Winter 2014-2015 Thread

[mitchnick]

major gains in snowfall around the Pole today

yesterday:

 

today:

[WxUSAF]

Booo.  We don't want snowcover now. 

[mattie g]

Mitch...As long as there aren't big gains below 60N until we get into mid-October, it can snow all it wants up near the poles.

[T. August]

Mitch...As long as there aren't big gains below 60N until we get into mid-October, it can snow all it wants up near the poles.

What's wrong with snow cover now?

[WxUSAF]

Mitch...As long as there aren't big gains below 60N until we get into mid-October, it can snow all it wants up near the poles.

Is 60N a cutoff?  We want snow in eastern Siberia in October and there isn't much of that below 60N. 

[mitchnick]

Booo.  We don't want snowcover now. 

 

 

Mitch...As long as there aren't big gains below 60N until we get into mid-October, it can snow all it wants up near the poles.

1) last year, the snow cover expanded rapidly at the end of September and look what happened

2) Cohen's forecast last year for the winter based upon the lack of snow cover in Siberia in October was a huge fail

http://www.nsf.gov/news/special_reports/autumnwinter/predicts.jsp

 

I'll take as much buildup of snow around the Pole as soon possible

[WxUSAF]

What's wrong with snow cover now?

The best statistical correlation to a winter -AO is the RATE of increase of snowcover in Siberia during October.  So, if we start off low, we have a better chance at getting good gains during the month.  Last year there was lots of snow early in October, which gave us a low rate of increase and we ended up with a +AO in the means for the winter.

[WxUSAF]

1) last year, the snow cover expanded rapidly at the end of September and look what happened

2) Cohen's forecast last year for the winter based upon the lack of snow cover in Siberia in October was a huge fail

http://www.nsf.gov/news/special_reports/autumnwinter/predicts.jsp

 

I'll take as much buildup of snow around the Pole as soon possible

His forecast was a fail, but the basic prediction of a +AO verified. 

[mitchnick]

His forecast was a fail, but the basic prediction of a +AO verified. 

-EPO ruled the roost, and the GOA remains favorable for that to continue

back to Cohen's stuff, I have a hard time believing that snow around the Pole a few days before 10/1 hurts, especially if snow cover keeps building after 10/1

[WxUSAF]

-EPO ruled the roost, and the GOA remains favorable for that to continue

back to Cohen's stuff, I have a hard time believing that snow around the Pole a few days before 10/1 hurts, especially if snow cover keeps building after 10/1

We'll see on the EPO.  But, we're a week before 10/1.  I doubt it's precise enough for 9/30 or 10/2, but it's still early.

[mitchnick]

We'll see on the EPO.  But, we're a week before 10/1.  I doubt it's precise enough for 9/30 or 10/2, but it's still early.

right, and you can see the substantial snow cover build up at the end of September, 2002 (days 268-273)

http://www.natice.noaa.gov/pub/ims/ims_gif/ARCHIVE/NHem/2002/

[mattie g]

-EPO ruled the roost, and the GOA remains favorable for that to continue

back to Cohen's stuff, I have a hard time believing that snow around the Pole a few days before 10/1 hurts, especially if snow cover keeps building after 10/1

There was a lot of discussion last year about how people didn't think that the massive gains in coverage in September would likig the chances of the AO going negative. But while Cohen's temperature forecast didn't verify, as WxUSAF mentioned, the AO was positive overall because the rate of advance was low in October.

Obviously, last year showed us that the AO isn't the be-all and end-all - the Pacific ruled the roost and drove our winter - but getting the SAI on our side would certainly be a good feeling. I agree about an increase in coverage way up north - it can increase all it wants up there at this point!

[T. August]

The best statistical correlation to a winter -AO is the RATE of increase of snowcover in Siberia during October.  So, if we start off low, we have a better chance at getting good gains during the month.  Last year there was lots of snow early in October, which gave us a low rate of increase and we ended up with a +AO in the means for the winter.

Ah, okay. Thanks.

[mattie g]

Is 60N a cutoff? We want snow in eastern Siberia in October and there isn't much of that below 60N.

From what I've read, the best correlation is the rate of advance in Eurasia (especially Siberia) between 25N-60N. I'm certainly open to be proven wrong on that.

[JoMo]

Since you all are talking AO, and it's almost October. How well did that "October Pattern Index" do last year?

