Breaking Down October NH Snowcover Data

[Bob Chill]

As most know, I'm a #'s guy. I will always look at #'s to help better understand the likely outcomes. Not just about weather either. Since long term predictions in weather have so much numerical data it makes it quite fun for me to do exercises like this.

There's been a lot of buzz recently about the October snow cover anomaly and its potential effects on winter in the US. Dr. Cohen did a great write up about the theory in 2011. It's a pretty short and easy to understand read so I recommend spending a few minutes reading it:

http://www.nws.noaa....ohen_062211.pdf

If you don't want to read it the short version is that there appears to be a connection with the amount of snow cover in the Northern Hemisphere during the month of October and the potential for increased blocking and cold air in the CONUS the following winter. Some of the finer details are the location of the anomaly (Eurasia is better) and the time frame that it builds up (2nd half of October seems best).

This year seems to have hit the preferred time frame, location, and anomaly. Eurasian snow cover in October appears to be the 3rd highest on recent record behind 1976 and 2002. The majority of the increase happened during the second half of the month.

I pulled snow cover data for the last 40+/- years, sorted it, and pulled the top 10. I gave the most weight to the highest #'s in Eurasia so the data is sorted for the to 10 on the other side of the globe but you can see in the image below, the top 10 in the NH are included as well. Doesn't look like focusing on Eurasia makes all that much difference because it only swaps out 1977 in exchange for 2004. The rankings on those years are 9th and 12th overall.

Here's the top 10s:

I put together the DJF monthly NAO, AO and trimonthly ENSO for the 10 years:

It's interesting to see the various states of ENSO. Definitely seems to be a bias towards warm enso (7 occurances) versus cold enso (3 occurances). I have no idea if that means anything but I don't think it matters one way or the other for this simplistic look at everything.

NAO and AO are relatively evenly divided between + years and - years. I guess the bias would be towards overall negative but not by all that much. However, it is notable that some of the years had a very -AO and were quite cold. Some very notable months of -NAO as well. It is also notable that the state of the AO in December have strong implications as to what to expect going forward. We discussed this at length last year so most here already know this.

I pulled DJF temp and precip maps for the top 10 and they look like this:

I'm not sure that is the best representation though because there are some very strong Ninos and 3 solid Ninas in the mix. I pulled the weak Ninos from the list and the temp and precip looks like this:

There is winter in particular that really stands out and that is 1976-77. Not only was it the top spot on the snow cover list, it was also a weak Nino following a multi-year Nina. Even though this year may not even hit weak Nino status it will still most likely be a warm neutral year. Here's the monthly temp breakdown of 76-77 and the season's precip:

It is also notable that with the exception of 2006, the weak Ninos on the list featured a negative to very negative AO with plenty of blocking. It is encouraging to see the -NAO/AO showing their face already this year but it would be unreasonable to assume that will continue. All eyes are on Dec at this point. There seems to be quite a high likelihood that if we can establish a solid -AO in Dec then we could be in for quite a cold winter. Snow is always a wildcard. This far south temps are always our enemy so lets see if we can get the cold in here and worry about precip later.

[Kmlwx]

This is excellent work! Thanks for putting it together. Seems at least some good news for the snow geese.

[TUweathermanDD]

Bob, excellent job man! I love it, and your reasoning is quite supportive.

[Ellinwood]

Thanks for collecting the data! 1976 is also a good analog because it has a -PDO, which seems to be what we'll get this winter. It did have the opposite sign on the AMO, though (but to me the AMO isn't as important for the CONUS in winter as a whole compared to the other factors).

[Bob Chill]

Thanks for collecting the data! 1976 is also a good analog because it has a -PDO, which seems to be what we'll get this winter. It did have the opposite sign on the AMO, though (but to me the AMO isn't as important for the CONUS in winter as a whole compared to the other factors).

The one thing I should have added about 76-77 was it was almost snowless. It was good for ice skating on the local ponds but pretty crappy for sledding.

Doesn't really matter though. The vast majority of times it's the temps that do us in down here. I'm sure most of us would agree that we're better off taking are chances with cold temps and a dryer than normal pattern and not the other way around. Not implying this winter will be dry anyway. I really have no idea about what to expect precip wise this winter.

