And we begin.

[HM]

 

I wouldn't say AO prediction is not in its infancy stage.  Again, it's a start.  And a good one so far.

 

Wait, what? I'm not sure what your first sentence means. But, if you are saying that AO long range predictions are, too, in their infancy, like the sun, then you would be incorrect. LR-solar predictions are significantly behind LR-weather predictions. 

The last few days have seen a solid jump in snowcover in the Eurasian side.

I don't mean to pile-on here, but this is why we have an index like the SAI. Human eyes can be correct or incorrect. But in the case of snow cover, yes usually you see a quick advancement day-to-day as storm systems move through. To keep saying that we have seen a "jump" each passing day in autumn means very little. We don't know what the rate is without the numbers. I look forward to seeing if the rate has indeed improved or not.  

[mcwx]

and....
http://www.bloomberg.com/news/2013-10-24/if-new-york-freezes-in-january-blame-siberian-snow-now.html

[GaWx]

and....

http://www.bloomberg.com/news/2013-10-24/if-new-york-freezes-in-january-blame-siberian-snow-now.html

 

 From this link:

 

 "'It’s the best winter predictor that we have,” Cohen said in a telephone interview. “We haven’t made a forecast yet, but we’re watching it closely and the snow cover has definitely been above normal so far.'

The more ground covered by snow across northern Europe and Asia at the end of October, the greater the chances of triggering a phenomenon known as the negative phase of the Arctic Oscillation. That would flood North America, Europe and East Asia with polar air and possibly erect a blocking effect in the North Atlantic that would bottle up the cold in the U.S."

 

 This article says nothing about an Oct. snow advance index. Cohen, himself, implies that the average Oct snow is the crucial factor as opposed to Oct. snow advance. They are not the same thing. So, which is it? Does heavy snow in late Sep. help or hurt -AO chances? If we're talking about average Oct. snow, then there's no doubt that widespread late Sep. snowfall would tend to help since that leads to a head-start in early Oct. However, if Oct. snow advance is crucial, widespread late Sep. snow may not necessarily tend to help as 2013 has been showing to this point with that big head-start leading to a more gentle slope of the line. It wouldn't surprise me if late Sep. snow tends to hurt since that means there's less bare ground left to which to add snowcover. Let's take it to an extreme to illustrate my point. Let's say that all of Eurasia was covered by snow as of 9/30. Then the Oct. snow advance index can't be higher than 0. Yes, I know that this extreme example is assuming an impossible scenario. However, that's irrelevant to the point I'm making. So, IF widespread late Sep. snow actually does tend to reduce the Oct snow advance index on average, that would mean that late Sep. snowfall is negatively correlated with the winter -AO per the Oct. snow advance theory. If so, wouldn't that be counterintuitive? I think that this is sort of the point that "Wow" is making. So, I feel this is a legit item to discuss/debate.

 

 

 Also, this is in it: "'It’s running well above normal,' said Matt Rogers, president of Commodity Weather Group LLC, a commercial forecaster in Bethesda, Maryland. 'Through the last week of September, it’s the highest snow total in Eurasia since 1977.'" 

 

 Matt seems to me to be considering the last week in Sep.'s high snow total as a positive though I could be misinterpreting him.

[NJwinter23]

 From this link:

 

 "'It’s the best winter predictor that we have,” Cohen said in a telephone interview. “We haven’t made a forecast yet, but we’re watching it closely and the snow cover has definitely been above normal so far.'

The more ground covered by snow across northern Europe and Asia at the end of October, the greater the chances of triggering a phenomenon known as the negative phase of the Arctic Oscillation. That would flood North America, Europe and East Asia with polar air and possibly erect a blocking effect in the North Atlantic that would bottle up the cold in the U.S."

 

 This article says nothing about an Oct. snow advance index. Cohen, himself, implies that the average Oct snow is the crucial factor as opposed to Oct. snow advance. They are not the same thing. So, which is it? Does heavy snow in late Sep. help or hurt -AO chances? If we're talking about average Oct. snow, then there's no doubt that widespread late Sep. snowfall would tend to help since that leads to a head-start in early Oct. However, if Oct. snow advance is crucial, widespread late Sep. snow may not necessarily tend to help as 2013 has been showing to this point with that big head-start leading to a more gentle slope of the line. It wouldn't surprise me if late Sep. snow tends to hurt since that means there's less bare ground left to which to add snowcover. Let's take it to an extreme to illustrate my point. Let's say that all of Eurasia was covered by snow as of 9/30. Then the Oct. snow advance index can't be higher than 0. Yes, I know that this extreme example is assuming an impossible scenario. However, that's irrelevant to the point I'm making. So, IF widespread late Sep. snow actually does tend to reduce the Oct snow advance index on average, that would mean that late Sep. snowfall is negatively correlated with the winter -AO per the Oct. snow advance theory. If so, wouldn't that be counterintuitive? I think that this is sort of the point that "Wow" is making, which seems like a legit reason to discuss/debate this.

