A Case Study of Anomalous Snowfall with an Alberta Clipper

1Department of the Earth Sciences, The College at Brockport, State University of New York, Brockport, NY, USA 2Department of Soil, Environmental and Atmospheric Sciences, University of Missouri, Columbia, MO, USA 3NOAA/NWS Operational Proving Ground, Cooperative Institute for Meteorological Satellite Studies, University of Wisconsin-Madison, Kansas City, MO, USA 4The Weather Channel, Atlanta, GA, USA


Introduction
The prediction of wintertime precipitation can be challenging, and the consequences of even correct forecasts can be significant.In mid-January 2004 up to 27 cm (11 in) of snow fell across western New York state, accompanied by thunder and lightning, as an Alberta clipper moved over the region.Such systems are not generally known for producing such intense precipitation, as noted by Bluestein [1] and Thomas and Martin [2], and the vigorous nature of this particular system surprised even seasoned local forecasters who are well acquainted with prolific snowfalls.
The Alberta clipper is generally known to be a fastmoving extratropical cyclone that originates in western Canada and is characterized by low moisture content, defined in Glickman [3].It is a product of lee cyclogenesis (see Glickman [3]).After formation, clippers typically move toward south-central Canada or the north-central United States, and then toward the east coast of North America (Hutchinson [4]).This track represents one of the major trajectories for cold-season North American cyclones (Petterssen [5]; Reitan [6]; Zishka and Smith [7]; Murray and Simmonds [8]; Hoskins and Hodges [9]; Simmonds and Keay [10]; Ulbrich et al. [11]; Ward [12]).As a result of their fast motions and general lack of available moisture, clipper-type storms typically produce modest precipitation amounts (7.5-15 cm in a 3-6 h period) along a narrow path generally north of the track (Harms [13]; Beckman [14]; Hutchinson [4]; Thomas and Martin [2]).Enhanced snowfall amounts are possible via an influx of Great Lakes moisture (Harms [13]; Cinzani and Changnon Jr. [15]; Silberberg [16]; Angel and Isard [17]) or through the effect of intense upper-level forcing in the presence of reduced static stability (Silberberg [16]; Gallus Jr. and Bresch [18]).
The interaction of transverse circulations associated with two separate upper-level (UL) jet streaks, along with its effects on observed lower-and middle-tropospheric weather features (and their ultimate surface impacts), has been documented in several studies.Uccellini [19] presented four cold-season precipitation episodes over the north-central

Data and Methods
For this case, analysis of synoptic-scale features was performed using the Eta model (Black [24]; Mesinger et al. [25]), while the Rapid Update Cycle (RUC; Benjamin et al. [26]) model was utilized to examine the UL jet streaks in finer detail, via both plan view and cross-section.The Eta model output was interpolated onto a horizontal grid with 80-km spacing and a 50-hPa vertical spacing, while the RUC model output was interpolated onto a similar grid with 25-hPa vertical intervals.Data were displayed and interrogated using the General Meteorological Package (GEMPAK; Koch et al. [27]).Plan-view analyses were investigated to diagnose the synoptic situation, determine the role of traditional quasi-geostrophic (QG) forcing mechanisms for vertical motion, and track the progress of UL jet streaks across the affected area.Cross-section analyses were created to diagnose ageostrophic circulations, vertical motion fields, and moisture distributions.In addition, backward trajectories were produced by the Hybrid Single Particle Lagrangian Integrated Trajectory Model (HYSPLIT; Stein et al. [28]) to determine the source regions of air parcels arriving over western New York during the event and to interrogate the potential role of the Great Lakes in the production of heavy snowfall during this episode.North American Regional Reanalysis (NARR; Mesinger et al. [29]) data were used to populate the HYSPLIT model.Radar reflectivity imageries from the US National Weather Service (NWS) Buffalo were included to track snowfall areas and determine their intensities during the storm's passage over western New York and its environs.
As part of the effort to understand how this storm created locally intense snowfall, we have adopted an ingredientsbased methodology (e.g., Doswell III et al. [30]), in which we attempt to identify the contributions of the following features to the overall event: (i) Synoptic setup (ii) Mesoscale forcing for ascent (iii) Moisture (iv) Potential Great Lakes influences (v) Instability.

