Thanksgiving Weekend Snowstorm (Click on images for larger images)
(Click on images for larger images)
Miller type A, southern stream, storm tapping gulf moisture developed along the gulf coast and tracked northeast along a frontal boundary eventually passing around 50 to 60 nautical miles southeast of Cape Cod around 00Z/November 27th. The situation was complicated by an approaching northern stream short wave.
The ECMWF consistently provided strong indications that a widespread major snowstorm would impact the region during the Wednesday-Thursday time frame as early as the prior weekend (Nov 22-23). The other global models, particularly the GFS and GEFS, proved quite unsuccessful at detecting this possible storm a few days in advance, and actually distracted the forecast process by suggesting the storm would be a far miss until closer to the actual event. It seems that the ECMWF may be more accurate in long-term predicting of southern- and northern-stream interactions, and development, track and evolution of Miller-A type east coast storm systems.
The ECMWF resolved the storm in the medium range, but it still struggled with the exact track and placement of the system up to 3 days in advance. Some runs of the ECMWF had the storm phasing with the northern stream energy and moving over NYC and central New England. This would have been a rain storm scenario for much of central and eastern New England, and a mixed precipitation scenario for much of western New England, and locations from the Capital Region south and east. This run was the 12Z/NOV 23 cycle. Also, the QPF from the ECMWF still has a lot to be desired, since it had a solid 1-1.5+ inches of QPF over most of the forecast area up until 00Z/NOV 26. This was a bit on the heavy side from this model. Again, the storm was there in the medium range but details were still uncertain. The GFS had a slightly further east track in Days 3 to 5, but its track for the storm inside of 3 days was not bad.
A possible reason that the ECMWF may have an edge, especially in hinting at possible coastal development in the medium-range, is that it samples oncoming waves off the Pacific much better than the GFS due to the way that the ECMWF assimilation scheme is currently set up. This may allow for better sampling of upper-level waves/energy moving onshore the western CONUS farther out in time which may help explain why the ECMWF seems to "catch-on" to coastal systems much earlier than the GFS.
All sources of guidance were not in good agreement as to the northwestward extent of the higher precipitation amounts even within 24-36 hours prior to the storm.
Leading up to the event, the ECMWF was consistently on the westward side of the model guidance envelope of potential tracks, as it hinted at at least some influence of northern stream energy interacting with the more dominant southern stream. The GFS (and most GEFS members) leading up to the event were consistently on the eastward side of the model guidance envelope, which favored a track further off the coast that would have less of an impact across our area than the ECMWF scenario. A possible explanation for why the GFS was on the east side of the track envelope is that it maintained almost a purely southern stream system, in which the coastal low formed far downstream of a positively-tilted trough and raced off to the northeast without any northern stream energy "capturing" it and tugging it closer to the coast.
However, within about 48 hours prior to storm onset, about a third of the GEFS members did hint at a more westward track of the coastal low bringing it closer to the coast. While deterministic models like the ECMWF are always great to look at, probabilistic guidance like the GEFS can offer clues towards a possible trending solution. The NAM was consistently far east with its track until just prior to the event when it finally caught up to the rest of the model guidance. The NAM seems to becoming increasingly apparent that it struggles outside of 36 hours, especially with storm track and QPF. Within 36 hours, the GFS caught on to the ECMWF, and began trending westward with the track with each successive model run leading up to the event.
If the ECMWF was resolved the storm better 3 to 5 days in advance, but the GFS did very well inside a couple of days. It actually closed the 700 hPa low over upstate NY in the WED afternoon into the evening which helped capture some of the moisture for the heavy snow totals from the Capital Region north and east.
Our region was in the right entrance region of a 300 hPa jet streak with magnitude of 180-200KT in Canada, which aided in upper divergence over the region and low level warm and moisture advection.
Preceding the event, GFS/NAM models indicated mesoscale banding would be possible looking at cross-sections due to strongly sloped frontogenesis and strong omega within the dendritic zone. The highest confidence for banding was for areas south and east of the Capital District based on where the best frontogenesis and omega intersecting the snow growth region were forecast. From Albany north and west, it appeared sloped frontogenesis would be present, but of weaker magnitude and displaced fairly high aloft (at least 600-500 mb and higher). Also, the best omega was expected to be south and east of Albany (although not too far), with the snow growth region also unusually elevated. This data gave enough confidence to mention possible 1-2" per hour snowfall rates for areas south and east of Albany the day before the storm.
The models did NOT do a good job depicting the mesoscale banding that also developed to the north and west of the Capital Region, evident in the radar imagery during the storm and the GIS snowfall analysis map that was created after. Much of the Mohawk Valley and central/southeast Adirondacks had a secondary maximum in snowfall as a result of the enhanced banding in this area, in addition to the other (anticipated) maximum across portions of the Taconics, Berkshires, southern Vermont, and northern Litchfield County. The multi-bands were not captured well by some of the guidance. Snow rates of 1 to 2 inches per hour persisted well north and west of the Capital Region. A lull in the precipitation equated to lower totals to the south and east.
A possible reason for why the mesoscale banding set up further to the north and west (which the models did not capture well at all), was the influence of northern stream energy mentioned at the very top of this section. As the coastal low took shape, it tracked up the eastern seaboard as it was pumped with southern stream energy and moisture. However, water vapor loops Wednesday afternoon/night revealed a rather potent northern stream shortwave diving southward across the central CONUS and eventually into the Tennessee Valley before hooking northeast Thanksgiving into Friday. This shortwave was depicted in the model guidance, but was depicted much weaker and slower than what actually occurred. The acceleration of this shortwave may have been aided by a stronger than expected Pacific jet, in which the ECMWF hinted at several days out in advance of the storm (possibly because it sampled this region better than the rest of the model guidance). Again, another clue as to why the ECMWF hinted at the storm so far out in time and with the farthest westward track.
