If the snow falls furiously in Siberia in October, it means a cold, snowy winter in the eastern United States.
Cohen has applied this theory over the years in his predictions, which generally have a strong track record. However, last fall, when the snow piled up fast over Siberia, Cohen predicted a cold, snowy winter for the eastern United States based on the theory, and the opposite occurred.
Unfazed by last year’s miss, Cohen stands by his theory — but has adapted it some. On Thursday, AER, which is supported by the National Science Foundation, released Cohen’s 2017-2018 outlook.
The snow cover was once again extensive in Siberia in October, and that has him leaning toward a cold and snowy winter for the eastern United States. But he said additional indicators he uses to make this year’s outlook aren’t all that clear — so his confidence is guarded.
Tell us, in brief, how much snow cover we had in Siberia in the fall and how fast it advanced. How does this compare to other years you’ve investigated?
I monitor two snow indexes. The first is the Eurasian snow cover extent for October. That index was above normal but less than recent Octobers.
The second index, the snow advance index, measures the rate of snow cover advance during the month of October, and that came in slightly below normal.
The snow cover started the month well above normal and then advanced slower than normal for the remainder of the month. I would conclude that the snow cover in Siberia this year provides a mixed signal. The last time that occurred was in 2013.
What are your predictions for the Arctic Oscillation in the winter based on the snow cover evolution?
When snow cover is above normal, that predicts a negative Arctic Oscillation, and when it is below normal, it predicts a positive winter Arctic Oscillation. Given that the snow indexes are mixed, I don’t think there is a strong signal for the winter Arctic Oscillation. However, I lean toward a negative winter Arctic Oscillation [based on the above normal snow cover extent].
I think the snow advance index is a better predictor of the timing of when we might anticipate a weakening of the polar vortex. I think that slower snow cover advance indicates a delayed weakening of the polar vortex [compared with normal], which is accurate. However, because of melting sea ice and accelerated warming in the Arctic, I do believe that the scales are tipped toward a weakening of the polar vortex eventually in the winter. Following a weakening or disruption of the polar vortex is when I would expect the most intense winter weather.
For the eastern United States, how do you see the winter shaping up compared to average and compared to last year — for temperatures?
We are predicting a colder than normal winter in the northern and eastern United States and warmer than normal winter in the southwestern and south-central United States. Our forecast is based on La Niña, which favors cold temperatures in the northwestern U.S., Eurasian snow cover, September Arctic sea ice concentration, which was below normal, and an index of high latitude blocking which was active in October. All of the three high latitude predictors favor cold temperatures in the eastern United States. I have to admit that I don’t have much confidence in the forecast.
There’s the saying you’re only as good as your last forecast, and last year’s winter outlook wasn’t your best. Some folks are wondering whether fall Siberian snow cover is as good a predictor as once thought for eastern North America winter weather. How do you respond to criticism, and have you had to adapt your methods at all?
I admit that last year’s winter forecast was poor in the eastern United States, but I do believe you learn more from your failures than your successes. I have argued in my research that above normal snow cover in the fall favors a weakening of the polar vortex and any value in using snow cover as a predictor is related to that atmospheric response. Nothing that I observed last winter makes me believe differently.
I recognize that snow cover is far from a perfect predictor, and I am trying to include new predictors like Arctic sea ice and improve the accuracy of the forecast, but it is a very difficult problem. Studying snow cover has led me to the polar vortex, and I remain convinced studying the polar vortex is the most expedient way to improve long-range winter forecasts.
How much snow are you thinking this year for D.C. and Boston?
I ran our snowfall forecast model, and it is predicting 17 inches for Washington and 64 inches for Boston. In general, based on our model, all the large cities in the northeast should expect above normal snowfall. I do think that La Niña favors the axis of heavy snowfall relative to normal further north across New England in the winter. Therefore, I am more confident of above normal snowfall in New England than in the Mid-Atlantic.