As the Washington area braces for its biggest snow storm since Commutageddon (January 26, 2011), a number of similar historical events, or analogs, suggest disruptive accumulations will likely occur from Snowquester.
An analog is a weather pattern that had been observed in the past, and shares similarities with the pattern that’s either currently in place or forecast to develop. The linkages between the analog and the current or forecast pattern provide useful guidance to meteorologists, who can analyze the similarities and assess the quality of the comparison. Analog forecasting follows the principle “if it has happened before, it can happen again.”
Several analog storms and associated patterns align well with the one in progress. All of them can be found on Saint Louis University’s CIPS Analog Webpage, which provides an interactive suite of analog-based guidance that allows forecasters to identify high-impact events. The webpage generates new analog data with every other run (12 hours apart) of the GFS and NAM forecast models. Forecasters select the weather feature of interest, and the site directs them to a new page with a scoreboard ranking the top analogs to the relevant feature.
For the storm currently threatening the region, eight analogs have consistently appeared near the top of the rankings following recent runs of the GFS and NAM. These include dates on which significant snow storms impacted the Washington, D.C. area.
The chart below shows snow accumulations at Reagan National Airport (DCA) and Dulles International Airport (IAD) from each of the eight top analog storms.
Note that in all but two events, higher snow totals occurred at Dulles. In fact, averaged over all eight storms, nearly 3.5 inches more fell at Dulles (10.8 inches vs. 7.4 inches at Reagan).
It’s true that the first February blizzard of the 2009-10 “Snowmaggedon” winter skews the average quite a bit higher at Dulles, where 32.4 inches accumulated from that one system. Even after removing this storm from the averages, Dulles commands almost a two-inch advantage (7.8 vs. 6.0) over its urban neighbor.
Regardless of the statistics one prefers to use, these analog storms signal a potential for 6 inches or more with the Snowquester storm. Each system did possess unique characteristics that will not be mimicked by Snowquester, but also featured enough similarities to make the comparisons valid.
Take, for instance, the most recent analog: February 9-10, 2010. This second of two blizzards traced its origins to a Clipper system diving southeastward across the Midwest (see the map of observed snowfall amounts with an overlaid storm track below). The initial storm center weakened as it reached the Appalachian mountains, but redeveloped off the North Carolina coast. Areas from southeast Pennsylvania through northern Maryland received in excess of 18 inches within the storm’s bull’s-eye.
Among the more striking similarities between that event and the one on our doorstep is the orientation and strength of the low pressure center at 500 mb (the upper low). As the reanalysis below shows, the low had closed off (see the circular shape of the contour pattern) and had packed high amounts of vorticity (or energy, as represented by the orange and red colors).
What’s more, the center of the closed low and associated energy passed by very close to DC, exposing the city and especially its far outlying northern suburbs to incredibly high snowfall rates. While this was not one of those typical marginally cold situations for snow, evaporation from the rapidly falling precipitation supported robust cooling of the air temperature. The District and especially the southern and eastern suburbs may need to rely on this evaporational cooling process to receive the higher-end snow totals expected with Snowquester.
So, what’s a Snowquester to do? Mimic the one that came before – that’s what. I’ve shown below the forecast 500 mb heights and vorticity pattern for a time step during the storm: late morning Wednesday. No, the upper low isn’t nearly as “deep” (or intense) as its February 2010 counterpart; the former featured a minimum height of less than 516 dm while the latter is forecast to drop lower than 528 dm (lower heights signify stronger systems). The vorticity, on the other hand, is almost equivalent. All of this energy should go toward generating heavy amounts of precipitation and, as part of a positive feedback loop, cool the near-surface air through evaporation.
A second feature helps denote the similarities between our two snow storms, that being the jet stream configuration. At a distance 30,000 feet above the ground, the jet stream marks a high-speed current of air along which circulations move. Meteorologists refer to embedded higher speeds, or maxima, within the jet stream as jet streaks.
Look back up at the two graphics that I previously discussed. The jet streak on the right panel of the first graphic is depicted in purple (150+ knots). On the right panel of the second graphic, it’s depicted in midnight blue (130-150 knots), indicating a weaker jet streak. Importantly, both jet stream graphics show the Mid-Atlantic in a favorable position – near the left front part of the jet streak (flip your frame of reference horizontally to see this) - for heavy precipitation.
While, yes, the similarities are real, there are a few major differences to keep in mind. First, Snowmaggedon’s second blizzard (aka “Snoverkill”) deepened to a minimum central pressure lower than 980 mb off the Mid-Atlantic coast; the latest model guidance deepens Snowquester to a weaker sub-988 mb. Second, the February 9-10, 2010, storm worked with very cold temperatures (teens and low 20s) and, hence, produced anomalously snow-to-liquid ratios of 20:1 (for every inch of liquid, 20 inches of snow will accumulate); Snowquester will probably have to work with ratios of less than 10:1 in town. And finally, third, the blizzard tapped into a juicy Gulf of Mexico that had infused several storms with rich amounts of moisture throughout the El Nino winter of 2009-10; the Gulf may not add as much juice to the playing field this time around.
This is where the real, yet limited, use of analog forecasting becomes apparent. A forecaster can take the objective example of an analog pattern and identify its similarities, but also key in on the differences. It’s these differences that determine whether an epic event like that seen in February 2010 or a tamer yet disruptive storm like that which we expect with Snowquester will come to pass.
To conclude this post, I have mined the data for notable March snow events (those producing 4”+) since 1950 at Reagan and, from this search, have created the chart below.
Note that the chart also plots snow event totals for Hagerstown (HGR) and Baltimore (BWI); these were included with the goal of capturing any geographical bias associated with late-season snowfalls. Interestingly, while the spread between each city’s averages is fairly low (1.5 inches), Baltimore actually sports the highest average of 7.1 inches per event. Hagerstown places second (6.9 inches), while DC finishes third (5.6 inches). Hagerstown and Baltimore – not surprisingly – share a propensity to record a high outlying amount (three 11”+ events at HGR and three 8”+ events at BWI).
All in all, a pure analog approach to the forecast would favor a disruptive snow storm for Washington, D.C., with amounts nearing or just exceeding 6 inches. A combination of other forecast guidance, including that gathered from model, radar and satellite data; in situ observations; and pattern recognition, will also go a long way toward crafting a successful forecast.
* Meteorologist Rick Grow blogs about the weather for The Frederick News-Post . He has a degree in atmospheric sciences from UNC-Asheville and previously worked at MDA EarthSat as a forecaster.