The St. Patricks Day Snowstorm

(Click on thumbnails for larger images)


Meteorological Conditions:


            This storm occurred while the NAO was positive, which is unusual for a big northeastern snowstorm (Figure 1a).  However, the PNA was slightly positive (Figure 1b), which supported a digging upper trough over the eastern U.S.  The MJO was in a neutral phase (Figure 1c), which was the case for the Martin Luther King Day Storm and the Valentine’s Day Storm.  The MJO had been trending toward phase 8 and 1 but went neutral, possibly because of the lingering effects of El Nino on the atmosphere in the southern stream.  However, the El Nino was trending toward neutral in mid March.  Phase 8 and 1 of the MJO is characterized by general low pressure along the east coast of North America.


Figure 1.  Displays of a)NAO and b) PNA for December through March, and c) the MJO February (green)and March (blue).  Note the positive phase of the NAO in mid March, and the negative phase of the PNA.  Also note the MJO was neutral after entering phase 6 prior to the Martin Luther King Valentine’s Day and St. Patrick’s Day storms.

The potential for a significant storm became evident 3 to 5 days before the storm, with increasing confidence by 13 March, as our region was beginning to enjoy the first of several days featuring springtime warmth in the 50s and 60s.  There was considerable disagreement from long range guidance of the exact track of the storm, with many solutions in ensembles and in the ECMWF, suggesting a storm track far enough east, that only Long Island and New England would be affected.  The broad range of solutions was very evident in the 13 March MREF Plumes (Figs. 2a-b) with very large spreads in QPF, precipitation type and timing of precipitation from each of the ensemble members.   Probability of 0.50” (Figs. 3a-c) from the MREF illustrates two general periods of precipitation, one with the cold front Wednesday night and Thursday, and another more significant period of precipitation late Friday into Saturday, with the highest probabilities over New England.  The 13 March SREF Plumes (Figs. 4a-b) and SREF probability of 0.30” (Figs. 5a-b) were very similar to the MREF Plumes and probabilities.


Figure 2.  Plume diagrams for Albany, NY from a) 06Z 13 March MREF, and b) 12Z 13 March MREF.  Note the broad range of precipitation types, timing and amounts for each ensemble member, suggesting low confidence forecasts.



Figure 3.  MREF probability of 0.50” for 24 hours a) 06Z 13 March valid 18Z 14 March through 18Z 15 March, b) 12Z 13 March valid 18Z 14 March through 18Z 15 March, and c) same as b except valid 12Z 16 March through 12Z 17 March.  Note the large spread in each suggesting some masking of potentially higher probabilities of QPF, and the suggestion of generally 2 periods of organized precipitation.


Figure 4.  SREF plume diagrams for Albany, NY from a) 03Z 13 March and b) 09Z 13 March.  Note the large spread in timing and amounts of precipitation through the period suggesting a low confidence forecast.



Figure 5.  SREF probability of 0.30” in 12 hours from a) 03Z 13 March valid 15Z 15 March through 03Z 16 March, and b) 09Z 13 March valid 18Z 15 March through 06Z 16 March.  Note the spread, possibly masking some of the potentially higher probabilities.  These probabilities were associated with the cold frontal passage prior to the St. Patrick’s Day storm.


However, by 15 March, the forecast models and ensembles had converged on a solution that southern stream energy along the Gulf Coast States would phase with digging northern stream energy tracking out of Canada and the western Great Lakes, combined with strong surface high pressure over southeastern Canada, anchoring the low-level cold air over our region (Figs 6a-b).  This evolution was evident in the MREF and SREF mean and spread for MSLP (Figs 7a-b), 850 Mb temperatures (Figs. 8a-b), low-level winds (Figs 9a-b), Plumes Figs 10a-d) and probabilities for 1.00” (Figs. 11a-b).  The U and V winds at 850 Mb were considerably more anomalous in the operational NAM and GFS (Figs. 12a-b), but even the SREF and MREF showed significant wind anomalies despite some smoothing by the spread of each 15 member ensemble. 


