Friday Weather Discussion: Super Bowl Snowstorm

6 February 2015

Case Overview

  • February 1-2, 2015 National Infrared Satellite loop
  • Note the development of the characteristic comma shape in the cloud field as the cyclone moves eastward.
  • Low cloud elements moving poleward from the Gulf of Mexico, but as I’ll soon show, there was not really a Gulf moisture connection for this snow storm.
  • Note cold brightness temperatures to the north throughout and then in the wake of the cyclone at later times, reflective of the cold air mass and fresh snowpack, respectively, present across the region.
  • February 1-2, 2015 Upper Mississippi Valley Radar loop
  • Note the inferred transition from warm conveyor belt-driven precipitation to deformation band-driven precipitation to lake effect precipitation.

Local Storm Impact

  • NWS Milwaukee/Sullivan CWA/Regional Snow Totals
  • Highest totals found in southern Milwaukee, Racine, and Kenosha counties, with even higher totals found to the south in Illinois.
  • Officially: 12.0” at MKE on 0.61” liquid-equivalent, for a mean snow-to-liquid ratio of 19.67:1.
  • NWS Chicago StormSummary
  • Fifth-largest snowstorm on record for Chicago (19.3”); most one-day snow (16.2” on the 1st) in February ever and fourth-largest for any day.
  • Officially: 19.3” at Chicago on 1.04” liquid-equivalent, for a mean snow-to-liquid ratio of 18.56:1.
  • Note local and regional snow accumulation maps and MODIS satellite image from February 2 showing the deep snowpack.
  • Snow-to-Liquid Equivalent Climatology
  • Ratios of 18-20:1 are toward the right-tail of the distribution of all snowfall events, but still only approximately one standard deviation above normal.
  • Unknown: how do these ratios stratify for minor, moderate, and heavy events?

Upper Air Charts

  • 300 hPa Charts, 31 January – 2 February
  • Longwave trough over eastern Canada, with a shortwave rotating northeastward through the northeast United States and eastern Canada.
  • Northern stream shortwave ejects east-southeast from the northern Rocky Mountains, interacting with a quasi-cutoff disturbance in the southwestern US.
  • Strong diffluence/divergence signature on February 1 across central US, partially in conjunction with the right entrance and left exit regions of two upper tropospheric jet streaks, one from the Great Lakes into New England and another along the Front Range of the northern Rockies.
  • 500 hPa Charts, 31 January – 2 February
  • Note the amplification of the upstream shortwave as it comes off of the Rockies and as it interacts with the southern stream quasi-cutoff disturbance. This is a typical evolution for shortwaves and is readily understood in light of vortex tube stretching arguments downwind of large mountain ranges.
  • 700 hPa Charts, 31 January – 2 February
  • Note the feed of moisture from the tropical eastern Pacific Ocean. The richest moisture (Td > -4°C) does not quite reach the Great Lakes, but does penetrate to between the I-70 and I-80 corridors.
  • 850 hPa Charts, 31 January – 2 February
  • Note implied deformation zone northeast of the developing cyclone, particularly on February 1, with implied frontogenesis.
  • Paltry moisture feed from the south; note quite dry dewpoints on January 31 across the southern Plains that progress eastward with the cyclone. Gulf moisture does return, but it remains south and east of our region.

Isentropic Charts

  • 1200 UTC 1 February, 285 K and 290 K surfaces
  • 2-3 g kg-1 mixing ratio in the 800-650 hPa layer, with isentropic ascent (flow across the isobars) noted from the southern Mississippi Valley into northern Illinois and southern Wisconsin.
  • Note the deformation zone, particularly on the 285 K surface, across the western Great Lakes, with a near-zero degree angle between the isobars and the axis of dilatation. This implies strong frontogenetical forcing here.
  • 0000 UTC 2 February, 285 K and 290 K surfaces
  • Eastward shift of the entire system, deformation pattern and all, albeit with weak ascent remaining over southern Lake Michigan on the 290 K surface.
  • This is coincident with the transition from synoptic-scale to lake effect snows.

