February 13-14 Historical Snowstorm

The NWS at ALY forecast area was impacted by a Miller Type-A Nor'easter 13-14 February 2014 that brought a widespread significant snowfall. The coastal low developed over the GA/SC coast Wednesday evening on 12 Feb 2014, as it produced significant ice and snow amounts for the Southeast and mid-Atlantic Region. The low deepened and intensified as it moved northeast along the Eastern Seaboard Wednesday morning. It was a "synoptic bomb" deepening from 997 hPa near VA/NC coast at 12Z 13 FEB 2014 to 973 hPa near eastern ME at 12Z 14 FEB 2014 (24 hPa or greater fall in 24 hours or less). The water vapor loop nicely showed a northern stream short-wave phasing with the southern stream disturbance Wednesday morning. The phasing of the streams allowed the mid-level low to close off over the mid-Atlantic and the long wave trough to become negatively tilted. This allowed the Nor'easter to track closer to the coast, which the NWP guidance had a hard time resolving in the medium range (even the short-range), as the NAM/GFS/GEFS favored a track further south of the region compared to the ECMWF/CAN GGEM a few days before the event.

The ECMWF was the most consistent model during the days preceding the event, not just in terms of the surface low track and more accurate QPF, but correctly depicting the mid/upper low closing off earlier and farther south and west than other sources of guidance were indicating. This aspect was a key component in forecast the evolution of this system. The ECMWF was also consistent depicting the upper-level structure to the system as well, being the most aggressive showing the phasing of the northern and southern stream Potential Vorticity anomalies. The GFS/NAM eventually caught on to this evolution scenario, but did so much later than the ECMWF.

The surface cyclone reached the NC/VA border by early Wednesday morning. Strong isentropic lift on the 285/290K surface ahead of the deepening and intensifying cyclone caused a few very impressive mesoscale snow bands to impact the lower Hudson Valley/NYC metro/Long Island prior to daybreak. This heavy snow moved into the southern portion of the ALY forecast area in the late morning...and then into the Capital Region...Berkshires...northern Catskills...and southern VT by the early afternoon. Eventually the band(s) moved into the northern tier of the forecast area by the late afternoon (Southern Adirondacks...Lake George and Northern Saratoga Region). Numerous Special Weather Statements were issued to cover the snow band as it move from south to north across the forecast area. Eventually this band pivoted to the north to northwest by the late afternoon and early evening. Snow rates were 1-2+ inches per hour with the main snow band. CSTAR research has shown these west to east oriented snow bands can form well in advance of the surface cyclone and its associated closed low in a strong warm advection regime. A pronounced low level baroclinic zone near the right entrance region of 125+kt jet streak helped enhanced the snow bands in the snow shield.

As the coastal wave reached the Delmarva Region, a strong dry slot moved up the coast and poked in the southern half of the region. The precipitation lightened in intensity from the Capital Region south and east. This allowed for some light mixed precipitation (sleet and freezing rain), and even for a brief pause prior to deformation snowfall and mesoscale snow band arriving.

The late afternoon HRRR runs and the NAM/GFS/ECMWF all indicated the heaviest snowfall would impact the majority of the forecast area between 9 pm and 6 am. An intense mesoscale snow band would impact eastern NY and western New England as the towards midnight as the 700/500 hPa closed circulation moved over eastern PA, southern NY and western New England. A secondary upper level jet streak would move close to the region...and it would be under the left exit region of a 125-140 kt jet streak. Lapse rates would be steepening, and tremendous upward vertical motion would be generated from the differential thermal and vorticity advections. The latest CSTAR research hinted at the development of a N-S oriented intense mesoscale snow band tapping tremendous of Atlantic moisture. A pivoting snow band would form. Cross-section down the Hudson River Valley from the NAM/GFS showed deep upward vertical motion through the troposphere with steeply sloped frontogenesis beneath an area of -EPV. The dendritic growth zone was coincident with the strong omega, strong FGEN, and -EPV. Some snow thunder was also possible in the moist symmetrically unstable environment. Snow rates of 3+ inches per hour occurred (i.e. 1 am to 2 am at KALB). Mixed precipitation of sleet reached the Capital Region and Southern Vermont before turning heavy snow. We even had a brief shower of freezing rain at the NWS office in Albany.