 

Here's the relevant threads from last year:

 

http://www.americanwx.com/bb/index.php/topic/41379-october-pattern-index-predicting-winter-ao-from-october-with-90-accuracy/

http://www.americanwx.com/bb/index.php/topic/41639-seasonal-winter-20132014-forecast-based-on-opi-index/

[mitchnick]

idk about the AO

I know Wes has shown many of our decent "events" have come during -AO periods, this monthly list of AO readings shows that a -AO winter does not guarantee a snowy winter...in fact, there are a few real stinkers in there snow-wise for us during -AO winters and some decent winters with +AO

http://www.cpc.ncep.noaa.gov/products/precip/CWlink/daily_ao_index/monthly.ao.index.b50.current.ascii.table

[mitchnick]

Since you all are talking AO, and it's almost October. How well did that "October Pattern Index" do last year?

 

Here's the relevant threads from last year:

 

http://www.americanwx.com/bb/index.php/topic/41379-october-pattern-index-predicting-winter-ao-from-october-with-90-accuracy/

http://www.americanwx.com/bb/index.php/topic/41639-seasonal-winter-20132014-forecast-based-on-opi-index/

thanks for that 

[StudentOfClimatology]

-EPO ruled the roost, and the GOA remains favorable for that to continue

back to Cohen's stuff, I have a hard time believing that snow around the Pole a few days before 10/1 hurts, especially if snow cover keeps building after 10/1

The GOA temps don't correlate to the amplitude of the ridge, though. There's plenty of warm-GOA analogs that failed to extend the ridge into AK and ended up torching us..usually under a +AO.

The statistics suggest you don't want to build up the snowpack yet.

[Bob Chill]

The +ao verified in Dec last year. We should all count our blessings that it didn't bleed well into Jan. Maybe it broke down because it went big + during Nov. Maybe the epo overwhelmed and took control.

There is solid data supporting the fact that a stout +ao in December is terrible here. The snowcover stuff is not perfect. Do we want to test it again? I sure don't.

[mattie g]

The +ao verified in Dec last year. We should all count our blessings that it didn't bleed well into Jan. Maybe it broke down because it went big + during Nov. Maybe the epo overwhelmed and took control.

There is solid data supporting the fact that a stout +ao in December is terrible here. The snowcover stuff is not perfect. Do we want to test it again? I sure don't.

Yup. It's not perfect in predicting what kind of winter we'll have overall, but it's a good indicator of the AO state for DJF.

[Cold Rain]

There is broad agreement that the rate of snow cover build up over eastern Siberia through October correlates well to the winter AO state.  But it's not the driver.  It would be interesting to understand what mechanism drives the snow cover advance.  If that was well understood, it might be able to see how snow cover might build somewhat in advance of October.  Are there certain atmospheric patterns or configurations of SSTs that correlate to building snow cover in key areas in October as opposed to earlier?  Maybe this is already known and I just missed it.

[Eskimo Joe]

There is broad agreement that the rate of snow cover build up over eastern Siberia through October correlates well to the winter AO state.  But it's not the driver.  It would be interesting to understand what mechanism drives the snow cover advance.  If that was well understood, it might be able to see how snow cover might build somewhat in advance of October.  Are there certain atmospheric patterns or configurations of SSTs that correlate to building snow cover in key areas in October as opposed to earlier?  Maybe this is already known and I just missed it.

Some nice gains in Russia recently.

 

[millwx]

Some nice gains in Russia recently.

Day268-2014.png

 

If you believe some of the previous discussions on this board, this is irrelevant.  And, in fact, it may actually be a bad thing.  Why?  Because it's TOO soon.  It is only the gains in October that matter.  I've even heard some (attempt to) attribute some physical processes as to why late September snowfall doesn't matter.

 

Frankly, it's bull, and you are correct to point it out.  This snow cover is a good thing.  I've run the numbers myself.  I admit, I'm kind of a data nerd, lol.  But I just get skeptical trusting other's analyses when we have all the data we need on hand.  Snow data is readily available online, as is the AO data.  I've run it all myself...

