Below are his forecasts for temperature followed by the actual observed temperature for two 3-month periods: December-January-February and January-February-March
What’s stunning is how well Cohen captured the cold (relative to average) in northern Europe and central and northern Asia and the warmth (relative to average) in southern Asia, northern Africa, and Greenland. Although he incorrectly called for colder than normal conditions over much of the Lower 48 in the December-January-February period, that call worked out incredibly well in the January-Februrary-March period.
As a reminder, Cohen’s develops his forecasts based on a relationship between the amount of snow cover in Siberia and its rate of advance in October. He has found that these characteristics can predict the prevailing state of the winter North Atlantic and Arctic Oscillations which are highly predictive of temperature.
In short, when the snow advances quickly in Siberia in October and there’s lots of it as in last fall – it favors a negative phase of these oscillations which are associated with cold weather over eastern North America and western Europe in the coming winter.
As evident in the figure below, Cohen’s Snow Advance Index (SAI) has an excellent track record in predicting the state of the winter Arctic Oscillation (AO) which has led to skillful winter outlooks.
“The SAI did a great job of predicting the strongly negative AO this winter, which if nothing else I am very pleased about since that is the first order prediction that we use the SAI for,” Cohen said in an email.
Cohen’s accurate outlook for a relatively cold winter over Northern Hemisphere land areas stands in sharp contrast to the average forecast from the large government operation centers, including NOAA’s Climate Prediction Center (CPC) which called for warmer than average conditions. The figures below show 1) average temperature predictions from an aggregate of government outlooks 2) NOAA’s temperature outlook for the U.S.
Compare what’s shown in the figures above with the observed temperatures in the relevant time frames in the top two figures in this post. These government-issued forecasts are, by-and-large, too warm.
NOAA has hesitated from using Cohen’s work in its outlook.
“The Arctic Oscillation can certainly play a large a role in outcome of winter, but at this point can only forecast on weekly to two-weekly basis,” CPC director Mike Halpert said upon issuing NOAA’s winter outlook last fall.
Yet Cohen’s research would suggest his technique can skillfully predict the AO months in advance.
Perhaps it’s time NOAA and other government centers begin to incorporate Cohen’s methodology into the preparation of their outlooks.