Originally by Matt Ross, with contributions from Josh Larson and Jason Samenow, published October 25, 2005

Executive Summary

We believe this winter will, on average, feature normal temperatures with slightly above normal snowfall. Unlike last winter which featured frequent oscillations from warm to cold and back, this winter will feature more persistent and longer lasting periods of both cold and warmth. Although it is difficult to predict with accuracy so far in advance, we believe that the winter will play out as follows:

Temperatures at Reagan National Airport (DCA):

Colder than average air will arrive sometime in the period from late November to mid December and last several weeks. Then as early as mid/late December to as late as early/mid January, warmer air will take over for an extended period of time. This warm period could last well into February, but likely at least into the early part of the month. The cold will return sometime in February and last well into March before springtime asserts itself. Because the "warm" period will actually take place during the time when temperatures are typically coldest and as the cold periods will occur both early and late in the season, winter could feel quite cold and long compared to last year.

December: Normal
January: +1.5 (degrees F above average)
February: Normal
March: -2

Overall temperatures for the December-March period: Normal, but leaning slightly toward below normal

Snowfall across the region:

We believe that snowfall will be slightly above average for the whole area. The cold periods described above will likely be the key times for meaningful snow events. However, the warm period isn't likely to be snowless as it occurs over the traditional heart of winter. Chances for a 6"+ widepsread event are at or above 50-50, and best guess timeframe for such an event(s) would be early/mid December to early January and/or mid-February to early March.

DCA: 16"-19"
Dulles(IAD): 24"-27"
Loudoun Co./Northern Montgomery Co.: 25"-30"
Outside the Beltway(Fairfax, Rockville, Mclean, Silver Spring): 20"-25"
Inside the Beltway(DC/Arl/Alex), PG Co.: 15"-20"

Overall snowfall: Slightly Above Normal

Additional Winter Forecast Detail and Rationale

We evaluated the current state of a number of climatic factors (such as temperature anomalies,large scale patterns and hurricane activity), and compared them to seasons past to identify the seasons sharing similar characteristics to this year. Those seasons from the past with similar overall characteristics to this season were selected as the "analogs" that we applied to project this winter's conditions (based on the winters from those analog years). The following factors were evaluated to develop the analogs and shape our winter outlook:

Summer Temperatures

The past summer was almost among the top 10 hottest and all 3 months were above average with increasing warm departures each month from May-September. Even by the increasingly warm standard of the last 20-30 years this summer was hot, primarily based on warm nights. We didn't hit the century mark, but we hit 90 about the average amount and a number of times in September. Among the related factors that were considered in selecting analog years:

• Hot summers past, adjusting for different climatology when necessary


• Summers with a similar temperature progression with respect to normal


• Summers with preceding cold Mays and proceeding warm Septembers


• Summers with the warmest anomalies centered over the Great Lakes region

These factors had a substantial, if not overwhelming impact on analog years and other years that were influential.

El Nino Southern Oscillation (ENSO)

It is becoming apparent based on a number of factors that ENSO will be neutral this winter, perhaps in the positive range. In neutral winters, there is a lot of variability in sensible weather. The extreme example of this is 1989-90 with the record cold December, much above average January and February, 90 degrees in March, and then several April snow events. We will not likely see extreme variability this winter (89-90 was neutral following La Nina, see below), but I think we have to factor in some variability as a likelihood. The following related factors were considered in selecting analogs:

• Preferred neutral winters that followed Weak Nino/neutral years or periods as opposed to neutral winters that followed La Nina years or periods (typically mild for us with long warm periods, see 2001-02).


• The evolution of sea surface temperature (SST) patterns

Quasi-biennial Oscillation (QBO)

This stratospheric oscillation has recently been in its easterly phase much more than it has been historically (-28 is about as low as it has gotten). It looks like we have bottomed out and the index will remain steady or rise as we head through the next 6 months. However, it is likely that it will still remain negative/easterly throughout winter with it possibly turning positive toward the end. Some, if not a lot of consideration was given to the QBO -- mostly to eliminate years when it was very positive in selecting possible analogs for the coming winter. A caveat is that data only go back to 1948 so the sample size is small.

Hurricane Season

We have clearly entered a period of more active seasons and earlier years should be adjusted as such. Nevertheless this season is memorable and perhaps record breaking and should be a factor in selecting analogs (i.e. we looked for other active years). The following was given consideration:

• Seasons with high tropical activity


• Seasons with a lot of Gulf of Mexico activity and Atlantic recurvature.

North Atlantic Oscillation (NAO)/Atlantic setup

There are multiple indications of above average blocking this year (or negative NAO). For snow lovers, we don't always execute in the good pattern setups, but this pattern, if realized, could give us a shot. The Atlantic will be of greater weight than usual since ENSO is less of a factor.

Pacific Decadal Oscillation (PDO)

There is some evidence that the Decadal index may not have switched in the late 90s to negative. We are not sure that it matters a whole lot whether we are with or against the prevailing index period this season. Climatology at this time seems to be split over whether we will average positive or negative for winter. Strong positive PDO periods, such as 2002-03, seem to support +PNA (Pacific North American pattern) setups which are usually good storm/cold setups for us. Given the uncertainty of how it will average out over winter (not as predictable as QBO), we treated the importance of huge +PDO winters with some skepticism.

