Monday, June 10, 2019

Does Siberian Snow Cover in October Really Predict North American Winters?

Over the last several winters, an interesting phenomenon that has been discussed is that of Siberian snow cover trends over the month of October. The idea is that if snow cover is anomalously widespread over Siberia in October, the following winter in North America will lean towards being colder than normal. Similarly, if snow cover over Siberia is lackluster in October, temperatures across North America will tend to be warmer than normal during the following winter.

This theory has been advanced by Dr. Judah Cohen of MIT, and I have been particularly fascinated by the concept. In an effort to see how reliable this method could be, in this post I present an admittedly-crude analysis of Eurasian snow cover in the month of October from 1967-2018 and view the subsequent winter seasons.

I would like to make it explicitly clear that this is only meant to be a crude analysis of this phenomenon - I have a great deal of respect for Dr. Cohen, and there are doubtless other far more intelligent and far more able-minded scientists who can analyze this topic more closely and more accurately. This post is more of a preliminary and basic analysis on the topic. Feel free to read more about the concept as well: link

---

Let's begin with an analysis of the data I will use in this analysis. I use data from Rutgers University's Global Snow Lab, which provides content on the extent of snow cover across different swaths of the Northern Hemisphere, as well as the Northern Hemisphere in the aggregate. This data is almost completely continuous, save for October data in the year 1969. As such, we will work with 51 samples in this analysis. For this post, I am using snow cover data for the Eurasia region specifically, over the month of October only. I am using data on an anomaly basis: per the Rutgers GSL, the average snow cover in the month of October in Eurasia is approximately 9,497,000 square kilometers.


I. Winters Following Top Ten October Eurasia Snow Cover Anomalies

First, I'll create a composite of winter-season temperature anomalies over North America for the winters following the ten years of highest positive snow cover anomalies in October over Eurasia. The composite is presented below. Click on any image to enlarge it.

Surface temperature anomalies for the ten winters following the ten Octobers featuring the highest Eurasia snow cover anomalies.
Source: ESRL
The ten years featuring the highest positive anomalies of snow cover in October over Eurasia, in order, are 1976, 2014, 2016, 2002, 1970, 2013, 2015, 1971, 1968 and 1972. I have graphed the composite surface temperature anomaly (in degrees Celsius) for the December-January-February period that followed each year in the above graphic. When putting these top ten years / top ten winters all together, the result is actually pretty in line with what this concept states, that stronger snow cover over Eurasia in the October preceding these winters should result in colder than normal winters for North America in the winters themselves.


II. Winters Following Bottom Ten October Eurasia Snow Cover Anomalies

Next, let's look at the temperature composite for the bottom ten years - that is, the winters that followed the ten Octobers where the snow cover anomaly over Eurasia was the lowest. The image is shown below.

Surface temperature anomalies for the bottom winters following the ten Octobers featuring the lowest Eurasia snow cover anomalies.
Source: ESRL
The picture becomes less clear-cut when looking at the ten winters which followed the ten Octobers with the lowest Eurasian snow cover anomalies. While the below-normal temperature anomalies in North America have become less centralized from what we saw in the first temperature composite, we do not see predominantly above-normal temperatures over these ten winters. Indeed, the majority of the country is actually at neutral or slightly below-normal temperatures, with much of Canada entrenched in firmly below-normal temperature anomalies.

However, I do see an area where this concept becomes useful.


III. 500-Millibar Geopotential Height Anomalies

I want to take a look at those same top-ten and bottom-ten winters, except this time instead of viewing surface temperature anomalies over North America we'll go over 500-millibar geopotential height anomalies over the Northern Hemisphere.

500-millibar geopotential height anomalies for the ten winters following the ten Octobers featuring the highest Eurasia snow cover anomalies.
Source: ESRL

500-millibar geopotential height anomalies for the ten winters following the ten Octobers featuring the bottom Eurasia snow cover anomalies.
Source: ESRL
Where the surface temperature anomalies were rather murky when trying to differentiate the winters following top-ten and bottom-ten Octobers in terms of snow cover anomalies over Eurasia, the picture becomes far more focused when expanding to 500-millibar geopotential height anomalies.

The first image of the above two shows 500-millibar height anomalies for those ten winters following the ten Octobers with the highest positive snow cover anomalies over Eurasia. That composite shows that the winters featured a firmly-disrupted tropospheric polar vortex, with strong ridging evident from the north Pacific through the Bering Sea, across the Arctic Circle and into western Europe. This led to the tropospheric polar vortex being forced to lower latitudes, as shown by the deep negative anomalies in North America, far western Europe, and northern Eurasia. In contrast, the second image - showing 500-millibar height anomalies for those ten winters following the ten Octobers with the deepest below-normal snow cover anomalies over Eurasia - shows essentially the opposite. A stronger than normal tropospheric polar vortex is observed across the upper latitudes, with generalized and widespread ridging prevailing around the mid-latitude regions.

This does seem to validate Dr. Cohen's theory, with more expansive snow cover over Siberia in October leading to a weaker stratospheric *and* tropospheric polar vortex in the Northern Hemisphere for the following winter.


To Summarize:
- A crude analysis of Eurasian snow cover data seems to confirm Dr. Judah Cohen's theory that anomalously high (low) snow cover over Eurasia in October corresponds to an anomalously weak (strong) stratospheric and tropospheric polar vortex in the following winter, with broadly-attendant chances for colder (warmer) than normal temperatures for the winter.
- While I was admittedly skeptical of this tool, the data in this preliminary analysis don't lie: this does seem to be a worthwhile tool for seasonal forecasting.

Andrew