Analysis and Discussion
3.1.Synoptic Setup.The 1200 UTC 11 January 2004 run of the Eta model was used to illustrate the progression of the synoptic-scale features of this case, which involved a weak Alberta clipper system that was forecast to move from Minnesota to Ontario province over the course of less than 24 h (Figures 1(a)-1(d)).The clipper's central pressure was forecast to drop from 1006 to 1005 hPa during its eastward track and then weaken by 1200 UTC 12 January 2004.The 500-hPa height field was equally unimpressive, with a broad trough over the eastern half of North America and an embedded shortwave over the surface cyclone at 1200 UTC 11 January 2004, which subsequently weakened with time.An examination of METAR observations from Buffalo and Niagara Falls (not shown) indicated that light snow had been falling intermittently throughout the early morning of 11 January, with steady snow occurring after 1500 UTC.The Eta model's quantitative precipitation forecast (QPF, not shown) for the Buffalo region during this period was on the order of 12 mm (0.5 in).Local forecasters were skeptical of this figure, believing it to be overestimated, as revealed in this excerpt from the Warning Forecast Office (WFO) Buffalo area forecast discussion (AFD) issued at 2006 UTC 11 January 2004: A clipper crosses Lake Ontario late tonight and should bring a general 2 to 4 inches as it passes by.Relatively mild air will hang around during Monday before another short wave pushes a second clipper through late Monday night with a following cold front during Tuesday.The models show the omega field remaining to the north so only chance pops are expected. ... As the afternoon and evening progressed, conditions worsened.Here is an excerpt from the 0147 UTC 12 January 2004 WFO Buffalo AFD:  Local forecasters were not anticipating this system to be exceptional, as Alberta clippers are not generally known to be prolific snowfall producers.Indeed, most stations in the Northeast US and Great Lakes regions reported no more than a few inches of snow on 11 January (Figure 2(a)).Nevertheless, areas of western New York reported up to 27 cm (11 in) of snow (Figure 2(b)), accompanied by lightning and thunder around 0300 UTC 12 January 2004.
For an event of this type, lift and moisture are certainly anticipated, along with a temperature profile that will support snow.However, this event featured lightning, which merits investigation of static stability., with neutral to cold air advection (CAA) across western New York at 1200 UTC.We acknowledge that the so-called "traditional" method of evaluating forcing for vertical motion via the individual terms of the quasigeostrophic (QG) omega equation is not ideal (Durran and Snellman [31]; Billingsley [32,33]).Yet, an examination of mid-tropospheric Q-vectors (not shown) did not indicate significant QG forcing for vertical motion across western New York during the period in question.Meanwhile, the Eta model 500-hPa omega fields indicated a weak, disorganized region of upward motion over and around the upper Great Lakes region east of the surface cyclone at 1800 UTC 11 January 2004 (Figure 5 In order to better evaluate the role of upper-level jet streak dynamics in this case, 0-h RUC model forecasts were The two jets continued their eastward progress at 0000 UTC 12 January 2004 (Figure 6(c)), with a broad but still distinct coupling signal over the Great Lakes.500-hPa vertical motion had weakened slightly but still showed a relative minimum omega of −6 b s −1 (+9 cm s −1 ) over southern Lake Huron.Midlevel frontogenesis was examined during this period (not shown), revealing a modest NW-SE band across the Niagara Peninsula, quite likely attributable to latent heat release via the strong ascent due to the jet interaction.