 

 

 Also, this is in it: "'It’s running well above normal,' said Matt Rogers, president of Commodity Weather Group LLC, a commercial forecaster in Bethesda, Maryland. 'Through the last week of September, it’s the highest snow total in Eurasia since 1977.'" 

 

 Matt seems to me to be considering the last week in Sep.'s high snow total as a positive though I could be misinterpreting him.

Please stop with this. Do not rely on some chooch over at Bloomberg to properly decipher and choose quotes from Cohen to plug in their article.

 

 

Here are some links to read up on

http://www.nws.noaa.gov/ost/climate/STIP/FY11CTBSeminars/jcohen_062211.pdf

http://web.mit.edu/jlcohen/www/papers/CohenandJones_GRL11.pdf

[GaWx]

Please stop with this. Do not rely on some chooch over at Bloomberg to properly decipher and choose quotes from Cohen to plug in their article.

 

 

Here are some links to read up on

http://www.nws.noaa.gov/ost/climate/STIP/FY11CTBSeminars/jcohen_062211.pdf

http://web.mit.edu/jlcohen/www/papers/CohenandJones_GRL11.pdf

 

 Thanks for the links (although the facepalm was unnecessary). Based on the 2nd link, Cohen is clearly going with the SAI (snowcover increase) as being better than the SCE (average snowcover). That clarifies that. Now, let's consider the idea that widespread late Sep. snowfall may actually REDUCE the SAI. From this link:

 

"Another possibility is that the SAI is sensitive to the

timing of snowfall, where snowfall at the end of the month

contributes to higher values of the SAI and snowfall at the

beginning of the month contributes to lower SAI values

while the monthly‐mean SCE is insensitive to the timing of

snowfall."

 

 So, Cohen has addressed what wow, myself, and others were discussing. Thank you! Cohen is actually saying that beginning of Oct. snowfall does, indeed, lower the SAI. He's saying what I've been thinking may be the case: a higher starting point for Oct. snowcover actually does tend to reduce the SAI after all as opposed to increasing it. So, it appears that widespread snowfall in late Sep. to ~10/1 does, indeed, tend to lead to a lower SAI. Therefore, Cohen is, indeed, implying that widespread snowfall in late Sep. to ~10/1, such as occurred in 2013, is negatively correlated to the average winter -AO! Does anyone here disagree? I'm not debating it at all. He's the expert. However, I do find that to be counterintuitive for obvious reasons. Therefore, this negative correlation is pretty fascinating!

 

 Can anyone think of a logical reason why late Sep. to very early Oct. sub 60 degree N Eurasian snowfall would be negatively correlated to the subsequent winter's average -AO? Again, I'm not doubting Cohen. I'm just wondering about the mechanism since it seems counterintuitive.

[NJwinter23]

 Thanks for the links (although the facepalm was unnecessary). Based on the 2nd link, Cohen is clearly going with the SAI (snowcover increase) as being better than the SCE (average snowcover). That clarifies that. Now, let's consider the idea that widespread late Sep. snowfall may actually REDUCE the SAI. From this link:

 

"Another possibility is that the SAI is sensitive to the

timing of snowfall, where snowfall at the end of the month

contributes to higher values of the SAI and snowfall at the

beginning of the month contributes to lower SAI values

while the monthly‐mean SCE is insensitive to the timing of

snowfall."

 

 So, Cohen has addressed what wow, myself, and others were discussing. Thank you! Cohen is actually saying that beginning of Oct. snowfall does, indeed, lower the SAI. He's saying what I've been thinking may be the case: a higher starting point for Oct. snowcover actually does tend to reduce the SAI after all as opposed to increasing it. So, it appears that widespread snowfall in late Sep. to ~10/1 does, indeed, tend to lead to a lower SAI. Therefore, Cohen is, indeed, implying that widespread snowfall in late Sep. to ~10/1, such as occurred in 2013, is negatively correlated to the average winter -AO! Does anyone here disagree? I'm not debating it at all. He's the expert. However, I do find that to be counterintuitive for obvious reasons. Therefore, this negative correlation is pretty fascinating!

 

 Can anyone think of a logical reason why late Sep. to very early Oct. sub 60 degree N Eurasian snowfall would be negatively correlated to the subsequent winter's average -AO? Again, I'm not doubting Cohen. I'm just wondering about the mechanism since it seems counterintuitive.

 

The face palm is becuase everything you have mentioned in both of your posts has long been clarified already in the prior 3 pages of this thread. Please go back and read through it, everything including your final question in the 2nd post is covered in this thread. It is only 3 pages.