Figure 6.  Daily weather maps valid at a)12Z 15 March and b) 12Z 16 March.  Note the strong surface high pressure center building east through southern Canada, and the associated cold air driving south into the northeastern U.S.   Also note the low pressure organizing over the southeastern U.S., which would become the St. Patrick’s Day storm.


Figure 7.  MSLP, anomalies and spread for a) 09Z 15 March SREF, and b) 12Z 15 March MREF.  Note the relatively small spread and decent consensus between the mean and spreads of the SREF and MREF.



Figure 8.  Spread, anomalies and mean 850 Mb temperatures for a) 09Z 15 March SREF, and b) 12Z 15 March MREF.  Note the relatively little spread, and the extremely tight thermal gradient in the mean across southern NY and southern New England.



Figure 9.  Wind barbs and anomalies (shaded) for a) 09Z 15 March SREF and b) 12Z 15 March MREF.  Note the U winds at 3 to 4 SD below normal and V winds at 3 to 4 SD above normal.




Figure 10.  Plume diagrams for Albany, NY from a) 03Z 15 March SREF, b) 09Z 15 March SREF, c) 00Z 15 March MREF and d) 12Z 15 March MREF.  Note the improved consensus from each ensemble and members of each ensemble with the timing of the periods of precipitation.  Also note the continued large spread in forecasted amounts of precipitation and precipitation types, but even the driest ensemble members suggested significant QPF.





Figure 11.  Probability of 1.00” in 36 hours from a) 09Z 15 March SREF valid 06Z 16 March through 18Z 17 march, and b) 12Z 15 March MREF valid 00Z 16 March through 12Z 17 March.  Note the spatial consistency in each ensemble resulting in above normal forecast confidence.



Figure 12.  Wind barbs and anomalies at 850 Mb (shaded) from a) 12Z 15 March NAM and b) 12Z 15 March GFS.  Note the magnitude of the U and V wind anomalies are greater than in the SREF and MREF displays in figure 8. 


Precipitable water anomaly forecasts, plume diagrams and 1” QPF probabilities pointed to extreme precipitation potential, as long as the full phasing occurred as forecasted (Figs. 10-11).  However, the forecasted QPF was less than for the Valentine’s Day Storm.  Rain potential in New England, however, was potentially over 2 inches.  So, the maximum QPF was not forecasted to be within the snowfall zone, but it would be close.  The forecasted thermal gradient forecasted at 850 Mb was extreme (Fig. 8), suggesting frontogenesis consistent with past storms that produced large areas of 18” or more.  Analyzing frontogenesis at any level or layer between 925 mb and 700 mb is often very helpful.

Based on figures 9 and 12, low-level wind anomaly forecasts in the MREF, SREF and operational models just barely exceeded the threshold for historical snowstorms, peaking between 4 and 5 SD below normal.  The V wind anomalies at 850 Mb were nearly 4 SD above normal, signaling the strong warm advection expected associated with the southern stream system merging with the northern stream system.  Upper-level U wind anomalies were below the threshold for a historical storm, peaking between 1.5 and 2 SD below normal, so the storm was not expected to be of a long enough duration for widespread 18” or more (Figs 13a-d).  Upper-level jet structure suggested considerable upper-level divergence in the right-entrance region of the upper jet over the northeastern U.S., with 250 Mb V wind anomalies over 4 SD above normal.  So, in summary, the wind anomalies didn’t quite point to a widespread 18” or more, but the extreme frontogenesis suggested some areas of 18”-24”, and most areas would see 8”-16”.



Figure 13.  Wind barbs and U and V anomalies at 250 Mb (shaded) from a) 09Z 15 March SREF valid 18Z 17 March, b) 12Z 15 March MREF valid 18Z 17 March, c) 12Z 15 March NAM valid 18Z 17 March, and d) 250 U wind anomalies from the 12Z 15 march GFS valid 18Z 17 March.  Note the SREF and MREF not reaching the -2.5 SD U wind anomaly threshold for a prolonged storm, but V wind anomalies were 4 or more SD above normal, with our region in the right entrance region of the upper jet.  Also note the NAM and GFS U and V wind anomalies were of higher magnitude than the SREF and MREF.