Mesoscale Forcing for
At 0300 UTC 12 January 2004 (Figure 6(d)), the jets propagated eastward while maintaining their strengths.The coupled jet signal followed a similar path over western New York.The southern jet's left exit region covered western New York and northwestern Pennsylvania at this time, while the right entrance region of the northern jet passed over the St. Lawrence River Valley of Quebec.The divergence and omega fields (−16 b s −1 [+24 cm s −1 ]) had significant maxima over the Niagara Peninsula of southern Ontario province.It was at this time that the snowband was at its most vigorous.Finally, the 0600 UTC analysis (Figure 6(e)) has the northern jet streak pulling off the coast of Nova Scotia, while the southern jet pushed eastward into central Pennsylvania, placing a well-focused divergent region over western and central New York state, along with a nearly collocated omega minimum (−8 b s −1 [+12 cm s −1 ]).By this time, the precipitation shield was becoming less organized across western New York.
The RUC initial fields were used to create cross-sections matching the times of the jet streak analyses detailed in the preceding section.The cross-section axes were drawn from the western edge of the strongest isotach of the northern jet streak to the eastern edge of the southern jet streak and are highlighted on the jet streak analyses shown previously.Siting of the cross-section axes was attempted to capture the coupled region between the two upper-level jet streaks.
The 1800 UTC 11 January 2004 cross-section (Figure 7(a)) highlights the presence of two jet streaks, the stronger one to the north of the slightly weaker one.A broad area of moist air can be found underneath the coupled region, extending upward through the mid-troposphere.Embedded in this region was an area of UVM, with a midlevel maximum close to the center of the coupled region.Three hours later (Figure 7(b)), the moist layer had expanded horizontally beneath the coupled region, with two distinct midlevel UVM maxima corresponding to the divergent regions of the two jets, the southern maximum being significantly stronger (−10 b s The 0000 UTC 12 January 2004 cross-sectional analysis (Figure 7(c)) reveals a region of nearly saturated air sloping from underneath the exit region of the southern jet upward and northward toward the entrance region of the northern jet.Two UVM maxima were present once again (∼−6 b s −1 [+9 cm s −1 ]): a near-surface region under the exit region of the southern jet and another located near 700 hPa in the coupled region, located in the 90% relative humidity swath.Three hours later (Figure 7(d)), the strongest signal for jet coupling and its consequences were found.A broad area of RH ≥ 70% was present from the surface through the mid-troposphere underneath the coupled region, characterized by strongly divergent ageostrophic wind vectors in the upper troposphere.Incredibly strong midlevel UVM was present underneath the coupling region, with omega values of −16 b s −1 (approximately +24 cm s −1 ).The 0600 UTC cross-section (Figure 7(e)) shows drier air starting to work into the northern half of the cross-section at low levels.The signal of the southern jet was not nearly as distinct (although still present, as evidenced in Figure 6(e)), while the UVM maximum (−7 b s −1 [+11 cm s −1 ]) was weaker than at 0300 UTC but extended through a deeper layer of the troposphere.

Moisture.
Anomalous precipitation in this system suggests the possibility of above normal moisture available for snow creation.As a means to assess both moisture abundance and origin, 72-hour backward trajectory ensembles were created using HYSPLIT, using three-hourly NARR fields as input data.The 27 ensemble trajectories were created by shifting the meteorology grid by one grid point in each of the horizontal directions and 0.01 sigma units in the vertical.In this section, we examined four different levels, terminating close to the surface (Figure 8(a)), as well as terminating near traditional meteorological levels, such as 1500 m above ground level (AGL; Figure 8(b)), 3000 m AGL (Figure 8(c)), and 5400 m AGL (Figure 8(d)); these latter three values approximate 850 hPa, 700 hPa, and 500 hPa, respectively.The end time for all trajectories was 0300 UTC 12 January 2004, near the time of heaviest snow.