[GaWx]

Perhaps the SAI in some way does address the issue of anomalous snow cover to start off October. Maybe it is the stronger wave driving from later in the month with a strong snow cover advancement that is capable of disturbing the PV early winter. If you start things too soon, the PV will still be developing and the wave strength will not be as strong. We are seeing now the consequences of early wave 1 disturbances, actually aiding in the vortex intensification mostly since it is still in the formative stages.

 

 This is the only post in this thread offering a possible explanation for why anomalous snowfall in late Sep. to very early Oct. on average leads to a less -AOish winter (consistent with Cohen's SAI). Cohen, himself, didn't even offer a possible explanation in his SAI paper. Sounds plausible and is absolutely fascinating. Next Sep. I'll certainly be hoping for a non-snowy Sep. below 60N in Eurasia! This is not how many here, including myself, had previously been thinking about Sep. in the past.

[bluewave]

Anyone know how this other recent  paper compares to Cohen's research?

 

http://onlinelibrary.wiley.com/doi/10.1002/joc.2341/abstract

[NJwinter23]

 This is the only post in this thread offering a possible explanation for why anomalous snowfall in late Sep. to very early Oct. on average leads to a less -AOish winter (consistent with Cohen's SAI). Cohen, himself, didn't even offer a possible explanation in his SAI paper. Sounds plausible and is absolutely fascinating. Next Sep. I'll certainly be hoping for a non-snowy Sep. below 60N in Eurasia! This is not how many here, including myself, had previously been thinking about Sep. in the past.

 

Sorry I may have overreacted. I interpreted your first post as questioning the entire existence  of the SAI and that may have not been what you were doing. The below 60N in Eurasia part has to do with reflecting sunlight (that still exists an ample amount that far south in October) and the feedback it has on the Mongolian high.

[Ginx snewx]

Anyone know how this other recent  paper compares to Cohen's research?

 

http://onlinelibrary.wiley.com/doi/10.1002/joc.2341/abstract

Hers another one shown to me by Radarman. http://water.columbia.edu/files/2011/11/Gong2003RelativeImpacts.pdf  The area and extent in the area, seems to be more valuable then the actually change. Seems intuitive

[Snow_Miser]

Updated comparison between the first of this month and now. 

 

[Snow_Miser]

You can see how awful the SAI was in 2011. There was barely any snow south of 60 N in Eurasia even late into the month.

 

[cfbaggett]

Here's the updated SAI. It still has a gentle slope. There are only 4 days left to October, so I will update the graph at the end of this week. At this point, my hope is that we manage to nudge the slope up to 175,000 km^2/day. I am perfectly content taking our chances with a neutral AO winter. I feel a winter which features transitions has the potential to yield storms.

 

BTW, Mallow and I are personally meeting Dr. Cohen in a few weeks. He's coming to Penn State to give a talk and meet with faculty and students. I'd like to solicit any and all questions from this forum that you may wish to address to him. I will ask him these questions and (assuming I have his permission) will post his responses. So please post your questions, and I will collate them.

[GaWx]

Here's the updated SAI. It still has a gentle slope. There are only 4 days left to October, so I will update the graph at the end of this week. At this point, my hope is that we manage to nudge the slope up to 175,000 km^2/day. I am perfectly content taking our chances with a neutral AO winter. I feel a winter which features transitions has the potential to yield storms.

 

BTW, Mallow and I are personally meeting Dr. Cohen in a few weeks. He's coming to Penn State to give a talk and meet with faculty and students. I'd like to solicit any and all questions from this forum that you may wish to address to him. I will ask him these questions and (assuming I have his permission) will post his responses. So please post your questions, and I will collate them.

 

 2013's slope (green) appears to be about tied with 2007 (blue) for the most gentle of those on the graph. Assuming that holds, does that mean the 2013 slope actually favors a +AO rather than a neutral AO? Why are only 2007, 2009, and 2012 on this graph?

[Jonger]

Western Canada is about to gain a big chunk of coverage over then next 2 days.

[Ginx snewx]

Here's the updated SAI. It still has a gentle slope. There are only 4 days left to October, so I will update the graph at the end of this week. At this point, my hope is that we manage to nudge the slope up to 175,000 km^2/day. I am perfectly content taking our chances with a neutral AO winter. I feel a winter which features transitions has the potential to yield storms.

 

BTW, Mallow and I are personally meeting Dr. Cohen in a few weeks. He's coming to Penn State to give a talk and meet with faculty and students. I'd like to solicit any and all questions from this forum that you may wish to address to him. I will ask him these questions and (assuming I have his permission) will post his responses. So please post your questions, and I will collate them.

Please ask him about this recent paper and the main difference between his theory and this paper. Do regions and extent matter more than actual gain?