Derived Charts

  • 700 hPa height, temperature, Q-vectors, and Q-vector convergence
  • Q-vector divergence (and thus quasi-geostrophic forcing for descent) trailing longwave trough across Manitoba and Ontario, Q-vector convergence (and thus quasi-geostrophic forcing for ascent) ahead of amplifying shortwave in Upper Midwest.
  • Note orientation of Q-vectors: parallel to and in same direction as the temperature gradient, from cold to warm. This strongly implies frontogenetical forcing on the 700 hPa isobaric surface as well, akin to what we saw on adjacent isentropic surfaces.
  • Why is frontogenetical forcing important? It typically accompanies rising motion (the things that lead to frontogenesis also lead to ascent, and vice versa), and is often a focus for banded precipitation features (which we won’t cover here today).
  • Sea level pressure, 1000-500 hPa thickness, 250 hPa wind, and precipitable water
  • Strong warm frontogenesis occurs, as implied by the tightening of the 1000-500 hPa thickness contours, as cyclogenesis occurs.
  • Note cyclogenesis in lee of the central Rockies but also the eastward extension of the trailing anticyclone in Canada, presumably as driven by the quasi-geostrophic descent noted above. This helps to lock seasonably cold air in place across the region, north of the surface warm front.
  • See, for example, this surface chart valid at 1800 UTC 1 February 2015.

Sounding Analysis and Lake Effect Snow Transition

  • 0.25° GFS-derived skew-T/ln-p diagrams for Milwaukee (MKE): 1800 UTC 1 February, 0000 UTC 2 February, 0600 UTC 2 February
  • Deeply moist, with very deep dendrite growth zone (-10°C to -20°C layer, roughly 300 hPa or more deep with forcing for ascent – e.g., as seen on Q-vector charts – through that layer).
  • Inversion height around 900 hPa (roughly 1 km above ground level).
  • 850 hPa temperatures decreasing from -11°C to -14°C to -16°C over the twelve hour period; coupled with lake temperatures of roughly +1°C, this implies a lake to 850 hPa delta-T of 12°C (favorable for lake enhancement; marginally favorable for lake effect) to 17°C (favorable for lake effect).
  • Northeasterly winds that become north-northeasterly and then northerly as the synoptic-scale cyclone shifts eastward; lake effect ends as winds turn lake-parallel and, eventually, offshore.
  • During lake effect period, note little to no directional shear beneath the inversion with modest speed shear (roughly 15 kt), primarily confined toward the top of the sub-inversion layer. This is generally favorable for lake effect.
  • By no means was this an extreme, or even particularly notable, lake effect event, but it did add that extra “oomph” to lead to the higher snow totals near the lake versus away from the lake.

Forecast Evolution, Challenges, and Standardized Anomalies

Note: many of the charts within this section are time-sensitive.

  • SREF, GEFS, etc. Ensemble Mean MSLP and Standardized Anomalies
  • Focus on forecasts valid 0000 UTC 2 February or, for SREF, 0300 UTC 2 February.
  • This was not a particularly anomalous feature, with roughly +1 standard deviation MSLP north and -2 standard deviation MSLP south.
  • The implied anomalous pressure gradient is large, however, as manifest by -3 standard deviation 850 hPa and 925 hPa u-winds (see change to 20150205, 00 cycle, 850WIND parameter). A post to Albany’s Map Discussion list from David Beachler with NWS Chicago implied that this was a similar pattern to what was seen with the Groundhog’s Day blizzard of 2011:
  • Note, however, that this is a 54-h forecast rather than a 0-h analysis, and that the anomalies with the 2011 storm are of greater magnitude than in 2015.
  • Other fields are similar in anomaly magnitude to MSLP.
  • The same set of charts provides a good perspective on forecast evolution in the ~48 h leading up to the verifying time. Both the SREF and GEFS, in their ensemble means, as well as the deterministic GFS did reasonably well with the track of the cyclone and its precipitation over the two day lead up to the event.
  • Levi Cowan/Tropical Tidbits d(prog)/dt charts
  • Note that link is the 0000 UTC 2 February 2015 GFS analysis of 500 hPa height and mean sea level pressure; use the arrows at bottom near “Prev. Runs” to cycle through as d(prog)/dt.
  • For the GFS, the event was consistently well-forecast starting 60-72 h out, though with some northward trend in the SLP minimum noted in that time.
  • Overall, the key uncertainties appeared to be with the amplitude of the shortwave responsible for the event (deeper with shorter lead time) and with the amplitude/orientation of the longwave trough in Canada (not as high amplitude, located more to north); as these came into focus, so too did the forecast.
  • The trend for the ECMWF was similar, as was that for the GEFS (in its ensemble mean) and the ECMWF EPS (in its ensemble mean).
  • Question: what degree of predictability should we expect for an event like this? Though I don’t have the individual ensemble member forecasts available, what should one do in the lead-up to an event – say, 3-4 days out – given cases where the ensemble is fairly certain about an event but ultimately is wrong in that certainty? What role can the human play in all of this – pattern matching/analogs (there’s quite a bit of variability here, even in 48 h forecast analogs, with many bearing no strong resemblance to what actually happened), deterministic vs. probabilistic forecast communication, and the like?