The lowest level trajectory (Figure 8(a)) reveals parcels that have spent the previous 72 hours almost entirely over a continental surface, with an initial northerly flow becoming southerly in the latter portion of most paths.The 24-72-hour time frame is dominated by parcel relative humidity values of 50-60%, even though many descended from above 500 m AGL (not shown) in that period.Parcel relative humidity values increased with time over the last 24 hours of their path, while most of the paths hovered between 0 and 250 m AGL.Thus cooling and/or moistening of parcels occurred in the last 24 hours, though adiabatic ascent was not the cause here.Even so, near-surface relative humidity values average a clearly unsaturated 87% at 0300 UTC 12 January 2004.
For parcels terminating at 1500 m AGL (Figure 8(b)), a similar horizontal life history is depicted, with a northerly flow again turning southerly, though that change comes ∼48 hours prior to termination at 1500 m, as opposed to ∼24 hours prior to termination at 250 m.Relative humidity values gradually dropped for most parcel paths, from ∼80% at 0300 UTC 09 January 2004 (72 h out) to ∼60% at 0300 UTC 11 January 2004 (24 h out).Meanwhile, most parcel trajectories show very little vertical motion and spend most of their path at or below 500 m AGL (not shown).In the last 12 hours of the trajectory swarm, most of the parcels ascended ∼1000 m (∼2 cm s −1 ), and mean relative humidity values escalated rapidly to 91% by trajectory termination.
The parcel swarm terminating at 3000 m AGL (Figure 8 The final parcel swarm examined terminated at 5400 m AGL (Figure 8(d)).Most of these parcels also spent much of their time over the North American continent, though they all do originate over the Pacific.All paths in this backward parcel swarm began off the California coast, at ∼30 ∘ N, at a typical elevation of 2500 m AGL.Not surprisingly, the last 18 hours in these paths also reveals a significant increase in elevation, typically 3000 m (∼4.5 cm s −1 ).Relative humidity values again increased in the last day or so before termination but dropped off slightly in the last 6 hours to 73%.
These trajectory swarms show, both individually and collectively, that this was not a storm having an overabundance of water vapor with which to create anomalous snowfall.True, these values evince the moisture available to a typical wintertime storm, but the trajectories do not originate over regions with abundant moisture, nor do the parcels have much of a life history of near saturation.Given that the relative humidity is a temperature dependent variable, we also offer as additional evidence the precipitable water (12.3 mm) from the 0000 UTC 12 January 2004 sounding from Buffalo,    NY (Figure 9).This value falls within the first standard deviation and is approximately half of the all-time maximum for that date in Buffalo (NOAA [35]).
We close the trajectory analysis by noting that the last 12-24 hours of each trajectory swarm, when viewed together, reveals a veering wind profile.This signature corroborates the idea of deep warm advection as a forcing mechanism for ascent.Yet, the previous analyses, as well as the trajectories themselves, suggest modest synoptic-scale ascent values.This idea, coupled with typical atmospheric moisture values, would tend to suggest a less significant storm, with accumulations of the expected 8 to 15 cm range (e.g., Thomas and Martin [2]).Thus we turn to analyses of lake influence and static stability to determine a possible means of enhanced mesoscale upward motion in this system.