 

It is true, as the SAI promoters suggest, that the highest correlation between Eurasian snow cover and the AO is on the advance of the snow cover, NOT on an absolute value.  However, that correlation (-0.5472) occurs on the delta between Week 43 and Week 39.  Week 39 is 9/23 - 9/30.  So, the argument that last September snowfall is "too soon" or irrelevant is just total b.s.  Mother Nature doesn't look at a calendar and flip a switch on October 1st.  And the numbers hold to that.  Mind you, the correlation using the delta between Week 43 and 40 (Week 40 starts at 10/1) isn't much lower (-0.5217), but it IS lower.  This late September snow advance is relevant and it is good.  Arguably, since we're IN week 39 and looking at the delta, maybe we don't want Week 39 to be high yet (don't want to subtract off a large number), but just AFTER... which could imply that what I'm saying is wrong and, in fact, we don't want to start watching until Week 40.  BUT... I've done the correlations on all of the week pairings, and I can tell you that even Week 43 minus Week 38 (9/16 - 9/22) has a higher correlation (-0.5373) than Week 43-40.  Bottom line... late September matters... unquestionably.

 

Moreover, while I said that the correlations support the SAI proponents, I'm not totally sold on the increase being most important anyway.  The correlation between AO and just the total snow cover in a given week (not the increase) barely comes in under the peak increase correlation.  The top correlation to raw snow cover values is -0.5259 (in Week 43 ...not coincidentally, I'm sure, the end-week on the peak increase correlation... indicating that the first week that you're subtracting off is of almost negligible importance).

 

Furthermore, the AO is not a perfect harbinger of U.S. winter temps anyway.  The correlation between Eurasian snow cover and population-weighted U.S. temperatures (which is what we use in my industry... but the population in the U.S. leans heavily to the East... so, this is VERY relevant to this board) actually does NOT peak on a snow increase value.  It peaks on an absolute snow cover value.  And, in fact, it's not even remotely close.  The highest correlation you get between U.S. winter temps and Eurasian snow increase is a putrid +0.1673 (correlation is positive because it's on U.S. heating degree days, not temperature)... that's on Week 43 minus Week 37.  But whatever... that's a pretty poor correlation.  The BEST correlation is on the absolute snow cover value... NOT on the increase!  The top correlation (still modest, lower than the AO correlation itself, but MUCH better than the HDD correlation with the snow increase), is +0.3024.  That is the correlation of the U.S. winter temps with Week 42 (10/15 - 10/21).

 

So, for U.S. (mainly Eastern U.S.) temperatures what matters most in Eurasian snow cover is the absolute amount by mid or mid-to-late October.  It's really as simple as that.  Sometimes I think we over-complicate matters.  If Eurasian snow cover is high in mid-October, that's a positive sign for a cold winter (not a guarantee... a 0.3 correlation is far, far from a home run... just a good indicator).  Simple as that.  (And, as an aside, if the GFS ends up even in the rough ballpark of correctness, snow cover by mid-Oct should be above normal... but that's pretty far out there... and it's the GFS... so, we'll see.)

[Eskimo Joe]

Nice write up Millwx...thanks for sharing your expertise.  I have always been curious as to what the snow depth is before sites such as Rutgers display it.

[Cold Rain]

Yup, agree with EJ.  Great write-up and very interesting.  Thanks for taking the time to research and present the data.

[mitchnick]

excellent Millwx....excellent

[NEchestercountydude]

Wow millwx what a write up, we're having a discussion in the main weather forum about SAI and SCE and this helps explain some of my questions greatly. Hope you don't mind but I posted it in the main forum. Hopefully we can all learn more from your findings.

[WxUSAF]

Nice work, millwx.  That corrects a couple of my posts.  

[millwx]

Thnx all.  I should add that the correlation between Week 39 (the current week) and Week 42 snowfall is NOT high... but it is positive and significant (0.243).  Further proof that late September snowfall matters (easier to retain above normal snowpack if you start early... I don't think there's anything magical in that... I don't think late September snow magically creates early to mid-October snow; more likely it's just related to a dominant pattern which would favor both).

 

Also, that correlation increase to 0.412 in Week 40 (10/1 - 10/7) and to 0.670 in Week 41 (10/8 to 10/14).

 

And the correlation suffers due to scatter.  If you're merely curious as to whether or not the weeks "match" (e.g., how often does Week 39 above normal snow cover couple with above or below normal Week 42 snow cover).  Well, in just being directionally predictive - which is ultimately what matters - Week 39 has the same signal as Week 42 60% of the time.  Week 40 has the same signal 66% of the time.  Week 41 matches Week 42, as you might expect, a whopping 80+% of the time.  So, by Week 41 (2nd week in Oct) we should have a solid idea what signal this providing.  And right now, we already have some "clues".

[EastCoast NPZ]

Thanks millwx.  That was an enlightening and fascinating post.

 

I think it's time for snow weenies everywhere to pool their money to buy and ship thousands of snow-making machines to Russia.  Maybe Midlo will donate his....