The strong upward motion associated with closed off 500 hPa low produced lightning thunder up and down the east coast from the VA towards southern NY. Several reports (through social media) of lightning/thunder accompanied the heaviest precipitation of mixed sleet/snow as it moved across the mid-Hudson Valley during the late evening hours. It was put into the Special Weather Statements that thunder/lightning could accompany the heavy precipitation as far north as the Capital Region, but the thunder didn’t quite make it up this far. However, several cellular-looking convectively driven snow showers were ahead of the main batch of precipitation associated with the 500 hPa low. These contained heavily rimed snowflakes, due to the strong upward motion. It's possible these were mistaken for sleet in the Capital Region, as the true warm nose never made it quite this far north and the water-coated snowflakes appeared large and icy such as sleet pellets. This, along with some freezing drizzle due to the absence of cold cloud tops in the break between precipitation rounds, could be the reason why a brief period of mixed precipitation was reported as far north as Albany. This precipitation, though, had little effect on the overall storm total QPF and snow amounts would be nearly identical even without these phenomena occurring.

Diagnostics for not only the existence of potential banding, but also characterization of banding was determined a good 24 to 36 hours before the event…mentioning the potential for 3+ inch per hour snowfall rates for Thursday night well in advance.

2. What was learned from this event:

Miller Type A storms can be complex to forecast and can result in very high snowfall totals. Several pieces of energy, both in the northern and southern stream, came together to form this singificant Nor'Easter. With the forecast models having a difficult time resolving the main features 2 to 3 days before the event.  This storm was not as certain as the Feb. 4-5 event was two to three days prior.

The QPF from most of the models (excluding the European which was the overall model of choice) was very poor. The QPF from NAM 84 hours out actually had the storm partially correctly forecasted. Later runs kept most if not all of the QPF south and east of Albany, and some runs even had light amounts to the south and east of Albany. The disturbing thing is that the NAM was consistent with this thinking until about 24 hours prior to the event. The GFS did not do much better. That model had the storm identified out around Day 10. Later runs took it out to sea with barely a flake in our entire country warning area. This trend continued until about three days before it "trended" a little closer to the coast but still with a light QPF, especially Albany north and west. The Canadian Model was close to the GFS, again with surprising low QPF amounts, especially Albany north and west before finally ramping up amounts one cycle before the GFS.

It took the 12Z NAM on Feb. 13 and 18z GFS on the same day, to finally get the second portion of the storm right, the "snow bomb" that ultimately led to the huge snowfall amounts Thursday night. Again, the ECMWF had this storm relatively well-resolved several days out although initially it appeared to have it for the "wrong" reasons. The ECMWF showed the heavy precipitation burst coming AFTER the upper low past well to our north and east along with the surface low, which was not necessarily intuitive. As it turned out, later runs, and of course reality, indicated the upper low did not reach our region until after midnight, and at that point "captured" the surface low near Cape Cod. It made much more sense why we got the snow burst at that point.

Also, no model really forecasted the extreme and unusual "dry slot" that cut snow to almost nothing for more than 6 hours. Had snow continued throughout the storm we would have been looking at a true "blockbuster" storm.

