPa flirted with -2.5 SD for multiple MREF and SREF runs prior to the storm (not shown), so the guidance did suggest a potentially long duration storm.  Additionally, the plume diagrams suggested at least 24 hours of continuous measurable precipitation.  The HPC WWD graphics were consistent in snowfall amounts and locations of maxima with each forecast issuance from 12 February until the onset of the storm (not shown).

A) B)


Figure 2.  A)  03Z 11 February SREF 850 hPa forecasted U wind anomalies (top) and V wind anomalies(bottom) valid 12Z 14 February.  B)  18Z 11 February MREF 850 hPa forecasted U wind anomalies (top) and MSLP anomalies (bottom) valid 00Z 15 February.




Figure 3.  A)  15Z 12 February SREF probability for 1.00” liquid equivalent and spread valid 06Z 15 February.  B) 12Z 12 February MREF probability for 1.00” liquid equivalent and spread valid 06Z 15 February.




Figure 4.  Plume diagrams for Albany, NY from A) 15Z 12 February SREF and B) 12Z 12 February MREF.



The consensus from all sources of guidance, the GFS, NAM, MREF and SREF was nearly unprecedented 48 to 72 hours before the storm.  There were subtle differences in the storm track and precipitation transition zone between the GFS and NAM.  The SREF and MREF solutions were definitely biased by their respective ensemble members, with the SREF mean and spread favoring the NAM and MREF favoring the GFS.  Still, the slight differences showed the GFS/MREF slightly further east with the track of the storm and the precipitation transition zone. 

Based on the water vapor (Figs. 5A-D), infrared and visible satellite pictures (Figs. 6A and 6B) of the storm as it tracked out of the eastern Pacific across the southwestern U.S. and into the southern plains, this was a very well-developed system (evident in 500 hPa analyses in figs. 7A-D) with associated convection suggesting downstream upper ridging would likely produce a more northwestern storm track and precipitation transition zone.  As the storm lifted out of the lower Mississippi Valley into the Tennessee Valley, severe convection broke out over the northern Gulf Coast States, increasing the confidence that the slightly more north and west NAM/SREF depiction of the storm track and precipitation transition zone would be most realistic.  Dynamic tropopause parameters from the models suggested enhanced convection and downstream ridging that would contribute to rapid deepening of the surface low once the northern and southern stream features phased over New England late on Wednesday 14 February.  Upper jet dynamics were supportive of a major storm as well, with an indication of dual jet structure, along with a very strong southern stream jet (Figs 8A and 8B).






Figure 5.  Water Vapor satellite imagery with 15 minute lightning plot overlay for A) 1200 UTC 11 February, B) 1145 UTC 12 February, C) 2045 UTC 13 February and D) 1945 UTC 14 February.





Figure 6.  Visible satellite imagery with 15 minute lightning plot overlay for A) 2045 UTC 13 February and B) 1945 UTC 14 February.


Still, BUFKIT, SREF/MREF probabilities for 1.00” QPF, and Plume diagrams showed QPF ranges unusually high for a nearly all snow event for the Albany area.  The big uncertainty was the snow to liquid ratio.  However, even with a 1:10 ratio, much of our forecast area could expect 9” or more of snow or a combination of snow and sleet.  Another question was whether there would be enough wind for blizzard conditions, since the visibilities certainly would drop to less than ¼ mile at times in many areas.





Figure 7.  500 Mb heights, contours, wind barbs and temperatures on A)1200 UTC 12 February, B) 1200 UTC 13 February, C) 1200 14 February, D)1200 UTC 15 February.




Figure 8.  250 Mb wind barbs, isotachs and streamlines for A) 1200 UTC 14 February and B) 1200 UTC 15 February.


What was learned from this event?