Since this northern stream shortwave moved faster, and was stronger (due to a stronger than expected Pacific jet), it was able to interact more with the coastal low. This may have allowed for enhanced downstream upper-level diffluence and deformation across our region as it turned the height field aloft. This could have set the stage for the frontogenesis to set up further to the north and west as a more supportive environment for frontogenesis was in place further north and west with a better deformation field than indicated in the model guidance. Another possible reason for why the band set up to the north and west was rather impressive isentropic lift evidenced on the 300K and 305K surfaces, in which the wind field was virtually perpendicular to the pressure. With this setup, isentropic lift was maximized across the region which would have enabled more efficient diabatic processes that could have acted to destabilize the upper-levels and enhance lift. There was also a hint of some EPV within the dendritic growth zone as well which in the presence of the favorable synoptic lift, could have supported a period of CSI, which was noted in the radar imagery Wednesday night with banding of snow showers across the Mohawk Valley/Capital Region resembling convective roll clouds.
The NAM indicated the possibility of a single band or multi bands with cross-sections from the 12Z/26 NOV cycle. The entry describes the banding, but past CSTAR research indicates multi-bands are possible when the quasi geostrophic lift is generated by the differential thickness advection. It would have been interesting in what the Q-vectors looked like with this system.
Frontogenesis in the 700-500 hPa layer may have organized the moisture and vertical motion into snow bands into the western and central Mohawk Valley and southern Adirondacks, further northwest than guidance suggested.
There was consistency in all sources of guidance depicting very strong isentropic lift between 285-295K extending well into eastern and northern NY as well as upper divergence between the upper energy along/off the east coast and the trailing upper energy that tracked through the OH Valley, which may have allowed moisture and upward motion to extend into the western and central Mohawk Valley and southern Adirondacks.
PV analysis for this event offered little support in improving confidence with the variable model guidance solutions. Although the PV analysis did suggest a dominant southern stream system, it was hard to discern any influence of northern stream energy as there was an already anomalous amount of northern stream energy present across the central CONUS from the prior week. This can resemble a giant "ball" of PV that can muddle any finer scale details. While PV analysis can be useful in meteorological analysis, it still must be used in conjunction with other principles, so that a full picture of a forecast can be understood. PV is typically good with big picture concepts, but the devil is in the details which this kind of analysis can lose at times.
a) b) c) d) e)
Above: 4-panel displays from the GFS (upper left), ECMWF (upper right), GFSEnsemble (lower left) and NAM12 (lower right) initialized 12Z 13 November and valid 00Z 27 November for a) 500 hpa Heights, b) MSLP, c) 850 hPa winds, d) 850-500 hPa frontogenesis and e) liquid equivalent QPF and GFSEnsemble probability of 1 inch in 24 hours.
Above: Loops of GFSEnsemble a) 500 hPa heights and anomalies (shaded), b) MSLP and anomalies (shaded) and c) 850 hPa winds and anomalies (shaded) initialized 12Z 26 November.
Above: Loops of SREF a) 500 hPa heights and anomalies (shaded), b) MSLP and anomalies (shaded) and c) 850 hPa winds and anomalies (shaded) initialized 09Z 26 November.
a) b) c) d)
Above: Loops of GFSEnsemble plumes initialized 00Z 24 November for a) Albany, NY, b) Watertown, NY, c) Burlington, VT and d) Monticello, NY.
a) b) c)
Above: Loops of SREF plumes initialized 03Z 24 November for a) Albany, NY, b) Watertown, NY and c) Monticello, NY.
a) b) c) d)
Above: Loops of GFSEnsemble plumes initialized 00Z 25 November for a) Albany, NY, b) Watertown, NY, c) Burlington, VT and d) Monticello, NY.
a) b) c)
Above: Loops of SREF plumes initialized 03Z 25 November for a) Albany, NY, b) Watertown, NY and c) Monticello, NY.
a) b) c) d)
Above: Loops of GFSEnsemble plumes initialized 00Z 26 November for a) Albany, NY, b) Watertown, NY, c) Burlington, VT and d) Monticello, NY.
a) b) c)
Above: Loops of SREF plumes initialized 03Z 26 November for a) Albany, NY, b) Watertown, NY and c) Monticello, NY.
a) b) c)
Above: Inintialized 700 hPa heights and frontogenesis from the a) 18Z 26 November NAM12, b) 18Z 26 November GFS and c) 12Z 26 November ECMWF.
a) b) c)
Above: Weather Prediction Center probabilities for a) 4" of snow, b) 8" of snow and c) 12" of snow issued 2004Z 23 November and valid 00Z 26 November through 00Z 27 November.
a) b) c)
Above: Loops of Weather Prediction Center probabilities for a) 4" of snow, b) 8" of snow and c) 12" valid 00Z 26 November through 00Z 27 November.
a) b) c)
Above: Loops of Weather Prediction Center probabilities for a) 4" of snow, b) 8" of snow and c) 12" valid 12Z 26 November through 12Z 27 November.
Above: Weather Prediction Center 3 day snowfall forecasts issued at a) 0639Z 24 November and b) 1821Z 24 November.
a) b) c)
Above: Weather Prediction Center snowfall forecasts for a) day 2 issued at 0611Z 25 November and b) day 1.5 issued at 1748Z 25 November and c) day 1 issued at 0502Z 26 November.
Above: Upper air soundings at 00Z 27 November for Albany, NY (KALB) and Upton, NY (KOKX).
a) b) c)
Above: Loops of a) water vapor satellite imagery, b) Infrared satellite imagery and c) Visible imagery.
Above: Radar reflectivity loops from a) KENX and b) mosaic of northeastern U.S. radars.