The increase in snow cover extent this October in Eurasia was fast and furious. That’s a compelling signal, says pioneering seasonal forecaster Judah Cohen, that the eastern U.S. faces a cold and snowy winter.
Cohen, who directs seasonal forecasting efforts at the firm Atmospheric and Environmental Research (AER), a unit of Verisk Climate, discovered the linkage between the behavior of Eurasian fall snow cover and eastern U.S. winters nearly 15 years ago. He has since applied the relationship in his winter outlooks and established an impressive track record. His outlooks have been on the money many years and at least in the ballpark most others.
Cohen’s method proposes that when snow increases rapidly and over a large area over Eurasia during October, it is a strong indication that a weather pattern known as the Arctic Oscillation (AO) will average in its negative phase during winter. When the AO is negative, it favors cold air outbreaks into the eastern U.S. and western Europe, that can also set the stage for snowstorms. (Conversely, when the Siberian snow is cover is minimal and slow to advance, the AO is favored to be positive, with cold air locked up in the Arctic and mild conditions farther south.)
I recently interviewed Cohen about this fall’s Eurasian snow cover, his winter outlook, and reactions to his outlook from professional peers. (The interview has been lightly edited.)
Briefly give us a sense of how Eurasian snow cover evolved this fall. How unusual was it?
Cohen: The snow cover was the second highest observed going back to 1972 and the speed of the advance was the fastest observed going back to 1997 (the first year that daily snow cover data became available). The signal from the snow cover was both strong and consistent.
Last year was very different in that the extent was high but the pace of the advance was slow, which provided an unusually divergent or inconsistent signal.
Given the snow cover situation, how confident are you that the AO will average negative this winter?
Cohen: As much as snow cover provides a useful prediction of the winter AO, this year is about as close to a highly confident forecast as you can get. There was some melting [of the snow] at the end of the month, which was fairly significant in western Russia and has me wondering but, for now, I am considering it just as noise.
What is your temperature outlook for the eastern U.S. (including the D.C. area)? Do you expect the first (December-January) or second (February-March) half to be harsher?
Cohen: We are predicting a temperature pattern that strongly resembles that associated with a negative winter AO, including cold in Washington, D.C. and much of the eastern U.S. (The AER winter forecast is posted to the National Science Foundation (NSF) Web site, which has funded much of my Eurasian snow-winter AO research that led to the development of the model.)
Typically the [fall Eurasian] snow signal doesn’t arrive [or manifest itself] in North America until sometime in January. So based on [fall Eurasian] snow cover alone, the forecast from January through March is clearer. So for years with high snow cover [like this one], I expect the period from late January through March to be colder than December through early-to-mid January. There have been some great examples of this recently but especially 2012-13.
Besides Eurasian snow cover, what other important factor(s) informed your outlook?
Cohen: Like everyone else, we include the El Nino Southern Oscillation (El Nino). But the developing El Nino looks to be weak and was not much of a factor in our model. I am encouraged that the warming sea surface temperatures have become stronger in the central equatorial Pacific relative to the eastern equatorial Pacific, which is complementary to our forecast based on snow cover, favoring cold temperatures in the eastern U.S.
We also use a proxy index for the atmospheric circulation in October across northern Eurasia that represents to what we refer to as a “tropospheric precursor.” We and other researchers have shown that tropospheric precursors influence the strength of the stratospheric polar vortex.
Finally, we used Arctic sea ice anomalies from September but this was not much of a factor.
I realize your outlook is temperature only, but do you think the odds for significant snow events along the East Coast are elevated?
Cohen: We have found that when snow cover is extensive across Eurasia in October, winter snowfall in the cities that stretch along the northeastern I-95 corridor is more likely to be above normal. So I would definitely consider the odds for significant snow events in the Eastern cities to be elevated. That will be especially true if a major sudden stratospheric warming is observed this winter and the polar vortex splits into two.
Briefly, how did your winter outlook perform last year and what it its long-term track record?
Cohen: Last winter was definitely a challenge. We now have two snow indices that we use in our winter forecasts and they both were strongly divergent; something that I frankly did not anticipate. For our December-February forecast, we weighted the snow index that indicated a positive winter AO. Hemispherically, it wasn’t a bad forecast but was too warm for the eastern U.S.
For January-March, we weighted the snow index that was indicating a negative AO (based on more active wave driving in December). That forecast captured the pattern across North America very well but performed more poorly across Eurasia.
Overall, I feel that our seasonal forecasts have performed very well but the January-March forecasts have been truly exceptional every winter going back to at least 2009. For the December through February period, the forecasts hemispherically have been excellent but more mixed for the U.S. [I’m] not sure if I fully understand why but possibly because the snow signal doesn’t arrive in North America until sometime in January making December much more of a wild card.
The National Weather Service has assigned “equal chances” for a cold or warm winter for the Mid-Atlantic and Northeast. Why do you think it’s reluctant to incorporate the AO-Eurasian snow cover relationship – which would suggest cold – more prominently in its outlook?
Cohen: It has been my opinion for my entire career that El Nino Southern Oscillation (ENSO) is too heavily emphasized in seasonal forecasting, especially for temperature, at all the national forecast centers, not just the National Weather Service.