A possible reason why the ECMWF did so well with this event could be with the latest upgrade in its model resolution that was done during the summer/fall of 2013. The ECMWF did miss some key features such as the dry slot.  However the ECMWF was the most consistent in shunting the brunt of the dry slot south of the CWA due to the tropopause fold.  Both the horizontal and vertical resolutions were increased, allowing for better resolving of key synoptic features utilizing a global-scale model. While not perfect, the low-resolution version of the ECMWF does expand across portions of the Northern Pacific, which can at a minimum provide a rough idea of potential model-resolved upper-level features developing across the Northern Pacific before they eventually move onshore the western U.S, giving a qualitative approach in addition to analyzing satellite water vapor loops of shortwaves, etc. The increase in vertical resolution in November 2013 changed the vertical levels from 91 to 137, which allows synoptic scale upper-level features to be better resolved. This allows upper-level features, particularly model-derived Potential Vorticity fields to be better resolved, as the increase in vertical resolution allows upper-level potential temperature and vorticity fields to be better resolved theoretically increasing the accuracy of these Potential Vorticity traces. Also, the ECMWF had an upgrade in its ocean model, which is used heavily in the boundary conditions and assimilation scheme. Combining all of these performance upgrades together can explain why in big synoptic events such as this Nor'easter why the ECMWF is able to handle the evolution and timing better than other models, as well as pick up on potential storms such as this one so far out ahead in time, due to the increases in model resolution and physics being able to resolve features out across the Northern Pacific (where often times a lot of potent northern stream Potential Vorticity anomalies originate) as a better first-guess than waiting for other models such as the GFS to only begin to resolve synoptic features once they move onshore (being resolved by upper air data). While the ECMWF will never be perfect as each storm is different, these model improvements can increase forecaster confidence/awareness of the potential for significant storm systems, especially when being shown consistently from model run to model run 3-5 days out. We will have to see in the coming months how the NCEP model upgrade will perform as well, and if they can follow the increasing performance trend in medium-range prediction that the ECMWF has been able to achieve.

Interesting too, the snow to liquid ration was close to 20:1 with the first band of heavy snow that fell midday Thursday. The radar signatures were more yellow with this band (well over 30 DBZ) and snow was indeed very "fluffy" with large aggregated dendrites. However, the snow to liquid ratio with the second portion of the storm (Thursday night) was closer to 10:1, much heavier, more like cement. The flakes were not quite as large, but there were plenty of them and the visibility was even lower. The difference in snow to liquid ratio was not intuitive, since the second portion of the snowfall was when the column was cooling and more dynamical and should have yield much higher snow to liquid ratios and lower QPF.

This was quite an interesting storm given the impressive dry slot, and even more impressive heavy snow bomb accompanying the upper level portion of the storm. This storm had similarities to the March 7th-8th 1996 that had a once two punch similar to this one. Ironically, we ended up with 14 inches of snowfall with this storm was well!

 

Forecast models - Initialized 10 February - Note the inconsistencies between all sets of guidance leading to

lower levels of predictability in potential sensible weather expected associated with the upcoming storm.

 

Above:  4-panel of 500 hPa heights and vortiticy initialized at 12Z 10 February from the GFS (upper left), NAM (upper right), ECMWF (lower left) and GFSEnsemble mean (lower right) and valid at 0000 UTC 14 February.

Above:  4-panel of 850 hPa winds and isotachs initialized at 12Z 10 February from the GFS (upper left), NAM (upper right), ECMWF (lower left) and GFSEnsemble mean (lower right) and valid at 0000 UTC 14 February.

Above:  4-panel of MSLP initialized at 12Z 10 February from the GFS (upper left), NAM (upper right), ECMWF (lower left) and GFSEnsemble mean and members (lower right) and valid at 0000 UTC 14 February.

Above:  4-panel of quantitative precipitation forecast initialized at 12Z 10 February from the GFS (upper left), NAM (upper right), ECMWF (lower left) and GFSEnsemble probability of 1 inch of liquid equivalent precipitation in 24 hours (lower right) and valid at 0000 UTC 14 February.

 

Forecast models - Initialized 11 February - Note the inconsistencies between all sets of guidance leading to

lower levels of predictability in potential sensible weather expected associated with the upcoming storm.

However, there is less spread in the guidance.  So, predictability is slowly increasing.

 

Above:  4-panel of 500 hPa heights and vortiticy initialized at 12Z 11 February from the GFS (upper left), NAM (upper right), ECMWF (lower left) and GFSEnsemble mean (lower right) and valid at 0000 UTC 14 February.

Above:  4-panel of 850 hPa winds and isotachs initialized at 12Z 11 February from the GFS (upper left), NAM (upper right), and 300 hPa winds and isotachs from the GFS (lower left) and NAM (lower right) and valid at 0000 UTC 14 February.

Above:  4-panel of MSLP initialized at 12Z 11 February from the GFS (upper left), NAM (upper right), ECMWF (lower left) and GFSEnsemble mean and members (lower right) and valid at 0000 UTC 14 February.

Above:  Temperatures at 850 hPa from the GFS and NAM initialized at 1200 UTC 11 February and valid at 0000 UTC 14 February.