[EastCoast NPZ]

The one thing I should have added about 76-77 was it was almost snowless. It was good for ice skating on the local ponds but pretty crappy for sledding.

Doesn't really matter though. The vast majority of times it's the temps that do us in down here. I'm sure most of us would agree that we're better off taking are chances with cold temps and a dryer than normal pattern and not the other way around. Not implying this winter will be dry anyway. I really have no idea about what to expect precip wise this winter.

Was just going to point this out after reading your first (and excellent) post. But as you said, we need the cold first before we can worry about precip (or lack thereof). This was the winter that formed (and likely skewed) my early childhood memories of cold winters and ice-skating. Great memories.

[WVclimo]

1976-77 winter was outrageously cold at Martinsburg:

111 below freezing nights for the entire cold season. 61 sub 20-degree lows, 19 nights under 10 degrees, and 4 below zero readings.

Not sure if we'll ever see something that extreme again.

[GaWx]

Thanks for collecting the data! 1976 is also a good analog because it has a -PDO, which seems to be what we'll get this winter. It did have the opposite sign on the AMO, though (but to me the AMO isn't as important for the CONUS in winter as a whole compared to the other factors).

The PDO was negative through June of 1976. However, it then turned positive and was solidly positive throughout the winter of 1976-7 (per both the Univ. of Washington and the NOAA tables).

If we can get back into steady weak Nino territory and it lasts long enough to get us a weak Nino peak this fall/winter as opposed to neutral positive (see my updates from today in the ENSO thread within the main wx forum that illustrate how much Nino 3.4 has been warming in recent days), 1976-7 would be a good analog since it was a weak Nino following a Nina. Two other weak Ninos that followed a Nina, 1939-40 and 1904-5, also had a strong turnaround from a -PDO to a +PDO sometime during the year of the oncoming Nino.

Because of these three analogs, I have been saying for a couple of months that there's more of a shot at getting a +PDO averaged over DJF than one might think given the recent very -PDO and -PDO regime (I put it at ~1 in 3 chance). The PDO hit its most negative around Sep. 10. However, it rose substantially from Sep. 10 (when it was well below -2) through 10/18 (when I roughly estimated that it had reached -0.50) thanks largely to the predominant -EPO. I believe it is now still close to -0.50. I am expecting that the U of Wash. Oct. PDO will be no lower than ~-1.25 and it may be as high as ~-0.75. It may fall back some during the first half of Nov. due to the lack of a forecasted predominant -EPO. Regardless, due to the fact that the PDO has risen so much since Sep. 10, I think that the odds of a +PDO averaged over the upcoming DJF have risen above 1 in 3...say, to about 40% now.

By the way, 1976-7 is the coldest winter on record for Atlanta.

[Bob Chill]

The PDO was negative through June of 1976. However, it then turned positive and was solidly positive throughout the winter of 1976-7 (per both the Univ. of Washington and the NOAA tables).

By the way, 1976-7 is the coldest winter on record for Atlanta.

I just went through the monthlies. You are correct. Although the 70's were clearly entirely in a longer term -pdo cycle, the winter of 76-77 was out on an island with the PDO index. I'm sure that had something to do with the intense cold but I don't think a +pdo is necessary to have a cooler than normal winter. Sure seems to help though.

Interestingly, in 77-78 the pdo was negative (modestly) much of the fall and early winter but flipped positive from Jan-Jun of 78.

It seems the chance of having a winter as cold as 76-77 nowadays is all but nonexistent. Probably a myriad of factors including global warming. However, there is no reason to discount having a good old fashioned cold winter. Look at what europe had last year. I don't know where to find climate data for europe but last year's cold and snow was one for the books.

[Ellinwood]

The PDO was negative through June of 1976. However, it then turned positive and was solidly positive throughout the winter of 1976-7 (per both the Univ. of Washington and the NOAA tables).

Oops, you're right! I was confusing it with another year.

I have also been wondering about the likelihood the PDO could go positive this year. It certainly has precedence as you have suggested, and given the recent trends it would not surprise me if it in fact did go positive for even just part of the season.