Potential Great Lakes
Influences.The snowband's location invokes the possibility of lake enhancement of the precipitation.The NARR 3-hour precipitation (Figure 10) matches well with the previous snowfall maps (see Figures 2(a)-2(b)).There is some question of lake-enhanced precipitation.Within the NARR, a precipitation band aligned along and parallel to the southern shore of Lake Erie.There, maximum 3-hour values of 2.5 mm were generated, ending at 0300 UTC 12 January 2004.Yet it is clear that the bulk of the NARR precipitation accumulation is transient, borne of larger-scale forcing.To underscore this point, the NARR sensible heat flux (Figure 11) and latent heat flux (Figure 12) at 0000 UTC 12 January 2004 match well with typical January values in the region (NOAA [36]).Significant lake contributions were unlikely.3.5.Instability.In the absence of significant lake surface influences and in the presence of fairly typical moisture profiles, we look to the thermodynamic profiles to assess instability as a mechanism for generating anomalously large snowfall totals.We will look first at the NARR soundings in the 0000 UTC to 0600 UTC 12 January 2004 time frame.Being constructed with its 4D assimilation method, the NARR reanalyses retain continuity with the preceding analyses in this study.

040112/0000V003 SFC FXSH03
Of the times examined, the 0600 UTC 12 January 2004 NARR sounding for Buffalo, New York, (Figure 13(a)) was the least statically stable of the group.The NARR soundings for Buffalo immediately prior (0000 UTC and 0300 UTC, not shown) are less impressive, having lifting parcel levels of 650 hPa (both), Lifted Index values (Lifted Index is the algebraic difference between the temperature of a adiabatically lifted parcel (based on a near-surface mixed layer) and that of the environment at 500 hPa; positive values indicate a parcel that is cooler than the environment, indicating static stability) of +4 (0000 UTC) and +2 (0300 UTC), and 700-500-hPa lapse rates of 5.7 K km −1 (0000 UTC) and 6.4 K km −1 (0300 UTC).However the 0600 UTC sounding is quite similar to the model suggested by Market et al. [37]  for midwestern United States convective snow cases, with a Lifted Index of 0, and a 700-500-hPa lapse rate of 7.2 K km −1 .However, no most unstable convective available potential energy (MUCAPE) is diagnosed at any of these times.
It is worth noting that an experimental run with the Weather Research and Forecasting (WRF; Skamarock and Klemp [38]) model for this event produced a sounding profile (Figure 13(b)) for Buffalo at 28 hours (0400 UTC 12 January 2004) that was quite similar to the NARR output at 0600 UTC (Figure 13(a)).This 13-km hydrostatic solution identified 46 J kg −1 of MUCAPE for a parcel lifted from 675 hPa.All of the soundings shown herein point to an atmosphere conducive to lightning, which was observed.
Thus the release of midlevel potential instability occurred to aid in both lightning generation and snow production.
3.6.Radar.The 1758 UTC 11 January 2004 0.5 ∘ elevation radar reflectivity image from the Buffalo NWS (Figure 14(a)) used to prodigious snowfall, so much snow in such a short period of time from a fast-moving synoptic system was most unusual.Reports of thundersnow are not uncommon for this region, but generally such events accompany vigorous Type I lake-effect snowbands (Moore and Orville [39]; Niziol et al. [40]).As a result, local forecasters were caught somewhat by surprise regarding the locally intense nature of the snowfall in this event, in spite of their wealth of accumulated knowledge and experience in heavy snow forecasting.
Postmortem analysis of this event reveals that the likely culprit of this event was the interaction of two upper-level jet streaks as they propagated over the area during a roughly 12-h period, along with localized convective instability that, while modest, was sufficient to yield charge separation.Trajectory analysis revealed that most air parcels in the lower and midtroposphere did not originate over moisture-rich areas; in addition, it appears that while the local Great Lakes might have provided some influence, the bulk of the heavy precipitation was due primarily to locally strong UVM via ULJ coupling in the presence of elevated convective instability.The divergent regions of the two jet streaks worked in concert over the lower Great Lakes region at the height of the storm, resulting in enhanced upward vertical motion and moisture transport, and ultimately snowfall totals not typically associated with fast-moving synoptic-scale disturbances.The 24-h liquid precipitation ending at 1200 UTC 12 January 2004 was 16.5 mm (0.65 in.), with a snow-liquid ratio of approximately 15 : 1, which is close to the climatological normal established by Baxter et al. [41].
Alberta clippers are most assuredly a consequence of a combination of meteorology and geography, forming in northwesterly flow regimes.Such cold-season airflow patterns are common around the world, often resulting in the development of significant snowbands in the Great Lakes region of the United States (e.g., Mulholland et al. [42]), the Gulf of Finland (e.g., Savijärvi [43]), and the Sea of Japan (e.g., Ueda et al. [44]), as well as the English Channel and Irish Sea (e.g., Norris et al. [45]).The difference in this case to these other examples is that while the significant snowfall occurred over the lower Great Lakes region of the United States, Lake Erie appeared to be of little consequence in the generation of the localized heavy snowfall amounts.
The analysis performed here obviously has the benefit of hindsight and access to finescale data without the typical demands and time pressures of an operational forecaster.While the period of interaction was brief and the affected area somewhat limited (albeit densely populated), the operational availability of short-term forecast data such as the High Resolution Rapid Refresh model (HRRR; Benjamin et al. [46]) should lead to better anticipation of mesoscale influences on synoptic-scale disturbances and their consequences.

Disclosure
Any opinions, findings, conclusions, or recommendations expressed herein are those of the authors and do not necessarily reflect the views of NSF.

Figure 1 Fast
Figure 1 Ascent.Evaluation of 500-hPa geopotential height and absolute vorticity fields from the Eta model between 1800 UTC 11 January and 1200 UTC 12 January 2004 (Figures 3(a)-3(d)) revealed a shortwave trough that transited from central Minnesota to central Pennsylvania and generally weakened as it did so.A broad area of positive vorticity advection (PVA) was found ahead of the moving trough over the period, entering extreme western New York at 0000 UTC 12 January 2004, and subsequently replaced by negative vorticity advection (NVA) by 1200 UTC.A large region of low-level warm air advection (WAA) was present ahead of the 850-hPa trough through most of the period (Figures 4(a)-4(d)) (a)), which became more focused and intense (−10 to −12 b s −1 [approximately +15 to +18 cm s −1 ]) over southern Ontario province and western New York state between 0000 and 0600 UTC 12 January 2004 (Figures 5(b)-5(c)), followed by weak forecasted vertical motion by 1200 UTC (Figure 5(d)).