One more factor to consider was the thermal profile within the zone of maximum vertical motion.  The consistent message from all sources of guidance was that the temperatures though the saturated layer with maximum vertical motion would mainly be at -12C or warmer.  So snow rates and ratios would not be extreme, and not be supportive of snow amounts observed in the Valentine’s Day Storm.

As stated earlier, by 15 March, most short range guidance reached enough of a consensus on the storm track and thermal profiles across the region for Winter Storm Watches to be issued.  By early morning on 16 March, Heavy Snow and Winter Storm Warnings were issued, with the prospects for 8”-16” over a large area, and nearly 2 feet within any mesoscale bands that could form.  A mix with sleet and freezing rain was possible for areas south and east of the Capital District, including the southern Catskills, Berkshires and northwestern CT, but not enough of a mix to cut snow amounts below Warning criteria. 


What happened?


            The storm evolved very much as expected with the northern and southern streams phasing, along with the surface high in Canada providing low-level cold air to our region, as evidenced in satellite imagery (Figs. 14a-g).  The snow began over southern areas before noon, the Capital District and Berkshires around noon, southern Vermont by mid afternoon, and the Adirondacks and Schoharie Valley during the late afternoon on Friday 16 March (Fig. 15).  One mesoscale band developed from the southern Catskills through northwestern CT and the southern Berkshires by late afternoon on Friday 16 March (Fig. 16), and slowly shifted north and west through the evening (Figs. 17 a-b), but not quite to the Capital District.  Sleet mixed with the snow on the southeastern periphery of the enhanced precipitation band. 




Figure 14.  Water vapor satellite imagery at a) 23Z 15 March, b) 23Z 16 March and c)16Z 17 March, infrared satellite imagery from d) 23Z 16 March and e) 16Z 17 March, and visible satellite imagery with lightning overlay from 19Z 16 March and 16Z 17 March.  Note the system along the Gulf Coast that phased with northern stream energy.  Also note the classic baroclinic leaf.


Figure 15.  Northeastern radar reflectivity mosaic at 17Z 16 March, showing the onset of precipitation in southern NY, and New England.


Figure 16.  Northeastern radar reflectivity mosaic at 00Z 17 March, showing the mesoscale band in southern NY, and New England.



Figure 17.  Northeastern radar reflectivity mosaics at a) 03Z 17 March, and b) 05Z 17 March, showing the mesoscale band in southern NY, and New England.



The snow ended before sunrise on Saturday 17 March in most locations, except in southern VT and parts of the Berkshires, where the precipitation ended around mid morning.  Southern VT and the Berkshires did see a few hours of light freezing rain and freezing drizzle before the precipitation ended.  Generally 7” to 14” snow and sleet mixed fell in the southern Catskills through Poughkeepsie, northwestern CT, the Berkshires and southern VT.  Areas from the northern Catskills through the Capital District and Saratoga area saw 1’ to 2’, with the 2’ amounts in the northern Catskills.  The Capital District received 13” to 16”.  The Adirondacks received 6” to 10” with around 10” in the Lake George area.  See figure 18 for the snowfall map.


Figure 18.  Observed snowfall for St. Patrick’s Day storm.


            The warm advection precipitation became very detached from the upper deformation area, as the phased upper energy produced a cut-off low pressure center over the eastern Great Lakes and Western NY (Figs. 14 c, e and g).  Another area of lighter snows affected NY, the Berkshires and southern VT during the night of Saturday 17 March into early Sunday 18 March.


What was learned from this event?


Big snowstorms can occur in our region when the NAO is positive.  This was the third major storm this season during a neutral MJO, namely the Martin Luther King Day Ice Storm, the Valentine’s Day Storm, and the St. Patrick’s Day Storm (Fig. 19).  A positive PNA was in place for the three major storms this season as well.  The MJO was trending toward phase 8 and 1 prior to each storm, which is favorable for low pressure along the east coast of North America.  However, the MJO was just weak enough to go neutral as it was about to enter a weak phase 8.  This was the third time the MJO trended this way since January, possibly due to lingering effects of the El Nino.