Above:  4-panel of pressure and omega (pressure advection) at 290K initialized at 12Z 11 February from the GFS (upper left), NAM (upper right), and 850-700 hPa QG frontogenesis from the GFS (lower left) and NAM (lower right) and valid at 0000 UTC 14 February.

Above:  4-panel of quantitative precipitation forecast initialized at 12Z 11 February from the GFS (upper left), NAM (upper right), ECMWF (lower left) and GFSEnsemble probability of 1 inch of liquid equivalent precipitation in 24 hours (lower right) and valid at 0000 UTC 14 February.

Above:  4-panel of SREF probability for 4 inches of snow in 12 hours (upper left), 8 inches of snow in 12 hours (upper right), 850 hP temperature (lower left) and 12 hour snow accumulation (lower right) initialized at 09Z 11 February and valid 0000 UTC 14 February.

Above:  4-panel of probabilities for .60" liquid equivalent precipitation in 24 hours (upper left) and 1.00" (upper right) from the GFS Ensemble initialized at 12Z 11 February and valid 0000 UTC February 14 and probabilities for .50" liquid equivalent precipitation in 24 hours (lower left) and 1.00" (lower right) from the SREF initialized at 09Z 11 February and valid 0000 UTC February 14.

Above:  Time sections from the NAM initialized at 1200 UTC 11 February for Albany, NY (left), Glens Falls, NY (center) and Poughkeepsie, NY (right).

Above:  Time sections from the GFS initialized at 1200 UTC 11 February for Albany, NY (left), Glens Falls, NY (right).

 

Forecast models - Initialized 12 February - Note the continued decrease inconsistencies between all sets of

guidance resulting in small but incremental increases in predictability in potential sensible weather

expected associated with the upcoming storm.

 

Above:  4-panel of 500 hPa heights and vortiticy initialized at 12Z 12 February from the GFS (upper left), NAM (upper right), ECMWF (lower left) and GFSEnsemble mean (lower right) and valid at 1200 UTC 14 February.

Above:  4-panel of 850 hPa winds and isotachs initialized at 12Z 12 February from the GFS (upper left), NAM (upper right), and 300 hPa winds and isotachs from the GFS (lower left) and NAM (lower right) and valid at 0000 UTC 14 February.

Above:  4-panel of MSLP initialized at 12Z 12 February from the GFS (upper left), NAM (upper right), ECMWF (lower left) and GFSEnsemble mean and members (lower right) and valid at 0600 UTC 14 February.

Above:  Temperatures at 850 hPa from the GFS and NAM initialized at 1200 UTC 12 February and valid at 0000 UTC 14 February.

Above:  4-panel of pressure and omega (pressure advection) at 290K initialized at 12Z 12 February from the GFS (upper left), NAM (upper right), and 850-700 hPa QG frontogenesis from the GFS (lower left) and NAM (lower right) and valid at 0000 UTC 14 February.

Above:  4-panel of pressure and omega (pressure advection) at 290K initialized at 12Z 12 February from the GFS (upper left), NAM (upper right), and 850-700 hPa QG frontogenesis from the GFS (lower left) and NAM (lower right) and valid at 0600 UTC 14 February.

Above:  4-panel of quantitative precipitation forecast initialized at 12Z 11 February from the GFS (upper left), NAM (upper right), ECMWF (lower left) and GFSEnsemble probability of 1 inch of liquid equivalent precipitation in 24 hours (lower right) and valid at 1200 UTC 14 February.

Above:  4-panel of probabilities for .60" liquid equivalent precipitation in 24 hours (upper left) and 1.00" (upper right) from the GFS Ensemble initialized at 12Z 12 February and valid 0600 UTC February 14 and probabilities for .50" liquid equivalent precipitation in 24 hours (lower left) and 1.00" (lower right) from the SREF initialized at 15Z 12 February and valid 0600 UTC February 14.

Above:  4-panel of SREF probability for 4 inches of snow in 12 hours (upper left), 8 inches of snow in 12 hours (upper right), 850 hP temperature (lower left) and 12 hour snow accumulation (lower right) initialized at 15Z 11 February and valid 0300 UTC 14 February.

Above:  Cross section of frontogenesis and vertical velocity from the GFS initialized 1200 UTC 12 February and valid 0600 UTC 14 February.