Analog Years

Analog years are used as guidance for what will occur this winter. Some years carry more weight than others and often they are used for different reasons. Rather than focus on any one analog year, it is good to view them in their totality without taking the data too literally. Three good periods to focus on for the upcoming winter are: 1932-1937, 1958-1963, and 1978-1982. In addition, we factored in some additional winters including 1900-01, 1966-67, 2003-04.

Although a number of past winters helped shape the winter outlook, the 4 below were the most influential (in chronological order):

1900-01

Positives:

• Very hot summer with increasing departures every month. Heat held into September and October like this year.


• Heat nicely centered over East Coast and Great Lakes.


• Perhaps the best overall temp matchup of all 4 analogs.


• PDO went negative in September after months being positive and not an overwhelming factor.

Possible Negatives/Drawbacks:

• Old data/climo not as relevant/reliable.


• No QBO data.


• Nice tropical season match in terms of tracks, but only 7 named storms.


• ENSO possibly borderline; Nino following moderate Nino. A tad warmer in region 3.4 than expected this year.

DC temps adjusted for climatology:

• DEC: Normal


• JAN: Slightly above


• FEB: Well below


• MAR: Slightly above

Snow:

• DC: 9.1"


• Baltimore: 8.7"


• NYC/BOS: Well below

1932-33

Positives:

• Very nice progression of temps locally from May-October.


• ENSO is nice neutral matchup following a several year period of generally warm values.


• PDO generally not a factor.


• Excellent tropical track matchup with above average(11) if not a remakable number of named storms

Possible Negatives/Drawbacks:

• Followed a historically warm and snowless winter: 1931-32


• No QBO data.


• Summer temperature anomaly locations are ok, but even when adjusted for climatology simply not as hot as 2005, just warm.

DC Temps adjusted:

• DEC: Normal to Slightly above


• JAN: Well above


• FEB: Slightly Above


• MAR: Slightly/Somewhat Below

Snowfall:

• DC: 23.8"


• BAL: 27.9"


• NYC/BOS: Normal

1936-37

Positives:

• While not perfect, good temperature profile both nationally and locally with nice progression matchup.


• Excellent ENSO comparitive.


• Good if not great tropical track matchup and very active season with 16 storms.

Possible Negatives/Drawbacks:

• No QBO data


• PDO ok and not a big factor but followed extended period of very elevated positive values.


• Warm May

DC Temperatures adjusted for climatology:

• DEC: Somewhat Above


• JAN: Well Above


• FEB: Normal


• MAR: Somewhat to Well below

Snow:

• DC: 20.2"


• BAL: 25.6"


• NYC/BOS: Somewhat to well below

1959-60

Positives:

• Excellent temperature profile both nationally and locally with very nice progression.


• Good if not ideal ENSO matchup with dimishing but then plateauing values coming out of a warm period.


• Decent tropical track matchup with 11 storms. Not great but moderately better than your average profile.


• PDO decent and not a strong factor.

Possible Negatives/Drawbacks:

• QBO had minor positive values and was falling instead of rising.


• Very warm May.

DC Temperatures:

• DEC: Somewhat Above normal


• JAN: Slightly to Somewhat above normal


• FEB: Normal


• MAR: Well below normal

Snow:

• DC: 24.3"


• BAL: 34.1"


• NYC/BOS: Boston normal, NYC somewhat above

Summary

DECEMBER

The 4 analog years average slightly above normal. But the 2 warm Decembers (1936, 1959) had cool to cold Novembers. None of the analogs suggest that the warmth we have had holds all the way through December. It is a matter of duration and over what period it is centered. So December may have normal temperatures as opposed to slightly above normal temperatures, especially if the NAO is negative (a possibility).

JANUARY:

All the analogs agree on a warm January. The warmth is unlikely to hold for the whole month, but warm is the way to go. If any month is a good candidate for much above average temperatures, it would be January.

FEBRUARY:

No great analog agreement, but the cold will return. It is just a matter of when it returns. We could be tempted by the analogs to go for slightly to somewhat below average February, but we think it gets mitigated by the following factors:

• February has been skewed cold in the 1971-2000 profile and is actually colder than the 1940-1970 profile.


• Other than 2003, every February since 1996 has been warm and some very warm. The climate may be changing/adjusting.

MARCH:

Data is very strong for cold March. Hot September correlates nicely with a cold March. Even 1900-01 which was above normal had a very cold Febuary. It fits the profile for neutral winters and variability.

SNOW:

1900-01 is the outlier, but even the other years are skewed high given the move to DCA (which receives less snow due airport/river location than original recording station) and a lot of late season events to pad the totals that might not even stick to the grass at DCA. We think the pattern is dynamic though that with bigger storms and with a -NAO, we will have our chances. General climatology for the type of years considered in addition to the analog years is a low range of 8-12" and an upper range of 25-30". We are leaning toward the middle of the two. This is also supported by the warmest period coinciding with the coldest month where it still can snow even if it is warm. Normal/cold bookends in this pattern suggests slightly above climatology.

The 30 yr running average snowfall at DCA is 15.5" with a median of 12.1".