The unusual consistency seen in each 15 member ensemble and the operational models resulted in the highest level of confidence possible for issuing Heavy Snow and Winter Storm Watches during the early morning of 12 February.  Heavy Snow and Winter Storm Warnings were issued during the early morning of 13 February.  The Winter Storm Warnings were issued from the southern Catskills through the POU area, NW CT and the southern Berkshires, where snow amounts would be reduced by sleet, but still a combination would add up to 9” or more.  Based on the evolution seen in satellite pictures and radar, some sleet was put in the forecast for the Capital District through southern VT.

Collaboration with surrounding offices through 12Planet was smooth as there were few if any disagreements on the forecast evolution.  HPC WWD graphics were helpful, and defined the axis of heaviest snow very well.  It was felt, however, that their snowfall amounts were a little low, and zone of freezing rain was too far north and west, into our forecast area.  The Grid editing and formatters worked well, with no problems noted, and snowfall amounts in the text consistent with Grids.

As far as snowfall amounts, the SREF and MREF plumes were in unusual agreement for 1.25” to over 2.5” of liquid equivalent, and the probabilities for 2.00” in 36 hours were above 50% for a significant portion of the forecast area.  We thought 18” to 36” of snow was realistic for areas where mostly snow was predicted, and 10” to 20” in areas that were expecting varying degrees of sleet mixed in.  All parameters pointed to unusually strong frontogenesis and banding (Figs. 9A and 9B), and wherever the band or bands set up, there could be more than 36”.  During the storm, the SPC issued timely and helpful mesoscale discussions and graphics outlining precipitation types, regions and intensities (Figs. 10A-C).




Figure 9.  Base reflectivity from A) KENX at 1759 UTC 14 February, and B) a mosaic of composite reflectivity at 1700 UTC February 14.



Figure 10.  Mesoscale graphics from the SPC highlighting precipitation types, locations and intensities during the storm impact.


Snowfall forecasts of this magnitude are rare, but the consensus from all sources of guidance was rare, and the pattern recognition supported the extreme snowfall amounts.  The consensus on the forecasted storm track from 2 separate 15 member ensembles and operation model runs, tracking the storm through Cape Cod (Figs. 11A and 11B), a favored track for snowstorms in our region, was nearly unprecedented. Waiting until the Warnings were issued on 13 February to include these “epic” storm total snowfall forecasts was definitely the best decision and provided at least 36 hours of lead time for users to take precautions and make decisions before warning level snowfall was observed.




Figure 11.  0000 UTC 15 February A) MSAS NWS MSLP and B) NAM12 initial 500 hPa heights and MSLP.  Note the surface low over interior cape Cod and an upper low over eastern NY.


The NAM/SREF precipitation transition zone worked out well, as areas from the Capital District through the Berkshires and southern VT did see a brief period of sleet (see KALY soundings, Figs. 12A and 12B).  So, using satellite imagery, radar imagery and Dynamic Tropopause information, it was recognized that there would be enhanced downstream upper ridging and warming further north and west, affecting the location of the precipitation transition zone.




Figure 12.  KALY sounding from A) 1200 UTC 14 February and B) 1800 UTC February.  Note the 1200 UTC sounding is completely below freezing, but the 1800 UTC sounding just touches the 0C isotherm around 800 hPa.  A period of sleet was reported in Albany.


Blizzard Warnings were issued during the morning of 2/14, based on potential wind gusts to 35 MPH during the last few hours of the heaviest snow.  The last band of snow lingered until around midnight in the Capital District as the upper deformation zone and TROWAL tracked through.  The last band of snow also looked like it was enhanced by the local river convergence zone (Fig. 13).



Figure 13.  Base reflectivity from KENX 0430 UTC 15 February.  Note the enhanced band oriented north-south through the mid Hudson Valley and Capital District.


The snow to liquid ratios were nearly 10:1, even with surface temperatures over much of the region between 5˚F and 20˚F.  There was a warm layer aloft around -5˚C to -10˚C that probably reduced the dendritic growth at least in the Capital District.  The ratios were likely higher within the enhanced band that developed northwest of the Capital District.  There was one primary enhanced band (Figs. 9A and 9B), not multiple bands, this could be researched.