Figure 19.  MJO phases for January (red), February (green) and March (blue).  Note the MJO transitioned through phases 2 through 6 before going neutral, possibly because of a lingering El Nino.  The Martin Luther King storm, Valentine’s Day Storm and St. Patrick’s Day storm occurred as the MJO was in a neutral phase, but near a weak phase 7/8/1, phases that are favorable for low pressure in eastern North America.


The long range guidance could not resolve this system very well 3 to 5 days prior to the storm, although pattern recognition did help add value to the HPC extended grids.  Even 1 to 2 days before the storm, issues such as the location of the precipitation transition zone and QPF were not well-resolved.  The 12Z model runs on Wed Mar 14 were the first to latch on to explosive coastal cyclogenesis, especially the GFS/UKMET. The main reason for the big change was the upper level trough across the western Great Lakes was forecast to negatively tilt and actually close off at 500 hPa, allowing for a more intense low pressure center and a track closer to the coast.  Even though model forecasts have improved over the last few years, this is a good example of the potential for uncertainty as little as 48 hours before an event.  Independent of the location of the transition zone and QPF, even the ensemble guidance with the least QPF suggested warning-level snow potential.  It was just the snow ranges and timing of onset and end of precipitation that had to be determined. 

The lack of a -12˚C to -18˚C region suggested any mesoscale banding would not feature extreme snowfall amounts like what was observed in the Valentine’s Day Storm.  In fact, the tight thermal gradient suggested more of a potential of a mix of sleet within any enhanced band. 

The wind anomalies at 850 Mb and 250 Mb suggested a quick hitting period of extreme precipitation, not as intense and not as prolonged as the Valentine’s Day Storm, and the snowfall forecasts and observations were consistent with this reasoning.  The precipitable water anomalies and QPF suggested the heaviest precipitation would be east of the snowfall zone, and within the snowfall zone, less liquid equivalent precipitation than the Valentine’s Day Storm would occur.  So the guidance was in very good agreement 12 to 24 hours before the storm.  Observed upper and low-level wind anomalies (Figs. 20a-d), 850 Mb temperatures (Figs. 21), frontogenesis (Figs. 22a-b) and surface features (Figs. 23) were very similar to the forecasts from 15 March.



Figure 20.  Initial wind barbs and U and V wind anomalies from a) 03Z 17 March SREF at 850 Mb, b) 00Z 17 March MREF at 850 Mb, c) 00Z 17 March NAM at 850 Mb, and d) 00Z 17 March NAM at 250 Mb.  Note the 850 Mb U wind anomalies almost 5 SD below normal in the ensembles and operational NAM, suggesting extreme convergence, frontogenesis and precipitation production.  Also note the 250 Mb U wind anomaly between 1.5 and 2 SD below normal, suggesting the storm was not very well cut off from the steering flow, and a long duration storm was not expected.


Figure 21.  Initial temperatures at 850 Mb from the 00Z 17 March GFS.  Note the extreme temperature gradient across southern NY and New England.



Figure 22.  Initial 00Z 17 March GFS a) frontogenesis at 850 Mb, and b) cross section from south of Long Island to northern NY, depicting frontogenesis, EPV and ageostrophic winds.  Note the extreme frontogenesis in central and southern New England.  Also note the upslope nature to the frontogenesis from south to north in the cross section.  Note there wan not much if any negative EPV right above the frontogenesis, but ageostrophic winds suggested strong vertical motion.


Figure 23.  MSLP from MSAS at 12Z 17 March.  Note the track of the surface low across Cape Cod, which is a favored track for heavy snow in Albany.  Also note the presence of a strong surface high pressure center in Canada, that provided the low level cold air.

The snow to liquid ratios were 8-10:1, even with surface temperatures over much of the region between 15˚F and 25˚F.  As stated before, there was no layer between -12˚C and -18˚C within the region of maximum vertical motion. There was one primary enhanced band, not multiple bands, this could be researched.