[usedtobe]

Bob,

Correlations with snow cover along certainly are not high enough to forecast a neg ao or NAO. The snow cover change index correlations for October offered by Cohen are quite a bit higher but the really high correlation he found were with a rather short data set making it tough to know how much weight to give it. If in November the stratospheric polar vortex is weaker than normal then I'd probably run with a negative AO especially if we get a weak nino.

[brad1551]

Bob Chill, I enjoyed reading your post on the Eurasia snow cover. I did not realize at all that we are the 3rd highest in snow cover behind 76 & 02.

Nice write up and thanks again!!!!

[Bob Chill]

Bob,

Correlations with snow cover along certainly are not high enough to forecast a neg ao or NAO. The snow cover change index correlations for October offered by Cohen are quite a bit higher but the really high correlation he found were with a rather short data set making it tough to know how much weight to give it. If in November the stratospheric polar vortex is weaker than normal then I'd probably run with a negative AO especially if we get a weak nino.

I totally agree Wes. I wasn't drawing conclusions as much as just tying the present the data in a reasonable format. It would be foolish to draw a black and white conclusion using a 10 year sample of a 40 year pool of data. I wish we had satellites and computers 1,000 years ago. LR forecasting would be a science and not the art it is today.

The one thing that I wonder about is the chicken/egg thing. IF there is a correlation is it the snow cover itself or what is causing the snow cover to happen in the first place? Meaning, I'm curious about pattern that builds the snow cover could potentially be the same pattern that drives things down the road. That it's simply not a function of the snow cover itself. But I really don't know squat about the whys.

I was going to include this in my original post but I wasn't finished looking at it and it would have crowded things. I went ahead an pulled the 10 lowest snow cover years. Makes sense to look at it on both sides to see if there is a notable difference. Even though the sample is quite small the differences are quite notable.

Here's a grid comparing the AO/NAO for the 2 sets of years:

Small sample aside, it's tough not to acknowledge the difference. There appears to be a clear bias towards + indicies with the low snow cover years. Again, is it a function of what caused the low snow anomaly to begin with or is Dr. Cohen really hitting on something?

Here's the DJF temp map for the 10 lowest snow cover years:

I could make the assumption about availability of cold air. Years with substantially less snow cover in fall at the higher latitudes could imply that the source region doesn't have as much bottled up to dump down on us. Or it could mean that the mechanism that dumps cold air on us is affected by low snow cover in the fall. Or it could simply mean nothing because the data set is ridiculously small.

[usedtobe]

I totally agree Wes. I wasn't drawing conclusions as much as just tying the present the data in a reasonable format. It would be foolish to draw a black and white conclusion using a 10 year sample of a 40 year pool of data. I wish we had satellites and computers 1,000 years ago. LR forecasting would be a science and not the art it is today.

The one thing that I wonder about is the chicken/egg thing. IF there is a correlation is it the snow cover itself or what is causing the snow cover to happen in the first place? Meaning, I'm curious about pattern that builds the snow cover could potentially be the same pattern that drives things down the road. That it's simply not a function of the snow cover itself. But I really don't know squat about the whys.

I was going to include this in my original post but I wasn't finished looking at it and it would have crowded things. I went ahead an pulled the 10 lowest snow cover years. Makes sense to look at it on both sides to see if there is a notable difference. Even though the sample is quite small the differences are quite notable.

Here's a grid comparing the AO/NAO for the 2 sets of years:

Small sample aside, it's tough not to acknowledge the difference. There appears to be a clear bias towards + indicies with the low snow cover years. Again, is it a function of what caused the low snow anomaly to begin with or is Dr. Cohen really hitting on something?

Here's the DJF temp map for the 10 lowest snow cover years:

I could make the assumption about availability of cold air. Years with substantially less snow cover in fall at the higher latitudes could imply that the source region doesn't have as much bottled up to dump down on us. Or it could mean that the mechanism that dumps cold air on us is affected by low snow cover in the fall. Or it could simply mean nothing because the data set is ridiculously small.

I think the info on Allan Huffman's stats on the October negative AO years in his Raleigh Examiner blog suggests that it might be related to early blocking and warmth at high latitudes. I'm starting to favor a negative AO for the winter though I attest to having no skill at making such a guess.