Above:  MSLP and Quasigeostrophic frontogenesis from the GFS initialized 1200 UTC 12 February and valid 0600 UTC 14 February.

Above:  Cross section of temperature and the dendritic growth zone shown as a black band, from the GFS initialized 1200 UTC 12 February and valid 0600 UTC 14 February.

Forecast models - Ensembles - Notice the GEFS consistently predicted less liquid equivalent precipitation

than the SREF but both ensembles showed a rapid increase in liquid equivalent precipitation in the last 2

initializations.

Above:  Loops of plumes for Albany, NY beginning with the 1200 UTC 10 February GEFS (left) and 2100 UTC 10 February SREF (right).

Above:  Loops of plumes for Binghamton, NY beginning with the 1200 UTC 10 February GEFS (left) and 2100 UTC 10 February SREF (right).

Above:  Loops of plumes for Monticello, NY beginning with the 1200 UTC 10 February GEFS (left) and 2100 UTC 10 February SREF (right).

Above:  Loops of plumes for Burlington, VT beginning with the 1200 UTC 10 February GEFS (left) and 2100 UTC 10 February SREF (right).

Above:  Loop of plume for Hartford, CT beginning with the 2100 UTC 10 February SREF.

 

Above:  Plumes for Albany, NY (upper left), Glens Falls, NY (upper right), Poughkeepsie, NY (lower left) and Pittsfield, MA (lower right) from the 0300 UTC 12 February SREF.

 

Forecast models - Short Range Mesoscale models - HRRR - Note the enhanced reflectivity band that is

predicted to affect some part of the eastern NY and western New England area.

Above:  Loops of selected runs of the HRRR 1 km AGL reflectivity valid at 0000 UTC 14 February (left) and 0600 UTC 14 February (right).

Above:  Loop of column maximum reflectivity from the 1900 UTC 13 February HRRR.

Above:  1 km AGL reflectivity from the 1200 UTC 12 February ARW HRRR valid 0600 UTC 14 February (left) and the 1200 UTC 12 February NMM HRRR valid 0600 UTC 14 February (right).

Above:  1 km AGL reflectivity from the 1200 UTC 12 February S.U.N.Y. Stonybrook GFS valid 0200 UTC 14 February (left) and the 1200 UTC 12 February S.U.N.Y. Stonybrook NAM valid 0600 UTC 14 February (right).

 

WPC Winter Weather Desk Forecasts - Note the increase in snowfall predictions over the northeastern U.S.

as the onset of the storm approached.

 

Above:  Forecasts from the Weather Prediction Center  of 25th percentile (left), 50th percentile (center) and 75th percentile (right) snowfall issued 1200 UTC 12 February valid 0000 UTC 15 February.

Above:  Forecasts from the Weather Prediction Center  of 50th percentile snowfall issued 1200 UTC 13 February valid 1200 UTC 14 February.

Above:  Forecasts from the Weather Prediction Center  of 24 hour probability for ≥ 8" snowfall issued 1200 UTC 13 February valid 1200 UTC 14 February.

Above:  Forecasts from the Weather Prediction Center  of total snowfall issued 0619 UTC 12 February (left) and 0602 UTC 13 February (right).

 

The peak of the storm - the analyses and anomalies indicated a classic historical heavy snowstorm

signature.

Above:  MSLP and anomalies initialized from the 1200 UTC 14 February GEFS (left) and 0900 UTC 14 February SREF (right).

Above:  Winds at 850 hPa and anomalies initialized from the 1200 UTC 14 February GEFS (left) and 0900 UTC 14 February SREF (right).

Above:  Winds at 250 hPa and anomalies initialized from the 1200 UTC 14 February GEFS (left) and 0900 UTC 14 February SREF (right).

Data displays of the storm in action

Above:  Water vapor Satellite Imagery.

Above:  Infrared Satellite Imagery.

Above:  Visible Satellite Imagery on 13 February (left) and 14 February (right).

Above:  KENX radar reflectivity loops on 13 February (left) and 14 February (right).

Above:  Regional radar reflectivity loop from 13-14 February (Courtesy of The College of DuPage).

 

Above:  Nesis ranking of 3.