 

Regions of autumn Eurasian snow cover and associations with North American winter temperatures

1. Thomas L. Mote^*, 
2. Emily R. Kutney
3.  

[cfbaggett]

 2013's slope (green) appears to be about tied with 2007 (blue) for the most gentle of those on the graph. Assuming that holds, does that mean the 2013 slope actually favors a +AO rather than a neutral AO? Why are only 2007, 2009, and 2012 on this graph?

 

2007 has the smallest slope in the dataset while 2009 has the steepest. Thus, I include those two years to give some idea of the range of possible slopes. I started making this chart last year so that is why you see 2012 on it. I probably should have deleted it.

[cfbaggett]

 2013's slope (green) appears to be about tied with 2007 (blue) for the most gentle of those on the graph. Assuming that holds, does that mean the 2013 slope actually favors a +AO rather than a neutral AO? Why are only 2007, 2009, and 2012 on this graph?

And yes, if the current slope holds, the SAI is indicating a positive AO winter. However, if we see snowcover growth over the next few days, then the slope could be pushed up to a more neutral AO signal.

[Snow_Miser]

Pretty large gain in Eurasian snowcover yesterday. Optimistic that the SAI will signal a neutral AO during winter. 

[Snow_Miser]

Snowcover took a big hit today on the IMS Snowmaps. Not good for the October SAI. 

[The_Global_Warmer]

The SIA slope will probably be even more unfavorable looking at the IMS like snowlover pointed out above and now the Rutgers snow lab charts also show a huge retreat on the Western flank of the Snow cover in Eurasia.

 

 

This is where I wonder how a few days even matter.  By tomorrow a massive region of SIberia might see snow cover go from above normal to way below.  How much can this really matter?  How do we decifer around this kind of short term change to ensure the predictive value isn't to skewed?

 

[Jonger]

Probably has a lot to do with the AO being so positive, cold isn't dropping far south yet, for once I wished Jennifer Francis was right.

[Wow]

If you shift the 31 day window back 4 days (Sept 27 - Oct 28), I imagine the all-important slope would be significantly different.

[cfbaggett]

If you shift the 31 day window back 4 days (Sept 27 - Oct 28), I imagine the all-important slope would be significantly different.

 

That is an interesting question, so I tested it by calculating the slope from 9/27 to 10/27 (31 days). Unfortunately, the change in the slope is insignificant. There's no way of getting around the fact that late September and early October had anomalously high snow cover.

[Sunny and Warm]

 

If you shift the 31 day window back 4 days (Sept 27 - Oct 28), I imagine the all-important slope would be significantly different.

I believe in the science behind the SAI work, but I still get a little queasy using hard and fast calendar dates to predict weather phenomena. Ma Nature doesn't understand that it's the 28th of Sep or the 2nd of Oct, and do those four days really matter in the development of the features that may/will influence future oscillation signals?? I don't know. It seems too arbitrary, but then again, it appears to work, so maybe I'm barking up the wrong tree.

[forkyfork]

That is an interesting question, so I tested it by calculating the slope from 9/27 to 10/27 (31 days). Unfortunately, the change in the slope is insignificant. There's no way of getting around the fact that late September and early October had anomalously high snow cover.

we went from higher than average snow cover to nearly below average. pretty much worst case for those rooting for a -AO

[heavy_wx]

 

I believe in the science behind the SAI work, but I still get a little queasy using hard and fast calendar dates to predict weather phenomena. Ma Nature doesn't understand that it's the 28th of Sep or the 2nd of Oct, and do those four days really matter in the development of the features that may/will influence future oscillation signals?? I don't know. It seems too arbitrary, but then again, it appears to work, so maybe I'm barking up the wrong tree.

 

Except the SAI is calculated using a linear regression of the snow cover each day of the month of October. Changing a few of the values at the end and beginning of the period is likely not going to have much of an impact on the regression coefficient. It makes sense to use a regression coefficient for this reason because it doesn't weight the snow cover values at the beginning or the end of the month more than any of the intermediate values.

[Sunny and Warm]

 

Except the SAI is calculated using a linear regression of the snow cover each day of the month of October. Changing a few of the values at the end and beginning of the period is likely not going to have much of an impact on the regression coefficient. It makes sense to use a regression coefficient for this reason because it doesn't weight the snow cover values at the beginning or the end of the month more than any of the intermediate values.

Thank you. That makes perfect sense now.

[skierinvermont]

Calendar dates do matter. The stratospheric PV develops at a particular time of year. If snow cover increases too early, the mechanism for disrupting the PV (whatever that may be) may not be effective. HM has already explained this.

 

In addition, the slope of SAI is likely not all that sensitive to a change in start date of only a few days. As shown above, in this year shifting the date 4 days earlier made little difference to the slope. It's a linear regression derived using all 31 data points. Adding or subtracting 4 isn't going to radically alter it.

[cfbaggett]

Here's the SAI for October. It seems to be indicating a moderately positive AO winter.