One reason for the comfort in utilizing ENSO in seasonal forecasts is that the ENSO cycle can now be simulated in the dynamical models and it is accepted that the atmospheric response to ENSO is also correctly simulated in the models. However, the dynamical models do not simulate snow variability very well and they don’t simulate the atmospheric response to snow forcing as observed unless the snow cover is fixed and not allowed to vary freely. I believe that the poor model simulation of snow cover and the atmospheric response has contributed to the reluctance to adopt snow cover in seasonal forecasting
Meteorologists in the private sector and on TV on the East Coast are – almost universally – referring to the AO/Eurasian snow cover relationship and using it as an input in their forecasts. Are you glad to see your work hitting the mainstream?
Cohen: I was surprised by the wide usage of Eurasian snow cover in discussing and predicting the upcoming winter and the possible outcomes. But, of course, I am heartened by the greater acceptance of my ideas especially since this has been twenty years in the making.
But I am also nervous. I have always felt more comfortable being the contrarian. Also I wonder if the idea will lose much of its credibility if the forecasts do not do well. I recognize that no single predictor is perfect and that includes snow cover. For that reason very early on, I made great efforts to understand under what atmospheric conditions does snow cover perform well as a predictor. This led to the six steps or model for how snow cover influences winter weather.
Have some fun: Take a guess at D.C.’s snowfall total this year (our median snowfall is 12”). We won’t hold you to it.
Cohen: I think that the potential is high for an above normal snowfall this winter in all the large northeastern cities, including Washington. It seems to me that. at least for a while, it was feast and famine for all the cities and if there was high snowfall in one city it was shared by all, such as in the winters of 1995-96 and 2002-03.
But recently there seems to be a division somewhere near New York City (NYC) where, to the south and north of NYC, the fates of the northeastern cities have diverged. Washington to Philadelphia did exceptionally well in 2009-10 and last winter, but Boston not so much. Boston did exceptionally well in 2010-11 and 2012-13, but Philadelphia and Washington recorded much less snow. If that pattern continues this winter, that is an extra level of difficulty. So even if some Northeastern cities receive high snowfall again, I think that it may not be equally distributed.
So ,mostly because last year the snowfall was weighted to the south of NYC, I think that this year it may be focused further north (for full disclosure, I am usually overly optimistic on snowfall in my own locale). However, because it is likely to be an El Nino winter, this is in Washington D.C.’s favor. I will forecast 24 inches.
What could go wrong with your outlook?
Cohen (technical discussion): I will limit myself to what I currently see as the biggest challenge to the forecast.
[In our methodology linking Eurasian snow cover and the eastern U.S. winter outlook], we have identified six steps in a cycle starting with the advance of the October Eurasian snow cover and ending with the winter AO. These steps happen sequentially over the course of several months. We are still early in the cycle, at step two, which is the buildup of a strong and expansive Siberian high following the extensive snow cover from this past October. I wouldn’t say that the hand-off from the snow cover to the atmosphere has been fumbled but rather bobbled. A deep polar low has taken residence across northwestern Asia and Barents–Kara seas (B-K) region that has not allowed the Siberian high to expand northward that is optimal for wave driving necessary to break down the stratospheric polar vortex. Somewhat ironically sea ice in the region has been running at or above normal this fall, which is very different from what we have observed over the recent past where the largest winter sea ice deficits have been in in the B-K region.
In a recent review article on the possible link between Arctic amplification [more climate warming in the Arctic compared to lower latitudes] and extreme weather, my co-authors and myself hypothesized that high Eurasian snow cover and low B-K sea ice can constructively interfere to weaken the polar vortex. The relatively high sea ice in the B-K this year is more hospitable for low pressure. Therefore, at least in my opinion, snow cover is on its own in forcing a weakening of the polar vortex this winter [which could cause pieces of it to break off, resulting in outbreaks of frigid air over the eastern U.S. and Europe].
Though the snow signal typically doesn’t arrive in North America until January, it is always more comforting to see the winter begin on the cold side. December 2014 is predicted to begin very mild and our own forecast model predicts a mild December. Therefore, I don’t consider it a challenge given that a warm December is baked into the winter forecast. But anomalies established in December tend to persist and if it turned very mild across the US for most of not all of December, that might prove too much of a deficit to overcome the remainder of the winter.
Where can people go to get updates on your winter outlook and related commentary?
Cohen: My colleague, Jason Furtado, and I have started a new Arctic Oscillation blog on the AER website that is updated weekly. In the blog, we discuss analysis of the AO and forecasts in the coming days, weeks and even months. So all readers, but especially weather enthusiasts, can follow how we expect AO variability to evolve.
Jason even generated an animation of the hemispheric temperature anomalies based on high minus low October Eurasian snow cover composites from September through February. I think that it is definitely eye candy for a weather weenie like myself and I hope that others will enjoy it as well. It should not be treated as a forecast but possibly can highlight periods when Arctic outbreaks are more favored. Starting with December 1, periods of significant cold are first favored across Eurasia and then later in North America. It will be fun to see how well it compares to the observed temperature anomalies this winter.
Related reading: Capital Weather Gang’s 2014-15 winter outlook | Fall snow bonanza in North America and Siberia may portend brutal winter | Judah Cohen’s winter outlook: a downer for East Coast winter weather lovers (from 2013-14) | AER’s Judah Cohen produces amazingly accurate winter outlook (from 2012-13) | Leading forecaster Judah Cohen favors harsh winter for Washington, D.C., East Coast (from 2012-13) |Siberian snowfall may help improve U.S. weather forecasts, meteorologist says (2011) | The Siberia to East Coast snow connection (2011)